Improvement in d-xylose utilization and isobutanol production in S. cerevisiae by adaptive laboratory evolution and rational engineering

Abstract

As the effects of climate change become apparent, metabolic engineers and synthetic biologists are exploring sustainable sources for transportation fuels. The design and engineering of microorganisms to produce gasoline, diesel, and jet fuel compounds from renewable feedstocks can significantly reduce our dependence on fossil fuels as well as lower the emissions of greenhouse gases. Over the past 2 decades, a considerable amount of work has led to the development of microbial strains for the production of advanced fuel compounds from both C5 and C6 sugars. In this work, we combined two strategies—adaptive laboratory evolution and rational metabolic engineering—to improve the yeast Saccharomyces cerevisiae’s ability to utilize d-xylose, a major C5 sugar in biomass, and produce the advanced biofuel isobutanol. Whole genome resequencing of several evolved strains followed by reverse engineering identified two single nucleotide mutations, one in CCR4 and another in TIF1, that improved the yeast’s specific growth rate by 23% and 14%, respectively. Neither one of these genes has previously been implicated to play a role in utilization of d-xylose. Fine-tuning the expression levels of the bottleneck enzymes in the isobutanol pathway further improved the evolved strain’s isobutanol titer to 92.9 ± 4.4 mg/L (specific isobutanol production of 50.2 ± 2.6 mg/g DCW), a 90% improvement in titer and a 110% improvement in specific production over the non-evolved strain. We hope that our work will set the stage for an economic route to the advanced biofuel isobutanol and enable efficient utilization of xylose-containing biomass.

Electronic supplementary material

The online version of this article (10.1007/s10295-020-02281-9) contains supplementary material, which is available to authorized users.

Introduction

Concerns over global warming and unsustainable demands have increased interest in renewable fuels and chemicals [1, 2]. Microbial production of biofuels using cost-competitive and environmentally friendly bioprocesses is a viable alternative to the traditional production of transportation fuels [3–5]. Notable examples of microbially produced biofuels include branched-chain and higher alcohols as gasoline replacements and fatty acid ethyl ester (FAEE) and long-chain alkanes as diesel replacements [6–12]. In many of these examples, the budding yeast S. cerevisiae was the host of choice because of its ideal properties as a chemical production platform. These properties include (1) high tolerance to many industrial stresses, including low pH and high osmotic pressure; (2) high tolerance to the biofuel compound of interest and contaminants from the pretreatment steps; (3) availability of “omics” data and genetic tools for bottom-up strain engineering; and (4) resistance to phage infection. A major drawback of S. cerevisiae, however, is its inability to utilize d-xylose, a five-carbon (C5) sugar that is the second most abundant sugar in lignocellulosic biomass (~ 30%–40%), as a carbon source [13].

Our lab and others have previously engineered the yeast S. cerevisiae to produce the advanced biofuel isobutanol directly from d-xylose (Fig. 1) [14–17]. In the seminal study by Brat and Boles, overexpression of the xylose isomerase XylA from Clostridium phytofermentans along with isobutanol pathway enzymes in S. cerevisiae led to an engineered strain that produced isobutanol at a yield of 0.16 mg/g d-xylose [14]. More recently, Zhang and coworkers engineered S. cerevisiae to express the xylose isomerase XylA from Piromyces sp., the xylulokinase Xyl3 from Pichia stipitis, as well as with the mitochondrially targeted isobutanol pathway enzymes. Deletion of several endogenous genes that have previously been shown to improve xylose utilization resulted in a final engineered strain that produced isobutanol at a yield of 35.8 mg/g d-xylose [17].

Fig. 1

Engineered isobutanol production from d-xylose in S. cerevisiae. HXT7  _F79S high-affinity glucose transporter with an F79S point mutation , XR xylose reductase , XDH xylitol dehydrogenase , XKS xylulokinase , Ilv2 acetolactate synthase , Ilv5 acetohydroxyacid reductoisomerase , Ilv3 dihydroxyacid dehydratase , kivDmit mitochondrially targeted keto-acid decarboxylase , ADH7mit mitochondrially targeted alcohol dehydrogenase . Enzymes highlighted in red are overexpressed

While both of these works utilized the xylose isomerase pathway (XI pathway), our previous effort employed the xylose reductase–xylitol dehydrogenase pathway (XR-XDH pathway) [15]. In specific, our strategy involved expression of the xylose-specific sugar transporter to enable transport of d-xylose into the cell; overexpression of xylulokinase (XK), xylose reductase (XR) and xylitol dehydrogenase (XDH) from Scheffersomyces stipitis; overexpression of the mitochondrially targeted isobutanol pathway enzymes; and deletion of the endogenous aldose reductase gene GRE3 and the alkaline phosphatase gene PHO13. The resulting strain (PWY2353) was able to grow in a yeast minimal medium containing d-xylose as the sole carbon source and produced isobutanol at a titer of 19.7 ± 2.4 mg/L (specific isobutanol production of 11.0 ± 1.4 mg/g DCW) after 48 h of fermentation.

Adaptive laboratory evolution (ALE) is a powerful and widely used technique for untargeted strain improvement and has been employed in a variety of microbial hosts [18]. Notable examples of ALE in S. cerevisiae include the identification of genetic targets that lead to tolerance against a variety of industrially relevant stresses such as high temperature [19], high glucose concentration [20], and high medium-chain alcohols’ content [21]. However, ALE has not been applied to improve isobutanol production from d-xylose before. Here, we used ALE to improve our engineered yeast PWY2353′s ability to utilize d-xylose. We subjected PWY2353 to an ALE experiment in the growth medium containing 2% d-xylose as the sole carbon source to enrich for S. cerevisiae strains with improved d-xylose utilization. After twelve 2-day rounds of evolution, we identified strains with significantly enhanced growth in the d-xylose-containing medium. Next, we performed whole genome re-sequencing of four selected evolved strains to identify the genetic targets that conferred the evolved strains with this trait. In specific, we found an A638S mutation in CCR4 and an A79S mutation in TIF1, two proteins with previously known roles in regulatory responses to cellular stresses but no direct links to xylose utilization or isobutanol production. Reverse engineering of the evolved strains by integrating these two specific mutations into the non-evolved PWY2353 strain confirmed their roles in improving growth on d-xylose. Finally, we fine-tuned the expression levels of the bottleneck isobutanol pathway enzymes in the best evolved strain (PWY2353ev34). The resulting strain (PWY2353ev34b) produced isobutanol at a titer of 92.9 ± 4.4 mg/L, a 90% improvement over the levels observed in the parental strain PWY2353.

Material and methods

Yeast strain, media and transformation

The yeast strains used in this study were constructed from BY4742 (derivative of S288C (Mat α; his3Δ1; leu2Δ0; lys2Δ0; ura3Δ0)) and PWY2353 [15] (Table 1). The CRISPR–Cas9 plasmids used in this study for genome editing were generated from pRPR1-gRNA handle-RPR1t [22] and p414-TEF1p-Cas9-CYC1t [23]. The primers used in this study are listed in Table S1. Yeast and bacterial strains were stored in 25% glycerol at − 80 °C. E. coli was grown in Luria–Bertani medium. Ampicillin at 100 μg/mL was added to the medium when required. Yeast strain without plasmid was cultivated in YPD medium (10 g/L yeast extract, 20 g/L Bacto Peptone and 20 g/L glucose). Yeast cells were transformed using the LiAc/SS Carrier DNA/PEG method as previously described [24, 25]. Selection of yeast transformants with URA3 was done on a yeast minimal medium (6.7 g/L of Yeast Nitrogen Base (Difco), 20 g/L glucose, and a mixture of appropriate nucleotide bases and amino acids with Uracil dropouts (CSM-URA).

Overview of different microbial strains generated in this study

Strain name	Genotype	Description	Specific growth rate in xylose (h−1)a	References
BY4742	Mat α; his3Δ1; leu2Δ0; lys2Δ0; ura3Δ0	Laboratory strain	N.D	Lab strain
PWY2353	BY4742 ARS208:: P  TEF1  -SsXKS; YMRWΔ15:: P  TEF1  -HXT  F79S; ARS308:: P  TEF1  -SsXYL1; ARS720:: P  TEF1  -SsXYL2; YORWΔ22:: P  TDH3   –LlkivDmt-T2A-ScADH7mt; ARS1309:: P  TEF1  -ScIlv2-T2A-ScIlv5-T2A-ScIlv3; GRE3Δ; PHO13Δ	BY4742 overexpressing XKS, XYL1 and XYL2 from Scheffersomyces stipitis, HXT7  F79S, Ilv2, Ilv5, Ilv3, ADH7 from Saccharomyces cerevisiae, and kivD from L. lactis with GRE3 and PHO13 deleted	0.0229 ± 0.0005	Promdonkoy et al . [15]
PWY2353ev17	PWY2353; adaptive laboratory evolution clone No. 17	PWY2353 subjected to adaptive laboratory evolution, clone No. 17	0.0300 ± 0.0011	This study
PWY2353ev34	PWY2353; adaptive laboratory evolution clone No. 34	PWY2353 subjected to adaptive laboratory evolution, clone No. 34	0.0363 ± 0.0003	This study
PWY2353ev36	PWY2353; adaptive laboratory evolution clone No. 36	PWY2353 subjected to adaptive laboratory evolution, clone No. 36	0.0341 ± 0.0013	This study
PWY2353ev38	PWY2353; adaptive laboratory evolution clone No. 38	PWY2353 subjected to adaptive laboratory evolution, clone No. 38	0.0290 ± 0.0011	This study
PWY2363	PWY2353 ARS416:: P  TDH3   –LlkivDmt-T2A-ScADH7mt	PWY2353 overexpressing one extra copy each of ADH7 from S. cerevisiae and kivD from L. lactis	0.0260 ± 0.0002	This study
PWY2373	PWY2353 ARS416:: P  TDH3   –LlkivDmt-T2A-ScADH7mt; ARS1014:: P  TEF1  -ScIlv2-T2A-ScIlv5-T2A-ScIlv3	PWY2353 overexpressing one extra copy each of Ilv2, Ilv5, Ilv3, ADH7 from S. cerevisiae and kivD from L. lactis	0.0273 ± 0.0005	This study
PWY2353ev34a	PWY2353ev34 ARS416:: P  TDH3  –LlkivDmt-T2A-ScADH7mt	PWY2353ev34 overexpressing one extra copy each of ADH7 from S. cerevisiae and kivD from L. lactis	0.0323 ± 0.0004	This study
PWY2353ev34b	PWY2353ev34 ARS416:: P  TDH3   –LlkivDmt-T2A-ScADH7mt; ARS1014:: P  TEF1  -ScIlv2-T2A-ScIlv5-T2A-ScIlv3	PWY2353ev34 overexpressing one extra copy each of Ilv2, Ilv5, Ilv3, ADH7 from S. cerevisiae and kivD from L. lactis	0.0317 ± 0.0007	This study
PWY2383	PWY2353 CCR4::CCR4  A638S	PWY2353 whereby G at position 1912 of CCR4 is mutated to T resulting in the amino acid change Ala638Ser	0.0282 ± 0.0009	This study
PWY2393	PWY2353 CCR4::CCR4  A638S  ; TIF1::TIF1  A79S	PWY2353 whereby G at position 1912 of CCR4 is mutated to T resulting in the amino acid change Ala638Ser and G at position 235 of TIF1 is mutated to T resulting amino acid change Ala79Ser	0.0323 ± 0.0016	This study

N.D. ‘not detected’ due to strain’s inability to grow

^aSpecific growth rates were determined using the BioLector Pro microbioreactor system

Plasmid construction

Plasmid p414-TEF1p-Cas9-CYC1t-HIS

The TRP1 selectable marker in p414-TEF1p-Cas9-CYC1t was replaced with the HIS3 selectable marker. The HIS3 gene fragment was amplified from the pSH62 plasmid using primers HIS3-hom-F and HIS3-hom-R. The 1.2-kb PCR band was gel-purified and then mixed with the purified MfeI-digested p414-TEF1p-Cas9-CYC1t fragment to form p414-TEF1p-Cas9-CYC1t-HIS via homologous recombination in S. cerevisiae.

Plasmid pRPR1-gRNA-CCR4

The gRNA-CCR4 fragment was amplified from the genomic DNA of PWY2353ev34 using primers CCR4_gRNA_F and univl_gRNA_R. The 0.13-kb PCR band was gel-purified and ligated to the HindIII/XhoI site of pRPR1-gRNA handle-RPR1t to yield pRPR1-gRNA-CCR4.

Plasmid pRPR1-gRNA-TIF1

The gRNA-TIF1 fragment was amplified from the genomic DNA of PWY2353ev34 using primers TIF1_gRNA_F and univl_gRNA_R. The 0.13-kb PCR band was gel-purified and ligated to the HindIII/XhoI site of pRPR1-gRNA handle-RPR1t to yield pRPR1-gRNA-TIF1.

Plasmid pJET1.2-CCR4_A638S-donor

PCR-based Site-Directed Mutagenesis was employed to generate the CCR4_A638S-donor DNA fragment that contains the point mutation in CCR4 observed in PWY2353ev34 as well as to remove the PAM site adjacent to the crRNA target. Two overlapping fragments were amplified from the genomic DNA of PWY2353ev34 using two sets of primers: upstream mutation point, CCR4_seq_F and CCR4_don_R1; downstream mutation point, CCR4_don_F2 and CCR4_seq_R. To combine the upstream and downstream PCR fragments, the second PCR was performed using primers CCR4_seq_F and CCR4_seq_R. The 1-kb PCR band was gel-purified and ligated with the blunt pJET1.2 cloning vector (Thermo Fisher Scientific) to yield pJET1.2-CCR4_A638S-donor.

Plasmid pJET1.2-TIF1_A79S-donor

PCR-based site-directed mutagenesis was employed to generate the TIF1_A79S-donor DNA fragment that contains the point mutation in TIF1 observed in PWY2353ev34 as well as to remove the PAM site adjacent to the crRNA target. Two overlapping fragments were amplified from the genomic DNA of PWY2353ev34 using two sets of primers: upstream mutation point, TIF1_seq_F and TIF1_don_R1; downstream mutation point, TIF1_don_F2 and TIF1_seq_R. To combine the upstream and downstream PCR fragments, the second PCR was performed using primers TIF1_seq_F and TIF1_seq_R. The 1-kb PCR band was gel-purified and ligated with the blunt pJET1.2 cloning vector (Thermo Fisher Scientific) to yield pJET1.2-TIF1_A79S-donor.

Adaptive laboratory evolution (ALE) of PWY2353

The parental strain (PWY2353) was cultured in 5 mL yeast minimal medium SD-URA containing 2% glucose [6.7 g/L of Yeast Nitrogen Base (Difco), 20 g/L glucose, and a mixture of appropriate nucleotide bases and amino acids with Uracil dropouts] in a 50-mL tube at 30 °C and 250 rpm in an orbital shaking incubator for 16 h. The overnight cultures were transferred to fresh 20 mL SX-URA medium containing 2% d-xylose [6.7 g/L of Yeast Nitrogen Base (Difco), 20 g/L d-xylose, and a mixture of appropriate nucleotide bases and amino acids with Uracil dropouts (CSM-URA)] to obtain an initial OD₆₀₀ of 0.01 and the cells were grown until the exponential phase was reached. Then, the cells were subcultured again into fresh 20 mL SX-URA medium and this process was repeated for a total of 12 rounds. The cells from the last round were diluted and spread on SX-URA agar plates. The plates were incubated at 30 °C for 3 days and the 40 biggest colonies were selected and cultured in SX-URA broth. OD₆₀₀ measurement of cells was performed to verify growth at various times till 48 h.

Specific growth rates for four selected strains were obtained using the BioLector microfermenter system (m2p labs). The strains were pre-cultured in 5 mL SD-URA medium at 30 °C in an orbital shaking incubator at 250 rpm for 16 h. The overnight cultures were washed and diluted with water to obtain an OD₆₀₀ of 2.0, and 25-μL aliquots of the diluted cultures were used to inoculate 975 μL SX-URA medium in a 48-well microflower plate. The cultures were cultivated in the BioLector microfermenter (m2p-labs) at 30 °C at 1100 rpm for 48 h for real-time measurement of biomass, pH and dissolved oxygen. The experiments were performed in biological triplicate. Scattered light measurements from the BioLector were converted to biomass (dry cell weight) by generating a standard curve using exponentially grown S. cerevisiae BY4742 cells and comparing a dilution series on the BioLector and dry cell weight.

Isolation and sequencing of genomic DNA from evolved strains

The genomic DNA of evolved strains PWY2353ev17, PWY2353ev34, PWY2353ev36, PWY2353ev38 and the parental strain (PWY2353) was isolated using Wizard Genomic DNA Purification Kit (Promega). Purified genomic DNA samples were submitted to Novogene for whole genome re-sequencing. Quality control of the 150 nt pair-end reads was performed using FASTP [26], and the clean sequences were then mapped against the genome sequence of S. cerevisiae S288C via BWA-MEM version 0.7.17 [27]. The genomic variant calling for all strains was analyzed using the Genome Analysis Toolkit (GATK4) [28]. The mutated genes CCR4 and TIF1 were amplified using primers CCR4_seq_F and CCR4_seq_R, TIF1_seq_F and TIF1_seq_R, respectively, for sequence confirmation. The sequencing data were deposited in Genbank as follows (strain, accession number): PWY2353, SAMN14558819; PWY2353ev17, SAMN14558820; PWY2353ev34, SAMN14558821; PWY2353ev36, SAMN14558822; PWY2353ev38, SAMN14558823.

Markerless genome editing of PWY2353 to generate point mutations

The CRISPR/Cas9 system was employed to edit the genome of PWY2353 to generate specific point mutations. Donor DNA for CCR4_A638S and TIF1_A79S was amplified from plasmids pJET1.2-CCR4_A638S-donor and pJET1.2-TIF1_A79S-donor using primers CCR4_seq_F and CCR4_seq_R, and TIF1_seq_F and TIF1_seq_R, respectively. The 1-kb PCR band of each donor DNA was gel-purified and then transformed into competent PWY2353 cells along with the corresponding gRNA plasmid and p414-TEF1p-Cas9-CYC1t-HIS. Colony PCR of transformants were performed using the corresponding primers. These are primers CCR4_seq_F and CCR4_seq_R for PWY2353 CCR4::CCR4  _A638S (named PWY2383); primers TIF1_seq_F and TIF1_seq_R for PWY2353 CCR4::CCR4  _A638S  TIF1::TIF1  _A79S (named PWY2393). The PCR products were gel-purified and sequenced to verify the point mutation in the genome. Specific growth rates for PWY2383 and PWY2393 were obtained using the BioLector microfermenter system (m2p-labs) as described in “Adaptive laboratory evolution (ALE) of PWY2353”.

Integrating an additional copy of the isobutanol pathway genes into the genome of PWY2353ev34

To integrate one additional copy each of LlkivD and ScADH7 into the genome of PWY2353ev34, the LlkivDmit-T2A-ScADH7mit expression construct, which also contains the URA3 marker, was amplified from plasmid pUG72-TDH3-LlkivDmit-2A-ScADH7mit [15] using primers ARS416_int_F and ARS416_int_R. These primers contain 42-bp overhangs on either side that are homologous to the ARS416 integration site. This integration site, along with the ARS1014 integration site used for the second round of integration, has previously been shown to lead to high expression level of the inserted gene [29]. The PCR product was gel-purified and transformed into PWY2353ev34 to create strain PWY2353ev34a. The marker gene (URA3) was removed by overexpressing the Cre recombinase to excise the selection marker between the loxP sites in the disruption cassette. This enables subsequent rounds of genomic integrations. Cre recombinase was expressed using the inducible GAL1 promoter on plasmid pSH62 [30]. The strain harboring pSH62 was grown in SD medium plus 1 g/L 5-fluoroorotic acid to encourage loss of the URA3 [31]. To verify the genetic stability of the engineered strain, PWY2353a’s genomic DNA was isolated (Promega Wizard Genomic DNA Purification kit) and then subjected to a diagnostic PCR amplification that amplified regions both upstream and downstream of the integration/deletion sites.

To integrate one additional copy each of ScIlv2, ScIlv5 and ScIlv3 into the genome of PWY2353ev34a, the ScIlv2-T2A-ScIlv5-T2A-ScIlv3 expression construct, which also contains the URA3 marker, was amplified from plasmid pRS416Tef1-ScIlv2-2A-ScIlv5-2A-ScIlv3 [15] using primers ARS1014_int_F and ARS1014_int_R. These primers contain 42-bp overhangs on either side that are homologous to the ARS1014 integration site. The PCR product was gel-purified and transformed into PWY2353ev34a to create strain PWY2353ev34b. Removal of the URA3 marker gene and strain verification were performed as described above.

Quantification of isobutanol production in engineered strains

Engineered strains were pre-cultured in 5-mL aliquots in a yeast minimal medium SD-URA overnight and used to inoculate 10 mL minimal yeast medium SX-URA in 50-mL Falcon tubes to achieve an initial OD₆₀₀ of 0.05. The cultures were grown at 30 °C and 250 rpm in an orbital shaking incubator. Samples were taken at 10, 24, 48, 72 and 144 h to determine OD₆₀₀, biomass, extracellular metabolites and production of higher alcohols. The amount of isobutanol and other extracellular metabolites was determined using high-performance liquid chromatography (HPLC) as previously reported with small modifications [15]. In brief, 1 mL of culture was centrifuged at 18,000g for 5 min and the supernatant was filtered through 0.2-μm nylon syringe filter (Filtrex). The purified sample was then applied to an Agilent 1100 series HPLC equipped with an Aminex HPX-87H ion exchange column (Bio-Rad). The HPLC program was performed using 5 mM H₂SO₄ as the solvent at a flow rate of 0.68 mL/min for 30 min. The column was maintained at 60 °C. All metabolites were detected with Agilent 1200 series DAD and RID detectors.

RNA isolation and transcript quantification

Strains were pre-cultured in 5-mL aliquots in yeast minimal medium SD-URA overnight and used to inoculate 5 mL minimal yeast medium SX-URA in 50-mL falcon tubes to achieve an initial OD₆₀₀ of 0.05. After 48 h, a 1-mL aliquot of each culture was collected and centrifuged for 2 min at 9000g. Total RNA was extracted using the QIAgen RNeasy Kit under the manufacturer’s protocol. Contaminating genomic DNA was removed from the RNA samples by DNaseI (NEB) digestion using the manufacturer’s protocol. The RNA quantity was analyzed using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies), and samples were stored at − 80 °C until RT-PCR analysis. cDNA was obtained using RevertAid Reverse Transcriptase (Thermo Fisher Scientific) using the manufacturer′s protocol. Relative expression levels of LlkivD-ScADH7, ScIlv2-ScIlv5-ScIlv3 and ScATF1 were quantified using iQ SYBR Green Kit (Bio-Rad) on CFX96 Touch Real-time PCR Detection System (Bio-Rad). Real-time PCR was performed in triplicates, and ScTAF10, a gene that encodes a subunit of transcription factor IID (TFIID), was used to normalize the amount of the total mRNA in all samples. Primers for real-time PCR are listed in Table S1.

Results and discussion

Adaptive laboratory evolution (ALE) of engineered S. cerevisiae led to a strain with improved d-xylose utilization

We first subjected PWY2353 to an ALE experiment in the growth medium containing 2% d-xylose as the sole carbon source to enrich for S. cerevisiae strains with improved d-xylose utilization. The culture was grown until the exponential phase and then the cells were subcultured again into fresh medium. After repeating this process for a total of 12 rounds, we isolated 40 strains from this population as single colonies on plates and assessed their preliminary growth profile in yeast minimal medium containing 2% d-xylose as the sole carbon source (Fig. 2a). From these, we picked the top four strains, PWY2353ev17, PWY2353ev34, PWY2353ev36 and PWY2353ev38, based on their final OD₆₀₀ values after 48 h, and performed a detailed study of their growth profile compared to parental (non-evolved) strain PWY2353 using the BioLector Pro microfermenter (m2p-labs) (Fig. 2b). Overall, the four evolved strains exhibited 26.6–58.3% higher specific growth rates compared to the value for PWY2353 (0.0229 ± 0.0005 h⁻¹) (Table 1 and Fig. 2c). In particular, strain PWY2353ev34 had the highest specific growth rate at 0.0363 ± 0.0003 h⁻¹, followed by strain PWY2353ev36 at 0.0341 ± 0.0013 h⁻¹.

Fig. 2

Adaptive laboratory evolution of PWY2353 resulted in strains with improved specific growth rates. Growth profile of 40 selected strains after 12 rounds of sub-culturing in liquid minimal medium with 2% d-xylose as the sole carbon source (a). Detailed growth profiles (b) and specific growth rates (c) of the four fastest-growing evolved strains and control strain (non-evolved PWY2353) in yeast minimal medium with 2% d-xylose as the sole carbon source. Specific growth rates were determined using the BioLector Pro microfermenter (m2p-labs)

In addition to cell growth, we also looked at isobutanol production of the selected evolved strains. Interestingly, there appears to be no direct correlation between isobutanol titers and specific growth rates in the selected evolved strains. Only one evolved strain, PWY2353ev34, produced isobutanol at a higher titer (60.6 ± 1.3 mg/L) than the level observed in the non-evolved strain PWY2353 (48.9 ± 2.8 mg/L) (Table 2). This increase in isobutanol was accompanied by a decrease in ethanol production with strain PWY2353ev34 producing 0.254 ± 0.002 g/L, a 55% decrease from the PWY2353’s value of 0.569 ± 0.022 g/L. The levels of xylitol, the only other major metabolite detected, were similar in both strains. Strain PWY2353ev36 produced isobutanol at similar titer to the PWY2353 at 49.2 ± 3.6 mg/L, while PWY2353ev17 and PWY2353ev38 produced isobutanol at lower titers of 36.0 ± 2.5 mg/L and 25.6 ± 2.3 mg/L, respectively.

Production of isobutanol and other metabolites in engineered and evolved strains

Strain name	Isobutanol titer in xylose (mgL−1)	Specific isobutanol production (mg gDCW−1)	Isobutanol productivity after 2 days (6 days) in xylose (mg gDCW−1 h−1)	Ethanol titer (gL−1)	Xylitol titer (gL−1)
PWY2353	48.9 ± 2.8	23.9 ± 1.4	0.291 ± 0.037 (0.166 ± 0.010)	0.569 ± 0.022	0.928 ± 0.175
PWY2353ev17	36.0 ± 2.5	20.1 ± 1.5	0.206 ± 0.009 (0.140 ± 0.010)	0.190 ± 0.006	0.412 ± 0.024
PWY2353ev34	60.6 ± 1.3	31.6 ± 1.8	0.392 ± 0.032 (0.220 ± 0.013)	0.254 ± 0.002	1.014 ± 0.014
PWY2353ev36	49.2 ± 3.6	24.9 ± 1.9	0.307 ± 0.015 (0.173 ± 0.013)	0.598 ± 0.027	1.224 ± 0.091
PWY2353ev38	25.6 ± 2.3	12.0 ± 1.1	0.148 ± 0.015 (0.083 ± 0.007)	0.321 ± 0.004	1.037 ± 0.027
PWY2363	63.8 ± 8.4	30.7 ± 0.8	0.441 ± 0.027 (0.213 ± 0.005)	0.546 ± 0.007	1.218 ± 0.052
PWY2373	66.7 ± 2.1	32.4 ± 1.2	0.385 ± 0.022 (0.225 ± 0.008)	0.683 ± 0.010	1.089 ± 0.036
PWY2353ev34a	74.1 ± 3.0	39.6 ± 2.3	0.487 ± 0.008 (0.275 ± 0.016)	0.289 ± 0.014	0.999 ± 0.062
PWY2353ev34b	92.9 ± 4.4	50.2 ± 2.6	0.588 ± 0.040 (0.348 ± 0.018)	0.318 ± 0.009	0.880 ± 0.017

Cumulatively, our results indicate that d-xylose utilization is a phenotype that can be improved by adaptive laboratory evolution and underscore the robustness of the strategy in improving desirable, industrially useful phenotypes in microbial hosts in a relatively short amount of time. Conversely, there remain many traits—for example, production of isobutanol, a compound that serves our interest but provides no apparent benefit to survival of the cell—that cannot be improved by adaptive laboratory evolution in a straightforward manner. For these traits, linking them to cell survival is critical to the success of the evolution strategy [18].

Genome sequencing of evolved strains identified several key mutations

To identify the specific genetic targets that lead to improved d-xylose utilization, we sequenced the genomic DNA from the top four evolved strains. We identified mutations by comparing the genome sequence of each strain with that of the parental strain (PWY2353), which was also sequenced. All of the mutations for each strain are listed in Table 3 and shown in Fig. 3a. Interestingly, all four evolved strains had an A638S mutation in Carbon Catabolite Repression 4 gene (CCR4), which encodes the 3′-5′-exoribonuclease subunit of the core CCR4-NOT complex. Present in all eukaryotes, the CCR4-NOT complex is a multisubunit complex that performs crucial roles in several cellular pathways [32]. First, it regulates transcription at multiple steps from production and processing of messenger RNAs (mRNAs) in the nucleus to their degradation in the cytoplasm via deadenylation-dependent mRNA decay [33]. Second, the CCR4-NOT complex is involved in DNA replication and regulation of cell cycle [34]. Third, the CCR4-NOT complex controls the activity of the environmental stress transcription factor MSN2 [35]. Taken together, these previous studies demonstrate the CCR4-NOT complex role in helping the cells respond to changes in environmental conditions. Though we were unable to find any report that implicates CCR4 in d-xylose utilization, several works indicate that the protein is required for expression of many essential genes under glucose-depletion conditions [36, 37].

Summary of mutations in evolved strains identified by whole-genome resequencing

Strain	Mutation	Gene	Protein function
PWY2353ev17	A638S	CCR4	3′-5′-exoribonuclease subunit of the core CCR4-NOT complex
PWY2353ev34	A638S	CCR4	3′-5′-exoribonuclease subunit of the core CCR4-NOT complex
	A79S	TIF1	a subunit of the translation initiation factor 4F complex
PWY2353ev36	A638S	CCR4	3′-5′-exoribonuclease subunit of the core CCR4-NOT complex
	− 577T > C (nucleotide change in the promoter region)	HXT15	Hexose transporter
PWY2353ev38	A638S	CCR4	3′-5′-exoribonuclease subunit of the core CCR4-NOT complex
	S155L	GIP2	a regulatory subunit of phosphatase Glc7p protein

 

Fig. 3

Genome re-sequencing of evolved strains revealed several SNPs that enable the engineered S. cerevisiae to more efficiently utilize d-xylose. Distribution of four genes with single nucleotide variants (SNVs) found in the four selected evolved strains (a). Detailed growth profiles (b) and specific growth rates (c) of strains PWY2383 and PWY2393, which harbor SNVs found in the evolved strain PWY2353ev34. Growth was determined using the BioLector Pro microfermenter (m2p-labs) in liquid minimal medium with 2% d-xylose as the sole carbon source

Another mutation of note was the A79S mutation in the translation initiation factor eIF-4A gene (TIF1), an essential gene that encodes a subunit of the translation initiation factor 4F complex [38]. Translation initiation factor eIF4A is an ATP-dependent D-E-A-D box RNA helicase that accelerates the recruitment of mRNA during the assembly of 48S preinitiation complex [39]. The role of the alanine residue at position 79 of eIF-4A has previously been characterized [40]. Located directly adjacent to the ATPase A motif, A79 was proposed to affect ATP binding and ATPase activity of the enzyme. Importantly, while the yeast eIF-4A protein contains alanine at this position, the mouse variant contains a serine residue, providing evidence that the eIF-4A_A79S mutation found in PWY2353ev34 would not hamper the activity of the enzyme.

Reverse engineering of PWY2353 with CCR4  _A638S and TIF1  _A79S mutations improves d-xylose utilization

Given that PWY2353ev34 had the highest xylose specific growth rate among the evolved strains, we decided to investigate whether the mutations found in PWY2353ev34 contributed to the improved d-xylose utilization phenotype. In specific, we reversed engineered the parental strain PWY2353 to harbor the CCR4  _A638S and TIF1  _A79S mutations. First, we employed the markerless CRISPR–Cas9 system to edit the genome of PWY2353 to harbor CCR4  _A638S [22, 23]. The strategy enabled us to edit a single base pair (1912G > T) in the yeast genome without the use of a selectable marker and without leaving any “scar” in the genome (Fig. 4). The resulting strain, PWY2383, had a d-xylose specific growth rate of 0.0282 ± 0.0009 h⁻¹, which is 23.0% higher than PWY2353′s value of 0.0229 ± 0.0005 h⁻¹ (Fig. 3). The A638S mutation occurs in the nuclease domain of CCR4. The crystal structure of the yeast CCR4-NOT complex has recently been solved but the alanine residue at position 638 was disordered [41]. Bioinformatic analysis of the CCR4  _A638S amino acid sequence using several open-source phosphorylation site prediction software such as NetPhos and Scansite suggested that the serine residue at 638 can serve as a phosphorylation site [42, 43]. However, when we mutated the alanine residue at position at 638 with threonine, another phosphorylatable residue, the resulting strain had a petite phenotype (Figure S2). Having verified the role CCR4  _A638S played in improving xylose growth, we next integrated TIF1  _A79S into the genome of PWY2383 to create strain PWY2393 (Figure S1). This further improved d-xylose specific growth rate to 0.0323 ± 0.0016 h⁻¹, an additional 14.4% improvement over PWY2383’s value and a total of 40.6% improvement over the PWY2353’s value.

Fig. 4

Strategy to reverse engineer PWY2353 with CCR4  _A638S to generate PWY2383

Further strain improvement by fine-tuning expression of bottleneck enzymes

Because adaptive laboratory evolution only improved the yeast’s specific growth on xylose, we turned to rational metabolic engineering to improve the isobutanol production titer. For this effort, we chose the evolved strain PWY2353ev34 due to its highest specific growth on xylose as well as its isobutanol titer. To improve the isobutanol production titer in the evolved strain PWY2353ev34, we fine-tuned the expression levels of the biosynthetic pathway genes. A substantial amount of research in metabolic engineering suggests that the expression levels of metabolic pathway genes can impact the production levels of the compound of interest [5]. Too low an expression level leads to bottlenecks in the pathway, while too high an expression level will deplete the cellular resources that would otherwise be used to generate the target compound.

In our previous work, we varied the pathway enzyme expression levels by adding additional gene copies via the use of CEN6/ARS4 (low-copy) and 2µ (high-copy) plasmid expression systems [15]. We observed the biggest improvement in isobutanol production (a 40% increase in isobutanol titer over the parental strain) when the final two enzymes (LlkivD and ScADH7) were overexpressed using the CEN6/ARS4 plasmid. The second best improvement (an 18% increase in isobutanol titer) was observed when we overexpressed the endogenous l-valine biosynthetic pathway genes (ScIlv2, ScIlv5 and ScIlv3)—also using the CEN6/ARS4 plasmid. Building upon these observations, we integrated one additional copy each of LlkivD and ScADH7 into the genome of PWY2353ev34. The genes were linked together with the self-cleaving 2A peptide sequence from the foot-and-mouth disease virus (FMDV) [44] and placed behind the strong constitutive promoter P_TDH3. Polycistronic expression using the 2A peptide allows multiple proteins to be encoded as polyproteins, which dissociate into individual proteins upon translation. Such a strategy has previously been used to introduce multi-enzyme pathways into several microbial hosts including S. cerevisiae and the methylotrophic yeast Pichia pastoris [45–47]. The resulting strain, PWY2353ev34a, produced isobutanol at a titer of 74.1 ± 3.0 mg/L (specific isobutanol production of 39.6 ± 2.3 mg/g DCW), which corresponds to a 22% improvement in titer over the evolved strain (Fig. 5b, d, Table 2). Encouraged by these results, we next integrated one additional copy each of ScIlv2, ScIlv5 and ScIlv3 into the genome of PWY2353ev34a. The three genes were again linked together with the self-cleaving 2A peptide sequence from FMDV, and placed behind another strong constitutive promoter P_TEF1. The resulting strain, PWY2353ev34b, was able to produce isobutanol at a titer of 92.9 ± 4.4 mg/L (specific isobutanol production of 50.2 ± 2.6 mg/g DCW), which corresponds to a further 25% titer improvement over strain PWY2353ev34a and a 52% improvement in titer over the evolved strain PWY2353ev34 (Fig. 5b, d, f). Moreover, this is a 90% improvement in titer and a 110% improvement in specific production over the non-evolved strain. As a comparison, we also integrated the LlkivD-T2A-ScADH7 gene construct, either alone or in a combination with the ScIlv2-T2A-ScIlv5-T2A-ScIlv3 gene construct, into the genome of the non-evolved strain (PWY2353) to create strains PWY2363 and PWY2373, respectively. The isobutanol titers in these strains were 63.8 ± 8.4 mg/L and 66.7 ± 2.4 mg/L, while their specific isobutanol production were 30.7 ± 0.8 mg/g DCW and 32.4 ± 1.2 mg/g DCW, respectively (Fig. 5a, c, e).

Fig. 5

Strain improvement by integrating an additional copy of isobutanol pathway genes. Fermentation of strains PWY2353 (a). Fermentation of the evolved strain PWY2353ev34 (b). Fermentation of PWY2363, which contains one additional copy each of isobutanol pathway genes LlkivD and ScADH7 (c). Fermentation of PWY2353ev34a, which is the evolved strain PWY2353ev34 that has one additional copy each of isobutanol pathway genes LlkivD and ScADH7 (d). Fermentation of PWY2373, which contains one additional copy each of isobutanol pathway genes LlkivD, ScADH7, ScIlv2, ScIlv5 and ScIlv3 (e). Fermentation of PWY2353ev34b, which is the evolved strain PWY2353ev34 that has one additional copy each of isobutanol pathway genes LlkivD, ScADH7, ScIlv2, ScIlv5 and ScIlv3 (f). Concentration of xylose was plotted on the left, divided by 10, such that the maximum value of 2.5 on the scale indicates 25 g/L of xylose

To verify that the increase in isobutanol production correlated with increased expression levels of the l-valine pathway genes and the keto-acid degradation pathway genes, we performed RT-PCR in the four engineered strains, PWY2363, PWY2373, PWY2353ev34a and PWY2353ev34b, as well as the parental strains PWY2353 and PWY2353ev34 (Fig. 6). The expression levels of LlkivD and ScADH7 were higher in strains PWY2363 and PWY2353ev34a, both of which have an additional copy each of the two genes integrated into the genome, compared to the levels observed in their corresponding parental strains (Fig. 6a, b). Similarly, the expression levels of ScIlv2, ScIlv5 and ScIlv3 were higher in strains PWY2373 and PWY2353ev34b, both of which have an additional copy each of the three genes integrated into the genome, when compared to the levels observed in their corresponding parental strains (Fig. 6c, d). As a control, the expression level of the acetyl-O-acyltransferase gene ScATF1 remained relatively unchanged between the engineered strains and their corresponding parental strains (Fig. 6e, f).

Fig. 6

RT-PCR analysis of isobutanol biosynthetic pathway genes including LlkivD-ScADH7 (a and b), ScIlv2–ScIlv5–ScIlv3 (c and d) and the alcohol O-acyltransferase gene ScATF1 (e and f) in engineered yeast strains

Conclusions

In this work, we improved the yeast Saccharomyces cerevisiae’s ability to utilize d-xylose and produce the advanced biofuel isobutanol using a combination of adaptive laboratory engineering and rational metabolic engineering. Through the former strategy, we identified two single point mutations that improved yeast’s growth on d-xylose: a single point mutation in CC4, which encodes a subunit of the core CCR4-NOT complex that performs crucial roles in several cellular pathways and a single point mutation in TIF1, which encodes a subunit of the translation initiation factor 4F complex required for protein translation. Importantly, neither one of these genes has previously been implicated in the yeast’s ability to utilize d-xylose. Reverse engineering of the parental, non-evolved strain with the two mutations improved the yeast’s specific growth by 23% and 14%, respectively. Finally, using rational metabolic engineering, we fine-tuned the expression levels of the bottleneck enzymes in the isobutanol pathway and further improved the evolved strain’s isobutanol titer to 92.9 ± 4.4 mg/L (specific isobutanol production of 50.2 ± 2.6 mg/g DCW). This is a 90% improvement in titer and a 110% improvement in specific production over the non-evolved strain. While these values are lower in comparison with glucose-based isobutanol production systems, given that xylose comprises a significant portion of biomass, we hope that our work will pave the way for an economic route to the advanced biofuel isobutanol and enable efficient utilization of xylose-containing biomass.

Funding

This work conducted by the Thailand National Center for Genetic Engineering and Biotechnology (BIOTEC) was supported by the Thailand Research Fund (TRF) under Contract No. TRG6180006.

Acknowledgements

We thank Professor George Church (Harvard University) and Professor Timothy K. Lu (Massachusetts Institute of Technology) for the CRISPR/Cas9 plasmids. We thank Justin W. Henceroth for his critical reading of this manuscript.

Author contributions

PP carried out most of the plasmid construction, ALE experiment, strain construction and characterization. WM and VC contributed to the genome resequencing of the evolved strains. ST contributed to the experimental design and drafted the manuscript. WR conceived and designed the study, and drafted the manuscript.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.