Solid phase synthesis of diamides as potential bone resorption inhibitors

Article abstract of DOI:10.1016/S0960-894X(00)00025-1

Unsymmetrical diamide libraries have been prepared by a general and versatile solid phase route, using diacid templates in combination with aromatic and aliphatic amines chosen with the help of statistical experimental design. The compounds were tested as

Full text of DOI:10.1016/S0960-894X(00)00025-1

Bioorganic & Medicinal Chemistry Letters 10 (2000) 503±507
Solid Phase Synthesis of Diamides as Potential Bone
Resorption Inhibitors
,
Karin M. Edvinsson,x * Margareta Herslof, Patrik Holm, Nina Kann,y,*
David J. Keeling, Jan P. Mattsson, Bo Norden{ and Vladimir Shcherbukhin
Preclinical R&D, AstraZeneca R&D Molndal, S-431 83 Molndal, Sweden
È È
Received 19 October 1999; accepted 13 January 2000
AbstractÐUnsymmetrical diamide libraries have been prepared by a general and versatile solid phase route, using diacid templates
in combination with aromatic and aliphatic amines chosen with the help of statistical experimental design. The compounds were
tested as potential inhibitors of osteoclast vacuolar ATPase. # 2000 Elsevier Science Ltd. All rights reserved.
The rapidly developing ®eld of combinatorial chemistry
and, in particular, solid phase synthesis, has had a great
impact on the drug discovery process of most pharma-
ceutical companies.1 Instead of the time consuming
method of synthesising potential drug candidates one by
one, libraries of compounds can be produced either as
distinct substances in an array format or as mixtures
using split-and-mix techniques.
points 2 and 3 with variations both in the amino func-
tionalised side chains as well as the central diacyl unit.
Design of Building Blocks
Obvious templates for library synthesis would be I and
II (Table 1). We also selected another ®ve dicarboxylic
acids (III±VII) structurally similar to calculated low
energy conformations of 1, yielding a pool of seven
dicarboxylic acids.4
The osteoclast V-ATPase plays an essential role in bone
resorption, and inhibition of this acid pump has been
shown to inhibit bone resorption both in vitro and in
vivo.2 As part of a program to identify inhibitors of
bone resorption, potential isosteres of 1, a recently
published compound shown to inhibit the osteoclast
vacuolar ATPase (IC50 49 nM),3 were synthesised. Such
inhibitors are potentially useful for the treatment of
osteoporosis, i.e. the loss of bone. Diamides 2 and 3 were
found to inhibit the osteoclast vacuolar acid pump in vitro
with IC50 values of 1.5 and 4.9 mM respectively (Fig. 1).
To start with, we made calculations for a number of
amine and aniline structures from in-house databases.
Statistical experimental designs of amine derivatives and
anilines were based on molecular properties that were
calculated for both two-dimensional (2D) and three-
dimensional (3D) structures. The number of freely
rotatable bonds and rings (some of the parameters cal-
culated by our in-house SaSA software5) were used as
2D descriptors, describing conformational ¯exibility at
the topological level. The octanol/water partition coef-
®cients (known to be crucially important for various
types of biological activities) were calculated using the
method by Al Leo and implemented in the Daylight
toolkit.6 Using the Spartan package,7 the 3D structure
of the molecules was taken into account by another
group of descriptors. These consisted of two subgroups
basically describing spatial (molecular volume and sur-
face, moments of inertia etc.) and electronic (dipole
moment, hardness etc.) properties of the molecules, or
both properties simultaneously (octanol and water sol-
Aiming at increasing the in vitro potency, we set out
to prepare analogues in a library format of the starting
*Corresponding authors.
xCurrent address: KaroBio, Novum, S-141 57 Huddinge, Sweden.
Tel.: +46-8-608-6000; fax: +46-8-774-8261; e-mail: karin.edvinsson
@karobio.se
yCurrent address: Department of Organic Chemistry, Chalmers
University of Technology, S-412 96 Goteborg, Sweden. Tel.: +46-31-
7723065; fax: +46-31-7723657; e-mail: nina.kann@oc.chalmers.se
{Current address: Department of Medicinal Chemistry, AstraZeneca
R&D Lund, S-221 87 Lund, Sweden.
0960-894X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00025-1

504
K. M. Edvinsson et al. / Bioorg. Med. Chem. Lett. 10 (2000) 503±507
Figure 1.
Table 1. Building blocks for the diamide library
Anilines
Diacid templates
Amines
Diverse
Similar
Aromatic
Aliphatic
vation energies). Principal components (PC) analysis
was used in order to eliminate the redundancy between
these highly correlated descriptors, and to reduce the
property space dimensionality to a few (normally 2±3)
orthogonal principal components.
closeness in PC space. This maximum coverage algo-
rithm started with picking some arbitrary compound
(usually in the centre of the PC space) and then itera-
tively adding new ones maximising the sum of Eucli-
dean distances between the selected compounds. Both
the maximum coverage and the similarity processes can
be supervised by the chemist to avoid the selection of
compounds which may be inappropriate due to some
considerations (e.g. high price or interfering functional
groups); in this case the selected compound was
replaced by one of its nearest neighbours.
Compound selection was performed in such a way as to
achieve either the maximum coverage of the PC space
by a smaller subset of the original library using a space-
®lling design protocol implemented in SaSA,5 or to
achieve maximum similarity of derivatives based on

K. M. Edvinsson et al. / Bioorg. Med. Chem. Lett. 10 (2000) 503±507
505
With these criteria for selection in hand, a total of 20
anilines (Table 1: A±J similar to 3,4-dichloroaniline, A0±
J0 as diverse as possible) and 16 amines containing
additional nitrogen-based functional groups (Table 1:
a±h aromatic and a0±h0 aliphatic), were chosen as start-
ing materials for the combinatorial libraries.
mat, i.e. using 10 anilines (A±J or A0±J0)Â1 diacidÂ8
amines (a±h or a0±h0) for each plate. The selected ani-
lines were loaded onto the Rink chloride resin batchwise
and then distributed into the columns of a Flexchem
solid phase synthesis block21,22 yielding two types of
starter plates; one containing anilines similar to 3,4-
dichloroaniline, the other containing diverse anilines.
The synthetic steps were carried out in analogy with the
test reaction, using a Quadra 96 pipetting robot for
addition of the reagents as well as for the washing steps.
Puri®cation was also carried out in a 96-well format,
using pre-packed SPE plates with silica as the matrix.
The products were subsequently weighed and analysed
by MS(ES) and HPLC (254 nm).
Chemistry
Having completed the selection of diacids and amines,
we then set out to develop a solid phase route for the
assembly of the desired diamide products.8 Rink chlo-
ride,9 prepared from Rink acid resin, has previously
been used to attach anilines to a polymer backbone and
thus seemed a suitable choice of linker. Furthermore,
additional functionality for the resin linkage was not
necessary as the amino group itself functions both as a
point of attachment and a nucleophilic site. In order to
avoid selectivity problems with unsymmetrical diacids,
as well as to limit cross-linking and increase the solubi-
lity of the carboxylic acids, we chose to protect these as
their monoesters (methyl or ethyl).10±16 Our strategy
was thus to attach the anilines to Rink chloride, couple
them to our pool of diacid templates, hydrolyse the ester
and then perform a second coupling step. Standard
conditions for the amide bond formation (PyBOP/
DIPEA or DIC/DMAP) gave incomplete reaction for
most anilines, while electron poor anilines did not react
at all. A more successful method was to convert the
acids to their corresponding acid chloride in situ using
Ghosez' reagent (1-chloro-N,N,2-trimethylpropenyl-
amine).17 The ester was subsequently hydrolysed using
potassium trimethylsilanolate,18,19 and activation of the
acid that was formed was again e€ectuated using Gho-
sez' reagent. Finally, the resin bound acid chloride was
reacted with an amine, and the product was then
cleaved from the resin using 5% TFA in dichlor-
omethane. Small amounts of unreacted acid were
removed by ®ltration through a plug of silica gel. An
example of a test sequence is given in Scheme 1.20
Ten plates of compounds, based on four diacid tem-
plates (I±IV), were synthesised to start with. Approxi-
mately half of the expected 800 compounds were
formed, the yields being heavily dependent on the ani-
lines and amines used. As a general trend, electron poor
anilines (similar to 3,4-dichloroaniline) gave lower yields
than the diverse pool of anilines; aliphatic amines (a0±h0)
gave good results while aromatic amines (a±h) per-
formed poorly (in part due to problems with solubility).
Steric hindrance in the dicarboxylic acid component was
also found to be of importance; test reactions with
template VII gave low yields even with reactive anilines
and aliphatic amines. The best results were obtained
when template IV was reacted with diverse anilines and
aliphatic amines; MS analysis of this plate indicated
that 80% of the desired products had been formed. An
additional six compounds, based on the three remaining
diacid templates (V±VII), were synthesised in parallel.
The results from a representative example of a diamide
library is shown in Table 2. In this case template II was
reacted with diverse anilines and aliphatic amines giving
the desired products in about 50% of the cases. Anilines
A, D, F and G as well as amine c0 yielded no product
with template II.
Biology
Once we had ascertained that synthesis of the desired
diamides was possible on solid phase, we switched our
attention to the preparation of diamide libraries. For
practical purposes we chose to work in an 80-well for-
The 10 diamide libraries were screened at 9 mM for
e€ect on V-ATPase mediated proton transport in mem-
brane vesicles derived from chicken medullary bone,
Scheme 1. Synthesis of diamides on solid support. Reagents and conditions: (a) 4-benzyloxyaniline (3.25 equiv), iPr2Net (3.4 equiv), ClCH2CH2Cl,
rt, 48 h; (b) acid chloride of II (generated in situ from II and Ghosez' reagent, 3 equiv), pyridine (6 equiv), CH2Cl2, rt, 2Â2 h; (c) KOSiMe3 (5 equiv),
THF, rt, 3 h, then washed with 5% AcOH in THF; (d) Ghosez' reagent (4 equiv), rt, 1 h; (e) b0 (Table 1, 5 equiv), pyridine (10 equiv), CH2Cl2, rt,
12+3 h; (f) 5% TFA in CH2Cl2, rt, 2Â30 min; (g) ®ltration through silica.

506
Table 2. Yields and purities in a representative example of a diamide librarya
Anilines
K. M. Edvinsson et al. / Bioorg. Med. Chem. Lett. 10 (2000) 503±507
Diacid II
Amines
A
B
C
D
E
F
G
H
I
J
a0
b0
c0
d0
e0
f0
x
x
x
x
x
x
x
x
72 (78)
80 (81)
x
69 (66)
84 (87)
74 (85)
79 (81)
54 (86)
x
43 (33)
x
x
51 (56)
x
x
x
x
x
x
x
x
x
x
x
68 (86)
78 (84)
x
57 (83)
100 (58)
60 (91)
65 (90)
59 (93)
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
75 (92)
68 (90)
x
71 (83)
75 (89)
86 (92)
71 (91)
82 (89)
60 (80)
58 (64)
x
42 (65)
67 (68)
53 (72)
30 (61)
52 (90)
32 (81)
60 (80)
x
27 (77)
32 (89)
66 (86)
67 (83)
48 (79)
g0
h0
aPurities (in parentheses) determined by HPLC at 254 nm; x=desired product not found by MS(ES).
using pH dependent quenching of the ¯uorescent weak
base acridine orange (AO).23,24 None of the compounds
showed greater than 50% inhibition of the control rate
of proton transport, thus no IC50 values were measured.
In the same experiment, ba®lomycin A1 (30 nM), a spe-
ci®c V±ATPase inhibitor, completely abolished proton
transport. The most potent compound in the diamide
libraries was one of the starting compounds (compound
2, prepared in one of the libraries as well) which gave
51% inhibition.
compound IV and to Kerstin Carlsson and Karin
Rydstrom for biological screening of the compound
libraries.
References and Notes
1. For a recent review on solid phase synthesis, see Booth, S.;
Hermkens, P. H. H.; Ottenheijm, H. C. J.; Rees, D. C. Tetra-
hedron 1998, 54, 15385.
2. Keeling, D. J.; Herslof, M.; Mattsson, J. P.; Ryberg, B. Acta
Physiol. Scand. 1998, 163 (Suppl 643), 195.
3. Gagliardi, S.; Nadler, G.; Consolandi, E.; Parini, C.; Mor-
van, M.; Legave, M.-N.; Bel®ore, P.; Zocchetti, A.; Clarke, G.
D.; James, I.; Nambi, P.; Gowen, M.; Farina, C. J. Med.
Chem. 1998, 41, 1568.
Discussion and Conclusions
In summary, we have shown that unsymmetrical dia-
mides can be conveniently prepared using solid phase
methods. Attachment of aromatic amines to Rink
chloride provides an opportunity for linking without
necessitating any extra functionality; a method that can
be generalised for linking amine derivatives to solid
phase. Activation of carboxylic acids towards reaction
with amines and anilines using Ghosez' reagent works
smoothly even on solid phase, thus avoiding side
products leading to puri®cation problems. Using pre-
packed SPE plates was found to be a convenient way to
purify products in a 96-well format.
4. Conformational analysis of
1 was performed using
Macromodel V6.0 (Columbia University) with the Amber
force ®eld.
5. Shcherbukhin, V. V.; Olsson, T. SaSA: synthesis and
structure administratorÐan in-house program for descriptor
calculation, enumeration, chemical ®ltering etc.
6. Daylight Users Manual, Release 4.41, Copyright # 1992±
1995 by Daylight Chemical Information Systems, Inc., Irvine,
CA.
7. Spartan 5.1.2 Copyright # 1991±1998 by Wavefunction,
Irvine CA.
8. Another study concerning the solid phase synthesis of dia-
mides was recently published (Wahhab, A.; Leban, J. Tetra-
hedron Lett. 1999, 40, 235). Only symmetrical diacids and
aliphatic amines were used in this case however. Also, addi-
tional functionality (a carboxylic group or a second amino
group) was necessary for attachment of the ®rst amine to the
resin.
9. Garigipati, R. S. Tetrahedron Lett. 1997, 38, 6807±6810.
10. Compound I was obtained via esteri®cation of trans,trans-
muconic acid followed by selective hydrolysis with Ba(OH)2 in
CH2Cl2/MeOH.
Statistical experimental design has proven to be a useful
tool when choosing building blocks, thus reducing the
need to make a large number of compounds, while still
keeping the information content of diverse/similar
compounds as high as possible. Due to the moderate
potency of the library compounds, no clear SAR data
could be obtained.
The screening of the libraries showed that there was no
compound with better in vitro potency than the starting
point, i.e. diamide 2, suggesting that further production
of diamide libraries would not be worthwhile.
11. Hydrolysis and esteri®cation of 3-cyanobenzyl nitrile, pre-
pared from 3-cyanobenzyl bromide and acetone cyanohydrin
(Dowd, P.; Wilk, K. B.; Wlostowski, M. Synth. Commun.
1993, 23, 2323) yielded II.
12. Compound III was used as purchased (Aldrich).
13. Selective hydrolysis of dimethyl trans-1,2-cyclopropanedi-
carboxylate with KOH in wet MeOH gave IV.
14. Compound V was synthesised from ethyl-2-bromo-pro-
pionoate and glyoxylic acid by use of a Wittig reaction (Zum-
brunn, A.; Uebelhart, P.; Eugster, C. H. Helv. Chim. Acta
1985, 68, 1519).
Acknowledgements
We would like to thank Professor Mark Bradley, Uni-
versity of Southampton and Dr. Ingemar Nilsson and
Dr. Lars Swanson, AstraZeneca R&D Molndal, for
fruitful ideas concerning solid phase synthesis. We are
grateful to Magnus Johansson for the synthesis of
15. Pyridine-2,5-dicarboxylic acid was selectively esteri®ed
giving VI (Isagawa, K.; Kawai, M.; Fushizaki, Y. Nippon
Kagaku Zasshi 1967, 88, 553; CA 68840h 1968).

K. M. Edvinsson et al. / Bioorg. Med. Chem. Lett. 10 (2000) 503±507
507
16. Dimethyl trans-1,4-cyclohexanedicarboxylate was hydro-
lysed selectively using KOH/MeOH yielding VII (Smith, H.
A.; Fort, T., Jr. J. Am. Chem. Soc. 1956, 78, 4000).
17. Haveaux, B.; Dekoker, A.; Rens, M.; Sidani, A. R.; Toye,
J.; Ghosez, L. Org. Synth. Coll. 1988, VI, 282. We found that
using triphosgene instead of phosgene worked equally well for
the preparation of the reagent.
18. Hoekstra, W. J.; Greco, M. N.; Yabut, S. C.; Hulshizer, B.
L.; Maryano€, B. E. Tetrahedron Lett. 2629, 1997, 38.
19. Conjugate addition of KOSiMe3 to the double bond was a
side reaction for template III, but could be avoided if the
reaction time was limited to 3 h.
followed by amine b0 (12 mg, 0.075 mmol, 3 equiv) in CH2Cl2
(0.5 mL). The mixture was left over night. After removal of
the solvent followed by washing with CH2Cl2, the procedure
was repeated. The solvent was removed after 3 h at room
temperature and the resin washed with CH2Cl2 followed by
THF, MeOH, CH2Cl2:MeOH 1:1 and CH2Cl2. Cleavage: the
resin was treated with 5% TFA in CH2Cl2 (2 mL) for 2 Â 30
min; the ®ltrate and the washings were combined and con-
centrated. The crude material was mixed with EtOAc (0.5 mL)
and added to a silica SPE column (200 mg/3mL). Elution of
the product with CH2Cl2:MeOH 9:1 followed by evaporation
1
of the solvents yielded 6 (10 mg, 80%). H NMR (600 MHz,
20. General procedure exempli®ed by the synthesis of 6. To a
stirred solution of II (43 mg, 0.225 mmol) in dry CH2Cl2 (1.5
mL) was added 30 mg (0.225 mmol) 1-chloro-N,N,2-tri-
methylpropenylamine (Ghosez' reagent). The resulting solu-
tion was stirred for 1 h at room temperature. To a 3 mL
polypropylene tube ®tted with a frit containing 4 (50 mg, 0.025
mmol) swollen in CH2Cl2, was added pyridine (12.2 mL, 0.15
mmol, 6 equiv in 0.5 mL CH2Cl2) followed by 0.5 mL (0.075
mmol, 3 equiv) of the acid chloride solution above. After 2 h
with occasional mixing, the solvent was removed and the resin
washed with CH2Cl2 (3Â1 mL). The procedure above was
repeated. After removal of the solvent the resin was washed
with CH2Cl2, CH2Cl2:MeOH 1:1, MeOH and THF. KOSiMe3
(16 mg, 0.125 mmol, 5 equiv) in dry THF was added and the
mixture was left for 3 h with occasional mixing. The solvent
was removed and the resin washed with 5% AcOH in THF
followed by THF, MeOH, CH2Cl2:MeOH 1:1 and CH2Cl2.
Ghosez reagent (17 mg, 0.125 mmol, 5 equiv) in dry CH2Cl2
(1 mL) was added. After 1 h at room temperature with
occasional mixing, the solvent was removed and the resin
washed with CH2Cl2 followed by THF and CH2Cl2. Pyridine
(12.2 mL, 0.15 mmol, 6 equiv) in CH2Cl2 (0.5 mL) was added
CD3OD) d 7.79 (s, 1H), d 7.71 (d, 1H), d 7.52 (d, 1H), d 7.40±
7.45 (m, 5H), d 7.35 (dd, 2H), d 7.28 (t, 1H), d 6.93 (m, 2H), d
5.17 (s, 2H), d 4.47 (m, 1H), d 3.72 (s, 2H), d 2.14 (dd, 2H), d
1.64 (dd, 2H), d 1.57 (s, 6H), d 1.46 (s, 6H). MS ES+ 395.5,
ES 393.5.
21. Robbins Scienti®c, Sunnyvale, USA.
22. Distribution of the resin was achieved by suspending the resin
in CH2Cl2:THF 2:1 and pipetting the suspension into the wells.
23. Mattsson, J. P.; Lorentzon, P.; Wallmark, B.; Keeling, D.
J. Biochim. Biophys. Acta 1993, 1146, 106.
24. Vesicles (20 mg protein, prepared as described23 using cal-
cium deprived egg-laying hens) diluted in bu€er containing
DTT (®nal concentration 0.5 mM), bu€er (5 mM Hepes/Tris;
pH 7.4, 3 mM MgSO4, 125 mM KCl, 5 mM acridine orange,
1 mM valinomycin) and the compounds (9 mM, ®nal con-
centration) or DMSO (control) were pre-incubated for 10 min
at room temperature. Proton transport was then initiated by
the addition of 3 mM Tris±ATP and the decrease in AO
absorbance (l=490 nm) was followed in a 96-well plate spec-
trophotometer (Molecular Devices, CA, USA). The rate of
proton transport was taken to be the maximum rate of
decrease of AO absorbance.