Freshly harvested sugarcane stalk was purchased from local farmers in Khon Kaen, Thailand. It was milled to obtain the juice, which was subsequently concentrated by boiling to a total sugar concentration of ca. 600 g/L, and stored at −20°C until use. The concentrated sugarcane juice (SJ) contained sucrose as the major component. Potassium, magnesium, and calcium were the major elements in the concentrated SJ. Protein was also present in the juice, but at a very low level. The compositions of the concentrated SJ are given in Table 1.

Paracoccus sp. KKU01 was used as the PHB producer. It was isolated from a soil sample collected at a cassava starch production plant, Kaensiri Starch Co., Ltd., Phra Yuen, Khon Kaen, Thailand. The bacterium was screened and isolated using the method of Kitamura and Doi (1994), and was identified at the Thailand Bioresource Research Center (TBRC), Pathum Thani, Thailand, using 16S rDNA sequencing analysis. According to the TBRC report, two primers were used, i.e., 27F (5’-AGA GTT TGA TCM TGG CTC AG-3’) or 800R (5’-TAC CAG GGT ATC TAA TCC-3’) and 518F (5’-CCA GCA GCC GCG GTA ATA CG-3’) or 1492R (5’-TAC GGY TAC CTT GTT ACG ACT T-3’). The identification was carried out using the BLASTN program against 16S rDNA sequence database of validly published prokaryotes. The isolated strain was found to have 99.93% similarity to Paracoccus denitrificans (Accession no. MT491243). Preliminary study showed that Paracoccus sp. KKU01 could utilize glucose and sucrose for growth, and could accumulate PHB under nitrogen-limited conditions (unpublished data). The bacterium was maintained on nutrient agar, supplemented with 20 g/L of sucrose as the carbon source, at 30°C. Inocula were prepared by transferring one loop-full of the bacterium growing on nutrient agar into nutrient broth containing 20 g/L of sucrose, and incubating at 30 °C, 150 rpm, pH 6.9 for 15 h. PHB production medium was prepared following the formula of Kalaiyezhini and Ramachandran (2015). The medium comprised (per L) 5 g of K₂HPO₄, 0.5 g of KH₂PO₄, 1 g of MgSO₄, and 2 ml of trace element solution. SJ (previously diluted to the designated total sugar concentration) and (NH₄)₂SO₄ were used as the carbon and nitrogen sources, respectively. The trace element solution consisted of (per 1 L of 1 N HCl) 10 g of CaCl₂·2H₂O, 4.98 g of FeSO₄·7H₂O, 0.23 g of ZnCl₂, 0.78 g of CuSO₄·5H₂O, 0.24 g of Na₂MoO₄·2H₂O, and 0.61 g of MnSO₄·H₂O. The SJ, MgSO₄, and trace element solution were sterilized separately at 121°C for 15 min, and aseptically mixed after being cooled to room temperature.

Effects of carbon to nitrogen (C/N) ratio and initial sugar concentration were investigated in 500-mL Erlenmeyer flasks, with a working volume of 100 mL, using one-factor-at-a-time approach. In C/N ratio optimization, concentrated SJ was diluted with distilled water to a total sugar concentration of 30 g/L. Then, the diluted SJ was supplemented with minerals according to the mineral medium formula, and added with calculated amounts of (NH₄)₂SO₄ to adjust the C/N ratios to 4 to 53 (by mass). The C/N ratio of four was selected based on the elemental compositions of Paracoccus sp. (CH_1.8O_0.48N_0.25) reported by Van Verseveld and Stouthamer (1978), and the ratio of 53 was obtained from the ratio between carbon and nitrogen in the mineral medium used to grow Paracoccus denitrificans (Kalaiyezhini and Ramachandran, 2015). The medium was autoclaved at 121°C for 15 min, and mixed with sterile MgSO₄ and trace element solutions before use. The fermentation was carried out at 30°C, pH 6.9, and 150 rpm, using an initial cell concentration of 0.2 unit of optical density at 600 nm (OD₆₀₀). Effects of the initial sugar concentration in a range 30–150 g/L was subsequently examined using the optimum C/N ratio. The cultivation conditions were the same as described above. Samples were taken to follow the bacterial growth (in terms of OD₆₀₀), and the consumption of total sugar during the fermentation. The experiments were conducted in triplicate, and the mean and standard deviation of the mean are reported.

Using the optimum C/N ratio and initial total sugar concentration, effects of dissolved oxygen (DO) level on growth of Paracoccus sp. KKU01 was examined. The fermentation was performed in batch mode in a 2-L bioreactor (Bioneer-neo-2L, B.E. Marubishi, Thailand), with a working volume of 1 L and an aeration rate of one vvm. The DO set point was varied in a range 10–50% of the oxygen saturation value. DO concentration was maintained at the designated level by automatic control of the agitation rate in the range 300 to 1,500 rpm. The bioreactor containing sterile medium was inoculated with Paracoccus sp. culture to an initial OD₆₀₀ of 0.2 unit, and the fermentation was carried out at 30 ± 2°C, and pH 6.9 ± 0.2 (controlled using 10 N NaOH and 10 N HCl). Samples were taken regularly for the determination of cell dry mass (CDM) and total sugar concentrations.

Fed-batch fermentation was carried out in the 2-L bioreactor with an initial volume of 0.8 L. The fermentation temperature, pH, and DO set point were controlled at 30 ± 2°C, 6.9 ± 0.2, and 20%, respectively. The experiment was started as a batch fermentation with the optimum initial total sugar concentration (100 g/L), and was allowed to proceed until the sugar concentration decreased to below 30 g/L. Then, calculated amount of fresh medium (SJ that was previously supplemented with (NH₄)₂SO₄ to attain a C/N ratio of 10) was fed into the bioreactor to increase the sugar concentration to around 100 g/L. The feeding of the medium was performed twice. The final volume of the medium was approximately 1 L. Samples were taken at 6-h intervals for the determination of CDM, total sugar, ammonia-nitrogen (NH₃-N) and PHB concentrations. This experiment was conducted in duplicate.

Economic yield (EY), defined as monetary unit (USD) of PHB produced per monetary unit (USD) spent in nutrients, was calculated using Eq. 1, where Y is the concentration of PHB (kg/L), C is the cost of each nutrient used in medium preparation (USD/kg), N is concentration of each nutrient (kg/L), and I is the selling price of PHB (USD/kg) (Bustos et al., 2004; Ma et al., 2021). It should be noted that the price of raw sugar (0.14 USD/kg) (Soccol et al., 2017) was used in the calculation due to the unavailability of concentrated SJ price in the literature. The selling price of PHB (3.50 USD/kg) was taken from Salakkam and Webb (2018), while the prices of other chemicals were taken from www.alfa.com (Supplementary Table S1). E Y = Y ∑ ( C N ) I (1)

Cell growth in terms of OD₆₀₀ was measured using a UV-VIS spectrophotometer (UVmini-1240, Shimadzu, Japan). CDM (g/L) was measured gravimetrically after washing the cells twice with distilled water, and drying at 80°C until a constant weight was reached. Residual cell dry mass (RCDM) concentration was calculated as the difference between CDM and PHB concentrations. Total sugar concentration was determined using the phenol-sulfuric method (Dubois et al., 1956), using sucrose as the standard. Total sugar consumption rate at time interval was calculated by dividing total sugar consumed (g/L) by fermentation time (h). Sugar compositions of fermentation broth were determined using high performance liquid chromatography (HPLC) following the method of Khamtib and Reungsang (2012). Specific growth rate of the bacterium was estimated as the largest slope of a log-linear plot between OD₆₀₀ and time. NH₃-N concentration was determined using the phenate method (Baird et al., 2017), using ammonium sulfate as the standard. Polymer identification was conducted using Fourier transform infrared spectroscopy (FTIR) analysis at the Faculty of Associated Medical Sciences, Khon Kaen University, Thailand, using commercial poly (3-hydroxybutyrate) as the standard (Supplementary Figure S1). The PHB standard was purchased from Sigma-Aldrich (Cat. no. STBB9669V). Extraction of PHB from the biomass for the identification was performed using the method of Hahn et al. (1995). PHB concentration was determined by HPLC (Waters, United States), equipped with an Aminex HPX-87H column (Bio-Rad, United States) and a UV detector, using the conditions reported by Fernandes et al. (2011). The commercial PHB was used as the external standard. PHB production rate at time interval was calculated by dividing PHB concentration (g/L) by fermentation time (h). Kinetic parameters for biomass growth and PHB production were obtained by curve fitting using the modified Gompertz model (Eq. 2), where Y _m is the maximum CDM, RCDM, and PHB concentration (g/L), R _m is the maximum rate of CDM, RCDM, and PHB production (g/(L·h)), e is the Euler’s number (2.7183), λ is the lag time for the production of CDM, RCDM, and PHB (h), and t is the fermentation time (h). The Solver function of Microsoft Excel 2019 was used for curve fitting and estimation of the kinetic parameter values. Y = Y m × exp [ − exp [ R m × e Y m ( λ − t ) + 1 ] ] (2)

To achieve a high cell density cultivation, it is very important that the fermentation conditions are adjusted to support the best cell growth. It is well known that carbon and nitrogen are essential nutrients for microbial growth and cellular components synthesis, and the ratio of which directly affects the growth and functions of microorganisms. In the present study, C/N ratio of the fermentation medium was optimized to examine a proper balance between carbon and nitrogen sources. Growth curves of the bacterium at all the C/N ratios tested revealed that the bacterium grew rapidly without observable lag phase upon the inoculation. The exponential growth phase lasted for around 9–12 h (Figure 1A). Calculation of the specific growth rate (µ) during the first 9 h revealed that the bacterium grew equally well at all the C/N ratios tested; the value of µ varied in a narrow range of 0.424–0.484 1/h (Table 2). This implied that, at all the C/N ratios tested, nitrogen was sufficient for the bacterial multiplication during the early growth phase. However, after 9 h, the growth at C/N ratio of 53 started to be lower than that at other C/N ratios. At the C/N ratio of 53, with the initial total sugar concentration of 26.7 ± 1.0 g/L (equivalent to 11.23 g-carbon/L, assuming that all the sugar in SJ was sucrose), only ca. 1.0 g/L of (NH₄)₂SO₄ was used in the medium (equivalent to 0.21 g-nitrogen/L). It could, therefore, be possible that most of the nitrogen was consumed during the exponential phase, resulting in insufficient nitrogen for the later stage of growth. As for the growths at other C/N ratios, both the µ and maximum OD₆₀₀ values were very similar. It was initially presumed that C/N ratio of four would be suitable for cell growth since the C/N ratio of Paracoccus sp. has been reported to be around 4 (Van Verseveld and Stouthamer, 1978). Nevertheless, good growth could still be achieved at C/N ratios up to 30. Similar results have been reported by Sawant et al. (2015) that growths of Paracoccus sp. LL1 during the first 36 h were more or less the same at C/N ratios between 6.7 and 20, and further increasing the ratio to 40 led to a large drop of cell growth.

Determination of total sugar concentration revealed that sugar was rapidly consumed when the cells were grown at C/N ratios of 4 and 10, and this decreased with increasing C/N ratio up to 53 (Figure 1B and Table 2). It is well known that nitrogen is the macronutrient essential in the syntheses of various cellular components, e.g., proteins, enzymes, DNA, and RNA (Zhou et al., 2007; Kampen, 2014; Kuypers et al., 2018). Therefore, the low availability of nitrogen at high C/N ratios could possibly lead to low microbial metabolic activity. Based on the results of the bacterial growth and sugar consumption (Figure 1), C/N ratio of 10 was considered optimum for growth of Paracoccus sp. KKU01, as at this C/N ratio, lesser amount of nitrogen source ((NH₄)₂SO₄) was used compared with C/N ratio of 4. This was consistent with a report of Sawant et al. (2015) that C/N ratio of 13 was optimum for growth of Paracoccus sp. LL1.

Using the optimum C/N ratio of 10, Paracoccus sp. KKU01 was grown at initial sugar concentrations of 30–150 g/L. Figure 2A and Table 3 reveal that the final biomass concentration increased with increasing initial sugar concentration. However, the specific growth rate tended to decrease. The specific growth rate decreased by 74% when the initial sugar concentration was increased from 30 to 150 g/L. Likewise, biomass productivity and yield decreased by 36 and 39%, respectively. The lower growth observed at high initial sugar concentrations could be a result of increased osmotic pressures at higher solutes [sugars and (NH₄)₂SO₄] concentrations, which can suppress cell growth (Sawant et al., 2015). The sugar consumption profile shown in Figure 2B reveal that the bacterium could completely consume sugar when grown at the initial sugar concentration as high as 120 g/L, while further increasing the initial sugar concentration to 150 g/L resulted in around 18.96 g/L of sugar remaining after the cultivation. Based on the results of biomass concentration, biomass yield, and sugar consumption rate, the initial sugar concentration of 100 g/L was selected as the suitable concentration for cell growth.

Using the C/N ratio and initial sugar concentration of 10 and 100 g/L, respectively, growth of the bacterium was examined at DO set point ranging from 10 to 50% of the oxygen saturation level in the medium. The bacterial growth tended to increase as a function of DO set point (Figure 3A). However, µ only increased, from 0.220 to 0.328 1/h, when the DO set point was raised to 40%. At 50% DO set point, a large drop of µ to 0.169 1/h was observed (Figure 3B). This was considered due to the toxicity of oxygen at too high concentration. Although Paracoccus sp. is able to grow aerobically, the presence of oxygen at too high concentrations could lead to the formation of reactive oxygen species (ROS), e.g., superoxide (O₂ ⁻) and hydrogen peroxide (H₂O₂), inside the cells. These ROS can be accumulated to a concentration that overwhelms the antioxidant defenses of the cells, leading to cell inactivation. Some enzymes, particularly dehydratases and metalloenzymes, have also been reported to be inactivated by the presence of superoxide in the system (Baez and Shiloach, 2014).

Trend of sugar utilization correlated well with the bacterial growth. Increasing the DO set point resulted in higher rate of total sugar consumption (Figure 3C). At the DO setpoints of 10–30%, sugar was depleted at 30 h, while at 40% and 50% DO setpoints, more rapid sugar consumptions were observed, with the depletion of total sugar recorded at 24 h. Growth parameters of the bacterium at different DO setpoints are summarized in Table 4. As can be seen in the table, the bacterium consumed almost all the sugar (over 95%) in the system within 24–30 h, and the final biomass concentration and biomass yield were more or less the same at all the DO set points tested. It should be noted that the biomass productivity was higher at 40% and 50% DO setpoints because these were calculated using the time at which the total sugar was completely consumed. Considering that oxygen is inhibitory at 50% DO set point, and using 30% and 40% DO setpoint could lead to high operating costs, while using 10% DO set points led to low total sugar consumption rate, DO set point at 20% was selected for the subsequent experiments. The use of 20% DO setpoint was in accordance with previous studies using ≥20% DO setpoint to grow Paracoccus sp. for PHAs production (Kalaiyezhini and Ramachandran, 2015; Sawant et al., 2015; Kumar et al., 2018; Kumar and Kim, 2019).

Using the C/N ratio, initial sugar concentration, and DO set point of 10, 100 g/L, and 20%, respectively, Paracoccus sp. KKU01 was cultivated in batch mode until total sugar concentration decreased to below 30 g/L, then fresh medium was fed into the bioreactor to start the fed-batch cultivation. Figure 4A shows that the biomass concentration increased slowly during the first 12 h. After that, the cells grew rapidly, reaching 51.2 g/L at 22.5 h, corresponding to the large drops of total sugar and NH₃-N concentrations to 31.7 and 0.07 g/L, respectively (Figure 4B). PHB was detected at as early as 12 h at 0.2 g/L, indicating the commencement of the PHB accumulation phase. Accordingly, fresh medium was fed into the bioreactor at 22.5 h to provide carbon source for PHB synthesis. It is noteworthy that although nitrogen deficiency is preferable for PHB production, complete deficiency of nitrogen might not always be supportive for PHB production as microbial PHB synthetic activities could be damaged (Suzuki et al., 1986). Therefore, nitrogen should be provided in the feed to support the catalytic activity and maintenance of the cells (Khanna and Srivastava, 2006; Norhafini et al., 2019). For this reason, the present study used fresh medium (SJ supplemented with (NH₄)₂SO₄) as the feeding solution. Provision of nitrogen source in the feed has been shown to be beneficial for PHB production in several PHB producers, e.g., Pseudomonas extorquens sp. K. (Suzuki et al., 1986), Ralstonia eutropha NRRLB14690 (Khanna and Srivastava, 2006), Wautersia eutropha NRRLB14690 (Khanna and Srivastava, 2008), Cupriavidus malaysiensis USMAA1020 (Norhafini et al., 2019), Halomonas venusta (Stanley et al., 2018), and Paracoccus sp. LL1 (Khomlaem et al., 2021). Noted that R. eutropha and W. eutropha are now known as Cupriavidus necator. Following the feeding, total sugar and NH_3-N concentrations were 111.5 g/L, and 1.3 g/L, respectively. NH₃-N was depleted shortly after the feeding, implying that nitrogen was required by the cells even if these were in the PHB accumulation phase. With the depletion of NH₃-N, excess carbon and limited nitrogen conditions were achieved, which are favorable for PHB synthesis by Paracoccus sp. (Kojima et al., 2004). Vigorous PHB accumulation proceeded until 48 h, attaining 28.1 g/L of PHB. At which time, total sugar and NH₃-N concentration reduced to 27.3 and 0 g/L, respectively. The second feeding was performed at 60 h when total reducing sugar concentration decreased to 15.2 g/L, which increased the total sugar and NH₃-N concentration to 107.5 and 1.5 g/L, respectively. However, neither significant biomass growth nor PHB production were observed after the second feeding. At the end of the fermentation, the CDM, RCDM, and PHB concentration were 87.9, 56.5, and 32.1 g/L, respectively. The PHB content in the cell was 36.2% and PHB productivity was 0.27 g/(L·h).

Profiles of CDM, RCDM, and PHB concentrations were fitted well with the modified Gompertz model (Eq. 2), which is commonly used to predict population growth and bioproduct formation (Zeb et al., 2019). The kinetic parameters for CDM, RCDM, and PHB, are given in Table 5. Based on the curve fittings (Figure 4A), it can be seen that RCDM (non-PHB biomass) concentration increased rapidly during 12–22.5 h, coinciding with the rapid sugar consumption before the first feeding of fresh medium, then the concentration stayed relatively constant until the end of the fermentation. The total sugar consumption rate shown in Figure 4C confirmed that the sugar was mainly consumed for biomass growth during this period. On the other hand, vigorous PHB production occurred after 18 h and lasted until 48 h. The PHB production rate decreased afterward and no further production was observed until the end of the fermentation. The profile of PHB production rate (Figure 4C) confirmed that most of the sugar present the bioreactor after the first feeding was used for PHB synthesis. Analysis of the sugar compositions in the fermentation broth at different times during the process, i.e., 0 h, 22.5 h (before and after the first feeding), 60 h (before and after the second feeding), and 120 h, revealed that the bacterium utilized sucrose, glucose, and fructose for growth and PHB production (Figure 5). However, it was interesting that there were only little biomass growth and PHB production after the second feeding at 60 h, yet the sugar was continually consumed. This was possibly due to the utilization of the sugar in other metabolic pathways for cell maintenance and for syntheses of other products. Astaxanthin, for instance, could be produced simultaneously with PHAs by Paracoccus sp. According to a report of Khomlaem et al., 2021, up to 8.51 ± 0.20 and 10.2 ± 0.24 mg/L of intracellular and extracellular astaxanthin were observed during a continuous fermentation of Paracoccus sp. LL1 in a membrane bioreactor. Additionally, secondary metabolites such as polyketides, bacteriocins, and non-ribosomal peptides might also be produced (Leinberger et al., 2021). Since the production of PHB was not observed after 60 h (Figure 4A), the process could be terminated at this time, which would improve the PHB productivity to 0.49 g/(L·h).

With the total sugar consumption of 251.6 g/L and the production of 32.1 g-PHB/L, a PHB yield was calculated to be 0.13 g-PHB/g-sugar. This was around 26% of the biochemical stoichiometric yield 0.5 g-PHB/g-sucrose reported by Yamane (1993). Based on the results of total sugar consumption, it was also possible to estimate the quantity of carbon utilized for biomass growth and PHB production, using an approach similar to that proposed by Bormann and Roth (1999). Since the elemental formula for Paracoccus sp. is CH_1.8O_0.48N_0.25 (molecular weight 25.03 g/mol) (Van Verseveld and Stouthamer, 1978; Araujo-Granda et al., 2020), and the monomer of PHB is C₄H₆O₂ (molecular weight 86 g/mol) (Yamane, 1993), carbon accounts for 47.9% and 55.8% of the biomass and PHB, respectively. With 56.5 g/L of RCDM and 32.1 g/L of PHB, the carbon in RCDM and PHB were calculated to be 27.1 and 17.9 g/L, respectively. This demonstrated that around 45 g/L of carbon, equivalent to 42.5% of the total carbon consumed, could be referred to RCDM and PHB. The elemental formula for Paracoccus sp. KKU01 accumulating 36.6% PHB could also be calculated as seen in Eq. 3. ( 56.5 25.03 ) CH 1 . 8 O 0 . 48 N 0 . 25 + ( 32.1 86 ) C 4 H 6 O 2 → C 3 . 75 H 6 . 3 O 1 . 84 N 0 . 56 (3)

The key results for PHB production from the present study along with those from recent publications are shown in Table 6. It can be seen that PHAs production varies depending on the microbial strain, substrate, and fermentation mode and conditions. It is obvious that Cupriavidus necator could be grown to high cell concentrations (148–164 g/L) and produced PHB at high concentrations (109–125 g/L) and productivities [1.5–3.1 g/(L·h)]. Nevertheless, Paracoccus sp. LL1 could also be grown to a high cell density (113 g/L) using a continuous fermentation process with a cell retention system (Khomlaem et al., 2021), though PHB production and productivity were lower. In the present study, Paracoccus sp. KKU01 was grown to a high cell density of 87.9 g/L and produced 32.1 g/L of PHB. These figures, although not the highest, were in the upper range among the results compared, suggesting that Paracoccus sp. KKU01 is a potential PHB producer from SJ. Better still, the bacterium may be used with other low-cost feedstocks containing sucrose, e.g., cane molasses (Palmonari et al., 2020) and fruit wastes (Choi et al., 2015; Nanda et al., 2016), which would be beneficial in reducing raw material cost. However, despite the promising results, the medium-based EY calculation revealed that the process used in the present study gave a low EY of 0.14, indicating that the revenue of PHB production is less than the cost of the medium. It is therefore necessary to further optimize the process to increase PHB production and to reduce the cost of medium. Increasing PHB production may be achieved by precise control of oxygen supply during the process (Blunt et al., 2018), by the use of genetic engineering to overexpress the pha genes involved in PHB synthesis (Khetkorn et al., 2016), or by using an effective process configuration such as multi-stage chemostat continuous fermentation (McAdam et al., 2020). The use of nutrients limitation strategy could also be further investigated with other nutrients. The limitation of phosphorus in the medium has been demonstrated to be effective in enhancing PHAs production in C. necator DSM 545 (Grousseau et al., 2014), while this improved the growth of Burkholderia sacchari LMG19450 on xylose (Oliveira-Filho et al., 2020). Double nutrients limitation, e.g., nitrogen-oxygen limitation, might also be applicable as this has been shown to enhance PHAs production in Halomonas boliviensis (Garcia-Torreiro et al., 2016). However, the expenditure of the process must be taken into consideration. Optimization of the feed composition, as well as feeding strategy, could also be performed to further enhance the PHB production. On the other hand, since the EY calculation showed that the cost of K₂HPO₄ accounted for over 80% of the total medium cost, medium optimization is worth the investigation. SJ already contains appreciable concentrations of potassium and phosphorus as shown in Table 1 and reported by De Souza et al. (2015). Therefore, it might be possible to reduce the concentration of K₂HPO₄ in the medium to reduce cost of medium preparation.