A metagenome-level analysis of a microbial community fermenting ultra-filtered milk permeate

Abstract

Fermentative microbial communities have the potential to serve as biocatalysts for the conversion of low-value dairy coproducts into renewable chemicals, contributing to a more sustainable global economy. To develop predictive tools for the design and operation of industrially relevant strategies that utilize fermentative microbial communities, there is a need to determine the genomic features of community members that are characteristic to the accumulation of different products. To address this knowledge gap, we performed a 282-day bioreactor experiment with a microbial community that was fed ultra-filtered milk permeate, a low-value coproduct from the dairy industry. The bioreactor was inoculated with a microbial community from an acid-phase digester. A metagenomic analysis was used to assess microbial community dynamics, construct metagenome-assembled genomes (MAGs), and evaluate the potential for lactose utilization and fermentation product synthesis of community members represented by the assembled MAGs. This analysis led us to propose that, in this reactor, members of the Actinobacteriota phylum are important in the degradation of lactose, via the Leloir pathway and the bifid shunt, and the production of acetic, lactic, and succinic acids. In addition, members of the Firmicutes phylum contribute to the chain-elongation-mediated production of butyric, hexanoic, and octanoic acids, with different microbes using either lactose, ethanol, or lactic acid as the growth substrate. We conclude that genes encoding carbohydrate utilization pathways, and genes encoding lactic acid transport into the cell, electron confurcating lactate dehydrogenase, and its associated electron transfer flavoproteins, are genomic features whose presence in Firmicutes needs to be established to infer the growth substrate used for chain elongation.

Keywords: butyric; chain elongation; dairy coproduct; fermentation; hexanoic; lactic; microbiome; succinic.

FIGURE 1

(A) Extracellular concentrations of carbohydrates and fermentation products during bioreactor operation. The black vertical lines demarcate 5 periods (A-E) in which different profiles of extracellular fermentation products were present. The “soluble COD” line represents the experimentally determined soluble COD concentration in gCOD/L at each time point. The “total influent COD” line represents the average total COD concentration (gCOD/L) determined for the UFMP feedstock. See also Supplementary Table S5. (B) Relative abundance of 10 abundant MAGs (Table 2) during bioreactor operation. The black line represents the percentage of DNA sequences that mapped to the set of 217 non-redundant MAGs. The black vertical lines demarcate periods A-E in which different profiles of extracellular fermentation products were identified. See also Supplementary Table S6.

FIGURE 2

Phylogenetic tree of 10 abundant MAGs in the UFMP-fed bioreactor and related genomes. The tree was based on sequence comparisons of 120 concatenated bacterial single-copy marker genes; 1000 bootstraps were used with bootstrap values shown as a percentage. The scale bar represents evolutionary distance and indicates the number of nucleotide substitutions per sequence site. When available, metabolite production from a species reported in the literature was indicated with colored circles: green, lactic acid; magenta, ethanol; yellow, succinic acid; blue, acetic acid; purple, butyric acid (Wüst et al., 1995; Ramos et al., 1997; Hall et al., 2002; Wu and Yang, 2003; Zhu and Yang, 2004; Jumas-Bilak et al., 2007; Von Ah et al., 2007; Ritalahti et al., 2011; Dwidar et al., 2013; Wu et al., 2014; Li et al., 2015; Rosero et al., 2016; Bidzhieva et al., 2020; Eckel et al., 2020).

FIGURE 3

(A) Metabolic network of pathways predicted to be involved in the fermentation of lactose in the bioreactor. (B) Predicted presence of genes within the 10 abundant MAGs. Genes within Embden-Meyerhof-Parnas (EMP) glycolysis, lactic acid production, or ethanol production that are also shared with the bifid shunt and/or phosphoketolase pathway are depicted in the heatmap only once (i.e., omitted from the bifid shunt and/or phosphoketolase pathway sections), for ease of reading. Several variations for anaerobic succinic acid production exist (Song and Lee, 2006; Cao et al., 2013) and only those present within the abundant MAGs are represented here. See Supplementary Figure S1 for other succinic acid production pathways and predicted genes for each variation. Gene, pathway intermediate, and coenzyme abbreviations are provided in Supplementary Table S8.

FIGURE 4

Phylogenetic tree comparing electron transfer flavoprotein B (EtfB) amino acid sequences of ACID1, CLOS1, and LCO1 with other EtfB sequences. 500 bootstraps were used with bootstrap values shown as a percentage. Bootstrap values below 50 are excluded from the figure. The scale bar represents evolutionary distance and indicates the number of amino acid substitutions per sequence site. Locus numbers for etfB genes are provided. Three EtfB proteins with biochemically characterized enzymatic function were used as references and indicated with bold text; the protein with which they complex is indicated in parentheses (ACD, acyl-CoA dehydrogenase; ecLDH, electron confurcating lactate dehydrogenase) (Li et al., 2008; Chowdhury et al., 2014; Weghoff et al., 2015). The predicted protein sequence of Paracoccus denitrificans RS12675 was used as an outgroup etfb from a non-Firmicute isolate with no biochemical evidence for electron bifurcation (Roberts et al., 1999; Costas et al., 2017). EtfB sequences derived from isolated organisms are indicated with a black circle at the node. Gene neighborhood designations describe the presence of one or more relevant genes flanking or nearby the etfB genes with ecLDH/LacT (electron confurcating lactate dehydrogenase/lactate permease) relevant to lactic acid utilization and ACD/ACAT/HAD/ECoAH (acyl-CoA dehydrogenase/acetyl-CoA acetotransferase/3-hydroxyacyl-CoA dehydrogenase/enoyl-CoA hydratase) relevant to reverse β-oxidation. Brackets indicate proposed clusters of EtfBs that participate in either lactic acid utilization or reverse β-oxidation. The raw sequences used to generate this tree are available in Supplementary Table S4.

FIGURE 5

Model of metabolic networks in the microbial community established in the UFMP-fed bioreactor. Arrows indicate the direction of carbon flow from the lactose of the feedstock and through intermediates as they are transformed by microbial community members.