Synthesis and Conformational Analysis of FR901464-Based RNA Splicing Modulators and Their Synergism in Drug-Resistant Cancers

FR901464 is a cytotoxic natural product that binds splicing factor 3B subunit 1 (SF3B1) and PHD finger protein 5A (PHF5A), the components of the human spliceosome. The amide-containing tetrahydropyran ring binds SF3B1, and it remains unclear how the substituents on the ring contribute to the binding. Here, we synthesized meayamycin D, an analogue of FR901464, and three additional analogues to probe the conformation through methyl scanning. We discovered that the amide-containing tetrahydropyran ring assumes only one of the two possible chair conformations and that methylation of the nitrogen distorts the chair form, dramatically reducing cytotoxicity. Meayamycin D induced alternative splicing of MCL-1, showed strong synergism with venetoclax in drug-resistant lung cancer cells, and was cancer-specific over normal cells. Meayamycin D incorporates an alkyl ether and shows a long half-life in mouse plasma. The characteristics of meayamycin D may provide an approach to designing other bioactive L-shaped molecules.


■ INTRODUCTION
Natural products FR901464, 1−3 pladienolide, 4−6 herboxidiene, 7,8 and thailanstatin A 9,10 (Figure 1a) modulate RNA splicing and exhibit potent anticancer activity.The tetrahydropyran rings of FR901464 bind splicing factor 3b subunit 1 (SF3B1) of the spliceosome. 11,12−29 Despite this, little is known about the atomic detail of the amide-containing tetrahydropyran ring in the SF3B1 active site.Prior to the discoveries of FR901464, a simplified variant of the amide fragment was studied by the Vasella group.They found that 34% of 3-acetamido-tetrahydropyran adopted the N-axial chair conformation in D 2 O at 298 K (Figure 1b). 30hey proposed that an intramolecular hydrogen bond was formed between the N−H group with the ring oxygen, stabilizing the N-axial conformer.
We previously reported FR901464 analogues, meayamycins, and conducted a structure−activity relationship (SAR) study. 31owever, it remained unclear which of the two chair conformers depicted in Figure 1c accounts for SF3B1 binding.The Webb group addressed the active conformation of the pyran ring of FR901464; their computational modeling placed the 3-acetamido-tetrahydropyran in the N-equatorial chair conformation (Figure 1d). 32In addition, the Nicolaou group found 3,6-dihydro-2H-pyran-containing analogues to be substantially less potent. 28n this paper, we describe the synthesis and conformational analysis of four new analogues to systematically probe the amide-containing six-membered ring conformation in meayamycin D (Figure 1e).Additionally, meayamycin D has activity comparable to meayamycin A in cancer cell lines and is significantly more metabolically stable.We show that meayamycin D has synergistic effects with venetoclax and osimertinib in lung cancer cell lines and reveal promising cancer specificity in colon cell lines with meayamycin D and meayamycin E. Our results provide insight into the active binding conformation of FR901464-related molecules and highlight O-alkyl ethers as compatible functional groups for future analogue developments.

■ RESULTS AND DISCUSSION
Our first objective was to replace the esters or carbamates at the C4′-position with an ether group to prevent Z-to-E isomerization of the side chain during the amide coupling we previously reported. 33Specifically, methoxymethyl (MOM) ether was chosen as a suitable replacement, which proved to suppress undesired isomerization. 33We also anticipated that the MOM ether might be stable in plasma. 34Therefore, our first synthetic target was meayamycin D. Following the route developed for meayamycin B, 35 (S)-(−)-ethyl lactate 1 was alkylated with in situ generated methoxymethyl chloride (the commercial supply of MOMCl was unstable during this study) to yield MOM ether 2 in 95% yield (Scheme 1a).Next, this compound was subjected to a sequential DIBALH reduction and Z-selective Horner−Wadsworth−Emmons reaction using Ando-type phosphonate 5 36 and t-BuOK to afford Z-enoate 3 in 69% yield, with a Z/E ratio of >99:1.Enoate 3 was hydrolyzed to acid 4 in 96% yield, completing the synthesis of the side chain fragment.Amine 6 was prepared using our previously reported method. 33Trifluoroacetic-acid-mediated Boc-deprotection followed by coupling with acid 4 using HATU formed amide 8 in 61% yield with retention of the olefin configuration (Scheme 1b, top).Next, this amide underwent olefin cross-metathesis with methacrolein, using nitro-Grela catalyst, to give aldehyde 9 in 74% yield.Finally, a Wittig olefination with Ph 3 P = CH 2 afforded diene 10 in 86% yield.
We turned our attention toward synthesizing N-methyl meayamycin D to eliminate the intramolecular hydrogen bond (Scheme 1b, bottom).In the original report by Vasella, Nmethylation of the 3-acetamido-tetrahydropyran yielded only the N-equatorial conformer. 30Therefore, we hypothesized that N-methyl meayamycin D might be constrained with the side chain in the equatorial conformation.The Boc-protected amine 6 was reduced with LiAlH 4 to generate N-methyl amine 16 quantitatively.The coupling of amine 16 and acid 4 with HATU gave the tertiary amide 17 in 55% yield.Due to the amide bond's slow rotation, the NMR spectrum of the tertiary amide presented signal peaks for two rotamers.Attempts to resolve these peaks by variable-temperature NMR were unsuccessful; therefore, one-dimensional nuclear Overhauser effect (NOE) experiments were used to confirm the presence of a single diastereomer (Figures S1 and S2). 37After confirming the structure, amide 17 was subjected to olefin cross-metathesis to give aldehyde 18 in 40% yield, which, upon Wittig olefination, delivered the N-methylated coupling partner 19 in 87% yield.
Analysis of the 15-H coupling constants in tetrahydropyrans 17, 18, and 19 led us to believe these compounds do not exist in a similar conformation to their non-N-methylated counterparts (8, 9, and 10).We asked whether removing the C12 methyl group enabled the tetrahydropyran ring to adopt the Naxial conformation more readily.Therefore, we synthesized a C12-desmethyl analogue (i.e., 23) of diene 19.We previously speculated that the hydrophobicity of the C12 methyl group might be important for binding. 31Based on this speculation, the N-methyl substituent might mimic the hydrophobicity of the C12 methyl group when oriented with the methyl group directed toward the C12 hydrogen and compensate for the direct loss of the C12 methyl group.The synthesis of N-methyl C12-desmethyl analogue 23 follows the same sequence as the N-methyl analogue 19 (Scheme 1b, bottom); Boc-protected amine 31 11 was reduced with LiAlH 4 to the corresponding Nmethyl amine 20 quantitatively.Amine 20 was coupled with acid 4 using HATU to afford amide 21 in 54% yield.The additional NMR signals of comparable hydrogens between amides 21 and 17 were shifted almost identically, implying that amide 21 is similarly observed as rotamers.Amide 21 underwent olefin cross-metathesis and Wittig olefination giving Finally, the C12-desmethyl analogue of meayamycin D (henceforth denoted as meayamycin E) was synthesized to investigate the effect on the diaxial interaction between the C12 and C14 positions.Our group previously synthesized and evaluated the morpholino carbamate analogue of meayamycin E. 31 The synthesis follows the same general route as its Nmethylated counterpart (Scheme 1b, top; see the Experimental Section for details).Employing nitro-Grela catalyst, olefin cross-metathesis of fragments 10, 15, 19, and 23 with the righthand fragment 24 was achieved to provide analogues meayamycin D, N-methyl meayamycin D, N-methyl meayamycin E, and meayamycin E in yields ranging between 2 and 4% after HPLC purification (see Scheme 1c for specific yields).
With the representative compounds in hand, the 3 J 15,14 value was used to characterize the conformation of the pyran ring and compare its resemblance to the natural product ring conformation (Figure 2).The Karplus equation describes the relationship between the vicinal coupling constant ( 3 J) observed in 1 H NMR and the dihedral angle between those protons. 38,39In the ideal chair conformation of the tetrahydropyran, the dihedral angle between 14-H and 15-H the N-equatorial conformer predominates in protic solvent due to the competing intermolecular hydrogen bond formed between solvent molecules and the substrate (see Figure 1b).However, unlike Vasella's model system, the N-equatorial conformation in meayamycin D presents a 1,3-diaxial interaction between the C11 and C15 substituents.Therefore, compounds 8, 13, 17, and 21 were analyzed in CD 3 OD to determine whether any analogues adopt a different conformation in a protic environment.In CD 3 OD, compounds 8 and 13 had similar 3 J 15,14 values of 2. Next, we aimed to quantify the percentage of the secondary amide 8 in the N-axial and N-equatorial conformations.At room temperature, the observed coupling constants and chemical shifts are averages of all conformers.However, the 1 H NMR signal for the hydrogen vicinal to the acetamide group (14-H in meayamycin D) is distinguishable at cryogenic temperatures. 30Therefore, we performed the 1 H NMR study of amide 8 in CD 2 Cl 2 from 21 to −81 °C (Figures 3a and  S12).Surprisingly, the 14-H signal was not split at any of the temperatures scanned.Additionally, we performed IR experiments to characterize the ring conformation further.The Vasella group studied the conformation of 3-acetamido-pyrans by IR spectroscopy to show a splitting of the N−H band based on the distribution between the unbound state and intramolecular-bound state. 30We acquired the IR spectrum of amide 8 in CCl 4 at 5 mM; CCl 4 was chosen to mimic the hydrophobicity of the binding pocket 12 and exclude the effects of solvation or aggregation.We observed only one peak for the N−H stretching at 3453 cm −1 (Figure 3b), indicating that amide 8 exists as a single conformer in CCl 4 . 1 H NOE analysis of the 11-, 13-, and 15-Hs signifies that the sole conformation observed is the N-axial conformation (Figures S13−S15).To gain further insight, we performed DFT calculations of amide 8.The lowest energy conformation located for amide 8 in dichloromethane placed the tetrahydropyran ring in the Naxial conformation (Figure S16).Additionally, the 14 H−15 H dihedral angle was calculated to be 57°, matching closely with the angle determined by 1 H NMR coupling.Finally, an X-ray crystal structure of amide 8 was obtained and confirmed that the solid-state conformation of amide 8 is solely the N-axial conformation (Figure 3c).Taken together, this data indicates that the amide-containing ring of FR901464 is essentially "locked" in one chair conformation and does not require further rigidity to improve the binding affinity.
Meayamycin D, meayamycin E, and their N-methyl analogues were evaluated for their cytotoxicity against various cancer cell lines (Table 1 and Figure S17).Meayamycin D exhibited GI 50 comparable to meayamycin A, ranging from 2.0 to 3.9 nM.Meayamycin E exhibited a slightly higher GI 50 , which may be caused by the loss of the CH-π interaction 40 between the C12 methyl group and Y36 of PHF5A (Figure 4). 12N-Methyl meayamycin D and N-methyl meayamycin E showed GI 50 over 1000-fold greater than their non-Nmethylated counterparts, highlighting a necessity for the C14−N-axial conformation.However, we cannot disregard the possibility that the substitution of the methyl group might remove a necessary hydrogen bonding interaction which could be more critical than the altered conformation of the Nmethylated analogues.It is also possible that these Nmethylated analogues are still weakly bound to SF3B1 and inhibit growth through alternative means as observed with other inactive SF3B1-inhibitors. 41Finally, the potent compounds were evaluated in normal colon epithelial cells (CCD-841CoN) and liver hepatocytes (FL83B).The GI 50 values for meayamycin A and meayamycin D were slightly higher (ca.2to 4-fold) in normal cells.Meanwhile, there was a 3 to 6-fold difference in the GI 50 for normal cells treated with meayamycin E. These results suggest meayamycins may be further tuned to selectivity target cancer cells and serve as an avenue for future drug development.
We evaluated the ability of meayamycin D and meayamycin E to alter RNA splicing, leading to differential protein expression.We observed a dose-dependent decrease in phosphorylation of Thr313 on SF3B1 (Figure 5a).Since SF3B1 is phosphorylated in the early stages of splicing, this result suggests that the compound is binding SF3B1 in the early stage of splicing. 42Meayamycin D and meayamycin E dose dependently decreased the expression of myeloid cell leukemia-1 (MCL-1L; long isoform) with a new band resulting from the alternatively spliced MCL-1S (short isoform).We previously showed that the protein and mRNA levels of MCL-1L and MCL-1S are correlated, indicating that the Western blot analysis of MCL-1 in the current study reflects their RNA transcripts. 43,44These meayamycin analogues also resulted in p27 truncation, previously reported by Yoshida et al. 11 Similar results were observed in DMS114 cells (Figure 5b).Interestingly, we observed an alternatively spliced form of p21 (Figure S18).This larger p21 isoform was recently reported by the Lunec group, which demonstrated that when cells were treated with E7107 (pladienolide analogue), p21 L retained an intron and lost its nuclear localization signal, resulting in a loss of CDK inhibitory activity. 45In addition, they observed alternative splicing of p53 with the pladienolide analogue E7107, which we did not observe with the meayamycins.Further studies are warranted to validate alternative splicing of p21.Taken together, these results indicate that these new analogues affect alternative splicing in the same manner as meayamycin A.   Cancer cells can innately resist anticancer drugs by overexpressing antiapoptotic proteins such as BCL-2 and MCL-1L. 46Therefore, inhibiting both proteins kills cancer cells more effectively than inhibiting only one of them. 47We investigated the combination of venetoclax (FDA-approved BCL-2 inhibitor) and meayamycin D in both venetoclaxresistant small-cell lung cancer (SCLC) DMS114 cells and venetoclax-sensitive SCLC DMS53 cells.The Chou−Talalay method defines the additive effect as the combination index (CI) = 1, synergism as CI < 1, and antagonism as CI > 1. 48,49 Figure 5c,d shows that the drug combination was strongly synergistic (CI < 0.5).
Drug resistance poses a therapeutic challenge, especially in patients with advanced EGFR-mutated NSCLC.Despite FDA approval of a third-generation mutant-EGFR inhibitor (osimertinib), acquired resistance occurs in an EGFR-dependent and -independent manner. 50One of the EGFRindependent mechanisms may involve RNA splicing. 51Patients with high levels of alternative splicing events have poorer overall survival. 51Further, siRNA knockdown of PHF5A, which is overexpressed in NSCLC, reduced cell proliferation and induced apoptosis. 52Given these findings, we evaluated the combination of meayamycin D and osimertinib in osimertinib-resistant NSCLC HCC827AR cells.This drug combination was synergistic (CI = 0.8, Figure 5e) in these resistant cells.In addition, the MCL-1 inhibitor S63845 was strongly synergistic with osimertinib as previously demonstrated (Figure 5f). 53inally, we assessed the half-lives of meayamycin D and meayamycin E in mouse plasma using procaine as a positive control (Figure 6a).Plasma contains many nonspecific esterase enzymes that will rapidly cleave the acetyl group of meayamycin A, affecting its ADME properties.As expected, the hydrolysis of meayamycin A was nearly complete after 10 min (t 1/2 = 1−2 min).Meanwhile, meayamycin D and meayamycin E were significantly more stable, with a half-life of 13 and 8 h, respectively.Furthermore, in the crude extracts of plasma treated with meayamycin A, we detected a mass peak corresponding to the acetate-hydrolysis product at the C4′ position.We also observed the time-dependent increase of this species (Figure 6b) that mirrors the decreasing signal of meayamycin A. While the rapid cleavage of the C4′-acetyl in meayamycin A is not surprising, it suggests that other FR901464 analogues such as spliceostatin A are also metabolically labile at that position.At the time of submission, there were no metabolic studies reported for spliceostatin A. However, our results suggest that spliceostatin A and other acetate-containing analogues are likely susceptible to nonspecific esterase enzymes.When analyzing the fraction of meamyamycin D and meayamycin E after 48 h, we did not detect the corresponding hydrolysis product, indicating the greater stability of MOM ether at the C4′-position in mouse plasma.
Combination Index Analysis.Cells were plated in 96-well plates at an initial density of 1500 or 5000 cells per well in RPMI-1640 medium (100 μL) and were incubated for 24 h prior to compound addition.The compounds were prepared separately as 10 mM in 100% DMSO or 10 μM in 100% DMSO.Serial dilution in sterile water gave 10× dilutions that were added directly to the cells alone or in combinations as 100× dilutions to give the desired concentration of compound: 0.5−3 nM meayamycin D, 1.25−10 μM S63845, 25 nM to 8 μM venetoclax, and 0.25−4 μM osimertinib.The cells were then incubated for additional 72 h.Cell proliferation was measured by using the commercial WST-1 dye reduction assay.To determine whether drug combinations were synergistic, we used CompuSyn software to calculate the combination index (CI) using nonconstant drug ratios. 48A CI of less than 1.0 was considered to be indicative of synergism, and this interaction was further classified as strong synergism (CI < 0.3), synergism (CI of 0.3−0.7),and slight to moderate synergism (CI of 0.7−0.9).

■ ASSOCIATED CONTENT
* sı Supporting Information

Figure 6 .
Figure 6.(a) In vitro plasma stability of meayamycin A, meayamycin D, and meayamycin E in mouse CD1 plasma over 48 h.n = 2. (b) Degradation of meayamycin A over 6 min.*The curve was determined by the appearance of species with m/z 464.3070 corresponding to C4′−OH meayamycin (calcd for [M + H] + C 26 H 42 NO 6 464.3007).

Table 1 .
Half-Maximal Growth Inhibition Concentrations (GI 50 ) for Meayamycin A, Meayamycin D, and Analogues a a Values reported as mean ± standard deviation (SD) of n ≥ 3 individual experiments (see Figure S17 for growth inhibition curves).b Normal cells.c p < 0.05.d p < 0.01.e p < 0.001, significance between HCT116 or SW48 and normal cells.