Báo cáo khoa học: The role of Tyr71 in Streptomyces trypsin on the recognition mechanism of structural protein substrates ppt

The specificity at this site in the chymotrypsin-like serine proteases has been explained using the structure of the S1 pocket, which comprises three b-sheets residues 189–192, 214–216 an

recognition mechanism of structural protein substrates Yoshiko Uesugi*, Hirokazu Usuki, Masaki Iwabuchi and Tadashi Hatanaka

Research Institute for Biological Sciences, Okayama, Japan

Introduction

Serine proteases play key roles in physiological and

cellular functions, including protein processing, tissue

remodelling, immunity, cell differentiation and blood

clotting [1] Serine proteases of clans SA

(chymotryp-sin-like) [2], SB (subtili(chymotryp-sin-like) [3] and SC (a⁄

b-hydro-lase fold) [4] maintain a strictly conserved catalytic

site geometry comprising serine, histidine and aspartic

acid residues They catalyse peptide bond hydrolysis,

which generally proceeds in a three-step mechanism:

the formation of an enzyme–substrate complex; acyla-tion of the active site serine; and hydrolysis of the acyl-enzyme intermediate [5]

Substrate recognition, especially the specificity at the S1 site, has been studied extensively The specificity at this site in the chymotrypsin-like serine proteases has been explained using the structure of the S1 pocket, which comprises three b-sheets (residues 189–192, 214–216 and 224–228) and the oxyanion-binding site

Keywords

collagenolytic enzyme; repeat-length

independent and broad spectrum (RIBS)

in vivo DNA shuffling; serine protease;

Streptomyces; topological specificity

Correspondence

T Hatanaka, Research Institute for

Biological Sciences, Okayama, 7549-1

Kibichuo-cho, Kaga-gun, Okayama 716-1241,

Japan

Fax: +81 866 56 9454

Tel: +81 866 56 9452

E-mail: hatanaka@bio-ribs.com

*Present address

Department of Biomaterials, Field of Tissue

Engineering, Institute for Frontier Medical

Sciences, Kyoto University, Japan

(Received 7 April 2009, revised 26 July

2009, accepted 4 August 2009)

doi:10.1111/j.1742-4658.2009.07256.x

Studies of substrate recognition by serine proteases have focused on speci-ficities at the primary S1–Sn sites, but topological specispeci-ficities (i.e recogni-tion at distinct three-dimensional structural motifs) have not been established This is the first report to identify the key amino acid residue conferring topological specificity A serine protease from Streptomyces omi-yaensis (SOT), which is a trypsin-like enzyme, was chosen as a model enzyme to clarify the recognition mechanism of structural protein sub-strates in serine proteases We have found previously that the topological specificities of SOT and S griseus trypsin (SGT) for high molecular mass substrates differ greatly, even though the enzymes have similar primary structures In this study, we constructed chimeras between SOT and SGT using an in vivo DNA shuffling system and several mutants to identify the key residues involved in topological specificities By comparing the sub-strate specificities of chimeras and mutants, we found that residue 71 of SOT, which is separate from the catalytic triad, contributes to the topologi-cal specificity Using site-directed mutagenesis, residue 71 of SOT was also found to be crucial for catalytic efficiency and enzyme conformation

Abbreviations

ADAMTS, a disintegrin and metalloproteinase with thrombospondin motifs; ANS, 1-anilinonaphthalene-8-sulfonic acid; CBD, collagen-binding domain; FITC, fluorescein isothiocyanate; LB, Luria–Bertani; MMP, mammalian matrix metalloprotease; RIBS, repeat-length independent and broad spectrum; SGT, Streptomyces griseus trypsin; SOT, Streptomyces omiyaensis serine protease; Z-Gly-Pro-Arg-MCA,

benzyloxycarbonylglycyl- L -prolyl- L -arginine-4-methylcoumaryl-7-amide.

(Gly193 and Ser195) in the C-terminal b-barrel

domain Specificity is usually determined by the

resi-dues at positions 189, 216 and 226 [6,7] For example,

the combination of Ser189, Gly216 and Gly226 creates

a deep hydrophobic pocket in the chymotrypsin

enzyme that accounts for S1 specificity Furthermore,

Asp189, Gly216 and Gly226 create a negatively

charged S1 site that confers specificity of trypsin for

substrates containing arginine or lysine at the P1

posi-tion [8,9] Surface loops 1, 2 and 3 (residues 184–195,

213–228 and 169–175, respectively) are also important

for substrate specificity The specificities of S2–Sn sites

have also been investigated using elastase [10]

How-ever, the mechanism by which serine proteases

recog-nize the structure of protein substrates is not known

Various structural features govern interactions between

protease and substrate, and therefore insight into the

mechanism is necessary to explain substrate

recogni-tion Data on the topological specificities are available

only for the metalloproteinase ADAMTS (a disintegrin

and metalloproteinase with thrombospondin motifs)

and the mammalian matrix metalloproteases (MMP),

which were obtained using triple-helical and

single-stranded fluorogenic substrates [11]

Recently, serine protease-catalysed degradation of

recalcitrant animal proteins, such as collagen [12,13],

keratin [14], those involved in blood clotting [15,16]

and amyloid prion proteins [17], has been of interest

because of potential industrial waste and medical

applications We therefore selected collagens as

‘hard-to-degrade’ substrates for our study of how serine

pro-teases recognize protein substrate structure At least 25

different types of collagen have been identified [18]

For example, collagen types I, II, III, V and IX are

classical fibrillar collagens, whereas collagen type IV

forms sheet-like networks [19] Collagen degradation

products have biological activities of industrial and

medical interest In addition, from the screening of

2000 soil isolates, we obtained a serine protease from

Streptomyces omiyaensis(SOT) with high

collagenolyt-ic activity [12] SOT belongs to the trypsin family and

the peptidase family S1 (subfamily S1A) The primary

structure of SOT is 77% identical to that of the trypsin

from Streptomyces griseus (SGT, EC 3.4.21.4) [20–22],

but the topological specificity of the former differs

sub-stantially from that of the latter Therefore, we used

SOT as a model for enzymes that hydrolyse

hard-to-degrade proteins in order to clarify how serine

proteases recognize structural protein substrates

For the identification of the amino acid residues

conferring topological specificity of our model serine

protease, we first constructed chimeras of SOT with

SGT using repeat-length independent and broad

spec-trum (RIBS) in vivo DNA shuffling [23], and com-pared their substrate specificities Using type I and type IV collagens as typical protein substrates with different structures, we identified a key residue on the substrate recognition site that conferred specificity for the substrates

Results and Discussion

Construction of chimeras using RIBS in vivo DNA shuffling

In a previous study, we demonstrated that SOT had wide substrate specificity for types I and IV collagens, gelatin and casein, whereas SGT only showed high substrate specificity towards type I collagen [12] To investigate which domain confers the different topolog-ical specificities, we constructed a chimeric gene library between SOT and SGT using RIBS in vivo DNA shuf-fling (Fig 1A) This system is a method of random chimeragenesis based on the combination of highly fre-quent deletion formation in the Escherichia coli ssb-3 strain with an rpsL-based chimera selection system

We have demonstrated previously the substrate recog-nition mechanism in Streptomyces phospholipase D using this system [23–25]

We obtained various chimeras with recombination sites widely distributed over the entire chimeric gene The DNA sequences of the parental sot gene and the trypsin gene (sprT) encoding SGT (in S griseus NBRC 13350) [21] are 82% identical Therefore, these genes are suitable for chimeragenesis using the RIBS in vivo DNA shuffling system We chose eight typical chime-ras, as presented in Fig 1B, for gene expression and protein purification The purified chimeras appeared as

a single band on SDS-PAGE, and the molecular mass (approximately 23 kDa) was similar to that of SOT and SGT (data not shown)

Comparison of substrate specificities for chimeras towards protein substrates The substrate specificities of eight chimeras were evalu-ated using fluorogenic bovine skin collagen type I and human placenta collagen type IV as substrates, and these were compared with those of parental SOT and SGT (Fig 2) In a previous study, we showed that the hydrolytic activities of SOT and SGT towards fluores-cein-labelled collagens agreed well with those towards native collagens [12] Figure 2A, B shows that the spe-cific activities of chimeras A and B towards type I and

IV collagens resemble those of SGT Chimeras A and

B show much lower activities towards type IV collagen

than do other chimeras or SOT Interestingly, the

hydrolytic activity of chimera F towards both types of

collagen was the highest among the chimeras and

higher than that of parental SOT and SGT These

spe-cific activities were used to calculate the ratio of the

hydrolytic activity towards type IV collagen relative to

that towards type I collagen (collagen IV⁄ I) (Fig 2C)

The ratios of chimeras C–H were comparable with that

of SOT They were, however, significantly different

from those of chimeras A, B and SGT These results

suggest that the region between chimeras B and C

(corresponding to residues 52–72 of SOT) confers

sub-strate specificity Therefore, we examined this region

further

Identification of amino acid residue(s) related to topological specificity

Chimera B differed from chimera C in five amino acid residues (Fig 3A) Therefore, we constructed four chimera B mutants (B-1 to B-4, the primary sequence of which is presented in Fig 3A) and evaluated their speci-ficities towards two collagen types For type I collagen, the specific activity of the chimera B mutant increased

as substituted residues accumulated (Fig 3B) In con-trast, the specific activity of the chimera B mutant towards type IV collagen changed considerably between B-3 and B-4 (Fig 3C) These results were reflected in collagen IV⁄ I (Fig 3D) Figure 3A shows that chimeras

A

P T7-lac

pACTI2b

(sot/GmrrpsL + /sprT)

lacl q

Cm r

E coli MK1019 ssb-3

Transformation

Gm rrpsL+

P T7-lac

P T7-lac

Cm r

Cm r

Gm r

lacl q

lacl q

sot sprT rpsL+

SGT 223 aa

in vivo DNA shuffling

Chimeric genes SOT

Chimera A

Chimera B

Chimera C

Chimera D

Chimera E

Chimera F

Chimera G

Chimera H

223 aa

223 aa

E

E

H

Fig 1 Random chimeragenesis between sot and sprT genes by RIBS in vivo DNA shuffling The random chimeragenesis strat-egy (A) is detailed in ref 23 For construc-tion of the shuffling vector, two

homologous genes (sot and sprT) were placed in the same direction, and a cassette containing the Gm r and E coli rpsL + genes was inserted between them rpsL + encodes the ribosomal protein S12 [23], the target of

Sm The transformation of E coli MK1019 [ssb-3 rpsL(Sm r )] with pACTI2b (sot ⁄ Gm r -rpsL+⁄ sprT) altered the phenotype of cells from Sm r to Sm s (and also Gm s to Gm r ) because the Sm s ribosome was reconsti-tuted with the wild-type RpsL protein encoded by the plasmid The Gm r -rpsL + cassette is simultaneously deleted from the plasmid and the cells reverse their pheno-type from Sm s ⁄ Gm r to Sm r ⁄ Gm s when recombination occurs between two homolo-gous genes Consequently, the intact form

of the chimeric gene is selectable by Sm and Cm without expression The primary structures of SGT, SOT and the eight chime-ras used in this study are illustrated sche-matically The recombination sites of chimeras A–H are shown in (B) The primary structures of SOT (upper) and SGT (lower) are shown The recombination sites of the chimeras are boxed The names of the chi-meras are indicated in bold capital letters The catalytic residues are indicated in red.

B-3 and B-4 differed in only one amino acid residue:

res-idue 71 From these results, we inferred that Tyr71 of

SOT (corresponding to residue 89 of a-chymotrypsin

numbering) is a key amino acid residue conferring

sub-strate specificity To visualize the location of Tyr71 in SOT, an overall structure of SOT was constructed based

on the crystal structure of SGT [22] Interestingly, the residue is located in the b-sheet separate from the cata-lytic triad of His37, Asp82 and Ser172 (corresponding

to residues 57, 102 and 195 of a-chymotrypsin number-ing; Fig 3E)

Effect of mutations on substrate specificity Next, we tried to prepare 11 SGT mutants and an SOT mutant, in which residue 71 was substituted for other amino acids, to investigate the role of Tyr71 of SOT in topological specificity We obtained mutants as active forms, except for an SGT mutant with proline substituted for Leu71 (SGT-L71P) Figure S1 shows SDS-PAGE data from several culture supernatants of SGT mutants Unlike other mutants, the band at the molecular mass of SGT-L71P was hardly observed, but many other bands were seen In addition, the cul-ture supernatant had no activity In the case of inac-tive SOT and SGT, in which Ser172 of the catalytic triad was substituted with alanine, many high molecu-lar mass bands were observed (data not shown), although these bands were not observed in active forms It is probable that correctly folded mutants could hydrolyse these high molecular mass proteins For SGT-L71P, low molecular mass bands were also observed at positions different from the case of active mutants We speculate that SGT-L71P, which was not folded correctly, was presumably hydrolyzed by other proteases in the culture These results suggest that residue 71 also affects the folding of SGT

Five purified SGT mutants (substitution of Leu71 with tyrosine, phenylalanine, tryptophan, alanine or histidine) showed higher activity towards type IV colla-gen than did wild-type SGT (Fig 4B, left) In parti-cular, the collagen IV⁄ I values of SGT-L71Y and SGT-L71H were twice as high as that of SGT (Fig 4C, left) Furthermore, mutant SOT-Y71L showed significantly lower specific activity towards type IV collagen than did wild-type SOT, although SOT-Y71L showed high activity towards type I colla-gen, similar to SOT (Fig 4A, B, right) From these results, it is interesting to note that the substitution with hydrophobic or bulky amino acids in wild-type SGT imparts a significant change in specific activity towards type IV collagen

We further evaluated the effect of mutations on the specificities for native type I collagen substrates from bovine Achilles’ tendon and type IV collagen from human placenta (Fig 5) The activities of SGT-L71Y (which showed the greatest change in substrate

150

200

250

Collagen type I

A

0

50

100

SGT A B C D E F G H SOT

SGT A B C D E F G H SOT

B

40

60

80

100

120

Collagen type IV

0

20

SGT A B C D E F G H SOT

0.4

0.5

0.6

Type IV/I

C

0

0.1

0.2

0.3

Fig 2 Comparison of the specific activities for chimeras A–H,

parental SGT and SOT towards different fluorescein-conjugated

substrates The reactions were performed using bovine skin

DQ-collagen type I (A) and human placenta DQ-DQ-collagen type IV (B) in

50 m M Tris ⁄ HCl (pH 8.0) containing 10 m M CaCl2at 37 C Enzyme

activity was measured by monitoring the fluorescence (FITC)

release (excitation, 395 nm; emission, 415 nm) The data are

expressed as the means ± SD of three independent experiments.

(C) Ratio of the hydrolytic activity towards type IV collagen to that

towards type I collagen.

specificity among the SGT mutants), SOT-Y71L,

wild-type SOT and SGT were compared The

colla-genolytic activity of SGT-L71Y for type IV collagen

was 3.3-fold higher than that of SGT (Fig 5B),

although the activity of SGT-L71Y towards type I

collagen was somewhat lower than that of SGT

(Fig 5A) In contrast, SOT-Y71L and SOT

hydroly-zed type I collagen similarly, but the activity of the

former towards type IV collagen was 2.2-fold lower

than that of the latter As shown in Fig 5C, collagen

IV⁄ I of SGT-L71Y was 4.8-fold higher than that of

SGT, whereas it was lower for SOT-Y71L than SOT

These results showed good agreement with those

obtained using fluorescein-labelled collagens

There-fore, we conclude that Tyr71 of SOT confers

topo-logical specificity

The ability of the mutation to hydrolyze the collagen

mimic substrate was determined using

benzyloxycarbon-ylglycyl-l-prolyl-l-arginine-4-methylcoumaryl-7-amide (Z-Gly-Pro-Arg-MCA), which is frequently employed

as a substrate in studies of the collagenolytic cysteine proteinase, cathepsin K [26,27] The specific activity was 1.4-fold higher in SGT-L71Y (90.3 lmolÆmin)1Æmg)1) than SGT (66.1 lmolÆmin)1Æmg)1), and 2.6-fold lower

in SOT-Y71L (190.2 lmolÆmin)1Æmg)1) than SOT (501.5 lmolÆmin)1Æmg)1) For this short peptide sub-strate, substitution of residue 71 has little effect, unlike the response seen with the structural protein substrates The kcatvalue of SGT-L71Y (51.3 s)1) was 1.6-fold higher than that of SGT (33.0 s)1), although their Km values were equivalent (Table 1) Conversely, both Km and kcat values of SOT-Y71L were approximately three-fold higher than those of SOT, resulting in a dra-matic decrease in the catalytic efficiency (kcat⁄ Km value) of SOT-Y71L The value was lower than that of SOT by one order of magnitude These results suggest

D82

Y71

S172

H37

A

Amino acid sequence

0.2

0.3

0

0.1

B B-1 B-2 B-3 B-4 C

120

140

Collagen type I

0

20

40

60

80

100

B B-1 B-2 B-3 B-4 C

50

0 10 20 30 40

B B-1 B-2 B-3 B-4 C

Fig 3 Identification of an amino acid residue related to topological specificity Primary structures of chimera B and C mutants (A) and their specific activities towards DQ-collagen type I (B) and DQ-col-lagen type IV (C) These activities were measured in 50 m M Tris ⁄ HCl (pH 8.0) con-taining 10 m M CaCl2at 37 C The data are expressed as the mean ± SD of three inde-pendent experiments (D) The ratio of the hydrolytic activity towards type IV collagen

to that towards type I collagen (E) The key residue (Tyr71 of SOT) for the overall structure of SOT is represented using the Swiss-pdb viewer based on the SGT crystal structure Tyr71 is indicated in pink with the residues in an active site.

that residue 71 affects both topological specificity and

catalytic efficiency significantly

Conformational change of SOT and SGT induced

by the substitution of residue 71

We measured the CD and fluorescence spectra of

SGT-L71Y, SOT-Y71L, SOT and SGT to investigate

the effect of the mutation at residue 71 on the tertiary

and secondary structures of SOT and SGT Figure 6A shows that the CD spectra of SOT-Y71L and SGT-L71Y were drastically different from those of wild-type SOT and SGT The spectra of the tryptophan fluores-cence emissions of the wild-type and mutants were also dramatically different (Fig 6B, C) SOT and SGT have five and four tryptophan residues, respectively, that contributed to their fluorescence emission spectra Figure 6B, C shows that the emission maximum,

A

B

60

80

100

120

100

150

200

Collagen type I

SGT-L71X

0

20

40

0

50

SOT- Y71L SOT

C

SGT-L71X

30

100

Collagen type IV

10

20

20

40

60

80

SOT- Y71L SOT

SGT-L71X

0.2

0.3

0.4 0.3

0.5

0.6

Type IV/I

0

0.1

0 0.1 0.2

SOT- Y71L SOT

Fig 4 Effects of mutations on topological

specificity Specific activities of SGT-L71X

mutants (left) and SOT-Y71L (right) towards

DQ-collagen type I (A) and DQ-collagen type

IV (B) were measured in 50 m M Tris ⁄ HCl

(pH 8.0) containing 10 m M CaCl 2 at 37 C.

The data are expressed as the mean ± SD

of three independent experiments (C) The

ratio of hydrolytic activity towards type IV

collagen to that towards type I collagen.

around 330 nm, was the same for all enzymes.

SOT-Y71L showed an extremely low fluorescence

emission intensity compared with that of wild-type

SOT In contrast, SGT-L71Y showed a higher fluores-cence emission intensity than did wild-type SGT Moreover, the CD and fluorescence spectra of SGT-L71F resembled those of SGT-L71Y, and the fluores-cence spectrum of SGT-L71S resembled that of SGT (data not shown) The measurements were performed under the same pH conditions for the above assays (pH 8; optimum pH of SOT and SGT) Of course, the autolysates of the mutants were not observed on SDS-PAGE Furthermore, we confirmed that these confor-mational changes were not caused by the disruption of the enzyme form using the hydrophobic fluorescence probe 1-anilinonaphthalene-8-sulfonic acid (ANS) (Fig S2) The fluorescence of ANS increases substan-tially when the hydrophobic core regions of proteins, which are inaccessible to the dye in the native struc-ture, are exposed The respective fluorescence intensi-ties of ANS with SOT, SGT and their mutants were similar to those without the enzymes, suggesting that the substitution of residue 71 induced conformational changes in the secondary and tertiary structures of wild-type SOT and SGT without disruption of the enzyme form From these results, we consider that the conformational changes induced by the substitution of residue 71 affect the topological specificity

New insights into substrate recognition The reaction mechanism within the catalytic triad and the specificities of the S1–Sn sites of chymotrypsin-like serine protease have been studied extensively [5–9] However, the recognition mechanism for structural protein substrates remains unclear Studies of cathepsin

K showed that its unique collagenolytic activity among cathepsins is caused by a preference for arginine and lysine at the P1 position and proline and glycine at the P2 and P3 positions [28]; neither of these residues, espe-cially proline, is preferred by other human cathepsins Previously, we have studied the residue preferences in SOT and SGT using fluorescence energy transfer

A

Collagen type I

3000

4000

5000

0

1000

2000

SGT SGT-L71Y SOT-Y71L SOT

B

Collagen type IV

30 000

SGT SGT-L71Y SOT-Y71L SOT

5000

10 000

15 000

20 000

25 000

0

SGT SGT-L71Y SOT-Y71L SOT

C

Type IV/I

4

5

6

7

0

1

2

3

Fig 5 Hydrolytic activities of SOT, SGT and their mutants towards

native collagens Collagenolytic activity was determined by the

fol-lowing method After preincubation of 2 mg of insoluble type I

colla-gen from bovine Achilles’ tendon (A) or type IV collacolla-gen from human

placenta (B) in 50 m M Tris ⁄ HCl (pH 8.0) containing 10 m M CaCl 2 ,

enzyme was added and incubated at 37 C for 30–60 min Next, the

reaction was terminated by the addition of 1 lL of 0.2 M HCl; the

rate of increase in free amino groups was measured using the

ninhy-drin method Clostridium histolyticum collagenase type I was also

estimated as a reference enzyme One collagen digestion unit

liber-ates peptides from collagen by collagenase type I equivalent in

nin-hydrin colour to 1.0 lmol of leucine in 5 h at pH 7.4 at 37 C in the

presence of CaCl2 Data are expressed as the mean ± SD of three

independent experiments (C) The ratio of hydrolytic activity towards

type IV collagen to that towards type I collagen.

Table 1 Kinetic parameters for the hydrolysis of a short peptide substrate by SOT, SGT and their mutants.

K m (l M ) k cat (s)1) k cat ⁄ K m (l M)1Æs )1 )

SGT-L71Y 17.2 ± 5.7 51.3 ± 7.8 3.2 SOT-Y71L 17.5 ± 2.9 93.0 ± 17.1 5.4

The assay was carried out using 0.03–0.5 m M Z-Gly-Pro-Arg-MCA

in 50 m M Tris ⁄ HCl (pH 8.0) containing 10 m M CaCl 2 at 37 C The data are expressed as the mean ± SD of three independent experi-ments.

substrate (FRETS) combinatorial libraries [29–31], which consist of a total of 475 peptide substrates In contrast with other studies, we found that the P1, P2 and P3 preferences of SOT resembled those of SGT However, they showed significantly different specifici-ties towards structural protein substrates [12] There-fore, the other regions, distinct S1, S2 and S3 sites, are presumed to confer topological specificity Our previous study showed that the N-terminal region of SOT included a key residue that conferred topological speci-ficity [12] We now expand these findings using RIBS

in vivoDNA shuffling to show that Tyr71 of SOT (cor-responding to residue 89 of a-chymotrypsin numbering) contributes to the topological specificity This finding has not been inferred from structural information The crystal structure of SGT consists of 15 b-sheets and two a-helices [22] (Fig 7A) SGT is divisible into two domains formed by six antiparallel b-sheets with a similar topology The catalytic triad lies in the cleft between the two domains Figure 7A shows that resi-due 71 in the b-sheet of the N-terminal b-barrel domain is located adjacent to Trp83, approximately at

3 A˚ (corresponding to residue 103 of a-chymotrypsin numbering) Therefore, we speculate that the substitu-tion of residue 71 affects the interacsubstitu-tion between this residue and Trp83, and subsequently causes a change

in the local environment around catalytic Asp82 This hypothesis is supported by the dramatic change in cat-alytic efficiency of SOT and SGT when carrying a mutation at residue 71 (Table 1)

Compared with other serine proteases for the struc-ture, the positional relationship between residue 71 and the catalytic triad in SOT almost resembles that in bovine trypsin (PDB ID: 1k1n) [32] and a-chymotryp-sin (PDB ID: 5cha) [33] Ile89 and Phe89 in bovine trypsin and a-chymotrypsin, respectively, correspond

to Tyr71 of SOT Residue 89 in these enzymes is situ-ated a short distance from Ile103, approximately 5 A˚ (corresponding to Trp83 of SOT) Thus, we speculate that these residues are not able to interact with each other The topological specificities of these serine pro-teases are assumed to be changed by the substitution

of residue 89 with other amino acids in order to allow for an interaction with Ile103

Figure 6 shows that the conformations of SOT and SGT differ, although the enzymes show similar hydro-lytic activity towards type I collagen The contribution

of an induced-fit mechanism to the substrate specificity

of aminoacyl-tRNA synthetase and adenylate kinase has been reported recently [34,35] Moreover, the 60-loop of thrombin lining the upper rim of the active site entrance is rearranged by binding exosite I of throm-bin and the protease-activated receptor PAR3 [36]

A

0

0.5

2 ·d

Wavelength (nm)

–1

SGT SGT-L71Y

C

300

400

500

SGT SGT-L71Y

0

100

200

Wavelength (nm)

SOT

100

200

300

0

Wavelength (nm)

Fig 6 Conformation of SOT, SGT and their mutants CD spectra

(A) and tryptophan fluorescence emission spectra (B, C) of SOT,

SGT and their mutants were measured at room temperature in

10 m M Tris ⁄ HCl (pH 8.0) containing 10 m M CaCl2, as described in

Materials and methods.

Indeed, the 60-loop shifts 3.8 A˚ upwards and causes a

180 flip of W60d that projects the indole ring into the

solvent and opens up the active site fully, when PAR3

binds to exosite I, although the indole ring of W60d

partially occludes access to the active site and restricts

the specificity towards physiological substrates and

inhibitors in free thrombin Based on these findings,

we propose that residue 71 triggers induced fitting

when the enzyme–substrate complex is formed in the

first step of the reaction mechanism Consequently, the

environment around the active site presumably changes

to allow structural protein substrate attachment; it

engenders change in topological specificity

Finally, this is the first report to identify the key

amino acid residue conferring topological specificity in

Streptomyces trypsin The general collagenases, such

as MMP and Clostridium histolyticum collagenase,

have a collagen-binding domain (CBD) [37,38], and

Tyr994 in this domain is the critical residue for

inter-action with collagen [38] The hydroxyl group of this

residue probably forms hydrogen bonds with

main-chain atoms to form a protein–collagen complex

Because the molecular size and structure of

trypsin-like serine proteases and these collagenases differ, the

domain corresponding to the CBD remains elusive

Nevertheless, residue 71 might also contribute to

colla-gen-binding Figure 7B shows that the residue is

located in the basic surface charged region Thus, SOT

and SGT might possess other collagen-binding regions

that act synergistically with residue 71 to promote the

binding of the acidic surface of collagens to the basic

surface charged region of the enzyme Studies are

underway using surface plasmon resonance analysis to

determine the role of residue 71 in collagen binding

The elucidation of the mechanism of structural protein

substrate recognition in serine proteases should help

advance therapeutic research into the prevention and

treatment of thrombotic and neurodegenerative dis-eases caused by ‘hard-to-degrade’ proteins

Materials and methods

Materials

A spin column (Vivapure S; Vinascience, Sartorius AG, Aubagne, France) and DQ-collagens (Molecular Probes Inc., Eugene, OR, USA) were used for this study Type I collagen from bovine Achilles’ tendon, type IV collagen from human placenta and Clostridium histolyticum collage-nase type I were purchased from Sigma-Aldrich Inc (St Louis, MO, USA) Z-Gly-Pro-Arg-MCA was obtained from Peptide Institute Inc (Minoh, Osaka, Japan) All other unspecified chemicals were of the highest purity available

Construction of plasmid for RIBS in vivo DNA shuffling

The plasmid for RIBS in vivo DNA shuffling was con-structed as described previously [23] A plasmid containing the sot and sprT genes cloned in tandem was constructed as follows The sot gene in plasmid pUC18 was digested with NdeI and KpnI The sprT gene in pCR-Blunt II-TOPO was digested with KpnI and HindIII Next, the sot and sprT genes were ligated into the NdeI–HindIII gap of plasmid pACTI2b [23], yielding pACTI2b(sot⁄ sprT) The Gmr -rpsL+cassette from pNC124 [23] was digested with KpnI, and the cassette was inserted into the KpnI site between the sot and sprT genes in pACTI2b(sot⁄ sprT) to construct pACTI2b(sot⁄ Gmr-rpsL+⁄ sprT) (Fig 1A) The resulting vector was used for RIBS in vivo DNA shuffling

Random chimeragenesis Figure 1 shows the random chimeragenesis strategy First,

E coli MK1019 [ssb-3 rpsL(Smr)] was transformed with

S172 D166

37H D82

Y71

B

S172

D166

W83 37H

D82

Y71

A

Fig 7 The amino acid residues related to topological specificity and the conformation

of the three-dimensional structures (A) The overall structure of SOT is portrayed using the Swiss-pdb viewer based on the crystal structure of SGT The key residue is indi-cated in red (Tyr71 of SOT) The residues in the catalytic site are also shown (B) The surface charge is represented using the Swiss-pdb viewer Acidic and basic surface charges of SOT are shown as red and blue, respectively.

pACTI2b(sot⁄ Gmr

-rpsL+⁄ sprT) by electroporation The transformants were selected using Luria–Bertani (LB) plates

containing chloramphenicol and gentamicin The Cmr⁄ Gmr

transformant sensitivities to streptomycin (Sms) were

con-firmed on LB plates containing chloramphenicol,

gentami-cin and streptomygentami-cin The Sms transformants from each

host were cultivated in LB medium containing

chloramphe-nicol Cultures were spread on LB plates containing

chl-oramphenicol and streptomycin Plasmids isolated from 48

selected Smrrevertants from each host were analysed using

agarose gel electrophoresis The 0.8 kb DNA fragment

con-taining the target gene on isolated plasmids was amplified

by PCR using LA taq with a GC buffer system (TaKaRa

Holdings, Inc.), with primers incorporating the NdeI site of

sot (5¢-CATATGCAGAAGAACCGACTCGTCC-3¢) and

the HindIII site of sprT (5¢-TGCCGGTACGAAGCTTCA

GAGCGTGCG-3¢) Recombination sites were determined

in detail using DNA sequencing

Construction of expression vectors

To prepare chimeras, we applied a novel expression system

[39] using the expression vector (pTONA5a), which

included a promoter from Streptomyces

metalloendopepti-dase, with Streptomyces lividans 1326 as a host strain The

chimeric gene was digested using NdeI and HindIII and

ligated into the NdeI–HindIII gap of pTONA5a to obtain

the expression vector

Construction of expression vectors of chimera B

and C mutants

To identify the amino acid residues related to topological

specificity, we constructed chimera B and C mutants using

PCR amplification To prepare the mutants (B-2 and B-3),

the following two mutagenic sense primers were synthesized

[the XhoI site (in italic type) was substituted with a silent

mutation]: 5¢-TCCAGTC(G fi C)TC(C fi G)AGCGCC

(Gfi A, Val fi Ile)TCAAG-3¢ (corresponding to nucleotides

281–303 from sprT) and 5¢-TC(G fi C)TC(C fi G)AGC

GCC(Gfi A, Val fi Ile)TCAAGGTCCGCTCCACCAAG

(Gfi A, Val fi Ile)TC-3¢ (corresponding to nucleotides

286–321 from sprT) The target mutation was introduced

with primer sets of 5¢-TGCCGGTACGAAGCTTCA

GAGCGTGCG-3¢ (a reverse primer, corresponding to the

HindIII site of sprT) and each of the mutagenic primers,

using KOD-Plus (version 2, Toyobo Co Ltd.) The partial

sotgene was amplified using PCR with a combination of a

forward primer (5¢-CATATGCAGAAGAACCGACTCG

TCC-3¢, corresponding to the NdeI site of sot) and a reverse

primer [5¢-GATGGCGCT(GCT fi CGA)GGACTGGAG

GT-3¢ for silent mutation of the XhoI site (in italic type)

and corresponding to nucleotides 284–306 from sot] The

amplified DNA fragments were cloned into pCR-Blunt

II-TOPO (Invitrogen Corp.); the resulting plasmids were

confirmed by DNA sequencing The plasmids representing B-2 and B-3 were digested with XhoI and HindIII The plasmid representing the partial sot gene was digested with NdeI and XhoI The fragments were ligated into the NdeI–HindIII gap of pTONA5a to construct the expression vector

The following mutagenic antisense primer, in which the KpnI site (in italic type) was substituted with a silent muta-tion, was synthesized to prepare the mutant B-4: 5¢-CGG T(Gfi A)CCGTTGTAGCCGGGGGCCTGG(AG fi TA, Leufi Tyr)-3¢ (corresponding to nucleotides 322–349 from sprT) The target mutation was introduced with primer sets

of 5¢-CATATGCAGAAGAACCGACTCGTCC-3¢ (a for-ward primer, corresponding to the NdeI site of sot) and the mutagenic primer described above, using KOD-Plus The partial sprT gene was amplified by PCR with a combina-tion of a forward primer [5¢-CCGGCTACAACGG(C fi T)ACCGGCAA-3¢ for silent mutation of the KpnI site (in italic type), corresponding to nucleotides 332–353 from sprT] and a reverse primer (5¢-TGCCGGTACGAAGCTT CAGAGCGTGCG-3¢, corresponding to the HindIII site of sprT) The amplified DNA fragments were cloned into pCR-Blunt II-TOPO, and the resulting plasmids were confirmed by DNA sequencing The plasmid repre-senting B-4 was digested with NdeI and KpnI The plasmid representing the partial sprT gene was digested with KpnI and HindIII The fragments were ligated into the NdeI–HindIII gap of pTONA5a to construct the expression vector To prepare the mutant B-1, the chime-ric gene obtained by RIBS in vivo DNA shuffling was digested using NdeI and HindIII, and ligated into the NdeI–HindIII gap of pTONA5a to construct the expression vector

Construction of SGT-L71X mutants and SOT-Y71L

We constructed SGT-L71X mutants and SOT-Y71L to investigate the effect of distinct residues on the recognition

of the substrates To prepare SGT-L71X mutants, the mutagenic gene was amplified using PCR with a combina-tion of a forward primer (5¢-CAACATATGAAGCACT TCCTGCGTGC-3¢, corresponding to the NdeI site of sprT) and a reverse primer [5¢-CGGT(G fi A)CCGTTGTAGC CGGGGGCCTG(GAGfi XXX)GACCTTG-3¢ for silent mutation of the KpnI site (in italic type), corresponding to nucleotides 315–349 from sprT] When XXX was GTA, GAA, CCA, GGC, GCC, GCG, GTG, GTC, GTT, GGA and GGG, leucine was substituted with tyrosine, phenylala-nine, tryptophan, alaphenylala-nine, glycine, argiphenylala-nine, histidine, aspar-tic acid, asparagine, serine and proline, respectively The amplified DNA fragments were then cloned, sequenced and digested with NdeI and KpnI The plasmid representing the partial sprT gene with the KpnI site described above was digested with KpnI and HindIII The fragments were ligated