Predatory Bacteriovorax Communities Ordered by Various Prey Species

The role of predation in altering microbial communities has been studied for decades but few examples are known for bacterial predators. Bacteriovorax are halophilic prokaryotes that prey on susceptible Gram-negative bacteria. We recently reported novel observations on the differential selection of Bacteriovorax phylotypes by two different prey, Vibrio parahaemolyticus and Vibrio vulnificus. However, the conclusion is restricted by the limited number of prey tested. In this study, we have conducted two independent investigations involving eight species of prey bacteria while using V. vulnificus and V. parahaemolytics as reference strains. Water samples collected from Dry Bar, Apalachicola Bay were used to establish microcosms which were respectively spiked with prey strains Vibrio cholerae, Escherichia coli or Pseudomonas putida to examine the response of native Bacteriovorax to freshwater bacteria. Indigenous Vibrio sp., Pseudoalteromonas sp., Photobacterium sp. and a clinical strain of V. vulnificus were also tested for the impact of saltwater prey on the Bacteriovorax community. At 24 hour intervals, optical density of the microcosm samples and the abundance of Bacteriovorax were measured over five days. The predominant Bacteriovorax plaques were selected and analyzed by 16S rRNA gene amplification and sequencing. In addition, the impacts of prey on predator population and bacterial community composition were investigated using culture independent denaturing gradient gel electrophoresis. Strikingly, Cluster IV was found consistently as the predominant phylotype produced by the freshwater prey. For all saltwater prey, subgroups of Bacteriovorax phylotype IX were the major predators recovered. The results suggest that prey is an important factor along with temperature, salinity and other environmental parameters in shaping Bacteriovorax communities in aquatic systems.


Introduction
Predation is an important service to the environment in maintaining population balances among organisms and food webs [1,2]. The control of bacterial communities by predation has been known for several decades; however, the greatest progress in uncovering the types and roles of predators has occurred in the past 50 years with the discovery of Bdellovibrio and like organisms (BALOs) [3], millions of bacteriophage in aquatic systems [4,5] and an improved understanding of the activities of protists [6]. Most investigations on bacterial predation have focused on the role of viruses and protists. The impact of BALOs in altering bacterial communities through predation is only beginning to be understood. Genera of these obligate predators share a unique life cycle consisting of two distinct phases, the predatory extracellular, attack phase in which the cells are highly motile to facilitate predation, and an intraperiplasmic growth phase in which they penetrate the cell wall and become lodged in the periplasmic space where they grow, multiply and finally free themselves by lysis of the prey cell [7]. Evidence strongly suggests they exert a potential sideways control, a mechanism by which bacterial predators prey on other prokaryotic or eukaryotic cells. In this way, they alter the structure, function, and dynamics of bacterial communities [8,9].
BALOs attack Gram negative bacteria, however, not all are susceptible to the predators and among those that are, not all are preyed upon with equal efficiency [10,11,12]. We recently reported novel observations on the differential selection of Bacteriovorax, a saltwater genus of BALOs, by two different prey, V. parahaemolyticus and V. vulnificus [13]. When an environmental water sample was amended with high numbers of the respective prey, the progeny yield from V. vulnificus over a five day period was primarily restricted to two Bacteriovorax phylotypes, Clusters IX and X. Conversely, V. parahaemolyticus yielded multiple phylotype clusters, up to five in one case, which typically varied from day to day.
However, that study included only two bacteria prey, both from the same genus which limit the conclusions that can be drawn. To further explore this phenomenon we have conducted two independent investigations involving eight species of prey bacteria. V. vulnificus and V. parahaemolyticus were included as reference strains. Moreover, the impacts of the amended prey and subsequent increase in the predator population and bacterial community composition (BCC) were investigated using a culture independent approach. The results are described in this report.

Sample collection
Water samples were collected at site Dry Bar in Apalachicola Bay, Florida USA (N 29u409130; W85u059390) on three occasions designated as DB4, DB5 and DB6. On each occasion, bottom water was collected from both sides of a National Estuarine Research Reserve research vessel (25-foot, C-Hawk) using a sterile sampler at a depth of approximately 1.74 m. Environmental parameters were measured and recorded on site ( Table 1). The water samples were stored on ice and transported to the laboratory at Florida A&M University for the setup of the microcosms within 6 h of collection. No specific permits were required for sampling in the above location.

Bacterial strains and culture conditions
Two prey species, V. vulnificus FLA042 (Vv) and V. parahaemolyticus strains P-5 (Vp) were included as reference strains in all microcosm experiments because their impact on shaping predator communities has been previously reported [13]. Other bacteria used in microcosm experiments included freshwater bacteria strains Vibrio cholerae, Escherichia coli or Pseudomonas putida which were selected from our laboratory culture collection and are known to be susceptible to Bacteriovorax. The saltwater bacteria used were indigenous species isolated from DB5 waters onto Luria-Bertani (LB) culture plates (Difco, Sparks, MD, USA). After obtaining pure cultures, DNA of the isolates was extracted by boiling and 16S rRNA fragments were PCR amplified, sequenced and blasted against NCBI database to obtain their phylogenetic identity. Subsequently, isolates of Vibrio sp., Pseudoalteromonas sp., and Photobacterium sp. were selected for the microcosm experiments for their known susceptibility to Bacteriovorax.
Prey suspensions used to amend the microcosms and for plating for Bacteriovorax recovery were prepared by adding 5 mL of sterile 70% artificial sea water (Instant Ocean Aquarium Systems, Inc., Mentor, OH, USA) (pH 8, Salinity 22 p.p.t.) to an overnight culture on LB plates. The bacterial colonies were suspended in the liquid for subsequent usage.

Establishment of laboratory microcosms
For both experiments, water samples were mixed and filtered through a 0.8 mm filter to remove debris and larger organisms such as some protists for the establishment of the microcosms. Five hundred ml of the filtrate was dispensed into each of four 2 L Erlenmeyer flasks. For subsequent analysis by denaturing gradient gel electrophoresis (DGGE), another 500 ml of the filtrate was filtered through a 0.1 mm filter to capture the microbial populations, including Bacteriovorax, on the filters which were stored at 220uC.
To investigate the response of Bacteriovorax communities to the freshwater bacteria, microcosms established with DB4 water samples were amended with V. cholerae, E. coli or P. putida, respectively, in addition to the two reference strains, Vv and Vp.
To access the effect of salt water prey bacteria indigenous to Dry Bar water, Vibrio sp., Pseudoalteromonas sp., and Photobacterium sp. isolates were spiked into microcosms consisting of DB6 water. A clinical strain of V. vulnificus mo6 (Vv2) was also included in this experiment to test against reference strain Vv for strain-specific variations in selecting for Bacteriovorax.
Suspensions of the prey bacteria were spiked into the respective flasks described above to yield an optical density (OD) measurement of 0.7 at 600 nm except Photobacterium sp. of which OD was adjusted to 0.26. This corresponds to approximately 5610 8 cells ml 21 as predetermined by enumeration on LB agar plates. Control microcosms established to monitor prey abundance without interference from Bacteriovorax or other microorganisms consisted of equal volumes of prey as in the test microcosms in autoclave-sterilized environmental water. The microcosm flasks were shaken at room temperature and monitored at 24 h intervals through 120 h. Subsequent sample processing assays were adapted from Chen et al. [13]. Briefly, at each 24 h interval, OD measurements (at 600 nm) were taken of samples from the test and control microcosms. Aliquots were removed from test microcosms, serially diluted and plated using the double agar overlay technique [14] for isolation of predominant Bacteriovorax strains using the same prey as in the microcosm. Following incubation, plaque-forming units (PFU) were counted and recorded.
To identify the predominant population, Bacteriovorax plaques appearing on plates of the highest dilution were picked for 16S rRNA gene amplification using Bacteriovoracaceae specific primers (Bac-676F, Bac-1442R) [15]. This was followed by purification with the QIAquick PCR-Purification Kit (QIAGEN) and sequencing using Bac-676F primer at the DNA Sequencing Laboratory at Florida State University. DNA sequences and homology searches were analyzed with the Basic Local Alignment Search Tool (BLAST) server from the National Center of Biotechnology Information (www.ncbi.nlm.nih.gov). Phylogenetic clusters of Bacteriovorax were assigned based on 96.5% or higher 16S rRNA gene sequence similarities with the strains described in previous reports [9,16,17]. Complementary to this cultural dependent technique, denaturing gradient gel electrophoresis (DGGE) was also used to monitor shifts in the Bacteriovorax and bacterial communities in the microcosms using universal bacterial primer GM5F-GC and 907R. DGGE bands representing the most prominent operational taxonomic units (OTUs) were excised and the DNA eluted and re-amplified using primers without the GC clamp and sequenced. These sequences were checked for chimeras using the Bellerophon Chimera Check (version 3) [18] and have been submitted to the GenBank databases under accession numbers (JQ612074-JQ612128). The sequences were taxonomically characterized by Basic Local Alignment Search Tool (BLAST).

Data analyses
The abundance of predator and prey (log transformed) were analyzed by analysis of variance (ANOVA) to detect significant differences among the numbers of bacteria in the various microcosm treatments. When ANOVA tests were passed, the Holm-Sidak test was performed. The T-test was used to compare two groups of treatments when normality and equal variance tests were passed. All statistical analyses were performed using the Sigmastat, version 3.5, software package.  For DGGE analysis, Quantity-one software (version 4.0, Bio-Rad, USA) was used to determine the presence, intensity and relative position of each band. Each DGGE band was assumed to represent an OTU or phylotype.

Reduction in prey abundance by Bacteriovorax predation
In the two independent experiments with laboratory microcosms, the predator and prey responses exhibited similar patterns ( Fig. 1, 2 , 3, 4). In all cases, the inoculated prey bacteria decreased significantly (ANOVA, p,0.01) with a simultaneous increase in Bacteriovorax numbers, indicating the predation of the spiked prey by the predator. These results are consistent with that of our previous report [13]. The greatest decreases in OD measurements were generally observed after 24 to 48 h except for the microcosms spiked with V. cholerae, which occurred after 72 h indicating a delayed response of Bacteriovorax to this prey ( Fig. 1). As expected, the initial concentrations of Bacteriovorax in the water samples before adding the test prey were very low, ranging from below detectable levels up to 10 PFU ml 21 . However, the Bacteriovorax population grew rapidly on the spiked bacteria. In microcosms with freshwater prey, Bacteriovorax numbers peaked around 10 8 PFU ml 21 at 72 h and remained relative constant until 120 h (Fig. 2). This is a typical response of the predators to high concentrations of prey bacteria and shows their ability to rapidly increase in number going from a rare dormant species to a dominant active population to control and reduce the prey population. The Bacteriovorax numbers in microcosms with saltwater prey also showed similar trends, albeit at a higher growth efficiency as revealed by a sharp increase within the first 24 h that peaked after 48 h of incubation. Notably, Bacteriovorax grew at a slower rate on Photobacterium sp., gradually reaching 10 6 PFUs ml 21 after 120 h of inoculation (Fig. 4).

Analysis of predator communities amended with freshwater prey
Bacteriovorax phylogenetic Cluster IV was the predominant predator population emerging from the three freshwater prey (Fig. 5). Strikingly, Cluster IV was not recovered from the microcosms of the two reference prey, Vv nor Vp, (Fig. 5 a, b). This is consistent with our previous report that Vv typically and consistently selected for Bacteriovorax Cluster IX throughout the experiment Vv whereas with Vp the predominant predator phylotype population was less stable changing on a nearly daily basis. The number and diversity of phylotypes were also higher with several phylotypes being represented among those that were  dominant. Bacteriovorax population was observed in Vp microcosms. Although the dominant Bacteriovorax recovered from all three freshwater prey belonged to Cluster IV, the patterns were still slightly different. Bacteriovorax Cluster V appeared in the V. cholerae microcosm during the first 48 h and was gradually taken over by Cluster IV (Fig. 5 c). Bacteriovorax phylotype Cluster X appeared in high abundance at 48 h in E. coli microcosm (Fig. 5 d) and Cluster IX was found in P. putida microcosm at 48 h and 72 h (Fig. 5 e).

DGGE analysis of microcosms established with freshwater bacteria in DB4 water
A total of 32 prominent bands in the DGGE gel images were excised and sequenced (position shown in Fig. 6). The major predators responsible for Vv reduction are clustered to phylotype IX. For Vp microcosm, a prominent band (band 9) related to Cluster III is shown at 48 h, however, at later time points Cluster X was predominant (band 13). Cluster XI was also found in Vp microcosms (band 17). Bands related to Cluster IV were detected in V. cholerae, E. coli and P. putida microcosms (band 20, 24, 28, 30 and 34). Bands related to Bacteriovorax Cluster V were observed only in V. cholerae microcosms (band 21 and 22). Cluster IX was only detected in the P. putida microcosm at 72 h ( Table 2). The results reveal that the predator phylotypes profiled by DGGE are consistent with that of the culture dependent method described above.
Other bacterial species were also identified in the PCR-DGGE gel bands. Bacterioidetes were observed to thrive in all the DB4 microcosms after prey inoculation. The Vv and Vp amended microcosms enhanced the growth of Thalassospira sp. (Fig. 6. a), whereas, the E. coli uniquely facilitated the growth of Flexibacter sp. and Coccinistipes sp. (Fig. 6 b).

Analysis of predator community arising from saltwater indigenous prey bacteria
Strikingly, Bacteriovorax Cluster IX was the predominant predator recovered by the culture method in microcosms established with the saltwater native prey bacteria. The DGGE banding patterns of the microbial populations in these same microcosms are shown in Fig. 7. DNA sequences obtained from a total of 47 bands on the DGGE gels were used to determine phylogenetic analysis. The nearest neighbor and phylogenetic groups of each sequence are listed in Table 3. Consistent with the results of plaque analyses, bands related to Bacteriovorax phylotype IX were present in all microcosms. In addition, Bacteriovorax Cluster XII was found in Pseudoalteromonas sp. microcosm at 72 h and Cluster XIII was found in Photobacterium sp. microcosm after 48 h. Interestingly, another clade of Bacteriovorax phylotype which shared a similarity index of 89% with Peredibacter sp., a freshwater BALO sp., was also revealed by DGGE analysis (bands 15, 22, 26 and 37). This clade was identified in all microcosms after 72 h except for the one amended with Photobacterium sp.
Bacteria other than the prey and predators in the microcosms were detected by DGGE. Ruegeria sp. was detected from all microcosms except for the Vv microcosm. Thalassospira sp. was detected in Vp, Vibrio sp. and Pseudoalteromonas sp. microcosms but not in the Vv, Vv2 and Photobacterium sp. microcosms. Notably, sequences affiliated with the phylum Bacterioidetes were abundant in microcosms established with freshwater bacteria but were not detected in this experiment with indigenous prey. A sequence closely related to Roseivirga sp. 16S rDNA was found in the Photobacterium sp. microcosm (band 44).

Discussion
Previously, we reported the first direct evidence of the influence of prey bacteria on selection of predatory Bacteriovorax phylotypes. In that study we observed that two prey species of the same genus, Vv and Vp [13] typically yielded different Bacteriovorax phylotypes, although both were added to the same environmental sample and thus available to the same native predator population. This result was consistent across water bodies of varying salinities, including ocean, estuarine and the Gulf of Mexico, water temperatures and seasons.
In the current study, we conducted a more comprehensive investigation which included more bacteria of different genera with representatives from both freshwater and saltwater environments. Vv and Vp were also included as reference strains since we had prior knowledge of their impact from our previous study. The results confirmed our previous findings that the bacterial prey of Bacteriovorax influences the strains of predators that multiply within them and are released into the environment when the prey cell is lysed. Another surprising observation made was a distinct difference between the Bacteriovorax population produced by the freshwater and saltwater test prey bacteria. The three freshwater prey bacteria consistently yielded Bacteriovorax phylotype Cluster IV as the predominant predator. In contrast, Cluster IV was not found in the microcosms amended with the reference halophilic strains, Vv and Vp (Fig. 5). This is interesting since Cluster IV was consistently isolated using Vp as prey from low salt waters of the Chesapeake Bay. However, since Cluster IV is typically found in low salt aquatic environments it may encounter both fresh and salt water prey bacteria and has become adapted to preying on both, but apparently, in this case, more efficiently on the freshwater strains.
We also observed that Bacteriovorax Cluster V was only detected in the V. cholerae microcosm. Cluster V, like cluster IV, is a distinct estuarine strain which has been recovered only from low salinity areas of Chesapeake Bay and other estuarine systems , but not from marine or high salt ecosystems [19,20]. Perhaps Cluster V has a preference for freshwater prey such as V. cholerae and may thrive best in those regions where the preferred prey is present.
Contrary to the Bacteriovorax Clusters IV and V produced from the freshwater prey, the predominant predator phylotype recovered from the halophilic prey was Cluster IX. Isolates of Cluster IX have been reported to be ubiquitous and the most abundant Bacteriovorax group cultured from marine environments [19]. Wen et al, [21] attributed their ability to adapt to saltwater environments to the heterogeneity of their 16S rRNA operons, as only Cluster IX displayed multiple visible V3 bands in DGGE gels. This was also observed in our study which, in one case, showed that two DGGE bands from the same sample were related to Cluster IX with 16S rRNA percentage similarities of 99% and 96% (Fig. 7b, bands 17and 18). This unique characteristic of Cluster IX may explain their superiority in controlling halophilic prey species. Table 2. Identification of bacteria (based on 16S rDNA sequence similarity to the nearest neighbor from NCBI database) in samples retrieved from microcosms established with freshwater bacteria in DB4 water (See Fig. 6 for position of the bands). Previous reports have revealed that freshly isolated autochthonous bacteria appeared to be more lucrative prey for BALO isolates from the same habitat than laboratory maintained prey strains [22,23,24,25] from other sources. Our results support this conclusion as greater prey reduction was observed in the microcosms spiked with the native Vibrio sp. than in the  Table 3. Identification of bacteria (based on 16S rDNA sequence similarity to the nearest neighbor from NCBI database) in samples retrieved from microcosms established with indigenous saltwater prey species in DB6 water (See Fig. 7 for position of the bands). microcosms amended with the three laboratory Vibrio strains, Vv, Vv2 and Vp (Fig. 3). It warrants noting that in microcosms established with indigenous prey, another clade of BALO phylotype which clustered within Peredibacter, a freshwater genus of BALOs, was also revealed by DGGE analysis (Table 3). Chauhan et al., [25] also discovered this clade in water samples from Apalachicola Bay which were bio-stimulated with yeast extract. The fact that the bands only were detected in microcosms after 72 h implies that the growth of this group may be slower in salt water environments. Their later growth and appearance may have been stimulated by an increase of available organic matter in the microcosms resulting from the massive lysis of prey by other Bacteriovorax phylotypes. If this clade is indeed a member of freshwater BALO, it also suggests it is a more versatile predator which has higher salt tolerance than the freshwater strains which are typically restricted by salt concentrations as low as 0.5%.

DGGE
Our results demonstrate that not only are environmental physicochemical pressures among the factors which determine BALO phylotypes recovered from nature, but also the prey community upon which they feed. Moreover, not all phylotypes can be detected by a single, particular prey. BALOs which have either a different prey preference than the bacterium used in isolation or are uncultivable will remain unrepresented in biodiversity studies.
Both culture dependent and independent methods were applied in this study to detect BALO phylotypes and were found to result in detection of more phylotypes than either method alone. Typically the predator communities revealed by the two methods were largely consistent although slight individual variations were observed. For example, a more diverse Bacteriovorax community was depicted by culture methods than by DGGE in DB4 microcosm spiked with Vp (Fig. 5 b), whereas, in the DB6 water microcosm Clusters XIII and XII were detected only by the DGGE method (Table 3 bands 38, 43). One advantage of using DGGE with a universal primer is that it is not limited to detecting known Bacteriovorax taxa and therefore can identify novel BALO isolates. Using DGGE, a BALO clade related to Peredibacter was detected in DB6 water that could not be amplified with the Bacspecific primer. Overall, both methods when combined yielded a more detailed structure of the predator community, but it should be noted that they are only able to detect predominant members represented in the samples.
In the study, we also used DGGE profiling to evaluate the temporal variations in the total bacterial community structure following prey inoculation into the microcosms. The bacteria that became abundant after prey inoculation were Bacterioidetes, Ruegeria sp. and Thalassospira sp. Bacteriodetes are common microbes in coastal waters especially during algal blooms and are known for their ability to rapidly degrade complex dissolved organic matter (DOM) [26,27]. It is also not surprising to find Ruegeria sp. and Thalassospira sp. predominant in the microcosm. Both are groups of alpha-Proteobacteria which is also a major component of BCC in marine environments dominating particularly in low nutrient conditions [28]. It should also be noted that both genera are gram-negative bacteria. Since Bacteriovorax preferentially prey on certain gram negative bacteria, it may provide the opportunity for Bacteriovorax resistant bacteria to thrive. Interestingly, even prey of the same species (Vv and Vv2) caused BCC to vary (Fig. 7 a). For instance, bands related to Ruegeria sp. were identified in the Vv2 microcosm but not in the Vv microcosm (Table 3, band 13). This suggests that predator-prey interaction has a complex role in regulating BCC and the impact is strain specific.
These results should be interpreted with some caution as the numbers of selected prey introduced into the natural water microcosms were several fold higher than typically detected in environmental waters. This may have influenced the population dynamics of the other bacteria in a way that may not be observed when the prey numbers are within their natural range. It was necessary to use such large numbers of prey in this study to amplify and detect the response of the Bacteriovorax community. Although the typical numbers of bacteria and prey in the water column may be below the threshold for highly active Bacteriovorax predation, there are instances in which spikes in bacterial populations occur, for example in the environment niches, sediments and biofilm, and events such as phytoplankton blooms, sewage input and decomposition of dead plants or animals. Under these circumstances, we believe the Bacteriovorax and other BALOs respond as we have observed in the microcosms in this study, by aggressively attacking and killing its prey until the prey population is substantially reduced. Although BALO numbers are typically very low, and sometimes undetectable, in environmental waters, the results of this and other studies [13] have shown the predators are able to rapidly increase by seven or eight logs in response to high numbers of prey bacteria.
The results from this study of eight bacteria representing different genera and species and our previous report [13] confirm that prey bacteria play an important role in determining the community composition and structure of Bacteriovorax communities in salt water systems. This particular function of prey bacteria was not known previously and unveils another aspect of the mystique and complexity of the predator-prey relationship of the BALOs and bacteria that serve as their prey. Further, the results revealed distinct differences in the Bacteriovorax phylotype communities derived from growth of the predators on freshwater and saltwater prey bacteria. Also, this study confirmed that in terms of methods for investigating Bacteriovorax native communities, the most accurate approaches include the use of both culture and non culture molecular techniques.
Since our studies were performed using microcosms made from natural water and the native bacterial population, it provided the opportunity to observe changes in the total bacterial population and especially those that may have been influenced by the amended prey and emerging BALO population. The collective results represent a major advance in the understanding of Bacteriovorax predator-prey interactions. Future studies are necessary to gain greater understanding of the mechanisms which governs the observed selective predation which leads to production of diverse Bacteriovorax population.