Polyamine Transporters and Polyamines Increase Furfural Tolerance during Xylose Fermentation with Ethanologenic Escherichia coli Strain LY180

Abstract

Expression of genes encoding polyamine transporters from plasmids and polyamine supplements increased furfural tolerance (growth and ethanol production) in ethanologenic Escherichia coli LY180 (in AM1 mineral salts medium containing xylose). This represents a new approach to increase furfural tolerance and may be useful for other organisms. Microarray comparisons of two furfural-resistant mutants (EMFR9 and EMFR35) provided initial evidence for the importance of polyamine transporters. Each mutant contained a single polyamine transporter gene that was upregulated over 100-fold (microarrays) compared to that in the parent LY180, as well as a mutation that silenced the expression of yqhD. Based on these genetic changes, furfural tolerance was substantially reconstructed in the parent, LY180. Deletion of potE in EMFR9 lowered furfural tolerance to that of the parent. Deletion of potE and puuP in LY180 also decreased furfural tolerance, indicating functional importance of the native genes. Of the 8 polyamine transporters (18 genes) cloned and tested, half were beneficial for furfural tolerance (PotE, PuuP, PlaP, and PotABCD). Supplementing AM1 mineral salts medium with individual polyamines (agmatine, putrescine, and cadaverine) also increased furfural tolerance but to a smaller extent. In pH-controlled fermentations, polyamine transporter plasmids were shown to promote the metabolism of furfural and substantially reduce the time required to complete xylose fermentation. This increase in furfural tolerance is proposed to result from polyamine binding to negatively charged cellular constituents such as nucleic acids and phospholipids, providing protection from damage by furfural.

INTRODUCTION

Lignocellulosic biomass can be used as a renewable carbohydrate feedstock for the production of fuels and chemicals (1). Unlike starch, which is used by plants for temporary energy storage, lignocellulose is a structural component designed to resist microbial enzymes and chemical deconstruction. A pretreatment step such as dilute acid hydrolysis is required to increase the accessibility of cellulase enzymes for more efficient carbohydrate hydrolysis (2).

Dilute acid pretreatments have been widely investigated. Such pretreatments effectively open the lignocellulose structure by hydrolyzing hemicellulose into pentose monomers. Dilute acid pretreatments also produce unwanted side products, such as furfural and hydroxymethylfurfural (HMF), that retard fermentation and increase process cost (3–7). Increasing concentrations of furans have been correlated with toxicity of hydrolysates. Addition of furfural to purified hydrolysates has been shown to restore toxicity (8, 9).

Genes have been described that increase the resistance of bacterial and yeast biocatalysts to furfural and hydroxymethylfurfural (5, 7, 10–20). However, to date, none have completely solved the toxicity problem. In Saccharomyces cerevisiae, overexpression of the transcription factor Yap1 has been shown to increase tolerance to both furfural and HMF (17). Disruption of pentose phosphate pathway (PPP) genes (ZWF1, GND1, RPE1, and TKL1) led to increased sensitivity to furfural and HMF in S. cerevisiae (10). Disruption of PPP genes is proposed to decrease the levels of NADPH, the cofactor for many enzymes involved in cellular defense mechanisms and for biosynthesis (10). A mutated form of the alcohol dehydrogenase ADH1 from S. cerevisiae was shown to confer tolerance by reducing HMF to 2,5-bis-hydroxymethylfuran using NADH instead of NADPH (18).

Microarray analysis of oxidoreductases in a furfural-resistant mutant of Escherichia coli (strain EMFR9) revealed that the NADPH-dependent YqhD is involved in furan tolerance (19). Native expression of chromosomal yqhD was found to inhibit the growth of E. coli in media containing furfural (6, 19). Blocking the functional expression of yqhD increased resistance to furfural. The detrimental effect of YqhD (low K_m for NADPH) has been attributed to competition for NADPH, limiting biosynthesis. In contrast, increased expression of an NADH-dependent furfural reductase gene (fucO) was beneficial for furfural tolerance (20). Both enzymes reduce furfural (aldehyde) to the less toxic furfuryl alcohol. Recently, a new genomic tool, multiscale analysis of library enrichments (SCALE), was developed and identified four genes in E. coli that increase furfural tolerance (15): thyA (DNA biosynthesis), lpcA (lipopolysaccharide biosynthesis), groES, and groEL (GroES-EL chaperonin complex). Increased expression of the groES, groEL, grpE, and clpB genes (21, 22) has also been shown to increase ethanol tolerance.

The precise mechanism of furfural toxicity is unknown. Although methods have been developed to mitigate or remove inhibitors (23), all of them increase process complexity. Furfural has been shown to cause strand breaks in DNA (24), to damage membranes (25), and to react with other cellular components (10, 26).

Polyamines can bind both nucleic acids and phospholipid membranes, reducing exposure of these vital cellular constituents to furfural. In E. coli, intracellular levels of polyamines are regulated by biosynthesis, degradation, and expression of transporter genes (27). In this paper, we report that plasmids containing polyamine transporter genes and polyamine supplements can be used to increase furfural tolerance in ethanologenic E. coli LY180.

MATERIALS AND METHODS

Strains, media, and growth conditions.

Table 1 contains a list of strains, plasmids, and primer sequences used in this study. Table S1 in the supplemental material lists primers used for reverse transcription-PCRs (RT-PCRs) and for sequencing genes cloned into plasmids. Ethanologenic E. coli strain LY180 (E. coli W derivative [19]) and derivatives were used to investigate furfural tolerance for growth and ethanol production. A furfural-resistant mutant of LY180 was previously isolated after 53 serial transfers of cultures in the presence of furfural and designated EMFR9 (19). A second strain, EMFR35, was subsequently isolated from the same enrichment culture after 107 transfers. These two mutants contained the same IS10 insertion in yqhC, silencing expression of yqhD (30).

TABLE 1.

Bacterial strains, plasmids, and primers used in this study

Strain, plasmid, or primer use	Relevant characteristic(s)	Reference or source
E. coli strains
E. coli W (ATCC 9637)	Wild type	ATCC
LY180	ΔfrdBC::(frgZm celYEc) ΔldhA::(frgZm casABKo) adhE::(frgZm estZPp FRT) ΔackA::FRT rrlE::(pdc adhA adhB FRT) ΔmgsA::FRT	19
RG100	LY180 ΔpuuP	This study
RG101	LY180 ΔpotE	This study
RG102	LY180 ΔpuuP ΔpotE	This study
RG105	EMFR9 ΔpotE	This study
EMFR9	LY180 furfural-resistant mutant (yqhD silenced)	19
EMFR35	LY180 furfural-resistant mutant (yqhD silenced)	This study
XW092	LY180 ΔyqhD	13
TOP10F′	F′[lacIq Tn10 (Tetr)] mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 recA1 araD139 Δ(ara-leu)7697 galU galK rpsL endA1 nupG	Invitrogen (Carlsbad, CA)
Plasmids
pLOI5249	potE gene in NcoI-BamHI-digested pTrc99A	This study
pLOI5408	potABCD gene in AvaI-XbaI-digested pTrc99A	This study
pLOI5409	potFGHI gene in EcoRI-XbaI-digested pTrc99A	This study
pLOI5410	ydcSTUV gene in EcoRI-XbaI-digested pTrc99A	This study
pLOI5411	plaP gene in EcoRI-XbaI-digested pTrc99A	This study
pLOI5412	puuP gene in EcoRI-XbaI-digested pTrc99A	This study
pLOI5414	mdtJI gene in EcoRI-XbaI-digested pTrc99A	This study
pLOI5415	cadB gene in EcoRI-XbaI-digested pTrc99A	This study
pKD4	bla FRT-kan-FRT	28
pKD46	bla γ β exo (Red recombinase), temperature-conditional replicon	28
pCP20	FLP+ λ cI857+ λ pR Repts bla catF	29
pTrc99A	pTrc bla oriR rrnB lacIq	Laboratory collection
Primer uses
potABCD cloning	For: CCCCCGGGCAAGGTGGTTAACCACAAACC	This study
	Rev: GCTCTAGACGAATTGAAAATTAGCGTGTAA
potFGHI cloning	For: CGGAATTCGTTAACGAACTTTCAGAAGGAA	This study
	Rev: GCTCTAGAATTTGTGTCAGCAGATATAGCCA
ydcSTUV cloning	For: CGGAATTCGAACAATTAATTACGACAGGAGTAAG	This study
	Rev: GCTCTAGACAGCGGTTTGCCACAATTAC
plaP cloning	For: CGGAATTCGCGACGGTTATCACCGTAAA	This study
	Rev: GCTCTAGATGCGATTATTTTTCGCGAGA
potE cloning	For: CATGCCATGGAACCTGTTGCCAGGTTTTGCA	This study
	Rev: CGGGATCCAGCTTCCTCGGTGAAGAACA
puuP cloning	For: CGGAATTCCAAACCTTATTACGCAGGGGAG	This study
	Rev: GCTCTAGACATGTTGGGCTTCTTCGCTG
mdtJI cloning	For: CGGAATTCACTTTGGTTTCGCTGAATTAAG	This study
	Rev: GCTCTAGAAGGCGGGATATCCTGAAGAT
cadB cloning	For: CGGAATTCTGACCCGGACTCCAAATTCAA	This study
	Rev: GCTCTAGAACAACGGCAGGTTCTCGTTCA
Deletion of puuP	For: TTTCAGGTCGACACGACCGCAAACCTTATTACGCAGGGGAGGCAGCAATTGTGTAGGCTGGAGCTGCTTC	This study
	Rev: TGGCGCGGCGCATTACCCTCAGGCAGGATAATGCGGCGCGCATCCGACTACATATGAATATCCTCCTTAGT
Deletion of potE	For: GTCATCAAGCCTCGTGATGCGCAAAGCACCCTGTTGAAAGGGGAAAAATTGTGTAGGCTGGAGCTGCTTC	This study
	Rev: TTTAATAAAAAAAGGGCGGTCGCAAGATCGCCCTTTTTTACTTTGCTTTTCATATGAATATCCTCCTTAGT

Strains and pTrc99A-based plasmids were constructed using Luria-Bertani medium. After construction, cultures were grown in AM1 mineral salts medium (31). All media were supplemented with xylose (20 g liter⁻¹ for solid medium, 50 g liter⁻¹ for broth cultures, and 100 g liter⁻¹ for pH-controlled fermentations). Ampicillin (50 mg liter⁻¹) and isopropyl-β-d-thiogalactopyranoside (IPTG; 0.1 mM) were added as needed. Stocks of these were prepared in 70% ethanol. Chromosomal deletions of puuP and potE were constructed using the method of Datsenko and Wanner (28). Constructions were confirmed by sequencing, PCR analysis, and phenotype.

Furfural tolerance.

Tolerance was examined by measuring growth and ethanol production after 48 h (37°C) using tube cultures (13 mm by 100 mm) containing 4 ml of AM1 medium. Ampicillin (12.5 mg liter⁻¹), furfural, IPTG, and other supplements were added as indicated. Inocula were grown overnight on AM1 xylose plates (solid medium). Fresh colonies were scraped, resuspended in medium, and adjusted to an optical density at 550 nm (OD₅₅₀) of 1.0. Tube cultures were inoculated to an initial OD₅₅₀ of 0.1 (43 mg [dry cell weight {dcw}] liter⁻¹) and incubated in a reciprocating water bath (50 oscillations min⁻¹).

Construction of polyamine transporter plasmids.

Genes encoding PotE and PuuP were amplified (including ribosomal binding site and terminator region) from strain E. coli W (ATCC 9637) chromosomal DNA by using PCR. These fragments were cloned into the NcoI and BamHI sites of pTrc99A to produce pLOI5249 and pLOI5412, respectively. Genes encoding six other polyamine transporters were also cloned in a similar manner using primers with flanking restriction sites (Table 1). After ligation, plasmids were transformed into E. coli TOP10F′. Plasmids were purified using a QIAspin Spin Miniprep kit (Qiagen, Valencia, CA). Clones were verified by digestion with restriction enzymes, gel analysis of PCR products, and sequencing.

Data and analyses for tube experiments.

Cell mass was measured as OD₅₅₀ using a Bausch & Lomb Spectronic 70 spectrophotometer. An OD₅₅₀ of 1 is equivalent to 430 mg (dcw) liter⁻¹. Ethanol was measured using an Agilent 6890N gas chromatograph (Palo Alto, CA) equipped with flame ionization detectors and a 15-m HP-Plot Q Megabore column. Data presented are averages for three or more experiments with standard deviations. Note that the presence of ampicillin and plasmids caused a small reduction in growth and ethanol production.

Fermentations.

The effect of furfural addition on pH-controlled fermentations was investigated as described previously (19) using AM1 medium containing 100 g liter⁻¹ of xylose. Ethanol was measured by gas-liquid chromatography. Cell mass was estimated from measurements of absorbance at 550 nm. Furfural was measured using a Beckman Coulter DU 800 spectrophotometer (32). Results represent averages of three or more experiments, with standard deviations.

Genome sequencing.

Genomic DNA samples from LY180, EMFR9, and EMFR35 were purified according to the bacterial genomic DNA isolation protocol from the DOE Joint Genome Institute (http://jgi.doe.gov). Next-generation sequencing was performed using Illumina paired-end short read technology provided by the Tufts University Core Facility (Boston, MA). Sequence data for LY180 were assembled using Geneious software (Auckland, New Zealand). Sequencing of EMFR9 and EMFR35 has not been completed. Genome annotation of LY180 was provided by the Prokaryotic Genomes Automatic Annotation Pipeline (PGAAP) of the National Center for Biotechnology Information (NCBI).

Microarray analysis.

Expression of mRNA was analyzed as previously described (19, 30). For each data set, 4 cultures were grown separately in 500-ml fermentors (350-ml working volume) and sampled at an OD₅₅₀ of 1.5 (650 mg [dcw] liter⁻¹) prior to furfural addition. Furfural was then added (5 mM for EMFR9 and LY180 control and 15 mM for EMFR35 and LY180). Cultures were sampled again after 15 min of incubation with furfural. Each sample was processed separately for RNA extraction. RNA from each set of 4 samples was pooled and submitted to Roche NimbleGen for microarray analyses. Data from NimbleGen were imported into ArrayStar (DNA Star) for analysis. Each experiment was performed twice and averaged. Expression results for each gene are reported as the arithmetic ratio of mutant transcripts divided by the parent LY180.

RT-PCR methods.

RNA was isolated (Qiagen RNeasy Protect Bacteria minikit) from cultures grown to an OD₅₅₀ of 1.5 (650 mg [dcw] liter⁻¹) as described in the previous section. Residual genomic DNA was removed by DNase treatment using the Ambion Turbo DNA-free kit (Life Technologies, Grand Island, NY). Real-time RT-PCR was performed in 25-μl reaction mixtures, each containing 125 ng of RNA (Qiagen QuantiTect SYBR green RT-PCR kit). Primer sequences for RT-PCR analysis of polA, potE, and puuP are listed in Table S1 in the supplemental material. The number of cycles to reach the threshold (C_T value) was measured for each primer pair in triplicate. Using expression of polA as a reference for comparisons among strains, fold differences were calculated by dividing expression in the mutant by expression in the parent. Results represent an average of 3 reactions, with standard deviations.

Nucleotide sequence and microarray data accession numbers.

The fully assembled genome of LY180 has been deposited in NCBI's GenBank (http://www.ncbi.nlm.nih.gov/genbank) under accession number CP006584. Microarray data for gene expression in LY180, EMFR9, and EMFR35 have been deposited in NCBI's Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) with GEO series accession numbers GSE17786 and GSE46442.

RESULTS

Isolation of furfural-resistant mutants.

Two furfural-resistant mutants of ethanologenic LY180 were isolated after 53 and 107 serial transfers in minimal medium containing furfural, EMFR9 (19) and EMFR35 (this study), respectively. Furfural tolerance of these strains was compared by measuring growth and ethanol production in tube cultures (Fig. 1A and B). For each strain, curves for growth and ethanol production exhibited very similar trends as a function of furfural concentration. The parent, LY180, was the most sensitive to furfural. Furfural tolerance was highest for EMFR35, followed by EMFR9. For these strains, the MICs (Fig. 1C) of furfural were 20.0 mM (EMFR35), 16 mM (EMFR9), and 12.5 mM (LY180).

FIG 1.

Comparison of furfural tolerances in the parent LY180 and mutants isolated from serial transfers in AM1 containing furfural. EMFR9 was isolated after 53 serial transfers in AM1 medium containing xylose (100 g liter⁻¹) and furfural. EMFR35 was isolated from a later stage of the same culture but after 107 serial transfers. (A) Growth in different concentrations of furfural; (B) ethanol production in different concentrations of furfural; (C) summary comparison of furfural tolerances (MIC) among strains and constructs. Genetic traits conferring furfural tolerance included plasmid-based expression of polyamine transporters potE and puuP (pLOI5249 and pLOI5412, respectively) and deletion of oxidoreductase gen yqhD from the parent, LY180, producing strain XW092. Constructs with a single genetic trait were more resistant to furfural than the parent containing empty vector and more sensitive to furfural than constructs carrying two traits.

Upregulation of polyamine transporters in furfural-resistant mutants.

Microarray studies of oxidoreductase expression in EMFR9 previously identified the silencing of yqhD (NADP-dependent furfural reductase) as an important mutation for furfural tolerance (19, 30). This mutation was present in both EMFR9 and EMFR35, consistent with a common ancestor. Further investigation of the EMFR9 microarray data revealed a 100-fold upregulation of potE expression (putrescine uptake) relative to that in the parent, LY180 (Table 2). This was the largest increase among all genes in this strain. Although E. coli contains 8 polyamine transporters encoded by 18 genes, only the expression of potE exhibited a large expression increase in EMFR9 relative to LY180 (Table 2). Upregulation of potE in EMFR9 was confirmed using RT-PCR (27-fold upregulation) (see Table S2 in the supplemental material). Addition of furfural did not affect potE expression (Table 2) in EMFR9 compared to LY180 (parent).

TABLE 2.

Microarray expression ratios comparing the relative expressions of polyamine transporter genes in furfural-resistant mutants (EMFR9 and EMFR35) and the parent (LY180)^a

Transporter	Gene	Locus tagb	Gene expression ratio (mutant/LY180)
			EMFR9 (0 mM furfural)	EMFR9 (5 mM furfural)	EMFR35 (0 mM furfural)	EMFR35 (15 mM furfural)
ABC transporter, spermidine/putrescine	potA	b1126	0.8	1.0	0.8	2.2
	potB	b1125	1.1	1.3	0.7	3.0
	potC	b1124	0.9	1.1	0.5	3.1
	potD	b1123	0.8	1.3	1.2	1.0
ABC transporter, putrescine	potF	b0854	1.0	0.7	0.9	1.4
	potG	b0855	1.0	0.7	1.0	1.2
	potH	b0856	0.8	0.8	1.2	1.1
	potI	b0857	0.8	0.9	1.3	1.2
ABC transporter, spermidine/putrescine	ydcS	b1440	1.0	1.0	1.3	1.1
	ydcT	b1441	1.0	1.3	1.5	1.6
	ydcU	b1442	0.9	1.2	0.9	1.4
	ydcV	b1443	0.9	1.1	1.0	1.2
Putrescine symporter	plaP	b2014	2.4	1.9	1.7	10
Putrescine symporter	potE	b0692	120	100	1.0	0.8
Putrescine symporter	puuP	b1296	0.9	1.1	750	590
Spermidine antiporter	mdtJ	b4132	1.3	1.1	1.0	0.3
	mdtI	b1599	1.7	1.5	1.0	1.9
Cadaverine symporter	cadB	b4132	0.6	1.4	0.8	0.7

^aComparisons were made using cells grown in the presence and absence of furfural. Two genes that are highly upregulated are shown in bold.

^bLocus tags are based on E. coli K-12 strain MG1655 classification (GenBank accession number NC_000913.3).

EMFR35 was isolated after 107 serial transfers in AM1 medium containing furfural, a continuation of the same enrichment culture as used to isolate EMFR9 (53 transfers). Unlike for EMFR9, expression of potE was not upregulated in EMFR35 relative to the parent, LY180 (microarray comparison). However, a different polyamine transporter (puuP) was expressed at 700-fold-higher levels in EMFR35 than in the parent (Table 2). This unusually large upregulation of puuP in EMFR35 was confirmed by RT-PCR using polA expression as a reference (see Table S2 in the supplemental material). RT-PCR measurement confirmed that puuP was highly expressed in EMFR35, over 1,500-fold higher than in the parent.

Smaller changes were also noted in EMFR35 for two other polyamine transporters (Table 2), a 10-fold increase in plaP expression and a potential furfural-induced increase in expression of the potABCD operon. Expression ratios for genes encoding other polyamine transporters in EMFR9 and EMFR35 were low, with an average of 1.1 relative to LY180. In each furfural resistant mutant (EMFR9 and EMFR35), a single polyamine transport gene exhibited the largest increase in expression among all chromosomal genes relative to LY180.

Plasmids containing polyamine transporters increase furfural tolerance in LY180.

The unusually high expression of a single polyamine transporter in each furfural-resistant mutant suggests that these transporters may contribute to furfural resistance. To test this hypothesis, all 8 polyamine transporters in E. coli were cloned into pTrc99A and transformed into LY180. LY180 containing empty vector served as a control. The largest differences in growth and ethanol production were observed with 10 mM furfural for potE and puuP (Fig. 2). At this furfural concentration, cell mass and ethanol were 4-fold higher with pLOI5249 (PotE) and pLOI5412 (PuuP) than with LY180 containing empty vector.

FIG 2.

Effects of polyamine transporter plasmids on furfural tolerance in LY180. Plasmids containing potE (pLOI5249) or puuP (pLOI5412) increased the MIC for growth and ethanol production. LY180 containing empty vector (pTrc99a) served as a control. (A) Growth in different concentrations of furfural; (B) ethanol production in different concentrations of furfural; (C) growth of LY180 containing plasmids with each of the 8 polyamine transporters (10 mM furfural). Four of the 8 polyamine transporters were beneficial for growth in the presence of 10 mM furfural (without addition of IPTG). Addition of 0.1 mM IPTG decreased furfural tolerance of all constructs containing polyamine transporters on plasmids. Note that IPTG caused a small increase in growth of LY180 containing empty vector (pTrc99A).

Figure 2C shows a comparison of furfural tolerances (10 mM) for LY180 strains containing individual transporter plasmids. Four transporter plasmids more than doubled the growth of LY180 (empty vector control) in the presence of 10 mM furfural. Three of the beneficial transporters (PotE, PuuP, and PlaP) are single-gene proton symports for putrescine uptake (33). The fourth, potABCD, encodes an ATP-dependent ABC transporter for spermidine and putrescine (33). The four beneficial transporters increased growth relative to the empty-vector control in the following order: PuuP > PotE > PlaP > PotABCD. Half of the transporter plasmids (PotFGHI, YdcSTUV, CadB, and MdtJI) did not affect furfural tolerance. The two transporters that were highly expressed in furfural-resistant mutants based on RT-PCR and microarrays (potE in EMFR9 and puuP in EMFR35) were also the most beneficial for furfural tolerance in LY180 when expressed from plasmids pLOI5249 and pLOI5412, respectively. With all polyamine transporters, leaky expression of polyamine transporter genes from pTrc99A plasmids was sufficient to confer an increase in furfural tolerance (Fig. 2C).

Addition of IPTG decreased furfural tolerance with all transporter constructs (Fig. 2C) but caused a small increase in furfural tolerance in the empty-vector control. Addition of IPTG decreased the growth of all strains carrying cloned transporter genes even in the absence of furfural (see Fig. S1 in the supplemental material). Similar detrimental effects with inducer have been observed for several E. coli genes, including pntAB (6) and trehalose biosynthetic genes (34).

Deleting chromosomal potE and puuP in LY180 decreased furfural tolerance.

The importance of native puuP and potE genes was examined by constructing deletions in LY180, designated RG100 (ΔpuuP), RG101 (ΔpotE), and RG102 (ΔpuuP ΔpotE). Furfural tolerance in these strains was compared by measuring growth with 10 mM furfural (Fig. 3A). Deletion of puuP decreased growth by 70% (0.15 g [dcw] liter⁻¹ for RG100, compared to 0.50 g [dcw] liter⁻¹ for LY180). Deletion of potE alone also decreased furfural tolerance in comparison to that of LY180 but to a lesser extent. Deletions of both puuP and potE from LY180 decreased growth in 10 mM furfural by 80%, indicating that native expression levels of these genes contribute to furfural tolerance in the parent.

FIG 3.

Effects of puuP and potE deletions on furfural tolerance. (A) Growth of LY180 and derivatives with a deletion in puuP (strain RG100), potE (strain RG101), and both genes (strain RG102). (B and C) Growth (B) and ethanol production (C) of EMFR9 and a mutant containing a potE deletion (RG105). Transporter deletions in the LY180 chromosome reduced furfural tolerance below that of the unmodified parent. Deletion of potE in EMFR9 reduced furfural tolerance to that of the parent (see also Fig. 1). Addition of pLOI5249 containing potE substantially restored furfural tolerance in a derivative of EMFR9 (RG105) carrying a potE deletion.

Deleting potE in EMFR9 reduced furfural tolerance.

The large increase in potE expression in EMFR9 (relative to LY180) was also presumed to contribute to furfural tolerance. Deletion of the potE coding region in EMFR9 to produce strain RG105 resulted in a substantial loss of furfural tolerance (Fig. 3B and C). With 10 mM furfural, growth and ethanol production were reduced by more than 70% and the furfural MIC was reduced from 15 mM to 12.5 mM. Addition of plasmid pLOI5249, encoding PotE, restored furfural resistance in strain RG105 to near that of EMFR9 containing empty vector, confirming the importance of the gene.

Gene duplication contributes to increased expression of puuP in EMFR35.

Attempts to delete the puuP gene in EMFR35 were unsuccessful with the same genetic tools as employed successfully with LY180 and EMFR9. Although the kanamycin gene was readily integrated, this integration did not remove the chromosomal puuP gene. A persistent full-length coding region for puuP remained in the chromosome. These observations suggested that multiple copies of puuP may be present in the chromosome.

We have previously identified multiple chromosomal copies of genes by examining chromosomal coverage using DNA sequence reads (35). Illumina sequence reads for EMFR35 were mapped against LY180 using Geneious software (Fig. 4). No mutations were found in the primary sequence of puuP or in the nearby puuR regions (repressor). However, this comparison of sequence reads revealed that an 8.8-kbp block of chromosomal DNA containing the puuP gene is duplicated at least 4.5-fold in comparison to flanking sequences. The repressor puuR was outside the amplified region and appears to be present as a single copy. Amplification of puuP without amplification of the repressor (puuR) may be responsible for the increase in expression observed in EMFR35 (Table 2). Beneficial tandem repeats are often selected under stress conditions and can collapse when environmental conditions improve. These are frequently associated with increased resistance to antibiotics (36) but could also contribute to furfural tolerance.

FIG 4.

Alignment of Illumina sequencing reads from EMFR35 on the puuP region of LY180 (parent). Template coordinates are shown at the bottom. Sequence coverage of the 8.8-kbp region containing puuP is increased 4.5-fold compared to that of puuR and flanking regions of the genome. Numbers near the bottom indicate average sequence reads for regions within arrows. Genes listed in the bottom have been modified to allow labeling and are not to scale.

No other gene amplifications were observed in the potE region, plaP region, or potABCD regions in furfural-resistant mutants (EMFR9 and EMFR35). No specific repressors are known for these genes. The mechanisms by which expression of potE is increased in EMFR9 and expression of plaP (10-fold) and potABC (2- to 3-fold) are increased in EMFR35 remain unknown but may involve unidentified regulatory genes.

Reengineering furfural tolerance based on genetic traits in EMFR9 and EMFR35.

Two genetic traits were identified in each of the furfural-resistant mutants (EMFR9 and EMFR35), silencing of yqhD expression by a regulatory mutation (30) and upregulation of a polyamine transporter gene (potE and puuP, respectively). Strains with analogous traits were constructed by transforming plasmids expressing polyamine transporter genes into an LY180 derivative containing a yqhD deletion (XW092). Furfural tolerance in the resulting strains was compared to those of the mutants EMFR9 and EMFR35 (Fig. 5). LY180 containing empty vector (pTrc99A) was the most sensitive to furfural, followed by XW092(pTrc99A) (Fig. 5A and B). Addition of the potE plasmid (pLOI5249) to XW092 increased furfural tolerance to near that found in EMFR9 containing pTrc99A, indicating that both genetic traits are needed. Results with the puuP plasmid (pLOI5412) were similar (Fig. 5C and D). Strain XW092(pLOI5412) was also more resistant to furfural than LY180(pTrc99A) and EMFR9(pTrc99A). However, EMFR35(pTrc99A) was more furfural tolerant than XW092(pLOI5412), indicating that additional mutations are involved. Constructs with a single genetic trait for furfural tolerance were more resistant to furfural than the parent containing empty vector and more sensitive to furfural inhibition than constructs carrying two traits. Figure 1C summarizes MICs for different strains and constructs.

FIG 5.

Reconstructing furfural tolerance in LY180 based on EMFR9 and EMFR35. (A) Growth of XW092 and EMFR9; (B) ethanol production with XW092 and EMFR9; (C) growth of XW092 and EMFR35; (D) ethanol production with XW092 and EMFR35. Strain XW092 is a derivative of LY180 with a deletion in yqhD, similar to the silencing of yqhD expression in EMFR9 and EMFR35. This strain was used as a host for plasmids expressing polyamine transporters. Plasmids pLOI5249 (PotE) and pLOI5412 (PuuP) were added and the resulting constructs compared to EMFR9 and EMFR35, respectively. Furfural resistance of XW092(pLOI5249) was equivalent to that of EMFR9 containing empty vector, indicating that these two mutations are sufficient to reconstitute furfural tolerance. In contrast, XW092(pLOI5412) was less resistant to furfural than EMFR35 containing empty vector. Additional unknown mutations appear to be needed to fully reconstruct the level of furfural resistance present in EMFR35.

Supplementation with agmatine, putrescine, and cadaverine increased furfural tolerance.

E. coli contains 3 polyamines (in order of abundance): putrescine, spermidine, and cadaverine (37). Each of these was tested as a supplement for LY180 in AM1 medium containing 10 mM furfural (Fig. 6A). Agmatine, a precursor of putrescine, was also included. All polyamines except spermidine caused a small increase in furfural tolerance. The addition of spermidine reduced furfural tolerance, though only a small decrease in growth was observed when spermidine was added in the absence of furfural (see Fig. S2 in the supplemental material). The addition of 5 mM putrescine increased growth and ethanol production in AM1 medium containing furfural (Fig. 6B and C).

FIG 6.

Effect of polyamine supplements on furfural (10 mM) tolerance in LY180. Growth (A, B, and D) and ethanol production (C) were used as measures of furfural tolerance. (A) Polyamine supplements (1.0, 5.0, and 10.0 mM) added to AM1 medium containing furfural; (B and C) effect of putrescine (5 mM) on furfural tolerance; (D) combined effects of putrescine (1.0 mM) and a polyamine transporter plasmid (PotE, pLOI5249; PuuP, pLOI5412).

The combination of suboptimal levels of putrescine (1.0 mM) with transporter plasmids (potE or puuP) increased the growth of LY180 in the presence of 10 mM furfural (Fig. 6D). In both cases, combinations were more effective than putrescine or transporter plasmid alone. The beneficial activity of plasmid-borne polyamine transporters for furfural tolerance appears to be partially replaced by supplementing with polyamines.

Batch fermentations with furfural.

In the absence of furfural (AM1 medium containing 100 g liter⁻¹ of xylose), growth and ethanol production began immediately with all LY180 constructs (Fig. 7A and B). Xylose fermentation was complete within 48 h. Addition of 10 mM furfural completely inhibited the growth and fermentation of LY180(pTrc99A) for over 96 h. Furfural was partially metabolized to furfuryl alcohol (Fig. 7C) (19) during the initial 24 h of incubation. Thereafter, the rate of furfural metabolism declined progressively. Less than half of the furfural was metabolized by LY180(pTrc99A) after 96 h.

FIG 7.

Effect of polyamine transporter plasmids on the fermentation of strain LY180 in AM1 medium containing furfural (10 mM) and100 g liter⁻¹ of xylose. (A) growth; (B) ethanol production; (C) furfural metabolism during fermentation.

In contrast, LY180(pLOI5249; potE) and LY180(pLOI5412; puuP) completed the fermentation of xylose (100 g liter⁻¹) in AM1 medium supplemented with 10 mM furfural within 96 h (Fig. 7). Furfural addition (10 mM) caused an initial lag of 48 h, during which furfural was fully metabolized. After this lag, rates of growth and ethanol production with LY180(pLOI5249) and LY180(pLOI5412) were similar to those for the LY180(pTrc99A) control lacking furfural. Plasmid-based expression of polylamine transporters (puuP and potE) increased the metabolism of furfural, decreased the initial lag, and decreased the time required to complete xylose fermentation.

DISCUSSION

Inhibitors such as furfural are formed during dilute acid pretreatment of lignocellulosic biomass (3–7). These pose a significant challenge for the fermentation of hemicellulose sugars into fuels and chemicals (9, 23). Furfural-resistant mutants of E. coli (6, 20) and yeasts (38) have been described that overexpress oxidoreductases which reduce furfural to the less toxic furfuryl alcohol (39, 40). Furfural has been shown to cause single-strand breaks and DNA mutagenesis, primarily in AT-rich regions (24). Allen et al. (25) have reported that reactive oxygen species (ROS), which are also known to damage DNA, proteins, and phospholipid membranes, accumulate in yeast cells during growth in the presence of furfural. In E. coli, polyamines protect DNA from damage by agents of oxidative stress (41, 42). Overexpression of thyA, thymidylate synthase, has also been demonstrated to increase furfural resistance in E. coli (14) and may facilitate repair of furfural-damaged DNA.

Many genes that are beneficial for furfural tolerance in E. coli (fucO, pntAB, and ucpA [13]) directly or indirectly promote the NADH-dependent reduction of furfural to the less toxic and less reactive compound furfuryl alcohol (39, 40). Given the reactive nature of furfural, cellular damage may be ongoing until this compound has been fully metabolized. In our studies, furfural concentrations above 3 mM appear sufficient to prevent rapid growth and metabolism in the presence or absence of furfural tolerance genes (6, 13, 14, 19, 20, 43). Plasmids with polyamine transporter genes appear to function in a similar manner (Fig. 7C). With 10 mM furfural and 100 g liter⁻¹ of xylose, furfural metabolism by LY180 (empty vector control) slows progressively after the first 24 h and does not reach completion even after 96 h. In contrast, LY180 containing either PotE or PuuP plasmids continues to metabolize furfural at a relatively constant rate until the process has been completed.

Polyamines are essential for cell division (33, 44). Concentrations are maintained within a specific range (45) and elevated during exponential growth (46). Polyamines bind anionic structures such as plasma membranes, ribosomes, and DNA (42, 47). This binding protects DNA from damage by some reactive compounds (41), regulates gene expression, and modulates translational fidelity (45, 48). Excess polyamines are toxic, displacing magnesium from ribosomes and inhibiting protein synthesis (49, 50). Our studies suggest that polyamine transporters such as PotABCD, PlaP, PotE, and PuuP and polyamine supplements in the medium protect cellular processes from furfural damage and allow cells to complete the reduction of furfural to the less toxic furfuryl alcohol. Increased expression of polyamine transporters and polyamine supplements represent a new approach to increase furfural tolerance in E. coli and may prove useful with other organisms.