Galacturonate Metabolism in Anaerobic Chemostat Enrichment Cultures: Combined Fermentation and Acetogenesis by the Dominant sp. nov. “Candidatus Galacturonibacter soehngenii”

To study the composition and physiology of anaerobic microbial enrichment cultures on d-galacturonate, duplicate chemostat cultures were grown on a synthetic medium in which this pectin monomer was the sole carbon and energy substrate. Both bioreactors were started by adding anaerobic shake flask precultures on the same medium, which had been inoculated with rotting orange peel and cow rumen content. The bioreactors were operated as batch cultures until all galacturonate was consumed and then were switched to continuous operation at a dilution rate of 0.13 h⁻¹. This dilution rate and other parameters were chosen to enable direct comparison with previous experiments in glucose-limited chemostat cultures (8). Subsequently, product concentrations and carbon dioxide production were monitored over a period of 6 weeks. Constant rates of CO₂ production and of metabolite concentrations in the chemostats indicated that a “metabolic steady state” was reached after 2 weeks of continuous cultivation (approximately 63 generations of the overall microbial population) (Fig. 2). In the steady-state cultures, the d-galacturonate concentration remained below the detection limit of 0.1 mM, consistent with the medium composition, which was designed to make d-galacturonate the growth-limiting nutrient. After 1 month of steady-state operation (approximately 135 generations), five samples were taken over a period of 5 days for further physiological and metagenomic analysis.

To investigate galacturonate metabolism, fermentation products were analyzed in the culture supernatants and off-gas of the steady-state enrichment cultures. In both reactors, acetate was the main catabolic product of d-galacturonate fermentation, with additional production of H₂ and formate (Table 1). Average recoveries of carbon (92%) and electrons (95%) in the steady-state cultures indicated that the major fermentation products were accounted for (see Table S1 in the supplemental material).

In contrast to previously analyzed anaerobic chemostat enrichment experiments on glucose, which were grown under the same conditions (8) as used in the present study, the galacturonate-grown enrichment cultures did not produce detectable amounts of reduced fermentation products such as butyrate or ethanol (Table 1). Qualitatively, the formation of acetate, formate, CO₂, and H₂ as major fermentation products was consistent with previously described pathways for bacterial galacturonate metabolism (Fig. 1). However, in contrast to the product distribution expected from such pathways and to previous observations on glucose-grown chemostat enrichment cultures (8), the sum of the acetyl-CoA-derived products significantly differed from the combined formate and H₂ concentrations. In bioreactors 1 and 2, the combined yields of hydrogen and formate on d-galacturonate (mole mole⁻¹) were 37% ± 2% and 37% ± 1% lower, respectively, than the corresponding yields of acetate (Table 1). The observed metabolite profiles would be consistent with a contribution of acetogenesis via the Wood-Ljungdahl (WL) pathway, which enables conversion of fermentatively produced hydrogen (or alternatively, reduced ferredoxin derived from a pyruvate:ferredoxin oxidoreductase, EC 1.2.7.1) and carbon dioxide to acetate (equation 1). The combination of acetogenesis with fermentative conversion of d-galacturonate into acetate via a classical modified Entner-Doudoroff (ED) route (equation 2; Fig. 1) allows for conversion of d-galacturonate to acetate without net production of hydrogen (equation 3; Fig. 3).

Based on the assumption that d-galacturonate catabolism in the enrichment cultures indeed reflected simultaneous d-galacturonate fermentation via an ED pathway and acetogenesis via the WL pathway, the distribution of carbon and electrons over both pathways can be easily estimated from measured rates of d-galacturonate consumption and of hydrogen, formate, and acetate production in the steady-state cultures (three equations with two unknowns [28]). Based on these calculations (see calculations in the supplemental material), the average specific rates of acetate formation from pyruvate generated in the ED pathway (q_acetate,ED) and of acetogenesis via the WL pathway (q_acetate,WL) were estimated to be 6.9 ± 0.4 mmol (g biomass)⁻¹ h⁻¹ and 1.7 ± 0.2 mmol (g biomass)⁻¹ h⁻¹, respectively. In such a scenario, 16% ± 3% of the produced hydrogen and formate generated from d-galacturonate would have been used for acetogenesis (Fig. 3).

To study microbial population dynamics during enrichment and to assess if the hypothesis outlined above was supported by the presence of acetogenic microorganisms, 16S-rRNA gene amplicon sequencing was performed on samples from both bioreactors. The inoculum used to inoculate both bioreactors contained species related to the genera Clostridium sensu stricto and Lachnoclostridium as well as to uncultured Lachnospiraceae (Fig. 4). After complete consumption of d-galacturonate in the batch phase preceding the continuous culture, bioreactor 1 was dominated by bacteria related to Raoultella and Yersinia, with the latter showing a 99% identity to Yersinia massiliensis in the SILVA small subunit (SSU) database release 128. At the same sampling time, bioreactor 2 was dominated by bacteria related to Klebsiella and Enterobacter. Despite this difference in population composition, product profiles were similar for the two bioreactors at this time point, with acetate as the dominant organic catabolic product and formate as the second major product (Fig. 2).

Although the microbial community compositions were different at the start of the continuous enrichment, similar populations developed during prolonged continuous operation of the two bioreactors. In both reactors, a bacterium related to Lachnotalea became dominant, with a side population of a Klebsiella-related bacterium (Fig. 4). The steady-state communities in the two replicates were highly similar, indicating that the chosen cultivation conditions generated a selective advantage for these species and their conversions.

To further quantify the abundance of dominant community members during the steady-state measurements, fluorescence in situ hybridization (FISH) analysis (29, 30) was performed (Fig. 5). A general probe (EUB338) was used to stain all bacterial cells, while based on the results of the 16S rRNA gene analysis, a probe specific for the Lachnospiraceae family (Lac435) and a probe for the Enterobacteriaceae family (ENT) were used to quantify bacteria related to Lachnotalea and Klebsiella, respectively. Microorganisms belonging to the Lachnospiraceae were found to be abundant in both reactors, with another abundant population of bacteria belonging to the Enterobacteriaceae (Fig. 4 and 5). These observations correlated with the identification of a Lachnotalea- and a Klebsiella-related bacterium, respectively, in the 16S rRNA gene analysis.

To investigate whether the enriched Lachnotalea-related bacterium harbored genes encoding enzymes of the WL pathway and for DNA-based identification of the dominant microorganisms, a metagenomic analysis was performed on the steady-state cultures. DNA of the entire community was extracted from both steady-state enrichment cultures and sequenced. In total, 14.9 million paired-end reads remained posttrimming, and 7.9 Gbp of sequenced bases were obtained for each of the enrichments. After de novo assembly and metagenome binning-based sequence composition and differential coverage, draft genomes of the dominant microorganisms were assembled (31–35), checked for completeness and contamination, and annotated with the RAST server (36–39).

The genome with the highest coverage (577-fold ± 94-fold; completeness, 98%) in both enrichments was identified to be a member of the Lachnospiraceae by analysis of the small subunit rRNA gene of the constructed genome. An identity of 93% was found within the SILVA SSU database release 132 with the closest cultured relative, Lachnotalea glycerini, indicating that the dominant microorganism belongs to a novel genus within the Lachnospiraceae family (40, 41). We propose the name “Candidatus Galacturonibacter soehngenii” (Ga′.lac.tu.ro.ni.bac.ter. N.L. masc. n., galacturoni, referring to the utilization of galacturonate; bacter, the equivalent of Gr. neut. n. bactrum, a rod or staff, referring to the shape; soeh.ng.en′.ii., N.L. masc. n. gen., named after Nicolaas Söhngen, first Ph.D. student in Delft University of Technology and pioneer in anaerobic microbiology) for this novel microbe. The assembled and annotated genome of this organism harbored all structural genes for the enzymes of the canonic ED pathways for d-galacturonate catabolism, as well as multiple signature genes encoding enzymes of the WL pathway for acetogenesis, including fhs, which encodes formate-tetrahydrofolate ligase (EC 6.3.4.3) and is specific for the WL pathway (42–44) (Table 2). However, as shown in Table 2, the genes cooS, encoding carbon-monoxide dehydrogenase (EC 1.2.7.4), and acsBCD, which encodes the CO-methylating acetyl-CoA synthase complex (EC 2.1.3.169), were not identified.

A complete set of WL pathway genes were identified in the metagenomic data set within the genome of a species closely related to a Sporomusa species (99% identity found in SILVA SSU database release 132), within a genus known for harboring homoacetogenic bacteria (45). However, the genome of the Sporomusa species had a coverage of only 9-fold ± 2-fold and a completeness of 80%. This coverage corresponds to approximately 1% ± 0% of the total coverage. The other genome with a high coverage (166-fold ± 12-fold; completeness, 100%), was closely related to Klebsiella oxytoca (100% identity of the 16S rRNA gene in the SILVA SSU database release 132). This observation was consistent with the 16S rRNA gene amplicon sequencing data, which showed a side population of a bacterium closely related to Klebsiella species. The K. oxytoca genome did not harbor any WL pathway genes but, in accordance with literature (46), did contain all the genes for the adapted ED pathway for d-galacturonate metabolism.

A continuously stirred reactor of 1.2-liter capacity (0.5-liter working volume) was used (mechanical stirring, 300 rpm). Water was recirculated to maintain a constant temperature at 30°C. The reactor was sparged with nitrogen gas at a flow rate of 120 ml min⁻¹ to maintain anaerobic conditions. The pH was controlled at pH 8 ± 0.1 by automatic titration (ADI 1030 Bio controller) with a 1 M NaOH solution. The dilution rate was 0.13 h⁻¹, and the working volume was kept constant by peristaltic effluent pumps (Masterflex; Cole-Parmer, Vernon Hills, IL, USA) coupled to electrical level sensors.

To characterize the product spectrum, the reactor was run in continuous mode until stable product composition and biomass concentration were established. Steady state was reached after 2 weeks (63 generations). On-line analysis of system stability was achieved by online monitoring of the gas productivities, in the form of CO₂, and base addition. On-line gas detection of CO₂ was done with a Rosemount Analytical NGA 2000 MLT 1 multicomponent analyzer (infrared detector). Data acquisition was achieved with SCADA software (Sartorius BBI systems MFCS/win 2.1). When these rate measurements were stable or varied within a limited range (±10%) without a trend showing increase or decrease, a set of samples were taken during the subsequent cycles. The concentrations of soluble organic fermentation products, hydrogen produced, and biomass in the reactor volume were determined.

The reactor was inoculated (1%, vol/vol) with open mixed precultures started with a mixture of two different sources. The first inoculum was obtained from the content of rumen offal of a grass-fed cow (Est, Netherlands). The second inoculum was a sample from organic waste containing a large amount of citrus peels (Orgaworld Nederland B.V., Netherlands). Shake flasks were inoculated in duplicate with compost (2%, vol/vol) and rumen extract (2% vol/vol), and the initial pH was adjusted to 8 with a 2 M KOH solution. The shake flasks were cultivated in an anaerobic chamber (Bactron III; Shel Lab, Cornelius, USA) at 30°C with a gas composition of 89% N₂, 6% CO₂, and 5% H₂.

The cultivation medium contained the following (in grams liter⁻¹): d-galacturonate, 4.0; NH₄Cl, 1.34; KH₂PO₄, 0.78; Na₂SO₄·10H₂O, 0.130; MgCl₂·6H₂O, 0.120; FeSO₄·7H₂O, 0.0031; CaCl₂, 0.0006; H₃BO₄, 0.0001; Na₂MoO₄·2H₂O, 0.0001; ZnSO₄·7H₂O, 0.0032; CoCl₂·H₂O, 0.0006; CuCl₂·2H₂O, 0.0022; MnCl₂·4H₂O, 0.0025; NiCl₂·6H₂O, 0.0005; EDTA, 0.20. The d-galacturonate and mineral solutions were prepared and fed separately. The d-galacturonate solution was sterilized by filtration (0.2 μm Mediakap Plus; Spectrum Laboratories, Rancho Dominguez, CA, USA), and the mineral solution was autoclaved for 20 min at 121°C. Three milliliters PluronicPE 6100 antifoam (BASF, Ludwigshafen, Germany) was added per 40 liters mineral solution to avoid excessive foaming.

To determine substrate and extracellular metabolite concentrations, reactor sample supernatant was obtained by centrifugation of culture samples (Biofuge Pico; Heraeus, Hanau, Germany) and subsequent filtration (0.2 μm Millex-HV; Millipore-Merck, Darmstadt, Germany). Concentrations of galacturonate and extracellular metabolites were analyzed using an Agilent 1100 Affinity high-pressure liquid chromatography (HPLC) machine (Agilent Technologies, Amstelveen, Netherlands) with an Aminex HPX-87H ion-exchange column (Bio-Rad, Hercules, CA, USA) operated at 60°C with a mobile phase of 5 mM H₂SO₄ and a flow rate of 0.6 ml min⁻¹. Measurements of H₂ and CO₂ during steady-state analysis were performed off-line using gas bags (3.8 liters; Tedlar, Saint Gobain, France) and analyzed using a mass spectrometer (Prima BT MS; Thermo Scientific, USA).

Culture dry weight was measured by filtering 20 ml of culture broth over predried and preweighed membrane filters (0.2-μm Supor-200; Pall corporation, New York, NY, USA), which were then washed with demineralized water, dried in a microwave oven (Easytronic M591; Whirlpool, Benton Harbor, MI, USA), and weighed again. Carbon balances and electron balances were established based on the number of carbon atoms and electrons per mole, while a standard biomass composition of CH_1.8O_0.5N_0.2 was assumed (70).

The bacterial DNA for the community analysis over time was obtained by extracting 2 ml broth and pelleting the biomass before storing it at −80°C for further analysis. The DNA was extracted using the UltraClean DNA isolation kit (MOBIO laboratories Inc., USA) as per the manufacturer's instructions. The 16S rRNA genes of days 1 and 38 of bioreactor 1 and day 38 of bioreactor 2 were analyzed using amplicon sequencing with an Illumina HiSeq (Novogene Bioinformatics Technology Co., Ltd.); for this, the primers 341F (5′-CCTAYGGGRBGCASCAG-3′) and 805R (5′-GGACTACNNGGGTATCTAAT-3′) were used, generating 250-bp paired-end reads. Sequences were analyzed according to the method of van den Berg et al. (71). Representative sequences for each operational taxonomic unit were submitted for BLAST analysis under default conditions. The extracted genomic DNAs of the inoculum, from days 9 and 14 of bioreactor 1 and days 1, 9, and 14 of bioreactor 2, were subsequently used for a two-step PCR targeting the 16S rRNA gene of most bacteria and archaea. For this, the primers U515F (5′-GTGYCAGCMGCCGCGGTA-3′) and U1071R (5′-GARCTGRCGRCRRCCATGCA-3′) were used according to the method of Wang et al. (72). The first amplification was performed to enrich for 16S rRNA genes, using a quantitative PCR assay with 2× iQ SYBR green Supermix (Bio-Rad, CA, USA), 500 nM primers each, and 1 to 50 ng of genomic DNA was added per well to a final volume of 20 μl. The thermocycles consisted of a first denaturation at 95°C for 5 min, 20 cycles at 95°C for 30 s, 50°C for 40 s, and 72°C for 40 s, and a final extension of 72°C for 7 min. 454-adapters (Roche) and MID tags at the U515F primer were added to the produced PCR products. The second amplification was similar to the first one, with the exception of the use of Taq PCR master mix (Qiagen Inc., CA, USA); the program was run for 15 cycles, and the template (the product from step one) was diluted 10 times. After the second amplification, the PCR products were pooled equimolar and purified over an agarose gel using a GeneJET gel extraction kit (Thermo Fisher Scientific, Netherlands). The resulting library was sent for 454 sequencing and run in 1/8 lane with titanium chemistry by Macrogen Inc. (Seoul, South Korea). After analysis, the reads library was imported into CLC genomics workbench v7.5.1 (CLC Bio, Aarhus, DK), quality trimmed (limit, 0.05) to a minimum of 200 bp and an average of 284 bp, and demultiplexed. A build-it SILVA 123.1 SSURef Nr99 taxonomic database was used for BLASTn analysis on the reads under default conditions. The top result was imported into an Excel spreadsheet and used to determine taxonomic affiliation and species abundance.

FISH was performed as described in reference 73, using a hybridization buffer containing 25% (vol/vol) formamide. Probes were synthesized and 5′ labeled either with the Fluos dye or with one of the sulfoindocyanine dyes Cy3 and Cy5 (Thermo Hybaid Interaciva, Ulm, Germany) (Table 3). The general probe EUB338 labeled with Cy5 was used to indicate all eubacterial species in the sample. Slides were observed with an epifluorescence microscope (Axioplan 2; Zeiss, Sliedrecht, Netherlands), and images were acquired with a Zeiss MRM camera and compiled with the Zeiss microscopy image acquisition software (AxioVision version 4.7; Zeiss). The probes were subjected to excitation at 550 nm for Cy3, 494 for Fluos, and 649 nm for Cy5, and images were captured at emissions of 570 nm, 516 nm, and 670 nm, respectively.

A 250-ml volume of the culture broth was centrifuged for 10 min at 4,696 × g (Sorvall Legend X1R centrifuge; Thermo Fisher Scientific), the supernatant was discarded, and the pellet was washed with Tris-EDTA (TE; pH 8) buffer and stored at −20°C. DNA for library construction was extracted with the genomic DNA kit in combination with Genomic-tips 100/G (Qiagen Inc., CA, USA) according to the protocol, except for the addition of 2.6 mg ml⁻¹ Zymolyase (20T; Amsbio, UK) and 4 mg ml⁻¹ lysozyme. Cells were subsequently lysed by application of 2.9 × 10⁵-kPa pressure with a French press (Constant Systems Ltd., UK). After purification with the genomic tip kit, the starting amount of DNA was quantified with the Qubit double-stranded DNA (dsDNA) HS assay kit (Thermo Fisher Scientific) and the quality of DNA was assessed with NanoDrop 2000. Sequencing was performed in-house on an Illumina MiSeq Sequencer (Illumina, San Diego, CA, USA), with MiSeq Reagent kit v3 with 2 × 300-bp read length, and the DNA libraries were prepared using the Nextera XT DNA sample preparation kit (Illumina). Quality trimming, adapter removal, and contaminant filtering of paired-end sequencing reads were performed using BBDUK (BBTOOLS version 37.17; BBMap [B. Bushnell, https://sourceforge.net/projects/bbmap/, unpublished]). Trimmed reads for all samples were coassembled using metaSPAdes v3.10.1 (74) at default settings. MetaSPAdes iteratively assembled the metagenome using kmer sizes 21, 33, 55, 77, 99, and 127. Reads were mapped back to the metagenome for each sample separately using the Burrows-Wheeler Aligner 0.7.15 (BWA), employing the “mem” algorithm (75). The generated sequence mapping files were handled and converted as needed using SAMtools 2.1 (76). Metagenome binning was performed using five different binning algorithms: BinSanity v0.2.5.9 (32), COCACOLA (33), CONCOCT (34), MaxBin 2.0 2.2.3 (35), and MetaBAT 2 2.10.2 (31). The five bin sets were supplied to DAS Tool 1.0 (77) for consensus binning to obtain the final optimized bins. The quality of the generated bins was assessed using CheckM 1.0.7 (36). The bins with the highest coverage were annotated by the RAST server for further analysis (37). For the identification of the dominant individual bins, the complete 16S rRNA genes were submitted to the SILVA database, release 132 (41).

Genome sequence data of the amplicon sequences of the 16S rRNA gene analysis have been deposited at the NCBI GenBank (https://www.ncbi.nlm.nih.gov/GenBank/) under accession numbers MG982381 to MG982451. Genome sequence data of the metagenomic assembly of the galacturonate-fermenting enrichment culture metagenome, biosample SAMN08337232, have been deposited in the NCBI genomic database (www.ncbi.nlm.nih.gov/genome/) under BioProject ID PRJNA429181.