Synthesis and biological evaluation of menthol-based derivatives as inhibitors of plasminogen activator inhibitor-1 (PAI-1)

Article abstract of DOI:10.1016/S0960-894X(03)00686-3

Compound 1 was identified by high throughput screening as a novel PAI-1 inhibitor. Optimization of the B and C-segments of 1 resulted in a series of structurally simplified compounds with improved potency. The synthesis and SAR data of these compounds are presented here.

Full text of DOI:10.1016/S0960-894X(03)00686-3

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3361–3365
Synthesis and Biological Evaluation of Menthol-Based Derivatives
as Inhibitors of Plasminogen Activator Inhibitor-1 (PAI-1)
Bin Ye,* Shawn Bauer, Brad O. Buckman, Ameen Ghannam, Brian D. Griedel,
Seock-Kyu Khim, Wheeseong Lee, Karna L. Sacchi, Kenneth J. Shaw, Amy Liang,
Qingyu Wu and Zuchun Zhao*
Discovery Research, Berlex Biosciences, 2600 Hilltop Drive, PO Box 4099, Richmond, CA 94804-0099, USA
Received 3 January 2003; accepted 24 April 2003
Abstract—Compound 1 was identified by high throughput screening as a novel PAI-1 inhibitor. Optimization of the B and C-seg-
ments of 1 resulted in a series of structurally simplified compounds with improved potency. The synthesis and SAR data of these
compounds are presented here.
# 2003 Elsevier Ltd. All rights reserved.
Plasminogen activator inhibitor-1 (PAI-1) is a naturally
occurring serine protease inhibitor, or serpin, that
rapidly inhibits the activity of several proteases, includ-
ing tissue plasminogen activator (tPA) and urokinase-
type plasminogen activator (uPA), by forming equimo-
lar, irreversible complexes that are internalized and
degraded.1 PAI-1 plays a major role in preventing fibri-
nolysis by decreasing the activity of tPA and/or uPA,
and consequently, the level of plasminogen converted to
plasmin.2 Plasmin is an enzyme critical to the lysis of
fibrin clots and works by cleaving fibrin to small soluble
products.3 Clinically, PAI-1 is considered to be a
thrombotic risk factor.4 Elevated levels of PAI-1 have
been described to correlate with an increased risk of
deep vein thrombosis,5 atherosclerosis,6 unstable angina
and myocardial infraction.1,4 In addition, elevated levels
of PAI-1 are also associated with a poor prognosis in
cancer patients,7 and PAI-1 is believed to play a role in
angiogenesis,8 cancer invasion,8 and metastasis.9 Thus,
inhibition of PAI-1 would be expected to be of ther-
apeutic benefit in a variety of cardiovascular and cancer
diseases. To date, several PAI-1 inhibitors including
antibodies,10 peptides,11 and small molecules1,12 have
been reported.
compound library led to the identification of the lead
compound 1 with an IC50 value of 1.4 mM. The lead 1
has a unique chemical structure, good potency, and
moderate oral bioavailability. However, its complex
chemical structure and poor aqueous solubility render it
suboptimal as a drug candidate. Thus, an effort was
made to simplify the structure and to improve the
potency as well as the pharmacokinetic profile. Structu-
rally, compound 1 can be divided into three segments
designated as A, B, and C (Fig. 1). Optimization of the
B and C-segments using a combined medicinal and
combinatorial chemistry approach is detailed in this
paper.
We started the optimization with the modification of the
B-segment( Table 1). The synthesis of the series in Table 1
is exemplified by Scheme 1. Amidation of (ꢀ)-men-
thoxyacetyl chloride 2 with 4-nitrobenzylamine 3 under
basic conditions afforded 4 in quantitative yield.
Hydrogenation of 4, followed by reductive-amination
with commercially available Psoromic acid 6, afforded
compound 1 in 78% yield over two steps. The rest of
analogues were prepared in a similar manner using dif-
ferentamines. Compared wiht 1, shortening the B-seg-
To develop small molecule PAI-1 inhibitors for the
treatment of thrombosis, we established a protein-based
PAI-1 assay.13 High throughput screening of our
*Corresponding author. Tel.: +1-510-669-4211; fax: +1-510-669-
4310; e-mail: bin_ye@berlex.com
Figure 1. Lead compound 1 from library.
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00686-3

3362
B. Ye et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3361–3365
Table 1. Replacementof B-segment
which functional group of the Psoromic acid backbone
is important to the inhibitory activity. Treatment of 1
with sodium methoxide and with ammonia afforded the
ring-opened products, 13 and 14, respectively (Scheme 2).
The ring-opened products maintained similar potency
to compound 1 (Table 2), indicating that diaryl ethers
mightbe suiatble C-segmentreplacemenst. For furht er
elaboration of this series, we defined the two rings of the
diaryl ether group as X and Y, and numbered the ring
position as shown in Scheme 2. A series of simplified C-
segmentpieces (e.g., 15–16) were designed and prepared
(Scheme 3). Coupling of phenol 17 with difluoro-
benzaldehyde 18 under basic conditions afforded 19,
which was hydrolyzed to give 15. Displacementof hte
chloride of 21 with phenol 20 under basic conditions,
followed by saponification afforded 16. The remaining
analogues were prepared in a similar fashion. All the C-
No.
Segment-B
IC50 (mM)a
1
1.4
7
1.1
3.9
8
9
1.4
segmentanalogues were coupled whit aniline
5 by
reductive-amination to afford the compounds listed in
Table 2. Mostof ht ese analogues ( 24–31) had potency
similar to or better than 1, exceptfor ht e simple diphe-
nyl ether analogue 23 which is devoid of a carboxylic
group. Furthermore, the methyl or ethyl ester deriva-
tives of the carboxylic acids (24–31) were inactive in the
primary assay (data not shown), suggesting that an
acidic group is essential for activity of this template. The
need for an acid moiety is also consistent with the lit-
erature findings.14 Compound 24 with a 20-carboxylic
acid and a 40-methoxy on the Y-ring had the same
potency as 1, and introduction of fluoride at the 3-posi-
tion of the X-ring (25) had no effecton poetncy. Com-
pounds with a nitro group on the X- or Y-ring also had
similar potency (27, 29, and 30). Improved potency was
seen for the compounds with polar groups such as 60-
hydroxy (26) and 40-amino group (28) on the Y-ring,
and for a compound with a heterocycle as the Y-ring
(31). Based on these results, we sought to test much
simpler C-segments. Commercially available aromatic
aldehydes were reacted with aniline 5 using solution
phase parallel synthesis. The resulting library was tested
without purification, and some of the compounds
10
11
12
5.0
0.53
>15
aIC50 values are averaged from multiple determinations (nꢁ2), and
the standard deviations are <30% of the mean.
ment by one carbon retained potency (7), while length-
ening by one carbon (8) decreased potency 3-fold.
Changing the substitution pattern to meta (9) did not
change potency, whereas blocking the hydrogen of the
amide bond in 1 with N-propyl decreased potency 3-fold
(10). The only modestimprovementwas found by
inserting the heterocyclic spacer 2,5-pyridinediamine
(11), which increased potency 2-fold. Non-aromatic
spacers were also investigated, with most of the replace-
ments decreasing potency. For this series, only one
example (12) is listed, which is almost inactive.
We next explored the optimization of the C-segment.
We initially did degradation studies on 1 to determine
Scheme 2. Conditions: (a) NaOꢂMe, MeOH, 0 ꢂC to rt, 2 h, yielding 13
.
(57%); (b) NH3 H2O, THF, 50 C, 1 h, yielding 14 (62%).
Scheme 1. Conditions: (a) Et3N, CH2Cl2, 3, 0 ꢂC to rt, 2 h, 100%;
(b) Pd/C, H2, MeOH, rt, 2 h; (c) NaBH3CN, MeOH, rt, 2 h, 6, 78% in
two steps.
Scheme 3. Conditions: (a) DMSO, NaH, rt, 10 h, 18 or 21; (b) LiOH,
THF/H2O, rt, 6 h.

B. Ye et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3361–3365
Table 2. Simplified C-segment: diphenyl ether
3363
No.
Segment-C
IC50 (mM)a
No.
Segment-C
IC50 (mM)a
No.
Segment-C
IC50 (mM)a
>15
1
1.4
26
0.71
32
13
1.8
27
1.2
33
>15
14
23
24
25
0.6
>15
1.4
28
29
30
31
0.7
0.71
2.6
0.5
34
35
36
37
>15
7.8
3.2
1.4
2.7
aIC50 values are averaged from multiple determinations (nꢁ2), and the standard deviations are ꢃ30% of the mean.
Scheme 4. Conditions: (a) EDC, DMAP, N-Boc-glycine, CH2Cl2, rt ,
overnight; (b) TFA, CH2Cl2, rt,
ClCH2CH2Cl, rt, 3 h.
2 h; (c) NaBH(OAc)3, 15,
showed moderate inhibitory activity. Table 2 lists some
typical examples (32–37) which were re-synthesized,
purified, and re-tested. Basically, neutral (32), basic (33)
and strong acidic (34) analogues were detrimental to
potency. Acid (35 and 36) and phenol (37) substituents
reduced potency 2- to 5-fold.
Scheme 5. Conditions: (a) NH3, ꢀ78 ꢂC to rt, 12 h; (b) DIEA, DMAP,
CH2Cl2, 2, 0 ꢂC to rt, 2 days, 44%; (c) CH3CN, 44, reflux, 4 h, K2CO3,
80%; (d) NaH, DMSO, 45, 80 ꢂC, 2 h, 78%.
After finding that the C-segment could be simplified
without a significant loss in potency, we looked at sim-
plifying the B and C fragments together. Reductive-
amination of aldehyde 15 with amine 39 prepared from
(ꢀ)-menthol 38 (Scheme 4) afforded compound 40 with
an IC50 value of 1.5 mM, which has equal potency as
compound 25. Although the structure of compound 40
did notlook like an exactdeletion of the B-segment(4-
amino-benzylamine) of compound 25, itappeared ht at
the B-segment may not be necessary for the potency.
Therefore, intermediate 43 was prepared (Scheme 5),
because a large number of analogues can be prepared by
simple displacement of the 5-fluoride with different
nucleophiles. Since the carboxylic acid group is essential
for the potency of this template, only necleophiles con-
taining acid were selected. Most of the amino-acid
adducts (46–52) maintained potency (Table 3). Com-
pounds 49, 50, and 51 were a mixture of diastereomers
since they were prepared from racemic amino-acids.
Additionally, 3- and 4-hydroxybenzoic acid, and 6-
hydroxy-3-pyridinecarboxylic acid were investigated
(53–55) because the corresponding final products con-
tain a diphenyl ether acid moiety, which was thought to
be an important functional group for potency. Interest-
ingly, compared with 1, compounds 53 and 54 have 4-
fold higher potency, while 55 is about 2-fold less potent.
In contrast to the case of 31, changing from benzene
(53) to pyridine (55) caused a decrease in potency.
We evaluated most of these compounds in a functional
15
clotlysis assay,
and explored their pharmacokinetic

3364
B. Ye et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3361–3365
Table 4. Selected data for PK and different assay
Table 3. Simplified analogues
No.
IC50 (mM)a
Functional
F% (rat)b
Primary
1
1.4
1.2
0.94
0.38
0.40
0.1
0.1
0.01
0.01
0.1
23
94
27
46
53
54
<10c
24
No.
X
IC50 (mM)a
<10c
46
0.94
aIC50 values are averaged from multiple determinations (nꢁ2), and
the standard deviations are <30% of the mean.
bCompounds were dosed by IV/PO in cannulated conscious rats with
a dose amount of 2 mg/kg. 94% PEG/2% 0.1HCl/4% EtOH solution
was used as a vehicle.
47
48
49
50
51
52
53
54
55
1.1
0.9
cBelow limit of detection.
0.89
2.9
Acknowledgements
The authors would like to thank Dao Lentz and Steven
Jones for high throughput screening and uPA-based
PAI-1 assay, Faye Wu and Kathy Tran for clot-lysis
assays, Marilyn Lam, Jun Shen, and Jih-Lie Tseng for
pharmacokinetic studies, and Monica Kochanny and
Gary Phillips for critical reading of this manuscript.
2.2
2.9
0.38
0.40
2.6
References and Notes
1. (a) Loskutoff, D. J.; Sawdey, M.; Mimuro, J. Prog. Hae-
most. Thromb. 1989, 9, 87. (b) Wu, Q.; Zhao, Z. Curr. Drug
Targets 2002, 2, 27.
2. Booth, N. A. Haemost. Thromb. 1992, 699.
3. Declerk, P. J.; Juhan-Vaguo, I.; Felez, J.; Wiman, B. J.
Intern. Med. 1994, 236, 425.
aIC50 values are averaged from multiple determinations (nꢁ2), and
the standard deviations are <30% of the mean.
4. Salomaa, V.; Stinson, V.; Kark, J. D.; Folsom, A. R.;
Davis, C. E.; Wu, K. K. Circulation 1995, 91, 284.
5. (a) Erikson, L. A.; Fici, G. J.; Lund, J. E.; Bogle, T. P.;
Polites, H. G.; Marotti, K. R. Nature 1990, 346, 74. (b)
Wiman, B. Haemost. Thromb. 1995, 74, 71.
6. Schneiderman, J.; Sawdey, M. S.; Keeton, M. R.; Bordin,
G. M.; Bernstern, E. F.; Dilley, R. B.; Loskutoff, D. J. Proc.
Nat. Acad. Sci. U.S.A. 1992, 89, 6998.
7. Frandesen, T. L.; Stephenes, R. W.; Pedersen, A. N.;
Engelholm, L. H.; Holst-Hansen, C.; Brunner, N. Drugs
Future 1998, 873.
8. Bajou, K.; Noel, A.; Gerard, R. D.; Masson, V.; Brunner,
N.; Holst-Hansen, C.; Skobe, M.; Fusening, N. E.; Carmeliet,
P.; Collen, D.; Foidart, J. M. Nat. Med. 1998, 4, 923.
9. Tsuchiya, H.; Katsuo, S.; Matsuda, E.; Sunamaya, C.;
Tomita, K.; Veda, Y.; Binder, B. Gen. Diag. Pathol. 1995, 141,
41.
properties in rat. Table 4 shows the data for the selected
compounds. Compound 27 has equal potency as 1 in
both the primary and functional assays, but its oral
bioavailability (F%) in rat was improved from 23 to
94%. Compounds 46 and 53 are 10-fold more potent
than 1 in the functional assay, and 53 has the same level
of oral bioavailability as 1. It is interesting to note that
54 is 10-fold less potent than 53 in the functional assay
even though they are equally potent in the primary
assay. The reason for this discrepancy remains unclear.
Compounds 46 and 54 both showed low plasma level
when dosed orally.
In summary, we have explored structure–activity rela-
tionships around the novel PAI-1 inhibitor 1 by modifi-
cation of the B- and C-segments. The B-segment could
be replaced with a variety of aromatic diamines without
loss of potency, but aliphatic diamines were not toler-
ated. Efforts to optimize the C-segment resulted in the
identification of a series of novel diphenyl ether car-
boxylic acid derivatives having improved potency,
moderate to good functional activity, and oral bio-
availability. Further simplification led to the discovery
of the most potent inhibitor 53 in both the primary and
functional assays. Optimization of the A-segment will
be reported in due course.
10. Abrahamsson, T.; Nerme, V.; Stromqvist, M.; Akerblom,
B.; Legnehed, A.; Petterson, K.; Westin-Eriksson, A. Hae-
most. Thromb. 1996, 75, 118.
11. (a) Eitzman, D. T.; Fag, W. P.; Lawrence, D. A.; Francis-
Chmura, A. M.; Shore, D.; Olson, S. T.; Ginsburg, D. J. Clin.
Invest. 1995, 95, 2416. (b) Guzzo, P. R.; Trova, M. P.;
Inghardt, T.; Linschoten, M. Tetrahedron Lett. 2002, 43, 41.
12. (a) Charlton, P. Drugs Future 1997, 22, 45. (b) Folkes, A.;
Brown, S. D.; Canne, L. E.; Chan, J.; Engelhardt, E.; Epsh-
teyn, S.; Faint, R.; Golec, J.; Hanel, A.; Kearney, P.; Leahy,
J. W.; Mac, M.; Matthews, D.; Prisbylla, M. P.; Sanderson, J.;
Simon, R. J.; Tesfai, Z.; Vicker, N.; Wang, S.; Webb, R. R.;
Charlton, P. Bioorg. Med. Chem. Lett. 2002, 12, 1063. (c)
Wang, S.; Golec, J.; Miller, W.; Milutinovic, S.; Folks, A.;

B. Ye et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3361–3365
3365
Williams, S.; Brooks, T.; Hardman, K.; Charlton, P.;
Wren, S.; Spencer, J. Bioorg. Med. Chem. Lett. 2002, 12,
2367.
the Hill equation with an automated analysis method using a
computer spreadsheet.
14. Bjoerquist, P.; Ehnebom, J.; Inghardt, T.; Hasson, L.;
¨
13. PAI-1 primary assay: PAI-1 activity was determined by its
inhibition on the activity of uPA. Assays were performed in
96-wells microtiter plates at room temperature in a total
volume of 200 mL. Human recombinantPAI-1 (6 nM, Molec-
ular Innovation) was incubated with compounds for 2 min in
assay buffer containing 50 mM TrisHCl, 150 mM NaCl, 2.5
mM CaCl2, and 0.1% polyethylene glycol 6000, pH 7.5.
Human kidney urokinase (0.0175 units/mL; Sigma) was then
added and incubated for another 2 min. Reactions started by
addition of a chromogenic peptide substrate, S2444 (32 mM;
Diapharma). Reaction rates were determined by measuring
the rate of the absorbance change at 405 nm in a ThermoMax
microplate reader (Molecular Devices). Controls without
compounds and with uPA only or with uPA plus PAI-1 were
assayed in the same plate. IC50 values for the inhibitors were
determined from the dose response curve by fitting the data to
Lindberg, M.; Linschoten, M.; Stroemqvist, M.; Deinum, J.
Biochemistry 1998, 37, 1227.
15. Plasma clot lysis assay: A plasma clotlysis assay was also
used to evaluate the potency of small molecule PAI-1 inhibi-
tors. In this assay, pooled human plasma was diluted (1:3) in a
buffer containing 150 mM NaCl, 2 mM CaCl2, 20 mM Hepes,
pH 7.4. Fibrin clot formation was initiated by addition of
human thrombin (30 nM). The newly formed fibrin clot
remained stable at 37 ꢂC for atleast2 h. If exogenous human
tPA (4 nM) was included in the assay, the fibrin clot will be
lysed within 20 min at 37 ꢂC, which was monitored by optical
absorbance at405 nm. In the presence of recombinanthuman
PAI-1 (4.3 nM), the activity of tPA was reduced and clot lysis
time was delayed. When both PAI-1 and PAI-1 inhibitors were
added, the PAI-1 activity was significantly inhibited, as
demonstrated by the shortened clot lysis times.