Similarities and differences in affinity and binding modes of tricyclic pyrimido- and pyrazinoxanthines at human and rat adenosine receptors

Bioorganic & Medicinal Chemistry 24 (2016) 4347–4362

Contents lists available at ScienceDirect

Bioorganic & Medicinal Chemistry

journal homepage: w ww.elsevier.com/locate/bmc

Similarities and differences in afﬁnity and binding modes of tricyclic

pyrimido- and pyrazinoxanthines at human and rat adenosine

receptors

a

b

´

Ewa Szymanska , Anna Drabczynska , Tadeusz Karcz , Christa E. Müller , Meryem Köse ,

c

a

´

Janina Karolak-Wojciechowska , Andrzej Fruzinski , Jakub Schabikowski , Agata Doroz-Płonka ,

a

a,

⇑

´

Jadwiga Handzlik , Katarzyna Kiec-Kononowicz

^aDepartment of Technology and Biotechnology of Drugs Jagiellonian University Medical College, Medyczna 9, PL 30-688 Kraków, Poland

^bPharmaCenter Bonn, Pharmaceutical Institute, Pharmaceutical Chemistry I, University of Bonn, An der Immenburg 4, 53121 Bonn, Germany

c

_

´

Institute of General and Ecological Chemistry, Technical University of Łódz, Zwirki 36, 90-924 Łódz, Poland

a r t i c l e i n f o

a b s t r a c t

Article history:

Received 31 March 2016

Revised 9 July 2016

Accepted 15 July 2016

Available online 18 July 2016

A new series of 32 pyrimido- and 5 tetrahydropyrazino[2,1-f]purinediones was obtained and evaluated

for their adenosine receptors (ARs) afﬁnities. The 1,3-dibutyl derivative of 9-(4-(2-(dimethylamino)

ethoxy)phenyl)-6,7,8,9-tetrahydropyrimido[1,2-f]purine-2,4(1H,3H)-dione was found to be the most

potent A₁AR antagonist of the present series, showing selectivity over the other AR subtypes. The struc-

ture–activity for the obtained purinediones was established. Docking experiments of the investigated

library to homology models of the human and rat A₁and A_2AARs allowed to compare the expected bind-

ing modes for selected compounds. The detailed analysis of binding cavities within individual AR sub-

types indicated small but signiﬁcant structural variations that may underlie the observed differences

in binding afﬁnities of purinediones at particular subtypes and species.

Keywords:

Adenosine receptor

Docking

Homology modeling

Purine-2,6-dione

Species differences

Subtype selectivity

Xanthines

Ó 2016 Published by Elsevier Ltd.

1. Introduction

effects on measures of locomotor activity, schedule-controlled

behavior, drug self-administration, and learning and memory.

Methylxanthines, including caffeine, theobromine and theo-

phylline, occur naturally in a number of species of plants belonging

to 28 genera and over 17 families, but the most common sources

are coffee, tea and cacao.¹The history of the medicinal use of cacao,

both as a primary remedy and as a vehicle to deliver other medici-

nes, dates back to the Olmecs, the ﬁrst elaborate pre-Columbian

civilization of Mesoamerica (1200–400 BCE), where a beverage

made from of Theobroma cacao beans ‘Xocolatl’ was highly valued

for its stimulating effect and healing properties.²Nowadays

methylxanthines, therapeutically used as bronchodilators and for

other indications, are probably the most widely self-administered

psychostimulatory drugs in the world.¹Natural xanthines (e.g. caf-

feine 1, Fig. 1A) and their synthetic derivatives exhibit a variety of

physiological effects, such as positive inotropic and chronotropic

effects on the heart, decreased airway resistance in the lung and

respiratory tract, stimulation as well as signiﬁcant behavioral

Most of these effects are likely due to the non-selective inhibition

of adenosine receptors.^3–5

Adenosine, an endogenous purine nucleoside, acts as a neuro-

modulator in both the central and peripheral nervous systems by

interacting with the P1 group of purine receptors, belonging to

the class A family of G protein-coupled receptors (GPCR). Four ade-

nosine receptor (AR) subtypes, A₁, A_2A, A_2Band A₃, are known and

have been pharmacologically characterized. Activation of the A₁

and A₃receptors inhibits the production of cyclic AMP via G_i/opro-

tein, while A_2Aand A_2Breceptor activation stimulates the activity

of the adenylate cyclase via G_sprotein, inducing an increase in

cAMP levels.⁶

The therapeutic potential of adenosine receptors ligands

depends on the diverse distribution of ARs throughout the body,

both in the central nervous system (CNS) and in peripheral tissues.

Thus, subtype-selective AR antagonists have been of interest as

potential kidney-protective (A₁AR), antiﬁbrotic (A_2AAR), neuro-

protective (A_2AAR), antiasthmatic (A_2BAR), antiglaucoma

(A₃AR), and anti-cancer (A_2A, A_2B) drugs.^3,7–10The A₁AR-selective

⇑

Corresponding author. Tel.: +48 012 620 55 80; fax: +48 012 620 55 96.

E-mail address: mfkonono@cyf-kr.edu.pl (K. Kiec´-Kononowicz).

http://dx.doi.org/10.1016/j.bmc.2016.07.028

0968-0896/Ó 2016 Published by Elsevier Ltd.

´

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

4348

Figure 1. (A) Structures of xanthine-based antagonists of adenosine receptors. (B) Scheme of structural modiﬁcations within the current study. Please note that according to

the IUPAC system the numbering of atoms for tricyclic compounds is different than for bicyclic xanthine derivatives.

xanthine-derived antagonists tonapofylline and rolofylline have

been explored for clinical applications in heart failure, for improv-

ing renal function and for the treatment of acute renal failure.^11,12

Among the A_2AAR antagonists, the xanthine-based istradefylline

(2, Fig. 1A) has been evaluated in clinical trials for Parkinson’s dis-

ease (PD) and depression and has been recently approved in Japan

as an antiparkinsonian drug.¹³The blockade of A_2AARs localized in

the brain has been proposed not only for the treatment of motor

deﬁcits in PD, but also for Alzheimer’s disease and Huntington’s

disease.^14–17

In the present project we focused on a new series of tricyclic

9-aryl/arylethyl-tetrahydropyrimido[1,2-f]purine-2,4-diones. As a

starting point for modiﬁcations two previously described com-

pounds were chosen, with their constrained structure mimicking

the structure of (E)-8-styrylxanthines previously developed as

A_2AAR ligands (Fig. 1B): 3a (A_2AAR antagonist)¹⁸and 3b (dual

A_2Aand A₁AR antagonist).¹⁹The primary goal of this work was

to increase the afﬁnity and selectivity for either A_2Aor A₁ARs

and to overcome the problem of low solubility.

For these purposes the following structural changes have been

introduced:

The structure–activity relationships (SAR) of xanthine-derived

AR antagonists, including polycyclic fused ring systems, at various

adenosine receptors subtypes has been intensively studied by our

groups as well as others.^18–24In general, modiﬁcation of the xan-

thine scaffold at the 8-position with aryl, alkylaryl or cycloalkyl

groups increases afﬁnity at the A₁AR or may result in high afﬁnity

for A_2BAR. Introducing an alkene spacer into this position con-

nected to an aromatic ring, e.g., the styryl residue, typically leads

to increased potency and selectivity for the A_2AAR (e.g.,

KW6002).^24–26A separate class of xanthines with modiﬁed 8-posi-

tion is represented by tricyclic compounds containing a third ring

fused to the f-bond of the 2,6-purinedione system. As previously

reported by our groups, tricyclic dihydroimidazo-, tetrahydropy-

rimido- and tetrahydrodiazepino[1,2-f]purinediones can act as rel-

atively potent A_2AAR (or A_2A/A₁AR) antagonists, in which the

annelated pyrimidine ring appears to be beneﬁcial for activity (3,

Fig. 1A).^18–22From a structural point of view, these compounds

can be treated as sterically ﬁxed and conﬁgurationally stable

analogs of (E)-8-styrylxanthines, the scaffold that is contributing

to the high A_2AAR afﬁnity of istradefylline. Unfortunately, the

drawback of this class of xanthine derivatives is their low

water-solubility leading to poor bioavailability.²⁴

(1) In addition to the two previously reported hydroxyphenyl

derivatives 4 and 8 included in this paper as reference com-

pounds,^18,19six new 4-hydroxyphenyl compounds were

synthesized and evaluated (Tables 1–3);

(2) An aliphatic tertiary amino group which increases the basic-

ity and solubility of the parent compound has been intro-

duced at the alkoxy chain attached to the benzene ring.

Dimethylamine, diethylamine, pyrrolidine, piperidine and

morpholine derivatives were prepared (Tables 1 and 2);

(3) The tetrahydropyrimidine ring was replaced by a tetrahy-

dropyrazine ring; this regioisomeric change shifts the

N-arylalkylamino substituent from the 9- to the 8-position

(Table 3). The N8 nitrogen atom is expected to exhibit

increased basicity due to its increased aliphatic character,

compared to the N9 atom.²⁷

For all synthesized tricyclic xanthines their afﬁnity to native rat

A₁(rA₁) ARs and A_2A(rA_2A) ARs as well as to human recombinant

A_2B(hA_2B) ARs and A₃(hA₃)ARs was determined in radioligand

binding assays. Selected compounds were further evaluated for

´

4349

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

Table 1

Afﬁnities of N9-aryl-tetrahydropyrimido[2,1-f]purinediones at A₁–A₃adenosine receptors

O

R¹

N

O

N

R¹

O

R²

Compd R²

QPlogP_o/w

QPlogS^*

rA₁

rA_2A

rA_2B

rA₃

hA₁

hA_2A

[³H]MSX-2 [³H]PSB-603 [³H]PSB-11

hA_2B

hA₃

*

[³H]CCPA

[³H]MSX-2 [³H]PSB-603 [³H]NECA [³H]CCPA

K_iSEM [nM]

(% inh. SEM at 1 lM)

R

¹= methyl

3a¹⁸

–CH₃

2.741

2.209

ꢀ4.244

ꢀ4.453

>25,000

(29 3)^a

998 70

(94 6)^a

nd

4¹⁸

–H

35,000 2000 1620 310 nd

>1000

(5 5)

>>1000

(6 3)

(42)^a

>1000

(2 2)

>>1000

(3 8)

>1000

(1 6)

>>1000

(ꢀ1 5)

12

13

2.029

2.587

ꢀ1.589

nd

N

>1000

(ꢀ3 3)

>1000

(3 3)

>>1000

(12 10)

>1000

(ꢀ5 2)

>1000

(3 3)

>>1000

(3 5)

ꢀ1.830

ꢀ2.006

>>1000

(13 5)

>>1000

(0 5)

14

15

16

2.469

2.631

nd

N

>1000

(0 1)

>>1000

(ꢀ4 9)

>>1000

(14 7)

>1000

(ꢀ1 3)

>1000

(ꢀ7 3)

>>1000

(6 8)

6

ꢀ2.237

ꢀ1.072

1.597

>1000

>>1000

(ꢀ5 4)

O

2.278^**

ꢀ3.271^**(ꢀ1 4)

R

¹= ethyl

5

–H

3.014

2.830

ꢀ4.986

ꢀ2.090

>1000

(14 2)

893 97

(53 2)

nd

>1000

(34 9)

>>1000

(1 2)

297 32

(36 2)

>1000

(27 4)

>1000

(29 4)

>>1000

(ꢀ1 7)

17

N

R

¹= n-propyl

6

3.749

ꢀ5.717

ꢀ2.522

>500

(25 3)

785 253

(53 8)

nd

>1000

(20 9)

>>1000

(18 2)

180 22

(47 3)

>1000

(44 7)

>1000

(17 2)

>>1000

(17 5)

18

19

3.473

4.128

nd

N

567 106

(30 3)

1790 510

(51 7)

>1000

(22 7)

>1000

(20 3)

ꢀ3.178

ꢀ3.342

175 44

(53 5)

>1000

(28 4)

>1000

(4 2)

>1000

(10 9)

1990 140 2380 450 >1000

(22 3)

>1000

(20 3)

20

21

22

4.011

4.177

N

(29 6)

(10 4)

254 24

(44 3)

>1000

(44 5)

>1000

(18 6)

>1000

(29 4)

ꢀ3.470

ꢀ1.932

nd

3.128

>1000

1930 460

(45 5)

>1000

(3 3)

>1000

(23 3)

nd

O

3.896^**

ꢀ4.739^**(14 2)

R

¹= n-butyl

7

–H

4.501

ꢀ6.524

ꢀ3.196

174 19

(48 1)

738 141

(64 3)

nd

>1000

(33 7)

>1000

(40 4)

56

8

1050 90

(47 2)

>1000

(9 4)

>1000

(15 8)

539 74

(43 5)

1030 130 >500

(52 1) (40 5)

>1000

(38 8)

23

24

4.185

4.802

N

(72 2)

137

9

1460 250 >1000

(49 6)

>1000

(16 7)

1190 130 1010 180 >1000

(30 5)

>1000

(44 2)

ꢀ3.650

ꢀ4.208

(55 2)

(6 3)

(54 4)

(33 5)

72

8

989 200

(49 4)

>1000

(15 1)

>1000

(20 4)

658 69

(41 5)

666 128

(68 2)

>1000

(34 1)

813 140

(48 2)

25

26

27

4.793

4.927

N

(69 1)

127 25

(65 3)

1480 170 >1000

(49 8)

>1000

(20 2)

713 119

(42 4)

921 29

(53 4)

>500

(41 2)

854 69

(48 2)

ꢀ4.307

ꢀ2.683

(9 11)

3.912

268 14

>1000

(31 2)

>1000

(27 8)

>1000

(38 5)

nd

O

4.318^**

ꢀ4.128^**(35 2)

nd—not determined.

a

Tested at 25 lM.

Protonated form (charge +1).

Not protonated (neutral) form.

*

**

their afﬁnity to recombinant human A₁(hA₁)ARs, human A_2A

(hA_2A)ARs, rat (rA_2B) A_2BARs, and rat A₃(rA₃) ARs. Additionally,

antagonistic properties of selected tricyclic xanthines were

examined in cAMP accumulation assay. The solubility of the syn-

thesized series (QPlogS) was estimated by in silico calculation,

and determined experimentally for selected analogues as well as

´

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

4350

Table 2

Afﬁnities of N9-arylethyl-tetrahydropyrimido[2,1-f]purinediones at A₁–A₃adenosine receptors

O

R¹

N

O

N

R¹

O

R²

rA₃

Compd R²

QPlogP_o/w

QPlogS^*

rA₁

rA_2A

rA_2B

hA₁

hA_2A

[³H]MSX-2 [³H]PSB-603 [³H]PSB-11

hA_2B

hA₃

*

[³H]CCPA

[³H]MSX-2 [³H]PSB-603 [³H]NECA [³H]CCPA

K_iSEM [nM]

(% inh. SEM at 1 lM)

R

¹= methyl

3b¹⁹

8¹⁹

–CH₃

–H

2.868

2.672

ꢀ2.763

ꢀ3.554

3850 430 370 20

nd

630 350

nd

ca. 25,000

230 10

>25,000

7200 600^b

>10,000

(46)^a

(34)^a

(9)^c

>1000

(ꢀ2 3)

653 84

(49 3)

>1000

(9 4)

>>1000

(ꢀ5 7)

28

29

2.597

3.137

ꢀ1.074

nd

N

>1000

(4 2)

339 48

(53 4)

>1000

(ꢀ2 5)

>1000

(6 3)

>1000

(1 4)

559 89

(57 4)

>1000

(6 3)

>>1000

(ꢀ3 4)

>>1000

(3 4)

ꢀ1.426

ꢀ1.535

>1000

(7 2)

>1000

(34 11)

>1000

(13 6)

30

31

32

2.995

3.301

nd

N

>1000

(3 1)

645 244

(65 5)

>1000

(2 4)

>>1000

(1 7)

ꢀ2.039

ꢀ0.549

2.232

>1000

585 150

(71 6)

>1000

(ꢀ5 2)

>>1000

(ꢀ3 3)

O

2.306^**

ꢀ0.932^**(0 1)

R

¹= ethyl

9

–H

3.350

ꢀ3.807

ꢀ0.870

>1000

(14 2)

271 16

(80 2)

>1000

(14 1)

>1000

(19 4)

2020 70

(20 3)

2180 390 >500

>>1000

(6 4)

(30 5)

(42 3)

2.817

>1000

859 23

(51 2)

>1000

(36 9)

>>1000

(7 3)

33

nd

N

2.803^**

ꢀ0.888^**(17 2)

R

¹= n-propyl

10

4.183

4.526

ꢀ5.516

370 37

(33 2)

464 66

(67 1)

nd

663 130

(58 4)

>1000

(24 2)

347 33

(31 1)

658 115

(65 3)

338 14

(66 4)

633 52

(53 3)

34

35

ꢀ2.312

ꢀ1.287

nd

N

3.440

>500

>1000

(44 1)

853 73

(50 2)

>1000

(20 4)

O

3.576^**

ꢀ1.800^**(26 3)

R

¹= n-butyl

11

–H

4.812

5.216

ꢀ5.379

184 57

(51 3)

518 79

(63 5)

nd

131 22

(80 2)

619 110

(44 4)

80

5

566 43

(64 2)

1100 100

(38 3)

>1000

(15 3)

1550 180 546 67

(27 3)

109 11

(87 1)

310 48

(75 2)

36

37

ꢀ2.931

ꢀ1.689

N

(69 2)

(64 4)

4.143

172 35

1370 270

(47 5)

337 79

(62 4)

890 84

(48 5)

nd

4.161^**

ꢀ2.029^**(47 3)

nd – not determined.

a

Tested at 25

lM.

b

c

Tested versus [³H]ZM241385.

Tested at 10 lM.

Protonated form (charge +1).

Not protonated (neutral) form.

*

**

two lead structures. To rationalize the observed subtype selectivity

as well as the species selectivity of the new compounds, molecular

modeling and docking studies using homology models of both

human and rat adenosine receptors were performed.

(Scheme 1) or 1,3-dimethyl urea and cyanoacetic acid (Scheme 2)

was applied using modiﬁcations of literature methods^18–21to yield

the key dihalogen intermediates. Formation of the tetrahydropy-

rimidine ring in 5–7, 9–11 was achieved by cyclization of

8-bromo-7-(3-chloropropyl)-1,3-dialkylpurinediones with either

4-aminophenol or tyramine, in analogy to the preparation of the

previously reported compounds 4 and 8.^18,19Phenol 38 was

prepared by a similar cyclization reaction of 7-(2-chloroethyl)-8-

(chloromethyl)-1,3-dimethyl-purinedione.

2. Results and discussion

2.1. Chemistry

To access the library of the target ethers 12–37 and 39–42, the

corresponding phenols were reﬂuxed with commercial 2-

chloroethylamines (alkyl or heterocyclic) in 2-butanone as a sol-

vent, in the presence of potassium carbonate as an absorbing agent

of hydrogen chloride released during the reaction.

The general procedure for the synthesis of the target series is

outlined in Schemes 1 and 2. A several-step procedure, starting

from 1,3-diethyl/1,3-dipropyl urea and cyanoacetic acid (1,3-

dimethyl- and 1,3-dibutyltheophylline are commercially available)

´

4351

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

Table 3

Afﬁnities of N8-arylethyl-tetrahydropyrazino[2,1-f]purinediones at A₁–A₃adenosine receptors

O

N

O

N

R

O

N

QPlogS^*

rA₁

rA_2A

rA_2B

[³H]PSB-603 [³H]NECA [³H]CCPA [³H]MSX-2 [³H]PSB-603 [³H]PSB-11

rA₃

hA₁

hA_2A

hA_2B

hA₃

*

Compd

R

QPlogP_o/w

[³H]CCPA

[³H]MSX-2

K_iSEM [nM]

(% inh. SEM at 10

l

M)

(% inh. SEM at 1 lM)

38

–H

1.642

ꢀ2.735

>1000

>10,000

(35 4)

nd

>1000

(11 7)

>1000

(3 6)

1.716^**

ꢀ3.352^**

(4 2)^a

>1000

(4 2)

>1000

(6 7)

39

40

1.259–1.396 0.768–1.091

1.761–1.884 0.634–0.901

1370 318 6400 990

nd

N

>10,000

>1000

(19 10)

>1000

(0 1)

7850 798

(42 4)

>1000

(13 1)

>1000

(0 3)

41

42

1.934–2.055 ꢀ0.021–0.167 1740 378 5640 1070

nd

N

0.961–1.007 0.973–1.784

>10,000

(22 5)

>1000

(4 3)

>1000

(3 2)

11,500 2650 nd

O

0.936^**

1.097^**

nd—not determined.

a

Tested at 1 lM.

Protonated form(s) (total charge +1 or +2).

Not protonated (neutral) form.

*

**

Scheme 1. Synthesis of the target tetrahydropyrimido[2,1-f]purinediones. Reagents and conditions: (i) DMF; (ii) 2-butanone, K₂CO₃.

The structures of the obtained tricyclic xanthines were con-

spectra of tetrahydropyrimido[2,1-f]purinediones a bathochromic

shift of a k_maxvalue of ca. 275 nm to about 300 nm, typical for

8-aminoxanthines,²⁹could be observed.

ﬁrmed by UV, IR and NMR spectra. All compounds showed IR

absorption bands typical for xanthine derivatives.²⁸In the UV

´

4352

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

Scheme 2. Synthesis of the target tetrahydropyrazino[2,1-f]purinediones. Reagents and conditions: (i) DMF; (ii) 2-butanone, K₂CO₃.

2.2. Biological activity

The effect on the rA₁AR strongly depends on the size of the

alkyl substituents at the 1- and 3-positions; within both, the N9-

aryl and the N9-arylethyl series: the afﬁnity increases with

increasing length of the alkyl chains. Among the most potent

derivatives are the 1,3-dibutyl-substituted compounds with A₁

AR afﬁnity (K_i) in the range of 56–268 nM, while 1,3-dipropyl

derivatives bind with an afﬁnity of 175–567 nM (except for the

non-active compounds substituted with the morpholine fragment

(22, 35 as well as the unsubstituted 6). A similar tendency of

increasing afﬁnity with increased alkyl chain length can also be

observed among the N9-arylethyl compounds for both, human

A_2Band A₃ARs (hA_2B: 109–337 nM for 1,3-dibutyl, 338–853 nM

for 1,3-dipropyl), while in case of the N9-aryl series, due to low

afﬁnity, this effect for rA_2AAR, hA_2BAR and hA₃AR is much less

pronounced and can only be estimated. At the same time, the afﬁn-

ity of the N9-arylethyl compounds to rat A_2AARs seems to drop

with enlargement of the substituents at the 1- and 3-positions.

All of these observations are well in agreement with previous liter-

ature data describing structure–activity relationships (SARs) for

xanthine-based adenosine receptors antagonists.²⁴

2.2.1. Adenosine receptor afﬁnity

The synthesized tetrahydropyrimido- and tetrahydropyrazino

[2,1-f]purinediones were evaluated in vitro in radioligand binding

assays for their afﬁnity at the rat A₁and A_2Aadenosine receptors in

brain cortical membrane, and brain striatal membrane prepara-

tions, respectively. Additionally, all compounds were tested for

their afﬁnity at human recombinant A_2Band A₃ARs stably

expressed in Chinese hamster ovary (CHO) cells. Selected tricyclic

xanthines were examined at all four AR subtypes of rat and human.

The following radioligands were employed: A₁: [³H]2-chloro-N6-

cyclopentyladenosine ([³H]CCPA); A_2A: [³H]3-(3-hydroxypropyl)-

7-methyl-8-(m-methoxystyryl)-1-propargylxanthine

2); A_2B

([³H]MSX-

[³H]4-(2-[7-amino-2-(2-furyl)-[1,2,4]-triazolo-[2,3-a]-

:

[1,3,5]-triazin-5-ylamino]ethyl)-phenol ([³H]ZM241385) and [³H]

8-(4-(4-(4-chlorophenyl)piperazine-1-sulfonyl)phenyl)-1-propylx-

anthine ([³H]PSB-603); A₃: [³H]phenyl-8-ethyl-4-methyl-(8R)-

4,5,7,8-tetrahydro-1H-imidazo[2,1-i]purine-5-one ([³H]PSB-11)

and [³H]5⁰-N-ethylcarboxamidoadenosine ([³H]NECA).

Exchange of the 4-methoxy group of the reference compounds

3a and 3b for a polar hydroxyl function at the benzene ring does

not appear to improve solubility of the compounds (QPlogS value,

Tables 1 and 2, compounds 4 and 8). At the same time, this

modiﬁcation triggered only small changes in rA_2AAR afﬁnity:

decrease for the N9-phenyl (K_i= 998 nM for 3a and 1620 nM for

2.2.2. Structure–activity relationships

All of the synthesized tricyclic xanthines are based on the same

core structure of 1,3-dialkyl-1H-purine-2,6(3H,7H)-dione, how-

ever, due to differences in their substitution pattern they can be

subdivided into three separate series:

4) and

a small increase for the N9-phenylethyl derivative

(1) tetrahydropyrimido[2,1-f]purinediones with N9-aryl substi-

tution: 3a, 4–7, 12–27 (Table 1);

(2) tetrahydropyrimido[2,1-f]purinediones with N9-arylethyl

substitution: 3b, 8–11, 28–37 (Table 2);

(3) tetrahydropyrazino[2,1-f]purinediones with N8-arylethyl

substitution: 38–42 (Table 3).

(K_i= 370 nM for 3b and 230 nM for 8). In the latter case the

introduction of a hydroxyl group also led to a total loss of afﬁnity

at the rA₁AR.

In general, most of the investigated 4-hydroxyphenyl deriva-

tives show submicromolar afﬁnity at rA_2AARs. The rA_2AAR afﬁnity

within the N9-aryl series for 4–7 slightly increases with prolonga-

tion of the R¹alkyl chain: K_i= 1620 nM (4) > 893 nM (5) > 785 nM

(6) > 738 nM (7), while among the N9-arylethyl compounds 8–11

this effect is inverse, and a weak drop in afﬁnity can be observed:

K_i= 230 nM (8) < 271 nM (9) < 464 nM (10) < 518 nM (11). The 1,3-

dibutyl- derivative 11 is also found to be a moderately potent

antagonist at the rat A₁AR and the human A_2BAR as well as a weak

Within the ﬁrst two groups, as with other reported tetrahy-

dropyrimido[2,1-f]purinediones,^18–22most of the compounds

show measurable afﬁnity at rA₁and/or rA_2AARs, while afﬁnity

for hA_2Band hA₃ARs is observed mainly within the N9-arylethyl

series.

´

4353

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

antagonist at the A₃AR with K_ivalues of 184 nM, 131 nM and

619 nM, respectively.

tetrahydropyrimidine for the tetrahydropyrazine ring was

intended as a modiﬁcation leading to increased water solubility.

Insertion of a basic amino group into the N9-substituent

resulted in a signiﬁcant increase in the calculated solubility of

compounds. Under neutral pH conditions (pH = 7 2) the tertiary

aliphatic amines are protonated and form highly soluble ammo-

nium salts, including morpholine, which is present as a mixture

Compounds with this structure were expected to bind to A₁A

and/or A_2AARs.²⁷Indeed, N8-arylethyl-substituted tetrahydropy-

razino[2,1-f]purinediones containing an additional amino group

(39–42) were predicted to show very high solubility and low

lipophilicity, compared to the rest of the synthesized compounds

within our library (Table 3). At the same time the rat A_2AAR afﬁnity

of 39–42 is reduced by at least 8–23-fold, compared to the tetrahy-

dropyrimidine regioisomers 28, 29 and 30, 31, and completely

abolished in case of the 4-hydroxyphenyl derivative 38 compared

to 8. On the other hand, the rat A₁AR afﬁnity of some amine com-

pounds—39 (dimethylamine) and 41 (piperidine)—increases to

micromolar levels (K_i= 1370 nM and 1740 nM, respectively) in

comparison to the analogues 28 and 31. Compounds 38, 40 and

42 do not bind to rA₁ARs in our assay.

of

a

cation and the free base (for N-ethylmorpholine

pK_a= 7.67).³⁰From an SAR point of view, the presence of 4-(2-ami-

noethoxy) fragments at the benzene ring reduces rat A_2AAR afﬁn-

ity in both, the N9-aryl and the N9-arylethyl series to varying

degrees, compared to the 4-hydroxyl compounds. Within the N9-

arylethyl series the smallest reduction in rA_2AAR binding afﬁnity

(in comparison with the 4-OH analogues) can be observed in case

of the diethylamine: K_i= 339 nM (29) versus 230 nM (8), 658 nM

(34) versus 464 nM (10), 566 nM (36) versus 518 nM (11).

On the other hand, the exchange of the hydroxyl for a 4-(2-ami-

noethoxy) group seems to exert a strong positive inﬂuence on rat

A₁AR afﬁnity, especially among the N9-aryl-substituted com-

pounds. N,N-Dimethylamine and pyrrolidine were found to be ben-

eﬁcial for A₁AR potency. Even in case of a small ethyl R¹

substituent, N,N-dimethylamine derivative 17 shows quite high

potency at the rat A₁AR, and with systematic enlargement of R¹

the afﬁnity increases: K_i= 297 nM (17) > 180 nM (18) > 56 nM

(23). A similar tendency is seen for pyrrolidine compounds:

K_i= 175 nM (20) > 72 nM (25). Moderate to high afﬁnity for the

rA₁AR is displayed among the 1,3-dipropyl- and 1,3-dibutyl-sub-

stituted analogues belonging to the N9-aryl or the N9-arylethyl

series and containing N,N-diethylamine (19, 24, 34, 36), piperidine

(21, 26) and morpholine (27, 37) groups.

The most potent antagonists at the rA₁AR, 23 and 25, belonging

to the N9-aryl-substituted derivatives, with K_ivalues of 56 nM and

72 nM, respectively, display at least an 11-fold preference for rA₁

AR over rA_2AA, hA_2BA and hA₃ARs. At the same time, xanthine

36, the most potent A₁AR antagonist within the N9-arylethyl series

(K_i= 80 nM), turned out to be less selective, showing a similarly

high afﬁnity also for the hA_2BAR (K_i= 109 nM) and moderate afﬁn-

ity for the hA₃AR and the rA_2AAR (K_i= 310 nM and 566 nM,

respectively). In general, afﬁnity for hA_2Band hA₃ARs can be

observed mainly (A₃AR) or even exclusively (A_2BAR) among the

N9-arylethyl-substituted xanthine derivatives. 1,3-Dipropyl- and

1,3-dibutyl-substituted derivatives of this series are almost

equipotent at rA₁and hA_2BARs (while being several-fold less

potent at rA_2AAR and hA₃AR). Thus, compounds 11 (unsubstituted

hydroxyl derivative) and 36 (diethylamine) are the most potent

hA_2BAR ligands out of all synthesized xanthines, with K_ivalues

of 131 nM and 109 nM, respectively.

None of the synthesized tetrahydropyrazino[2,1-f]purinediones

shows measurable afﬁnity for the hA_2Bor to hA₃subtypes of ARs.

2.2.3. Species differences

ARs show a high percentage of amino acid sequence identity for

human and rat orthologues: A₁AR 95%, A_2AAR 82%, A_2BAR 86%, and a

bit lower for A₃AR: 73%.³¹Despite their high homology, differences

in afﬁnity between human and other species orthologues of ARs are

described in the literature, and the same phenomenon was also

reported for other families of GPCRs.^{10,27,31–34}The A₁AR-selective

antagonist rolofylline displays a 42-fold preference for the rat over

the human A₁subtype (rA₁AR K_i= 0.19 nM, hA₁AR K_i= 8 nM)^10,24

and the non-selective AR antagonist XAC does not bind to rat A₃

ARs but shows a high afﬁnity of 71 nM for the human A₃subtype.²⁴

As rat models are often employed in preclinical studies, we

decided to determine the potency for the most potent compounds

at the whole range of human and rat ARs in order to establish

potential species differences in afﬁnity and selectivity.

Within the present library of synthesized tricyclic xanthines, rat

A₁AR afﬁnity of potent N9-aryl-substituted derivatives (K_iin range

of 56–175 nM) turns out to be 5–11-fold higher than at the human

subtype (K_iin range of 539–1990 nM). The best rA₁AR ligand of the

N9-arylethyl series (36) is over 19-fold more potent at rat than at

human A₁ARs (rA₁K_i= 80 nM, hA₁K_i= 1550 nM). At the same

time, 36 shows a 10-fold preference for the human versus the rat

A_2BARs (rA_2BK_i= 1100 nM, hA_2BK_i= 109 nM), and at least

3-fold preference for the human over the rat A₃subtype

(rA₃K_i> 1000 nM, hA₃K_i= 310 nM).

In case of A_2AAR afﬁnity, the N9-phenylethyl-substituted

derivatives show usually higher potency at rat than at human

receptors. The highest, almost 10-fold difference can be observed

for 4-hydroxyphenyl compound 9: rA_2AAR K_i= 271 nM, hA_2AAR

K_i= 2180 nM. On the other hand, compounds with an N9-aryl

structure show equipotency or even a weak preference at human

A_2AAR compared to the rA_2AAR (23–26).

No clear correlation between afﬁnities and the estimated solu-

bility of the synthesized compounds has been found (Tables 1–3).

Exchange of the 4-methoxy substituent in 3a and 3b for a 4-hydro-

xyl group resulted in a small drop in calculated solubility, while

insertion of an amine moiety, especially a morpholine ring, into

the tetrahydropyrimido[2,1-f]purinedione structure signiﬁcantly

increased the estimated solubility compared to 3a, 3b. The series

of N9-arylethyl compounds generally appear to show higher solu-

bility compared to N9-aryl tetrahydropyrimido[2,1-f]purinediones,

with the most soluble one being the morpholine derivatives 32 and

33 (in the protonated form of amine). In both series, with prolon-

gation of alkyl chains in positions 1 and 3 the solubility of com-

pounds decreases, while in most cases their afﬁnity for rat and

human adenosine receptors increases. The most potent rA₁AR

antagonist 23 displays high lipophilicity (QPlogP_o/w= 4.19) and a

quite low predicted solubility (QPlogS = ꢀ3.20).

2.2.4. Functional assays

Our previous studies on pyrimido[2,1-f]purinediones revealed

that all compounds belonging to this group displayed antagonistic

properties. Nevertheless, we performed cAMP accumulation assays

in CHO cells expressing human recombinant A_2AAR or A_2BARs for

the most active ligands in order to conﬁrm their expected antago-

nistic properties. The second messenger responses to various con-

centrations of the non-selective agonist NECA were measured in

the presence or absence of ﬁxed concentrations of test compound.

Gaddum/Schild shift analysis was performed for the obtained data

to determine the K_bvalues of the antagonists.

In addition to the series of tetrahydropyrimido[2,1-f]purine-

diones, several tetrahydropyrazino[2,1-f]purinediones were

Two selected potent compounds were further characterized in

functional assays at A_2AARs. The results obtained in cAMP accumu-

lation assays conﬁrmed that the compounds behave as competitive

synthesized

and

investigated.

The

exchange

of

the

´

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

4354

antagonists (Figs. 2A and 2B). They cause a right-shift of the con-

centration–response curve for the agonist NECA, resulting in an

increase of the agonist’s EC₅₀value in the presence of the antago-

lead structures (3a, 3b) and their modiﬁcations (19, 23, 29 and

36) was determined by UV spectroscopy on the basis of the method

described earlier.³⁵Results clearly indicate that introduction of 2-

aminoethoxy moiety improved water solubility within the consid-

ered series of purinediones, compared to lead compounds 3a and

3b possessing the methoxy group. Although all compounds dis-

played rather low water solubility (<1 mg/mL), the values deter-

mined for both lead structures are signiﬁcantly lower than those

of their modiﬁcations 19, 23, 29 and 36 (Table 4). The presence

of a linker between N9 atom and the benzene ring as well as the

length of alkyl chains at 1,3-positions seem to play an important

role. N9-arylethyl compounds (29, 36) demonstrated the highest

solubility in this group, whereas analogues belonging to the

N9-aryl series (19, 23) showed the 3-fold lower solubility in the

assay.

nists. Xanthine 29 at a concentration of 5

value of NECA from 68 nM to 460 nM, whereas the hydroxyl

derivative 8, used at a concentration of 7 M, causes a shift of

lM changes the EC₅₀

l

the EC₅₀value of NECA from 77 nM to 302 nM. The K_bvalue deter-

mined for 29 in the functional assay (839 nM) is well in accordance

with the K_ivalue determined in radioligand binding studies

(559 nM). Compound 8 has been found to be somewhat weaker

in the functional assay (K_b= 2220 nM) than in the radioligand

binding assay (K_i= 630 nM), but the results are still consistent.

Similarly to the above-mentioned observations, xanthine

36—the most potent non-selective A_2BAR ligand of the present

work—was also classiﬁed as a competitive antagonist, according

to the Gaddum/Schild EC₅₀shift analysis in cAMP accumulation

assays using CHO cells stably expressing the human A_2BAR. Com-

pound 36 caused an EC₅₀value shift for NECA from 73 nM to

335 nM, and the corresponding K_bvalue was calculated to be

547 nM (Fig. 2C).

2.4. Molecular modeling

To address the observed A₁/A_2AAR subtype selectivity as well as

species differences, docking both to the published X-ray structure

of the human A_2AAR (PDB entry 3EML³⁶), and to homology models

of human and rat A₁and A_2AARs was performed. For that we

decided to use the premade homology models of ARs published

by Moro and coworkers, which are freely accessible on the

Adenosiland platform.³⁷

2.3. Water solubility of selected compounds

The water-solubility of selected N9-aryl and N9-arylethyl sub-

stituted tetrahydropyrimido[2,1-f]purinediones, including two

A (cmpd. 8)

B (cmpd. 29)

K_b= 839 41 nM; K_i= 559 89 nM

K_b= 2 220 310 nM; K_i= 630 350 nM

C (cmpd. 36)

K_b= 547 48 nM; K_i= 109 11 nM

Figure 2. cAMP accumulation studies in CHO cells expressing the human adenosine A_2A(A, B) and A_2B(C) receptors. The dose–response curves for NECA-induced stimulation

of cAMP accumulation were generated in the absence or in the presence of ﬁxed concentrations of 8 (A), 29 (B), and 36 (C). Graphs from two independent experiments

performed in duplicates with mean values SEM are shown. All three investigated compounds shifted the concentration–response curve for NECA in a parallel manner to the

right, indicating competitive antagonism.

´

4355

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

Table 4

pocket structures, the type of residues and the position of side

chains in the range of 5 Å from the docked poses of selected ligands

were taken into account (Fig. 3, Ballesteros–Weinstein GPCR num-

bering of residues is used^42–44).

Water solubility determined for selected compounds

Compound

Solubility [mg/mL]

Solubility [lmol/mL]

3a

3b

19

23

29

36

0.051

0.063

0.260

0.255

0.709

0.974

0.150

0.169

0.539

0.528

1.560

1.808

To examine the ligand–protein non-bonded and polar interac-

tions energy for each ﬁnal docking pose, the overall terms of the

docking score function (electrostatic interactions, van der Waals

interactions and hydrogen bonding energy), as well as the energy

contributions calculated for the individual residues, were analyzed.

Since the ﬁrst crystal structure of the human A_2AAR in complex

with the antagonist ZM241385 was published,³⁶several other X-

ray structures of the hA_2AAR, comprising agonist-bound and inac-

tive conformations have become available. Their analysis allowed

the identiﬁcation of residues responsible for ligand recognition

and crucial for ligand selectivity. Structural insight derived from

adenosine receptor X-ray structures has also allowed for more pre-

cise molecular modeling of 3D structures of other AR subtypes.^45,46

The comparison of available crystal structures of the human

adenosine A_2AAR in complex with xanthine- and triazine-based

antagonists (and inverse agonists)^36,47–49allows to perceive ﬂexi-

bility of some amino acid side chains within the receptor binding

All the structures—the homology models of the hA₁, rA₁and

rA_2AAR as well as 3EML—were prepared by the same method

implemented in Maestro and validated by docking in Glide.³⁸Sol-

vent molecules present in 3EML were deleted prior to docking. As

referential set of subtype-selective, xanthine-based literature

antagonists, the following compounds were used: A₁AR-selective

tonapofylline and rolofylline as well as A_2AAR-selective istrade-

fylline.^10,24All the homology models as well as the experimental

3EML structure showed poor discrimination of the described set

of AR ligands. For this reason 3EML was replaced by the 3EML-

based homology model of the hA_2AAR,³⁷and all four models were

further optimized in a ligand-guided way by induced-ﬁt docking³⁸

using the above-mentioned set of reference xanthines. After the

validation step, the optimal, highest-ranked models were selected

from the obtained clusters of AR models. Their stereochemical

quality and compatibility was assessed by PROCHECK,³⁹

RAMPAGE⁴⁰and ANOLEA.⁴¹Each ‘improved’ homology model

together with the high-scored docking poses of the reference

antagonists (rolofylline and tonapofylline for hA₁/rA₁, istradefylline

for hA_2A/rA_2AARs) were additionally analyzed in the Maestro pro-

gram with respect to the model 3D structure, disulﬁde bonds and

binding cavity.

Finally, the library of the synthesized tricyclic xanthine-based

compounds was prepared. The tricyclic core of tetrahydropyrim-

ido[2,1-f]purinedione was built, based on the reported crystallo-

graphic data for N9-phenyl- and N9-(2-benzyloxy)ethyl

derivatives.^18,19Models of tetrahydropyrazino[2,1-f]purinediones

were constructed using the 3D information obtained from the

solved X-ray structures of 38 and 42. The ligand library was pre-

pared with the LigPrep module (low energy ionization/protonation

state at pH = 7 1, tautomeric state), optimized by conformational

search (MacroModel) and docked using Glide SP mode with calcu-

lation of per-residue interactions.³⁸Obtained poses were ranked

according to docking scores implemented in Glide and analyzed

in terms of important interactions described in the literature for

each subtype of AR. According to this ranking the following dock-

ing poses were selected for further analysis: 9 at hA_2AAR, 9 at rA_2A

AR, 23 at hA₁AR, 23 at rA₁AR. For detailed analysis of the binding

cleft. Wide variations can be observed, e.g., in case of Glu169^5.30

,

and Asn253^6.55, two amino acids described as crucial residues for

A_2AAR antagonist binding (residue numbering as for the human

A_2AAR sequence P29274,⁵⁰in superscript Ballesteros–Weinstein

GPCR numbering, extended for extracellular loops^36,42–44). In the

structures of hA_2AAR with bound xanthine derivatives, XAC and

caffeine (PDB entries 3REY and 3RFM, respectively), the

Glu169^5.30side chain does not form a polar interaction with a

ligand, like in case of the triazine-based ligands, and adopts a

rotated conformation compared to the 3EML structure.³⁶The side

chain of Asn253^6.55in 3REY has a reversed position compared to

other structures including 3RFM. Nevertheless, in both xanthine-

bound crystal structures the ligands share a common binding

mode, with the same carbonyl group (in the 6-position of a bicyclic

purinedione core) forming an H-bond contact with the terminal

amino group of Asn253^6.55. In our calculations the observed bind-

ing system of tricyclic core resembles the canonical way of binding

reported for XAC and caffeine, with the H-bond between the con-

served Asn^6.55and the ligand carbonyl group in the 4-position (cor-

responding to C@O in the 6-position of bicyclic 2,6-purinediones).

2.4.1. Binding mode of the xanthine 9 at the human and rat A_2A

AR models

Comparison of the observed docking poses for the hA_2Aand the

rA_2AARs is illustrated with the compound 9, a potent rA_2AAR

antagonist showing 10-fold selectivity over the hA_2AAR. In case

of the hA_2AAR model, the tricyclic core of 9 adopts the similar

Figure 3. Alignment of hA_2A, rA_2A, hA₁and rA₁AR residues in the range of 5 Å from the ligands (according to the best docking poses of selected tricyclic purinediones at the

homology models: (1) hA_2AAR + compd 9, (2) rA_2AAR + compd 9, (3) hA₁AR + compd 23, (4) rA₁AR + compd 23. Residue numbering for human A_2AAR sequence P29274,

Ballesteros–Weinstein GPCR numbering given at the top). Residues situated farther than 5 Å from the docked ligand in the particular model are marked in brown.

´

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

4356

location inside the binding pocket as the triazolotriazine system of

ZM241385 in the 3EML X-ray structure (Fig. 4); the N3-ethyl sub-

stituent is buried in the same cavity as the furan ring of ZM241385.

The purinedione scaffold of 9 is anchored by a hydrogen bond

between the Asn253^6.55and the carbonyl group in the 4-position

(numbering according to the hA_2AAR sequence P29274). Addition-

proximity of the top TM-II, and the 4-hydroxy group of the ligand

forms the hydrogen bond with Ile66^2.64. In case of the rat model,

the benzene ring is shifted towards the top of TM-VII and a hydro-

gen bond is formed between the 4-hydroxyl function of 9 and

Tyr266^7.36(residue numbering as for the rA_2AAR sequence

P30543⁵⁰).

ally, the ﬂat heterocyclic core of 9 is involved in

p

–

p

stacking with

In this part of the binding site, small but signiﬁcant structural

differences can be observed for the human and rat A_2AARs

sequences. In the hA_2AAR Met^7.35forms numerous hydrophobic

contacts with the benzene ring of the ligand, while in the rat coun-

terpart it is rotated and directed towards the fused pyrimidine ring

of 9 (Fig. 5). Orientation of Met^7.35seems to be at least partially

triggered by the close presence of a residue in position 7.32 at

the very top of helix TM-VII: a ﬂexible Leu267 in the human A_2A

AR sequence is replaced with a rigid Pro262 in the rA_2AAR. The dif-

ference in the 3D structure of this part of the binding site is

reﬂected in the stronger van der Waals interactions between 9

and the individual residues of rA_2AAR.

the conserved Phe168^5.29(the extracellular loop 2, EL2) and in

hydrophobic aliphatic contacts with Glu169^5.30(EL2), Met177^5.38

and Leu85^3.33from one side and Leu249^6.51from the other side

(Figs. 4 and 5). Almost identical location and analogous binding

of the tricyclic scaffold of 9 can also be seen for the rA_2AAR model

(Fig. 5). According to the site-directed mutagenesis data for the

hA_2AAR, some of above-mentioned amino acids are important or

even crucial for ligand binding. Replacement of Glu169^5.30

,

His250^6.52Asn253^6.55Ile274^7.39or His278^7.43with alanine

,

resulted in a total loss of XAC afﬁnity at the hA_2AAR, while muta-

tion of Val84^3.32decreased the afﬁnity of antagonists. Recently

published mutagenesis data also suggest, that conserved residues

Phe168^5.29and Leu249^6.51play a central role in coordinating the

bicyclic core present in both agonist and antagonist structures,

while mutation of Met177^5.38to alanine impeded antagonist

binding.^51,52

Small differences between docking poses of 9 at the human and

rat A_2AAR models are observed for the location of the hydrox-

yphenyl group (Fig. 5). In both cases the benzene ring of the ligand

is situated near the extracellular region (the extracellular loops EL1

and EL2), close to the salt bridge between Glu^5.30(EL2) and His^6.66

(EL3). In the hA_2AAR binding site this aromatic function is placed in

Apart from the mentioned Leu267/Pro262^7.32exchange, in the

range of 5 Å distance from the ligand 9, there is only one more

replacement in position 3.28: Ile80 in the hA_2AAR is replaced by

Phe77 in the rA_2AAR. However, in both of our models, side chains

of these residues seem to be too far away to have any important

effect on afﬁnity, and component terms of van der Waals interac-

tions and Coulomb interactions energy, calculated for both resi-

dues, are small and insigniﬁcant.

Analysis of overall Glide energy terms, calculated for the ﬁnal

poses of 9 docked to the hA_2Aand the rA_2AARs models, showed

similar values for the hydrogen bonding and van der Waals

Figure 4. Superposition of 3EML X-ray structure and the hA_2AAR model with the docked compound 9. The ligands ZM241385 (pink) and the compound 9 (yellow) are shown

in sticks. Selected residues of 3EML protein (purple) and the hA_2AAR homology model (green) are shown in wires. For clarity only side chains are depicted (except Ile^2.64).

Residue numbering according to the Ballesteros–Weinstein GPCR numbering.

´

4357

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

Figure 5. Superposition of the compound 9 docking poses in the human and rat A_2AAR models. The selected residues of the hA_2Amodel (green wires) and the rA_2Amodel

(purple wires) are shown. The docking poses of the compound 9 are depicted in yellow sticks (docked to the hA_2Amodel) and pink sticks (docked to the rA_2Amodel). For

clarity only side chains are depicted (except Ile^2.64).

interactions. The electrostatic interactions, however, are more

favorable in case of the rat A_2AAR model and this result signiﬁ-

cantly contributes to the higher docking score of 13 to the rat

A_2AAR compared to the human counterpart.

In the obtained A₁AR models there are only two changes in the

sequences within 5 Å distance from the docked ligand 23. In TM-

VII the residue Thr270^7.35of the hA₁AR sequence corresponds to

Ile270 in the rA₁AR (and Met270 in the hA_2AAR). Both side chains

differ in lipophilicity, but in our models adopt a comparable posi-

tion forming interactions with the ligand 23 in a similar way. Even

though the hydroxyl group of threonine is not involved in any

hydrogen bond with 23 in the observed docking poses, its direct

proximity to the ligand scaffold might contribute to the species dif-

ference in the afﬁnity of the compounds at the ARs.

Another structural difference is observed in the third extracellu-

lar loop EL3, where His264^6.66in the hA₁AR corresponds to Gln264

in the rA₁AR (and His264 in hA_2AAR). In our hA₁AR model

His264^6.66does not form a salt bridge with Glu172^5.30(EL2), as in

case of A_2AAR models, but instead forms a hydrogen bond with

the backbone carbonyl group of Thr257^6.58, thereby stabilizing

the position of the EL3. In the rA₁AR model the corresponding glu-

tamine Gln264 plays a similar role, interacting by its side chain

amide group with the backbone of Lys265^6.67(EL3) from one side,

and the hydroxyl group of Thr257^6.58from the other side. At the

same time Gln264 is placed 5Å from the docking pose of 23 (acting

by weak repulsive Coulomb forces), while in case of the hA₁AR

model the distance of the docking pose of 23 from His264^6.66is

almost 7 Å (van der Waals and electrostatic interactions between

them are negligible).

2.4.2. Binding mode of the xanthine 23 at the human and rat A₁

AR models

To compare the docking poses of the tricyclic xanthines and

their interaction with the rA₁and the hA₁AR binding pockets,

the example of the compound 23 was chosen for its highest afﬁnity

at the rA₁AR. In the docking pose observed for both, hA₁and rA₁AR

models, 23 forms a hydrogen bond between Asn^6.55and the car-

bonyl group in the 4-position (Fig. 6). The purinedione core of

the ligand is involved in hydrophobic interactions with Phe^5.29

from one side and Leu^6.51, Ile^7.39and Asn^6.55from the other side.

The long butyl substituent in N3-position of the docked 23 occu-

pies a similar cavity as the N3-ethyl chain of 13 in A_2AAR models,

forcing a shift of the whole ligand tricyclic core towards the extra-

cellular top of the receptor, compared to the position of 9.

The benzene ring of 23 is stabilized by van der Waals interac-

tions with surrounding residues of EL2 (Glu^5.28, Glu^5.30) as well

as Tyr^7.36and the residue in position 7.35. In both A₁AR models

the side chains of the conserved Glu172/Glu172 in position 5.30

(numbering according to P30542 and P25099 for human and rat

sequences, respectively⁵⁰) forms single hydrophobic contacts and

attractive electrostatic interactions with the ligand scaffold, but

no H-bond is observed. The glutamate in position 5.28 (corre-

sponding to leucine in hA_2Aand rA_2AAR) adopts in A₁AR models

two different conformations, but in both cases the carboxyl group

of Glu^5.28forms a salt bridge with the protonated amino group of

23 (Fig. 6).

The proximity of Gln264^6.66to rA₁AR-bound ligands (not neces-

sarily xanthines) and the hydrogen bond feature of its side chain

make this residue one of probable factors inﬂuencing the observed

hA₁/rA₁AR selectivity. A putative ligand-Gln264^6.66hydrogen bond

might be formed directly or by means of the water network. A

presence of water molecules in this part of the binding site was

´

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

4358

Figure 6. Superposition of the compound 23 docking poses in the human and rat A₁AR models. The selected residues of the hA₁model (green wires) and the rA₁model

(purple wires) are shown. The docking poses of the compound 23 are depicted in yellow sticks (docked to the hA₁model) and pink sticks (docked to the rA₁model). For clarity

only side chains are depicted.

described in case of 3EML crystal structure of the hA_2AAR, where in

the direct vicinity of the His264^6.66residue there is a water mole-

cule involved in the hydrogen bond with the ligand ZM241385.

In terms of the stabilization energy, the compound 23 is better

accommodated in the binding site of rA₁AR than hA₁AR. The

higher docking score of 23 in rA₁AR is driven mainly by van der

Waals energy, but also electrostatic interactions and hydrogen

bonds terms contribute to this result. The better docking score is

reﬂected in the higher afﬁnity of 23 to this receptor compared to

hA₁AR afﬁnity.

Met265^7.35. One of the most important replacements is undoubt-

edly the exchange in position 5.28 of EL2, which, in all our homol-

ogy models is in the direct proximity of the docked ligands 9 and

23. In this position leucine, present in both rat and human A_2A

AR sequence is exchanged into glutamate in A₁AR sequence. The

energy contributions, calculated for interactions: compound 23-

Glu^5.28in the A₁AR models and compound 9-Leu^5.28in the A_2A

AR models show that the presence of glutamate is beneﬁcial for

the ligand binding. Apart from the hydrogen bond between the car-

boxyl function of Glu^5.28and the charged amine group of the xan-

thine 23, reﬂected in H-bond and electrostatic energy

contributions, also van der Waals interactions between glutamate

and the ligand are favorable, compared to leucine in this position.

Observed differences within the sequences are located to a large

extent in the extracellular region, especially within the second and

the third extracellular loop. Among adenosine receptors EL2 is

described to be essential for receptor activation^53–56and plays an

important role in ligand recognition, binding and subtype selectiv-

ity,^53,54while EL3 may be involved in receptor activation and bind-

ing of ligands.^54,55However, due to their ﬂexibility and high

variability of sequences, loops are difﬁcult subject of unambiguous

prediction using the homology modeling method. The most prob-

lematic for mapping is the highly divergent EL2 region, containing

33 residues in the rA₁AR sequence and 31 residues in the rA_2A

AR.^43,44Moreover, the extracellular part of the receptor is exposed

to water, often involved in H-bond network with the receptor and

the ligand. Therefore, in most cases both the position and the inﬂu-

ence of individual structural differences within the extracellular

loops on the observed ligand afﬁnity and selectivity is difﬁcult to

unambiguously estimate.

2.4.3. Comparison of the binding modes at the rA₁and rA_2AAR

models

The analysis of binding sites in the rat A₁and A_2AAR models

shows numerous differences in the sequences of both individual

receptors (Figs. 3, 5 and 6). The most important mutations within

thebindingpocketof thesubtypesrA₁and rA_2Aare placedin proxim-

ity of the fused pyrimidine ring in the tricyclic core and in the prox-

imity of the N9-aryl tail of the ligands, while the residues in the deep

binding site, surrounding the purinedione scaffold, are highly con-

served. This seems to indicate that the position, size and kind of sub-

stituent in the pyrimidine ring, interacting with the top part of the

receptor, has a big inﬂuence on the subtype selectivity, while con-

tacts of the purinedione system with the conserved transmembrane

domains are responsible for the afﬁnity to both subtypes.

In the range of 5 Å distance from the ligands 9 or 23, the differ-

ences between the rA₁and rA_2AAR sequences are: Asn70/Ser64^2.65

,

Ile71/Thr65^2.66, Val83/Phe77^3.28

, Lys168/Thr160 (EL2), Glu170/

Leu162 (EL2), Lys173/Asp165 (EL2), Leu253/Ile247^6.54, Gln264/

His259 (EL3), Lys265/Ala260 (EL3), Ser267/Pro262^7.32and Ile270/

´

4359

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

The synthesis and physicochemical properties of the com-

pounds 3a, 3b, 4 and 8 were reported previously.^18,19Procedures,

yields and physical data of the selected compounds are given

below, for all other obtained xanthine derivatives the analytical

data can be found in Supplementary Data.

3. Conclusion

A new series of 32 tetrahydropyrimido- and 5 tetrahydropy-

razino[2,1-f]purinediones was designed, synthesized and evalu-

ated for their rat and human adenosine receptor afﬁnities.

Insertion of a basic amino group into the N9-substituent as well

as the exchange of the fused pyrimidine ring to a pyrazine ring

resulted in a signiﬁcant increase in the calculated solubility of

compounds, compared to the lead structures: 4-methoxyphenyl-

or 4-methoxyphenylethyl-substituted derivatives of 1,3-dialkyl-

6,7,8,9-tetrahydropyrimido-[1,2-f]purine-2,4(1H,3H)-diones. The

increased water solubility of the N9-aryl and especially

N9-arylethyl analogues was conﬁrmed by the experimental assay

performed for selected compounds.

In the group of the pyrimido derivatives most compounds

showed measurable afﬁnity at rA₁and/or rA_2AARs, and among

the 1,3-dibutyl derivatives possessing an amine group potent rA₁

AR antagonists were identiﬁed with K_ivalues in the range of

56–80 nM. Afﬁnity for hA_2Band hA₃ARs was shown mainly within

the N9-arylethyl series, in particular for 1,3-dibutyl compounds

with K_iat hA_2BAR in range of 109–337 nM.

4.1.1. General procedure for the synthesis of 4-hydroxyphenyl/

4-hydroxyphenylethyl substituted derivatives of 1,3-dialkyl-

6,7,8,9-tetrahydropyrimido-[1,2-f]purine-2,4(1H,3H)-diones (4–

11)

A mixture of 2 mmol of 1,3-dialkyl-8-bromo-7-3-chloropropyl-

xanthine and 4 mmol of commercial available 4-aminophenol or

tyramine (obtained ex tempore from saturated aqueous solution

of tyramine hydrochloride by alkalization with 25% Na₂CO₃) was

reﬂuxed in DMF for 5–10 h. Raw products were precipitated by

addition of water to the reaction mixture, ﬁltered off and puriﬁed

by crystallization from ethanol.

1,3-Diethyl-9-(4-hydroxyphenyl)-6,7,8,9-tetrahydropyrimido

[1,2-f]purine-2,4(1H,3H)-dione (5)

The reaction was carried out for 8 h. The raw product was crys-

Docking experiments of the investigated library to the homol-

ogy models of the human and rat A₁and A_2AARs allowed to com-

pare the expected binding modes for selected tricyclic xanthines.

Analysis of the binding site structures revealed a highly conserved

protein region in close proximity of ligands, especially within the

rat and the human orthologues of a particular receptor. The contri-

butions to the overall interaction energies calculated for the indi-

vidual residues show that single amino acid replacements found

in the closest surroundings of the docking poses 9 and 23

(Leu267/Pro262^7.32in case of hA₁/rA₁AR, Thr270/Ile270^7.35and

His264/Gln264^6.66in case of hA_2A/rA_2AAR) may have a signiﬁcant

inﬂuence on the species selectivity, but in case of the investigated

xanthines the observed 7–10 fold difference in the rat/human AR

afﬁnity is driven mainly by overall electrostatic and van der Waals

interactions. More structural divergences can be found within

the binding sites of the A₁and the A_2AAR subtypes, especially in

the extracellular regions surrounding the N9-substituent. Among

those differences the Glu/Leu^5.28in EL2 seems to be the most sig-

niﬁcant one for ligand binding from an energy point of view.

tallized from ethanol. Yield: 76%; mp 238–240 °C. Anal. for

C

₁₈H₂₁N₅O₃: Calcd: C, 60.83; H, 6.96; N, 19.71. Found: C, 60.75;

H, 6.68; N, 19.48; LC/MS: purity 100%, m/z 356.35 [M⁺+1]. ¹H

NMR (CDCl₃) d: 1.02–1.15 (m, 6H, 2CH₃CH₂), 2.12–2.23 (m, 2H,

CH₂CH₂CH₂), 3.68 (t, J = 5.3 Hz, 2H, CH₂CH₂CH₂), 3.77–3.91 (m,

4H, 2CH₃CH₂), 4.17 (t, J = 5.8 Hz, 2H, CH₂CH₂CH₂), 6.75 (d,

J = 8.7 Hz, 2H, Ar), 7.22 (d, J = 8.7 Hz, 2H, Ar), 9.42 (s, 1H, OH). ¹³C

NMR (CDCl₃) d: 13.1, 13.4, 21.6, 36.2, 38.6, 42.1, 49.1, 103.3,

116.7, 126.5, 134.9, 147.5, 150.7, 150.8, 153.9, 155.2. IR

3169 (OH), 1694 (C@O), 1650 (C@O); UV k_max(nm): 305.

m

(cm^ꢀ1):

4.1.2. The synthesis of 8-(4-hydroxyphenethyl)-1,3-dimethyl-

6,7,8,9-tetrahydropyrazino[2,1-f]purine-2,4(1H,3H)-dione (38)

A mixture of 2 mmol of 7-(2-chloroethyl)-8-(chloromethyl)-

1,3-dimethyl-1H-purine-2,6(3H,7H)-dione and 4 mmol of tyra-

mine (obtained ex tempore from saturated aqueous solution of

tyramine hydrochloride by alkalization with 25% Na₂CO₃) was

reﬂuxed in DMF medium for 20 h. After cooling and addition of

water the brown solid precipitated and was ﬁltered off. A raw pro-

duct was puriﬁed by crystallization from ethanol.

4. Experimental protocols

4.1. Chemistry

Yield: 63%; mp 259–261 °C. Anal. for C₁₈H₂₁N₅O₃ꢁH₂O: Calcd:

C, 57.89; H, 6.21; N, 18.76. Found: C, 57.77; H, 6.29; N, 19.04.

LC/MS: purity 100%, m/z 356.35 [M⁺+1]. ¹H NMR (CDCl₃) d: 2.47–

2.52 (m, 4H, CH₂CH₂Ar), 2.92–3.00 (m, 2H, CH₂CH₂), 3.21 (s, 3H,

N¹CH₃), 3.40 (s, 3H, N³CH₃), 3.44–3.54 (m, 4H, CH₂CH₂+CH₂), 4.68

(s, 1H, OH), 6.71 (d, J = 8.6 Hz, 2H, Ar), 7.09 (d, J = 8.6 Hz, 2H, Ar).

¹³C NMR (CDCl₃) d: 27.9; 29.7, 32.8, 44.3, 54.0, 57.6, 59.3, 106.6,

All reagents were purchased from commercial suppliers and

were used without further puriﬁcation. Melting points (mp.) were

determined on a MEL-TEMP II (LD Inc., USA) melting point appara-

tus and are uncorrected. ¹H NMR spectra were recorded at

300 MHz using a Varian-Mercury-VX 300 MHz PFG spectrometer

or a Bruker AMX 300 spectrometer (Bruker, Germany). Chemical

shifts were expressed in parts per million (ppm). ¹³C NMR spectra

were recorded at 75 MHz on Varian-Mercury-VX 300 MHz PFG or

400 MHz spectrometer. IR spectra were measured as KBr pellets

on FT/IR-410 Spectrometer (Jasco) or FT/IR Nicolet iS5 Spectrome-

ter (ThermoScientiﬁc). UV spectra were recorded on Jasco UV/Vis

V-530 spectrometer in a concentration of 10^ꢀ2g/L or 10^ꢀ3g/L in

methanol. Mass spectra (LC/MS) were performed on Waters TQ

Detector mass spectrometer coupled to a Waters Acquity Ultra Per-

formance Liquid Chromatography (UPLC). The UPLC purity of all

ﬁnal compounds was determined (%). Elemental analyses (C, H,

N) were performed on an Elemental Analyser Vario El III (Hanau,

Germany). For column chromatography puriﬁcation silica gel 60

(0.063–0.20 mm; Merck) was used and the mixture of dichloro-

methane with methanol was applied as a mobile phase.

114.7, 129.6, 131.6, 148.0, 148.5, 151.7, 157.3, 162.8. IR

3336 (OH), 1683 (C@O), 1624 (C@O); UV k_max(nm): 229.

m

(cm^ꢀ1):

4.1.3. General procedure for the synthesis of 4-(2-aminoethoxy)

phenyl and 4-(2-aminoethoxy)phenethyl derivatives of 1,3-

dialkyl-6,7,8,9-tetrahydropyrimido- and 1,3-dimethyl-6,7,8,9-

tetrahydropyrazino[1,2-f]purine-2,4(1H,3H)-diones (12–37, 39–

42)

A mixture of 2 mmol of:

9-(4-hydroxyphenyl)-1,3-dialkyl-6,7,8,9-tetrahydropyrimido-

(4–7) or

9-(4-hydroxyphenethyl)-1,3-dialkyl-6,7,8,9-tetrahydropyrim-

ido- (8–11) or

8-(4-hydroxyphenethyl)-1,3-dimethyl-6,7,8,9-tetrahydropy-

razino[1,2-f]purine-2,4(1H,3H)-dione (38)

´

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

4360

and the hydrochloride of the appropriate aminoethylchloride

(dimethyl, diethyl, morpholine, pyrrolidine, piperidine) in 2 fold

excess was reﬂuxed in 2-butanone for 5-18 h, in the presence of

K₂CO₃(4 mmol). The progress of the reaction was monitored by

thin layer chromatography (TLC). The hot reaction mixture was ﬁl-

tered off to remove inorganic salts, cooled and concentrated in

vacuo. The raw product was puriﬁed by crystallization or column

chromatography (dichloromethane/methanol).

2CH₃CH₂CH₂CH₂), 1.62 (quin, J = 7.5 Hz, 2H, CH₃CH₂CH₂CH₂), 1.74

(quin, J = 7.4 Hz, 2H, CH₃CH₂CH₂CH₂), 1.97–2.10 (m, 2H,

CH₂CH₂CH₂), 2.77 (br s, 4H, 2CH₂CH₃), 2.88 (t, J = 7.2 Hz, 2H,

CH₂CH₂), 2.99 (br s, 2H, CH₂CH₂), 3.19 (t, J = 5.5 Hz, 2H, OCH₂CH₂),

3.68 (t, J = 7.2 Hz, 2H, OCH₂CH₂), 3.96 (t, J = 7.0 Hz, 2H,

CH₂CH₂CH₂), 4.05 (t, J = 7.2 Hz, 2H, CH₂CH₂CH₂), 4.09–4.23 (m, 4H,

2CH₃CH₂CH₂CH₂), 6.83 (d, J = 8.5 Hz, 2H, Ar), 7.11 (d, J = 8.5 Hz, 2H,

Ar). ¹³C NMR (CDCl₃) d: 42.9, 44.5, 45.4, 47.8, 51.7, 66.4, 102.9,

114.6, 129.7, 131.0, 149.0, 151.3, 151.4, 153.8, 157.5. IR

m

(cm^ꢀ1):

9-(4-(2-(Diethylamino)ethoxy)phenyl)-1,3-dipropyl-6,7,8,9-

tetrahydropyrimido[1,2-f]purine-2,4(1H,3H)-dione (19)

1698 (C@O), 1654 (C@O), 1247 (CAO); UV k_max(nm): 302.

The reaction was carried out for 10 h. The raw product was crys-

tallized from the mixture of ethanol and water. Yield: 71%; mp 94–

96 °C. Anal. for C₂₆H₃₈N₆O₃: Calcd: C, 64.71; H, 7.94; N, 17.42.

Found: C, 64.52; H, 7.91; N, 17.23; LC/MS: purity 98%, m/z 483.59

[M⁺+1]. ¹H NMR (CDCl₃) d: 0.82–0.97 (m, 6H, 2CH₃CH₂CH₂), 1.06

(t, J = 7.1 Hz, 6H, 2CH₂CH₃), 1.57–1.76 (m, 4H, 2CH₃CH₂), 2.27

(quin, J = 5.7 Hz, 2H, CH₂CH₂CH₂), 2.64 (q, J = 7.0 Hz, 4H, 2CH₂CH₃),

2.88 (t, J = 6.2 Hz, 2H, OCH₂CH₂), 3.76 (t, J = 5.5 Hz, 2H, OCH₂CH₂),

3.90 (q, J = 7.5 Hz, 4H, 2CH₂CH₂CH₃), 4.05 (t, J = 6.3 Hz, 2H, CH₂CH₂-

CH₂), 4.32 (t, J = 6.0 Hz, 2H, CH₂CH₂CH₂), 6.89 (d, J = 9.0 Hz, 2H, Ar),

7.30 (d, J = 9.0 Hz, 2H, Ar). ¹³C NMR (CDCl₃) d: 47.8, 51.8, 66.7,

4.2. Pharmacology

4.2.1. Adenosine receptor binding assays

Adenosine receptor binding assays were performed as previ-

ously described³¹using rat brain cortical membrane preparations

for A₁and rat brain striatal membrane preparations for A_2AAR

assays. Frozen rat brains (unstripped) were obtained from Pel-

Freez, Rogers, Arkansas, USA. For assays at human A₁, A_2A, A_2B

and A₃ARs as well as at rat A_2Band A₃ARs, cell membranes of

CHO cells expressing recombinant receptors were used as

described.³¹The following compounds were used as radioligands:

A₁: [³H]2-chloro-N6-cyclopentyladenosine ([³H]CCPA); A_2A: [³H]

3- (3-hydroxypropyl)-7-methyl-8-(m-methoxystyryl)-1-propar-

103.1, 114.7, 124.7, 136.3, 149.0, 149.9, 151.3, 154.1, 156.2. IR

m

(cm^ꢀ1): 1693 (C@O), 1652 (C@O), 1245 (CAO); UV k_max(nm): 305.

gylxanthine ([³H]MSX-2); A_2B

:

[³H]4-(2-[7-amino-2-(2-furyl)-

1,3-Dibutyl-9-(4-(2-(dimethylamino)ethoxy)phenyl)-6,7,8,9-

tetrahydropyrimido[1,2-f]purine-2,4(1H,3H)-dione (23)

[1,2,4]-triazolo[2,3-a][1,3,5]triazin-5-ylamino]ethyl)phenol ([³H]

ZM241385), or [³H]8-(4-(4-(4-chlorophenyl)piperazine-1-sul-

fonyl)phenyl)-1-propylxanthine ([³H]PSB-603), respectively; A₃:

[³H]phenyl-8-ethyl-4-methyl-(8R)-4,5,7,8-tetrahydro-1H-imidazo

[2,1-i]purine-5-one ([³H]PSB-11) and [³H]5⁰-N-ethylcarboxami-

doadenosine ([³H]NECA).

Initially, a single high concentration of compound was tested in

three (A₁, A_2A) or two (A_2B, A₃) independent experiments. For

potent compounds, full concentration-inhibition curves were

determined using different concentrations of test compounds

spanning at least 3 orders of magnitude. Data were analyzed using

the PRISM program version 4.0 or higher (Graph Pad, San Diego,

CA, USA).

The reaction was carried out for 10 h. The raw product was crys-

tallized from the mixture of ethanol and water. Yield: 96%; mp

123–125 °C. Anal. for C₂₆H₃₈N₆O₃: Calcd: C, 64.71; H, 7.94; N,

17.42. Found: C, 63.36; H, 8.05; N, 17.24; LC/MS: purity 96%, m/z

483.52 [M⁺+1]. ¹H NMR (CDCl₃) d: 0.87–0.99 (m, 6H, 2CH₃CH₂CH₂-

CH₂), 1.25–1.45 (m, 4H, 2CH₃CH₂CH₂CH₂), 1.56–1.75 (m, 4H, 2CH₃-

CH₂CH₂CH₂), 2.23–2.31 (m, 2H, CH₂CH₂CH₂), 2.34 (s, 6H, 2CH₃),

2.74 (t, J = 5.6 Hz, 2H, OCH₂CH₂), 3.79 (t, J = 5.5 Hz, 2H, OCH₂CH₂),

3.96 (q, J = 7.0 Hz, 4H, 2CH₃CH₂CH₂CH₂), 4.08 (t, J = 5.6 Hz, 2H, CH₂-

CH₂CH₂), 4.34 (t, J = 5.9 Hz, 2H, CH₂CH₂CH₂), 6.92 (d, J = 9.0 Hz, 2H,

Ar), 7.33 (d, J = 9.0 Hz, 2H, Ar). ¹³C NMR (CDCl₃) d: 42.9, 45.9, 47.8,

58.3, 66.2, 103.1, 111.8, 124.6, 136.3, 148.2, 149.9, 151.3, 154.0,

156.1. IR

m

(cm^ꢀ1): 1697 (C@O), 1660 (C@O), 1249 (CAO); UV k_max

4.2.2. Functional assays

(nm): 306.

cAMP accumulation experiments were essentially performed as

previously described.³¹Stably transfected CHO cells expressing the

human A_2Aor A_2Breceptor were grown in DMEM-F12 medium

(Invitrogen) with 10% fetal calf serum, 100 U/ml penicillin G,

100 mg/ml streptomycin and 1% ultraglutamine at 37 °C with 5%

CO₂. On the day of the experiment cells were transferred to 24-well

plates at a density of 200,000 cells per well. After 24 h the medium

was removed and the cells were washed with 500 ml of 37 °C

warm Hank’s Balanced Salt Solution (HBSS; 20 mM HEPES,

13 mM NaCl, 5.5 mM glucose, 5.4 mM KCl, 4.2 mM NaHCO₃,

1.25 mM CaCl₂, 1 mM MgCl₂, 0.8 mM MgSO₄, 0.44 mM KH₂PO₄

and 0.34 mM Na₂HPO₄, pH adjusted to 7.3) containing 1 U/ml of

adenosine deaminase (ADA, Sigma). The cells were then incubated

in 300 ml of HBSS with ADA at 37 °C and 5% CO₂for 2 h. Then, the

phosphodiesterase inhibitor Ro20-1724 (Hoffmann La Roche) was

9-(4-(2-(Diethylamino)ethoxy)phenylethyl)-1,3-dimethyl-

6,7,8,9-tetrahydropyrimido[1,2-f]purine-2,4(1H,3H)-dione (29)

The reaction was carried out for 6 h. The raw product was crys-

tallized from the mixture of ethanol and water. Yield: 59%; mp

150–152 °C. Anal. for C₂₄H₃₄N₆O₃: Calcd: C, 63.42; H, 7.54; N,

18.49. Found: C, 63.25; H, 7.67; N, 18.37; LC/MS: purity 98%, m/z

455.40 [M⁺+1]. ¹H NMR (CDCl₃) d: 1.07 (t, J = 7.2 Hz, 6H, 2CH₃),

2.03 (quin, J = 5.8 Hz, 2H, CH₂CH₂CH₂), 2.64 (q, J = 6.9 Hz, 4H, 2CH₂-

CH₃), 2.85–2.91 (m, 4H, OCH₂CH₂+ CH₂CH₂), 3.18 (t, J = 5.5 Hz, 2H,

CH₂CH₂), 3.37 (s, 3H, N³CH₃), 3.53 (s, 3H, N¹CH₃), 3.70 (t, J = 7.3 Hz,

2H, OCH₂CH₂), 4.02 (t, J = 6.3 Hz, 2H, CH₂CH₂CH₂), 4.16 (t,

J = 6.0 Hz, 2H, CH₂CH₂CH₂), 6.84 (d, J = 8.5 Hz, 2H, Ar), 7.11 (d,

J = 8.5 Hz, 2H, Ar). ¹³C NMR (CDCl₃) d: 66.5, 102.8, 114.6, 129.7,

130.9, 149.2, 151.5, 151.9, 153.8, 157.5. IR

m

(cm^ꢀ1): 1697 (C@O),

added to each well at a ﬁnal concentration of 40 lM and the cells

1649 (C@O), 1246 (CAO); UV k_max(nm): 302.

were incubated for 15 min at 37 °C. Subsequently various dilutions

of the agonist 5⁰-N-ethylcarboxamidoadenosine (NECA, Sigma) in

the presence or absence of a single concentration of test compound

in HBSS containing 2.5% DMSO were added in duplicates. After

15 min of incubation at 37 °C the supernatant was removed and

1,3-Dibutyl-9-(4-(2-(diethylamino)ethoxy)phenylethyl)-6,7,8,9-

tetrahydropyrimido[1,2-f]purine-2,4(1H,3H)-dione (36)

The reaction was carried out for 11 h. The raw product was

crystallized from the mixture of acetone and water. Yield: 31%;

mp 72–74 °C. Anal. for C₃₀H₄₆N₆O₃: Calcd: C, 66.88; H, 8.63; N,

15.6. Found: C, 66.86; H, 8.57; N, 15.54; LC/MS: purity 99%, m/z

539.26 [M⁺+1]. ¹H NMR (CDCl₃) d: 0.88–1.03 (m, 6H,

2CH₃CH₂CH₂CH₂), 1.15 (br s, 6H, CH₃CH₂), 1.32–1.47 (m, 4H,

500 ll of 90 °C hot lysis buffer consisting of 4 mM EDTA and

0.01% Triton X-100 with the pH adjusted to 7.3 were added. After

1 h of mixing on ice, cAMP amounts in the cell lysates were deter-

mined by competitive radioligand binding experiments. cAMP

competition experiments were performed in a ﬁnal volume of

´

4361

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

120 ml containing 50

in lysis buffer (ﬁnal concentration 3 nM) and 40

protein diluted in the same buffer (50 g per sample). For deter-

mining cAMP concentrations 50 l of various cAMP concentrations

l

l of cell lysates, 30

l

l of [³H]cAMP solution

l of cAMP binding

4.5. Molecular modeling

l

All models were downloaded from the Adenosiland website³⁷

and, together with the 3EML X-ray structure prepared to docking

using the Protein Preparation Wizard option as a part of Schrödin-

ger package.³⁸In case of 3EML X-ray structure all non-protein

atoms and the intracellular T4-lysozyme insertion were removed.

The ligand binding cavity was conﬁned to a box with 20 Å size.

Docking studies for the reference adenosine receptor antagonists

to the rigid receptor binding pocket were performed using Glide

program,³⁸with the standard precision (SP) mode. The amino

hydrogen atom of the Asn^6.55side chain in each models was used

as a constrained H-bond donor for a ligand.

l

were measured instead of cell lysates to obtain a standard curve.

Total binding was determined by adding radioligand and binding

protein to the lysis buffer, and the background was determined

without addition of binding protein. The mixture was incubated

for 60 min on ice and ﬁltered through a GF/B glass ﬁber ﬁlter using

a cell harvester (Brandel). The ﬁlters were washed three times,

each with 2 ml of ice-cold 50 mM Tris–HCl buffer, pH 7.4, and sub-

sequently transferred into scintillation vials. The liquid scintilla-

tion counting of the ﬁlters started after 9 h of incubation in

2.5 ml of scintillation cocktail (Lumag AG, Basel). Three separate

experiments were performed. The amount of cAMP was deter-

mined by comparison to a standard curve generated for each

experiment.

4.5.1. Induced ﬁt docking protocol

All the downloaded models were improved by induced ﬁt dock-

ing procedure in Schrödinger Suite³⁸using reference adenosine

receptors antagonists (rolofylline and tonapofylline for hA₁/rA₁,

istradefylline for hA_2A/rA_2A). In the initial step the side chains

around the ligand were automatically trimmed (based on B-factor),

then the side chains of the residues around 5A of ligand poses were

reﬁned using Prime application.

The stereochemical quality and compatibility of all ﬁnal models

was assessed by PROCHECK,³⁹RAMPAGE⁴⁰and ANOLEA.⁴¹Each

‘improved’ homology model together with the high-scored docking

poses of the reference antagonists (rolofylline and tonapofylline for

hA₁/rA₁, istradefylline for hA_2A/rA_2A^10,24) was additionally analyzed

in Maestro program with respect to the model 3D structure, disul-

ﬁde bonds (Table S1, Supplementary data) and binding cavity.

4.3. Water solubility determination

Determination of the water solubility of selected compounds

(3a, 3b, 19, 23, 29 and 36) was performed using UV spectroscopy

on the basis of methods described earlier.³⁵Saturated solutions

of compounds (basic form) were prepared by suspending each

compound (10 mg) in H₂O (2 mL). The suspensions were mixed

and boiled for 5 min, then left overnight at 20 °C and ﬁltered off

using ﬁlter Macherey–Nagel MN 619 de. Each ﬁltrate was diluted

in MeOH (from 10 to 80-fold) and analyzed by UV spectroscopy

as a solution in MeOH/H₂O (90%v/v). Standard curves were deter-

mined using known concentrations of each compound. Each stock

solution was prepared (2 mg in 2 mL of the 90% MeOH) and further

diluted to obtain seven different concentrations ranging from 10^ꢀ3

to 10^ꢀ1mg/mL. The concentration of the saturated solutions for the

compounds was determined by linear regression of two vicinal

points from the standard curves and multiplication by the degree

of dilution using MS Excel.

4.5.2. Docking of tricyclic xanthine-based compounds

The 3D molecule models of tricyclic xanthines were built using

Schrödinger Suite molecular modeling environment³⁸based on the

reported crystallographic data for N9-phenyl- and N9-(2-benzy-

loxy)ethyl derivatives. Models of tetrahydropyrazino[2,1-f]purine-

diones were constructed using the 3D information obtained from

the solved X-ray structures of 38 and 42. The ligands library was

prepared with LigPrep module (low energy ionization/protonation

state in pH = 7 1, tautomeric state) optimized by conformational

search (MacroModel). The geometry optimization was performed

using the multiple minimization method as implemented in

MacroModel 9.7 with MMFFs force ﬁeld and Truncated Newton

Conjugate Gradient (TNCG) options and terminated when the root

means square (RMS) of conjugate gradient was below

0.05 kJ mol^ꢀ1Å^ꢀ1. The minimization was carried out in vacuum,

4.4. X-ray structure analysis

Inmoleculardockingcalculationstheknowledgeon the3D struc-

ture of tetrahydropyrimido- and tetrahydropyrazino[2,1-f]purine-

dione-based ligands adopted in the crystal state is useful. The

library of tetrahydropyrimido[2,1-f]purinediones 4–37 was built,

based on the reported crystallographic data for N9-phenyl- and

N9-(2-benzyloxy)ethyl- derivatives.^18,19Models of tetrahydropy-

razino[2,1-f]purinediones were constructed using the 3D informa-

tion obtained from the solved X-ray structures of 38 and 42 (Fig. 7).

with dielectric constant 1.0 as

interactions.

a way to treat electrostatic

Cmpd. 38

Cmpd. 42

Figure 7. Crystal structures of tetrahydropyrazino[2,1-f]purinedione-based compounds 38 and 42.

´

4362

E. Szymanska et al. / Bioorg. Med. Chem. 24 (2016) 4347–4362

21. Drabczy n ´ ska, A.; Yuzlenko, O.; Köse, M.; Paskaleva, M.; Schiedel, A. C.; Karolak-

Docking simulations of low-energy conformations for all com-

´

Wojciechowska, J.; Handzlik, J.; Karcz, T.; Kuder, K.; Müller, C. E.; Kiec-

pounds were performed with SP mode to all improved homology

models of adenosine receptors, with the constrained H-bond

between the side chain amino group of Asn^6.55and the ligand. Five

poses obtained after docking for each ligand (RMS deviation higher

than 0.5 Å) were post-minimized, and ﬁnal poses were kept and

analyzed according to the obtained docking score values.

The selected physicochemical properties, namely partition coef-

ﬁcient (QPlogP_o/w) and water solubility (QPlogS) were evaluated

using QikProp module.³⁸

Kononowicz, K. Eur. J. Med. Chem. 2011 , 46 , 3590.

22. Drabczy n ´ ska, A.; Karcz, T.; Szyma n ´ ska, E.; Kose, M.; Müller, C. E.; Paskaleva, M.;

Karolak-Wojciechowska, J.; Handzlik, J.; Yuzlenko, O.; Kie c ´-Kononowicz, K.

Purinergic Signal. 2013 , 9 , 395.

´

23. Szymanska, E.; Mazurkiewicz, J.; Kiec-Kononowicz, K. Heterocycl. Commun.

2013 , 19 , 297.

24. Müller, C. E.; Jacobson, K. A. In Methylxanthines ; Fredholm, B. B., Ed.; Springer:

Berlin Heidelberg, 2011; Vol. 200, pp 151–199.

25. Kim, S. A.; Marshall, M. A.; Melman, N.; Kim, H. S.; Müller, C. E.; Linden, J.;

Jacobson, K. A. J. Med. Chem. 2002 , 45 , 2131.

26. Van der Walt, M. M.; Terre’Blanche, G.; Petzer, A.; Lourens, A. C. U.; Petzer, J. P.

Bioorg. Chem. 2013 , 49 , 49.

27. Brunschweiger, A.; Koch, P.; Schlenk, M.; Pineda, F.; Kuppers, P.; Hinz, S.; Kose,

M.; Ullrich, S.; Hockemeyer, J.; Wiese, M.; Heer, J.; Müller, C. E. ChemMedChem

2014 , 9 , 1704.

For the graphic presentation of the selected structures with the

highest docking scores, representing individual clusters of poses,

PyMOL software was used.⁵⁷

28. Rybar, A.; Antos, K. Collect. Czech. Chem. Commun. 1970 , 35 , 1415.

29. Bergmann, F.; Dikstein, S. J. Am. Chem. Soc. 1955 , 77 , 691.

30. Perrin, D. D. Dissociation Constants of Organic Bases in Aqueous Solution ;

Buttersworth: London, 1972.

Acknowledgements

31. Alnouri, M. W.; Jepards, S.; Casari, A.; Müller, C. E. Purinergic Signal. 2014 , 10 ,

769.

Financial support of this research by the National Science Cen-

tre Poland (DEC/2012/04/M/NZ4/00219) is gratefully acknowl-

edged. C.E.M. is grateful for support by the Alzheimer Forschung

Initiative (AFI) e.V.

32. Schnell, D.; Strasser, A.; Seifert, R. Biochem. Pharmacol. 2010 , 80 , 1437.

33. Soriano-Ursua, M. A.; Ocampo-Lopez, J. O.; Ocampo-Mendoza, K.; Trujillo-

Ferrara, J. G.; Correa-Basurto, J. Comput. Biol. Med. 2011 , 41 , 537.

34. Auchampach, J. A.; Kreckler, L. M.; Wan, T. C.; Maas, J. E.; van der Hoeven, D.;

Gizewski, E.; Narayanan, J.; Maas, G. E. J. Pharmacol. Exp. Ther. 2009 , 329 , 2.

35. Vollmann, K.; Qurishi, R.; Hockemeyer, J.; Müller, C. E. Molecules 2008 , 13 , 348.

36. Jaakola, V. P.; Grifﬁth, M. T.; Hanson, M. A.; Cherezov, V.; Chien, E. Y. T.; Lane, J.

R.; IJzerman, A. P.; Stevens, R. C. Science 2008 , 322 , 1211.

37. Floris, M.; Sabbadin, D.; Medda, R.; Bulfone, A.; Moro, S. Eur. J. Med. Chem. 2012 ,

58 , 248.

Supplementary data

Supplementary data associated with this article can be found, in

the online version, at http://dx.doi.org/10.1016/j.bmc.2016.07.028.

38. Schrödinger, Maestro, version 9.3.5; LigPrep, version 2.5; QikProp, version 3.5;

Glide, version 5.8; Prime, version 3.1; Induced Fit Docking, Schrödinger Suite 2012

update 2 ; Schrödinger: New York, NY, 2012.

References and notes

39. Laskowski, R. A.; MacArthur, M. W.; Moss, D. S.; Thornton, J. M. J. Appl.

Crystallogr. 1993 , 26 , 283.

40. Lovell, S. C.; Davis, I. W.; Arendall, W. B.; de Bakker, P. I. W.; Word, J. M.; Prisant,

M. G.; Richardson, J. S.; Richardson, D. C. Proteins: Struct., Funct., Bioinf. 2003 , 50 ,

437.

1. McKim, W. A.; Hancock, S. In Drugs & Behavior: An Introduction to Behavioral

Pharmacology ; Pearson, 2012; pp 209–227.

2. Dillinger, T. L.; Barriga, P.; Escarcega, S.; Jimenez, M.; Lowe, D. S.; Grivetti, L. E. J.

Nutr. 2000 , 130 , 2057S.

3. Jacobson, K. A. In Adenosine Receptors in Health and Disease ; Wilson, C. N.,

Mustafa, S. J., Eds.; Springer: Berlin Heidelberg, 2009; Vol. 193, pp 1–24.

4. Francis, S. H.; Sekhar, K. R.; Ke, H.; Corbin, J. D. In Methylxanthines ; Fredholm, B.

B., Ed.; Springer: Berlin Heidelberg, 2011; Vol. 200, pp 93–134.

5. Howell, L. L.; Cofﬁn, V. L.; Spealman, R. D. Psychopharmacology 1997 , 129 , 1.

6. Fredholm, B. B.; IJzerman, A. P.; Jacobson, K. A. J. L.; Müller, C. E. Pharmacol. Rev.

2011 , 63 , 1.

7. Chen, J. F.; Eltzschig, H. K.; Fredholm, B. B. Nat. Rev. Drug Disc. 2013 , 12 , 265.

8. Jacobson, K. A.; Müller, C. E. Neuropharmacology 2016 , 104 , 31.

9. Sachdeva, S.; Gupta, M. Saudi Pharm. J. 2013 , 21 , 245.

10. Müller, C. E.; Jacobson, K. A. Biochim. Biophys. Acta 2011 , 1808 , 1290.

11. Biogen. Oral BG9928 in patients with heart failure and renal insufﬁciency,

ClinicalTrials.gov [Internet]. Bethesda (MD): National Library of Medicine (US).

2000- [cited 2015 Aug 31]. Available from: http://www.clinicaltrials.gov/ct2/

show/NCT00745316 Identiﬁer: NCT00745316.

12. Madden, D. R.; Armstrong, N.; Svergun, D.; Perez, J.; Vachette, P. J. Biol. Chem.

2005 , 280 , 23637.

13. Dungo, R.; Deeks, E. D. Drugs 2013 , 73 , 875.

14. Preti, D.; Baraldi, P. G.; Moorman, A. R.; Borea, P. A.; Varani, K. Med. Res. Rev.

2015 , 35 , 790.

15. Petzer, J. P.; Petzer, A. Curr. Med. Chem. 2015 , 22 , 975.

16. Tarazi, F. I.; Sahli, Z. T.; Wolny, M.; Mousa, S. A. Pharmacol. Ther. 2014 , 144 , 123.

17. Laurent, C.; Burnouf, S.; Ferry, B.; Batalha, V. L.; Coelho, J. E.; Baqi, Y.; Malik, E.;

Mariciniak, E.; Parrot, S.; Van der Jeugd, A.; Faivre, E.; Flaten, V.; Ledent, C.;

D’Hooge, R.; Sergeant, N.; Hamdane, M.; Humez, S.; Muller, C. E.; Lopes, L. V.;

Buee, L.; Blum, D. Mol. Psychiatry 2016 , 21 , 97.

41. Melo, F.; Feytmans, E. J. Mol. Biol. 1998 , 277 , 1141.

42. Ballesteros, J. A.; Weinstein, H. Methods Neurosci. 1995 , 25 , 366.

43. Isberg, V.; Mordalski, S.; Munk, C.; Rataj, K.; Harpsøe, K.; Hauser, A. S.; Vroling,

B.; Bojarski, A. J.; Vriend, G.; Gloriam, D. E. Nucleic Acids Res. 2016 , 44 , D356.

44. Isberg, V.; Vroling, B.; van der Kant, R.; Li, K.; Vriend, G.; Gloriam, D. Nucleic

Acids Res. 2014 , 42 , D422.

45. Piirainen, H.; Ashok, Y.; Nanekar, R. T.; Jaakola, V. P. Biochim. Biophys. Acta,

Biomembr. 2011 , 1808 , 1233.

46. Martinelli, A.; Ortore, G. Methods Enzymol. 2013 , 522 , 37.

47. Dore, A. S.; Robertson, N.; Errey, J. C.; Ng, I. W.; Hollenstein, K.; Tehan, B.;

Hurrell, E.; Bennett, K.; Congreve, M.; Magnani, F.; Tate, C. G.; Weir, M.;

Marshall, F. H. Structure 2011 , 19 , 1283.

48. Hino, T.; Arakawa, T.; Iwanari, H.; Yurugi-Kobayashi, T.; Ikeda-Suno, C.;

Nakada-Nakura, Y.; Kusano-Arai, O.; Weyand, S.; Shimamura, T.; Nomura, N.;

Cameron, A. D.; Kobayashi, T.; Hamakubo, T.; Iwata, S.; Murata, T. Nature 2012 ,

482 , 237.

49. Congreve, M.; Andrews, S. P.; Dore, A. S.; Hollenstein, K.; Hurrell, E.; Langmead,

C. J.; Mason, J. S.; Ng, I. W.; Tehan, B.; Zhukov, A.; Weir, M.; Marshall, F. H. J.

Med. Chem. 2012 , 55 , 1898.

50. UniProt Consortium Nucleic Acids Res. 2015 , 43 , D204.

51. Kim, S. K.; Gao, Z. G.; Van Rompaey, P.; Gross, A. S.; Chen, A.; Van Calenbergh,

S.; Jacobson, K. A. J. Med. Chem. 2003 , 46 , 4847.

52. Jaakola, V. P.; Lane, J. R.; Lin, J. Y.; Katritch, V.; Ijzerman, A. P.; Stevens, R. C. J.

Biol. Chem. 2010 , 285 , 13032.

53. Seibt, B. F.; Schiedel, A. C.; Thimm, D.; Hinz, S.; Sherbiny, F. F.; Muller, C. E.

Biochem. Pharmacol. 2013 , 85 , 1317.

54. Peeters, M. C.; van Westen, G. J.; Li, Q. Trends Pharmacol. Sci. 2011 , 32 , 35.

55. Peeters, M. C.; Wisse, L. E.; Dinaj, A.; Vroling, B.; Vriend, G.; Ijzerman, A. P.

Biochem. Pharmacol. 2012 , 84 , 76.

56. Peeters, M. C.; Li, Q.; van Westen, G. J.; Ijzerman, A. P. Purinergic Signal. 2012 , 8 ,

23.

57. Pymol, Open-Source PyMOL Molecular Graphics System v. 1.5.0.4. DeLano

Scientiﬁc.

18. Drabczy n ´ ska, A.; Müller, C. E.; Lacher, S. K.; Schumacher, B.; Karolak-

´

Wojciechowska, J.; Nasal, A. P. K.; Yuzlenko, O.; P e ˛k ala, E.; Kiec-Kononowicz,

K. Bioorg. Med. Chem. 2006 , 14 , 7258.

´

19. Drabczynska, A.; Müller, C. E.; Schiedel, A.; Schumacher, B.; Karolak-

Wojciechowska, J.; Fruzinski, A.; Zobnina, W.; Yuzlenko, O.; Kie c ´-

Kononowicz, K. Bioorg. Med. Chem. 2007 , 15 , 6956.

´

20. Drabczynska, A.; Müller, C. E.; Karolak-Wojciechowska, J.; Schumacher, B.;

Schiedel, A.; Yuzlenko, O.; Kie c ´-Kononowicz, K. Bioorg. Med. Chem. 2007 , 15 ,

5003.

Article Doi

Similarities and differences in affinity and binding modes of tricyclic pyrimido- and pyrazinoxanthines at human and rat adenosine receptors

DOI: 10.1016/j.bmc.2016.07.028

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.bmc.2016.07.028

Full text of DOI:10.1016/j.bmc.2016.07.028

Products guided by the article

R&D Labs maybe for 955121-59-2

Relevant to this article

Hot Product