Characterization of the Role of para-Aminobenzoic Acid Biosynthesis in Folate Production by Lactococcus lactis

Abstract

The pab genes for para-aminobenzoic acid (pABA) biosynthesis in Lactococcus lactis were identified and characterized. In L. lactis NZ9000, only two of the three genes needed for pABA production were initially found. No gene coding for 4-amino-4-deoxychorismate lyase (pabC) was initially annotated, but detailed analysis revealed that pabC was fused with the 3′ end of the gene coding for chorismate synthetase component II (pabB). Therefore, we hypothesize that all three enzyme activities needed for pABA production are present in L. lactis, allowing for the production of pABA. Indeed, the overexpression of the pABA gene cluster in L. lactis resulted in elevated pABA pools, demonstrating that the genes are involved in the biosynthesis of pABA. Moreover, a pABA knockout (KO) strain lacking pabA and pabBC was constructed and shown to be unable to produce folate when cultivated in the absence of pABA. This KO strain was unable to grow in chemically defined medium lacking glycine, serine, nucleobases/nucleosides, and pABA. The addition of the purine guanine, adenine, xanthine, or inosine restored growth but not the production of folate. This suggests that, in the presence of purines, folate is not essential for the growth of L. lactis. It also shows that folate is not strictly required for the pyrimidine biosynthesis pathway. L. lactis strain NZ7024, overexpressing both the folate and pABA gene clusters, was found to produce 2.7 mg of folate/liter per optical density unit at 600 nm when the strain was grown on chemically defined medium without pABA. This is in sharp contrast to L. lactis strains overexpressing only one of the two gene clusters. Therefore, we conclude that elevated folate levels can be obtained only by the overexpression of folate combined with the overexpression of the pABA biosynthesis gene cluster, suggesting the need for a balanced carbon flux through the folate and pABA biosynthesis pathway in the wild-type strain.

Many plants and bacteria have the ability to produce folate. Folate is essential for most animals, including humans, and insufficient intake of folate may lead to physiological disorders, such as anemia and neural tube defects in newborns (20). Among elderly people, folate deficiency may also lead to mental disorders such as psychiatric syndromes and decreased cognitive performance (6, 13). It is also assumed that folate has protective properties against cardiovascular diseases and a few types of cancer (4, 5, 20). Metabolic engineering of fermentative microbes can be used to produce food products with elevated folate contents.

Folate biosynthesis proceeds via the conversion of GTP in seven consecutive steps to the biologically active cofactor tetrahydrofolate (THF). Two condensation reactions take place in the biosynthesis pathway of THF. The first is the condensation of para-aminobenzoic acid (pABA) with 2-amino-4-hydroxy-6-hydroxymethyl-7,8-dihydropteridine to produce dihydropteroate. The second is the reaction of glutamate with dihydropteroate to form dihydrofolate (8) (Fig. 1). pABA itself is synthesized from the pentose phosphate pathway; in this pathway, d-erythrose 4-phosphate is condensed with phosphoenolpyruvate to ultimately lead to chorismate (Fig. 1). Chorismate serves as a branching point for the synthesis of the aromatic amino acids (tryptophan, phenylalanine, tyrosine) and pABA (21). In Escherichia coli, chorismate is converted via chorismate synthetase components I and II (PabB and PabA, EC 6.3.5.8) into 4-amino-4-deoxychorismate. Subsequently, pyruvate is cleaved by 4-amino-4-deoxychorismate lyase (PabC, EC 4.1.3.38), to result in pABA (10, 26) (Fig. 1). pABA knockout (KO) strains of E. coli are unable to grow in the absence of pABA or folate in a minimal medium (9, 15). Without pABA, no THF can be produced, and THF is essential as the donor and acceptor of one-carbon groups (i.e., methyl, formyl, methenyl, and methylene) in the biosynthesis of purines and pyrimidines, formyl-methionyl tRNA^fMet, and some amino acids (24, 37). Interestingly, it was found that Streptococcus faecalis R, incapable of producing folate, was able to grow on medium without folate. Growth was achieved in a chemically defined medium (CDM) supplemented with medium components which require folate for biosynthesis, i.e., methionine, serine, thymine, adenine, and guanine (33).

FIG. 1.

Biosynthesis pathway of pABA and folate, showing the conversion of GTP into THF, the route for pABA biosynthesis from chorismate, and the incorporation of pABA into THF.

An extensive genetic metabolic engineering approach has been applied to the folate biosynthesis genes in Lactococcus lactis (38). The overexpression of the folate gene cluster (folB folKE folP folQ folC) resulted in a folate-overproducing strain, but only when pABA was added to the medium.

In this paper, we describe the involvement of the pABA genes in the production of folate in L. lactis. First, the genes coding for pABA biosynthesis were characterized. Second, a pABA KO strain was constructed; in this strain, folate production and growth performance were analyzed in the presence and absence of pABA, glycine, serine, and nucleobases/nucleosides (guanine, adenine, uracil, xanthine, inosine, thymidine, and orotic acid). Finally, we analyzed the effect of overexpression of the folate biosynthesis genes combined with overexpression of the pABA biosynthesis genes on folate pools.

MATERIALS AND METHODS

Bacterial strains, growth conditions, and cultivations.

The bacterial strains, plasmids, and primers used in this study are listed in Table 1. The Lactococcus lactis NZ9000-derived strains were grown at 30°C on CDM (25, 27, 29) and M17 broth and agar (43). CDM contains the following medium components: 3 g/liter K₂HPO₄, 3 g/liter KH₂PO₄, 1 g/liter sodium acetate, 0.6 g/liter ammonium citrate, 11 g/liter glucose · 1H₂O, 0.5 g/liter ascorbic acid, 0.25 g/liter l-tyrosine, 0.001 g/liter Ca-(d)-(+)-pantothenate, 0.0025 g/liter d-biotin, 0.001 g/liter 6,8-thiotic acid, 0.001 g/liter nicotinic acid, 0.005 g/liter pyridoxamine HCl, 0.002 g/liter pyridoxine HCl, 0.001 g/liter riboflavin, 0.001 g/liter thiamine HCl, 0.001 g/liter vitamin B₁₂, 0.005 g/liter inosine, 0.005 g/liter thymidine, 0.005 g/liter orotic acid, 0.2 g/liter MgCl₂ · 6H₂O, 0.05 g/liter CaCl₂ · 2H₂O, 0.016 g/liter MnCl₂ · 4H₂O, 0.003 g/liter FeCl₃ · 6H₂O, 0.005 g/liter FeCl₂ · 4H₂O, 0.005 g/liter ZnSO₄ · 7H₂O, 0.0025 g/liter CoSO₄ · 7H₂O, 0.0025 g/liter CuSO₄ · 5H₂O, 0.0025 g/liter (NH₄)₆Mo₇O₂₄ · 4H₂O, 0.24 g/liter alanine, 0.125 g/liter arginine, 0.42 g/liter aspartic acid, 0.13 g/liter cysteine-HCl, 0.50 g/liter glutamic acid, 0.175 g/liter glycine, 0.15 g/liter histidine, 0.21 g/liter isoleucine, 0.475 g/liter leucine, 0.44 g/liter lysine, 0.125 g/liter methionine, 0.275 g/liter phenylalanine, 0.675 g/liter proline, 0.34 g/liter serine, 0.225 g/liter threonine, 0.05 g/liter tryptophan, 0.325 g/liter valine, 0.01 g/liter guanine, 0.01 g/liter adenine, 0.01 g/liter uracil, and 0.01 g/liter xanthine. CDM was used to evaluate the growth rate and the pools of pABA and folate. M17 broth and M17 agar were used for the construction of the engineered strains. The CDM described here may lack nucleobases, nucleosides, glycine, serine, and pABA. When these components were added separately, the concentrations as described above were used. Spent medium was made by growth of the NZ9000ΔpABA strain on CDM lacking nucleobases, nucleosides, glycine, serine, and pABA, and the spent medium was filter sterilized with a 0.22-μm filter. The pH of the spent medium was adjusted to the original pH of 6.5, and 0.5% additional glucose was added; afterwards, this medium was filter sterilized again. The spent medium was subsequently used as the growth medium to test specific nutrient requirements of L. lactis NZ9000ΔpABA. The E. coli DH5α strain (47), used for the construction of the pABA KO strain, was grown at 37°C in TY medium plates (30). Growth rates for the L. lactis strains were determined in 96-well microtiter plates by measuring turbidity at 600 nm using the Spectra Max384 spectrophotometer (Molecular Devices). To select and maintain the plasmid, we used the following antibiotics: chloramphenicol at 10 mg/liter, tetracycline at 10 mg/liter, and erythromycin at 10 mg/liter for L. lactis and at 50 mg/liter for E. coli.

TABLE 1.

List of strains, constructed plasmids, and primers used

Strain, plasmid, or primer	Relevant feature(s)a	Source or reference
Strains
E. coli DH5α	Cloning host	47
L. lactis NZ9000	MG1363 pepN::nisRK, cloning host	16
L. lactis NZ7024	L. lactis NZ9000 containing pNZ7019 and pNZ7023	This study
L. lactis NZ9000ΔpABA	L. lactis NZ9000 lacking pabA and pabBC	This study
L. casei ATCC 7469	L. casei, the indicator strain for the folate microbiological assay	11
Plasmidsb
pNZ8148	Cmr; empty vector, nisin promoter	16
pNZ7020	Cmr pNZ8148 derivative, containing the pabA and pabBC genes of L. lactis NZ9000	This study
pNZ7021	Cmr pNZ8148 derivative, nisin promoter replaced by pepN promoter	This study
pNZ7022	Cmr RpNZ7021 derivative, containing the pabA and pabBC genes of L. lactis NZ9000	This study
pNZ7017	Cmr, pNZ8148 derivative containing folKE of L. lactis	39
pNZ7019	Cmr, pNZ8148 derivative containing the folB, folP, folKE, folQ, 1and folC gene cluster of L. lactis	46
pNZ5319	Cmr Eryr, double-crossover deletion vector	17
pNZ7027	Cmr Eryr, pNZ5319 vector containing a linear PCR product; surface protein gene upstream of pabA, including the final 15 bases pairs of pabA	This study
pNZ7028	Cmr Eryr, pNZ7028 containing a linear PCR product and the final 15 base pairs of pabBC up to 346 bp in yneH	This study
pIL253	Eryr	35
pNZ7023	Eryr, pIL253 derivative, containing the nisin promoter along with pabA and the pabBC gene cluster of L. lactis	This study
pNZ5327	Tetr pGHOST derivative, containing the resolvase gene for recombination of lox sites	This study
pNZ5348	Tetr NZ7110 derivative, containing the resolvase gene for recombination of lox sites	17
pGHOST8	Tetrori(Ts)	2
Primers
FLLSURFPRxho-F	CGTTCTCGAGTGATCATATATC
RLLpabA15nucl-R	TAAAAGTAATCGCATTTTATC
FLLpabB/C15nucl-F	GATGGTATTTCTTAATTATAAAG
RLLyneHbgl2-R	GTACGAAGATCTTCATTATTG
FpRB36catF3	CATTACCGAAGTAATCGTTAAAC
RLLyneHupstream	CTTTTAAATGTAGCAAAAAC
66doRCS85	GTTTTTTTCTAGTCCAAGCTCACA
FLLSURFPRupstream	GTTTTATAAAACATGAAAG
lclpabAsphI2	AGTCGCATGCGAGGAGGATAAAATGCGATTAC
lclpabB/CXbaI	CAAGGCTTTTCTAGATTAAGAAATACCATCG
pnisF	TAGTCTTATAACTATACTGAC
pabB/CrevR	CTAGATTAAGAAATACCATCG

^aThe XhoI digestion sequence is underlined in the forward primer FLLSURFPRxho-F, as are the BglII digestion sequence in the reverse primer RLLyneHBgl2-R, the SphI site in the lclpabAsph primer, and the XbaI site in the reverse primer lclpabBCxbaI.

^bWhere indicated, all pNZ plasmids contain the pSH71 or PWV01 origin of replication.

Genetic constructs and DNA methods.

E. coli DH5α was transformed by the CaCl₂ procedure (32) and L. lactis strains by electroporation (7). PCRs for the construction of KO and overproduction vectors were performed with Pfx polymerase (Invitrogen, Breda, The Netherlands). Taq polymerase (Promega Biotech, Roosendaal, The Netherlands) was used for the PCRs to check the lengths of inserts. L. lactis genomic DNA was isolated using established procedures (18, 45), and plasmid DNA was isolated using Jetstar columns (Genomed GmbH, Bad Deynhausen, Germany). All restriction enzymes but Ecl136II (Fermentas UAB, Vilnius, Lithuania) were purchased from Invitrogen. The preintegration vector for the inactivation of the pabA-pabBC gene cluster was constructed by using the Cre-lox system (17). The sequence with the up- and downstream regions of the pABA gene cluster of L. lactis MG1363 was used for amplification by PCR. This sequence appears in the GenBank nucleotide sequence library under accession number AM406671. The first amplified linear fragment of DNA is 1,359 base pairs in length (the pabA fragment); the second fragment, with a length of 550 base pairs, is designated the pabBC fragment. The pabA fragment contains the upstream region of pabA, annotated as a surface protein gene, and ends with the first 15 base pairs of pabA. For the PCR, the forward primer FLLSURFPRxho-F (the sequence was modified to introduce an XhoI digestion sequence [Table 1]) and the reverse primer RLLpabA15nucl-R were used (Table 1). The second fragment, the pabBC fragment, contains the final 15 base pairs of pabBC, the intergenic regions, and 346 base pairs of the neighboring open reading frame yneH. The forward primer FLLpabBC15nucl-F and reverse primer RLLyneHBgl2-R (the sequence was modified to introduce a BglII digestion sequence [Table 1]) were used for the PCR. After amplification of the two PCR fragments, plasmid pNZ5319 was digested with PmeI and XhoI and the pabA fragment was digested with XhoI. Subsequently, the digested pabA fragment was ligated into the PmeI/XhoI-digested pNZ5319 vector with T4 DNA ligase (Invitrogen) and was then cloned in the E. coli DH5α strain. Transformants with the expected insert length were selected on TY agar plates containing erythromycin. The selected colonies were checked for the proper digestion profile. The resulting plasmid, designated pNZ7027, was isolated and used for Ecl136II and BglII digestion, and the pabBC fragment was digested by BglII and subsequently ligated into the Ecl136II/BglII-digested pNZ7027 vector with T4 ligase. E. coli DH5α was subsequently transformed with the ligation mix. After 2 days of growth, a few colonies were picked and cultivated. From these strains, the pABA KO integration vector (pNZ7028) was isolated, checked for the presence of the pabBC fragment by PCR, and digested to check for restriction profiles. The pNZ7028 pABA KO vector was transferred to competent cells of the L. lactis NZ9000 strain. The transformed cells were selected on M17 plates containing chloramphenicol. To select colonies with the double-crossover genotype, erythromycin-sensitive colonies were selected by colony streaking on plates containing chloramphenicol with or without erythromycin. Erythromycin-sensitive L. lactis colonies were then analyzed by PCR to confirm the double-crossover genotype. The analysis was performed by using the forward primer FLLsurfpRupstream and the reverse primer 66doRCS85 to amplify the upstream region of the surface protein gene up to the chloramphenicol resistance gene. The downstream region of yneH was also checked by PCR. For this purpose, the forward primer FpRB36catF3 and the reverse primer RLLyneHupstream were used to amplify the chloramphenicol resistance gene in the downstream region of yneH. A clean deletion mutant was obtained by the addition of a plasmid which contained the resolvase gene to recombine the two lox loci on both sides of the chloramphenicol resistance gene. The plasmid for the recombination of the two lox sites in L. lactis was constructed as follows. The resolvase gene was obtained by digestion of plasmid pNZ5348 (17) with HindIII and KpnI. Subsequently, plasmid pGHOST8 (2) was digested with SmaI. Afterwards, both digested DNA fragments were mixed and used for T4 DNA ligation. After the ligation, SmaI digestion was performed to prevent self-ligation. Then the ligated DNA fragment was transferred to L. lactis NZ9000 and tetracycline-resistant strains were selected and checked for the correct restriction profile. The vector with the proper orientation was named pNZ5327. Growth experiments with the transformants containing pNZ5327 were performed at 20°C. The clean KO was made by transferring pNZ5327 into competent L. lactis NZ9000ΔpABA. Tetracycline- and chloramphenicol-resistant colonies were picked up and grown on M17 at 42°C for 1 h; thereafter, the cells were further cultivated at 30°C. This temperature shift leads to the instability of the maintenance of pNZ5327. After prolonged incubation for approximately 24 h, the vector is lost. The clean KO genotype of strain NZ9000ΔpABA was tested by PCR using the forward primer FLLsurfpRupstream and the reverse primer RLLyneHupstream. A single L. lactis colony lacking the pABA gene cluster was selected, cultivated, and stored as a glycerol stock. This strain was designated NZ9000ΔpABA.

Overproduction of pABA by using the pNZ8148 and the pIL253 vector.

For the overproduction of pABA and complementation of strain NZ9000ΔpABA, two nisin-inducible vectors were constructed. One vector was based on pNZ8148 (16) and the other one on pIL253 (35). First, the pNZ8148-derived vector was constructed; the L. lactis pABA biosynthesis gene cluster pabA-pabBC was amplified by PCR using the forward primer lclpabAsphI2 and the reverse primer lclpabB/CxbaI. The lclpabAsphI2 and lclpabB/CxbaI primers were modified in their sequences to introduce two restriction sites. The forward primer contained an SphI site, and the reverse primer contained an XbaI restriction site (Table 1). The vector pNZ8148 was digested with XbaI and SphI, and the same digestion was performed on the amplified DNA of the pABA gene cluster. Subsequently, both fragments were mixed, ligated with T4 DNA ligase, and transferred to the L. lactis NZ9000 wild-type strain and to the NZ9000ΔpABA strain. The resulting vector was named pNZ7020. For the second overproduction vector based on pIL253, plasmid pNZ7020 (containing the pABA gene cluster behind the nisin promoter) was used as a template for a PCR using forward primer pnisF and reverse primer pabBCrevR. The PCR product, containing the nisin promoter and the pABA gene cluster, was mixed with a SmaI-digested pIL253 vector and subsequently ligated using T4 DNA ligase. After ligation, the mixture was again digested with SmaI to prevent the self-ligation of pIL253. The ligation mix was transferred to competent cells of L. lactis NZ9000, NZ9000ΔpABA, and L. lactis harboring the folate overexpression vector pNZ7019 (46). Transformed cells capable of growing on erythromycin were selected, checked by PCR for proper orientation, and further checked by restriction profiling. The resulting vector was named pNZ7023. L. lactis harboring pNZ7019 and pNZ7023 was designated L. lactis NZ7024. Cultivation of this strain requires the presence of both chloramphenicol and erythromycin.

A pNZ8148 derivative was used for the constitutive overexpression of pABA. The pNZ8148 vector was first digested using BglII and SphI. The same digestion was also performed on the pNZ7017 vector (39), thereby excising the constitutive pepN promoter. The digested pNZ8148 vector and pepN promoter were mixed, and both fragments were ligated using T4 ligase. Subsequently, the ligation mixture was added to competent cells of L. lactis NZ9000 for transformation by electroporation. Transformants capable of growing on M17 plates containing chloramphenicol were selected and checked for the presence of the vector. The resulting vector was named pNZ7021. Plasmid pNZ7021 was then isolated from L. lactis. To construct the vector for constitutive pABA overproduction, first the pABA gene cluster of L. lactis was amplified using the forward primer lclpabAsphI2 and the reverse primer lclpabB/CxbaI (whose sequences were modified to introduce an SphI and XbaI digestion sequence [Table 1]). Plasmid pNZ7021 and the PCR product were digested with SphI and XbaI. These two DNA fragments were ligated and transferred into electrocompetent cells of L. lactis NZ9000 and NZ9000ΔpABA. The constructed vector was designated pNZ7022. In addition, the L. lactis wild-type and NZ9000ΔpABA strains were transformed with pNZ8148 to deliver experimental control strains.

Folate and pABA analyses.

Folate was quantified using a microbiological assay with Lactobacillus casei ATCC 7469 as the indicator strain. Enzymatic deconjugation of polyglutamate tails was part of the sample preparation (11, 40). The detection limit of the microbiological folate assay was determined to be 2 μg/liter. To analyze folate production in the pABA KO strain grown on CDM lacking pABA and/or nucleobases/nucleosides, the following concentration step was employed to detect low folate concentrations. A culture (250 ml) was centrifuged and washed twice with 50 ml wash buffer (100 mM sodium acetate and 1% vitamin C) to prevent oxidation of the folate molecules. The pellet was resuspended in 5 ml wash buffer and subsequently extracted by bead beating (three times for 30 s at speed 4 using a FastprepFP120 apparatus [Qbiogene Inc., France]). The broken cells were boiled for 3 min. Subsequently, the supernatant was collected by centrifugation and the centrifugation step was repeated several times to remove all cell debris. The clear supernatant was finally lyophilized, and the resulting cell extract was dissolved in 0.5 ml 0.1 M NaOH and was further used for folate analyses in the microbiological assay.

The pABA production levels were determined using a procedure based on the high-performance liquid chromatography (HPLC) method described previously (41, 42). Instead of using an HPLC separation gradient of 75% elution liquid B (MilliQ plus 1.5% formic acid) and 25% elution liquid C (80% MilliQ-20% methanol plus 1.5% formic acid), a ratio of 60% elution liquid B to 40% elution liquid C was used for the proper separation of pABA on the column. For the pABA standard, 10 mg/liter of pABA was used.

Chemicals.

All chemicals were reagent grade and obtained from commercial sources.

RESULTS

Identification of the pABA genes in L. lactis.

The pabC gene, essential for the conversion of 4-amino-4-deoxychorismate to pabA, was initially not identified in the genomes of Lactococcus lactis SK11 and IL-1403 (ERGO database). The pabC gene could also not be identified in the pABA gene cluster of MG1363. Nevertheless, strains SK11 and MG1363 were able to produce folate (41), suggesting that the pABA biosynthesis pathway is complete in these strains. Interestingly, all three L. lactis strains contain the pabA and pabB genes, which are essential for the conversion of chorismate into 4-amino deoxychorismate. To identify the missing pabC gene, the pabB gene of IL-1403 was compared to all genomes present in the ERGO database by BLAST P (31) analysis. This revealed that pabB in Clostridium difficile 630 was very similar to the first part of pabB in L. lactis IL-1403. However, pabB of L. lactis contains an additional sequence with a length of 741 base pairs, which is homologous to pabC of C. difficile 630 (Fig. 2). The pabC of C. difficile 630 shares homology to known 4-amino deoxychorismate lyases of both E. coli and Bacillus subtilis (10, 36). Based on this in silico analysis, we conclude that L. lactis contains a pabBC fusion gene. The BLAST search also revealed that such pabBC fusion genes are present in both gram-positive and gram-negative species, such as Streptococcus pyogenes M5 and Helicobacter pylori J99, respectively (Fig. 2). Based on the L. lactis MG1363 sequence, we hypothesize that the three L. lactis strains possess all genes needed for pABA production, which would be in agreement the observed folate production in the L. lactis NZ9000 strain, when cultivated in the absence of pABA (41).

FIG. 2.

Organization of the pabBC fusion gene in Lactococcus lactis MG1363 compared to those of the pabB (pABA synthetase component I) and pabC (4-amino-4-deoxychorismate lyase) genes of Clostridium difficile 630, Lactococcus lactis IL-1403, Streptococcus pyogenes M5, and Helicobacter pylori J99. The E score for the similarity between the pabC gene of C. difficile 630 and the pabC part of the L. lactis IL-1403 pabBC gene upon BLAST P analysis was 1.e⁻⁷, suggesting similar functions.

Inactivation of the pabA-pabBC gene cluster.

The presence, activity, and physiological role of the pABA biosynthesis pathway in L. lactis was evaluated by inactivating the pABA gene cluster through the construction a double-crossover mutant. Strain L. lactis NZ9000ΔpABA lacks 98.8% of the pABA gene cluster. The growth rate, final optical density at 600 nm (OD₆₀₀), and total folate pools were determined for the wild-type strain and strain NZ9000ΔpABA, grown on CDM in the presence or absence of pABA and nucleobases/nucleosides (Table 2). The growth rate and final OD₆₀₀ of strain NZ9000ΔpABA on medium lacking pABA and nucleobases/nucleosides are severely impaired, reaching a maximum growth rate of only 0.02 h⁻¹ and a final OD₆₀₀ of 0.5, which are much lower than the growth rate of 0.31 h⁻¹ and final OD₆₀₀ of 3.2 for the wild type. The fact that the NZ9000ΔpABA strain was able to grow poorly but consistently on medium lacking pABA and nucleobases/nucleosides is somewhat unexpected. Therefore, we checked whether the CDM contained trace amounts pABA, folate, glycine, serine, or nucleobases/nucleosides. To test this, the following experiment was conducted. Strain NZ9000ΔpABA was grown on CDM lacking pABA glycine, serine, and nucleobases/nucleosides until no further increase in optical density was observed (OD₆₀₀ of 0.5). The spent medium of strain NZ9000ΔpABA was filter sterilized and adjusted to the initial pH, and finally, 0.5% additional glucose was added. The freshly inoculated NZ9000ΔpABA strain was unable to grow on the spent medium, even though glucose and other nutrients were still present in sufficient amounts. The addition of pABA to the spent medium resulted in good growth of the KO mutant, showing that the medium still supports growth. The spent CDM was subsequently supplemented with either adenine, guanine, inosine, or xanthine (purines); orotic acid, thymidine, or uracil (pyrimidines); glycine/serine; or pABA. Interestingly, the growth of the pABA KO strain was restored in the presence of the purine nucleobases (adenine, guanine, and xanthine) as well as the purine nucleoside inosine. The addition of pyrimidines (uracil, thymidine, and orotic acid) as well as the mixture of glycine and serine did not support the growth of the mutant in the spent medium (Table 3). The results of this experiment, together with the data supplied in Table 2, show that, in L. lactis, folate is essential for purine biosynthesis. The experiments also show that folate is not strictly required for the pyrimidine biosynthesis pathway. Finally, we conclude that the folate-dependent interconversion of glycine and serine is not essential for growth (24, 37).

TABLE 2.

Growth rates, final OD₆₀₀ values, and total folate production levels of the wild-type and NZ9000ΔpABA strains grown on CDM with and without pABA and nucleobases/nucleosides

Presence of pABA	Presence of nucleobase/nucleoside	Wild type		NZ9000ΔpABA		Mean concn (μg/liter) of folate (SD) produced in:
		Mean growth rate (h−1) (SD)	OD600	Mean growth rate (h−1) (SD)	OD600	Wild type	NZ9000ΔpABA
−	−	0.31 (0.03)	3.2	0.02 (0.001)	0.5	91.7 (10.5)	0.05 (0.01)a
+	−	0.31 (0.01)	3.3	0.29 (0.03)	3.1	129.9 (7.9)	92.0 (12.3)
−	+	0.32 (0.05)	3.5	0.26 (0.04)	3.3	103.7 (9.6)	0.05 (0.01)a
+	+	0.32 (0.03)	3.5	0.35 (0.06)	3.3	90.3 (16.0)	109.7 (10.3)

^aThis sample was concentrated 500-fold prior to its addition in the microbiological assay. The actual concentration given in this table is the calculated value based on analysis of the concentrated sample in the assay.

TABLE 3.

Growth of the NZ9000ΔpABA strain on spent medium without and with supplementation

Medium component(s) added to spent medium	Growtha
None	−
Adenine	+
Guanine	+
Inosine	+
Xanthine	+
Orotic acid	−
Thymidine	−
Uracil	−
Glycine and serine	−
pABA	+
All nucleobases/nucleosides	+

^aNo growth is expressed as −, and growth is expressed as +.

Folate pools were determined in the wild type and in strain NZ9000ΔpABA on CDM containing or lacking pABA and nucleobases/nucleosides (Table 2). Regardless of whether pABA and nucleobases/nucleosides were added to the medium, wild-type L. lactis accumulated folate pools ranging from 90.3 to 129.9 μg/liter. This is consistent with folate pools in L. lactis strains measured in another study (41). However, the pABA KO strain was unable to accumulate folate pools when cultivated on CDM lacking pABA, thereby reaching folate levels of only 0.05 μg/liter. This low level of accumulated folate is possibly a consequence of the uptake of trace amounts of folate or folate precursors from the medium (see above). Interestingly, the growth rate of strain NZ9000ΔpABA was restored by the addition of nucleobases/nucleosides to the medium, although folate pools remained very low (0.05 μg/liter). The addition of pABA to a culture of the NZ9000ΔpABA strain boosted folate pools back to wild-type levels, i.e., approximately 100 μg/liter (Table 2). These experiments suggest that folate production in L. lactis is essential for growth only when no nucleobases/nucleosides are present in the medium. In the presence of purines, folate is not essential for growth.

Effect of pABA overproduction on the folate and pABA pools in the wild-type and NZ9000ΔpABA strains.

The pabA-pabBC genes for the biosynthesis of pABA were cloned on different plasmids to determine the impact of pABA overproduction on the production of folate in the wild-type and NZ9000ΔpABA strains. Strain NZ9000 and NZ9000ΔpABA were complemented with two nisin-inducible pABA overexpression vectors (pNZ7020 and pNZ7023) and one constitutive pABA overexpression vector (pNZ7022). Folate analysis, performed on strain NZ9000ΔpABA harboring pNZ7020, showed that pools were restored to wild-type levels, regardless of whether nisin was added for induction (data not shown). The folate analyses performed on the wild-type strain harboring pNZ7020 displayed a shift in folate distribution across the intracellular and extracellular compartment upon induction of expression with nisin (Fig. 3). The L. lactis wild-type strain harboring the empty vector pNZ8148 showed no shift in folate distribution upon induction with nisin, indicating that this shift was caused by the overproduction of pABA and not by the nisin itself. The shift in folate distribution was also observed in strain NZ9000 (wild type) harboring pNZ7022 (constitutive overexpression) and pNZ7023 (nisin-induced overexpression) as well as in the NZ9000ΔpABA strain harboring the same pABA overproduction vectors, pNZ7020, pNZ7022, and pNZ7023 (data not shown). In none of the strains were increased folate pools observed. In conclusion, overproduction of pABA did not increase folate production levels, suggesting the need for a balanced carbon flux through the folate and the pABA biosynthesis pathway.

FIG. 3.

Folate production (μg/ml per OD₆₀₀ unit) of the L. lactis wild-type strain (wt) harboring pNZ8148 (empty vector) and pNZ7020 (nisin-inducible pABA overexpression vector). The folate pool quantities of intracellular, extracellular, and total fractions are shown. The uninduced folate levels are depicted in gray bars, and the induced folate levels are shown in black.

The levels of pABA production in the L. lactis wild type, the NZ9000ΔpABA strain, and the NZ9000ΔpABA strain harboring pNZ7020, pNZ7022, and pNZ7023 were quantified. All strains were cultivated on CDM with nucleobases/nucleosides in the absence of pABA. The levels of pABA production were determined by HPLC. The accumulation of pABA could not be detected in the L. lactis wild-type or NZ9000ΔpABA strain or in the uninduced NZ9000ΔpABA strain carrying pNZ7020 or pNZ7023. However, after nisin induction, the NZ9000ΔpABA strains harboring pNZ7020 and pNZ7023 were found to accumulate pools of 7.8 and 5.3 mg of pABA/liter, respectively, per OD₆₀₀ unit in the cell extract. An equivalent amount of pABA was found to be produced by strain NZ9000ΔpABA containing pNZ7022 (5.3 mg/liter per OD₆₀₀ unit). Assuming complete retention of pABA and an intracellular volume of 3.6 μl/mg cell protein (an OD₆₀₀ value of 1 corresponds with 0.2 mg cell protein per ml) (28), the intracellular pABA pool reaches a size of 57 to 84 mM. In conclusion, overexpression of the pabA-pabBC genes in L. lactis resulted in a high intracellular accumulation of pABA, proving their involvement in the biosynthesis of pABA.

Overexpression of folate and pABA.

Overexpression of the pABA gene cluster in strain NZ9000ΔpABA boosted the pABA production but did not lead to elevated folate pools. Sybesma (38) has shown that pABA biosynthesis controls the overproduction of folate in a strain which overexpresses the entire folate gene cluster. We hypothesize that the expression of the pteridine pathway and that of the pABA pathway are tightly coupled. In previous work (38), an L. lactis strain (NZ9000 containing the pNZ7019 vector) that produces high folate levels was constructed; the folate biosynthesis genes (folB, folKE, folP, folQ, and folC) were cloned on a high-copy-number vector (pNZ8148 derivative) under the control of the pepN promoter. This strain was cultivated in CDM in the presence and absence of pABA. Only when pABA was supplied to the medium were high folate levels (2.3 mg/liter per OD₆₀₀ unit) found to be produced. In the absence of pABA, no elevated folate levels were observed (Fig. 4.). Vector pNZ7023 (pIL253 derivative), containing the pabA-pabBC gene cluster under the control of the nisin promoter, was transferred to competent cells of L. lactis NZ9000 harboring pNZ7019. The resulting strain, named L. lactis NZ7024, was cultivated on CDM without pABA. Interestingly, after nisin induction, this strain was able to produce 2.7 mg of folate/liter per OD₆₀₀ unit, independently of pABA supplementation of the growth medium (Fig. 4). Without induction, only 0.4 mg of folate/liter per OD₆₀₀ unit was produced. In the L. lactis NZ7024 strain, the expression of the pABA gene cluster results in the conversion of chorismate into high pABA levels. Simultaneously, the overexpression of the folate gene cluster leads to the conversion of GTP into THF, which can be achieved only when high pABA levels are present. Therefore, we conclude that the overexpression of the pABA genes combined with the overexpression of the folate biosynthesis genes is essential for the overproduction of folate. The overexpression of only pABA or the folate gene cluster separately cannot boost folate production.

FIG. 4.

Total folate production levels (mg/liter per OD₆₀₀ unit) of three different strains: L. lactis NZ9000 (wild type [wt]), L. lactis NZ9000 harboring pNZ7019 (folate overexpression vector) cultivated in the presence and absence of pABA, and L. lactis NZ7024 (NZ9000 strain harboring the folate and pABA overexpression vector) cultivated without pABA prior to and after the induction with nisin.

DISCUSSION

Based on the observation in a previous study that L. lactis MG1363 is capable of producing folate (41), it was expected that all genes for the synthesis of pABA are present in L. lactis. And indeed after bioinformatics analysis it was found that L. lactis IL-1403 (3), a strain closely related to L. lactis MG1363, possesses the three genes needed for pABA production. L. lactis MG1363 contains pabA as a separate open reading frame and pabBC as a fused open reading frame. These merged pabBC genes were found not only in L. lactis but also in other gram-positive bacteria as well as in several gram-negative bacteria, suggesting that this gene fusion was the result of an ancient event. Another interesting feature of the pABA genes is the homology they share with the genes for tryptophan biosynthesis. The pabA gene is homologous with the trpG gene, and pabB is homologous with trpE (23). However, no pabC homologue has been found, suggesting that pabC is very specific for converting 4-amino-4-deoxychorismate into pABA. In the tryptophan biosynthesis pathway, chorismate is converted via TrpE and TrpG into 2-ABA (anthranilate), whereas in the pABA biosynthesis pathway, chorismate is converted via PabA, PabB, and PabC into 4-ABA (pABA). The facts that two enzymes are required to produce 2-ABA and that three enzymes are needed for the synthesis of pABA highlight the important role of PabC and the crucial difference between the tryptophan and the pABA biosynthesis pathway.

The pABA biosynthetic enzymes (PabA, PabB, and PabC) play an important role in the biosynthesis of folate. The deletion of the pabA-pabBC gene cluster in L. lactis led to a strain that was unable to grow in the absence of pABA, glycine, serine, and nucleobases/nucleosides. Adding all nucleobases/nucleosides to the medium restored the growth of strain NZ9000ΔpABA completely but did not restore folate production. It was demonstrated that single additions of purines to the spent medium restored the growth of the NZ9000ΔpABA strain. From this it can be concluded that the addition of purines circumvents the requirement for folate as a cofactor for the synthesis of IMP. Purines are synthesized from 5-phospho-d-ribosyl-1-pyrophosphate via several steps into IMP (14). In this biosynthetic pathway, 10-formyltetrahydrofolate is used twice as a cofactor, showing the key role of 10-formyltetrahydrofolate in the formation of purines. But the role of 5,10-methylenetetrahydrofolate is not so clear for both the conversion of dUMP to dTMP and the conversion of glycine to serine. Glycine, serine, and pyrimidines can be omitted from the medium, and still the NZ9000ΔpABA strain is able to grow, providing the presence of purines in the medium. This demonstrates that the pyrimidine biosynthesis pathway seems to operate in the absence of 5,10-methylenetetrahydrofolate. This raises the question of which cofactor, other than folate, is used by thymidylate synthase (EC 2.1.1.45) for the conversion of dUMP to dTMP. However, the complex defined medium contains components that can circumvent the necessity of 5,10-methylenetetrahydrofolate as a cofactor for the glycine-serine interconversion. From the model of the metabolic network of L. plantarum WCFS1 (44), it could be deduced that threonine can be converted into glycine via threonine aldolase (EC 4.1.2.5.). In addition, the reconstructed metabolic network also predicts that cysteine can be converted into serine by cysteine synthetase (EC 4.2.99.8) and serine O-acetyltransferase (EC 2.3.1.30). In a similar experiment, the inhibitory effect of antifolates like methotrexate and trimethoprim can also be counteracted by the addition of the end products of folate biosynthesis (1, 12). These antifolates compete with folate in the cell for the enzyme dihydrofolate reductase (EC 1.5.1.3.), thereby depleting the folate pool which serves as the cofactor in the nucleotide biosynthesis pathway.

One other interesting feature of the medium depletion experiments is the fact that somehow protein translation is initiated in the absence of folate. This was previously investigated in S. faecalis R by Samuel et al. (33), who suggested that there might be an alternative C₁ donor for the initiation of protein synthesis or that no formylation is needed to initiate translation in S. faecalis R.

Overexpression of the pABA gene cluster on three different vectors, namely, two nisin-inducible vectors (pNZ7020 and pNZ7023) and one constitutive vector (pNZ7022), led to the accumulation of pABA. The overproduction of pABA, however, did not lead to elevated folate pools. This confirms earlier results of Sybesma (38), who demonstrated that folate levels could not be increased solely by the overexpression of the folate biosynthesis genes of L. lactis. Hence, we constructed a strain (L. lactis NZ7024) overexpressing the pABA and the folate biosynthesis gene clusters simultaneously. This strain was found to produce high folate levels independently of pABA supplementation.

Although overproduction of pABA alone did not result in elevated (total) folate pools, we observed a shift in the folate distribution across the cytoplasmic membrane (accumulated versus secreted folate). The overproduction of pABA leads to relatively low intracellular folate pools and a relatively high secretion of folate. An explanation for this phenomenon is that elevated pABA pools might inhibit the activity of the enzyme responsible for the elongation of the polyglutamate tail of the folate molecule folylpolyglutamate synthetase. It is known that monoglutamate THF molecules diffuse more readily out of the cell than the more highly charged polyglutamate THF derivates (19, 22, 34). In L. lactis, it was found that overexpression of the gene coding for folylpolyglutamate synthetase (folC) resulted in increased polyglutamate tails, which caused higher intracellular retention (42).

The observation that folate production was already restored in the pABA KO strain without nisin induction suggests that the inducible nisin promoter in the vectors pNZ7020 and pNZ7023 exhibited a low but significant activity in the absence of its inducer molecule (nisin). As a result, the wild-type phenotype is restored in the absence of the inducer nisin. The use of high- and intermediate-copy-number plasmids, such as pNZ7020 and pNZ7023, respectively, can amplify the effect of low promoter activity, resulting in the restoration of the wild-type phenotype.

This study shows that the activities of the pathways for folate and pABA biosynthesis in L. lactis are tightly correlated. First, the deletion of the pABA genes in L. lactis eliminated its ability to synthesize folate, causing a complete inability to grow in the absence of purine nucleobases/nucleosides. In the presence of purine nucleobases/nucleosides, folate is not required for growth. Furthermore, we have shown that folate is not strictly required for the pyrimidine biosynthesis pathway. The overproduction of folate or pABA alone did not result in a strain with increased folate pools. However, the combined overexpression of folate and pABA biosynthesis pathways led to a strain that produces a high folate concentration and that does not rely on the supplementation of precursors in the medium. This strongly suggests that in the wild-type cells, the production of pABA and 2-amino-4-hydroxy-6-hydroxymethyl-7,8-dihydropteridine (the intermediate of the folate biosynthesis pathway at the pABA junction) are tightly correlated. In fact, both pathways need to be overexpressed simultaneously to increase total folate pools 80-fold in the absence of an external pABA source.