INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial sigma () factors confer promoter specificity to transcription initiated by the RNA polymerase holoenzyme (₂'). On the basis of structural and functional criteria,  factors fall into two major classes. Most  factors are similar to the major vegetative  factor of Escherichia coli, ⁷⁰ (18, 26). This sigma factor recognizes sequences similar to the canonical 35/10 type of promoter and directs transcription of many housekeeping genes. Besides ⁷⁰, this family contains several alternative  factors allowing cells to respond to many different environmental stimuli, each controlling a specific process such as sporulation, heat shock response, and flagellation (26). The second type of  factor, ⁵⁴ (RpoN, NtrA, or GlnF), shows little sequence similarity to the first class and has been identified so far in many gram-negative and gram-positive bacterial species. Although it was originally recognized for its role in nitrogen metabolism, it is clear that ⁵⁴ also controls the expression of genes responding to many other physiological needs (24, 31). Promoters recognized by RpoN share conserved DNA sequences of which the consensus is 5'-CTGGCAC-N₅-TTGCA-3' (24/12 type of promoter; the invariant dinucleotides GG/GC are in boldface) (2).

Whereas the ⁷⁰-holoenzyme complex often initiates transcription in the absence of transcriptional activators, transcription from all known ⁵⁴-dependent promoters has an absolute dependence on the presence of an activator protein, such as NifA or NtrC (9, 24, 49). In the latter case, activators often bind to sequences located more than 100 bp upstream from the transcriptional start site (5, 9). This difference is probably correlated with the ability of ⁵⁴, but not ⁷⁰, to bind (in vitro) to the promoter in the absence of RNA polymerase core enzyme (6, 12). The strong interaction of ⁵⁴ with its promoter keeps the ⁵⁴ holoenzyme-promoter complex in a closed conformation which may then require activation by another protein to induce local DNA melting and initiate transcription (38, 46).

In most bacteria in which the rpoN region has been analyzed, a common organization of downstream open reading frames (ORFs) is found (31). The function of the conserved ORFs has been studied in only a few bacterial species, and their role has not been clearly defined yet. These genes are thought to modulate the activity of ⁵⁴ (30). In Klebsiella pneumoniae, the inactivation of the two ORFs, ORF95 and ptsN, located immediately downstream from rpoN increases the expression from several ⁵⁴-dependent promoters (30, 32), while a mutation in the fourth gene (ORF90) reduces the activity of these same promoters (32). For Pseudomonas aeruginosa, it was suggested that, together with rpoN, ORF2 functions as a coinducer of genes involved in the assimilation of glutamine (20). In Caulobacter crescentus, ORF159 modulates the expression from a ⁵⁴-dependent flagellar promoter.

Bacteria belonging to the genera Rhizobium, Bradyrhizobium, Azorhizobium, Mesorhizobium, and Sinorhizobium (collectively called rhizobia) elicit nitrogen-fixing nodules on the roots of their leguminous host plant. The rpoN genes in four rhizobial species, Rhizobium meliloti (43), Bradyrhizobium japonicum (23), Azorhizobium caulinodans (48), and Rhizobium sp. strain NGR234 (52), have been characterized. In these organisms, ⁵⁴ has been shown to control C₄-dicarboxylate utilization (23, 43, 48), nitrate assimilation (23, 43, 48), and several symbiotic functions (23, 43, 48, 52). However, only partial DNA sequence information on the downstream regions of rhizobial rpoN genes is available (23, 43, 48). In addition, the role and regulation of the conserved ORFs located 3' to rpoN have not been investigated yet.

Here we describe the identification and analysis of the rpoN region of Rhizobium etli, the nodulating symbiont of the common bean plant, Phaseolus vulgaris. In contrast to the case for most other bacterial species, the rpoN gene is separated by approximately 1.6 kb from the conserved ORFs, ORF191 and ptsN, which are normally found immediately downstream from rpoN. We analyzed the regulation of rpoN transcription and the phenotypes of rpoN and ptsN mutant strains. Moreover, we identified an R. etli Tn5 mutant strain displaying reduced melanin production. A further examination of this mutant revealed that the phenotype resulted from a mutation in ptsA, the gene coding for enzyme I of the phosphoenolpyruvate:sugar phosphotransferase system (PTS). Our results indicate that in R. etli, ptsN and ptsA may coregulate several RpoN-dependent activities, including the activation of nifH, the production of the pigment melanin, and the assimilation of alanine. In addition, the ptsN and ptsA mutant strains display an increased sensitivity to the toxic effects of high concentrations of malate and succinate. Therefore, ptsN and ptsA are thought to be part of the same regulatory cascade which may be involved in the sensing of C₄-dicarboxylates.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmids, and culture conditions. The bacterial strains and plasmids used in this study are listed in Table 1. Some are also diagrammed in Fig. 1. Plasmids used merely for sequencing are not shown. The Rhizobium isolate used in the present study is CNPAF512 (CNPAB/EMBRAPA culture collection), a Brazilian isolate from P. vulgaris nodules. Based on our nucleotide sequence analysis of a 260-bp fragment of the 16S rRNA gene, amplified by PCR with the primers Y1 (5'-TGGCTCAGAACGAACGCTGGCGGC-3') and Y2 (5'-CCCACTGCTGCCTCCCGTAGGAGT-3'), and multilocus enzyme electrophoresis of this isolate (26a), CNPAF512 should be classified as R. etli. R. etli strains were routinely grown in liquid TY (0.5% tryptone, 0.3% yeast extract, 7 mM CaCl₂) at 30°C and maintained on yeast-mannitol agar plates. E. coli was grown in Luria-Bertani medium at 37°C. Antibiotics supplied to the medium were at the following concentrations: nalidixic acid and neomycin, 40 µg/ml; kanamycin and gentamicin, 30 µg/ml; and ampicillin, 100 µg/ml. Tetracycline was added to a final concentration of 1 µg/ml (for R. etli) or 10 µg/ml (for E. coli).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Physical map of the R. etli rpoN region. The 1.8-, 4.6-, and 6.1-kb EcoRI fragments hybridizing to the A. tumefaciens probe are shown above the physical map of the 5.6-kb region that was sequenced. Triangles represent insertions of the -Km interposon or the uidA-aphII cassette. The positions and orientations of the identified ORFs are indicated below the restriction map. Restriction sites are abbreviated as follows: B, BstXI; H, HindIII; E, EcoRI; N, NotI; P, PstI.

Growth tests. Tests of growth of R. etli in liquid medium were carried out in acid minimal salts (AMS) medium (37) containing 1 mM CaCl₂. Cells were first grown overnight in TY, washed twice in 10 mM MgSO₄, brought to an optical density (OD) at 600 nm of 0.02 in a Perkin-Elmer lambda 2 spectrometer, and diluted 100 times in AMS medium. Carbon and nitrogen sources were added to the appropriate concentrations with a sterile concentrated stock solution at pH 7.0. Cell growth was monitored in terms of turbidity at 600 nm in a microtiterplate reader. Carbon sources were D-()-mannitol, D-(+)-glucose, sucrose, L-(+)-arabinose, D-(+)-galactose, D-sorbitol, D-()-fructose, glycerol, trisodium citrate, cis-aconic acid, DL-isocitric acid, -ketoglutaric acid, succinic acid, fumaric acid, DL-malic acid, oxaloacetic acid, and pyruvic acid. Nitrogen sources were NH₄Cl, L-alanine, L-glutamine, and KNO₃. For growth tests at various pH values, 3 (N-morpholino)propanesulfonic acid (pH 7.0) and 2(N-morpholino)ethanesulfonic acid (pH 6.5, 6.0, and 5.5) were used at a concentration of 30 mM.

DNA methods. General DNA manipulations were performed as described previously (3, 45). DNA fragments were recovered from agarose gels by using the Nucleotrap kit (Macherey-Nagel). Southern blotting and hybridizations were carried out as described previously (3, 34, 45). DNA probes were labeled with a digoxigenin labeling and detection kit (Boehringer). To generate blunt ends to incompatible DNA fragments, DNA was incubated with Klenow or T4 DNA polymerase in the presence of the four deoxynucleoside triphosphates. Automated DNA sequencing was performed on a Pharmacia A.L.F. sequencer with fluorescein-labeled universal and synthetic oligonucleotide primers. Both strands of overlapping pUC18 subclones covering the 5,600-bp DNA fragment were read with multiple sequencing.

PCR conditions. Amplification of DNA fragments by PCR was performed in a TRIO-Thermoblock (Biometra). Twenty-five-microliter reaction mixes, containing 0.65 U of Taq DNA polymerase (Boehringer), each of the deoxynucleoside triphosphates at 200 µM, and each of the primers at 1 µM, were subjected to 30 cycles of incubation at 94°C for 60 s, 60°C (primers ojm065 and ojm072) for 60 s, and 72°C for 210 s.

Construction of mutants. To mutagenize the R. etli rpoN gene, the 1.8-kb EcoRI fragment of pFAJ1150 was blunt-end ligated into the SmaI site of pJQ200-UC1. The resulting plasmid was digested with PstI. This plasmid was used in two separate reactions. First, it was blunt-end ligated to the 1.8-kb BamHI fragment from pHP45-Km, resulting in two constructs in which the -Km fragment is inserted in opposite directions. The -Km interposon contains the aphII gene from Tn5 and confers resistance to kanamycin and neomycin. Second, to construct a transcriptional rpoN-gusA fusion, it was also blunt-end ligated to the 3.8-kb BamHI fragment from pWM6. Two plasmids carrying the gusA-Km^r cassette in opposite orientations were obtained. These four insertional mutations (-Km and gusA) were finally recombined into the R. etli CNPAF512 chromosome. Insertion of these mutations was verified by Southern blot hybridization with the appropriate probes. In this way, the following mutants were constructed: FAJ1154 (orientations of -Km and rpoN are opposed), FAJ1155 (same orientations), FAJ1156 (orientations of gusA and rpoN are the same), and FAJ1157 (opposite orientations).

Nucleotide sequence accession number. The nucleotide sequence of the 5,600-bp DNA fragment containing R. etli rpoN and associated genes has been deposited with DDBJ-EMBL-GenBank under accession no. U23471.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Cloning of the R. etli rpoN gene. To detect the rpoN gene, EcoRI-restricted genomic DNA from R. etli CNPAF512 was hybridized with, as a probe, the 1.7-kb EcoRI-HindIII fragment from pRJ7694 carrying the B. japonicum rpoN1 gene (23). One EcoRI fragment of 1.8 kb strongly hybridized to the probe. Also, only one hybridizing fragment was observed on HindIII- and SalI-restricted genomic DNA. The same probe was therefore used to screen a gene library of R. etli CNPAF512 maintained in E. coli HB101. This library was constructed in the EcoRI site of pLAFR1. One hybridizing cosmid, pFAJ1150, was identified. When pFAJ1150 was digested with EcoRI and hybridized with a 3.5-kb EcoRI fragment containing Agrobacterium tumefaciens rpoN and flanking DNA (53), three fragments of 1.8, 4.6, and 6.1 kb hybridized strongly. A physical map of the R. etli rpoN region is shown in Fig. 1.

Sequence analysis of the rpoN region. To further characterize the R. etli rpoN gene, the nucleotide sequence of a 5,600-bp DNA fragment was determined. Examination of this nucleotide sequence revealed the presence of one partial and four complete (ORFs) (Fig. 1). All of the ORFs are transcribed in the same direction as rpoN. The initiation codons were assigned on the basis of sequence homology with homologous ORFs in other bacterial species.

Expression of R. etli rpoN. To monitor expression of R. etli rpoN, an rpoN-gusA (gusA in both orientations) fusion was inserted into the chromosome by site-directed mutagenesis, thereby inactivating the rpoN gene. The resulting strains are FAJ1156 (correct orientation of gusA) and FAJ1157 (opposite orientation). R. etli FAJ1156 was also complemented by using pFAJ1150. FAJ1156(pFAJ1150) did not differ from the wild type with respect to growth on nitrate or succinate, melanin production, and symbiotic properties (data not shown). Expression of rpoN was assayed in different media (Table 2). No difference between the expression levels in these media was observed, except that in FAJ1156 the expression level of rpoN was slightly lower when the strain was cultured in TY than when it was cultured in AMS medium containing ammonia and succinate. Also, expression was unaffected by the ammonia concentration (from 1 to 10 mM) (data not shown). However, under all free-living conditions tested, expression of rpoN-gusA was higher in FAJ1156 than in the complemented strain (Table 2). No -glucuronidase activity was measured in the control strain FAJ1157. From these data it appears that the rpoN gene is expressed independently of the nitrogen concentration and is negatively autoregulated.

Growth of R. etli rpoN mutants. To investigate the function of rpoN, the -Km interposon was inserted in both orientations in the R. etli rpoN gene. These mutations were recombined into the chromosome by site-specific mutagenesis, yielding mutants FAJ1154 and FAJ1155 (Fig. 1 and Table 1). Insertion of the interposon at the correct site was confirmed by Southern hybridization. The rpoN mutants were tested for growth on defined media containing amino acids, ammonium, or KNO₃ as a nitrogen source and mannitol or succinate as a carbon source.

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Effects of malate and succinate concentrations on the growth curves of the R. etli wild-type strain CNPAF512 (A) and rpoN (B), ptsN (C), and ptsA (D) mutants. Cultures were grown in AMS medium containing NH4Cl (20 mM) as a nitrogen source and mannitol (20 mM) (), succinate (10 mM [] or 20 mM []), or malate (10 mM [] or 20 mM []) as a carbon source. OD595, OD at 595 nm.

Growth of R. etli ptsN mutants. Insertion mutants of ptsN (FAJ1164 and FAJ1165) were constructed as detailed in Materials and Methods. The insertion of the -Km interposon in both mutants was controlled by Southern hybridization with the appropriate probes.

Production of melanin and expression of pnifH-gusA in R. etli rpoN and ptsN mutants. For R. etli CNPAF512, the production of the black pigment melanin was previously demonstrated to depend on the nifA gene (34). In addition, when reconstituted in E. coli, this phenotype was shown to be RpoN dependent (17). We therefore tested melanin synthesis in the rpoN mutant. These experiments were also performed with the ptsN mutant, since this gene was previously demonstrated to affect the expression of RpoN-dependent genes (30). Melanin production was first assayed on TY plates containing tyrosine and CuSO₄ (17). As a control, we used the nifA mutant strain Rp1000, which is Mel (34). Both rpoN mutant strains were unable to produce the black pigment melanin, while no difference between the wild-type strain and ptsN mutants was observed. However, since minor differences in the production of melanin cannot readily be detected by this method, melanin synthesis was also determined quantitatively on media containing different carbon sources (Fig. 3A and B). These quantified data clearly show a reduction of melanin synthesis in all mutants tested. Pigment production was abolished in the rpoN mutants and reduced two- to threefold in the ptsN mutant, depending on the composition of the growth medium.

View larger version (30K): [in this window] [in a new window]   FIG. 3.   Melanin production (A and B) and expression of nifH (C and D) in wild-type R. etli (wt) and rpoN (FAJ1154), ptsN (FAJ1165), and ptsA (FAJ1166) mutants. All data are the means from four independent replicates. Error bars denote the standard deviations. Precultures were grown overnight in TY medium at 30°C, diluted 20-fold in the different media, and incubated overnight with 0.5% oxygen (34). The nitrogen source used was alanine (20 mM). The carbon sources were mannitol (black bars), succinate (stippled bars), and malate (white bars) at 5 mM (A and C) and 20 mM (B and D). To quantify melanin production, cultures were lysed at 37°C in the presence of a solution containing sodium dodecyl sulfate (1%), CuSO4 (10 µg/ml), and tyrosine (30 µg/ml). The OD of the culture after lysis was measured at 340 nm in a microplate reader after 60 and 120 min of incubation. The difference between the ODs at 340 nm was used to calculate the units. Units are expressed as the ratio of the change in OD at 340 nm to the OD at 595 nm. -Glucuronidase activities of the translational pnifH-gusA fusion plasmid pFAJ21 are expressed as Miller units.

Analysis of the R. etli Mel strain FAJ1166. In an attempt to isolate regulatory mutants affected in the production of melanin, a library of Tn5-mob mutants of R. etli CNPAF512 was previously screened on plates. Several mutants with a reduced level of melanin production were isolated. One of these mutants, FAJ1166 (which has the lowest level of melanin production among the isolated mutants), was further characterized. By the quantitative assay, FAJ1166 was shown to have a two- to fivefold reduction in melanin synthesis (Fig. 3A and B). Also, expression of pFAJ21 was reduced approximately twofold (Fig. 3C and D). These phenotypes were not caused by reduced expression of the nifA gene as demonstrated by the analysis of a pnifA-gusA fusion plasmid (data not shown). Also, under the same growth conditions, expression of pFAJ302 was not reduced in this mutant (data not shown). To further analyze this mutant, chromosomal DNA was partially digested with EcoRI and ligated to the cosmid vector pSUP205 (47). The ligation mixture was packaged into phage heads and used to infect E. coli HB101 cells. Cosmids containing the Tn5 insert were selected on kanamycin. Next, a fragment of approximately 14 kb carrying the Tn5 transposon was subcloned into pUC18. By using a primer derived from the sequence of the Tn5 inverted repeat (5'-GGTTCCGTTCAGGACGCT-3'), the DNA region of one site of the Tn5-flanking DNA was sequenced. This sequence coded for a peptide with homology to E. coli enzyme I of the PTS system (Fig. 4). Further analysis indicated that growth of mutant FAJ1166 was considerably impaired on all solid media tested, although the effect was strongest on medium containing succinate (Tables 3 and 4). However, growth of this mutant did not depend solely on the carbon source used. Higher growth rates were observed when serine instead of alanine, glutamine, or ammonia was used in combination with mannitol or glucose. Finally, in contrast to the wild type or the ptsN mutants, FAJ1166 had lost its mucoid morphology on all media tested.

View larger version (36K): [in this window] [in a new window]   FIG. 4.   DNA sequence flanking the Tn5-mob insertion of R. etli FAJ1166. The deduced amino acid sequence (RE) is compared to that of the E. coli (EC) phosphoenolpyruvate-protein phosphotransferase PtsA (enzyme I; accession no. P32670) from amino acid 111 to 226. The conserved phosphorylation site (H189) is underlined. Identical and similar (S-T-A, L-V-I-M, K-R, Q-N, and F-Y-W) residues are indicated below the amino acid sequences (cons). Stars denote similar residues.

Growth on TCA cycle acids. As observed on solid medium, growth of ptsN and ptsA mutants is impaired in the presence of succinate (Table 4). Therefore growth of both mutants was also tested in liquid AMS medium containing mannitol or the C₄-dicarboxylate succinate, fumarate, or malate as a carbon source and NH₄Cl, alanine, or glutamine as a nitrogen source. Growth in the presence of NH₄Cl is presented in Fig. 2. Growth of the ptsN and ptsA mutants was inhibited on malate and succinate. Inhibition was shown to depend on the concentration of the dicarboxylate. Complete growth inhibition of the ptsN and ptsA mutants, but not the wild-type strain, occurred in the presence of 20 mM malate (Fig. 2) or 30 mM succinate but not 30 mM fumarate (data not shown). Mannitol at concentrations of up to 30 mM or malate and succinate at a concentration of 5 mM did not interfere with growth (data not shown). Growth inhibition of these mutants decreased upon lowering of the concentration of malate or succinate. These observations were independent of the nitrogen source used.

View larger version (22K): [in this window] [in a new window]   FIG. 5.   Toxic effects of malate on cell growth in the presence of mannitol and succinate. The strains tested are the R. etli wild-type strain CNPAF512 (A) and rpoN (B), ptsN (C), and ptsA (D) mutants. Cultures were grown in AMS medium containing alanine (20 mM) as a nitrogen source and mannitol (20 mM) () or succinate (10 mM) () alone or in combination with 20 mM malate (, mannitol and malate; , succinate and malate). OD595, OD at 595 nm.

View larger version (20K): [in this window] [in a new window]   FIG. 6.   Influence of pH on growth of R. etli wild-type strain CNPAF512 (A) and rpoN (B), ptsN (C), and ptsA (D) mutants. Alanine (20 mM) was used as a nitrogen source. Malate was at 20 mM. Symbols: , pH 5.5; , pH 6.0; , pH 6.5; , pH 7.0. OD595, OD at 595 nm.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The rpoN locus of R. etli was isolated by using the B. japonicum rpoN1 gene as a probe. The nucleotide sequence of this region was determined, and four complete ORFs (ORF258, rpoN, ORF191, and ptsN) were identified. In most organisms analyzed so far, the organization of ORFs linked to the rpoN gene is remarkably well conserved (31). Exceptions are Rhodobacter capsulatus and Rhodobacter sphaeroides, in which rpoN is cotranscribed with nitrogen fixation genes (27, 28). The products of the conserved ORFs located upstream from rpoN genes, including R. etli ORF258, share features with the family of the ATP-binding cassette-type permeases. However, unlike those in other bacterial species, the two conserved downstream ORFs in R. etli (ORF191 and ptsN) are not adjacent to the rpoN gene but are separated by approximately 1.6 kb. Also, in C. crescentus homologs of these two genes are located 0.8 kb downstream from rpoN (19). The ptsN gene is homologous to genes coding for the enzyme IIA protein of the PTS (42). In bacteria, the PTS facilitates the uptake of many sugars (44) (see below). In E. coli and K. pneumoniae, sequence analysis of the DNA region downstream from ptsN has revealed the presence of two additional ORFs: ORF284 (also partially sequenced in Pseudomonas putida and P. aeruginosa [20]) and ORF90 (21, 32, 39). The product of the latter gene, NPr, is homologous to Hpr-like proteins of the PTS. No homologs of these two genes were detected in the DNA region downstream from R. etli ptsN. In addition, a rho-independent termination sequence was found 19 bp downstream from ptsN, suggesting that transcription did not extend further downstream. For E. coli, it was suggested that the NPr protein modulates the phosphorylation status of PtsN and as a result determines the activity of the protein (39) (see below). It is therefore possible that mechanisms of fine tuning of PtsN activity in R. etli are different from those in E. coli.

Expression of rpoN genes in E. coli and K. pneumoniae is at a low constitutive level (7, 29). In contrast, expression of the R. etli rpoN gene is increased in an rpoN mutant background. A similar situation exists in B. japonicum (23) and P. putida (22), where the rpoN gene (rpoN2 of B. japonicum) is also negatively autoregulated. Inspection of the R. etli rpoN promoter region revealed the presence of an RpoN consensus binding site, reading in the leftward direction (on the template strand). On the basis of sequence conservation between the R. meliloti and R. etli rpoN promoter regions (the R. etli rpoN gene can also functionally substitute for R. meliloti rpoN), this RpoN-binding site was shown to overlap the 10 promoter region and the transcription start site, as determined for R. meliloti (Fig. 7). Moreover, we observed that the promoter sequences and the oppositely oriented RpoN-binding sites were also conserved in the promoter regions of Rhizobium sp. strain NGR234 rpoN (one nucleotide missing) and B. japonicum rpoN2 (Fig. 7). We could not detect such a binding site in the 5' region of the E. coli and K. pneumoniae rpoN genes, which are not autoregulated but are transcribed at a low constitutive level (7, 29). In P. putida, a putative RpoN-binding sequence also was identified three nucleotides downstream from the 10 region (22). However, this sequence was located on the noncoding strand. Also, in the Acinetobacter calcoaceticus rpoN gene, which is negatively autoregulated, a 24/12 promoter consensus sequence is found on the noncoding strand near the transcriptional start site (13). The RpoN protein is able to bind in vitro to 24/12-type promoters in the absence of the RNA polymerase core enzyme or of an activator protein (6, 11). It is therefore possible that negative autoregulation of R. etli and B. japonicum rpoN genes occurs by direct interference with E⁷⁰ functioning through the binding of ⁵⁴ (or E⁵⁴) to the 10 to +1 promoter region of the rpoN gene, thereby inhibiting transcription initiation. Negative regulation of ⁷⁰ promoters by binding of the repressor to the 10 to +1 region of the promoter is well documented for E. coli (16).

View larger version (16K): [in this window] [in a new window]   FIG. 7.   Alignment of promoter regions from rhizobial rpoN genes. The aligned rpoN sequences are from R. meliloti (Rm), R. etli (Re), Rhizobium sp. strain NGR234, and B. japonicum rpoN2 (BjrpoN2). The 10 and 35 regions from the R. meliloli rpoN promoter are overlined, and the transcription start site is indicated by an arrow (43). The inverse/complement of the consensus (Cons) sequence of a 24/12 promoter is indicated; conserved GC and CC residues are in boldface. Nucleotides identical to the consensus 24/12 sequence are underlined. Ec, E. coli.

R. etli rpoN mutants were constructed by insertion of the -Km interposon in both directions in the gene. These mutations are polar and may thus affect the functioning of the genes located downstream from and belonging to the same operon as rpoN. However, the phenotypes of an R. etli strain carrying a mutation immediately downstream from the rpoN gene (data not shown) and of the ptsN mutants were different from that of the rpoN mutants. In addition, insertion of a gusA-aphII cassette into the rpoN gene, giving rise to nonpolar mutations (33), produced the same phenotypic defects as in the interposon mutants. Therefore, the phenotype of the rpoN mutants is not caused by a polar effect on downstream genes but can be attributed to inactivation of rpoN.

Growth defects of the R. etli rpoN mutants on different C and N sources were observed. When growth was tested on plates containing different amino acids as carbon and/or nitrogen sources, growth of the mutant, but not that of the wild-type strain, was inhibited on plates containing either serine or alanine as the sole C and N source. Growth defects may be attributed to the inability of the rpoN mutant strain to catabolize serine and alanine or to assimilate its degradation products. In E. coli, the metabolism of serine and that of alanine involve different genes (41). Both amino acids are degraded to pyruvate and ammonium. The ammonia produced is used to synthesize glutamate and glutamine. The corresponding R. etli genes may also be subject to an RpoN-type regulation. However, serine and alanine also inhibit growth of the R. etli rpoN mutants on medium containing ammonium and mannitol. The supplementary addition of glutamine to these media relieved growth inhibition. In E. coli, alanine and serine inhibit the activity of glutamine synthetase. A similar inhibitory action on the glutamine synthetase enzyme in R. etli might explain the observed phenotypes. The analysis of ammonia assimilation in R. etli is complicated by the presence of three different glutamine synthetase genes, glnA (coding for GSI), glnII (GSII), and glnT (GSIII). Rhizobium GSI enzymes are similar to the GSI of E. coli and might therefore also be inhibited by alanine and serine. GSII is similar to eukaryotic glutamine synthetases. In R. etli, transcription of glnII is RpoN dependent (36), while that of the glnA gene is not (8). Possibly, the wild-type strain synthesizes GSI and/or GSII, while the rpoN mutant produces only GSI. The observed phenotypes may therefore be attributed to the inability of the rpoN mutant to assimilate ammonia produced by the degradation of alanine and serine. A defect in the assimilation of ammonia by the rpoN mutant strain was also observed in liquid medium containing ammonium as the sole nitrogen source. The presence of a second glutamine synthetase, encoded by glnII, may thus confer a selective advantage under those conditions where GSI activity is inhibited. It is not known whether Rhizobium is subject to such conditions during its life cycle.

Growth of the rpoN mutants on succinate, fumarate, and malate was delayed compared to that of the parental strain. Therefore, in R. etli, an RpoN-dependent system for the uptake of C₄-dicarboxylates is likely to exist. However, after an extended lag phase, the mutants grew as fast as the wild type, indicating that in the absence of RpoN, the activation of alternative control mechanisms of the RpoN-dependent C₄-dicarboxylate transport (dct) system or the existence of a second, inducible, dct system may account for the observed growth kinetics. This is in contrast to the case for R. meliloti, Rhizobium sp. strain NGR234, and A. caulinodans (43, 48, 52). rpoN mutants of these bacteria cannot grow on succinate as the sole carbon source. Alternatively, a B. japonicum rpoN double mutant is not impaired in the assimilation of dicarboxylates. For this species, the existence of an RpoN-independent dicarboxylate uptake system has been suggested (23).

In R. etli, interposon mutagenesis of ptsN results in reduced expression of pnifH and decreased production of the pigment melanin. Due to the presence of a putative rho-independent terminator, the ptsN gene is probably not cotranscribed with downstream genes, and the observed phenotype is likely due to its inactivation. Growth of the ptsN mutant in media containing mannitol, glucose, and dicarboxylic acids in combination with different nitrogen sources (alanine, serine, glutamine, and ammonium) was tested. Although growth inhibition of the mutant on plates containing alanine and serine in the presence of mannitol and glucose was observed, the strongest inhibition resulted from the presence of the C₄-dicarboxylates. Also, in liquid medium, growth of the ptsN mutant was impeded in the presence of malate or succinate but not fumarate or mannitol, independently of the nitrogen source (alanine, ammonia, or glutamine) used. This growth inhibition was dependent on the concentration of the dicarboxylic acid. Malate was inhibitory at lower concentrations than succinate. Also, a clear effect of pH on the toxicity of malate was observed; toxicity was greatly reduced upon lowering of the pH of the growth medium. Finally, the inhibitory activity of malate is dominant over the assimilation of additional carbon sources that are normally metabolized (mannitol or low concentrations of succinate). These observations indicate that malate inhibits an essential cellular function or metabolic activity.

Independently of the analysis of the rpoN locus, we identified a Tn5-induced R. etli mutant with a reduced production of the black pigment melanin. Partial sequence analysis indicated that the mutated gene corresponds to a homolog of the E. coli gene coding for enzyme I of the PTS. Interestingly, this mutant also displayed reduced expression of pnifH. In addition, growth of this mutant was inhibited in the presence of malate or succinate, similar to what was observed with the ptsN mutant. Presently, we cannot exclude the possibility that the phenotype of FAJ1166 may be due to polarity of the Tn5 transposon on a thus-far-unidentified downstream gene.

From these observations, two conclusions can be drawn. First, the rpoN, ptsN, and ptsA mutants display several similar phenotypes. The expression of nifH or the production of melanin is abolished in the rpoN mutant and reduced in the ptsN and ptsA mutant strains. Also, depending on the carbon source used, growth of the ptsN mutant is impaired in the presence of alanine and serine, as observed in the rpoN mutant. These data indicate that PtsN and PtsA may act as coregulators of RpoN-dependent genes. However, no effect of these mutations on the activation of the RpoN-dependent amtB promoter was observed, indicating that not all of the RpoN-regulated genes are also controlled by PtsN and PtsA. Analysis of the conserved ORFs located downstream from rpoN genes has been carried out with only a few bacterial species. In K. pneumoniae, mutations in the two ORFs immediately downstream from rpoN (ORF95 and ptsN) increase the expression from the ⁵⁴-dependent pnifH, pnifL, and pglnAp2 (30). However, inactivation of the fourth gene (encoding an HPr-like product) results in decreased expression from these same promoters (32). In P. aeruginosa, ptsN positively controls genes involved in the assimilation of glutamine and does not influence pilin or flagellin genes (20). In E. coli, induction of the glnAp2 promoter is not dependent on ptsN, although ptsN is required for maximal growth on alanine or adenosine as the sole nitrogen source. In addition, ptsN facilitates the use of these compounds as organic nitrogen sources in the presence of additional carbon sources, especially TCA cycle intermediates. Finally, ptsN suppresses mutations in the GTPase Era (39). In C. crescentus, ptsN is involved in fine tuning the expression from the fljK promoter but not from those of other ⁵⁴-dependent flagellar genes (19). Clearly, the phenotypes of mutations in the rpoN-linked genes largely differ between different species and between the experimental systems used. Merrick and Coppard (30) proposed that the ORF95 and PtsN proteins control the stability of either the E⁵⁴ complex or the E⁵⁴-DNA complex. Alternatively, it was recently suggested that these downstream ORFs may also control the level of reinitiation of transcription by affecting the rate of release or conformational change of the ⁵⁴ protein bound at the 24/12 promoter after transcription has initiated (50). In both models, since the primary sequence of the promoter affects the stability of ⁵⁴ binding, this sequence may also account for the differential effects of mutations in the downstream ORFs on the expression of these genes.

Second, it is striking that both the rpoN mutant and the ptsN (or ptsA) mutant have growth defects on C₄-dicarboxylates. However, these defects clearly result from two different mechanisms. The ptsN and ptsA mutant strains have wild-type growth on low concentrations of malate and succinate, while these concentrations are strongly inhibitory for the rpoN mutant. In addition, increasing concentrations of these dicarboxylates stimulate growth of the rpoN mutant strain (Fig. 2 and 5) but repress growth of the ptsN and ptsA mutants. The defects of the rpoN mutant are probably due to the absence of a high-affinity dct system (see above). Growth inhibition of the ptsN and ptsA mutants is relieved by low pH. Assuming that the deprotonation of C₄-dicarboxylates increases their permease-mediated uptake, it is likely that at pH 7.0, the toxic effects of both dicarboxylates are caused by their increased cellular concentrations. Possibly, the input of larger amounts of dicarboxylates can affect the TCA cycle activities, resulting in growth inhibition. In R. etli, the catabolism of glucose is arrested in the presence of dicarboxylates (25). Therefore, a sensing system for dicarboxylates is likely to be operative. This system might signal to the cell the concentration of C₄-dicarboxylates in the growth medium. Depending on the concentration of these compounds, cells may, for example, have to decrease their uptake. Given the well-described role of the PTS in signalling in other microorganisms and since toxicity of malate and succinate is strongly increased in ptsN and ptsA mutants, this system might involve ptsN and ptsA. In other microorganisms, the PTS consists of two general proteins, enzyme I and Hpr, and the sugar-specific permease enzyme II. Enzyme II comprises up to four different polypeptides or domains (IIA, IIB, IIC, and IID). A phosphoryl group is sequentially transferred from phosphoenolpyruvate to enzyme I, Hpr, enzyme II, and, concomitant with its uptake, the sugar (44). PtsN, which is homologous to the enzyme IIA subunit, from E. coli is phosphorylated in vitro by the standard PTS phosphorelay involving phosphoenolpyruvate, enzyme I, and Hpr (39). Similar results were obtained with the K. pneumoniae ORF162 product PtsN (4, 32). Since ptsA and ptsN mutants have similar phenotypes, the phosphotransfer cascade of the PTS system in R. etli possibly controls the phosphorylation of the PtsN protein, thereby modulating its activity.