Design of an organic sequence suitable for the solid phase combinatorial synthesis of libraries of macro-heterocycles

Article abstract of DOI:10.1016/S0040-4039(00)00710-3

The solid as well as solution phase synthesis of two 14-membered macro- heterocycles is reported. The macrocycles are built out of four or three major building blocks and contain serine, a trans alkene and the malonate unit inside the framework. The macrocyclization was an SN2 reaction between a nucleophilic malonate anion and an allylic chloride leaving group. The first macrocycle, an ether, was obtained with an overall yield of 8% for the eleven steps on solid support similar to the 11% yield in solution. The overall yield for the second target macrolactone was 45% on solid phase for the six steps and was higher than in solution (35%). The stability of the DHP linker to various reagents and reactions conditions is noteworthy. (C) 2000 Published by Elsevier Science Ltd.

Full text of DOI:10.1016/S0040-4039(00)00710-3

Tetrahedron Letters 41 (2000) 4743±4749
Design of an organic sequence suitable for the solid phase
combinatorial synthesis of libraries of macro-heterocycles
Mahesh Ramaseshan,a John W. Ellingboe,b Yves L. Dorya,*
and Pierre Deslongchampsa,*
a
Á
Â
Laboratoire de Synthese Organique, Departement de Chimie. Institut de Pharmacologie de Sherbrooke,
e
Â
Â
Universite de Sherbrooke, 3001, 12 Avenue Nord, Sherbrooke, Quebec, J1H 5N4, Canada
bWyeth-Ayerst Research, Division of Chemical Sciences, Pearl River, NY 10965, USA
Received 26 January 2000; revised 17 April 2000; accepted 18 April 2000
Abstract
The solid as well as solution phase synthesis of two 14-membered macro-heterocycles is reported. The
macrocycles are built out of four or three major building blocks and contain serine, a trans alkene and the
malonate unit inside the framework. The macrocyclization was an SN2 reaction between a nucleophilic
malonate anion and an allylic chloride leaving group. The ®rst macrocycle, an ether, was obtained with an
overall yield of 8% for the eleven steps on solid support similar to the 11% yield in solution. The overall
yield for the second target macrolactone was 45% on solid phase for the six steps and was higher than in
solution (35%). The stability of the DHP linker to various reagents and reactions conditions is noteworthy.
# 2000 Published by Elsevier Science Ltd.
Introduction. Whilst searching for an ecient way to build macrocycles from several building
blocks on solid support, we have been studying tyrosine as a ®rst building block.1 According to
the chemistry sequences developed then, the trifunctionalized amino acid tyrosine was attached to
the resin through its acid part. However, in the name of molecular diversity it is desirable to use
other groups as well for that purpose. In this paper we wish to report a new strategy where the
®rst building block is attached to the resin by means of an alcohol. Such stratagem allowed us to
carry out new chemistry to increase the bank of reactions compatible with the construction of large
sca€olds on solid support. Ultimately this knowledge could then be used to design a new sequence
of reactions amenable to automation and that could deliver, in a swift manner, a combinatorial
library of biologically interesting macrocycles.2 This paper relates this e€ort which ®nally provides
us with a very ecient six-step synthetic scheme leading to macrolactones.
Use of serine and DHP resin. Serine like tyrosine is a tripod synthon which is usually linked to
the resin through its acid end during SPPS (solid phase peptide synthesis). This means that ®nal
* Corresponding authors. Fax: (819) 820-6823; e-mail: yves.dory@courrier.usherb.ca
0040-4039/00/$ - see front matter # 2000 Published by Elsevier Science Ltd.
PII: S0040-4039(00)00710-3

4744
compounds, like macrocycles,1 incorporating these a-amino acids used in a conventional manner
would always bear acid pendant groups or their corresponding esters. Nevertheless, such amino
acids possess two extra groups that could be used as a linkage to the resin; in particular serine can
also be attached to the resin by its amine or its alcohol. Since the technology related to the alcohol
is already known,3 it was decided to explore that direction. Thus, following Ellman's procedure,4
the DHP linker was attached to the Merri®eld resin and serine could be coupled to the resulting
DHP resin via its alcohol. The acetal which in fact corresponds to a THP ether should withstand
a wide range of conditions including strong bases. Two targets have been studied in a sequential
manner: The ®rst target 1 is a 14-membered cyclic ether that required eleven steps to assemble it
on solid support out of four building blocks (Scheme 1), whereas the second target 2 is an isosteric
14-membered lactone for which only six steps were necessary for its synthesis, because it was
made from three instead of four building blocks.
Scheme 1.
Design and synthesis of the ®rst target. The ®rst target is a 14-membered ring bearing one
insaturation. Its construction requires eleven consecutive steps of SPOC (solid phase organic
synthesis) amongst which ®ve are coupling reactions; the remaining six steps being either depro-
tection, cleavage or functionnal group interconversion. Thus, the ®rst coupling consists in the
attachment of suitably protected serine to the DHP resin. Two chains are then sequentially added
to the two remaining groups of serine to yield a fully alkylated serinol derivative. The last two
coupling reactions allow the introduction of the last malonate connector and the macrocyclization
of its corresponding anion by SN2 attack on the allylic chloride moiety. This sequence based on

4745
the DHP resin rendered possible the reduction of serine methyl ester to serinol by means of
lithium aluminium hydride. Finally, in order to ®nd the right conditions on the DHP resin, all the
reactions were carried out in solution ®rst with benzyl DHP resin model 3.
The protected serine 45 was coupled to the DHP resin by means of PPTS in 1,2-dichloroethane
(yield of 83% for the same reaction in solution using benzyl DHP resin model 3) (Scheme 2). The
resulting tosylate 5 was subjected to Mitsunobu coupling conditions with the alcohol 61 to obtain
the ether 7 with an excellent yield of 95% (yield in solution). Then, lithium aluminum hydride
reduction of the ester group a€orded the serinol derivative 8 (77% yield in solution)6. A long (18 h)
saturated aqueous ammonium chloride treatment had to be carried out to secure thorough
removal of metal hydroxides. The alcohol 8 underwent ether formation with the iodide 97 using
silver tri¯ate (despite the fact that silver nitrate was giving good results as well in solution; it
proved not successful on solid support because it is less soluble than the tri¯ate8). Under such
conditions 10 was formed with a good yield of 80% (solution phase). Steps 5 and 6 consisted in
Scheme 2.

4746
the cleavage of the PMB ether followed by activation of the resulting alcohol 11 to give the iodide
12. Noteworthy, the DDQ reaction conditions9 did not lead to resin-product cleavage at all;
rather the reaction was entirely selective and high-yielded (87% in solution). The malonate connector
was added (step 7) with a yield of 83% (solution) (Scheme 3). The remaining silyl protecting
group was removed at step 8 by means of TBAF (91% in solution); the alcohol 14 was trans-
formed into the corresponding allylic chloride 15 using NCS, triphenylphosphine and imidazole
in a yield of 76% for the same reaction in solution. Finally the macrocyclization (step 10) went
very well at 70ꢀC by simply adding Cs2CO3 in acetonitrile to the allylic chloride precursor 15 and
with a yield of 75% for the reaction in solution. This result con®rms that the SN2 mode of
cyclization is very powerful and that 14-membered rings bearing only one double bond are strain-
free and consequently cyclize readily.10 Treatment of 16 (solution model and solid phase) with
TFA provided the same macrocycle 1.11
Scheme 3.
The overall yield of 8% for the SPOC synthesis of target 1 was evaluated according to the
loading of the original Merri®eld resin and compares well with the solution phase overall yield of
11%. We can therefore safely consider that all the yields observed for the solution phase individual
steps were virtually identical to those obtained on solid support for the equivalent reactions. This
indicates that all the reactions discussed here are perfectly suitable for SPOC and combinatorial
chemistry.
Design and synthesis of the second target. In order to build libraries of macrocycles short sequences
of SPOC should be envisaged for reasons of overall yields, purity and ease of manipulation. The
DHP resin approach that we describe in this work appears to be versatile in terms of diversity at
the amino acid site because all amino acids with an alcohol or a phenol group can be used.

4747
However, the sequence shown above on the DHP resin is 11-step long because it took a lot of
e€ort to introduce the malonate unit. It is therefore possible to shorten it dramatically the same
way the tyrosine series was already reduced from 11 to six steps1 as shown in Scheme 1. Thus, in
order to simplify the previous DHP series, the number of building blocks was reduced from four
to three, and the ether formation from an alcohol and an iodide with silver tri¯ate was replaced
by a mild and more versatile carbodiimide coupling between an alcohol and an acid. In summary,
the prototype sequence of our future library gives a 14-membered lactone and is carried out in six
steps. The three building blocks, from which diversity will emerge, are assembled by means of
four coupling reactions: a THP protection with the DHP resin,4 a carbodiimide coupling to form
an ester, a Mitsunobu coupling and the cyclization reaction.
In solution, the methyl ester 5 (see Schemes 2 and 4) was reduced by means of lithium aluminum
hydride (78%). Then step 3 was a CDI coupling12 between the resulting alcohol 17 and the acid
1813 with a yield of 82%. The following reaction allowed the introduction of the allylic chloride
moiety and consisted in a Mitsunobu coupling between the tosylamide group of 19 and the
alcohol 201 (85%). Noteworthy, this reaction takes place at a later stage (step 4) than in the
previous similar six-step sequence.1 Clearly, this represents an improvement because the resulting
allylic chloride 21 is used right away. No doubt this improvement in¯uences the ®nal yield in a
Scheme 4.

4748
very positive way. Then the macrocyclization was performed in the usual manner by means of
cesium carbonate in acetonitrile at 70ꢀC and with a yield of 90%. Finally, the THP protection
was removed by means of TFA to give the ®nal macrocycle 214 (85%). The same procedure was
then repeated on the DHP resin with a better overall yield.
The overall yield of 45% for the SPOC synthesis of macrocycle 2 (Scheme 4) was evaluated
from the original loading of the Merri®eld resin used to prepare the DHP resin. This yield is 10%
higher than that obtained for the equivalent six-step sequence of reactions carried out in solution;
it corresponds to the very high average yield of 88%.
Conclusion. This episode in the investigation of SPOC technology, ultimately destined to the
creation of a library of macrocycles, led us to the establishment of a very ecient protocol suitable
for combinatorial chemistry. Whilst keeping the complexity of the two target molecules 1 and 2 to
the same degree during this work, it was ®nally possible to reduce the length of the sequence from
11 steps to merely six steps. The synthetic schemes take advantage of the DHP resin which allows
us to carry out reductive reactions safely; the coupling reactions comprehend the very mild diimide
and Mitsunobu conditions, resulting in the formation of N-alkylated and O-acylated serinol
derivatives. Then, the macrocyclization occurs by means of SN2 attack of a malonate anion on
an allylic chloride leaving group. Due to the very high overall yield obtained on solid support for
macrocycle 2, the prospect is very high and work is currently being pursued to actually build a
large library of distinct macro-heterocycles. The three blocks 4, 18 and 20 used in the model series
can be swapped for other related synthons in a very straightforward way, hence the possibility of
generating large and diverse libraries.
Acknowledgements
Financial support from Wyeth-Ayerst Research Laboratories and NSERC (Canada) through a
University-Industry research grant is greatly appreciated. We wish to thank Dr Magid Abou-
Gharbia for stimulating discussions.
References
1. Ramaseshan, M.; Robitaille, M.; Ellingboe, J. W.; Dory, Y. L.; Deslongchamps, P. Tetrahedron Lett. 2000, 41,
4737.
2. Kiely, J. S.; Hayes, T. K.; Grith, M. C.; Pei, Y. In A Practical Guide to Combinatorial Chemistry; Czarnik, A. W.;
DeWitt, S. H., Eds.; ACS: Washington, DC, 1996; pp. 99±122.
3. Nugiel, D. A.; Cornelius, L. A. M.; Corbett, J. W. J. Org. Chem. 1997, 62, 201.
4. Thompson, L. A.; Ellman, J. A. Tetrahedron Lett. 1994, 35, 9333.
5. Serine was sequentially treated with tosyl chloride and 1 M NaOH in ethyl acetate and water, then with MeOH
and HCl to yield 4 (52%).
6. Gargipati, R. S.; Adams, B.; Adams, J. L.; Sankar, S. K. J. Org. Chem. 1996, 61, 2911.
7. 1,4-Butanediol was monoprotected with PMB-Cl and DMF in THF (52%). The remaining alcohol group was
transformed into the corresponding iodide with Ph3P, MeI and DEAD to give 9 (84%).
8. Burk, R. B.; Gac, T. S.; Root, M. B. Tetrahedron Lett. 1994, 35, 8111.
9. Oikawa, Y.; Yoshioka, T.; Yonemistsu, O. ibid 1982, 23, 885.
10. Girard, S.; Deslongchamps, P. Can. J. Chem. 1992, 70, 1265.
11. Macrocycle 1 analytical data: 1H NMR ꢀ (CDCl3): 1.2±1.65 (6H, m, 3 CH2); 2.3±2.45 (4H, m, CH2-CHˆCH-CH2);
2.49 (3H, s, CH3-Ph-SO2); 3.14 (2H, t, J=4.5 Hz, CH2N); 3.4±3.5 (4H, m, CH2OCH2); 3.68 (2H, d, J=4 Hz,
CH2OH); 3.70 (3H, s, OCH3); 3.74 (3H, s, OCH3); 4.02 (1H, m, CHN); 5.3±5.5 (2H, m, CHˆCH); 7.3±7.65 (4H,

4749
m, Ph). 13C NMR ꢀ (CDCl3): 18.75, 20.82, 29.43, 31.36, 34.08, 34.90, 44.86, 52.10, 52.66, 59.38, 60.97, 62.84,
125.87, 127.16, 129.80, 132.19, 135.46, 143.56, 171.48, 171.66. IR n (cm^1): 3500br (OH); 1733 (CˆO). HRMS:
calcd [M+H]+: 498.1349; found: 498.1358.
12. Hassner, A.; Alexanian, V. Tetrahedron Lett. 1978, 46, 4475.
13. The acid 18 was obtained by Jone's oxidation of the corresponding alcohol (62%) (see Ref. 1).
14. Macrocycle 2 analytical data: 1H NMR ꢀ (CDCl3): 1.4±1.8 (4H, m, 2 CH2); 2.2±2.4 (4H, m, CH2-CHˆCH-CH2);
2.43 (3H, s, CH3-Ph-SO2); 2.60 (2H, t, J=6.5 Hz, -CH2-CO); 3.21 (2H, m, CH2N); 3.4±3.6 (4H, m, HO-CH2-CH-
CH2O); 3.71 (3H, s, OCH3); 3.73 (3H, s, OCH3); 3.97 (1H, tt, J=6 and 6.5 Hz, CHN); 5.15 (1H, m, CHˆCH);
5.45 (1H, m, CHˆCH); 7.25±7.35 (2H, m, Ph); 7.65±7.75 (2H, m, Ph). 13C NMR ꢀ (CDCl3): 19.10, 21.52, 30.82,
32.38, 34.27, 34.92, 45.02, 52.57, 52.66, 59.26, 60.80, 62.81, 125.87, 127.39, 129.75, 132.38, 137.48, 143.57, 171.48,
171.64, 172.81. IR n (cm^1): 3300br (OH); 1733 (CˆO). HRMS: calcd [M+H]+: 512.1954; found 512.1958.