Design, Synthesis and Biological Evaluation of Novel C14-C3’BzN–Linked Macrocyclic Taxoids

Abstract

Novel macrocyclic paclitaxel congeners were designed to mimic the bioactive conformation of paclitaxel. Computational analysis of the “REDOR-Taxol” structure revealed that this structure could be rigidified by connecting the C14 position of the baccatin moiety and the ortho position of C3’N-benzoyl group (C3’BzN), which are ca. 7.5 Å apart, with a short linker (4−6 atoms). 7-TES-14β-allyloxybaccatin III and (3R,4S)-1-(2-alkenylbenzoyl)-β-lactams were selected as key components and the Ojima-Holton coupling afforded the corresponding paclitaxel-dienes. The Ru-catalyzed ring-closing metathesis (RCM) of paclitaxel-dienes gave the designed 15- and 16-membered macrocyclic taxoids. However, the RCM reaction to form the designed 14-membered macrocyclic taxoid did not proceed as planned. Instead, the attempted RCM reaction led to the occurrence of an unprecedented novel Ru-catalyzed diene-coupling process, giving the corresponding 15-membered macrocyclic taxoid (SB-T-2054). The biological activities of the novel macrocyclic taxoids were evaluated by tumor cell growth inhibition (i.e., cytotoxicity) and tubulin-polymerization assays. Those assays revealed high sensitivity of cytotoxicity to subtle conformational changes. Among the novel macrocyclic taxoids evaluated, SB-T-2054 is the most active compound, which possesses virtually the same potency as that of paclitaxel. The result may also indicate that SB-T-2054 structure is an excellent mimic of the bioactive conformation of paclitaxel. Computational analysis for the observed structure-activity relationships is also performed and discussed.

Introduction

Cancer has become the leading cause of death for people under the age of 85 in the US.^1,2 Paclitaxel is currently one of the most widely used drugs in the fight against cancer among a variety of chemotherapeutic agents.^3,4 Paclitaxel binds to the β-tubulin subunit, accelerates the polymerization of tubulin and stabilizes the resultant microtubules, which leads to apoptosis through cell-signaling cascade.^5,6 Investigation into the bioactive conformation of paclitaxel^7,8 would lead to the development of novel drugs with much simpler structures and same or even higher activity.

The polar and nonpolar conformations of paclitaxel were first proposed based on the structural analysis in solution and in the solid state using NMR and X-ray crystallography.^9-13 Organic and medicinal chemists designed a series of structurally restrained taxoids to mimic these two conformations, but all of the synthesized taxoids were substantially less active than paclitaxel.^7,14-16 The structural biology study of paclitaxel did not start until the first cryo-electron microscopy (cryo-EM) structure of paclitaxel-bound Zn²⁺-stabilized α,β-tubulin dimer was reported in 1998 at 3.7 Å resolution (1TUB structure),¹⁷ which used the docetaxel crystal structure for display. The 1TUB structure was later refined to 3.5 Å resolution with a paclitaxel molecule (1JFF structure) in 2001.¹⁸ However, the resolution of these cryo-EM structures was not high enough to solve the binding conformation of paclitaxel. Thus, a computational study of the electron-density map was performed, which led to the proposal of the “T-Taxol” conformation.¹⁹ To prove the validity of the “T-Taxol” structure, rigidified paclitaxel congeners were designed, synthesized and assayed for their tubulin polymerization ability and cytotoxicity.^7,20-23 Among those “T-Taxol” mimcs, C4-C3’–linked macrocyclic taxoids showed higher activities than paclitaxel in the cytotoxicity and tubulin-polymerization assays.^20,22

In the mean time, two ¹³C-¹⁹F intramolecular distances of the microtubule-bound 2-(4-fluorobenzoyl)paclitaxel were experimentally obtained by means of the REDOR NMR study in 2000.²⁴ On the basis of the REDOR distances, MD analysis of paclitaxel conformers, photoaffinity labeling²⁵ and molecular modeling studies using the 1TUB coordinate¹⁷, we proposed the “REDOR-Taxol” as a valid microtubule-bound paclitaxel structure in 2005.⁸ We have further refined the “REDOR-Taxol” using the 1JFF coordinate and found that the “REDOR-Taxol” is not only fully consistent with the new REDOR-experiments,²⁶ but also accommodates highly active macrocyclic paclitaxel analogs designed based on the “T-Taxol” structure (see Supplementary Information). The critical difference between these two structures is the orientation of the C2’-OH group. In the “REDOR-Taxol”, the C2’-OH group interacts with His²²⁷ as the hydrogen bond donor,⁸ while the H-bonding is between the C2’-OH and the backbone carbonyl oxygen of Arg³⁶⁹ in the “T-Taxol”.¹⁹ Accordingly, we have designed novel macrocyclic taxoids by linking the C14 and C3’BzN groups, which mimic the “REDOR-Taxol” structure to examine whether these conformationally restricted taxoids exhibit the same level of biological activity as that of paclitaxel.

Results and Discussion

Design of novel C14-C3’BzN–linked macrocyclic taxoids

Computational analysis of the “REDOR-Taxol” revealed that this structure could be rigidified by connecting the the C14 position of the baccatin moiety and the ortho position of C3’N-benzoyl group (C3’BzN), which are ca. 7.5 Å apart, with a short linker (4−6 atoms) (Figure 1). Also, the same analysis indicated that the use of such linkers would not interfere with the amino acid residues in the paclitaxel binding site. Since the naturally occurring 14β-OH-10-deacetylbaccatin (14-OH-DAB) is readily available to us and we have performed substantial SAR studies on the taxoids derived from this material, we chose 14-OH-DAB as a key starting material for the synthesis of novel macrocyclic taxoids. Thus, the C14 position of the designed novel taxoids was fixed to β-ether functionality. The structures of the designed macrocyclic taxoids are shown in Figure 2. As the overlays in Figure 3 illustrates, the designed macrocyclic taxoids 1a and 1c appear to mimic the REDOR-Taxol structure very well, while the C3’N-benzoyl group of 1b deviates from the rest.

FIGURE 1.

REDOR-Taxol-1JFF and the atom-atom distance between C14 and the ortho carbon of the C3’N-benzoyl group.

FIGURE 2.

Designed novel C14-C3’BzN–linked macrocyclic taxoids

FIGURE 3.

Overlays of REDOR-Taxol (green) with 1a (cyan), 1b (red) and 1c (pink).

In addition, we included an C3’-(2-methylpropen-2-yl) analog of 1a, i.e., 1d, for comparison, since the replacement of the 3’-phenyl moiety with a 3’-(2-methylpropen-2-yl) group in paclitaxel and docetaxel produced the corresponding taxoids with higher potency.^27,28

Synthesis of macrocyclic taxoids

As the retrosynthetic analysis in Scheme 1 shows, the designed C14-C3’BzN–linked macrocyclic taxoids 1a-c can be synthesized through ring-closing metathesis (RCM)²⁹ of diene key intermediates 7a-c, which can be obtained from the corresponding (3R,4S)-1-(2-alkenylbenzoyl)-β-lactam 3a-c and 14β-allyloxy-baccatin 6 through the Ojima-Holton coupling.^{14,15,27,30,31} A similar strategy worked very well previously for the synthesis of the C2-C3’-linked^14,32 and C2-C3’N-linked³³ macrocyclic taxoids in our laboratory. Macrocyclic taxoid 1d can be synthesized just by using (3R,4S)-4-(2-methylpropen-2-yl)-β-lactam 3d in place of 3a-c.

SCHEME 1.

Retrosynthetic analysis of the designed macrocyclic taxoids 1a-c.

As Scheme 2 shows, (3R,4S)-β-lactams 3a-d were prepared in excellent yields through the acylation of enantiopure β-lactams 1a,b with the corresponding 2-alkenylbenzoyl chlorides 2a-c^34-36 in the presence of triethylamine and DMAP.7-TES-14β-allyloxy-baccatin 6 was prepared, following the method previously reported by us from 14-OH-DAB through selective acetylation of the C10-hydroxyl group³⁷ and subsequent TES protection of the C7-hydroxyl group (83% for two steps, Scheme 3).⁸ The Ojima-Holton coupling of β-lactams 3a-d with baccatin 6 was carried out under the standard conditions (LiHMDS in THF at −30 °C) to give the corresponding paclitaxel-dienes 7a-d bearing olefinic groups at the C14 position as well as the ortho position of the C3’BzN moiety (Scheme 3).

SCHME 2.

SCHEME 3.

As Scheme 4 shows, the RCM reactions of paclitaxel-dienes 7a and 7d catalyzed by the “1st-generation Grubbs catalyst”, Cl₂Ru(=CHPh)(PCy₃)₂, proceeded smoothly at room temperature to give the corresponding macrocyclic taxoids 8a and 8d, respectively, which were isolated in high yields by passing through a short silica gel column to remove the Ru catalyst. The crude 8a and 8d, thus obtained, were subjected to the deprotection of all silyl groups with HF-pyridine to give the designed macrocyclic taxoids 1a and 1d, respectively, in high yields. In both products, E-isomer was formed exclusively. We reported the synthesis of 1a (SB-T-2053) in 2005,⁸ but we needed to resynthesize this macrocyclic taxoid for additional cytotoxicity assay for comparison purpose. We confirmed that the reported synthesis is perfectly reproducible.

SCHEME 4.

The RCM reaction of 7c, proceeded slowly at room temperature. Thus, the reaction mixture was heated at reflux for 3 days until 7c had disappeared. The reaction gave E and Z isomers at the alkene moiety formed, i.e., 8c-E and 8c-Z, in 40% and 42% yields, respectively. The observed low reactivity of 7c and the lack of stereoselectivity in the alkene formation can be attributed to the larger ring size of the product(s). The silyl groups of 8c-E and 8c-Z were removed by HF-pyridine to give 1c-E and 1c-Z, respectively, in high yields (Scheme 5) .

SCHEME 5.

In contrast to the RCM reactions described above, the reaction of paclitaxel-diene 7b did not proceed as anticipated. The attempted RCM reaction of 7b was sluggish. Thus, the reaction was run in refluxing dichloromethane for 5 days. As often observed in RCM reactions, deactivation of the Ru catalyst appeared to have occurred. Accordingly, more Ru catalyst was added during the reaction (0.25 equiv. × 3). The reaction stopped at 75% conversion (i.e., 25% of the unreacted 7c was recovered), giving ring-closed product 8b in 50% yield. Then, the silyl groups were removed by HF-pyridine to give the corresponding macrocyclic taxoid 9b in high yield. However, the LCMS analysis of 9b showed its molecular weight higher than the anticipated RCM product 1b by 14 mass units, which was quite puzzling. The ¹H and ¹³C NMR and 2D NMR analyses suggested that 9b should possess a butenylene unit between the C14-O and the ortho position of the C3’N-benzoyl moiety and the newly formed double bond should be conjugated to the phenyl group. This proposed structure was confirmed by the X-ray crystallographic analysis of 9b (SB-T-2054) as shown in Figure 4. This unexpected reaction is shown in Scheme 6.

FIGURE 4.

X-ray crystal structure of 9b (SB-T-2054)

SCHEME 6.

A plausible mechanism for this unique reaction is proposed in Scheme 7. The first step of this process would be the formation of a Ru-carbene intermediate M-1 through a normal Ru-carbene exchange reaction. In the next step, however, a normal metalacyclobutane intermediate M-2 is not formed, which is very likely due to the macrocyclic ring strain for the formation of a 14-membered ring. Since M-2 is the precursor of the RCM product diTES-1b, the expected RCM reaction is blocked at this stage. Instead, regioisomeric metalacyclobutane intermediate M-3 is generated, forming a 15-membered ring. There are two possible pathways from M-3, i.e., (i) reductive elimination to form a cyclopropane 10 or (ii) ring-opening through β-hydride elimination to form a σ-allylic Ru intermediate M-5a and/or its regioisomer M-5b, which may well be equilibrated via π-allylic Ru intermediate M-5. Apparently, the first pathway did not happen since the formation of cyclopropane product 10 was not observed. Thus, the reaction should have proceeded through the possible pathway (ii). Since M-5a has a clear stabilization through double bond conjugation to the benzene ring, the reaction gives thermodynamically favorable product 8b through reductive elimination. To the best of our knowledge, this is the first example of a unique ring-closing olefin-olefin coupling reaction promoted by the Grubbs RCM catalyst.

SCHEME 7.

Plausible mechanism for the formation of 8d via novel ring-closing coupling

It should be noted that this Ru-carbene complex – promoted reaction does not regenerate the living Ru-carbene complex, different from RCM reactions, since the final step in this process is the reductive elimination of the alkene product 8b and Ru(II) species without a carbene bond. The fact that 75% conversion was attained by using 0.75 equiv. (total) of Cl₂Ru(=CHPh)(PCy₃)₂ may indicate that the reaction proceeded very well as it should be, although we did not recognize it at the time of the experiment.

Evaluation of Biological Activities

Cytotoxicity of macrocyclic taxoids against human breast, ovarian and colon cancer cell lines

Novel macrocyclic taxoids, thus obtained, were evaluated for their cytotoxicity against drug-sensitive and drug-resistant human breast cancer cell lines, drug-resistant human ovarian cancer cell line, as well as drug-sensitive human colon and overian cancer cell lines. Results are summarized in Table 1.

TABLE 1.

Cytotoxicity of C14-C3’BzN–linked macrocyclic taxoids (IC₅₀ nM)^a

Taxoids	MCF-7b	NCI/ADRc	LCC6-WTd	LCC6-MDRe	A2780f	HT-29g
Paclitaxel	1.85	395	2.45	110	36.1	7.28
1a (SB-T-2053)	12.3	592	12.2	300	114	29.2
9b (SB-T-2054)	3.49	183	2.09	129	31.0	17.0
1c-E	1,650	10,010	2,067	2,285	1,475	523
1c-Z	196	1,135	348	1,924	198	62.0
1d	398	1,000	130	618	---	---

^aConcentration of compound which inhibits 50% (IC₅₀, nM) of the growth of human tumor cell line after 72 h drug exposure.

^bMCF7: human breast carcinoma cell line.

^cNCI/ADR: multi-drug resistant human ovarian carcinoma cell line

^dLCC6-WT: human breast carcinoma cell line (Pgp−).

^eLCC6-MDR: mdrl transduced cell line (Pgp+).

^fA2780 human ovarian carcinoma cell line.

^gHT-29 human colon carcinoma cell line.

As Table 1 shows, 9b (SB-T-2054) is the most potent compound among the macrocyclic taxoids synthesized and evaluated. It is worthy of note that the potency of 9b is almost equivalent to that of paclitaxel against all 6 cell lines, i.e., 9b is more potent than paclitaxel against NCI/ADR, LCC6-WT, and A2780 cell lines, but less potent than paclitaxel against MCF7, LCC6-MDR and HT-29. The results may suggest that 9b is very closely mimicking paclitaxel's bioactive conformation. This is also confirmed by the tubulin polymerization assay described below. Macrocyclic taxoid 1a (SB-T-2053) is an olefin regioisomer of 9b, which is 2−3 times less potent than 9b. The results appear to indicate high sensitivity of the potency to the subtle difference in the position of C3’N-benzoyl group and the rigidity of the macrocyclic structure. Larger ring analogs 1c-E and 1c-Z exhibit substantial loss of potency. Also, there is a marked difference between the E isomer and the Z isomer, wherein the Z isomer (1c-Z) is 6−8 times more potent than the E isomer, except for LCC6-MDR cell line. The result further indicates the highly delicate effects of ring size and rigidity of the macrocyclic taxoids on their potency. Macrocyclic taxoid 1d, which is a 3’-(2-methylpropen-2-yl)-analog of 1a, shows 2−30 times less potency than 1a. This result is a surprise for us since the replacement of the 3’-phenyl moiety of paclitaxel and docetaxel with a 2-methylpropen-2-yl group has been shown to increase the potency.^27,28 Thus, the result may suggest that the 2^nd-generation taxoids bearing 3’-(2-methylpropen-2-yl) moiety have a slightly different binding site from that of paclitaxel or these taxoids have a different bioactive conformation that that of paclitaxel.

Microtubule polymerization assay

The activity of 9b (SB-T-2054) was evaluated in in vitro tubulin polymerization assay. Paclitaxel was used as the standard for comparison. Changes in absorbance in this spectrophotometric assay provide the direct measure of turbidity, hence indicating the extent of tubulin polymerization. As Figure 5 shows, 9b induced tubulin polymerization in the absence of GTP in a manner similar to paclitaxel. The microtubules formed with 9b as well as paclitaxel were stable against Ca²⁺-induced depolymerization. We have already reported that 1a (SB-T-2053) also exhibited virtually the same or slightly better activity than that of paclitaxel in the same assay.⁸ This result also support our observation that 9b is closely mimicking paclitaxel's bioactive conformation, as mentioned above.

FIGURE 5.

Tubulin polymerization with 9b and paclitaxel: microtubule protein 1mg/mL, 37 °C, GTP 1 mM, Drug 10 μM

Electron microscopy analysis

The microtubules formed with 9b were analyzed further by electron microscopy for their morphology and structure in comparison with those formed by using GTP and paclitaxel. As Figure 6 shows, the microtubules formed with 9b and paclitaxel are very similar, while those formed with GTP are longer and more uniformed.

FIGURE 6.

Electromicrographs of microtubule: (a) GTP; (b) paclitaxel; (c) 9b (SB-T-2054)

Molecular Modeling Analysis

The microtubule-bound structure of 9b (SB-T-2054) was energy minimized in the 1JFF coordinate and overlaid with the “REDOR-Taxol” structure in the 1JFF β-tubulin. The overlay of 9b with the “REDOR-Taxol” is shown in Figure 7. The H-bond between the C2’-OH of 9b and His²²⁷ is 2.7 Ǻ.

FIGURE 7.

Overlays of REDOR-Taxol (green) with 9b (SB-T-2054) (purple).

Figure 8 shows the overlay of 9b with 1a (SB-T-2053). Those two structures are very close, but there is a slight difference in the position of the C3’N-benzoyl group as well as the rigidity of the macrocyclic structure, i.e., 9b is more rigid than 1a, due to the conjugation of the ethenyl group to the benzoyl group. These subtle differences appear to be correlated to the 2−3 time difference in their potency,

FIGURE 8.

Overlay of 9b (purple) and 1a (SB-T-2053) (cyan).

Figure 9 shows the overlay of 1c-E, 1c-Z and the “REDOR-Taxol”. As the cytotoxicity assay revealed (see Table 1), there is 6−8 times difference in the potency of these two stereoisomers, i.e., 1c-Z is substantially more potent than 1c-E. It is clear that 1c-Z has a C3’N-benzoyl group orientation closer to that of 1a and 9b as compared to that of 1c-E, and 1c-Z is more rigid than 1c-E. Monte Carlo conformational analysis of 1c-E and 1c-Z in a simulated aqueous environment supports the substantial difference in the rigidity of these two stereoisomes (see Supporting Information).

FIGURE 9.

Overlay of REDOR-Taxol (green) with 1c-E (pink) and 1c-Z (yellow)

We also performed the Monte Carlo conformational analysis of 9b and 1c under the same conditions. We started from their “REDOR-Taxol” conformation as well as “T-Taxol” conformation, but both converged to the same distribution of conformers (see Supporting Information). The results indicate that the X-ray crystal structure of 9b (see Figure 5) is just one of the preferred conformations in a crystalline state. Also, the dihedral angle of C13-C1’-C2’-O2’ of 9b does not match those of the “REDOR-Taxol” and the “T-Taxol”, which may suggest that both proposed bioactive conformations of paclitaxel need further refinements.

In summary, novel C14-C3’BzN–linked macrocyclic taxoids, 1a (SB-T-2053), 1c-E, 1c-Z, 1d and 9b (SB-T-2054), were synthesized based on the computational design, mimicking the “REDOR-Taxol” structure. The Ru-catalyzed RCM reaction was employed in the key macrocyclization step of the paclitaxel-dienes 7a-d. The formation of 15- and 16-membered macrocyclic taxoids (1a, 1c and 1d) proceeded as anticipated, while that of 14-membered macrocyclic taxoid 1b did not go through. However, a novel Ru-catalyzed intramolecular coupling of two ethenyl moieties was discovered instead, which gave 15-membered macrocyclic taxoid 9b (SB-T-2054). This new process is likely to involve a metalacyclobutane intermediate, regioisomeric to the RCM intermediate, followed by β-hydride elimination and reductive elimination. Evaluation of the biological activities of these novel macrocyclic taxoids has revealed that their potency is highly sensitive to the subtle conformational differences as well as the rigidity of the compounds. Computational analysis for the structure-activity relationships of these novel macrocyclic taxoids is consistent with this observation. Macrocyclic taxoid 9b is found to be the most potent compound among those synthesized and assayed. It is noteworthy that 9b exhibits virtually the same potency as that of paclitaxel in the cytotoxicity and tubulin-polymerization assays as well as morphology analysis of microtubules by electron microscopy, which may indicate that 9b structure almost perfectly mimics the bioactive conformation of paclitaxel.

Experimental Section

General Methods

¹H and ¹³C NMR spectra were measured on a Varian 300, 400, 500 or 600 MHz NMR spectrometer or a Bruker AC-250 NMR spectrometer. Melting points were measured on a Thomas Hoover Capillary melting point apparatus and are uncorrected. Specific optical rotations were measured on a Perkin-Elmer Model 241 polarimeter. IR spectra were recorded on a Perkin-Elmer Model 1600 FT-IR spectrophotometer. TLC was performed on Merck DC-alufolien with Kieselgel 60F-254 and column chromatography was carried out on silica gel 60 (Merck; 230−400 mesh ASTM). Purity was determined with a Waters HPLC assembly consisting of dual Waters 515 HPLC pumps, a PC workstation running Millennium 32, and a Waters 996 PDA detector, using a Phenomenex Curosil-B column, employing CH₃CN/water (2/3) as the solvent system with a flow rate of 1 mL/min. High-resolution mass spectra were obtained at the Mass Spectrometry Laboratory, University of Illinois at Urbana-Champaign, Urbana, IL or the Mass Spectrometry Facility, University of California, Riverside Riverside, CA.

Materials

The chemicals were purchased from Aldrich-Sigma Co. and used as received or purified before use by standard methods. Tetrahydrofuran was freshly distilled from sodium metal and benzophenone. Dichloromethane was also distilled immediately prior to use under nitrogen from calcium hydride. N,N-Dimethylformamide (DMF) was distilled over 4A molecular sieves under reduced pressure. 4-(N,N-Dimethylamino)pyridine (DMAP) was uses as received. 14β-Hydroxyl-deacetylbaccatin III (DAB, 4) was obtained from Indena, SpA, Italy. (3R,4S)-3-triethylsiloxy-4-phenylazetidin-2-one (1a)¹⁵ and (3R,4S)-3-tert-butyldimethylsiloxy-4-(2-methylpropen-2-yl)azetidin-2-one (1b)²⁷ were prepared by the methods we reported. (3R,4S)-1-(2-Allylbenzoyl)-3-triethylsiloxy-4-phenylazetidin-2-one (3a), 7-triethylsilyl-14β-hydroxybaccatin III (5), 14β-allyloxy-3’N-debenzoyl-3’N-(2-allylbenzoyl)-7,2’-triethylsilylpaclitaxel (7a), and macrocyclic taxoid 1a (SB-T-2053) were synthesized using our methods previously reported.⁸

1-(2-Alkenylbenzoyl)-β-lactams (3a-d)

Procedure for the synthesis of β-lactam 3d is described as a typical example. To a solution of (3R,4S)-3-triisopropylsiloxy-4-phenylazetidin-2-one (1b) (106 mg, 0.365 mmol), triethylamine (0.2 mL, 1.46 mmol) and DMAP (8 mg, 0.073 mmol) in CH₂Cl₂ (2 mL), was added 2-allylbenzoyl chloride (2a) (2 equiv.) in a solution of CH₂Cl₂ (2 mL) dropwise at room temperature. The mixture was stirred overnight at room temperature and the reaction was quenched with saturated aqueous NH₄Cl solution (2 mL). The mixture was then extracted with CH₂Cl₂ (10 mL × 3). The combined extracts were dried over anhydrous MgSO₄ and concentrated in vacuo. The crude product was purified on a silica gel column using hexanes:EtOAc (9/1) as the eluent to afford (3R,4S)-1-(2-allylbenzoyl)-3-triisopropylsiloxy-4-(2-methylprop-1-enyl)azetidin-2-one (3d) as a colorless oil (126 mg, 80% yield): ¹H NMR (300 MHz, CDCl₃) δ 0.99 (m, 21 H), 1.81 (br s, 6 H), 3.42−3.61 (m, 2 H), 4.98−5.04 (m, 4 H), 5.36 (d, J = 8.2 Hz, 1 H), 5.87−6.01 (m, 1 H), 7.27 (m, 2 H), 7.37 (m, 1 H), 7.50 (m, 1 H); ¹³C NMR (75.0 MHz, CDCl₃) δ 12.2, 17.4, 18.3, 26.0, 37.3, 56.3, 116.0, 117.7, 125.8, 128.5, 129.9, 131.4, 133.6, 136.8, 138.4, 140.4, 165.6, 166.7. HRMS (EI) m/z calcd. for C₂₆H₃₉NO₃SiH⁺ : 442.2777, found: 442.2776 (Δ = 0.1 ppm).

In a similar manner, β-lactams 3a-c were prepared from (3R,4S)-3-triethylsiloxy-4-phenylazetidin-2-one (1a) (For 3a, see Materials, vide supra).

(3R,4S)-1-(2-Ethenylbenzoyl)-3-triethylsiloxy-4-phenylazetidin-2-one (3b)

Colorless oil; ¹H NMR (600 MHz, CDCl₃) δ 0.44 (m, 6 H), 0.77 (m, 9 H), 5.15 (d, J = 6.6 Hz, 1 H), 5.33 (m, 2 H), 5.69 (d, J = 17.4 Hz, 1 H), 6.94 (dd, J = 17.4, 10.8 Hz, 1 H), 7.33 (m, 6 H), 7.47 (m, 1 H), 7.52 (m, 1 H), 7.59 (d, J = 6.8 Hz, 1 H); ¹³C NMR (100.5 MHz, CDCl₃) δ 4.3, 6.16, 61.4, 76.9, 117.3, 126.1, 127.2, 128.1, 128.1, 128.3, 128.5, 131.3, 132.1, 133.5, 133.8, 136.9, 165.3, 166.4. HRMS (FAB/DCM/NaCl) m/z calcd. for C₂₄H₂₉NO₃SiH⁺ : 408.1995 found: 408.1995 (Δ = 0 ppm).

(3R,4S)-1-(2-Allyloxybenzoyl)-3-triethylsiloxy-4-phenylazetidin-2-one (3c)

Colorless oil; ¹H NMR (300 MHz, CDCl₃) δ 0.40−0.50 (m, 6 H), 0.75−0.81 (m, 9 H), 4.62 (dd, J = 5.7, 1.2 Hz, 2 H), 5.10 (d, J = 6.0 Hz, 1 H), 5.22 (dd, J = 10.5, 0.9 Hz, 1 H), 5.36 (d, J = 6.0 Hz, 1 H), 5.45 (dd, J =17.4, 1.8 Hz, 1 H), 6.04−6.20 (m, 1 H), 6.96−7.03 (m, 2 H), 7.32−7.45 (m, 7 H); ¹³C NMR (75 Hz, CDCl₃) δ 4.34, 6.15, 61.5, 69.7, 70.1, 112.2, 118.3, 120.5, 124.2, 127.8, 127.9, 128.0, 129.4, 132.8, 132.9, 133.6, 156.9, 164.6, 165.0; HRMS (FAB/DCM/NaCl) m/z calcd for C₂₅H₃₁NO₄SiH⁺: 438.2100, found: 438.2088 (Δ = −2.7 ppm).

14β-Allyloxy-7-triethylsilylbaccatin III (6)

To a solution of 5 (427 mg, 0.596 mmol) in DMF (20 mL) was added 1.0 M sodium bis(trimethylsilyl)amide (NaHMDS) in THF (0.656 mL, 0.656 mmol) at −40 °C and the reaction mixture was stirred for 5 min and then allyl iodide (0.060 mL, 0.656 mmol) was added. The reaction was quenched after 10 min with saturated aqueous NH₄Cl (5 mL). The reaction mixture was extracted with ethyl acetate (100 mL), washed with water (10 mL × 3) and brine (10 mL). The organic layer was dried over anhydrous MgSO₄, filtered and the solvent was removed under reduced pressure. The residue was purified on a silica gel column using hexanes:EtOAc (6/1) as the eluant to afford 6 as a white solid (361 mg, 80%): mp 133−135 °C; ¹H NMR (300 MHz, CDCl₃) δ 0.55 (m, 6 H), 0.90 (m, 9 H), 1.19 (s, 3 H), 1.60 (m, 2 H), 1.66 (s, 3 H), 1.84 (m, 1 H), 2.00 (m, 3 H), 2.12 (s, 3 H), 2.29 (s, 3 H), 2.50 (m, 1 H), 3.22 (d, J = 4.2 Hz, 1 H), 3.69 (d, J = 5.7 Hz, 1 H), 3.77 (m, 2 H), 4.06 (m, 1 H), 4.18 (m, 1 H), 4.29−4.47 (m, 3 H), 4.69 (bs, 1 H), 4.87−4.98 (m, 3 H), 5.61−5.72 (m, 1 H), 5.83 (d, J = 6.9 Hz, 1 H), 7.42 (m, 2 H), 7.55 (m, 1 H), 8.06 (m, 2 H); ¹³C NMR (62.9 MHz, CDCl₃) δ 5.1, 5.4, 6.6, 6.7, 9.7, 10.0, 14.7, 20.8, 21.4, 22.4, 26.2, 37.1, 42.7, 46.4, 58.7, 72.1, 73.0, 74.5, 75.6, 75.7, 76.2, 76.4, 80.9, 81.5, 82.0, 84.0, 117.4, 128.4, 129.7, 129.8, 133.3, 133.6, 133.7, 141.2, 165.6, 169.3, 170.1, 201.9. HRMS (FAB/DCM/NaCl) m/z calcd. for C₄₀H₅₆O₁₂SiH⁺: 757.3619, found: 757.3643 (Δ = −3.1 ppm).

7,2’-Disilyl-paclitaxel-diene analogs (7a-d), substrates for RCM reactions

Procedure for the synthesis of 7d is described as a typical example (For 7a, see Materials, vide supra). To a solution of 6 (100 mg, 0.13 mmol) and β-lactam 3d (3 equiv.) in THF (15 mL) was added 1 M LiHMDS in THF (0.2 ml, 0.20 mmol) at −40 °C. The reaction mixture was warmed up to −20 °C over 1 h and then quench with saturated aqueous NH₄Cl solution (10 mL) and extracted with EtOAc (30 mL × 3). The organic layers were combined and solvent was removed by reduced pressure vacuum. The residue was purified by using chromatography column on silica gel using hexanes/EtOAc (9/1) to afford 7-triethylsilyl-14β-allyloxy-2’-triisopropyl-3’-dephenyl-3’-(2-methylprop-1-enyl)-3’N-debenzoyl-3’N-(2-allylbenzoyl)paclitaxel (7d) as a white solid (115 mg, 73%): ¹H NMR (300 MHz, CDCl₃) δ 0.60−0.65 (m, 6 H), 0.93−1.00 (m, 9 H), 1.18 (s, 21 H), 1.32 (s, 3 H), 1.77 (s, 3 H), 1.83 (s, 3 H), 1.91−2.03 (m, 3 H), 1.92 (s, 3 H), 2.05 (s, 3 H), 2.20 (s, 3 H), 2.43 (s, 3 H), 2.50−2.54 (m, 1 H), 3.45−3.61 (m, 2 H), 3.74 (s, 1 H), 3.84 (m, 2 H), 3.92−3.98 (m, 1 H), 4.30−4.32 (d, J = 8.4 Hz, 1 H), 4.43−4.58 (m, 3 H), 4.69 (s, 1 H), 4.79 (d, J = 9.9 Hz, 1 H), 4.94−5.09 (m, 3 H), 5.14−5.20 (m, 1 H), 5.45 (d, J = 9.0 Hz, 1 H), 5.67−5.81 (m, 1 H), 5.91−6.11 (m, 3 H), 6.48−6.51 (m, 2 H), 7.22−7.26 (m, 2 H), 7.30−7.54 (m, 5), 8.14−8.16 (m, 2 H); ¹³C NMR (62.9 MHz, CDCl₃) δ 5.3, 6.7, 9.8, 12.6, 14.5, 17.9, 18.1, 18.7, 20.1, 22.7, 25.7, 26.1, 37.2, 43.3, 46.0, 50.8, 58.7, 72.2, 72.5, 74.5, 75.0, 75.1, 76.4, 78.7, 78.9, 81.6, 84.1, 115.8, 118.2, 122.2, 126.2, 127.3, 128.5, 129.6, 129.9, 130.1, 133.0, 133.4, 135.4, 136.0, 136.2, 136.4, 137.4, 137.9, 165.5, 168.5, 169.3, 171.3, 172.0, 201.1. HRMS calcd. for C₆₆H₉₅NO₁₅Si₂H⁺: 1198.6318, found 1198.6320 (Δ= −0.1 ppm).

14β-Allyloxy-3’N-debenzoyl-3’N-(2-ethenylbenzoyl)-7,2’-triethylsilylpaclitaxel (7b)

White solid; mp 99−102 °C; [α]_D²⁰ −42 (c 0.6, CHCl₃); ¹H NMR (600 MHz, CDCl₃) δ 0.34−0.47 (m, 6 H), 0.55−0.63 (m, 6 H), 0.78 (m, 9 H), 0.93 (m, 9 H), 1.15 (s, 3 H), 1.31 (s, 3 H), 1.74 (s, 3 H), 1.95 (m, 1 H), 2.03 (s, 3 H), 2.19 (s, 3 H), 2.56 (m, 1 H), 2.60 (s, 3 H), 3.67 (s, 1 H), 3.82 (d, J = 7.2 Hz, 1 H), 3.84 (d, J = 8.0 Hz, 1 H), 3.93 (dd, J = 10.8, 4.2 Hz, 1 H), 4.28 (d, J = 9.0 Hz, 1 H), 4.32 (dd, J = 10.5, 4.0 Hz, 1 H), 4.47 (m, 3 H), 4.62 (d, J = 10.5 Hz, 1 H), 4.71 (d, J = 1.2 Hz, 1 H), 5.00 (d, J = 9.0 Hz, 1 H), 5.32 (d, J = 7.6 Hz, 1 H), 5.36 (m, 1 H), 5.61 (d, J = 8.4 Hz, 1 H), 5.70 (d, J = 11.4 Hz, 1 H), 5.99 (d, J = 7.8 Hz, 1 H), 6.21 (d, J = 7.0 Hz, 1 H), 6.47 (s, 1 H), 6.99 (m, 2 H), 7.26−7.48 (m, 8 H), 7.50−7.58 (m, 3 H), 8.10 (d, J = 8.4 Hz, 2 H); ¹³C NMR (75.5 MHz, CDCl₃) δ 4.3, 5.3, 6.4, 6.7, 9.9, 14.4, 20.8, 23.3, 26.1, 37.3, 43.5, 45.9, 56.1, 58.8, 72.3, 72.5, 74.8, 74.9, 76.5, 76.6, 78.6, 78.6, 81.7, 84.1, 117.2, 117.8, 126.3, 126.5, 127.7, 128.0, 128.7, 129.5, 129.9, 130.6, 132.7, 133.5, 134.2, 134.8, 135.6, 136.2, 136.5, 138.6, 165.6, 168.3, 169.4, 169.7, 171.4, 201.0. HRMS (FAB/DCM/NaCl) m/z calcd. for C₆₄H₈₅NO₁₅Si₂H⁺: 1164.5536 found: 1164.5535 (Δ = 0.1 ppm).

14β-Allyloxy-3’N-debenzoyl-3’N-(2-allyloxybenzoyl)-7,2’-triethylsilylpaclitaxel (7c)

White solid; mp 205−207 °C; ¹H NMR (400 MHz, CDCl₃) δ 0.28−0.48 (m, 6 H), 0.52−0.66 (m, 6 H), 0.76−0.86 (m, 9 H), 0.88−0.98 (m, 9 H), 1.11 (s, 3 H), 1.25 (s, 3 H), 1.73 (s, 3 H), 1.88−1.98 (m, 1 H), 2.01 (s, 3 H), 2.17 (s, 3 H), 2.50−2.60 (m, 1 H), 2.66 (s, 3 H), 3.74 (s, 1 H), 3.76−3.88 (m, 3 H), 4.20 (dd, J = 10.8, 4.4 Hz, 1 H), 4.28 (d, J = 8.4 Hz, 1 H), 4.42−4.55 (m, 4 H), 4.70 (d, J = 1.6 Hz, 1 H), 4.74−4.83 (dd, J = 13.2, 5.2 Hz, 1 H), 4.83−4.89 (dd, J = 12.8, 5.6 Hz, 1 H), 5.0 (d, J = 8.4 Hz, 1 H), 5.20−5.30 (m, 1 H), 5.34 (d, J = 10.4 Hz, 1 H), 5.44 (d, J = 16.4 Hz, 1 H), 5.64 (d, J = 6.4 Hz, 1 H), 5.96 (d, J = 7.2 Hz, 1 H), 6.16−6.30 (m, 2 H), 6.45 (s, 1 H), 6.92−7.0 (m, 2 H), 7.26−7.42 (m, 6 H), 7.48 (t, J = 7.2 Hz, 2 H), 7.59 (t, J = 7.6 Hz, 1 H), 8.02 (dd, J = 1.6, 8 Hz, 1 H), 8.10 (d, J = 7.2 Hz, 2 H), 9.09 (d, J = 7.6 Hz, 1 H); 13C NMR (100 MHz, CDCl₃) δ 4.3, 5.3, 6,5, 6.7, 9.90, 14.3, 20.9, 22.4, 23.4, 26.0, 29.7, 37.3, 43.4, 45.9, 56.6, 58.8, 70.3, 72.2, 72.7, 74.7, 74.9, 75.1, 76.6, 78.2, 78.5, 81.6, 84.2, 113.0, 117.2, 118.6, 121.1, 126.6, 127.6, 128.5, 128.6, 129.6, 129.9, 132.5, 132.8, 132.9, 133.4, 135.4, 136.4, 139.0, 157.1, 164.6, 165.6, 169.4, 170.0, 171.2, 201.0; HRMS (FAB/DCM/NaCl) m/z calcd for C₆₅H₈₇NO₁₆Si₂H⁺: 1194.5462, found: 1194.5509 (Δ = 3.9 ppm).

Macrocyclic taxoid 1c (E and Z)

To a solution of 7c (108 mg, 0.09 mmol) in dry CH₂Cl₂ (25 mL) was added Cl₂Ru(=CHPh)(PCy₃)₂ (17 mg, 0.022 mmol) in dry CH₂Cl₂ (1 mL) under a nitrogen atmosphere. The reaction was stirred for 3 days at reflux to convert all of the starting materials. Solvent was removed in vacuo. Two products were separated by flash chromatography on silica gel (hexanes/EtOAc = 4/1 ∼ 3/1) to afford 8c-E as a crude yellow solid (42 mg, 40%) and 8c-Z as a crude yellow solid (44 mg, 42%).

To a solution of 8c-E (35 mg) in CH₃CN (0.7 mL) and pyridine (0.7 mL) was added HF-pyridine (70:30, 0.35 mL) at 0 °C under nitrogen, and the mixture was stirred overnight. Ethyl acetate (60 mL) was added to the reaction mixture and the resulting solution was washed with saturated aqueous NaHCO₃ (10 mL × 2), aqueous CuSO₄ (8 mL × 3), water (10 mL × 3), and brine (5 mL). The organic layer was dried over anhydrous MgSO₄ and solvent was removed in vacuo. Flash chromatography of the residue on silica gel (hexanes/EtOAc = 1/2) afforded 1c-E as a white solid (24 mg, 85 %). In the same manner, 8c-Z (48 mg) was deprotected to give 1c-Z as a white solid (35 mg, 91%).

Macrocyclic taxoid 1c-E

White solid; mp 194−196 °C; [α]_D²⁰ −107 (c 12.7, CHCl₃); ¹H NMR (400 MHz, CDCl₃) δ 1.08 (s, 3 H), 1.12 (s, 3 H), 1.60 (s, 3 H), 1.64 (s, 3 H), 1.78−1.86 (m, 1 H), 2.13 (s, 3 H), 2.34 (s, 1 H), 2.36 (s, 3 H), 2.44−2.56 (m, 1 H), 3.27 (s, 1 H), 3.69−3.77 (m, 3 H), 4.08−4.19 (m, 3 H), 4.30−4.33 (m, 3 H), 4.60−4.68 (m, 2 H), 4.92 (d, J = 8.4 Hz, 1 H), 5.57 (d, J = 15.6 Hz, 1 H), 5.82 (dd, J = 3.2, 9.2 Hz, 1 H), 5.86 (d, J = 7.2 Hz, 1 H), 5.94 (m, 1 H), 6.16 (s, 1 H), 6.25 (d, J = 6.0 Hz, 1 H), 6.89 (d, J = 8.4 Hz, 1 H), 7.04 (t, J = 7.6 Hz, 1 H), 7.17−7.24 (m, 3 H), 7.40 (t, J = 7.6 Hz, 3 H), 7.47 (d, J = 7.2 Hz, 2 H), 7.53 (t, J = 7.2 Hz, 1 H), 7.92 (dd, J = 7.6, 1.2 Hz, 1 H), 8.02 (d, J = 7.6 Hz, 2 H), 8.52 (d, J = 9.2 Hz, 1 H); ¹³C NMR (100 MHz, CDCl₃) δ 9.4, 15.3, 20.7, 22.2, 22.5, 26.2, 29.7, 35.6, 43.2, 45.1, 54.8, 58.9, 69.1, 72.1, 72.9, 74.5, 75.3, 76.3, 77.1, 81.5, 83.5, 84.3, 112.3, 121.5, 122.7, 123.9, 127.5, 127.6, 128.1, 128.6, 129.6, 129.9, 131.7, 132.0, 132.9, 133.6, 134.4, 136.9, 138.7, 156.4, 165.2, 165.7, 170.3, 171.0, 202.9; HRMS m/z calcd for C₅₁H₅₅NO₁₆H⁺, 938.3599; found, 938.3590 (Δ = −0.9 ppm).

Macrocyclic taxoid 1c-Z

White solid; mp 166−168 °C; [α]_D²⁰ −73 (c 3.7, CHCl₃); ¹H NMR (500 MHz, CDCl₃) δ 1.22 (s, 3 H), 1.24 (s, 3 H), 1.69 (s, 3 H), 1.84 (s, 3 H), 1.87 (m, 1 H), 2.04 (s, 3 H), 2.24 (s, 3 H), 2.40 (d, J = 4.0 Hz, 1 H), 2.50−2.51 (m, 1 H), 2.97 (d, J = 8.5 Hz, 1 H), 3.44 (s, 1 H), 3.74 (d, J = 7.5 Hz, 1 H), 3.85 (d, J = 8.0 Hz, 1 H), 4.14 (d, J = 11.0 Hz, 1 H), 4.19 (d, J = 8.0 Hz, 1 H), 4.20 (d, J = 8.5 Hz, 1 H), 4.34−4.44 (m, 3 H), 4.60 (dd, J = 12.5, 8.5 Hz, 1 H), 4.91 (d, J = 9.0 Hz, 1 H), 5.03 (dd, J = 8.5, 3.5 Hz, 1 H), 5.49 (dd, J = 5.5, 4.0 Hz, 1 H), 5.70−5.77 (m, 1 H), 5.82 (t, J = 9.0 Hz, 1 H), 5.92 (d, J = 7.5 Hz, 1 H), 6.20 (d, J = 6.5 Hz, 1 H), 6.28 (s, 1 H), 7.00 (d, J = 8.0 Hz, 1 H), 7.24−7.26 (m, 1 H), 7.34−7.44 (m, 7 H), 7.60−7.66 (m, 2 H), 8.03 (d, J = 7.5 Hz, 2 H), 8.35 (dd, J = 7.5, 1.0 Hz, 1 H), 8.49 (d, J = 6.0 Hz, 1 H); ¹³C NMR (100.5 MHz, CDCl₃) δ 9.4, 14.7, 20.8, 22.2, 23.4, 26.2, 35.4, 43.4, 44.9, 58.6, 59.5, 63.3, 71.1, 71.9, 72.8, 72.9, 75.2, 75.9, 77.2, 78.0, 78.6, 78.8, 81.0, 84.3, 111.8, 122.2, 123.2, 128.1, 128.6, 128.7, 129.7, 129.9, 132.8, 133.5, 133.6, 134.6, 134.7, 135.4, 138.2, 156.3, 165.1, 165.6, 169.9, 171.1, 174.1, 202.9; HRMS m/z calcd for C₅₁H₅₅NO₁₆H⁺, 938.3599; found, 938.3589 (Δ = −1.0 ppm).

Macrocyclic taxoid 1d (SB-T-2061)

To a solution of 7d (45 mg, 0.04 mmol) in CH₂Cl₂ (20 mL) was added Cl₂Ru(=CHPh)(PCy₃)₂ (4 mg, 0.01 mmol) in CH₂Cl₂ (10 mL). The reaction was stirred overnight solvent was removed by reduced pressure vacuum. The residue was passed through a short silica gel column to remove the catalyst to afford 8d as a yellow solid (80%).

To a solution of 8d thus obtained in CH₃CN (1.0 mL) and pyridine (1.0 mL) was added HF-pyridine (70:30, 0.5 mL) and the reaction mixture was stirred overnight. The reaction mixture was quenched with EtOAc (100 mL) and washed with saturated aqueous NaHCO₃ solution (10 mL), CuSO₄ solution (10 mL × 3), water (10 mL × 3) and brine (3 mL). The organic layer was dried over anhydrous MgSO₄ and solvent was removed under reduced pressure. The residue was purified by a column chromatography on silica gel using hexanes:EtOAc (2:1) as the eluent to afford 1d as white solid (83%): mp 198−200 °C; ¹H NMR (300 MHz, CDCl₃) δ 1.19 (s, 3 H), 1.23 (s, 3 H), 1.72 (s, 3 H), 1.90 (s, 3 H), 1.91 (s, 3 H), 1.92 (s, 3 H), 2.27 (s, 3 H), 2.39 (s, 3 H), 3.23−3.28 (m, 2 H), 3.80−3.82 (m, 2 H), 3.88 (d, J = 5.1 Hz, 1 H), 3.98−4.05 (m, 2 H), 4.25 (d, J = 5.4 Hz, 1 H), 4.40−4.48 (m, 2 H), 4.54 (d, J = 3.6 Hz, 1 H), 5.02 (d, J = 5.4 Hz, 1 H), 5.19−5.31 (m, 2 H), 5.49 (d, J = 9.3 Hz, 1 H), 5.68−5.77 (dt, J = 15.0, 6.3 Hz, 1 H), 5.99 (d, J = 7.2 Hz, 1 H), 6.16 (dd, J = 7.8, 1.8 Hz, 1 H), 6.31−6.35 (m, 2 H), 7.18−7.21 (m, 1 H), 7.29−7.34 (m, 1 H), 7.38−7.55 (m, 4 H), 7.64−7.66 (m, 1 H), 8.09−8.13 (m, 2 H); ¹³C NMR (62.9 MHz, CDCl₃) δ 9.4, 14.9, 18.8, 20.8, 22.4, 22.9, 26.1, 35.5, 36.1, 43.4, 45.0, 49.8, 58.8, 72.0, 72.5, 74.5, 75.2, 77.1, 80.3, 81.5, 84.3, 118.5, 126.4, 126.8, 127.6, 128.6, 129.4, 130.0, 130.5, 131.1, 131.7, 133.7, 134.5, 135.7, 137.4, 138.3, 140.8, 165.2, 168.9, 170.0, 171.1, 171.8, 202.9. HRMS calcd. for C₄₉H₅₇NO₁₅H⁺: 900.3806, found 900.3805 (Δ= 0.2 ppm).

Macrocyclic taxoid 9b (SB-T-2054)

To a sulotion of 7b (43 mg, 0.036 mmol) in CH₂Cl₂ (22 mL) was added Cl₂Ru(=CHPh)(PCy₃)₂ (8 mg × 3, 0.027 mmol) in CH₂Cl₂ (0.08 mL). The reaction was refluxed for 5 days and solvent was removed by reduced pressure vacuum. The residue was passed through a short column (eluent: hexanes/EtOAc = 3/1) to remove the catalyst to afford 8b as a crude yellow solid (20 mg, 50%) with unreacted diene (7b, 11 mg, 25%) recovered.

To a solution of the 8b (20 mg) in CH₃CN (0.4 mL) and pyridine (0.4 mL) was added HF-pyridine (70:30, 0.2 ml) and the reaction mixture was stirred overnight. The reaction mixture was quenched with EtOAc (50 mL) and washed with saturated aqueous NaHCO₃ solution (10 mL × 2), CuSO4 solution (10 mL × 3), water (10 mL × 3) and brine (3 mL). The organic layer was dried over anhydrous MgSO₄ and solvent was removed under reduced pressure. The residue was purified by a column chromatography on silica gel using hexanes:EtOAc (1:1) as the eluent to afford 9b as a white solid (13 mg, 80%); mp 203−205 °C; [α]_D²² −28 (c 0.92, CHCl₃); ¹H NMR (600 MHz, CDCl₃) δ 1.19 (s, 3 H), 1.23 (s, 3 H), 1.72 (s, 3 H), 1.73 (s, 3 H), 1.91 (m, 1 H), 1.94 (m, 1 H), 2.23 (s, 3 H), 2.24 (m, 1 H), 2.57 (m, 1 H), 2.63 (s, 3 H), 3.35 (s, 1 H), 3.49 (t, J = 9.6 Hz, 1 H), 3.82 (m, 3 H), 4.26 (d, J = 8.4 Hz, 1 H), 4.38 (m, 1 H), 4.40 (d, J = 8.4 Hz, 1 H), 4.78 (m, 1 H), 5.00 (d, J = 6.8 Hz, 1 H), 5.62 (d, J = 7.8 Hz, 1 H), 5.92 (ddd, J = 16.8, 8.4, 2.8 Hz, 1 H), 6.02 (d, J = 7.2 Hz, 1 H), 6.31 (s, 1 H), 6.36 (d, J = 7.8 Hz, 1 H), 6.52 (d, J = 16.2 Hz, 1 H), 6.90 (d, J = 8.4 Hz, 1 H), 7.22 (d, J = 7.2 Hz, 1 H), 7.34−7.63 (m, 10 H), 8.10 (d, J = 7.8 Hz, 2 H); ¹³C NMR (75.5 MHz, CDCl₃) δ 9.1, 15.1, 20.8, 23.0, 23.3, 26.2, 29.7, 32.7, 35.7, 43.6, 44.9, 54.6, 58.9, 72.1, 72.7, 73.3, 75.2, 76.4, 77.2, 77.9, 79.0, 81.6, 84.3, 126.5, 126.7, 126.8, 128.3, 128.4, 128.8, 128.9, 129.0, 129.2, 129.9, 130.0, 131.1, 133.7, 134.9, 135.0, 136.6, 137.9, 138.4, 165.4, 169.4, 170.1, 171.1, 173.2, 202.9. HRMS (FAB/DCM/NaCl) m/z calcd. for C₅₁H₅₅NO₁₅H⁺: 922.3650 found: 922.3643 (Δ = −0.8 ppm).

X-ray single crystal analysis of 9b (SB-T-2054)

A colorless single crystal (0.2×0.3×0.5 mm³) of 9b grown in a NMR tube by slow diffusion of hexanes (0.5 mL) into a solution of 9b (5 mg) in a mixture of dichloromethane (0.1 mL) and ethyl actate (0.05 mL) was selected for X-Ray crystallographic analysis. Since the crystals obtained in this way were sensitive to air, the test single crystal was sealed along with some mother liquid in a capillary tube. Intensity data collection was carried out with a Bruker AXS SMART diffractometer equipped with a CCD detector using a Siemens graphite-monochromated Mo radiation tube (λ = 0.71073 Å) at room temperature. The unit cells were determined by a least-squares analysis using the SMART software package. The raw frame data were integrated into SHELX-format reflection files and corrected for Lorentz and polarization effects using SAINT. Corrections for incident and diffracted beam absorption effects were applied using SADABS. The structure was solved by direct method and refined by full-matrix least-squares method on F₂ using the SHELXTL 97 software package. All non-hydrogen atoms were refined anisotropically. Hydrogen atoms were placed in calculated positions and added as fixed contributions. There are solvent molecules trapped in the crystals. However, we were not able to fully solved the structure due to their highly disordered behavior. Ignorance of these solvent molecules did not terribly harm the final refinement, thus we simply treated this single crystal as a solvent free sample. The cif file is attached as Supporting Information.

In vitro cell growth inhibition assay

Tumor cell growth inhibition was determined according to the method established by Skehan et al.³⁸ Human cancer cells LCC6-WT (Pgp−), MCF-7 (Pgp−), LCC6-MDR (Pgp+), NCI/ADR (Pgp+), A2780 (Pgp−) and HT-29 (Pgp−) were plated at a density of 400−2,000 cells/well in 96-well plates and allowed to attach overnight. These cell lines were maintained in RPMI-1640 medium (Roswell Park Memorial Institute growth medium) supplemented with 5% fetal bovine serum and 5% Nu serum (Collaborative Biomedical Product, MA). Macrocyclic axoids were dissolved in DMSO and further diluted with RPMI-1640 medium. Triplicate wells were exposed to various treatments. After 72 h incubation, 100 μL of ice-cold 50% trichloroacetic acid (TCA) was added to each well, and the samples were incubated for 1 h at 4 °C. Plates were then washed five times with water to remove TCA and serum proteins, and 50 μL of 0.4% sulforhodamine B (SRB) was added to each well. Following a 5 min incubation, plates were rinsed five times with 0.1% acetic acid and air-dried. The dye (SRB) was then solubilized with 10 mM Tris base (pH 10.5) for 5 min on a gyratory shaker. Optical density was measured at 570 nm. The IC50 values were then calculated by fitting the concentration-effect curve data with the sigmoid-E_max model using nonlinear regression, weighted by the reciprocal of the square of the predicted effect. ³⁹

Tubulin polymerization assay

Assembly and disassembly of calf brain microtubule protein (MTP) was monitored spectrophotometrically (Beckman Coulter DU 640, Fullerton, CA) by recording changes in turbidity at 350 nm at 37 °C.^40,41 MTP was diluted to 1mg/mL in MES buffer containing 3 M glycerol. The concentration of tubulin in MTP is 85% and that is taken into consideration when the ratios of tubulin to drug are presented in Figure 6. Microtubule assembly was carried out with 10 μM 9b (SB-T-2054). Paclitaxel (10 μM) and GTP (1 mM) were also used for comparison purpose. Calcium chloride (6 mM) was added to the assembly reaction after 50 min to follow the calcium-induced microtubule depolymerization.

Electron microscopy

Aliquots (50 μL) were taken from in vitro polymerization assays at the end of the reaction and placed onto 300-mesh carbon-coated, formavar-treated copper grids. Samples were then stained with 20 μL of 2% uranyl acetate and viewed with a JEOL model 100CX electron microscope.

Computational methods

The structures of the C14-C3’BzN–linked macrocyclic taxoids, 1a, 1c-E, 1c-Z and 9b, in the 1JFF were produced by directly introducing the linker into the “REDOR-Taxol” in the 1JFF complexes using the Builder module in the InsightII 2000 program (CVFF). Then, these structures were energy-minimized in 5000 steps or till the maximum derivative being < 0.001 kcal/A by means of the conjugate gradients method using the CVFF force field and the distance-dependent dielectric. The backbone of the protein was fixed throughout the energy minimization. After the energy minimization, the snapshots were overlaid by superimposing the backbones of the proteins. During the energy minimizations of the macrocyclic taxoids, the C’2-OH--N(His²²⁷) H-bond was very stable in all cases. Monte Carlo conformational searches on the C14-C3’BzN–linked macrocyclic taxoids were performed with energy minimization (5000 MC steps, minimization for 1000 steps with Polak-Ribiere conjugated gradients) by Macromodel program using a generalized born with surface area term (GBSA) continuum solvent description of water solvation⁴².