[FeFe] Hydrogenase Genetic Diversity Provides Insight into Molecular Adaptation in a Saline Microbial Mat Community

Abstract

Degenerate primers for the [FeFe] hydrogenase (hydA) were developed and used in PCRs to examine hydA in microbial mats that inhabit saltern evaporative ponds in Guerrero Negro (GN), Mexico. A diversity of deduced HydA was discovered that revealed unique variants, which may reflect adaptation to the environmental conditions present in GN.

Hydrogen (H₂) is an important intermediate in the decomposition of organic matter in anaerobic environments and is the basis for many syntrophic interactions that occur in the food chain such as those between acetogens and methanogens and/or sulfate reducers. Little is known concerning the potential role of fermentative bacteria and/or H₂ cycling in phototrophic microbial mat systems, such as the 6-cm-thick and extremely diverse microbial mats inhabiting the saline ponds at the Exportadora de Sal SA saltern in Guerrero Negro (GN), Baja California Sur, Mexico. Recent 16S rRNA gene surveys in GN revealed an abundance of bacteria in the upper 2 mm of the mat that, based on their phylogenetic affiliation, are thought to harbor fermentative metabolisms (6, 8, 13, 16, 20, 28). Similarly, previous studies have shown that the flux of H₂ from the surface of the phototrophic mats present in GN is ∼150-fold higher during the night than during the day (9). Together, these results suggest the potential for fermentative metabolisms in H₂ and carbon cycling in the GN saline mat environment.

The [FeFe] hydrogenase has a central role in primary and secondary substrate fermentations, catalyzing the oxidation of excess reducing equivalents coupled to the reduction of protons, yielding H₂ (2, 17, 29, 30). The [FeFe] hydrogenase is only known to occur in anaerobic bacteria and a small number of green algae (23, 29, 30), making this a useful biomarker for examining the diversity and distribution of this functional class of organism. Eighty-two putative [FeFe] hydrogenase large-subunit protein sequences (HydA) present in the GenBank database that represent the known diversity of putative HydA were screened for the L1 ([FLI]TSC[C/S]P[GAS]W[VIQH]) and L2 ([IVLF]MPCx[ASRD]K[KQ]xE) (conserved residues are in bold and underlined, and bracketed positions indicate “semiconserved” residues at that position) signature sequence motifs (17, 30). Degenerate PCR primers, corresponding to positions 272 to 279 (AD[M/L]TIMEE) and positions 420 to 427 (TGGVMEAA) in the Clostridium pasteurianum protein sequence, were designed for use in specific gene amplifications.

Mat core samples (1 by 6 cm) were collected at 2:00 pm from pond 4 (35°C, pH 8.0, and 80 ppt salt) near pond 5 at the Exportadora de Sal saltworks, GN, on 13 February 2005. Mat cores were sectioned in 1-mm increments on site and immediately flash frozen in liquid nitrogen as previously described (16). Genomic DNA was extracted by bead beating in the presence of phenol-chloroform-isoamyl alcohol and sodium dodecyl sulfate as previously described (7) and was quantified by using the High DNA Mass Ladder (Invitrogen, Carlsbad, CA). Approximately 500-bp fragments of hydA were amplified in triplicate using primers FeFe-272F (5′-GCHGAYMTBACHATWATGGARGA-3′, where H = A, C, T; Y = C, T; M = A, C; B = C, G, T; W = A, T; R = A, G; 432-fold degeneracy) and FeFe-427R (5′-GCNGCYTCCATDACDCCDCCNGT-3′, where N = A, C, T, G; Y = C, T; D = A, G, T; 864-fold degeneracy) and 35 cycles of PCR as previously described (4). An empirically determined annealing temperature of 56.5°C, a MgCl₂ concentration of 1.5 mM, and a primer concentration of 1 μM for each forward and reverse primer were utilized in 50-μl PCR mixtures containing 10 ng of genomic DNA as the template. Equal 40-μl volumes of three replicate PCR products were pooled, purified using the Promega Wizard purification kit (Madison, WI), quantified using the Low Mass DNA Ladder (Invitrogen), cloned using the pGEM Easy Vector System (Promega), and sequenced by using the M13F-M13R primer pair as previously described (5).

Deduced amino acid sequences were screened for the presence of the L1 and L2 HydA sequence motifs as described above. ClustalX (version 2.0.8) (15) was employed to align inferred amino acid sequences and to create distance matrices for use in identifying and clustering operational taxonomic units (OTUs) with DOTUR (26). Calculations of Shannon diversity and Chao1 and Ace1 deduced amino acid sequence richness were completed with DOTUR by using the furthest-neighbor algorithm and a precision of 0.01. The phylogenetic position of putative HydA was assessed with MRBAYES (10, 25). Tree topologies were sampled every 500 generations for 2,000,000 generations (burnin = 1,000,000) by using the WAG evolutionary model with fixed amino acid frequencies and gamma-shaped rate variation with a proportion of invariable sites as recommended by ProtTest (1). The Saccharomyces cerevisiae Narf protein, a homolog of HydA, was used as the outgroup. The phylogenetic tree was projected using TreeView (version 1.6.6) (21).

An unexpectedly diverse assemblage of putative HydA variants was present in the top 1 mm of the GN microbial mat, corresponding to the photic zone. At a sequence identity threshold (SIT) of 100%, the predicted Shannon diversity index was 3.84 and the mean Chao1 HydA richness was 187 unique OTUs (see Fig. S1a in the supplemental material). When a 99% SIT was applied, the predicted Chao1 richness estimate decreased 56% to 82 unique OTUs and the Shannon diversity index decreased slightly to 3.55 (data not shown). A further decrease in the SIT from 99 to 85% resulted in a slight decrease (23%) in the Chao1 predicted richness and a modest decrease in the Shannon diversity index from 3.55 to 3.28, suggesting that the majority of the putative HydA OTUs present in the top 1 mm of the mat is a result of recent evolution or speciation.

The putative HydA variants from GN resolved into 42 OTUs which clustered into one of seven distinct phylogenetic clusters (Fig. 1A), although clusters 5 and 7 could not be adequately resolved. Intriguingly, the mean identity scores determined from sequence alignment matrices for members of each of the seven sequence clusters recovered from GN, when compared to sequences in the GenBank database, were very low (47.5 to 73.2%) (Table 1), suggesting that the putative HydA sequences recovered from GN represent novel sequence space.

FIG. 1.

(A) Phylogenetic tree based on deduced putative HydA amino acid sequences from GN and reference HydA sequences. Nodes with posterior probabilities of <50% were collapsed in this analysis (posterior probabilities of 100 are denoted by asterisks). Bar, one substitution per 10 sites. (B) Partial-length deduced amino acid sequences of environmental and reference deduced HydA sequences illustrating the phylogenetic coherence of novel substitutions and insertions in and upstream of L1 sequence motif. Cluster designations (C1 to C7) correspond to those presented in Table 1.

TABLE 1.

Phylogeny of partial-length putative HydA sequence clusters

Sequence clustera	No. of clones in cluster	Maximal intracluster sequence divergence (%)b	Most closely related sequencec	Phylumd	Mean % sequence identity (range)e
C1	22	49.4	Moorella thermoacetica ATCC 39073	Firmicutes	61.8 (56.4-67.3)
C2	16	41.6	Opitutus terrae PB90-1	Verrucomicrobia	64.3 (56.3-71.1)
C3	6	28.8	Bacteroides thetaiotaomicron VPI-5482	Bacteroidetes	72.1 (69.8-73.2)
C4	9	38.6	Alkaliphilus oremlandii OhILAs	Firmicutes	65.1 (61.4-66.4)
C5	3	38.6	Moorella thermoacetica ATCC 39073	Firmicutes	61.4 (NA)
C6	7	51.8	Heliobacillus mobilis	Firmicutes	50.3 (47.5-54.6)
C7	2	32.9	Thermoanaerobacterium saccharolyticum JW/SL-YS485	Firmicutes	61.2 (58.5-64.2)

^aSequence clusters correspond to those presented in Fig. 1.

^bMaximum intracluster sequence divergence among sequences within a cluster as calculated by the ClustalX sequence identity matrix following alignment using the Gonnet protein weight matrix (gap opening penalty of 13, gap extension penalty of 0.05).

^cThe most closely related sequence is defined as the taxon with the HydA sequence most closely related to that of members within the cluster as presented in Fig. 1.

^dPhylum designation of most closely related HydA sequence as defined in Fig. 1.

^eSequence identity means and ranges were calculated by using the ClustalX sequence identity matrix as described in footnote b. NA, not available.

The highest proportion of putative HydA sequences recovered from the top 1-mm transect of the GN microbial mat were related to acetogens most closely affiliated with the Firmicutes (66.2% of the clones) and the Verrucomicrobia (24.6%), with a lower proportion related to the Bacteroidetes (9.2%). These observations support those of previous studies that identified the highest abundance of 16S rRNA genes related to fermentative/acetogenic organisms within the phyla Firmicutes, Verrucomicrobia, and Bacteroidetes in the upper few millimeters of microbial mat (16), coinciding with the mat transect where nighttime H₂ concentrations were also elevated (9). The recovery of putative HydA sequences in the photic zone of the GN microbial mat is consistent with the results of previous studies conducted in the phototrophic mats in Yellowstone National Park, WY, which documented the potential for the secondary fermentation of cyanobacterial primary fermentation products in the top photic layers of the mat at night (3). Importantly, cyanobacterial fermentation may also contribute to the production of H₂ in the mats during periods of anoxia (9).

Deduced amino acid sequences from GN had significantly lower (P = <0.01 for both GRAVY and aliphatic indices) hydropathy indices than nonhalotolerant HydA sequences from GenBank (see Table S1 in the supplemental material), a feature which may reflect molecular adaptation to salinity. These observations are consistent with the results of previous investigations of deduced amino acid sequences from a variety of halophilic organisms that indicate hydrophobicity indices lower than those of their nonhalophilic counterparts, which is hypothesized to reduce the effects of salt-driven protein misfolding and/or aggregation (11, 12, 14, 24, 27, 31). In addition, many of the putative HydA sequences recovered from GN exhibited previously unobserved substitutions in the L1 motif and novel insertion domains upstream from the L1 motif (Fig. 1B). Residues within L1 are involved in the coordination of the oxygen-labile [4Fe-4S] subcluster of the H cluster of HydA (18, 19, 22). Substitutions in this region may have implications for the redox properties of this [4Fe-4S] cluster and thus may be important in conferring stability in the presence of oxygen (23).

The discovery of a rich assemblage of putative HydA variants in the top 1 mm of the GN microbial mat underscores the utility of these primers in examining the natural diversity of putative HydA and fermentative organisms in complex microbial ecosystems. The composition of putative HydA sequences from the top 1 mm of the microbial mat suggests adaption to the conditions present in the evaporative salterns in GN.

Nucleotide sequence accession numbers.

A total of 65 putative hydA gene sequences determined in the present study have been deposited in the GenBank, DDBJ, and EMBL databases under accession numbers FJ623894 to FJ623958.