Functional Analysis of Sinorhizobium meliloti Genes Involved in Biotin Synthesis and Transport

Abstract

External biotin greatly stimulates bacterial growth and alfalfa root colonization by Sinorhizobium meliloti strain 1021. Several genes involved in responses to plant-derived biotin have been identified in this bacterium, but no genes required for biotin transport are known, and not all loci required for biotin synthesis have been assigned. Searches of the S. meliloti genome database in combination with complementation tests of Escherichia coli biotin auxotrophs indicate that biotin synthesis probably is limited in S. meliloti 1021 by the poor functioning or complete absence of several key genes. Although several open reading frames with significant similarities to genes required for synthesis of biotin in gram-positive and gram-negative bacteria were found, only bioB, bioF, and bioH were demonstrably functional in complementation tests with known E. coli mutants. No sequence or complementation evidence was found for bioA, bioC, bioD, or bioZ. In contrast to other microorganisms, the S. meliloti bioB and bioF genes are not localized in a biotin synthesis operon, but bioB is cotranscribed with two genes coding for ABC transporter-like proteins, designated here bioM and bioN. Mutations in bioM and bioN eliminated growth on alfalfa roots and reduced bacterial capacity to maintain normal intracellular levels of biotin. Taken together, these data suggest that S. meliloti normally grows on exogenous biotin using bioM and bioN to conserve biotin assimilated from external sources.

Microorganisms from the gram-negative genera Rhizobium, Sinorhizobium, Bradyrhizobium, Mesorhizobium, and Azorhizobium, collectively termed rhizobia, are well known for their capacity to establish N₂-fixing symbioses with legume plants (2). Although the molecular basis of rhizobial N₂ reduction is defined (17) and a foundation has been constructed for understanding other bacterial genes expressed in the plant (25), our knowledge of how rhizobia grow on plant roots is less complete. Good growth of S. meliloti strain 1021 (Rm1021) on alfalfa roots requires external biotin (33), often supplied by plant roots (27), which regulates a gene, bioS, that helps S. meliloti compete under such conditions (13, 34). Whether biotin is essential or simply stimulatory for rhizobial growth has been long debated (36-38), but clearly, cell densities of Rm1021 and many other rhizobia under biotin-limiting conditions are increased greatly by small amounts of biotin (8). Growing S. meliloti serially under biotin-limited conditions produces several physiological and metabolic changes, including the accumulation of polyhydroxybutyrate and a significant reduction in cell size (8, 14). Biotin-dependent enzymes such as pyruvate carboxylase are also affected under biotin-limited conditions, and several tricarboxylic acid cycle auxiliary enzymes show decreased activities (4, 5).

Biotin is formed in bacteria by a well-defined pathway (Fig. 1) (6, 7, 15, 16, 23, 26). Many microorganisms, including Mesorhizobium loti, have biotin synthesis genes organized in operons (1, 18, 19, 21, 28, 35), but neither the organization of bio genes in S. meliloti nor the molecular basis of this organism's dependence on external biotin has been defined.

FIG. 1.

Biotin synthesis pathways in gram-positive and gram-negative bacteria, with emphasis on rhizobia. M. loti R7A carries a bioZ gene for pimeloyl-CoA synthesis, but no bioC and bioH genes have been identified (35). The three S. meliloti bio genes shown to be functional are underlined. No possible bioD gene was detected in BlastX and BlastP searches. In gram-positive bacteria, the first steps of biotin biosynthesis require the bioW and bioI genes, and in gram-negative microbes, the bioC and bioH genes are required for this step. The last step requires bioB in gram-negative microbes and bioY in gram-positive microbes. DAPA, 7,8-diaminopelargonic acid; KAPA, 7-keto-8-aminopelargic acid.

The present study therefore was initiated to answer two questions. First, do new genomic data (10; http://sequence.toulouse.inra.fr/rhime/public/Access/RhimeFormRA.html) support the concept that S. meliloti is a biotin auxotroph? Second, does S. meliloti have a biotin transport system that explains how it can depend so strongly on external sources of this vitamin? Our results suggest that biotin synthesis in S. meliloti is limited because bioA, bioC, and bioD are either absent or function only at very low levels and that two genes, bioM and bioN, which are cotranscribed with bioB, may be involved in biotin transport.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Microbiological materials used in the present work are listed in Table 1. Escherichia coli was grown at 37°C on Luria-Bertani (LB) or M9 medium (29) supplemented with appropriate antibiotics, and S. meliloti 1021 was grown in GTS medium (20). Tests for rhizosphere growth and colonization of alfalfa roots were done as reported previously (33).

TABLE 1.

Biological materials

Strain or plasmid	Relevant characteristics	Sourcea or reference
E. coli
ATCC 33767	K-12ΔH1 derivative carrying a deletion in the bio region, Δ(bio-uvrB)	ATCC
DH5α	recA1 ΔlacZ	Gibco-BRL
S17-1	Modified RP4 plasmid integrated into genome	31
R872	E. coli bioF3 mutant	3
R875	E. coli bioB17 mutant	3
R877	E. coli bioD19 mutant	3
R878	E. coli bioC18 mutant	3
R879	E. coli bioA24 mutant	3
BM7086	E. coli bioH mutant Δ(mal-bioH) gal	12
S. meliloti
Rm1021	S. meliloti 1021 parent strain, Smr	24
Rm1021-B3	Rm1021 carrying Tn5-B30 in Y00962, Tetr Neor, requires biotin	33
Rm1021-B6	Rm1021 carrying a Tn5-B30 in Y00963, Tetr Neor, requires biotin	33
M. loti Gö1	Wild type	This study
Lactobacillus plantarum ATCC 8014	Biotin auxotroph	DSMZ
Plasmids, cosmids, etc.
pBBR1MCS2	Broad-host-range cloning vector, Kmr	22
pBSK+	pBluescriptSK+ multicopy cloning vector, Ampr	Stratagene
pBKS+	pBluescriptKS+ multicopy cloning vector, Ampr	Stratagene
pHSG399	Low-copy-number cloning vector, Cmr	11
pCosHE2	pWE15 containing a 30-kb insert with bio genes from a microbial consortium	9
pbioB-MLGö1	M. loti Gö1 bioD in pBBR1MCS2 under Ptac in the EcoRV site	This study
pbioBD-HE2-trp	pCosHE2 bioBD in pBBR1MCS2 under control of Ptrp	This study
pbioBD-HE2-nodD3	pCosHE2 bioBD in pBBR1MCS2 under control of PnodD3	This study
pbioABCFD	pBBR1MCS2 carrying the pCosHE2 bio operon as an EcoRI fragment	This study
pY20277-Rm1021	Rm1021 ORF Y20277 in pBSK+ in lacZ direction of transcription	This study
pY1426-Rm1021	Rm1021 ORF Y1426 possible bioC in pBSK+ in lacZ direction of transcription	This study
pbioF-Rm1021b	Rm1021 ORF Y01565 in pHSG399 in lacZ direction of transcription	This study
ppcaD-Rm1021b	Rm1021 ORF pcaD in pHSG399 in lacZ direction of transcription	This study
pfabH-Rm1021	Rm1021 ORF fabH in pBKS+ in lacZ direction of transcription	This study
Tn5-B30	pSup102 with transposable nptII promoter probe, Tetr	32

^aATCC, American Type Culture Collection; DSMZ, Deutsche Sammlung von Mikroorganismen and Zellkulturen.

Construction of bio clones and complementation tests.

DNA was cloned with standard methods (29), and DNA-modifying enzymes were used as specified by the manufacturer. To amplify the different open reading frames (ORFs) from S. meliloti, M. loti, Agrobacterium tumefaciens, and pCosHE2, PCRs with standard conditions (29) were employed. Complementation tests in E. coli biotin auxotrophs were initiated in liquid medium, and then possible transformants carrying the various bio constructs were selected on M9 agar medium supplemented with avidin (0.056 U/ml) and vitamin-free Casamino Acids (0.2%, wt/vol; Difco). If required, additional amino acids were supplied at 5 to 10 mM. For complementation studies with Rm1021, cells were grown in GTS liquid medium lacking biotin and rinsed twice with fresh medium before generating and selecting the transformants as specified for E. coli.

Biotin measurements in culture supernatants.

Biotin was measured by Lactobacillus growth and with a competitive enzyme-linked immunosorbent assay (ELISA) (9). For the latter, microtiter plates were coated for 2 h with anti-rabbit immunoglobulin G diluted 1:5,000 in phosphate-buffered saline (PBS), washed and treated for 1 h with blocking solution (vitamin-free casein), and washed with PBS. Extravidin-alkaline phosphatase conjugate (Sigma, Heidelberg, Germany) was added to the plates in a 1:20,000 dilution in PBS-Tween (0.025%, vol/vol); then 100-μl aliquots of samples were added to the wells, and plates were incubated at 37°C for 30 min. Serial dilutions of the various samples were performed prior to the tests, and a standard with known biotin concentrations was included in each microtiter plate. After repeated washing of the microtiter plates with PBS, substrate buffer was added (100 mM Tris-HCl, 100 mM NaCl, 50 mM MgCl₂, 2 mg of para-nitrophenylphosphate per ml, pH 9.5), and the color development was recorded at 405 nm in a microplate reader.

Biotin uptake studies.

Rhizobial cells were grown to late log phase in GTS medium containing 40 nM [¹²C]biotin, centrifuged, and washed twice in fresh medium lacking biotin. Cells were then depleted of biotin for 2 h by shaking at 150 rpm in fresh medium at 30°C. Biotin uptake by the depleted cells was measured with [¹⁴C]biotin (34). To determine the affinity (K_m) and velocity (V_max) of S. meliloti, biotin uptake data were collected after incubating biotin-depleted bacteria for 2 min with either [¹⁴C]biotin or [³H]biotin.

Nucleotide sequence accession numbers.

The M. loti Gö1 bioB and bioD genes were submitted to GenBank (accession numbers AF416911 and AF416912), and functionality of the S. meliloti bio genes (e.g., bioB and bioF) was communicated to the S. meliloti genome server.

RESULTS AND DISCUSSION

Identification and functional tests of S. meliloti biotin synthesis genes.

Results from this study show that S. meliloti 1021 clearly has part of the pathway required for biotin synthesis. Detailed Blast searches of the S. meliloti genome database (http://sequence.toulouse.inra.fr/rhime/public/Access/RhimeFormRA.html) identified seven ORFs with significant similarities to bio genes in other microorganisms (Table 2). The bioF gene was highly similar to the corresponding protein from the gram-positive bacterium Kurthia sp. S. meliloti bioB was similar to the gram-positive Deinococcus radiodurans biotin synthase gene, and only the other putative bio genes showed high similarities to bio genes in closely related gram-negative bacteria. Observed similarities and identities for putative S. meliloti bio genes with corresponding genes derived from other microbes were significant but highly variable. The overall identities observed ranged from 73% for the putative bioZ to only 15% for the possible bioC. Surprisingly, no bioD gene was found in BlastX and BlastP searches using homologues from a diverse range of other microbes, and only the ORF identified as possible bioB was organized in an operon (Fig. 2 and Table 2).

TABLE 2.

S. meliloti strain 1021 ORFs with highest significant similarities to known bio genes and their tested function in E. coli and RM1021 biotin biosynthesis^a

S. meliloti ORF	Length (amino acids)	% Identity/% similarity/E valueb by Blast	Organism	Putative function in S. meliloti biotin biosynthesisc	Functionality
Y20277	447	143/217/2e-54 (430)	Bacillus sphaericus	DAPA amino transferase, bioA	No complementation of E. coli R879
Y00964	187	63/84/2e-14 (187)	Deinococcus radiodurans	Biotin synthase, bioY	Desthiobiotin utilization in Rm1021 but not in mutants B3 and B6
Y01426	242	38/53/9e-09 (93)	Escherichia coli	Synthesis of pimely1-CoA, bioC	No complementation of E. coli R878
Y01565	395	136/219/4e-65 (385)	Kurthia sp.	KAPA synthase, bioF	Complements E. coli R872
pcaD	268	76/117/3e-16 (254)	Kurthia sp.	Pimeloyl-CoA synthesis, bioH	No complementation of E. coli BM7086
Y01812	415	123/223/3e-48 (404)	Bacillus subtilis	Pimelic acid synthesis, bioI	Not tested
fabH	323	236/261/e-127 (323)	Mesorhizobium loti	3-Oxoacyl-(acyl-carrier protein) synthase, bioZ	No complementation of E. coli BM7086

^abio genes were tested for complementation (Table 1) on agar and liquid M9 medium.

^bNumbers of amino acids used to determine percent identities and similarities are indicated in parentheses.

^cDAPA, 7,8-diamino pelargic acid; KAPA, 7-keto-8-aminopelargic acid.

FIG. 2.

Physical organization of the two functional S. meliloti bioB and bioF genes. (A) Physical organization of bioM and bioN in S. meliloti. The bioB gene, which codes for the final step in biotin synthesis (Fig. 1), is cotranscribed with bioM and bioN, which affect retention of biotin in the cell. ORF numbers were assigned by the S. meliloti genome sequencing project. ORF Y00960 shows similarities to TldD proteins; ORF Y00961 is similar to the E. coli YcaQ protein; ORF Y00965 is similar to bacterial long-chain fatty acid ligases; and ORF Y00965 is similar to acetyl-CoA acetyl transferases. (B) Physical organization of bioF in S. meliloti. The bioF gene is involved in condensation of pimeloyl-CoA (Fig. 1). The S. meliloti bioF is flanked by genes involved in DNA recombination (recQ), a threonine dehydrogenase (tdh), and a sigA gene.

Genetic and biochemical tests of putative ORFs established the functionality of several bio genes in S. meliloti (Table 2). When S. meliloti bioF was expressed under control of the tac promoter, it conferred growth on E. coli R872. Unexpectedly, ORF Y20277, which showed significant similarities to various bioA genes, did not restore growth in the E. coli bioA mutant, nor did other possible bio gene homologues of S. meliloti restore growth of the corresponding E. coli bioH and bioC mutants (Table 2). In further tests, the parent strain Rm1021 was complemented with a complete bio operon by using a plasmid carrying bioABCFD, and this transconjugant grew as a clear biotin prototroph (data not shown). This observation suggests that Rm1021 carries a bioH homologue which is required for the synthesis of pimeloyl coenzyme A (CoA). Because the ORF identified as a putative bioH in Rm1021 did not complement the corresponding E. coli mutant, it is possible that a different, as yet unidentified gene is involved in this early step in rhizobial biotin biosynthesis.

Transconjugants of Rm1021 containing extra copies of bioB showed better growth in the presence of desthiobiotin than wild-type Rm1021 in the same medium (data not shown). Thus, bioB gene transcription or copy number limits biotin production in Rm1021 cells supplied with desthiobiotin. Interestingly, while exogenous desthiobiotin promoted growth of the Rm1021 parent, this compound had no effect on growth of either Rm1021-B3 or Rm1021-B6. The same mutants, however, showed excellent growth in the presence of desthiobiotin when extra copies of bioB were present (data not shown). These results clearly suggest that bioB is not transcribed in these two mutants. Altogether, the data obtained from Blast searches, complementation tests, and precursor feeding studies suggested that the bioB, bioH, and bioF genes and their products are functional in S. meliloti, but the data also suggested that biotin synthesis is limited in S. meliloti because several other key genes (bioA, bioC, and bioD) are absent or nonfunctional.

Data presented here add new details to the long-standing debate over biotin auxotrophy in rhizobia. Wilson and Wilson (38) purified medium components to remove biotin and then demonstrated slow growth in the absence of added biotin directly by viable-cell counts through 16 transfers for S. meliloti and 40 transfers for Rhizobium leguminosarum biovar trifolii. Watson and coworkers (36) purified medium components to remove biotin and then demonstrated no significant increase in the optical density of 27 S. meliloti strains under biotin-limited conditions. Both studies contributed other information, but the dominant point of each report was that biotin stimulated growth. We concur in that conclusion and offer evidence here that the problem for S. meliloti 1021 lies in the absence of sufficient BioA, BioC, and BioD activity. This conclusion differs in detail, but not in overall effect, from the suggestion that S. meliloti lacks the entire known biotin biosynthetic pathway (36).

Two novel genes linked to S. meliloti bioB.

Further analysis of the mutations in Rm1021-B3 and Rm102-B6 indicated their close relationship to bioB. Nucleotide sequencing showed that the transposon in Rm1021-B3 and Rm1021-B6 inserted immediately upstream of the bioB gene (Fig. 2) and interrupted two adjacent ORFs (Y00962 and Y00963), which are designated here bioM and bioN, respectively, because their mutation eliminates growth under biotin-limited conditions. Blast analyses suggest that bioM is similar to cbiO, an ATP-binding protein in an ABC transporter from D. radiodurans, but bioN shows no significant homology to known proteins or genes. Weak similarities (>0.001%) hint that bioN may also be part of an ABC transporter, and both bioM and bioN show weak relationships to cobalt transporters in other microbes (data not shown).

The failure of Rm1021-B3 and Rm1021-B6 to grow on desthiobiotin suggests that bioB is cotranscribed with the mutated genes bioM and bioN. This assumption was validated by demonstrating that bioB from M. loti Gö1 in the self-replicating pbioB-MLGö1 stimulates growth of both Rm1021-B3 and Rm1021-B6 in the presence of exogenous desthiobiotin (data not shown). Additional growth studies, however, suggested that the lack of bioB in the original Rm1021-B3 and Rm1021-B6 mutants was only of minor importance for the observed growth phenotypes. This conclusion is based on the fact that Rm1021-B3/pbioB-MLGö1 and Rm1021-B6/pbioB-MLGö1 did not grow under biotin-limited conditions, while the parent strain grew under these conditions (data not shown).

The possibility that poor expression of the M. loti bioB in S. meliloti caused this problem was eliminated by fusing the heterologous bioB to different promoters, including Ptrp and P-nodD3, which are expressed in S. meliloti. In those experiments, constructs in Rm1021-B3 and Rm1021-B6 containing bioBD controlled by either a trp or nodD3 promoter did not restore growth under biotin-limited conditions (data not shown). These results with additional BioB protein in the cells support the conclusion that the growth phenotypes of Rm1021-B3 and Rm1021-B6 result primarily from the mutation in bioM and bioN rather than from any downstream effect on the cotranscribed copy of bioB.

The possible homology between bioM and genes involved in cobalt transport in other microbes encouraged us to test whether the mutation in Rm1021-B3 affects synthesis of the Co-containing vitamin B₁₂ or pathways directly requiring vitamin B₁₂, such as methionine synthesis. Supplementing cultures with vitamin B₁₂, additional cobalt, or methionine did not restore growth in Rm1021-B3 or Rm1021-B6 (data not shown). We conclude, therefore, that mutations affecting a cobalt transporter, vitamin B₁₂ synthesis, or a B₁₂-dependent metabolic activity are not responsible for the failure of these mutants to grow under biotin-limited conditions. Finally, no major influence on growth of the parent strain was observed in the defined medium when small amounts of B₁₂ or methionine were added.

BioM and BioN help retain cellular biotin.

Because growth of wild-type S. meliloti is so responsive to exogenous biotin (33, 36) and because transcription of bioM and bioN showed weak similarities to that of ABC transporters (30), we hypothesized that BioM or BioN protein may be involved in biotin transport. Tests using [¹⁴C]biotin showed that initial transport of biotin into biotin-starved cells of Rm1021 and Rm1021-B3 was identical over the first 8 min (Fig. 3). Measurements taken 2 to 8 min after exposing cells to [¹⁴C]biotin detected no difference in uptake between the parent Rm1021 and either of the two mutants, Rm1021-B3 or Rm1021-B6 (data not shown). More detailed analyses confirmed this finding and indicated that the affinities for biotin, i.e., the apparent K_m and the V_max, were identical in Rm1021 and Rm1021-B3 cells. A K_m of approximately 2.2 nM and a V_max for biotin of 4.7 pmol/min were measured for both Rm1021-B3 and the Rm1021 parent (data not shown). After an initial 8-min uptake period, however, Rm1021-B3 mutant cells retained less biotin than Rm1021 cells (Fig. 3). This marked drop in cellular biotin content, as measured by ¹⁴C, was observed in both Rm1021-B3 and Rm1021-B6 relative to the parent Rm1021 cells.

FIG. 3.

Biotin uptake by S. meliloti parent Rm1021 and mutant Rm1021-B3. Cells grown to late log phase were starved for 2 h and then supplied with [¹⁴C]biotin. Data represent mean values of at least three measurements, and data were highly reproducible.

Tests with recombinant strains supported the conclusion that mutants Rm1021-B3 and Rm1021-B6 retained less biotin than the parent Rm1021. Transconjugants of Rm1021, Rm1021-B3, and Rm1021-B6 containing a complete biotin synthesis operon in plasmid pbioABFCD released vastly different amounts of biotin during a 5-day growth period. In tests using a defined medium without added biotin, Rm1021/pbioABCFD excreted 2.4 ng of biotin/ml of culture supernatant, while Rm1021-B3/pbioABCFD and Rm1021-B6/pbioABCFD released 4.5 and 5.0 ng of biotin/ml of culture supernatant, respectively, in data from three experiments. Taken as a whole, these findings support the hypothesis that the mutations affect a system required to maintain constant levels of biotin in the bacterial cells.

Rhizosphere growth of Rm1021-B3 and Rm1021-B6.

Analyses in rhizosphere microcosms showed that the mutations in Rm1021-B3 (Fig. 4) and Rm1021-B6 (data not shown) abolished the capacity to colonize alfalfa roots (Fig. 4). When approximately 100 cells were inoculated onto the alfalfa seedlings, Rm1021 cells doubled more than nine times in 144 h to titers of 6 × 10⁴ cells per seedling. In the same test, populations of Rm1021-B3 cells declined from the initial inocula and then remained static, with no significant increase in cell numbers between day 1 and day 6. This phenotype is consistent with the importance of bioB in biotin synthesis and the apparent role of the bioM and bioN genes in retaining biotin in the cell after it is taken up from plant exudates. Interestingly, this phenotype was partially corrected when a complete bio operon was expressed in RM1021-B3 or -B6 (data not shown) (33).

FIG. 4.

Alfalfa root colonization by S. meliloti parent Rm1021 and mutant Rm1021-B3. Bacteria were recovered from the roots at various times after inoculation at time zero. Data points are the means ± standard deviation from at least nine replicate plants.

The discovery here of bioM and bioN offers preliminary insight into the molecular mechanisms associated with biotin transport. The significant similarity between BioM and proteins that are commonly found in ABC transporters (30) documents a genetic basis that complements our direct functional evidence from uptake experiments. The observed K_m for biotin transport in both Rm1021 and mutant Rm1021-B3 was approximately 2.2 nM biotin, which indicates a markedly higher affinity than the E. coli K_m of 140 nM for biotin (26). Such a high-affinity uptake system for biotin may offset the lack of a fully functional biotin biosynthesis pathway and help this organism persist in soil as a biotin auxotroph. No other genetic locus has been assigned to microbial biotin transport in any prokaryotic microorganism. Therefore, defining how these genes function and the other genes that contribute to sensing and using external biotin in S. meliloti remains a worthwhile future objective.