INTRODUCTION

Top Abstract Introduction Materials and Methods Results and Discussion References

Lactobacilli are fastidious gram-positive bacteria with complex nutritional requirements resulting from numerous genetic lesions in metabolic pathways which often revert to prototrophy (24). Natural auxotrophies involved in carbamoyl phosphate (CP) biosynthesis were shown to be reversed in most cases by incubating lactobacilli in CO₂-enriched air (2; F. Bringel, unpublished data). Metabolically, lactobacilli are at the threshold of the transition from anaerobic to aerobic life (16); Lactobacillus plantarum grows in aerobiosis but prefers microaerobiosis or increased CO₂ concentration.

The arginine and pyrimidine biosynthetic pathways share CP as a common precursor. CP synthetase (CPS) catalyzes the synthesis of CP from bicarbonate, glutamine, and two molecules of ATP via a complex reaction mechanism that leads to several unstable intermediates. The X-ray crystal structure of CPS in Escherichia coli has revealed the location of three separate active sites connected by two molecular tunnels that run through the interior of the protein (33) and that would allow substrate channeling and subsequent protection of the reactive intermediates (reviewed in reference 22). Furthermore, CP biosynthesis in Pyrococcus abyssi also includes metabolic channeling, a process whereby the product of one enzyme is directly transferred to the next enzyme in the pathway without being released in the bulk solvent; CPS may interact with two CP-utilizing enzymes in the pyrimidine (aspartate carbamoyltransferase) and the arginine (anabolic ornithine carbamoyltransferase) biosynthetic pathways (29). The heterodimeric CPS enzyme is composed of a small subunit (GLN) which functions as a glutamine amidotransferase and a large synthetase subunit (SYN) that fulfills the other catalytic properties (for a review, see reference 4). Allosteric regulation via the carboxy-terminal part of SYN subunit by effectors which are intermediates in the pyrimidine, arginine, and purine pathways has been identified for some prokaryotic CPSs (4) but not for others (27, 38).

Prokaryotic CPSs are encoded by two genes, commonly named carA and carB, which are organized as an operon with the gene order carAB. Prokaryotes harbor a single CPS regulated by both the pyrimidines and arginine levels in the cell (E. coli); alternatively, a single CPS regulated by the pyrimidines and CP is also produced by arginine degradation (Lactobacillus leichmannii and possibly Lactococcus lactis and Enterococcus faecalis), or a set of two CPSs specifically pyrimidine or arginine regulated is produced (Bacillaceae). In L. plantarum, earlier studies suggested the presence of two CPSs, CPS-P, and CPS-A. CPS-P is part of the pyr operon regulated by transcriptional attenuation under the control of the regulator PyrR (9, 10). The existence of L. plantarum CPS-A was suspected but not proven. Partial cloning of a carA-like sequence contiguous to the arginine-repressed biosynthetic genes (3) suggested the existence of CPS-A in L. plantarum. Furthermore, Lyman et al. observed in 1947 that L. plantarum (formerly named L. arabinosus) grew in the presence of uracil if incubated in air enriched with CO₂. This CO₂-conditional uracil sensitivity was confirmed in our laboratory. These data argue for the presence of a CPS-A in L. plantarum which is not inhibited by uracil and not functional in cells grown with ordinary air unless supplemented with CO₂. In this study, we demonstrate that L. plantarum indeed harbors two functional CPSs. In contrast to other microorganisms, wild-type L. plantarum CP synthesis is dependent on higher CO₂ requirements.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results and Discussion References

Gram-positive bacteria strains and culture conditions. L. plantarum CCM-1904 was grown on MRS medium (Difco Laboratories) supplemented with arginine (500 µg/ml), uracil (100 µg/ml), and erythromycin (2.5 µg/ml) when necessary. Nutritional requirements were tested at 30°C on agar plates of defined medium DLA (3) supplemented with 50 µg of arginine or uracil per ml in aerobiosis, anaerobiosis, or CO₂-enriched air. Aerobiosis was obtained by incubation in ordinary air (P_O2, 21%; P_CO2, 0.01%) with vigorous aeration when liquid cultures were performed. Anaerobiosis was obtained in a hermetically closed jar either with GENbox anaer or GazPak CO₂ + H₂ (bioMérieux SA, Marcy l'Etoile, France), providing less than 1% residual P_O2 and approximately 20 or 4% P_CO2 respectively. CO₂-enriched aerobiosis was controlled either using a water-jacketed CH/P incubator (Forma Scientific) calibrated at 4 or 20% P_CO2 or in a hermetically closed jar with GENbox CO₂ (bioMérieux) providing 3 to 6% P_CO2.

Nucleic acid techniques. L. plantarum, B. subtilis, and E. coli DNA transformation and nucleic acid preparations were performed as previously described (3). RNA extractions were performed according to the Lactobacillus bulgaricus hot phenol protocol (17) adapted to L. plantarum by adding -mercaptoethanol (1.9 mM, final concentration) and increasing lysozyme 30-fold and sodium dodecyl sulfate 10-fold (9). Nucleic acid hybrids on Hybond positively charged nylon membranes (Amersham) were detected with a nonradioactive digoxigenin DNA labeling and detection kit using the alkaline phosphatase chemiluminescent substrate CDP-Star (Boehringer Mannheim). Routine PCR amplifications were done with the Sigma Taq DNA polymerase. For cloning purposes, higher-fidelity amplifications were obtained with the native Pfu DNA polymerase (Stratagene). DNA primers were ³²P end labeled with the T4 polynucleotide kinase (30), and primer extension was performed as previously described (13). DNA sequencing was done with a Thermo Sequenase radiolabeled terminator cycle sequencing kit (Amersham Pharmacia Biotech).

Construction of the L. plantarum CPS mutants using pGID023-derived plasmids. The 8.2-kb cloning vector pGID023 is stable in E. coli and unstable in gram-positive bacteria under nonselective conditions. Mutagenesis by allelic replacement is therefore possible in L. plantarum by two successive homologous recombinations: an insertion followed by excision (14). Erythromycin resistance was the selective marker in E. coli and in gram-positive bacteria.

(i) FB331. Complete deletion of the two genes pyrAa and pyrAb, coding for the two subunits of CPS-P and part of the pyrRBCAaAbDFE operon (10), was designed to prevent mutations in the adjacent pyr genes.

(ii) FB335. CPS-A encoded by the carAB genes was inactivated by removing the majority of CarA and 16% of the beginning of CarB. This was done to avoid deregulation of the contiguous and divergently transcribed arg cluster since the beginning of carA harbors a putative ARG box suggested to be part of the argC operator (3). Therefore, in this CPS-A mutant, a fusion protein of the first 27 amino acids (aa) of CarA and the CarB C-terminal part will be synthesized. To avoid translation of the remaining carB gene, two stop codons in each frame were added at the junctions of the deletion.

(iii) HN217. To construct an L. plantarum strain with no functional CPS, plasmid pGID023car harboring the CPS-A deletion was introduced by electroporation in FB331, a strain having the CPS-P deleted. After the two steps of homologous recombination, pGID023car integration at the car locus, and excision of the vector, erythromycin-sensitive excisants were analyzed as described for FB335. Strain HN217 was an erythromycin-sensitive excisant deleted of both CPS-A and CPS-P as demonstrated by Southern-type hybridization, PCR amplifications, and sequencing (data not shown).

Computer analysis. Database searches and sequence analyses were performed using the Genetics Computer Group package from the University of Wisconsin (6) and the advanced BLAST program (1). The ClustalX program (35) was used for protein alignments.

Nucleotide sequence accession number. The sequence data reported here have been submitted to DDB/EMBL/GenBank as an update of accession number X99978.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and Methods Results and Discussion References

Sequencing strategy without cloning. CPSs are generally encoded by two carA and carB clustered genes (see Introduction). We cloned part of carA while cloning the argF gene by complementation of a B. subtilis argF mutant since argF is 4.5 kb distant from carA (3). Our attempts to clone the L. plantarum car genes by functional complementation in a CPS- deficient E. coli (strain C600carB8) were unsuccessful (see discussion of arginine repression studies for possible explanations). Thus, PCR without cloning was chosen to complete the carA sequence and find carB (Fig. 1). We added 3,954 bp to the 7,226-bp arg cluster sequence previously published.

View larger version (13K): [in this window] [in a new window]   FIG. 1.   Sequencing strategy of the L. plantarum carAB operon. Relevant restriction sites are indicated on the top. The arrows indicate primer orientations and localizations. Primers P6 and P7 define the ends of the CPS-A deletion constructed in strains FB335 and HN217. Steps in our sequencing strategy are marked with capital letters in rectangles as follows. Region A is part of a previously sequenced 7.2-kb fragment (3). Segment B, a 1.8-kb fragment, was PCR amplified with a universal primer (M13Rev) and the P9 (5'-CAAGTACTAACGGCTACC-3') internal carA primer deduced from region A. The template was a ligation mixture of the SmaI-digested pGID023 vector (14) ligated with a complete HincII digest of the L. plantarum genomic DNA. Fragment C was PCR amplified with a universal primer (M13Rev) and the P10 primer (5'-AGATCAAGAATCCAGTCACGC-3') deduced from segment B. The template was a ligation mixture of the HindIII-digested pUC19 vector (26) ligated with a complete HindIII digest of the L. plantarum DNA. Region D was sequenced on one strand directly from a 4-kb gel-purified BclI digestion of L. plantarum DNA, PCR amplified, and subsequently sequenced on the complementary strand.

Sequence analysis. (i) GLN subunit. Divergently oriented with respect to the arg cluster, an open reading frame (ORF) of 355 aa shares more than 40% identity with the GLN small subunit of different bacterial CPSs (B. caldolyticus, B. stearothermophilus, B. subtilis, and E. coli; CPS-P of L. plantarum) and corresponds to the carA gene. Important residues were identified for E. coli GLN (Cys²⁶⁹, His³⁵³, Glu³⁵⁵, and His³¹²) by biochemical studies, site-directed mutagenesis, and X-ray crystal structure analysis (33) and found in the L. plantarum GLN subunit (Cys²⁴⁰, His²²⁶, Glu²²⁸, and His²⁸⁴).

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Prokaryote CPS-A and CPS-P comparisons of GLN subunit subdomain junctions. The framed 8-aa-long deletion (CPS-A) was found when comparing the smaller GLN subunits of the gram-positive bacteria CPS-A with the larger gram-positive CPS-P and the single E. coli CPS. Numbers on the right indicate the studied sequence positions within the different whole GLN subunits. The underlined E. coli Leu153 marks the border between the bilobal GLN subunit subdomains: the NH2-terminal and catalytic COOH-terminal subdomains (33). Underneath the sequences, three characters (*, :, and .) indicate positions which harbor a single conserved residue having fully, strongly, and weakly conserved groups, respectively, as defined by the ClustalX multiple sequence alignment program (35).

(ii) SYN subunit. The second ORF encodes 1,020 aa sharing significant identity with the SYN large subunit of several bacterial CPSs (L. plantarum CPS-P, 62%; B. caldolyticus, 59%; L. lactis, 59%; E. coli, 48%). Therefore, this ORF was named carB. The 7-nt overlap of the GLN and SYN subunits (Fig. 3) suggests stoichiometric translational coupling. E. coli CPS crystal structure studies revealed an internal tunnel mediating substrate channeling; 21 conserved residues lining this tunnel (34) were also found in L. plantarum CPS-A (data not shown). The L. plantarum SYN subunit is the smallest of the arginine-specific SYN subunits described so far. Like CPS-A in B. subtilis (1,027 aa [27]) and B. stearothermophilus (1,047 aa [38]), the L. plantarum CPS-A SYN subunit had a C-terminal truncation which was not observed in the five known gram-positive bacterial CPS-P: L. plantarum (1,058 aa), L. lactis (1,064 aa), B. caldolyticus (1,065 aa), B. subtilis (1,071 aa), and B. stearothermophilus (1,064 aa) (protein alignments not shown). Yang et al. (38) correlated the difference in size (average of 34 aa) between CPS-A and CPS-P with the presence of the pyrimidine effector binding site in the carboxy terminus of CPS-P. So, is L. plantarum CPS-A, like other gram-positive bacteria CPS-A, unable to bind allosteric ligands?

View larger version (21K): [in this window] [in a new window]   FIG. 3.   Transcriptional and translational elements of the L. plantarum carAB operon. Numbers correspond to the citrulline biosynthetic cluster sequence (accession number X99978). The +1 labels indicate transcriptional start sites, which are marked with thick arrows. Residues involved in the carA promoter are double underlined. A B. subtilis ARG box-like sequence (23) around the +1 site is indicated with a blackened box. The proposed translational initiation and stop codons are indicated with boxes for the CPS-A GLN (carA gene) and SYN (carB gene) subunits and for the adjacent divergently oriented ORF. Good putative ribosome binding sites which exhibit complementarity to the L. plantarum 3'-terminal and 5'-terminal 16S rRNA sequence (EMBL accession number M58827) are boldface and underlined, respectively. The free energy (G [36]) characterizes the base pairing between the different RNAs. Two arrows indicate the inversely repeated (IR) sequences preceding a row of T residues, which suggests a transcriptional rho-independent terminator (estimated G of 12 kcal/mol for IR pairing).

(iii) Third ORF. Another gene was found with the opposite orientation of carB and located 106 bp after the carB stop codon. This ORF encodes a 124-aa protein truncated at its N terminus that shared no similarity with known proteins in the databank. Therefore, we designated this gene usg (upstream gene of the citrulline biosynthetic cluster).

Transcription studies and arginine repression. (i) carB is cotranscribed with carA and transcribed monocistronically. mRNAs extracted from L. plantarum grown on DLA without arginine were probed with a carA fragment. A unique 4.2-kb transcript was detected (Fig. 4A), demonstrating cotranscription of the carAB genes. The carA transcription initiation site was localized using primer extension (Fig. 5), and the corresponding promoter was found (Fig. 3) by similarity with the Lactobacillus consensus sequence (35 box [TTGACA] and 10 box [TATAAT] 17 nt apart) (28). Since the carB gene is transcribed monocistronically in P. aeruginosa and in L. lactis (18, 20), independent carB transcription was assessed in L. plantarum using a specific carB probe in Northern hybridizations. Indeed, we detected a 4.2-kb transcript (Fig. 4B) but not the 3.1-kb band corresponding to the predicted carB transcript. Nonetheless, carB may be weakly transcribed. Therefore, primer extension experiments were done on mRNAs extracted from cells grown in the absence of arginine, and a transcription start site was identified (Fig. 5) with two different primers (data not shown). We concluded that carB is independently transcribed. However, in the vicinity of the transcription start site, no promoter-like sequence could be found. Moreover, as no carB transcript was revealed by Northern blots, the carB mRNA is probably not abundant. This could result from weak transcription in the tested conditions or carB mRNA unstability.

View larger version (73K): [in this window] [in a new window]   FIG. 4.   Transcription studies and arginine repression with Northern hybridizations using mRNAs extracted from L. plantarum grown on defined medium without and with arginine (tracks 1 and 2, respectively) and probed with a carA (A) or carB (B) probe. The size of the transcript was determined using the Gibco BRL 0.24- to 9.5-kb RNA ladder (track M), which was revealed by hybridization with digoxigenin-labeled  DNA.

View larger version (77K): [in this window] [in a new window]   FIG. 5.   Primer extension mapping of the 5' end of the carAB and carB mRNAs. RNAs were extracted from L. plantarum CCM1904 grown on defined medium without arginine and no agitation at 30°C. Sequencing ladders generated with the oligonucleotide (P11 [5'-GTCCGCTAAAGTTAAGTATTTG-3'] for carAB; P12 [5'-AGCAATCGAATGAATGGCATTG-3'] for carB) used for the primer extension were loaded next to the reaction mixture (track RT). On the DNA sequence of the sense strand, the first nucleotide in the transcript is indicated as +1 and the promoter 10 sequence is boxed.

(ii) Arginine repression studies. In L. plantarum, arginine inhibition of the biosynthetic ornithine transcarbamoylase activity was correlated to the presence of 11 E. coli-like ARG boxes in the intergenic carA/argC promoter region (3). Northern hybridizations with either a carA or a carB probe were assessed on mRNA extracted from L. plantarum grown in DLA with or without arginine. As a positive control, transcription of the constitutive gene ldhL (11) was visualized with a 1-kb band in all cases (data not shown). No transcripts were detected with the car probes in presence of arginine (Fig. 4), demonstrating arginine repression of the car operon. Two regions were similar to the B. subtilis ARG box consensus sequence (23). The first region (5'-CATtAAccAAAATGaAAT-3'; mismatches are in lowercase) was located on the carA initiation codon (Fig. 3); the second region is also similar to the E. coli ARG box consensus and was previously designated R9 (3). The lack of success in cloning L. plantarum carAB in E. coli may be explained by the arginine repressor binding to the ARG boxes present on the insert and not to its E. coli target. Moreover, plasmid instability may occur since the arginine repressor is an accessory protein involved in ColE1 multicopy plasmid stability (reviewed in reference 32). We conclude that L. plantarum may harbor an arginine repressor binding in the presence of its corepressor to a specific ARG box, as found for E. coli or other gram-positive bacteria, B. subtilis (23), and B. stearothermophilus (7).

Physiological analysis of the L. plantarum deletion mutants. Since CPS produce CP for arginine and pyrimidine biosynthetic pathways, CPS-deficient derivatives were tested for arginine and uracil auxotrophy. Moreover, different growth conditions were tested such as aerobiosis and anaerobiosis because oxygen limitations induce arginine catabolism in some microorganisms (12), and CP produced from arginine catabolism provides pyrimidine biosynthesis in some lactobacilli (15). Our anaerobiosis conditions implied both oxygen depletion (less than 2%) and CO₂ enrichment from 4 to 20%. Thus, to evaluate the effect of CO₂ alone, we tested growth conditions in ordinary air enriched with either 4 or 20% CO₂ and no change of oxygen concentration. We found no difference of growth between anaerobiosis and CO₂-enriched air at either 4 or 20% CO₂ (Table 1). On the contrary, incubation with less than 0.01% of CO₂ (ordinary air) had a dramatic effect on L. plantarum growth in the presence of uracil alone. Thus in our tests, the important factor to alleviate uracil sensitivity in L. plantarum was CO₂ and not oxygen.

(i) In L. plantarum, no CP-producing system other than CPS-A and CPS-P is functionally significant. To test if L. plantarum can produce CP for arginine and pyrimidine biosynthetic pathways using a system different from the two CPSs, strain HN217 (double CPS deletant) was tested for its ability to grow on defined media. Such an alternative CP-producing system may result from carbamate kinase activity. A carbamate kinase-like CPS has been identified in the prokaryote Pyrococcus furiosus (8) and shown to be enzymatically and structurally a carbamate kinase that produced rather than degraded CP in vivo (37). To assess if an anabolic carbamate kinase may be present in L. plantarum, the growth requirement of HN217 was tested. The absence of both CPSs in HN217 resulted in arginine and pyrimidine auxotrophy (Table 1). This result was confirmed in the single CPS deletants under conditions repressing the remaining CPS (Table 1; FB335 was uracil sensitive and FB331 was unable to grow in the presence of arginine alone in CO₂-enriched air). Thus, in the tested conditions, no CP-producing system other than CPS-A and CPS-P is present or functionally significant in L. plantarum.

(ii) CPS-P is the major source of CP in L. plantarum. To assess the CPS-P relative role in L. plantarum and whether CPS-P can supply CP necessary for arginine and pyrimidine biosynthesis, we tested the growth requirements of strain FB335 with only a functional CPS-P. FB335 was prototrophic for arginine and uracil (Table 1, without supplement). Thus, CPS-P alone can provide sufficient CP for the pyrimidine and arginine biosynthetic pathways. Despite its natural lability, CP can be transferred from the L. plantarum pyrimidine-specific CPS-P to the arginine biosynthetic CP-consuming enzyme, ornithine carbamyltransferase. This is reminiscent of the absence of CP pool compartmentalization found in B. subtilis, which also harbors two CPSs, and is consistent with the case of prokaryotes having a unique CPS such as E. coli but is different from the eukaryote Neurospora (27). Since unlike CPS-P deletion (see below), CPS-A deletion did not lead to auxotrophy, we concluded that CPS-P is the major source of CP in L. plantarum.

(iii) CPS-A is functional in L. plantarum grown in CO₂-enriched air and provides CP only for arginine biosynthesis. With only a functional CPS-A, FB331 was unable to grow on DLA (Table 1, without supplement). Thus, CPS-A is unable to provide CP for both the arginine and pyrimidine biosynthetic pathways. Therefore, unlike CPS-P, CPS-A is not essential for L. plantarum growth in minimal medium. FB331 grew in the presence of both arginine and uracil, demonstrating that only these corresponding biosynthetic pathways were impaired in FB331. FB331 grew in the presence of uracil only when CO₂ was added to air (similar to the wild-type strain, where CPS-P was repressed and only CPS-A was theoretically available for CP biosynthesis [Table 1, with uracil in ordinary air]). The conditional pyrimidine auxotrophy suggests that CPS-A provides enough CP for arginine biosynthesis and is active only in CO₂-enriched air. Since unlike CPS-P, CPS-A is not able to provide CP for pyrimidine biosynthesis, CPS-P may be more abundant than CPS-A, or possibly CPS-A activity remains lower than CPS-P activity even in presence of increased CO₂ concentrations.

How can CO₂ stimulate CP biosynthesis in L. plantarum? CO₂ may regulate CPS-A expression as described for various microbial genes (31). To assess if CO₂ was absolutely required for carAB transcription, Northern hybridization and primer extensions were done on mRNAs extracted from cells cultivated in air with agitation. The 4.2-kb carAB transcript was detected, and primer extension of the 5' end of the carB transcript was obtained (data not shown). These transcriptional results do not exclude CO₂ as a modulator of the car genes but strongly suggest that the main effect of CO₂ stimulation on CPS-A is not at the transcriptional level.