l-Hydroxyproline and d-Proline Catabolism in Sinorhizobium meliloti

ABSTRACT

Sinorhizobium meliloti forms N₂-fixing root nodules on alfalfa, and as a free-living bacterium, it can grow on a very broad range of substrates, including l-proline and several related compounds, such as proline betaine, trans-4-hydroxy-l-proline (trans-4-l-Hyp), and cis-4-hydroxy-d-proline (cis-4-d-Hyp). Fourteen hyp genes are induced upon growth of S. meliloti on trans-4-l-Hyp, and of those, hypMNPQ encodes an ABC-type trans-4-l-Hyp transporter and hypRE encodes an epimerase that converts trans-4-l-Hyp to cis-4-d-Hyp in the bacterial cytoplasm. Here, we present evidence that the HypO, HypD, and HypH proteins catalyze the remaining steps in which cis-4-d-Hyp is converted to α-ketoglutarate. The HypO protein functions as a d-amino acid dehydrogenase, converting cis-4-d-Hyp to Δ¹-pyrroline-4-hydroxy-2-carboxylate, which is deaminated by HypD to α-ketoglutarate semialdehyde and then converted to α-ketoglutarate by HypH. The crystal structure of HypD revealed it to be a member of the N-acetylneuraminate lyase subfamily of the (α/β)₈ protein family and is consistent with the known enzymatic mechanism for other members of the group. It was also shown that S. meliloti can catabolize d-proline as both a carbon and a nitrogen source, that d-proline can complement l-proline auxotrophy, and that the catabolism of d-proline is dependent on the hyp cluster. Transport of d-proline involves the HypMNPQ transporter, following which d-proline is converted to Δ¹-pyrroline-2-carboxylate (P2C) largely via HypO. The P2C is converted to l-proline through the NADPH-dependent reduction of P2C by the previously uncharacterized HypS protein. Thus, overall, we have now completed detailed genetic and/or biochemical characterization of 9 of the 14 hyp genes.

IMPORTANCE Hydroxyproline is abundant in proteins in animal and plant tissues and serves as a carbon and a nitrogen source for bacteria in diverse environments, including the rhizosphere, compost, and the mammalian gut. While the main biochemical features of bacterial hydroxyproline catabolism were elucidated in the 1960s, the genetic and molecular details have only recently been determined. Elucidating the genetics of hydroxyproline catabolism will aid in the annotation of these genes in other genomes and metagenomic libraries. This will facilitate an improved understanding of the importance of this pathway and may assist in determining the prevalence of hydroxyproline in a particular environment.

INTRODUCTION

4-Hydroxyproline and 3-hydroxyproline in animal and plant proteins are formed through the posttranslational modification of proline residues by the enzyme proline hydroxylase. In some bacteria, hydroxyproline is synthesized from free l-proline, where it is employed in the synthesis of secondary metabolites (1, 2). 4-Hydroxyproline has two chiral carbon atoms, and of its four isomeric forms, trans-4-hydroxy-l-proline (trans-4-l-Hyp) and cis-4-hydroxy-d-proline (cis-4-d-Hyp) appear to be the two most common isomers. trans-4-l-Hyp is one of the most abundant amino acids in animals, where it is a major constituent of collagen (3). In plants and algae, several abundant cell wall proteins and glycoproteins are rich in hydroxyproline (4, 5). Other proline derivatives, such as hydroxyproline betaine and proline betaine (N,N-dimethylproline, stachydrine), are present in alfalfa and some other plants, and these are generally associated with osmotic adaptation (6–8).

Hydroxyprolines represent rich sources of nitrogen and carbon for microorganisms in soil and other environments containing decomposing biological material (9, 10). Microorganisms that utilize hydroxyproline for growth have been isolated from soil, and a pathway for the catabolism of trans-4-l-Hyp by Pseudomonas was elucidated by Adams and coworkers in the 1960s (see reference 11 and references therein). In this pathway, trans-4-l-Hyp is epimerized to cis-4-d-Hyp by hydroxyproline 2-epimerase. The cis-4-d-Hyp is then oxidized to Δ¹-pyrroline-4-hydroxy-2-carboxylate (HPC) by a d-hydroxyproline dehydrogenase enzyme containing flavin cofactors (12). This enzyme was previously referred to as cis-4-d-Hyp oxidase (11) (Fig. 1). HPC is subsequently deaminated to α-ketoglutarate semialdehyde (α-KGSA) by Δ¹-pyrroline-4-hydroxy-2-carboxylate deaminase, and in the final step, α-KGSA is oxidized by α-KGSA dehydrogenase to the central tricarboxylic acid cycle intermediate α-ketoglutarate (α-KG). Until recently, there have been few molecular analyses of the underlying genes and enzymes involved in the hydroxyproline catabolic pathway (12,−,14). Hence, the genes coding for hydroxyproline transport and metabolism are poorly annotated in bacterial genomes. The abundance of genome, metagenome, and transcriptome sequences is increasing the utility and insights from having well-defined gene-function relationships for catabolic pathways. Thus, elucidating the gene-function relationships for metabolic pathways is of broad importance.

FIG 1.

Genetics and biochemistry of hydroxyproline metabolism in S. meliloti. (A) Diagram of the hydroxyproline transport and catabolic locus; (B) schematic diagram of the hydroxyproline catabolic pathway, as described by Adams and Frank (11). (A and B) Gene annotations, promoters, and enzymatic functions as deduced through this study and previous work (13, 14, 18). HypR, negative regulator; HypD, Δ¹-pyrroline-4-hydroxy-2-carboxylate deaminase; HypT, unknown; HypS, Δ¹-pyrroline-2-carboxylate reductase; HypH, α-ketoglutarate semialdehyde dehydrogenase; HypMNPQ, l-hydroxyproline ABC-type transport system; HypO, cis-4-hydroxy-d-proline dehydrogenase; HypRE, hydroxyproline 2-epimerase; HypX, unknown; HypY, unknown, a possible proline racemase pseudogene; HypZ, unknown.

Sinorhizobium meliloti grows as a free-living bacterium in soil, and it also forms N₂-fixing root nodules on alfalfa. This bacterium grows rapidly on 4-hydroxyproline as a sole carbon and nitrogen source, and in earlier work, we identified a cluster of 14 hyp genes, located on the S. meliloti pSymB chromid, whose expression is induced by trans-4-l-Hyp (13, 14). The trans-4-l-Hyp catabolic pathway in S. meliloti appears to be the same as that originally characterized in Pseudomonas (11). In S. meliloti, the hyp gene cluster includes five operons (Fig. 1), and transcription of each is repressed by the negative regulator, HypR. Repression is relieved by trans-4-l-Hyp and more strongly by cis-4-d-Hyp, the first catabolite (14). Uptake of trans-4-l-Hyp and cis-4-d-proline is mediated by the HypMNPQ ABC-type transport system (13), and the first step in the catabolic pathway (epimerization) is performed by the hypRE gene product (14). The hypO, hypD, and hypH genes were predicted to encode cis-4-d-Hyp dehydrogenase, (HPC) deaminase, and α-KGSA dehydrogenase, respectively, but were not biochemically illustrated (14).

Watanabe and coworkers recently identified and characterized the d-hydroxyproline dehydrogenase and HPC deaminase enzymes from Pseudomonas putida and Pseudomonas aeruginosa (12). The d-hydroxyproline dehydrogenase enzymes from these organisms appear to have arisen via convergent evolution, where the P. aeruginosa enzyme contained three different subunits while the P. putida enzyme was a single-subunit protein (12).

In the present report, we present genetic and biochemical data that confirm the assignments of the HypD and HypH proteins and the involvement of the HypO protein in hydroxyproline catabolism. We present the high-resolution structure of HypD, which shows strong similarity to members of the N-acetylneuraminate lyase (NAL) enzyme subfamily of (α/β)₈ barrel proteins, most significantly at the active site (12, 15). We also show that S. meliloti can catabolize d-proline as a carbon and nitrogen substrate and that the hyp gene cluster is involved in the catabolic pathway. We characterize the hypS gene and show that it encodes a pyrroline-2-carboxylate reductase that reduces pyrroline-2-carboxylate to l-proline. This activity is required for the growth of S. meliloti on d-proline but not on trans-4-l-Hyp or cis-4-d-Hyp.

MATERIALS AND METHODS

Bacterial strains, plasmids, and mutant construction.

All S. meliloti strains are derived from RmP110, which is Rm1021 carrying a wild-type pstC gene (16). The nonpolar deletions of the hypD, hypS, hypO, and hypH genes were constructed by replacing the coding regions with the FLP recombination target (FRT)-Kan^r-FRT cassette using λ Red recombinase (17) as described for the ΔhypRE mutant (14). In all cases, the target gene together with approximately 300 bp of flanking DNA was PCR amplified and cloned into the Gm^r suicide plasmid, pUCP30T. The gene cloned in pUCP30T was replaced by a 1.4-kb FRT-Kan^r-FRT PCR product using pKD13 as the template and oligonucleotides whose 5′ ends were 50-bp sequences immediately 5′ and 3′ to the gene to be deleted. The resulting plasmid was then integrated into the genome of RmP110 to generate Nm^r Gm^s double-crossover recombinants in which the target gene was replaced by the FRT-Kan^r-FRT cassette. The kan gene was then removed using Flp recombinase (supplied on vector pTH2505) to generate nonpolar deletion mutants of RmP110 (14). Plasmid pTH2505 is unstable in S. meliloti and was easily cured. The final constructs were checked by PCR and sequencing across the FRT junctions. The mutant strains were RmP2506 ΔhypO (Δsmb20267), RmP2510 ΔhypD (Δsmb20259), RmP2514 ΔhypS (Δsmb20261), and RmP2516 ΔhypH (Δsmb20262). The araE ΔhypH double mutant (RmP3174) was made by transducing an araE::Tn5-B20 mutation (18) into RmP2516. To construct the ΔhypS hypS⁺ merodiploid strain RmP3272, plasmid pTH2685, containing hypS (smb20261) plus 300 nucleotides (nt) upstream and downstream in pUCP30T, was transferred into RmP2514, and Gm^r single-crossover recombinants were selected. The proline auxotroph RmP3155 was constructed by transducing ΔB161 (an ∼53-kb deletion of the pSymB plasmid that removes the smb20003 gene; Nm^r Gm^r) into a proC::Tn5-B20 strain (19, 20). The putA mutant SmFL5502 was identified in our previously constructed pTH1522 reporter gene fusion library of RmP110 (21). Reporter gene fusion strains and β-glucuronidase (GusA) activity were measured, as previously described (13, 14).

Growth studies.

To test for the ability of S. meliloti strains to utilize and grow with various compounds as sole carbon sources, strains were grown in M9 mineral salts medium containing 48 mM Na₂HPO₄, 22 mM NaH₂PO₄, 8.6 mM NaCl, 18.6 mM NH₄Cl, 1 mM MgSO₄, 0.25 mM CaCl₂, 0.005 μg/ml biotin, 10 ng/ml CoCl₂, and a carbon source at 10 or 15 mM. When we tested for growth on nitrogen sources, NH₄Cl was omitted from the M9 growth medium. For growth curves, inoculum cultures grown overnight in LBmc (19) were washed with 0.85% NaCl and inoculated into each test medium to an optical density at 600 nm (OD₆₀₀) of 0.05. The growth profiles of 0.15-ml cultures were measured for 48 h with shaking in 96-well microtiter plates at 30°C, and the data were analyzed as previously described (22). The OD₆₀₀ values presented are not corrected for path length, and unless stated otherwise, generation times were calculated between uncorrected OD₆₀₀ values of 0.1 and 0.3.

Overexpression and purification of recombinant proteins HypD, HypH, and HypS.

For overexpression of Hyp proteins with an N-terminal His₆ tag in Escherichia coli, the hyp genes were each cloned into pET28a and introduced into E. coli strain BL21/DE3. For overexpression, each strain was cultured at 30°C with aeration to the mid-log phase in LB medium with 25 μg/ml kanamycin. Expression of the recombinant protein was then induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG) to 0.5 mM, and the cultures were incubated for an additional 4 h. The cultures were then cooled on ice, and the cells were harvested and resuspended in buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, and 7.5% glycerol. The cells were disrupted by a French press, and the lysates were cleared by centrifugation at 24,000 × g for 45 min.

Each of the recombinant proteins was purified to 95% purity with BD Talon cobalt resin, using batch elution as described in the manufacturer's instructions. The purified protein was stored at −80°C in buffer containing 20 mM HEPES (pH 7.5), 150 mM KCl, 1 mM dithiothreitol (DTT), 1 mM EDTA, and 10% glycerol. The protein concentrations were determined using Bio-Rad protein assay reagent (Bio-Rad) and bovine serum albumin as a standard.

HPC synthesis and HPC deaminase assays.

As Δ¹-pyrroline-4-hydroxy-2-carboxylate (HPC) is an unstable compound that is not commercially available, all of the HPC substrate used in reactions was enzymatically synthesized and used immediately upon preparation (23, 24). HPC was synthesized from cis-4-d-Hyp in reaction mixtures that contained cis-4-d-Hyp and membrane preparations containing either S. meliloti HypO or purified cis-4-d-Hyp dehydrogenase from P. aeruginosa or P. putida as described by Watanabe et al. (12).

The purified S. meliloti HypD protein carrying an N-terminal His₆ tag was assayed for HPC deaminase activity by coupling its reaction to an α-ketoglutarate semialdehyde dehydrogenase (α-KGSADH) and monitoring the increased absorbance at 340 nm. The α-KGSADH enzyme used in these reactions was either purified HypH from S. meliloti or purified α-KGSADH protein from Acinetobacter baylyi (25). Similar deaminase activities were detected using either enzyme. The deaminase activity was also monitored by coupling the reaction to a purified glutamate dehydrogenase enzyme (α-KG + NH₃ + NADH + H⁺ ⇔ glutamate + NAD⁺) and detecting the production of ammonia. However, the detection of NH₃ formation was much less sensitive than the detection of deaminase activity using the α-KGSADH assay.

P2C synthesis and P2C reductase assays.

The Δ¹-pyrroline-2-carboxylate (P2C) used in the P2C reductase reactions was synthesized from d-proline with the enzyme d-amino acid oxidase as described by Visser et al. (26). The reaction mixture consisted of 10 mmol d-proline, 0.25 mg porcine d-amino acid oxidase (Sigma-Aldrich), and 1,000 to 2,500 units bovine catalase (Sigma-Aldrich) in 25 mM ammonium bicarbonate buffer (pH 8.3), and the reaction was allowed to proceed overnight at 37°C. Electrospray mass spectroscopic (MS) analysis of the reaction mixtures before and after incubation showed the complete conversion of the d-proline to Δ¹-pyrroline-2-carboxylate (data not shown).

The HypS P2C reductase activity (27) was quantified at room temperature by measuring the decrease in absorbance at 340 nm in a 1-ml reaction mixture containing 13 μg HypS, 0.5 mM NADPH, and various concentrations of P2C in a 25 mM ammonium bicarbonate buffer (pH 8.3). The reverse reaction was quantified by measuring the increase in absorbance at 340 nm in a 1-ml reaction consisting of 100 mM Tris-HCl (pH 10.0), 13 μg HypS, 1 mM NADP⁺, and various concentrations of l-proline.

α-Ketoglutarate semialdehyde dehydrogenase enzyme kinetics.

α-KGSADH activity was quantified at room temperature by measuring the increase in absorbance at 340 nm in 1-ml reaction mixtures containing 50 mM HEPES buffer (pH 7.5), 100 mM NaCl, 0.5 mM MgCl₂, 2 to 3 μg HypH enzyme, α-KGSA, and NAD(P)⁺. The K_m value for α-KGSA was determined using various concentrations of α-KGSA (4.45, 11.1, 17.8, 22.3, 35.6, 44.5, 89.0, 133.6, and 178.1 μM) and a constant concentration of 0.5 mM NADP⁺. Kinetic measurements for NAD(P)⁺ coenzyme were performed with 178 μM α-KGSA and various concentrations of NADP⁺ (10, 25, 50, 100, 300, and 500 μM) or NAD⁺ (0.2, 0.5, 1.0, 2.5, 4.0, 5.0, and 8.0 mM). α-Ketoglutarate semialdehyde was synthesized from d-glucarate using the A. baylyi enzymes d-glucarate dehydratase (ACIAD0128) and d-5-keto-4-deoxyglucarate dehydratase (ACIAD0130), which were overexpressed from E. coli as described in the work of Aghaie et al. (25).

Crystallization and structure determination of HypD.

Purified HypD protein carrying an N-terminal His tag was crystallized at room temperature using the hanging-drop method, with 0.5 μl of 19.5 mg/ml protein solution (including 20% [wt/vol] glycerol) mixed with 0.5 μl of reservoir solution (0.2 M potassium-sodium tartrate, 20% [wt/vol] polyethylene glycol 3350 [PEG 3350]). The crystal was cryoprotected with Paratone oil. Diffraction data at 100 K were collected at the Structural Genomics Consortium (Toronto, ON, Canada) on a Rigaku FR-E SuperBright rotating copper anode source with a Rigaku R-Axis HTC detector. X-ray diffraction data were reduced with HKL-3000 (28), and evaluation of the reduced data quality by phenix.xtriage (29) revealed significant merohedral twinning no matter which space group was chosen, so the highest symmetry space group of P2₁3 was selected for refinement (L test for acentric data value of 0.385, with perfect twin of 0.375 and twin law of l, −k, h). The structure of HypD was solved by molecular replacement using Phenix.phaser with the structure of dihydrodipicolinate synthase DapD from Agrobacterium tumefaciens (PDB code 2HMC), the sequence of which is 88% identical to that of HypD, as the search model for two copies in the asymmetric unit. B-factors were refined as isotropic using translation-libration-screw rotation (TLS) parameterization (groups were chain A residues 23 to 44, 45 to 129, 130 to 216, 217 to 280, and 281 to 340 and chain B residues 22 to 40, 41 to 45, 46 to 132, 133 to 213, and 214 to 340). The final atomic mode of HypD included residues 23 to 340 and 22 to 340 of the two chains in the asymmetric unit. The average B-factor and bond angle/length root mean square deviation (RMSD) values were calculated using Phenix. All model geometries were verified using the Phenix and Coot (30) validation tools plus the wwPDB validation server (31). The structure has good backbone with percentages of residues of 92.1% in the most favored, 7.5% in the additional allowed, 0% in the generously allowed, and 0.4% in the disallowed regions of the Ramachandran plot (corresponding to Leu-128 of each protein chain, which is found at the noncrystallographic axis that is coincident with a putative protein-protein interface forming a HypD hexamer).

Structure similarity searches of the Protein Data Bank were performed using the PDBeFold and Dali servers (32, 33). Structure superpositions were performed with the Mustang algorithm (34). Protein-protein interfaces were determined using the PDBePISA server (35). Electrostatic potential surfaces were calculated using Chimera (36). The HypD active-site cavity was identified and analyzed using the CASTp server (37). PDB files were visualized with Jmol software (38).

The X-ray diffraction data collection statistics are listed in Table S1 in the supplemental material.

Protein structure accession number.

The structure factors and atomic coordinates for the HypD complex structures were deposited in the Protein Data Bank with the PDB code 5CZJ.

RESULTS

Growth phenotypes of S. meliloti hypD, hypH, hypO, and hypS mutants.

Previously in analyzing the S. meliloti hyp gene cluster, we showed that the hypMNPQ transport cassette and the hypRE epimerase gene are required for growth on hydroxyproline but that hypY (an epimerase-like pseudogene) is not (13, 14). To determine if the remaining predicted enzymes of the hyp gene cluster are required for growth on hydroxyproline, strains carrying nonpolar deletions of the hypO, hypD, hypS, and hypH genes were constructed, and the abilities of these strains to grow on various carbon sources were examined (Table 1). As all of the mutant strains grew at rates similar to that of the wild type in minimal medium with sucrose or l-proline as the carbon source, we conclude that the mutations appeared to have no effect on general carbon metabolism or l-proline catabolism (Table 1). With trans-4-l-Hyp as the sole source of carbon, the hypD mutant failed to grow, while the hypO and hypH mutants grew at about half the rate of the wild type, and, thus, these genes appear to play a role in trans-4-l-Hyp metabolism. The growth properties of the hypS mutant are discussed later.

TABLE 1.

Generation times of the ΔhypO, ΔhypD, ΔhypH, and ΔhypS mutants with various carbon sources

Strain	Result (h) with carbon sourcea:
	l-Hyp	l-Proline	Sucrose
RmP110 (wild type)	5.3 (0.6)	6.0 (0.1)	4.2 (0.2)
RmP2514 (ΔhypS)	6.2 (0.6)	5.9 (0.2)	4.3 (0.1)
RmP2506 (ΔhypO)	10.2 (0.3)	6.0 (0.3)	4.4 (0.1)
RmP2510 (ΔhypD)	No growth	6.2 (0.3)	4.3 (0.2)
RmP2516 (ΔhypH)	10.2 (0.1)	6.1 (0.1)	4.2 (0.1)

^aValues are the means from triplicate samples, with the standard deviations given in parentheses. Generation times were calculated between OD₆₀₀ values of 0.1 and 0.3. l-Hyp, trans-4-hydroxy-l-proline.

hypH encodes an α-KGSADH.

To confirm that hypH encodes an α-ketoglutarate semialdehyde dehydrogenase (α-KGSADH) enzyme (EC 1.2.1.4), the S. meliloti HypH protein was purified from E. coli as an N-terminal His₆-tagged protein and assayed for α-KGSADH activity. As the α-ketoglutarate semialdehyde (α-KGSA) substrate used for assaying α-KGSADH is not commercially available, it was synthesized from d-glucarate using purified d-glucarate dehydratase and 5-keto-4-deoxy-glucarate dehydratase enzymes (see Materials and Methods and reference 25). The purified recombinant HypH protein catalyzed the NAD(P)-dependent reduction of α-KGSA, and the kinetic behavior of the enzyme at various concentrations of α-KGSA or NADP⁺ or NAD⁺ showed typical Michaelis-Menten hyperbolic kinetics (Table 2). The enzyme showed a clear preference for NADP⁺ as a cofactor with a k_cat/K_m value for NADP⁺ that was 200-fold higher than that for NAD⁺ (Table 2). The substrate specificity of HypH was assessed using a standard reaction mix containing 0.5 mM NADP⁺ and 1 mM aldehyde substrate. Specific activities of 21.1 ± 2.6, 0.8 ± 0.1, 1.1 ± 0.2, and 1.7 ± 0.2 μmol mg⁻¹ min⁻¹ were measured for α-KGSA, propionaldehyde (C₃), valeraldehyde (C₅), and octylaldehyde (C₈), respectively. This suggests that HypH activity is fairly specific for α-KGSA as a substrate.

TABLE 2.

Kinetic parameters of the HypH α-ketoglutarate semialdehyde dehydrogenase enzyme

Parameter	Result fora:
	α-KGSA	NADP+	NAD+
Vmax (μmol−1 min−1 mg)	11	10	67
Km (μM)b	40	47	1,420
kcat (min−1)c	629	550	3,691
kcat/Km (μM−1 min−1)	16	12	0.06

^aValues are the means from triplicate assays, and similar results were obtained in two independent experiments.

^bThe K_m for α-KGSA was determined with a constant concentration of 0.5 mM NADP⁺. The K_m values for NADP⁺ and NAD⁺ were determined with a constant concentration of 178 μM α-KGSA.

^ck_cat = V_max [E]⁻¹.

The crystal structure of the S. meliloti HypH (Smb20262) protein was previously deposited in the Protein Data Bank (PDB code 3V4C); however, the specific role of the protein was not identified, and there is no associated publication. In addition, the amino acid sequence of HypH is 49% identical to that of an NADP⁺-dependent aldehyde dehydrogenase (VH-ADH) from Vibrio harveyi, whose structural features involved in cofactor specificity and catalysis have been examined (PDB codes 1EZO and 1EYY) (39–42). Like HypH (Table 2), VH-ADH employs NADP⁺ as a cofactor, and VH-ADH was shown to contain a cleft within which NAD(P)⁺ binds. The adenine ring of NADP⁺ interacts with the side chain and guanidium group of Arg-210 of VH-ADH (39), and in HypH, R-220 fulfills this role. In VH-ADH, there are extensive interactions between the 2′ phosphate of NADP⁺ and Lys-172, Thr-175, and Arg-210, and the equivalent residues in HypH are Lys-182, Ser-185, and Arg-220. The small size of Thr-175 in VH-ADH provides a pocket in which the phosphate group from NADP can bind, and its small size is retained by Ser-185 of HypH. These structural data confirm the designation of HypH as an NADP⁺-dependent α-KGSADH whose mechanism of cofactor specificity is conserved with that of VH-ADH.

hypD encodes an HPC deaminase.

The S. meliloti hypD gene is required for growth on trans-4-l-Hyp, and this phenotype was fully complemented by a plasmid carrying only the hypD gene (Table 1 and data not shown). We previously suggested that hypD encodes a Δ¹-pyrroline-4-hydroxy-2-carboxylate (HPC) deaminase that forms α-KGSA and NH₃ from HPC (14), and indeed, HypD shares 71% and 27% identity, respectively, with the HPC deaminase PP1257 protein from P. putida and the PA1254 protein from P. aeruginosa (12).

To verify that hypD encodes an HPC deaminase that converts HPC to α-KGSA and NH₃, the S. meliloti HypD protein was purified from E. coli as an N-terminal His₆-tagged protein and assayed for HPC deaminase activity. The HPC used in these assays was synthesized enzymatically from cis-4-d-Hyp, and the HPC deaminase activity was monitored by coupling its activity to α-KGSADH (see Materials and Methods). The purified HypD protein was found to catalyze the HPC-dependent formation of α-KGSADH with a specific activity of 1.6 μmol⁻¹ min⁻¹ mg⁻¹ protein. The HypD-dependent formation of NH₃ from HPC was also detected via a glutamate dehydrogenase enzyme, but this assay was much less sensitive than the α-KGSADH-coupled assay. Because of the instability of the HPC substrate, further kinetic analysis of the HypD protein was not performed.

The HypD HPC deaminase protein is a member of the (α/β)₈ barrel DapA family of proteins.

Sequence analysis revealed that HypD is a member of the N-acetylneuraminate lyase (NAL) subfamily of (α/β)₈ barrel proteins (EC 4.1.3.3). The enzymatic activities of a number of enzymes in this subfamily, including Escherichia coli dihydrodipicolinate synthase (DHDPS) (43), E. coli NAL (44), Pseudomonas putida trans-o-hydroxybenzylidenepyruvate hydratase-aldolase (HBPHA) (45), P. putida d-4-deoxy-5-oxoglucarate dehydratase (DOGDH) (46), and more recently P. putida Δ¹-pyrroline-4-hydroxy-2-carboxylate deaminase (LhpC), have been characterized (12). The HypD of S. meliloti characterized in this study has 31%, 26%, 26%, 27%, and 74% identities to the above-named enzymes, respectively (see Fig. S1 in the supplemental material). It is not surprising that HypD shows the highest identity to LhpC (74%) as they are homologous proteins that are involved in the same hydroxyproline catabolic pathway in S. meliloti and P. putida (12). The other proteins (DHDPS, NAL, HBPHA, and DOGDH) function in very different pathways: lysine biosynthesis, sialic acid concentration regulation, naphthalene degradation, and glucarate metabolism, respectively.

Although the DHDPS, NAL, HBPHA, and DOGDH enzymes function in disparate pathways, they all share a reaction mechanism with a covalent Schiff base intermediate that forms between the substrate (which varies, depending on the enzyme) and a completely conserved lysine residue in the active site (47). A tyrosine residue in the substrate binding pocket is also completely conserved and is predicted to be involved directly in the reaction mechanism or to function in stabilizing the position of the lysine (47, 48). The similarities of the sequences of HypD and LhpC to those of the other characterized members of this group and the fact that both have the conserved active-site lysine and tyrosine residues (see Fig. S1 in the supplemental material and reference 12) predict that the reaction mechanism of the bacterial HPC deaminase also involves a Schiff base intermediate.

The recombinant HypD protein was characterized in more detail through determination of the three-dimensional structure of the crystallized protein by X-ray crystallography. The asymmetric unit of the solved structure is two monomers, and putative protein-protein interfaces suggested the formation of a homotetramer. As predicted, each monomer has an (α/β)₈ barrel fold with eight β sheets and eight α helices alternating from the N terminus toward the C terminus (Fig. 2). Consistently with other solved structures of the NAL subfamily, E. coli DHDPS and E. coli NAL (43, 44), HypD has a C-terminal extension beyond the last alpha helix (α8) of the barrel fold. The structures of HypD and E. coli DHDPS are very close; the deviation between the aligned alpha carbons of the two structures is 2.44 Å (see Fig. S2 in the supplemental material). The C-terminal extension of HypD consists of 6 alpha helices (α9 to α14). Together, α10 and α11 correspond to α10 of DHDPS, which has an extension of 5 alpha helices rather than 6.

FIG 2.

Schematic representation of the HypD crystal structure. (A) View from the “side” of the (α/β)₈ barrel. The barrel motif is to the right of the image, and alpha helices 9, 10, 13, and 14 of the C-terminal extension are labeled (α9, α10, α13, and α14) for reference. The N terminus, starting at A3 in the refined structure, is labeled with an “N.” (B) View “down” the barrel from the C-terminal face into the active site. The conserved active-site residue K184 is shown. All alpha helices are labeled (α1 to α14), as are the beta sheets (β1 to β8) of the (α/β)₈ barrel motif. (A and B) Structures are colored from the N terminus (blue) to the C terminus (red).

In DHDPS and NAL, the active site is centered at the C-terminal face of the barrel (43, 44, 47). The lysine residue that is involved in Schiff base formation is extended in the substrate-binding pocket, with the conserved active-site tyrosine side chain just above it. In HypD, the positions of these residues, K164 and Y136, are extremely similar, with the lysine exposed for potential substrate binding (Fig. 2B). In the superposed structures of HypD and DHDPS, the deviation between K164 of HypD and the corresponding K161 of DHDPS (in both cases on β6) is 1.17Å (see Fig. S2 in the supplemental material). These results strongly suggest that the reaction mechanism of HypD includes a covalent Schiff base intermediate which forms between the substrate and K164 in the enzyme's active site.

hypS encodes a P2C reductase that is required for growth with d-proline as the carbon source.

The HypS protein is annotated as a malate/l-lactate dehydrogenase family protein; however, assays with the purified His₆-tagged HypS protein showed no reduction in NAD(P)⁺ with either malate or lactate as the substrate electron donor (data not shown). A screen for activity with other compounds identified l-proline as a substrate (Table 3). When Δ¹-pyrroline-2-carboxylate (P2C) was assayed as an electron acceptor, HypS was found to catalyze its NADPH-dependent reduction (Table 3). The pH optima for P2C reduction and l-proline oxidation were 10 and 7, respectively (data not shown). A kinetic analysis of P2C reduction and l-proline oxidation was performed (Table 3). The elevated k_cat/K_m with P2C as a substrate (2,914) compared to that of l-proline (0.87) showed that HypS has a clear preference toward the reduction of P2C to form l-proline (Table 3). This suggested that, under physiological conditions, HypS functions as a P2C reductase in the formation of l-proline.

TABLE 3.

Kinetic parameters of HypS in the reduction of Pyr2C and oxidation of l-proline

Parameter	Result fora:
	Pyr2C	l-Proline
Vmax (μmol−1 min−1 mg)	41	1.7
Km (μM)	0.89	74
kcat (min−1)b	1,469	61
kcat/Km (μM−1 min−1)	1,650	0.82

^aValues are the means from triplicate assays, and similar results were obtained in two independent experiments. For the pyrroline-2-carboxylate (Pyr2C) reduction and l-proline oxidation reactions, NADPH and NADP⁺ were present at concentrations of 0.5 mM and 1 mM, respectively.

^bk_cat = V_max [E]⁻¹.

Since the oxidation of d-proline produces Δ¹-pyrroline-2-carboxylate, we investigated whether wild-type S. meliloti and hyp mutants can grow on d-proline. The wild-type RmP110 and the hypRE and hypD mutants all grew on minimal medium containing 10 mM d-proline as the sole carbon source, whereas the hypS mutant failed to grow unless it was complemented in trans with an integrated copy of the hypS gene (Fig. 3B). Furthermore, the hypO and hypMNPQ mutants showed impaired growth with d-proline (see Fig. S3 in the supplemental material). The involvement of the hyp cluster in the catabolism of d-proline and the observed relatively slow growth of wild-type S. meliloti on d-proline (Fig. 3B) led us to investigate whether hyp gene transcription is induced by d-proline (Fig. 4). Reporter fusions of gusA to hypM, hypS, and the regulator hypR were found to be upregulated by d-proline relative to succinate or l-proline but to a lesser extent than l-Hyp (Fig. 4). As these results suggested that the slow growth of the wild type on d-proline may be caused by limited expression of some hyp genes, we tested a strain in which hyp transcription is constitutive because the negative regulator, hypR, was deleted. We found that the ΔhypR strain grew at a rate similar to that of the wild type on d-proline (data not shown), and hence hyp transcription does not appear to limit growth on d-proline.

FIG 3.

Growth profiles of S. meliloti hypS and related mutants. Shown are growth profiles of several S. meliloti strains in minimal medium containing as a sole source of carbon l-proline (A), d-proline (B), or trans-4-hydroxy-l-proline (C). Data points are the means from triplicate samples, and the error bars indicate the standard deviations. Strains shown are wild-type RmP110, RmP2514 (ΔhypS), RmP3155 (proC::Tn5-B20 ΔB161), RmP3153 (proC::Tn5-B20 ΔB161 ΔhypS), and RmP3272 [ΔhypS pTH2685(hypS⁺)].

FIG 4.

Induction profiles of the hypM, hypS, and hypR genes. The expression levels of the three hyp genes were measured following 40 h of growth in four different carbon sources using a gusA transcriptional fusion. β-Glucuronidase activities were derived from triplicate assays (±the standard errors of the mean), and similar results were obtained in two independent experiments. Strains shown are wild-type RmP110, RmP1886 (hypM::gusA), RmP239 (hypS::gusA), and RmFL2315 (hypR::gusA).

Catabolism of d-proline proceeds via l-proline in S. meliloti.

The finding that growth on d-proline required hypS suggested that d-proline catabolism proceeds via Δ¹-pyrroline-2-carboxylate to l-proline, and a schematic diagram of this pathway is shown in Fig. 5. We performed two sets of experiments to confirm that d-proline is converted to l-proline in S. meliloti. First, we investigated whether mutants defective in l-proline utilization are also defective in the utilization of d-proline. In S. meliloti, l-proline is oxidized to glutamate by a proline dehydrogenase and a Δ¹-pyrroline-5-carboxylate dehydrogenase, with both enzyme activities present in the single bifunctional PutA protein (Fig. 5 and reference 49). We identified a putA mutant (SmFL5502) in our previously constructed pTH1522 reporter gene fusion library (21), and growth experiments employing this mutant revealed that the growth of S. meliloti on l-proline or d-proline but not on trans-4-l-Hyp was dependent on putA (Fig. 6). A slight reduction in growth was observed for the putA mutant versus the wild type when grown with trans-4-l-Hyp as the carbon source, and we suggest that may result from constitutive expression of the reporter genes in the putA::lacZ-gfp fusion strain (21) (http://info.mcmaster.ca/fusionlibrary.html).

FIG 5.

Schematic illustrations of l- and d-proline metabolism in Sinorhizobium meliloti. The biosynthetic and catabolic pathways for l-proline biosynthesis from l-glutamate are shown, as is the proposed pathway for d-proline catabolism. l-Proline and d-proline are highlighted by gray shading. The association of proteins to each biochemical reaction is based on the work reported here and elsewhere (20, 49, 64, 65, 67). Abbreviations: Pyr5C, Δ¹-pyrroline-5-carboxylate; Pyr2C, Δ¹-pyrroline-2-carboxylate; γ-glutamyl-5-P, γ-glutamyl-5-phosphate; Pi, phosphate.

FIG 6.

Effect of a putA mutation on growth with proline compounds. The growth profiles of wild-type S. meliloti RmP110 and S. meliloti RmFL5502 (putA::pTH1522) are shown in minimal medium containing as a sole source of carbon l-proline (A), d-proline (B), or trans-4-hydroxy-l-proline (C). Data points are the means from triplicate samples, and the error bars indicate the standard deviations.

Catabolism of d-proline to l-proline predicts that growth of an S. meliloti l-proline auxotroph should occur whether the auxotroph is supplemented with l-proline or d-proline. To investigate this prediction, we employed an S. meliloti l-proline auxotroph with two genes disrupted, proC and smb20003, both of which encode Δ¹-pyrroline-5-carboxylate reductase enzymes required for l-proline synthesis (Fig. 5) (20). As this mutant (RmP3155) is unable to synthesize l-proline from l-glutamate, either l-proline supplementation or the addition of l-proline precursors for separate l-proline biosynthetic pathways is required for growth in minimal medium. When examined, this auxotroph grew in minimal medium containing either l-proline or d-proline (Fig. 3). We are aware of one other report in which an uncharacterized l-proline auxotroph of P. putida KT2440 grew upon supplementation of the medium with d-proline (50). Presumably, the growth resulting from that addition of d-proline would be dependent on the dpkA-encoded Δ¹-pyrroline-2-carboxylate reductase present in P. putida KT2440, although this was not tested (27).

DISCUSSION

We have presented evidence that S. meliloti hypD encodes an HPC deaminase and hypH an α-KGSADH, with each involved in the catabolism of trans-4-l-Hyp to α-KG (Fig. 1). As expected, hypD mutants failed to grow with trans-4-l-Hyp as the sole carbon source (Table 1). On the other hand, both hypH and hypO mutants continued to grow with trans-4-l-Hyp, albeit at half the rate of the wild type (Table 1). The partial-growth as opposed to the no-growth phenotype of the hypH and hypO mutants can be attributed to the presence of alternate enzyme activities. HypO is predicted to be a cis-4-d-Hyp oxidase/dehydrogenase (14), and several proteins are annotated as d-amino acid oxidase enzymes in S. meliloti, e.g., Smb20877 and Smc03265. As these enzymes generally have broad specificity (51), it seems likely that the residual growth of the hypO mutant on trans-4-l-Hyp is due to these oxidase activities; however, those enzymes were not examined further. A cis-4-d-Hyp dehydrogenase enzyme from P. putida was recently purified and characterized (12), and the sequence of that protein (PP1255) is 35% identical to that of the S. meliloti HypO protein. However, when S. meliloti hypO was cloned into expression plasmids, attempts to overexpress a His₆-tagged S. meliloti HypO protein either in E. coli or in S. meliloti were unsuccessful (data not shown). Thus, while we have not directly demonstrated an activity for HypO, its similarity to the P. putida PP1255 protein and the partial growth defect of the ΔhypO mutant with trans-4-l-Hyp as a carbon source support its role as a cis-4-d-Hyp dehydrogenase enzyme.

The in vitro catalysis of α-KGSA to α-KG by HypH (Table 2) showed that HypH can function as an α-KGSADH. α-KGSA dehydrogenase enzymes involved in the catabolism of l-arabinose, d-glucarate, d-galactarate, and hydroxy-l-proline have also been characterized (25, 52, 53). To investigate whether the reduced-growth phenotype of the hypH mutant on trans-4-l-Hyp may be due to the presence of other α-KGSADH-like enzymes, we searched the S. meliloti genome for other annotated dehydrogenases and identified the most similar as AraE, which functions as an α-KGSADH in the catabolism of l-arabinose (18). A hypH araE double mutant was constructed, and unlike the hypH mutant, which showed a moderate growth phenotype, and the araE mutant, which grew like the wild type, the hypH araE double mutant (RmP3174) failed to grow with trans-4-l-Hyp as the carbon source (data not shown). None of these strains had a growth defect with glucose as the sole carbon source. As the growth of the hypH mutant on trans-4-l-Hyp was dependent on araE, this growth presumably resulted from the activity of the araE-encoded α-KGSADH isoenzyme. This represents another example of genetic redundancy in S. meliloti (20), and taken together, these data illustrate that hypH encodes the primary α-KGSADH involved in the trans-4-l-Hyp catabolic pathway.

The inability of a hypD-null mutant to grow with trans-4-l-Hyp as a carbon source suggested that it is essential in the l-Hyp catabolic pathway. We previously suggested that hypD encodes a Δ¹-pyrroline-4-hydroxy-2-carboxylate (HPC) deaminase that forms α-KGSA and NH₃ from HPC (14). Additionally, LhpC (PP1257) of P. putida, which also catabolizes hydroxyproline, catalyzes the HPC deaminase reaction (12). In this study, we found that purified HypD is capable of converting HPC to α-KGSA and NH₃. Thus, HypD functions as the HPC deaminase in the l-Hyp catabolic pathway in S. meliloti.

HypD is a member of the N-acetylneuraminate lyase (NAL) subfamily of (α/β)₈ barrel proteins, which includes the DHDPS, NAL, HBPHA, and DOGDH enzymes described above in Results. Consistent with what is known about the catalytic mechanism of these enzymes, the HypD structure has an (α/β)₈ barrel fold with a C-terminal extension of alpha helices, and a putative active site which includes the conserved lysine and tyrosine residues (Fig. 2). The proposed mechanism for the NAL subfamily is the formation of a covalent Schiff base intermediate between the nitrogen of the active-site lysine side chain and substrate. However, the reactions catalyzed vary, depending on the enzyme, and include aldol cleavage, condensation, and decarboxylation (47, 48). Addition of HypD and LhpC to the subfamily adds deamination to the reactions performed by this group of proteins (12; this study).

Of the remaining uncharacterized genes in the S. meliloti hyp cluster, we have observed that hypS orthologs are often located in the vicinity of the hypD and hypH orthologs within genomes that carry hyp gene clusters (data not shown). However, a hypS mutant did not impact the ability of S. meliloti to grow with l-hydroxyproline. Characterization of HypS revealed that it functions as a Δ¹-pyrroline-2-carboxylate (P2C) reductase, forming l-proline, in an NADPH-dependent fashion (Table 3). HypS shares 41% amino acid identity with the NADPH-dependent P2C reductase DpkA protein from P. putida KT2440 (27), and the two showed similar kinetic properties because the reductase activities of both were inhibited by high concentrations of the substrate P2C (data not shown) and the two enzymes had similar pH optima. Although P2C is not an intermediate in the catabolism of trans-4-l-Hyp, it is in the catabolism of d-proline, and we showed that growth of S. meliloti on d-proline is dependent on HypS. HypMNPQ and HypO were also involved but were not essential for d-proline catabolism (see Fig. S3 in the supplemental material). These results are consistent with a metabolic pathway in S. meliloti through which d-proline is transported into the cell by the HypMNPQ uptake system, and it is subsequently oxidized to Δ¹-pyrroline-2-carboxylate by HypO and then converted to l-proline by HypS. A ΔhypMNPQ mutant was previously shown to grow poorly on trans-4-l-Hyp, and it was suggested that the residual growth likely resulted from trans-4-l-Hyp uptake via an alternate transport system (13). Similarly, HypMNPQ and HypO may transport and oxidize d-proline, and in their absence, other systems can partially fulfill these functions.

The above results prompted us to investigate whether the catabolic pathway for trans-4-l-Hyp might result in the synthesis of small quantities of d- or l-proline and hence whether the l-proline auxotroph might grow in minimal medium containing trans-4-l-Hyp (Fig. 3). Growth experiments revealed that incubation of a proline auxotroph (proC smb20003) with trans-4-l-Hyp (10 mM) resulted in slight growth; however, this growth was also observed for the proC smb20003 hypS triple mutant and hence was hypS independent (Fig. 3). We suspect that this is due to background ornithine cyclodeaminase activity, encoded by ocd, which can convert the l-arginine biosynthetic intermediate l-ornithine into l-proline (20, 54, 55), allowing partial growth of the S. meliloti l-proline auxotroph.

The ability of S. meliloti to metabolize and grow on d-proline coupled with the identified requirement for hypS in d-proline metabolism raises the question as to whether this is a primary role in the etiology of hypS. Low concentrations of d-proline have been detected in plants and soil, and hence the ability to catabolize d-proline might benefit the bacteria (56–58). However, an alternate role for hypS is in the reduction of Δ¹-pyrroline-2-carboxylate that is formed from the metabolism of trans-3-hydroxy-l-proline (trans-3-l-Hyp). Visser and coworkers identified a human trans-3-hydroxy-l-proline dehydratase enzyme, similar to proline racemases, which converts trans-3-l-Hyp to Δ¹-pyrroline-2-carboxylate (26). More recently, Watanabe and colleagues showed that Azospirillum brasilense can grow on trans-3-l-Hyp and that this ability is dependent on both a trans-3-l-Hyp dehydratase enzyme (GenBank accession no. AB894494.1) and a Δ¹-pyrroline-2-carboxylate reductase (GenBank accession no. AB845355.1), the sequence of which is 51% identical to that of HypS (59). When tested, we found that S. meliloti cannot grow on trans-3-l-Hyp as a carbon source. Moreover, while the sequence of the S. meliloti HypY protein shows similarity to that of the trans-3-l-Hyp dehydratase (proline racemase), it lacks a cysteine residue important for catalysis and is presumably inactive for the dehydration reaction (14, 26, 59). We speculate that hypS is situated within the hyp cluster, as it was involved in the metabolism of hydroxyproline derivatives other than trans-4-l-Hyp, whose catabolism generates Δ¹-pyrroline-2-carboxylate.

The study of hyp genes in S. meliloti is also of interest because among the most highly expressed genes in N₂-fixing alfalfa bacteroids is a FixLJ microaerobically regulated gene, smc03253 (60), which encodes a 2-oxoglutarate-dependent dioxygenase that converts l-proline to cis-4-hydroxy-l-proline (61). This raises the possibility that cis-4-Hyp could be synthesized by bacteroids (60), and if this occurred, we would expect to see induction of the hyp gene cluster during symbiosis (14). However, in alfalfa root nodules, the concentration of free hydroxyproline appears to be low, as the S. meliloti hyp genes are expressed at low levels in bacteroids (13).

In conclusion, the findings reported here and earlier (14) confirmed that trans-4-hydroxy-l-proline is catabolized to α-ketoglutarate in S. meliloti via the pathway shown in Fig. 1B. Together with other reports (12, 25, 59, 62–64), this report defines the genes that encode specific enzyme activities required for trans-4-hydroxy-l-proline catabolism. These studies should allow the identification of hydroxyproline catabolic genes in other organisms where many of these genes are annotated as genes of unknown function. In this respect, detailed knowledge of both the biochemistry and the regulation of the hyp genes of S. meliloti should be useful. The role of the hyp cluster in d-proline metabolism serves as a reminder that the hyp genes likely play roles in the metabolism of hydroxyproline- and proline-related compounds, many of which are found in plants, soils, and sediments (65, 66), as was recently shown for the osmolytes trans-4-hydroxy-l-proline betaine and cis-4-hydroxy-d-proline betaine (63).

Funding Statement

This work was also supported by funding from the National Science and Engineering Research Council of Canada through an NSERC CGS-D scholarship to G.C.D. The funder had no role in study design, data collection and interpretation, or the decision to submit the work for publication.