Regulation of the sol Locus Genes for Butanol and AcetoneFormation in Clostridium acetobutylicum ATCC 824 by a
Putative Transcriptional RepressorRAMESH V. NAIR,1† EDWARD M. GREEN,2 DAVID E. WATSON,2 GEORGE N. BENNETT,2
AND ELEFTHERIOS T. PAPOUTSAKIS1*
Department of Chemical Engineering, Northwestern University, Evanston, Illinois 60208,1 andDepartment of Biochemistry and Cell Biology, Rice University, Houston, Texas 772512
Received 4 June 1998/Accepted 28 October 1998
A gene (orf1, now designated solR) previously identified upstream of the aldehyde/alcohol dehydrogenasegene aad (R. V. Nair, G. N. Bennett, and E. T. Papoutsakis, J. Bacteriol. 176:871–885, 1994) was found to encodea repressor of the sol locus (aad, ctfA, ctfB and adc) genes for butanol and acetone formation in Clostridiumacetobutylicum ATCC 824. Primer extension analysis identified a transcriptional start site 35 bp upstream ofthe solR start codon. Amino acid comparisons of SolR identified a potential helix-turn-helix DNA-binding motifin the C-terminal half towards the center of the protein, suggesting a regulatory role. Overexpression of SolRin strain ATCC 824(pCO1) resulted in a solvent-negative phenotype owing to its deleterious effect on thetranscription of the sol locus genes. Inactivation of solR in C. acetobutylicum via homologous recombinationyielded mutants B and H (ATCC 824 solR::pO1X) which exhibited deregulated solvent production character-ized by increased flux towards butanol and acetone formation, earlier induction of aad, lower overall acidproduction, markedly improved yields of solvents on glucose, a prolonged solvent production phase, andincreased biomass accumulation compared to those of the wild-type strain.
Several solventogenic genes (aad [42] or adhE [22], bdhAand bdhB [50, 65], adc [23, 48, 49], and ctfA and ctfB [22, 49])have recently been cloned and sequenced from Clostridiumacetobutylicum and another solventogenic Clostridium species(29) (adh1 [74]). These genes share a common induction pat-tern in that they are all expressed only at the onset of solven-togenesis during the late exponential growth stage. Speculationabounds as to the factors that are responsible for triggeringsolventogenesis. Some of these are believed to be pH, thresh-old butyrate concentration (62), and nutrient limitations (51).A Spo0A-mediated regulation of events during stationary-phase metabolism is implicated in a Clostridium beijerinckiistrain (68). A repressor protein (similar to the LacI family ofrepressors) encoded by regA from a solventogenic Clostridiumspecies (formerly C. acetobutylicum P262 [29]) is believed to beinvolved in the regulation of starch degradation (15, 70). Todate, no gene encoding a regulatory protein that modulatessolvent formation genes has been cloned from C. acetobutyli-cum. Recently the sigA product (57) from strain DSM 792(which is grouped with the type strain ATCC 824 [29, 30]) wasidentified. However, so far sigma factors involved in transcrip-tion of solventogenic genes have not been found in C. aceto-butylicum (20, 41).
Clustering of genes involving both mono- and polycistronicoperons in C. acetobutylicum has been reported elsewhere (6,9, 22, 42, 49, 65). The only polycistronic operon that has so farbeen cloned from C. acetobutylicum involving solvent pathwaygenes is that of aad-ctfA-ctfB (42) (a virtually identical systemin C. acetobutylicum DSM 792 is the sol operon involving the
genes adhE and ctfA amd ctfB [22]). This operon containsgenes involved in both butanol and acetone formation, the twopredominant solvents produced by C. acetobutylicum. We haverecently reported that in C. acetobutylicum ATCC 824, thisoperon and the adc gene are located on a large 210-kb plasmid(pSOL1) and not on the chromosome (13). Understanding ofthe regulation of these solventogenic genes is crucial for met-abolically engineering (37) this organism to improve produc-tion of butanol and acetone. In a search for regulatory proteinsof this polycistronic operon, sequencing further upstream ofaad was initiated, which resulted in the discovery of a 957-bpopen reading frame (ORF) (orf1) 663 bp upstream of aad onthe same DNA strand (42). The proximal location on pSOL1of orf1 (now designated solR) to aad and the size of the protein(;37 kDa) that may be encoded by this gene suggested thatthis may be a regulatory protein. The present study examinesthe possible regulatory role played by this solR-coded product.
MATERIALS AND METHODS
Bacterial strains and plasmids. All bacterial strains and plasmids used in thisstudy are shown in Table 1.
Growth conditions and maintenance. All Escherichia coli strains were grownaerobically at 37°C in Luria-Bertani (LB) medium. For recombinant strains,media were appropriately supplemented with ampicillin (50 to 60 mg/ml), eryth-romycin (ERM) (200 mg/ml), and chloramphenicol (32 mg/ml). Both recombi-nant and wild-type strains were stored at 285°C in 15% (vol/vol) glycerol (53).
C. acetobutylicum ATCC 824 was maintained as spores in corn mash glucosemedium (CMG) (51) at 4°C under nitrogen. Spores were activated by heating at70 to 80°C for 10 min. Recombinant clostridial strains were stored frozen in 15%(vol/vol) glycerol at 285°C or as single colonies on reinforced clostridial agar(RCA; Difco Laboratories, Detroit, Mich.), pH 6.8. In 10 ml-tube cultures, C.acetobutylicum was grown under anaerobic conditions at 37°C in 23 YTG (45),reinforced clostridial medium (RCM; Difco Laboratories), or clostridium growthmedium (CGM) (51). Recombinant C. acetobutylicum cells (carrying macrolide-,lincosamide-, and streptogramin B-resistant [MLSr] plasmids) were cultured inthe above media supplemented with 40 mg of ERM per ml on plates and 100 mgof ERM per ml in liquid culture.
Controlled-pH fermentor experiments. Large-scale batch fermentations (5.5liters) of various C. acetobutylicum strains were performed in a BioFlo II fer-mentor (New Brunswick Scientific, Edison, N.J.) with a culture volume of 5 liters
(CGM with 80 g of glucose per liter instead of 50 g/liter), as previously described(42).
Glucose and fermentation product analysis. Residual glucose concentration inculture supernatants was measured with a Select biochemistry analyzer (model2700; YSI, Yellow Springs, Ohio) and a YSI dextrose membrane according to themanufacturer’s instructions. The concentrations of butanol, acetone, ethanol,butyrate, and acetate were determined with a Varian Vista 6000 gas chromato-graph (Varian, Walnut Creek, Calif.) (36).
DNA isolation, transformation, and manipulation. Plasmid isolation from E.coli was done by the method of Birnboim and Doly (7), with the additional stepsof the procedure of Wu and Welker (71) when the DNA was to be sequenced.Large-scale plasmid isolation was undertaken with a QIAGEN Plasmid Maxi Kit(QIAGEN, Chatsworth, Calif.). Plasmid DNA was desalted and concentratedusing Microcon-100 microconcentrators (Amicon, Beverly, Mass.). BacterialDNA was prepared from 10 ml of exponential-phase C. acetobutylicum cells(optical density at 600 nm of ;0.8) with a Puregene DNA Isolation Kit (GentraSystems, Minneapolis, Minn.). Previously published methods were used for elec-trotransformation of E. coli (19) and C. acetobutylicum (39). Prior to transfor-mation of C. acetobutylicum, plasmids pCO1 and pO1X were methylated in E.coli(pAN1) by the B. subtilis phage f3TI methyltransferase, which protects theplasmid DNA from restriction by the clostridial endonuclease Cac824I (38).Approximately 15 mg of methylated nonreplicating plasmid pO1X DNA wasused to transform C. acetobutylicum.
Southern hybridization. Plasmid (pO1X) and bacterial (wild-type, mutant B,and mutant H) DNAs were digested to completion with either EcoRI or ScaI.The DNA was transferred from an agarose gel to a HYBOND-N1 nylon mem-brane (Amersham Life Science, Arlington Heights, Ill.) by capillary blotting (60)and then probed with a radiolabeled solR gene fragment, isolated from pO1X.The gene fragment was labeled with [a-32P]dATP using a random primingDECAprime II DNA Labeling Kit (Ambion, Austin, Tex.), and unincorporatedradionucleotides were removed by exclusion chromatography on Sephadex G-50.The prehybridization, hybridization, and washing steps were performed at 42°Cin accordance with the membrane manufacturer’s instructions, and the radioac-tive membranes were visualized after exposure to X-ray film.
Primer extension. Total RNA was isolated from C. acetobutylicum as previ-ously described (65). Primer extension reactions were performed as previouslydescribed (24) with Moloney murine leukemia virus reverse transcriptase (Am-ersham) using 20 mg of total RNA, unless otherwise stated. To determine the 59end of solR mRNA, an end-labeled oligonucleotide, BORFU-PE (59-CGCAATAGGTATGACATATG-39) complementary to the N-terminal region of solR wasused in a primer extension reaction. Oligonucleotide primers were end labeledwith [g-32P]ATP (NEN Research Products) as previously described (41). Theprecise transcriptional start site of solR mRNA was determined by sequencing ofplasmid pCO1 around the N-terminal end of solR by the Sanger dideoxy chain-termination method (54), as previously described (42), using the same syntheticoligonucleotide (20-mer) primer (BORFU-PE). RNA for solR primer extensionstudies was isolated from wild-type and recombinant C. acetobutylicum [strain
ATCC 824(pCO1)] cells collected during the acidogenic (early exponentialgrowth phase, stage A, 5 h) and early solventogenic (late exponential growthphase, stage B, 10 h) stages in batch fermentations with a controlled pH (pH $4.5). RNA for the time course primer extension experiments was isolated fromATCC 824(pCO1) and mutant B (ATCC 824 solR::pO1X) cells isolated duringthe early exponential (stage A, 5 h), late exponential (stage B, 10 h), earlystationary (stage C, 25 h), and late stationary (stage D, 50 h) stages in batchfermentations with a controlled pH (pH $ 4.5). The presence of mRNA corre-sponding to solR, aad, and adc genes in each of the above four stages was verifiedby performing primer extension reactions using end-labeled 20-mer syntheticoligonucleotides BORFU-PE, BYDH-PE (59-TTTACTGTTGTGACTTTCAT-39), and N-ADC (59-TTCATCCTTTAACATAAAAG-39) that are complemen-tary to the N-terminal ends of the respective genes.
Northern analysis. The 0.9-kb clostridial EcoRI fragment from pO1X waslabeled with [a-32P]dATP (NEN) using the random-priming Prime-It II Kit(Stratagene, La Jolla, Calif.) per the manufacturer’s instructions. Unincorpo-rated radionucleotides were removed by using NucTrap probe purification col-umns (Stratagene). Northern blotting was performed as previously described(64) with the following modifications. Transfer of the RNA to 0.2-mm-pore-sizeMaximum-Strength Nytran Plus nylon membrane (Schleicher & Schuell, Keene,N.H.) was done by using a TurboBlotter Nytran System (Schleicher & Schuell)according to the manufacturer’s instructions. Prehybridization and hybridizationsteps were carried out for 24 h at 29°C, while all washes were performed at 44°C.The air-dried filter was exposed at 285°C to XOMAT-AR film for 5 days with aDuPont Cronex Xtra Life Lightning-Plus Intensifying Screen (E. I. DuPont deNemours, Wilmington, Del.) used according to the manufacturer’s instructionsto amplify the signal.
Construction of plasmids. (i) pSOLR. The DNA fragment containing solR andthe promoter region of aad was amplified by PCR, as described earlier (42), withplasmid (pHXS5) used as the template DNA. The upstream primer DAP-UP(59-ATGGTCGGCGTGAATTCGTGAACAATTG-39) was generated by sub-stituting a G for a T at nucleotide position 220 (42) and a T for a A at nucleotideposition 224 to provide an internal EcoRI site (underlined). The downstreamprimer DAP-DN (59-TGCTGCCATTGCTGCAGTTCTAAAGATT-39) wasgenerated by substituting a G for a T at nucleotide position 2171 (42) on thecomplementary strand to provide an internal PstI site (underlined). The 1,979-bpamplified DNA fragment was digested sequentially with EcoRI (cuts above theengineered site) and PvuII (cuts at nucleotide position 1768 [42]) to generate a1,550-bp fragment containing solR, its natural promoter, and the two putativerho-independent terminators downstream of solR (42). This 1.55-kb DNA frag-ment was ligated into EcoRI-SmaI-digested pUC19 vector to yield the ;4.2-kbplasmid pSOLR (Fig. 1a).
(ii) pCO1. pSOLR digested with XbaI and plasmid pIM13 (35) digested withHindIII were then treated with DNA polymerase I large (Klenow) fragment(New England Biolabs), according to the manufacturer’s instructions, in order togenerate blunt-ended termini. The B. subtilis plasmid pIM13 provides MLSr anda gram-positive bacterial origin of replication (35). The two blunt-ended frag-ments (linearized pSOLR and the larger, ;2.0-kb HindIII fragment frompIM13) were then ligated to yield the ;6.2-kb plasmid pCO1 (Fig. 1b).
(iii) pO1X. A 890-bp internal DNA fragment of solR was amplified by PCRwith plasmid (pSOLR) DNA as the template (42). The upstream primerORFX-UP (59-TGCGATATGTAGAATTCTTCCAATATTT-39) was gener-ated by substituting a G for a T at nucleotide position 491 (42) and a T for a Aat nucleotide position 495 to provide an internal EcoRI site (underlined). Thedownstream primer ORFX-DN (59-TTTTTATCATCGAATTCTATGCCTAAAT-39) was generated by substituting a A for a T at nucleotide position 1357 (42)on the complementary strand to provide an internal EcoRI site (underlined).The amplified DNA fragment was digested with EcoRI to generate a 0.9-kbfragment corresponding to bp 492 to 1358 (42) and ligated into EcoRI-digestedpJC4 vector (33) to yield the ;6.2-kb plasmid pO1X (Fig. 1c). DNA sequenceanalysis showed that the insert in pO1X is derived from the solR gene andconfirmed the sequence of this segment of the solR region reported by Fischer etal. (22).
PCR experiments. PCR primers were designed to correspond to the regions ofthe solR gene (see Fig. 4a). The solR453 forward primer (59-GAGTTGAATTTAGCATGAATTTATTA-39; bp 428 to 453) (42), the solR1361 reverse primer(59-AATTTTCCGTTAAGTATTTTTTTATCAT-39; bp 1361 to 1388) (42),primer Tc239 (59-CATAGAAATTGCATCAACGCATA-39; bp 239 to 261) (61)for the tetracycline resistance gene of pO1X, and primer Em373 (59-CAATTGTTTTATTCTTTGGTTGAGTAC-39; bp 373 to 399) (63) for the erythromycinresistance (MLSr) gene of pO1X were synthesized by Genosys (The Woodlands,TX).
The primers were used with C. acetobutylicum DNA isolated from ATCC 824,solR mutant B, and solR mutant H to probe the solR region and insertedsequences (if any) in these strains. The PCR conditions used with primer pairssolR453-Tc239 and Em373-solR1361 were as follows: 13 PCR optimizationbuffer D (Invitrogen, Carlsbad, Calif.), 0.4 mM (each) primer, 250 mM finalconcentrations of each deoxynucleoside triphosphate, 1 ml of template (;0.8 mgof either ATCC 824, SolR-B, or SolR-H DNA), and 1 U of Taq polymerase ina 50-ml reaction mixture volume. The PCR cycling conditions used were an initialdenaturation step (2 min at 94°C) followed by 35 cycles, with each cycle consist-ing of 45 s at 94°C for denaturation and 1 min at 72°C for annealing-extension,
TABLE 1. Bacterial strains and plasmids used in this study
Strain or plasmid Relevant characteristicsa Source orreference
StrainsC. acetobutylicum ATCC
824ATCCb
E. coli ER2275 recA lacZ mcrBC New EnglandBiolabsc
PlasmidspUC19 Apr 72pIM13 MLSr 35pAN1 Cmr f3T I 38pJC4 Tcr MLSr 33pHXS5 Apr aad ctfA ctfB orf1 42pSOLR Apr solR This studypCO1 Apr MLSr solR This studypO1X Tcr MLSr This study
a The relevant characteristics are genotypes or phenotypes. The genes causethe following effect or encode the following protein(s): recA, homologous re-combination abolished; lacZ, b-galactosidase; mcrBC, methylcytosine-specificrestriction system; aad, aldehyde/alcohol dehydrogenase; ctfAB, acetoacetyl co-enzyme A:acetate/butyrate:coenzyme A-transferase subunits; orf1 (now desig-nated solR), putative repressor of sol locus genes. Abbreviations: Apr, ampicillinresistant; MLSr, macrolide, lincosamide, and streptogramin B resistant; Cmr,chloramphenicol resistant; f3T I, f3T methylase; Tcr, tetracycline resistant.
b ATCC, American Type Culture Collection, Manassas, Va.c New England Biolabs, Beverly, Mass.
and a final extension step (5 min at 72°C). When primers solR453 and solR1361were used, Perkin-Elmer’s (Foster City, Calif.) XL (extralong) PCR kit was used.Following the kit instructions, an optimal magnesium acetate level of 1.1 mMand the following cycling times were used: an initial denaturation step (2 min at94°C), 16 cycles with each cycle consisting of 15 s at 94°C for denaturation and5 min at 62°C for annealing and extension, then 12 additional cycles in which thedenaturation conditions remained 94°C for 15 s and the annealing-extension timewas increased successively by 15 s from the corresponding time of the previouscycle, and at the end a final extension step of 10 min at 72°C. PCR products wereanalyzed by gel electrophoresis on agarose gels and staining with ethidiumbromide.
Computer programs. The Wisconsin Genetics Computer Group (17) sequenceanalysis software package (version 9.1, September 1997) was used for programsBestFit, PeptideStructure, PlotStructure, and FindPatterns.
Homology searches using BLAST (release 2.0, September 1997) (1) were doneon the WWW BLAST Server (www.ncbi.nlm.nih.gov). The BLASTP programwas used to search the nr peptide sequence database (all nonredundant Gen-Bank CDS translations, PDB, SwissProt, and PIR).
Additional homology searches were done with the Blocks WWW Server (www.blocks.fhcrc.org) to look for the most highly conserved regions in groups ofproteins. The database searched was BLOCKS (version 10.1, April 1998) (26).Wherever presented, consensus patterns are from the patterns section of PRO-SITE (release 14.0, November 1997) (4) obtained with the ScanProsite tool.
Homology searches were also done on the PRINTS (release 18.0, March 1998)protein motif fingerprint database (www.biochem.ucl.ac.uk) with the Finger-PRINTScan tool to look for conserved motifs characteristic of protein families(3).
RESULTS
Sequence analysis. Upon sequencing upstream of aad onplasmid pHXS5, a 957-bp ORF (orf1, now solR) (42) was lo-cated, which based on homology searches of protein databasesand experimental evidence (this report), appears to encode aputative repressor protein (SolR) involved in negative regula-tion of solvent formation genes. A putative ribosomal bindingsite (59-GGAAAGAG-39), similar in sequence and spacing tothose of other C. acetobutylicum genes (47), was found 11 bpupstream of the solR start codon. Two inverted repeat seg-ments (42) were identified in the region of DNA between solRand aad (DG 5 220.0 kcal/mol [75], positions 1399 to 1439;DG 5 219.4 kcal/mol, positions 1617 to 1657). Northern blotanalysis of the identical gene in C. acetobutylicum DSM 792coding for ORF5 (22) showed two transcripts 1.3 and 1.0 kblong, indicating that both terminator structures are utilized.The solR gene is terminated by a single (UAA) stop codon.
Homology searches. The solR gene codes for a protein(SolR) containing 319 amino acid (aa) residues. The calculatedmolecular mass of the SolR protein is 36,916 Da.
The highest scoring fingerprint obtained via the Finger-PRINTScan performed on the SolR sequence was that of HO-MEOBOX. Most proteins containing homeobox domains areknown to be sequence-specific DNA-binding transcription fac-tors. The domain binds DNA through a helix-turn-helix (HTH)structure (59). HOMEOBOX is a three-element fingerprintthat provides a signature for the homeobox domain. The threeelements identified by FingerPRINTScan within SolR are NAYITRERIYFY (starts at residue 66), LGEPERALKYF (startsat residue 112), and KFKELIAKTK (starts at residue 286).
Proteins containing HTH DNA-binding motifs are charac-teristic of the cyclic AMP (cAMP) receptor protein (CRP)-fumarate and nitrogen regulatory protein (FNR) family ofregulatory proteins, as will be discussed below. Overall, at theamino acid level, SolR exhibits a 20.5% identity (44.9% simi-larity) with the 210-aa CRP (46) from E. coli, a 17.2% identity(43.7% similarity) with the 250-aa FNR (46) from E. coli, anda 19.0% identity (46.0% similarity) with the 219-aa FNR-likeprotein (FLP) (28) from Lactobacillus casei, which is the firstdiscovered member of the CRP-FNR family in a gram-positiveorganism. Figure 2 shows an alignment of a-helical DNArecognition sequences (HTH motif) of 33 DNA-binding pro-teins. The PROSITE (4) consensus pattern [LIVM]-[STAG]-[RHW]-X2-[LI]-[GA]-X-[LIVMFYA]-[LIVS]-G-X-[STAC]-X2-[MT]-X-[GST]-R-X-[LIVMF]-X2-[LIVMF], where the letterswithin the brackets represent the different possible amino acidresidues at each position and the subscript numbers representthe number of occurrences of the indicated residue(s), hasbeen presented for the HTH DNA-binding motif (within theGTR motif) of several repressor proteins. The correspondingputative region in SolR is presented in Fig. 2.
FIG. 1. Schematic representations of the plasmids pSOLR (a), pCO1 (b),and pO1X (c). P is the solR promoter, and T1 and T2 are transcriptionalterminators identified previously (42) downstream of solR. solR9 is the 0.9-kbinternal fragment of solR (see construction of pO1X in Materials and Meth-ods).
VOL. 181, 1999 TRANSCRIPTIONAL REPRESSOR OF SOL LOCUS GENES 321
SolR protein secondary structure. The Wisconsin GeneticsComputer Group sequence analysis software package was usedto predict the SolR secondary structure by the method of Chouand Fasman (11) using programs PeptideStructure and Plot-Structure. The putative DNA-binding site of SolR presented inFig. 2 spans amino acid residues 164 to 187. The Chou-Fasmanmethod predicts a-helical regions at the extreme ends of this24-aa region with the conserved glycine residue (Gly-173) lyingoutside the turn region. Figure 2 clearly shows that DNA-binding regions (HTH motifs) within the CRP-FNR family ofregulatory proteins lie predominantly in the C-terminal regionsof the proteins. The putative DNA-binding region within SolRis present in the C-terminal half of the protein close to thecentral region.
Isolation and characterization of solR mutants B and H. Thesuicide plasmid pO1X was introduced into C. acetobutylicumATCC 824 by electroporation. The resulting transformants
(selected on ERM-containing plates) were grown for 48 to 72 hin CGM tube cultures and then analyzed for product concen-trations. In these tube cultures, two such solR mutants, ATCC824 solR::pO1X (designated mutants B and H), compared tothe wild-type strain, produced ca. threefold-more butanol andacetone, ca. four- to fivefold-more ethanol, but only ca. 0.3- to0.6-fold-more butyrate and acetate. Mutant B produced themost solvents and hence was used in further fermentationstudies.
Total cellular DNA from the integrants (mutants B and H)and parental strain (ATCC 824) was characterized by Southernhybridization. DNA was digested with either ScaI or EcoRIand was then probed with the labeled 0.9-kb EcoRI fragmentfrom pO1X (containing an internal fragment of solR). ScaI wasinitially chosen because this enzyme cuts at a single site in thebackbone of the vector pJC4 but not in the clostridial solRinsert. If a single copy of pO1X integrated into the bacterial
FIG. 2. Amino acid alignment of the DNA-binding HTH domains of the CRP-FNR family of regulatory proteins. Hydrophobic residues (h) and conserved residuesin three different positions (boxed) are indicated at the top. CRP_ECOLI, CRP (catabolite gene activator protein) from E. coli (46); CAP_KLEAE, CAP-like proteinfrom Klebsiella aerogenes (44); CAP_SHIFL and CAP_SALTY, CAP from Shigella flexneri and Salmonella typhimurium, respectively (14); CRP_HAEIN, CRP fromHaemophilus influenzae (10); CLP_XANCA, CAP-like protein from Xanthomonas campestris (16); CRP_SALTY, CRP from S. typhimurium (58); BifA_ANABA,CRP-like protein from Anabaena sp. (67); CysR_SYNEC, regulatory protein from Synechococcus sp. (31); Hin, product of DNA inversion gene (46); TnpR_Tn3, Tn3resolvase (46); Resolvase_gd, resolvase from transposon gd (46); AraC_ECOLI, arabinose regulatory protein from E. coli (46); ftz_DROME, product of segmentationgene fushi tarazu from Drosophila melanogaster (32); smox-2_SCHMA, Schistosoma mansoni homeobox-containing DNA-binding protein (66); MATa1_YEAST andMATa2_YEAST, Saccharomyces cerevisiae proteins that specify a and a diploid functions including sporulation (32); Cro_434, regulatory protein from bacteriophage434 (46); C2_P22, repressor from bacteriophage P22 (56); CI_f80 and Gene30_f80, DNA-binding proteins from bacteriophage f80 (43); CII_l, regulatory proteinfrom bacteriophage l (46); LacR_ECOLI, lactose repressor from E. coli (46); GalR_ECOLI, galactose repressor from E. coli (46); CI_l, repressor from bacteriophagel (55); Cro_l, regulatory protein from bacteriophage l (46); FNR_ECOLI, FNR from E. coli (46); HlyX_ACTPL, regulatory protein from Actinobacillus pleuropneu-moniae (34); AadR_RHOPA, FNR-CRP member from Rhodopseudomonas palustris (18); FixK_RHILE, FNR-like protein from Rhizobium leguminosarum (12);FixK_RHIME, FNR-CRP member from Rhizobium meliloti (5); FixK_BRAJA, FNR-like protein from Bradyrhizobium japonicum (2); FLP_LACCA, FNR-like proteinfrom Lactobacillus casei (28); SolR_CLOAC, putative repressor protein from Clostridium acetobutylicum. Numbers preceding sequences represent amino acid positionswithin each protein. Previously reported (32, 46) consensus sequence alignments were used.
DNA, digestion with ScaI should generate two solR-hybridiz-ing fragments, whose combined size equals the combined sizeof the integrational plasmid (pO1X) and the ScaI fragment onthe parental DNA that contains the homologous gene. EcoRIbacterial DNA fragments were similarly analyzed, and Fig. 3combines all observations regarding mutants B and H and theparental ATCC 824 strain. The physical map of the clostridialinsert in plasmid pHXS5 (42) earlier showed an EcoRI site lessthan 1 kb upstream of solR between the XbaI and ScaI sites.More-recent restriction analysis of this plasmid showed thatthis EcoRI site is actually absent in this region and that theEcoRI restriction site observed earlier in this vicinity was ac-tually the site in the pUC19 polylinker adjacent to this end ofthe insert. Also, sequencing of the solR9 region in pO1X re-vealed two errors in the sequence reported earlier (42) be-tween nucleotides 220 to 239 in the solR structural gene. Con-sequently, a BglII site predicted by the earlier (42) sequence ofsolR in this region is actually absent. These sequence errorsand a few others have been corrected in the GenBank entry(accession no. L14817) published previously. The actual nucle-otide positions mentioned throughout this article are based onthe previously published sequence (42) and may vary slightlyfrom those based on the revised GenBank sequence entry.
In the ScaI digestions shown in Fig. 3A, the probe hybridizedto the expected 6.2-kb band generated by integrational plasmidpO1X (lanes a and e), a single fragment from the parentalstrain (with a migration position of 7.6 kb [lane b]), and twofragments (with migration positions of 6.4 and 7.4 kb) eachfrom mutant B (lane c) and mutant H (lane d). Since thecombined size of the ScaI fragments from the two mutants, Band H, equals the size of the integrational plasmid and the solRfragment on the bacterial DNA, it appears that a single copy ofpO1X was integrated into the bacterial DNA via single-cross-over homologous recombination. However, the similarity ofthe sizes of the bands in mutants B and H (;7.4 and 6.4 kb) tothose of the parent strain (;7.6 kb) and plasmid pO1X (;6.2kb) limits the ability to precisely map or distinguish the pres-ence of an additional band which would have resulted from anadditional in-tandem insertion of pO1X. It is possible that theretardation of bands may be due to an artifact; therefore,EcoRI digestion and PCR analysis were also conducted tobetter analyze the position of insertion of pO1X.
Analysis of EcoRI digests also shows disruption of the nativefragment containing solR. In the EcoRI digestions shown inFig. 3B, the probe hybridized to the expected ;0.9-kb frag-ment from the integrational plasmid pO1X (lane a), the ex-pected one ;9 kb from the parental strain (lane b), threefragments from mutant B (0.9, 3, and 7 kb [lane c], and twofragments from mutant H (3 and 7 kb [lane d]. Since theinternal EcoRI fragment of pO1X is ;0.9 kb, these findingssuggest that mutant B contains at least an additional copy ofpO1X inserted tandemly. Such additional inserts are fre-quently found in these types of recombination events (40) andare due to a tandem insertion.
The DNA isolated from solR mutants H and B was alsoanalyzed by PCR amplification using primers designed to am-plify the junction between the vector part of pO1X and theends of the solR gene not present on the plasmid. First, the useof the solR453 and solR1361 primers gave, as expected, a;0.9-kb fragment with C. acetobutylicum ATCC 824 DNA asthe template; however, small amounts of this fragment werefound with DNA from mutants H and B also (Fig. 4b, gel A).The low intensity of the ;0.9-kb fragment with templates pre-pared from mutants H and B may result from amplification ofa small contamination or the presence of a small population ofrevertant cells in the sample which have lost the plasmid byexcision which occurs at a low frequency. Another possibility isthat the integration events occurred in only a fraction of thepSOL1 plasmids in a cell; the copy number of pSOL1, whichcarries the sol locus, is not yet known. A more intense fragmentof ;7 kb was also found with mutant H, consistent with oneinsert of the pO1X plasmid into mutant H (Fig. 4b, gel A). Inthe case of mutant B, the large fragment (of at least 13 kb,which would be expected if one additional copy of pO1X wastandemly inserted) was not found. Apparently, due to its size itwas not produced well in the PCR amplification even whenextralong PCR conditions were used. Second, the junctionfragments of solR with the integrated vector, as indicated bypositive amplification of the expected-size fragment with thesolR453-Tc239 primer pair and the Em373-solR1361 primerpair, were the same when DNA from both mutant, B and Hwas used as the template (Fig. 4b, gels B and C). No amplifi-cation was observed when these primer pairs were used withATCC 824 DNA. Additional evidence for a tandem insertionwas found when the Em373 and Tc239 primer pairs were usedwith mutant B DNA. In this case, a fragment of 2.2 kb wasobtained (data not shown), indicating that these genes areoriented as they would be in a tandem insertion. Such a frag-ment was not found with mutant H (data not shown), theexpected result for a single integration mutant. These com-bined results show that the pO1X insertions in mutants B andH were within the solR gene and the solR gene would beexpected to be nonfunctional as a result of these insertions. Ofcourse, we cannot rule out the possible presence of alterationsin other genes which were not investigated within the derivedstrains.
Batch fermentation studies. Batch fermentations of C. ace-tobutylicum ATCC 824(pCO1) and mutant B were performedat a pH of $4.5 (Fig. 5), and the final product concentrationswere compared to those obtained earlier with fermentations ofwild-type and recombinant (carrying plasmid pCCL) strains(42) at the same pH (Table 2). Plasmid pCCL carries a trun-cated form of aad, and ATCC 824(pCCL) fermentations pro-duce higher solvent titers than those of the wild-type strain forreasons which are still unknown. These data show that over-expression of solR in strain ATCC 824 [strain ATCC824(pCO1)] results in loss of butanol and acetone, while eth-anol production was reduced [five- to sevenfold from that of
FIG. 3. Southern analysis. Hybridization of a 0.9-kb solR fragment to ScaI- orEcoRI-digested bacterial DNA from C. acetobutylicum mutant B (lanes c), mu-tant H (lanes d), wild-type (lanes b), and plasmid pO1X (lanes a and e). Sizemarkers (in kilobase pairs) are BstEII-digested l DNA.
VOL. 181, 1999 TRANSCRIPTIONAL REPRESSOR OF SOL LOCUS GENES 323
FIG. 4. Schematic representations (a) and results of PCR analysis (b) on wild-type (WT) C. acetobutylicum ATCC 824 and solR mutants B and H using primers(a) designed to amplify the junction between the vector portion of pO1X and the solR gene. For each gel in panel b, lane 1 contains HindIII-digested lambda marker,lane 2 contains the WT genomic template, lane 3 contains the solR mutant B template, and lane 4 contains the solR mutant H as the template. (Gel A) Extralong PCRusing primers solR453 and solR1361 designed to amplify the complete insert. In lane 2, WT DNA shows an expected ;0.9-kb band. This band is also seen, but muchweaker, in both lanes 3 and 4. In addition, lane 4 contains a band with an apparent size of ;7 kb (marked with an arrow), consistent with the presence of one insertof pO1X into solR of mutant H. (Gel B) PCR results using primers solR453 and Tc238. A band of ;1.2 kb can be seen in lanes 3 and 4 with no product in lane 2.(Gel C) PCR results using primers Em373 and solR1361. A band of ;2.1 kb is visible in lanes 3 and 4. Again, no product was observed with WT DNA (lane 2).
the control strains [ATCC 824 and ATCC 824(pCCL)]. As aresult, considerably larger amounts of butyrate and acetate[2.5- to 15.0-fold and 1.5- to 6.4-fold increase, respectively,over the final levels produced by the control strains ATCC 824and ATCC 824(pCCL)] accumulated.
Inactivation of solR (mutant B) led to higher butanol andacetone levels. There was a 3.2-fold and a 1.7-fold increase inthe final butanol concentration over those of the wild-type andATCC 824(pCCL) strains, respectively. There was a 2.1-foldand 1.5-fold increase in the corresponding levels of acetone.While butyrate and acetate levels were only half that of thewild-type strain, there was a 3.7-fold increase in the final levelsof butyrate over that of strain ATCC 824(pCCL). In the mu-tant B fermentation, earlier induction of the AAD protein(results not shown) resulted in an increased flux towards bu-tanol formation and a decreased flux towards acid accumula-tion early on, and consequently, acid uptake did not contributeas significantly to solvent formation. Final solvent titers were17.8 g of butanol per liter, 8.1 g of acetone per liter, and 1.0 gof ethanol per liter for a net solvent titer of ;27.0 g/liter.Mutants B and H were also tested in several other fermenta-tions which produced similar high solvent titers.
Earlier induction of solventogenic genes also leads to a moreefficient conversion of glucose to solvents. The molar yields ofbutanol and acetone on glucose (final concentration [millimo-lar] of solvent/initial glucose concentration [millimolar]), formutant B were 0.54 and 0.32, respectively, while those for thewild-type strain were only 0.17 and 0.15 and those for strainATCC 824(pCCL) were 0.31 and 0.21, respectively. StrainATCC 824(pCO1) produced no detectable butanol or acetone.
Transcriptional start site of solR. Primer extension analysisusing primer BORFU-PE had a two-fold objective: (i) to iden-tify the precise 59 end of the solR transcript and (ii) to comparetranscriptional levels of solR in the wild-type (ATCC 824)strain and the recombinant [ATCC 824(pCO1)] strain thatoverexpresses solR. Results with mRNA from both wild-typeand ATCC 824(pCO1) cells obtained from early exponential(stage A, 5 h) and late exponential (stage B, 10 h) growthstages (Fig. 6) show that solR mRNA is present at very lowlevels in the wild-type strain with bands in lanes 1 and 2 barelyvisible despite starting with twice the standard amount of totalRNA (40 mg).
The solR transcriptional start site is shown in Fig. 6 at nu-cleotide position 407 (42), which is 35 nucleotides upstreamfrom the initiation codon with C as the first transcribed nucle-otide. Looking further upstream of this start site, a putativepromoter structure TCGATA(17 bp)TATTAT was identifiedwith 7 bp separating the 210 region of the promoter structureand the transcriptional start site. The 210 and 235 sequencesare similar to those identified previously (73) as part of aconsensus clostridial promoter with differences in only 3 of 12positions. The location of the solR transcriptional start site andpromoter region are in agreement with the results obtained foran identical gene coding for ORF5 in C. acetobutylicum DSM792 (22).
Overexpression of solR in strain ATCC 824(pCO1) is clearlyevident in lanes 7 and 8 (Fig. 6). Based on relative bandintensities, it appears that solR is transcribed more actively inacidogenic cells (lane 7) than in solventogenic cells (lane 8).The low transcriptional levels of solR in wild-type cells makesuch a direct comparison based on relative band intensitiesmore difficult. However, careful examination of the faint bandsin lanes 1 and 2 (Fig. 6) appears to indicate that this could alsobe true in wild-type cells.
Expression of aad, ctfA, ctfB, and adc in a strain overpro-ducing SolR. Primer extension analysis (Fig. 7) was used to
FIG. 5. Product concentration and optical density (OD) profiles for con-trolled-pH (pH $ 4.5) batch fermentations with C. acetobutylicum strain ATCC824(pCO1) (a) and mutant B (b). Zero time indicates the time at which thebioreactor was inoculated with a 1/10 (vol/vol) preculture. Symbols: µ, glucose;F, ethanol; Œ, acetone; ■, butanol; E, acetate; h, butyrate; ‚, optical density at600 nm; J, pH.
VOL. 181, 1999 TRANSCRIPTIONAL REPRESSOR OF SOL LOCUS GENES 325
examine the effects of SolR overproduction on the transcriptlevels of the solventogenic genes aad, ctfA, ctfB, and adc (sollocus). Since aad, ctfA, and ctfB form a polycistronic operon(22), the transcriptional levels of aad and ctfAB are identical.Based on the locations of the transcriptional start sites of solR(this study), aad (42), and adc (24) and the positions of primersBORFU-PE, BYDH-PE, and N-ADC, the primer extensionproducts for solR and adc are 77 and 105 bp, respectively, whilethose of aad are 103 and 263 bp, respectively (the shorterproduct predominates as seen earlier [42]). Figure 7 shows thetranscript levels corresponding to solR (lanes 1 to 4), aad (lanes5 to 8), and adc (lanes 9 to 12) at four different stages of theATCC 824(pCO1) fermentation. solR appears to be stronglyinduced in the early exponential growth phase (Fig. 7, lane 1)with continually decreasing transcription levels during the restof the fermentation (Fig. 7, lanes 2 to 4). The primer extensiongel (shown all the way to the top up to the loading wells in Fig.7) shows that aad and adc (which are normally inducedstrongly at the end of the exponential growth phase, whichcorresponds to the onset of solvent formation [24, 41]) are nottranscribed (Fig. 7, lanes 5 to 12). This appears to be a directresult of overexpression of solR. While homology searchessuggested that SolR was perhaps a regulatory protein, thisexperiment offers the first tentative link between SolR over-production and transcriptional shutdown of solventogenicgenes. This suggests that SolR is a putative repressor proteinthat regulates the transcription of butanol and acetone forma-tion genes of the sol locus (aad, ctfA, ctfB, and adc genes).
Expression of aad and adc in the solR-inactivated mutant B.Samples from stages A, B, C, and D of a fermentation of C.acetobutylicum mutant B were used for primer extension anal-ysis (Fig. 8) to examine the effects of solR inactivation on thetranscript levels of aad and adc. Faint bands corresponding toa 77-bp primer extension product are barely visible in lane 1(wild type) and in lanes 2 to 5 (mutant B). This implies that atranscript corresponding to solR exists in mutant B, but aNorthern analysis would be needed to determine the alteredsize of this transcript due to insertional inactivation of solR.Strong expression of aad (Fig. 8, lanes 7 to 9) and adc (Fig. 8,lanes 11 and 12) is apparent. The expected 103- and 263-bpbands for aad and the 105-bp adc band are clearly visible (Fig.8). Apparently, solR inactivation elevates predominantly theaad transcriptional levels from the proximal strong (42) pro-moter (103-bp band). The intensity of the 263-bp band corre-sponding to transcription from the distal promoter is probablyidentical to that in the wild-type strain (42).
Alteration of solR transcription in mutant B. Northern anal-ysis (Fig. 9) was used to compare solR transcript sizes fromwild-type and mutant B cells, since the presence of a transcriptin mutant B at levels comparable to that in the wild-type cellswas evident (Fig. 8, lanes 1 to 5). Based on the location of themapped solR transcriptional start site and the locations of thetwo terminator structures (42), both of which structures areused (22), the expected solR transcript sizes are 1.01 and 1.23kb. In wild-type cells, a broad band centered at about 1.15 kb
is observed (Fig. 9, lane 1) and this could be accounted for bythe overlapping of the two expected solR transcripts. The cor-responding band in mutant B appears as a smear centered atabout 4.47 kb (Fig. 9, lane 2). From Fig. 4a, it is apparent thechromosomal solR promoter can generate a transcript whichwould contain the solR9 region and since the native solR tran-scription terminator is not present in the insert of pO1X, thetranscript would continue into the vector where it may termi-nate near the ori or the MLSr gene. The longer transcript isunlikely to arise from transcription of solR9 from the MLSr
gene promoter as this promoter is oriented in the directionopposite that of the solR9 insert.
DISCUSSION
The top-scoring fingerprint match to SolR from the PRINTSdatabase was that of a class of transcription factors with HTHDNA-binding motifs. A putative DNA-binding motif in SolRhas been identified based on the corresponding region in theCRP-FNR family of regulatory proteins among others (Fig. 2).A Chou-Fasman secondary structure prediction of SolR con-firmed this to be a HTH region. This is evidence that SolR isa DNA-binding transcriptional regulator. SolR overexpressionand inactivation studies presented here back up this contentionwhile qualifying the regulatory role of SolR to be that of arepressor of butanol and acetone formation genes.
A search of protein databases using the BLASTP programrevealed homology (24.7% identity and 48.1% similarity), es-pecially in the C-terminal region, of SolR to Spo0KA (545 aa,61.5 kDa) from B. subtilis (52), an oligopeptide permease re-quired for sporulation and competence. Another gene involvedin initiation of sporulation in B. subtilis is spo0A (responseregulator, transcription repressor/activator) (27). Spo0A (267aa, 29.7 kDa) from B. subtilis (21), a DNA-binding protein thatcontrols the expression of genes that are involved in the tran-sition from growth to the stationary phase, is activated byphosphorylation and has two tightly folded domains, an N-terminal phosphorylation domain and a C-terminal DNA-binding domain with specificity for the 0A box 59-TGNCGAA-39 (25, 27, 52). A search of the 4,797-bp sequence in therevised GenBank entry (accession no. L14817 revised from the4,800-bp sequence published previously [42]) for putative 0Aboxes using the FindPatterns program (allowing one mis-match), as done earlier (68), located one such sequence in thesolR-aad intergenic region (sequence 59-TGGCGTA-39 on thenoncoding strand ending at nucleotide position 1712 of therevised GenBank sequence entry), with 16 other sequenceslocated within the coding regions of either solR or aad oneither strand. A BestFit alignment of SolR and Spo0A from B.subtilis (21) (15.7% identity and 42.1% similarity) revealed thatthe putative HTH motif in SolR (Fig. 2) is aligned to a similarsequence, including the conserved Gly residue (Gly-165 inSpo0A [21]), present within the region in Spo0A that has beenimplicated in DNA binding. Putative HTH motifs believed tobe responsible for binding 0A boxes have been identified in
TABLE 2. Final product levels in C. acetobutylicum fermentor experiments at pH 4.5
C. acetobutylicumATCC 824
Final concn (mM)Reference
Butanol Acetone Ethanol Butyrate Acetate
Wild-type 74 66 15 36 53 42Harboring pCCL 138 95 21 6 12 42Harboring pCO1 0 0 3 90 77 This studyMutant B 240 140 21 22 29 This study
spo0A gene products from Bacillus and Clostridium sp. (8). TheSolR protein shares a 20.1% identity (42.9% similarity) with a166-aa fragment of the Spo0A protein (GenBank accession no.U09978) from C. acetobutylicum ATCC 4259 and a 18.2%
identity (43.0% similarity) with a 223-aa fragment from theSpo0A protein (GenBank accession no. U09979) from C. bei-jerinckii (formerly C. acetobutylicum) NCIMB 8052.
After weighing all available evidence, experimental (Fig. 5,
FIG. 6. Primer extension analysis. Primer extension products made with primer BORFU-PE complementary to the N-terminal end of solR are shown. RNA forthese experiments was obtained from C. acetobutylicum ATCC 824 cells (lanes 1 and 2) and from C. acetobutylicum ATCC 824(pCO1) cells (lanes 7 and 8). Cell sampleswere collected during early exponential (stage A, 5 h) and late exponential (stage B, 10 h) growth phases. The late exponential growth phase is also the solventogenicphase of ATCC 824 cells. Regions of the plasmid pCO1 were sequenced with the same primer, and the resulting DNA sequences are shown in lanes 3 to 6. Thecorresponding 59-to-39 DNA sequence of the complementary (coding) strand is indicated to the right of the gel (sequence continues onto a second line), wherein theboxed nucleotide is the transcriptional start site. The arrow indicates the position of the nearly invisible bands in lanes 1 and 2. The total RNA (20 mg) used for lanes7 and 8 corresponding to strain ATCC 824(pCO1) was the same as that loaded in primer extension reactions performed previously (42) to map the transcriptional startsite of aad. However, low transcriptional levels are characteristic of regulatory proteins, since they are present in low copy numbers and hence, on this suspicion, twicethe standard amount of total RNA (40 mg) was used for lanes 1 and 2 (corresponding to wild-type strain ATCC 824) in order to obtain visible bands.
VOL. 181, 1999 TRANSCRIPTIONAL REPRESSOR OF SOL LOCUS GENES 327
7, and 8) and theoretical, it appears that SolR is a putativeDNA-binding transcriptional repressor that negatively regu-lates the onset of solventogenic (primarily butanol and ace-tone) metabolism. Induction of this putative repressor in wild-type cells during the acidogenic (early to late exponentialgrowth) phase and considerably lower expression during thesolventogenic (beyond late exponential growth) phase wouldensure the well-known induction pattern of solventogenicgenes (beginning during the late exponential growth phase),which is in keeping with the proposed role of SolR. The in-ducer responsible for derepression of solventogenic genes
could be one of several factors that are believed to triggersolventogenesis, including pH, threshold butyrate concentra-tions (62), and nutrient limitations.
The fermentation of mutant strain B, without any effort foroptimization towards increased solvent yields, is one of themost impressive ever reported in the literature in terms ofsolvent production (27.0 g of total solvents per liter) and bu-tanol tolerance (17.8 g/liter). This performance (Fig. 5) wouldmake this genetically characterized strain quite attractive in-
FIG. 7. Time course primer extension analysis of ATCC 824(pCO1) cells.RNA for the time course primer extension experiments was isolated from ATCC824(pCO1) cells isolated during the early exponential growth (stage A, 5 h), lateexponential growth (stage B, 10 h), early stationary (stage C, 25 h), and latestationary (stage D, 50 h) stages in a batch fermentation with a controlled pH(pH 4.5). The presence of mRNA corresponding to solR (lanes 1 to 4) aad (lanes5 to 8) and adc (lanes 9 to 12) genes in each of the above four stages was verifiedby performing primer extension reactions with 20 mg of total RNA, using end-labeled 20-mer synthetic oligonucleotides BORFU-PE, BYDH-PE, and N-ADCthat are complementary to the N-terminal ends of the respective genes.
FIG. 8. Time course primer extension analysis of mutant B cells. RNA forthe time course primer extension experiments was isolated from mutant B cellssampled from the early exponential growth (stage A, 5 h), late exponentialgrowth (stage B, 10 h), early stationary (stage C, 25 h), and late stationary (stageD, 50 h) stages in a batch fermentation with a controlled pH (pH $ 4.5). An earlyexponential growth phase (stage A) sample from an ATCC 824 pH 4.5 fermen-tation was examined for the solR transcript (lane 1). The presence of mRNAcorresponding to solR (lanes 2 to 5), aad (lanes 6 to 9), and adc (lanes 10 to 13)genes in each of the above four stages was verified by performing primer exten-sion reactions with 20 mg of total RNA, using end-labeled 20-mer syntheticoligonucleotides BORFU-PE, BYDH-PE, and N-ADC that are complementaryto the N-terminal ends of the respective genes.
dustrially (69). The results presented here indicate that earlierinduction of solventogenic genes (deregulated solvent produc-tion) as opposed to overexpression of the same genes in thesolventogenic phase (37, 64) is essential for generating indus-trially significant solvent-producing strains. So far, manipulat-ing one gene (like solR) with a global effect appears to be themost-effective approach for strain improvement to increasesolvent yields and titers.
ACKNOWLEDGMENTS
This research was supported by NSF grants BES-9632217 and BES-9604562.
We thank Neil Welker (Northwestern University) for constructivediscussions and suggestions.
REFERENCES
1. Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang, W. Miller,and D. J. Lipman. 1997. Gapped BLAST and PSI-BLAST: a new generationof protein database search programs. Nucleic Acids Res. 25:3389–3402.
2. Anthamatten, D., B. Scherb, and H. Hennecke. 1992. Characterization of afixLJ-regulated Bradyrhizobium japonicum gene sharing similarity with theEscherichia coli fnr and Rhizobium meliloti fixK genes. J. Bacteriol. 174:2111–2120.
3. Attwood, T. K., M. E. Beck, D. R. Flower, P. Scordis, and J. N. Selley. 1998.The PRINTS protein fingerprint database in its fifth year. Nucleic Acids Res.26:304–308.
4. Bairoch, A. 1991. PROSITE: a dictionary of sites and patterns in proteins.Nucleic Acids Res. 19(Suppl.):2241–2245.
5. Batut, J., M.-L. Daveran-Mingot, M. David, J. Jacobs, A. M. Garnerone, andD. Kahn. 1989. fixK, a gene homologous with fnr and crp from Escherichiacoli, regulates nitrogen fixation genes both positively and negatively in Rhi-zobium meliloti. EMBO J. 8:1279–1286.
6. Bennett, G. N., and F. B. Rudolph. 1995. The central metabolic pathwayfrom acetyl-CoA to butyryl-CoA in Clostridium acetobutylicum. FEMS Mi-crobiol. Rev. 17:241–249.
7. Birnboim, H. C., and J. Doly. 1979. A rapid alkaline extraction procedure forscreening recombinant plasmid DNA. Nucleic Acids Res. 7:1513–1523.
8. Brown, D. P., L. Ganova-Raeva, B. D. Green, S. R. Wilkinson, M. Young, andP. Youngman. 1994. Characterization of spo0A homologues in diverse Ba-cillus and Clostridium species reveals regions of high conservation within theeffector domain. Mol. Microbiol. 14:411–426.
9. Cary, J. W., D. J. Petersen, E. T. Papoutsakis, and G. N. Bennett. 1988.Cloning and expression of Clostridium acetobutylicum phosphotransbutyry-lase and butyrate kinase genes in Escherichia coli. J. Bacteriol. 170:4613–4618.
10. Chandler, M. S. 1992. The gene encoding cAMP receptor protein is requiredfor competence development in Haemophilus influenzae Rd. Proc. Natl.Acad. Sci. USA 89:1626–1630.
11. Chou, P. Y., and G. D. Fasman. 1974. Prediction of protein conformation.Biochemistry 13:222–245.
12. Colonna-Romano, S., W. Arnold, A. Schluter, P. Boistard, A. Puhler, andU. B. Priefer. 1990. An Fnr-like protein encoded in Rhizobium leguminosa-rum biovar viciae shows structural and functional homology to Rhizobiummeliloti FixK. Mol. Gen. Genet. 223:138–147.
13. Cornillot, E., R. V. Nair, E. T. Papoutsakis, and P. Soucaille. 1997. The
genes for butanol and acetone formation in Clostridium acetobutylicumATCC 824 reside on a large plasmid whose loss leads to degeneration of thestrain. J. Bacteriol. 179:5442–5447.
14. Cossart, P., E. A. Groisman, M.-C. Serre, M. J. Casadaban, and B. Gicquel-Sanzey. 1986. crp genes of Shigella flexneri, Salmonella typhimurium, andEscherichia coli. J. Bacteriol. 167:639–646.
15. Davison, S. P., J. D. Santangelo, S. J. Reid, and D. R. Woods. 1995. AClostridium acetobutylicum regulator gene (regA) affecting amylase produc-tion in Bacillus subtilis. Microbiol. 141:989–996.
16. de Crecy-Lagard, V., P. Glaser, P. Lejeune, O. Sismeiro, C. E. Barber, M. J.Daniels, and A. Danchin. 1990. A Xanthomonas campestris pv. campestrisprotein similar to catabolite activation factor is involved in regulation ofphytopathogenicity. J. Bacteriol. 172:5877–5883.
17. Devereux, J., P. Haeberli, and O. Smithies. 1984. A comprehensive set ofsequence analysis programs for the VAX. Nucleic Acids Res. 12:387–395.
18. Dispensa, M., C. T. Thomas, M.-K. Kim, J. A. Perrotta, J. Gibson, and C. S.Harwood. 1992. Anaerobic growth of Rhodopseudomonas palustris on 4-hy-droxybenzoate is dependent on AadR, a member of the cyclic AMP receptorprotein family of transcriptional regulators. J. Bacteriol. 174:5803–5813.
19. Dower, W. J., J. F. Miller, and C. W. Ragsdale. 1988. High efficiency trans-formation of Escherichia coli by high voltage electroporation. Nucleic AcidsRes. 16:6127–6145.
20. Durre, P., R.-J. Fischer, A. Kuhn, K. Lorenz, W. Schreiber, B. Sturzen-hofecker, S. Ullmann, K. Winzer, and U. Sauer. 1995. Solventogenic en-zymes of Clostridium acetobutylicum: catalytic properties, genetic organiza-tion, and transcriptional regulation. FEMS Microbiol. Rev. 17:251–262.
21. Ferrari, F. A., K. Trach, D. LeCoq, J. Spence, E. Ferrari, and J. A. Hoch.1985. Characterization of the spo0A locus and its deduced product. Proc.Natl. Acad. Sci. USA 82:2647–2651.
22. Fischer, R. J., J. Helms, and P. Durre. 1993. Cloning, sequencing, andmolecular analysis of the sol operon of Clostridium acetobutylicum, a chro-mosomal locus involved in solventogenesis. J. Bacteriol. 175:6959–6969.
23. Gerischer, U., and P. Durre. 1990. Cloning, sequencing, and molecularanalysis of the acetoacetate decarboxylase gene region from Clostridiumacetobutylicum. J. Bacteriol. 172:6907–6918.
24. Gerischer, U., and P. Durre. 1992. mRNA analysis of the adc gene region ofClostridium acetobutylicum during the shift to solventogenesis. J. Bacteriol.174:426–433.
25. Grimsley, J. K., R. B. Tjalkens, M. A. Strauch, T. H. Bird, G. B. Spiegelman,Z. Hostomsky, J. M. Whiteley, and J. A. Hoch. 1994. Subunit compositionand domain structure of the Spo0A sporulation transcription factor of Ba-cillus subtilis. J. Biol. Chem. 269:16977–16982.
26. Henikoff, S., and J. G. Henikoff. 1991. Automated assembly of protein blocksfor database searching. Nucleic Acids Res. 19:6565–6572.
27. Hoch, J. A. 1993. Regulation of the phosphorelay and the initiation ofsporulation in Bacillus subtilis. Annu. Rev. Microbiol. 47:441–465.
28. Irvine, A. S., and J. R. Guest. 1993. Lactobacillus casei contains a member ofthe CRP-FNR family. Nucleic Acids Res. 21:753.
29. Johnson, J. L., J. Toth, S. Santiwatanakul, and J. S. Chen. 1997. Cultures of“Clostridium acetobutylicum” from various collections comprise Clostridiumacetobutylicum, Clostridium beijerinckii, and two other distinct types based onDNA-DNA reassociation. Int. J. Syst. Bacteriol. 47:420–424.
30. Keis, S., C. F. Bennett, V. K. Ward, and D. T. Jones. 1995. Taxonomy andphylogeny of industrial solvent-producing clostridia. Int. J. Syst. Bacteriol.45:693–705.
31. Laudenbach, D. E., and A. R. Grossman. 1991. Characterization and mu-tagenesis of sulfur-regulated genes in a cyanobacterium: evidence for func-tion in sulfate transport. J. Bacteriol. 173:2739–2750.
32. Laughon, A., and M. P. Scott. 1984. Sequence of a Drosophila segmentationgene: protein structure homology with DNA-binding proteins. Nature 310:25–31.
33. Lee, S. Y., G. N. Bennett, and E. T. Papoutsakis. 1992. Construction ofEscherichia coli-Clostridium acetobutylicum shuttle vectors and transforma-tion of Clostridium acetobutylicum strains. Biotechnol. Lett. 14:427–432.
34. MacInnes, J. I., J. E. Kim, C.-J. Lian, and G. A. Soltes. 1990. Actinobacilluspleuropneumoniae hlyX gene homology with the fnr gene of Escherichia coli.J. Bacteriol. 172:4587–4592.
35. Mahler, I., and H. O. Halvorson. 1980. Two erythromycin-resistance plas-mids of diverse origin and their effect on sporulation in Bacillus subtilis.J. Gen. Microbiol. 120:259–263.
36. McLaughlin, J. K., C. L. Meyer, and E. T. Papoutsakis. 1984. Gas chroma-tography and gateway sensors for on-line state estimation of complex fer-mentations (butanol/acetone fermentation). Biotechnol. Bioeng. 27:1246–1257.
37. Mermelstein, L. D., E. T. Papoutsakis, D. J. Petersen, and G. N. Bennett.1993. Metabolic engineering of Clostridium acetobutylicum for increasedsolvent production by enhancement of acetone formation enzyme activitiesusing a synthetic acetone operon. Biotechnol. Bioeng. 42:1053–1060.
38. Mermelstein, L. D., and E. T. Papoutsakis. 1993. In vivo methylation inEscherichia coli by Bacillus subtilis phage f3T I methyltransferase to protectplasmids from restriction upon transformation of Clostridium acetobutylicumATCC 824. Appl. Environ. Microbiol. 59:1077–1081.
FIG. 9. Northern analysis. Total RNA was prepared from exponential-phaseATCC 824 (lane 1) and mutant B (lane 2) cells from pH $ 4.5 fermentations andprobed with the solR internal fragment from pO1X.
VOL. 181, 1999 TRANSCRIPTIONAL REPRESSOR OF SOL LOCUS GENES 329
39. Mermelstein, L. D., N. E. Welker, G. N. Bennett, and E. T. Papoutsakis.1992. Expression of cloned homologous fermentative genes in Clostridiumacetobutylicum ATCC 824. Bio/Technology 10:190–195.
40. Monedero, V., M. J. Gosalbes, and G. Perez-Martinez. 1997. Cataboliterepression in Lactobacillus casei ATCC 393 is mediated by CcpA. J. Bacte-riol. 179:6657–6664.
41. Nair, R. V. 1995. Ph.D. thesis. Northwestern University, Evanston, Ill.42. Nair, R. V., G. N. Bennett, and E. T. Papoutsakis. 1994. Molecular charac-
terization of an aldehyde/alcohol dehydrogenase gene from Clostridium ace-tobutylicum ATCC 824. J. Bacteriol. 176:871–885.
43. Ogawa, T., H. Ogawa, and J.-I. Tomizawa. 1988. Organization of the earlyregion of bacteriophage f80. Genes and proteins. J. Mol. Biol. 202:537–550.
44. Osuna, R., and R. A. Bender. 1991. Klebsiella aerogenes catabolite geneactivator protein and the gene encoding it (crp). J. Bacteriol. 173:6626–6631.
45. Oultram, J. D., M. Loughlin, T.-J. Swinfield, J. K. Brehm, D. E. Thompson,and N. P. Minton. 1988. Introduction of plasmids into whole cells of Clos-tridium acetobutylicum by electroporation. FEMS Microbiol. Lett. 56:83–88.
46. Pabo, C. O., and R. T. Sauer. 1984. Protein-DNA recognition. Annu. Rev.Biochem. 53:293–321.
47. Papoutsakis, E. T., and G. N. Bennett. 1993. Cloning, structure, and expres-sion of acid and solvent pathway genes of Clostridium acetobutylicum, p.157–199. In D. R. Woods (ed.), The clostridia and biotechnology. Butter-worth-Heinemann, Stoneham, Mass.
48. Petersen, D. J., and G. N. Bennett. 1990. Purification of acetoacetate decar-boxylase from Clostridium acetobutylicum ATCC 824 and cloning of theacetoacetate decarboxylase gene in Escherichia coli. Appl. Environ. Micro-biol. 56:3491–3498.
49. Petersen, D. J., J. W. Cary, J. Vanderleyden, and G. N. Bennett. 1993.Sequence and arrangement of genes encoding enzymes of the acetone-production pathway of Clostridium acetobutylicum ATCC 824. Gene 123:93–97.
50. Petersen, D. J., R. W. Welch, F. B. Rudolph, and G. N. Bennett. 1991.Molecular cloning of an alcohol (butanol) dehydrogenase gene cluster fromClostridium acetobutylicum ATCC 824. J. Bacteriol. 173:1831–1834.
51. Roos, J. W., K. McLaughlin, and E. T. Papoutsakis. 1985. The effect of pHon nitrogen supply, cell lysis, and solvent production in fermentations ofClostridium acetobutylicum. Biotechnol. Bioeng. 27:681–694.
52. Rudner, D. Z., J. R. LeDeaux, K. Ireton, and A. D. Grossman. 1991. Thespo0K locus of Bacillus subtilis is homologous to the oligopeptide permeaselocus and is required for sporulation and competence. J. Bacteriol. 173:1388–1398.
53. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: alaboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold SpringHarbor, N.Y.
54. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequencing withchain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74:5463–5467.
55. Sauer, R. T. 1978. DNA sequence of the bacteriophage l cI gene. Nature276:301–302.
56. Sauer, R. T., J. Pan, P. Hopper, K. Hehir, J. Brown, and A. R. Poteete. 1981.Primary structure of the phage P22 repressor and its gene c2. Biochemistry20:3591–3598.
57. Sauer, U., J. D. Santangelo, A. Treuner, M. Buchholz, and P. Durre. 1995.
Sigma factor and sporulation genes in Clostridium. FEMS Microbiol. Rev.17:331–340.
58. Schroeder, C. J., and W. J. Dobrogosz. 1986. Cloning and DNA sequenceanalysis of the wild-type and mutant cyclic AMP receptor protein genes fromSalmonella typhimurium. J. Bacteriol. 167:616–622.
59. Scott, M. P., J. W. Tamkun, and G. W. Hartzell. 1989. The structure andfunction of the homeodomain. Biochim. Biophys. Acta 989:25–48.
60. Southern, E. M. 1975. Detection of specific sequences among DNA frag-ments separated by gel electrophoresis. J. Mol. Biol. 98:503–517.
61. Sutcliffe, J. G. 1979. Complete nucleotide sequence of the Escherichia coliplasmid pBR322. Cold Spring Harbor Symp. Quant. Biol. 43(Part 1):77–90.
62. Terracciano, J. S., and E. R. Kashket. 1986. Intracellular conditions requiredfor initiation of solvent production by Clostridium acetobutylicum. Appl.Environ. Microbiol. 52:86–91.
63. Trieu-Cuot, P., C. Poyart-Salmeron, C. Carlier, and P. Courvalin. 1990.Nucleotide sequence of the erythromycin resistance gene of the conjugativetransposon Tn1545. Nucleic Acids Res. 18:3660.
64. Walter, K. A. 1993. Ph.D. thesis. Northwestern University, Evanston, Ill.65. Walter, K. A., G. N. Bennett, and E. T. Papoutsakis. 1992. Molecular char-
acterization of two Clostridium acetobutylicum ATCC 824 butanol dehydro-genase isozyme genes. J. Bacteriol. 174:7149–7158.
66. Webster, P. J., and T. E. Mansour. 1992. Conserved classes of homeodo-mains in Schistosoma mansoni, an early bilateral metazoan. Mech. Dev.38:25–32.
67. Wei, T.-F., T. S. Ramasubramanian, F. Pu, and J. W. Golden. 1993.Anabaena sp. strain PCC 7120 bifA gene encoding a sequence-specific DNA-binding protein cloned by in vivo transcriptional interference selection. J.Bacteriol. 175:4025–4035.
68. Wilkinson, S. R., D. I. Young, J. G. Morris, and M. Young. 1995. Moleculargenetics and the initiation of solventogenesis in Clostridium beijerinckii (for-merly Clostridium acetobutylicum) NCIMB 8052. FEMS Microbiol. Rev.17:275–285.
69. Woods, D. R. 1995. The genetic engineering of microbial solvent production.Trends Biotechnol. 13:259–264.
70. Woods, D. R., and S. J. Reid. 1995. Regulation of nitrogen metabolism,starch utilisation and the b-hbd-adh1 gene cluster in Clostridium acetobuty-licum. FEMS Microbiol. Rev. 17:299–306.
71. Wu, L., and N. E. Welker. 1991. Cloning and characterization of a glutaminetransport operon of Bacillus stearothermophilus NUB36: effect of tempera-ture on regulation of transcription. J. Bacteriol. 173:4877–4888.
72. Yanisch-Perron, C., J. Vieira, and J. Messing. 1985. Improved M13 cloningvectors and host strains: nucleotide sequences of the M13mp18 and pUC19vectors. Gene 33:103–119.
73. Young, M., N. P. Minton, and W. L. Staudenbauer. 1989. Recent advancesin the genetics of the clostridia. FEMS Microbiol. Rev. 63:301–326.
74. Youngleson, J. S., W. A. Jones, D. T. Jones, and D. R. Woods. 1989. Molec-ular analysis and nucleotide sequence of the adh1 gene encoding anNADPH-dependent butanol dehydrogenase in the gram-positive anaerobeClostridium acetobutylicum. Gene 78:355–364.
75. Zuker, M., and P. Stiegler. 1981. Optimal computer folding of large RNAsequences using thermodynamics and auxiliary information. Nucleic AcidsRes. 9:133–148.
