Experimental evolution reveals an effective avenue torelease catabolite repression via mutations in XylRChristian Sieverta,b, Lizbeth M. Nievesa, Larry A. Panyona, Taylor Loefflera, Chandler Morrisa, Reed A. Cartwrighta,b,1,and Xuan Wanga,1

aSchool of Life Sciences, Arizona State University, Tempe, AZ 85287; and bThe Biodesign Institute, Arizona State University, Tempe, AZ 85287

Edited by Arnold L. Demain, Drew University, Madison, NJ, and approved June 1, 2017 (received for review January 14, 2017)

Microbial production of fuels and chemicals from lignocellulosicbiomass provides promising biorenewable alternatives to the con-ventional petroleum-based products. However, heterogeneous sugarcomposition of lignocellulosic biomass hinders efficient microbialconversion due to carbon catabolite repression. The most abundantsugar monomers in lignocellulosic biomass materials are glucose andxylose. Although industrial Escherichia coli strains efficiently use glu-cose, their ability to use xylose is often repressed in the presence ofglucose. Here we independently evolved three E. coli strains from thesame ancestor to achieve high efficiency for xylose fermentation.Each evolved strain has a point mutation in a transcriptional activatorfor xylose catabolic operons, either CRP or XylR, and these mutationsare demonstrated to enhance xylose fermentation by allelic replace-ments. Identified XylR variants (R121C and P363S) have a higheraffinity to their DNA binding sites, leading to a xylose catabolic ac-tivation independent of catabolite repression control. Upon introduc-ing these amino acid substitutions into the E. coli D-lactate producerTG114, 94% of a glucose–xylose mixture (50 g·L−1 each) was used inmineral salt media that led to a 50% increase in product titer after96 h of fermentation. The two amino acid substitutions in XylR en-hance xylose utilization and release glucose-induced repression indifferent E. coli hosts, including wild type, suggesting its potentialwide application in industrial E. coli biocatalysts.

metabolic engineering | XylR | carbon catabolite repression |adaptive laboratory evolution | E. coli

Microbial biocatalysts such as Escherichia coli and Saccharo-myces cerevisiae have been developed to convert sugars to an

array of value-added chemicals, ranging from simple fermentationproducts to complex terpenoids like artemisinic acid (1). Ligno-cellulosic biomass represents a promising renewable feedstock thatcan support large-scale microbial production processes for fuelsand specialty chemicals without interfering with human food supply(2, 3). Lignocellulose is a complex matrix present in plant cell wallsand is mainly composed of polysaccharides and phenolic polymers(2). D-glucose (the sole monomer of cellulose) and D-xylose (thepredominant sugar in hemicellulose) are major sugars found intypical lignocellulosic materials (2, 3). Although the sugar contentin lignocellulosic materials (e.g., agricultural wastes such as cornstover) is higher than 50% of their dry weight, the heterogeneousnature of lignocellulosic sugars inhibits efficient microbial catabo-lism and thus decreases production (2, 3). Industrial microbes suchas S. cerevisiae and Zymomonas mobilis do not natively metabolizexylose, and a foreign xylose catabolic pathway must be integratedinto these hosts for xylose utilization (4, 5). Even for bacteria likeE. coli that natively contain the xylose catabolic pathway, xyloseutilization rates and growth rates on xylose are low (6). More im-portantly, utilization of xylose is repressed in the presence of glu-cose due to a global regulatory mechanism called carbon cataboliterepression (CCR), a common phenomenon observed in bacteriaand fungi, which results in abundant amounts of xylose unusedwhen cells ferment a glucose–xylose mixture (7, 8).As one of classic global regulatory mechanisms, CCR is well

characterized in E. coli (7, 8). The global transcriptional regulatorCRP (cAMP receptor protein) plays a central role in modulating

transcriptional activation of catabolic operons for secondary sugarssuch as xylose, arabinose, and galactose (7, 8). The phosphoenol-pyruvate:sugar phosphotransferase system (PTS) and the membrane-bound enzyme adenylate cyclase (AC; catalyzing the conversionof ATP to cAMP) are also involved in glucose-induced re-pression of xylose utilization in E. coli (7, 8). The phosphoryla-tion state of EIIAGlc, a PTS component encoded by crr in E. coli,plays a pivotal role in regulating AC activities and cAMP levelsaccording to glucose concentrations (9). When glucose concen-trations are high, the phosphate from EIIAGlc is drained towardthe sugars, and the dephosphorylated EIIAGlc is not able to ac-tivate AC, which results in low levels of cAMP (10). WithoutcAMP, CRP cannot activate the transcription of xylose catabolicoperons. In contrast, at low glucose concentrations, copious amountsof phosphorylated EIIAGlc exist and are able to activate AC andpromote cAMP synthesis. If xylose is present, CRP activated bycAMP and the xylose-specific activator XylR (activated when boundby xylose) together coactivate the xylose catabolic operons, xylAB andxylFGH (Fig. 1A) (11). After xylose is imported by XylFGH, a xylose-specific ATP-binding cassette transporter protein, xylose is convertedto xylulose through a reversible one-step reaction catalyzed byxylose isomerase, XylA. Xylulose is then converted to xylulose-5-phosphate by the xylulokinase, XylB, for further degradation viathe pentose phosphate pathway and glycolysis (Fig. 1A) (12).Releasing CCR by genetic engineering allows microbial bio-

catalysts to simultaneously use glucose and other secondary sugarsderived from lignocellulosic biomass and leads to a more efficientfermentation process (7). Common strategies to engineer sugarcoutilization in E. coli include the inactivation of PTS components,such as PtsG and PtsI (13–15), and mutagenesis of CRP (16, 17).

Significance

Lignocellulosic biomass is a promising renewable feedstock formicrobial production of fuels and chemicals. D-glucose andD-xylose are the most abundant sugars in lignocellulosic mate-rials. Economically feasible bioconversion of lignocellulose re-quires simultaneous utilization of both sugars, which is primarilyprevented by carbon catabolite repression. Here we characterizethe genetic basis of three independent laboratory evolution tra-jectories for improved xylose fermentation and reveal an effec-tive way to release carbon catabolite repression in Escherichia coli.

Author contributions: C.S., R.A.C., and X.W. designed research; C.S., L.M.N., L.A.P., T.L.,C.M., and X.W. performed research; C.S., L.M.N., L.A.P., T.L., C.M., R.A.C., and X.W. ana-lyzed data; and C.S., R.A.C., and X.W. wrote the paper.

Conflict of interest statement: Results of this study are part of the nonprovisional patentapplication 15/176,866.

This article is a PNAS Direct Submission.

Freely available online through the PNAS open access option.

Data deposition: The sequence reported in this paper has been deposited in the SRAdatabase (accession no. SRP083931).1To whom correspondence may be addressed. Email: wangxuan@asu.edu or cartwright@asu.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1700345114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1700345114 PNAS | July 11, 2017 | vol. 114 | no. 28 | 7349–7354

APP

LIED

BIOLO

GICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

18,

202

0

However, both approaches have caveats, and only limited successhas been achieved, especially for conditions more relevant to in-dustrial practice such as high sugar concentrations and low-costmedia. Inactivation of PTS components also impairs glucose up-take, and thus, extra efforts to compensate this defectiveness areneeded (13, 18). Theoretically, a cAMP-independent CRP variantshould be able to activate the catabolic operons of secondary sugarseven in the presence of glucose. However, these cAMP-independentCRP mutants often cannot activate the target operons at the sameefficiency as wild-type CRP bound with cAMP (16). In addition, asan important global regulator, CRP directly regulates the tran-scriptional expression of more than 400 genes, and CRP mutantsoften have slow growth phenotypes potentially due to unpredictableexpressional changes of other important genes (19). Here, weevolved E. coli for enhanced xylose fermentation and identified theconvergent genetic basis that increases xylose utilization. By char-acterizing the transcriptional activation mechanism of xylose cata-bolic genes, we discovered a simple and effective genetic approach torelease CCR in E. coli. We identified single nucleotide mutations inxylR that increase xylose utilization up to fourfold in different E. colistrains when fermented in a glucose–xylose mixture (50 g·L−1 of eachsugar). This discovery has the potential to enable different E. colibiocatalysts to simultaneously convert major sugars from lignocel-lulosic biomass into value-added chemicals.

ResultsIdentification of Primary Genetic Changes of E. coli Adaptation forImproved Xylose Fermentation to D-Lactate. A previously engi-neered D-lactate–producing E. coli strain XW043 significantly im-proved its xylose fermentation ability by laboratory adaptation, butthe causative genetic mechanism remains elusive (20). Due to theabsence of competing fermentation pathways in XW043, the evo-lutionary trajectories are likely strictly constrained to xylose–lactate

conversion. We hypothesized that characterization of repeated evo-lutionary trajectories would reveal the convergent causative muta-tions for this adaptation. XW043 was evolved three timesindependently in mineral salt media containing 100 g·L−1 xylose (Fig.1B). The bacterial population was maintained at the exponential orearly stationary phase in a fermentation vessel by serially transferringcultures into new media as previously described (20, 21). In all threeevolutionary trajectories, a rapid adaptation occurred that simulta-neously increased xylose catabolism, lactate titer, and cell growth(Fig. 1B and Fig. S1). Because lactate production is the only fer-mentation pathway supporting cell growth in XW043 under oxygen-limiting experimental conditions (20), increased xylose catabolismrates would lead to higher cell growth rates. At the end of threeevolution experiments, there was an approximately fivefold increaseof lactate titers at 48 h compared with the ancestor strain XW043,accompanied with an increase in yield from 0.6 to 0.8–0.9 g lactate/1 g xylose (Fig. 1B). To understand the genetic changes responsiblefor the increased xylose catabolism, we sequenced the genomes ofthe ancestor XW043 and three representative evolved clones (onefrom each evolved population), designated as strains TL1, CM2, andLP2 using Illumina paired-end sequencing. Each clone was se-quenced twice with an average 24-fold coverage per library. By ap-plying a comprehensive analysis pipeline (details in SI Materials andMethods), overall 5 point mutations, a 1-bp deletion, 15 duplications,and 7 deletions in mostly uncharacterized genes were found in threeevolved isolates compared with XW043 (Table S1). Additionally, wedetected 60 breakpoints as an indicator for structural rearrange-ments occurring in all three evolved clones (Table S1). Of all of themutations detected, three point mutations, one per clone, occurredat the transcriptional coactivators CRP and XylR that are critical forxylose catabolism (Fig. 1A), suggesting a potential result of conver-gent evolution to relieve the predominant metabolic constraints inthe ancestor strain. The three independently isolated point muta-tions result in amino acid changes in CRP (G141D in LP2) andXylR (R121C in CM2 and P363S in TL1). In addition, an IS10 in-sertion was identified in focA reading frame for all three evolvedclones at different positions and orientations (Fig. 1C and Fig. S2A),thereby suggesting an independent origin for each insertion.

Characterization of Physiological Effects of Convergent Mutations.Insertion of IS10 into focA is a common event for all threeevolved isolates, suggesting that the loss of focA function may play abeneficial role for xylose to lactate conversion in XW043. However,an in-frame deletion of focA in XW043 only showed a marginalpositive effect on xylose fermentation (Fig. S2B). Moreover, com-plementation of focA from a plasmid in all evolved strains did notrestore the ancestral phenotype (Fig. S2C), suggesting that the lossof focA function is not the main cause for improved xylose fer-mentation. The pflB gene is located downstream to focA, and IS10insertion into focAmay interfere with the complex promoter systemfor pflB that ranges from the upstream region of focA to the up-stream region of pflB (22) (Fig. S2A). Indeed, the transcript levels ofpflB in all evolved strains were lower than that of the ancestor asshown by qRT-PCR (Fig. S2D). Induced expression of pflB from aplasmid in all evolved strains restored the ancestral phenotype at alarge degree with only less than 25 g·L−1 lactate produced at 96 h(Fig. S2C). This complementation result indicates that the de-creased pflB expression is beneficial for fermentative lactate pro-duction that is also extensively supported by previous work (23, 24).Next, we tested if the missense mutations in crp and xylR have abeneficial effect on xylose fermentation. Replacement of the wild-type copy crp of XW043 with the mutated form (G141D) in thechromosome doubled both xylose utilization and D-lactate titerafter 96 h fermentation, and growth rate and final biomass werealso increased (Fig. 2A and Fig. S3), thereby confirming the positiverole for this mutation to improve xylose fermentation. A 32-kbchromosomal region containing xylR and other xylose catabolicgenes is duplicated in both XW043 and all evolved strains as

A

C

B

Fig. 1. Overview of the adaptive laboratory evolution experiment. (A) Sche-matic drawing of the xylose to D-lactate pathway. PPP, the pentose phosphatepathway; Xylulose-5-P, xylulose-5-phosphate. (B) Serial transfers of E. coli strainXW043 were cultured independently three times, resulting in three evolvedpopulations, from which CM2, LP2, and TL1 were isolated. Extracellular xyloseand D-lactate were determined for each transfer after 48 h of fermentation.(C) Primary convergent mutations were characterized using genome sequenc-ing and allelic replacements.

7350 | www.pnas.org/cgi/doi/10.1073/pnas.1700345114 Sievert et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

0

indicated by genome sequencing (Fig. S4) probably because theprecursor strain of XW043 was adapted for using hemicellulosehydrolysates as the carbon source (25). Only one of the two xylRcopies was mutated (R121C in CM2 and P363S in TL1), therebyconferring a quasi-heterozygous genotype (Fig. S4 C and D). Wewere not able to replace xylR with its mutant alleles in XW043 usinga λ-red recombinase-mediated two-step method (26) presumablybecause the second copy of wild-type xylR tended to replace theselection marker due to a favored intrachromosomal recombination.Instead, using a λ-red recombinase-mediated intrachromosomalrecombination approach (details in SI Materials and Methods), wesuccessfully replaced the xylR mutations with a wild-type copy inboth CM2 and TL1 strains. The restoration of the wild-type alleledecreased xylose utilization rates of the initial 24 h by 40% and65% compared with CM2 and TL1, respectively (Fig. 2 B and C),

indicating a positive effect of these xylR mutations on xylose fer-mentation. Accordingly, the modified strains had decreased lactateproduction and cell growth rates (Fig. 2 B and C and Fig. S3).Restoring wild-type xylR did not produce a complete ancestralphenotype, suggesting the presence of a potential synergy effect be-tween beneficial mutations. We further restored both wild-type focAand xylR in evolved background TL1 to reconstruct an ancestralphenotype. Interestingly, this strain (CS4: TL1 xylR wt/wt, focA wt)performed even worse than XW043 (Fig. 2C), indicating that boththe xylR mutation and focA IS10 insertion are primary causativemutations with a synergy effect and that there are also further mu-tations epistatically interacting with the described mutations.To test if the identified xylR mutations have a universal effect and

if the heterozygous xylR copies are required to improve xylose utili-zation, we substituted wild-type xylR with its mutant version (R121C,P363S, or both) in wild-type E. coli W (ATCC 9637) (Fig. 2D). Allmutants increased xylose utilization and cellular fermentative growthto different degrees (Fig. 2D). Xylose utilization rates were increased2.4- and 4.3-fold compared with the wild-type strain within the initial24 h for xylR R121C and P363S substitutions, respectively (Fig. 2D).We next combined both xylR mutations in a wild-type strain andobserved a slightly additive effect (Fig. 2D). At the end of 96 hfermentation, 40 g·L−1 xylose remained unused in the broth for thewild-type strain, whereas only 10 g·L−1 remained for xylR P363Ssubstitution or xylR R121C and P363S substitution (Fig. 2D).

XylR Variants Show Enhanced Activity in Vitro. CRP G141D waspreviously identified as a mutant with an altered allosteric mecha-nism, and position 141 is involved in the interfacial interactions be-tween subunits and domains (16, 27). Potentially due to the CRPG141Dmutation in the evolved strain LP2, the transcription of xylosecatabolic genes was up-regulated by more than 10-fold as measuredby qRT-PCR (Fig. S5). Unlike CRP and its mutant variants, whichare well characterized, much less is known about XylR.We examinedthe effects that xylR mutations potentially had on the transcriptionallevels for its responsive genes, which are organized as two operons,xylAB and xylFGHR (Fig. 3A). The xylAB and xylFGH transcriptswere increased at least 20-fold in CM2 with xylR (wt/R121C) and 10-fold in TL1 with xylR (wt/P363S) compared with the ancestor strainXW043 (Fig. 3B). The xylR transcript alone shows only between 3.1-and 3.6-fold up-regulation, suggesting that the xylFGHR transcript ispartially degraded at the 3′ end or read-through of the RNA poly-merase is prevented by clashing with proteins bound at the xylR-specific promoter. Because the crp transcript level is not significantlychanged (Fig. 3B), we concluded that the XylR mutations are theprimary reason for up-regulation of the xyl operons.To further study the molecular mechanism causing enhanced

transcriptional activation by XylR variants, we conducted electro-phoretic mobility shift assays (EMSAs) using three purifiedN-terminal His-tagged XylR variants (wild type, R121C, and P363S)and DNA fragments containing the binding sites IA and IF (up-stream of and xylA and xylF, respectively) located in the intergenicregion between xylAB and xylFGH (Fig. 3A) (11, 28). The dissoci-ation constant KD for each site was determined to measure theirbinding affinity (Fig. 3 C and D and Fig. S6 A and B). Although thegel-based assay allows for only a rough approximation, our data forthe wild-type XylR are consistent with a previous report showing aKD of 33 ± 0.8 nM for IA and 25 ± 0.6 nM for IF, respectively (Fig.3D) (28). The evolved XylRs have a significantly lower KD, rangingfrom 3- to 14-fold depending on which binding site and variant (Fig.3D), suggesting that a higher binding affinity leads to a more stabletranscription initiation complex and a subsequently higher tran-scriptional rate. Moreover, wild-type XylR could not bind opera-tor sequences in the absence of xylose, as the majority of DNAfragments remained unbound in the gel (Fig. S6C). In contrast,both XylR variants bound DNA even without xylose (Fig. S6C),suggesting that these XylR variants have xylose-independentactivities. A twofold increase in binding of wild-type XylR to its

A

B

C

D

Fig. 2. Evaluation of focA, crp, and xylR mutations. Through allelic replace-ments, the genotypes of appropriate genes were modified in different geneticbackgrounds: the ancestor XW043 (A), the evolved CM2 (B) and TL1 (C), andthe wild-type strain W (D). Fermentation performance was determined bymeasuring the cell optical density (Left) and extracellular xylose concentration(Right) over 96 h. Point mutations are indicated by the relevant amino acidsubstitution. Error bars represent SEM with n = 3. WT, wild type.

Sievert et al. PNAS | July 11, 2017 | vol. 114 | no. 28 | 7351

APP

LIED

BIOLO

GICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

18,

202

0

responsive site was observed when 100 μM xylose was added, butsimilar enhancement was much less significant for both XylRvariants (Fig. S6C).

Amino Acid Substitutions in XylR Release CCR. The higher bindingaffinity of the evolved XylR and the concomitant xylose independencyled to the hypothesis that these amino acid substitutions in XylR areable to release CCR. Wild-type E. coli W was used as the test strainfor different genetic modifications to release CCR. Batch fermenta-tions were conducted using a mineral salt medium AM1 containing100 g·L−1 of a glucose–xylose mixture (50 g·L−1 each). The wild-typestrain consumed all glucose but only 16% of the xylose after 96 h offermentation (Fig. 4A and Fig. S7A). In comparison, the XylR vari-ants R121C or P363S released CCR, and glucose and xylose wereconsumed simultaneously, eventually leading to 61–69% of the xyloseused (Fig. 4A and Fig. S7A). LN6, the strain with both R121C andP363S substitutions, used 87 g·L−1 of the total sugars compared with57 g·L−1 for wild type (Fig. 4A). The xylose utilization rate was in-creased fourfold compared with wild type (81% of the xylose used

after 96 h fermentation) and only 3 g·L−1 glucose remained unused inthe fermentation broth (Fig. 4A and Fig. S7A). CRP G141D alone ortogether with xylRmutations did not enhance sugar coutilization (Fig.4A). Investigation of known mutations that relieve CCR such asΔmgsA (29) and crp* (I112L, T127I, and A144T) (30, 31) was alsoconducted in the same wild-type background. CRP* only increasedxylose utilization to 37% with impaired cell growth, whereas themgsAdeletion had no benefit in wild-type background (Fig. 4A and Fig.S7A). Two additional sugar compositions (the ratio of glucose toxylose is 1:2 or 2:1 by mass with the total sugar concentration at100 g·L−1) were used to further compare the effect of crp* and xylRmutations on sugar coutilization. The xylR mutations (R121C andP363S) yielded more total sugar utilization than crp* by 58% and14% for these two conditions, respectively (Fig. S7B).To further demonstrate the application of our discovery in con-

verting sugar mixtures into renewable chemicals, we substituted wild-type xylR with xylR (R121C and P363S) of a previously engineeredD-lactate E. coli producer TG114, which efficiently converts glucoseinto D-lactate (21). Fermenting TG114 in a glucose–xylose mixture,50 g·L−1 of each sugar, showed that only 34% of the xylose was useddue to CCR (Fig. 4B). Introducing R121C and P363S substitutionsin xylR enabled the modified strain to use 88% of the xylose. In-terestingly, the decreased glucose utilization caused by xylR muta-tions in wild-type background did not occur in TG114 background,and 50 g·L−1 glucose was completely fermented (Fig. 4 and Fig. S7)potentially due to the enhanced glucose fermentation ability ofTG114 through a long-term adaptation (21). Consequently, D-lactateproduction increased to 1.5-fold compared with TG114 with a finaltiter at 86 g L−1 and a yield of 0.91 g g−1 sugars.

DiscussionWe characterized the primary genetic changes of the adaptation forxylose utilization in an E. coli D-lactate producer and found a wayto effectively release CCR in E. coli. Due to the strong selective

D E

A

B

C

Fig. 3. Molecular mechanism of enhanced xylose utilization by XylR muta-tions. (A) Schematic drawing of the xyl operons (gray box) including theirtranscription start point (arrow). Binding sites of CRP and XylR are indicated asyellow and blue boxes. XylR has two known binding sites, IA and IF. The givensize does not reflect real proportions. (B) Relative transcript levels of selectedgenes of strains CM2 and TL1 with xylR variants R121C and P363S, respectively.The data indicate the fold change of transcript level in relation to the ancestorXW043 background and are normalized by the transcript level of 16S ribosomalRNA (rrsA), n = 4, unpaired Student’s t test indicates significance at *P < 0.05.(C) Fitted data and (D) the determined dissociation constant (KD) from EMSAsto determine the binding affinity of different XylR variants with their knownbinding sites, IA and IF, respectively. (E) Modeled structure of the dimeric wild-type XylR (28). The following features are highlighted: xylose (purple), mutatedresidues (red), and DNA binding domain (blue).

A

B

Fig. 4. Coutilization of glucose–xylose mixtures by batch fermentations.(A) E. coli W was engineered with different genotypes including previouslyreported genetic traits such as mgsA deletion (ΔMgsA), CRP*, and identifiedCRP and XylR variants to compare their effect of releasing CCR. Batch fer-mentations were performed using glucose–xylose mixtures (50 g·L−1 of eachsugar) for 96 h. WT, wild type. (B) The fermentation of TG114 and its modifiedstrain LN23C with the R121C/P363S substitution in XylR using sugar mixtures.Cell optical density (OD550nm), xylose, and D-lactate concentrations were de-termined. Error bars represent SEM, n = 3.

7352 | www.pnas.org/cgi/doi/10.1073/pnas.1700345114 Sievert et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

0

pressure and short adaptation time, the convergent mutation pat-terns occurred in three parallel evolved populations (Fig. 1C). Incontrast to the convergent mutations in crp and xylR identified inthis work, previous research using similar experimental evolutionaryapproaches showed that beneficial mutations for enhanced xylosefermentations in E. coli biocatalysts occurred at sugar transportergenes, such as galP and gatC (32, 33). This suggests that there arepotentially multiple evolutionary solutions for the same problem,and bacterial genotypic backgrounds may predetermine evolu-tionary trajectories. Current high-throughput next-generationsequencing capacity has significantly outpaced reverse engi-neering processes to characterize the causative mutations. Onetime-consuming bottleneck is to distinguish the causative muta-tions from mutational noise and sequencing errors (34). Here,two strategies were used to effectively identify most critical muta-tions for the adapted phenotype. First, study of multiple in-dependent evolutionary trajectories from the same ancestorfacilitated the process of searching for important convergentmutations. Second, as sequencing technical errors were excludedby sequencing each genome twice with high coverage, linkagedisequilibrium is probably the only source of noise, which wefurther reduced by focusing on only an individual evolved cloneinstead of the whole bacterial population.We identified a G141D substitution in CRP enhancing tran-

scription of xyl operons in E. coli. This mutation was previouslyreported to alter allosteric regulation (16). Position 141 is part of ahinge important for the intramolecular transduction of the activa-tion signal (35). CRP G141D did not release CCR (Fig. 4A), sug-gesting that this variant remains to be responsive to cAMP, which isconsistent with other previously reported mutations at this position(27). We hypothesize that a polar residue Asp at position 141 ori-ents the DNA binding domain for a better interaction with oper-ator sequences, resulting in an increased expression compared withwild type. Interestingly, an intrinsically active CRP homolog of theplant pathogen Xanthomonas campestris has an Asp residue at thecorresponding position to G141 of E. coli CRP. A substitution ofthis Asp residue with the nonpolar residue Ala reduces the bindingaffinity to its promoter by ∼12-fold (36). In contrast to CRPG141D, the identified XylR variants not only enhance xylose uti-lization but also release CCR, leading to an efficient coutilization ofglucose–xylose mixtures (Fig. 4). The AraC-type transcription fac-tor XylR has two helix-turn-helix (HTH) motifs, and its N-terminalligand binding domain contains a unique periplasmic-binding pro-tein fold that is structurally related to LacI/GalR transcriptionfactors (28). Upon xylose binding, XylR undergoes a conforma-tional change orienting the two HTH motifs of the DNA bindingdomain in an active conformation (28). In one instance, a Proresidue of XylR was substituted with Ser (P363S) at the sitethat connects the two HTH motifs, likely causing a significantreorientation and tighter binding to the DNA (Fig. 3E). Thesecond variant is a R121C substitution, which is neither in closeproximity to the DNA nor in close proximity to the ligand bindingdomain (Fig. 3E). The region is also not known to participate insignal communication from the N-terminal ligand binding to theC-terminal DNA binding domain (28). However, the effect on thebinding affinity is similarly increased in both variants. A possiblemode of action might be an interaction of residue 121 with Thr185,which is directly connected to a helix motif that promotes di-merization of XylR. Cys is less bulky than Arg, which might stabilizedimerization due to a reduced steric hindrance.CCR in E. coli can be dramatically released simply by amino

acid substitution in XylR as demonstrated here (Fig. 4). The likelymechanism is that the enhanced binding of XylR variants to theoperator sequences between xylAB and xylFGH operons enablesindependency from both CRP and xylose, thereby resulting in theobserved sugar coutilization. Multiple genetic engineering ap-proaches have previously been developed to release CCR in E. coli.Inactivation ofmgsA significantly enhanced sugar coutilization in an

ethanologenic E. coli (29). However, glucose-induced repression isstill severe in two tested strains with inactivated mgsA, LN24B(E. coliW ΔmgsA) and TG114 (mgsA is deleted to prevent productimpurity) (Fig. 4), suggesting its limited application in a broad rangeof E. coli catalysts. Several cAMP-independent CRP mutants werediscovered to release CCR to some extent (16, 37). But cell growthis often stunted by crp mutations, and sugar coutilization is not ef-ficient, which was also observed in this work (Fig. 4A and Fig. S7A).Disruption of the PTS system (deletion of ptsG or ptsI) is an ef-fective approach to release CCR, but glucose uptake is significantlyimpaired and needs to be reengineered for glucose utilization,which involves extra genetic modifications or adaptation (7, 13, 18).In addition, this approach does not enhance the native xylose ca-tabolism of E. coli. In this study, by simply substituting two residuesin XylR of an established D-lactate producer, this modified strain isable to simultaneously ferment 50 g·L−1 glucose and 43 g·L−1 xyloseand produce 86 g·L−1 lactate within 96 h in a mineral salt medium(Fig. 4B), which shows advantages compared with previously engi-neered strains for lactate fermentative production using sugarmixtures in terms of the xylose utilization rate, lactate yield, andtiter under a similar batch fermentation condition (14, 38). Thisgenetic approach may be effective for other E. coli biocatalysts, asthese XylR variants have a positive effect in a wild-type strain (Fig.4A). Moreover, these XylR variants may also reduce arabinose-induced repression, which is caused by a competitive binding ofAraC (activated by D-arabinose) to the regulatory regions of xyloperons (39). Enhanced binding of these XylR variants to the op-erator sequences will potentially compete against AraC and achievecoutilization of arabinose and xylose.

Materials and MethodsDetailed description of the materials and methods can be found in SI Ma-terials and Methods.

Bacterial Strains, Plasmids, and Growth Conditions. The strains used in this studyare listed in Table S2. The D-lactate–producing E. coli XW043 (20) and TG114 (21)were kindly provided by Lonnie Ingram’s laboratory at the University of Florida,Gainesville, FL. PC05 containing CRP* (31) was generously provided by PatrickCirino’s laboratory at University of Houston, Houston. All other strains are de-veloped in this study. The plasmids and primers used for chromosomal integra-tions, purification of His-tagged XylR variants, and induced expression of focA andpflB are summarized in Table S2. During strain construction, cultures were grownaerobically in shake flasks containing Luria broth. Ampicillin (50 mg·L−1),kanamycin (50 mg·L−1), or chloramphenicol (40 mg·L−1) was added as appropri-ate. All fermentations were performed using AM1 mineral salt media supple-mented with different carbon sources (20). For batch fermentations experiments,E. coli was grown at 37 °C in a pH-controlled fermentation vessel with the initialcell density at 0.022 g·dry cell weight (dcw)·L−1 as previously described (20, 21).The media contain the defined sugar compositions with the total sugar con-centration at 100 g·L−1. For laboratory evolution experiments, strain XW043 wasconsecutively transferred into fresh media in the above-mentioned fermentationvessels at ∼24–48-h time periods for a total of 20 transfers (∼150 generations).Sugars and lactate were measured by HPLC (20, 21).

Genetic Methods and Genome Sequencing.All chromosomalmodificationswereconducted using a λ-red recombinase-mediated two-step method as previouslydescribed (26). Additional details are included in SI Materials and Methods.Genomic DNA from strains XW043, TL1, LP2, and CM2 was extracted and pu-rified using the Promega Wizard genomic DNA purification kit. Extracted DNAwas sequenced using the MiSeq Illumina sequencer by the DNASU Next Gen-eration Sequencing Core at Arizona State University, generating 2 × 300 ntpaired-end reads, and all sequencing reads were deposited in NCBI (SRA ac-cession no. SRP083931). The procedures for genome assembly and identifica-tion of de novo mutations are included in SI Materials and Methods.

Real-Time RT-PCR Analysis. Relative transcript levels were determined using thecomparative CT method (40). The data were normalized with the level of 16Sribosomal RNA (rrsA).

EMSA. His-tagged XylR proteins were purified using a column packed with Ni-NTA agarose. We amplified 200-bp DNA fragments containing one XylR

Sievert et al. PNAS | July 11, 2017 | vol. 114 | no. 28 | 7353

APP

LIED

BIOLO

GICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

18,

202

0

binding site (IA and IF) by PCR and purified them. EMSA was performed aspreviously described with minor modifications (41). The detailed experimentalprocedures are included in SI Materials and Methods.

ACKNOWLEDGMENTS.We thank Scott Bingham, David Winter, and Kael Daifor technical assistance. This work was supported by Arizona State Universityand NIH Grant R01-HG007178.

1. Nielsen J, Keasling JD (2016) Engineering cellular metabolism. Cell 164:1185–1197.2. Saha BC (2003) Hemicellulose bioconversion. J Ind Microbiol Biotechnol 30:279–291.3. Gírio FM, et al. (2010) Hemicelluloses for fuel ethanol: A review. Bioresour Technol

101:4775–4800.4. Ho NW, Chen Z, Brainard AP (1998) Genetically engineered Saccharomyces yeast capable

of effective cofermentation of glucose and xylose. Appl EnvironMicrobiol 64:1852–1859.5. Zhang M, Eddy C, Deanda K, Finkelstein M, Picataggio S (1995) Metabolic engineering of a

pentose metabolism pathway in ethanologenic Zymomonas mobilis. Science 267:240–243.6. Beall DS, Ohta K, Ingram LO (1991) Parametric studies of ethanol production form xylose

and other sugars by recombinant Escherichia coli. Biotechnol Bioeng 38:296–303.7. Kim JH, Block DE, Mills DA (2010) Simultaneous consumption of pentose and hexose

sugars: An optimal microbial phenotype for efficient fermentation of lignocellulosicbiomass. Appl Microbiol Biotechnol 88:1077–1085.

8. Görke B, Stülke J (2008) Carbon catabolite repression in bacteria: Many ways to makethe most out of nutrients. Nat Rev Microbiol 6:613–624.

9. Bettenbrock K, et al. (2007) Correlation between growth rates, EIIACrr phosphorylation,and intracellular cyclic AMP levels in Escherichia coli K-12. J Bacteriol 189:6891–6900.

10. Hogema BM, et al. (1998) Inducer exclusion in Escherichia coli by non-PTS substrates:The role of the PEP to pyruvate ratio in determining the phosphorylation state ofenzyme IIAGlc. Mol Microbiol 30:487–498.

11. Song S, Park C (1997) Organization and regulation of the D-xylose operons in Es-cherichia coli K-12: XylR acts as a transcriptional activator. J Bacteriol 179:7025–7032.

12. Lawlis VB, Dennis MS, Chen EY, Smith DH, Henner DJ (1984) Cloning and sequencingof the xylose isomerase and xylulose kinase genes of Escherichia coli. Appl EnvironMicrobiol 47:15–21.

13. Gosset G (2005) Improvement of Escherichia coli production strains by modification ofthe phosphoenolpyruvate:sugar phosphotransferase system. Microb Cell Fact 4:14.

14. Dien BS, Nichols NN, Bothast RJ (2002) Fermentation of sugar mixtures using Es-cherichia coli catabolite repression mutants engineered for production of L-lacticacid. J Ind Microbiol Biotechnol 29:221–227.

15. Lindsay SE, Bothast RJ, Ingram LO (1995) Improved strains of recombinant Escherichiacoli for ethanol production from sugar mixtures. Appl Microbiol Biotechnol 43:70–75.

16. Aiba H, Nakamura T, Mitani H, Mori H (1985) Mutations that alter the allosteric na-ture of cAMP receptor protein of Escherichia coli. EMBO J 4:3329–3332.

17. Kolb A, Busby S, Buc H, Garges S, Adhya S (1993) Transcriptional regulation by cAMPand its receptor protein. Annu Rev Biochem 62:749–795.

18. Chiang CJ, et al. (2013) Systematic approach to engineer Escherichia coli pathways forco-utilization of a glucose-xylose mixture. J Agric Food Chem 61:7583–7590.

19. Khankal R, Chin JW, Ghosh D, Cirino PC (2009) Transcriptional effects of CRP* ex-pression in Escherichia coli. J Biol Eng 3:13.

20. Wang X, et al. (2011) Increased furfural tolerance due to overexpression of NADH-dependent oxidoreductase FucO in Escherichia coli strains engineered for the pro-duction of ethanol and lactate. Appl Environ Microbiol 77:5132–5140.

21. Grabar TB, Zhou S, Shanmugam KT, Yomano LP, Ingram LO (2006) Methylglyoxalbypass identified as source of chiral contamination in L(+) and D(-)-lactate fermen-tations by recombinant Escherichia coli. Biotechnol Lett 28:1527–1535.

22. Sawers G, Suppmann B (1992) Anaerobic induction of pyruvate formate-lyase geneexpression is mediated by the ArcA and FNR proteins. J Bacteriol 174:3474–3478.

23. Utrilla J, Gosset G, Martinez A (2009) ATP limitation in a pyruvate formate lyasemutant of Escherichia coli MG1655 increases glycolytic flux to D-lactate. J IndMicrobiol Biotechnol 36:1057–1062.

24. Jarboe LR, Grabar TB, Yomano LP, Shanmugan KT, Ingram LO (2007) Development ofethanologenic bacteria. Adv Biochem Eng Biotechnol 108:237–261.

25. Geddes CC, et al. (2011) Simplified process for ethanol production from sugarcanebagasse using hydrolysate-resistant Escherichia coli strain MM160. Bioresour Technol102:2702–2711.

26. Jantama K, et al. (2008) Eliminating side products and increasing succinate yields inengineered strains of Escherichia coli C. Biotechnol Bioeng 101:881–893.

27. Cheng X, Lee JC (1994) Absolute requirement of cyclic nucleotide in the activation of theG141Q mutant cAMP receptor protein from Escherichia coli. J Biol Chem 269:30781–30784.

28. Ni L, Tonthat NK, Chinnam N, Schumacher MA (2013) Structures of the Escherichia colitranscription activator and regulator of diauxie, XylR: An AraC DNA-binding familymember with a LacI/GalR ligand-binding domain. Nucleic Acids Res 41:1998–2008.

29. Yomano LP, York SW, Shanmugam KT, Ingram LO (2009) Deletion of methylglyoxalsynthase gene (mgsA) increased sugar co-metabolism in ethanol-producing Escher-ichia coli. Biotechnol Lett 31:1389–1398.

30. Eppler T, Boos W (1999) Glycerol-3-phosphate-mediated repression of malT in Es-cherichia coli does not require metabolism, depends on enzyme IIAGlc and is medi-ated by cAMP levels. Mol Microbiol 33:1221–1231.

31. Cirino PC, Chin JW, Ingram LO (2006) Engineering Escherichia coli for xylitol pro-duction from glucose-xylose mixtures. Biotechnol Bioeng 95:1167–1176.

32. Sawisit A, et al. (2015) Mutation in galP improved fermentation of mixed sugars tosuccinate using engineered Escherichia coli AS1600a and AM1 mineral salts medium.Bioresour Technol 193:433–441.

33. Utrilla J, et al. (2012) Engineering and adaptive evolution of Escherichia coli forD-lactate fermentation reveals GatC as a xylose transporter. Metab Eng 14:469–476.

34. Conrad TM, Lewis NE, Palsson BO (2011) Microbial laboratory evolution in the era ofgenome-scale science. Mol Syst Biol 7:509.

35. Popovych N, Tzeng SR, Tonelli M, Ebright RH, Kalodimos CG (2009) Structural basis forcAMP-mediated allosteric control of the catabolite activator protein. Proc Natl AcadSci USA 106:6927–6932.

36. Chin KH, et al. (2010) The cAMP receptor-like protein CLP is a novel c-di-GMP receptorlinking cell-cell signaling to virulence gene expression in Xanthomonas campestris.J Mol Biol 396:646–662.

37. Ryu S, Kim J, Adhya S, Garges S (1993) Pivotal role of amino acid at position 138 in theallosteric hinge reorientation of cAMP receptor protein. Proc Natl Acad Sci USA 90:75–79.

38. Abdel-Rahman MA, et al. (2015) Fed-batch fermentation for enhanced lactic acidproduction from glucose/xylose mixture without carbon catabolite repression. J BiosciBioeng 119:153–158.

39. Desai TA, Rao CV (2010) Regulation of arabinose and xylose metabolism in Escherichiacoli. Appl Environ Microbiol 76:1524–1532.

40. Schmittgen TD, Livak KJ (2008) Analyzing real-time PCR data by the comparative C(T)method. Nat Protoc 3:1101–1108.

41. Shin JH, Roh DH, Heo GY, Joo GJ, Rhee IK (2001) Purification and characterization of aregulatory protein XylR in the D-xylose operon from Escherichia coli. J MicrobiolBiotechnol 11:1002–1010.

42. Kitagawa M, et al. (2005) Complete set of ORF clones of Escherichia coli ASKA library(a complete set of E. coli K-12 ORF archive): Unique resources for biological research.DNA Res 12:291–299.

43. Datsenko KA, Wanner BL (2000) One-step inactivation of chromosomal genes in Es-cherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97:6640–6645.

7354 | www.pnas.org/cgi/doi/10.1073/pnas.1700345114 Sievert et al.

Dow

nloa

ded

by g

uest

on

June

18,

202

0