Two coal samples were collected from coalbeds in Hancheng (110°45′E, 35°47′N), Shanxi Province, and Bayannaoer (108°11′E, 40°45′N) in the Inner Mongolia Autonomous Region, located at the southeastern and northwestern borders of the Ordos Basin, respectively. The Hancheng coal (H) was a meager lean coal formed in the Carboniferous–Permian period from marine-continental interactive sedimentation. The sample was collected from a coal seam depth of 651–652 m with an in situ temperature of 26.7°C. The Bayannaoer coal (Zhongqi, ZQ) was brown coal formed in the middle and lower Jurassic period from continental sedimentation. The sample was collected from a 38–40 m deep coal seam with an in situ temperature of 20.5°C. Both coals were sampled as big intact blocks (cubes of approximately 10 cm × 10 cm × 10 cm), directly from coal seams which were in production. Coal blocks were immediately put into sterile plastic bags and taken to the lab at 5–7°C within 24 h. Approximately 2 cm of the outer layer of the samples was removed with sterile tools, as the inner part was stored aseptically at -80°C until culture. The coal properties, including moisture, ash content, volatile matter content, sulfur content, and gas content were detected as reported by Tang et al. (2012). The mixed interlayer water was collected from different layers in Hancheng area for media preparation.

About 60 g of each coal sample was grinded using sterile mortars and screened with 16 meshes. The coal powder was then mixed with 300 ml of interlayer water. The mixer was rotated at a speed of 800 rpm at 30°C for 20 h. After natural precipitation, 1 ml of supernatant was removed and serially diluted. Three kinds of media were used to isolate bacteria from all samples: (i) coal medium (M) (60 g coal with 300 ml interlayer water mixed and rotated for 20 h in a 30°C water bath, and then sterilized at 121°C for 1 h; the mixture was centrifuged and the supernatant was diluted 10-fold; agar was added at 2.0% w/v, and media was again sterilized); (ii) mineral medium (W) [L^-1: 5 g NaCl, 1 g K₂HPO₄, 1 g NH₄H₂PO₄, 1 g (NH₄)₂SO₄, 0.2 g MgSO₄⋅7H₂O, 3 g KNO₃, and 20 g agar, sterilized at 121°C for 1 h]; and (iii) coal/mineral medium (MW) [5 g NaCl, 1 g K₂HPO₄, 1 g NH₄H₂PO₄, 1 g (NH₄)₂SO₄, 0.2 g MgSO₄⋅7H₂O, 3 g KNO₃, and 20 g agar in 1 L M medium and sterilized at 121°C for 1 h]. Each sample was serially diluted and spread on the three media agar plates in triplicate, and cultured at 30°C for 3–8 days. All isolates growing on the plates were picked and transferred to the same freshly prepared media, until pure cultures were obtained; isolates were stored for further investigation. The isolates were named after the sample location, medium name, and the colony series numbers. For example, the first and fifth strains originating from Hancheng, and isolated from M medium, were named as HM1 and HM5, respectively.

A single colony was transferred to a glass tube with 1 ml of liquid medium (corresponding to the original medium used for isolation) without agar. The cultures were incubated for 3–5 days at 30°C with shaking at 180 rpm. The cultured cells were collected and concentrated and the extraction of genomic DNA and PCR amplification of the 16S rRNA gene were performed as described by Wang et al. (2007). The 16S rRNA gene was amplified from genomic DNA of all isolates using a bacterial universal primer set as described previously (Wang et al., 2007) (Supplementary Table S1). The PCR products were purified with the QIAquick PCR Purification Kit (Qiagen, Shanghai, China) and digested with Rsa I and Msp I (TaKaRa, Dalian, China). The isolates were digested firstly by Rsa I at 37°C for 5 h, then the isolates with same restriction enzyme digestion patterns were selected to be digested by Msp I at 37°C for 5 h. The total reaction system was 20 μl with 5 μl PCR products, 2 μl 10 × buffer (including 0.1% BSA), 0.5 μl restriction endonucleases and 12.5 μl deionized water. According to patterns of 16S rRNA gene fragment restriction digestion. The 16S rRNA genes, from 876 isolates, were classified into 77 patterns, and the PCR products of each pattern were further sequenced (Cook and Meyers, 2003; Sun et al., 2013). The obtained DNA sequences were aligned in GenBank using the BLAST tools¹. The reference sequences were retrieved from GenBank. After multiple sequence alignment of the sequences by CLUSTAL X and manually correction, 1283 bp was selected to construct phylogenetic tree using the neighbor-joining method in the MEGA software package version 5.0 (Tamura et al., 2011).

To rapidly distinguish bacterial species and strains of the dominant Thauera genus, we randomly selected 145 of 224 total Thauera isolates (according to 16S rRNA gene sequences and 16S rRNA gene fragment restriction digestion patterns), and investigated their distributions of repetitive DNA elements in the 16S rRNA genes according to BOX-PCR and repetitive extragenic palindromic PCR (REP-PCR) patterns. Primers (Sangon, Shanghai, China) used for BOX-PCR and REP-PCR are listed in Supplementary Table S1. Amplification was performed using rTaq DNA polymerase (TaKaRa, Dalian, China) in a MJ Mini Personal Thermal Cycler (Bio-Rad, Hercules, CA, USA) as follows: (1) BOX-PCR: 5 min at 95°C, 35 cycles of 1 min at 53°C, 4 min at 72°C, and 4 min at 72°C, with a single final extension of 10 min at 72°C; (2) REP-PCR: 5 min at 94°C, 30 cycles of 30 s at 95°C, 60 s at 45°C, and 8 min at 65°C, with a single final extension 16 min at 65°C. The reaction products were stored at 4°C until they were subjected to gel electrophoresis using a 2.0% (w/v) agarose gel with a 100 bp DNA ladder (Tiangen, Beijing, China).

To detect the ability of bacterial oxidation of lignin coupled to the reduction of molecular dioxygen to water (Couto and Herrera, 2006), GU-WA medium (10 g NaH₂PO₄⋅12H₂O, 2 g KH₂PO₄, 0.5 g NaCl, 0.5 g NH₄NO₃, 0.5 g MgSO₄⋅7H₂O, 0.1 mg FeSO₄7⋅H₂O, 0.1 mg CaCl₂, 1 mg CuSO₄, 0.2 g guaiacol, 2 g 80 mesh Eucalyptus wood powder, and 20 g agar in 1 L distilled water, adjusted to a final pH of 7.0 and then sterilized; Jung et al., 1995; Li et al., 2006), was prepared for culturing 77 representative strains (based on 77 patterns achieved through restriction digestion of PCR-amplified 16S rRNA gene) from Hancheng and Zhongqi. The strains were pre-cultured in W medium with 0.01% yeast extract and 0.01% tryptone for 3 days, after which 3 μl inoculum was transferred to GU-WA medium and cultured at 30°C. After 7–10 days, the color of the circle surrounding inoculation sites was assessed for the presence or absence of a color change.

Previously described primers used to amplify the LMCO gene are listed in Supplementary Table S1. DNA fragments corresponding to the correct target size for each gene were cloned into the pGEM-T Easy Vector System I (Promega, Madison, WI, USA) and sequenced. The obtained DNA sequences were aligned in GenBank using the Blastx ^{footnotenumfont 1}. The functionally identified reference sequences or those with the highest similarities to the query sequences were retrieved from GenBank. All sequences were aligned using CLUSTAL X with manual correction. The phylogenetic tree of the LMCO genes based on amino acid sequences was constructed using the neighbor-joining method in the MEGA software package version 5.0 (Tamura et al., 2011). Trees were bootstrapped using 1000 replications. The stability of tree topology was evaluated with maximum-likelihood and maximum parsimony algorithms.

One strain Massilia sp. ZQW3 was inocubated by Luria-Bertani (LB) liquid medium at 30°C for 3 days with shaking of 180 rpm. Bacteria cells were collected (3200 × g, 5 min) and washed three times by basal salt medium (10 g NaH₂PO₄⋅12H₂O, 2 g KH₂PO₄, 0.5 g NaCl, 0.5 g NH₄NO₃, 0.5 g MgSO₄⋅7H₂O, 0.1 mg FeSO₄⋅7H₂O, 0.1 mg CaCl₂, 1 mg CuSO₄ in 1 L distilled water, adjusted to a final pH of 7.0). Then cells were inocubated to basal salt medium with lignin (10 g/L) and glucose (10 g/L) as sole carbon source, respectively. The final optional density (OD) was 1.0. The cultures were incubated for 24 h at 30°C with shaking at 180 rpm, and the total RNAs were obtained according to RNAprep Pure Cell/Bacteria Kit (TIANGEN). Primers used in quantitative real-time PCR (qRT-PCR) were 16S rRNA-F (5′-CTCCTACGGGAGGCAGCAGT-3′), 16S rRNA-R (5′-CGTATTACCGCGGCTGCTGG-3′), LMCO-F (5′-ATCCTGGTGCC-GGCGGAGAT-3′), and LMCO-R (5′-GTTTTGCTGGAGCTTGAACT-3′).

The reverse transcription PCR for complementary DNA (cDNA) was performed followed by qRT-PCR using the SYBR^® Green method as the instruction of SYBR Premix ExTaq^TM GC (TaKaRa). Cycling parameters of qRT-PCR reactions were programmed with an initial step of 30 s at 95°C followed by 50 cycles consisting of denaturation at 95°C for 5 s, annealing at 55°C for 30 s, then 10 s at 95°C, melt curve 65 to 95°C, increment 0.5°C for 5 s. The relative quantification was analyzed with the 2^-ΔΔCt method. Technical triplicates were performed for each biological replicate, and the average values were used for quantification. 16S rRNA herein was used as internal control to normalize the relative transcription of the analyzed gene.

A total of 393 isolates and 483 isolates were obtained from the coal of Hancheng and Zhongqi, respectively. The 16S rRNA genes were amplified by PCR and classified according to 16S rRNA gene fragment restriction digestion patterns (using two endonuclease cleaving enzymes). One 16S rRNA gene amplicon of each pattern was selected randomly for sequencing. Through sequencing and alignment, the sequences having >97% similarity were grouped into a cluster. Based on restriction enzyme digestion results and sequencing results, all bacterial strains could be classified into 77 clusters, which contained 27 genera from Actinobacteria (339 isolates from 8 genera), Firmicutes (23 isolates from 5 genera), and Proteobacteria (514 isolates from 14 genera) (Figures 1A–C; Table 1). These isolates from Hancheng and Zhongqi showed 98.80–100.00% and 98.00–100.00% 16S rRNA gene identities to valid reference strains, respectively. Principal component analysis and clustering analysis were performed on all the bacterial patterns (Figures 2A,B), which showed that isolates were site-specific more significantly than medium-specific.

Isolates from the genus Thauera dominated those from Hancheng, accounting for 57% of the total; the remaining isolates were from the genera Arthrobacter, Rhizobium, Staphylococcus, Brachybacterium, Micrococcus, Methylobacterium, Tessaracoccus, Brevibacterium, Janibacter, Leucobacter, Devosia, Acinetobacter, Aerococcus, Diaphorobacter, Providencia, Lysinibacillus, Bacillus, Shinella, Proteus, Ancylobacter, and Bosea (Figure 3A). In contrast, Arthrobacter dominated all isolates from Zhongqi (64%), and the remaining isolates were Rhizobium, Staphyloco-ccus, Brachybacterium, Micrococcus, Methylobacterium, Terra-bacter, Brevundimonas, Curtobacterium, Massilia, and Pseudom-onas (Figure 3B). Isolates from Hancheng coal had higher Shannon-Weiner index values, 2.83, at the 16S rRNA gene similarity level, compared to 2.13 of Zhongqi coal (Table 1), suggesting higher bacterial diversity in Hancheng. Strains belonging to Arthrobacter, Rhizobium, Staphylococcus, Brachybacterium, Micrococcus, and Methylobacterium genera were isolated from both Hancheng and Zhongqi, which occupied 32.06 and 97.31% of total isolates, respectively. Strains belonging to Rhizobium could be isolated from all the media from both Hancheng and Zhongqi (Figure 4A). Thauera, Tessaracoccus, Brevibacterium, Janibacter, Leucobacter, Devosia, Acinetobacter, Aerococcus, Diaphorobacter, Providencia, Lysinibacillus, Bacillus, Shinella, Proteus, Ancylobacter, and Bosea were only found in Hancheng (Figure 4B; Supplementary Table S4), and Terrabacter, Brevundimonas, Curtobacterium, Massilia, and Pseudomonas were only found in Zhongqi (Figure 4B; Supplementary Table S4).

In this study, we used three different types of media to isolate bacteria from coal. There were 135, 66, and 192 isolates obtained from Hancheng using M, W, and MW media, respectively. The Shannon-Weiner index values of the isolates using M, W, and MW media were 2.467, 2.311, and 2.369, respectively (Table 1). This showed the highest diversity of isolates was obtained using M medium and the lowest diversity was obtained using W medium. In addition, M medium also obtained the highest diversity of isolates from Zhongqi. The Shannon-Weiner index values of isolates from Zhongqi using M, W, and MW media were 2.061, 1.799, and 0.562, respectively (Table 1). Results of both Hancheng and Zhongqi showed that M medium could achieve the highest diversity of isolates, whereas it might activate more bacteria living from a certain environment than other media. Strains belonging to Rhizobium, Staphylococcus, Bacillus, Methylobacterium, and Thauera genera were isolated in all three media from Hancheng, whereas Arthrobacter and Rhizobium genera were isolated from Zhongqi using all three media (Figure 4C).

Among all isolates from the two coal samples, isolates of Thauera, Arthrobacter, and Rhizobium were the most prevalent. There were 224 Thauera isolates obtained from Hancheng coal with 32, 20, and 172 isolates obtained from M, W, and MW media, respectively. (Figure 3A; Supplementary Table S4). No Thauera isolates were identified from Zhongqi. These Thauera isolates were closely related to the species Thauera phenylacetica B4P (T), based on 16S rRNA gene sequence, with the highest similarities being 99.40%. Further BOX-PCR and REP-PCR analyses showed that there were 27 different patterns for Thauera (Supplementary Table S5). There were 309 Arthrobacter strains isolated from Zhongqi coal, including 7, 121, and 181 isolates from M, W, and MW media, respectively (Figure 3A). Among these, 117 and 179 isolates were closely related to Arthrobacter defluvii 4C1-a(T) and A. oryzae KV-651(T), with 16S rRNA gene sequence similarities being 99.50-99.57 and 98.50-100.00%, respectively (Supplementary Table S2). BOX-PCR and REP-PCR analyses showed 10 patterns of Arthrobacter (Supplementary Table S6). Rhizobium was the only genus that could be found in both coals and could be isolated using all media. A total of 246 Rhizobium isolates were isolated from Hancheng and Zhongqi, among which, 145 isolates from ZQM were closely related to Rhizobium pusense NRCPB10(T), with 16S rRNA gene sequence similarities being 99.80%; 76 isolates (70 isolates from HM, 1 isolate from HW, and 5 isolates from HMW) were closely related to R. selenitireducens B1(T) with 16S rRNA gene sequence similarities being 98.80-98.90%; 16 isolates (1 isolate from HM, 11 isolates from HW, 1 isolate from ZQM, 2 isolates from ZQW, and 1 isolate from ZQMW) were closely related to R. massiliae 90A, with 16S rRNA sequence similarities being 99.70-100.00%, and one isolate, only from ZQM, was closely related to R. cellulosilyticum ALA10B2(T), with 16 rRNA gene sequence similarity being 99.20%; two isolates from ZQM were closely related to R. alkalisoli CCBAU 01393(T), with 16S rRNA gene sequence similarities being 98.29%, and six isolates were closely related to R. borbori DN316(T) isolated from HW, with 16S rRNA gene sequence similarities being 99.69%. All Rhizobium isolates could be assigned to 12 patterns based on BOX-PCR and REP-PCR analyses (Supplementary Table S7).

Of all 876 isolates, 218 isolates (comprising 35 species) from Hancheng and 394 isolates (comprising 19 species) from Zhongqi showed positive results for lignin degradation, with varying abilities, on GU-WA agar plates (Figure 5A). Of these, 27, 335, and 250 isolates had very strong, strong, and weak lignin degradation ability, respectively (the color reaction results indicating different degrees of lignin degradation ability are shown in Figure 5B). The strains with very strong lignin degradation belonged to Arthrobacter (10/310 isolates), Brachybacterium (2/6 isolates), Methylobacterium (10/19 isolates), Rhizobium (1/246 isolate), and Massilia (4/5 isolates) (Supplementary Table S4). Genera having the highest number of lignin-degrading isolates were Arthrobacter (80.65% of Arthrobacter isolates) and Thauera (40.18% of Thauera isolates) in Hancheng and Zhongqi, respectively. Of the Rhizobium genus, 78.86% of isolates from both Hancheng and Zhongqi had lignin-degrading abilities (Supplementary Table S4).

From Hancheng, 90 of 224 isolates of Thauera, and 85 of 94 isolates of Rhizobium showed lignin-degrading abilities. From the remaining genera, 12 of 16 isolates of Methylobacterium, five of nine isolates of Tessaracoccus, four of five isolates of Brachybacterium, and four of eight isolates of Bacillus were positive for lignin degradation. In contrast, 255 of 309 isolates of Arthrobacter and 129 of 152 isolates of Rhizobium, from Zhongqi, had lignin-degrading abilities. From the remaining genera, four of five isolates of Massilia and two of two isolates of Pseudomonas had lignin degrading ability (Supplementary Table S4). Among all lignin degrading isolates, 233, 151, and 228 isolates were detected from M, W, and MW media, respectively. In contrast, lignin degradation abilities were not detected with isolates from the genera Curtobacterium, Brevibacterium, Devosia, Brevundimonas, and Proteus.

For Arthrobacter, 82.26% strains had lignin degradation ability. Among these functional strains, nine isolates had very strong lignin degradation abilities. These isolates were related to A. chlorophenolicus A6(T) (one isolate) and A. oryzae KV-651(T) (eight isolates), based on 16S rRNA gene similarities of 99.48% and 98.50–100.00%, respectively. For Rhizobium, 87.0% of isolates had lignin degradation ability. Among all 214 functional strains, one isolate was very strong for this function and was related to R. massiliae 90A, based on a 16S rRNA gene similarity of 100.00% (Supplementary Table S2). For Thauera, 40.17% strains had lignin degradation ability. However, none showed very strong lignin degradation ability.

Along with the colony morphological characteristics, BOX-PCR, and REP-PCR analyses, 27, 12, and 10 Thauera, Rhizobium, and Arthrobacter isolates, covering all patterns, as well as 53 other strains representing separate clusters, were selected for detection the of LMCO genes.

From 13 isolates, belonging to 10 genera of 102 original isolates, 16 LMCO genes with different amino acid sequences were detected (Figure 6A). This is the first report of LMCO genes discovered in the genera Aerococcus, Shinella, and Massilia. Phylogenetic analysis showed that these laccase genes were separated into two major clusters. Cluster I contained a mix LMCO genes from different taxa, such as Rhizobium, Providencia, Pseudomonas, Massilia, and Thauera from Proteobacteria, as well as Aerococcus, Janibacter, and Staphylococcus, from Actinobacteria and Firmicutes. Cluster II contained LMCO genes from Providencia, Shinella, Sinorhizobium, and Rhizobium. The topology of the LMCO gene based phylogenetic tree was much different from that of 16S rRNA gene-based phylogenetic tree (Figures 6A,B). For example, LMCO genes from Rhizobium were separated into different groups, whereas the 16S rRNA gene analysis resulted in these isolates being clustered together. The LMCO gene from Rhizobium sp. ZQM163 was closely related to that of Pseudomonas rather than Rhizobium. LMCO sequences from Thauera sp. HMW163 and Thauera sp. HMW58 were closely related to Roseovarius and Agrobacterium, respectively. However, the 16S rRNA gene sequences from these isolates shared 100.00% similarity and were closely related to Agromyces. Two LMCO genes were found in Providencia sp. HW8, one of which (HW8_45) was closely related to that from Pseudomonas, whereas the other (HW8_21) was closely related to that of Rhizobium. Moreover, genes in Cluster II, from both Proteobacteria and Actinobacteria, showed higher similarities (85.00–100.00%), which suggested that these genes might have been generated from the same ancestor. The results indicated that horizontal gene transfer of LMCO might happen frequently during these strains.

To detect the gene expression, one of stains that the LMCO were first detected was selected. Massilia sp. ZQW3 was chosed to detect LMCO gene expression. Expression levels were significantly upregulated with the stimulation of lignin, but in relative low levels when induced by glucose (Figure 7).

In summary, 876 strains were isolated from low and high rank coal of the Ordos Basin, using three kinds of media. Most isolates could degrade lignin, suggesting the potential for the application of bioconversion of coal. From 13 isolates, 16 LMCO genes with different amino acid sequences were detected. This is the first report of LMCO genes discovered in the genera Aerococcus, Shinella, and Massilia. High LMCO gene expression level was also detected in the genera Massilia. Moreover, LMCO genes were absent in most lignin degrading bacteria, suggesting the presence of other enzymes mediating lignin degradation. However, the indigenous microbial community of coal environments, especially the culturable strains, is still far from clear. Considering the poor culturability of oligotrophic bacteria, and the limitations of degenerate PCR approaches, more bacteria with novel genes and functions might be found in the future.