Insight into autoproteolytic activation from the
structure of cephalosporin acylase: A protein
with two proteolytic chemistries
Jin Kwang Kim*†, In Seok Yang*†, Hye Jeong Shin‡, Ki Joon Cho*, Eui Kyung Ryu*§, Sun Hwa Kim*¶, Sung Soo Park*,
and Kyung Hyun Kim‡储**
*Department of Life Sciences and Biotechnology, School of Life Sciences and Biotechnology, and ‡Department of Bio-Microsystem Technology, Korea
University, Seoul 136-701, Korea; and 储Department of Biotechnology, College of Science and Technology, Korea University, Jochiwon 339-700, Korea
Edited by Earl W. Davie, University of Washington, Seattle, WA, and approved December 19, 2005 (received for review September 12, 2005)
Cephalosporin acylase (CA), a member of the N-terminal nucleo-
phile hydrolase family, is activated through sequential primary and
secondary autoproteolytic reactions with the release of a pro
segment. We have determined crystal structures of four CA mu-
tants. Two mutants are trapped after the primary cleavage, and the
other two undergo secondary cleavage slowly. These structures
provide a look at pro-segment conformation during activation in
N-terminal nucleophile hydrolases. The highly strained helical pro
segment of precursor is transformed into a relaxed loop in the
intermediates, suggesting that the relaxation of structural con-
straints drives the primary cleavage reaction. The secondary auto-
proteolytic step has been proposed to be intermolecular. However,
our analysis provides evidence that CA is processed in two sequen-
tial steps of intramolecular autoproteolysis involving two distinct
residues in the active site, the first a serine and the second a
glutamate.
autoproteolysis 兩 precursor activation 兩 intermediate structure 兩
pro segment
P osttranslational autoproteolysis is a mechanism used to activate
many proteins via self-catalyzed peptide bond rearrangements,
which play an essential role in a wide variety of biological processes.
They include activation cascades such as blood coagulation and
fibrinolysis, cell death, embryonic development, protein targeting
and degradation, viral protein processing, and zymogen activation
(1–6). Increasing evidence suggests that autoproteolysis may be
involved in more biological activation processes than previously
appreciated. These processes are all mediated by intramolecular or
intermolecular autocleavage reactions. Four types of intramolecu-
lar protein modifications have been studied extensively: hedgehog
proteins, inteins, N-terminal nucleophile (Ntn) hydrolases, and
pyruvoyl enzymes (4, 7–9).
Members of the Ntn hydrolase family have divergent sequences,
but share an ␣␤␤␣ sandwich structural motif and a common
enzyme mechanism (7, 8). In addition, they catalyze internal
peptide bond rearrangement through an N 3 O or N 3 S acyl shift
by a nucleophilic Ntn amino acid created by autoproteolytic pro-
cessing (4, 7). Structures of mature Ntn hydrolases, including
cephalosporin acylase (CA), penicillin G acylase (PA), penicillin V
acylase, glutamine 5-phosphoribosyl-1-pyrophosphate amidotrans-
ferase, 20S proteasome ␤ subunit (PRO), glycosylasparaginase
(GA), and L-aminopeptidase, have been determined (10–17).
Structures of the precursors of CA, PA, PRO, and GA have been
reported, providing some insight into the activation mechanism (11,
18–22). However, the understanding of mechanism cannot be
complete without structural information on the first cleavage
intermediates.
Ntn hydrolases are of particular interest because they include
acylase enzymes used in the biosynthesis of the ␤-lactam antibiotics
cephalosporin and penicillin. We have focused on CA. This enzyme
is expressed as a 76-kDa precursor that is activated by two sequen-
tial autocleavage steps (11, 23) (Fig. 1A). A primary cleavage
1732–1737 兩 PNAS 兩 February 7, 2006 兩 vol. 103 兩 no. 6
between Gly-169 and Ser-170 generates the ␣⬘ subunit (the ␣
subunit with a 9-aa pro segment) and the ␤ subunit. A subsequent
secondary autocleavage between Gly-160 and Asp-161 releases the
pro segment from the ␣⬘ subunit, resulting in an active heterotet-
ramer (␣␤)2. The release of the pro segment may be required for
optimal CA activity. Ser-170, which becomes the N-terminal resi-
due (Ser-1)†† of the ␤ subunit of CA is invariant and known to play
a critical role in both autoproteolytic activation and enzyme catal-
ysis (23–25).
In an effort to gain a better understanding of the autoproteolyic
mechanism, we have designed constructs containing mutations that
partially prevent or retard autocleavage and carried out the struc-
tural determination of these mutants. Here, we report structural
evidence that CA is processed in two sequential steps of intramo-
lecular autoproteolysis involving two distinct proteolytic mecha-
nisms, the first mediated by a serine residue and the second by a
glutamate.
Results
Choice of Mutant Proteins. A single amino acid mutation of
Ser-170 to alanine (S170A) prevents primary cleavage, pro-
ducing an enzymatically inactive single-polypeptide precursor
(Fig. 1B). The crystal structures of precursor (S170A) and
mature CA have been reported (11). In this study, we have
focused mainly on four mutants. S170C and E159Q underwent
primary autocleavage but no secondary autocleavage. Y202L
and R226K, slow processing mutants, showed a delayed release
of the pro segment in comparison with the mature enzyme.
Tyr-202 and Arg-226 are located in the substrate binding site
of CA. The Y202L and R226K mutants were initially analyzed
as altered-activity mutants and found to be capable of sequen-
tially producing ␣⬘ and ␣ subunits during activation. In the case
of Y202L, it took ⬇24 h at 37°C to make a complete transition
from the uncleaved to the cleaved form (Fig. 1C, Y202L
control). Thus, early (Y202L) and late (Y202L,L) stages of
Conflict of interest statement: No conflicts declared.
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: CA, cephalosporin acylase; Ntn, N-terminal nucleophile.
Data deposition: The atomic coordinates have been deposited in the Protein Data Bank,
www.rcsb.org (PDB ID codes 2ADV, 2AE5, 2AE4, and 2AE3 for Y202L, S170C, E159Q, and
R226K, respectively).
†J.K.K. and I.S.Y. contributed equally to this work.
§Present address: Department of Chemistry, Pohang University of Science and Technology,
Pohang 790-784, Korea.
¶Present address: Department of Microbiology and Immunology, College of Medicine,
Catholic University, Seoul 137-701, Korea.
**To whom correspondence should be addressed. E-mail: khkim@korea.ac.kr.
††The amino acid residue is numbered according to the precursor (residues 1– 691) com-
prising the ␣ subunit (residues 1–160), the pro-segment region (residues 161–169), and
the ␤ subunit (residues 170 – 691). The amino acid residue of the ␤mature subunit (precursor
number minus 169) is numbered in parentheses, if applicable.
© 2006 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0507862103

Fig. 1. Autoproteolytic cleavage of glutaryl 7-aminocephalosporanic acid
acylase. (A) Schematic diagram of two-step autoproteolytic cleavage. The
precursor is activated through a primary cleavage to generate the ␣ subunit
containing the pro segment (␣⬘ subunit) and the ␤ subunit. A subsequent
secondary autocleavage releases the pro segment. (B) SDS兾PAGE analyses of
the autoproteolytic cleavage of CA. Lane M, molecular weight markers; S170A
is a precursor form. Y202L,L was obtained by incubating Y202L at 37°C for 24 h.
(C) Time-course experiments of secondary autocleavage. Y202L or S170C was
incubated with mature enzyme (WT) or 1-ethyl-3-(3-dimethylaminopropyl-
)carbodiimide (EDC) or at different pH values. Incubation and reconstitution
experiments (Inc兾Rec). Lane 1, freshly prepared Y202L; lane 2, crystals of fresh
Y202L; lane 3, crystals of Y202L incubated at 37°C; lane 4, reconstitution with
the combination of ␣mature and ␤mature subunits; lane 5, reconstitution with
␣⬘S170C and ␤mature subunits; and lane 6, reconstitution with ␣⬘E159Q and ␤mature
subunits.
Y202L during the time course of autoproteolysis could be
captured for crystallization (Fig. 1B, lanes Y202L and
Y202L,L). The presence or absence of the pro segment linked
to the ␣ subunit of the mutant (␣⬘␤)2 or (␣␤)2 was confirmed
by mass spectrometric and crystallographic results (see Fig. 6,
which is published as supporting information on the PNAS web
site). The electron density of the pro segment in Y202L,L
revealed a clear discontinuity after Gly-160, indicating a
Kim et al.
Fig. 2. Structural comparison of CA proteins. (A) Superposition of the overall
structures of S170A, S170C, E159Q, Y202L, R226K, and the mature CA. Each
structure is shown in yellow, cyan, dark green, green, slate blue, and marine,
respectively. To emphasize the differences of the pro segments, the pro
segment of Y202L is shown in red. Residues Gly-160, Ser-170, Tyr-202, and
Arg-226 are shown in a ball-and-stick representation in magenta. (B) Stereo-
view of the pro-segment conformation of Y202L. The electron density map
corresponds to the 2 Fo ⫺ Fc map calculated at 0.5 ␴.
complete release of the pro segment, which was consistent with
the SDS兾PAGE result (Fig. 1B, lane Y202L,L).
Comparison of Mutant Structures and Pro Segments. The refined
overall structures of the mutants, Y202L, R226K, S170C, and
E159Q (see Table 1, which is published as supporting information
on the PNAS web site), are similar to the precursor and mature
structures reported earlier (11) (Fig. 2A). In brief, they adopt the
␣␤␤␣ motif formed by two ␤-sheets packed against each other,
sandwiched by two layers of ␣-helices. After superposition of the C␣
atoms, the rms deviations of mutants against the precursor S170A
range from 0.37 to 0.53 Å. The structural ␣␤␤␣ core is highly
conserved in the Ntn hydrolase superfamily (8), as in structures of
glutarylamidase from Pseudomonas sp. and CA from Pseudomonas
diminuta (10, 24). However, there are significant differences at the
entryway to the active site, which is a deep cleft at the tip of the
␣␤␤␣ sandwich motif, where the pro segment is located.
In precursor CA the pro segment is ␣-helical from Pro-163 to
Ala-166 (11). Proline is not favored in helices, and significant
deviations from helical geometry suggest that this is a high-energy
conformation, generating a peptide flip for initiation of primary
autoproteolysis. A similar strained conformation, which was sug-
gested to regulate the activation of glycosylasparaginase (21) and a
pyruvoyl enzyme (22), is believed to be the driving force for
BIOPHYSICS
autocleavage to break the scissile peptide bond (26). In the inter-
mediate mutants of CA, which have undergone the initial but not
the secondary cleavage, the pro segments adopt strikingly different
conformations from that of the precursor (Fig. 2 A). First, the pro
segments show a loop conformation without regular secondary
PNAS 兩 February 7, 2006 兩 vol. 103 兩 no. 6 兩 1733
structure. While their N termini are covalently attached to the ␣
subunit, the C termini upon primary cleavage are moved from
residue 170 and may be partly exposed to solvent. In addition, there
are no significant geometric deviations from ideal geometry in the
pro segments, indicating unstrained conformations. Second, they
are highly mobile with high B factors (68–83 vs. 18–35 Å2 in
precursor). The electron densities for the pro segments are visible
only at a low contour level, as shown in Fig. 2B (and see Fig. 7A,
which is published as supporting information on the PNAS web
site). The pro segment may become highly mobile upon primary
cleavage, a characteristic of a terminal end, which would probably
facilitate secondary autocatalysis and subsequent dissociation of
the pro segment. Thus the highly strained helical conformation
of the precursor pro segment is transformed into a highly flexible
and relaxed loop conformation in the intermediates, supporting
the notion that the relaxation of structural constraints drives the
primary autocleavage reaction.
A comparative analysis of the mutant structures reveals that
the pro segments exist in a variety of conformations (see Fig. 7B).
Although Y202L, R226K, and E159Q follow similar backbone
paths from Glu-159 to Pro-162, moving down to the deep cleft at
Glu-159, they extend away from each other at Pro-163. By contrast,
residues Glu-159 and Gly-160 in S170C have ␸ conformational
angles opposite to the signs of other mutants and the pro segment
moves downward at Gly-160 to the deep cleft. The differences in the
backbone path place Pro-163 of S170C and Y202L at positions ⬇10
Å apart from each other. Moreover, S170A precursor structure
reveals extensive contacts between the pro segment and protein
residues (see Tables 2 and 3, which are published as supporting
information on the PNAS web site), but there are significantly
fewer contacts in the intermediate mutants. Although only partial
models of the pro segments could be built, it is clearly shown that
the number of polar and van der Waals contacts is greatly dimin-
ished in the intermediate mutants (see below). It is likely that
conformational constraints in precursor may be confined within the
pro segment during protein folding with expenditure of the non-
covalent interaction energies, which are released upon primary
autocleavage.

Primary Autoproteolytic Site. The active sites of mutants formed

upon primary autocleavage display structural features similar to the
mature enzyme. First, an intact catalytic pseudotriad of Ser-His-
Glu becomes registered in all of the mutants, involving the free
␣-amino group of (Ser-1), not the usual hydroxyl group, as in the
wild-type structures (11, 24). The side chain of (His-23) in the
wild-type contributes to a reduction in the pKa of the ␣-amino
group of (Ser-1), which then acts as a general base to enhance the
nucleophilicity of its own hydroxyl group (data not shown). Second,
Ser-170(Ser-1), which is buried in the precursor (11, 18), becomes
partially exposed to solvent in the mutants; it is completely open to
solvent in mature enzyme (11). The differential burial of Ser-170
may correspond to a structural activation signal for enzyme catal-
ysis. Third, despite the partial enclosure of the active site by the pro
segment, a solvent molecule emerges near the hydroxyl group of
(Ser-1) at the active site in Y202L and R226K. The solvent molecule
at this position plays a crucial role in the enzyme activity of mature
CA (10, 11, 24), as recent studies revealed the role of the bound
water near the catalytic residue in Ntn hydrolases (18, 21, 26).
Secondary Autoproteolysis: Is It Intermolecular or Intramolecular? In
contrast to the primary autocleavage mechanisms for the Ntn
hydrolases (11, 19, 20, 22), the secondary cleavage mechanism of
CA is not clearly understood. An intermolecular cleavage mecha-
nism for CA activation was proposed based on reconstitution
experiments (23). CA takes glutaryl 7-aminocephalosporanic acid
(7-ACA) as a primary substrate and catalyzes its hydrolysis into
glutaric acid and 7-ACA. It was reported that the mutations in the
substrate binding site that inhibited deacylation activity also pre-
Fig. 3. Secondary autocleavage site of CA proteins. The interactions be-
tween pro-segment and protein residues in Y202L, precursor S170A, R226K,
S170C, and E159Q. Those involved in hydrophobic interactions in S170A are
superimposed in a transparent space-filling representation. Hydrogen bonds
are indicated by dotted lines, and water molecules are represented as red
spheres. The distances between the side chain of Glu-159 and the backbone
carbon atom of Gly-160 via WAT1 in Y202L and R226K are highlighted by thick
dotted lines.
vented secondary autoproteolysis (25). CA secondary autoprote-
olysis might be correlated with catalytic activity. Nevertheless, the
question remains: how is secondary cleavage initiated in the ab-
sence of the active mature enzyme? Importantly, the release rate of
the pro segment of Y202L was unchanged, and the autocleavage of
S170C was not restored when incubated with the mature enzyme
(Fig. 1C, Y202L⫹WT and S170C⫹WT), providing experimental
evidence against intermolecular secondary cleavage. More con-
vincingly, Y202L crystals were incubated at 37°C for 24 h, where
intermolecular autocleavage is not allowed because of crystal
packing, and SDS兾PAGE analysis of the incubated Y202L crystals
demonstrated the generation of the cleaved ␣ subunit (Fig. 1C,
Y202L Inc兾Rec, lane 3).
In the precursor structure, the carboxyl group of Glu-159 is
within hydrogen-bonding distance of Gln-75 (Fig. 3, S170A). The
side chain of Asp-161 is hydrogen-bonded to those of Ser-152 and
Arg-155. Moreover, Leu-165 is involved in hydrophobic contacts

1734 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0507862103 Kim et al.

with Leu-148, Tyr-199, and Phe-200. There are extensive contacts
between the pro segment and protein residues in the deep cleft (see
Table 2). In the intermediate mutants, however, there are fewer
contacts and no hydrophobic interaction is found between Leu-165
and protein residues. Notably, Gln-75 points away toward the
solvent and its hydrogen-bonding to Glu-159 is absent. In Y202L,
Glu-159 is linked to Thr-76 by a network of hydrogen bonds
involving a series of water molecules (Fig. 3, Y202L). A positive
density at the Fo ⫺ Fc difference Fourier map could be modeled as
solvent molecule WAT1, which is hydrogen-bonded to the carbonyl
oxygen of Asp-161 and another solvent molecule. Interestingly,
WAT1 is also close to the side chain of Glu-159 (4.0 Å) and is
positioned on top of the carbonyl carbon of Gly-160 (4.0 Å) at the
scissile peptide bond of secondary autocleavage. R226K has a very
similar backbone configuration to that of Y202L at the secondary
cleavage site (Fig. 3). Notably, a solvent molecule corresponding to
WAT1 of Y202L could be placed in hydrogen-bonding distance to
Glu-159 (3.5 Å) and close to the carbonyl carbon of Gly-160 (4.0 Å).
Both mutant structures thus reveal that WAT1 could be modeled
only 3.5–4.0 Å from Glu-159 adjacent to the conserved secondary
site, suggesting that it is well within the range to facilitate proton
abstraction.
In the S170C mutant structure, however, strong hydrogen bonds
could be made between a water molecule and the backbone oxygen
and nitrogen atoms of Glu-159 and Gly-160, respectively, instead of
the side chain of Glu-159 (Fig. 3). S170C displays different back-
bone conformation of the secondary cleavage site, in contrast to
other mutants (see Fig. 7B). In E159Q, where the local backbone
conformation is similar to that of Y202L or R226K, no significant
electron densities for solvent molecules could be observed at the
secondary site.
Carboxyl Proteolytic Autocleavage. In an effort to derive a detailed
molecular mechanism of secondary autocleavage, we further car-
ried out a number of biochemical and mutational experiments. At
first, it appeared that the sequential arrangement of two carboxylic
acids, Glu-159 and Asp-161, together with WAT1 has some resem-
blance to aspartyl protease. A carboxyl protease inhibitor, a car-
bodiimide-mediated modification of Y202L, blocked the secondary
cleavage (Fig. 1C, Y202L⫹EDC). Monitoring the pH dependence
of the secondary autocleavage also indicated that deprotonation of
the carboxyl group was required for autocatalysis (Fig. 1C, Y202L
pH 5.0–7.0). However, the mutants D161L or D161N gave a fully
processed enzyme (Fig. 1B). Mutations in other candidate residues
that are close to the scissile peptide bond, S152A, R155K, and
R155M, also produced a successfully processed enzyme. Only two
mutants, E159M and E159Q, lost their secondary autocleavage
activities. To test the possibility of general base catalysis by gluta-
mate, the ␣⬘ subunits from S170C and E159Q mutants were
purified separately in the denatured condition (␣⬘S170C and ␣⬘E159Q,
respectively) and reconstituted with the denatured ␤ subunit
(␤mature) of the mature enzyme by refolding (Fig. 1C, Y202L
Inc兾Rec, lanes 4–6). The secondary cleavage occurred only when
the enzyme was reconstituted with the combination of ␣⬘S170C and
␤mature subunits, but did not when it was reconstituted with the
␣⬘E159Q and ␤mature subunits. Our results convincingly demonstrated
that Glu-159 is pivotal for this secondary processing.
Cation–␲ Interaction. The oxyanion binding of Arg-155 could be
stabilized by the carbonyl oxygens of Tyr-149 and Val-150 (Fig. 3,
Y202L). Importantly, the side chains of Tyr-149 and (Phe-177)
form a near edge-face aromatic interaction and (Phe-177) is in a
favorable arrangement for a cation–␲ interaction (27) with (Arg-
57) in the substrate binding site (Fig. 4). Similar interactions are
recognized as crucial for the structural stabilization of caspase BIR
domain (28) and neutrophil collagenase (29). Moreover, (Phe-177)
is the only residue of 682 modeled amino acid residues in a
disallowed region of the Ramachandran plot in all mutant crystal
Kim et al.
Fig. 4. Close-up view of the active site in Y202L, R226K, S170C, and E159Q
structures. Cation–␲ and aromatic interactions are shown in a ball-and-stick
representation. The electron densities from the 2 Fo ⫺ Fc maps were contoured
at 1.0 ␴. Hydrogen bonds are indicated by dotted lines, and water molecules
are represented as red spheres.
structures as well as in precursor and mature CA (see Fig. 8, which
is published as supporting information on the PNAS web site).
F177P mutant remains a nonprocessed precursor with no catalytic
activity (25). Notably, Y202L, R226K, and S170C mutants have the
defects localized in this region, where Tyr-202(Tyr-33), Arg-
226(Arg-57), and (S1) are replaced by leucine, lysine, and cysteine,
respectively. In Y202L, solvent water molecules are incorporated in
hydrogen-bonding networks, similar to precursor or mature CA.
The cation–␲ interaction in R226K may become weak because of
the mutation, resulting in a much slower processing mutant than
Y202L (data not shown). S170C and E159Q, on the other hand,
lack the hydrogen-bonding networks, probably interrupted by an
unexplained strong positive density found midway between (Arg-
57) and (Cys-1) or (Ser-1). Therefore, the perturbation in the
integrity of this region caused by mutation may result in a serious
defect in the secondary autocatalytic activity, suggesting that the
cation–␲ and hydrogen-bonding interactions may play an important
structural role in supporting proper autocatalysis.
Discussion
A glutamate residue acting as a general base was first discovered in
histone acetyltransferase, where the conserved glutamic acid is an
essential catalytic residue by deprotonating the histone substrate
(30). N-myristoyl transferase also has a conserved glutamic acid
residue that is essential in catalysis (31). Interestingly, chloram-
phenicol acetyltransferase uses a histidine residue as a general base
to abstract a proton from the primary hydroxyl group of chloram-
phenicol. A glutamic acid replacement of this histidine can substi-
tute as the general base, albeit with lower catalytic activity (32).
Mutational studies in other homologous CA proteins support our
notion that Glu-159 plays a critical role in secondary autoprote-
olysis: The Glu-159 mutants, E159L, E159M, and E159Q, of CA
from P. diminuta also underwent primary autocleavage but lost the
secondary autocleavage activity (18). In the crystal structure of
E159Q, the backbone conformation from Gly-158 to Asp-161 is
BIOPHYSICS
fairly similar to that of the Y202L and R226K forms (Fig. 3).
Based on the intermediate structures, we therefore propose that
CA may have evolved separate autoproteolytic mechanisms, the
first a serine proteolysis and the second a carboxyl proteolysis.
Upon protein folding, precursor CA undergoes the primary auto-
PNAS 兩 February 7, 2006 兩 vol. 103 兩 no. 6 兩 1735

Fig. 5. Proposed autoproteolytic activation mechanisms for CA. (A) Primary
serine autoproteolytic cleavage. A peptide flip (open arrow) under the influ-
ence of the strained conformation creates a hydrogen bond between WAT1
and the carbonyl group of Gln-168. WAT1, held by hydrogen bonds in
pseudotetrahedral geometry, acts as a general base and deprotonates the
hydroxyl group of Ser-170 for initiation of primary cleavage (curved arrow).
Dotted lines indicate hydrogen-bond interactions. WAT and WAT1 represent
a water molecule. (B) Secondary carboxyl autoproteolytic cleavage. It is initi-
ated by a general-base-catalyzed attack of WAT1 on the Gly-160 –Asp-161
peptide bond, where Glu-159 acts as a general base (curved arrows). The
main-chain NH of Arg-155 stabilizes the oxyanion of Gly-160.
proteolysis that uses a N 3 O acyl shift (Fig. 5A). Initially, a
high-energy state of the pro segment caused by conformational
constraints may trigger a peptide flip. WAT1, held by hydrogen
bonds in pseudotetrahedral geometry, acts as a general base to
deprotonate the hydroxyl group of Ser-170 to enhance its nucleo-
philicity. The hydroxyl group carries out the nucleophilic attack on
the peptide bond between Gly-169 and Ser-170. This primary
cleavage results in generating a helix-to-loop transition of the pro
segment, which consequently triggers an activation signal switch for
initiation of the secondary carboxyl autoproteolysis. Glu-159 acting
as a general base may accept a proton from WAT1 to enhance its
nucleophilicity (Fig. 5B). The nucleophilic attack on the peptide
bond between Gly-160 and Asp-161 is carried out by WAT1,
producing a tetrahedral transition state. Formation of the ester
intermediate (a N 3 O acyl shift) requires stabilization of the
oxyanionic transition state, which may be undertaken by the
backbone nitrogen of Arg-155. The secondary autoproteolysis
results in cleavage and subsequent dissociation of the pro segment,
producing the ␣ subunit. Autoproteolytic cleavage in Ntn hydro-
lases is mediated by intramolecular autocleavage reactions. CA then
1736 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0507862103
truly belongs to the Ntn hydrolase family of intramolecular self-
cleavage.
Two mutants, Y202L and R226K, showed significant decreases
in the rates of autocleavage reactions, thus being capable of
kinetically trapping the uncleaved and the cleaved forms of the pro
segment. We postulate that Y202L and R226K may represent
intermediate structures in transition from an immature to a mature
arrangement during activation, whereas S170C and E159Q may be
regarded as trapped intermediates. However, we cannot rule out
the possibility that the pro segment containing the Glu-159–WAT1
pair in Y202L or R226K may not represent a legitimate interme-
diate during autoproteolysis of CA or may adopt a slightly altered
conformation from that of a native intermediate. Nevertheless, our
results are strongly suggestive that the proposed mechanisms rep-
resent a common molecular mode of action for CA autoproteolysis.
In conclusion, our finding of dual catalysis at intervals of 9 aa,
both serine and a previously undiscovered carboxyl autoproteolysis,
may have novel implications for understanding the mechanisms
of autoproteolysis that are important in such diverse biological
processes.
Materials and Methods
Plasmids, Strains, and Site-Directed Mutagenesis. The pET23d plas-
mids harboring the cloned mature CA gene from Pseudomonas sp.
strain GK16 and the S170A mutant gene have been described (11).
The mutants S152A, R155K, R155M, E159M, E159Q, D161L,
D161N, S170C, Y202L, and R226K were constructed by site-
directed mutagenesis, using a PCR. Each PCR product was cloned
into vector pET22b or pET23d (Novagen), which carries a sequence
coded with a C-terminal His tag downstream from the T7 RNA
polymerase promoter site. The construct was then transformed into
an Escherichia coli DH5␣. The bacteria were grown at 37°C, and the
plasmids were purified by using a plasmid mini kit (Qiagen,
Valencia, CA). All mutations were verified by automated DNA
sequencing (data not shown). E. coli strains were grown in LB
medium.
Protein Expression and Purification. The clones were inserted into E.
coli BL21兾DE3 (Novagen) for protein overexpression, and bacteria
were grown at 37°C until the culture reached an absorbance of 0.6
at 600 nm. Protein expression was induced by addition of isopropyl
1-thio-␤-D-galactopyranoside (Sigma) to a final concentration of
0.4 mM, which was followed by an additional 4 h of incubation. The
purification of all of the mutant CA proteins was performed
according to published procedures (11). All of the mutant proteins
yielded proteins of sufficient purity for crystallization.
Analysis of Autocleavage. The extent of primary and secondary
autocleavage for the purified mutants was analyzed by SDS兾PAGE.
Y202L showed an approximately equimolar mixture of cleaved and
uncleaved ␣ subunits subsequent to purification (Y202L early:
Y202L). Both the ␣ subunit linked with the pro segment (␣⬘
subunit) and the ␣ subunit itself were subjected to MALDI-TOF
mass spectrometry (Korea Basic Research Institute, Seoul). The
time-course monitoring of in vitro secondary processing of S170C
or Y202L was carried out in 20 mM Tris (pH 8.0) at 37°C for 24 h.
S170C remained in the uncleaved form because of the lack of
secondary cleavage. For Y202L, it took ⬇24 h at 37°C, to make a
complete transition from the uncleaved to the cleaved form (Y202L
late: Y202L,L). From the solution processed at 37°C, 5-␮l aliquots
were sampled at 3-h intervals. The addition of the mature enzyme
to each mutant solution or 0.1 M 1-ethyl-3-(3-dimethylaminopro-
pyl) carbodiimide (EDC) to Y202L in 1.0 M glycine ethyl ester at
pH 5.0 at 25°C was used to detect changes in secondary processing.
For pH analysis of the secondary autocleavage of Y202L, time-
course monitoring was carried out at different pH values, ranging
from 5.0 to 8.0. For reconstitution experiments, ␣, ␣⬘, or ␤ subunit
was separately purified from mature CA, S170C, and E159Q in the
Kim et al.
denatured condition (␣mature and ␤mature, ␣⬘S170C, and ␣⬘E159Q,
respectively). They were reconstituted with the combination of
␣mature and ␤mature, ␣⬘S170C and ␤mature, or ␣⬘E159Q and ␤mature by
refolding. The results of reconstitution and time-course monitoring
experiments were analyzed by SDS兾PAGE.
Crystallization and Data Collection. Unlike autocleavage analysis,
only four mutants were crystallized for structural analysis at the
cleavage sites. The crystals of the two Y202L forms (Y202L and
Y202L,L), R226K, S170C, and E159Q were grown at 22°C, using
the hanging-drop vapor diffusion method. One to two microliters of
protein solution (30 mg兾ml) was mixed with an equal volume of well
solution and equilibrated against 1 ml of the well solution of 0.1 M
Na-cacodylate (pH 6.5), 0.2 M MgCl2, and PEG 6000. For Y202L,
PEG 3000 was used. Crystals were produced with a well developed
bipyramidal morphology within 3–4 days. Diffraction data were
collected with the crystals flash cooled at 100 K by using either a
laboratory x-ray source or a synchrotron radiation source, beamline
4A at Pohang Light Source (Pohang, Korea). All intensity data
were processed and scaled by using the programs DENZO and
SCALEPACK (33). The unit cell dimensions of the mutant protein
crystals of P41212 in this study were almost identical (see Table 1).
Structure Solution and Refinement. The mutant crystal structures of
Y202L, Y202L,L, R226K, S170C, and E159Q were solved by using
CNS (34) for molecular replacement, using the mature structure as
a search model (11). The initial solution was optimized by rigid body
refinement, which produced a clearly interpretable electron density
for the overall ␣ and ␤ subunit structures and the mutation site.
Manual adjustment of the backbone and side chain and the manual
building of the pro-segment residues was conducted in O (35).
Crystallographic refinement was carried out by using the program
REFMAC5 (36). The sigma A weighted 2 Fo ⫺ Fc maps were used to
perform a visual inspection during the rebuilding of the pro
segment with O (35). After a few rounds of model rebuilding, a
continuous electron density for the residues of the pro segment
1. Davie, E. W. (1995) Thromb. Haemostasis 74, 1–6.
2. Salvesen, G. S. & Dixit, V. M. (1999) Proc. Natl. Acad. Sci. USA 96, 10964–10967.
3. Dissing, M., Giordano, H. & DeLotto, R. (2001) EMBO J. 20, 2387–2393.
4. Perler, F. B. (1998) Cell 92, 1–4.
5. Babe, L. M. & Craik, C. S. V. (1997) Cell 91, 427–430.
6. Kahn, A. R. & James, M. N. G. (1998) Protein Sci. 7, 815–836.
7. Perler, F. B. (1998) Nat. Struct. Mol. Biol. 5, 249–252.
8. Brannigan, J. A., Dodson, G., Duggleby, H. J., Moody, P. C., Smith, J. L., Tomchick, D. R.
& Murzin, A. G. (1995) Nature 378, 416–419.
9. van Poelje, P. D. & Snell, E. E. (1990) Biochemistry 29, 132–139.
10. Kim, Y., Yoon, K., Khang, Y., Turley, S. & Hol, W. G. (2000) Structure (London) 8,
1059–1068.
11. Kim, J. K., Yang, I. S., Rhee, S., Dauter, Z., Lee, Y. S., Park, S. S. & Kim, K. H. (2003)
Biochemistry 42, 4084–4093.
12. Duggleby, H. J., Tolley, S. P., Hill, C. P., Dodson, E. J., Dodson, G. & Moody, P. C. (1995)
Nature 373, 264–268.
13. Suresh, C. G., Pundle, A. V., SivaRaman, H., Rao, K. N., Brannigan, J. A., McVey, C. E.,
Verma, C. S., Dauter, Z., Dodson, E. J. & Dodson, G. G. (1999) Nat. Struct. Mol. Biol. 6,
414–416.
14. Smith, J. L., Zaluzec, E. J., Wery, J. P., Niu, L., Switzer, R. L., Zalkin, H. & Satow, Y. (1994)
Science 264, 1427–1433.
15. Lowe, J., Stock, D., Jap, B., Zwickl, P., Baumeister, W. & Huber, R. (1995) Science 268,
533–539.
16. Oinonen, C., Tikkanen, R., Rouvinen, J. & Peltonen, L. (1995) Nat. Struct. Mol. Biol. 2,
1102–1108.
17. Bompard-Gilles, C., Villeret, V., Davies, G. J., Fanuel, L., Joris, B., Frere, J. M. & Van
Beeumen, J. (2000) Structure (London) 8, 153–162.
18. Kim, Y., Kim, S., Earnest, T. N. & Hol, W. G. J. (2001) J. Biol. Chem. 277, 2823–2829.
19. Hewitt, L., Kasche, V., Lummer, K., Lewis, R. J., Murshudov, G. N., Verma, C. S., Dodson,
G. G. & Wilson, K. S. (2000) J. Mol. Biol. 302, 887–896.
20. Ditzel, L., Huber, R., Mann, K., Heinemeyer, W., Wolf, D. H. & Groll, M. (1998) J. Mol.
Biol. 279, 1187–1191.
Kim et al.
could be obtained at the 0.5- to 0.8-␴ level depending on mutants.
At 1.0 ␴, there was no electron density visible for the region. In the
case of Y202L, the pro-segment atoms were modeled with half-
occupancy, based on the equimolar ratio of the cleaved and
uncleaved ␣ subunits on SDS兾PAGE (Fig. 1C, Y202L Inc兾Rec,
lanes 1 and 2). Nevertheless, the electron density for the pro-
segment backbone was quite clear in the mutant structures. The
residues up to Ala-166, Leu-165, Asp-164, and Asp-164 of the pro
segments could be modeled in the Y202L, R226K, S170C, and
E159Q structures, respectively. More importantly, the electron
density map for the secondary cleavage site including residues from
Glu-159 to Gly-160 was readily interpretable. Water molecules were
added by using the Fo ⫺ Fc map peaks ⬎3.0 ␴, if the putative water
molecule made possible hydrogen-bonding contacts and had no
adverse van der Waals contacts. Some water molecules contacting
the pro segment were added by using the Fo ⫺ Fc map peaks ⬍3.0
␴, if the B factors were ⬍60 Å2 after refinement. The free R value
was used as an indicator to validate the water picking and refine-
ment procedure and to guard against possible overfitting of the data
(37). The stereochemical analysis of all of the refined structures by
using PROCHECK (38) showed four prolines [Pro-131 of the ␣
subunit and (Pro-253), (Pro-379), and (Pro-466) of the ␤ subunit]
in a cis-peptide conformation and one residue (Phe-177) of the ␤
subunit in a disallowed region of the Ramachandran plot (see Fig.
8). Fig. 3 was prepared with the programs MOLSCRIPT (39) and
RASTER3D (40). Figs. 2, 4, 5, 6B, and 7 were prepared with the
program PYMOL (www.pymol.org; ref. 41).
We thank the staff at beamline 4A, Pohang Light Source for help with
data collection and Prof. T. Blundell, Dr. Z. Dauter, and Dr. K. H. Han
for comments on the manuscript. This work was supported by Ministry
of Science and Technology Grant R05-2004-10651, Korea Research
Foundation Grant KRF-2005-C00036, and Biogreen 21 Program 2005-
0301034369, Rural Development Administration, Suwon, Korea. J.K.K.,
I.S.Y., K.J.C., E.K.R., and S.H.K. were supported by the BK21 program
of the Ministry of Education, Seoul.
21. Xu, Q., Buckley, D., Guan, C. & Guo, H. (1999) Cell 98, 651–661.
22. Albert, A., Dhanaraj, V., Genschel, U., Khan, G., Ramjee, M. K., Pulido, R., Sibanda, B. L.,
von Delft, F., Witty, M. & Blundell, T. L., et al. (1998) Nat. Struct. Mol. Biol. 5, 289–293.
23. Lee, Y. S. & Park, S. S. (1998) J. Bacteriol. 180, 4576–4582.
24. Fritz-Wolf, K., Koller, K. P., Lange, G., Liesum, A., Sauber, K., Schreuder, H., Arets, W.
& Kabsch, W. (2002) Protein Sci. 11, 92–103.
25. Kim, S. & Kim, Y. (2001) J. Biol. Chem. 276, 48376–48381.
26. Qian, X. Q., Guan, C. & Guo, H. (2003) Structure (London) 11, 997–1003.
27. Zacharia, N. & Dougherty, D. A. (2002) Trends Pharmacol. Sci. 23, 281–287.
28. Luque, L. E., Grape, K. P. & Junker, M. (2002) Biochemistry 41, 13663–13671.
29. Gavuzzo, E., Pochetti, G., Mazza, F., Gallina, C., Gorini, B., D’Alessio, S., Pieper, M.,
Tschesche, H. & Tucker, P. A. (2000) J. Med. Chem. 43, 3377–3385.
30. Tanner, K. G., Trievel, R. C., Kuo, M. H., Howard, R. M., Berger, S. L., Allis, C. D.,
Marmorstein, R. & Denu, J. M. (1999) J. Biol. Chem. 274, 18157–18160.
31. Weston, S. A., Camble, R., Colls, J., Rosenbrock, G., Taylor, I., Egerton, M., Tucker, A. D.,
Tunnicliffe, A., Mistry, A., Mancia, F., et al. (1998) Nat. Struct. Mol. Biol. 5, 213–221.
32. Lewendon, A., Murray, I. A., Shaw, W. V., Gibbs, M. R. & Leslie, A. G. (1994) Biochemistry
33, 1944–1950.
33. Otwinowski, Z. & Minor, W. (1997) Methods Enzymol. 276, 307–326.
34. Brunger, A. T., Adams, P. D., Clore, G. M., Delano, W. L., Gros, P., Grosse-Kunstleve,
R. W., Jiang, J. S., Kuszewski, J., Nilges, M., Pannu, N. S., et al. (1998) Acta Crystallogr. D
54, 905–921.
35. Jones, T. A., Zou, J. Y., Cowan, S. W. & Kjeldgaard, M. (1991) Acta Crystallogr. A 47,
110–119.
36. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997) Acta Crystallogr. D 53, 240–255.
37. Brunger, A. T. (1992) Nature 355, 472–475.
38. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thorton, J. M. (1993) J. Appl.
Crystallogr. 26, 283–291.
39. Kraulis, P. J. (1991) J. Appl. Crystallogr. 24, 946–950.
40. Merritt, E. A. & Murphy, M. E. P. (1994) Acta Crystallogr. D 50, 869–873.
41. DeLano, W. L. (2002) The PYMOL Molecular Graphics System (DeLano Scientific, San
Carlos, CA).
BIOPHYSICS
PNAS 兩 February 7, 2006 兩 vol. 103 兩 no. 6 兩 1737