INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The taxon Shewanella putrefaciens ("Pseudomonas putrefaciens") comprises a group of gram-negative oxidative and nonoxidative bacilli whose chief phenotypic attribute is the production of hydrogen sulfide gas (H₂S) on TSI slants (22). Although S. putrefaciens has been implicated occasionally as a human pathogen, it is most frequently recovered from nonhuman sources, including aquatic reservoirs (marine, freshwater, and sewage), natural energy reserves (oil and gas), soil, and fish, poultry, dairy, and beef products (14, 18, 20, 21). Most human isolates of S. putrefaciens occur as part of a mixed bacterial flora, clouding their clinical significance. However, a number of monomicrobic illnesses due to S. putrefaciens have been documented and include bacteremia, soft tissue infections, and otitis media (1, 2, 4, 13).

For over two decades, it has been known that S. putrefaciens is a genetically heterogeneous species. These conclusions are based upon the wide variation in G+C content (44 to 54 mol%), results of DNA-DNA reassociation studies, and numerical taxonomy investigations (14, 17, 18, 20, 21). As late as the 1980s, Gilardi (6) recognized three distinct biovars within S. putrefaciens, while the Centers for Disease Control and Prevention (CDC) (22) recognized two biotypes based upon carbohydrate oxidation patterns and growth on salmonella-shigella (SS) medium and nutrient agar containing high salt (~6%) concentrations. In 1990, Simidu et al. (19) proposed the name Shewanella alga for a tetrodotoxin-producing isolate recovered from red algae. Subsequent to this publication, Nozue and colleagues (15) found that strains of S. putrefaciens with high G+C contents (52 to 54 mol%) were genetically related to S. alga and not to the type strain of S. putrefaciens (ATCC 8071). Furthermore, the latter study found that all S. alga strains produced a hemolytic reaction on sheep blood agar while S. putrefaciens isolates lacked this activity.

Of 40 clinical isolates of S. putrefaciens reidentified by Nozue and others (15), 33 (83%) were found to be S. alga based upon DNA homology values and phenotypic criteria. Domínguez et al. (5) later reported the first two Danish cases of S. alga bacteremia, which had originally been mistakenly attributed to S. putrefaciens. A follow-up 1997 systematic investigation of 76 Shewanella strains found that 16 of 19 human isolates (84%) resided within the S. alga group, based upon various taxonomic criteria including 16S rRNA sequencing, ribotyping, and whole-cell protein profiles (21). These studies suggest that S. alga may be the predominant human pathogen within the genus.

Presently, it is unclear how easy it is to distinguish S. putrefaciens from S. alga in the clinical laboratory and whether or not differences in host tropism do occur and correlate with pathogenic characteristics and overt virulence. In light of these questions, we have analyzed the biochemical and enzymatic capabilities of 49 Shewanella strains from diverse sources (human and nonhuman) to address these issues and have performed pathogenicity studies on representative strains from each group. These studies serve as the basis of this report.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains. Forty-nine strains of Shewanella spp. (human [n = 31] and nonhuman [n = 18]) were investigated in this study. Five of these strains were cultures submitted to the Microbial Diseases Laboratory for identification. Two reference strains (ATCC 8071 and ATCC 49138) were obtained from the American Type Culture Collection (Rockville, Md.). The remaining 42 reference or wild-type strains were kindly provided by the following persons: A. Carnahan (Baltimore, Md.), A. Chipman (West Sacramento, Calif.), J. Graf (Bern, Switzerland), C. Kaysner (Bothell, Wash.), V. Knight (New Brunswick, N.J.), D. Lies (Milwaukee, Wis.), V. Pünter (Zurich, Switzerland), and T. Robin (New York, N.Y.). Most of these strains were received as S. putrefaciens. Working cultures were maintained in motility deeps at room temperature; typically, these isolates produced an orangish discoloration at the surfaces of these deeps upon prolonged storage. All assays were performed at 35°C, with the exception of eight strains recovered from Lake Michigan, which grew poorly or failed to grow at elevated temperatures; these strains were tested at 25°C.

Identification. Isolates that were motile, had oxidative metabolisms, were oxidase and catalase positive, ornithine decarboxylase positive, and DNase positive, and produced H₂S on triple sugar iron slants within 72 h of incubation were identified as belonging to the phenospecies S. putrefaciens. Several Shewanella strains failed to grow on DNase test agar (Difco Laboratories, Detroit, Mich.), and six isolates were nonmotile. All other reactions were uniformly positive for the 49 Shewanella strains studied. The biovar and biotype of each isolate were determined according to Gilardi (6) and Weyant et al. (22), respectively, based on acid production from sucrose and maltose, growth on SS agar, and growth in the presence of 6.5% NaCl. Species designations were determined by the criteria of Nozue et al. (15) by using the following tests: hemolysis on sheep blood agar, growth at 42°C, growth on nutrient agar containing 6.5% NaCl, growth on SS agar, and acid production from maltose and L-arabinose. Five strains each of S. alga and S. putrefaciens of clinical origin were also tested on the API 20E (bioMérieux, Hazelwood, Mo.), API NFT (bioMérieux), RapID NF Plus (Innovative Diagnostic Systems, Norcross, Ga.), and Vitek (bioMérieux) systems. All tests were performed according to the manufacturers' instructions.

Phenotypic characterization. Seventeen Shewanella strains (S. alga [n = 8] and S. putrefaciens [n = 9]) were evaluated further for production of a number of enzymatic activities with the API-ZYM test (bioMérieux), according to the manufacturer's instructions. Relative activity was defined as the summation of numerical values for each isolate divided by the total number of strains tested for each species (8). Similarly, the cellular fatty acid profiles for 14 Shewanella strains (S. alga [n = 7] and S. putrefaciens [n = 7]) were determined by the MIDI microbial identification system (Microbial ID Inc., Newark, Del.).

Virulence-associated factors and resistance markers. Ten Shewanella strains (S. alga [n = 5] and S. putrefaciens [n = 5]) were evaluated for potential virulence-associated characteristics, including resistance markers to select antimicrobial agents. MICs of penicillin, ampicillin, and tetracycline were determined for each strain by using the AB Biodisk E test (Remel, Lenexa, Kans.), as previously described (11). Hemolytic activity was tested by plate and broth assays and an agar overlay method; a strain of Edwardsiella tarda expressing a cell-associated hemolysin served as a positive control (9). The adhesive, invasive, and cytotoxic capabilities of each Shewanella isolate were evaluated with HEp-2 cell monolayers prepared in chamber slides (Lab Tek; Miles Laboratory, Naperville, Ill.), as previously described (7). The relative pathogenicity of each S. alga and S. putrefaciens strain was determined by 50% lethal dose (LD₅₀) assays with female Swiss Webster mice (12).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Biotyping of Shewanella isolates according to two separate schemes provided similar results (Table 1). Most clinical isolates (74%) belonged to Gilardi biovar 2 (CDC biotype 2), being sucrose and maltose negative while growing on SS agar and on media containing high NaCl concentrations. The only major difference noted between the Gilardi classification scheme (6) and that of Weyant et al. (22) was that biovar 3 isolates (sucrose, maltose, SS, and NaCl negative) were ungrouped by the CDC biotyping system. In contrast to the human isolates, biovar 1 (CDC biotype 1) predominated (67%) among nonhuman strains. These strains typically produced acid from maltose and/or sucrose and failed to grow on high salt-containing or SS agar. Based upon the recent taxonomic proposals of Nozue and others (15), biovar 2 (CDC biotype 2) strains would be identified as S. alga while all biovar 1 (CDC biotype 1) and 6 of 7 biovar 3 strains would be designated S. putrefaciens; the remaining biovar 3 strain was subsequently identified as S. alga. Clinically, S. alga was found to predominate (77%), while the majority of nonhuman isolates (89%) were confirmed to be S. putrefaciens (Table 1).

All 10 Shewanella isolates, when tested on the API 20E, API NFT, RapID NF Plus, and Vitek systems, were identified as S. putrefaciens, with one exception. Four of five S. putrefaciens isolates produced unacceptable profile numbers on the API 20E system (no. 0602026 and 0602006); the fifth strain generated a rare biotype number for S. putrefaciens. All S. alga strains yielded excellent identification as S. putrefaciens on the API 20E. The API NFT system identified all 10 Shewanella isolates as S. putrefaciens with good to excellent identifications, with one exception, also an S. putrefaciens strain (low discriminatory value, 48 h). RapID NF Plus identified all 10 Shewanella isolates as S. putrefaciens, with 99.9% accuracy. Similarly, all strains were identified by Vitek (97 to 99% accuracy) as S. putrefaciens, although three S. putrefaciens strains required 5 to 9 h of incubation before final identification, in contrast to the 4-h results for the other 7 strains. Carbohydrate reactions (arabinose and maltose) on the API 20E, API NFT, and Vitek systems, however, do permit most strains to be correctly assigned to the relevant taxa (S. putrefaciens and S. alga) if read manually after final identifications as S. putrefaciens.

Comparison of the biochemical and enzymatic properties of Shewanella species revealed a number of differences (Table 2). Hemolysis on sheep blood agar, as reported by Nozue et al. (15), was detected in all S. alga strains but only in a couple of S. putrefaciens isolates. Most S. alga strains exhibited this phenotype only after prolonged incubation (48 to 72 h), and the area of hemolysis was often irregular and difficult to detect. Other activities previously found to aid in the separation of S. alga and S. putrefaciens, such as growth at 42°C, growth on media containing high salt (6.5%) concentrations, and acid production from L-arabinose, sucrose, and maltose, were confirmed. We found a substantially larger number of S. putrefaciens strains that grew on SS agar than previously reported; most of these originated from nonhuman sources. With the exception of ribose, production of acid from carbohydrate oxidation was uniquely associated with S. putrefaciens. Sugar patterns, however, varied considerably among these isolates, with some being arabinose, maltose, and sucrose positive while others were positive for maltose only or were asaccharolytic (biovar 3 strains). Several new enzymatic activities were detected among select Shewanella isolates that to our knowledge have not been previously reported. These included tyrosinase, alkylsulfatase, chitinase, and elastase activities (Table 2); most of these enzymes were found in nonhuman isolates of S. putrefaciens. Most S. alga and S. putrefaciens strains produced a siderophore, as determined by Chrome Azurol S assays. This activity was weak, and five isolates (three S. alga and two S. putrefaciens isolates) failed to grow on this medium.

Select Shewanella isolates that grew well at 35°C were further characterized for enzymatic activity by using API-ZYM (Table 3) and for cellular fatty acid profiles with the MIDI system. Of the nine substrates attacked by one or both Shewanella species with API-ZYM, higher overall activities for seven of these enzymes were associated with S. alga. The single activity found to be stronger in S. putrefaciens was valine arylamidase, although this activity was extremely weak even in these strains. Both species produced uniformly strong alkaline phosphatase activity. An additional observation was that all S. alga strains consistently produced eight of these nine enzymes, the only exception being valine arylamidase. In contrast, S. putrefaciens was more heterogeneous, with four of the nine enzymes detected being not universally present in all isolates. Analysis of 14 Shewanella strains indicated that the predominant fatty acids were i-15:0, 17:18c, and 16:0; some strains produced large amounts of 16:17c (9 to 18%), while others produced negligible quantities. While most fatty acid peaks were fairly consistent among S. alga and S. putrefaciens strains tested, several differences were noted (Table 4). Higher mean values of pentadecanoic acid and cis-9-heptadecenoic acid (17:18c) were noted for S. alga, while the converse held true for S. putrefaciens regarding hexadecanoic and dodecanoic acids. For hexadecanoic acid, no overlap in the total fatty acid range between S. alga and S. putrefaciens was observed; for pentadecanoic acid and 17:18c, only one S. putrefaciens or S. alga isolate produced a value that fell within the other's range. While no single peak was diagnostic, the use of all four peaks together clearly separated the 14 Shewanella strains into two groups along species lines.

It was found that the 23 strains of S. putrefaciens could be broken down into three separate groups based upon several phenotypic characteristics (Table 5). Group 1, which consisted of eight strains, including ATCC 8073, produced acid from maltose, sucrose, and arabinose and utilized urocanic acid. Group 1 strains were equally divided among clinical and nonhuman isolates. Group 2 strains (n = 6), which included ATCC 8071, were chiefly distinguished from group 1 strains by the inability to oxidize sucrose and maltose. Again, half of these strains originated from clinical material. Group 3 strains (n = 9), all of environmental origin (Lake Michigan area), differed dramatically from groups 1 and 2. They grew poorly or failed to grow at 35°C, produced chitinase, and were nonpigmented on L-tryptophan agar. All nine group 3 strains initially produced -glucosidase, but upon retesting only three strains were consistently positive. Maltose was oxidized, but not sucrose or arabinose. Unlike groups 1 and 2, urocanic acid was not utilized as a source of energy.

Recently, Vogel and colleagues (21) have noted differences between S. alga and S. putrefaciens in their susceptibilities to certain antimicrobial agents, including penicillin, ampicillin, and tetracycline. This, coupled with the report linking hemolytic activity with S. alga and its apparent association with human disease, suggests possible differences in pathogenicity between these two species (Table 6). We therefore selected 10 strains (5 from each species) for further analysis. Although we did not notice major differences in susceptibility to penicillin, ampicillin, and tetracycline between these two groups based upon the susceptibility category (susceptible, intermediate, or resistant), we did notice that the mean MICs for S. alga of penicillin, ampicillin, and tetracycline (~200, 56, and 5.2 µg/ml, respectively) were greater than the corresponding MICs for S. putrefaciens (3, 1.3, and 1.1 µg/ml, respectively).

Four of five S. putrefaciens strains attached weakly (+) to strongly (+++) (Table 6) to HEp-2 cells in adherence assays; in contrast, no S. alga strains exhibited similar adhesive characteristics, although four strains bound strongly to the glass slide background. Invasive activities were not detected in any Shewanella strain. Although a delayed hemolytic reaction was observed on sheep blood agar for all five S. alga strains (Table 2), beta-hemolysis was not detected with the agar overlay technique or broth assays (Table 6); control E. tarda strains were positive in 1 h in both tests. For all five S. alga strains (and one S. putrefaciens isolate), a weak cytotoxic reaction was sometimes observed during adhesion and invasion studies. This cytotoxic reaction was manifested by the appearance of HEp-2 cells with abnormal cellular morphology, including cell debris (ghosts). Differences in mouse pathogenicity between these two species were found, however, as the mean LD₅₀ in Swiss Webster mice for S. alga was 1.9 × 10⁸ CFU, while that for S. putrefaciens was 8.4 × 10⁸ CFU (P < 0.02).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In agreement with previous investigators (15, 21), we found S. alga, rather than S. putrefaciens, to be the predominant Shewanella species associated with clinical specimens. These species were easy to distinguish from one another by using a number of recommended phenotypic tests, such as growth at 42°C, growth in 6.5% NaCl, production of a hemolytic substance, and utilization of various carbohydrates (Table 2). Growth on SS agar, a differential trait noted in various studies, was found to be less useful in the present survey, as 52% of the S. putrefaciens strains exhibited this characteristic. However, this result may be a somewhat spurious observation, as most SS-positive S. putrefaciens strains originated from one geographic region (Lake Michigan and Green Bay). Although the S. alga strains typically were unable to utilize sugars, 35% of the isolates tested produced acid from D-ribose. Domínguez and colleagues (5) also found that two bacteremic strains of S. alga, plus the type strain, used D-ribose, while the type strain of S. putrefaciens did not. Thus, production of acid from D-ribose may be a marker for S. alga strains. Another species-associated trait was alkylsulfatase activity; >70% of the S. putrefaciens strains elaborated this enzyme, while only 8% of the S. alga strains were positive. We also found the S. alga strains to produce uniformly higher levels of enzymatic activities against a number of substrates in the API-ZYM system (Table 3). Of nine compounds utilized by either or both species, S. alga produced approximately two- to threefold-higher mean enzymatic activity against five of these substrates than did S. putrefaciens. This suggests that isolates might be presumptively screened for species determinations with this 4-h test, based upon the overall higher enzymatic activity of S. alga strains. Finally, we also noted that comparisons of fatty acid profiles of the two species revealed quantitative differences in four fatty acids that may be useful in species determination (Table 4). However, both of the latter sets of results need confirmation by evaluation of a larger number of isolates of each species.

A reflection of the phenotypic diversity found within S. putrefaciens was the recognition of three distinct biogroups within the species (Table 5). These biogroups were distinguished from one another by a number of phenotypic tests, including several new differential characteristics (-glucosidase, chitinase, and urocanic acid). Nozue and others (15) found that S. putrefaciens as presently defined is still heterogeneous at the DNA level, with at least three detectable genomospecies based upon DNA homology values. Clearly, further work needs to be undertaken to link specific phenotypic markers to each of these hybridization groups.

Why S. alga appears to be the predominant Shewanella species associated with human infections remains unclear. Results of limited pathogenicity studies on select Shewanella strains suggest that adherence to epithelial cells is not a major determinant. However, we did notice significant differences between S. alga and S. putrefaciens in mouse LD₅₀ studies, suggesting a possible explanation for the preferential association of the former species with human disease. One attractive explanation for this observation is the hemolytic reaction produced by virtually all S. alga strains, and some authors have suggested possible exotoxin involvement in S. putrefaciens cellulitis (3). Although we were able to detect this beta-hemolytic-type reaction upon prolonged incubation of isolates on sheep blood agar (48 to 72 h), other methods traditionally positive for the presence of a cytolytic toxin (beta-hemolysin), such as broth, agar overlay, and epithelial cell cytotoxicity assays, were either uniformly negative or weakly positive at best. Such results raise considerable questions as to the nature of this hemolytic reaction and whether or not it is caused by a true beta-hemolysin. Regardless, there appears to be increasing evidence that, between S. alga and S. putrefaciens, S. alga causes the most human illnesses and that significant differences exist between these two species regarding resistance to antimicrobial agents, mouse pathogenicity, and certain virulence factors (hemolysis and adhesion). Whether these factors translate into true in vivo differences with respect to disease spectrum, pathogenicity, and therapy remains to be determined.