Control of hydroxyproline catabolism in Sinorhizobium meliloti

Authors


E-mail finan@mcmaster.ca; Tel. (+1) 905 525 9140 ext. 22932; Fax (+1) 905 522 6066.

Summary

Hydroxyproline (Hyp) in decaying organic matter is a rich source of carbon and nitrogen for microorganisms. A bacterial pathway for Hyp catabolism is known; however, genes and function relationships are not established. In the pathway, trans-4-hydroxy-l-proline (4-l-Hyp) is epimerized to cis-4-hydroxy-d-proline (4-d-Hyp), and then, in three enzymatic reactions, the d-isomer is converted via Δ-pyrroline-4-hydroxy-2-carboxylate (HPC) and α-ketoglutarate semialdehyde (KGSA) to α-ketoglutarate (KG). Here a transcriptional analysis of cells growing on 4-l-Hyp, and the regulation and functions of genes from a Hyp catabolism locus of the legume endosymbiont Sinorhizobium meliloti are reported. Fourteen hydroxyproline catabolism genes (hyp), in five transcripts hypR, hypD, hypH, hypST and hypMNPQO(RE)XYZ, were negatively regulated by hypR. hypRE was shown to encode 4-hydroxyproline 2-epimerase and a hypRE mutant grew with 4-d-Hyp but not 4-l-Hyp. hypO, hypD and hypH are predicted to encode 4-d-Hyp oxidase, HPC deaminase and α-KGSA dehydrogenase respectively. The functions for hypS, hypT, hypX, hypY and hypZ remain to be determined. The data suggest 4-Hyp is converted to the tricarboxylic acid cycle intermediate α-ketoglutarate via the pathway established biochemically for Pseudomonas. This report describes the first molecular characterization of a Hyp catabolism locus.

Introduction

Four per cent of the total amino acids found in animal proteins is estimated to be trans-4-hydroxy-l-proline (4-l-Hyp), as this amino acid constitutes a large proportion of the abundant protein collagen (Gorres and Raines, 2010). In plants, glycoproteins rich in 4-l-Hyp are present in cell walls and root nodules, and they function in plant growth and development, including cell wall assembly and root hair elongation (Benhamou et al., 1991; Velasquez et al., 2011). In animals and plants, hydroxyproline (Hyp) is synthesized post-translationally by prolyl hydroxylase enzymes that catalyse the hydroxylation of proline residues within specific proteins. In animals, prolyl hydroxylase domain proteins also play a key role in oxygen sensing via interactions with hypoxia-inducible transcription factors (Nakayama, 2009). In some microorganisms, hydroxyproline is incorporated into peptide antibiotics and there it is synthesized via the direct hydroxylation of free proline (Katz et al., 1979).

Hydroxyproline is a rich source of reduced carbon and nitrogen. In mammals, Hyp is released upon protein degradation and it is rapidly reabsorbed and catabolized by a pathway that is different to that employed by bacteria (Adams and Frank, 1980). The bacterial pathway for Hyp catabolism was elucidated by Adams and colleagues for a Pseudomonas strain that grew on 4-l-Hyp as a carbon source (Fig. 1A). Four enzymatic steps were identified that result in the conversion of 4-l-Hyp to α-ketoglutaric acid (α-KG), an intermediate of the tricarboxylic acid (TCA) cycle (Fig. 1A). Following its transport into cells, hydroxyproline epimerase (EC 5.1.1.8) catalyses the isomerization of 4-l-Hyp to cis-4-hydroxy-d-proline (4-d-Hyp; previously called allohydroxy-d-proline) (Finlay and Adams, 1970), and this is then oxidized to Δ1-pyrroline-4-hydroxy-2-carboxylate by d-amino acid oxidase (EC 1.4.3.3) (Yoneya and Adams, 1961). Δ1-Pyrroline-4-hydroxy-2-carboxylate is converted to α-ketoglutarate semialdehyde (α-KGSA) by Δ1-pyrroline-4-hydroxy-2-carboxylate deaminase (EC 3.5.4.22) (Singh and Adams 1965) and in the fourth step α-KGSA is oxidized to α-ketoglutarate by the enzyme α-KGSA dehydrogenase (KGSADH) (Singh and Adams, 1964). While several reports have identified mutants in this pathway (e.g. Koo and Adams, 1974; Manoharan and Jayaraman, 1979) and activity measurements showed that these enzymes were induced upon growth on Hyp (Gryder and Adams, 1969), no detailed characterization of the genes or genetic components involved in this pathway has been reported.

Figure 1.

Hydroxyproline catabolic pathway and hyp gene cluster of S. meliloti. A. Catabolic pathway as determined by Adams et al. using Pseudomonas putida as a model (Adams and Frank, 1980). Trans-4-hydroxy-l-proline (4-l-Hyp) is the initial substrate of the pathway; the identities of the metabolites are cis-4-hydroxy-d-proline (4-d-Hyp), Δ-pyrroline-4-hydroxy-2-carboxylate (HPC), α-ketoglutaric semialdehyde (α-KGSA) and finally the tricarboxylic acid cycle intermediate α-ketoglutarate (2-oxoglutaric acid; α-KG). B. A schematic of the hyp gene cluster found on pSymB of S. meliloti, as described by MacLean et al. (2009) and in this study. Gene annotations and functions are shown in Table 1.

Recently, Minoprio and colleagues examined 12 proteins from various pathogens that were annotated as proline racemase (PRAC)-like, but only one had PRAC activity (Chamond et al., 2003; Goytia et al., 2007). Instead, they found that proteins from Pseudomonas aeruginosa, Burkholderia pseudomallei and three Brucella species were hydroxyproline-2-epimerase enzymes (HypRE) (Goytia et al., 2007). Whereas PRAC and HypRE enzymes each contained two conserved cysteine residues in the active site, three amino acid residues that discriminated proline racemases and hydroxyproline epimerases were annotated. No defined bacterial mutants in these genes were identified.

Several reports have drawn attention to catabolic pathways that converge in the synthesis of α-ketoglutarate semialdehyde. In addition to 4-l-Hyp, this includes the catabolic pathways of the pentose sugars, d-glucarate, d-galactarate and l-arabinose. These pathways contain isozymes for the α-ketoglutaric semialdehyde dehydrogenase in which α-ketoglutaric semialdehyde is oxidized to α-ketoglutaric acid (Brouns et al., 2006; Watanabe et al., 2007; Aghaie et al., 2008).

The nitrogen-fixing plant endosymbiont Sinorhizobium meliloti grows rapidly on 4-l-Hyp or 4-d-Hyp (4-Hyp) as a sole source of carbon and nitrogen. We use the abbreviation Hyp when referring to 4-l-Hyp or 4-d-Hyp. Other forms of hydroxyproline, such as trans-3-hydroxy-l-proline are referred to by name. The paucity of information on the genes involved in 4-l-Hyp metabolism and the possibility that Hyp may play a role in the interaction of S. meliloti with alfalfa prompted us to analyse the catabolic pathway for this amino acid. Earlier we identified an ABC transport system responsible for Hyp uptake into S. meliloti and showed that deletion of this system (ΔhypMNPQ) resulted in poor growth on 4-l-Hyp (MacLean et al., 2009). Here we identify and characterize genes linked to the hypMNPQ transport gene cluster and report on enzyme activities associated with some of the gene products. We examine their transcriptional regulation by 4-l-Hyp, 4-d-Hyp and hypR. In addition to the hyp cluster, we also identify other genes whose expression is induced upon catabolism of 4-l-Hyp.

Results

Gene prediction within the hyp gene cluster

Previously, we identified an ABC-type uptake system (encoded by hypMNPQ) that is specific to and required for 4-l-Hyp uptake. A GntR-type-negative regulator (hypR) that acts at the hypMNPQ promoter was also identified (MacLean et al., 2009). As illustrated in Fig. 1B and Table 1 and outlined further below, these genes are part of a 14-gene cluster that is involved in growth of S. meliloti on hydroxyproline as a sole source of carbon and nitrogen. In addition to hypMNPQ and hypR, six genes in the hyp cluster encode proteins with predicted enzymatic activities, and three ORFs have no similarity to proteins of known function.

Table 1.  Annotations and predicted functions of the hyp cluster genes of pSymB of S. meliloti.
Gene namesAnnotationPredicted function in hydroxyproline catabolism
  • a. 

    The function of hypR as a regulator is demonstrated here and in MacLean et al. (2009).

  • b. 

    The function of these genes in hydroxyproline transport has been demonstrated previously (MacLean et al., 2009).

  • c. 

    The function of HypRE as a hydroxyproline epimerase is demonstrated in this study. Epimerase activity was not detected for HypY.

hypR smb20258 GntR-type transcription regulatorRegulationa
hypD smb20259 DapA family memberDeaminase
hypT smb20260 Conserved hypothetical proteinUnknown
hypS smb20261 Malate dehydrogenaseUnknown
hypH smb20262 Semialdehyde dehydrogenaseα-Ketoglutaric semialdehyde dehydrogenase
hypM smb20263 Periplasmic solute-binding proteinTransportb
hypN smb20264 PermeaseTransportb
hypP smb20265 PermeaseTransportb
hypQ smb20266 ATP-binding proteinTransportb
hypO smb20267 d-amino acid oxidase d-amino acid oxidase
hypRE smb20268 Proline racemaseHydroxyproline epimerasec
hypX smb20269 Conserved hypothetical proteinUnknown
hypY smb20270 Proline racemasePossible pseudogene, inactivec
hypZ smb20271 Conserved hypothetical proteinUnknown

As indicated in Table 1, three of the predicted enzymes (Smb20262, Smb20267 and Smb20268) are strong candidates for three of the four enzymatic steps of the previously described Hyp catabolic pathway shown in Fig. 1A. Both Smb20268 and Smb20270 have similarity (53% and 39% respectively) to the previously described hydroxyproline epimerase protein from Pseudomonas aeruginosa (PaHypRE) (Goytia et al., 2007). Hydroxproline epimerases and the closely related proline racemases share a reaction mechanism that requires two active site cysteines (Chamond et al., 2003; Buschiazzo et al., 2006; Goytia et al., 2007). An alignment across the active site region of Smb20268 and Smb20270 with characterized racemases/epimerases of this family shows that the former gene product has both active site cysteines (cys 90 and 253) while Smb20270 does not (Fig. S1). This suggests that Smb20268 is the most likely candidate to encode an active HypRE. As discussed further below, we predicted that Smb20259 was the most likely candidate to have the deaminase activity, based on known reaction mechanisms of related proteins of the dihydrodipicolinate synthase (DapA) and N-acetylneuraminate lyase subfamily (Lawrence et al., 1997).

The putative HypRE (Smb20268) is absolutely required for growth on 4-hydroxyproline

To determine if hydroxyproline catabolism in S. meliloti proceeds via the pathway shown in Fig. 1A, we studied the hypRE-like genes hypRE (smb20268) and hypY (smb20270) in greater detail. HypRE activity is unique to the pathway shown in Fig. 1A, and is required for the first step (Adams, 1959; Gryder and Adams, 1969). To our knowledge no other biological function of HypRE activity is known.

To determine if either smb20268 or smb20270 is required for catabolism of 4-l-Hyp, we constructed a non-polar deletion of each gene and compared the growth of each mutant to wild-type S. meliloti on a number of different carbon sources. A deletion of smb20270 had no effect on growth on any of the carbon sources tested, including both forms of 4-Hyp (4-l-Hyp and 4-d-Hyp; Fig. 2). In contrast, while the smb20268 deletion mutant exhibited wild-type growth on succinate and proline, no growth was observed with 4-l-Hyp as a sole carbon source (Fig. 2, compare A and B with C). However, wild-type growth was observed for this mutant on 4-d-Hyp (Fig. 2D).

Figure 2.

Growth of wild-type S. meliloti (open squares), a hypRE deletion mutant (filled circles) and a hypY deletion mutant (filled triangles) in M9 minimal medium containing (A) succinate as the sole source of carbon, or (B) l-proline, (C) 4-l-Hyp or (D) 4-d-Hyp as sole sources of carbon and nitrogen. Optical density was measured in triplicate (three cultures per strain); data points shown are the averages of each with error bars.

To verify that the 4-l-Hyp-deficient growth phenotype observed for Δsmb20268 was solely due to the loss of smb20268, the mutant was complemented with the native smb20268 gene expressed from its own promoter (PhypM) on a plasmid. As shown in Fig. S2, growth of the complemented mutant on 4-l-Hyp was restored to wild-type levels.

Therefore, Smb20268 (designated here as HypRE) is absolutely required for catabolism of 4-l-Hyp in S. meliloti, while Smb20270 (HypY) is not, consistent with the observation that it is missing active site cysteine residues (Fig. S1). Further, the observation that the smb20268 deletion mutant is capable of growing on 4-d-Hyp (the product of the HypRE reaction in Fig. 1A) strongly suggests that this gene encodes a HypRE activity, and that hydroxyproline catabolism in S. meliloti occurs via the Pseudomonas putida pathway elucidated by Adams and colleagues (Adams, 1959; Gryder and Adams, 1969; Adams and Frank, 1980).

Smb20268, but not Smb20270, has HypRE activity in vitro

To further address the HypRE activity of HypRE and HypY, we overexpressed each protein with a C-terminal His6 fusion from the T7 promoter of pET28 in Escherichia coli BL21/DE3. Using purified protein, racemase/epimerase activity was assayed in vitro using a capillary electrophoresis method (to resolve and quantify proline and hydroxyproline stereoisomers) that we have previously reported (Gavina et al., 2010).

Purified recombinant protein was incubated at a concentration of 500 nM with 8 mM substrate, either l-proline, 4-l-Hyp or 4-d-Hyp. Reactions were incubated at 25°C for 5 min, and samples removed every 30 s and quenched. Each sample was derivatized with Fmoc-Cl, and then separated by capillary electrophoresis for stereoisomer identification and quantification. With 4-l-Hyp as substrate, epimerase activity of recombinant HypRE protein was high, as indicated by the rapid accumulation of its stereoisomer 4-d-Hyp with a Km of 10 mM and Vmax of 87 µM min−1 (Fig. 3). The kinetics of the reverse reaction (4-d-Hyp to 4-l-Hyp) were very similar (Km = 8 mM; Vmax = 78 µM min−1; Fig. 3), while isomerization of l-proline to d-proline was not detected (data not shown). No conversion was measured in control reactions with substrate and heat-inactivated protein (data not shown). In a similar set of reactions with purified recombinant Smb20270, no measurable HypRE or PRAC activity was detected in the presence of any of the substrates tested (data not shown).

Figure 3.

Non-linear regression of Michaelis–Menten kinetics highlighting the reversibility of 4-d/l-Hyp conversion by SmHypRE. Reactions were performed with 50 nM enzyme at 30°C and heat-quenched prior to analysis by capillary electrophoresis with UV detection (Gavina et al., 2010). Error from replicate measurements (n = 3) of initial enzyme reaction rates had an average coefficient of variance of 8%.

All predicted genes of the hyp gene cluster are induced by 4-Hyp

Previously, we have shown that the ABC-type uptake genes hypMNPQ are induced by 4-Hyp (Mauchline et al., 2006; MacLean et al., 2009). The hyp gene cluster of S. meliloti shown in Fig. 1B is quite extensive, and includes more genes than those predicted above to be required for the hydroxyproline catabolic pathway. Here, we examine conditions for expression of these genes to predict whether or not they are likely to function in hydroxyproline catabolism.

To study expression, we used gusA or lacZ reporter fusions to each hyp gene, and integrated these via single cross-over recombination with the homologous gene of the hyp gene cluster on pSymB. Assays were conducted with cells grown on glucose, succinate or glycerol and glycerol supplemented with 4-l-Hyp or 4-d-Hyp or l-proline as carbon sources. Glycerol alone did not significantly induce any of the reporter constructs tested (Table 2).

Table 2.  All hyp cluster genes are hydroxyproline-inducible.a
FusionRelative activityb
GlucoseSuccinateGlycerol l-pro4-l-Hyp4-d-Hyp
  • a. 

    Strains containing gusA and lacZ fusions to each hyp cluster gene were cultured in M9 minimal medium containing 15 mM glucose, 15 mM succinate or 0.5% glycerol as sole carbon sources. For cultures containing amino acids as carbon sources [5 mM l-proline (l-pro), 4-l-Hyp or 4-d-Hyp], each was supplemented with 0.5% glycerol as many of the hyp reporter gene fusions disrupt the hyp cluster genes.

  • b. 

    Expression levels of the gusA and lacZ fusions were normalized by setting the hypM::gusA and hypM::lacZ fusions both at 100. Assays were performed in triplicate (from three separate cultures); data shown are averages with standard deviation.

None (WT) gusA0.1 ± 0.00.2 ± 0.00.3 ± 0.00.2 ± 0.00.2 ± 0.00.1 ± 0.0
None (WT) lacZ1.6 ± 0.12.4 ± 0.15.7 ± 0.02.5 ± 0.12.7 ± 0.12.7 ± 0.1
hypR::gusA2.8 ± 0.13.8 ± 0.23.5 ± 0.03.3 ± 0.031.2 ± 0.328.0 ± 1.1
hypD::gusA4.6 ± 0.05.7 ± 0.34.0 ± 0.26.0 ± 0.1113.0 ± 3.3120.2 ± 0.1
hypS::lacZ10.8 ± 0.913.2 ± 0.312.3 ± 0.115.2 ± 0.086.6 ± 0.774.6 ± 0.4
hypH::gusA20.4 ± 0.827.6 ± 0.519.9 ± 0.323.5 ± 0.159.7 ± 2.267.2 ± 1.0
hypM::gusA11.8 ± 0.111.9 ± 0.313.9 ± 0.112.9 ± 0.1100.0 ± 1.4b104.0 ± 1.2
hypM::lacZ6.5 ± 0.07.1 ± 0.116.3 ± 1.36.9 ± 1.0100.0 ± 7.0b104.5 ± 0.9
hypO::lacZ4.9 ± 0.25.7 ± 0.217.5 ± 0.08.1 ± 0.140.5 ± 0.292.7 ± 0.3
hypRE::gusA8.2 ± 0.19.4 ± 0.19.1 ± 0.110.0 ± 0.162.5 ± 3.598.2 ± 1.8
hypX::lacZ9.5 ± 0.29.9 ± 0.217.5 ± 0.312.3 ± 0.476.3 ± 3.0110.2 ± 2.5
hypY::gusA4.2 ± 0.25.8 ± 0.23.8 ± 0.14.5 ± 0.055.5 ± 1.956.7 ± 2.3
hypZ::gusA8.0 ± 0.511.7 ± 0.18.9 ± 0.110.7 ± 0.160.7 ± 0.559.3 ± 1.0

For all genes within the hyp cluster, induction by 4-l-Hyp or 4-d-Hyp was at least twofold above that in the presence of any of the other carbon sources tested, including l-proline (Table 2). For most of the genes, there was no significant difference in induction by the two Hyp diastereomers. However, hypO, hypRE and hypX were induced more strongly in the presence of 4-d-Hyp than 4-l-Hyp. The hypD::gusA fusion was the most strongly induced gene in the cluster.

Identification of transcription start sites of hyp cluster promoters

Previously, we identified the promoter for the operon hypMNPQ using primer extension analysis (Fig. 1B) (MacLean et al., 2009). Here, we perform similar experiments to identify and map transcription start sites upstream of hypR, hypD, hypS and hypH.

For primer extension assays, RNA was isolated from wild-type S. meliloti grown on either glycerol alone (as a sole source of carbon), or on 4-l-Hyp alone as a sole source of both carbon and nitrogen. One clear transcription start site was identified upstream of hypR, one upstream of hypD and one upstream of hypS (Fig. 4A–C).

Figure 4.

Identification of hyp cluster promoters. Primer extension reactions for PhypR (A), PhypD (B), PhypS (C) and PhypH (D) were performed using RNA isolated from wild-type S. meliloti cultured in M9 minimal medium with glycerol as the sole source of carbon (lane 1), or 4-l-Hyp as the sole source of carbon and nitrogen (lane 2). Additional reactions were performed for PhypR (A), PhypD (B) and PhypH (D) with RNA isolated from the hypR deletion mutant cultured in medium with glycerol (lane 3), or 4-l-Hyp as above (lane 4). Hydroxyproline-inducible transcript start sites are marked in bold and are underlined. A possible constitutive promoter for hypH expression is also indicated with an asterisk. (E) Sequences of the hyp-inducible promoters identified in this study and of the hypM promoter identified previously for comparison (MacLean et al., 2009). Promoter elements [−35, −10 and transcript start sites (+1)] are underlined, as are the ATG translation start sites. The predicted HypR binding site, 5′-TTTGTnnACna-3′, is shown in italics in bold font.

These sites are 16 nucleotides, 69 nucleotides and 51 nucleotides upstream of each predicted translation start site respectively (Fig. 4E). In wild-type S. meliloti, accumulation of each of these transcripts requires the presence of 4-l-Hyp (Fig. 4A–C, compare lanes 1 and 2), which is consistent with the hydroxyproline-inducible expression of these genes.

For hypR and hypD (smb20259), primer extension assays were also performed with RNA isolated from cells of a hypR mutant. In both cases, the hydroxyproline-inducible transcript accumulated in cells of the hypR mutant that were grown in both glycerol alone (sole source of carbon) and hydroxyproline alone as a sole source of carbon and nitrogen (Fig. 4A and B, compare lanes 3 and 4). This constitutive expression suggests that these genes are regulated by the previously described HypR-negative regulator (MacLean et al., 2009).

A clear hydroxyproline-inducible transcript start site was also identified 166 nucleotides upstream of hypH (Fig. 4D and E). Like the above described promoters, accumulation of this transcript was constitutive in the hypR mutant (Fig. 4D, lanes 3 and 4). An additional promoter with a transcription start site 93 nucleotides upstream of hypH was observed. This was active when wild-type cells are grown in glycerol alone as well as with 4-l-Hyp. This promoter may account for the relatively high background expression of hypH gene fusions that is observed in cells grown in the absence of 4-l-Hyp (Tables 2 and 3). The α-KGSADH reaction catalysed by the predicted HypH enzyme is known to occur in several other catabolic pathways (see Introduction) and hypR-independent transcription of hypH would allow HypH to function in other pathways in S. meliloti. We found that S. meliloti hypH mutants grow on 4-l-Hyp and assume this growth occurs because alternate α-KGSADH enzymes are active in cells growing on 4-l-Hyp.

Table 3.  Glucose utilization has a minimal effect on 4-l-hyp-induced hyp gene expressiona.
FusionMiller unitsb
Glucose + NHyp + NHyp + glucose + NHypHyp + glucose
  • a. 

    Strains containing integrated gusA fusions to each hyp promoter were cultured in M9 minimal medium containing excess ammonia as a nitrogen source and 15 mM glucose as the sole carbon source (Glucose + N), 10 mM 4-l-Hyp (Hyp + N), or both glucose and 4-l-Hyp (Hyp + glucose + N). Assays were also performed with 4-l-Hyp provided as the sole nitrogen source, either alone (Hyp) or with the addition of glucose (Hyp + glucose).

  • b. 

    Assays were performed in triplicate (from three separate cultures); data shown are averages with standard deviation.

None (WT)5 ± 0.16 ± 0.15 ± 0.114 ± 0.49 ± 0.6
hypR::gusA61 ± 0.2620 ± 34.1603 ± 28.1713 ± 10.3663 ± 5.2
hypD::gusA90 ± 2.73372 ± 38.02973 ± 56.13263 ± 136.02261 ± 51.8
hypS::gusA42 ± 1.12099 ± 43.01538 ± 65.62235 ± 80.01546 ± 19.3
hypH::gusA353 ± 19.5792 ± 48.4893 ± 21.61136 ± 15.2942 ± 18.0
hypM::gusA210 ± 19.12896 ± 236.21804 ± 43.72236 ± 40.51661 ± 44.8

HypR is a member of the helix–turn–helix GntR (FadR) family of transcription regulators (Rigali et al., 2002) and we previously identified a sequence 5′-TTTGTnnAC-3′ in the −10 promoter region of hypR that is likely a HypR binding site (MacLean et al., 2009). This sequence is present in the −35 region of the hypD promoter, and 15 nucleotides upstream of the −35 region of the hypM promoter. The data revealed that the divergently transcribed hypS and hypH transcripts arise from overlapping promoters. The predicted HypR binding site, TTTGTnnACna, is present directly upstream of the −10 region of the hypS promoter and on the opposite strand, this HypR binding site overlaps the −10 region of the hypH promoter. We therefore suggest that HypR binding to the TTTGTnnACna site occludes RNA polymerase from simultaneously binding to both the hypS and the hypH promoters.

All identified hyp cluster promoters are regulated by HypR, and induced most strongly by a metabolite of the pathway

The hypMNPQ promoter is repressed by HypR, and repression is relieved in the presence of 4-Hyp (MacLean et al., 2009). The above primer extension data suggest that HypR negatively regulates additional promoters within the hyp cluster.

To test whether HypR regulates each of the hyp cluster promoters, we compared their expression in both the wild-type and hypR mutant backgrounds (Fig. 5). In the wild-type, expression of each promoter fusion is at least threefold higher when the cells are cultured in media containing either 4-l-Hyp or 4-d-Hyp as compared to glycerol alone, consistent with the previous results. However, in the hypR mutant background, the expression of each promoter fusion is at similar levels at all growth conditions tested. This loss of hydroxyproline-dependent expression in the hypR mutant strongly suggests that HypR is a negative regulator at each promoter, and repression is relieved in the presence of either 4-l-Hyp or 4-d-Hyp.

Figure 5.

HypR regulates expression at each hyp cluster promoter. Wild-type (WT), hypR mutant or hypRE mutant strains containing gusA and lacZ fusions to each hyp cluster promoter were cultured in M9 minimal medium containing 0.5% glycerol (gly), 5 mM 4-l-Hyp supplemented with 0.5% glycerol (l-Hyp) or 5 mM 4-d-Hyp supplemented with 0.5% glycerol (d-Hyp). Expression levels of the gusA and lacZ fusions were normalized by setting the hypM::gusA and hypM::lacZ fusions in the wild-type background both at 100. Assays were performed in triplicate (from three separate cultures) and the data shown are averages with error bars representing standard deviation.

HypRE performs the first step of hydroxyproline catabolism and, as shown above, is absolutely required for growth on 4-l-Hyp. To determine if 4-l-Hyp or a downstream catabolite is the direct inducer of hyp gene expression, we also measured the activities of the above reporter fusions in the hypRE deletion background. Expression of each of the five hyp promoters is hydroxyproline-inducible in the hypRE mutant background as expected (Fig. 5). However, in the presence of 4-d-Hyp, the expression of each fusion is at least 2.5-fold higher than in the presence of 4-l-Hyp. For the promoter fusions to hypD (smb20259), hypS (smb20261) and hypM, expression levels in the presence of 4-l-Hyp are higher than in glycerol alone; however, this is not the case for the fusions to hypR and hypH (smb20262). These data indicate that 4-d-Hyp is the effector that interacts with HypR and the results suggest a role of positive feedback in the regulation of hyp catabolism, as 4-d-Hyp (or a downstream catabolite) is a stronger inducer than the initial substrate 4-l-Hyp.

hyp gene expression is not strongly affected by alternate carbon and nitrogen sources

To test whether or not hyp gene expression is affected by the presence of an alternate carbon source, we compared the induction of the five hyp promoters in the absence or presence of glucose. In control reactions, cells were grown in media containing NH3 as a nitrogen source and either glucose or 4-Hyp as a carbon source. Induction of the hypR, hypD, hypS and hypM promoters is at least 10-fold greater in 4-l-Hyp than in glucose, while induction of hypH is only twofold greater in 4-l-Hyp than in glucose (Table 3), consistent with results shown in Table 2. When both glucose and 4-l-Hyp are supplied, expression levels are somewhat lower than in 4-l-Hyp alone for hypD, hypS and hypM. Moreover, employing succinate as an alternate carbon source did not alter the 4-l-Hyp-mediated expression of hypM or hypH gene fusions from that observed for glucose. In both cases the effect was less than twofold (Table 3 and data not shown). Interestingly, the effect of glucose on hyp gene induction is similar even when 4-l-Hyp is provided as the sole source of nitrogen, suggesting that any effect is due to the presence of an alternate carbon source, rather than nitrogen limitation.

To further investigate nitrogen regulation and hyp gene expression, gene fusion strains were grown with glucose and ammonia, nitrate, l-proline or 4-l-Hyp as sources of nitrogen. Gene fusions to three nitrogen stress response (Ntr-regulated) genes, nrtA, sma0585 and livK (encoding two putative transport binding proteins for nitrate and an amino acid respectively), were examined (Galibert et al., 2001; Davalos et al., 2004; Mauchline et al., 2006). These were expressed at higher levels with either nitrate or l-proline as nitrogen sources than with ammonia (a preferred nitrogen source) or 4-l-Hyp (Table 4). Therefore, growth on l-hydroxyproline as a sole source of nitrogen does not induce the Ntr response.

Table 4.  The hyp cluster genes and nitrogen regulation.a
FusionRelative activityb
NH3NO3ProHypHyp + NH3
  • a. 

    Strains containing gusA or lacZ fusions were cultured in M9 minimal medium containing 15 mM glucose as a carbon source and the following as nitrogen sources: excess ammonium chloride (NH3), nitrate (NO3), l-proline (Pro), 4-l-Hyp (Hyp), or both 4-l-Hyp and excess ammonium chloride (Hyp + NH3).

  • b. 

    Expression levels of the gusA and lacZ fusions were normalized by setting the hypM::gusA and hypM::lacZ fusions both at 100. Activity assays were performed in triplicate (from three separate cultures); data shown are averages with standard deviation.

None (WT) Gus0 ± 0.01 ± 0.00 ± 0.00 ± 0.00 ± 0.0
None (WT) lacZ3 ± 0.19 ± 0.13 ± 0.14 ± 0.14 ± 0.2
nrtA::lacZ14 ± 0.362 ± 2.474 ± 0.616 ± 0.210 ± 0.6
sma0585::lacZ5 ± 0.774 ± 1.512 ± 0.49 ± 1.06 ± 0.4
livK::gusA19 ± 0.350 ± 1.740 ± 0.923 ± 0.615 ± 0.2
hypR::gusA3 ± 0.13 ± 0.03 ± 0.122 ± 0.524 ± 0.5
hypD::gusA4 ± 0.14 ± 0.15 ± 0.1115 ± 1.9110 ± 3.5
hypS::lacZ18 ± 0.321 ± 1.323 ± 0.572 ± 2.289 ± 1.2
hypH::gusA13 ± 0.113 ± 0.416 ± 0.736 ± 0.839 ± 1.2
hypM::gusA11 ± 0.112 ± 0.314 ± 0.0100 ± 2.4b74 ± 0.7
hypM::lacZ14 ± 0.216 ± 0.614 ± 0.1100 ± 2.1b81 ± 1.9

In the presence of ammonia, nitrate or l-proline the hyp genes were expressed at similar low levels, while expression in the presence of 4-l-Hyp was much higher (Table 4). Therefore, the nitrogen stress response does not activate the hyp cluster genes. In addition, hyp gene expression levels in the presence of 4-l-Hyp versus 4-l-Hyp plus ammonia were similar, suggesting that the presence of a preferred nitrogen source does not repress expression of the hyp genes, even in the presence of an alternate carbon source.

Global gene expression analysis

In order to determine if additional genes (other than those of the hyp cluster) are differentially regulated in the presence of 4-Hyp versus alternate carbon and/or nitrogen sources, we performed transcriptome studies using whole-genome arrays. For these experiments, RNA was extracted from mid-log phase wild-type S. meliloti cells grown in minimal medium containing glycerol or 4-l-Hyp as sole carbon sources (both with NH3), and from cells grown with 4-l-Hyp as both the sole carbon and nitrogen source.

As expected, in a comparison of growth on 4-l-Hyp versus glycerol as sole carbon sources (both with NH3), expression levels of all hyp cluster genes were at least twofold greater in the presence of 4-l-Hyp than in glycerol (Tables 5 and S2). An additional set of transport genes encoding a TRAP-T system (smb20320, smb20321, smb20322) was also Hyp-inducible, consistent with previously published results (Mauchline et al., 2006). A number of genes predicted to be involved in amino acid metabolism were also induced and one of these, smc04388, is a putative ω-aminotransferase (class III). We have found that Smc04388 shares key amino acid residues that are conserved in other ω-aminotransferases including Cc3143 (AptA) from Caulobacter crescentus (Hwang et al., 2008). The physiological role of smc04388 will be interesting to determine as relatively little is known of the biological roles of these enzymes. In addition, whether the induction of smc04388 relates to the accumulation of an amino donor related to Hyp catabolism is an interesting possibility.

Table 5.  Transcriptome analysis of S. meliloti: representative genes whose expression in cells grown on 4-l-Hyp as carbon source was greater than expression with glycerol as carbon source.
GeneSignal (net)aRatiobAnnotation
HYPGLYHYP/GLYProbability
  • a. 

    Genes that are significantly upregulated in HYP-relative GLY; NS, not significantly above background (P > 0.001).

  • b. 

    Change in expression (HYP vs GLY); NS, not significantly greater than background (P > 0.01).

  • RmP110 cells were grown in M9 with 0.5% glycerol and in M9 with 4-l-hydroxyproline (in both cases, NH4Cl was added as a nitrogen source).

  • Substrate prediction based on Mauchline et al. (2006).

  • Abbreviations: DH, dehydrogenase; SU, subunit; N, nitrogen; C, carbon; con., conserved; TCA, tricarboxylic acid cycle; TRAP-T, transport system; ABC-T, transport system.

Hydroxyproline metabolism     
 hypR5 76744413.00.0000GntR regulator
 hypD12 43629042.90.0000DapA member
 hypT9 77916758.60.0000Hypothetical (L = 130 aa)
 hypS1 0811806.0 (NS)NSMalate DH
 hypH2 4246663.6 (NS)0.0028Semialdehyde DH
 hypM7 96554014.80.0000Periplasmic SBP
 hypN8 23936922.30.0000Permease
 hypP7 02121133.30.0000Permease
 hypQ5 11035114.60.0000ATP-binding protein
 hypO1 076161 (NS)6.7 (NS)NS d-amino acid oxidase
 hypRE1 8952876.60.0006Proline racemase
 hypX1 185113 (NS)10.5 (NS)NSCon. hypothetical protein
 hypY1 273106 (NS)12.0 (NS)NSProline racemase
 hypZ1 2813114.1 (NS)NSCon. hypothetical protein
Unknown transport     
 SM_b203208 07116449.10.0000SBP, TRAP-T
 SM_b203212 71476 (NS)35.60.0000LPT, TRAP-T
 SM_b203223 31785 (NS)38.80.0000SPC, TRAP-T
Carbon metabolism     
 pckA2 81115118.60.0000PEPCK
 lpdA24 9899225.4 (NS)0.0040E3 of αKGDH
 fbaB1 5621719.10.0001Fruc bisphos aldolase
 1 sucB1 9254883.9 (NS)NSE2 of αKGDH
 2 sucA12 37915278.1 (NS)NSE1 of αKGDH
 3 sucD14 20922816.20.0000Succ-CoA syn. α-SU
 4 sucC34 75983874.10.0000Succ-CoA syn. β-SU
 5 mdh23 33141165.70.0000Malate DH
Amino acid and nitrogen metabolism     
 SMc032599 35732572.90.0006Putative DHDPS
 SMc043882 071102 (NS)20.3 (NS)0.0019Omega aminotransferase
 putA801381.7 Proline DH
Calvin cycle genes     
 1 ppe540117 (NS)4.6 (NS)NSRibulose-phos 3-epim.
 2 cbbX1 3482994.5 (NS)NSATPase of AAA+
 3 cbbS10 12782012.40.0000RuBisCO small SU
 4 cbbL4 09521019.50.0000RuBisCO large SU
 5 cbbA1 489138 (NS)10.8 (NS)0.0057Fruc-bispho aldolase
 6 cbbT41742 (NS)9.8 (NS)NSTransketolase
 7 cbbP2 360122 (NS)19.3 (NS)NSP-ribulokinase
 8 cbbF2973 (NS)90 (NS)NS d-fruc-1,6-bisPase
 9 cbbR113 (NS)117 (NS)1.0 (NS)NSTranscript regulator
Growth (replication, stress)     
 dnaA14 81875412.00.0008Replication initiation factor
 gst23 3196774.90.0000Glutathione S-transfer
Thiamine biosynthesis     
 1 thiC36 64475414.90.0000Thi biosynthesis
 2 thiO4 3049894.4 (NS)0.0052Thi biosynthesis
 3 thiG2 6526694.0 (NS)0.0011Thi biosynthesis
 4 thiE2 4966493.8 (NS)0.0043Thi biosynthesis

l-proline catabolism in S. meliloti requires the putA gene encoding proline dehydrogenase and pyrroline-5-carboxylate dehydrogenase (Jiménez-Zurdo et al., 1997; Soto et al., 2000). We note that the microarray data revealed that putA gene transcription occurred at similar levels whether S. meliloti cells were growing on Hyp or glycerol as sole carbon sources. Moreover, no induction of hyp gene expression was observed in cells growing with l-proline as C source (Table 2). Thus despite the similarity of l-proline and 4-l-Hyp, there appears to be no regulatory or metabolic link between their catabolic pathways.

The TCA cycle genes mdh, sucC, sucD and sucA are all expressed at significantly higher levels in 4-l-Hyp (fourfold or greater), while a number of genes implicated in polyol and glycerol transport and metabolism are expressed at significantly higher levels in cells grown in glycerol (compare ‘carbon transport and metabolism’ of Tables 5 and S3). In particular, an operon of eight genes (glpR-smc02514) and a separate gene, glpK, encoding a putative glycerol kinase (Arias and Martinez-Drets, 1976), were highly induced upon growth with glycerol as carbon source (Table S3). Orthologues of these genes were recently characterized and shown to be involved in glycerol uptake and catabolism in Rhizobium leguminosarum (Ding et al., 2012).

Surprisingly, cbbS, cbbL and cbbP were also induced in cells growing on 4-l-Hyp (Table 5). These genes are part of a cluster that encodes components of the Calvin–Benson–Bassham CO2 fixation cycle. While the expression of this gene cluster has not been studied in S. meliloti, this pathway is required for formate-dependent autotrophic growth in S. meliloti (Pickering and Oresnik, 2008).

In a comparison of global gene expression in cells grown on 4-l-Hyp as sole nitrogen source versus cells grown in the presence of ammonia, a number of growth and nitrogen uptake and metabolism genes were differentially expressed. However, as expected, there were few differences in expression levels of the hyp cluster genes observed between these two conditions (Tables  S2, S4 and S5).

Discussion

The S. meliloti hydroxyproline catabolism locus is a cluster of 14 genes whose expression is induced by 4-Hyp. These genes are organized in five transcripts, three with single genes, one with two genes and a transcript containing nine genes that appears to be driven from a promoter upstream of hypM. Promoter and expression analyses indicate that transcription of the hyp genes is negatively regulated by the GntR (FadR)-like regulator HypR, through binding to a site (5′-TTTGTnnAC-3′) in the promoter region of the hypR-regulated genes. As 4-d-Hyp but not 4-l-Hyp induced hyp gene transcription in the hypRE mutant background, we can infer that 4-d-Hyp or a downstream metabolite is the physiological effector that interacts with the HypR protein.

Expression of the hyp genes was not subject to global nitrogen control (Ntr) or any apparent carbon source control (Tables 3 and 4). Instead, induction of the hyp genes appears to be entirely dependent on the presence of 4-Hyp in the media. This is in comparison to the transcription of genes involved in both nitrate assimilation and leucine utilization, which showed global nitrogen control as transcription of these genes was induced in media with nitrate or l-proline as the sole nitrogen sources and transcription was not induced in media with ammonia as the nitrogen source (Table 4). Conversely, cells utilizing 4-l-Hyp as a nitrogen source showed minimal induction of the Ntr-regulated nrtA, sma0585 or livK genes and thus behaved similarly to cells with the preferred nitrogen source ammonia (Table 4 and 4-l-Hyp vs NH3). The absence of nitrogen control of hyp gene transcription is consistent with the inferred −35 and −10 hexanucleotide sequences of promoter regions of the hypM, hypR, hypD, hypS and hypH genes. These sequences appear to be similar to the consensus RpoD-like promoter sequence previously described in S. meliloti (5-CTTGAC-N17-CTATAT; MacLellan et al., 2006) as opposed to the −12 −24 RpoN-dependent promoters (Thöny and Hennecke, 1989) employed by genes subject to global nitrogen control (Fig. 4E). The lack of carbon or nitrogen source control of hyp gene transcription is similar to that reported for the proline utilization gene, putA (Jiménez-Zurdo et al., 1997; Soto et al., 2000).

One of the goals of this work is to establish whether the previously described Hyp catabolic pathway functions in S. meliloti and to define gene–protein function relationships. Thus of the 14 hyp genes, hypR is a regulatory gene, hypMNPQ encode Hyp transport proteins (MacLean et al., 2009) and hypRE encodes a hydroxyproline epimerase. While not definitively established here, hypO appears to encode a 4-d-Hyp oxidase, hypH an α-ketoglutarate semialdehyde dehydrogenase and hypS a malate dehydrogenase-like protein. Preliminary data suggest that hypD encodes the Δ-pyrroline-4-hydroxy-2-carboxylic acid deaminase responsible for the deamination of Δ-pyrroline-4-hydroxy-2-carboxylate to α-ketoglutarate semialdehyde (Fig. 1). We are currently characterizing this enzyme further.

HypY has amino acid sequence similarity to proline racemase and is annotated as a probable proline racemase enzyme. However, purified HypY showed no epimerase activity with l-proline, 4-l-Hyp or 4-d-Hyp as substrates. Both cysteine residues that play key roles in proline and hydroxyproline epimerase catalysis are absent from HypY (Fig. S1). In addition to the two cysteine residues that are important for epimerase activity in PRAC and HypRE enzymes, Goytia et al. (2007) identified three other residues (R1, R2 and R3) that distinguished these two enzymes. The amino acid residues at these positions in the HypY protein (positions 62, 224 and 268) differed from those in the PRAC and HypRE enzymes, whereas the serine, histidine and alanine residues at these positions in the S. meliloti HypRE protein are those predicted for hydroxyproline epimerase enzymes. Thus while the evidence that HypRE is a hydroxyproline epimerase is clear, the role (if any) for HypY is unclear. HypY may play a role, possibly in regulating HypRE activity through protein–protein interactions, or it may have a catalytic activity. Interestingly, a family of PRAC-like proteins in which one cysteine is replaced by a threonine was recently shown to catalyse the dehydratation of trans-3-hydroxy-l-proline to Δ1-pyrroline-2-carboxylate (Visser et al., 2012).

Of the three remaining genes (hypT, hypX and hypZ), hypT encodes a 130-amino-acid protein with similarly sized orthologues of unknown function (DUF1232) in many other organisms, hypX encodes a 563-amino-acid protein with sequence similarity to the aconitase/3-isopropylmalate dehydratase family of proteins and hypZ encodes a 143-amino-acid protein that is a member of the Pfam OsmC protein family (P23929) that contains stress-induced proteins including an osmotically induced protein (OsmC) and an organic hydroperoxide detoxification protein (Ohr). It is possible that HypZ may play a role in protecting the cell against reactive oxygen species generated from the oxidation of 4-d-Hyp. In S. meliloti, the hypT and hypXYZ genes are not essential for hydroxyproline catabolism as a hypT mutant and mutants in which hypXYZ expression is eliminated exhibit wild-type growth with 4-Hyp as a carbon source (data not shown). Thus, either these genes are not required for Hyp catabolism, or their functions are redundant with respect to some other activities present in S. meliloti. The global transcriptome of S. meliloti cells utilizing hydroxyproline showed that the 14 hyp genes of the hyp gene cluster were all induced in cells utilizing hydroxyproline. Another operon-like cluster (smb20320–20322) that is also located on the pSymB megaplasmid was also clearly induced in cells utilizing hydroxyproline. This TRAP-T uptake system was previously shown to be induced by hydroxyproline. However, this system does not appear to facilitate the uptake of hydroxyproline (Mauchline et al., 2006; MacLean et al., 2009). A HypR-like binding site was not located in the smb20320 promoter and it seems likely that transcription of the smb20320–20322 genes is regulated by the divergently transcribed gntR family gene smb20323.

Additional genes whose expression was also induced in cells growing with 4-Hyp as carbon source include those encoding the TCA cycle enzymes α-ketoglutarate dehydrogenase (sucA), succinyl–CoA synthetase (sucC and sucD), malate dehydrogenase (mdh), and the gluconeogenic enzymes phosphoenolpyruvate carboxykinase (pckA) and fructose bisphosphate aldolase (fbaB). The induction of these genes and presumably their enzyme activities is entirely consistent with a pathway in which hydroxyproline is catabolized to α-ketoglutaric acid, which is then converted to oxaloacetic acid via TCA cycle enzymes. In S. meliloti, gluconeogenesis initiates via the synthesis of phosphoenolpyruvate from oxaloacetic acid by PCK (pckA) and glyceraldehyde-3-phosphate and dihydroxyacetone phosphate are formed via the lower half of the Embden–Myerhoff–Parnas pathway (Finan et al., 1991) with fructose bisphosphate aldolase (fbaB) catalysing synthesis of fructose-1-6-bisphosphate.

Other genes whose expression was induced upon hydroxyproline catabolism include a gene cluster (cbbFBTALSX) encoding enzymes of the Calvin–Benson cycle CO2 fixation pathway (Pickering and Oresnik, 2008). Perhaps, this induction is linked to a high rate of CO2 evolution in cells catabolizing 4-Hyp. Genes involved in thiamine biosynthesis were also induced (Table 5; thiC, thiO, thiG and thiE) and this induction may reflect increased use of enzymes containing thiamine pyrophosphate (TPP) as a cofactor in cells catabolizing 4-Hyp. α-Ketoglutarate dehydrogenase and transketolase (cbbT) are two such enzymes. A number of reports have linked metabolic demand and thiamine synthesis and there is evidence that TPP present in enzymes undergoes a slow destruction during catalysis (McCourt et al., 2006). Thus, cells experiencing a high flux through pathways that include enzymes with TPP as a cofactor would have an increased demand for thiamine. In Saccharomyces cerevisiae, the DNA-binding protein, Pdc2, appears to regulate glycolytic genes involved in carbon metabolism and genes involved in TPP synthesis (Mojzita and Hohmann, 2006). Thiamine synthesis genes are also regulated by the concentration of intracellular NAD+ via the NAD+-dependent histone deacetylase and, to a lesser extent, Sir2 (Li et al., 2010).

In conclusion, the data for S. meliloti are very consistent with a pathway in which hydroxyproline is metabolized to the TCA cycle intermediate α-ketoglutarate. The gene/function relationships for all of the steps in this pathway are established or inferred and studies should be completed for the HypO, HypD and HypH proteins. The functions for four genes (hypXYZ and hypS) remain to be determined and it is possible these encode for accessory activities related to hydroxyproline catabolism. These studies will aid annotation of hyp gene orthologues in other bacteria.

Experimental procedures

Bacterial strains, plasmids, media and growth conditions

The bacterial strains and plasmids used in this work are listed in Table S1. Ecoli cultures were grown at 37°C in LB (Luria–Bertani media), and Smeliloti cells were grown at 30°C in LB containing 2.5 mM CaCl2 and 2.5 mM MgSO4. M9-succinate medium contained 1 × M9 salts (Difco) supplemented with 0.25 mM CaCl2, 1 mM MgSO4, 0.5 µg ml−1 biotin and 43 nM CoCl2 (10 ng ml−1 CoCl2.6H2O). Carbon sources were added as indicated in the text at the following concentrations (unless indicated otherwise): 0.5% glycerol, 15 mM glucose, 15 mM succinate, 10 mM l-proline or 10 mM trans-4-hydroxy-l-proline, or 10 mM cis-4-hydroxy-l-proline. Amino acids (purchased from Sigma) were added as nitrogen sources to final concentrations of 10 mM. Ammonium chloride or potassium nitrate was added as N sources to final concentrations of 1 mg ml−1 or 0.72 mg ml−1 respectively. Antibiotics were used at the following concentrations: streptomycin, 200 µg ml−1; gentamicin, 60 µg ml−1 (for Ecoli 10 µg ml−1); tetracycline, 10 µg ml−1 (for Ecoli 15 µg ml−1).

DNA and RNA manipulations and microarrays

DNA isolation, transformation, digestion and ligation were performed by standard procedures. Oligonucleotide synthesis (Table S6) and DNA sequencing were performed at the MobixLab (McMaster University, Hamilton, Ontario, Canada).

To identify the transcriptional start sites, RNA isolation and primer extension reactions were performed as described previously (MacLean et al., 2009). Briefly, total RNA was isolated from RmP110 or RmP1724 (a hypR mutant) grown in M9 medium with 0.5% glycerol (and ammonium chloride) or 10 mM 4-l-Hyp as the sole carbon and nitrogen source. A primer for each promoter (CWF196 for PhypR, CWF197 for PhypD, CWF65 for PhypS and CWF195 for PhypH; Table S6) was end-labelled with γ-32P-ATP and following the primer extension reaction, the product was loaded onto a 6% acrylamide, 7 M urea sequencing gel and electrophoresed alongside a sequencing ladder generated by using the same primer with the plasmid pFL2315 (for PhypR), pTH2681 (for PhypD) or pFL866 (for PhypS and PhypH). Microarray chips were purchased from NimbleGen Systems, Madison, Wisconsin. Wild-type S. meliloti (strain RmP110) was cultured in M9 minimal medium containing 0.5% glycerol, or 15 mM 4-l-Hyp, or M9 without NH4Cl and with 15 mM 4-l-Hyp as the sole carbon and nitrogen source. Cells were harvested and RNA was isolated as previously described (MacLellan et al., 2006) from aerated mid-log phase cultures (OD600 0.4 to 0.6). Following extraction, the RNA was further purified and DNA removed using RNeasy minikits (Qiagen).

RNA quality analysis, cDNA synthesis and hybridizations were performed by NimbleGen following their procedures. The custom-made 4-plex arrays contained ∼ 72 K probes of 60 oligonucleotides that targeted sequences within annotated start and end positions of 6269 annotated S. meliloti features, mostly protein-coding sequences, but some RNA sequences. Raw data probe intensities were quantile normalized across all experimental replicates (4 experiments × 2 replicates or 8 arrays). Normalized probe intensities were subjected to a filter that required at least six non-redundant probes per feature. This reduced the number of features analysed to 6096. The median intensity of the pooled, filtered probes within an annotated region was used as an uncorrected measure of gene expression for each experiment. A set of 1786 null probes that did not match any S. meliloti genome sequence was used to estimate background. The probability that the median of gene expression for a feature was above background was estimated by taking 10 000 bootstrap samples. The number of times the median of the bootstrap sample from the feature set was less than the median of the null set sample was used as the estimate of probability. The feature expression was called significantly above background if P < 0.001. A similar bootstrap resampling was made to assess the difference in expression of a feature between two experiments. A difference was called significantly above background if P < 0.01. Differences in expression between experiments were also expressed as ratios of net expression above background, the significance of a ratio being called from the bootstrap call for the expression difference. The microarray data are available at Geo (http://www.ncbi.nlm.nih.gov/geo/) under the Accession Number GSE38087 and an excel spreadsheet of the data are presented in Table S5.

Construction of plasmid integration mutants and reporter gene fusion strains

GusA or LacZ reporter fusions to hypD, hypO and hypZ were constructed in this study (the other fusions were made previously; see Table S1). To generate a gusA fusion to hypD, 507 bp at the 3′ end of hypD was amplified using oligonucleotides CWF104 and 105, and the resulting PCR product was cloned into the PmlI and XbaI sites of pTH1360 (Zaheer et al., 2009) to obtain a hypD::gusA fusion (pTH2690). Single cross-over homologous recombination of pTH2690 into the S. meliloti genome (of strain RmP110) resulted in the fusion of hypD to gusA and preserved a functional copy of the promoter and all genes at this locus. A lacZ fusion to hypO was constructed similarly, except that 520 bp of the 3′ end of hypO was amplified using oligonucleotides CWF135 and CWF189, and the product cloned into the XhoI site of pTH1522 (Cowie et al., 2006) to generate pTH2697. Plasmid pTH2697 was integrated into RmP110 to yield strain RmP2551. A gusA fusion to hypZ was constructed by cloning the 425 bp PCR product (amplified with CWF137 and 138) of the 3′ end of that gene into the XhoI site of pTH1522 to yield pTH2696. Integration of that plasmid into RmP110 yielded strain RmP2543.

To construct a non-polar deletion mutant of hypRE, we used λ-red recombinase and the strains described by Datsenko and Wanner (2000). A 1602 bp PCR product which includes hypRE and flanking sequence was amplified from RmP110 DNA with oligonucleotides CWF34 and 35, and cloned into the XmaI and SacI sites of pUCP30T to generate pTH2534. A 1400 bp PCR product was then amplified from pKD13 using oligonucleotides CWF50 and CWF51. The PCR product contained an FRT-kan-FRT cassette, flanked on either side by approximately 40 bp of sequence corresponding to the ends of the hypRE gene. This PCR product was then incorporated (via electroporation) into strain BW25113 (pKD46) (pTH2534) to replace the hypRE gene in pTH2534 with the FRT-kan-FRT cassette by λ-red recombination. The resulting plasmid (pTH2551) was then integrated into the genome of RmP110 to generate RmP2507, where hypRE is replaced by FRT-kan-FRT in a double cross-over. The kan gene was then removed using FLP (supplied on vector pTH2505), to generate a non-polar deletion mutant of hypRE (strain RmP2508). The plasmid pTH2505 is unstable in S. meliloti and easily cured. All constructs were analysed as in the Datsenko and Wanner (2000) method, and final constructs were checked by PCR and sequencing across the FRT–pSymB junctions.

To complement the hypRE mutant, a plasmid was generated that expressed hypRE from its own promoter (PhypM). The hypM promoter region (802 bp) was amplified with oligonucleotides CWF79 and 81, and cloned into the HindIII site of pTH1582 to generate pTH2574. The hypRE gene was then amplified with oligonucleotides CWF91 and 92 and cloned into the BglII site of pTH2574, downstream of PhypM, to generate pTH2689.

To construct gusA and lacZ fusions to hyp gene promoters in the hypR mutant background, plasmids pTH2690 (hypD::gusA) and pTH2494 (hypM::gusA) were integrated into S. meliloti strain RmP1724 to generate strains RmP2524 and RmP2525 respectively. To examine gusA and lacZ fusions to hyp gene promoters in the hypRE mutant background, plasmids pFL2315 (hypR::gusA), pTH2690 (hypD::gusA), pFL866 (hypS::lacZ and hypH::gusA), pTH2494 (hypM::gusA) and pFL7003 (hypM::lacZ) were each integrated into S. meliloti strain RmP2508 to generate strains RmP2526, RmP2527, RmP2528, RmP2529 and RmP2530 respectively.

To measure β-galactosidase (LacZ) activity and β-glucuronidase (GusA) activity of the reporter fusions described above and in Table S1, assays were performed as previously described, with the carbon and nitrogen sources indicated (MacLean et al., 2006).

Overexpression and purification of recombinant proteins

The hyp genes hypRE and hypY from pSymB (Fig. 1B) were each cloned into pET28a for overexpression in E. coli with a C-terminal His6 tag, from the T7 promoter. To clone hypRE, the gene was amplified with oligonucleotides CWF 7 and 9 (Table S6), and the resulting PCR product was digested with BbsI and HindIII, and cloned into the NcoI and HindIII sites of pET28a, to generate pTH2479. To clone hypY, the gene was amplified with oligonucleotides CWF54 and 55, and the PCR product was cut with BsmBI and HindIII, and cloned into the NcoI and HindIII sites of pET28a, to generate pTH2555.

Each plasmid was introduced into E. coli strain BL21/DE3 (Novagen). For overexpression, each strain was cultured at 30°C with aeration to mid-log phase in LB medium with 200 µg ml−1 ampicillin. Expression of recombinant protein was then induced by the addition of isopropyl-β-d-thiogalactopyranoside 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 20 mM HEPES (pH 7.5), 150 mM KCl and 7.5% glycerol. Cells were disrupted by French press and the lysates were each cleared by centrifugation at 160 000 g for 20 min.

Each of the recombinant proteins was then purified to 95% purity using HisPur Cobalt Resin (Thermo Scientific), using batch elution as described in the manufacturer's instructions. Purified protein was stored at −70°C in buffer containing 20 mM HEPES (pH 7.5), 150 mM KCl, 1 mM DTT, 1 mM EDTA and 50% glycerol. Protein concentrations were determined using Bio-Rad Protein Assay Reagent (Bio-Rad) and bovine serum albumin as a standard.

In vitro assays for hydroxyproline epimerase activity

The purified recombinant HypRE and HypY proteins were tested for HypRE activity and kinetics with 4-l-Hyp or 4-d-Hyp as substrate. Reactions were performed with enzyme at a concentration of 50 nM in a buffer containing 50 mM HEPES (pH 8.0), 5 mM β-mercaptoethanol, 1 mM EDTA and substrate (4-l-Hyp or 4-d-Hyp) at 0.0, 0.5, 0.75, 1.0, 2.0, 4.0 or 8.0 mM. Reaction volumes were 240 µl, and each reaction (at each substrate concentration) was incubated at 30°C. At 30 s intervals, from 0 to 5 min, 20 µl was removed from each reaction and quenched at 95°C for 20 min. In control reactions, heat-killed enzyme (incubated at 95°C for 20 min) was used. No activity was detected in these controls. To detect and quantify activity (conversion of 4-l-Hyp to 4-d-Hyp or vice versa), reactions were derivatized and separated by capillary electrophoresis exactly as previously described (Gavina et al., 2010).

Acknowledgements

We are grateful to Paola Minoprio for sharing information and clones with us and to Jane Fowler, Allyson MacLean, Vladimir Jokic and Guianeya Perez Hernandez for assistance. This work was supported by grants to T. M. F. and P. B. M. from the Natural Sciences and Engineering Council of Canada. T. M. F. also acknowledges support from Genome Canada through the Ontario Genomics Institute and the Ontario Research and Development Challenge Fund.

Ancillary