BPA-degrading microbial community and meta-omics analysis pipeline
We enriched a BPA-degrading culture from activated sludge using BPA as the sole electron donor and carbon source. BPA concentrations were increased from 20 to 50 mg L−1 over a 5-month period. A workflow was developed for analyzing and integrating various meta-omics datasets in order to investigate the metabolic capabilities and correlations between different bacterial populations in the BPA-degrading community. This workflow was also used to integrate pure culture isolation and analysis to test the proposed metabolic interaction model revealed from meta-omics analyses (Fig. 1).
We used three major types of integrated analysis to identify differences in encoded and expressed microbial functions in the context of metabolite variation in the BPA-degrading microbial communities. First, since many transient intermediates were not detected by liquid chromatography in conjunction with tandem mass spectrometry (LC-MS/MS), we integrated the metabolic capacity identified by functional annotation of metagenomic data to the metabolite analysis to reconstruct the community-wide BPA-mineralizing pathway. Second, we combined the metabolic capabilities identified for each microbial genome from the metagenomics data with gene expression profiles obtained from the metatranscriptomics to obtain the gene expression of specific dominant pathways. 16S rRNA gene analysis confirmed the dominance of the bacterial populations identified from metagenomic data. Mapping of dominant gene expression to the community-wide BPA-mineralizing pathway thus revealed the pathway profile of dominant species and the potential roles of individual species to the overall degradation. Finally, we isolated dominant strains from the enrichment to confirm the metabolic interactions identified from meta-omics analyses.
Reconstruction of community-level BPA-mineralizing pathway using metabolite analysis and functional annotation of metagenome
To characterize the BPA-mineralizing pathway in the enrichment, we monitored the dynamics of BPA and its known degradation intermediates over 48 h in four batch experiments amended with either BPA or one of its previously reported degradation intermediates, i.e., 1-BP, 2-BP, and 4-DM (Fig. 2). These degradation profiles indicate that the community transformed BPA by two divergent pathways via either 1-BP or 2-BP as the respective major intermediate. Further transformation of 1-BP generated 4-DM that was then transformed to either 4-HDB or 4-HAP. 2-BP was further transformed to 2,4-BP and 3,4-BP.
In the experiments with BPA amendment, higher concentrations of 2-BP pathway intermediates were detected than 1-BP pathway intermediates, indicating the accumulation of these intermediates (Fig. 2a). Indeed, comparison of culture amended with either 2-BP or 1-BP showed that 1-BP was more readily and rapidly degradable while 2-BP was more recalcitrant to biodegradation (Fig. 2b, d). When the culture was amended with 10 mg L−1 2-BP or 2,4-BP, accumulation of 2,4-BP and 3,4-BP were also observed during the 48-h incubation. In contrast, the intermediates in the culture amended with 1-BP degraded fairly quickly. Although 4-HBD was observed when the enrichment was amended with 50 mg L−1 BPA, it was absent when either 1-BP or 4-DM was amended, indicating the readily degradable nature of these compounds (Fig. 2b, c).
Since the LC-MS/MS analysis did not detect any downstream metabolites potentially involved in the conversion of 4-HBD/4-HAP/2,4-BP/3,4-BP to the intermediates in TCA cycle, we sought to identify the lower pathway of conversion of 4-HBD/4-HAP/2,4-BP/3,4-BP using functional annotation of assembled open reading frames (ORFs) in metagenomics analysis (details described in the next section). These analyses indicate that the BPA-degrading community possessed the genes encoding the transformation of 1-BP pathway downstream intermediates, 4-HBD and 4-HAP (Fig. 2e). 4-HBD could be further transformed to either oxoacetate/pyruvate or succinyl-CoA via 3,4-dihydroxybenzoate (3,4-DHB). 4-HAP could be further transformed to succinyl-CoA via 3-oxoadipate (3-ODP). Currently, it is unclear how 2,4-BP and 3,4-BP are transformed to TCA cycle intermediates. These integrated data of metabolites and metagenomics indicate that the enrichment culture mineralized BPA mainly through the 1-BP pathway via either 4-HDB or 4-HAP to the TCA cycle.
Integration of metagenomic and metatranscriptomic data to investigate the roles of individual microbial populations in BPA mineralization
To profile the microbial metabolic capability and activity, a total of ~ 36 million metagenomic raw reads (~ 8.8 Gbp), obtained from two different metagenomic libraries, and ~ 256 million metatranscriptomic raw reads (~ 56.4 Gbp), from eight different metatranscriptomic libraries, in addition to ~ 10,000 raw reads of 16S rRNA gene sequences were obtained from biomass samples collected at two different time points from the enrichment culture using 50 mg L−1 BPA as the sole substrate (detail of samples and sampling please refer to method sections).
To determine the identities and functions of microbial populations in the enrichment, assembled contigs from metagenomics sequencing were binned using bi-dimensional coverage plots. This analysis identified ten bacterial genomes. While the completeness of seven genomes was higher than 96%, that of the remaining three genomes was below 60% (Fig. 3 and Additional file 1: Table S1). The functions of 62.3 to 92.5% of predicted ORFs of the recovered genomes were identified using BLASTp against NCBI-non-redundant protein sequences, KEGG, and Brenda databases (Additional file 2 and Additional file 3: Table S2). Taxonomic annotation using genome taxonomy database (GTDB) [33] indicated that the genomes belong to eight genera (Additional file 1: Table S1). Two genomes of the genus Sphingomonas were found to possess all known genes necessary for degrading BPA. Four genomes from the genera, Pseudomonas, Leucobacter, Pusillimonas, and Pandoraea, respectively, contained genes only encoding either full or partial intermediate-degrading pathways (Additional file 1: Table S1).
The 16S rRNA gene 454-pyrosequencing analysis revealed that four bacteria, including two Sphingomonas spp., Sph-1 and Sph-2, a Pseudomonas sp., and a Pusillimonas sp., were the predominantly populations after BPA amendment, accounting for 38.2 ± 1.5%, 4.5 ± 0.6%, 20.8 ± 1.5%, and 5.8 ± 1.0% of relative abundance of community members, respectively (Fig. 3). No obvious abundance shifting of the four strains was observed during the degradation process. Our metatranscriptomic analyses therefore focused on these four strains because of their potential involvement in BPA biodegradation and high abundance as well as high genome completeness and low genetic contamination (≤ 1%). We selected four BPA biodegradation stages: phase I was 24 h after the enrichment culture inoculated into basal medium without BPA; phases II and III were 2 h and 14 h, respectively, after BPA amendment; and phase IV was 24 h after BPA was provided (Fig. 2a).
Metagenomic analysis identified 3707 and 6077 ORFs in Sph-1 and Sph-2, respectively (Additional file 2). Both Sph-1 and Sph-2 contained genes encoding enzymes predicted to convert BPA to 1-BP and 2-BP. Nine putative genes were predicted to encode cytochrome P450 enzymes (CYP) in Sph-1 and Sph-2. One of the proposed CYP gene that was present in both genomes shares 99% similarity in nucleotide sequence (across 98% of its full-length 1290 bps) with a CYP gene that was previously characterized as the principal BPA-degrading enzyme of Sphingomonas sp. AO1 [34]. The same scaffold where the CYP gene was found also contained the other component of the BPA-degrading enzyme system, a ferredoxin gene which shares 100% similarity with the ferredoxin from the strain AO1 [34]. Although the degradation reactions have been reported after the initial transformation of BPA to 1-BP or 2-BP, the enzymes involved in these steps are still unknown. The two Sphingomonas genomes also carried the genes encoding the conversion of 4-HBZ to oxaloacetate/pyruvate and 4-HAP to 4-HPAT (Fig. 4). Specific genes encoding the conversion of 4-HPAT to hydroquinone (HQN), and 4-hydroxyphenacyl alcohol (4-HPAH) to 4-HBZ, were only present in Sph-1. The gene encoding 4-HBD conversion seemed to be absent from either Sphingomonas genome. Instead, a gene was found in both Sphingomonas genomes to transform salicyladehyde, a 4-HBD isomer, indicating novel pathways might be involved in the conversion of 4-DM to 4-HBZ.
To our knowledge, the two Sphingomonas genomes recovered herein were the first two reported draft genomes of BPA-degrading Sphingomonas spp. To achieve a better understanding of their phylogenetic relationship with other Sphingomonas spp., nearly full-length 16S rRNA gene sequences were obtained from 16S rRNA gene cloning. Phylogenetic analysis indicates both Sph-1 and Sph-2 were distantly related to each other and to previously reported BPA-degrading Sphingomonas spp. (Additional file 2 and Additional file 3: Figure S1).
Metatranscriptomics analysis of Sph-1 and Sph-2 indicated that there were two major response groups of the BPA pathway genes that we identified as A (arched)-shaped expression patterns (up-regulated in phases II and III; flat or downregulated in phase IV) and U-shaped expression patterns (downregulated in phases II and III; flat or upregulated in phase IV). The genes encoding the CYP and the ferredoxin exhibited consistent A-shaped patterns, indicating their important roles in BPA biodegradation (Fig. 4). Similarly, the expression of most genes in the 4-HBZ to pyruvate/oxaloacetates pathways (pobA, ligAB, ligC, galD, ligJ, and ligK) and the TCA cycle genes were also A-shaped (Fig. 4). In contrast, hapA and hapB, involved in the conversion of 4-HAP to HQN were U-shaped as was dhad, indicating the inactivity of these genes in our experiments. Interestingly, although the genes involved in 2-BP degradation are unknown, a few oxidoreductases were upregulated only in the final phase where 2-BP pathway intermediates were dominant, suggesting their potential correlation to 2-BP degradation (Additional file 2 and Additional file 4: Figure S2). These analyses suggest that Sph-1 and Sph-2 were the major BPA-degrading populations in the community.
Metagenomic analysis indicated that the Pseudomonas and Pusillimonas in the enrichment contained 5672 and 4050 ORFs, respectively (Additional file 2). The Pseudomonas sp. lacked genes responsible for initial BPA degradation to either 1-BP or 2-BP but possessed a complete pathway that converts 4-HBD to succinyl-CoA and the gene encoding the 4-HBZ transporter across the membrane (Fig. 4). Similarly, the Pusillimonas sp. possessed two almost complete pathways for converting 4-HBZ to either oxaloacetate/pyruvate or succinyl-CoA and the 4-HBZ transport gene (Fig. 4). Also, the Pseudomonas genome contained a complete pathway for the transformation of 4-HAP to 3-ODP. The Pusillimonas genome encoded a lower 4-HAP pathway converting HQN to 3-ODP, but the genes responsible for transformation of 4-HAP to HQN were missing.
Pseudomonas sp. exhibited A-shaped patterns for those genes involved in 4-HBZ transporter, conversion of 4-HBZ to succinyl-CoA, and the TCA cycle. The genes for converting 4-HBD to 4-HBZ (pchA) and 4-HAP to 3-ODP (hapA, hapB, hapC, hapE, hapF) were U-shaped, suggesting they might not participate in BPA biodegradation under the tested conditions (Fig. 4). Unlike Pseudomonas sp., Pusillimonas sp. exhibited inconsistent expression patterns for genes encoding either pathways for the conversion of 4-HBZ downstream intermediates, suggesting that it might use a hybrid pathway to mineralize 4-HBZ. The lower 4-HAP-degrading pathway (from HQN to succinyl-CoA) also exhibited U-shaped regulation. The upregulation of genes in this pathway in phase IV indicates their functions at the late stage of BPA biodegradation (Fig. 4).
The integration of 16S-sequencing, metagenomic, and metatranscriptomic analyses revealed an interesting metabolic interdependence between Sphingomonas spp. and Pseudomonas sp. or Pusillimonas sp. (Fig. 4). This interaction model suggests that two Sphingomonas species were the key BPA-degraders, converting BPA to 4-HBZ and other intermediates that likely supported the growth of non-degrading microbial populations. (Fig. 4).
Confirmation of microbial interactions in BPA biodegradation using bacterial isolates and consortium from the BPA-degrading community
To determine if the hypothesized substrate cross-feeding played a role in the BPA-degrading efficiency, we designed isolation strategies in order to capture both the BPA-degrading and lower pathway metabolite-utilizing bacteria by using either BPA-containing or non-selective medium. This approach isolated a Sphingomonas sp. and a Pseudomonas sp. from the BPA-containing and non-selective medium, respectively. Genomic analysis of the draft genome of these isolates showed the average nucleotide identities were 100 ± 0.48% and 100 ± 0.04% similar to the binned genomes of Sph-2 and Pseudomonas sp. respectively.
In batch axenic culture, Sph-2 quickly degraded BPA, 1-BP, 4-DM, 4-HBD, and 4-HBZ, but slowly degraded 2-BP and 4-HAP. Sph-2 was inefficient at degrading 2,4-BP, 3,4-BP, and 4-HPAT that were accumulative during the incubation (Fig. 5a and Additional file 5: Figure S3a, b). These results are in agreement with the downregulation of the genes for conversion of 4-HAP and 4-HPAT, and upregulation of genes involved in 4-HBD and 4-HBZ degradation (Fig. 4). Consistent with our prediction from the integrated meta-omics analysis, Pseudomonas sp. was efficient at degrading 4-HBZ, but not BPA, 1-BP, 4-DM, and 2-BP. Pseudomonas sp. also demonstrated the abilities to degrade 4-HBD, 4-HAP, and 4-HPAT with higher efficiencies to degrade 4-HBD and 4-HBZ than 4-HAP and 4-HPAT (Additional file 5: Figure S3c), but lacking capacity to degrade BPA, 1-BP, 2-BP and 4-DM, which is consistent with metagenomics prediction. The downregulation of genes involved in the transformation of 4-HAP and 4-HPAT (hapA and hapB) observed in the community probably reflected that they were less favorable for Pseudomonas sp.
The co-culture of Sph-2 and Pseudomonas sp. demonstrated faster and more complete BPA mineralization than the Sph-2 axenic culture, even though comparable BPA degradation rates were observed in both sets of cultures (Fig. 5a–c). For example, after 24-h incubation, about 69 ± 0.5% of TOC was removed in the co-culture, whereas only about 40 ± 0.6% disappeared in the axenic culture. At the end of the incubation (72 h), about 84 ± 0.4% and 77 ± 0.4% of TOC were found in the co-culture and axenic culture, respectively. The higher TOC removal efficiency, especially in the first 24 h, observed in the co-culture was related to the disappearance of intermediates such as 1-BP, 4-DM, 4-HBD, 4-HBZ, 4-HAP, and 4-HPAT. In fact, 4-HBD, 4-HBZ, and 4-HPAT were not detected in the co-culture, indicating their fast degradation by Pseudomonas sp. (Fig. 5b). Although Pseudomonas sp. was incapable of degrading 4-DM, 4-DM was not detected in the co-culture. The disappearance of 4-DM was probably caused by a faster consumption by Sph-2 as a result of the fast removal of 4-DM downstream metabolites by Pseudomonas sp. In agreement with the fast utilization of BPA and intermediates, both cell numbers of Sph-2 and Pseudomonas sp. increased significantly over 24 h. Similar Sph-2 cell numbers ((1.4 ± 0.4) × 106 cell mL−1) were inoculated in both sets of cultures while 3.2 ± 0.3 × 105Pseudomonas were inoculated to the co-culture to mimic the relative abundance observed in the enrichment (Fig. 3a). Sph-2 increased to the similar amount (4.8–5.0 ± 0.8 × 107 cell mL−1) in both of the co-culture and the Sph-2 axenic culture, while the Pseudomonas sp. increased to 3.3 ± 0.7 × 107 cell mL−1 after 72-h incubation (Fig. 5d). These results indicate that even though Pseudomonas sp. consumed BPA degradation products from Sph-2, it did not affect the growth of Sph-2.