ACS Medicinal Chemistry Letters
Letter
ethylammonium chloride (BTEAC) as a phase-transfer
catalyst, resulting in 7-(2-bromoethyl)-8-bromotheophylline
(VI) or 7-(4-bromobutyl)-8-bromotheophylline (VII), respec-
tively. Compounds VI and VII were then cyclized with 3-(2-
aminoethyl)indole in dimethylformamide (DMF) or 2-
methoxyethanol, yielding diazepino- and imidazo-[2,1-f ]-
purine-2,4-diones 16 and 17. The structures of the synthesized
1
compounds were confirmed by MS, UV, IR, and H NMR
spectra. The purity of all compounds was confirmed to be
>95%.
Structure−Activity Relationships. To study initial SARs,
we modified the indolylethyl substituent on the xanthine core.
Any replacement of the indolyl residue with other aromatic or
aliphatic substituents (see 6−15 and 18) abolished the GPR18
activity. (See Table 1.) This indicates the importance of the
indole residue attached to the tricyclic xanthine scaffold for
GPR18 agonistic activity.
Previously, we had reported on several natural products
bearing indole moieties, which were able to inhibit THC-
induced GPR18 activation.38,39 In fact, these compounds had
been the first antagonists described for GPR18. In light of the
present findings, the observed requirement of an indole moiety
not only for agonists but also for various antagonists suggests
that the reported antagonists and the newly discovered
agonists may share the same binding site on GPR18.
Next, we studied the tricyclic xanthine core itself. Extension
of the annelated six-membered tetrahydropyrimidine ring
present in agonist 5 to a seven-membered ring appeared to
slightly increase the activity (16, PSB-KD477, EC50 0.454
0.156 μM). However, decreasing the ring size to a five-
membered imidazolidine ring significantly reduced the activity
by about 10-fold (17, EC50 = 5.68 1.54 μM).
Figure 4. Concentration-dependent inhibition of GPR18 activation
by an EC80 concentration of the GPR18 agonist THC (blue curve)
and the GPR18 agonist 5 (red curve). IC50 values were 0.650 μM (vs
THC, complete inhibition) and 0.944 μM (vs 5, 64% maximal
inhibition), determined in a β-arrestin enzyme complementation assay
using CHO cells stably expressing the human GPR18 receptor. Data
points are means of three independent experiments, each performed
in duplicate.
These results may be indicative of different binding sites or
receptor conformations to which the structurally very different
GPR18 agonists, the lipid-like THC, on the one hand, and the
heterotricyclic xanthine derivative 5, on the other hand, are
binding. Whereas the lipophilic antagonist 20 appears to bind
to the same binding site as the lipid-like agonist THC, this is
probably not the case for the xanthine-based agonist.
To confirm this hypothesis, we studied the concentration-
dependent activation of GPR18 by xanthine agonist 5 in the
absence and in the presence of different concentrations of the
lipidic antagonist 20. In fact, increasing concentrations of 20
led to a lowering of the maximal effect induced by agonist 5.
No significant rightward shift of the concentration−response
curve could be observed (Figure 5), and the EC50 values were
not significantly different from each other. (See Table S2.)
This is consonant with an allosteric mechanism of inhibition
Agonist 16 was also somewhat more potent at GPR55 (38%
activation at 10 μM) and at the CB2 receptor (46% inhibition
of radioligand binding at 10 μM) than lead structure 5. (See
Table 1 and Table S1.) Therefore, we continued modifying the
more selective derivative, the six-ring-annelated xanthine
derivative 5.
In a subsequent step, we investigated the substituents in the
N1- and N3-positions of the xanthine core. Replacement of the
methyl groups in agonist 5 (which are also present in the
naturally occurring xanthine alkaloids caffeine and theophyl-
line) by the longer propyl residues (compound 19) drastically
reduced the GPR18 activity. These steep SARs indicate limited
space in the binding pocket of GPR18.
None of the compounds evaluated in this study, including
those that showed no GPR18-agonistic activity, inhibited
THC-induced β-arrestin recruitment in GPR18-expressing
cells (Table 1). This shows that the investigated tricyclic
compounds that failed to activate GPR18 also did not block
the receptor.
Finally, we characterized the new agonist 5 by trying to
block its effect by the THC-competitive GPR18 antagonist
PSB-CB-27 (20; see Figure 4).24 In our previous study, we had
shown that 20 was able to completely block THC-induced
GPR18 activation in a β-arrestin recruitment assay, displaying
an IC50 value of 0.650 μM. Agonist 5 was employed at a
concentration of 1 μM, which corresponds to its EC80 value,
and inhibition of its effect by GPR18 antagonist 20 was studied
(Figure 4). Surprisingly, antagonist 20 was not able to fully
inhibit GPR18 activation induced by the tricyclic xanthine
agonist 5 (IC50 value of 0.944 μM, 65% maximal inhibition;
see Figure 4).
Figure 5. Activation of GPR18 by 5 in the absence and presence of
different concentrations of 20. EC50 values and maximum effects are
collected in Table S2. Determined by the β-arrestin enzyme
complementation assay using CHO stably expressing the human
GPR18 receptor. Data points shown are means of three independent
experiments performed in duplicate.
E
ACS Med. Chem. Lett. XXXX, XXX, XXX−XXX