|Home | About | Journals | Submit | Contact Us | Français|
A synthetic approach to palau’amine is described that exploits veiled symmetry in the structure. Bis-alkylidenes i have been prepared and found susceptible to halogenative desymmetrization using t-BuOCl. This oxidation forms the imbedded spirocyclopentane motif observed in the natural product. A host of atypical reactions and processes developed during these studies are discussed, as are plans for completing total syntheses of this compound class.
Palau’amine (1 - Figure 1)1 is a prominent member of the pyrrole/imidazole family of marine derived alkaloids - a group whose numbers continue to grow with research on selected pacific sponges. Relationships within the group have been discussed extensively, relating mainly to proposals for unified biosynthetic schemes.1b, 2 While direct evidence is lacking, an early consensus viewed the class as originating with 3-amino-1-(2-aminoimidazolyl)-prop-1-ene (2).3 One can trace this motif in scores of natural products - described throughout four decades of literature. In this scenario, aminoimidazole 2 and pyrrole-2-carboxylic acid are constituents of oroidin,4 itself the feedstock from which the larger family derives2a - perhaps wherein genes encoding enzymatic machinery manipulating intermediates have transferred and evolved across producing species. An alternate view stems from Al-Mourabit’s recent insightful work on autooxidative incorporation of guanidine into partially oxidized proline dimers.5 This direct access to displacamides from a non-oroidin source adds another dimension to biosynthesis considerations.
Origins notwithstanding, palau’amine and its relatives harbor two units of aminoimidazole 2. It is these dimeric conjugates that make up the most structurally and biochemically interesting members of the set. Palau’amine was isolated as an anti-fungal constituent of S. aurantium.1 The substance was also reported to possess anti-microbial activity and to inhibit stimulated human T-cell proliferation in vitro, while being relatively innocuous to resting lymphocytes. To date these functions have not been examined further or characterized at a molecular level. Actually, the same is true of the various biological effects attributed to other palau’amine type bis-guanidines. The stalemate derives from a scarcity of the materials in nature and the fact that assembling related structures in the laboratory has proven an unusually difficult problem.
The imbedded cyclopentane scaffolding the two guanidine units within 1 is its signature feature. Installing this motif becomes de facto the central event in any projected synthesis. Several laboratories have published creative models on the topic and tactics range from putatively biomimetic to decidedly abiotic.6 Overman’s beautiful work in the area has reached an advanced stage.7 Baran et al. have very recently reported the first synthesis of axinellamines and massadines - truly seminal achievements.8a,b Our own experiments fall in the biomimetic vein wherein we view the carbocycle in 1 as derivative of oxidative halogenation; in particular of seco bis-alkylidenes 3.9 This choice makes symmetry in the problem clear, provides options for varying diastereochemical outcomes and, assuming 4 to be a viable intermediate en route to 1, reduces the initial task to that of controllably dimerizing a dispacamide alkaloid.5b We targeted fused heterobicycle 6 as a dispacamide synthon for this purpose, such that dimerization could occur via a derived dienolate oxidation. We describe here methods to synthesize the requisite monomers and show their dimerization can be uniquely executed in both a regio and stereocontrolled manner. We further demonstrate that derived bis-alkylidenes 3 are in fact susceptible to halogenative spirocyclization, validating our blueprint for syntheses of palau’amine type alkaloids generally.
The ring system in 6 had not been synthesized previously, although a generic relationship to simpler components was apparent (Figure 1). The Claisen condensation product of γ-butyrolactone and (CO2Et)2 was degraded with HBr / AcOH and the crude reaction treated with MeOH / cat H2SO4 to afford brominated α-keto ester 7 (Scheme 1).10 This material was exposed to hydrazine to generate a tetrahydropyridazine carboxylate11 that was subsequently allylated and saponified to afford 8. When carboxylic acid 8 was condensed with o-xylyl diamine-derived methylisothiourea 9,12 the adduct (10) could be induced to cyclize in situ by concentrating the reaction mixture at 70 °C in vacuo. This served to generate the desired N-amino glycocyamidine 11 via net extrusion of methyl mercaptan.
With 11 in hand, we set out to dimerize the material using enolate oxidation techniques.13 Enamide 11 could be deprotonated readily at low temperature with KHMDS. The resultant dienolate was prone to oxygenation, 14 but if anaerobic conditions were established, addition of FeCl2(DMF)3FeCl415 generated dimeric products efficiently. With the original assignment of palau’amine stereochemistry1,16 (namely, 1a) in mind, the desired outcome was a meso dimer resulting from homocoupling at the γ position of monomer 11 (namely 14a - Scheme 2). However, little of that material was formed. Rather, we obtained predominately a diastereomeric mixture of α,γ adducts 12 along with lesser amounts of α,α regioisomer 13.17 The trace amount of 14 observed was present as a 1:1 mixture of meso and C2 forms.18 We subsequently found that α,α coupling product 13 was unstable, and rearranged to α,γ forms 12 in toluene solution at 85 °C.19 Rather than disproportionate or Cope rearrange to γ,γ isomers 14, the molecule undergoes a net 1,3-sigmatropic shift. It was possible to purify 13 and photolyze the compound to generate additional 14, although this protocol was inefficient and continued to afford substantial amounts of 12. This result, and the fact we were unable to independently convert 12 to its γ,γ counterparts 14,20 necessitated that attention turn to controlling kinetic selectivity in the oxidative dimerization itself.
We observed that substituting I2 for FeCl2(DMF)3FeCl4 as the oxidant (Scheme 2) provided for a similarly efficient dimerization, although 14 was no longer detectable in the crude reaction mixture and the ratio of dimeric products now favored isomers 13. This was interpreted in terms of a monomeric radical dimerizing through an early, reactant-like transition state.13c,21 Independently, FeCl2(DMF)3FeCl4 was found not to convert 13 into 14 in THF solution at rt.22 It therefore seemed reasonable to infer the Fe(III)-based procedure (entry 1) involved an organometallic species whose homocoupling was more responsive to ground state stability of incipient products. If true, furthering that trend could make the γ,γ outcome competitive, perhaps favored - provided regioisomers 14 were indeed lower in relative energy.23 Replacing Fe(III) with Cu(II) did increase the proportion of 14 formed (entry 3), although the reaction remained impractical for preparative purposes. The breakthrough came when the potassium dienolate was treated with an equivalent of Cp2TiCl2 prior to oxidation. In that case, we obtained largely desired regioisomers 14 in good yield (Scheme 3). When [i-PrCp]2TiCl2 was used, selectivity for 14 was essentially complete - regioisomers 12 or 13 could no longer be detected in crude reaction mixtures. Our understanding of the mechanism operating is limited. However, reasoning by analogy to Schmittel’s findings on the electrochemical oxidation of simpler titanocene enolates,24 we interpret the outcome in terms of Cu(II) engaging an intermediate of type 15 to initiate dimerization via an electron deficient species whose Ti-O bond is intact at the point of C-C bond formation.25 This would effectively block the dienolate α position. Attempts to further characterize details of this atypical oxidative coupling are ongoing.
Our initial plans for palau’amine synthesis called for removing one allyl unit from meso 14 - en route to a substance carrying the acylpyrrole unit on N2 (see Figure 1) needed to construct the dipyrrolopyrazinone motif found in 1 (vide infra). Unfortunately, the seemingly simple task of deallylating 14a/b proved untenable, at least by direct means.26 Thus, a more ambitious goal was pursued wherein the acylpyrrole component would be installed early and carried through the synthesis - giving our dimeric intermediates the bis-pyrrole composition common to axinellamine, massadine, carteramine, and konbu’acidin.6,16 The monomer required for this purpose was synthesized beginning with tetrahydropyridazine 16 (Scheme 4). This material was treated with acid chloride 1727 and the adduct saponified to afford carboxylic acid 18. Condensation with methylisothiourea 9 subsequently provided amide 19 from which target 20 was derived by mild thermolysis in the presence of HgCl .28
The potassium dienolate of 20 decomposed readily at -78 °C. However, if 20 was treated with [i-PrCp]2TiCl2 prior to KHMDS, subsequent oxidation with Cu(OTf)2 afforded dimers 21 (3.3:1 meso:C2) in respectable yield (Scheme 5). Notably, changing the oxidant to FeCl2(DMF)3FeCl4 reversed the selectivity to now favor the C2 isomer (1:3 meso:C2). The yield in this case was diminished due to competing formation of α,γ regioisomeric dimers. Nonetheless, selectivity and efficiency observed were encouraging given the complexity of the construction.
With meso 21 assembled, we again faced a desymmetrization, yet one quite different than confronted in 14. Plans called for saturating a single N-N sigma bond en route to a compound of type 24 (Scheme 6A). Models suggested relieving one such ring constraint would provide flexibility necessary for the intended spirocyclization (as drawn), while the N-N bond remaining would enforce a syn relationship between C-Cl bond and C18 aminomethyl branch within polycycle 25 generated upon chlorination. Of course, this outcome would require chemoselective halogenation of the vinyl hydrazide in 24, itself a tenuous proposal. However, the alternative was to delay desymmetrization and break both N-N bonds in 21 en route to bis-alkylidenes of type 3 (Figure 1). The concern there was that stacked conformers having minimal allylic strain in the tether connecting glycocyamidine units would react preferentially to give products having a trans relationship between ring substituents at carbons 17 and 18. This was unacceptable in light of Scheuer’s original palau’amine assignment (1a), and thus means to generate structures 24 took priority.
A two-electron reduction of 21 having the effect of saturating one N-N bond was an interesting problem. We chose to pursue a conjugate addition manifold, the thought being hydride addition to one enamide could generate a transient ketene aminal (e.g. 22) perhaps able to β-eliminate, wherein the ring-opened product (23) would tautomerize upon protonation to give 24. In practice, working with meso 21, procedures employing various soft metal hydrides, either as isolated reagents or generated in situ, were ineffective for this purpose. However, we did observe that Wilkinson’s complex could catalyze partial reduction of 21 to afford 27 in the presence of PhMe2SiH,29 provided that 21 was pre-treated with 2 equiv. NH4PF6 to suppress catalyst poisoning. Isolated yields were at first poor, although the method was improved significantly using rhodium carbenoid 3130 complexed with Buchwald’s ‘DavePhos’31 as catalyst. This procedure gave 27 in yields indicative of multiple catalyst turnovers. The next question was whether the silylketene aminal putatively formed in situ (namely 26) could be coaxed to fragment en route to a structure of type 24. Fluoride treatment of crude hydrosilylation mixtures decreased the isolated yield of 27, but not productively. We presumed the NH4PF6 pre-treatment used to facilitate conversion had converted meso 21 into a salt form wherein 26 would be protonated and could self-immolate in situ. Lewis acids, rather than NH4PF6, were therefore screened as alternate promoters. When MgBr2 etherate was employed, we continued to isolate 27, but the major product proved to be a single diastereomer of polycycle 28 - an outcome rationalized in terms of a Lewis acid-coordinated form of 26 undergoing an internal Mukaiyama-Michael addition. This unintended result was potentially valuable. The core diazabicycle in 28 was identical to that targeted in 25, just lacking halogen on the methano bridge. X-ray diffraction analysis of crystals grown from samples of SEM-deprotected 28 showed desired relative stereochemistry throughout the structure. Moreover, treating 28 with KHMDS efficiently cleaved the hydrazide N-N bond adjacent to the enolizable methine to afford 30. Rendering that material more like we draw the natural product (as shown) makes its relation to the original palau’amine assignment clear. One could envision scenarios in which a series of controlled reductions might achieve that end.
With that topic left for a future date, we returned to the issue of generating 24. We were unable to steer the hydrosilylation of meso 21 towards 24 in situ, yet the observation that 28 cleanly fragmented to 30, presumably via its potassium enolate, upon KHMDS treatment suggested the 22→23 conversion was viable - despite the N-N bond to be cleaved being oriented near orthogonal to the enolate π system. We thus treated isolated 27 with KHMDS. Unlike 28, this caused extensive decomposition. However, if we dissolved 27 in DMF and added an equivalent of anhydrous LiCl, followed by DBU,32 we could isolate a ring-opened material in good yield. Unfortunately, ring opening occurred cleanly within the oxidized ‘half’ to afford extended chromophore 32 as a canary yellow solid (Scheme 7).
Enolizing a mono-reduced dimer was evidently not a method to accesss 24. While paths to circumvent the problem seemed available, our next move was actually dictated by external events. In early 2007, a succession of papers called into question Scheuer’s original assignment of palau’amine relative stereochemistry.16 Characterization of newly discovered relatives, along with re-interpretation of spectroscopic data collected on palau’amine itself, suggested the molecule was epimeric at carbons 12 and 17, wherein 1b (Figure 1) would be the true form of the natural product. Diastereomer 1b was a fascinating variant and fortunately, one of numerous potential stereochemical outcomes downstream of a 3→4 oxidation (Fig 1). Up to this point we had targeted 24 as a halogenation substrate intending to enforce an ‘all syn’ substituent array about the cyclopentane core as assigned in 1a. If 1b were the natural product, the N-N bond in 24 would be an unnecessary constraint and we could simply target symmetric bis-alkylidenes 3 as per the generic plan.
This was examined first in the meso series. Altering the stoichiometry of silane to 2.2 equiv. and using a combination of NH4PF6 and MgI2 additives, rhodium catalyzed reduction of meso 21 provided doubly saturated compounds 33 in high yield (Scheme 8). All three possible stereoisomers were formed - two meso forms 33a/b along with non-symmetric material 33c. This was of no preparative consequence. After careful experimentation, we found the isomeric mixture could be treated with excess Cy2BOTf, followed by brief exposure to KHMDS to afford doubly ring-opened, bis-alkylidene 34 in good yield. The reaction generates a mixture of geometric isomers upon hydrolytic workup, however these converge to a single, symmetric form during silica gel chromatography or standing in chloroform solution at room temperature. We tentatively assign Z stereochemistry at both alkenes in 34.33
With 34 in hand, we could examine the central tenant in our approach - the propensity of such compounds to form spirocyclic products upon oxidative halogenation. Gratifyingly, when 34 was dissolved in THF containing MgCl2, cooled to -78 °C and treated with freshly prepared t-BuOCl, we obtained three chlorinated products - all of which data suggests contain the target spirocyclopentane motif. Minor product 35 is analogous to the alkylidene formed in our earlier model studies,9 except the material is now fully functionalized and properly chlorinated. A vicinal C17H/C18H coupling constant of 12.8 Hz implies the trans stereochemistry shown,16 although we have yet to assign configuration at C16. Interestingly, the major products of the reaction appear to be C10 epimers of polycycle 36. This unique structure is rationalized in terms of a rebound addition of the guanidine unit at N1 to N6 of an acyl iminium species generated transiently upon spirocycle formation (Scheme 8B) - rather than the deprotonation of Ha that presumably affords 35.34 While we have yet to characterize 36 crystallographically, all spectroscopic measurements as well as mass spectra are consistent with the assignment. Note that the C17H/C18H cis (J = 3.9 Hz) stereochemistry proposed in 36 differs from 35. Our initial argument that the hydrazide constraint in 24 would be required to establish this arrangement seems in error. That said, several reaction variables are ambiguous: the reacting alkylidene geometry, the reversibility of the spirocyclization, and the identity of the halogen transfer species. Nevertheless, upon chlorination we find crude mixtures contain alkylidene 35 having data consistent with C17/C18 trans stereochemistry whereas isolated polycycles 36 display the corresponding substituents in a cis relation. A lack of coupling between C11H and C12H and a weak NOESY correlation between C11H and the C13 methylene protons imply C11/C12 trans stereochemistry in 36 and, to the extent the diazabicyclooctane motif can be assumed cis fused, the C16 C-N bond would be oriented beta.
Another intriguing finding was made using MgBr2·Et2O in place of MgCl2 during the halogenation. In that case we obtained only 38, wherein C17 was substituted with bromine rather than chlorine.35 The yield of product had nearly tripled and the corresponding alkylidene 37 could not be detected in crude reaction mixtures. Like 36, pentabromide 38 was isolated as a mixture of C10 epimers. The major epimer was treated with Et3BHLi followed by anhydrous TiF4 powder in toluene to afford a mixture of SEM-deprotected hemiaminal isomers.36 We tentatively assign structure 39 to these materials.
The above results were encouraging. However, we were uncertain if comparable reactivity could be observed in the C2-symmetric series, which is configured appropriately to reach revised palau’amine structure 1b. Catalyzed hydrosilylation remained an effective method of reducing the dimer, in this case with C2-21 affording saturated isomers 40 (Scheme 9). However, unlike the synthesis of 34 from 33, isomers 40 reacted chaotically with KHMDS, giving intractables, both with and without added Cy2BOTf. This was a persistent difficulty and considerable time passed before an alternative was found. Ultimately, treating a CH3CN/DMF solution of 40 with excess proazaphosphatrane 4237 at ambient temperature provided doubly ring-opened, partially debrominated bis-alkylidene 43. If care was taken to avoid decomposition during isolation, it was possible to obtain 43 in 40% yield, contaminated with only trace impurities. We know of no precedent for this remarkable process wherein two N-N sigma bonds are cleaved and two heteroaryl carbon bromine bonds are reduced in a single operation.
Bis-alkylidene 43 is prone to decomposition although, when pure, the material could be dissolved in cold CH2Cl2 and treated with t-BuOCl to generate spirocyclic mono-alkylidene 44. In contrast to reactions in the meso series, added MgCl2 or MgBr2 was neither beneficial nor steered the outcome towards polycycles analogous to 36. In fact, Lewis acidic additives were generally detrimental. Our best result employed t-BuOCl alone, affording 44 in 20-32% yield, with the mass balance being intractable, seemingly oligomeric materials.
This result validates our thesis and, despite low yields at present, we anticipate optimizations. Chloride 44 is the sole isolable product from the oxidation, which is the 10th linear step in our sequence. The C17H/C18H trans relation is supported by a vicinal coupling constant of 11.6 Hz.16 The C18H/C12H trans relation is thought maintained during the oxidation and is consistent with ROESY correlations in 44 between C18H and the C13 methylene protons as well as between C12H and C17H. While the alkylidene geometry and relative configuration at C16 are not yet known, the former will likely be inconsequential and the latter potentially adjustable (via equilibration) in the context of more complete intermediates.39,40 Most importantly, confident that bis-alkylidenes of type 34/43 can be precursors to the halogenated core common to palau’amine and its relatives, we can now focus on tactical refinements needed for total syntheses. These include substrate modifications to 1) better manage stability and basicity of intermediates 2) permit more ready unveiling of free guanidine functionality41 3) tunably dictate stereochemical outcomes, and 4) generate optically-active end products. Studies along these lines are ongoing.
A mixture of 19 (34 g, 50 mmol), HgCl2 (19 g, 67 mmol) and pyridine (12.1 mL, 150 mmol) in CH3CN (250 mL) was heated at reflux for 3 h. A second portion of HgCl2 (2.7 g, 13.4 mmol) was added and reflux continued for 1 h. Upon cooling to rt, a white solid was removed by filtration and the solvent was evaporated in vacuo. The reaction mixture was taken up in EtOAc and the undissolved solids were removed by filtration. The organic layer was washed with 1N NaOH whereupon a white precipitate formed. The precipitate was filtered and the NaOH wash / filtration sequence continued until no further solid formed (6-7 ×). The organic layer was then washed with water and brine and dried over NaSO4. Concentration in vacuo and purification by silica gel chromatography (1:3 EtOAc/hexanes) afforded 20 as a white foam (15.0 g). Mixed fractions were purified on a second silica gel column (1:9 EtOAc/CH2Cl2) to provide a further 5.0 g of 20 (total yield = 70%). Rf = 0.7 (1:9 CH3CN: CHCl3); IR (film): 2951, 1741, 1635, 1402, 1093, 859, 758, 793 cm-1. 1H NMR (400 MHz, CD3CN): δ 7.38-7.28 (m, 4H), 6.81 (s, 1H), 5.82 (t, J = 4.7 Hz, 1H), 5.60 (s, 2H), 4.89 (s, 2H), 4.66 (s, 2H), 3.92-3.75 (m, 2H), 3.57 (t, J = 12 Hz, 2H), 2.33 (dd, J = 5.3, 10.4 Hz, 2H), 0.88 (t, J = 12 Hz, 2H), 0.0 (s, 9H). 13C NMR (100 MHz, CD3CN): δ 163.8, 159.3, 141.6, 141.4, 134.9, 129.5, 129.2, 129.1, 129.0, 126.3, 118.7, 118.0, 112.6, 104.2, 100.2, 76.1, 66.7, 49.5, 45.7, 43.6, 23.5, 18.2, -1.3. HRMS (ESI-TOF) calcd for C25H30Br2N5O3Si (M+H)+: 634.0479; found: 634.0483.
Monomer 20 (311 mg, 0.49 mmol) and solid [i-PrCp]2TiCl2 (190 mg, 0.57 mmol) were dissolved in anhydrous THF (2.5 mL). The brandy colored solution was degassed via consecutive freeze-pump-thaw cycles (3×) and cooled to -78 °C. KHMDS (1.08 mL, 0.5 M in toluene) was added dropwise and the resultant dark green slurry was stirred at -78 °C for 1.5 h. A fine suspension of Cu(OTf)2 (260 mg, 0.74 mmol) in dry, degassed THF was added over 15 min and the dark red/brown slurry stirred at -78 °C for 3h and then at rt for 1.5 h. The reaction was treated with pH 8.0 aq EDTA (5 mL, 0.35 M) and the volatiles were removed in vacuo. The residue was diluted in EtOAc, washed with additional pH 8.0 EDTA solution (until blue color no longer observed), H2O and brine. The organics were dried over Na2SO4, filtered and concentrated. Purification by column chromatography on silica gel (1% CH3CN/CHCl3) afforded 21 meso (150 mg, 49%) and 21 C2 (45 mg, 15%). 21 meso: light yellow solid; Rf = 0.35 (10% CH3CN/CHCl3); IR (film): 1745, 1745, 1698, 1447, 1396, 1093, 934, 836 cm-1; 1H NMR (400 MHz, CD3CN, 70°C): δ 7.40-7.20 (m, 8H), 6.71 (s, 2H), 5.66-5.48 (m, 6H), 4.95 (s, 4H), 4.62 (s, 4H), 3.77 (bs, 4H), 3.57 (t, J = 8.0 Hz, 4H), 2.49 (s, 2H), 0.86 (t, J = 8.0 Hz, 4H), 0.01 (s, 18H); 13C NMR (75 MHz, CDCl3, 55 °C): δ 163.0, 158.4, 140.7, 140.2, 133.6, 130.3, 129.3, 128.8, 128.7, 128.5, 125.3, 118.0, 112.4, 104.0, 100.2, 75.6, 66.5, 49.6, 45.6, 43.6, 36.5, 18.2, -1.2; HRMS (ESI-TOF) calcd for C50H57Br4N10O6Si2 (M+H)+ 1265.0729, found 1265.0730. Crystals of 21 meso suitable for X-ray diffraction were grown from MeOH (slow evaporation). Details of the crystallographic analysis are provided in a separate CIF file. 21 C2: light yellow solid; Rf = 0.2 (10% CH3CN/CHCl3); IR (film): 1745, 1698, 1448, 1397, 1093, 934, 836 cm-1; 1H NMR (400 MHz, CD3CN, 70°C): δ 7.40-7.20 (m, 8H), 6.72 (s, 2H), 5.72 (s, 2H), 5.51(s, 4H), 4.95 (s, 4H), 4.61(S, 4H), 3.87 (bs, 4H), 3.58 (t, J = 8 Hz, 4H), 2.49 (s, 2H), 0.86 (t, J = 8 Hz, 4H), 0.01 (s, 18H); 13C NMR (75 MHz, CDCl3, 55 °C): δ 163.0, 158.5, 140.7, 140.2, 133.6, 130.2, 129.3, 128.8, 128.5, 125.0, 118.1, 112.5, 103.7, 100.2, 75.6, 66.6, 49.6, 46.4, 43.6, 36.7, 18.2, -1.2; HRMS (ESI-TOF) calcd for C50H57Br4N10O6Si2 (M+H)+: 1265.0729; found: 1265.0707.
A 25 mL flame-dried flask was charged with 21 C2 (2.0 g, 1.58 mmol), NH4PF6 (770 mg, 4.72 mmol), LiI (233 mg, 1.74 mmol) and THF (7.9 mL) and the mixture was stirred at rt for 10 min. The solvent was removed in vacuo. A separate 25-mL flame-dried flask was charged with 100 mg of cyclooctadiene N, N-dimethylimidazolium rhodium (I) iodide 41 (0.24 mmol, 15 mol%), 2-dicyclohexylphosphino-2′-(N,N-dimethylamino)biphenyl (102 mg, 0.26 mmol, 17 mol%) and CH2Cl2 (7.9 mL). PhMe2SiH (644 mg, 4.72 mmol) was added and the resulting solution was stirred at rt for 5 min. This catalyst solution was added to the flask containing 21 and the resulting suspension was heated at 50 °C for 72 h. Upon cooling to rt the reaction mixture was quenched with saturated NaHCO3 and diluted with CH2Cl2. The aqueous layer was separated and extracted with CH2Cl2. The combined organics were dried over MgSO4, filtered and concentrated in vacuo. Purification via column chromatography on silica gel (gradient from 5%→50% CH3CN/CHCl3) gave 928 mg of 40a that was roughly 60% pure. This was followed by 40b (840 mg, 42%). Pure 40a was obtained by triturating with CH3CN (yield ~30%). 1H NMR (500 MHz, CDCl3): δ 7.37-7.25 (m, 8H), 6.82 (s, 2H), 5.81 (d, J = 10.5 Hz, 2H), 5.36 (d, J =10.5 Hz, 2H), 4.90 (d, J = 19.2 Hz, 2H), 4.87 (d, J = 19.2 Hz, 2H), 4.73 (d, J = 19.2 Hz, 2H), 4.70 (d, J = 19.2 Hz, 2H), 4.45 (d, J = 12.8 Hz, 2H), 3.73-3.70 (m, 2H), 3.62 (dd, J = 11.6, 4.5 Hz, 2H), 3.45 (ddd, J = 9.0, 6.9, 1.9 Hz, 4H), 2.49 (dd, J = 12.7, 11.3 Hz, 2H), 2.11-2.08 (m, 2H), 1.59-1.54 (m, 2H), 1.20-1.07 (m, 2H), 0.89-0.74 (m, 2H), -0.06 (s, 18H). 13C NMR (100 MHz, CDCl3): δ 169.4, 163.7, 145.1, 140.1, 133.0, 129.4, 128.6, 128.5, 128.4, 128.3, 125.2, 117.3, 111.0, 99.4, 75.1, 66.1, 56.2, 49.2, 43.3, 34.9, 29.9, 17.9, -1.4. MS for C50H61Br4N10O6Si2 (ESI+) m/z (relative intensity): 1273 (M++1, 100). 40b: 1H NMR (500 MHz, CD3CN @ 70 °C): δ 7.37-7.27 (m, 7H), 7.24-7.20 (m, 1H), 6.77 (s, 1H), 6.75 (s, 1H), 5.61 (d, J = 10.7 Hz, 1H), 5.59 (s, 2H), 5.54 (d, J = 10.7 Hz, 1H), 4.95-4.87 (m, 4H), 4.75 (d, J = 14.5 Hz, 1H), 4.70 (d, J = 14.4 Hz, 1H), 4.62 (d, J = 14.5 Hz, 1H), 4.60 (d, J = 14.5 Hz, 1H), 4.56-4.43 (m, 1H), 4.12 (t, J = 6.1 Hz, 1H), 4.01 (dd, J = 12.9, 7.3 Hz, 1H), 3.81 (dd, J = 11.6, 4.9 Hz, 1H), 3.59 (t, J = 8.1 Hz, 2H), 3.57 (t, J = 8.1 Hz, 2H), 3.26 (dd, J = 12.9, 5.3 Hz, 1H), 2.33 (t, J = 12.2 Hz, 1H), 2.18-2.16 (m, 1H), 1.91-1.86 (m, 1H), 1.78-1.69 (m, 2H), 1.28-1.21 (m, 1H), 1.04 (q, J = 12.2 Hz, 1H), 0.97-0.82 (m, 4H), 0.01 (s, 9H), 0.00 (s, 9H). 13C NMR (100 MHz, CD3CN): δ 172.1, 171.3, 164.9, 164.6, 149.1, 148.0, 135.8, 135.7, 130.8, 130.7, 130.3, 130.2, 130.0, 129.9, 128.7, 128.5, 118.3, 111.8, 111.5, 100.8, 100.7, 77.5, 67.9, 58.0, 50.5, 50.3, 45.1, 44.9, 34.8, 34.4, 31.9, 27.9, 19.6, 19.5, -0.4. HRMS (ESI-TOF) calcd for C50H61Br4N10O6Si2 (M+H)+: 1269.1042; found: 1269.1038.
A flame-dried 5 mL flask was charged with 40a or 40b (100 mg, 78.6 μmol), DMF (830 μL) and CH3CN (330 μL) at rt. 2,8,9-Triisobutyl-2,5,8,9-tetraaza-1-phosphabicyclo[3.3.0]undecane (269 mg, 786 μmol) was added and the resulting mixture was stirred at rt for 22 h. The dark reddish-purple mixture was diluted with EtOAc and washed with saturated aqueous NH4Cl, water, saturated aqueous NaHCO3, water and brine. The organic layer was dried over Na2SO4, filtered and concentrated in vacuo. The crude residue was purified by column chromatography on silica gel slurry packed (1% Et3N-CH2Cl2) and eluted (2% MeOH-CH2Cl2) to give the desired product in yields ranging from 30-50%. Pure 43 was obtained using HPLC (5 μm C-18 column, 250 × 10 mm, 85 % CH3CN / H2O, 8 mL/min, UV detection at 254 nm; TR = 6.0 min). 1H NMR (500 MHz, CD3CN): δ 7.48 (bs, 2H), 7.42-7.31 (m, 8H), 6.97 (d, J = 1.8 Hz, 2H), 6.63 (d, J = 1.8 Hz, 2H), 5.60 (d, J = 10.2 Hz, 2H), 5.51 (d, J = 9.2 Hz, 2H), 5.42 (d, J = 10.2 Hz, 2H), 4.91 (d, J = 18.2 Hz, 2H), 4.88 (d, J = 18.2 Hz, 2H), 4.62 (d, J = 14.9 Hz, 2H), 4.52 (d, J = 14.8 Hz, 2H), 3.46-3.38 (m, 6H), 3.18-3.06 (m, 4H), 0.79-0.70 (m, 4H), -0.12 (s, 18H). 13C NMR (100 MHz, CD3CN): δ 167.5, 161.6, 140.3, 136.5, 130.3, 130.0, 129.7, 128.7, 127.4, 116.2, 115.9, 96.4, 77.7, 67.1, 46.1, 44.2, 41.7, 30.8, 18.7, 18.6, 0.0. MS calcd for C50H63Br2N10O6Si2 (ESI+) m/z (relative intensity): 1115.30; found: 1115.00.
Symmetric bis-alkylidene 43 (18 mg, 16.1 μmol) was dissolved in CH2Cl2 (150 μL) and the resulting solution was cooled to -78 °C. A stock solution of tert-butylhypochlorite (20 μL) in CH2Cl2 (500 μL) was prepared fresh at rt prior to use. 50 μL of this stock solution was added to the solution of 43 and stirring was continued at -78 °C for 10 min. After warming to rt, the mixture was diluted with CH2Cl2 and washed with saturated aq NaHCO3. The organic layer was dried over Na2SO4, filtered and concentrated in vacuo. Purification by preparative thin layer chromatography (9% MeOH/CH2Cl2) gave 6.4 mg (36%) of 44. Pure 44 was obtained after HPLC: analytical (5 μm C-18 column, 250 × 4.6 mm, 85 % CH3CN / H2O, 1 mL/min, UV detection at 254 nm, TR = 13.3 min); preparative (5 μm C-18 column, 250 × 10 mm, 85 % CH3CN / H2O, 8 mL/min, UV detection at 254 nm, TR = 7.5 min). 1H NMR (500 MHz, CD3CN): δ 8.15 (s, 1H), 7.48-7.45 (m, 1H), 7.42-7.36 (m, 2H), 7.35-7.26 (m, 6H), 7.02 (d, J = 1.9 Hz, 1H), 7.00 (d, J = 1.9 Hz, 1H), 6.83 (d, J = 1.7 Hz, 1H), 6.75 (d, J = 1.9 Hz, 1H), 5.61 (d, J = 10.3 Hz, 1H), 5.59 (d, J = 10.4 Hz, 1H), 5.58 (d, J = 10.3 Hz, 1H), 5.56 (d, J = 10.3 Hz, 1H), 4.92 (s, 2H), 4.73 (s, 2H), 4.56 (d, J = 14.8 Hz, 1H), 4.49 (d, J = 15.0 Hz, 1H), 4.49 (d, J = 15.0 Hz, 1H), 4.38 (d, J = 15.0 Hz, 2H), 4.01 (d, J = 11.6 Hz, 1H), 3.77 (ddd, J = 13.1, 4.3, 4.3 Hz, 1H), 3.65 (ddd, J = 14.3, 6.0, 4.3 Hz, 1H), 3.49-3.38 (m, 4H), 3.46 (t, J = 2.0 Hz, 2H), 3.13 (ddd, J = 8.8, 4.1, 4.1 Hz, 1H), 2.32-2.25 (m, 1H), 0.79 (t, J = 2.0 Hz, 2H), 0.77-0.71 (m, 2H), -0.10 (s, 9H), -0.14 (s, 9H). 13C NMR (100 MHz, CD3CN): δ 178.9, 166.5, 162.5, 162.4, 158.9, 158.7, 140.3, 139.8, 138.5, 136.6, 136.4, 130.8, 130.3, 130.2, 130.1, 129.9, 129.8, 129.7, 129.3, 129.0, 128.3, 127.8, 127.3, 116.8, 116.5, 96.4, 96.3, 78.0, 77.6, 76.6, 69.7, 67.2, 67.0, 47.8, 46.0, 45.7, 45.2, 45.0, 44.1, 42.9, 40.2, 18.6, -1.0. HRMS (ESI-TOF) calcd for C50H63Br2ClN10O6Si2 (M+H)+: 1147.2442; found: 1147.2432.
Funding was provided by the NIH (RO1-GM60591), the Robert A. Welch Foundation, the Chilton Family Endowment, the Donald J. & Jane M. Cram Endowment, and unrestricted research awards from Merck and Pfizer. A large portion of this work was carried out at the University of Texas Southwestern (UTSW) Medical Center where Q. Li was a Robert A. Welch Foundation Graduate Fellow. We are grateful to Drs. Ranny Mathew Thomas, Jian Wang and Christopher Lindsey for valuable experimental assistance.