GPR84, a Gi protein-coupled receptor that is activated by medium-chain (hydroxy)fatty acids, appears to play an important role in inflammation, immunity, and cancer. Recently, 6-octylaminouracil (4) has been reported to act as an agonist at GPR84. Here, we describe the synthesis of 69 derivatives and analogs of 4, 66 of which represent new compounds. They were evaluated in (a) cyclic adenosine monophosphate accumulation and (b) β-arrestin assays in human GPR84-expressing cells. Potent nonbiased as well as G protein-biased agonists were developed, e.g., 6-hexylamino-2,4(1H,3H)-pyrimidinedione (20, PSB-1584, EC50 5.0 nM (a), 3.2 nM (b), bias factor: 0) and 6-((p-chloro- and p-bromo-phenylethyl)amino)-2,4(1H,3H)-pyrimidinedione (47, PSB-16434, EC50 7.1 nM (a), 520 nM (b), bias factor: 1.9 = 79-fold Gi pathway-selective; 48, PSB-17365, EC50 2.5 nM (a), 100 nM (b), bias factor 1.3 = 20-fold selective), which were selective versus other free fatty acid-activated receptors. Compounds 20 and 48 were found to be metabolically stable upon incubation with human liver microsomes. A pharmacophore model was created on the basis of structurally diverse lipidlike GPR84 agonists.
GPR84, a Gi protein-coupled receptor that is activated by medium-chain(hydroxy)fatty acids, appears to play an important role in inflammation, immunity, and cancer. Recently, 6-octylaminouracil (4) has been reported to act as an agonist at GPR84. Here, we describe the synthesis of 69 derivatives and analogs of 4, 66 of which represent new compounds. They were evaluated in (a) cyclic adenosine monophosphate accumulation and (b) β-arrestin assays in humanGPR84-expressing cells. Potent nonbiased as well as G protein-biased agonists were developed, e.g., 6-hexylamino-2,4(1H,3H)-pyrimidinedione (20, PSB-1584, EC50 5.0 nM (a), 3.2 nM (b), bias factor: 0) and 6-((p-chloro- and p-bromo-phenylethyl)amino)-2,4(1H,3H)-pyrimidinedione (47, PSB-16434, EC50 7.1 nM (a), 520 nM (b), bias factor: 1.9 = 79-fold Gi pathway-selective; 48, PSB-17365, EC50 2.5 nM (a), 100 nM (b), bias factor 1.3 = 20-fold selective), which were selective versus other free fatty acid-activated receptors. Compounds 20 and 48 were found to be metabolically stable upon incubation with human liver microsomes. A pharmacophore model was created on the basis of structurally diverse lipidlikeGPR84 agonists.
G
protein-coupled receptors (GPCRs) are seven transmembrane receptors
that constitute one of the largest gene families in the human genome.[1,2] They are considered attractive targets for the development of drugs
for many human diseases, and it has been estimated that about 30%
of modern drugs target GPCRs, including many of the world’s
best-selling pharmaceuticals.[3,4] More than 100 GPCRs
represent orphan receptors, whose natural ligands are still unknown
or unconfirmed.[5] GPR84 is a so far poorly
investigated GPCR that is activated by micromolar concentrations of
medium-chain fatty acids with carbonchain lengths of 9–14,
which couples to the Gi/o pathway.[6−8] The receptor
is predominantly expressed in immune system-related tissues and cells,
such as bone marrow, spleen, lung, lymph nodes,[9] and brain microglia. Moreover, it is expressed in adipose
tissue.[10,11] GPR84 expression is upregulated in macrophages
upon lipopolysaccharide (LPS) stimulation, where it regulates the
production of interleukin-12 (IL-12), a proinflammatory cytokine that
controls the balance of T helper (Th1/Th2) responses and plays an
important role in inflammatory diseases such as rheumatoid arthritis
and inflammatory bowel disease.[12] The potential
role of GPR84 in this context has been supported by the finding that
knockout of GPR84 in T cells leads to augmented production of the
anti-inflammatory cytokine IL-4.[12] It has
been demonstrated that the zebrafishGPR84 (zGPR84) is involved in
the accumulation of lipid droplets, and undecanoic acid was shown
to amplify lipopolysaccharide-induced production of the proinflammatory
cytokine IL-12p40 through GPR84, indicating that GPR84 may play a
role in directly linking fatty acid metabolism to immune responses.[13] In animal models of endotoxemia and multiple
sclerosis (experimental autoimmune encephalomyelitis), it was found
that GPR84 is upregulated in microglia, suggesting that GPR84 may
also regulate neuroinflammatory processes.[10] A recent study reported that GPR84 may act as a sensor in amyloid
pathology, and its deficiency led to reduced microgliosis and dendritic
homeostasis in a mouse model of Alzheimer’s disease.[14] Thus, GPR84 is a proinflammatory receptor.[8] Potent and selective GPR84 agonists and antagonists
are required to further study its (patho)physiological roles and its
potential as a future drug target. GPR84 agonists might be useful
for the treatment of cancer by activating the immune response (immuno-oncology),
whereas desensitization of GPR84 by agonists might lead to a functional
blockade of the receptor, resulting in anti-inflammatory effects and
possibly antiproliferative effects on acute myeloid leukemia.[15]In 2006, GPR84 was shown to be activated
by medium-chain free fatty
acids with chain lengths of C9–14. In particular, decanoic
acid (C10, 1, Figure ), undecanoic acid (C11) and lauric acid (C12) were
found to be among the most potent agonists of GPR84.[7] Other studies by Suzuki et al. reported 2- or 3-hydroxy
fatty acids to be comparatively more potent than the nonhydroxylated
fatty acids,[8] and recently, Hoffmann et
al. have found that 3-hydroxytetradecanoic acid effectively activates
GPR84 (compounds 2 and 3, see Figure ).[16] Moreover, 6-octylaminouracil (4, Figure )[8] and the structurally
related (hexylthio)pyrimidine-4,6-diol (5, Figure ) and its derivatives were
identified as synthetic GPR84 agonists.[17,18] Compound 4, which displays a lipid-mimetic structure, was reported
to activate humanGPR84 with an EC50 value of 512 nM determined
in [35S]GTPγS binding assays.[8] Compound 5 exhibited GPR84 agonistic activity with
an EC50 of 139 nM determined in a calcium mobilization
assay in HEK293 cells recombinantly expressing the humanGPR84.[17,18]
Figure 1
Structures
of reported GPR84 agonists.
Structures
of reported GPR84 agonists.Embelin (6, Figure ), a natural product isolated from the plant Embelia ribes (Myrsinaceae), also
activates GPR84 (EC50 795 nM, cyclic adenosine monophosphate
(cAMP) assay, see Table ).[19] Embelin was reported to have anticancer,
antioxidant, anthelmintic, antifertility, antitumor, antiapoptotic,
antimicrobial, analgesic, and anti-inflammatory activities.[20−22] The safety and toxicity profile of embelin in rodents and nonrodents
was investigated and the compound was found to be safe up to 3 g/kg
(orally) when tested in rodents after acute exposure. Embelin was
reported to act as an X-linked inhibitor of apoptosis (IC50 4.1 μM) besides interaction with other targets,[23] but it is even more potent as an agonist of
GPR84. Diindolylmethane (DIM, 7),[8,24] another
natural product-derived compound that activates GPR84, also possesses
anticancer activity.[25] In a previous study,
we investigated the structure–activity relationships (SARs)
of DIM derivatives and analogues and developed a potent and selective
GPR84 agonist derived from DIM, compound 8 (PSB-16671),
with an EC50 value of 41.3 nM (Figure ).[26] DIM derivatives
were characterized as agonists with an ago-allosteric mechanism of
action as compared to fatty acids.[26,27] In the present
study, we used the potent and selective GPR84 agonist 4 as a lead structure. Our goal was to study the SARs of 6-aminouracil
derivatives and analogues as GPR84 agonists, to improve their potency,
optimize their properties, including selectivity and metabolic stability,
and develop Gi protein-biased agonists.
Table 1
Agonistic Activity of 6-Alkylaminouracils
and Standard Agonists at GPR84a
human
GPR84
cAMP assayb
β-arrestin assay
compound
R1
EC50 ± SEM (μM)
(or percent receptor activation at 10 μM)b [efficacy]c
EC50 ± SEM (μM)
(or percent receptor activation at 10 μM) [efficacy]d
EC50 values are shown
in bold. Efficacy is in italic.
Inhibition of forskolin (10 μM)-induced
decrease in cAMP accumulation.
Efficacy (Emax) relative to the max.
effect of decanoic acid (100 μM)
(=100%).
Efficacy (Emax) relative to the max. effect of embelin
(10 μM) (=100%).
Bias
factor was calculated as described
in Experimental Section.
n.d., not determined.
A literature value of 0.144 μM
was reported for this compound, determined in a homogeneous time-resolved
fluorescence assay (Cisbio) on HEK293 cells stably expressing human
GPR84.[18]
EC50 values are shown
in bold. Efficacy is in italic.Inhibition of forskolin (10 μM)-induced
decrease in cAMP accumulation.Efficacy (Emax) relative to the max.
effect of decanoic acid (100 μM)
(=100%).Efficacy (Emax) relative to the max. effect of embelin
(10 μM) (=100%).Bias
factor was calculated as described
in Experimental Section.n.d., not determined.A literature value of 0.144 μM
was reported for this compound, determined in a homogeneous time-resolved
fluorescence assay (Cisbio) on HEK293 cells stably expressing humanGPR84.[18]
Results and Discussion
Compound Design
To get insights
into the SARs of uracil
derivatives as GPR84 agonists, the following structural modifications
of the lead agonist 4 were made (Figure ): (A) optimization of the alkyl chain length;
(B) introduction of aromatic residues into the side chain, probing
the importance of the hydrogen bond donors: (C) N1–H, (D) N3–H,
and (E) N6–H; and (F) introduction of a substituent
at position 5 of the uracil moiety.
Figure 2
Structural modifications of lead compound 4.
Structural modifications of lead compound 4.
Chemistry
All
target compounds were synthesized by
coupling of the 6-chlorouracil derivative 12 or 13 with the appropriate amines. The commercially available
barbituric acid 9 was converted to 2,4,6-trichloropyrimidine
(10) by treatment with phosphorus oxychloride in the
presence of dimethylaniline. The key intermediate 6-chlorouracil (11) was obtained by heating 10 to reflux with
aqueous sodium hydroxide solution. 6-Chloro-1-methyluracil (12a) and 6-chloro-3-methyluracil (12b) were synthesized
by treatment of 11 with methyl iodide in the presence
of potassium carbonate, followed by silica gel column chromatography
using methanol (1%) in dichloromethane. 1,3-Dimethyl-6-chlorouracil
(12c) was synthesized by treating 11 with
an excess of methyl iodide. Finally, 6-chlorouracils 12a–c and 5-bromouracil 13, respectively,
were treated with the appropriate amine in 1-butanol under reflux
overnight to produce the desired products 4 and 14–71. Bromination was achieved by treatment of the
appropriate uracil derivative with N-bromosuccinimide
(NBS) to yield 72–75, whereas reacting 6, 20, 21, 42, or 43 with sodium nitrite under acidic conditions led to the
5-nitrosouracil derivatives 76–80. The structures of all synthesized final products were confirmed
by 1H and 13C NMR, or attached proton test (13Capt) NMR spectroscopy. The purity of the compounds
was determined by high-performance liquid chromatography (HPLC) coupled
to UV and electrospray ionization mass spectrometry (ESI-MS), confirming
a purity of at least 95% (Scheme ).
Scheme 1
Synthesis of 5-Alkylamino or 6-(Ar)Alkylamino-Substituted
Uracil
Derivatives (4 and 14–80)
Reagents and condition: (i) POCl3, dimethylaniline, 50 °C, 7 h and 125 °C, 20 min;
(ii) aqueous NaOH, reflux, 3 h; (iii) K2CO3,
CH3I, dimethyl sulfoxide (DMSO), room temperature (rt),
2 h or K2CO3, N,N′-dimethylformamide (DMF), 6 h for 12a and 12b or K2CO3, CH3I, DMF,
50 °C, 12 h for 12c; (iv) amines, 1-butanol, reflux,
12 h; (v) pyridine, N-bromosuccinimide (NBS), 80
°C, 2 h for 72–75 from 20, 6, 47, and 43;
or NaNO2, CH3COOH/H2O (1:1), 60–70
°C, 1 h and cooled to 4 °C for 76–80 from 6, 20, 21, 42, and 43, 67–79%; (vi) hexylamine, 1-butanol,
reflux, 12 h.
Synthesis of 5-Alkylamino or 6-(Ar)Alkylamino-Substituted
Uracil
Derivatives (4 and 14–80)
Reagents and condition: (i) POCl3, dimethylaniline, 50 °C, 7 h and 125 °C, 20 min;
(ii) aqueous NaOH, reflux, 3 h; (iii) K2CO3,
CH3I, dimethyl sulfoxide (DMSO), room temperature (rt),
2 h or K2CO3, N,N′-dimethylformamide (DMF), 6 h for 12a and 12b or K2CO3, CH3I, DMF,
50 °C, 12 h for 12c; (iv) amines, 1-butanol, reflux,
12 h; (v) pyridine, N-bromosuccinimide (NBS), 80
°C, 2 h for 72–75 from 20, 6, 47, and 43;
or NaNO2, CH3COOH/H2O (1:1), 60–70
°C, 1 h and cooled to 4 °C for 76–80 from 6, 20, 21, 42, and 43, 67–79%; (vi) hexylamine, 1-butanol,
reflux, 12 h.
Pharmacological Evaluation
All synthesized compounds were initially investigated in cAMP accumulation
assays at a concentration of 10 μM for their potency to inhibit
forskolin (10 μM)-induced cAMP accumulation. Chinese hamster
ovary (CHO) cells stably expressing the Gi protein-coupled
humanGPR84 were employed. Full concentration–response curves
were determined, and EC50 values were calculated for compounds
that showed more than 50% inhibition of cAMP accumulation in preliminary
tests (see Tables –3). Efficacy of the compounds was
determined by comparing their maximal effects to the maximal signal
induced by decanoic acid (100 μM; EC50 7.42 μM).
Standard GPR84 agonists were tested under the same conditions for
comparison (see Table ). Selected compounds that did not activate the receptor were screened
in antagonist assays versus decanoic acid (20 μM) at a concentration
of 10 μM. GPR84 specificity of the observed effects was confirmed
for the most potent compounds by testing them in the same assay, but
in nontransfected CHO cells lacking GPR84 expression. Selected compounds
were also investigated in β-arrestin 2 recruitment assays using
the β-galactosidase fragment complementation technology (PathHunter,
DiscoverX)[28,29] (see Tables –3). Potent
GPR84 agonists were additionally studied at human free fatty acid
receptors FFAR1 and FFAR4 to explore their selectivity (see Tables S1 and S2).[26]
Table 3
Various 6-Substituted and 5,6-Disubstituted
Uracil Derivatives
Inhibition of forskolin
(10 μM)-induced
decrease in cAMP accumulation.
Efficacy (Emax) relative to the max.
effect of decanoic acid (100 μM)
(=100%).
Efficacy (Emax) relative to the max. effect of embelin
(10 μM) (=100%).
Bias
factor was calculated as described
in Experimental Section.
n.d., not determined.
Structure–Activity Relationships
cAMP Accumulation Assays
Previously, 6-octylaminouracil
(4) was identified as a GPR84 agonist displaying an EC50 value of 512 nM determined in [35S]GTPγS
binding assays using Sf9 cell membranes expressing a humanGPR84-Gαi fusion protein.[8] This relatively
high EC50 value may be due to the highly artificial test
system that was employed. In our cAMP accumulation assay using CHO
cells transfected with the humanGPR84, 4 induced an
inhibition of forskolin-induced cAMP accumulation with an EC50 value of 17 nM (see Figure ). It was about 440-fold more potent than the standard agonist
decanoic acid (EC50 7400 nM, p = 0.0003).
We confirmed that the effect seen with 4 in the cAMP
assay was clearly due to GPR84 activation as it had no effect in nontransfected
CHO cells (see Figure ).
Figure 6
Concentration–response curves of selected compounds at human
GPR84 in cAMP assays (A, C) and in enzyme fragment complementation
β-arrestin recruitment assays (B, D). Mean values ± SEM
from three to four independent experiments performed in duplicates
are shown. For EC50 values, see Tables and 2.
Figure 4
Evaluation of selected compounds at wild-type
(wt) CHO cells in
cAMP accumulation assays. CHO wt cells were preincubated with the
respective test compounds at the indicated concentrations for 5 min.
Then, 10 μM forskolin was added and the cells were incubated
for additional 15 min. The maximal forskolin-induced cAMP accumulation
in the absence of test compound stimulation was defined as 100%. Mean
values ± standard error of the mean (SEM) from three independent
experiments performed in duplicates are shown.
To gain deeper insights into the SARs of uracil derivatives
as agonists of GPR84, we initially focused on modifying the hydrophobic
alkyl tail: a change in alkyl chain length ranging from C1 to C7 and
C9 to C10 (see Table ) demonstrated that the right alkyl chain
length was essential for high potency of the compounds at GPR84. A
short chain length of C2–3 as in 13 and 14, as well as a branched alkyl chain as in 15–18, yielded inactive uracil derivatives, whereas 19 with an alkyl chain length of five carbon atoms displayed
moderate agonistic activity with an EC50 of 460 nM. Increasing
the chain length by one more methylene unit to hexyl (20) led to a highly potent agonist displaying an EC50 value
of 5.0 nM, 92-fold more potent than 19 (p = 0.0391). Further extension of the alkyl chain length to C7 (21, EC50 12 nM), C9 (22, EC50 30 nM), or C10 (23, EC50 21 nM) led to slightly
reduced activities. Branching of the alkyl chain as in N6-(R,S)-(2-ethyl)hexyluracil (24) abolished activity, indicating limited space. N6-Methylthiopropyluracil (26, EC50 23 000 nM), an analogue of N6-pentyluracil (19, EC50 460 nM), in
which a methylene group was exchanged for a (lipophilic) sulfur atom,
was surprisingly 50-fold less potent than 19. The rank
order of potency for the length of the alkyl chain attached to the
uracil core was as follows: C6 (20, EC50 5.0
nM, vs 4, p = 0.0185) ≥ C7 (21, EC50 12 nM) ≥ C8 (4, EC50 17 nM) ≥ C10 (23, EC50 21
nM) ≥ C9 (22, EC50 30 nM) > C5 (19, EC50 460 nM, p = 0.0268).
Next, polar groups such as hydroxy or carboxy were introduced at the
end of the alkyl chain yielding compounds 27, 28 or 29, 30. Among them, only the hydroxyheptyl
derivative 28 (EC50 2000 nM) showed moderate
activity; its potency was significantly decreased in comparison to
the lead compound 4 (p = 0.0130); the
other polar derivatives were all inactive, again indicating that a
highly lipophilic pocket harbored the alkyl chain.Our next
effort was to investigate the importance of the NH functions,
N1–H, N3–H, and N6–H of the 6-aminouracil
derivatives. Methylation of N3 reduced the agonistic potency by more
than 40-fold (compare 31 (EC50 720 nM) with 4 (EC50 17 nM, 42-fold difference), 32 (EC50 2000 nM) with 22 (EC50 30
nM, 67-fold difference), and 33 (EC50 1900
nM) with 23 (EC50 21 nM, 90-fold difference)).
Methylation of N1 (34–35) or N1,N3-dimethylation
(36) virtually abolished potency of the compounds. Thus,
both NH atoms are important and may serve as hydrogen bond donors,
with N1–H being more important than the N3–H atom. Substitution
of the hydrogen atom at the 6-amino group (N6–H)
of the uracil core with a methyl group also led to a reduction in
potency of the hexyl-substituted derivative (compare 37 (EC50 110 nM) with 20 (EC50 5.0
nM), 22-fold reduction). However, surprisingly, N6-methylation
of the octyl-substituted lead structure 4 only led to
an insignificant (2-fold) decrease in potency (compare 38 (EC50 38 nM) with 4 (EC50 17
nM)). Taken together, these studies suggest that all NH functions
in the 6-aminouracil derivatives, N1–H, N3–H, and N6–H, should be ideally unsubstituted, but a free NH
function appears to be particularly important at the N1-position.Next, we moved the octylamino substituent of lead compound 4 from the 6- to the 5-position of the uracil core, resulting
in 39, which turned out to be completely inactive (EC50 > 10 μM). This confirms that the position of the
hydrophobic
tail is very important for its interaction with the receptor.Subsequently, we introduced a large variety of aromatic residues
attached to the N6-alkyl chain (see Table ). Benzyl (40) and (1-naphthyl)methyl
substitution (41) led to inactive compounds, whereas
longer alkyl or alkoxy linkers of two to four atoms between N6 of the aminouracil core and the aromatic ring resulted in
moderately to highly potent GPR84 agonists. A phenylethyl substituent
led to moderate activity (42, EC50 200 nM).
Further elongation of the carbonchain in the phenylpropyl derivative 43 reduced activity by 9-fold (EC50 1700 nM); however,
an additional methylene group (in the phenylbutyl derivative 44, EC50 24 nM) dramatically increased potency
to a value similar to that determined for the octyl-substituted lead
structure 4. The lipophilic chains of both compounds, 4 and 44, have about the same length. Encouraged
by these results, our next effort was to introduce substituents on
the phenyl ring of the relatively potent phenylethyl-substituted compound 42 with the goal to improve its potency. The following substituents
were introduced at the p-position of the phenyl ring: p-methyl (45, EC50 13 nM), p-fluoro (46, EC50 65
nM), p-chloro (47, EC50 7.1
nM), p-bromo (48, EC50 2.5
nM), p-ethyl (49, EC50 4.3
nM), p-methoxy (50, EC50 57
nM), and p-tert-butyl (51, EC50 44 nM). The obtained results showed that the substituents
led to an increase in potency, and hydrophobic residues were particularly
favorable. For example, the bulky p-bromo substituent
in 48 displayed a 23-fold increased potency compared
to the corresponding p-methoxy derivative 50, and in fact, 48 (EC50 2.5 nM) was the most
potent GPR84 agonist of the present series. The rank order of potency
for substituents at the p-position of the phenyl
ring in N6-phenethyluracil derivatives was as follows: p-bromo (48, EC50 2.5 nM) ≥ p-ethyl (49, EC50 4.3 nM) ≥ p-chloro (47, EC50 7.1 nM) > p-methyl (45, EC50 13 nM, p = 0.0166) > p-tert-butyl
(51, EC50 44 nM; p = 0.0008)
≥ p-methoxy (50, EC50 57 nM) ≥ p-fluoro (46, EC50 65 nM) ≥ p-unsubstituted (42, EC50 200 nM).
Table 2
Agonistic Activity of 6-Arylalkylamino-Substituted
Uracil Derivatives
Inhibition of forskolin
(10 μM)-induced
decrease in cAMP accumulation.
Efficacy (Emax) relative to the max.
effect of decanoic acid (100 μM)
(=100%).
Efficacy (Emax) relative to the max. effect of embelin
(10 μM) (=100%).
Bias
factor was calculated as described
in Experimental Section.
n.d., not determined.
Inhibition of forskolin
(10 μM)-induced
decrease in cAMP accumulation.Efficacy (Emax) relative to the max.
effect of decanoic acid (100 μM)
(=100%).Efficacy (Emax) relative to the max. effect of embelin
(10 μM) (=100%).Bias
factor was calculated as described
in Experimental Section.n.d., not determined.Subsequently, the effects of substituents at the o- or m-position of the phenyl moiety were
investigated.
The results showed that substituents in these positions generally
reduced potency compared to compounds with substituents at the p-position.
The rank order of potency for substituents at the m-position of the
phenyl ring in N6-phenethyluracil derivatives was as follows: m-bromo (55, EC50 64 nM) ≥ m-methyl (52, EC50 79 nM) ≥ o-fluoro (57, EC50 100 nM, p = 0.0299 compared to 55) = o-chloro (58, EC50 100 nM) ≥ m-chloro (54, EC50 130 nM) ≥
o,m-unsubstituted (42, EC50 200 nM) > m-fluoro (53, EC50 390 nM, p = 0.0007 vs 54) > o-methyl
(56, EC50 3200 nM, p <
0.0001 vs 53) ≫ o-bromo (59, EC50 > 10 000 nM). m-Methoxy substitution was not well tolerated as demonstrated by the
moderate potency of the m,p-dimethoxyphenyl
derivative 60 (EC50 810 nM, compared to that
of the p-methoxy derivative 50, EC50 57 nM, p = 0.0327). Dichloro substitution
in the m- and p-positions (61, EC50 44 nM)
combined the positive effect of the p-chloro substituent
and the negative effect of the m-chloro substituent.
Bioisosteric replacement of the phenyl ring in 42 (EC50 200 nM) with a 2-thienyl ring (63, EC50 190 nM) was well tolerated, whereas exchange for a 3-indolyl residue
significantly decreased potency (62, EC50 1400
nM, p = 0.0052). Introducing an ether linkage into
the side chain to increase polarity was well tolerated (compare the
phenoxypropyl derivative 65, EC50 73 nM, with
its phenylbutyl analog 44, EC50 24 nM) and
even led to 17-fold enhanced potency in case of the phenoxyethyl derivative 64 (EC50 100 nM) compared to the analogous phenylpropyl
derivative 43 (1700 nM).In a next small series
of compounds, we rigidified the N6-substitutent by integrating
the 6-amino group into cyclic structures,
piperidine or piperazine (see Table ). The 4-phenylpiperidinyl
derivative 66 was inactive, whereas the corresponding,
somewhat more flexible, 4-benzylpiperidinyl-substituted compound 67 (EC50 130 nM) was equipotent to the N6-phenylethyl-substituted analogue 42 and 13-fold more
potent than the N6-phenylpropyl-substituted compound 43. Four 6-piperazinyluracil derivatives were obtained, which
were substituted on the free piperazine N-atom (68–71). All four compounds showed submicromolar potency with
IC50 values between 38 and 150 nM, indicating that various
residues are tolerated ranging from substituted phenyl (68, 69) to indanyl (70) and even to (1-naphthyl)methyl
(71).Inhibition of forskolin
(10 μM)-induced
decrease in cAMP accumulation.Efficacy (Emax) relative to the max.
effect of decanoic acid (100 μM)
(=100%).Efficacy (Emax) relative to the max. effect of embelin
(10 μM) (=100%).Bias
factor was calculated as described
in Experimental Section.n.d., not determined.Finally, we investigated the effect of substituting
the uracil
5-position of 5-(ar)alkylaminouracil derivatives (72–80, Table ). We selected
potent agonists discussed above (4, 20, 21, 42, 43, and 48)
and introduced a bromo or nitroso substituent at position 5. The resulting
derivatives showed only moderate to weak potency compared to their
parent compounds (compare for 5-bromo substitution 72 vs 20 (1500-fold difference, p = 0.0094), 73 vs 4 (119-fold difference, p = 0.0287), 74 vs 48 (148-fold difference, p < 0.00001), and 75 vs 43 (>6-fold
difference) and for 5-nitroso substitution 76 vs 20 (1570-fold difference, p = 0.0001), 77 vs 21 (40-fold difference, p = 0.0006), 78 vs 4 (36-fold difference, p = 0.0022), and 79 vs 42 (>51-fold
difference, p < 0.0001)).Selected compounds
that were inactive as agonists were subsequently
tested for potential blockade of the receptor in cAMP assays. However,
none of them was found to be an antagonist of GPR84 (data not shown).The N6-substituted 1,3-dimethyl-6-aminouracil derivative
uradipil (see Table ), an α1-adrenoreceptor antagonist and serotonin 5-HT1A receptor agonist, which is therapeutically used as an antihypertensive
drug in Europe, neither activated nor inhibited GPR84 as determined
in cAMP accumulation assays.Figure summarizes
the structure–activity relationships of the investigated compounds
as agonists of the humanGPR84: (i) the length of the alkyl chain
plays an important role in determining the potency; an optimal chain
length of six carbon atoms was determined. Introducing an aromatic
residue into the alkyl chain was well tolerated and in some cases
led to increased potency. However, branching of the alkyl chain or
attachment of polar groups (−OH or −COOH) at its terminus
reduced or abolished the activity; (ii) free N3–H and particularly
N1–H functions in the uracil core structure were found to be
very important for activity; (iii) the substitution at the 6-position
of uracil was crucial, whereas 5-substitution abolished the activity.
Figure 3
Structure–activity
relationships of uracil derivatives at
human GPR84 as determined in cAMP assays.
Structure–activity
relationships of uracil derivatives at
humanGPR84 as determined in cAMP assays.
Efficacy in cAMP Assays
The maximal effect of the physiological
agonist decanoic acid (1) was set at 100% and compared
to the maximal effects observed for the investigated uracil derivatives
(see Tables –3). Lead structure 4 displayed a similar
efficacy (104%), and most of the potent uracil derivatives showed
the same or even higher efficacy. For example, the most potent agonists 20, 47, and 48 had efficacies of
127–137%. Our results showed that potent uracil-derived GPR84
agonists fully activate the Gi protein-coupled pathway.
Specificity of Effects Determined in cAMP Accumulation Assays
Selected potent GPR84 agonists, namely, 20, 47, 48, 49, as well as the lead
compound 4, were investigated for their ability to inhibit
forskolin-induced cAMP accumulation in CHO wild-type (wt) cells, which
do not express GPR84. None of the tested compounds showed any significant
effect at concentrations of up to 1 μM. At high concentrations
of up to 100 μM, 4 and 20 showed no
effect, whereas the other investigated compounds induced a minor inhibition,
which never exceeded 50% at a very high concentration of 100 μM
(Figure ). This indicates that the effects measured in cAMP
assays on GPR84-transfected cells (Tables –3) were clearly
due to GPR84 activation.Evaluation of selected compounds at wild-type
(wt) CHO cells in
cAMP accumulation assays. CHO wt cells were preincubated with the
respective test compounds at the indicated concentrations for 5 min.
Then, 10 μM forskolin was added and the cells were incubated
for additional 15 min. The maximal forskolin-induced cAMP accumulation
in the absence of test compound stimulation was defined as 100%. Mean
values ± standard error of the mean (SEM) from three independent
experiments performed in duplicates are shown.
β-Arrestin Recruitment Assays
Selected agonists
that had shown potency in the cAMP assays were further evaluated in
β-arrestin recruitment assays using the β-galactosidase
complementation assay technology (DiscoverX) (see Tables –3 and Figure B,D).
The standard agonist embelin (6, Figure ) was used in these assays as a reference
compound. Embelin is more suitable than decanoic acid (1) because it is more potent and has a somewhat higher efficacy in
the β-arrestin assays than the physiological standard agonist 1. Both lipidic agonists, embelin and decanoic acid, were
unbiased, showing nearly identical EC50 values in the two
different assays (decanoic acid: 7420 and 6080 nM, respectively; embelin:
795 and 424 nM, respectively; see Table ). In contrast, lead structure 4 and the structurally related (hexylthio)pyrimidine-4,6-diol (5) were nearly 7- to 10-fold more potent in the Gi-dependent cAMP assay compared to the β-arrestin recruitment
assay (4, 17 vs 110 nM; 5, 6.6 vs 78 nM).
As a direct comparison of EC50 values obtained in different
assays may be misleading due to different degrees of signal amplification,
we calculated the bias factors. The Hill slope of the curves was in
most cases close to 1, between 0.9 and 1.3. Because both assays were
conducted in the same cellular background, the reference compound
will control for any systematic bias.[33] For each of the two intracellular pathways, Δlog(Emax/EC50) was computed, followed by calculating
the pathway bias factor as ΔΔlog(Emax/EC50) according to a recently described and
validated method.[33] A bias factor of 0
means no bias, whereas a factor of 1 corresponds to a 10-fold preference,
and a factor of 2 to a 100-fold selectivity for the Gi-coupled
pathway. For many compounds, we observed an ca. 8- to 20-fold preference
for the cAMP pathway (bias factor of 0.9–1.3) compared to β-arrestin
recruitment. Nevertheless, the SARs of the uracil derivatives determined
in β-arrestin assays were quite similar to those observed in
cAMP assays (see Figure ). Among the tested compounds, three very potent unbiased agonists
could be identified, revealing almost identical EC50 values
in both assays (see Figures and 6A,B): 20, EC50 cAMP 5.0 nM, β-arrestin 3.2 nM,
bias factor: 0.0; 22, EC50 cAMP 30.0 nM, β-arrestin
34.0 nM (bias factor: 0.0); 51, EC50 cAMP
44.0 nM, β-arrestin 39.0 nM (bias factor −0.1). The unbiased
compound 20 with a C6 alkyl tail was the most
potent agonist of the present series in the β-arrestin assay.
The rank order of potency in the β-arrestin assay with regard
to the length of the alkyl chain at the uracil core was as follows:
C6 (20, 3.2 nM) > C9 (22, EC50 34 nM, p = 0.0113) ≥ C8 (4,
EC50 110 nM) > C10 (23, EC50 360
nM, p = 0.0033) ≈ C7 (21, EC50 390 nM) > C5 (19, EC50 4800
nM, p = 0.0292). A compound with a polar hydroxy
group at the end of the lipophilic tail exhibited identical potency
in both assay systems (28, EC50 1980 vs 2120
nM, bias factor: −0.2). In contrast, compounds 37 (EC50 110 nM (cAMP) vs 4200 nM (β-arrestin)) and 38 (EC50 38 nM (cAMP) vs 640 nM (β-arrestin))
displayed bias factors of 1.2 and 0.6, respectively. The most biased
agonists displaying a bias factor (ΔΔlog(Emax/EC50)) ≥ 1.3 and an EC50 < 10 nM were 47, 48, and 49 (Figure ), all of
which represent p-substituted 6-(phenylethylamino)uracil derivatives.
The most pathway-selective agonist of the whole series was 47 with an IC50 value in the cAMP assay of 7.1 nM and a
bias factor of 1.9 corresponding to a 79-fold preference for Gi coupling versus β-arrestin recruitment (see Figure B).
Figure 5
(A) Correlation between
the pEC50 values determined
in cAMP assays and pEC50 values determined in β-arrestin
assays. (Number of pairs: 34; p value (two-tailed)
= 0.0008). The most potent unbiased GPR84 agonists 20, 22, and 51 are marked in red; the most
potent Gi-biased agonists 47, 48, and 49 are marked in turquois. (B) Bias factors [ΔΔlog(Emax/EC50)] calculated for the most
potent biased and unbiased GPR84 agonists; positive values indicate
bias for the Gi protein-dependent over β-arrestin
recruitment pathway.
(A) Correlation between
the pEC50 values determined
in cAMP assays and pEC50 values determined in β-arrestin
assays. (Number of pairs: 34; p value (two-tailed)
= 0.0008). The most potent unbiased GPR84 agonists 20, 22, and 51 are marked in red; the most
potent Gi-biased agonists 47, 48, and 49 are marked in turquois. (B) Bias factors [ΔΔlog(Emax/EC50)] calculated for the most
potent biased and unbiased GPR84 agonists; positive values indicate
bias for the Gi protein-dependent over β-arrestin
recruitment pathway.Concentration–response curves of selected compounds at humanGPR84 in cAMP assays (A, C) and in enzyme fragment complementation
β-arrestin recruitment assays (B, D). Mean values ± SEM
from three to four independent experiments performed in duplicates
are shown. For EC50 values, see Tables and 2.Concentration–response curves for selected
agonists including
the lead structure 4 are depicted in Figure A–D.
Efficacy in
β-Arrestin Assays
The maximal effect
of embelin (6) in the β-arrestin assays was set
at 100% efficacy. Decanoic acid (1) was slightly less
efficacious (92%), whereas lead structure 4 showed a
higher efficacy of 189%. Most of the potent uracil derivatives displayed
high efficacies of around 150–250%, with a few exceptions.
Particularly high efficacy was observed for the N6-methylated
6-aminouracil derivatives 37 (278%) and 38 (309%) and 6-(p-bromophenethylamino)uracil (48, 283%), whereas the 6-alkylaminouracil derivatives 20, 23, 77, and 78 showed
efficacies of only around 100%. For comparison, the previously published
GPR84 agonist 5 displayed a high efficacy of 242%. The
availability of compounds with a range of efficacies in β-arrestin
assays will be useful to study its pharmacological significance.
Receptor Selectivity
Because the investigated uracil
derivatives can be envisaged as mimics of fatty acids such as dodecanoic
acid, their selectivities versus the G protein-coupled free fatty
acid receptors FFAR1 and FFAR4 were subsequently studied (see Tables S1 and S2 for the entire set of compounds).
In both, agonist and antagonist assays, none of the compounds activated
or inhibited FFAR1 or FFAR4. These results suggest that the developed
uracil derivatives can be considered as selective GPR84 agonists versus
FFAR1 and FFAR4.
Metabolic Stability
For subsequent
in vivo studies,
metabolically stable drugs will be required. Therefore, we studied
selected potent compounds, namely, 20, 22, 42, and 48, for their metabolic stability
in human liver microsomes. Although 6-nonylaminouracil (22) showed a very short half-life of 2 min (Clint 880 μL/min
mg protein), 6-phenethylaminouracil (42) and the two
most potent GPR84 agonists of the present series, 6-hexylaminouracil
(20) and 6-(p-bromophenethylamino)uracil
(48), proved to be highly stable (see Figure for 20, 22, and 48 and Figure S1 for 42). Even after an incubation period of 60 min,
degradation of 20, 42, and 48 was negligible. In addition, we investigated the metabolic stability
of 2-hexylthiopyrimidine-4,6-diol (5), a GPR84 agonist
recently reported by Liu et al.[18] (Figure ). Compound 5 was metabolized, and after 30 min of incubation in human
microsomes, almost 40% of the drug had disappeared. In comparison,
after the same incubation time, no degradation of 6-aminouracil derivatives 20, 42, and 48 could be observed.
This indicates that the potent GPR84 agonists 20 and 48 are metabolically stable and should be useful tools for
in vivo studies.
Figure 7
Metabolic stability of 6-aminouracil derivatives 20, 22, and 48, and 2-hexylthiopyrimidine-4,6-diol
(5) in human liver microsomes (0.5 mg/mL, mixed gender,
pooled). Compounds were tested at a concentration of 1 μM. Data
points represent mean values ± SD (for details, see Experimental Section).
Metabolic stability of 6-aminouracil derivatives 20, 22, and 48, and 2-hexylthiopyrimidine-4,6-diol
(5) in human liver microsomes (0.5 mg/mL, mixed gender,
pooled). Compounds were tested at a concentration of 1 μM. Data
points represent mean values ± SD (for details, see Experimental Section).
Pharmacophore Modeling
To understand and rationalize
important structural features of lipidlikeGPR84 agonists, we compared
the uracil derivatives 4 (lead structure) and 48 (potency in the low nanomolar range) to those of the known GPR84
agonists 3-hydroxydodecanoic acid (2) and embelin (6), which are >100-fold less potent than 48. Figure A shows
an overall
flexible alignment of the selected GPR84 agonists 2, 4, 6, and 48. All four structures
overlay very well with several features designated F1–F5 that
likely contribute to high GPR84 potency:
Figure 8
Overall alignment of
the selected GPR84 agonists 2 (green), 4 (cyan), 6 (blue), and 48 (orange). (A)
Left: five pharmacophore features were identified
(F1: H bond acceptor; F2, F3, and F5: H bond donor; and F4: hydrophobic).
Right: the distances between the pharmacophore features are shown
as red lines, and the distance (Å) between the features is reported
in the table. (B) Each individual GPR84 agonist is shown with the
five identified pharmacophore features (F1–F5); oxygen atoms
are colored red, nitrogen atoms blue, bromine atoms dark red, and
hydrogen atoms silver white; nonpolar hydrogen atoms are omitted.
(C) Pharmacophore features are indicated in two-dimensional (2D) representation
for 2, 4, 6, and 48 (cyan, hydrogen bond acceptor; magenta, hydrogen bond donor; green,
hydrophobic/aromatic domain). (D) Alignment of the previously published
agonist 5 with the created pharmacophore model in 2D
representation.
A hydrogen bond acceptor (F1) is found
in all structures: the carbonyl of the carboxylic acid in 2, the C1-carbonyl group in 6, and the C2-carbonyl group
in the uracil derivatives 4 and 48. This
pharmacophore feature is shown the cyan-colored sphere in Figure A,B.Three hydrogen bond donors (F2, F3,
and F5 shown as the magenta-colored spheres in Figure A,B) are found in the most potent agonists, 4 and 48, namely, N1–H (F2), N6–H (F3), and N3–H (F5), whereas agonist 2 has F3 (3-OH group), but not F2 and F5, and agonist 6 features only F2 (2-OH group).A long aliphaticchain or an arylalkyl
residue (F4 shown as the green-colored sphere in Figure A,B). This feature is found
in all agonists. The aromatic ring replacing the terminal part of
the aliphaticchain improves hydrophobic interactions and therefore
increases potency.Overall alignment of
the selected GPR84 agonists 2 (green), 4 (cyan), 6 (blue), and 48 (orange). (A)
Left: five pharmacophore features were identified
(F1: H bond acceptor; F2, F3, and F5: H bond donor; and F4: hydrophobic).
Right: the distances between the pharmacophore features are shown
as red lines, and the distance (Å) between the features is reported
in the table. (B) Each individual GPR84 agonist is shown with the
five identified pharmacophore features (F1–F5); oxygen atoms
are colored red, nitrogen atoms blue, bromine atoms dark red, and
hydrogen atoms silver white; nonpolar hydrogen atoms are omitted.
(C) Pharmacophore features are indicated in two-dimensional (2D) representation
for 2, 4, 6, and 48 (cyan, hydrogen bond acceptor; magenta, hydrogen bond donor; green,
hydrophobic/aromatic domain). (D) Alignment of the previously published
agonist 5 with the created pharmacophore model in 2D
representation.The steric and electronic
fit of the four structures is excellent
and in agreement with the observed SARs. The additional hydrogen bond
donors in the aminouracil derivatives 4 and 48 may be the reason for their improved GPR84 potency as compared to
the standard agonists 2 and 6. The uracil
ring, which lacks flexibility, keeps the hydrogen bond acceptor and
donor features in a fixed position for interaction with the amino
acid residues of GPR84.We subsequently investigated whether
the published GPR84 agonist
2-hexylthiopyrimidine-4,6-diol (5) matches the developed
pharmacophore model (see Figure D). In fact, it can be superimposed displaying most
of the identified interaction features.
Conclusions
In
conclusion, a series of 69 6-aralkylamino- and alkylamino-substituteduracil derivatives was synthesized, of which 66 (13–39, 41–65, 66–68, and 70–80) are new compounds that are not previously reported in the literature.
Starting from 6-octylaminouracil (4) as a lead structure,
our goal was to study the SARs of this class of GPR84 agonists, improve
their potency determined by Gi protein-dependent inhibition
of intracellular cAMP formation, and obtain metabolically stable compounds
suitable for in vivo studies. Many 6-(ar)alkylamino-substituted uracil
derivatives showed high GPR84 agonistic activity, whereas substitution
at the 5-position reduced or abolished activity. The length of the
carbonchain attached at the 6-position of the uracil core determined
the potency of compounds. Moreover, the introduction of an aromatic
residue into the alkyl chain further improved potency. 6-Hexylamino-2,4(1H,3H)-pyrimidinedione (20,
EC50 5.0 nM), 6-((p-chlorophenylethyl)amino)-2,4(1H,3H)-pyrimidinedione (47,
EC50 7.1 nM), and 6-((p-bromophenylethyl)amino)-2,4(1H,3H)-pyrimidinedione (48,
EC50 2.5 nM) were found to be the most potent GPR84 agonists
of the present series showing high efficacy. Potencies of the selected
compounds were further determined in β-arrestin assays, which
indicated that the phenethyl-substituted 6-aminouracil derivatives 47–49 are biased toward Gi-mediated adenylate
cyclase inhibition (20- to 79-fold), whereas the alkyl-substituted 20 displayed the same EC50 values in cAMP and β-arrestin
assays and was unbiased. Biased and unbiased agonists may display
different pharmacological profiles. For example, it had been suggested
that the μ-opioid receptor agonist morphine and related opioids
display their severe side effects, such as fatal respiratory depression,
by signaling through the β-arrestin pathway. Therefore, agonists
that are Gi-protein-biased and do not induce β-arrestin
recruitment have been developed.[34] Gi-protein-biased GPR84 agonists may be devoid of inducing receptor
desensitization and thus exhibit longer-lasting effects, or they might
trigger different signaling pathways compared to nonbiased agonists.The new compounds showed high selectivity for GPR84 versus the
related GPCRs FFAR1 and FFAR4 that display an overlapping ligand preference
regarding fatty acids. The new GPR84 agonists, which exhibit excellent
metabolic stability, will be useful tool compounds for elucidating
the physiologic roles and therapeutic potential of GPR84
Experimental
Section
General Methods
All commercially available reagents
were used as purchased (Acros, Alfa Aesar, Sigma-Aldrich, ABCR or
TCI). Solvents were used without additional purification or drying
except for dichloromethane, which was distilled over calcium hydride.
The reactions were monitored by thin-layer chromatography (TLC) using
aluminum sheets with silica gel 60 F254 (Merck). Column
chromatography was performed with 0.060–0.200 mm silica gel
with pore diameter of ca. 6 nm. All synthesized compounds were finally
dried in vacuum at 8–12 Pa (0.08–0.12 mbar) using a
sliding vane rotary vacuum pump (Vacuubrand GmbH). 1H and 13C NMR data were collected on a Bruker Avance 500 MHz NMR
spectrometer at 500 and 126 MHz, respectively. If indicated, NMR data
were collected on a Bruker Ascend 600 MHz NMR spectrometer at 600
MHz (1H) and 151 MHz (13C). DMSO-d6 was employed as a solvent at 303 K, unless otherwise
noted. Chemical shifts are reported in parts per million (ppm) relative
to the deuterated solvent, that is, DMSO, δ 1H: 2.49
ppm; 13C: 39.7 ppm. Coupling constants J are given in hertz, and spin multiplicities are given as s (singlet),
d (doublet), t (triplet), q (quartet), sext. (sextet), m (multiplet),
and br (broad). Melting points were determined on a Büchi 530
melting point apparatus and are uncorrected. The purities of isolated
products were determined by ESI-mass spectra obtained on an liquid
chromatography–mass spectrometry (LC–MS) instrument
(Applied Biosystems API 2000 LCMS/MS, HPLC Agilent 1100) using the
following procedure: the compounds were dissolved at a concentration
of 1.0 mg/mL in acetonitrile containing 2 mM ammonium acetate. Then,
10 μL of the sample was injected into an HPLC column (Macherey-Nagel
Nucleodur 3 μ C18, 50 × 2.00 mm2). Elution was
performed with a gradient of water/acetonitrile (containing 2 mM ammonium
acetate) from 90:10 to 0:100 for 20 min at a flow rate of 300 μL/min,
starting the gradient after 10 min. UV absorption was detected from
200 to 950 nm using a diode array detector. Purity of all compounds
was determined at 254 nm. The purity of the compounds was generally
≥95%.Compounds 13(8) and 69(31,32) have previously been described.
Compound 5 was synthesized according to a reported synthetic
procedure.[18] The synthesis and structural
characterization of the new compounds are described below. Data for
compounds 14–18, 24–27, 29, 30, 34–36, 39–41, 47, 59, 66, 72, 75, 76, and 79 are reported in Supporting Information.
General Synthetic Procedure
for the Preparation of 4 and 13–71
A suspension of 6-chlorouracil
(10 mmol) and the appropriate amine (50 mmol, 5 equiv) in 1-butanol
(20 mL) was refluxed at 125 °C for 12 h. The reaction was cooled
to rt, and half of the solvent was removed by evaporation under reduced
pressure. The precipitated solid was filtered off, washed with 1-butanol
(∼20 mL), followed by diethyl ether (∼20 mL), and dried
under vacuum.
General Synthetic Procedure for the Preparation
of 72–75
To a solution of the appropriate
6-alkylaminouracil derivative
(10 mmol) in dry pyridine (2 mL) was added N-bromosuccinimide
(NBS, 10 mmol) under an Ar atmosphere. The solution was heated to
80 °C for 2 h and cooled to rt. The precipitated solid was filtered
off, washed with diethyl ether (∼25 mL), and subjected into
column chromatography to afford the desire product.
General Synthetic
Procedure for the Preparation of 76–80
To a suspension of 5-alkylaminouracil (5 mmol) in water/acetic
acid (6.0:6.0 mL) heated at 60–70 °C was added a solution
of sodium nitrite (10 mmol, 2.0 equiv) in water (2 mL). The resulting
solution was stirred at the same temperature for 1 h and then cooled
to 4 °C. The resulting suspension was filtered off and washed
with cold water (2 × 50 mL) to afford the product.
The recombinant CHO cell line expressing
the humanGPR84 (CHO-hGPR84 cells) with a β-galactosidase fragment
and β-arrestin 2 containing the complementary fragment of the
enzyme for performing β-arrestin recruitment (Pathhunter) was
purchased from DiscoverX (Fremont, CA). The CHO-hGPR84 cells were
cultured in F12 medium supplemented with 10% FCS, 100 units/mL penicillin
G, 100 μg/mL streptomycin, 800 μg/mL G 418, 300 μg/mL
hygromycin B, and 1% ultraglutamin (Invitrogen, Carlsbad, CA, or Sigma-Aldrich,
St. Louis, MO). Stock solutions of compounds were prepared in DMSO.
The final DMSO concentration in the assays did not exceed 1%. Data
analysis was performed using GraphPad Prism (version 6.02). Concentration–response
data were fitted by nonlinear regression to estimate EC50 values (Prism 6.02). The unpaired t test was used
for statistical comparisons. Pathway bias was calculated as described
by Winpenny et al.[33] Δlog(Emax/EC50) was calculated by the following
equation: Δlog(Emax/EC50) = log(EmaxB/EC50B) – log(EmaxA/EC50A), where A is the reference agonist decanoic acid
and B is the test compound. ΔΔlog(Emax/EC50) was determined using the following equation:
ΔΔlog(Emax/EC50) = Δlog(Emax/EC50)Gi pathway – Δlog(Emax/EC50)ß-arrestin pathway.
GPR84 cAMP Accumulation Assays
cAMP assays were performed
as previously described.[26,35] In short, CHO-hGPR84
cells were stimulated by the addition of forskolin (10 μM) in
the absence (control) or presence of test compounds for 15 min. The
reaction was stopped with hot (90 °C) lysis solution (4 mM ethylenediaminetetraacetic
acid, 0.01% Triton X-100 in water). cAMP levels were quantified by
a radioactive assay using [3H]cAMP (PerkinElmer, Rodgau,
Germany) and a cAMP-binding protein prepared from bovine adrenal medulla.
The forskolin-induced increase in cAMP concentration in the presence
of agonists was expressed as a percentage of the response to forskolin
in the absence of agonists (% of control). Three independent experiments,
each in duplicate, were performed.
GPR84 β-Arrestin
Recruitment Assays
β-Arrestin
assays were performed as previously described.[26] Briefly, CHO-hGPR84 cells (20 000 per well) were
incubated with compound dilutions (in DMSO, final concentration: 1%)
for 90 min before adding the detection reagent (DiscoverX). After
60 min of incubation at rt, the luminescence was measured using an
NXT plate reader (PerkinElmer, Meriden, CT). Three to five independent
experiments were performed, each in duplicate.
FFAR4 β-Arrestin
Recruitment Assays
β-Arrestin
assays were performed as previously described.[26] In brief, CHO-hFFAR4 cells were incubated with test compound
or the reference agonist 4-[(4-fluoro-4′-methyl[1,1′-biphenyl]-2-yl)methoxy]benzenepropanoic
acid (TUG-891) for 90 min. In antagonist assays, the cells were preincubated
with the antagonist for 30 min before adding the reference agonist.
Next, the detection reagent was added and the mixture was incubated
for 60 min at rt. The luminescence was measured using an NXT plate
reader (PerkinElmer, Meriden, CT). All compounds were tested at a
final concentration of 10 μM. Test results were normalized to
values obtained by determining the background and the signal induced
by 30 and 4 μM TUG-891 in agonist and antagonist assays, respectively.
Three to four independent experiments were performed in duplicate.
FFAR1 Calcium Mobilization Assay
β-Arrestin assays
were performed as previously described.[26] In brief, 1321N1 astrocytoma cells recombinantly expressing humanFFAR1 were incubated with Hank’s balanced salt solution buffer
solution supplemented with 3 μM of the calcium dye Fluo-4-AM
(Life Technologies, Darmstadt, Germany) and 0.06% Pluronic F-127 for
60 min. After exchanging the buffer, test compound solutions were
added to each well using a FlexStation 3 plate reader (Molecular Devices,
Sunnyvale, CA). In antagonist assays, cells were preincubated for
30 min with test compound solutions before adding the reference agonist
3-(4-(o-tolylethynyl)phenyl)propanoic acid (TUG-424,
final concentration: 1 μM ≈ EC80). All compounds
were tested at a final concentration of 10 μM. Signals induced
by the test compounds were normalized to the signals induced by 1
and 10 μM TUG-424 in antagonist and agonist assays, respectively.
Three to four independent experiments were performed in duplicate.Pooled human liver microsomes (0.5
mg/mL, mixed gender, pooled) were employed. Compounds were tested
at a concentration of 1 μM. These experiments were performed
by Pharmacelsus (Contract Research Organisation), Saarbrücken,
Germany. Data points represent means of two separate experiments performed
in duplicate. Standard deviations were around ±20%.
Pharmacophore
Modeling
The selected GPR84 agonists 2, 4, 6, and 48 were
flexibly aligned using the flexible alignment module implemented in
Molecular Operating Environment (MOE 2014.09).[36] The flexible alignment method utilizes the stochastic search
procedure and simultaneously searches for the conformational space
of the defined molecules and the space of alignment of those molecules.[37] The method aligns the GPR84 agonists by maximizing
steric and feature overlap using the MMFF94x force field.[38] Each resulting alignment is given a score that
quantifies the quality of the alignment in terms of both internal
strain and overlap of molecular features. The similarity terms, including
hydrogen bond donor, acceptor, hydrophobicity, and volume, were used
in the flexible alignment with the default settings for the other
parameters. On the basis of the overall score, strain energy, and
visual inspection of the alignment, the presented flexible alignment
of the GPR84 agonists was selected. This alignment was then used as
a template for the pharmacophore model generation. The pharmacophore
model was generated using the pharmacophore elucidator feature in
MOE 2014.09. The pharmacophore features were calculated automatically
with the consensus pharmacophore function. This function clusters
the features into potential pharmacophore features, which are more
conserved than a tolerance and threshold value. For the presented
pharmacophore model, a tolerance distance of 1.5 Å and a threshold
of 50% conservation were used.
Authors: Sarah J Mancini; Zobaer Al Mahmud; Laura Jenkins; Daniele Bolognini; Robert Newman; Matt Barnes; Michelle E Edye; Stephen B McMahon; Andrew B Tobin; Graeme Milligan Journal: Sci Rep Date: 2019-02-12 Impact factor: 4.379
Authors: Amit Mahindra; Laura Jenkins; Sara Marsango; Mark Huggett; Margaret Huggett; Lindsay Robinson; Jonathan Gillespie; Muralikrishnan Rajamanickam; Angus Morrison; Stuart McElroy; Irina G Tikhonova; Graeme Milligan; Andrew G Jamieson Journal: J Med Chem Date: 2022-08-10 Impact factor: 8.039
Authors: Amadeus Samuel Schulze; Gunnar Kleinau; Rosanna Krakowsky; David Rochmann; Ranajit Das; Catherine L Worth; Petra Krumbholz; Patrick Scheerer; Claudia Stäubert Journal: iScience Date: 2022-09-06
Authors: Jamshid Amiri Moghaddam; Antonio Dávila-Céspedes; Stefan Kehraus; Max Crüsemann; Meryem Köse; Christa E Müller; Gabriele Maria König Journal: Mar Drugs Date: 2018-10-08 Impact factor: 5.118
Figure 1
Figure 2
Scheme 1
Figure 6
Figure 4
Figure 3
Figure 5
Figure 7
Figure 8