INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Essentially all multicellular eukaryotes normally exist in association with a suite of cooperative microorganisms, many of which have been recognized to play essential, benign roles in the development and health of their hosts (5). Such interactions can be either consortial, being composed of many microbial species, as in the enteric tracts of all animals (6), or monospecific, as in the root nodules of plants (42) and the light-emitting organs of marine squids and fishes (34). In most of these associations, microorganisms are transmitted horizontally to each new host generation when bacteria present in the ambient environment come in contact with and colonize a particular tissue in their animal and plant hosts. Species specificity is important in these relationships, yet remarkably little is known about how each association is established and persists over the lifetime of the host. One common mechanism of cell-cell recognition is specific attachment structures, which provide the bacterium with the ability to first identify and then remain in close contact with the appropriate target tissue.

Bacteria utilize a number of external structures to attach to specific host tissues and, in some instances, to bring about changes in the biochemical and cellular activity of the host cells to which they adhere (12). Among the best studied of these adhesive structures are pili (fimbriae) and outer membrane proteins (OMPs, or porins), both of which project into the bacterium's environment. The regulation, structure, and specificity of fimbriae and OMPs have been best described in certain pathogenic bacterial species (23, 29, 35), including the human intestinal pathogen Vibrio cholerae (2). In contrast, less is known about the mechanisms by which benign bacteria initiate specific, cooperative, and often obligate associations that can persist throughout the life of the host.

The marine bacterium Vibrio fischeri is the specific symbiont of the light-emitting organ of the sepiolid squid Euprymna scolopes (20). The nascent light organ of a newly hatched E. scolopes juvenile is axenic, but cells of V. fischeri present in the surrounding seawater serve as an inoculum that passes through pores on the surface of the organ and proliferates within epithelium-lined internal crypt spaces (27). The colonization process requires that the bacteria migrate past several different host cell types on their way into the organ (26) and then become securely associated with the microvillar surface of the crypt cells (14). Periodic expulsion of over 95% of the symbiotic bacterial population every morning (33) may further select for closely adhering V. fischeri cells, which become increasingly invested in the microvilli during the first few days after colonization (14). In addition, the presence of the bacteria induces both reversible and irreversible stages in the program of normal light organ development (22, 44), suggesting that signaling is occurring between the bacteria and their host.

The mechanisms by which V. fischeri cells attach to and colonize the light organ tissues of juvenile hosts are just beginning to be described in detail. For example, aggregation of V. fischeri cells in a host-derived mucus-like matrix is an early event that is required for these cells to find and enter the pores that lead to the nascent light organ crypts (27). In addition, evidence exists that mannose residues present on the cells lining the crypts may function as receptors for the colonizing bacteria (21) and that bacterial fimbriae are involved in this process (B. Feliciano and E. G. Ruby, Abstr. 99th Annu. Meet. Am. Soc. Microbiol., abstr. 462, 1999). Reports that the OmpU protein of some strains of V. cholerae might serve in attachment of this pathogen to host tissue (38) suggested to us that an examination of the OMPs of V. fischeri might lead to a better understanding of the role of extracellular structures in the symbiotic colonization of the squid light organ.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and media. V. fischeri strain ESR1, a rifampin-resistant derivative of wild-type strain ES114 (8), was used as the parent strain for all mutant constructions (Table 1). Escherichia coli strain DH5 (4) was the recipient for most cloning experiments, and plasmids were passaged through a dam mutant E. coli strain prior to introduction into V. fischeri by electroporation (45).

Protein analyses. Cellular protein extracts were obtained from cultures of V. fischeri cells grown to mid-exponential phase (OD of approximately 0.2) in SWT medium. Cells were harvested by centrifugation and washed with seawater. Total soluble proteins were extracted from washed cell pellets that were resuspended in a cold lysing buffer containing 50 mM Tris-HCl (pH 7.9), 50 mM EDTA, 15% (wt/vol) sucrose, and lysozyme (final concentration, 0.5 mg/ml). This suspension was incubated for 30 min on ice and centrifuged for 5 min at 12,000 × g. OMP-enriched fractions were obtained from the resulting pellet using a 1% N-lauroylsarcosine detergent extraction as previously described (19). Soluble proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and visualized by staining with Coomassie brilliant blue. Western blot analyses using a Photobacterium profundum OmpL antibody were performed as described by Welch and Bartlett (47).

Cloning procedures. V. fischeri DNA was manipulated using previously described methods (10, 39, 46). DNA fragments obtained by restriction endonuclease digestion were separated by agarose gel electrophoresis, and the desired fragments were extracted from gel slices using GeneClean (Bio 101, Inc., Vista, Calif.). T4 DNA ligase was used to join two fragments together, the ligated fragments were transformed into E. coli DH5 cells made competent by CaCl₂ treatment (4), and plasmid-carrying strains were isolated on selective antibiotic-containing media. Enzymes were obtained from Promega, Inc. (Madison, Wis.) or New England Biolabs (Beverly, Mass.).

Squid colonization assays. Juveniles of E. scolopes were inoculated within 4 h of hatching with V. fischeri strains as described previously (32). Briefly, individual squids in vials containing 4 ml of seawater were exposed for 3 h to an inoculum of either the ompU mutant or its parent strain. After this inoculation, the animals were transferred to symbiont-free seawater and maintained for up to 3 days. Measurements of the luminescence of the juvenile squid were performed using a Turner 20/20 luminometer (Turner, Sunnyvale, Calif.) and used as an indication of successful colonization (33). At specific times following inoculation, the juveniles were homogenized, and dilutions of the homogenates were spread on SWT agar medium to determine the number of CFU in the light organ (1). To examine whether the ompU mutant had a competitive disadvantage compared to the parent strain, juvenile squids were inoculated with a 1:1 mixture of the mutant and the parent, and the proportion of the resulting symbiotic population was determined by plating light organ homogenates on antibiotic-containing media that differentiated between the two strains (46).

Nucleotide sequence accession number. The GenBank nucleotide accession number for the V. fischeri ompU gene sequence is AY050511.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

OMPs of V. fischeri. A characterization of the presence and regulation of OMPs in V. fischeri cells was begun by growing the bacteria in media limited for either iron or phosphate, two nutrients that often require transport through specific porins. SDS-PAGE analysis of OMPs present in cell extracts of V. fischeri strain ES114 (or its derivative, ESR1) grown in SWT medium revealed three major protein bands with molecular masses of approximately 34, 40, and 41 kDa (Fig. 1). Two additional minor proteins (approximately 74 and 88 kDa) were also present and were significantly induced in cells cultured in medium containing the iron chelator EDDHA, suggesting that they might be in the class of iron siderophore porins described for other Vibrio species (17). Similarly, growth in a phosphate-limited medium resulted in enhancement of the 40-kDa OMP band relative to the 41-kDa band (Fig. 1). There was no indication that growth in any of these media resulted in a significant change in the intensity of the 34-kDa band.

View larger version (90K): [in this window] [in a new window]   FIG. 1.   SDS-polyacrylamide gel of OMPs isolated from cells of V. fischeri strain ESR1 grown under different culture conditions. Lane 1, cells grown in nutrient medium SWT (7.2 µg of protein loaded); lane 2, cells grown in SWT under iron limitation (6.5 µg of protein loaded); lane 3, cells grown in a minimal medium under phosphate limitation (8.0 µg of protein loaded). Molecular mass standards (stds) are indicated. The arrow indicates the 34-kDa band.

View larger version (17K): [in this window] [in a new window]   FIG. 2.   Inhibition of light organ colonization by V. fischeri cells treated with an antiserum directed against P. profundum OmpL, a homolog of V. fischeri OmpU. Groups of between 7 and 10 newly hatched, uncolonized juveniles of E. scolopes were individually inoculated by placing them for 3 h in 5 ml of seawater containing approximately 104 cells of V. fischeri ESR1 that were either untreated () or treated with OmpL antiserum () or preimmune serum (). Control animals () were incubated in seawater without added V. fischeri cells. The data shown are representative of those obtained in four separate experiments.

Isolation of V. fischeri ompU homolog. To identify the presence of a V. cholerae ompU homolog in the genome of V. fischeri, we first aligned the V. cholerae gene (37) and the homologous ompL gene from P. profundum (48). Based on a conserved sequence at the C terminus, two oligonucleotide PCR primers, OMP3 (5'-GACGCTACTTACTACTTC-3') and OMP4 (5'-AAGTCGTAACGTACACC-3') that would amplify a predicted 134-bp fragment corresponding to nucleotide positions 928 to 1061 in the V. cholerae ompU gene were designed. PCR amplification using these primers and V. fischeri chromosomal DNA as the target resulted in a 137-bp fragment that, when sequenced, aligned with 80% identity to the expected corresponding region of the V. cholerae ompU sequence.

Alignment of V. fischeri ompU gene homolog with other OMP genes. As expected, the complete ORF1 aligned well with the V. cholerae ompU and P. profundum ompL genes. There was 50.8 and 45.5% identity between the deduced amino acid sequence of the V. fischeri ORF1 and those of V. cholerae OmpU and P. profundum OmpL, respectively, leading us to propose that ORF1 should be designated the V. fischeri ompU gene. The N terminus of the deduced V. fischeri OmpU protein (Fig. 3) contains a 21-amino-acid leader sequence that is 52% identical to the processed 21-amino-acid signal peptide of P. profundum OmpL (48). Based on the assumption that this peptide sequence is also processed in V. fischeri, the mature OmpU protein of V. fischeri would be 300 residues long, with a deduced molecular mass of 32.5 kDa. This molecular mass is consistent with the OMP SDS-polyacrylamide gel band at approximately 34 kDa (Fig. 1). Twenty-eight base pairs downstream of the stop codon of ompU is a putative rho-independent terminator, beyond which is ORF2, which encodes a partial protein sequence with highest identity to the E. coli penicillin-binding protein 4 gene, called dacB. The V. cholerae ompU gene is also located directly upstream of a dacB gene homolog (11).

View larger version (29K): [in this window] [in a new window]   FIG. 3.   Alignment of deduced amino acid sequence encoded by the V. fischeri (VFIS) ompU gene with those of V. cholerae (VCHO) ompU and P. profundum (PPRO) ompL genes. Residues that are identical to those in the V. fischeri sequence are indicated by an asterisk (*), and gaps that were inserted to optimize alignment are indicated by periods.

Expression of OmpU in V. fischeri. Experiments were conducted to determine whether OmpU was differentially expressed during growth in either minimal medium or the tryptone-based SWT medium. Cells of V. fischeri strain ESR1 were harvested at the early, middle, and late phases of exponential growth, and SDS-polyacrylamide protein gels were run to examine the relative intensity of the OmpU 34-kDa band. While generally produced to the same level in all media tested, this band appeared somewhat more prominent in cells grown to late exponential phase in SWT medium (data not shown). Culturing the cells under conditions that were limited for iron, phosphate (Fig. 1), or oxygen (i.e., anaerobic culture) had no significant effect on the intensity of the 34-kDa band.

Construction of V. fischeri ompU null mutant. To determine the possible function(s) of OmpU in V. fischeri, we constructed a mutation in the ompU gene by marker exchange. Briefly, a 1.1-kbp PstI/EcoRV fragment from pKV36 (provided by K. Visick), which contains a chloramphenicol resistance (Cm^r) gene, was used to replace a 560-bp PstI/HincII fragment within the ompU coding region in pFA3 to form pFA5 (Fig. 4, step 1). A trimethoprim resistance (Tp^r) gene was obtained as a NotI fragment from pEVS20 (provided by E. Stabb) and inserted into the NotI site in pFA5, creating pFA8 (Fig. 4, step 2). pFA8 was electroporated (45) into V. fischeri ESR1, and Cm^r clones were selected. Because the ColE1 origin does not replicate well in V. fischeri, plasmid pFA8 is lost over time in the absence of selection. Clones were passaged several times on medium without antibiotic to select cells that had integrated the plasmid by homologous recombination. Such clones were then serially passaged on chloramphenicol-containing medium and periodically patched onto trimethoprim-containing medium to screen for a second recombinational event that had led to excision of the plasmid (resulting in the loss of Tp^r) and replacement of the wild-type copy of ompU with the insertionally inactivated one. One of a few such Cm^r and Tp^s clones obtained was designated V. fischeri strain OM3. Gene replacement in OM3 was confirmed by PCR amplification of the ompU locus from chromosomal DNA. Only a single band was produced, and it had the size expected for the disrupted ompU gene rather than the full-length wild-type gene.

View larger version (23K): [in this window] [in a new window]   FIG. 4.   Two-step scheme for constructing an insertional mutation in the V. fischeri ompU gene. In step 1, a Cmr cassette is used to replace an internal portion of the ompU gene in pFA3, resulting in pFA5. In step 2, to facilitate screening of recombinants, a Tpr cassette is also added to pFA5, yielding pFA8. This plasmid was then introduced into V. fischeri to produce an ompU mutation by double recombination. Abbreviations: lacZ, -galactosidase gene; ColE1 ori, ColE1 origin of replication; F1 ori, F1 origin of replication; Apr, ampicillin resistance gene; Cmr, chloramphenicol resistance gene; Knr, kanamycin resistance gene; Tpr, trimethoprim resistance gene.

View larger version (100K): [in this window] [in a new window]   FIG. 5.   SDS-polyacrylamide gel of proteins isolated from different strains of V. fischeri. Cultures were grown at room temperature in SWT medium for 14 to 18 h with shaking. Lane 1, KR-tox1 (toxR omp+); lane 2, ESR1 (parent strain, ompU+); lane 3, OM3 (ompU gene replacement mutant); lane 4, OM3 pFA9 (OM3 complemented with pFA9, ompU+). Positions of molecular mass standards (stds) are indicated. The arrow indicates the 34-kDa OmpU protein.

Growth characteristics of V. fischeri ompU mutant in culture. When grown in either SWT or LBS, which are rich nutrient media containing tryptic peptides and yeast extract, OM3 and its parent ESR1 grew at the same rate and to about the same cell yield (Fig. 6 and data not shown), suggesting that the absence of OmpU creates no significant defect in general bacterial metabolism. However, growth of strain OM3 is restricted by a significantly lower concentration of bile salts, the antimicrobial peptide protamine sulfate, or SDS compared to the ompU⁺ parent strain (Fig. 6). A similar growth defect was observed with cells of OM3 grown in the presence of deoxycholate (data not shown). This defect in OM3 could be complemented by pFA9, which carries an intact ompU gene (Table 1), but not by the vector control, pVO8 (Fig. 6C). Perhaps the apparently increased SDS resistance of the complemented OM3 strain reflected the increased gene dosage of ompU resulting from carriage of the multicopy pFA9 plasmid. Taken together, these data suggest that OmpU is required to maintain normal cell integrity.

View larger version (26K): [in this window] [in a new window]   FIG. 6.   Effects of bile, protamine, and SDS on the growth of strains OM3 and ESR1. Cultures of the ompU mutant strain OM3 (hatched bars) or its parent, ESR1 (solid bars), were inoculated into LBS medium containing different concentrations of bovine bile, protamine sulfate, or SDS. (A) After 24 h of growth at 28°C, dilutions of the bile-containing cultures were spread on LBS agar medium, and the numbers of CFU in the cultures were calculated. The value of each bar is the average of two separate experiments. Similarly, LBS cultures containing either protamine sulfate (B) or SDS (C) were incubated for 24 h at 28°C, and the resulting relative cell density was estimated spectrophotometrically. Cultures of strain OM3 carrying either a wild-type copy of ompU on pFA9 (gray bars) or the vector control, pVO8 (open bars), were also tested on SDS-containing LBS medium. Error bars indicate standard errors of the means.

Colonization characteristics of V. fischeri ompU mutant. Experiments were conducted to determine whether the mutation in ompU resulted in a symbiotic defect in V. fischeri. A comparison of the light organ colonization efficiency of strain OM3 relative to its parent, ESR1, indicated that when presented at a concentration of between 1,000 and 2,000 CFU/ml of seawater, the mutant strain was less effective at initiating the symbiosis (Fig. 7). Specifically, while about 70% of the animals exposed to strain ESR1 were colonized at 14 h postinoculation, fewer than 30% of the animals exposed to the ompU mutant were infected. Similarly, by 38 h about 90% of the ESR1-exposed animals were colonized, while only 65% of the mutant-exposed animals were.

View larger version (24K): [in this window] [in a new window]   FIG. 7.   Colonization efficiency of V. fischeri ompU mutant strain. Groups of newly hatched, uncolonized juvenile squids were placed into seawater containing between 700 and 1,500 cells of either the ompU mutant, OM3, or its parent strain, ESR1, per ml for 3 h. In each experiment the OM3 inoculum was determined to have at least as many cells as the ESR1 inoculum. Starting 14 h after inoculation and continuing at different intervals, the animals were assayed for the production of luminescence, an indication of successful colonization. The percentage of animals in the OM3-inoculated group () and the ESR1-inoculated group () that produced luminescence was determined. Data points are the means of three separate experiments, and error bars indicate the standard errors of the means.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial OMPs constitute a class of cell envelope proteins whose members are involved in an array of diverse surface-mediated phenomena that include attachment, nutrient acquisition, and host-cell signal transduction (13, 35). In the genus Vibrio, OMPs function as porins in iron, phosphate, and sugar accumulation (15, 19, 41) and in attachment to inanimate and animate surfaces (37, 40). Considerable emphasis has been placed on understanding the role of Vibrio OMPs in pathogenesis, yet important issues have remained unresolved. For instance, in V. cholerae strain O1, OmpU has been reported to be involved in bacterial adhesion to host tissue (37), while other evidence suggests it is not required for adherence to the intestinal epithelium of rabbits (24). We report here the presence in V. fischeri of an ompU homolog that encodes a 32.5-kDa OMP. A V. fischeri ompU null mutant grew normally in both complex and minimal nutrient media; however, it was more susceptible to growth inhibition in the presence of either anionic detergents or the cationic antimicrobial peptide protamine. In addition, the mutant had a decreased ability to colonize the light organ of juvenile E. scolopes squid except when it was presented together with its ompU⁺ parent. This work constitutes the first report that ties the presence of a specific bacterial OMP to the initiation of a successful cooperative association with an animal host.

The V. fischeri ompU gene has a number of characteristics in common with its homologs in V. cholerae and P. profundum, including a putative N-terminal processing site and similar adjacent genetic loci. However, there are several differences as well. First, while these two other bacterial species express a second OMP whose abundance is regulated in an inverse manner (e.g., when V. cholerae OmpU expression is depressed, OmpT is increased, and vice versa) (30, 48), no analogous protein was detected in V. fischeri (Fig. 5). A second difference is that the regulation of expression of the V. fischeri OmpU was distinct from that reported for either OmpU in V. cholerae or OmpL in P. profundum. For instance, unlike these two OMPs (30), expression of the V. fischeri OmpU protein is not significantly affected by a null mutation in the porin regulatory protein ToxR (Fig. 5). This difference may not be too surprising, since other Vibrio species have also shown variations in OMP numbers and regulation (16, 17, 19, 31). Similarly, during the evolution of some but not all Vibrio species, a porin regulon has even been recruited to control the expression of horizontally transferred virulence determinants (30). Taken together, these findings suggest that there has been considerable functional and adaptive divergence in the biology of OMPs within the Vibrionaceae.

The fact that the expression of certain Vibrio OMPs responds to physiological changes in the environment has long been recognized (3, 24) and has indicated a role for these proteins in (i) the transport of specific nutrients and (ii) protecting the bacterium from disruptive chemical agents. An examination of the growth of the V. fischeri ompU mutant in media of different compositions did not identify any nutrients (e.g., peptides, inorganic iron, or phosphate) whose transport required OmpU. In contrast, the V. fischeri ompU mutants expressed an increased sensitivity to bile and other detergents (Fig. 6) that was similar to what has been reported for a V. cholerae ompU mutant (31). The absence of an obvious chemical or structural similarity between the anionic detergents and the cationic protamine further suggests that the effects of these agents may not be due to a change in the mutant of a specific structure in the outer membrane. Thus, while the functional basis for the V. fischeri ompU mutant's increased sensitivity to cell membrane-disrupting agents remains unclear, it may simply indicate that this protein plays a significant role in the integrity of the outer membrane. The fact that 30 to 60% of the OMPs in V. cholerae are OmpU (2) indicates the potential importance of this protein and provides support for this hypothesis.

Perhaps the most intriguing phenotype of the V. fischeri ompU mutant is its inability to colonize the squid light organ normally. The process by which juvenile E. scolopes become infected is typically a well-programmed and predictable one (22). If they are present in the ambient seawater, V. fischeri cells become aggregated on mucous strands emanating from the nascent light organ of newly hatched squid (27). After a period of 2 or 3 h, these cells begin to move out of the aggregates and towards the pores that lead to epithelium-lined crypts inside the nascent light organ. Only V. fischeri cells are able to survive and complete this process, and within 5 h they have begun to proliferate into a symbiotic population of several hundred thousand (44). The progression of this process is remarkably consistent and is reflected in the onset and level of luminescence emitted by the squid (33). The ompU mutant, while able to colonize the light organ, is defective in at least two aspects of the process. First, it initiates the colonization significantly more slowly and with less than 70% of the effectiveness of its ompU⁺ parent (Fig. 7). Interestingly, the delay they exhibit is similar to that seen when wild-type cells are treated with an antibody that reacts with OmpU (Fig. 2), suggesting that the inability of the cell to present this protein on its surface is responsible for this initiation defect.

The second defect expressed by the mutant is that, on average, the population level it can achieve in the light organ is only 20 to 25% of that of the parent strain (Table 2). The reason for this reduced colonization effectiveness is unknown, but it may indicate either that the cells have a diminished capacity to survive some condition within the crypts or, alternatively, that the host reacts to the mutant by providing fewer nutrients to support the proliferation of the bacterial population (9).

A method that has proven useful for examining the basis of symbiotic defects in V. fischeri mutants has been to determine their ability to compete with wild-type cells in mixed-colonization experiments. Previous work has suggested that such competition assays reveal subtle defects in mutants that, as a monoculture, would otherwise colonize to normal levels (43, 46). Thus, it was surprising to see no evidence of a competitive disadvantage in the ompU mutant. Instead, when it was coinoculated with its parent, there were twice as many cells of the mutant present in the light organ (average = 3.4 × 10⁵) than when it was the sole strain in the inoculum (average = 1.6 × 10⁵; Table 2). This result suggests that the presence of the parent strain provided an activity that complemented the ompU defect in the mutant, i.e., both the mutant and parent shared the benefit resulting from the activity. The basis for this effect is as yet unknown, but we hypothesize that one way in which OmpU may function in the symbiosis is to attach to a receptor on the host epithelium. This attachment may initiate a host response, such as provision of nutrients, that is required to support a normal level of symbiotic colonization. Support for this hypothesis comes from reports of bacterial OMPs that bind to host tissue and specifically modify its activities. Examples of such OMPs include Opa and Opc of Neisseria spp. (18) and OmpA-like proteins in Acinetobacter spp. (28) and members of the family Enterobacteriaceae (36). Future investigations will be focused on testing the hypothesis in V. fischeri and identifying the mechanism(s) underlying the role of OmpU in promoting benign colonization by V. fischeri.