Introduction

Most ecosystems rely on phototrophic carbon fixation, or, in the absence of light, an external supply of phototrophically-fixed carbon into the ecosystem. Exceptions are deep sea hydrothermal vents, where carbon is derived from chemosynthesis using energy sources other than light (reviewed by Lutz and Kennish, 1993; Van Dover et al., 2002; Campbell, 2006). Movile Cave, located near the coast of the Black Sea in Mangalia, Romania, is an underground cave system that has been completely sealed off from the outside world for several million years (Sarbu et al., 1996). Unlike other cave systems, where dissolved and particulate organic carbon enters the cave with meteoric waters from above, the food web in Movile Cave is sustained exclusively by non-phototrophic carbon fixation. Since its discovery in 1986, Movile Cave has provided an excellent natural ecosystem to study a highly unusual, light-independent, microbially-driven food web (Sarbu et al., 1994; Sarbu and Kane, 1995; Sarbu et al., 1996; Vlasceanu et al., 1997; Rohwerder et al., 2003; Hutchens et al., 2004; Porter et al., 2009; Chen et al., 2009). Movile Cave harbours rich and diverse populations of cave-adapted invertebrates, all of which are sustained by chemolithoautotrophic microorganisms that thrive along the redox interface created between the oxygenated atmosphere and the high concentrations of reduced compounds such as hydrogen sulfide (H2S) and methane (CH4) present in the water (Sarbu and Kane, 1995). Microbial mats composed of bacteria, fungi and protists float on the water surface (kept afloat by CH4 bubbles) and also grow on the limestone walls of the cave (Sarbu et al., 1994).

Methylotrophs are organisms capable of using one-carbon (C1) compounds, that is, compounds lacking carbon–carbon bonds, as their sole source of carbon and energy (Anthony, 1982; Lidstrom, 2006; Chistoserdova et al., 2009). In addition to CH4, C1 compounds such as methanol and methylated amines are important carbon and energy sources for a range of methylotrophic bacteria in Movile Cave (Hutchens et al., 2004; Chen et al., 2009). Methylated amines are typically associated with saline environments (Gibb et al., 1999; Fitzsimons et al., 2006) where they are formed by the degradation of glycine betaine and trimethylamine N-oxide, osmolytes commonly found in marine organisms (Barrett and Kwan, 1985; Lin and Timasheff, 1994). There are fewer studies on the distribution of methylated amines in terrestrial and freshwater environments, although the dissolved organic nitrogen fraction as a whole is increasingly being recognised as an important source of microbial nitrogen nutrition (Berman and Bronk, 2003, Worsfold et al., 2008). Generally, environments with high concentrations of organic matter have a high potential for dissolved organic nitrogen generation (Neff et al., 2003). We hypothesise that in Movile Cave, degradation of the extensive, organic-rich microbial mats produces large amounts of methylated amines, which are used as growth substrates by certain microorganisms that are the subject of this study.

Methylotrophs that use methylated amines as a carbon source are phylogenetically diverse, ubiquitous in the environment and often metabolically versatile (for example, Bellion and Hersh, 1972; Colby and Zatman, 1973; Levering et al., 1981; Anthony, 1982; Bellion and Bolbot, 1983; Brooke and Attwood, 1984; Kalyuzhnaya et al., 2006b; Boden et al., 2008). New methylotrophs are still being identified from a wide range of environments, including genera not previously associated with methylotrophy, and novel metabolic pathways (see recent reviews by Chistoserdova et al., 2009; Chistoserdova, 2011).

Methylated amines are also a nitrogen source for a wide range of non-methylotrophic bacteria. While utilisation of monomethylamine (MMA) as a bacterial nitrogen source was reported over 40 years ago (Budd and Spencer, 1968; Bicknell and Owens, 1980; Anthony, 1982; Murrell and Lidstrom, 1983; Glenn and Dilworth, 1984), details of the metabolic pathways involved have only recently been identified (Chen et al., 2010b).

The key intermediates in methylotrophic metabolism are formaldehyde or formate, respectively, as they represent the branching point at which carbon is either oxidised further to CO2, or assimilated into cell carbon. Carbon is assimilated from formaldehyde via the ribulose monophosphate cycle, or from formate via the serine cycle (Anthony, 1982; Chistoserdova et al., 2009; Chistoserdova, 2011). In the metabolism of methylated amines, there are two possible pathways for the oxidation of MMA (Supplementary Figures S1a and b): In the well-characterised, direct MMA-oxidation pathway, a single enzyme oxidises MMA to formaldehyde, releasing ammonium. In methylotrophic Gram-positive bacteria the enzyme responsible is MMA oxidase, while in Gram-negative methylotrophs it is MMA dehydrogenase (Anthony, 1982). PCR primers are available for mauA (Neufeld et al., 2007a), the gene coding for the small subunit of MMA dehydrogenase. However, these primers do not detect all MMA-utilising bacteria. An alternative, indirect pathway oxidises MMA not to formaldehyde but to 5,10-methylenetetrahydrofolate (CH2=THF) in a stepwise conversion via the methylated amino acids gamma-glutamylmethylamide (GMA) and/or N-methylglutamate (NMG) (Latypova et al., 2010; Chistoserdova, 2011). Although this pathway has been known since the 1960s (Kung and Wagner, 1969), the enzymes and genes involved have only recently been characterised (Latypova et al., 2010; Chen et al., 2010a): MMA is converted to GMA by GMA synthetase (gmaS), GMA is then converted to NMG by NMG synthase (mgsABC), and finally to CH2=THF by NMG dehydrogenase (mgdABCD). A variation of this pathway is found in Methyloversatilis universalis FAM5, where gmaS is not essential for oxidation of MMA to CH2=THF via NMG (Latypova et al., 2010). Importantly, the GMA-/NMG-mediated pathway is also found in bacteria that use MMA only as a nitrogen (but not carbon) source (Chen et al., 2010b; Chen 2012). In a recent study (Chen, 2012), PCR primers targeting gmaS from marine Roseobacter clade (MRC) bacteria were developed for the detection of MMA utilisers in marine environments, highlighting the potential of the gmaS gene as a biomarker for MMA utilisation.

The objectives of this study were to determine the role of methylated amines as carbon and nitrogen sources for microorganisms in Movile Cave, and to identify active MMA utilisers in this unique ecosystem using DNA stable isotope probing (DNA-SIP) (Radajewski et al., 2000; Murrell and Whiteley, 2011). DNA-SIP has been successfully applied in the study of methanotrophic and autotrophic communities in Movile Cave (Hutchens et al., 2004; Chen et al., 2009). Time-course SIP experiments with 13C-labelled MMA were set up in order to monitor changes in the methylotrophic community. Cultivation-based studies were also used to isolate and characterise methylated amine-utilising bacteria from the cave. The distribution of genes for the GMA-dependent MMA-oxidation pathway in Movile Cave microbes was examined using new PCR primer sets developed to target gmaS from non-marine bacteria.

Material and methods

Study site and sampling

Movile Cave near Mangalia on the coast of the Black Sea is located in an area rich in hydrothermal activity with numerous sulfurous springs and lakes, as well as creeks bubbling with methane. The cave consists of a network of passages, including a dry, upper level and a lower level, which is partly flooded by thermal sulfidic waters (for a detailed cross-section of the cave see Supplementary Figure S2). A small lake room (ca 3 m in diameter) is located between the dry and the flooded sections of the cave, and two air bells are located in the submerged region. The temperature in the cave is a constant 21 °C (Sarbu and Kane, 1995). The atmosphere in the air bells shows O2 depletion (7–10% v/v) and is rich in CO2 (2.5% v/v) and CH4 (1–2% v/v) (Sarbu and Kane, 1995). The water contains H2S (0.2–0.3 mM), NH4+ (0.2–0.3 mM) and CH4 (0.02 mM) and is buffered by high amounts of bicarbonate from the limestone walls at ∼pH 7.4 (Sarbu, 2000). Dissolved O2 decreases to less than 1 μM after the first few centimetres from the water surface, with the deeper water being essentially anoxic (Sarbu, 2000). Methylamine concentrations in the cave water were measured by our recently developed ion chromatography method with a detection limit of ∼1 μM for MMA (Lidbury et al., 2014). Preliminary measurements carried out using this assay suggested that the in situ concentration of MMA in Movile Cave water is below the detection limit of 1 μM, which could indicate rapid turnover of MMA by bacteria in the cave.

Water and floating mat samples for enrichment and isolation experiments were collected from the lake room and the two air bells in October 2009, stored at 4 °C in the nearby field station and processed within 48 h. Biofilm covering the limestone walls of both air bells was scraped off into sterile tubes. Similar samples for further isolation experiments, SIP enrichments and nucleic acid extractions were obtained from Movile Cave in April 2010. Material for DNA work was concentrated by centrifugation within 1 h of sampling and frozen at −20 °C for storage until processing.

DNA-SIP with 13C-MMA

SIP incubations were set up at the field station in Mangalia, within 1 h of sampling, using water from Airbell 2. For each incubation, a 20 ml aliquot of cave water was added to a pre-sterilised 120 ml serum vial containing 50 μmol of labelled (13C) or unlabelled MMA–HCl (dissolved in 0.2 ml sterilised distilled water). Control incubations with no added MMA (referred to as ‘no-substrate controls’ from here on) were also set up. All serum vials were immediately sealed with a butyl rubber cap and an aluminium crimping lid and incubated at 21 °C in the dark. Samples for t0 (t=0 days) were prepared by centrifugation of 20 ml of cave water, discarding the supernatant and freezing the pellet at −20 °C. SIP incubations and no-substrate controls were harvested in the same way at time intervals of 48 h (t1), 96 h (t2) and 4 weeks (t3). In future SIP experiments, the recently developed ion chromatography method for measuring MMA (see above, Lidbury et al., 2014) could be used to monitor consumption of substrate over time. From each sample, up to 1 μg of total extracted DNA was added to caesium chloride (CsCl) solutions for isopycnic ultracentrifugation and gradient fractionation following published protocols (Neufeld et al., 2007b).

Enrichment and isolation of methylated amine-utilising bacteria from Movile Cave

Methylotrophic bacteria capable of using methylated amines as a carbon and nitrogen source were selectively enriched using MMA, dimethylamine (DMA) and trimethylamine (TMA). Separate enrichments were set up for each of the three substrates by adding a final concentration of 1 mM substrate to 20 ml cave water in sterile 120 ml serum vials. For mats and biofilms, 2 g sample material was placed into 27 ml serum vials and made up to a final volume of 4 ml with nitrogen-free dilute basal salts (DBS) medium. DBS medium (modified after Kelly and Wood, 1998) contained (per litre): 0.1 g MgSO4· 7H2O, 0.05 g CaCl2 · 2H2O, 0.11 g K2HPO4, 0.085 g KH2PO4, adjusted to pH 7. The medium was supplemented with a vitamins solution as described by Kanagawa et al. (1982) and 1 ml of a trace element solution (modified after Kelly and Wood, 1998) containing (per 1 L): 50 g EDTA, 11 g NaOH, 5 g ZnSO4· 7H2O, 7.34 g CaCl2 · 2H2O, 2.5 g MnCl2 · 6H2O, 0.5 g CoCl2 · 6H2O, 0.5 g (NH4)6Mo7O24· 4H2O, 5 g FeSO4 · 7H2O, 0.2 g CuSO4 · 5H2O, adjusted to pH 6.0. After flushing the headspace of each vial with N2, the headspace was made up to a final concentration of 7% (v/v) O2 and 3.5% (v/v) CO2 to resemble the cave atmosphere. Enrichments were incubated at 21 °C in the dark. After 4 weeks, 10 ml (for water samples) or 4 ml (for mat samples) of fresh DBS medium were added and cultures were spiked with 20 mM MMA, 10 mM DMA or 10 mM TMA. After amending the headspace as previously, enrichment cultures were incubated at 21 °C in the dark. When enrichments became turbid (after a further 2 weeks), dilutions were spread onto agar plates (DBS medium, 1.5% agar) containing 5 mM MMA, DMA, or TMA as the only added carbon and nitrogen source. Plates were incubated at 21 °C in the dark until colonies became visible (2–10 days). In order to achieve isolation of a variety of methylotrophs, individual colonies were examined by microscopy and a selection of morphotypes was transferred onto fresh plates containing the same substrates as before. Cells were observed at × 1000 magnification in phase-contrast under a Zeiss Axioskop 50 microscope (Carl Zeiss Ltd, Cambridge, UK). Isolates were submitted to a series of transfers on plates and microscopy was used routinely to check purity before transferring individual isolates into liquid media (containing 5 mM MMA, DMA or TMA). Once grown in liquid (2–7 days), isolates were transferred back onto methylated amine plates.

In a separate enrichment approach, non-methylotrophic bacteria capable of using methylated amines as a nitrogen (but not carbon) source were isolated. These enrichments were set up in the same manner and using the same sample material as described above for the methylotrophs. In addition to 1 mM of MMA, DMA or TMA, a mixture of alternative carbon compounds (comprising glucose, fructose, succinate, glycerol, pyruvate and acetate) was added to a final concentration of 5 mM. Isolates obtained in this way were additionally tested for growth in liquid medium containing no alternative carbon source to detect any co-enriched methylotrophs, as well as in liquid medium containing carbon sources but no methylated amines to eliminate the possibility that they might be fixing N2 rather than using methylated amines as nitrogen source.

DNA extraction and PCR amplification of bacterial 16S rRNA genes

DNA from cave samples, SIP enrichments and bacterial isolates was extracted as previously described (Neufeld et al., 2007a). DNA from soil and lake sediment samples retrieved from the University of East Anglia campus (used for gmaS primer validation, see later) was extracted using the FastDNA SPIN Kit for soil by MP Biomedicals LLC (Santa Ana, CA, USA). Bacterial 16S ribosomal RNA (rRNA) genes from SIP enrichments were amplified using primer set 341f-GC/907r (Muyzer et al., 1993; Lane, 1991) for analysis by denaturing gradient gel electrophoresis (DGGE). For cloning and sequencing, bacterial 16S rRNA genes from isolates were amplified with primer set 27f/1492r (DeLong, 1992; Lane et al., 1985).

Denaturing gradient gel electrophoresis (DGGE)

DGGE analysis of bacterial 16S rRNA gene fragments was carried out as described by Neufeld et al. (2007a) using the DCode Universal Mutation Detection System (Bio-Rad, Hercules, CA, USA). After electrophoresis for 16 h at 60 °C and 80 V, gels were stained using SYBR Gold Nucleic Acid Gel Stain (Invitrogen, Paisley, UK) and viewed under the Bio-Rad Gel Doc XR gel documentation system using Amber Filter 5206 (Bio-Rad). For gene sequence analysis, well-defined DNA bands were physically excised from the gel for re-amplification using the same PCR conditions and primers described above, followed by sequencing analysis using primer 341f (Muyzer et al., 1993).

DGGE, when compared with amplicon pyrosequencing, is a relatively low resolution technique. However, the DGGE technique enabled us to accurately compare SIP enrichments across different CsCl gradient fractions (heavy to light) and also to compare 13C-incubations to 12C-incubated controls. This first study on MMA degraders in Movile Cave thereby allowed us to identify key players in the microbial food web. Building on data obtained in this study, more detailed studies involving pyrosequencing of amplicons can be carried out in the future.

Functional gene PCR and development of gmaS primers

mauA genes were amplified using PCR primer set mauAf1/mauAr1 (Neufeld et al., 2007a). Currently there is one gmaS PCR primer set available (Chen, 2012) which targets specifically the MRC. This PCR primer set therefore may not detect gmaS from non-marine bacteria. Three new gmaS PCR primers were designed in this study, based on multiple alignment of 34 gmaS sequences derived from (i) methylotrophic isolates confirmed to use the NMG-/GMA-mediated pathway and (ii) bacterial genomes published on the Integrated Microbial Genomes (IMG) platform (Markowitz et al., 2010) of the Joint Genome Institute (JGI) Genome Portal (http://genome.jgi.doe.gov). Genomes were screened for gmaS-related sequences using gmaS from Methylocella silvestris as a query sequence (Chen et al., 2010a). Corresponding full length sequences included both gmaS and glutamine synthetase type III (glnA) sequences, due to the high level of sequence similarity between the two genes. In order to identify genuine gmaS sequences, the gene neighbourhood of all obtained sequences was manually inspected for predicted neighbouring open reading frames typically found adjacent to gmaS (genes encoding NMG dehydrogenase and NMG synthase). Confirmed gmaS sequences included many sequences apparently mis-annotated as glnA. For primer design, multiple sequence alignments of chosen sequences were established with the Clustal X program (Thompson et al., 1997) and viewed using the GeneDoc software (Nicholas et al., 1997). Because of their sequence similarity to gmaS, a number of glnA sequences were included in the alignment in order to identify suitable primer-binding regions specific only to gmaS (for a complete list of all gmaS and glnA sequences used for primer design, see Supplementary Table S1). The resulting forward primer gmaS_557 f (5′-GARGAYGCSAACGGYCAGTT-3′) was used in all cases, with the reverse primers α_gmaS_970r (3′-TGGGTSCGRTTRTTGCCSG-5′) and β_γ_gmaS_1332r (3′-GTAMTCSAYCCAYTCCATG-5′) being used to target the gmaS gene of non-marine Alphaproteobacteria and that of Beta- and Gammaproteobacteria, respectively. Touchdown PCR protocols for gmaS amplification were used as follows: for gmaS_557f/α_gmaS_970r, an initial step at 94 °C for 5 min was followed by 10 cycles of denaturation at 94 °C for 45 s, annealing at variable temperatures for 45 s and extension at 72 °C for 1 min. In the first cycle, the annealing temperature was set to 60 °C, and for each of the nine subsequent cycles the annealing temperature was decreased by 1 °C. This was followed by 30 cycles of 45 s at 94 °C, 45 s at 56 °C and 1 min at 72 °C, and a final extension time of 8 min at 72 °C. For gmaS_557 f/β_γ_gmaS_1332r, a modification of the first touchdown protocol was used; the annealing temperature was set to 55 °C in the first cycle and decreased by 1 °C for each of the nine subsequent cycles. The first 10 cycles were followed by 35 cycles with an annealing temperature of 52 °C.

The primer sets were tested for their specificity by (i) amplification and sequencing of gmaS sequences from genomic DNA of the following bacterial strains known to use the indirect MMA-oxidation pathway: Sinorhizobium meliloti 1021, Mesorhizobium loti MAFF303099, Rhizobium leguminosarum bv. viciae 3841, Agrobacterium tumefaciens C58 and Pseudomonas fluorescens SBW25 (Chen et al., 2010b). For further validation of the primers, gmaS was amplified from DNA extracted from (ii) MMA enrichments from Movile Cave, (iii) fresh Movile Cave mat and (iv) soil and freshwater sediment from a small lake (the ‘Broad’) on the University of East Anglia campus. gmaS-based clone libraries were constructed for (ii)–(iv) and a total of 30 clones were randomly selected for sequencing.

DNA sequencing and phylogenetic analysis

DNA sequencing employed the Sanger method on a 3730A automated sequencing system (PE Applied Biosystems, Warrington, UK). To determine approximate phylogenetic affiliations, partial 16S rRNA gene sequences were analysed with the Basic Local Alignment Search Tool (BLAST) on the NCBI GenBank database (Altschul et al., 1990). Amino acid and nucleotide-based phylogenetic trees were established using the MEGA5 program (Tamura et al., 2011). The evolutionary history was inferred by neighbour-joining (Saitou and Nei, 1987). For nucleotide-based trees (Supplementary Figures 1a and b), the evolutionary distances were computed using the maximum composite likelihood method (Tamura et al., 2004). For amino-acid-based trees, the evolutionary distances were computed using the Poisson correction method (Zuckerkandl and Pauling, 1965). All positions containing gaps and missing data were eliminated. Bootstrap analysis (1000 replicates) was performed to provide confidence estimates for phylogenetic tree topologies (Felsenstein, 1985). Phylogenetic analysis of gmaS genes was carried out at the amino-acid level (135–250 amino-acyl residues).

Nucleotide sequence accession numbers

Nucleotide gene sequences obtained from this study were deposited in the GenBank nucleotide sequence database under the accession numbers KM083620–KM083705.

Results

Active methylotrophic bacteria identified by DNA-SIP with 13C-MMA

DNA-SIP enrichments with 13C-labelled MMA were set up from Movile Cave water in order to identify active, methylotrophic bacteria capable of using MMA as a carbon source. DNA was extracted from microcosms enriched with 13C-labelled and unlabelled MMA after 48 h (t1), 96 h (t2) and 4 weeks (t3). The bacterial communities in the microcosms were investigated by DGGE analysis of bacterial 16S rRNA gene fragments. Comparison of DGGE profiles from unfractionated DNA from the different time points revealed a significant shift in the bacterial community over time, which was similar between 12C-MMA and 13C-MMA incubations (Figure 1).

Figure 1
figure 1

Denaturing gradient gel electrophoresis analysis of bacterial 16S rRNA gene fragments in native (unfractionated) DNA from incubations of Movile Cave water with 13C-MMA (left) and unlabelled MMA (right), after 48 h (t1), 96 h (t2) and 4 weeks (t3).

Full size image

For identification of active methylotrophs, DNA extracted from all time points was subjected to density gradient centrifugation and fractionation, allowing separation of 13C-labelled DNA (contained in heavy fractions) from unlabelled 12C-DNA (contained in light fractions). Bacterial 16S rRNA gene fragments were amplified from all DNA fractions and analysed by DGGE and sequencing. Time point t1 (48 h) did not show any significant enrichment in 13C-DNA and was therefore not further analysed. DGGE analysis of heavy and light DNA fractions from time points t2 and t3 (13C-MMA incubation) revealed major differences in the community profiles of the heavy fractions (Figures 2a and c): A single band dominated the heavy fractions at t2 (96 h, Figure 2a) but was absent at t3 (4 weeks, Figure 2c). Sequence analysis of the excised band revealed that the sequence affiliated with Methylotenera mobilis (99% identity), an obligate methylotroph (Kalyuzhnaya et al., 2006a) known to be abundant in Movile Cave from previous studies (Chen et al., 2009). At t3, several different phylotypes appeared in the heavy fractions of the 13C-MMA incubation (Figure 2c), that is, a more diverse bacterial community had incorporated the label following extended incubation with MMA.

Figure 2
figure 2

Denaturing gradient gel electrophoresis analysis of bacterial 16S rRNA gene fragments in light and heavy DNA fractions from 13C-MMA incubations of Movile Cave water after 96 h (a) and 4 weeks (c) DGGE profiles of unfractionated DNA of both time points (b) are shown for reference.

Full size image

Bacterial 16S rRNA gene sequences from these DGGE bands affiliated with well-characterised methylotrophs such as Methylobacterium extorquens (100% identity) and Methylovorus (97% identity to Methylovorus menthalis), but also included Catellibacterium (98% identity to Catellibacterium caeni), Cupriavidus (99% identity to Cupriavidus necator), Porphyrobacter (99% identity to Porphyrobacter neustonensis) and Altererythrobacter (99% identity to Altererythrobacter epoxidivorans), none of which have previously been reported to grow methylotrophically. The Catellibacterium sequence identified from DGGE shared 98–100% sequence identity with a novel organism subsequently isolated from Movile Cave during this study (see below) and cloned 16S rRNA gene sequences from 13C-labelled DNA from t3 (data not shown, refer to Supplementary Figure S3a).

The non-methylotrophic bacterial community co-enriched in 13C-MMA incubations was investigated by PCR-DGGE of 16S rRNA bacterial genes present in the light fractions (12C-DNA). Light fractions harboured a diversity of mostly heterotrophic bacterial sequences (Figure 2a and b), namely Rhodobacter, Acinetobacter, Azospirillum, Oleomonas and Hydrogenophaga and a number of sequences not closely related to cultivated representatives (as little as 84–87% identity).

All bacterial sequences obtained from DGGE bands in lanes loaded with heavy DNA (that is, 13C-labelled organisms) were exclusive to MMA enrichments, and not seen in no-substrate controls (data not shown). Two sequences detected in the light fractions from MMA incubations (Acinetobacter and Azospirillum) also appeared to be absent in the no-substrate controls, suggesting that these bacteria may have been selectively enriched due to their capability of using MMA as a nitrogen source (but not as a carbon source, and so their DNA was not labelled). One of these sequences (Acinetobacter lwoffi) did indeed correspond to a bacterium isolated from Movile Cave in this study with MMA as the only nitrogen source (see below).

Methylotrophic and non-methylotrophic isolates from Movile Cave

To complement data from 13C-MMA-SIP experiments, methylated amine-utilising bacteria were isolated from different locations (lake room, Airbell 1 and Airbell 2) in Movile Cave. Methylotrophs were isolated with DBS medium containing MMA, DMA or TMA as sole added source of carbon, energy and nitrogen. A selection of isolates differing in colony and cell morphology was transferred into liquid DBS medium containing the respective methylated amine (to distinguish true methylotrophs from organisms growing on agar). Seven methylotrophic strains were isolated, identified based on 16S rRNA gene sequencing analysis (Table 1, Supplementary Figure S3a). The highest diversity of methylotrophs was obtained on MMA enrichments (based on morphology and 16S rRNA gene sequencing data), while DMA and TMA enrichments were dominated by Xanthobacter tagetidis (Padden et al., 1997). Notably, no Methylotenera isolates were obtained (even after using a variety of different cultivation media which are commonly used for methylotrophic bacteria, changing incubation conditions such as temperature, pH, ionic strength of media and dilution-to-extinction experiments), despite the active role of this methylotroph in MMA metabolism as determined by DNA-SIP results (see above), and its apparent abundance in Movile Cave (Chen et al., 2009). In addition to well-characterised methylotrophs such as M. extorquens, two novel methylotrophs were also isolated. A member of the relatively new genus Catellibacterium (Tanaka et al., 2004; Liu et al., 2010, Zheng et al., 2011; Zhang et al., 2012), provisionally named Catellibacterium sp. LW-1 was isolated from lake water enrichments with MMA. 16S rRNA gene sequences relating to this organism were also detected in heavy DNA fractions from 13C-MMA enrichments (see above, Figure 2c, Supplementary Figure S1b), indicating that Catellibacterium may have a significant role in the cycling of methylated amines in Movile Cave. In addition, a new member of the genus Mesorhizobium (a genus not currently known to contain any methylotrophic species), was isolated from an MMA enrichment set up with floating mat from Airbell 1. All methylotrophic isolates were facultative, that is, also able to use sugars or carboxylic acids for growth. Notably, all methylotrophs could use all three methylated amines as sole growth substrates, with the exception of Catellibacterium sp. LW-1 which did not grow on DMA (Table 1).

Table. 1 Growth of bacterial isolates from Movile Cave on methylated amines with and without added carbon
Full size table

In a separate experiment, heterotrophic bacteria capable of using methylated amines as a nitrogen but not carbon source were enriched and isolated using the same sample material as above. MMA, DMA or TMA were the only added nitrogen sources in these enrichments and a mixture of sugars and carboxylic acids were added as carbon and energy source. A diversity of non-methylotrophic methylated amine-utilising bacteria was obtained—in total eight bacterial species, as determined by 16S rRNA gene sequencing analysis (Table 1, Supplementary Figures S3a and b). All of these isolates used MMA as a nitrogen source, while only some could use DMA and TMA (Table 1), suggesting that many lack the enzymes for de-methylation of secondary and tertiary methylated amines to MMA. None of the isolates grew methylotrophically with MMA, DMA or TMA. While all methylotrophic isolates obtained in this study belonged to the Alphaproteobacteria, non-methylotrophic MMA utilisers also included Beta- and Gammaproteobacteria (Table 1). A. lwoffi, isolated from Airbell 2 water with MMA as a nitrogen source, was also detected in 12C-DNA fractions from MMA-SIP incubations (see above), while not seen in control incubations without added MMA. These results suggest that Acinetobacter (and other non-methylotrophs) may have an active role in the cycling of methylated amines in Movile Cave.

Development and validation of functional gene primer sets targeting gmaS

The gene gmaS codes for GMA synthetase, the enzyme catalysing the first step in the conversion of MMA to CH2=THF in the recently characterised indirect MMA-oxidation pathway (Latypova et al., 2010; Chen et al., 2010a, 2010b). We selected gmaS as a functional biomarker to assess the distribution of this pathway among MMA-utilising bacteria. Since currently available gmaS primers are specific to the MRC (Chen, 2012), we designed two new primer sets covering gmaS of non-marine bacteria. Suitable primer regions were identified by alignment of gmaS sequences obtained from (i) isolates confirmed to use the GMAS-/NMG-mediated pathway and (ii) published bacterial genomes. Due to sequence similarity between the two genes, a number of glnA gene sequences were included in the alignment to enable identification of suitable gmaS primer-binding regions not found in glnA.

Sequence alignment and establishment of nucleotide-based and amino-acid-based phylogenetic trees clearly separated glnA from gmaS genes and revealed two distinct gmaS clusters dividing (i) Alphaproteobacteria and (ii) Beta- and Gammaproteobacteria (Figure 3). The alphaproteobacterial gmaS cluster was further split into two major subgroups: ‘Group 1’ contained MRC-associated sequences (in a separate sub-cluster), as well as sequences belonging to soil and freshwater bacteria from the orders Rhodobacterales and Rhizobiales, while ‘Group 2’ contained only gmaS sequences from non-marine bacteria of the orders Rhodospirillales, Rhizobiales and Sphingomonadales (Figure 3). For primer design, sequences associated with the MRC were removed from the alignment as they were too divergent from the other sequences to be targeted by the same primers. A common region shared by all remaining gmaS sequences was used to design the forward primer (gmaS_557f). Two different reverse primers were designed for Alphaproteobacteria (α_gmaS_970r) and Beta- and Gammaproteobacteria (β_γ_gmaS_1332r) because no further region of sufficient similarity shared by both groups could be identified (alignments in Supplementary Figures S4a–c).

Figure 3
figure 3

Phylogenetic relationship of gmaS sequences (135–250 amino acids) derived from published bacterial genomes, methylotrophic (solid rectangles/orange font) and non-methylotrophic (hollow rectangles/blue font) bacterial isolates and clone library sequences (triangles/bold print) from Movile Cave. glnA sequences present the outgroup. The tree was established using the neighbour-joining method (1000 bootstrap replicates) and the Poisson correction method for computing evolutionary distances. 1Rubrobacter xylanophilus is a member of the Actinobacteria although its gmaS sequence affiliates with the beta- and gammaproteobacterial cluster. *gmaS sequences containing a total of more than two mismatches across the forward/reverse primer set designed for the respective gmaS clusters are marked with an asterisk. MRC, marine Roseobacter clade.

Full size image

Specificity of these PCR primer sets was confirmed by amplification and sequencing of gmaS from (i) five bacteria known to use the gmaS-dependent pathway (as specified in Material and methods) (ii) MMA enrichments from Movile Cave (iii) Movile Cave biofilm and (iv) soil and lake sediment from a different environment (UEA campus; as described in Material and methods). All PCR products obtained were of the expected size, that is, ∼410 bp (alphaproteobacterial gmaS) and ∼770 bp (beta- and gammaproteobacterial gmaS). With DNA from MMA enrichments, a slightly larger, second band was obtained in addition to the gmaS product when using 557f/1332r. This gene fragment shared high sequence identity with a viral coat protein and could not be eliminated by using more stringent PCR conditions due to extremely high similarity with the target gene in the primer-binding regions. This alternative amplification product was restricted to Movile Cave enrichment DNA and was avoided by gel excision of the gmaS band. All sequences obtained from genomic DNA (i) and clone libraries (a total of 30 randomly selected clones from (ii), (iii) and (iv)) were identified as gmaS (Figure 3, Supplementary Figure S5), confirming specificity of the primers.

The gmaS sequences obtained from Movile Cave DNA affiliated with gmaS from both methylotrophic and non-methylotrophic bacteria—namely Methylobacterium, Catellibacterium, Pseudomonas and Acinetobacter (99–100% similarity, Figure 3)—identified by DNA-SIP and isolation work in this study. A further sequence loosely affiliated with Methylotenera, Methylovorus and Methylophaga (89–90% similarity with all three genera). A final gmaS sequence was related to gmaS from the methylotroph Hyphomicrobium (99% similarity) which had not been detected by DNA-SIP or isolation.

Distribution of gmaS and mauA genes in Movile Cave isolates

To assess the distribution of the direct and indirect MMA-oxidation pathways in Movile Cave, bacterial isolates were screened for the presence of mauA and gmaS genes. While the mauA-dependent, direct MMA-oxidation pathway is so far only known to exist in bacteria using MMA as a carbon source (that is, methylotrophs), the gmaS-dependent, indirect pathway has recently been shown to also exist in bacteria using MMA for nitrogen nutrition only (that is, non-methylotrophs) (Chen et al., 2010b). Using the gmaS primer sets developed in this study, PCR and sequence analysis of DNA from isolates revealed the presence of gmaS in all eight non-methylotrophic MMA-utilising bacteria and in all seven methylotrophic MMA utilisers (Table 2). Phylogenetic analysis placed the retrieved gmaS sequences within the alphaproteobacterial and the beta-/gammaproteobacterial clusters as expected. Interestingly however, gmaS from Aminobacter, Paracoccus, Catellibacterium, Mesorhizobium and Rhodobacter formed a distinct subgroup within the Alphaproteobacteria, separate from the other freshwater and soil group, and separate from the marine group (Figure 3). mauA was detected in addition to gmaS in four of the seven methylotrophic isolates. These data indicate that the gmaS gene is widespread among MMA-utilising bacteria in Movile Cave.

Table 2 Methylated amine metabolism and the presence of functional gene markers in Movile Cave isolates
Full size table

Discussion

Methylated amine-utilising methylotrophs in Movile Cave

The combination of SIP and cultivation proved very effective for the identification of methylotrophs. DNA-SIP results revealed M. mobilis as one of the major MMA-utilising methylotrophs in Movile Cave, which is in agreement with previous studies which showed high abundance of this organism (Chen et al., 2009). While resisting all isolation attempts, at 96 h in SIP incubations M. mobilis was the first organism that responded to addition of MMA.

The combination of cultivation-based studies and SIP furthermore revealed that a new methylotroph, Catellibacterium sp. LW-1, is an active MMA utiliser in Movile Cave. Growth studies were essential in consolidating DNA-SIP results and confirming Catellibacterium as a novel methylotroph and active MMA-utilising bacterium in Movile Cave. These results also highlight the benefit of analysing SIP enrichments at different time points.

Data from SIP enrichments also suggested that Cupriavidus, Porphyrobacter and Altererythrobacter might have a major role in methylotrophic MMA utilisation alongside known methylotrophs such as Methylobacterium and Methylovorus. While these organisms were not isolated from the cave and have hence not been tested for growth with methylated amines, published genomes of some Cupriavidus/Ralstonia species contain gmaS (refer to trees in Figure 3, Supplementary Figure S5).

Use of methylated amines by non-methylotrophic bacteria in Movile Cave

The large variety of bacterial isolates in Movile Cave using methylated amines as nitrogen sources but not as carbon sources is intriguing, considering the relatively high standing concentrations of ammonium present in Movile Cave water. It is possible that ammonium-depleted areas exist within the microbial mats where utilisation of MMA is advantageous. The fact that nitrogen in the mat is isotopically light while ammonium in the cave water is heavy (Sarbu et al., 1996) could be explained by isotopic fractionation during ammonium assimilation and nitrification. However, it may also indicate that a nitrogen source other than ammonium is used. When growing methylotrophically, some bacterial species have been shown to use the nitrogen of MMA, even when high ammonium concentrations are present (Bellion et al., 1983). The high concentrations of ammonium may even be partly due to the release of excess nitrogen by bacteria using MMA as both a carbon and nitrogen source.

Distribution of the gmaS gene and its use as a biomarker

The newly developed PCR primers targeting gmaS were successful in the detection of MMA-utilising bacteria not covered by currently available primers which target mauA-containing methylotrophs. Results from functional gene screening of non-methylotrophic Movile Cave isolates support previous findings (Chen et al., 2010a) which showed that the gmaS-dependent pathway is used by the non-methylotroph Agrobacterium tumefaciens. Taken together, these results suggest that the gmaS pathway may be the major mode of MMA utilisation in bacteria using MMA as a nitrogen, but not as a carbon, source. Based on our results, the gmaS-dependent pathway also appears to be present in the majority of methylotrophic MMA-utilising bacteria. The direct MMA dehydrogenase (mauA)-dependent pathway, which was detected in a number of methylotrophic isolates in addition to gmaS, seems to be restricted to certain groups of methylotrophic bacteria. It will be interesting to understand how the two pathways are regulated under different growth conditions in organisms containing both.

Conclusions

Combining DNA-SIP and isolation studies, key methylotrophs in Movile Cave were identified and it was shown that methylated amines are important intermediates in Movile Cave, serving as a source of carbon, energy and/or nitrogen for a wide range of bacteria. The GMAS-/NMG-mediated pathway appears to be widespread among both methylotrophic and non-methylotrophic MMA utilisers and newly developed primer sets targeting gmaS have great potential as biomarkers for identification of MMA-utilising bacteria.