The Use of Polymer-Assisted Solution-Phase Synthesis and Automation for the High-Throughput Preparation of Biologically Active Compounds

STEVEN V. LEY, MARK LADLOW AND EMMA VICKERSTAFFE

1 Introduction

In recent years, the drug discovery process has been revolutionised by progress in the areas of proteomics and genomics, together with advances in high-throughput screening (HTS) of compounds for activity in various biological assays. This in turn has created a much higher demand for the rapid production of novel and functionally diverse compounds, thereby driving chemists to look for new ways to simplify, expedite and automate the process of organic synthesis. The importance of synthe-sising high-quality arrays of discrete compounds, from which meaningful structure activity relationships can be derived and also act as assets for future screening protocols is now well recognised.

The origins of high-throughput organic chemistry can be traced back to the work of Merrifield who pioneered solid-phase peptide synthesis. The development of this approach enabled the subsequent automation of peptide synthesis. Early attempts at high-throughput chemistry utilised this strategy, exploiting the advantages of solid-phase chemistry, which facilitates reaction work-up and rapid sample processing.

Today, although solid-phase organic synthesis (SPOS) remains a powerful technique for some aspects of high-throughput chemistry, there are a number of limitations that restrict its application. For example, while in some instances, by-product formation on the resin can be monitored without cleavage from the resin, separation of these materials from the desired product is not feasible until the end of the reaction sequence. The consequences are often cumulative, resulting in a complex final purification step. Additionally, even though a number of well-established methods for monitoring resin-bound intermediates have been developed, accurate quantitative measurement of immobilised material still represents a major challenge. Techniques such as magic angle spinning (MAS)-NMR and IR enable the analysis of molecules covalently bound to the resin, whereas MS techniques or the use of analytical constructs, require that the analyte be cleaved from the resin. The development of new chemistry on the solid support is therefore difficult and often protracted, requiring extended reaction rehearsal and optimisation.

Although SPOS is well suited to the synthesis of large, but relatively simple compound libraries, typically using combinatorial methods, the time required for route development and limitations as to the synthetic transformations that can be reliably performed is often restrictive. Increasingly, therefore, higher quality, smaller arrays are being synthesised by employing a much wider range of precedented solution-phase chemistries. Historically, attempts to increase the throughput of solution-phase chemistries have been confounded by the need for extensive work-up and purification procedures. An increasingly popular approach to circumvent many of these drawbacks involves the use of supported reagents." In this way, the advantages of performing chemistry in solution, namely, straightforward reaction monitoring and optimisation, are maintained. Excess reagents and their associated by-products may be readily separated from the desired solution-phase reaction product by simple filtration. A wide variety of reagents and scavenger resins have been developed in recent years to extend the scope of this new paradigm.

The concept of a reagent being immobilised on a solid-support was first exploited in catalytic applications as early as 1946. However, it was not until the end of the twentieth century that solid-supported nucleophiles and electrophiles immobilised on a polystyrene (PS) support were applied to the rapid purification of compounds prepared using standard solution-phase procedures. This application and its subsequent extension to the immobilisation of reagents has stimulated an explosion in the number of publications describing the development of novel polymer-bound reagents, catalysts and scavengers. As an ever-increasing number of these reagents becomes commercially available, interest in applying polymer-assisted solution-phase (PASP) techniques within industrial settings escalates.

PASP synthesis can be divided into two main approaches; these being (a) the use of supported reagents and scavengers, and (b) the adoption of a catch-and-release strategy (Figure 1). Both of these techniques allow the production of clean products, without the need to resort to traditional purification techniques, such as column chromatography. In an ideal case, the incubation of a substrate with a supported reagent causes its complete transformation into a new chemical entity, with any excess or spent reagent being removed by a simple filtration. This process circumvents the need for further purification and allows reactions to be driven to completion through the use of reagent excesses. Furthermore, reaction progress can conveniently be monitored by standard solution-phase techniques (TLC, LC/MS etc.), thereby minimising the time required to optimise a transformation. However in reality, many reactions are not totally selective and by-products are often formed. In many cases, these can be removed by incubation with an appropriate scavenger resin, which targets a specific functional group, unique to these by-products. For example, in Figure 1a, following condensation of the amine and carboxylic acid in the presence of a polymer-supported carbodiimide, remaining excess amine is removed by attachment to the strongly acidic polymer-supported sulfonic acid resin, which in turn affords the desired carboxamide product in high purity.

Conversely, purification can sometimes be conveniently performed by selectively capturing the desired reaction product using a suitable resin, washing away the byproducts, and then releasing the immobilised product back into solution. This is typically referred to as solid-phase extraction (SPE) and is particularly useful when the target molecule contains acidic or basic functionality.

An alternative approach utilises the concept of catch-and-release whereby one of the reactants is first attached to the solid phase by an activated bond that is subsequently cleaved by exposure to a conjugate reactant. For example, in Figure 1b, the carboxylic acid is initially immobilised as a tetrafluorophenyl ester. Although intrinsically stable, the activated ester is highly susceptible to nucleophilic cleavage, and in the presence of a substoichiometric amount of amine, the desired carboxamide product is cleanly released into solution.

Immobilisation also allows the combination of two or more otherwise mutually incompatible reagents to be used in the same pot. This was elegantly demonstrated by Parlow, who, in an important early contribution, utilised a combination of three different polymeric reagents simultaneously in a single reaction vessel to perform a three-step synthesis of pyrazoles. Notably, the yield of the pyrazole product 1 was greater when the supported reagents were combined in one flask as opposed to being used sequentially (Scheme 1).

In addition, supported reagents have been demonstrated to be effective under reaction conditions when either thermal or microwave heating is employed. They have also been utilised in traditional batch synthesis, stop-flow methods and continuous flow processes. However, one caveat is that the immobilisation of reagents can change their reactivity. For example, polymer-supported borohydride selectively reduces α,β-unsaturated carbonyl compounds to the α,β-unsaturated alcohol in contrast to the behaviour of the solution-phase counterpart, which additionally causes double bond reduction.

To date, immobilised reagents have mainly been PS based. PS supports have the advantages of being inexpensive, relatively easy to handle, have reasonably high loadings and, additionally, there is a vast amount of literature precedent available for their use. Reagents can either be covalently attached to the polymer, as in the polymer-supported Schwesinger base, 2-tert-butylimino-2-diethylamino- 1,3-dimethylperhydro-1,3,2 diazaphosphorin (PS-BEMP) 2, or electrostatically bound, for example, in the macroporous polymer-supported tetraalkylammonium triace-toxyborohydride 3 (Figure 2). Macroporous resins contain a higher degree of cross-linking than PS supports. In practice, this means that in contrast to microporous PS supports, the resin-bound functional groups of a macroporous resin come into contact with reagents by diffusion through the network of pores and therefore do not require the use of a solvent that will normally swell the resin. More recently, catalysts have been microencapsulated within a polymeric matrix.' However, issues with the degradation of resins under certain reaction conditions have led to the development of alternative supports such as controlled pore glass, monoliths, cellulose, zeolites and silicas.

2 PASP Synthesis Approaches to Biologically Active Compounds

2.1 Applications to the Synthesis of Commercial Drug Molecules

A number of syntheses targeting commercially available drugs have been reported, which demonstrate the utility and effectiveness of supported reagents for the rapid and efficient preparation of drug-like scaffolds. The introduction of Sildenafil 4 for the treatment of male erectile dysfunction has been incredibly successful, resulting in it becoming one of the largest selling globally marketed prescription drugs. Sildenafil acts by inhibiting the phosphodiesterase enzyme PDE5, which is the main phosphodiesterase present in the smooth muscle of the corpus cavernosum. Upon sexual stimulus, nitric oxide is released from nerve terminals in the corpus cavernosum. The nitric oxide activates guanylate cyclase to produce cyclic guanosine monophosphate (cGMP), causing the intracellular levels of cGMP within the smooth muscle cells of the penis to increase. In healthy tissue, the elevated cGMP is returned to basal levels by the action of the PDE5 enzyme. Inhibition of the PDE5 enzyme prevents the breakdown of cGMP and thus potentiates the smooth muscle relaxation. This increases the blood flow in the cavernosum causing an erection.

The polymer-assisted synthesis to Sildenafil (Scheme 2) follows a precedented route,' which concludes with the convergent coupling of the two key fragments 5 and 6. These fragments were prepared using PASP techniques, without the need for column chromatography, the former in a two-step sequence and the latter utilising seven different transformations. Fragment 5 was found to be contaminated with approximately 10% of the bis-esterified material 7, which could be removed in the subsequent catch-and-release step. Notably by using a catch-and-release strategy in the penultimate amide-coupling step, this transformation acts as an in-line purification step while concomitantly activating the acid group to nucleophilic attack. Introduction of 6, followed by scavenging with isocyanate resin to remove any unre-acted amine, cleanly yielded the amide. Assembly of the pyrimidine ring system is performed using microwave heating to effect the rapid dehydration of 8. A simple removal of the water formed during the cyclisation step was achieved with MgSO4 and the subsequent evaporation of the solvent gave a quantitative yield of Sildenafil 4. This convergent synthetic pathway is clearly amenable to library production.

Rosiglitazone 9, an agonist of peroxisome proliferator activated receptor-γ (PPARγ), is a recently introduced antihyperglycemic thiazolidinedione effective in the treatment of noninsulin dependant diabetes mellitus (type II diabetes). A seven-step synthesis was developed (Scheme 3), which utilised supported reagents in combination with in-line SPE purifications. The introduction of the pyridine moiety provided a convenient molecular handle by which to purify the molecule throughout the synthesis. Notably, the overall yield (46%) for this synthesis was higher than the yield reported in the initial shorter solution-phase synthesis (31%).

The synthesis of chiral drug molecules has only recently been reported using a supported reagent approach. Although many supported reagents systems have been reported for use in enantioselective reactions, the paucity of the enantioselectivity achieved has prevented their extensive application in synthesis. A single isomer of the drug Salmeterol 10, a potent and long-acting β22 adrenoceptor agonist, has been prepared via a chiral auxiliary approach (Scheme 4). The key chiral reduction of ketone 11 to alcohol 12 is dependent upon the introduction of the (S)-phenylglycinol functionality. Treatment of 11 with calcium chloride at 0 °C followed by the addition of PS-borohydride resin delivered the desired amino alcohol 12 as a 10:1 mixture of diastereoisomers favouring the desired (R)-alcohol. A single recrystallisation afforded diastereomerically pure material. It is hypothesised on the basis of 1H NMR shift studies that the reduction proceeds via a chelated intermediate 13 where the phenyl substituent points away from the crowded centre. The approach of the reducing agent then occurs preferentially from the convex face of the complex. From here, introduction of the long liphophilic chain and selective removal of the chiral auxiliary and ace-tonide-protecting group affords (R)-Salmeterol 10 in >97% ee.

2.2 Applications of PASP to the Synthesis of Biologically Active Natural Products

Some of the most extensive applications of PASP strategies have been directed towards the synthesis of biologically active natural products. The total synthesis of the cytotoxic antitumour natural product Epothilone C 14 is a tour de force, the synthetic strategy demanding the exploitation and development of new immobilised reagent methods, to meet the goal of a chromatography free synthesis. The route to the 16-membered macrocycle Epothilone C, followed similar published strategies involving coupling of three core fragments via a stereoselective C6–C7 aldol reaction, prior to either C1–C15 macrolactonisation or C12–C13 ring closing metathesis (Figure 3). The target molecule was prepared in high overall yield in 29 steps, with the longest sequence of linear steps being just 17. Impressively, considering the size, stereogenic centres and complexity of the molecule, only a single short column chromatography purification step was necessary at the end of the sequence to remove residual impurities and small quantities of the minor diastereoisomers that had been carried through the synthesis. Moreover, routes to fragments 1 and 3 are reported where the stereogenic centres are installed during the synthesis, via Oppolzer's sultam chiral auxiliary approach or an asymmetric Mukaiyama aldol reaction for fragment 1 and an asymmetric Brown allylation for fragment 3. However, the optimum synthesis to fragment 2 and ultimately fragment 3 relied on the use of chiral-starting materials to generate the desired stereochemistry.

The use of supported reagents has also been applied to the total synthesis of natural products, which have not previously been synthesised by traditional methods. An example of this is the first total synthesis of (+)-plicamine 15 and its unnatural enantiomer (Figure 4). (+)-Plicamine is a member of the amaryllidaceae alkaloids, which exhibit antitumour, immunosuppresive and analgesic activity and have recently found application in the therapeutic treatment of Alzheimer's disease. Other examples of biologically active natural products synthesised using supported reagents include the alkaloids ([+ or -])-oxomaritidine 16 and ([+ or -])-epimaritidine 17, and the potent analgesic ([+ or -])-epibatidine 18 isolated from the Ecuadorian poison frog Epipedobates tricolor.