The fungal endophyte, Ascocoryne sarcoides, produced aviation, gasoline and diesel-relevant hydrocarbons when grown on multiple substrates including cellulose as the sole carbon source. Substrate, growth stage, culturing pH, temperature and medium composition were statistically significant factors for the type and quantity of hydrocarbons produced. Gasoline range (C5-C12), aviation range (C8-C16) and diesel range (C9-C36) organics were detected in all cultured media. Numerous non-oxygenated hydrocarbons were produced such as isopentane, 3,3-dimethyl hexane and d-limonene during exponential growth phase. Growth on cellulose at 23°C and pH 5.8 produced the highest overall yield of fuel range organics (105 mg * g·biomass-1). A change in metabolism was seen in late stationary phase from catabolism of cellulose to potential oxidation of hydrocarbons resulting in the production of more oxygenated compounds with longer carbon chain length and fewer fuel-related compounds. The results outline rational strategies for controlling the composition of the fuel-like compounds by changing culturing parameters.
Many strategies are being explored to replace or supplement petroleum-derived transportation fuels. These include microbial conversion of cellulosic sugars after extensive pretreatment, as well as chemical and thermal conversion technologies [
Ascocoryne sarcoides (NRRL 50072) is a cellulolytic fungal endophyte capable of direct conversion of cellulosic materials to fuel-related hydrocarbon compounds such as alkanes, alcohols and aromatics [
Microbial utilization of cellulose as a carbon source requires production of cellulase enzymes. Almost 400 cellulase and cellulose-related genes have been identified in the annotated A. sarcoides genome, and the expression of many of these genes has been observed when A. sarcoides was grown on cellobiose and cellulose, but not when grown on a cellulose-free medium [
Here we report the results of pH, growth medium, and temperature on fuel-related hydrocarbon production by A. sarcoides by measuring fuel compounds in both the gas and liquid phases. Multiple methods of analyzing VHC were used to identify and quantify the compounds produced. This work demonstrates the potential of A. sarcoides to produce a variety of VHC compounds with fuel potential and provides guidance for efforts being made on other organisms having this capability.
A. sarcoides (NRRL 50072) was grown in 250 mL baffled flasks or bottles on a minimal cellulose medium (CM) consisting of (per liter): 50 µm microcrystalline cellulose (20 g) (Acros Organics), ammonium chloride (5 g) (Fisher Chemical), NaH2PO4·2H2O (2.75 g) (Fisher Chemical), MgSO4·7H2O (0.86 g) (Fisher Chemical), Ca(NO3)2·4H2O (0.28 g) (Fisher Chemical, yeast extract (0.05 g) (Fisher BioReagents), and trace salts: KCl (60 mg), KNO3 (80 mg), FeCl3 (2 mg), MnCl2 (5 mg), ZnSO4 (2.5 mg), H3BO3 (1.4 mg), and KI (0.7 mg). 100 mmol KH2PO4 was added as a buffer. CM was modified with glucose (Fisher Chemical) in place of cellulose for glucose medium (GM), and with 90,000 average molecular weight sodium carboxymethyl cellulose (CMC) (Sigma Aldrich) in place of cellulose for CMC medium. To test the impact of nutrient limitation, the GM recipe listed above was increased three fold for all components except calcium and magnesium compounds. Initial pH was adjusted to 3.5, 5.0, 5.6 and 6.5 with 2 M NaOH or HCl.
Response surface design is a statistical method that determines the minimum number of experiments necessary to predict the optimal response from a system. Response surface design and analysis have been used for optimization of fungal systems [
In the experiments using CM and GM, the variables included temperature (13˚C, 16.5˚C, 20˚C, 23˚C and 28˚C), pH (3.5, 4.5, 5, 5.6, 6 and 6.5), oxygen condition (aerobic, sealed batch and anoxic) and nutrient level of the medium (1×, 2×, 2.3× and 3×). The combination of variables was guided by the experimental design. Cultures covered with permeable cloth were considered aerobic; those sealed with stoppers containing ambient air in the headspace were denoted sealed batch; those sparged with N2 then sealed with oxygen impermeable stoppers were denoted anoxic. Cultures were shaken in temperature controlled incubators at 150 rpm. All cultures were inoculated with a 7-day culture prepared with MicrobankTM Bead frozen stocks (−80˚C) and grown in CM with cellulose replaced with 0.5 g cellobiose and 0.45 g yeast extract.
Biomass was determined by cell dry weight as described previously [
Throughout the growth period, samples of culture liquid were collected ranging from 7 to 14 days for glucose cultures and up to 60 days for sealed cellulose cultures. The liquid samples were centrifuged to remove cells, and the cell-free culture liquid added to a Nuclear Magnetic Microscopy (NMR) tube with 10% deuterium oxide (D2O) by volume. Proton NMR was run on a Bruker DRX600 spectrometer operating at a proton frequency of 600.13 MHz. Data were collected using a 1D Nuclear Overhauser Effect Spectroscopy (NOSEY) experiment with 100 ms mixing time and 2 second presaturation of the water signal. A total of 32,000 data points were collected, with a sweepwidth of 7183.91 Hz. The 64 scans were averaged for each data set with a total recycle time of 4.28 seconds. All data sets were collected with the same receiver gain, so the intensity of the signals could be directly compared.
The data were processed using an exponential window with a 0.5 Hz line broadening and a final size of 16,000 real points. All results were corrected for background with un-inoculated controls. This technique measures signal of protons (H+) on the liquid bound compounds and is therefore quantitative on a molecular level. The signal peaks and areas are grouped based on structure and elemental compositions and compared with control samples. The peak spectra were grouped by aromatics, non-oxygenated hydrocarbons and sugars. Oxygenated hydrocarbons are included in the sugars peak, so cannot be differentiated with this technique. The results are nonspecific as far as carbon length, but can indicate structure (e.g. branching).
Statistical analysis of VHC concentration data from NMR spectra was completed using linear mixed effects (lme) models fit by the lme4 package [
A. sarcoides (NRRL 50072) was grown in 10 L bottles with 4 L of media CM, GM, both at three fold concentration, and Potato Dextrose Broth (PDB) at 24 g/L. CM and GM were brought to pH 5.8 with 2 M NaOH. Continuous house air supply at 1 L/min passed through the bottle headspace, and the cultures were incubated at 23˚C with shaking at 100 rpm. At the conclusion of the experiments, biomass was determined by cell dry weight as described previously [
The air passing through the bottle headspace was routed with teflon tubing to an external column containing 10 g of CarbotrapTM B and 10 g of CarbotrapTM C (Supelco, Bellefonte, PA)(Booth et al., 2011). CarbotrapTM B collects compounds with a carbon range of C5-C12 and CarbotrapTM C collects compounds from C12-C20. Volatiles were collected on one column from day 5 to day 17 of growth and for cellulose, on another from day 17 to day 32. Volatiles were also collected from un-inoculated controls and subtracted from culture results. CarbotrapTM columns were conditioned before collection as described previously [
Liquid from the 4 L sparged bioreactor cultures was extracted using the U.S. Environmental Protection Agency (EPA) method 3510. The extracted liquid was analyzed by GC-MS by Pace Analytical Labs (Billings, MT) for gasoline range organics (GRO) by EPA method 8015/8021 and for diesel range organics (DRO) by EPA 8015 modified. Peak areas were used to quantify total DRO and GRO, correcting for un-inoculated controls. DRO and GRO are reported separately from the external column collection and desorption described below.
Volatile compounds collected on the external Carbotrap column were desorbed and analyzed as described previously with only slight modifications [
Data processing was performed with MassHunter Pro B.04.00 and Mass Profiler Pro B.04.00 (Agilent Technologies, Santa Clara, CA). Cross-comparison of spectra across all substrates was completed before compound identification to eliminate errors associated with database searches and identity allocation. Spectra from the uninoculated controls were compared with samples in Mass Profiler Pro to subtract peaks similar to control peaks. The software compared peak spectra with similar retention times, so slight differences in retention time did not impact the results. A fold change of 10,000 was used to report spectra values, though a fold change of 10 gave the same results. The reported mass spectra were compared to library spectra from the National Institute of Standards and Technology (NIST) Standard Reference Database, 2.0 f. A quality match cutoff of 75% reported spectra with 75% or greater similarity with the database spectra. The library match process was repeated and when the library identified multiple compounds as matches for a single compound, manual comparison of the sample spectra determined the final identification. Sample spectra that could not be rectified with the library spectra were considered “unknown” even if the library considered it to have a quality match above 75%. All identified compounds are listed in the NIST Chemistry WebBook terminology [
The quantities of volatiles desorbed from the fiber were calculated from the peak area of the total ion current measured by the mass spectrometer minus the peak area from un-inoculated controls. As previously shown [
Analytical results from liquid and gas-phase measurements are described for both small-scale shake flask studies as well as larger-scale gas-purged bioreactors. Small-scale shake flask cultures permitted a larger matrix of variables to be tested, and analysis using NMR gave VHC results specific to chemical classification. The largerscale gas-purged batch bioreactor experiments were used to expand the knowledge of VHC identities and distributions. These analyses differed from previously reported results based on the experimental and analytical methods used. The results presented here reflect quantitative data not possible with the previous HS-SPME analyses and demonstrate a new technique for reporting HS-SPME GC-MS results using an external Carbotrap collection column as designed by Booth et al. [
The culturing environment influenced VHC production by the fungus significantly. Over 130 culture supernatant samples were measured with NMR and the resulting peak areas were analyzed to determine which variables had a statistically significant effect. NMR spectra from a 15-day old culture grown at 20˚C and pH 5 showed distinct VHC peaks between 2 and 1 ppm, while culture-free cellulose medium samples had none (
The statistical analysis of the 132 NMR samples showed no statistically significant impact on mean VHC concentration from carbon source alone (glucose and cellulose). Individual models for each medium elucidated further interactions of environmental variables. For CM samples, statistically significant inputs included oxygen condition, pH, temperature and age of the culture. The most important factor was pH. The effect of pH on VHC
production was at least an order of magnitude higher than the other effects (Supplementary Tables S1-S3) (pvalue = 8.26 × 10−26). However, the oxygen condition in which the cultures were grown had a significant interaction with pH (p-value = 1.34 × 10−3) and therefore, the model predicts different VHC concentration maximums at different pH-oxygen level combinations. The statistical analysis predicted maximum VHC production at a pH of 6.1 for aerobic oxygen conditions and at a pH of 5.1 for both sealed batch and anaerobic oxygen conditions. The maximum VHC near the top of the pH range (3.5 - 6.5) confirmed preliminary observations that at the lower pH levels around 3 or 3.5 growth and metabolic function of A. sarcoides was inhibited. Overall, lower oxygen concentrations indicated higher VHC concentrations (p-values ≤ 1.44 × 10−4).
Analysis of the GM samples showed there was a statistically significant impact on mean VHC production based on nutrient levels in the media (p-value: 0.036). As the medium concentration (e.g. 1× and 2×) increased, the VHC concentrations decreased. Other variables did not show a statistically significant impact on VHC production after accounting for all other environmental effects for the GM grown cultures (p-value ≥ 0.16). This could be due to the smaller number of samples processed for GM versus CM. For either medium, explicit regression equations can be used to predict VHC levels under specific conditions (Tables S1-S3).
Building on these results, larger scale batch bioreactor experiments were conducted to compare volatile and liquid VHC production on three different substrates at 23˚C, a starting pH of 5.8, and continuous flow of gas through headspace. In addition, the culture age impact seen for the CM cultures was explored by sampling at two distinct time periods (Section 3.2).
Medium substrate and headspace oxygen concentration were important factors for the growth rate of A. sarcoides. Growth rates and final biomass concentrations were calculated for CMC and GM in aerobic conditions and for CM, CMC, and GM in sealed batch conditions (data not shown). Lower oxygen values in sealed batch cultures resulted in slower specific growth rates on CMC (0.038 ± 0.006 hr−1 versus 0.026 ± 0.001 hr−1), but not on GM (both ~0.02 hr−1) which had a similar rate to CM (0.017 ± 0.01 hr−1). However, despite comparable growth rates, a higher final biomass concentration on GM (0.871 ± 0.033 g·L−1) was reached with sealed batch conditions as compared to 0.503 ± 0.009 g·L−1 under aerobic conditions. Growth on CM achieved maximum biomass (0.797 ± 0.062 g·L−1) under aerobic conditions.
Substrate had a marked impact on VHC production type and frequency in the gas-purged bioreactors. Growth on cellulose (CM) and potato dextrose broth (PDB) showed more than twice the number of compounds as growth on glucose (GM). This is not surprising due to the differences in initial substrate complexity. CM showed increased product diversity from GM, with 50 compounds distinct from GM, and 16 in common with PDB, indicating the diverse metabolic capability of the organism to produce different compounds on complex substrates (
After compound identification by GC-MS (64 of the 114 compounds were identifiable based on a conservative quality match cutoff of 75%), the compounds were sorted into classes based on structure: alkane, alkene, aldehyde, ketone, aromatic, alcohol, acid, and ester. Compounds with non-aromatic ring structures were classified based on bonding in the ring (e.g. a double carbon bond was put in the alkene class). Comparing the results from days 5-17 (See Section 3.2.2 for discussion on multiple cellulose time points),
largest diversity were the aromatics and esters (
The quantity of VHC being produced is crucial to evaluating cellulose as an appropriate substrate for fuel production. Looking at the overall modest yields of fuel-range organic VHC in
experiments imply that the total VHC potential of A. sarcoides was underestimated by the NMR technique that did not quantify oxygenated VHC.
The total fuel-range VHC production yields shown in
Two time points from the 4 L gas-sparged bioreactor cellulose culture were taken to assess the importance of culture age on VHC production. Significant changes were seen in the VHC composition between the two time points. The number of compounds detected was nearly constant before (60 total) and after day 17 (61 total) (
Of the identifiable compounds, there was a large change in speciation of the VHC produced for the different time periods.
A possible metabolism change in stationary phase is the oxidation of non-oxygenated VHC to gain electrons. For example, 2-heptanone, 6-methylwas only present during the second time period and could have been oxidized from the alkene, 1-heptene, 6-methyl- (
Many of the compounds identified in the gas-purged bioreactors have fuel potential, such as hexane, 3-methyland cyclopropane, propyl- (
a. Octane ratings are the average of Motor Octane Number and Research Octane Number [
within the boiling point requirement (25˚C - 230˚C) and carbon chain length range, C5-C12, for gasoline fuel, while the majority fit the boiling point requirement (126˚C - 287˚C) and carbon chain length, C8-C16, for aviation fuel (
Aviation fuel is composed of 70% - 85% paraffins including isoparaffins, cycloparaffins and naphthenes [
In addition, other non-oxygenated compounds were observed for which boiling points, octane ratings and/or enthalpies of combustion were not readily available in the literature: 1,3-hexadiene, 3-ethyl-2-methyl-; cyclohexane, propylidene-; and 1,5-cyclooctadiene, 1,5-dimethyl- (an olefin, a cycloparaffin, and a cyclo-olefin, respectively). However, these compounds have the correct carbon chain length and branching properties for gasoline fuel. Compounds produced which were more suited for kerosene or diesel fuel included an isoparaffin: dodecane, 2,6,10-trimethylat a boiling point of 253˚C and a paraffin: nonadecane at 330˚C.
The compounds with fuel potential in
The diverse metabolic capability of A. sarcoides to utilize multiple carbon sources to produce gasoline, diesel and aviation range organics was demonstrated. The use of multiple analytical methods to assess VHC production by A. sarcoides cultures in both the liquid and gas phases yielded information on the type and quantities of VHC produced. Statistically significant differences in VHC production was observed with carbon source, oxygen concentration, pH and temperature of the culture indicating the potential to further optimize production by varying process and growth parameters. Gasoline range (C5-C12), aviation range (C8-C16), and diesel range (C9-C36) organics were detected in all cultured media. Among multiple substrates, the highest levels of recovered VHCs were measured from growth on cellulose at a pH of 5.8˚C and 23˚C, and the highest level of biomass was predicted at pH 5.6˚C and 20˚C. Cellulose stood out as the preferable substrate for VHC production and fuel-related compounds based on the relatively inexpensive substrate and the quantity and variety of fuel compounds produced. There was a pronounced shift to more oxygenated compounds, longer carbon chain length, and fewer fuel-related VHCs as the cellulose culture progressed in stationary phase indicative of a change in carbon metabolism. Therefore, future refinement of production mechanisms by A. sarcoides and other similar fungi is warranted to increase VHC yields for renewable production of liquid fuel compounds.
The authors would like to thank the National Science Foundation, Emerging Frontiers in Research and Innovation #093761 for funding this research. The NMR data was collected by the help of by COBRE award P20 RR024237 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH), but the conclusions from these data are solely the responsibility of the authors and do not necessarily represent the official view of NCRR or NIH. Our appreciation goes to Eric Booth for training on thermal desorption, to Dr. W. Berk Knighton for advice on analytical chemistry techniques, to Isabel M. Holtmeyer for proofreading and positive feedback, and to Heidi Schoen for
1. In Tables S1-S3, the superscript F indicates that the variable is included in the regression model as a categorical factor. In
VHC = 15.36 + 35.81ICarbon = Glucose + 221.9IO2 = Anaeoboc + 396.4IO2 = Sealed Batch − 126.8Conc − 23.11Temp + 165.6pH + 2.88Age.
The notation I denotes an indicator function, where I = 1 if the condition specified in the subscript is met; and I = 0 otherwise. For example, when Ascocoryne sarcoides is grown on cellulose; under anaerobic conditions and a medium concentration of 2×; at a temperature of 30 degrees C, pH = 7; and aged for 5 days; then the regression equation predicts that the VHC production of A. sarcoides is:
VHC = 15.36 + 221.9(1) − 126.8(2) − 23.11(30) + 165.6(7) + 2.88(5) = 463.96 (total area as measured by NMR).