Fully automated multi-step solution phase synthesis using polymer supported reagents: preparation of histone deacetylase inhibitors.

Article abstract of DOI:10.1039/b305713e

The first fully automated multi-step polymer assisted solution phase (PASP) synthesis is described. An array of histone deacetylase (HDAc) inhibitors was prepared by an unattended 4-5 step sequence incorporating in-line 'catch and release' purification.

Full text of DOI:10.1039/b305713e

View Article Online / Journal Homepage / Table of Contents for this issue
Fully automated multi-step solution phase synthesis using polymer
supported reagents: preparation of histone deacetylase inhibitors
a
a
a
b
Emma Vickerstaffe, Brian H. Warrington, Mark Ladlow* and Steven V. Ley*
a
GlaxoSmithKline Cambridge Technology Centre, University of Cambridge, Lensfield Road,
Cambridge, UK CB2 1EW. E-mail: Mark.2.Ladlow@gsk.com; Fax: ϩ44 (0) 1223 331532;
Tel: ϩ44 (0) 1223 336349
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge,
UK CB2 1EW. E-mail: svl1000@cam.ac.uk; Fax: ϩ44 (0) 1223 336442; Tel: ϩ44 (0) 1223
b
3
36398
Received 22nd May 2003, Accepted 18th June 2003
First published as an Advance Article on the web 20th June 2003
The first fully automated multi-step polymer assisted
solution phase (PASP) synthesis is described. An array of
histone deacetylase (HDAc) inhibitors was prepared by an
unattended 4–5 step sequence incorporating in-line ‘catch
and release’ purification.
reactions to completion, and facile reaction monitoring
in real time using conventional analytical methods, are
exploited.
However, whereas automation of solid phase chemistry is
already well established, to the best of our knowledge full
automation of multi-step solution phase synthesis has not
5
In recent years, the high-throughput screening of synthetic
compound libraries has emerged as a key strategy within the
pharmaceutical industry as a means of identifying new starting
been reported. Nevertheless, polymer assisted solution phase
(PASP) methods are intrinsically well suited to automation, in
that they typically utilise an iterative series of incubation
and filtration steps. Therefore, fully automated PASP synthesis
should be achievable. In particular, when the multi-step
solution phase synthesis can be performed in a single solvent,
full automation is feasible using existing robotic platforms.
Here, we describe the first fully automated multi-step PASP
synthesis of an array of histone deacetylase inhibitors (HDAc),
1
points for medicinal chemistry programs. This in turn has
created an increased demand for novel compounds, and has
led to the development of strategies to simplify, expedite and
2
automate the process of organic synthesis.
Solid phase organic synthesis is currently widely used to
implement high-throughput chemistry in an attempt to meet
3
6
these demands. However, this approach often suffers from
based upon the reported lead structure 1. HDAc inhibitors
unacceptably long development times, together with the
inability to separate by-products prior to cleavage from the
resin. This has led to a re-evaluation of solution phase
approaches with the objective of introducing strategies and
methods that lead to higher throughput. In particular, the use
of polymer supported reagents and scavenger resins in library
synthesis has been widely adopted. In this way, the inherent
advantages of both solid and solution phase methods, namely
simplified work up, the use of reagent excesses to drive
play a key biological role in chromatin remodelling and in the
regulation of gene transcription. As such, they have consider-
7
able potential as the basis for new therapeutic approaches in
8
9
the treatment of cancer, and as novel anti-protozoal and
anti-viral agents.
10
Initially, we devised a PASP synthesis leading to the
hydroxamic acid 1 based upon the reported solution phase
synthesis but incorporating changes to facilitate automation
(Scheme 1). The synthesis was first developed and performed
4
Scheme 1 Reagents and conditions: a) DMF, 40 ЊC, 1 h, ×2; b) DMF, rt, 2 h; c) DMF, RX, rt, 18 h; d) acrylic acid, tributylamine, DMF, 90 ЊC, 18 h;
e) DIPEA, NH OTHP, DMF, 50 ЊC, 18 h; f ) MeOH, DMF, rt, 3 h.
2
T h i s j o u r n a l i s © T h e R o y a l S o c i e t y o f C h e m i s t r y 2 0 0 3
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 1 9 – 2 4 2 2
2419

View Article Online
manually, and utilised immobilised reagents for each step
chosen such that the entire sequence could be performed in a
single solvent (DMF), thereby removing the additional
complexity associated with the need for solvent interchanges. In
addition, to avoid the need for conventional chromatographic
purification of intermediates, in-line purifications were
incorporated utilising a combination of scavenger resins and a
Heck olefination of 3 with acrylic acid to afford 6 (R = H)
was investigated using a variety of immobilised palladium
catalysts to facilitate product work-up and these reactions were
profiled using the Reactarray SK233 automated reaction
15
sampling system (Fig. 1). In this way, we determined that the
competing dehalogenation pathway leading to 4 was minimised
16
using microencapsulated Pd(OAc)2 (Pd EnCat) as the
source of palladium and tributylamine as base. At this stage,
although it was possible to remove the palladium catalyst
by simple filtration, a number of contaminants remained in
solution. Bearing in mind the subsequent synthetic steps, in
particular the need to activate the acid 6 prior to hydroxamate
formation, the presence of tributylamine was acceptable.
Further, the presence of any unreacted acrylic acid could be
minimised by limiting the amount introduced at the Heck
olefination stage. However, the dehalogenated contaminant
4 was not removed by filtration of the reaction mixture. A
much better work-up procedure, therefore, was to utilise the
carboxylic acid functionality present in the desired product 6 in
an in-line ‘catch and release’ purification step. This was made
even more attractive by the possibility of effecting concomitant
activation of the acid functionality. This was achieved by
incubation of the supernatant from the Heck reaction with
11
‘catch and release’ strategy.
The new synthesis also benefits from other advantages
associated with the use of polymer supported reagents. For
example, when sulfonylation of the aniline 2 is performed
conventionally in solution in the presence of a sulfonyl chloride,
a mixture of mono- and bis-sulfonamides is typically obtained.
Therefore, this transformation is usually performed with a
sufficient excess of the sulfonyl chloride to yield the bis-
sulfonamide exclusively which is subsequently hydrolysed to the
mono-sulfonamide in a separate step. However, we quickly
established that the mono-sulfonamide 3 could be prepared
directly by treating the aniline 2 with the sulfonyl transfer
reagent derived from the immobilisation of an appropriate
aryl sulfonyl chloride on polymer-supported dialkylamino-
12
pyridine in DMF. Any unreacted aniline was then readily
removed following incubation with the sulfonic acid
ion-exchange resin Amberlyst H-15 to afford 3 in high purity
2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium
hexa-
13
14
18
(
>98% by LC-MS ) and acceptable yield (66%).
fluoroantimonate on polystyrene (PS-HBTU) resin to gener-
ate the resin-bound activated HOBt ester of 6 directly without
the need for any other additional reagents. This resin was then
washed free of contaminants before treatment with a solution
of O-THP hydroxylamine to release the hydroxamate 7 into
solution. Exposure of 7 to Amberlyst H-15 effected final
O-deprotection of the tetrahydrofuranyl ether to afford the
desired hydroxamic acid 1. Although 1 was obtained directly
in good purity (87% by LC-MS), on this occasion further
purification was performed by HPLC to afford the hydroxamic
19
acid 1 in 33% overall yield (purity >99% by LC-MS).
At this stage we recognised that an extra point of diversity
could be incorporated into the synthesis by N-alkylation of the
intermediate sulfonamide 3 prior to Heck olefination to afford
5
. In practise, this could be reliably achieved in the presence of
20
the polymer-supported Schwesinger base PS-BEMP to afford
(R = Me) in 90% yield and >98% purity by LC-MS. The use
5
of an excess of less volatile alkylating agents is also acceptable
because any residual alkylating agent is effectively scavenged in
the following Heck reaction by conversion to the quaternary
ammonium salt of tributylamine at 90 ЊC in DMF.
Our attention next turned to full automation of the array
synthesis. A number of factors influenced our choice of robotic
Fig. 1 Reactions profiles for the Heck olefination of 3 in the presence
1
6,17
of immobilised palladium catalysts
leading to 4 and 6.
1
2
3
Fig. 2 Flowchart for auto-PASP synthesis of array 8 {R , R , R }.
2
420
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 1 9 – 2 4 2 2

View Article Online
1
2
3
Table 1 Monomers used to prepare the array of HDAc inhibitors
inhibitors 8 {R , R , R } incorporating the monomers shown in
Table 1. The robot was pre-loaded with all the resins and
reagents necessary and the 180 reactions required were allowed
to run unattended over a 4 day period. Finally, the array 8 was
subjected to quality control analysis by LC-MS which con-
firmed that 34 of the 36 desired compounds had been obtained
and that these were the major products in each case. In general,
the compound purities were between 55–80% and 10–20 mg of
crude material was obtained in each case (Fig. 3). These results
compared well with those obtained for the automated synthesis
of the single hydroxamic acid 1. Following a single final
autopreparative purification step, all compounds were obtained
1
2
3
8
{R , R , R } by auto-PASP synthesis
1
in greater than 95% purity according to LC-MS and H NMR
analysis.
platform. Firstly, for convenience, we wished to utilise a
commercially available robotic platform. Secondly, a suitable
robotic synthesiser should be able to: (i) dispense solutions
of reagents into reaction vessels in a range of sizes, and (ii)
efficiently mix and filter concentrated resin slurries and transfer
the supernatant solutions between reaction vessels/blocks.
Fig. 3 Yield (%) and HPLC purities (area @ 254 nm) of HDAc array
1
2
3
8{R , R , R } prepared by auto-PASP synthesis.
Conclusions
Thirdly, for efficiency,
a multi-channel synthesiser was
preferred. A top filtration platform with 4 channels and high
speed vortexing mixing capability was therefore identified as
being most suitable for our purposes.
In summary, we have demonstrated that immobilised reagents
and scavenger resins can be used to perform a series of sequen-
tial transformations that may be executed in the same solvent.
The transformations required can be automated using a
commercially available robotic synthesiser to facilitate the
preparation of drug-like motifs. To exemplify this strategy, we
have completed the first unattended fully automated PASP
synthesis of an array of HDAc inhibitors 8. Moreover, the
synthesis was performed more quickly than the corresponding
manual process (3.5 days vs. >5 days). Importantly, this study
shows for the first time the utility of polymer assisted strategies
to facilitate full, unattended automation of solution phase
synthesis.
21
After preliminary studies to establish that the synthesiser
could handle concentrated resin slurries, we developed modular
protocols suitable for performing each synthetic transform-
ation. These involved either the addition of reagent solutions to
the various resins followed by incubation and vortexing, or the
filtration and washing of resins combined with dispensing of
the supernatant solutions to the next reaction block. Separate
reaction blocks were pre-loaded with all the reagent and
scavenger resins required for the complete automated synthesis.
Intermediate resin wash steps were performed with the
minimum amount of solvent to limit serial dilution of the
reaction mixtures during the linear stages of the synthesis.
The automated protocols were then combined in an iterative
manner to provide the fully automated sequence necessary for
the array synthesis shown in Fig. 2.
The automated synthesis was validated by the preparation
of the hydroxamic acid 1, collecting aliquots at each stage
and subjecting these to analysis by LC-MS. Reassuringly, the
reaction profile obtained was similar to that observed for the
corresponding manual synthesis. Importantly, it was possible to
run the complete synthesis on the robot without the need for
any manual intervention. The process required 3.5 days to run
and afforded 1 directly in good purity (76% by LC-MS) and
Acknowledgements
We would like to thank Tony Parkhouse (GlaxoSmithKline),
Adrian Pipe (GlaxoSmithKline) and Fiona Payne (Zinsser
Analytic) for their help with the automation, and GlaxoSmith-
Kline for generous financial support of this work.
Notes and references
1
(a) B. A. Kenny, M. Bushfield, D. J. Parry-Smith, S. Fogarty and
J. M. Treherne, Progress Drug Res., 1998, 51, 245; (b) A. Golebioski,
S. R. Klopfenstein and D. E. Portlock, Curr. Opin. Chem. Biol.,
2
001, 5, 273.
2
8% overall yield. With further optimisation of the individual
2
3
I. Hughes and D. Hunter, Curr. Med. Chem., 2001, 8, 985.
transformations it is anticipated that the overall run time could
be reduced even further.
The automated PASP protocol was then applied to the prep-
aration of a 36 member combinatorial array of potential HDAc
(a) J. Yoshida and K. Itami, Chem. Rev., 2002, 102, 3693; (b)
A. Ganesan, Drug Discovery Today, 2002, 7, 47; (c) F. Z. Dörwald,
Organic Synthesis on Solid Phase, Wiley-VCH, Weinheim, 2000; (d )
D. Obrecht, J. M. Villalgordo, Solid-Supported Combinatorial and
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 1 9 – 2 4 2 2
2421

View Article Online
Parallel Synthesis of Small-Molecular-Weight Compound Libraries,
Pergamon, Oxford, 1998 .
opportunity for failure of the entire synthesis. In this case, we found
it advantageous to routinely repeat the sulfonylation step twice in
separate reaction blocks (1 and 2, Fig. 2), prior to scavenging.
15 (a) C. Jamieson, M. S. Congreve, D. F. Emiabata-Smith and
S. V. Ley, Synlett, 2000, 11, 1603; (b) C. Jamieson, M. S. Congreve,
D. F. Emiabata-Smith, S. V. Ley and J. J. Scicinski, Org. Proc. Res.
Develop., 2002, 6, 823.
16 Pd EnCat; This reagent is available from Avecia
(www.avecia.com), cf: (a) C. Ramarao, S. V. Ley, I. M. Shirley and
N. J. DeAlmeida, Chem. Commun., 2002, 1132; (b) S. V. Ley,
C. Ramarao, R. S. Gordon, A. B. Holmes, A. G. Morrison,
I. F. McConvey, I. M. Shirley, S. C. Smith and M. D. Smith, Chem.
Commun., 2002, 1134.
4
(a) S. V. Ley, I. R. Baxendale, R. N. Bream, P. S. Jackson,
A. G. Leach, D. A. Longbottom, M. Nesi, J. S. Scott, R. I. Storer
and S. J. Taylor, J. Chem. Soc., Perkin Trans. 1, 2000, 3815; (b)
A. Kirschning, H. Monenschein and R. Wittenberg, Angew. Chem.,
Int. Ed. Engl., 2001, 40, 650; (c) D. H. Drewry, D. M. Coe and
S. Poon, Med. Res. Rev., 1999, 19, 97.
5
6
7
(a) Sugawara and D. G. Cork, Comb. Chem., 2000, 373; (b)
R. A. Tommasi, L. W. Whaley and H. R. Marepalli, J. Comb. Chem.,
2
000, 2, 447.
R. Lavoie, G. Bouchain, S. Frechette, S. H. Woo, E. A. Khalil,
S. Leit, M. Fournel, P. T. Yan, M. Trachy-Bourget, M. Beaulieu,
Z. Li and J. Besterman, Bioorg. Med. Chem. Lett., 2001, 11, 2847.
(a) C. M. Grozinger and S. L. Schreiber, Chem. Biol., 2002, 9, 3;
17 FibreCat 1001: 2.71 mol% Pd was supplied by Johnson Matthey
(www.chemicals.matthey.com).
(
1
b) C. A. Hassig and S. L. Schreiber, Curr. Opin. Chem. Biol., 1997,
, 300.
18 A pre-production trial batch of this polymer supported reagent
was kindly supplied by Argonaut Technologies Inc.
(www.argotech.com).
8
9
W.-G. Zhu and G. A. Otterson, Curr. Med. Chem.: Anti-Cancer
Agents, 2003, 3, 187–199.
P. T. Meinke and P. Liberator, Curr. Med. Chem., 2001, 8, 211.
0 E. Shestakova, M.-E. Bandu, J. Doly and E. Bonnefoy, J. Virol.,
19 Selected data for 1: Mp 233–234 ЊC; HRMS C H NO S requires
1
6
15
4
ϩ
Ϫ1
(MH) 318.0800, found 318.0809; νmax (cm ): 3255, 1674, 1627,
1322, 1155; δ (400 MHz, DMSO-d ): 12.40 (br s, 1H), 10.5 (br s,
1
1
1
H
6
2
001, 75, 3444.
1H), 7.65 (d, 2H, J = 8 Hz), 7.51 (d, 2H, J = 8 Hz), 7.42 (d, 1H, J = 16
1 L. A. Thompson, Curr. Opin. Chem. Biol., 2000, 4, 324; D. L. Flynn,
Med. Res. Rev., 1999, 19, 408.
2 S. V. Ley, M. H. Bolli, B. Hinzen, A.-G. Gervois and B. J. Hall,
J. Chem. Soc., Perkin Trans. 1, 1998, 2239.
Hz), 7.32 (d, 2H, J = 8 Hz), 7.08 (d, 2H, J = 8 Hz), 6.34 (d, 2H, J = 16
Hz), 2.30 (s, 3H); δC (100 MHz, DMSO-d ) 167.9, 143.9, 143.5,
6
140.0, 136.9, 130.2, 129.9, 129.7, 127.1, 119.5, 118.3, 21.3.
20 2-tert-Butylimino-2-diethylamino-1,3-dimethylperhydro-1,3,2-dia-
Ϫ1
1
1
3 HPLC purities were measured at 254 nm.
zaphosporine on polystyrene, 2.3 mmol g available from Fluka
4 In general, in automated synthesis, an operationally simpler
protocol may be preferred over a more complex sequence, even at
the expense of a slightly reduced yield, since it introduces less
(order No. 20026).
21 The Zinsser Sophas M6 robotic synthesiser (Zinsser Analytic,
www.zinsser-analytic.com) was used in this study.
2
422
O r g . B i o m o l . C h e m . , 2 0 0 3 , 1, 2 4 1 9 – 2 4 2 2