Thermal and alkaline pre-treatments of inoculum halt methanogenesis and enables cheese whey valorization by batch acidogenic fermentation

Carboxylates like volatile fatty acids (VFAs) can be produced by acidogenic fermentation (AF) of dairy wastes like cheese whey, a massive residue produced at 160.67 million m3 of which 42% are not valorized and impact the environment. In mixed-culture fermentations, selection pressures are needed to favor AF and halt methanogenesis. Inoculum pre-treatment was studied here as selective pressure for AF demineralized cheese whey in batch processes. Alkaline (NaOH, pH 8.0, 6 h) and thermal (90°C for 5 min, ice-bath until 23°C) pre-treatments, were tested together with batch operations run at initial pH 7.0 and 9.0, food-to-microorganism (F/M) ratios of 0.5 to 4.0 g COD g-1 VS, and under pressurized and non-pressurized headspace, in experiments duplicated in two institutes. Acetic acid was highly produced (1.36 and 1.40 g CODAcOH L-1) at the expense of methanogenesis by combining a thermal pre-treatment of inoculum with a non-pressurized batch operation started at pH 9.0. Microbial communities comprised of VFAs and alcohol producers, such as Clostridium, Fonticella, and Intestinimonas, and fermenters such as Longilinea and Leptolinea. Communities also presented the lipid-accumulating and bulk and foaming Candidatus Microthrix and the metanogenic Methanosaeta regardless of no methane production. An F/M ratio of 0.5 g COD g-1 VS led to the best VFA production of 1,769.38 mg L-1. Overall, inoculum thermal pre-treatment, initial pH 9.0, and non-pressurized headspace acted as a selective pressure for halting methanogen and producing VFAs, valorizing cheese whey via batch acidogenic fermentation.


5
Headspace gas composition and pressure play a role in product formation and metabolic 86 pathways preferentially used. However, most studies still focus on hydrogen production (Darvekar 87 et al., 2019). According to Zhou (2018) and Sarkar (2017), low hydrogen pressure favors VFA 88 formation. Finally, the F/M ratio, which is inoculum and substrate-dependent, is a parameter that 89 impacts acidogenesis, with lower F/M ratios being beneficial to VFAs production (Pang et al.,90 2019; Shah et al., 2015). 91 In this work, we aimed at (i) identifying how abiotic factors (i.e., pH, inoculum pre-treatment, 92 headspace pressure, and F/M ratio) influence the products spectra of cheese whey via AF, (ii) 93 validating thermal pre-treatment efficiency for halting methanogenesis, and (iii) identifying the 94 parameters that increase acetate level of production for other biological processes (e.g., microalgal 95 photoorganoheterotrophic biomass production). inoculated at a concentration of 6.6 g total volatile solids (TVS) L -1 , as depicted in Mockaitis et al. 154 (2020). Batch reactors were continuously agitated at 50 rpm (Orbital Incubator Marconi -MA420) 155 at a mesophilic temperature of 35°C for 30 days. 156

Batches headspace pressure assays 157
The influence of headspace pressure (pressurized and non-pressurized batches) on VFAs 158 production was investigated according to Peixoto et al., (2011). Manometric pressure was 159 measured, then, 3 mL samples of gas were collected with a syringe containing a pressure lock 160 (Thermo Fisher). Each sample corresponded to a batch condition. 161 10 In pressurized headspace assays, butyl rubber stoppers of each flask were covered with silicone 162 sealant after manometric measurement and gas sample collection, allowing gas accumulation in the 163 batch headspace. 164 In non-pressurized experiments, the headspace of each flask was punctured with a needle after 165 pressure manometric reading and gas sampling, allowing the remaining gas to be released until 166 reaching the atmospheric pressure value. After this step, butyl rubber stoppers were also covered 167 with silicone sealant avoiding any gas release into the atmosphere. 168 Once collected, gas samples were analyzed with a gas chromatograph (GC) equipped with a 169 thermal conductivity detector GC-TCD (Construmac, Brazil) with hydrogen as a carrier gas. 170 The volumetric production of biogas and its constituents were inferred through equation 1 as a 171 punctual function from sampling timestep t 0 = 0 to t for non-pressurized assays. 172 Where ߁ is the volumetric production of a gas of interest (N 2 , H 2 , CH 4 or CO 2 ) in standard 173 temperature and pressure (STP) equivalents (L), at a given time n; P i is the measured pressure at 174 sampling time (atm); is the headspace volume before sampling (L); T is system's temperature 175 Inoculum thermal pre-treatment and non-pressurized headspace assays were reproduced to 196 confirm the efficiency of such imposed conditions in halting methanogenesis while enhancing 197 VFAs production. Each factor was tested at one level, based on the effects identified at UNICAMP 198 (section 2.3): inoculum thermal pre-treatment, initial pH of 9.0, and non-pressurized headspace. A 199 12 batch with no inoculum pre-treatment, initial pH of 9.0, and non-pressurized headspace acted as a 200 control. The experimental design with three factors applied to this set-up is shown in Table 2. 201 Similar to the first experiments conducted, the initial working volume of the 1 L Duran flasks 202 was 750 mL with an initial headspace volume of 398 mL, for a total working volume of 1,148 mL. 203 The inoculum concentration was identical at 6.6 g TVS L -1 as depicted in Mockaitis et al., (2020). 204 Batch reactors were incubated for 10 days at a mesophilic temperature of 35°C in a Certomat ® BS1 205 incubator (Sartorius Stedim Biotech, Germany) under continuous agitation (145 rpm). The 206 headspace was released with the aid of Cole-Palmer stopcocks with Luer connection, a 1-way male 207 lock (Cole-Parmer, USA). Experiments were performed in triplicate and lasted 12 days. 208

VFAs measurements 209
VFAs and carbohydrates were quantified by HPLC. One mL of mixed liquors was 210 centrifuged at 5,073 x g for 5 minutes and supernatants were filtered using a 0.45 μ m syringe filters. 211 BioRad HPX-87H (300 x 7.8 mm) with a BioRad Cation-H refill cartridge (30 x 4.6 mm) guard 216 column (BioRad, USA) and the column oven was built in-house. The flow rate of the pump was 0.6 217 mL min -1 and the temperature of the column was set at 59°C. The RI detector was operated at 218 30 °C. The mobile phase was 1.5 mmol L -1 phosphoric acid diluted in ultrapure water (MilliQ, 219 Merck Millipore). The VFAs measured were acetic acid, butyric acid, formic acid, caproic acid, 220 propionic acid, valeric acid, iso-butyric acid, iso-caproic acid, and iso-valeric acid. 221 13 2.5 Ideal F/M ratio for optimal VFAs production with thermally pre-treated inocula 222 F/M ratio is a measurement used to determine the amount of substrate needed for the number of 223 microorganisms present in a system. It is an important parameter to evaluate in an approach aiming 224 at maximizing VFA production. To determine the best FM ratio for VFA production when working 225 with a thermal pre-treated sludge at non-pressurized headspace, four different F/M ratios were 226 tested: 0.5, 1.0, 2.0, and 4.0,g COD g VS -1 . These ratios were obtained by dividing the COD of the 227 substrate by the VS of the sludge. The initial pH of all experiments was 9.0. In section 2.4, both 228 control and thermal batches presented an F/M ratio of 0.5 g COD g VS -1 . g COD g VS -1 acted as control. 229 Assays were conducted in duplicates for 14 days. The experimental design is depicted in Table 2. 230

Biogas measurements 233
The presence or absence of CH 4 in the gas was detected by injecting 10 mL samples of gas 234 Inoculum alkaline pre-treatment halted methanogenesis in both pressurized and non-306 pressurized assays but only with initial pH 7.0, presenting negligible to no levels of total methane 307 production of 0.0 and 0.54 mL, respectively. Initial pH 9.0 presented a total methane production of 308 about 52.29 mL for pressurized assays and 0.67 mL for non-pressurized ones. 309 As a strategy to maximize VFA production, inoculum thermal pre-treatment appears to be the 310 best selective pressure for impairing methanogens in batch processes. It is of utmost importance to 311 confirm whether this selective pressure is also successful to halt methanogenesis considering 312 different types of inoculum (e.g., waste activated sludge and secondary sludge), reactors (e.g., 313 sequence-batch reactors, and continuous-stirred tank reactors) and configurations (e.g., organic 314 loading rate, hydraulic retention time, coupled reactors). If so, then inoculum thermal pre-treatment 315 can be implemented in full-scale AF processes, decreasing their overall costs (Corti and Lombardi, Another result of interest regarding inoculum alkaline pre-treatment was the correlation 343 between acetate and methane production. While there was no methane production in inoculum 344 alkaline treatment with initial pH 7.0 in both non-pressurized and pressurized headspace batches, 345 acetate production differed 185.6% from non-pressurized headspace assay to pressurized headspace 346 19 (i.e., A PT 7NP 1,056.04 mg COD AcOH L -1 and A PT 7P 369.82 mg COD AcOH L -1 ). So, headspace 347 pressure can favor acetate production in acetoclastic methanogens. 348 The detected levels of CO 2 in both A PT 7NP and A PT 7P (i.e., 1.84 mL and 60.78 mL, 349 respectively) were low as was the average of the final pH for both conditions (i.e., 4.8). Hence, 350 instead of methanogenesis being inhibited due to disturbance in the buffer system (Casallas-Ojeda 351 et al., 2020), we can infer that most CO 2 was dissolved in the liquid phase, decreasing the pH 352 (Deublein and Steinhauser, 2011). Figure 2 shows the methane and carbon dioxide production in 353 alkaline and thermal batches with both pressurized and non-pressurized headspaces. Cheese whey has a natural tendency to acidify and since pH was not controlled during the 361 experiments (pressurized and non-pressurized batches displayed an average pH of 4.96 and 5.26, 362 respectively), it is likely that hydrolysis was not a limiting step in the AF process. 363 An additional hypothesis to justify the different production of VFAs and alcohol in all batches 364 was that together with uncontrolled pH, the headspace internal pressure variation from pressurized 365 and non-pressurized batches played a key role on the spectra and quantity of VFAs produced. In 366 pressurized batches, the pressure would contribute to the gas phase being accumulated within the 367 liquid phase, which influenced both gas and VFA spectra. In non-pressurized batches, due to 368 headspace release, gas and VFAs production restarted daily, after each sampling. 369 Overall, non-pressurized batches produced more VFAs than pressurized ones. Special 370 highlights can be given to A PT 9 NP (21,360.54 mg COD compound L -1 ), T PT 9 NP (4,108.71 mg 371 COD compound L -1 ), and T PT 7 NP (3,414.15 mg COD compound L -1 ), when compared to their counterparts 372 (A PT 9 P 1,940.14 mg COD compound L -1 , T PT 9 P 2,151.99 mg COD compound L -1 and T PT 7 P 2,140.62 mg 373 COD compound L -1 ). Both pressurized and non-pressurized controls and A PT 7 did not present a 374 substantial difference with an increase of approximately 2.9% in VFAs produced in non-pressurized 375 batches compared to pressurized batches as seen in table 3. 376 It is important to stress that although A PT 9 NP, T PT 9 NP, and T PT 7 NP produced greater 377 quantities of metabolites, they did not produce a great variety of VFAs and alcohol. A PT 9 NP 378 produced 7,677.30 mg COD IBA L -1 of iso-butyric acid and 13,683.24 mg COD EtOH L -1 of ethanol. 379 These metabolites production together with the absence of acetic acid production are a clear result 380 22 of a metabolic shift that was favored due to the imposed selective pressures in this batch (i.e., 381 alkaline inoculum pre-treatment, initial pH of 9.0, and non-pressurized headspace). 382 This was in opposition to all other conditions tested. It is likely that in such operational 383 conditions, acetate served as the substrate for fermentative bacteria to form iso-butyric acid and 384 ethanol, as already described by Thatikayala et al. (2021) and Liu et al. (2022). Both chemicals are 385 of great industrial interest (i.e., pharmaceutical, feed, chemical, biofuels). Optimization of AF 386 processes towards the production of either ethanol or iso-butyric acid or both can be an additional 387 approach for cheese whey valorization. 388 T PT 9 NP and T PT 7 NP displayed a production of acetic acid of 1,479.13 and 1,558.17 mg 389 COD AcOH L -1 , respectively. As shown in figure 2, there was no production of methane. Acetic acid 390 was consumed in A PT 9NP, whereas in A PT 7NP its production was slightly lower than both T PT 9 NP 391 and T PT 7 NP with 1,056.04 mg COD AcOH L -1 . 392 Interestingly, CNP also presented a higher acetic acid production of 1,109.13 mg COD AcOH L -1 , 393 which shows that headspace pressure is a more suitable selective pressure parameter than inoculum 394 pre-treatment for acetic acid production. The produced acetic acid can thus be utilized as an organic 395 carbon source for higher added-value biological processes (i.e., microalgal 396 photoorganoheterotrophic biomass production). 397 The imposed parameters (i.e., inoculum pre-treatments, initial pH, and headspace pressure) also 398 acted as selective pressures in the batch experiments modifying the microbial fermentation end-399 products. In general, thermal pre-treatment, non-pressurized headspace, and initial pH 9.0 displayed 400 a trend for higher VFA production. Nevertheless, it is not possible to affirm what the best 401 combination of parameters would be since there was not a pattern of production observed for all 402 VFAs in this given setup. 403 23 However, the combination of parameters can be suitable for the production of specific volatile 404 acids or ethanol, as seen in T PT 7NP and T PT 9NP for the production of acetic (1,479.13 and 1,558.17 405 mg COD AcOH L -1 , respectively) and butyric acids (1,207.46 and 1,425.36 mg COD BTA L -1 ), and 406 A PT 9NP for iso-butyric acid and ethanol (7,677.30 mg COD IBA L -1 and 13,683.24 mg COD EtOH L -1 ). 407 Non-pressurized batches presented a global production of 3 times more VFAs than pressurized 408 batches. Besides common VFAs, pressurized headspace batches produced formic and lactic acids 409 but in negligible quantities as shown in table 3. 410 Table 3. VFAs and alcohols product spectra of pressurized (P) and non-pressurized (NP) headspace batches with alkaline (A PT ) and thermal (T PT ) pre-411 treatments and initial pH of 7.0 and 9.0. Experiments lasted for 30 days. 412 Batch operational conditions (mg COD compound L-1) alkaline and thermal pre-treatments with initial pH 7.0 and 9.0 (A PT 7, A PT 9, T PT 7, and T PT 9) with pressurized (P) and non-pressurized (NP) headspaces. 423 Analyses considered are: chemical oxygen demand (COD), total organic carbon (TOC), total nitrogen (TN), total solids (TS), total volatile solids 424 In the replicate experiment performed in TU Delft, both control and inoculum thermal pre-429 treated microbial communities' composition and evolution were determined through 16S rRNA 430 gene amplicon sequencing (Figures 3 and 4). Initial microbial communities were highly similar, 431 with few quantitative and only minor qualitative differences being observed. environments. Although chlorophyll (Chl) deficiency can be caused by low levels of oxygen, these 445 microorganisms can regulate Chl production by activating oxygen-independent oxidases or by 446 inducing the transcription of genes that encodes enzymes that work in microoxic conditions (Fujita 447 et al., 2015). In the given scenario, facultative anoxygenic photosynthesis uses sulfide as the 448 electron donor with photosystem I-driven photoassimilation (Hamilton et al., 2018). 449 29 Control and inoculum thermal pre-treated microbial communities evolved differently along the 450 process. The final populations presented highly similar compositions at the phylum, class, order, 451 and family levels, sharing all or almost all the main detected taxons with only quantitative 452 differences being perceived. Noticeable qualitative and quantitative differences were however 453 observed at the genus level (Figures 3 and 4). They can also compete for nitrate content in cheese whey (Oliveira et al., 1995).   As shown in Table 6, F/M 0.5 ratio showed the best VFA production (3,317.88 mg COD compound L -516 1 ), whereas the F/M ratio 2.0 (345.93 mg L -1 ) had the lowest. Non-pressurized thermal pre-treatment 517 batch in the replicate experiment had a VFAs production of 2,568.26 mg L -1 . There was no 518 production of methane in all samples which corroboratto the results observed in previous 519 experiments. 520 35 Table 6 -VFAs spectra of F/M ratios experiments. F/M 0.5, F/M 1.0, F/M 2.0, and F/M 4.0: 0.5, 521 1.0, 2.0 and 4.0 g COD g VSS -1 , respectively. F/M 0.5 acted as a positive control as it has the same 522 0.5 g COD g VSS -1 ratio as the replicated thermal pre-treatment batch in section 3.

F/M ratios affected microbial communities' evolution 525
Microbial communities evolved differently during the F/M experiments. As shown in Figure 5, 526 increasing F/M ratios significantly impacted the microbial populations' evolution, with the 527 strongest impact observed at the highest ratio of 4.0. Under such conditions, the microbial 528 population was ultra-dominated by a single genus, Clostridium (Figures 3 and 4). At lower F/M 529 ratios, the community was mainly composed of the same four genera, which altogether represented residues such as cheese whey. Inoculum pre-treatments can play as selective pressures on microbial 547 communities, selecting microorganisms that can thrive in imposed conditions. The main 548 conclusions from this work are the following: 549 1. Thermal pre-treatment of inocula was efficient to halt methanogenesis regardless of 550 headspace pressure, pH, and F/M ratios. 551 2. Contrary to the literature, alkaline pre-treatment did not improve methanogenesis. However, 552 non-pressurized headspace alkaline pre-treatment pH 9.0 did produce significant quantities 553 of iso-butyric acid (7,677.30 mg COD IBA L -1 ) and ethanol (13,693.24 mg COD EtOH L -1 ). 554 Further studies on the mechanisms of this pre-treatment are necessary to optimize the 555 process. 556 3. Methanosaeta was present in the replicate of the control (pH 8.25) and thermal (pH 9.0) 557 batches, despite the absence of methane production. Thermal pre-treatment could have 558 either inactivated or destroyed this microorganism, which in its turn enabled the detection of 559 Methanosaeta in the 16S rRNA gene amplicon sequencing analysis. 560 4. A low F/M ratio of 0.5 g COD g -1 VS selected a microbial community producing a high 561 level of VFAs by AF. 562 5. The setup of parameters used as selective pressure during cheese whey AF must vary 563 according to the desired end product. 564 6. Headspace pressure directs the metabolic pathways for VFAs and alcohol production in AF. 565 Pressurized headspace batches had a more di while non-pressurized batches produced higher 566 amounts of VFAs. The exception would be iso-butyric acid and ethanol produced in non-567 pressurized alkaline pH 9.0 batches. 568 38 7. Another important parameter was the Initial pH of 9.0. Regardless of the headspace 569 pressure and inoculum pre-treatment, an initial pH of 9.0 produced the greatest number of 570 products. The drastic pH drop influences the redox potential of the medium facilitating the 571 uptake of compounds while setting the grounds for ecologic relationships within the 572 microbial community. 573 This work aimed to verify if imposed parameters (i.e., headspace pressure, inoculum thermal 574 and alkaline pre-treatment, and initial pH) in cheese whey AF would work as selective pressures on 575 the inoculum microbial community. Our objectives were met when we managed to halt 576 methanogenesis and increase acetic acid, iso-butyric acid, and ethanol. Acetate is an organic carbon 577 source that can be used for different biological conversions (e.g., microalgal 578 photoorganoheterotrophic processes), whereas iso-butyric and ethanol are of interest of 579 pharmaceutical, feed, chemical, and biofuels industries. 580 Still, there is plenty to be understood, especially regarding the influence of these parameters 581 on metabolic pathways and microbial interaction. The fact that thermal pre-treatment time and 582 temperature were optimized and methanogenesis was halted, despite the presence of methanogens. 583 Understanding the mechanism behind this can overcome competition for substrate. 584 Thermal inoculum pre-treatment can be used in short fermentation processes. Nevertheless, 585 the interaction among the factors with different substrates, inoculum and reactor configurations 586 must be studied to certify thermal pre-treatment efficiency. Thus, it is necessary to evaluate other 587 configurations for process implementation and optimization. 588 Understanding the VFAs production and consumption mechanisms is crucial for driving the 589 spectrum of fermentation products to desired compounds. Future 16S rRNA metagenomics analyses 590 will be fundamental to decrease these knowledge gaps.