Keywords

7.1 Introduction

Chemical and biological synthetic approaches for producing natural products with complex structures and intriguing biological properties have largely progressed as separate fields, each aiming to efficiently access the target compounds [1,2,3,4]. Both approaches have distinct advantages and challenges, with limited exploration of their combined potential for the total synthesis of natural products and their analogues. Several robust enzymes, such as lipases and alcohol dehydrogenases, have traditionally been employed for optical resolution and enantioselective reduction [5,6,7]. Recent advances in gene analysis, gene synthesis, and genetic databases have propelled late-stage functionalization using enzymes such as P450s for site- and stereoselective oxidation of natural product scaffolds [8]. Concurrently, non-natural reactions through directed evolution techniques have progressed rapidly [9]. The complex skeletons of natural products are constructed by enzymes, typically non-ribosomal peptide synthetases, polyketide synthases, and terpene cyclases. However, the generally large size and unfavorable physical characteristics of these enzymes make them challenging to manipulate, even using current heterologous expression techniques, resulting in their infrequent use in synthetic applications.

Fig. 7.1
Chemical structures of saframycin A 1, cyanosafracin B 2, jorunnamycin A 3, renieramycin M 4, ecteinascidin 743 and lubinectedin 6. The structures are from the alkaloidal family.

The THIQ alkaloidal family and a synthetic derivative (6)

Full size image

The bis-tetrahydroisoquinoline (THIQ) alkaloids, which include saframycins (1), safracins (2), jorunnamycins (3), and renieramycins (4), are important families of antitumor antibiotics [10] (Fig. 7.1). These natural products are characterized by a complex, densely functionalized pentacyclic framework constructed by non-ribosomal peptide synthetases (NRPS) that can alkylate DNA via electrophilic iminium species generated from an aminonitrile or a hemiaminal group at C21 [11,12,13,14,15]. Representative THIQ alkaloid ecteinascidin 743 (5) which has a macrolactone and an additional THIQ system shows exceptional antitumor activity. It has been approved for treating ovarian neoplasms and sarcoma [16]. Additionally, in 2020, the U.S. FDA sanctioned the use of a semi-synthetic compound, lurbinectedin (6), to treat small cell lung cancer [17, 18]. This semi-synthetic analog is distinguished by its spiro-fused β-tetrahydrocarboline unit, which replaces the THIQ unit of the macrolactone of 5. These anticancer agents are prepared semi-synthetically in more than 20 chemical steps from fermentation-derived cyanosafracin B (2) [19, 20]. The significant therapeutic potential of these alkaloids, coupled with increasing demand for their synthetic production, highlights the importance of the THIQ scaffold as a key target for both chemical synthesis and engineered biosynthesis [21, 22].

7.2 Biosynthetic Machinery of THIQ Alkaloids

The biosynthesis of saframycin A (1) involves the modification of l-tyrosine to generate 7, catalyzed by SfmD/M2/M3, followed by construction of a complex pentacyclic bis-THIQ scaffold originating from l-Ala, Gly, and two molecules of 7 [11, 23, 24] (Fig. 7.2). This latter process is orchestrated by three modules of non-ribosomal peptide synthetase (NRPS), SfmA–C, which assemble the amino acids and a fatty acid [12, 13]. This biosynthetic assembly line for 1 is characterized by three distinct phases. First, the incorporation and subsequent removal of a fatty acid moiety are involved in both the pre- and post-assembly stages of the pentacyclic scaffold. Second, the unique PS domain responsible for the Pictet–Spengler (PS) reaction in the N-terminal of SfmC plays an important role in forming the THIQ rings, replacing the typical peptide condensation reaction commonly observed in the NRPS machinery. Lastly, the Red domain is crucial for reducing three thioesters tethered on the PCP (peptidyl carrier protein) domain to release the resultant aldehyde intermediates. This SfmC module is pivotal to the process, enabling the sequential assembly of two tyrosine-derived molecules with two aldehyde intermediates generated by reduction of their corresponding thioesters.

Fig. 7.2
A chemical pathway. L tyrosine is modified with methyl and methoxy followed by reduction, loading with A T P, Pictet Spengler reaction, reduction with N A D P H, reloading, Pictet Spengler reaction, reduction with N A D P H, cyclization, N methylation oxidation, removal of fatty acid and secretion, and treatment with S f m C y 2 to give saframycin S and A.

Proposed biosynthetic machinery of saframycin A (1)

Full size image

Even though saframycin A (1) lacks fatty acyl chains, the SfmA–C-catalyzed biosynthetic process begins by incorporating a long fatty acyl unit, like myristic acid. This step is facilitated by the enzyme SfmA, which contains an acyl ligation (AL) domain. Subsequent amide bond formation with l-Ala and Gly is catalyzed by SfmA and SfmB, respectively, to form an N-myristoyl alanyl-glycidyl thioester, intermediate A (Fig. 7.2). The crucial module in this pathway, SfmC, is equipped with a reduction (Red) domain that uniquely reduces thioester A to aldehyde 8, bypassing traditional amide bond formation. The subsequent imine formation and Pictet–Spengler (PS) reaction of the resultant aldehyde 8 and the primary amine in 7, which is tethered on the PCP domain, yields a bicyclic THIQ thioester B. This PS reaction incorporates an sp3 chiral center at C1 and completes the first stage of reactions facilitated by SfmC. The second stage of the sequential assembly process begins with the Red domain-catalyzed thioester reduction of intermediate B to liberate aldehyde intermediate 9. Reloading of 7 onto the PCP domain followed by PS cyclization forms intermediate C with a stereocenter at C11. Intermediate C is characterized by two THIQ segments connected with adjacent chiral centers (C3 and C11). The final transformation of intermediate C involves thioester reduction to release aldehyde 10, and subsequent spontaneous ring closure through intramolecular nucleophilic addition of the secondary amine. This leads to formation of the pentacyclic bis-tetrahydroisoquinoline core scaffold 11 containing a hemiaminal group at C21.

The next steps involve N-methylation at the N12 position of the secondary amine 11 and oxidation of two phenol rings on both wings (ring-A and E), leading to bis-quinone 12 [25]. Biosynthetic intermediate 12 then undergoes further modification by SfmE, a membrane-bound peptidase, which liberates the fatty acid moiety from the side chain at C1 [26]. The primary amine 13 is then secreted by transmembrane efflux protein SfmG. The final stages include oxidative deamination catalyzed by SfmCy2, the FAD-binding oxidoreductase, in extracellular region that installs a carbonyl group at the C25 position, thus producing saframycin S (14) [26]. Saframycin A (1) is then obtained through the cyanation of saframycin S using KCN.

The biosynthesis of saframycins, which uniquely involves the incorporation and subsequent removal of long fatty acyl chain, utilizes a mechanism also found in the NRPS biosynthetic machinery of other relevant THIQ alkaloids such as SF-1739/quinocarcin [27], naphthyridinomycin [28, 29], safracin [30], and ecteinascidin [31, 32]. This process, particularly the installation of fatty acid moieties, is believed to protect the terminal primary amine of the l-Ala component throughout the sequential transformations conducted by three NRPS modules SfmA–C. Furthermore, the enhanced lipophilicity is believed to play several roles. Firstly, it likely prevents the diffusion of aldehyde intermediates (8 and 9) away from the enzymatic machinery. These intermediates are released through the reduction of the corresponding thioesters tethered to PCP domains of SfmB and SfmC. Secondly, the enhanced lipophilicity may play a key role in the substrate recognition at presumably hydrophobic active sites within the SfmB–SfmC protein complexes. The fatty acid moiety on the C1 position also reduces the DNA alkylating capability of the resulting bis-THIQ scaffolds [33]. NRPS machineries responsible for the biosynthesis of saframycin, safracin, and naphthyridinomycin are thus believed to facilitate the production of less toxic prodrugs within cells. These antibiotics are then released into the extracellular space after the fatty acyl chains are cleaved off.

7.3 Integrating SfmC-Catalyzed Enzymatic Processes with Chemical Syntheses: Chemo-enzymatic Total Synthesis of THIQ Alkaloids

7.3.1 Impact of Fatty Acyl Chain Length on SfmC-Catalyzed Enzymatic Conversions

To develop a chemo-enzymatic hybrid process aimed at the total synthesis of THIQ alkaloidal natural products using the unique enzyme SfmC, we initially varied the chain length of fatty acid moiety to assess the effect of the substrate structure on its enzymatic conversion efficiency [12, 15, 34]. A series of peptidyl aldehydes was synthesized, each featuring a hydrophobic acyl chain with a different length. Specifically, we synthesized three kinds of aldehyde substrates, denoted as 15, 8, and 16, each carrying a fatty acid moiety consisting of twelve, fourteen, and sixteen carbon atoms, respectively. The aldehydes were subjected to SfmC-mediated transformations with chemically synthesized tyrosine derivative 7 [35]. To prepare an active holo-form SfmC equipped with a phosphopantetheinyl arm, phosphopantetheinyl transferase Sfp was co-expressed.

We evaluated the conversion rate of the three synthetic substrates, 15, 8, and 16, by comparing the UV absorbance peak areas of corresponding products 17–19 possessing the pentacyclic scaffolds as chromophores. Substrate 8, with a C14 lipophilic chain derived from myristic acid, demonstrated the best conversion efficiency among the tested aldehydes. The relative conversion rates of 15 and 16 bearing C12 and C16 fatty acyl chain, having just two less or two more methylene groups than substrate 8, resulted in markedly lower at 27% and 17%, respectively. These results underscore the critical importance of attaching fatty acyl moieties like myristic acid (C14) in the biosynthetic construction of bis-THIQ scaffolds. Recent gene deletion studies in Pseudomonas fluorescens A2-2 indicated the attachment of palmitic acid (C16) on intermediates is indispensable in the safracin B biosynthetic pathway [30], despite notable differences between the Streptomyces and Pseudomonas species. Our findings to date also underscore the critical importance of attaching fatty acid moieties involving myristic (C14) and palmitic (C16) acid in the NRPS biosynthetic machinery for bis-THIQ scaffolds [12]. Given the substantially hydrophobic character of peptidyl aldehydes, aldehyde substrate analog 8 with a C14 chain was selected as being optimal for maximizing conversion efficiency in subsequent in vitro chemo-enzymatic reactions (Fig. 7.3).

Fig. 7.3
A chemical pathway. Lauroyl, myristoyl and palmitoyl are treated with holo S f m C, N A D P H, M g C l 2, M n C l 2 and phosphate buffer followed by K C N, 2 pic B H 3 and H C H O to form pentacyclic compounds. A plot of relative conversion rates versus peak area is also plotted.

Conversion rates for synthetic substrates (15, 8, and 16) with tyrosine derivative 7 into pentacyclic compounds (17–19) via enzymatic reactions catalyzed by SfmC and following chemical conversions (cyanation, N-methylation). *Quantitative analysis derived from UV absorbance peak area of the bis-THIQ scaffold chromophores. **Error bars show calculated standard error of the mean (SEM) based on three replicates

Full size image

7.3.2 Chemo-enzymatic Transformation of Substrate Analogs with Cleavable Linkages

In our efforts to broaden the range of substrates used and to facilitate additional chemical modifications at the substituent at the C1 position, a series of peptidyl aldehydes (20–22) were designed and chemically synthesized. These aldehydes incorporate a chemically cleavable functional group, including ester or allylic carbamate, substituting the peptide linkage found in the biosynthetic intermediate 8. Although SfmE, a membrane-associated peptidase, selectively hydrolyzes the peptide bond connecting the fatty acid segment [26], it is still challenging to achieve chemoselective cleavage of one of the several peptide bonds in the intricate intermediate 18. We therefore incorporated a chemically cleavable ester or an allylic carbamate moiety, instead of the peptide linkage in the biosynthetic intermediate 8 [36]. Although altering amide bonds reduced the efficiency of the SfmC-catalyzed conversions, aldehyde 20 having an ester group close to the aldehyde moiety, nonetheless demonstrated a high conversion rate [34]. This led to the formation of compound 23, with about 55% efficiency as compared with the transformation of 8 to 18. Meanwhile, other synthetic substrates 21 and 22, which involve an ester or an allylic carbamate group substituting the peptide linkage near the fatty acid moiety, were transformed into the pentacyclic scaffolds 24 and 25, each with conversion rates of 14% in comparison to the transformation of 8 into 18. These findings suggest that the two peptide linkages in substrate 8 significantly influence the sequential reactions catalyzed by SfmC, especially the peptide linkage connecting the fatty acid moiety, which appears to be more critical than the peptide bond in the vicinity of the aldehyde moiety (Fig. 7.4).

Fig. 7.4
A chart of 8 chemical structures includes peptide aldehydes, ester and fatty acid moieties. A bar graph compares the peak areas for transitions from 8 to 18, 20 to 23, 21 to 24 and 22 to 25 in decreasing order.

Comparative conversion rates for synthetic substrates (8, 20–22) in SfmC-catalyzed conversions with tyrosine derivative 7, followed by cyanation and N-methylation, to yield the respective bis-THIQ scaffolds (18, 23–25). *Quantitative analysis derived from UV absorbance peak area of the bis-THIQ scaffold chromophores. **Error bars show calculated standard error of the mean (SEM) based on three replicates

Full size image

7.3.3 Chemo-enzymatic Total Syntheses of Saframycin A and Jorunnamycin A

The total synthesis of bis-THIQ alkaloids were carried out after establishing the SfmC-catalyzed chemo-enzymatic conversion of substrate analogs bearing cleavable linkers into pentacyclic scaffolds. We first applied the chemo-enzymatic strategy to saframycin A (1), demonstrating the flexibility of this approach to access to various bis-THIQ alkaloids [36] (Fig. 7.5). The methyl ketone moiety at the terminal of 1 was designed to be installed through a simple basic hydrolysis of the ester linkage using substrate analog 21 followed by oxidation of the resulting secondary alcohol. Merging enzymatic sequential assembly of synthetic substrates (21, 7) and chemical installation of an aminonitrile and N-Me group led to the formation of pentacyclic 24 with 13% isolated yield. Due to the instability of the intermediate 26, the extractive work up of 26 was followed immediately by its conversion into stable tertiary amine 24 using 2-picolineborane [37]. While the overall conversion efficiency for 21 was less than half that for 20 in analytical scale, we nonetheless achieved a semi-preparative scale synthesis to obtain 12.2 mg of the desired bis-THIQ core scaffold 24 by repeating the optimal in vitro chemo-enzymatic conversions three times. The combined yield for the pentacyclic 24 at this semi-preparative scale exceeded that achieved at an analytical scale. Unlike conventional multi-step syntheses, this method does not require the isolation of intermediates, significantly reducing the number of labor-intensive steps and enabling rapid production. Basic hydrolysis of 24 removed the fatty acyl chain, producing 27 in 91% yield. Subsequent oxidative conversions, including the Salcomine-mediated conversion of the two phenols followed by Swern oxidation of the resultant secondary hydroxyl group on the C1 side chain, enabled the five-pot chemo-enzymatic total synthesis of saframycin A (1) starting from two simple synthetic substrates, 7 and 21. Compared to the previously reported total synthesis of 1, this chemo-enzymatic approach could represent a distinct achievement, offering rapid access to medically important and structurally intricate THIQ natural products [38,39,40,41,42,43].

Fig. 7.5
A reaction schematic begins with L tyrosine modified methoxy and methyl and a fatty acid allylic carbamate moiety followed by 8 reaction steps to obtain saframycin A.

Chemo-enzymatic total synthesis of saframycin A (1)

Full size image

Next, we focused on the expeditious chemo-enzymatic total synthesis of jorunnamycin A [36] (Fig. 7.6, 3). By treatment with SfmC, synthetic substrates 7 and 20 were enzymatically converted to pentacyclic scaffold 29. This intermediate was then subjected to cyanation in a single-pot reaction to form aminonitrile 30. After removing the enzyme through centrifugation, further N-methylation of 30 was achieved using formaldehyde and 2-picolineborane, giving rise to 23 in just 30 min. This chemo-enzymatic process produced the pentacyclic intermediate 23 (6.5 mg) as a yield of 18% in a single day with the precise installation of multiple functional groups. Subsequent two-pot chemical conversions, including basic hydrolysis of the ester linkage in 23 followed by Salcomine-catalyzed oxidation of two phenols in 31 to their corresponding bis-quinones, allowed the chemo-enzymatic total synthesis of jorunnamycin A (3) using only four reaction vessels, starting from the synthetic substrates 7 and 20. While Zhu, Chen, Stoltz, and Yang previously achieved elegant total synthesis of 3, our chemo-enzymatic method offers an alternative approach to accessing this naturally occurring alkaloid [44,45,46,47,48].

Fig. 7.6
A reaction schematic begins with L tyrosine modified methoxy and methyl and a fatty acid allylic carbamate moiety, followed by 7 reaction steps to obtain jorunnamycin A.

Chemo-enzymatic total synthesis of jorunnamycin A (3)

Full size image

7.3.4 Chemo-enzymatic Synthesis of N-Fmoc Saframycin Y3

We further adapted our chemo-enzymatic method to synthesize a variant of saframycin A (1), known as saframycin Y3, which features a free primary amine moiety in its l-Ala unit [36] (Fig. 7.7). To this end, we employed substrate analog 22 possessing an allylic carbamate moiety instead of the amide linkage found in substrate 8. Designed substrate 22 allowed the site-selective removal of the fatty acid moiety under mild conditions while forming the primary amine in the saframycin Y3 structure. Conducting a chemo-enzymatic process in two pots using both 22 and 7 two-times provided more than 12 mg of compound 25 through 32. Efficient palladium-catalyzed scission of the allylic carbamate linker in 25 was achieved using a catalytic amount of Pd(PPh3)4 in dichloromethane. Phenylsilane was employed as a reductant for the resulting π-allyl palladium intermediate [49]. We anticipated that Salcomine-catalyzed oxidative conversions of 33 to bis-quinones would enable rapid access to saframycin Y3. However, our attempts to isolate saframycin Y3, which has a primary amine proximal to a quinone ring, were found to be challenging due to its instability. Therefore, the primary amino group was protected prior to oxidation in order to isolate N-Fmoc saframycin Y3 (35). The two-step conversions, including the palladium-catalyzed cleavage of allyl carbamate to remove the acyl chain followed by Fmoc protection, afforded intermediate 34 in 80% yield for the two steps. Subsequent oxidation of phenols in A and E-rings yielded N-Fmoc saframycin Y3 (35) with a yield of 59%. This chemo-enzymatic approach enabled the concise syntheses of 33 and 34, which have lower A-ring oxidation states compared to naturally occurring THIQ alkaloids. Interestingly, synthetic variant 33, lacking the C5 oxygen functional group exhibited higher DNA alkylating ability toward various double stranded DNAs bearing 5′-GGG-3′, 5′-GGC-3′, 5′-CGG-3′, 5′-AGC-3′, and 5′-AGT-3′ sequences in comparison to the commercially available natural product, cyanosafracin B (2), produced through fermentation of Pseudomonas fluorescens [33].

Fig. 7.7
A reaction schematic begins with L tyrosine modified methoxy and methyl and a fatty acid allylic carbamate moiety, followed by 8 reaction steps to obtain N F moc saframycin Y 3.

Chemo-enzymatic synthesis of N-Fmoc saframycin Y3 (35)

Full size image

7.3.5 Chemo-enzymatic Assembly of bis-THIQ Scaffolds Incorporating Diverse Amino Acid Derivatives

To expand the range of artificial substrates suitable for the established chemo-enzymatic transformations, we explored eight kinds of synthetic substrates [34] (Fig. 7.8a, 36a–h). These synthetic substrates featured a series l- or d-amino acids replacing the l-Ala in the biosynthetic intermediate 8. Analog 36a, which incorporates an l-Leu residue, exhibited the highest enzymatic conversion efficiency, reaching 91% relative to the biosynthetic intermediate 8 despite the increased steric bulk of the isopropyl sidechain in l-Leu than the methyl group in 8. The relative conversion rates for 36b and 36c, containing l-Met and l-Phe, were 33% and 25%, respectively, compared to 8. These experimental results suggested that SfmC can accommodate additional (methylthio)methyl or phenyl groups on the methyl group in 8.

Fig. 7.8
a. A chart compares the conversion rates of L-alanine, leucine, methionine, phenylalanine, valine, isoleucine, proline, D alanine and D valine. b. A reaction schematic presents the conversion of L methionine to a pentacyclic scaffold through 2 steps.

a Relative conversion rates of substrate analogs (36a–36h) with tyrosine derivative 7 compared to the biosynthetic intermediate 8 into their respective bis-THIQ scaffolds (37a–37h, 18), as catalyzed by SfmC enzymatic action, and subsequent cyanation and N-methylation. b Preparative-scale synthesis of pentacyclic scaffold 37b with an l-Met unit. *Quantitative analysis derived from UV absorbance peak area of the bis-THIQ scaffold chromophores. **Error bars show calculated standard error of the mean (SEM) based on three replicates

Full size image

Conversely, 36d and 36e, which respectively have l-Val and l-Ile, showed diminished conversion efficiencies of 23% and 6%, indicating that branched substituents on the l-Ala side chain in 8 substantially impede enzymatic conversion. While an increase in steric hindrance generally had a detrimental effect, we unexpectedly discovered that SfmC tolerated the installation of l-Pro in place of l-Ala, as evidenced by the modest conversion (12%) of 36f to pentacyclic 37f. Furthermore, SfmC could accommodate the reversal of the stereogenic center: substrates 36g and 36h bearing either a d-Ala of d-Val component, were transformed into the corresponding bis-THIQ scaffolds 37g and 37h, respectively, albeit with lower efficiencies (2% and 13%). Our findings imply that SfmC is somewhat flexible, allowing for the incorporation of amino acids, such as l-Leu, l-Met, or l-Phe, with additional substituents on the methylene moiety adjacent to the alpha carbon of l-Ala. However, changing to l-Val or l-Ile components bearing branched substituents, resulted in a significant drop in conversion efficiency. The feasibility of incorporating l-Pro, d-Ala, and d-Val was also demonstrated, albeit with modest conversion efficiency.

Notably, two-pot semi-preparative scale conversions of 7 and 36b allowed rapid generation of greater than 2 mg of pentacyclic 37b (2.16 mg) by utilizing a synthetic substrate with the replacement of l-Ala with l-Met [34] (Fig. 7.8b). This semi-preparative hybrid synthetic method provides a rapid and convenient means to secure the minimum amount of sample (approximately 1–2 mg) necessary for structural analysis and in vitro assays of novel compounds closely relevant to natural products that are otherwise challenging to access. Therefore, this chemoenzymatic approach offers a rapid synthetic platform for accelerating the discovery of drug leads from natural products and their analogs.

7.4 Conclusion

In this chapter, we paid attention to the NRPS biosynthetic assembly line for artificial biosynthesis of bis-THIQ alkaloids and their analogs. We developed a chemo-enzymatic hybrid process that integrates enzymatic sequential one-pot conversions with the precise chemical synthesis of designed substrates and functional group manipulation of the products of the enzymatic conversions. We confirmed the previously cryptic roles of long-chain fatty acyl groups, especially the C14-chain, in streamlining the cascade one-pot enzymatic transformations. Furthermore, we demonstrated that SfmC possesses a relatively wide substrate tolerance and can assemble the tyrosine derivative and various synthetic peptidyl aldehydes into the corresponding pentacyclic scaffolds of bis-THIQ alkaloids. Our approach enabled the expeditious total synthesis of saframycin A and jorunnamycin A, demonstrating the effectiveness and flexibility of this chemo-enzymatic hybrid synthetic platform. These relatively unexplored approaches, featuring integration of enzymatic and chemical synthesis, could facilitate further development of hybrid processes to gain rapid, robust, and customizable access to therapeutically valuable natural products-based molecules.