Activation Control of pur Gene Expression in Lactococcus lactis: Proposal for a Consensus Activator Binding Sequence Based on Deletion Analysis and Site-Directed Mutagenesis of purC and purD Promoter Regions

Abstract

A comparison of the purC and purD upstream regions from Lactococcus lactis revealed the presence of a conserved ACCGAACAAT decanucleotide sequence located precisely between −79 and −70 nucleotides upstream from the transcriptional start sites. Both promoters have well-defined −10 regions but lack sequences resembling −35 regions for ς⁷⁰ promoters. Fusion studies indicated the importance of the conserved sequence in purine-mediated regulation. Adjacent to the conserved sequence in purC is a second and similar region required for high-level expression of the gene. A consensus PurBox sequence (AWWWCCGAACWWT) could be proposed for the three regions. By site-directed mutagenesis we found that mutation of the central G in the PurBox sequence to C resulted in low levels of transcription and the loss of purine-mediated regulation at the purC and purD promoters. Deletion analysis also showed that the nucleotides before the central CCGAAC core in the PurBox sequence are important. All results support the idea that purC and purD transcription is regulated by a transcriptional activator binding to the PurBox sequence.

Lactococcus lactis is a gram-positive bacterium related to members of the Streptococcus, Lactobacillus, and Bacillus genera (30). It obtains all energy by fermenting sugars to lactic acid and can be isolated from raw milk. While most strains of L. lactis are multiply auxotrophic for both amino acids and vitamins (30), the ability to synthesize both pyrimidine (15) and purine (21) nucleotides de novo is retained in all isolates of L. lactis tested. The de novo synthesis of purine nucleotides requires 10 enzymatic steps leading to IMP, which functions as a precursor for both AMP and GMP nucleotides (35). Purine nucleotides can also be formed by salvage reactions from exogenous purine nucleosides or bases (22). Whereas the de novo pathway appears to be conserved among most organisms, the salvage pathway can vary between organisms (22), and so far only one gene involved in purine salvage is known in L. lactis (21).

The organization of bacterial genes involved in purine metabolism and the regulation of the expression of these genes have been best studied in Escherichia coli, Salmonella typhimurium (19, 36), and Bacillus subtilis (19, 34, 35). In the gram-negative bacterium E. coli, the purine biosynthetic genes are scattered around the chromosome (19). However, the transcription of all of these genes is repressed by a single regulatory protein, the purR-encoded purine repressor (8, 13, 16, 23). Binding of the E. coli PurR repressor to its target DNA sequences (PurBox’s) is stimulated by the corepressors guanine and hypoxanthine (17, 24). In the gram-positive bacterium B. subtilis, all de novo genes are organized in a single transcriptional unit, the purine operon (4). The transcription of the operon is controlled by two independent mechanisms. Initiation is controlled by a repressor which is encoded by the purR gene and which, at low 5-phosphoribosyl-1-pyrophosphate (PRPP) concentrations, binds specifically to a DNA sequence in the promoter region (5, 33). A rationale for the use of PRPP as an indicator of purine availability was put forward by Weng and coworkers (33). Upon uptake, adenine is converted to AMP, consuming PRPP in the process. The subsequent phosphorylation of AMP yields ADP, which is the primary inhibitor of PRPP synthetase. Thus, the combined inhibition of PRPP synthesis and increased PRPP consumption may explain why high extracellular adenine pool levels are correlated with low PRPP pool levels in B. subtilis (26). The purR genes from B. subtilis and E. coli are unrelated, and the B. subtilis enzyme shows a high degree of similarity with purine phosphoribosyltransferases (1, 12, 33), while the E. coli enzyme is a classical lacI-type repressor (28). The second mechanism controlling the expression of the pur operon in B. subtilis involves a terminator-antiterminator structure located between the promoter and the translation start site of the first gene. The formation of the antiterminator structure is believed to be prevented by the binding of an unidentified RNA binding protein in the presence of the purine base guanine (4) and hypoxanthine (1), thus resulting in premature termination of transcription.

We recently reported the nucleotide sequence and characterization of the purDEK operon from L. lactis CHCC285 (20). The transcription of the genes was shown to be down regulated more than 30-fold upon the addition of purines to a chemically defined medium. Deletion analysis of the region upstream of the purD reading frame made it possible to localize the promoter to a 133-bp fragment. By monitoring the promoter activity from different DNA fragments in a promoter fusion vector, we found that the 133-bp fragment also retained full purine regulation. A specific deletion mutant in which sequences from 78 bp upstream from the transcriptional start site was removed showed greatly reduced promoter activity. This result suggested that the affected region was a positive regulatory element (20).

Here we report the further characterization of the regulatory region located in front of the purD gene in L. lactis CHCC285. After cloning of the purC gene from the same strain, we were able to identify a common motif, designated a PurBox, which is present in two copies in purC and in one copy in purD. We present evidence that the PurBox is a positive regulatory site, most likely the binding site for a transcriptional activator.

MATERIALS AND METHODS

Bacterial strains and growth conditions.

The strains used in the present study are listed in Table 1. Cultures of L. lactis were grown in either DN medium (3) or SA medium (11) supplemented with 1% glucose (GSA medium) and erythromycin at 2 μg/ml when required. Purine additions were made at the following final concentrations: guanosine (30 μg/ml), adenine (15 μg/ml), and hypoxanthine (15 μg/ml). SR plates (9) were used for plating transformants of L. lactis after electroporation.

TABLE 1.

Bacterial strains

Strain	Relevant genotype or phenotype	Source or reference
Escherichia coli DH5α	endA1 hsdR17(rK− mK+) supE44 thi-1 recA1 gyrA relA1 Δ(lacZYA-argF)U169/φ80dlacΔ(lacZ)M15	Laboratory collection
Lactococcus lactis
MG1363	Wild type	6
CHCC285	Wild type	21
DN207	MG1363 purC	3
MK189	MG1363/pMK1013	This work
SH1	MG1363/pLN95	This work
SH4	MG1363/pSH4	This work
SH5	MG1363/pSH5	This work
SJ2	MG1363/pSJ2	This work
SJ3	MG1363/pLN96	This work
SJ4	MG1363/pLN97	This work
SJ5	MG1363/pSJ5	This work
SW1	MG1363/pSW1	This work
SW2	MG1363/pSW2	This work
SW5	MG1363/pSW5	This work

Oligonucleotide primers.

The oligonucleotide primers used in the present study are listed in Table 2 and were obtained from T-A-G-Copenhagen ApS, Copenhagen, Denmark.

TABLE 2.

Oligonucleotide primers

Primer	Sequencea	Commentb
PAK80	TCCTTTCAAAGTTACCC	pAK80 (Peter L. Madsen)
PAK80ERM	TTTCAACTGCCTGGC	pAK80 (Peter L. Madsen)
MKP17	AAAAAGCTTGCTGGTTCAGTCATGTGG	purC 30–47
MKP21	AAACTGCAGCCACAATATCCTCCATCC	purC 258–241
MKP31	AAAACTGCAGAGTCAAAGTGACGATAGCTC	purD 174–155
MKP36	AAAAAGCTTGAGTATTTCCGAACATTCAGC	purC 120–140
MKP37	AAAAAGCTTCAGCAAAACCGAACAATAAATTG	purC 137–159
MKP40	AAAACTGCAGGCTCTAGCGATAAAACTAGAAC	purC 186–165
MKP51	AAAAAGCTTGAGTAAATACCGAACAATCTCTC	purD 39–61
MKP58	CCTTCATAGAGCAACTTTTCTTTTTCCACAATATCC	purC 283–248
MKP80	AAAAAGCTTGAGTAAATACCCAACAATCTCTC	purD 39–61
MKP109	AAAAAGCTTCAGCAAAACCCAACAATAAATTG	purC 137–159

^aThe nucleotide sequences are shown with the 5′ end to the left. Non-Lactococcus linkers are underlined. 

^bNumbers refer to the nucleotide positions in the purC and purD sequences shown in Fig. 1. 

Cloning of the purC gene.

Chromosomal DNA from L. lactis CHCC285 was digested with HindIII and inserted into the HindIII site of the L. lactis-E. coli shuttle vector pCI3340 (7). After transformation of E. coli DH5α with the ligation mixture, plasmid DNA was extracted from the resulting pool of transformants. Subsequently, the pur mutant DN207 (3) was transformed to chloramphenicol resistance with the pCI3340 library. Upon subsequent analysis of the transformants on solid DN medium in the absence and the presence of purines, purine prototrophic transformants were selected.

Nucleotide sequence determination.

The nucleotide sequence was determined with either a Sequenase 2.0 sequencing kit (Stratagene) or a Thermosequenase kit (containing nucleotides labelled with ³³P-dideoxynucleoside triphosphates) (Amersham LifeScience).

Construction of purC-lacLM and purD-lacLM fusion plasmids in pAK80.

The construction of the fusion plasmids pLN95, pLN96, and pLN97 has been described elsewhere (20). The DNA fragments inserted in plasmids pMK1013, pSH2, pSH4, pSH5, pSJ2, pSJ5, pSW1, pSW2, and pSW5 were all generated by PCR amplification of either CHCC285 chromosomal DNA or purified pLN95 plasmid DNA. The DNA template and oligonucleotide primers used in each construction are listed in Table 3. Each upstream primer has a HindIII site incorporated and each downstream primer has a PstI site incorporated for convenient insertion into the vector. pAK80 plasmid DNA was digested with the two restriction enzymes, and the linearized vector DNA was purified with a High Pure PCR purification kit from Boehringer. Each of the PCR-generated fragments was likewise digested with PstI and HindIII and purified. About 500 ng of digested pAK80 and 100 ng of digested PCR product were ligated in a total volume of 25 μl. Two microliters was used to transform MG1363 to erythromycin resistance, with selection on SR plates (9) containing erythromycin (2 μg/ml) and supplemented with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) (200 μg/ml) for the detection of promoter activity in the inserts. After purification of the transformants, the plasmid DNA was extracted (25). The nucleotide sequence of the inserts was determined after PCR amplification of the regions on the plasmid and the PAK80 and PAK80ERM primers (Table 2), specific for regions flanking the cloning sites (data not shown).

TABLE 3.

Plasmids and details on the construction of plasmids

Plasmid	Source of DNA	PCR amplificationa with the following primer:		Comment
		1	2
pAK80				lacLM transcriptional fusion plasmid (10)
pCI3340				E. coli-L. lactis shuttle vector (7)
pLN63	CHCC285			HindIII fragment in pCI3340 (this work)
pLN95	CHCC285			purD-lacLM fusion in pAK80 (20)
pLN96	CHCC285			purD-lacLM fusion in pAK80 (20)
pLN97	CHCC285			purD-lacLM fusion in pAK80 (20)
pMK1013	CHCC285	MKP109	MKP21	purC-lacLM fusion in pAK80 (Fig. 3)
pSH2	CHCC285	MKP17	MKP21	purC-lacLM fusion in pAK80 (Fig. 3)
pSH4	pLN95	PAK80	MKP31	purD-lacLM fusion in pAK80 (Fig. 3)
pSH5	pLN95	PAK80	MKP32	purD-lacLM fusion in pAK80 (Fig. 3)
pSJ2	CHCC285	MKP51	MKP31	purD-lacLM fusion in pAK80 (Fig. 3)
pSJ5	CHCC285	MKP80	MKP31	purD-lacLM fusion in pAK80 (Fig. 3)
pSW1	CHCC285	MKP36	MKP21	purC-lacLM fusion in pAK80 (Fig. 3)
pSW2	CHCC285	MKP36	MKP40	purC-lacLM fusion in pAK80 (Fig. 3)
pSW5	CHCC285	MKP37	MKP21	purC-lacLM fusion in pAK80 (Fig. 3)

^aInsert DNA generated by PCR. All primer 1 oligonucleotides contained PstI linkers, and all primer 2 contained HindIII linkers. The PCR fragments were inserted into the PstI and HindIII sites of transcription fusion vector pAK80. 

Growth conditions and determination of β-galactosidase activity.

The plasmid-containing strains were grown exponentially at 30°C in 50 ml of purine-free GSA medium supplemented with erythromycin (2 μg/ml) with slow magnetic stirring. For purine-excess conditions, guanosine (30 μg/ml), adenine (15 μg/ml), and hypoxanthine (15 μg/ml) were added. Growth was monitored by measuring the optical density at 450 nm (OD₄₅₀) of the culture. At an OD₄₅₀ of about 0.8, 10 ml of bacterial culture was harvested and washed in 0.9% NaCl. After resuspension in 500 μl of Z buffer (18), cells were disrupted by sonication two times for 1 min each time with an amplitude of 8 μm. The cell debris was removed by centrifugation at 20,000 × g for 10 min, and β-galactosidase activity was determined from 10 and 50 μl of the crude extract in an assay volume of 600 μl. Specific activity is given as micromoles of ortho-nitrophenol (ONP) formed per minute per milligram of protein, with a molar extinction coefficient for ONP of 0.045 at 420 nm (18). Protein concentration was determined according to the method of Lowry et al. (14).

RNA extraction.

For analysis of the purC transcriptional start site in L. lactis CHCC285, RNA was extracted from bacterial cultures after growth in purine-free GSA medium and GSA medium containing guanosine, adenine, and hypoxanthine. At an OD₄₅₀ of 0.6, 5 ml of culture was mixed with 5 ml of EPS solution (60% ethanol, 2% phenol, 0.9% NaCl) prechilled at −30°C to rapidly cool the cell culture (32). After centrifugation at 5,000 × g for 5 min at 4°C, the pellet was washed in EPS solution mixed 1:1 with 0.9% NaCl. The pellet was frozen at −80°C and lyophilized in a vacuum centrifuge without heating. The lyophilized pellet was ground with acid-washed glass beads (100 μm; Sigma) with the rounded tip of a melted pasteur pipette as a pestle, and RNA was extracted essentially as described by Vogel et al. (32). For detection of the start sites in fusion plasmids pLN95 and pSW1, plasmid-containing strains SH1 and SW1 were grown as described above except that erythromycin (2 μg/ml) was added to the cultures. RNA was extracted as described above.

Primer extension analysis.

Primer extension analysis of purC transcripts from CHCC285 was done as previously described (20) with primer MKP58. A slightly modified procedure was used for the mapping of start sites from fusion plasmids in MG1363. About 10 pmol of oligonucleotide primer was end labeled at 37°C in 10 μl of kinase buffer (50 mM Tris-Cl [pH 7.6], 10 mM MgCl₂) containing 5 μl of γ-[³²P]ATP (>5,000 mCi/mmol, ∼3 mM) and 1 U of polynucleotide kinase. After incubation for 30 min, the enzyme was inactivated at 95°C for 10 min. To 5 μl of this mixture was added 20 μg of RNA, 9 μl of 5× first-strand buffer (250 mM Tris-Cl [pH 8.3], 375 mM KCl, 15 mM MgCl₂, SuperScript reverse transcriptase [Gibco BRL Life Technologies]), and water to a final volume of 30 μl. Hybridization was performed with a thermocycler for 10 min at 94°C, followed by 5 min at each of the following temperatures: 75, 73, 71, 69, 67, 65, 63, 61, 59, 57, and 55°C. During incubation at 55°C, the following were added to the sample: 8.5 μl of preheated water, 4.5 μl of 100 mM dithiothreitol, 1 μl of 10 mM deoxynucleoside triphosphates (dNTPs) (10 mM each dATP, dGTP, dCTP, and dTTP), and 1 μl of SuperScript II reverse transcriptase (Moloney murine leukemia virus RNase H negative; Gibco BRL). Incubation was continued for 15 min. The nucleic acids were precipitated by the addition of 160 μl of TE (10 mM Tris-Cl, 1 mM EDTA) (pH 8) and 20 μl of 3 M sodium acetate (pH 4.8), followed by 500 μl of absolute ethanol. After centrifugation and drying of the pellet, the RNA-cDNA was dissolved in 6 μl of H₂O, and 4 μl of stop solution (from the Thermosequenase kit) was added. A 2.5-μl sample was applied to a sequencing gel beside a sequence ladder obtained with the same primer and a PCR-amplified fragment from MG1363 chromosomal DNA as a template.

For the mapping of purD transcriptional start sites, the primer was not end labeled, but labeling was incorporated during the extension reaction by replacing 1 μl of 10 mM dNTPs with 2 μl of α-[³²P]ATP (>5,000 mCi/mmol, ∼3 μM) plus 1 μl of a solution of dNTPs (each at 150 μM) in the first-strand buffer described above. The primers used were MKP58 for the detection of the purC start site in CHCC285 and PAK80 for the detection of start points for transcription into the lacLM genes in pAK80.

Nucleotide sequence accession number.

The nucleotide sequence for purC has been assigned GenBank accession no. AF054888.

RESULTS AND DISCUSSION

Cloning of the purC gene from L. lactis subsp. lactis CHCC285.

Purine prototrophic transformants were selected on purine-deficient DN agar plates after transformation of the uncharacterized pur mutant DN207 (3) with recombinant plasmids from a library of CHCC285 DNA in plasmid pCI3340 (see Materials and Methods). Plasmid pLN63 was extracted from one such transformant. A preliminary determination of the nucleotide sequence (data not shown) of the 4.3-kb insert in plasmid pLN63 showed that the insert carried sequences homologous to the purC and purQ genes of B. subtilis (data not shown).

Nucleotide sequence of the purC promoter region and determination of the transcriptional start site.

Figure 1A shows the nucleotide sequence of a region from about 250 bp before to 50 bp after the start codon in the purC gene present in plasmid pLN63. Translation of the purC gene is most likely initiated from a GTG start codon at position 255 which is preceded by a good ribosome binding site (GGAGG). This start site is consistent with the start site of purC from E. coli (31) (having the N-terminal sequence H₂N-MQKQAELYRGKAK-, where underlined amino acids are identical to those in the lactococcal enzyme). Upstream of the purC gene an open reading frame was detected from position 1 to position 110, where it encounters a stop codon (orfC; Fig. 1A). Following this reading frame is a putative terminator structure at positions 160 to 191, indicating that transcription from orfC does not proceed into the purC gene. Thus, the intercistronic space between orfC and purC is composed of 146 bp.

FIG. 1.

purC (A) and purD (B) promoter regions from L. lactis CHCC285. Transcriptional start sites (+1), putative −10 regions, putative operator sites (PurBox), start codons, and ribosome binding sites (Shine-Dalgarno [SD]) are shown in boldface, underlined, and marked above the nucleotide sequence, while the amino acids specified by the structural genes are shown below the sequence . End points of inserts in promoter fusion plasmid pAK80 (marked with the plasmid designations pLN95 to pLN97) are shown below the sequence as arrows, with the ending nucleotide aligned with the wavy line. Oligonucleotide primers (marked by an MKP designation) used for the generation of PCR products are also shown below the sequence, with the wavy lines aligned with the nucleotides present. The nucleotide sequence of the purD promoter region was modified from that given in reference 20.

The presence of a promoter on the 300-bp DNA fragment shown in Fig. 1A was verified by insertion of a PCR-generated fragment covering position 30 (primer MKP17) through position 258 (primer MKP21) into the promoter fusion vector pAK80 (see Materials and Methods). When the resulting plasmid, pSH2, was introduced into strain MG1363, it expressed higher levels of β-galactosidase activity when the strain was grown in the absence of exogenous purines than when the strain was grown in the presence of purines (see Fig. 3). Thus, both the promoter and the regulatory region are present in plasmid pSH2.

FIG. 3.

Expression of β-galactosidase from purC-lacLM (A) and purD-lacLM (B) fusions. The locations of the DNA fragments inserted in each fusion plasmid are shown below the structures of the promoter regions. Black boxes correspond to structural genes, while open and gray boxes correspond to intact and mutated PurBox sequences, respectively. Specific β-galactosidase activities are given as micromoles of ONP produced per minute per milligram of protein. The strains (MG1363 transformed with the indicated plasmids) were grown in GSA medium containing erythromycin (−Pur) or in the same medium containing guanosine, adenine, and hypoxanthine (+Pur). MG1363 transformed with vector pAK80 showed a low and constant level of background activity (<2 U/mg of protein) under both growth conditions. Fold regulation is given as the values for −Pur/+Pur.

To identify the position of the transcriptional start site, the end of the purC transcript was mapped by primer extension analysis. RNA was extracted from strain CHCC285 that had been growing exponentially in GSA medium in the absence or presence of purines. After the extension reaction, the products were separated on a sequencing gel. One prominent band, which corresponded to an A at position 223 in Fig. 1A, was much more intense in lane 2 of Fig. 2A than in lane 1, suggesting that it represented a purine-regulated transcript. Preceding this start site was a sequence, TAGAAT, which might function as a promoter −10 region (Fig. 1A). No sequences matching the −35 consensus sequence could be found.

FIG. 2.

Primer extension analysis of purC and purD transcriptional start sites. Primer extension experiments were performed by using 10 μg of RNA extracted from CHCC285 (purC) and primer MKP58 (A), SW1 (purC-lacLM) and primer PAK80 (B), and SH1 (purD-lacLM) and primer PAK80 (C). RNA was extracted from cells growing exponentially in GSA medium (lanes 2) or in the same medium supplemented with purines (lanes 1). Lanes G, A, T, and C, sequencing reactions with the same primers as in the primer extension experiments and with PCR-generated DNA as a template. Asterisks indicate the positions of the flanking nucleotides shown in the sequences adjacent to the lanes with the extension products. The picture was scanned at 300 dpi with a Scan Jet 4c/T (Hewlett-Packard Co.) and DeskScan II version 2.3 software. The TIF file was imported into Top Draw version 3.1 for the addition of text.

Identification of a putative regulatory region.

When the purC promoter region was compared to the promoter region of the purDEK operon (20) from the same strain, we found that they share a number of features. Both promoters have A residues as the starting nucleotides, they both have reasonable −10 regions (TAGAAT for purC and TAAGAT for purD), and neither contains sequences with similarity to −35 regions (Fig. 1). Most striking, however, is the presence of an identical decanucleotide stretch (ACCGAACAAT) which can be found at exactly the same positions (positions −79 to −70) relative to the transcriptional start site (Fig. 1). The absence of well-defined −35 regions for the promoters and the conservation of the decanucleotide at about position −75 relative to the transcriptional start site suggested that the decanucleotide could serve as the binding site for a transcriptional activator (27). In a search for sequences resembling the decanucleotide, we found a number of regions in the purC and purD genes. These sequences are shown in Table 4. As shown, the similarity between the sequences extends past the decanucleotide, and a consensus sequence for the five sequences can be found (AWWWCCGAACWWT). Since it is likely that these regions represent binding sites for a purine-specific activator, we have termed the sequences PurBox’s. Most interesting is a sequence (PurBox C1) positioned just upstream of the previously identified PurBox (PurBox C2) in purC. These PurBox sequences are so close that two activator proteins bound to each site would be likely to come in contact.

TABLE 4.

Comparison of PurBox sequences

Operator site	Sequence	Locationa	Verification
PurBox C1	ATTTCCGAACATT	−93	Deletion
PurBox C2	AAAACCGAACAAT	−76	Deletion, mutation
PurBox C3	AAAACCGAACAAT	∼+263	None
PurBox D1	AATACCGAACAAT	−76	Deletion, mutation
PurBox D2	ATTTCCGAACTAT	∼+289	None
Consensus sequence	AWWWCCGAACWWT
Nucleotide position	1.....7.....13
Mutant PurBoxb
PurBox D1-LN96	TTCGCCGAACAAT	−76	Fusion plasmid pLN96
PurBox D1-G7C	AATACCCAACAAT	−76	Fusion plasmid pSJ5
PurBox C2-G7C	AAAACCCAACAAT	−76	Fusion plasmid pMK1010

^aThe location is given for the central G7 residue relative to the transcriptional start site. 

^bStructures of the PurBox’s in the mutants analyzed in the present study. 

Deletion of PurBox sequences destroys purine regulation and lowers promoter activity.

In order to analyze the importance of the PurBox sequences for purC and purD promoter activities, we constructed a set of promoter fusions to a reporter gene for each promoter region. The fusions were constructed by inserting DNA fragments in front of the promoterless lacLM operon in the transcriptional fusion vector pAK80. The resulting plasmids were introduced into L. lactis MG1363. After exponential growth of the transformants in GSA medium with or without purine additions, the level of plasmid-encoded β-galactosidase activity was determined. Figure 3B shows the localization of the purD DNA fragments present in front of lacLM together with the resulting β-galactosidase levels. The end points of the fragments, which were all generated by PCR amplification of DNA derived from strain CHCC285, are shown in Fig. 1B (see also Table 3 for details). When the levels of β-galactosidase produced from the purD transcriptional fusion plasmids were analyzed (Fig. 1), the region upstream of the PurBox was found to play no role in purine regulation (compare pSH4 and pSJ2). The promoter −10 and +1 regions, however, were crucial for purD-lacLM transcription (compare pSH4 and pSH5). The importance of the PurBox is indicated by the constitutive low level of transcription which was produced from the promoter when the G7 residue in the PurBox was mutated to a C (compare pSJ5 and pSJ2).

β-Galactosidase production from plasmids pLN95, pLN96, and pLN97 was reported previously (20). Despite the difference in genetic backgrounds and growth conditions, we found that pLN95 was regulated 34-fold, in accordance with the previous report. Deletion of the region upstream of the PurBox in pLN96, where the fusion to the pAK80 sequences changes the PurBox sequence from AATACCGAACAAT to TTCGCCGAACAAT (Table 4), resulted in a drastic reduction in promoter activity, and regulation was reduced to 17-fold. These results indicate that the nucleotides before the CCGAAC hexamer in the PurBox are important for its proper function. A sixfold-higher level of β-galactosidase expression was found for pLN95 compared to pSH4, which carries a deletion of all DNA downstream of position 50. The difference in β-galactosidase levels could be explained by differences in mRNA stability or could be the result of increased translation of lacLM resulting from translational coupling to the truncated purD gene. Upon sequencing of the different fusion constructs, we noted that the three translation stop sites in pAK80 were located so close to the lacL start site that lacL was likely to be translationally coupled to an upstream gene. The possibility that sequences within the deleted region are required for maximal purD transcription in the intact gene cannot be ruled out, however.

The levels of β-galactosidase produced from the purC transcriptional fusion plasmids are shown in Fig. 3A. As for purD, we detected no significant differences in promoter activity when the region upstream of the two PurBox’s was removed (compare pSH2 and pSW1), while deletion of the promoter −10 and +1 regions led to a loss of purC-lacLM transcription (compare pSW1 and pSW5). Deletion of the purC distal PurBox (PurBox C1) resulted in a drastic decrease in promoter activity under both growth conditions (Fig. 3), to only 10% the activity achieved with both PurBox’s present (compare pSW2 and pSW1). The importance of the remaining PurBox is indicated by the residual low level of transcription produced from the promoter when the G7 residue in the PurBox was mutated to a C (compare pMK1013 with pSW2). We have no explanation for why plasmid pMK1013 resulted in ninefold purine regulation whereas the analogous purD fusion (pSJ5) resulted in unregulated expression. To be able to extrapolate our conclusions from purine regulation on plasmids to regulation of the genes present on the chromosome, we wanted to ensure that transcription was initiated at the same site. Therefore, we performed primer extension analysis of both purC and purD transcripts from pLN95 and pSW1. The data in Fig. 2B and C show that the transcriptional start sites were identical in the two situations (compare Fig. 2A and B and compare Fig. 2C with the previously determined start site shown in Fig. 1B [20]), whether the promoter regions were present in the promoter fusion vectors in MG1363 (L. lactis subsp. cremoris) or on the chromosome in CHCC285 (L. lactis subsp. lactis).

Conclusion.

From the analysis of the purine-mediated regulation of purC and purD transcription, we can conclude that efficient transcription requires a PurBox in the promoter region. We do not have enough data to conclude that the activating PurBox has to be located with the central G7 exactly at position −76 relative to the transcriptional start site, but the fact that both promoters share this feature leads us to expect that this is the case. It is interesting that the number of PurBox’s in the purC promoter affects only the activity and not the regulation of the promoter. This finding may not, however, be too surprising when it is taken into account that the two PurBox’s are spaced by 17 bp, which corresponds to 1.65 helical turns or an angular spacing of approximately 130°. So, if the RNA polymerase is able to make contact with an activator protein bound to PurBox C2, as we expect, it should be unable to make a similar contact with the corresponding region on an activator protein bound to PurBox C1, as this is turned 130° away. It is more likely that two activators are able to form protein-protein contacts while bound to the DNA and that the resulting dimer has an enhanced binding affinity due to the presence of two DNA binding sites, as has been seen for many repressors (2). If more PurBox’s were present, it could be envisioned that the multimerization of activators would result in a helix-like protein structure along the DNA. This suggestion may seem a little premature in the present context, but in the accompanying report (12) we identified a regulatory gene, purR, which is required for pur gene expression, and we present genetic evidence that the PurBox’s identified in the present study are binding sites for the PurR protein. Interestingly, the L. lactis PurR activator is very similar to the purR product from B. subtilis, which is a repressor of pur expression (33). Although the regulatory effects of the two PurR proteins are opposite, in the accompanying report (12) we present a unifying model which focuses on binding of the PurR proteins to PurBox’s. Sequences resembling PurBox’s are present in all PurR-regulated genes in both L. lactis and B. subtilis (12). The binding of the B. subtilis PurR repressor to its target DNA has been found to involve an extended DNA region of approximately 60 to 80 bp (29) which was proposed to be due to wrapping of the DNA around PurR (33). The data, however, could also be interpreted as the result of PurR multimerization along the DNA, as proposed above for the L. lactis PurR activator.