Bivalent ligands of CXCR4 with rigid linkers for elucidation of the dimerization state in cells

Article abstract of DOI:10.1021/ja107447w

To date, challenges in the design of bivalent ligands for G protein-coupled receptors (GPCRs) have revealed difficulties stemming from lack of knowledge of the state of oligomerization of the GPCR. The synthetic bivalent ligands with rigid linkers that ar

Full text of DOI:10.1021/ja107447w

Published on Web 10/25/2010
Bivalent Ligands of CXCR4 with Rigid Linkers for Elucidation of the
Dimerization State in Cells
Tomohiro Tanaka, Wataru Nomura,* Tetsuo Narumi, Akemi Masuda, and Hirokazu Tamamura*
Department of Medicinal Chemistry, Institute of Biomaterials and Bioengineering, Tokyo Medical and Dental
UniVersity, 2-3-10 Kandasurugadai, Chiyoda-ku, Tokyo 101-0062, Japan
Received August 18, 2010; E-mail: nomura.mr@tmd.ac.jp; tamamura.mr@tmd.ac.jp
Abstract: To date, challenges in the design of bivalent ligands
for G protein-coupled receptors (GPCRs) have revealed difficul-
ties stemming from lack of knowledge of the state of oligomer-
ization of the GPCR. The synthetic bivalent ligands with rigid
linkers that are presented here can predict the dimer form of
CXCR4 and be applied to molecular probes in cancerous cells.
This “molecular ruler” approach would be useful in elucidating
the details of CXCR4 oligomer formation.
The chemokine receptor CXCR4 is a membrane protein belong-
ing to the family of G protein-coupled receptors (GPCRs). In current
drugs, ∼60% of drug target molecules are located at the cell surface,
and half of them are GPCRs.1 Recent studies have indicated a
pivotal role for homo- and heterooligomerization of CXCR4 in
cancer metastasis, and the significance of oligomeric forms of GPCR
has been gaining acceptance.2 However, the functional implications
proposed for these oligomers, which include signal transduction
and internalization, are poorly understood and require additional
studies.3 Efforts to understand those correlations have used
photochemical analyses such as bioluminescence resonance energy
transfer (BRET) analysis,3,4 but the elucidation of the native state
of CXCR4 in living cells is complicated by conformational or
functional changes resulting from mutations. Estimates of the
precise distance between ligand binding sites in the dimer form
would permit the development of bivalent ligands of CXCR4 having
improved binding affinity and specificity.5 In spite of the enormous
effort devoted to the design of bivalent ligands, rational design of
such linkers has been difficult because of the lack of knowledge
concerning the dimeric form of GPCRs. Therefore, there is an
increasing demand for a novel strategy for the analysis of the precise
distance between ligand binding sites.6
In this study, we designed and synthesized novel CXCR4 bivalent
ligands consisting of two molecules of an FC1317 analogue, [cyclo(-
D-Tyr-Arg-Arg-Nal-D-Cys-)] [Nal ) L-3-(2-naphthyl)alanine, 1a],
connected by a poly(L-proline) or a PEGylated poly(L-proline)
linker. Poly(L-prolines) have been utilized as rigid linkers between
the two functional units, which require a predetermined separation
for activity.8 Linkers consisting of poly(L-prolines) were expected
to maintain constant distances of 2-8 nm between the ligands. Our
bivalent ligands with linkers of various lengths were used to
determine the distance between two binding sites of ligands
consisting of CXCR4 dimers. Acetamide-capped FC131 (1b), in
which Gly is replaced by D-Cys and the thiol group of Cys is capped
with an acetamide group, was synthesized as a monomer unit of
the ligand (Figure 1). Although this substitution caused a 2-fold
decrease in binding to CXCR4, the binding affinity was still
adequate for analyses. Poly(L-proline) helices are known to maintain
a length of 0.9 nm per turn.9 In this study, polyproline- and
Figure 1. Design of bivalent ligands against chemokine receptor CXCR4.
As CXCR4 binding moieties, D-Cys FC131 (R ) CH2SH, 1a) and
acetamide-capped FC131 (R ) CH2SCH2CONH2, 1b) were prepared.
Poly(L-proline) (2a-h) and PEG-conjugated poly(L-proline) (3a-f) with
CXCR4 binding moieties on both ends were synthesized. As monomer
binding ligands with linkers, Ac6pro FC131 (4) and AcPEG FC131 (5)
were synthesized. Tetramethylrhodamine (TAMRA)-labeled 2e (6) and 4
(7) were prepared for the imaging experiments.
PEGylated polyproline-type linkers with lengths of 2-8 nm were
synthesized.10 The synthetic linkers and their conjugated bivalent
ligands were characterized by high-resolution mass spectrometry
(HRMS) (Tables S3 and S5 in the Supporting Information), and
their CD spectra clearly showed the presence of a type-II polypro-
line helix (Figure S3 in the Supporting Information). As monomer
controls, FC131 analogues 4 and 5 with hexaproline and poly(eth-
ylene glycol) (PEG) linkers, respectively, that were acetylated at
the other end were also prepared.
The binding affinities of the synthetic ligands were evaluated in
a competitive binding assay against [125I]-SDF-1R, as reported
previously.7d The binding assay showed that the binding affinity
of our bivalent ligands is clearly dependent on the linker length.
Ligands of the poly(L-proline) type with the highest affinities were
2e and 2f. Among the PEGylated poly(L-proline)-type ligands, 3c
and 3d showed the highest affinity. The linker-optimized bivalent
ligands, 2f and 3d, showed 7.3- and 21-fold increases in binding
affinity relative to 4 and 5, respectively (Table 1). These results
9
10.1021/ja107447w  2010 American Chemical Society
J. AM. CHEM. SOC. 2010, 132, 15899–15901 15899

C O M M U N I C A T I O N S
Table 1. Summary of Binding Affinities of Synthetic Bivalent and
Monovalent Ligands Analyzed by [125I]-SDF-1R Competition Assay
compd
Ki (nM)a
linker length (nm)
compd
Ki (nM)a
linker length (nm)
FC131
1b
2a
2b
2c
2d
2e
2f
2g
31.5
53.4
51.2
45.4
64.4
59.5
13.2
9.9
-
3a
3b
3c
3d
3e
3f
4
87.2
45.6
17.8
13.9
49.3
83.3
72
3.8
4.7
5.6
6.5
7.4
8.3
-
-
1.8
2.7
3.6
4.5
5.4
6
5
7
294
119
-
22.5
45.8
6.9
8.1
-
2h
a Ki values are the concentrations corresponding to 50% inhibition of
125I]-SDF-1R binding to Jurkat cells.
Figure 2. Binding of TAMRA-labeled FC131-derived monovalent and
bivalent ligands to EGFP-CXCR4-transfected HeLa cells. Bivalent ligand
6 with an optimized linker length was utilized. The pictures show the binding
of (A) 6 (25 nM) and (B) 7 (50 nM). Each panel is divided into four sections
as follows: upper left, EGFP emission; upper right, differential interference
contrast (DIC) image; lower left, TAMRA emission conjugated to ligands;
lower right, merged image. Orange bars in the panels represent 50 µm.
[
indicate successful bivalent binding of the ligands, which has been
known to be responsible for an increase in binding affinity.5a It
should be noted that the maximum increase in binding affinity was
observed for ligands of the two linker types having similar lengths
(5.5-6.5 nm). In the dimer state of CXCR4, there are several forms
of assembly (head-to-head, tail-to-tail, and head-to-tail).5a These
forms have different distances between the binding sites of the
ligands. Molecular modeling studies of FC131 with CXCR4
suggested that amino acids in transmembrane (TM) 7 are important
for FC131 binding.11 Through the use of the rhodopsin structure,
it was revealed that in the TM 4 and 5 assembly form, the linear
distance between ligand binding sites is 5.3 nm. In the other forms
of possible assembly, the linear distances were determined to be
3.5 and 3.9 nm for TM 1 and 2 assembly and the combination of
TM 1-4 and TM 2-5 assembly, respectively (Figure S4). The
changes in binding affinity were relatively moderate, and although
the existence of different assembly forms is possible, a majority of
the population should be in the TM 4 and 5 assembly form.
Figure 3. FACS analysis to evaluate the dependence of ligand binding on
the levels of CXCR4 expression. The columns show the binding of (A) 6
and (B) 7 to Jurkat (white), HeLa (gray), and K562 (black) cells. The fold
increase values were calculated by dividing the MFIs of the above cells in
the presence of ligands by the corresponding values in the absence of ligands.
The results are means of three independent experiments; error bars indicate
standard errors of the mean.
From the increased binding affinity of linker-optimized bivalent
ligands, a hypothesis was derived that such ligands could be applied
as probes specific to CXCR4 on the cell surface because the
receptors are overexpressed in several kinds of malignant cells12
and that the dimer formation of the receptor should depend on the
expression level. Accordingly, compound 2e, which showed high
binding affinity, was chosen for labeling with tetramethylrhodamine
(TAMRA) and applied to the imaging of CXCR4. The TAMRA
moiety was conjugated to an N-teminal of the proline linker via
γ-butyric acid. To confirm that the ligands specifically bind to
CXCR4, a CXCR4-EGFP fusion protein (EGFP ) enhanced green
fluorescent protein) was transiently expressed in HeLa cells. The
increase in binding affinity of the bivalent ligand was clearly
reflected in the imaging of CXCR4, as a merged image of TAMRA-
labeled 2e (6) and EGFP-fused CXCR4 was observed (Figure 2).
When a control monomer, TAMRA-labeled 4 (7), was utilized for
detection, only a trace of binding was observed. Additionally,
binding to mock HeLa cells at the same concentration of ligands
was not observed for either ligand (Figure S5).
of ligand 6 at 1 µM, although no significant increase in MFI was
observed at 25 or 250 nM 6, which corresponds with the imaging
experiment (Figure S5). Meanwhile, the monovalent ligand 7 at 2
µM showed similar binding to Jurkat and HeLa cells, involving
1.1- and 1.4-fold increases in MFI, respectively. These results
suggest that it is difficult to distinguish the expression level of
CXCR4 by molecular imaging using the monovalent ligand. On
the other hand, it is of special interest that the bivalent ligand
showed distinguishability of the differences in CXCR4 expression
levels. Furthermore, the binding of our CXCR4 ligands would be
responsive to CXCR4, as no binding of either ligand to K562 cells,
which express a trace of CXCR4, was observed. These results
provide evidence in support of the hypothesis that the bivalent ligand
binds preferentially to the constitutive dimer of CXCR4. Molecular
imaging of CXCR4 on the cell surface by specific antibodies, such
as c835214 or the monomer ligand T140,15 has been previously
reported. In the present system, however, it is possible that the
bivalent ligands could distinguish the density of CXCR4 on the
cell surface.
To further assess whether our bivalent ligand could distinguish
between cancerous and normal cells by the imaging method, A549
and Human Umbilical Vein Endothelial Cells (HUVEC) were
employed for staining as adhesive cell lines. A549 cells are human
lung adenocarcinomic human alveolar basal epithelial cells, which
are known to possess high CXCR4 expression levels.16 HUVEC
were chosen as a normal cell line without CXCR4 expression. It
has been reported that the expression of CXCR4 on HUVEC is
induced by fibroblast growth factor (FGF), which is highly
expressed in the embryonic stage.17 Thus, HUVEC was cultured
To further evaluate the binding specificity and dependence on
CXCR4 expression levels, fluorescence-activated cell sorting
(FACS) analyses utilizing Jurkat, K562, and HeLa cells were
performed (Figure 3). The cells were adopted on the basis of their
different levels of CXCR4 expression (Jurkat > HeLa > K562).13
The binding was evaluated by changes in mean fluorescence
intensity (MFI) of the above cells in the presence and absence of
ligands. The bivalent ligand 6 showed intense binding to Jurkat
cells, which highly express CXCR4, as evidenced by the 2.3- and
3.3-fold increases in MFI at 25 and 250 nM, respectively. For
binding to HeLa cells, the MFI was increased 2.4-fold by binding
9
15900 J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010

C O M M U N I C A T I O N S
Acknowledgment. The authors thank Prof. Kazunari Akiyoshi
(Tokyo Medical and Dental University) for access to the laser
scanning microscope. T.T. was supported by JSPS Research
Fellowships for Young Scientists. This research was supported in
part by New Energy and Industrial Technology Development
Organization (NEDO).
Supporting Information Available: Curve-fitting data for the
binding analyses, CD spectra, docking study of bivalent ligand binding,
imaging analyses of mock cells, histogram and MFI of FACS analysis,
imaging analyses of HeLa cells cultured with FGF, experimental
procedures, and spectral and analytical data for all new compounds.
This material is available free of charge via the Internet at http://
pubs.acs.org.
References
(1) Overington, J. P.; Al-Lazikani, B.; Hopkins, A. L. Nat. ReV. Drug. DiscoVery
2006, 5, 993–996.
(2) Wang, J.; He, L.; Combs, C. A.; Roderiquez, G.; Norcross, M. A. Mol.
Cancer Ther. 2006, 5, 2474–2483.
(3) Percherancier, Y.; Berchiche, Y. A.; Slight, I.; Volkmer-Engert, R.;
Tamamura, H.; Fujii, N.; Bouvier, M.; Heveker, N. J. Biol. Chem. 2005,
280, 9895–9903.
(4) (a) Babcock, G. J.; Farzan, M.; Sodroski, J. J. Biol. Chem. 2003, 278, 3378–
3385. (b) Luker, K. E.; Gupta, M.; Luker, G. D. FASEB J. 2009, 23, 823–
834.
(5) (a) Handl, H. L.; Sankaranarayanan, R.; Josan, J. S.; Vagner, J.; Mash,
E. A.; Gillies, R. J.; Hruby, V. J. Bioconjugate Chem. 2007, 18, 1101–
1109. (b) Zheng, Y.; Akgu¨n, E.; Harikumar, K. G.; Hopson, J.; Powers,
M. D.; Lunzer, M. M.; Miller, L. J.; Portoghese, P. S. J. Med. Chem. 2009,
52, 247–258.
Figure 4. Imaging of CXCR4 by TAMRA-labeled FC131-derived monov-
alent and bivalent ligands on cancerous and normal primary cells. The panels
show the binding of (A) 6 (1 µM) and (b) 7 (2 µM) to A549 cells and (C)
6 (250 nM) and (D) 7 (500 nM) to HUVEC cells. Each panel is divided
into four sections as follows: upper left, TAMRA emission image; upper
right; DIC image; lower left, merged image; lower right, focused image.
Orange bars in the panels represent 50 µm.
(6) Panetta, R.; Greenwood, M. T. Drug DiscoVery Today 2008, 13, 1059–
1066.
(7) (a) Tamamura, H.; Xu, Y.; Hattori, T.; Zhang, X.; Arakaki, R.; Kanbara,
K.; Omagari, A.; Otaka, A.; Ibuka, T.; Yamamoto, N.; Nakashima, H.;
Fujii, N. Biochem. Biophys. Res. Commun. 1998, 253, 877–882. (b) Fujii,
N.; Oishi, S.; Hiramatsu, K.; Araki, T.; Ueda, S.; Tamamura, H.; Otaka,
A.; Kusano, S.; Terakubo, S.; Nakashima, H.; Broach, J. A.; Trent, J. O.;
Wang, Z.; Peiper, S. C. Angew. Chem., Int. Ed. 2003, 42, 3251–3253. (c)
Tamamura, H.; Araki, T.; Ueda, S.; Wang, Z.; Oishi, S.; Esaka, A.; Trent,
J. O.; Nakashima, H.; Yamamoto, N.; Peiper, S. C.; Otaka, A.; Fujii, N.
J. Med. Chem. 2005, 48, 3280–3289. (d) Tanaka, T.; Tsutsumi, H.; Nomura,
W.; Tanabe, Y.; Ohashi, N.; Esaka, A.; Ochiai, C.; Sato, J.; Itotani, K.;
Murakami, T.; Ohba, K.; Yamamoto, N.; Fujii, N.; Tamamura, H. Org.
Biomol. Chem. 2008, 6, 4374–4377.
in the absence of FGF. Ligand 6 showed clear binding to A549
cells (Figure 4A) but not to HUVEC (Figure 4C) at concentrations
of 1 µM and 250 nM, respectively. On the other hand, monomer
ligand 7 showed a trace of binding to each cell line (Figure 4B,D).
Bivalent ligand 6 showed binding to HUVEC cultured with FGF
at 250 nM (Figure S7). Thus, the bivalent ligands can detect
cancerous cells that are in a state of high CXCR4 expression in a
specific manner.
(8) (a) Arora, P. S.; Ansari, A. Z.; Best, T. P.; Ptashne, M.; Dervan, P. B.
J. Am. Chem. Soc. 2002, 124, 13067–13071. (b) Sato, S.; Kwon, Y.;
Kamisuki, S.; Srivastava, N.; Mao, Q.; Kawazoe, Y.; Uesugi, M. J. Am.
Chem. Soc. 2007, 129, 873–880.
(9) (a) Kuemin, M.; Schweizer, S.; Ochsenfeld, C.; Wennemers, H. J. Am.
Chem. Soc. 2009, 131, 15474–15482. (b) Schuler, B.; Lipman, E. A.;
Steinbach, P. J.; Kumke, M.; Eaton, W. A. Proc. Natl. Acad. Sci. U.S.A.
2005, 102, 2754–2759.
(10) For synthesis details, see the Supporting Information.
(11) Va˚benø, J.; Nikiforovich, G. V.; Marshall, G. R. Chem. Biol. Drug Des.
2006, 67, 346–354.
(12) Balkwill, F. Semin. Cancer Biol. 2004, 14, 171–178.
(13) (a) Carnec, X.; Quan, L.; Olson, W. C.; Hazan, U.; Dragic, T. J. Virol.
2005, 79, 1930–1933. (b) Majka, M.; Rozmyslowicz, T.; Honczarenko,
M.; Ratajczak, J.; Wasik, M. A.; Gaulton, G. N.; Ratajczak, M. Z. Leukemia
2000, 14, 1821–1832.
(14) Schwartz, V.; Lue, H.; Kraemer, S.; Korbiel, J.; Krohn, R.; Ohl, K.; Bucala,
R.; Weber, C.; Bernhagen, J. FEBS Lett. 2009, 583, 2749–2757.
(15) Nomura, W.; Tanabe, Y.; Tsutsumi, H.; Tanaka, T.; Ohba, K.; Yamamoto,
N.; Tamamura, H. Bioconjugate Chem. 2008, 19, 1917–1920.
(16) Murdoch, C.; Monk, P. N.; Finn, A. Immunology 1999, 98, 36–41.
(17) Salcedo, R.; Wasserman, K.; Young, H. A.; Grimm, M. C.; Howard,
O. M. Z.; Anver, M. R.; Kleinman, H. K.; Murphy, W. J.; Oppenheim,
J. J. Am. J. Pathol. 1999, 154, 1125–1135.
(18) Sohy, D.; Parmentier, M.; Springael, J. J. Biol. Chem. 2007, 282, 30062–
30069.
(19) Contento, R. L.; Molon, B.; Boularan, C.; Pozzan, T.; Manes, S.; Marullo,
S.; Viola, A. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 10101–10106.
(20) Pello, O. M.; Mart´ınez-Mun˜oz, L.; Parrillas, V.; Serrano, A.; Rodr´ıguez-
Frade, J. M.; Toro, M. J.; Lucas, P.; Monterrubiol, M.; Mart´ınez-A, C.;
Mellado, M. Eur. J. Immunol. 2008, 38, 537–549.
In summary, we have presented experimental results concerning
the elucidation of the native state of the CXCR4 dimer utilizing
bivalent ligands. These lead to a more precise understanding of
the oligomerization state. Such a “molecular ruler” approach could
be utilized in the design of bivalent ligands for any GPCR. It has
been suggested that several GPCRs also exist as heterodimer forms,
and CXCR4 has been hypothesized to form heterodimers with
CCR2,18 CCR5,19 CXCR7,4b and the δ-opioid receptor.20 Although
the biological significance of GPCRs in homo- or heterooligomer-
ization is still unclear and controversial, the approach described
here involving rigid linkers conjugated to ligands specific to each
GPCR would lead to elucidation of these issues. Furthermore,
through the avidity shown as the specific binding affinity for the
dimeric form of CXCR4, the fluorescent-labeled bivalent ligands
have been shown to be powerful tools for cancer diagnosis on the
basis of their ability to distinguish the density of CXCR4 on the
cell surface. Our approach has the advantages that the ligand can
directly capture dimeric forms of GPCRs and that the linkers can
be applied to virtually any known GPCR.
JA107447W
9
J. AM. CHEM. SOC. VOL. 132, NO. 45, 2010 15901