|Home | About | Journals | Submit | Contact Us | Français|
Haloarchaea are the dominant microbial flora in hypersaline waters with near-saturating salt levels. The haloarchaeal diversity of an Australian saltern crystallizer pond was examined by use of a library of PCR-amplified 16S rRNA genes and by cultivation. High viable counts (106 CFU/ml) were obtained on solid media. Long incubation times (≥8 weeks) appeared to be more important than the medium composition for maximizing viable counts and diversity. Of 66 isolates examined, all belonged to the family Halobacteriaceae, including members related to species of the genera Haloferax, Halorubrum, and Natronomonas. In addition, isolates belonging to a novel group (the ADL group), previously detected only as 16S rRNA genes in an Antarctic hypersaline lake (Deep Lake), were cultivated for the first time. The 16S rRNA gene library identified the following five main groups: Halorubrum groups 1 and 2 (49%), the SHOW (square haloarchaea of Walsby) group (33%), the ADL group (16%), and the Natronomonas group (2%). There were two significant differences between the organisms detected in cultivation and 16S rRNA sequence results. Firstly, Haloferax spp. were frequently isolated on plates (15% of all isolates) but were not detected in the 16S rRNA sequences. Control experiments indicated that a bias against Haloferax sequences in the generation of the 16S rRNA gene library was unlikely, suggesting that Haloferax spp. readily form colonies, even though they were not a dominant group. Secondly, while the 16S rRNA gene library identified the SHOW group as a major component of the microbial community, no isolates of this group were obtained. This inability to culture members of the SHOW group remains an outstanding problem in studying the ecology of hypersaline environments.
Hypersaline waters, with salinities at or near saturation, are extreme environments, yet they often maintain remarkably high cell densities (≥107 cells per ml) and are biologically very productive ecosystems (27). The dominant microorganisms in such systems are haloarchaea, specifically members of the family Halobacteriaceae (5, 7, 27). A small percentage are extremely halophilic Bacteria, such as the recently described Salinibacter ruber (2).
Attempts have been made to identify the major microbial groups in salt lakes or salterns by the use of both molecular ecological and cultivation-based methods (3-5, 8, 23). While the results of these studies vary considerably, in general they indicate that there are dominant microbial groups that have not been cultured and that the most readily isolated and studied genera may not be the most significant in the in situ community. For example, the square haloarchaea of Walsby (SHOW group) can represent from 40 to 80% of cells present in salt lakes (1, 37), but no member has yet been cultured (4, 5, 27). In contrast, members of the genera Halobacterium, Haloarcula, and Haloferax are estimated from 16S rRNA analyses to make up only small proportions of in situ communities, but they are commonly found in cultivation studies (1, 3, 29, 37).
The first important steps toward an understanding of the ecology of hypersaline lakes include identifying the organisms present, assessing their numerical importance, and growing the dominant organisms in pure culture. While the sequencing of PCR-amplified 16S rRNA genes from DNAs extracted from environmental samples has proved to be a powerful and very reliable means of identifying the prokaryotes that are present and for estimating their approximate significance, cultivation has traditionally been considered far less useful. In many cases, dominant microbial groups are absent from or poorly represented in cultivation studies (4, 16). Recently, this picture has changed with dramatic improvements in culture methods for the dominant environmental species found in marine waters (32) and soil (20), providing considerable optimism that microorganisms identified by molecular ecological approaches in other environments can also be isolated and studied in laboratory culture.
Australia has many salt lakes and solar salterns, and the crystallizer ponds of commercial salterns are particularly useful model ecosystems due to their managed nature, as the salt concentrations are kept relatively constant. This system has proven to be a useful source of haloarchaea (for a review, see reference 14) and their viruses (11). While random isolations can provide useful model organisms, a knowledge of environmentally relevant haloarchaea and their viruses is of particular interest for a better understanding of this ecosystem. For this study, the archaeal composition of an Australian saltern crystallizer pond was examined by analysis of a library of PCR-amplified 16S rRNA genes and by parallel cultivation-dependent methods.
A sample from a crystallizer pond was collected aseptically from the Cheetham Salt Works (Geelong, Victoria; 38°09.841′S, 144°25.274′E) in March 2002. A portion was sent to a commercial water chemistry laboratory for analysis.
Crystallizer pond water and haloarchaeal cultures were examined by phase-contrast and fluorescence microscopy under a Leitz Diaplan microscope (Leica Microsystems AG, Wetzlar, Germany). For fluorescence microscopy, acridine orange was used (10 μg/ml) with para-phenylene diamine (0.5% [wt/vol]) as an antifading agent. Direct cell counts were made by use of a Neubauer hemocytometer (Weber, United Kingdom).
A 25% salt solution was used for plate cultivation experiments. It was prepared by dissolving 200 g of NaCl, 25 g of MgCl2 · 6H2O, 29.2 g of MgSO4 · 7H2O, 5.84 g of KCl, and 0.034g of NaNO3 in 700 ml of distilled water; 4.16 ml of 1 M CaCl2 was then slowly added, and the volume was brought up to 800 ml with distilled water. The solution was dispensed in 200-ml lots into 500-ml glass bottles along with 50 ml of a suspension of 15 g of washed Difco Bacto agar (final concentration, 1.5% [wt/vol]; see below). The agar was dissolved at 100°C and the medium was sterilized in an autoclave at 121°C for 15 min. After the medium was cooled to 55°C, the following solutions were added to each bottle: 5 ml of (NH4)2HPO4 from a 20 mM stock solution (final concentration, 0.4 mM), 0.5 ml of SL10 trace element solution (38), 0.25 ml of vitamin solution 1, 0.75 ml of vitamin solution 2 (18), and 0.25 ml of a selenite-tungstate solution (36). Finally, individual substrates were added to the final concentrations described below and plates were poured. The substrates and final concentrations in the five minimal media were as follows: (i) 0.01% nutrient broth (Oxoid Australia Pty. Ltd., Melbourne, Australia); (ii) glucose, galactose, arabinose, and xylose, 50 μM (each); (iii) glycerol, glycolic acid, and the sodium salts of pyruvate, l-lactate, and acetate, 50 μM (each); (iv) glycine betaine, 200 μM; and (v) 0.05% amino acid mix (15) with the addition of 0.8 g of a 100-ml stock solution of l-tryptophan. These media will be referred to (in the order described above) as 0.01% NB, sugar, organic acid, glycine betaine, and amino acid media. The medium MGM was prepared with 23% salts as described in the online protocol manual, the Halohandbook (http://www.microbiol.unimelb.edu.au/micro/staff/mds/HaloHandbook). Solid MGM was prepared with unwashed 1.5% (wt/vol) agar (Difco Bacto) as a gelling agent. Washed agar was prepared by adding 16.5 g of agar (Difco Bacto) (10% allowance for loss during washing) to distilled water, stirring vigorously for 5 min, allowing the agar to settle for 30 min, and then decanting the supernatant. This process was repeated until the supernatant was clear (two to four rinses). The agar was used by stirring it into suspension, pouring it off in uniform lots of 50 ml, and mixing it with the desired medium as described above.
Aerobically incubated plates were placed in sealed plastic containers to prevent moisture loss over the extended incubation period. These were opened periodically for colony counts (at 3, 8, and 12 weeks). Plates incubated under microaerophilic conditions were placed into airtight vessels, and the oxygen tension was reduced with CampyGen (Oxoid Australia Pty. Ltd.) sachets. These vessels remained unopened until plate counts were performed after 12 weeks. All plates were incubated at 37°C in the dark.
Isolate nomenclature was based on the sample source (Cheetham Salt Works) and on numbers referring to the medium from which the isolate was cultivated (1 to 5, as described above, and 6 for MGM), an isolate number, and the plating dilution, e.g., CSW 6.14.5 is a Cheetham Salt Works isolate from medium 6 (MGM) with an isolate number of 14 and was taken from a 10−5 dilution plate.
Rapid DNA preparations suitable for PCRs were made by a modified version of a method described by M. Pfeiffer in the Halohandbook (http://www.microbiol.unimelb.edu.au/micro/staff/mds/HaloHandbook). Briefly, 1 ml of each broth culture was centrifuged for 3 min (for visibly turbid cultures) or 10 min (for nonturbid cultures) in a microcentrifuge at 16,000 x g. The supernatant was aspirated, the sample was centrifuged for an additional 2 min, and again the supernatant was aspirated. For tubes containing visible pellets, 400 μl of sterile distilled water was added and the cell pellet was lysed by vigorous mixing with a pipette. For tubes without a visible pellet, 200 μl of distilled water was used. In each case, 1 μl of the lysate was added to a 50-μl PCR mix. The primers used for this study were as follows: 344mod, ACGGGGCGCAGCAGGCGCG (modified from reference 33); F1, ATTCCGGTTGATCCTGC (17); 1492Ra, ACGGHTACCTTGTTACGACTT (13); Top168r, ATGTTGTGTGGAATTGTGAGCGG (this study); GEM2987f, CCCAGTCACGACGTTGTAAAACG (24); 1391R, GACGGGCGGTGTGTRCA (9); 723f, AACCGGATTAGATACCC (17); 1114F, GCAACGAGCAGAACCC (this study); and 1114FA, GCAACGAGCGAGATCC (this study).
All PCRs were performed in an MJ Research PTC-100 thermal cycler in a volume of 50 μl, with the exception of sequencing PCRs, which are described separately.
Each reaction comprised 1.75 mM MgCl2, PCR buffer (Qiagen Pty. Ltd., Clifton Hill, Australia), 200 μM (total) deoxynucleoside triphosphates, 50 pmol of forward primer F1 (ATTCCGGTTGATCCTGC) (17) or 344mod (ACGGGGCGCAGCAGGCGCG) (modified from reference 33), 50 pmol of reverse primer 1492R (ACGGHTACCTTGTTACGACTT) (13), and 2 U of HotStart Taq DNA polymerase (Qiagen). The balance of the volume to 49 μl was distilled water, and the master mix was UV irradiated to cross-link any contaminating DNAs (to preventing them from participating in the subsequent PCR). One microliter of a DNA preparation (lysate or DNA extract) was added, and the following thermal cycling parameters were used: 15 min at 95°C; 30 cycles of 1 min at 95°C, 30 s at 46°C, and 2 min at 72°C; and 10 min at 72°C. A Haloferax volcanii lysate was used as a positive control, and distilled water was used as a negative control.
The protocol for the PCR analysis of 16S rRNA genes cloned into E. coli used two separate pools of reagents, a master mix and a hot start mix. The master mix comprised Mg-free PCR buffer (Promega Corp., Annandale, Australia), 1 mM MgCl2, 50 pmol of the forward primer GEM2987f (CCCAGTCACGACGTTGTAAAACG) (24), 50 pmol of reverse primer (ATGTTGTGTGGAATTGTGAGCGG) (this study), and distilled water to 40 μl. The template DNA was mixed with 40 μl of master mix, and then a hot start mix containing Taq DNA polymerase was added and the reaction was held at 95°C for 2 min. This prevented nonspecific polymerase action before the reaction reached 95°C. Plastic pipette tips were used to pick up small samples of isolated colonies and to transfer the cells to separate reactions (40 μl). A drop of molecular-biology-grade mineral oil (Promega Corp.) was added to prevent aerosol contamination. After 2 min, the hot start mix, comprising 200 μM deoxynucleoside triphosphates (total), 1.25 U of Taq DNA polymerase (Promega Corp.), and distilled water to 10 μl, was added by use of a separate pipette tip for each sample to the existing reaction. The reaction was then cycled 35 times for 15 s at 94°C, 15 s at 56°C, and 1 min at 72°C. The negative controls were (i) E. coli cells containing the vector plasmid without inserted DNA and (ii) distilled water. E. coli cells containing a plasmid with a known 16S rRNA gene clone were used as a positive control.
PCR cleanup was performed with an UltraClean PCR Clean-Up DNA purification kit (MoBio Laboratories, Solana Beach, Calif.) used according to the manufacturer's instructions, with elution into a final volume of 50 μl of distilled water. The DNAs were then used for cloning or sequencing.
Between 30 and 100 ng of PCR product was used for sequencing reactions. Reactions contained 3.2 pmol of sequencing primer and 4 μl of ABI Prism Big Dye terminator mix, version 2 or 3 (Applied Biosystems, Scoresby, Australia). Sequencing primers included the PCR primers used for the amplification of 16S rRNA genes (see above), the internal rRNA gene primers 723f (AACCGGATTAGATACCC) (17), 1114F (GCAACGAGCAGAACCC) (this study), and 1114FA (GCAACGAGCGAGATCC) (this study), and for cloned genes, the GEM2987f primer. Thermal cycling was performed with a Gene Amp PCR system 9600 instrument (Applied Biosystems) and consisted of 25 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C. The reaction products were precipitated in 65% (vol/vol) isopropanol and pelleted by centrifugation. The reactions were analyzed at the Australian Genome Research Facility DNA Sequencing Laboratory (Parkville, Melbourne, Australia).
Five-microliter samples of PCRs were mixed with 2 μl of Orange G loading dye (0.1 mM EDTA, 40% [vol/vol] glycerol, 0.15 g of Orange G dye [ICN Biomedicals Australasia Pty. Ltd., Girraween, Australia]) prior to being loaded into wells of 1.2% (wt/vol) agarose gels containing TAE (40 mM Tris, brought to pH 8.0 with acetic acid, and 1 mM EDTA, brought to pH 8.0 with NaOH) or TBE (9 mM Tris [pH 8.0], 0.2 mM EDTA, 89 mM boric acid, 25 mM NaOH) buffer. A pGEM DNA marker (Promega Corp.) or a Precision molecular mass ruler (Bio-Rad Laboratories Pty. Ltd., Regents Park, Australia) was used as a size standard. Electrophoresis was done at 120 V for 20 min for routine PCRs or at 90 V for 75 min for PCRs requiring a higher resolution. Gels were stained with ethidium bromide and photographed over a UV transilluminator, and band intensities were analyzed with the Kodak 1D 2.0.2 software program (Kodak Australasia Pty. Ltd., Coburg, Australia).
Cells from 10 ml of saltern water were pelleted by centrifugation (5,000 × g for 20 min), the supernatant was removed, and DNAs were extracted from the cell pellet by bead beating (1 min at 2,500 rpm and room temperature) (Mini-BeadBeater; Biospec Products Inc., Bartlesville, Okla.) according to a method based on that of Pitcher et al. (31). Five replicate PCRs were performed with primers F1 and 1492R and an additional five reactions were performed with primers 344mod and 1492R, with both sets following the protocol for PCRs of 16S rRNA genes from crystallizer pond DNAs (described above). Prior to the addition of DNA, each reaction was UV treated to cross-link any contaminating DNAs (20 mJ/cm2 for 3 min) (Spectrolinker XL-1000 UV cross-linker; Spectronics Corporation, Lincoln, Nebr.). The five reactions with the same DNA sample were pooled, and the DNA was purified and concentrated to 50 μl with a Promega Wizard DNA purification kit. A dA tail was added to the DNA in a 15-μl volume by the use of HotStart Taq DNA polymerase (Qiagen Pty. Ltd.). The A-tailed products were electrophoresed through an agarose gel, and a DNA between 1.1 and 1.5 kb was excised and extracted by use of a Geneclean kit (Qbiogene Inc., Carlsbad, Calif.) according to the manufacturer's instructions. This product was cloned into the plasmid vector pCR 2.1-TOPO and introduced into E. coli TOP10 cells by use of a TOPO 2.1 cloning kit (Invitrogen Australia Pty. Ltd., Mount Waverley, Australia).
The nomenclature of cloned 16S rRNA sequences reflects the source of the gene (Cheetham Salt Works) and the forward primer used (F1 or 344mod), e.g., CSWM048 is Cheetham Salt Works clone sequence number 48 generated with the forward primer 344mod (and the reverse primer 1492Ra). Fifty-nine cloned genes were sequenced, which included 31 clones from the library generated with the F1 primer and 28 clones from the sequence library generated with the 344mod primer.
Exponential-phase cultures of three isolates from this study, specifically CSW 6.08.5 (Haloferax sp.), CSW 3.10.5 (Halorubrum group 1), and CSW 6.01.5 (Halorubrum group 2), were counted by phase-contrast microscopy using a hemocytometer slide and then were mixed to give equal cell concentrations of each isolate. The mixture was diluted in salt water diluent (identical to that used in basic 25% salt medium) to a turbidity similar to that of the original saltern water sample. A 16S rRNA sequence library was generated with primers F1 and1492R as described above.
Clone inserts were amplified by the PCR protocol for the analysis of 16S rRNA genes cloned into E. coli plasmids (described above) with primers TOP168r and GEM2987f, and the product of each reaction was purified as described for the other PCR protocols. BamHI was selected to identify the cloned 16S rRNA sequences, as this enzyme gives characteristic digestion fragments for the three isolates. Restriction digests were electrophoresed in an agarose gel alongside DNA size markers and BamHI digests of PCR products of the three strains.
Primer design and phylogenetic tree reconstructions were performed by using the ARB phylogeny package and the 16S rRNA sequence database of Phil Hugenholtz (http://rdp.cme.msu.edu/html/alignments.html), updated with haloarchaeal sequences from the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov/). Primer sequences were rigorously checked against all known archaeal, and particularly, haloarchaeal 16S rRNA gene sequences by use of the ARB package to ensure that they showed a high level of complementarity, and where necessary, a high specificity. The ARB software (21; http://www.arb-home.de/) was ported and run on an Apple Mac OS X platform (http://fink.sourceforge.net/pdb/package.php/arb, http://www.microbiol.unimelb.edu.au/staff/mds/ARB_OSX/ARB_to_MacOSX.html). The identification of new sequences was performed with ARB and the NCBI BLAST web-based nucleotide-nucleotide search program (http://www.ncbi.nlm.nih.gov/BLAST). Phylogenetic tree reconstructions were inferred by using the algorithms and tools available in ARB or PAUP (D. L. Swofford, Phylogenetic analysis using parsimony, v. 4, Sinauer Associates, Sunderland, Mass., 2003). Chimeric 16S rRNA sequences were detected with the Bellepheron server (http://foo.maths.uq.edu.au/~huber/bellerophon.pl) and were discarded. For phylogenetic tree reconstructions, a backbone tree was constructed, using only sequences of >1,300 nucleotides (nt), and was then aligned within ARB. The alignment was exported to PAUP, and tree reconstruction was performed with the maximum likelihood algorithm. Bootstrap values were derived from 1,000 replicate tree reconstructions by the use of PAUP (DNA parsimony algorithm). The tree was imported into ARB, and partial sequences of cloned genes and isolates (all with >400 nt) were then added to the tree by the parsimony add option within ARB.
Isolates and cloned 16S rRNA sequences were assigned to groups that were clearly defined as coherent genera on the basis of sequence differences (12). Four of the groups were named after genera of the family Halobacteriaceae (e.g., the Haloferax group) because the sequences within those groups were similar to those of recognized species, but the group name was purely operational and does not imply any formal taxonomic proposals for the isolates in this study. Members of the Halorubrum group were clearly separated into two subgroups, which we termed Halorubrum group 1 and Halorubrum group 2. The amount of gene sequence difference displayed by members of the SHOW group suggests that they belong to one genus, and they have been grouped together on this basis.
The longer 16S rRNA sequences (>1,300 nt) from this study have been deposited in GenBank under accession numbers AY498639 to AY498650.
The saltern pond system used for sample collection is located on the foreshore of Corio Bay, Victoria, Australia (38°09.841′S, 144°25.274′E). The crystallizer pond was sampled in March 2002, and its chemical composition is given in Table Table1.1. The pH was 8.1. The total salt concentration was 33% and the general composition was typical of marine solar salterns (26, 34). The water was visibly turbid, with a pink color, and light microscopy (phase-contrast; ×1,000 magnification) revealed large numbers of square cells, typical of SHOW group haloarchaea. A direct count of cells under a phase-contrast microscope gave a value of 1.2 × 107 ± 0.2 × 107 cells/ml.
Microbial growth on solid media was examined with substrates, most at low concentrations (see Materials and Methods), that are likely present in the natural environment and/or likely to initiate growth of a wide variety of microorganisms (4, 19, 28). Given that the crystallizer pond sample had a salt concentration of 33%, a salt concentration of 25% was chosen for use in the growth media to allow the widest spectrum of halophile growth, from those with low salt optima (e.g., Haloferax spp., with an optimum of about 18%) to those with high optima (e.g., Halobacterium spp., with optima of about 25 to 30%). The effects of medium composition, colony density, and time of incubation on viable counts and diversity are described below.
The viable counts obtained on all media after 12 weeks of incubation were similar (Fig. (Fig.1).1). The colony density was found to significantly affect the viable count, so instead of the expected trend of a 10-fold reduction in the inoculum leading to a 10-fold decrease in CFU per plate, a trend towards a higher viable count was obtained as the inoculum was decreased. Terminal dilutions yielded viable counts between 0.9 × 106 and 1.2 × 106 CFU/ml (with an average colony density of 14), which was two to three times higher than counts using less dilute inocula (Fig. (Fig.1).1). This effect has been commonly observed for plate counts of soil microorganisms (25) and is thought to be due to reduced crowding and competition. All plate isolates could be subcultured on high-substrate-concentration complex MGM medium, and it appears that the low-substrate-concentration approach, which has been successful in improving isolate diversity in other environments such as soil (19) and freshwater lakes (10), is not necessarily a requirement for hypersaline waters. This could be due to nutrient inputs, as fertilizer is often added to promote microbial growth in saltern ponds. In addition, moderate viral lysis rates may increase nutrient fluxes (30), and the nutrient distribution in the water column may be less heterogeneous than in soils.
After inoculation, the plates were examined at 3, 8, and 12 weeks and colony counts were recorded. Viable counts increased with incubation time in a similar pattern for all five media, and the data for three of these media (organic acids, 0.01% NB, and MGM) are shown in Fig. Fig.2.2. Other media yielded similar results. Extending the incubation time from 3 to 12 weeks produced a significant rise in viable counts (two- to threefold). The rise was most pronounced in the 3- to 8-week incubation period, particularly for 0.01% NB, and tailed off over the 8- to 12-week period, suggesting that extending the incubation period past 12 weeks would not result in a significantly larger number of colonies. The high-substrate-concentration medium, MGM, yielded significantly higher viable counts at the early time points; the colonies grew faster and had a much larger final size than those on the low-substrate-concentration media, causing them to be detected earlier. As with the other media, MGM continued to produce new colonies (albeit at a lesser rate) throughout the entire incubation period. The average viable count for all six media, using terminal dilution figures, was 1.1 × 106 CFU/ml, and the direct cell count of the water sample was 1.2 × 107 cells/ml. While there are significant uncertainties associated with both values (sampling errors in the former and the unknown fraction of viable cells in the latter), they give an estimate of cultivability of about 10%.
The diversity of microorganisms growing on different media and after various times of incubation was assessed by comparative sequence analysis of the 16S rRNA genes of 66 selected isolates. Colonies appearing after 3 weeks were all isolates that fell within two previously described genera, Haloferax and Halorubrum (Fig. (Fig.3).3). The Halorubrum isolates fell into two sequence groups (Halorubrum groups 1 and 2; see below). Colonies that appeared after 8 weeks of incubation also included members of three other groups: these were an isolate related to Natronomonas spp., an isolate related to Halogeometricum borinquense, and members of a sequence group that was previously detected in an Antarctic hypersaline lake (Deep Lake) (Fig. (Fig.3).3). Members of the Antarctic Deep Lake (ADL) sequence group, described by Bowman et al. (8), have not previously been cultivated. Hereafter, these isolates are referred to as belonging to the ADL group. The isolates of the ADL and Natronomonas groups were slow growing, even when grown on a rich medium (MGM; unpublished data), indicating that extending the incubation time was a significant factor in their successful isolation.
From 8 to 12 weeks of incubation, the viable counts rose on all media, but no increase in genus-level diversity was observed.
No isolates with 16S rRNA gene sequences belonging to the Bacteria were detected. Evidence that the media were capable of culturing halophilic bacteria was provided by plating water from a saltern pond with a lower salinity on the same media as those used for the crystallizer pond sample. In this case, halophilic bacteria such as Halomonas were readily isolated (data not shown). A culture of S. ruber (2), kindly provided by A. Oren, was also able to grow well in liquid MGM and to form colonies on solid MGM (data not shown).
All media supported the isolation of organisms belonging to the genera Haloferax and Halorubrum (groups 1 and 2). The five members of the ADL group were isolated on four different media (0.01% NB, amino acids, sugars, and MGM). The single Natronomonas-like isolate (CSW 4.3.5) was isolated on glycine betaine medium.
Plates incubated under a microaerophilic atmosphere yielded viable counts that were at least 10-fold lower than those for the same media incubated under full aerobic conditions. All of the colonies had a very similar morphology, and sequence analysis of the 16S rRNA genes of five randomly selected colonies revealed that only Haloferax spp. had developed on these plates (data not shown).
The microbial diversity in the saltern pond was examined by analyzing libraries of PCR-amplified 16S rRNA genes generated from cells in the water sample used for the cultivation study.
Two libraries were generated, one by the use of primers F1 and 1492Ra, amplifying bases 23 to 1491 (E. coli numbering), and one by the use of primers 344mod and 1492Ra, amplifying bases 363 to 1491 (see Materials and Methods). A comparative sequence analysis of 59 cloned inserts from both sequence libraries (average, about a 550-nt sequence/clone) revealed that the 16S rRNA genes of 57 of the clones were derived from five groups of haloarchaea (Halorubrum groups 1 and 2, the ADL group, the SHOW group, and the Natronomonas group). Two cloned 16S rRNA sequences were chimeric and were discarded. All five groups were represented in the 344mod library, and all but the Natronomonas group were represented in the F1 library. The sequences were pooled and treated as one library. The cloned, PCR-amplified 16S rRNA genes were dominated by sequences similar to those of members of the genus Halorubrum (49% of the total cloned sequences), followed in abundance by sequences similar to those of the SHOW group (33%) and the ADL group (16%). One sequence was most similar (97.7%) to the 16S rRNA gene of Natronomonas pharaonis. Studies of the microbial populations in salterns in Spain and Israel have also found that they are predominantly made up of members of the genus Halorubrum and of the SHOW group (1, 3-5, 8, 23), but the studies in Israel did not detect members of the ADL group. The proportion of ADL group sequences in our library and that found in the Antarctic study of Bowman et al. (8) were similar (16 and 18%, respectively).
The levels of microbial diversity discerned by the sequence library and by cultivation differed significantly in two ways (Fig. (Fig.4).4). Cultivation produced no SHOW group isolates (a well-known problem), whereas 33% of the 16S rRNA genes in the sequence library were derived from members of this group. It was recently demonstrated that SHOW group organisms take up exogenous amino acids (35), but our use of a medium containing free amino acids (0.05%) did not facilitate the isolation of members of this group. It is possible that the concentration used was inhibitory; for example, the dominant marine bacterium SAR11 is strongly inhibited by peptone at levels as low as 0.001% (32).
To test if there was a bias against recovering Haloferax 16S rRNA genes in the sequence library (including both the DNA extraction and PCR steps), we constructed a library from a mixture of a Haloferax sp. and two different Halorubrum spp., and the frequency of sequences from each organism was estimated. Three isolates corresponding to the three most common isolate groups, Haloferax, Halorubrum group 1, and Halorubrum group 2, were selected. The cultures were grown to mid-exponential phase, cell concentrations were determined by direct counting under a phase microscope, and the volumes of the cultures were mixed to give equal concentrations of the three organisms. Immediately after mixing, a sequence library was prepared. Cloned PCR products were identified by restriction fragment length polymorphism analysis using BamHI. One-third of the sampled clones (15 of 48 [31%]) were derived from the Haloferax sp., and the remaining sequences were approximately equally divided between those from isolates of Halorubrum group 1 and those from isolates of Halorubrum group 2 (29 and 40%, respectively). The results indicated that there was no significant bias against Haloferax 16S rRNA gene sequences, although the possibility remains that the PCRs of DNA extracts from natural water samples could have been biased by components in the water. If it is assumed that there was little or no bias, then the high representation of isolates belonging to the Haloferax group is likely due to their high cultivability compared to the other haloarchaea.
All sequence and isolate 16S rRNA gene sequences were placed into a phylogenetic tree by use of the ARB software package. The groups represented in this study make up a relatively minor component of the known diversity of the family Halobacteriaceae (Fig. (Fig.5).5). Representatives of the genus Halorubrum made up the dominant group in the sequence library (49%) and among the isolates (74%), and almost all of these isolates and clone sequences fell into two groups. Group 1 sequences branched with Halorubrum sodomense and shared ≥97% sequence identity with the 16S rRNA of this species, while group 2 sequences formed a separate clade and shared 93 to 96% similarity with the Halorubrum sodomense sequence.
Representatives of the ADL sequence group (16% of the sequence library) appeared to form at least three main clades (Fig. (Fig.6),6), with isolate CSW 6.21.5 being related the closest to previously reported ADL sequences, such as DEEP-5 (accession no. AF142879) (4, 8). The aligned sequences of ADL isolates showed pairwise sequence similarities ranging from 93 to 99%, indicating that this group may contain more than one genus (12). Some of the isolates and sequences shared identical or highly similar 16S rRNA gene sequences (Fig. (Fig.6).6). Isolates belonging to the genus Haloferax were abundant (15% of analyzed isolates), but no library sequences could be assigned to this group. The 16S rRNA sequences of all of the isolates were very similar to each other, with identities of ≥99% with each other and with the 16S rRNA gene of Haloferax volcanii.
The SHOW group made up the second largest group within the sequence library (33%), but no isolates were obtained. The cloned sequences in this study all had 99% identity with the SHOW clone sequence (accession no. X84084) reported by Benlloch et al. (6). The sequence of one isolate, CSW 2.24.4, branched just outside the SHOW group (91% identity to sequence X84084 [Fig. [Fig.5])5]) and was phylogenetically most similar to Halogeometricum borinquense (92% identity). One isolate and one clone sequence were assigned to the Natronomonas group. Their 16S rRNA gene sequences showed 98 to 99% identity with each other and 97 to 98% identity with N. pharaonis. The latter organism was isolated from a soda lake and is unable to grow below pH 8 (its optimum is pH 9 to 9.5) (14), but closely related isolates have been found that are not alkaliphilic, such as isolate Nh.2 (included in the Natronomonas group in Fig. Fig.5),5), which was recovered from a salt mine (22).
The combination of molecular ecological and cultivation-dependent methods was able to identify the major microbial groups within the crystallizer pond and to isolate representatives of many of them for further detailed study. The success in this study of relatively simple cultivation procedures along with molecular techniques in isolating and identifying representatives of significant, and in some cases, previously uncultured members of the microbial community of a solar saltern demonstrates that cultivation remains a viable and important technique for assessing microbial diversity and ecology.
Isolation of SHOW group organisms has now been achieved using media based on natural saltern water (unpublished data).
We thank Wally Rickard and Noel Taylor (Cheetham Salt Ltd., Geelong, Australia) for allowing us to sample crystallizer ponds, Jianping Lin (Water Studies Centre, University of Melbourne) for his help with water analyses, and Ben Hines, who was instrumental in bringing the ARB package to Mac OS X.
This work was partly funded by a grant from the Australian Research Council.