Using Transcriptomics To Improve Butanol Tolerance of Synechocystis sp. Strain PCC 6803

Abstract

Cyanobacteria are emerging as promising hosts for production of advanced biofuels such as n-butanol and alkanes. However, cyanobacteria suffer from the same product inhibition problems as those that plague other microbial biofuel hosts. High concentrations of butanol severely reduce growth, and even small amounts can negatively affect metabolic processes. An understanding of how cyanobacteria are affected by their biofuel product can enable identification of engineering strategies for improving their tolerance. Here we used transcriptome sequencing (RNA-Seq) to assess the transcriptome response of Synechocystis sp. strain PCC 6803 to two concentrations of exogenous n-butanol. Approximately 80 transcripts were differentially expressed at 40 mg/liter butanol, and 280 transcripts were different at 1 g/liter butanol. Our results suggest a compromised cell membrane, impaired photosynthetic electron transport, and reduced biosynthesis. Accumulation of intracellular reactive oxygen species (ROS) scaled with butanol concentration. Using the physiology and transcriptomics data, we selected several genes for overexpression in an attempt to improve butanol tolerance. We found that overexpression of several proteins, notably, the small heat shock protein HspA, improved tolerance to butanol. Transcriptomics-guided engineering created more solvent-tolerant cyanobacteria strains that could be the foundation for a more productive biofuel host.

INTRODUCTION

Cyanobacteria are attractive biochemical production platforms due to their minimal nutrient requirements, ease of genetic manipulation, and wide natural diversity. Several model cyanobacteria have been engineered as hosts for production of chemicals and biofuels such as hydrogen (1), ethanol (2), isobutanol (3, 4), and n-butanol (5). The modified cyanobacterial strains contained enzymes from fermentative organisms, such as Clostridium, Lactococcus, and Zymomonas species. The alcohols were produced at levels (up to 450 mg/liter) that were detectable but well below the titers of 10 to 20 g/liter achieved by industrial Clostridium and Saccharomyces strains (6). In addition to restrictions on metabolic flux (7, 8), hosts may suffer from product inhibition even at low levels of product (9). Biofuel hosts that are more tolerant to the product may be more efficient producers even at levels below toxicity (10, 11), though this phenomenon is not universal (12).

The deleterious effects of solvents on a wide range of microbes have been reported and recently reviewed (13). Solvents compromise the cell membrane and alter membrane fluidity (14). A compromised cell membrane can leak metabolites, resulting in loss of the transmembrane electrochemical gradient and hence of the proton-motive force. Reactive oxygen species (ROS) may accumulate as the cell modulates respiration to recover lost ATP (9, 15). These effects have been observed for bacteria, archaea, and yeast. Recent studies have begun to describe the complex response of cyanobacteria to solvents such as ethanol (16), n-butanol (17), and hexane (18). Photosynthetic organisms may be more susceptible to the detrimental effects of solvents due to an extreme sensitivity to the redox state of key molecules such as plastoquinone and the intricate organization of the membrane-bound photosynthetic apparatus. Understanding the solvent-induced stress response in cyanobacteria could guide genetic engineering of solvent tolerance. A guided engineering approach typically involves assessing the cell response to exogenous solvents at either the transcript or protein level followed by deletion or overexpression of selected genes identified from the analysis (19–21).

Butanol is a large-market chemical with high-value applications as a chemical precursor and low-value applications as a fuel additive. In this study, the transcriptomic stress response to exogenous butanol of the model cyanobacterium Synechocystis sp. strain PCC 6803 was measured with transcriptome sequencing (RNA-Seq). We studied the effects of butanol at concentrations that approximate reported production titers. By including two butanol concentrations, we attempted to identify a butanol dosage-dependent response. In the second phase, we used the physiology and transcriptomics data to select genes for overexpression with the aim of improving butanol tolerance. This work demonstrates that transcriptomics-based engineering can be used to improve Synechocystis growth rates and cell viability in the presence of butanol.

MATERIALS AND METHODS

Culture conditions and mRNA extraction.

Wild-type Synechocystis sp. strain PCC 6803 (Pasteur Culture Collection) and a mutant strain, ssJA004 (Δslr1829-slr1830::Km^r), were cultivated and used for RNA extractions. Cultures were grown photoautotrophically in BG-11 or BG-11₀ medium (22) at 30°C, with 50 μE s⁻¹ m⁻² illumination, atmospheric CO₂, and 150 rpm shaking. In total, four cultures were prepared in biological duplicates: (i) B−, Synechocystis in BG-11; (ii) B+, Synechocystis in BG-11 with butanol at 40 mg/liter; (iii) B++, Synechocystis in BG-11 with butanol at 1 g/liter; and (iv) N−, ssJA004 in BG-11₀. All cultures (40 ml) were first grown in a preparatory phase in BG-11 without butanol until an optical density at 730 nm (OD₇₃₀) of 0.5 was reached. Cells were collected by centrifugation at 4,000 × g for 10 min and washed with either BG-11 (B+, B++, and B− samples) or BG-11₀ (N− sample) and resuspended in the same before addition of butanol to B+ and B++ samples. All cultures were then incubated for 12 h before collection of cells at 16,000 × g and 4°C for 1 min. Pellets were immediately used for RNA extraction. Total RNA was isolated and DNase treated using an Ambion RiboPure-Bacteria kit according to the manufacturer's protocol with the following exception: the cell lysis time in step C5 was extended to 30 min. mRNA was enriched using an Ambion MICROBExpress kit according to the manufacturer's protocol and an extended rRNA capture step of 1 h. The quality of mRNA was assessed with a Bioanalyzer 2100 (Agilent). The RNA integrity number (RIN) was >8 for all samples, indicating intact mRNA.

Transcriptome sequencing and analysis.

All samples were sequenced multiplexed on a single lane on an Illumina HiSeq 2000 sequencing system (2 × 101 bp). Raw reads were subjected to trimming of poor-quality ends and removal of adaptor contamination using the FastqMcf tool of ea-utils (23). Each sample was mapped to a PCC 6803 reference genome (GenBank accession number NC_017277.1) using the aligner MOSAIK 2.1, with a hash size of 12 bp. Potential PCR duplicates were eliminated using the samtools rmdup program (24). The number of reads mapping to each gene was determined using the htseq-count script from the HT-Seq framework (25). Normalization of read counts and differential gene expression analysis was done using DESeq (26) and comparing each of the conditions to the control (B−), with a cutoff at least 1.5-fold change and a false discovery rate of 10%.

Growth rate determination.

Synechocystis cultures were grown photoautotrophically in BG-11 at 30°C, with 50 μE s⁻¹ m⁻² illumination, atmospheric CO₂, and 150 rpm shaking. Cells were cultured (3 ml) in 24-well culture plates in BG-11 medium buffered to pH 7.5 with 25 mM HEPES. Cultures were inoculated from a mature culture to an OD₇₃₀ of 0.05. For butanol-challenged cultures, butanol was added at the initial inoculation. The culture plate was agitated using a Multi-Genie MicroPlate Shaker (Fisher Scientific). Cell density measurements were taken at approximately 24-h intervals for 8 days. Cell cultures were performed in triplicate. Growth rate was determined through linear fit of the logarithmic growth equation. The butanol concentration of the culture medium was measured twice daily during the cultivation via gas chromatography. Butanol was added to cultures twice daily to account for evaporation.

Measurement of ROS and autofluorescence with flow cytometry.

Flow cytometry was used to simultaneously measure reactive oxygen species (ROS) through fluorescence emission of the CellROX dye (Life Technologies), chlorophyll (Chl) content, and phycocyanin (PC) content through autofluorescence analysis. Data corresponding to the approximate spectral properties of Chl and PC were taken from Beutler et al. (27). Cell cultures (50 μl) at an OD₇₃₀ of 0.1 were challenged with butanol at various concentrations for 1 h at 30°C, with 50 μE s⁻¹ m⁻² illumination and atmospheric CO₂. The CellROX staining was done according to the manufacturer's instructions; the CellROX reagent was added directly to butanol-challenged cell cultures after 1 h, and the mixture was incubated in darkness for an additional 30 min. Cells were then analyzed on a Beckman Gallios flow cytometer (Beckman Coulter) without washing. The fluorescent channels were FL-1 (488-nm excitation [ex], 525-nm emission [em]; CellROX reagent), FL-3 (488-nm ex, 685-nm em; Chl fluorescence), and FL-6 (620-nm ex, 640-nm em; PC fluorescence). Reported fluorescent intensity values represent arithmetic means of the results determined for 10,000 analyzed cells. Chlorophyll was extracted from cells after pelleting by addition of 100% methanol and a 10-min vortex treatment. Cell debris was pelleted and the supernatant analyzed with an absorbance spectrophotometer.

Cell viability.

Cell viability measurements were performed by butanol challenge, plating of culture, and counting of resultant colonies. Cells at the exponential-growth phase (50 μl; OD₇₃₀ of 0.5) were incubated in buffered BG-11 media with added butanol for 1 h at 30°C, with 50 μE s⁻¹ m⁻² illumination and atmospheric CO₂. After incubation, 2 μl of cells was plated in a serial dilution. Colonies were counted after 5 days. Viability was defined as the ratio of colonies on the butanol-challenged plates to those with no added butanol. The viability assay was performed in duplicate.

Construction of cyanobacterial strains.

The pJA2 broad host replicating vector is a variant of the pPMQAK1 vector constructed by Huang et al. (28) and was generated through introduction of promoter P_psbA2 upstream of the BioBrick cloning site and a terminator (BioBrick B0015; http://partsregistry.org) downstream of the cloning site. P_psbA2 was taken as −360 to −6 upstream of the slr1311 start codon; nucleotides from −6 to −1 ATG were replaced by an XbaI site for cloning. The terminator region contained upstream SpeI and PstI sites for cloning. The pJA2 backbone was prepared for subcloning by PCR (Hot-Start Phusion polymerase; Thermo Scientific) and subsequent XbaI/PstI restriction digestion, phosphatase treatment (FastDigest enzymes; Fermentas), and DpnI treatment (New England BioLabs). Synechocystis genes for overexpression were amplified by PCR from the start codon to the stop codon with primers containing 5′ XbaI (forward) or 5′ PstI (reverse) overhangs and were restriction digested before ligation into XbaI/PstI-digested pJA2. The pJA2-trxA tpx construct was constructed in two steps. The trxA gene was first cloned into pJA2 in the same way as described for the previous inserts. The pJA2-trxA vector was then PCR amplified with 5′ PstI (forward) and 5′ SpeI (reverse) primers, using annealing downstream and at the 3′ end of trxA, respectively, and SpeI/PstI digested. The 3′ SpeI site is compatible with the 5′ XbaI site of the subsequent insert. The tpx gene was amplified with 5′ XbaI (forward) and 5′ PstI (reverse) overhangs and XbaI/PstI digested before ligation into pJA2-trxA. The genes chosen for overexpression were hspA (sll1514), sodB (slr1516), trxA and tpx (slr0623 and sll0755), fdXIII (slr1828), cccS (slr1667), and ssr0692.

Plasmid constructs were subcloned in Escherichia coli and transformants cultured (10 ml) for plasmid purification using Miniprep. Synechocystis was transformed by electroporation (10 to 100 ng plasmid DNA) as described by Chiaramonte et al. (29) and grown photoautotrophically.

Quantitative PCR measurement of gene expression.

Overexpression of genes from the Synechocystis pJA2 mutants was assessed with quantitative PCR from total RNA. The wild-type strain and pJA2 mutants were cultured photoautotrophically to mid-log phase (OD₇₃₀ = 1.0). A 5-ml quantity was collected and total RNA extracted using a Qiagen RNEasy Minikit, including on-column DNase treatment to remove genomic DNA. The DNase treatment differed from the recommended instruction with a 45-min incubation at 28°C. Approximately 1 μg of total RNA was retrieved. A SYBR green reverse transcriptase-PCR (RT-PCR) kit (Bio-Rad) was used to measure the amount of each transcript from total RNA of the wild-type strain and the pJA2 mutant. Calibration using known amounts of PCR amplicons allowed transcript quantification. The quantity of each transcript was normalized to that of rpoB (sll1787), and overexpression data represent this ratio as determined by comparisons of pJA2 mutants to the wild type. Absence of genomic DNA contamination was confirmed with reverse transcriptase (−) controls, and primer annealing efficiencies were over 85%.

RESULTS AND DISCUSSION

Butanol effect on growth rate and intracellular ROS of Synechocystis sp. strain PCC 6803.

We hypothesized that exogenous n-butanol concentrations below the tolerance threshold would still have a distinct effect on cellular processes and would more accurately simulate butanol production conditions. Two butanol concentrations were chosen. B+ (40 mg/liter) approximates the highest reported titers of n-butanol produced by a genetically modified cyanobacterium (7); B++ (1 g/liter) corresponds to the highest exogenous butanol concentration that does not affect growth rates (Fig. 1A). Growth rates were markedly reduced at concentrations higher than B++. These results are consistent with a recent study showing that butanol is more toxic to Synechocystis than ethanol (30). Slower growth at these high n-butanol concentrations was accompanied by an increase in intracellular ROS levels (Fig. 1B), indicating that this could be one mechanism of solvent-induced toxicity. Solvent-induced ROS accumulation has been observed in nonphotosynthetic organisms such as E. coli (31) and Saccharomyces cerevisiae (32). Alteration of electron transfer pathways in the photosystem electron transport chain (PETC) can produce ROS through several mechanisms (33). A small but significant butanol-dependent increase in the fluorescent signatures of two light-harvesting pigments, phycocyanin (PC) and chlorophyll (Chl), was observed (Fig. 1B). The pigment content was not affected by butanol treatment; the increase in fluorescence emission may have been due to reduced photochemical quenching arising from a closed or damaged photosystem (34).

Fig 1.

Butanol affects growth rate, ROS, and autofluorescence of Synechocystis sp. strain PCC 6803. (A) Specific growth rates determined during exponential growth (triplicate data). Concentrations of n-butanol chosen for RNA-Seq analysis are indicated. (B) Intracellular ROS levels determined with CellROX Green and measured via flow cytometry after 1 h of incubation with n-butanol. Autofluorescence of phycocyanin (Phyc) and chlorophyll (Chl) was measured via flow cytometry after 1 h of incubation with n-butanol.

Transcriptomics analysis of short-term butanol stress response.

Synechocystis cultures at the exponential-growth phase were incubated for 12 h with exogenous butanol at concentrations corresponding to B+ (40 mg/liter) and B++ (1 g/liter). Additionally, we included a NaNO₃-depleted sample (N−) (see Materials and Methods). This sample had no exogenous butanol added. Comparison of butanol-challenged samples to those starved of nitrogen would help to distinguish between general stress responses and butanol-specific stress responses. Cell culture, RNA purification, and data analyses were performed in biological duplicates.

mRNA-enriched RNA was isolated and analyzed with RNA-Seq (see Materials and Methods). A total of 97.5% of all annotated protein-coding Synechocystis genes were represented by at least 10 reads in at least one sample. Criteria for significant up- or downregulation were fold change (FC) ≥ 1.5 or FC ≤ 0.67 relative to normal conditions and P value ≤ 0.05, allowing a false discovery rate of 0.1.

Compared to a control sample with no n-butanol added, 78 genes were differentially expressed under the B+ conditions (Fig. 2; see also File S1 in the supplemental material). Under the B++ conditions, 276 genes, representing 8.3% of the open reading frames (ORFs) in the Synechocystis chromosomal genome, were affected (35). Nitrogen starvation affected 469 genes, but this response had little overlap with the butanol response: only 30% of genes differentially expressed in the butanol samples were also affected during nitrogen starvation (Fig. 2; see also File S1 in the supplemental material). These genes included downregulated PETC proteins such as cytochrome b6f and phycobilisome proteins.

Fig 2.

Numbers of differentially expressed genes in Synechocystis sp. strain PCC 6803 after butanol challenge or nitrogen starvation (12 h) compared to a no-butanol, nitrogen-replete sample. B+, 40 mg/liter; B++, 1 g/liter. The N⁻ sample was a Synechocystis strain lacking PHB synthase genes phaEC (see Materials and Methods). The criteria for significance were FC ≥ 1.5 and P ≤ 0.05.

The KEGG (Kyoto Encyclopedia of Genes and Genomes) database was used to assign differentially expressed KEGG Orthology (KO) groups (36). Most (∼65%) butanol-affected genes could not be classified as part of a certain KO group, as their function is unknown. Furthermore, many genes belong to several KO groups. The largest KO group among upregulated genes corresponded to photosynthesis. Upregulation of respiration-related proteins such as ubiquinone oxidoreductase during butanol stress has been reported for E. coli (31). However, several photosynthesis group genes, including those encoding enzymes for chlorohyll biosynthesis, were also downregulated (Fig. 3B). Taken together with the changes in fluorescence emission from photosynthetic pigments, these results suggest a damage and repair cycle for photosystem proteins stemming from impaired electron flow. A previous study on the hydrogen peroxide response of Synechocystis also reported contrasting responses among some photosystem II (PSII) components and NADH dehydrogenase subunits (37), indicating an overlap of the butanol and peroxide stress responses. The largest KO group among downregulated genes corresponded to amino acid metabolism; nucleotide and protein biosynthesis groups were also downregulated. Several ABC transporters and two-component-system genes were found in both up- and downregulated gene sets. A complete table of KEGG assignments is available in File S1 in the supplemental material.

Fig 3.

Distribution of differentially expressed genes into KEGG pathways. Only genes with annotated KEGG pathways (34/153 for upregulated genes, 65/137 for downregulated genes) are included. The remainder of the differentially expressed genes are categorized as hypothetical and are not assigned a KEGG pathway.

Effect of butanol on photosynthesis and oxidative stress-related genes.

The large group of photosynthesis genes upregulated upon butanol challenge includes plastocyanin, ferredoxins, and photosystem proteins, mainly from the PSII reaction center. Three ferredoxin variants were found upregulated in a butanol-dependent manner (Fig. 4). Recent proteomics studies of butanol and ethanol responses in Synechocystis sp. strain PCC 6803 reported upregulation of ferredoxin proteins and components of PSI and PSII (17, 38). This aspect of the butanol response is similar to reported UV-B or high-intensity white light responses which included upregulation of PSII genes, phycobilisome degradation genes, and plastocyanin (39, 40). The oxidative stress-related genes encoding IsiA (sll0247), superoxide dismutase (SOD; slr1516), thioredoxin A (slr0623), and glutathione peroxidase (slr1992) were upregulated approximately 2-fold at either B+ or B++ but did not meet the criteria for significant upregulation. Flavodoxin (isiB [sll0248]), previously reported as upregulated under iron deprivation conditions as a potential replacement for ferredoxin (41), was detected at low levels and was not affected by butanol stress. Transcription of the gene encoding the small heat shock protein HspA (sll1514) was induced 4.6-fold under the B+ conditions. This gene is known to be induced under conditions of temperature stress, osmotic stress (42), and hydrogen peroxide stress (37). In addition to its role in stabilizing misfolded proteins, HspA associates with thylakoid membranes to modulate membrane fluidity during heat shock of Synechocystis (43). A HspA homolog in Synechococcus was also found to be associated with membrane-bound photosystems and phycobilisomes during oxidative stress (44). The strong upregulation of the hspA gene during butanol stress is consistent with solvent-induced deterioration of periplasmic and thylakoid membranes, even at low butanol concentrations.

Fig 4.

Butanol-dependent stress response. (A) Fold change of genes differentially expressed under either B+ or B++ conditions relative to no-butanol conditions. (B) Selected significantly affected genes that showed a strong butanol-dependent response. Gene identifications: copM, sll0788; pilA1, sll1694; nblA1, ssl0452; rfrP, slr0967; cccS, slr1667; fdx I, ssl0020; fdx II, sll1382; fdx III, slr1828; fdx IV, slr0150; pntB, slr1434; rpoA, sll1818; hypothetical protein, slr2052.

Effect of butanol on transporters and two-component systems.

Most of the transporters and two-component systems that were induced upon butanol challenge mediate metal homeostasis, and the data suggest severe perturbation of levels of intracellular iron and copper, possibly through leakage from membrane-bound proteins. Photosynthetic organisms such as cyanobacteria are heavily dependent on iron and copper for photosynthetic and respiratory oxidases and photosynthetic electron transfer proteins (plastocyanin). The three ABC transporter genes that were significantly upregulated by butanol stress were all involved in iron transport. Expression of two of these, futA1 and futA2 (slr1295 and slr0513), has also been reported to be induced by hydrogen peroxide stress (45). Iron homeostasis is perturbed by the presence of ROS such as superoxide and H₂O₂, where the latter can oxidize ferrous iron and release damaging peroxide radicals through the Fenton reaction (46). Induction of iron transport proteins may therefore signify a change of intracellular ferrous iron levels through heme-protein degradation, membrane leakage, or oxidizing reactions. Expression of both components of the copper-sensing and regulator pair Hik31 and Rre34 (sll0790 and sll0789) was induced over 4-fold under the B+ and B++ conditions. The induction of expression of this operon was likely due to degradation of the electron carrier plastocyanin and subsequent release of bound copper. A similar upregulation was previously observed upon addition of DBMIB (2,5-dibromo-3-methyl-6-isopropyl-p-benzoquinone), which blocks electron transfer to PC and oxidizes the PC pool (47).

Three of the upregulated two-component-system genes encoded pilus assembly proteins. Two of them, pilA1 (sll1694) and pilA4 (slr1456), were also upregulated over 4-fold by benzyl alcohol (48). These proteins could possibly play a role in the cell aggregation seen in cultures containing high butanol concentrations. Additionally, PilA1 has been found to interact with chlorophyll and has been implicated in delivering newly synthesized chlorophyll to photosystem proteins (49).

A large portion of the downregulated transporters and two-component-system gene products were related to biosynthesis and nitrogen metabolism, such as nitrate transport proteins NrtA, NrtB, and NrtC and glutamate-ammonia ligases GlnA and GlnN. Other downregulated members of two-component systems were genes encoding chemotaxis proteins and twitching-motility proteins.

Stress response correlates with butanol concentration.

One motivation for including two butanol concentrations in this study was to identify genes that are affected in a butanol dosage-dependent manner. We observed that among the 291 genes affected at either B+ or B++, the B++ response was generally stronger (Fig. 4A). However, of the 78 genes affected at a low butanol concentration (B+), few showed further induction or repression at a higher butanol concentration (B++) (Fig. 4B); instead, the cell response was much more diverse at B++ (Fig. 2).

Dosage-dependent regulation was defined as genes affected at B+ that also showed an additional 1.5-fold change under the B++ conditions. Genes meeting these criteria include those encoding parts of the copper-sensing regulon CopMRS (sll0788 and sll0789), the phycobilisome degradation protein NblA1 (ssl0452), the pilin-like protein PilA1 (slr1456), two ferredoxins (Fdx I [ssl0020] and Fdx III [slr1828]), and two hypothetical proteins (slr1667 and slr0967) (Fig. 4B). The genes encoding phycobilisome protein NblA have previously been reported as upregulated in the presence of H₂O₂, i.e., under oxidative stress conditions, and appear to be partly regulated by Hik16 and PerR (45), which were also upregulated in the butanol samples. The hypothetical proteins, together with other hypothetical proteins strongly affected by butanol stress, are described in Table 1.

Table 1.

Hypothetical genes up- or downregulated more than 3.5-fold and significantly differentially expressed in B+ and B++

Gene	Symbol	Fold change				Information from literature
		B+/B−	B++/B−	B−/B+	B−/B++
Upregulated hypothetical genes
sll0788	copM	6.2 ± 2.9	12.3 ± 5.0			Upregulated by copper; under the control of Hik31-Rre34 (47)
ssr0692		5.7 ± 1.6	5.3 ± 1.9			Upregulated by CO2 limitation (50) and benzyl alcohol stress (48)
slr0967	rfrP	2.2 ± 0.3	4.6 ± 0.7			Salt and hyperosmotic stress-inducible gene (51)
slr1667	cccS	2.0 ± 0.3	3.9 ± 0.6			Involved in construction of cell surface components and formation of thick pili (52)
ssr2062		3.8 ± 0.2	2.9 ± 0.8			Upregulated in LexA-depleted mutant (53)
slr1908		2.8 ± 0.8	3.7 ± 1.0			Probable porin detected in the outer and plasma membranes (54)
Downregulated hypothetical genes
sll0783				13.8 ± 14	10.4 ± 5.8	Upregulated upon nitrogen starvation; necessary for PHB production (55)
slr1593	ylmD			6.5 ± 1.9	4.4 ± 2.0	Dispensable cell-division protein (56)
sll1049				3.6 ± 0.5	2.5 ± 0.4	Contains transglutaminase-like domain; possible glutaredoxin target (57)
slr0955				3.3 ± 0.4	4.2 ± 1.9	Probable tRNA/rRNA methyltransferase induced by cold (58)
slr0151				2.8 ± 0.6	4.0 ± 0.9	Putative photosystem II assembly protein (59)
slr2052				2.3 ± 0.2	4.1 ± 1.2	Downregulated by UV-B or high-intensity white light irradiation (40)

Dosage-dependent downregulated genes include genes encoding a transhydrogenase subunit, PntB (slr1434; subunit PntA was also downregulated under butanol stress conditions), a ferredoxin (Fdx IV slr0150), and RNA polymerase subunit alpha (sll1818). The transhydrogenase enzyme could potentially catalyze the interconversion of NADPH and NADH and thus influence redox balance in the cell. However, to our knowledge, no significant transhydrogenase activity has been confirmed in this cyanobacterium (60).

Effect of butanol on hypothetical proteins.

The majority of the differentially expressed genes in the butanol samples were not annotated as part of a certain pathway in the KEGG database and were classified as hypothetical in CyanoBase. However, some of these genes have been mentioned in other stress response or biochemical characterization studies. Hypothetical genes showing strongly differential levels of expression in B+ and B++ are most relevant to the butanol response and are summarized in Table 1, together with the suggested functions derived from the literature. Their strong regulation in the presence of butanol may help to elucidate their functions.

Targets for engineering solvent tolerance.

Our physiological and transcriptomics data describe a cyanobacterial cell with a compromised membrane, accumulated ROS, and altered metal homeostasis in the presence of butanol. We hypothesized that overexpression of key genes could improve growth rates in the presence of butanol and viability after butanol challenge. We selected genes with the aim of restoring membrane integrity and reducing oxidative stress. An additional criterion was gene upregulation in the RNA-Seq data set. Genes selected for overexpression are listed in Table 2.

Table 2.

Proteins selected for overexpression in Synechocystis and their induction under conditions of butanol stress

Gene	Symbol	Fold change		Abundance in WT (RPKMb)	Expression from pJA2 (factor over WT)a	Function
		B+/B−	B++/B−
sll1514	hspA	4.6 ± 1.7	3.0 ± 0.6	400	10	Stabilizes thylakoid membranes (43)
slr1828	fdxIII	2.4 ± 1.5	6.2 ± 2.2	10	16	Low-abundance ferredoxin could alleviate overreduced PETC
slr1516	sodB	2.1 ± 0.3	1.8 ± 0.4	400	7	Protects against oxidative stress
slr0623	trxA	1.6 ± 0.1	1.3 ± 0.2	380	6	Noniron electron shuttle protein
sll0755	tpx	0.6 ± 0.2	0.7 ± 0.2	70	9	Peroxidase activity when encoding gene is paired with trxA (61)
slr1667	cccS	2.0 ± 0.3	3.9 ± 0.6	5,200	18	Construction of cell surface proteins (52)
ssr0692	ssr0692	5.7 ± 1.6	5.3 ± 1.9	1,600	7	Unknown; interacts with NDH subunit H (62)

^aGenes were cloned into a low-copy-number replicating plasmid, pJA2, and expressed under the control of the P_psbA2 promoter. Overexpression was measured with quantitative PCR (qPCR) (see Materials and Methods).

^bWT, wild type; RPKM, normalized reads per kilobase per million mapped reads.

The small heat shock protein HspA (sll1514) and the hypothetical protein cccS (slr1617) were chosen due to their reported roles in thylakoid membrane stabilization (43) and cell surface construction (52), respectively. Three redox-related proteins were selected with reference to the accumulated ROS during butanol shock. Superoxide dismutase (SOD) is critical to oxidative stress tolerance in cyanobacteria (63). The low-abundance ferredoxin (slr1828) and ssr0962, encoding a hypothetical protein, were also strongly affected by butanol. This hypothetical protein was found to interact with an NADH dehydrogenase subunit in a two-hybrid system (62) and may therefore be associated with the electron transport chain. Finally, we included the thioredoxin A (TrxA [slr0623]) and 2-Cys peroxidase (Tpx [sll0755]) pair, although these genes were not upregulated under butanol stress conditions. This pair has been reported to catalyze the reduction of H₂O₂ in vitro (61). Furthermore, a peroxiredoxin was recently shown to reduce accumulated ROS in the cyanobacterium Anabaena sp. strain PCC 7120 (64). All genes were inserted into a low-copy-number, replicative plasmid, pJA2, under the control of the light-induced P_psbA2 promoter. Overexpression of genes relative to the wild type was confirmed with quantitative PCR (Table 2).

Butanol tolerance of the mutant strains was first assessed via growth during an extended cultivation at 4 g/liter n-butanol or isobutanol. Isobutanol was included to assess the generality of solvent protection mechanisms. At this concentration, the wild type shows an approximately 50% reduction in growth rate. The pJA2-hspA, pJA2-ssr0692, and pJA2-cccS strains showed improved growth during the extended cultivation at 4 g/liter butanol and isobutanol relative to wild type (Fig. 5A). In particular, the pJA2-hspA strain had accumulated more biomass at the end of the cultivation period in n-butanol (Fig. 5B). The pJA2-sodB strain showed improved growth relative to the wild type in the presence of n-butanol but did not grow significantly faster in the presence of isobutanol. The improved growth of the pJA2-sodB and pJA2-ssr0692 strains in n-butanol may be due to the ability of these proteins to alleviate the accumulation of oxidative stress during long exposure to butanol. The Synechocystis sodB has been shown to increase tolerance to methyl viologen when expressed heterologously in E. coli (65).

Fig 5.

Overexpression of cyanobacterial proteins improves growth rate and viability in the presence of butanol. Selected genes were overexpressed from the low-copy-number replicating pJA2 vector under the control of the P_psbA2 promoter (see Materials and Methods). The pJA2 sample was empty vector. For gene details, see Table 2. (A) Growth rates of Synechocystis strains in the presence of 4 g/liter n-butanol or isobutanol. The dashed line indicates the growth rate of the wild type at a rate of 0 g/liter butanol. (B) Growth curves for wild-type and pJA2-hspA strains in the absence and presence of butanol. (C) Viability of Synechocystis strains in response to butanol shock (20 g/liter, 1 h). Relative CFU data are relative to each strain incubated with 0 g/liter butanol.

The pJA2 strains were also assayed for viability after exposure to a short, high-concentration n-butanol shock. Cells were treated with 20 g/liter n-butanol for 1 h before plating on agar plates was performed. Viability was defined according to CFU levels in reference to cells not treated with butanol. Wild-type Synechocystis was viable at <2% after this treatment (Fig. 5C). Of the tested strains, pJA2-hspA and pJA2-cccS showed viability that was improved over 10-fold at 20 g/liter n-butanol relative to the wild-type results. However, no strains were viable after a 4-h treatment at 20 g/liter n-butanol.

The improved tolerance of the pJA2-hspA strain under conditions of extended cultivation and short butanol shock is consistent with reports that overexpression of heat shock proteins can improve solvent tolerance severalfold in E. coli (20), Lactobacillus (66), and Clostridium (21). HspA belongs to one of several heat shock protein families that are induced in response to changes in membrane fluidity in cyanobacteria (67). HspA confers resistance to temperature and UV shock in several cyanobacteria species through a mechanism involving membrane insertion and stabilization (43), which likely explains its beneficial effect on solvent resistance.

The pJA2-cccS strain may also have a more robust cell membrane that can better withstand the high-butanol-concentration shock. The cccS gene product localizes at the cell membrane and contributes to cell surface construction of membrane proteins and pilins (52).

We note that the response to exogenous butanol addition may differ from the response to internally produced butanol. Small hydrophobic compounds such as ethanol and butanol are assumed to diffuse passively across the cell membrane (68) so that internal and external concentrations are rapidly equalized. However, a recent study found that E. coli was more sensitive to internally produced medium-chain (C₈-C₁₄) fatty acids than to exogenous fatty acids. This may be due to an inability of the E. coli to efficiently import and export these larger products (69).

Conclusions.

The results from our performed RNA-Seq analysis showed unexpectedly little similarity to the results from a previous proteomics study of the butanol stress response in Synechocystis (17). The poor correlation could possibly be explained to at least some extent by three significant differences in the experimental execution, namely, differing butanol incubation times (12 h in our study compared to 24 h and 48 h) and butanol concentrations (40 mg/liter and 1 g/liter in our study compared to 1.6 g/liter) and analysis of transcripts versus proteins in the previous study. For instance, transcription of the psbA2 and psbA3 genes encoding the photosystem II reaction center D1 subunit has been shown to have been upregulated upon UV-B irradiation whereas the D1 protein level was decreased, confirming that differential expression results on the transcript level do not always correlate with protein expression levels (39).

Our results show that the cyanobacterium cell responds strongly to butanol at low concentrations. Upregulation of genes encoding cell surface proteins, metal homeostasis regulators, and several electron shuttle proteins suggests membrane damage, metal leakage, and impaired photosynthetic electron transfer. Overexpression of oxidative stress proteins such as SodB and of membrane stabilization proteins such as HspA and CccS can improve solvent tolerance. Strains with stabilized membranes and oxidative stress tolerance are likely to be more tolerant to a range of solvents. Such transcriptomics-informed engineering could lead to the creation of a cyanobacterium that is more tolerant to its product and a better solvent producer.