Supplementary information to
Structural alterations for proton translocation in the M state of wild-type
bacteriorhodopsin
Hans Jürgen Sass*, Georg Büldt*, Ralf Gessenich*, Dominic Hehn*, Dirk Neff*, Ramona
Schlesinger*, Joel Berendzen† & Pal Ormos‡
* Institute of Structural Biology, Research Centre Jülich, 52425 Jülich, Germany
† Biophysics Group, Mail Stop D454, Los Alamos Laboratory, Los Alamos, NM 87545, USA
‡ Institute of Biophysics, Biological Research Centre of the Hungarian Academy of Sciences, Szeged,
P.O.Box 521, 6701 Szeged, Hungary
Additional Results
To illustrate the effect of changes in the tertiary structure together with the reorientation of
side chains in our M2 state structure the figure below displays the cavities within the molecule
of the ground state and the M intermediate. The cavities were determined with the program 1
GRASP using a probe radius of 1 Å. The most important changes are found in the cytoplasmic
part of the molecule (top). In the M2 state model a cavity system, connected to the protein
surface by a tube (too narrow to allow the passage of water molecules in a rigid bR molecule)
has established which could allow a fluctuating water molecule to access the large cavity close
to the amino acid Asp96, as described in the main text.
Figure : The images of the BR model (a) and the M2 model (b) display the cavities in yellow
and the indentation from the outside surfaces in blue, as determined with a probe radius of 1
Å. The ground state (a) has a number of cavities but no clear connections to the cytoplasmic
or extracellular surfaces. In the M2 state structure (b) extended cavities are detected in the
cytoplasmic domain, necessary for proton transfer from Asp96 to the Schiff base nitrogen of
the retinal and for the reprotonation of Asp96 from the cytoplasmic surface.
2
Table 1 Diffraction data
Data set
Space group
Cell dimensions (Å)
Resolution (Å)
Unique reflections, all
Unique reflections, I > 2 
Completeness (%), all
Completeness (%), I > 2 
Rsym (%) ¶
Mosaicity (°)
Twinning (%)
Illuminated
P63
a=b=61.08, c=110.40
20 - 2.25
10792
9284
97.2
20.0 - 2.25 Å
99.4
2.33 - 2.25 Å
83.6
20.0 - 2.25 Å
62.4
2.33 - 2.25 Å
8.2
20 - 2.25 Å
32.8
2.33 -2.25 Å
0.6
38
ground state
P63
a=b=60.81, c=110.64
20 - 2.0
14913
13992
94.9
20.0 - 2.0 Å
89.1
2.07 - 2.0 Å
89.0
20.0 - 2.0 Å
71.0
2.07 - 2.0 Å
4.7
20.0 - 2.0 Å
31.9
2.07 - 2.0 Å
0.79
44
¶ Rsym(l) = ΣhklΣiIhkl,i - Ihkl/ΣhklΣiIhkl,i, with the average intensity lhkl of the multiple observations lhkl,i for
symmetri-related reflections.
Table 2 Refinement statistics
Data set
M occupancy (%) §
Test set (%) #
R factor (%) ||
Rfree (%) &
Illuminated
35
5.3
16.7
20 - 2.25 Å
23.6
20 - 2.25 Å
ground state
0
5.5
17.9
20.0 - 2.0 Å
20.8
20.0 - 2.0 Å
r.m.s. bond distance dev. (Å) $
r.m.s. angle dev. (°) $
0.008
1.19
0.006
0.98
Average B factor (Å2) 
24.8 / 20.9
31.4
§ The BR model with an occupancy of 65% was fixed during refinement.
# structure factors for Rfree calculation selected in several thin resolution shells to avoid bias due to the presence of
merohedral twinning.
|| R factor = ΣhklFo - Fc/ ΣhklFo, Fo and Fc are observed and calculated structure factors, respectively.
& Rfree = ΣhklFo - Fc/ ΣhklFo, R factor of the test set omitted from the refinement
$ root-mean-square deviation.
 two B factors are given for the illuminated crystal : M model / ground state model.
Deposited in the Protein Data Bank: 1CWQ
Methods
This is a more complete version than in the main text where space limitations prevented the
full description.
Crystallisation and data collection
Crystals were grown according to the protocol of Landau and Rosenbusch 2. Diffraction
experiments at 95 K were performed at the X8C beam line of the NSLS Brookhaven with a
wavelength  = 1.214 Å.
Illumination of the crystals
bR ground state: The crystal was flash-cooled in the light-adapted ground state to liquid
nitrogen temperature. Small amounts of the K intermediate, possibly produced by the room
light, were driven back to the BR ground state by illumination with red light of λ = 685 mm.
Formation of M intermediate: To generate the late intermediates of the photocycle it is
critical to consider that the protein conformation, shown to be correlated to effective proton
pumping 3, can be observed only above 260 K . Our procedure for M state formation was: (i)
Heating the crystal to room temperature by blocking the nitrogen cold stream. (ii) Illuminating
with green light (33 mW, 514 nm) for 1 second. (iii) Cooling by unblocking the cold stream
while the green light is still on. (iv) 1 second after cooling has started, the illumination is
turned off. (v) Thereafter illuminating with weak red light, to drive the molecules possibly
trapped in the K intermediate back to BR.
Microspectroscopy of the crystals: Spectroscopic control experiments under identical
conditions were performed with a microscope interfaced to a spectrometer. Difference spectra
in the visible region were produced by measuring single beam spectra before and after
illumination. The amount of components in the photoproduct was estimated on the basis of
known spectra of the intermediates 4. On average 60 to 70 % bR molecules in each crystal
were accumulated in the M state, small amounts in the N intermediate and 30 to 40 % in BR
state. For the ground state it was verified that more than 90 % of bR were driven to the light
adapted ground BR state.
Treatment of Diffraction Data
Determination of lattice parameters (P63 :  =  = 90°,  = 120°; ground sate aBR = bBR =
60.81 Å, cBR = 110.64 Å; illuminated aI = bI = 61.08 Å, cI = 110.40 Å), integration, scaling and
merging of the intensities (ground state 20-2.0 Å, Rsym = 4.7 %, illuminated 20-2.25 Å, Rsym =
8.2 %) were done using the HKL suite 5 (Table 1). Conversion to amplitudes was performed
with the programs TRUNCATE and MTZUTIL of the CCP4 program package 6. CNS 7,
version 0.4, was used for most of the further processing of the data. For quality evaluation
about 5 % of the data were selected with the program xdlDATAMAN 8 in resolution shells to
be used only for calculation of the Rfree-value. The twinning ratio was first determined by
using
the
twinning
server
at
UCLA
university
(http://www.doe-
mbi.ucla.edu/Services/Twinning/, Yeates, T.O.) and then refined by inspecting the quality of
the Rfree after rigid body refinements with different twinning ratios. The used values of the
twinning ratio are 0.44 for the ground state crystal and 0.38 for the illuminated one. To cope
with the problem of twinning, the available CNS version was slightly modified in the way that
for any comparison of measured amplitudes with model amplitudes a twinned version of the
model amplitudes is used and for difference calculations, in addition, the intensities were
detwinned with the known equations 9 before used in density calculations.
Ground state refinement: As a starting model the one of Luecke et. al. 10, PDB:1BRX, was
used in normal molecular replacement and rigid body refinement calculations with CNS. The
model of the ground state was then further refined through several circles of simulated
annealing calculations (using 4000 K and the standard slow cool protocol
11
) and model
building with the molecular graphics program 12 O to complete the E-F loop, to extend the Cterminus to amino acid residue Ser239 and to reduce the model bias. In the beginning retinal
and water molecules were not included in the refinement. Residues Arg82, Asp85, Asp96,
Glu194, Glu204 and Lys216 were omitted and their conformations carefully checked. At a
later stage of the refinement, water molecules were selected with the standard input in CNS
and especially the internal ones were checked during further refinement for proper density.
Elongated densities in the lipid region were assigned to carbon chains of various length to
simulate possible lipid or detergent molecules but without trying to determine real models of
lipids or detergents. The final R-values are R = 17.9 % and Rfree = 20.8 %, corresponding to an
estimated coordinate error using Rfree of r = 0.211 Å, from the Luzzati plot, and r = 0.131Å,
from the SIGMAA analysis, as determined with standard input-files of CNS.
5
M state refinement: The BR model was first oriented in the new unit cell with a molecular
replacement calculation and a rigid body refinement. Six different approaches have been used
for cross-checking the molecular alterations found for the M2 intermediate: (i) Fourier
difference density maps (FoI - FoBR , of the observed structure factors of the illuminated and
the ground state crystal) at different resolutions have been calculated and used to locate the
changes in the molecule visually with the graphics program 12 O. (ii) A model of two complete
bR molecules including water molecules and hydrocarbon chains, one declared as M and one
as BR was refined against the data. During the energy minimization refinement the atomic
positions of the BR model were fixed whereas the atomic coordinates of the M model with the
complementary occupancy were allowed to change. In the same way simulated annealing
calculations with standard slow cool protocol 11 at 2500 K and 4000 K were carried out which
resulted in M state models comparable to that obtained from simple energy minimisation.
Larger changes were observed in the loop regions. (iii) The observed structure factors FoI of
the illuminated crystal contain a fraction s of the BR state and a fraction 1-s of the M state.
Therefore, extrapolateded difference density maps
13
Δρex = (│FoI│ - s│FcBR│)exp(φcBR) ∕
(1-s) were calculated using amplitudes FcBR and phases φcBR of the ground state model. These
maps should show the electron density of the M intermediate for the correct fraction s of the
ground state contribution to FoI. (iv) Omit density maps
14
Δρexo were calculated by the Δρex
formula using phases φcBR determined without including the retinal and/or Arg82, Asp85,
Glu194, Glu204, Lys216, Phe219 and all water molecules (Fig. 2a). These maps were used for
visual comparison of the M state conformations of these residues with the resulting densities.
(v) Complete omit density maps Δρo were calculated with φcBR as well as FcBR determined
from a BR model without the above mentioned amino acids and/or the retinal. Varying
fractions of the M and BR state conformations of the omitted residues were correlated to the
Δρo densities to define the fraction of the M conformation. The correlations resulted in about
65 % 13-cis retinal and only about 35 % M state conformation for the respective amino acid
side chains. If we consider that the M intermediate consists of the early and the late M state
contributions and assume that the early M state structure of these residues are very similar to
their BR state conformations as verified at low resolution, one would draw the conclusion that
on average 35 % of bR molecules in an illuminated crystal are in the ground state, another 35
% in the late M state and 30 % in the early M. (vi) Selecteded difference density maps Δρs =
(│FoI│ - 0.65│FcBR│)exp(φcM) ∕ (1-0.65) were calculated using amplitudes FcBR of the
ground state model and phases φcM of the M intermediate model (Fig.2b). These maps show
6
the final electron density of the M intermediate. The final R-values of the M state model,
including the fixed ground state model, are R = 16.7 % and Rfree = 23.6 %, corresponding to
an estimated coordinate error using Rfree of r = 0.279 Å, from the Luzzati plot, and r =
0.239 Å, from the SIGMAA analysis, as determined with standard input-files of CNS.
The refinement statistics is summerized in Table 2 and in the header of the Protein Data Bank
file 1CWQ.
Reference List
1. Nicholls, A., Sharp, K.A. & Honig, B. Protein folding and association: insights from the
interfacial and thermodynamic properties of hydrocarbons. Proteins 11, 281-296 (1991).
2. Landau, E.M. & Rosenbusch, J.P. Lipidic cubic phases: a novel concept for the
crystallization of membrane proteins. Proc Natl Acad Sci U S A 93, 14532-14535
(1996).
3. Ormos, P. Infrared spectroscopic demonstration of a conformational change in
bacteriorhodopsin involved in proton pumping. Proc Natl Acad Sci U S A 88, 473-477
(1991).
4. Zimanyi, L. & Lanyi, J.K. Deriving the intermediate spectra and photocycle kinetics
from time- resolved difference spectra of bacteriorhodopsin. The simpler case of the
recombinant D96N protein. Biophys J 64, 240-251 (1993).
5. Otwinowski, Z. & Minor, W. Processing of X-Ray Diffraction Data Collection in
Oscillation Mode. Methods Enzymol 276, 307-326 (1997).
6. Collaborative Computational Project, No.4. The CCP4 suite: programs for protein
crystallography. Acta Crystallogr D 50, 760-763 (1994).
7. Brunger, A.T. et al. Crystallography & NMR system: A new software suite for
macromolecular structure determination. Acta Crystallogr D 54, 905-921 (1998).
7
8. Kleywegt, G.J. & Jones, T.A. xdlMAPMAN and xdlDATMAN-Programs for
reformating, analysis and manipulation of biomacromolecular electron-density maps and
reflection data set. Acta Crystallogr D 52, 826-828 (1996).
9. Yeates, T.O. Detecting and overcoming crystal twinning. Methods Enzymol 276, 344358 (1997).
10. Luecke, H., Richter, H.T. & Lanyi, J.K. Proton transfer pathways in bacteriorhodopsin at
2.3 angstrom resolution. Science 280, 1934-1937 (1998).
11. Adams, P.D., Pannu, N.S., Read, R.J. & Brunger, A.T. Extending the limits of molecular
replacement through combined simulated annealing and maximum-likelihood
refinement. Acta Crystallogr D Biol Crystallogr 55 ( Pt 1), 181-190 (1999).
12. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Improved methods for binding
protein models in electron density maps and the location of errors in these models. Acta
Crystallogr A 47, 110-119 (1991).
13. Genick, U.K. et al. Structure of a protein photocycle intermediate by millisecond timeresolved crystallography. Science 275, 1471-1475 (1997).
14. Bhat, T.N. & Cohen, G.H. OMITMAP: An electron density map suitable for the
examination of errors in a macromolecular model. J Appl Cryst 17, 244-248 (1984).
15. Evans, S.V. SETOR: hardware-lighted three-dimensional solid model representations of
macromolecules. J Mol Graph 11, 134-138 (1993).
8