Technical-Economical Approach for PHB Production by Ralstonia Eutropha Strain Using Concentrated Vinasse as Carbon Source and Other Biotechnological Applications


 The Brazilian ethanol industry is one of the most important in the global market, however these important industrial activities have been generating significant amounts of vinasse and its management has become costly for distilleries. In this study, the aim was to evaluate concentrated and in natura vinasse as basal culture media for biotechnological processes. Different bacteria and processes were assessed: L-threonine production by E. coli THR14, with glucose as carbon source; PHB production by halophilic strain Halomonas sp. HG03, with sucrose as carbon source; and PHB biosynthesis by R. eutropha L359PCJ, which used glycerol from vinasse as carbon source. Strains were evaluated firstly in shake flasks cultivations using vinasse-based media. E. coli THR14 had no statistical difference for biomass and L-threonine concentrations among control and vinasse-based treatments (up to 50% v v-1 of in natura vinasse). Halomonas sp. HG03 and R. eutropha L359PCJ were cultivated in mineral media diluted by in natura (50% and 75% v v-1) and concentrated (50% and 75% v v-1) vinasses. Higher vinasse concentrations resulted in higher cellular growth rather than PHB accumulation for both bacteria. In vinasse-based treatments, Halomonas sp. HG03 had PHB content between 19.6 – 75.2% and R. eutropha L359PCJ, 48.4 – 68.5%. 50% (v v-1) of concentrated vinasse was the most attractive condition for PHB production by both bacteria. Further experiments in CSTR bioreactors used this nutritional condition and R. eutropha L359PCJ had PHB content of 66.3%, concentrations of residual cell dry weight (rCDW) = 9.4 g L-1 and PHB = 18.6 g L-1, with YX/S = 0.16 g gGLYCEROL-1, YP/S = 0.32 g gGLYCEROL-1 and 0.25 gPHB Lh-1. Halomonas sp. HG03 had PHB content of 45.7%, rCDW = 9.8 g L-1, PHB = 8.3 g L-1 and YX/S = 0.18 g gSUCROSE-1, YP/S = 0.16 g gSUCROSE-1 and 0.12 gPHB Lh-1. Finally, cost reductions of PHB production by R. eutropha L359PCJ with concentrated vinasse-based medium were evaluated in silico by using SuperPro Designer. As a partial source of glycerol and other nutrients for PHB production by R. eutropha L359PCJ, vinasse reduced overall production costs by 13%. Simulated processes that used concentrated vinasse-based media combined with improvements of PHB productivity and higher cellular densities had production costs between US$ 3.9 – 7.5/kgPHB and 2.6 – 7.3 years of payback time.


31
Global demand for green fuels has increased in recent years and it is expected to increase even 32 further. Bioethanol is one of the most important among these, along with biodiesel, biogas and 33 others. Together, the United States and Brazil account for about 85% of the global ethanol supply.

34
Brazil has a sucrose-based industry, established on sugarcane exploit by using juice, molasse

38
Vinasse is the main wastewater produced by Sucroenergetic activities as it is the remaining 39 fermented broth after distillation. All residual organic matter, metabolites and nutrients from 40 fermented broth are found in vinasse, except for ethanol, which is efficiently recovered. About 10-41 15 L of vinasse are generated along with 1 L of ethanol, which implies an annual vinasse supply 42 of over 300 billion L. Its composition is dependent on the ethanol process, so different sorts and 43 quality of feedstocks, as well as process conditions and operations affect vinasse characteristics 44 (Cortez, 2010).

45
In general, vinasses have very high polluting potential (COD between 21,000 -34,000 mgO2 L -1 ) 46 and they are a source of carbon compounds and salts: glycerol has been frequently described to    Vinasse was obtained from a distillery in São Paulo State, Brazil, which produces ethanol from 94 sugarcane molasses. In our laboratory, vinasse was concentrated to 34.4ºBrix and kept at 4ºC.

95
In order to apply vinasse in chemical analyses and experiments, the highly concentrated material 96 was diluted in distilled water (m m -1 ) so final concentrations could be obtained. In this study, we 97 named in natura vinasse the material diluted to 3.4ºBrix, simulating the concentration found in the 98 output stream of bioethanol process.

99
Before all analytical procedures and shake flasks experiments, vinasses were filtered in vacuum 100 filtration system, which consisted of a Kitassato flask, Büchner funnel and qualitative paper filter.

101
Prior to all experiments, vinasses had pH set to 6.8 < pH < 7.8 with NaOH 10 M or NH4OH 4 M 102 solutions, followed by autoclave sterilization, and then vinasses were added to sterile culture 103 media. In natura vinasse had pH = 4.3 and its chemical composition is presented in Table 1 For L-threonine production experiments, it was utilized the strain E. coli MG1655 ΔmetJIQ ΔlysA All microorganisms were kept in glycerol solution (20% v v -1 ) at -80ºC. Before experiments, they 119 were all cultivated in their respective solid culture media, kept at 30ºC for 24-36 h and then a 120 loopful from a single colony was inoculated into fresh liquid medium for inoculum preparation.

181
LB culture medium was used for R. eutropha L359PCJ culture in solid medium and for inoculum 182 preparation, which was carried with the same procedures described for Halomonas sp. HG03.

183
The mineral medium MMRe was adapted from Rocha et al. (2008) and it had the same composition

208
All experiments in shake flasks had treatments conducted in triplicates and analytical procedures 209 were carried out in duplicates. Data were analyzed using ANOVA and means were compared 210 with Tukey's test (5% probability), by using ExpDes.pt package in RStudio software (Ferreira,

216
Bioprocesses were carried on in consecutive batches: two initial batches had growth-inducing 217 conditions, followed by batches that induced nutritional conditions for PHB accumulation. The aim 218 was to achieve higher concentration of rCDW before imposing nitrogen limitation for PHB 219 accumulation. Besides, we aimed to minimize any osmotic stress in growth phases that could be 220 a result of higher concentrations of carbon source, salts and vinasse.

221
Inoculum preparation followed the same procedures described above for shake flasks 222 fermentations. Both R. eutropha L359PCJ and Halomonas HG03 strains were cultivated with 223 initial volume of 6 L, inoculum ratio of 10% (v v -1 ) and nutritional conditions from MMReV507 and 224 MMHaV507 treatments were chosen for PHB production in bioreactors.

225
Bacteria were cultivated in culture media composed of mineral medium MMB, adapted from 226 Rocha et al. (2008), and a volumetric dilution of 50% of concentrated vinasse (7ºBrix) was used 227 (

232
Similarly to flask cultivations with R. eutropha L359PCJ, glycerol was partially provided by vinasse 233 (7.5 g L -1 ), and a concentrated solution of pure glycerol (300 g L -1 ) was used to set a final 234 concentration of 15 g L -1 in all batch cultivations.

235
Once the carbon source reached limiting concentration, below 5 g L -1 , the first batch was 236 considered finished. The second batch was initiated by adding a concentrated solution (0.7 L), so 237 nutrients from mineral medium could be set to concentrations described in Table 2. A 238 concentrated solution of pure glycerol and highly concentrated vinasse (34.4ºBrix) provided the 239 final concentration of 15 g L -1 of glycerol (1:1). Finally, working volume was 6.7 L.

240
Once carbon source in the second batch became limiting, further batches aiming to PHB 241 accumulation were initiated by adding concentrated vinasse (34.4ºBrix) and pure glycerol solution 242 (300 g L -1 ) as the sole nutrients source. These materials were added in variable volumes, so the 243 final concentrations of glycerol from vinasse (7.5 g L -1 ) and pure glycerol (7.5 g L -1 ) were kept the 244 same for all batches. PHB accumulation phase was ceased once the culture reached stationary 245 growth phase.

246
Halomonas sp. HG03 used sucrose as carbon source, which was fully provided by adding a 247 concentrated solution (600 g L -1 ) so final concentration could be set to 15 g L -1 .

248
Once sucrose reached limiting concentration, below 5 g L -1 , the first batch was finished and the 249 second batch was carried on similarly to the procedure described for R. eutropha L359PCJ. A 250 concentrated solution (0.7 L) containing nutrients from mineral medium (Table 2), a concentrated 251 solution of sucrose (600 g L -1 ) and highly concentrated vinasse (34.4ºBrix) were added. Glycerol 252 from vinasse was not the preferable carbon source of Halomonas sp. HG03, but its concentration 253 in culture medium (7.5 g L -1 ) was efficiently used as an indicator of vinasse dilution in bioreactor, 254 so the volumetric dilution of 50% could be kept the same for all batches.

298
Vinasse refractive index was determined as Brix Degrees (ºBrix), by using digital refractometer.

306
In E. coli THR14 cultures, cell dry weight was determined by gravimetric method. Culture media 307 were centrifuged at 10,000 g, for 10 minutes, at 10ºC. Pellets were then resuspended in distilled

315
In Halomonas sp. HG03 and R. eutropha L359PCJ cultivations, cell dry weight was determined 316 by lyophilization of cells, so PHB extraction could be performed next. Sample volumes of 10 mL 317 were centrifuged, resuspended in saline solution (0.85% m v -1 ), next washed in the same solution 318 and centrifuged. Remaining pellets were then submitted to lyophilization overnight. Resulting dry 319 mass was measured and cell dry weight was calculated in order to determine total biomass 320 concentration (CDW g L -1 ).

321
The same dry mass was used for PHB extraction method by propanolysis (Riis & Mai, 1988).

328
PHB cellular content was determined as the ratio of PHB concentration in total CDW. Residual 329 cell dry weight concentration (rCDW) was then obtained by subtracting PHB mass from CDW.

330
Even though vinasses were previously filtrated, specific analyses were performed in order to

343
A base scenario was designed and named SC1. The bioprocess flowsheet is detailed in Figure   344 1.

353
Economical evaluations aimed to estimate production costs that could be reduced by using 354 vinasse as a partial source of glycerol and salts for PHB production by R. eutropha L359PCJ.

355
Technical-economical scenarios were simulated in order to analyze how PHB productivity and 356 cell density might impact on production costs and profitability indicators.

357
Experimental data were previously converted into input data for SuperPro Designer simulation 358 model as described below.

360
were used in global stoichiometric equations used in both unit procedures seed fermenter SFR-361 101 (R1) and production bioreactor FR-101 (R2). All mineral medium components were 362 normalized to glycerol (Table 3).
Equation 5 was used to calculate the term (tf -ti). Experimental data of μmáx (h -1 ) and Xi (g L -1 ) 370 were used. The term ti was set to t = 0 h; the term Xf simulated high cell densities, named as Xf-

387
Finally, the simulated global productivities (PpSi g Lh -1 ) were calculated as it follows: Firstly, the annual production capacity was 10,000 tons of PHB. By taking into consideration that 396 vinasse would provide 50% of carbon source, such production capacity would be compatible to 397 vinasse production by a distillery that produces 295,245 -443,000 m 3 of ethanol annually (10 -15 398 L of vinasse for 1 L of ethanol), providing 44,300 tons of glycerol annually (in natura vinasse with 399 10 g L -1 of glycerol, Table 1).

400
Secondly, the annual campaign had 5760 operating hours, which is compatible to the operating 401 hours in bioethanol campaign and sugarcane harvest (240 days year -1 ) so vinasse supply would 402 be guaranteed.

403
These two requirements defined major aspects of model design, however the base scenario SC1

408
The simulation model was set to batch mode and it consisted of three sections: Upstream,

409
Bioprocess and Downstream ( Figure 1). The Upstream section included culture medium 410 preparation and seed fermentation for inoculum production.

411
A blending tank was set to be charged with medium components, followed by in situ sterilization.

491
Both L-threonine (YP/S g g -1 ) and biomass yields from glucose (YX/S g g -1 ) showed no significant 492 difference between MMEc, MMEcV253 and MMEcV503. MMEcV753 showed the lowest biomass 493 production, but glucose consumption was consistently low (9.7 g L -1 of residual glucose) and 494 resulting biomass yield did not differ from those in other treatments (Table 5).  In this study, biomass yields were consistent with those previously reported on E. coli yield on 506 glucose (0.5 gbiomass gglucose -1 ) (Shiloach & Fas, 2005), and no statistical difference was found 507 among treatments. Furthermore, MMEcV253 and MMEcV503 treatments had significantly lower 508 acetic acid production than control, which made these vinasse-based treatments interesting for 509 E. coli THR14 cultivation.

510
In bioprocesses with E. coli strains using glucose as carbon source, acetic acid synthesis may 511 lead to longer processes and consequently lower productivities. Glucose is assimilated and

517
There is a large concern about acetate production in bioprocesses by E. coli strains. In our study,

522
(0.5 ± 0.2 g L -1 ) were, respectively, 75% and 79% lower than that in control treatment. These 523 results indicate that those treatments provided balanced nutritional conditions for E. coli THR14 524 metabolism.

525
As for MMEcV753 treatment, results indicate that low efficiency was most likely due to nutritional 526 limitation rather than inhibition. As stated in Material and Methods section, our experiments aimed 527 to supply all macronutrients that are known to limit cellular growth. However, micronutrients from 528 the mineral medium were diluted by vinasse.

529
Since biomass yield in MMEcV753 treatment was satisfactory, it is possible that as long as cells 530 had favorable nutritional conditions for growth, they were able to consume glucose for biomass 531 production. Once the culture reached some possible nutritional limitation, cellular growth ceased 532 and L-threonine was not favored. Acetic acid concentration in MMEcV753 supports the hypothesis 533 of limitation rather than inhibition. In MMEcV753 treatment, glucose uptake by E. coli THR14 did 534 occur and acetyl-CoA was synthesized and driven to acetic acid pathway. The metabolic switch 535 from cellular growth to acetic acid synthesis might have occurred once nutritional conditions 536 started to limit biomass synthesis due to lacking (micro) nutrients, so available acetyl-CoA was 537 driven towards acetic acid pathway.

538
In this study, L-threonine production was below some results described by other authors. In

542
Ours results indicate that much can improved in the L-threonine-producing strain E. coli THR14.

546
This study, however, aimed to investigate the potential of using vinasse for L-threonine 547 production. To our knowledge, there is no available study about amino acids production using

559
In L-threonine production by E. coli THR14, vinasse was a source of water and mineral salts,

635
However, higher rCDW concentration in MMReV757 contributed to its final PHB concentration (3.9 636 ± 0.3 g L -1 ), which differed only from control treatment and MMReV503.

637
In shake flasks cultures, using glycerol as carbon source, R. eutropha strains were described by 638 other authors with residual biomass production between 0.66 -2.2 g L -1 , and PHB cellular content

643
Since PHB biosynthesis is not growth related, as substrate conversion into biomass increased, 644 substrate conversion into product decreased.

668
Concerning treatments with 75% (v v -1 ) of concentrated vinasse, the unsatisfactory results for 669 PHB production suggest that nutritional conditions favored cellular growth and they were not 670 unbalanced enough for PHB accumulation, instead, these treatments provided the most efficient 671 cellular growth conditions in our study.

672
Similarly to L-threonine production by E. coli THR14, processes with Halomonas sp. HG03 673 exploited vinasse mostly as a source of water, salts, residual amino acids and an external carbon 674 source had to be supplemented.

697
In our study we also had interest on vinasse as carbon source, so the glycerol-consuming R.

774
Sucrose was satisfactorily consumed until t = 48 h, but sucrose accumulation was observed later.

819
By the end of the growing phase (t = 32 h), conversion yield of glycerol into rCDW was YX/S = 0.52 820 g g -1 , and into PHB, YP/S = 0.14 g g -1 . At the end of the seventh batch the global yields were YX/S 821 = 0.16 g g -1 , YP/S = 0.32 g g -1 and the final concentration of 18.6 g L -1 PHB was reached after 74 822 h, which resulted into PHB productivity of 0.25 g Lh -1 .

825
Thus, the PHB production by R. eutropha L359PCJ with concentrated vinasse-based medium in 826 bioreactor had conversion yields consistent with those determined in shake flasks fermentations 827 and PHB content satisfactorily improved.

828
The production results were also consistent with those described by other authors with glycerol 829 as carbon source, which supports the use of concentrated vinasse as glycerol source for PHB

852
In comparison to results obtained in shake flasks (MMHaV507 treatment), the consecutive batches 853 operation mode was not successful in achieving higher PHB content by Halomonas sp. HG03, 854 which had been 50.7 ± 6.9%. As mentioned above, further investigation on Halomonas sp. HG03 855 nutritional requirement is needed and such information could be valuable for improving 856 bioprocess strategies for PHB production in bioreactors with concentrated-vinasse medium.

857
PHB production by Halomonas sp. HG03 indeed was less attractive than that showed by R.

867
Other authors have also reported relatively important PHB production by Halomonas sp. strains 868 during growth phases. Besides, in bioprocesses that used complex media, such as hydrolysates, 869 the authors have also described that rCDW continued to increase despite the depletion of the

881
In silico study of PHB production process by R. eutropha L359PCJ with concentrated vinasse-882 based culture medium 883 884 The model process considered a seed fermenter with reaction described reation R1 (Table 3) and 885 the glycerol conversion yields determined experimentally during growth phase were used (YX/S = 886 0.52 g g -1 ; YP/S = 0.14 g g -1 ).

887
Settings in the production bioreactor considered initial biomass (rCDW) of 0.3 g L -1 ; PHB 888 accumulation phase would require the increase from 21.4% PHB up to 66.3% (m m -1 ); and 889 glycerol conversion yields applied in R2 (Table 3) were YX/S = 0.16 g g -1 and YP/S = 0.32 g g -1 , as 890 obtained experimentally. In order to determine the time needed for growth phase and PHB 891 accumulation phase, the experimental data µmax = 0.255 h -1 and µP = 0.05 g gh -1 (t = 57 h) were 892 used. These assumptions about growth phase and PHB accumulation phase were considered 893 during calculations of simulation parameters, however, settings in SuperPro Designer treated 894 these parameters globally, as detailed in Table 8.

906
907 Leong et al. (2017) also analyzed in silico the PHB production process with pure glycerol as the 908 carbon source. The authors did not specifically relate the study to experimental data and the 909 simulated process was described by PHB productivity of 2.86 g Lh -1 . Some simulation parameters 910 used by Leong et al. (2017) are compared to those applied in this study, as detailed in Table 9. The authors achieved maximum PHB content of 62%, which resulted in PHB titer of 51.1 g L -1 918 and productivities ranged from 0.84 to 1.5 g Lh -1 .

919
In comparison to the study described by Leong et al. (2009), the simulated productivity we 920 calculated in this study, PpSi = 1.1 g Lh -1 was low, although comparable to those experimentally

935
As detailed in Table 9, scenario SC1 had lower production costs related to raw materials (37.4%).

936
However, facility-dependent production costs and investments were very important (56.9%) and 937 remarkably disadvantageous in comparison to simulated results by Leong et al. (2017). As a 938 consequence, despite lower costs with raw material, unit production cost in SC1 (US$ 6.2 kgPHB -939 1 ) was not more competitive than that described by Leong

942
Production costs were also analyzed by process section. The Upstream section, which included 943 culture medium and inoculum preparation, had the most important share in costs, 39.5%. Next,

944
Bioprocess accounted for 37.6%, which included mostly the operation of FR-101 units. Lastly, the 945 Downstream section accounted for 22.9% of total production costs. These results implyed that optimizing bioreactor use was imperative for reduction of production costs, but cheaper culture 947 medium were also needed.

948
Firstly, different prices of glycerol were evaluated, as well as the use of vinasse as a partial source 949 of nutrients for culture medium. As expected, higher costs with glycerol resulted in higher 950 production costs in general. Scenarios that did not consider the use of vinasse (SC1, SC2 and 951 SC3) had annual production costs reduced by 6% due to the lower glycerol prices. In scenarios 952 SC4, SC5 and SC6, the higher the glycerol price, more important were production costs due to 953 vinasse use as nutrients source. Both scenarios SC1 and SC4 had glycerol price at US$ 400 ton -954 1 , and costs in SC4 were 13.1% lower. As for SC2 and SC5 (glycerol at US$ 300 ton -1 ), SC5 had

966
Vinasse was also a partial source of mineral components in culture medium.

976
Despite reduction of production costs by decreasing costs with culture medium, facility-dependent 977 costs were significant and invariable ( Figure 12). So in order to decrease process costs even 978 further, improvements in productivity are needed.

1847
B. PHB production as tons per batch. C. Unit production cost.