[Molecules: 8] [Related articles/posters: 043 114 050 032 115 ]

ECHET96 Article 109 Michael A. Walters

Solid-phase parallel synthesis applied to lead development: potent analogues of the GPIIb/IIIa antagonist RWJ-50042

William J. Hoekstra,* Bruce E. Maryanoff, Patricia Andrade-Gordon, Judith H. Cohen, Michael J. Costanzo, Bruce P. Damiano, Robert Falotico, Barbara J. Haertlein, Bruce D. Harris, Jack A. Kauffman, Patricia M. Keane, David F. McComsey, John A. Mitchell, Frank J. Villani, Jr. and Stephen C. Yabut

Drug Discovery and Chemical Development Department, The R. W. Johnson Pharmaceutical Research Institute, Spring House, Pennsylvania 19477, USA

Abstract

A series of b-turn peptide mimetics with a nipecotic acid heterocyclic scaffold was designed by NMR analysis of the C-terminal g-chain of fibrinogen to provide the lead GPIIb/IIIa (fibrinogen receptor) antagonist RWJ-50042 (1). We have employed solid-phase parallel synthesis for the preparation of over 200 analogues of this lead with a protocol of optimization cycles. This strategy produced some nipecotamide analogues with a 100-fold improvement in potency.

Combinatorial chemistry methods have unleashed a powerful strategy for diversity-based discovery of new leads.¹ Convergent, resin-based parallel synthesis of discrete chemical libraries is a focused adaptation of combinatorial chemistry that provides an attractive lead development tool for the refinement of biological activity.² Herein, we report on the solid-phase parallel synthesis of analogues of a prototypical fibrinogen receptor (GPIIb/IIIa) antagonist lead, RWJ-50042 (1). This approach resulted in a marked compression of the drug lead-to-clinical-candidate timeline vis-a-vis traditional solution-phase techniques.

RWJ-50042 is an orally active antagonist of the platelet fibrinogen receptor (binding IC₅₀ = 0.0045 uM; sustained ex vivo inhibition of collagen-induced platelet aggregation at 10 mg kg^-1 in dogs), which was discovered by using the solution structure of the C-terminal g-chain of fibrinogen as a basis for drug design.^3-7 This substituted nipecotic acid modelled the b-turn structure contained within the KQADG sequence of the g-chain (residues 406-410). Given the competitive environment surrounding this type of antithrombotic therapy,⁸ we pursued synthetic methodology to expedite the study of RWJ-50042 analogues.

Synthesis of analogues

Solid-phase parallel synthesis was employed to prepare analogues of RWJ-50042 quickly. Given that this lead molecule consists of two amide bonds/three structural components, variation of each of these elements represents a practical, systematic strategy toward potency improvement (Scheme 1). Since variants of the b-amino acid component are readily available as esters, commercially or synthetically, a dozen were selected initially for fibrinogen receptor antagonist synthesis. Our apparatus is arranged in a three-by-four array to effect production of twelve products for any given set of molecules; the N-terminal pseudodipeptide unit is held constant in a matrix. Importantly, we chose a strategy of N-terminal attachment to the resin to allow for coupling of the numerous b-amino esters relatively late in the synthesis, thereby timing the 12-way resin division immediately before the second amide bond coupling (Scheme 2). A corresponding C-terminal attachment strategy would have incorporated less-available N-protected b-amino acids and compelled resin splitting prior to the first amide bond-forming step.

Scheme 1 Strategy for preparation of a "virtual" library of 288 variants of RWJ-50042 (6 x 4 x 12)

Isonipecotic acid and five- and seven-membered-ring variants of nipecotic acid were chosen to study scaffold size/conformation change of the "central ring" in relation to the other two components. While 288 variants (6 x 4 x 12) were targeted in principle for synthesis, a concurrent refinement process was implemented to select relatively optimal components for subsequent analogue synthesis/evaluation. This process of using a "virtual" compound library discards inferior components to avoid the unnecessary synthesis and bioassay of weakly active agents.

Resin-based preparation of the 3-phenyl-3-aminopropionic acid analogue (15) of RWJ-50042 typifies our strategy of convergent, high-throughput synthesis. N-attachment of 2-chlorotrityl chloride resin to allyl 4-piperidinepropanoate furnished "N-protected" intermediate A. Allyl ester removal under mild, reproducible palladium(0) conditions and then DIC-mediated coupling to allyl piperidinepropanoate afforded pseudo-dipeptide B. Starting at the N-terminus allows one to accrue large quantities of resin-bound intermediate B in a common reaction vessel. Once B was saponified, the resin was split for coupling with twelve readily available b-amino esters, leading to final products. For modifications at the tertiary amide, analogous urethane and urea couplings at the N-nipecotyl position were performed by using standard p-nitrophenylchloroformate conditions⁹ with the appropriate resin-bound primary alcohol or primary amine (e.g., 2 and 3). Solution-phase coupling of CBZ-4-piperidineethanesulfonyl chloride with ethyl nipecotate gave the corresponding sulfonamide intermediate (e.g., 4).¹⁰ Intermediate B was saponified and coupled to methyl 3-phenyl-3-aminopropionate to render C. To isolate a variety of carboxylic acid targets from readily available methyl or ethyl b-amino esters, an organic solution method of KOSiMe₃/THF saponification was adapted to intermediates such as C. This method allows the resin to swell suitably to complete ester cleavage when basic aqueous conditions fail. Potassium carboxylates were then acidified with dilute AcOH and cleaved with CF₃CO₂H to give products, exemplified by 3-phenyl derivative 15.

Scheme 2

Racemic 3-substituted b-amino esters were purchased, or prepared by using a modified Knoevenagel procedure (RCHO/ammonium acetate/malonic acid).¹¹ Biological testing of racemic, matrix-produced targets revealed highly active antagonists of interest for enantiospecific solution-phase scale-up synthesis. Enantiomerically enriched 3-aryl-3-amino esters were synthesized by using (R)-1-phenylethylamine Michael addition to arylacrylates.¹² Lithium acetylide addition to 4-benzoyloxy-2-azetidinone followed by ring opening and chromatographic chiral resolution of the desired O-methylmandelamide derivative expedited preparation of acetylene intermediates.¹³ Racemic 3-aryl-3-aminopropanoic acids were also resolved as their phenylacetamide derivatives by using penicillin amidase (Scheme 3).¹⁴

Scheme 3

Scheme 4

A representative synthesis that uses the asymmetric Michael addition as the key step is shown for methylenedioxybenzene derivative 25 (Scheme 4). Conjugate addition of (R)-(+)-1-phenylethylamine to ethyl 3,4-methylenedioxybenzeneacrylate proceeds with >90% diastereomeric excess. Hydrogenolysis of the a-methylbenzyl group and HBTU-mediated coupling with Boc-(R)-(-)-nipecotic acid¹⁵ gives a Boc-nipecotamide intermediate which is then N-terminally deprotected and coupled iteratively with Boc-piperidinepropanoic acid. Purifications were performed on fully protected coupling products via chromatography on silica gel. Nipecotamide 25, for instance, was isolated in 17% overall yield.

Discussion

Compounds 1-25 are representative examples of more than 250 solid-phase synthesis products that were tested in vitro (Table 1). Systematic changes at the piperidine linker unit Y indicate that a tertiary amide at this site is preferred (e.g., 1). While 12 b-amino acid variants of urethane (2) or urea (3) linkers were prepared on the resin, sulfonamide (4), N-Me-piperidine (5), and piperazine (6) examples represent singular samplings from solution-phase synthesis. Since the preferred central ring turned out to be nipecotic acid (n = 1, 1), about 180 variants were prepared while 12 each of examples 7-9 were isolated and tested. In vitro testing of b-amino acid variants 12-22 indicated the possibility of activity improvement with 3-alkyl (13, 14), 3-aryl (15), or 2-oxy (18) substitutions. Aspartate methyl ester 16 is regarded as a "3-alkyl" case since its diacid derivative is inactive (IC50 > 25 uM). Testing of 29 alkyl, alkenyl, or alkynyl 3-substituted compounds gave antagonists with as much as fourfold potency improvement (e.g., phenylacetylene 23 and tert-butylacetylene 24, Table 2) over the active enantiomer of RWJ-50042 (i.e., 10). Large hydrophobic groups in this part of the molecule rendered the best activity; thus, numerous 3-aryl cases were prepared (120 compounds tested).

This parallel synthesis methodology rapidly produced greater than a dozen analogues of RWJ-50042, which were targeted for enantiospecific scale-up synthesis and subsequent in vivo evaluation (some are shown in Table 2). From an in vitro standpoint, some of the more promising compounds were the doubly racemic 3-(3,4-methylenedioxyphenyl) and 3-(3-quinoline) congeners. Indeed, enantiospecific synthesis of these fibrinogen receptor antagonists afforded the most potent compounds in this series (3S-enantiomers 25 and 28). 3-Substituted b-aminopropionic acid analogues with the R absolute configuration are only weakly active.

To address potential oral bioavailability limitations of our amino acid-like antagonists, compound 25 was "capped" at the N- or C-terminus. N-Methylpiperidine 26 and ethyl ester 27 exhibit inferior in vitro and in vivo characteristics relative to 25, however. The power of parallel synthesis played a decisive role in overcoming this bioavailability hurdle. Due to the ample selection of potent, solid-phase synthesis-derived compounds, analogues were identified with useful systemic availability (15-20%) without the need for prodrug modifications (oral canine studies).

Table 1 Inhibition of human platelet aggregation and fibrinogen binding by RWJ-50042 analogues (uM)

Pl. Agg. Bndg Pl. Aggr. Bndg

# Y Z IC₅₀* IC₅₀⁺ # n IC₅₀* IC₅₀⁺

1 COCH₂ CH 0.66 0.005 1 1 0.66 0.005

2 COO CH 4.7 0.027 7 0 7.0 0.004

3 CONH CH 6.7 0.016 8 2 7.6 0.013

4 SO₂CH₂ CH 11.0 0.025 9 1 >25 (isonipec.) >25

5 COCH₂ CH 2.4 0.006 10 1 0.34 (3R) 0.005

(N-Me-piperidine) 11 1 2.93 (3S) 0.004

6 COCH₂ N >25 >25

Pl. Agg. Bndg Pl. Agg. Bndg

# X IC₅₀* IC₅₀⁺ # X IC₅₀* IC₅₀⁺

1 b-Ala 0.66 0.005 17 2-Me-b-Ala >25 >25

12 N-Me-b-Ala 27.5 0.20 18 2-OH-b-Ala 0.85 0.005

13 3-Me-b-Ala 2.0 0.003 19 4-oxo-nipecotic acid >25 0.33

14 3-Buⁱ-b-Ala 4.1 0.0025 20 3-NH-c-C₆H₁₀-CO₂H >25 0.20

15 3-Ph-b-Ala 3.6 0.003 21 NH(CH₂)₂SO₃H 10.8 0.18

16 L-Asp-OMe 1.1 0.003 22 NH(CH₂)2-5-tetrazole 25.5 1.39

* Thrombin-induced gel-filtered platelet aggregation (uM, n = 3).³

⁺ Inhibition of biotinylated fibrinogen binding to immobilized GPIIb/IIIa (uM, n = 2).³

Table 2 In vitro data for three-substituted b-amino acid GPIIb/IIIa antagonists

Fg Binding⁺ Human GFP*

# R1 R2 R3 IC₅₀ (uM) IC₅₀ (uM)

10 H H H 0.005 0.34

23 H C C-Ph H 0.0002 0.080

24 H C C-Bu^t H 0.021 0.079

25 H 3,4-methylenedioxy-Ph H 0.0005 0.028

26 Me 3,4-methylenedioxy-Ph H 0.0002 0.84

27 H 3,4-methylenedioxy-Ph Et 0.046 36

28 H 3-quinoline H 0.0002 0.019

29 H 2-thiophene H 0.0002 0.090

Xemlofiban (SC-54684) 0.0006 0.31

* Thrombin-induced gel-filtered platelet aggregation (n = 3).³

⁺ Inhibition of biotinylated fibrinogen binding to immobilized GPIIb/IIIa (n = 2).³

Table 3 Canine ex vivo data for four GPIIb/IIIa antagonists

Dog PRP* Dog ex vivo platelet aggregation⁺

Compd IC₅₀ (uM) PO Dose Duration

10 0.41 3 mg kg^-1 120 min

24 0.45 10 mg kg^-1 90 min

25 0.015 3 mg kg^-1 >180 min

Xemlofiban 1.20 3 mg kg^-1 >180 min

* Collagen-induced platelet-rich plasma aggregation (n = 3).

⁺ Oral dose required to inhibit collagen-induced canine ex vivo platelet aggregation at least 50% (3 dogs).

Conclusion

Solid-phase parallel synthesis facilitated lead improvement/development of the GPIIb/IIIa antagonists in the following manner. First, it produced numerous potent analogues, in vitro and in vivo, of RWJ-50042 quickly. Classical solution-phase synthesis would have, in all probability, resulted in a similar quality of improvement, but not in the time period exhibited here. Second, the number of improved antagonists yielded by resin-based synthesis allowed for study of other critical in vivo properties such as consistency of oral absorption, plasma halflife, duration of action, etc. Overall, the timeline for lead discovery-to-development candidate selection within the series was compressed relative to a traditional medicinal chemistry approach.

Acknowledgment

We thank Dr Michael N. Greco for advice on the solid-phase synthesis.

References

(a) Gordon, E. M.; Barrett, R. W.; Dower, W. J.; Fodor, S. P. A.; Gallop, M. A. J. Med. Chem. 1994, 37, 1385; (b) Thompson, L. A.; Ellman, J. A. Chem. Rev. 1996, 96, 555.
DeWitt, S. H.; Kiely, J. S.; Stankovic, C. J.; Schroeder, M. C.; Cody, D. M. R.; Pavia, M. R. Proc. Natl. Acad. Sci. USA 1993, 90, 6909.
Hoekstra, W. J.; Beavers, M. P.; Andrade-Gordon, P.; Evangelisto, M. F.; Keane, P. M.; Press, J. B.; Tomko, K. A.; Fan, F.; Kloczewiak, M.; Mayo, K. H.; Durkin, K. A.; Liotta, D. C. J. Med. Chem. 1995, 38, 1582.
Fan, F.; Mayo, K. H. J. Biol. Chem. 1995, 270, 24693.
Mayo, K. H.; Fan, F.; Beavers, M. P.; Eckardt, A.; Keane, P.; Hoekstra, W. J.; Andrade-Gordon, P. FEBS Lett. 1996, 378, 79.
Mayo, K. H.; Fan, F.; Beavers, M. P.; Eckardt, A.; Keane, P.; Hoekstra, W. J.; Andrade-Gordon, P. Biochemistry 1996, 35, 4434.
Hoekstra, W. J.; Press, J. B.; Bonner, M. P.; Andrade-Gordon, P.; Keane, P. M.; Durkin, K. A.; Liotta, D. C. Bioorg. Med. Chem. Lett. 1994, 4, 1361.
For a recent review, see: Zablocki, J. A.; Rao, S. N.; Baron, D. A.; Flynn, D. L.; Nicholson, N. S.; Feigen, L. P. Curr. Pharm. Design 1995, 1, 533.
Hutchins, S. M.; Chapman, K. T. Tetrahedron Lett. 1994, 35, 4055.
(a) DeGaw, J. I.; Kennedy, J. G. J. Heterocycl. Chem. 1966, 3, 90; (b) Hearst, P. J.; Noller, C. R. Org. Synthesis 1950, 30, 58.
Profft, E.; Becker, F. J. J. Prakt. Chem. 1965, 30, 18.
Rico, J. G.; Lindmark, R. J.; Rogers, T. E.; Bovy, P. R. J. Org. Chem. 1993, 58, 7948.
Zablocki, J. A.; Rico, J. G.; Garland, R. B.; Rogers, T. E.; Williams, K.; Schretzmann, L. A.; Rao, S. A.; Bovy, P. R.; Tjoeng, F. S.; Lindmark, R. J.; Toth, M. V.; Zupec, M. E.; McMackins, D. E.; Adams, S. P.; Miyano, M.; Markos, C. S.; Milton, M. N.; Paulson, S.; Herin, M.; Jacqmin, P.; Nicholson, N. S.; Panzer-Knodle, S. G.; Haas, N. F.; Page, J. D.; Szalony, J. A.; Taite, B. B.; Salyers, A. K.; King, L. W.; Campion, J. G.; Feigen, L. P. J. Med. Chem. 1995, 38, 2378.
Soloshonok, V. A.; Fokina, N. A.; Rybakova, A. V.; Shishkina, I. P.; Galushko, S. V.; Sorochinsky, A. E.; Kukhar, V. P.; Savchenko, M. V.; Svedas, V. K. Tetrahedron: Asymmetry 1995, 6, 1601.
Although (R)-(-)-nipecotamides were identified early on as more potent antagonists than their epimers,³ racemic allyl nipecotate was employed in high-throughput matrix synthesis. Boc-(R)-(-)-nipecotic acid was prepared in two steps from (R)-(-)-nipecotic acid ethyl ester (Chemi S.p.A.).