JOURNAL OF BACTERIOLOGY, June 1992, p. 3423-34280021-9193/92/113423-06$02.00/0Copyright © 1992, American Society for Microbiology

Vol. 174, No. 11

MINIREVIEW

Determinants of Extracellular Protein Secretion inGram-Negative Bacteria

STEPHEN LORY

Department ofMicrobiology, School of Medicine, University of Washington,Seattle, Washington 98195

INTRODUCTION

Protein localization has been the subject of extensiveresearch using models ranging from bacteria to mammaliancells. The central question which prokaryotic- and eukary-otic-cell biologists have attempted to answer is how proteinssynthesized in the cytoplasmic compartment traverse thehydrophobic membrane barrier or become integrated into it.This work has led to the identification of targeting signalswithin secreted proteins and of a secretion machinery (49,56).The release of proteins from gram-negative bacteria has

recently been recognized as an important biological processwhich is particularly common among human and plantpathogens. The export pathway of a protein destined to bereleased from a cell starts in the cytoplasmic compartmentand crosses the inner membrane, periplasm, and outermembrane. Because all of these compartments contain anormal array of compartment-specific proteins, correct sort-ing is of utmost importance. This not only assures that theexported proteins are targeted to the extracellular space butalso prevents their retention in any one of the compartmentsof the cell envelope.One significant finding to have emerged from approxi-

mately two decades of study of bacterial protein export is thesurprising uniformity in the overall secretion mechanisms forboth membrane and periplasmic proteins. In contrast, themechanisms that are responsible for extracellular proteinlocalization in gram-negative bacteria are much more di-verse. This review summarizes the recent advances in un-derstanding the process of extracellular localization of pro-teins by gram-negative bacteria. The key features of theextracellular protein secretion pathways discussed here aresummarized in Table 1. The major emphasis will be onmechanisms that lead to export of soluble enzymes frombacteria without affecting the integrity of the cell envelope.Several reviews on extracellular protein secretion haveappeared in recent years (19, 20, 46, 53), wherein the readermay find a more detailed discussion of export systems notcovered here.

EXTRACELLULAR SECRETION OF PROTEINSSYNTHESIZED WITH SIGNAL SEQUENCES

The N-terminal signal sequences found in precursors ofproteins destined for extracellular localization resemblethose of membrane and periplasmic proteins. Release ofsuch proteins from the gram-negative bacterium has beenassumed to proceed via a two-stage process. In the firststage, the N-terminal signal sequence is thought to direct theproteins into the maJor secretion pathway, where they insert

into or are secreted across the inner membrane by theirinteraction with the products of the sec (prl) genes. Thesignal peptide is then cleaved off by the leader peptidase.Direct proof of the involvement of the general secretionmachinery in this first stage of extracellular protein secre-tion, however, is lacking, since sec (prl) mutants of only afew bacteria other than Escherichia coli have been isolated.In the second stage of the export process, the protein isreleased from the cell. This either is self-promoted or re-quires participation of an apparatus of extracellular secre-tion.The specializing sorting process needed for extracellular

protein secretion relies on the recognition of informationencoded within the exported polypeptide. This encodedinformation is absent from proteins that remain in thecytoplasm. Extracellularly secreted proteins which enter theexport pathway via their signal sequences contain additionaldeterminants that assure their external localization. Thereare no obvious differences in signal sequences on the aminotermini of precursors of extracellularly secreted proteins,which might differentiate them from those that remain in theperiplasm or the outer membrane. Therefore, determinantsof extracellular localization must be part of the maturepolypeptide. They can, however, be located in any part ofthe polypeptide chain.

Self-promoted extracellular secretion: immunoglobulin Aprotease family. Several bacterial proteases, such as thosemade by gram-negative pathogenic species of Neisseria,Haemophilus, and Serratia, are exported from the cell via amechanism which appears to be self-promoted and does notinvolve a specialized machinery for extracellular secretion.The Neisseria gonorrhoeae immunoglobulin A protease andthe serine protease of Serratia marcescens are synthesizedas large precursors in which the domain that will become themature enzyme is flanked by a typical secretion signalsequence at the amino terminus and a large helper domain atthe carboxy terminus (44, 57). Both of these regions partic-ipate in the export process. The signal sequence is respon-sible for interacting with the normal secretion apparatus ofthe cell and for traversal across the inner membrane, whilethe C-terminal peptide directs the protease across the outermembrane, where it remains embedded via the same C-ter-minal peptide (44). The enzyme is then released from thebacterial cell by autocatalytic proteolysis, leaving the helperpeptide embedded in the outer membrane.The role of the C-terminal helper peptide in facilitating the

extracellular secretion of the catalytic portion of the prote-ases is not known. The C-terminal domain, when fused topassenger proteins, can promote their transit across theouter membrane, leading to surface exposure (27). These

3423

on Septem

ber 3, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3424 MINIREVIEW

TABLE 1. Characteristics of major protein excretion systems in gram-negative bacteriaa

Presence of:Prototype Other possible members ofrototypeN-terminal Excretion Periplasmic prototype family

signal sequence machinery intermediate

N. gonorrhoeae IgAb protease Yes No S. marcescens serine protease, other IgAproteases, P. solanacearum endoglucanase

K oxytoca pullulanase Yes Yes Yes Exoenzymes of P. aeruginosa, X. campestnis,E. chrysanthemi

E. coli hemolysin No Yes No RTX cytotoxins, B. pertussis adenylate cyclase,colicin V, E. chrysanthemi metalloproteases

a Y. enterocolitica Yops and multisubunit toxins cannot be assigned to a prototype family at this time.b IgA, immunoglobulin A.

studies were carried out by using a bacterial species (E. coli)which would not normally release the passenger protein intothe surrounding medium. This result therefore suggests thatthe C-terminal transit peptide either actively participates inthe extracellular secretion process or serves as a signal forcellular machinery found in most bacteria. A wide range ofproteins could be exported by this mechanism, provided thatthey reach the outer surface in a form which allows theirrelease by the severing of the membrane-associated anchorfrom the exposed portion of the protein. If the extracellularlysecreted protein is a protease, the enzyme may mediate itsown release. However, the release need not be autocata-lytic, since forms of the Serratia serine protease that aremutant in the catalytic site of the enzyme can be efficientlyreleased from E. coli. The cleavage of the C-terminal peptideat an alternative site is presumably carried out by anotherenzyme (37).A potential variant of the mechanism exemplified by the

Neisseria and Serratia proteases has been found in Pseudo-monas solanacearum; in this mechanism, an extracellularendoglucanase is extracellularly secreted in a two-step pro-cess. A precursor form of this enzyme is acylated, its signalsequence is removed, and it is transported to the surface.There it remains bound via the outer-membrane-associatedlipid anchor. The release of the enzyme from the cell resultsfrom proteolytic cleavage at a site between the matureendoglucanase and a short peptide which contains the lipidanchor (23). Proteases capable of releasing surface-exposedproteins may be quite prevalent.

Multisubunit toxins. A number of multisubunit bacterialtoxins, such as cholera toxin and pertussis toxin, are activelyexported from their respective gram-negative cells. Thecompositions of the holotoxins vary: cholera toxin consistsof one A subunit and five identical B subunits (10), whilepertussis toxin is composed of six subunits, five of which arenonidentical (52). The subunits for cholera and pertussistoxins are synthesized with normal secretion signal se-quences. Release from the cell is preceded by assembly ofthe subunits into the holotoxin in the periplasm (17). Thesubsequent translocation of the toxin across the outer mem-brane is directed entirely by the assembled B oligomers,since mutants unable to synthesize A subunits release sub-stantial amounts of the B oligomers, while in the absence offunctional B subunits, there is no detectable export of the Asubunits (2, 18, 43). The release of the holotoxin, or of the Boligomers alone, is not self-promoted, and it very likelyrequires some cell machinery.Evidence for involvement of a cell-specified apparatus is

based on studies of two closely related enterotoxins thatdiffer in their localization, depending on the bacterium

producing them. The E. coli heat-labile enterotoxin is se-creted into the periplasm, while the highly homologouscholera toxin is released into the surrounding media byVibrio cholerae. When cholera toxin genes are expressed inE. coli, the toxin is efficiently assembled but remains withinthe periplasm (42). In contrast, the introduction of E. coliheat-labile enterotoxin genes into V. cholerae leads to theirefficient export into the medium, suggesting that E. coli doesnot have the required machinery of extracellular secretionfor these toxins (18, 38). Evidence that such machineryexists has been provided by the isolation of export-defectivemutants of V. cholerae (21), although they remain unchar-acterized.

Pullulanase family. One of the earliest described proteinsexported from a gram-negative bacterium was an extracel-lular lipoprotein, pullulanase, produced by Klebsiella oxy-toca (46). This protein is acylated and transferred across theouter membrane, where it is anchored via the fatty acidsattached to the N-terminal cysteine of the mature polypep-tide. Following this surface exposure, a substantial fractionof the enzyme is released from the cell as a multimericcomplex. E. coli carrying clones of the pullulanase genecould extracellularly secrete the enzyme, provided thatadditional genes flanking the pullulanase structural genewere coexpressed. This led to the subsequent identificationof a cluster of 12 genes, all necessary for complete export ofpullulanase. Recent studies have shown that a similar appa-ratus of extracellular secretion is present in a wide variety ofgram-negative bacteria which actively export proteins, in-cluding Pseudomonas aernginosa, Erwinia chrysanthemi,and Xanthomonas campestris (3, 7, 9, 15, 40, 41). None ofthe proteins released by bacteria other than K oxytoca arelipoproteins; therefore, a surface-exposed intermediate maybe unique to pullulanase.Most of the genes encoding the proteins of the extracellu-

lar secretion apparatus ofK oxytoca have been sequenced,their products have been identified, and their localization inthe cytoplasm or cell envelope has been determined. Four ofthese proteins, PulG, PulH, PulI, and PulJ, share significanthomology, at their amino termini, with type IV pilins (40).The product of the last gene of the pul operon, pulO, ishomologous with PilD (3, 40, 48), a protein required for thebiogenesis of type IV pili in P. aeruginosa (39). PilD is anendopeptidase responsible for cleaving the leader sequencefrom precursors of type IV pilin subunits (40). The absenceof functional PilD not only results in a block in processing ofprepilin and failure of bacteria to assemble pili but also leadsto a pleiotropic defect in extracellular protein secretion. Anumber of extracellular enzymes of P. aeruginosa accumu-late in the periplasm in fully processed form, indicating that

J. BACTERIOL.

on Septem

ber 3, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

MINIREVIEW 3425

in addition to prepilin, one or several components of theextracellular secretion machinery also require processing byPilD (51). Genes for four proteins, PddA, PddB, PddC, andPddD, involved in the extracellular secretion of severalenzymes from P. aeruginosa have been recently identified,and PddC was shown to be processed by PilD (41). Notsurprisingly, PddA, -B, -C, and -D are homologs of PulG,-H, -I, and -J. Amino-terminal cleavage of the leader se-quence from the precursor of PulG by PulO has also beenconfirmed (47). Other homologs of type IV pilins involved inextracellular secretion of enzymes from E. chrysanthemi andX. campestris have also recently been described (7, 15). PilDand related peptidase enzymes are the only components ofthe extracellular secretion apparatus with an identified bio-chemical activity.Two additional proteins involved in the biogenesis of P.

aeruginosa type IV pili share sequence homology with puloperon proteins, as well as with proteins encoded by thegenes involved in extracellular protein secretion from Pseu-domonas, Erwinia, and Xanthomonas spp. PilB shares sig-nificant homology with PulE, while PilC is similar to PulF(38a, 45). The most striking feature of PilB and its homologsis the presence of a conserved ATP-binding site in each ofthese proteins, suggesting that its function is perhaps linkedto energy-dependent translocation of proteins across thegram-negative cell envelope. Such extensive conservation ofsequence among proteins involved in the biogenesis of typeIV pili, as well as among the various homologs of the pilinsthemselves, raises the possibility that some of these proteinsparticipate in protein export from the bacterial cell indirectlyand that their primary role may be the assembly of afunctional apparatus of extracellular protein secretion, com-posed of both the cytoplasmic and cell envelope compo-nents.Two sets of signals are required for extracellular secretion

of proteins that contain cleavable signal sequences (46). Theinitial interaction with the normal sec-encoded machineryrequires the amino-terminal signal sequence, which iscleaved by one of the normal signal (leader) peptidases. Thesubsequent sorting which leads to translocation across theouter membrane, therefore, relies on signals found on themature portion of the polypeptide. Because a number ofproteins with little or no sequence similarity are exported bythe same pathway (41a), the signals recognized by theextracellular secretion machinery very likely are structuralmotifs and not a linear sequence of specific amino acids.These signals may take on a specific secondary structure ina region within the polypeptide chain, or they may be part ofa domain assembled from different portions of the foldedpolypeptide.Hybrid proteins carrying the N-terminal portion of pullu-

lanase have been used to detect the extracellular secretionsignals of this enzyme. Surface localization of pullulanase-P-lactamase and pullulanase-alkaline phosphatase requiredat least three-fourths of the 1,090-amino-acid mature protein;any further narrowing of the signals was not possible (28).Only the pullulanase-,B-lactamase hybrid was released fromthe surface, suggesting that this specific pullulanase-alkalinephosphatase hybrid protein folded into a conformationwhich did not permit the release of the fusion protein fromthe outer membrane. Deletion analysis of P. aeruginosaexotoxin A has narrowed a region essential for extracellularsecretion to within a 30-amino-acid sequence at the aminoterminus of the mature polypeptide (14). However, when theexotoxin A signal sequence and the first 30 amino acids ofthe mature protein were fused to alkaline phosphatase, the

hybrid protein was not exported beyond the P. aeruginosaperiplasm (30a). The requirement for a specific export-competent conformation, and the likely existence of noncon-tiguous localization determinants, will make decoding extra-cellular secretion signals experimentally difficult.

EXTRACELLULAR SECRETION OF PROTEINSLACKING CLEAVABLE SIGNAL SEQUENCES

Hemolysin (RTX) family. A number of gram-negativebacteria extracellularly secrete proteins which lack typicalN-terminal signal sequences, presumably by completelybypassing the normal secretion pathway. The first protein ofthis family to be identified was a hemolytic (and thereforecytotoxic) protein produced by a few pathogenic strains ofE. coli (8). This family is named for a tandem nine-amino-acid repeat called RTX (repeats in toxin), which is commonto the E. coli hemolysin and some other cytolysins in thefamily (29, 55). Other members of the family, produced by avariety of gram-negative bacteria, include colicin V (11),bacterial metalloproteases (30, 53), and a cytolysin-adenyl-ate cyclase (12). These varied proteins are related by theirextracellular secretion machinery, as shown by signal se-quence homology. Extracellular secretion of the E. colihemolysin requires three accessory proteins: HlyB andHlyD, encoded by genes linked to the structural gene forhemolysin, and TolC, the product of an unlinked gene (54).A fourth protein coded in the gene cluster, HlyC, is requiredfor activation of the RTX cytotoxin (24). This linkage ofgenes specifying export determinants with those encodingthe transported enzymes is conserved among all gram-negative bacteria utilizing the hemolysin-type extracellularsecretion mechanism.The most intriguing aspect of the hemolysin-type appara-

tus is that HlyB homologs participate in the transport of awide range of unrelated molecules not only across thebacterial cytoplasmic membrane but also across the eukary-otic plasma membrane (5, 16). This superfamily of transportproteins includes, in bacteria, those responsible for import ofsugars, amino acids, and oligopeptides, as well as thoseinvolved in export of cyclic glucans. The best-characterizedeukaryotic homologs of HlyB include the P-glycoproteininvolved in expulsion of hydrophobic drugs from mammaliancells, the protein responsible for resistance to chloroquin inPlasmodium spp., the yeast pheromone transporter (STE6),and the ion channel (CFTR) whose defect is involved incystic fibrosis. The most significant homology among thesesequences is found in a putative ATP-binding domain ofapproximately 200 amino acids.How the extracellular secretion apparatus facilitates the

translocation of hemolysin and related proteins across thecell envelope is not known. The most appealing modelinvokes a selective pore or a transporter, consisting of HlyBand HlyD in an inner membrane complex (5), bound to ToICin the outer membrane (54). This would allow the exportedprotein to completely bypass the periplasmic compartment.This arrangement may be necessary to provide a desirablemembrane topology for the activation of hemolysin byacylation mediated by the product of the hlyC gene (24).Because all systems with proteins which structurally resem-ble hemolysin express an HlyC homolog from a linked gene,it is probable that a similar modification takes place duringexport of all of the proteins that utilize the HlyB-HlyD-likepathway of extracellular secretion.

Classical genetic approaches to decode export signals relyon the isolation of mutants with mutations in a specific

VOL. 174, 1992

on Septem

ber 3, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3426 MINIREVIEW

region of an exported protein which alters its localization.The E. coli hemolysin was the first extracellularly secretedprotein for which some of the localization signals wereidentified. By use of deletion analysis, the extracellulartargeting signal was found to be in a region corresponding tothe 50 endmost C-terminal amino acids of the polypeptide(13). This was subsequently confirmed by demonstrating theexport of hybrid proteins generated by fusing various C-ter-minal portions of the hemolysin to the outer membraneprotein OmpF (32) or to a cytoplasmic protein such aschloramphenicol acetyltransferase, 3-galactosidase, ormammalian prochymosin (25). In each case, only a fractionof hybrid protein synthesized was released from E. coli;however, the observed extracellular secretion was alwaysdependent on the presence of the HlyB-HlyD translocatorproteins. Similarly, deletion and deletion-fusion analysis ofthe E. chrysanthemi metalloprotease B has identified a40-amino-acid C-terminal region which functions as a signalfor extracellular secretion (6). In addition, this same local-ization signal was also somewhat inefficiently recognized bythe hemolysin export determinants HlyB and HlyD in E.coli. The overall features of targeting signals appear to beconserved, allowing secretion of many extracellular proteinsby heterologous translocation machinery (30, 33). Recogni-tion of heterologous export signals is somewhat surprising,given the lack of sequence similarities at the extreme Ctermini of any of these proteins (50).

Extensive mutational analysis of the C-terminal region ofthe E. coli hemolysin has confirmed the role of a specific48-amino-acid region as the targeting signal for the extracel-lular secretion of this protein (50). These amino acids arearranged in a structural motif consisting of an amphipathichelix containing two clusters of charged residues, followedby a brief stretch of uncharged amino acids and terminatingat the extreme C terminus with eight amino acids that aregreatly (>50%) enriched for the hydroxylated amino acidsserine and threonine. Examination of the secondary struc-ture of the C-terminal region in a number of proteins whichare exported by a machinery analogous to the hemolysinexport functions reveals the presence of a potential amphi-pathic domain in regions previously implicated in extracel-lular secretion. This region may serve as a signal to thetranslocation apparatus HlyB-HlyD. Alternatively, the am-phipathic character of the signal may allow spontaneousinsertion of the protein into the inner membrane in anorientation which allows HlyB-HlyD to continue the trans-location of the remainder of the polypeptide chain across theinner and outer membranes.

Virulence plasmid-encoded Yops. Pathogenic species ofYersinia express a number of virulence plasmid-encodedproteins termed Yops (Yersinia outer membrane proteins).Recently, a number of the Yops have been shown to beexported into the surrounding medium (36). When the de-duced amino acid sequences for the yop genes and primaryYop protein sequence were compared, it was found thatthese Yops lack N-terminal signal sequences and are re-leased by Yersinia enterocolitica without N-terminal proc-essing. Mutations in at least one region of the virulenceplasmid (the virC locus) block extracellular secretion ofYops. The virC locus contains 13 open reading frames, oneof which, yscC, encodes a protein sharing limited homologywith PulD, a protein required for export of pullulanase fromK oxytoca (35). No sequence similarity between any of theother proteins encoded in the virC locus and members of thepullulanase or of the hemolysin family of translocators hasbeen found. Therefore, the mechanism of extracellular se-

cretion of the Yops may be the prototype of a novel exportmechanism involving proteins without classical signal se-quences.

Deletion analysis of genes encoding the exported Yops ofY. enterocolitica led to expression of stable, truncatedproteins; many of these proteins, despite lacking substantialportions of their respective carboxy termini, were efficientlyexported from the bacterial cell. The minimal N-terminalsequence required for export of the truncated proteinsranged from 48 amino acids for YopH and 76 amino acids forYopQ to 98 amino acids for YopE (34). The N-terminal 65amino acids of YopH, when fused to alkaline phosphataselacking its own signal sequence, led to efficient extracellularlocalization of the hybrid protein, suggesting that in YopH,as well as in many other Yops, the amino terminus containsall of the information recognized by the plasmid-encoded, orpossibly chromosomally encoded, extracellular secretionapparatus.

EXTRACELLULAR SECRETION PATHWAYS

Identification of the machinery responsible for extracellu-lar protein localization by gram-negative bacteria has unam-biguously shown that the process involves specific proteintargeting and not merely the release of a specialized class ofproteins from the outer membrane. It is remarkable howvery little is known about the precise pathways which lead tothe final steps in the localization of the outer membraneproteins. Therefore, one cannot exclude the possibility thatthe mechanisms of extracellular protein secretion will sharesome additional similarities with the process of outer mem-brane assembly.The pathway allowing extracellular secretion of a protein

from a gram-negative bacterium has to lead through threedistinct compartments (inner membrane, periplasm, andouter membrane) that separate the cytoplasm, the site ofsynthesis, from the cell exterior. Demonstration of a path-way by pulse-labeling of proteins and cellular fractionationduring the chase has been very difficult for most proteins thatare exported from a bacterial cell. This includes proteinswhich utilize the normal secretion pathway during the initialstages of interactions with the inner membrane. Membraneperturbation can lead to outer membrane localization ofsome extracellularly secreted proteins (31), suggesting thatthe extracellular secretion machinery is located at sites ofadhesion between the inner and outer membranes, where thetwo membranes may be contiguous. These adhesion zones,first demonstrated by Bayer (4), not only may allow transferof newly synthesized proteins from the cytoplasm to theexterior, bypassing the periplasm, but also may permit directchanneling of cytoplasmic energy into the extracellular se-cretion process.Some proteins enter the periplasm as part of their normal

extracellular secretion pathway. These include the V. chol-erae enterotoxin (17), the Aeromonas hydrophila aerolysin(22), and the P. aeruginosa elastase (26). While these pro-teins, and perhaps all extracellularly secreted proteins, mayat one time pass through the periplasm, the Bayer junctionsmight still function in sorting polypeptides, especially ifsome of the components of the extracellular secretion ma-chinery are assembled in these adhesion zones. This mech-anism would have the ability to distinguish between differ-entially compartmentalized proteins, whether they enter theexport pathway from the cytoplasm, the inner membrane, orthe periplasm, always leading to correct localization ofextracellular proteins.

J. BACTERIOL.

on Septem

ber 3, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

MINIREVIEW 3427

In the absence of direct evidence that Bayer junctionsparticipate in protein export, it is still a formal possibilitythat extracellular secretion occurs by a two-step mechanism,wherein the newly synthesized proteins are first secretedinto the periplasm and then translocated across the outermembrane. This model has been supported not only bydemonstrating that periplasmic pools exist for several of theabove-mentioned extracellularly secreted proteins but alsoby showing periplasmic accumulation of exported enzymesin strains that carry mutations in components of the extra-cellular secretion machinery (22, 51). In fact, some of theseproteins have assumed the native, fully folded conformation,including the correct formation of intrachain disulfide bonds.Translocation of such proteins from the periplasm across theouter membrane would necessitate significant rearrangementof their structure during passage through an outer membranechannel. Following translocation, the proteins would have torefold into their native conformation. Alternatively, if foldedpolypeptides were released from the periplasm, this processwould then require the presence of pores in the outermembrane. Such pores would have to allow unidirectionaltransit of a broad class of extracellularly secreted proteins,without concomitant release of the periplasmic contents.Such selective gates have not been demonstrated for anygram-negative outer membrane.

CONCLUSION

A great deal of progress has been made during the past 5years in defining components of the extracellular proteinsecretion machinery in gram-negative bacteria. Our under-standing of precisely how these components act together inextracellular secretion, however, remains rather limited. It isapparent that there are diverse pathways employed bygram-negative bacteria to export polypeptides from the cell,and this review has focused on the better-characterizedpathways shared by a variety of gram-negative microorgan-isms.There are striking sequence similarities among some of the

proteins involved in membrane translocation of a variety ofunrelated substrates. The family of proteins related to the E.coli hemolysin transporter HlyB has members in both eu-karyotic and prokaryotic cells (16). This protein familyfunctions in the import as well as the export of a wide rangeof substrates. Similarly, several components analogous tothose of the pullulanase transport machinery mediate exportof proteins and assembly of pili in gram-negative bacteria, aswell as the uptake of DNA in gram-positive bacteria (1).Understanding of the basic mechanisms by which proteinsare released from gram-negative bacteria may providegreater insight into the principles of molecular traffickingshared by both prokaryotic and eukaryotic cells.

ACKNOWLEDGMENTS

I thank C. Manoil, S. Moseley, M. Starnbach, M. Strom, K.Rhodes, and D. Nunn for stimulating discussions and criticalreading of the manuscript.The research in my laboratory has been supported by Public

Health Service grant AI-21451 from NIH and by grants from theCystic Fibrosis Foundation.

REFERENCES1. Albano, M., R. Breitling, and D. Dubnau. 1989. Nucleotide

sequence and genetic organization of the Bacillus subtilis comGoperon. J. Bacteriol. 171:5386-5404.

2. Antoine, R., and C. Locht. 1990. Roles of the disulfide bond and

the carboxy-terminal region of the S1 subunit in the assemblyand biosynthesis of pertussis toxin. Infect. Immun. 58:1518-1526.

3. Bally, M., G. Ball, A. Badere, and A. Lazdunski. 1991. Proteinsecretion in Pseudomonas aeruginosa: the xcpA gene encodesan integral membrane protein homologous to Klebsiella pneu-moniae secretion function protein PulO. J. Bacteriol. 173:479-486.

4. Bayer, M. E. 1979. The fusion sites between outer membraneand cytoplasmic membrane of bacteria: their role in membraneassembly and virus infection, p. 167-202. In M. Inouye (ed.),Bacterial outer membranes: biogenesis and functions. JohnWiley & Sons, Inc., New York.

5. Blight, M. A., and I. B. Holland. 1990. Structure and function ofhemolysin B, P-glycoprotein and other members of a novelfamily of membrane translocators. Mol. Microbiol. 4:873-880.

6. Delepelaire, P., and C. Wandersman. 1990. Protein secretion inGram-negative bacteria. The extracellular metalloprotease Bfrom Erwinia chrysanthemi contains a C-terminal secretionsignal analogous to that of Escherichia coli alpha hemolysin. J.Biol. Chem. 265:17118-17125.

7. Dums, F., J. M. Dow, and M. J. Daniels. 1991. Structuralcharacterization of protein secretion genes of bacterial phyto-pathogen Xanthomonas campestris pathovar campestris: relat-edness to secretion systems of other Gram-negative bacteria.Mol. Gen. Genet. 229:357-364.

8. Felmlee, T., S. Pellet, and R. A. Welch. 1985. Escherichia colihemolysin is released extracellularly without cleavage of asignal peptide. J. Bacteriol. 163:88-93.

9. Filoux, A., M. Bally, G. Ball, M. Akrim, J. Tommassen, and A.Lazdunski. 1990. Protein secretion in Gram-negative bacteria:transport across the outer membrane involves common mecha-nisms in different bacteria. EMBO J. 9:4323-4329.

10. Finkelstein, R. A. 1973. Cholera. Crit. Rev. Microbiol. 2:553-623.

11. Gilson, L., H. K. Mahanty, and R. Kolter. 1990. Geneticanalysis of an MDR-like export system: the secretion of colicinV. EMBO J. 9:3875-3884.

12. Glaser, P., H. Sakamoto, J. Bellalou, A. Ullmann, and A.Danchin. 1988. Secretion of cyclolysin, the calmodulin-sensitiveadenylate cyclase-hemolysin bifunctional protein of Bordetellapertussis. EMBO J. 7:3997-4004.

13. Gray, L., N. Mackman, J.-M. Nicaud, and I. B. Holland. 1986.The carboxy-terminal region of hemolysin 2001 is required forsecretion of the toxin from Escherichia coli. Mol. Gen. Genet.205:127-133.

14. Hamood, A. N., J. C. Olson, T. S. Vincent, and B. H. Iglewski.1989. Regions of toxin A involved in toxin A excretion inPseudomonas aeruginosa. J. Bacteriol. 171:1817-1824.

15. He, S., M. Lindeberg, A. C. Chaterjee, and A. Collmer. 1991.Cloned Erwinia chrysanthemi out genes enable Escherichia colito selectively secrete a diverse family of heterologous proteinsto its milieu. Proc. Natl. Acad. Sci. USA 88:1079-1083.

16. Higgins, C. F., S. C. Hyde, M. M. Mimmack, U. Gileadi, D. R.Gill, and M. P. Galagher. 1990. Binding protein-dependenttransport systems. J. Bioenerg. Biomembr. 22:571-591.

17. Hirst, T. R., and J. Holmgren. 1987. Transient entry of entero-toxin subunit into the periplasm occurs during secretion fromVibrio cholerae. J. Bacteriol. 169:1037-1045.

18. Hirst, T. R., J. Sanchez, J. B. Kaper, S. J. S. Hardy, and J.Holmgren. 1984. Mechanism of toxin secretion by Vibrio chol-erae investigated in strains harboring plasmids that encodeheat-labile enterotoxins of Escherichia coli. Proc. Natl. Acad.Sci. USA 81:7752-7756.

19. Hirst, T. R., and R. A. Welch. 1988. Mechanism for secretion ofextracellular proteins by Gram-negative bacteria. Trends Bio-chem. Sci. 13:265-268.

20. Holland, I. B., M. A. Blight, and B. Kenny. 1990. The mecha-nism of secretion of hemolysin and other polypeptides fromGram-negative bacteria. J. Bioenerg. Biomembr. 22:473-491.

21. Holmes, R. K., M. L. Vasil, and R. A. Finkelstein. 1975. Studiesof toxigenicity in Vibrio cholerae. III. Characterization ofnontoxigenic mutants in vitro and in experimental animals. J.

VOL. 174, 1992

on Septem

ber 3, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

3428 MINIREVIEW

Clin. Invest. 55:551-560.22. Howard, P. S., and J. T. Buckley. 1985. Protein export by a

gram-negative bacterium: production of aerolysin by Aeromo-nas hydrophila. J. Bacteriol. 161:1118-1124.

23. Huang, J., and M. A. Schell. 1990. Evidence that extracellularexport of endoglucanase encoded by egl of Pseudomonas so-lanacearum occurs by a two-step process involving a lipopro-tein intermediate. J. Biol. Chem. 265:11628-11632.

24. Issartel, J.-P., V. Koronakis, and C. Hughes. 1991. Activation ofEscherichia coli prohemolysin to the mature toxin by acylcarrier protein-dependent acylation. Nature (London) 351:759-761.

25. Kenny, B., R. Haigh, and I. B. Holland. 1991. Analysis ofhemolysin transport process through the secretion from Esche-nichia coli of PCM, CAT or 1-galactosidase fused to the HlyC-terminal signal domain. Mol. Microbiol. 5:2557-2568.

26. Kessler, E., and M. Safrin. 1988. Synthesis, processing, andtransport of Pseudomonas aeruginosa elastase. J. Bacteriol.170:5241-5247.

27. Klauser, T., J. Pohlner, and T. F. Meyer. 1990. Extracellulartransport of cholera toxin B subunit using Neisseria IgA prote-ase B-domain: conformation-dependent outer membrane trans-location. EMBO J. 9:1991-1999.

28. Kornacker, M. G., and A. P. Pugsley. 1990. The normallyperiplasmic enzyme 1 lactamase is specifically and efficientlytranslocated through the Escherichia coli outer membrane whenit is fused to the cell-surface enzyme pullulanase. Mol. Micro-biol. 4:1101-1109.

29. Koronakis, V., M. Cross, B. Senior, E. Koronakis, and C.Hughes. 1987. The secreted hemolysins of Proteus mirabilis,Proteus vulgaris, and Morganella morganii are geneticallyrelated to each other and to the alpha-hemolysins of Eschenichiacoli. J. Bacteriol. 169:1509-1515.

30. Letoffe, S., P. Delepelaire, and C. Wandersman. 1991. Cloningand expression in Escherichia coli of the Serratia marcescensmetalloprotease gene: secretion of the protease from E. coli inthe presence of the Erwinia chrysanthemi protease secretionfunctions. J. Bacteriol. 173:2160-2166.

30a.Lory, S. Unpublished data.31. Lory, S., P. C. Tai, and B. D. Davis. 1983. Mechanism of

excretion of proteins by gram-negative bacteria: Pseudomonasaeruginosa exotoxin A. J. Bacteriol. 156:695-702.

32. Mackman, N., K. Baker, L. Gray, and I. B. Holland. 1987.Release of a chimeric protein into the medium from Escherichiacoli using the C-terminal secretion signal of hemolysin. EMBOJ. 6:2835-2841.

33. Masure, H. R., D. C. Au, M. K. Gross, M. G. Donovan, andD. R. Storm. 1990. Secretion of Bordetella pertussis adenylatecyclase from Escherichia coli containing the hemolysin operon.Biochemistry 29:140-145.

34. Michiels, T., and G. R. Cornelis. 1991. Secretion of hybridproteins by the Yersinia Yop export system. J. Bacteriol.173:1677-1685.

35. Michiels, T., J.-C. Vanoothegan, C. L. DeRouvroit, B. China, A.Gustin, P. Boudry, and G. R. Cornelis. 1991. Analysis of virC,an operon involved in secretion of Yop proteins by Yersiniaenterocolitica. J. Bacteriol. 173:4994-5009.

36. Michiels, T., P. Wattiau, R. Brasseur, J.-M. Ruysschaert, and G.Cornelis. 1990. Secretion of Yop proteins by yersiniae. Infect.Immun. 58:2840-2849.

37. Miyazaki, H., N. Yanagida, S. Horinouchi, and T. Beppu. 1989.Characterization of the precursor of Serratia marcescens serineprotease and COOH-terminal processing of the precursor duringits excretion through the outer membrane of Escherichia coli. J.Bacteriol. 171:6566-6572.

38. Neil, R. J., B. E. Ivins, and R. K. Holmes. 1983. Synthesis and

secretion of the plasmid-coded heat-labile enterotoxin of Esch-erichia coli in Vibrio cholerae. Science 221:289-291.

38a.Nunn, D. Unpublished data.39. Nunn, D., S. Bergman, and S. Lory. 1990. Products of three

accessory genes, pilB, pilC, and pilD, are required for biogen-esis of Pseudomonas aeruginosa pili. J. Bacteriol. 172:2911-2919.

40. Nunn, D., and S. Lory. 1991. Product of Pseudomonas aerugi-nosa gene pilD is a prepilin leader peptidase. Proc. Natl. Acad.Sci. USA 88:3281-3285.

41. Nunn, D., and S. Lory. 1992. Components of the protein-excretion apparatus of Pseudomonas aeruginosa are processedby the type IV prepilin peptidase. Proc. Natl. Acad. Sci. USA89:47-51.

41a.Nunn, D., and S. Lory. Unpublished data.42. Pearson, G. D. N., and J. Mekalanos. 1982. Molecular cloning of

Vibrio cholerae enterotoxin genes in Escherichia coli K12.Proc. Natl. Acad. Sci. USA 79:2976-2980.

43. Pizza, M., M. Bugnoli, R. Manetti, A. Covacci, and R. Rappuoli.1990. The subunit S1 is important for pertussis toxin secretion.J. Biol. Chem. 29:17759-17763.

44. Pohlner, J., R. Halter, K. Beyreuther, and T. F. Meyer. 1987.Gene structure and extracellular secretion of Neisseria gonor-rhoeae IgA protease. Nature (London) 325:458-462.

45. Possot, O., C. d'Enfert, I. Reyss, and A. P. Pugsley. 1992.Pullulanase secretion in Escherichia coli K-12 requires a cyto-plasmic protein and a putative polytopic cytoplasmic membraneprotein. Mol. Microbiol. 6:95-105.

46. Pugsley, A., C. d'Enfert, and M. G. Kornacker. 1990. Geneticsof extracellular protein secretion by Gram-negative bacteria.Annu. Rev. Genet. 24:67-90.

47. Pugsley, A. P., and B. Dupoy. 1992. An enzyme with type IVprepilin peptidase activity is required to process components ofthe general extracellular protein secretion pathway of Klebsiellaoxytoca. Mol. Microbiol. 6:751-760.

48. Pugsley, A. P., and I. Reyss. 1990. Five genes at the 3' end of theKlebsiella pneumoniaepulC operon are required for pullulanasesecretion. Mol. Microbiol. 4:365-379.

49. Schatz, P. J., and J. Beckwith. 1990. Genetic analysis of proteinexport in Escherichia coli. Annu. Rev. Genet. 24:215-248.

50. Stanley, P., V. Koronakis, and C. Hughes. 1991. Mutationalanalysis supports a role for multiple structural features in theC-terminal secretion signal of Escherichia coli hemolysin. Mol.Microbiol. 5:2391-2403.

51. Strom, M. S., D. Nunn, and S. Lory. 1991. Multiple roles of thepilus biogenesis protein PilD: involvement of PilD in excretionof enzymes from Pseudomonas aeruginosa. J. Bacteriol. 173:1175-1180.

52. Tamura, M., K. Nogimori, S. Murai, M. Yajima, K. Ito, T.Katada, M. Ui, and S. Ishii. 1982. Subunit structure of islet-activating protein, pertussis toxin, in conformity with the A-Bmodel. Biochemistry 21:5516-5522.

53. Wandersman, C. 1989. Secretion, processing and activation ofbacterial extracellular proteases. Mol. Microbiol. 3:1825-1831.

54. Wandersman, C., and P. Delepelaire. 1990. TolC, an Escherichiacoli outer membrane protein required for hemolysin secretion.Proc. NatI. Acad. Sci. USA 87:4776-4780.

55. Welch, R. A. 1991. Pore-forming cytolysins of Gram-negativebacteria. Mol. Microbiol. 5:521-528.

56. Wickner, W., A. J. M. Dreissen, and F. Hartl. 1991. Theenzymology of protein translocation across the Escherichia coliplasma membrane. Annu. Rev. Biochem. 60:101-124.

57. Yanagida, N., T. Uozumi, and T. Beppu. 1986. Specific excre-tion of Serratia marcescens protease through the outer mem-brane of Escherichia coli. J. Bacteriol. 166:937-944.

J. BACTERIOL.

on Septem

ber 3, 2020 by guesthttp://jb.asm

.org/D

ownloaded from