Article https://doi.org/10.

1038/s41467-023-40482-9

Molecular insights into intrinsic transducer-

coupling bias in the CXCR4-CXCR7 system

Received: 11 April 2023 Parishmita Sarma1, Carlo Marion C. Carino2, Deeksha Seetharama1,
Shubhi Pandey1, Hemlata Dwivedi-Agnihotri1, Xue Rui3, Yubo Cao 4,
Accepted: 31 July 2023
Kouki Kawakami 2, Poonam Kumari5, Yu-Chih Chen6, Kathryn E. Luker 7
,
Prem N. Yadav5, Gary D. Luker7,8, Stéphane A. Laporte 4,9, Xin Chen3,
Asuka Inoue 2 & Arun K. Shukla 1
Check for updates

Chemokine receptors constitute an important subfamily of G protein-coupled

1234567890():,;
1234567890():,;

receptors (GPCRs), and they are critically involved in a broad range of immune
response mechanisms. Ligand promiscuity among these receptors makes
them an interesting target to explore multiple aspects of biased agonism.
Here, we comprehensively characterize two chemokine receptors namely,
CXCR4 and CXCR7, in terms of their transducer-coupling and downstream
signaling upon their stimulation by a common chemokine agonist, CXCL12,
and a small molecule agonist, VUF11207. We observe that CXCR7 lacks G-
protein-coupling while maintaining robust βarr recruitment with a major
contribution of GRK5/6. On the other hand, CXCR4 displays robust G-protein
activation as expected but exhibits significantly reduced βarr-coupling com-
pared to CXCR7. These two receptors induce distinct βarr conformations even
when activated by the same agonist, and CXCR7, unlike CXCR4, fails to activate
ERK1/2 MAP kinase. We also identify a key contribution of a single phos-
phorylation site in CXCR7 for βarr recruitment and endosomal localization.
Our study provides molecular insights into intrinsic-bias encoded in the
CXCR4-CXCR7 system with broad implications for drug discovery.

Chemokines are small secreted proteins that typically exert their promiscuity where not only a single chemokine can bind to, and acti-
actions via chemokine receptors belonging to the large superfamily of vate multiple chemokine receptors, but a given receptor can also be
G protein-coupled receptors (GPCRs), also known as seven trans- activated by several different chemokines4,5. Chemokine receptors
membrane receptors (7TMRs)1,2. Chemokines and chemokine recep- typically couple to, and signal through, heterotrimeric G-proteins and
tors contribute to a diverse array of physiological processes, especially β-arrestins (βarrs), as expected for prototypical GPCRs6. Interestingly
in various aspects of immune response activation and regulation2,3. A however, there are several examples of chemokine receptors that
peculiar aspect in the chemokine-chemokine receptor system is ligand exhibit a significant deviation from this paradigm, especially with

1
Department of Biological Sciences and Bioengineering, Indian Institute of Technology, Kanpur 208016, India. 2Graduate School of Pharmaceutical Sciences,
Tohoku University, Sendai, Miyagi 980-8578, Japan. 3Department of Medicinal Chemistry, School of Pharmaceutical Engineering and Life Science, Changzhou
University, Changzhou, Jiangsu 213164, China. 4Department of Pharmacology and Therapeutics, McGill University, Montréal, QC H3G 1Y6, Canada. 5Neu-
roscience and Ageing Biology Division, CSIR-Central Drug Research Institute Sector 10, Sitapur Road, Lucknow 226031 Uttar Pradesh, India. 6Department of
Computational and Systems Biology, Department of Bioengineering, University of Pittsburgh, Pittsburgh, PA, USA. 7Center for Molecular Imaging, Department
of Radiology, University of Michigan, Ann Arbor, MI, USA. 8Department of Biomedical Engineering, Department of Microbiology and Immunology, University of
Michigan, Ann Arbor, MI, USA. 9Department of Medicine, McGill University Health Center, McGill University, Montréal, QC H4A 3J1, Canada.
e-mail: [email protected]

Nature Communications | (2023)14:4808 1

Article https://doi.org/10.1038/s41467-023-40482-9

respect to their transducer-coupling patterns and downstream in various aspects of cancer onset and progression, cardiac disorders
signaling responses7–11. and autoimmune diseases13. Interestingly, CXCR7, but not CXCR4, also
The CXC chemokine receptor subtype 4 (CXCR4) and subtype 7 recognizes another chemokine referred to as CXCL1114. CXCR4 is
(CXCR7; also known as Atypical Chemokine Receptor 3, ACKR3) con- widely considered a prototypical GPCR with coupling to Gαi subfamily
stitute an interesting pair as they both recognize a common natural of G-proteins as measured in terms of inhibition of cAMP, and it also
chemokine agonist, referred to as CXCL12, also known as the stromal recruits βarrs upon agonist-stimulation15. On the other hand, stimula-
cell-derived factor 1 (SDF1)12 (Fig. 1a) These two receptors are involved tion of CXCR7 by CXCL12 or CXCL11 fails to elicit any measurable Gαi

a c

b d
Gi1 dissociation Gi2 dissociation Gi1 dissociation
1.1 1.1 1.1
1.0 1.0 1.0
0.9 0.9 0.9
0.8 0.8 0.8
0.7 Mock 0.7 Mock 0.7 Mock
CXCR4 CXCR4 CXCR4
0.6 CXCR7 0.6 CXCR7 0.6 CXCR7
0.5 0.5 0.5
-10 -9 -8 -7 -10 -9 -8 -7 -8 -7 -6 -5

Gi3 dissociation Go dissociation Gi3 dissociation Go dissociation

1.1 1.1 1.1 1.1
1.0 1.0 1.0 1.0
0.9 0.9 0.9 0.9
0.8 0.8 0.8 0.8
0.7 Mock 0.7 Mock 0.7 Mock 0.7 Mock
CXCR4 CXCR4 CXCR4 CXCR4
(Fold normalized)

0.6 CXCR7 0.6 CXCR7 0.6 CXCR7 0.6 CXCR7

0.5 0.5 0.5 0.5
-10 -9 -8 -7 -10 -9 -8 -7 -8 -7 -6 -5 -8 -7 -6 -5

Gs dissociation Gq dissociation Gs dissociation Gq dissociation

1.1 1.1 1.1 1.1
1.0 1.0 1.0 1.0
0.9 0.9 0.9 0.9
0.8 0.8 0.8 0.8
0.7 Mock 0.7 Mock 0.7 Mock 0.7 Mock
CXCR4 CXCR4 CXCR4 CXCR4
0.6 CXCR7 0.6 CXCR7 0.6 CXCR7 0.6 CXCR7
0.5 0.5 0.5 0.5
-10 -9 -8 -7 -10 -9 -8 -7 -8 -7 -6 -5 -8 -7 -6 -5

G12 dissociation G13 dissociation G12 dissociation G13 dissociation

1.1 1.1 1.1 1.1
1.0 1.0 1.0 1.0
0.9 0.9 0.9 0.9
0.8 0.8 0.8 0.8
0.7 Mock 0.7 Mock 0.7 Mock 0.7 Mock
CXCR4 CXCR4 CXCR4 CXCR4
0.6 CXCR7 0.6 CXCR7 0.6 CXCR7 0.6 CXCR7
0.5 0.5 0.5 0.5
-10 -9 -8 -7 -10 -9 -8 -7 -8 -7 -6 -5 -8 -7 -6 -5
log (CXCL12) [M] log (VUF11207) [M]
e f g
Gαi-coupling assay (cAMP inhibition) Gαs-coupling assay (cAMP accumulation) Gαq-coupling assay (Ca++ response)
CXCR4 (CXCL12) CXCR4 (VUF11207) V2R (AVP) CXCR4 5-HT2C (Serotonin) CXCR4 (VUF11207)
CXCR7 (VUF11207) CXCR7 (VUF11207) (VUF11207) CXCR7 (VUF11207)
120 120 120
(% normalized)
(% normalized)

(% normalized)

80 80 80

40
40 40
0
0 0
-12 -10 -8 -6 -4 -12 -10 -8 -6 -4 40 80 120
log (agonist) [M] log (agonist) [M] Time (s)

Nature Communications | (2023)14:4808 2

Article https://doi.org/10.1038/s41467-023-40482-9

Fig. 1 | Lack of G-protein activation upon stimulation of CXCR7. a CXCL12, a CXC taken as 1). e Agonist-induced decrease in forskolin-induced cAMP level measured
type chemokine, is a common agonist for both, CXCR4 and CXCR7 (created with using the GloSensor assay for the indicated receptor-ligand combinations as a
BioRender.com). b NanoBiT-based assay for CXCL12-induced dissociation of het- readout of Gαi-activation (mean ± SEM; n = 4; normalized with the signal at minimal
erotrimeric G-proteins for CXCR4 and CXCR7 (mean ± SEM; n = 4-5 independent ligand dose for CXCL12-CXCR4 combination as 100%). f Agonist-induced increase
experiments; i.e., for Gi1 dissociation: Mock, n = 4; CXCR4, n = 5; CXCR7, n = 4; for in cAMP level measured using the GloSensor assay for the indicated receptor-ligand
Gi2 and Gi3 dissociation: n = 5; for Go dissociation: Mock, n = 4; CXCR4, n = 5; combinations as a readout of Gαs-activation (mean ± SEM; n = 3; normalized with
CXCR7, n = 5; for Gs, Gq, G12, and G13 dissociation: n = 4; normalized with lumi- maximal signal for V2R as 100%). V2R (vasopressin receptor subtype 2) is used as a
nescence signal under unstimulated condition taken as 1). Mock represents empty positive control. g Agonist-induced increase in Ca++ level measured using the
vector transfected cells as a negative control. c VUF11207 is a small molecule GCaMP sensor for the indicated receptor-ligand combinations as a readout of Gαq-
agonist for CXCR7 but its efficacy for CXCR4, if any, is not known (created with activation (mean ± SEM; n = 4; normalized with maximal signal for serotonin as
BioRender.com). d NanoBiT-based assay for VUF11207-induced dissociation of 100%). 5-HT2C receptor is used as a positive control. Source data are provided as a
heterotrimeric G-proteins for CXCR4 and CXCR7 (mean ± SEM; n = 3 independent source data file.
experiments; normalized with luminescence signal under unstimulated condition

activation, although there are reports that suggest its ability to recruit of its binding and/or agonism at CXCR4, which is further corroborated
βarrs16,17. Interestingly, a small molecule ligand known as VUF11207 has by the absence of any measurable response even upon CXCR4 over-
also been reported to promote CXCR7-βarr interaction18, although its expression (Fig. 1e–g). We also measured surface expression of CXCR4
interaction with other CXCRs and complete transducer-coupling pro- and CXCR7 in these assays, which was always higher than mock-
file has not been evaluated thus far. Therefore, the CXCR4-CXCR7 pair transfected cells although in some experiments, CXCR7 expression
represents an intriguing system to probe the molecular and structural was only marginally above the mock-transfection (Supplementary
details of intrinsic transducer-coupling bias. Fig. 1b, c). Therefore, we repeated these experiments with higher
Here, we present a comprehensive investigation of agonist- expression levels of CXCR7 by increasing the amount of transfected
induced G-protein-coupling, contribution of GRKs in βarr recruit- DNA but still did not observe any response for CXCR7 (Supplementary
ment and conformational signatures, and ERK1/2 MAP kinase activa- Fig. 1d, e). Taken together, these data establish the inherent inability of
tion downstream of CXCR4 and CXCR7 using the shared natural G-protein activation by CXCR7 upon stimulation by CXCL12 or
agonist CXCL12 and a small molecule compound, VUF11207. Our study VUF11207.
provides the molecular details of the intrinsic bias encoded in the
CXCR4-CXCR7 system and their functional divergence. These findings Agonist-induced βarr recruitment
not only offer important insights to better understand biased agonism Although previous studies have shown βarr coupling to CXCR415 and
at 7TMRs but also present an experimental framework that may guide CXCR716, a comprehensive side-by-side analysis of recruitment of both
analogous exploration of other chemokine receptors. βarr isoforms i.e. βarr1 and 2, to both receptors has not been described
thus far. Therefore, we used two different assays to measure recruit-
Results ment of βarr1 and 2 to both these receptors in response to CXCL12 and
Agonist-induced G-protein-activation and second messenger VUF11207. First, we measured CXCL12-induced βarr2 recruitment in
response PRESTO-Tango assay21, where the receptors are engineered to contain
CXCR7, unlike CXCR4, is considered to lack G-protein-coupling in the carboxyl-terminus of vasopressin subtype 2 receptor (i.e. CXCR4-
response to CXCL12 stimulation, although the experimental evidence V2 and CXCR7-V2). While we observed a robust response for CXCR7 in a
is limited primarily to the lack of canonical Gαi-activation17. Therefore, ligand dose-dependent manner, the basal luminescence signal in
we first set out to comprehensively probe CXCL12-induced G-protein CXCR4-expressing cells was high and it increased only slightly in
activation using a NanoBiT-based heterotrimer dissociation assay19 for response to CXCL12 (Fig. 2a and Supplementary Fig. 2a). Considering
CXCR4 and CXCR7. In this assay, agonist-stimulated G-protein activa- that PRESTO-Tango constructs use a chimeric receptor, we also gen-
tion is measured as a decrease in luminescence signal arising from erated new Tango assay constructs for CXCR4 and CXCR7 without the
dissociation of the NanoBiT-engineered heterotrimer consisting of the V2R carboxyl-terminus fusion, and measured βarr2 recruitment in
Gα-LgBiT, SmBiT-Gγ2 and untagged Gβ1 subunits. We observed that response to CXCL12. We observed a robust agonist-induced response
CXCR4 robustly activates G-proteins of the Gαi subfamily but CXCR7 for CXCR7, however, CXCR4 displayed a significantly lower Emax for
remains silent in this assay not only for Gαi but other subtypes as well βarr2 recruitment compared to CXCR7 (Fig. 2b and Supplementary
(Fig. 1b). As mentioned earlier, a small molecule ligand has also been Fig. 2b). On the other hand, VUF11207 effectively promoted βarr2
described for CXCR7 although its characterization remains limited to recruitment for CXCR7 but not for CXCR4 further suggesting its spe-
binding studies and βarr recruitment in a BRET assay (Fig. 1c). We cificity at CXCR7 (Fig. 2c, d). We also noted that CXCR7 expression in
therefore decided to also test VUF11207 in the NanoBiT-based het- PRESTO-Tango assay was only marginally higher than mock-
erotrimer dissociation assay to probe its ability to activate G-proteins, transfection (Supplementary Fig. 2a), and therefore, we repeated
if any. We observed that like CXCL12, VUF11207 also fails to elicit any these experiments in cells transfected with higher amount of CXCR7-
measurable G-protein activation from CXCR7 (Fig. 1d). Moreover, V2 plasmid. While the surface expression of CXCR7 improved only
VUF11207 also did not promote any G-protein activation for CXCR4, moderately over mock-transfection (Supplementary Fig. 2c), βarr2
suggesting its selectivity for CXCR7 (Fig. 1d). In these assays, surface recruitment nearly doubled in these experiments (Supplementary
expression of both the receptors was optimized to be at comparable Fig. 2d) further corroborating the ability of CXCR7 to efficiently
levels as measured using a flow-cytometry based assay (Supplemen- recruit βarr2.
tary Fig. 1a). We found the weaker βarr2 recruitment response for CXCR4
We also tested VUF11207 in second messenger assays based on compared to CXCR7 in the Tango assay intriguing, and therefore, we
cAMP response and calcium release, and did not observe a measurable generated NanoBiT assay constructs for CXCR4 and CXCR7, which
response for CXCR7, which further confirms the inability of CXCR7 to enable us to measure βarr response in real time, to probe this further.
activate G-proteins (Fig. 1e–g). A previous study has reported that Here, we generated and tested several combinations of CXCR4 and
stimulation of HEK-293 cells with CXCL12 results in measurable cal- βarr1/2 NanoBiT constructs (i.e. CXCR4-SmBiT+LgBiT-βarr1/2 vs. βarr1/
cium mobilization due to endogenous CXCR420. Thus, the absence of 2-LgBiT, and CXCR4-LgBiT+SmBiT-βarr1/2 vs. βarr1/2-SmBiT), and we
calcium response upon VUF11207 stimulation further indicates the lack observed that CXCR4-SmBiT+LgBiT-βarr1/2 combination yielded

Nature Communications | (2023)14:4808 3

Article https://doi.org/10.1038/s41467-023-40482-9

a b c d
βarr2 recruitment βarr2 recruitment βarr2 recruitment βarr2 recruitment
(PRESTO-Tango assay) (Tango assay) (PRESTO-Tango assay) (Tango assay)
6 CXCR4-V2 40 CXCR4 6 CXCR4-V2 40 CXCR4
(Fold normalized)

5 CXCR7-V2 CXCR7 5 CXCR7-V2 CXCR7

30 30
4 4
3 20 3 20
2 2
10 10
1 1
0 0 0 0
-12 -11 -10 -9 -8 -7 -6 -5 -12 -11 -10 -9 -8 -7 -6 -5 -11 -10 -9 -8 -7 -6 -5 -4 -12 -11 -10 -9 -8 -7 -6 -5 -4
log (CXCL12) [M] log (CXCL12) [M] log (VUF11207) [M] log (VUF11207) [M]

e f
βarr1 recruitment (NanoBiT assay) βarr2 recruitment (NanoBiT assay)
(Receptor-SmBiT+LgBiT-βarr1) (Receptor-SmBiT+LgBiT-βarr2)
20 CXCR4 1.6 10 CXCR4 1.6
(Fold normalized)

CXCR7 8 CXCR7
15 1.4 1.4
6
10 1.2 1.2
4
5 1.0 2 1.0

0 0.8 0 0.8
-13 -12 -11 -10 -9 -8 -7 -6 -5 -13 -12 -11 -10 -9 -8 -7 -6 -5 -13 -12 -11 -10 -9 -8 -7 -6 -5 -13 -12 -11 -10 -9 -8 -7 -6 -5
log (CXCL12) [M] log (CXCL12) [M]
g h i j

βarr1 recruitment (NanoBiT assay) βarr2 recruitment (NanoBiT assay) βarr1 recruitment βarr2 recruitment
(Receptor-SmBiT+LgBiT-βarr1) (Receptor-SmBiT+LgBiT-βarr2) (Receptor-SmBiT+LgBiT-βarr1) (Receptor-SmBiT+LgBiT-βarr2)
20 CXCR4 10 CXCR4 15 AMD3100 + CXCL12 8 AMD3100 + CXCL12
(Fold normalized)

CXCR7 8 CXCR7 AMD3100 AMD3100

15 CXCL12 6 CXCL12
10
6 VUF11207 VUF11207
10 4
4
5
5 2 2

0 0 0 0
-12 -11 -10 -9 -8 -7 -6 -5 -4 -12 -11 -10 -9 -8 -7 -6 -5 -4 -13-12-11-10 -9 -8 -7 -6 -5 -4 -13-12-11-10 -9 -8 -7 -6 -5 -4
log (VUF11207) [M] log (VUF11207) [M] log (agonist) [M] log (agonist) [M]

Fig. 2 | β-arrestin recruitment to CXCR7. a, b CXCL12-induced βarr2 recruitment luminescence signal at minimal ligand dose treated as 1). Response for CXCR4 is
to CXCR4 and CXCR7 in PRESTO-Tango and Tango assays, respectively (mean ± also shown separately in the right panels. g, h VUF11207-induced βarr1/2 recruit-
SEM; n = 3 independent experiments; normalized with the luminescence signal at ment to CXCR4 and CXCR7 in NanoBiT assay (mean ± SEM; n = 4 independent
minimal ligand dose treated as 1). The PRESTO-Tango assay uses a chimeric experiments; normalized with luminescence signal at minimal ligand dose treated
receptor construct with the carboxyl-terminus of V2R while the Tango assay uses as 1). i, j A side-by-side comparison of CXCL12- vs. VUF11207-induced βarr1 and 2
native receptors. c, d VUF11207-induced βarr2 recruitment to CXCR4 and CXCR7 in recruitment to CXCR7, respectively (mean ± SEM; n = 4 independent experiments;
PRESTO-Tango and Tango assays, respectively (mean ± SEM; n = 3 independent normalized with luminescence signal at minimal ligand dose treated as 1). A CXCR4-
experiments; normalized with the luminescence signal at minimal ligand dose specific antagonist AMD3100 is used either alone, or as pre-treatment to CXCL12, as
treated as 1). e, f CXCL12-induced βarr1/2 recruitment to CXCR4 and CXCR7 in a negative control and to rule out the possibility of any contribution from endo-
NanoBiT assay (mean ± SEM; n = 4 independent experiments; normalized with genous CXCR4. Source data are provided as a source data file.

maximal response although it was still only two-fold over basal signal relatively weaker signal for βarr recruitment upon stimulation of
(Supplementary Fig. 3a–f). Subsequently, we used CXCR4/CXCR7- CXCR4 is intriguing, especially considering that CXCR4 harbors sev-
SmBiT+LgBiT-βarr1/2 combination to measure agonist-induced βarr eral potential phosphorylation sites in the carboxyl-terminus, and they
recruitment for the two receptors. Similar to Tango assay data, we have been implicated in agonist-induced βarr recruitment20,22. There-
observed a stronger recruitment in terms of Emax of βarr1 and 2 for fore, additional studies are warranted to further probe this interesting
CXCR7 compared to CXCR4 upon stimulation with CXCL12 (Fig. 2e, f observation with respect to receptor phosphorylation and distinct
and Supplementary Fig. 2e, f) while VUF11207 selectively promoted modes of βarr engagement23–27.
βarr1 and 2 recruitment to CXCR7 but not to CXCR4 (Fig. 2g, h). These A closer analysis of βarr recruitment to CXCR7 in the NanoBiT
data not only suggest the selectivity of VUF11207 for CXCR7 but more assay suggested a difference in the EC50 between CXCL12 and
importantly, also underscore a relatively lower propensity of βarr- VUF11207. Therefore, we compared these two ligands side-by-side and
coupling to CXCR4 compared to CXCR7. We acknowledge that we confirmed that CXCL12 is more potent for βarr recruitment over
cannot unequivocally rule out the possibility that βarr recruitment to VUF11207, although their Emax values are comparable (Fig. 2i, j). In
CXCR4 is comparable to that of CXCR7, however, differential con- order to rule out the contribution of endogenous CXCR4 in relatively
formational restraints in βarrs lead to poor efficiency of protease higher potency of CXCL12 compared to VUF11207, we pre-treated the
cleavage and fragment complementation in the Tango and NanoBiT cells with AMD3100, a CXCR4 antagonist, followed by
assays, respectively, resulting in weaker response. Still however, the CXCL12 stimulation. However, we did not observe any change in

Nature Communications | (2023)14:4808 4

Article https://doi.org/10.1038/s41467-023-40482-9

a βarr2 recruitment b βarr2 recruitment

(PRESTO-Tango assay) (Tango assay)
25 CXCR1-V2 100 CXCR1

(Fold normalized)
(Fold normalized)
CXCR2-V2 CXCR2
20 CXCR3-V2 80 CXCR3
CXCR4-V2 CXCR4
15 CXCR5-V2 60 CXCR5
CXCR6-V2 CXCR6
10 CXCR7-V2 40 CXCR7
Mock Mock
5 20
0 0
-11 -10 -9 -8 -7 -6 -5 -4 -11 -10 -9 -8 -7 -6 -5 -4
log (VUF11207) [M] log (VUF11207) [M]

✱✱✱✱
c d
✱✱✱✱
Gαi-coupling assay (cAMP inhibition)
✱✱✱✱

120

Migration distance (μm)

1000
(% normalized)

100
CXCR1 800
80 CXCR2
60 CXCR3 600
CXCR4
40 CXCR5 400
CXCR6
20 CXCR7 200
C5aR1 (C5a)
0
0
-13-12-11-10 -9 -8 -7 -6 -5 -4

07

07
tro

L1
log (VUF11207) [M]

12

12
on

XC

F1

F1
C

VU

VU
+
2
L1
XC
C
Fig. 3 | Receptor sub-type selectivity of VUF11207 and its ability to promote cell over 2 independent experiments, One-way ANOVA, Šídák’s multiple comparisons
migration. a, b VUF11207-induced βarr2 recruitment for all the CXC chemokine test). The exact p-values are as follows: Control vs. CXCL12 (p < 0.0001), Control vs.
receptors (CXCR1-7) in the PRESTO-Tango and Tango assay, respectively (mean ± VUF11207 (p < 0.0001), Control vs. CXCL12 + VUF11207 (p < 0.0001). The lower and
SEM); n = 5 independent experiments; normalized with the luminescence signal at upper whiskers represent the 5th percentile and the 95th percentile respectively
minimal ligand dose treated as 1. c VUF11207-induced Gαi-coupling for all the CXC and the bounds of box correspond to 25th and 75th percentile. The cross inside the
chemokine receptors (CXCR1-7) in the GloSensor assay (mean ± SEM; n = 3 inde- box indicates the mean (Control: 220.798, CXCL12: 390.201, VUF11207: 341.980,
pendent experiments; normalized with luminescence signal at minimal ligand dose CXCL12 + VUF11207: 325.656). The solid line at the 50th percentile indicates the
treated as 100%). d Migration of MDA-MB-231 stably expressing CXCR7 in response median (Control: 111.150, CXCL12: 325.000, VUF11207: 291.850, CXCL12 +
to indicated ligands. The experiment was carried out using a 2D microfluidic device VUF11207: 257.07)5. (****p < 0.0001). Source data are provided as a source data file.
and each point on the graph represents an individual cell (n = 600 cells examined

CXCL12-induced βarr1/2 recruitment, and AMD3100 also did not elicit activation by CXCL1213,28. Considering the selectivity of VUF11207 for
any response by itself as expected (Fig. 2i, j). CXCR7 over CXCR4, we envisioned that it may be a powerful tool to
specifically measure the contribution of CXCR7 in migration and
Selectivity profiling of VUF11207 on CXCRs invasion of cancer cells. Therefore, we measured the migration of
Inspired by the selectivity of VUF11207 for CXCR7 over CXCR4, we MDA-MB-231 breast cancer cells that were stably transfected with
decided to test it on other CXCRs as well. There are seven CXCRs CXCR7 in response to VUF11207 using a 2D microfluidic device29. We
(CXCR1-CXCR7) in the human genome, and we measured VUF11207 observed that stimulation of these cells with VUF11207 resulted in
response for all of them in parallel using the GloSensor assay for Gαi- efficient migration as compared to vehicle-treated cells, and that
coupling and PRESTO-Tango/Tango assay for βarr2 coupling. We VUF11207 response was comparable to that of CXCL12 (Fig. 3d).
observed that VUF11207 was able to induce βarr2 recruitment only for Simultaneous addition of CXCL12 and VUF11207 did not result in any
CXCR7 and no other CXCRs (Fig. 3a, b). Moreover, VUF11207 was synergistic effect on migration of these cells (Fig. 3d). Taken together
completely silent for every CXCR tested in the GloSensor-based cAMP with the striking sub-type selectivity of VUF11207, these data provide
assay as a readout of Gαi-coupling (Fig. 3c). We observed some varia- direct evidence for a contribution of CXCR7 in the migration of breast
tions in the relative receptor expression in these assays, although they cancer cells that aligns with previous studies30. It would be interesting
all expressed at levels that are higher than the mock-transfected cells to further probe the mechanistic aspect of this observation, for
(Supplementary Fig. 4a–c). Taken together, these findings establish example, with respect to the interplay of CXCR4 and CXCR7 and the
VUF11207 as a CXCR7 selective agonist and therefore, provides a pre- contribution of βarr signaling.
validated tool compound to probe the structure and function of
CXCR7 in future studies. Contribution of GRKs in βarr recruitment
Previous studies have reported that both, CXCR4 and CXCR7 As receptor phosphorylation is a key determinant of βarr
promote migration and invasion of several cancer cell lines upon recruitment, we next tested the contribution of different GRKs in

Nature Communications | (2023)14:4808 5

Article https://doi.org/10.1038/s41467-023-40482-9

a βarr1 recruitment (NanoBiT assay) b βarr2 recruitment (NanoBiT assay)

(CXCR7-SmBiT+LgBiT-βarr1) (CXCR7-SmBiT+LgBiT-βarr2)
15 15 Parent ΔGRK5/6
Parent ΔGRK5/6

(Fold normalized)

(Fold normalized)
ΔGRK2/3 ΔGRK2/3/5/6 ΔGRK2/3 ΔGRK2/3/5/
10 6
10

5 5

0 0
-10 -9 -8 -7 -10 -9 -8 -7
log (CXCL12) [M] log (CXCL12) [M]

c βarr1 recruitment (NanoBiT assay) d βarr2 recruitment (NanoBiT assay)

(CXCR7-SmBiT+LgBiT-βarr1) (CXCR7-SmBiT+LgBiT-βarr2)

15 Parent ΔGRK5/6 15 Parent ΔGRK5/6

(Fold normalized)
(Fold normalized)

ΔGRK2/3 ΔGRK2/3/5/6 ΔGRK2/3 ΔGRK2/3/5/6

10 10

5 5

0 0
-8 -7 -6 -5 -8 -7 -6 -5
log (VUF11207) [M] log (VUF11207) [M]

e βarr1 recruitment (NanoBiT assay) f βarr2 recruitment (NanoBiT assay)

(CXCR7-SmBiT+LgBiT-βarr1) (CXCR7-SmBiT+LgBiT-βarr2)
10 8 Parent ΔGRK5/6
Parent ΔGRK5/6
(Fold normalized)

(Fold normalized)

ΔGRK2/3 ΔGRK2/3/5/6 ΔGRK2/3 ΔGRK2/3/5/6

8 6
6
4
4
2
2
0 0
-10 -9 -8 -7 -10 -9 -8 -7
log (CXCL11) [M] log (CXCL11) [M]
Fig. 4 | Contribution of GRKs in β-arrestin1/2 recruitment to CXCR7. for βarr2 recruitment: Parent, n = 5; ΔGRK2/3, n = 4; ΔGRK5/6, n = 4; ΔGRK2/3/5/6,
a, b CXCL12-induced βarr1/2 recruitment to CXCR7 in GRK knock-out cells using n = 4; normalized with luminescence signal under unstimulated condition treated
the NanoBiT assay (mean ± SEM; n = 4–5 independent experiments; i.e., for βarr1 as 1). e, f CXCL11-induced βarr1 and 2 recruitment to CXCR7 in GRK knock-out cells
recruitment: Parent, n = 5; ΔGRK2/3, n = 4; ΔGRK5/6, n = 4; ΔGRK2/3/5/6, n = 5; for using the NanoBiT assay (mean ± SEM; n = 3–4 independent experiments; i.e., for
βarr2 recruitment: Parent, n = 5; ΔGRK2/3, n = 4; ΔGRK5/6, n = 4; ΔGRK2/3/5/6, βarr1 recruitment: Parent, n = 3; ΔGRK2/3, n = 4; ΔGRK5/6, n = 4; ΔGRK2/3/5/6,
n = 4; normalized with luminescence signal under unstimulated condition treated n = 4; for βarr2 recruitment: Parent, ΔGRK2/3, ΔGRK5/6, and ΔGRK2/3/5/6, n = 4;
as 1). c, d VUF11207-induced βarr1/2 recruitment to CXCR7 in GRK knock-out cells normalized with luminescence signal under unstimulated condition treated as 1).
using the NanoBiT assay (mean ± SEM; n = 4–5 independent experiments; i.e., for Source data are provided as a source data file.
βarr1 recruitment: Parent, n = 5; ΔGRK2/3, n = 4; ΔGRK5/6, n = 4; ΔGRK2/3/5/6, n = 5;

agonist-induced βarr recruitment to CXCR7 using GRK knock-out βarr1 than βarr2. These observations converge with the lack of
cell lines31. We observed that knock-out of GRK5/6 nearly ablates G-protein-coupling to CXCR7 because GRK2/3 translocation to the
CXCR7-βarr1/2 interaction in response to both agonists, CXCL12 plasma membrane and activation has previously been shown to
and VUF11207 while knock-out of GRK2/3 did not influence βarr require Gβγ release32. We also assessed agonist-induced CXCR7-βarr
recruitment to CXCR7 (Fig. 4a–d). In addition to CXCL12, another interaction in presence of pertussis toxin (PTX), however, the pat-
chemokine agonist CXCL11 also binds and activates CXCR717, and tern of recruitment in response to either of the agonists did not
therefore, we also measured the effect of GRK knock-out on CXCL11- change significantly (Supplementary Fig. 5a–f). This observation
induced βarr1/2 recruitment for CXCR7. As presented in Fig. 4e, f, rules out activation-independent contribution of Gαi on βarr
CXCL11-induced βarr1/2 recruitment was also sensitive primarily to recruitment, and therefore, establishes CXCR7 as a model βarr-
GRK5/6 knock-out although GRK2/3 knock-out also appears to have coupled receptor system to further investigate the functional con-
some effect on βarr recruitment, which is more pronounced for tribution of βarrs.

Nature Communications | (2023)14:4808 6

Article https://doi.org/10.1038/s41467-023-40482-9

a Mock CXCR4 CXCR7 b CXCL12-induced

ERK1/2 phosphorylation

WB:pERK1/2 WB:tERK1/2
Time (min) 0 5 15 0 5 15 0 5 15
54kDa Mock CXCR4 CXCR7
200
ns ns ns
43kDa

(% normalized)
✱✱ ✱✱✱✱ ✱✱
150
33kDa
54kDa 100
43kDa
50
33kDa
0
Time (min) 0 5 15 0 5 15 0 5 15

c Mock CXCR7 d CXCL12-induced

ERK1/2 phosphorylation

WB:pERK1/2
Time (min) 0 5 0 5 0 5 0 5 Mock CXCR7
200
✱ ✱✱✱
54kDa

(% normalized)
43kDa 150 ✱✱✱✱ ns ✱✱✱✱ ns

33kDa
100

WB:tERK1/2
54kDa
43kDa 50

33kDa 0
CXCL12 CXCL12 CXCL12 CXCL12 CXCL12 - + - + - + - +
+AMD +AMD AMD3100 - - + + - - + +

e Mock CXCR4 CXCR7 f VUF11207-induced

ERK1/2 phosphorylation
Time (min) 0 5 15 0 5 15 0 5 15
WB:pERK1/2 WB:tERK1/2

10 Mock CXCR4 CXCR7

54kDa
(Fold normalized)

43kDa 8 ns ns ns

33kDa 6 ns ns ns

54kDa 4
43kDa 2
33kDa 0
Time (min) 0 5 15 0 5 15 0 5 15

Fig. 5 | Agonist-induced ERK1/2 phosphorylation for CXCR4 and CXCR7. Mock(-AMD3100): 5 min vs. Mock(+AMD3100): 5 min (p = 0.0192), Mock(+-
a, b CXCL12-induced ERK1/2 phosphorylation in Mock, CXCR4 or CXCR7 trans- AMD3100): 0 min vs. Mock(+AMD3100): 5 min (p = 0.8299), CXCR7(-AMD3100):
fected HEK-293 cells measured by Western blotting. Densitometry-based quantifi- 0 min vs. CXCR7(-AMD3100): 5 min (p < 0.0001), CXCR7(-AMD3100): 5 min vs.
cation (mean ± SEM; n = 7 independent experiments, normalized with the 5 min CXCR7(+AMD3100): 5 min (p = 0.0004), CXCR7(+AMD3100): 0 min vs.
signal for CXCR4 as 100%, Two-way ANOVA, Tukey’s multiple comparison test). The CXCR7(+AMD3100): 5 min (p = 0.3619). e, f VUF11207-induced ERK1/2 phosphor-
exact p-values are as follows: Mock: 0 min vs. Mock: 5 min (p = 0.0095), Mock: ylation in Mock, CXCR4 or CXCR7 transfected HEK-293 cells as measured by
0 min vs. Mock: 15 min (p > 0.9999), CXCR4: 0 min vs. CXCR4: 5 min (p < 0.0001), Western blotting. Densitometry-based quantification (mean ± SEM; n = 7 indepen-
CXCR4: 0 min vs. CXCR4: 15 min (p = 0.9940), CXCR7: 0 min vs. CXCR7: 5 min dent experiments, normalized with respect to the 0 min signal for each condition
(p = 0.0031), CXCR7: 0 min vs. CXCR7: 15 min (p = 0.5552). c, d CXCL12-induced treated as 1, Two-way ANOVA, Tukey’s multiple comparison test). The exact p-
ERK1/2 phosphorylation in CXCR7 expressing cells is blocked by pre-treatment with values are as follows: Mock: 0 min vs. Mock: 5 min (p = 0.9793), Mock: 0 min vs.
AMD3100 (10 μM, 30 min). Densitometry-based quantification (mean ± SEM; n = 6 Mock: 15 min (p > 0.6170), CXCR4: 0 min vs. CXCR4: 5 min (p > 0.9999), CXCR4:
independent experiments, normalized with CXCL12-induced signal for CXCR7 as 0 min vs. CXCR4: 15 min (p > 0.9999), CXCR7: 0 min vs. CXCR7: 5 min (p > 0.9999),
100%, Two-way ANOVA, Tukey’s multiple comparison test). The exact p-values are CXCR7: 0 min vs. CXCR7: 15 min (p = 0.9633) (*p < 0.05, **p < 0.01, ***p < 0.001,
as follows: Mock(-AMD3100): 0 min vs. Mock(-AMD3100): 5 min (p < 0.0001), ****p < 0.0001, ns = non-significant). Source data are provided as a source data file.

Agonist-induced ERK1/2 activation and conformations of βarr2 phosphorylation in HEK-293 cells upon CXCL12-stimulation that
for CXCR4 vs. CXCR7 arises from endogenous CXCR4 and further enhanced by CXCR4
In order to test if βarr recruitment results in ERK1/2 MAP kinase overexpression20, which agrees with the data presented here. Fur-
activation, we measured agonist-induced ERK1/2 phosphorylation thermore, CXCL12-induced ERK1/2 phosphorylation in CXCR4-
for CXCR4 and CXCR7 in response to CXCL12 in transfected HEK-293 expressing cells was also completely abolished by pre-treatment
cells. Although we observed robust ERK1/2 phosphorylation in with pertussis toxin (PTX) suggesting a major dependence on Gαi
CXCR4- and CXCR7-expressing cells, there was a similar response in activation (Supplementary Fig. 6c, d). This is supported by previous
mock-transfected cells as well (Fig. 5a, b). Moreover, ERK1/2 phos- studies showing that CXCL12-induced ERK1/2 phosphorylation
phorylation was effectively blocked by the pre-treatment of does not require βarrs, using either siRNA-based knock-down of
AMD3100 (CXCR4 antagonist) (Fig. 5c, d and Supplementary Fig. 6a, βarrs in HeLa cells33, or, βarr knock-out MEFs (mouse embryonic
b). Thus, CXCL12-induced ERK1/2 phosphorylation likely arises fibroblasts)34. Interestingly, VUF11207 failed to elicit any measurable
from the endogenous CXCR4 expressed in HEK-293 cells16. We also ERK1/2 activation at either CXCR4 or CXCR7 (Fig. 5e, f). In these
note that a previous study has reported measurable ERK1/2 experiments, the receptors were expressed at significantly higher

Nature Communications | (2023)14:4808 7

Article https://doi.org/10.1038/s41467-023-40482-9

a b
CCPGCC Tetra cysteine motif)

F1 F2 F3 F4 F5 F6
βarr2
FlAsH
40 140 171 225 263 410
βarr2 conformation (FlAsH-BRET)
β-arrestin 2

F1

F2

F3

F4

F5

F6
R-Luc N domain C domain
ns ns ns ✱✱ ns ✱
C-terminus 0.10
0.05

ΔBRET
0.00
-0.05
-0.10 CXCR4
-0.15 CXCR7

Fig. 6 | Conformational changes in βarr2 upon interaction with CXCR4 vs. (mean ± SEM), n = 6 independent experiments; the difference of the net-BRET ratio
CXCR7. a Schematic representation of intramolecular BRET-based sensors of βarr2 between the stimulated and unstimulated condition were plotted, Two-way
conformation where the N-terminus of βarr2 harbors R-Luc (Renilla luciferase) as ANOVA, Sidak’s multiple comparison test. The exact p-values are as follows:
the BRET donor, and the FlAsH motif are engineered at indicated positions in βarr2 F1:CXCR4 vs. F1:CXCR7 (p = 0.9746), F2:CXCR4 vs. F2:CXCR7 (p > 0.9999),
as BRET acceptor. The structural representation in the lower panel is designed F3:CXCR4 vs. F3:CXCR7 (p = 0.9991), F4:CXCR4 vs. F4:CXCR7 (0.0010), F5:CXCR4
based on alpha fold generated model of βarr2. b CXCL12-induced BRET signal vs. F5:CXCR7 (p = 0.9052), F6:CXCR4 vs. F6:CXCR7 (p = 0.0420) (*p < 0.05,
measured in HEK-293 cells expressing the indicated receptor and sensor constructs **p < 0.01, ns non-significant). Source data are provided as a source data file.

level than mock-transfected cells (Supplementary Fig. 6e, f). While Identification of the key phosphorylation-site cluster in CXCR7
the lack of VUF11207-induced ERK1/2 response at CXCR4 is expected, As receptor phosphorylation is a key determinant of βarr binding27,37,38,
its inability to activate ERK1/2 at CXCR7, in spite of robust βarr we analyzed the carboxyl-terminus of CXCR7 and identified two dis-
recruitment, is intriguing. Taken together, these data reveal that tinct phosphorylation clusters, each containing three potential phos-
CXCR7 activation, either by CXCL12 or VUF11207, does not initiate phorylation sites (Fig. 7a). These two clusters, referred to as cluster1
ERK1/2 MAP kinase activation despite robust βarr recruitment. These and 2 from here onwards, harbor PXXPXXP and PXPXXP type pattern
findings also underscore the notion that βarr recruitment does not of phosphorylation sites, respectively, where P represents a Serine or
always translate to ERK1/2 activation and it may depend on specific Threonine. We have previously determined the structure of βarr2 in
conformation adopted by βarrs in complex with 7TMRs. complex with a phosphopeptide corresponding to cluster1, which
Prompted by the lack of ERK1/2 phosphorylation despite robust revealed a partially-active βarr2 conformation in terms of the inter-
βarr recruitment, we decided to probe the conformation of βarr2 domain rotation39. We generated two different CXCR7 constructs by
upon its interaction with CXCR4 and CXCR7. We used a BRET-based mutating the phosphorylation sites in these clusters, and monitored
approach where tetra-cysteine FlAsH motifs are engineered at six agonist-induced βarr1/2 recruitment and endosomal localization. We
distinct sites in βarr2 as BRET acceptor while the Renilla luciferase (R- observed that the mutation of cluster2 phosphorylation sites nearly
Luc) is engineered at the N-terminus as BRET donor35,36 (Fig. 6a). A abolished βarr1/2 recruitment and endosomal localization (Fig. 7b–e
change in BRET signal therefore reports conformational changes in and Supplementary Fig. 7a–d). On the other hand, cluster1 mutation
βarr2, and a side-by-side comparison of BRET signal for these six had a modest effect only on βarr1 recruitment without affecting βarr2
different sensors offers readout of conformational changes in βarr2 recruitment, and βarr1/2 endosomal localization (Fig. 7b–e). This
imparted by its interaction with a receptor. Interestingly, we observation potentially hints at differential contribution of specific
observed that the changes in BRET signal for at least two of these phosphorylation sites in the recruitment of βarr isoforms, and it would
sensors were significantly different between CXCR4 and CXCR7 in be an interesting direction to investigate further in future studies.
response to CXCL12 (Fig. 6b). For example, in case of sensor F4 where Taken together, these data establish cluster2 as the major determinant
the FlAsH label is localized between β-strand 14 and 15, there was a of βarr interaction and endosomal localization for CXCR7, and
decrease in BRET signal for CXCR4 but an increase for CXCR7. prompted us to investigate the contribution of individual sites in this
Similarly, in case of sensor F6 where the FlAsH label is localized at phosphorylation cluster.
position 410 in the carboxyl-terminus, there was an increase in BRET
signal for CXCR4 but a decrease for CXCR7. In other sensors, the A key phosphorylation site in CXCR7 drives βarr recruitment
pattern of BRET change was similar between the two receptors. and endosomal localization
Taken together, these data suggest that βarr2 adopts distinct con- We generated a series of CXCR7 mutants lacking either one or a
formations upon its recruitment to CXCR4 vs. CXCR7 in response to combination of phosphorylation sites in cluster2, and tested their
CXCL12 although additional studies would be required to identify the ability to promote βarr1/2 recruitment upon stimulation with CXCL12
precise nature of these conformational changes in more detail. and VUF11207 (Fig. 8a, b and Fig. 9a–d). We observed that the mutation
Moreover, as noted earlier, CXCL11 also promotes βarr recruitment of a single phosphorylation site i.e. Thr352 resulted in a dramatic
through CXCR7, and therefore, it would be interesting to compare it decrease in βarr1/2 recruitment, while the other two sites had relatively
alongside CXCL12 and VUF11207 in terms of βarr conformation in modest effect individually (Fig. 8a, b and Fig. 9a–d). A combination of
future studies. two sites where one was Thr352 had an additive effect in terms of

Nature Communications | (2023)14:4808 8

Article https://doi.org/10.1038/s41467-023-40482-9

a
CXCR7 C terminus

320R N Y R Y E L M K A F I F K Y S A K T G L T K L I D A S R V S E T E Y S A L E Q S T K362

Cluster1 Cluster2
(PXXPXXP) (PXPXXP)

b βarr1 recruitment c βarr2 recruitment

(NanoBiT assay) (NanoBiT assay)
25 WT 15 WT

(Fold normalized)
(Fold normalized)

20 Cluster1 Cluster1
Cluster2 10 Cluster2
15
10
5
5
0 0
-13 -12 -11 -10 -9 -8 -7 -6 -5 -13 -12 -11 -10 -9 -8 -7 -6 -5
log (CXCL12) [M] log (CXCL12) [M]

d βarr1 endosomal localization e βarr2 endosomal localization

(NanoBiT assay) (NanoBiT assay)
1.8 WT 1.8 WT
(Fold normalized)

(Fold normalized)
1.6 Cluster1 1.6 Cluster1
Cluster2 Cluster2
1.4 1.4
1.2 1.2
1.0 1.0
0.8 0.8
-12 -11 -10 -9 -8 -7 -6 -5 -12 -11 -10 -9 -8 -7 -6 -5

log (CXCL12) [M] log (CXCL12) [M]

Fig. 7 | Identification of the key phosphorylation site cluster in CXCR7. a The independent experiments; normalized with luminescence signal at minimal ligand
carboxyl-terminus of CXCR7 harbors two potential phosphorylation site clusters dose treated as 1). d, e Dose response curves of CXCL12-induced endosomal loca-
indicated as cluster 1 and 2 with PXXPXXP and PXPXXP type phosphorylation codes lization of βarr1/2 for the indicated CXCR7 constructs measured using the NanoBiT
(P is Ser/Thr; X is any other amino acid) (the schematic for the receptor is created assay (Receptor+SmBiT-βarr1/2+LgBiT-FYVE) (mean ± SEM; n = 4 independent
with BioRender.com; same as in Fig. 1a/c). b, c Dose response curves for CXCL12- experiments; normalized with the luminescence signal at minimal ligand dose
induced βarr1 and 2 recruitment to indicated CXCR7 constructs measured using treated as 1). Source data are provided as a source data file.
the NanoBiT assay (Receptor-SmBiT+LgBiT-βarr1/2) (mean ± SEM; n = 4

attenuation in βarr binding, while the mutant lacking all three phos- promiscuity compared to other prototypical GPCRs, and therefore,
phorylation sites completely lost βarr recruitment. In addition to the may contain several examples of naturally-encoded ligand and recep-
data presented in Fig. 8a, b and Fig. 9a–d where only two saturating tor bias45,46. Our data now establish the CXCR4-CXCR7 system as an
doses of agonists were used, we also carried out complete dose intriguing example of GPCR-ACR pair, and uncover intrinsic bias
response experiments on selected mutants, which further recapitu- encoded at the level of transducer-coupling and functional responses.
lated the same pattern of βarr1/2 recruitment (Fig. 8c, d and Moreover, our findings also establish VUF11207 as a highly selective
Fig. 10a–d). In order to test if diminished βarr recruitment also trans- agonist for CXCR7 and demonstrate that its efficacy is comparable to
lates to reduced functional response, we measured agonist-induced CXCL12, although the potency is slightly weaker.
endosomal localization of βarr1/2, and observed near-complete loss of While we comprehensively establish the lack of G-protein activa-
endosomal localization of βarr1 and 2 for Thr352Ala mutation (Fig. 8e, f tion and second messenger response for CXCR7, the underlying
and Fig. 10e, f), which mirrors the recruitment pattern presented in molecular mechanism remains to be explored. For example, the aty-
Fig. 8c, d and Fig. 10a–d. In these experiments, the surface expression pical chemokine receptors lack the signature motifs such as the DRY
of different mutants was optimized to be comparable to the wild-type and NPXXY that are present in prototypical GPCRs although recon-
receptor (Supplementary Fig. 8a–d and Supplementary Fig. 9a–f). stitution of these motifs by site-directed mutagenesis does not
Taken together, these data suggest that Thr352 is a key residue involved necessarily result in robust gain of G-protein-coupling and
in directing βarr recruitment and endosomal localization for CXCR7. activation10,11,14,47–49. On the other hand, CXCR7 contains both, the DRY
and NPXXY motif similar to prototypical GPCRs but still lacks func-
Discussion tional G-protein-coupling. These data suggest a possible conforma-
The conceptual framework of biased agonism has focused primarily on tional mechanism wherein the agonist-induced conformation of
ligands that induce distinct preferences of transducer-coupling40–43, CXCR7 is somewhat in an intermediate state with a partial opening on
although biased receptor mutants have also been described for a the intracellular side of the receptor incompatible with G-protein-
handful of receptors43,44. Chemokine receptors are peculiar in this coupling and/or activation50. It is also important to note that a recent
context as they display a significantly higher degree of ligand cryo-EM structure of CXCL12-CXCR7 complex has been determined,

Nature Communications | (2023)14:4808 9

Article https://doi.org/10.1038/s41467-023-40482-9

a βarr1 recruitment (NanoBiT assay) b βarr1 recruitment (NanoBiT assay)

200 CXCL12 150 VUF11207
ns

(% normalized)
ns (μM) (μM)

(% normalized)
150
0 100 0
0.1 ****
100 1
**** **** 0.2 50 10
50 **** **** ****
**** **** **** ****
**** ****
0 0
S3 T
T3 0A

52 T3 A
5 S A

S3 T
T3 A

5 T A
50 S3 A
S3 A
C 55A

2
50 S3 A
S3 A
C 5 5A
er

er
W

W
T3 0A+ 355
52

50

T3 0A+ 355
S3 A+ 52
A+ 55

52

S3 2A+ 352
A+ 55
5

st

st
lu

lu
5 S
S3

S3
c βarr1 recruitment d βarr1 recruitment
(NanoBiT assay) (NanoBiT assay)
120 WT 120 WT
(% normalized)

(% normalized)
T352A T352A
80 S355A 80 S355A
Cluster2 Cluster2
40 40

0 0
-12 -11 -10 -9 -8 -7 -6 -5 -11 -10 -9 -8 -7 -6 -5 -4
log (CXCL12) [M] log (VUF11207) [M]

e βarr1 endosomal locallization f βarr1 endosomal localization

(NanoBiT assay) (NanoBiT assay)
2.0 WT 2.0 WT
(Fold normalized)
(Fold normalized)

1.8 T352A 1.8 T352A

1.6 Cluster2 1.6 Cluster2
1.4 1.4
1.2 1.2
1.0 1.0
0.8 0.8
-13 -12 -11 -10 -9 -8 -7 -6 -5 -13-12-11-10 -9 -8 -7 -6 -5 -4
log (CXCL12) [M] log (VUF11207) [M]

Fig. 8 | Key phosphorylation sites in CXCR7 driving βarr1 recruitment and S355A (p < 0.0001), WT vs. S350A + T352A (p < 0.0001), WT vs. T352A + S355A
endosomal localization. a, b CXCL12- and VUF11207-induced βarr1 recruitment, (p < 0.0001), WT vs. S350A + S355A (p < 0.0001), WT vs. Cluster2 (p < 0.0001).
respectively, to the indicated phosphorylation site mutants of CXCR7 using the c, d Dose response curves of CXCL12- and VUF11207-induced βarr1 recruitment to
NanoBiT assay (mean ± SEM); n = 3–4 independent experiments; i.e., for CXCL12- selected phosphorylation site mutants of CXCR7 in the NanoBiT assay (Receptor-
induced βarr1 recruitment: WT, S350A, T352A, S355A, S350A + T352A, T352A + SmBiT+LgBiT-βarr1) (mean ± SEM; n = 3 independent experiments; normalized with
S355A, and Cluster2, n = 4; S350A + S355A, n = 3; for VUF11207-induced βarr1 luminescence signal at maximal ligand dose for wild-type treated as 100%). e, f Dose
recruitment: WT, S350A, T352A, S355A, S350A + T352A, T352A + S355A, and Clus- response curves of CXCL12- and VUF11207-induced βarr1 endosomal localization
ter2, n = 4; S350A + S355A, n = 3; normalized with luminescence signal at maximal for the selected phosphorylation site mutants of CXCR7 in the NanoBiT assay
ligand dose for wild-type treated as 100%, Two-way ANOVA, Dunnett’s multiple (receptor+SmBiT-βarr1 + LgBiT-FYVE) (mean ± SEM; n = 3–7 independent experi-
comparisons test. The exact p-values are: for CXCL12-induced βarr1 recruitment; ments; i.e., for CXCL12-induced βarr1 endosomal localization: WT, n = 6; T352A,
WT vs. S350A (p = 0.9972), WT vs. T352A (p < 0.0001), WT vs. S355A (p < 0.0001), n = 3; and Cluster2, n = 3; for VUF11207-induced βarr1 endosomal localization: WT,
WT vs. S350A + T352A (p < 0.0001), WT vs. T352A + S355A (p < 0.0001), WT vs. n = 7; T352A, n = 3; and Cluster2, n = 4; normalized with the luminescence signal at
S350A + S355A (p < 0.0001), WT vs. Cluster2 (p < 0.0001). For VUF11207-induced minimal ligand dose treated as 1). (****p < 0.0001, ns non-significant). Source data
βarr1 recruitment; WT vs. S350A (p = 0.4700), WT vs. T352A (p < 0.0001), WT vs. are provided as a source data file.

and the authors proposed that the positioning of ICL2 may interfere context such as presence of specific lipid environment, accessory
with efficient G-protein-coupling to the receptor51. It is also fascinating proteins and post-translational modifications that may also contribute
to explore whether lack of G-protein activation as demonstrated here to the lack of G-protein activation by CXCR7 measured in HEK-
using heterotrimer dissociation assay, and second messenger 293 cells.
response reported in several studies, reflects an absence of physical It is also noteworthy that despite a robust recruitment of βarrs,
interaction between the receptor and G-proteins or, the inability to CXCR7 fails to elicit any measurable ERK1/2 phosphorylation in
activate G-proteins despite physical coupling. This becomes particu- response to either CXCL12 or VUF11207. Previous studies using a
larly important considering the reports that CXCR7 and G-proteins combination of direct binding assays and biophysical methods have
may exist in close proximity in cellular context although agonist- correlated βarr conformation and activation with the interaction of c-
induced increase in their proximity and/or interaction was not Raf1, MEK-1 and ERK2 binding, and ERK2 phosphorylation in-vitro52,53.
apparent51. Finally, we cannot rule out the contribution of cellular Thus, it is plausible that the lack of ERK1/2 activation downstream of

Nature Communications | (2023)14:4808 10

Article https://doi.org/10.1038/s41467-023-40482-9

a b
βarr2 recruitment (Tango assay) βarr2 recruitment (Tango assay)
150 CXCL12 150 VUF11207

(% normalized)
(μM) (μM)

(% normalized)
****
****
100 **** 0 100 0
0.1 1
**** ****
0.2 50
**** 10
50
**** ****
**** **** **** **** **** ****

0 0

S3 T
T3 A

5 T A
50 S 3 A

2
50 S3 A
S3 A
C 55A
S3 T
T3 A

5 T A
50 S 3 A

2
50 S3 A
S3 A
C 55 A

er
W
50

T3 A+ 55
er

52

S3 2A+ 352
A+ 55
W
50

T3 A+ 55
52

S3 2A+ 352
A+ 55

st
st

lu
lu

S3
S3

c d
βarr2 recruitment (NanoBiT assay) βarr2 recruitment (NanoBiT assay)
150 CXCL12 150 VUF11207
ns ns
(μM)

(% normalized)
(μM)
(% normalized)

****
100 *** 0 100 0
**** 0.1 1
**** ****
0. 50
**** 10
50
**** **** 2 **** **** ****
****

0 0
S3 T
T3 A

5 T A
5 S A

2
50 S3 A
S3 A
C 55A
S3 T
T3 A

5 T A
50 S 3 A

2
50 S3 A
S3 A
C 55A

er
W
er

50

T3 0A+ 355
52

S3 2A+ 352
A+ 55
W
50

T3 A+ 55
52

A+ 55
S3 2A+ 352

st
st

lu
lu

S3
S3

Fig. 9 | Contribution of different phosphorylation sites in CXCR7-mediated S355A (p < 0.0001), WT vs. S350A + S355A (p < 0.0001), WT vs. Cluster2
βarr1/2 recruitment. a, b CXCL12- and VUF11207-induced βarr2 recruitment, (p < 0.0001). c, d CXCL12- and VUF11207-induced βarr2 recruitment, respectively,
respectively, to the indicated phosphorylation site mutants of CXCR7 using the to the indicated phosphorylation site mutants of CXCR7 using the NanoBiT assay
Tango assay (mean ± SEM); n = 4–6 independent experiments; i.e., for CXCL12: WT, (mean ± SEM); n = 3 independent experiments; normalized with luminescence sig-
S350A, T352A, S355A, S350A + T352A, T352A + S355A, and Cluster2, n = 6; S350A + nal at maximal ligand dose for wild-type treated as 100%, Two-way ANOVA, Dun-
S355A, n = 4; for VUF11207-induced βarr2 recruitment: WT, S350A, T352A, S355A, nett’s multiple comparisons test. The exact p-values are: for CXCL12; WT vs. S350A
S350A + T352A, T352A + S355A, and Cluster2, n = 6; S350A + S355A, n = 4; normal- (p = 0.9994), WT vs. T352A (p < 0.0001), WT vs. S355A (p = 0.0001), WT vs.
ized with luminescence signal at maximal ligand dose for wild-type treated as 100%, S350A + T352A (p < 0.0001), WT vs. T352A + S355A (p < 0.0001), WT vs. S350A +
Two-way ANOVA, Dunnett’s multiple comparisons test. The exact p-values are: for S355A (p < 0.0001), WT vs. Cluster2 (p < 0.0001). For VUF11207; WT vs. S350A
CXCL12-induced βarr2 recruitment; WT vs. S350A (p < 0.0001), WT vs. T352A (p = 0.8957), WT vs. T352A (p < 0.0001), WT vs. S355A (p < 0.0001), WT vs.
(p < 0.0001), WT vs. S355A (p < 0.0001), WT vs. S350A + T352A (p < 0.0001), WT vs. S350A + T352A (p < 0.0001), WT vs. T352A + S355A (p < 0.0001), WT vs. S350A +
T352A + S355A (p < 0.0001), WT vs. S350A + S355A (p < 0.0001), WT vs. Cluster2 S355A (p < 0.0001), WT vs. Cluster2 (p < 0.0001) (***p < 0.001, **** p < 0.0001, ns
(p < 0.0001). For VUF11207; WT vs. S350A (p < 0.0001), WT vs. T352A (p < 0.0001), non-significant). Source data are provided as a source data file.
WT vs. S355A (p < 0.0001), WT vs. S350A + T352A (p < 0.0001), WT vs. T352A +

CXCR7 despite robust βarr recruitment reflects a βarr conformation also critical in determining βarr binding and functional outcomes,
that is not conducive to scaffolding and/or activation of ERK1/2 MAP in addition to the phosphorylation bar-code as suggested
kinase. In fact, our data with intramolecular BRET sensor of βarr2 hints previously54–57.
at distinct conformations in βarr2 induced by CXCR4 vs. CXCR7. A In summary, our study provides molecular insights into the
change in BRET for an intramolecular sensor typically reflects a change intrinsic-bias encoded in the CXCR4-CXCR7 system, and present an
in the distance and/or orientation of the donor and acceptor moieties, important advance to better understand the intricate details of
and therefore, directionally opposite changes in BRET signal for 7TMR-βarr interaction and signaling with potential therapeutic
CXCR4 vs. CXCR7 for the same sensor suggest a difference in distance implications.
and/or orientation of the FlAsH sites relative to the N-terminus (i.e.
R-Luc fusion site). However, direct structural visualization of βarrs in Methods
complex with CXCR4 and CXCR7 at high resolution is required to General reagents, plasmids, and cell culture
better understand the precise conformational differences, and how Most of the general reagents were purchased from Sigma Aldrich
they are linked to resulting functional outcomes for these two recep- unless mentioned otherwise. Dulbecco′s Modified Eagle′s Medium
tors. Taken together, these data support the notion that βarr-binding (DMEM), Fetal-Bovine Serum (FBS), Dulbecco’s Phosphate buffer
does not always translate into ERK1/2 activation at 7TMRs, and saline (PBS), Trypsin-EDTA, Hank’s balanced salt solution (HBSS), and
receptor-bound conformations of βarrs fine-tune the resulting penicillin-streptomycin solution were purchased from Thermo
signaling outcomes. The identification of Thr352 as a key site driving Fisher Scientific. HEK-293 cells (ATCC) and HTLA cells were main-
CXCR7-βarr recruitment and endosomal localization of βarrs presents tained in DMEM (Gibco, Cat. no. 12800-017) supplemented with 10%
an interesting paradigm where the spatial positioning of a single site is FBS (Gibco, Cat. no. 10270-106) and 100 U ml−1 penicillin (Gibco, Cat.

Nature Communications | (2023)14:4808 11

Article https://doi.org/10.1038/s41467-023-40482-9

a βarr2 recruitment b βarr2 recruitment

(Tango assay) (Tango assay)
120 WT 120 WT

(% normalized)
(% normalized)
S350A S350A
80 T352A 80 T352A
Cluster2 Cluster2
40 40

0 0
-12 -11 -10 -9 -8 -7 -6 -5 -11 -10 -9 -8 -7 -6 -5 -4
log (CXCL12) [M] log (VUF11207) [M]

c
βarr2 recruitment d
βarr2 recruitment
(NanoBiT assay) (NanoBiT assay)
120 WT 120 WT
(% normalized)

(% normalized)
T352A T352A
80 S355A 80 S355A
Cluster2 Cluster2
40 40

0 0
-12 -11 -10 -9 -8 -7 -6 -5 -11 -10 -9 -8 -7 -6 -5 -4
log (CXCL12) [M] log (VUF11207) [M]

e βarr2 endosomal localization f βarr2 endosomal localization

(NanoBiT assay) (NanoBiT assay)
1.8 WT 2.4 WT
(Fold normalized)

(Fold normalized)

1.6 T352A T352A

Cluster2 2.0 Cluster2
1.4
1.6
1.2
1.0 1.2

0.8 0.8
-13 -12 -11 -10 -9 -8 -7 -6 -5 -12 -10 -8 -6 -4
log (CXCL12) [M] log (VUF11207) [M]
Fig. 10 | Effect of Thr352Ala mutation in CXCR7 on βarr2 recruitment and phosphorylation site mutants of CXCR7 in the NanoBiT assay (Receptor+SmBiT-
endosomal localization. a–d Dose-response curves of CXCL12- and VUF11207- βarr2 + LgBiT-FYVE) (mean ± SEM; n = 3–7 independent experiments; i.e., for
induced βarr2 recruitment to selected phosphorylation site mutants of CXCR7 in CXCL12-induced βarr2 endosomal localization: WT, n = 6; T352A, n = 3; and Clus-
the Tango and NanoBiT assays (mean ± SEM; n = 6 independent experiments for ter2, n = 3; for VUF11207-induced βarr2 endosomal localization: WT, n = 7; T352A,
a, b and n = 3 independent experiments for c, d; normalized with luminescence n = 3; and Cluster2, n = 4; normalized with the luminescence signal at minimal
signal for WT at maximal ligand dose treated as 100%). e, f Dose-response curves of ligand dose treated as 1). Source data are provided as a source data file.
CXCL12- and VUF11207-induced βarr2 endosomal localization for the selected

no. 15140122) and 100 μg ml−1 streptomycin (Gibco, Cat. no. 15140- G-protein dissociation assay, helical domain of Gα subunit was tag-
122) at 37 °C in 5% CO2. The cDNA coding region for CXCR1-7 were ged with large fragment (LgBiT) and small fragment (SmBiT) was
cloned in pcDNA3.1 with an N-terminal FLAG tag and an HA signal fused at the N-terminus of Gγ. All constructs were verified by DNA
sequence. PRESTO-Tango assay constructs were acquired from sequencing (Macrogen). The oligonucleotide primers used in this
Addgene (Cat. no. 1000000068) while the Tango assay constructs study are listed in Supplementary Table 1. Recombinant CXCL12 was
were generated in our laboratory58. CXCR7 phosphorylation site purchased from PeproTech (Cat. no. 300-28 A), and VUF11207 was
mutants were generated by site-directed mutagenesis using Q5 Site- from Sigma (Cat. no. SML0669). In addition, we also synthesized
Directed Mutagenesis Kit (NEB, Cat. no. E0554S). For the NanoBiT VUF11207 following a previously published protocol18 and used it in
assay, receptor constructs with carboxyl-terminus SmBiT were gen- some experiments. The antibodies used in this study were HRP-
erated. For nanoBiT based βarr recruitment assay, large fragment conjugated anti-FLAG M2 (1:2,000; Sigma-Aldrich, Cat. no. A8592),
(LgBiT) of the nanoluciferase enzyme was fused at the N-terminus of anti-pERK1/2 (1:5,000; Cell Signaling Technology, Cat. no. 9101), anti-
βarr. To study endosomal localization, βarr was tagged with small tERK1/2 (1:5,000; Cell Signaling Technology, Cat. no. 9102), and HRP-
fragment (SmBiT) of nanoluciferase and large fragment (LgBiT) was coupled anti-rabbit IgG secondary antibody (1:10,000 dilution;
fused to FYVE domain of human endofin with a flexible linker. For Genscript, Cat. No. A00098).

Nature Communications | (2023)14:4808 12

Article https://doi.org/10.1038/s41467-023-40482-9

Synthesis of VUF11207 experiments presented in Fig. 1e, 0.25 μg of CXCR4 and 5 μg of CXCR7
The schematic of VUF11207 synthesis is shown in the Supplementary were used along with 2 μg of 22 F cAMP biosensor. For the Gαs-
Fig. 10. A detailed protocol of VUF11207 synthesis has been published coupling assay (Fig. 1f), 0.5 μg of CXCR4, 0.5 μg of V2R and 4 μg of
earlier18. Briefly, Aldol condensation of 2-fluorobenzaldehyde with CXCR7 plasmids were used along with 3 μg of 22 F cAMP biosensor. For
propionaldehyde under basic condition gave (E)-3-(2-Fluorophenyl)-2- the experiments presented in Fig. 3c, 3.5 μg of CXCR1, CXCR3, and
methylacrylaldehyde (1) in 72% yield. In the presence of acetic acid, CXCR4 plasmids were used while 5 μg of plasmids were used for
aldehyde 1 reacted with 2-(1-methylpyrrolidin-2-yl) ethanamine in CXCR2, CXCR5, CXCR6, and CXCR7 along with 2 μg of 22 F cAMP
methanol, and the resulting imine was reduced with NaBH(OAc)3 to biosensor.
generate (E)-3-(2-fluorophenyl)-2-methyl-N-(2-(1-methylpyrrolidin-2-
yl)ethyl) prop-2-en-1-amine (2) in 50% yield. Under the standard amide Calcium flux assay
coupling conditions (EDCI/HOBt/DIEA), amine 2 was coupled with HEK-293 cells were transiently transfected with pGP-CMV-GCaMP6s
3,4,5-trimethoxybenzoic acid to afford VUF11207 with 67% yield. The (Ca2+ Sensor plasmid, Addgene, Cat. no. 40753; 4 µg), 5HT2c receptor
structures of target compound and the intermediates were confirmed (as positive control, cDNA.org, Cat. no. HTR02CTN00; 4 µg) or CXCR4/
by their spectral properties. CXCR7 receptor (4 µg) using PEI max in a ratio of 1:4 (DNA:PEI max) and
plated at a density of 5 × 104 cells well−1 in black optical bottom plate in
NanoBiT-based G-protein dissociation assay complete DMEM media (10% FBS). After 14-16 h of transfection, media
Ligand-induced G-protein activation was measured using NanoBiT- from the plate was aspirated and 100 μl of Ca2+/Mg2+ free HBSS buffer
based G-protein dissociation assay19. Briefly, a NanoBiT-G-protein (pH 7.2) was added, cells were further incubated at 37 °C for 10 min in
consisting of Gα subunit tagged with LgBiT and SmBiT-tagged the Flex Station 3 (Molecular Devices) before the assay was initiated.
Gγ2 subunit along with the untagged Gβ1 subunit were co-expressed Ligand induced change in relative fluorescence unit (RFU) was mea-
with the indicated receptor constructs and ligand-induced change in sured at excitation 485 nm and emission 525 nm (cut off 515 nm) with
luminescence signal was measured. Typically, HEK-293 (Thermo Fisher the settings of 6 reads well−1. Basal fluorescence of each well was
Scientific) cells were transfected with a plasmid mixture consisting of recorded for 15 s, and then 20 µl of 6X concentration of each agonist
100 ng Gα-LgBiT, 500 ng Gβ1, 500 ng SmBiT-Gγ2 (C68S) with either (Serotonin and VUF11207) as indicated was added using robotic
50 ng of CXCR4 plasmid or 3.5 μg of CXCR7. The receptor constructs pipetting of FlexStation system and RFU was recorded at 2 s interval
used in this assay contain N-terminal HA signal sequence, FLAG tag and for a total of 135 s. The changes in RFU (ΔRFU) for each treatment
a flexible linker sequence. To enhance NanoBiT-G-protein expression group was calculated by subtracting the average basal response (RFU
for Gs, Gq and G12/13, 100 ng of RIC8B plasmid (isoform 2; for Gs) or before ligand addition) from RFU of each well at each time points after
RIC8A (isoform 2; for Gq, G12, and G13) were also co-transfected. 24 h ligand addition. ΔRFU for each ligand was plotted and analyzed using
post-transfection, cells were harvested with EDTA-containing GraphPad Prism 9.5.0 software.
PBS, centrifuged, and suspended in 2 ml of HBSS (Gibco, Cat. no.
14065-056) containing 0.01% bovine serum albumin (BSA fatty Receptor surface expression
acid–free grade, SERVA) and 5 mM 4-(2-hydroxyethyl)-1-piper- In order to measure the surface expression of the receptors in various
azineethanesulfonic acid (HEPES), pH 7.4 (assay buffer). Afterwards, assays, we used a previously described whole cell-based surface ELISA
cells were dispensed in a white 96-well plate (80 μl well−1), incubated assay60. Transfected cells from the corresponding assays were seeded
with 20 μl of 50 μM coelenterazine (Carbosynth, Cat. no. EC175526), into a 24-well plate pre-coated with 0.01% poly-D-Lysine at a density of
and 2 h later, baseline luminescence was measured (SpectraMax L, 2 × 105 cells well−1 and incubated at 37 °C for 24 h. Afterwards, cells
Molecular Devices). Subsequently, 20 μl of 6X agonist, serially diluted were washed once with ice-cold 1XTBS, fixed with 4% PFA (w/v in
in the assay buffer, were manually added and the plate was immedi- 1XTBS) on ice for 20 min, washed again three times with 1XTBS, and
ately read for the second measurement in a kinetic mode. Lumines- blocked at room temperature for 1.5 h with 1% BSA prepared in 1XTBS.
cence counts recorded from 3-5 min post-agonist addition were Subsequently, the cells were incubated with anti-FLAG M2-HRP anti-
averaged, corrected with the baseline signals, normalized with respect body (Sigma, Cat. no. A8592) (1:2,000 for 1.5 h at room temperature)
to vehicle control plotted using the GraphPad Prism 9.5.0 software. followed by three washes in 1% BSA and incubation with TMB-ELISA
substrate (Thermo Fisher Scientific, Cat. no. 34028) until the light blue
GloSensor-based cAMP assay color appeared. The signal was quenched by transferring 100 μl of the
In order to assess agonist-induced coupling of Gαs and Gαi, we used colored solution to another 96 well plate containing 100 μl of 1 M
GloSensor-based cAMP assay59. Briefly, HEK-293 cells were co- H2SO4, and the absorbance was measured at 450 nm. For normal-
transfected with FLAG-tagged receptor constructs and luciferase- ization, TMB substrate was removed, cells were washed twice with
based 22 F cAMP biosensor (Promega, Cat. no. E2301) using poly- 1XTBS, and incubated with 0.2% (w/v) Janus Green (Sigma, Cat. no.
ethylenimine (PEI) linear (Polysciences, Cat. no. 19850) at a ratio of 1:3 201677) for 15 min at room temperature. The excess stain was removed
(DNA:PEI linear) as transfection reagent. After 14-16 h of transfection, by washing the cells with water followed by addition of 800 μl of 0.5 N
cells were detached from the plates, resuspended in assay buffer HCl in each well, and 200 μl of this solution was transferred to a 96 well
(1XHBSS, 20 mM HEPES, pH7.4) containing D-luciferin (0.5 mg ml−1, plate for measuring the absorbance at 595 nm. The signal intensity was
GoldBio, Cat. no. LUCNA-1G) and seeded into 96 well white plates normalized by calculating the ratio of A450/A595 values and plotted
(Corning) at a density of 2 × 105 cells well−1 in a volume of 100 μl. After using the GraphPad Prism 9.5.0 software. The surface expression data
an incubation of 1.5 h at 37 °C and 30 min at room-temperature, in every experiment were normalized with respect to mock-
baseline luminescence readings were recorded. For Gαs-coupling transfection (i.e. pcDNA) treated as 1.
assay, ligands prepared in the assay buffer were added at indicated For the NanoBiT-based G-protein dissociation assay, surface
final concentration after baseline readings while for Gαi-coupling expression of the receptors was measured using flow-cytometry based
assay, 5 μM forskolin (Sigma, Cat. no. F6886) was added to the cells method. Briefly, a small amount of HEK-293 cells from the corre-
and luminescence readings were recorded till they stabilized (5-10 sponding assays were harvested with 0.5 mM EDTA-containing PBS
cycles) followed by ligand addition. The change in luminescence signal and transferred to a 96 well V-bottom plate. Cells were fluorescently
was recorded using a microplate reader (Victor X4; Perkin Elmer) for labeled using anti-FLAG monoclonal antibody (Clone 1E6, FujiFilm
60 min, and data were normalized as indicated in the respective figure Wako Pure Chemicals; 10 μg ml−1 diluted in 2% goat serum + 2 mM
legends and plotted using GraphPad Prism 9.5.0 software. For the EDTA-containing PBS) followed by incubation with Alexa Fluor

Nature Communications | (2023)14:4808 13

Article https://doi.org/10.1038/s41467-023-40482-9

488-conjugated goat anti-mouse IgG secondary antibody (Thermo bands were detected by Western blotting using corresponding anti-
Fisher Scientific; 10 μg ml−1). Subsequently, the cells were washed with bodies (rabbit phospho-ERK1/2 antibody, 1:5,000 dilution; rabbit total
PBS, resuspended in 2 mM EDTA-containing PBS, filtered through a ERK1/2 antibody, 1:5,000 dilution; anti-rabbit HRP-coupled secondary
40 μm filter and the fluorescent intensity of single cells was quantified antibody, Genscript, Cat. No. A00098, 1:10,000 dilution). ECL solution
using a flow cytometer. Fluorescent signal from Alexa Fluor 488 was from Promega (Cat. no. W1015) was used as a substrate for the HRP,
recorded and analyzed using the FlowJo software. Mean fluorescence and the signals were developed using ChemiDoc (BioRad). The signals
intensity from about 20,000 cells per sample were used for analysis. were quantified using densitometry in BioRad Image Lab software,
normalized as indicated in the figure legend, and data were plotted
Tango assay for βarr recruitment using GraphPad Prism 9.5.0 software. For the experiments presented
In order to assess the βarr2 recruitment to indicated receptors, Tango in Fig. 5c, d and Supplementary Fig. 6a, b, cells were pre-treated with
assay was used61. Briefly, HTLA cells were transfected with indicated 10 μM of AMD3100 for 30 min prior to CXCL12 stimulation, and for the
receptor constructs and 24 h post-transfection, cells were trypsinized, experiments presented in Supplementary Fig. 6c, d, cells were pre-
resuspended in complete DMEM, and seeded into 96 well white plates treated with Pertussis toxin (100 ng μl−1) for 12 h during serum
at a density of 1 × 105 cells well−1. After another 24 h, cells were stimu- starvation step.
lated with the indicated dose of ligands and incubated at 37 °C for
additional 7-8 h. Afterwards, the culture media was changed with assay BRET assay for βarr2 conformational change
buffer (1XHBSS, 20 mM HEPES, pH 7.4 and 0.5 mg ml−1 D-luciferin). Intramolecular FlAsH-based BRET sensors were used to monitor the
Luminescence readings were measured in a microplate reader (Victor conformational changes in βarr235,36. Briefly, HEK-293SL cells were
X4; Perkin Elmer), normalized as mentioned in the corresponding seeded at a density of 1.5 × 105 cells well−1 in 6 well plates and trans-
figure legends, and plotted using GraphPad Prism 9.5.0 software. For fected with the indicated receptor constructs along with the βarr2-
the data presented in Figs. 2a, c, and 3a and Supplementary Fig. 2d, FlAsH sensors using calcium phosphate. 24 h post-transfection, cells
PRESTO-Tango constructs were used. PRESTO-Tango constructs har- were detached and seeded into poly-ornithine-coated 96 well white
bors V2R tail at the end of the native receptor followed by a TEV plates at a density of 2.5 × 104 cells well−1. After another 24 h, cells were
cleavage site and tTA transcription factor. For the data presented in washed and incubated with Tyrode’s buffer (140 mM NaCl, 2.7 mM KCl,
Figs. 2b, d and 3b, Tango assay constructs were generated by engi- 1 mM CaCl2, 12 mM NaHCO3, 5.6 mM D-glucose, 0.5 mM MgCl2,
neering a TEV protease cleavage site and tTA transcription factor at the 0.37 mM NaH2PO4, 25 mM HEPES, pH 7.4) for 1 h at room temperature.
end of the receptor coding sequence in pcDNA3.1 vector backbone. Subsequently, FlAsH reagent solution was prepared by mixing 1.75 μl
of FlAsH-EDT2 stock reagent with 3.5 μl of 25 mM EDT solution in
NanoBiT assay for βarr recruitment DMSO and left for 10 min at room temperature. 100 μl of Tyrode’s
Agonist-induced βarr1/2 recruitment for CXCR4 and CXCR7 was also buffer was added to this mixture followed by an additional incubation
measured using NanoBiT-based assay62. Briefly, HEK-293 cells were for 5 min at room temperature and then the volume was adjusted to
transfected with CXCR4 (1 µg) and CXCR7 (7 µg) harboring carboxyl- 5 ml with Tyrode’s buffer. Cells were incubated with 60 μl of the
terminus fusion of SmBiT and βarr1/2 constructs (2 µg) with N-terminal labeling solution for 1 h at 37 °C followed by washing with BAL wash
fusion of LgBiT. The cells were stimulated with varying doses of buffer and Tyrode’s buffer. Finally, 90 μl of Tyrode’s buffer was added
respective ligands followed by measurement of luminescence signal to each well and the plate was incubated at 37 °C for 1 h before ligand
using a multimode plate reader for 10-15 cycles and average data from stimulation. Coelenterazine H was added at a final concentration of
5th to 10th cycle are used for analysis and presentation. In order to 2 μM, cells were stimulated with 100 nM CXCL12 and six consecutive
evaluate the contribution of different GRKs in βarr recruitment to BRET measurements were taken using a Victor X; PerkinElmer plate
CXCR7, we used previously described GRK knock-out cell lines31. reader with a filter set (center wavelength/band width) of 460/25
(donor) and 535/25 (acceptor). BRET ratios (intensity of light emitted
Microfluidic chemotaxis assay by the acceptor/intensity of light emitted by the donor) were calcu-
We quantified chemotaxis using a microfluidic device that tracks lated and net-BRET ratio was determined after subtracting the back-
movement of single cells toward a gradient63 and we used MDA-MB-231 ground BRET ratio i.e. the difference between the FlAsH-EDT2-labeled
human breast cancer cells (purchased from the ATCC, Manassus, VA, BRET ratio and the unlabeled condition. The difference of the net-
USA) stably transduced with CXCR7 fused to GFP64. Briefly, we intro- BRET ratio for ligand-stimulated condition vs. vehicle-treatment was
duced MDA-MB-231 cells stably transduced with CXCR7 fused to GFP plotted using GraphPad Prism 9.5.0 software.
into the device at a concentration of 1 × 105 cells ml−1 in complete
DMEM medium with 10% serum and 1% GlutaMAX. After allowing cells βarr recruitment for the phosphorylation site mutants of CXCR7
to adhere for 10 min, we replaced medium in the seeding port with The phosphorylation site mutants of CXCR7 as indicated in Fig. 8a, b
serum-free DMEM. We added a chemoattractant, 100 ng ml−1 CXCL12 and Fig. 9a–d were generated using Q5 site-directed mutagenesis kit
(R&D Systems) and/or 100 nM VUF11207 (Cayman Chemical) in serum- followed by βarr recruitment in Tango and NanoBiT assays. Surface
free DMEM with 0.1% Probumin (Millipore), to the opposite side of the expression of the indicated mutants was first optimized to be at
device. We quantified chemotaxis of single cells after 16 h in the device. comparable levels followed by the Tango and NanoBiT assays. For the
Tango assay, HTLA cells were transfected with 7 µg of the wild-type and
ERK1/2 MAP kinase phosphorylation assay mutant receptor constructs except the CXCR7T352A+S355A for which, 5 μg
Agonist-induced ERK1/2 MAP kinase phosphorylation was measured DNA was transfected. Cells were treated with indicated concentration
using the Western blot assay65,66. Briefly, HEK-293 cells were trans- of agonists for 8 h at 37 °C followed by the addition of luciferin and
fected with CXCR4 (0.25 µg), CXCR7 (4 µg) or empty vector (pcDNA3.1; luminescence was recorded. Ligand-induced luminescence signal was
7 µg), and 24 h post-transfection, they were seeded into a 6 well plate at normalized with respect to the minimal ligand dose concentration
a density of 1 × 106 cells well−1. Subsequently, the cells were serum taken as 1 and plotted in GraphPad Prism 9.5.0.
starved for 12 h followed by agonist-stimulation (100 nM CXCL12 and
10 µM VUF11207) as indicated in the corresponding figure legends. NanoBiT assay for βarr endosomal localization
Afterwards, the cells were harvested, lysed in 2XSDS loading buffer, Agonist-induced βarr1/2 endosomal localization was monitored using
heated at 95 °C for 15 min followed by centrifugation at 21000 x g for NanoBiT assay. HEK-293 cells were transiently transfected with
15 min. 10 μl of lysate was then separated by SDS-PAGE and ERK1/2 receptor, N-terminal SmBiT-tagged βarr1/2 constructs and N-terminal

Nature Communications | (2023)14:4808 14

Article https://doi.org/10.1038/s41467-023-40482-9

LgBiT-tagged FYVE constructs. The amount of DNA for receptor, βarr1/ 17. Rajagopal, S. et al. Beta-arrestin- but not G protein-mediated sig-
2 and FYVE was kept as 7 µg, 2 µg and 5 µg, respectively. naling by the “decoy” receptor CXCR7. Proc. Natl Acad. Sci. USA
107, 628–632 (2010).
Data quantification and statistical analysis 18. Wijtmans, M. et al. Synthesis, modeling and functional activity of
All the experiments described here were carried out at least three substituted styrene-amides as small-molecule CXCR7 agonists. Eur.
times and data (mean ± SEM) are plotted and analyzed using GraphPad J. Med Chem. 51, 184–192 (2012).
Prism software (9.5.0). The data were normalized with respect to 19. Inoue, A. et al. Illuminating G-protein-coupling selectivity of GPCRs.
proper experimental controls and appropriate statistical analyses were Cell 177, 1933–1947.e25 (2019).
performed as indicated in the corresponding figure legends. 20. Busillo, J. M. et al. Site-specific phosphorylation of CXCR4 is
dynamically regulated by multiple kinases and results in differential
Reporting summary modulation of CXCR4 signaling. J. Biol. Chem. 285, 7805–7817
Further information on research design is available in the Nature (2010).
Portfolio Reporting Summary linked to this article. 21. Kroeze, W. K. et al. PRESTO-Tango as an open-source resource for
interrogation of the druggable human GPCRome. Nat. Struct. Mol.
Data availability Biol. 22, 362–369 (2015).
All the relevant data are included in the manuscript and the Supple- 22. Zhuo, Y., Crecelius, J. M. & Marchese, A. G protein-coupled receptor
mentary Information files. Source data are provided with this paper. kinase phosphorylation of distal C-tail sites specifies beta arrestin1-
Any additional information can be obtained from the corresponding mediated signaling by chemokine receptor CXCR4. J. Biol. Chem.
authors upon request. Source data are provided with this paper. 298, 102351 (2022).
23. Kumari, P. et al. Functional competence of a partially engaged
References GPCR-beta-arrestin complex. Nat. Commun. 7, 13416 (2016).
1. Hughes, C. E. & Nibbs, R. J. B. A guide to chemokines and their 24. Kumari, P. et al. Core engagement with beta-arrestin is dispensable
receptors. FEBS J. 285, 2944–2971 (2018). for agonist-induced vasopressin receptor endocytosis and ERK
2. Griffith, J. W., Sokol, C. L. & Luster, A. D. Chemokines and chemo- activation. Mol. Biol. Cell 28, 1003–1010 (2017).
kine receptors: positioning cells for host defense and immunity. 25. Shiraishi, Y. et al. Biphasic activation of beta-arrestin 1 upon inter-
Annu Rev. Immunol. 32, 659–702 (2014). action with a GPCR revealed by methyl-TROSY NMR. Nat. Commun.
3. Proudfoot, A. E. Chemokine receptors: multifaceted therapeutic 12, 7158 (2021).
targets. Nat. Rev. Immunol. 2, 106–115 (2002). 26. Grimes, J. et al. Plasma membrane preassociation drives beta-
4. Zlotnik, A., Yoshie, O. & Nomiyama, H. The chemokine and che- arrestin coupling to receptors and activation. Cell 186, 2238–2255
mokine receptor superfamilies and their molecular evolution. e20 (2023).
Genome Biol. 7, 243 (2006). 27. Maharana, J., Banerjee, R., Yadav, M. K., Sarma, P. & Shukla, A. K.
5. Steen, A., Larsen, O., Thiele, S. & Rosenkilde, M. M. Biased and g Emerging structural insights into GPCR-beta-arrestin interaction and
protein-independent signaling of chemokine receptors. Front functional outcomes. Curr. Opin. Struct. Biol. 75, 102406 (2022).
Immunol. 5, 277 (2014). 28. Yagi, H. et al. A synthetic biology approach reveals a CXCR4-G13-
6. Zhao, S., Wu, B. & Stevens, R. C. Advancing chemokine GPCR Rho signaling axis driving transendothelial migration of metastatic
structure based drug discovery. Structure 27, 405–408 (2019). breast cancer cells. Sci. Signal 4, ra60 (2011).
7. Pandey, S. et al. Intrinsic bias at non-canonical, beta-arrestin- 29. Song, J. W. et al. Microfluidic endothelium for studying the intra-
coupled seven transmembrane receptors. Mol. Cell 81, 4605–4621 vascular adhesion of metastatic breast cancer cells. PLoS One 4,
e11 (2021). e5756 (2009).
8. Ulvmar, M. H., Hub, E. & Rot, A. Atypical chemokine receptors. Exp. 30. Qian, T. et al. CXCR7 regulates breast tumor metastasis and angio-
Cell Res. 317, 556–568 (2011). genesis in vivo and in vitro. Mol. Med. Rep. 17, 3633–3639 (2018).
9. Nibbs, R. J. & Graham, G. J. Immune regulation by atypical che- 31. Kawakami, K. et al. Heterotrimeric Gq proteins act as a switch for
mokine receptors. Nat. Rev. Immunol. 13, 815–829 (2013). GRK5/6 selectivity underlying beta-arrestin transducer bias. Nat.
10. Graham, G. J., Locati, M., Mantovani, A., Rot, A. & Thelen, M. The Commun. 13, 487 (2022).
biochemistry and biology of the atypical chemokine receptors. 32. Pitcher, J. A. et al. Role of beta gamma subunits of G proteins in
Immunol. Lett. 145, 30–38 (2012). targeting the beta-adrenergic receptor kinase to membrane-bound
11. Cancellieri, C., Vacchini, A., Locati, M., Bonecchi, R. & Borroni, E. M. receptors. Science 257, 1264–1267 (1992).
Atypical chemokine receptors: from silence to sound. Biochem Soc. 33. Malik, R., Soh, U. J., Trejo, J. & Marchese, A. Novel roles for the E3
Trans. 41, 231–236 (2013). ubiquitin ligase atrophin-interacting protein 4 and signal trans-
12. Heuninck, J. et al. Context-dependent signaling of CXC chemokine duction adaptor molecule 1 in G protein-coupled receptor signal-
receptor 4 and atypical chemokine receptor 3. Mol. Pharm. 96, ing. J. Biol. Chem. 287, 9013–9027 (2012).
778–793 (2019). 34. Hitchinson, B. et al. Biased antagonism of CXCR4 avoids antagonist
13. Shi, Y., Riese, D. J. 2nd & Shen, J. The role of the CXCL12/CXCR4/ tolerance. Sci. Signal 11, eaat2214 (2018).
CXCR7 chemokine axis in cancer. Front Pharm. 11, 574667 35. Lee, M. H. et al. The conformational signature of beta-arrestin2
(2020). predicts its trafficking and signalling functions. Nature 531,
14. Naumann, U. et al. CXCR7 functions as a scavenger for CXCL12 and 665–668 (2016).
CXCL11. Plos One 5, e9175 (2010). 36. Nuber, S. et al. beta-Arrestin biosensors reveal a rapid, receptor-
15. Luo, J., Busillo, J. M., Stumm, R. & Benovic, J. L. G protein-coupled dependent activation/deactivation cycle. Nature 531,
receptor kinase 3 and protein kinase C phosphorylate the distal 661–664 (2016).
C-terminal tail of the chemokine receptor CXCR4 and mediate 37. Isaikina, P. et al. A key GPCR phosphorylation motif discovered in
recruitment of beta-arrestin. Mol. Pharm. 91, 554–566 (2017). arrestin2⋅CCR5 phosphopeptide complexes. Mol. Cell 83,
16. Nguyen, H. T. et al. CXCR7: a beta-arrestin-biased receptor that 2108–2121.e7 (2023).
potentiates cell migration and recruits beta-arrestin2 exclusively 38. Maharana, J. et al. Structural snapshots uncover a key phosphor-
through Gbetagamma subunits and GRK2. Cell Biosci. 10, ylation motif in GPCRs driving beta-arrestin activation. Mol. Cell 83,
134 (2020). 2091–2107.e7 (2023).

Nature Communications | (2023)14:4808 15

Article https://doi.org/10.1038/s41467-023-40482-9

39. Min, K. et al. Crystal Structure of beta-Arrestin 2 in Complex with 62. Dwivedi-Agnihotri H, S. P., Deeksha, S., Kawakami, K., Inoue, A. &
CXCR7 Phosphopeptide. Structure 28, 1014–1023.e4 (2020). Shukla, A. K. An intrabody sensor to monitor conformational acti-
40. Shukla, A. K., Xiao, K. & Lefkowitz, R. J. Emerging paradigms of beta- vation of β-arrestins. Methods Cell Biol. 169, 267–278 (2022).
arrestin-dependent seven transmembrane receptor signaling. 63. Chen, Y. C. et al. Single-cell migration chip for chemotaxis-based
Trends Biochem Sci. 36, 457–469 (2011). microfluidic selection of heterogeneous cell populations. Sci. Rep.
41. Violin, J. D. & Lefkowitz, R. J. Beta-arrestin-biased ligands at seven- 5, 9980 (2015).
transmembrane receptors. Trends Pharm. Sci. 28, 416–422 (2007). 64. Luker, K. E., Steele, J. M., Mihalko, L. A., Ray, P. & Luker, G. D.
42. Reiter, E., Ahn, S., Shukla, A. K. & Lefkowitz, R. J. Molecular mechan- Constitutive and chemokine-dependent internalization and recy-
ism of beta-arrestin-biased agonism at seven-transmembrane cling of CXCR7 in breast cancer cells to degrade chemokine
receptors. Annu Rev. Pharm. Toxicol. 52, 179–197 (2012). ligands. Oncogene 29, 4599–4610 (2010).
43. Wisler, J. W., Xiao, K., Thomsen, A. R. & Lefkowitz, R. J. Recent 65. Kumari, P., Dwivedi, H., Baidya, M. & Shukla, A. K. Measuring
developments in biased agonism. Curr. Opin. Cell Biol. 27, agonist-induced ERK MAP kinase phosphorylation for G-protein-
18–24 (2014). coupled receptors. Methods Cell Biol. 149, 141–153
44. Shenoy, S. K. et al. beta-arrestin-dependent, G protein-independent (2019).
ERK1/2 activation by the beta2 adrenergic receptor. J. Biol. Chem. 66. Baidya, M. et al. Allosteric modulation of GPCR-induced beta-
281, 1261–1273 (2006). arrestin trafficking and signaling by a synthetic intrabody. Nat.
45. Eiger, D. S., Boldizsar, N., Honeycutt, C. C., Gardner, J. & Rajagopal, Commun. 13, 4634 (2022).
S. Biased agonism at chemokine receptors. Cell. Signal. 78,
109862 (2021). Acknowledgements
46. Anderson, C. A., Solari, R. & Pease, J. E. Biased agonism at che- Research in A.K.S.’s laboratory is supported by the Senior Fellowship of
mokine receptors: obstacles or opportunities for drug discovery? J. the DBT Wellcome Trust India Alliance (IA/S/20/1/504916) awarded to
Leukoc. Biol. 99, 901–909 (2016). A.K.S., Science and Engineering Research Board (EMR/2017/003804,
47. Bachelerie, F. et al. International Union of Basic and Clinical Phar- SPR/2020/000408, and IPA/2020/000405), Council of Scientific and
macology. [corrected]. LXXXIX. Update on the extended family of Industrial Research [37(1730)/19/EMR-II], Indian Council of Medical
chemokine receptors and introducing a new nomenclature for research (F.NO.52/15/2020/BIO/BMS), Young Scientist Award from Lady
atypical chemokine receptors. Pharm. Rev. 66, 1–79 (2014). Tata Memorial Trust, and IIT Kanpur. H.D.A. was supported by a BioCare
48. Nomiyama, H. & Yoshie, O. Functional roles of evolutionary con- grant from DBT (BT/PR31791/BIC/101/1228/2019). A.I. was funded by the
served motifs and residues in vertebrate chemokine receptors. J. LEAP JP20gm0010004, and the BINDS JP20am0101095 from the Japan
Leukoc. Biol. 97, 39–47 (2015). Agency for Medical Research and Development (AMED); Japan Society
49. Hoffmann, F. et al. Rapid uptake and degradation of CXCL12 for the Promotion of Science (JSPS) KAKENHI grants 21H04791,
depend on CXCR7 carboxyl-terminal serine/threonine residues. J. 21H05113, JPJSBP120213501 and JPJSBP120218801; FOREST Program
Biol. Chem. 287, 28362–28377 (2012). JPMJFR215T and JST Moonshot Research and Development Program
50. Sarma, P., Banerjee, R. & Shukla, A. K. Structural snapshot of a beta- JPMJMS2023 from Japan Science and Technology Agency (JST); The
arrestin-biased receptor. Trends Pharm. Sci. 44, 1–3 (2023). Uehara Memorial Foundation; and Daiichi Sankyo Foundation of Life
51. Yen, Y. C. et al. Structures of atypical chemokine receptor 3 reveal Science. This work was also supported by a grant from the Canadian
the basis for its promiscuity and signaling bias. Sci. Adv. 8, Institutes of Health Research (CIHR) (MOP-74603) to S.A.L., and Y.C. is
eabn8063 (2022). supported by a doctoral training scholarship from the Fonds de
52. Perry-Hauser, N. A. et al. The two non-visual arrestins engage ERK2 recherche santé Québec. The research work in XC’s laboratory is sup-
differently. J. Mol. Biol. 434, 167465 (2022). ported by a grant from the National Science Foundation of China
53. Qu, C. et al. Scaffolding mechanism of arrestin-2 in the cRaf/MEK1/ (21272029). G.D.L. is supported by United States NIH grants
ERK signaling cascade. Proc. Natl Acad. Sci. USA 118, R01CA238023, U24CA237683, R01CA238042, U01CA210152,
e2026491118 (2021). R33CA225549, and R37CA222563. K.E.L. is supported by NIH grant
54. Dwivedi-Agnihotri, H. et al. Distinct phosphorylation sites in a pro- R50CA221807. We thank Manish K. Yadav, Jagannath Maharana and
totypical GPCR differently orchestrate beta-arrestin interaction, Ramanuj Banerjee for helping with manuscript preparation and revision,
trafficking, and signaling. Sci. Adv. 6, eabb8368 (2020). Annu Dalal and Nashra Zaidi for helping with surface expression assays,
55. Zarca, A. et al. Differential involvement of ACKR3 C-Tail in beta- and Manisankar Ganguly for preparing the structural schematic
Arrestin recruitment, trafficking and internalization. Cells 10, in Fig. 6a.
618 (2021).
56. Baidya, M. et al. Key phosphorylation sites in GPCRs orchestrate the Author contributions
contribution of beta-Arrestin 1 in ERK1/2 activation. EMBO Rep. 21, P.S. performed the cAMP assays, βarr recruitment and endosomal
e49886 (2020). localization assays, ERK1/2 phosphorylation assay, site-directed muta-
57. Chen, J. et al. Individual phosphorylation sites at the C-terminus of genesis with help from D.S., S.P., H.D.-A.; C.M.C.C. and K.K. performed
the apelin receptor play different roles in signal transduction. Redox the G-protein dissociation and βarr recruitment in GRK knock-out cells
Biol. 36, 101629 (2020). under the supervision of A.I.; X.R. synthesized VUF11207 under the
58. Barnea, G. et al. The genetic design of signaling cascades to record supervision of X.C., Y.-C.C. carried out the BRET experiment under the
receptor activation. Proc. Natl Acad. Sci. USA 105, 64–69 (2008). supervision of S.A.L.; P.K. performed the Calcium assay with the gui-
59. Kumar, B. A., Kumari, P., Sona, C. & Yadav, P. N. GloSensor assay for dance from P.N.Y.; K.E.L. generated cell lines used for migration
discovery of GPCR-selective ligands. Methods Cell Biol. 142, experiments performed by Y.-C.C. under supervision of G.D.L.; A.K.S.
27–50 (2017). supervised and managed the overall project; all authors contributed to
60. Pandey, S., Roy, D. & Shukla, A. K. Measuring surface expression and data analysis, interpretation and manuscript writing.
endocytosis of GPCRs using whole-cell ELISA. Methods Cell Biol.
149, 131–140 (2019). Competing interests
61. Dogra, S., Sona, C., Kumar, A. & Yadav, P. N. Tango assay for ligand- G.D.L. and K.E.L. receive research funding from InterAx AG administered
induced GPCR-beta-arrestin2 interaction: application in drug dis- through the University of Michigan. All other authors declare no com-
covery. Methods Cell Biol. 132, 233–254 (2016). peting interests.

Nature Communications | (2023)14:4808 16

Article https://doi.org/10.1038/s41467-023-40482-9

Additional information Open Access This article is licensed under a Creative Commons
Supplementary information The online version contains Attribution 4.0 International License, which permits use, sharing,
supplementary material available at adaptation, distribution and reproduction in any medium or format, as
https://doi.org/10.1038/s41467-023-40482-9. long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if
Correspondence and requests for materials should be addressed to changes were made. The images or other third party material in this
Arun K. Shukla. article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not
Peer review information Nature Communications thanks the anon- included in the article’s Creative Commons license and your intended
ymous reviewer(s) for their contribution to the peer review of this work. A use is not permitted by statutory regulation or exceeds the permitted
peer review file is available. use, you will need to obtain permission directly from the copyright
holder. To view a copy of this license, visit http://creativecommons.org/
Reprints and permissions information is available at licenses/by/4.0/.
http://www.nature.com/reprints
© The Author(s) 2023
Publisher’s note Springer Nature remains neutral with regard to jur-
isdictional claims in published maps and institutional affiliations.

Nature Communications | (2023)14:4808 17