Research Article pubs.acs.org/synthbio
Engineering an NADPH/NADP+ Redox Biosensor in Yeast Jie Zhang,† Nikolaus Sonnenschein,† Thomas P. B. Pihl,† Kasper R. Pedersen,† Michael K. Jensen,*,† and Jay D. Keasling†,‡,§,∥,⊥ †
The Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Kongens Lyngby, 2800, Denmark Joint BioEnergy Institute, Emeryville, California 94608, United States § Biological Systems & Engineering Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States ∥ Department of Chemical and Biomolecular Engineering and ⊥Department of Bioengineering, University of California, Berkeley, California 94709, United States ‡
S Supporting Information *
ABSTRACT: Genetically encoded biosensors have emerged as powerful tools for timely and precise in vivo evaluation of cellular metabolism. In particular, biosensors that can couple intercellular cues with downstream signaling responses are currently attracting major attention within health science and biotechnology. Still, there is a need for bioprospecting and engineering of more biosensors to enable real-time monitoring of specific cellular states and controlling downstream actuation. In this study, we report the engineering and application of a transcription factor-based NADPH/NADP+ redox biosensor in the budding yeast Saccharomyces cerevisiae. Using the biosensor, we are able to monitor the cause of oxidative stress by chemical induction, and changes in NADPH/NADP+ ratios caused by genetic manipulations. Because of the regulatory potential of the biosensor, we also show that the biosensor can actuate upon NADPH deficiency by activation of NADPH regeneration. Finally, we couple the biosensor with an expression of dosage-sensitive genes (DSGs) and thereby create a novel tunable sensor-selector useful for synthetic selection of cells with higher NADPH/NADP+ ratios from mixed cell populations. We show that the combination of exploitation and rational engineering of native signaling components is applicable for diagnosis, regulation, and selection of cellular redox states. KEYWORDS: redox, biosensor, dosage-sensitive genes, yeast cofactors and their ratios. As such, it was first discovered in the 1960s that NADH and NADPH could be detected by their endogenous fluorescence.7 However, this method generally suffers from low sensitivity and cannot discriminate NADPH and NADH. Since then, other commonly used methods include liquid chromatography−mass spectrometry (LC−MS)-based analytics8 and spectrophotometric-based enzymatic assays.9 However, because of fast turnover rates of NAD and NADP redox couples, quantification methods are often laborious, costly, and low-throughput. Also, for eukaryotic cells, none of these methods can analyze redox cofactors in different compartments separately, such as cytoplasm and mitochondrion, that are known to have separate redox environments.10 As such, the ability to measure these molecules in vivo would have major implications for how to study redox perturbations and the targeted manipulation of them. Recently, genetically encoded biosensors have attracted enormous attention.11 For monitoring redox states this group
R
edox cofactors, such as nicotinamide adenine dinucleotide (NAD) and nicotinamide adenine dinucleotide phosphate (NADP), are essential for all living cells. Apart from being a coenzyme for regulation of enzyme activity, NAD and NADP redox couples also play a role in cellular processes related to oxidative stress,1 aging,2 and cell survival.3 For this reason, it is of key importance to maintain a proper cellular redox state, which is determined by the ratios between reduced and oxidized forms of these redox cofactors. Although both NAD and NADP are important for redox homeostasis, they have very different biological roles and distinct redox states. Indeed, although no consensus values have been reported, several studies have shown that cytosolic NADH/NAD+ ratios are low, ranging from 0.001 to 0.01,4,5 while cytosolic NADPH/NADP+ ratios range from 15 to 60.6 The vast differences in these ratios make sense given their roles in cell physiology: the low NADH/NAD+ ratio allows respiration and oxidative phosphorylation to proceed efficiently, while the high NADPH/NADP+ ratio drives biosynthesis of fatty acids and nucleic acids and oxidative stress defense.1 Because of their importance for cellular homeostasis, various methods have been developed for quantification of redox © 2016 American Chemical Society
Received: May 2, 2016 Published: July 15, 2016 1546
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
Figure 1. Regulatory mechanism for Yap1p activation and translocation between cytoplasm and nucleus in yeast. Yap1p can be oxidized (Yap1pox shown as red oval) upon oxidative stress and consequently accumulates in the nucleus. Oxidized Yap1p can be reduced (Yap1pre shown as blue hexagon) by reduced thioredoxin (Trxre), the level of which is maintained by thioredoxin reductase (Trr1p) using NADPH as an electron donor.
includes engineered fluorescent proteins based on domains that are either sensitive to redox changes or that directly interact with a redox cofactor.5,12−14 Despite their real-time reporting of changing NAD(P)H/NAD(P)+ ratios, these biosensors cannot transmit the signal to a downstream actuation. As an alternative, more recent studies have harnessed native transcriptional responses to altered redox states as biosensors. In yeast this includes the simple introduction of the green fluorescent protein gene under the control of the GPD2 promoter, which is induced under conditions of increased NAPH/NAD+ ratios.15 In bacteria, another study reported an NADPH/NADP+ sensor based on the transcriptional regulator SoxR.16 Though both studies adopted transcriptional regulation as modus operandi for coupling changes in the redox states with changes in the transcriptional output from a reporter promoter, neither coupled changes in the redox states with actuation of a metabolic pathway or some other system that would correct the imbalance in redox cofactors. Though this should simply require the exchange of the reporter gene with an actuator gene, the direct transplantation of SoxR into eukaryotes is hampered by the distinct iron−sulfur protein assembly mechanisms between bacteria and eukaryotes.17 In search of more general physiological responses amenable for engineering redox biosensors, disulfide-bridge formation between cysteine residues is acknowledged as a universal response to oxidative stress. For this purpose, almost all life from bacteria to humans employs glutathione, thioredoxin, and to a lesser extent glutaredoxin, to detoxify reactive oxygen species produced during respiration or externally induced.18 At the core of this defense system is a functionally conserved family of transcriptional regulators that can sense changes in cellular redox states and activate the expression of components in the antioxidant defense system.19 In yeast, this oxidative response pathway consists of the transcription factor (TF) Yap1p, thioredoxin, glutathione, and their reductases, which use NADPH as the final electron donor.20 Both NADH/NAD+ and NADPH/NADP+ are key players in maintaining redox homeostasis; however, it is the redox state of the NADPH/ NADP+ couple that regulates the states of thioredoxin and glutathione in equilibrium with their oxidized forms. At the
molecular level, Yap1p is oxidized upon oxidative stress (OS) and this leads to the formation of intraprotein disulfide bonds.19,21 The conformational change of Yap1p masks the Cterminal nuclear export sequence (NES) recognized by the exportin Crm1p22 and consequentially amplify its nuclear accumulation mediated by the importin Pse1p.23 In the nucleus, oxidized Yap1p activates the expression of genes involved in OS defense, including the biosynthetic genes for glutathione and thioredoxin, as well as their reductases24 (Figure 1). Oxidation of Yap1p can be reversed by reduced thioredoxin (Trx-(SH)2), and the resulting oxidized thioredoxin (Trx-S2) can be directly reduced by thioredoxin reductase 1 (encoded by TRR1) using electrons from cytosolic NADPH.20 Therefore, the NADPH/ NADP+ redox state indirectly regulates the equilibrium between oxidized and reduced Yap1, and consequently its cytoplasmicnuclear distribution and transcriptional potential. In this study, we developed an NADPH/NADP+ redox sensor by exploiting the native Yap1p regulatory pathway and rational engineering of a Yap1p target promoter. Initially, using green fluorescent protein (GFP) as a reporter gene, we showed that this biosensor is useful for in vivo monitoring of the NADPH/NADP+ redox state when perturbed by either an external oxidative stimulus or endogenous genetic modifications. Furthermore, as a novel selection mechanism, we regulated the expression of dosage-sensitive genes in order to create a tunable sensor-selector that couples cellular redox state to cell growth. As a proof of principle, we applied this sensorselector to demonstrate in a mixed cell population the synthetic selection of yeast strains with improved NADPH generation capacity.
■
RESULTS
Characterization and Optimization of Sensor-Reporter Performance. To exploit oxidative stress signaling through Yap1p as a NADPH/NADP+ redox sensor, we initially characterized the induction of yeast-enhanced green fluorescent protein (yEGFP) controlled by the wild-type Yap1p target promoter TRX2 (PTRX2_wt) upon treatment of diamide, a thiol oxidant. Despite the correlation between the GFP intensity and diamide concentrations (0.25 to 2.5 mM), the dynamic range of 1547
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
Figure 2. Characterization and optimization of the sensor-reporter combination. (A) Yap1BS‑1 (TTACTAA) is responsible for Yap1-mediated oxidative stress response. An additional YRE in the TRX2 promoter (PTRX2_v1) improves the diamide inducibility of the promoter, and either knockout of YAP1 (yap1Δ) or removal of Yap1BS‑1 (PTRX2_v2) completely abolishes the response to diamide treatment. Red, blue, purple, and green bars indicate TATA box, Yap1BS‑1, Yap1BS‑2, and YRE, respectively. (B) Fusion of one to five tandem repeats of the UAS from the TRX2 promoter increases the sensor output without significantly changing the basal activity of the promoter. Error bars represent standard deviations from independent triplicate measurements.
this, we used a strategy reported by Blazeck et al.30 to create a series of synthetic hybrid PTRX2 variants by fusing tandem repeats29 of the YRE-containing the TRX2 UAS to the core of a strong promoter (PTDH3) or a weak promoter (PREV1), which span an expression range of at least 2 orders of magnitude (Supplementary Figure S3). This generated a set of promoters with various output strengths upon Yap1p oxidation (Supplementary Figure S4). Overall, promoter variants with the weak PREV1 core showed the lowest basal activities compared to those with PTRX2 and especially compared to promoters with a PTDH3 core. For comparison, some of the hybrid promoters (PTDH3_3/4/5xUAS) could even be induced to a level that surpassed the strength of the TEF1 promoter, which is one of the strongest constitutive promoters commonly used in yeast. Yap1p Can Sense Changes in NADPH Regeneration. To test if the biosensor can report changes in intracellular NADPH/NADP+ redox states caused by genetic modifications, we introduced the sensor-reporter in an oxidative stresssensitive zwf1Δ strain, which is known to have insufficient NADPH supply due to inactivation of the oxidative branch of pentose phosphate pathway (oxPPP).31,32 In line with the oxPPP being the primary source of NADPH, we observed an approximately 10-fold higher GFP intensity in the zwf1Δ strain compared to that in the wild type strain (Figure 3a), indicating that Yap1p is considerably oxidized. Moreover, GFP in the zwf1Δ strain was significantly more induced compared to that in cells treated with 2 mM diamide (Figures 2b and 3a), suggesting that deficiency in NADPH generation has a larger impact on Yap1p oxidation as compared to an externally applied oxidative stress, such as diamide. Interestingly, though concentrations of diamide above 2 mM caused a decreased growth rate in the wild type strain, this growth defect was not observed in the oxPPP-inactivated zwf1Δ strain (Supplementary Figure S5). Besides generating NADPH as the reducing equivalent for anabolism, PPP also provides pentose phosphate for the biosynthesis of nucleotides, which is a component of nucleic acids (DNA and RNA), energy molecules (e.g., ATP), and cofactors (e.g., NADH). To make sure that high induction of reporter yEGFP was not mainly due to an impaired de novo synthesis of NADPH, we investigated whether Yap1-dependent transcriptional regulation of an NADP+-dependent acetaldehyde dehydrogenase (encoded by ALD6) expressed under the
the GFP reporter was only 2-fold when treated with 2 mM diamide. Moreover, high concentrations of diamide also caused a dramatic reduction in cell growth (Supplementary Figure S1). To increase the dynamic range for better applicability of the sensor, we engineered the PTRX2 controlling GFP expression to include more Yap1p binding sites.25 Our results showed that a PTRX2 variant with an extra Yap response element (YRE) (TTACGTAA)26 (PTRX2_v1) significantly increased expression by diamide treatment compared to that from PTRX2_wt (Figure 2a). To examine the Yap1-dependent activation of PTRX2, we also removed each of the two formerly suggested Yap1p binding sites, namely, Yap1BS‑1 (TTACTAA) and Yap1BS‑2 (TTAGTAA), resulting in two PTRX2 variants PTRX2_v2 and PTRX2_v3, respectively. While PTRX2_v3 behaved similarly compared to the wild type (PTRX2_wt), PTRX2_v2 was completely irresponsive to diamide treatment (Figure 2a). This result is consistent with the previous finding that TTACTAA is the preferred Yap1p binding site,27 yet it also shows that Yap1BS‑1 is the dominant site for Yap1-mediated induction of PTRX2 during oxidative stress. For comparison, the reporter construct PTRX2_v1-yEGFP expressed in a yap1Δmutant did not respond to diamide treatment, confirming that the increased inducibility was largely caused by oxidative stress in a Yap1-dependent manner (Figure 2a), which is also consistent with a previous finding showing that Yap1p is essential for TRX2 activation.24 To further improve the dynamic range of the sensor-reporter combination, we created a series of TRX2 promoter variants by adding tandem repeats, ranging from one to five, of the 140-bp (located −275 to −136) upstream activating sequence (UAS) including the YRE from PTRX2. These fusion promoters with various numbers of UASes exhibit a similar expression pattern when cells were grown in synthetic noninducer medium; their inducibility by diamide incrementally increased with the number of UASes added, reaching 5.5-fold induction in GFP intensity with five UASes (PTRX2_5xUAS) (Figure 2b). This is consistent with a previous result that the TRX2 promoter engineered to contain three UASes gave an approximately 3fold increase in GFP intensity induced by diamide.28 To test if other oxidizing agents known to perturb NADPH/NADP+ ratios would also affect the NADPH/NADP+ redox biosensor output, we cultivated cells in minimal medium supplemented with either H2O2 or furfural.29 Here, we observed a 2.9- and 1.3-fold increase in GFP output from cells treated with H2O2 and furfural, respectively (Supplementary Figure S2). Following 1548
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
Figure 3. In vivo monitoring of the cellular redox states using the redox sensors. (A) Test of the NADPH sensor-reporter combination (PTRX2_5xUASyEGFP) in a zwf1Δ mutant strain and effect of the suppressor ALD6 expressed under various promoters. (B) Proposed model for ALD6 suppression of the NADPH deficiency in a zwf1Δ strain. (C) Test of an NADH redox sensor (PGPD2-yEGFP) in the wild type and adh1Δ strains in the absence or presence of diamide. (D) Test of the NADPH redox sensor in the wild type and adh1Δ strains. Error bars represent standard deviations from independent triplicate measurements.
green fluorescence in the adh1Δ mutant harboring PGPD2yEGFP was 2.5-fold higher than that in the wild type strain harboring PGPD2-yEGFP (Figure 3c), confirming that the NADH/NAD+ ratio is increased by ADH1 deletion. In contrary, the NADPH/NADP+ biosensor applied in the adh1Δ strain only gave a mild increase in GFP signal compared to the wild type strain (Figure 3d), suggesting that the increased redox potential of NADH/NAD+ has no major impact on the NADPH/NADP+ biosensor. Model-Guided Identification of Gene Targets to Improve NADPH Availability. NADPH is a key cofactor in both maintenance of cellular redox homeostasis and biosynthesis of fatty acids. It has previously been demonstrated that genome-scale metabolic models can be used to search for gene targets that can improve NADPH availability.35 In line with this work, we used a number of constraint-based metabolic modeling methods, such as the parsimonious flux balance analysis (pFBA),36 the linear version of the minimization of metabolic adjustment (MOMA),37 and the regulatory on off minimization (ROOM),38 to predict the steady-state fluxes for all viable single knockout strains. In particular, we compared NADPH turnover (vnadph), which is the sum over all NADPHproducing fluxes (or consuming fluxes), between each mutant and the wild-type strain (wild-type fluxes have been computed by pFBA). Subsequently, all knockout mutants were ranked after normalizing Δvnadph by the predicted growth rate for each of the knockout strains in order to remove changes in turnover that are simply due to changes in growth rate (see Methods for further details). Among the list of single mutants predicted to have an increased NADPH flux, 10 targets from three categories in the central metabolism, namely, upper glycolysis, TCA cycle, and serine/glycine metabolism (Table 1), were experimentally tested using the redox biosensor (PTRX2‑5×UASyEGFP). As a control we also included the zwf1Δ strain, which
control of PTRX2 variants would suppress the biosensor output observed in the zwf1Δ strain. Here we observed that ALD6 controlled by the TRX2 promoter lacking Yap1BS‑1 (PTRX2_v2) could only mildly suppress zwf1Δ, whereas the oxidative stressinducible promoter PTRX2_5xUAS substantially restored the NADPH deficiency in the zwf1Δ strain (Figure 3a). For comparison, when controlling the expression of ALD6 using the strong TEF1 promoter in the zwf1Δ strain background the biosensor output was comparable to the biosensor output observed in the wild type background (Figure 3a). Taken together, these results suggest that overexpression of ALD6 complements the NADPH deficiency in a zwf1Δ strain, presumably through higher NADPH generation coupled with acetate production catalyzed by Ald6p. Furthermore, because of the autoregulatory cycle in which the output of PTRX2_5xUAS decreases when ALD6 is induced, the GFP level in this strain was still slightly higher than that in the strain where ALD6 was constitutively overexpressed under the TEF1 promoter (Figure 3b). To validate the NADPH/NADP+ redox biosensor output in the genetic variants, we quenched cells and quantified the NADPH/NADP+ ratio using a calorimetric assay. In line with our biosensor output, we observed the zwf1Δ strain to have lowered the NADPH/NADP+ ratio compared to the wild type strain (Supplementary Figure S6), which is consistent with earlier finding that zwf1Δ leads to a lower NADPH.31,32 However, the calorimetric assay did not allow us to discriminate the NADPH/NADP+ ratios between the wild type strain and the strain expressing ALD6 in the zwf1Δ strain background (Supplementary Figure S6). To test the sensor specificity, we applied our NADPH redox sensor (PTRX2_5xUAS-yEGFP) as well as a previously reported NADH biosensor in yeast (PGPD2-yEGFP) in an adh1Δ mutant strain that is known to have elevated NADH levels.33,34 The 1549
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
PGI1), suggesting that the NADPH/NADP+ ratios in these strains were higher than the wild type strain (Figure 4c). Interestingly, although the model can only predict that steadystate fluxes and changes of NADPH turnover (vnadph) are used as a proxy for changes in the actual NADPH/NADP+ ratio, these results generally correlate very well with the trends in our experimental measurements. Construction of a Novel Sensor-Selector Combination. Biosensor-reporter combinations can be useful for realtime, and even high throughput, screening of large strain libraries that are often generated for functional genomics studies or metabolic pathway engineering.39 However, as the techniques for genome editing and library construction advance rapidly, as well as the price for de novo DNA synthesis continuously drops, the size of strain libraries may exceed available resolution and screening capacity (i.e., FACS).40 This challenge can be tackled using a sensor−selector combination, in which the sensor output controls the expression of the selector gene that can be coupled to the cell growth.41 For example, when using an antibiotic resistant gene as the selector controlled by a sensor that responds to a phenotypic improvement, only strains that harbor beneficial mutations can grow (or grow relatively faster) in the presence of antibiotic. Subsequently, the fraction of these mutant strains in the whole population will increase continuously. However, since our redox sensor is activated by a decreased NADPH/ NADP+ ratio, and it is often the opposite (i.e., a higher NADPH/NADP+ ratio) that is desirable, it requires a selector gene whose expression leads to slower cell growth. For this reason, we decided to exploit a novel engineering strategy using the so-called dosage-sensitive genes (DSGs) as endogenous selectors for the screening of strains with altered NADPH generation capacity. The transcription of these DSGs is so tightly regulated that cell growth is hampered by even a small increase in their native expression level,42 which should make them suitable as selectors when coupled with the redox sensor. To identify the best DSG candidate and its critical expression level, we first characterized the growth rates of strains expressing three previously reported DSGs, namely ACT1, CDC14, and TPK2,42 under the control of promoters with various strengths spanning at least 2 orders of magnitude.43 Here, we identified the critical expression level for these three DSGs that would cause a minor reduction in growth rate (0.29−0.31 h−1) as compared to the wild type (0.34 h−1), and then further increased their expression level beyond the critical level which led to a substantial decrease in growth rates (Figure 5a). These critical expression levels depend on the native abundance of the DSGs, suggesting that it is the relative change in protein abundance that regulates cell growth. Among the DSGs tested, we chose to proceed with the sensor-selector combination CDC14 controlled by PTRX2‑5xUAS because (i) this promoter variant had the highest dynamic range upon oxidative stress (Figure 2b) and (ii) this sensor-selector combination was significantly more sensitive than other combinations to oxidative stress induced by diamide (Figure 5b). As a proof-of-concept for the sensor−selector design, we first created three strains: wild type, zwf1Δ, and a PTEF1-ZWF1 overexpression strain, which contained mKate2, CFP, and yEGFP (each under control of PTEF1) in the genome, respectively. Next, we used FACS to screen the abundance of the individual fluorophores during a three-strain cocultivation running through two passages over 48 h (Figure 6a). In the control group without the selector, the percentage of the zwf1Δ
Table 1. Genome-Scale Metabolic Model (GSMM) Prediction for Single Knockout Mutant Strainsa gene knockout
method
growth rate (h−1)
normalized ΔvNADPH
PGI1
MOMA pFBA ROOM pFBA pFBA MOMA ROOM MOMA ROOM MOMA ROOM MOMA ROOM MOMA ROOM MOMA ROOM MOMA pFBA ROOM MOMA pFBA ROOM
0.1488 0.2230 0.1316 0.0718 0.0718 0.2796 0.2782 0.2699 0.2783 0.2711 0.2782 0.2796 0.2782 0.2628 0.2780 0.2753 0.2759 0.2775 0.2856 0.2782 0.2710 0.2857 0.2813
3.6780 4.2964 4.1242 2.3444 2.3434 0.0307 0.3507 0.0733 0.2861 0.0348 0.3643 0.0307 0.3583 0.0561 0.7528 0.0082 0.1948 0.1930 0.1439 0.2731 −0.7967 −0.0956 −0.0316
TPI1 PFK1 LSC1 SDH2 FUM1 KGD2 LPD1 SHM2 SER1
ZWF1
a
Genome-scale model iMM904 was used for simulation to predict changes in the NADPH/NADP+ redox state in all possible (viable) single gene knockouts compared to a wild-type reference. The flux distributions for these single mutant strains were predicted using three different methods, namely parsimonious flux balance analysis (pFBA), minimization of metabolic adjustment (MOMA), and regulatory on off minimization (ROOM), which are based on distinct assumptions on how the cells will respond to the genetic modifications. For more information, see Experimental procedures.
was predicted by the model to have a substantially decreased total NADPH flux. Apart from PGI1, all nine other knockout strains were viable. Introduction of the redox biosensor into the knockout strains showed that the pf k1Δ and tpi1Δ strains had approximately 2fold lower GFP levels, suggesting that the NADPH/NADP+ ratio is increased in these mutant strains (Figure 4a). In addition to pf k1Δ and tpi1Δ, all knockout mutants of genes encoding TCA cycle enzymes showed little effect on the biosensor output, except lsc1Δ, which had a slightly lower GFP level compared to the wild type strain. Deletion of genes involved in serine/glycine metabolism had a 1.2- to 2-fold increase in GFP levels, indicating that these knockouts had a negative impact on the NADPH/NADP+ balance (Figure 4a). However, deletion of PFK1 or TPI1 led to a substantially slower growth as compared to the wild type strain (Figure 4b). To examine if the lower GFP level was due to slower growth and insufficient protein synthesis, we knocked down PFK1 and TPI1 as well as PGI1 (pgi1Δis inviable on glucose or fructose) by replacing their native promoters with the REV1, RNR2, and PFK1 promoters, respectively. The strength of these promoters was estimated on the basis of the yEGFP reporter data (Supplementary Figure S3). As compared to the wild type strain, these knockdown mutants only showed a slightly decreased growth (Figure 4b). Yet, all knockdown mutant strains gave a lower GFP signal, although to a lesser extent compared to the corresponding knockout strains (except for 1550
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
Figure 4. FBA model-guided identification of knockout targets for improved NADPH generation capacity. (A) FBA model prediction of total NADPH flux (left) and experimental measurements of NADPH redox sensor-reporter combination (PTRX2_5xUAS- yEGFP): (∗) p-value < 0.005; (∗∗) p-value < 0.0005. (B) Effect of PFK1, PGI1, and TPI1 knockout or knockdown on cell growth. (C) Biosensor output for knockout and knockdown mutants compared to the wild type strain. Data for pgi1Δ were not available since this mutant strain could not grow on glucose as the carbon source. Error bars represent standard deviations from independent triplicate measurements.
Figure 5. Characterization of DSGs on cell growth. (A) Growth rate affected by DSGs expressed under promoters of various strengths. (B) Growth of strains with various sensor−selector constructs in Delft medium with diamide concentration. Error bars represent standard deviations from independent triplicate measurements.
strain in the population decreased to 17% of the culture after 1 day and further decreased to 10% after 2 days, which can be attributed to its slightly slower growth rate as compared to the wild type strain (Supplementary Figure S5). In the cocultivation of strains with the sensor-selector integrated, the fraction of zwf1Δ cells almost disappeared within 24 h
(approximately 1% of the total population) and was not detectable after 2 days (Figure 6b). We also noticed that, in the control group, the ZWF1 overexpression strain was a smaller fraction of the population than the wild type strain (34% vs 49% after 24 h and 39% vs 52% after 2 days, respectively), presumably due to slower growth of the ZWF1 overexpression 1551
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
Figure 6. Selection of strains with improved NADPH generation capacity using the sensor-selector combination. (A) Experimental approach for the selection of the fastest growing strains in a three-strain cocultivation. (B) FACS analysis of two-day coculture of three yeast strains (wild type, zwf1Δ and ZWF1 overexpression by TEF1 promoter) with or without the sensor-selector combination integrated in their genomes. (C) Same setup as in (B) except that the mutant strain zwf1Δ ALD6 overexpression instead of ZWF1 overexpression strain was cocultivated with the wild type and zwf1Δ mutant strain. Error bars represent standard deviations of population distributions in triplicate cocultures.
screening of populations that have different NADPH redox states and therefore useful for screening larger libraries of mutant strains generated by mutagenesis or other advanced genome editing techniques.
strain. In the sensor−selector group, however, the populations for the wild type and ZWF1 overexpression strains were almost equal after 24 h (48% vs 50%) and after 48 h the ZWF1 overexpression strain started to outcompete the wild type strain (62% vs 38%). This suggests that the ZWF1 overexpression improves the overall NADPH/NADP+ redox state during the two-day culture thereby slightly increasing the growth rate as a result of lower CDC14 expression. This growth-coupled selection using the sensor−selector combination was much clearer for the cocultivation of the wild type, zwf1Δ and zwf1Δ ALD6 overexpression strains. Similar to the other (i.e., wild type/zwf1Δ/ZWF1 overexpression) cocultivation, the fraction of zwf1Δ mutant strain in the nonselector group decreased to 16% after 1 day and 9% after 2 days (Figure 6c), and the zwf1Δ ALD6 overexpression strain grew slightly slower compared to the wild type (37% vs 47% after 1 day). In the selector group, the zwf1Δ strain also disappeared after 1 day of cocultivation, and the zwf1Δ ALD6 overexpression strain was the majority (82% of the population) and became dominant (95% of total population) after 2 days (Figure 6c). These data show that although ZWF1 has an important role in NADPH hemostasis, deletion of this gene causes severely impaired NADPH regeneration, and overexpression of this gene adds very little to the cells capability in generating NADPH (note that the ZWF1 overexpression strain only started to outcompete the wild type strain after 2 days of cocultivation in the selector group, Figure 6b). Instead, ALD6 overexpressed in the zwf1Δ mutant resulted in a strain that grew clearly faster than the wild type strain in the selector group, which suggests that overexpression of ALD6 is more effective than ZWF1 overexpression for NADPH generation. This selection would not have been possible without the synthetic sensor−selector, since neither the ZWF1 overexpression nor ALD6 overexpression in zwf1Δ background provides growth advantage over the wild type strain (Supplementary Figure S5). Our approach shows that a novel selection mechanism based on endogenous sensing and actuation is robust for the
■
DISCUSSION
In this study, we engineered an NADPH/NADP+ redox biosensor based on the oxidative stress responsive TF Yap1p in the budding yeast S. cerevisiae. Genetically encoded biosensors are useful tools and given the conservation of this protein family our engineering strategy could be applicable for investigating real-time cell physiology and many other dynamic processes in other eukaryotic systems as well.44 One application of the NADPH sensor is that it can be implemented for in vivo monitoring of cellular oxidative stress and functionally characterizing genetic components involved in NADPH metabolism. NADPH homeostasis has been proposed as a potential therapeutic target for cancer since several recent studies have shown that maintaining a high NADPH/NADP+ ratio can promote cancer cell growth and proliferation by fueling anabolism as well as protecting cancer cells against oxidative stress during nutrient limitation.45,46 Given S. cerevisiae’s role as a model eukaryote, the NADPH redox sensor developed in this study should be able to facilitate the investigation of the yeast counterparts of these relevant pathways and gain knowledge on how cells generally acquire robustness against oxidative stress. In line with this, we applied the sensor-reporter to characterize genetic modifications predicted by a flux balance model to have an effect on NADPH turnover (Figure 4a and Table 1). Using the sensorreporter we found that disruptions of serine/glycine biosynthesis lead to low NADPH or high oxidative stress, which is consistent with previous findings that the serine/glycinecoupled folate metabolism has an important role in NADPH generation and was previously suggested as a target for proliferating cells such as cancer cells.47 Therefore, it is conceivable that this NADPH/NADP+ biosensor could be used for high throughput screening of bioactive compounds that 1552
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
this study offers more facile and cost-effective monitoring of cellular NADPH/NADP+ ratios compared to tedious and sensitive calorimetric assays (Supplementary Figure S6). In short, our NADPH redox sensor should be useful for large-scale studies of cellular redox states and screening of genetic components involved in redox homeostasis. The novel DSGs-based sensor−selector combinations provide an alternative mechanism for selection of large libraries for desired phenotypes and should be possible to be implemented in genetic circuits for other synthetic biology applications. The sensor may also be applied in industrial fermentations, in which oxidative stress can conditionally activate a rescuing mechanism to improve cell survival and robustness of such processes.
specifically target oxPPP or other important pathways related to NADPH homeostasis in proliferating cells. From an engineering perspective, the NADPH sensor can also be useful to identify knockout targets for improved NADPH cofactor balancing, since this cofactor is heavily demanded in the biosynthesis of many industrially relevant compounds, such as 3-hydroxypropionic acid or fatty acid-based products.48,49 In order to fully explore the potential of very large libraries generated by DNA synthesis combined with advanced assembly or genome-editing techniques, a preselection may be necessary to reduce the library size and enrich the cells that have improved performance before they are screened using FACS.39 Antibiotic resistance is commonly used for this type of selections, in which the expression of the resistance gene can be conditionally activated by the sensor output. However, antibiotic resistance is not a suitable selection mechanism to use together with our NADPH redox sensor, because the sensor output is inversely correlated with the NADPH/NADP+ ratios. This would mean cell survival in the presence of an antibiotic would be linked to low NADPH levels, which is generally unfavorable for cell homeostasis. Although an inverter could be applied to change the sensor output from ON to OFF during the selection, this construction would introduce an additional layer of complexity in the system. Moreover, antibiotic resistance-based selection may suffer from lower resolution due to the instability of antibiotics for long-term selection, such as adaptive laboratory evolution (ALE). Finally, antibiotics may also be degraded by cells that express the resistance gene at very high levels (especially on agar solid media), thus reducing the selection pressure on the whole population. In contrast, our sensor/DSG-based selection can delicately control cell growth depending on an endogenous signal (e.g., the intracellular NADPH/NADP+) of that individual cell, thereby selecting the cell that has the lowest signal. In principle, this novel sensor-selector strategy does not require use of any external chemical selector (e.g., antibiotic) and can facilitate screening of large libraries in liquid as well as solid medium with a low false positive rate that may arise from inhomogeneity due to degradation of antibiotics. Inspired by several metabolic engineering efforts using biosensor-regulator in bacteria,50,51 we also anticipate that this TF-based redox sensor can be implemented in regulatory circuits, and could thereby enable dynamic control of the metabolic networks in yeast and potentially other eukaryotes.11 Finally, it has been reported that Yap1p TF regulates expression of over 100 genes.52 To minimize the interference with the native regulatory network, we decided to use the TRX2 promoter, a downstream target of Yap1p, as readout of its oxidation and nuclear accumulation. Our data with PTRX2 variants confirm the indispensability of the Yap1p binding site (Figure 2a) for the oxidant response of this promoter. Regarding the sensor specificity, we also wondered whether the NADH/NAD+ redox state has any impact on the reversible oxidation of Yap1p. Our finding that deletion of ADH1 led to a dramatic increase in NADH/NAD+ (determined by the NADH sensor PGPD2-yEGFP) but little change in the NADPH sensor output (Figure 3c,d) shows that these two redox sensors are quite specific to changes in NADH and NADPH, respectively. These results also suggest that an elevated cytosolic NADH/ NAD+ ratio does not improve the cytosolic NADPH/NADP+ redox state significantly, in which case a decreased GFP level would have been observed in the adh1Δ mutant strain. Finally, and most importantly, the engineered biosensor presented in
■
METHODS Plasmids and Strains. Plasmids (Supplementary Table S1) were constructed using USER cloning and propagated in Escherichia coli DH5alpha strain as described.53 All yeast strains used in this study (Supplementary Table S2) have CEN.PK MATa background and are maintained following standard protocol. Yeast strains were transformed with linearized or episomal plasmids using the lithium acetate/single-stranded carrier DNA/PEG method.54 Genes were knocked out by transforming the parent strains with linear DNA fragments containing a KanMX4 cassette with both flanking regions homologous to the up- and downstream of the targeted loci, followed by selection on appropriate growth media containing 100 mg/L G418-Sulfate (Sigma-Aldrich). For genome integration, the expression cassettes were always USER cloned into integrative plasmids, which were linearized and transformed into the parent strains followed by selection on appropriate media. ALD6 and ZWF1 overexpression cassettes were cloned into integrative plasmids containing a synthetic ble gene and transformed into yeast followed by selection on growth media containing 10 mg/L phleomycin (InvivoGen). In the flow cytometry analysis, each strain was transformed with two linearized integrative plasmids: one contained the Kluyveromyces lactis URA3 gene (KlURA3) and one of the fluorescence proteins followed and the other contained the Schizosaccharomyces pombe HIS5 gene (SpHIS5) and one of the sensor−selector combinations, followed by selection on yeast synthetic drop-out media lacking uracil or both uracil and histidine. For knock-down strains, the parent strain CEN.PK102-5B was first transformed with linearized sensor− reporter combination (without any selection markers), and the integration was facilitated by CRISPR-Cas9 (KanMX4) together with a guide RNA (sequence ATATGTCCTAATTTTGGAA) (NatMX) targeting at EasyClone site XI-3, followed by selection on medium containing 100 mg/L G418-sulfate and 50 mg/L nourseothricin. The resulting strain was further transformed with PREV1 (1000 bp), PRNR2 (761 bp), and PPFK1 (840 bp), which were amplified using primers whose 5′-ends was homologous to PPFK1 (500 bp), PPGI1 (124 bp), and PTPI1 (400 bp), respectively. Guide-RNA plasmids (LEU2) that target PPFK1 (sequence: TGAAGTAAGAATAACAATAT), PPGI1 (sequence: AAAATGGGACGAAACAAATA), and PTPI1 (sequence: TATAAAGGGCAGCATAATTT) were cotransformed to facilitate the integration by homologous recombination. All mutant strains were verified by yeast colony PCR. Media and cell cultures. All E. coli strains were cultured in standard LB medium containing 100 mg/L ampicillin (SigmaAldrich). All yeast strains were cultured on either yeast extract− peptone dextrose (YPD) or yeast synthetic drop-out media 1553
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology
were predicted using three different methods that are based on distinct assumptions on how the cells will cope with the perturbation: (i) pFBA, as already used for predicting wild type fluxes, assumes that the cells will be able to attain the next optimal state after the gene loss, which might be an unrealistic expectation in the short term but might hold true after a longer period of adaptation. (ii) MOMA minimizes the Euclidean distance of the mutant and wild type flux distribution and thus finds the closest feasible but not necessarily optimal solution to the original wild type solution.37 It has been shown that MOMA exhibits a higher accuracy in predicting gene essentiality than flux balance analysis. In this work we minimized the Manhattan distance instead to avoid overproportional punishment of changes in large fluxes. (iii) ROOM on the other hand minimizes the number of fluxes that need to be switched on and off in order to reach a new feasible solution.38 The authors of ROOM have argued that ROOM models predict more accurately the transcriptional regulatory events that need to take place after the genetic perturbation. Reactions to be removed after gene deletions were determined using the Boolean gene-protein-reaction rules included in the model. Except for TPI1 and PFK1, only gene knockout targets that are predicted by at least two methods to have increased NADPH turnover (positive ΔvNADPH) were selected. Further details and mathematical formulations of the methods are provided in the Supporting Information: Constraint-based metabolic modeling.
(lacking the proper amino acids for selection) (Sigma-Aldrich). For synthetic drop-out media containing antibiotics, sodium glutamate instead of ammonia was used as the nitrogen source. For all fluorescence measurements using the microtiter plate reader, yeast cells were cultured in a minimal mineral (Delft) medium.53 All media, chemicals, and antibiotics (except phleomycin) were purchased from Sigma-Aldrich. Fluorescence Measurements. A yeast-enhanced green fluorescent protein (yEGFP)55 was used as the main reporter. When expressed in episomal plasmids, another far-red fluorescent protein mKate256 was expressed under the control of the TEF2 constitutive promoter in order to normalize the GFP signal for variation in plasmid copy number. GFP and mKate2 signals for all strains were measured in 96-well microtiter plates with a Biotek Synergy 2 plate reader (Holm & Halby) using excitation/emission wavelengths of 485/515 nm and 588/633 nm, respectively. All fluorescence measurements were background corrected and normalized by absorbance at 600 nm (OD600). Flow Cytometry Analysis. For easy identification, all strains used in the flow cytometry analysis had one fluorescent protein (yEGFP, mKate2 or CFP) expressed under the control of the TEF1 constitutive promoter. Precultures were grown in SC-dropout media to mid log phase (OD600 around 1.0) and mixed at equal ratios to a final OD600 of 0.005 at Day 0 and incubated at normal growth condition in a 24-well deep plate (with 1 mL culture in each well). After 24 h (day 1), the mix cultures were transferred to fresh medium to OD600 0.25 and grown for another 24 h (day 2). Cell cultures from both day 1 and day 2 were analyzed using BD FACS Aria III equipped with lasers for green, red, and cyan fluorescent proteins (Figure 6A). The flow cytometry data were analyzed using FlowJo. Cofactor Quantification by Calorimetric Assay. NADPH and NADP+ were extracted and measured using the NADPH/NADP+ Quantitation Kit (Sigma-Aldrich) following the manufacturer’s instruction with slight modification. Cells were grown to mid-log phase and harvested by centrifugation at 11 000g for 1 min. Cell pellets were suspended with 400 μL of extraction buffer from the kit and transferred to a new tube with 50 μL of acid-washed glass beads, and kept on dry ice for 10 min, followed by incubating at room temperature until cells started to thaw (but still frozen). Cells were then lysed on a bead-beater homogenizer for 10 s and immediately put on ice for 2−3 min, and centrifuged at 17 000g for 10 min. The supernatant was filtered using a Pierce protein concentrator (10K MWCO PES, Thermo Scientific) to remove proteins according to the kit instruction. All steps between cell harvest and enzymatic assay were carried out at 0 °C. Constraint-Based Metabolic Modeling. The genomescale model iMM90457 was used for all simulations. The aerobic glucose minimal media condition as set in the original model (obtained from the Supporting Information) was left as is. Parsimonious flux balance analysis (pFBA),36 which maximizes biomass production while minimizing flux magnitude in order to simulate maximum growth under optimal proteome allocation, was used to predict the flux distribution in the wild type strain. Since constraint-based models such as iMM904 do not have metabolite concentrations as state variables, predicted changes in NADPH turnover rates (i.e., the sum over all producing or consuming fluxes) were used as a proxy for changes in the NADPH/NADP+ redox state, following a similar approach as described by Chemler et al.35 Mutant flux distributions for all possible single gene deletions
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acssynbio.6b00135. Pasmids and yeast strains used in this study; additional figures supporting the text; constraint-based metabolic modeling (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Author Contributions
J.Z., M.K.J., and J.D.K. conceived the project. J.Z., T.P.B.P., and K.R.P. constructed plasmids and strains and all experiments. N.S. performed constraint-based metabolic modeling. J.Z., M.K.J., and J.D.K. analyzed and interpreted the data. J.Z., M.K.J., and J.D.K. wrote the manuscript and all authors assisted. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Dushica Arsovska (NNF CFB, DTU) for technical assistance in flow cytometry analysis and funding from the Novo Nordisk Foundation.
■
REFERENCES
(1) Carmel-Harel, O., and Storz, G. (2000) Roles of the glutathioneand thioredoxin-dependent reduction systems in the Escherichia coli and saccharomyces cerevisiae responses to oxidative stress. Annu. Rev. Microbiol. 54, 439−461. (2) Sauve, A. a, Wolberger, C., Schramm, V. L., and Boeke, J. D. (2006) The biochemistry of sirtuins. Annu. Rev. Biochem. 75, 435−465.
1554
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology (3) Filomeni, G., De Zio, D., and Cecconi, F. (2015) Oxidative stress and autophagy: the clash between damage and metabolic needs. Cell Death Differ. 22, 377−388. (4) Canelas, A. B., van Gulik, W. M., and Heijnen, J. J. (2008) Determination of the cytosolic free NAD/NADH ratio in Saccharomyces cerevisiae under steady-state and highly dynamic conditions. Biotechnol. Bioeng. 100, 734−43. (5) Hung, Y. P., Albeck, J. G., Tantama, M., and Yellen, G. (2011) Imaging cytosolic NADH-NAD(+) redox state with a genetically encoded fluorescent biosensor. Cell Metab. 14, 545−54. (6) Hedeskov, C. J., Capito, K., and Thams, P. (1987) Cytosolic ratios of free [NADPH]/[NADP+] and [NADH]/[NAD+] in mouse pancreatic islets, and nutrient-induced insulin secretion. Biochem. J. 241, 161−7. (7) Chance, B., Cohen, P., Jobsis, F., and Schoener, B. (1962) Intracellular oxidation-reduction states in vivo. Science (Washington, DC, U. S.) 137, 499−508. (8) Seifar, R. M., Ras, C., Deshmukh, A. T., Bekers, K. M., SuarezMendez, C. a., da Cruz, A. L. B., van Gulik, W. M., and Heijnen, J. J. (2013) Quantitative analysis of intracellular coenzymes in Saccharomyces cerevisiae using ion pair reversed phase ultra high performance liquid chromatography tandem mass spectrometry. J. Chromatogr. A 1311, 115−120. (9) Wagner, T. C., and Scott, M. D. (1994) Single extraction method for the spectrophotometric quantification of oxidized and reduced pyridine nucleotides in erythrocytes. Anal. Biochem. 222, 417. (10) Veech, R. L., Eggleston, L. V., and Krebs, H. a. (1969) The redox state of free nicotinamide-adenine dinucleotide phosphate in the cytoplasm of rat liver. Biochem. J. 115, 609−619. (11) Zhang, J., Jensen, M. K., and Keasling, J. D. (2015) Development of biosensors and their application in metabolic engineering. Curr. Opin. Chem. Biol. 28, 1−8. (12) Hanson, G. T., Aggeler, R., Oglesbee, D., Cannon, M., Capaldi, R. a., Tsien, R. Y., and Remington, S. J. (2004) Investigating Mitochondrial Redox Potential with Redox-sensitive Green Fluorescent Protein Indicators. J. Biol. Chem. 279, 13044−13053. (13) Yano, T., Oku, M., Akeyama, N., Itoyama, A., Yurimoto, H., Kuge, S., Fujiki, Y., and Sakai, Y. (2010) A novel fluorescent sensor protein for visualization of redox states in the cytoplasm and in peroxisomes. Mol. Cell. Biol. 30, 3758−66. (14) Zhao, Y., Jin, J., Hu, Q., Zhou, H. M., Yi, J., Yu, Z., Xu, L., Wang, X., Yang, Y., and Loscalzo, J. (2011) Genetically encoded fluorescent sensors for intracellular NADH detection. Cell Metab. 14, 555−566. (15) Knudsen, J., Carlquist, M., and Gorwa-Grauslund, M. (2014) NADH-dependent biosensor in Saccharomyces cerevisiae: principle and validation at the single cell level. AMB Express 4, 81. (16) Siedler, S., Schendzielorz, G., Binder, S., Eggeling, L., Bringer, S., and Bott, M. (2014) SoxR as a single-cell biosensor for NADPHconsuming enzymes in Escherichia coli. ACS Synth. Biol. 3, 41−47. (17) Lill, R., and Mühlenhoff, U. (2005) Iron-sulfur-protein biogenesis in eukaryotes. Trends Biochem. Sci. 30, 133−141. (18) Cremers, C. M., and Jakob, U. (2013) Oxidant sensing by reversible disulfide bond formation. J. Biol. Chem. 288, 26489−26496. (19) Toone, W. M., Morgan, B. a, and Jones, N. (2001) Redox control of AP-1-like factors in yeast and beyond. Oncogene 20, 2336− 2346. (20) Lu, J., and Holmgren, A. (2014) The thioredoxin antioxidant system. Free Radical Biol. Med. 66, 75−87. (21) Wood, M. J., Storz, G., and Tjandra, N. (2004) Structural basis for redox regulation of Yap1 transcription factor localization. Nature 430, 917−21. (22) Yan, C., Lee, L. H., and Davis, L. I. (1998) Crm1p mediates regulated nuclear export of a yeast AP-1-like transcription factor. EMBO J. 17, 7416−29. (23) Isoyama, T., Murayama, A., Nomoto, A., and Kuge, S. (2001) Nuclear Import of the Yeast AP-1-like Transcription Factor Yap1p Is Mediated by Transport Receptor Pse1p, and This Import Step Is Not Affected by Oxidative Stress. J. Biol. Chem. 276, 21863−21869.
(24) Kuge, S., and Jones, N. (1994) YAP1 dependent activation of TRX2 is essential for the response of Saccharomyces cerevisiae to oxidative stress by hydroperoxides. EMBO J. 13, 655−64. (25) Sharon, E., Kalma, Y., Sharp, A., Raveh-Sadka, T., Levo, M., Zeevi, D., Keren, L., Yakhini, Z., Weinberger, A., and Segal, E. (2012) Inferring gene regulatory logic from high-throughput measurements of thousands of systematically designed promoters. Nat. Biotechnol. 30, 521−530. (26) Rodrigues-Pousada, C. A., Nevitt, T., Menezes, R., Azevedo, D., Pereira, J., and Amaral, C. (2004) Yeast activator proteins and stress response: An overview. FEBS Lett. 567, 80−85. (27) Fernandes, L., Rodrigues-Pousada, C., and Struhl, K. (1997) Yap, a novel family of eight bZIP proteins in Saccharomyces cerevisiae with distinct biological functions. Mol. Cell. Biol. 17, 6982−93. (28) Jayaraman, M., Radhika, V., Bamne, M. N., Ramos, R., Briggs, R., and Dhanasekaran, D. N. (2005) Engineered Saccharomyces cerevisiae strain BioS–OS1/2, for the detection of oxidative stress. Biotechnol. Prog. 21, 1373−1379. (29) Allen, S. A., Clark, W., McCaffery, J. M., Cai, Z., Lanctot, A., Slininger, P. J., Liu, Z. L., and Gorsich, S. W. (2010) Furfural induces reactive oxygen species accumulation and cellular damage in Saccharomyces cerevisiae. Biotechnol. Biofuels 3, 2. (30) Blazeck, J., Garg, R., Reed, B., and Alper, H. S. (2012) Controlling promoter strength and regulation in Saccharomyces cerevisiae using synthetic hybrid promoters. Biotechnol. Bioeng. 109, 2884−2895. (31) Hector, R. E., Bowman, M. J., Skory, C. D., and Cotta, M. A. (2009) The Saccharomyces cerevisiae YMR315W gene encodes an NADP(H)-specific oxidoreductase regulated by the transcription factor Stb5p in response to NADPH limitation. New Biotechnol. 26, 171−180. (32) Castegna, A., Scarcia, P., Agrimi, G., Palmieri, L., Rottensteiner, H., Spera, I., Germinario, L., and Palmieri, F. (2010) Identification and functional characterization of a novel mitochondrial carrier for citrate and oxoglutarate in Saccharomyces cerevisiae. J. Biol. Chem. 285, 17359−17370. (33) Cordier, H., Mendes, F., Vasconcelos, I., and Francois, J. M. (2007) A metabolic and genomic study of engineered Saccharomyces cerevisiae strains for high glycerol production. Metab. Eng. 9, 364−378. (34) Ng, C. Y., Jung, M., Lee, J., and Oh, M.-K. (2012) Production of 2,3-butanediol in Saccharomyces cerevisiae by in silico aided metabolic engineering. Microb. Cell Fact. 11, 68. (35) Chemler, J. a, Fowler, Z. L., McHugh, K. P., and Koffas, M. a G. (2010) Improving NADPH availability for natural product biosynthesis in Escherichia coli by metabolic engineering. Metab. Eng. 12, 96−104. (36) Lewis, N. E., Hixson, K. K., Conrad, T. M., Lerman, J. A., Charusanti, P., Polpitiya, A. D., Adkins, J. N., Schramm, G., Purvine, S. O., Lopez-Ferrer, D., Weitz, K. K., Eils, R., König, R., Smith, R. D., and Palsson, B. Ø. (2010) Omic data from evolved E. coli are consistent with computed optimal growth from genome-scale models. Mol. Syst. Biol. 6, 390. (37) Segrè, D., Vitkup, D., and Church, G. M. (2002) Analysis of optimality in natural and perturbed metabolic networks. Proc. Natl. Acad. Sci. U. S. A. 99, 15112−7. (38) Shlomi, T., Berkman, O., and Ruppin, E. (2005) Regulatory on/ off minimization of metabolic flux changes after genetic perturbations. Proc. Natl. Acad. Sci. U. S. A. 102, 7695−7700. (39) Dietrich, J. A., McKee, A. E., and Keasling, J. D. (2010) Highthroughput metabolic engineering: advances in small-molecule screening and selection. Annu. Rev. Biochem. 79, 563−590. (40) Esvelt, K. M., and Wang, H. H. (2013) Genome-scale engineering for systems and synthetic biology. Mol. Syst. Biol. 9, 641. (41) Raman, S., Rogers, J. K., Taylor, N. D., and Church, G. M. (2014) Evolution-guided optimization of biosynthetic pathways. Proc. Natl. Acad. Sci. U. S. A. 111, 17803. (42) Makanae, K., Kintaka, R., Makino, T., Kitano, H., and Moriya, H. (2013) Identification of dosage-sensitive genes in Saccharomyces 1555
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
Research Article
ACS Synthetic Biology cerevisiae using the genetic tug-of-war method. Genome Res. 23, 300− 311. (43) Lee, M. E., Aswani, A., Han, A. S., Tomlin, C. J., and Dueber, J. E. (2013) Expression-level optimization of a multi-enzyme pathway in the absence of a high-throughput assay. Nucleic Acids Res. 41, 10668− 10678. (44) Toone, W. M., and Jones, N. (1999) AP-1 transcription factors in yeast. Curr. Opin. Genet. Dev. 9, 55−61. (45) Jiang, P., Du, W., Mancuso, A., Wellen, K. E., and Yang, X. (2013) Reciprocal regulation of p53 and malic enzymes modulates metabolism and senescence. Nature 493, 689−93. (46) Jeon, S.-M., Chandel, N. S., and Hay, N. (2012) AMPK regulates NADPH homeostasis to promote tumour cell survival during energy stress. Nature 485, 661−5. (47) Fan, J., Ye, J., Kamphorst, J. J., Shlomi, T., Thompson, C. B., and Rabinowitz, J. D. (2014) Quantitative flux analysis reveals folatedependent NADPH production. Nature 510, 298−302. (48) Chen, Y., Bao, J., Kim, I. K., Siewers, V., and Nielsen, J. (2014) Coupled incremental precursor and co-factor supply improves 3hydroxypropionic acid production in Saccharomyces cerevisiae. Metab. Eng. 22, 104−109. (49) d’Espaux, L., Mendez-Perez, D., Li, R., and Keasling, J. D. (2015) Synthetic biology for microbial production of lipid-based biofuels. Curr. Opin. Chem. Biol. 29, 58. (50) Xu, P., Li, L., Zhang, F., Stephanopoulos, G., and Koffas, M. (2014) Improving fatty acids production by engineering dynamic pathway regulation and metabolic control. Proc. Natl. Acad. Sci. U. S. A. 111, 11299. (51) Zhang, F., Carothers, J. M., and Keasling, J. D. (2012) Design of a dynamic sensor-regulator system for production of chemicals and fuels derived from fatty acids. Nat. Biotechnol. 30, 354. (52) Zhang, C., Li, Z., Zhang, X., Yuan, L., Dai, H., and Xiao, W. (2016) Transcriptomic profiling of chemical exposure reveals roles of Yap1 in protecting yeast cells from oxidative and other types of stresses. Yeast 33, 5−19. (53) Jensen, N. B., Strucko, T., Kildegaard, K. R., David, F., Maury, J., Mortensen, U. H., Forster, J., Nielsen, J., and Borodina, I. (2014) EasyClone: Method for iterative chromosomal integration of multiple genes in Saccharomyces cerevisiae. FEMS Yeast Res. 14, 238−248. (54) Gietz, R. D., and Schiestl, R. H. (2008) High-efficiency yeast transformation using the LiAc/SS carrier DNA/PEG method. Nat. Protoc. 2, 31−35. (55) Cormack, B. P., Bertram, G., Egerton, M., Gow, N. a, Falkow, S., and Brown, a J. (1997) Yeast-enhanced green fluorescent protein (yEGFP): a reporter of gene expression in Candida albicans. Microbiology 143, 303−11. (56) Shcherbo, D., Murphy, C. S., Ermakova, G. V., Solovieva, E. A., Chepurnykh, T. V., Shcheglov, A. S., Verkhusha, V. V., Pletnev, V. Z., Hazelwood, K. L., Roche, P. M., Lukyanov, S., Zaraisky, A. G., Davidson, M. W., and Chudakov, D. M. (2009) Far-red fluorescent tags for protein imaging in living tissues. Biochem. J. 418, 567−74. (57) Mo, M. L., Palsson, B. Ø., and Herrgard, M. J. (2009) Connecting extracellular metabolomic measurements to intracellular flux states in yeast. BMC Syst. Biol. 3, 37.
1556
DOI: 10.1021/acssynbio.6b00135 ACS Synth. Biol. 2016, 5, 1546−1556
