Review

Orthologue selectivity and ligand bias:

translating the pharmacology of GPR35
Graeme Milligan
Molecular Pharmacology Group, Institute of Neuroscience and Psychology, College of Medical, Veterinary and Life Sciences,
University of Glasgow, Glasgow G12 8QQ, UK

GPR35 is a poorly characterized G protein-coupled receptor (GPCR) that has been suggested as a potential
therapeutic target for the treatment of diabetes, hypertension and asthma. Two endogenously produced
ligands have been suggested as activators of GPR35,
although the relevance of these remains unclear. Recently, a series of surrogate agonist ligands and the first
antagonists of GPR35 have been identified. However,
marked differences in the potency of agonists at species
orthologues of GPR35 have been noted, and this presents substantial challenges in translating the pharmacology at the cloned human receptor to ex vivo and in
vivo studies of the physiological function of this receptor
in animal models. Currently identified agonists will probably not display high selectivity for GPR35. By contrast,
comparisons of the potency of ligands at species orthologues of GPR35 have provided insight into the nature of
the ligand binding pocket and could result in the identification of more potent and selective ligands.
Introduction
GPR35 is a poorly characterized 7-transmembrane domain
G protein-coupled receptor (GPCR) first identified more
than 10 years ago. It was derived from an open reading
frame corresponding to 309 amino acids located in humans
on chromosome 2, region q37.3 [1]. In these initial studies,
expression was examined in a range of tissues but was
detected only in the intestine of the rat; it was also reported
to be lacking in a number of regions of human brain [1].
Subsequently, this same sequence (and a further sequence
encoding a second form of GPR35 that appears to be a
differentially spliced isoform containing an N-terminal
extension of 31 amino acids) (Figure 1) was identified from
a cDNA library produced from human gastric cancer cells
[2]. Again, expression was also detected in normal intestinal mucosal cells [2], and because these cDNAs were able to
transform NIH-3T3 cells, it was suggested that GPR35
might be oncogenic and play a role in the generation of
gastric cancers [2]. The significance of the N-terminally
extended form of GPR35 remains to be defined, but messenger RNA encoding this variant has been reported to be
present at higher levels than the shorter form [2].
In humans, the GPR35 gene displays significant polymorphic variability, with a number of non-synonymous
variants within the open reading frame resulting in alterations in amino acid sequence [3,4] (Figure 1). However, to
Corresponding author: Milligan, G. (Graeme.Milligan@glasgow.ac.uk).

date, apart from a very cursory examination of the Ser294Arg variant [5], which has been associated with the propensity to develop coronary artery calcification [6] (Table 1),
effects of these variations on signal transduction and pharmacology have yet to be reported. A further single nucleotide
polymorphism, located in the 50 ’ untranslated
region of the GPR35 gene, has been linked to early-onset
inflammatory bowel disease in a genome-wide association
study [7] but no further information on this is currently
available.
Potential endogenous agonists of GPR35
The first endogenously produced chemical that was shown
to be able to activate GPR35 was the tryptophan metabolite kynurenic acid [8]. When human GPR35 was expressed
along with a mixture of promiscuous and chimeric G
proteins [9,10] (Box 1) in CHO cells, addition of kynurenic
acid elevated [Ca2+]i in a concentration-dependent fashion
[8]. Importantly, other intermediates of tryptophan metabolism, including the non-carboxylate kynurenine, were
inactive [8]. This demonstrated the probable importance
of the acidic moiety of kynurenic acid for binding and/or
function. Furthermore, although each of the human, rat
and mouse orthologues of GPR35 was activated by kynurenic acid, it was already noted that kynurenic acid was less
potent at human GPR35 than at the rodent orthologues [8].
These observations were difficult to interpret fully, however, because the studies were performed after transient
transfection of CHO cells and without any indication of
the relative expression levels of the orthologues of GPR35
[8]. Further studies indicated that GPR35 was probably
able to couple to pertussis toxin-sensitive Gi-family G
proteins, because chimeric G protein a subunits containing
only the C-terminal five or nine amino acids from such G
proteins were able to transduce signals. Furthermore,
kynurenic acid-stimulated binding of [35S]GTPgS to membranes of CHO cells expressing GPR35 was prevented by
pretreatment with pertussis toxin [8], which blocks signal
transduction via this class of G proteins. A series of further
studies has confirmed the agonist action of kynurenic acid
at GPR35 [5,1113]. Moreover, the initial report of variation in potency of kynurenic acid at human versus rodent
orthologues of GPR35 has been confirmed and extended.
For example, Oka et al. [14] struggled to generate a response to kynurenic acid at human GPR35 in Ca2+ assays,
whereas Jenkins et al. [13] reported the EC50 of kynurenic
acid as >1103 M at human GPR35 but 7105 M at the
rat orthologue using a bioluminescence resonance energy

0165-6147/$ see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.tips.2011.02.002 Trends in Pharmacological Sciences, May 2011, Vol. 32, No. 5

317

()TD$FIG][ Review

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

MLSGSRAVPTPHRGSEELLKYMLHSPCVSLT
GPR35b: 31 aa insert at N-terminus
NH2

T253M
3.32

A25T
V29I

V76M
R

3.36

T108M

R125S

S294R
COOH
TRENDS in Pharmacological Sciences

Figure 1. Important structural features of human GPR35. Two isoforms of human GPR35 differ by the presence (sequence bar) or absence of a 31 amino extracellular
N-terminal sequence. Non-synonymous polymorphic variations in sequence within the open reading frame are shown as red circles with the alternative amino acids
defined by their one letter code. The Ser294Arg (i.e. S294R) variation has been associated with the propensity to develop coronary artery disease [6]. Arginine (R) (position
3.36) and tyrosine (Y) (position 3.32) residues in transmembrane domain III that play an important role in ligand recognition and/or function are highlighted in yellow.

transfer (BRET)-based GPR35-b-arrestin 2 interaction assay (Box 2) (Figure 2). Indeed, the very low potency of
kynurenic acid at human GPR35 prompted Jenkins and
colleagues [13] to question the potential relevance of this
ligand as a functional endogenous agonist, at least in
humans. Given the higher potency of kynurenic acid at
rodent orthologues of GPR35 and reported micromolar
concentration of kynurenic acid in rat small intestine
[15], effects of this ligand via GPR35 in rodents should
be anticipated.
A second group of endogenously produced ligands that
have been reported to activate GPR35 are lysophosphatidic
acids [14], particularly 2-acyl lysophosphatidic acids [14].
Responses to such ligands were difficult to assess in [Ca2+]i
elevation assays because vector transfected cells also produced a robust stimulation [14]. This probably reflects endogenous expression of one or more members of the
lysophospholipid receptor group of GPCRs [16,17]. However,
2-oleoyl lysophosphatidic acid caused internalization of an
epitope-tagged form of human GPR35, whereas kynurenic
acid had little effect [14]. Furthermore, in cells expressing
human GPR35, 2-oleoyl lysophosphatidic acid promoted
GTP loading on to the small GTP binding protein Rho A,
and this was maintained over a substantially longer
time period than in vector-transfected cells [14]. This is
particularly interesting given recent information on the
G protein-coupling profile of GPR35 (see below). Although,
on phylogenetic trees of GPCR sequences expressed in
humans and rodents, GPR35 does not reside in the same
318

region as the lysophospholipid receptors, it is most closely

related to GPR23. This receptor has been reported to respond to lysophosphatidic acid and has previously been
referred to as both the P2Y9 receptor and the lysophosphatidic acid LPA-4 receptor [18]. Furthermore, another relatively closely related receptor is GPR55. GPR55 was
originally discussed in terms of being a potential atypical
cannabinoid receptor [19], but it is certainly able to respond
to lysophosphatidylinositol [20]. Lysophosphatidic acids or
other endogenously produced lipids might represent true
endogenous ligands for GPR35 and for other related receptors such as GPR87 and GPR92 [21].
Surrogate ligands for GPR35
Although identification of endogenously produced chemicals with agonist action at GPR35 is of considerable importance, the ligands described above are far from ideal to
probe the roles of GPR35. Surrogate ligands are therefore
required. Until recently, the key GPR35 agonist has been
zaprinast (2-(2-propyloxyphenyl)-8-azapurin-6-one) (Table
1). Zaprinast was first identified as a GPR35 agonist by
Tanaguchi et al. [22]. Like kynurenic acid, zaprinast was
considerably more potent at rat than human GPR35, an
observation that has also subsequently been confirmed by
others [13,23] (Figure 2). Importantly, however, zaprinast
is substantially better known as an inhibitor of cGMP
phosphodiesterases (PDEs), particularly PDE5 and
PDE6, for which it displays low micromolar potency. If
zaprinast is used as the probe, it could be difficult in many

Review

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

Table 1. Chemical structures of a range of GPR35 ligands

Structure
Zaprinast

Action
Full agonist

Comments
Key surrogate ligand: potency at rat > human

References
[12,13,22]

[TD$INLE]

Kynurenic acid

Full agonist

Potential endogenous agonist: potency at rat > human

[8,13,34]

Partial agonist

Highest potency ligand at human. Low potency at rat

[12,23]

Full agonist

Clinically used anti-asthma medication

[23,33]

High efficacy agonist

Equipotent at human and rat

[23]

Partial agonist at rat

Limited activity at human

[23]

Antagonist

Action only described at human

[12]

[TD$INLE]

Pamoic acid/pamoate

[TD$INLE]

Cromolyn
O

OH

[TD$INLE]

HO

OH

O
O

Dicumarol
HO

OH

[TD$INLE]
O

Luteolin
OH
OH
HO
[TD$INLE]

OH

CID2745687
O
O
S

[TD$INLE]

N
H

N
H

settings to disentangle the contribution of the elevation of

cGMP from that resulting from the activation of GPR35.
As a consequence, two groups screened the Prestwick
Chemical Library1 of 1120 small molecule marketed
drugs and drug-like molecules for ligands able to act as
agonists at human [12] or both human and rat [23] GPR35.
Although both groups employed GPR35-b-arrestin 2 interaction assays, the bases of these were distinct (Box 2).
Zhao et al. [12] reported two hits from the primary screen:

the previously characterized ligand zaprinast and oxantel

pamoate. However, in follow-up studies pamoate
(4,40 -methylenebis(3-hydroxy-2-naphthoic acid)), rather
than the supposed active ingredient oxantel (1-methyl-2(3-hydroxyphenylethenyl)-1,4,5,6-tetrahydropyrimidine),
was identified as the GPR35 active ligand, displaying an
EC50 value of 80 nM [12]. By comparison, Jenkins et al. [23]
reported a wider range of hits at human GPR35 in
their primary screen. These included zaprinast but also
319

Review

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

Box 1. Promiscuous and chimeric G proteins

Promiscuous G protein a subunits (e.g. Ga16 in humans [9,10] and the
related rodent homologue Ga15, as well as a wide range of chimeric G
protein a subunits [9,10]) have been used widely in GPCR deorphanization studies and in ligand identification campaigns.
Although expressed endogenously in only limited sets of immune
cells, heterologous expression of Ga15 and/or Ga16 has been shown to
allow a wide range of GPCRs to elevate [Ca2+]i [10,39], an endpoint
favoured in many ligand screening campaigns, at least in part
because fluorescent kinetic plate readers and liquid-handling technology has allowed massive throughput [4042].
Although often described as promiscuous or universal G proteins
[9,10], Ga15 and Ga16 do not interact with all GPCRs, including GPR35
[6]. The desire to develop robust and generic assays that employ GPCRinduced elevation of intracellular [Ca2+] as the end point has, therefore,
resulted in the widespread use of chimeric G protein a subunits (in
which the extreme C-terminal region of non-Gq-family G proteins is
used to replace the equivalent region of Gaq, to produce Ga subunits
that couple different GPCR groups to this end point). This reflects the
fact that the extreme C-terminal region plays a crucial role in

determining which GPCRs interact productively with a G protein a

subunit, whereas the downstream signalling mechanisms are regulated by more internal sections of the G protein a subunit sequence.
Usually, such chimeras involve replacement of between five and nine
amino acids from the C-terminal tail. Mixtures of such chimeric
constructs are often transfected in combinations in de-orphanization
studies because, inherently, little or nothing is known in advance about
the G protein preference of the GPCR being studied. This was the
strategy employed in the first study to identify kynurenic acid as an
agonist at GPR35 [6]. This basic concept has been adapted to produce a
family of Ga16-Gax chimeras to attempt to extend the utility of Ga16 [42]
and a family of Gas-Gax chimeras [43] to switch signal output to the
elevation of cAMP levels. A further variation on the same basic concept
that has been used to screen for ligands at GPR35 reflects the limited
repertoire of G protein-coupled signals in the yeast Saccharomyces
cerevisiae [44]. Following replacement of the yeast GPCR Ste2 with
human GPR35 and of the yeast G protein a subunit Gpa1 with a Gpa1Ga13 chimera, GPR35 agonists promoted yeast cell growth and bgalactosidase activity via an appropriate gene reporter construct [23].

Box 2. b-Arrestin-based ligand identification

If occupied by agonist for a significant period of time, the vast majority
of GPCRs are able to interact effectively with a b-arrestin [45,46].
Although believed initially only to provide a means to terminate G
protein signalling by preventing access of the GPCR to G protein,
interactions between GPCRs and b-arrestins are now believed to also
trigger alternative signal pathways [45,46]. Regardless of the relevance
of this, because such interactions occur in an agonist-dependent
manner, they have been recognized to offer a means to identify ligands
that occupy GPCRs in assays that are independent of G proteincoupling preference [47]. In early studies, such assays invariably
employed visual detection of the cellular translocation of a b-arrestin
tagged with an autofluorescent protein, usually green fluorescent
protein. Although effective, such studies required the parallel development of high content imaging hardware and software [48,49] to provide
reasonable throughput and robust pharmacology. This approach was
used by Zhao et al. [12] to identify pamoate as a high potency agonist at
human GPR35. The need for high content screening equipment led to
the development of other assays that detect interactions between
GPCRs and b-arrestins. The most commonly employed are based
on either enzyme complementation [50] or BRET [51]. Both bgalactosidase complementation [13] and BRET [13,23] based GPR35b-arrestin 2 interaction assays have been used to either confirm the

cromolyn (5,50 -(2-hydroxypropane-1,3-diyl)bis(oxy)bis(4oxo-4H-chromene-2-carboxylic acid)) disodium, dicumarol

(3,30 -methylenebis(4-hydroxy-2H-chromen-2-one)), niflumic acid (2-{[3-(trifluoromethyl)phenyl]amino}nicotinic acid) and, importantly, both oxantel pamoate and pyrvinium
(4-[(3-carboxy-2-hydroxynaphthalen-1-yl)methyl]-3-hydroxynaphthalene-2-carboxylic acid; 1-methyl-2-[(E)-2-thiophen-2-ylethenyl]-5,6-dihydro-4H-pyrimidine)) pamoate
(Table 1).
The presence of the supposed inactive drug congener
pamoate in two separate primary screen hits alerted these
researchers to the possibility that this was the common
link in the responses and subsequent confirmation of pamoate as an agonist at human GPR35 with an EC50 value
of 50 nM [23]. The recognition that supposedly inactive
components of a mixture might have activity at a distinct
target has been discussed further in light of these results [24].
Importantly, previous studies by Jenkins et al. [13] that had
confirmed the species orthologue selectivity of zaprinast
encouraged the authors to repeat the primary screen using
320

activity of previously reported GPR35 agonists [13] or identify novel

GPR35 ligands from small scale chemical libraries [23]. The BRET-based
GPR35-b-arrestin 2 assay was reported to have a high signal to
background ratio and excellent screening statistics for both human
and rat orthologues of GPR35 [23] and to identify agonists with varying
efficacy [23]. A feature of b-arrestin-based assays of particular use in
screens at orphan or poorly characterized GPCRs is that, at least
theoretically, the potency of the ligands in such assays is anticipated to
provide a good measure of ligand affinity, because there should be a
direct correlation between GPCR occupancy and potency. This can
generate a structureactivity relation profile to underpin medical
chemistry in programmes targeting the development of agonist ligands
without the need for direct affinity measurements. These would
normally be provided via ligand binding assays but such studies might
be impractical if the ligand series is of low potency/affinity. For a
number of GPCRs a range of ligands has been shown to promote
receptor-b-arrestin interaction but has not been observed to activate G
protein-dependent signalling pathways. Such ligand bias or functional
selectivity [52,53] could have therapeutic implications but also implies
that ligands detected in such assays must be re-examined in more
conventional G protein-dependent assays to fully appreciate their
capacity for signal regulation.

rat GPR35 [23]. Although a number of hits at human GPR35

were also identified at rat GPR35, two novel hits, the
closely related flavenoids luteolin (2-(3,4-dihydroxyphenyl)-5,7-dihydroxy-4-chromenone) and quercetin (2-(3,4dihydroxyphenyl)- 3,5,7-trihydroxy-4H-chromen- 4-one),
were now also identified. Furthermore, neither oxantel
pamoate nor pyrvinium pamoate was identified when
screening against rat GPR35 [23]. This suggested that
pamoate might be significantly selective for human
GPR35, and when pamoate was assessed in parallel at
the two species orthologues, selectivity of at least 1000-fold
was observed. It was clear that pamoate was not an antagonist at the rat orthologue [23] because pamoate failed to
alter the potency of zaprinast to promote interactions between rat GPR35 and b-arrestin 2 [23]. This is one of the
most notable examples to date of ligand selectivity at mammalian GPCR species orthologues.
Interestingly, certain other hits at GPR35 (such as
niflumic acid) also displayed substantial selectivity for
the human orthologue [23]. This was not universal; a

()TD$FIG][ Review

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

BRET

GPR35

+ coelentrazine h

YFP

GPR35

YFP

NET BRET (mBRET)

250
arr2

Rluc

+ agonist

200

Key:
Human
Rat

150
100
50
0
-50

arr2

-11 -10

GPR35

-9

-8

-7

-6

-5

-4

-3

Log [zaprinast]M

Rluc
arr2
YFP
Rluc
< 80A

TRENDS in Pharmacological Sciences

Figure 2. Developing a BRET-based GPR35-b-arrestin 2 interaction assay. Representation of the GPR35-b-arrestin 2 interaction assay. (a) GPR35 tagged at the C-terminal tail
with enhanced yellow fluorescent protein (YFP) is cotransfected into cells along with a Renilla luciferase (RLuc) tagged form of b-arrestin 2. Following addition of a GPR35
agonist (black triangle) GPR35-YFP interacts with b-arrestin 2-RLuc. With addition of the luciferase substrate coelentrazine-h, light emitted upon substrate oxidation by the
luciferase is transferred to YFP and subsequently re-emitted at a longer wavelength if GPR35 and b-arrestin 2 have brought YFP and RLuc within a BRET-compliant distance
(<80 A). No such effect occurs if YFP and RLuc are not in proximity. (b) These effects can be quantified. Responses to varying concentration of the GPR35 agonist zaprinast
were measured after cotransfection of b-arrestin 2-RLuc with YFP tagged forms of either rat or human GPR35. Data are adapted with permission from [23].

number of compounds were essentially equipotent at these

two orthologues (including the anti-asthma medication
cromolyn disodium and the vitamin K antagonist dicumarol) [23]. However, like zaprinast, luteolin displayed significant selectivity for the rat receptor [23]. A further feature
of the various ligands was that they were partial agonists
when compared to zaprinast as the reference compound [23].
For example, increasing concentrations of pamoate reduced
human GPR35-b-arrestin 2 interactions produced by maximally effective concentrations of zaprinast [23], while the
partial agonist nature of luteolin and quercetin at rat
GPR35 was also demonstrated [23]. The potential selectivity
of these compounds for GPR35 over other related GPCRs
and potential therapeutic targets remain to be determined.
This is an important issue that needs to be explored before
roles for GPR35 are defined based on ex vivo or in vivo use of
such ligands. This is also the case for a series of thiazolidinedione ligands originally described as GPR35 agonists in a
patent from Arena Pharmaceuticals [25] and confirmed as
such by Jenkins and colleagues [13]. Currently the only
described antagonists of GPR35 are based on methyl5[(tert-butylcarbamothioylhydrazinylidene)methyl]-1-(2,4difluorophenyl)pyrazole-4-carboxylate; that is, CID2745687
[12] (Table 1), an nM inhibitor of the human orthologue. Use
of this ligand and identification of further antagonists from
distinct chemical series will probably be central in defining
the key functions of GPR35.
The mode of binding of ligands to GPR35
As noted above, although kynurenic acid is an agonist at
GPR35, this is true for neither kynurenine [8] nor

kynurenic acid ethyl ester [13]. This implicates a key

role for the carboxylate group in binding and/or activation of GPR35. Importantly, in studies of the L-lactate
receptor GPR81 [26], a number of receptors related to
GPR35 (and which have acidic ligands) were noted to
have a conserved arginine in transmembrane domain III.
This is at position 3.36 in the nomenclature of Ballesteros
and Weinstein [27] (in which the most conserved residue
in transmembrane domain X is designated X.50, whereas
the amino acid X.49 is one residue closer to the N
terminus and X.51 is one closer to the C terminus). This
residue was predicted to provide an ionic interaction with
the carboxylate [26]. Following alteration of this residue
to alanine, neither rat nor human GPR35 responded to
kynurenic acid [13] (Figure 3). Furthermore, the agonist
action of zaprinast at each orthologue was also eliminated by this mutation [13] (Figure 3). Although lacking a
formal negative charge, zaprinast does contain an acid
bioisostere (a group with similar physical or chemical
properties that provides functional characteristics broadly similar to a chemical compound). Furthermore, alteration of the tyrosine residue to alanine at position 3.32,
which is predicted to be on the same face of transmembrane domain III but one turn of the helix further towards the extracellular face of the receptor, also
eliminated responses to both kynurenic acid and zaprinast [13]. Although very preliminary, these studies have
begun to identify key residues of the binding pocket of
GPR35, and this will be investigated further by mutagenesis and analysis of the effects of a wider range of
ligands at such mutants.
321

()TD$FIG][ Review

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

OH

(a)

(b)

H
N

HN
N

300

hGPR35 WT

250

hGPR35 R(3.36)A

200

rGPR35 WT

150

rGPR35 R(3.36)A

100
50
0
-50

N
N

hGPR35 WT

400

Key:
NET BRET (mBRET)

NET BRET (mBRET)

350

Key:

CO2H

hGPR35 R(3.36)A

350
300

rGPR35 WT

250

rGPR35 R(3.36)A

200
150
100
50
0
-50

-8

-7

-6

-5

-4

-3

Log [kynurenic acid] M

-11

-10

-9

-8

-7

-6

-5

-4

-3

Log [zaprinast] M
TRENDS in Pharmacological Sciences

Figure 3. The role of arginine 3.36 in orthologues of GPR35. YFP tagged forms of wild-type (filled symbols) and Arg3.36Ala (open symbols) human (red) or rat (blue) GPR35
were cotransfected with b-arrestin 2-RLuc into HEK293 cells. BRET measurements as in Figure 1 were then performed in the presence of kynurenic acid (a) or zaprinast (b).
The chemical structure of these ligands is shown. Data are adapted with permission from [13].

G protein-coupling profile of GPR35

Although some of the earliest studies of GPR35 detected
ligand activation via transfection of a mixture of chimeric
and promiscuous G proteins [8,22], they noted selective
interaction of GPR35 with chimeric G proteins containing
the receptor recognition regions of Gao and Gai [8]. By
contrast the promiscuous G protein Ga16 (Box 1) did not
appear to couple to GPR35 [8]. Standard [35S]GTPgS binding studies are most suited to detect activation of Gi-family
G proteins [28,29]. Prevention of stimulation of
[35S]GTPgS binding by kynurenic acid in membranes
of CHO cells expressing human GPR35, by prior treatment
of the cells with pertussis toxin [8], was consistent with
this. Furthermore, following heterologous introduction of
the human isoforms of GPR35 into rat sympathetic neurons, the ability of both kynurenic acid and zaprinast to
inhibit N-type calcium channels was blocked by prior
treatment with pertussis toxin [5]. The ability of endogenously expressed GPR35 to inhibit forskolin-stimulated
cAMP levels in rat dorsal root ganglion was also ablated
by pertussis toxin pretreatment [11]. Despite these observations, Jenkins et al. [13] reported that, following expression in HEK293 cells, human GPR35 generated only very
modest increases in binding of [35S]GTPgS in response to
kynurenic acid; therefore, they explored possible interactions with other G proteins. Although they were unable to
record elevation of [Ca2+]i in cells cotransfected with Gaq
and either human or rat GPR35, the presence of a GaqGa13 chimera generated robust [Ca2+]i responses to zaprinast via both orthologues, whereas equivalent experiments
with a Gaq-Ga12 chimera did not [13]. Use of an antibody
able to identify only the GTP-bound, active state of Ga13
provided further support for interaction with this G protein
[13]. Subsequent development of an immunocapture assay
using an epitope-tagged form of Ga13 confirmed ligand
stimulation of [35S]GTPgS binding to this G protein [23]
(Figure 4). The ability of GPR35 expressed in HEK293 cells
to promote binding of GTP to Rho A [14] is also consistent
with a role for Ga13 because activation of Ga13 is generally
322

upstream of this effect [30]. By contrast, pertussis toxinmediated inhibition of interleukin 4 release from alphagalactosylceramide-activated human invariant natural
killer T cells via GPR35 [31], and of ERK activation in
U2OS cells expressing GPR35 [12], both support a role for
Gi-family G proteins (as does the capacity of both kynurenic acid and zaprinast to reduce forskolin-elevated cAMP
levels in cultured mouse glial cells [32]). It appears, therefore, that GPR35 can couple to both Ga13 and pertussis
toxin-sensitive Gi-family G proteins. It will be instructive
to determine whether there is ligand bias (Box 2) between
these pathways or predominance of one over another in
different cells and tissues, because such effects might
generate distinct signals from GPR35 in different cell
types. Although interactions between GPR35 and
b-arrestin-2 have been employed to develop assays to identify novel GPR35 ligands, and presumably occur in cells that
express GPR35 endogenously, their possible role in generating G protein-independent signals also remains to be
investigated.
Expression profile of GPR35
As noted above, initial studies indicated expression of
GPR35 in rat intestine [1] and stomach [2]. Subsequent
studies have confirmed significant expression levels in the
small intestine, colon and stomach, and this might be
relevant in the association between a GPR35 polymorphic
variant and early-onset inflammatory bowel disease [7].
GPR35 is also expressed in a range of other rat tissues
including lung, uterus, dorsal root ganglion and spinal cord
[22,32]. Wang et al. [8] were the first to record expression in
the spleen and white cells in both humans and mice,
whereas Yang et al. [33] have demonstrated that GPR35
is expressed in human mast cells, basophils and eosinophils, and that GPR35 mRNA is upregulated upon challenge with IgE antibodies. Furthermore, Barth et al. [34]
have suggested that GPR35 is highly expressed by human
peripheral monocytes, and messenger RNA encoding
GPR35 is upregulated substantially in primary human

()TD$FIG][ Review
(a)

(b)

DN

pc

E)

(E

3
G1

% of Background [35S]GTPS

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

Key:

180

pcDNA
160

G13(EE)

***

140
***
120
100
80
t

e
cl

hi
Ve

as

in
pr

Za

L D K L G E P DY I P S Q Q D I L L A R
LDKLGEPEYMPTEQDILLAR

le
ic

h
Ve

te

as

Za

in
pr

oa

m
Pa

G13
G13 (EE) (181-199)
TRENDS in Pharmacological Sciences

Figure 4. Activation of Ga13 by human GPR35. HEK293 cells were transfected to express human GPR35 with or without a form of Ga13 containing a modified sequence to
incorporate the so called Glu-Glu (EE) epitope tag. (a) Membrane preparations from these cells were resolved by SDSPAGE and immunoblotted to detect Ga13(EE). The
sequence of both wild -type Ga13 and the modified EE form of Ga13 is shown for amino acids 181199. (b) Membranes were then used in a [35S]GTPgS binding assay [29]
into which was added vehicle, zaprinast or pamoate. Ga13(EE) was subsequently immunoprecipitated with anti-EE and, after washing, binding of [35S]GTPgS assessed. Data
are adapted with permission from [23].

macrophages exposed to benzo(a)pyrene [35]. The functional significance of this has not yet been reported.
Physiological roles of GPR35 and potential therapeutic
opportunities
The expression of GPR35 in pancreatic b-cells, coupled
with the ability of thiazolidinedione ligands that have
agonist action at GPR35 to enhance glucose-stimulated
insulin release in model cell lines and to improve oral
glucose tolerance tests [25], has suggested a potential
use for agonists of GPR35 in the treatment of diabetes
and related metabolic disorders (Table 2). Expression of
GPR35 in the intestine and colon might also be relevant
because other GPCRs expressed on b-cells that regulate
insulin secretion, such as FFA1 (also designated GPR40),
are also expressed on enteroendocrine cells and regulate
the secretion of incretins such as GLP-1 [36], hence providing dual pathways of control. This is also true of the free
fatty acid receptor GPR120 [37], and could be a general
property of GPCRs that link nutrient sensing and energy
homeostasis. Although the specific distribution of GPR35
within the gut remains unclear, this is important to understand. Further studies using GPR35 agonists unrelated

to the thiazolidinediones will also be vital to better define

the contribution of GPR35 to the control of insulin
secretion.
GPR35 has also been implicated recently in the control of
blood pressure. GPR35 was identified in an effort to explore
genes associated with heart failure [38], but this was a
poorly powered study involving only 12 patients with a
variety of underlying conditions and marked differences
in disease severity. More interestingly and convincingly,
the blood pressure of mice lacking expression of GPR35 was
reported to be elevated by 37.5 mmHg compared with wildtype littermates [38]. GPR35 agonists might therefore be
anticipated to lower blood pressure. Substantial numbers of
patients have poorly controlled hypertension despite the use
of combinations of current front-line therapies, so new
therapeutic targets are needed for this population. Although
evidence of involvement of GPR35 is preliminary, this receptor is clearly worthy of study in this area.
The identification of both cromolyn disodium [23,33]
and nedocromil sodium [33] as GPR35 agonists (Table 1)
is of particular clinical interest. Both of these drugs are
approved anti-asthma medications and regulators of mast
cell sensitization and histamine release. However, they are

Table 2. Therapeutic potential for GPR35 ligands

Disease indication
Diabetes
Hypertension
Coronary artery disease
Asthma
Pain
Early-onset inflammatory
bowel disease

Supporting evidence
Thiazolidinediones with agonist action at GPR35 promote glucose-dependent
insulin secretion Such ligands also improve glucose handling
GPR35 knockout mice have markedly elevated blood pressure
Association with Ser294Arg polymorphism
Anti-asthma medications cromolyn disodium and nedocromil sodium are
agonists of GPR35
Expression of GPR35 in mouse dorsal root ganglion and spinal cord Effects
of agonist ligands in acetic acid-induced writhing models
Genetic linkage to a 50 untranslated single nucleotide polymorphism of GPR35

References
[25]
[38]
[6]
[23,33]
[12,32]
[7]

323

Review
considered orphan drugs because their mode of action has
been unclear. The fact that they can be shown to have
agonist action at GPR35 in cells transfected to express this
receptor does not mean that their mechanism of action in
vivo has been defined. However, as noted above, various
white blood cells, including mast cells, do express GPR35
and the growing availability of both agonist and antagonist
ligands will allow the contribution of GPR35 to the therapeutic actions of cromolyn disodium and nedocromil sodium to be more fully assessed.
Concluding remarks
Several orphan GPCRs have expression profiles that indicate they are worthy of consideration as therapeutic targets. This view can be supported via various transgenic
techniques, and it would be interesting to have wideranging phenotypic information on GPR35 knockout mice.
Based on the number of GPR35 active compounds identified recently in very small-scale screens [12,23,33], there is
reason to hope that more extensive screens, along with
follow-up medical chemistry programmes, will rapidly increase the pharmacological uses of this receptor. Such tools
will allow investigations of the function of GPR35. The
disease areas highlighted in this review (Table 2) are all
active topics for research with unmet clinical need. This is
likely to result in rapid progress in efforts to validate
GPR35 as a therapeutic target. Although only approximately 20 publications directly address the expression,
pharmacology and function of this receptor, this number
is likely to increase as pharmacological tools that modulate
the activity of GPR35 become widely available and the
potential of GPR35 as a therapeutic target becomes better
appreciated.
References
1 ODowd, B.F. et al. (1998) Discovery of three novel G-protein-coupled
receptor genes. Genomics 47, 310313
2 Okumura, S. et al. (2004) Cloning of a G-protein-coupled receptor that
shows an activity to transform NIH3T3 cells and is expressed in gastric
cancer cells. Cancer Sci. 95, 131135
3 Shrimpton, A.E. et al. (2004) Molecular delineation of deletions on
2q37.3 in three cases with an Albright hereditary osteodystrophy-like
phenotype. Clin. Genet. 66, 537544
4 Horikawa, Y. et al. (2000) Genetic variation in the gene encoding
calpain-10 is associated with type 2 diabetes mellitus. Nat. Genet.
26, 163175
5 Guo, J. et al. (2008) Inhibition of N-type calcium channels by activation
of GPR35, an orphan receptor, heterologously expressed in rat
sympathetic neurons. J. Pharmacol. Exp. Ther. 324, 342351
6 Sun, Y.V. et al. (2008) Application of machine learning algorithms to
predict coronary artery calcification with a sibship-based design. Genet.
Epidemiol. 32, 350360
7 Imielinski, M. et al. (2009) Common variants at five new loci associated
with early-onset inflammatory bowel disease. Nat. Genet. 41,
13351340
8 Wang, J. et al. (2006) Kynurenic acid as a ligand for orphan G proteincoupled receptor GPR35. J. Biol. Chem. 281, 2202122028
9 Milligan, G. and Kostenis, E. (2006) Heterotrimeric G-proteins: a short
history. Br. J. Pharmacol. 147 (Suppl. 1), S4655
10 Kostenis, E. et al. (2005) Techniques: promiscuous Ga proteins in basic
research and drug discovery. Trends Pharmacol. Sci. 26, 595602
11 Ohshiro, H. et al. (2008) GPR35 is a functional receptor in rat dorsal
root ganglion neurons. Biochem. Biophys. Res. Commun. 365, 344348
12 Zhao, P. et al. (2010) Targeting of the orphan receptor GPR35 by pamoic
acid: a potent activator of extracellular signal-regulated kinase and
b-arrestin2 with antinociceptive activity. Mol. Pharmacol. 78, 560568

324

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

13 Jenkins, L. et al. (2011) Agonist activation of the G protein-coupled
receptor GPR35 involves transmembrane domain III and is transduced
via Ga(13) and b-arrestin-2. Br. J. Pharmacol. 162, 733748
14 Oka, S. et al. (2010) GPR35 is a novel lysophosphatidic acid receptor.
Biochem. Biophys. Res. Commun. 395, 232237
15 Kuc, D. et al. (2008) Micromolar concentration of kynurenic acid in rat
small intestine. Amino Acids 35, 503505
16 Chun, J. et al. (2010) International Union of Basic and Clinical
Pharmacology. LXXVIII. Lysophospholipid receptor nomenclature.
Pharmacol. Rev. 62, 579587
17 Tigyi, G. (2010) Aiming drug discovery at lysophosphatidic acid
targets. Br. J. Pharmacol. 161, 241270
18 Noguchi, K. et al. (2003) Identification of p2y9/GPR23 as a novel G
protein-coupled receptor for lysophosphatidic acid, structurally distant
from the Edg family. J. Biol. Chem. 278, 2560025606
19 Sharir, H. and Abood, M.E. (2010) Pharmacological characterization
of GPR55, a putative cannabinoid receptor. Pharmacol. Ther. 126,
301313
20 Oka, S. et al. (2007) Identification of GPR55 as a lysophosphatidylinositol
receptor. Biochem. Biophys. Res. Commun. 362, 928934
21 Ishii, S. et al. (2009) Non-Edg family lysophosphatidic acid (LPA)
receptors. Prostaglandins Other Lipid Mediat. 89, 5765
22 Taniguchi, Y. et al. (2006) Zaprinast, a well-known cyclic guanosine
monophosphate-specific phosphodiesterase inhibitor, is an agonist for
GPR35. FEBS Lett. 580, 50035008
23 Jenkins, L. et al. (2010) Identification of novel, species selective
agonists of the G protein-coupled receptor GPR35 that promote
recruitment of b-arrestin-2 and activate Ga13. Biochem. J. 432,
451459
24 Neubig, R.R. (2010) Mind your salts: when the inactive constituent
isnt. Mol. Pharmacol. 78, 558559
25 Leonard, J. et al. Arena Pharmaceuticals Inc.GPR35 and modulators
thereof for the treatment of metabolic-related disorders, WO 2005/
119252 A2.
26 Liu, C. et al. (2009) Lactate inhibits lipolysis in fat cells through
activation of an orphan G-protein-coupled receptor, GPR81. J. Biol.
Chem. 284, 28112822
27 Ballesteros, J.A. and Weinstein, H. (1995) Integrated methods for the
construction of three-dimensional models and computational probing
of structure-function relations in G protein-coupled receptors. Methods
Neurosci. 25, 366428
28 Strange, P.G. (2010) Use of the GTPgS ([35S]GTPgS and Eu-GTPgS)
binding assay for analysis of ligand potency and efficacy at G proteincoupled receptors. Br. J. Pharmacol. 161, 12381249
29 Milligan, G. (2003) Extending the utility of [35S]GTPgS binding assays.
Trends Pharmacol. Sci. 24, 8790
30 Worzfeld, T. et al. (2008) G(12)/G(13)-mediated signalling in
mammalian physiology and disease. Trends Pharmacol. Sci. 29,
582589
31 Fallarini, S. et al. (2010) Expression of functional GPR35 in human
iNKT cells. Biochem. Biophys. Res. Commun. 398, 420425
32 Cosi, C. et al. (2010) G-protein coupled receptor 35 (GPR35) activation
and inflammatory pain: Studies on the antinociceptive effects of
kynurenic acid and zaprinast. Neuropharmacology DOI: 10.1016/
j.neuropharm.201011.014
33 Yang, Y. et al. (2010) G-protein-coupled receptor 35 is a target of the
asthma drugs cromolyn disodium and nedocromil sodium. Pharmacology
86, 15
34 Barth, M.C. et al. (2009) Kynurenic acid triggers firm arrest of
leukocytes to vascular endothelium under flow conditions. J. Biol.
Chem. 284, 1918919195
35 Sparfel, L. et al. (2010) Transcriptional signature of human
macrophages exposed to the environmental contaminant
benzo(a)pyrene. Toxicol. Sci. 114, 247259
36 Edfalk, S. et al. (2008) Gpr40 is expressed in enteroendocrine cells and
mediates free fatty acid stimulation of incretin secretion. Diabetes 57,
22802287
37 Ichimura, A. et al. (2009) Free fatty acid receptors act as nutrient
sensors to regulate energy homeostasis. Prostaglandins Other Lipid.
Mediat. 89, 8288
38 Min, K.D. et al. (2010) Identification of genes related to heart failure
using global gene expression profiling of human failing myocardium.
Biochem. Biophys. Res. Commun. 393, 5560

Review
39 Kostenis, E. (2006) G proteins in drug screening: from analysis of
receptor-G protein specificity to manipulation of GPCR-mediated
signalling pathways. Curr. Pharm. Des. 12, 17031715
40 Emkey, R. and Rankl, N.B. (2009) Screening G protein-coupled receptors:
measurement of intracellular calcium using the fluorometric imaging
plate reader. Methods Mol. Biol. 565, 145158
41 Liu, K. et al. (2010) A multiplex calcium assay for identification of
GPCR agonists and antagonists. Assay Drug Dev. Technol. 8, 367379
42 Ueda, T. et al. (2009) Development of generic calcium imaging assay for
monitoring Gi-coupled receptors and G-protein interaction. J. Biomol.
Screen. 14, 781788
43 Hsu, S.H. and Luo, C.W. (2007) Molecular dissection of G protein
preference using Gsalpha chimeras reveals novel ligand signaling of
GPCRs. Am. J. Physiol. Endocrinol. Metab. 293, E1021E1029
44 Dowell, S.J. and Brown, A.J. (2009) Yeast assays for G protein-coupled
receptors. Methods Mol. Biol. 552, 213229
45 Luttrell, L.M. and Gesty-Palmer, D. (2010) Beyond desensitization:
physiological relevance of arrestin-dependent signaling. Pharmacol.
Rev. 62, 305330
46 DeWire, S.M. et al. (2007) Beta-arrestins and cell signaling. Annu. Rev.
Physiol. 69, 483510

Trends in Pharmacological Sciences May 2011, Vol. 32, No. 5

47 Verkaar, F. et al. (2008) G protein-independent cell-based assays for
drug discovery on seven-transmembrane receptors. Biotechnol. Annu.
Rev. 14, 253274
48 Garippa, R.J. et al. (2006) High-throughput confocal microscopy for
beta-arrestin-green fluorescent protein translocation G proteincoupled receptor assays using the Evotec Opera. Methods Enzymol.
414, 99120
49 Haasen, D. et al. (2006) Comparison of G-protein coupled receptor
desensitization-related beta-arrestin redistribution using confocal and
non-confocal imaging. Comb. Chem. High Throughput Screen. 9, 3747
50 Eglen, R.M. (2007) Assessing GPCR activation using protein
complementation: a novel technique for HTS. Biochem. Soc. Trans.
35, 746748
51 Kocan, M. and Pfleger, K.D. (2009) Detection of GPCR/beta-arrestin
interactions in live cells using bioluminescence resonance energy
transfer technology. Methods Mol. Biol. 552, 305317
52 Violin, J.D. and Lefkowitz, R.J. (2007) Beta-arrestin-biased
ligands at seven-transmembrane receptors. Trends Pharmacol.
Sci. 28, 416422
53 Kenakin, T. (2010) Functional selectivity and biased receptor
signaling. J. Pharmacol. Exp. Ther. DOI: 10.1124/jpet.110.173948

325