Water-soluble PDE4 inhibitors for the treatment of dry eye

Article abstract of DOI:10.1016/j.bmcl.2010.03.023

PDE4 inhibitors have the potential to alleviate the symptoms and underlying inflammation associated with dry eye. Disclosed herein is the development of a novel series of water-soluble PDE4 inhibitors. Our studies led to the discovery of coumarin 18, whic

Full text of DOI:10.1016/j.bmcl.2010.03.023

Bioorganic & Medicinal Chemistry Letters 20 (2010) 2928–2932
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Water-soluble PDE4 inhibitors for the treatment of dry eye
a
a
b
a
a
Steven P. Govek a, , Guy Oshiro , John V. Anzola , Clay Beauregard , Jasmine Chen , Avery R. Coyle ,
Daniel A. Gamache b, Mark R. Hellberg b, Jennifer N. Hsien a, Julia M. Lerch a, John C. Liao b,
James W. Malecha a, Lena M. Staszewski a, David J. Thomas a, John M. Yanni b,
Stewart A. Noble a, Andrew K. Shiau a
*
a Kalypsys, Inc., 10420 Wateridge Circle, San Diego, CA 92121, United States
b Alcon Laboratories, Inc., 6201 South Freeway, Fort Worth, TX 76134, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
PDE4 inhibitors have the potential to alleviate the symptoms and underlying inflammation associated
with dry eye. Disclosed herein is the development of a novel series of water-soluble PDE4 inhibitors.
Our studies led to the discovery of coumarin 18, which is effective in a rabbit model of dry eye and a tear
secretion test in rats.
Received 5 January 2010
Revised 28 February 2010
Accepted 4 March 2010
Available online 7 March 2010
Ó 2010 Elsevier Ltd. All rights reserved.
Keywords:
PDE4
PDE4 inhibitor
Dry eye
Dry eye is a disease of the tears and ocular surface.1 Numerous
factors contribute to the onset of the disease, but once dry eye has
developed, inflammation of various ocular surface tissues propa-
gates the disease as both cause and consequence of ocular surface
damage.2 Individuals with dry eye suffer from ocular discomfort
(dry, gritty feeling; itching; stinging/burning; pain/soreness) and
blurred vision.3 Improvement in these symptoms can be affected
by administration of artificial tears, but the relief is transitory as
the underlying inflammation persists.4 Therefore, an agent capable
of reducing inflammation and inducing tear secretion should be an
effective therapy for dry eye.
The phosphodiesterase 4 (PDE4) enzymes regulate a host of bio-
logical processes by degrading the intracellular second messenger
cAMP.5 PDE4 inhibitors have been intensively investigated as anti-
inflammatory therapies because increases in cAMP levels are
known to attenuate inflammatory responses in multiple cell
types.6 Other agents that increase cAMP have been shown to in-
duce tear secretion.7 Therefore, PDE4 inhibitors should serve the
dual role of reducing inflammation and inducing tear secretion
providing an effective treatment for dry eye.
Piclamilast (1) is a potent, selective PDE4 inhibitor.8 The cocrys-
tal structure of piclamilast bound to PDE4B reveals a binding mode
within the active site that results in an exposed region of piclami-
last encompassing the amide carbonyl and C2 of the phenyl ring
(Fig. 1).9 We postulated that appending a flat, fused ring in this re-
gion would provide novel PDE4 inhibitors (2, Fig. 2). Furthermore,
substituents at C2 of 2 would have minimal interaction with the
protein which should allow for a variety of solubilizing groups to
be incorporated without adversely impacting potency. Indeed, a re-
lated series of compounds, represented by 3, has been reported,10
but substitution at N2 is not possible with this flat, unsaturated
phthalazine ring.
In this letter, we describe our efforts to create a novel, potent
series of PDE4 inhibitors with aqueous solubility compatible with
topical ocular delivery. Furthermore, we disclose efficacy in an
in vivo model for dry eye and an in vivo test for tear secretion.
* Corresponding author. Tel.: +1 858 342 4395; fax: +1 858 334 4848.
E-mail address: sgovek@san.rr.com (S.P. Govek).
Figure 1. Structure of piclamilast (1) bound to PDE4B (pdb:1XM4).
0960-894X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2010.03.023

S. P. Govek et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2928–2932
2929
pyridine ring would not be ideal, but the possibility of C2 substitu-
tion prompted us to prepare the compound. As expected, 7 is a very
weak PDE4 inhibitor. Initial attempts to prepare analogs with oxy-
gen-containing six-membered rings proved challenging, but even-
tually, chroman 8 was accessed. The lack of activity is not
unexpected when considering the deviation from planarity that
exists within the dihydropyran ring. Coumarin 9 was prepared to
restore planarity. With a biochemical potency of 5 nM and a cell-
based potency of 28 nM, coumarin 9 represents a novel scaffold
of PDE4 inhibitors that is comparable to piclamilast.
Figure 2. Generic structure 2 (X, Z = CH, N, O, or a bond; SG = solubilizing group)
and phthalazine 3.
Although we were successful in appending a flat, fused ring, the
carbonyl of coumarin 9 is positioned where we initially proposed
to append solubilizing groups. In an effort to evaluate other oppor-
tunities for substitution, we cocrystallized coumarin 9 with PDE4B
(Fig. 3).14 In this complex, the PDE4B enzyme adopts the same con-
formation as that observed in the piclamilast complex.15 Further-
more, 9 binds to the enzyme in an orientation that is nearly
identical to that of piclamilast and more importantly, exposes
one side of the cyclopentyl group. We postulated that if the cyclo-
pentane was replaced with an alkyl chain of sufficient length to
exit the hydrophobic pocket and further attached to a polar, solu-
bilizing group, the resultant PDE4 inhibitors would be both potent
and water soluble.
New scaffolds were prepared as detailed in Table 1. Compounds
were assayed11 for their intrinsic potency against human PDE4B
enzyme,12 as well as their ability to inhibit cAMP hydrolysis in a
cell-based biosensor assay.13 Our initial efforts focused on simple
nitrogen-containing heterocycles where SG = H (4–6). Disappoint-
The results of exploring chain length with a carboxylic acid as
the solubilizing group are shown in Table 2. Initially, acids 10–13
were prepared, and it became immediately clear that our strategy
was feasible. As the length of the alkyl chain was extended from
n = 1 to n = 5 (10–13), the compounds became increasingly more
ingly, a dramatic loss in potency (2 nM ? 0.5–3.3 lM) discouraged
further exploration. Under the presumption that the polar, basic
nitrogens were at the root of potency loss, compounds with oxy-
gen-containing rings were prepared. For benzofuran 7, we ex-
pected that the spatial positioning of the catechol motif and
potent in both the biochemical (1.2 lM ? 17 nM) and cell-based
Table 1
In vitro potency data for novel scaffolds
Compound
Structure
PDE4B IC50
,
l
Ma
CNG IC50, l
Ma
1
3
—
—
0.002
0.4
0.013
1.1
4
5
3.3
0.5
8.2
na
Figure 3. Structure of coumarin 9 bound to PDE4B (pdb: 3LY2).
Table 2
In vitro potency and solubility data for acid derivatives
6
7
8
2.3
16
10
>30
>30
Compound
n
PDE4B IC50
,
l
Ma
CNG IC50
,
l
Ma
Solubility mg/mLb
0.0003
9
10
11
12
13
14
15
16
—
1
3
4
5
6
7
8
0.005
1.180
0.080
0.043
0.017
0.002
0.002
0.001
0.03
5.35
0.89
0.73
0.28
0.12
0.04
0.03
>30
0.5
0.9
1.4
0.3
0.3
9
0.005
0.028
a
Value is mean of three or more experiments except 10 (n = 2).
Shake-flask solubility (0.1 M phosphate buffer; pH = 7.4).
b
a
Value is mean of two or more experiments (na: not active at 70 lM).

2930
S. P. Govek et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2928–2932
assays. Because the potency was still increasing at n = 5, the alkyl
chain was extended further (14–16), and the potency continued
to improve (17 ? 1 nM). With respect to aqueous solubility, com-
pounds 12–16 had moderate to good solubility, but no clear trend
was observed. Overall, the alkanoic acid side chain with n = 5–6 (13
and 14) provides for potent PDE4 inhibitors that have good water
solubility (0.9, 1.4 mg/mL).
We also employed a basic amine as a solubilizing group. The re-
sults of exploring chain length with a dimethylamino group are
shown in Table 3. As seen with the acids, an improvement in po-
tency was observed as the alkyl chain was lengthened. There was
a dramatic potency enhancement (670 ? 5 nM) from n = 3 to
n = 6 (17–20), but further improvement was minimal (5 ? 2 nM)
from n = 6 to n = 8 (20–22). With respect to aqueous solubility,
amine 18 (n = 4) had very good solubility (1.8 mg/mL), but there
was a significant drop-off (0.06 mg/mL) for 19 (n = 5). Within this
subseries, the dimethylaminobutyl side chain provides for a po-
tent, soluble PDE4 inhibitor.
Having identified a novel series of water soluble PDE4 inhibi-
tors, we sought to demonstrate in vivo efficacy. To explore the
anti-inflammatory nature of our compounds, we utilized the rabbit
model of lacrimal gland inflammation-induced dry eye, in which
corneal staining was used to measure ocular surface health.16 Of
the many new PDE4 inhibitors screened in this model (data not
shown), compound 18 looked to be the most promising, so a full
dose response study was run (Fig. 4). Compound 18 was extremely
effective at protecting the eye, equivalent to the corticosteroid
Figure 4. Inhibition of corneal staining in the rabbit model of lacrimal gland
inflammation-induced dry eye. Inhibition of corneal staining corresponds with
protection of the ocular surface. Percent inhibition is relative to vehicle (*p <0.05,
ANOVA, Dunnett’s post-test). Compounds were administered topically b.i.d. for
4 days. Positive control is the corticosteroid dexamethasone (100 lg/mL).
dexamethasone from 10 ng/mL to 10
of our compounds to induce tear secretion, we utilized the phenol
red thread test in rats (Fig. 5).17 At doses of 10
g/mL and 100 g/
lg/mL. To explore the ability
l
l
mL, compound 18 was effective at inducing tear secretion. Thus,
compound 18 demonstrates the potential effectiveness of PDE4
inhibitors for the treatment of dry eye.
In summary, we have disclosed the development of a novel ser-
ies of water-soluble PDE4 inhibitors. Our studies led to the discov-
ery of coumarin 18, which is effective in a rabbit model of dry eye
and a tear secretion test in rats. Coumarin 18 substantiates the idea
that PDE4 inhibitors can serve the dual role of reducing inflamma-
tion and inducing tear secretion providing an effective treatment
for alleviating the symptoms and underlying inflammation associ-
ated with dry eye.
Chemistry:18 Compounds 4–6 were synthesized as illustrated in
Scheme 1. Aniline 24 was prepared by nitration, alkylation, and
then reduction of guiacol (23). Heating 24 with Meldrum’s acid
and trimethylorthoformate gave an intermediate enamine that cy-
clized to quinolinone 25 by thermolysis in diphenyl ether. Alkyl-
ation of 26 followed by condensation with malonic acid gave
Figure 5. Increase in tear secretion in rats. Tear secretion was measured by the
phenol red thread test 10 min after topical administration of compound. Percent
increase is relative to untreated animals. Statistical significance is relative to saline
(*p <0.05, ANOVA, Dunnett’s post-test). Positive control is the beta-adrenergic
agonist isoproterenol (10 lg/mL).
acrylic acid 27. Formation of the acyl azide and thermolysis in
diphenylmethane gave isoquinolinone 28. Nitration of 29 followed
by alkylation gave 30. Reduction of the nitro group gave the aniline
which cyclized to quinazolinone 31 upon heating in formamide.
Intermediates 25, 28, and 31 were transformed into their respec-
tive chlorides by treatment with phosphoryl chloride. These chlo-
rides were converted into the desired compounds 4, 5, and 6 by
displacement with 4-amino-3,5-dichloropyridine.
Table 3
In vitro potency and solubility data for Me2N derivatives
Compounds 7–9 were synthesized as illustrated in Scheme 2.
Bromination of ketone 32 gave an a-bromo ketone. Treatment with
base followed by bromocyclopentane resulted in first, cyclization
onto the adjacent hydroxyl and then, alkylation of the remaining
hydroxyl. The resultant furanone was reduced with sodium boro-
hydride to give 33. Elimination to the benzofuran was spontaneous
when the hydroxyl of 33 was activated. Bromination followed by
coupling with 4-amino-3,5-dichloropyridine gave the desired ben-
zofuran 7. Towards 8, 32 was transformed into a chromenone by
treatment with triethylorthoformate in perchloric acid. Alkylation
and reduction gave chromanone 34. Treatment of 34 with hydrox-
Compound
n
PDE4B IC50
,
l
Ma
CNG IC50
,
l
Ma
Solubility mg/mLb
17
18
19
20
21
22
3
4
5
6
7
8
0.670
0.045
0.021
0.005
0.003
0.002
>30
1.82
0.66
0.18
0.15
0.15
1.8
0.06
a
Value is mean of three or more experiments.
Shake-flask solubility (0.1 M phosphate buffer; pH = 7.4).
b

S. P. Govek et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2928–2932
2931
Scheme 3. Reagents and conditions: (a) concd HCl, rt, 1 h; (b) Br(CH2)nCO2Et or
Br(CH2)nBr, NaH, DMSO, rt, 3–9 h; (c) LiOH, THF, MeOH, rt, 2–4 h; (d) 2 M Me2NH in
THF, DMSO, rt, 1–5 h.
man 8. Towards 9, 32 was alkylated with bromocyclopentane, acyl-
ated with diethylcarbonate, and then cyclized to hydroxycoumarin
35. Heating 35 with ammonium acetate in m-xylene provided an
aminocoumarin which was reacted with 3,4,5-trichloropyridine
to give the desired coumarin 9.
Compounds 10–22 were synthesized as illustrated in Scheme 3.
Dealkylation of 36 (prepared in analogous fashion as 9) with concd
HCl gave phenol 37. Subsequent alkylation with an ethyl bro-
moalkanoate or with a 1,n-dibromoalkane gave intermediates 38
and 39, respectively. Saponification of esters 38 gave the desired
acids 10–16. Displacement of bromides 39 with dimethylamine
gave amines 17–22.
Scheme 1. Reagents and conditions: (a) NaNO3, cat. NaNO2, H2SO4, Et2O, rt, 14 h;
(b) bromocyclopentane, Cs2CO3, CH3CN, reflux, 40 h; (c) 1 atm H2, 10% Pd/C, MeOH,
21 h; (d) Meldrum’s acid, HC(OMe)3, reflux 5 h; (e) Ph2O, 250 °C, 15 min; (f)
bromocyclopentane, K2CO3, CH3CN, reflux, 17 h; (g) malonic acid, piperidine,
pyridine, 100 °C, 2.5 h; (h) Ph2P(O)N3, Et3N, PhMe, rt, 3 h; (i) Ph2CH2, 200–225 °C,
5 h; (j) iPrONO2, nBu4NHSO4, H2SO4, CH2Cl2, 0 °C to rt, 45 min; (k) formamide,
190 °C, 6 h; (l) POCl3, cat. DMF, 1,2-dichloroethane, reflux, 15 h; (m) 4-amino-3,5-
dichloropyridine, NaH, DMF, 120 °C, 15 h from 25 and 28, or rt, 30 min from 31.
References and notes
1. International Dry Eye WorkShop (2007). Report of the Definition and
Classification Subcommittee. Ocul. Surf. 2007, 5, 75.
2. (a) Baudouin, C. Surv. Ophthalmol. 2001, 45, S211; (b) Stern, M. E.; Pflugfelder, S.
C. Ocul. Surf. 2004, 2, 124.
3. Perry, H. D. Am. J. Manag. Care 2008, 14, S79.
4. International Dry Eye WorkShop (2007). Report of the Management and
Therapy Subcommittee. Ocul. Surf. 2007, 5, 163.
5. (a) Conti, M.; Richter, W.; Mehats, C.; Livera, G.; Park, J.-Y.; Jin, C. J. Biol. Chem.
2003, 278, 5493; (b) Houslay, M. D.; Adams, D. R. Biochem. J. 2003, 370, 1.
6. (a) Spina, D. Br. J. Pharmacol. 2008, 155, 308; (b) Kodimuthali, A.; Jabaris, S. S. L.;
Patel, M. J. Med. Chem. 2008, 51, 5471.
7. Gilbard, J. P.; Rossi, S. R.; Heyda, K. G.; Dartt, D. A. Invest. Ophthalmol. Vis. Sci.
1990, 31, 1381.
8. Ashton, M. J.; Cook, D. C.; Fenton, G.; Karlsson, J.-A.; Palfreyman, M. N.;
Raeburn, D.; Ratcliffe, A. J.; Souness, J. E.; Thurairatnam, S.; Vicker, N. J. Med.
Chem. 1994, 37, 1696.
9. Card, G. L.; England, B. P.; Suzuki, Y.; Fong, D.; Powell, B.; Lee, B.; Luu, C.;
Tabrizizad, M.; Gillette, S.; Ibrahim, P. N.; Artis, D. R.; Bollag, G.; Milburn, M. V.;
Kim, S.-H.; Schlessinger, J.; Zhang, K. Y. J. Structure 2004, 12, 2233.
10. Napoletano, M.; Norcini, G.; Pellacini, F.; Marchini, F.; Morazzoni, G.; Ferlenga,
P.; Pradella, L. Bioorg. Med. Chem. Lett. 2000, 10, 2235.
11. All compounds were tested to a minimum of n = 2 (Table 1) or n = 3 (Tables 2
and 3; except 10 (n = 2)). The difference in IC50 was generally <twofold with
respect to the values quoted in the tables.
12. Bioluminescent biochemical assay: Govek, S. P.; Shiau, A. K.; Noble, S. A.;
Thomas, D. J. WO 2008/006051 A2, Jan 10, 2008.
Scheme 2. Reagents and conditions: (a) CuBr2, EtOAc, CHCl3, reflux, 16 h; (b)
K2CO3, CH3CN, rt, 2 h, then bromocyclopentane, reflux, 16 h; (c) NaBH4, THF, MeOH,
rt, 1 h; (d) CBr4, PPh3, CH2Cl2, 0 °C to rt, 30 min; (e) Br2, CS2, 0 °C, 1 h, then NaOEt,
EtOH, rt, 16 h; (f) 4-amino-3,5-dichloropyridine, Pd2(dba)3, X-phos, NaOtBu, PhMe,
110 °C, 2 h; (g) HClO4, HC(OEt)3, rt, 30 min, filter, then H2O, 90 °C, 5 min; (h)
bromocyclopentane, K2CO3, DMF, 100 °C, 3 h; (i) 1 atm H2, 10% Pd/C, PhMe, 7 d; (j)
NH2OHꢁHCl, K2CO3, EtOH, 80 °C, 6 h; (k) 1 atm H2, 10% Pd/C, THF, MeOH, 3.5 d; (l)
nBuLi, THF, ꢂ78 to 0 °C, then 3,4,5-trichloropyridine, reflux, 2 h; (m) bromocycl-
opentane, K2CO3, DMF, 110 °C, 17 h; (n) NaH, Et2CO3, reflux, 1 h; (o) KOtBu, tBuOH,
reflux, 4 d; (p) NH4OAc, m-xylene, reflux, 3 h; (q) NaH, 3,4,5-trichloropyridine,
DMSO, rt, 4 h.
13. HEK-293 cells stably expressing
a mutant cyclic nucleotide-gated (CNG)
channel (BD ACTOne, BDBiosciences) were plated at 1 ꢀ 106 cells/mL (25
lL) in
black 384-well, clear-bottom plates. Cells were incubated at 37 °C for 16 h.
Cells were then loaded by adding 25 lL of FLIPR membrane potential dye
(R8034, Molecular Devices) per manufacturer’s instructions for 4 h at room
temperature (rt). Compounds, previously arrayed in 11 or 21 point 1/2 log
dilutions, were then added as 10 lL solutions containing 1% DMSO in
Dulbecco’s phosphate-buffered saline. After a 15 min incubation at rt, 10 lL
of NECA (50-(N-ethylcarboxamido)adenosine, Sigma) in phosphate-buffered
saline was added to stimulate production of cAMP. The concentration of NECA
was titrated to yield an EC10 for maximal cAMP production (generally about
0.2 lM). After an additional 45 min at rt, fluorescence intensity was read on a
Molecular Devices Acquest plate reader. PDE4 inhibition results in higher
fluorescence signal due to decreased cAMP hydrolysis. Raw fluorescence data
ylamine followed by hydrogenation gave an aminochroman which
was reacted with 3,4,5-trichloropyridine to give the desired chro-

2932
S. P. Govek et al. / Bioorg. Med. Chem. Lett. 20 (2010) 2928–2932
were normalized to negative (DMSO) and positive (Roflumilast, a known PDE4
inhibitor, IC50 = 1 nM) controls. Data analysis was performed using Spotfire
(Spotfire, Inc.) and Kalypsys proprietary software.
The coordinates and structure factors have been deposited in the Protein Data
Bank (accession code 3LY2). (a) McCoy, A. J.; Grosse-Kunstleve, R. W.; Adams, P.
D.; Winn, M. D.; Storoni, L. C.; Read, R. J. J. Appl. Crystallogr. 2007, 40, 658. (b)
Muller, K.; Amman, H. J.; Doran, D. M.; Gerber, P. R.; Gubernator, K.; Schrepfer,
G. Bull. Soc. Chim. Belg. 1998, 97, 655. (c) Brünger, A. T.; Adams, P. D.; Clore, G.
M.; DeLano, W. L.; Gros, P.; Grosse-Kunstleve, R. W.; Jiang, J. S.; Kuszewski, J.;
Nilges, M.; Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren G. L. Acta
Crystallogr., Sect. D 1998, 54, 905.
14. An N-terminally His6-tagged catalytically active fragment of human PDE4B
was constructed by inserting the sequence for residues 152-487 into pET-28a.
The protein was expressed in Escherichia coli (BL21(DE3)) and purified using
metal affinity (Ni-NTA resin) and anion exchange (HiTrap
Q column)
chromatography. Protein (14 mg/mL) was mixed with 1 mM compound in
DMSO and incubated on ice for 1 h. Equal volumes of the protein stock and well
buffer (1.8–2.0 M ammonium sulfate, 1–2% PEG 400, 100 mM sodium
cacodylate pH 6.0) were mixed and then suspended over 1 mL well buffer
reservoirs at 4 °C. Crystals were soaked in a cryoprotecting solution (well
buffer with the PEG concentration increased to 30%) prior to freezing in liquid
nitrogen. Data were collected to 2.60 Å (Rsym = 12.1%) at the Advanced Light
Source beam line 5.0.3 (Lawrence Berkeley National Laboratory). The crystals
lie in the space group C2221 with cell constants a = 214.8 Å, b = 234.0 Å,
c = 165.3 Å. The structure was determined by molecular replacement as
implemented in PHASERa using the protein component of the PDE4B
roflumilast complex (PDB ID:1XMU) as the search model. The model was
rebuilt using Molocb and refined using non-crystallographic symmetry
constraints and a maximum-likelihood target (CNX)c to an R/Rfree = 20.5/23.3.
15. The conformational similarity of PDE4B bound to
9 and piclamilast was
assessed by calculating the root mean square deviation (rmsd) from a least
squares-based superimposition of the relevant coordinates of the alpha carbon
atoms of each amino acid. The rmsd was 0.57 Å.
16. Nagelhout, T. J.; Gamache, D. A.; Roberts, L.; Brady, M. T.; Yanni, J. M. J. Ocul.
Pharmacol. Ther. 2005, 21, 139.
17. The PRT test was performed on one eye of ophthalmologically normal, male
Lewis rats (4–6 weeks old, 10 per group). For a description of PRT methodology
in normal guinea pigs see: Trost, K.; Skalicky, M.; Nell, B. Vet. Ophthalmol. 2007,
10, 143.
18. All final compounds displayed spectral data (NMR, MS) that were
consistent with the assigned structure. Experimental details can be found
in Ref. 12.