Influence of the total concentration and the profile of volatile fatty acids on polyhydroxyalkanoates (PHA) production by mixed microbial cultures

The inoculum used in the batch tests comes from a sequential system of two semi-continuous reactors coupled in series.

The first reactor is a 5L-anaerobic and thermophilic stirred tank reactor (55 °C) which has been fed with ESBC that produces hydrogen-rich biogas and a VFA-rich effluent (acetic, propionic, butyric, valeric, etc.). The reactor was started from an anaerobic reactor that operated under methanogenic anaerobic digestion conditions. In this reactor, to produce the complete washing of the methanogenic microbiota in 21 days, a strategy of reducing the hydraulic retention time (HRT), from 18 to 6 days, and the pH until 5.5 as set point was applied. This reactor was in stable operation for more than 250 days with an average production of biogas (hydrogen and carbon dioxide) at the moment to perform the batch tests over 1.5–2.0 L/L_reactor·day and with an average hydrogen percentage of 30–40% v/v in the biogas.

The liquid fraction of the acidogenic effluent from this first reactor was used to feed a second reactor destined for the PHA production, with a stable population of microorganisms selected from mixed sludge from a wastewater treatment plant (WWTP). A solution of nutrients recommended in the literature was added to this liquid fraction to enhance the PHA production [23]. The total volatile acidity concentration in the acidogenic effluent was maintained in the range of 10–20 g HAc/L.

In the second 5L-reactor for the PHA production, a microbiota capable of producing PHAs was selected, applying a strategy based on alternating oxic and anoxic periods of 24 h, coinciding with periods of scarcity and abundance of the substrate (famine-feast cycles). During the oxic periods, the soluble oxygen concentration was kept between 6 and 8 mg/L. The operating HRT, which was initially set at 12.5 days, was reduced to 9 days once the selection of the PHA-producing microbiota was made to increase the rate of the process. Throughout this process to obtain a stable PHA-producing inoculum, the PHA production yields were among 20–40 mg PHA/L, with values close to 60 mg PHA/L (13% dry cell weight) when it was used for the inoculation of batch reactors. The reduction of organic matter, in terms of soluble chemical oxygen demand (COD), was around 90–95% with an organic loading rate of 2.5 g COD/L_reactor·day at the moment in which it was used as inoculum for the start-up of the batch tests.

To carry out this study, nine duplicated tests were carried out operating in batch mode and using 1-L reactors with a 50:50 (v/v) substrate to inoculum ratio. The inoculum used comes from the inoculation PHA reactor described in the previous section. The substrate was composed of a synthetic mixture of acetic, propionic, butyric, and valeric acids in a nutrient solution previously indicated. All the trials were performed by combining these conditions according to Table 1. Albuquerque et al. [24] have reported that no significant variations were observed in terms of PHA yield, maximum PHA content, and PHA composition between the synthetic and fermented feedstocks.

At the inoculation time, in the effluent from the acidogenic reactor, the average proportions of acids were 83.2:1.0:43.9:2.0 in terms of acetic:propionic:butyric:valeric, respectively. In all the synthetic mixtures, concentrations of acetic and propionic acids were incorporated in the same proportion of the acidogenic effluent obtained from the DF reactor but modifying the proportions between butyric and valeric acids, with respect to their original proportion with acetic acid, in order to obtain the valeric:butyric ratios to be tested (10:1; 1:1; and 1:10).

Given the results obtained for the first set of assays with high butyric acid proportion, it was decided to modify the initially planned experimental set up in the study. Thus, it was decided to check, in the second set of assays (neutral ratio assays), a range of total acidity concentrations higher than the previous ones (1.5–3.5 g/L) but maintaining the butyric acid concentration lower than 1 g HBu/L.

Finally, for the third set of assays (high valeric assays), it was considered appropriate to continue increasing the total volatile acidity concentration but limiting the presence of butyric acid in order to analyze the independent influence of valeric acid on the PHA production in a more evident way.

Each set of tests consisted of six reactors (three conditions by duplicate) with synthetic mixtures. The tests have been carried out at 25 °C and with a constant aeration rate of 20 L_air/L_reactor·h in order to maintain an oxic environment and to promote the PHA production. The system used the forced aeration itself for homogenizing the content of the reactors.

The analytical determinations were carried out according to the standard methods for wastewater [25] adapted to high organic content waste. For the monitoring of the reactors, the following parameters were analyzed: pH (4500-H⁺), dissolved oxygen (DO) (4500-O), volatile suspended solids (VSS) (2540-B/E), COD (5220-C), and dissolved organic carbon (DOC) (5310-B).

For the determination of DO, an oximeter (Labprocess® OXY 70) was used based on an optical luminescence cell. The DOC was determined by means of a carbon analyzer (Analytik Jena® multi N/C 3100) with samples previously filtered by 0.47 μm.

Additionally, the quantification of the total PHA concentration was performed by UV spectrophotometry at 235 nm after an extraction process according to the procedure described in the literature [26, 27].

The individual VFAs were measured by gas chromatography in a Shimadzu® GC-2010 equipped with a flame ionization detector (FID) and a fixed phase column of the Nukol® type (polyethylene glycol modified with terephthalic acid). Hydrogen was used as carrier gas at a flow rate of 50 mL/min. The temperature of the injection port and the detector were 200 °C and 250 °C, respectively [28].

The biogas generated by the acidogenic anaerobic reactor was collected in Tedlar® gas bags (SKC® 232 series) and it was quantified using a wet rotary drum gas flow meter (Ritter® TG5) connected to a Laboport gas suction pump (KNF®, model KT18) with a capacity of 5.5 L/min under standard conditions of pressure and temperature. The biogas composition from the acidogenic anaerobic reactor was determined by a gas analyzer EMERSON® X-Stream.

PHA-producers bacteria have been identified in the effluent of the PHA inoculum reactor. For that, the sample was serially diluted and plated on LB (Luria–Bertani) and VKM culture medium. VKM medium contained acidogenic effluent:nutrient solution:distilled water, 3:4:3 ratio, supplemented with yeast extract (1 g/L) and agar (25 g/L). The plates were incubated at 25 °C for 7 days. For each medium, several representative types of colonies were identified and purified after several passages on the same type of medium.

The most representative colonies were isolated and genomic DNA was extracted following the protocol described by Saba et al. [29]. The isolates were typed by sequencing the 16S rRNA gene that had been amplified by PCR from genomic DNA [30]. Similarity searches were performed with the Basic Local Alignment Search Tool (BLAST) [31] on the DNA sequences with the Genbank® database.

A total of 20 bacteria were isolated from the PHA production bioreactor. The 16S ribosomal RNA sequences of the isolates were matched with the contents of the databank by Nucleotide BLAST. At a percent identity score of  ≥ 97%, a total of 8 different bacteria genera were identified: Agrococcus (98.4%), Corynebacterium (97.2%), Cupriavidus (97.5%), Enterobacter (98.9%), Microbacterium (98.9%), Salana (97.1%), Stenotrophomonas (99.7%), and Pseudoclavibacter (99.1%). Cupriavidus (n = 6 isolates tested) was the most common genus in the inoculum, followed by Corynebacterium (n = 5) and Enterobacter (n = 4), while Salana, Pseudoclavibacter, Agrococcus, Stenotrophomonas, and Microbacterium were identified in one isolate. Of these, the genera Enterobacter [32], Corynebacterium [33], Cupriavidus [34, 35], and Microbacterium [36, 37] have been previously described as PHA producers. In addition, Cupriavidus necator is the most widely used Gram-negative bacterium due to its high PHA production capacity [8].

Except on day 0 (start-up) when some reactors had pH below 7, which was immediately corrected, the pH values always remained in a very stable range between 8 and 9, with a slight trend to increase. This range of values agrees with data reported by Chua et al. [38] using batch reactors fed with urban wastewaters.

On the other hand, DO concentration showed the opposite behavior to the PHA accumulation, decreasing gradually as the VFAs were consumed (see Sect. 3.3). Thus, the lowest values were reached when PHA production peaked while the initial DO level was recovered once the microorganisms begin to consume the PHAs formed due to a shortage of other available carbon sources. Similar results were obtained by Mannina et al. [13] working with acetic- and acetic-propionic-based feedings.

In order to illustrate this behavior, the evolution of pH, DO oxygen, and PHA production in the assay called 1.5B has been selected to be shown in Fig. 1. The rest of the tests have similar evolutions of the mentioned parameters.

This section presents the evolution of COD, DOC, and total volatile acidity as representative parameters of the organic matter available for the PHA production in the batch tests.

Figure 2 shows the evolution of the COD in the reactors grouped by the valeric:butyric ratio. All evolutions show a clear decreasing trend except for the assay named 3.5B where the decrease was slow and weak and, hence, as far as the trial progresses, the amount of organic matter is lower due to the degradative activity of the microorganisms.

However, the diminishing does not occur in the same way for tests that have the same valeric:butyric ratio but different total volatile acidity concentration. In general, the decline is rapid (reaching the final levels in 6–8 days) except for test 3.5B where the high butyric acid concentration seems to be a problem for COD reduction. The literature [39] indicates that if C. necator has optimal growth conditions (including enough carbon and nutrient sources), the acetyl-CoA is massively incorporated into the Krebs cycle. This metabolic process involves great CoA production. Consequently, the excess of CoA inhibits the actuation of 3-ketothiolase (the enzyme related to acetoacetyl-CoA production which is the first step in the PHA production pathway). Therefore, in optimal condition, the PHA production could be stopped in the first step [39]. An increase in substrate concentration could derive in a PHA production inhibited by the substrate as some authors have confirmed [40‐43]. While the other assays with the same ratio valeric:butyric (1.5B and 2.5B) reduce their concentrations to less than half in 4 days, in the 3.5B trial, the degradation is weak and delayed. Nevertheless, when the proportion of valeric acid is highest (5.0 V, 7.0 V, and 9.0 V), the degradation is still fast and very uniform even when the total volatile acidity increases. This fact shows that the system presents a lower sensitivity to high valeric acid concentrations (up to 4 g/L in assay 9.0 V) as it will be seen in the following sections.

The other two parameters related to available organic matter that have been analyzed are DOC and total volatile acidity. Their evolutions are shown in Figs. 3 and 4, respectively. As for the previous COD graphs, the results have also been grouped by their valeric:butyric ratio.

In a much more evident way than in the case of COD, a gradual decrease in DOC concentration has been observed. It should be noted that the final levels after 8–9 days are very similar regardless of the initial concentration of total volatile acidity (419–453 mg C/L and 183–227 mg C/L in assays V, and N, respectively), except for the highest butyric concentration assay (503 mg C/L in 3.5B versus 93–179 mg C/L in 1.5B, and 2.5B, respectively). This test, as it was previously noted in the COD evolution, seems to have different behavior and it could have developed an incomplete substrate degradation for the duration of the test. Moreover, the 3.5B test also showed an initial period of 4–5 days with a notably lower COD consumption rate than the other tests.

In the case of the total volatile acidity, determined as the weighted sum of the concentrations of individual VFAs (C2-C7) and expressed in units of mg HAc/L, decreasing is gradual and clear. VFAs are the ultimate substrate that would lead to the PHA production.

It could be observed that high valeric acid concentration assays (Fig. 4c) have shown a sharper and more continuous VFAs degradation rate than high butyric (Fig. 4a) and neutral ratio assays (Fig. 4b). Thus, a much more uniform trend can be observed for the degradation of total volatile acidity in the high valeric acid content assays.

As it was observed for COD and DOC evolutions, the behavior of the 3.5B test is also differentiated. In fact, VFAs degradation was slower and delayed in 3.5B compared with 1.5B and 2.5B.

In short, the analysis of DOC, COD, and total volatile acidity denotes that all the tested conditions, except 3.5B which corresponds to the highest butyric acid concentration, have been successful in removing a high proportion of organic matter in a very short time.

The analysis of the individual evolution of the four VFAs used in the synthetic feed allows observing how their consumption is coupled to the PHA production. For each condition, a graph has been built in order to analyze how the total volatile acidity and valeric:butyric ratio influence the PHA production (see Fig. 5).

The PHA yield should not be measured in terms of concentration, as the biomass concentration changes throughout the process due to cell growth. Therefore, the PHA production yields have been calculated by dividing the total PHA obtained by the VSS. In this case, the VSS were assumed to represent the amount of cell biomass in the assays.

Furthermore, this ratio (g PHA/g VSS) is a key factor for the PHA extraction stage since if its value is high the recovery of PHA (downstream industrial processes) is easier and cheaper [13]. Figure 7 shows the evolution of the PHA yield estimated in this way.

Table 2 shows the maximum values of the PHA yield (expressed as percentage) with an indication of the day on which they have been obtained. The average PHA yield values are also given for the period from the start of the test to the day when the maximum PHA production yield was reached. Additionally, Table 2 includes the values of the average daily rate (ADR) for PHA production. ADR represents the ratio between the maximum PHA yield of the process and the time required to reach it.

The ADR is a parameter that allows comparing the different conditions tested in order to select that leading to maximum PHA production in the minor time, which could be of industrial interest. Results of ADR in Table 2 show that better conditions are corresponding to the minor total volatile acidity concentration in each set of assays: 3.0 N, 1.5B, and 5.0 V.

Clearly, the lower acid concentration assays (1.5B, 3.0 N, and 5.0 V) offered the best yields within each valeric:butyric ratio. As the acid concentration increases in each ratio, the MMC were less capable of directing carbon consumption towards PHA production. It is noticeably evident (assays 3.5B, 5.0 N, and 9.0 V) that the ADR of PHA production is visibly affected by the high butyric and valeric concentration. However, the sensitivity to each acid is not the same. The high butyric acid assays at low concentrations of total volatile acidity (1.5B and 2.5B tests) gave better PHA yields (23.0 and 17.1%, respectively) than the equivalent tests with high valeric:butyric ratio. Thus, in the last case (5.0 V and 7.0 V tests), the maximum PHA yield reached was 14.2 and 8.5% despite the butyric acid concentration was similar or even lower than in 1.5 B.

Table 3 shows data from different authors of the maximum accumulation capacity for PHA production, using VFAs, by MMC. As it can be seen, the PHA yield values are between 23 and 53%. The results obtained in this work are in the lower range, although they are practically the same as those obtained by other authors who have worked with VFAs with MMC.

PHA titer values are highly dependent on the operational conditions and configuration of the process for PHA production from VFAs by MMC. Ciesielski and Przybylek [51] observed PHA titers between 227.8 (acetic as feeding) and 673 mg/L (propionic as feeding) in batch reactors inoculated with activated sludge from wastewater treatment plant. In the present study, propionic was always under 70 mg/L in feeding while acetic acid was fed between 704 and 2426 mg/L and the PHA titer was over 200 mg/L just in 1.5B and 2.5B assays where acetic was in lower levels but higher butyric concentrations where supplied. It is should be remarked that butyric acid follows the same metabolic pathway than acetic acid for PHA production [52]. However, PHA titer from MMC could be considerably higher under most favorable conditions as it can be seen in the study of Pérez-Zabaleta et al. [16].

The results obtained for the maximum PHA production yield indicated in Table 2 can be related to the two main variables analyzed in this study: the concentration of total volatile acidity fed to the system and the valeric:butyric ratio.

Thus, representing the maximum values of PHA production obtained in each test versus the initial total acidity (expressed as acetic acid), Fig. 8a is obtained. In this figure, it can be seen that the results corresponding to the tests with a ratio of butyric:valeric of 10:1 (B) and 1:1 (N) have the same trend, while the tests corresponding to the ratio 1:10 (V) show a differentiated behavior. In all the cases, the trends observed are fitted to linear correlations. In Fig. 8b and 8c, the maximum PHA yields are plotted versus the initial concentrations of butyric and valeric acids, respectively. These last two graphs clearly show that the data also cluster in linear distributions for each of the butyric:valeric ratios tested.

As it can be seen in Fig. 8, the evolutions of the maximum PHA production yield adopt a distribution with respect to the initial acid concentrations that clearly differentiate the set of tests carried out with each butyric:valeric ratio. In all the cases, the observed evolution indicates that the maximum PHA production yield decreases as the initial concentration of the considered acid increases. This implies that the VFAs, although they are the substrate used by the microorganisms in the generation of PHA, have an inhibitory effect on the process. This effect has been observed by other authors [40‐43, 53].

Concretely, Tamang et al. [53] have estimated the inhibitory effect of VFAs in some industrial wastewaters between 7.1 and 82.1 mM by means of measuring the diminishing in the specific oxygen uptake rate (SOUR) over diluted samples of wastewaters. With dairy wastewaters, the SOUR decrease by 83% when the total volatile acidity concentration was incremented from 4 to 15.7 mM.

As it can be seen in Fig. 8, the different sets of data have been fitted by linear regression. Table 4 shows the representative parameters of each linear adjustment. Letters B, N, and V in the column named as ratio indicate the set of points used for fitting according to the nomenclature defined in Table 1. The slope is related to the inhibitory effect of each acid, as it was previously commented. The y-interception corresponds to the maximum PHA yield value achievable under the operating conditions. These values are in the same range as described by other authors [24, 54‐56] for the maximum PHA accumulation in the cell (expressed in %) when using MMC and carbon sources corresponding to wastes and not pure compounds and well-defined media.