Characterization of Erwinia chrysanthemi extracellular proteases ...

Vol. 169, No. 11

JOURNAL OF BACTERIOLOGY, Nov. 1987, p. 5046-5053 0021-9193/87/115046-08$02.00/0

Copyright © 1987, American Society for Microbiology

Characterization of Erwinia chrysanthemi Extracellular Proteases: Cloning and Expression of the Protease Genes in Escherichia coli CICILE WANDERSMAN,* PHILIPPE DELEPELAIRE, SYLVIE LETOFFE, AND MAXIME SCHWARTZ

Unite de Genetique Moleculaire, Institut Pasteur, 75724 Paris Cedex 15, France Received 30 March 1987/Accepted 17 August 1987

Erwinia chrysanthemi, a phytopathogenic enterobacterium, secretes three antigenically and structurally distinct proteases, A, B, and C, and produces a protease inhibitor, a low-molecular-weight, heat-stable protein which remains mostly intracellular and which binds specifically to the A, B, and C proteases. The structural genes for proteases A, B, and C and for the inhibitor are clustered on a ca. 40-kilobase DNA fragment present in cosmid pEW4. Escherichia coli strains harboring pEW4 secrete the three proteases into the medium during the exponential phase of growth, without intracellular accumulation and in the absence of detectable cell lysis. An 8.5-kilobase EcoRI fragment derived from the cosmid encodes proteases B and C and the inhibitor as well as functions involved in the synthesis or secretion (or both) of the proteases. The inhibitor is not required for protease synthesis or secretion.

pMMB33 were used in this study (3, 9, 17). All media have been described previously (21). E. chrysanthemi and E. coli were grown in LB medium at 30 and 37°C, respectively. Extraction and manipulation of plasmids. Isolation of plasmids, transformation of bacteria, and restriction endonuclease mapping were performed as described by Maniatis et al. (17). Tn5 mutagenesis (27) and chemical mutagenesis (21) were described previously. Isolation or protease hyperproducing mutant. Spontaneous hyperproducing mutants were obtained by plating a B374 culture on skim milk agar (29). Small papillae growing out of the bacterial lawn were picked after 2 days of incubation at 30°C. Clones exhibiting the largest hydrolysis zones around the colonies were purified and assayed for proteolytic activity. Supernatant concentration. At the times indicated in the different figures, samples of the cultures were taken and centrifuged for 15 min at 5,000 x g at 4°C. The supernatant media were concentrated by ammonium sulfate precipitation (80% saturation) for 4 h at 4°C. The precipitate was centrifuged for 30 min at 15,000 x g at 4°C, suspended in a minimal volume of 100 mM Tris hydrochloride, pH 8.0, and extensively dialyzed against 100 mM Tris hydrochloride, pH 8. Proteolytic activity was stable during this concentration procedure. The concentration factor was 100. Alternatively, inactive concentrated culture supernatants were prepared by precipitation with 10% trichloroacetic acid for 1 h at 4°C. The precipitate was centrifuged for 10 min at 10,000 x g, washed once in 80% acetone, and suspended in 100 mM Tris hydrochloride, pH 8.0. Localization of enzyme activities in the different compartments. Two successive samples (200 ml each) were taken from a culture, one during the exponential growth phase (optical density at 600 nm [OD600], 1.0), the other in the early stationary phase (OD600, 2.5). The supematant medium, which contains the extracellular proteins, was assayed either directly for 1-lactamase activity (a periplasmic protein) and 3-galactosidase activity (a cytoplasmic protein) or after concentration by ammonium sulfate for protease activity. The cell pellets were washed once with 100 mM Tris hydrochloride, pH 8.0, and subjected to osmotic shock (23). The shocked cells were disrupted by sonication for 10 s at 50

Many gram-positive and gram-negative bacteria secrete proteins, such as enzymes and toxins, into the medium (25). The mechanisms whereby these enzymes are released from the cells have been extensively studied in gram-positive bacteria, which use a signal peptide-dependent pathway similar to that used by gram-negative bacteria to export proteins to the periplasm (18, 19). The mechanism of protein secretion in gram-negative bacteria, which involves the crossing of both the cytoplasmic membrane and outer membrane, is comparatively less well documented (24, 25). Whereas Escherichia coli secretes few proteins, other gramnegative bacteria, including enterobacteria closely related to E. coli, secrete many different proteins into the medium (25). Several specific and independent pathways appear to be used for the secretion of different proteins. We have chosen to study the exoproteases of Erwinia chrysanthemi as an example of proteins secreted by gram-negative bacteria (29). E. chrysanthemi is a gram-negative phytopathogenic bacterium which causes soft rot disease in various plants (5). It produces many extracellular enzymes, including pectinases, cellulases, and proteases (5, 10, 29). Whereas the pectinases and cellulases are released during the stationary phase of growth (1), the proteases are released during exponential growth, without concomitant lysis (29). To characterize the genes involved in protease synthesis and secretion, we have studied proteolytic clones of E. coli K-12 harboring a cosmid which carries a fragment of E. chrysanthemi B374 DNA (26). In this work, we characterized three E. chrysanthemi proteases and studied their production and secretion in E. coli. We show that, when produced by E. coli, the E. chrysanthemi proteases are secreted in the absence of cell lysis and without prior intracellular accumulation. MATERIALS AND METHODS Strains, plasmids, and media. E. chrysanthemi B374 has been described previously (28). Strain HP1 is a spontaneous hyperproteolytic mutant of B374 (this work). Two E. coli strains were used: C600 and HB101, which bears a recA mutation (17). Plasmids pBR322 and pACYC184 and cosmid * Corresponding author. 5046

E. CHRYSANTHEMI PROTEASE GENE EXPRESSION

VOL. 169, 1987

W in a Branson Sonifier model B12. Intact cells were sedimented by centrifugation for 5 min at 5,000 x g. Again, protease activity, 3-galactosidase activity, and 1-lactamase activity were assayed in the soluble fraction of the shocked cells (periplasm) and within the sonicated cells (cytoplasm). Alternatively, inactive concentrated cell extracts were prepared by boiling whole cells in sodium dodecyl sulfate (SDS)-sample buffer before protein separation by SDSpolyacrylamide gel electrophoresis (PAGE). Measurements of enzyme activities. Protease activity was measured on concentrated culture supernatants or in cytoplasmic extracts by monitoring the hydrolysis of azocasein as previously described (29). Enzyme activity was expressed by the increase in A44 x 108 min1 ml-' per OD6 unit. 3-Galactosidase and ,B-lactamase assays were performed as described in references 21 and 15, respectively. After SDS-PAGE of the concentrated culture supernatants and the purified proteins, proteolytic activity was revealed as follows. The polyacrylamide gel was layered on top of a 1-mm-thick agar gel (2% Noble agar [Difco], 5% skim milk [Difco], 100 mM Tris hydrochloride, pH 8.0) and they were incubated at 37°C for 4 h, during which time the proteases diffused from the polyacrylamide gel into the agar gel and hydrolyzed skim milk proteins; the protease activity was visualized as a cleared band on the opaque background. Peptide mapping. The B and C proteases were purified from concentrated E. coli C600 (pRUW1) culture supernatant by two cycles of preparative electrophoresis on 7.5% SDS-polyacrylamide gels, followed by electroelution of the proteins from the gel strip. They were then precipitated with 1 2 3 4 5 6 7 8 ;M ~ . .^: 1:. t .-I

....

q-

-~~~

40

qwp

A

_

_

_

A

B

FIG. 1. Exoproteases produced by E. chrysanthemi HP1 and by several E. coli C600 recombinant clones. Concentrated culture supernatants were subjected to SDS-PAGE analysis. Polyacrylamide concentration was 7.5% (panel A) or 12% (panel B). (A) Detection of proteolytic activity on skim milk agar (see Materials and Methods) for the supernatant from late stationary cultures. Lanes: 1, E. chrysanthemi HP1; 2, E. coli C600 (pEW4); 3, E. coli C600; 4, E. coli C600(pRUW1); 5, 6, and 7, purified proteases A, B, and C, respectively. The amounts loaded on each lane correspond to (lanes): 1, 280 ,ul; 2, 400 ,ul; 3, 140 ,ul; 4, 140 ,u1 of culture supernatant. (B) Coomassie blue staining of the supernatants from late stationary culture. Lanes: 1, molecular weight markers (92, 66, 45, 31, 21.5, and 14.4 kilodaltons); 2, 3, and 4, purified proteases A, B, and C, respectively; 5, E. coli C600(pBR); 6, E. chrysanthemi HP1; 7, E. coli C600(pEW4); 8, E. coli C600(pRUW1). The amounts loaded on each lane correspond to (lanes): 5, 3 ml; 6, 3 ml; 7, 3 ml; 8, 3 ml of culture supernatant. Arrows on right point to protease positions.

5047

20% trichloroacetic acid, washed in 80% acetone, and suspended in 125 mM Tris hydrochloride (pH 6.8)-0.5% SDS10% glycerol. They were digested with increasing amounts of staphylococcal V8 protease (see Fig. 3) or chymotrypsin (data not shown) for 30 min at 37°C as described by Cleveland (7). After digestion, the samples were analyzed on a 15% SDS-polyacrylamide gel. Crude preparation of inhibitor. One liter of an overnight culture of C600(pRUW1) grown in LB medium was centrifuged for 30 min at 5,000 x g at 4°C. The cell pellet was washed once with 100 mM Tris hydrochloride, pH 8, and suspended in 20 ml of the same buffer, and the cells were then disrupted by sonication in a Branson Sonifier model B12 (six successive sonications at 50 W for 15 s each; each sonication was followed by cooling of the cells on ice). The disrupted cells were centrifuged for 30 min at 20,000 x g. The supernatant was incubated in a boiling water bath for 5 min and centrifuged at 10,000 x g for 10 min, and the resulting supernatant was precipitated with ammonium sulfate (80% saturation) for 30 min at 4°C. After centrifugation for 10 min at 10,000 x g, the supernatant was extensively dialyzed against 100 mM Tris hydrochloride, pH 8.0, and precipitated with acetone (80%); after centrifugation at 10,000 x g for 10 min at 4°C, the pellet was suspended in 1 ml of 100 mM Tris hydrochloride, pH 8.0. Similar preparations were made for E. coli C600 and C600(pRUW1::TnS120). Protease inhibitor activity was tested in cell extracts from different strains against a concentrated culture supernatant from E. coli C600(pRUW1) as described by Millet (22). A concentrated culture supernatant (200 ,ul) was incubated for 20 min at room temperature with 40 pl of the inhibitor preparation at different concentrations or with 40 ,uI of 100 mM Tris hydrochloride, pH 8.0. Protease activity was then measured by monitoring azocasein hydrolysis as described above. The inhibition of protease activity AOD440 x 108 min-1 is the difference of protease activity without and with the inhibitor preparation. The dilution of the inhibitor preparation was adjusted to give 30 to 60% inhibition. The inhibitor activity is expressed as AOD440 x 108 min-' ml of inhibitor preparation-' per OD6 unit of inhibitor preparation, and 1 U of inhibitor activity will inhibit 1 U of protease activity. Immunological techniques. A rabbit antiserum against protease A was prepared by using purified protease A. A rabbit antiserum against proteases B and C was prepared by using the crude concentrated supernatant of the E. coli C600 (pRUW1) culture, which contains mostly B and C proteases (Fig. 1B, lane 8). Immunoblotting was done as described previously (4). The second antibody was a mouse anti-rabbit immunoglobulin G coupled to peroxidase and was revealed by the H202diaminobenzidine technique. A sandwich technique was used to identify in intracellular fractions protein(s) able to bind to the B and C proteases. Partially purified inhibitor preparations were run on a 15% polyacrylamide-SDS gel in the presence of 8 M urea. The separated proteins were transferred onto a nitrocellulose sheet alongside molecular weight markers. After transfer, the nitrocellulose sheet was stained with Ponceau red, which allowed the positioning of the markers. It was then saturated with phosphate-buffered saline containing 5% skim milk for 1 h at 37°C, washed twice in phosphate-buffered saline-3% bovine serum albumin, and incubated for 1 h at 37°C in phosphate-buffered saline-3% bovine serum albumin containing a concentrated culture supernatant of C600 (pRUW1) during which proteases could bind to the inhibitor. It was

5048

WANDERSMAN ET AL.

finally processed for immunodetection of proteases bound to the inhibitor by using an anti-B+C antibody. RESULTS Characterization of E. chrysanthemi proteases. Proteolytic activity in the cell-free medium from an E. chrysanthemi B374 culture was very unstable (data not shown). All attempts to concentrate this activity led to significant losses (data not shown). A spontaneous mutant (HP1) exhibiting a higher proteolytic activity was selected by growth on skim milk agar as described in Materials and Methods. The supernatant from a culture of strain HP1 could be concentrated without significant loss of the proteolytic activity. The concentrated supernatant from an overnight stationaryphase culture of strain HP1 was subjected to SDS-PAGE analysis and either blotted onto skim milk agar (Fig. 1A, lane 1) or stained with Coomassie blue (Fig. 1B, lane 6). Three protein bands with apparent molecular masses of 50, 53, and 55 kilodaltons exhibited proteolytic activity. The three secreted proteases are referred to as A, B, and C in order of their increasing apparent molecular masses. Cloning of the E. chrysanthemi genes for secreted proteases. A genomic library of E. chrysanthemi B374 DNA was constructed in cosmid pMMB33 and introduced into E. coli K-12 by Reverchon et al. (26). Recombinant E. coli clones expressing proteolytic activity were generously provided by S. Reverchon. The cosmid (pEW4) carried by one proteolytic recombinant was isolated and used to transform E. coli C600 with a selection for kanamycin resistance. All of the kanamycin-resistant transformants were proteolytic, confirming that pEW4 carries a protease gene(s). Plasmid pEW4 was digested with EcoRI and ligated with EcoRI-linearized pBR322. The ligation mixture was used to transform E. coli C600 with selection for ampicillin resistance, and transformants were screened for proteolytic activity on skim milk agar. One protease-positive clone, carrying a plasmid (pRUW1) with an 8.5-kilobase (kb) EcoRI insert, was selected for further studies. Characterization of the proteases produced by recombinant E. coli clones. All of the proteolytic activity produced by E. coli C600(pEW4) or C600(pRUW1) was found in the cell-free culture medium. The concentrated supernatant media from an overnight culture of C600(pEW4) and C600(pRUW1) were subjected to SDS-PAGE analysis and either blotted onto skim milk agar (Fig. 1A) or stained with Coomassie blue (Fig. 1B). In the case of C600(pEW4), three major bands with proteolytic activity were detected corresponding to the same apparent molecular weight as proteases A, B, and C from E. chrysanthemi HP1 (Fig. 1A and B). In the case of C600(pRUW1), two bands with proteolytic activity were detected corresponding to the same apparent molecular weight as proteases B and C from HP1 (Fig. 1A and B). Further comparison of the proteases produced by C600(pEW4) and C600(pRUW1) with those synthesized by HP1 was performed by immunological tests using antibodies raised against purified protease A or against the mixture of proteases B and C (see Materials and Methods). The immunoblot shown in Fig. 2 reveals that these antibodies always recognized the corresponding proteases A or B and C, irrespective of the producing strain (E. coli or E. chrysanthemi). The results shown in Fig. 1A and B and 2 indicate that cosmid pEW4 still encodes proteases A, B, and C, whereas plasmid pRUWl encodes proteases B and C but not protease A. Proteases B and C were produced in larger amounts by the cells harboring pRUWl than by those

J. BACTERIOL.

carrying pEW4 (Fig. 1B, lanes 7 and 8). This can be accounted for by the difference in plasmid copy number: pRUW1, a derivative of pBR322, is present at about 30 copies per cell (11), whereas pEW4, an RSF1010 derivative, is present at 10 copies per cell (2). The three proteases A, B, and C could be purified separately by two cycles of preparative SDS-PAGE and still retained their activity in our assay on skim milk agar (Fig. IA, lanes 5, 6, 7, and B, lanes 2, 3, and 4). It should be noticed that during purification protease C was converted into a slightly slower-migrating form (Fig. 1A, lane 7, and B, lane 4), whereas proteases A and B did not change mobility. The immunological tests performed on the concentratead culture supernatants (Fig. 2) showed that there was no immunological cross-reactivity between protease A and proteases B and C. Protease A is therefore distinct from proteases B and C. We also showed, using limited proteolysis by staphylococcal V8 protease, that B and C display different patterns (Fig. 3), indicating that proteases B and C are distinct proteins. Similar results were obtained with chymotrypsin (data not shown). Protease secretion in E. coli. The analysis by SDS-PAGE of concentrated stationary-phase culture supernatants shown in Fig. 1B indicates that the proteases are the major bands present in the supernatant of C600(pEW4) or C600(pRUW1). The level of the other proteins stayed the same in E. coli C600 as well as in C600(pEW4) or C600(pRUW1), indicating that there is no appreciable cell lysis in the proteaseproducing cells. E. coli carrying pEW4 or pRUW1 was able to secrete proteases into the culture medium during the stationary phase. We investigated whether this protease secretion occurs also during the exponential phase of growth, which is associated with the lowest cell lysis. In E. coli (pEW4) or (pRUW1) cultures, the proteolytic activity could already be detected during the exponential phase of growth (Table 1). This was also the case for E. chrysanthemi HP1 exponential-growth-phase culture (data not shown). The SDS-PAGE analysis of concentrated supernatant of exponential-growth-phase cultures indeed showed the presence of proteases but the interpretAtion of the SDS-PAGE was hindered by the electrophoretic heterogeneity of the exoproteases (data not shown). To confirm that the release 1 2 3 4

1 2 3 4

(anti-A)

(anti-B+C)

FIG. 2. Immunodetection of proteases A, B, and C in culture supematants of the different strains. Concentrated culture supernatants from late stationary culture were subjected to SDS-PAGE (7.5%) analysis and subsequently transferred onto nitrocellulose. Antibodies against protease A (I) or against the mixture of proteases B and C (II) were used. Lanes: 1, E. chrysanthemi HP1; 2, E. coli C600(PEW4); 3, E. coli C600(PRUW1); 4, E. coli C600. The amounts loaded on each lane correspond to the following volume of culture supernatants: 1, 3 ml; 2, 6.2 ml; 3, 1.4 ml; 4, 3.5 ml.

VOL. 169, 1987

1 2 3 4 5 6 7

FIG. 3. Digestion of B and C proteases by staphylococcal V8 protease. Purified protease C (lanes 1, 3, and 5) and protease B

(lanes 2, 4, and 6)

were digested with different concentrations of staphylococcal V8 protease: 25 p.g/ml (lanes 1 and 2), 50 ~~.g/ml (lanes 3 and 4), and 250 ~Lg/ml (lanes 5 and 6). Lane 7 corresponds to staphylococcal V8 protease alone (250 p.g/ml) incubated under the same conditions. The samples by SDS-PAGE (15%). Staining was with silver nitrate.

were'analyzed

of proteases

arose

related with

a

from

a specific secretion process uncorlysis in recombinant E. ccli clones, samples of a C600(pEW4, pBR322) or C600(pRU W1) culture were harvested at two different growth stages (exponential 2.5), and the cells ODN= 1.0, and early stationary OD6wo

cellular

=

were

5049

E. CHRYSANTHEMI PROTEASE GENE EXPRESSION

fractionated

as

described in Materials and Methods.

The distribution of protease activity, activity (a cytoplasmic marker enzyme), and P-Iactamase(a periplas-

activity is already released from E. coli C600(pRUW1) or C600(pEW4) during the exponential growth phase (OD600 = 1). On the other hand, the fractionation experiments shown in Table 1 clearly indicate that the cytoplasmic and periplasmic enzymes are retained in the E. coli host in all cases studied: C600(pBR322), C600(pEW4, pBR322), and C600 (pRUW1). Similar results were obtained for the early stationary culture (OD600 = 2.5 [not shown]). It appears, therefore, that protease secretion in recombinant E. coli clones does not result from cellular lysis but from a specific process. Identification of a protease inhibitor. In E. chrysanthemi HP1 cultures, as well as in E. coli C600 harboring pEW4 or pRUW1, proteolytic activity was detected only in the supernatant medium (Table 1). The lack of proteolytic activity in soluble cell extracts might be caused by a protease inhibitor. Preliminary protease assays were performed to test this hypothesis. Concentrated culture supernatant from E. coli C600(pRUW1) was mixed with increasing amounts of cell extracts from E. coli C600, C600(pEW4), or C600(pRUW1) or E. chrysanthemi HP1. With all the cell extracts but C600, the protease activity decreased with increasing amounts of cell extract (data not shown). This result suggested an inhibitory activity within the protease-producing cells. The same results were obtained once the cell extracts had been boiled for 10 min, indicating that this inhibitory activity was heat stable. This activity could be quantified by measuring the inhibitor activity, as described in Materials and Methods. Cell extracts of three different strains were tested for inhibitory activity: C600, C600(pRUW1), and C600(pRUW1::Tn5-120). The last strain had a TnS insertion in the inh gene (see below). Table 2 shows that no inhibitory activity could be detected in cell extracts from either C600 or C600(pRUW1::Tn5-120) at concentrations for which cell extracts from C600(pRUW1) fully inhibited the protease activity. The inhibitor was found to be active against the A, B, and C proteases but inactive against other proteases tested (trypsin, subtilisin, papain, and proteinase K [data not shown]). The inhibitory activity was sensitive to immobilized proteinase K (data not shown),

3-galactosidase

mic marker enzyme) determined (Table 1).

in

the

different compartments

Table 1

shows

that

the

was

TABLE 2. Detection of protease-inhibitory activity in cell

extracts from E. coli C600 harboring different plasmidsa

proteolytic

IProtease ( OD ~activity

Inhibition factor

Ianchtibvityr (U/ODvU.)

Control (+ buffer) E. coli C600 cell extracts (x 1,000 concentrated) E. coli C600(pRUW1) cell extract x 1,000

3,000 3,000

99% >99% 40%