New Sipanmycin Analogues Generated by Combinatorial Biosynthesis and Mutasynthesis Approaches Relying on the Substrate Flexibility of Key Enzymes in the Biosynthetic Pathway

Combinatorial biosynthesis has proved its usefulness in generating derivatives of already known compounds with novel or improved pharmacological properties. Sipanmycins are a family of glycosylated macrolactams produced by Streptomyces sp. strain CS149, whose antiproliferative activity is dependent on the sugar moieties attached to the aglycone. In this work, we report the generation of several sipanmycin analogues with different deoxysugars, showing the high degree of flexibility exerted by the glycosyltransferase machinery with respect to the recognition of diverse nucleotide-activated sugars. In addition, modifications in the macrolactam ring were introduced by mutasynthesis approaches, indicating that the enzymes involved in incorporating the starter unit have a moderate ability to introduce different types of β-amino acids. In conclusion, we have proved the substrate flexibility of key enzymes involved in sipanmycin biosynthesis, especially the glycosyltransferases, which can be exploited in future experiments.

ABSTRACT

The appearance of new infectious diseases, the increase in multidrug-resistant bacteria, and the need for more effective chemotherapeutic agents have oriented the interests of researchers toward the search for metabolites with novel or improved bioactivities. Sipanmycins are disaccharyl glycosylated macrolactams that exert antibiotic and cytotoxic activities. By applying combinatorial biosynthesis and mutasynthesis approaches, we have generated eight new members of the sipanmycin family. The introduction of plasmids harboring genes responsible for the biosynthesis of several deoxysugars into sipanmycin-producing Streptomyces sp. strain CS149 led to the production of six derivatives with altered glycosylation patterns. After structural elucidation of these new metabolites, we conclude that some of these sugars are the result of the combination of the enzymatic machinery encoded by the introduced plasmids and the native enzymes of the d-sipanose biosynthetic pathway of the wild-type CS149 strain. In addition, two analogues of the parental compounds with a modified polyketide backbone were generated by a mutasynthesis approach, feeding cultures of a mutant strain defective in sipanmycin biosynthesis with 3-aminopentanoic acid. The generation of new sipanmycin analogues shown in this work relied on the substrate flexibility of key enzymes involved in sipanmycin biosynthesis, particularly the glycosyltransferase pair SipS9/SipS14 and enzymes SipL3, SipL1, SipL7, and SipL2, which are involved in the incorporation of the polyketide synthase starting unit.

IMPORTANCE Combinatorial biosynthesis has proved its usefulness in generating derivatives of already known compounds with novel or improved pharmacological properties. Sipanmycins are a family of glycosylated macrolactams produced by Streptomyces sp. strain CS149, whose antiproliferative activity is dependent on the sugar moieties attached to the aglycone. In this work, we report the generation of several sipanmycin analogues with different deoxysugars, showing the high degree of flexibility exerted by the glycosyltransferase machinery with respect to the recognition of diverse nucleotide-activated sugars. In addition, modifications in the macrolactam ring were introduced by mutasynthesis approaches, indicating that the enzymes involved in incorporating the starter unit have a moderate ability to introduce different types of β-amino acids. In conclusion, we have proved the substrate flexibility of key enzymes involved in sipanmycin biosynthesis, especially the glycosyltransferases, which can be exploited in future experiments.

INTRODUCTION

Discovery and development of bioactive compounds have become of great concern in recent years, not only because some infectious diseases are emerging but also because several pathogens are becoming resistant to commonly used clinical treatments. Furthermore, there is a need to generate novel compounds to be used in cancer chemotherapy, due to the rapid development of resistance to chemotherapeutic agents, the high toxicity associated with these drugs, their undesirable side effects, and the demand for novel antitumor chemical entities that are active against untreatable tumors, with fewer side effects or with greater therapeutic efficiency (1, 2).

Natural products are the agents most commonly used against both animal and human diseases. Among them, the most abundant are those produced by bacteria, specifically from the genus Streptomyces. The rate of new drug discovery has decreased in past decades, due to the low success rates for classical screening programs and the reisolation of already known compounds. Researchers have focused their efforts on searching for new bioactive compound producers in poorly studied ecosystems, as marine sediments, or associated with other organisms, such as ants, wasps, or sponges (3–5). In addition, the improvements in sequence data analysis and genetic molecular modification techniques have opened new fields in drug development through (i) the awakening of silent gene clusters leading to the production of bioactive compounds, as, for example, in the case of stambomycin, a glycosylated polyketide that was discovered through the overexpression of a LuxR transcriptional activator in Streptomyces ambofaciens ATCC 23877 (6), and (ii) combinatorial biosynthesis strategies that have been successfully applied to generate, among others, several paulomycin derivatives with different sugar moieties attached to the core structure (7) and different caboxamycins bearing distinct substitutions in the aryl ring (8).

Glycosylated natural compounds with important biological properties have been described, and some of them, such as erythromycin and doxorubicin, have been widely used for antibacterial and anticancer treatments, respectively. Glycosylation has a great influence on the biological activity of compounds; it can modify drug pharmacokinetic profiles, solubility, and transport, and it also participates in the recognition of the drug molecular target (9, 10). The cytotoxic activity of the polyketide jadomycin depends on the sugar attached to the aglycone (11). Furthermore, mithramycin bioactivity was improved by combinatorial biosynthesis leading to derivatives with different saccharides (12). Thus, glycodiversification could be a valuable tool to obtain improved new analogues of core scaffolds by changing the sugar moieties attached to them.

Sipanmycins are glycosylated 24-membered macrolactams isolated from Streptomyces sp. strain CS149 that have been described as antibacterial and cytotoxic agents (13). Their biosynthesis starts with the incorporation of a 3-aminobutyrate molecule (generated from α-glutamic acid through the action of seven enzymes, SipL1 to SipL7) into the polyketide synthase (PKS) machinery. Then, six units of malonyl-coenzyme A (CoA), three units of methylmalonyl-CoA, and one isobutylmalonyl-CoA (in sipanmycin A [SIP-A]) or 2-(2-methylbutyl)-malonyl-CoA (in sipanmycin B [SIP-B]) are condensed by five PKS enzymes to give rise to the complete macrolactam ring skeleton. Finally, two aminodeoxysugars are attached to the aglycone, first UDP-d-xylosamine (synthesized from N-acetylglucosamine by SipS1, SipS2, and SipS3) and then TDP-d-sipanose (synthesized by seven enzymes, i.e., SipS7, SipS6, SipS10, SipS11, SipS13, SipS8, and SipS5). It was demonstrated by knockout experiments that each glycosylation step requires the coordinated activity of two different glycosyltransferases (GTs); the SipS4/SipS15 pair attaches UDP-d-xylosamine to the aglycone, and the SipS9/SipS14 pair (with the aid of the additional helper protein SipO2) introduces TDP-d-sipanose (14) (see Fig. S1 at https://figshare.com/s/87fcece083b744468230). The disaccharide attached to the sipanmycin aglycone by its unique GT machinery has been proved to be essential for antibacterial and cytotoxic activities. Similarly, the presence or absence of a hydroxy group at the C-10 position of the aglycone determines the biological activity of this class of compounds (14). Thus, both the sugar moiety and the aglycone could be promising modification targets to obtain new analogues with improved pharmacological properties.

In this work, we have explored the versatility of some enzymes involved in sipanmycin biosynthesis, to obtain novel derivatives. We have tested the substrate flexibility of the GT pairs SipS4/SipS15 and SipS9/SipS14, which naturally transfer UDP-d-xylosamine and TDP-d-sipanose, respectively, to the sipanmycin aglycone (14). In addition, we have substituted the natural β-amino acid unit that serves as a starter in the biosynthesis of the macrolactam ring. By using combinatorial biosynthesis and mutasynthesis approaches, we have generated novel sipanmycin analogues and tested their biological activities.

RESULTS

Glycosylated sipanmycin derivatives obtained by combinatorial biosynthesis.

Plasmids pLNRT, pFL844T, and pLNBIVT (directing the biosynthesis of d-olivose, l-amicetose, and l-digitoxose, respectively) (see Fig. S2 at https://figshare.com/s/87fcece083b744468230) (15) were introduced into wild-type Streptomyces sp. strain CS149 by intergeneric conjugation (Escherichia coli-Streptomyces). Clones harboring these sugar biosynthesis plasmids were grown on R5A liquid medium, and samples were extracted with ethyl acetate at 3, 5, and 7 days and analyzed by ultraperformance liquid chromatography (UPLC) and high-performance liquid chromatography-mass spectrometry (HPLC-MS).

Comparative analysis of production profiles for the wild-type CS149 strain and the CS149(pLNBIVT) strain showed the production of compounds SIP-A and SIP-B in both strains and two extra peaks in the latter strain, sharing UV absorption spectra with sipanmycins (Fig. 1A and B) but differing in their [M+H]⁺ ions (m/z 753 for compound 1 and m/z 767 for compound 2). These compounds were purified by HPLC, and their chemical structures were determined by high-resolution mass spectrometry (HRMS) and nuclear magnetic resonance (NMR) spectroscopy (see Fig. S3 to S14 and Table S1 for compound 1 and Fig. S15 to S21 and Table S2 for compound 2 at https://figshare.com/s/87fcece083b744468230). Structural elucidation of compounds 1 and 2 identified them as α-l-digitoxyl-(1→4′)-3′-O-demethylsilvalactam, which we designated SIP-A1, and its congener carrying an extra carbon in the aliphatic substituent at C-2, which we designated SIP-B1, respectively, following the same trivial nomenclature we originally employed for wild-type SIP-A and SIP-B. SIP-A1 and SIP-B1 thus correspond to analogues of compounds SIP-A and SIP-B, respectively, in which the second sugar moiety, β-d-sipanose, has been replaced by α-l-digitoxose (Fig. 1B).

FIG 1.

Novel sipanmycin analogues obtained by combinatorial biosynthesis. UPLC chromatograms (at 320 nm) of culture extracts of wild-type Streptomyces sp. strain CS149 (A), strain CS149(pLNBIVT) (B), strain CS149(pLNRT) (C), and strain CS149(pFL844T) (D) are shown. Molecular structures of the corresponding derivatives are included next to each chromatogram. As expected, a negative-control strain harboring the empty vector pEM4T did not show any difference in its production profile, compared to the wild-type strain (data not shown).

Similarly, the CS149(pLNRT) strain produced compounds SIP-A and SIP-B but also two extra compounds (compounds 3 and 4) (Fig. 1A and C), with [M+H]⁺ ions at m/z 753 and m/z 767. Compounds 3 and 4 were purified by HPLC and their structures were determined by HRMS and NMR (see Fig. S22 to S29 and Table S3 for compound 3 and Fig. S30 to S40 and Table S4 for compound 4 at https://figshare.com/s/87fcece083b744468230). Structural elucidation of compounds 3 and 4 identified them as β-d-olivosyl-(1→4′)-3′-O-demethylsilvalactam, which we designated SIP-A2, and β-d-olivomycosyl-(1→4′)-3′-O-demethylsilvalactam, which we designated SIP-A2b, respectively. Both compounds shared the aglycone and the first deoxysugar with compound SIP-A but differed in the second one (β-d-olivose for SipA2 and β-d-olivomycose for SIP-A2b) (Fig. 1C).

In culture extracts of the CS149(pFL844T) strain, compounds SIP-A and SIP-B were not observed but two novel peaks (compounds 5 and 6), showing the characteristic absorption spectra of sipanmycins, with UPLC retention times only slightly different from those expected for compounds SIP-A and SIP-B (Fig. 1A and D) and with [M+H]⁺ ions at m/z 764 and m/z 778, were identified. After purification and HRMS and NMR analyses, their structures were elucidated as β-d-forosaminyl-(1→4′)-3′-O-demethylsilvalactam, which we designated SIP-A3, and its congener carrying an extra carbon in the aliphatic substituent at C-2, which we designated SIP-B3, respectively. Compared with wild-type SIP-A and SIP-B, the terminal sugar in SIP-A3 and SIP-B3 has been substituted with β-d-forosamine (corresponding to β-3-demethyl,3-deoxy-d-sipanose) (see Fig. S41 to S52 and Table S5 for compound 5 and Fig. S53 to S61 and Table S6 for compound 6 at https://figshare.com/s/87fcece083b744468230) (Fig. 1D).

All of the novel sipanmycin analogues described so far have altered sugars in the second (terminal) position of the disaccharide chain. In order to evaluate the possibility of obtaining new sipanmycin analogues containing alternative sugars replacing the first sugar moiety, d-xylosamine, a mutant defective in the biosynthesis of UDP-xylosamine (149ΔXyl) was generated by replacement of the sipS1, sipS2, and sipS3 genes (encoding putative N-acetylglucosaminyl deacetylase, UDP-glucose-6-dehydrogenase, and nucleoside diphosphate [NDP] sugar epimerase, respectively) by an apramycin resistance cassette (see Fig. S62 at https://figshare.com/s/87fcece083b744468230). Culture extracts of the149ΔXyl mutant strain showed production of neither compounds SIP-A and SIP-B nor even the aglycones of the compounds. However, no novel glycosylated sipanmycin derivatives were produced by 149ΔXyl strains harboring the pLNBIVT, pLNRT, or pFL844T plasmid.

Novel sipanmycin derivatives generated by mutasynthesis.

In an attempt to modify the sipanmycin aglycone by replacing the natural starter unit (β-glutamic acid) with other β-amino acids, a Streptomyces sp. strain CS149 non-sipanmycin-producing mutant was generated by replacement of sipL4 (encoding lysine 2,3-aminomutase) by the apramycin resistant cassette, using plasmid pUH149ΔLAM (see Fig. S63 at https://figshare.com/s/87fcece083b744468230). The 149ΔLAM mutant strain obtained was cultivated in R5A liquid medium, and the production of sipanmycins was checked by UPLC analysis. Neither SIP-A nor SIP-B could be detected in this mutant strain (Fig. 2A and B). Production of sipanmycins was restored when 10 mM β-glutamic acid was added to 24-h-old R5A cultures of the 149ΔLAM mutant strain (Fig. 2C), thus proving the essential role of the SipL4 enzyme in the biosynthesis of β-glutamic acid, the proposed starter unit in the biosynthesis of sipanmycins (14).

FIG 2.

Generation of SIP-C and SIP-D in mutasynthesis experiments. (A to D) UPLC chromatograms (at 320 nm) of culture extracts of the wild-type CS149 strain (A), the 149ΔLAM mutant strain (B), and the 149ΔLAM strain fed with β-glutamic acid (C) or with 3-APA (D). (E) Chemical structures of SIP-C and SIP-D.

Twenty-four-hour R5A cultures of the 149ΔLAM mutant strain were independently fed with 10 mM dl-β-leucine, dl-β-phenylalanine, β-alanine, or racemic 3-aminopentanoic acid (3-APA), and samples were extracted with ethyl acetate at 24 and 48 h postfeeding. Analysis of UPLC chromatograms showed that two different peaks with the characteristic UV absorption spectra of sipanmycins were produced only in cultures fed with 3-APA (Fig. 2D). Compound 7 ([M+H]⁺ at m/z 808) and compound 8 ([M+H]⁺ at m/z 822) were purified and analyzed by HRMS and NMR, allowing their structural elucidation (see Fig. S64 to S72 and Table S7 for compound 7 and Fig. S73 to S77 and Table S8 for compound 8 at https://figshare.com/s/87fcece083b744468230). Compound 7, 28-methylsipanmycin A, which we propose to name SIP-C, and its congener carrying an extra carbon in the aliphatic substituent at C-2 (compound 8), which we propose to name SIP-D, correspond to SIP-A and SIP-B analogues, respectively, in which the starter unit 3-aminobutanoic acid (derived from β-glutamic acid) has been replaced by 3-APA (Fig. 2E).

Antibacterial activity and in vitro cytotoxicity of compounds.

The novel sipanmycin analogues showed neither antibacterial activity against Gram-negative bacteria Escherichia coli and Pseudomonas aeruginosa nor antifungal activity against Candida albicans, as reported previously for the parental compounds (14), but they exerted moderate antibacterial activity against the Gram-positive bacteria Micrococcus luteus and Staphylococcus aureus (Table 1). In particular, some of the analogues (SIP-A2b, SIP-A3, SIP-B3, and SIP-C) showed MIC values against Staphylococcus aureus that were lower than those of the corresponding parental compounds. In vitro cytotoxicity assays were also carried out with several tumor cell lines. The results indicated that the strongest activity was exerted by the parental compound SIP-A (Table 2), with 50% inhibitory concentration (IC₅₀) values ranging between 96 nM and 1.72 μM, depending on the cell line tested. All of the derivatives in which d-sipanose has been replaced by another sugar demonstrated higher IC₅₀ values, suggesting that the second sugar moiety plays an essential role in the bioactivity of sipanmycins. For most of the cell lines tested, SIP-B and SIP-C showed similar IC₅₀ values, compared to SIP-A, indicating that minor changes in the structure of the aglycone have minor effects on sipanmycin cytotoxicity.

TABLE 1.

Antibiotic activity of sipanmycins and derivatives

Compound	MIC (μg/ml)
	Micrococcus luteus	Staphylococcus aureus	Pseudomonas aeruginosa	Escherichia coli	Candida albicans
SIP-A	3.22 ± 1.87	10.58 ± 6.55	>100	>100	>100
SIP-B	3.63 ± 2.19	12.15 ± 6.22	>100	>100	>100
SIP-A1 (compound 1)	10.66 ± 7.06	16.31 ± 9.39	>100	>100	>100
SIP-B1 (compound 2)	19.17 ± 3.56	26.84 ± 8.60	>100	>100	>100
SIP-A2 (compound 3)	9.41 ± 2.34	22.59 ± 3.76	>100	>100	>100
SIP-A2b (compound 4)	4.79 ± 0.25	8.87 ± 0.71	>100	>100	>100
SIP-A3 (compound 5)	4.45 ± 0.55	6.68 ± 2.06	>100	>100	>100
SIP-B3 (compound 6)	7.29 ± 3.43	9.72 ± 1.21	>100	>100	>100
SIP-C (compound 7)	4.71 ± 0.58	6.06 ± 2.85	>100	>100	>100

TABLE 2.

In vitro cytotoxicity of sipanmycins and derivatives

Compound	IC50 (μM)
	3T3 (fibroblasts)	A549 (lung)	HT29 (colon)	HL60 (leukemia)	CAPAN-1 (pancreas)	MDAMB231 (breast)
SIP-A	0.134	0.096	0.201	1.72	0.189	0.181
SIP-B	0.103	0.096	0.185	2.54	0.206	0.213
SIP-A1 (compound 1)	0.410	1.15	0.783	1.52	1.99	1.26
SIP-B1 (compound 2)	0.436	0.430	0.718	2.16	2.10	0.466
SIP-A2 (compound 3)	0.618	0.529	0.647	1.26	4.03	0.633
SIP-A2b (compound 4)	0.448	1.28	0.436	0.686	2.81	2.05
SIP-A3 (compound 5)	0.184	0.229	0.410	3.39	0.676	0.448
SIP-B3 (compound 6)	0.187	0.233	0.364	2.13	0.959	0.469
SIP-C (compound 7)	0.124	0.144	0.146	3.91	0.289	0.194

DISCUSSION

Searching for new bioactive compounds to fight multidrug-resistant bacteria or emerging infectious diseases or to function as more effective chemotherapeutic agents has become a great challenge in recent years. Two of the strategies that are being employed to achieve this goal are the isolation of drug producers from unexploited ecosystems and the modification of already known bioactive metabolites. Glycosylated compounds have proved to be one of the most important types of metabolites regarding their use in clinical treatments, as in the case of erythromycin (antibiotic), amphotericin B (antifungal), and doxorubicin (chemotherapy) (9). In this sense, glycodiversification is a powerful tool directed toward expanding the structural diversity of glycosylated metabolites, but it requires the presence of a GT capable of recognizing several aglycones and sugars as substrates. Combinatorial biosynthesis approaches have been successfully used to generate new derivatives of already known metabolites of diverse chemical structures (16, 17). In the case of glycosylated compounds, this strategy led to the generation of novel analogues of mithramycin (12), paulomycin (7), and jadomycin (11), to mention just a few examples. Thus, we focused our work on determining the promiscuity of the GTs involved in sipanmycin glycosylation, as well as on the enzymes responsible for the introduction of the β-amino acid that serves as a starter unit for the PKS machinery and the possibility of generating novel derivatives by combinatorial biosynthesis and mutasynthesis.

Plasmids encoding enzymes for the biosynthesis of l-digitoxose, l-amicetose, and d-olivose (pLNBIVT, pFL844T, and pLNRT, respectively) (15) were introduced into wild-type Streptomyces sp. strain CS149 by intergeneric conjugation. Applying this combinatorial biosynthetic approach, six novel sipanmycin derivatives were obtained (Fig. 1). SIP-A1 and SIP-B1 were the result of the replacement of the β-d-sipanose residue (the native terminal deoxyaminosugar in SIP-A and SIP-B) by an α-l-digitoxose residue, the sugar synthesized by pLNBIVT (Fig. 3).

FIG 3.

Proposed biosynthetic pathway of the new sugars attached to sipanmycin derivatives obtained in this work. Sip enzymes involved are shown in bold letters. Squares mark the sugars in the final molecules of SIP-A1/SIP-B1 (blue), SIP-A2 (light orange), SIP-A2b (dark orange), and SIP-A3/SIP-B3 (green).

Similarly, when pLNRT was introduced into the wild-type CS149 strain, the β-d-sipanose moiety was replaced by β-d-olivose (SIP-A2) (Fig. 3). In this case, we could not detect the sipanmycin analogue of SIP-B containing d-olivose (the expected SIP-B3 congener), probably due to its low production levels. Surprisingly, we also observed the production of sipanmycin derivatives in which the second sugar moiety was replaced not by the sugar determined by the plasmid introduced into the wild-type strain but rather by a modified or unexpected sugar. Presumably, the biosynthetic machinery of the native CS149 d-sipanose (14) interfered with the deoxysugar biosynthesis directed by the introduced plasmid, yielding an unexpected terminal deoxysugar in the isolated products, or, alternatively, the biosynthetic machinery related to the introduced plasmid interfered with the biosynthesis of the native CS149 d-sipanose, yielding an unexpected terminal deoxyaminosugar in the isolated products. In the context of the first scenario, TDP-4-keto-2,6-dideoxy-d-glucose (also known as TDP-4-keto-d-olivose), which is a common intermediate derived from both native sip genes and ole genes from the different introduced plasmids, might be methylated at the C-3 position by the C-methyltransferase SipS5, the enzyme responsible for the 3-methylation of d-sipanose (14), and then the 4-keto group might be reduced by UrdR (from the introduced pLNRT plasmid directing d-olivose biosynthesis) to generate the d-olivomycose present in SIP-A2b (Fig. 3).

In line with the second scenario, the aminodeoxysugar d-forosamine present in SIP-A3 and SIP-B3 could be biosynthesized by the coordinated action of genes involved in native CS149 d-sipanose biosynthesis with the extra participation of a gene from pFL844T (the plasmid directing l-amicetose biosynthesis). The reductase UrdQ from the l-amicetose biosynthetic pathway would accept the common intermediate TDP-4-keto-2,6-dideoxy-d-glucose, catalyzing its C-3 deoxygenation to yield TDP-4-keto-2,3,6-trideoxy-d-glucose, which, by the action of the aminotransferase SipS13, would be converted into TDP-4-amino-2,3,6-trideoxy-d-glucose, which later would be dimethylated by SipS8 (as in d-sipanose biosynthesis [Fig. 3]), yielding the final TDP-d-forosamine. In conclusion, the GT pair SipS9/SipS14, which is involved in the second sipanmycin glycosylation step and acts as an inverting GT, could accept as activated TDP-glycosides different d-configured sugars, including neutral and aminodeoxysugars, as well as deoxysugars from the l-configured series, such as l-digitoxose, thus yielding sipanmycin analogues in which the terminal sugar residue has a β-d or α-l absolute configuration, according to Klyne’s rule (18).

Very interestingly, the SipS9/SipS14 pair is capable of transferring d-forosamine (as found in the new analogues SIP-A3 and SIP-B3), the same monosaccharide that is natively transferred by the spinosyn forosaminyltransferase SpnP (a member of the GT-1 family), which is involved in the last glycosylating step in the biosynthetic pathway for spinosyns (19). Thus, it is logical to propose SipS9 and/or SipS14 as a homologue of SpnP. We had already indicated (14) that SipS14 (but not SipS9) contains the putative motif involved in GT-auxiliary protein interactions (H-X-R-X₅-D-X₅-R-X_12–20-D-P-X₃-W-LX_12–18-E-X4-G), as described for the GT SpnP (20).

Fortunately, the structural studies carried out with SpnP (20) have identified key residues for the glycosyl transfer, such as the basic residue (H13) that is involved in deprotonation of the acceptor hydroxyl in the aglycone and the key negatively charged residues (D356 and E357) that interact with the positively charged tertiary amine at C-4 of the aminodeoxysugar forosamine. Similarly, key residues forming hydrophobic interactions with the thymine moiety (L254, Y315, and L318) or the asparagine/threonine pair (N230 and T335) in which the N230 amino group forms a hydrogen bond with the 3′-OH of the TDP unit (this asparagine/threonine pair sterically excludes UDP by clashing with its 2′-OH) have been identified, together with key residues interacting with the pyrophosphate moiety via backbone NH groups (T335 and T336) or side chain contacts (S11, S12, H331, S333, T335, and T336) (20). Alignment of the complete sequence of the structurally characterized GT SpnP with those of both SipS14 and SipS9 (and their homologues in incednine biosynthesis, IdnS14 and IdnS9, respectively, which are involved in the transfer of N-demethyl-d-forosamine) (see Fig. S78 at https://figshare.com/s/87fcece083b744468230) revealed that the key basic residue (H13) and the key residues involved in C-4-deoxyaminosugar recognition (D356 and E357) are present in SipS14 (and IndS14) but not in SipS9 (or IndS9), confirming the latter protein as a mere auxiliary protein not having native GT activity for transferring C-4 aminodeoxysugars, in agreement with our previous results in which residual production of SIP-A was observed only in the ΔsipS9 mutant and not in the ΔsipS14 mutant (14).

Similarly, some of the SpnP key residues involved in the interaction with the TDP moiety were observed only in SipS14 (and IndS14) and not in SipS9 (or IndS9), whose classification as a GT is thus doubtful (21). Overall, SpnP and SipS14 (and IdnS14) must be structurally similar GTs (belonging to the GT-1 family) that natively transfer C-4 aminodeoxysugars. The flexibility of SipS14 for transferring neutral and C-4 aminodeoxysugars of both l-configured and d-configured series described in this work not surprisingly parallels the flexibility already reported for its homologue SpnP, which showed the ability to transfer l-mycarose (22) and l-olivose (19) instead of the native d-forosamine in biotransformation experiments with engineered strains of Saccharopolyspora erythraea, leading to the generation of new spinosyn analogues. In a similar manner, the substrate-flexible GT DesVII (which also requires an auxiliary protein, DesVIII), which is responsible for transfer of the d-desosamine unit (a C-3 aminodeoxysugar) in the glycosylation step involved in the biosynthesis of narbomycin and the macrolide antibiotic YC-13, demonstrated its capacity to transfer a variety of l- and d-configured neutral sugars instead of the native aminosugar (d-desosamine) in combinatorial biosynthesis experiments, leading to new narbomycin analogues (23) and new antibiotic YC-13 analogues (24). Other flexible GTs worth mentioning include StaG, one of the GTs involved in indolocarbazole staurosporine biosynthesis, which was able to transfer several NDP-activated sugars (including l-configured, d-configured, neutral, and aminated sugars) in heterologous expression experiments (25), and AraGT, the GT involved in aranciamycin biosynthesis, which accepted different nucleotide-activated neutral sugars of both l- and d-configured series (such as d-amicetose, l-rhodinose, l-rhamnose, and l-axenose) in heterologous expression experiments (26).

Unfortunately, no sipanmycin derivatives carrying differences in the first aminosugar were found, even when plasmids harboring genes for sugar biosynthesis were introduced into a mutant strain in which the three genes responsible for UDP-xylosamine synthesis were replaced by the apramycin resistance cassette. A reasonable explanation would be the inability of the GT pair SipS4/SipS15 to recognize a deoxysugar activated by TDP instead of UDP. A few examples of GTs that can recognize and transfer sugars activated with different NDPs have been described, however. This is the case for VinC, the GT that is responsible for the attachment of TDP-vicenisamine to vicenistatin in vivo but is also able to transfer UDP- and ADP-activated vicenisamine in vitro (27).

It has been demonstrated that the nature of the starter unit can determine important characteristics of the polyketide (28). In the case of the macrolactams, the starter unit is a β-amino acid that is recognized by an adenylation enzyme, ligated to a stand-alone acyl carrier protein, and finally transferred to the PKS machinery (29). We explored the possibility of generating sipanmycin derivatives by substituting the 3-aminobutanoic acid (derived from β-glutamic acid) used by the PKS as a starter unit. Feeding experiments with 149ΔLAM mutant strain cultures using β-Leu, β-Phe, or β-Ala did not yield any sipanmycin analogue production, but compounds SIP-C and SIP-D were generated with the addition of 3-APA. These results are in accordance with those obtained by Cieślak and coworkers in 2017 (30). In vitro studies of the substrate specificity of IdnL1 (the adenylation enzyme involved in the recognition of 3-aminobutanoic acid as the starter unit for the biosynthesis of incednine) showed its preference for short-chain 3-amino fatty acids such as 3-aminobutanoic acid or 3-APA. In contrast, IdnL1 exhibited weak activity against β-Ala, medium-chain fatty acids, and the aromatic β-amino acid β-Phe. Thus, the failure to obtain sipanmycin derivatives with β-Leu, β-Ala, or β-Phe could be due to the high specificity of the adenylation enzyme SipL1. Amino acid sequence comparisons between SipL1 and IdnL1 pointed out their high similarity level (76.4%), which might explain the in vivo behavior observed in sipanmycin biosynthesis. Indeed, amino acids of the substrate-binding pocket showed that SipL1 could be classified into the short-chain fatty acid recognition type of adenylation enzymes (the group that includes IndL1) (29).

In conclusion, with the generation of eight novel sipanmycin analogues, we have shown the substrate flexibility of some of the enzymes involved in the biosynthesis of these compounds, particularly those participating in the incorporation of the polyketide backbone starter unit and the GTs involved in the attachment of d-sipanose, which could recognize different deoxysugars and aminodeoxysugars. The biosynthetic flexibility of those enzymes contrasts with the apparent inflexibility of the biological activity shown by the novel compounds generated, which in general exert lower cytotoxic or antibiotic activity than their parental compounds. The only exception to that observation involves derivatives SIP-A2b, SIP-A3, SIP-B3, and SIP-C, which were more active against Staphylococcus aureus than the parental compounds SIP-A and SIP-B.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The sipanmycin producer Streptomyces sp. strain CS149 (13) was used as the host strain during this work. For metabolite production, 30 ml of tryptic soy broth was inoculated with spores, which were grown for 24 h at 30°C and 250 rpm. This seed culture was used to inoculate 50 ml of R5A medium (31) to a final optical density at 600 nm (OD₆₀₀) of 0.2. Medium A (32) was used for sporulation, and mannitol soy medium containing 20 mM MgCl₂ (33) was used for intergeneric conjugation. Escherichia coli DH5α (34) was used for routine cloning procedures, and E. coli ET12567(pUB307) (35, 36) was used for intergeneric conjugation; both strains were grown in 2× tryptone-yeast extract (TY) medium. Culture media were supplemented with antibiotics, i.e., apramycin (100 μg/ml for E. coli or 25 μg/ml for Streptomyces), thiostreptone (25 μg/ml), kanamycin (25 μg/ml), tetracycline (10 μg/ml), chloramphenicol (25 μg/ml), and/or nalidixic acid (50 μg/ml), when needed.

DNA manipulation and plasmids used in this work.

DNA manipulations were performed according to standard procedures for E. coli (37) and Streptomyces (38). All PCR amplifications were carried out with the high-fidelity polymerase Herculase II Fusion (Agilent Technologies), following the manufacturer’s instructions. Plasmids pLNRT, pLNBIVT, and pFL844T (15) were used for combinatorial biosynthesis (see Fig. S2 at https://figshare.com/s/87fcece083b744468230). pEM4T (39) was used as an empty vector for comparative purposes.

pUH149ΔXyl was generated to eliminate the genes involved in UDP-xylosamine biosynthesis (sipS3, sipS2, and sipS1) in Streptomyces sp. strain CS149. Upstream and downstream flanking regions were amplified by PCR (oligonucleotide primer pairs 149dXyl.5F [5′-TAT GAA TTC AAC TGG ACA TCG TCG CTG A-3′]/149dXyl.5R [5′-TAT AAG CTT GCA CGC TCG ACG AGA TCA T-3′] and 149dXyl.3F [5′-TAT CAT ATG GCG CAA CCA CTA TCA GGA GT-3′]/149dXyl.3R [5′-TAT TCT AGA GCC AGG ACC ATC TTC ATC AC-3′]) and cloned into pUO9090 (40). After digestion with SpeI, the replacement cassette was cloned into pHZ1358 (41) (see Fig. S62 at https://figshare.com/s/87fcece083b744468230).

pUH149ΔLAM was constructed in order to obtain a nonproducing sipanmycin strain (149ΔLAM mutant strain) through replacement of the sipL4 gene (coding for a lysine 2,3-aminomutase). As described above, flanking regions were amplified by PCR using the oligonucleotide primer pairs 149dLAM64.5F (5′-TAT GAA TTC AGA CTG TAG ATG TGC GTG CG-3′)/149dLAM64.5R (5′-TAT AAG CTT CCA CCT CGT CCA TGT GC TG-3′) and 149dLAM64.3F (5′-TAT CAT ATG TGG ATG GAC CAT CTG GAG CT-3′)/149dLAM64.3R (5′-TAT TCT AGA GTC CGG GTA CAC GTA GAA GC-3′). Amplicons were cloned into pUO9090 (40) at both sides of the apramycin resistance gene. The resulting plasmid was digested with SpeI, and the replacement cassette was cloned into the pHZ1358 (41) XbaI site to obtain the final conjugative vector for knockout experiments (see Fig. S63 at https://figshare.com/s/87fcece083b744468230).

Feeding experiments.

All reagents were purchased from Sigma-Aldrich, including β-glutamic acid (product no. G1763-50MG), dl-β-leucine (product no. 17988-1G-F), dl-β-phenylalanine (product no. 159492-5G), β-alanine (product no. 146064-25G), and racemic 3-APA (product no. BBO000720-1G). Aqueous solutions of these reagents were added to 24-h cultures of the 149ΔLAM strain in R5A medium to a final concentration of 10 mM. Samples were extracted at 24 and 48 h postfeeding to check the compound production.

Extraction, analysis by UPLC and HPLC-MS, and isolation of compounds by semipreparative HPLC.

Whole cultures (1 ml) of selected strains were extracted with 1 volume of ethyl acetate at different times and analyzed by UPLC and HPLC-MS, as described previously (13). Two-liter cultures of mutant CS149 strains in R5A medium were used to purify sipanmycin derivatives by semipreparative HPLC, as described previously (13).

Structural elucidation of novel compounds.

Compounds 1 to 8 were analyzed by liquid chromatography-diode array detection-electrospray ionization–time of flight mass spectrometry (LC-DAD-ESI-TOF MS) to determine their UV-visible (DAD) spectra and their molecular formulae, based on the experimental accurate masses and the corresponding isotopic distribution. The structural elucidation of each compound was carried out by detailed analysis of one-dimension and two-dimension NMR spectra, assisted by comparison with the spectroscopic data reported for incednine (42), silvalactam (43), and especially SIP-A and SIP-B (13). Relative configurations were determined by coupling constants and nuclear Overhauser effect analyses, assisted by comparison with the NMR data for SIP-A and SIP-B (13). Absolute configuration proposals were supported by biosynthetic arguments. A detailed description of the structural elucidation of each compound is presented, with the corresponding spectral data, in the supporting material (posted at https://figshare.com/s/87fcece083b744468230).

LC-DAD-ESI-TOF MS and NMR analyses.

HRMS and UV-visible spectra were obtained by LC-DAD-ESI-TOF MS analyses performed using an Agilent 1200RR HPLC system, equipped with a SB-C₈ column (2.1 by 30 mm; Zorbax), coupled to a Bruker maXis mass spectrometer. Chromatographic and ionization conditions were identical to those we employed for our dereplication routines (44), which also were previously employed for SIP-A and SIP-B (13) and the first series of derivatives we obtained while studying their biosynthesis (14). NMR spectra were recorded in CD₃OD at 24°C with a Bruker Avance III 500-MHz NMR spectrometer (500 and 125 MHz for ¹H and ¹³C NMR, respectively) equipped with a 1.7-mm TCI MicroCryoProbe, using the residual solvent signal as an internal reference.

In vitro cytotoxicity and antibiotic activity assays.

The cytotoxicity of compounds was tested against the following human tumor cell lines: colon adenocarcinoma (HT29), non-small-cell lung cancer (A549), breast adenocarcinoma (MDA-MB-231), promyelocytic leukemia (HL-60), and pancreatic cancer (CAPAN-1). The mouse embryonic fibroblast cell line NIH/3T3 was used as a control, to evaluate cytotoxicity against nonmalignant cells. These analyses were carried out as described previously (14).

Antibiotic activity tests (MIC assays) were performed in 96-well microtiter plates. Fresh cultures of each microorganism (Micrococcus luteus, Staphylococcus aureus, Escherichia coli, Pseudomonas aeruginosa, and Candida albicans) were used as seed cultures to inoculate the plates, with the appropriate compound concentration, to a final OD₆₀₀ of 0.1 and a total volume of 150 μl per well. Plates were incubated overnight at 37°C (30°C for C. albicans).