University of Groningen Interaction between ArgR and AhrC controls regulation of arginine metabolism in Lactococcus lactis

The expression of arginine metabolism in Lactococcus lactis is controlled by the two homologous transcriptional regulators ArgR and AhrC. Genome sequence analyses have shown that the occurrence of multiple homologues of the ArgR family of transcriptional regulators is a common feature of many low-G + C Gram-positive bacteria. Detailed studies of ArgR type regulators have previously only been carried out in bacteria containing single regulators. Here, we present a first characterization of the two L. lactis arginine regulators by means of gel retardation and DNase I footprinting. ArgR of L. lactis was shown to bind to the promoter regions of both the arginine biosynthetic argCJDBF operon and the arginine catabolic arcABD1C1C2TD2yvaD operon, but in an arginine-independent manner. Surprisingly, AhrC alone was unable to bind to DNA. Arginine-dependent DNA binding was obtained by mixing the two regulators in gel retardation assays. With both regulators present, the addition of arginine led to increased binding of ArgR-AhrC to the biosynthetic argC promoter but also to diminished binding to the catabolic arcA promoter. Footprinting showed ArgR-AhrC protection of regions containing ARG box operator sequences preceding argC. In the absence of AhrC, ArgR protected sites in the arcA promoter region with similarity to ARG box half-sites, here called ARC boxes. We propose a model for repression of arginine biosynthesis and activation of catabolism by anti-repression, involving arginine-dependent interaction between the two L. lactis regulator proteins, ArgR and AhrC.


Interaction between ArgR and AhrC Controls Regulation of Arginine Metabolism in Lactococcus lactis*
Despite differences in the organization of genes involved in arginine metabolism, experimental evidence indicates that the mechanism of arginine-dependent regulation of these genes is highly conserved among a range of different organisms, including Gram-negative, Gram-positive and extremophilic bacteria (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12). Regulation is exerted by binding of single transcriptional regulators of the ArgR family to so-called ARG operator sites preceding the relevant target genes, generally leading to repression of arginine biosynthetic genes and activation of catabolic genes, in the presence of arginine.
Crystal structures of the ArgR type transcriptional regulators of Escherichia coli (ArgREc (13,14)), Bacillus stearothermophilus (ArgRBst (15)), and Bacillus subtilis (AhrCBsu (16)) have revealed these to be structurally similar proteins, making up a complex of six identical subunits. The subunits are arranged in hexameric structures, which are organized as dimers of trimers. In E. coli and B. subtilis, the hexameric structure is maintained both in the absence and presence of arginine (4,17), whereas the regulator of B. stearothermophilus mainly exists as a trimer that assembles into hexamers dependent of the concentrations of arginine, protein, and DNA (5,15). Six arginine molecules are bound at the trimer-trimer interface, strengthening the interaction between the trimers and at the same time introducing a conformational change in the regulator, thus increasing its affinity for operator binding (4).
An ArgR monomer consists of an N-terminal DNA-binding domain, a central hinge region, and a C-terminal multimerization and arginine-sensing domain. In hexameric form, the DNA-binding domains surround the core of C-terminal domains (14 -16). Mutagenesis studies of mainly ArgREc have allowed identification of specific amino acid residues making up the N-terminal winged helix-turn-helix DNA binding region (18). Additionally, a range of residues in the C-terminal domain has been shown to be important for either subunit multimerization or arginine binding (18 -20).
ARG operator sites consist of pairs of 18-bp palindromic sequences (called ARG boxes), of which the 5Ј-TnTGnATwww-wATnCAnA-3Ј (where conserved residues are capitalized, n represents any nucleotide, and w represents A or T) consensus sequence in E. coli (21) is conserved with only small variations in various other organisms studied (22). The distance between the ARG boxes varies between 2 bp (e.g. for the B. stearothermophilus argC operator) and 3 bp (for the E. coli biosynthetic argC O1 operator). This spacing means that the boxes are aligned on the same side of the DNA helix. Also, single ARG boxes can confer regulator binding and regulation. This is exemplified by the arginine catabolic rocABC and rocDEF operons of B. subtilis (23)(24)(25) and the biosynthetic argGHCJBD operon of Thermotoga maritima (9). ARG box sequence variation, spacing, and location are factors that determine the strength of regulator-DNA interaction.
Whereas single ArgR-type regulators have been studied in detail, the continuously increasing number of bacterial genome sequences becoming available make it clear that several low-G ϩ C Gram-positive organisms harbor multiple homologues of ArgR type regulators (see overview by Belitsky (26)). A few recent investigations have proven that these ArgR homologues are not merely orthologous but fulfill distinct functions in these organisms. A study in Enterococcus faecalis revealed the presence, upstream of the arginine catabolic arcABCRD operon, of two genes named argR1 and argR2 (10). Although the function of the E. faecalis ArgR-type regulators was not investigated, it was proven that the divergently transcribed argR1 and argR2 genes were differentially expressed in response to arginine and glucose, possibly via putative ARG boxes preceding the genes (10). In our laboratory, a random knock-out screening led to the identification of the argR and ahrC genes in L. lactis, the gene products of which were responsible for repression of the arginine biosynthetic gltSargE operon (12). Further characterization showed that both ArgR and AhrC of L. lactis are necessary for repression of the arginine biosynthetic argCJDBF, glt-SargE, and argGH operons; they do not complement each other. Interestingly, arginine-dependent regulation of the arginine catabolic arcABD1C1C2TD2yvaD operon also required both ArgR and AhrC, but in a manner different from that of arginine biosynthesis. Whereas deletion of argR resulted in constitutively increased expression of the arc genes, deletion of ahrC gave constitutively decreased expression. However, arc expression was increased in an L. lactis argR ahrC double mutant, indicating that AhrC is not a classical activator of arc expression and, additionally, that ArgR might act as a repressor of arc expression (12). A thorough recent study of arginine regulation in Lactobacillus plantarum showed that repression of arginine biosynthesis was abolished when point mutations were introduced in either one of two separate genes encoding putative ArgR-type regulators or in promoter regions containing ARG box-like sequences (11).
In this work, we have sought to clarify the molecular basis for the complex dual mechanism of ArgR-AhrC-mediated regulation in L. lactis. To this end, purified ArgR and AhrC were investigated for their function in DNA binding and arginine sensing, with respect to both repression of arginine biosynthesis and activation of catabolism. The experimental evidence allowed us to propose a comprehensive model of ArgR-AhrCmediated gene regulation in L. lactis.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Media-Strains of Lactococcus lactis ssp. lactis (listed in Table I) were routinely cultivated at 30°C in M17 (27) medium containing 0.5% (w/v) glucose (GM17). For primer extensions and citrulline determinations, cells were grown in a chemically defined medium (CDM15) as described previously (28), with 0.5% (w/v) glucose as carbon source and free amino acids as nitrogen source. Arginine was added to the CDM15 as described throughout. When required, 4 g/ml erythromycin (Em), 4 g/ml chloramphenicol (Cm), or 40 g/ml 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) was added to the growth medium. Chemicals and antibiotics were purchased from Merck and Sigma. For induction of genes cloned behind the nisA promoter, Nisaplin (Aplin & Barnett Ltd., Beaminster, Dorset, UK) was suspended 1:1 (w/v) in 50% ethanol, thoroughly vortexed, and centrifuged (5 min at 12,000 rpm), after which the supernatant was added 1:1 ϫ 10 Ϫ6 (v/v) to the culture, unless stated otherwise.
DNA Isolation and Manipulation-General molecular techniques were performed as described by Sambrook et al. (29). Chromosomal and plasmid DNA was isolated from L. lactis according to Johansen and Kibenich (30) and Birnboim (31), respectively. L. lactis and E. coli were transformed with plasmid DNA by electroporation as described by Holo and Nes (32) using a Bio-Rad Gene Pulser (Bio-Rad). All DNA modification enzymes were purchased from Roche Applied Science, and used according to the manufacturer's directions. PCRs were performed using Pwo DNA polymerase (Roche Applied Science) and purified with the Roche PCR purification kit (Roche Applied Science). Primers (listed in Table II) were purchased from Biolegio BV (Malden, The Netherlands).
Construction of Regulator Deletion Mutants of NZ9000 -Since L. lactis strains MG1363 and NZ9000 are isogenic, the argR and ahrC deletion plasmids pORI⌬argR and pORI⌬ahrC, made with MG1363 chromosomal DNA as template (12), were used to delete these genes from NZ9000. Single crossover integration and excision in NZ9000 was done using pVE6007 (33) as helper plasmid, as described before (12), yielding L. lactis NZ⌬ahrC and L. lactis NZ⌬argRahrC (Table I). Chromosomal deletions were confirmed by PCR and by Southern blotting. Probe-labeling, hybridization, and detection was done with the ECL direct nucleic acid labeling system (Amersham Biosciences), according to the manufacturer's instructions.
Overexpression and Isolation of His-tagged ArgR and AhrC Proteins-The argR and ahrC genes were amplified from MG1363 chromosomal DNA with the primer pairs argR-Nhis1/argR-MAL2 and ahrC-Nhis1/ahrC-His2, respectively, thereby introducing N-terminal hexahistidine tags (His tags). The PCR products were cloned as NcoI/ HindIII and NcoI/XbaI fragments, respectively, in the multiple cloning Construction of Regulator Point Mutations-The argR and ahrC genes, amplified by PCR using the argR-NZ/argR-MAL2 and ahrC-NZ/ ahrC-5 primer pairs, respectively, were blunt end-cloned into the SmaI restriction site of pUC19 (35). The proper constructs were picked up in E. coli XL1-Blue. Point mutations were introduced in argR and ahrC using the protocol of the QuikChange TM site-directed mutagenesis kit (Stratagene, La Jolla, CA). Whole-plasmid PCR was performed using the native Pfu DNA polymerase (Stratagene) to make the following ArgR and AhrC mutations (primers used are shown in parentheses): Table II). Mutations were verified by nucleotide sequencing, and mutated genes were subsequently cloned as RcaI/ HindIII and RcaI/XbaI restriction fragments into the NcoI/HindIII and NcoI/XbaI sites of the PnisA expression vector pNG8048E, respectively, using L. lactis NZ9000 as the cloning host. The plasmid constructs were again verified by nucleotide sequencing and used to transform L. lactis strains NZ⌬ahrC and NZ⌬argRahrC.
DNase I Footprinting Assays-His 6 -ArgR and His 6 -AhrC DNA binding sites were analyzed by DNase I footprinting (protection) assays, largely according to the protocol of the Sure Track Footprinting Kit (Amersham Biosciences). The PargC region was amplified with the argC-7 (forward) and argC-2 (reverse) primers, one of which was endlabeled (2 h at 37°C) with [␥-32 P]ATP using T4 polynucleotide kinase (Amersham Biosciences) according to the manufacturer's instructions, before standard PCR with the respective unlabeled primers. The ParcA region was likewise amplified and labeled, using the arcA-1 (forward) and arcA-7rev (reverse) primers. Binding reactions were performed as described for the gel retardation assays (see above), except that the final volume was 40 l, and 150,000 cpm of labeled DNA was used per lane. Concentrations of His 6 -ArgR, His 6 -AhrC, and arginine were as specified. DNase I (Amersham Biosciences) degradation and fragment separation by polyacrylamide gel electrophoresis (National Diagnostics) were performed as described previously (37). Detection was performed as described for the gel retardations (see above). Maxam-Gilbert sequencing reactions were made from the footprinting probes according to Sambrook et al. (29) and were run next to the footprinting lanes to determine the sizes of degradation fragments.  Table II. End-labeled probes were incubated with His 6 -ArgR or His 6 -AhrC, and retardation was investigated by electrophoresis on 6% polyacrylamide gels. A, His 6 -ArgR was used in the following concentrations (in monomer equivalents). Lanes 1, no regulator. RNA Isolation and Primer Extension-RNA was isolated from cells grown to the midexponential phase of growth (A 600 ϭ 0.6 -0.7) in CDM15 with 0.1 or 10 mM arginine. RNA isolation was carried out using macaloid to remove DNA and the High Pure RNA isolation kit (Roche Applied Science). RNA quantity was determined spectrophotometrically (29), and RNA quality was verified on an Agilent Bioanalyzer 2100 using RNA 6000 LabChips (Agilent Technologies Netherlands BV, Amstelveen, The Netherlands). Oligonucleotide arcA-px was end-labeled with [␥-33 P]ATP using T4 polynucleotide kinase (Amersham Biosciences) and purified with the QIAquick nucleotide removal kit (QIAgen GmbH, Hilden, Germany). The labeled oligonucleotide was used for synthesis of first strand cDNA with 5 g of total RNA as template, using the SuperScriptIII reverse transcriptase (Invitrogen) by incubating for 10 min at 25°C and 40 min at 42°C. The enzyme was inactivated by heating at 70°C for 15 min. Primer extension products were analyzed by electrophoresis (National Diagnostics) next to a Maxam-Gilbert sequencing ladder, made from an arcA-px/arcA-1 PCR product, using the [␥-33 P]ATP end-labeled arcA-px oligonucleotide, as described for footprinting assays (see above).
Citrulline Determination-Intracellular citrulline concentrations were determined in cell-free extracts of L. lactis strains harvested at the midexponential phase of growth in CDM15 with 10 mM arginine. Citrulline measurements were done essentially according to Archibald (38).

Isolation of His 6 -ArgR and His 6 -AhrC Reveals an Unusually
Stable Multimeric Complex of AhrC-DNA fragments encoding N-terminally hexahistidine-tagged derivatives of the two arginine regulators ArgR and AhrC of L. lactis MG1363 were cloned behind the nisin-inducible nisA promoter in pNG8048E. The His-tagged regulators, His 6 -ArgR and His 6 -AhrC, were overproduced in L. lactis NZ9000⌬argRahrC to prevent copurification with the wild-type regulator proteins and isolated to near purity as determined by SDS-PAGE (Fig. 1). His 6 -ArgR appeared as a protein of ϳ15 kDa, which is well in agreement with the expected size of the regulator in monomeric form (Fig.  1A). During our studies, His 6 -AhrC consistently appeared in several bands corresponding to high molecular weight proteins when investigated by SDS-PAGE (Fig. 1A). Expression of the wild-type AhrC protein in L. lactis as well as in E. coli BL21(DE3) also yielded high molecular bands in SDS-PAGE, in addition to a band expected for the monomeric form of the protein, despite sample boiling prior to electrophoresis and electrophoresis under denaturing conditions (data not shown). Thus, the stable His 6 -AhrC complexes were not caused by the His tag. The ability of the His-tagged regulators to complement the argR and ahrC deletion mutants was examined by measuring intracellular citrulline. These studies showed that the His tags did not abolish regulator functionality (data not shown). To make sure that the high molecular weight bands observed during SDS-PAGE of His 6 -AhrC were not caused by contaminating, co-purified proteins, proteins in these bands were proven to be identical to AhrC of L. lactis by MALDI-TOF analysis (data no shown). Furthermore, purified samples of His 6 -AhrC were denatured by incubation and electrophoresis in 8 M urea (Fig. 1B). Samples were boiled for increasing periods of time until complete dissociation to the monomeric form was observed (Fig. 1B). Surprisingly, boiling of up to 30 s in 8 M urea was required for complete denaturation of the high molecular weight His 6 -AhrC form, which indicates that this regulator forms unusually stable multimeric structures. Whether this is a result of overexpression or in vitro purification conditions remains to be determined.
Gel Retardation Experiments Reveal Differences between ArgR and AhrC-The functions of ArgR and AhrC were initially investigated by gel retardation experiments. Our earlier studies have shown that both regulators are involved in transcriptional repression of the arginine biosynthetic genes. A more complex mechanism, also requiring both regulators, is responsible for regulation of arginine catabolism (12). DNA fragments covering the biosynthetic argC and the catabolic arcA promoter regions were chosen as probes in gel retarda- experiments. An intragenic glnA fragment was used as negative control. His 6 -ArgR was able to shift all three probes ( Fig. 2A). The apparent dissociation constants (K d ) for His 6 -ArgR were calculated to be 1.4 ϫ 10 Ϫ9 M for PargC, 5.3 ϫ 10 Ϫ10 M for ParcA, and 1.3 ϫ 10 Ϫ7 M for the unspecific glnA probe. Besides the ϳ2.5-fold higher affinity of His 6 -ArgR for ParcA than for PargC, more protein-DNA complexes were obtained with ParcA than with PargC ( Fig. 2A), suggesting the presence of more operator regions in the former. Surprisingly, His 6 -AhrC did not retard any of these probes (Fig. 2B) despite the fact that the protein contains a highly conserved N-terminal DNA binding domain. Importantly, cell-free extracts of L. lactis with overexpressed AhrC also failed to shift any of these probes (data not shown). Retardation experiments using either His 6 -ArgR or His 6 -AhrC were performed without arginine or in the presence of 10 mM arginine, but no difference was observed. The binding of His 6 -ArgR to the supposedly unspecific glnA probe revealed that ArgR has an intrinsic DNA binding ability and allowed us to work at His 6 -ArgR concentrations that were specific for the argC and arcA promoter fragments during the remainder of the study.
Interaction between ArgR and AhrC Is Necessary for Regulation in Response to Arginine-The specific binding of His 6 -ArgR to the catabolic arcA promoter as well as to the biosynthetic argC promoter, the lack of His 6 -AhrC-DNA interaction, and the knowledge that both regulators are required for arginine-dependent regulation, led us to perform gel retardation experiments in the presence of both proteins. Using concentrations of His 6 -ArgR that only partially shifted the free probes, His 6 -AhrC was added, with and without 10 mM arginine (Fig. 3). The addition of His 6 -AhrC had no effect on His 6 -ArgR-mediated band shifts in the absence of arginine (Fig. 3, A and B). However, in the presence of arginine, clear but opposite effects were observed for the two different promoter fragments. Whereas His 6 -AhrC increased the affinity of His 6 -ArgR for PargC (Fig.  3C), His 6 -ArgR-mediated interaction with ParcA was completely lost (Fig. 3D).
His 6

-ArgR and His 6 -AhrC Interact with ARG Box-like Operators in the Biosynthetic argC Promoter Region-
The binding of His 6 -ArgR-His 6 -AhrC to the argC promoter region was further investigated by DNase I footprinting. The concentration of His 6 -AhrC was increased in the presence of a fixed, low amount of His 6 -ArgR and 10 mM arginine. Two operator sites of 20 -25 bp, here called argC O1 and argC O2 , were protected in both strands of the argC promoter fragment in an His 6 -AhrC-dependent manner (Fig. 4A). Visual inspection of the protected residues showed that the two sites have high similarity to classical ARG box operators known to be required for binding of ArgR-type regulators in several organisms (5Ј-TnTGnATwww-wATnCAnA-3Ј, where n represents any nucleotide, w is A or T, and capitalized residues are highly conserved) (Fig. 4, B and C). The two ARG boxes are separated by a 32-bp spacer region that contains hypersensitive residues on both strands, suggesting that bending of DNA takes place between the two sites as a result of His 6 -ArgR-His 6 -AhrC binding (Fig. 4). DNase I footprinting experiments using PargC and His 6 -ArgR alone did not give clearly protected sites (data not shown), possibly because of the weak affinity of His 6 -ArgR for PargC, compared with that of His 6 -ArgR-His 6 -AhrC.

His 6 -ArgR Binds to Several ARG Box Half-sites in the arcA
Promoter Region-In contrast to the argC promoter region, no consensus ARG box(es) could previously be identified in the promoter region of arcA. Since His 6 -AhrC diminishes the binding of His 6 -ArgR to ParcA (Fig. 3), footprinting of this promoter region was performed with His 6 -ArgR alone. Although binding of His 6 -ArgR was weak, protected regions could still be discerned in ParcA (Fig. 5A). Interestingly, inspection of the protected sites revealed a high similarity of these to ARG box half-sites of the sequence 5Ј-TGnATAWW-3Ј (where n represents any nucleotide; W is A or T; and capital letters represent conserved residues) (Fig. 5, B and C). Some of these ARG-half sites (called ARC boxes below) are positioned immediately next to each other without spacing, whereas others are present as single boxes (Fig. 5B). Weakly hypersensitive sites were identified between the sites denoted as C 1 C 2 and D 1 D 2 , shown in Fig. 5, located on the predicted P2 and P1 promoter regions, respectively (Fig. 5).
To confirm the protection assays, overlapping ParcA fragments of the same size (ϳ100 bp) were used in gel retardation assays. All fragments (except the arcA and glnA intragenic controls) gave low molecular weight complexes, whereas fragments containing the central B and C 1 C 2 regions additionally resulted in a high molecular weight complex (Fig. 6). However, the relative amounts of shifted versus free probes define a center of binding around the 5/5rev fragment (Fig. 6C). Although direct correlation of these results to those of the protection assays is not possible, a number of important conclusions can still be drawn. First, His 6 -ArgR binds specifically to several sites spanning the ParcA region. Second, the double ARC box at location C 1 C 2 (Fig. 6C) is not essential for His 6 -ArgR binding, since fragments that only contain the ARC boxes A or D were still retarded (Fig. 6). Third, strong His 6 -ArgR binding was centered around the ARC boxes B and C, which cover the putative P2 promoter.
The arcA P1 Promoter Is Regulated in Response to Arginine-The arcA promoter region contains two core promoter sequences, suggesting that transcription of the arc operon genes might initiate and/or be regulated at two different sites. To answer this question, primer extension analysis was performed using total RNA isolated from the wild type strain L. lactis MG1363, the arginine regulator single mutants MG⌬argR and MG⌬ahrC, and the double mutant MG⌬argRahrC, grown in high (10 mM) or low (0.1 mM) concentrations of arginine. A primer annealing ϳ100 bp downstream from the Ϫ10 region of arcA P1, was used in the reverse transcription reactions. Only very weak bands were observed in the 33-bp space between P2 and P1, suggesting that P2 has no or only low activity under the conditions applied. In contrast, a strong band appeared at a T residue 6 bp downstream of P1, indicating that P1 most likely is the main arc promoter (Fig. 7). Additionally, the primer extensions showed that transcription from arcA P1 is strongly regulated by the availability of arginine in the wild type strain (Fig. 7). In the ahrC deletion strain, no expression was seen, and in the argR single mutant and the argR ahrC double mutant, high expression was observed irrespective of the arginine concentration in the growth medium (Fig. 7).
AhrC The open circles indicate residues protected in footprints on the forward strand (above the sequence) and reverse strand (below the sequence). Hypersensitive residues are indicated by vertical arrows. C, alignment of the L. lactis ARG box operators and resulting consensus sequence: argC O1 and argC O2 are from this study; argG O1 , argG O2 , gltS O1 , and gltS O2 are predicted from promoter sequences (12). The E. coli ARG box consensus is according to Maas (21). The convergent arrows indicate ARG box dyad symmetry. structures of ArgR-type regulators from E. coli, B. stearothermophilus, and B. subtilis have shown that arginine bound to the proteins interacts with two conserved aspartate residues in the C-terminal sensing domain. However, the situation is different in L. lactis and other low-G ϩ C Gram-positive organisms (Fig. 8A). ArgR of L. lactis has only one of the two Asp residues, whereas AhrC has three (12) (Fig. 8). In order to evaluate the importance of these Asp residues in the regulators in L. lactis, two mutations were introduced in each regulator (see Fig. 8B). The function of the mutated regulators was determined by expression in ahrC and argR ahrC mutants of L. lactis NZ9000. The intracellular concentration of citrulline in the strain was determined as a measure of arginine degradation via the arc operon-encoded arginine deiminase (ADI) pathway. ArgR(D127G) and ArgR(A126D) behaved like wild type ArgR. Thus, the conserved ArgR(D127) is not important for arginine sensing in L. lactis, and the introduction of an Asp residue at ArgR(Ala 126 ) could not complement the AhrC deletion. Considering that ArgR(Asp 127 ) and AhrC(Asp 126 ) of L. lactis are conserved in all aligned regulators (Fig. 8A), it was surprising that also AhrC(D126G) activity was almost that of the wild type AhrC. However, the "extra" Asp 124 of AhrC is of major importance for activity, since AhrC(D124G) resulted in a drastic reduction of citrulline production via the ADI pathway (Fig. 8B). DISCUSSION The work presented in this paper was aimed at clarifying the specific functions of the two arginine regulators ArgR and AhrC in arginine metabolism and regulation of L. lactis. We approached the question mainly by gel retardation and footprinting analysis. Eventually, the obtained results, to be discussed below, have led us to propose the regulatory model shown in Fig. 9.
Since putative ARG box operators could be predicted in the argC promoter region (12), His 6 -ArgR binding to this promoter fragment was expected. However, His 6 -ArgR showed even higher affinity (ϳ2.5-fold) for the arcA promoter, which lacks consensus ARG box sequences, than for the argC promoter. The observation that the addition of arginine had no effect on His 6and the ribosomal binding site is underlined. The open circles indicate residues protected in footprints on the forward strand (above the sequence) and reverse strand (below the sequence). Hypersensitive residues are indicated by vertical arrows. C, alignment of the L. lactis ARC boxes and resulting consensus sequence. The arrow underneath the alignment refers to one-half of an ARG box (compare with Fig. 4C).  2-8 in A). D, schematic representation of the argS-arcA intergenic region. The black squares indicate the Ϫ10 and Ϫ35 motifs of the P1 and P2 core promoter regions, and the transcription start site (TSS) is shown with a bent arrow. A terminator structure is indicated by a lollipop. The white boxes below the DNA line represent ARC operator sites, labeled A-D. PCR fragments used in the gel retardation assays in A, 1/1rev to 7/7rev, are indicated as horizontal lines.
ArgR binding suggested that ArgR does not carry out arginine regulation alone or that arginine is not the actual ArgR effector molecule. Equally surprising was the lack of probe-DNA binding by His 6 -AhrC although the protein contains an N-terminal H-T-H DNA-binding domain that is highly conserved among ArgR-type regulators (12). The observation that deletion of ahrC results in derepression of arginine biosynthesis led to the initial conclusion that AhrC binds to ARG operators preceding the arginine biosynthetic genes. The stable His 6 -AhrC complex could explain these inconsistencies. However, the ability of His 6 -AhrC to complement an ahrC mutation, the fact that a high molecular weight complex was also observed when overexpressing wild type AhrC in E. coli as well as in L. lactis, and the failure of overexpressed wild type AhrC to mediate DNA binding as well motivated us to proceed with His 6 -AhrC in these studies.
The arginine-independent DNA binding of His 6 -ArgR and the lack of binding by His 6 -AhrC, together with the knowledge that both regulators are required for arginine regulation (12), prompted us to perform gel retardation experiments, using both His 6 -ArgR and His 6 -AhrC. Indeed, arginine-dependent interaction with DNA fragments containing ARG or ARC boxes only took place in the presence of both regulators. His 6 -AhrC increased the His 6 -ArgR (or His 6 -ArgR-His 6 -AhrC) affinity for PargC considerably but decreased the His 6 -ArgR affinity for ParcA. A peculiarity, however, was seen in the shifts of the argC promoter fragment. Since His 6 -ArgR already forms a complex with the PargC probe, the increased shift correlating with the increase in the concentration of His 6 -AhrC could be expected to result in the formation of one or more additional retardation complexes. This was not the case, since even under conditions where an almost complete shift (Fig. 3C, lane 8) of the PargC probe was seen, only a single retardation band was observed. One explanation for this result could be that AhrC transmits the arginine signal to ArgR, without actually binding FIG. 7. Primer extension of the arcA promoter. Primer extension reactions were performed with primer arcA-px on total RNA isolated from MG1363 (wild type), MG⌬argR, MG⌬ahrC, and MG⌬arg-RahrC, harvested at midexponential phase of growth in CDM15, with 0.1 and 10 mM arginine. C ϩ T, Maxam-Gilbert sequencing ladder made from PCR product of primer arcA-px (labeled) and arcA-1 (unlabeled). Part of the arcA promoter region, including the Ϫ10 region of promoter P1, indicating the exact transcriptional start site is shown on the right. to ArgR and/or DNA. Alternatively, ArgR and AhrC complexes could be able to exchange subunits. Since all ArgR-type regulators characterized in detail so far have trimeric or hexameric quarternary structure (as dimers of trimers (13)(14)(15)(16)), it is possible that each regulator is able to form homogenic dimers of trimers and, in the presence of arginine, form a heterogenic structure (e.g. consisting of an ArgR trimer bound to a trimer of AhrC). Examination of B. stearothermophilus ArgR has shown that binding of arginine in the trimer-trimer interface results in a rotation of one trimer relative to the other, which is proposed to increase the specificity for interaction with ARG box operators (15). It is tempting to speculate that a putatively hexameric ArgR of L. lactis has affinity for ARC operators in the absence of arginine but that the interaction with arginine and AhrC results in increased affinity and specificity for ARG operators concomitant with a decrease in the affinity for ARC operators. Along the same line of reasoning, an interaction between the two arginine regulators, ArgR1 and ArgR2 of L. plantarum, was proposed in order to explain the observation that introduction of point mutations in any one of the two DNA binding domains resulted in complete derepression of arginine biosynthesis (11).
The operator sites in the argC promoter region are highly similar to the 18-bp ARG boxes of E. coli (21). The presence of identical motifs in the promoter regions of the two other arginine biosynthetic operons, gltSargE and argGH (12), is in good agreement with this result and suggests that a similar mechanism of transcriptional regulation takes place at these promoters. By combining the ARG boxes of all three biosynthetic promoters, we were able to construct an ARG box consensus sequence for L. lactis (Fig. 4C). Considering the increased affinity of His 6 -ArgR-His 6 -AhrC for the ARG operators compared with that of His 6 -ArgR alone, plus the conserved DNA binding domain of L. lactis AhrC, it is most likely that His 6 -AhrC takes part in ARG box binding. Since ArgR is able to bind ARG box half-sites, one half of an ARG box might be occupied by ArgR, and the other half might be occupied by AhrC. This would also explain why the operators of the biosynthetic promoters and the catabolic promoter are different, namely to achieve differential regulation. The ARG boxes of L. lactis differ from those of most other systems by the presence of a large interoperator spacer region. Such spacer regions are generally 2-3 bp in E. coli, B. stearothermophilus, B. subtilis, and Thermotoga neapolitana (8,21,25), compared with 32 bp for the PargC operators and possibly 75 and 10 bp for the PgltS and PargG operators, respectively (with ARG box lengths of 18 bp). No clear difference in affinity of the regulators for argC O1 and argC O2 was apparent, and the presence of hypersensitive residues in the DNA footprint of the PargC region between the two operators suggests that DNA bending takes place. Bending could be the result of interaction between two regulators occupying the two sites or of looping of promoter DNA, leading to interaction of the DNA with two different DNA-binding regions of one regulator, as suggested for argC O1 and argC O2 of B. subtilis (4). Investigation of single box affinities will be required for elucidating the exact mechanism.
ArgR interacts with multiple operator sites (here called ARC sites), which are highly similar to ARG box half-sites and are present at various (about six) portions of the arginine catabolic arcA promoter region (Figs. 5 and 6). Interestingly, footprinting shows that ArgR protects single as well as double ARC boxes, and electrophoretic mobility shift assays using ParcA subclones suggest that ArgR-mediated regulation is centered around the C 1 C 2 double ARC box (Fig. 6). Except for the D 1 D 2 double box, all ARC boxes are located upstream of arcA P1, with the putative P2 core promoter sequence covered by as many as three boxes. Nevertheless, arginine-dependent transcriptional regulation appears to initiate at the arc operon proximal promoter arcA P1. An earlier ParcA deletion analysis using a low copy plasmid-encoded lacZ expression system revealed that expression of the arcA P1 minimal promoter was independent of arginine (12). By including the arcA P1 upstream region, corresponding to the 5Ј-ends of ParcA fragments 5/5rev and 6/6rev (Fig. 6C), arginine-dependent regulation was restored (12). The lack of regulation of arcA P1 lacking the upstream region, despite clear His 6 -ArgR binding, can be explained in two ways; the low copy plasmid system may lead to insufficient in vivo levels of ArgR to repress the promoter, or, alternatively, interaction between regulator subunits binding to the D 1 D 2 sites and the upstream A, B, and C 1 C 2 sites might be required for efficient arcA P1 regulation. Expression and regulation of arcA P2 cannot be unequivocally excluded, but under the conditions applied, P2 does not seem to be regulated in response to arginine. It is noteworthy that the biosynthetic ARG boxes are composed of converging ARC boxes, explaining why His 6 -ArgR (without His 6 -AhrC) is able to shift the ARG box-containing fragments as well as those containing only ARC boxes. The necessity of ArgR binding to the A and B operator sites of ParcA is unclear but may be a drafting mechanism to attract ArgR molecules to the catabolic promoter.
Mutation of the double Asp residues in the C-terminal domain of ArgREc has been shown to be detrimental for arginine sensing (18,19), and structural studies have suggested that these residues interact directly with arginine in the interface between the two ArgR trimers (13,15,16). Whereas double-Asp residues are conserved in ArgR regulators in organisms with a single ArgR regulator, large deviations in this region are observed in organisms with multiple ArgR-type regulators (Fig.  8A). Surprisingly, the fully conserved Asp 129 (ArgREc numbering), was not essential for the functioning of ArgRLl and Ahr-CLl. Moreover, since ArgR(A126D) was unable to replace the function of AhrCLl, these two residues are apparently not involved in arginine sensing in ArgRLl. The additional Asp 124 residue in AhrC, which is also present in ArgR4 of E. faecalis, was found to be of major importance for AhrCLl functioning. Possibly, this Asp residue of AhrCLl is able to complement the missing Asp residue of ArgRLl. It is tempting to speculate that AhrCLl and ArgR4Efa are responsible for arginine sensing, whereas the task of ArgRLl and (at least one of) the other ArgR regulators of E. faecalis is DNA binding.
Based on the results presented here, we propose a model describing the functions of ArgR and AhrC in arginine-mediated transcriptional regulation in L. lactis (Fig. 9). In the absence of arginine, the higher affinity of ArgR for ParcA than for PargC, possibly due to the additional ARC sites in the former, suggests that ArgR mainly occupies the arcABD1C1C2TD2yvaD promoter, preventing arginine degradation via the ADI pathway. At the same time, this leaves expression of the arginine biosynthetic argCJBDF, gltSargE, and argGH operons unrepressed, allowing for de novo arginine production. The addition of arginine leads to association of ArgR and AhrC in a complex with high affinity for the ARG box operators. ArgR is shifted from the arcA promoter to the ArgR-AhrC complex, which represses expression of the arginine biosynthetic genes. Accordingly, the arginine catabolic arc operon is now derepressed, allowing for utilization of the arginine as a nitrogen and energy source via the ADI arginine degradation pathway (Fig. 9). With ArgR acting as the main transcriptional repressor, AhrC appears to have the unusual dual function of co-repressor and anti-repressor.
Despite the high conservation between ArgR-type regulators of different bacterial species, we show that the mechanisms by which these proteins function are not conserved. This study extends our understanding of transcriptional regulation of arginine metabolism in organisms harboring more than one ArgR-type regulator, but intriguing questions remain to be answered. The subunit multimerization and overall structure of L. lactis ArgR and AhrC proteins is of major interest. Performing band shift or gel filtration experiments, using one wild-type regulator in combination with a functional fusion construct of the other, would verify the hypothesis of argininedependent interaction between ArgR and AhrC. Additionally, the multiple regulator system of L. lactis proposes that regulatory targets might exist in addition to the genes of the arginine metabolic pathways. Finally, the results presented here pose the question of why such a complex regulatory mechanism is operating in L. lactis, a renowned model organism because of its metabolic simplicity and low number of gene paralogues (39,40).