Reproducible, high-yielding, biological caproate production from food waste using a single-phase anaerobic reactor system

Background Nowadays, the vast majority of chemicals are either synthesised from fossil fuels or are extracted from agricultural commodities. However, these production approaches are not environmentally and economically sustainable, as they result in the emission of greenhouse gases and they may also compete with food production. Because of the global agreement to reduce greenhouse gas emissions, there is an urgent interest in developing alternative sustainable sources of chemicals. In recent years, organic waste streams have been investigated as attractive and sustainable feedstock alternatives. In particular, attention has recently focused on the production of caproate from mixed culture fermentation of low-grade organic residues. The current approaches for caproate synthesis from organic waste are not economically attractive, as they involve the use of two-stage anaerobic digestion systems and the supplementation of external electron donors, both of which increase its production costs. This study investigates the feasibility of producing caproate from food waste (FW) without the supplementation of external electron donors using a single-phase reactor system. Results Replicate leach-bed reactors were operated on a semi-continuous mode at organic loading of 80 g VS FW l−1 and at solid retention times of 14 and 7 days. Fermentation, rather than hydrolysis, was the limiting step for caproate production. A higher caproate production yield 21.86 ± 0.57 g COD l−1 was achieved by diluting the inoculating leachate at the beginning of each run and by applying a leachate recirculation regime. The mixed culture batch fermentation of the FW leachate was able to generate 23 g caproate COD l−1 (10 g caproate l−1), at a maximum rate of 3 g caproate l−1 day−1 under high H2 pressure. Lactate served as the electron donor and carbon source for the synthesis of caproate. Microbial community analysis suggested that neither Clostridium kluyveri nor Megasphaera elsdenii, which are well-characterised caproate producers in bioreactors systems, were strongly implicated in the synthesis of caproate, but that rather Clostridium sp. with 99% similarity to Ruminococcaceae bacterium CPB6 and Clostridium sp. MT1 likely played key roles in the synthesis of caproate. This finding indicates that the microbial community capable of caproate synthesis could be diverse and may therefore help in maintaining a stable and robust process. Conclusions These results indicate that future, full-scale, high-rate caproate production from carbohydrate-rich wastes, associated with biogas recovery, could be envisaged. Electronic supplementary material The online version of this article (10.1186/s13068-018-1101-4) contains supplementary material, which is available to authorized users.

. Strategies applied to improve caproate production from food waste SRT: Solid retention time; FW: food waste; VS: volatile solid; a leachate from previous batch was diluted 15 times and used as starting liquid; b half of the leachate from the liquid bed was removed and replaced with water.
Section S1. Enrichment culture assays Enrichment cultures were developed using the following selected substrates: whatman filter paper 1 (WP), skimmed milk (SM), xylan, oleate and palmitate as cellulose, protein, hemicellulose, and fat sources respectively. Briefly, granular sludge was used to inoculate 19 ml of sterile anaerobic mineral medium [1], to which sterile stock of substrate was provided to a final concentration of 3 g l -1 except for oleate and palmitate provided at 1 mM. All cultures were incubated at 37°C and agitated. Substrates degradation was indirectly monitored by measuring the biogas accumulated in the headspace of the bottles. Distinct enrichment series were obtained by successive transfers of active cultures (10%) into fresh medium containing the relevant substrate.

Section S2. Substrate utilisation assays using enriched cultures
The ability of each enriched culture to metabolise the substrate more rapidly than the initial granular sludge was investigated by following the same procedure as the enrichment culture assays described above. This time, sacrificial vials were set up for the direct monitoring of substrate degradation. The protein concentration of the skimmed milk powder was determined by using the RC DC™ Protein Assay kit (BIO -RAD) which is based on the Lowry protocol.
Degradation of WP, xylan, palmitic and oleic acid were established gravimetrically through the reduction of the corresponding total solid contents.

Section S3. Enrichment culture processes successfully selected for cellulose and hemicellulose degraders
The hydrolysis constant kh for WP (0.33 d -1 ) and xylan (0.14 d -1 ) were found to have improved when using enriched cultures as compared to the granular sludge for which Kh for WP and xylan were 0.13 d -1 and 0.07 d -1 respectively. The Kh for SM (0.67 d -1 and 0.74 d -1 ) did not show any significant improvement likely due to the previous adaptation of the initial inoculum (granular sludge from dairy waste water facility) to similar substrate. As rates of degradation of oleate and palmitate were very slow, hydrolysis constants were found to be close to zero and were not reported.

Section S4. Bacteroides graminisolvens and Porphyromonodaceas are implicated in cellulose and hemicellulose hydrolysis, respectively
The microbial community composition from the seed sludge and the 18th generation of xylan while less than 1% methanogenic community was found in the hemicellulose enriched culture.
The enrichment cultures were set up with the intention to enrich for hydrolysers rather that fermenters and methanogens.

Section S5. Bio-augmentation Assay
Bio-augmentation experiment was performed using the cellulose, hemicellulose, protein and fat enriched cultures. Five percent of each enriched culture obtained after several sub-culturing were mixed in 50 ml tubes and centrifuged at low speed (indicated the speed here please). The supernatants were discarded and the cells pellets were added to 500 ml bottles containing mixture of digestate (from R1, R2 and R3, day 7 of batch 18) and FW to a ratio of 0.25 (digestate/FW, on the VS basis). The operational conditions were similar to the one described for leach-bed reactors. After 7 days of incubation at 37 o C, the content of each bottle was mixed with some digestate from day 7 of batch 19 and the mixture was used to inoculate reactor R1, R2 and R3 (containing fresh FW) at the ratio of 0.25 (inoculum/FW on the VS basis).

Section S6. Characterisation of the food waste and digestate
The lipid fractions of the FW and digestate were determined using the methanol-chloroform method previously described by Folch et al. [2]. Hemicellulose and cellulose fractions were determined using a combination of dilute and strong acid hydrolysis at high temperature.
Briefly, 4 ml of 6 M sulphuric acid was added to 100 mg of dried FW or digestate. The mixture was incubated at room temperature (RT) for 30 min followed by the addition of 17.5 ml of water and autoclaving at 121 o C (15 min holding). The tubes were centrifuged at 6000 × g for 15 min and supernatants containing hydrolysed sugar derived from hemicellulose were stored at room temperature (RT). The pellet was dried overnight at 60 o C before concentrated sulphuric acid (18 M) was added. The tubes were incubated at RT for 30 min followed by addition of 18 ml water and autoclaving at 121 o C (with 15 min holding). The tubes were centrifuged and the resulting supernatant containing glucose released from cellulose hydrolysis stored at RT. The resulting sugar solutions were analysed using the phenol-sulfuric method by Dubois et al. [3]. The sample is combusted with oxygen and the nitrogen released is then converted to protein using a conversion factor.

Figure S1
Microbial communities profiling from the granular seed sludge, whatman filter paper 1 (WP) and xylan enrichment cultures (18th generation) assigned from the 16S rRNA gene sequencing from DNA samples.

Section S7. Ribonucleic acid extraction
Briefly, 0.5 ml of 10% hexadecyltrimethylammonium bromide (CTAB) extraction buffer and 0.5 ml of phenol-chloroform-isoamyl alcohol (25:24:1) were added to 2 ml Eppendorf tubes containing frozen microbial cell pellets (from -80 o C). CTAB extraction buffer was prepared by mixing equal volumes of 10% (wt/vol) CTAB (Sigma, Poole, United Kingdom) in 0.7 M NaCl with 240 mM potassium phosphate buffer, pH 8.0. To each tube, 0.5 mm and 0.1 mm diameter zirconia beads (Thistle Scientific) were added and the microbial cells were lysed by bead beating for 5 min using a vortex (IKA Vortex Genius 3). The top aqueous layer were recovered after 10 min of centrifugation at 16, 000 × g and transferred to clean tubes. To remove residual phenol from previous step, 0.5 ml of chloroform-isoamyl alcohol (24:1) were added and the aqueous layer was recovered after 5 min of centrifugation at 16,000 × g and transferred to clean tubes. Subsequent precipitation of total nucleic acids were performed by adding two volumes of 30% (wt/vol) polyethelene glycol (PEG) 6000 (Fluka BioChemika)-1.6 M NaCl and incubated at 4 o C for 2 hours followed by 20 min centrifugation at 16, 000 × g. Supernatant were carefully discarded and the pellets were washed by adding 1 ml ice cold 70% ethanol followed by 30 min centrifugation at 16,000 × g. Ethanol was completely removed and the pellets left to dry at room temperature for 2 min. Finally, the pellets containing DNA and RNA were resuspended in 50 µl RNase free water.

Figure S8
Profile of individual VFA production in leach-bed reactors R1, R2 and R3 during Phase 2 and 3 (period 2). Batch (B) 13 and 14 were selected to represent phase 2 and 3 corresponding to high and low VFA concentrations in starting liquid respectively. Data at each point represent average of duplicate measurements.

Figure S9
Profile of the lactate concentration in triplicate leach-bed rectors (R1 -R3) during period 1 and 2 (phase 2 and 3). Batch (B) 7 was selected to represent period 1 corresponding to 14-day SRT. B13 and 14 were selected to represent phase 2 and 3 corresponding to high and low VFA concentrations in starting liquid respectively. Value at each point is the average of duplicate measurements.
na: methane not detected in biogas