Extent of Genetic Lesions of the Arginine and Pyrimidine Biosynthetic Pathways in Lactobacillus plantarum, L. paraplantarum, L. pentosus, and L. casei: Prevalence of CO₂-Dependent Auxotrophs and Characterization of Deficient arg Genes in L. plantarum

Abstract

Lactic acid bacteria require rich media since, due to mutations in their biosynthetic genes, they are unable to synthesize numerous amino acids and nucleobases. Arginine biosynthesis and pyrimidine biosynthesis have a common intermediate, carbamoyl phosphate (CP), whose synthesis requires CO₂. We investigated the extent of genetic lesions in both the arginine biosynthesis and pyrimidine biosynthesis pathways in a collection of lactobacilli, including 150 strains of Lactobacillus plantarum, 32 strains of L. pentosus, 15 strains of L. paraplantarum, and 10 strains of L. casei. The distribution of prototroph and auxotroph phenotypes varied between species. All L. casei strains, no L. paraplantarum strains, two L. pentosus strains, and seven L. plantarum strains required arginine for growth. Arginine auxotrophs were more frequently found in L. plantarum isolated from milk products than in L. plantarum isolated from fermented plant products or humans; association with dairy products might favor arginine auxotrophy. In L. plantarum the argCJBDF genes were functional in most strains, and when they were inactive, only one gene was mutated in more than one-half of the arginine auxotrophs. Random mutation may have generated these auxotrophs since different arg genes were inactivated (there were single point mutations in three auxotrophs and nonrevertible genetic lesions in four auxotrophs). These data support the hypothesis that lactic acid bacteria evolve by progressively loosing unnecessary genes upon adaptation to specific habitats, with genome evolution towards cumulative DNA degeneration. Although auxotrophy for only uracil was found in one L. pentosus strain, a high CO₂ requirement (HCR) for arginine and pyrimidine was common; it was found in 74 of 207 Lactobacillus strains tested. These HCR auxotrophs may have had their CP cellular pool-related genes altered or deregulated.

Lactic acid bacteria (LAB) are gram-positive bacteria that have adapted to rich environments. As a result, in addition to sugars as energy and carbon sources these organisms require nucleobases, vitamins, cations, and amino acids (18). For example, some LAB associated with particular fermented foods have developed auxotrophies for specific growth factors, including orotic acid present in milk for Lactobacillus delbrueckii subsp. bulgaricus (32), a small peptide present in freshly prepared yeast extracts for the sourdough bacterium Lactobacillus sanfrancisco (1), and d-mevalonic acid for rice wine spoilage lactobacilli (33). Some authors (21) consider LAB to be a highly specialized form of life in view of their complex nutritional requirements and restricted habitats. The complex nutritional needs of LAB may be the result of two opposing evolutionary processes. A primitive LAB may have had restricted metabolism and gradually acquired new enzymatic activities. Alternatively, a chemoorganotrophic ancestor with many biochemical abilities may have evolved by progressively loosing unnecessary genes upon closer association with plants, animals, or humans. Research data favor the latter hypothesis. Recently, genome analysis has revealed that Lactococcus lactis has the genetic potential to synthesize most amino acids, even though growth of this organism in synthetic media requires six amino acids (2). In genetic studies, a systematic attempt to isolate mutants that no longer require each of the necessary amino acids was undertaken for different Lactobacillus (26), Enterococcus, Pediococcus, and Lactococcus species (10). When amino acid prototrophic revertants were obtained, minor genetic lesions, such as point mutations, were postulated to be present in the parental strain. When no mutants were obtained, more extensive mutations may have inactivated the amino acid biosynthetic pathways. Defective LAB genes have been shown to be inactivated by point mutations (6, 14, 16), frameshift mutations (11, 29, 35), deletions (11, 16), or mobile element insertions (20). However, most of these studies included a limited number of strains with restricted habitats. Only a few LAB species are ubiquitous in nature; L. plantarum is found in many different fermented vegetable products (silage, sauerkraut, pickles, sourdough, and beverages such as wine and beer), as well as in fermented animal foods (fish, ripened dairy products, and meats). L. plantarum is also found on plant surfaces, in mammalian cavities (mouths, intestinal tracts, stools), and in sewage. It has been found occasionally in cases of bacteremia or endocarditis in humans after surgery or in patients with reduced immunity (17). Among the LAB growing in rich specialized habitats, L. helveticus is associated with milk products, while L. acidophilus and L. johnsonii are found in mammalian digestive tracts. Genome sequencing of L. johnsonii revealed a lack of all amino acid biosynthesis pathways (19). On the other hand, since L. plantarum may be exposed to prototrophic conditions, it might not have accumulated as many genetic lesions (26). As a matter of fact, only branched-chain amino acid biosynthesis pathway-encoding genes were not present in the L. plantarum genome (19). By analyzing the genetic lesions found in naturally occurring auxotrophs of LAB related to L. plantarum, we intended to characterize the first steps in the process of genome evolution towards cumulative DNA degeneration.

We investigated the auxotrophies found in a collection of L. plantarum and L. plantarum-related strains isolated from various geographical areas and ecological niches (Table 1). We evaluated the molecular basis of Lactobacillus auxotrophy in the arginine and pyrimidine biosynthetic pathways for three reasons. First, these pathways have been characterized in many organisms and, in particular, in an L. plantarum prototroph, strain CCM 1904 (equivalent to ATCC 8014) (3, 13, 28). Second, an arginine precursor, citrulline, is always found at low levels in a wide variety of fermented foods and beverages (25). This theoretically alleviates the selective pressure to maintain the metabolic pathway responsible for citrulline synthesis. Third, the arginine and pyrimidine biosynthetic pathways have a common intermediate, carbamoyl phosphate (CP), which is required in protein and nucleic acid synthesis. In L. plantarum two CP synthetases (CPS) synthesize CP from glutamine, ATP, and bicarbonate, the dissolved form of CO₂ (Fig. 1A). CPS-A is the arginine-regulated CPS encoded by the carAB operon (28), and CPS-P is the pyrimidine-regulated CPS encoded by the pyrAaAb genes (13) present in the pyr operon regulated by transcriptional attenuation (Fig. 1B). Air enriched with 5 to 10% CO₂ enhanced surface growth of L. plantarum on solid media (18), and recently, CO₂ concentration was found to modulate the growth of L. plantarum CCM 1904 with CPS-P deleted (28). Unlike CPS-P, CPS-A provided sufficient CP only if L. plantarum was cultured in CO₂-enriched air. In the presence of uracil, CPS-P was not synthesized; growth depended on CPS-A and therefore on the CO₂ concentration to supply CP for arginine biosynthesis. Therefore, in the absence of arginine, uracil inhibited L. plantarum's growth in air but not in CO₂-enriched air. For this reason, we screened our Lactobacillus strain collection for both an arginine requirement and a pyrimidine requirement not only in air but also in CO₂-enriched air. The extent of high-CO₂-requiring (HCR) arginine and pyrimidine auxotrophs was unexpected, and such organisms accounted for more than one-third of the L. plantarum and L. pentosus population and 80% of the L. paraplantarum population. On the other hand, auxotrophy for only pyrimidine was uncommon and was found in only one L. pentosus strain. Arginine auxotrophy was found in all L. casei strains but in less than 6% of L. plantarum, L. pentosus, and L. paraplantarum strains. In this paper, we describe the gene mutations responsible for L. plantarum arginine auxotrophy. The implications of our results for LAB ecological niches and LAB evolution are also discussed below.

TABLE 1.

Strains grouped according to their nutritional needs^a

Nutritional groupb	Species	Bacterial strain(s)c	Origin
Prototrophs	L. paraplantarum	CNRZ 745	Silage
		H41, H48	Polish sauerkraut
	L. pentosus	CNRZ 1218, NCFB 1059	Cheese
		CNRZ 1537, CNRZ 1538, CNRZ 1539, CNRZ 1540, CNRZ 1542, CNRZ 1543, CNRZ 1544, CNRZ 1545, CNRZ 1559, CNRZ 1560, CNRZ 1564, CNRZ 1565, CNRZ 1569	Fermented olives
		KOG9	Pickled vegetables
	L. plantarum	54, 57.2, A1, A2, A4, A7, A9, DK38, DKO20A, DKO22, SF2A33, SF2A39, SF2A35B, SF2B37-1	Sour cassava starch fermentations
		Agrano 15b, CNRZ 424, CNRZ 432, FB101, FB108	Bread dough
		CCM 1904, CNRZ 738, CNRZ 739, LP85-2, NCIMB 6105, NCIMB 8299	Silage
		229v, CIP 102359, CIP 104439, CIP 104441, CIP 104442, CIP 104447, CIP 104448, CIP 104449, CIP 104450, CIP 104452, NCIMB 1406, NCIMB 8825	Human
		CCM 4279, CIP 104454, CNRZ 764, CNRZ 1228, CNRZ 1229, CNRZ 1246, CST 12007, CST 12008, DK32, DSM 9296, KOG24, LMAB2, NCFB 1204	Cheese and milk products
		CNRZ 184, CNRZ 1838, CNRZ 1849, CNRZ 1850, CST 10952, Hd4; Hd17, Lactolabo, LP80, NCIMB 5914, NCIMB 8016, NCIMB 8102	Unknown
		CST 10967, CST 11031	Beer
		DK19	Kenkey, Nigeria
		DK21	Fermented oil bean, Nigeria
		CNRZ 1890, DK15, DK36, DKO2A, DKO7, DKO8, DKO9, DKO12, NCIMB 12120	Fermented cereals, Nigeria
		FB115	Fermented taro, United States
		FOEB 8402, FOEB 9106, FOEB 9113, FOEB 9235	Grapes or wine
		DK9, DKO18, JCL1267, JCL1268, JCL1269, JCL1271, JCL1275, JCL1278, JCL1284, JCL1285	Fermented cucumber
		KOG7, KOG14, KOG18, NCIMB 11974T	Pickled vegetables
Uracil auxotrophs	L. pentosus	CNRZ 1547	Fermented olives, Spain
Arginine auxotrophs (ornithine)	L. plantarum	KOG5, FB400 NCFB 772, NCFB 963, NCFB 965, NCFB 2171	Pickled vegetables Cheese
Arginine auxotrophs (citrulline)	L. casei	5A, 5B2, 12CST 10927	Cow udder, Algeria Sirup
		CIP 103918T, CIP 53.166, CIP 103164	Unknown
		DSM 20020, CIP 104456	Human
		NCIMB 3254	Cheese
	L. pentosus	CNRZ 1555, CNRZ 1570	Fermented olives, Spain
	L. plantarum	CCM 3626	Cheese
HCR auxotrophs, class 1	L. paraplantarum	CNRZ 188861D	Unknown Human feces
		CNRZ 1885T, CNRZ 1887	Beer
		KOG1, KOG3, KOG16, KOG25	Pickled vegetables
	L. pentosus	CNRZ 1546, CNRZ 1552, CNRZ 1553, CNRZ 1558,     CNRZ 1561, CNRZ 1562, CNRZ 1566	Fermented olives
		CNRZ 1245 (bradytrophy for Orn)	Cheese
		Hansen	Commercial starter
		KOG17	Pickled vegetables
		NCFB 363T	Sawdust fermentation
		NCIMB 8531	Waste sulfite liquor fermentation
	L. plantarum	13, 1188, LN32, LT	Cow udder, Algeria
		A12, SF2A31B, SF2B41-1	Sour cassava starch fermentations
		CIP 102021, CIP 71.39, NCIMB 6461	Unknown
		CIP 104440, CIP 104451	Human
		CST 10928, CST 11023	Beer, France
		CST 12009, NCFB 773, NCFB 1206, NCFB 1988	Cheese and dairy products
		CNRZ 1890	Fermented millet, Nigeria
		DSM 2648	Silage
		KOG2, KOG4, KOG6, KOG8, KOG10, KOG11, KOG12,     KOG13, KOG21, KOG22, CIP 104453, LMAB1, NCIMB 7220	Pickled vegetables/PICK>
HCR auxotrophs, class 2	L. pentosus	CNRZ 1573	Fermented olives, Spain
	L. plantarum	IN5	Cow udder, Algeria
		JCL1279, JCL1280, JCL1281, JCL1282, JCL1283	Fermented cucumber, Spain
		ALAB20, 38AA	Sour cassava starch fermentation
		CNRZ 1395	Cheese
		NCIMB 8826	Human saliva
HCR auxotrophs, class 3	L. paraplantarum	H43 KOG15, KOG20	Polish sauerkraut Pickled vegetables
		NCFB 1088	Cheese
	L. plantarum	CIP 104446	Human
		CNRZ 1220, NCFB 1042	Cheese
		CST 11019	Beer
		DK30, KOG19	Pickled vegetables

^aThe strain collection included four Lactobacillus species isolated from different ecological niches. The survey was a phenotypic profile of arginine and pyrimidine prototrophy and auxotrophy.

^bNutritional needs for arginine and two of its precursors (citrulline and ornithine) were tested in combination with uracil in air and in CO₂-enriched air. Four major phenotypes were obtained; prototrophs; uracil auxotrophs; arginine auxotrophs which were rescued by either citrulline or ornithine (indicated in parentheses), and HCR auxotrophs unable to grow without both arginine and uracil in the absence of CO₂ enrichement. The HCR auxotrophs were classifed in three classes as shown in Table 2.

^cATCC, American Type Culture Collection, Rockville, Md.; CCM, Czechoslovak Collection of Microorganisms, Brno, Czech Republic; CIP, Collection of Bacterial Strains of Institut Pasteur, Paris, France; CNRZ, Centre National de Recherches Zootechniques, Jouy-en-Josas, France; CST, Collection souches Tepral des brasseries Kronenbourg, Strasbourg, France; NCFB; National Collection of Food Bacteria, Reading, United Kingdom; NCIMB, National Collection of Industrial and Marine Bacteria, Aberdeen, Scotland.

FIG. 1.

Arginine and pyrimidine biosynthesis in L. plantarum. Arginine synthesis and pyrimidine synthesis have a common intermediate, CP. CPS catalyzes the formation of CP from one molecule of glutamine, two molecules of Mg²⁺-ATP, and one molecule of bicarbonate, which is the dissolved form of CO₂. Two types of CPS are present in L. plantarum: CPS-A, the arginine-regulated CPS, and CPS-P, the pyrimidine-regulated CPS (28). (A) Summarized scheme of the arginine biosynthesis pathway (thick arrows) and the pyrimidine biosynthesis pathway (thin arrows). Ornithine and citrulline are precursors of arginine. The argCJBD genes are involved in ornithine synthesis from glutamate with the recycling of N-acetyl intermediates. The argF gene encodes OTCase responsible for citrulline synthesis from ornithine and CP. (B) Gene organization. The pyr operon (pyrRBCAaAbDFE) containing the de novo pyrimidine synthesis genes (13) is controlled by transcriptional attenuation by the regulatory protein PyrR. The genes involved in citrulline biosynthesis are organized in two divergently transcribed operons, carAB and argCJBDF, which are repressed by arginine (3). The genes associated with octagons were mutated in seven L. plantarum strains isolated from cheese or pickled vegetables.

MATERIALS AND METHODS

Bacterial strains and species identification.

The origins of the strains are shown in Table 1. On the basis of biochemical ability (5) (9) or 16S ribosomal DNA sequence (8), L. plantarum can be misclassified as L. paraplantarum or L. pentosus. The strains used in this study were identified by using a polyphasic approach (5) or by using 16S-23S ribosomal DNA spacer region PCR amplification (30). The collection included 150 strains of L. plantarum, 32 strains of L. pentosus, and 15 strains of L. paraplantarum. Ten strains, including strains CIP 103164 and CIP 104456 listed as L. plantarum strains by the Institut Pasteur collection, were classified as L. casei-related strains on the basis of PCR amplification of 23S-5S rRNA intergenic spacer regions (7; data not shown).

Bacterial propagation and physiological tests.

Routine cultivation of lactobacilli was done in MRS (Difco Laboratories) at 30°C. To test nutritional requirements, a colony grown on MRS agar was suspended and diluted in sterile physiological water to obtain around 10⁸ and 10⁷ cells/ml. A drop (around 15 μl) of each dilution was put on the surface of an agar plate of the defined rich medium DLA (3). At final concentrations of 50 μg/ml, uracil, arginine, citrulline, and ornithine were tested individually and in combination. Plates were incubated at 30°C for 3 to 5 days in ordinary air (partial pressure of CO₂ [P_CO2], <0.01%) and in 4% CO₂-enriched air obtained by incubation in a water-jacketed CH/P incubator (Forma Scientific) calibrated at a P_CO2 of 4%.

Spontaneous arginine revertants were searched for as follows. After reaching the stationary growth phase in 20 ml of liquid DLA containing arginine (50 μg/ml), cells were harvested by centrifugation, washed twice with sterile physiological water, and then suspended in 1 ml of physiological water. Both the cell suspension (100 μl) and a 1/10 dilution of the cell suspension (100 μl) were spread on DLA agar plates that were incubated at 30°C in 4% CO₂-enriched air in the absence of light for 4 to 5 days. Total viable cells were counted on MRS agar plates. The reversion rate was defined as the number of revertants growing on DLA divided by the total viable cells counted on the MRS plates. Since at least 5 × 10¹⁰ cells were spread on the revertant selection plates, the reversion rate detection threshold was less than 10⁻¹¹.

Functional complementation assays with prototrophic genes.

Plasmids pJIM407 and pJIM207 harbor the L. plantarum functional argF and argCJBDF genes, respectively, on a pAMβ1-derived plasmid (3). Plasmids pJIM407 and pJIM207 were electroporated into naturally arginine-deficient strains. Two electroporation protocols were used in our attempt to electroporate recalcitrant lactobacilli (4, 22). Erythromycin-resistant transformants were selected and tested to determine their arginine and uracil requirements on DLA plates in air and in CO₂-enriched air.

DNA preparation.

Total DNA was liberated from lysozyme-treated lactobacilli by using the detergent sodium dodecyl sulfate. DNA was separated from sodium perchlorate-precipitated cell debris and proteins by phenol-chloroform treatments as previously described (5).

PCR amplification and sequence analysis.

From the functional L. plantarum CCM 1904 argCJBDF gene sequence, which codes for arginine synthesis via ornithine (GenBank accession number X99978), primers were deduced to PCR amplify and sequence the corresponding genes in the natural auxotrophs. The nucleic acid sequences of different strains of L. plantarum exhibited at least 99% identity. The same primers did not amplify the corresponding genes in other species due to sequence divergence between different Lactobacillus species (data not shown). Primers carAP4 (5′-GTCCGCTAAAGTTAAGTATTTG-3′) and revargF (5′-AATCACGACCGTGCTCCCAGCTGCCTGCGC-3′) were used for argCJBDF amplification. The Expand Long Template PCR system (Roche Molecular Biochemicals) was used for PCR amplification of fragments that were 5 kb long or longer. After incubation for 2 min at 92°C, 10 cycles of denaturation for 10 s at 92°C, annealing for 30 s at 50°C, and elongation for 10 min at 68°C were performed. For the additional 19 cycles, the annealing temperature was increased to 54°C, and the polymerization step was extended by 20 s for each cycle. A postelongation step consisting of 68°C for 15 min completed the polymerization. PCR fragments were purified by gel filtration (Amersham Pharmacia Biotech MicroSpin columns) and sequenced (Applied Biosystems 373 DNA sequencer). The GCG package from the University of Wisconsin (12) was used for nucleotide sequence analysis.

RESULTS

Arginine and pyrimidine nutritional requirements of 207 lactobacilli.

The strain collection contained mesophilic lactobacilli belonging to the following four taxa: L. plantarum (150 strains), L. pentosus (32 strains), L. paraplantarum (15 strains), and L. casei-related strains (10 strains). The extents of genetic lesions in these strains were evaluated by testing their abilities to synthesize arginine and pyrimidine nucleotides. To localize the enzymatic steps impaired in these pathways, permeable specific substrates and biosynthetic intermediates were tested, as follows: (i) CO₂ as a substrate for CP synthesis (Fig. 1) involved in both de novo biosynthetic pathways; and (ii) citrulline and ornithine as biosynthetic precursors of arginine. Although the de novo pyrimidine biosynthesis pathway leads to UMP synthesis, UMP and its precursors (ureidosuccinic acid, dihydroorotate, orotate, and orotidine monophosphate) were not able to save L. plantarum pyr mutants (23). On the other hand, uracil restored the growth of mutants with impaired de novo pyrimidine synthesis (24). Thus, arginine and pyrimidine requirements were assessed in this strain collection with arginine or one of its precursors in combination with uracil in air and in CO₂-enriched air. Prototrophs (113 strains) grew in the absence of arginine and uracil and in all conditions tested except in the presence of uracil in normal air (Table 2). Auxotrophs represented around 45% of the L. plantarum and L. pentosus strains, 80% of the L. paraplantarum strains, and 100% of the L. casei strains. Auxotrophs were classified in three nutritional groups: uracil auxotroph (strain CNRZ 1547), arginine auxotrophs (19 strains), and HCR auxotrophs (74 strains). A nutritional requirement for the pyrimidine alone was rare. Among the 207 lactobacilli tested, only L. pentosus CNRZ 1547 had the phenotype of a uracil auxotroph. CNRZ 1547 spontaneous revertants obtained on DLA in the presence of arginine at frequencies of 10⁻⁷ had the phenotype of spontaneous derivatives obtained with FB331, a strain with CPS-P deleted (data not shown). Thus, among the genes required for de novo pyrimidine biosynthesis (13), it remains to be determined whether CNRZ 1547 harbors a revertible mutation within its CPS-P-encoding gene.

TABLE 2.

Arginine and pyrimidine nutritional requirements in normal and CO₂-enriched air

Phenotype	Group	No. of strains				Concn of CO2 in air (%)	Growth in DLA supplemented witha:
		L. plantarum (n = 150)	L. pentosus (n = 32)	L. paraplantarum (n = 15)	L. casei (n = 10)		Nothing	Ura + Arg or Ura + Cit	Arg or Cit	Ura	Orn
Prototrophs		94	16	3	0	<0.01	+	++	++	−	+
						4	++	++	++	++	++
Auxotrophs for uracil		0	1	0	0	<0.01	−	++	−	−	−
						4	−	++	−	++	−
Auxotrophs for Arginine	Ornithine auxotrophs	6	0	0	0	<0.01	−	+	—b	−	−
						4	−	++	++	−	+
	Citrulline auxotrophs	1	2	0	10	<0.01	−	+	—c	−	−
						4	−	++	++	−	−
HCR auxotrophs for arginine and uracil	Class 1	33	12	8	0	<0.01	−	++	−	−	−
						4	++	++	++	++	++
	Class 2	10	1	0	0	<0.01	−	++	+d	−	−
						4	++	++	++	−	++
	Class 3	6	0	4	0	<0.01	−	++	+	−	−
						4	++	++	++	++	++

^aGrowth was tested on defined DLA agar plates supplemented with combinations of uracil, arginine, citrulline, and ornithine (Ura, Arg, Cit, and Orn, respectively). −, no growth; +, good growth; ++, very good growth. c, No growth for CST 10927.

^bNCFB 963, NCFB 965, NCFB 2171, and FB400 grew, but KOG5 and NCFB 772 did not grow.

^cCST 10927 did not grow. CCM 3626 grew after 3 days. Other strains were bradytrophs (they grew very slowly in the absence of arginine or citrulline), so three additional days of incubation was necessary to detect growth.

^dNCIMB 8826, ALAB20, IN5, CNRZ 1573, CNRZ 1395, and JCL1279 did not grow in the presence of arginine in ordinary air.

The distribution of prototrophic and auxotrophic groups was different in different species (Table 2). All L. casei strains were citrulline auxotrophs. Citrulline auxotrophy was rare in the other species; it was found only in L. plantarum CCM 3626 and L. pentosus CNRZ 1555 and CNRZ 1570. Ornithine auxotrophy was observed in 4% of L. plantarum strains; for the other species, the sample sizes might have been too small to detect this auxotrophy.

The most striking observation was that most auxotrophs had HCR phenotypes. The HCR auxotrophs did not grow in DLA (in the absence of both arginine and uracil) in normal air unless CO₂ was provided. All the L. paraplantarum auxotrophs had this phenotype. HCR auxotrophs were divided into three phenotypic classes, which were designated classes 1 to 3 (Table 2). Most HCR auxotrophs were unable to grow in the presence of arginine in normal air (Table 2, class 1). Some of the other HCR auxotrophs able to grow in the presence of arginine (Table 2, classes 2 and 3) were inhibited by uracil (class 2).

Analysis of the nutritional needs with respect to strain origin.

The sources of the strains were analyzed to establish possible correlations with the nutritional requirements.

For analysis of natural arginine auxotrophs, the following three source criteria were used: fermented plant products (89 strains), dairy products (25 strains), and human isolates (16 strains). For 20 strains, either the origin was unknown or the source classification was unclear (cow udder, for instance). Among the 150 L. plantarum tested, 7 arginine auxotrophs were found (Table 1). Five arginine auxotrophs originated from cheese; three of these, the three strains isolated from New Zealand cheese, had the same sugar fermentation profiles (as determined with API 50 CH strips with API 50CHL medium [BioMérieux]) (data not shown) and had identical argCJBD nucleic acid sequences (Table 2). These similarities suggest that NCFB 963, NCFB 965, and NCFB 2171 were isolates of a single strain. If these three strains were considered a single isolate, only three different arginine auxotrophs were detected among 23 strains isolated from dairy products. Two auxotrophs were isolated from pickled vegetables. No arginine auxotrophs were detected in L. plantarum of human origin. In summary, for L. plantarum the percentages of arginine auxotrophs isolated from dairy products (3 of 23 strains), from plant products (2 of 89 strains), and from humans (0 of 16 strains) were 13, 2, and 0% of the total population, respectively. Thus, our data suggest that L. plantarum strains associated with milk products were more likely to lose the ability to synthesize arginine than L. plantarum strains isolated from plant products or humans were.

Since most natural auxotrophs required high levels of CO₂ for arginine and pyrimidine synthesis, the lactobacilli isolated from P_CO2-enriched environments, such as bread sourdough and beer, were analyzed in view of their nutritional phenotypes. All five bread sourdough isolates were prototrophs (Table 1). Of the seven beer isolates, two L. plantarum strains were prototrophs and the others were HCR auxotrophs (three L. plantarum strains and two L. paraplantarum strains). Therefore, the presence of high CO₂ levels in the source of isolation of the strains does not necessary lead to CO₂ dependence since in bread sourdough all of the isolates were prototrophs. On the other hand, in beer, five of the seven isolates tested had evolved towards CO₂ dependence.

Rates of reversion to arginine prototrophy in CO₂-enriched air.

If arginine auxotrophs harbor minor lesions in one of the arg genes (argCJBDF in Fig. 1), then spontaneous revertants able to grow on defined DLA without arginine may be selected. However, some arginine auxotrophs were unable to grow in the presence of arginine in normal air unless CO₂-enriched air was provided (Table 2). Since this HCR phenotype may have resulted from other genetic lesions not linked to the arg cluster, we searched for arginine prototroph derivatives under CO₂-enriched conditions. The reversion test was carried out with 17 arginine auxotrophs of three Lactobacillus species, L. plantarum (7 strains), L. pentosus (1 strain), and L. casei (9 strains), which included all of the arginine auxotrophs except L. casei 5A and L. pentosus CNRZ 1570 (Table 1).

Reversion rates lower than 10⁻¹¹ were not detected (see Materials and Methods). With this detection threshold, only L. plantarum revertants were obtained. Thus, four of seven L. plantarum strains (KOG5, NCFB 2171, NCFB 965, and NCFB 963), as well as the arginine auxotrophs of L. casei and L. pentosus tested, harbored more than one revertible point mutation in one of the arginine biosynthetic genes.

With L. plantarum NCFB 772, CCM 3626, and FB400, revertant frequencies ranging from 10⁻⁸ to 10⁻⁹ were obtained. The following independent revertants were selected: R-NCFB772 from NCFB 772; R-CCM3626-A, R-CCM3626-K, and R-CCM3626-P from CCM 3626; and R-FB400-D7, R-FB400-B4, R-FB400-A1, and R-FB400-B3 from FB400. These L. plantarum revertants were used to localize minor mutations by comparison of the arg cluster sequences of the revertants and the parental arginine auxotrophs (Table 3). Furthermore, CO₂-dependent growth was tested in the revertants to evaluate if additional mutations conferring the HCR phenotype were present.

TABLE 3.

Comparison of arginine cluster sequences in arginine-deficient L. plantarum natural auxotrophs and in arginine prototroph spontaneous derivatives^a

Protein	No. of amino acids	Major mutations in strain(s)b:			Minor mutations in strain(s)c:
		NCFB 2171, NCFB 965, and NCFB 963	KOG5	NCFB 772	R-NCFB772	FB400, R-FB400-D7, and R-FB400-B4	R-FB400-A1	CCM 3626	R-CCM3626-A
ArgC	341	M14-P341del (deletion; 4915-4997)d	A234V (GCA → GTA; 5573)	A234V (GCA → GTA; 5573), G333E (GGA → GAA; 5870)e	A234V (GCA → GTA; 5573), E333G (GAA → GGA; 5870)f	V35M (GTG → ATG; 4976)	V35M (GTG → ATG; 4976)	NSg	NS
ArgJ	404	A179V (GCC → GTC; 6458), A354V (GCC → GTC; 6983)	F16C (TTT → TGT; 5969), G305D (GGC → GAC; 6836)	Y150H (TAC → CAC; 6371)	Y150H (TAC → CAC; 6371)	T143M (ACG → ATG; 6351)	T143M (ACG → ATG; 6351)	NS	NS
ArgB	248	Q32H (CAG → CAT; 7247)	Q32H (CAG → CAT; 7247), L86P (CTT → CGT; 7409)	Q32H (CAG → CAT; 7247), T50N (ACT → AAT; 7301), A81V (GCG → GTG; 7394)	Q32H (CAG → CAT; 7247), T50N (ACT → AAT; 7301), A81V (GCG → GTG; 7394)	Q51stop (CAG → TAG; 7304)h	stop51Y (TAG → TAT; 7304)i	NS	NS
ArgD	389	K296E (AAA → GAA; 8782)	47insY (ins TAT; 8037), K296E (AAA → GAA; 8782), S301A (TCC → GCC; 8797), I330M (ATC → ATG; 8884)j	H62Y (CAT → TAT; 8080), K296E (AAA → GAA; 8782), P364S (CCG → TCG; 8986)	NS	N136T (AAT → ACT; 8302)	N136T (AAT → ACT; 8302)	NS	NS
ArgF	340	NS	NS	NS	NS	NS	NS	P18S (CCA → TCA; 9110), T257A (ACG → GCG; 9827), Q290stop (CAG → TAG; 9926)k	P18S (CCA → TCA; 9110), T257A (ACG → GCG; 9827), stop290L (TAG → TTG; 9926)l

^aThe argCJBDF reference sequence used was the sequence of the arginine prototroph strain CCM 1904 (accession number X99978). Nucleic acid differences leading to amino acid changes are underlined. Silent mutations are not shown.

^bNo spontaneous arginine prototrophic derivatives were obtained.

^cSpontaneous arginine prototrophic revertants were selected. Each revertant was designated by using R- followed by a modified version of the designation of the parental arginine-auxotrophic strain.

^dThe amino acids between M14 and P341 were deleted due to an 82-nucleotide deletion. The mutation led to a truncated protein and therefore to arginine auxotrophy.

^eThe G333E mutation led to a nonfunctional protein and therefore to arginine auxotrophy.

^fThe E333G mutation restored arginine prototrophy.

^gNS, not sequenced.

^hThe amber mutation led to a nonfunctional protein and therefore to arginine auxotrophy.

ⁱThe mutation restored arginine prototrophy. Another revertant, R-FB400-B3, was analyzed and had a stop51Q mutation (TAG → CAT) in argB, generating an ArgB protein identical to CCM 1904 ArgB.

^jins, insertion. The 47insY mutation led to a nonfunctional protein and therefore to arginine auxotrophy.

^kThe amber mutation led to a nonfunctional protein and therefore to arginine auxotrophy.

^lThe stop290L mutation led to a functional protein and therefore to arginine prototrophy. Two other revertants were analyzed: R-CCM3626-K (stop290E; TAG → CAG) and R-CCM3626-P (stop290Y; TAG → TAT).

Ornithine-requiring L. plantarum strains had mutations in the argCJBD genes.

The products of the argCJBD genes catalyzed ornithine synthesis in L. plantarum (3). When these functional genes present on plasmid pJIM207 were introduced by electroporation, the Orn⁻ L. plantarum strain NCFB 772 became Orn⁺ in CO₂-enriched air. This demonstrated that this gene cluster was deficient in NCFB 772. In order to characterize the mutations responsible for the ornithine requirement, the argCJBD genes in the six natural L. plantarum ornithine auxotrophs were PCR amplified and sequenced (see Materials and Methods). The prototroph L. plantarum CCM 1904 sequence was used as a reference for primer design. The argCJBD gene sequences were compared with the sequences of the reference prototroph, as well as the natural Orn⁻ and Orn⁺ derivatives, when available. The results of these comparisons are shown in Table 3 for major and minor mutations, which correlated with the abilities of the natural auxotrophs to revert to prototrophy.

The following minor mutations were found in two ornithine auxotrophs: transition point mutations G→A and C→T in NCFB 772 and FB400, respectively. In strain NCFB 772, a glutamate residue at position 333 replaced the glycine residue found in the Orn⁺ derivative R-NCFB772 and in the prototroph CCM 1904. In strain FB400, a premature stop codon (TAG) in the N-terminal part of argB reduced the size of the functional 248-amino-acid ArgB protein to 49 amino acids. A comparison of the sequences of the argCJBD genes in FB400 and two independent Orn⁺ revertants revealed a single transversion point mutation (TAT in R-FB400-A1 and CAG in R-FB400-B3) which reestablished the 248-amino-acid ArgB protein (Table 3). Thus, ornithine auxotrophy in FB400 was linked to a point mutation in argB.

Major mutations were detected in four ornithine auxotrophs. An 82-nucleotide deletion in argC resulted in truncation of 90% of ArgC. The same deletion was found in three strains (NCFB 963, NCFB 965, and NCFB 2171) which also had identical argCJBD nucleotide sequences (Table 3). Therefore, the presence of the deletion explained why no spontaneous revertants were obtained from these natural auxotrophs. In the Orn⁻ auxotroph KOG5, a 3-nucleotide insertion in argD introduced a tyrosine residue at position 47. Moreover, eight missense mutations resulting from four transversion mutations and four transition point mutations in the argCJBD genes were found (Table 3). KOG5 did not spontaneously revert to ornithine prototrophy. Genetic complementation with the functional argCJBD genes present in plasmid pJIM207 could not be tested in this strain since no transformants were obtained with two different electroporation protocols.

Citrulline-requiring L. plantarum strain CCM 3626 had a nonsense mutation within argF.

The natural isolate CCM 3626 and AM 1215, the control ArgF (G129E, G209E) mutant derived from CCM 1904 (3), grew in the presence of arginine or citrulline but not with ornithine. This phenotype suggested that argF was inactive in CCM 3626. When a functional argF gene present on plasmid pJIM407 was introduced by electroporation, arginine prototrophy was recovered. To characterize the defective CCM 3626 argF gene, the arg cluster was PCR amplified (see Materials and Methods) and sequenced. A nonsense mutation (TAG) was found to delete the last 50 residues of ArgF. A leucine codon (TTG), a glutamate codon (CAG), or a tyrosine codon (TAT) replaced the stop codon in three independent revertants (Table 3). These revertants did not require arginine for growth in CO₂-enriched air. Thus, the nonsense point mutation in argF was responsible for the citrulline-requiring phenotype of CCM 3626.

In FB400, the amber mutation in argB may be suppressed.

A premature stop codon in argB was found to be responsible for the ornithine auxotrophy of FB400 (Table 3). Two kinds of spontaneous prototrophs were selected from FB400. In revertants R-FB400-A1 and R-FB400-B3, the argB premature stop codon was mutated so that a functional ArgB product was generated. In other Arg⁺ derivatives (R-FB400-B4 and R-FB400-D7), the sequence of the argCJBD genes was identical to that of the parental strain (Table 3 and data not shown). Thus, the amber mutation found in argB was suppressed by a mutation elsewhere in the DNA.

DISCUSSION

Comparison of arginine and pyrimidine synthesis in four Lactobacillus species revealed conserved and species-specific impaired metabolic steps.

The arginine and pyrimidine requirements were investigated in a collection of lactobacilli that included L. plantarum (150 strains), L. pentosus (32 strains), L. paraplantarum (15 strains), and L. casei (10 strains). In the absence of both arginine and pyrimidines, the growth of 74 lactobacilli was CO₂ dependent. CO₂ is a substrate for CP synthesis, and CP is a common intermediate in both pathways (Fig. 1A). HCR auxotrophs represented 80, 40, and 33% of the total natural isolates of L. paraplantarum, L. pentosus, and L. plantarum, respectively. To characterize the molecular basis of the HCR phenotype, the genes involved directly or indirectly in CP metabolism remain to be investigated (see below) to demonstrate that CP metabolism is a preferred target for DNA evolution in these lactobacilli.

Among the 207 lactobacilli tested in the presence of 4% CO₂-enriched air, only L. pentosus natural isolate CNRZ 1547 was an auxotroph for the pyrimidines alone. Since spontaneous revertants had the phenotype of a strain with CPS-P deleted, CNRZ 1547 may have a revertible mutation in its CPS-P-encoding gene. Hence, the functionality of the pyrBCDEF genes involved in the de novo pyrimidine biosynthesis pathway (Fig. 1) is a common feature of all the lactobacilli tested. Arginine could be replaced by citrulline without changing the phenotype. Therefore, the two enzymatic steps required for arginine biosynthesis from its precursor, citrulline, which are encoded by the argGH genes (Fig. 1), are functional in all the lactobacilli tested.

Examination of citrulline and ornithine auxotrophies revealed differences in nutritional needs between species (Table 2). All the L. casei strains, no L. paraplantarum strains, and a few strains of L. pentosus (6%) and L. plantarum (0.7%) were citrulline auxotrophs. A lack of ornithine biosynthesis was detected in only 4% of the L. plantarum strains analyzed (Fig. 1B). Unlike the other species, L. casei may be characterized by an irreversible lack of ornithine carbamoyltransferase (OTCase) activity encoded by the argF gene (Fig. 1A), which may have resulted from major mutations impairing either the argF gene or other factors modulating OTCase activity. We demonstrated that in the L. plantarum citrulline auxotroph CCM 3626, the argF gene was inactivated by an amber mutation. In order to understand the biological significance of the specific metabolic evolution in L. casei, the mutations will be characterized in L. casei when its genome sequence is available. For the biosynthetic pathways studied, metabolic biodiversity was found not only in different species but also within species.

Characterization of L. plantarum deficient arginine biosynthetic genes suggests that there are random mutations.

The molecular basis of arginine auxotrophy in seven L. plantarum natural auxotrophs was investigated (Table 2). In three strains (NCFB 772, FB400, and CCM 3626), single transversion point mutations were solely responsible for arginine auxotrophy, as demonstrated with Arg⁺ revertants. The four other auxotrophs harbored nonrevertible lesions. An 82-nucleotide deletion in argC was identified in three strains (NCFB 963, NCFB 965, and NCFB 2171, which were considered to be isogenic based on identity of sequences, physiological parameters, and source of isolation). A 3-nucleotide insertion within argD combined with missense mutations in argJB was found in strain KOG5 (Table 2). We found that three of five different natural arginine auxotrophs harbored a single revertible point mutation in different arg genes (Fig. 1B). Different argCJBDF genes were mutated; no gene was preferentially inactivated in the natural process of single point mutation shutoff of citrulline and ornithine synthesis. We suggest that random mutations generated L. plantarum arginine auxotrophs.

argB paralogs in L. plantarum.

In the natural arginine auxotroph FB400 isolated from pickled vegetables, an amber mutation in the argB gene led to ArgB truncation. The following two kinds of spontaneous arginine derivatives were obtained: (i) derivatives with the stop codon removed and (ii) derivatives with suppressor mutations having an unchanged arg operon compared to the parent strain (Table 2). The latter may be due to a nonsense suppressor tRNA. Another possibility is the presence of an argB-like gene with N-acetylglutamate kinase activity that is inactive in parental strain FB400 or is weakly expressed and can be recruited after mutation. The genome sequence of L. plantarum (M. Kleerebezem, personal communication) revealed the presence of a second copy of argB. This second copy of argB may be involved in the suppressor mutation of the amber mutation of the argB gene within the arg operon. Sequence analysis of this locus in strains R-FB400B4 and R-FB400D7 is necessary to test this hypothesis.

Implications of this work for CP metabolism in L. plantarum, L. paraplantarum, and L. pentosus.

Several points are discussed below.

(i) All prototrophs were uracil sensitive.

The uracil sensitivity in air of the prototrophic strain L. plantarum CCM 1904 was demonstrated to result from an absence of CP synthesis for arginine biosynthesis. In normal air (with a low P_CO2), CPS-A was inactive and CPS-P was absent when pyr operon transcription was inhibited by uracil (28). In CO₂-enriched air, CPS-A was active, allowing cell growth in the presence of uracil. The phenotype of strain CCM 1904 was found in all prototrophs tested, including L. plantarum (94 strains), L. paraplantarum (3 strains), and L. pentosus (16 strains). Therefore, these prototrophs might also harbor a CPS inhibited by uracil, as well as a second CPS expressed in the presence of uracil, which requires a higher P_CO2 for efficient CP synthesis.

(ii) HCR auxotrophs may result from a loss of regulation or mutations involved in CP metabolism.

The inability of a strain to grow at a low P_CO2 without both arginine and pyrimidines suggests that genes involved in CP metabolism are deficient. The HCR phenotype may result from genetic lesions in CP synthesis or pathways that alter cellular CP pools (27). Mutation of the pyr operon transcription attenuators decreased pyr operon transcription, thereby lowering CPS-P expression (A. Elagöz, J.-C. Hubert, and B. Kammerer, unpublished data). These mutants had the same phenotype as the HCR auxotrophs. Increasing CP demand in a mutant defective in nucleic acid recycling (with a mutation in the uracil phosphoribosyltransferase gene) resulted in HCR auxotrophy (H. Nicoloff, J.-C. Hubert, and F. Bringel, unpublished data). These data suggest that the natural HCR auxotrophs may also be impaired in CP metabolism.

(iii) Most L. plantarum natural arginine auxotrophs have mutations not only in argCJBDF but also in genes involved in CP metabolism.

The presence of multiple lesions was determined in five L. plantarum arginine auxotrophs. In strain NCFB 772, mutations conferring the HCR phenotype were present in addition to the revertible argC mutation. A spontaneous Orn⁺ revertant (R-NCFB772) and NCFB 772 transformed with plasmid pJIM207 harboring a functional L. plantarum argC gene had the phenotype of HCR auxotrophs (Table 2, class 1). Since no prototrophic revertants from R-NCFB772 were obtained on DLA agar plates in air, major mutations in addition to the argC mutation are present in strain NCFB 772. In the auxotroph CCM 3626, uracil sensitivity was detected in derived ArgF⁺ revertants and in a derivative harboring plasmid pJIM407 with a functional argF gene. Similarly, the argCΔ82nt auxotrophs NCFB 965, NCFB 963, and NCFB 2171 were unable to grow in the presence of ornithine when uracil was added to the medium (data not shown). Uracil sensitivity suggests that these strains have an inactive CPS-A, as demonstrated in strain FB331 (28), but other enzymatic activities or genes may also be deficient.

Ecological niches and auxotroph occurrence.

We observed that L. plantarum arginine auxotrophs represented a higher percentage of the population isolated from milk products (13%) than of the population isolated from humans (0%) or plant products (2%). The number of L. plantarum natural arginine auxotrophs was low (7 auxotrophs, 143 prototrophs), and the sample sizes of the isolates obtained from the three sources were not equivalent (25 strains from dairy products, 16 strains from humans, and 89 plant-associated strains). However, our data suggest that an association with dairy products favors selection of natural L. plantarum arginine auxotrophs.

HCR auxotrophs were commonly found in lactobacilli (74 of 207 strains). In natural niches and manufactured products, lactobacilli are exposed to various levels of CO₂. In the gastrointestinal tracts of mammals, CO₂ is produced by intestinal mucosa metabolic activity. In the food industry, packing under a CO₂-controlled atmosphere is used to increase the shelf life of fresh meat, fish, or vegetables. In ecological niches with mixed or sequential microbial populations, CO₂-producing cells might provide the endogenous CO₂ necessary for HCR auxotrophic growth of lactobacilli. CO₂-producing organisms, such as heterofermentative LAB, Saccharomyces cerevisiae, or Propionibacterium, ferment products in which lactobacilli are also found (for instance, sourdough, beer; kefir, or certain types of cheeses). For the origins of these isolates, we identified two P_CO2-enriched environments: bread sourdough and beer. Although in theory HCR auxotrophs may have appeared in both fermented products, they were detected only in beer, in which three of five strains had evolved towards CO₂ dependence. It is not always understood how specific auxotrophies occur in some habitats. For instance, even though the amounts of branched-chain amino acids limit growth in milk, 94% of the Lactococcus lactis strains isolated from dairy products have auxotrophies for branched-chain amino acids, while most strains isolated from different fermented plant products are prototrophs (16). The difficulty in correlating the occurrence of auxotrophy with specific habitat adaptations may result from the complexity inherent in numerous metabolic pathway interconnections and nutritional exchange between organisms present simultaneously in fermented foods or ecological niches.

Lactobacillus evolution.

A few isolates of L. plantarum, a Lactobacillus species that is ubiquitous in different rich natural niches, have lost the ability to synthesize arginine. Analysis of the arg genes impaired revealed no gene that was preferably altered. These data support the hypothesis that LAB evolve by progressively losing unnecessary genes upon adaptation to specific habitats, with the genome evolving towards cumulative DNA degeneration. Some mutations may improve growth or cell viability in stressful or peculiar conditions, and thus the organism appears to have adapted to its environment. Recent studies have suggested that in starved or stressed culture conditions, a small subpopulation is hypermutable and involved in adaptive (stationary-phase) mutagenesis (15, 31). Lactobacilli encounter multiple stresses in their natural habitats, such as the stress during gastrointestinal tract transit, during long-term incubation (like that used in cheese ripening or beer, silage, or pickled vegetable fermentation), or during their stay on plant surfaces. The occurrence of mutations in nondividing or stressed cells has been documented in L. plantarum. In a wild-type tryptophan-requiring strain (ATCC 8014), tryptophan prototrophs were recovered in a Trp-depleted medium after 8 days of incubation (34) or after incubation at 30°C but not at 37°C after mutagenesis (26). Thus, in L. plantarum, stationary-phase stress could trigger adaptive mutation and reactivate cryptic pathways or inactivate unnecessary metabolic pathways. LAB metabolic diversity may be the result of stress-induced evolution. The ecological niches in which L. plantarum strains are found offer nutriment abundance, complex microbial population interactions, and multiple kinds of stresses. These factors could favor the occurrence of adaptive mutagenesis and subsequently contribute to the metabolic diversity of LAB.