Selective fermentation of carbohydrate and ... - Wiley Online Library

ARTICLE Selective Fermentation of Carbohydrate and Protein Fractions of Scenedesmus, and Biohydrogenation of its Lipid Fraction for Enhanced Recovery of Saturated Fatty Acids YenJung Sean Lai,1 Prathap Parameswaran,1 Ang Li,2 Alyssa Aguinaga,1 Bruce E. Rittmann1 1

Swette Center for Environmental Biotechnology, The Biodesign Institute at Arizona State University, P.O. Box 875701, Tempe, Arizona 85287-5701; telephone: þ1-480-727-0849; fax: þ1-480-727-0889; e-mail: [email protected] 2 State Key Laboratory of Urban Water Resource and Environment, School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin, People’s Republic of China

Introduction ABSTRACT: Biofuels derived from microalgae have promise as carbon-neutral replacements for petroleum. However, difficulty extracting microalgae-derived lipids and the co-extraction of nonlipid components add major costs that detract from the benefits of microalgae-based biofuel. Selective fermentation could alleviate these problems by managing microbial degradation so that carbohydrates and proteins are hydrolyzed and fermented, but lipids remain intact. We evaluated selective fermentation of Scenedesmus biomass in batch experiments buffered at pH 5.5, 7, or 9. Carbohydrates were fermented up to 45% within the first 6 days, protein fermentation followed after about 20 days, and lipids (measured as fatty acid methyl esters, FAME) were conserved. Fermentation of the non-lipid components generated volatile fatty acids, with acetate, butyrate, and propionate being the dominant products. Selective fermentation of Scenedesmus biomass increased the amount of extractable FAME and the ratio of FAME to crude lipids. It also led to biohydrogenation of unsaturated FAME to more desirable saturated FAME (especially to C16:0 and C18:0), and the degree of saturation was inversely related to the accumulation of hydrogen gas after fermentation. Moreover, the microbial communities after selective fermentation were enriched in bacteria from families known to perform biohydrogenation, i.e., Porphyromonadaceae and Ruminococcaceae. Thus, this study provides proof-ofconcept that selective fermentation can improve the quantity and quality of lipids that can be extracted from Scenedesmus. Biotechnol. Bioeng. 2016;113: 320–329. ß 2015 Wiley Periodicals, Inc. KEYWORDS: fermentation; lipids; microbial community; Scenedesmus

Correspondence to: P. Parameswaran Contract grant sponsor: ASU LightWorks Received 20 May 2015; Revision received 17 July 2015; Accepted 24 July 2015 Accepted manuscript online 29 July 2015; Article first published online 18 September 2015 in Wiley Online Library (http://onlinelibrary.wiley.com/doi/10.1002/bit.25714/abstract). DOI 10.1002/bit.25714

320

Biotechnology and Bioengineering, Vol. 113, No. 2, February, 2016

Dependency on fossil fuels and an increasing threat of global warming from carbon dioxide emissions are primary drivers in the search for renewable energy. Biofuels derived from lipids produced by photosynthetic microorganisms—algae and cyanobacteria— have the promise of being carbon-neutral replacements for petroleum (Chisti, 2007; Rittmann, 2008). Microalgae biomass targeted as feedstock for transportation fuel has focused mainly on microalgae species that accumulate a high level of triacylglycerides (TAG) having an appropriate fatty acid profile (Griehl et al., 2012; Islam et al., 2013). Efficient extraction of the TAG from microalgae is energy intensive and normally requires strong and toxic solvents, such as chloroform and methanol, along with pretreatment techniques, such as acid/alkali catalysis, thermal, ultrasonication, and pulsed electric fields, to lyse and disrupt the cell walls and membranes (Cho et al., 2013; Lai et al., 2014; Laurens et al., 2015; Sheng et al., 2011a; Zbinden et al., 2013). Moreover, current lipidextraction strategies co-extract non-lipid components from microalgae, necessitating further downstream processing to recover the valuable fuel precursors (Lai et al., 2014). Ideal characteristics of microalgae-derived lipids for fuel include a low content of polyunsaturated fatty acids and a high content of saturated fatty acid (Alonso et al., 2010; Chisti, 2007; Greenwell et al., 2009; Knothe, 2006). Microalgae species known to accumulate high lipid content are Chlorella, Scenedesmus, and Nannochloropsis, but they generally have more poly-unsaturated fatty acids (C16:1, C18:1, and C18:3) than saturated fatty acids (C14:0 to C18:0). Hence, significant reforming of the fuel feedstock from microalgae often is required before production of the fuel (Alonso et al., 2010; Greenwell et al., 2009; Knothe, 2006). Hydrogenation, which catalytically converts unsaturated fatty acids to saturated forms (Chisti, 2007; Dijkstra, 2006), is energy intensive and adds cost to the fuel production. An appealing alternative to catalytic ß 2015 Wiley Periodicals, Inc.

hydrogenation is microbial biohydrogenation, which is known to occur within Porphyromonodaceae, Bacteroidales, Clostridiales, and Ruminococcaceae families in the rumen microbial ecosystems of animals (Castro-Carrera et al., 2014; Huws et al., 2011). To date, no published studies document biohydrogenation of microalgaederived lipids. Fermentation is a mature biotechnology for capturing electrons from a variety of organic wastes in the form of volatile fatty acids (VFAs) and hydrogen gas (H2) (Gavala et al., 2003; Guo et al., 2010). In general, microbial communities that ferment complex organic matter ferment lipids more slowly than carbohydrates and proteins (Christ et al., 2000; O’Rourke, 1968; Rittmann and McCarty, 2001). Specifically, bacteria responsible for hydrolysis and fermentation of lipids, being slow growers, are washed out of mixed-culture fermentation systems at solid retention times (SRTs) that allow transformation of carbohydrates and proteins present in primary wastewater sludge to volatile acids. For example, lipid-fermenting bacteria were washed out at an SRT of around 4 days at 35 C (O’Rourke, 1968; Rittmann and McCarty, 2001). This difference in washout SRT should make it feasible to operate a fermentation system that selectively ferments the carbohydrates and proteins to VFAs, but leaves the lipids intact, although in a form much more readily extractable because the algal cells have been disrupted and much of the carbohydrate and protein fractions have been transformed to volatile fatty acids. This concept of selective fermentation has not yet been demonstrated for microalgal biomass, and the co-occurrence of biohydrogenation would further elevate the quality of the conserved lipids for use as liquid transportation fuel. Fermentation of organics is influenced by the operating pH, source of inoculum, and selection based on pH, temperature, and the nature of the feed organics. An important example is the distribution of fermentation products, which depends strongly on the pH (Feng et al., 2009; Lee et al., 2008) and leads to a concomitant change to the microbial community (Cheng et al., 2014; Lee et al., 2009; Zheng et al., 2013). Major efforts have been channeled toward understanding how pH affects products during fermentation of carbohydrates and protein degradation (Lee et al., 2009; Walker et al., 2005), but almost no attention has been given to the fate of the lipid fraction. While recent research has shown that fermentation of algae can produce high yields of volatile fatty acids (VFA) and hydrogen gas (Choi et al., 2011; Pham et al., 2012), the work does not include comprehensive electron balances needed for understanding the relative degradation of carbohydrates, proteins, and lipids, as well as the feasibility of biohydrogenation during fermentation. Here, we carried out proof-of-concept testing of selective fermentation at three pH conditions—5.5, 7, and 9—using anerobic digester sludge as the inoculum. The main goals of our study were (i) to evaluate selective fermentation of the carbohydrate and protein fractions for the microalga Scenedesmus sp.; (ii) to determine if selective fermentation led to biohydrogenation of unsaturated fatty acids and to enhanced lipid recovery (as fatty acid methyl esters or FAMEs); and (iii) to elucidate linkages between microbial community structure and function during selective fermentation.

Materials and Methods Biomass Forty liters of Scenedesmus biomass, freshly harvested from a pilotscale photobioreactor, was obtained from the Arizona Center for Algal Technology and Innovation (AzCATI), Mesa, AZ. After transport to the Swette Center for Environmental Biotechnology (30 min transit time), the biomass samples were immediately characterized and utilized for selective-fermentation assays. Selective Fermentation Selective-fermentation assays were set up with Scenedesmus biomass and an inoculum of anerobic digested sludge obtained from the anerobic digesters at the Mesa Northwest Wastewater Reclamation Plant (MNWWRP). Three pH conditions were established—5.5, 7, and 9—by buffering with 40 mM 2-(N-morpholino) ethanesulfonic acid (MES), sodium phosphate, and sodium borate, respectively. Methanogenesis was inhibited with 10 mM 2-bromoethanesulfonic acid sodium (BES). A reducing agent, 40 mM sodium sulfide, also was amended to scavenge any oxygen carried over in the biomass or produced by photosynthetic microorganisms. One biological replicate for each pH condition was conducted, the reactor (500-mL screw-cap serum bottle, VWR, Radnor, PA) was modified with a side sampling port for gas measurement, and two PTFE tubes (Kimble Chase, Vineland, NJ) were inserted through the rubber stopper (Wheaton, Millville, NJ) for the liquid sampling. All reactors were well mixed with an incubator shaker at 210 rpm (New Brunswick Scientific, Enfield, CT) and kept at 37 C. Fermentation studies were carried out in two batches. The first batch, using anerobic digester sludge, was used to enrich for a community well adapted to the given pH conditions. The volume ratio of anerobic digester sludge: Scenedesmus biomass was 1:10 (50 mL inoculum: 450 mL Scenedesmus biomass). After 32 days of the first batch fermentation, three volumes of fermented and adapted inoculum from the first batch were combined with seven volumes of fresh algae biomass, and the pH was buffered to approximately the same value (5.5, 7, or 9) as for the first batch. The second batch fermentations are the focus of the evaluation of selective fermentation, and they began after the reactors were sparged with nitrogen gas followed by addition of 40 mM reducing agent. Slurry and gas samples were taken at regular intervals for the measurements outlined in analytical methods. DNA samples for microbial-community analysis were collected for anerobic sludge inoculum and at the end of the second batch fermentation for microbial community analysis. Analytical Methods Slurry samples were assayed directly for total chemical oxygen demand (TCOD), total suspended solids (TSS), and volatile suspended solids (VSS). TSS and VSS were determined by dry weight according to Standard Methods (Rice et al., 2012). Four parameters were assayed after 0.2-mm membrane filtration (Pall Science, Port Washington, NY): soluble-protein, soluble-carbohydrate, volatile fatty acids (VFA), and ammonium-N (NH4þ-N). Total COD was measured using a HACH COD kit (concentration range 10–1,500 mg/L). Carbohydrate and protein

Lai et al.: Selective Fermentation of Scenedesmus Biomass Biotechnology and Bioengineering

321

were analyzed by a colorimetric method (DuBois et al., 1956) and the bicinchoninic acid (BCA) method (Brown et al., 1989), respectively; concentrations were determined using calibration curves with glucose and bovine serum albumin standards, respectively. Volatile fatty acids were measured with an HPLC (Shimadzu North America, Columbia, MD) equipped with an Aminex HPX-87H column (Parameswaran et al., 2009). Gas samples of CH4 and H2 were analyzed by gas chromatography (Parameswaran et al., 2009). NH3-N was measured using a HACH NH3-N kit (concentration range 0–50 mg/L as N) (HACH, Loveland, CO) using a spectrophotometer (DR-2800, HACH).

We use the following nomenclature for each fatty acid: Cx:n, where x is the number of carbons in the chain and n is the number of double bonds in the fatty acid chain. For instance, C16:0 refers to a fatty acid with 16 carbon and no double bonds, while C18:3 refers to a fatty acid with 13 carbons and three double bonds. To quantify the degree of biohydrogenation between the initial sample and after selective fermentation, we define the differential saturation degree with Equation (2) Differential saturation degree ð%Þ ¼ ½½Saturated FAME ð%Þfinal ½Saturated FAME ð%Þinitial = ½Saturated FAME ð%Þinitial 100%

Crude Lipids and Fatty Acid Methyl Ethers (FAME) Lipids were extracted from the biomass at the end of batch fermentation using methods adapted from Sheng et al. (2011b). For minimizing effects of the differences in sample pH on lipid extraction, all the samples were washed with distilled water and rinsed with MES buffer (40 mM, pH 4.5) to maintain the same low pH, which improves lipid extraction (Liang et al., 2012). Then, about 5 g of biomass (dry weight) was freeze dried using a FreeZone Benchtop instrument (Labconco, Kansas City, MO) and then mixed with 3 mL of Folch solvent (Folch et al., 1957) with a solvent-tobiomass ratio of 200:1 mL/g. The mixture was vortexed for 3 h using a vortex mixer (Scientific Industries, Bohemia, NY) at room temperature, and solvent extracts were obtained after removing the biomass debris by filtration through a 0.2-mm PVDF membrane (Pall Science, Port Washington, NY). The crude lipid weight was obtained by evaporating the solvent in a N2 evaporator (Labconco RapVap, Kansas City, MO), weighing the tube containing the dried lipid, and subtracting the weight of the empty tube plus the weight of material released from the syringe filter through which the solvent alone was passed. Concentrations of fatty acid methyl esters (FAME) were obtained in two ways. The first way was through trans-esterification of dried crude lipids. We added 2 mL of 3-N methanolic HCl (Sigma– Aldrich, St.Louis, MO) to the entire dried lipid in a test tube and incubated the mixture at 85 C in an oven for at least 2.5 h (Sheng et al., 2011b). The extractable FAME recovery was evaluated using Equation (1).

ð2Þ

Mass Balance Based on COD We established COD mass balances by computing the COD equivalents of measured carbohydrate, protein, lipids (expressed as FAME), volatile fatty acids, H2, and CH4. COD conversion units were adapted from the previous studies (Lee et al., 2008; Miron et al., 2000; Rittmann and McCarty, 2001): 1.07 mg COD per mg carbohydrate; 1.5 mg COD per mg protein; 0.35, 1.07, 1.33, 1.51, 1.82, 2.04 mg COD per mg, respectively, for the VFAs formate, lactate, acetate, propionate, butyrate, and valerate; 2.72, 2.81, 2.88, and 2.93 mg COD per mg fatty acid, respectively, for the four major FAME compounds (lauric, myristic, palmitic, and stearic); and 0.63 and 2.52 mg COD per mL gas (37 C) for H2 and CH4. The overall mass balance involved the addition of the individual COD values from carbohydrate, protein, FAME VFA, H2, and CH4. In addition, we estimated the net change of the COD of total FAME using Equation (3) Net change of total FAME ð%Þ ¼ ½½FAME to biomass ratio ðmg=mgÞ TSSinitialðmg=volÞ vol COD equivalenceinitial ½FAME to biomass ratio ðmg=mgÞ TSSfinal ðmg=volÞ vol COD equivalencefinal =

FAME recovery ratio ð%Þ ¼ ½½FAME to crude lipidfinal

½FAME to biomass ratio ðmg=mgÞ

½FAME to crude lipidinitial = ½FAME to crude lipidinitial 100%

TSSinitial ðmg=volÞ vol COD equivalenceinitial 100%

ð1Þ The second way was direct transesterification (Laurens et al., 2012), in which we directly amended dried biomass with 2 mL methanolic HCl. Direct transesterification gave the maximum value for total FAME, as it was not limited by extraction efficiency. All FAME components were quantified against a 37-component FAME-mix standard (Supelco, Bellefonte, PA) using a gas chromatograph (Shimadzu GC 2010, Columbia, MD) equipped with a Supelco SP-2380 capillary column (30 m 0.25 mm 0.20 mm) and flame ionization detector (FID). We identified the compounds by comparing peak retention times to those of standard compounds.

322


ð3Þ

DNA Extraction Pellets were prepared from the fermentation reactors at the end of batch operation and from the anerobic digested sludge that served as the inoculum. Approximately 0.30 g dry weight was used for DNA extraction with a PowerSoil DNA isolation kit (MoBio laboratories, Inc., Carlsbad, CA). We assessed the quantity and quality of extracted DNA by using a nanodrop spectrophotometer at 260 and 280 nm (Ruiz et al., 2014).

Pyrosequencing Analysis Extracted DNA was sent to the Research and Testing Laboratory (Lubbock, TX) for Roche 454 pyrosequencing. The “blue” primer set (104F, 530R) targeted the V2-V3 regions of the bacterial 16S rRNA gene (Miceli et al., 2012). Data received from the testing laboratory were analyzed using QIIME (Caporaso et al., 2010) after discarding sequences shorter than 25 bp, longer than 450 bp, or labeled as chimeric sequences. After screening, primer sequences were trimmed off, and taxonomic classification was performed using the RDP classifier (Cole et al., 2009) at the 80%-confidence threshold. The total number of sequence reads for each sample after screenings were: anerobic inoculum ¼ 5,865, pH 5.5 end of batch ¼ 9,703, pH 7 end of batch ¼ 11,650, and pH 9 end of batch ¼ 11,519.

Results Kinetics of Carbohydrate and Protein Fermentation to VFAs Table I summarizes key parameters at the beginning and the end of the second batch-fermentation experiments; the same data set for the first batch experiments are shown in Supplementary Table S1. All three pH conditions showed loss of VSS and an increase in ammonium. Figure 1a shows a loss of about 90 mg net COD carbohydrate/g TCOD for pH 5.5 and 7 and 60 mg net COD/g TCOD loss for pH 9 within 6 days. Figure 1b shows no significant degradation in total protein for the first 20 days, but subsequently about 49, 27, and 5 mg net COD protein/g TCOD was degraded for pH 7, 5.5, and 9, respectively. The first batch showed a similar trend as the second batch, i.e., protein degradation following carbohydrate degradation, but conservation of lipids (in Supplementary Figs. S1 and S3). The COD of volatile fatty acids (VFA) increased with time (Fig. 1c), and VFA production was greater for pH 7 than for pH 5.5 or pH 9. Most of the VFA production occurred within the first 6 days, which corresponds to carbohydrate degradation (Fig. 1a). VFA production after 6 days came mostly from degradation of protein, and the sum of COD associated with loss of carbohydrate and protein match with the net production of VFA-COD for pH 5.5 and 7. In terms of VFA composition shown in Figure 2 (Supplementary Fig. S2 for the first batch), acetate, propionate,

butyrate, and valerate were the four major components for all batch runs. Acetate was predominant at pH 7 and 9 and threefold more than at pH 5.5. Propionate and butyrate were significant, along with acetate, at pH 5.5. Lipid (FAME) Conservation Figure 3a clearly shows that the net change of total FAME (obtained according to the Equation (3) and involving normalization from the dried solids before and after fermentation, respectively corresponding to the pHs) was less than 4% at all pHs, which indicates minimal release of free fatty acids into liquid phase after fermentation. Figure 3b shows that fermentation enhanced the extraction efficiency of the crude lipids and FAME. For example, extractability of crude lipids after fermentation improved by 12, 28, and 14% for pH 5.5, 7, and 9, respectively, compared to biomass before fermentation. Selective fermentation also led to a simultaneous increase in the FAME-to-crude lipid ratios by 17, 10, and 1% for pH 5.5, 7, and 9, respectively. The results from the first batch had similar trends for lipid conservation and FAME extraction (in Supplementary Fig. S3). COD Mass Balance During Fermentation The COD distributions based upon COD equivalence for each measured component are shown in Figure 4 for the second batch experiment. The COD mass balances at the end of fermentation show that carbohydrate fermentation contributed most of the COD equivalents of fermentation products (mostly VFAs), a trend also seen in the first batch experiment (Supplementary Fig. S4). H2 was a minor component in the COD balance, 0.2% for pHs 5.5 and 7 and 0.4% for pH 9. COD conservation during the fermentation experiments was good, with deviations < þ8% based upon the COD of the components (in Table I) and 2% to þ8% based on direct TCOD measurements (Supplementary Table S2). FAME Profile Change After Selective Fermentation Prior to fermentation, raw microalgal biomass mixed with anerobic digested inoculum for both batch runs showed FAME profiles with similar compositions that were dominated by C18:1 (unsaturated) and with some C16:0 (saturated). This is shown in Figure 5a for the

Table I. Key parameters for the second batch fermentation experiments, which used a 3:7 (v/v) ratio of inoculum from the first batch to fresh algal biomass.

Conditions

pH

TSS (mg/L)

Raw algal biomass Initial pH 5.5 Final pH 5.5 Initial pH 7 Final pH 7 Initial pH 9 Final pH 9

8.3 5.5 5.5 7 6.5 9.1 8.9

3800 30 5200 340 4100 140 5000 160 3500 40 5400 370 4900 20

VSS (mg/L)

(VSSin-VSSout)/ VSSin (%)

Ammonium (mg N/L)

Estimateda TCOD (mg)

Difference between initial and final conditions for estimated assay (%)

3700 20 4900 170 3900 90 4100 30 2800 40 4100 80 3500 110

NA 21

NA 3 28 0 31 1 23

NA 3450 3350 3230 3360 3339 3616

NA 3

32 14

4 8

The standard deviations are for duplicate measurements of each parameter. in, initial fermentation condition; out, final fermentation condition; NA, not attempted. a The estimated COD is the sum of COD equivalent values from carbohydrate, protein, lipid, volatile fatty acid, and methane/hydrogen. The COD equivalent values for individual carbons are documented in the Materials and methods section.


323

Figure 1.

Changes of (a) total carbohydrate, (b) total protein, and (c) volatile fatty acids during the second batch fermentation. The vertical-axis values represent the fraction expressed as COD normalized to the TCOD concentrations obtained from the HACH COD assays. The values were obtained from duplicate samples.

second batch fermentation and in Supplementary Figure S5 for the first batch fermentation. At the end of the fermentations at pH 5.5 and 7, the dominant fatty acids were fully saturated 16:0, C18:0, and C14:0. As illustrated in Supplementary Figures S1 and S2, selective fermentation also occurred at pH 9 in the first batch study, but the FAME profile hardly changed from the raw microalgae mixed with anerobic digested sludge (Fig. 5a and Supplementary Fig. S5). Thus, saturation of the lipids only occurred for selective fermentation at pH values of 5.5 and 7. The quantity of FAME in the anerobic sludge inoculum was only 0.6% (Supplementary Fig. S5b), a value much lower than the total FAME in the microalgae (22%); thus, mixing the anerobic sludge with the microalgal biomass had

324


Figure 2. VFA profiles during the second batch fermentation for (a) pH 5.5, (b) pH 7, and (c) pH 9. insignificant impact on the quantity and profile of FAME in the fermentation experiments. Figure 5b clearly demonstrates that the increase in the degree of saturation after fermentation was inversely proportional to the H2 COD equivalents at the end of the batch fermentations. In other words, less accumulation of H2 in the headspace was associated with more production of saturated fatty acids during fermentation.

Microbial Communities Associated With Selective Fermentation Microbial community analysis by pyrosequencing revealed that selective fermentation significantly altered the microbial community from the anerobic inoculum. Figure 6a shows that the anerobic

Figure 3.

(a) The net change of total fatty acid methyl esters (FAME) after fermentation for the second batch. Estimation was according to Equation (3). (b) The crude lipids and FAME recovery obtained using Folch-solvent extraction before and after fermentation for the second batch. The values, normalized to the dried biomass before and after fermentation for each pH condition, were obtained from duplicate samples.

Figure 4. The COD mass balance for the second batch before (initial) and after fermentation (final). The carbon fractions were measured from individual assays and converted to COD equivalents values and represented as a fraction of the total COD.

Figure 5. (a) FAME profiles at different pHs before and after fermentation (initial and final), as well as the feeding algal biomass (raw FB) for the second batch. The data are the average fraction of FAME obtained from duplicate samples. (b) Accumulated H2 after fermentation at different pHs for the second batch; the H2 accumulation was inversely correlated to the differential saturation degree obtained from initial and final saturated FAME profile change.

sludge was diverse (at least 10 identifiable phyla), with Proteobacteria (33% of total reads) and Bacteroidetes (18%) being most significant. After selective fermentation, the number of significant phyla (defined as greater than 2% of the total reads) declined: 4 phyla at pHs 5.5 and 7. The Bacteroidetes phylum predominated at pH 5.5 (64%) and pH 7 (58%), but was absent at pH 9. Cyanobacteria represented the major dominant phylum at pH 9 (80%) and was present at pH 7 (28%), while Firmicutes represented the second predominant phylum at pH 5.5 (25%) and pH 9 (14%), although they represented less than 5% of the total reads at pH 7. Family-level information is in Figure 6b. After fermentation, Porphyromonadaceae from the Bacteroidetes phylum and Ruminococcaceae from the Firmicutes phylum predominated at pH 5.5 and 7, respectively. However, Anaerobrancaceae (5%), Clostridiaceae (3%), and Peptococcaceae (2.4%) were observed at pH 9, all of which are associated with alkaline fermentation of sugars and proteins (Mesbah et al., 2007; Seckbach et al., 2000; Zhilina et al., 2009). A high fraction of a cyanobacterium, Chlamydamonodaceae, was present at pH 9 and 7, likely due to carry over from the algal biomass samples and then maintained at both pHs.


325

Figure 6. High-throughput pyrosequencing analysis for the anaerobic inoculum and for the end of fermentation for the three pH conditions (5.5, 7, and 9) for the second batch at the (a) phylum and (b) family levels.

Discussion The concept of selective fermentation was demonstrated in the batch studies with Scenedemus biomass. Degradation of volatile solids confirmed that hydrolysis of Scenedesmus biomass was a prerequisite for successful selective fermentation, and hydrolysis was significant at pH 5 and 7. Fermentation degraded carbohydrates most rapidly, protein degradation followed (with the ammonium-N increases reinforcing the degradation of protein), and lipids were conserved (Fig. 3). All of these trends are consistent with the anerobic digestion studies that motivated the concept of selective fermentation (Christ et al., 2000; O’Rourke, 1968; Rittmann and McCarty, 2001). In addition, our protein results are similar to anerobic protein degradation observed for microalgae by Velasquez-Orta et al. (2009).

326


Fermentation of carbohydrate and protein yielded commensurate quantities of volatile fatty acids (Fig. 2). The quantity and the composition of VFAs were highly affected by pH, which is consistent with previous fermentation research (Ginkel et al., 2001; Lee and Rittmann, 2009). In general, acetate was the predominant fermentation product, but lower pH (5.5 vs. 7 and 9) shifted the pathway, leading to more butyrate and propionate production (along with acetate). The low H2 levels (Supplementary Fig. S6) observed in the headspace at pH 5.5 and 7 indicate that H2 produced by fermentation was consumed by hydrogen-oxidizing bacteria. Crude-lipid recovery by extraction was higher after selective fermentation, since lipids (assayed by direct transesterification) were conserved, while remaining lipid-containing biomass was more susceptible to solvent penetration. The extracts were enriched

in FAME, because non-lipid materials had been fermented or separated from the lipids during fermentation. Selective fermentation led to FAME profiles becoming more saturated (Fig. 5a). The reduction of unsaturated fatty acids to saturated forms probably was driven by oxidation of the H2 generated from fermentation of carbohydrate and protein (as shown in Supplementary Table S3). If no H2 sinks were present, H2 accumulation would have followed the order of pH 5 > pH 7 >> pH 9. However, if biohydrogenation were one of the H2 sinks (homoacetogenesis being the other), a higher degree of biohydrogenation should lead to less H2, a trend that we observed in Figure 5b, as shown by this order, pH 7 < pH 5