The molecular surface of proteolytic enzymes has an important role
in stability of the enzymatic activity in extraordinary environments
Youhei Yamagata
1
, Hiroshi Maeda
1
, Tasuku Nakajima
1
and Eiji Ichishima
2
1
Laboratory of Molecular Enzymology, Division of Life Science, Graduate School of Agricultural Science, Tohoku University,
Aoba-ku, Sendai, Japan;
2
Department of Biotechnology, Faculty of Engineering, Soka University, Hachioji, Tokyo, Japan
It is scientifically and industrially important
1
to clarify the
stabilizing mechanism of proteases in extraordinary envi-
ronments. We used subtilisins ALP I and Sendai as models
to study the mechanism. Subtilisin ALP I is extremely
sensitive to highly alkaline conditions, even though the
enzyme is produced by alkalophilic Bacillus, whereas sub-
tilisin Sendai from alkalophilic Bacillus is stable under
conditions of high alkalinity. We constructed mutant
subtilisin ALP I enzymes by mutating the amino acid
residues specific for subtilisin ALP I to the residues at the
corresponding positions of amino acid sequence alignment
of alkaline subtilisin Sendai. We observed that the two
mutations in the C-terminal region were most effective for
improving stability against surfactants and heat as well as

alkalophilic Bacillus sp. G-825-6, categorized as an
alkaline subtilisin, is very stable under highly alkaline
conditions.
Maeda et al. reported that the inactivation of subtilisin
ALP I at high alkalinity was caused by the instability of
its molecular surface structure and autolysis in the
N-terminal region and/or the C-terminal region [12,13].
We hypothesized that the divergence of the properties of
ALP I from the alkaline subtilisins might depend on the
structure of the enzyme. In particular, the instability of
ALP I in highly alkaline conditions might be caused by
the existence of consensus amino acid sequences of
ALP I and the neutral subtilisins and/or the peculiar
residues in the amino acid sequence of ALP I. We
selected 12 consensus amino acid residues from the
amino acid sequence alignment of ALP I and the neutral
subtilisins. These candidate residues did not occur at the
corresponding positions of the alkaline subtilisins. Fur-
thermore, on the basis of the predicted three-dimensional
structure of ALP I, we believed that the C-terminal
region was located on the molecular surface and was
exposed to the solvent phase; therefore two unique
residues in the C-terminal region were replaced by the
residues at corresponding positions of amino acid
sequence of Sendai. As a result of analysing the mutant
ALP I s, two amino acid residues in the C-terminal
region were found to play important roles in maintaining
stability in highly alkaline conditions. The double muta-
tions prolonged the half-lifetime by more than 120-fold.
The substitutions of the amino acid residues also

mk
+
), phoA,
supE44, k-, thi-1, gyrA96, relA1] was used for cloning with
M13 derivatives mp18 and mp19. E. coli MV1184 [ara,
D(lac-proAB), rpsL, thi (F80 lacZDM15) D(srl-recA)306::
Tn10 (tet
r
)/F¢ (tra36, proAB
+
, lacI
q
, lacZDM15) [14], and
BMH71-18 mutS [D(lac-proAB), supE, thi, mutS215:: Tn10
(tet
r
)/F¢ (tra36, proAB
+
, lacI
q
, lacZDM15] was used for site-
directed mutagenesis. B. subtilis KN2 (phe-I, lys-I, nprR2,
nprE18, aprE3, ispA) [15] was used for protein expression.
Plasmids pUC119 and pUC118 [14] were used as the vectors
for construction of the mutant enzymes and for site-directed
mutagenesis. Plasmids pALP3 [3], pALP1 [11] and pTnat3
[7] were the recombinant plasmids containing intact ALP I
gene (aprQ), Sendai gene (aprS)andNATgene(aprN),
respectively. Plasmid pUB110 [16] was used for transfor-
mation of B. subtilis.

BPN' 1 AQSVPYGVSQIKAPALHSQGYTGSNVKVAVIDSGIDSSHPDLKVAGGASMVPSETNPFQDNNSHGTHVAG
MECE 1 AQSVPYGISQIKAPALHSQGYTQSNVKVAVIDSGIDSSHTDLQVRGGASFVPSETNPYQPGSSHGTHVAG
80 90 100 110 120 130 140
PB92 69 TIAALNNSIGVLGVAPNAELYAVKVLGASGSGSVSSIAQGLEWAGNNGMHVANLSLGSPSPSATLEQAVN
221 69 TIAALNNSIGVLGVAPSAELYAVKVLGASGSGSVSSIAQGLEWAGNNGMHVANLSLGSPSPSATLEQAVN
SAVI 68 TIAALNNSIGVLGVAPSAELYAVKVLGASGSGSVSSIAQGLEWAGNNGMHVANLSLGSPSPSATLEQAVN
SEND 69 TIAALNNSIGVVGVAPNAELYAVKVLGANGSGSVSSIAQGLQWTAQNNIHVANLSLGSPVGSQTLELAVN
YAB 68 TIAALNNSIGVLGVAPNVDLYGVKVLGASGSGSISGIAQGLQWAANNGMHIANMSLGSSAGSATMEQAVN
* *** * ***** * ** *** *** * * * * ***
ALP1 68 TVAALNNSYGVLGVAPGAELYAVKVLDRNGSGSHASIAQGIEWAMNNGMDIANMSLGSPSGSTTLQLAAD
* *** * ***** * ** *** *** * * * * ***
CARL 70 TVAALDNTTGVLGVAPSVSLYAVKVLNSSGSGTYSGIVSGIEWATTNGMDVINMSLGGPSGSTAMKQAVD
DY 70 TVAALDNTTGVLGVAPNVSLYAIKVLNSSGSGTYSAIVSGIEWATQNGLDVINMSLGGPSGSTALKQAVD
NAT 71 TIAALNNSIGVLGVAPSASLYAVKVLDSTGSGQYSWIINGIEWAISNNMDVINMSLGGPTGSTALKTVVD
E 71 TIAALNNSIGVLGVSPSASLYAVKVLDSTGSGQYSWIINGIEWAISNNMDVINMSLGGPTGSTALKTVVD
J 71 TIAALNNSIGVLGVSPSASLYAVKVLDSTGSGQYSWIINGIEWAISNNMDVINMSLGGPSGSTALKTVVD
AMYL 71 TIAALNNSIGVLGVSPSASLYAVKVLDSTGSGQYSWIINGIEWAISNNMDVINMSLGGPSGSTALKTVVD
BPN' 71 TVAALNNSIGVLGVAPSASLYAVKVLGADGSGQYSWIINGIEWAIANNMDVINMSLGGPSGAAALKAAVD
MECE 71 TIAALNNSIGVLGVAPSSALYAVKVLDSTGSGQYSWIINGIEWAISNNMDVINMSLGGPTGSTALKTVVD
150 160 170 180 190 200 210
PB92 139 SATSRGVLVVAASGNSGA-GSIS YPARYANAMAVGATDQNNNRASFSQYGAGLDIVAPGVNVQSTYP
221 139 SATSRGVLVVAASGNSGA-GSIS YPARYANAMAVGATDQNNNRASFSQYGAGLDIVAPGVNVQSTYP
SAVI 138 SATSRGVLVVAASGNSGA-GSIS YPARYANAMAVGATDQNNNRASFSQYGAGLDIVAPGVNVQSTYP
SEND 139 QATNAGVLVVAATGNNGS-G TVSYPARYANALAVGATDQNNNRASFSQYGTGLNIVAPGVGIQSTYP
YAB 138 QATASGVLVVAASGNSGA-G N-VGFPARYANAMAVGATDQNNNRATFSQYGAGLDIVAPGVGVQSTVP
* * * ** * * * * **** ** ** * * *** **
ALP1 138 RARNAGVLLIGAAGNSGQQGGSNNMGYPARYASVMAVGAVDQNGNRANFSSYGSELEIMAPGVNINSTYL
* * * ** * * * * **** ** ** * * *** **
CARL 140 NAYARGVVVVAAAGNSGSSGNTNTIGYPAKYDSVIAVGAVDSNSNRASFSSVGAELEVMAPGAGVYSTYP
DY 140 KAYASGIVVVAAAGNSGSSGSQNTIGYPAKYDSVIAVGAVDSNKNRASFSSVGAELEVMAPGVSVYSTYP

of ALP I; *, consensus sequence in the sub-
tilisins; n, catalytic triad. ALP I, subtilisin
ALP I (accession number; BAA06158); PB92,
serine protease from B. alcalophilus PB92
(A49778); 221, no. 221 protease from alkalo-
philic Bacillus sp. no. 221 (S27501); SAVI,
Savinase
TM
(P29600); SEND, subtilisin Sen-
dai (BAA06157), YAB, alkaline esterase Ya-B
(P20724); CARL, subtilisin Carlsberg
(P00780); DY, subtilisin DY (P00781): NAT,
subtilisin NAT (JH0778); E, subtilisin E
(P04189); J, subtilisin J (P29142); AMYL,
subtilisin amylosacchariticus (P00783); BPN¢,
subtilisin BPN¢ (P00782); MECE, mecente-
ricopeptidase (P07518).
4578 Y. Yamagata et al. (Eur. J. Biochem. 269) Ó FEBS 2002
coding region of aprQ was designated pNALP3. Plasmid
pNALP3 digested with EcoRI was ligated with pUB110
digested with EcoRI. The shuttle vector carrying aprQ was
designated pNALP3B. It was used in the protoplast
transformation of B. subtilis KN2 [18]. Plasmid pALP1,
carrying the Sendai gene, aprS [11], was digested with EcoRI
and ligated with pUB110. The constructed plasmid was
named pSen6B. It was also introduced into B. subtilis KN2.
Construction of mutant enzymes
Oligonucleotides for introducing the mutation to the
enzymes are shown in Table 1. The oligonucleotides were
used to replace the amino acid residues of ALP I with the

active fraction was dialysed with 10 m
M
Mes at pH 6.5
containing 2 m
M
CaCl
2
. The enzyme solution was loaded to
an FPLC-Hitrap SP (Amersham Pharmacia Biotech)
equilibrated with the same buffer. The enzyme active
fraction was eluted with a 0–0.5
M
NaCl linear gradient.
The purified enzymes were monitored by SDS/PAGE [19]
and immunoblot analysis [20]. Purified enzyme was also
blotted onto poly(vinylidene difluoride) (PVDF) membrane
[21]. Amino-terminal amino acid sequence analysis of each
enzyme blotted onto PVDF membrane was performed with
an ABI protein sequencer Model 491 (Applied Biosystems).
Assay of enzymatic activities
Protease activities towards milk casein were examined as
described in an according to a previous report [22]. Fluoro-
metric assays were conducted as described previously [23].
Protein was measured by Lowry’s method using BSA
fraction V (Seikagaku ko-gyo, Tokyo, Japan) as the
standard. The alkaline stability was measured at 30 °C
and pH 10.0 using succinyl-
L
-alanyl-
L

NAT (NAT) from B. subtilis (natto) was used for expres-
sion. The open reading frame of the ALP I gene, aprQ,was
ligated of the downstream of the promoter region of the
pALP1
Amp
r
aprQ
PA
Pst
I
Eco
RI
pTnat3
Amp
r
aprN
PN
Pst
I
Nde
I
pALP1+Nd
Amp
r
aprQ
PA
Pst
I
Eco
RI

r
Neo
r
Eco
RI
Eco
RI
Nde
I
Pst
I
Introduction of a new
Nde
I site
Nde
I and
Pst
I
Eco
RI
Eco
RI
Nde
I and
Pst
I
Introduction of a new
Nde
I site
Fig. 2. Construction of the expression plasmid for ALP I. Thick arrows

were consistent with those of the wild-type enzymes
(Table 2).
Stability under alkaline conditions
ALP I lost its enzymatic activity after only a few minutes’
incubation in 0.1
M
Na
2
HPO
4
/NaOH buffer pH 12
(Fig. 3A). After 2 min, ALP I showed only 27% of the
original activity, and after 10 min the enzyme showed
just 1% of its original activity. On the other hand, Sendai
was stable under these conditions and held 63% of the
original activity after 6 h at pH 12 (Fig. 3B). Two mutant
enzymes, D266N/Y269A- and D266N/Y269A/A271T/
Q272R-ALP I, were most stable retaining 60% of the
original activity after 1 h, and 30% after 6 h.
2
The D266N-
ALP I showed 40 and 20% of the original activity after
1 h and 6 h of incubation, respectively. The stability of
Y269A/A271T/Q272R- and Y269A-ALP I in alkaline
Table 2. Specific activities of the mutant subtilisins.
Enzyme
Specific activity (katÆkg
)1
)
a

I108L 5¢-ATTCATCGCCCAcTCgAgTCCTTGAGCAAT-3¢ XhoI
D117H 5¢-GTTGGCAATATgCATCCCATTATT-3¢ EcoT221
D137N 5¢-CTAGCGCGGTtTGCTGCcAgcTGCAGGGTTGT-3¢ PvuII
A150T 5¢-TTGTCCTGAGTTaCCgGtCGCCCCAATTAA-3¢ AgeI
S170N 5¢-TCCAACAGCCATaACgttTGCATAGCGCHC-3¢ AclI
V177T 5¢-TCCATTTTGGTC CgtCGCTCCAACAGC-3¢
E192G
5¢-AATCTCAAGTcCgGATCCATAGCT-3¢ BamHI
M196V 5¢-TAATATTGACcCCgGGCGCCAcAATCTCAAG-3¢ SmaI
Y259Q 5¢-GCCATTTCCATAtTgaGTaCTGTTACCAAG-3¢ ScaI
D266N 5¢-CATACTCAGCgTtaACTAAGCCATTTC-3¢ HpaI
Y269A 5¢-TTGAGCCGCAgcCTCAGCgTCgAC TAAGCCATTTC-3¢ SalI
Q272R 5¢-CTTAGGGATTAacGAGCCGCATACTCAGCgTCgAC TAAGCCATTTC-3¢ HpaI
D266N/Y269A 5¢-ATTGAGCCGCAgcCTCAGCgTtaACTAAGCCATTTC-3¢ HpaI
A271T/Q272R 5¢-CTTAGGGATTAacGcGtCGCATACTCAG-3¢ MluI
Y269A/A271T/ Q272R 5¢-CTTAGGGATTAacGcGtCGCAgcCTCAGCATCC-3¢ MluI
D266N/Y269A/ 5¢-CTTAGGGATTAacGcGtCGCAgcCTCAGCATtCACTAAGCCA-3¢ MluI
A271T/Q272R s-N263D 5¢-TGCAGCTTCTGCgTcgACAAGTCCACTGCC-3¢ SalI
s-N263D/A266Y 5¢-AATATAAGCTTAaCgcGTTGCAtaTTCTGCgTcgAcAAGTCCACTGCC-3 ¢ MluI/SalI
4580 Y. Yamagata et al. (Eur. J. Biochem. 269) Ó FEBS 2002
conditions was also improved. We did not observe
improved stability under alkaline conditions in the other
mutant enzymes. They lost the activity within 10 min as did
wild-type ALP I.
Resistance to surfactants
Residual activities of the mutant enzymes were measured
after incubation with 0.1% SDS in 0.1
M
H
3

after incubation for 10 min at pH 10.0 and at a variety of
temperatures (Fig. 5). The substitution of Asp266Asn and
Asp266Asn/Tyr269Ala improved the thermostability by
 10 °C, and Tyr269Ala substitution improved the thermo-
stability by  5 °C.
Protein denaturation by thermal treatment
Our investigation of enzymatic stability against alkalinity
and surfactants showed that the Asp266Asn and Tyr269Ala
6543210
0
20
40
60
80
100
120
Time (h
)
Time (h
)
A
6543210
0
20
40
60
80
100
120
B

mutations were most effective (Fig. 6). Thermostability (T
m
:
mid point in the thermally induced transition from the
folded to the unfolded state) of wild-type ALP I, Sendai and
D266N/Y269A-ALP I were estimated by differential scan-
ning calorimetry (DSC). The T
m
of D266N/Y269A-ALP I
was 74.4 °C. It was higher than that of the wild-type ALP I,
70.2 °C, and almost the same as that of wild-type Sendai,
73.6 °C.
Stability of mutant Sendai in alkaline conditions
The substitutions of Asp266Asn and Tyr269Ala were most
effective in improving the stability of ALP I. To estimate the
effects of the corresponding amino acid residues in Sendai,
N266D- and N266D/Y269A-Sendai were constructed by
using primers s-N263D and s-N263D/A266Y (Table 1).
Compared with the stability of wild-type Sendai, both
mutant enzymes, N266D- and N266D/Y269A-Sendai
showed decreased stability under alkaline conditions
(Fig. 3B). Wild-type Sendai was stable at pH 12 and it
maintained 80% of the original activity after 6 h. The
activity of N263D-Sendai decreased to 45% and 10% of the
original after 2 and 6 h of incubation at pH 12, respectively.
N263D/A266Y-Sendai showed only 10% of the original
activity after 2 h, and little enzymatic activity was observed
after 4 h.
DISCUSSION
Based on our previous results we hypothesized that the

in ALP I restrain the conformational changes of the
molecular surface responsible for degradation.
The substitution of Asp266Asn, Tyr269Ala and
Asp266Asn/Tyr269Ala were also effective in increasing
resistance of ALP I to anionic surfactants. The unfolding
caused by surfactants occurred in a moderate manner in
comparison with denaturation by high alkalinity, and the
structural change of the molecular surface proceeded slowly.
In conditions of high alkalinity, a hydroxyl ion probably
Fig. 6. Thermostability of the enzyme structure. The denaturing
temperatures of the enzymes (3.3 nmol) were measured by DSC. The
arrowheads indicate the midpoints of the thermally induced phase
transitions.
7060504030
0
20
40
60
80
100
120
Tem p. ( C
)
Fig. 5. Thermostability of the mutant-ALP Is. The enzymes
(0.1 mgÆmL
)1
) were incubated for 10 min at 30, 40, 50, 55, 57, 60, 63,
65, 70 °C and pH 10.0, and then the residual activities were measured
with Suc-Ala-Ala-Pro-Phe-MCA as a substrate at 30 °C and pH 10.0.
s, wild-type ALP I; d, D266N-ALP I; n, Y269A-ALP I; m, D266N/

increments between the inactive temperature and the T
m
should indicate that the mutations improved the stability of
surface region. ALP I was not denatured at 55 °C, but the
enzymatic activity was lost. This indicates degradation of
the ALP I molecular begins as soon as the conformational
change occurs on the surface region. As improvement of
structural stability at the molecule surface would repress
autolysis, the inactivation temperature increases. However,
the effect of the mutation should not extent the whole
protein and so the T
m
did not increase likewise.
Stability of N263D- and N263D/A266Y-Sendai were
observed in alkaline conditions. The mutated residues of
Asn263 and Ala266 in Sendai correspond to Asp266 and
Ala269 in ALP I, respectively. The mutant Sendai became
sensitive to high alkalinity. The double-mutated Sendai was
additively more sensitive to alkalinity than N263D-Sendai.
The mutations at these positions in Sendai should promote
instability of the surface region. As the C region of Sendai
also would play an important role in restraining the
conformational change of the surface regions, wild-type
Sendai could be resistant to highly alkaline conditions.
The putative three-dimensional models of the enzymes
were constructed to clarify the location of substituted
residues and their interactions with surrounding residues.
The effective mutation sites of Asp266 and Tyr269 in the
C-terminal region were located on the back surface of a
catalytic triad, and it was understandable that the substi-

a scenario in which the first cleavage site of ALP I occurs at
Glu18–Gly19 in the N-terminal region, and the next is
located in the C-terminal region [13]. The mutation
Asp117His contributed to the resistance to surfactants.
Aspartic acid at position 117 was located in the bottom of
the depression on the surface of ALP I, and it was adjacent
to Lys26, which was the last residue of N-terminal region on
the molecular surface. As a result of substitution of Asp117
to His, the side chain is larger. The mutation seemed to fill in
the gap between surrounding residues of the depression, and
the side chain of the mutated residue might restrain the
mobility of the N-terminal region on the surface by
interaction with the main chain of Lys26 by van der Waals’
forces (data not shown).
Altering core packing, helix stabilization, introduction of
surface salt bridges and reduction of flexibility in surface
loops are proposed mechanisms for the thermostability of
proteins [25–29]. The stability of ALP I under alkaline
conditions was caused by the stabilization of the surface
structure. Similar results are obtained from the structural
studies of shuffled p-nitrobenzyl esterases with improved
solvent stability and thermostability. The enzyme obtains a
17 °C increase in thermostability with 13 amino acid
residues replacements out of 484 residues with the eight
times reiterative random mutations [29]. Some of the
mutations decrease the conformational freedom. The
mutations fix disordered loops of esterase.
We selected the amino acid residues to mutate on the
basis of the predicted three-dimensional protein structure
and the alignment of amino acid sequences of the subtilisins.

Enzymology 19 (Perlmann, G.E. & Lorand, L., eds), pp. 199–215.
Academic press, New York, London.
2. Bryan, P.N. (2000) Protein engineering of subtilisin. Biochim.
Biophys. Acta 1543, 203–222.
3. Yamagata, Y., Sato, T., Hanzawa, S. & Ichishima, E. (1995) The
structure of subtilisin ALP I from alkalophilic Bacillus sp. NKS-
21. Curr. Microbiol. 30, 201–209.
4. Wells, J.A., Ferrari, E., Henner, D.J., Estell, D.A. & Chen, E.Y.
(1983) Cloning, sequencing, and secretion of Bacillus amylolique-
faciens subtilisin in Bacillus subtilis. Nucleic Acids Res. 11, 7911–
7925.
5. Jacobs, M., Eliasson, M., Uhlen, M. & Flock, J.I. (1985) Cloning,
sequencing and expression of subtilisin Carlsberg from Bacillus
licheniformis. Nucleic Acids Res. 13, 8913–8926.
6. Stahl, M.L. & Ferrari, E. (1984) Replacement of the Bacillus
subtilis subtilisin structural gene with an in vitro-derived deletion
mutation. J. Bacteriol. 158, 411–418.
7. Nakamura, T., Yamagata, Y. & Ichishima, E. (1992) Nucleotide
sequence of the subtilisin NAT gene, aprN,ofBacillus subtilis
(natto). Biosci. Biotechnol. Biochem. 56, 1869–1871.
8. Kaneko, R., Koyama, N., Tsai, Y.C., Juang, R.Y., Yoda, K. &
Yamasaki, M. (1989) Molecular cloning of the structural gene for
alkaline elastase YaB, a new subtilisin produced by an alkalophilic
Bacillus strain. J. Bacteriol. 171, 5232–5236.
9. Takami, H., Kobayashi, T., Kobayashi, M., Yamamoto, M.,
Nakamura, S., Aono, R. & Horikoshi, K. (1992) Molecular
cloning, nucleotide sequence, and expression of the structural gene
for alkaline serine protease from alkaliphilic Bacillus sp. 221.
Biosci. Biotechnol. Biochem. 56, 1455–1460.
10. Betzel, C., Klupsch, S., Papendorf, G., Hastrup, S., Branner, S. &

17. Carter,P.,Bedouelle,H.&Winter,G.(1985)Improvedoligo-
nucleotide site-directed mutagenesis using M13 vectors. Nucleic
Acids Res. 13, 4431–4443.
18. Chang, S. & Cohen, S.N. (1979) High frequency transformation of
Bacillus subtilis protoplasts by plasmid DNA. Mol. Gen. Genet
168, 111–115.
19. Laemmli, U.K. (1970) Cleavage of structural proteins during the
assembly of the head of bacteriophage T4. Nature 227, 680–685.
20. Burnette, W.N. (1981) ÔWestern blottingÕ: electrophoretic transfer
of proteins from sodium dodecyl sulfate–polyacrylamide gels to
unmodified nitrocellulose and radiographic detection with
antibody and radioiodinated protein A. Anal Biochem. 112,
195–203.
21. LeGendra, N. & Matsudaira, P. (1989) Purification of proteins
and peptides by SDS-PAGE. In A Practical Guide to Protein and
Peptide Purification for Microsequencing (Matsudaira, T.P., ed.),
pp. 49–69. Academic Press, San Diego, New York, Berkley,
Boston, London, Sydney, Tokyo, Toronto.
22. Tsuchida, O., Yamagata, Y., Ishizuka, T., Arai, T., Yamada, J.,
Takeuchi, M. & Ichishima, E. (1986) An alkaline proteinase of an
alkalophilic Bacillus sp. Curr. Microbiol. 14, 7–12.
23. Yamagata, Y. & Ichishima, E. (1989) A new alkaline proteinase
with pI 2.8 from alkalophilic Bacillus sp. Curr. Microbiol. 19, 259–
264.
24. Takagi, H., Takahashi, T., Momose, H., Inouye, M., Maeda, Y.,
Matsuzawa, H. & Ohta, T. (1990) Enhancement of the thermo-
stability of subtilisin E by introduction of a disulfide bond
engineered on the basis of structural comparison with a thermo-
philic serine protease. J. Biol. Chem. 265, 6874–6878.
25. Vogt, G., Woell, S. & Argos, P. (1997) Protein thermal stability,