Anastasios Melis, Department of Plant and Microbial Biology, University of California, Berkeley, CA 94720-3102, USA. E-mail: email@example.com
Aims: Isoprene (2-methyl-1,3-butadiene; C5H8) is naturally produced by photosynthesis and emitted in the atmosphere by the leaves of many herbaceous, deciduous and woody plants. Fermentative yeast and fungi (Ascomycota) are not genetically endowed with the isoprene production process. The work investigated whether Ascomycota can be genetically modified and endowed with the property of constitutive isoprene production.
Methods and Results: Two different strategies for expression of the IspS gene in Saccharomyces cerevisiae were employed: (i) optimization of codon usage of the IspS gene for specific expression in S. cerevisiae and (ii) multiple independent integrations of the IspS gene in the rDNA loci of the yeast genome. Copy number analysis showed that IspS transgenes were on the average incorporated within about 25% of the endogenous rDNA. Codon use optimization of the Pueraria montana (kudzu vine) IspS gene (SckIspS) for S. cerevisiae showed fivefold greater expression of the IspS protein compared with that of nonoptimized IspS (kIspS). With the strategies mentioned earlier, heterologous expression of the kudzu isoprene synthase gene (kIspS) in S. cerevisiae was tested for stability and as a potential platform of fermentative isoprene production. The multi-copy IspS transgenes were stably integrated and expressed for over 100 generations of yeast cell growth and constitutively produced volatile isoprene hydrocarbons. Secondary chemical modification of isoprene to a number of hydroxylated isoprene derivatives in the sealed reactor was also observed.
Conclusion: Transformation of S. cerevisiae with the Pueraria montana var. lobata (kudzu vine) isoprene synthase gene (IspS) conferred to the yeast cells constitutive isoprene hydrocarbons production in the absence of adverse or toxic effects.
Significance and Impact of the Study: First-time demonstration of constitutive isoprene hydrocarbons production in a fermentative eukaryote operated through the mevalonic acid pathway. The work provides concept validation for the utilization of S. cerevisiae, as a platform for the production of volatile hydrocarbon biofuels and chemicals.
Efforts are currently being made to develop technologies for the renewable generation of fuels and chemicals to meet increasing demands for energy and synthetic chemistry feedstock, without contributing to global warming, environmental degradation or disruption of food supply (Stephanopoulos 2007; Rittmann 2008). Such technologies need to be sustainable, secure, renewable and environmentally friendly. Possible renewable energy carriers include biologically generated hydrocarbon molecules and their derivatives. Hydrocarbons are desirable products because they have a high energy density, both on a mass as well as a volume basis (Fischer et al. 2008; Peralta-Yahya and Keasling 2010). Isoprene (2-methyl-1,3-butadiene; C5H8) is a volatile hydrophobic terpenoid hydrocarbon molecule that meets these requirements, plus it affords a number of practical advantages in terms of generation and separation from the biomass (Lindberg et al. 2010; Bentley and Melis 2012). It also serves as feedstock in the synthetic chemistry industry, for example, production of rubber. Isoprene is currently derived from the petroleum refining process, as a by-product of ethylene production (Sharkey and Singsaas 1995).
Isoprene is naturally synthesized in the chloroplast of many herbaceous, deciduous and woody plants species and released in the atmosphere from the leaves in a process induced by heat-stress (Sharkey and Yeh 2001; Sharkey et al. 2008). The biological significance of isoprene release by plants is unclear; however, it has been suggested that plants utilize isoprene emission as a thermotolerance mechanism (Singsaas et al. 1997; Sharkey et al. 2001; Behnke et al. 2007; Sasaki et al. 2007). Plant isoprene production is not suitable for commercial exploitation, owing to the difficulty of harvesting this volatile product from the extended canopy of leafy plants.
It was recently demonstrated that isoprene could be synthesized in the photosynthetic apparatus of transgenic cyanobacteria (Lindberg et al. 2010) and collected in suitably designed photobioreactors (Bentley and Melis 2012). Isoprene emission was also reported from bacteria, such as Escherichia coli (Miller et al. 2001; Zhao et al. 2011) and Bacillus subtilis (Julsing et al. 2007). However, it is not known whether heterologous transformation of eukaryotic yeast and fungi could also result in stable and constitutive isoprene hydrocarbons production. This is an important question, as many Ascomycota are commercially valuable fermentative microorganisms. However, they are substantially different from photosynthetic plastid and bacterial systems because (i) they are fermentative eukaryotes lacking a plastid structure and the attendant photosynthetic apparatus and (ii) they lack the deoxyxylulose phosphate/methylerythritol phosphate (DXP-MEP) isoprenoid pathway that exclusively operates in isoprene-producing photosynthetic systems.
Saccharomyces cerevisiae (budding yeast) is a model organism that employs the mevalonic acid (MVA) pathway to generate isoprenoids essential for cell growth, such as geranyl, farnesyl and geranylgeranyl derivatives (Jackson et al. 2003; Oswald et al. 2007; Shiba et al. 2007; Takahashi et al. 2007). In the MVA pathway, three molecules of acetyl-coenzyme A (acetyl-CoA) condense successively to form the 6-carbon intermediate 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA). The HMG-CoA is then reduced to MVA, which is subsequently phosphorylated twice and decarboxylated to form isopentenyl diphosphate (IPP). IPP can be isomerized to DMAPP (Kuzuyama 2002) through the action of the IPP isomerase (Fig. 1). Thus, the MVA pathway of yeast cells is substantially different from the DXP-MEP pathway of plant plastids and bacteria. Moreover, the yeast genome does not encode for an isoprene synthase (IspS), and therefore, it is not known whether yeast cells are capable of converting the metabolically critical DMAPP into isoprene upon introduction of the IspS transgene in the cells. Unknown also is the ability of yeast cells to tolerate overexpression of the IspS transgenic protein or whether they might be subject to toxicity effects (Withers et al. 2007) from the IspS product.
In this work, S. cerevisiae were genetically engineered for isoprene production upon heterologous expression of the codon-optimized kudzu vine IspS gene. Stability of transgenes and their expression, constitutive production of isoprene and absence of adverse or toxic effects are reported. Thus, it is possible to drive the MVA pathway in the cytoplasm of a eukaryote and to constitutively produce volatile isoprene hydrocarbons from the fermentation metabolism of the host organism. The work provides a new dimension for the potential utilization of yeast (Sudbery 1996; Huang et al. 2008), as a platform for the production of volatile hydrocarbon biofuels and other useful chemicals.
Materials and Methods
Strains and culture conditions
Escherichia coli DH5α [F′/endA1 hsdR17 (rk− mk+) supE44 thi-1 recA1 gyrA (Nalr) relA1Δ(lacZYA-argF)u169: (m80ΔlacZM15)] (Invitrogen, Carlsbad, CA, USA) was used to amplify plasmid DNA using standard procedures (Ausubel et al. 2008). Saccharomyces cerevisiae 20B-12, a peptidase mutant (α trp1 pep4-3; ATCC 20626, Manassas, VA) was used as the recipient strain of plasmid pSckIspStg containing S. cerevisiae codon-optimized kudzu vine (Pueraria montana var. lobata) IspS gene and plasmid pkIspStg containing the original sequence of the kudzu vine IspS gene, respectively. Escherichia coli was cultured on LB medium (Luria-Bertani; 1% tryptone, 0·5% yeast extract, 1% sodium chloride) or LB containing appropriate antibiotics for competent cell preparation and plasmid isolation (Maniatis et al. 1989). Saccharomyces cerevisiae strain 20B-12 (recipient strain) was grown on YPD medium (1% yeast extract, 2% peptone, 2% glucose). We used YNBD medium (0·67% yeast nitrogen base without amino acids, 2% glucose) as a selective medium for the isolation of yeast transformants (Lopes et al. 1989). For measurements of isoprene production, yeast transformants were cultured in YPG medium (2% galactose instead of 2% glucose in YPD). Galactose was chosen as the carbohydrate feedstock in these measurements because the culture supports lower amounts of ethanol relative to glucose, without negatively affecting the growth rate of the yeast cells (unpublished observations).
Construction of plasmids containing the S. cerevisiae codon-optimized IspS or the nonoptimized kudzu vine IspS
Plasmid pNGVF served as the plasmid backbone. The pNGVF was constructed using pAPNGFPVFNB and pGFPV (Hong and Linz 2008). The pAPNGFPVFNB was digested with NotI and AscI, and the 3·7-kb fragment from the pAPNGFPVFNB replaced 2·8-kb fragment in pGFPV digested with the same enzymes, resulting in pNGVF. A 0·65-kb GPD promoter was generated by PCR with Pfu DNA polymerase (Stratagene, La Jolla, CA, USA), appropriate primers and p424 GPD (ATCC 87357, Manassas, VA, USA) as a template using standard procedures (Ausubel et al. 2008) (see Table S1 for primer sequences). PCR was performed in a PTC-200 Peltier Thermal Cycler (Bio-Rad, Hercules, CA, USA). The reaction conditions were as follows: 94°C for 5 min followed by 35 cycles of 94°C for 1 min, annealing for 1 min (see Table S1 for annealing temperatures) and extension at 72°C (2 min/1 kb). The reaction was completed with a final extension at 72°C for 10 min. The PCR fragment was digested with NotI and SgfI and cloned into pNGVF cut with the same enzymes, resulting in pGVF. A 0·8-kb TRP1d gene fragment was generated by PCR with Pfu DNA polymerase, appropriate primers and p424 GPD plasmid as a template. Similar PCR conditions were used as for pGVF (Table S1). The PCR fragment was digested with PacI and NotI and cloned into pGVF cut with the same enzymes, resulting in pTGVF. To subclone a CYC1 terminator fragment, the 0·26-kb fragment was generated by PCR with Pfu DNA polymerase, appropriate primers and p424 GPD as a template. Similar PCR conditions were used as for pGVF (Table S1). The PCR fragment was cloned into the SmaI site of pUC18, resulting in pUCCYC1t. To subclone rDNA fragments, the 0·67-kb and 0·59-kb fragments were generated by PCR with Pfu DNA polymerase, appropriate primers and genomic DNA from S. cerevisiae recipient strain as a template using standard procedures (Ausubel et al. 2008) (for primers, see Table S1). Genomic DNA isolation from S. cerevisiae was performed according to the method of Philippsen and co-workers with minor modifications (Philippsen et al. 1991). Similar PCR conditions were employed as for the construction of pGVF (Table S1). The 0·67-kb ribosomal DNA (rDNA) fragment was subcloned into pUCCYC1t cut with AsiSI and AscI, resulting in pUCCD. DNA fragments containing the CYC1 terminator and the rDNA were then subcloned from pUCCD into pTGVF cut with FseI and AscI, resulting in pTGVCD. The 0·59-kb rDNA fragment was subcloned into pTGVCD cut with SbfI and PacI, resulting in pDTGVCD.
A 1·7-kb S. cerevisiae codon-optimized kudzu vine IspS gene (cDNA; SckIspS) was designed on the basis of the codon usage table for S. cerevisiae (Nakamura et al. 2000) using the GeneDesigner software, synthesized by DNA 2·0, Inc. (Menlo Park, CA, USA), and subcloned to generate plasmid pSckIspS. DNA fragments containing the CYC1 terminator and the 0·67-kb rDNA were then subcloned from pUCCD into pSckIspS cut with FseI and AscI, resulting in pSckIspSCD. Finally, DNA fragments containing the SckIspS and the CYC1 terminator were subcloned from pSckIspSCD into pDTGVCD cut with AsiSI(=SgfI), resulting in plasmid pSckIspStg (Fig. 2). To construct the plasmid containing the 1·7-kb original sequence of the kudzu vine IspS gene (cDNA; kIspS), the fragment was generated by PCR with Pfu DNA polymerase, appropriate primers and pBA2kIA2 (Lindberg et al. 2010) as a template. Similar PCR conditions were used as for pGVF (Table S1). The 1·7-kb kudzu vine IspS fragment was subcloned into the SmaI site of pUC18, resulting in pUCkIspS. DNA fragments containing the CYC1 terminator and the 0·67-kb rDNA were then subcloned from pUCCD into pUCkIspS cut with FseI and AscI, resulting in pUCkIspSCD. Finally, DNA fragments containing the kIspS and the CYC1 terminator were subcloned from pUCkIspSCD into pDTGVCD cut with AsiSI(=SgfI), resulting in plasmid pkIspStg.
Saccharomyces cerevisiae transformations with plasmid DNA
For yeast transformation, plasmids were digested with restriction enzymes SbfI and AscI and the DNA fragments containing the selectable marker (TRP1d), IspS gene and rDNA regions were agarose gel-purified. The DNA fragments were targeted for integration into the NTS2 (nontranscribed spacer 2) and 5′ ETS (external transcribed spacer) region in the rDNA. The rDNA of S. cerevisiae consists of 100–200 tandemly repeated copies of a 9·1-kb unit on the right arm of chromosome XII (Venema and Tollervey 1999). Each repeated unit contains the genes for 5S, 18S, 5·8S and 25S rRNAs, as well as three types of spacer regions: internal transcribed spacers (ITS1, ITS2), external transcribed spacers (5′ ETS, 3′ ETS) and nontranscribed spacers (NTS1, NTS2) (Fig. S1). Each ribosomal RNA is present in a single copy in an 80S yeast ribosome. 18S rRNA is a component of the 40S ribosomal subunit, and the 5S, 5·8S and 25S rRNAs are components of the 60S subunit (Spahn et al. 2001). Yeast transformations were performed by the lithium acetate method with minor modifications (Gietz et al. 1995).
Yeast colony PCR
Yeast cells were transferred in 0·2% sodium dodecyl sulphate (SDS) solution and were ruptured by heating at 90°C for 4 min (Akada et al. 2000). After centrifugation, 1 μl of supernatant was used for PCR reaction. PCR was performed with primers (see Table S2 for primer sequences) specific to each coding region. The reaction conditions were as follows: 95°C for 2 min followed by 35 cycles of 95°C for 1 min, annealing at 56°C for 1 min and extension at 72°C for 3 min. The reaction was completed with a final extension at 72°C for 10 min.
Southern hybridization and genomic PCR analyses
Genomic DNA was isolated by the method of Philippsen and co-workers with minor modifications (Philippsen et al. 1991). Southern hybridization analyses were conducted using standard procedures (Maniatis et al. 1989). Approximately 1 μg of genomic DNA cut with BglII or PvuII were separated by agarose gel electrophoresis and were transferred onto a nylon membrane (Amersham Hybond- N+, GE Healthcare, Little Chalfont, UK) by capillary action. The DNA probe was generated with the Amersham Gene Images AlkPhos Direct Labelling and Detection System (GE Healthcare) and the 0·57-kb rDNA PCR fragment (see Table S2 for primer sequences) specific to NTS2 (nontranscribed spacer 2) in a 9·1-kb rDNA unit using a procedure provided by the manufacturer. Detection of the DNA fragments was performed using the Amersham Detection System (GE Healthcare). PCR analyses were performed with genomic DNA and primers specific to a set of the NTS2 and TRP1d or another set of the CYC1 terminator and 18S rDNA (for primer sequences, see Table S2) to confirm the integration site of the DNA fragment.
Copy number analysis
Copy number of the integrated DNA fragment containing the selectable marker (TRP1d), IspS gene and rDNA regions was calculated by comparison of the intensity of the new band with that of the endogenous ribosomal band from each transformant on X-ray film generated by Southern hybridization analyses. The intensity of the bands was measured by Lab Works Image Acquisition and Analysis Software (UVP Bioimaging System, UVP Inc., Upland, CA, USA).
Western blot analysis of IspS proteins
Yeast cells were cultured in 50 ml of YNBD (+Trp) medium at 30°C for 24 h with shaking at 200 rev min−1 by 5% inoculation of overnight seed culture in the same medium. Protein from yeast cells was extracted using a standard glass beads procedure (Ausubel et al. 2008). Protein concentration in extracts was determined using a DC Protein Assay (Bio-Rad Laboratories). Approximately 25 μg of protein extracts were separated by 8% SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Immunodetection was carried out with a specific polyclonal antibody against the IspS protein (Lindberg et al. 2010) as a primary antibody, a goat anti-rabbit antibody conjugated to horseradish peroxidase (Bio-Rad Laboratories) as a secondary antibody, and a Supersignal ECL detection kit (Pierce, Thermo Fisher Scientific Inc., Rockford, IL, USA) using a procedure provided by the manufacturer. A PageRuler Prestained Protein Ladder (Fermentas Inc., Burlington, Ontario, Canada) was used as a molecular mass marker. The intensity of the bands was measured by Lab Works Image Acquisition and Anaylsis Software (UVP Bioimaging System, UVP Inc., Upland, CA, USA).
Yeast cells were harvested at OD600 = 1·0. Total RNA was isolated using Trizol Reagent (Invitrogen) according to the manufacturer’s instructions. Total RNA was treated with DNaseI (Invitrogen). RT-PCR was performed with 0·5 μg of total RNA, primers specific to the coding region of the SckIspS and kIspS, respectively (see Table S2 for primer sequences) using the Qiagen OneStep RT-PCR kit (Qiagen, Valencia, CA, USA). The reaction conditions were as follows: 50°C for 30 min and 95°C for 15 min followed by 25 cycles of 94°C for 1 min, annealing at 56°C for 1 min and extension at 72°C for 3 min. The reaction was completed with a final extension at 72°C for 10 min.
Measurement of isoprene production
For measurements of isoprene production, yeast cells were cultured in 10 or 50 ml of YPG medium at 30°C with shaking at 200 rev min−1. After 48 or 72 h incubation under aerobic conditions, the flasks were sealed with rubber stoppers. One ml of gas from the headspace of sealed cultures was sampled and analysed by gas chromatography using a Shimadzu 8A GC (Shimadzu, Columbia, MD, USA) equipped with a Porapak N 80/100 column and a flame-ionization detector (FID). This column is designed to optimally separate short-chain hydrocarbons, namely C1 to C5 molecules. Amounts of isoprene produced by yeast transformants were estimated by comparison with an isoprene standard (Acros Organics, Fair Lawn, NJ, USA).
For GC-MS analysis, 200 μl of gas from the headspace of sealed cultures was injected into the GC inlet port of Agilent Technologies 6890 GC/5973 MSD equipped with a DB-XLB column (0·25 mm i.d. × 0·25 μm × 30 m, J &W Scientific). The DB-XLB column is suited for the separation of a wide range of molecules including pesticides, herbicides, PCBs and PAHs. Programming of the GC oven entailed initial setting at 40°C (hold for 4 min), 5°C min−1 ramp to 80°C, 12 min of total run time and 260°C of inlet temperature. Size selection conditions were chosen to detect 48–100 Da products and to eliminate smaller or larger molecules from the analysis. This GC-MS size selection range effectively eliminated ethanol signals from the GC-MS output.
Mitotic stability determination
Stability of the multi-copied integrative DNA fragment was determined by the method of Nieto and co-workers (Nieto et al. 1999). Transformants were cultured in 5 ml of YPD medium at 30°C with shaking at 200 rev min−1 starting at OD600 =0·001 by inoculation of an overnight seed culture and were diluted into the same medium to start at the same OD600 value each day during the yeast culture until 100 generations. The cells were plated onto YPD agar medium after growth for 100 generations and replica-plated onto YNBD and YNBD+Trp agar media. The copy number of the fragment was analysed as described earlier. The IspS activity was tested by gas chromatography (GC) after transformants were cultured in YPG medium.
Statistical analyses were performed by student’s t-test using Excel statistical software (Microsoft Corp., Redmond, WA, USA). Statistical significance between samples was defined by a P value not >0·05 (P ≤ 0·05).
Transformation of Saccharomyces cerevisiae recipient strain with the IspS plasmid
For efficient expression of the isoprene synthase gene (IspS) in S. cerevisiae, a codon-optimized IspS cDNA (SckIspS) was designed on the basis of the codon usage table for S. cerevisiae (Nakamura et al. 2000). The SckIspS gene was synthesized without the putative transit peptide, consisting of amino acid 45–608 of the kudzu IspS sequence. The SckIspS gene was cloned in plasmid pSckIspStg (Fig. 2), in which TRP1d and rDNA fragments were used, respectively, for transformant selection and multi-copy integration into the rDNA region via double homologous recombination. To test the effect of the codon-optimization on the efficiency of the IspS gene expression in S. cerevisiae, a control plasmid was also constructed (pkIspStg, not shown) in which the codon-optimized SckIspS gene was replaced with the native kudzu vine kIspS cDNA. Transformation of the S. cerevisiae recipient strain with plasmids pSckIspStg or pkIspStg generated 20 and 29 colonies, respectively. Yeast colony PCR analysis was performed to screen these transformants for the isolation of IspS multi-copy integrants. Four transformant lines (ScT2, ScT11, ScT16 and ScT17) from the pSckIspStg and three transformant lines (KT8, KT16 and KT24) from the pkIspStg showed a high copy number of the integrated plasmid (not shown). These transformants were chosen for further analyses.
Analyses of integration sites of the IspS plasmid in the yeast chromosome
To identify the integration sites of the IspS transgenes in the yeast chromosome, Southern hybridization and genomic PCR analyses were performed with the recipient strain and the seven isolated transformants. Southern blot analyses were designed with primers and a probe such that hybridization bands from the transformants would be larger in size compared with that from the endogenous region of the recipient strain, the difference being due to integration of the plasmid SckIspS or kIspS into the rDNA region (Fig. 3a). This strategy is schematically depicted in the genomic DNA map of the recipient strain (Fig. 3a, upper) and the IspS transformant strains (Fig. 3a, lower). Integration of the IspS (SckIspS or kIspS) expression construct into the yeast rDNA locus is shown in Fig. 3a to occur between the NTS2 and 5′ETS DNA regions. Restriction enzyme sites for BglII (=B) and PvuII (=P) are indicated. The expected band sizes are also indicated above the yeast chromosome after integration of the SckIspS expression fragment and below the chromosome map after integration of the kIspS expression fragment (Fig. 3a, lower). Primer positions for genomic PCR analysis are shown as P1, P2, P3 and P4.
Southern blot analysis of the S. cerevisiae recipient strain (Fig. 3b, lanes 1 and 6), using the NTS2 (nontranscribed spacer 2) as a probe, showed a single hybridization band at 4·58 kb (lane 1) and 2·06 kb (lane 6). Transformants carrying the SckIspS construct generated a new hybridization band at 7·06 kb (Fig. 3b, lanes 2–5) and 4·1 kb (Fig. 3b, lanes 7–10). These results are evidence that plasmid SckIspS integrated within the rDNA repeat units in each of the four SckIspS transformants.
Similarly, Southern blot analysis of the S. cerevisiae recipient strain (Fig. 3c, lanes 1 and 5) showed a single hybridization band at 4·58 kb (lane 1) and 2·06 kb (lane 5). Transformants carrying the kIspS construct generated a new hybridization band at 8·04 kb (Fig. 3c, lanes 2–4) and 5·52 kb (Fig. 3c, lanes 6–8). These results also showed that plasmid kIspS integrated within the rDNA repeat units in each of the three kIspS transformants.
Genomic PCR analysis was independently employed to confirm integration sites of the IspS construct in the rDNA locus in the transformants (Fig. 3d). Two sets of primers (P1-P2 or P3-P4, Fig. 3a) specific for the IspS construct and rDNA region were employed in this PCR analysis. Recipient strains (Fig. 3d, lanes 1 and 2) showed no PCR product with either of the two sets of primers. All four SckIspS and three kIspS transformants showed PCR products with both the P1-P2 and P3-P4 sets of primers. These results further strengthened the notion that plasmids SckIspS and kIspS integrated within the rDNA repeat units in each of the two different types of IspS transformants.
Copy number of the integrated IspS DNA construct was calculated from the Southern blot analysis results as relative band intensity, by way of comparison of the intensity of the new (transgene) band with that of the endogenous rDNA band on the blot (Fig. 3b and c). Results showed that the intensity of the new (transgene) band was on the average about 25% of that from the endogenous rDNA band in the various lanes of the blot. Given that as many as 100–200 copies of the rDNA unit are present on the yeast chromosome per cell, it was estimated that 25–50 copies of the IspS expression construct were integrated into the rDNA repeats in each of the seven transformants.
Expression of the IspS transgenes
Transcription of the IspS transgenes was tested upon RT-PCR analysis. This was conducted on the seven transformant lines to test whether the SckIspS or kIspS genes are transcribed under physiological growth conditions. The RT-PCR reactions were performed with primers specific to the coding regions of the SckIspS and kIspS constructs, respectively. The result in Fig. 4a showed that the SckIspS gene was transcribed, as the respective mRNA accumulated in all four of the SckIspS transformants (Fig. 4a, lanes 5, 7, 9 and 11). Similarly, the result in Fig. 4b showed that the kIspS gene was transcribed, as the respective mRNA accumulated in all three of the kIspS transformants (Fig. 4b, lanes 5, 7 and 9). These results confirm the presence of IspS mRNA in all seven S. cerevisiae transformants harbouring the IspS transgenes. In contrast, no transcript was detected in the negative controls, where the reverse transcriptase was not added (Fig. 4a, lanes 4, 6, 8 and 10 and Fig. 4b, lanes 4, 6 and 8).
Translation of the IspS transgenes from the corresponding mRNAs was tested in Western blot analyses with specific polyclonal antibodies raised against the kIspS protein (Lindberg et al. 2010). This analysis confirmed the presence of the IspS protein as a 65-kD protein in the transformants (Fig. 5a, lanes 2–8) but not in the recipient strain (Fig. 5a, lane 1). A Coomassie-stained SDS-PAGE of the same samples (Fig. 5b) served as an internal control for this measurement, showing no discernible differences in protein pattern or loading. Interestingly, the intensity of the antibody-kIspS cross-reaction was on the average fivefold greater in the protein extracts from the four transformants carrying the S. cerevisiae codon-optimized SckIspS construct (Fig. 5a, lanes 2–5) compared with three transformants carrying the original unmodified kudzu vine kIspS (Fig. 5a, lanes 6–8), affording evidence for the enhancement in IspS gene expression upon codon usage optimization. These results clearly show that the IspS protein is synthesized and accumulates in all seven IspS transformants and that transformants carrying codon-optimized SckIspS synthesize and accumulate the IspS protein more efficiently than transformants carrying the nonoptimized kIspS.
Isoprene production by IspS transformants
Vapour samples from the sealed headspace of recipient and transformant strains carrying the SckIspS or kIspS transgenes were examined by GC-MS gas chromatography (Agilent Technologies 6890 GC/5973 MSD equipped with a DB-XLB column). In the GC-MS approach, a 48–100 Da size selection was employed, which effectively eliminates ethanol signals from the GC-MS output. The recipient strain headspace showed two peaks with retention times of about 1·55 and 6·1 min (Fig. 6a, recipient strain). MS analysis of peak  showed multiple lines of low abundance reflecting unspecific molecules (noise level signal, see Fig. S2a). MS analysis of peak  showed a dominant line ‘57’ corresponding to 1,6- hexanediol. This compound is probably generated in the presence of the YPG media as a by-product of the yeast cell metabolism.
GC-MS analysis of vapour from the headspace of IspS transformants is shown in Fig. 6b. Peak  with retention time of 1·55 min displayed dominant 53, 67 and 68 lines attributed to isoprene (Fig. S2b and c). Peaks  through  with retention times of 2·2–3·8 min were dominated by lines (55, 56, 57 and 70) indicative of 5-carbon branched alcohols (2-methyl-3-buten-1-ol and 3-methyl-1-butanol), probably derived from the nonspecific hydroxylation of isoprene. Peak  with retention time of 5·71–5·74 min displayed a single GC-MS line (96) attributed to 5-hydroxypyrimidine, whereas peak  with retention times of 6·15–6·26 min displayed a single GC-MS line (57) attributed 1,6-hexanediol (also present in the GC-MS profile of the recipient strain). Table 1 provides a detailed summary of the GC-MS analysis results, correlating GC retention time, corresponding MS lines, and the chemical species that corresponds to each of the GC peaks shown in Fig. 6.
Table 1. Summary of GC-MS analysis results from Fig. 6
GC retention, time, min
Peak numbers were assigned on the basis of retention time in the GC, and chemical species identification was enabled upon comparative analysis of the resultant lines with known standards. Peaks  and  are from Fig. 6a. Peaks  through  are from Fig. 6b.
*Line 72 was putatively identified as one of several different molecules, including (i) 3-phenyl-2-propenoic acid, 2-methyl-2-propenyl ester, (ii) N,N′-diisopropylethylenediamine, (iii) 2-fluoro-1,3-butadiene, or (iv) 3-methoxy-1-propanamine. MS lines in parentheses indicate low relative abundance.
Multiple lines, noise level
In addition to the GC-MS analysis, we conducted a GC-FID measurement of headspace gases from control (recipient strain) and the IspS ScT16 transformant strain of S. cerevisiae.Fig. S2d shows that the headspace of the recipient strain cultures contained CO2 and ethanol, the latter appearing as a dominant GC peak with a 5·4 min retention time (Fig. S2d, left panel). The headspace of SckIspS and kIspS transformants contained CO2, ethanol and isoprene (Fig. S2d, middle panel). All seven IspS transformant lines examined, but only the transformants produced isoprene, which appeared as a separate and distinct GC peak with a 6·8 min retention time (Fig. S2d, middle and right panels).
These results clearly show that the expressed IspS transgenic protein is active in the yeast transformants. It is also evident that chemical modification of the isoprene molecule takes place in the yeast growth medium, with various hydroxylated isoprene derivatives being the dominant product. The ratio of secondary product to isoprene differed among different yeast cultures and preparations, suggesting a variable and time-dependent process for the secondary modification of isoprene (see also below). Quantitative analysis of the SckIspS-specific products showed a yield of over 500 μg of isoprene and isoprene derivatives accumulated in the headspace per l culture in the course of 72 h of anaerobic incubation. This was a rather conservative estimate of product accumulation in the headspace. Moreover, it did not take into account hydroxylated isoprene derivatives that remained in the liquid medium of the culture.
Similar kinetics of isoprene accumulation were noted in both of the SckIspS and kIspS transformants, reaching a maximum at about 72 h after sealing of the culture flasks (Fig. 7a and b). It is of interest to observe that SckIspS transformants (Fig. 7a) accumulated about twice as much isoprene as their kIspS counterparts (Fig. 7b). It was reproducibly observed that, among all transformants, strain ScT16 was more efficient in isoprene production. These quantitative isoprene production results (Fig. 7a) are consistent with the Western blot analyses (Fig. 5a), showing that codon use optimization is an important tool in efforts to improve the performance of transgenes.
Stability of multi-copy integrated IspS in transformants
To investigate whether the multi-copy integration of the IspS transgenes was stably maintained in the yeast transformants, we tested for tryptophan prototrophy, copy number of IspS and IspS activity as a function of time over many generations of yeast cell growth. For tryptophan prototrophy, 100 colonies were tested after the ScT16 strain was cultured in YPD as a nonselective rich medium until 100 generations of growth. All ScT16 colonies showed prototrophy to tryptophan after growth for 100 generations (not shown). Copy number of IspS was determined by Southern blot analysis using genomic DNA isolated from the ScT16 strain before and after 100 generations of growth. The intensity of the transgenic band at 4·1 kb was approximately 1/3 of that of the endogenous rDNA band at 2·06 kb in the samples from both before (Fig. 8a, lanes 1 and 2) and after propagation for 100 generations (Fig. 8a, lanes 3 and 4). Importantly in this respect, the ratio of transgenic/endogenous bands intensity (4·1 kb/2·06 kb) remained the same before and after strain propagation for 100 generations (Fig. 8a).
IspS enzymatic activity of the transformants was also compared, based on isoprene production, which was measured using the ScT16 cultures before and after propagation for 100 generations. No statistically significant differences could be observed between ScT16 cultures before and after 100 generations (Fig. 8b). These results support the notion of stability of the IspS transgene in the S. cerevisiae transformants. They are also consistent with the earlier finding that a high proportion of yeast transformants (> 97% of the colonies) retain their prototrophy characteristics and are stable for over 200 generations under nonselective media conditions (Lopes et al. 1989; Nieto et al. 1999).
Yeast and fungi are commercially employed in large-scale fermentations to generate industrially valuable products. However, it was unclear whether the eukaryotic metabolism of Ascomycota can express the isoprene synthase (IspS) transgene, and whether isoprene might be toxic to the cells. Evidence was presented in this work to show stable heterologous transformation of Saccharomyces cerevisiae with a plant isoprene synthase gene (IspS), conferring to the yeast cells constitutive volatile isoprene (C5H8) hydrocarbons production. Two specific criteria for expression of the IspS gene in S. cerevisiae were applied, aiming to enhance prospects of isoprene hydrocarbons production. One criterion entailed multiple independent integrations of the IspS gene into the rDNA locus in the yeast genome so as to express as high amounts of the IspS protein as possible and thus to generate a genetically stable yeast strain with sustained isoprene production properties. The other criterion was optimization of codon usage of the IspS gene for specific application in S. cerevisiae.
It was previously reported that multi-copy integration can be obtained in S. cerevisiae if a defective selectable marker is used (Lopes et al. 1989; Nieto et al. 1999). In this work, we used TRP1d (conferring tryptophan auxotrophy) as a defective selectable marker to target multiple integrations of the IspS gene into the rDNA region on the yeast chromosome. Conferring tryptophan auxotrophy is an attractive selectable marker for industrial strain construction, as undefined media do not contain this amino acid after sterilization (Lopes et al. 1991). Saccharomyces cerevisiae transformants carrying multi-copy integrations (25–50 copies) of the IspS construct were generated in which the IspS gene was expressed under control of the GPD promoter, a yeast promoter designed to confer constitutive expression of the IspS gene (Fig. 2). Analysis of the transformants showed that they constitutively produced isoprene and showed high mitotic stability, which are desirable features for long-term culturing of yeast in industrial fermentations.
An alternative approach to express the IspS gene would be to employ a standard episomal yeast expression vector for isoprene production. However, episomal expression vectors could easily be lost from the cells in the course of continuous long-term cultivation. An additional disadvantage of episomal yeast expression vectors is the required presence of selectable markers (antibiotics) for the maintenance of the vector. The latter is not desirable for isoprene commercialization conditions, as large-scale use of antibiotics substantially increases cost and also raises environmental issues.
Our experimental approach entailed sealing the cultures to prevent escape of the volatile isoprene gas, followed by a subsequent sampling of the gas in the headspace of the cultures. In the context of this experimental approach, we noted that substantial amounts of isoprene converted into primarily hydroxylated products, that is, 2-methyl-3-buten-1-ol and 3-methyl-1-butanol. Quantitative analysis of the headspace of one of the IspS transformants showed a hydroxylated isoprene-to-isoprene ratio of about 100 : 1. This ratio is probably >100 : 1, as hydroxylated isoprene molecules have greater than isoprene solubility in the aqueous medium; hence, they were not quantitatively present in the reactor headspace. Accordingly, total productivity of isoprene and its hydroxylated products in the IspS transformants is much greater than the 500 μg l−1 estimated in this work. Thus, the average rate of isoprene accumulation in this experiment was >7 μg per l culture per h.
Rates of isoprene accumulation in different transformants are available from the literature. The cyanobacterium Synechocystis photoproduced 4 μg isoprene per l culture per h (Lindberg et al. 2010; Bentley and Melis 2012). A study of isoprene release by Bacillus subtilis indicated rates of 50 μg isoprene per l culture per h (Julsing et al. 2007). Biosynthesis of isoprene in E. coli via the methylerythritol phosphate (MEP) pathway reportedly generated 2·8 mg isoprene per l culture per h (Zhao et al. 2011). In contrast, isoprene production in E. coli batch cultures via the MVA pathway showed rates of 18 mg isoprene per l culture per h (Zurbriggen et al. 2012). These measurements are preliminary and their significance should be moderated by their respective commercialization value. A techno-economic analysis for commercial feasibility of the above-mentioned isoprene production processes is outside the scope of this work. Nevertheless, it is clearly understood that improvements in endogenous substrate partitioning towards isoprene rather than biomass need be achieved with all of the above-mentioned systems before anyone could contemplate commercialization. Approaches to increase the isoprene yield and to prevent or regulate the chemical modification of isoprene in the yeast cell and/or growth media are currently under investigation in this lab. For example, alcohol dehydrogenase gene disruption would prevent endogenous substrate flux towards ethanol and, thus, improve the isoprene-to-ethanol and isoprene-to-biomass carbon-partitioning ratio of the cultures.
Commercially, isoprene is currently generated as a by-product of ethylene production from a petroleum feedstock. However, this work showed that isoprene can be generated from S. cerevisiae through heterologous transformation of the cell with a codon-optimized IspS construct, and by driving isoprene biosynthesis through the MVA pathway. Isoprene is a volatile and highly hydrophobic small hydrocarbon that can apparently pass from the cytosol, where it is generated, and through the yeast cell membranes without exerting any toxic effects, ending up in the headspace of the culture. The spontaneous separation from the cells is a property that simplifies product isolation. Spontaneous and efficient product separation from the biomass also alleviates potential feedback inhibition or product toxicity to the cells, which applies when products accumulate in microbial cells (Fischer et al. 2008). Our results demonstrate the potential of S. cerevisiae serving as a platform for the production of renewable isoprene hydrocarbons through further optimization of the MVA pathway.
Award of Swiss National Science Foundation Postdoctoral Fellowship, grant no. PBBEP3_128360 to A.Z. is gratefully acknowledged. Authors state that they have no conflict of interest to declare.