Mechanistic studies of a AAA+ Protease
by
Andrew R. Nager
B.S. Chemistry, Molecular and Cellular Biology
Vanderbilt University, Nashville, TN, 2008
SUBMITTED TO THE PROGRAM OF BIOLOGY (COURSE 7) IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
ARCHVES
December 2012
MASSACHUSETTS INSTIUTE
©2012 Massachusetts Institute of Technology.
All rights reserved.
LiBRARIES
71-'\
1i)
Signature of Auth.
Biology
December 04, 2012
Certified by:
Robert T. Sauer
Salvador E. Luria Professor of Biology
Thesis Supervisor
Accepted by:
-W.0of
Stephen Bell
Professor of Biology
Co-Chair, Biology Graduate Committee
I
Mechanistic studies of a AAA+ Protease
by
Andrew R. Nager
Submitted to the program of Biology (Course 7) on December 04, 2012
in partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Biology
at the Massachusetts Institute of Technology
Abstract
AAA+ proteases are present in all branches of life and responsible for the energy-dependent
degradation of most cytosolic proteins. Substrates for AAA+ proteases are unfolded and
translocated into a compartmental peptidase. The requirement for protein unfolding raises several
questions. How easily are proteins unfolded within the native environment of a cell? Are some
proteins more difficult to unfold than others, and, if so, why? How do AAA+ ATPases convert
the chemical energy of ATP binding and hydrolysis into mechanical unfolding and translocation?
ClpXP is a AAA+ protease that consists of the hexameric ClpX unfoldase and polypeptide
translocase and the ClpP compartmental peptidase. ClpX binds a substrate by an unstructured
degradation tag and then, by multiple rounds of ATP-binding and hydrolysis, unfolds and
translocates the substrate into the proteolytic chamber of ClpP. To study the features that allow a
protein to resist unfolding, I investigate the degradation of degron-tagged Green Fluorescent
Protein (GFP; Chapter 2). By engineering GFP substrates, I determine the steps of GFP
unfolding and how structure local to the degron can hinder ClpX-mediated unfolding. In later
chapters, my collaborators and I use ensemble and single-molecule fluorescent assays to study
the mechanochemical cycle of ClpX 6 . By these assays, we observe that subunits adopt unique
classes which differ in structure and nucleotide binding and hydrolysis, subunit classes switch in
a thermally-driven probabilistic fashion that is decoupled from the chemical cycle, and ClpX 6
form a staircase architecture similar to AAA+ helicases.
Thesis supervisors: Robert T. Sauer, Tania, A. Baker
Titles: Salvador E. Luria Professor of Biology, Whitehead Professor of Biology
2
Acknowledgements
Many people have contributed to my graduate school experience and, ultimately, this thesis. My
advisors, Bob Sauer and Tania Baker, have shaped and encouraged my work. Furthermore, they
have been role models, showing me how to be a productive scientist, teacher, speaker, and writer.
Similar, Prof. Monty Krieger has been an invaluable mentor, expanding my view of cell biology
and teaching. Thanks to Prof. Matt Lang who has taught me much about single-molecule
techniques and is an endless source of excitement (even when I suggest horrible experiments).
Many thanks to Profs. Stephen Bell, Amy Keating, and Thomas Schwartz. Each has influenced
me either through classes, committee meetings, or one-on-one conversations. Additionally, I still
think Stephen Bell took a risk letting me into the graduate program. Thank you Profs. Richard
Schwartzstein, Robert Sackstein, and Elazer Edelman for letting me audit coursework at Harvard
Medical School and Health Sciences & Technologies.
To the members of the Sauer, Baker, Krieger, Keating, Schwartz, Laub, Gilbert, and King labs
for collegial discussions and improving the work (and after work) environment. Thank you Joey
Davis, Mary Lee, J Sohn, Shankar Sundar, Jiejin Chen, Vlad Baytshtok, Anna de Regt, Jon
Grabenstatter, Seokhee Kim, Karl Schmitz, Ben Stinson, Ohad Yosefson, Santiago Lima, Randall
Mauldin, Peter Chien, Leonid Gaidukov, Ayce Yesilaltay, Marsha Penman, Scott Chen, Jen
Kaplan, Chris Negron, Karl Hauschild, Evan Thompson, Kevin Knockenhauer, Brian Sosa,
Kasia Gora, Pavan Vaidyanathan, and Oksana Sergeeva. Thank you to the students that I have
mentored and learned from, both as a teaching assistant and graduate resident tutor. I would
especially like to thank Steven Glynn, Orr Ashenberg, Sandra Kim, Adrian Olivares, Josh
Arribere, Mary Kay Thompson, and Nathaniel Schafheimer. Steve trained me in crystallography
and how to eat large numbers of gummy bears at the synchrotron. Orr is a continuing resource
on computational biology and we have shared many unproductive evenings at bars. Sandy
encouraged my interests in medicine and influenced my decision to study primary cilia. Adrian,
in addition to being Beer Tzar and commissioner of our fantasy football league, has been an
intellectual resource and eager debater of kinetics. Josh, Mary Kay, and Nathaniel have added
much color (often from red wine) to my graduate school experience.
Special thanks to Brenda Pepe, the lab administrator. I have incorrectly filed and ordered many
things over the past four years. Thank you for your never ending help.
Finally, thank you to my parents and family for their unwavering support. To my Dad who
shows off each of my papers to his work colleagues (even the second author publications!). To
my Mom who, when I need to relax, is close and caring and, when I need to work, gives me
distant support. Then to my grandfather who has long fostered my love for science and instilled
in me a multi-disciplinary, open approach to inquiry.
3
To my grandfather Maxwell Nager who taught me that science can be creative.
4
Table of Contents
Chapter 1 AAA+ Proteases.....................................................................................................
10
Intracellular Proteolysis ......................................................................................................
11
AAA+ Proteases ......................................................................................................................
13
Protein-unfolding in a native environm ent.......................................................................
16
Structures of AAA+ M achines ...........................................................................................
21
Subunit coordination within ATP-fueled m otors .............................................................
25
The F1 ATPase: a sequential rotary engine ........................................................................
26
Cytoplasm ic Dynein walkers: a stochastic stum ble ...........................................................
29
AAA+ unfoldases have multiple classes of subunits.........................................................
31
AAA+ m otors: stochastic, concerted, or sequential?.............................. . ........................ . . 31
References................................................................................................................................
36
Chapter 2 Stepwise unfolding of a p barrel protein by the AAA+ ClpXP protease.......... 44
Abstract....................................................................................................................................
45
Introduction.............................................................................................................................
46
Results ......................................................................................................................................
49
ClpXP extraction of terminal P strands in split-GFP variants ..........................................
49
GFP fluorescence during ClpXP stalling supports terminal-strand extraction .................. 52
54
Stalling behavior of circularly permuted GFP variants ....................................................
Stalling substrates have lower maximal rates of ClpXP degradation ................................
60
Equilibrium and kinetic stability...........................................................................................
61
Dependence of rates of ATP hydrolysis and degradation on ATP concentration............... 62
Discussion.................................................................................................................................
64
68
M aterials and M ethods.......................................................................................................
Protein Expression, Purification, and Cleavage................................................................
68
TAM RA-labeled fluorescent peptides ...............................................................................
70
70
Biochem ical Assays ..............................................................................................................
71
Acknowledgem ents .................................................................................................................
72
References................................................................................................................................
Chapter 3 Nucleotide binding and conformational switching in the hexameric ring of a
79
AAA+ m achine ............................................................................................................................
80
Abstract....................................................................................................................................
Introduction.............................................................................................................................
81
84
Results ......................................................................................................................................
84
New crystal structures.......................................................................................................
85
Evidence supporting 4L:2U and 5L:1U subunit arrangem ents.........................................
86
A test of subunit switching................................................................................................
89
An assay for subunit-specific nucleotide binding .............................................................
92
Subunit-specific conformational changes ........................................................................
Locking subunits in the L conformation prevents unfolding and degradation .................. 94
96
Subunit communication and ATP hydrolysis ....................................................................
98
Discussion.................................................................................................................................
98
Setting and resetting the configuration of the ClpX ring ..................................................
100
Evidence for subunit switching...........................................................................................
102
Structural and functional classes of ClpX subunits ............................................................
103
Using CoM ET assays to study multimeric proteins ...........................................................
5
Experimental procedures .....................................................................................................
M aterials .............................................................................................................................
104
10 4
P ro tein s ...............................................................................................................................
10 4
Crystallization and structure determination........................................................................
105
Fluorescence assays ............................................................................................................
107
Biochemical assays .............................................................................................................
108
Acknowledgements ................................................................................................................
110
References...............................................................................................................................111
Chapter 4 The stochastic mechanism of a AAA+ machine observed by single-molecule
CoM ET........................................................................................................................................116
Abstract...................................................................................................................................117
Introduction............................................................................................................................118
Results.....................................................................................................................................119
CoM ET of a ClpX subunit...................................................................................................119
L *U switching is independent of nucleotide hydrolysis...................................................
122
L *U kinetic analysis .........................................................................................................
123
Hydrolysis-dependent L motions ......................................................................................
124
Discussion...............................................................................................................................
126
L @U switching and L motions are uncoupled...................................................................
126
Subun it classes ....................................................................................................................
12 6
ClpX mechanism and other AAA+ machines.....................................................................
128
Acknowledgements ...............................................................................................................
129
References..............................................................................................................................
130
Chapter 5 Polarized TIRFM of ClpX rigid bodies differentiates subunits within a hexamer
..........................................................................
133
Introduction..........................................................0
.......................................
Experimental Design......................................................................................................
Results .........................................................................................................................
Discussion.......................................................................................................................
References................................................................................
. .....
Appendix A Stalling of cp6a-SFGFP-ssrA .............................................................................
Extraction of an a helix .................
....................................
134
136
140
143
148
152
153
Appendix B Supplement for Ensemble CoMET of ClpX.........................
155
Appendix C Supplement for Single-Molecule CoMET of ClpX......................
Supplemental M ethods .......................................................................................................
Appendix D Catalog of ClpX mutations ....................................
Tether and IGF loop truncations for ClpX crystallography .............................................
Cysteine mutations.......... .......................................................
.......................
cCoM ET mutations and pairs..............................................
o.........................................
Mutations for single-molecule nanometry ....................................
ClPPplatform......................................................................
........................................
Synthetic ClpX constructs......
........................................
165
169
170
171
173
177
180
180
181
6
List of Figures
Chapter 1 AAA+ Proteases
12
Figure 1.1 ClpXP, AAA+ Protease......................................................................................
15
Figure 1.2 Single-molecule nanometry...............................................................................
16
Figure 1.3 Unfolding by AAA+ unfoldases........................................................................
17
Figure 1.4 Kinetic and global stability. ..............................................................................
....
19
ClpXP
degradation.
and
the
rate
of
Figure 1.5 Relationship between global stability
Figure 1.6 Structures of substrates used for ClpXP-degradation studies....................... 20
22
Figure 1.7 Asymmetric structure of ClpX. .......................................................................
23
Figure 1.8 Rigid-body subunit interface..........................................................................
24
Figure 1.9 ClpX pore loops..................................................................................................
27
Figure 1.10 Sequential nucleotide cycle of F1 ATPase. .....................................................
28
Figure 1.11 Asymmetric interactions of the 7 stalk...........................................................
30
Figure 1.12 Cytoplasmic Dynein walkers ..........................................................................
Figure 1.13 Models for coordination of nucleotide hydrolysis within AAA+ hexamers... 32
35
Figure 1.14 Models for mixed subunit coordination.........................................................
Chapter 2 Stepwise unfolding of a P barrel protein by the AAA+ ClpXP protease
48
Figure 2.1 Green Fluorescent Protein ...............................................................................
51
Figure 2.2 A stable 10-stranded barrel...............................................................................
53
Figure 2.3 An unfolding intermediate is populated during stalling ...............................
Figure 2.4 Circularly permuted GFP variants show stalling and non-stalling ClpXP
55
degradation..............................................................................................................................
Figure 2.6 Effects of ClpXP versus ClpX extraction of terminal peptides from thrombin56
split substrates.........................................................................................................................
59
..............................................................................
assays
Figure 2.7 Strand-replacement
61
Figure 2.8 Equilibrium and kinetic stability of GFP variants ........................................
63
Figure 2.9 Two-step unfolding ..........................................................................................
Chapter 3 Nucleotide binding and conformational switching in the hexameric ring of a
AAA+ machine
83
Figure 3.1 ClpX structure ...................................................................................................
Figure 3.2 VI mutations alter the ATP dependence of ClpX function............................ 89
90
Figure 3.3 nCoMET detects nucleotide binding to specific subunits .............................
Figure 3.4 cCoMET detects conformational changes in specific subunits..................... 94
96
Figure 3.5 Effects of L-lock disulfides on ClpX function .................................................
Figure 3.6 ATP hydrolysis by a variant with one binding site does not support function 97
98
Figure 3.7 Model for ring-setting and ring-resetting reactions ......................................
Chapter 4 The stochastic mechanism of a AAA+ machine observed by single-molecule
CoMET
121
Figure 4.1 Single-molecule cCoMET of a ClpX subunit ...................................................
Figure 4.2 Dwell time probability distributions reveal hidden L classes ......................... 124
125
Figure 4.3 Transition density plots ......................................................................................
127
Figure 4.4 Two mechanical cycles........................................................................................
a
hexamer
within
Chapter 5 Polarized TIRFM of ClpX rigid bodies differentiates subunits
135
Figure 5.1 Rotations at the hinge.........................................................................................
137
Figure 5.2 Hinge rotations position adjacent subunits ......................................................
138
Figure 5.3 Polarized TIRF microscopy ...............................................................................
7
Figure 5.4 polTIRFM of BFR-Cl X6 ..-.......---...............................................................
Figure 5.4 Construction of ClpPp lform ....
...............................
Figure 5.5 polTIRFM trajectories and hidden Markov fits..............................................
Figure 5.6 ClpX rigid bodies occupy five distinct angles ..................................................
Figure 5.7 One-step dwell-time distributions.....................................................................
Figure 5.8 Ring structure and potential tilt........................................................................
Figure 5.9 Transitions between ring positions....................................................................
Appendix A Stalling of cp6a-sFGFP-ssrA
Figure 1 Circularly permutated GFP..................................................................................
Figure 2 Extraction of an a helix, not a P strand, results in stalling for cp6a.................
Appendix B Supplement for Ensemble CoMET of ClpX
Figure 1 ATP hydrolysis and degradation by tethered ClpX trimers ..............................
Figure 2 Size exclusion chromatography of W-W-W and W-VI-W...............
Figure 3 Stoichiometry of nucleotide binding to W-VI-W and W 6 . .
. . . . . . . . . . . . . .
139
140
140............
141
142
143
145
146
154
154
156
157
157
Figure 4 ATP dependence of substrate degradation by ClpX hexamers with single mutant
subunits ..................................................................................................................................
158
Figure 5 Co2+ supports ClpX activity..................................
.................
......... 158
Figure 6 Co2+ inhibits peptide cleavage by ClpP................................................................
159
Figure 7 ATP hydrolysis by M363C labeled and unlabeled ClpX variants ..................... 159
Figure 8 ADP binding to W-VI-W by nCoMET.................................................................
160
Figure 9 ADP-dependent conformation changes of W-VI-W by cCoMET ..........
160
Figure 10 nCoMET and cCoMET of W-W-W ..................................................................
161
Figure 11 ClpP pore opening with L-locked variants ........................................................
162
Figure 12 nCoMET quenching is specific to ATP ..............................................................
162
Figure 13 Changes in cCoMET fluorescence depende on Ni 2 +-NTA................................ 163
Figure 14 cCoMET across the rigid-body interface...........................................................
163
Appendix C Supplement for Single-Molecule CoMET of ClpX
Figure 1 Single-molecule trajectories in the presence of saturating ATP........................ 166
Figure 2 Ni2+-NTA binding to the ClpX His7 -X3-His76 motif ........................................... 166
Figure 3 Single-molecule trajectories of D
CTAMRA K 3 3 0CTAMRA
76
contact quenching with
saturating ATP ......................................................................................................................
166
Figure 4 Nucleotide-occupancy switches the average L:U subunit ratio 4L:2U to 5L:1U
... ...................................
.............................................
. ............... 167
Figure 5 U dwell-time probability distribution with different ADP concentrations....... 167
Figure 6 L dwell-time probability distribution with 1 mM ATPyS................ 167
Figure 7 Simulated L dwell-time distributions...........
...................................
168
Figure 8 Dwell-time probability distributions for a low-affinity subunit at saturating
AD P .........................
. .......... ....... ...................... ...............................................
168
Appendix D Catalog of ClpX mutations
Figure 1 Tether truncations......................................
.......... ............................................ 172
Figure 2 IGF loop truncations .............................................................................................
173
Figure 3 ClpX cysteine m utations ................ ................................................................
175
Figure 4 cCoM ET design..................................................................................................
178
Figure 5 cCoMET with different pairs...............................................................................
179
Figure 6 Expression of the N-terminal sortase ClpX trimer............................................. 181
8
List of Tables
Chapter 1 AAA+ Proteases
Table 1.1 Degradation of ClpXP substrates with different stabilities............................. 18
Table 1.2 ClpXP degradation of substrates with ssrA tags located at different positions in
the structure. ...........................................................................................................................
20
Chapter 2 Stepwise unfolding of a P barrel protein by the AAA+ ClpXP protease
60
Table 2.1 Properties of ssrA-tagged GFP substrates.........................................................
Chapter 3 Nucleotide binding and conformational switching in the hexameric ring of a
AAA+ machine
Table 3.1...................................................................................................................................
91
Chapter 4 The stochastic mechanism of a AAA+ machine observed by single-molecule
CoMET
121
Table 4.1 Dwell times of cCoMET transitions....................................................................
155
Appendix B Supplement for Ensemble CoMET of ClpX......................................................
156
Table 1 Crystallographic statistics.......................................................................................
164
Table 2 cCoMET/TT fit parameters....................................................................................
170
Appendix D Catalog of ClpX mutations.................................................................................
173
1
IGF
loop
truncations
...............................................................................................
Table
177
Table 2 Cysteine and alanine mutations .............................................................................
179
Table 7.3 cCoMET i/i+4 histidine motifs ............................................................................
180
Table 4 Pore loop and ATPase mutations ...........................................................................
181
Table 5 ClPpp "at"rm
............................................................................................................
182
Table 6 Synthetic ClpX multimers .................................................
182
..........................
Table 7 Monomer-to-multimer primers..........
9
Chapter 1
AAA+ Proteases
10
Intracellular Proteolysis
The proteome is continuously molded by controlled protein degradation to clear misfolded
proteins, maintain steady-state protein concentrations, or facilitate a rapid response to stress. For
example, during heat shock, if protein misfolding were left unchecked, the exposure of
hydrophobic residues within the crowded cellular environment would lead to the formation of
toxic aggregates. 1
As another example, proteins can be targeted for degradation as a
consequence of signaling pathways. Because protein degradation is irreversible, degradation is
involved in decisive signaling pathways such as inflammation and apoptosis.2 However, both
examples require that a target protein be degraded within the cellular milieu, while off-target
proteins are left unscathed.
For a cell under any condition, it would be detrimental to degrade a spectrum of off-target
proteins, and thus degradation must be highly specific. If a weakly-specific protease, such as
trypsin, were expressed within a cell, it would proteolyze both desired and undesired targets. As
such, cells have evolved two strategies for controlled, intracellular proteolysis.
One strategy
enlists highly-specific proteases that cleave substrates only after binding unique amino-acid
sequences.
For example, caspase-3 is a protease that, upon activation, cleaves at a DEVDG
motif found in caspase-6 and -7 and thereby commits a cell to undergo apoptosis. 2 Highlyspecific proteases are useful for signaling pathways because a single or small number of effector
substrates can be targeted, but specific proteases are poorly suited for general degradation of
many different proteins, as the armada of specific-proteases would be enormous. A second
strategy for intracellular degradation is to selectively sequester proteins within a degradation
11
chamber. Whereas entrance to the degradation chamber is carefully regulated, once within the
chamber, a protein substrate is exposed to non-specific proteolytic sites and is degraded to small
peptides. Sequestered degradation is highly adaptable to different substrates and accounts for
more than 90% of the protein turnover inside the cell.3 The E. coli AAA+ ClpXP protease
exemplifies sequestered degradation and consists of the barrel-shaped ClpP peptidase capped by
a AAA+ ClpX ATPase (Fig. 1.1). The proteolytic chamber of ClpP can be accessed by a narrow
channel on either end of the barrel. ClpX binds proteins with specific degradation tags and
unfolds and translocates these substrates into the chamber of ClpP, which requires the energy of
adenosine triphosphate (ATP) binding and hydrolysis. The narrow entryway of ClpP excludes
off-target proteins, but necessitates that a target substrate be unfolded before being threaded into
the proteolytic chamber.4 This raises several questions. How easily are proteins unfolded within
the native environment of a cell? Are some proteins more difficult to unfold than others, and, if
so, why? How do AAA+ ATPases convert the chemical energy of ATP binding and hydrolysis
into mechanical unfolding and translocation?
lkd
translocase
peptidase
3tag
rP
free substrate
and enzyme
tag-dependent
recognition
AP
1
lon
tranelcalon
degradation
Figure 1.1 CIpXP, AAA+ Protease
Recognition occurs when a degradation tag (brown) binds the central pore of the AAA+ ClpX unfoldase and translocase (dark
purple). Then, by multiple rounds of ATP binding and hydrolysis, the substrate is unfolded and translocated into ClpP, a selfcompartmentalized peptidase (light purple). Adapted from Baker, T.A. & Sauer, R.T. (2012).
12
AAA+ Proteases
AAA+ proteases are present in all branches of life and can be divided into families based on
sequence homology.5
The 26S proteasome and CDC48-20S families are present in the
eukaryotic cytosol, and the ClpXP, ClpAP, HslUV, FtsH, and Lon family proteases can be found
both in the eubacterial cytosol and eukaryotic organelles. Archaebacteria contain the CDC4820S and PAN-20 families, as well as membrane-bound Lon. Members of different protease
families often recognize different degradation tags or determinants and can also vary in their
ability to degrade model substrates with different mechanical or thermodynamic stabilities. 6
Nevertheless, all AAA+ proteases share a similar architecture: a barrel-shaped peptidase capped
by a ring-shaped AAA+ ATPase. AAA+ enzymes (ATPases associated with various cellular
activities) are characterized by large and small AAA+ sub domains, typically form ring hexamers
or filaments, and couple ATP hydrolysis to mechanical work, including protein disaggregation
and degradation, microtubule transport, and DNA translocation.
The reaction cycle of a AAA+ protease is best understood for degradation of ssrA-tagged
proteins by E. coli ClpXP (Fig 1.1).
The ssrA tag (AANDENYALAA)
is added to the C
terminus of nascent proteins when translation stalls and is recognized by loops within the axial
pore of the ClpX ring. 7 Rounds of ATP binding and hydrolysis result in translocation of the tag
through the pore. If the tag is attached to a folded domain, the collision of the domain being
pulled against the narrow axial pore generates an unfolding force. Many pulling events can be
required before the domain is unfolded, allowing translocation of the unfolded polypeptide into
ClpP for degradation.8
ClpXP can recognize degradation tags other than the ssrA tag, including
13
the C-terminal kO degron.9 Other AAA+ proteases have similar mechanisms: bind a polypeptide
tag and translocate sequentially from the tag to distal terminus. For instance, Lon can recognize
~20-residue degradation tags attached to either the N or C terminus of a protein, and translocates
towards the other terminus.' 0
Recently, the reaction cycle of ClpXP was observed using single-molecule nanometry. AubinTam et al. used optical trapping nanometry to monitor ClpXP-mediated unfolding and
translocation of filamin, a multi-domain model substrate (Fig 1.2)."
First, a filamin-ssrA was
attached to a polystyrene bead and subsequently trapped in a strong-force clamp. Next, a
ClpXP-labeled polystyrene bead, in a weak-force clamp, was brought in close proximity to the
filamin-ssrA bead to allow ClpXP to engage the substrate.
Following engagement, ATP-
dependent changes in the bead-bead distance were observed as ClpXP unfolded and then
translocated each domain of filamin. The processing of each domain involved a dwell of no
movement, followed by a sudden increase in bead-bead distance caused by unfolding, followed
by a decrease in distance caused by translocation (Fig. 1.2 gray inset). The dwell results when
ClpX encounters a folded domain but a single cycle of ATP hydrolysis fails to unfold the protein.
In this case, the process is repeated until, by chance, a power stroke coincides with transient
destabilization of the substrate, permitting successful unfolding. While attempting to unfold and
translocate some domains, ClpX may hydrolyze 100s of ATPs. Once unfolding occurs, the
tension between the beads causes an increase in distance proportional to the length of the
unfolded polypeptide. Following unfolding, ClpX translocates the unfolded polypeptide until
another domain is reached, shortening the distance between the beads. Close inspection of the
14
translocation phase reveals that ClpX translocates the unfolded polypeptide in discrete steps of
5-8 amino acids with a frequency expected for hydrolysis of one ATP per step (Fig. 1.2 white
inset).
DNA tether
bead-to-beaddistance
r
I
-T
goo
I
Figure 1.2 Single-molecule nanometry.
Unfolding and translocation of multi-domain filamin by a single ClpXP protease. Filamin-ssrA is attached to a bead by a
protein-DNA linkage and trapped in a laser beam. ClpXP is attached to a second bead by a biotin-streptavidin linkage and held
under tension with a weak trap. (Gray inset) Following engagement of the substrate by ClpXP, the distance between the beads
changes as ClpXP unfolds and translocates individual domains of filamin. Unfolding increases bead-bead distance, whereas
translocations decrease bead-bead distance. (White inset) Translocation of an unfolded Filamin domain occurs in discrete steps.
Adapted from Aubin-Tam, M.E., et al. (2011).
As seen above, ClpXP unfolding of very stable domains can be slow and energetically costly. In
solution experiments, one can determine the average time spent and ATP used for unfolding and
translocation by comparing the rates of degradation of native versus chemically-denatured
15
substrates. For instance, titin-127-ssrA, a model native substrate with an immunoglobulin fold, is
degraded at a ~15-fold slower maximal rate and requires hydrolysis of ~500 more molecules of
ATP than the same substrate unfolded by carboxymethylation of cysteines in the hydrophobic
core (Table 1.1, Fig. 1.3).
Thus, protein unfolding is the rate-determing step in substrate
degradation. The structural features of a domain that may allow a substrate to resist unfolding
are discussed in the next section.
t~ ylef
Sn
t.
A
substrate
'fo
Isa
W
Figure1.3
nstratem
unfodases
Unoldin
by AA+
con\ msktWo
Figure 1.3 Unfolding by AAA+ unfoldases.
Successful unfolding is often a low probability event. First, a folded substrate is engaged by the unfoldase (light blue). Then, a
cycle of ATP hydrolysis allows the unfoldase to tug on the substrate, creating a strained state in which force is applied on the
substrate (dark blue). At this point, the substrate either slips, is unfolded, or is released from the enzyme. The likelihood of
unfolding depends on the structure of the substrate. Following unfolding, the unfolded polypeptide is translocated by additional
cycles of ATP hydrolysis. Adapted from Sauer, R.T., et al. (2004).12
Protein-unfolding in a native environment
Is the equilibrium or kinetic stability of a protein responsible for its ability to resist unfolding
when a AAA+ protease tugs at its terminus, or is the stability of local structural elements near the
gripped terminus more important for resisting unfolding? These questions have been investigated
for ClpXP-mediated unfolding. How frequently a native protein samples the denatured state
16
(kinetic stability) is determined by the unfolding energy barrier (Fig. 1.4). For example, P22 Arc
repressor has a low barrier and unfolds ~8 times per minute, whereas green fluorescent protein
(GFP) has a much higher barrier with a half-life for spontaneous unfolding of -20 years.' 3 "4 If
ClpXP simply trapped a spontaneously unfolded state, then it would degrade Arc-ssrA rapidly
but require years to degrade GFP-ssrA. Instead, ClpXP degrades both substrates at similar rates
(1.8 min~' enz~1 for Arc-ssrA; 1.2 min~' enz~1 for GFP-ssrA).15 Thus, ClpXP accelerates the rate
of GFP unfolding by ~107, representing an ~10 kcal mol-1 decrease in the energy of the transition
state and a rate enhancement similar to that observed in a strong denaturant like 7 M GndHCl.
These results indicate that ClpXP actively denatures GFP-ssrA, rather than passively capturing
spontaneously unfolded protein.
*
0
C
Kinetic
w
Stability (k)
0
Unfolded
Global
Stability
(AGe,KU)
Folded
Figure 1.4 Kinetic and global stability.
Gibbs free-energy diagram for the two-state folding of a protein. Kinetic stability (ku; red) is determined by the energy difference
between the folded state and the transition state. The global equilibrium stability (AGu, Ku; purple) is determined by the energy
difference between the folded and unfolded states. f denotes the transition state.
17
ssrA-tagged
substrate
Arc
GFP
titin-127
carboxymethylatedtitin-127
titin-127-V4A
titin-127-V9P
titin-127-V11P
titin-127-V13P
titin-127-V15P
His 6 -RNase-H*
degradation rate
(min~' enz'1)
1.8
1.2
0.25
3.7
AGu
(kcal mol-1)
1.3
4.6
6.4
unfolded
0.36
1.5
2.9
3.1
0.85
4.2
4.4
4.5
3.5
2.9
4.6
12
Ku
1.2*10~'
5.7*10-4
3.1*10-5
7.9*10-4
6.7*10-4
3.4*103
9*10-3
5.7*10-4
3.4*10'
Table 1.1 Degradation of ClpXP substrates with different stabilities.
Arc data is from Burton et al. (2001); GFP data is from Nager et al. (2011); titin-127 data is from Kenniston et al. (2003);
RNaseH* data is from Kenniston et al. (2004).
The equilibrium stability of a protein is determined by the energy difference between the folded
and unfolded states (Fig. 1.4), which in turn is a function of the favorable and unfavorable
enthalpic and entropic interactions in both states. Does the equilibrium stability of an ssrAtagged protein dictate its resistance to ClpXP disruption of the structure?
Kenniston et al.
compared the degradation rates of titin-127-ssrA variants with different global stabilities (Table
1.1; Fig. 1.6A).
Although there was a reciprocal trend between global stability and ClpXP
degradation rates, a greater than 3000-fold change in Ku resulted in only a -12 fold change in the
rate of degradation. Likewise, there was no obvious correlation (Fig. 1.5) when degradation
rates and global stabilities were compared between different substrate families (Arc, GFP, titin127, RNase-H*). For example, the global stability of T thermophilus RNase-H*-ssrA (12 kcal
mol-1) was almost twice as high as titin-127-ssrA (6.4 kcal mol-1), but the more-stable protein was
degraded almost 15 fold faster than the less-stable protein.16 These results show that the
equilibrium stability of a protein substrate is a poor predictor of the rate of ClpXP-catalyzed
18
unfolding and degradation.
increasing stability
4-
g
2
-
y
01
AG. (kcal/mole)
Figure 1.5 Relationship between global stability and the rate of ClpXP degradation.
There was no strong correlation between the maximal rates of degradation of different ssrA-tagged substrates by ClpXP and the
equilibrium stabilities of these substrates. Titin-127 variants are shown as circles, Arc variants are shown as squares, GFP is a
triangle, and RNase-H* is a diamond. Adapted from Kenniston et al. (2004).
If the global kinetic and equilibrium stabilities of substrates correlate poorly with ClpXP
unfolding and degradation, then the stability of local structural elements near the degradation tag
may determine resistance to ClpXP. 17 In support of this idea, titin-127-ssrA variants with
destabilizing mutations near the ssrA tag (Y9P, V15P) were degraded 3-6 fold faster than a
variant with a destabilizing mutation distal from the tag (V4A; Table 1; Fig. 1.6A).
To
investigate the effects of local structure in a different protein, Kenniston et al. generated singlecysteine RNase-H* L114C, R140C, or A166C variants and crosslinked them to a cysteinereactive (sulfo-MBS)-ssrA peptide (Table 1.2, Fig. 1.6B). When the ssrA peptide was crosslinked
to the C-terminal residue (A166C), the resulting substrate was degraded at the same rate as
RNase-H*-ssrA. Importantly, when the ssrA tag was crosslinked to L114C or R140C, ClpXP
degraded these substrates much more slowly, even though these proteins had global stabilities
similar to the rapidly degraded substrates (Table 1.2).
One might argue that slow ClpXP
degradation of the ssrA-tagged L144C and R140C constructs occurs because ClpXP must
19
simultaneously translocate two polypeptide chains to unfold and translocate these substrates.
However, experiments with disulfide-crosslinked substrates show that ClpXP can readily
translocate two polypeptide chains (Burton, R.E, et al. 2001). 18 Thus, it appears that the protein
structure near the ssrA tag can greatly influence the rate of degradation by ClpXP.
substrate
degradation rate
Tm ("C)
(min-' enz-1)
RNaseH*-ssrA
RNaseH* Ll14C-ssrA
RNaseH* R140C-ssrA
RNaseH* A166C-ssrA
4.2
<0.1
0.2
4.2
82
82
81
83
Table 1.2 ClpXP degradation of substrates with ssrA tags located at different positions in the structure.
Data are from Kenniston et al. (2004).
A
Titin 27
B
RNase-H*
Figure 1.6 Structures of substrates used for CIpXP-degradation studies.
(A) Structure of titin-127 (PDB: 1TIT).' 9 Amino acids that were mutated in destabilized variants are marked with numbered
circles. Destabilizing mutations near the C-terminal ssrA tag had a greater effect on the rate of degradation. (B) Structure of T.
thermophilus RNase-H* (PDB: IRIL). 0 Amino acids that were mutated to cysteine and reacted with a cysteine-reactive ssrA tag
are marked with numbered ovals. When an ssrA tag was crosslinked to residue 114, 140, or 166, ClpXP must attempt to denature
RNase-H* by pulling on different parts of the structure.
Taken together, these results suggest that the global kinetic or equilibrium stability of a substrate
has a relatively small influence on the rate of degradation by ClpXP, whereas the stability of the
structure near the site of enzyme-mediated pulling can have a larger impact. In chapter 2, I
examine how local structure near the ssrA tag affects ClpXP degradation of the p-barrel structure
20
of GFP-ssrA and circularly permuted variants of this protein. I found that the initial step in GFPssrA unfolding by ClpXP is extraction of the C-terminal P strand. This event does not result in
cooperative unfolding of the rest of the
P barrel,
which ClpXP must denature in a second step.
Interestingly, this second step fails at low ATP concentrations. There appears to be a decisive
moment when ClpXP must either denature the folded intermediate or release the extracted
strand, which permits refolding and negates the work done. The probability of either outcome
depends on the engine speed of ClpXP, which is determined by the ATP concentration, the
stability of the intermediate, and the rate at which the local structure refolds.
Structures of AAA+ Machines
AAA+ machines actively unfold substrates by converting the energy of ATP binding and
hydrolysis into conformational changes that are used to perform mechanical work. X-ray crystal
structures
of AAA+
machines, including ClpX,
are available
and
suggest potential
conformational changes but whether these structural motions are functionally relevant and how
they relate to ATP binding and hydrolysis is unknown.
The AAA+ module of a ClpX subunit consists of a large domain, a short hinge, and a small
domain (Fig. 1.7A). N-terminal of the AAA+ module is a family-specific Zn2+-binding domain
involved in the recognition of some substrates, but this domain is dispensable for the degradation
of ssrA-tagged substrates.2 1
The hinged interface between the domains forms a potential
nucleotide-binding site, but the domain-domain orientation varies greatly between individual
21
subunits in a homohexamer and creates two general categories of subunits (Fig. 1.7BC).2
ATP-
loadable (L) subunits can bind nucleotide. By contrast, an 80-90 hinge rotation in ATPunloadable (U) subunits destroys the nucleotide-binding pocket, in part by placing an a helix
where ATP/ADP would normally bind (Fig. 1.7CD). In several different crystal structures of
ClpX, including ones with no nucleotide, subunits within the hexamer were arranged in a L-L-UL-L-U order around the ring (Fig. 1.6B). Upon soaking ATP, ATPyS, or ADP into nucleotide-free
crystals, small conformational changes were observed in both classes of subunits as a
consequence of nucleotide binding to the L subunits.
A
iie
arg
~
C
135 A
F
C
Iinuclootide
J!
smal
domnains
Loadable
82*
Uratoadable
cae~n~k
Loadable
Unloadable
Figure 1.7 Asymmetric structure of ClpX.
(A) Diagram of the two-domain AAA+ module in a ClpXAN subunit. (B) X-ray crystal structure of a ClpX hexamer (PDB:
3HTE). Subunits are marked A-F and differently colored. A schematic of the crystal structure is shown on the right. ATPloadable (L) and ATP-unloadable (U) subunits are indicated. For an ATP-soaked structure, nucleotide (nuc) bound to the four L
subunits. (C) Comparison of domain rotation in L and U subunits. The large domain is shown in gray and the small domain in
green. (D) Effect of domain rotation on the hinge. For U subunits, movement of the hinge obstructs the nucleotide-binding
pocket (PDB: 3HWS). Adapted from Glynn et al. (2009).
22
In addition to the variably hinged domain-domain interface within subunits, crystal structures
revealed a conserved "rigid-body" interface between the large and small domains of adjacent
subunits (Fig. 1.8). To test if this "rigid-body" interface remained intact during ClpX function
and substrate processing, Glynn et al. designed disulfide bonds to crosslink the subunit-subunit
interface. Despite placing two crosslinks across each of the six interfaces in a hexamer, the ClpX
enzyme was still able to unfold and degrade GFP-ssrA in combination with ClpP.2 2 In contrast,
minor mutations in the hinge greatly reduced substrate processing.
These results support a
model in which motions at the hinged interface, rather than the subunit interface, are involved in
driving substrate unfolding and translocation.
major subunit interface
large
domain
small
domain
Figure 1.8 Rigid-body subunit interface.
Comparison of the major subunit interface of subunits within a nucleotide-bound ClpX hexamer (PDB: 3HWS). Alignment of
the large AAA+ domains of each subunit (gray) resulted in alignment of the small domain of the counterclockwise subunit.
Similar results were observed for subunit-subunit interactions in other nucleotide-bound and nucleotide-free hexamers. Adapted
from Glynn et al. (2009).
The axial pore of a ClpX hexamer physically binds some degradation tags and translocates
polypeptides into ClpP for degradation.2 3 The pore is lined with three classes of loops - the
RKH, GYVG, and pore-2 loops - contributed by the large domain of each subunit (Fig. 1.9).24-26
Mutations in these loops confirm their importance in substrate binding and translocation.2 7 The
tyrosine of the GYVG loop appears to be especially important for substrate translocation, as
substitution of this residue with alanine in just two subunits of a hexamer results in polypeptide
23
slippage and highly costly or failed degradation of substrates. 2 8 This tyrosine may act as a
paddle that nonspecifically pushes or pulls polypeptide sequences through the pore. 29 Based
upon the crystal structures, one can model conformational changes at the hinged interface that
would propagate via the rigid-body interfaces around the ClpX ring and move specific GYVG
pore loops towards ClpP along the axial pore. In one set of models, subunits are allowed to flip
between U and L conformations, resulting in 90 rotation of the hinge and movement of the axial
pore loops. In a second set of models, the identities of U and L subunits remain fixed, but
smaller hinge movements generate a power stroke by cycling between different L conformations.
For instance, comparing L conformations in nucleotide-free and nucleotide-bound structures
shows that nucleotide binding causes a flexing of the hinged interface and consequentially
pushes the pore loops downward.
Although these models suggest possible conformational
changes, it remains to be determined if crystal structures represent active conformations and if
modeled conformational changes occur. For instance, do U and L subunits interconvert during
function? How are nucleotide binding and hydrolysis coupled to these events?
RKH loops
pore-2 loops
Figure 1.9 ClpX pore loops.
The RKH, GYVG, and pore-2 loops of ClpX help to bind degradation tags and translocate substrates. The RKH loops are
positively charged and are important for discriminating between different classes of degradation tags. The GYVG loops are
critical for the translocation of substrates. The pore-2 loops are involved both in substrate binding, translocation, and ClpX
interactions with ClpP. Adapted from Baker & Sauer (2012).
24
Is an L-L-U-L-L-U arrangement of subunits observed in other AAA+ machines? Dynein, a
microtubule motor protein, has six distinct AAA+ domains connected in a single polypeptide
chain; in crystal structures, dynein adopts a dimer-of-trimers arrangement (L-L-U-L-L-U) similar
to ClpX or a low symmetry L-L-L-U-L-U architecture.3
Some crystal structures of HslU, the
AAA+ unfoldase of the HslUV protease, show an L-L-L-L-L-L pattern of subunits.3 2 A cryo-EM
reconstruction of Rpti 6 , the hetero-hexameric AAA+ unfoldase of the 26S proteasome, shows a
hexamer with 5L subunits and lU subunit.33
Similarly, the AAA+ El helicase forms a
homohexamer with five subunits in contracted L-like conformations and one U-like subunit in an
extended-gap conformation.
In chapters 3 and 4, I describe ensemble and single-molecule
experiments that suggest that nucleotide binding to ClpX changes its structure from a 4L:2U
arrangement to a 5L:lU architecture.
Additionally, in chapter 4, I use single-molecule
fluorescence experiments to observe conformational switch between different classes of ClpX
subunits.
Subunit coordination within ATP-fueled motors
Protein motors work by changing conformation in response to nucleotide occupancy. The tight
coupling between mechanical and chemical steps is critical for efficient work. However, this
concept is less clear for multimeric motors. Although the activities of subunits in the Fi ATPase,
kinesin, and myosin V motors are sequentially coordinated, the movement of cytoplasmic dynein
along
microtubules occurs
in a stochastic, uncoordinated
stumble,
suggesting weak
mechanochemical coupling.34-37 Nevertheless, all of these motors can generate processive,
directional work. To better understand subunit coordination (or lack thereof), I will review the
25
mechanisms of Fi ATPase and cytoplasmic dynein. Next, I will outline arguments for subunit
coordination within hexameric AAA+ rings. Later, in chapters 3 and 4, I will investigate
coordination between AAA+ modules in ClpX hexamers, and show that (i) there are distinct
subunit-classes based on hexamer geometry; and (ii) the mechanical cycle of ClpX consists of
both coordinated and uncoordinated elements.
The F1 ATPase: a sequential rotary engine
The F1 ATPase converts the energy of ATP binding, hydrolysis, and product release to rotation of
a central camshaft.
The minimal complex (
subunits encircling an asymmetric y-stalk.
3
p37)
consists of three nucleotide-binding as
When nucleotide binds and is subsequently
hydrolyzed and released by designated as subunits, changing interactions between the as
subunits and y-stalk cause the stalk to rotate. The scheme outlined in Fig. 1.10 has subunits in a
a43y complex designated as ap', ap2, and ap 3, with ATP initially bound only to ap' (state A).
Upon binding ATP to ap 2, the y-stalk rotates 90 counterclockwise (state B). At this point, the
ATP bound to ap is hydrolyzed (state C) and released, rotating the y-stalk 30 counterclockwise
and resetting the cycle (state A'). Each chemical step is gated by conformational changes of the
stalk and vice versa, ensuring tight mechanochemical coupling and counterclockwise rotation.
38 39
Indeed, if one conformation is locked by a disulfide bond, the chemical cycle is inhibited. ,
26
(A)
(B)
UPIK
(C)
D
APATP
binding
hydrolysis
release
T*
TP
TP
2
(A')
aPa
TN
D
Figure 1.10 Sequential nucleotide cycle of F1 ATPase.
The F, ATPase (a3 p3y) rotates counterclockwise with ATP binding and product release. In state A, ap' is bound to ATP. ATP
binds to ap 2, causing the asymmetric y-stalk to rotate 90' counterclockwise (state B). Then, the ATP in ap3 is hydrolyzed (state
C) and ADP and phosphate are released, causing a 30' rotation and restarting the cycle (state A'). There are two potential points
in the cycle where a backwards step could occur: (1) in state A, if ATP bound to ap 3 instead of ap2, or (2) in state B, if the ATP
bound to ap 2 was hydrolyzed instead of ap1 , then this would lead to release of nucleotide from ap 2 and a back step. Either backstep scenario could only occur if two unbound or bound subunits were functionally equivalent. This situation does not occur
because the asymmetric y stalk controls the conformations of each ap subunit so that no two as subunits are equivalent.
In the absence of a y stalk, as subunits, either as a as monomer or a3p 3 , adopt the nucleotideempty conformation in which the nucleotide-binding site is distorted. 40' 4 1 However, when
encircling the y stalk, as subunits assume unique conformations based on their interactions with
the asymmetric stalk: ATP bound, ADP bound, empty (Fig. 1.11).42 Because ap conformations
are dictated by interactions with the asymmetric y stalk, a 1:1:1 ratio of as classes is maintained
and the conformations of different as subunits are allosterically connected. Moreover, because
the conformation of the as nucleotide-binding pocket reflects interactions with the y stalk,
chemical
steps
and
mechanical
output
are
coupled.
Tight
coupling
between
the
mechanochemical steps of different subunits ensures that the motor works in a sequential,
counterclockwise rotation. For the scheme in Fig. 1.10, two opportunities when subunit coupling
prevents the rotor from backward steps are highlighted. A backwards step could occur at either
state A, if ATP bound to the empty Cp3 instead of the empty ap', or state B, if ATP was
hydrolyzed by ATP-bound ap2 instead of ATP-bound ap
.44
Both scenarios are avoided as the y
stalk imposes asymmetry at all times, energetically forcing subunits into unique roles or classes.
27
fi-AnF
a fl-
Figure 1.11 Asymmetric interactions of the y stalk.
Side and top views of the F1 ATPase (asp3y) with 120' rotations (PDB: 1E79).4 The y stalk (orange) makes asymmetric
interactions with each p subunit (green). The top of the y stalk makes numerous contacts with the ATP-bound and ADP-bound
subunits, but not the empty subunit, leading to deformation of the nucleotide-binding site of pempty. The slope of the y stalk
causes there to be different contacts along the length of pATP and OADP. Adapted from Kinosita et al. (2004).
Strict subunit classes and coupling prevents wasteful backward steps. Consider that an ATPhydrolyzing protein can adopt at least five states: empty, ATP bound, ADP+Pi bound, ADP
bound, and Pi bound. For an ap subunit of the F1 ATPase, each state either pulls, pushes, or
doesn't contact the y shaft. If subunits were uncoordinated, then the trimer could assume at least
53 or 125 different states, each exerting different forces on the y shaft and few resulting in
rotation. Strict subunit coordination insures that trimer states are populated in the correct order,
and thus that essentially every bound and hydrolyzed ATP molecule results in clockwise rotation.
A strictly sequential mechanism, in which a motor proceeds through a predetermined series of
states, results in stalling if a component of the motor fails. For example, the F1 ATPase normally
takes ~400 120" rotations per second, but occasionally pauses for longer than 60 seconds. 4 5
Pauses occur when solution ADP-Mg
non-productively binds a catalytic site, jamming the rotor
at a waiting-for-dissociation 90 step.
The motor resets only after slow off-pathway ATP
28
association to an empty site. 4,47 Similarly, the F1-ATPase can be jammed by single-subunit
mutations, inhibitors, and conformational events. 4 8-51 For motors that handle more diverse tasks
than the F1 ATPase or translocate heterogeneous tracks, such as AAA+ unfoldases translocating
denatured polypeptides, a strictly sequential mechanism may lead to frequent motor failure.
Cytoplasmic Dynein walkers: a stochastic stumble
A dimeric dynein walker is a 1.2 MDa complex that uses ATP to power minus-end directed
microtubule transport. A dynein subunit consists of a microtubule-binding domain, a AAA+
ring, and a N-terminal dimerization domain (Fig. 1.12). Although it is not known how the
AAA+ modules within a dynein ring are coordinated, a single AAA+ module, called AAAl, is
responsible for most ATP hydrolysis and is critical for stepping of the microtubule-binding
domain. For this section, movement of dynein will be considered in terms of the coordination
between microtubule-binding domains within a dimer.
Dimerization
Domain
AAAI
AAA+ ring
MicrotubuleBinding
Domain
29
AAAI
Figure 1.12 Cytoplasmic Dynein
Each subunit of a dynein dimer
domain. The microtubule-binding
AAA+ modules within the AAA+
the dimerization domain.
walkers
consists of three domains: a microtubule-binding domain, AAA+ ring, and a dimerization
domain and AAA+ ring of different subunits are distinguished by color (yellow vs. blue). The
ring are designated AAA1 through AAA6. An extended linker connects the AAA1 module to
DeWitt et al. labeled the microtubule-binding domains of a dimeric dynein walker with different
colored quantum dots and observed how the domains stepped relative to each other. 37 Although
most steps alternated between subunits, -30% of steps were non alternating, with one subunit
taking multiple consecutive steps before the other subunit moved. Step size varied from 0-50 nm
with the trailing domain tending to take larger steps as it was pulled forward by intramolecular
tension. This stochastic stepping behavior suggests that there is little coordination between the
chemical cycles of dynein subunits. Indeed, there is no structural evidence that AAA1 modules
in different subunits communicate. Subunit coordination does not appear to occur through the
N-terminal dimerization domain, as this domain can be replaced with an artificial dimerization
domain without altering stepping behavior.52 Moreover, the AAA+ rings likely do not physically
touch as the microtubule-binding domains separate by as much as 50 nm. Taken together, these
results suggest the mechanochemical cycles of dynein subunits are not tightly coupled,
permitting the motor to walk in a stochastic and uncoordinated manner.
Although the lack of subunit coordination may decrease the efficiency of a dynein walker,
stochastic stepping increases the variability of step sizes and resilience to failure. In fact, dynein
frequently back steps or side steps, wasting energy. A stochastic mechanism can prevail despite
component failure. De Witt et al. constructed a heterodimeric dynein with one functional and
one inactivated AAA1
domain and observed that the functional subunit stepped along
microtubules, dragging the inactive subunit along. 37
30
Thus, dynein translocates along
microtubules in a highly flexible manner which may be beneficial when multiple dynein motors
cooperate with kinesin motors to transport cellular cargo. 53 Flexibility would be valuable for a
AAA+ unfoldase, which must deal with proteins that have very different unfolding barriers and
also encounter highly variable protein sequences during translocation.
AAA+ unfoldases have multiple classes of subunits
For the ClpX, HslU, and PAN unfoldases, there is evidence for three classes of nucleotide
binding sites: empty, weak, and tight.54 -56 For a ClpX variant in which a Walker-B mutation
essentially eliminates ATP hydrolysis, stoichiometry experiments showed that at least two
subunits did not bind nucleotide. Additionally, dissociation of bound nucleotide was biphasic,
reflecting tight- and weak-binding sites. Multiple classes of subunits within homomeric AAA+
unfoldases suggest that subunits coordinate during the slow step of the reaction cycle, ATPhydrolysis. In chapter 3, I use nucleotide-binding mutations in ClpX to differentiate three classes
of subunits and show that nucleotide binding to the tight and weak sites has different effects on
the conformation of the hexamer. Importantly, ATP hydrolysis and substrate processing only
occurs when U and L subunit classes can switch.
AAA+ motors: stochastic, concerted, or sequential?
Several attempts have been made to determine if subunits of AAA+ motors hydrolyze ATP in a
stochastic, concerted, or sequential order (Fig. 1.13). For three AAA+ machines - SV40 large T
antigen (LTag), papillomavirus El, and E. coli ClpX - different models for subunit coordination
31
have been proposed.
a
Concerted hydrolysis
C Probabilistic hydrolysis
b Sequential hydrolysis
0
ATP hydrolysis
Figure 1.13 Models for coordination of nucleotide hydrolysis within AAA+ hexamers.
ATP hydrolysis in hexamers could be coordinated in concerted (A), strictly sequential (B), or probabilistic/stochastic (C) manner.
For any model, variations where a fewer number of subunits participate are possible. For instance, if two subunits remain
nucleotide free, then the remaining four subunits could hydrolyze ATP in a concerted, sequential, or stochastic manner. Adapted
from Martin et al. (2005).
LTag and El are homohexameric AAA+ helicases, which have been proposed to have concerted
or sequential models based on crystal structures.
However, in both cases, subsequent
biochemical experiments have not validated these models. LTag crystallized in several states in
which an asymmetric hexamer was bound to no nucleotide, 6 ATPs, or 6 ADPs. 57 Because
nucleotide binding in the crystal structure was all-or-none, the authors proposed that all six
subunits act in concert, cooperatively binding, hydrolyzing, and releasing nucleotide.
A
crystallographic argument was also posed for the El helicase. El helicase bound to singlestranded DNA crystallized with 5 ADP-bound subunits forming a spiral staircase, with each
clockwise subunit slightly offset. 58 The staircase was followed by an empty, extended subunit
that closed the ring. As three of the subunits bound a chloride ion at the position expected for the
y phosphate of ATP, three ADP-bound subunits were assigned as "ATP-bound" subunits. This
assignment resulted in a clockwise arrangement of ATP-ATP-ATP-ADP-ADP-empty subunits,
32
which the authors argued was evidence for a sequential mechanism similar to F1 ATPase.
However, an identical El structure was solved without bound nucleotide or substrate, suggesting
that the staircase state may be an energy minima and not strictly coupled to chemical steps. 59 At
present, an enormous variety of AAA+ crystal structures have been solved with few conserved
features. This situation suggests either that these hexameric machines utilize many different
configurations during a reaction cycle or that the flexibility of these machines leads to
crystallization artifacts.
In either case, biophysical experiments that observe conformational
changes during a reaction cycle are necessary to validate crystal structures and determine
dynamic mechanisms.
To explore subunit coordination within ClpX hexamers, Martin et al. devised a strategy to place
ATPase mutations in specific subunits of pseudo hexamers in which AAA+ modules were
covalently linked by 20 amino-acid tethers.60 Importantly, there was no arrangement of ATPase
active and inactive modules that stalled ATP hydrolysis or substrate degradation. If ClpX
subunits must sequentially hydrolyze ATP, then even a single inactive subunit should stall the
motor, but this was not observed. Furthermore, a hexamer with a single ATPase-active subunit
supported substrate degradation, albeit slowly. If a particular arrangement of subunits had to
hydrolyze ATP in a concerted fashion, then a hexamer with a single ATPase-competent subunit
would be expected to be inactive.
The flexibility of the ClpX-reaction cycle suggests that
subunits hydrolyze ATP stochastically or that there is a probabilistic step within a coordinated
reaction cycle.
33
A purely stochastic model in which ClpX subunits are entirely uncoordinated is unlikely. For
example, the detection of subunit classes in ensemble measurements indicates coordination for
the slow step of the reaction cycle. Secondly, single-molecule nanometry of ClpX observes a
somewhat regular step size with relatively few back steps, consistent with subunit coordination.
It is possible that both coordinated and uncoordinated components of the reaction cycle are
possible and occur.
However, to my knowledge, no well-studied examples of similar
mechanisms exist. Here are two theoretical possibilities:
(1) A probabilistic step within a coordinated reaction cycle. At one point in the reaction
cycle, multiple subunits adopt equivalent conformations. When this occurs, each subunit
has an equal probability of proceeding to the next step. Once a subunit proceeds, all
other subunits undergo coordinated conformational changes, forming subunit classes for
the remainder of the reaction cycle (Figure 1.14; Model #1). This is similar to the MWC
allosteric model in which all tense subunits are equivalent but ligand binding can stabilize
the relaxed conformation of subunits.i
(2) Simultaneous coordinated and uncoordinated cycles. In this model, the motor performs
work by a coordinated cycle similar to classically studied motors. However, a second
cycle, which is not tightly coupled to the chemical cycle, frequently resets the motor
(Figure 1.14; Model #2).
Take the example of ADP-Mg inhibition of F1 ATPase.
Inhibition occurs because, given a current assignment of subunit classes, ADP binds to
the wrong site.
Hypothetically, if a slow conformational isomerization changed the
assignment of subunit classes, the motor could resume work.
Example taken from Baker & Sauer (2012).
34
Model #1
Model #2
A probabilistic step
Two cycles
AA
A' B'
B'A
B'
BA
AB
A2
BC
CB
A B'
AA
AB'
Figure 1.14 Models for mixed subunit coordination
Models for how a dimer could have steps with and without subunit coordination. At each step, subunit classes within the dimer
are represented by letters. Conformational changes within a subunit classes are represented by superscript numbers. For Model
#1, there is a step of the reaction cycle were subunits are equivalent and, depending on chance, the reaction cycle follows one of
two mirror paths. For Model #2, two cycles occur simultaneously. One cycle retains subunit classes and proceeds through a
sequence of conformational changes. A second cycle resets the motor with new subunit classes.
These models hedge efficiency and adaptability. During coordinated steps, subunit classes are
assumed and highly efficient work is possible.
During uncoordinated steps, efficiency is
sacrificed for adaptability. In chapter 4, I investigate the conformational changes within a single
AAA+ module using single-molecule fluorescence techniques. What emerges is two mechanical
cycles, supported by the second model. In one cycle, highly-coupled subunit classes undergo
fast conformational changes tied to the chemical cycle. In a second cycle, slow conformational
changes, uncoupled from ATP-hydrolysis, isomerize the ring. Ensemble experiments suggest
that both cycles are critical for function.
35
References
1. Dobson, C.M. (2003) Protein folding and misfolding. Nature 426, 884-90.
2. Fernandes-Alnemri, T., Takahashi, A., Armstrong, R., Krebs, J., Fritz, L., Tomaselli, K.J.,
Wang, L., Yu, Z., Groce, C.M., Salveson, G., Earnshaw, W.C., Litwack, G., & Alnemri, E.S.
(1995) Mch3, a novel human apoptotic cysteine protease highly related to CPP32. Cancer Res.
55, 6045-52.
3. Gottesman, S. (2003). Proteolysis in bacterial regulatory circuits. Annu. Rev. Cell Dev. Biol.
19, 565-87.
4.
Sauer, R.T. & Baker, T.A. (2011) AAA+ proteases: ATP-fueled machines of protein
destruction. Annu. Rev. Biochem. 80, 587-612.
5. Snider, J. & Houry, W.A. (2008) AAA+ proteins: diversity in function, similarity in structure.
Biochem. Soc. Trans. 36, 72-7.
6. Gottesman, S., Roche, E., Zhou, Y., & Sauer, R.T. (1998) The ClpXP and CipAP proteases
degrade proteins with carboxy-terminal peptide tails added by the SsrA-tagging system. Genes
Dev. 12, 1338-47.
7. Baker, T.A., & Sauer, R.T. (2012) ClpXP, an ATP-powered unfolding and protein-degradation
machine. Biochim. Biophys. Acta. 1823, 15-28.
8. Flynn, J.M., Neher, S.B., Kim, Y.I., Sauer, R.T., & Baker, T.A. (2003). Proteomic discovery of
cellular substrates of the ClpXP protease reveals five classes of ClpX-recognition signals. Mol.
Cell 11, 671-83.
36
9. Gur, E., Vishkautzan, M., & Sauer, R.T. (2012). Protein Unfolding and degradation by the
AAA+ Lon protease. ProteinSci. 21, 268-78.
10. Aubin-Tam, M.E., Olivares, A.O., Sauer, R.T., Baker, T.A., & Lang, M.J. (2011). Singlemolecule protein unfolding and translocation by an ATP-fueled proteolytic machine. Cell 145,
257-67.
11.
Sauer, R.T., Bolon, D.N., Burton, B.M., Burton, R.E., Flynn, J.M., Grant, R.A., Hersch,
G.L., Joshi, S.A., Kenniston, J.A., Levchenko, I., Neher, S.B., Oakes, E.S., Siddiqui, S.M., Wah,
D.A., Baker, T.A. (2004) Sculpting the proteome with AAA(+) proteases and disassembly
machines. Cell 119, 9-18.
12. Burton, R.E., Siddiqui, S.M., Kim, YI., Baker, T.A., & Sauer, R.T. (2001). Effects of protein
stability and structure on substrate processing by the ClpXP unfolding and degradation machine.
EMBOJ. 20, 3092-100.
13.
Kim, Y., Burton, R.E., Burton, B.M., Sauer, R.T., & Baker, T.A. (2000). Dynamics of
substrate denaturation and translocation by the ClpXP degradation machine. Mol. Cell 5, 639-48.
14. Nager, A.R., Baker, T.A., & Sauer, R.T. (2011). Stepwise unfolding of a
P barrel
protein by
the AAA+ ClpXP protease. J. Mol. Biol. 413, 4-16.
15. Kenniston, J.A., Burton, R.E., Siddiqui, S.M., Baker, T.A., & Sauer, R.T. (2004). Effects of
local protein stability and the geometric position of the substrate degradation tag on the
efficiency of ClpXP denaturation and degradation. J. Struct. Biol. 146, 130-40.
16.
Lee, C., Schwartz, M.P., Prakash, S., Iwakura, M., & Matouschek, A. (2001). ATP37
dependent proteases degrade their substrates by processively unraveling them from the
degradation signal. Mol. Cell 7, 627-37.
17. Bolon, D.N., Grant, R.A., Baker, T.A., & Sauer, R.T. (2004). Nucleotide-dependent substrate
handoff from the SspB adaptor to the AAA+ ClpXP protease. Mol. Cell 16, 343-50.
18. Improta, S., Politou, A.S., & Pastore, A. (1996). Immunoglobulin-like modules from titin Iband: extensible components of muscle elasticity. Structure 4, 323-37.
19.
Ishikawa, K., Okumura, M., Katayanagi, K., Kimura, S., Kanaya, S., Nakamura, H., &
Morikawa, K. (1993). Crystal structure of ribonuclease H from Thermus thermophiles HB8
refined at 2.8 A resolution. J. Mol. Biol. 230, 529-42.
20.
Banecki, B., Wawrzynow, A., Puzewicz, J., Georgopoulos, C., & Zylicz, M. (2001).
Structure-function analysis of the zinc-binding region of the Clpx molecular chaperone. J Biol.
Chem. 276, 18843-8.
21.
Glynn, S.E., Martin, A., Nager, A.R., Baker, T.A., & Sauer, R.T. (2009). Structures of
asymmetric ClpX hexamers reveal nucleotide-dependent motions in a AAA+ protein-unfolding
machine. Cell 139, 744-56.
22. Glynn, S.E., Nager, A.R., Baker, T.A., & Sauer, R.T. (2012). Dynamic and static components
power unfolding in topologically closed rings of a AAA+ proteolytic machine. Nat. Struct. Mol.
Biol. 19, 61623. Ortega, J., Singh, S.K., Ishikawa, T., Maurizi, M.R., & Steven, A.C. (2000). Visualization of
substrate binding and translocation by the ATP-dependent protease, ClpXP. Mol. Cell 6, 1515-21
38
24. Farrell, C.M., Baker, T.A., & Sauer, R.T. (2007). Altered specificity of a AAA+ protease.
Mol. Cell 25, 161-6.
25. Siddiqui, S.M., Sauer, R.T., & Baker, T.A. (2004). Role of the processing pore of the ClpX
AAA+ ATPase in the recognition and engagement of specific protein substrates. Genes Dev. 18,
369-74.
26.
Martin, A., Baker, T.A., & Sauer, R.T. (2007). Distinct static and dynamic interactions
control ATPase-peptidase communication in a AAA+ protease. Mol. Cell 27, 41-52.
27.
Martin, A., Baker, T.A., & Sauer, R.T. (2008). Diverse pore loops of the AAA+ ClpX
machine mediate unassisted and adaptor-dependent recognition of ssrA-tagged substrates. Mol.
Cell 29, 441-50.
28. Martin, A., Baker, T.A., & Sauer, R.T. (2008). Pore loops of the AAA+ ClpX machine grip
substrates to dirve translocation and unfolding. Nat. Struct. Mol. Biol. 15, 1147-51.
29. Barkow, S.R., Levchenko, I., Baker, T.A., & Sauer, R.T. (2009). Polypeptide translocation
by the AAA+ ClpXP protease machine. Chem. Biol. 16, 605-12.
30. Kon, T., Oyama, T., Shimo-Kon, R., Imamula, K., Shima, T., Sutoh, K:, & Kurisu, G.
(2012). The 2.8 A crystal structure of the dynein motor domain. Nature 484, 345-50.
31. Carter, A.P., Cho, C., Jin, L., & Vale, R.D. (2011). Crystal structure of the dynein motor
domain. Science 331, 1159-65.
32. Wang, J., Song, J.J., Franklin, M.C., Kamtekar, S., Im, Y.J., Rho, S.H., Seong, I.S., Lee,
C.S., Chung, C.H., & Eom, S.H. (2001). Crystal structures of the HslVU peptidase-ATPase
39
complex reveal an ATP-dependent proteolysis mechanism. Structure 9, 177-84.
33. Lander, G.C., Estrin, E., Matyskiela, M.E., Bashore, C., Nogales, E., & Martin, A. (2012).
Complete subunit architecture of the proteasome regulatory particle. Nature 482, 186-91.
34. Noji, H., Yasuda, R., Yoshida, M., Kinosita, K. Jr. (1997) Direct observation of the rotation
of F1-ATPase. Nature 386, 299-302.
35. Yildiz, A., Tomishige, M., Vale, R.D., & Selvin, P.R. (2004) Kinesin walks hand-over-hand.
Science 303, 676-8.
36. Yildiz, A., Forkey, J.N., McKinney, S.A., Ha, T., Goldman, Y.E., & Selvin, P.R. (2003)
Myosin V walks hand-over-hand: single fluorophore imaging with 1.5-nm localization. Science
300, 2061-5.
37. DeWitt, M.A., Chang, A.Y., Combs, P.A., & Yildiz, A. (2012). Cytoplasmic dynein moves
through uncoordinated stepping of the AAA+ ring domains. Science 335, 221-5.
38. Scanlon, J.A., Al-Shawi, M.K., & Nakamoto, R.K. (2008). A rotor-stator cross-link in the
F1-ATPase blocks the rate-limiting step of rotational catalysis. J Biol. Chem. 283, 26228-40.
39. Tsunoda, S.P., Muneyuki, E., Amano, T., Yoshida, M., & Noji, H. (1999). Cross-linking of
two beta subunits in the closed conformation in Fl-ATPase. J Biol. Chem.274, 5701-6.
40.
Shirakihara, Y, Leslie, A.G.W., Abrahams, J.P., Walker, J.E., Ueda, T., Sekimoto, Y,
Kambara, M., Saika, K., Kagawa, Y., & Yoshida, M. (1997). The crystal structure of the
nucleotide-free alpha 3 beta 3 subcomplex of Fl -ATPase from the termophilic Bacillus PS3 is a
symmetric trimer. Structure 5, 825-36.
40
41. Yagi, J., Tozawa, K., Sekino, N., Iwabuchi, T., Yoshida, M., & Akutsu, H. (1999). Functional
conformational changes in the TF 1-ATPase P subunit probed by 12 tyrosine residues. Biophys. J.
77, 2175-83.
42. Abrahams, J.P., Leslie, A.G., Lutter, R., & Walker, J.E. (1994). Structure at 2.8 A resolution
of F1-ATPase from bovine heart mitochondria. Nature 370, 621-8.
43. Konisita, K., Adachi, K., & Itoh, H. (2004). Rotation of Fl -ATPase: how an ATP-driven
molecular machine may work. Annu. Rev. Biophys. Biomol. Struct. 33, 245-68.
44. Gibbons, C., Montgomery, M.G., Leslie, A.G., & Walker, J.E. (2000). The structure of the
central stalk in bovine F(1)-ATPase at 2.4 A resolution. Nat. Struct. Biol. 7, 1055-61.
45. Hirono-Hara, Y., Noji, H., Nishiura, M., Muneyuki, E., Hara, K.Y., Yasuda, R., Kinosita, K.
Jr., & Yoshida, M. (2001). Pause and rotation of F(l)-ATPase during catalysis. Proc. Nati. Acad.
Sci. U.S.A. 98, 13649-54.
46.
Matsui, T., Muneyuki, E., Honda, M., Allison, W.S., Dou, C., & Yoshida, M. (1997).
Catalytic activity of the alpha3beta3 gamma complex of Fl-ATPase without noncatalytic
nucleotide binding site. J Biol. Chem. 272, 8215-21.
47. Jault, J.M., Dou, C., Grodsky, N.B., Matsui, T., Yoshida, M., & Allison, W.S. (1996). The
alpha3beta3 gamma subcomplex of the F1 -ATPase from the thermophilic bacillus PS3 with the
betaT165S substitution does not entrap inhibitory MgADP in a catalytic site during turnover. J.
Biol. Chem. 271, 28818-24.
48. Ariga, T., Muneyuki, E., & Yoshida, M. (2007). F1-ATPase rotates by an asymmetric,
41
sequential mechanism using all three catalytic subunits. Nat. Struct. Mol. Biol. 14, 841-6.
49. Bowler, M.W., Montgomery, M.G., Leslie, A.G., & Walker, J.E. (2006). How azide inhibits
ATP hydrolysis by the F-ATPases. Proc.Natl. A cad. Sci. U.S.A. 103, 8646-9.
50.
Gledhill, J.R., Montgomery, M.G., Leslie, A.G., & Walker, J.E. (2007). Mechanism of
inhibition of bovine F1-ATPase by resveratrol and related polyphenols. Proc. Natl. Acad Sci.
U.S.A. 104, 13632-7.
51.
Saita, E., Lino, R., Suzuki, T., Feniouk, B.A., Kinosita, K. Jr., & Yoshida, M. (2010).
Activation and stiffness of the inhibited states of F1-ATPase probed by single-molecule
manipulation. J. Biol. Chem. 285, 11411-7.
52. Reck-Peterson, S.L., Yildiz, A., Carter, A.P., Gennerich, A., Zhang, N., & Vale, R.D. (2006).
Single-molecule analysis of dynein processivity and stepping behavior. Cell 126, 335-48.
53. Derr, N.D., Goodman, B.S., Jungmann, R., Leschziner, A.E., Shih, W.M., & Reck-Peterson,
S.L. (2012). Tug-of-War in motor protein ensembles revealed with a programmable DNA
origami scaffold. Science 338, 662-665.
54. Hersch, G.L., Burton, R.E., Bolon, D.N., Baker, T.A., & Sauer, R.T. (2005). Asymmetric
interactions of ATP with the AAA+ ClpX6 unfoldase: allosteric control of a protein machine.
Cell 121, 1017-27.
55. Yakamavich, J.A., Baker, T.A., & Sauer, R.T. (2008). Asymmetric nucleotide transactions of
the HslUV protease. J Mol. Biol. 380, 946-57.
56.
Smith, D.M., Fraga, H., Reis, C., Kafri, G., & Goldberg, A.L. (2011). ATP binds to
42
proteasomal ATPases in pairs with distinct functional effects, implying an ordered reaction cycle.
Cell 144, 526-38.
57.
Gai, D., Zhao, R., Li, D., Finkielstein, C.V., & Chen, X.S. (2004). Mechanisms of
conformational change for a replicative hexameric helicase of SV40 large tumor antigen. Cell
119, 47-60.
58. Enemark, E.J., & Joshua-Tor, L. (2006) Mechanism of DNA translocation in a replicative
hexameric helicase. Nature 442, 270-5.
59. Sanders, C.M., Kovalevskiy, O.V., Sizov, D., Lebedev, A.A., Isupov, M.N., & Antson, A.A.
(2007). Papillomavirus El helicase assembly maintains an asymmetric state in the absence of
DNA and nucleotide cofactors. Nucleic Acids Res. 35, 6451-7.
60.
Martin, A., Baker, T.A., & Sauer, R.T. (2005). Rebuilt AAA+ motors reveal operating
principles for ATP-fuelled machines. Nature 437, 1115-20.
43
Chapter 2
Stepwise unfolding of a 1 barrel protein by the AAA+ ClpXP
protease
This chapter contains experiments from A.R. Nager, T.A. Baker, and R.T. Sauer (2011) J Mol
Biol 413, 4-16, as well as unpublished experiments that will form the basis of a manuscript to be
written. I performed all of the experiments.
44
Abstract
In the AAA+ ClpXP protease, ClpX uses the energy of ATP binding and hydrolysis to unfold
proteins before translocating them into ClpP for degradation. For proteins with C-terminal ssrA
tags, ClpXP pulls on the tag to initiate unfolding and subsequent degradation. Here, I
demonstrate that an initial step in ClpXP unfolding of the 11-stranded
P barrel
of superfolder
GFP-ssrA involves extraction of the C-terminal P strand. The resulting 10-stranded intermediate
is populated at low ATP concentrations, which stall ClpXP unfolding, and at high ATP
concentrations, which support robust degradation. To determine if the C-terminal
p strand causes
low-ATP stalling, I designed and characterized circularly permuted GFP variants. Notably,
stalling was observed for a variant whose 10-stranded intermediate could reassociate with the
extracted strand, but not for a variant with a stable intermediate that could not. Additionally, I
observe stalling for a circularly permutated GFP variant with a C-terminal a helix rather than a
strand.
P
A stepwise-degradation model in which the rates of terminal-structure extraction,
refolding or recapture, and unfolding of the intermediate all depend on the rate of ATP hydrolysis
by ClpXP accounts for the observed changes in degradation kinetics over a broad range of ATP
concentrations. My results suggest that unfolding intermediates will play important roles in
determining whether forced enzymatic unfolding requires a minimum rate of ATP hydrolysis.
45
Introduction
In cells ranging from bacteria to mammals, AAA+ proteases bind specific target proteins and
then use cycles of ATP binding and hydrolysis to unfold them and to translocate the denatured
polypeptide into a compartmental peptidase for degradation.1 Although these ATP-fueled
machines can unfold substrates with diverse structures and stabilities, some proteins resist
proteolysis or are only partially degradedf. Inhibitory or slippery sequences, highly stable
domains, or stable unfolding intermediates have all been proposed to play roles in helping
proteins resist degradation.
The ClpXP protease of Escherichia coli consists of the AAA+ ClpX unfoldase and the associated
ClpP compartmental peptidase. 8 -9 Peptide signals that bind in the axial pore of a hexameric ClpX
ring, including the 11-residue ssrA tag, target substrates for ClpXP degradation.10 - 13 ATPdependent translocation is thought to pull the peptide tag through the pore, eventually unfolding
attached domains that cannot pass through the narrow channel in a native conformation (Fig.
2.1A). For stable proteins, unfolding is generally the rate-limiting step in ClpXP degradation,
often requiring hydrolysis of hundreds of ATP molecules. 14-15 Once the substrate is unfolded,
ClpX translocates the unfolded polypeptide into the degradation chamber of ClpP in steps of 5-8
amino acids per power stroke.16-17
Because the ssrA tag is a C-terminal degradation signal,
translocation of ssrA-tagged substrates begins at the C-terminus and proceeds to the N-terminus.
SsrA-tagged green fluorescent protein (GFP-ssrA) is an excellent model substrate for ClpXP,
because unfolding and degradation can be monitored by loss of native fluorescence.'14'
8
Interestingly, however, ClpXP degradation of GFP-ssrA ceases or stalls at low ATP
concentrations, whereas degradation of other stable proteins slows linearly as the rate of ATP
46
hydrolysis decreases.7 To explain these observations, Martin et al. proposed that unfolding of
GFP-ssrA by ClpX requires two steps: initial extraction of the C-terminal P strand, followed by
unfolding of the resulting 10-stranded barrel. 7 They also suggested that a thresh-hold rate of ATP
hydrolysis was needed for degradation because capture of the strand-extracted intermediate
required multiple rapid cycles of ATP hydrolysis to prevent refolding of the extracted strand.
The fluorescence properties of GFP depend on its folded structure. For example, denatured GFP
displays very low fluorescence because of solvent quenching. In native GFP, by contrast, an 11
-
stranded P barrel shields the enclosed chromophore (residues 65-67), which contains a phenolic
side chain that equilibrates slowly between protonated and unprotonated states (Fig. 2. 1ABC). 19 2
The unprotonated chromophore absorbs 467-nm light and emits 511 -nm light (hereafter called
467-nm fluorescence). The protonated chromophore absorbs 400-nm light but also emits a 511nm photon (hereafter called 400-nm fluorescence), because absorption transiently leads to
deprotonation via an excited state proton transfer (ESPT) reaction (Fig. 2. lC).2- Importantly,
the proton acceptor (Glu22 2) for this transfer reaction is on
GFP, and the Glu 222 -Gln
p-strand
11, near the C-terminus of
mutant displays normal 467-nm fluorescence but no 400-nm
fluorescence.24 In the studies below, these fluorescence properties allow detection of a GFP
species in which the P barrel is largely intact but the C-terminal p-strand is not.
47
AB
SFGFP
400-nrn
Glu
strand 11
tn
protontated
chromophore
sarA
H
tagH.H
0
CIPX
H
ESPTjr,.511nn
chr~mohor
H
excited
unprotonated
CipP
chror
- phton
H
H-O H
Figure 2.1 Green Fluorescent Protein
(A) Cartoon showing ClpXP after engaging the ssrA tag of SFGFP-ssrA but before unfolding, translocation, and degradation. The
GFP chromophore is shown in CPK representation (nitrogen, blue; oxygen, red; carbon, green). (B) Diagram of the secondary
structure of SFGFP-ssrA. P strands are the same color as in the structure in panel A. (C) Absorption of 400-nm light results in
excited state proton transfer (ESPT) in which the phenolic proton moves to Glu22 2 on strand 11. Return to the ground state is
accompanied by fluorescence emission at 511 nm. Mutation of Ser 20s or Glu 222 prevents ESPT and fluorescence after excitation
with 400-nm light. Fluorescence arising from excitation of the deprotonated chromophore with 467-nm light does not depend on
Ser 20 s or Gly 222 .
In this paper, I use ssrA-tagged variants of superfolder GFP (SF GFP; ref. 25) to probe the
mechanism of unfolding and degradation by ClpXP. Using fluorescence signals that monitor
different protein species, I show that ClpXP produces a strand-extracted intermediate of sFGFPssrA, which is substantially populated both at low ATPase rates where degradation stalls, and at
high ATPase rates where robust degradation occurs. The rates of appearance and disappearance
of this intermediate suggest an on-pathway role in unfolding and degradation.
SFGFP
contains
multiple stabilizing mutations, 25 which allowed me to design, purify, and characterize circularly
permuted variants in which different
P strands
or loops of SFGFP contained the C-terminal ssrA
tag. One of these variants showed ClpXP stalling at low ATPase rates, whereas two others did
48
not. I also engineered sites for thrombin cleavage between the C-terminal and penultimate
strands of SFGFP-ssrA and the permuted variants, cleaved these proteins to produce split proteins,
and then removed the C-terminal strand by ClpXP extraction to test if the resulting 10-stranded
barrels maintained metastable structures. Even though all variants had stable intermediates,
stalling was only observed for substrates in which the strand-extracted intermediate could
reassociate with the extracted
p
strand. In combination, these results show that ClpXP unfolds
GFP in a stepwise fashion and support a model in which the rates of terminal-strand extraction,
strand refolding, and unfolding of the 10-stranded intermediate all depend on the rate of ATP
hydrolysis.
Results
ClpXP extraction of terminal P strands in split-GFP variants
GFP lacking its
1 1 th
strand maintains a folded structure. 7 ,26 I inserted a site for thrombin cleavage
between strands 10 and 11 of
SFGFP-ssrA
(SFGFP- 10/11 -ssrA; Fig. 2.2A), incubated the purified
protein with thrombin, and confirmed that cleavage had occurred by SDS-PAGE (Fig. 2.2B,
lanes 1 and 2). This split protein had absorbance and fluorescence spectra similar to those of the
uncleaved protein (Fig. 2.2C, 2D). Next, I incubated the split substrate with ClpXP and
saturating ATP, and monitored fluorescence emission after excitation at 400 or 467 nm (hereafter,
called 400-nm and 467-nm fluorescence). Importantly, I observed complete time-dependent loss
of 400-nm fluorescence (initial rate ~1.5 min-' enz-1) with a small increase in 467-nm
fluorescence (Fig. 2.2E). SDS-PAGE confirmed that incubation of the split
SFGFP-10/11-ssrA
substrate with ClpXP destroyed the small fragment, corresponding to the ssrA-tagged
49
1 1 th
strand,
but did not alter the large fragment, corresponding to the remaining structural elements of GFP
(Fig. 2.2B, lane 3). Previous studies have shown that 467-nm fluorescence depends on shielding
of the GFP chromophore from water by the P barrel, 2 7 whereas 400-nm fluorescence requires
excited state proton transfer (ESPT) from the chromophore to the side chain of Glu 22 2 in strand
11 (Fig. 2.1 C).22,24 In combination, these results indicate that ClpXP removes the ssrA-tagged
1 1 th
strand of the split substrate without denaturing the remaining 10-stranded structure.
I also constructed a
SFGFP-9/10-ssrA
variant, containing a thrombin-cleavage site between
strands 9 and 10 (Fig. 2.2A), cleaved this protein with thrombin (Fig. 2.2B, lanes 4 and 5), and
performed experiments similar to those described above. In this case, incubation of the 9/10-split
substrate with ClpXP resulted in loss of both 400-nm and 467-nm fluorescence with an initial
rate of ~0.5 min enz-1 (Fig. 2.2F). I conclude that ClpXP extraction of the
leaves the
P barrel largely
intact, whereas extraction of both the
1 0 th
and the
1 1 th
1 1 th
strand of GFP
strands leads to
denaturation of the barrel, allowing solvent to quench the chromophore. The finding that ClpXP
extraction of strand 11 from the 10/11-split substrate occurred faster than extraction of strands 10
and 11 from the 9/10-split substrate is consistent with a sequential model of extraction of these
elements of secondary structure in the intact native protein.
50
A
1-10
A
P4
thrombin
thrombin
4
srA
-0-11
4
SFGFP1/11-ssrA
B
LVPRGS
LVPRGS
4
1
12
SFGFP9111-ssrA
1
j-CPK
uncleaved cleaved -F
S p1-10
- CIpP
131-9=rA
srA
C-terminal
fragment
C
no
no
2.
yes yes
no yes
no
no
yes yes thrombin
no yes CIpXP
D
1.0.
0
1.
cleaved
SFGFP-10/11-ssrA
0.6.
:
*.
e
0
.
404
Dos.
a
e
:
*
E 0.4.
cleaved
SFGFP-1 0/11 -ssrA ".
uncleaved
0
00
350
460
e
0.2.
uncleaved
FGFP-10/11-ssrA
lses
0
A
450
480
500
520
540
emission wavelength (nm)
500
wavelength (nm)
F
E 1.0
0.8
1.0<
L1-
10-11
A
0.6
0
rA
ClpxP4
0.8-
0
UIL
*
SFGFP-10/11-ssrA
131-9
0.6467-nm
L
0.
0.4-
0.2
0.2400-nm
0
200
400
600
800
1000
time (s)
0
1000
2000
300
time (s)
Figure 2.2 A stable 10-stranded barrel
(A) A thrombin cleavage sequence (LVPRGS) was inserted between strands 10/11 or 9/10 in the SFGFP-10/1 l-ssrA and SFGFP9/10-ssrA proteins (NCBI accession codes JF951868 and JF951869, respectively), allowing creation of split proteins. (B) SDSPAGE showing SFGFP-10/l l-ssrA or SFGFP-9/10-ssrA (10 ptM each) before and after thrombin cleavage and ClpXP (0.3 pM
ClpX 6 ; 0.9 pM ClpP 14) extraction/degradation. The gel is a composite, with the lower portion taken from a gel containing 8-fold
more sample than the upper portion. CPK, creatine phosphokinase. (C) Absorbance spectra of SFGFP-10/1l-ssrA before (closed
circles) or after (open circles) thrombin cleavage. (D) Fluorescence emission spectrum of SFGFP-10/ll-ssrA before (closed
circles) or after (open circles) thrombin cleavage. (E) Incubation of 10 pM thrombin-cleaved SFGFP-10/l1-ssrA with 1 pM
ClpXP (1 pM ClpX 6; 2 pM ClpP 14) and 4 mM ATP resulted in loss of 400-nm (open circles) but not 467-nm (closed circles)
fluorescence. (F) Incubation of 10 gM thrombin-cleaved SFGFP-9/10-ssrA with 1 pM ClpXP and 4 mM ATP resulted in loss of
400-nm (open circles) and 467-nm (closed circles) fluorescence. The experiments in panels E and F contained an ATPregeneration system.
51
GFP fluorescence during ClpXP stalling supports terminal-strand extraction
At the low ATPase rates that result in stalling, competition between ClpXP extraction and
subsequent refolding of the eleventh strand of GFP was proposed to result in populations of the
strand-extracted and native structures that depend on the rates of each reaction.7 This model
predicts that stalling conditions should result in lower values of 400-nm fluorescence (a measure
of intact GFP) as compared to 467-nm fluorescence (intact GFP plus the strand-extracted
species). Indeed, using ClpXP (1 [pM),
SFGFP-ssrA
(10 pM), and a low ATP concentration (50
pM), I observed a time-dependent decrease in 400-nm fluorescence but almost no change in 467nm fluorescence (Fig. 2.3A). After -1000 s, the 400-nm fluorescence stabilized, suggesting that
equilibrium had been reached. I observed no change in 400-nm or 467-nm fluorescence in the
absence of ClpXP but observed rapid loss of both signals when 4 mM ATP was added to the
stalled reaction after 2500 s (data not shown).
52
A
C
SFGFP-ssrA
5OpJM ATP
0
401.
50 pM ATP
467-nm
M ft
LL 0.
0
B
H148D-SFGFP-ssrA
500
1000
1500
2000
400-nm
010
LL
4100-nm
2500
D
1
0
I
09 0
47n
0.9
1
200
00
300
400
500
600
1.0-
*0
0.8-
0.640
/
ULL
LL
%* 467-nm
400-nm
$,
0.4-
0.2-
00016"--.
* 00
4 mM ATP
0 .1
2500
time (s)
0
100
200
300
400
500
600
time (s)
Figure 2.3 An unfolding intermediate is populated during stalling
(A) Changes in 400-nm fluorescence (open circles) or 467-nm fluorescence (closed circles) following incubation of sFGFP-ssrA
(10 pM) with ClpXP (1 IM ClpX 6 ; 2 pM ClpP14) and 50 pM ATP. (B) Same proteins as in panel A but using 4 mM ATP. The
inset shows the concentration of the strand-extracted intermediate after 250 s as a function of ClpXP concentration from 4 mM
ATP experiments like the one in the main panel. Values plotted are averages (n=4) ± 1 standard deviation. (C) Changes in 400nm fluorescence (open diamonds) or 467-nm fluorescence (closed diamonds) following incubation of H148D-SFGFP-ssrA (10
RM; NCBI accession code JF951865) with ClpXP (1 PM ClpX 6 ; 2 IM ClpP 14 ) and 50 pM ATP. (D) Same proteins as in panel B
but using 4 mM ATP. An ATP-regeneration system was used in all experiments.
To determine if the strand-extracted intermediate of SFGFP-ssrA accumulated under robust
degradation conditions, I assayed changes in 400-nm and 467-nm fluorescence in a reaction
containing 10 pM substrate, 1 tM ClpXP, and 4 mM ATP (Fig. 2.3B). As noted above, 467-nm
fluorescence includes contributions from native GFP plus the intermediate, whereas 400-nm
fluorescence depends only on the native GFP concentration. Thus, the concentration of the
strand-extracted intermediate can be calculated as a function of the initial GFP concentration
(GFPo), and the normalized 400-nm and 467-nm fluorescence: [I] = GFPo-(
53
467F/ 467Fo
-
4 00
F/ 4 00 Fo). The concentration of the intermediate was substantial (~25%
of the ClpXP
concentration) during most of the degradation reaction (Fig. 2.3B). Moreover, when I varied the
ClpXP concentration but kept the
SFGFP-ssrA
(10 pM) and ATP (4 mM) concentrations constant,
the amount of the intermediate increased linearly with ClpXP concentration (Fig. 2.3B, inset).
This result is expected for an enzyme-bound intermediate in degradation. An on-pathway
intermediate would need to form at a faster rate than overall degradation. Indeed, the rate of
ClpXP extraction of strand 11, calculated from the 10/11 -split GFP experiment (Fig. 2.2E), was
~4-fold faster than the degradation rate in the experiment using 1 ptM ClpXP and 4 mM ATP
(Fig. 2.3B).
The His 148 -Asp GFP mutation on p-strand 7 provides an alternative ESPT acceptor and restores
222
-Gln
400-nm fluorescence to Glu9
GFP. 2 ' Thus, H148D-SFGFP should not lose 400-nm
fluorescence even upon extraction of Glu22 2 and strand 11 from the
P barrel.
Indeed, incubation
of H148D-SF GFP-ssrA with ClpXP caused no change in 400-nm or 467-nm fluorescence using
50 pM ATP (Fig. 2.3C), which results in stalling, but caused concurrent loss of both signals
during degradation using 4 mM ATP (Fig. 2.3D).
Stalling behavior of circularly permuted GFP variants
To investigate if a stable unfolding intermediate is sufficient for stalling, I constructed and
purified three circularly permuted variants of GFP, in which different C-terminal structural
elements would be initially extracted by ClpXP (Fig. 2.4A). One variant, cp6a-SF GFP-ssrA, in
54
which the ssrA tag was attached to an a helix following the
6 th
strand, displayed "stalling"
behavior and was degraded by ClpXP at high but not low ATP concentrations as assayed by
SDS-PAGE (Fig. 2.4B) or by loss of 467-nm fluorescence (Fig. 2.4C). By contrast, permuted
variants with strand 7 (cp7- SF GFP-ssrA) or strand 8 (cp8-SFGFP-ssrA) at the C-terminus showed
non-stalling behavior and were degraded at ATP concentrations as low as 50 PM (Fig. 2.4BC).
A
order of 1 strands
JFJDaDlIJ51ssrA
SFGFP-ssrA
[[U
Cp6a-SFGFP-ssrA
cp7-SFGFP-ssrA
cp8-SFGFP-ssrA
@
@
1
(])-ssrA
J D
3I
U
C
1
c
S
0.8-
@-ssrA
Ef-ssrA
SFGFP-ssrA
C
cp6a-SFGFP-ssrA
B
GFP-ssrA
SFGFP-ssrA
0.4-
cp7-SFGFP-ssrA
S0.2.
cp8-SFGFP-ssrA
7
cp6a-SFGFP-ssrA
0
50
cp7.SFGFP-ssrA
0
300
0
[ATP] (pM)
50
100
200
150
[ATP] (pM)
250
300
Figure 2.4 Circularly permuted GFP variants show stalling and non-stalling ClpXP degradation.
(A) Cartoon representation of the order of P strands in SFGFP-ssrA and circularly permuted variants. (B) Permuted variants (1
1 iM)
were incubated overnight with ClpXP (1.25 pM ClpX 6 ; 2.5 9M CIpP 14), the SspB adaptor (1 pM), and 0, 50, or 300 pM
ATP before assaying degradation by SDS-PAGE. (C) End-point experiments like those in panel B were performed but
degradation was assayed by reduced 467-nm fluorescence. GFP-ssrA (circles); SFGFP-ssrA (diamonds); cp6a-sFGFP-ssrA
(upward triangles); cp7-SFGFP-ssrA (triangles); cp8-SFGFP-ssrA (squares). The lines are fits to a modified form of the Hill
equation. In the panel-B and panel-C experiments, an ATP-regeneration system was used.
To test if the circular permutants also display stable intermediates in ClpXP unfolding, I
engineered thrombin-cleavage sites before the C-terminal element of structure. For the cp7SFGFP-6/7-ssrA
and cp8-SFGFP-7/8-ssrA variants, a thrombin-cleavage site was inserted between
the penultimate and C-terminal
between the
6 th
p strands. For cp6a-SFGFP-6/a-ssrA,
a thrombin site was inserted
p strand and the C-terminal a helix. ClpXP extraction of the terminal peptide of
each thrombin-split protein resulted in species with 400-nm fluorescence that were 68-105% of
55
the starting value (Fig. 2.6ABC), indicating that each extracted protein retained a folded
structure. Because all of these proteins form stable intermediates upon ClpXP extraction of their
C-terminal peptides and yet the parental proteins show stalling and non-stalling behaviors at low
ATP concentrations, I conclude that formation of a stable intermediate, by itself, is not sufficient
to explain stalling.
A
B
cp7.SFGFP-6/7-ssrA
1
1.05-
cp8-SFGFP-7/8-ssrA
0
CIpXP
A
0.8
LU
E
C
Cpx
0.6
U.
U-
Non-Stalling
500
100
0.4
I 500
time (s)
D
1500
time (s)
SFGFP-10/11-ssrA
0
.i
LU
0.
E
0.
Stalling
U.
0
500
1000
1
Figure 2.6 Effects of ClpXP versus ClpX extraction of terminal peptides from thrombin-split substrates
Thrombin-cleaved GFP variants (10 pM) were incubated with ClpXP (1 pM ClpX6 ; 2 pM ClpP 14 ; circles) or ClpX (1 pM ClpX6 ;
triangles). All reactions included ATP (4 mM) and an ATP-regeneration system. For non-stalling constructs, cp7-SFGFP-(6/7)ssrA (A) and cp8-SFGFP-7/8-ssrA (B), identical changes in fluorescence were observed upon treatment with ClpX or ClpXP. For
stalling constructs, cp6a-SFGFP-6/a-ssrA (C) and SFGFP-(10/1 1)-ssrA (D), extraction and degradation of the C-terminal structural
element resulted in larger changes than extraction alone. The ~15% decrease in SFGFP-(10/1 1)-ssrA fluorescence following ClpX
treatment was reversed once the ATP-regeneration system was exhausted.
56
In principle, stalling might not occur for cp7- SF GFP-ssrA and cp8- SFGFP-ssrA because the
extracted peptide sequence does not refold. To test if an extracted peptide from a split variant can
reassociate with the intermediate and reverse changes in fluorescence, I used ClpX to extract but
not degrade the C-terminal sequence. For the non-stalling substrates, cp7-SFGFP-6/7-ssrA and
cp8-SF GFP-7/8-ssrA, treatment with ClpX and ClpXP resulted in identical changes in 400-nm
fluorescence, suggesting that the extracted C-terminal strand does not reassociate (Fig. 2.6AB).
This result could be explained if the strand-extracted intermediate rapidly changes conformation
(e.g., forming a closed 10-straded barrel) and thus prevents reassociation. By contrast, ClpX
extraction from the stalling constructs, cp6a- SFGFP-6/a-ssrA or SFGFP-10/11-ssrA, resulted in no
permanent change in 400-nm fluorescence (Fig. 2.6CD). For
SFGFP-10/11-ssrA,
there was a
small decrease in 400-nm fluorescence in the presence of ClpX representing a steady-state
population of strand-extracted GFP, but fluorescence was regained once the ATP-regeneration
system was exhausted or EDTA was added to terminate ATP hydrolysis and steady-state
extraction (Fig. 2.6D; data not shown).
The results presented above support a model in which ClpX extraction of the C-terminal
sequence of a GFP substrate can lead to two outcomes. The first is partitioning between refolding
of the sequence at low ATP concentrations and processive unfolding of the intermediate at high
ATP concentrations. This outcome explains the low-ATP substrate stalling observed for
SFGFP-
ssrA and cp6a-SFGFP-ssrA. The second outcome posits that the extraction step is effectively
irreversible, leading to non-stalling behavior for substrates like cp7- SFGFP-ssrA and cp8- SFGFPssrA.
57
To test this model more directly, I designed a fluorescence assay for association of ClpXPextracted, thrombin-split SFGFP-10/11-ssrA (stalling) or cp8-SFGFP-7/8-ssrA (non-stalling)
proteins with TAMRA-labeled synthetic peptides identical to the extracted C-terminal peptide
but with no ssrA tag. If the TAMRA-labeled peptide associates with the newly formed 10stranded barrel, then a FRET change should occur as the TAMRA dye would be positioned
within 20 A of the GFP chromophore, well within the F~rster radius of -58 A. Indeed, as ClpXP
extracted the C-terminal strand of the split
SFGFP- 10/11
-ssrA protein, there was a time-dependent
decrease in GFP fluorescence and increase in TAMRA fluorescence when equimolar substrate
and peptide were used (Fig. 2.7A). The increase in TAMRA florescence and decrease in GFP
fluorescence occurred with similar kinetics, suggesting that ClpXP extraction of the ssrA-tagged
p strand rather than association of the TAMRA labeled peptide
was the slow step in the reaction.
Using higher concentrations of the TAMRA-labeled peptide did not result in substantial FRET
increases, as expected if binding of the TAMRA-labeled peptide was reasonably tight and only
one peptide bound each 10-stranded barrel (Fig. 2.7B). Thus, for the
SFGFP-
10/11 -ssrA stalling
substrate, extraction of the C-terminal strand does not preclude rebinding of an equivalent
sequence. ClpXP extraction of split cp8-SFGFP-7/8-ssrA in the presence of a TAMRA-peptide
corresponding to strand 8 also resulted in a time-dependent decrease in GFP fluorescence (Fig.
2.7C). However, there was no concomitant increase in TAMRA fluorescence (Fig. 2.7C; 2.7D),
as would be expected from FRET if the TAMRA peptide could bind following terminal-strand
extraction from this non-stalling substrate.
58
A
SFGFP-10/11-ssrA
B
SFGFP-10/11-ssrA
1*
1.6P
., JMWemission
X
S0.8ILL
E 1
.'0.4
U-
GFPemission
0
0.2-
0.40
0
100 200 300 400 500 600 700
2
m1
time (s)
3
5
4
molar ratio strand #11:GFP
C
cp8.SFGFP-718-ssrA
D
cp8-SFGFP-7/8-ssrA
1.6-
1.2
1.
Li
I-
E
TAMRAemission
1
0.8.
U0.6.
IF
C
U.
a
0
0.4.
C
0.2.
0.4-
10
01
260 360 460 560 600 700
0.1
time (s)
*
*
1
*
*
.
,0.
10
100
molar ratio strand #8:GFP
Figure 2.7 Strand-replacement assays
(A)
Thrombin-split SFGFP-(10/11)-ssrA
TAMRAHMVLLEFVTAAG (10 pM), ATP (4
(10 pM) was incubated with ClpXP (1 pM ClpX 6; 2 pLM ClpP 14),
mM), and an ATP-regeneration system. ClpXP degradation of the terminal p strand
permitted the TAMRA-labeled peptide corresponding to strand 11 to associate with the 10-stranded p barrel, decreasing GFP
emission and increasing TAMRA emission by FRET. (B) Final FRET ratios for experiments like that shown in panel A with
increasing concentrations of the TAMRA-labeled strand-Il peptide. The fit is to a quadratic equation for tight 1:1 binding with a
KD of 111 ± 30 nM. (C) Thrombin-split cp8-SFGFP-(7/8)-ssrA (10 pM) was incubated with ClpXP (1 IM ClpX ; 2 pM ClpP ),
14
TAMRAIKANFKIRHNV (10 pM), ATP (4 mM), and an ATP-regeneration system. No change in TAMRA 6
fluorescence was
observed, although GFP fluorescence decreased. (D) Increasing the concentration of the TAMRA-labeled strand-8 peptide in
experiments like that shown in panel D resulted in no significant TAMRA fluorescence changes, following excitation at 400 nm.
In combination, these results support a model in which the "stalling" phenotype requires both a
stable unfolding intermediate and the ability of the extracted sequence to refold or reassociate
with the intermediate. Specifically, the
SFGFP-ssrA
and cp6a-SFGFP-ssrA proteins stall ClpXP
and removal of their C-terminal structural elements results in f barrels that are stable but can
reassociate with the extracted strand or helix. By contrast, the cp7-SFGFP-ssrA and cp8--SFGFPssrA proteins did not stall ClpXP and extraction of their C-terminal strands resulted in stable
intermediates that failed to rebind the extracted sequence.
59
Stalling substrates have lower maximal rates of ClpXP degradation
Using saturating ATP, I determined steady-state kinetic parameters for ClpXP degradation of
different concentrations of GFP-ssrA,
SFGFP-ssrA
cp6a- SF GFP-ssrA, cp7-SFGFP-ssrA, and cp8-
SFGFP-ssrA,
(Table 1; Fig. 2.8A). Interestingly, cp7-SFGFP-ssrA and cp8-SF GFP-ssrA, the
substrates which did not stall, also had the highest Vmax value. Although the "stalling" substrates
displayed a range of Vmax values, with
SFGFP-ssrA
being the slowest, the overall correlation
suggests that substrates whose degradation stalls at low ATPase rates are also more difficult to
degrade at high ATPase rates.
degradation
Km
Vmax
stalls
ATP hydrolysis
Km
Hill
Vmax
protein stability
AG
meq
k,
m,
minenz-1
kcal/mol
kcal/mol
M-1
s-1
OM
GuHCl
kcal/mol
*
pm
minenz-1
Pm
yes
0.36
0.39
114
1.85
159
6.9
1.6
1.6E-9
1.1
SFGFP-ssrA
no
0.60
2.21
114
2.2
306
5.4
1.5
ND
ND
cp7-SFGFP-ssrA
no
0.33
1.48
115
2.2
336
4.9
1.9
2.9E-9
1.5
cp6-
yes
0.79
1.02
131
1.85
203
4.4
3.0
4.2E-7
1.2
yes
1.3
1.19
133
1.96
167
4.6
2.3
5.9E-7
0.79
SFGFP-ssrA
cp8-
SFGFP-ssrA
GFP-ssrA
Table 2.1 Properties of ssrA-tagged GFP substrates
*Stalling substrates displayed < 5% of the maximal degradation rate when the ATPase rate was 50% of maximal (Fig. 2.8B).
Errors for Km and k, (± 50%); errors for other values (± 10%). ND: Not Determined.
60
B
A
equilibrium unfolding
CIpXP degradation
2
SFGFP-ssrA
C.)
C
1
0
cp8-SFGFP-ssrA
8C
*0
U
E
[substrate] (pM)
cp7-SFGFP-ssrA
C
*
kinetic stability
0
0
10-1
cp6a-SFGFP-ssrA
cp7-SFGFP-ssrA
. *
SFGFP-ssrA
GFP-ssrA
10-5L5.0
5.5
6.0
0
6.5
[GuHCI] (M)
GFP-ssrA
0
0
1
2
3
4
5
6
[GuHC] (M)
Figure 2.8 Equilibrium and kinetic stability of GFP variants
(A) Michaelis-Menten plots of ClpXP degradation (0.1 pM ClpX 6 ; 0.2 pM ClpP 14 ) of ssrA-tagged variants of GFP and SFGFP in
the presence of ATP (4 mM) and an ATP-regeneration system. Initial degradation rates were calculated from changes in 467-nm
fluorescence. The lines are fits to the equation rate = Vmax[S]/(KM + [S]). Error bars (± 1 SD) based on four independent
replicates. Km and Vmax values for each substrate are listed in Table 2.1. (B) Proteins (0.5 pM) were incubated with different
concentrations of GuHCl for 2 weeks, and denaturation was assayed by 467-nm fluorescence. The solid lines are fits to a twostate unfolding model. Cp6-SFGFP-ssrA has an unusual native baseline and higher meq value than the other proteins (Table 2.1).
These properties could reflect an increase in the solvent accessibility of the unfolded protein and/or the presence of a populated
unfolding intermediate. (C) Rates constants for unfolding (ku) were determined by single-exponential fits of changes in 467-nm
fluorescence after jumps to different concentrations of GuHCl. The values plotted are averages of three independent experiments
± 1 standard deviation. The final protein concentration in each assay was 0.5 pM.
Equilibrium and kinetic stability
I determined the thermodynamic and kinetic stabilities of the GFP-ssrA,
SFGFP-ssrA,
cp7-SF GFP-ssrA,
SFGFP-ssrA,
cp6a-
and cp8-SF GFP-ssrA proteins at different concentrations of
GuHCl. The equilibrium stability of the
SFGFP-ssrA
protein was substantially greater than the
stabilities of GFP-ssrA, cp6a- SFGFP-ssrA, cp7- SFGFP-ssrA, and cp8- SFGFP-ssrA (Fig. 2.8B;
61
Table 2.1). At 5 M GuHCl, the order of kinetic stabilities from most to least stable was sFGFPssrA > GFP-ssrA > cp7-SF GFP-ssrA > cp6a-SF GFP-ssrA (Fig. 2.8C); extrapolation to 0 M
denaturant gave
SFGFP-ssrA
> cp7- SFGFP-ssrA > GFP-ssrA > cp6a-SFGFP-ssrA (Table 2.1).
These results show that neither ClpXP stalling nor the maximal rate of degradation correlate with
the equilibrium or kinetic stabilities of the GFP variants, a result that is consistent with previous
studies of different ClpXP substrates.4, 29 Indeed, the non-stalling substrates, cp7-SFGFP-ssrA and
cp8-SFGFP-ssrA,
were degraded at the fastest rates but had stabilities intermediate between those
of the stalling substrates.
Dependence of rates of ATP hydrolysis and degradation on ATP concentration
I determined the rates at which ClpXP hydrolyzed different concentrations of ATP in the
presence of 10 [iM GFP-ssrA,
SFGFP-ssrA,
cp6a-SFGFP-ssrA cp7-SF GFP-ssrA, or cp8-SF GFP-
ssrA. Fig. 2.9A shows normalized ATPase rates plotted as a function of ATP concentration. For
each substrate, the v/Vma curves were similar and were fit well by the Hill version of the
Michaelis-Menten equation, with half-maximal rates at ATP concentrations of 115-135 pM and
positively cooperative n values of 1.8 to 2.2 (Table 1). There were, however, substrate-dependent
Vmax differences, with the values for GFP-ssrA,
SF GFP-ssrA,
and cp6a-SFGFP-ssrA being
roughly similar (159-203 min-' enz-1), whereas the value for the non-stalling substrates, cp7SFGFP-ssrA
and cp8-SFGFP-ssrA, was substantially higher (306-336 min-' enz-1).
Next, I determined rates of degradation for GFP-ssrA,
62
SFGFP-ssrA,
cp6a-SFGFP-ssrA, cp7-
SGFP-ssrA,
or cp8-SF GFP-ssrA over a wide range of ATP concentrations by measuring loss of
467-nm fluorescence. Fig. 2.9B shows the fractional degradation rate plotted as a function of the
fractional ATPase rate (defined as a). For GFP-ssrA,
SFGFP-ssrA,
and cp6a-SFGFP-ssrA,
degradation fell off very steep non-linear fashions. Indeed, a modest initial decline in the ATPase
rate from 100 to 90% of maximal resulted in a ~5-fold decrease in the degradation rate of these
substrates. These substrates have stable unfolding intermediates that can rebind the extracted Cterminal sequence. By contrast, ClpXP degradation of cp7-SFGFP-ssrA and cp8- SFGFP-ssrA
decreased in a roughly linear manner with the ATPase rate. For both of these substrates, the
strand-extracted intermediate cannot reassociate with the extracted C-terminal sequence.
B
A
-~1.0-
-- c
0.6S 0.4
0.8
0.8
.p7FGFP-ssrA
GFP-ssrA
0.8
o1 06
2
cp7.SFGFP-ssrA -
06toSFGFPssA
G-0
o~~~GP
SF
C 8-F~
0
FCP7SFGFpssrA
-
0.4
0.6
sr
SFGCp60.SFGFP
cp8SFGFP-srA
-
0
cpsFGPssA5
0.4A
V
cp6F .SFGFP-ssrA
GFP-ssrA
SGFP-ssrA
C 0.2-
00.0
0
100
200
300
400
1
500
[ATP] (pM)
C
klea
ES :'
El
k. 1(1 -a)
0.2
0.4
0.6
0.8
fractional ATPase rate (a)
0
k2 ea
E+P
Figure 2.9 Two-step unfolding
(A) Fractional rates of ATP hydrolysis (a = v/Vmax) by ClpXP (1 PM ClpX 6 ; 2 IM ClpP 14 ) at different concentrations of ATP in
the presence of 10 pM GFP-ssrA (closed squares), SFGFP-ssrA (circles), cp6a-SFGFP-ssrA (triangles), cp7-SFGFP-ssrA
(diamonds), or cp8-SFGFP-ssrA (open squares). The solid line is a fit of the SFGFP-ssrA data to c = 1/(1+(Km/[ATP])"), the Hill
form of the Michaelis-Menten equation. Km, Vmax, and n values for each protein substrate are listed in Table 1. (B) Fractional
degradation rates
(v/Vmax)
for ClpXP (1 pM ClpX6 ; 2
M ClpP 14 ) proteolysis of 10 pM GFP-ssrA (diamonds),
SFGFP-ssrA
(circles), cp6a- SFGFP-ssrA (downward triangles), cp7-SFGFP-ssrA (upward triangles), and cp8-SFGFP-ssrA (squares) are plotted
as a function of the fractional rate of ATP hydrolysis (a). Solid lines are fits to the equation kga 2 /(kb(1-aC)+x) for GFP-ssrA,
cp7-SFGFP-ssrA, and cp8-SFGFP-ssrA, where ka = kik 2 /(ki+k 2) and kb = k. 1/(ki+k 2). (C) Single-intermediate model for enzymatic
unfolding, where ES represents the complex of ClpXP with intact GFP and El represents a complex in which the terminal p
strand of the substrate has been extracted, leaving a 10-stranded P barrel.
63
I found that a single-intermediate model (Fig. 2.9C) generally accounted for both the stalling and
non-stalling behaviors of the different GFP substrates. In this model, the ClpXP-substrate
complex (ES) forms a strand-extracted intermediate (EI) with a rate constant of ki-a, and El is
subsequently denatured/degraded with a rate constant of k2 -c or refolds to ES with a rate
constant of k.1-(1-ax). The latter term, disfavors refolding of the strand-extracted intermediate at
high ATPase rates, when most ClpXP enzymes are ATP bound. Assuming steady-state for [EI]
and substrate saturation, this model predicts that the fractional degradation rate equals
ki-k 2 -a2/(k.-(1-c)+(ki+k2)-ac). Fig. 2.9B shows fits to this equation for one of the stalling
substrates (GFP-ssrA; R > 0.99) and for the non-stalling substrates (cp7- SF GFP-ssrA and cp8SFGFP-ssrA;
R > 0.99 and > 0.97). Good fits were also obtained for
SFGFP-ssrA
(R > 0.98). Although the single-intermediate model recapitulates most of the
SFGFP-ssrA
and cp6a-
observed stalling features, it seems likely that additional unfolding intermediates are populated to
some degree, potentially including intermediates which the C-terminal
P strand is only
partially
dislodged or extracted.
Discussion
To explain why degradation of GFP-ssrA ceases at low rates of ATP hydrolysis, Martin et al.
proposed that ClpXP extracts the ssrA-tagged terminal strand from the GFP P barrel but the
intermediate refolds before further unfolding and degradation can occur.7 Our results strongly
support this stepwise model for ClpXP unfolding of GFP, which leads to futile cycles of strand
extraction and refolding at low concentrations of ATP. For example, I designed a split variant in
which the ssrA-tagged
1 1 th
strand was non-covalently bound to the remaining GFP structure and
64
found that ClpX extracted this strand without denaturing the rest of the protein and the extracted
sequence could reassociate with the 10-stranded barrel. Moreover, the rate of strand extraction by
ClpXP was fast enough to account for the rate of GFP proteolysis, as expected for an on-pathway
step in the degradation reaction. Notably, when ClpXP extracted the
1 1 th
and I0
th
strands of
another split GFP variant, the remaining portions of GFP were also denatured. Thus, GFP lacking
its C-terminal
1 1 th
strand is reasonably stable but subsequent extraction of the adjacent
1 0 th
strand results in cooperative unfolding. GFP-ssrA lacking its C-terminal strand loses 400-nm
fluorescence, which depends upon Glu2 22 in strand 11, but maintains normal 467-nm
fluorescence. Importantly, I found that ClpXP produces GFP species with a reduced ratio of
400/467 fluorescence both under low-ATP stalling conditions and high-ATP degradation
conditions, as predicted if a strand-extracted intermediate is populated under these conditions.
I designed a circularly permuted variant of superfolder GFP (cp6-SF GFP-ssrA), which formed a
stable intermediate and was able to reassociate with the extracted C-terminal strand. Like nonpermuted
SFGFP-ssrA,
this variant was resistant to ClpXP degradation at low ATP-hydrolysis
rates. By contrast, two other circularly permuted variants (cp7- SFGFP-ssrA, cp8- SFGFP-ssrA)
could not reassociate with an extracted C-terminal strand and did not stall ClpXP. Thus, the
simple presence of a stable unfolding intermediate is not sufficient to stall ClpXP degradation at
low ATP
concentrations.
Rather, stalling
requires
that
the extracted
sequence
can
refold/reassociate with the intermediate. At saturating ATP, ClpXP degraded stalling GFP
variants more slowly than the non-stalling variants. Thus, refolding of the extracted C-terminal
strand appears to slow normal degradation. Changes in the rates of strand extraction and/or in the
65
rates of unfolding of the 10-stranded barrel probably also play roles in determining the overall
rate of degradation.
As ClpX extracts the C-terminal strand of cp7-SFGFP-ssrA or cp8-SFGFP-ssrA, the resulting
p
barrel appears to undergo a conformational rearrangement that prevents reassociation of the
extracted sequence. Serpin proteins normally contain a 6-stranded
P sheet,
but a central strand
can spontaneously unfold and form a disordered loop. The resulting structure then rearranges to a
5-stranded sheet, forming new hydrogen bonds between the strands that originally flanked the
central strand and eliminating the binding site of the extracted strand. 30 A similar rearrangement
may occur for the non-stalling cp7-SFGFP-ssrA or cp8-SF GFP-ssrA upon ClpXP extraction of the
C-terminal strand. Presumably, energetics dictate whether unfolding intermediates maintain an
essentially wild-type structure that can rebind the missing structural element or change
conformation to eliminate this possibility.
ClpXP appears to unfold proteins using a power-stroke mechanism. 15-17 Specifically, each cycle
of ATP hydrolysis by ClpX is thought to result in an attempt to translocate a segment of the
substrate polypeptide, thereby pulling the native protein against the narrow axial channel and
creating a transient unfolding force. For stable substrates, like the titin127 domain, hundreds of
cycles of ATP hydrolysis can be required before denaturation becomes statistically probable.' 5
This result suggests that a power stroke must coincide with a stochastic decrease in protein
stability to successfully extract the terminal structural element of the substrate. For titin12 7 , this
66
initial ClpXP-mediated unfolding event appears to cause global denaturation.7 Studies of ClpXP
unfolding of a multi-domain filamin substrate assayed by optical-trapping nanometry also
support this model.16 For example, in different single-molecule experiments, the dwell time
before unfolding of a specific filamin domain varied from a few to more than 100 s, with the
latter time being sufficient to hydrolyze several hundred ATP molecules. However, once
unfolding of a filamin domain commenced, highly cooperative denaturation was typically
complete in less than 1 ms. Subsequent ATPase cycles then resulted in translocation of the
unfolded protein in steps of 5-8 amino acids at an average rate of-30 residues s- .
Our current view of the mechanism by which ClpXP unfolds GFP-ssrA begins with repeated
enzymatic tugging on the 11 th or C-terminal strand that is attached to the ssrA tag. During this
process, enzymatic pulling will occasionally partially or completely dislodge the terminal strand,
leaving a native 10-stranded barrel. Multiple cycles of ATP hydrolysis are then required to finish
translocation of the extracted strand and preceding turn (18
residues) and to unfold the
remaining 10-stranded structure to allow degradation. The time required for completion of these
events will increase as the ATPase rate decreases. Refolding of the extracted strand before
unfolding of the 10-stranded barrel would restore the original substrate, necessitating renewed
attempts to begin denaturation of the 11-stranded barrel. Strand refolding probably requires
slipping of the substrate from the grip of ClpXP and is therefore more likely to occur at low
ATPase rates when a higher fraction of enzymes are in an ATP-free state. Indeed, good fits of the
observed ATP dependence of GFP degradation by ClpXP required a strand-refolding rate
proportional to 1 minus the fractional ATP-hydrolysis rate.
67
The protein-unfolding activities of different AAA+ proteases display considerable variation. 32
For example, the HslUV and FtsH proteases fail to degrade GFP proteins with suitable Cterminal recognition tags, whereas Lon degrades such substrates about 200-fold more slowly
than ClpXP or ClpAP. 3 3 -37 For the proteases that degrade these GFP variants poorly, it is
presently unclear if these AAA+ enzymes fail to dislodge the C-terminal strand or if they fail in a
subsequent step in unfolding and degradation. Our results suggest that this question could be
resolved by assaying 400-nm and 467-nm fluorescence of appropriate GFP variants during
unfolding attempts by Lon, HslUV, and FtsH. Moreover, preliminary experiments suggest that
appropriately tagged versions of some of our circularly permuted GFP proteins will be useful
model substrates for Lon and HslUV, allowing convenient fluorescence-based assays of the
unfolding and degradation activities of these AAA+ proteases. ClpXP extraction of the terminal
elements of split proteins also provides a powerful new tool with which to investigate the
sequence determinants of protein structure and function or potentially to replace parts of split
proteins with synthetic peptides with fluorescent dyes or other modifications for studies of
structure-function relationships. GFP-fusion proteins are commonly used to study protein
localization and turnover in vivo, but interpretations can be complicated if partial degradation
generates free GFP. This problem could potentially be overcome by fusing proteins to a
circularly permutated GFP that can be completely degraded.
Materials and Methods
Protein Expression, Purification, and Cleavage
E. coli ClpX and E. coli ClpP were purified as described. 4'3 8 A His 6-tagged variant of E. coli
68
SspB was purified by Ni2+-NTA chromatography and S200 size-exclusion chromatography. 39
GFP substrates were expressed in E. coli strain X90, which had been transformed with
appropriate overproducing plasmids. Cells were grown at 37 C to OD 60 0 = 0.8, protein
expression from the T7 promoter was induced by addition of 1 mM IPTG, and growth was
continued at room temperature for 3.5 h, before the cells were harvested and lysed.
The coding sequence for SFGFP was obtained from the Registry of Standard Biological Parts
(BBa 1746916).
SFGFP
variants were cloned into pCOLADuet-1 with an N-terminal H6 tag
(MGSHHHHHH) and a C-terminal ssrA tag (AANDENYALAA). Table 2.1 lists the different
SFGFP
or GFP variants used for these studies. Most variants were constructed in the super-folder
GFP sequence background (SFGFP). 25 Circularly permuted GFP variants are designated with a
cp# prefix, where # represents the C-terminal
P strand
of the permuted structure, and were
cloned with a GGTGGS sequence connecting the residues corresponding to wild-type N
terminus and C terminus. 40 In some GFP variants, a GGTEGSLVPRGSGESGGS sequence for
thrombin cleavage was inserted into the loop between two P strands to allow production of split
proteins. These variants have names like
SFGFP-10/11-ssrA,
where 10/11 indicates insertion of
the cleavage site between strands 10 and 11.
All GFP variants were purified by Ni 2 +-NTA affinity (Qiagen) and S200 size-exclusion
chromatography, and were stored in PD buffer (50 mM HEPES [pH 7.5], 200 mM KCl, 5 mM
MgCl 2 , 10% glycerol). Cleavage of appropriate substrates with thrombin (GE Healthcare; 4
69
units/mg substrate) was performed in 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2.5 mM CaC12
for 2 h at 37 C.
TAMRA-labeled fluorescent peptides
Synthetic peptides corresponding to GFP strand 11 (TAMRAHMVLLEFVTAAG-COOH) or strand
8 (TAMRAIKANFKIRHNV) were ordered from the Koch Biopolymers and Proteomics Core
Facility and purified by HPLC.
Biochemical Assays
Degradation and unfolding assays were performed at 30 C in PD buffer and were monitored by
SDS-PAGE, and/or by loss of fluorescence emission at 511 nm after excitation at 400 nm or at
467 nm. Degradation reactions contained ssrA-tagged substrates, E. cli ClpX 6 , E. cli ClpP 14,
and an ATP-regeneration system (16 mM creatine phosphate, 6 pg/mL creatine phosphokinase).
Some degradation reactions also contained E. coli SspB, which helps deliver ssrA-tagged
substrates for ClpXP degradation. 4 1 Rates of ATP hydrolysis were determined at 30 'C in PD
buffer using an assay in which production of ADP is coupled to enzymatic oxidation of
NADH. 42-43 Equilibrium and kinetic stability assays were performed at 30 'C in PD buffer
supplemented with different concentrations of GuHCl and were monitored by changes in 467-nm
fluorescence.
70
Acknowledgements
I thank Peter Chien, Santiago Lima, and Randall Mauldin for helpful discussions. Supported by
NIH grant Al- 15706. T.A.B. is an employee of the Howard Hughes Medical Institute.
71
References
1. Baker, T.A. & Sauer, R.T. (2006). ATP-dependent proteases: recognition logic and operating
principles. Trends Biochem. Sci. 31, 647-653.
2. Palombella, V.J., Rando, O.J., Goldberg, A.L., & Maniatis, T. (1994). The ubiquitinproteasome pathway is required for processing the NF-KB 1 precursor protein and the activation
of NF-KB. Cell 78, 773-785.
3. Levitskaya, J., Sharipo, A., Leonchiks, A., Ciechanover, A., & Masucci, M.G. (1997).
Inhibition of ubiquitin/proteasome-dependent protein degradation by the Gly-Ala repeat domain
of the Epstein-Barr virus nuclear antigen 1. Proc. Natl. Acad. Sci. USA 94, 12616-12621.
4 Lee, C., Schwartz, M.P., Prakash, S., Iwakura, M. & Matouschek, A. (2001). ATP-dependent
proteases degrade their substrates by processively unraveling them from the degradation signal.
Mol. Cell 7, 627-637.
5. Kenniston, J.A., Baker, T.A. & Sauer, R.T. (2005). Partitioning between unfolding and release
of native domains during ClpXP degradation determines substrate selectivity and partial
processing. Proc. Nati. Acad. Sci. USA 102, 1390-1395.
6. Tian, L., Holmgren, R.A. & Matouschek, A. (2005). A conserved processing mechanism
regulates the activity of transcription factors Cubitus interruptus and NF-kappaB. Nat. Struct.
Mol. Biol. 12, 1045-1053.
72
7. Martin, A., Baker, T.A. & Sauer, R.T. (2008). Protein unfolding by a AAA+ protease: critical
dependence on ATP-hydrolysis rates and energy landscapes. Nat. Struct. Mol. Biol. 15, 139-145.
8. Wojtkowiak, D., Georgopoulos, C. & Zylicz, M. (1993). Isolation and characterization of
ClpX, a new ATP-dependent specificity component of the Clp protease of Escherichia coli. J.
Biol. Chem. 268, 22609-22617.
9. Gottesman, S., Clark, W.P., de Crecy-Lagard, V. & Maurizi, M.R. (1993). ClpX, an alternative
subunit for the ATP-dependent Clp protease of Escherichia coli. Sequence and in vivo activities,
J Biol. Chem. 268, 22618-22626.
10. Gottesman, S., Roche, E., Zhou, YN. & Sauer, R.T. (1998). The ClpXP and ClpAP proteases
degrade proteins with C-terminal peptide tails added by the SsrA tagging system. Genes Dev. 12,
1338-1347.
11. Siddiqui, S.M., Sauer, R.T. & Baker, T.A. (2004). Role of the protein-processing pore of
ClpX, an AAA+ ATPase, in recognition and engagement of specific protein substrates. Genes
Dev. 18, 369-374.
12. Martin, A., Baker, T.A. & Sauer, R.T. (2008). Pore loops of the AAA+ ClpX machine grip
substrates to drive translocation and unfolding. Nat. Struct. Mol. Biol. 15, 1147-1151.
13. Martin, A., Baker, T.A. & Sauer, R.T. (2008). Diverse pore loops of the AAA+ ClpX machine
mediate unassisted and adaptor-dependent recognition of ssrA-tagged substrates. Mol. Cell 29,
73
441-450.
14. Kim, Y.I., Burton, R.E., Burton, B.M., Sauer, R.T. & Baker, T.A. (2000). Dynamics of
substrate denaturation and translocation by the ClpXP degradation machine. Mol. Cell 5, 639648.
15. Kenniston, J.A., Baker, T.A., Fernandez, J.M. & Sauer, R.T. (2003). Linkage between ATP
consumption and mechanical unfolding during the protein processing reactions of an AAA+
degradation machine. Cell 114, 511-520.
16. Aubin-Tam, M.E., Olivares, A.O., Sauer, R.T., Baker, T.A. & and Lang, M.J. (2011). Singlemolecule protein unfolding and translocation by an ATP-fueled proteolytic machine. Cell 145,
256-267.
17. Maillard, R.A., Chistol, G., Sen, M., Righini, M., Tan, J., Kaiser, C.M., Hodges, C., Martin,
A., & Bustamante, C. (2011). ClpX(P) generates mechanical force to unfold and translocate its
protein substrates. Cell 145, 459-469.
18. Singh, S. K., Grimaud, R., Hoskins, J. R., Wickner, S. & Maurizi, M. R. (2000). Unfolding
and internalization of proteins by the ATP-dependent proteases ClpXP and ClpAP. Proc. Natl.
Acad. Sci. USA 97, 8898-8903.
19. Heim, R., Prasher, D.C., & Tsien, R.Y. (1994). Wavelength mutations and posttranslational
autoxidation of green fluorescent protein. Proc. Natl. Acad. Sci. USA 91, 12501-12504.
74
20. Cubitt, A.B., Heim, R., Adams, S.R., Boyd, A.E., Gross, L.A., & Tsien, R.Y. (1995).
Understanding, improving and using green fluorescent proteins. Trends Biochem. Sci. 20, 448455.
21. Orm6, M., Cubitt, A.B., Kallio, K., Gross, L.A., Tsien, R.Y & Remington, S.J. (2006).
Crystal structure of the Aequorea victoria green fluorescent protein. Science 273, 1392-1395.
22. Chattoraj, M., King, B.A., Bublitz, G.U. & Boxer, S.G. (1996). Ultra-fast excited state
dynamics in green fluorescent protein: multiple states and proton transfer. Proc. Nat!. Acad Sci.
USA 93, 8362-8367.
23. Kent, K.P., Childs, W. & Boxer, S.G. (2008). Deconstructing green fluorescent protein. J
Am. Chem. Soc. 130, 9664-9665.
24. Stoner-Ma, D., Melief, E.H., Nappa, J., Ronayne, K.L., Tonge, P.J. & Meech, S.R. (2006).
Proton relay reaction in green fluorescent protein (GFP): Polarization-resolved ultrafast
vibrational spectroscopy of isotopically edited GFP. J Phys. Chem. B. 110, 22009-22018.
25. Pedelacq, J.D., Cabantous, S., Tran, T., Terwilliger, T.C. & Waldo, G.S. (2006). Engineering
and characterization of a superfolder green fluorescent protein. Nat. Biotechnol. 24, 79-88.
26. Cabantous, S., Terwilliger, T.C. & Waldo, G.S. (2005). Protein tagging and detection with
engineered self-assembling fragments of green fluorescent protein. Nat. Biotechnol. 223,102107.
75
27. Reid, B.G. & Flynn, G.C. (1997). Chromophore formation in green fluorescent protein.
Biochemistry 36, 6786-6791.
28. Stoner-Ma, D., Jaye, A.A., Ronayne, K.L., Nappa, J., Meech, S.R., & Tonge, P.J. (2008). An
alternate proton acceptor for excited-state proton transfer in green fluorescent protein; rewiring
GFP. J. Am. Chem. Soc. 130, 1227-1235.
29. Kenniston, J.A., Burton, R.E., Siddiqui, S.M., Baker, T.A. & Sauer, R.T. (2004). Effects of
local protein stability and the geometric position of the substrate degradation tag on the
efficiency of ClpXP denaturation and degradation. J Struct. Biol. 146, 130-140.
30. Tsutsui, Y, Dela Cruz, R., & Wintrode, P.L. (2012). Folding mechanism of the metastable
serpin al-antitrypsin. Proc. Natl. Acad Sci. USA. 109, 4467-72.
31. Peterson, C.N., Levchenko, I., Rabinowitz, J.D., Baker, T.A., & Silhavy, T.J. (2012). RpoS
proteolysis is controlled directly by ATP levels in Escherichia coli. Genes Dev. 26, 548-53.
32. Koodathingal, P., Jaffe, N.E., Kraut, D.A., Prakash, S., Fishbain, S., Herman, C. &
Matouschek, A. (2009). ATP-dependent proteases differ substantially in their ability to unfold
globular proteins. J Biol. Chem. 284, 18674-18684.
33. Flynn, J.M., Levchenko, I., Seidel, M., Wickner, S.H., Sauer, R.T. & Baker, T.A. (2001).
Overlapping recognition determinants within the ssrA degradation tag allow modulation of
proteolysis. Proc.Natl. Acad Sci. USA 11, 10584-10589.
76
34. Herman, C., Prakash, S., Lu, C.Z., Matouschek, A. & Gross, C.A. (2003). Lack of a robust
unfoldase activity confers a unique level of substrate specificity to the universal AAA protease
FtsH. Mol. Cell 11, 659-669.
35. Kwon, A.R., Trame, C.B. & McKay, D.B. (2004). Kinetics of protein substrate degradation
by HslUV. J. Struct. Biol. 146, 141-147.
36. Choy, J.S., Aung, L.L. & Karzai, A.W. (2007). Lon protease degrades transfer-messenger
RNA-tagged proteins. J. Bacteriol. 189, 6564-6571.
37. Gur, E. & Sauer, R.T. (2008). Recognition of misfolded proteins by Lon, a AAA+ protease.
Genes Dev. 22, 2267-2277.
38. Neher, S.B., Sauer, R.T. & Baker, T.A. (2003). Distinct peptide signals in the UmuD and
UmuD' subunits of UmuD/D' mediate tethering and substrate processing by the ClpXP protease.
Proc. Natl.Acad Sci. USA 100, 13219-13224.
39. Bolon, D.N., Wah, D.A., Hersch, G.L., Baker, T.A. & Sauer, R.T. (2004). Bivalent tethering
of SspB to ClpXP is required for efficient substrate delivery: a protein-design study. Mol. Cell
13, 443-449.
40. Baird, G.S., Zacharias, D.A. & Tsien, R.Y. (1999). Circular permutation and receptor
insertion within green fluorescent proteins. Proc. Natl. Acad Sci. USA 96, 11241-11246.
77
41. Levchenko, I., Seidel, M., Sauer, R.T. & Baker, T.A. (2000). A specificity-enhancing factor
for the ClpXP degradation machine. Science 289, 2354-2356.
42. Norby, J.G. (1988). Coupled assay of Na+,K*-ATPase activity. Methods Enzymol. 156, 116119.
43. Burton, R.E., Siddiqui, S.M., Kim, Y.I., Baker, T.A. & Sauer, R.T. (2001). Effects of protein
stability and structure on substrate processing by the ClpXP unfolding and degradation machine.
EMBO J 20, 3092-3 100.
78
Chapter 3
Nucleotide binding and conformational switching in the hexameric
ring of a AAA+ machine
This chapter contains experiments from B.M. Stinson*, A.R. Nager*, S.E. Glynn*, K.R.
Schmitz, T.A. Baker, and R.T. Sauer (2013) in submission (*-equal contribution). I developed
cCoMET and worked with Stinson to develop nCoMET. I conducted all experiments involving
cCoMET and disulfide-linked constructs and crystallized the 6L ClpX 6 structure.
79
Abstract
ClpX, a AAA+ ring homohexamer, uses the energy of ATP binding and hydrolysis to power
conformational changes that unfold and translocate target proteins into the ClpP peptidase for
degradation. In multiple crystal structures of ClpX rings, some subunits adopt nucleotideloadable conformations, others adopt unloadable conformations, and each class exhibits
substantial variability. Using mutagenesis of individual subunits in covalently tethered hexamers
together with new fluorescence methods to assay the conformations and nucleotide-binding
properties of these subunits, we demonstrate that dynamic interconversion between loadable and
unloadable conformations is required to couple ClpX ATP hydrolysis to mechanical work. ATP
binding to different classes of subunits initially drives staged allosteric changes, which set the
conformation of the ring to allow hydrolysis and linked mechanical functions.
Reciprocal
subunit switching between loadable and unloadable conformations subsequently isomerizes or
resets the configuration of the nucleotide-loaded ring, which may produce a power stroke or
obviate stalling after malfunction.
80
Introduction
In all branches of life, AAA+ molecular machines harness the energy of ATP binding and
hydrolysis to degrade, disaggregate, and secrete proteins, to remodel macromolecular complexes,
to transport nucleic acids, and to drive vectorial transport along microtubules.' A central
unsolved challenge in dissecting the mechanisms of these complicated multi-protein machines is
to determine how ATP interacts with different subunits and coordinates the conformational
changes that ultimately power machine function. Although many different models of this
orchestration are possible, most proposals in the literature depend upon multiple untested
assumptions (for example, ref. 2 and 3), and the paucity of methods to test specific models has
limited our understanding of these machines.
ClpXP is an ATP-dependent protease that consists of a self-compartmentalized barrel-shaped
peptidase (ClpP) and a hexameric-ring AAA+ unfoldase (ClpX), which recognizes, unfolds, and
translocates protein substrates into an internal ClpP chamber for degradation (for review, ref. 4).
Insight into ClpX function has come from biochemistry, protein engineering, and singlemolecule biophysics. For example, translocation of a peptide degron through the axial pore of the
ClpX ring drives substrate unfolding, and processive translocation can occur against substantial
resisting forces.5- 8 Although ClpX is a homohexamer, it is asymmetric and nucleotides fail to
bind some ClpX subunits and bind other subunits with different affinities. 9 A single active
subunit in the hexameric ring is sufficient to power mechanical unfolding and translocation,
subunit-subunit communication appears to be important for controlling and coupling ATP
hydrolysis to function.10 Because processive ClpXP proteolysis of a single polypeptide can
81
require hundreds of ATP-binding and hydrolysis events,11 understanding how these nucleotide
transactions are coupled to mechanical work is a critical aspect of mechanism.
Crystal structures of the hexameric ClpX ring reveal two basic classes of subunits. 12 In four
loadable (L) subunits, the orientation of the large and small AAA+ domains creates a binding
cleft in which nucleotide can contact each domain, the intervening hinge, and a neighboring
subunit (Fig. 3.1A). The exact structure and properties of these binding sites can differ depending
upon the position in the hexamer and bound nucleotide. By contrast, these sites are destroyed in
two unloadable (U) subunits by a hinge rotation that reorients the flanking domains. In the
known hexamer structures, these subunits are arranged in an L/U/L/L/U/L pattern with
approximate two-fold symmetry (Fig. 3.1B). For all subunits, the large AAA+ domain packs
against the small AAA+ domain of the counterclockwise subunit in a conserved rigid-body
fashion, and crosslinks across these interfaces are compatible with full ClpX function.13 Thus,
the functional ring can be viewed as six rigid-body units connected by hinges (Fig. 3.1B).
Nucleotide-dependent changes in hinge geometry provide a potential way to couple ATP binding
and hydrolysis in one subunit to conformational changes in neighboring subunits. However,
direct evidence for such allosteric conformational changes is lacking, and it is not known if L and
U subunits interconvert and/or if a 4:2 ratio of L:U subunits is maintained during function.
82
A
domain (B)
smll
domain (A)
Cr
E=
+b ATP hdro1sis
y
large domain
y
Usubunit
hinge
hL
domain (A)
Lsubunit following
ATP hydrolysis
oc e
L subunits
subunit
subunit
6L
subunits
domain (F)
B
subunits
Lclockwise
small domains
ATPgS
non-switching models
m
non-witchig
4L2
hexamer
axis
+LU
D
hinge
0.3
4
ATP
mallnJ~rigid-by
domain
uni
domain
large
o da ~~0.1-
ATP
ADP
~inouaI
orsubunit
00
ATI gS'
~
hin
44L:2U
ATP
w
ADP
0.2
8
CJI-4!
switching models
-4
5L:1 U
O24§iMA%
0.2 4mMAT
mixed
0 0 10
200 400 600 800
[nucleotide] (pM)
(M)
1000
Figure 3.1 ClpX structure
(A) Nucleotide binds between the large and small AAA+ domains of a ClpX subunit and also contacts the neighboring large
domain of the clockwise subunit. Each small domain and the large domain of the neighboring clockwise subunit form a rigidbody unit. The structure shown is from an ATPyS-bound hexamer. (B) In most crystal structures, the ClpX ring consists of four L
subunits, which can bind nucleotide, and two U subunits, which cannot bind nucleotide. The ring consists of six rigid-body units.
Changes in the conformations of the hinges that connect the large and small domain of each subunit are responsible for major
conformational changes in the hexameric ring. (C) After aligning the large AAA+ domains of each crystallographically
independent subunit in eight crystal structures of E. coli ClpX hexamers, the attached small domains were represented by a vector
corresponding to a single c helix (residues 332-343). Although the vectors in U subunits are very different then those in L
subunits, substantial variations within each subunit class are evident. (D) Addition of ATP, ATPyS, or ADP reduced the
fluorescence of a W-W-WTT pseudo hexamer (0.3 pM) to a level expected for a 5L: lU hexamer or a mixture of 4L:2U and 6L
hexamers. The lines are fits to a Hill equation (Y = a - b*[ATP]n/([ATP]" + (K p)")). The same final fluorescence value was
observed in the presence of 4 mM ATP with or without ClpP (1 RM) and a titinn substrate (10 piM) unfolded by reaction with
fluorescein-5-maleimide (inset). (E) Models of ClpX function in which ClpX subunits retain their U or L identities (non
switching models) or subunits adopt U and L conformations at different points in the reaction cycle (switching models).
To address these questions, we have developed and applied assays for subunit-specific nucleotide
binding (nCoMET) and conformational changes (cCoMET), where CoMET signifies coordinated
metal energy transfer, a method that relies on short-distance quenching of a fluorescent dye by a
transition-metal ion (tmFRET; ref. 14). Our results show that nucleotide binding to ClpX
83
subunits with tight and weak affinities allosterically alters the conformations of neighboring
subunits in a stepwise fashion, support a model in which L and U subunits in the ClpX ring
dynamically interconvert during the functional cycle, and suggest that nucleotide binding
stabilizes a ring with five L-like subunits, reminiscent of structures observed in the AAA+ rings
of the El helicase and 26S proteasome.15'
6
The operating principles and tools developed here
should be broadly applicable to the study of other AAA+ machines and multimeric assemblies.
Results
The family specific N domain of ClpX, which is not required for machine function,''7 ' 8 was
deleted in the variants used here. ClpX variants were typically expressed from genes encoding
two, three, or six subunits connected by polypeptide tethers, as linking subunits in this way
allows ClpP-mediated degradation of ssrA-tagged substrates, does not affect pseudo-hexamer
formation, and permits mutations to be introduced into specific subunits for functional analysis
or introducing fluorescent probes.10
New crystal structures
AN
In previous structures with and without nucleotide, a covalently tethered ClpXA
trimer, with
ATPase-defective E185Q (E), R370K (R), and E185Q/R370K (ER) mutations in an E-E-ER
pattern, crystallized as a pseudo hexamer with an L/U/L/L/U/L arrangement of subunits.12 We
obtained six new pseudo-hexamer structures. Most had the L/U/L/L/U/L pattern, including an EE-ER with bound ATPyS, E-R dimers, W-W-R trimers (W is a wild-type subunit), W-W-W
84
trimers, and W-W-W trimers with bound ADP (Appendix B Table 1). Thus, the L/U/L/L/U/L
arrangement is not a consequence of bound nucleotide, the number of covalent tethers, or the
presence of specific mutations. However, one W-W-W structure revealed a, L/L/L/L/L/L or 6L
arrangement of subunits (Appendix B Table 1).
We aligned the large domains of each subunit from eight crystal structures and represented the
small domains by a vector corresponding to one helix (Fig. 3.1 C). As expected, there were two
major categories, corresponding to L and U conformations, but substantial variations were
evident in each class. For example, compared to single reference vectors, the average angular
variability was 16 +/- 70 (maximum 270) among L subunits and 18 +/- 130 (maximum 450)
among U subunits, highlighting the variability in the conformations of individual subunits that
comprise the ClpX ring. This variability allowed us to build a plausible 5L:1U ring structure
using subunits taken from the observed 4L:2U and 6L structures.
Evidence supporting 4L:2U and 5L:1U subunit arrangements
Contact between two rhodamine-family dyes, such as TAMRA, results in quenching that
displays an all-or-none character.19 To address which arrangements of ClpX subunits might be
populated in solution, we produced a W-W-WTT trimer in which TT designates TAMRA dyes
attached to K330C in the small domain and D76C in the large domain of the third subunit (the
TAMRA-labeled
protein was active in ATP hydrolysis and supported ClpP-mediated
degradation; Appendix B Fig. 1). Modeling showed that the TAMRA dyes were close enough for
85
contact quenching in L subunits but were >25 A apart in U subunits.
Compared to an
unquenched control, we would therefore expect -33% fluorescence for a population of 4L:2U
hexamers, -16%
fluorescence for a population of 5L:lU hexamers, and no substantial
fluorescence for a population of 6L hexamers. In the absence of nucleotide, the fluorescence of
the W-W-WT pseudo hexamer was -28% of a control sample of the same protein denatured in 3
M urea, as expected for a predominant population of 4L:2U structures (Fig. 3. 1D). Addition of
saturating concentrations of different nucleotides resulted in a decrease to ~16% of the
unquenched control, consistent with a 5L:1 U arrangement. These results are also consistent with
a roughly equal mixture of 4L:2U and 6L hexamers at saturating nucleotide, but we consider this
possibility less likely, as similar final fluorescence values at saturating ATP were also obtained
when ClpX was bound to ClpP or was translocating an unfolded substrate into ClpP for
degradation (Fig. 3.1D inset). Thus, if 4L:2U and 6L ClpX species were equally populated at
saturating nucleotide, this equilibrium would have to be independent of the identity of the bound
nucleotide and independent of ClpP binding and ATP-fueled protein degradation.
A test of subunit switching
In principle, L subunits and U subunits could maintain their conformations during the chemomechanical cycle of ClpX, or switch dynamically. Fig. 3.1E shows several non-switching and
switching models, but many more are possible, including variations in which ATP hydrolysis
occurs sequentially or probabilistically among the L subunits that bind nucleotide and/or models
in which subunit switching is not coupled to ATP hydrolysis.
86
One way to determine if L and U subunits maintain their conformations during ATP hydrolysis
and protein unfolding by ClpX is to reduce ATP-binding affinity to one or a few subunits in the
hexamer and test for effects on the ATP concentrations required for these activities. The logic is
that non-switching models would allow low-affinity subunits to adopt U conformations, and the
concentration of ATP required for activity should not be significantly altered. By contrast, if
nucleotide must bind to each subunit in the ring at some point in a cycle, as required by most
switching models, then substantially higher concentrations of ATP would be required for
equivalent levels of ClpX activity. To weaken nucleotide affinity, we engineered V78A/179A
substitutions (hereafter called VI) to truncate the wild-type side chains and reduce packing with
the adenine base of ATP (Fig. 3.2A). As anticipated, substantially higher concentrations of ATP
were required to support ATP hydrolysis and ClpP-mediated protein degradation by the VI
homohexamer compared to the parental enzyme (Fig. 3.2B; Appendix B Fig. 2A).
To test the predictions of a 4L:2U non-switching model, we constructed a covalently tethered WVI-W trimer, which ran as a pseudo hexamer in gel filtration (Appendix B Fig. 2) and bound 3-4
ADPs in isothermal titration calorimetry (ITC) (Appendix B Fig. 2A-C).
This value is similar to
the stoichiometry of ATP binding to ATPase-defective ClpXE18 5Q hexamers. 9 Notably, protein
unfolding, ATP hydrolysis, and ATPyS hydrolysis by W-VI-W required -10-fold higher
ATP/ATPyS concentrations to achieve activities comparable to the W-W-W parent (Fig. 3.2C-E;
Table 3.1). We also introduced the El 85Q mutation into the VI subunit to generate W-VIE-W, as
this mutation should only affect activity if nucleotide binds the VIE subunit. Importantly, WVIE-W had much lower ATP-hydrolysis activity than W-VI-W (Fig. 3.2D). These results are
87
inconcsistent with a non-switching 4L:2U model and suggest that robust activity requires ATP
occupancy and hydrolytic activity by at least one VI or VIE subunit in these pseudo hexamers.
To test the 5L:lU non-switching model, we constructed a W-W-W-W-W-VIE enzyme. Again,
higher concentrations of ATP were required for function compared to the parental W-W-W-WW-W enzyme and the maximal activity of W-W-W-W-W-VIE was reduced (Fig. 3.2F). These
results suggest that ATP binds to the single VIE subunit of this pseudo hexamer, a result
inconsistent with non-switching models.
Additional pseudo hexamers with five wild-type
subunits and one subunit with multiple mutations affecting ATP binding and/or hydrolysis also
required increased concentrations of ATP for function (Appendix B Fig. 2D).
A
B
C
3
6
Vi,
A V78
179
*D
ATPgS
w
1
W-W-W
10K
2m
W
-
W6
1,0 10
4-
E 2.
.0w-VI-w
10
1000
[ATP] (mM)
D
E
F
6.
-
200-
1000
1.5.
7
300-
100
[ATP] (mM)
W-W-W-W-W-W
C
N4.
+
1.0-
WW
W-W-
100-w'w
.a-
W-W-W
0.5.
2.
W-VIE-W
W-W-W-W-W-VIE
W-VI-W
1
10 10 1000
[ATP] (mM)
1
10
100 1000
[ATPgS] (mM)
88
10
100 1000
[ATP] (mM)
Figure 3.2 VI mutations alter the ATP dependence of ClpX function
(A) The side chains of Val78 and le79 contact the adenine base of bound nucleotide. (B) The VI mutations (V76A/179A) in a
non-tethered hexamer (V1 6 ) increased the concentration of ATP required to support degradation of cp7-CFP-ssrA (20 pM) by
ClpP 14 (0.9 pM) compared to an otherwise identical hexamer (W6 ) without the VI mutations. The V16 and W 6 concentrations
were 0.3 piM. (C) ATP dependence of the unfolding of cp7-CFP-ssrA (10 pM) by the W-W-W and W-VI-W ClpX variants (1
pseudo hexamer). This experiment and those in panels D and E contained 10 mM C02+ and no Mg 2+. (D) ATP dependence of
the rate of ATP hydrolysis for W-W-W, W-VI-W, or W-VIE-W (0.3 pM pseudo hexamer). (E) ATPyS dependence of the rate of
1iM
ATPyS hydrolysis for W-W-W (0.1 pM pseudo hexamer) and W-VI-W (2 pM pseudo hexamer). (F) ATP dependence of the
degradation of cp7-CFP-ssrA (10 pM) by ClpP (0.3 jM) supported by the W-W-W-W-W-W or W-W-W-W-W-VIE ClpX
variants (0.1 jM pseudo hexamer).
An assay for subunit-specific nucleotide binding
We developed nCoMET to measure nucleotide binding to specific subunits in the ClpX ring.
Mg2+ and nucleotide normally bind ClpX together, but Co2+ substitutes for Mg2+ and can quench
fluorescence of a nearby Oregon Green dye with a calculated F6rster radius (Ro) of ~13 A (Fig.
3.3A). We attached this dye to ClpX residue M363C in just one subunit of the W-VI-W trimer,
which should position the dye ~10-15 A from the metal in the nucleotide-binding site of the
same subunit. By contrast, the closest neighboring nucleotide-binding site was -35 A away, a
distance at which nucleotide/Co2+ binding would cause less than 1% quenching. Although Mg2+
is normally required for ClpX function, Co2+ supported ATP/ATPyS hydrolysis and protein
unfolding by W-VI-W and W-W-W (Appendix B Fig. 3A-C), although it inhibited the ClpP
peptidase (Appendix B Fig. 3D).
Indeed, the assays shown in Fig. 3.2C, 3.2D, and 3.2E
contained Co2+ but no Mg2+ to allow comparisons of ClpX function and nucleotide binding under
the same conditions. Modification of M363C with the Oregon Green dye was also compatible
with ClpX function (Appendix B Fig. 3E).
To allow nCoMET binding assays to different subunits, we generated and purified W*-VI-W, WVI*-W, and W-VI-W* pseudo hexamers, where the asterisk indicates the subunit containing the
89
nCoMET probes
Titration experiments were performed using ATP (Fig. 3.3B), ATPYS (Fig.
3.3C) or ADP (Appendix B Fig. 3F). In each case, binding to the rightmost W* subunit was tight
(Kapp 2-14 pM), binding to the leftmost W* subunit was weaker (Kapp 60-90 pLM), and binding to
the VI* subunit was even weaker (Kapp 430 pM) and undetectable (Fig. 3.3D; Table 3.1A). For
ATP and ATPyS, Kapp values are a function of the rate constants for nucleotide association and
dissociation, the rate constant for hydrolysis, and the rate constant for ADP dissociation, and thus
are larger than the true KD. Nevertheless, ADP and ATP/ATPyS bound subunits over similar
concentration ranges (Fig. 3.3D), a finding we return to in the Discussion.
A
adjacent subunit
same subunit
1.0-
nCoMET binding assay
-
CO
00.5.
440UnR]
42000 10
iM i
0.0 :
2
30
40
M
ATPgS
ATP
ADP
4001-
metal-to-dye distance (A)
80-
B
C 0
W-VI-W*
6-
60
0.6
C
W-V-W*
40
M 0.4-
0.4-
~0
_
02
W*-V-W
Cr
W*-VI-W
W-Vl*-W
C..
0.0
1
10
100 1000
[ATPgS] (mM)
1
10
0 El
w-vi-w* w*--iw w-v1*-w
W-Vl*-W
Y
-
-""
200
100 1000
[ATPI (mM)
Figure 3.3 nCoMET detects nucleotide binding to specific subunits
2
(A) In the nCoMIET assay, nucleotide binds ClpX and coordinates a Co + ion, which quenches the fluorescence of an Oregon-
Green dye attached to M363C in the small AAA+ domain of a ClpX subunit. Given the calculated Ro, quenching would be
2
substantial from nucleotide/Co 2+ bound in the same subunit but minimal from nucleotide/Co + bound in neighboring subunits. (B)
assayed by nCoMET. The lines are
W-VI-W*
and
W-VI*-W,
of
W*-VI-W,
hexamers
- (C) ATPyS and ATP binding to pseudo
fits to a hyperbolic equation (Y = a*[nuc]/([nuc] + Kapp)). Kappvalues are listed in Appendix B Table 2. The panel-B experiment
contained 0.1 pM nCoMET variants (pseudo hexamer equivalents). The panel-C experiment contained 0.5 pM pseudo hexamer
and 10 pM cp7-CFP-ssrA. (D) Summary of fitted Kapp values for nucleotide binding to different classes of subunits.
90
For ATPyS hydrolysis by W-VI-W, Km (470 pM) was similar to Kapp (430 riM) for nCoMET
binding to the VI subunit in W-VI*-W (Table 3.1A). For ATP hydrolysis by W-VI-W in the
presence of protein substrate, Km (-4 mM) was 20-fold greater than the ATP concentration
required for nCoMET binding to the weakest wild-type subunits in W*-VI-W or W-VI-W* (Fig.
3.3D; Table 3.1A). Thus, as expected for a subunit-switching model, the high ATP/ATPyS
concentrations required to support W-VI-W function appear to reflect binding of these
nucleotides to the VI subunit.
A. nCoMET binding assays.
variant
nucleotide
ADP
W-W-W*
ATP
ATPyS
ADP
ATP
ATPyS
W-VI-W*
ADP
W*-VI-W
ATP
ATPyS
ADP
ATP
ATPyS
W-VI*-W
hyperbolic fit
Kapp (pM)
R2
single
25 ± 4
0.990
4
single
double
single
single
single
amplitudes were 0.18 0.04 (tight) and 0.28 0.04 (weak);
amplitudes were 0.12 ± 0.02 (tight) and 0.08 ± 0.02 (weak).
a
2; 66 ±18
79± 8
17 13; 170 ±77
17 2
1.8 0.1
14 1
13 1
10 2
3± 1; 88 ±60
68 14
59 8
not detected
not detected
430 200
doublea
single
double
single
single
single
single
b
amplitudes were 0.11
0.07 (tight) and 0.23
0.998
0.996
0.998
0.994
0.998
0.996
0.998
0.982
0.996
0.984
0.992
0.947
0.07 (weak);
B. Activity assays
variant
W-W-W
nucleotide
assay
Km or Ku2 (pM)
ATP
hydrolysis
unfolding
hydrolysis
230 3
410 23
9 1
3900± 350
3300 ±210
470± 37
ATPyS
W-VI-W
ATP
ATPyS
hydrolysis
unfolding
hydrolysis
Hill constant
1.3 ±0.1
1.4 ±0.1
1.3 ±0.2
not determined
1.6 ±0.1
0.9± 0.1
Table 3.1
Nucleotide-interaction parameters obtained from nCoMET assays of binding (A) or activity assays (B).
91
R2
0.999
0.998
0.998
0.998
0.999
0.999
Subunit-specific conformational changes
We developed cCoMET to assay how the conformations of specific subunits in the ClpX ring
were altered by nucleotide binding. In this variation of tmFRET, 4 quenching is determined by
the distance between a TAMRA dye attached to K330C in the small AAA+ domain and a Ni 2 +
ion bound to an a-helical His-X 3-His motif in the large domain of the same subunit (Fig. 3.4A).
The His-X 3 -His site was engineered by introducing N72H and D76H mutations in combination
with H68Q to remove an alternative Ni2+ binding site, and nitrilotriacetic acid (NTA) was
included in assays to minimize Ni 2+ binding to nucleotides. The calculated Ro for the Ni2+_
TAMRA pair is ~14 A, and thus strong quenching should occur in L subunits (modeled distance
8-15 A) and weak or no quenching should be observed in U subunits (19-31 A). The mutations
and modifications required for this assay did not affect ATP hydrolysis or substrate degradation
(Appendix B Fig. 4A,B).
We introduced the cCoMET modifications (§) to generate W§-VI-W, W-Vl§-W, and W-VI-W§
pseudo hexamers and assayed fluorescence quenching using conditions differing from nCoMET
only in the divalent metals. Changes in cCoMET quenching were determined as a function of
ATP with protein substrate present (Fig. 3.4B), as a function of ATPyS with no substrate (Fig.
3.4C), and as a function of ADP without substrate (Appendix B Fig. 4C). As nucleotide
increased, quenching increased from an initial value to a plateau that depended on the variant and
nucleotide. Several conclusions follow from these assay results. (a) Nucleotide binding to the
hexamer generally decreased the average distance between the dye and the Ni2 + in W§ and in VI§
92
subunits. (b) Conformational changes in both the leftmost and rightmost W subunits occurred at
nucleotide concentrations that resulted in only the rightmost subunits, which have strong
nucleotide affinity, being substantially occupied in nCoMET assays (Fig. 3.4B,C). Thus, the first
nucleotide-binding events cause allosteric changes in both bound (rightmost W) and unbound
(leftmost W) subunits of the ring, possibly altering the hinge conformations and rigidifying the
domain-domain interfaces in L subunits. (c) At low ATP concentrations, where the high-affinity
(tight) sites were bound in nCoMET assays, the Ni2+-dye distance in VP subunits increased (Fig.
3.4C). At substantially higher concentrations, where the low-affinity (weak) W subunits were
also occupied and binding to VI subunits was expected, the VP Ni 2 +-dye distance decreased
substantially.
Thus, nucleotide binding to weak W and VI subunits stabilizes a different
conformation (possibly a 5L:lU ring) than binding to tight subunits. (d) Although there were
small differences depending on the nucleotide, the magnitude of maximal quenching in VP and
W subunits was generally similar, suggesting that these subunits spend roughly comparable
amounts of time in L and U conformations because of switching.
We also performed cCoMET and nCoMET assays using W-W-W and W-W-W* constructs
(Appendix B Fig. 3G, 4D). In these cases, signal amplitudes were similar to those observed with
the W-VI-W proteins, but averaging over all types of subunits precluded rigorous determination
of interaction constants for individual classes.
93
A
cCoMET conformational assay
loadable subunit
quenching
C.00. 5 -
RO 14.5
C,
unloadable subunit (no quenching)
U
10
20
30
metal-to-dye distance (A)
B 0 .5
tight
C
weak
0.5-
tight
weak
W,-VI-W
D 0.4.9
o 0.4-
I
W-VI-WI
WS-VI-W:
WV-~
W-Vis-W
Cr 0.3 -
0.3-
W-V -W
0.2
0.2)
1
10
100
1
1000o
10
100
1000
[ATP] (mM)
[ATPgS] (mM)
Figure 3.4 eCoMET detects conformational changes in specific subunits
(A) The cCoMET assay measures quenching of a TAMRA dye attached to K330C in the small domain of a ClpX subunit by a
Ni2+ ion bound to an cc-helical His-Xr-His motif in the large domain. Based on crystal structures, L subunits should display
moderate quenching and U subunits should display little or no quenching. (B) ATPyS-dependent changes in the conformations of
subunits containing cCoMET probes ()were assayed for pseudo hexamers (0.3 pM). Lines are fits to a Hill equation with Kapp
and n values of 22 +/- I pM and 1.4 +-0.1 (W§-VI-W), 70 +/- 2 gM and 1.5 +/- 0.1 (W-VI§-W), or 25 +/- I pM and 1.3 +/- 0.1
(W-VI-W§). The dashed lines marked "tight" and "weak" represent Kapp values for wild-type subunits from W-VI-W nCoMET
experiments, which were performed under very similar conditions. (C) ATP-dependent cCoMET conformational changes using
0.3 pM pseudo hexamers and 10 pM cp7-CFP-ssrA. Lines are either fits to a Hill equation. Kapp and n values of 45 +/- 4 pM and
0.8 +/- 0.1 for W§-VI-W, and 21 +/- I pM and 1.2 +/- 0.1 for W-VI-W§, or a hyperbolic plus Hill equation for W-VI§-W
(hyperbolic phase, Kapp = 12 +/- 18 pM, amplitude = -0.03; Hill phase, Kapp = 230 +/- 13 gM, n = 2.5 +/- 0.3, amplitude = 0. 15).
Locking subunits in the L conformation prevents unfolding and degradation
To prevent L-U switching, we engineered disulfide bonds to lock a single subunit or two
opposed subunits of a pseudo hexamer in the L conformation. A T147C cysteine (TC) in the
large domain of one subunit can form a disulfide with a E205C cysteine (EC) in the large domain
of the clockwise subunit, only when the TC subunit adopts the L conformation. We constructed a
94
linked trimer with the EC mutation in the first subunit and the TC mutation in the third subunit
and a linked hexamer with the EC mutation in the first subunit and the TC mutation in the sixth
subunit. Disulfide formation between two trimers forms a covalently closed hexameric ring, with
subunit 3 of each trimer in the L-lock conformation (called double L-lock; Fig. 3.5A). Disulfide
formation in the linked hexamer covalently closes the ring and locks subunit 6 in the L
conformation (called single L-lock; Fig. 3.5A).
We purified these variants, catalyzed oxidation with copper phenanthroline, and confirmed that
disulfides were formed by non-reducing SDS-PAGE (Fig. 3.5B), although 10-15% of the single
L-lock protein remained reduced. The disulfide-bonded L-lock enzymes hydrolyzed ATP (Fig.
3.5C) at maximum rates comparable to the reduced enzymes but faster than a W-W-W control.
In the presence of ClpP, which binds both variants (Appendix B Fig. 5), the disulfide-bonded
enzymes showed poor or undetectable degradation of a folded protein substrate (cp7-GFP-ssrA)
compared to the reduced proteins or W-W-W (Fig. 3.5D). Similarly, the disulfide bonded double
L-lock enzyme failed to degrade an unfolded substrate (titin -ssrA with core cysteines modified
by fluorescein-5-maleimide) in the presence of ClpP, but degraded this substrate well following
reduction (Fig. 3.5E). Following oxidation, the single L-lock enzyme displayed some
degradation of the unfolded substrate (Fig. 3.5E), but at a level similar to the amount of reduced
protein remaining (Fig. 3.5B). ClpX rings topologically closed by formation of different
disulfide bonds are fully active.' 3 Thus, blocking L->U switching in one or two subunits of the
ClpX ring uncouple ATP hydrolysis from efficient substrate unfolding and translocation.
95
A
00
double L-lock single L-lock
B
disulfide bond
red
ox
red
ox
S-S
bonded
)
1
s-S
bonded
reduced
reduced double
L-lock
C
single
L-lock
cp7-GFP-ssrA
degradation
D
E
CM-titin127-ssrA
degradation
double
L-locc
ii
I,!
double
L-Iock
/
single
L-loCk
sit
U
I'
/
Figure 3.5 Effects of L-lock disulfides on ClpX function
(A) Cartoon depiction of L-lock disulfide bonds between cysteines in the adjacent large AAA+ domains of subunits in the L
conformation. (B) Non-reducing SDS-PAGE of the single and double L-lock proteins before and after treatment with 20 mM
copper phenanthroline. For the single L-lock enzyme, -15% of the sample was not disulfide bonded after oxidation. (C) Maximal
rates of ATP hydrolysis were determined by Michaelis-Menten experiments using the indicated ClpX variants (0.3 gM pseudo
hexamers). Km values were 100 ± 19 pM (W-W-W), 2700 ± 350 gM (disulfide bonded double L-lock), 1250 ± 190 pM (reduced
double L-lock), 440 ± 45 pM (disulfide bonded single L-lock), and 490 ± 120 pM (reduced single L-lock). (D) Rates of
degradation of cp7-GFP-ssrA (10 sM) by the indicated ClpX variants (0.3 ptM pseudo hexamers) and ClpP 14 (0.5 pM) in the
presence of ATP (4 mM) and an ATP-regeneration system. (E) Rates of degradation of titin12 7-ssrA (20 jM) denatured by
reaction with fluorescein-5-maleimide by the indicated ClpX variants (1 pM pseudo hexamers) and ClpP1 (1 M) in the
presence of ATP (10 mM) and an ATP-regeneration system.
Subunit communication and ATP hydrolysis
To determine if ATP hydrolysis requires communication between different nucleotide-binding
sites, we designed mini-ClpXAN, which contains two rigid-body units that encompass a single
nucleotide-binding site (Fig. 3.6A). Mini-ClpXAN eluted as a pseudo dimer in gel-filtration
experiments (Fig. 3.6B), as expected because the surface buried between different rigid-body
units is small' 2 and thus a mini-ClpXAN pseudo hexamer would not be stable. Importantly, mini96
ClpXAN hydrolyzed ATP at approximately 75% of the basal rate of a ClpXAN hexamer on a per
site basis, and the KM for ATP was similar for mini-ClpXAN and W-W-W (230 pM; Table 3.1B),
and hydrolysis activity was abolished by an E185Q mutation in the single nucleotide-binding site
in mini-ClpXAN (Fig. 3.6C). Thus, hexamer formation and communication between different
nucleotide-binding sites are not required for ATP binding and hydrolysis. However, mini-ClpXAN
did not bind ClpP (Fig. 3.6D) and exhibited no evidence of interacting with or unfolding protein
substrates (Fig. 3.6C,E), suggesting that hexamer formation is required for these activities.
A
B
AN subunit
tether
hinge
N
C
rigid body
670
tether
158 44 17 kDA
I
I I
C
rigid body
ATP-
miniCipXAN
I
-Jrngid
hbinding
body
C
I
elution volume (mL)
E
D
CIpXAN
-;r 30 0.03.
20
ILILI
Co C
- 4
10
0.02
0.01
0substrate
CIpP
- + - - +
-+-
- - +
-+-
- - +
0
3
6
9
12
[CIpX variant] (pM)
15
0.
CIpxa
C
N
Figure 3.6 ATP hydrolysis by a variant with one binding site does not support function
(A) Cartoon depiction of the domain structure of mini-ClpXAN. This variant contains two rigid-body units but only one complete
ATP-binding site. (B) Mini-ClpXAN (loading concentration 12 pM) chromatographed at a position expected for a pseudo dimer
on a Superose 6 gel-filtration column. The elution position of a single-chain ClpXAN hexamer is shown for reference. (C) MiniClpXAN hydrolyzed ATP at approximately 75% of the basal rate of single-chain ClpXAN when activities were normalized for the
number of active sites. Unlike ClpXAN, ATP hydrolysis by mini-ClpXAN was not stimulated by protein substrate (10 FM V15Ptitin-27 -ssrA) or repressed by ClpP 14 (3 pM). (D) The rate of cleavage of a fluorescent decapeptide (15 pM) by ClpP 14 (50 nM) in
the presence of increasing concentrations of single-chain ClpXAN or mini-ClpXAN were determined in the presence of 1 mM
ATPyS. (E) Unfolding rates of photo-cleaved Kaede-ssrA (5 jM) by single-chain ClpXAN or mini-ClpXAN (2 pM) were
determined in the presence of 2.5 mM ATP and a regeneration system.
97
Discussion
Setting and resetting the configuration of the C1pX ring
Our results support a model in which (i) the conformation of the hexameric ClpX ring must
initially be "set" in a staged nucleotide-binding reaction to allow ATP hydrolysis, and (ii) the
configuration of the nucleotide-loaded ring can subsequently be "reset" via isomerization that
involves reciprocal U@L subunit switching. A plausible model for these ring-setting and ringresetting reactions is shown schematically in Fig. 3.7. In this model, ATP binding to ClpX
causes one subunit in a 4L:2U ring to switch from a U->L conformation, resulting in a 5L:lU
ring. The variations in subunit conformation that we observe in ClpX hexamers with 4L:2U and
6L structures make a 5L:lU ring plausible, and structures with five L-like subunits and one Ulike subunit are observed in the AAA+ rings of the El helicase and 26S proteasome.
4L:2U
Setting the Ring
Resetting the Ring
Nucleotide loading
ATP hydrolysis and
mechanical function
4L:2U
15,16
5L:1U
Figure 3.7 Model for ring-setting and ring-resetting reactions
The left side of the figure shows the proposed ring-setting reaction. In the absence of nucleotide, ClpX hexamers mainly adopt a
4L:2U ring structure with two U subunits (triangles), two L subunits with high affinity for nucleotide (dark gray circles), and two
L subunits with low affinity for nucleotide (light gray circles). ATP (red oval) binding to the high-affinity subunits changes the
conformations of both classes of loadable subunits (designated by color darkening), but this intermediate is inactive in ATP
hydrolysis. Subsequent ATP binding to the low-affinity L subunits stabilizes a 5L: IU configuration of the ring, which is active in
98
ATP hydrolysis, but the new L subunit (white circle) has very low nucleotide affinity and is generally empty. The right side of
the figure shows ring-resetting reactions (gray arrows), in which reciprocal L@U subunit switching changes the configuration of
the ring. These isomerization reactions are required for efficient protein unfolding and translocation of substrates by ClpX.
Resetting of the ring could occur sequentially, with subunit switching proceeding clockwise or counterclockwise in rotary order
around the perimeter, or stochastically, with subunits switching to any other isomer in a probabilistic fashion. Many variations of
the model shown are possible.
The key feature of the ring-setting reaction is the stepwise conformational changes, driven
initially by ATP binding to high-affinity subunits and subsequently by binding to lower-affinity
subunits, rather than the exact structure of the nucleotide-loaded ring. Our results indicate that
only the fully loaded ring can perform mechanical work and show that little or no ATP hydrolysis
occurs in the partially loaded intermediate with just the high-affinity ATP sites occupied. Thus,
the ring-setting reaction minimizes ATP hydrolysis before it can be coupled to functional work.
However, the mini-ClpXAN pseudo dimer, which has just one nucleotide-binding site, hydrolyzes
ATP efficiently with a KM similar to wild-type hexamers. Thus, structural constraints imposed
by formation of the hexameric ring appear to keep the high-affinity subunits in the partially
loaded intermediate in a conformation poorly suited for ATP hydrolysis.
In our model, ring resetting involves reciprocal U@L subunit switching in the nucleotide-loaded
ClpX hexamer, resulting in new isomers or ring configurations (Fig. 7). More complicated
models, allowing reciprocal switching between 4L:2U rings or incorporating 6L intermediates in
5L: lU switching, are possible but are not necessary to account for current results. As we discuss
below, resetting the ring appears to be required for the mechanical functions of the ClpX
machine. Ring resetting could be linked to ClpX activity in two ways. First, reciprocal switching
could be a normal consequence of ATP hydrolysis and be required to generate a power stroke in
the ClpX ring. By this model, ATP-hydrolysis events would lead to coupled L 4U and U+L
99
switching events. Alternatively, ring resetting could provide a way to escape stalling in situations
where machine function might otherwise be compromised. These situations could include failed
protein-unfolding attempts, as fewer than 1%of ATP-hydrolysis events result in ClpX unfolding
of very stable proteins, 1 or binding of an inappropriate nucleotide. For example, in the
sequential and tightly coupled F1 ATPase, binding of ADP instead of ATP stalls the motor for
very long periods until the ADP dissociates (see Chapter 1). Notably, we find that ADP binds
ClpX subunits over concentration ranges similar to or lower than ATP, and thus ring
isomerization through subunit switching might allow ClpX to eject improperly bound ADP and
avoid prolonged stalling. Ring resetting could also explain how ClpX hexamers with only one or
two hydrolytically active subunits escape stalling as they unfold and translocate protein
substrates. 1
Evidence for subunit switching
Our results support L*U switching during ClpX function. For example, we find that decreasing
the nucleotide affinity of one or two subunits in a hexamer increases the concentration of
ATP/ATPyS required for ClpX activity, a result expected for a switching model in which
nucleotide must bind to every subunit in the hexameric ring at some time during the multiple
enzyme cycles required for protein unfolding and translocation. In a non-switching model, by
contrast, a low-affinity subunit could simply assume a U conformation and would only alter the
ATP dependence of enzyme function by restricting the number of active ring configurations,
which would produce much smaller effects than those we observe. Moreover, when we crosslink
one or two ClpX subunits in the L conformation using disulfide bonds, these enzymes hydrolyze
100
ATP and bind ClpP but did not unfold protein substrates and/or translocate unfolded substrates
into ClpP efficiently.
L@U switching is also supported by the cCoMET results. Notably, each type of subunit in the
W-VI-W pseudo hexamer displayed roughly similar quenching at saturating concentrations of
nucleotide. Thus, in a nucleotide-loaded ring, the average Ni 2 -TAMRA distance must be similar
in high-affinity and low-affinity W subunits as well as in VI subunits. Based on crystal
structures, moderate quenching is expected for loadable subunits and very low quenching for
unloadable subunits. Thus, our cCoMET results suggest that no single type of subunit in W-VIW adopts just an L conformation or just a U conformation, but rather that all subunits sample
both conformations. This result is inconsistent with non-switching models, in which the VI
subunits would preferentially adopt the U conformation. For ATP and ATPyS, hydrolysispowered L@U switching could explain these results. For ADP, however, subunit switching
would need to be thermally driven. Indeed, ongoing single-molecule studies using the assays
described here show subunit switching in the presence of saturating ADP and should help clarify
whether subunit switching and/or ATP hydrolysis proceed in a strictly sequential or probabilistic
fashion during protein unfolding and translocation by ClpX (Chapter 4).
Is subunit switching important for other AAA+ machines? The answer is not known, but we note
that certain AAA+ modules in the covalent ring of dynein are observed in conformations
amenable to nucleotide binding in some crystal structures and in non-binding conformations in
101
other structures (see Chapter 1).
Structural and functional classes of ClpX subunits
In our crystal structures, substantial conformational variations occur within the general L and U
classes of ClpX subunits. Indeed, this variation could allow each L subunit in a 4L:2U or 5L: 1U
ring to have different nucleotide-binding properties. 9 Subsequent studies with the hexameric
HslU and PAN AAA+ unfoldases revealed similar nucleotide-binding categories and ratios.3,2 0
Our current studies support multiple classes of nucleotide-binding sites, and suggest a basic ring
pattern of [weak-empty-tight-weak-empty-tight] sites, with the proviso that empty sites may bind
nucleotide very weakly. Based on our nCoMET and cCoMET studies, ATP binding to tight
subunits alters the conformations of weak subunits, whereas ATP binding to weak subunits alters
the conformations of empty subunits. These allosteric effects can be viewed as being propagated
to the subunit clockwise from the bound subunit, probably through the shared rigid-body unit.
Because the conformational changes observed in different classes of subunits occur over
different ranges of nucleotide, allosteric models in which there are just two conformations of the
ClpX ring can be rejected.
Although our results suggest that ATP must bind to every subunit in the ClpX ring at some point
in the reaction cycle, previous studies showed that R-W-E-R-W-E rings, which contain just two
hydrolytically active subunits, displayed ~30% of the wild-type unfolding and translocation
activity. 10 In combination, these results demonstrate that ATP binding is functionally important
102
even when a bound ClpX subunit cannot hydrolyze this nucleotide. Our model provides a role
for ATP binding to the four non-hydrolytic subunits in the R-W-E-R-W-E variant in setting and
maintaining the hydrolytically active conformation of the ring.
Using CoMET assays to study multimeric proteins
CoMET assays are based upon tmFRET14 and provide information about nucleotide binding and
conformation in the ClpX ring that could not have been acquired by traditional methods. Because
the half-maximal distance for quenching of a fluorescent dye by a transition metal ion is very
short (10-15 A) relative to pairs of standard FRET dyes (30-100 A), CoMET-based assays are
ideally suited to measure ligand binding to specific subunits or conformational changes within or
between subunits in a multimer. Additionally, because a CoMET pair requires a single
fluorophore, it should be possible to design protein complexes with multi-color CoMET pairs to
make simultaneous measurements of ligand binding and/or conformational changes in either
population or single-molecule experiments, thereby probing how the activities of different
subunits are coordinated. For homohexamers, like ClpX, the ability to covalently link subunits
allows nCoMET or cCoMET probes to be placed in specific subunits. However, many AAA+
machines are hetero multimers (DNA clamp loader, Rpti-6 ring of the 26S proteasome, Mcm 2 -7
helicase, etc.) or naturally consist of linked AAA+ modules (dynein), which should make
CoMET assays in these systems straightforward. The sensitivity of cCoMET to small
conformational differences and its use of an a-helical His-X 3-His motif would also be ideally
suited for testing of models that involve changes in helix-helix packing registration.
103
Experimental procedures
Materials
PD buffer contained 25 mM HEPES-KOH (pH 7.5), 100 mM KCl, 10% (v/v) glycerol, and 0.5
mM EDTA. IEXA buffer contained 20 mM HEPES (pH 7.8), 150 mM KCl, 10% (v/v) glycerol,
and 1 mM EDTA. GF buffer contained 50 mM Tris (pH 7.0), 300 mM KCl, and 10% (v/v)
glycerol. ATP (Sigma), ADP (Sigma), ATPyS (Roche), and GTP (Sigma) were dissolved in PD
buffer, and adjusted to pH 7.0 by addition of NaOH.
Proteins
Unless noted, all ClpX variants were derived from E. coli ClpXAN (residues 61-423), contained
the Cl 69S mutation to remove an accessible cysteine, and were constructed by PCR and purified
generally as described.'o,22 During purification of variants with reactive cysteines, buffers were
degassed, argon sparged, and contained 0.5 mM EDTA to minimize oxidation. Buffer exchange
and desalting steps were performed using a PD10 column (GE Healthcare).
ClpX variants used for nCoMET experiments initially contained an N-terminal His6 -SUMO
domain. After Ni -NTA affinity chromatography (Qiagen), these variants were exchanged into
IEXA buffer, and the H6 -SUMO domain was cleaved by incubation with equimolar Ulpl
protease for 2 h at room temperature. Cleavage was confirmed by SDS-PAGE, and the mixture
was chromatographed on a MonoQ column (GE Healthcare) using a gradient from 150 to 500
mM KCl. Fractions containing the nCoMET variant were incubated with Oregon Green 488
104
Maleimide (Invitrogen; 3 equivalents for each cysteine) for 30 min at room temperature, and a
final purification step on a Superdex S-200 column (GE Healthcare) equilibrated in GF buffer
was performed.
ClpX variants used for cCoMET initially contained a TEV-cleavable C-terminal His6 tag. After
Ni 2+-NTA affinity chromatography, these variants were exchanged into PD buffer, incubated
with TAMRA-5-maleimide (1.5 equivalents for each cysteine) for 30 min at room temperature,
DTT (1 mM) was added to quench the reaction, and excess dye was removed by desalting.
Proteins were incubated with equimolar TEV protease for ~1 h at room temperature to remove
the His6 tag, and a final purification step was performed using a Superdex S-200 column
equilibrated in PD buffer. Disulfide-bond formation in L-lock variants was performed as
described. 13
The cp7-CFP-ssrA substrate was generated from cp7-GFP-ssrA23 by PCR incorporation of the
Y66W, A206K, and N1461 mutations and initially contained a cleavable N-terminal His 6 tag.
ep7-CFP-ssrA was purified by Ni-NTA2+ affinity chromatography, exchanged into IEXA buffer,
incubated with 10 units of PreScission protease (GE Healthcare) for 2 h at room temperature to
remove the His6 tag, and the sample was diluted 5-fold into water and purified on a MonoQ
column using a gradient from 30 to 500 mM KCl.
Crystallization and structure determination
105
Crystallized proteins included a tethered E-R dimer and tethered W-W-W, W-W-R and E-E-ER
trimers, where ER designates E185Q/R370K subunits (Appendix B Table 3.1). These proteins
did not contain the C169S mutation. Polypeptide tethers were twenty (T 20) or zero (To) residues
and compatible with ClpX function.' 3 Variants were crystallized at room temperature by
hanging-drop vapor diffusion after mixing 1 ptL of well solution with 1 ptL of protein solution
(~40 [M pseudo hexamer). The composition of well solutions are listed in Appendix B Table
3.1. The nucleotide-bound form of W-W-W with T2 0 tethers was obtained using a previously
established soaking procedure.12 To obtain the structure of E-E-ER bound to ATPyS, nucleotidefree crystals were soaked in 3.4 M sodium malonate, 75 mM sodium acetate (pH 4.8), 4 mM
ATPyS, and 4 mM MgCl 2 for approximately 2 h. All crystals were cryo-protected by coating in
Paratone-N (Hampton Research) and flash-frozen in liquid nitrogen. Data were collected at the
24-ID-C beamline of the Advanced Photon Source, Argonne National Laboratories. Unit-cell
volumes indicated that crystals with the space groups P212 121 and P6 3 contained six and two
subunits in the asymmetric unit, respectively. Data collection and refinement statistics are listed
in Appendix B Table 3.1.
Diffraction data were integrated and scaled using HKL2000,2 TRUNCATE,2 MOSFLM, 2 6 and
SCALA. 2 5 The structures of the large and small AAA+ domains of F. coli ClpX12 were used as
molecular-replacement search models in PHASER. 2 7 Each domain was placed independently to
avoid bias towards previously observed conformations. Manual model building and real-space
refinement were carried out in COOT.28 Rigid-body refinement, TLS refinement, and grouped
atomic displacement parameter refinement were performed using PHENIX. 2 9 Individual large
106
and small AAA+ domains were defined as rigid bodies and TLS groups, with either individual
domains or individual residues defined as atomic displacement parameter groups based on the
resolution and quality of the data.
For crystals soaked in nucleotide, examination of calculated mFO-DFc maps revealed strong
peaks of positive density in the nucleotide-binding pockets of loadable subunits. The bound
nucleotides
in the ATPyS-soaked
W-W-W (T 20 ) structure
could not be determined
unambiguously, probably because of mixed occupancy by nucleotide and sulfate ions, and ADP
provided the best fit for the electron density. For the E-E-ER crystal soaked in sodium malonate
and ATPyS, the bound nucleotides were unambiguously modeled as ATPyS.
Structural validation was performed using MolProbity. 30 Superposition of structures was carried
out using LSKQAB. 2 5 The atomic coordinates for all structures have been deposited in the
Protein Data Bank, with accession codes listed in Appendix B Table 3.1.
Fluorescence assays
Unless noted, assays were performed at room temperature in PD buffer, with nucleotides, metals,
and substrate added as required. All nCoMET assays contained 10 mM CoCl 2 , and nucleotidedependent
changes
in
fluorescence
were
measured
using
a
PTI
QM-20000-4SE
spectrofluorimeter (excitation 500 nm; emission: 520 nm). Signal contamination from
107
fluorescent substrate, when present, was less than 2%. Addition of 1 mM GTP did not result in
nCoMET quenching (Appendix B Fig. 8), confirming specificity. Ni2+ can also be used for
nCoMET and supports ClpX function, but Co2+ results in a larger Ro when paired with the
Oregon-Green dye. nCoMET assays were limited to nucleotide concentrations below 2 mM, as
higher concentrations appeared to alter fluorescence indirectly by binding Co2+
All cCoMET assays contained 500 pM Ni 2 +-NTA and 10 mM MgCl 2 . Ni 2 +-NTA (KD ~1 nM) was
used to bind the oa-helical His 7 2-X 3-His 7 6 motif because it had weaker affinity than Ni2+ for free
and ClpX-bound nucleotides. Titration of ATP against W-W-W in the absence of Ni2+-NTA
resulted in no quenching (Appendix B Fig. 9A). Titration of Ni2+-NTA against W-W-W resulted
in ~25% quenching with an affinity of 160 ± 45 pM (Appendix B Fig. 9B). cCoMET can detect
small conformational changes. For example, when a cCoMET pair was placed within a single
rigid-body unit, ATP binding to high-affinity subunits resulted in quenching (Appendix B Fig.
10). Conformational changes of 3-4 A caused by a minor change in the rigid body or side-chain
movements that alter the dye-quencher distance, could account for these results.
The degree of TAMRA-TAMRA contact quenching for W-W-WTT was determined relative to a
control experiment with W-W-WTT in 3 M urea (a W-W-WTT sample degraded with elastase
showed the same fluorescence as the sample in urea).
Biochemical assays
108
ATP hydrolysis was measured by a coupled assay. 31 ATPyS hydrolysis was analyzed by ion-pair
chromatography on a Shimadzu Class-VP HPLC. Time points were taken by quenching the
reaction with excess DTT (when Co2 was present) or by the addition of one-quarter volume of
50% trichloroacetic acid. Samples were directly loaded onto a Waters Delta-Pak C18 column
(300 A, 5 pM, 3.9 x 150 mm) equilibrated in 30 mM triethylammonium phosphate (pH 5.5) and
eluted isocratically. Hydrolysis was measured by quantifying the ADP and ATPyS peaks
(retention times ~6 min and ~15 min, respectively) and calculating the amount of ADP formed as
a fraction of total nucleotide. ATP-hydrolysis rates measured by the coupled assay and the HPLC
assay were within experimental error.
The rate of cp7-CFP-ssrA unfolding was calculated from the initial rate of loss of fluorescence
measured with an SF-300X stopped flow instrument (KinTek). Pre-mixed ClpX and substrate
were rapidly mixed with an equal volume of ATP solution and substrate fluorescence was
monitored (excitation 435 nm; emission 495 nm long-pass filter).
ClpP peptide-cleavage assays were performed using a succinyl-Leu-Tyr-AMC dipeptide or a
decapeptide containing an N-terminal 2-aminobenzoic acid fluorophore and nitro-tyrosine
quencher at residue nine as described.
109
Acknowledgements
We thank C. Lukehart, C. Drennan, R. Grant, and T. Schwartz for help and discussions. This
work was supported by NIH grant AI-15706. T.A.B. is an employee of the Howard Hughes
Medical Institute. Studies using the NE-CAT beamline were supported by the NCRR
(5P41RR015301-10) and NIGMS (8 P41 GM103403-10). Use of the Advanced Photon Source at
Argonne National Laboratory was supported by the US DOE (contract DE-AC02-06CH11357).
110
References
1. Hanson, P.I., and Whiteheart, S.W. (2005). AAA+ proteins: have engine, will work. Nat. Rev.
Mol. Cell. Biol. 6, 519-529.
2. Wang, J., Song, J.J., Seong, I.S., Franklin, M.C., Kamtekar, S., Eom, S.H., and Chung, C.H.
(2001). Nucleotide-dependent conformational changes in a protease-associated ATPase HsIU.
Structure 9, 1107-1116.
3.
Smith, D.M., Fraga, H., Reis, C., Kafri, G., and Goldberg AL. (2011). ATP binds to
proteasomal ATPases in pairs with distinct functional effects, implying an ordered reaction cycle.
Cell 144, 526-538.
4.
Baker, T.A., and Sauer, R.T. (2012) ClpXP, an ATP-powered unfolding and protein-
degradation machine. Biochim. Biophys. Acta. 1823, 15-28.
5. Martin, A., Baker, T.A. and Sauer, R.T. (2008a). Diverse pore loops of the AAA+ ClpX
machine mediate unassisted and adaptor-dependent recognition of ssrA-tagged substrates. Mol.
Cell 29, 441-450.
6. Martin, A., Baker, T.A. and Sauer, R.T. (2008b). Pore loops of the AAA+ ClpX machine grip
substrates to drive translocation and unfolding. Nat. Struct. Mol. Biol. 15, 1147-1151.
7. Aubin-Tam, M.E., Olivares, A.O., Sauer, R.T., Baker, T.A., and Lang, M.J. (2011). Singlemolecule protein unfolding and translocation by an ATP-fueled proteolytic machine. Cell 145,
257-67.
8. Maillard, R.A., Chistol, G., Sen, M., Righini, M., Tan, J., Kaiser, C.M., Hodges, C., Martin,
111
A., and Bustamante, C. (2011). ClpX(P) generates mechanical force to unfold and translocate its
protein substrates. Cell 145, 459-469.
9. Hersch, G.L., Burton, R.E., Bolon, D.N., Baker, T.A. and Sauer, R.T. (2005). Asymmetric
interactions of ATP with the AAA+ ClpX 6 unfoldase: allosteric control of a protein machine. Cell
121, 1017-1027.
10. Martin, A., Baker, T.A., and Sauer, R.T. (2005). Rebuilt AAA+ motors reveal operating
principles for ATP-fueled machines. Nature 437, 1115-1120.
11. Kenniston, J.A., Baker, T.A., Fernandez, J.M., and Sauer, R.T. (2003). Linkage between ATP
consumption and mechanical unfolding during the protein processing reactions of an AAA+
degradation machine. Cell 114, 511-520.
12. Glynn, S.E., Martin, A., Nager, A.R., Baker, T.A., and Sauer, R.T. (2009). Crystal structures
of asymmetric ClpX hexamers reveal nucleotide-dependent motions in a AAA+ proteinunfolding machine. Cell 139, 744-756.
13.
Glynn, S.E., Nager, A.R., Baker, T.A., and Sauer, R.T. (2012). Dynamic and static
components power unfolding in topologically closed rings of a AAA+ proteolytic machine. Nat.
Struct. Mol. Biol. 19, 616-622.
14. Taraska, J.W., Puljung, M.C., Olivier, N.B., Flynn, G.E., and Zagotta, W.N. (2009) Mapping
the structure and conformational movements of proteins with transition metal ion FRET. Nat.
Methods. 6, 532-537.
15. Enemark, E.J., and Joshua-Tor, L. (2006). Mechanism of DNA translocation in a replicative
112
hexameric helicase. Nature 442, 270-275.
16. Lander, G.C., Estrin, E., Matyskiela, M.E., Bashore, C., Nogales, E., and Martin, A. (2012).
Complete subunit architecture of the proteasome regulatory particle. Nature 482, 186-191.
17. Singh, S.K., Rozycki, J., Ortega, J., Ishikawa, T., Lo, J., Steven, A.C. and Maurizi, M.R.
(2001). Functional domains of the ClpA and ClpX molecular chaperones identified by limited
proteolysis and deletion analysis. J. Biol. Chem. 276, 29420-29429.
18. Wojtyra, U.A., Thibault, G., Tuite, A. and Houry, W.A. (2003). The N-terminal zinc binding
domain of ClpX is a dimerization domain that modulates the chaperone function. J. Biol. Chem.
278, 48981-48990.
19.
Zhou, R., Kunzelmann, S., Webb, M.R., and Ha, T. (2011). Detecting intramolecular
conformational dynamics of single molecules in short distance range with subnanometer
sensitivity. Nano Lett. 11, 5482-5488.
20. Yakamavich, J.A., Baker, T.A., and Sauer, R.T. (2008). Asymmetric nucleotide transactions
of the HslUV protease. J. Mol. Biol. 380, 946-957.
21.
Hirono-Hara, Y., et al. (2001). Pause and rotation of F(1)-ATPase during catalysis. Proc.
Natl. Acad. Sci. USA. 98, 13649-54.
22. Martin, A., Baker, T.A., and Sauer, R.T. (2007). Distinct static and dynamic interactions
control ATPase-peptidase communication in a AAA+ protease. Mol. Cell 27, 41-52.
23. Nager, A.R., Baker, T.A., and Sauer, R.T. (2011) Stepwise unfolding of a P barrel protein by
the AAA+ ClpXP protease. J. Mol. Biol. 413, 4-16.
113
24. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in
oscillation mode. Methods Enzymol. 276, 307-326.
25. Winn, M.D., et al. (2011). Overview of the CCP4 suite and current developments. Acta
Cryst. D67, 235-242.
26. Leslie, A.G.W. and Powell, H.R. (2007). Processing Diffraction Data with Mosfim. Evolving
Methods for Macromolecular Crystallography, 245, 41-51.
27. McCoy, A.J., Grosse-Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C. and Read,
R.J. (2007). Phaser crystallographic software. J. Appl. Cryst. 40, 658-674.
28. Emsley, P., Lohkamp, B., Scott, W.G. and Cowtan, K. (2010). Features and development of
Coot. Acta Cryst., D66, 486-501.
29. Adams, P.D., Afonine, P.V., Bunkoczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J.,
Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., McCoy, A.J., Moriarty, N.W., Oeffner, R.,
Read, R.J., Richardson, D.C., Richardson, J.S., Terwilliger, T.C. and Zwart, P.H. (2010).
PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta
Cryst., D66, 213-221.
30. Chen, V.B. Arendall, W.B. 3 d, Headd, J.J., Keedy, D.A., Immormino, R.M., Kapral, G.J.,
Murray, L.W., Richardson, J.S., & Richardson, D.C. (2010). MolProbity: all-atom structure
validation for macromolecular crystallography. Acta. Cryst., D66, 12-21.
31. Norby, J.G. (1988). Coupled assay of Na+,K*-ATPase activity. Methods Enzymol. 156, 116119.
114
32. Lee, M.E., Baker, T.A., Sauer, R.T. (2010). Control of substrate gating and translocation into
ClpP by channel residues and ClpX binding. J. Mol. Biol. 399, 707-718.
115
Chapter 4
The stochastic mechanism of a AAA+ machine observed by singlemolecule CoMET
This chapter contains experiments from A.R. Nager*, Y Shin*, H. Manning, T.A. Baker, M.J.
Lang, and R.T. Sauer (in preparation) (* equal contribution). I developed cCoMET and
conducted experiments and analyzed data in collaboration with Yongdae Shin and Harris
Manning.
116
Abstract
Hexameric ring-shaped AAA+ ATPases harness the energy of ATP to power a variety of cellular
tasks including protein degradation, DNA translocation, and microtubule transport, but it is not
known what conformational changes these machine-like enzymes undergo or how subunits
within a hexamer coordinate their activities. ClpXP is a AAA+ protease that consists of the
hexameric ClpX unfoldase and polypeptide translocase and the ClpP compartmental peptidase.
ClpX binds a substrate by an unstructured degradation tag and then, by multiple rounds of ATPbinding and hydrolysis, unfolds and translocates the substrate into the proteolytic chamber of
ClpP. By using a novel short-distance, transition metal ion FRET technique (CoMET), we show
that a subunit within the AAA+ ClpX ring undergoes sub-nanometer conformational changes that
can be associated with ATP hydrolysis. Surprisingly, there were large conformational changes
that were thermally driven and likely uncoupled from nucleotide hydrolysis. Furthermore, the
kinetics of these conformational changes cannot be explained by strictly sequential or concerted
models, but can be described by an asymmetric hexamer with multiple subunit classes that
interconvert infrequently. We have observed similar results using two independent, singlemolecule fluorescence assays. Other AAA+ enzymes may operate by similar mechanisms.
117
Introduction
AAA+
polypeptide
translocases
are
hexameric
ATP-fueled
machines
that
remodel
macromolecular complexes, and refold, secrete, disaggregate, and degrade proteins. These
machines appear to work by a common mechanism: (1) six AAA+ subunits form a ring and bind
an unfolded polypeptide tag to loops in the central pore; (2) ATP hydrolysis causes
conformational changes that translocate the tag stepwise through the pore; and (3) attached
folded domains are repeatedly pulled against the pore until they are denatured and can be
translocated. For energy-dependent protein degradation carried out by the 26S proteasome in
eukaryotes, the Cdc48-20S and PAN-20S proteasomes in archaea, and the ClpXP, ClpAP, HslUV,
Lon, and FtsH proteases in bacteria, a AAA+ translocase threads polypeptides into the
proteolytic chamber of an associated compartmental peptidase.1 Protein translocation is a
difficult task, as it involves movement of diverse polypeptide tracks with very different chemical
properties and structural preferences. For ClpXP, the unfolding and translocation of some
substrates requires 100s of ATP-hydrolysis events. 2
ClpX is a homohexameric AAA+ motor that binds and translocates ssrA-tagged proteins into the
ClpP peptidase. The AAA+ module of each ClpX subunit consists of a large and small domain
connected by a hinge. The hinged interface between domains forms a nucleotide-binding pocket
and undergoes conformational changes in response to ATP binding and hydrolysis. These
structural changes are propagated via rigid-body interfaces to neighboring subunits in the ClpX
ring. ClpX subunits appear to play distinct roles during function, as unloadable subunits fail to
bind nucleotide and loadable subunits have different nucleotide affinities, abilities to hydrolyze
nucleotide, and mechanisms of allosteric communication with neighboring subunits (Chapter 3).
118
Some workers in the field have interpreted subunit classes as evidence for a sequential
mechanism, in which each subunit passes through different conformations and nucleotide-bound
states in a strictly ordered fashion that depends upon ATP hydrolysis. 3 If function required
subunits to hydrolyze ATP in a strict order, than a hexamer with an inactive subunit should stall.
However, mutagenesis of covalently-linked ClpX pseudo hexamers shows that a ring with one
hydrolysis-active subunit can unfold and translocate substrates, 4 suggesting some type of nonsequential mechanism must allow the active subunit to productively hydrolyze ATP rather than
being locked in an ATP-empty class, futilely waiting for the hydrolytically-dead subunits to fire.
How such a mechanism would work is not known.
Results
CoMET of a ClpX subunit
To observe conformational changes of a ClpX subunit within a hexamer, we applied a
conformational CoMET assay" that measures distance-dependent quenching of a TAMRA dye
attached to K330C in the small AAA+ domain by Ni 2+-NTA bound to an a-helical His72-X3His76 motif in the large domain of the same subunit. Covalently-linked pseudo hexamers with a
C-terminal biotin tag were bound to a streptavidin-coated flow cell and observed by total internal
reflection fluorescence (TIRF) microscopy. Initially, ClpXTAMRA was imaged without Ni 2+-NTA
to determine the unquenched fluorescence. After -10 s, 500 pM Ni 2+-NTA, 300 nM ClpP, and
nucleotide were flowed into the cell, and time-dependent changes in fluorescence were
monitored until the TAMRA dye was photobleached (Fig. 4.1A,B, additional trajectories in
appendix C Fig. 1). In the presence of saturating ATP, unquenched and quenched states were
" Coordinated Metal Energy Transfer
119
observed. The unquenched state is expected for an unloadable (U) subunit, in which a rotation at
the hinged interface blocks the nucleotide-binding pocket and separates the cCoMET pair by as
much as 31 A, a distance far greater than the calculated Fdrster radius (Ro = 14.5 A). Although
an unquenched state could represent Ni2+*NTA unbinding, this possibility is unlikely for several
reasons. First, a saturating concentration of Ni 2+-NTA was used (500 jiM; appendix C Fig. 2).
Second, the frequency and distribution of unquenched dwell times vary with nucleotide
concentration (described below). Third, a U state with similar kinetics was observed by contact
quenching of TAMRA dyes attached to K330C and D76C (appendix C Fig. 3). The different
quenched states observed by CoMET probably represent different conformations of loadable (L)
subunits, as the CoMET pair is separated by 8-15 A in loadable subunits observed in different
crystal structures. 5 In the single-molecule trajectories, the proportion of time spent in the U
conformation was 30-40% at low nucleotide concentrations and 16% with saturating nucleotide,
consistent with bulk studies that indicated a hexamer switched from a 4L:2U to a 5L:lU
configuration when multiple nucleotides were bound (U/(L+U) row in Table 4.1; appendix C Fig.
4; Chapter 3). Switching between L and U subunit classes supports a model in which subunits
classes are not fixed within the hexamer but rather interconvert in a dynamic manner.
120
A
2
+.500 pM Ni' NTA, 1 mM ADP
U class
C
0
L classes
(L*''and L**"')
D
Time (sec)
B
80
0U.
1.0
U class
0.4
L classes
(LxP and Lamma)
0
Time (sec)
Figure 4.1 Single-molecule cCoMET of a CIpX subunit
Trajectories are shown for of individual fluorescent spots with saturating (A) ADP and (B) ATPyS. 500 pM Ni 2 -NTA,
nucleotide, and 300 nM ClpP 14 were flowed in after -10 s (marked with an arrow) resulting in quenching. Over time, unquenched
states (expected for U subunits) and quenched states (expected for L subunits) were found to interconvert. Data was fit by a
Hidden Markov Model (red line). At least two types of L subunits (L*P and Lganna) were inferred by kinetic analysis (described
below). ATP trajectories are shown in appendix C, Fig. 1.
Condition
No
ATP
nucleotide
U dwell
L dwell
U/(L+U)
L motions
(sec)
4.75
7.3
0.4
none
ATPyS
(sec)
2.9
7.7
0.26
slow
30 piM
3.5
13.3
0.21
ND
Iable 4.1 Dwell times of cuoVIET transitions
For W-VIN-W, a W-I-W-W-VI-W hexamer was used with the CoMET probe in the
Fig. 4.2A. All averaged times are in seconds. ND -not determined.
121
(sec)
100 pM 1 mM
3.3
2.9
13.8
14.9
0.17
0.16
ND
1
ADP
(sec)
1 mM
1 mM
3.9
14.7
0.17
7.3
4.3
14.8
0.21
none
§ marked subunit. Dwells are diagramed in
L'#U switching is independent of nucleotide hydrolysis
In the presence of ATP, the labeled subunit in the hexameric ring slowly switched between L and
U classes and also rapidly changed between L conformations, which could represent
conformational changes of 2-4
A (called L motions).
L motions only occurred in the presence of
ATP and ATPyS. Surprisingly, however, L@U switching occurred without nucleotide and with
saturating ATP, ATPyS, ADP (Fig. 4.1A,B). In the without nucleotide and ADP cases, class
switching must be thermally driven by stochastic fluctuations in ring conformation. Indeed, the
simplest model is that class switching is thermally driven in all cases, irrespective of the presence
or type of nucleotide, because the U+L switching rate was similar under all conditions. For
example, the average U dwell for the labeled subunit was ~3-5 s under all nucleotide conditions
(Table 4.1). With increasing nucleotide concentrations, the dwell time between successive U
states increased from -7 to ~15 s, consistent with a decrease in the average number of U subunits
in the hexamer from ~2 to -1 (Table 4.1). The average time between ClpX hydrolysis events at
saturating ATP is ~0.8 sec, whereas the average U+L with saturating ATP was ~3 s. In a tightly
coupled 5L:lU model , each ATP hydrolysis event would result in a U+L switch. Thus, class
switching appears to occur too slowly to be tightly coupled with ATP hydrolysis. For ATPyS
hydrolysis, the average time between hydrolysis events is -7 s (Chapter 3). Although ATPyS
hydrolysis is ~10-fold slower than ATP hydrolysis, the U-L switching rate was similar for both
nucleotides. This result shows that switching does not occur after a fixed number of hydrolysis
events and is consistent with a model in which subunit switching is hydrolysis independent, as
expected for a thermally driven process. 6
122
L@U kinetic analysis
We analyzed the dwell-time probability distributions for U subunits (U dwell) and L subunits (L
dwell, defined as the time between the end of one U dwell and the beginning of the next U dwell;
Fig. 4.2A). At saturating ATP, the distribution of U dwells was fit well by a single exponential,
suggesting that the U->L transition occurs with a single rate-limiting step with a time constant of
3 s (Fig. 4.2B). At low nucleotide concentrations, the U-dwell distribution was best fit by a
gamma function, suggesting at least two 3-s steps, one of which is sensitive to nucleotide binding
(appendix C Fig. 5).7 The L-dwell distribution was exponentially distributed in the absence of
nucleotide, consistent with every L subunit has an equal probability of switching to a U subunit
(appendix C Fig. 7A). Intriguingly, saturating ATP changed the L-dwell distribution to an
exponential-plus-gamma distribution (Fig. 4.2C,D, appendix C Fig. 6). This exponential-plusgamma distribution suggests there are two classes of L subunits, LexP subunits that can switch to
U subunits in one step and Lgamma subunits that switch in multiple steps (appendix C Fig. 7B).
Importantly, the exponential-plus-gamma distribution is incompatible with strictly sequential
models in which a subunit must consecutively pass through multiple L states with similar rate
constants before switching to the U class (appendix C Fig. 7C). One possibility is that
nucleotide-empty L subunits can switch to the U class in one step (LexP) whereas nucleotidebound L subunits switch in multiple steps (Lganua). First, Lgamma subunits are dependent on
nucleotide. Second, if one observes CoMET of a nucleotide-empty subunit by using nucleotidebinding mutations (V78A,179A (VI) mutation; Chapter 3), one observes that a nucleotide-empty
subunit switches between U and L conformations with exponential kinetics similar to Lexp in the
presence of 1 mM ATP (appendix C Fig. 8). Taken together, these results support a model in
which there are three subunit classes that switch stochastically; U subunits, nucleotide-empty
123
L*xP
subunits, and nucleotide-occupied Lgan subunits.
B
A
U dwell
U class
8 1.0
i
0
LI.
0.3
0.4
L classes
0.2
L dwell
0
Time
0.3
I
exponential fit
0.1
0
C
U dwell
1 mM ATP
0.4
0
10
D
11
L dwell
no nucleotide
20
time (sec)
40
30
L dwell
I mM ATP
0.11
0.2
0
exponential-plus-gamma fit
exponential fit
0.1
U
D
20
40
30
time (sec)
50
60
30
40
time (sec)
50
60
Figure 4.2 Dwell time probability distributions reveal hidden L classes
(A) Diagram of a fluorescent trajectory with the U dwell and L dwells highlighted. A single L dwell can include many L motion
transitions as well as hidden transitions between L*P and gamma subunits. (B) U-dwell distribution with saturating ATP fit to a
single exponential (k = 0.3 s-1). The U dwell for saturating ADP fit well to a single exponential, whereas low nucleotide
concentrations fit to a gamma distribution (appendix C, Fig. 5). (C) L-dwell distribution without nucleotide fit to a single
exponential. (D) L-dwell distribution for saturating ATP was best fit by an exponential-plus-gamma function. Similar
distributions were observed for saturating ADP and ATPyS (data not shown; appendix C, Fig. 6).
Hydrolysis-dependent L motions
L-class subunits underwent fast hydrolysis-dependent motions. With saturating ATP, for
example, subunits exchanged between 60 and 80% quenched states with dwell times between
0.5 and 1.5 s (Table 4.1; Fig. 4.3A), comparable to the average time between ATP-hydrolysis
events (0.8 s). Moreover, with saturating ADP, there were no L motions and subunits remained
124
~60% quenched until they switched to the U state (Fig. 4.3B). At low ATP or saturating ATPyS,
the rate of exchange between quenched states slowed, and a 40% quenched states was also
observed (Table 4.1, Fig. 4.3C). Any given state (e.g., 60% quenched) may include multiple
conformations. Indeed, in a hexamer, each of five L subunits have the potential to adopt slightly
different conformations based on nucleotide state and the identities of neighboring subunits,
many of which could display similar CoMET quenching. The LexP and Lganua classes may
undergo different conformational changes in response to ATP hydrolysis, but appear similar by
our assays. However, our results show changes between L conformations at a fast hydrolysisdependent rate. Additionally, if one compares nucleotide-free and nucleotide-bound crystal
structures of ClpX, small conformational changes at the hinge of L-class subunits propagate
throughout the ring and drive larger motions of the loops involved in substrate translocation. The
fast hydrolysis-dependent L motions observed by CoMET potentially could move the central
pore loops and translocate polypeptides.
A
ATP
B
.
0.6.
0.8
Initial
0.6
State
0.4
C
ADP
0.8'
*
0.2-
0.8
.
0.6.
.0.4
0.2,
0
0.2 0.4 0.6 0.8
1
0
?
0.
0.2
0
0
ATPgS
0
0
0.2
.4 0.6 0.8
1
0
0.2 0.4 0.6 0.8
1
Final State
Figure 4.3 Transition density plots
Representation of all state transitions within (A) ATP, (B) ADP, and (C) ATPyS data sets. No nucleotide was similar to ADP
(data not shown). U-+L switches are boxed in red and L motions to lower fluorescent states are boxed in brown. The diagrams
appear mirrored because (i) U4L transitions are followed by L+U transitions, and (ii) because of the likely numerous states
within L motions, only two states are assigned within a trajectory and, for ATP and ATPyS, these states are continuously
exchanged. For ATP (panel A), L motions occur at lower fluorescent levels (0.2-0.6) compared to ATPyS (panel B; 0.4-0.6). For
ADP, L motions were rare and simple L@YU switching was the most common transition.
125
Discussion
L4#U switching and L motions are uncoupled
Slow L*U switching appears to thermally driven, whereas fast L motions are dependent on
nucleotide hydrolysis. Because L@U switching and L motions have different rate-determining
steps, these processes must be largely uncoupled (Fig. 4.4A). Consistent with two uncoupled
cycles, all quenched states are able to switch to the unquenched state and vice versa (Fig.
4.3A,B,C). More specifically, an L-class subunit can switch to a U subunit at any point of the L
motion. Because the U conformation destroys the nucleotide-binding pocket, any nucleotide
bound to an L subunit that switches - hydrolyzed or not - must be released either before or as
part of the switching reaction. Similarly, a U subunit that switches to a L conformation will
initially be nucleotide free, although the conformation of the newly formed L subunit will be
influenced by rigid-body interactions with its neighbors, some of which are likely to be
nucleotide bound. After switching to an L conformation, a subunit can resume hydrolysisdependent L motions. Thus, the L-*U-*L cycle effectively resets a subunit, first displacing
bound nucleotide and then permitting hydrolysis-dependent motions to restart.
Subunit classes
Nucleotide influences the likelihood of an L-U switch as nucleotide-empty LeP subunits switch
in one step and occupancy-dependent Lgamma subunits switch at a slower rate in two steps.
Presumably, the Lga
U switch requires breaking bonds with nucleotide, increasing the
energy barrier for switching (Fig. 4.4B). Ensemble studies suggest that the 5L: lU ClpX 6 binds
126
only 3-4 nucleotides at saturation, so one or two L-class subunits may remain nucleotide free and
have a higher probability of L-U switching. Importantly, class switching need not be sequential
or tied to ATP hydrolysis. In contrast, the Fi ATP synthase consists of three as subunits that
adopt empty, ADP, and ATP classes, and sequentially switch classes with every round of
nucleotide hydrolysis and release. In future work, it will be important to determine the effect of
an L-U switch on adjacent subunits. Ensemble studies show that subunits clockwise and
counterclockwise to a low-affinity mutant subunit assume time-averaged conformations with
different nucleotide affinities, conformations, and allosteric communication with other subunits
(Chapter 3). If the empty U subunit dictates the L class of adjacent subunits, then for each L-+>U
switch, adjacent L subunits would undergo coordinated switches to new classes of L subunits,
effectively isomerizing subunit classes for the entire hexameric ring of ClpX.
A EOo
1a
A
ATP Hydrolysis
01
0.0
c
i
00
4
._10
,0
4
o4
Ue4o,
cum---L
0
*[0J0
*nMdd430lisbund L
4
o hyd-lyels
B
011
L!xP
Swf0chb
00*
Iamm
ne
900l
5
S.o
*
o
e*0
Nudeotide
DlSissciadw
@0
weakly-towndsebuift
Figure 4.4 Two mechanical cycles
(A) Conformational changes of ClpX 6 can be explained by two mechanical cycles, thermal-driven class switching and ATPhydrolysis tied L motions. These cycles occur simultaneously. (B) There are two types of L subunits, one that can switch to the
U conformation in one step and another that switches in multiple steps. The Lgamma requires a breakdown in symmetry.
127
ClpX mechanism and other AAA+ machines
Our results suggest that thermally driven and thus stochastic L OU switching changes the class
of subunits within the ClpX ring, providing a mechanism to reset the motor if a
mechanochemical step fails. For instance, if subunits hydrolyzed ATP sequentially, then a
hexamer with an inactive subunit would stall as soon as hydrolysis by that subunit was required.
In our probabilistic switching model, ring isomerization could simply change which subunits are
poised for ATP hydrolysis. Similarly, if a AAA+ translocase could not complete a mechanical
step, either failed unfolding of a substrate or slipping on a polypeptide track, probabilistic
isomerization would allow the machine to reset. By this model, following each unsuccessful
attempt, the translocase could transiently release substrate (which might or might not dissociate),
rebind ATP, and reapply force. For the subunit that pulled on the polypeptide but failed,
switching to a nucleotide-free U conformation could release that subunit from the substrate and
permit other subunits to engage for another attempt.
By contrast, the strictly sequential
mechanism of the Fi ATP synthase stalls if a single subunit binds to the wrong nucleotide, has an
ATPase mutation, or is bound by a small molecule inhibitor. 8~0 An off pathway reset can make
an engine more resilient to failure."-1 In F1 , for example, nucleotide inhibition can be resolved
by off-pathway interactions of multiple subunits with tentoxin.1 4
Peptide translocases are part of a larger AAA+ family including microtubule transporters, DNA
helicases, and viral-packaging motors. In addition to the AAA+ fold, several of these motors
have structural features similar to ClpX 6 suggesting a common mechanism. Crystal structures of
dynein show a variable 4L:2U arrangement, suggesting that L and U classes can switch. 15'1 6 The
Rpti. 6 ring of the 26S proteasome, El helicase, and RecA-like Rho helicase form 5L:lU
128
arrangements.17-19 As stochastic L@U switching of ClpX 6 occurs in the absence of nucleotide,
switching is a thermodynamic consequence of the hexameric ring structure. In comparison,
subunit-subunit interfaces are similar to ClpX 6 for Rpti- 6, but quite different for El and Rho,
potentially permitting a different mechanism for these helicases. A further extreme is the <p29
DNA packaging motor which forms homo pentamers and, by single-molecule nanometry,
appears to have much greater coordination between subunits.20 Single-molecule fluorescence
techniques will be necessary to determine and compare the mechanisms of these diverse
machines.
Acknowledgements
We thank Adrian Olivares and Ben Stinson for helpful discussion.
129
References
1.
Sauer, R.T., & Baker, T.A. (2011). AAA+ proteases, ATP-fueled machines of protein
destruction. Annu. Rev. Biochem. 80, 587-612.
2. Kenniston, J.A., Baker, T.A., Fernandez, J.M., & Sauer, R.T. (2003). Linkage between ATP
consumption and mechanical unfolding during the protein processing reactions of an AAA+
degradation machine. Cell 114, 511-20.
3.
Smith, D.M., Fraga, H., Reis, C., Kafri, G., & Goldberg, A.L. (2011). ATP binds to
proteasomal ATPases in pairs with distinct functional effects, implying an ordered reaction cycle.
Cell 144, 526-38.
4.
Martin, A., Baker, T.A., & Sauer, R.T. (2005). Rebuilt AAA+ motors reveal operating
principles for ATP-fuelled machines. Nature 437, 1115-20.
5.
Glynn, S.E., Martin, A., Nager, A.R., Baker, T.A., & Sauer, R.T. (2009). Structures of
asymmetric ClpX hexamers reveal nucleotide-dependent motions in a AAA+ protein-unfolding
machine. Cell 139, 744-56.
6.
Burton, R.E., Baker, T.A., & Sauer, R.T. (2003). Energy-dependent degradation: linkage
between ClpX-catalyzed nucleotide hydrolysis and protein-substrate processing. Protein Sci. 12,
893-902.
7. Yildiz, A., Forkey, J.N., McKinney, S.A., Ha, T., Goldman, Y.E., & Selvin, P.R. (2003).
130
Myosin V walks hand-over-hand: single fluorophore imaging with 1.5-nm localization. Science
300, 2061-5.
8. Hirono-Hara, Y, Noji, H., Nishiura, M., Muneyuki, E., Hara, K.Y, Yasuda, R., Kinosita, K.
Jr., & Yoshida, M. (2001). Pause and rotation of F(1)-ATPase during catalysis. Proc.Nati. Acad
Sci. U.S.A. 98, 13649-54.
9.
Ariga, T., Muneyuki, E., & Yoshida, M. (2007). F1-ATPase rotates by an asymmetric,
sequential mechanism using all three catalytic subunits. Nat. Struct. Mol. Biol. 14, 841-6.
10. Bowler, M.W., Montgomery, M.G., Leslie, A.G., & Walker, J.E. (2006). How azide inhibits
ATP hydrolysis by the F-ATPases. Proc. Nati. Acad Sci. U.S.A. 103, 8646-9.
11. Subramanian, R., & Gelles, J. (2007). Two distinct modes of processive kinesin movement
in mixtures of ATP and AMP-PNP. J. of Gen. Phys. 130, 445-555
12. Thoresen, T., & Gelles, J. (2008). Processive mocement by a kinesin heterodimer with an
inactivating mutation in one head. Biochemistry 47, 9514-21.
13. Guydosh, N.R., & Block, S.M. (2006). Backsteps induced by nucleotide analogs suggest the
front head of kinesin is gated by strain. Proc. Nati. Acad Sci. U.S.A. 103, 8054-9.
14. Meiss, E., Konno, H., Groth, G., & Hisabori, T. (2008). Molecular processes of inhibition
and stimulation of ATP synthase caused by the phytotoxin tentoxin. J Biol. Chem.283, 24594-9.
131
15. Kon, T., Oyama, T., Shimo-Kon, R., Imamula, K., Shima, T., Sutoh, K., & Kurisu, G.
(2012). The 2.8 A crystal structure of the dynein motor domain. Nature 484, 345-50.
16. Carter, A.P., Cho, C., Jin, L., & Vale, R.D. (2011). Crystal structure of the dynein motor
domain. Science 331, 1159-65.
17. Lander, G.C., Estrin, E., Matyskiela, M.E., Bashore, C., Nogales, E., & Martin, A. (2012).
Complete subunit architecture of the proteasome regulatory particle. Nature 482, 186-91.
18. Enemark, E.J., & Joshua-Tor, L. (2006). Mechanism of DNA translocation in a replicative
hexameric helicase. Nature 442, 270-5.
19.
Thomsen, N.D., & Berger, J.M. (2009). Running in reverse: the structural basis for
translocation polarity in hexameric helicases. Cell 139, 523-34.
20. Moffitt, J.R., Chemla, YR., Aathavan, K., Grimes, S., Jardine, P.J., Anderson, D.L., &
Bustamante, C. (2009). Intersubunit coordination in a homomeric ring ATPase. Nature 457, 44650.
132
Chapter 5
Polarized TIRFM of ClpX rigid bodies differentiates subunits within
a hexamer
This chapter contains experiments that are being prepared for publication by A.R. Nager*, H.
Manning*, Y. Shin, T.A. Baker, M. Lang, and R.T Sauer (* designates equal contribution). I
designed the experiment, constructed materials, conducted initial experiments with Harris
Manning, and performed the initial data analysis.
133
Introduction
AAA+ ATPases are multimeric machines that disaggregate and degrade proteins, transport cargo
along microtubules, unwind DNA, and perform other cellular functions that involve mechanical
work. ATP-dependent protein degradation is carried out by AAA+ proteases, which consist of a
hexameric AAA+ unfoldase and a self-compartmentalized peptidase. AAA+ proteases include
the 26S proteasome in eukaryotes, PAN-20S and Cdc48-20S in archaea, and ClpXP, ClpAP,
HslUV, Lon, and FtsH family proteases in bacteria and eukaryotic organelles. Typically, the
AAA+ unfoldase binds a degradation tag, unfolds attached protein domains, and then
translocates the unfolded polypeptide into the peptidase for degradation (reviewed in Chapter 1).
Except for the final degradation step, all other steps require ATP binding and hydrolysis, but it is
not well understood how subunits within the AAA+ unfolding ring coordinate their activities
during function.
The ClpX unfolding ring consists of six AAA+ modules that have identical sequences but
assume different asymmetric conformations in crystal structures (Chapter 3).1 The AAA+ module
consists of a large domain, a short hinge, and a small domain (Fig. 5.1A). There are two general
classes of subunit conformations, ATP loadable (L) and unloadable (U), that differ in their hinge
structures and thus in the orientation of the large and small AAA+ domains (Fig. 5.1B). The
current paradigm is that nucleotide-dependent motions of the hinge are critical for function. By
contrast, the interface between small and large domains of adjacent subunits is relatively
invariant, forming a rigid-body unit.2 As such, a change in the hinged interface of one subunit
propagates through rigid-body interactions to adjacent subunits. Crystal structures of ClpX
hexamers show subunits arrangements in L/L/U/L/L/U and L/L/L/L/L/L configurations.
134
Nevertheless, bulk and single-molecule measurements described in Chapters 3 and 4 suggested
that ClpX adopts a 5L: lU conformation when bound to nucleotide. Structures of some hexameric
AAA+ ATPases show five L-like subunits is a staircase arrangement in which each subunit is
progressively offset along the central axis (Fig. 5.lC). 3 5 A sixth, U-like subunit, closes the
staircase (omitted in Fig. 5.lC). Based on these staircase conformations, other groups proposed
that subunits within AAA+ machines operate sequentially, with each subunit progressing through
successive positions down the staircase. Thus, we sought to determine if ClpX does form a
5L: lU staircase and if positions along the staircase move in a sequential fashion.
A
~Inge
Hng
laLarg
domains
small
domains-
Loadable
82
5 consecutive L subunits
Unloadable
Figure 5.1 Rotations at the hinge
(A) Diagram of the two-domain AAA+ module in a ClpXAN subunit. (B) Comparison of domain rotations in ClpX L and U
subunits. The large domain is shown in gray and the small domain in green (PDB: 3HTE). Adapted from Glynn et al. (2009). (C)
Cryo-EM reconstruction of 5L subunits within the Rpti1 6 ring of the 26S proteasome. The sixth subunit forms an extended U
conformation and is omitted from this image. To better view the staircase, a single tyrosine is highlighted in each subunit.
Our single-molecule studies suggest that ClpX subunits undergo two conformational cycles: a
fast cycle tied to ATP hydrolysis and a slow cycle that appears to be independent of chemical
steps (Chapter 4). During the fast cycle, L subunits interconvert between similar conformations
135
at a nucleotide-dependent rate. By contrast, for the slow cycle, two subunits appear to switch
reciprocally between L and U conformations. The dwell times between U conformations suggest
two classes of L subunits, those that can switch to U in one step and those that require multiple
steps to switch to U. In structures of 5L staircases, the U subunit marks the top and bottom of the
5L staircase.3-5 As such, one would expect that whenever an L subunit switches into the U
conformation, another subunit within the newly marked staircase boundaries must switch to an L
conformation.
We used polarized total internal reflection fluorescence microscopy (polTIRFM) to infer the
relative angles of individual rigid-body units within ClpX hexamers and show that subunits form
a staircase of angles. 6 Rigid bodies within the staircase switch position in a single hydrolysisindependent step with kinetics comparable to the U->L switch. Moreover, switching does not
occur in a strictly sequential manner in which subunits predictably pass through each position of
the staircase. Intriguingly, rigid bodies also undergo fast nucleotide-dependent conformational
changes that appear to depend on the location within the staircase. Further analysis will be
necessary to determine if staircase position dictates subunit class.
Experimental Design
The hinged interface of one ClpX subunit positions the domains of the adjacent subunits through
rigid-body interactions. For example, the rotation of a L-subunit hinge in a hexamer rotates the
adjacent small domain 400 along the Z-axis (Fig. 5.2; compare blue and green highlighted small
domains). The U-subunit hinge has an opposing rotation, rotating the connected rigid body 80900 in the opposite direction along the Z-axis (Fig. 5.2; compare green and red highlighted small
domains). The consequence of these domain rotations in 4L:2U rings is that any three adjacent
136
subunits occupy distinct angles relative to the Z-axis. In the 5L:lU structures of the Rpti- 6 ring
and El helicase, all six subunit have unique angles relative to the Z-axis. Five adjacent L
conformations result in a staircase, in which each subunit is slightly rotated from the adjacent
subunit (Fig. 5.1C). The sixth, U subunit closes the ring by a large opposing rotation. Using
polTIRFM, one can measure the change in angle relative to the Z-axis and infer the ring position
of a subunit within a hexamer.
Z-axis
U
boundL
Figure 5.2 Hinge rotations position adjacent subunits
Side view of nucleotide-free and nucleotide-bound ClpX 6 rings with the small domains of two L subunits (blue and green) and
one U subunit (red) highlighted (PDB: 3HTE, 3HWS). The Z-axis passes through the central pore of the ClpX 6 ring. To highlight
rotations, a purple arrow is extended from a common alpha helix in each small domain. The position of a subunit within the ring
has the greatest effect on the angle of the arrow. Nucleotide occupancy has a smaller effect. Most changes in angle are with
respect to the Z-axis. Adapted from Glynn et al. (2009).
The excitation of a fluorophore depends on the angle of the dye dipole (p) to the polarization of
the excitation light. This phenomenon can be exploited to determine the spatial angle of a
fluorophore. Consider this scenario, the dipole of a fluorophore is fixed on the Z-axis and
illuminated with polarized light either along the Z-axis or normal to the Z-axis (Fig 5.3A). Upon
illumination with light parallel to the fluorescent dipole, there is maximum absorption and
consequent fluorescence. However, illumination with light polarized normal to the Z-axis results
137
in no absorption. By using two polarizations, one can calculate the angle of the dipole with the
Z-axis (0) and the angle within the XY-plane (<b; Fig. 5.3B).
A
z
B
Y
Y
Parallel
Polarization
z
Normal
%Polarization
Figure 5.3 Polarized TIRF microscopy
Diagram of polTIRFM. (A) Hypothetical situation where a fluorophore (purple) is oriented with its dipole (p) perpendicular to
the glass slide. When the fluorophore is excited with polarized light parallel to the Z-axis (green), there is maximum excitation.
However, there is no excitation with light normal to the dipole (red). (B) Any dipole can be defined by two angles, the polar
angle (0) and azimuthal angle (#). With data from two polarizations, one can solve for both angles.
For the angle of a protein-attached fluorophore to reflect protein conformation, the fluorophore
must undergo minimal motions that do not reflect movement of the attached protein. For
example, if a fluorophore were attached by a single carbon-sulfur bond (e.g., fluorophoremaleimide reacted with cysteine), the fluorophore might rotate freely and sample random dipole
angles. This situation can be prevented by attaching the fluorophore to the protein using multiple
sites on the dye and protein (Fig. 5.4A). 7 ,8 Moreover, the fluorescently-labeled protein ideally
would be also be attached to the glass slide by multiple tethers, to avoid changing the orientation
of the dipole with respect to the polarized light by simple rotation around a single tethering point.
Additionally, the mechanism of attachment should give a point of reference to allow measured
angles to be related to the structure. For polTIRFM studies of myosin, actin filaments were
secured to a glass slide by numerous attachments and the movement of fluorescent myosin across
the filament defined the XY-plane to relate changes in polarized excitation to conformational
138
changes during stepping.9'10
A
B12.
Large
(subunit n+1)
0 N,
0
NH
HIN
A
~
O
10-15 A
(subunit n)
BFR
CipX
CipPlatform
Streptavidin-Blotin
Glass Slide
Figure 5.4 poITIRFM of BFR-CIpX 6
(A) A bifunctional rhodamine (BFR) dye with two cysteine-reactive alkyl halides. The alkyl halides are separated by 10-15 A.
(B) A single rigid body from a ClpX hexamer composed of a small domain (green) and the large domain of the clockwise subunit
(blue). The BFR dye was reacted with one cysteine on the small domain (K330C) and another cysteine on the adjacent large
domain (S105C). The C,-Ce distance between these cysteines is 12.5 A. (C) Diagram of BFR-ClpX 6 tethered to a glass slide by
ClpPplatfom.
I labeled a ClpX hexamer with a single bifunctional rhodamine (BFR) via covalent attachment to
two cysteines on stable loops (S105C, K330C) within neighboring domains in a single rigidbody unit (Fig 5.4B). Polarization-dependent excitation was observed for BFR-ClpX 6 , but not for
ClpX labeled with a mono-functional TAMRA dye (data not shown). I then tethered BFR-ClpX 6
to a glass slide using an asymmetric ClpP1 4 molecule (called ClpPpiatfo"). ClpP1 4 normally
consists of two identical heptameric rings. By contrast, one ring of ClpPplatform contains seven
wild-type subunits that bind ClpX through multiple interactions,"
whereas the second ring
contains seven biotinylated M5A subunits which cannot bind ClpX but can make multiple tethers
to the streptavidin coated slide.' 2 ClpP tetradecamers can be reversibly split into heptameric rings
by washing with salt at 4 'C.1'3 14 To make ClpPPiatfon, I dialyzed ClpP-M5A-His 6-biotin and an
139
excess of wild-type ClpP against 150 mM ammonium sulfate, then dialyzed the mixture into a
low-salt buffer, and finally purified the chimera by Ni2 -NTA affinity (Fig. 5.4A,B). As expected,
the activity of the resulting ClpPafo"" complex in degradation of cp7-SFGFP-ssrA was the same
with one or two equivalents of ClpX 6 , as each ClpPPlafo" bind only a single ClpX 6 (Fig. 5.4C).
By single-molecule TIRFM, attachment of BFR-ClpX 6 to the streptavidin surface was dependent
on ClpPPlatform and nucleotide (data not shown).
A
150 mM AmSQ
49or
m
B
Dialyze
Ni2 tNTA Pu
C
S
Washes
,if_t
' Fllllq
eq.CIpX
6
3,
2.5.
Elution
Input
2-
1
EM
1
- ClpPMSCblotinhistag
qw-pp
0-
P
CIpP USC
Figure 5.4 Construction of CipPPla'er-n
(A) Schematic for construction of ClpPPiato". Wild type ClpP 14 (purple) is separated into Cl P 7 rings by high salt, mixed with
ClpP7 M5A H 6-Biotin rings (red), dialyzed, and affinity purified to yield asymmetric ClpPP"" complexes. (B) SDS-PAGE of
Ni 2 -NTA purification of ClpPafor"'. Wild type and M5C ClpP can be differentiated by size as M5C ClpP has a C-terminal H6 biotin tag. The eluted complex (inset) contains a 1:1 mixture of the wild-type and mutant ClpP rings. (C) cp7-SFGFP-ssrA
degradation by CIpX 6 with ClpP variants. 1 pM ClpP 14 was incubated with 10 pM substrate, 4 mM ATP, an ATP regeneration
system, and 0.5, 1, or 2 pM ClpX6 . Wild-type ClpP 14 support higher degradation with 2 equivalents of ClpX 6 , M5C ClpP 14
supported almost no degradation, and ClpPatfo"" supported almost the same level of degradation with 1 or 2 equivalents of
ClpX6 .
Results
Polarized TIRFM experiments using BFR-ClpX 6 bound to ClpPiafo"
were performed with
saturating ATP or ATPyS, an analog that is slowly hydrolyzed by ClpX.15 For both experiments,
140
there were no time-dependent changes in emission following excitation by vertically polarized
light (informative of XY-plane, azimuthal angle *), showing that ClpX does not rotate on top of
the ClpPPiatfo". In contrast, there were time-dependent changes in fluorescence following
excitation by horizontally polarized light (informative of the Z-axis, polar angle 0, and the XYplane,
#),
suggesting that conformational changes within ClpX change the orientation of rigid
bodies with respect to the Z-axis. For this chapter, I have analyzed raw fluorescence following
excitation by horizontally polarized light. In future work, the absolute angle of fluorophores will
be calculated and used to assign subunit angles more accurately.
2
0
(I)
0
0
25
50
75
100
Time (sec)
Figure 5.5 poITIRFM trajectories and hidden Markov fits
Sample trajectory of single BFR-ClpX-ClPPP~afrm spots following excitation with horizontally polarized light (blue). Changes in
fluorescence are largely caused by changes in the polar angle (0), the angle of the dipole with respect the Z-axis. Trajectories
were fit with a hidden Markov model (red).
141
The intensities of single fluorescent spots were recorded for 120 s and fit by a hidden Markov
model (Fig. 5.5).16 The last background-fluorescence state was assumed to be a photobleaching
event and was omitted from analysis. On average, an ATP trajectory had 3.2 ± 0.9 states and an
ATPyS trajectory had 4.0 ± 0.9 states. No trajectory had more than five states (Fig. 5.6A). A low
or zero fluorescence state was the most frequently sampled and may include both bona fide
conformations and photo-blinking as the fluorophore turns off and quickly back on. The
fluorescence of fitted states was normalized by the lowest fluorescent state and plotted (Fig.
5.6B). The distribution of fluorescent states for ATP and ATPyS shows five peaks. For an
asymmetric hexamer, one would expect six discrete states, one for each rigid-body unit. In our
data, the sixth state may be a low-fluorescence state that is not discriminated from the zero
fluorescence state (marked 1) or a high-fluorescence state that is highly variable in angle
(arbitrary fluorescence units 0.45-0.65; Fig. 5.6B). Importantly, the five differentiated states were
roughly equally spaced by fluorescence units and solved angles, as one would expect for a
staircase in which each rigid-body unit is offset from the adjacent unit by a common angle.
A
B 6o0
25-
0
ATP
ATP
20-
10-
2
205-
0
,
2
3
5
4
10-
3
4
5
ii0
6
# Fluorescent States
0
0.1
0.2
0.3
0.4
0.5
0.6
State Fluorescence
Figure 5.6 ClpX rigid bodies occupy five distinct angles
(A) The number of fluorescent states detected within trajectories for ATP (black) and ATPyS (gray). (B) Within ATP (black) and
ATPyS (gray) trajectories, fitted states were normalized by the lowest fluorescence state and the resulting arbitrary fluorescence
units plotted. The five groups of states are marked by red numbers.
142
The dwell times for each state were determined. Because the low-fluorescence state may include
photo-blinking, transitions to and from this state were omitted from analysis. The dwell times for
all analyzed states fit moderately well to single exponentials with time constants of 1.99 ± 0.09 s
for ATP and 2.11 ± 0.11 s for ATPyS (Fig. 5.7A). Moreover, in the presence of ATP, the dwell
times for a single state (0.1 fluorescence; marked 2 in Fig. 5.6B), fit moderately well to a single
exponential with a dwell time 2.05 ± 0.16 s (Fig 5.7B). For all of the distributions, there were a
larger than expected number of very short dwell times (<0.3 seconds) for a single exponential
distribution (Fig. 5.7A,B). Because the dwell times of fluorescent states were similar for ATP and
ATPyS, it is likely that these conformations reflect the slow hydrolysis-independent cycle of
ClpX. Indeed, the dwell times were similar to those observed for the U+L and the fastest L-U
switch in the single-molecule studies described in Chapter 4.
A
B
Dwell time distribution of
all states above background
Dwell time distribution of
0.1 fluorescence state
ATPyS
-
AT,01sate
ATPExponential
Ft
Density
0
1
2
3
4
5
6
7
8
9
10
Time (seconds)
0
1
2
3
4
5
6
7
8
9
10
Time (seconds)
Figure 5.7 One-step dwell-time distributions
Dwell time distributions for ATP (black) and ATPyS (gray). (A) Dwell times for transitions between all fluorescent states above
the lowest state (peak 1 in Fig. 5.6B). The red line is an exponential fit of the ATP data. (B) Distribution for all transitions from a
low fluorescent state (peak 2 in Fig. 5.6B).
Discussion
In past TIRFM studies, we observed that a ClpX hexamer seems to assume a 5L:1U
143
conformation with L subunits stochastically switching to the U conformation during function.
The L->U switch was independent of ATP hydrolysis and either occurred in one or multiple
steps. Because some L subunits only switch to the U conformation after multiple steps, this
suggests a hierarchy of subunit classes where some L subunits must first switch to a different
class of L conformation before switching to the U conformation. However, past techniques have
been unable to differentiate L subunit classes because all L conformations resulted in relatively
similar fluorescent signals. We designed a polTIRFM experiment that measures the angle of a
subunit relative to the central axis of ClpP and differentiates at least five subunit positions. As the
angle of each position was roughly equally spaced, our results support a 5L staircase, where
consecutive L subunits are progressively offset (Fig. 5.8A). In the 4L:2U architecture observed in
crystal structures, the dimer of trimers symmetry would result in only 3 unique angles with
respect to the Z-axis. Similarly, a symmetric 6L conformation would have only one unique angle
(Chapter 3).
17
Subunits switched between fluorescent states in a single rate-determining step that
did not depend of the nucleotide-hydrolysis rate with kinetics similar to the U4L switch
observed by cCoMET and TT quenching (Chapter 4). This result suggests that when one subunit
switches to the U conformation, all other subunits switch to new ring positions.
In cryo-electron microscopy of the 26S and PAN-20S proteasomes, the AAA+ ring was
occasionally tilted on top of the compartmental peptidase (Fig. 5.8B,C).
8,19
This may be a cryo-
EM artifact, a low-resolution perspective of a 5L staircase, or a true tilt on top of the peptidase.
For the 26S proteasome, the tilts were too small to account for the 90 or greater change we
observe by polTIRFM (0 = 00 to 2.50 for Xenopus 26S; 00 to 50 or 00 to 110 for Drosophila
26S).18,20 Moreover, tilting has not been observed for ClpXP (Fig 5.8D), ClpAP or HslUV and
144
may be a unique characteristic of AAA+ unfoldases interacting with the asymmetric 20S core
particle. 2 1-2 6 Additionally, tilting would change the spatial relationship of ClpX 6 to ClPPPaotfo,
but not the relationship of individual ClpX subunits. As such, tilting would change all measured
angles by a constant value and would not affect comparisons between angles. Thus, a tilt
between ClpX and ClpP would not explain the discrete states we observe by polTIRFM or affect
comparisons of these states.
A
Ring Structure
B
Rpt
C
Drosophila 26S
Lid
L idorange)
Tilt
archael PAN-20S
PAN
20S
Flat
D
CipX
E.coli CIpXP
Flat
CIpP
20S
Flat
Slight Tilt
Large Tilt
Increasing F
Slight Tilt
+ Offset
Figure 5.8 Ring structure and potential tilt
(A) Staircase model for ClpX. Increasing color refers to greater fluorescence following excitation by horizontally polarized light
(high fluorescence when fluorophore aligned with Z axis). Here, rigid bodies are assigned to the subunit that contributes the small
domain. The U subunit is shown as a red square and L subunits are shown as circles. If ClpX adopted a L-L-U-L-L-U or 6L
conformation, one would expect only 3 or 1 fluorescent states, respectively. (B) Cryo-EM model for a conformation of the
Drosophila 26S proteasome. The AAA+ Rpti- 6 ring is highlighted orange. The top ring is tilted relative to the 20S core, whereas
the bottom ring is not. The axis of both rings is offset relative to the axis of the 20S core. The offset axis would not affect
measured angles and has only been observed for the 26S and PAN-20S proteasomes (shown in B and C, panel 4). Adapted from
Nickell et al. (2009). (C) Four averages showing different conformations of the PAN-20S protease. The AAA+ PAN ring is
marked by an orange line. Conformations differ by the tilt of PAN on top of the 20S core particle. Adapted from Smith et al.
(2005). (D) EM image of ClpXP. The AAA+ ClpX 6 ring is marked by an orange line and rests flat on top of ClpP 14 . Adapted
from Ortega et al. (2002).
Single-molecule CoMET and TT quenching experiments suggest that some L subunits can
switch directly to the U conformation, whereas others cannot (Chapter 4). Thus, one would
expect disallowed transitions between states observed by polTIRFM. Until we have a larger data
set with angle-defined classes, we cannot definitively assign transitions between states. However,
145
my preliminary analysis suggests that not all transitions are allowed (Fig. 5.9A). Instead, there is
a hierarchy of subunits with "hub" subunits that can switch to many ring positions and other
subunits whose transitions are more restricted (Fig. 5.9B). Additionally, the likelihood of forward
and backward transitions is similar (Fig. 5.9A). Several groups have proposed that subunits
within 5L staircases switch in a unidirectional sequential order (e.g. states 5
+
4 -+ 3 -
2
without back steps). Our preliminary results support a stochastic model in which multiple
pathways are possible, and the energy barriers between forward and backward steps are similar.
A
2
Final State
3
4
B
5
2
3
Initial State
4
C
5
Figure 5.9 Transitions between ring positions
(A) Heat map of transitions between states 2-5 in which red signifies an increased probability of the transition. The state prior to
the transition is shown on the Y axis and the state after the transition is shown on the X axis. (B) Diagram of results in panel A.
Ring position 2 can switch with all subunits while positions 4 is restricted from transitions with positions 3 and 5. All transitions
have similar probabilities.
Ensemble measurements (Chapter 3) show that there are multiple classes of nucleotide-bound L
subunits.2 7 One model posits that the position of a subunit relative to the empty U subunit
dictates subunit class. In fact, our ensemble studies with nucleotide-binding mutations show that
wild-type subunits clockwise and counterclockwise to the mutant subunit have different apparent
nucleotide affinities and time-averaged conformations (Chapter 3). Upon closer inspection of our
polTIRFM ATP results, some ring positions have fast fluctuations that were not considered in my
preliminary analysis. Importantly, the fast conformational changes appear to be unique to ring
position, as some positions display stable fluorescence. Further experiments will be necessary to
146
determine if these fast fluctuations reflect ATP-hydrolysis dependent conformational changes in
subunits.
147
References
1. Glynn, S.E., Martin, A., Nager, A.R., Baker, T.A., & Sauer, R.T. (2009). Structures of
asymmetric ClpX hexamers reveal nucleotide-dependent motions in a AAA+ protein-unfolding
machine. Cell 139, 744-56.
2. Glynn, S.E., Nager, A.R., Baker, T.A., & Sauer, R.T. (2012). Dynamic and static components
power unfolding in topologically closed rings of a AAA+ proteolytic machine. Nat. Struct. Mol.
Biol. 19, 616-22.
3. Enemark, E.J., & Joshua-Tor, L. (2006). Mechanism of DNA transocation in a replicative
hexameric helicase. Nature 442, 270-5.
4. Thomsen, N.D., & Berger, J.M. (2009). Running in reverse: the structural basis for
translocation polarity in hexameric helicases. Cell 139, 523-34.
5. Lander, G.C., Estrin, E., Matyskiela, M.E., Bashore, C., Nogales, E., & Martin, A. (2012).
Complete subunit architecture of the proteasome regulatory particle. Nature 482, 186-91.
6. Beausang, J.F., Sun, Y., Quinlan, M.E., Forkey, J.N., & Goldman, Y.E. (2012). Orientation and
rotational motions of single molecules by polarized total internal reflection fluorescence
microscopy (polTIRFM). Cold Spring Harb. Protoc.
7. Julien, 0., Mercier, P., Spyrcopoulos, L., Corrie, J.E.T., & Sykes, B.D. (2008). NMR studies of
the dynamics of a bifunctional rhodamine probe attached to Troponin C. J Am. Chem. Soc. 130,
148
2602-9.
8. Beausang, J.F., Sun, Y, Quinlan, M.E., Forkey, J.N., & Goldman, YE. (2012). Fluorescent
labeling of calmodulin with bifunctional rhodamine. Cold Spring Harb. Protoc.
9. Sun, Y, Schroeder, H.W. 3 rd, Beausang, J.F., Homma, K., Ikebe, M., & Goldman, YE. (2007).
Myosin VI walks "wiggly" on actin with large and variable tilting. Mol. Cell 28, 954-64.
10. Lewis, J.H., Beausang, J.F., Sweeney, H.L., & Goldman, YE. (2012). The azimuthal path of
myosin V and its dependence on lever-arm length. J. Gen. Physiol. 139, 101-20.
11. Martin, A., Baker, T.A., & Sauer, R.T. (2007). Distinct static and dynamic interactions control
ATPase-peptidase communication in a AAA+ protease. Mol. Cell 27, 41-52.
12. Bewley, M.C., Graziano, V., Griffin, K., & Flanagan, J.M. (2006). The asymmetry in the
mature amino-terminus of ClpP facilitates a local symmetry match in ClpAP and ClpXP
complexes. J. Stuct. Biol. 153, 113-28.
13. Maglica, Z., Kolygo, K., & Weber-Ban, E. (2009). Optimal efficiency of ClpAP and ClpXP
chaperone-proteases is achieved by architectural symmetry. Structure 17, 508-16.
14. Maurizi, M.R., Singh, S.K., Thompson, M.W., Kessel, M., & Ginsburg, A. (1998). Molecular
properties of ClpAP protease of Escherichia coli: ATP-dependent association of ClpA and clpP.
Biochemistry 37, 7778-86.
149
15. Burton, R.E., Baker, T.A., & Sauer, R.T. (2003). Energy-dependent degradation: Linkage
between ClpX-catalyzed nucleotide hydrolysis and protein-substrate processing. Protein Sci. 12,
893-902.
16. McKinney, S.A., Joo, C., & Ha, T. (2006). Analysis of single-molecule FRET trajectories
using hidden Markov modeling. Biophys. J.91, 1941-51.
17. Wang, J., Song, J.J., Franklin, M.C., Kamtekar, S., Im, YJ., Rho, S.H., Seong, I.S., Lee, C.S.,
Chung, C.H., & Eom, S.H. (2001). Crystal structures of the HslVU peptidase-ATPase complex
reveal an ATP-dependent proteolysis mechanism. Structure 9, 177-84.
18. Walz, J., Erdmann, A., Kania, M., Typke, D., Koster, A.J., & Baumeister, W. (1998). 26S
proteasome structure revealed by three-dimensional electron microscopy. J Struct. Biol. 121, 1929.
19. Smith, D.M., Kafri, G., Cheng, Y, Ng, D., Walz, T., & Goldberg, A.L. (2005). ATP binding to
PAN or the 26S ATPases causes association with the 20S proteasome, gate opening, and
translocation of unfolded proteins. Mol. Cell 20, 687-698.
20. Nickell, S., Beck, F., Scheres, S.H., Korinek, A., Forster, F., Lasker, K., Mihalache, 0., Sun,
N., Nagy, I., Sali, A., Plitzko, J.M., Carazo, J.M., Mann, M., & Baumeister, W. (2009). Insights
into the molecular architecture of the 26S proteasome. Proc. Natl. Acad Sci. U.S.A. 106, 119437.
150
21. Ortega, J., Lee, H.S., Maurizi, M.R., & Steven, A.C. (2002). E.MB.O. J. 21, 4938-49.
22. Grimaud, R., Kessel, M., Beuron, F., Steven, A.C., & Maurizi, M.R. (1998). J Biol. Chem.
273, 12476-81.
23. Ortega, J., Lee, H.S., Maurizi, M.R., & Steven, A.C. (2004). ClpA and ClpX ATPases bind
simultaneously to opposite ends of ClpP peptidase to form active hybrid complexes. J Struct.
Biol. 146, 217-226.
24. Effantin, G., Ishikawa, T., Donatis, G.M., Maurizi, M.R., & Steven, A.C. (2010). Local and
global mobility in the ClpA AAA+ chaperone detected by cryo-electron microscopy functional
connotations. Structure 18, 553-62.
25. Sousa, M.C., Trame, C.B., Tsuruta, H., Wilbanks, S.M., Reddy, V.S., & McKay, D.B. (2000).
Crystal and solution structures of an HslUV protease-chaperone complex. Cell 103, 633-43.
26. Saeki, Y, & Tanaka K. (2007). Unlocking the proteasome door. Mol. Cell 27, 865-7.
27. Hersch, G.L., Burton, R.E., Bolon, D.N., Baker, T.A., & Sauer, R.T. (2005). Asymmetric
interactions of ATP with the AAA+ ClpX6 unfoldase: allosteric control of a protein machine.
Cell 121, 1017-27.
151
Appendix A
Stalling of cp6a-SF GFP-ssrA
This appendix contains experiments from Nager, Baker, and Sauer (2011) J Mol Biol 413, 4-16,
as well as unpublished experiments that will form the basis of a manuscript to be written. I
performed all of the experiments.
152
Extraction of an a helix
To test if stalling depended on a C-terminal 0 strand, I constructed and purified a circularly
permutated ssrA-tagged variant in which an a helix following strand 6 was at the C-terminus of
the P barrel (cp6a- SFGFP-ssrA; Fig. IA). ClpXP degraded ep6a-SFGFP-ssrA at high but not low
ATP concentrations as assayed by SDS-PAGE (Fig. 1B) or by loss of 467-nm fluorescence (Fig.
1C). To test if a stable intermediate could be formed following extraction of C-terminal elements
of structure, I engineered thrombin-cleavage sites either between strand 6 and the a helix (cp6aSFGFP-6/a-ssrA)
or between strands 5 and 6 (cp6a- SFGFP-5/6-ssrA). Following thrombin
cleavage, ClpXP extraction of the terminal peptide of cp6a- SFGFP-6/a-ssrA resulted in a stable
barrel with ~88% of native fluorescence (Chapter 2). However, ClpXP extraction of the a helix
and strand 6 of thrombin-split cp6a-SFGFP-5/6-ssrA caused complete loss of native fluorescence
(Fig. 2A). After strand extraction by ClpXP, the cp6a- SFGFP-5 protein appeared to aggregate and
eluted in the void volume of an S200 gel-filtration column. Moreover, in comparison with the
split protein prior to strand extraction, the absorbance spectrum of the gel-filtered protein had a
blue-shifted absorbance maximum near 395 nm and had lost an absorbance peak near 490 nm
(Fig. 2B). These properties are similar to those of uncleaved cp6a-SFGFP-ssrA after acid
denaturation (Fig. 2B). Thus, ClpXP degradation of cp6a-SFGFP-ssrA stalls at low ATP
concentrations, and extraction of the C-terminal a helix results in a stable intermediate. However,
extraction of this terminal helix and the neighboring strand does not produce a stable
intermediate. Thus, stalling can occur following extraction of diverse structural elements. In fact,
the degradation of RpoS, an entirely a helical protein, occurs with a stalling ATP dependence
similar to GFP (Peterson et al. Genes Dev. 26, 548-543, 2012).
153
A
C
order of 0 strands
SFGFP-ssrA
cp6a-SFGFP-ssrA
cp7.SFGFP-ssrA
cp8-SFGFP-ssrA
ssrA
@WWEJ(]a(DOOO)[
@o0
@OJUEJ
oo
W
D3h
c
-ssrA
a@-ssrA
@UUU1iGIssrA
B
SFGFP-ssrA
cp7-SFGFP-ssrA
II.
cp6a.SFGFP-ssrA
0
50
300
[ATP] (pM)
[ATPJ (PM)
Figure 1 Circularly permutated GFP.
(A) Cartoon representation of the order of p strands in SFGFP-ssrA and circularly permuted variants. (B) Permuted variants (1
RM) were incubated overnight with ClpXP (1.25 tM ClpX 6; 2.5 pM CIpP 14), the SspB adaptor (1 RM), and 0, 50, or 300 pM
ATP before assaying degradation by SDS-PAGE. (C) End-point experiments like those in panel B were performed but
degradation was assayed by reduced 467-nm fluorescence. GFP-ssrA (circles); SFGFP-ssrA (diamonds); cp6a-SFGFP-ssrA
(upward triangles); cp7-SFGFP-ssrA (triangles); cp8-SFGFP-ssrA (squares). The lines are fits to a modified form of the Hill
equation. In the panel B and C experiments, an ATP-regeneration system was used.
A
B
cp6a-SFGFP-5
1.0.i
1.0-
A7-11a-ssrA
1'O
CIPXP
A.
C 0.8-
A
A
o
cleaved
cp6A-SFGFP-5/6-ssrA
A
0.8-
0.6-
0.6-
E
rr 0.4-
0.4-
A
A
An
A
A
An
A
AN
A,
A
AA
'UA
A A
"A
WA
06A
A
A
SA
AA
mA
LE0.2-
0.21
U.
0
2000
4 00 0
6000
A
AA
A
iacid-denatured
cp6a-SFGFP-ssrA
350
time (s)
A
A
A%
WAO
=
400
450
wavelength (nm)
500
Figure 2 Extraction of an a helix, not a P strand, results in stalling for cp6a.
(A) The cp6a-SFGFP-5/6-ssrA protein (10 LM; NCBI accession code JF951870) was cleaved with thrombin and incubated with
ClpXP (1 pM ClpX 6 ; 2 pLM ClpP 14), 4 mM ATP, and an ATP-regeneration system. ClpXP extraction of the terminal a helix
and P strand resulted in time-dependent loss of 467-nm fluorescence and 400-nm fluorescence (data not shown). The initial rate
of this reaction (0.2 min-' enz-) was slow, but within error of the rate of ClpXP degradation of uncleaved cp6a-SFGFP-5/6-ssrA
(data not shown). (B) Absorbance spectra of thrombin-cleaved cp6a- sFGFP-5/6-ssrA (closed triangles), the cp6a-sFGFP-5 protein
after ClpXP strand extraction and purification by S200 gel filtration (open triangles), and cp6a-SFGFP-ssrA denatured by
incubation at pH 2 (squares).
154
Appendix B
Supplement for Ensemble CoMET of ClpX
This appendix contains experiments from B.M. Stinson*, A.R. Nager*, S.E. Glynn*, K.R.
Schmitz, T.A. Baker, and R.T. Sauer (2013) in submission (*-equal contribution). I developed
cCoMET and worked with Stinson to develop nCoMET. I conducted all experiments involving
cCoMET and disulfide-linked constructs and crystallized the 6L ClpX 6 structure.
155
Ta bl e 1 C
a
t ll
a
hi
c'Lt
ti s i cs!_______%,.
ClpX variant
E-E-ER
W-W-W
W-W-W
W-W-R
E-R
W-W-W
PDB code
in progress
in progress
in progress
in progress
in progress
in progress
tether length
20
20
20
20
20
0
bound nucleotide
crystallization well solution
Data collection
space group
ATPyS
a
ADP
a
none
a
none
b
none
b
none
c
P2 12 12 1
P2 12 12 1
P2 12 12 1
P2 12 12 1
P2 12 12 1
P6 3
57.9
199.2
211.9
90, 90, 90
55.9
181.9
201.4
90, 90, 90
55.2
199.9
222.3
90, 90, 90
55.2
201.2
222.6
90, 90, 90
119.4
119.4
111.7
90, 90, 120
50.0 - 4.5
50.0 - 5.7
60.0 - 4.5
7.7 (27.8)
17.1(4.3)
4.0 (3.2)
92.2 (80.4)
9.6 (35.2)
20.7(2.9)
8.5 (5.0)
99.4 (96.2)
18.6 (27.2)
5.5(4.1)
6.4 (6.6)
94.6 (95.7)
R7.4
(I)/sig(I)
redundancy
completeness (%)
(84.3)
24.7(1.6)
6.7 (6.5)
98.8 (95.5)
7.7 (30.9)
16.9(3.1)
3.8 (3.4)
93.9 (95.3)
58.3
199.6
203.4
90, 90, 90
50.0 -4.1
5.9 (29.4)
24.8(3.8)
5.7 (4.3)
97.2 (84.1)
Refinement
resolution (A)
Rwork/Rfree (%)
41.0-3.9
27.7/30.7
45.4-3.7
27.5/33.4
38.5 -4.1
28.5/31.0
48.3-4.5
30.9/33.0
49.1-5.7
29.4/31.9
49.1-5.0
32.2/35.2
bond angles (0)
bond lengths (A)
allowed Ramachandran (%)
0.002
0.417
100
0.003
0.508
100
0.002
0.420
100
0.004
0.641
100
0.008
0.579
100
0.004
0.613
100
unit-cell lengths (a, b, c) (A)
unit-cell angles (a,p,y)
(0)
50.0
resolution (A)
-
50- 3.7
3.8
Values in parentheses refer to me hignest resolution snell. Well solution a is 1.9 IVIanmonum sulate, i5 mVI sodum acetate
(pH 4.8). Well solution b is 2.2 M ammonium sulfate, 0.2 M amonium bromide, 0. 1 M bicine (pH 9.0). Well solution c is 2 M
ammonium sulfate, 0. 15 M potassium sulfate, 4 mM ATP, 4 mM MgCl 2 chloride, and 50 mM EDTA.
%~%~>
250,
*.~
1.5,
-
-
-
-
-
0
0
4
0.5~
V.
I-
-+
-+
-+
-+
-+
- +
Ni 2+-NTA
-+
-+
-+
-+
-+
-+
Ni 2+-NTA
Figure 1 ATP hydrolysis and degradation by tethered ClpX trimers
Tethered ClpX trimers (0.3 pM pseudo hexamer) containing the TT modifications (D76CTAlA; K330CTAMA) or cCoMET (J)
modifications (K330CTAMA; H68Q/N72H/D76H) were active in hydrolyzing 4 mM ATP (left panel) and in supporting
2
degradation of 10 jiM cp7-GFP-ssrA by 0.5 pM ClpP (right panel) in the presence or absence of Ni +-NTA (500 pM).
156
670
1584417
1.0-
kDa
1 1 1
1
w-vI-w.- w-w-w
-o
0
.0
*-
E
.
0.5-
-.
"0 Go
C
0%W
-.
5
10
15
20
elution volume (mL)
Figure 2 Size exclusion chromatography of W-W-W and W-VI-W
W-VI-W and W-W-W chromatographed at positions expected for pseudo hexamers on a Superose 6 gel-filtration column.
lini
B
0
60
vtnii
120
190
a
.5.
-10~
molar r1so
1
U-
AD NHTVW)
0
0
1iJ 2Q 3"o 40
.5
enooar ratw
ADP)(Wd
moor rM*a
ADFPWVWA),
Figure 3 Stoichiometry of nucleotide binding to W-VI-W and W 6
Binding of ADP to W-VI-W (panel A) or W 6 (panel B) assayed by isothermal titration calorimetry. The initial
concentrations in pseudo hexamer equivalents of the ClpX variants were 15.8 pM (W-VI-W) and 54.0 pM (W6 ).
Binding isotherms were fit to a one-site model using MicroCal Origin software. Data also fit well two a two-site
model (W-VI-W; site 1: KD = 3 ± 1 gM, N = 2.0± 0.6; site 2: KD = 26 ± 10 gM, N = 1.3 ±0.8; W6; site 1: KD = 5
2 pM, N = 2.0± 1.1; site 2: KD = 46 ± 20 RM, N = 2.0 ±0.8).
157
1.5
L~1.0
+C
0.5
0.0
10
100
1000
[ATP] (gM)
Figure 4 ATP dependence of substrate degradation by ClpX hexamers with single mutant subunits
ATP dependence of the rate of cp7-CFP-ssrA (20 pM) degradation by covalent ClpX hexamers (0.2 ptM pseudo hexamer; 0.5 pM
CIpP 14) containing a single nucleotide-binding-deficient subunit. The K substitution (K125M) disrupts the conserved lysine of
the Walker A motif and also prevents hydrolysis in the subunit bearing the substitution. Fit parameters, Table 2B.
8-
10-
300I
M0-.
200U~.
I
8-
2.
6-
N
zo
-0
4.
I
6-
-E
4-
00
2
200-
0.
0
10mM
10mM
10mM
10mM
Mg +
C02+
Mg +
C02+
2
2
0-
10 mM
10 mM
Mg2+
C02+
Figure 5 Co2 + supports ClpX activity
Comparison of the Mg 2+ versus C02+ supported activities of W-W-W ClpX (1 pM pseudo hexamer) in unfolding 10
pM cp7-CFP-ssrA (left panel), of W-W-W (0.3 tM pseudo hexamer) in hydrolysis of ATP in the presence of 10 pLM
cp7-CFP-ssrA (center panel), and of W-W-W (1 pM pseudo hexamer) in hydrolysis of ATPyS (right panel). Assays
were performed in PD buffer supplemented with the appropriate divalent metal at room temperature.
158
1
Co
CL
cc
a)
0-4
10mM
10mM
10mM
Mg 2 +
C02+
Mg 2 +
no CIpP
Figure 6 C02+ inhibits peptide cleavage by ClpP
Co 2 + inhibits ClpP cleavage. The rate of cleavage of a succinyl-Leu-Tyr-AMC dipeptide (50 RM) by ClpP 14 (1 M) was assayed
by changes in fluorescence (excitation 345 nm; emission 440 nm) in PD buffer supplemented with 10 mM MgCl 2 or 10 mM
CoCl 2. Rates were normalized to the rate with 10 mM MgCl 2.
200
S('
-
150-
SN
2 -0
100
-
50
-
CL
SE
0-
W-W-W
W-W-W*
Figure 7 ATP hydrolysis by M363C labeled and unlabeled ClpX variants
Rates of hydrolysis of ATP (5 mM) by the W-W-W and Oregon-Green labeled W-W-W* ClpX variants (0.3 pM pseudo
hexamer) in the presence of cp7-CFP-ssrA (10 pM).
159
0.6
0.4
Cr
0.2
0 .0
5f'
1
10
100 1000
[ADP] (RM)
Figure 8 ADP binding to W-VI-W by nCoMET
ADP binding to pseudo hexamers (0.1 pM) of W*-VI-W, W-VI*-W, and W-VI-W* assayed by nCoMET. The lines are fits to a
hyperbolic equation (Y = a-[nuc]/([nuc] + Kapp)). Kapp values are listed in Table S2.
0.5
c>)
0.4
C-
0.3
0.2
I
10
100
1000
[ADP] (sM)
Figure 9 ADP-dependent conformation changes of W-VI-W by cCoMET
ADP-dependent changes in the conformations of subunits containing cCoMET probes (§) were assayed for pseudo
hexamers (0.3
sM) of W-VI-W, W-VI-W,
and W-VI-W1 . Lines are fits to a Hill equation.
160
A
ATP
Cr
S0.30.2
1
10
100
1000
[nucleotide] (pM)
0.5-
ATPyS
0.4-
AT
ADP
Cr 0.3-
0.2
1
100
1 0
10
[nucleotide] (pAM)
Figure 10 nCoMET and cCoMET of W-W-W
(A) Nucleotide binding to W-W-W* (0.1 gM pseudo hexamer for ATPyS and ADP titrations; 0.5 gM pseudo hexamer plus 10
[tM cp7-CFP-ssrA for ATP titration) assayed by nCoMET. Lines are fits to single or double hyperbolic functions. Kapp values are
listed in Table S2. (B) Nucleotide-dependent changes in the rightmost subunit of W-W-W (0.3 gM pseudo hexamer for ATPyS
and ADP titrations; 0.3 jM pseudo hexamer plus 10 pM cp7-CFP-ssrA for ATP titration) assayed by cCoMET. Lines are fits to a
Hill equation.
161
1.0-
0.8-
0
'U
0.6-
I-
0.4-
0.204
V
1\0%
V,
I'm
0+
Nd
Figure 11 CIpP pore opening with L-locked variants
Rates of cleavage of a fluorescent decapeptide (15 gM) by a cysteine-free ClpP 14 variant (50 nM) were determined in the
presence of different ClpX variants (0.2 pM pseudo hexamer) and 1 mM ATPyS. The disulfide-bonded L-lock enzymes bind
ClpP and enhance peptide cleavage, although they do not support degradation of protein substrates.
1.0E
0,
0.80.60.40.20-
nonuc
U(ll
(1 mM)
Al'
(1 mM)
Figure 12 nCoMET quenching is specific to ATP
Specificity of nCoMET quenching. In PD buffer plus 10 mM CoCl 2 , ATP reduced the fluorescence of W-W-W* but GTP or
buffer with no nucleotide did not result in quenching.
162
A
B
1
~
0.3CD
u. 0.5 -
0.2-
no Ni~-NTA
Cr
0.1-
o.u
o.0o
O
0
500 1000 1500 2000
[ATP] (pM)
200
400
600
[Ni**-NTA] (pM)
Figure 13 Changes in cCoMET fluorescence depende on Ni 2+-NTA
(A) In the absence of Ni2+-NTA, titration of ATP against W-W-WV (0.3 pM) did not result in quenching. (B) Titration of Ni 2+_
NTA against W-W-W (0.1 pM) gave -30% quenching The line is a single hyperbolic fit with with Kapp = 25 ± 2 pM.
0.55S
K
gO
0.45Cr
0.35 4
0
50
1000 1500
[ATP] (RM)
2000
Figure 14 cCoMET across the rigid-body interface
ATP dependence of cCoMET quenching for a W-W-W variant containing TAMRA-labeled S389C in the small
AAA+ domain of the second subunit and the His7 2 -X 3 -His 76 mutations in the large domain of the third subunit. This
cCoMET pair spans a single rigid-body unit. The line is a single hyperbolic fit with Kapp = 30 ± 9 pM. At nucleotide
concentrations at which the low-affinity sites are occupied and ATP hydrolysis occurs, no major changes were
observed indicating that the conformational changes monitored with other cCoMET pairs, which correlate with ATP
hydrolysis, involve changes across the hinged interfaces of ClpX rather than the rigid-body interfaces.
163
Table 2 cCoMET/TT fi parameters
variant
W-W-W§
W-VI-W§
w§-VI-w
W-V1-W
TT
*amplitude
=
Kapp (pM)
nucleotide
66 5
150 5
27 1
6 2
21 1
25 ±1
11 ±1
45 ±4
22 ±1
266 ±8
2 ± 18*
± 13**
70 ±2
15 ±2
41 ±6
9 ±1
ADP
ATP
ATPyS
ADP
ATP
ATPyS
ADP
ATP
ATPyS
ADP
ATP
ATP226
ATPyS
ADP
ATP
ATPyS
-0.03 **amplitude = 0.15
164
Hill
1.4 ±0.1
1.6 ±0.1
1.6 ±0.1
1.1 ±0.2
1.2 ±0.1
1.3 ±0.1
1.1 ±0.1
0.8 ±0.1
1.4 ±0.1
2.2 ±0.1
n/a
2.5 ± 0.3
1.5 ±0.1
n/a
n/a
n/a
Appendix C
Supplement for Single-Molecule CoMET of ClpX
This chapter contains experiments from A.R. Nager*, Y. Shin*, H. Manning, T.A. Baker, M.J.
Lang, and R.T. Sauer (in preparation) (* equal contribution). I developed cCoMET and
conducted experiments and analyzed data in collaboration with Yongdae Shin and Harris
Manning.
165
8
0
X-
0
120
0
120
Time (sec)
Time (sec)
Figure 1 Single-molecule trajectories in the presence of saturating ATP
A ClpX variant (W-W-W-W-W-W) with a single CoMET-labeled subunit was observed by total internal reflection fluorescence
microscopy. After -10 seconds, Ni2+-NTA, ClpPI4 , and 1 mM ATP were flowed into the cell. Switching between quenched and
unquenched states (U@L switching) and between different quenched states (L motions) was observed.
A
B
Ni2'-NTA binding curve
dwell time probability of
unquenched states
S 10015
IO
0
25
3
e
E
0..
0
200
400
800
[N12.-NTA] (pM)
I00
1000
1200
Time (sec)
Figure 2 Ni 2 -NTA binding to the CIpX His7 2 -X3 -His7 1motif
Single-molecule experiments were conducted with 1 mM ADP and increasing concentrations of Ni 2+-NTA. (A) Plot of the
fraction of time spent in a quenched state at a concentration of Ni 2 +-NTA. The portion of time that is not quenched with
saturating Ni2+-NTA is likely to represent bona fide extended conformations. (B) Dwell-time probability distribution for
unquenched states with 100, 500, and 1000 pM Ni2+-NTA. For 500 pM and greater concentrations of Ni 2 +-NTA, the unquenched
dwell-time distribution was similar, consistent with these dwell times representing extended conformations rather than fast
Ni 2 +-NTA binding-unbinding events.
U
o
Time (sec)
Time (ee)
Figure 3 Single-molecule trajectories of D 7 6 CTAMRA K330CTAmRA conta
uenching with saturating ATP
A ClpX variant (W-W-WIT-W-W-W) with a single TT subunit (D76CT
K330C1AlR) was observed by single-molecule
TIRF. The TAMRA-dye pair is in close proximity and undergoes contact quenching in the L conformation but is separated and
unquenched in the U conformation. Switching between a high-fluorescent state and a quenched state was observed in this
experiment with ATP (1 mM) and also with saturating ADP (0.75 mM) and ATPyS (1 mM).
166
DA
A
Fraction U subunits
ensemble
tingle-molecule
[ATP) (pM)
Figure 4 Nucleotide-occupancy switches the average L:U subunit ratio 4L:2U to 5L:1U
Single-molecule CoMET experiments using W-W-W-W-W-W were performed with increasing concentrations of ATP, and the
proportion of time spent in a U conformation was plotted (squares). Also plotted is the estimated proportion of time spent in a U
conformation from an ensemble measurement using W-W-WTT pseudo hexamers (Chapter 3).
U-dwell as a function of [ADP]
25!
1mMA[P
20
1OuMADP
lololAp
IUMAEP
20-
15
#
10
1 3 5
7 9 11 13 15 17 19 21 23
Time (s)
Figure 5 U dwell-time probability distribution with different ADP concentrations
Single-molecule CoMET experiments were conducted with increasing concentrations of ADP. At low ADP concentrations (green
and black bars), the U dwell-time distribution was best fit by a gamma function indicating multiple rate-determining steps. At
nucleotide concentrations above KD for ADP binding (red and blue bars), the U-dwell distribution was fit well by an exponential
function. These results suggest that at least one rate-determining step is accelerated by ADP occupancy.
L dwell time distribution, 1 mM ATPgS
expon ntial-pls-gamra fit
10
2M
W
time (sec)
40
5D
C
Figure 6 L dwell-time probability distribution with 1 mM ATPyS
The distribution was best fit by an exponential-plus-gamma function supporting two classes of L subunits in the presence of 1
mM ATPyS.
167
A
stochastic
Sequential
\*
LI
7
time (aarbAiry)
B
Class Hierarchy
time (arbtry)
Figure 7 Simulated L dwell-time distributions
Simulated L dwell time-distributions for (A) stochastic, (B) stochastic with class hierarchy, and (C) strictly sequential models
with saturating nucleotide. For each model, I assumed that a hexamer contains 5L and lU subunit at all times, a U class (squares)
switches to an L class (circles) in one step, and all rates (arrows) are equal. For the stochastic model (A), at any moment, every L
subunit can switch to the U class in a single step. Thus, the L-dwell distribution is best fit by an exponential function. The strictly
sequential model (C) requires a specific switching order in which a subunit must sequentially pass through every position in the
hexamer. A series of steps before the L4U switch results in a gamma distribution. For class hierarchy models (B), there are two
classes of L subunits, one class (open circles) that can switch to U in one step and another (closed circles) that can switch to U in
two steps. A class hierarchy results in an exponential-plus-gamma distribution. Many variants of this model result in a similar
distribution. For instance, depending on relative rates, there can be different proportions of each class of L subunits. Additionally,
L subunits could switch in stochastic or ordered schemes as long as the relationship with the U subunit remains unchanged.
Dwell time probability distribution
W-VIP-W, 0.75 mM ADP
*-U
dwell
40
,L
Ldwel
W'
time(sec)
n5
45
4
Figure 8 Dwell-time probability distributions for a low-affinity subunit at saturating ADP
U and L dwell-time distributions for CoMET of a W-VIN-W-W-VI-W hexamer in which the subunit marked § contains the
CoMET pair. The VI mutation reduces the affinity for nucleotide (Chapter 3) and thus the probability of the VIP subunits being
unoccupied is higher than for W subunits. 0.75 mM ADP.
168
Supplemental Methods
Covalently-linked ClpX trimers and ClpP1 4 were purified and labeled as described (Martin et al.
Nature 437, 1115-20, 2005; Chapter 3).
To construct stable hexamers for single-molecule
imaging, a fluorescently-labeled trimer with C-terminal LPETGG was fused with a biotinlabeled trimer with N-terminal GGGG by sortagging (Popp et al. Nat. Chem. Biol.3, 707-8,
2007). Streptavidin-coated flow chambers were prepared as described (Shin et al. PNAS 106,
19340-5, 2009).
169
Appendix D
Catalog of ClpX mutations
This chapter contains an unpublished catalog and commentary of ClpX mutations. I performed
all of the experiments.
170
For mutations of other constructs (GFP, CIpS, Synzips, DegP, etc.), please refer to my
plasmids/primer spreadsheet. Primers listed as "ATH" refers to the "Around-the-Horn" cloning
strategy. All ClpX constructs are AN domain. Most primers are designed for Andy Martin's ClpX
monomer.
Tether and IGF loop truncations for ClpX crystallography
Past structures of ClpX hexamers were solved using genetically-tethered trimers in which three
ClpX subunits were connected by 20 amino acid tethers (L20; Fig. 1A). To improve crystal
contacts, I progressively shortened the tethers to 0 amino acids (L10, L6, L4, L2, LO). Because
the N- and C-termini of ClpX are unstructured, I designed LM# (minus # amino acids) constructs
that include truncations of the termini (LM2, LM4). A modeled 0-length tether could easily
bridge the visible termini in ClpX crystal structures (Fig. lB,C, data not shown). Tethered dimers
and trimers as short as 0 amino acids expressed, were soluble, and hydrolyzed ATP (Fig.
1D,E,F). Dimers with minus tethers (LM2, LM4) had low solubility. LO trimers formed crystals
rapidly (3 days; Fig. 2D), but had anisotropic diffraction (Chapter 3).
171
A
D
CIPX2
B
E
C~pX
ATP/min
Tmni(Ct
Dio
Typ I -I
H->rI24.58
II->I
A
18.89 A
.
27A
ATP/
min
Figure 1 Tether truncations
(A) Diagram of a ClpX hexamer with 20-aa tethers shown in red (taken from Martin et al (2005) Nature). (B) A rigid body with
the small (green) and large (blue) domains of adjacent subunits. The N- and C-terminal residues (Ser 62 & Tyr 413) are 27 A
apart (PDB: 3HTE; Glynn et al (2009) Cell). (C) Distance between termini for different types of subunits. Note: Type 1 = ATP
Loadable, Type 2 = ATP Unloadable. (D) SDS-PAGE of ClpX dimers with different length tethers. The black illustrates the
change in distance when comparing constructs. (E) ATP hydrolysis by ClpX dimers with different length tethers. (F) ATP
hydrolysis by ClpX trimers with different length tethers.
In an attempt to improve crystal contacts, I made progressive deletions of the IGF loops (residues
264-281), which are flexible but necessary for binding ClpP (Fig. 2A). The nearby R261 is
necessary for hexamerization (Table 1). I made progressive deletions up to 21 residues: aa262282. All truncations were soluble, but the 21 residue truncation had reduced ATPase in the
presence of substrate (Fig. 2B,C). The A21 variant had hexamerization defects (trailing on S200,
[ClpX] titration by ATPase; data not shown). I constructed a ClpX trimer with 19 residue IGF
truncation and 0 amino acid tethers. Unfortunately, this construct did not form large crystals and
frequently precipitated in trays (Fig. 2D).
172
A
B
Deletion 13 15 17 19 21
IGFLoopdeletionMonomers
C
eatsunme
JDP
aN
too
ATP/i400
min
300
613
D
d15
dt7
di9
d21
LOTether,
d19IGF
Trimer
LOTether
Trimer
Figure 2 IGF loop truncations
(A) Side view of a ClpX hexamer. A single subunit is highlighted in green. Inset: The top of the IGF loop. Most of the loop is
disordered. (B) SDS-PAGE of ClpX monomers with progressive deletions of the IGF loop. (C) ATP hydrolysis by ClpX
monomers with IGF loop deletions. Shown with and without CM-127-ssrA. (D) Crystal drop of a ClpX trimer with short tethers
versus short tethers and no IGF loop. Short-tether trimers gave large plate-shaped crystals whereas the IGF deletion only resulted
in small crystals.
Mutation
Primers
ClpX construct
IGF
IGF
IGF
IGF
IGF
165/156 ATH
164/154 ATH
163/152 ATH
162/150 ATH
161/148 ATH
Andy's
Andy's
Andy's
Andy's
Andy's
l3aa
15aa
17aa
l9aa
21aa
Table 1 IGF loop truncations
Progressive deletion of the IGF loops.
Cysteine mutations
This section contains a list of cysteine mutations (Table 2). For additional mutagenesis sites,
consider those in Table 3. I selected residues to mutate to cysteine based on solvent accessibility
(to avoid folding defects), nearby basic residues (to enhance reactivity), non-conservation in
173
ClpX orthologs (to avoid residues critical for folding or function), and general spread over the
molecule. A collection of cysteine mutants and their activities are shown Fig. 3A and listed in
Table 2. The most difficult area to place mutations was near the nucleotide-binding site. There
were few non-conserved residues and several mutations inactivate the enzyme. Interestingly, the
sequence of the a-helix consisting of residues 318-328 was not strongly conserved, but some
mutations/labeling were not tolerated (Fig. 3B,C, Table 2). E327CDYE was inactive (Fig. 3B).
E328C was hyperactive (hydrolyzing ATP 7-fold faster than wild type and degrading CM-titin12 7 _
ssrA 4-fold faster) but also hypersensitive to stalling (data not shown). Histidine mutations in this
helix were tolerated, but addition of a metal ion halted ATPase activity (Fig. 3C, Table 3). By
contrast, I had success labeling cCoMET constructs in the residue 77-80 loop (Table 2). In my
hands, these constructs hydrolyzed ATP, degraded substrates, and underwent conformational
changes with an [ATP] dependence similar to other ClpX variants. However, the 179A/G80A
mutation reduced affinity for nucleotide (Chapter 3). Either the dye-labeled sites do not weaken
nucleotide binding or my functional assays did not detect the reduced affinity. For additional
residues near the nucleotide-binding site, consider residues near the arginine finger. A second
area intolerant to mutations were the helices and turns near the rigid-body interface. Mutation of
a331-344 often resulted in hexamerization defects (Table 2, 3). I would guess a similar effect for
mutations in a371-391, as mutations at 386 and 389 reduced ATPase and cp7-ssrA degradation
(Table 2).
The underside of the small domain (352-365) can be mutated with little or no consequence
(Table 2,3). Similarly, the top of the large domain (a60-76, a158-168, a206-217, loop 105,106)
can mostly be mutated. At one point, I was curious about the large number of charged residues in
this region and considered a possible function as an "unfolding surface." I mutated many
174
residues in pairs (and subsequently combined pairs) without effecting ATPase or degradation of
GFP-ssrA or CM-127-ssrA. The only residues that affected activity had side chains that were not
surface accessible (Table 2). I made a few mutations near the IGF loops (250-290). A few sites
disrupted hexamerization, but others were tolerated, including A252CDYE
For L-lock constructs, in which a disulfide locks a subunit in the L conformation, I tried three
combinations: E150C/E205C, E150C/E209C, and T147C/E205C. A fourth combination,
T147C/E209C, was never tested, but was present only in nucleotide-bound structures.
E150C/205C was not favorable in nucleotide-free or nucleotide-bound structures, and, as
expected, was poor at forming crosslinks. E150C/209C, which is present in the nucleotide-free
structures, formed best in the absence of nucleotide. In comparison, T147C/E205C was present
in all structures and formed well with and without nucleotide. I continued with T147C/E205C
(Chapter 3), but one could use other cysteine pairs to lock different L conformations.
A
GFP degradation
GFP/
min*
B
K327C
WT
ytm
C
E320H, Q324H
WT
ytiew
WI W/NJ
miew
MMtI
Figure 3 ClpX eysteine mutations
(A) GFP degradation by ClpX monomers with cysteine mutations. Constructs have not been reacted with fluorescent dyes. (B)
K327C loses activity when reacted with monobromobimane or fluorescein-5-maleimide. (C) A i/i+4 histidine motif that inhibits
ATPase when bound to metal.
175
Mutation
E69C
N72C
D76C
Y77C
179C
G80C
R100C
S105C
B106C
G107C
L127C
D137C
T147C
E150C
E156A
D157A
E159A
N160A
Q163A, K164A
Q167A, K168A
Q174C
E205C
Q208A, Q209A
Q209C
K213A
E216A
A252C
H260C
R261C
E263C
E283C
A288C
S318C
K327C
E328C
K330C
K336C
D356C
A357C
K360C
M363C
A364C
Primers
581/582
167/168
169/170
490/493
491/493
492/493
171/172
583/585 ATH
584/585 ATH
173/174
494/495
175/176
462/464 ATH
463/464 ATH
472/473 ATH
474/475 ATH
476/477 ATH
could not find
478/479 ATH
480/481 ATH
205/178
465/467 ATH
482/483 ATH
466/467 ATH
484/485 ATH
484/486 ATH
179/180
181/182
could not find
183/184
206/207
187/188
189/190
191/192
326/327
328/329
330/331
510/497 ATH
510/579 ATH
510/580 ATH
499/500 ATH
503/504 ATH
193/194 or 503/505
ClpX construct
His-TEV-Andy's, D76C
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Synthetic B
Synthetic B
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's, M363C
Andy's, A364C
Andy's
Andy's
Andy's
176
Activity
WT
WT
WT
WT? see text
WT? see text
Fig. 3A
WT
WT
Fig. 3A
dead, Fig. 3B
Fig. 3A
increased ATPase
WT
WT
dead
dead
dead
WT
WT
Fig. 3A
WT
dead
WT
WT
WT
near-WT
Fig. 3A
hexamer defect
Fig. 3A
hexamer defect
expect WT
Fig. 3A
dead, Fig. 3B
hyperactive
WT
hexamer defect
WT
WT
WT
WT
WT
D386C
S389C
E394C
G405C
ATH
586/587 ATH
503/508 ATH
586/588 ATH
195/196
208/209
Synthetic C
Andy's
Synthetic C
Andy's
Andy's
half WT
low activity
Fig. 3A
Fig. 3A
Table 2 Cysteine and alanine mutations
Most primers not listed as "ATH" are site-directed mutagenesis. Check excel spreadsheet for details.
cCoMET mutations and pairs
I designed or planned four classes of cCoMET pairs (Fig. 4). The L vs. U pair differentiates
between these conformations and is described in Chapters 3 and 4 (Fig. 4A). In general, high
nucleotide concentrations caused an increase in quenching (Fig. 5B). This is most likely due to
hexamers switching from 4L:2U to 5L: lU ring architectures. I planned a similar pair that was
closer to the pore (Fig. 4C). These pairs can be designed to amplify motions detected by L vs. U
pairs. Additionally, pairs can be made with pore loops, permitting one to investigate local
motions of the pore loops. Mutations near the pore are tolerated (Table 2). The third class was
termed "snake jaws" (Fig. 4B). These pairs are designed to quench both the U and L
conformation. The snake jaws model posits that substrates would cause the U conformation to
expand, expanding the pore to accommodate substrates and reducing quenching. I did not
observe a change with substrate, but multi-chain histag-cleaved substrates were not tested.
Interestingly, this pair had a biphasic response to nucleotide (Fig. 5C), with opposing changes
upon nucleotide binding to the tight and weak sites. Although many snake jaws pairs work, I
would suggest using K360H, M363H paired with Y77CTAMRA because these pairs are spatially
close. I79CTAMRA and G8 0CTAMRA are closer, but may effect nucleotide binding (Chapter 3).
The final cCoMET pairs were across the rigid body interface (Chapter 3, Fig. 4D). Upon
nucleotide binding to the tight sites, there was a small change in quenching (Fig. 5D). Nucleotide
177
may "set the ring," forming the rigid body interface and poising the hexamer for future
nucleotide-binding and hydrolysis events. There were no changes in fluorescence at higher
nucleotide concentrations, consistent with the rigid-body interface remaining static during
function.
For cCoMET, one can use Ni2+ or Cu2+. However, for ClpX, Cu 2 + binds to a secondary site. The
secondary site is within the large domain and is near a60-76. Because the Cu2+ site is specific,
one can still use Cu2+ for initial measurements with a longer Ro.
A
L vs. U cCoMET
ATPLoadable
B
C Near-pore cCoMET
Canoant
Snake Jaws cCoMET
ATPLoadable
ATPUnloadable
D
utations
ATPUnioadabe
Rigid Body cCoMET
Smal(n subunit)
Figure 4 cCoMET design
(A) i/i+4 motif at residues 72/76 and C at 330. Quenching in the L, but not U, conformation (Chapter 3, 4). (B) The cCoMET pair
is split between the -75 helix and ~360 helix. In this pair, both L and U conformations are detected. By comparing with the pair
in (A), one can differentiate between translations and rotations of the two domains. Additionally, these pairs could test the snake
jaws model. (C) I considered moving pairs further away from the nucleotide-binding pocket. This can be done to amplify signals.
(D) Across the rigid body interface.
178
Mutation
H68Q
Primers
453/454 ATH
461/454 ATH
N72H,D76H
V78A, 179A
E320H, Q324H
Q324H, K327H
T334H, Q338H
K335H, A339H
D356H, K360H
A359H, M363H
K360H, M363H
IGQE78hexahis
IGQE78CTPHPFM
K330W
210/211 ATH
597/598 ATH
212/213 ATH
332/333 ATH
334/335 ATH
334/336 ATH
510/498 ATH
499/501 ATH
499/502 ATH
593/594 ATH
595/596 ATH
589/590
ClpX Construct
Andy's, N72H, D76H
His-TEV-Andy's,
N72H, D76H
Andy's
Andy's, N72H, D76H
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Activity
WT
WT
Chapter 3
Fig. 7.3C
similar Fig. 7.3C
hexamer defect
hexamer defect
WT
WT
WT
not tested
not tested
WT
Table 7.3 cCoMET i/i+4 histidine motifs
A
CipX Hexamer
B
C
FSM-labeled L vs. U
FF
K
D
FSM-labeled Snake Jaws
*.
K,,, = 30 pM
120 pM
1ATP(CM)
TAMRA-labeled RI
(ATeI
(PM)
[ATPI(pM)
Figure 5 cCoMET with different pairs
(A) ClpX hexamer with a subunit highlighted (large: cyan, small: pink). Inset: single domain with measurements highlighted. 1,
2, and RI refer to (B), (C), and (D), respectively. (B) L vs. U pair. In general, nucleotide caused contraction with a % max similar
to ATP hydrolysis. (C) Snake jaws pair. All pairs of this type give a similar biphasic result indicative of two binding sites with
differing affinities. (D) Rigid body pair. At very low nucleotide concentrations, there is a small rearrangement of the rigid body.
179
Mutations for single-molecule nanometry
I made a series of Yl53A ClpX hexamers for study by single-molecule nanometry. The goal was
to investigate slipping and the consequences of ortho-, meta-, and para-Y1 53A mutations. These
constructs were given to Ohad Yosefson for further analysis. In addition, I made AA insertions
near the GYVG loop to see how this would affect translocation. Monomers with AA insertions
could not degrade CM-titin12 7-ssrA. For a while, we thought that E185Q mutations were too
severe for the optical-tweezer setup, so I designed less severe mutations of the Walker B motif.
These mutations are useful for slowing down, but not eliminating, ATP hydrolysis.
Mutation
Primers
ClpX Construct
Y153A
538/539
540/541
542/543
544/545
546/547
548/549
468/469 ATH
470/471 ATH
599/600 ATH
599/601 ATH
602/603 ATH
602/604 ATH
602/605 ATH
Synthetic A
Synthetic B
Synthetic C
Synthetic D
Synthetic E
Synthetic F
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
Andy's
insert AA after T 147
Insert AA after D157
Y181F
Y181A
D184E
D184N
E185D
Table 4 Pore loop and ATPase mutations
ClpPplatform
ClpPPiatfon is described in Chapter 5, but these constructs are applicable for a large number of
experiments. For instance, the first version of ClpPPiatfon (ClpP-sortase tag) was designed for
constructing ClpXP fusions for crystallography. The ClpP-TEV-Histag-Biotin and ClpP-TEVHistag-Sortase constructs are autocatalytically clipped at the C-terminus. To prevent this, remove
180
any one of the three tags.
Mutation
Primers
Add Sortase tag to C-terminus
552/553
554/555
606/607
608/609
646/647
M5A
Add biotin to C-terminus
Remove TEV from C-terminus
ClpP construct
ClpP-TEV-Histag
ClpP-Histag
ClpP
ClpP-TEV-Histag
ClpP-TEV-Histag-Biotin
ATH
ATH
ATH
ATH
ATH
Table 5 CIPPp''f"rm
Synthetic C1pX constructs
The original N-terminal sortase (polyG) trimers did not express in F. coli (Fig. 6). Expression
was rescued by deletion of the N-terminal polyG motif (data not shown). To retain the polyG
motif, I inserted a Ulpl-cleavable Sumo domain N-terminal to the trimer. This construct
expressed well, but not as well as the C-terminal sortase trimer (Fig. 6, data not shown). In Table
6 are additional primers for adding/removing sequences from the termini of synthetic constructs.
No Sumo
e
250
150
Sumo
Induction
wi
250
I50
h
ductio
CIPX
Figure 6 Expression of the N-terminal sortase ClpX trimer
SDS-PAGE of N-sortase trimer expression without (left) and with (right) a preceding Sumo domain.
181
Mutation
Add Sumo-polyG to Nterminus
Add Sumo-polyA to Nterminus
Remove TEV-Histag
Add TEV
Remove Biotin-TEV-Histag
Remove Biotin
Remove Flag
Primers
528/529 insert (Neol/SacI)
ClpX Construct
Nsort trimer
528/530 insert (Neol/SacI)
Nsort trimer
531/532
535/536
648/649 insert (NotI/XhoI)
693/694 insert (NotI/XhoI)
690 (pair with 3' synA primer)
Nsort trimer-biotin-TEV-Histag
Csort trimer-LPETG-Histag
Nsort trimer-biotin-TEV-Histag
Hexamer-biotin-TEV-Histag
Csort/Hexamer, T66C
Table 6 Synthetic ClpX multimers
Below are primers for cloning Andy's ClpX monomer into the synthetic hexamer. The hexamer
can tolerate several ClpX monomers without noticeable recombination. When cloning, some
sites are more difficult than others. For instance, Andy's monomer does not clone well into site A
or site D. By contrast, Andy's monomer is better at cloning into sites C, E, or F than the synthetic
monomers. If you have trouble cloning, switching between Andy's and the synthetics is a
possible strategy.
Synthetic Site
X1 (synA)
X2 (synB)
X3 (synC)
X4 (synD)
X5 (synE)
X6 (synF)
Restriction Enzymes
NdeI/KpnI
KpnI/SpeI
Spel/SacI
SacI/PstI
PstI/BamHI
BamHI/NotI
Table 7 Monomer-to-multimer primers
182
Primers
573/574
577/578
523/524
575/576
637/638
533/534
insert
insert
insert
insert
insert
insert