Specific Binding of the E2 Subunit of Pyruvate Dehydrogenase to the Upstream Region of Bacillus thuringiensis Protoxin Genes*

During sporulation, Bacillus thuringiensis produces inclusions comprised of different amounts of several related protoxins, each with a unique specificity profile for insect larvae. A major class of these genes designatedcry1 have virtually identical dual overlapping promoters, but the upstream sequences differ. A gel retardation assay was used to purify a potential regulatory protein which bound with different affinities to these sequences in three cry1 genes. It was identified as the E2 subunit of pyruvate dehydrogenase. There was specific competition for binding by homologous gene sequences but not by pUC nor Bacillus subtilis DNA; calf thymus DNA competed at higher concentrations. The B. thuringiensis gene encoding E2 was cloned, and the purified glutathioneS-transferase-E2 fusion protein footprinted to a consensus binding sequence within an inverted repeat and to a potential bend region, both sites 200–300 base pairs upstream of the promoters. Mutations of these sites in the cry1A gene resulted in decreased binding of the E2 protein and altered kinetics of expression of a fusion of this regulatory region with the lacZ gene. Recruitment of the E2 subunit as a transcription factor could couple the change in post exponential catabolism to the initiation of protoxin synthesis.

During sporulation, Bacillus thuringiensis produces inclusions comprised of different amounts of several related protoxins, each with a unique specificity profile for insect larvae. A major class of these genes designated cry1 have virtually identical dual overlapping promoters, but the upstream sequences differ. A gel retardation assay was used to purify a potential regulatory protein which bound with different affinities to these sequences in three cry1 genes. It was identified as the E2 subunit of pyruvate dehydrogenase. There was specific competition for binding by homologous gene sequences but not by pUC nor Bacillus subtilis DNA; calf thymus DNA competed at higher concentrations. The B. thuringiensis gene encoding E2 was cloned, and the purified glutathione S-transferase-E2 fusion protein footprinted to a consensus binding sequence within an inverted repeat and to a potential bend region, both sites 200 -300 base pairs upstream of the promoters. Mutations of these sites in the cry1A gene resulted in decreased binding of the E2 protein and altered kinetics of expression of a fusion of this regulatory region with the lacZ gene. Recruitment of the E2 subunit as a transcription factor could couple the change in post exponential catabolism to the initiation of protoxin synthesis.
Most Bacillus thuringiensis subspecies contain multiple, plasmid-encoded protoxin genes that are very actively transcribed primarily during sporulation (1,2). There is extensive synthesis of the related protoxins which are often packaged into the same inclusion. Each of these protoxins has a somewhat different specificity (7), and there may be synergism among some (8,9).
Many but not all of these genes contain very similar overlapping promoters (2, 3) recognized in the mother cell during sporulation by E and K forms of RNA polymerase (2,4). Dual promoters ensure transcription of these particular genes (cry1) throughout much of sporulation 1 (5), but they are differentially transcribed (6). Control of expression of the cry genes is necessary not only to ensure a balance of transcription with mothercell spore genes which utilize the same forms of RNA polymerase but for regulating the relative amounts of the various protoxins and their assembly into an inclusion (2).
Given the similarity of the overlapping promoter regions for three of these cry1 genes, the sequences upstream for ϳ1 kbp 2 were examined and found to differ substantially (10). 3,4 These regions appear to be important for regulation because expression of cry1-lacZ fusion plasmids in B. thuringiensis was enhanced by their presence. 1 Employing a gel retardation assay, a novel DNA binding protein was identified and purified, and its gene was cloned. The binding sites in the upstream regions of two of these cry genes were determined by footprinting. The effects of mutations in these sites indicated that this protein is likely to have a role in regulating the expression of this class of protoxin genes.

EXPERIMENTAL PROCEDURES
Cell Growth-B. thuringiensis subsp. kurstaki HD1, strain 80-21 (11), subsp. aizawai HD133, and a plasmid-cured (with mitomycin C) acrystalliferous derivative of B. thuringiensis subsp. kurstaki HD1 designated Mit9 4 were grown in G-Tris medium at 30°C (12) in a New Brunswick incubator shaker. This medium contains 0.2% glucose as the principal carbon source, which was replaced with 0.1% potassium gluconate for the ␤-galactosidase assays (see below). Bacillus subtilis JH642 was grown in nutrient sporulation medium at 37°C. Growth was monitored by A 600 in a Perkin Elmer Model 35 spectrophotometer and sporulation in the phase microscope.
Isolation of Regions Upstream of the Promoters-A region of 280 bp upstream of the cryIAb gene was prepared by PCR using oligonucleotides 5Ј-AATAGGATCCTTCCTATATTTACTTTGCCC-3Ј, containing a BamHI site, and 5Ј-GGTTTGAATTCCGTTAACTTATTTTAAAGT-3Ј, containing an EcoRI site.
The region upstream of the promoters of the cry1C gene was isolated as a 656-bp BglII/HindIII fragment from a 7-kbp EcoRI fragment containing this gene, including 2.5 kbp upstream of the promoters (13). This fragment was isolated from low melting agarose with GELase (Epicentre Technologies) and cloned into the HindIII and BamHI sites of pUC18. For gel retardation, this clone was digested with EcoRI and HindIII, and the 656-bp fragment was reisolated as described above.
The cryID gene (14) was cloned as a 3.8 kbp KpnI fragment from B. thuringiensis subsp. aizawai HD133, strain 5 (11), into the Escherichia coli/B. thuringiensis shuttle vector pHT3101 (15). A 2.2-kbp NdeI fragment embracing most of the coding region was deleted from this plasmid, creating p⌬ID from which the cryID upstream region including the promoters was isolated as a 1.6-kbp KpnI1-NdeI fragment.
The E2 binding site containing a potential bend region was mutated from 5Ј-CTCAATTTGTATATGTAAAATAGGAAAAGTG to 5Ј-CTCAG-TCTGTCTATGTAGAACAGGACAAGTG (bold letters indicate changes including the creation of an MspHI site for screening) employing oligonucleotide 5Ј-CACTTCTCCTGTTGTACATAGACAGACTGAG. The inverted repeat (IR; see Fig. 4) was mutated from 5Ј-CCTGCAATTCAT-CTTGAATTGTAAATGC to 5Ј-CCTGCAGTTAAGCCTGAATTGTAAA-TGC with the introduction of a PstI site for screening. The cry1Ab upstream region as a 310-bp HindIII fragment (see Fig. 4) was cloned into pGEM 11(ϩ) for production of single-stranded template in E. coli CJ236 (dut Ϫ , ung Ϫ ). Mutagenesis followed the procedure of Kunkel (16), and plasmids produced in E. coli DH5␣ were screened initially for the * This work was supported by United States Public Health Service Grant GM 34035. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF039908.
‡ Supported by a Predoctoral Fellowship from the National Science Foundation. To whom correspondence should be addressed. presence of an additional MspHI or PstI site in the insert and then sequenced to confirm the changes. The mutagenized HindIII fragment was isolated from the pGEM clone as described above, and the fragment inserted into the single HindIII site in a lacZ fusion vector containing the cry1A promoter region (see below). The orientation was established on the basis of the sizes of HpaI restriction fragments.
Purification of the Binding Protein-Strain Mit9 was grown in 2 liters of G-Tris medium until about 70% of the cells contained phase bright endospores. B. subtilis JH642 was grown in 500 ml of nutrient sporulation medium until Ͼ50% contained phase bright endospores (the maximum is about 70% versus Ͼ90% for Mit9). Cells were harvested by centrifugation and washed once with 20 ml of 100 mM KCl, 5 mM EDTA, pH 8.0. The pellets were suspended in 1/5 volume of buffer A (50 mM Tris, pH 7.4, 1 mM EDTA, 100 g/ml phenylmethylsulfonyl fluoride), washed twice with this buffer, resuspended in 1/50 the original volume of buffer A plus 1 mM dithiothreitol, and lysed in a French press at 9000 p.s.i. The lysate was centrifuged at 4500 ϫ g for 10 min at 4°C. The supernatant was withdrawn and centrifuged in an Eppendorf microcentrifuge for 10 min at 4°C. This supernatant was then heated to 45°C for 10 min to inactivate DNases while leaving binding activity unaltered.
The heated extract was fractionated by the addition of solid (NH 4 ) 2 SO 4 to final concentrations of 25, 40, 50, 60, 80, and 100% of saturation. The protein precipitate after each step was collected by centrifugation at 8000 ϫ g for 20 min at 4°C, dissolved in 0.5 ϫ TBE (0.04 M Tris, 0.04 M sodium borate, 2 mM EDTA, pH 8.0), and dialyzed overnight at 4°C against two changes of 2 liters each of this buffer. Each fraction was then tested for binding activity as described below.
The active 50 -60% (NH 4 ) 2 SO 4 fraction from Mit9 was further fractionated by passage over a heparin-agarose column (17). One column volume was loaded in 0.5 ϫ TBE at 4°C, and the effluent was passed back over the column ten times. After washing with 4 column volumes of 0.5 ϫ TBE, the column was eluted with a KCl gradient from 0 to 0.4 M in 0.5 ϫ TBE. One-ml fractions were collected, concentrated 4-fold by lyophilization, and dialyzed against 0.5 ϫ TBE. Fractions were assayed for binding activity as described below.
Purification was also performed by excising and eluting the retarded DNA band from polyacrylamide gels. Elution was performed in a Little Blue Tank elutrap system (ISCO) using 0.05 ϫ TBE in the sample wells and 0.5 ϫ TBE in the electrophoresis chamber. The DNA-protein complex concentrated in this way was dissociated by addition of KCl to 0.4 M for 8 h at 27°C. The DNA was then digested for 1 h at 37°C with 20 units of DNase I, and the protein was fractionated in 10% SDS-PAGE (16). The gel was transferred to polyvinylidene difluoride and stained (18). Protein bands were excised, and the sequence of the first 25 amino acids was determined in an automated sequenator (Purdue Center for Macromolecular Structure).
DNA Labeling-DNA was treated with calf intestinal alkaline phosphatase (CIP; 2 units per mg of DNA) at 37°C for 45 min, extracted with phenol and precipitated with ethanol. The DNA was then dissolved in 10 mM Tris, 20 mM ␤-mercaptoethanol, 10 mM MgCl 2 , pH 7.5, and incubated with l0 units of T4 polynucleotide kinase and 300 Ci of [␥-32 P]ATP (150 mCi ml Ϫ1 ) per g of DNA for 90 min at 37°C in a total volume of 25 l. To remove the label from one end, the 32 P-DNA was digested with BamHI for footprinting the transcribed strand and with EcoRI for the nontranscribed strand. The labeled DNA was purified by excision from a low-melting agarose gel and digestion with GELase.
Gel Retardation Assays-Retardation assays (19) were performed in 3.5% native polyacrylamide gels (acrylamide:bisacrylamide ratio of 60:1) as per Fried and Crothers (20) and Garner and Revzin (21). After investigating several buffers, it was found that the best retardation was observed when protein and 32 P-labeled DNA were mixed in 20 l of 0.5 ϫ TBE and incubated at 16°C for 20 min. Five l of bromphenol blue-xylene cyanol in 5 M sucrose was added to the samples, which were loaded onto vertical gels prepared in 0.5 ϫ TBE and subjected to electrophoresis at 4°C and 10 V/cm until the bromphenol blue dye front was about 3/4 of the way down the gel (approximately 2.5 h for a 15-cm gel). The gel was dried at 80°C under vacuum and autoradiographed.
Cloning the pdhC Gene from B. thuringiensis-The pdhC gene encoding the E2 subunit of pyruvate dehydrogenase (PDH) from B. thuringiensis was cloned as a fusion with the glutathione S-transferase gene in the pGEX-KG expression vector (22). PCR oligonucleotides: 5Ј-TAGGAGGTCGGGATCCGTGGCATTTGAATT-3Ј containing a BamHI site and 5Ј-ATAGGGAAATCTCGAGCTACCATAACATTA-3Ј containing a XhoI site were based on the sequences of the B. subtilis pdhC gene (23) and used to clone a 1350-bp region of DNA from B. thuringiensis corresponding to its pdhC gene. This gene was cloned in-frame as a BamHIXhoI fragment into the pGEX-KG vector to produce plasmid pCB117. It was sequenced and had a deduced open reading frame encoding a polypeptide of the expected size.
A fusion protein consisting of B. thuringiensis PDH-E2 fused to glutathione S-transferase was produced after transformation of pCB117 into E. coli TG1 and induction with isopropyl-1-thio-␤-D-galactopyranoside (22). The fusion protein that was purified by elution from a glutathione-agarose column (24) reacted with anti-PDH-E2 antibody from Staphylococcus aureus, kindly provided by Dr. H. Hemila (23). The glutathione S-transferase could not be removed by thrombin digestion without disrupting the B. thuringiensis PDH-E2 protein, which contains an internal thrombin cleavage site.
To obtain a less extensive E2 fusion protein, this gene was also cloned into the His 6 expression vector, pQE30 (Qiagen). The BamHI/XhoI fragment was cloned into pUC18 and then excised as a BamHI/SphI fragment for cloning in-phase in pQE30. E. coli DH5␣ containing this clone was grown in LB-ampicillin (25 g ml Ϫ1 ) and expression was induced by addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h. Cells were lysed as per the Qiagen manual. Although there was substantial His 6 -E2 in the soluble fraction, the E2 protein solubilized from the pellet with 6 M urea in 0.05 M Na 2 HPO 4 , 0.3 M NaCl, pH 8.0 was most active in gel retardation. The latter was fractionated on a Ni 2ϩ -agarose column using a step gradient of 0.1-0.3 M imidazole in the above buffer. After dialysis for 18 h at 4°C against 4000 volumes of 0.5 ϫ TBE, fractions were assayed for gel retardation activity, and the most active fraction was stored at Ϫ80°C.
DNase I Footprinting-Fifty ng of purified, end-labeled DNA was incubated with varying amounts of the E2 protein and 5 g of poly(dI⅐dC) in 20 l of 0.5 ϫ TBE. The binding reaction was carried out for 20 min at 16°C, after which 0.05 units of DNase I (Boehringer Mannheim) in 12.5 mM MgCl 2 was added and the tubes incubated at 25°C for 90 s. The reaction was stopped by the addition of 5 l of 30 mM EDTA and extraction of the mixture with phenol-chloroform. The DNA was precipitated with 2 volumes of ethanol and dissolved in Sequenase stop buffer (United States Biochemical). After heating at 90°C for 5 min, samples were loaded onto a 6% polyacrylamide gradient gel containing 8 M urea in 90 mM Tris, 89 mM borate, 2.5 mM EDTA, pH 8.3 (16) and electrophoresed at 1000 V and 75 mA. Gels were then dried and autoradiographed.
DNA sequence was obtained according to the standard method for double-stranded DNA sequence analysis (16) with the addition of Mn 2ϩ to the reaction mix to allow sequence determination near the primer (25).
␤-Galactosidase Activity-A cryIAb promoter-lacZ fusion was constructed in an E. coli/Bacillus shuttle vector. 1 A HindIII-digested, PCR-cloned 310-bp upstream fragment from the cryIAb gene (containing the region from Ϫ218 to Ϫ528; see  ers. This fragment containing the mutated bend or IR regions was also inserted into this vector, and the plasmids were electroporated (26) into strain 80 -21. Cells were grown in G-Tris medium (12) containing 0.l% yeast extract and 0.1% potassium gluconate plus 5 g/ml chloramphenicol. Differences in induction of ␤-galactosidase (but not growth) were more evident when potassium gluconate rather than glucose was the primary carbon source. Samples of 0.2 ml were taken at various times during growth and sporulation, sonicated for 10 s, and 30 -50 l were assayed for ␤-galactosidase activity (27) as modified by Giacomini et al. (28). Growth was monitored by A 600 in the Perkin Elmer spectrophotometer and sporulation followed in the phase microscope, measuring the percent phase dull, phase white, and phase bright endospores in each sample.

RESULTS
Binding Activity in Sporulating Cells-The 50 -60% (NH 4 ) 2 SO 4 fraction of a crude extract from sporulating but not from growing cells of B. thuringiensis strain Mit9 retarded a 280-bp fragment from the cry1A upstream region, a 656-bp fragment from cry1C, and an ϳ1.6-kbp fragment from cry1D (Fig. 1). In all cases, at least a 1,000-fold excess of poly(dI⅐dC) was present. The further addition of a small excess of homologous unlabeled DNA competed for binding, whereas the same concentration of pUC18 did not ( Fig. 2A). Even a 100-fold excess of pUC18 DNA (either linear or as 0.3 ϩ 2.4-kbp PvuII fragments) or sonicated B. subtilis DNA did not compete. 4 There was some inhibition of retardation by a 50-fold excess of sonicated calf thymus DNA (Fig. 2B).
The specificity of this binding was confirmed by footprinting ( Figs. 4 and 6) as well as by the effectiveness of a B. thuringiensis protein fraction as compared with that from B. subtilis (Fig. 2C). There was greater retardation of the cry1A fragment by equivalent amounts of E2 in the 50 -60% ammonium sulfate fraction from B. thuringiensis Mit9 than that from B. subtilis JH642 (Fig. 2C). A comparable preparation from E. coli did not retard.
Retardation of 1 pmol of the cry1Ab fragment was complete when incubated with about 2 g of the 50 -60% ammonium sulfate fraction from B. thuringiensis. Complete retardation of the cry1C and cry1D DNAs required 10 -20-fold more protein, indicating lower affinities and/or more binding sites. There were several degrees of retardation of these DNAs (arrows on the right panel of Fig. 1) in contrast to the cry1A DNA. The presence of multiple retarded complexes may be because of several binding sites with different affinities for the protein.
Characterization of the Binding Protein-The binding protein was isolated by electroelution of an excised, retarded band followed by dialysis and digestion with DNase I. This procedure resulted in recovery of a single protein of ϳ60 kDa (Fig. 3). The sequence of the first 25 residues of this band was 92% identical to that of dihydrolipoamide acetyltransferase, the E2 subunit of PDH from B. subtilis and Bacillus stearothermophilus ( Table  I). The E2 subunit is a 48-kDa protein containing lipoic acid (which decreases its mobility in SDS-PAGE).
Because sufficient quantities of E2 could not be renatured after elution from the retarded complex, the 50 -60% (NH 4 ) 2 SO 4 fraction was further purified by elution from a heparin-agarose column with a KCl gradient. The greatest binding activity (based on retardation of cry1A DNA) was in the 0.2-0.3 M KCl fraction that contained the E2 protein as Ͼ50% of the total based on staining and confirmed by immunoblotting with antiserum against the B. subtilis PDH complex. The E1␣ and E1␤ subunits of PDH eluted at a lower salt concentration, and this fraction retarded poorly (41).
Footprinting of the Binding Sites-The purified glutathione S-transferase fusion protein protected three regions in the 280-bp cry1Ab DNA (Fig. 4, A and B and Fig. 5). The most distal of these and the inverted repeat share a common sequence. The third region was within a stretch of intrinsically bent DNA (29) with a 10-bp spacing between each of the protected regions. There were also several hypersensitive sites on both strands (arrows). A major protected site in the cry1C DNA was within an inverted repeat (Fig. 6) almost identical in sequence to that in the cry1Ab DNA. There were also a multiplicity of hypersensitive sites in this DNA but no detectable bend region. The footprint with the heparin-agarose fraction was the same as that obtained with the purified fusion protein.
Function of the Upstream Region in cry1 Gene Transcription-The binding site within the bend region and the IR were mutated as described under "Experimental Procedures." Both were found to have lower affinities for purified His 6 -E2 protein than the wild type fragment (Fig. 7). The apparent K d value for the wild type, assuming that it was the monomer of the His 6 -E2 adduct that bound, was 4 -6 nM. E2 is a multimer in the PDH complex (30), however, and it is likely that a multimeric form of this protein is required for binding to DNA. As mentioned above, only His 6 -E2 extracted from the crude membrane fraction with 4 -6 M urea in buffer was active. The extent of gel retardation by His 6 -E2 purified from the E. coli clone was greater than that of E2 purified from B. thuringiensis (Figs. 1  and 7), implying different aggregation states perhaps because of concentration effects or the presence or absence of lipoamide.
Cells containing lacZ fusions with either the cry1A wild type Boxed sequences indicate those protected in the DNase footprints (Fig. 4). The boxed inverted repeat (convergent overlying arrows) is based on footprinting results with both the transcribed and nontranscribed strands. Start sites of transcription from the BtI ( E ) and BtII ( K ) promoters are indicated by arrows; the ribosome binding site is overlined. Lowercase letters indicate oligonucleotide primers used to construct a 310-bp HindIII PCR fragment of the upstream region. Letters in bold are the HindIII sites.
or mutant (bend and IR) upstream regions were sampled during sporulation, and the kinetics of ␤-galactosidase synthesis was determined (Fig. 8). Addition of the upstream DNA resulted in an enhancement of ␤-galactosidase synthesis. Both the initial rate and the final activity were reduced in strains containing the mutated upstream regions. DISCUSSION Regulation of protoxin genes is of interest not only because of their insecticidal properties but from the perspective of how a cell recruits regulatory elements for a group of structural genes that have very likely become part of the genetic repertoire of this Bacillus relatively late in evolution. One aspect of the regulation is the presence of dual overlapping promoters, which ensures a constant rate of transcription of these cry genes during an extended period of sporulation. In addition, each of these genes is independently regulated, so factors other than the promoters must be involved. The sequences upstream of the promoters for the differentially regulated cry1A, cry1C, and cry1D genes differ substantially (although some features are shared; see below) so we began a search for DNA binding proteins.
Extracts of sporulating but not vegetative cells of B. thuringiensis subsp. kurstaki contained a protein that bound to regions of DNA upstream of the cry1Ab, cry1C, and cry1D gene promoters. The major binding protein in the heparin-purified fraction from B. thuringiensis was identified as the E2 subunit of PDH. The purified GST-E2 or His 6 -E2 fusion proteins footprinted to specific sites in the cry1A and cry1C upstream regions. The presence of three close binding sites in the cry1A but not the cry1C sequence may account for the higher affinity for the E2 oligomer by the former (Fig. 1) and thus a basis for the differential regulation of these cry 1 genes.
The evidence for specific binding of this novel DNA binding protein is 1) the presence of a consensus binding sequence (5Ј-cAAGAT/gG/tAA) in two of the three sites in the cry1A sequence and within the inverted repeat in the cry1C sequence. There was also binding to an intrinsically bent region in the cry1A DNA. The binding site(s) in the cry1D DNA have not been mapped. 2) There was optimal competitive binding by the homologous DNA and very little or no competition by nonspecific DNAs such as poly(dI⅐dC), pUC18, or B. subtilis DNAs. Sonicated calf thymus did compete at higher concentrations (Fig. 2B), probably because of the complexity of sequences in this DNA including potential bend regions. 3) There was no binding to a digest of pUC18, which included a fragment similar in size (about 0.3 kbp) to that of the cry1A upstream fragment. 4) There was higher affinity binding by the E2 protein from B. thuringiensis as compared with that from B. subtilis. While the deduced sequences of the B. thuringiensis and B. subtilis E2 proteins are Ͼ80% identical, there is considerably less homology in and around the so-called hinge region. 4 This region of E2 links the lipoyl domain to the E1 and E3 binding sites (30), and it is a potential DNA binding region (37).  1-3), mutated IR (lanes 4 -7), and mutated bend region (lanes 8 -10) DNAs. Lanes 1, 4, and 8, no protein; lanes 2 and 5, 1.0 g of His 6 -E2; lanes 3, 6 and 9, 2.0 g of His 6 -E2; lanes 7 and 10, 3.0 g of His 6 -E2. The stoichiometry and patterns of DNA retardation indicated extensive cooperativity in the binding, probably involving conformational changes to allow some form of the E2 protein to bind. The icosohedral core of PDH is comprised of 60 E2 subunits (30), so it is likely that some multimeric form is involved in binding as indicated by the requirement of a urea extraction in the purification protocol of His 6 -E2 from E. coli. The role of lipoamide, which is a component of E2 in the PDH complex, is not known. Whatever conformational changes of the DNA occurred could be reversed by treatment with protease K, demonstrating that the continued presence of E2 was essential. In many respects, the binding was similar to that of the Lrp protein (31), which may serve as a useful paradigm.
The importance of these E2 binding sites was indicated by the effect of mutations on the binding of His 6 -E2 and on the expression of lacZ fusions (Figs. 7 and 8). Both the rates and final amounts were reduced, but the former may be particularly important. These parasporal inclusions have a crystalline array (32) so that the deposition of the disulfide cross-linked protoxins (33) must be an orderly process, very likely dependent upon chaperones and other factors (2). A decrease in the rate of protoxin accumulation such as that resulting from mutations in the upstream binding sites for the E2 protein (Fig. 8) could substantially alter the relative amount of the Cry1A protoxin in the inclusion.
The consensus binding sequence is also present close to the start site of transcription of the pdhC gene in B. subtilis (23) as well as near the origin of replication (34,35). The latter was identified as a region of the chromosomal DNA bound to the membrane primarily by a protein of 60 kDa, which was subsequently identified as E2. 5 The E2 protein in Neurospora crassa (designated MRP3) appears to be a mitochondrial ribosomal protein (36) so its recruitment for other functions is not without precedence.
A role for the E2 subunit in regulation implies a connection between catabolism and protoxin synthesis. It has been known for some time that the protoxin yield per spore or per mg of dry weight varies considerably with the subspecies studied and with the media (38). During growth on glucose, bacilli excrete acetic acid, pyruvic acid, and acetoin (39) which are catabolized during sporulation (40). The pyruvate is utilized rapidly, and PDH is then no longer required for catabolism. At about this time, the E2 subunit is found in the soluble fraction of sporulating cells 4 (35). The source of the soluble E2 could be dissociation of the PDH complex and/or transcription from its own promoter (23), which does function at the end of growth. 4 The E2 consensus binding sequence in this promoter region is thus all the more intriguing.
The presence of soluble E2 would signal the end of growth on sugars, and its recruitment for regulating the plasmid-encoded cry genes would provide a mechanism for selectively enhancing their transcription in postexponential cells. The amount of soluble E2 in sporulating cells would depend upon prior growth conditions and perhaps autoinduction of the pdhC gene. Such factors could integrate cell growth with the subsequent transcription of the cry genes.