Enhanced polyhydroxyalkanoate (PHA) production from the organic fraction of municipal solid waste by using mixed microbial culture

Percolates production from OFMSW

Experimental apparatus

The acidogenesis (SSAD) trials were carried out by developing a pilot-scale anaerobic percolation biocell reactor (APBR) that was made up of an insulated vertical cylinder (100 L of volume) of PVC material with a hermetic cap. Inside the reactor, there was a stainless steel basket with holes at the bottom, into which OFMSW was introduced from the APBR cap, as well as the digestate used to irrigate the organic wastes (Fig. 1). From the bottom of the APBR, the percolate generated daily due to digestate irrigation was extracted and weighed. Digestate irrigation was used to buffer the acidity produced during OFMSW fermentation and to remove OA produced in the APBR. This encouraged high hydrolysis performance from the organic waste, which is the limiting step in producing large amounts of OA from organic substrates [20, 21]. Moreover, large-scale OA production becomes toxic for methanogenic bacteria, limiting methane production during the trials.

Percolates treatment before PHA production

The three percolates obtained from anaerobic acidic fermentation trials were submitted to two pre-treatments before their use for polyhydroxyalkanoate (PHA) production. The first pre-treatment was to centrifuge the percolates at 20,000g for 15 min, reducing suspended organic carbon from percolates. Supernatants obtained underwent a second pre-treatment adopted to remove ammonia by stripping. This step was performed by bringing supernatant-pH to 11 by adding 6-mol L⁻¹ KOH and stirring it (magnetic stirring at 200 rpm) under the hood until a final C:N close to 10 was reached [11]. Then ammonia stripping was stopped and the pH was brought again to the initial value of 8 by using 3 mol L⁻¹ H₂SO₄. Ammonia stripped was not trapped at lab scale and it was not considered in this work. Obviously, this step needs to be considered and further studied on a larger scale.

After pre-treatments, the three supernatants were diluted with deionized water, to produce a final COD content close to 1300 mg L⁻¹ [11]. These OFMSW-derived supernatants (OFMSW-supernatants_in) were used for subsequent inoculum selection.

Substrates (OFMSW-supernatants_acc.) used for PHA accumulation tests were derived from percolates that had undergone the same treatments shown for those destined for the inoculum selection. Three OFMSW-supernatants_acc. were obtained but, in this case, the ammonia content was lower than those reported for OFMSW-supernatants_in, since it has been reported that N starvation can determine a greater conversion of carbon into PHA because of cell growth limitation [22]. Moreover, OFMSW-supernatants_acc. were less diluted, giving a final COD concentration of about 7500 mg L⁻¹, in order to avoid an excessive dilution of the biomass in the accumulation reactor [10].

Inoculum production and PHA-producing bacteria enrichment

The enrichment of PHA-producing bacteria was performed by using an inoculum constituted by an activated sludge (5 g total suspended solids L⁻¹) collected from the secondary sedimentation tank of a wastewater treatment plant (5.2 × 10⁵ equivalent inhabitants) located at Peschiera Borromeo (Milan, Italy).

The enrichment in PHA-producing bacteria was carried out in a Sequencing Batch Reactor (SBR) with a working volume of 800 mL, applying an aerobic dynamic feeding (ADF) strategy [11]. In brief, the SBR cycle length was of 12 h, consisting of four discrete phases: (1) influent filling (4 min), (2) aeration (675 min), (3) settling (40 min) and (4) withdrawal of the exhausted effluent (5 min), with 1 day of hydraulic retention time (HRT) and 5 days of Sludge Retention Time (SRT), keeping the temperature at 25 °C and the pH at 8.8, this latter controlled automatically by adding 1 mol L⁻¹ HCl. Aeration and agitation were provided by supplying air at 6 L min⁻¹ and stirring set at 110 rpm. Pumping, aeration and stirring were automatically controlled.

The selection process lasted for 3 months; for each month, a different pre-treated supernatant was used as substrate: OFMSW-supernatant_in 1, OFMSW-supernatant_in 2 and OFMSW-supernatant_in 3, respectively, for the first, second and third month of selection.

The selection trend was monitored by determining the duration of the feast phase achievable by using the dissolved oxygen (DO) concentration in the selection media [10], measured by an optical probe (FDO 925, WTW, Germany).

In particular, the feast (h)-to-famine (h) ratio (F/F ratio) was calculated as the ratio between the lengths in hours of the two phases. For a correct selection of PHA-storing bacteria, the F/F ratio had to be equal or less than 0.33 [23]. To carry out the selection of PHA-accumulating bacteria, 400 mL of activated sludge was used as the inoculum feed for each cycle, with 400 mL of OFMSW-supernatant_in. Organic Loading Rate (OLR) was kept close to 1300 mg COD L⁻¹ day⁻¹ and the C:N:P ratio was of about 100:13:0.4 mol C: mmole N: mmole P. Every month, two cycles were monitored in order to evaluate the performance of the selected culture. In particular, monitoring was performed between cycles 31 and cycle 45 since it was reported that the cultures reach stability after three SRTs from the beginning of the trial (cycle 31) [24].

PHA accumulation

The ability of the MMC to accumulate PHA was assessed by fed-batch assays carried out in a 300-mL working volume glass reactor, with continuous aeration and stirring. These assays consisted of feeding the substrate to 160 mL of enriched culture (at least 3 SRTs from the beginning of the selection) [24] adopting a pulse-wise feeding method. The assays were monitored continuously by measuring the concentration of the dissolved oxygen in the accumulation media [10]. In particular, substrate (OFMSW-supernatant_acc.) was fed to the reactor when DO showed a strong increase [10]. Total C dosed was calculated taking into account that the ratio of the carbon to the microorganisms had to be the same as that inside the selection reactor. The assays were stopped when no DO variation followed the substrate feeding.

For the accumulation tests, the operating conditions used were those adopted in the selection reactor, i.e. temperature of 25 °C, aeration of 6 L min⁻¹ and stirring at 110 rpm.

The biomass from the selection process was submitted to accumulation tests using the same substrate as the carbon source; every month, two accumulation trials were performed in duplicate, between the cycles 31 and 45.

PHA extraction

The biomass collected after PHA accumulation tests was centrifuged at 8000g for 15 min, washed with 0.9% sodium chloride solution and centrifuged again at 8000g for 20 min. The pellet obtained was lyophilized and then suspended in chloroform (in a ratio of ca. 40 mL CHCl₃ g⁻¹ dried cells) and left to dissolve for a period of 3 days at 37 °C [10]. The solution was then filtered to remove all undissolved material. The extracted PHA in chloroform was then precipitated by the addition of 5 volumes of methanol and allowed to settle down for 30 min [25]. The white precipitate formed was then filtered, suspended in chloroform and used to fill glass Petri dishes. Finally, chloroform was evaporated allowing polymer recovery in the form of a thin film.

Analytical procedures

Percolation trials

Percolates obtained were similar in terms of chemical characteristics apart from the percolate coming from the third trial that appeared more diluted as regarded OA and N-NH₄ ⁺ content (Table 1).

Trial 2 optimized OFMSW hydrolysis and subsequent acidogenesis fermentation, producing the highest amount of OA.

OFMSW-supernatants

OFMSW-supernatants used to perform PHA accumulation tests (OFMSW-supernatants_acc.) were again quite similar to each other, with the exception for P content and HB/HV precursors ratio which were higher for OFMSW-supernatant_acc. 3 than for the other OFMSW-supernatants_acc. (Table 2).

Mixed microbial culture enrichment

The culture fed with OFMSW-supernatant_in 3 gave different results as the polymer content at the end of the feast phase and PHA storage yield on COD consumed were lower by 36 and 20% in comparison with the data reported for OFMSW-supernatants_in 1 and 2. On the other hand, PHA storage yield on OA consumed increased by 34%. In any case, performance obtained with OFMSW-supernatant_in 3 was better than that reported in the literature for similar substrates (OFMSW leachate) [17, 18] and it was comparable to those obtained with other waste substrates (e.g. fermented molasses and fermented cheese whey) [10, 29]. Probably, differences found in terms of reduction of polymer content were because OFMSW-supernatant_in 3 was characterized by a higher presence of P that reduced PHA accumulation [30].

Feast-to-famine ratio obtained during all selection processes using the three OFMSW-supernatants_in (F/F ratio ranged from 0.03 to 0.15) (Fig. 2), confirmed the good performance of the selection procedures, indicating successful enrichment in PHA-accumulating bacteria [23].

PHA production

Accumulation tests (Table 3) revealed a better performance for the cultures selected with OFMSW-supernatants_in 1 and 2 than that of 3 in terms of PHA accumulation in relation to VSS. Maximum PHA content at the end of the tests fed with OFMSW-supernatants_acc. 1 and 2 was of 467 ± 41 g kg⁻¹ VSS and 465 ± 21 g kg⁻¹ VSS, respectively, which were about 13% higher than that obtained in the tests fed with OFMSW-supernatant_acc. 3 (Table 3). These results were similar to those obtained in previous work using other waste substrates (e.g. palm oil mill effluent, fermented olive oil mill pomace, crude glycerol) [31–33], indicating a reliable performance of the cultures in accumulation tests. In terms of PHA storage yields (expressed as COD) on COD consumed, PHA accumulation carried out by using the three supernatants gave substantially similar results when OFMSW-supernatants_acc. 2 and 3 were used, i.e. 0.5 mg mg⁻¹ COD_cons.. These data were higher than those detected for the accumulation trial using OFMSW-supernatant_acc. 1.

PHA storage yields (expressed as COD) referred to as consumed-OA were very different for different supernatants. For example, PHA yield was reduced from 1.13 to 0.66 mg mg⁻¹ COD_OA-cons. when comparing the trial performed using the OFMSW-supernatant_acc. 1 to the one using the OFMSW-supernatant_acc. 3. Residual OA could be potentially reused to produce PHA or destined to the production of biogas and energy, supporting the entire process throughout anaerobic digestion. These solutions have not been considered in this work.

Trials performed using the OFMSW-supernatants_acc. 1 and 2, gave the highest PHA production with respect to the total COD and OA fed, i.e., as average, 130 ± 16 g kg⁻¹ COD_in and 223 ± 28 g kg⁻¹ OA_in, that were 30% higher than data obtained for the culture fed with the OFMSW-supernatant_acc. 3. Lower PHA production was probably due to the decrease of biomass which occurred in the SBR during the last month (data not reported).

PHA composition was constant among the three different accumulation tests performed, with an average value of 56 ± 3% of HB and 44 ± 3% of HV, as detected by the GC–MS approach. This composition was similar to that obtained by Amulya et al. in 2015 [17] using fermented OFMSW as substrate.

PHA production and mass balance

Results above discussed indicated that supernatants obtained worked well as substrates to select and enrich MMC in PHA-storing bacteria. In particular, PHA content at the end of the feast phase and PHA yield on COD and on OA consumed obtained in this work were comparable or higher than those reported in the literature for selection processes of mixed microbial cultures obtained by feeding similar OFMSW leachate [17, 18] or other waste substrates (e.g. fermented molasses and fermented cheese whey) [10, 29]. Best performances obtained could be due to the fact that in this work culture enrichment was performed at 30 °C, which is an optimal temperature for microbial activity, and because a more balanced C:N ratio (close to 10) was adopted with respect to the lower values adopted by other authors [15, 18].

The PHA yields obtained, i.e. 0.44–0.52 mg mg⁻¹ COD_cons. and 0.65–1.13 mg mg⁻¹ COD_OA-cons. (Table 4), were very good and comparable to those reported in the literature by using other waste substrates to produce PHA (e.g. fermented olive oil mill effluent, fermented molasses, fermented cheese whey) [10–12]. Moreover, results from this work were higher than that reported by Zhang et al. in 2014 [16] concerning the use of fermented food waste and sewage sludge as the substrate for PHA production (Table 4).

Percolation trials indicated that the process parameters adopted for Trial 2 determined the highest percolate and OA production per kg of OFMSW treated, giving the best trial results among the three tested. This evidence together with the very good performance in converting OA into PHA (second stage) using the OFMSW-supernatant_acc. 2, resulted in the highest PHA production among the three two-stage processes, i.e. 33.22 ± 4.2 g kg⁻¹ of OFMSW_w.w., corresponding to 114.4 ± 14.5 g kg⁻¹ of OFMSW_TS, almost 5 times higher than that reported by Rhu et al. in 2003 [15] (25 g of PHA kg⁻¹ of dry food waste).

PHA characterization

Mass balance previously discussed indicated that Trial 2 gave the best results in terms of total PHA yield. Therefore, PHA characterization was focused on the products obtained in Trial 2.

Solid-state ¹³C NMR was used to evaluate the content of PHA in the generated biomass and to monitor the PHA extraction process. ¹³C cross polarization NMR technique, based on the carbon magnetization transferred from ¹H protons, could not be used for quantification of different chemical species/functional groups. In contrast, single-pulse excitation (SPE) NMR is based on the direct ¹³C nuclei polarization and, even if less sensitive than CP MAS NMR, can be optimized for quantitative evaluation analysis. The most critical parameter, recycle delay, was set at 160 s, in order to cover T1 values of each ¹³C type [36]. The obtained ¹³C SPE NMR of the reference PHA standard, biomass enriched in PHA and the biomass after PHA extraction are shown in Fig. 3.

Together with the PHA signals, the biomass spectrum contains signals compatible with proteins (large CO peak at 170–185 ppm), minor polysaccharides (large peak of anomeric O–C–O at 100–110 ppm), etc. It is worth noting that PHA-related peaks are sharper than the other signals, indicating that these polymers were present, at least partially, in their semi-crystalline form.

In spite of the optimized acquisition parameters, quantitative analysis of PHA in biomass was limited by strong signal overlapping, especially in the aliphatic region (0–80 ppm) (Fig. 3). As a tentative for PHA quantitative evaluation in the original biomass, a ratio between CO^PHA and CO^biomass was estimated as 30–35 molar %, using a Topspin software for peak integral simulation. Since the result significantly depends on the estimated signal width, such an approach could not be applied for the exhausted biomass, where the CO^PHA peak is not sufficiently high. It still can be clearly observed from the decreased signals of CO^PHA (~172 ppm) (Fig. 3) and well isolated CH ₃ ^3HV (~10 ppm) (Fig. 3) that the biomass treated with chloroform contained PHA in lower quantity. Notably, even before extraction, it can be seen that the synthesized PHA are characterized by higher content of 3HV monomer that is in agreement with GC–MS data. Altogether, these data can be useful for monitoring the PHA content in biomass during their production and extraction.

In order to provide more detailed structural features of the extracted PHA polymers, solution ¹H NMR, GC–MS and GPC-TDA were applied along with solid-state NMR. By doing so, extracted PHA was characterized by ¹H NMR using standard PHA^88/12 for signal assignment (Fig. 4).

Gel permeation chromatography with a TDA detection system was used to describe the molecular weight distribution of the biosynthesized PHA. Separation and detection conditions (including dn/dc calculation) were optimized using a commercial PHA^88/12 sample. Table 6 reports the average molecular weight Mw, polydispersity and hydrodynamic radius Rh determined for the reference PHA^88/12 and extracted PHA sample. In spite of a similar polydispersity, the biosynthesized PHA sample was characterized by higher molecular weight and hydrodynamic radius. Mark–Houwink parameters a and logK, reflecting conformational behaviour of polymers in solution, are also reported in Table 6. The obtained a values for reference PHA^88/12 and extracted PHA are compatible with the values characteristic for flexible polymers in solution. Interestingly, these properties do not depend on the differences in molecular weight and chemical composition between the two samples.