16S-rDNA-based analysis on high-strength liquor from grass acidification
Mechanically ground mixed grass (Graminidae) was acidified in three subsequent batch reactions under anaerobic conditions at mesophilic and thermophilic temperatures (Fig. 1). pH was automatically adjusted to 5.5 to prevent it dropping below that value. Acidification occurred in tap water as a result of microbial activity. The second and third batch received 5% Inoculum from the previous batch. Samples for VFA analysis were taken daily and every two days for 16S-rDNA amplicon sequencing. The mixed grass microbiome was analysed prior to entering acidification reactors, and it proved rich in Cyanobacteria- and Proteobacteria-related taxa. Upon transference into the reactors, the taxonomic profile rapidly switched to the one dominated by members of the phylum Firmicutes. This happened under both mesophilic and thermophilic conditions (Fig. 1).
After just two days, hardly any Proteobacteria and Cyanobacteria remained. As often occurs with 16S-rDNA-based analyses of plant material, cyanobacteria-related sequences may correlate to plant chloroplasts. On day four, most of the Firmicutes were suppressed by Bacteroidetes at mesophilic temperatures, while the proportion of Firmicutes remained high at 55 °C. The acidification process was repeated three times in a row and Bacteroidetes were also the dominating phylum at mesophilic temperatures. At thermophile temperatures the dominant phylum was Firmicutes, although at two of the sampling points a strong but transitory shift towards Proteobacteria was observed (Fig. 1a). In the second and third week an inoculum from the previous stages was used; however, this hardly influenced the taxonomic profile, which was constantly dominated by Bacteroidetes.
Upon termination of each acidification cycle, the high-strength liquors produced were transferred into bottles filled with nitrogen and stored at room temperature thereafter (Fig. 1b). The microbial composition in the stored liquor was analysed (Fig. 1, right) and yielded no significant changes at mesophilic temperature. However, a strong shift in the stored liquor originating from the thermophilic reaction was observed after incubation at room temperature (RT). After four days at RT, numbers of Bacteroidetes dramatically increased, yielding a stable taxonomic profile very similar to the one of the mesophilic acidification step. The microbial profile of the thermophilic samples upon RT storage was not accompanied by any changes in the concentration of chemical oxygen demand (COD) or VFA (Data not shown).
The results are in concordance with a previous work describing high titres of Bacteroidetes and Firmicutes during acidification of alginate under mesophilic conditions [24]. A microbiome dominated by Bacteroidetes and Firmicutes has also been reported for one-stage processes at mesophilic temperatures [9, 31, 32], but not for sewage sludge [7, 8].
There are no previous reports on the microbiome of acidification at thermophilic temperatures; however, a shift to Clostridia (Firmicutes) has been described for one-stage digesters [33, 34], similar to the increased titre of Firmicutes described in the present results.
Environmental parameters
Production of total volatile fatty acids (TVFAs) was more effective at mesophilic temperatures than at thermophilic ones (Fig. 2). With 200 mg TVFA per gram of input COD, the mesophilic stage yielded twice as many TVFAs as at thermophilic temperatures (Fig. 2a). At 37 °C, the relative amount of acetic acid and propionic acid were much higher than at 55 °C (Fig. 2b). By contrast, an accumulation of butyric acid was observed at thermophilic temperatures.
To the best of our knowledge, there are no previous reports comparing taxonomic profiles of mesophilic and thermophilic biogas acidification stages. There are reports, however, that thermophilic processes in one-stage digesters result in higher degradation efficiency compared to mesophilic ones [34,35,36,37]. Previous works have reported long incubation times for adaption of the biocoenosis to thermophile temperatures, ranging from several months [35] to up to one year [37]. Therefore, successful adaption to high temperatures and well-chosen seed sludge might be crucial for a separated acidification step.
In concordance with the reduced acid production in the thermophilic acidification, two of the corresponding methane stages (leach bed and semicontinuous batch with sewage) yielded more methane per gram of input COD with thermophilic liquor than with mesophilic. (Fig. 2c). However, in the system containing seed sludge from a co-digester (CD), the yield from the thermophilic-treated liquor was higher than in the one receiving mesophilic liquor. This might be related to the higher total solids (TS) content, high heterogeneity and high gasification activity also causing very high gas yields in the negative control from the CD sludge (Fig. 2b). In concordance with a previous work [38], the liquefied COD from the produced high-strength liquor was efficiently transformed into methane, indicating no inhibitory effects.
Usage of the high-strength liquor produced
High-strength liquor was stored in bottles at RT upon production, which were always flushed with nitrogen after opening to keep anoxic conditions. The liquor was semicontinuously fed into various methane stages (Figs. 1b, 2, 3).
The used seed sludge from the co-digester was very rich in Firmicutes, Synergistetes and Bacteroidetes, while the seed sludge from the sewage plant (SW) consisted mainly of Proteobacteria, Bacteroidetes, Spirochaeta and Chloroflexi (Matrix at day 7, Fig. 3). Both findings are in concordance with our previous report on several co-digester microbiomes [8]. The starting samples for the leach-bed systems (Matrix at day 7, Fig. 3) were taken 24 h after refilling the leach bed with sewage seed sludge. Compared to the original sewage, there was a dramatic decrease in Actinobacteria. This may be due to the high sensitivity of Actinobacteria to environmental changes, as sensitivity to environmental changes has been described for Actinobacteria in soil [39]. The two leach-bed systems were both rich in Chloroflexi, especially in the leach-bed biofilm (Fig. 3, Leach Bed). This is in concordance with other works describing high abundance of Chloroflexi in deep biofilm layers on building walls [40] and in the sediments of Winogradsky columns [41]. The input of the high-strength liquor, rich in Firmicutes and Bacteroidetes, did not result in an increase in those phyla in the sewage sludge batches or in the leach-bed systems (SW and Leach samples from Day 11 to Day 27, Fig. 3). Samples from both systems remained rich in Chloroflexi and Spirochaeta, even though they received a daily microbial input rich in Firmicutes and Bacteroidetes. This highlights the stability of the underlying biocoenosis and suggests the potential of separated acidification as an important step in preventing the occurrence of major microbial disturbances in the biocoenosis of the respective sewage digesters. For example, an additional thermophilic acidification stage could be included in co-digestion in sewage digesters in order to improve the robustness of the active microbiome. The positive effect of co-digesting organic matter with sewage sludge (e.g. food waste or energy grass) on the reactor performance has recently been reported [42, 43]. Moreover, the application of leachate in sewage digestion has been proposed too [44]. Our results indicate that using liquefied grass biomass (after separation from solids) might be a promising method for co-digestion with sewage. Large amounts of unused grass biomass, could still be valorised [29]. Although there have been attempts to add grass biomass into sewage sludge for co-digestion [45], co-digestion of liquefied grass biomass with sewage has not been demonstrated until now.
During the experiment, the lowered temperature in the storage bottle of the high-strength liquor at room temperature (Storage 3/Week 4, Fig. 1) resulted in a dramatically modified community composition of the thermophilic liquor after two days at RT. Thus, the transference of thermophilic high-strength liquor into a mesophilic sewage digester might destabilize the microbial community in the liquor and provide an advantage to the existing biocoenosis from the sewage digester. Using the high-strength liquor for co-digestion prevented the entry of solids into the water treatment circle.
In the last days of the experiment, the sludge samples from the co-digester exhibited low levels of Firmicutes and Bacteroidetes, and high rates of Actinobacteria. This coincided with a reduction in the production of biogas (Fig. 2). Interestingly, we have previously reported a concurrence between increasing amounts of Actinobacteria and low methane production [8]. Occurrence of Actinobacteria in anaerobic digester plants has been reported repeatedly [46, 47]. Actinobacteria were described in previous works as important key players in the degradation of plant material in compost [48] with effective enzymes that can allow large-scale application for breakdown of cellulosic plant material [49]. This is not necessarily contradictory with our results, since Actinobacteria survival and its efficiency for degradation of plant material could vary greatly at different nutrient concentrations due to their sensitivity to environmental changes as mentioned before. Although further work is needed to confirm this link, it is tempting to propose the identification and quantification of microbial key players as a marker of process efficiency.
In the case of the leach-bed system, the last part of the experiment was characterized by higher amounts of Bacteroidetes in the liquid phase (Leach samples from Day 11 to Day 27, Fig. 3). It has to be stressed that the biofilm became denser during the experiment and thus a biofilm filtering effect could be responsible for the very clear supernatant observed at the end of the process, which might, in turn, have affected the microbial composition of the leachate.
Archaea were also detected through 16S-rDNA amplicon sequencing and identification (Fig. 4). The genus Methanoculleus was the most abundant one in most of the samples. The co-digester sludge contained small amounts of Methanobacterium and Methanosarcina, as previously reported for the same plant [8] (Matrix at day 7, Fig. 4). However, upon transferring the sludge into the batch systems, a rapid shift was observed, in terms of an overwhelming abundance of Methanoculleus (CD, SW and Leach-Bed Samples at day 11, Fig. 4). This might be related to stress factors caused by the sludge transference (e.g. changing reactor conditions or short-time exposure to oxygen), and it could be hypothesized that Methanoculleus is more resistant to these changes. This is consistent with previous reports on the robustness of Methanoculleus, which is particularly resistant to inhibitors such as ammonia [50], phenol [51] or paraffin [52].
After eight days of incubation under constant conditions Methanosaeta and Methanobacterium started to recover in the batch reactions with the sewage seed sludge (Fig. 4), although no significant increases were observed for the leach-bed system. However, Methanosaeta proved frequent in the biofilm from the leach bed, (Fig. 4, Leach Bed). The occurrence of Methanosaeta in biofilms has been reported previously [53, 54]. This result highlights the need for a separate analysis of leach-bed samples and associated biofilms. In the co-digesters, Methanosarcina were also recovered as of day 23 (CD samples at day 23–27, Fig. 4).
Proteomic analysis on the high-strength liquor produced
Proteins were extracted from the samples d2, d4, d6 and d8 from the first cycle of acidification. The proteome at mesophilic and thermophilic temperatures proved strikingly different in the previous SDS-PAGE analysis (Additional file 1: Figure S3). This observation was approved by a principal component aggrupation (PCA) from mass spectroscopy raw data (peptide) analysis, where samples not only separated into two groups by temperature (X-axis, Fig. 5b), but also showed dynamic changes in time (Y-axis, Fig. 5b).
At the first stages, plant proteins were detected in the greatest amounts, as expected from the mixed grass biomass used in all assays. However, in line with increasing incubation time, the ratio between plants and bacteria shifted due to massive microbial growth and/or degradation of plant material at 37 °C (Fig. 5a). However at 55 °C, there was a constant plant:bacteria ratio in the protein abundance, indicating a decrease in the total microbial population.
An abundance of enzymes involved in carbohydrate metabolism and degradation in metaproteomes from both series of samples were identified using a protein database search for Bacteria and Archaea domains, although additionally diverse chaperones and heat-shock proteins (e.g. 10 and 60 kDa chaperonins, and GroEL) were overrepresented in the thermophilic samples (Additional file 2: Table S4). Among the most abundant proteins detected in all analysed samples, there was an almost complete set of glycolytic enzymes (glucose-6-phosphate isomerase, fructose-bisphosphate aldolase, triosephosphate isomerase, glyceraldehyde-3-phosphate dehydrogenase, phosphoglycerate kinase, enolase), as well as components of sugar transport systems (like the phosphotransferase system, PTS). These results indicate that the microbial population is basically engaged in the degradation and catabolism of sugars in the fermentative phase of short-chain acid production, an observation that is coherent with previous reports on the metaproteome [28] and metametabolome [55] of this kind of microbial communities.
Label-free quantitative proteome analysis was performed to determine differentially expressed proteins between mesophilic and thermophilic temperatures (Additional file 3: Table S5). A total of 1731 proteins were quantified from samples d2, d4, d6 and d8 collected from the first cycle of acidification: 556 proteins increased and 176 decreased between mesophilic and thermophilic conditions (37 vs. 55 °C). Samples were compared using the Limma statistics software package. Differences in protein abundances clearly separated samples into two clusters corresponding to culture temperature, with the subset of proteins showing an increased expression that was richer at 37 than 55 °C (Fig. 6a). On comparing protein abundances during sampling time, 120 (out of 1731) proteins showed differential expression at 37 °C, whereas at 55 °C, the differentially expressed proteins were only 5 (out of 1731) (Fig. 6b). Remarkably, most differences were observed when comparing d2 and d4 samples, and d2 and d8 at mesophilic conditions, whereas at thermophilic conditions a small set of differential proteins was only detected when comparing samples d2 and d8 (Fig. 6b). Among the differentially expressed proteins at mesophilic conditions there is a notable representation of ribosomal proteins indicating a dynamic state of these microbial communities. The taxonomic profiles of metaproteome samples were in agreement with the presented metagenomic data.
Among the differentially expressed proteins in d2 samples at 37 °C, noteworthy was the presence of several membrane transport systems from Firmicutes species involved in sugar uptake. These were the PTS HPr-related protein and the cellobiose-specific PTS IIB component, and the PTS phosphocarrier protein HP. There was also an increase in haemolysin-type calcium-binding protein, with a predicted hydrolytic activity on O-glycosyl compounds and a carbohydrate-binding domain (CBM) type 2 from an Alpha-proteobacterium in d2 samples. Previous studies on mesophilic biogas-producing, cellulolytic communities have indicated the abundance of sugar transporters and enzymes involved in polysaccharide degradation [9, 28, 56].