Microbial lipid production from AFEX™ pretreated corn stover

Abstract

Lipids having high carbon to heteroatom ratios can be upgraded to bio-diesel and jet fuels which are more advanced drop-in fuels compared to ethanol. The present study investigated microbial lipid production from Ammonia Fiber Expansion (AFEX) pretreated and hydrolyzed corn stover (CS) using an oleaginous yeast strain Lipomyces tetrasporus NRRL Y-11562. Process conditions were optimized for carbon to nitrogen ratio of fermentation medium, fermentation temperature and pH, and solid loading of AFEX–CS. The inhibitory effect of AFEX degradation products on lipid fermentation was also investigated. Both separate hydrolysis and fermentation (SHF) and Rapid Bioconversion with Integrated recycle Technology (RaBIT) processes were used for lipid production. From 1 kg AFEX–CS, 36.7 g lipids were produced via SHF at a titer of 8.4 g L⁻¹ with a yield of 0.08 g g⁻¹ consumed sugar. A yeast meal stream (97.9 g) was also generated. L. tetrasporus NRRL Y-11562 grew better in AFEX–CS hydrolysate, but produced fewer lipids compared to synthetic medium. Minimal washing of AFEX–CS improved the lipid yield and titer to 0.10 g g⁻¹ consumed sugar and 10.7 g L⁻¹, respectively. RaBIT on washed AFEX–CS generated a similar amount of lipids compared to SHF with 35% lower enzyme loading. Economic analysis does not favor lignocellulosic lipid production with current lipid yields.

---

1. Introduction

Liquid fuels with high energy density, such as diesel and jet fuel in particular, are necessary for aviation and heavy vehicle transport. With fossil fuel reserves diminishing, it is necessary to find alternative resources for liquid fuel.¹ Biomass is probably the sole renewable resource that can be used for liquid fuel production based on currently available technology. The development of second generation biorefineries using lignocellulosic biomass, which is abundant and does not compete in the food market, as feedstock has recently attracted much attention. The production of biodiesel from lignocellulosic biomass via microbial bioconversion is also being explored.^2,3

Production of biodiesel from lignocellulosic biomass involves pretreatment to open up the complex plant cell wall followed by enzymatic hydrolysis to release sugars, microbial conversion of the sugars to lipids, and trans-esterification of the lipids into fatty acid esters (biodiesel). Pretreatment prior to enzymatic hydrolysis is essential for achieving high sugar yields from lignocellulosic biomass. Ammonia Fiber Expansion (AFEX) pretreatment, an alkaline pretreatment process that creates physicochemical alterations in the lignocellulose ultra and macro structure, is a leading pretreatment that provides a biomass substrate with high digestibility and high fermentation potential.^4,5 The high fermentability is achieved by preserving nutrients that are naturally in the plant biomass and by forming fewer inhibitory degradation products.^6,7 Due to its very high potential, AFEX has been scaled up to a pilot scale (1 ton per day) by MBI (Lansing, Michigan, USA). The successful demonstration of AFEX pretreatment at the pilot scale is paving the way for AFEX commercialization.

Enzymes are a major cost item for lignocellulosic biofuel production. Recently, a novel integrated biological process-Rapid Bioconversion with Integrated recycle Technology (RaBIT) (a.k.a Biomass Conversion Research Laboratory [BCRL] separate hydrolysis and fermentation [SHF] & simultaneous saccharification and co-fermentation [SSCF]) was developed at BCRL at Michigan State University for reduction of enzyme loading and improvement of ethanol productivity.⁸ The process performs enzymatic hydrolysis for only 24 h to exploit high initial hydrolysis rates and recycles unhydrolyzed biomass solids to the subsequent hydrolysis cycles. Since a large portion of enzymes are adsorbed on the unhydrolyzed solids at 24 h, these enzymes are recycled by this approach. The process was demonstrated on AFEX corn stover with five cycles. Enzyme loading was reduced by 38% and ethanol productivity was enhanced by three fold.

Oleaginous microorganisms that accumulate ≥20% of their body biomass as lipids have been widely investigated for lipids production.⁹ The fatty acid profile of microbial lipids is very similar to vegetable oils,⁹ rendering microbial lipids a suitable feedstock for biodiesel production. Besides, microbial lipids can be used for various applications, such as production of surfactants, lubricants, coatings, polymers, and solvents.^10,11 The theoretical maximum lipid yield during culture on glucose and xylose (two major sugars derived from lignocellulosic biomass) is estimated to be 0.32 and 0.34 g lipid per g consumed sugar, respectively.^9,12 However, a yield below 0.25 is typically reported, even when using synthetic media.^12–14 With degradation products acting as inhibitors in lignocellulosic biomass hydrolysate, reported lipid yields are even lower.¹⁵ Lipids accumulation in oleaginous microbes usually occurs during nitrogen source limiting conditions.⁹ The optimal C/N ratio of a medium for lipid fermentation is between 65 and 100.^16,17 AFEX corn stover hydrolysate has a C/N ratio of around 75 (calculated based on the free sugar, ammonium and free amino acids data reported in⁵) rendering it suitable medium for lipid fermentation.

Our previous screening research discovered a promising oleaginous yeast strain (Lipomyces tetrasporus NRRL Y-11562) that ferments both glucose and xylose efficiently in AFEX corn stover hydrolysate and accumulates a high lipid content.¹⁸ In the present work, we investigate various factors that affect lipids production by this strain in AFEX pretreated corn stover. RaBIT was adapted for lipids production and detailed mass balances and techno-economics were analyzed.

2. Materials and methods

2.1 AFEX pretreated corn stover

AFEX pretreated corn stover (AFEX–CS) with glucan and xylan contents of 33.9% and 20.9%, respectively, was used in this study. The AFEX pretreatment procedure was the same as described previously.¹⁹ Pretreatment conditions included: ammonia to biomass loading 1.0 g g⁻¹ dry biomass, water loading 0.6 g g⁻¹ dry biomass, temperature 140 °C and residence time 30 min.

2.2 Microorganisms and seed culture preparation

Lipomyces tetrasporus NRRL Y-11562, which was obtained as a lyophilized culture from the Agricultural Research Service Culture Collection, National Center for Agricultural Utilization Research, Peoria, Illinois, was used in this work. The seed culture of this strain was prepared in a 250 mL baffled conical flask with 50 mL YEP medium (10 g L⁻¹ yeast extract, 20 g L⁻¹ peptone, and 50 g L⁻¹ glucose). A frozen glycerol stock was used for seed culture inoculation. The initial optical density (OD₆₀₀) was approximately 0.1. Seeds were cultured at 29 °C and 200 rpm under aerobic conditions for 48 h. The cell density of seed culture reached an OD₆₀₀ of about 14. Initial OD₆₀₀ of 0.5 was used for all fermentation experiments.

2.3 Optimization of fermentation conditions in synthetic medium

Synthetic medium (SM) used in this work was adapted from¹⁴ with slight modification. The SM (with C/N ratio of 16) contained: 60.0 g L⁻¹ glucose, 30.0 g L⁻¹ xylose, 10.0 g L⁻¹ yeast extract (carbon content: 12%, nitrogen content: 7%), 8.0 g L⁻¹ (NH₄)₂SO₄, 1.0 g L⁻¹ KH₂PO₄, and 1.0 g L⁻¹ MgSO₄·7H₂O. To investigate the effect of nutrient level on lipids accumulation, nitrogen source (yeast extract and ammonia sulfate) contents in the medium were altered (see Table 1), resulting in media with C/N ratios of 53, 75, 105, and 173. To determine the optimal pH and temperature, lipid fermentations were carried out at various pHs (5.0–6.5) and temperatures (25–32 °C) in SM with C/N ratio of 75, which was the same as SM with C/N ratio of 16 except that yeast extract concentration was changed to 7.0 g L⁻¹ and (NH₄)₂SO₄ was removed. An aliquot of the culture was withdrawn for analyses at designated times during the cultivation. The results shown in figures and tables represent the mean of duplicate experiments.

Table 1 Effect of nutrient (yeast extract and ammonia sulfate) level on lipid fermentation by L. tetrasporus NRRL Y-11562 in synthetic media

Yeast extract (g L^−1)	Ammonia sulphate (g L^−1)	C/N^a	Time^b (h)	DCW (g L^−1)	Lipid content (%)	Lipid yield (g g^−1)	Lipid conc. (g L^−1)
a To calculate the C/N ratio in the medium, it was assumed that yeast extract contained 12 wt% of carbon and 7 wt% of nitrogen.b The time required to reach the maximal lipid concentration.c AFEX–CS hydrolysate has a C/N ratio of 75. Besides yeast extract and ammonia sulfate indicated in the table, synthetic media contained 60.0 g L^−1 glucose, 30.0 g L^−1 xylose, 1.0 g L^−1 KH2PO4, and 1.0 g L^−1 MgSO4·7H2O.
10.0	8.0	16	72	21.70 ± 0.35	57.27 ± 6.61	0.14 ± 0.01	12.42 ± 1.23
10.0	0.0	53	72	25.35 ± 0.35	49.05 ± 6.14	0.14 ± 0.01	12.41 ± 1.38
7.0	0.0	75^c	72	22.33 ± 1.31	58.52 ± 0.97	0.15 ± 0.01	13.06 ± 0.55
5.0	0.0	105	96	20.23 ± 0.18	64.61 ± 1.05	0.15 ± 0.01	13.07 ± 0.33
3.0	0.0	173	96	19.73 ± 0.11	67.41 ± 0.55	0.15 ± 0.01	13.30 ± 0.97

---

2.4 Minimal washing of AFEX–CS

Washing of AFEX–CS was conducted by spraying distilled water on AFEX–CS at a ratio of 5 mL water per g dry AFEX–CS as described before.⁴ The wetted AFEX–CS was soaked for a while and then was pressed to reduce moisture content to 57 ± 2% (total weight basis). The water extract was used for the fermentation inhibition study and washed AFEX–CS was enzymatically-hydrolyzed.

2.5 Enzymatic hydrolysis and yeast cultures in hydrolysate

Enzymatic hydrolysis of unwashed AFEX–CS was conducted at glucan loadings of 6.0, 7.5 and 9.0 wt% glucan loadings, corresponding to solids loadings of 17.7, 22.1 and 26.5 wt%, respectively. Enzymatic hydrolysis of washed AFEX–CS was performed at 7.5 wt% glucan loading. Commercial enzyme cocktails Cellic Ctec3 and Cellic Htec3 (Novozymes, Franklinton, NC) were used for enzymatic hydrolysis with a loading of 11.2 mg g⁻¹ glucan for Ctec3 and 11.3 mg g⁻¹ glucan for Htec3. Enzymatic hydrolysis was carried out for 48 h at 50 °C and 250 rpm in a 250 mL baffled flask with a working mixture of 90 g. After enzymatic hydrolysis, liquid hydrolysate was obtained by centrifuge and sterile filtration. Liquid hydrolysate was then fermented as is (without supplementation of any nutrients) aerobically using Lypomyces sp. in a 250 mL baffled flask with 50 mL working volume at 27 °C, pH 5.5, initial OD₆₀₀ = 0.5 and 200 rpm.

2.6 Conventional SHF process vs. RaBIT process for lipid production

For this comparison, the conventional SHF process performed enzymatic hydrolysis at 7.5% glucan loading of both washed and unwashed AFEX–CS for 72 h under the same conditions for enzymatic hydrolysis as described above, followed by lipid fermentation using liquid hydrolysate as described above.

The RaBIT process performed enzymatic hydrolysis for 24 h at 7.5% glucan loading of washed AFEX–CS, followed by centrifugation at 5300 rpm for 20 minutes (Fig. 1). The resulting unhydrolyzed solids were recycled to the next enzymatic hydrolysis tank. The supernatant (hydrolysate) was used for lipid fermentation at 27 °C, pH 5.5, initial OD₆₀₀ = 0.5 and 200 rpm. Enzymatic hydrolysis for cycles 1, 2, 3 was conducted in a 250 mL baffled flask while cycles 4 and the last step were performed in a 500 mL baffled flask. The enzyme loading for the first cycle was the same as the conventional processes. 60% of the cycle 1's enzyme loading was applied in cycle 2. 50% of the cycle 1's enzyme loading was used for cycles 3 to 4. For enzymatic hydrolysis in the last step, 60 mL water was added without any addition of enzymes or fresh biomass. All the experiments were conducted in duplicates with the average and standard deviation shown in figures.

Fig. 1 Comparison of conventional SHF and RaBIT process for lipid production (adapted from Jin et al. 2012).

2.7 Analytical methods and mass balance

Sugars were analyzed by using a Shimadzu HPLC-RI detector system. An Aminex HPX-87H carbohydrate analysis column (Bio-Rad, Hercules, CA) equipped with a guard cartridge (Bio-Rad) was used for quantifying sugars in the samples. Degassed 5 mM aqueous sulfuric acid was used as the mobile phase at 0.6 mL min⁻¹ at a column temperature of 60 °C. The injection volume was 10 μL with a run time of 20 min.

Yeast cell biomass concentration, expressed as dry cell weight (DCW) per liter, was calculated based the OD₆₀₀ and a calibration curve in which the absorbance and dry weight of samples were correlated. The colorimetric method based on the sulfo-phospho-vanillin reaction was used for the determination of total lipid in microbial cells.²⁰ Lipid yield was defined as the amount of lipid produced per g consumed sugar. Lipid content was defined as the percentage of lipid accumulated per 100 g DCW. Mass balances and consumed sugars were calculated based on the analysis of monomeric, oligomeric and polymeric sugars in samples as described earlier.⁴

2.8 Techno-economic analysis

NREL 2011 model for a 2000 ton per day lignocellulosic ethanol biorefinery²¹ was modified and used for techno-economic analysis in Microsoft Excel spreadsheets. The modified model assumed biomass handling and pretreatment were conducted in the local biomass processing depots (LBPDs)²² and thus were removed from the model. The biomass cost, pre-processing cost, and ammonia usage were assumed to be %50 per tonne, %60 per tonne, and 3.85 kg ammonia per tonne biomass. Onsite cellulase production was also removed in the model and a fixed enzyme cost (cellulases: %3600 per tonne protein; hemicellulases: %4500 per tonne protein) estimated based on NREL model was used. Ethanol fermentation and distillation were replaced by lipid fermentation and lipid processing in Aspen. Lipid fermentation and lipid processing was modeled according to Koutinas et al.'s work.²³ Capital cost, raw material cost, energy cost, total lipid and electricity revenue were calculated based on inputs regarding biomass composition, enzymatic hydrolysis and fermentation performance.

3. Results and discussion

3.1 Optimization of lipid fermentation conditions in synthetic medium

Culture conditions known to be critical for lipid production including nutrient level, temperature and pH²⁴ were investigated for L. tetrasporus NRRL Y-11562 in the synthetic medium. For investigating the effect of nutrient level, carbon source glucose and xylose were fixed at 60 g L⁻¹ and 30 g L⁻¹, respectively (modeled on 6% glucan loading AFEX–CS hydrolysate), and nitrogen source/nutrient source (yeast extract and ammonia sulfate) concentrations were varied, resulting in SM with different C/N ratios (Table 1). Ammonia sulfate was only included in the SM with C/N ratio of 16. As shown in Table 1, the SM with C/N ratio of 16 resulted in a DCW of 21.7 g L⁻¹ with lipid content of 57.3%, lipid yield of 0.14 g g⁻¹ consumed sugar and lipid concentration of 12.4 g L⁻¹. As (NH₄)₂SO₄ was removed from the fermentation medium, which enhanced the C/N ratio to 53, surprisingly, the DCW increased to 25.4 g L⁻¹ while lipid content decreased to 49.1% with lipid yield and lipid concentration unchanged. A further decrease in nutrient level by reducing the concentration of yeast extract, which increased the C/N ratio, resulted in decreased the DCW and increased the lipid content as expected. The maximum lipid content (67.4%) was achieved in the SM with a C/N ratio of 173 at a low nutrient (yeast extract) level. Lipid yield and lipid concentration were also slightly increased with a lower nutrient level. However, the increase in lipid concentration, which is the most important parameter for lipids production, was insignificant, which means almost the same amount of lipids was produced in the tested SM. Moreover, it is worthy to note that SM with C/N ratios of 105 and 173 required 96 h to reach the maximum lipid production instead of 72 h for lower C/N ratios (Table 1). Since AFEX–CS hydrolysate has a C/N ratio of around 75, the SM with C/N ratio of 75 was used for the following experiments.

The optimal temperature and pH for oleaginous yeast cultures are typically observed between 25 °C and 30 °C, and 4.0–7.0, respectively.^9,12,17 Several temperatures and pHs in these ranges were investigated (Fig. 2). Culturing at 27 °C led to the best combination of growth (e.g. DCW) and lipid production, while higher temperature decreased lipid production (Fig. 2A). No significant difference was observed among the pHs tested (Fig. 2B). Enzymatic hydrolysate has a pH of approximately 5.0. To minimize alkaline consumption for pH adjustment, pH 5.5 was chosen for the following experiments.

Fig. 2 Effect of temperature (A) and pH (B) on lipid fermentation in synthetic medium with C/N ratio of 75. Fermentations were performed for 72 h. The initial pH used for (A) was 5.5 and the temperature used for (B) was 27 °C.

3.2 AFEX–CS hydrolysate vs. synthetic medium for lipid fermentation

Lipid fermentations with L. tetrasporus NRRL Y-11562 in SM (C/N = 75) and AFEX–CS hydrolysates derived from enzymatic hydrolysis at 6.0%, 7.5% and 9.0% glucan loadings were compared under identical culture conditions (Fig. 3). The 6.0% glucan loading AFEX–CS hydrolysate contained similar sugar concentrations and C/N ratio as SM. However, much lower lipid concentration and lipid yield were achieved (Fig. 3A). It is interesting to note that DCW was significantly higher compared to SM though the lipid content was far lower (Fig. 3B). AFEX–CS hydrolysate contains numerous nutrient elements including vitamins and trace elements, which might have stimulated growth.⁵ Nevertheless, the presence of degradation products generated during pretreatment likely inhibited lipid biosynthesis. With increased glucan loading, DCW increased likely due to more nutrients and sugars available for cell growth, while lipid content decreased (Fig. 3B). Since 7.5% glucan loading reached the same lipid yield and higher lipid concentration compared to 6.0% glucan loading, 7.5% glucan loading was chosen for further studies.

Fig. 3 Effect of AFEX–CS solids loading on lipid fermentation (A & B) and time course of lipid fermentation (C & D). (A) Effect of solids loading on lipid concentration and lipid yield; (B) effect of solids loading on lipid content and DCW; (C) time course of lipid fermentation in synthetic medium (C/N = 75); (D) time course of lipid fermentation in 7.5% glucan loading AFEX–CS hydrolysate. Fermentations were performed at 27 °C and pH 5.5. The fermentation times required to reach the maximal lipid concentration (shown in the figure) for SM (C/N = 75), 6.0%, 7.5%, and 9.0% glucan loading AFEX–CS hydrolysate were 72, 120, 120, 168 h, respectively. Enzymatic hydrolysis was conducted for 48 h. The glucose concentrations for the 6.0%, 7.5% and 9.0% glucan loading AFEX–CS hydrolysates were 56.5, 67.4 and 81.1 g L⁻¹, respectively. The xylose concentrations for the 6.0%, 7.5% and 9.0% glucan loading AFEX–CS hydrolysates were 30.5, 37.8 and 45.1 g L⁻¹, respectively.

The time-course of lipid fermentation showed that lipid concentration reached the maximum at 72 h of fermentation in SM with both glucose and xylose nearly exhausted (Fig. 3C). Cell growth and lipid production occurred simultaneously with lipid production rate higher during late fermentation phase. A long lag phase (close to 48 h) was observed during culturing in 7.5% glucan loading AFEX–CS hydrolysate (Fig. 3D), which might be caused by the presence of degradation products generated during pretreatment. The sugars were then consumed very rapidly and lipid production peaked at 120 h (Fig. 3D). Xylose fermentation in lignocellulosic hydrolysate for ethanol production is slow (taking several days to complete) due to issues of ethanol inhibition, degradation products inhibition and anaerobic fermentation properties and has been limiting ethanol productivity.^8,25,26 It is interesting to note that xylose was consumed very rapidly during this aerobic lipid fermentation if long lag phase (∼48 h) is not counted (Fig. 3D). Directed evolution is a promising strategy for shortening or even eliminating the lag phase.²⁷

3.3 The cause for reduced lipid production in AFEX–CS hydrolysate

Since nutrients and sugar concentrations almost proportionally increased as solids loading increased (sugar concentrations were shown in the caption of Fig. 3), the C/N ratios of different solids loadings were similar. As different solids loading cultures were compared, lipid content and lipid yield decreased with increased solids loadings (Fig. 3). The degradation products might have inhibited lipid biosynthesis. To further investigate the effect of degradation products, water extract of AFEX–CS was added into SM (C/N = 75) to reach a final concentration of 0–6.8% glucan loading equivalent and lipid fermentations were conducted (Fig. 4A and B). With additions of as low as 1.6% and 3.3% glucan loading equivalent water extract, significant increases of cell growth (DCW) and significant reductions of lipid content were observed (Fig. 4B). Water extract contained both degradation products and nutrients. Some particular nutrients might be present in AFEX–CS water extract rather than in yeast extract as increasing the yeast extract concentration did not boost the growth that much (Table 1). Stimulation of growth likely channeled carbon flux into cell biomass production rather than lipid production. Increasing the water extract concentration to 4.9% glucan loading equivalent further enhanced cell growth. However, this enhancement was not significant, while the reduction of lipid content was significant. It is likely that starting from 4.9% glucan loading degradation products inhibition became a dominating factor that affected lipid production. Overall, addition of water extract into SM led to lower lipid concentration and lipid yield (Fig. 4A).

Fig. 4 Effect of degradation products in AFEX–CS on lipid fermentation. (A) Effect of AFEX–CS water extract concentration on lipid concentration and lipid yield (SM with C/N = 75 was used); (B) effect of AFEX–CS water extract concentration on lipid content and DCW (SM with C/N = 75 was used); (C) effect of minimal washing of AFEX–CS on lipid concentration and lipid yield; (D) effect of minimal washing of AFEX–CS on lipid content and DCW. Fermentations were performed at 27 °C and initial pH 5.5. Water extract was added into the synthetic medium at various concentrations. Water extract of AFEX–CS at the concentration of 6.8% glucan loading equivalent contained 0.3 g L⁻¹ glucose and 0.8 g L⁻¹ xylose. Minimal washing effect was investigated based on 7.5% glucan loading for enzymatic hydrolysis and fermentation. The fermentation times required to reach the maximal lipid concentration for unwashed AFEX–CS hydrolysate and washed AFEX–CS hydrolysate were 120 h and 96 h, respectively.

Another study was conducted to further confirm the above observations. AFEX–CS hydrolysate (6.0% glucan loading) was diluted by 1.25, 1.67 and 2.5 fold, and then sugars were added to make both glucose and xylose concentrations the same as in undiluted hydrolysate. Fermentation results showed that 1.25 fold dilution significantly improved cell growth and lipid content (ESI Fig. S1†), while further dilution had minor effects on each. These results confirmed that degradation products inhibition was the dominating factor affecting lipid fermentation when the solids loading was above approximately 5% glucan loading. Diluting the degradation products below a critical threshold improved lipid production. When degradation products were not dominating, further dilution, which further decreased degradation products and nutrient level, did not enhance lipid production.

Due to the ammonolysis reaction occurring during AFEX pretreatment, degradation products generated were majorly in amide form rather than in acid form.⁶ There were around 6 g L⁻¹ acetamide and 3.6 g L⁻¹ total phenolic amides present in 7.5% glucan loading AFEX–CS hydrolysate (ESI Table S1†), which likely played a critical role in affecting lipid production, although amides are less inhibitory than acids.²⁸ Degradation products were also found significantly inhibiting lipid production by other researchers.^29,30

As reduction of degradation products level below 5% glucan loading is critical to achieve more lipid production, minimal washing of AFEX–CS was implemented. The washed AFEX–CS was then enzymatically hydrolyzed and fermented. By this approach, the lipid concentration, lipid yield, lipid content were increased from 8.4 g L⁻¹, 0.08 g g⁻¹ consumed sugar, and 24% to 10.7 g L⁻¹, 0.10 g g⁻¹ consumed sugar and 33%, respectively (Fig. 4C and D). The culture time required to reach the maximal lipid concentration was shortened from 120 h to 96 h.

3.4 Lipid production through RaBIT process

RaBIT process reduces enzyme loading, quickens enzymatic hydrolysis and hence reduces overall processing cost.⁸ Therefore, lipids production using this process was evaluated on washed AFEX–CS at 7.5% glucan loading. The enzyme loading of 22.5 mg g⁻¹ glucan, the same as for regular enzymatic hydrolysis or SHF, was used for the RaBIT cycle 1 enzymatic hydrolysis. Enzyme loadings for cycle 2, cycle 3, and cycle 4 were 60%, 50% and 50%, respectively of the cycle 1. Four cycles of enzymatic hydrolysis resulted in consistent sugar concentrations (∼60 g L⁻¹ glucose and 30 g L⁻¹ xylose) (Fig. 5A). The last step resulted in lower sugar concentration since no fresh biomass or enzyme was added (Fig. 1). The resulting hydrolysates were used for lipids fermentation and led to production of around 10 g L⁻¹ lipid for cycle 1–4 and 6.6 g L⁻¹ lipid for the last step. The lipid yield reached 0.1 g lipid per g consumed sugar (Fig. 5B). These results validate the RaBIT process for lipid production.

Fig. 5 Lipid production through RaBIT process. (A) Sugar concentrations after 24 h enzymatic hydrolysis for cycle 1–4 and 48 h for last step; (B and C) fermentation performance. The enzyme loading for cycle 1 is 22.5 mg g⁻¹ glucan (100%). Enzyme loadings used for cycles 2, 3, and 4 were 60%, 50%, and 50%, respectively of the cycle 1. The average enzyme loading was 14.6 mg g⁻¹ glucan.

3.5 Mass balance comparison of different processes for lipid production

The processes compared include SHF on unwashed AFEX–CS, SHF on washed AFEX–CS and RaBIT on washed AFEX–CS (Fig. 6). SHF on unwashed AFEX–CS converted 79.2% of glucan and 81.9% of xylan in AFEX–CS to soluble sugars (monomeric and oligomeric sugars) in 72 h. It then took 120 h for the lipid fermentation to reach maximum lipid production. This process yielded 36.7 g lipids and 97.9 g yeast meals from 1 kg AFEX–CS (Fig. 6A). It is interesting to note that a large portion of oligosaccharides were consumed during fermentation, which was also observed during fermentations in SHF and RaBIT on washed AFEX–CS (Fig. 6B and C). The cellulases and hemicellulases might be still actively hydrolyzing the oligosaccharides in the hydrolysate during fermentation to monosaccharides which were then consumed by the L. tetrasporus NRRL Y-11562. L. tetrasporus might also directly consume oligosaccharides like Cryptococcus curvatus (another oleaginous yeast) did,³¹ which needs further investigation. Minimal washing of AFEX–CS removed little amount of sugars (mostly oligosaccharides, Fig. 6B). SHF on washed AFEX–CS yielded similar amount of soluble sugars during enzymatic hydrolysis, but more lipids during fermentation compared to SHF on unwashed AFEX–CS (Fig. 6B). This is due to the enhancement of lipid yield from 0.08 to 0.10 g g⁻¹ consumed sugar through removal of degradation products via washing. With less degradation products inhibition, fermentation time was shortened from 120 h to 96 h. Washing also removed some nutrients and caused the reduction of yeast meal production (Fig. 6B). RaBIT used 35% less enzyme for hydrolysis and yielded similar amount of glucose (monomeric plus oligomeric) but less xylose (Fig. 6C). Lipid and yeast meal produced through RaBIT was 39.1 g and 69.2 g, respectively, per kg AFEX–CS, which was slightly lower than SHF on washed AFEX–CS.

Fig. 6 Mass balance of different processes. (A) SHF process without washing, (B) SHF process with washing and (C) RaBIT process with washing. 7.5% glucan loading was used for these experiments.

3.6 Comparison of lignocellulosic ethanol and lipids production

Production of ethanol from lignocellulosic biomass has been extensively studied, while study of lipid production from lignocellulosic biomass has just received attention in recent years. Theoretically, ethanol fermentation (anaerobic) can recover 82.1% of the energy and 66.5% of the carbon present in sugar substrates (glucose and xylose). Lipid fermentation (aerobic) recovers similar amount of energy (79.4%) and carbon (∼61.4%). From 1 kg of corn stover, 312.6 g ethanol or 200.9 g lipids can potentially be produced (Fig. 7). However, our experimental results showed that ∼200 g ethanol or ∼40 g lipids were produced from 1 kg of corn stover with the energy recovery of 52.5% (ethanol) or merely 15.8% (lipids). Sugar yield of the enzymatic hydrolysis step is limiting the final ethanol yield and the ethanol yield during fermentation has achieved as high as >90% of the theoretical maximum.³² In contrast, lipid fermentation yield (0.08 g g⁻¹ consumed sugar, 23.5% of the theoretical maximum) is limiting the lipid production from lignocellulosic biomass. Besides, lipid fermentation is an aerobic process and thus requires more energy input during fermentation compared to ethanol fermentation. Although lipid fermentation generated much more yeast meals (90 g) compared to ethanol fermentation (27 g), which should serve as a value-added product, the economics of lipid production from lignocellulosic biomass is currently not comparable to ethanol production.

Fig. 7 Comparison of ethanol and lipid production from corn stover via AFEX pretreatment, enzymatic hydrolysis and fermentation. Ethanol data was adapted from (Lau and Dale 2009b). Lipid production data of SHF on unwashed AFEX–CS (Fig. 6A) was used for this comparison. Energy densities of sugar (glucose/xylose), ethanol and lipids used for this calculation were 15.55, 25.00, and 37.70 MJ kg⁻¹, respectively. Energy content in yeast meal was calculated based on 50% protein content in yeast meal and protein energy density of 16.80 MJ kg⁻¹.

3.7 Techno-economic analysis of lignocellulosic lipids production

Techno-economic analysis based on our experimental data (SHF without washing) showed that the cost for microbial biodiesel production from corn stover was ∼%15 per gallon with considering glycerol and yeast meal as co-products, which have prices of %120 and %380 per metric ton, respectively. This cost is too high compared to current biodiesel price (∼%4 per gallon). Sensitivity analysis showed that lipid production cost was very sensitive to lipid yield and sugar conversion. While there is still some room for sugar conversion improvement (via advancing pretreatment and enzyme technologies), it is limited since ∼80% sugar conversion has already been achieved (Fig. 6). In contrast, there is much room for lipid yield improvement (the current yield only reached 23.5% of the maximum). The cost could be reduced to ∼%5 per gallon by increasing lipid yield to 0.24 g g⁻¹ consumed sugar which has been achieved in synthetic media.¹⁴ As compared in a recent review paper, most of the lipid yields currently achieved on lignocellulosic biomass were around 0.08–0.2 g g⁻¹ consumed sugar.³⁰ Identification of responsible degradation products that caused low liquid yield, removal/reduction of degradation products inhibition (Fig. 4), direct evolution of fermentation organism and metabolic engineering can be efficient ways to enhance lipid yield.^30,33 Reducing enzyme use (e.g. via RaBIT) during enzymatic hydrolysis could further reduce the production cost. However, co-production of high-value products (e.g. omega-6 and omega-3 fatty acids, which have ∼100 times higher price than typically observed lipids) seems essential to explore to make lignocellulosic lipid biorefinery economically viable.³⁰

4. Conclusions

Lipids production from AFEX corn stover using L. tetrasporus NRRL Y-11562 via SHF and RaBIT processes was investigated. Degradation products inhibited lipids fermentation at a level >5% glucan loading. RaBIT reduced enzyme loading. However, the key issue was low lipid yield, which resulted in high cost of lignocellulosic lipids production. The economics of lipid production from lignocellulosic biomass is currently not comparable to ethanol production. Reduction of degradation products inhibition, improvement of lipid yield and co-production of high value products could potentially make lignocellulosic lipids economically viable.

Acknowledgements

This work is supported by the China Scholarship Council. We would like to thank Novozymes for supplying us commercial enzymes for this work, Charles Donald Jr for preparing AFEX-pretreated corn stover and Christa Gunawan for analyzing HPLC samples. We would also like to thank the members of the Biomass Conversion Research Laboratory (BCRL) at Michigan State University for their valuable suggestions. We also thank Dr Cletus P. Kurtzman for supplying the culture for this research.

References

1. B. E. Dale and R. G. Ong, Biotechnol. Prog., 2012, 28, 893–898.
2. C. Huang, X. F. Chen, L. Xiong, X. D. Chen, L. L. Ma and Y. Chen, Biotechnol. Adv., 2013, 31, 129–139.
3. A. Yousuf, Waste Manag., 2012, 32, 2061–2067.
4. M. W. Lau and B. E. Dale, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 1368–1373.
5. M. W. Lau, B. D. Bals, S. P. S. Chundawat, M. Jin, C. Gunawan, V. Balan, A. D. Jones and B. E. Dale, Energy Environ. Sci., 2012, 5, 7100–7110.
6. S. P. S. Chundawat, R. Vismeh, L. N. Sharma, J. F. Humpula, L. D. Sousa, C. K. Chambliss, A. D. Jones, V. Balan and B. E. Dale, Bioresour. Technol., 2010, 101, 8429–8438.
7. M. W. Lau, C. Gunawan and B. E. Dale, Biotechnol. Biofuels, 2009, 2, 30.
8. M. Jin, C. Gunawan, N. Uppugundla, V. Balan and B. Dale, Energy Environ. Sci., 2012, 5, 7168–7175.
9. C. Ratledge and J. P. Wynn, in Advances in Applied Microbiology, ed. A. I. Laskin, J. W. Bennett and G. M. Gadd, 2002, vol. 51, pp. 1–51.
10. J. C. Ronda, G. Lligadas, M. Galià and V. Cádiz, Eur. J. Lipid Sci. Technol., 2011, 113, 46–58.
11. U. Biermann, U. Bornscheuer, M. A. R. Meier, J. O. Metzger and H. J. Schäfer, Angew. Chem., Int. Ed., 2011, 50, 3854–3871.
12. S. Papanikolaou and G. Aggelis, Eur. J. Lipid Sci. Technol., 2011, 113, 1031–1051.
13. C. Ratledge and Z. Cohen, Lipid Technol., 2008, 20, 155–160.
14. X. Zhao, C. M. Hu, S. G. Wu, H. W. Shen and Z. K. Zhao, J. Ind. Microbiol. Biotechnol., 2011, 38, 627–632.
15. H. B. Xie, H. W. Shen, Z. W. Gong, Q. Wang, Z. B. K. Zhao and F. W. Bai, Green Chem., 2012, 14, 1202–1210.
16. J. D. Weete and D. J. Weber, Lipid biochemistry of fungi and other organisms, Plenum Press, 1980.
17. J. M. Ageitos, J. A. Vallejo, P. Veiga-Crespo and T. G. Villa, Appl. Microbiol. Biotechnol., 2011, 90, 1219–1227.
18. P. Slininger, B. Dien, C. Kurtzman, B. Moser, R. Evangelista, B. Saha, M. Cotta, V. Balan, M. Jin, L. D. Sousa and B. Dale, 2013.
19. V. Balan, B. Bals, S. P. Chundawat, D. Marshall and B. E. Dale, Methods Mol. Biol., 2009, 581, 61–77.
20. J. Izard and R. J. Limberger, J. Microbiol. Methods, 2003, 55, 411–418.
21. D. Humbird, R. Davis, L. Tao, C. Kinchin, D. Hsu and A. Aden, NREL report, 2011.
22. B. D. Bals and B. E. Dale, Bioresour. Technol., 2012, 106, 161–169.
23. A. A. Koutinas, A. Chatzifragkou, N. Kopsahelis, S. Papanikolaou and I. K. Kookos, Fuel, 2014, 116, 566–577.
24. J. M. Ageitos, J. A. Vallejo, P. Veiga-Crespo and T. G. Villa, Appl. Microbiol. Biotechnol., 2011, 90, 1219–1227.
25. M. Jin, C. Sarks, C. Gunawan, B. D. Bice, S. P. Simonett, R. A. Narasimhan, L. B. Willis, B. E. Dale, V. Balan and T. K. Sato, Biotechnol. Biofuels, 2013, 6, 108.
26. M. Jin, V. Balan, C. Gunawan and B. E. Dale, Bioresour. Technol., 2012, 111, 294–300.
27. P. Slininger, S. Thompson, M. Shea-Andersh, V. Balan, L. Da Costa Sousa, B. Dien and B. Dale, US Pat., Application Serial No. 14/167,430 (USDA Docket No. 018311), 2014.
28. X. Tang, L. da Costa Sousa, M. Jin, S. P. Chundawat, C. K. Chambliss, M. W. Lau, Z. Xiao, B. E. Dale and V. Balan, Biotechnol. Biofuels, 2015, 8, 1.
29. S. Galafassi, D. Cucchetti, F. Pizza, G. Franzosi, D. Bianchi and C. Compagno, Bioresour. Technol., 2012, 111, 398–403.
30. M. Jin, P. J. Slininger, B. S. Dien, S. Waghmode, B. R. Moser, A. Orjuela, L. d. C. Sousa and V. Balan, Trends Biotechnol., 2015, 33, 43–54.
31. Z. Gong, Q. Wang, H. Shen, L. Wang, H. Xie and Z. K. Zhao, Biotechnol. Biofuels, 2014, 7, 13.
32. M. W. Lau and B. E. Dale, Proc. Natl. Acad. Sci. U. S. A., 2009, 106, 1368–1373.
33. M.-H. Liang and J.-G. Jiang, Prog. Lipid Res., 2013, 52, 395–408.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra01134e

---