Identification of Two Distinct Stereoselective Lysine 5-Hydroxylases by Genome Mining Based on Alazopeptin Biosynthetic Enzymes
Graphical Abstract
Enzymes that catalyze regio- and stereoselective hydroxylation of amino acids are valuable for pharmaceutical production. Two lysine 5-hydroxylases (Am_AzpK2 and Pp_AzpK2) that synthesize (2S,5S)- and (2S,5R)-5-hydroxylysine, respectively, were identified by genome mining. Several lysine 5-hydroxylases in the genome database were also characterized. Stereoselective dehydrogenases for (2S,5S)- and (2S,5R)-5-hydroxylysine were also discovered.
Abstract
Enzymes that catalyze regioselective and stereoselective hydroxylation of amino acids are useful tools for the synthesis of pharmaceuticals. AzpK is an unprecedented lysine 5-hydroxylase that is involved in alazopeptin biosynthesis, although its enzymatic activity has not been confirmed in vitro. Here, we identified two α-ketoglutarate/Fe2+-dependent dioxygenases in Actinosynnema mirum and Pseudomonas psychrotolerans (Am_AzpK2 and Pp_AzpK2, respectively) as lysine 5-hydroxylases, using genome mining based on the alazopeptin biosynthetic gene cluster. Interestingly, Am_AzpK2 and Pp_AzpK2 synthesized different isomers, (2S,5S)- and (2S,5R)-5-hydroxylysine, respectively. We also identified two AzpJ homologs as the dehydrogenases that specifically recognize the hydroxy groups of (2S,5S)- and (2S,5R)-5-hydroxylysine to synthesize a keto group. These dehydrogenases were shown to be useful tools for characterizing the stereochemistry of 5-hydroxylysine and evaluating the activity of lysine 5-hydroxylases. Furthermore, we identified three lysine 5-hydroxylases that synthesize (2S,5S)-5-hydroxylysine and four lysine 5-hydroxylases that synthesize (2S,5R)-5-hydroxylysine from the genome database. Genome scanning based on lysine 5-hydroxylases indicated the presence of undiscovered natural products with 5-hydroxylysine moieties. In conclusion, this study provides a fundamental technology for the stereoselective production of 5-hydroxylysine. Further analysis of the stereoselective lysine 5-hydroxylases would reveal how nature establishes highly stereoselective hydroxylation.
Introduction
The hydroxy group is one of the most universal functional groups in natural products, and several hydroxylase families have been discovered, including cytochrome P450,1 FAD-dependent monooxygenase,2 and α-ketoglutarate/Fe2+-dependent dioxygenase (α-KGD).3 Among these hydroxylase families, α-KGD is a useful catalyst because it can catalyze the regioselective and stereoselective hydroxylation of chemically stable methylene using simple redox partners, α-ketoglutarate (α-KG) and Fe2+ ion.4 When α-KGD catalyzes the hydroxylation reaction, one atom of an oxygen molecule is used to hydroxylate the substrate, and the other is incorporated into the co-substrate, α-KG, to form the carboxy group of succinic acid with the release of CO2.3 Importantly, the enzymes belonging to α-KGD are known to catalyze not only hydroxylation but also various chemical reactions, such as halogenation,5, 6 desaturation,7 demethylation,8, 9 cyclopropanation,10, 11 and epoxidation,12 which have attracted the attention of many researchers.13
Because of its usefulness in drug development, many attempts have been made to establish the site-selective hydroxylation of methylene using organic chemistry. However, this is generally challenging.14 Therefore, hydroxylases mentioned above have often been used for the industrial production of chemical compounds containing hydroxy groups. In particular, methylene hydroxylases of amino acids, such as l-Lys,15 l-Ile,16 l-Arg,17 l-Asp,18 and l-Pro,19-22 have been discovered so far, and some of them are used in industry.4 However, no enzyme that stereoselectively hydroxylates the carbon at position 5 (C-5) of free l-Lys has been identified so far. Note that the (2S,5R)-5-hydroxylysine present in collagen and histones is formed by hydroxylation of the lysine side chain during post-translational modification by α-KGD.23, 24 Commercially available 5-hydroxylysine is prepared by organic synthesis; however, stereoselective hydroxylation requires multiple reaction steps.25, 26 It has been reported that 5-hydroxylysine can be used for the synthesis of bengamides, which are potential antitumor agents.27 In addition, (2S,5S)-5-hydroxylysine can be used to produce cis-5-hydroxypipecolic acid, an important precursor of several pharmaceuticals.28 Therefore, identifying enzymes that synthesize 5-hydroxylysine by regioselective and stereoselective l-Lys hydroxylation is important for producing raw materials for pharmaceuticals.
Recently, we discovered the alazopeptin biosynthetic gene cluster (BGC) in two alazopeptin-producing strains (Streptacidiphilus griseoplanus and Kitasatospora azatica) and revealed the complete biosynthetic pathway of alazopeptin, a tripeptide consisting of two molecules of 6-diazo-5-oxo-l-norleucine (DON) and one molecule of l-Ala.29, 30 The initial reaction for DON biosynthesis was shown to be the hydroxylation of free l-Lys at position C-5 by AzpK to synthesize 5-hydroxylysine by gene inactivation of azpK and a feeding experiment of 5-hydroxylysine (a mixture of four different stereoisomers) to the ▵azpK mutant (Figure 1A).29 However, AzpK (WP_035850923.1, BCN13447.1) shows no homology to known oxidases and its enzymatic activity has not been observed in vitro using a recombinant enzyme. Therefore, the catalytic mechanism and redox partner requirements remain unknown.29 In addition, the stereoselectivity of hydroxylation catalyzed by AzpK remains unclear.29 Therefore, although AzpK catalyzes attractive reactions, its application in material production and other areas is difficult.
The biosynthetic pathway for DON in alazopeptin biosynthesis and comparison of azp-related gene clusters among the three species. (A) The biosynthetic pathway for DON in alazopeptin biosynthesis. The first step in this pathway is the hydroxylation of l-Lys at position C-5 by AzpK, which belongs to a novel family of oxidases, in S. griseoplanus. In A. mirum and P. psychrotolerans, the same reaction is supposed to be catalyzed by α-KGD (AzpK2). (B) Comparison of azp-related gene clusters among S. griseoplanus, A. mirum, and P. psychrotolerans. There are no azpK (green) homolog genes in the genomes of A. mirum and P. psychrotolerans; instead, α-KGD genes (azpK2, red) are present in the azp-related gene clusters. AzpL homologs and AzpJ homologs are shown in yellow and orange, respectively.
In this study, we identified α-KGDs (named AzpK2) that catalyze the stereoselective hydroxylation of l-Lys at position C-5 to synthesize 5-hydroxylysine from the BGCs that are homologous to the alazopeptin BGC (azp cluster) but contain α-KGD genes instead of azpK. Interestingly, AzpK2 enzymes from Actinosynnema mirum and Pseudomonas psychrotolerans showed different stereoselectivities and synthesized (2S,5S)- and (2S,5R)-5-hydroxylysine, respectively. We also characterized two AzpJ homologs (5-hydroxylysine dehydrogenases), which stereoselectively oxidized (2S,5S)- and (2S,5R)-5-hydroxylysine to produce 5-oxolysine. Furthermore, using AzpJ homologs with different substrate specificities, we identified three and four, S and R-selective lysine 5-hydroxylases, respectively, from the genome database. These lysine 5-hydroxylases can be used for the enzymatic synthesis of the two isomers of 5-hydroxylysine.
Results and Discussion
Comparative Analysis of Gene Clusters Homologous to the azp Cluster
In our previous study, we searched for BGCs homologous to the azp cluster in the genome database and found that azpJ, azpK, and azpL homologs, which probably encode the proteins responsible for DON biosynthesis from l-Lys and nitrous acid, are highly conserved in many bacteria, consisting of a variety of BGCs.29 However, we also found exceptional gene organization; the related BGCs in Actinosynnema mirum JCM3225 and Pseudomonas psychrotolerans NS383 contain a putative α-KGD gene instead of azpK (Figure 1, Tables S1, S2).31, 32 The BGC of A. mirum seems to direct the biosynthesis of alazopeptin because this BGC has all azp gene homologs except for azpK (Figure 1 and Table S1). In contrast, the BGC of P. psychrotolerans has no other azp gene homologs, indicating that this gene cluster directs the biosynthesis of DON or DON derivatives (Figure 1 and Table S2). We named the α-KGD gene in these two BGCs as azpK2. The name of the bacterial species is abbreviated and prefixed with the gene name hereafter, for example Am_azpK2 and Pp_azpK2 for azpK2 genes in A. mirum and P. psychrotolerans, respectively. Because hydroxylation of l-Lys is essential for the biosynthesis of DON, and these two strains possess homologs of the other genes (azpJ and azpL) required for DON biosynthesis, we expected that AzpK2 would catalyze the same reaction as AzpK.
To investigate whether the putative alazopeptin BGC of A. mirum could produce alazopeptin or DON derivatives, we cultured A. mirum, which was obtained from the Japan Collection of Microorganisms (JCM), and analyzed its metabolite. After considering various conditions, including culture medium and temperature, we succeeded in detecting alazopeptin and N-acetylated DON (AcDON) in the culture medium of A. mirum, demonstrating that this strain contains all the genes necessary for the biosynthesis of alazopeptin (Figure S1). No azpK homolog was found in A. mirum genome, which supports our hypothesis that Am_AzpK2 catalyzes the 5-hydroxylation of l-Lys.
In vitro Analysis of Am_AzpK2 and Pp_AzpK2
To confirm that Am_AzpK2 and Pp_AzpK2 catalyze the hydroxylation of l-Lys at position C-5, we prepared recombinant enzymes (Figure S2) and performed in vitro assays. As expected, both Am_AzpK2 and Pp_AzpK2 synthesized 5-hydroxylysine using l-Lys and α-KG (Figure 2B). Next, we determined the absolute configuration of 5-hydroxylysine synthesized by each enzyme. The enzymatically synthesized 5-hydroxylysine was converted to 5-hydroxypipecolic acid (5-HPA) by ornithine cyclodeaminase (OCD) and derivatized by Nα-(5-fluoro-2,4-dinitrophenyl)-l-alaninamide (FDAA) as described previously.33 The absolute stereochemistry of 5-HPA was determined by comparing the retention time with commercially available standards (Figure S3). As a result, the absolute configurations of the 5-hydroxylysines synthesized by Am_AzpK2 and Pp_AzpK2 were revealed to be (2S,5S) and (2S,5R), respectively (Figure 2A). Surprisingly, Am_AzpK2 and Pp_AzpK2 have different stereoselectivities, although they appear to be involved in the biosynthesis of the same compound, DON (Figure 2A).
In vitro analysis of Am_AzpK2, Pp_AzpK2, Sg_AzpJ, and Pp_AzpJ. (A) Schematic representation of the reactions clarified in this study. (B) In vitro analysis of Am_AzpK2 and Pp_AzpK2. Each enzyme synthesizes 5-hydroxylysine using l-Lys and α-KG as substrates. EICs for m/z=163.2, which corresponds to [M+H]+ of 5-hydroxylysine, are shown. A mixture of four stereoisomers of 5-hydroxylysine was used as the standard. (C) In vitro analysis of Sg_AzpJ and Pp_AzpJ. EICs for m/z=143.2, which corresponds to [M+H]+ of cyclodehydrated 5-oxolysine, are shown. Similar to our previous study, 5-oxolysine was not detected.29 Cyclodehydrated 5-oxolysine was synthesized when Am_AzpK2 and Sg_AzpJ reacted together, and when Pp_AzpK2 and Pp_AzpJ reacted together. These results indicate that Sg_AzpJ only recognizes (2S,5S)-5-hydroxylysine and Pp_AzpJ only recognizes (2S,5R)-5-hydroxylysine.
In vitro Analysis of AzpJ Homologs
In our previous study, AzpJ from Streptacidiphilus griseoplanus (Sg_AzpJ) was shown to catalyze the dehydrogenation of the hydroxy group of 5-hydroxylysine using NAD+ as an oxidant.29 However, it was not clarified whether Sg_AzpJ stereoselectively oxidizes the hydroxy group of 5-hydroxylysine because we could only purchase 5-hydroxy-dl-lysine, which is a mixture of four different stereoisomers, for in vitro analysis.29 To determine the stereoselectivity of Sg_AzpJ, we incubated recombinant Sg_AzpJ (Figure S2) with NAD+ and 5-hydroxylysine synthesized using Am_AzpK2 or Pp_AzpK2. LC–MS analysis of the reaction mixture showed that the cyclodehydrated form of 5-oxolysine was synthesized with Am_AzpK2, but not with Pp_AzpK2 (Figure 2C). Quantification of NADH by measuring absorbance at 340 nm confirmed that NADH was formed only when Am_AzpK2 was included in the reaction mixture (Figures 3A, S4A). Moreover, Sg_AzpJ did not recognize commercially available (2S,5R)-5-hydroxylysine as a substrate but produced 5-oxolysine from 5-hydroxy-dl-lysine (a mixture of isomers) (Figure S4B). These results indicated that Sg_AzpJ recognizes (2S,5S)-5-hydroxylysine, but not (2S,5R)-5-hydroxylysine, as a substrate (Figure 2A).
In vitro analysis of Am_AzpK2 and its homologs incubated with Sg_AzpJ or Pp_AzpJ. (A) The relative yield of NADH was quantified by measuring the increase in absorbance of the reaction mixture at 340 nm. Because Sg_AzpJ stereoselectively recognizes (2S,5S)-5-hydroxylysine and synthesizes 5-oxolysine by oxidation with NAD+, (2S,5S)-5-hydroxylysine synthesized by AzpK2 can be quantified by quantifying NADH. (B) The same assay was performed using Pp_AzpJ. Because Pp_AzpJ stereoselectively recognizes (2S,5R)-5-hydroxylysine and oxidizes it with NAD+, the yield of (2S,5R)-5-hydroxylysine synthesized by AzpK2 can be quantified by quantifying NADH. Error bars represent standard error (n=3).
Because Pp_AzpK2 synthesized (2S,5R)-5-hydroxylysine and Sg_AzpJ seemed to have strict substrate specificity, we assumed that the AzpJ homolog (Pp_AzpJ, 31.4 % amino acid sequence identity compared with Sg_AzpJ) in P. psychrotolerans specifically dehydrogenates (2S,5R)-5-hydroxylysine. Recombinant Pp_AzpJ was prepared (Figure S2) and the stereoselectivity of Pp_AzpJ was examined by performing the same experiments described above. As expected, the cyclodehydrated form of 5-oxolysine and NADH were detected only in the reaction mixture containing Pp_AzpK2 (Figures 2C, 3B, S4A). In addition, Pp_AzpJ catalyzed the reaction with commercially available (2S,5R)-5-hydroxylysine (Figure S4B). These results indicated that Pp_AzpJ recognizes (2S,5R)-5-hydroxylysine as a substrate to form 5-oxolysine (Figure 2A).
The AzpJ homologs of K. azatica (Ka_AzpJ) and A. mirum (Am_AzpJ) show high sequence identity (83.3 % and 73.4 %, respectively) with Sg_AzpJ, indicating that these two enzymes are likely to have the same function. Recombinant Ka_AzpJ and Am_AzpJ were prepared (Figure S2) and their stereoselectivity were examined. The results showed that Ka_AzpJ and Am_AzpJ also had the same stereoselectivity as Sg_AzpJ (Figure S4). These results suggested that Sg_AzpJ (Ka_AzpJ or Am_AzpJ) and Pp_AzpJ can be used to determine the stereochemistry of the hydroxy group at position C-5 of 5-hydroxylysine.
Mining and Functional Analysis of AzpK2 Homologs in the Database
To understand the distribution of lysine 5-hydroxylases, we analyzed AzpK2 homologs in the database. Using a BLAST search with the amino acid sequences of Am_AzpK2 and Pp_AzpK2 as queries, we obtained approximately 1,100 α-KGDs and carried out a sequence similarity network (SSN) analysis using the EFI sequence similarity tool (Figure 4).34, 35 The result of SSN analysis indicated that there were approximately 100 sequences classified in the same group as Am_AzpK2, which showed relatively high amino acid sequence similarity (>40 % identity) to Am_AzpK2 (Table 1). In contrast, no sequence was classified into the same group as Pp_AzpK2 (Figure 4). We were interested in the functions of enzymes classified in the same group as Am_AzpK2. We chose seven enzymes with different degrees of similarity to Am_AzpK2 estimated from the phylogenetic tree generated from the approximately 100 sequences described above (Figures S5 and S6): Al_AzpK2 from Actinoplanes lobatus, Lg_AzpK2 from Lentzea guizhouensis, Sd_AzpK2 from Saccharothrix deserti, Np_AzpK2 from Nocardia puris, Ns_AzpK2 from Nonomuraea soli, Sa_AzpK2 from Sphaerisporangium album, and Sl_AzpK2 from Streptomyces luteocolor. We assumed that these seven enzymes should also catalyze the hydroxylation of l-Lys. To test this hypothesis, the recombinant enzymes (Figure S2) were incubated with l-Lys and α-KG. As expected, the formation of 5-hydroxylysine was confirmed by LC–MS analysis (Figure S7). Therefore, all the enzymes classified in the same group as Am_AzpK2 by SSN analysis (Figure 4) were expected to synthesize 5-hydroxylysine.
Sequence similarity network analysis of α-KGDs generated using EFI sequence similarity tools.34, 35 Approximately 1,100 sequences obtained by the BALST search using Am_AzpK2 and Pp_AzpK2 as queries were used for the analysis. The network was generated by using an SSN score of 70. The SSN score was set to a value as small as possible to distinguish between AzpK2 homologs and the two related halogenases, BesD and HalB.
name |
product |
kcat (min−1) |
Km (μM) |
kcat/Km (min−1 μM−1) |
accession |
Identity[a] |
|---|---|---|---|---|---|---|
Am_AzpK2 |
(2S,5S)-5-hydroxylysine |
4.51±0.07 |
84.8±2.2 |
0.053 |
WP_049796942.1 |
100 % |
Pp_AzpK2 |
(2S,5R)-5-hydroxylysine |
9.31±0.17 |
65.0±2.3 |
0.14 |
WP_153007390.1 |
22.5 % |
Al_AzpK2 |
(2S,5S)-5-hydroxylysine |
10.2±0.50 |
97.4±13 |
0.11 |
WP_189638256.1 |
62.2 % |
Lg_AzpK2 |
(2S,5S)-5-hydroxylysine |
17.2±0.45 |
83.0±6.1 |
0.21 |
WP_065917003.1 |
59.9 % |
Sd_AzpK2 |
(2S,5S)-5-hydroxylysine |
7.87±0.48 |
366±55 |
0.022 |
WP_158841566.1 |
59.6 % |
Np_AzpK2 |
(2S,5R)-5-hydroxylysine |
3.60±0.21 |
56.7±9.7 |
0.064 |
WP_051047386.1 |
41.5 % |
Ns_AzpK2 |
(2S,5R)-5-hydroxylysine |
12.6±0.20 |
250±3.9 |
0.050 |
WP_181610242.1 |
47.5 % |
Sa_AzpK2 |
(2S,5R)-5-hydroxylysine |
11.5±0.14 |
90.0±3.8 |
0.13 |
WP_114027730.1 |
49.0 % |
Sl_AzpK2 |
(2S,5R)-5-hydroxylysine |
24.3±3.0 |
980±170 |
0.025 |
WP_055702164.1 |
50.6 % |
- [a] amino acid sequence identity with Am_AzpK2.
To determine the stereoselectivity of these seven enzymes, we exploited the coupling reaction with Sg_AzpJ or Pp_AzpJ and monitored the formation of NADH, which indicated 5-hydroxylysine oxidation (Figure 3). If NADH formation was observed only with Sg_AzpJ, this result indicated that (2S,5S)-5-hydroxylysine was synthesized by AzpK2. In contrast, if NADH formation was observed only with Pp_AzpJ, this result indicated that (2S,5R)-5-hydroxylysine was synthesized by AzpK2. Interestingly, the enzymes showing relatively high sequence similarity to Am_AzpK2 (Al_AzpK2, Lg_AzpK2, and Sd_AzpK2) synthesized (2S,5S)-5-hydroxylysine, whereas others (Np_AzpK2, Ns_AzpK2, Sa_AzpK2, and Sl_AzpK2) synthesized (2S,5R)-5-hydroxylysine (Figure 3). Thus, enzymes classified in the same group as Am_AzpK2 by SSN analysis (Figure 4) included two distinct stereoselective lysine 5-hydroxylases, although they are similar in their amino acid sequences (>40 % identity, Figure S6).
Kinetic Analysis of AzpK2 Homologs
To obtain further insights into the activity of AzpK2 homologs, kinetic analysis toward its substrate (l-Lys) was performed for all AzpK2 homologs identified in this study; the initial velocity was determined by observing NADH formed by the coupling reaction with Sg_AzpJ or Pp_AzpJ. The kcat/Km values of AzpK2 homologs ranged from ~2.0×10−2–~2.0×10−1 μM−1 min−1 (Figure S8 and Table 1). These kcat/Km values were reasonable compared to the kcat/Km value of lysine 4-hydroxylase (6.6×10−2 μM−1 min−1), which shows amino acid sequence similarity to Am_AzpK2 (32.2 % identity) and Pp_AzpK2 (24.1 %) (Figure 4).36 Among the (2S,5S)-5-hydroxylysine-synthesizing AzpK2 homologs, Lg_AzpK2 showed the highest kcat/Km value, while Pp_AzpK2 showed the highest value among the (2S,5R)-5-hydroxylysine-synthesizing AzpK2 homologs (Figure S8 and Table 1). Evaluation of 5-hydroxylysine synthesis activity using Sg_AzpJ or Pp_AzpJ was simple and accurate. Therefore, further evaluation of the activity of AzpK2 homolog enzymes in the genome database using this technique may lead to the identification of enzymes more suitable for industrial use for 5-hydroxylysine production.
Determination of the Stereoselectivity of HalB, which Synthesizes 5-Hydroxylysine as a Byproduct
Although AzpK2 is the first 5-hydroxylysine synthase that has been demonstrated to have in vitro hydroxylation activity, a lysine halogenase (HalB) that synthesizes 5-chlorolysine was reported to synthesize a trace amount of 5-hydroxylysine as a byproduct.6 Presumably because of its low yield, the stereochemistry of the 5-hydroxylysine synthesized by HalB was not determined. To prove the usefulness of Sg_AzpJ and Pp_AzpJ, we determined the absolute configuration of the 5-hydroxylysine synthesized by HalB using these stereoselective 5-hydroxylysine oxidases. Because the activity of several HalB homologs from different species was examined in a previous study,6 we targeted three of these homologs (Sa_HalB from Streptomyces afghaniensis, Si_HalB from Streptomyces iranensis, and Sw_HalB from Streptomyces wuyuanensis) in our experiments. First, the amount of NADH in the reaction mixture containing HalB and Sg_AzpJ or Pp_AzpJ was measured. However, no significant difference was detected, presumably due to the low amount of 5-hydroxylysine. Therefore, the 5-oxolysine synthesized by HalB and AzpJ was detected by LC–MS after derivatization with 9-fluorenylmethyl chloroformate (Fmoc-Cl). The results indicated that a larger amount of 5-oxolysine was produced in the presence of Sg_AzpJ than Pp_AzpJ, suggesting that HalB stereoselectively synthesizes (2S,5S)-5-hydroxylysine as a byproduct (Figure S9). Because the amount of 5-oxolysine in the presence of Pp_AzpJ appeared to be similar to that in the negative control without Pp_AzpJ, we concluded that only (2S,5S)-5-hydroxylysine was produced by HalB homologs. It is possible that a small amount of 5-oxolysine was produced even in the absence of AzpJ, because HalB homologs seem to have weak activity to further oxidize 5-hydroxylysine and/or 5-chlorolysine.6
Bioinformatic Analysis of BGCs that Contain a Lysine 5-Hydroxylase Gene
As described above, α-KGDs classified in the same group as Am_AzpK2 by SSN analysis were predicted to be lysine 5-hydroxylases. Most of these α-KGDs are encoded in the genomes of actinomycetes, and the α-KGD genes are expected to be included in putative secondary metabolite BGCs (Table S3). However, to the best of our knowledge, only a few natural products contain 5-hydroxylysine as a backbone, such as bengamide37 and odilorhabdin,38, 39 in addition to alazopeptin.29 Therefore, these BGCs are expected to be involved in the biosynthesis of unknown natural products. BGCs with these putative lysine 5-hydroxylase genes were analyzed using antiSMASH40 and BiG-SCAPE (Figure S10).41 Most of the BGCs did not show high similarity to known BGCs in the MiBIG database.42 In addition, azpJ homologs appeared to exist only in the BGCs that were homologous to azp BGC; these BGCs were assumed to synthesize DON derivatives. Among the 112 BGCs analyzed by BiG-SCAPE, 57 BGCs seemed to have no key enzyme genes for the biosynthesis of major secondary metabolites (e. g., polyketide and nonribosomal peptide, Figure S10). Among the remaining BGCs, BGCs that encode nonribosomal peptide synthetase (NRPS) were the most abundant (27 BGCs). There were also BGCs with ribosomally synthesized and post-translationally modified peptide (RiPP)-related genes (9 BGCs), polyketide synthase genes (4 BGCs), and PKS-NRPS hybrid genes (4 BGCs). These observations indicated that several microorganisms produce natural products with 5-hydroxylysine or its derivatives as a building block, although these compounds have not been discovered. These natural products will be discovered in the future by genome mining using the lysine 5-hydroxylase gene as a target.
Discussion
In this study, we identified two distinct stereoselective lysine 5-hydroxylases using genome mining based on alazopeptin biosynthetic enzymes.29, 30 In addition, Sg_AzpJ (Ka_AzpJ and Am_AzpJ) and Pp_AzpJ were shown to catalyze the stereoselective oxidation of (2S,5S)-5-hydroxylysine and (2S,5R)-5-hydroxylysine, respectively, to yield 5-oxolysine. The stereoselectivity of Sg_AzpJ suggests that AzpK synthesizes (2S,5S)-5-hydroxylysine in alazopeptin biosynthesis in S. griseoplanus and K. azatica, which was unclear in our previous study.29
From an evolutionary viewpoint, DON BGCs exhibit two interesting features. First, two different families of lysine 5-hydroxylases are used for the initial reaction in different bacteria; AzpK is a novel family of hydroxylases, whereas AzpK2 belongs to α-KGD. Second, the stereoselectivities of the set of enzymes (AzpK [or AzpK2] and AzpJ) involved in the first two reactions to yield 5-oxolysine are different between actinobacteria and Pseudomonas. Two possible scenarios are assumed to explain these features. The first scenario includes acquiring a gene with the same function, followed by the loss of the original gene. Since azpK, but not the α-KGD gene (azpK2), is present in most BGCs containing putative DON biosynthetic genes in the genome database, the ancestor of the DON biosynthetic gene clusters is expected to contain azpK.29 In some strains, such as A. mirum and P. psychrotolerans, it is assumed that an α-KGD gene was transferred from the outside of the BGC and azpK was deleted during evolution. The second scenario is based on convergent evolution. In this case, each bacterium individually obtained the lysine 5-hydroxylase gene, followed by evolution of the surrounding genes so that DON could be synthesized by chance. While the second scenario can easily explain the common stereoselectivity of cognate AzpK (or AzpK2) and AzpJ, it is difficult to explain the conservation of azpL, which encodes a diazotase, the most important enzyme in DON biosynthesis. In contrast, the first scenario can explain the conservation of azpL, but it is difficult to explain the occurrence of the two different stereoselectivities of the first two enzymes (AzpK/AzpK2 and AzpJ), although there seems to be no doubt that these two enzymes have co-evolved to catalyze the sequential reactions to produce 5-oxolysine from l-Lys. Thus, we cannot provide a clear conclusion regarding the evolution of DON BGCs. However, our study provides interesting insights into how secondary metabolite BGCs have been organized evolutionarily.
We focused on enzymes classified in the same group as Am_AzpK2 by SSN analysis (Figure 4) and showed that they included two distinct stereoselective lysine 5-hydroxylases: one group catalyzing the synthesis of (2S,5S)-5-hydroxylysine and the other synthesizing (2S,5R)-5-hydroxylysine (Figure S6). Since these enzymes have relatively high amino acid sequence similarities to each other (>40 % identity, Figure S6), further analysis of these enzymes should provide important insights into how these enzymes control stereoselectivity. Although it is difficult for chemical synthesis to achieve stereo- and regioselective introduction of hydroxy groups to chemically stable methylene,14 stereoselectivity and regiocontrol are essential for the production of pharmaceuticals and other useful compounds. Therefore, a better understanding of the reaction mechanism of AzpK2 enzymes should contribute to material production through enzyme engineering.
We successfully developed a simple method for determining the absolute stereochemistry of 5-hydroxylysine using two stereoselective 5-hydroxylysine dehydrogenases, Sg_AzpJ and Pp_AzpJ. Using these tools, we were able to rapidly identify the products of Am_AzpK2 homologs and evaluate their activity in detail, including kinetic analysis. This approach should enable high-throughput evaluation of 5-hydroxylysine synthesis activity during engineering of AzpK2 enzymes to obtain a variant with higher activity that is useful for the stereoselective production of 5-hydroxylysine. The usefulness of these tools was further proved by the determination of the stereochemistry of 5-hydroxylysine synthesized by HalB as a byproduct. Because only a trace amount of 5-hydroxylysine was produced by HalB, this model experiment emphasized the usefulness of our new method using Sg_AzpJ and Pp_AzpJ.
Conclusions
Genome mining based on the alazopeptin BGC followed by in vitro analysis of each enzyme led to the discovery of two distinct stereoselective α-KGDs that catalyze the hydroxylation of methylene at position C-5 of l-Lys. Furthermore, we identified NAD+-dependent dehydrogenases that selectively oxidize the hydroxy group of either (2S,5S)- or (2S,5R)-5-hydroxylysine. Using these dehydrogenases, a high-throughput method for evaluating lysine 5-hydroxylase activity was established. We believe that this study not only provides a fundamental technology for the stereoselective production of 5-hydroxylysine but also contributes to a better understanding of the diversity of microbial biosynthetic pathways.
Acknowledgments
This work was supported by Grants-in-Aid for Scientific Research on Innovative Areas (19H04645 to Y. K., 19H05685 to Y. O.), Grant-in-Aid for Transformative Research Areas (22H05130 to Y. K.), Grant-in-Aid for JSPS Fellows (22KJ1046 to S. K.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan, a grant (Development of production technology for bio-based products to accelerate the realization of carbon recycling) from NEDO (to Y. O.), the World-leading Innovative Graduate Study Program in Proactive Environmental Studies, The University of Tokyo (to S. K.), and JSPS A3 Foresight Program grant (to Y. O.).
Conflict of Interests
S. K., Y. K., and Y. O. are co-inventors of a patent application (PCT/JP2023/218522) filed by The University of Tokyo and API Corporation relating to work in this manuscript.
Open Research
Data Availability Statement
The data that support the findings of this study are available in the supplementary material of this article.
