Tài liệu Báo cáo khoa học: ˚ The 1.8 A crystal structure of a proteinase K-like enzyme from a psychrotroph Serratia species docx

The 1.8 A
˚
crystal structure of a proteinase K-like enzyme
from a psychrotroph Serratia species
Ronny Helland
1
, Atle Noralf Larsen
2
, Arne Oskar Smala
˚
s
1,3
and Nils Peder Willassen
1,2
1 Norwegian Structural Biology Centre, Faculty of Science, University of Tromsø, Tromsø, Norway
2 Department of Molecular Biotechnology, Faculty of Medicine, University of Tromsø, Tromsø, Norway
3 Department of Chemistry, Faculty of Science, University of Tromsø, Tromsø, Norway
Peptidases are enzymes that catalyze the hydrolysis of
peptide bonds in other proteins, and are often used in
biotechnology and other industries. About 60% of the
industrial enzymes sold are peptidases [1]. Neutral pep-
tidases are active within a relatively narrow range
(pH 5–8) and are often used in the food industry.
Alkaline peptidases are generally highly active around
pH 10, have broad substrate specificity and are active
at high temperatures (around 60 °C). Due to the vast
amount of sequence, structure and kinetic data, peptid-
ases are excellent model systems for studying factors
important for protein stability, substrate specificity and
catalytic efficiency. Knowledge about these factors can
be further exploited in the redesign of proteins in order

to give enzymes with altered specificity and ⁄ or bio-
physical properties ([2–5] for redesign of subtilases and
[6–8] for rational redesign in general).
Enzymes from microorganisms adapted to low tem-
peratures (0 °Cto4°C), so-called psychrophiles, are
often characterized by higher catalytic efficiency and
lower temperature stability than their homologues from
more temperate species. The increased efficiency appears
to be at the expense of the stability, and is believed to be
caused by a more flexible structure [9,10]. Increased
Keywords
proteinase K; psychrophilic; S’ binding;
substrate specificity; subtilase; subtilisin
Correspondence
R. Helland, Norwegian Structural Biology
Centre, Faculty of Science, University of
Tromsø, 9037 Tromsø, Norway
Fax: +47 77644765
Tel: +47 77646474
E-mail:
Website:
Enzymes
Serratia sp. peptidase (EC 3.4.21 ); protein-
ase K (EC 3.4.21.64); Vibrio sp. peptidase
(EC 3.4.21 ).
(Received 7 September 2005, revised 25
October 2005, accepted 31 October 2005)
doi:10.1111/j.1742-4658.2005.05040.x
Proteins from organisms living in extreme conditions are of particular
interest because of their potential for being templates for redesign of

enzymes both in biotechnological and other industries. The crystal struc-
ture of a proteinase K-like enzyme from a psychrotroph Serratia species
has been solved to 1.8 A
˚
. The structure has been compared with the struc-
tures of proteinase K from Tritirachium album Limber and Vibrio sp. PA44
in order to reveal structural explanations for differences in biophysical
properties. The Serratia peptidase shares around 40 and 64% identity with
the Tritirachium and Vibrio peptidases, respectively. The fold of the three
enzymes is essentially identical, with minor exceptions in surface loops.
One calcium binding site is found in the Serratia peptidase, in contrast to
the Tritirachium and Vibrio peptidases which have two and three, respect-
ively. A disulfide bridge close to the S2 site in the Serratia and Vibrio pep-
tidases, an extensive hydrogen bond network in a tight loop close to the
substrate binding site in the Serratia peptidase and different amino acid
sequences in the S4 sites are expected to cause different substrate specificity
in the three enzymes. The more negative surface potential of the Serratia
peptidase, along with a disulfide bridge close to the S2 binding site of a
substrate, is also expected to contribute to the overall lower binding affinity
observed for the Serratia peptidase. Clear electron density for a tripeptide,
probably a proteolysis product, was found in the S’ sites of the substrate
binding cleft.
Abbreviations
PRK, proteinase K; SPRK, Serratia proteinase K; Suc-XXXX-pNa, succinyl-(amino acid)
4
-p-nitroanilide where X represent the one letter code of
the amino acid; VPRK, Vibrio sp. PA44 proteinase K.
FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS 61
flexibility of the molecule reduces the activation energy
for formation of reaction intermediates, and results in

a more efficient turnover of the substrates. The low
temperature optimum and temperature stability makes
psychrophilic enzymes particularly useful in reactions
performed at low temperatures and in reactions that
needs an easy inactivation.
In a quest for finding peptidases with properties
which could be further evolved into biotechnological
or industrial use proteinase K, a serine peptidase of
the subtilisin superfamily, or subtilases, was chosen as
a model enzyme. Proteinase K is a very stable enzyme
with broad substrate specificity which previously has
been used primarily for removal of RNase activity
[11], but has found other applications such as identi-
fication and decontamination of the prion protein
responsible for scrapie [12,13]. The binding region in
the subtilases is capable of accommodating at least six
amino acid residues (P4–P2¢; notation according to
Schechter and Berger [14]) of a polypeptide substrate
or inhibitor [15], and both main chain and side chain
interactions contribute to binding. Although several
subsites contribute to specificity in the subtilases, the
most important interactions are those between the sub-
strate and the S1 and S4 binding sites [16], where favo-
rable P4–S4 interactions seem to eliminate or reduce
unfavorable effects introduced at other subsites [17].
A new proteinase K-like enzyme isolated from a
psychrotroph Serratia species (SPRK), sharing about
40% sequence identity with proteinase K from fungus
Tritirachium album Limber (PRK), has been identified,
cloned and expressed [18]. The biophysical characteri-

zation of SPRK did not reveal the classical cold adap-
ted features [19], but still initial comparative studies
showed that the catalytic turnover was at least twice
that of PRK, but substrate affinity was reduced. SPRK
was compared with PRK and was found to be remark-
able stable against SDS denaturation while being con-
siderable more labile towards the presence of EDTA.
In this paper, the 3D structure of SPRK is described
and compared to the structures of PRK (PDB 1ic6 [20]),
and Vibrio sp. PA44 (VPRK; PDB 1sh7 [21]), and an
attempt is made to give structural explanations for some
of the differences observed in biophysical properties. In
addition, the presence of a tripeptide bound to the prime
side (S¢) of the active site is reported.
Results and discussion
Crystallization, data collection and refinement
Crystals of the catalytic domain of the Serratia prote-
inase K (SPRK) grew within a few days to thin hexa-
gonal plates growing in bundles, and were up to
0.2 · 0.1 · 0.02 mm
3
. A complete data set to 1.8 A
˚
was collected. Data collection and refinement charac-
teristics are listed in Table 1. The crystals belonged to
space group C2 with unit cell of 86 · 42 · 74 mm
3
and b ¼ 110.2° and with one molecule in the asymmet-
ric unit. R
sym

(R
sym
¼ R ŒI ) <I>Œ⁄RI, where I is the
observed intensity and <I> is the average intensity)
was about 11%, I ⁄ rI was about 6 and multiplicity was
about 3.7. The final R-factors are 16.0% (R
work
) and
19.3% (R
free
).
The refined structure contains about 3900 protein
atoms, including a tripeptide found in the active site
binding cleft, and 225 solvent atoms including two
SO
4
2–
ions and one Ca
2+
ion. The average B-factor is
8.86 A
˚
2
for all enzyme atoms and 10.05 A
˚
2
for all
atoms. The electron density is well defined for the
whole molecule and only a small number of side chains
on the molecular surface could not be fitted into elec-

tron density. One residue was refined with alternate
conformations.
The structure of the Serratia peptidase
Overall fold
The sequence (Fig. 1) identity between SPRK and the
two other peptidases is about 40 and 64% (PRK and
VPRK, respectively), but still the fold of SPRK is virtu-
ally identical to PRK and VPRK (Fig. 2). The struc-
tures of the two bacterial peptidases are more similar to
each other than to PRK. SPRK superimposes on PRK
and VPRK with an root mean square deviation (rmsd)
value of 1.47 and 0.66 A
˚
, respectively, for all main chain
atoms, excluding the inserted or deleted regions.
Table 1. Data collection and refinement statistics for SPRK.
Resolution (A
˚
)1.8
Space group C2
Cell parameters
a-axis (A
˚
) 86.43
b-axis (A
˚
) 42.56
c-axis (A
˚
) 74.26

b angle (°) 110.19
Number of observations (24 – maximum resolution) 108140
Number of unique reflections 30408
R
sym
15.0 A
˚
– resolution limit (%) 11.4 (38.1*)
Completeness (%) 100 (100*)
I ⁄ rI 6.0 (1.9*)
Wilson B (A
˚
2
) 11.57
Multiplicity 3.7
R
work
(%) 16.0
R
free
(%) 19.3
*Outer shell: 1.9–1.8 A
˚
Structure of proteinase K from a Serratia sp. R. Helland et al.
62 FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS
Eight regions in the three proteinase K structures
can be considered to display differences in the fold of
the main chains. None of these includes any of the
binding sites, but some are located in the vicinity.
Three of the regions with different conformation

involve calcium binding and four of them involve
insertion or deletion of amino acids. This can probably
influence the flexibility and stability of the enzymes.
SPRK and VPRK have an insertion of three resi-
dues in the loop comprising residues 58–65 after resi-
due 60. At the end of this loop SPRK and VPRK
have a cysteine residue, Cys66, which forms a disulfide
bridge to residue 98, which is in the proximity of the
S2 binding site. Cys66 is also close to the catalytic
triad residue His69. The potential effect of the disulfide
bridges will be discussed further below. The region
Fig. 1. Structural alignment of SPRK, PRK and VPRK. Helices (red tubes) and sheets (yellow arrows) are according to SPRK. Residues
belonging to the S1 site are shaded blue and residues belonging to the S4 site are shaded light blue. *represents residues that are involved
both in the S1 and S4 sites. Residues forming the calcium binding site found in SPRK (and VPRK) are shaded green, residues forming the
‘strong’ calcium binding site in PRK (and in VPRK) are shaded khaki, and residues forming the calcium sites unique for either PRK or VPRK
are shaded pale green.
Fig. 2. Stereo plot of SPRK (red), PRK (green) and VPRK (blue). The catalytic triad residues of SPRK are represented as ball-and-stick models,
and calcium ions are represented as spheres and follow the coloring of enzymes. Sulfate ions found in SPRK are drawn as yellow spheres.
Cyan and magenta ball-and-stick models represent disulfide bridges in SPRK and VPRK, and yellow represent disulfide bridges in PRK. Selec-
ted loop legions with significant structural differences are labeled.
R. Helland et al. Structure of proteinase K from a Serratia sp.
FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS 63
comprising residues 58–63 (56–63 in VPRK) also binds
a calcium ion in VPRK and this will be further dis-
cussed below. Five residues are deleted in the loop
comprising the residues 120–127 in SPRK and VPRK.
PRK has one of its two disulfide bridges (Cys123–
Cys34) located in this region. The loop formed by resi-
dues 240–245 in the bacterial species has conformation
slightly different from PRK due to an insertion of one

residue. The only region where SPRK is different from
both of the other two enzymes is after residue 213,
where two residues are inserted. The length of the loop
comprising residues 16–29 is identical in the three
enzymes, but the conformation is slightly different. A
calcium ion is present in SPRK and VPRK in this
loop, but is absent in PRK. This will be discussed fur-
ther below. The helix from residues 140–151 is slightly
translated in SPRK and VPRK relative to PRK, poss-
ibly due to Val and Asn at positions 145 and 146 in
SPRK (both Ala in PRK) and Val and Glu in VPRK.
Small differences are also observed for residues 175–
177. These are two of the residues comprising the
‘strong’ calcium binding site in PRK and will be dis-
cussed below. Position 178 in PRK is occupied by a
cysteine which forms a disulfide bridge to residue 249.
Minor differences are observed in the region 263–270.
This is a region where VPRK has a deletion of one
residue.
Calcium binding site
Binding of calcium ions in subtilases has been shown to
be essential for correct folding and stability [15]. A total
of four calcium binding sites have been reported for the
proteinase K-like enzymes, and we define the sites as fol-
lows. Ca1 constitute the new binding site reported by
[21] and is formed by residues Asp11, Asp14, Gln15,
Asp21 and Asp23 (SPRK numbering), and is involved
in stabilizing the N-terminal region of the enzyme. Ca2
is the site observed in VPRK and thermitase [22] which
is formed by residues Asp58, Asp60D and Asp62

(SPRK numbering). This site stabilizes the loop between
strand b4 and helix a2 where VPRK and SPRK have an
insert of three residues. Ca3 is the so-called strong bind-
ing site found in PRK and formed by the side chain of
Asp200 and the carbonyl oxygen atoms of residues 175
and 177 and is involved in stabilizing strands b8 and b9.
Ca4 is the weak binding site in PRK, which is formed by
the side chain of Asp260 and the carbonyl oxygen of
Thr16, and stabilizes the N- and C-terminal regions of
the molecule.
Only one calcium binding site is found in SPRK, in
contrast to PRK which has two and VPRK which has
three. The Ca
2+
ion in SPRK binds at the Ca1 site,
and it is coordinated to the carboxyl oxygen atoms of
the side chains of Asp11, Asp14 and Asp21, the amide
oxygen atom of Gln15 and the carbonyl oxygen atoms
of Asp11 and Asn23 (Fig. 3) in an arrangement similar
to what is observed in VPRK. Both PRK and VPRK
have calcium bound at Ca3. SPRK also has an aspar-
tic acid residue at position at 200, and with the same
conformation as in the two other structures, but
Asp200 forms a salt bridge to Lys253 (Ala253 and
Asn250 in PRK and VPRK, respectively), and prob-
ably loses some of its potential as calcium ligand. The
Fig. 3. Stereo plot of the calcium binding site in SPRK.
Structure of proteinase K from a Serratia sp. R. Helland et al.
64 FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS
amine group of the lysine side chain is about 4.6–4.8

A
˚
from the position of calcium in PRK and VPRK.
Position 260 at the second calcium binding site in
PRK, Ca4, is occupied by an aspartic acid. This posi-
tion is a lysine in SPRK and VPRK, and thus prevents
binding of calcium. The side chain of this lysine forms
a hydrogen bond to Asn17 in SPRK (15 in VPRK). In
addition, Arg16 in SPRK forms hydrogen bonds to
residues 255 and 257, thus stabilizing the N- and C-ter-
minal regions which are stabilized by calcium in PRK.
The residues forming the second calcium binding site
in VPRK, Ca2, are identical in SPRK and VPRK, so
there is no obvious explanation to why calcium is
absent in this site in SPRK. Calcium binding in SPRK
could be prevented by the side chain of Arg94 which is
about 4.0 A
˚
from the position occupied by calcium in
VPRK. However, both VPRK and thermitase also
have an arginine at position 94.
SPRK loses about 40% of its activity when incuba-
ted at 37 °C for 2 h in the presence of EDTA [18],
while PRK is essentially unaffected. At 50 °C SPRK
was totally inactivated, while PRK still had approxi-
mately 50% activity remaining. The crystal structure
of SPRK shows that the calcium ion is tightly bound
with six protein atom binding partners, while the two
calcium ions are bound weaker to the protein in PRK,
with four and three protein ligands for binding sites

Ca1 and Ca2, respectively. This could probably explain
why SPRK is less stable than PRK towards EDTA,
and suggests that the calcium ion is more crucial for
stability in SPRK. VPRK has two binding sites that
bind calcium stronger than PRK, and this could prob-
ably explain why also the stability of VPRK depends
strongly on calcium [21].
Disulfide bridges
Two disulfide bridges are found in the crystal structure
of SPRK, as in PRK, whereas VPRK has three. The
disulfide bridges in SPRK occupy the same positions
as two of the bridges in VPRK (Figs 1 and 2), and
sequence alignment [18] suggests that the third disul-
fide bridge found at the C-terminal in VPRK could in
principle be present in SPRK too if the C-terminal was
not cleaved off. The bridges are formed by residues
66–98 and 167–198 in SPRK, 67–99, 163–194 and 277–
281 in VPRK (VPRK numbering) and 34–123 and
178–249 in PRK. One of the disulfide bridges in SPRK
and VPRK is close to the S2 binding site. It anchors a
tight loop (see below) which is part of the S2 site, to
the rest of the molecule, and may therefore affect the
catalytic activity of the enzyme by reducing the flexibil-
ity of the protein. Reduced flexibility may prevent effi-
cient binding of a substrate, something which can be
illustrated by the lower binding affinity SPRK has for
Suc-AAPF-pNA than PRK does [18]. Docking this
substrate into the active site of SPRK shows that the
C
d

atom of the P2 residue (here Pro) would be only
2.4 A
˚
away from the carbonyl oxygen atom of residue
100 (data not shown). The structure of proteinase K in
complex with the inhibitor methoxysuccinyl-AAPA-
chlorometylketone (PDB 3prk [23]) shows that the
main chain atoms of the residues forming the S2 site
(residues 96 and 100) move more than 1 A
˚
to accom-
modate a proline residue at the P2 position. The sec-
ond disulfide in SPRK (167–198) and VPRK (163–194)
is located in a region close to the bottom of the S1
pocket. This bridge can be considered to be involved
in the stabilization of the region which is stabilized by
calcium (Ca3) in PRK (residues 200 and 175–177).
Both disulfide bridges in PRK are further away from
the active site than in SPRK and VPRK, and are
expected to influence the activity to a lesser degree.
They could, however, influence stability.
Salt bridges
The three enzymes have 18, 19 and 15 residues forming
24, 23 and 18 interactions less than 4 A
˚
(SPRK, PRK
and VPRK, respectively) between atoms of positively
and negatively charged residues, of which 6, 3 and 5
include histidine. The average length is quite similar in
the three enzymes, 3.15 A

˚
in SPRK and 3.21 A
˚
in
PRK and VPRK. None of enzymes have extended salt
bridge networks but both SPRK and PRK have salt
bridges involving three residues [Lys265–Asp263–
Arg189 in SPRK (Pro, Asn, Arg in PRK, Arg, Asp,
Arg in VPRK) and Asp187–Arg12–Asp260 in PRK
(Asp, Arg, Lys in SPRK and VPRK)]. The three-resi-
due network observed in PRK is not possible in SPRK
and VPRK due to the Lysine residue in position 260.
A reduced number of salt bridges are generally accep-
ted to be one of the properties of enzymes adapted to
cold environments. The reduced number is believed to
increase the flexibility of an enzyme, and hence, the
catalytic efficiency, but at the expense of the stability
towards temperature (for reviews see [19,24,25]). SPRK
from the psychrotrophic Serratia sp. is not found to
be considerable less stable than PRK [18], but VPRK
is [26,27]. This could possibly in part be related to the
number of salt bridges in the three enzymes.
Position 98 close to the S2 site is involved in a disul-
fide bridge in SPRK and VPRK. The Asp residue in
this position in PRK forms a salt bridge to Lys94. It is
not yet clear to what extent this interaction would
affect PRK specificity at the S2 site.
R. Helland et al. Structure of proteinase K from a Serratia sp.
FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS 65
Structure in the binding region

The electrostatic surface calculated by icm (Fig. 4 [28])
shows that the bacterial peptidases have a more ani-
onic character than the fungal species. The amino acid
sequence in the binding region is very similar in the
three enzymes, and so is the conformation of the side
chains, but some differences exist. The amino acid
sequence in the S1 site is conserved, except at positions
162 and 169 located at the bottom of the pocket. The
neutral Asn162 in PRK is replaced by the polar and
smaller Ser residue in VPRK, and by the acidic Asp in
SPRK. The conformation of Asn162 and Asp162 is
practically identical in the two enzymes, but the acidic
residue would contribute to the more anionic character
of the S1 site in SPRK. Asn or Asp at position 162
would be about 5 A
˚
from the position occupied by a
docked P1 Phe residue (data not shown). Position 169
is occupied by Tyr in SPRK and PRK and Thr in
VPRK. The residue at this position in subtilisin BPN
has been subjected to extensive mutational studies in
order to modify specificity [4,5]. Tyr169 in SPRK and
PRK is not likely to interact directly with a P1 residue
of a substrate. For such interactions to take place, the
side chain will have to rotate, and hence, cause steric
clashes with the enzyme main chain that will require
energy to resolve. The most obvious difference between
the binding sites of the three enzymes is found in the
S4 site at residues 103, 104, 107 and 141 where all
three enzymes have different amino acid sequence.

This segment is Gln, Tyr, Ile and Val in PRK, Thr,
Thr, Val and Leu in VPRK and Ser, Asn, Val and Thr
in SPRK. The fungal enzyme thus has a smaller and
more hydrophobic binding site than the two other
enzymes. One of the most interesting differences
between the bacterial and fungal enzymes is located at
the periphery of the S2 site. As mentioned above,
Asp98 in PRK is replaced by a cysteine in the other
two proteins, but the residues on both sides of this
position are also of interest because the region 97–101
forms a tight loop which is stabilized differently in the
three enzymes. The side chains of residues Asn97,
Ser99 and Ser101 in SPRK (Asp, Asn, Ser and Ser,
Ser, Ser in PRK and VPRK, respectively) form a
strong hydrogen bond network which brings the con-
formation of the loop almost into a ‘hub and spokes’
arrangement (Fig. 5). The side chain of Asn99 in PRK
is too long to participate efficiently in a similar
arrangement, and the side chain of Ser97 in VPRK is
too short. One side of the loop is involved in the bind-
ing of the P2–P4 residues of a substrate. The other side
of the loop is close to Arg94 (Lys in PRK), which in
SPRK and VPRK are further stabilized by salt bridges
to residues in the 58–65 loop. Hence, the region con-
taining residues 97–101 is likely to be more rigid in
SPRK than in the other two enzymes. It is therefore
Fig. 4. Electrostatic surface potential of SPRK (A), PRK (B) and
VPRK (C). The surfaces are contoured at 5.0 kcalÆe.u.
)1
charge

where red regions represent negative potential and blue regions
represent positive potential.
Structure of proteinase K from a Serratia sp. R. Helland et al.
66 FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS
not unreasonable to believe that the three enzymes will
have a different specificity profile not only for the S4
site, but possibly also in the S2 site.
Tripeptide in the active site cleft
Extra electron difference density was observed in the
prime site of the substrate binding region of SPRK.
This was initially interpreted as a Pro-Ser dipeptide,
but was after a few cycles of refinement interpreted as
an Ala-Pro-Thr tripeptide (Fig. 6). Attempts were
made to fit the ligand into the initial difference density
as a buffer or cryprotectant (glycerol) molecule, but it
did not fit the electron density maps as well as a pep-
tide. Average B-factors for the peptide is about 20 A
˚
2
,
which is higher than the average for the enzyme (about
Fig. 5. Stereo plot illustrating the tight loop forming the S2 binding site. Red is SPRK, green is PRK and blue is VPRK. Labels and distances
in the loop refer to SPRK. The stabilizing hydrogen bonding network formed by residues Asn97, Ser99 and Ser101 in SPRK is displayed as
ball-and-stick models. A similar network is not as strong in PRK and VPRK due to a shorter Ser97 in VPRK and a too-long Asn99 in PRK. The
loop is anchored to the rest of the molecule in SPRK and VPRK by a disulfide bridge between Cys98 and Cys66 (Asp98 in PRK). The cata-
lytic His69 is displayed as ball-and-stick model in order to illustrate the orientation of the loop relative to the binding site.
Fig. 6. Stereo plot illustrating the electron density surrounding the Ala-Pro-Thr tripeptide found in the S1¢ and S2¢ binding site of SPRK. The
2fofc electron density is contoured at 1.0 r.
R. Helland et al. Structure of proteinase K from a Serratia sp.
FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS 67

9A
˚
2
), indicating reduced occupancy. The B-factor is,
however, in the same order as the average of the
solvent molecules. The tripeptide is most likely a pro-
teolysis product from the expression or purification
process. The gene for SPRK codes for a pre-pro
enzyme including an N-terminal chaperone like
domain, the catalytic domain and two similar C-ter-
minal domains with unknown structure and function.
The N-terminal pro domain is autocatalytically cleaved
off when the enzyme has obtained its active conforma-
tion, and the two C-terminal domains are cleaved off
by heat treatment at 50 °C. An Ala-Pro-Thr sequence
is located in the first C-terminal domain.
The tripeptide is not located in the primary binding
site (the S1 site), where most inhibitors, substrates and
substrate analogues are observed in the structures of
serine peptidase complexes. Instead it is located in the
S1¢ and S2¢ sites, and Thr3 and Pro2 occupy the same
space as P1¢ and P2¢ residues found for naturally
occurring protein inhibitors in complex with other sub-
tilases (Fig. 7); subtilisin BPN in complex with chymo-
trypsin inhibitor 2 (CI2; PDB 1lw6, 1tm1 and 2sni
[29–31]) or eglinC (PDB 1sbn [32]) and subtilisin Carls-
berg in complex with eglinC (PDB 1cse [33]), ovomu-
coid turkey inhibitor 3 (OMTKY3; PDB 1r0r [34]) or
the double-headed tomato inhibitor II (TI2; PDB 1oyv
[35]).

The tripeptide is hydrogen bonded to the enzyme by
interactions only through Thr3. The amine atom
(Thr3 N) forms a hydrogen bond to Ser221 O
(2.75 A
˚
). One of the terminal carboxyl oxygen atoms
of Thr3 is hydrogen bonded to the catalytic His69 N
e2
(2.88 A
˚
) and Ser224 O
c
(2.68 A
˚
) and the second ter-
minal carboxyl oxygen atom occupy a position in the
so-called oxyanion hole, with hydrogen bonds to
Asn161 O
d1
(2.90 A
˚
) and Ser224 O
c
(3.16 A
˚
). The
position of the second terminal oxygen atom of Thr3
atom is about 0.8–1.5 A
˚
from the position occupied by

the carbonyl oxygen atom of the scissile bond in other
subtilase–inhibitor complexes (data not shown). The
C
a
positions of the tripeptide occupy almost the same
positions as the C
a
positions of the inhibitors. Intro-
duction of the tripeptide causes a slight rotation of
Ser224, such that the hydroxyl group of the side chain
weakens its catalytic triad interaction with His69 N
e2
and instead forms hydrogen bonds to both terminal
oxygen atoms of Thr3.
Another interesting feature is that the direction of
the main chain of the tripeptide is antiparallel to the
naturally occurring inhibitors. It was first speculated
whether the C-terminal domain in an uncleaved state
could fold back into its own active site, or alternatively
into the active site of a neighboring molecule, some-
thing which could explain the antiparallel binding.
However, this can not be verified here since the C-ter-
minal is cleaved off.
The tripeptide has a conformation most similar to
OMTKY3, CI2 and the second binding region of TI2
(Fig. 7). The first binding region of TI2 and eglinC
have slightly different orientation from the P2¢ resi-
due, probably resulting from different scaffolds of the
inhibitors. Structures of PRK with inhibitors or sub-
Fig. 7. Stereo plot illustrating inhibitor binding in subtilases. The binding loop of chymotrypsin inhibitor II (CI2) in complex with subtilisin BPN

(cyan, PDB 2sni), the inhibitor fragment of lactoferrin in complex with PRK (green, PDB 1bjr), a hepta-peptide in complex with PRK (yellow,
PDB 1p7v) are docked on the ribbon figure of SPRK. The tripeptide found in SPRK is illustrated as red ball and sticks.
Structure of proteinase K from a Serratia sp. R. Helland et al.
68 FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS
strate analogues at the prime side of the binding
region do exist, e.g. hepta- or octapeptide substrate
analogues or an inhibiting decapeptide fragment of
lactoferrin. However, these peptides have a completely
different binding mode compared to the naturally
occurring inhibitors. They bind in the S1–S1¢ region,
but not always at the other subsites of the enzymes.
It is therefore not clear as yet whether there are two
distinct S’ binding sites in the proteinase K family or
whether the different binding modes can be related,
i.e. to d-amino acids in the P1¢ position of the pep-
tide inhibitors, thus forcing the amino acids on the
prime side of the binding site into a different confor-
mation.
Conclusion
The structure of a proteinase K-like enzyme of a psy-
chrotroph Serratia species has been solved to 1.8 A
˚
.
The fold of the protein reported here is practically
identical to the two other enzymes of this family with
known 3D structure, with only minor differences in
the surface loops. In addition, an Ala-Pro-Thr tripep-
tide is observed at the prime side of the active site,
and the direction of the tripeptide is antiparallel to
what would be expected of a naturally occurring pro-

teinaceous inhibitor. The protein presented in this
paper is more similar to the bacterial species (VPRK)
than the fungal species (PRK) with respect to disul-
fide bridges, calcium binding and surface loops. This
also coincides with the sequence identity, which is
about 64% to VPRK and about 40% to PRK. Only
one calcium binding site is observed in the structure
presented here, in contrast to the other proteinase K
structures where two or more calcium binding sites
are observed. The calcium ion in this structure, sim-
ilar to Ca1 in VPRK, is coordinated to six protein
atoms, indicating a very strong binding, whereas the
calcium ions in PRK are coordinated to not more
than four protein atoms. We therefore suggest that
the calcium binding site observed in this structure is
more crucial for the stabilization of the enzyme than
the calcium binding sites observed in PRK. This
assumption is supported by the observation that
SPRK loses about 40% of its activity when incubated
in the presence of 10 mm EDTA whereas PRK
remains essentially unaffected [18]. SPRK should have
the ability to bind one more calcium ion since it has
the same amino acid sequence as VPRK in the loop
comprising residues 58–62, but there is no obvious
structural explanation why this does not occur. The
two disulfide bridges in SPRK are found in the same
positions as VPRK, and they are completely different
from the positions found in PRK. One of the disul-
fide bridges (including Cys98) in SPRK and VPRK is
located next to a residue forming a part of the S2 site

of the enzyme. The disulfide bridge restricts the flexi-
bility of a tight loop which is part of the S2 site. The
tight loop is further stabilized by a tight hydrogen
bond pattern. This could confer more rigidity to the
active site of the enzyme and may explain the much
lower binding affinity (higher K
M
value) SPRK has
towards the synthetic suc-AAPF-nitroanilide substrate
than PRK has [18]. The volume of the S4 site in the
bacterial enzymes is larger than in PRK. It is there-
fore reasonable to believe that larger residues will be
favored at this site in SPRK and VPRK and more
kinetic data should therefore be obtained in order to
verify this assumption.
The electrostatic potential of SPRK is significantly
more negative than PRK. Since the N-terminal
protective succinyl group is negatively charged, one
cannot rule out that differences in activity towards
suc-AAPF-pNA when comparing SPRK and PRK
[18], at least partly, are due to electrostatic effects.
Methods and materials
Crystallization and data collection
Expression and purification of the peptidase from the Ser-
ratia sp. (SPRK) is described by Larsen and coworkers
[18]. Crystals of the catalytic domain of SPRK were
obtained from 0.1 m Bis ⁄ Tris and Tris buffer pH 6.5–7.5
and 45–47.5% ammonium sulfate at 4 °C by the hanging
drop vapor diffusion method. Protein concentration was
5mgÆmL

)1
in 25 mm Tris pH 7.5 and 10 mm CaCl
2
. X-ray
data to 1.8 A
˚
were collected at 100 K at Swiss Norwegian
Beamlines (BM01) at the European Synchrotron Radiation
Facility (ESRF) using reservoir solution containing 25%
glycerol as cryoprotectant. Data were integrated using
denzo [36], scaling and merging was carried out using
Scalepack [36] and scala of the CCP4 program suite [37].
Molecular replacement, model building and
refinement
The structure was solved using MolR ep [38] in CCP4i,
using proteinase K (PRK; PDB 1ic6 [20]), as search model.
The structure was built using a combination of auto build-
ing in ARP ⁄ wARP [39] and manual refitting of side chains
using O [40] based on 2FoFc and FoFc electron density
maps. Refinement was carried out in Refmac5 [41] of the
CCP4 suite. Solvent molecules were automatically added to
the structure using ARP ⁄ wARP. Solvent molecules that did
not fit into electron density, had no hydrogen binding
R. Helland et al. Structure of proteinase K from a Serratia sp.
FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS 69
partners or where the temperature factor exceeded 50 A
˚
2
were excluded from further refinement. One calcium and
two sulfate ions were added based on the electron density

and hydrogen binding partners.
Numbering of SPRK follows that of PRK, and inserted
residues get a letter extension after the number (i.e. 60B,
60C, etc.).
Structure analysis and comparison
Software for structure analysis and comparison was chosen
from the CCP4 suite and O. Numbering in the comparisons
refers to SPRK unless stated otherwise.
The alignment figure is generated using alscript [42].
Figures represented as cartoons are generated using mol-
script [43] and bobscript [44]. Electrostatic surfaces are
generated using ICM [28].
Acknowledgement
The authors thank the organizers at the Swiss-Nor-
wegian Beamlines (SNBL) at the European Synchrotron
Radiation Facilities (ESRF) for provision of beam time.
The present study was supported by the national Func-
tional Genomics Programme (FUGE) in The Research
Council of Norway and Biotec Pharmacon ASA.
References
1 Rao MB, Tanksale AM, Ghatge MS & Deshpande VV
(1998) Molecular and biotechnological aspects of micro-
bial proteases. Microbiol Mol Biol Rev 62, 597–635.
2 Rheinnecker M, Baker G, Eder J & Fersht AR (1993)
Engineering a novel specificity in subtilisin BPN¢.
Biochemistry 32, 1199–1203.
3 Rheinnecker M, Eder J, Pandey PS & Fersht AR (1994)
Variants of subtilisin BPN¢ with altered specificity pro-
files. Biochemistry 33, 221–225.
4 Wells JA, Powers DB, Bott RR, Graycar TP & Estell

DA (1987) Designing substrate specificity by protein
engineering of electrostatic interactions. Proc Natl Acad
Sci USA 84, 1219–1223.
5 Estell DA, Graycar TP, Miller JV, Powers DB, Burnier
JP, Ng PG & Wells JA (1986) Probing steric and hydro-
phobic effects on enzyme interactions by protein engi-
neering. Science 233, 659–663.
6 Cedrone F, Menez A & Quemeneur E (2000) Tailoring
new enzyme functions by rational redesign. Curr Opin
Struct Biol 10, 405–410.
7 Chen RD (2001) Enzyme engineering: rational redesign
versus directed evolution. Trends Biotechnol 19, 13–14.
8 Hult K & Berglund P (2003) Engineered enzymes for
improved organic synthesis. Curr Opin Biotechnol 14,
395–400.
9 Somero GN (1975) Temperature as a selective factor in
protein evolution – adaptational strategy of compro-
mise. J Exp Zool 194, 175–188.
10 Hochancha PW & Somero GN (1984) Temperature
Adaptation in Biochemical Adaptation, pp. 377–388. New
Jersey, USA, Princeton University Press.
11 Sweeney PJ & Walker JM (1993) Proteinase K. Methods
Mol Biol 16, 305–311.
12 Shaked GM, Shaked Y, Kariv-Inbal Z, Halimi M,
Avraham I & Gabizon R (2001) A protease-resistant
prion protein isoform is present in urine of animals and
humans affected with prion diseases. J Biol Chem 276,
31479–31482.
13 Jackson GS, McKintosh E, Flechsig E, Prodromidou
K, Hirsch P, Linehan J, Brandner S, Clarke AR, Weiss-

mann C & Collinge J (2005) An enzyme-detergent
method for effective prion decontamination of surgical
steel. J General Virol 86, 869–878.
14 Schechter I & Berger A (1967) On the size of the active
site in proteases. I. Papain. Biochem Biophys Res Com-
mun 27, 157–162.
15 Siezen RJ & Leunissen JA (1997) Subtilases: the super-
family of subtilisin-like serine proteases. Protein Sci 6,
501–523.
16 Gron H, Meldal M & Breddam K (1992) Extensive
comparison of the substrate preferences of two subtili-
sins as determined with peptide substrates which are
based on the principle of intramolecular quenching.
Biochemistry 31, 6011–6018.
17 Gron H & Breddam K (1992) Interdependency of the
binding subsites in subtilisin. Biochemistry 31, 8967–8971.
18 Larsen AN, Moe E, Helland R, Gjellesvik DR & Will-
assen NP (2006) Characterization of a recombinant
expressed proteinase K-like enzyme from a psychro-
trophic Serratia sp. FEBS J 273, 47–60.
19 Smalas AO, Leiros HK, Os V & Willassen NP (2000)
Cold adapted enzymes. Biotechnol Annu Rev 6, 1–57.
20 Betzel C, Gourinath S, Kumar P, Kaur P, Perbandt M,
Eschenburg S & Singh TP (2001) Structure of a serine
protease proteinase K from Tritirachium album limber
at 0.98 A resolution. Biochemistry 40, 3080–3088.
21 Arnorsdottir J, Kristjansson MM & Ficner R (2005)
Crystal structure of a subtilisin-like serine proteinase
from a psychrotrophic Vibrio species reveals structural
aspects of cold adaptation. FEBS J 272, 832–845.

22 Teplyakov AV, Kuranova IP, Harutyunyan EH,
Vainshtein BK, Frommel C, Hohne WE & Wilson KS
(1990) Crystal structure of thermitase at 1.4 A
˚
resolu-
tion. J Mol Biol 214, 261–279.
23 Wolf WM, Bajorath J, Muller A, Raghunathan S, Singh
TP, Hinrichs W & Saenger W (1991) Inhibition of pro-
teinase K by methoxysuccinyl-Ala-Ala-Pro-Ala-chloro-
methyl ketone: an x-ray study at 2.2-A
˚
resolution.
J Biol Chem 266 , 17695–17699.
Structure of proteinase K from a Serratia sp. R. Helland et al.
70 FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS
24 Feller G & Gerday C (1997) Psychrophilic enzymes:
molecular basis of cold adaptation. Cell Mol Life Sci
53, 830–841.
25 Feller G (2003) Molecular adaptations to cold in psy-
chrophilic enzymes. Cell Mol Life Sci 60, 648–662.
26 Kristjansson MM, Magnusson OT, Gudmundsson HM,
Alfredsson GA & Matsuzawa H (1999) Properties of a
subtilisin-like proteinase from a psychrotrophic Vibrio
species comparison with proteinase K and aqualysin I.
Eur J Biochem 260, 752–760.
27 Arnorsdottir J, Smaradottir RB, Magnusson OT,
Thorbjarnardottir SH, Eggertsson G & Kristjansson
MM (2002) Characterization of a cloned subtilisin-like
serine proteinase from a psychrotrophic Vibrio species.
Eur J Biochem 269, 5536–5546.

28 Abagyan R, Totrov M & Kuznetsov D (1994) icm –a
new method for protein modeling and design – applica-
tions to docking and structure prediction from the
distorted native conformation. J Comput Chem 15, 488–
506.
29 Radisky ES & Koshland DE Jr (2002) A clogged gutter
mechanism for protease inhibitors. Proc Natl Acad Sci
USA 99, 10316–10321.
30 Radisky ES, Kwan G, Karen Lu CJ & Koshland DE Jr
(2004) Binding, proteolytic, and crystallographic ana-
lyses of mutations at the protease–inhibitor interface of
the subtilisin BPN¢⁄chymotrypsin inhibitor 2 complex.
Biochemistry 43, 13648–13656.
31 McPhalen CA & James MN (1988) Structural compari-
son of two serine proteinase-protein inhibitor com-
plexes: eglin-c-subtilisin Carlsberg and CI-2-subtilisin
novo. Biochemistry 27, 6582–6598.
32 Heinz DW, Priestle JP, Rahuel J, Wilson KS & Grutter
MG (1991) Refined crystal structures of subtilisin novo
in complex with wild-type and two mutant eglins: com-
parison with other serine proteinase inhibitor complexes.
J Mol Biol 217, 353–371.
33 Bode W, Papamokos E & Musil D (1987) The high-
resolution X-ray crystal structure of the complex formed
between subtilisin Carlsberg and eglin c, an elastase
inhibitor from the leech Hirudo medicinalis: structural
analysis, subtilisin structure and interface geometry. Eur
J Biochem 166, 673–692.
34 Horn JR, Ramaswamy S & Murphy KP (2003) Struc-
ture and energetics of protein–protein interactions: the

role of conformational heterogeneity in OMTKY3 bind-
ing to serine proteases. J Mol Biol 331, 497–508.
35 Barrette-Ng IH, Ng KK, Cherney MM, Pearce G, Ryan
CA & James MN (2003) Structural basis of inhibition
revealed by a 1 : 2 complex of the two-headed tomato
inhibitor-II and subtilisin Carlsberg. J Biol Chem 278,
24062–24071.
36 Otwinowski Z & Minor W (1997) Processing of X-ray
diffraction data collected in oscillation mode. Methods
Enzymol 276, 307–326.
37 Collaborative Computational Project Number 4 (1994)
The CCP4 suite: programs for protein crystallography,
Acta Crystallogr Sect D Biol Crystallogr D50, 760–763.
38 Vagin A & Teplyakov A (1997) MOLREP: an auto-
mated program for molecular replacement. J Appl Crys-
tallogr 30, 1022–1025.
39 Perrakis A, Harkiolaki M, Wilson KS & Lamzin VS
(2001) ARP ⁄ wARP and molecular replacement. Acta
Crystallogr Sect D Biol Crystallogr 57, 1445–1450.
40 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991)
Improved methods for building protein models in elec-
tron density maps and the location of errors in these
models. Acta Crystallogr Sect A Found Crystallogr 47,
110–119.
41 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the maxi-
mum-likelihood method. Acta Crystallogr Sect D Biol
Crystallogr 53, 240–255.
42 Barton GJ (1993) ALSCRIPT: a tool to format multiple
sequence alignments. Protein Eng 6, 37–40.

43 Kraulis PJ (1991) MOLSCRIPT: a program to produce
both detailed and schematic plots of protein structures.
J Appl Crystallogr 24, 946–950.
44 Esnouf RM (1997) An extensively modified version of
MolScript that includes greatly enhanced coloring capa-
bilities. J Mol Graph Model 15, 112–113.
R. Helland et al. Structure of proteinase K from a Serratia sp.
FEBS Journal 273 (2006) 61–71 ª 2005 The Authors Journal compilation ª 2005 FEBS 71