Discovery of Lysine Hydroxylases in the Clavaminic Acid Synthase-Like Superfamily for Efficient Hydroxylysine Bioproduction

ABSTRACT

Hydroxylation via C—H bond activation in the absence of any harmful oxidizing reagents is technically difficult in modern chemistry. In this work, we attempted to generate pharmaceutically important hydroxylysine from readily available l-lysine with l-lysine hydroxylases from diverse microorganisms. Clavaminic acid synthase-like superfamily gene mining and phylogenetic analysis led to the discovery of six biocatalysts, namely two l-lysine 3S-hydroxylases and four l-lysine 4R-hydroxylases, the latter of which partially matched known hydroxylases. Subsequent characterization of these hydroxylases revealed their capacity for regio- and stereoselective hydroxylation into either C-3 or C-4 positions of l-lysine, yielding (2S,3S)-3-hydroxylysine and (2S,4R)-4-hydroxylysine, respectively. To determine if these factors had industrial application, we performed a preparative production of both hydroxylysines under optimized conditions. For this, recombinant l-lysine hydroxylase-expressing Escherichia coli cells were used as a biocatalyst for l-lysine bioconversion. In batch-scale reactions, 531 mM (86.1 g/liter) (2S,3S)-3-hydroxylysine was produced from 600 mM l-lysine with an 89% molar conversion after a 52-h reaction, and 265 mM (43.0 g/liter) (2S,4R)-4-hydroxylysine was produced from 300 mM l-lysine with a molar conversion of 88% after 24 h. This report demonstrates the highly efficient production of hydroxylysines using lysine hydroxylases, which may contribute to future industrial bioprocess technologies.

IMPORTANCE The present study identified six l-lysine hydroxylases belonging to the 2-oxoglutarate-dependent dioxygenase superfamily, although some of them overlapped with known hydroxylases. While the substrate specificity of l-lysine hydroxylases was relatively narrow, we found that (2S,3S)-3-hydroxylysine was hydroxylated by 4R-hydroxylase and (2S,5R)-5-hydroxylysine was hydroxylated by both 3S- and 4R-hydroxylases. Moreover, the l-arginine hydroxylase VioC also hydroxylated l-lysine, albeit to a lesser extent. Further, we also demonstrated the bioconversion of l-lysine into (2S,3S)-3-hydroxylysine and (2S,4R)-4-hydroxylysine on a gram scale under optimized conditions. These findings provide new insights into biocatalytic l-lysine hydroxylation and thus have a great potential for use in manufacturing bioprocesses.

INTRODUCTION

A method for C—H bond functionalization is considered a promising technology for its potential to alter fundamental synthetic strategies in organic chemistry (1); however, this is a technically challenging task. Biocatalysts, such as hydroxylases in living organisms, functionalize C—H bonds to yield C—OH bonds via oxygenation. This reaction, known as hydroxylation, is advantageous because of its high chemoselectivity, regioselectivity, and stereoselectivity.

Recently, an increasing demand has been placed on developing a hydroxylation technique to establish a manufacturing process with reduced costs, energy expenditures, and waste (2, 3). Over the years, extensive research on amino acid hydroxylases has been undertaken in pursuit of this goal. Recent microbial studies have characterized several nonaromatic amino acid hydroxylases, such as proline hydroxylases (4–7), arginine hydroxylase (VioC) (8), asparagine hydroxylase (AsnO) (9), isoleucine hydroxylase (IDO) (10), and N-succinylamino acid hydroxylase (SadA) (11). These biocatalysts belong to the 2-oxoglutarate-dependent dioxygenase superfamily and have a tremendous potential in modern bioindustry for their robust activity in comparison to that of other oxygenases requiring a partner reductase.

Recent whole-genome sequencing studies have facilitated the identification of potential amino acid hydroxylases. Hydroxylysine (Hyl), the basic hydroxylated amino acid, is found in a particular type of collagen peptide that is abundant in the extracellular matrix, where it stabilizes the collagen scaffold by subsequent O-glycosylation. l-Lysine residue hydroxylation is a posttranslational modification catalyzed by lysyl hydroxylase (EC 1.14.11.4). Unlike hydroxyproline, which is found in free and peptide-bound forms, little is known about the occurrence and physiological roles of free Hyl. In terms of chemical synthesis, a need exists for several Hyl regio- and stereoisomers, owing to their significant utility as a synthon for pharmaceutical agents and to expand our understanding of Hyl biological behavior. For example, (2S,3R)-3-Hyl serves as an intermediate during the synthesis of (−)-balanol, a potent protein kinase C inhibitor isolated from Verticillium balanoides metabolite (12, 13). (2S,4S)-4-Hyl and (2S,4R)-4-Hyl are promising precursors of functionalized piperidine-2-ones, which are highly versatile building blocks for the synthesis of many bioactive substances (14). (2S,5R)-5-Hyl is the most abundant Hyl isomer and a constituent of collagen peptides; therefore, an in-depth physiological understanding of Hyl may be valuable for collagen-relevant medicinal chemistry (15). Several stereoselective syntheses have sought to prepare a wide variety of optically pure Hyls; however, most chemical syntheses suffer from relatively low yield and poor selectivity (15), further emphasizing the need for a more efficient method.

The present study focused on biocatalytic Hyl synthesis to overcome the drawbacks of current methods. We have previously identified the six free lysine hydroxylases (16, 17), some of which were also recently reported (18); however, little was known about free l-lysine hydroxylases at the start of this work. Thus, we sought to identify novel amino acid hydroxylases capable of free l-lysine hydroxylation. Clavaminic acid synthase (CAS) (Fig. 1a) (19) and VioC (Fig. 1b) (8) are essential parts of biocatalysts that act on C—H bonds in basic amino acids to form clavulanic acid (20) and viomycin (21), respectively. Based on these biocatalysts, we constructed a phylogenetic tree comprised of 93 CAS-like superfamily members and then selected 36 representative genes for heterologous expression in Escherichia coli to assess their capacity for amino acid hydroxylation. Most significantly, we also demonstrated a one-step approach to synthesize (2S,3S)-3-Hyl and (2S,4R)-4-Hyl from l-lysine with an excellent conversion efficiency by using l-lysine hydroxylase-expressing E. coli as a whole-cell biocatalyst.

FIG 1.

Regio- and stereoselective hydroxylation reactions of basic amino acids. (a) Clavaminic acid synthase (CAS; EC 1.14.11.21); (b) l-arginine 3S-hydroxylase (VioC; EC 1.14.11.41); (c) l-lysine 3S-hydroxylase and 4R-hydroxylase found in this study.

RESULTS AND DISCUSSION

Classification of functionally unidentified CAS-like superfamily members.

CAS and VioC exhibit ∼32% amino acid sequence identity and have similar reaction mechanisms and substrate structures. Thus, we hypothesized that novel lysine-hydroxylating biocatalysts would be included in functionally unknown proteins and share a similar sequence identity to CAS and VioC. Figure 2 shows the distribution of VioC-related proteins in terms of amino acid sequence and CAS-like superfamily (cd00250) proteins according to the conserved protein domain database (22). Each annotated protein shown in the tree is summarized in Table S1 in the supplemental material. Furthermore, the two histidine residues and one glutamic acid residue thought to mediate hydroxylase activity were conserved throughout the identified proteins. Thus, of the 93 genes satisfying the selection criteria, 36 representative factors were cloned and individually expressed in E. coli.

FIG 2.

Phylogenetic analysis of clavaminic acid synthase-like superfamily proteins. The selected candidates are indicated in blue. The scale bar indicates the number of amino acid substitutions per site.

Detection of l-lysine hydroxylation activity.

Based on the phylogenetic analysis, we classified and cloned a wide spectrum of 36 experimentally unidentified proteins in addition to VioC (Fig. 2). The protein genes were heterologously expressed in E. coli Rosetta 2(DE3) and then evaluated for l-lysine hydroxylation activity in whole-cell reactions. High-performance liquid chromatography (HPLC) analysis revealed six reactions with a significant dose-dependent decrease in l-lysine corresponding with an increase of an unidentified product, in addition to the VioC control. The six positive proteins, namely those deposited under GenBank accession numbers ABS05421, EAR24255, ABQ06186, ACU60313, AEV99100, and EFK34737 and henceforth referred to as K3H-1, K3H-2, K4H-1, K4H-2, K4H-3, and K4H-4, respectively, were purified and showed apparent homogeneity, as judged by 12.5% SDS-PAGE and Coomassie brilliant blue staining (Fig. S1). All proteins were produced as C-terminally His₆-tagged forms with the exception of K3H-1, which failed to purify and thus was generated with an N-terminal His₆ tag.

l-Lysine hydroxylation reactions were carried out as stated above using purified proteins instead of E. coli whole cells, and the supernatants were analyzed by HPLC. This revealed l-lysine hydroxylation activity for all seven reactions; however, clearly distinct retention times were observed as follows: the products of K3H-1, K3H-2, and VioC appeared in 14.35 min, and the products of K4H-1, K4H-2, K4H-3, and K4H-4 appeared in 14.69 min (Fig. 3). Furthermore, the products gained a molecular mass of 16 as determined by high-resolution mass spectrometry (HR-MS), identical to the calculated mass of Hyls (Table 1). No further hydroxylated products, such as dihydroxylysine, were detected. Thus, these analyses identified six novel microbial l-lysine hydroxylases and showed that VioC displayed slight l-lysine hydroxylation activity (Fig. S2), despite a previous report contradicting this finding (23).

FIG 3.

HPLC chromatogram of l-lysine and biocatalytically synthesized hydroxylysines (Hyls). Solid line, (2S,3S)-3-Hyl; dashed line, (2S,4R)-4-Hyl; dotted line, l-lysine.

TABLE 1.

HR-MS analysis of Hyls obtained from l-lysine

Hydroxylase	Mass-to-charge ratio (m/z)
	[M+H]+	[M+Na]+
VioC	163.1085	185.0904
K3H-1	163.1083	185.0905
K3H-2	163.1085	185.0904
K4H-1	163.1086	185.0905
K4H-2	163.1086	185.0906
K4H-3	163.1085	185.0904
K4H-4	163.1084	185.0903

Stereochemistry of Hyls.

The hydroxylated l-lysine products were confirmed by HR-MS analysis, although the retention times on chromatograms were not identical to those of any commercially available 5-Hyls containing four stereoisomers [i.e., (2S,5S), (2S,5R), (2R,5S), and (2R,5R)]. To determine absolute configuration of Hyls, we isolated the hydroxylated products on a preparative scale as described in Materials and Methods.

The nuclear Overhauser enhancement spectroscopy (NOESY) spectra of the (2S,3S)-3-Hyl derivative are shown in Fig. S3. Although a cross peak was found between H⁴ and H⁵, no cross peak was found between H⁵ and H^1′, showing that the relative stereochemistry between H⁴ and H⁵ was cis configuration. Since the absolute configuration of lysine used in the reaction was only 2S, we could infer that 5-(3-benzyloxycarbonylaminopropyl)-2-oxo-4-oxazolidinecarboxylic acid and 3-Hyl were the (4S,5S) and (2S,3S) isomers, respectively. In addition, the NOESY spectra of the (2S,4R)-4-Hyl derivative are shown in Fig. S4. Cross peaks were found among H³, H^4a, and H⁵, indicating that the relative stereochemistry between H³ and H⁵ was cis configuration; thus, 3-trifluoroacetylamino-5-(2-trifluoroacetylaminoethyl)-2(3H)-dihydrofuranone and 4-Hyl were the (3S,5R) and (2S,4R) isomers, respectively. The same absolute configurations were independently observed by Baud et al. (18).

Collectively, these findings led to the identification of two l-lysine 3S-hydroxylases and four 4R-hydroxylases (Fig. 1c). We previously reported on l-lysine hydroxylases (22, 23); however, Baud et al. also recently identified one l-lysine 3S-hydroxylase and two 4R-hydroxylases using a genome-mining approach (18). Unlike our focus on CAS-like proteins, they identified factors based on sequence homology with known 2-oxoglutarate-dependent dioxygenases in combination with five motifs defined on the InterPro database (24) and subsequently expressed 131 candidate genes in E. coli. In comparison, our strategy enabled us to identify six l-lysine hydroxylases from 36 candidate proteins, although three of the seven α-keto acid-dependent oxygenases (KDO2, KDO3, and KDO4) identified by Baud et al. (18) were identical to K4H-2, K4H-1, and K4H-3, respectively, in the present study.

Substrate specificity of l-lysine hydroxylases.

We next evaluated substrate specificity for the two l-lysine 3S-hydroxylases and four l-lysine 4R-hydroxylases toward various amino acids and their derivatives. Consequently, these hydroxylases acted only on l-lysine. Other nonproteinogenic amino acids, including l-ornithine, l-pipecolic acid, N^α-methyl-l-lysine, N^ε-methyl-l-lysine, l-lysine methylester, l-lysine ethylester, diaminopimeric acid, l-ethionine, and d-lysine, were also inert to all hydroxylases but acted on (2S,5R)-5-Hyl (Table 2). Taken together, these data indicate a relatively narrow substrate flexibility in which unprotected carboxy and amino groups on an α-carbon with a primary amine group were essential for activity, while the C-5 hydroxy group on l-lysine was acceptable.

TABLE 2.

HR-MS analysis of dihydroxylysines obtained from Hyls

Hydroxylase type	Substrate	Mass-to-charge ratio (m/z)
		[M+H]+	[M+Na]+
3S-Hydroxylase	(2S,5R)-5-Hyl	179.1032	201.0851
4R-Hydroxylase	(2S,5R)-5-Hyl	179.1033	201.0851
	(2S,3S)-3-Hyl	179.1032	201.0851

A comparative analysis of the l-lysine hydroxylase amino acid sequences also revealed a fully conserved 2-His-1-carboxylic acid motif (25) (Fig. S5). Moreover, comparison of enzyme-ligand interactions in VioC and l-lysine hydroxylases revealed conserved serine and arginine residues (S158 and R334, respectively, in VioC) that interact with the arginine α-carboxy group in l-lysine 3S-hydroxylases (Fig. 4). The corresponding residues were also found in l-lysine 4R-hydroxylases but were not adjacent to substrate sites according to homology models, suggesting that these were not likely to be involved in substrate recognition. Actually, the S144A mutation in K3H-1 decreased hydroxylation activity; however, K4H-4 with the S161A mutation retained 90% activity, supporting this prediction (Fig. 5). The histidine and glutamate residues (H168 and E170, respectively, in VioC), which donate the 2-His-1-carboxylic acid triad, were also found in all cases. Moreover, the corresponding residues (H154 and E156 in K3H-1 and H170 and E172 in K4H-4) seem critical because catalytic activities were significantly decreased by mutagenesis (Fig. 5). Interestingly, an arginine guanidino moiety corresponding to the lysine ε-amino group was surrounded by two acidic residues (D268 and D270 in VioC), suggesting that one or both acidic residues act as substrate acceptors. Subsequent mutagenesis experiments demonstrated that both residues were likely important for activity, but K3H-1 D241 and K4H-4 D253, rather than D243 and E256, respectively, appear to be more critical for the interaction with l-lysine (Fig. 5). Further structural analyses of substrate-bound l-lysine hydroxylases are necessary to understand these predictions and the functional significance of the different oxygen insertion positions.

FIG 4.

Two-dimensional structure diagram of hydroxylase-ligand complex generated by PoseView (40). Putative substrate binding residues are predicted by SWISS-MODEL and indicated by the following colors: dark gray, VioC; magenta, K3H-1; blue, K3H-2; green, K4H-1; orange, K4H-2; cyan, K4H-3; red, K4H-4. Dashed lines represent hydrogen bonds.

FIG 5.

Relative activities of mutated K3H-1 (a) and K4H-4 (b). Data are the means ± standard deviations (SD) of results of three independent experiments.

The physiological role of l-lysine hydroxylases remains unclear. Most other well-characterized amino acid hydroxylases are involved in the production of secondary metabolites, such as etamycin (l-proline hydroxylase) (26), viomycin (VioC) (27), daptomycin-type antibiotic (AsnO) (9), and streptolidine (OrfP) (28). On the other hand, l-isoleucine hydroxylase functions as a tricarboxylic acid cycle shunt and in 2-amino-3-methyl-4-ketopentanoic acid biosynthesis, which serves as a metabolic antagonist during vitamin B₁₂ production (29). Indeed, some protein genes relevant to secondary metabolite biosynthesis, including genes encoding nonribosomal peptide synthetases, were located adjacent or proximal to some l-lysine hydroxylase genes in original microbial genomes. As such, further investigation is needed to clearly understand their physiological functions. It should also be noted that microbial l-lysine hydroxylases with mammalian lysyl hydroxylases belong to the same protein family, despite their poor sequence similarity.

Recently, Baud et al. reported on l-lysine 3S-hydroxylase (18), which was not obtained in this work, despite being the same protein as that deposited under accession number ACU69184, which is located in a neighbor branch with K3H-1 and K3H-2 in Fig. 2. This could result from an insufficient expression of the corresponding gene or differences in the detection method, as the present study utilized HPLC in the initial screening whereas Baud et al. used an enzyme-coupled spectrophotometric assay following 2-oxoglutarate consumption.

Optimal reaction conditions.

Using the purified proteins, we optimized the pH and temperature conditions for each hydroxylase and summarized the results in Table 3. Except for psychrophilic K3H-2 (15°C), reactions were most efficient with neutral pHs (6.0 to 8.0) and mesophilic temperatures (25 to 40°C). The reaction conditions were also strictly dependent on 2-oxoglutarate as a cosubstrate and Fe²⁺ as a cofactor (Fig. 1c), consistent with 2-oxoglutarate-dependent dioxygenase family proteins.

TABLE 3.

Optimal conditions for hydroxylases

Hydroxylase	Optimal pH	Optimal temp (°C)
K3H-1	6.0	40
K3H-2	7.5	15
K4H-1	6.0	40
K4H-2	8.0	35
K4H-3	7.5	40
K4H-4	7.5	25

Evaluation of biocatalysts suitable for Hyl production.

To investigate the feasibility of bioconversion, we assessed favorable biocatalysts for (2S,3S)-3-Hyl and (2S,4R)-4-Hyl production with whole-cell reaction mixtures containing 100 mM l-lysine, 120 mM 2-oxoglutarate, 5 mM FeSO₄, and E. coli cells (optical density at 600 nm [OD₆₀₀], 30). Notably, the productivity of K3H-1 was much higher than that of K3H-2 for (2S,3S)-3-Hyl production, while K4H-4 showed relatively higher productivity than the other 4R-hydroxylases in (2S,4R)-4-Hyl production (Fig. 6). Thus, we concluded that K3H-1 and K4H-4 were suitable biocatalysts for the production of (2S,3S)-3-Hyl and (2S,4R)-4-Hyl, respectively, and used in the following preparative reactions.

FIG 6.

Specific productivity of (2S,3S)-3-Hyl and (2S,4R)-4-Hyl in large-scale preparative reactions. Data are the means ± SD of results of three independent experiments. CDW, cell dry weight.

Preparative-scale Hyl production.

We then sought to utilize the characterized l-lysine 3S- and 4R-hydroxylases to establish a preparative method for large-scale (2S,3S)-3-Hyl and (2S,4R)-4-Hyl production. For (2S,3S)-3-Hyl production, K3H-1-expressing E. coli cells were first used in whole-cell reactions to convert 50 mM l-lysine to (2S,3S)-3-Hyl within 1.5 h with no further degradation. Subsequent reactions were performed with higher concentrations of l-lysine in a stepwise manner. Notably, a complete conversion was achieved with up to 500 mM l-lysine (Fig. 7a); however, productivity continued even with an initial concentration of 600 mM l-lysine, which yielded 531 mM (86.1 g/liter) (2S,3S)-3-Hyl after a 52-h reaction with an 89% molar conversion (Fig. 7b). Similar results were found with (2S,4R)-4-Hyl production, where 50 mM l-lysine was entirely converted to (2S,4R)-4-Hyl within 1 h, and higher (2S,4R)-4-Hyl productivity was accomplished with up to 300 mM l-lysine, albeit with a slightly decreased conversion. Interestingly, an initial l-lysine concentration of 400 mM resulted in a serious reduction in both product concentration and conversion efficiency, probably due to substrate inhibition (Fig. 8a). The largest amount of (2S,4R)-4-Hyl reached 265 mM (43.0 g/liter) after a 24-h reaction with an 88% molar conversion (Fig. 8b). Organic acids were also determined in both bioprocesses. Then, time-dependent 2-oxoglutarate consumption and succinate generation were confirmed (Fig. S6), which were accompanied by lysine hydroxylation, but the succinate level was gradually decreased, likely because of E. coli cellular metabolism.

FIG 7.

Production of (2S,3S)-3-Hyl using K3H-1-expressing E. coli as a whole-cell biocatalyst. (a) Impact of initial l-lysine concentration on the (2S,3S)-3-Hyl productivity (bars, left axis) and conversion efficiency (open circles, right axis). (b) Time course of the bioconversion of l-lysine (closed squares) into (2S,3S)-3-Hyl (closed circles). Data are the means ± SD of results of three independent experiments.

FIG 8.

Production of (2S,4R)-4-Hyl using K4H-4-expressing E. coli as a whole-cell biocatalyst. (a) Impact of initial l-lysine concentration on the (2S,4R)-4-Hyl productivity (bars, left axis) and conversion efficiency (open diamonds, right axis). (b) Time course of the bioconversion of l-lysine (closed squares) into (2S,4R)-4-Hyl (closed diamonds). Data represent the means ± SD of results of three independent experiments.

In conclusion, we have developed a practical production process for (2S,3S)-3-Hyl and (2S,4R)-4-Hyl. While two hydroxylases, K4H-1 and K4H-2, were identical to KDO3 and KDO2, which were found independently from this work (16), we successfully identified additional new hydroxylases. Especially, K3H-1 and K4H-4, which were discovered in this study, exhibited remarkable catalytic activity, allowing the highly efficient production of Hyls. It is noteworthy that the productivity and conversion efficiency of (2S,3S)-3-Hyl and (2S,4R)-4-Hyl were extremely high compared to those of previous production methods with 2-oxoglutarate-dependent dioxygenase (11, 30–33). While the basis for this increased efficiency should be elucidated experimentally, we suspect that it is attributable to the high water solubility of both the substrate and the product, as well as a high catalytic efficiency. Nevertheless, the (2S,3S)-3-Hyl and (2S,4R)-4-Hyl production demonstrated here serves as an excellent model for practical production and may be valuable for future industrial bioproduction.

MATERIALS AND METHODS

Reagents.

(2S,5R)-5-Hyl was obtained from Sigma-Aldrich (St. Louis, MO, USA). Unless otherwise stated, all other reagents were of analytical grade and purchased from Wako Pure Chemical (Osaka, Japan) or Kanto Chemical (Tokyo, Japan).

Phylogenetic analysis.

To classify CAS-like superfamily proteins, we queried the VioC amino acid sequence in the National Center for Biotechnology Information database using the protein BLAST algorithm (34). We selected 93 proteins with >25% amino acid sequence identity. Subsequent phylogenetic analyses were performed with ClustalW (35), and a tree was constructed in GENETYX-MAC software v.19 (Genetyx Corporation, Tokyo, Japan).

Plasmid construction.

Twenty-eight CAS-like superfamily genes were PCR amplified from microbial genomic DNA templates with KOD FX Neo (Toyobo, Osaka, Japan) or the GC-RICH PCR system (Roche, Basel, Switzerland) using the appropriate primer pairs listed in Table 4. The fragments were then ligated into the pET-21a(+) (NdeI with appropriate restriction sites) or pET-21d(+) (NcoI with appropriate restriction sites) expression plasmids. Eight artificial genes (accession numbers EHR60654, EFK99552, AEP14369, CAO97865, EHY77614, EAR24255, ABZ01301, and EGX55940) were synthesized by DNA2.0, Inc. (Menlo Park, CA, USA), inserted into the pJexpress 401 vector, and then introduced individually into E. coli Rosetta 2(DE3) for overexpression.

TABLE 4.

Primers used for gene cloning

Proteina	Template	Oligonucleotide (5′ to 3′)	Restriction site
CAM03547	Saccharopolyspora erythraea NRRL 2338	TTATCATATGTCGGTGGCAGTCCGCACCG	NdeI
		ATAATCATATGACGGCCGTACTCGACACCG	XhoI
CCB75394	Streptomyces cattleya NRRL 8057	ATAATCATATGACGGCCGTACTCGACACCG	NdeI
		AATAGCTCGAGGTCGAGGACGTAGCCGTCGTC	XhoI
CAH18567	Streptoalloteichus tenebrarius	ATAATCATATGTCCTCGGCGCTCACGATTCC	NdeI
		AATAGCTCGAGCCACAGGCCTCGGATGGTCG	XhoI
ACU75182	Catenulispora acidiphila DSM 44928	ATAATCATATGACCGTTCTGACCGCCTCC	NdeI
		AATAGCTCGAGGTGATGCACCCGGCGGTTC	XhoI
CCB72401	Streptomyces cattleya NRRL 8057	ATAATCATATGACCGTCATCGACCACACCAC	NdeI
		AATAGCTCGAGGAAGTGGACCCGGCGGTTGTC	XhoI
CBJ92070	Xenorhabdus nematophila ATCC 19061	ATAATCCATGGAAAGTAGAAATTTACTTG	NcoI
		AATAGCTCGAGAATAATAAAGCGTGTATTAATAAC	XhoI
EDY47125	Streptomyces clavuligerus ATCC 27064	ATAATCATATGGCCTCTCCGATAGTTGACTGC	NdeI
		AATAGCTCGAGGCGGCGCGGCGAGAACGAG	XhoI
ACU98305	Saccharomonospora viridis DSM 43017	ATAATCATATGACCACCACCGCCGAATCACC	NdeI
		AATAGAAGCTTCCGGGGCACGAACTTCACGAC	HindIII
ABS05421 (K3H-1)	Kineococcus radiotolerans SRS30216	ATTCACATATGTCCTCGCTGTTCCTCGACTC	NdeI
		AGCTTCTCGAGGCTGAAGCTGGCCTGCACG	XhoI
EAR24255b (K3H-2)	pJexpress 401 carrying CAS-like protein of marine actinobacterium PHSC20C1	TAATCATATGGAAACAATGTCAGCAATCGCC	NdeI
		AATAGCTCGAGGGAGTGGACTGCACCCAGGG	XhoI
ACU69184	Catenulispora acidiphila DSM 44928	ATAATCATATGAAGAACCTGTCTGCGTATGAAG	NdeI
		AATAGCTCGAGGCTGAACCTCGCAGAGACGAC	XhoI
CBJ90519	Xenorhabdus nematophila ATCC 19061	ATAATCATATGATGCCTGATACTCAGGAAG	NdeI
		AATAGCTCGAGTTTCGCCCTTAAAGTACCTGC	XhoI
BAC73330	Streptomyces avermitilis MA-4680	ATTCACATATGAGCACGGCAGCCGCACCTG	NdeI
		AGCTTCTCGAGGCGCGCGTGGGCTTCGATCG	XhoI
EDY49560	Streptomyces clavuligerus	ATAATCATATGAGCACGGCAGCCGCACC	NdeI
		AATAGCTCGAGGCGGACGTGGGCCTCCAGG	XhoI
EGG49088	Streptomyces griseoaurantiacus M045	ATAATCATATGGACACCGACGACGGCCTG	NdeI
		AATAGCTCGAGGGGATGTGCCACCAAGGCGGC	XhoI
ACZ86677	Streptosporangium roseum DSM 43021	ATAATCATATGGGCCTCAACGTGACCCCTG	NdeI
		AATAGCTCGAGCCGCTCCTCGTAGGGGTCGATC	XhoI
GAB25096	Gordonia polyisoprenivorans NBRC 16320	ATAATCATATGGCGATGATCGGCGCGGC	NdeI
		AATAGAAGCTTTACCAGCGCCCCGGCGTAC	HindIII
ZP_06561781	Saccharopolyspora erythraea NRRL 2338	ATAATCATATGCTTCTCGAAACGGCTTCCGC	NdeI
		AATAGAAGCTTTCCCGCGGACCGCAGTGAC	HindIII
EGG48212	Streptomyces griseoaurantiacus M045	ATAATCATATGCTCGGCCAGACCCCCACC	NdeI
		AATAGAAGCTTCCAGTGCGAACCGCCCGCC	HindIII
ABQ06186 (K4H-1)	Flavobacterium johnsoniae UW101	TTATCATATGAAATCACAATCATTAATTGAAGATGAG	NdeI
		TGTAATAGCTCGAGAGCCTGATCAAAAACTTTTCCTAAATG	XhoI
ACU60313 (K4H-2)	Chitinophaga pinensis DSM 2588	ATAATCATATGAGACCCTTAGACGTGACACCC	NdeI
		AATAGCTCGAGAAGGTTTGCCAGGTGAGCGCTATATAC	XhoI
AEV99100 (K4H-3)	Niastella koreensis GR20-10	ATAATCATATGGAAACTATCATTGAATCC	NdeI
		AATAGCTCGAGTTGTTGTGAATGAAACAATTTG	XhoI
EFK34737 (K4H-4)	Chryseobacterium gleum ATCC 35910	ATAATCATATGAATTCTACACAAATTTTAG	NdeI
		AATAGCTCGAGAAAATGTTGAAAGTTTTTACC	XhoI
CBJ90214	Xenorhabdus nematophila ATCC 19061	ATAATCCATGGACCCATCTATATATTCAATTG	NcoI
		AATAGCTCGAGTGGCAGTACATTAATGCGATC	XhoI
BAL15753	Streptomyces lavendulae subsp. lavendulae	ATAATCATATGTCGAACCTCACCGACCAGTCCAC	NdeI
		AATAGCTCGAGGCCCAGCAGCCGGGTGGCGG	XhoI
CCB71649	Streptomyces cattleya NRRL 8057	ATAATCATATGTCGCACAGCGCTGTCAGCGAC	NdeI
		AATAGCTCGAGCACCACCGCCCGGCCCGCGTC	XhoI
ACU72362	Catenulispora acidiphila DSM 44928	ATAATCATATGCACCGCTTGGCCCTGAC	NdeI
		AATAGCTCGAGGTAAATGACCCGGTCGTCCGG	XhoI
EDY47332	Streptomyces clavuligerus ATCC 27064	ATAATCATATGATCAAGGTTGAACACCGGCCC	NdeI
		AATAGCTCGAGCCAGGACAGTCCGGTGCTGAC	XhoI
ADL45379	Micromonospora aurantiaca ATCC 27029	ATAATCATATGAAGACCCTGGACAGGATCG	NdeI
		AATAGCTCGAGGAACAGCACCCGGTAGCTCG	XhoI

^aGenBank accession number.

^bThe gene encoding this protein was reintroduced into pET-21a(+) from pJexpress401.

Mutagenesis.

Mutants were generated with a QuikChange site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA, USA) using the mutagenic primer pairs (Table 5). PCR was performed as follows: 94°C for 2 min and 10 cycles each of 94°C for 20 s, 50°C for 30 s, and 68°C for 7 min. The remaining template was digested with DpnI, and the resulting PCR product was transformed into E. coli JM109. The mutations were subsequently confirmed by DNA sequencing.

TABLE 5.

Primers used for site-directed mutagenesis

Mutant	Oligonucleotide (5′ to 3′)
K3H-1
Q142A	GCGACTGCGCGGTCGCGGCGAGCCTGGCG
	CGCGACCGCGCAGTCGCAGGGCTCCCGGG
S144A	CTGCGCCTGCTGGGTCGCGGC
	CAGGCGCAGGGCTCCCGGGTG
H154A	GGTCGCCGCCTCCAGCTCCAC
	GCGGCGACCGAGCAGTGCTTC
E156A	CTGCGCGGTGTGCGCCTCCAG
	ACCGCGCAGTGCTTCTCCGAC
D241A	CTGCGCGACGAGCATCGTCGG
	GTCGCGCAGGACCTCATGCAC
D243A	GAGCGCCTGGTCGACGAGCAT
	CAGGCGCTCATGCACGGCATC
R305A	ACCCGCGGAGATGAACCGGTC
	TCCGCGGGTTTCGTCGTCCGC
K4H-4
Q158A	ATCCTGTCGCTGTTTCTTTCATTTTTTCATC
	AGAAACAGCGACAGGATCAGGATCCTCAACG
S161A	TCCCGCTCCTGTCTGTGTTTC
	GGAGCGGGATCCTCAACGGAT
H170A	TGTCGCTACATACAGATCCGT
	GTAGCGACAGAAGATGCTTTT
E172A	AAGCATCCGCTGTATGTACATACAGATCCG
	TACATACAGCGGATGCTTTTCTGAAACATCAGGC
D253A	CGCCGCAAATCGCATAAAAGG
	TTTGCGGCGGCGGAACAGCTT
E256A	CTGCGCCGCCGCATCAAATCG
	GCGGCGCAGCTTTTCAATTCC
R326A	TCTCGCTTCGCACGGCTGTTC
	GAAGCGAGAATTATGCTTCGG

Gene expression and protein purification.

Protein purification procedures have been described previously (36). Briefly, gene expression was induced with 0.1 mM isopropyl-β-d-thiogalactopyranoside in E. coli Rosetta 2(DE3) cells harboring the pET-21a(+), pET-21d(+), or pJexpress 401-derived vectors. Harvested cells were sonicated, and then His₆-tagged proteins were purified using a HisTrap HP column (GE Healthcare, Little Chalfont, United Kingdom). The bound proteins were eluted with phosphate buffer containing excess imidazole, which was subsequently exchanged by gel filtration chromatography using a PD-10 column (GE Healthcare) to remove imidazole and saline. Protein concentrations were determined by Bradford assays (37) with a bovine serum albumin standard, and then the purity was confirmed by 12.5% SDS-PAGE and Coomassie brilliant blue staining.

Whole-cell l-lysine hydroxylation activity assays.

The recombinant E. coli strains were harvested by centrifugation (4°C, 5,000 × g, 10 min) and washed twice with chilled 0.85% NaCl solution. The test tube-scale reaction solution contained 100 mM phosphate buffer (pH 7.0), 5 mM l-lysine, 10 mM 2-oxoglutarate, 0.1 mM FeSO₄, and E. coli whole cells (OD₆₀₀, 10) in a total volume of 5 ml. After incubation with reciprocal shaking at 150 rpm at 30°C for 3 h, the cells were immediately removed by centrifugation (4°C, 20,000 × g, 10 min) and the supernatants were harvested for HPLC analysis.

In vitro amino acid hydroxylation assays.

Amino acid hydroxylation activity was assessed in analytical scale using 5 mM amino acid as a substrate. Reactions were performed as described above but with 0.3 mg/ml purified protein (except for 1.0 mg/ml K3H-2) instead of E. coli whole cells. The effects of pH and temperature on hydroxylation activity were measured from pH 5.0 to pH 10.0 with a step size of pH 0.5 and from 10°C to 50°C with a step size of 5°C. After a 5-min incubation, the reaction was immediately terminated by heat denaturation at 80°C, and the supernatants were analyzed by HPLC.

Mutants were evaluated with a reaction solution containing 100 mM HEPES-NaOH buffer (pH 7.0), 5 mM l-lysine, 10 mM 2-oxoglutarate, 1 mM l-ascorbate, 0.5 mM FeSO₄, and the appropriate amount of cell extracts, which were prepared by ultrasonic disruption, followed by centrifugation in a total volume of 100 μl. After incubation at 30°C for 5 min, the reaction was immediately terminated by heat denaturation at 80°C, and the supernatants were analyzed by HPLC.

Large-scale preparative whole-cell reactions.

In large-scale bioconversions, 40-ml reaction mixtures containing 50 to 600 mM l-lysine, 60 to 720 mM 2-oxoglutarate (1.2-fold higher concentration than l-lysine), 5 mM FeSO₄, 1% (vol/vol) Triton X-100, and various concentrations of E. coli whole cells were incubated in a 500-ml baffled flask at 30°C with agitation at 200 rpm. A small aliquot was withdrawn from the flask at each sampling period and centrifuged, and the supernatant was analyzed by HPLC.

Analytical methods.

All amino acids were determined with an L-7000 series HPLC system (Hitachi, Tokyo, Japan) equipped with a Cosmosil AR-II column (4.6 by 250 mm; Nacalai Tesque, Kyoto, Japan). Column oven temperature and flow rate were kept at 40°C and 1 ml/min, respectively. Eluent A was an aqueous solution containing 0.2% (vol/vol) HCOOH. Eluent B was acetonitrile containing 0.2% (vol/vol) HCOOH. Elutions were performed with a linear gradient from 0 min (A:B = 95:5) to 12 min (A:B = 50:50). AccQ-derivatized amino acid was detected at A₂₆₀ after separation per the manufacturer's instructions (Waters, Milford, MA, USA).

For routine Hyl determination, a postcolumn method with o-phthalaldehyde (OPA) was performed with eluent C (20 mM KH₂PO₄, 1 g/liter sodium 1-heptanesulfonate, 15% [vol/vol] methanol, adjusted to pH 2.5 with phosphoric acid). The derivatization reagent contained 21.64 g/liter H₃BO₃, 12 g/liter NaOH, 2 g/liter N-acetyl-l-cysteine, and 0.8 g/liter OPA. Isocratic elutions were performed with eluent C at a flow rate of 0.5 ml/min, and online derivatization was done prior to detection with derivatization solution at a flow rate of 0.3 ml/min. The OPA-derivatized Hyl was detected with a fluorescence detector at 340-nm excitation and 450-nm emission.

The presence of organic acids was determined with an L-7000 series HPLC system equipped with an Ultron PS-80H column (8 by 300 mm; Shinwa Chemical, Kyoto, Japan) and aqueous perchloric acid solution (pH 2.1) as the eluent. Column oven temperature and flow rate were kept at 60°C and 1 ml/min, respectively. Organic acids were detected at A₂₁₀ after separation.

HR-MS analysis was carried out using a Xevo G2-XS quadrupole time-of-flight (QToF) mass spectrometer combined with an Acquity ultraperformance liquid chromatography (UPLC) H-class system (Waters).

NMR analysis.

A reaction mixture containing 7 mmol l-lysine, 8 g of E. coli cells expressing K3H-1 or K4H-1, 35 ml of 1 M potassium phosphate buffer (pH 7.0), 304 ml of distilled water, 14 mmol 2-oxoglutarate, 0.7 mmol sodium l-ascorbate, 0.07 mmol FeSO₄, and 0.35 g Adekanol LG-109 (Adeka, Tokyo, Japan) was incubated in a 1-liter jar fermentor for 17 to 19 h at 30°C with 500-rpm agitation and 2-vvm airflow. After cation exchange chromatography using a strong cation resin (Diaion SK-1B [H form]; Mitsubishi Chemical, Tokyo, Japan), the eluent with NH₄OH was concentrated under reduced pressure. The reactions generated 6.17 mmol (2S,3S)-3-Hyl and 6.79 mmol (2S,4R)-4-Hyl with 88% and 97% yields, respectively. Isolated Hyls were identified by ¹H nuclear magnetic resonance (NMR) spectroscopy recorded on an Avance spectrometer (Bruker, Billerica, MA, USA). Chemical shift values (δ) were set in relation to the solvent peak and given in parts per million. The coupling constants (J) were given in Hz. ¹H NMR of (2S,3S)-3-Hyl (400 MHz, D₂O): δ = 1.45–1.58 (2H, m), 1.63–1.73 (1H, m), 1.74–1.88 (1H, m), 2.93–3.04 (2H, m), 3.47 (1H, d, J = 4.3 Hz), 3.89 (1H, dt, J = 8.4, 4.5 Hz). ¹H NMR of (2S,4R)-4-Hyl (400 MHz, D₂O): δ = 1.50–1.65 (3H, m), 1.71 (1H, ddd, J = 14.1, 9.1, 4.3 Hz), 2.62–2.75 (2H, m), 3.32 (1H, dd, J = 8.6, 4.8 Hz), 3.72–3.80 (1H, m).

To determine the absolute configuration of (2S,3S)-3-Hyl, carbamoyl derivatization was done according the literature (38). First, 0.051 mmol (2S,3S)-3-Hyl was stirred overnight at room temperature in the presence of 0.54 mmol NaOH and 0.26 mmol benzyloxycarbonyl chloride. Then, 0.5 ml of tetrahydrofuran was added to the solution and kept at 60°C for an additional 2 h. After cooling, 2.4 mmol NaOH was added and the mixture was incubated overnight at room temperature. The resulting solution was washed twice with toluene-tetrahydrofuran (1:1) and then acidified with 250 μl of concentrated HCl, followed by washing with ethyl acetate for three times. The compound in aqueous layer was extracted four times with 1-butanol. After drying over MgSO₄, 0.045 mmol (4S,5S)-5-(3-benzyloxycarbonylaminopropyl)-2-oxo-4-oxazolidinecarboxylic acid was obtained with a yield of 89%. ¹H-NMR of (4S,5S)-5-(3-benzyloxycarbonylaminopropyl)-2-oxo-4-oxazolidinecarboxylic acid (400 MHz, CD₃OD): δ = 1.39–1.53 (3H, m, H^1′, 2H^2′), 1.59–1.68 (1H, m, H^1′), 3.02–3.08 (2H, m, 2H^3′), 3.60–3.64 (1H, m, H⁴), 3.93–4.00 (1H, m, H⁵), 4.90–5.02 (2H, m, Bn), 7.18–7.28 (5H, m, Bn).

To determine the absolute configuration of (2S,4R)-4-Hyl, lactone derivatization was done according the literature (39). First, 0.3 mmol (2S,4R)-4-Hyl was dissolved in 0.2 ml of H₂O containing 6 M HCl and incubated for 1 h at room temperature. After concentration, 67 mg of (3S,5R)-3-amino-5-(2-aminoethyl)-2(3H)-dihydrofuranone dihydrochloride was obtained as a crude white crystal. Then, 42 mg of the resulting compound was placed into a flask containing 4 mmol triethylamine and 1 ml of dichloromethane, followed by the addition of 1.0 mmol trifluoroacetic anhydride under cooling with ice. The mixture was concentrated following 2 h of stirring and purified by silica gel column chromatography. The resulting oily solution was dissolved in ethyl acetate and washed with a potassium carbonate solution and saturated saline. The aqueous layer was extracted by ethyl acetate, and the collected organic layer was dried over potassium sulfate. Consequently, 43 mg of oily brown compound was obtained after drying. NMR analysis revealed that this compound included 0.044 mmol (3S,5R)-3-trifluoroacetylamino-5-(2-trifluoroacetylaminoethyl)-2(3H)-dihydrofuranone with a yield of 23%. ¹H-NMR of (3S,5R)-3-trifluoroacetylamino-5-(2-trifluoroacetylaminoethyl)-2(3H)-dihydrofuranone (400 MHz, CDCl₃): δ = 1.93–2.22 (3H, m, H^4b, 2H^1′), 2.76 (1H, ddd, J = 12.6, 8.8, 5.6 Hz, H^4a), 3.51–3.56 (2H, m, 2H^2′), 4.52–4.60 (1H, m, H⁵), 4.75 (1H, dd, J = 11.9, 9.1 Hz, H³), 7.86 (1H, brs, NH), 8.87 (1H, brs, NH).