fVlolecular fVlicrobiology (1992) 6(23), 3521-3529

Secretion of the STA3 heat-stable enterotoxin of Escherichia coli: extracellular delivery of Pro-STA is accomplished by either Pro or Yuan Yang, Zeren Gao, Luz-Mana Guzman-Verduzco,^ Kathy Tachias and Yankel M. Kupersztoch* Department of f\/licrobiotogy, University of Texas Southwestern Medical Center, Dallas, Texas 75235-9048.

Summary The methanol-soluble, heat-stable enterotoxin of Escherichia coli is a protease-resistant extracellular peptide which is synthesized as a 72-amino-acid precursor Pre-Pro-STA3. The specific roles of Fre (19 amino acids), Pro {34 amino acids) and ST^^ (19 amino acids) in the secretion process were studied by functionally deleting each of the three domains. Deletion of the Pre signal sequence resulted in a short-lived cell-associated molecule with an M^ equivalent to that of Pro-STA3. Deletion of Pro (i.e., PreSTA3) resulted in the rapid extracellular accumulation o* STA3; the periplasmic intermediate found in the secretion of the wild-type toxin was undetected. Deletion of the STA3 domain resulted in a cell-associated Pre-Pro peptide; with time this form converted to periplasmic Pro which later became extracellular. When DNA encoding either STA3, by itself, or ProSTA3 (lacking the signal peptide) was expressed, these peptides were degraded intracellularly, with no periplasmic or extracellular forms detected. The results presented demonstrate that the signal peptide (Pre) is essential even for the export of small peptides to the periplasm, and that its absence causes the STAS domain to become susceptible to intracellular proteases. The rapid degradation of intracetlular STA3 indicates that its proteolytic resistance is acquired in a compartment other than the cytoplasm. The results also show that after the Pre domain is proteolytically cleaved from Pre-STA3 and Pre-Pro, the STA3 and Pro peptides can exit to the culture supernatant. Since STA3 and Pro have neither sequence nor conformational homology, it appears

Received 4 March, 1992; revised 9 July, 1992; accepted 15 July, 1992. tPresent address' Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston. Massachusetts. *For correspondence. Tel, (214)688 2210; Fax (214)688 2199.

that various peptide conformations are compatible with exit through the outer membrane. Introduction In Gram-negative bacteria, severai mechanisms have evolved for the extraceliuiar delivery of peptides, a process known as secretion (Pugsley ef ai, 1990). In one system, exemplified by the secretion of «-haemolysin, the gene product of hlyA. there is no proteolytic processing (Felmlee et ai. 1985), and extracellular targeting is directed by a specific C-terminal region. (Gray et ai, 1989). No periplasmic intermediates are found and hylA does not code for an amino-terminal signal sequence; rather additional gene products (HlyD and HlyB) are needed for secretion (Wagner etai. 1983); it has been proposed that they form transenvelope complexes through which «-haemolysin is secreted (Mackman etai, 1986). Other extracellular peptides use a similar pathway (Reviewed by Holland ef ai, 1990). Other proteins, like the pullulanase of Kiebsiella plantolytica, are secreted by a more complex mechanism. Expression of the putA structural gene in E. co//leads to the accumulation of the fatty acylated and processed enzyme on the periplasmic side of the inner membrane. Its exit to the culture supernatant requires approximately 13 additional genes (Kornacker ef ai. 1991). Other proteins are secreted by a mechanism that is accompanied by the permeabilization of the outer membrane by the action of accessory proteins. This group of proteins, exemplified by cloacin DF13 (DeGraaf and Oudega, 1986), lack a signal peptide and translocate the inner and outer membrane by an unknown mechanism that is dependent on the participation of the so-called release protein (Oudega et ai. 1984). It has been proposed that when the release protein is present in the inner membrane it favours the export of the precursor into the periplasm, and when it is present in the outer membrane it facilitates the extracellular secretion of the peptide (DeGraaf and Oudega, 1986). Yet another secretory mechanism used by Gram-negative bactena is illustrated by IgA protease whose precursor has an amino-terminal signal peptide which is cleaved as the protein reaches the periplasm (Pohlner ef ai. 1987). It has been proposed that the C-terminus of the periplasmic intermediate then forms a pore in the outer membrane through

3522

Y. Vang etal. INNER MEMBRANE

CYTOPLASM

Rg. 1. Model for the export to the periplasm and secretion to the supernatant of STftg. The 19amino acid signal peptide (Pre) of Pre Pro-STflg is cleaved as the molecule is exported through the inner membrane (1) delivering the 53 aminoacid Pro-STA3 into the periplasm. The tertiary structure determined by the 19 amino-acid ST^j (2a), by the 34-amino-acid Pro domain (2b). or by the interplay of both moieties (2c) fosters an interaction with elements of the secretory machinery. As a result. Pro-STA3 is secreted to the exterior of the cell (3), where the three disulphide bridges are formed (4) and Pro-ST^ is autoproteolytically matured to the 19-residue toxin (5).

OUTER MEMBRANE MEDIUM

PERIPLASM

CD

Pro

STa

Pre-Pro-ST. I

\Pro-Slt,

which the molecule translocates. The same precursor molecule has a proteolytic activity thought to be responsible for the cleavage of the outer membrane-associated form and the final release of the molecule to the external medium. The heat-stable enterotoxins of Escherichia coli, STA and STB. are small polypeptides which are preferentially secreted to the culture supernatant (Martinez-Cadena et at.. 1981; Kupersztoch etat., 1990). Their secretion does not cause the release of proteins normally found in the periplasm, and in contrast to other extracellular proteins, their secretion does not depend upon the presence of additional gene products. In wild-type E. coli, expression of either toxin gene {estA3, an estA allele or estS) is sufficient for that toxin's extracellular delivery (Kupersztoch et ai, 1990; Rasheed ef a/., 1990). Our model for export and secretion of ST^ is illustrated in Fig. 1: the toxin is synthesized as a precursor, Pre-Pro-STft with a 19-amino-acid signal peptide (Rasheed etai, 1990). Following cleavage of the signal peptides, Pro-STft translocates the inner membrane to become periplasmic; this intermediate, subsequently exits the bacteria and is secreted as a 53amino-acid precursor which is cleaved extracellularly (a process termed maturation), to the protease- (Alderete et at., 1988) and heat-resistant ST^. We had previously hypothesized (Guzman-Verduzco et ai, 1983), as shown in Fig. 1, that for Pro-ST^ to exit through the outer membrane, specific interactions had to occur between Pro and/or STA and elements of the secretory machinery. In this communication, we examine the role that the three domains, Pre, Pro, and STA, have in the export (translocation to the periplasm) and secretion (extracellular delivery) of STA3. We show that in the absence of the signal sequence, Pre, the intracellular peptides Pro-ST^s and STAS alone are susceptible to proteolysis and that the Pre domain is required for export.

We also show that both Pre-STA3 and Pre-Pro are first cleaved by signal peptidase and then delivered to the culture supernatant. These results are discussed in relation to the general process of secretion of extracellular peptides. Results Elimination of the STA3 domian It was previously shown and is illustrated in Fig. 1, that Pro-STA3 exits through the E. coli outer membrane (Rasheed et ai, 1990). Some polypeptides are targeted to either the periplasm or to the outer membrane and, it was therefore thought that secretion to the supernatant should be sequence- or conformation-dependent. In the case of STA, as shown in Fig. 1, either Pro, STAS, or regions defined by the interplay of both domains could participate in the interactions that result in the secretion of Pro-STA3 to the culture supernatant. To determine if the STA3 domain, known to be responsible for toxicity, was the determining element in secretion, a T residue was inserted at the junction of Pro and STA3. Plasmid pYK206, harbours this mutation (Table 1; Fig. 2). The resultant frameshift gave rise to three termination codons (TAA, TAG, TAG) immediately after the met-53 codon and should result in a 53-amino-acid polypeptide formed by the 19 amino acids of Pre plus the 34 residues of Pro. When such a mutant was expressed (Fig. 3), an intracellular form was detected with M, 5500 which is close to the theoretical molecular mass (5864 Da) calculated for PrePro. With time this peptide chased first to the periplasm as a species with M, 4400 and then to the supernatant; the calculated molecular weight of the 34-amino-acid Pro domain is 3704. These results suggested that the Pro domain is responsible for secretion from the periplasm to the media.

Secrefion 0/Escherichia coli ST. enterotoxin

3523

Table 1. Mutagenic oligonucleotides and expected amino acid sequences of the mutants. Plasmid/Mutation

Mutagenic oligonucleotides^/Codons

Resulting Sequence''

pYK206'=/

TTG CTA CTA TT® CAT GCT TTC amb amb och M53 S52 E51

AGC ATG TAA TAG TAG S52 M53 och amb amb

Avail A CTC TAC TGG AC^j AGC ATC CTG E26 V25 P24 G23 A22 D21 Q20

Aesf/13 (Pro 23-48)

GCT GGT CCT GAA AGC A22 G49 P50 E51 S52

tyr ATT GCT ACT ATT CAT GA'T ATC CTG N57 S56 S55 N54 M53 152 D21 Q20 AGG TG P17 S16

EcoRV CAG GAT ATC ATG AAT Q20 D21 152 M53 N54

pYK212'/ AesM3 (Pre 2-19)

TGG TTT AGC ATC CTG CAT ATT ACC TCC GGA P24 K23 A22 D21 Q20 Ml -1 -2 -3 -4

ATG CAG GAT GCT AAA Ml R20 D21 A22 K23

PYK213V AesM3 (Pre-Pro 2-53)

GTA ATT GCT ACT AGG CAT ATT ACC TCC GGA Y58 K57 S56 S55 N54 Ml -1 -2 -3 -4

ATG AAT AGT AGC AAT Ml N54 S55 S56 N57

CA

^estA3{Pro22-5^)

GTA Y58 AGC GAA A19 F18

All these plasmids are derived from pT7-3 (Tabor and Richardson, 1985), a ptasmid with the C0IEI replicon that has the bacteriophage T, 010 promoter preceding the multiple cloning site where the mutant DNAs were inserted. The plasmids also code for p-lactamase, a periplasmic enzyme, a. The mutagenic oligonucleotides are shown as the antisense strand in the 5'-3' orientatton. Under the bases, the equivalent sense codons and their positions are shown, b. Nucieotides of the sense strand. The complete mutant genes were sequenced; the codons are shown under the nucleotide sequence, c. The introduction of the T residue shown in the antisense strand as A encircled at position 654 caused a frame shift such that three termination codons follow M53 codon, d. This mutant was obtained in two steps: first, the wild-type bases shown in superscript were replaced by the bases shown in bold phase; the substitutions led to the introduction of an AvaW site (shown above the nucieotides) in position 562, The resulting change (K^jG) did not affect secretion,' The estA3 gene, isolated as part of a 513 EcoRI-H/ndlll fragment, was then digested at both the new and the natural (position 640) AvaW sites. The 78 bp AvaW fragment was eliminated: the 248 bp AvaW-EcoR] and the 187 bp HindU-AvaW fragments were ligated and recioned, e. Tnplet ACC (tyr) was replaced by GAT (llu); after looping out, it yielded the EcoRV site shown above the nucleotide sequence, 1. These mutants were obtained by looping out the DNA region that did not hybridize with the mutagenic oligonucleotide.

Plasmid/ Mutation

IMKKSILglFLSVLSFSPFAl QDAKPVESSKEKITLESKKCNIAKKSNKSGPESM

PYK206/ Aest:A3 (och

PYK210/ AestA3 (Pro 23-48) PYK211/ AestA3 (Pro 22-51) pYK212/ AestA3 (Pre 2-19) pYK213/ &estA3 (PrePro 2-53) Fig. 2. Amino add sequence and deletions ot Pre-Pro-STA3. Deleted regions are shown as empty spaces and the continuous line represents the remaining aminoacids. To illustrate, plasmid pYK212 harbours the deletion of residues 2 to 19 from the Pre region resulting in the fusion of amino acids M, toQ^o,

3524

y. Vanpetal. Supernatant

Cells I

I

PrePro 5'

PrePro 10' 30' 30'

0'

r

5'

10' 30'

Pre-plapla-

Fig. 3. Effect of elimination of the ST domain on the extraceiluiar delivery of Pro, Cultures of a strain lacking any esM3 sequences ((HB101)(pGP1-2)(pT7-3)), iabelied controi. and of an isogenic strain, labelled Pre Pro, harbouring estA3 functionaiiy deleted of the ST^j domain ((HB101 )(pGP1-2)(pYK206]) were pulsed for 1 min and chased for the time indicated. The bacteria were separated from the supernatant and both fractions submitted to SDS-PAGE and fluorography. The bands Iabeiled as Pre-Pro and Pro were absent from the control estA3 strain; in the ceilular fraction, Pre-p-iactamase and Pre-Pro chase to (i-lactamase and to Pro respectively. Pro is converted with time to an extracellular species.

PreProPro-

Deletion of the Pro domain To corroborate the interpretation of the previous section that the Pro region is responsible for the mobilization of Pro-STA to the culture supernatant, we isolated pYK210 (Table 1; Fig. 2), a plasmid which harbours a deletion of 26 of the 34 Pro domain codons (positions 23 to 48). As illustrated in Fig, 2, expression of this mutant should result in a peptide composed of the Pre signal peptide (M, to A,9). the signal peptidase recognition region, and the maturation region (G49 to Y72) which includes the STA3 domain (N54 to Y72). Pulse-chase experiments with this mutant gave rise shortly after the pulse (Fig, 4), to an extracellular polypeptide with an electrophoretic mobility slightly slower than STft3; this is to be expected for an STA3 derivative that includes mature STA3 plus eight additional Pro residues (Q2oD2iA22G49p5oE5iS52M53}. The presence of these eight extra amino-terminal residues was confirmed by Edman degradation (data not shown). As seen in Fig. 4, the extracellular peptide encoded by this mutant appears very fast, while the periplasmic intermediate observed in wild-type estA3 (Rasheed et ai, 1990) and in PYK206, the plasmid that harbours the mutant with deletion of ST^s (Fig. 3) was not detected. The exit of this peptide to the media was so fast that we questioned if its export pathway was the same as that of wild-type toxin. To determine if this mutant used the same SecA-dependent export route as the wild-type, plasmid pYK210 was introduced into a secA temperature-sensitive host (Fig. 5). At S/'^C, the non-permissive temperature, a band corresponding to the expected mobility of Pre-STA3 accumulates within the cells. At the permissive temperature (29°C), as is the case with secA* strains, the Pre-STA3 form is transiently found within the cell; shortly after the pulse an extracellular species was detected.

Similar results were obtained with plasmid pYK211 (Table 1 and Fig. 2) that harbours a deletion of Pro codons 22 to 51 (data not shown). These results indicate that PreSTA3. like the wild-type toxin, is secreted via a SecAdependent pathway. At the non-permissive temperature, Pre-STA3 accumulated in the cell and it was not chased nor degraded even after incubation for 30 min at 37"C (data not shown). Surprisingly, in the absence of the Pro domain, a periplasmic intermediate identified during the secretion of both wild-type and Pre-Pro was not detected. These results indicate that extracellular maturation (Fig. 1, Step 5) does not take place when amino acids 23 to 48 of the Pro domain are deleted and that STA3 alone is dearly able to exit through the outer membrane. In fact, the presence of Pro has the effect of retarding the exit of STA3to the medium.

Supernatant

Cells S'

10' 30'

5'

10' 30'

Pre-|)la-

lila-

ST,

•5T,

Fig. 4. Effect of the elimination of the Pro domain on the secretion of ST^, A strain harbouring a deletion of Pro (Table 1. Fig, 2;(HB101)(pGP1-2)(pYK210) AesM3(pro 23-48)) was pulse-chased as described in the legend of Fig. 3 and fractionated into cells (A) and supernatant (B), The only band present in the sampie denved from the Pro-deletion strain that is absent in the controi strain (not shown) migrates as a M, 3000; this band appears in the supernatant very shortly after the pulse and is barely visible in the cells.

Secre/Zono/Escherichia coli ST. enterotoxin

Pre - STA in sec A pT7-3 29 C

37°C

A PRO 29°C

37"C

0' 2' 5- 0' 2' 5' 0' r 2' 5' 0' V 2' 5' Pre - pla

Pre Fig. 5. Accumulation of Pre-ST^ in a secA ts mutant. Isogenic secA temperature-sensitive strains harbouring a control plasmid iacking estA3 sequences (pT7-3) and the same plasmid with estA3 deleted of the Pro domain (iPro) (pYK210:;esM5(pro 23-48)) were pulse-chased at 2 9 X and 37°C and processed as described in Fig. 3, The cellular fractions were submitted to SDS-PAGE. The band labelled Pre STA3 is absent from the control strain (pT7-3) and is just visible in APro when the experiment was performed at 29°C. the permissive temperature; at 37"C, the non permissive temperature, Pre-p-lactamase and Pro-ST^ remain unprocessed after 5 min of chase.

Deletion of Pre The data shown in the previous section demonstrate that Pre-STA3 or Pre-Pro could be delivered to the culture supernatant after Pre is removed. The unexpected observations that both the ST^g and Pro peptides individually can exit the cell suggested that small peptides of very different primary structures can translocate the inner and outer membranes. To determine if translocation of the inner membrane could be achieved by these small peptides in the absence of a signal sequence, Pre was deleted from Pre-Pro-ST^s (Table 1 and Fig. 2, pYK212). In the SDS-polyacrylamide gel electrophoresis (PAGE) analysis of the peptides synthesized by this mutant (Fig. 6) in the cellular fraction, a band of the expected size is detected during the pulse. Shortly after the chase, this band becomes undetectable. Analysis of the supernatant traction did not reveal the presence of any Pro-STA3 or STA3 peptides. These observations could be expfained if the small peptide was secreted in the absence of a signal peptide but that the detection methods used were inadequate; alternatively, the cell-associated forms could be susceptible to proteolysis. As shown in Fig. 7, when similar experiments were performed using a Ion" mutant (a mutation in the intracellular Ion protease gene), the peptide was stabilized within the cell; however, even under these conditions, neither periplasmic nor extracellular forms were detected. Furthermore, when cultures of a lon~ strain containing pYK212, (the plasmids that harbours the deletion of Pre; Fig. 2 and Table 1), were sonically disrupted and diluted back to the original culture volume, cell-associated Pro-ST^g was recovered using the

3525

same methodology that previously did not show extracellular Pro-STfts (data not shown). The expression of this mutant (deletion of signal peptide) in a secA ts host at the non-permissive temperature resulted in a very short-lived Pro-STfta, similar to that observed in a wild-type secA* strain (data not shown). These results suggest that the presence of Pre, when accumulated as Pre-Pro-ST^s or as Pre-STA3 (Fig. 5) in a secA ts strain, somehow causes the peptides to be less susceptible to proteolytic degradation. Without a signal peptide, the 53-amino-acid ProSTA3 polypeptide cannot translocate the inner membrane and is very susceptible to proteolysis regardless of the SecA background of the host strain

Deletion of Pre-Pro The inability of the 53-amino-acid Pro-ST^s to translocate the inner membrane could be due solely to its molecular size as the Pro and ST^g domains separately are able to cross the outer membrane. Therefore, the DNA region encoding Pre and Pro (codons 2 to 54) were deleted (Table 1 and Fig. 2, pYK213); the resulting open reading frame encoded only for the 19 amino acids of STA3 preceded by the initial methionine codon. Analysis of the expression of this mutation yielded results similar to those obtained when only the Pre signal peptide region was deleted (in plasmid pYK212). Intracellular STA3 is quickly degraded in either a secA* or secA~ background; its longevity is enhanced in a ton' background; and ST^a is not exported into the periplasm or secreted into the extracellular media (data not shown).

Pro - S T A in /on HpT7-3

0'

2'

5' 10' 20'

A Pre

0'

2'

5' 10' 20'

Pre pia

Fig. 6. Effect of the elimination of the signai peptide on the secretion and stability of Pro-STA, Cultures of HB101(pGP1-2)(pYK212) harbounng a deletion of the Pre domain and of an isogenic control strain harbouring the expression system devoid of any insertion, (HBi0i(pGP1-2](pT7-3)) were pulse-labelled for 30 s and chased for the ttmes indicated. The samples were processed as descnbed in the legend to Fig, 3, SDS-PAGE of the celluiar fraction showed that the band associated with the estA3 insert is barely detectable after 5 min of chase. No ST^ related band was detected in the culture supernatant (not shown).

3526

y. Vang etal. Pro -

in Ion -

STA

Cells I

Supernatant I I

0' 2'

5- 10- 20'

0' 2- 5'

I 10' 20'

Pre- pia

Fig. 7. Stabilization of Pro-STfl, in a ion background, A similar experiment to that performed in Fig 6 using strain Y1089. a Ion' protease mutant, allowed the detection of a cell-associated band with the electrophoretic migration expected for Pro-ST^ for at least 20 min after the chase. Over.exposure of the gel failed to unveil any bands in the super, natant with eiectrophoretic mobilities near that expected for Pro-STA, ST, and Pro,

Discussion The heat-stable enterotoxin of E. coli is synthesized as Pre-Pro-ST^a, a 72-amino-acid precursor, which is converted to periplasmic Pro-ST^s and then secreted to the culture supernatant where it matures to the protease and heat-resistant active STA3 toxin (Rasheed et ai, 1990). Previously we investigated if proteins normally residing in the periplasm could be mobilized to the extracellular media when their signal peptide was substituted by the Pre-Pro-STA3 polypeptide. When such a fusion was done with the 11 500 Da periplasmic LTg {the B subunit of the heat-labile enterotoxin), the fusion products remained cell-associated (Guzman-Verduzco and Kupersztoch, 1987; 1990) and were not secreted to the media (Kupersztoch et ai, 1990). Similarly Okamoto and Takahara (1990) reported that there was lack of secretion when either the first 53 amino acids (Pre-Pro) or the first 21 amino acids (Pre) of estA1 were fused to the 149 amino acids of staphylococcal nuclease A. This Gram-positive extracellular nuclease, however, was delivered to the periplasm of E. coti when fused to the OmpA signal peptide (Takahara etat., 1985). The heat-stable toxin derived Pre-Pro-nuclease and Pre-nuclease fusions resulted, in both cases, in nuclease activity localized in the periplasm. It is possible that one of the many periplasmic proteases could separate Pro from nuclease and that Pro could translocate the outer membrane. Nevertheless, under the above conditions, the Pro sequence did not guide periplasmic proteins to the extracellular milieu. Thus, these experiments suggested that there is a conformational requirement for outer membrane translocation, but they did not directly address whether Pro or ST^a by themselves can reach the culture supernatant.

We had previously hypothesized that Pro was responsible for Pro-STA3 secretion (Guzman-Verduzco ef ai, 1983). To test this model, a mutant lacking the ST^g region (Pre-Pro) was isolated (pYK206, estAS (ooh54); Table 1 and Fig. 2) and examined. Our results (Fig. 3) seem to support this role of Pro in the secretion of ProSTA3. Further indirect backing of this rote of Pro in the exit of Pro-STA3 through the outer membrane comes from the unpublished results of Y. Yang and Y, M. Kupersztoch in which mutations of 10 different residues of STA3 did not affect the extracellular delivery of the mutant peptides. It was further reasoned that in the absence of Pro, Pre-STA3 would result in periplasmic STA3 and that STA3 by itself would not be able to translocate the outer membrane. However, when such a mutant was analysed (Fig. 4), STA3 was found in the culture supernatant very shortly after the pulse and the cell-associated STA3 was not detected. Furthermore, under conditions that favour the detection of periplasmic Pro-STA3 — 29°C incubation of wild-type STA cultures with CCCP (carbonyl cyanide mchlorophenylhydrazone) and then inactivating the CCCP with 2-mercaptoethanol (Wolfe and Wickner, 1984) — no cell-associated STA3 was detected when Pro deletion mutants were expressed (data not shown). Similar results were obtained by Okamoto and Takahara (1990) when estA1 Pro codons 22 to 53 were deleted. Forty per cent of the wild-type toxin level was found extracellularty and no STA activity was found in the cellular extracts. In other experiments, when the STB signal peptide was fused to STA3 and set under the control of the inducible alkaline phosphatase promoter (Greenberg et ai, 1991), a high level of extracellular toxin (4mgmr'') was produced. From these three experiments, contrary to our original hypothesis, it is clear that in the absence of Pro, toxic and radiolabeiled forms of STAS are secreted to the culture supernatant. Based on these observations, we now propose that Pro is the domain responsible for the natural controlled secretion of the wild-type toxin. When Pro-STA3 reaches the periplasm it binds and translocates the outer membrane; in the absence of Pro, STA3 reaches the extracellular media by an alternative route, without or with a very shortlived periplasmic intermediate. Presumably the evolution of estA has selected two alternative mechanisms that guarantee the extracellular delivery of the toxin. The secretion of the 19-amino-acid STA3 could be explained if small polypeptides, solely because of their size, could reach the culture supernatant. p-Endorphine, a 43amino-acid peptide was delivered to the culture supernatant when its gene was fused to the ompF signal sequence and the fusion was expressed in E. coli (Nagahari et ai, 1985). Similarly, when human pancreatic trypsin inhibitor was fused to the OmpA gene signal peptide, the biologically active and properly processed

Secretion of Escherichia coli ST^ enterotoxin products (from 56 to 63 amino acids) were found in the culture medium (Maywald et ai, 1988). In contrast, the fusions of PhoA signal peptide to human epidermal growth factor (53 amino acids long; Oka etai, 1985) and of LamB signal peptide to bovine somatotropin (14 amino acids long; Klein et ai, 1991) as well as the expression of proinsulin (84 amino acids; Talmadge et ai, 1980) remain periplasmic. Therefore, it appears that small peptides do not exit the outer membrane owing to their size but require a yet unidentified element to reach the extracellular media. The Pro domains by itself can be secreted; but when fused to periplasmic peptides (like LTb, about 11 500 Da, or staphylococcal nuclease, about 17000 Da), the fusion products were not secreted. When present as Pro-STA3, the peptide appears to increase its periplasmic residence time. At present, it is not understood what conformation and/or signals (or lack of them) trigger the extracellular delivery of polypeptides. Having shown that Pro-STA3, Pro and STA3 translocate the outer membrane, the question arose as to whether the signal sequence is necessary for the translocation of these small peptides through the inner membrane. As seen in Fig. 6, the elimination of Pre and the deletion of Pre-Pro caused the mutant peptides to remain susceptible to proteolysis within the cells. The peptides showed more stability in a lon~ mutant (Fig. 6) than in wild-type lon^ strains. The instability of exportable peptides when they lack the signal sequence is contrasted with the apparent resistance to degradation of intracellular PrePro-STA3 which accumulates in a secA"" mutant. When deletions of the signal peptide were expressed in SecA mutants, they were as unstable as in secA* strains. Similar observations of stability in secA mutants have been reported for Pro-STe (Kupersztoch et ai. 1990). At the present time, it is unknown if the relative long life of PrePro-STA3 in a secA~ background is because the signal peptide fosters interactions with chaperonins (such as SecB, DnaK, GroEL; reviewed by Ellis and van der Vies, 1991), and this event protects STA from proteolysis. Alternatively, the membrane-cytoplasm cycling of SecA, found defective in secAts^^ (Cabelli et ai, 1991) could cause Pre-Pro-STA3 to be preferentially membrane-bound, and, therefore, inaccessible to proteases. These alternatives are being examined. In conclusion, deletion of the Pre signal peptide from estA3 causes the resulting molecules of Pro-STA3 to remain intracellular; intracellular Pro-ST^a or STA3 by itself are very susceptible to proteolysis. The Pro domain, when preceded by the signal peptide, is first found intracellularly and then it is exported to the periplasm. This form either as Pro alone, or as Pro-STA3, is then secreted to the culture supernatant. Deletion of the Pro domain (i.e., Pre-ST^s) causes a very fast delivery of STAsto the extracellular milieu. Therefore, it appears that Pro alone

3527

can be secreted, but when present as Pro-STA3, it retards the exit of STA3. Pre-STA3 is converted into STA3 and

secreted via a SecA- and energy-dependent pathway with either a very short, or no, periplasmic residence. Under the hyperproduction conditions used, both Pro and STA3 can translocate the inner and outer membranes when preceded by a signal peptide. However, the artificially high level of estA3 peptides produced by the expression system used, could possibly cause abnormal peptide localizations even though the fractionation markers (5-lactamase and its precursor were, as expected, cytoplasmic and periplasmic, respectively. Therefore, experiments are in progress to determine the localization of ST-related peptides synthesized from the natural estA3 promoter.

Experimental procedures E. coli strains, plasmids and media Strain HB101 was used to express the wild-type and the estA3 mutants (Boyer and Roulland-Dussoix, 1969), The effect of secA on the export of estA3 products was tested in strains MC4100 and a secA temperature-sensitive derivative, MM52 (Oliver and Beckwith, 1981). Strain TGI (Sambrook ef a/., 1989) was used to propagate male-specific single stranded phages. Strains CJ236 (Kunkel etai, 1987) and MV1190 were used to obtain uracil-containing DNA tor the isolation of sitedirected mutations as described by Kunkel etat. (1987), A lon'^ mutant, strain Y1089 (Young and Davies, 1983), was used to determine the effect of Lon protease on the stability of estA3 mutants. The T7 expression system composed of plasmids pGP1-2 and pT7-3 has been described (Tabor and Richardson, 1985) and was used to radiotabel preferentially wild-type and mutant estA3 products as well as p-lactamase. pGP1-2 (p15A replicon) is kanamycin resistant (Km'^) and codes for C|867. the phage X temperature-sensitive repressor. At 42''C, it is inactivated and the host RNA polymerase transcribes high levels of T7 RNA polymerase from the PL promoter, pT7-3 (ColEI replicon) is ampicillin resistant (Amp") and has a polycloning site immediately downstream from 010, the T7 promoter. The mutant and wild-type estA3were cloned into pT7-3 and expressed using this system. The Muta-Gene kit (BioRad Laboratories) was used to clone, mutagenize and sequence estA3. The origins of the estA3 gene and other plasmids employed have been previously described (Stieglitz et at., 1988). Low-sulphate C media was used when peptides were radioactively labelled (Guzman-Verduzco and Kupersztoch, 1987), When indicated, L media was supplemented with ampicillin (50 ^g ml"'') or kanamycin (50 ng ml"'').

Otigonucteotide-directed mutagenesis Site-directed mutagenesis was performed either by the double primer method of Zoller and Smith (1984) or with the MutaGene mutagenesis kit (BioRad Laboratories) which is based on the selective degradation of uracil-containing M13MP18 DNA (Kunkel et ai, 1987), Oligonucleotides for use in mutagenesis

3528

Y. Yang et al.

were synthesized on an Applied Biosystems 380B DNA synthesizer in the Biochemistry Department of the University of Texas Southwestern Medical Center at Dallas. When the double-primer mutagenesis method was employed, the second non-muiagenic primer was the Ml 3 universal sequencing primer. The mutagenic procedures and the resulting mutants are summarized in Table 1 and represented in Fig, 2, Deletion mutants pYK212 AesM3 (Pre 2-19), pYK213 AesM5 (Pre-Pro 2-53), and pYK211 A(Pro 22-51) were generated using the Muta-Gene system by looping out the specific regions of the wild-type sequence (Gil e/a/,, 1985), Mutants pYK210 (AesM3 (Pro 23-48)) and pYK206 {estA3 (och54)) were obtained by the double primer method (Zolter and Smith, 1984), To isolate AesM3 (Pro 23-48), an AvaW site was introduced at nucleotide 562 (Stieglitz etai, 1988) using the oligonucteotide shown in Table 1, The mutant estA3 DNA containing the new AvaW site was isolated as a 513 bp £coRI-H;ndlll fragment; upon AvaW digestion the 78 bp ^vall fragment was discarded and the 248 bp EcoV{\-AvaW fragment was ligated to the 187 bp AvaWH/ndlll fragment and to pT7-3 digested with EcoRI-H/ndlll. To isolate pYK206 {estA3 (och^^)), a T residue was inserted at position 654 (Stieglitz ef at., 1988); this insertion causes three termination codons to be in frame after M53 (Table 1).

modifying the method described by Rasheed et at. (1988). In brief, the concentration of Tris was increased from 0,375 to 0,75 M Tris, pH 8.8 as described by Fling and Gregerson (1986), and a linear polyacrylamide gradient (12 to 18%) containing 45% glycerol in the 18% acrylamide solution was used. The stacking gel was 4% polyacrylamide containing 0,1% SDS in 0,12 M Tris, pH 6.8. The running buffer was 0.05 M Tris, 0.192 M giycine and 0 . 1 % SDS, Electrophoresis was performed overnight at 100 V at 25°C in a Protean II slab gel chamber (BioRad) or for 1 h 45 min, 4 X at 200 V using a miniProtean II dual slab cell. After electrophoresis, the unfixed gels were directly prepared for fluorography (Rasheed etat.. 1990).

Expression of the mutations and their analysis

References

The method used is modified from that described previously (Rasheed et ai. 1990), The strains of interest were grown separately at 29"C in C broth to a density of about 2x10^ cells ml"''. The cells were concentrated by centrifugation, resuspended in 1/3 the original volume of C broth, incubated at 29''C for 10 min, and then shifted to 42''C for 15 min, Rifampicin (200ngmr') was added, and incubation continued for an additional 10 min and the suspensions were then shifted to 37^C for 10 min. The peptides synthesized under the T7 expression system were radiolabelied by incubation with [^^S]-cysteine (15 nCi mt ^) and, when indicated, also with [^^S]-methionine (15 |jCi m l ' ) . The pulses were terminated by the addition of an equal volume of C medium supplemented with non-radioactive cysteine and methionine (8 mg ml"' each); the samples were then chased for the indicated time periods. After chasing, onehalf volume of ice-cold stop solution (containing 0.04% chloramphenicol, 0.4 M sodium azide and 0.02 M 2-4-dinitrophenol) was added. Cells were separated from supernatant by centrifugation, and 1 % 2-mercaptoethanol was added to the culture supernatant which was then incubated at 25''C for 30 min. The supernatant was applied fo a C-18 reverse-phase chromatography cartridge (Sep-Pak, Waters Chromatography) and eiuted as described previously (Rasheed et at., 1990) by washing the cartridge in a step-wise fashion with 4 ml each of 10%, 20%, 30% and 40% methanol, ST^s and related peptides were eiuted with 3 ml of 50% methanol. When necessary, rifampicin was removed from the solution by extracting it with an equal volume of chloroform. The aqueous phase was removed and dried in a speed vac concentrator (Savant Instruments), The dried supernatant and the cellular pellet were resuspended in 60 mM Tris-HCI, pH 8,100 mM DTT, 2% SDS, 10% glycerol and 0,01% bromophenol blue, boiled for 2 min and fractionated by SDS-PAGE, Peptides were separated on discontinuous SDS-polyacrylamide gels (10:1 acrylamide/A/,W' diallyltartardiamide)

Acknowledgements We thank Heather Stieglitz. Lawrence Dreyfus and Gary Coombs for critically reading this manuscript and Cindy Patterson for typing it. This work was supported in part by Public Health Service Grant PO1-HD22766 and by the Advanced Research Program of the Texas Higher Education Board, We thank Kathryn Strauch and Don Oliver for bacterial strains.

Alderete, J.F,, and Robertson, D.C, (1978) Purification and chemical characterization of the heat-stable enterotoxin produced by porcine strains of enterotoxigenic Escherichia coti. infect tmmun A9: ^02^-^030. Boyer, H,W,, and Roulland-Dussoix. D, (1969) A complementation analysis of the restriction and modification of DNA in Escherichia coti. J Mot Biol 4 1 : 459-472, Cabetli, R.J., Dolan, K,M., Qian, L,, and Oliver, D.B. (1991) Characterization of membrane-associated and soluble states of secA protein from wild-type and secA51 (ts) mutant strains of E. coti. J Biot Chem 36: 24420-24427, DeGraaf, F,K,, and Oudega, B, (1986) Production and release of cloacin DF13 and related colicins, Curr Top Microbiot /mmuno/125: 183-205, Ellis, R,J. and van der Vies, S,M, (1991) Molecular chaperones. Annu Hev Biochem 60: 321-347, Felmlee, T., Pellett, S., Lee, E.-Y., and Welch, R, (1985) Escherichia co'/hemolysin is released extracellularly without cleavage of a signal peptide, J Bacteriot 163: 88-93, Fling, S,P,, and Gregerson, D.S, (1986) Peptide and protein molecular weight determination by electrophoresis using a high-molarity Tris buffer system without urea. Anat Biochem 155: 83-88, Gariepy, J., Judd, A,K,, and Schoolnik, G,K, (1987) Importance of disulfide bridges in the structure and activity of Escherichia coti enterotoxin ST1 b, Proc Natt Acad Sci USA 84:8907-8911, Gil, G,, Faust, D.J., Chin, D.J,, Goldstein, J.L,, and Brown, M,J, (1985) Membrane-bound domain of HMG CoA reductase is required for sterol-enhanced degradation of the enzyme, Ce//41: 249-258. Gray, L., Baker, K., Kenny, B,, Mackman, N., Haigh, R,, and Holland, I.B. (1989) A novel C-terminal signal sequence targets Escherichia coti haemolysin directly to the medium. J Cett Sci SupptM: 45-75. Greenberg, R,N,, Ping, Z., Biek, D.P., and Mann, D.M. (1991) High level expression and secretion of a lysine-containing

Secretion of Escherichia coli STA enterotoxin analog of Escherichia coti heat-stable enterotoxin. Protein Expression Purifl: 3 9 4 ^ 0 1 . Guzman-Verduzco, L,M., and Kupersztoch, Y,M, (1987) Fusion of Escherichia coti heat-stable enterotoxin and heatlabile enterotoxin B subunit, JSac/er/o/169: 5201-5208, Guzman-Verduzco, L.M., and Kupersztoch, Y,M, (1990) Export and processing analysis of a fusion between the extracellular heat-stable enterotoxin and the periplasmic B subunit of the heat-labile enterotoxin in Escherichia coli. Mol Microbiot 4:253-264, Guzman-Verduzco, L.M,, Fonseca, R., and Kupersztoch-Portnoy, Y,M, (1983) Thermoactivation of a periplasmic heatstable enterotoxin of Escherichia coti. J Bacteriot 154: 146-151. Holland, I.B,, Kenny, B., and Blight, M. (1990) Haemolysin secretion from E. coti. Biochimie72:131-141, Klein, B,K., S,R, Hill, C,S, Devine, E, Rowold, C E , Smith, S. Galosy, and P,0, Olins. (1991) Secretion of active bovine somatotropin in Escherichia coti. Bio/Technotogy 9: 669-872, Kornacker, M,G,, D, Faucher, and A,P. Pugsley, (1991) Outer membrane translocation of the extracellular enzyme pullulanese in Escherichia coti Kt2 does not require a fatty acylated N-terminal cysteine, J Biot Chem 266: 13842-13848, Kunkel, T,A,, Roberts, J.D., and Zakour, R.A, (1987) Rapid and efficient site-specific mutagenesis without phenotype secretion, Meth Enzymot ^5A•. 367-382, Kupersztoch, Y,M,, Tachias, K., Moomaw, CR,, Dreyfus, L,A,, Robert, U,, Slaughter, C , and Whipp. S. (1990) Secretion of methanol insoluble heat-stable enterotoxin (STb): Energy and secA dependent conversion of Pre-STb to an intermediate indistinguishable from the extracellular toxin, J Bacteriot 172:2427-2432, Kupersztoch, Y,M,, Powell, F,E,, and Guzman-Verduzco, L,M, (1990) Conditional lysis of Escherichia coti by the fusion of extracellular (STa) to periplasmic (LTb) enterotoxins: Apparent phenotypic suppression of lactose permease, Curr Microbiot 20: 3^-37. Mackman, N,. Nicaud. J,-M,, Gray, L., and Holland, I.B. (1986) Secretion of haemolysin by Escherichia coti. Curr Top Microbiot tmmunot 125: 159-181. Martinez-Cadena, M.G,, Guzman-Verduzco, L,M., Stieglitz, H., and Kupersztoch-Portnoy, Y,M, (1981) Catabotite repression of the heat-stable enterotoxin activity of E. coti. J Bacteno/145: 722-728, Maywald, F., T, Bbldicke, G, Gross, R, Frank, H. Blocker, A, Meyerhans, K, Schwellnus, J. Ebbers, W, Bruns, G, Reinhardt, E, Schnabel, W, Schroder, H, Fritz, and J._Collins, (1988) Human pancreatic secretory trypsin Inhibitor (PSTI) produced in active form and secreted from Escherichia coli. Gene 6B: 357-369. Nagahan, K,, Kanaya, S,, Munakata, K,, Aoyagi, Y,, and Mizushima, S, (1985) Secretion into the culture medium of a foreign gene product from Escherichia coti: use of the ompF gene for secretion of human beta-endorphin, EMBO J 4: 3589-3592. Oka, T,, S, Sakamoto, K,-l, Miyoshi, T. Fuwa, K, Yoda, M, Yamasaki, G, Tamura, and T. Miyake. (1985) Synthesis and secretion of human epidermal growth factor by Escherichia coti. Proc Natt Acad Sci USA B2: 7212-7216.

3529

Okamoto, K,, and Takahara, M, (1990) Synthesis of Escherichia coli heat-stable enterotoxin STp as a pre-pro form and role of the pro sequence in secretion, J Bacteriol 172:5260-5265, Oliver, D,. and Beckwith, J, (1981) E. co//mutant pleiotropically defective in the export of secreted proteins, Cett 25: 765-772, Oudega, B,, Ykema, A,, Keina, A,Y,, Stegehuis, F,, and DeGraaf, F,K, (1984) Detection and subcellular localization of mature protein H involved in excretion of cloacin DF13. FEMS Microbiot Lett 22: 101-108, Pohlner, J,, Halter, R,, Beyreuther, K,, and Meyer, T,F, (1987) Gene structure and extracellular secretion of Neisseria gonorrhoeae IgA protease. Nature 325: 458-^462, Pugsley, A,P,, D'Enfert, C , Reyss, I,, and Kornacker, M.G. (1990) Genetics of extracellular protein secretion by gramnegative bacteria- Ann Rev Genef 24: 67-90, Rasheed, JK,, Guzman-Verduzco, L,M,, and Kupersztoch, Y,M, (1988) Hyperproduction of heat-stable enterotoxin (STa4) of E. coti An analysis of the unusual electrophoretic behavior of reduced and alkylated forms of STas, Microbiat Pathogen 5: 333-343. Rasheed, K,, Guzman-Verduzco, L,M,, and Kupersztoch, Y,M, (1990) Two precursors ot the heat-stable enterotoxin of Escherichia coti: evidence of extracellular processing, Mol Microbiol^: 265-274, Sambrook. J,, Frisch, E,F,, and Maniatis, T, (1989) Motecutar ctoning: A Laboratory Manuat. Cold Spring Harbor Laboratory, New York: Cold Spring Harbor Laboratory Press. Stieglitz, H,, Cervantes, L,. Robledo. R,, Fonseca, R,, Covarrubias, L., Bolivar, F,, and Kupersztoch, Y,M, (1988) Cloning, sequencing and expression in Ficoll generated minicells of an E. co//heat-stable enterotoxin gene, Ptasmid20: 42-53, Tabor, S,, and Richardson, C C , (1985) A bacteriophage T7 polymerase/promoter system for controlled exclusive expression of specific genes. Proc Natt Acad Set USA 82: 1074-1078. Takahara, M,, Hibler, D.W,, Barr. P,J,, Gerit, J.A,, and Inouye, M, (1985) The OmpA signal peptide directed secretion of staphylococcal nuclease A by Escherichia coti. J Biot Chem 260:2670-2674, Talmadge, K,, J, Kaufman, and W, Gilbert. (1980) Bacteria mature preproinsulin to proinsulin. Proc Natt Acad Sci USA 77: 3988-3992, Wagner, W,, Vogel, M,, and Goebel, W, (1983) Transport of hemolysin across the outer membrane of Escherichia coli requires two functions, J Bacteriot \SA: 200-210, Wolf, P,B,, and Wickner, W. (1984) Bacterial leader peptidase, a membrane protein without a leader peptide, uses the same export pathway as pre-secretory proteins. Cell 36: 1067-1072, Young, R,A,, and Davis, R.W- (1983b) Yeast RNA polymerase II genes: Isolation with antibody probes. Science 222: 778-782, Zoller, M,J,, and Smith, M, (1984) Oligonucleotide-directed mutagenesis: A simple method using two oligonucleotide primers and a single-stranded DNA template. DNA 3: 479-488.