J. Mol. Biol. (1997) 269, 861±880
Crystal Structure of an Oligomer of Proteolytic
Zymogens: Detailed Conformational Analysis of the
Bovine Ternary Complex and Implications for
Their Activation
F. X. Gomis-RuÈth1*, M. GoÂmez-Ortiz1, J. Vendrell1, S. Ventura1
W. Bode2, R. Huber2 and F. X. AvileÂs1
1
Institut de Biologia
Fonamental i Departament de
BioquõÂmica i Biologia
Molecular, Universitat
AutoÁnoma de Barcelona
08193 Bellaterra, Spain
2
Abteilung fuÈr
Strukturforschung, MaxPlanck-Institut fuÈr Biochemie
82152 Martinsried, Germany
The pancreas of ruminants secretes a 100 kDa non-covalent ternary complex of the zymogen of a metalloexopeptidase, carboxypeptidase A, and
the proforms of two serine endopeptidases, chymotrypsin C and proteinase E. The crystal structure of the bovine complex has been solved and
re®ned to an R-factor of 0.192 using synchrotron radiation X-ray data to
Ê resolution. In this heterotrimeric complex, the 403 residue procar2.35 A
boxypeptidase A takes a central position, with chymotrypsinogen C and
proproteinase E attached to different surface sites of it. The procarboxypeptidase A subunit is composed of the active enzyme part and the 94
residue prodomain, similar to the monomeric porcine homologous form.
The 251 residue subunit chymotrypsinogen structure, the ®rst solved of
an anionic (acidic pI) chymotrypsinogen, exhibits characteristics of both
chymotrypsinogen A and elastases, with a potential speci®city pocket of
intermediate size (to accommodate apolar medium-sized residues)
although not properly folded, as in bovine chymotrypsinogen A; this
pocket displays a ``zymogen triad'' characteristic for zymogens of the
chymotrypsinogen family, consisting of three non-catalytic residues (one
serine, one histidine, and one aspartate) arranged in a fashion similar to
the catalytic residues in the active enzymes. Following the traits of this
family, the N terminus is clamped to the main molecular body by a disulphide bond, but the close six residue activation segment is completely
disordered. The third zymogen, the 253 residue proproteinase E, bears
close conformational resemblance to active porcine pancreatic elastase; its
speci®city pocket is buried, displaying the second ``zymogen triad''. Its
®ve N-terminal residues are disordered, although the close activation site
is ®xed to the molecular surface. The structure of this native zymogen
displays large conformational differences when compared with the
recently solved crystal structure of bovine subunit III, an N-terminally
truncated, non-activatable, proproteinase E variant lacking the ®rst 13
residues of the native proenzyme.
Most of the prosegment of procarboxypeptidase A and its activation sites
are buried in the centre of the oligomer, whilst the activation sites of chymotrypsinogen C and proproteinase E are surface-located and not
involved in intra or inter-trimer contacts. This organization confers a
functional role to the oligomeric structure, establishing a sequential proteolytic activation for the different zymogens of the complex. The large
surface and number of residues involved in the contacts among subunits,
Present address: F. X. Gomis-RuÈth, Centre d' Investigacio i Desenvolupament C.S.I.C.; Jordi Girona 18-26; 08034
Barcelona, Spain. (E-mail: xgrcri@cid.csic.es)
Abbreviations used: AS, activation segment; b, bovine; CTGA and CTGC, chymotrypsinogen A and C; EL, pancreatic
elastase; p, porcine; (P)CPA and (P)CPB, (pro)carboxypeptidase A and B; PDB, Brookhaven Protein Data Bank; (P)PE,
(pro)proteinase E; PRTR, prethrombin 2; (P)SBT, (pro)subtilisin; r.m.s, root-mean-square; TC, ternary complex;
Ê ˆ 0.1 nm.
1A
0022±2836/97/250861±20 $25.00/0/mb971040
# 1997 Academic Press Limited
862
Structure of a Complex of Zymogens
as well as the variety of non-bonded interactions, account for the high
stability of the native ternary complex.
# 1997 Academic Press Limited
*Corresponding author
Keywords: X-ray crystal structure; procarboxypeptidase A;
chymotrypsinogen C; proproteinase E; serine proproteinases
Introduction
The current view on the molecular determinants
of the structure, activation and functionality of
pancreatic digestive proprote(in)ases and active
prote(in)ases is guided by the knowledge on a few
members of these proteins, namely trypsinogen,
chymotrypsinogen A, elastase and procarboxypeptidase A1, considered as good models for the rest
of the members of their respective families, serine
proteinases and metallocarboxypeptidases. This
view disregards the signi®cant differences (or
unknowns) from several other members of these
families. This is the case for pancreatic procarboxypeptidases A (PCPA), which occur as A1 and A2
subfamilies in mammals (AvileÂs et al., 1993). Similarly with pancreatic chymotrypsinogens, there is a
signi®cant lack of knowledge about the three-dimensional structure and functionality of chymotrypsinogens B and C, among other, and their
occurrence in different species (Rinderknecht,
1993). There is even more confusion concerning the
family of the elastase and elastase-like proteins,
such as proteinases E, in relation to the molecular
properties of some isoforms and their speci®cities
and the basis of their ability or inability to degrade
elastin (Rinderknecht, 1993), leading to the grouping of these proteins into several subfamilies (elastases I, IIa, IIb, IIIa, IIIb and IV), whose structural
and functional differences have not been fully clari®ed (Largman et al., 1976; Shen et al., 1987; Tani
et al., 1988; Rinderknecht, 1993). We know equally
little about the oligomeric complexes of zymogens
of serine proteinases and metallocarboxypeptidases
found in several ruminant (ox, goat and camel;
Kerfelec et al., 1985) and non-ruminant species
(pig, human and rat; Kobayashi et al., 1978;
Pascual et al., 1990; Oppezzo et al., 1994), although
they constitute a major component of pancreatic
juice.
Some clari®cation may be achieved by studies of
the ternary complex (TC) of bovine procarboxypeptidase A (bPCPA), chymotrypsinogen C
(bCTGC) and proproteinase E (bPPE), a form that
can be isolated either from bovine pancreas or pancreatic juice (Brown et al., 1961; Kobayashi et al.,
1981a,b; Pascual et al., 1990). This complex was initially termed procarboxypeptidase A-S6, because
this proenzyme is the major and ®rst identi®ed
component of the complex and because it has a
sedimentation coef®cient of 6 S (Brown et al., 1963).
The three zymogens of the complex were named
as subunit I, II and III, although the latter, proproteinase E, was initially isolated and studied as an
N-terminally truncated form. The characterization
of the three constituting zymogens and the complex did not go so far, and a three-dimensional
structure was available for the enzyme core of only
the ®rst one, carboxypeptidase A (CPA), obtained
by complete autolysis and disaggregation (Rees
et al., 1983).
It should be added that it has not been fully
clari®ed yet whether the bPCPA subunit from the
bovine TC belongs to the A1 or to the A2 subfamily (Faming et al., 1991; AvileÂs et al., 1993) and that
no previous conformational information about its
activation segment (94 residues) was available. It
has been proposed but not shown that the bCTGC
subunit is homologous to the monomeric chymotrypsinogen C from porcine pancreas (Folk &
Schirmer, 1965), sharing an anionic character
(acidic pI) with chymotrypsinogen B, in contrast to
the cationic character of chymotrypsinogen A
(CTGA). Until now, no three-dimensional structure
of an anionic chymotrypsinogen has been solved.
On the other hand, the bPPE subunit is homologous to human and porcine proproteinases E (Shen
et al., 1987; Cambillau et al., 1988; AvileÂs et al.,
1989), all of them anionic, and evolutionarily related to proelastases I and II (Shirasu et al., 1988),
which are cationic. The lack of activity against elastin of the enzymes derived from the former group
of proenzymes allows a primary differentiation
from the latter. A clear differentiation would require detailed conformational comparisons, which
have not been possible due to the lack of available
three-dimensional structures of proteinases E. Recently, the crystal structure of the isolated, N-terminally truncated bovine proproteinase E, lacking its
®rst 13 residues, has been reported (Pignol et al.,
1994); this is a form derived from the native component of the TC by autolysis (Pascual et al., 1990).
This form, however, shows a reorganization of the
active site compared to serine endopeptidases
(such as porcine pancreatic elastase (pEL; Meyer
et al., 1988), chymotrypsin (Matthews et al., 1967)
or trypsin (Huber et al., 1974)), and is inactive and
cannot be activated.
With the aim of clarifying some of the abovementioned questions, we have crystallized the bovine TC and undertaken an X-ray crystallographic
study. Previously (Gomis-RuÈth et al., 1995b), we reported an overall view of the TC three-dimensional
structure, of the relative location of the subunits,
and of some properties of these components. Here,
we present the re®ned structure obtained by synchrotron X-ray analysis, in which the detailed conformational features of each subunit are related to
Structure of a Complex of Zymogens
potential functionality, activity, speci®city and proteolytic activation. The involvement of the most
important residues, among the 907 residues of the
complex (403 ‡ 252 ‡ 253 residues), in the structural and functional determinants of these proteins
has been carefully analysed. Also, the mutual in¯uence of the subunits in the proteolytic activation
processes is discussed. Finally, the study intends to
give insights into the structure and role of oligomeric complexes of digestive proteases, a ®eld still
with many unknowns.
Results and Discussion
Crystal packing of the oligomers and
quaternary structure
In the present crystal structure, one TC heterotrimer is in the asymmetric unit and nine oligomers
are present in the hexagonal unit cell. The assignment of each of the three different subunits to a
distinct TC can be made on the basis of the dominating number of intra versus inter-complex contacts. In agreement with previous studies in
solution (Puigserver & Desnuelle, 1977; Kerfelec
et al., 1985; Michon et al., 1991a), the central PCPA
molecule holds the complex in a Y-shaped disposition (see Figure 1). The two serine proproteinases
do not interact with one another in the TC but are
attached to two sides of the PCPA subunit, at an
Ê2
angle of about 80 . The contact surfaces of 1239 A
Ê 2 (bPCPA-bCTGC) are
(bPCPA-bPPE) and 1151 A
large and account for the high stability of the ternary complex against denaturants (AvileÂs et al.,
1993). The contacts with crystal symmetry-related
863
molecules are small and essentially made via
bPCPA-bPCPA and bPPE-bCTGC.
The structure of the central subunit, bPCPA
The three-dimensional structure of the PCPA
subunit in the bovine TC is similar to those previously solved for the monomeric porcine procarboxypeptidase A1 (pPCPA1; Guasch et al., 1992)
and for the monomeric porcine procarboxypeptidase B (pPCPB; Coll et al., 1991). Accordingly,
bPCPA, like its porcine homologous forms, consists
of two clearly distinguishable regions, the N-terminal 94 residue activation segment and the larger
309 residue enzyme moiety (see Figure 2).
The activation segment region (AS) is folded in a
globular domain (residues Lys4A to Glu80A; for
nomenclature see Material and Methods and PDB
access code 1pyt) followed by an extended segment (Asp81A to Arg99A) connecting it to the enzyme moiety. This AS almost completely covers
the CPA active-site groove, establishing several
interactions with residues important for substrate
recognition, but does not directly block the activesite residues. It particularly exhibits a loop directed
towards the catalytic zinc ion of the enzyme moiety and blocks the access of larger substrates to the
active-site cavity. The AS globular domain exhibits
an open-sandwich antiparallel a/antiparallel b
structure, with a four-stranded b-sheet covered by
two a-helices (see Figure 3). The b-sheet faces the
active site of the enzyme, which is blocked by a
loop protruding from the former. Similar to
pPCPA1, the connecting region folds in a threeturn a-helix and an adjacent open large loop that,
Figure 1. Stereo-ribbon plot of the TC in the chosen standard orientation, with the central AS of bPCPA (red) clamping the three subunits bCTGC (blue, to the left), bPPE (green, on the top), and the CPA moiety of bPCPA (yellow).
The two localized ions, calcium in bPPE (orange sphere) and zinc in CPA (magenta sphere) and the termini of each
chain are labelled.
864
Structure of a Complex of Zymogens
Figure 2. Stereo-cartoon showing the bovine (thick trace) and the porcine (thin line) PCPA Ca-structures displayed in
the same orientation as in Figure 1 after optimal least-squares-®tting of the corresponding CPA domains. As
observed, the AS present a slight relative displacement to each other. The termini of bPCPA (Lys4A and Tyr309), and
the last residue of the AS (Arg99A) are labelled.
Figure 3. (a) A diagram showing the regular secondary structure elements (b1-a1-b2-b3-a2-b4) of the activation
domains of bPCPA and PSBT (Gallagher et al., 1995). (b) Superimposition after least-squares-®t of the a-carbon structures of the AS of bPCPA (thick trace) and PSBT (thin line). Some residues of bPCPA (suf®x A) and of PSBT (pre®x
P) are labelled.
Structure of a Complex of Zymogens
through the primary proteolytic cleavage site, connects to the N terminus of the enzyme. Presumably
due to several interactions with the N-terminal region of the bCTGC subunit (see below), this activation loop exhibits a well-de®ned conformation.
These interactions might have an in¯uence on its
accessibility and the rate of its proteolytic cleavage.
This is in contrast to the mobile character of the
same loop in pPCPA1 and pPCPB (Coll et al., 1991;
Guasch et al., 1992). The potential primary site for
tryptic cleavage, the peptide bond Arg99A-Ala1, is
only slightly exposed; the four residues ¯anking
this bond exhibit main-chain conformation angles
close to those observed for canonical reactive site
loops of protein serine proteinase inhibitors (Bode
& Huber, 1994; Hubbard et al., 1994). The second
potential tryptic cleavage site, at Arg74A-Tyr75A,
is at the edge strand of the AS b-sheet; its conformation does not seem to be in¯uenced upon formation of the TC, although its accessibility to
activating proteases is restricted (see Figure 7a).
The structure of the bPCPA enzyme moiety is
very similar to that previously reported for active
bovine CPA (Rees et al., 1983); the r.m.s. deviation
Ê for the topologibetween both structures is 0.45 A
cally equivalent 308 residues. It shows a central
mixed b-sheet ¯anked on both sides by several ahelices packed against it. The active-site cavity is
located close to the C terminus of the central parallel b-strands and contains the (fully occupied)
catalytic zinc ion. Close to this cation are the coordinating protein residues His69, His196 and
Glu72, as well as those proposed to be involved in
the enzymatic mechanism, Glu270, Arg127,
Tyr248, Arg145 and Arg71 (Christianson &
Lipscomb, 1989; Kim & Lipscomb, 1991). No signi®cant change is observed on comparing both
structures. Just the exposed surface-located loop
132-137 displays a somewhat different conformation in bPCPA compared with CPA.
An apparent difference detected between the
structure of the bPCPA subunit and that of monomeric pPCPA1 (Guasch et al., 1992) is a slight rotation between the AS and the enzyme moiety.
Separate ®ts of both molecular parts, after omission
of four residues ¯anking the primary activation
site and ten residues that belong to surface loops,
yields deviations of the common a-carbon atoms of
Ê and 0.38 A
Ê , respectively (see Figure 2). The
0.50 A
bovine proenzyme contains an additional residue
(Tyr309) at the C terminus with respect to the porcine counterpart. The side-chain of this residue, as
well as that of the penultimate Leu308, is buried in
the molecular surface. Besides, the carboxylate
group of the C terminus is involved in an intramolecular salt-bridge with Lys190 Nz, which clamps
the C-terminal region of the proenzyme to the
main molecular body. Both features probably protect the proenzyme against autolysis.
Overall, the bPCPA subunit ®ts better with the
structure of a CPA1-like enzyme than with a CPA2
form, the new subdivision of the A forms (Gardell
et al., 1988; Faming et al., 1991; AvileÂs et al., 1993).
865
Among other distinct structural properties, it contains a side-chain in position A47 (GlnA47), which
establishes a hydrogen bond with the Ile244 carbonyl oxygen atom, similar to other equivalent residues in this position in A1 forms (porcine and
human; Guasch et al., 1992; Catasus et al., 1992). In
this position, the A2 forms display residue Cys244
disulphide-bridged to Cys210, a bridge that hinders the establishment of a similar A47-244 connection. This is of importance, as it affects the activesite ``shaping'' loop formed by residues 243 to 252,
the conformation of which is clearly different from
rat CPA2 (Faming et al., 1991). The occurrence of a
single disulphide bridge in the enzyme moiety of
the bPCPA subunit (between Cys138 and Cys161),
like in porcine CPA1, in contrast to the two disulphide bridges in CPA2-like enzymes (Cys138Cys161 and Cys210-Cys244; rat and human) is also
one of the features that favours classi®cation of
bPCPA into the A1 family. Furthermore, at position 268 bPCPA contains a threonine residue,
which forms an S01-subsite more appropriate to receive medium-sized groups from the substrate, as
usual for A1-type enzymes, than aromatic groups,
as usual for A2-type enzymes, with alanine at this
position (Catasus et al., 1995).
Comparison of the bPCPA activation segment
with the prosegment of prosubtilisin
The globular region of the bPCPA AS has been
shown to display a folding topology similar to
other protein domains with no obvious functional
relationship, such as the C-terminal fragment of
the ribosomal protein L7/L12, the RNA-binding
domain of the U1 small nuclear ribonucleoprotein
A and the N-terminal fragment of ribulose-biphosphate carboxylase (Coll et al., 1991; Guasch et al.,
1992). Recently, the structure of the serine endopeptidase subtilisin BPN0 from Bacillus amyloliquefaciens in complex with its prosegment has been
reported (Bryan et al., 1995; Gallagher et al., 1995;
PDB access code 1spb). In this zymogen-resembling complex, the AS is required for the proper
folding of the prosubtilisin (PSBT) molecule and is
autocatalytically removed during maturation. The
activation is postulated to cause a rearrangement
of the newly generated N terminus of the mature
subtilisin (SBT) molecule and movement over
Ê (Gallagher et al., 1995). In contrast to
about 25 A
PCPA and PCPB (AvileÂs et al., 1993), the globular
part of the AS does not directly cover the active
site of subtilisin, but is attached laterally to sites of
the cleft adjacent to the catalytic residues, with the
subsequent connecting segment running across the
active site, the extended, substrate-like binding Cterminal chain interacts with the active-site cleft,
causing inhibition. As the AS displays inhibitory
activity towards mature SBT, the separation of this
AS from the complex after activation is proposed
to occur by means of additional (auto)catalytic
cleavage events in the propart (Gallagher et al.,
1995).
866
The AS structure of PSBT is quite homologous to
the corresponding segment of bPCPA, despite no
signi®cant sequence homology. Likewise, the globular part is mainly composed of a four-stranded
antiparallel sheet and two helices of identical connectivity (see Figure 3(a)). Both structures may be
superimposed (see Figure 3(b)) with 58 structurally
equivalent Ca atoms (out of 71 residues in PSBT
Ê after optiand 94 in bPCPA) deviating by 2.31 A
mal least-squares ®t. Both N termini are surfacelocated and start at the same side (see Figure 3(b))
of the globular moiety and become topologically
equivalent from His10A (bPCPA) andGluP7
(PSBT) to Ala17A (PheP14), forming the ®rst strand
b1. A turn leads in both structures to helix a1 (see
Figure 3), slightly rotated and shifted in PSBT as
compared with PCPA, further connected to strand
b2 (Glu342-Arg39A of bPCPA, equivalent to
GlyP34-CysP39 of PSBT). At the end of strand b2,
the major differences between both structures
arise; the loop connecting strands b2 and b3 is
folded towards helix a1 in the AS of bPCPA as
compared with the equivalent part in PSBT. The
topological equivalence is continued at Pro51A
(AlaP46 in PSBT) of strand b3 and almost completely maintained beyond the second helix a2 up to
Phe67A (AspP62), including the turn connecting
both secondary structure elements b3 and a2. The
turn ®nally connecting a2 and b4 is three residues
longer in the bPCPA AS than in PSBT AS. At
Ile79A (AspP71), the main chains take different
courses and display distinct lengths: 20 residues in
the bPCPA AS to join the ®rst residue of the mature enzyme moiety at the left surface according
to Figure 3(b); seven residues in the cleaved form
of PSBT entering the active-site cleft of mature
SBT.
In summary, both propeptidases bPCPA and
PSBT are inhibited by AS of very similar topology,
but by completely different mechanisms. While the
active-site groove of CPA is covered by the compact domain of the activation segment in PCPA,
thus preventing the access of major substrates to
the active-site residues, it is solely the C terminus
of the AS that inhibits SBT in the PSBT structure.
The structure of subunit II, a
new chymotrypsinogen
According to its sequence and speci®city, the
bCTGC subunit can be classi®ed as a hybrid between a classical chymotrypsinogen and a proelastase according to sequence and speci®city analysis:
it exhibits 41% sequence identity with bovine
chymotrypsinogen A (bCTGA), 56% with bovine
proproteinase E (bPPE), and 54% with porcine
pancreatic elastase (pEL), and shows a preference
for medium-sized apolar residues at P1 (Folk &
Schirmer, 1965; Keil-Dlouha et al., 1972). This
classi®cation is in agreement with the three-dimensional structure. Thus, on one hand, its N-terminal
(albeit shorter) sequence starts with a cysteine residue, Cys701 (see Materials and Methods for num-
Structure of a Complex of Zymogens
bering), which (as in bCTGA (Cys1-Cys122)) is
disulphide-linked to Cys822. On the other hand, it
exhibits several insertions present also in pEL, such
as after positions 736, 763, 870, 888, 917 and 921.
The scaffold of the bCTGC subunit displays similar
r.m.s deviations from both (pro)proteinase protoÊ and 1.11 A
Ê in the comparisons with
types: 1.09 A
204 topologically equivalent a-carbon atoms of
bCTGA (see Figure 4(a)) and 198 a-carbon atoms
of pEL after omission of all mobile and some
strongly deviating residues. It may be mentioned
that bCTGA and pEL are cationic molecules,
whereas bCTGC is anionic.
The main structural elements characteristic for
chymotrypsin(ogen)-like serine (pro)proteinases
found in bCTGA and pEL are also in the bCTGC
subunit (see Figure 4(a)). Thus, the protein folds in
two b-barrel domains of six antiparallel strands
each, forming the ®rst four strands of a Greek key
motif and a b-hairpin the last two. Besides, the
structure contains two a-helical segments, a short
one in the middle of the protein (Asp864 to Ser869)
and another one at the C terminus (Ile935 to
Leu944). Several surface-located segments directly
related to the structure, activity and activation of
the protein are identi®ed in bCTGC as in the homologous forms.
The N-terminal region of bCTGC including the
``activation loop'' shows similarities to and differences from homologous serine proteinases. The
®rst eight residues of the 13 residue activation peptide, Cys701 to Pro708, are located on the molecular surface of the enzyme, as in bCTGA. The
subsequent six residue activation loop of sequence
Ser711-Ala712-Arg715*Val716-Val717-Gly718 ¯anking the scissile peptide bond (*) is fully exposed
and has a poorly de®ned conformation, making
this region a good target for tryptic hydrolysis. In
bCTGA, in contrast, the equivalent longer loop
(carrying two additional residues before the cleavable arginine residue) is arranged at the molecular
surface in a partially helical, well-de®ned noncanonical conformation; this loop is ®xed via two
hydrogen bonds between Val17 (bCTGA numbering; Wang et al., 1985) and the parallel segment
Gly187 to Ser189; in this case, however, Val17 (the
P02 residue in the tryptic proteolysis) exhibits an
unusual high-energy conformation (with mainchain angles located in high-energy regions of the
Ramachandran plot), which presumably facilitates
liberation of the scissile segment and attack by
trypsin.
The ®rst domain of bCTGC also contains the
potential (metal-lacking) ``calcium-binding loop'',
running from Lys770 to Ser781, the conformation
of which is similar to that of bovine trypsin/trypsinogen (Bode & Schwager, 1975; Fehlhammer et al.,
1977) and pEL (Meyer et al., 1988) in spite of an additional residue. The two calcium ligands found in
the calcium-binding species, Glu70 and Glu80
(bCTGA numbering), are replaced in bCTGC by a
lysine and a serine residue, respectively; this is reminiscent of Lys70 in thrombin (Bode et al., 1989)
Structure of a Complex of Zymogens
and Arg70 in porcine pancreatic kallikrein (Bode
et al., 1983). In contrast to thrombin, however, the
Lys770 side-chain of CTGC is not ®xed by intramolecular hydrogen bonds, but points away from this
867
loop; instead, Glu778 extends the loop from the opposite side, ®xing it through hydrogen bonds
made with the amide nitrogen atoms of residues
Asn771 and Asn772.
Figure 4. Stereo Ca plot approximately in the traditional standard orientation of chymotrypsin-like serine endopeptidases (see Huber & Bode, 1978) displaying, after optimal least-squares-®tting, the structures of the following (pro)enzyme pairs: (a) bCTGC (thick trace; ¯exible segments in the thinnest line) and bCTGA (thin tracing); some bCTGC
residues are labelled. (b) bCTGC (thick trace; ¯exible segments in the thinnest line) and bPPE (thin tracing); some
bCTGC residues are labelled. (c) bPPE (thick tracing) and pEL (thin line); some bPPE residues are labelled.
868
Adjacent to the calcium-binding loop of bCTGC
is the ``autolysis loop''. The second part of it is
very mobile while its ®rst part forms part of the
rim of the active-site cavity, with Leu844 partially
®lling it. This structure can be compared with
those of trypsinogen (Fehlhammer et al., 1977) and
bCTGA (Wang et al., 1985) where the autolysis
loop is completely and partially disordered, respectively.
A key region in bCTGC is the ``activation domain'', the central part of which is formed by residues 889 to 895. Presumably, after cleavage of the
activation loop, the oxyanion hole is formed,
which functions in stabilizing the tetrahedral transition intermediate during catalysis. If it follows
the same rearrangement as in other serine proteinase zymogens (Bode & Huber, 1986), the newly liberated amino group of Val716 should become
engaged in an internal salt-bridge with Asp894. In
bCTGC, as in bCTGA, the 889 to 895 region is
folded in such a way that it blocks the entrance of
substrates into the speci®city pocket, in contrast to
its exposure in active serine proteinases. Like in
trypsinogen and bCTGA, a ``zymogen triad'' can
be observed in the bCTGC structure, with Ser732
bound to His740 and this to Asp894 (Bode, 1979;
Madison et al., 1993) through a hydrogen bond/
salt-bridge network. This network helps to keep
the activation domain in an internal, de®ned,
location.
The essential residues in the catalytic site,
Ser895, His757 and Asp802, the ``catalytic triad'',
show similar positions and conformations in
bCTGC and in bCTGA. As expected, differences in
sequence and conformation are observed in the
speci®city pocket. Thus, the entrance frame-forming segment is rigid from Val913 to Gly916; the following kink-forming segment up to Cys920 bows
outward in a manner intermediate between
bCTGA and pEL (Wang et al., 1985; Meyer et al.,
1988), presumably due to the insertion at Gly917A,
reminiscent of porcine elastase. The disulphide
bridge Cys891-Cys920 connecting the entrance of
the speci®city pocket with the base exhibits a righthanded conformation and is differently placed
compared with all related structures. It should be
noted that in two crystallographically independent
(but chemically identical) bCTGA molecules
(Wang et al., 1985), the equivalent segment Val213Cys220 and the disulphide bond showed different
conformations, indicating a certain degree of ¯exibility. More important yet, due to the glycine residue in position 916 and a threonine residue in
position 926, the speci®city pocket generated in
bCTGC after activation cleavage should have a
size intermediate between those of bovine chymotrypsin A (Matthews et al., 1967) and pEL, that is
shaped to accommodate medium-sized residues.
This is in agreement with the chymotrypsin C preference for peptide substrates with leucine residues
at P1 (Keil-Dlouha et al., 1972) and with its weaker
activity towards tryptophan-containing substrates,
Structure of a Complex of Zymogens
as compared with chymotrypsin A, with a glycine
residue at position 926.
Structural determinants for the functionality
and inactivity of the third subunit, bPPE
The bPPE subunit may be classi®ed as a typical
proelastase considering its sequence (53% identity
with pEL) and speci®city, with a preferential cleavage after alanine, valine, serine or threonine residues (Kobayashi et al., 1978, 1981a,b). Its high level
of identity (85%) identi®es it as the real equivalent
to human proteinase E (Mallory & Travis, 1975;
Shen et al., 1987); thus, our bPPE model can be considered as a prototype for this human enzyme. The
human proteinase E has been shown to be identical
with the cholesterol-binding protein (Sziegoleit,
1982), potentially involved in transport and metabolism of cholesterol. Like bCTGC, bPPE exhibits
the characteristic features and overall folding topology of the trypsin/chymotrypsin family of serine
endopeptidases. Similar to pEL, upon activation,
the 11 residue N-terminal activation peptide is removed (AvileÂs et al., 1989). bPPE exhibits several
typical elastase-like insertions, such as after positions 436 (for numbering, see Materials and
Methods), 499, 570, 617 and 621. In the activated
enzyme, the side-chains of Val616 and Thr626
(same residues in pEL) will form an S01-pocket
shaped to accommodate small or medium-sized
side-chains (Kobayashi et al., 1981a,b). The introduction of an additional disulphide bond (Cys498Cys499B) does not lead to any signi®cant mainchain adaptations on comparison with the pEL
structure, and the two extra residues Ala486A
and Gly486B are allowed for by a small bulge
(see Figure 4(c)). The 198 topologically equivalent
a-carbon atoms of bPPE and pEL show a r.m.s.
Ê , compared with 1.24 A
Ê for
deviation of 0.91 A
196 equivalent a-carbon atoms of bCTGA (41%
sequence identity between bPPE and bCTGA).
Just three peptide segments signi®cantly diverge
in bPPE from the corresponding parts of pEL;
namely, those arranged around the speci®city
pocket, mainly re¯ecting conformational changes
occurring upon activation cleavage. Therefore, pEL
can be considered as a valid homology model for
active proteinase E. In analogy with bCTGC (see
Figure 4(b)) and bCTGA, segment Ser589 to
Asp594, which will form the ``lower lip'' of the S1pocket after activation, protrudes into the molecule
in bPPE, and the associated Asp594 side-chain projects out of the molecule forming with His440 and
Ser432 the ``second triad'' typical for proenzymes
of the serine endopeptidase family (Madison et al.,
1993). The preceding segment Gly585 to Gly590 is
ordered in bPPE, but takes a slightly different
course compared to pEL (see Figure 4(c)). The entrance-frame-forming peptide segment Thr613 to
Cys620 is, in contrast to bCTGA, bCTGC and trypsinogen (Wang et al., 1985; Bode et al., 1976;
Fehlhammer et al., 1977) uniquely de®ned in bPPE
and almost superimposable on the equivalent seg-
Structure of a Complex of Zymogens
ment of pEL (see Figure 4(c)); the adjacent disulphide bridge Cys591-Cys620 essentially permits
the considerable shift of Cys591 by a large internal
reorganization.
The extended surface-located N-terminal segment of bPPE is arranged in a hairpin loop-like
manner ``below'' the preformed speci®city pocket,
forming intramolecular contacts essentially via the
Val417 side-chain and through two polar hydrogen
bonds (Lys554 Nz-Val416 O and Asp421 OdAsn418 N). From Ser410 to Asn418, it simultaneously interacts with an adjacent symmetryrelated bCTGC molecule between Arg415 and
Asn418 forming a parallel three-rung ladder with
segment Thr787 to Val790 of this bCTGC molecule.
From Asn418 to Asp421, the bPPE chain folds via a
regular 1,4 tight turn of type II0 and then follows
the equivalent chain segment of the activated pEL.
Therefore, the Gly419-Glu420 pair would seem to
represent a hinge point in bPPE, about which segment Val416 to Gly420 would rotate after activation (tryptic proteolytic cleavage of peptide
bond Arg415-Val416) permitting the newly liberated Val416 N terminus to insert into the molecule
and form the buried salt-bridge with Asp594; this
salt-bride formation is presumed to induce the conformational changes leading to an open speci®city
pocket and a functional active site (Bode, 1979). On
comparing the N-terminal activation segment of
bPPE with the structures available for serine proproteinases, namely prethrombin2 (Vijayalakshmi
et al., 1994), bCTGA (Wang et al., 1985), and, now,
bCTGC, whose activation peptides are de®ned
(not the case in trypsinogen; Bode et al., 1976;
Fehlhammer et al., 1977), a major difference arises
(see Figure 5). Before the position 23 (bCTGA numbering; 423 in bPPE), common to all structures, the
869
chain takes a different course in bPPE, running
along the lower left front surface, while in prethrombin2, CTGA and CTGC, the N terminus is
located below and behind the strand immediately
preceding the autolysis loop and segment 25 to 32
(CTGA numbering), suggesting a different approach of an activating trypsin molecule.
The autolysis loop of bPPE is de®ned by electron
density. As in the other related zymogens, Gly542
acts as a hinge, with the following chain segment
taking a completely different course (up to Gly549)
compared with activated enzymes such as pEL.
This loop runs through the site, which after activation will become occupied by the ``lower lip'',
with Leu544 positioned at the entrance of the preformed pocket and somewhat plugging it through
contacts with Thr613 and Val616. The structurally
neighbouring calcium-binding loop is likewise well
de®ned in bPPE and exhibits, in spite of almost
completely different side-chains between residues
Glu470 and Glu480, a virtually identical conformation with that in pEL. The central, fully occupied calcium ion is similarly co-ordinated in a
pentagonal-pyramidal manner by Glu470 Oe1 and
Oe2, Glu480 Oe1, the carbonyl oxygen atoms of
Asp472 and Val475, and the carboxamide oxygen
atom of Gln477, with the seventh position occupied by a bulk solvent molecule (see Figure 6).
Recently, the structure of monomeric bovine
subunit III (Pignol et al., 1994) was published, displaying striking differences compared with bPPE.
Probably because of the absence of the ®rst two valine residues (Val416 and Val417, according to our
nomenclature) removed by autolysis (Pascual et al.,
1990), this truncated bPPE was found to have a
very low enzymatic activity (Kerfelec et al., 1986),
conceivably from its inability to form the internal
Figure 5. Superimposition of the N-terminal segments (before position 23, common to all of them) of the available
serine proproteinases structures, prethrombin2 (PRTR), bCTGA, bCTGC and bPPE in the traditional standard orientation of chymotrypsin-like serine endopeptidases (Huber & Bode, 1978). The N terminies of each structure, the P1arginine residue, after which tryptic activation occurs, are labelled (Cys1 of bCTGA almost coincides with Cys701 of
bCTGC). In all structures besides bPPE, the scissile peptide bond runs from left to right.
870
Structure of a Complex of Zymogens
Figure 6. Detail of the bPPE calcium-binding site and the corresponding loop Glu470 to Glu480 superimposed with
the ®nal 2Fobs ÿ Fcalc electron density contoured at 1.0 s. Each residue of the loop and the central ion are labelled.
salt-bridge (equivalent to Val416-Asp594) and to
stabilize an intact substrate-binding site and a correct active-site environment. The structure analysis
of subunit III (Pignol et al., 1994) revealed that the
N-terminal segment is just de®ned and distinctly
arranged from Pro424 or Ser426 onwards (depending on the crystal forms), and that the lower lipforming segment Ser589-Asp594 is folded inwards,
similar to what is observed in bCTGA, bCTGC and
bPPE. The second triad (Madison et al., 1993)
characteristic for the zymogen structures, clearly
present in bPPE as well as in the chymotrypsinogens and in trypsinogen (Bode et al., 1976;
Fehlhammer et al., 1977; Freer et al., 1970; Wang
et al., 1985), was not formed in subunit III, however; the calcium-binding loop had moved out of
its ``normally'' (i.e. in bPPE and bCTGC) observed
position occupying a surface region of the molecule
corresponding to Val423-Pro424 in pEL, the adjacent ``autolysis loop'' had turned towards the region ``normally'' accommodating this calciumbinding loop, and segment Gly515 to Leu520 was
found with a different conformation. In comparison to pEL and bPPE, these three surface loops
display shifts in their a-carbon atoms of up to
Ê.
almost 20 A
We suggest that the bPPE structure and that of
bovine subunit III (recently con®rmed by the crystallographic analysis of its homologous form from
porcine; see Pignol et al., 1995) represent two geometrically quite distinct states of similar free energy, in¯uenced by very few contacts made via the
N-terminal segment. The overall structural similarity of bPPE with bCTGA, bCTGC and trypsinogen suggests that the bPPE model is a
representative of the proproteinase E/proelastase
family and that the activation of these proenzymes
does occur following a mechanism similar to that
shown for the trypsinogen/chymotrypsinogen-like
proenzymes (Huber & Bode, 1978).
Modulation of the proteolytic activation by the
arrangement of the subunits
The activation process of oligomeric forms of
PCPAs in vitro is much slower than that observed
for other pancreatic zymogens and for monomeric
forms of PCPAs (Freisheim et al., 1967; Pascual
et al., 1990). To account for this phenomenon,
Neurath and co-workers postulated activation
mechanisms for the binary and ternary complexes
by trypsin whereby CTGC is activated ®rst yielding active chymotrypsin (Brown et al., 1963; Uren
& Neurath, 1972). The dissociation of the complex
was observed only at highly basic pH or after extensive treatment accounting for the subsequent activation of PCPA. More recently, the possibility of
activation of bCTGC without dissociation of the
bovine TC was discussed (Michon et al., 1991a,b).
Interestingly, an enhanced catalytic ef®ciency of
activated chymotrypsin C within the complex with
bPCPA is observed as compared with the monomeric form (Puigserver & Desnuelle, 1977), maybe
due to a more rigid conformation of the active site
of this enzyme when bound to bPCPA. It has been
further postulated that the slow rise in CPA activity during activation of its zymogen could be
due to the inhibitory action of its pro-segment (San
Segundo et al., 1982; Chapus et al., 1987; Vendrell
et al., 1990). According to this, after tryptic cleavage of the pro-segment the stability of the oligomeric complex would be weakened, given that the
activation segment of PCPA acts as a link between
the metalloenzyme moiety and PPE, as shown in
the porcine system (Vilanova et al., 1985). Based on
the previous considerations, a two-step activation
process by trypsin has been proposed where the
endopeptidase activity is expressed ®rst, whereas
CPA activity is detected only after complex dissociation (Kerfelec et al., 1985; Michon et al.,
1991a,b).
Most of the above-mentioned results and hypotheses become clari®ed when the three-dimensional structure of the bovine TC is analysed. As
shown in Figure 1, mainly the activation region of
bPCPA is in contact with the other two subunits
which, therefore, could perturb the activation process by external proteases. In contrast, the activation cleavage sites of bPPE and bCTGC subunits
are directed away from the bPCPA contacting surface and exposed to the bulk solvent, readily available for proteolytic attack to the same extent as in
871
Structure of a Complex of Zymogens
monomeric serine protease zymogens. Thus, the
bPPE subunit establishes an extensive interaction
with the bPCPA subunit through the activation
segment of the former, which acts as a link: 157
Ê are established beatom-atom contacts below 4 A
tween both proteins. bPPE essentially covers the
second surface helix of the AS with its active-site
groove but also partially touches other parts of it,
such as the N-terminal segment, the fourth (edge)
strand and the N-terminal pole of the helix connecting the activation globular domain and the enzyme moiety. Only bPPE residues of some exposed
loop segments surrounding its active site (Tyr435
to Thr441, His457 to Arg462A, Trp494 to Asp502,
Arg543 to Thr547, Trp572 to Thr575 and Ser614 to
Gly619) interacting with segments Glu5A to
Val8A, Leu29A to Leu34B, Phe58A to Ala70A and
Tyr75A to Asp87A of the bPCPA AS are involved
(see Table 1A). These are regions that (except the
autolysis loop) will presumably not undergo signi®cant conformational changes upon activation of
bPPE, as can be seen when superimposing the pEL
structure with the latter within the complex. Thus,
the same intermolecular contacts might occur in
complexes formed with activated proteinase E, apparently in agreement with experimental results
showing a stable oligomer also with activated proteinase E (AvileÂs et al., 1989). Signi®cantly, in the
TC the bPPE loop Asn497 to Val499 approaches
the edge strand of the AS of the central bPCPA
subunit. Here, the highest number of contacts
made by two single residues (11) is observed between Asn497 and Tyr75A, the P01-residue of the
secondary proteolytic cleavage point of bPCPA
(see Table 1A). Besides other noticeable hydrophobic and polar interactions, bPPE is hydrogen
bonded to the P01-carbonyl group via the side-chain
carboxamide group of its Asn497 (see Figure 7(a)).
Additionally, the P1-residue Arg76A is further
blocked by one interaction with Asn497. Steric hindrance by the adjacent bPPE and stabilization
through these many intermolecular contacts will
certainly help to prevent unfolding of this bPCPA
segment around Arg74A-Tyr75A; thus, this second
tryptic cleavage site seems to be essentially protected against tryptic attack in the TC, in agreement with experimental results (Chapus et al.,
1987; Puigserver & Desnuelle, 1977).
bCTGC likewise covers with its active-site
groove the bPCPA subunit, but touches not only
the AS, but also part of the enzyme domain of
bPCPA via some exposed loops by formation of
130 amino acid contacts. Here, we can observe four
clear interactions: between Glu91 and Arg99A, the
end of the C-terminal helix of the AS and the connecting segment, and Tyr735 to Thr741/Ser760 to
Leu762A of bCTGC; between Asn5 and Phe7 of
the N-terminal part of the mature CPA enzyme
moiety and loop Trp794 to Ser796 of bCTGC; between Trp81 and Glu88 of CPA and Phe797Leu798; ®nally, between segment Ser917 to Lys924
of bCTGC and three more short regions of CPA
(see Table 1B). Particularly, the bCTGC loop seg-
ment around Asn761 gets close to the scissile peptide bond Arg99A-Ala1 of bPCPA and is clamped
(via two hydrogen bonds) by means of the Asn761
carboxamide group to Arg99A N and O (see
Figure 7(b)). Additionally, the adjacent bCTGC
loop around Arg736A to Thr741 is folded over the
side-chains of the preceding (non-primed) bPCPA
residues Ser96A-Ser98A, contacting it through a
number of non-polar as well as polar interactions
(see Table 1B). Thus, this principal bPCPA cleavage
site seems to be sterically not accessible to tryptic
cleavage in the TC, in agreement with earlier proposals (Michon et al., 1991a,b).
In conclusion, in the TC the AS of bPCPA essentially acts as a clamp for the two other serine proproteinases, which stabilize and sterically protect
the ®rst and the second activation sites of bPCPA
against tryptic attack. Simultaneously, the activation sites of both serine proteinase zymogens are
exposed, as in the monomeric state, and are presumably free to attack by trypsin molecules. Thus,
these latter zymogens should be activated ®rst
(maybe in sequential order, bCTGC before bPPE;
see Kobayashi et al., 1981a,b), which might slightly
weaken the TC and facilitate quaternary rearrangements of the subunits or dissociation, only after
which the bPCPA component might become activatable (see Figure 8).
Role of the ternary complex
Depending on the species, PCPA secreted from
the pancreas (the widely occurring A1 form) appears either as a monomer or as non-covalently associated forms with other proenzymes (AvileÂs et al.,
1993). In contrast, PCPA2 and PCPB have been described only as monomers (Pascual et al., 1990;
AvileÂs et al., 1993; Oppezzo et al., 1994). Oligomeric
PCPA has been found in a binary complex with
the serine proteinase zymogens chymotrypsinogens B and C or with proproteinase E in human,
pig, whale and rat (for a review, see AvileÂs et al.,
1993) or in a ternary complex with the latter two
proenzymes in bovine, sheep, camel and goat
(Brown et al., 1961, 1963; Kerfelec et al., 1985). The
occurrence of PCPA in oligomeric complexes is
therefore quite widespread.
Arti®cial binary and ternary heterocomplexes between PCPA, CTGC and PPE (or subunit III) from
different species indicate that the speci®city of recognition and binding between these zymogens is
maintained within the mammals (Kerfelec et al.,
1985). It may also indicate that the spatial location
of the subunits in the corresponding oligomeric
complexes is similar. The three-dimensional structure of the bovine ternary complex might be a
model for the other oligomeric (binary or ternary)
complexes and a similar functional behaviour and
role could be expected for them. Related evidence
has been recently reported from the mouse mast
cell system, indicating that the serine endopeptidase chymase (mast cell protease 5) is required to
target translated mast cell carboxypeptidase A
E 5A
D 6A
V 8A
L 29A
L 32A
H 34A
L 34B
F 58A
P 59A
Q 62A
A 63A
K 65A
V 66A
F 67A
E 69A
A 70A
Y 75A
R 76A
I 77A
E 80A
D 81A
Q 83A
S 84A
D 87A
A.
PCPA
PPE
K
436A
2 (0;0)
1 (1;0)
6 (2;0) 7 (0;0)
8 (0;0)
Y
435
H
457
9 (2;0)
1 (0;0) 4 (0;0)
T
441
3 (1;0)
S
460
1 (1;0)
T
461
(1;1)
(0;0)
(1;0)
(0;1)
5 (0;0)
R
462A
1 (1;0) 5
3
2 (1;0) 4
1
S
462
1 (0;0)
W
494
C
498
2 (0;0)
2 (0;0)
4 (;0)
V
499
A
499A
2 (0;0) 7 (1;0) 8 (2;0) 4 (1;0)
N
497
11 (2;0)
1 (0;0)
2 (0;0) 1 (1;0)
3 (1;0)
S
496
1 (0;0)
D
502
5 (1;0)
T
547
5 (0;0) 1 (0;0)
R
543
T
575
2 (0;0)
1 (0;0) 1 (0;0)
W
572
Ê between (A) bPCPA and bPPE and (B) bPCPA and bCTGC in the ternary complex
Table 1. Matrix of intermolecular contacts < 4 A
F
615
5 (0;0)
1 (0;0) 2 (0;0)
S
614
1 (0;0)
S
617
F
618
G
619
3 (0;0)
3 (1;0) 5 (1;0) 1 (0;0)
1 (0;0)
A
617A
2 (0;0)
2 (0;0) 2 (0;0) 2 (0;0)
V
616
Sum
157 (22;2)
37
(9;2)
74
(7;0)
19
(3;0)
27
(3;0)
S
u
m
5 (3;0)
3 (0;0)
6 (0;0)
Y
735
2 (0;0)
CTGC
L
736
2 (0;1)
14 (0;0)
R
736A
5 (1;1)
R
739
1 (0;0)
T
741
1
3
8
2
(1;0)
(1;0)
(2;0)
(0;0)
S
760
6 (3;0)
8 (1;0)
N
761
4 (2;0)
1 (0;0)
T
762
2 (0;0)
3 (0;0)
L
762A
1 (0;0)
2 (0;0)
W
794
7 (1;0)
N
795
5 (1;0)
5 (1;0)
2 (1;0)
S
796
4
5
2
3
(0;0)
(0;0)
(0;0)
(0;0)
F
797
2 (0;0)
L
798
3 (0;0)
S
917
5 (0;0)
G
917A
2 (0;0)
2 (0;0)
L
918
2 (1;0)
2 (0;0)
K
924
Sum
130 (19;2)
16
(1;0)
16
(0;0)
22
(4;0)
76
(14;2)
S
u
m
The number of contacts between two residues is displayed at the intersection of both numbers, hydrogen bonds and salt-bridges are further indicated in parentheses (for numbering of bPCPA, bCTGC, and
bPPE, see Materials and Methods).
E 91A
F 94A
A 95A
S 96A
Q 97A
S 98A
R 99A
N5
T6
F7
W 81
K 84
K 85
E 88
L 233
Y 234
R 272
S 284
Q 285
P 288
B.
PCPA
Table 1. continued
874
Structure of a Complex of Zymogens
Figure 7. Detail of the TC structure around the interface between (a) bPPE (yellow) and the AS of bPCPA (red)
around the secondary bPCPA cleavage site (Arg74A-Tyr75A) superimposed with the ®nal 2Fobs ÿ Fcalc electron density contoured at 1 s and (b) bCTGC (orange) and bPCPA (red) around the primary bPCPA cleavage site (Arg99AAla1).
to the secretory granules or to protect the latter
neutral protease from autolysis (Stevens et al.,
1996).
Some hypotheses to explain the occurrence of
the bovine heterotrimer and other oligomeric
complexes have been advanced, based on its
putative role in the timing and potentiation of
the activation, in the proper co-ordination of the
appearance of the proteolytic activities in the
duodenum, in the modulation of the activity of
the subunits when still bound, and in the protection of PCPA against inactivation by the acidic
gastric juice at the beginning of the duodenum
(Uren & Neurath, 1972; Puigserver & Desnuelle
1977; Kerfelec et al., 1985; Chapus et al., 1987;
Michon et al., 1991a,b; AvileÂs et al., 1993), among
other proposals. However, the physiological sig-
ni®cance of the oligomeric complexes is still not
fully understood.
According to the present structural results, it can
be hypothesised that one of the roles of the complexes, at least for the ternary complex, is the burial of the proteolytic activation cleavage sites of
PCPA by the accompanying serine protease zymogens. This fact probably accounts for the differential appearance of the corresponding active forms,
the endopeptidases being activated ®rst (maybe in
sequential order, CTGC before PPE; see Kobayashi
et al., 1981b), PCPA being the last (see Figure 8).
This would ®t with the complementarity and required sequential action of the involved proteases
on alimentary proteins, which would be ®rstly converted in large pieces by the endopeptidases (chymotrypsin C and proteinase E) and then trimmed
875
Structure of a Complex of Zymogens
Figure 8. A scheme illustrating the proposed sequence of activation of the TC upon tryptic attack. In a ®rst step,
bCTGC and bPPE are activated, the TC structure probably being maintained even if a cleavage occurs at the primary
bPCPA scissile peptide bond; in this case, the strong binding between the severed (but still intact) AS and CPA
would still hold the oligomeric complex and keep CPA inactive. On further typtic treatment, both the primary and
secondary bPCPA scissile peptide bonds are cleaved, promoting ®nally the disassociation of the complex, fragmentation of the AS, and appearance of CPA activity.
by the carboxypeptidase. The maintenance of the
ternary complex until the occurrence of a deep proteolytic processing of the PCPA subunit could facilitate the protection of this subunit (and of the
others, in active form) against the destructive action of other proteases during its long passage
along the duodenum. According to this, co-ordination of activation and action, and protection
against proteolysis along the intestine, could be
two of the roles of the oligomeric complexes.
Materials and Methods
Materials
Phenylmethylsulphonylchloride, soybean trypsin inhibitor and synthetic peptide substrates were obtained
from Sigma. All other chemicals were purchased from
Merck. For column chromatography, an FPLC system
(Pharmacia) with a ®tted TSK-DEAE 5 PW column
(10 mm particle size, 100 nm pore size, 0.75 cm 7.5 cm),
supplied by Toso-Haas, was used. Cryschem crystallization dishes were purchased from Charles Supper (Natick, MA).
Protein purification
Acetone powders from fresh bovine pancreas were
used as the source for the TC. The acetone powders
were extracted for 30 minutes in 20 mM bis-Tris/HCl
(pH 6.5) containing 0.3 mg/ml of soybean trypsin inhibitor and 2 mM phenylmethylsulphonylchloride, at
4 C. The crude extracts were centrifuged at 8000 rpm,
the supernatants precipitated with 21% ammonium sulphate and left for 30 minutes at 4 C. The precipitate
was recovered by centrifugation at 8000 rpm, redissolved in buffer A (20 mM bis-Tris/acetate, pH 6.5),
and equilibrated with this buffer by gel-®ltration in
small Sephadex G-25 columns. A TSK-DEAE 5PW column previously equilibrated with buffer A was used
for the chromatography step. The proteins were loaded
and then eluted by a 100 minute linear gradient of this
buffer and buffer B (20 mM bis-Tris/acetate (pH 6.5),
0.8 M ammonium acetate) at 4 ml/minute at room temperature. Under these conditions, TC eluted as a main
peak at about 83 minutes. The fraction containing TC
was desalted with a small G-25 column and immediately rechromatographed in the same conditions as
above. The nature and purity of the eluted TC was
con®rmed by SDS/polyacrylamide gel electrophoresis
and activity measurements with synthetic peptide substrates.
Crystallization
After a series of preliminary trials applying the incomplete factorial approach using the sitting-drop vapour diffusion method, crystals were obtained after
four to six days from drops containing equal volumes
of aqueous protein solution (15 mg/ml) and a
solution of 0.2 M MgCl2, 30% (v/v) PEG 400, buffered
with 0.1 M Hepes to pH 7.1 at 4 C. These crystals are
Ê resolution using
well ordered, diffract beyond 2.8 A
CuKa radiation, and belong to the rhombohedral
space group R3 with cell constants (hexagonal setting)
Ê (for a ˆ b), and
varying between 187.6 and 188.5 A
Ê (for c) as observed for different
80.5 and 82.1 A
crystals, not isomorphous to each other. However,
these crystals displayed an almost perfect R32 symmetry, incompatible with the packing in the cell, due
to hemihedral twinning (Gomis-RuÈth et al., 1995a). Untwinned, well-ordered crystals were obtained from
drops with equal volumes of protein solution and
0.1 M CaCl2, 0.05 M Hepes, 15% PEG 400. Only one or
two crystals per drop of maximal dimensions
0.5 mm 0.5 mm 0.2 mm were obtained. These crysÊ resolution (using rotatingtals diffract beyond 2.6 A
anode CuKa radiation) and share the spacegroup with
the twinned ones, with just slightly modi®ed cell conÊ , c ˆ 82.5 A
Ê ), but the collected
stants (a ˆ b ˆ 188.5 A
diffraction data display no R32 symmetry. Crystals of
this type are isomorphous to each other.
876
Structure of a Complex of Zymogens
and reduced with rotavata/agrovata/truncate (CCP4,
1994). Table 2 provides a summary of the data collection
and processing.
Data collection and processing
Ê resolution were colX-ray diffraction data up to 2.6 A
lected on a 180-mm-MAR Research image plate detector
attached to a Rigaku-Denki rotating anode CuKa-generator operated at 5.4 kW at ÿ12 C. Furthermore, data (up
Ê ; I/s(I) ˆ 2.7 for data between 2.40 and 2.35 A
Ê)
to 2.35 A
were recorded on a 300-mm-MAR Research image plate
detector at the BW6 beam-line of the Deutsches Elektronensynchrotron (DESY) in Hamburg (project P-95-17)
Ê . The data were processed
using radiation of l ˆ 1.0 A
with MOSFLM v.5.23 (Leslie, 1991) and scaled, merged,
Structure solution
The orientation and position of the three molecules
was found with the Patterson-search method (Huber,
1965). The starting models used were those of porcine
procarboxypeptidase A (Guasch et al., 1992) for the bovine homologous form, pEL (Meyer et al., 1988) for
bPPE, and bCTGA (Wang et al., 1985) for bCTGC. The
Table 2. Data collection, processing and structure determination
Diffraction data (rotating anode ‡ synchrotron radiation)
Ê ) ‡ 116,068(to 2.35 A
Ê)
116,274(to 2.6 A
39,633
0.078
No. of measured re¯ections
No. of unique re¯ections
Ramerge
Completeness (%)
Ê
Data 20.0 to 2.35 A
Ê
Data 2.46 to 2.35 A
87.2
39.7
Patterson search and rigid-body re®nement
Cross-rotationb
a
(deg.)
b
(deg.)
g
(deg.)
Correlation
functiond
Rotation function for bPCPA
55.3
53.7
104.0
130.1
77.8
283.5
23.9
7.6
Rotation function for bPPE
99.8
103.4
53.5
79.5
158.9
341.5
14.4
11.5
Rotation function for bCTGC
91.3
0.2
52.1
66.8
1.2
346.3
10.8
8.4
Translationb,c
g
x
y
z
Correlation
functiond
Crystallographic
Rfactor (%)
Independent translation function for bPCPA
55.3
104.0
55.3
104.0
77.8
77.8
0.1564
0.2605
0.1614
0.4824
0.0000
0.0000
36.3
21.2
45.4
50.3
Independent translation function for bPPE
99.8
53.5
99.8
53.5
158.9
158.9
0.0754
0.0272
0.4538
0.0713
0.0000
0.0000
17.4
15.0
50.7
51.8
Independent translation function for bCTGC
91.3
52.1
91.3
52.1
1.2
1.2
0.5846
0.9249
0.0452
0.2763
0.0000
0.0000
13.5
13.4
51.9
52.1
a
b
Dependent 3-body translation functione
bPCPA
55.3
bPPE
99.8
bCTGC
91.3
104.0
53.5
52.1
77.8
158.9
1.2
0.1564
0.0784
0.5908
0.1614
0.4546
0.0431
0.0000
0.3529
0.9622
36.3
46.4
47.4
45.4
41.9
42.1
Rigid-body re®nement
bPCPA
bPPE
bCTGC
104.6
55.5
53.1
77.3
158.4
1.8
0.1573
0.0781
0.5902
0.1605
0.4547
0.0465
0.0010
0.3495
0.0679
54.6
39.0
56.4
101.4
91.0
a
Rmerge ˆ h ijI(h)i ÿ hI(h)ij/h I I(h)i; I(h)i is the observed intensity of the ith measurement of re¯ection h, and hI(h)i the mean
intensity of re¯ection h; calculated after loading and scaling, with rotavata/agrovata (CCP4, 1994).
b
a, b, g are given in Eulerian angles.
c
x, y, z are given in fractional cell coordinates.
d
De®ned as in the AMoRe suite (Navaza, 1994).
e
Correlation and Rfactor are cumulative after positioning of the ®rst, second, and third molecules, respectively. In the rotation and
translation functions, the second highest peaks are additionally displayed for comparison.
877
Structure of a Complex of Zymogens
Table 3. Final model of the procarboxypeptidase A ternary complex
Ê)
Resolution range used for re®nement (A
Re¯ections used
Protein atoms (non-hydrogen)
active
inactivea
Total solvent molecules
Ê 2)
Average temperature factor (A
Root-mean-square deviation from target
values
Ê)
Bonds (A
Angles (deg.)
Crystallographic Rfactor (free Rvalue)
a
6.0±2.35
30,541
7028
6933
95
381
45.4
0.012
1.822
0.192 (0.283)
Not included in re®nement or in map calculations.
rotation and translation functions were computed with
the AMoRe suite (Navaza, 1994) using data between 15
Ê resolution (see Table 2). No unique solution was
and 4 A
found for bCTGC employing pEL as searching model despite a very high level of sequence/structure homology.
Rigid body re®nement of the properly oriented and positioned molecules was performed with AMoRe (``®ting'',
Castellano et al., 1992)) dropping the crystallographic Rfactor (de®ned as jFobs ÿ Fcalcj/jFobsj) from 0.421 to
0.390 (see Table 2). The model was visually inspected
and corrected against (2Fobs ÿ Fcalc) and (Fobs ÿ Fcalc)
Fourier maps in successive cycles comprising manual
model building (using TURBO-Frodo version 5.0a, BioGraphics, Marseille (France)) on a Silicon Graphics workstation and crystallographic re®nement using X-PLOR
(BruÈnger et al., 1987). The crystallographic R-factor for
the ®nal model as re®ned against 30,541 re¯ections beÊ resolution is 0.192 (Rfree 0.283); the
tween 6.0 and 2.35 A
r.m.s. deviations from target values for bond lengths and
Ê and 1.822 , respectively (see Table 3).
angles are 0.012 A
For rationality reasons, in the bPPE residue numbering
400 counts have been added to the equivalent bCTGA
numbering, that is, Ser195, Asp102 and His57 of CTGA
are equivalent to Ser595, Asp502 and His457 in bPPE.
The bCTGC numbering starts with residue Cys701 (700
counts added to the corresponding bCTGA residues; the
bCTGC catalytic triad therefore consists of Ser895,
Asp802 and His757). The ®nal model comprises the residues from bPCPA (403 residues (sequence according to
Le HueÈrou et al., 1991), labelled 4A to 99A and 1 to 309,
numbering according to PDB access code 1pca; see
Guasch et al., 1992), bPPE (248 residues out of 253,
labelled 410 to 645), bCTGC (251 residues, labelled 701
to 945), one zinc and one calcium ion (labelled 350 and
650, respectively), and 381 solvent molecules) labelled
W100 to W480). Segments Ser712 to Val716, Tyr846 to
Thr847, and Val888 to Ser889 of bCTGC are not de®ned
by appropriate electron density; these nine residues and
eight surface-located side-chains have nevertheless been
tentatively traced, in order to preserve chain continuity,
and their occupancies set to zero. The ®rst ®ve residues
of bPPE (Phe405 to Phe409) are not de®ned by electron
density and have not been traced. An examination of
main-chain dihedral angles performed with PROCHECK
v. 3.0 (Laskowski et al., 1993), as implemented in CCP4
Program Suite release 2.10 (CCP4, 1994), showed four residues in disallowed regions. From these residues,
Ala852 is part of a ¯exible region, following the under®ned segment Tyr846 to Thr847 mentioned above.
Ser199 is embedded in a loop connecting two antiparallel
strands of the central CPA b-sheet and displays similar
``disallowed'' conformational angles ( ˆ 141; ˆ ÿ 12)
as observed in the monomeric porcine homologous
structure (133/ ÿ 4), being clearly de®ned by electron
density. Asp736B and Ala779 are also part of clearly de®ned surface-located loops.
Structure comparison and representations
Structure superimpositions were performed with
LSQKAB (Kabsch, 1978), as implemented in the CCP4package, and TURBO-Frodo, and displayed with the
latter program on a Silicon Graphics workstation.
Figures have been made with MOLSCRIPT (Kraulis,
1991) and TURBO-Frodo. The co-ordinates have been deposited with the PDB (access code 1pyt).
Acknowledgements
The kind help at early stages of protein puri®cation
and the structure analysis provided by V. Villegas, J.
Navaza, H. Brandstetter and I. Fita, as well as the ®nancial support provided by grant ERBCHRXCT9400535
(Human Capital and Mobility Programme of the European Union), by grand BIO95-0848 of CICYT (ComisioÂn
Interministerial de Ciencia y Tecnologia, Spain), and by
FPI-fellowship EX94 46121143 from the Ministerio de
EducacioÂn y Ciencia (Spain), as well as by Fundacio J.
Roviralta and by Centre de RefereÁncia en Biotecnologia
(Generalitat de Catalunya), are acknowledged.
References
AvileÂs, F. X., Pascual, R., SalvaÂ, M., Bonicel, J. &
Puigserver, A. (1989). Generation of a subunit
III-like protein by autolysis of human and porcine
proproteinase E in a binary complex with procarboxypeptidase A. Biochem. Biophys. Res. Commun.
163, 1191± 1196.
AvileÂs, F. X., Vendrell, J., Guasch, A., Coll, M. & Huber,
R. (1993). Advances in metalloprocarboxypeptidases. Emerging details on the inhibition mechanism and on the activation process. Eur. J. Biochem.
211, 381± 389.
Bode, W. (1979). The transition of bovine trypsinogen to
a trypsin-like state upon strong ligand binding. II.
The binding of the pancreatic trypsin inhibitor and
of isoleucine-valine and of sequentially related peptides to trypsinogen and to p-guanidinobenzoatetrypsinogen. J. Mol. Biol. 127, 357 ±374.
Bode, W. & Huber, R. (1986). Crystal structures of pancreatic serine endopeptidases. In Molecular and Cellular Basis of Digestion (Desnuelle, P., SjoÈstroÈm, H. &
NoreÁn, O., eds), pp. 213 ± 234, Elsevier, Amsterdam.
Bode, W. & Huber, R. (1994). Proteinase-protein inhibitor interactions. Fibrinolysis, 8, 161 ± 171.
Bode, W. & Schwager, P. (1975). The re®ned crystal
Ê resolution. II. Crystalstructure of b-trypsin at 1.8 A
lographic re®nement, calcium binding site, benzamidine binding and active site at pH 7.0. J. Mol.
Biol. 98, 693± 717.
Bode, W., Fehlhammer, H. & Huber, R. (1976). Crystal
Ê resolution.
structure of bovine trypsinogen at 1.8 A
I. Data collection, application of Patterson-search
techniques and preliminary structural interpretation. J. Mol. Biol. 106, 325±335.
878
Bode, W., Chen, Z., Bartels, K., Kutzbach, C., SchmidtÊ
Kastner, G. & Bartunik,, H. (1983). Re®ned 2 A
X-ray crystal structure of porcine pancreatic kallikrein A, a speci®c trypsin-like serine proteinase.
J. Mol. Biol. 164, 237±282.
Bode, W., Turk, D. & Karshikov, A. (1989). The re®ned
Ê X-ray crystal structure of D-Phe-Pro-Arg
1.9 A
chloromethylketone-inhibited human a-thrombin:
structure analysis, overall structure, electrostatic
properties, detailed active-site geometry, and structure-function relationships. Protein Sci. 1, 426 ± 471.
Brown, J. R., Cox, D. J., Greenshields, R. N., Walsh, H.,
Yamasaki, M. & Neurath, H. (1961). The chemical
structure and enzymatic functions of bovine procarboxypeptidase A. Proc. Natl Acad. Sci. USA, 47,
1554± 1562.
Brown, J. R., Greenshields, R. N., Yamasaki, M. &
Neurath, H. (1963). The subunit structure of bovine
procarboxypeptidase 1 - S6. Chemical properties
and enzymatic activities of the products of molecular disaggregation. Biochemistry, 2, 867 ± 876.
BruÈnger, A. T., Kuriyan, K. & Karplus, M. (1987). Crystallographic R factor re®nement by molecular
dynamics. Science, 235, 458± 460.
Bryan, P., Wang, L., Hoskins, J., Ruvinov, S., Strausberg,
S., Alexander, P., Almog, O., Gilliland, G. &
Gallagher, T. (1995). Catalysis of a protein folding
Ê strucreaction: mechanistic implications of the 2.0 A
ture of the subtilisin-prodomain complex. Biochemistry, 34, 10310 ±10318.
Cambillau, C., Kerfelec, B., Sciaky, M. & Chapus, C.
(1988). Subunit III of ruminant procarboxypeptidase
A-S6 complexes and pancreatic proteases E. A new
family of pancreatic serine endopeptidases?. FEBS
Letters, 232, 91 ±95.
Castellano, E. E., Oliva, G. & Navaza, J. (1992). Fast
rigid-body re®nement for molecular-replacement
techniques. J. Appl. Crystallog. 25, 281± 284.
Catasus, L., Villegas, V., Pascual, R., AvileÂs, F. X.,
Wicker-Planquart, C. & Puigserver, A. (1992).
cDNA cloning and sequence analysis of human
pancreatic procarboxypeptidase A1. Biochem. J. 287,
299± 303.
Catasus, L., Vendrell, J., AvileÂs, F. X., Carreira, S.,
Puigserver, A. & Billeter, M. (1995). The sequence
and conformation of human pancreatic procarboxypeptidase A2. cDNA cloning, sequence analysis,
and three-dimensional model. J. Biol. Chem. 270,
6651± 6657.
CCP4, (1994). The CCP4 suite: programs for protein
crystallography. Acta Crystallog. sect. D, 50, 760±
763.
Chapus, C., Kerfelec, B., Foglizzo, E. & Bonicel, J. (1987).
Further studies on the activation of bovine pancreatic procarboxypeptidase A by trypsin. Eur. J. Biochem. 166, 379 ±385.
Christianson, D. W. & Lipscomb, W. N. (1989). Carboxypeptidase A. Accts. Chem. Res. 22, 62 ± 69.
Coll, M., Guasch, A., AvileÂs, F. X. & Huber, R. (1991).
Three-dimensional structure of porcine procarboxypeptidase B: a structural basis of its inactivity.
EMBO J. 10, 1 ± 9.
Faming, Z., Kobe, B., Stewart, C. B., Rutter, W. J. &
Goldsmith, E. J. (1991). Structural evolution of an
enzyme speci®city. The structure of rat carboxypepÊ resolution. J. Biol. Chem. 266,
tidase A2 at 1.9 A
24606± 24612.
Fehlhammer, H., Bode, W. & Huber, R. (1977). Crystal
Ê resolution.
structure of bovine trypsinogen at 1.8 A
Structure of a Complex of Zymogens
II. Crystallographic re®nement, re®ned crystal structure and comparison with bovine trypsin. J. Mol.
Biol. 111, 415± 438.
Folk, J. E. & Schirmer, E. W. (1965). Chymotrypsin C. I.
Isolation of the zymogen and the active enzyme:
preliminary structure and speci®city studies. J. Biol.
Chem. 240, 181± 192.
Freer, S. T., Kraut, J., Robertus, J. D., Wright, H. T. &
Ê crysXuong, N. H. (1970). Chymotrypsinogen 2.5 A
tal structure, comparison with a-chymotrypsin, and
implications for zymogen activation. Biochemistry, 9,
1997± 2009.
Freisheim, J. H., Walsh, K. A. & Neurath, H. (1967). The
activation of bovine procarboxypeptidase A. II.
Mechanism of activation of the succinylated enzyme
precursor. Biochemistry, 10, 3010 ±3019.
Gallagher, T., Gilliland, G., Wang, L. & Bryan, P. (1995).
The prosegment-subtilisin BPN0 complex: crystal
structure of a speci®c `foldase'. Structure, 3, 907 ±
914.
Gardell, S. J., Craik, C. S., Clauser, E., Goldsmith, E. J.,
Stewart, C. B., Graf, M. & Rutter, W. J. (1988).
A novel rat carboxypeptidase, CPA2: characterization, molecular cloning, and evolutionary implications
on
substrate
speci®city
in
the
carboxypeptidase gene family. J. Biol. Chem. 263,
17828± 17836.
Gomis-RuÈth, F. X., Fita, I., Kiefersauer, K., Huber, R.,
AvileÂs, F. X. & Navaza, J. (1995a). Determination of
hemihedral twinning and initial structural analysis
of crystals of the procarboxypeptidase A ternary
complex. Acta Crystallog. sect. D, 51, 819± 823.
Gomis-RuÈth, F. X., Gomez, M., Bode, W., Huber, R. &
AvileÂs, F. X. (1995b). The three-dimensional structure of the native ternary complex of bovine pancreatic procarboxypeptidase A with proproteinase E
and chymotrypsinogen C. EMBO J. 14, 4387± 4394.
Guasch, A., Coll, M., AvileÂs, F. X. & Huber, R. (1992).
Three-dimensional structure of porcine pancreatic
procarboxypeptidase A. A comparison of the A and
B zymogens and their determinants for inhibition
and activation. J. Mol. Biol. 224, 141± 157.
Hubbard, S. J., Eisenmenger, F. & Thornton, J. M.
(1994). Modeling studies of the change in conformation required for cleavage of limited proteolytic
sites. Protein Sci. 3, 757± 768.
Huber, R. (1965). Die automatisierte FaltmolekuÈlmethode. Acta Crystallog. sect. A, 19, 353 ±356.
Huber, R. & Bode, W. (1978). Structural basis for the
activation and action of trypsin. Accts. Chem. Res.
11, 114± 122.
Huber, R., Kukla, D., Bode, W., Schwager, P., Bartels,
K., Deisenhofer, J. & Steigemann, W. (1974). Structure of the complex formed by bovine trypsin and
bovine pancreatic trypsin inhibitor. II. CrystalloÊ resolution. J. Mol. Biol.
graphic re®nement at 1.9 A
89, 73 ± 101.
Kabsch, W. (1978). A solution for the best rotation to
relate two sets of vectors. Acta Crystallog. sect. A, 32,
922± 923.
Keil-Dlouha, V., Puigserver, A., Marie, A. & Keil, B.
(1972). On subunit II of bovine procarboxypeptidase
A. Enzymatic speci®city and tryptic activation. Biochim. Biophys. Acta, 276, 531± 545.
Kerfelec, B., Chapus, C. & Puigserver, A. (1985). Existence of ternary complexes of procarboxypeptidase
A in the pancreas of some ruminant species. Eur. J.
Biochem. 151, 515± 519.
Structure of a Complex of Zymogens
Kerfelec, B., Cambillau, C., Puigserver, A. & Chapus, C.
(1986). The inactive subunit of ruminant procarboxypeptidase A-S6 complexes. Structural basis
of inactivity and physiological role. Eur. J. Biochem.
157, 531± 538.
Kim, H. & Lipscomb, W. N. (1991). Comparison of the
structures of three carboxypeptidase A-phosphonate
complexes determined by X-ray crystallography.
Biochemistry, 30, 8171± 8180.
Kobayashi, R., Kobayashi, Y. & Hirs, C. H. W. (1978).
Identi®cation of a binary complex of procarboxypeptidase A and a precursor of protease E in
procine pancreatic secretion. J. Biol. Chem. 253,
5526± 5530.
Kobayashi, R., Kobayashi, Y. & Hirs, C. H. W. (1981a).
The speci®city of porcine pancreatic protease E.
J. Biol. Chem. 256, 2460± 2465.
Kobayashi, Y., Kobayashi, R. & Hirs, C. H. W. (1981b).
Identi®cation of zymogen E in a complex with
bovine procarboxypeptidase A. J. Biol. Chem. 256,
2466± 2470.
Kraulis, P. J. (1991). MOLSCRIPT: a program to produce
both detailed and schematic plots of protein
structures. J. Appl. Crystallog. 24, 946 ± 950.
Largman, C., Broderick, J. W. & Geokas, M. C. (1976).
Puri®cation and characterization of two human
pancreatic elastases. Biochemistry, 15, 2491± 2500.
Laskowski, R. A., MacArthur, M. W., Hutchinson, E. G. &
Thornton, J. M. (1993). PROCHECK: a program
to check the stereochemical quality of protein
structures. J. Appl. Crystallog. 26, 283± 291.
Le HueÈrou, I., Guilloteau, P., Toullec, R., Puigserver,
A. & Wicker, C. (1991). Cloning and nucleotide
sequence of a bovine pancreatic preprocarboxypeptidase A cDNA. Biochem. Biophys. Res. Commun. 175,
110± 116.
Leslie, A. G. W. (1991). Macromolecular data processing.
In Crystallographic Computing V (Moras, D.,
Podjarny, A. D. & Thierry, J. C., eds), pp. 27 ± 38,
Oxford University Press, Oxford.
Madison, E. L., Kobe, A., Gething, M.-J., Sambrook,
J. F. & Goldsmith, E. J. (1993). Converting tissue
plasminogen activator to a zymogen: a regulatory
triad of Asp-His-Ser. Science, 262, 419± 421.
Mallory, P. A. & Travis, J. (1975). Human pancreatic
enzymes: characterization of a nonelastolytic
enzyme, protease E, resembling elastase. Biochemistry, 14, 722± 730.
Matthews, B. W., Sigler, P. B., Henderson, R. & Blow,
D. M. (1967). Three-dimensional structure of tosyla-chymotrypsin. Nature, 214, 652 ±656.
Meyer, E., Cole, G., Radahakrishnan, R. & Epp, O.
(1988). Structure of native porcine pancreatic elasÊ resolution. Acta Crystallog. sect. B, 44,
tase at 1.65 A
26± 38.
Michon, T., Granon, S., Sauve, P. & Chapus, C. (1991a).
The activation peptide of pancreatic procarboxypeptidase A is the keystone of the bovine procarboxypeptidase A-S6 ternary complex. Biochem. Biophys.
Res. Commun. 181, 449 ± 455.
Michon, T., Sari, J. C., Granon, S., Kerfelec, B. &
Chapus, C. (1991b). Microcalorimetric investigation
of the interactions between the subunits of the
bovine pancreatic procarboxypeptidase A complex.
Eur. J. Biochem. 201, 217± 222.
Navaza, J. (1994). AMoRe: an automated package for
molecular replacement. Acta Crystallog. sect. A, 50,
157± 163.
879
Oppezzo, O., Ventura, S., Bergman, T., Vendrell, J.,
Jornvall, H. & AvileÂs, F. X. (1994). Procarboxypeptidase in rat pancreas. Overall characterization and
comparison of the activation processes. Eur. J. Biochem. 222, 55± 63.
Pascual, R., Vendrell, J., AvileÂs, F. X., Bonicel, J., Wicker,
C. & Puigserver, A. (1990). Autolysis of proproteinase E in bovine procarboxypeptidase A ternary
complex gives rise to subunit III. FEBS Letters, 277,
37 ±41.
Pignol, D., Gaboriaud, C., Michon, T., Kerfelec, B.,
Chapus, C. & Fontecilla-Camps, J. C. (1994). Crystal
structure of bovine procarboxypeptidase A-S6 subunit III, a highly structured truncated zymogen E.
EMBO J. 8, 1763± 1771.
Pignol, D., Granon, S., Chapus, C. & Fontecilla-Camps,
J. C. (1995). Crystallographic study of a cleaved
non-activatable form of porcine zymogen E. J. Mol.
Biol. 252, 20 ± 24.
Puigserver, A. & Desnuelle, P. (1977). Reconstitution of
bovine procarboxypeptidase A-S6 from the free
subunits. Biochemistry, 16, 2497± 2501.
Rees, D. C., Lewis, M. & Lipscomb, W. N. (1983).
Re®ned crystal structure of carboxypeptidase A at
Ê resolution. J. Mol. Biol. 168, 367 ± 387.
1.54 A
Rinderknecht, H. (1993). Pancreatic secretory enzymes.
In The Pancreas. Biology, Pathobiology, and Disease
(Go, V. L. W., Dimagno, E. P., Gardner, J. D.,
Lebenthal, E., Reber, H. A. & Scheele, G. A., eds),
2nd edit., pp. 219±251, Raven Press, New York.
San Segundo, B., MartõÂnez, M. C., Vilanova, M.,
Cuchillo, C. M. & AvileÂs, F. X. (1982). The severed
activation segment of porcine pancreatic procarboxypeptidase A is a powerful inhibitor of the
active enzyme. Isolation and characterization of
the activated peptide. Biochim. Biophys. Acta, 707,
74±80.
Shen, W., Fletcher, T. S. & Largman, C. (1987). Primary
structure of human pancreatic protease E determined by sequence analysis of the cloned mRNA.
Biochemistry, 26, 3447± 3452.
Shirasu, Y., Takemura, K., Yoshida, H., Sato, Y., Iijima,
H., Shimada, Y., Mikayama, T., Ozawa, T., Ozawa,
T., Ikeda, N., Ishida, A., Tamai, Y., Matsuki, S.,
Tanaka, J., Ikenaga, H. & Ogawa, M. (1988). Molecular cloning of complementary DNA encoding
one of the human pancreatic protease E isozymes.
J. Biochem. (Tokyo), 104, 259± 264.
Stevens, R. L., Qui, D., McNeil, H. P., Friend, D. S.,
Hunt, J. E., Austen, K. F. & Zhang, J. (1996).
Transgenic mice that possess a disrupted mast cell
protease 5 (mMCP-5) gene can not store carboxypeptidase A (mMC-CPA) protein in their granules.
FASEB J. 10, A1307.
Sziegoleit, A. (1982). Puri®cation and characterization of
a cholesterol-binding protein from human pancreas.
Biochem. J. 207, 573 ± 582.
Tani, T., Ohsumi, J., Mita, K. & Takiguchi, Y. (1988).
Identi®cation of a novel class of elastase isozyme,
human pancreatic elastase III, by cDNA and
genomic gene cloning. J. Biol. Chem. 263, 1231±
1239.
Uren, J. R. & Neurath, H. (1972). Mechanism of activation of bovine procarboxypeptidase A S5. Alterations in primary and quaternary structure.
Biochemistry, 11, 4483± 4492.
Vendrell, J., Cuchillo, C. M. & AvileÂs, F. X. (1990).
The tryptic activation pathway of monomeric
880
Structure of a Complex of Zymogens
procarboxypeptidase A. J. Biol. Chem. 265, 6949±
6953.
Vijayalakshmi, J., Padmanabhan, K. P., Mann, K. G. &
Tulinsky, A. (1994). The isomorphous structures of
prethrombin2, hirugen-, and PPACK-thrombin:
changes accompanying activation and exosite binding to thrombin. Protein Sci. 3, 2254± 2271.
Vilanova, M., Burgos, F. J., Cuchillo, C. M. & AvileÂs,
F. X. (1985). Urea-gradient gel electrophoresis stu-
dies on the association between procarboxypeptidase A and B, proproteinase E and their tryptic
activation products, from guinea pig pancreas.
FEBS Letters, 191, 273 ± 277.
Wang, D., Bode, W. & Huber, R. (1985). Bovine chymotrypsinogen A. X-ray crystal structure analysis and
Ê resolution.
re®nement of a new crystal form at 1.8 A
J. Mol. Biol. 185, 595 ± 624.
Edited by I. A. Wilson
(Received 11 November 1996; received in revised form 4 March 1997; accepted 4 March 1997)