Cryptic Production of trans-3-Hydroxyproline in Echinocandin B Biosynthesis

ABSTRACT

Echinocandins are antifungal nonribosomal hexapeptides produced by fungi. Two of the amino acids are hydroxy-l-prolines: trans-4-hydroxy-l-proline and, in most echinocandin structures, (trans-2,3)-3-hydroxy-(trans-2,4)-4-methyl-l-proline. In the case of echinocandin biosynthesis by Glarea lozoyensis, both amino acids are found in pneumocandin A₀, while in pneumocandin B₀ the latter residue is replaced by trans-3-hydroxy-l-proline (3-Hyp). We have recently reported that all three amino acids are generated by the 2-oxoglutarate-dependent proline hydroxylase GloF. In echinocandin B biosynthesis by Aspergillus species, 3-Hyp derivatives have not been reported. Here we describe the heterologous production and kinetic characterization of HtyE, the 2-oxoglutarate-dependent proline hydroxylase from the echinocandin B biosynthetic cluster in Aspergillus pachycristatus. Surprisingly, l-proline hydroxylation with HtyE resulted in an even higher proportion (∼30%) of 3-Hyp than that with GloF. This suggests that the selectivity for methylated 3-Hyp in echinocandin B biosynthesis is due solely to a substrate-specific adenylation domain of the nonribosomal peptide synthetase. Moreover, we observed that one product of HtyE catalysis, 3-hydroxy-4-methyl-l-proline, is slowly further oxidized at the methyl group, giving 3-hydroxy-4-hydroxymethyl-l-proline, upon prolonged incubation with HtyE. This dihydroxylated amino acid has been reported as a building block of cryptocandin, an echinocandin produced by Cryptosporiopsis.

IMPORTANCE Secondary metabolites from bacteria and fungi are often produced by sets of biosynthetic enzymes encoded in distinct gene clusters. Usually, each enzyme catalyzes one biosynthetic step, but multiple reactions are also possible. Pneumocandins A₀ and B₀ are produced by the fungus Glarea lozoyensis. They belong to the echinocandin family, a group of nonribosomal cyclic lipopeptides that exhibit a strong antifungal activity. Chemical derivatives are important drugs for the treatment of systemic fungal infections. We have recently shown that in the biosynthesis of pneumocandins A₀ and B₀, three hydroxyproline building blocks are provided by one proline hydroxylase. Here we demonstrate that the proline hydroxylase from echinocandin B biosynthesis in Aspergillus pachycristatus produces the same hydroxyprolines, with an increased proportion of trans-3-hydroxyproline. However, echinocandin B biosynthesis does not require trans-3-hydroxyproline; its formation remains cryptic. While one can only speculate on the evolutionary background of this unexpected finding, proline hydroxylation in G. lozoyensis and A. pachycristatus provides an unusual insight into peptide antibiotic biosynthesis—namely, the complex interplay between the selectivity of a hydroxylase and the substrate specificity of a nonribosomal peptide synthetase.

INTRODUCTION

Proline hydroxylases (PHs) belong to the large family of non-heme Fe(II)- and 2-oxoglutarate (2OG)-dependent dioxygenases that typically hydroxylate C—H bonds through activation of molecular oxygen and oxidative decarboxylation of 2OG to succinate (1–3). In contrast to the peptide-modifying prolyl hydroxylases known, e.g., from mammalian collagen biosynthesis or the hypoxia-inducible factor (HIF) regulatory pathway, PHs exclusively convert the free amino acid l-proline (Pro) in a highly regio- and stereoselective manner. Bacterial PHs provide three of the four stable hydroxy-l-proline (Hyp) isomers (4–8) which are typically used for the biosynthesis of peptide antibiotics, such as etamycin (7) and telomycin (8). In biotechnology, bacterial PHs have been applied for the production of Hyp isomers as valuable chiral building blocks for the synthesis of chemicals and pharmaceuticals (9).

In fungi, trans-Hyps are structural elements of potent nonribosomal peptide antibiotics, such as echinocandins, or the bicyclic amatoxins. The latter are ribosomally produced by Amanita mushrooms and are highly toxic for mammals by inhibiting RNA polymerase (10). Echinocandins consist of a ring of six mostly nonproteinogenic amino acids, including two Hyps, and are equipped with a fatty acid side chain (Fig. 1). The lipopeptides strongly inhibit glucan biosynthesis, required for fungal cell wall formation, and thus are potent antifungal agents. Semisynthetic derivatives such as caspofungin, anidulafungin, and micafungin are important drugs for the clinical treatment of invasive mycoses (11). Echinocandin biosynthesis is found rather infrequently in Aspergillacea (Eurotiomycetes) and Helotiales (Leotiomycetes) (12, 13). The biosynthesis has been elucidated for pneumocandins produced by Glarea lozoyensis (Helotiales) through ¹³C labeling experiments (14, 15) and, more recently, on the genomic level for echinocandin B produced by Aspergillus pachychristatus (Aspergillaceae) (16, 17). However, a PH has not been described in the latter work.

FIG 1.

Structures of pneumocandins A₀ and B₀ and echinocandin B. Hydroxyprolines are shown in blue.

In 2014, we reported that the fungus Glarea lozoyensis has a PH (GloF) and that the encoding gene occurs in the pneumocandin biosynthetic gene cluster (18). Through in vitro conversions with the heterologously produced enzyme, it was found that GloF catalyzes the previously unknown trans-3-hydroxylation of Pro, albeit with the trans-4-hydroxylated product being predominant. Furthermore, we established that all three Hyp building blocks required for the biosynthesis of pneumocandins A₀ and B₀ are produced by this enzyme, namely, trans-4-hydroxy-l-proline (4-Hyp), trans-3-hydroxy-l-proline (3-Hyp), and (3S,4S)-3-hydroxy-4-methyl-l-proline (4-Me-3-Hyp) (Fig. 1) (18). Apart from GloF from pneumocandin biosynthesis, eight (putative) fungal PHs are known according to the eight additional echinocandin biosynthetic clusters reported so far (13). They form a monophyletic clade which is clearly distinct from other groups of fungal Fe(II)/2OG-dependent dioxygenases, including the recently discovered and related fungal pipecolic acid hydroxylases (19).

In contrast to the case with pneumocandin biosynthesis, congeners with 3-Hyp at position 6 are unknown from echinocandin B biosynthesis by Aspergillaceae (13). One hypothesis to explain this finding is that the corresponding adenylation (A) domain of the nonribosomal peptide synthetase (NRPS) is strictly substrate specific in aspergilli, while that of G. lozoyensis is more promiscuous. Indeed, a comparison of homology models of the A domains from G. lozoyensis and A. pachycristatus (previously classified as Emericella rugulosa and Aspergillus nidulans var. roseus [20]) revealed that the A domain from A. pachycristatus has a more “compact” binding pocket, suggesting a stricter substrate specificity (12). An alternative explanation could be that the strict product selectivity is due to an entirely trans-4-selective PH in the host fungus. In this case, 3-Hyp would not be formed at all and the specificity of the A domain would be irrelevant. To establish which explanation is correct, we decided to investigate the PH activity in A. pachycristatus NRRL 11440. The biosynthetic gene cluster of this strain has been identified and comprehensively characterized by the groups of Tang and Walsh (16, 17). More recently, the sequence has been revised (21).

In this work, we describe the heterologous overproduction and characterization of the PH HtyE from the echinocandin biosynthetic gene cluster in A. pachycristatus. The catalytic properties are compared to the previously characterized orthologous enzyme GloF from G. lozoyensis (18), and the relevance for the biosynthesis of echinocandin B in A. pachycristatus (Aspergillaceae) and the pneumocandins in G. lozoyensis (Helotiales) is discussed.

RESULTS

Enzymatic characterization and substrate screening of HtyE.

The gene htyE (NCBI accession no. AFT91391) encoding the PH from A. pachycristatus was synthesized and heterologously expressed in Escherichia coli BL21 Gold as the N- and C-terminally His-tagged protein in yields up to 75 mg liter⁻¹. The purified HtyE enzyme was incubated with Pro (4 mM) under standard conditions (see Materials and Methods). Analysis by fluorescence high-performance liquid chromatography (HPLC) (22) revealed two product peaks, which could be assigned to 4-Hyp and 3-Hyp by means of reference compounds (Fig. 2A). The product ratio of 4-Hyp to 3-Hyp was about 2:1. The masses of the products were confirmed by liquid chromatography-mass spectrometry (LC-MS) (see Table S2 in the supplemental material). Clearly 4-Hyp was the main product of HtyE catalysis; however, the proportion of 3-Hyp in the product mixture was significantly higher than that observed previously for the ortholog GloF from G. lozoyensis (4-Hyp/3-Hyp ratio ≈ 8:1) (Fig. 2 and 3A). To determine whether the product ratio is influenced by the assay conditions, Pro hydroxylation was investigated in different buffers (morpholineethanesulfonic acid [MES], HEPES, and Tris) with various pH values (6.0, 6.8, 7.0, and 8.0) (see Fig. S1); in particular at pH 6 and 8, the catalytic activity dropped significantly. Nevertheless, the trans-4/3-selectivity was only slightly affected (2.2 to 2.0:1), which is consistent with previous findings for GloF (18, 23). The best activity was observed in MES buffer at pH 6.8 and 20°C, in which 4-Hyp and 3-Hyp were produced in a ratio of 2.1:1.

FIG 2.

Conversions of Pro (1), 4-Me-Pro (2), and Pip (3) with HtyE (A) and GloF (B), determined by fluorescence HPLC. The substrates 4-Me-Pro (9.8 min) and Pip (9.3 min) were not detected. The peaks at 8.3 and 8.8 min originate from the staining reagent FMOC-Cl.

FIG 3.

(A) 4-Hyp/3-Hyp ratios from conversions of Pro with HtyE and GloF. (B) Conversion of different substrates with HtyE and GloF under standard conditions. No conversion was observed for all four Hyp isomers: 3,4-dehydro-l-proline, 4-thio-l-proline, and cis- and trans-4-fluoro-l-proline. For statistical data, see Table S1.

In addition to Pro, the second native substrate, trans-4-methyl-l-proline (4-Me-Pro), was tested. The predominant product peak in the HPLC chromatogram had the same retention time as the product of GloF conversions, which has been characterized as 4-Me-3-Hyp (Fig. 2) (18).

HtyE and GloF are part of nonribosomal peptide biosynthetic pathways, and Pro hydroxylation can principally occur at three stages of biosynthesis: at the amino acid as a free molecule, tethered to an acyl carrier protein, or as a residue in a pre-echinocandin peptide. Our experiments so far have clearly shown that free Pro and 4-Me-Pro are good substrates of both enzymes. To determine whether bound Pro is also accepted as a substrate, we tested the N-acetylcysteamine thioester of Pro, mimicking the phosphopantetheine anchor of a thiolation (T) domain, and also diverse Pro tripeptides (Gln-Pro-Ser, Thr-Pro-Tyr, and Tyr-Gln-Pro) with both enzymes. No conversion of these model substrates was detected by means of liquid chromatography-tandem mass spectrometry (data not shown) (23). These findings suggest that GloF and HtyE are true PHs that exclusively convert the free amino acids.

Bacterial PHs are known for a moderate substrate promiscuity concerning the hydrocarbon backbone (carbon atoms 2–4) (9, 24). Thus, we tested the fungal counterparts HtyE and GloF with various Pro congeners, most of which are known as substrates of at least one bacterial PH. However, both enzymes only accepted l-pipecolic acid (Pip) in addition to the native substrates. Notably, Pip was converted quantitatively in the assays, which indicates an activity even greater than that with Pro (80 to 90%) (Fig. 2 and 3B; see also Table S1). The Pip conversions yielded a single product, which had the same retention time as trans-4-hydroxy-l-pipecolic acid (4-Hypip), a known product of the fungal l-Pip trans-4-hydroxylase from A. nidulans (AnPip4H) (19).

Methyl group hydroxylation as a secondary reaction.

When HtyE was incubated with 4-Me-Pro under standard conditions (see Materials and Methods), the substrate was quantitatively converted into 4-Me-3-Hyp within 40 min. Notably, upon longer incubation, a second product slowly accumulated, to about 30% within 3 h, while the concentration of 4-Me-3-Hyp decreased significantly (Fig. 4A). LC-MS analysis showed that the mass of the secondary product corresponded to that of dihydroxylated 4-Me-Pro (m/z 162, M+H⁺) (Fig. 4B and C). To enable determination of the position of the second hydroxyl group by nuclear magnetic resonance (NMR) spectroscopy, 4-Me-Pro was converted on a 5-mg scale and the products were purified by ion-exchange chromatography. The one- and two-dimensional NMR spectra showed well-resolved signals of a second product, besides those of 4-Me-3-Hyp, which could be clearly assigned to 3-hydroxy-4-hydroxymethyl-l-proline (4-HyMe-3-Hyp) (see Fig. S5). In particular, the AB system of the CH₂OH group in the ¹H NMR spectrum was identified as a crucial difference from the methyl group doublet observed in the spectrum of 4-Me-3-Hyp (see Fig. S6).

FIG 4.

(A) Time-dependent conversion of 4-Me-Pro (5 mM) with HtyE analyzed by fluorescence HPLC. The increase in the secondary product 4-HyMe-3-Hyp was coupled with a decrease in 4-Me-3-Hyp (representative example of 4 experiments under slightly varying assay conditions). (B and C) HPLC (B) and LC-MS (C) ion chromatograms of a conversion of 4-Me-Pro with HtyE after 105 min. (D) Reaction scheme of 4-Me-Pro conversions with HtyE and GloF.

A prolonged conversion of 4-Me-Pro with GloF also led to a secondary product in trace amounts. Surprisingly, this was not 4-HyMe-3-Hyp, as evidenced by a different retention time on HPLC analysis. The mass of 146 (m/z, M+H⁺) indicated a second monohydroxylated product (see Fig. S3). Dihydroxylated species were not detected. Hence, the monohydroxylated product must have been produced from 4-Me-Pro in parallel to the main product 4-Me-3-Hyp. Through a comparison of retention times, the cis-diastereomers of 3- or 4-hydroxylated 4-Me-Pro (4-Me-cis-3-Hyp and 4-Me-cis-4-Hyp, respectively) could be excluded as putative products. References of these compounds were readily available by hydroxylation of 4-Me-Pro with cis-3-PH type II and the pipecolic acid hydroxylase GetF (25). We therefore speculate that the novel product is trans-4-hydroxymethyl-l-proline (4-HyMe-Pro), formed through methyl group hydroxylation in analogy to the generation of 4-HyMe-3-Hyp (Fig. 4D).

To confirm that 4-HyMe-Pro and 4-HyMe-3-Hyp are actually formed by enzyme catalysis, further experiments were performed. 4-Me-Pro was converted quantitatively into 4-Me-3-Hyp by incubation with HtyE for 40 min; virtually no dihydroxylation product was detected in the mixture at this time. Then the enzyme was removed via ultrafiltration. No 4-HyMe-3-Hyp was formed upon prolonged incubation of the enzyme-free solution, and the concentration of 4-Me-3-Hyp remained constant. In a further experiment, first 4-Me-Pro was converted quantitatively into 4-Me-3-Hyp with GloF, and then HtyE was added to the mixture. Only then was there formation of 4-HyMe-3-Hyp, as observed by HPLC. Interestingly, the monohydroxylated side product of GloF catalysis was barely affected upon incubation with HtyE. Although the concentration of the side product, possibly 4-HyMe-Pro (of which no reference was available), was too low to enable quantitative determinations, apparently this compound is not a preferred substrate for trans-3-hydroxylation by HtyE.

Kinetic characterization of HtyE.

In analogy to the kinetic characterization of GloF (18), the catalytic parameters were determined for HtyE with both physiological substrates (Pro and 4-Me-Pro). The substrate concentration was varied (0.1 to 40 mM), and conversions in the presence of excess oxygen (air), 2OG, and Fe(II) were analyzed after a reaction time of 5 min. The data were fitted into a Michaelis-Menten model (see Fig. S4), and K_m and k_cat values were determined (Table 1).

TABLE 1.

Kinetic parameters for Pro and 4-Me-Pro conversion with HtyE, and for comparison with GloF, according to a Michaelis-Menten model

Substrate/product	Km (mM)	kcat (s−1)	kcat/Km (mM−1 s−1)	Sp act (nmol min−1 mg−1)
HtyE
Pro	4.2 ± 0.58	0.647 ± 0.0215	0.153 ± 0.0217	930 ± 30.9
4-Hyp		0.390 ± 0.0156	0.0918 ± 0.0137	560 ± 22.5
3-Hyp		0.251 ± 6.0 × 10−3	0.0661 ± 8.1 × 10−3	362 ± 8.7
4-Me-Pro	1.9 ± 0.25	1.04 ± 0.02	0.560 ± 0.0748	1,500 ± 28.5
GloF (18)
Pro	8.7 ± 0.55	0.125 ± 0.0029	0.0141 ± 9.7 × 10−4	178 ± 2.9
4-Hyp		0.112 ± 0.0022	0.0129 ± 8.3 × 10−4	159 ± 3.1
3-Hyp		0.0134 ± 7.2 × 10−4	0.0012 ± 2.0 × 10−4	19 ± 1.0
4-Me-Pro	1.7 ± 0.25	0.095 ± 0.0026	0.055 ± 0.0081	135 ± 3.7

Comparing the enzymatic kinetic data (Table 1), it is striking that the turnover number (k_cat) for both substrates is significantly higher for HtyE than for GloF (5× increase for Pro and 10× increase for 4-Me-Pro). K_m and specific activity of the enzyme HtyE are of the same order of magnitude as for the recently discovered and distantly related fungal pipecolic hydroxylases, which convert Pro as a less preferred substrate into 4-Hyp (19). We assume that the substantial and well-reproducible difference in activity between HtyE and GloF is primarily due to the intrinsic properties of the enzymes; however, other interfering factors, such as the “quality” of the protein preparation [i.e., Fe(II) occupation, protein folding, denaturation, etc.] and nonoptimal reaction conditions, cannot be fully excluded.

The Michaelis constants (K_m) for 4-Me-Pro are similar for the two enzymes (1.9 mM for HtyE and 1.7 mM for GloF). Both values are significantly lower than those determined for Pro (4.2 mM for HtyE and 8.7 mM for GloF). For better comparability of the substrate and product selectivities, the turnover numbers (k_cat) were normalized using the corresponding k_cat for 4-Me-Pro as an internal standard (k_{cat, rel} [4-Me-Pro] 100%). By analogy, the catalytic efficiencies (k_cat × K_m⁻¹) were normalized (k_cat × K_m⁻¹_{, rel} [4-Me-Pro] 100%). In these relative terms, GloF converts Pro slightly faster than 4-Me-Pro (k_{cat, rel} [Pro] = 118%), while in HtyE catalysis the conversion of 4-Me-Pro is faster (k_{cat, rel} [Pro] = 62%). This difference is compensated by the lower K_m of HtyE for Pro, so the relative catalytic efficiencies for Pro are almost identical: 28% for HtyE and 26% for GloF.

For the echinocandin-producing fungi, physiological Pro concentrations far below K_m can be expected. Consequently, the reaction rate v approaches the catalytic efficiency (v ≈ k_cat × K_m⁻¹). This means that at identical substrate concentrations, the ratios of (3/4)-Hyp to 4-Me-3-Hyp may be similar for HtyE and GloF.

DISCUSSION

Like GloF, HtyE is an enzyme with a relatively strict substrate specificity and only accepted Pip besides the native substrates Pro and 4-Me-Pro. In conversions with 4-Me-Pro, the hydroxylation was not entirely trans-3 selective, though the side reactions were slow. When HtyE was used, considerable amounts of the primary product 4-Me-3-Hyp were slowly oxidized at the methyl group to yield 4-HyMe-3-Hyp. In GloF conversions, traces of the 4-Me-Pro substrate were hydroxylated to a second product besides 4-Me-3-Hyp. Indirect evidence obtained by HPLC and MS suggests that this additional compound is 4-HyMe-Pro. Although these side reactions may be irrelevant in vivo for the host fungi, they provide a clue as to the biosynthesis of the echinocandin cryptocandin by Cryptosporiopsis cf. quercina, in which 4-HyMe-3-Hyp is reported at position 6 (26).

A significant difference between the enzymes is found in the regioselectivity of Pro hydroxylation. The 4-Hyp/3-Hyp product ratio is about 8:1 for GloF and 2:1 for HtyE. In the case of GloF, this corresponds well with production of pneumocandins A₀ and B₀ in an approximate ratio of 7:1 (which requires a 4-Hyp/3-Hyp ratio of 8:1). However, the increased proportion of 3-Hyp in HtyE catalysis is counterintuitive, as this compound is not required for echinocandin B biosynthesis. Thus, 3-Hyp production seems to be a waste of metabolic resources, unless the compound is used in another, as-yet-unknown biosynthetic pathway. So far, there is no evidence for production of 3-Hyp by HtyE in vivo; nevertheless, our results for GloF have shown that this enzyme's product selectivities are very similar in vivo and in vitro. The limited regioselectivity of HtyE also indicates that the exclusion of 3-Hyp in echinocandin B biosynthesis must be attributed to the substrate specificity of the corresponding A domain 6 of the NRPS. This result is in accordance with a previous structural comparison of the corresponding A domains from A. pachycristatus and G. lozoyensis based on homology models (12).

Regarding echinocandin biosynthesis in general, 4-Me-3-Hyp (and not 3-Hyp) is found at position 6 in most structures (13), although the production of the precursor 4-Me-Pro requires additional metabolic energy. Therefore, it can be hypothesized that 4-Me-3-Hyp confers an advantage over 3-Hyp to the producer strain in its natural habitat, even though a strong bioactivity has been found for echinocandins with 3-Hyp under laboratory conditions (27). The preference for 4-Me-3-Hyp is also reflected in the significantly lower K_m values of the PHs for 4-Me-Pro than for Pro. In fact, the K_m values of 8.7 mM (GloF) and 4.2 mM (HtyE) for Pro indicate a remarkably low affinity for a native substrate to an enzyme. It should be noted, however, that the underlying hypothesis that 3-Hyp is a less favorable compound in echinocandins is currently only based on the dominance of 4-Me-3-Hyp in biosynthesis. To definitively assess the significance of the methyl group at the 3-Hyp residue, more meaningful data on the production in nature and the chemical ecology of the producing fungi are required.

If it is assumed that 4-Me-3-Hyp is the privileged building block at position 6 in echinocandins, an evolutionary explanation of the selectivity of the PHs becomes possible: in the case of echinocandin B biosynthesis by Aspergillaceae, the selectivity for the methylated Hyp is guaranteed by the substrate-specific A domain 6 of the nonribosomal peptide synthetase, and thus, the selectivity of HtyE is of minor importance. In contrast, in pneumocandin biosynthesis, the incorporation of the Hyp building block at position 6 is exclusively controlled by the selectivity of GloF. This increases the evolutionary force on this enzyme to be more trans-4 selective, even if this is accompanied by an impaired catalytic performance.

The more promiscuous A domain 6 of GloF is also interesting for generating chemical diversity, as it effectively allows the introduction of different Pro congeners into nonribosomal peptides, simply controlled by the supplementation of substrates. A detailed analysis of pneumocandin production by G. lozoyensis has shown that at least five l-prolines are accepted by this A domain [4-Me-3-Hyp, 3-Hyp, 4-Hyp, (3R,4S)-3,4-dihydroxy-Pro, and Pro] corresponding to pneumocandins A to E, respectively, whose portion can thus be influenced by the feeding of Pro and congeners (28).

The PH HtyE involved in echinocandin B biosynthesis from A. pachycristatus was overproduced in E. coli, and its catalytic properties were determined and compared to those of the previously characterized enzyme GloF from pneumocandin biosynthesis in G. lozoyensis. Although the functions of the two enzymes in biosynthesis are comparable and their sequences are very similar (64% identity), three major differences have been found. (i) HtyE is about 1 order of magnitude more active than GloF. (ii) Upon longer incubation, HtyE additionally hydroxylates the methyl group of the primary product 4-Me-3-Hyp, such that 4-HyMe-3-Hyp is obtained at about a 30% yield. This transformation was not found in GloF conversions. Instead, trace amounts of 4-Me-Pro were converted into a second monohydroxylated product besides 4-Me-3-Hyp, putatively 4-HyMe-Pro. This would mean that both enzymes catalyze methyl group hydroxylation, but of different substrates: while GloF exclusively accepts 4-Me-Pro, HtyE converts only 4-Me-3-Hyp. 4-HyMe-3-Hyp is also reported for the echinocandin cryptocandin produced by Cryptosporiopsis cf. quercina (Helotiales) (26). Information concerning the biosynthesis of cryptocandin is unavailable, and thus, the origin of the hydroxymethyl group remains speculative. (iii) Comparing the product selectivity of Pro hydroxylation, the proportion of 3-Hyp relative to 4-Hyp is significantly increased in HtyE catalysis. This is remarkable because 3-Hyp is not required for echinocandin B biosynthesis. If it is assumed that, in evolutionary terms, 4-Me-3-Hyp is the privileged building block at position 6 in echinocandins, it is probably the substrate-specific incorporation of 4-Me-3-Hyp by the NRPS in echinocandin B biosynthesis that relieves HtyE from the force toward a stricter trans-4 selectivity.

Pro hydroxylation with HtyE and GloF is an example of a non-product-selective yet well-controlled enzymatic step in secondary metabolism. While in pneumocandin biosynthesis this leads to considerable product diversity (13), in echinocandin B biosynthesis only one Hyp isomer is used and the biosynthetic pathway remains linear. Metabolites not utilized in the native pathway can, in principle, be of importance for the evolution of secondary metabolism, because they are readily available for new and developing biosynthetic routes. Cryptic enzymatic activities, such as proline trans-3-hydroxylation by HtyE, are often overlooked by genome mining approaches. Even if the product is metabolized to another natural product, it can be notoriously difficult to identify the enzyme, as the gene is not located in the expected biosynthetic gene cluster and the enzyme may already be assigned to another catalytic activity in a distant cluster.

MATERIALS AND METHODS

General.

Chemicals were purchased from Roth (Germany) or Sigma-Aldrich (Germany). 4-Me-Pro was purchased from Santa Cruz Biotechnology (USA), fluoroprolines were obtained from Bachem (Switzerland), thioproline was from Fluka (Germany), and hydroxyprolines were purchased from Acros (USA).

Protein production and purification.

The codon-optimized gene htyE (see the supplemental material) was synthesized by Invitrogen/Thermo Fisher Scientific (Germany), ligated into vector pET-28b(+) (Novagen, Germany) using the restriction sites NdeI and SalI (5′ and 3′ ends, respectively), and heterologously expressed in E. coli BL21(DE3) Gold (Terrific Broth medium at 20°C and 180 rpm). The N- and C-terminally polyhistidine-tagged enzyme was purified by nickel-nitrilotriacetic acid (Ni-NTA) affinity chromatography and eluted with a buffer containing MES (50 mM, pH 6.8), NaCl (100 mM), glycerol (10%), and imidazole (gradient from 75 to 175 mM).

Activity assay.

Standard conditions for analytical conversions were as follows. The enzyme (8 μM) was incubated with the substrate (4 mM), 2OG (6 mM), ascorbate (1.2 mM), and FeSO₄ (0.4 mM) in MES buffer (50 mM, pH 6.8) in a total volume of 100 μl. The samples were incubated at 20°C (HtyE and GloF) in a rotary shaker (450 rpm) for 2 h. The reactions were stopped by addition of acetonitrile to a final concentration of 50% (vol/vol).

HPLC analysis.

The samples were centrifuged (10 min at 14,000 × g). The supernatant was derivatized with 9-fluorenylmethoxycarbonyl chloride (FMOC-Cl) according to a standard procedure (22) and then analyzed on a Chromolith high-resolution RP18e column (1.15 μm; 100 by 4.6 mm; Merck KGaA, Germany) using a gradient of HPLC buffers A (24 mM sodium acetate, 20% acetonitrile [pH 5]) and B (6 mM sodium acetate, 80% acetonitrile [pH 5]).

Catalytic conversion in a milligram scale.

4-Me-Pro (5 mg) in reaction buffer (8 mM substrate; 5 ml) was incubated with HtyE (7 mg) for 16 h at 20°C. After separation of precipitated protein by centrifugation (30 min at 15,500 × g), the amino acids were purified by cation-exchange chromatography with Dowex 50WX8 resin in hydrogen form. Elution with aqueous NH₃ (12.5%) and concentration of the product solution under reduced pressure gave a solid product, which was analyzed by ¹H NMR spectroscopy.

LC-MS.

An HPLC series 1100 system (Agilent Technologies, USA) was used in combination with a triple quadrupole LC-MS/MS QT4500 mass spectrometer (AB Sciex, Germany). The samples from the activity assays were purified on a hydrophilic interaction liquid chromatography (HILIC) column (XBridge BEH amide; 2.5 μm; 2.1 by 100 mm; Waters, USA) with a gradient of acetonitrile and ammonium acetate buffer (10 mM, pH 5) (flow rate, 0.3 ml min⁻¹; column temperature, 45°C; injection volume, 10 μl).