Culturing environmental isolates
The microorganisms cultured from L. pertusa had 16S rRNA gene sequences that were closely related to bacteria previously isolated and cultured from a variety of marine environments. The growth medium used in this study (GASWA) is nonspecific and nutrient rich, selecting for bacteria that grow quickly on solid agar. GASWA is commonly used in culture-based studies of shallow-water coral microbial communities (Ritchie & Smith, 1995; Ritchie, 2006; Mao-Jones et al., 2010) and has also been used to culture bacteria associated with deep-sea gorgonians (Gray et al, 2011). The majority of isolates was related to Gammaproteobacteria, with a heavy representation of Vibrionaceae and Pseudoalteromonas (Tables 1 and 2). Culture-independent studies of bacteria associated with the white color morph of L. pertusa have found a higher diversity of bacterial 16S rRNA genes, with large numbers of clones representing the Alphaproteobacteria, Bacteroidetes, Tenericutes, and Actinobacteria in addition to Gammaproteobacteria (Neulinger et al., 2008; Kellogg et al., 2009). No overlap was observed between cultured isolates and clones from previous Lophelia investigations.
Isolates phylogenetically designated as Halomonadaceae (including Halomonas sp. and Cobetia sp.) were only cultured from corals at VK 826 and only from the two dives in 2005 (Table 1). All isolates were 99% identical to their top GenBank match. All of the GenBank matches were marine in origin, with some existing as free-living (planktonic) or sediment-associated bacteria. The top match for six Halomonadaceae isolates was ‘uncultured bacterium clone CB-14’ (accession #HM100916), which is potentially symbiotic or commensal with a marine sponge; the closest cultured GenBank matches for these isolates were all Cobetia spp.
Bacteria with high similarity (≥98%) to marine Pseudoalteromonas spp. were isolated from all dives (Tables 1–3). Many had GenBank matches that could be considered potentially symbiotic, with hosts including euphasiids, anemones, red algae, shallow-water coral mucus, and multiple bryozoan species. The majority of these GenBank matches come from cultured Pseudoalteromonas spp., not clones. Overlap between the two sample sites was observed, with isolates from both sites and both years matching the same GenBank top hit (i.e. Pseudoalteromonas sp. B149).
Psychrobacter-like bacteria were isolated from both sites, but only in 2004. Six bacterial isolates were identified as Psychrobacter spp., although they were most similar to only two sequences in GenBank, indicating the low culturable diversity of this genus on Lophelia. The GenBank matches were cultured from seawater and deep-sea sediment.
There was a high abundance of Vibrionaceae represented in the culture samples. Photobacterium, Aliivibrio, and Vibrio were all isolated from L. pertusa. All except three Vibrio-related VK 826 isolates and all VK 906/862 isolates were cultured in 2005, which could be an artifact of the culturing procedures (e.g. heterogeneity of mucus/tissue slurry) or an indication of a temporal variation in the coral-associated bacterial communities. Bacterial isolates from VK 826 were dominated by Photobacterium spp., particularly Photobacterium sp. 42X-1a4 (accession #EU440051), which was originally isolated from infected krill tissue, indicating potential pathogenicity. However, other Photobacterium-like isolates (e.g. 4878K2-B30 and 4878K2-B5) were most similar to bacteria that had been cultured from squid light organs, a symbiotic lifestyle. More diversity was seen among the Vibrionaceae at VK 906/862, with the L. pertusa isolates matching to sequences in GenBank that include microorganisms associated with fish intestines (FJ456798, AM181657), sediments (AM501622, EF114129), and sponges (HM100897).
Shewanella-like bacteria were only isolated in 2005 from corals at VK 906/862. The L. pertusa isolates were 98–100% identical to their closest blast matches in GenBank, all of which were cultured. Two isolates were most similar to Shewanella-like sequences that are potentially symbiotic: Shewanella sp. B225 (FN295775) was isolated from a bryozoan and Shewanella sp. J327 (AY369989) was isolated from a deep-water sponge. A single bacterium from VK 906/862 was most closely related to Pseudomonas sp. AM04 (GQ483506), which produces a biosurfactant.
The drastic differences between culturable bacteria and bacterial sequences recovered using culture-independent methods from the same samples have been documented previously (Rohwer et al., 2001). Bacterial isolates from dive 4753 (Tables 1 and 3) can be compared with the 16S rRNA genes cloned from the same corals in an experiment conducted by Kellogg et al. (2009). Branches from the same coral were either preserved at depth using DMSO/EDTA/salt buffer or brought to the surface without being preserved. The branch that was brought to the surface live was subdivided: DNA extraction was performed on several polyps (see Kellogg et al. 2009 for full details) and the rest were homogenized into a slurry to spread-plate on growth media. Comparison of the clone libraries from the DNA extractions and the number of cultured bacteria shows large differences. The clone library of one coral colony (4753K4) that was preserved at depth had representatives of Alpha- and Gammaproteobacteria, Tenericutes, and Bacteroidetes. The unpreserved sample from the same coral (brought to the surface live and immediately DNA extracted) was dominated by Tenericutes, with small fractions of Alpha- and Gammaproteobacteria and Bacteroidetes. The cultured representatives from that coral (Table 1) are only from the Gammaproteobacteria, comprised of Pseudoalteromonas spp. and Psychrobacter spp. Recognizing the inherent biases in culturing bacteria from the environment is important, but culturing and isolation of bacteria is necessary for detailed studies of physiology and ecological function. The use of an assortment of media types and growth condition variables can aid in increasing the diversity of microorganisms recovered by culturing, an ongoing experimental endeavor.
Potential function of cultured bacteria
Pseudoalteromonas spp. and Vibrio spp. are commonly recovered when culturing bacteria from corals (Rohwer et al., 2001; Dobretsov & Qian, 2004; Lampert et al., 2006; Ritchie, 2006; Bally & Garrabou, 2007; Brück et al., 2007; Chimetto et al., 2008; Raina et al., 2009) and sponges (Lee et al., 2007; Mangano et al., 2009; Menezes et al., 2009), indicating their ubiquity in the marine environment, as well as their possible symbiosis with marine organisms. Hypotheses on the ecological or the metabolic function of these community members are varied. Pseudoalteromonas spp. are known to secrete bioactive compounds with antifouling functions (Dobretsov & Qian, 2004), and both Vibrio spp. and Pseudoalteromonas spp. can produce antibiotics (Ritchie, 2006). These findings may suggest a protective role, aiding L. pertusa in clearing or preventing the settlement of epibiotic organisms. Additionally, Vibrio spp. associated with a Brazilian coral have been shown to fix nitrogen (Chimetto et al., 2008). Vibrio spp. have also been shown to break down recalcitrant carbon sources, such as cellulose and lignin (Neulinger et al., 2008) and dimethyl-sulfoniopropionate (a sulfur-organic compound released by phytoplankton, Raina et al., 2009). Unlike zooxanthellate shallow-water corals, L. pertusa relies completely on capture-feeding for nutrition (Freiwald & Roberts, 2005; Roberts et al., 2006) and could supplement feeding by maintaining a community of microorganisms that cycle necessary nutrients such as nitrogen or break down carbon sources unusable by the coral.
While Vibrio spp. are often implicated in diseases of corals (Kushmaro et al., 1997; Ben-Haim et al., 2003; Hall-Spencer et al., 2007; Thurber et al., 2009), the corals collected for this study were apparently healthy at the time of collection. No diseases affecting L. pertusa have been identified through sampling, video footage, or still photographs. In addition, the prevalence of culturable Vibrio spp. on healthy corals indicates that they are normal members of coral microbiomes (Dobretsov & Qian, 2004; Lampert et al., 2006; Ritchie, 2006; Brück et al., 2007; Chimetto et al., 2008; Raina et al., 2009). This suggests that Vibrio spp. are playing a symbiotic role that may be disrupted in times of stress for the coral, leading to a population explosion that results in physical signs of disease and over-representation in microbial studies of disease.
All bacteria in this study were isolated and cultured from plates that were maintained between 4 and 10 °C, in an effort to maintain in situ temperatures. As noted by Gray et al. (2011), temperature can bias the number and diversity of bacteria recovered from cold-water coral samples. They found that fewer bacteria were recovered from plates incubated at 22 °C than at 4 °C, and of those bacteria, there was a higher percentage of Vibrio spp. isolated at the warmer incubation temperature (Gray et al., 2011). This finding also indicates that high-temperature stress on in situ Lophelia colonies could cause the community of microorganisms to shift, allowing more Vibrio-like growth and inhibiting the growth of other microorganisms.
Kirby–Bauer antibiotic susceptibility testing
Effect of incubation length on the zone of inhibition size
In order to apply the Kirby–Bauer test to bacteria isolated from the cold-water coral L. pertusa, the procedure was altered so that the plates were incubated at 4–7 °C for up to 30 days, instead of 37 °C for 24–48 h. Maintaining the bacterial cultures under conditions as close as possible to their natural environment was deemed important to obtain environmentally relevant data, and long incubation times were required to observe sufficient bacterial growth. However, concerns were raised that increasing the incubation time could lead to antibiotic breakdown (leading to smaller zones of inhibition) or greater antibiotic diffusion into the agar (leading to larger zones of inhibition). To address these concerns, incubation length was measured in hours and plotted against the zone of inhibition diameter. An extremely low correlation was observed (maximum r2=0.19) for all regression lines, and the low significance (Fs<1.0 for all, anova) of these regression lines suggests that the size of the zones of inhibition could not be well explained by the length of incubation (Supporting Information, Fig. S1).
Variability of Kirby–Bauer profile among grouped/closely related bacteria
Discussion of antibiotic susceptibility as measured using the Kirby–Bauer test is limited first by the number of cultivable bacteria isolated out of the total bacterial community associated with the coral holobiont, and secondly, by the subset of those bacteria that can grow on Muller–Hinton II agar amended with salt. Surprisingly, very few Vibrionaceae grew on the Mueller–Hinton II agar, indicating that some essential nutrient was missing, was present in the wrong amount, or that an inhibitor specific to that family was in the medium. The purpose of this test was to examine antibiotic sensitivity differences between bacterial isolates classified as the same species based on their 16S rRNA gene sequences. Lampert et al. (2006) used the same six antibiotics to examine strain-level differences between bacterial cultures isolated from mucus of the shallow-water coral Fungia scutaria.
Overall patterns of susceptibility to the six antibiotics were represented across the different phylogenetic groups: the majority of isolates were susceptible to three or four antibiotics, while five isolates were susceptible to five of the antibiotics, and a single isolate (4746K8-B6) was susceptible to only two antibiotics (Fig. 2). Individual patterns of susceptibility could vary within the phylogenetic groups, but were repeated between groups (e.g. 4878K4-B1, 4746K6-B13, and 4753K4-B4 share the same profile, but are in different phylogenetic groups). A study of culturable chloramphenicol-resistant bacteria from coastal waters in Jiaozhou Bay, China, revealed that the majority was identified as Pseudoalteromonadaceae sp. (Dang et al., 2008). The single bacterial isolate from the current study that was intermediately susceptible, indicating slight resistance to chloramphenicol, is also a Pseudoalteromonas sp.; however, all of the remaining bacteria were susceptible (Fig. 2).
Differing patterns of antibiotic susceptibility would suggest strain-level differences in accessory or antibiotic-resistance genes either within the genome or encoded on plasmids, integrons, or transposons in the bacterial isolates (Thomas & Nielsen, 2005; Allen et al., 2010). It has been shown that while bacteria maintain a core set of ‘housekeeping genes’ necessary for function (i.e. 16S rRNA genes), there is continuous swapping of accessory genes (e.g. antibiotic resistance genes; Staley, 2006). The definition of bacterial species is almost solely anchored on phylogenies constructed with core genes; however, the ecological role of the bacteria is highly dependent on the suite of accessory genes (Tettelin et al., 2005; Staley, 2006). Testing bacterial function within the coral holobiont, as well as phylogenetic identity, will aid researchers in deciphering the role of bacterial communities associated with corals.
Isolates with shared GenBank matches, which putatively identified them as the same bacterial species, showed differences in their antibiotic resistance profiles. For example, bacteria related to Pseudoalteromonas sp. B149 were isolated from both sites during both 2004 and 2005, indicating that this bacterium is temporally stable at the two sites. However, the antibiotic resistance profiles between and within sites and years were variable, with differing susceptibilities to penicillin, chloramphenicol, and novobiocin (Fig. 2). Analogous variability can be seen among isolates related to ‘uncultured bacterium clone CB-14,’ a Halomonas species. These isolates were cultured from site VK 826 in 2005 on two different dives and show variable antibiotic resistance profiles. Notably, one of the isolates (4881K6-B4) was susceptible to clindamycin, one of only two strains to exhibit this trait.
Differing antibiotic resistance profiles among bacterial strains that would be grouped together based on their 16S rRNA gene phylogenies illustrate the difficulty of assigning species identifications anchored solely by a single gene. For example, the identification of Vibrio spp. to the species or the strain level requires the use of multilocus sequencing (Thompson et al., 2005; Pollock et al., 2010). The variability in antibiotic resistance profiles could indicate important strain-level differences. At the minimum, it cautions against relying too heavily on identifications based on single genes in order predict ecological roles.
Natural antibiotic resistance
Antibiotic resistance in nonclinical or environmental settings is not necessarily an indication of an anthropogenic influence, although pharmaceutical waste has been implicated in influencing antibiotic resistance in natural microbial populations (Allen et al., 2010). Rather, there seems to be a constant, low-level existence of antibiotic resistance genes flowing throughout natural populations (Yim et al., 2007; Allen et al., 2010). In addition, antibiotics are found at subinhibitory or sublethal concentrations in the natural environment (Yim et al., 2007) and are commonly produced by microorganisms during the stationary growth phase (Fajardo et al., 2009). Gene expression studies have shown that sublethal doses of antibiotics can induce phenotype changes (e.g. biofilm production) or transcription patterns shifts (Davies et al., 2006; Fajardo et al., 2009). The results of these studies suggest that the natural molecules from which clinical antibiotics are derived can function as cell-to-cell signals in the environment, an important ability for microorganisms existing in a community structure (Davies et al., 2006; Yim et al., 2007; Fajardo et al., 2009). The concentrations of antibiotics used in Kirby–Bauer testing are meant to be bacteriostatic or bactericidal. An interesting experiment would be to observe possible phenotype changes when bacterial isolates are exposed to sublethal doses of the same antibiotics.
Initial microbiological studies of bacteria associated with the cold-water coral L. pertusa have all relied on culture-independent methods (Yakimov et al., 2006; Neulinger et al., 2008; Kellogg et al., 2009; Schöttner et al., 2009). This paper is the first to present the results from culture-based bacterial surveys. Culture-based experiments are significantly limited in the bacterial diversity recovered (Fuhrman & Campbell, 1998); however, they provide important information about the physiological capabilities of the microorganisms. Our results show that bacterial function is not necessarily tied to phylogeny, hinting at a cryptic functional potential in these bacterial isolates.