Studies directed at the use of a parallel synthesis matrix to increase throughput in an in vivo assay

Article abstract of DOI:10.1021/jm990416r

Heparin is the anticoagulant of choice for hospitalized patients, but it is dosed only by injection because it is not absorbed following oral administration. We have discovered and prepared compounds (delivery agents) that facilitate the gastrointestinal absorption of heparin in rats, monkeys, and humans when given orally. We are currently developing a parallel synthesis approach to increase our delivery agent screening throughput in vivo. This approach has been used to produce micromolar quantities of compounds for testing in rats in a 5 x 5 parallel synthesis array. Using an amine benzoylation reaction sequence, 10 mixtures were prepared. These mixtures contained equal weight quantities of five N-substituted, non-α, amino acid delivery agents. Each of these mixtures was orally administered to rats in combination with heparin, and plasma clotting times (APTT) were measured to determine activity. Deconvolution of the data accurately identified the most active individual components. Independent synthesis of these compounds verified their activity. This parallel synthesis approach is an effective tool for the screening of oral heparin delivery agents and has increased screening throughput significantly.

Full text of DOI:10.1021/jm990416r

J . Med. Chem. 2000, 43, 3573-3576
3573
Brief Articles
Stu d ies Dir ected a t th e Use of a P a r a llel Syn th esis Ma tr ix to In cr ea se
Th r ou gh p u t in a n in Vivo Assa y
Andrea Leone-Bay,* J ohn Freeman, Doris O’Toole, Connie Rosado-Gray, Sirpa Salo-Kostmayer, Monica Tai,
Frank Mercogliano, and Robert A. Baughman
Emisphere Technologies, Inc., 765 Old Saw Mill River Road, Tarrytown, New York 10591
Received August 16, 1999
Heparin is the anticoagulant of choice for hospitalized patients, but it is dosed only by injection
because it is not absorbed following oral administration. We have discovered and prepared
compounds (delivery agents) that facilitate the gastrointestinal absorption of heparin in rats,
monkeys, and humans when given orally. We are currently developing a parallel synthesis
approach to increase our delivery agent screening throughput in vivo. This approach has been
used to produce micromolar quantities of compounds for testing in rats in a 5 × 5 parallel
synthesis array. Using an amine benzoylation reaction sequence, 10 mixtures were prepared.
These mixtures contained equal weight quantities of five N-substituted, non-R, amino acid
delivery agents. Each of these mixtures was orally administered to rats in combination with
heparin, and plasma clotting times (APTT) were measured to determine activity. Deconvolution
of the data accurately identified the most active individual components. Independent synthesis
of these compounds verified their activity. This parallel synthesis approach is an effective tool
for the screening of oral heparin delivery agents and has increased screening throughput
significantly.
Sch em e 1. Synthetic Scheme for the Preparation of
Delivery Agent Mixtures, M1-M10
In tr od u ction
Heparin is the anticoagulant of choice for hospitalized
patients, but it is dosed only by injection because it is
not absorbed following oral administration. We have
discovered a series of compounds (delivery agents) that
act to facilitate the oral delivery of heparin.1,2
As part of our continued efforts to develop an oral
heparin, we routinely prepare and screen novel, oral,
heparin delivery agent candidates. The candidate com-
pounds are prepared individually on a micromolar scale
and screened in vivo.3 In these screening studies, a
combination of one delivery agent and heparin is evalu-
ated per study group. This traditional approach requires
the preparation, purification, and structural identifica-
tion of each delivery agent individually, and each in vivo
study provides data on only one delivery agent.
To improve the throughput in these studies, a parallel
synthesis approach was investigated. Parallel synthesis
has become a successful tool for lead generation in the
field of medicinal chemistry.4 In general this technique
is most valuable as a method of preparation of large
numbers of compounds in nanomolar quantities for
evaluation in in vitro receptor binding assays. To date,
however, parallel synthesis has received little attention
for studies in which micromolar quantities of material
are needed for in vivo screening studies. This is most
likely due to the fact that animal studies are inherently
more variable than in vitro assays. As a result, inter-
pretation of the data can be more difficult. In this
particular case, the proposed studies were further
complicated by the variability of the APTT assay.8
Despite these challenges, we have been able to demon-
strate the use of a 5 × 5 parallel synthesis matrix to
improve the throughput of our in vivo oral heparin
delivery agent screen that requires about 500 mg of
material. The delivery agents tested in this matrix were
selected based on our previous work.3 This parallel
synthesis approach allows the screening of 25 delivery
agent candidates in 10 in vivo studies. The throughput
rate is significantly greater than the traditional ap-
proach of screening 25 delivery agents in 25 studies.
Ch em istr y
We have previously reported on a high-yield amine
benzoylation reaction sequence for the preparation of
N-substituted amino acids.6 This clean, high-yielding
reaction appeared to be a good candidate for use in a
parallel synthesis format.7
A dichloromethane solution of five amino acids (4-
aminobutyric acid, 6-aminocaproic acid, 8-aminocaprylic
* To whom correspondence should be addressed. Tel: 914-785-4733.
Fax: 914-593-8281. E-mail: ALeone-Bay@Emisphere.com.
10.1021/jm990416r CCC: $19.00 © 2000 American Chemical Society
Published on Web 08/26/2000

3574 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 19
Brief Articles
Sch em e 2. Parallel Synthesis Matrix Starting Materials and in Vivo Dataa
a
The numbers in parentheses report the mean peak APTT ((SEM) in seconds for each mixture and individual compound following a
single, colonic dose of the test compound in combination with heparin in rats.
acid, 10-aminodecanoic acid, and 12-aminododecanoic
acid) was divided into five equal portions, and each
portion was reacted according to Scheme 1 with one of
the following five acid chlorides: acetylsalicyloyl chlo-
ride, benzoyl chloride, 2-anisoyl chloride, 2-fluorobenzoyl
chloride, and 2-toluyl chloride. Conversely, a dichlo-
romethane solution of these acid chlorides was divided
into five portions and each portion was reacted accord-
ing to Scheme 1 with one of the five amino acids. The
resultant matrix is shown in Scheme 2. By design, each
reaction product contained five compounds in equal
amounts by weight. The mixtures were labeled M1-
M10 (Scheme 2). The components of each mixture in
this matrix were verified by independent synthesis and
co-injection on a high-performance liquid chromato-
graph. Each of these mixtures essentially contained the
five desired compounds as the only components and was
used without further purification. Figure 1 shows a
representative HPLC profile of a five-component mix-
ture (M8) prepared by this chemistry.
measured by activated partial thromboplastin time
(APTT).5 A delivery agent mixture was considered to
be active if a mean peak APTT level of >30 s was
obtained.
Figure 2a shows the pharmacodynamic profile ob-
tained following a single, oral dose of mixtures M1-
M5 in combination with heparin. Figure 2b shows
the pharmacodynamic profile obtained following admin-
istration of mixtures M6-M10 in combination with
heparin. Scheme 2 reports the mean peak APTT ((SEM)
values obtained in these studies. The data indicate that
the three most active delivery agent mixtures are M3,
M4, and M6. Of these, the two most active delivery
agent mixtures are M4 and M6 with mean peak APTT
levels of 242 ( 53 and 289 ( 59 s, respectively. The only
delivery agent common to both of these mixtures is 8
suggesting that 8 should be the most active oral heparin
delivery agent of the 25 compounds contained in the
matrix. Delivery agent mixture M3 also shows good
activity for oral heparin delivery producing a mean peak
APTT level of 168 ( 57 s. No compounds are common
to M3 and M4; however, 7 is contained in mixtures M3
and M6. Thus the data suggests that 7 should be a good
oral delivery agent for heparin and be slightly less active
than 8. M7 and M5 also exhibit reasonable activity.
Using the logic described above to analyze M3, M4, and
M6 data, compounds 9 and 12-14 would also be
expected to be effective oral heparin delivery agents.
To evaluate the ability of mixture testing to identify
effective heparin delivery agents, the 25 compounds
defined in the matrix were then each individually
prepared and tested in rats for their ability to facilitate
colonic heparin absorption. Some of these compound had
Resu lts a n d Discu ssion
Each compound mixture was tested colonically in rats
for its ability to facilitate the gastrointestinal absorption
of heparin. We have previously shown that colonic drug
absorption is a good model for oral drug absorption.8
Groups of five male Sprague-Dawley rats were admin-
istered by gavage a single, intracolonic (IC) dose of the
delivery agent mixture in combination with heparin.
Each experiment contained four dose groups. Blood
samples were collected prior to dosing and at 0.25, 0.5,
1.0, and 1.5 h after dosing. Evidence of heparin delivery
was indicated by an increase in blood clotting time

Brief Articles
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 19 3575
F igu r e 1. Representative reverse-phase HPLC of a five-component mixture.
delivery agents has increased screening throughput
been previously prepared and tested.3 The data obtained
from these studies is reported as mean peak APTT
(Figure 3) and shows that compounds 7 and 8 are the
most effective oral delivery agents for heparin. Com-
pound 9 also has good activity when tested with hep-
arin, and compounds 12-14 are significantly less active.
It is interesting that mixture M7 shows reasonable
activity; however, testing of its individual components
(10-14) indicates that they are less active than ex-
pected. This observation can be explained by the fact
that delivery agent activity is dependent on a variety
of factors. Among these, aqueous solubility is important.
Each of the compounds in mixture M7 has limited
aqueous solubility under the conditions used to dose the
discrete delivery agent/heparin combinations (50 mg/
mL delivery agent). Thus, compounds 10-14 appear to
have poor activity when tested individually. When these
compounds are administered as a mixture, the concen-
tration of each compound in the dosing solution is
reduced considerably (10 mg/mL) and each component
is soluble. Also, the heparin dose in the mixture dosing
solution is increased. These two factors can be used to
explain the apparent activity of M7. Thus, the matrix
screening approach has correctly identified the three
most active compounds in this group of 25 potential
delivery agents in only 10 in vivo studies.
significantly in this in vivo assay. This methodology
allows the testing of 25 compounds in 10 experiments,
while the traditional medicinal chemistry approach
requires 25 experiments to test 25 compounds. Both the
parallel synthesis and the traditional medicinal chem-
istry studies identified the same three delivery agents
as the most active compounds and agreed the order of
decreasing activity to be 8 > 7 > 9.
Exp er im en ta l Section
Ch em istr y. NMR spectra were recorded at 300 MHz in
either D2O or DMSO-d6. Combustion analyses were performed
by Microlit Laboratories, Madison, NJ , and were within
acceptable limits (C, H, N (0.4%). Thin-layer chromatography
(TLC) was performed using E. Merck Kieselgel 60 F-254 plates.
Reactions were monitored by high-pressure liquid chromotog-
raphy (HPLC) on a Vydac 25- × 4.6-mm C18 protein and
peptide column using a 0-50% gradient of acetonitrile in water
with 0.1% trifluoroacetic acid. Melting points were performed
using a Mel-Temp II from Laboratory Devices. All chemicals
used in the syntheses of mixtures 1-10 and compounds 5-29
were purchased from Aldrich Chemical Co., St. Louis, MO, and
prepared as described previously.3
Gen er a l P r oced u r e for th e P r ep a r a tion of Deliver y
Agen t Mixtu r es. P r ep a r a tion of M8. A mixture of 4-ami-
nobutyric acid (434 mg, 4.2 mmol), 6-aminocaproic acid (495
mg, 3.8 mmol), 8-aminocaprylic acid (543 mg, 3.5 mmol), 10-
aminodecanoic acid (582 mg, 3.1 mmol), and 12-aminodecanoic
acid (616 mg, 2.2 mmol) was dissolved in methylene chloride
(100 mL). Trimethylsilyl chloride (4.42 mL, 34.8 mmol) was
added in one portion and the resulting mixture was heated to
Con clu sion s
The use of a 5 × 5 parallel synthesis matrix to prepare
mixtures of compounds for testing as oral heparin

3576 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 19
Brief Articles
The reaction mixture was stirred at room temperature over-
night and then concentrated in vacuo. The residue was stirred
in saturated sodium carbonate and acidified with hydrochloric
acid (2 N). The resulting aqueous solution was extracted with
ether (2 × 100 mL), dried over sodium sulfate, filtered, and
concentrated in vacuo to give the desired five-component
mixture (4.31 g) that was used without further purification.
Dosin g Solu tion P r ep a r a tion . The delivery agent mixture
(150 mg) and heparin (75 mg) were mixed by vortex as dry
powders. This dry mixture was dissolved in 25% v/v aqueous
propylene glycol, and the apparent pH was adjusted to about
7 with 2 N aqueous sodium hydroxide. The final volume was
adjusted to 3.0 mL. The dosing solution was sonicated to
produce a clear solution with an apparent pH of about 7.
In Vivo Stu d ies. Male Sprague-Dawley rats, housed in
the animal facility at Emisphere Technologies, Inc., were
acclimated for a period of at least 5 days prior to dosing. The
animals weighed 300-350 g and were fasted for 12 h before
dosing. Groups of 5 rats were anesthetized with 44 mg/kg
ketamine hydrochloride intramuscularly immediately prior to
dosing. Each group was administered a single, colonic dose (1
mL/kg) of the delivery agent/heparin combination via a 7.5-
cm, 8-fr Rusch catheter attached to a 1-mL syringe. The dosing
catheter was inserted into the colon through the anus until
the tube was no longer visible, and the dosing solution was
expressed slowly into the colon. Citrated blood samples were
collected serially by cardiac puncture at 0, 0.25, 0.5, 1.0, and
1.5 h, plasma was harvested, and APTT values were measured.
Su p p or tin g In for m a tion Ava ila ble: Experimental data
for compounds 5-29. This material is available free of charge
via the Internet at http://pubs.acs.org.
F igu r e 2. (a) Pharmacodynamic response in rats following a
single, colonic dose of a delivery agent mixture (50 mg/kg) in
combination with heparin (50 mg/kg): squares, M1; circles,
M2; up-triangles, M3; down-triangles, M4; diamonds, M5. (b)
Pharmacodynamic response in rats following a single, colonic
dose of a delivery agent mixture (50 mg/kg) in combination
with heparin (50 mg/kg): squares, M6; circles, M7; up-
triangles, M8; down-triangles, M9;diamonds, M10.
Refer en ces
(1) Rivera, T.; Leone-Bay, A.; Paton, D. R.; Leipold, H.; Baughman,
R. A. Oral delivery of heparin in combination with sodium N-[8-
(2-hydroxybenzoyl)amino]caprylate: pharmacological consider-
ations. Pharm. Res. 1997, 14, 1830-1834.
(2) Brayden, D.; Creed, E.; O′Connell, A.; Leipold, H.; Agarwal, R.;
Leone-Bay, A. Heparin absorption across the intestine: effects
of sodium N-[8-(2-hydroxybenzoyl)amino]caprylate in rat in situ
intestinal instillations and in Caco-2 monoloayers. Pharm. Res.
1997, 14, 1772-1779.
(3) Leone-Bay, A.; Paton, D. R.; Freeman, J .; Lercara, C.; O′Toole,
D.; Gschneidner, D.; Wang, E.; Harris, E.; Rosado, C.; Rivera,
T.; DeVincent, A.; Tai, M.; Mercogliano, F.; Agarwal, R.; Leipold,
H.; Baughman, R. Synthesis and evaluation of compounds that
facilitate the gastrointestinal absorption of heparin. J . Med.
Chem. 1998, 41, 1163-1171.
(4) Berman, J .; Howard, R. J . In Combinatorial Chemistry and
Molecular Diversity Drug Discovery; Gordon, E., Kerwin, J . F.,
Eds.; Wiley-Liss: New York, 1998.
(5) Henry, J . B. Clinical Diagnosis and Management by Laboratory
Methods; W. B. Saunders: Philadelphia, 1979.
(6) Ho, K.-K.; Wang, N.-F.; Lercara, C., O′Toole, D.; Achan, D.;
Vuocolo, E.; Leone-Bay, A. Solution phase preparation of highly
pure amide mixtures via in situ chlorotrimethylsilane protection
and activation. Synth. Commun. 1997, 27, 883-895.
(7) Pirrung, M. C.; Chen, J . Preparation and screening against
acetylcholinesterase of a non-peptide “indexed” combinatorial
library. J . Am. Chem. Soc. 1995, 117, 1240-1245.
(8) Leone-Bay, A.; Ho, K.-K.; Agarwal, R.; Baughman, R. A.;
Chaudhary, K.; DeMorin, F.; Genoble, L.; McInnes, C.; Lercara,
C.; Milstein, S.; O′Toole, D.; Sarubbi, D.; Variano, B.; Paton, D.
R. 4-[4-[2-(Hydroxybenzoyl)amino]phenyl]butyric acid as a novel
oral delivery agent for recombinant human growth hormone. J .
Med. Chem. 1996, 39, 2571-2578.
F igu r e 3. Mean peak APTT in rats following a single, colonic
dose of a selected delivery agent (50 mg/kg) in combination
with heparin (25 mg/kg).
reflux for about 2 h. The reaction was cooled to room temper-
ature and triethylamine (3.64 mL, 26.1 mmol) was added
dropwise followed by anisoyl chloride (2.58 mL, 17.4 mmol).
J M990416R