Open Access

Rational and combinatorial approaches to engineering styrene production by Saccharomyces cerevisiae

  • Rebekah McKenna1,
  • Brian Thompson1,
  • Shawn Pugh1 and
  • David R Nielsen1Email author
Microbial Cell Factories201413:123

https://doi.org/10.1186/s12934-014-0123-2

Received: 18 July 2014

Accepted: 14 August 2014

Published: 21 August 2014

Abstract

Background

Styrene is an important building-block petrochemical and monomer used to produce numerous plastics. Whereas styrene bioproduction by Escherichia coli was previously reported, the long-term potential of this approach will ultimately rely on the use of hosts with improved industrial phenotypes, such as the yeast Saccharomyces cerevisiae.

Results

Classical metabolic evolution was first applied to isolate a mutant capable of phenylalanine over-production to 357 mg/L. Transcription analysis revealed up-regulation of several phenylalanine biosynthesis pathway genes including ARO3, encoding the bottleneck enzyme DAHP synthase. To catalyze the first pathway step, phenylalanine ammonia lyase encoded by PAL2 from A. thaliana was constitutively expressed from a high copy plasmid. The final pathway step, phenylacrylate decarboxylase, was catalyzed by the native FDC1. Expression of FDC1 was naturally induced by trans-cinnamate, the pathway intermediate and its substrate, at levels sufficient for ensuring flux through the pathway. Deletion of ARO10 to eliminate the competing Ehrlich pathway and expression of a feedback-resistant DAHP synthase encoded by ARO4 K229L preserved and promoted the endogenous availability precursor phenylalanine, leading to improved pathway flux and styrene production. These systematic improvements allowed styrene titers to ultimately reach 29 mg/L at a glucose yield of 1.44 mg/g, a 60% improvement over the initial strain.

Conclusions

The potential of S. cerevisiae as a host for renewable styrene production has been demonstrated. Significant strain improvements, however, will ultimately be needed to achieve economical production levels.

Keywords

StyreneYeastPhenylalanineAromatics

Background

Similar to most other monomers used in conventional plastics production, at present, styrene is derived exclusively from non-renewable feedstocks. More specifically, current styrene production predominantly involves the energy-intensive chemocatalytic dehydrogenation of petroleum-derived ethylbenzene [1,2]. With the global annual demand of styrene expected to surpass 41 million tons by 2020 [3], representing a > $28 billion U.S. market [4], net energy requirements associated with just this single conversion will amount to over 200 trillion BTU of steam each year [5]. Accordingly, concerns over depleting feedstock availability and deleterious environmental impacts, continue to drive interest in developing ‘green’ processes for producing biorenewable replacements to conventional petrochemicals, including monomers such as styrene.

Advances in metabolic and pathway engineering continue to expand the range of conventional monomer compounds that can be synthesized from renewable biomass feedstocks [6-9]. Along these lines, the engineering of a novel and non-natural pathway for styrene biosynthesis from biomass-derived glucose was recently reported using the bacterium Escherichia coli as host [10]. Said pathway, which is illustrated in Figure 1, utilizes endogenous phenylalanine as its immediate precursor. Phenylalanine is first deaminated to trans-cinnamate by phenylalanine ammonia lyase (PAL), encoded by PAL2 from Arabidopsis thaliana. Next, trans-cinnamate is decarboxylated to styrene via phenylacrylate decarboxylase, encoded by FDC1 from Saccharomyces cerevisiae [10,11]. Co-expressing PAL2 and FDC1 in a previously-engineered phenylalanine over-producing E. coli background resulted in styrene titers as high as ~260 mg/L (2.5 mM) in glucose minimal media after 29 h, representing a glucose yield of ~0.07 g/g (0.12 mol/mol; 25% of theoretical) [10]. As a commodity chemical, however, said production metrics must be improved for biologically derived styrene to emerge as a viable alternative to its conventional counterpart [12].
Figure 1

Styrene biosynthesis by engineered S. cerevisiae. Dashed arrows signify multiple steps are involved but not illustrated. Black arrows represent enzymes steps native to S. cerevisiae whereas gray arrows are heterologous; dotted arrows represent multiple enzymatic steps; feedback repression is shown using thin dotted lines with flat heads whereas transcriptional activation is shown using thin dotted lines with a round head and ‘+’; disruption of a gene or regulatory mechanism is signified by ‘X’. Abbreviations: phosphoenolpyruvate (PEP), erythrose-4-phosphate (E-4P), 3-deoxy-D-arabino-heptulosonate-7-phosphate (DAHP), hydroxyphenylpyruvate (HPP), 2-phenylacetaldehyde (PAA).

The engineering of more robust hosts for renewable chemical production is an important aim in industrial biotechnology [13] and, relative to E. coli, yeast biosynthetic platforms often afford several inherent attributes of importance to robust and large–scale renewable chemicals production. Said phenotypes most significantly include faster growth rates, elevated solvent tolerance, and the ability to withstand low temperatures and pH [14-18]. Amongst yeast, S. cerevisiae is a particularly attractive host for metabolic engineering efforts owing to its well characterized genetics, physiology, and metabolism, as well as due to the availability of diverse genetic toolkits for its engineering [19]. Past studies have demonstrated that S. cerevisiae is a suitable host for the renewable production of useful aromatic compounds, including protocatechuate, catechol, vanillin, naringenin, and 2-phenylethanol [20-23], production of styrene, however, has not yet been reported.

The objective of the present study was to demonstrate that styrene biosynthesis from glucose could be systematically engineered in S. cerevisiae. In addition to the aforementioned phenotypic advantages afforded by the use of S. cerevisiae, this aim was further motivated by other relevant factors. For instance, it was hypothesized that improved function of styrene pathway enzymes might be realized in S. cerevisiae since: i) PAL2 is of eukaryotic origin, and ii) FDC1 is native to S. cerevisiae. Furthermore, contingent upon the native regulation of FDC1 expression (as investigated in this study), it is plausible that a functional styrene pathway could be constructed by expressing a single heterologous enzyme (i.e., PAL2), thereby minimizing the effects of metabolic burden. Meanwhile, as S. cerevisiae lacks a natural transporter for phenylalanine efflux [24], it was hypothesized that increased intracellular retention of phenylalanine might enhance its availability to the engineered pathway. These unique features position S. cerevisiae as particularly promising host for renewable styrene production. In this study, classical anti-metabolite selection was first applied to evolve a S. cerevisiae strain capable of over-producing phenylalanine, the styrene pathway precursor. Rational genetic engineering approaches were used to construct the non-natural styrene pathway and further boost precursor availability. While not before applied for styrene bioproduction, this basic approach has been proven effective for engineering bacterial producers of other aromatic chemicals [25-28].

Results and discussion

Evolving phenylalanine over-production by S. cerevisiae

As phenylalanine serves as the immediate endogenous precursor to the styrene pathway, its over-production by S. cerevisiae is an essential pre-requisite to styrene biosynthesis. Thus, to develop a phenylalanine over-production phenotype in S. cerevisiae, a classic antimetabolite selection strategy was first employed [29]. In this case, m-fluoro-DL-phenylalanine was chosen to provide the necessary selection pressure whereas exposure to the chemomutagen EMS increased mutation rates and frequency [30]. In the first round of selection, a total of only two mutants were isolated when using either 18 mg/L (strain 18A) or 22 mg/L (strain 22A) m-fluoro-DL-phenylalanine. As seen in Figure 2, said mutants were subsequently characterized in shake flask cultures. Since phenylalanine is not exported from S. cerevisiae, the established practice of correlating enhanced flux through the pathway with net extracellular accumulation of 2-phenylethanol and 2-phenylacetate (both of which are naturally and readily produced as degradation products of phenylpyruvate, the precursor to phenylalanine; Figure 1) was employed [24]. In S. cerevisiae, it has previously been shown that all available phenylalanine is efficiently shuttled through one or both of these pathways [31], with the relative distribution of products being dictated by the cell’s redox state (specifically, the relative intracellular ratio of NAD + to NADH) [32]. For example, in glucose grown cultures with limited aeration (as would be expected in the sealed shake flasks used in this study), 9:1 mixtures of 2-phenylethanol:2-phenylacetate are typically observed [32]. Throughout this study, the mixture of products obtained was similarly consistent (an average of 89% 2-phenylethanol; Figure 2). Relative to BY4741 (the parent strain and control), strains 18A and 22A showed 3.3- and 6.4-fold improvements in net production of 2-phenylethanol and 2-phenylacetate (0.63 ± 0.05 and 1.13 ± 0.10 mM), respectively.
Figure 2

Evolution of phenylalanine overproducing mutants of S. cerevisiae . Mutants were evolved through the use of EMS mutagenesis and high-throughput selection on solid agar plates using m-fluoro-DL-phenylalanine as anti-metabolite. 2-Phenylethanol (gray) and 2-phenylacetate (hashed) production by isolated mutants was determined after 48 h of growth in SD media by measuring the concentration of 2-phenylethanol and 2-phenylacetate in the supernatant. Error bars reported at one standard deviation from triplicate experiments.

To further deregulate and enhance phenylalanine biosynthesis, a second round of mutagenesis and selection was performed, in this case using the isolated mutants 18A and 22A as parents. The selection pressure was accordingly elevated by increasing the content of m-fluoro-DL-phenylalanine in SD agar plates to 25 mg/L, 50 mg/L, or 75 mg/L. In contrast to the first round, in this case numerous colonies (i.e., > 50) were obtained at all three selection pressures. Thus, to screen just for the best performers, only the fastest growing mutants were selected (i.e., those whose colony forming units were largest after overnight incubation). A total of 18 additional mutants were chosen and subsequently characterized, as above. As seen in Figure 2, the top performing mutant, 22A75D, produced a combined total of about 2.17 ± 0.22 mM 2-phenylethanol and 2-phenylacetate in 48 h, representing ~3.3- and ~21-fold improvements over 22A and BY4741, respectively.

Investigating the evolved phenotypes

A series of characterizations were next performed on strains 22A and 22A75D (with BY4741 as control) to begin to understand the underlying factors responsible for imparting the evolved phenotypes. The native aromatic amino acid biosynthesis pathways of S. cerevisiae are shown in Figure 1, where it can be seen that two known control points are principally responsible regulating metabolite flux. The first occurs at DAHP synthase (for which S. cerevisiae possesses two isoenzymes), which is allosterically feedback inhibited by either phenylalanine (ARO3) or tyrosine (ARO4) [33-35]. The second, meanwhile, occurs at chorismate mutase (ARO7), which converts chorismate to prephenate, the precursor to both phenylalanine and tyrosine. Transcription of ARO7 is repressed in the presence of as little as 0.5 mM tyrosine but remains, however, insensitive to phenylalanine [33,36,37]. Here, overcoming feedback repression of ARO3 thus constitutes a key priority. However, whereas relief from tyrosine repression of ARO4 has been reported to result from a single mutation (K229L) [24], a phenylalanine feedback resistant mutant of ARO3 remains unreported to date.

Sequences of several key genes in the phenylalanine biosynthesis pathway (ARO3, ARO4, ARO7, ARO8, and PHA2) were first determined for all three strains (including coding regions as well as 500 bp upstream of each start codon). Interestingly, however, mutations were not observed in the sequence of any investigated gene, including with respect to both its coding and upstream non-coding regions. Transcription levels of all genes in the phenylalanine biosynthesis pathway (namely ARO1, ARO2, ARO3, ARO4, ARO7, ARO8, ARO9, and PHA2; see Figure 1) were next examined in the mutants 22A and 22A75D and quantified relative to that of the wild-type control (BY4741). The results are compared in Figure 3, wherein it can be seen that, in strain 22A75D, up-regulation of ARO8 was found to be most significant (a 9.3-fold increase), followed by ARO1 (6.8-fold), ARO2 (5.8-fold), and ARO3 (4.5-fold). Note that similar but less significant differences were also observed in strain 22A. Furthermore, only modest increases in ARO4 and ARO7 expression were observed in 22A75D (about 2.7- and 1.8-fold, respectively), with no significant changes occurring in 22A for either gene.
Figure 3

Transcriptional analysis of top phenylalanine overproducing S. cerevisiae mutants. Relative transcript levels of the top first (22A) and second (22A75D) round evolved yeast mutants, normalized to the parent (BY4741). Measured genes included ARO1 (black), ARO2 (right diagonal), ARO3 (dark gray), ARO4 (horizontal), ARO7 (light gray), ARO8 (left diagonal), ARO9 (no fill), and PHA2 (hashed). Error bars reported at one standard deviation from triplicate experiments.

The collective findings point to the prospects of several interesting mechanisms in the mutant strains. For example, with no change to its sequence, the evolved phenotype clearly did not arise as a result of relieving allosteric inhibition at the known bottleneck enzyme, ARO3. However, as a significant increase in its expression was observed in both mutants, this could suggest that up-regulation of wild type ARO3 occurred as an alternative strategy. That is, despite the fact that the wild type enzyme possesses lower specific activity in the presence of phenylalanine, with more copies net DAHP synthase activity may have been sufficiently high so as to effectively overcome the flux bottleneck. Simultaneous up-regulation of ARO1 and ARO2 (the two subsequent steps in the pathway), meanwhile, may have aided in this process by ensuring that produced DAHP was then promptly assimilated further along the pathway, thereby maintaining a maximum driving force. Up-regulated expression of ARO1 has been successfully employed in yeast as a rational approach for enhancing the biosynthesis of cis,cis-muconic acid – a product derived from the shikimate pathway intermediate 3-hydroshikimate [38]. Meanwhile, up-regulation of ARO8 was evolved perhaps as a mechanism to compete with the native activity of ARO10, ensuring that metabolite flux was efficiently routed towards phenylalanine biosynthesis rather than through the degradative Ehrlich pathway [32,39]. This prospect is further supported by the fact that, in contrast to ARO8, no appreciable change was observed with respect to the expression of ARO9, which functions primarily in the reverse direction (i.e., for phenylalanine assimilation from the culture medium) [40].

With no change to either the coding or upstream non-coding regions for any of the four up-regulated genes (ARO1, ARO2, ARO3, and ARO8) another factor must be responsible for this observed result. Increased copy number through genomic amplifications is one possibility, however, a more efficient mechanism may have involved the mutation of one or more transcription factors controlling their expression. GCN4 encodes one such major transcription factor [41], however, further sequencing revealed no changes there either. To identify other prospective transcription factors involved, the promoter regions of the four up-regulated genes were further investigated by aligning the 1000 bp sequences prior to each start codon. Whereas a possible consensus sequence of 5’-AACATC-3’ was located at positions −292, −307, −289, and −290 for ARO1, ARO2, ARO3, and ARO8, respectively, said sequence does not match binding site of any known transcription factor. Among all known transcriptional regulators, eleven are shared between the four up-regulated genes [42] (note, an annotated list is provided in Additional file 1: Table S3). The ability to determine which if any of these regulators are responsible for the evolved changes will only be possible through the collective analysis of their gene sequences, or better, to ensure full elucidation of all changes in the mutants, through whole-genome sequencing. However, given that the achievable titers are still quite modest (a total of only 2.17 ± 0.22 mM 2-phenylethanol and 2-phenylacetate), such an undertaking was deemed as unwarranted at this time. For now, and for the purpose of this study, these efforts were successful in developing a host strain to serve as a test platform for engineering styrene biosynthesis from glucose.

Investigating native FDC1 activity and factors influencing its expression

Although it had previously been shown that, when cultured in the presence of exogenous trans-cinnamate, S. cerevisiae is capable of catalyzing its decarboxylation to styrene [11], several factors related to the native function and expression of FDC1 remained initially unclear and deserving of further investigation. Most importantly, it was wholly unknown as to if, when, and how the native expression of FDC1 would be induced in the context of the styrene pathway and under the culture conditions of interest. BY4741 was initially cultured in SD minimal media supplemented with potential inducers of interest. In addition to trans-cinnamate and phenylalanine (the pathway intermediate and precursor, respectively), p-coumarate and ferulate were also screened as positive controls (note, both are structural homologs of trans-cinnamate and known inducers of trans-cinnamate decarboxylase activity [11]). As seen in Table 1, in vitro trans-cinnamate decarboxylase activity was detected in the lysates of cells cultured in the presence of each of trans-cinnamate, p-coumarate, and ferulate (with the former serving as the strongest inducer), but not with phenylalanine or in the control. With respect to the styrene pathway, this implies that native FDC1 expression will be wholly contingent upon the heterologous expression of PAL2 to provide trans-cinnamate as inducer (a realization that could be of benefit with respect to minimizing overall metabolic burden).
Table 1

Assaying the in vitro decarboxylase activity of FDC1 against a pool of structurally-related, phenylacrylic acid substrates

Compound

Induced activity

mU mg −1 total protein

trans-cinnamate

+

0.46 ± 0.02

p-coumarate

+

0.39 ± 0.02

ferulate

+

0.21 ± 0.03

phenylalanine

-

N.D.

none (control)

-

N.D.

Positive, ‘+’; Negative, ‘-‘; Not Detected, ‘N.D.’.

Evaluating the styrene pathway via the exogenous addition of phenylalanine

Preliminary studies were next performed to begin probing the functionality of the styrene pathway in wild type S. cerevisiae. Strains BY4741-PAL and BY4741ΔFDC1-PAL were first cultured in SD-Leu minimal media supplemented with 200 mg/L (1.21 mM) phenylalanine while monitoring the extracellular accumulation of trans-cinnamate, styrene, and 2-phenylethanol, 2-phenylacetate, and concomitant depletion of phenylalanine. As seen in Figure 4, while only 37% of supplied the phenylalanine was consumed by BY4741-PAL after 24 h, styrene and 2-phenylethanol constituted the major end-products, reaching titers of up to 20 ± 1 and 43 ± 1 mg/L (0.19 ± 0.01 and 0.35 ± 0.01 mM), respectively, with trans-cinnamate being undetected. In contrast, 2-phenylethanol was produced to a final titer of 98 ± 3 mg/L (0.80 ± 0.02 mM) by BY4741ΔFDC1-PAL with styrene being undetected throughout. In addition, in this case the extracellular accumulation of trans-cinnamate was also observed, reaching final concentration of 26 ± 3 mg/L (0.18 ± 0.02 mM) by 24 h. Only trace levels of 2-phenylacetate were observed throughout.
Figure 4

Assessing the trans-membrane export of trans -cinnamate. Depletion of exogenous phenylalanine (black; initially 200 mg/L) by growing cultures of wild-type S. cerevisiae BY4741-PAL and BY4741ΔFDC1-PAL and the resultant production of trans-cinnamate (diagonal), styrene (horizontal), and 2-phenylethanol (gray) after 24 h. Error bars reported at one standard deviation from triplicate experiments.

These results demonstrate several key points. First, styrene can be produced from phenylalanine by S. cerevisiae via the heterologous expression of PAL2 and native expression of FDC1. Second, both styrene and trans-cinnamate are naturally excreted from S. cerevisiae, at least to a certain degree. Third, since trans-cinnamate did not accumulate in BY4741-PAL cultures, this implies that the net activity afforded by native FDC1 expression was sufficiently high so as to preclude the creation of a flux bottleneck at the final pathway step (at least with respect to the specific PAL2 expression level examined). And, lastly, that under the conditions examined, synthesis of byproduct 2-phenylethanol significantly competes with the styrene pathway for precursor availability, even when PAL2 is constitutively expressed on a high copy number plasmid.

Styrene production from glucose

Based on the above findings, a series of strains were next constructed and evaluated with respect to their styrene production potential from glucose, the results of which are compared in Figure 5. Although only as much as ~5 mg/L (0.05 mM) styrene was detected in BY4741-PAL cultures, 22A75D-PAL accumulated up to 18 ± 2 mg/L (0.17 ± 0.02 mM) styrene in 48 h. In the latter culture, however, more significant accumulation of byproduct 2-phenylethanol was also observed, reaching up to 54 ± 5 mg/L (0.44 ± 0.04 mM; again, 2-phenylacetate did not accumulate above trace levels). As this again suggested that the native Ehrlich pathway was competitively inhibiting the styrene pathway with respect to precursor availability, the effect of deleting ARO10 – which converts phenylpyruvate to phenylacetaldehyde [32,43] – on styrene production was explored. As seen in Figure 5, deletion of ARO10 to preserve phenylpyruvate availability improved styrene production by 22A75D10-PAL by ~28%, reaching up to 23 ± 2 mg/L (0.22 ± 0.02 mM). Lastly, to further improve styrene production, a feedback resistant mutant of ARO4 – namely ARO4 K229L – was introduced into 22A75D10-PAL. Although ARO4 encodes a tyrosine-sensitive DAHP synthase, in previous works ARO4 K229L over-expression in S. cerevisiae was shown to increase flux through the shikimic acid pathway by as much as 4.5-fold [16,24]. Here, expressing ARO4 K229L , 22A75D104-PAL displayed an additional 25% increase in styrene titer, reaching up to 29 ± 2 mg/L (0.28 ± 0.02 mM) at a glucose yield of about 1.44 ± 0.11 mg/g (0.0025 ± 0.0002 mol/mol; or just 0.6% of theoretical). Meanwhile, trans-cinnamate was not detected in the culture media of any styrene producing strain at any time (data not shown).
Figure 5

Styrene biosynthesis from glucose by engineered S. cerevisiae strains. Styrene (gray) and 2-phenylethanol (lined) production by strains BY4741-PAL, 22A75D-PAL, 22A75D10-PAL, and 22A75D104-PAL after 48 h in SD-Leu minimal media in sealed shake flask cultures. Error bars reported at one standard deviation from triplicate experiments.

As a volatile product, it was also noted that styrene significantly accumulated within the headspace of each sealed flask. We have previously confirmed that said vapor–liquid partitioning behaves according to Henry’s Law, with a dimensionless Henry’s Law coefficient of 0.113 at 32°C [10]. For the conditions examined here, this meant that an additional ~45% styrene was produced and accumulated in each case. With this in mind, the maximum volumetric styrene production achieved here would be more accurately represented as 42 ± 3 mg/L (0.40 ± 0.03 mM) with a glucose yield of 2.09 ± 0.16 mg/g (0.0036 ± 0.0003 mol/mol; 0.9% of theoretical). The volatile nature of styrene may also prove useful as a product recovery strategy in the future [44].

Even at this adjusted output, however, achievable styrene production remains only 18% of the net production of 2-phenylethanol and 2-phenylacetate demonstrated by 22A75D (2.17 ± 0.22 mM) and is 65% of the net production of styrene and 2-phenylethanol by 22A75D-PAL (0.61 ± 0.02 mM). This suggests that multiple limiting factors may have arisen during strain construction. Although it was not anticipated to be problematic at such low aqueous titers, to ensure that end-product inhibition was not the central productivity-limiting factor, a cursory evaluation of styrene toxicity was performed. S. cerevisiae growth and viability was found to be only minimally disrupted in the presence of styrene at up to at least 200 mg/L (1.92 mM; data not shown), suggesting that styrene toxicity was not a critical barrier at this point. Over-expression of PAL2 appears to have had the greatest negative impact, reducing net aromatic production by 72%. This was likely due to the metabolic burden imposed by its expression from a high-copy plasmid. While decreasing PAL2 expression will likely reduce burden, as trans-cinnamate was never detected in styrene producing cultures, this already points to the fact that PAL2 activity was rate-limiting in the styrene pathway. Thus, future and careful optimization of PAL2 expression and/or the identification of other PAL homologs displaying greater inherent activity in S. cerevisiae will be key to achieving further improvements in styrene production.

Lastly, although seemingly low, it should be appreciated that level of styrene production demonstrated here agrees well with that of prior reports by others whom have engineered S. cerevisiae to produce aromatic chemicals (for which Curran et al. previously provided a comprehensive examination). Notable examples include p-hydroxycinnamate [45], p-aminobenzoic acid [46], p-hydroxybenzoic acid [46], and vanillin [20], which have been produced to maximal titers of up to 33.3, 34.3, 89.8, and 45 mg/L, respectively, at yields of 1.7, 2.3, 6.0, and 2.3 mg/g. Meanwhile, when compared with the baseline for styrene production established using an E. coli platform [10], the achievable titers and yields demonstrated with S. cerevisiae currently lag by about 9- and 21-fold, respectively. To achieve higher styrene titers with S. cerevisiae, further de-regulation of metabolite flux through its phenylalanine biosynthesis pathway will ultimately be required.

Conclusions

By coupling the classical approach of metabolic evolution with systematic strain and pathway engineering, styrene bioproduction directly from glucose by S. cerevisiae has been demonstrated for the first time. While providing an important demonstration of concept, future strain engineering efforts will be required to ultimately achieve economical production levels.

Materials and methods

Strains and media

All strains and plasmids used in this study are listed in Table 2. Custom oligonucleotide primers were synthesized by Integrated DNA Technologies (Coralville, IA) and are provided in Additional file 1: Table S1. All S. cerevisiae strains were purchased from Thermo Scientific (Waltham, MA). Yeast plasmids used were derived from the Gateway vector collection and purchased from AddGene (Cambridge, MA). Genomic DNA was prepared from S. cerevisiae whole cells using the ZR Fungal/Bacterial DNA MiniPrep (Zymo Research, Irvine, CA) according to vendor protocols. E. coli strain NEB10β (New England Biolabs, Ipswich, MA) was used for routine cloning and plasmid propagation, except for pDONR221 which was propagated in One Shot ccdB Survival 2 T1 E. coli (Life Technologies, Grand Island, NY). E. coli strains were routinely cultured at 37°C in Luria-Bertani (LB) broth supplemented with appropriate antibiotics, as necessary. Yeast strains were routinely cultured at 32°C in Yeast Extract Peptone Dextrose (YPD) medium, yeast synthetic dextrose (SD) medium, or yeast synthetic minimal (SD-Leu) medium. YPD medium was composed of 10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose. SD medium was composed of 6.7 g/L yeast nitrogen base, 20 g/L glucose, and 20 mg/L of each uracil, histidine, leucine, and methionine. SD-Leu medium was composed of 6.7 g/L yeast nitrogen base, 20 g/L glucose, 20 mg/L each uracil, histidine, and methionine.
Table 2

List of strains and plasmids engineered and/or used in this study

Strain

Genotype

Source

E. coli

  

NEB10β

araD139 ∆(ara,leu)7697 fhuA lacX74 galK16 galE15 mcrA f80d(lacZ∆M15)recA1 relA1 endA1 nupG rpsL rph spoT1∆(mrr-hsdRMS-mcrBC)

New England Biolabs

One Shot ccdB Survival 2 T1

F mcrA Δ(mrr-hsdRMS-mcrBC) Φ80lacZΔM15 ΔlacX74 recA1 araΔ139 Δ(ara-leu)7697galU galK rpsL (StrR) endA1 nupG fhuA::IS2

Life Technologies

S. cerevisiae

  

BY4741

MATa his3Δ0 leu2Δ0 met15Δ0 ura3Δ0

Thermo Scientific

BY4741ΔFDC1 (YDR539W)

MATa his3Δ0 leu2Δ0 met15Δ0 ura3Δ0 fdc1Δ

Thermo Scientific

22A

phenylalanine overproducer evolved from BY4741

This Study

22A75D

phenylalanine overproducer evolved from 22A

This Study

BY4741-PAL

MATa his3Δ0 leu2Δ0 met15Δ0 ura3Δ0 p425GPDPAL

This Study

BY4741ΔFDC1-PAL

MATa his3Δ0 leu2Δ0 met15Δ0 ura3Δ0 fdc1Δ p425GPDPAL

This Study

22A75D-PAL

22A75D harboring p425GPDPAL

This Study

22A75D10-PAL

22A75D aro10Δ harboring p425GPDPAL

This Study

22A75D104-PAL

22A75D aro10Δ::aro4 K229L harboring p425GPDPAL

This Study

Plasmids

  

pTrc99A

P trc , pBR322 ori, lacI q , Amp R

Prather Lab, MIT

pFA6-KanMX

KanR2, pBR322 ori, Amp R

AddGene

pDONR221

attP1-ccdB-Cm R -attP2 cassette, pUC ori, Kan R

Life Technologies

pDONR-PAL

PAL2 from A. thaliana inserted into pDONR221

This study

p425GPD

P GPD , attR1-ccdB-Cm R -attR2 cassette, pBR322 ori, LEU2, Amp R

AddGene

p425GPDPAL

PAL2 from A. thaliana inserted into p425GPD

This study

pACYCDuet-1

P T7lac , p15A ori, lacI q , Cm R

Novagen

pACYC-ARO4K229L-KanMX

pACYCDuet-1 with the integration cassette aro4 K229L -KanMX

This Study

pUN15-PAL2

Clone U12256 containing AT3G53260 (PAL2) from A. thaliana

ABRC

Evolution of phenylalanine over-producing mutants

Evolution of a phenylalanine over-producing phenotype in S. cerevisiae was achieved through random mutagenesis and high-throughput selection. The phenylalanine anti-metabolite m-fluoro-DL-phenylalanine provided selective pressure. S. cerevisiae BY4741 was first treated with ethylmethanesulphonate (EMS) according to standard protocols [47] before then being plated on minimal media supplemented with m-fluoro-DL-phenylalanine. In the first round of mutagenesis, selection occurred on SD media plates supplemented with either 18 or 22 mg/L m-fluoro-DL-phenylalanine. Note that the minimum inhibitory concentration of m-fluoro-DL-phenylalanine against BY4741 was initially determined as ~15 mg/L (data not shown). Mutants isolated from the first round were then subjected to a second round of mutagenesis and selection, however, in this case using 25, 50, or 75 mg/L m-fluoro-DL-phenylalanine. All isolated mutants were then cultured in SD media at 32°C for 48 h and periodically sampled and analyzed by high performance liquid chromatography (HPLC; as described below) to assess their comparative abilities with respect to producing 2-phenylethanol and 2-phenylacetate. Note that 2-phenylethanol and phenylacetate, which are endogenously produced from phenylpyruvate via ARO10 and either a native alcohol dehydrogenase (i.e., ADH1-6, and others putative dehydrogenases) or phenylacetaldehyde dehydrogenase (i.e., ALD1-6), respectively, were collectively used as surrogates to indicate enhanced flux through the phenylalanine biosynthesis pathway due to the fact that phenylalanine is not naturally exported by yeast to the extracellular media [24]. Several key pathway and regulatory genes (specifically, ARO3, ARO4, ARO7, ARO8, GCN4, and PHA2) were sequenced in each of BY4741 (wild type control) and top evolved mutants. In all cases, both the entire coding regions as well as 500 bp upstream of the start codon were sequenced.

Transcriptional analysis of phenylalanine over-producing mutants

Relative transcription levels of each of ARO1, ARO2, ARO3, ARO4, ARO7, ARO8, ARO9, and PHA2 were quantified at mid-log phase in each of BY4741 and the evolved mutants 22A and 22A75D. Approximately 1.5 × 108 cells of each of strain were collected by centrifugation at 17,000 x g for 1 min. The supernatant was discarded and RNA was extracted from the cell pellet using the YeaStar RNA Extraction Kit (Zymo Research, Irvine, CA). cDNA was synthesized using the SuperScript VILO cDNA Synthesis Kit (Life Technologies) whereas RT-qPCR was performed using SYBR Green (Life Technologies) based quantitative PCR according to manufacturer’s protocols. Custom oligonucleotide primers for RT-qPCR experiments, including those for the reference housekeeping gene 26S [48], were designed and synthesized, the sequences of which are provided in Additional file 1: Table S2. RT-qPCR was performed on an Applied Biosystems StepOne Real-Time PCR (Applied Biosystems) using a 60°C annealing temperature. Data analysis was performed using StepOne software with relative transcriptional levels calculated via the 2-ΔΔCt method [49].

Investigating native induction and activity of FDC1

A seed culture of S. cerevisiae BY4741 were prepared in 5 mL YPD broth at 32°C while shaking at 250 rpm overnight. The same seed (1 mL) was then used to inoculate a series of sealed shake flasks (250 mL) each containing 50 ml SD media. Cultures were grown until reaching an OD600 of ~0.6, at which point either trans-cinnamate, ferulate, p-coumarate, or phenylalanine were added to a final concentration of 0.2 mM. All cultures were incubated for an additional 12 h after which an equal number of cells (~7.4x107) were collected by centrifugation at 2,800 x g for 4 min. The cell pellet was lysed using Zymolyase (Zymo Research) before the sample was then centrifuged at 11,000 x g for 2 min to separate and collect the supernatant. Each sample was then assayed for its ability to catalyze the decarboxylation of trans-cinnamate to styrene by adding 5 μL of crude cell lysate to a 1 mL solution of 50 mM Tris–HCl buffer (pH 7.5) initially containing 250 mM trans-cinnamate. All samples were incubated at room temperature with the subsequent accumulation of styrene in the reaction mixture then followed at 247 nm for a total of 5 min at 20 s intervals using a UV/Vis spectrophotometer. A molar extinction coefficient of 10,000 M−1 cm−1 [50] and a 1 cm path length were used to establish activity in terms of mU mg−1 total protein. Total protein content in crude lysates was determined by Bradford Assay using bovine serum albumin (BSA) as an external standard.

Cloning of PAL2 from A. thaliana

The PAL2 encoding gene from A. thaliana was derived from a cDNA library plasmid containing the specific loci of interest (AT3G53260) and obtained from the Arabidopsis Biological Research Center (ABRC; Ohio State University, Columbus, OH). PAL2 was PCR amplified using Phusion DNA Polymerase (Finnzymes, Espoo, Finland) and custom oligonucleotide primers (Additional file 1: Table S1). Using Gateway Cloning Technology [19], amplified linear DNA fragments flanked with attB sequences were purified using the Zyppy Clean and Concentrator kit (Zymo Research). The BP reaction between the DNA fragment and pDONR221 (Life Technologies) was performed using Gateway BP Clonase II Enzyme Mix (Life Technologies) following manufacturer’s protocols. Transformations were performed using chemically competent NEB10β with transformants being selected by plating overnight on LB solid agar containing 34 mg/L kanamycin. The resultant donor plasmid, pDONR-PAL, was mixed with the desired destination plasmid, p425GPD, using the Gateway LR Clonase II Enzyme Mix (Life Technologies), with transformation and selection as previously performed. As listed in Table 2, this approach resulted in construction of p425GPDPAL, a high copy (2 μ) plasmid for the constitutive expression of PAL2 in S. cerevisiae.

Assaying the extracellular transport potential of trans-cinnamate

S. cerevisiae BY4741 and the FDC1 deletion mutant BY4741ΔFDC1 were each transformed with plasmid p425GPDPAL, resulting in strains BY4741-PAL and BY4741ΔFDC1-PAL, respectively. All yeast transformations were performed by lithium acetate method [51]. Both strains were then cultured in SD-Leu media. Upon reaching OD600 of ~0.6, 200 mg/L phenylalanine was then added to each culture, after which the extracellular accumulation of each of trans-cinnamate, styrene, and 2-phenylethanol were then periodically monitored via HPLC for a total period of 24 h.

Chromosomal disruption of ARO10 and integration of ARO4K229L

Targeted chromosomal disruption of ARO10 in strain 22A75D was performed via homologous recombination. Gene disruption cassettes were generated via PCR to contain 40 base pairs of homology on both sides of the targeted integration site (i.e., ARO10), in addition to the KanMX selectable marker (as obtained from pFA6-KanMX4). Following transformation, colonies were selected on YPD solid agar plates containing 200 mg/L G418. Clones successfully carrying the ARO10 disruption cassette were further confirmed by colony PCR. This resulted in the strain 22A75D10. In addition, a copy of the feedback resistant mutant ARO4 K229L , whose expression was driven by the native ARO4 promoter, was likewise integrated into 22A75D at the ARO10 locus, thereby also and simultaneously resulting in the chromosomal disruption of ARO10. In this case, however, the gene disruption cassette was first constructed in the E. coli expression vector pACYCDuet-1. The ARO4 point mutation (K229L) was generated via overlap extension using primers as listed in Additional file 1: Table S1. The resultant fragment containing ARO4 K229L was inserted between the BamHI and EcoRI sites of pACYCDuet-1, before the desired mutation was then confirmed by sequencing. Subsequently, the KanMX selectable marker with its promoter was PCR amplified from pFA6-KanMX4 before then being inserted downstream of ARO4 K229L to generate the plasmid pACYC-ARO4K229L-KanMX. The entire cassette was then PCR amplified using primers whose overhangs each contained 40 base pairs of homology to ARO10. The resultant fragment was then transformed into 22A74D, followed by selection on YPD solid agar plates containing 200 mg/L G418. Successful clones were further confirmed by colony PCR, resulting in strain 22A75D104.

Styrene production from glucose by S. cerevisiae

S. cerevisiae BY4741, 22A75D, 22A75D10, and 22A75D104 were each individually transformed with the plasmid p425GPDPAL. The resultant strains (BY4741-PAL, 22A75D-PAL, 22A75D10-PAL, and 22A75D104-PAL, respectively) were then each grown in 5 mL SD-Leu broth for 12 h at 32°C while shaking at 250 rpm to prepare seed cultures. Each seed (1 mL) was then used to inoculate 50 mL SD-Leu media in a 250 ml shake flask sealed with a glass cap. A closed system with large headspace was used to avoid volatile product (i.e., styrene) losses while also precluding the exhaustion of oxygen. Culturing continued for 48 h with periodic sampling performed to determine cell growth, residual substrate levels, and metabolite production.

HPLC analysis

Substrate and metabolite concentrations were determined via HPLC analysis. Samples were prepared by first removing 1 mL of culture from shake flasks and pelleting the cells at 11,000 x g for 2 min. The supernatant (0.75 mL) was then transferred to a glass HPLC vial which was then sealed with a Teflon-lined screw cap. Separation and analysis was performed using a Hewlett Packard 1100 series HPLC system (Palo Alto, CA) equipped with an auto sampler, diode array (UV/Vis) detector, and reverse-phase Hypersil Gold SBC18 column (4.6 mm x 150 mm; Thermo Fisher, USA), according to previously described methods [10]. The eluent was monitored at 215 nm for phenylalanine, 2-phenylethanol, and 2-phenylacetate, and at 258 nm for trans-cinnamate and styrene. Using the same HPLC system, glucose analysis was performed using an RID detector and an anion exchange column (Aminex HPX-87H; BioRAD, Hercules, CA). The column was eluted with 0.005 M H2SO4 at a constant flow rate of 0.8 ml/min.

Declarations

Acknowledgements

We thank Dr. Bo Wang (ASU) for his technical assistance with RT-qPCR experiments.

Authors’ Affiliations

(1)
Chemical Engineering, School for Engineering of Matter, Transport, and Energy, Arizona State University

References

  1. Wu C, Koylinski T, Bozik J: Preparation of styrene from ethylbenzene US Patent 4,255,599. ; 1981.Google Scholar
  2. Mimura N, Saito M: Dehydrogenation of ethylbenzene to styrene over Fe 2 O 3 /Al 2 O 3 catalysts in the presence of carbon dioxide. Catal Today 2000, 55:173–178.View ArticleGoogle Scholar
  3. James DH, Castor WM: Styrene. In Ullmann's Encyclopedia of Industrial Chemistry, Volume Volume 34. 7th edition. Edited by Arpe HJ, Biekert E, Davis HT, Gerhartz W. Weinheim: Wiley-VCH; 2011.Google Scholar
  4. SRI: Styrene. Rockville: Access Intelligence LLC Inc; 2010.Google Scholar
  5. United States Department of Energy, Office of Energy Efficiency and Renewable Energy, National Renewable Energy Laboratory: Steam system opportunity assessment for the pulp and paper, chemical manufacturing, and petroleum refining industries. 2002.Google Scholar
  6. Adkins J, Pugh S, McKenna R, Nielsen DR: Engineering microbial chemical factories to produce renewable "biomonomers". Front Microbiol 2012, 3:313.View ArticleGoogle Scholar
  7. Curran KA, Alper HS: Expanding the chemical palate of cells by combining systems biology and metabolic engineering. Metab Eng 2012, 14:289–297.View ArticleGoogle Scholar
  8. EB N, Winters P: Perspective on opportunities in industrial biotechnology in renewable chemicals. Biotechnol J 2012, 7:176–185.View ArticleGoogle Scholar
  9. Lee JW, Kim HU, Choi S, Yi J, Lee SY: Microbial production of building block chemicals and polymers. Curr Opin Biotechnol 2011, 22:758–767.View ArticleGoogle Scholar
  10. McKenna R, Nielsen DR: Styrene biosynthesis from glucose by engineered E. coli . Metab Eng 2011, 13:544–554.View ArticleGoogle Scholar
  11. Mukai N, Masaki K, Fujii T, Kawamukai M, Iefuji H: PAD1 and FDC1 are essential for the decarboxylation of phenylacrylic acids in Saccharomyces cerevisiae . J Biosci Bioeng 2010, 109:564–569.View ArticleGoogle Scholar
  12. Claypool J, Raman DR, Jarboe LJ, Nielsen DR: Technoeconomic evaluation of bio-based styrene production by engineered Escherichia coli . J Ind Microbiol Biotechnol 2014 DOI:10.1007/s10295-014-1469-5.Google Scholar
  13. Lo TM, Teo WS, Ling H, Chen BB, Kang A, Chang MW: Microbial engineering strategies to improve cell viability for biochemical production. Biotechnol Adv 2013, 31:903–914.View ArticleGoogle Scholar
  14. Demain AL, Vaishnav P: Production of recombinant proteins by microbes and higher organisms. Biotechnol Adv 2009, 27:297–306.View ArticleGoogle Scholar
  15. Sudbery PE: The expression of recombinant proteins in yeasts. Curr Opin Biotechnol 1996, 7:517–524.View ArticleGoogle Scholar
  16. Curran KA, Leavitt JM, Karim AS, Alper HS: Metabolic engineering of muconic acid production in Saccharomyces cerevisiae . Metab Eng 2013, 15:55–66.View ArticleGoogle Scholar
  17. Nevoigt E: Progress in metabolic engineering of Saccharomyces cerevisiae . Microbiol Molec Biol Rev 2008, 72:379–412.View ArticleGoogle Scholar
  18. Ostergaard S, Olsson L, Nielsen J: Metabolic engineering of Saccharomyces cerevisiae . Microbiol Molec Biol Rev 2000, 64:34–50.View ArticleGoogle Scholar
  19. Alberti S, Gitler AD, Lindquist S: A suite of Gateway® cloning vectors for high-throughput genetic analysis in Saccharomyces cerevisiae . Yeast 2007, 24:913–919.View ArticleGoogle Scholar
  20. Hansen EH, Moller BL, Kock GR, Bunner CM, Kristensen C, Jensen OR, Okkels FT, Olsen CE, Motawia MS, Hansen J: De novo biosynthesis of vanillin in fission yeast ( Schizosaccharomyces pombe ) and baker's yeast ( Saccharomyces cerevisiae ). Appl Environ Microbiol 2009, 75:2765–2774.View ArticleGoogle Scholar
  21. Jiang H, Wood KV, Morgan JA: Metabolic engineering of the phenylpropanoid pathway in Saccharomyces cerevisiae . Appl Environ Microbiol 2005, 71:2962–2969.View ArticleGoogle Scholar
  22. Stark D, Munch T, Sonnleitner B, Marison IW, von Stockar U: Extractive bioconversion of 2-phenylethanol from L-phenylalanine by Saccharomyces cerevisiae . Biotechnol Prog 2002, 18:514–523.View ArticleGoogle Scholar
  23. Koopman F, Beekwilder J, Crimi B, van Houwelingen A, Hall RD, Bosch D, van Maris AJA, Pronk JT, Daran JM: De novo production of the flavonoid naringenin in engineered Saccharomyces cerevisiae . Microb Cell Fact 2012, 11:155–169.View ArticleGoogle Scholar
  24. Luttik MAH, Vuralhan Z, Suir E, Braus GH, Pronk JT, Daran JM: Alleviation of feedback inhibition in Saccharomyces cerevisiae aromatic amino acid biosynthesis: Quantification of metabolic impact. Metab Eng 2008, 10:141–153.View ArticleGoogle Scholar
  25. Nijkamp K, van Luijk N, de Bont JA, Wery J: The solvent-tolerant Pseudomonas putida S12 as host for the production of cinnamic acid from glucose. Appl Microbiol Biotechnol 2005, 69:170–177.View ArticleGoogle Scholar
  26. Nijkamp K, Westerhof RG, Ballerstedt H, de Bont JA, Wery J: Optimization of the solvent-tolerant Pseudomonas putida S12 as host for the production of p-coumarate from glucose. Appl Microbiol Biotechnol 2007, 74:617–624.View ArticleGoogle Scholar
  27. Verhoef S, Wierckx N, Westerhof RG, de Winde JH, Ruijssenaars HJ: Bioproduction of p-hydroxystyrene from glucose by the solvent-tolerant bacterium Pseudomonas putida S12 in a two-phase water-decanol fermentation. Appl Environ Microbiol 2009, 75:931–936.View ArticleGoogle Scholar
  28. Wierckx NJ, Ballerstedt H, de Bont JA, Wery J: Engineering of solvent-tolerant Pseudomonas putida S12 for bioproduction of phenol from glucose. Appl Environ Microbiol 2005, 71:8221–8227.View ArticleGoogle Scholar
  29. Drake RH: The biochemistry of mutagenesis. Annu Rev Microbiol 1976, 45:11–37.Google Scholar
  30. Wloch DM, Szafraniec K, Borts RH, Korona R: Direct estimate of the mutation rate and the distribution of fitness effects in the yeast Saccharomyces cerevisiae . Genetics 2001, 159:441–452.Google Scholar
  31. Vuralhan Z, Morais MA, Tai SL, Piper MD, Pronk JT: Identification and characterization of phenylpyruvate decarboxylase genes in Saccharomyces cerevisiae . Appl Environ Microbiol 2003, 69:4534–4541.View ArticleGoogle Scholar
  32. Hazelwood LA, Daran JM, van Maris AJ, Pronk JT, Dickinson JR: The Ehrlich pathway for fusel alcohol production: a century of research on Saccharomyces cerevisiae metabolism. Appl Environ Microbiol 2008, 74:2259–2266.View ArticleGoogle Scholar
  33. Caspi R, Foerster H, Fulcher CA, Kaipa P, Krummenacker M, Latendresse M, Paley S, Rhee SY, Shearer AG, Tissier C, Walk TC, Zhang P, Karp PD, Caspi R, Foerster H, Fulcher CA, Hopkinson R, Ingraham J, Kaipa P, Krummenacker M, Paley S, Pick J, Rhee SY, Tisse C, Zhang P, Karp PD: MetaCyc: a multiorganism database of metabolic pathways and enzymes. Nucleic Acids Res 2006, 34:D511–D516.View ArticleGoogle Scholar
  34. Fukuda K, Watanabe M, Asano K, Ouchi K, Takasawa S: A mutated ARO4 gene for feedback-resistant DAHP synthase which causes both o-fluoro-dl-phenylalamine resistance and β-phenethyl-alcohol overproduction in Saccharomyces cerevisiae . Curr Genet 1991, 20:453–456.View ArticleGoogle Scholar
  35. Teshiba S, Furter R, Niederberger P, Braus G, Paravicini G, Hütter R: Cloning of the ARO3 gene of Saccharomyces cerevisiae and its regulation. Molec Gen Genet 1986, 205:353–357.View ArticleGoogle Scholar
  36. Schmidheini T, Sperisen P, Paravicini G, Hütter R, Braus G: A single point mutation results in a constitutively activated and feedback-resistant chorismate mutase of Saccharomyces cerevisiae . J Bacteriol 1989, 171:1245–1253.Google Scholar
  37. Kradolfer P, Zeyer J, Miozzari G, Huetter R: Dominant regulatory mutants in chorismate mutase of Saccharomyces cerevisiae . FEMS Microbiol Lett 1977, 2:211–216.View ArticleGoogle Scholar
  38. Weber C, Bruckner C, Weinreb S, Lehr C, Essl C, Boles E: Biosynthesis of cis, cis-muconic acid and its aromatic precursors, catechol and protocatechuic acid, from renewable feedstocks by Saccharomyces cerevisiae. Appl Environ Microbiol 2012, 78:8421–8430.View ArticleGoogle Scholar
  39. Dickinson JR, Salgado LE, Hewlins MJ: The catabolism of amino acids to long chain and complex alcohols in Saccharomyces cerevisiae . J Biol Chem 2003, 278:8028–8034.View ArticleGoogle Scholar
  40. Iraqui I, Vissers S, Cartiaux M, Urrestarazu A: Characterisation of Saccharomyces cerevisiae ARO8 and ARO9 genes encoding aromatic aminotransferases I and II reveals a new aminotransferase subfamily. Molec Gen Genet 1998, 257:238–248.View ArticleGoogle Scholar
  41. Hinnebusch AG: Translational regulation of gcn4 and the general amino Acid control of yeast. Annu Rev Microbiol 2005, 59:407–450.View ArticleGoogle Scholar
  42. Cherry JM, Hong EL, Amundsen C, Balakrishnan R, Binkley G, Chan ET, Christie KR, Costanzo MC, Dwight SS, Engel SR: Saccharomyces Genome Database: the genomics resource of budding yeast. Nucleic Acids Res 2012, 40:D700–D705.View ArticleGoogle Scholar
  43. Caspi R, Foerster H, Fulcher CA, Kaipa P, Krummenacker M, Latendresse M, Paley S, Rhee SY, Shearer AG, Tissier C, Walk TC, Zhang P, Karp PD: The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of pathway/genome databases. Nucleic Acids Res 2008, 36:D623–D631.View ArticleGoogle Scholar
  44. McKenna R, Moya L, McDaniel M, Nielsen DR: Comparing in situ removal strategies for improving styrene bioproduction. Bioproc Biosys Eng 2014. DOI:10.1007/s00449-014-1255-9.Google Scholar
  45. Vannelli T, Qi WW, Sweigard J, Gatenby AA, Sariaslani FS: Production of p-hydroxycinnamic acid from glucose in Saccharomyces cerevisiae and Escherichia coli by expression of heterologous genes from plants and fungi. Metab Eng 2007, 9:142–151.View ArticleGoogle Scholar
  46. Kromer JO, Nunez-Bernal D, Averesch NJH, Hampe J, Varela J, Varela C: Production of aromatics in Saccharomyces cerevisiae - a feasibility study. J Biotechnol 2013, 163:184–193.View ArticleGoogle Scholar
  47. Winston F: EMS and UV mutagenesis in yeast. In Current Protocols in Molecular Biology: John Wiley & Sons, Inc.; 2001.Google Scholar
  48. Martorell P, Querol A, Fernández-Espinar MT: Rapid identification and enumeration of Saccharomyces cerevisiae cells in wine by real-time PCR. Appl Environ Microbiol 2005, 71:6823–6830.View ArticleGoogle Scholar
  49. Livak KJ, Schmittgen TD: Analysis of relative gene expression data using real-time quantitative PCR and the 2(−Delta Delta C(T)) Method. Methods 2001, 25:402–408.View ArticleGoogle Scholar
  50. Murray MJ, Gallaway WS: Steric inhibition of resonance in 1-methyl-2-isopropenylbenzene. J Am Chem Soc 1948, 70:3867–3868.View ArticleGoogle Scholar
  51. Gietz RD, Woods RA: Transformation of yeast by lithium acetate/single-stranded carrier DNA/polyethylene glycol method. Methods Enzymol 2002, 350:87–96.View ArticleGoogle Scholar

Copyright

© McKenna et al.; licensee BioMed Central 2014

This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.