Synthesis of Oxa-Bridged Analogues of Farnesyltransferase Inhibitor RPR 115135

Download Synthesis of Oxa-Bridged Analogues of Farnesyltransferase Inhibitor RPR 115135

Post on 21-Feb-2017

212 views

Category:

Documents

0 download

Embed Size (px)

TRANSCRIPT

<ul><li><p>Synthesis of Oxa-Bridged Analogues of FarnesyltransferaseInhibitor RPR 115135</p><p>Celine Martin, Patrick Mailliet, and Jacques Maddaluno*,</p><p>Laboratoire des Fonctions Azotees et Oxygenees Complexes de l'IRCOF, UMR 6014 CNRS,Universite de Rouen, 76821 Mont St Aignan, France, and Aventis-Pharma SA, Centre de Recherche de</p><p>Vitry-Alfortville, 13 quai Jules Guesde, 94403 Vitry-sur-Seine Cedex, France</p><p>jmaddalu@crihan.fr</p><p>Received January 5, 2001</p><p>Two synthetic routes to new oxygen-bridged analogues of farnesyltransferase inhibitors are describedthat follow either a [3 + 2]/[4 + 2] or a [4 + 2]/[3 + 2] sequence of reactions. The first approach hasbeen achieved by reacting the in situ generated phenylisobenzofuran (PIBF) 4 with pyrroline 5aand has led stereoselectively to racemic 18, which was transformed in a few steps into the targetmolecule 2. The second pathway relies on a key intermediate 6, obtained either by condensation ofPIBF with methyl acrylate, followed by a deprotonation/selenation and an oxidation/eliminationsequence, or by cycloaddition between PIBF and R-phenylselenoacrylate 11, followed by the sameoxidation/elimination sequence. The reaction of 6 with amino dipole 7 gives diastereoselective accessto pyrrolidine 25, a precursor of the second target 3, an epimer of 2.</p><p>Introduction</p><p>The central role played by mutations of the Ras proteinin human cancerogenesis, especially in colon and pan-creas tumors, has been well documented lately.1 Threedifferent mutation types, the so-called Harvey-Ras, Neu-roblastoma-Ras, and Kirsten-Ras 4B, have been identi-fied and are characterized by their C-terminal variations.The early steps of the regular cell division process involvethe farnesylation and binding of the Ras-GDP protein tothe inner wall of the cell. This event, followed by aphosphorylation into Ras-GTP, induces a signal cascadeending with the translocation into the nucleus and thegene transcription activation. Because it is unable toswitch back from the Ras-GTP to the Ras-GDP form, themutated protein keeps on firing a continuous cell prolif-eration signal. Among the various biochemical strategiesstudied to reverse this phenomenon, the inhibition offarnesyl transferase (FT), which prevents the Ras-GDPfixation on the cell wall, has turned out to be the mostefficient one and has opened promising new directionsin cancer therapy. There is, however, one major difficultyassociated with this approach, and that is the selectiveinhibition of FT with respect to related enzymes such asgeranylgeranyltransferase (GGT), of which Ras is also asubstrate. Several classes of compounds have beendeveloped to reach an acceptable level of selectivity,among which is the benzo[f]perhydroisoindole (BPHI)class.</p><p>Compounds such as RPR 115135 (1, Figure 1), whichcombines a remarkably low GGT inhibition profile withan elevated cellular potency and moderate in vivo activi-ties,2 constitutes a particularly effective lead molecule inthis series. A large investigation on the structure-activity relationship has already been achieved aroundthis skeleton and has revealed key features such as the</p><p>presence of two aromatic rings, one on the lateral amidechain and the other one at the 9 position.2 However,several other important structural characteristics suchas the nature of the bridging element, or the position ofthe angular acid group, have not been taken into accountyet. The possible role played by these fragments in thedrug-enzyme interaction led us to prepare the two newanalogues displayed in Figure 1, i.e., 2 (RPR 225,370)and 3 (RPR 222,490). We present in this paper thedifferent synthetic pathways we have studied to accessthese compounds.</p><p>Results and Discussion</p><p>The synthesis of the skeleton of 2 and 3 has beentackled through the routes summarized in Figure 2,</p><p> Universite de Rouen. Rhone-Poulenc Rorer, currently Aventis-Pharma SA.(1) Bos, J. L. Cancer Res. 1989, 49, 4682.</p><p>(2) Mailliet, P.; Laoui, A.; Bourzat, J. D.; Capet, M.; Cheve, M.;Commercon, A.; Dereu, N.; Le Brun, A.; Martin, J. P.; Peyronel, J. F.;Salagnad, C.; Thompson, F.; Zucco, M.; Guitton, J. D.; Pantel, G.;Bissery, M. C.; Brealey, C.; Lavayre, J.; Lelievre, Y.; Riou, J. F.;Vrignaud, P.; Duchesne, M.; Lavelle, F. In Farnesyltransferase andGeranylgeranyltransferase Inhibitors in Cancer and CardiovascularTherapy; Sebti, S., Ed.; Humana Press: Totowas, NJ, in press.</p><p>Figure 1. Farnesyl transferase inhibitor RPR 115135 and itsanalogues.</p><p>3797J. Org. Chem. 2001, 66, 3797-3805</p><p>10.1021/jo0100188 CCC: $20.00 2001 American Chemical SocietyPublished on Web 05/10/2001</p></li><li><p>corresponding to disconnections based on either a [3 +2]/[4 + 2] or a [4 + 2]/[3 + 2] cycloaddition sequence. Thekey step in route A relies on the reaction betweenphenylisobenzofuran 4 and the pyroline 5. In route B,the tetracyclic skeleton is obtained by addition of dipole7 (derived from amine 9) on the double bond of 6. Thislatter compound can be obtained from the correspondingR-selenoester 8. Two possibilities open at that stage, thefirst one (B1) relying on ester 10, obtained by reactingPIBF 4 and an acrylate. The second path of the fork (B2)is based on the direct reaction between 4 and theR-selenoacrylate 11. We will see in the following that eachof these pathways A and B leads stereoselectively toprecursors of 2 and 3, respectively.</p><p>Route A: Access to Analogue 2</p><p>Let us first discuss the case of route A. The twopartners needed for this approach are relatively easy toprepare. The pyroline skeleton of 5 is indeed accessiblethrough the [3 + 2] scheme proposed by Sakurai andAchiwa (Scheme 1).3 The reaction of chloromethylbutylether 12 with lithium amide 13 provides the tertiaryamine 9. The action of a catalytic amount of trifluoro-acetic acid on the latter in methylene chloride generatesthe dipole 7, which reacts in situ with methyl propiolateat room temperature to yield pyroline 5a in 59% yield.</p><p>On the other hand, phenylisobenzofuran 4 is a well-known compound of which synthesis have been describedunder both acidic and basic conditions.4 An acidic me-dium was anticipated to be hardly compatible with thestrained oxygen bridge in the expected adducts. We havethus preferred to prepare the diene through a basic</p><p>process and resorted to the method described by Tobiaand Rickborn,4c which relies on phthalane 17 (Scheme2). This compound has been obtained following Rodrigoand co-workers convenient procedure.5 Thus, acetaliza-tion of o-bromobenzaldehyde 14 followed by bromine-lithium exchange and benzaldehyde condensation pro-vides alcohol 16, which cyclizes in methanol in thepresence of acidic Dowex 50W-X2. The phthalane 17obtained presents a syn/anti ratio of 43:57. Addition ofn-butyllithium to 17 triggers a -elimination and affords,after tert-butyl alcohol quenching,6 the 1-phenylisoben-zofuran (PIBF) 4. This diene has not been isolated orpurified and has been trapped in situ by pyrroline 5aunder mild thermal conditions (3 h 20 min at 55 C inTHF). A single adduct 18 was recovered after flashchromatography on silica gel in a mediocre 22% yield forthe two steps (Scheme 2).</p><p>The NMR analysis indicates that the ring junctionproton H11 appears as a doublet, indicative of both theregioselectivity and exo (with respect to the ester group)selectivity of this reaction. The coupling between H11 andH4 is indeed characteristic of the -orientation of H11. The</p><p>(3) (a) Hosomi, A.; Sakata, Y.; Sakurai, H. Chem. Lett. 1984, 1117.(b) Terao, Y.; Kotaki, H.; Imai, N.; Achiwa, K. Chem. Pharm. Bull.1985, 33, 2762.</p><p>(4) (a) Keay, B. A.; Plaumann, H. P.; Rajapaksa, D.; Rodrigo, R. Can.J. Chem. 1983, 61, 1987. (b) Hayakawa, K.; Yamaguchi, Y.; Kane-matsu, K. Tetrahedron Lett. 1985, 26, 2689. (c) Tobia, D.; Rickborn,B. J. Org. Chem. 1986, 51, 3849. (d) Porsey, S. P.; Rajapaksa, D.;Taylor, N. J.; Rodrigo, R. J. Org. Chem. 1989, 54, 4280.</p><p>(5) (a) Plaumann, H. P.; Smith, J. G.; Rodrigo, R. J. Chem. Soc.Chem. Commun. 1980, 354. (b) Kuroda, T.; Takahashi, M.; Kondo, K.;Iwasaki, T. J. Org. Chem. 1996, 61, 9560.</p><p>(6) Using methanol to quench this reaction leads to a doubleconjugated addition of lithium methylate on methyl propiolate, provid-ing to the known 2,2-dimethoxy methyl propionate. See, for instance:(a) Walia, J. S.; Walia, A. S. J. Org. Chem. 1976, 41, 3765. (b) Bertz,S. H.; Dabbagh, G.; Cotte, P. J. Org. Chem. 1982, 47, 2216. (c) Tietze,L. F.; Meier, H.; Voss, E. Synthesis 1988, 274. (d) Hosokawa, T.; Aoki,S.; Murahashi, S. I. Synthesis 1992, 558.</p><p>Figure 2. Three retrosynthetic routes to analogue 2 and 3 skeleton.</p><p>Scheme 1 Scheme 2</p><p>3798 J. Org. Chem., Vol. 66, No. 11, 2001 Martin et al.</p></li><li><p>latter, in the R-position, would correspond to a singlet,due to the 90 dihedral angle between H11 and H4, asevidenced below. This exo selectivity can be rationalizedby a putative - interaction taking place between thebenzylic part of 5a and the cyclohexadienic moiety of 4(Figure 3).</p><p>The following functionalization steps to convert adduct18 into analogue 2 first go through a debenzylation. Allattempts made to hydrogenolyze 18 in the presence ofpalladium hydroxide (Pearlmans catalyst) have re-mained unsuccessful, despite several attempts to varythe pressure, the solvent, and the temperature. TheN-dealkylation of tertiary amines is known to be possibleby the action of vinyl chloroformate (Voc-Cl).7 The addi-tion of 2 equiv of vinylchloroformate to 18 in dichlo-romethane in the presence of 2 equiv of pyridine providescarbamate 19 in 49% yield (Scheme 3). The unprotectedpyrrolidine 20 is then obtained in 41% yield by refluxing19 in a methanol solution of hydrochloric acid for 4 h.The final coupling between 20 and the known8 R-(2-methoxyphenyl)acrylic acid has been accomplished throughits chloride 21 in standard conditions.9 The low 26% yieldin the target amide 2 corresponds to an nonoptimizedreaction.</p><p>In conclusion, this study of route A indicates that the[4 + 2] cycloaddition between PIBF 4 and pyrroline 5a</p><p>provides the expected tetracyclic adducts in modest yieldsbut in a fully diastereoselective way. The following func-tionalization steps have been achieved to connect theresulting adduct to the lateral chain determined from apreliminary structure-activity relationship study.2 Thesamples thus obtained have been evaluated for their invitro activities as farnesyltransferase inhibitors (seebelow).10</p><p>Route B: Synthesis of Analogue 3</p><p>The most convergent B-type approach to the key-synthon 6 requires methylpropiolate and PIBF 4 (Figure4). Unfortunately, no reaction was observed betweenthese compounds (or propiolic acid and PIBF) at roomtemperature, while a complex mixture of products, fromwhich no cycloadduct could be identified, was recoveredafter 4 h 30 min at 100 C in toluene. We have thereforetried to prepare 6 through the oxidation of the corre-sponding R-selenoesters 8, accessible through routes B1or B2 (Figure 2).</p><p>We first considered the B1 approach by reacting freshlyprepared 4 with methyl acrylate at room temperature for3 h (Scheme 4). The adduct 23a is obtained as a mixtureof isomers in an overall 48% yield. Examples taken fromthe literature suggest that a mixture of regioisomers wasto be expected, even if mainly the ortho substitutionpattern could be anticipated.11 A careful NMR analysisled us to the conclusion that only the ortho regioisomer23a represented had been obtained as a 38:62 mixtureof endo and exo isomers.</p><p>(7) (a) Olofson, R. A.; Yamamoto, Y. S.; Wancowicz, D. J. Tetrahe-dron Lett. 1977, 1563. (b) Olofson, R. A.; Schnur, R. C.; Bunes, L.; Pepe,J. P. Tetrahedron Lett. 1977, 1567.</p><p>(8) Giraud, E.; Luttmann, C.; Lavelle, F.; Riou, J. F.; Mailliet, P.;Laoui, A. J. Med. Chem. 2000, 43, 1807.</p><p>(9) Only a slight excess of thionyl chloride is to be used to convertthe acid 25 into its chloride 26, since a conjugated addition of chlorideonto the acrylate double bond can occur.</p><p>(10) Martin, C.; Pichon, N.; Harrison-Marchand, A.; Maddaluno, J.;Mailliet, P., to be published.</p><p>Figure 3. Endo and exo approaches of phenylisobenzofuran21 by pyroline 3a.</p><p>Scheme 3</p><p>Scheme 4</p><p>Figure 4. Routes between phenylisobenzofuran 21 and ester22.</p><p>Synthesis of Analogues of RPR 115135 J. Org. Chem., Vol. 66, No. 11, 2001 3799</p></li><li><p>A peculiar coupling pattern between their protons H2and proton H1 is worth underlining. Molecular models(as well as the AM1 optimization of 23a endo, Figure 5)indeed show that the bridgehead proton H1 lies in a planemore or less perpendicular to the H2 syn to the bridgingoxygen (H2), leading to their almost null couplingconstant.12 This topological peculiarity can be confusingsince a superficial analysis of the NMR spectrum coulderroneously suggest that the meta regioisomer is ob-tained selectively. The preferred ortho regioselectivitycan be qualitatively rationalized on the basis of the AM1analysis of the HOMO coefficients of 4 (Table 1), albeitonly very small differences are obtained at this level ofcalculation for the reagent alone and in its steady state.Regarding the tendency to an exo-favored addition, it isoften reported for isobenzofurans13 but remains difficultto explain. The attribution of the endo and exo structuresto the isomers is based on the analysis of the NMRcoupling constants. H2 is indeed identified through its</p><p>coupling with H1, while the amplitude of its couplingconstant with H3 determines the syn or anti characterbetween H3 and H2, thus the exo or endo character of23a.</p><p>The possible influence of steric factors on both theregio- and endo/exo selectivities of this reaction has beenconsidered. The access to other regioisomers of 3 wasindeed attractive to extend the structure/activity rela-tionship study in this series, and reversing the regiose-lectivity of the above cycloaddition seemed possibleconsidering the low differences between orbitalar coef-ficients in the HOMO of 4 (Table 1). To increase thedienophile bulkiness, we resorted to tert-butyl acrylate.Its cycloaddition with PIBF takes place at reflux ofTHF in 2 h and provides adduct 23b in 73% yield.14The regioselectivity remains unaltered while the endoselectivity is slightly enhanced (endo/exo ) 57:43, Scheme4). Mellor and Webb have reported a comparable im-provement of the endo preference with the increase ofthe steric hindrance of the acrylates.15 The ease ofapproach of the isobenzofuran nucleus has been alsomodified replacing the phenyl group of 4 by an o-tolylappendage. The o-tolylisobenzofuran 22 has been pre-pared following the procedure described above (Scheme2). Its reaction with methyl acrylate takes place at roomtemperature and provides adduct 24 in 49% yield after3 h. The regioselectivity is once more unchanged and theendo/exo ratio is 44:56 (Scheme 4). This set of experi-ments tends to indicate that the regio- and the endose-lectivity can hardly be influenced by the choice of thepartners.</p><p>The oxidation step needed to set the double bond in 6has been achieved through deprotonation and selenationof 23. As reported previously,16 KHMDS is well suited tothis task, provided diphenyl diselenide is added to themediu...</p></li></ul>