TY - JOUR
AU - Steele, James L
AB - Abstract Bacterial contamination of corn-based ethanol biorefineries can reduce their efficiency and hence increase their carbon footprint. To enhance our understanding of these bacterial contaminants, we temporally sampled four biorefineries in the Midwestern USA that suffered from chronic contamination and characterized their microbiomes using both 16S rRNA sequencing and shotgun metagenomics. These microbiotas were determined to be relatively simple, with 13 operational taxonomic units (OTUs) accounting for 90% of the bacterial population. They were dominated by Firmicutes (89%), with Lactobacillus comprising 80% of the OTUs from this phylum. Shotgun metagenomics confirmed our 16S rRNA data and allowed us to characterize bacterial succession at the species level, with the results of this analysis being that Lb. helveticus was the dominant contaminant in this fermentation. Taken together, these results provide insights into the microbiome of ethanol biorefineries and identifies a species likely to be commonly responsible for chronic contamination of these facilities. Electronic supplementary material The online version of this article (10.1007/s10295-019-02254-7) contains supplementary material, which is available to authorized users. Introduction Reducing the emissions of greenhouse gases (GHG) is one of the most important challenges facing mankind. The USA produced the second highest amount of GHG in 2015 and is among the highest in per capita GHG generation [36]. Increasing the use of corn-based bioethanol in the US transportation sector has the potential to reduce GHG emissions, as this process produces 43% less GHG emissions than gasoline on an energy equivalent basis [15]. Although significant improvement in the carbon footprint of bioethanol manufacturing has occurred in the last 15 years, continued improvements in the efficiency of ethanol production are still necessary to further reduce the environmental impact of this industry and to enhance its economic sustainability [15]. A significant source of inefficiency within corn-based ethanol biorefineries is the loss of yield due to bacterial contamination. A variety of microorganisms contaminate these facilities, but lactic acid bacteria (LAB), particularly lactobacilli, are thought to be the organisms of greatest concern [7, 8, 27]. Importantly, LAB compete with yeast, the typical biocatalyst in biofuel fermentations, for nutrients and produce organic acids, rather than the desired ethanol end product [1]. Chronic contamination with LAB can result in the conversion of 1% of the fermentable carbohydrates into organic acids, thereby reducing production efficiencies and resulting in a loss of more than 1 million USD per year per plant (personal communication). When acute bacterial contamination occurs, the LAB-derived organic acids and other metabolites can inhibit the yeast, leading to “sluggish” or stuck fermentations [32, 41]. These events can cause ethanol biorefineries to shut down for cleaning and disinfection or kickstart fermentations by adding additional yeast or nutrients, all of which result in significant loss in efficiency and significant financial burden to biorefineries [1, 22, 28]. Microbial methods for controlling such contaminants include the use of antibiotics, primarily penicillin, and virginiamycin [31, 38]. However, the use of antibiotics can lead to the emergence of antibiotic-resistant contaminants [31] directly in the fermentation facility and indirectly through dried distillers grains with solubles, which are a by-product of ethanol production and widely used as animal feed [31]. This can potentially lead to antibiotic-resistance genes being transferred into agricultural systems, with possible downstream impacts in food production and other sectors. Therefore, highly effective antibiotic-free solutions for the control of bacterial contaminants in ethanol biorefineries are needed. The importance of bacterial contaminants in corn-based ethanol biorefineries has led to extensive investigations on the microbiota of these facilities, mainly through the use of culture-dependent methods [1–3, 11, 28]. In general, these studies have concluded that this bacterial community is dominated by Lactobacillus species, with Lb. delbrueckii, Lb. plantarum, Lb. casei, Lb. mucosae, and Lb. fermentum commonly being observed. However, the development of culture-independent methods to assess microbiota composition in a diversity of environments has clearly demonstrated that culture-dependent techniques are unable to accurately reflect the composition of the microbiota [12, 27, 29]. For example, Li et al. [27] used a culture-independent approach by sequencing the 16S rRNA gene to examine the microbial community of corn-based ethanol biorefineries. They found that these microbiotas were dominated by bacteria from the phyla Firmicutes and Proteobacteria. In addition, two of the five facilities they investigated were dominated by LAB, whereas the other three had a greater abundance of Proteobacteria, primarily members of the Pseudomonas and Escherichia-Shigella [27]. While these culture-independent methods provided substantial insights into the composition of the resident microbiota found in ethanol biorefineries, use of 16S rRNA sequencing is incapable of resolving bacterial classifications beyond the genus level [27]. To address this, and further provide insights into the temporal dynamics of the ethanol biorefinery microbiome, we used both 16S rRNA and shotgun metagenomics sequencing to characterize the microbial communities in four corn-based ethanol biorefineries in the Midwest region of the USA. The combination of these two sequencing approaches allows for the resolution of bacterial classifications to the species levels, in addition to functional knowledge concerning the gene content of the organisms present [17, 21, 46]. Materials and methods Sampling procedure Three visits per biorefinery were made over a 6-months period during 2016 to four corn-based ethanol biorefinery plants located in the Midwest of the USA, named EP1, EP2, EP3, and EP4 to obtain the primary 77 samples for this study (Table S1). Additional samples were obtained in 2018 from EP1 to evaluate the stability of the microbiota of this biorefinery. These facilities continuously produce corn kernel slurry, conduct batch fermentations, and have continuous distillation. Samples were drawn from three different stages of the bioethanol process: (1) cooled mash (CM); (2) fermentation vessels at different times during fermentation (Ferm); and (3) the beer well (BW), immediately placed on ice, transported to the University of Wisconsin–Madison, and then stored at − 20 °C. Bacterial isolation and enumeration Bacterial enumeration was conducted on De Man, Rogosa, and Sharpe (MRS) agar plus 0.001% cycloheximide (Difco, Franklin Lakes, NJ). The plates were incubated at 33 °C for 24–48 h and colonies were enumerated. Two hundred individual colonies were isolated from cooled mash (80), fermentation (90), and beer well (30) samples, randomly selected among the four ethanol biorefineries, and subjected to full-length 16S rRNA sequencing. Isolation of genomic DNA and sequencing Genomic DNA from environmental samples and bacterial isolates was extracted utilizing the DNeasy Blood and Tissue kit (QIAGEN, Hilden, Germany) following the manufacturer’s instructions. DNA concentration was determined fluorometrically using the Qubit® dsDNA HS Assay Kit (ThermoFisher Scientific, Waltham, MA, USA). The full-length 16S rRNA genes were amplified using 16S universal primers (ACTCCTACGGCAGGCAGCAGT/AGGCCCGGGAACGTATTCACCG), and the amplified fragments were purified using the PCR clean-up system (Pure Link® PCR Purification Kit, Thermo Fisher Scientific) and then sequenced by the University of Wisconsin–Madison Biotechnology Center. The V3–V4 region of the 16S rRNA gene was amplified by PCR utilizing the primer set UNI340F (CCTACGGGRBGCASCAG) and UNI806R (GGACTACNNGGGTATCTAAT) [44]. The PCR products were cleaned using a 0.7x volume of AxyPrep Mag PCR clean-up beads (Axygen Biosciences, Union City, CA) and submitted to the Biotechnology Center at the University of Wisconsin–Madison for library preparation, sequenced on an Illumina MiSeq using an Illumina MiSeq Reagent Kit v3 (600 cycle) sequencing kit. Raw sequence reads were deposited into the NCBI’s Short Read Archive and are publicly available under Bioproject Accession SUB5621668. 16S rRNA data processing and analysis Analysis of the V3–V4 16S rRNA sequences was conducted using mothur v.1.40.0 [43] following the MiSeq procedure essentially as described by Kozich et al. [23]. Chimeras were removed with chimera.unchime [14] and sequences not relevant to this study (Eukaryote, Archaea, Chloroplasts, and Cyanobacteria) were excluded from the final data set. Good’s coverage was calculated by sample. All sequences were then grouped into 97% similarity operational taxonomic units (OTUs) by uncorrelated pairwise distances and average neighbor clustering in mothur. Representative sequences from each OTU were then aligned to the SILVA v132 ribosomal RNA gene database [39] and classified with a bootstrap value cutoff of 80. Bacterial communities were normalized to the sample with the lowest number of sequences (5240) and only those sequences with a relative abundance > 0.1% were used in our analysis. Alpha and beta-diversity were assessed in the sample [9, 30], as calculated in R v3.4.3 [40]. Normality was verified with Shapiro–Wilk normality test and differences in community diversity and richness were evaluated by non-parametric Kruskal–Wallis and Wilcoxon rank test for non-normally distributed metrics. All tests were assessed at a significance p < 0.05. Beta-diversity was visualized by nonmetric multidimensional scaling (nMDS) plots of the Bray–Curtis dissimilarity metric calculated with square root transformed data in R (vegan package) [35]. Differences in the spread of Bray–Curtis values within groups were calculated using a permutation tests of multivariate homogeneity of group dispersions (PERMUTEST) and an analysis of similarities (ANOSIM) in vegan [35]. Communities were compared by ethanol biorefinery, processing steps, and fermentation timepoints. OTUs contributing to ethanol biorefineries differences observed in ANOSIM were identified by analysis of similarity percentages (SIMPER) in vegan [35] and their significance across ethanol plant groups was tested using Kruskal–Wallis [40]. OTUs contributing at least 1% of the variation up to a cumulative total of 70% in at least one pairwise comparison were considered significant. All tests were assessed at significance p < 0.05, and strong correlations were defined as either > 0.7 or < − 0.7. Shotgun metagenomic sequencing data and annotation Shotgun metagenomic sequencing was performed to identify the most abundant species in 7 samples from ethanol biorefinery 1. Paired-end sequence reads were generated on an Illumina HiSeq2500 with 2 × 125 bp paired-end reads (insert size > 250 bp). FASTQ sequence files were generated using bcl2fastq2 version 2.18 and initial quality assessment was based on data passing the Illumina chastity filtering [18–20]. The second quality assessment was based on the remaining reads using the FASTQC quality control tool version 0.11.5. High-quality reads were de novo assembled into scaffolds using a combination of metagenomic analysis toolkit MOCAT2 [24] and the metagenomic module of the SPAdes assembler [34]. In brief, raw reads were trimmed based on quality using the quality trimming module in MOCAT2 [24], and then assembled using metaSPAdes [34]. Scaffolds shorter than 150 bp were removed. Gene prediction, non-redundant gene catalog creation (clustering at 100% sequence identity, resulting in a catalog of 664,590 non-redundant genes), functional annotation, and functional abundance profiling were performed using the corresponding modules in the MOCAT2 framework [24]. The microbial community composition of the sample was calculated using MetaPhlAn2 [47] following species-specific marker genes. The resulting taxonomic profiles were reported in relative abundance. Raw shotgun metagenomic sequence reads were deposited into the NCBI’s Short Read Archive, and are publicly available under Bioproject Accession SUB5855114. Results Summary of sequencing results From the 77 samples obtained, bacterial 16S rRNA amplicons were prepared, sequenced, and filtered through mothur pipeline [43], yielding 1.2 million high quality sequences. Sequencing coverage was determined to be sufficient based on each sample having a Good’s coverage > 98% (Table S2). A total of 2592 Operational Taxonomic Units (OTUs) binned at 97% similarity were identified in this data set, with samples containing between 3 and 96 OTUs per sample after normalization (Table S2). OTU 0001, OTU 0003, and OTU 0005 accounted for 70% of all sequences, and the addition of the ten next most abundant OTUs increased this value to 90.5% (Table S3). Taxonomic resolution of the bacterial communities in ethanol biorefineries The vast majority of the OTUs from the ethanol biorefineries belonged to the Firmicutes (89%) and Proteobacteria (11%). On a per biorefinery basis EP1, EP2, EP3, and EP4 had 90, 94, 96, and 71% Firmicutes, respectively. The percentages of Firmicutes and Proteobacteria per sample are presented in Fig. S1. Within the Firmicutes, OTUs belonging to the genus Lactobacillus were dominant, making up 80% of all OTUs within this phylum, followed by Lactococcus (11%), Weissella (2.5%), and Clostridium (1.8%). On a per plant basis, the abundance of Lactobacillus ranged from 37 to 95% (Fig. 1). Proteobacteria that could be classified at the genus level belonged to three major genera (Table S4): Acetobacter (1.2%), Enterobacter (0.4%), and Gluconobacter (0.4%). The Proteobacteria that could not be classified at the genus level primarily belonged to the families Enterobacteriaceae and Acetobacteriacea. Fig. 1 Open in new tabDownload slide Microbial community at genus level in ethanol biorefineries. Each bar represents a sample from the specified ethanol plant. Only OTUs accounting for at least 0.1% of the population were plotted To add further depth to our understanding of the biorefinery microbiota, we conducted sub-genera analysis by assigning Lactobacillus and Pediococcus OTUs to their correspondent clade within the 24 resolved phylogenetic clades described by Zheng et al. [49]. Five clades of lactobacilli were identified, accounting for 75% of all OTUs, Lb. delbrueckii clade (55%), Lb. reuteri clade (15%), Lb. casei clade (1.4%), Lb. perolens clade (0.5%), and Lb. brevis clade (0.3%). The relative abundance of lactobacilli clades in each sample, organized by ethanol biorefinery plant, is shown in Fig. 2. Fig. 2 Open in new tabDownload slide Microbial community analysis at the sub-genera level. Lactobacillus were assigned to the 24 phylogenetic clades as described by Zheng et al. [49]. Each bar represents a sample from the specified ethanol plant. Only clades with OTUs accounting for at least 0.1% of the total sequence abundance were plotted Alpha diversity of bacterial communities in the ethanol biorefinery environment changes according to biorefinery and processing stage To evaluate the impact of biorefinery and processing step on the microbial composition of our samples, we calculated Shannon diversity and Chao richness (Table S5). Alpha-diversity metrics indicated that the diversity of bacterial communities within these samples is impacted by biorefinery (Fig. 3, Kruskal–Wallis p < 0.05, Table S5), which does not contribute to changes in Chao’s richness (Fig. S2, Table S5). EP2 and EP4 had the highest Shannon’s diversity, although the only significant difference in diversity was observed between EP1 and EP4. Fig. 3 Open in new tabDownload slide Shannon’s diversity within samples from different ethanol biorefineries. The microbial community diversity in ethanol biorefineries with different letters above the boxplots are significantly different (Wilcoxon rank test, p < 0.05) Shannon’s diversity and Chao’s richness were both impacted by the processing step from which the sample was taken (Kruskal–Wallis rank sum test, p < 0.05). Diversity was richest in the mash, decreased during fermentation, and decreased further in the beer well, indicating that bacterial community composition in the beginning of the corn fermentation process is distinct from later fermented samples (Fig. 4a, Wilcoxon p < 0.05, Table S5). Conversely, Chao richness between samples from cooled mash and fermentation was similar, but differed significantly from the beer well-bacterial communities (Fig. 4b, Wilcoxon p < 0.05, Table S5). Fig. 4 Open in new tabDownload slide Shannon’s Diversity (a) and Chaos Richness (b) of samples from different processing steps during ethanol production. Results are expressed as standard boxplots with medians, and outliers shown as dots. Different letters above boxplots indicate significant difference (Wilcoxon rank sum test, p < 0.05) The microbial community is plant specific and only 4 OTUs are the main contributors to beta-diversity Bray–Curtis diversity analysis between different ethanol biorefineries indicated that the microbial community from EP4 is the most distinct of all the biorefineries investigated in this study (Fig. 5, ANOSIM, Table S6). EP1 and EP3 had similar microbiota composition, while EP2 was the most variable according to a multivariate analysis (PERMUTEST, Table S6). OTU1, OTU3, OTU5, and OTU6 corresponded to 79% of the variation that contributes to differences in beta-diversity among the biorefineries (Fig. 6, SIMPER, Table S4). These OTUs were classified as Lb. helveticus, Lactococcus sp., Lb. pontis, and Lb. delbrueckii, respectively (mothur v.1.40.0 [43]) Fig. 5 Open in new tabDownload slide Non-metric multidimensional scaling (nMDS) plots of the Bray–Curtis dissimilarity metric for bacterial communities of ethanol biorefineries. Each point represents a sample from the indicated ethanol plants (EP1–EP4) Fig. 6 Open in new tabDownload slide Main OTUs impacting beta-diversity of ethanol biorefineries (SIMPER) presented as standard boxplots with medians. Outliers are shown as dots and groups with same letters above the plots are not significantly different (Kruskal test, p > 0.05). EP1, EP2, EP3, and EP4: ethanol biorefinery 1–4 Temporal changes in bacterial composition of EP1 The stability of the microbiota in EP1 was examined by comparing the 16S rRNA results from fermentation and beer well samples collected in 2016 with those obtained in 2018 (Table S7) at similar fermentation timepoints. The Lb. delbrueckii clade was dominant in both 2016 and 2018 samples at all fermentation timepoints. On the other hand, the Lb. casei and Lb. reuteri clades were only observed in the 2018 samples (Table S7). Microbial succession of lactobacilli clades was observed during the ethanol process (Fig. 7). Six lactobacilli clades were observed in the 2018 ethanol plant samples. The Lb. delbrueckii clade was the most abundant member of the bacterial community in the fermentation and beer well samples. This clade increased in abundance during the process from 10%, 57%, 91%, 85%, 84%, and 89% of the population in the corn mash, fermentation 4 h, fermentation 22 h, fermentation 30 h, fermentation 47 h, and beer well samples, respectively. Similarly, the Lb. reuteri clade was present in 5 of the 6 samples analyzed, and increased in abundance from 2%, 5%, 10%, 14%, and 11% of the population in the corn mash, 22 h fermentation, 30 h fermentation, 47 h fermentation, and beer well samples, respectively. Lb. casei was the most abundant member of the bacterial community in the corn mash, and then rapidly decreased in abundance during the process, reaching a level below the 1% cutoff after the 4 h fermentation. Ped. acidilactici was present at 2% in the corn mash and below the 1% cutoff in all subsequent samples. Lb. buchneri and Lb. coryniformis were present at 6 and 4% of the population, respectively, in the 4 h fermentation sample and below the 1% cutoff in all subsequent samples. Fig. 7 Open in new tabDownload slide Microbial succession of lactobacilli clades during a single batch of corn fermentation from EP1, obtained in 2018. Results are shown as relative abundance (%) and phylogenetic clade classifications are assigned as described by Zheng et al. [49]. Only those OTUs corresponding to at least 1% of the total sequence abundance was used to generate the clusters Shotgun metagenomics allows for increased taxonomic resolution of bacterial communities to the species level To characterize the microbiome in greater detail, three samples from EP1 taken in 2018 were submitted for shotgun metagenomics analysis. This analysis allows for resolution of the species within the main clades identified by 16S rRNA sequencing during ethanol production (Fig. 8). At 4 h of fermentation only three species were present above the threshold (> 1% of the microbial population in the sample); these were Lb. casei from the Lb. casei clade, Lb. helveticus from the Lb. delbrueckii clade, and Lb. buchneri from the Lb. buchneri clade, which made up 49, 41, 10% of the population, respectively. At 30 h of fermentation, the Lb. delbrueckii clade made up 89% of the population and was comprised primarily of Lb. helveticus (96%) with relatively small quantities of Lb. amylovorus (2%) and Lb. delbrueckii (2%). In addition, Lb. reuteri (8%), Lb. casei (1%), and Streptococcus infantarius (2%) were also detected in the 30 h fermentation samples. The beer well sample contained organisms from only the Lb. delbrueckii and Lb. reuteri clades, which made up 88 and 11% of the population, respectively. The Lb. delbrueckii clade was made up of primary of Lb. helveticus (96%), with relatively small quantities of Lb. amylovorus (2%) and Lb. delbrueckii (2%). The Lb. reuteri clade was made up Lb. reuteri, Lb. fermentum, and Lb. mucosae at 45, 36, and 18%, respectively. Fig. 8 Open in new tabDownload slide Relative abundance of bacterial clades identified by 16S rRNA sequencing (left) and species within bacterial clades identified by shotgun metagenomics approach (right). Only OTUs corresponding to at least 1% of the total sequence abundance were plotted Dynamics of yeast and bacterial relative abundances during ethanol production The ratio of yeast to bacteria during ethanol production was also assessed using total number of reads from shotgun metagenomics normalized based upon the relative genome size of S. cerevisiae (12 Mbp) and a typical Lactobacillus (3 Mbp). Yeast was the dominant microorganism present, comprising 99.2, 99.2, 93.5, 94.6, 89.0, and 37.6% of the population in the corn mash, 4 h fermentation, 22 h fermentation, 30 h fermentation, 47 h fermentation, and the beer well samples, respectively. A corresponding increase in the bacterial population in these samples was observed, with bacteria accounting for 0.2, 0.2, 1.7, 1.4, 3.0, and 29.3% of the population in the corn mash, 4 h fermentation, 22 h fermentation, 30 h fermentation, 47 h fermentation, and the beer well samples, respectively. Culture-dependent method does not accurately represent microbial community of ethanol biorefineries Thirteen samples from EP2, EP3, and EP4 were selected for culture-dependent analysis and compared to the culture-independent data (Fig. 9). The comparison was conducted at either the lactobacilli clade level, using the clades suggested by Zheng et al. [50], or at the genus level. Our results showed significant discrepancies between the composition of the bacterial populations described by the culture-dependent and culture-independent approaches (Fig. 9). The most prominent difference between these two approaches is that the culture-independent method detected the Lactococcus sp. clade as comprising 42% of the population, while this organism was not detected using the culture-dependent approach. In addition, the culture-independent method identified Bacillus sp. and Clostridium sp. at 2% of this population, while these organisms were not found using the culture-dependent approach. In contrast the culture-dependent method detected significantly higher (> twofold) levels of Lb. plantarum, Shingomonas sp., Weissella sp., Lb. casei, Ped. acidilactici, Enterococcus sp., and Lb. brevis species clades in this population, relative to the culture-independent. Fig. 9 Open in new tabDownload slide Species clades and bacterial genera identified by culture-dependent (Sanger sequencing of full-length 16S rRNA) and culture-independent methods (amplicon 16S rRNA). Lactobacilli clades were assigned as described by Zheng et al. [49] Discussion The financial impact of bacterial contamination during corn-based bioethanol production has resulted in numerous studies on the microbial community of this environment, with the vast majority of these employing culture-dependent approaches [1, 10, 11, 45]. Although these studies have made significant contributions to our understanding of the ethanol biorefinery microbiome, advances in culture-independent methods provide an opportunity to characterize these communities more broadly and in greater detail [37, 48]. This study utilized both 16S rRNA sequencing and shotgun metagenomics to characterize the bacterial contaminants in four ethanol biorefineries in the Midwest region of the USA. Ethanol biorefineries employ a number of different measures to control for bacterial contaminants. In a typical corn-based ethanol biorefinery, the corn kernel is ground using a hammer mill, hydrated with recycled water to form a slurry, heated at 85 °C for 3 h to gelatinize the starch in the presence of a heat-stable α-amylase, which liquefies the starch and produces oligosaccharides. This step also sterilizes the substrate. This mixture is subsequently cooled to 33 °C (referred to as cooled mash) and transferred to a sanitized fermentation vessel, where the yeast, nutrients, a glucoamylase, and antimicrobials are added. The EPs characterized in this study utilized a variety of different antimicrobial interventions prophylactically to reduce the impact of bacterial contamination on ethanol yield. The added glucoamylase converts the oligosaccharides to glucose which the yeast converts to ethanol and CO2 [13, 27]. A variety of antimicrobials are used, including stabilized chlorine dioxide, hop acids, or antibiotics, with antibiotics being the most commonly used antimicrobial [26, 31, 38, 42]. In addition, yeast metabolism reduces the pH and increases the ethanol concentration, both of which can limit bacterial growth. These antimicrobial interventions are additive, and their combined impacts determine which organisms persist and grow within these facilities. Overall, the microbiota of corn-based ethanol biorefineries examined in this study were not very complex, relative to other fermentation processes such as wine, cheese, and American coolship ale [4–6], with 13 total OTUs accounting for 90% of the bacterial population in these facilities. Our findings are similar to those reported by Li et al. [27], which employed 16S rRNA sequencing and also observed that 13 OTUs accounting for 90% of the bacterial population in the five ethanol plants they studied. The relatively low complexity of the bacterial community in this niche is likely a reflection of the low microbial load in the fermentation substrate (cooled mash), the sanitation practices used in these facilities, the addition of antimicrobials, and the restrictive environment imposed by fermentation, in particular, the low pH and high ethanol titer. The microbiota of the four corn-based ethanol biorefineries was dominated by Firmicutes, although a significant level of Proteobacteria were also observed. Li et al. [27] also found high levels of Proteobacteria, in particular those in the genus Pseudomonas, which were the third most abundant OTU overall and among the most abundant bacteria in the microbiotas in three of the five ethanol plants they studied. In our study, we did not detect any Pseudomonas OTUs and the Proteobacteria detected in our samples were determined to be Enterobacteriaceae and Acetobacteraceae (Table S4) which could not be resolved below the family level. The presence of Acetobacteraceae is not surprising, as this family contains the acetic acid bacteria which are commonly found in alcoholic environments. Among the Firmicutes, 80% of the OTUs classified to the genus Lactobacillus, which correspond to previously known contaminants of ethanol biorefineries. The predominance of Lactobacillus in our data is in agreement with both other culture-dependent and culture-independent investigations of bioethanol microbiomes [7, 25, 27, 28, 33]. Our findings also indicate that the diversity of the microbiota in ethanol plants was impacted by both biorefinery and processing steps (Figs. 3, 4), although these factors did not significantly alter richness. The diversity in EP1 was the lowest among the biorefineries we studied, and was significantly different from EP4, which also had the highest inter-sample variation (beta-diversity, Fig. 5); this is explained by the abundance of Proteobacteria in this facility. Multivariable analysis demonstrated that 4 OTUs, namely Lb. helveticus, Lactococcus sp., Lb. pontis, and Lb. delbrueckii were responsible for 79% of the variation that contributed to differences in beta-diversity among the biorefineries (Fig. 6, Table S4). We found that diversity and richness also varied in different parts of the processing plant, likely in response to the local environment and opportunity for contamination [16]. The diversity was highest in the cooled mash, decreased in the fermentation samples, and then decreased further in the beer well (Fig. 4). The high diversity in cooled mash is likely due to the contact of liquefied commercially sterile corn mash with heat exchangers used to cool this material, on which microbial biofilms are commonly found [41]. The diversity declines in the fermentation and beer well samples, likely due to the microbial hurdles present during fermentation, which includes antibiotics, low pH (< pH 4.0), and high ethanol (up to 19% v/v) [16]. The combination of these constraints likely reduced the diversity, even though additional contaminants may be introduced during fermentation due to biofilms in the fermentation vessel and the cooling heat exchangers. It was surprising that the differences in diversity between fermentation and beer well samples were not significant (p > 0.05, Table S5). This might be explained by the low incidence of cleaning and sanitizing of the beer well tank (approximately once every 6 months), while the fermentation vessels are cleaned and sanitized after each fermentation batch. We also found that three of our four ethanol plants had relatively stable microbiotas across broad timescales (6 months–2 years). For EP2, which was not stable, relatively to the other four ethanol plants, samples were drawn monthly for 6 months and the microbiota was dominated by Lactococcus during the first three months, with this organism not being detected in the last three months (data not shown). These last three months had highly variable microbiota dominated by unclassified Streptococcus/unclassified Lactobacillus, Lb. delbrueckii clade/unclassified Weissella, and Lb. delbrueckii clade/Lb. reuteri clade, respectively (data not shown). With the exception of EP2, we posit that, in general, the control measures implemented by ethanol biorefineries and the environmental constraints imposed by the ethanol fermentation process result in a fairly stable microbiota in these facilities. For example, EP1 had a very stable microbiota over the 28 months this plant was monitored, with the microbiota being dominated by organisms from the Lb. delbrueckii clade (Table S7). A direct consequence of our 16S rRNA analysis is the identification of potential contaminants within the ethanol biorefineries we studied, which primarily belong to the LAB. Our classification of OTUs within the genus Lactobacillus, which is known to contain approximately 230 species within 24 phylogenic clades [49], identified five lactobacilli clades that accounted for 75% of all OTUs in our samples. Using these five clades, we were able to track lactobacilli succession during a single fermentation in EP1 (Fig. 7). The results of this analysis suggest that the diversity of LAB decreases with fermentation time with members of the Lb. delbrueckii and Lb. reuteri clades dominating the microbiota by the end of fermentation (Fig. 7), which is a clear indication of their fitness for this environment. To better resolve the identified lactobacilli contaminants in EP1, we utilized shotgun metagenomics, which not only can identify bacteria to the species and strain level, but do not suffer from PCR bias inherent in 16S rRNA sequencing, providing a more accurate representation of species abundance. By applying deep shotgun metagenomic sequencing, we were also able to sequence the yeast present in the fermentation process. We found that the percentage of bacteria in the microbiota during the fermentation increased from 0.8% at 4 h to 11.0% at 47 h, indicating that the bacteria increased in abundance, relative to yeast, as the fermentation progressed. In addition, bacteria comprised 37.6% of the microbiota in the beer well sample, likely reflecting the ability of bacteria to persist in this vessel, which is rarely cleaned and sanitized. From our shotgun metagenomic data, we identified eight lactobacilli species that accounted for more than 80% of the bacterial population (Fig. 8). Importantly, Lb. helveticus was determined to be the species that dominated the later stages of fermentation and made up the vast majority of the organisms from the Lb. delbrueckii clade in this fermentation. Moreover, the predominance of Lb. helveticus correlates to the increase in the bacterial to yeast ratio observed as the fermentation proceeded. Interestingly, our culturing efforts did not result in the isolation of any Lb. helveticus (Fig. 9), which reinforces the importance of using culture-independent techniques. Conclusions The 16S rRNA sequencing and shotgun metagenomics approach utilized in this study has allowed for the characterization of the contaminants present in ethanol biorefineries and how this microbiota changes during fermentation. The identification of dominant yet non-culturable LAB contaminants is an important step towards the development of methods for their control and mining of these metagenomic sequences may provide insights into their physiology. Future efforts should focus on obtaining isolates of these important bioethanol contaminants. Acknowledgements This research was supported by grants from Wisconsin Alumni Research Foundation (WARF) and by Lallemand Inc. We acknowledge the ethanol biorefineries for providing samples, Professor Garret Suen and his team for helping with data analysis, and the analytical team from Mascoma, LLC for all their support during this project. Fernanda Firmino gratefully acknowledges the scholarship from CAPES (Brazil) to pursue her postgraduate studies. References 1. Beckner M , Ivey ML, Phister TG Microbial contamination of fuel ethanol fermentations Lett Appl Microbiol 2011 53 387 394 10.1111/j.1472-765X.2011.03124.x Google Scholar Crossref Search ADS PubMed WorldCat 2. Bischoff KM , Skinner-Nemec KA, Leathers TD Antimicrobial susceptibility of Lactobacillus species isolated from commercial ethanol plants J Ind Microbiol Biotechnol 2007 34 11 739 744 10.1007/s10295-007-0250-4 Google Scholar Crossref Search ADS PubMed WorldCat 3. Bischoff KM , Liu S, Leathers TD, Worthington RE, Rich JO Modeling bacterial contamination of fuel ethanol fermentation Biotechnol Bioeng 2009 103 117 122 10.1002/bit.22244 Google Scholar Crossref Search ADS PubMed WorldCat 4. Bokulich NA , Mills DA Next-generation approaches to the microbial ecology of food fermentations BMB Rep 2012 45 377 389 10.5483/BMBRep.2012.45.7.148 Google Scholar Crossref Search ADS PubMed WorldCat 5. Bokulich NA , Bamforth CW, Mills DA Brewhouse-resident microbiota are responsible for multi-stage fermentation of American coolship ale PLoS One 2012 7 e35507 10.1371/journal.pone.0035507 3329477 Google Scholar Crossref Search ADS PubMed WorldCat 6. Bokulich NA , Joseph CML, Allen G, Benson AK, Mills DA Correction: Next-generation sequencing reveals significant bacterial diversity of botrytized wine PLoS One 2013 10.1371/annotation/4d347090-bca1-4184-a5d5-bc2d38baec7d 3797221 Google Scholar OpenURL Placeholder Text WorldCat Crossref 7. Bonatelli ML , Quecine MC, Silva MS, Labate CA Characterization of the contaminant bacterial communities in sugarcane first-generation industrial ethanol production FEMS Microbiol Lett 2017 10.1093/femsle/fnx159 Google Scholar OpenURL Placeholder Text WorldCat Crossref 8. Carvalho-Netto OV , Carazzolle MF, Mofatto LS, Teixeira PJPL, Noronha MF, Calderón LAL, Mieczkowski PA, Argueso LL, Pereira GAG Saccharomyces cerevisiae transcriptional reprograming due to bacterial contamination during industrial scale bioethanol production Microb Cell Fact 2015 14 1 13 10.1186/s12934-015-0196-6 Google Scholar Crossref Search ADS PubMed WorldCat 9. Chao A Nonparametric estimation of the number of classes in a population Scand J Stat 1984 11 265 270 Google Scholar OpenURL Placeholder Text WorldCat 10. Cherubin RA (2003) Efeitos da viabilidade da levedura e da contaminação bacteriana na fermentação alcoólica. Universidade de São Paulo 11. Daniel JS , Dowe N, Ibsen KN, Riley CJ, Ruth MF, Lumpkin RE Contaminant occurrence, identification and control in a pilot-scale corn fiber to ethanol conversion process Bioresour Technol 2007 98 2942 2948 10.1016/j.biortech.2006.10.002 Google Scholar PubMed OpenURL Placeholder Text WorldCat 12. De Filippis F , Parente E, Ercolini D Metagenomics insights into food fermentations Microb Biotechnol 2017 10 91 102 10.1111/1751-7915.12421 Google Scholar Crossref Search ADS PubMed WorldCat 13. Devantier R , Pedersen S, Olsson L Characterization of very high gravity ethanol fermentation of corn mash. Effect of glucoamylase dosage, pre-saccharification and yeast strain Appl Microbiol Biotechnol 2005 68 622 629 10.1007/s00253-005-1902-9 Google Scholar Crossref Search ADS PubMed WorldCat 14. Edgar RC , Haas BJ, Clemente JC, Quince C, Knight R UCHIME improves sensitivity and speed of chimera detection Bioinformatics 2011 27 2194 2200 10.1093/bioinformatics/btr381 21700674 Google Scholar Crossref Search ADS PubMed WorldCat 15. Flugge MJ, Lewandrowski J, Rosenfeld J, Boland C, Hendrickson T, Jaglo K, Kolansky S, Moffroid K, Riley-Gilbert M, Pape D (2017) A life-cycle analysis of the greenhouse gas emissions of corn-based ethanol. Report prepared by ICF under USDA Contract No. AG-34142-D-16-0243 16. Gibson BR , Lawrence SJ, Leclaire JPR, Powell CD, Smart KA Yeast responses to stresses associated with industrial brewery handling FEMS Microbiol Rev 2007 31 535 569 10.1111/j.1574-6976.2007.00076.x Google Scholar Crossref Search ADS PubMed WorldCat 17. Hamady M , Knight R Microbial community profiling for human microbiome projects: Tools, techniques, and challenges Genome Res 2009 19 1141 1152 10.1101/gr.085464.108 3776646 Google Scholar Crossref Search ADS PubMed WorldCat 18. Illumina (2017) bcl2fastq2 Conversion v2.19. In: Illumina, User Guid. 15051736, v2. https://usermanual.wiki/Document/bcl2fastq2guide15051736v2.164202932/view. Accessed 10 Feb 2019 19. Illumina I (2017) Calculating percent passing filter for patterned and nonpatterned flow cells. In: Illumina, Tech. Note Informatics. chrome-extension://oemmndcbldboiebfnladdacbdfmadadm/https://www.illumina.com/content/dam/illumina-marketing/documents/products/technotes/hiseq-x-percent-pf-technical-note-770-2014-043.pdf. Accessed 10 Feb 2019 20. Illumina (2018) HiSeq X System Guide. In: Illumina, Hiseq x Syst. Guid. (15050091), v07. http://www.illumina.com/systems/hiseq-x-sequencing-system/performance-specifications.html. Accessed 10 Feb 2019 21. Kchouk M , Gibrat JF, Elloumi M Generations of sequencing technologies: From first to next generation Biol Med 2017 09 8 10.4172/0974-8369.1000395 Google Scholar Crossref Search ADS WorldCat 22. Khatibi PA , Roach DR, Donovan DM, Hughes SR, Bischoff KM Saccharomyces cerevisiae expressing bacteriophage endolysins reduce Lactobacillus contamination during fermentation Biotechnol Biofuels 2014 7 1 13 10.1186/1754-6834-7-104 Google Scholar Crossref Search ADS PubMed WorldCat 23. Kozich JJ , Westcott SL, Baxter NT, Highlander SK, Schloss PD Development of a dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence data on the miseq illumina sequencing platform Appl Environ Microbiol 2013 79 5112 5120 10.1128/AEM.01043-13 3753973 Google Scholar Crossref Search ADS PubMed WorldCat 24. Kultima JR , Coelho LP, Forslund K, Huerta-Cepas J, Li SS, Driessen M, Voigt AY, Zeller G, Sunagawa S, Bork P MOCAT2: a metagenomic assembly, annotation and profiling framework Bioinformatics 2016 32 2520 2523 10.1093/bioinformatics/btw183 4978931 Google Scholar Crossref Search ADS PubMed WorldCat 25. Leathers TD , Bischoff KM, Rich JO, Price NPJ, Manitchotpisit P, Nunnally MS, Anderson AM Inhibitors of biofilm formation by biofuel fermentation contaminants Bioresour Technol 2014 169 45 51 10.1016/j.biortech.2014.06.065 Google Scholar Crossref Search ADS PubMed WorldCat 26. Leite IR , Faria JR, Marquez LDS, Reis MHM, De Resende MM, Ribeiro EJ, Cardoso VL Evaluation of hop extract as a natural antibacterial agent in contaminated fuel ethanol fermentations Fuel Process Technol 2013 106 611 618 10.1016/j.fuproc.2012.09.050 Google Scholar Crossref Search ADS WorldCat 27. Li Q , Heist EP, Moe LA Bacterial community structure and dynamics during corn-based bioethanol fermentation Microb Ecol 2016 71 409 421 10.1007/s00248-015-0673-9 Google Scholar Crossref Search ADS PubMed WorldCat 28. Lucena BTL , dos Santos BM, Moreira JL, Moreira APB, Nunes AC, Azevedo V, Miyoshi A, Thompson FL, de Morais M Diversity of lactic acid bacteria of the bioethanol process BMC Microbiol 2010 10.1186/1471-2180-10-298 2999616 Google Scholar OpenURL Placeholder Text WorldCat Crossref 29. Montel MC , Buchin S, Mallet A, Delbes-Paus C, Vuitton DA, Desmasures N, Berthier F Traditional cheeses: rich and diverse microbiota with associated benefits Int J Food Microbiol 2014 177 136 154 10.1016/j.ijfoodmicro.2014.02.019 Google Scholar Crossref Search ADS PubMed WorldCat 30. Morris EK , Caruso T, Buscot F, Fischer M, Hancock C, Maier TS, Meiners T, Müller C, Obermaier E, Prati D, Socher SA, Sonnemann I, Wäschke N, Wubet T, Wurst S, Rillig MC Choosing and using diversity indices: Insights for ecological applications from the German biodiversity exploratories Ecol Evol 2014 4 3514 3524 10.1002/ece3.1155 4224527 Google Scholar Crossref Search ADS PubMed WorldCat 31. Murphree CA , Heist EP, Moe LA Antibiotic resistance among cultured bacterial isolates from bioethanol fermentation facilities across the United States Curr Microbiol 2014 69 277 285 10.1007/s00284-014-0583-y Google Scholar Crossref Search ADS PubMed WorldCat 32. Narendranath NV , Thomas KC, Ingledew WM Effects of acetic acid and lactic acid on the growth of Saccharomyces cerevisiae in a minimal medium J Ind Microbiol Biotechnol 2001 26 171 177 10.1038/sj.jim.7000090 Google Scholar Crossref Search ADS PubMed WorldCat 33. Neto PDO , De Lima FA, Da Silva KC, Da Silva DF, Carvalho AFA, Dos Santos C Chemical inhibition of the contaminant Lactobacillus fermentum from distilleries producing fuel bioethanol Braz Arch Biol Technol 2014 57 441 447 10.1590/S1516-8913201401214 Google Scholar Crossref Search ADS WorldCat 34. Nurk Sergey , Meleshko Dmitry, Korobeynikov Anton, Pevzner Pavel A metaSPAdes: a new versatile metagenomic assembler Genome Research 2017 27 5 824 834 10.1101/gr.213959.116 Google Scholar Crossref Search ADS PubMed WorldCat 35. Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O’Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E, Wagner H (2018) Vegan: Community ecology package. https://cran.r-project.org/. Accessed 10 Feb 2019 36. Olivier JGJ, Peters JAHW (2017) Trends in global CO2 and total greenhouses gas emissions. https://www.pbl.nl/en/publications/trends-in-global-co2-and-total-greenhouse-gas-emissions-2018-report. Accessed 10 Feb 2019 37. Pandya S , Ravi K, Srinivas V, Jadhav S, Khan A, Arun A, Riley LW, Madhivanan P Comparison of culture-dependent and culture-independent molecular methods for characterization of vaginal microflora J Med Microbiol 2017 66 149 153 10.1099/jmm.0.000407 Google Scholar Crossref Search ADS PubMed WorldCat 38. Paulus Compart DM , Carlson AM, Crawford GI, Fink RC, Diez-Gonzalez F, Dicostanzo A, Shurson GC Presence and biological activity of antibiotics used in fuel ethanol and corn co-product production J Anim Sci 2013 91 2395 2404 10.2527/jas.2012-5714 Google Scholar Crossref Search ADS PubMed WorldCat 39. Quast C , Pruesse E, Yilmaz P, Gerken J, Schweer T, Yarza P, Peplies J, Glöckner FO The SILVA ribosomal RNA gene database project: Improved data processing and web-based tools Nucleic Acids Res 2013 41 590 596 10.1093/nar/gks1219 Google Scholar Crossref Search ADS WorldCat 40. R Core Team (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Accessed 10 Feb 2019 41. Rich JO , Leathers TD, Bischoff KM, Anderson AM, Nunnally MS Biofilm formation and ethanol inhibition by bacterial contaminants of biofuel fermentation Bioresour Technol 2015 196 347 354 10.1016/j.biortech.2015.07.071 Google Scholar Crossref Search ADS PubMed WorldCat 42. Rückle L , Senn T Hop acids can efficiently replace antibiotics in ethanol production Int Sugar J 2006 108 139 147 Google Scholar OpenURL Placeholder Text WorldCat 43. Schloss PD , Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, Lesniewski RA, Oakley BB, Parks DH, Robinson CJ, Sahl JW, Stres B, Thallinger GG, Van Horn DJ, Weber CF Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities Appl Environ Microbiol 2009 75 7537 7541 10.1128/AEM.01541-09 2786419 Google Scholar Crossref Search ADS PubMed WorldCat 44. Takai K , Koki H Rapid detection and quantification of members of the archael community by quantitative PCR using fluorogenic probes Appl Environ Microbiol 2000 66 5066 5072 10.1128/AEM.66.11.5066-5072.2000 92420 Google Scholar Crossref Search ADS PubMed WorldCat 45. Thomas KC , Hynes SH, Ingledew WM Effect of lactobacilli on yeast growth, viability and batch and semi-continuous alcoholic fermentation of corn mash Appl Microbiol 2001 90 819 828 10.1007/s10295-004-0159-0 Google Scholar Crossref Search ADS WorldCat 46. Tran B , Brown AMK, Bedard PL, Winquist E, Goss GD, Hotte SJ, Welch SA, Hirte HW, Zhang T, Stein LD, Ferretti V, Watt S, Jiao W, Ng K, Ghai S, Shaw P, Petrocelli T, Hudson TJ, Neel BG, Onetto N, Siu LL, McPherson JD, Kamel-Reid S, Dancey JE Feasibility of real time next generation sequencing of cancer genes linked to drug response: results from a clinical trial Int J Cancer 2013 132 1547 1555 10.1002/ijc.27817 Google Scholar Crossref Search ADS PubMed WorldCat 47. Truong DT , Franzosa EA, Tickle TL, Scholz M, Weingart G, Pasolli E, Tett A, Huttenhower C, Segata N MetaPhlAn2 for enhanced metagenomic taxonomic profiling Nat Methods 2015 12 902 903 10.1002/ijc.27817 Google Scholar Crossref Search ADS PubMed WorldCat 48. Yarza P , Yilmaz P, Pruesse E, Glöckner FO, Ludwig W, Schleifer KH, Whitman WB, Euzéby J, Amann R, Rosselló-Móra R Uniting the classification of cultured and uncultured bacteria and archaea using 16S rRNA gene sequences Nat Rev Microbiol 2014 12 635 645 10.1038/nrmicro3330 Google Scholar Crossref Search ADS PubMed WorldCat 49. Zheng J , Ruan L, Sun M, Gänzle M A genomic view of lactobacilli and pediococci demonstrates that phylogeny matches ecology and physiology Appl Environ Microbiol 2015 81 7233 7243 10.1128/AEM.02116-15 4579461 Google Scholar Crossref Search ADS PubMed WorldCat 50. Zheng J , Zhao X, Lin XB, Gänzle M Comparative genomics Lactobacillus reuteri from sourdough reveals adaptation of an intestinal symbiont to food fermentations Sci Rep 2015 5 18234 10.1038/srep18234 4995734 Google Scholar Crossref Search ADS PubMed WorldCat Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. © Society for Industrial Microbiology 2020 This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © Society for Industrial Microbiology 2020
TI - Characterization of microbial communities in ethanol biorefineries
JF - Journal of Industrial Microbiology and Biotechnology
DO - 10.1007/s10295-019-02254-7
DA - 2020-02-01
UR - https://www.deepdyve.com/lp/oxford-university-press/characterization-of-microbial-communities-in-ethanol-biorefineries-qxJM7P2RQM
SP - 183
EP - 195
VL - 47
IS - 2
DP - DeepDyve
ER -