All fermentation experiments were carried out in duplicate at different time points (half a year period between replicates). This way, all different parameters (pH, temperature, presence of fungal biofilm, methanogenic activity) were tested twice with potentially different starting rumen microbial communities. Although differences were observed in the absolute values of the results obtained between the replicates, the same trends regarding SCFAs yields, productivities, and selectivities were observed from the two duplicate sets for all conditions under investigation. Analysis of variance (one-way ANOVA) was performed with the raw values of the replicates to reveal potential differences. Although duplicate experiments were carried out at different times of the year, no statistically significant differences between the replicates were observed on SCFA production (Table
1
). Similarly, no differences between the replicates were observed on the composition of the SCFA mixture.
Table 1 SCFA production using different amounts of rumen fluid volumes Effect of the rumen inoculum size on SCFA production
The dry matter of the rumen fluid varied from 3.5 to 4.5% w/v, while the SCFAs content ranged from 12 up to 20 g/L. To transfer as less dry matter and SCFAs as possible in the reactors and also to have an efficient rumen inoculum for the in vitro bioconversion of cellulose to SCFAs, fermentation experiments were performed using rumen fluid/fermentation volume ratios from 0.093 up to 0.372. Although small differences were observed on the production patterns of SCFAs as a function of time (Fig. S1), no statistically significant differences were observed (Table
1
) on the final SCFA yields and selectivities among the different ratios, apart from the butyric acid production between the ratios 0.186 and 0.372. With this being the only statistically significant difference, no trend could be observed regarding the production of total SCFAs, or that of butyric acid or any other acid, to justify the selection of a certain inoculum ratio.” Thus, the lower ratio tested (250 mL of rumen fluid in 2700 mL final volume, ratio 0.093) was applied in all later experiments with the rumen microbial community.
The presence of a cellulolytic aerobic fungus in MBM reactors enhances the cellulolytic activities and the SCFA yields
To prove the suitability, the usefulness and the efficiency of the MBM system for SCFA production using the rumen culture, we investigated whether a fungal aerobic biofilm would be beneficial for the overall process, assuming that a higher cellulolytic efficiency would be reflected into higher SCFA productivities and yields.
To confirm the importance of the cellulolytic enzyme activities for an efficient conversion of cellulose to SCFAs by the rumen microbial community in the MBM system, experiments were carried out only with the rumen microbiome (no fungal biofilm present) using glucose as the substrate. The results from these experiments are compared with the results on cellulose conversion in Fig.
1
. As shown, the rumen bacterial community converted glucose to SCFAs much more efficiently than cellulose, not only in terms of final yields but also in terms of productivities. It has been shown that neither crystallinity nor degree of polymerization is changing until enzymatic hydrolysis of Avicel reaches 50–60% conversion to sugars (Kafle et al. 2015). Taking also into account that Avicel is a quite homogeneous substrate and thus only a small increase to its recalcitrance during hydrolysis would be expected, it may be assumed that cellulose hydrolysis was a bottleneck of the process. This could be attributed to a progressive deactivation of the existing cellulolytic enzymes and a simultaneous modification of the composition of the rumen bacterial community, resulting in lower cellulolytic activities. More experiments were therefore designed to evaluate the ability of a fungal biofilm to enhance the cellulolytic activity during the conversion of cellulose to SCFAs by the rumen community. The temperature and pH optimum values for Trichoderma reesei and the rumen microbial community are very different and thus fermentations without T. reesei were carried out not only at pH 6 and T = 30 °C but also at 39 °C and pH 6.5 to eliminate any possible biases.
The results from fermentations carried out with or without a fungal (T. reesei) biofilm on the membrane of the reactor were compared. As shown in Fig.
2
, at 30 °C, the presence of fungal biofilm resulted in higher SCFAs yield and productivity. At these conditions, the MBM system with T. reesei biofilm reached 7.3 g/L of total SCFAs, corresponding to a 39% higher concentration than the rumen microbiome alone (5.1 g/L). T. reesei biofilm also resulted in faster increase in SCFA concentration in the broth. Seventy-five percent of the maximum SCFA concentration had been reached after 168 h of cultivation, while in the case of the rumen microbiome alone (at 30 °C), SCFAs production was very low during the first 290 h of fermentation. When the membrane of the MBM system was not flushed with air but with N2, the lag phase was not observed. Therefore, it could be assumed that the lag phase was due to not strict anaerobic conditions in the reactor, during the early stages of the fermentation. This was supported by the redox profiles between these two cases (Fig. S2). However, with N2 flushing through the membrane, the SCFA yield from the rumen microbiome alone was even lower than in the case of air flushing.
Among the fermentations performed without a fungal biofilm, the more productive was at 39 °C and pH 6.5, conditions which are identical to the rumen natural conditions (Weimer et al. 2009) (Fig.
2
). A lag phase (about 120 h), although shorter than the one at 30 °C, was again observed. At these conditions, the rumen microbial culture alone achieved a maximum concentration of 6.3 g/L (16% lower than the fermentation with T. reesei) from which 5.2 g/L is produced during the fermentation while 1.1 g/L was carried in the fermenter with the rumen fluid. Thus, the final SCFA concentration corresponded to a production yield of 0.32 g/g of cellulose, while when T. reesei was present, the production yield was 0.38 g/g of cellulose. The contribution of T. reesei in SCFA production and specifically to acetic acid production cannot be excluded since a paralog gene for aldehyde dehydrogenase (ALD1), which converts acetaldehyde to acetate, is not repressed by glucose (as it happens with the other paralog ALD2, and also with Saccharomyces cerevisae) (Chambergo et al. 2002). However, in previous experiments with T. reesei alone in the MBM system (Xiros and Studer 2017), acetic acid was not detected. Moreover, in all experiments, glucose and cellobiose were not detected in the medium throughout the cultivations indicating that cellulose hydrolysis was slower than sugar conversion to SCFAs.
Measurements of free cellobiohydrolase activity (CBH) in the liquid medium gave an indication of the cellulolytic efficiency of the consortium, since CBH is the most representative enzymatic activity for the substrate (microcrystalline cellulose) used in the fermentations (Sharma et al. 2018). As shown in Fig.
3
, at 30 °C, the presence of T. reesei resulted in higher CBH activities in the fermentation medium which is in accordance with the results on SCFA production shown in Fig.
2
a, b. The CBH activities during the fermentations at 39 °C were slightly better than in 30 °C (for both fermentations in the absence of T. reesei), indicating that probably some cellulolytic microorganisms present in the rumen remain more active at higher temperature.
Methanogenic activity during cellulose fermentations to SCFAs by the rumen microbial community
Carbon loss to by-products should be minimized in order to maximize the yields of the conversion process. Methane was considered as a major potential by-product, since methanogenic microbial strains are part of the rumen microbiome (Moss et al. 2000). Therefore, the carbon loss in the form of methane was investigated and the effect of methanogenic inhibitors on the SCFAs yields was evaluated. The methane production during the experiments was measured with or without the addition of a chemical inhibitor of the methanogenic activity (Fig. S3). As shown, during the experiments with T. reesei biofilm, at 30 °C, without the methanogens’ inhibitor (2-bromoethane sulfonate), the accumulated methane production reached 0.4 g at the end of fermentation, while it was eliminated when the inhibitor was added. The initial carbon available in the form of cellulose was 16.2 g. The carbon loss (0.3 g) due to the CH4 production corresponded to 2% of the total carbon mass available in the form of cellulose, and thus was considered negligible. This is also reflected on the production of SCFAs with or without inhibition of the methanogens, where as shown in Fig.
4
, no statistically significant difference was observed. Methane production during the fermentation with the rumen community alone was in all cases lower, even without the addition of a methanogens’ inhibitor, and even at 39 °C, which is the optimum temperature for rumen metabolism.
Effect of pH on production and composition of SCFAs
The pH range to be studied was selected on the basis of the rumen physiological pH as well as the optimum pH of the fungus which formed the biofilm. Rumen pH ranges from 6.2 to 6.5 while the optimum pH for Trichoderma reesei is 5. The range that was therefore tested in the experiments with this fungus was from 5 to 6.5. In Fig.
5
, the production of SCFAs for the different pH tested is shown. In the early stages of the fermentation, the acidogenic activity was higher at pH 5 and pH 6.5 than at pH 5.5 and pH 6. The highest SCFA yield (0.57 g/g of substrate) was achieved at pH 6.5 after 192 h of fermentation. At this point, the selectivities for acetic and butyric acid were 0.54 g/g and 0.22 g/g of total SCFAs, respectively (Fig. S4). However, at pH 6.5, the concentration of total SCFAs started to decrease after 200 h of fermentation. This decrease was not correlated to any methane production. The consumption of SCFAs could be attributed to consortium members which presumably are favored at higher pH values. At pH 5, a more stable production rate was maintained throughout the experiments probably also due to the better stability of the T. reesei cellulolytic enzymes at this pH resulting in a high cellulolytic efficiency for longer time.
Although duplicate experiments were performed at different time periods, and consequently they were inoculated with potentially different rumen microbial communities, the evolution of the SCFAs production did not change significantly from one period to the other (Fig.
5
). However, an increase in all absolute values of SCFA concentrations can be observed in the case of the second set of replicates. These concentrations (presumably attributed to differences in inoculum) are also higher than those reported in previous experiments (Fig.
2
, Fig.
4
).
Temperature strongly affects the composition of the produced SCFA mixtures
Fermentations were carried out at different temperatures to assess the effect of this parameter on SCFA production. Temperatures from 26 up to 32 °C were applied using T. reesei, while using Coprinopsis cinerea, fermentations were also performed at 37.5 °C. The fermentation temperature did not significantly influence the total SCFA production (online resource, Fig. S5). All experiments at temperatures from 26 to 37.5 °C gave very similar overall yields and productivities and only slightly higher overall yields were obtained from one replicate carried out at 32 °C. However, the temperature had a great impact on the composition of the SCFAs produced. As shown in Fig.
6
a, acetic acid yield increased while butyric acid yield decreased. The ratio of produced acetic to butyric acid was 0.83 at 26 °C, while it reached 2.0 at 32 °C.
Experiments at 37.5 °C and pH 6.5 were carried out to study the SCFA composition at conditions closer to the optimum for the rumen microorganisms. Coprinopsis cinerea was used as the biofilm forming, cellulolytic fungus. Although a fermentation temperature of 39 °C would be also interesting to study, preliminary experiments showed that this fungal strain was growing better at 37.5 °C, and therefore, this temperature was selected. As shown in Fig.
6
b, where the results of this experiment are compared with results obtained from fermentations at 30 °C (pH 6 and pH 6.5), this further increase of temperature had a dramatic effect on the composition of the SCFA mixture. Caproic acid was not detected in the production mixture throughout the fermentation and butyric acid concentration did not exceed 0.8 g/L, showing a decrease of about 400% compared with the results from lower temperatures. Acetic and propionic acids were practically the only acids produced, achieving selectivities (g of individual acid per g of total SCFAs produced) as high as 0.41 and 0.49, respectively. The acetic to butyric acid ratio was about 4.7 when the fermentations were stopped. The comparison of the three experiments in Fig.
6
b indicated that although pH affected the acetic to butyric ratio to a certain extent (see also online resource, Fig. S4), the temperature had a higher impact on the SCFA composition, showing a clear trend from longer to shorter SCFAs when the temperature increased.