CXCR7 Controls Competition for Recruitment of β-Arrestin 2 in Cells Expressing Both CXCR4 and CXCR7

Chemokine CXCL12 promotes growth and metastasis of more than 20 different human cancers, as well as pathogenesis of other common diseases. CXCL12 binds two different receptors, CXCR4 and CXCR7, both of which recruit and signal through the cytosolic adapter protein β-arrestin 2. Differences in CXCL12-dependent recruitment of β-arrestin 2 in cells expressing one or both receptors remain poorly defined. To quantitatively investigate parameters controlling association of β-arrestin 2 with CXCR4 or CXCR7 in cells co-expressing both receptors, we used a systems biology approach combining real-time, multi-spectral luciferase complementation imaging with computational modeling. Cells expressing only CXCR4 maintain low basal association with β-arrestin 2, and CXCL12 induces a rapid, transient increase in this interaction. In contrast, cells expressing only CXCR7 have higher basal association with β-arrestin 2 and exhibit more gradual, prolonged recruitment of β-arrestin 2 in response to CXCL12. We developed and fit a data-driven computational model for association of either CXCR4 or CXCR7 with β-arrestin 2 in cells expressing only one type of receptor. We then experimentally validated model predictions that co-expression of CXCR4 and CXCR7 on the same cell substantially decreases both the magnitude and duration of CXCL12-regulated recruitment of β-arrestin 2 to CXCR4. Co-expression of both receptors on the same cell only minimally alters recruitment of β-arrestin 2 to CXCR7. In silico experiments also identified β-arrestin 2 as a limiting factor in cells expressing both receptors, establishing that CXCR7 wins the “competition” with CXCR4 for CXCL12 and recruitment of β-arrestin 2. These results reveal how competition for β-arrestin 2 controls integrated responses to CXCL12 in cells expressing both CXCR4 and CXCR7. These results advance understanding of normal and pathologic functions of CXCL12, which is critical for developing effective strategies to target these pathways therapeutically.

qRT-PCR. We isolated RNA using Trizol (Life Technologies) and a RNeasy minikit (Qiagen). cDNA was synthesized using random hexamer primers and Reverse Transcription System (Promega). We measured expression of human CXCR4, CXCR7, and GAPDH by qRT-PCR with gene specific PCR primers as described previously. [1] PCR primers are listed in the table at the end of supplemental methods. For each cell line, we determined amounts of CXCR4 or CXCR7 relative to GAPDH as ΔΔCt values (n = 3 each). After 45 cycles of qRT-PCR, we verified sizes of PCR products by gel electrophoresis. For each RNA sample, no PCR product was produced in negative control samples performed without a reverse transcriptase reaction.
Receptor binding assays. We quantified cell surface CXCR4 and CXCR7 on various cell lines using binding of CXCL12 fused to Gaussia luciferase (CXCL12-GL) with minor modifications of our previous protocol. [1] We used parental MDA-MB-231 cells as controls. Briefly, we resuspended 5 x 10 5 cells per condition in 100 µl phenol red free DMEM with 0.2% BSA (Millipore) in microfuge tubes and incubated cells on ice for 30 minutes with increasing concentrations of non-bioluminescent CXCL12. In cells coexpressing CXCR4 and CXCR7 complementation reporters, we added 1 µM AMD3100 or CCX733 (ChemoCentryx) to selectively block CXCR4 or CXCR7, respectively. We added approximately 0.1 ng/mL CXCL12-GL to cells for an additional 30 minutes on ice. We washed cells 3 times with ice-cold PBS before quantifying cell-associated bioluminescence from CXCL12-GL (n = 4). We calculated numbers of CXCL12-GL binding sites per cell for CXCR4 or CXCR7 using GraphPad Prism 5 software. The calculation subtracts binding of CXCL12-GL to control MDA-MB-231 cells, so a value of 0 denotes binding at background levels.

Plasmids.
We used N-terminal and C-terminal fragments of click beetle green and red luciferases (Promega) comprising amino acids 2-413 and 395-542, respectively, for each spectral variant. [2] We designated N-terminal fragments as CBGN and CBRN for click beetle green and red, respectively, which confer spectral characteristics of each luciferase. The common C-terminal fragment, which complements with either Nterminal fragment, is designated CBC. We amplified these fragments by PCR using primers shown below. PCR products were digested with appropriate restriction enzymes and ligated to corresponding sites in our previously described firefly luciferase complementation plasmids for CXCR4, CXCR7, and β-arrestin 2. [3,4] The cloning procedure removed firefly luciferase fragments and replaced them with click beetle fragments.
To enable sorting of transduced cell populations, we modified lentiviral vector FUGW to replace green fluorescent protein (GFP) with either mTagBFP, citrine with a nuclear localization sequence, or FP650. [5][6][7] We generated the vector for citrine with a nuclear localization sequence by replacing AcGFP with citrine in the nuclear localization vector obtained from Clontech. We amplified click beetle complementation constructs by PCR, which transfers a CMV promoter and complementation reporter to the PacI site of the lentiviral vector. We cloned β-arrestin 2-CBC into the vector with FP650. We inserted CBGN fusions for CXCR4 or CXCR7 into a vector with co-expressed mTagBFP, and CBRN fusions were cloned into a vector with nuclear citrine. Amplified products for complementation reporters were confirmed by DNA sequencing. Individual species that contribute to β-arrestin 2 recruitment to CXCR4 in CXCR4 + cells (A), CXCR7 in CXCR7 + cells (B), CXCR4 in CXCR4 + -CXCR7 + cells (C), and CXCR7 in CXCR4 + -CXCR7 + cells (D). For all, CXCL12 = 1000 ng/ml. Species (free receptor bound to probe-labeled arrestin, ligand-bound receptor bound to probe-labeled arrestin, and internalized complexes) are as defined in Table 1.   Endogenous β-arrestin 2 was set to 0 and all kinetic parameters are as described in Table   2. Colors represent the fold change of β-arrestin 2 to CXCR4 or CXCR7; colorbar limits range from 0 to the maximum calculated fold change value. Consistent with Supplemental Figure 5, altering CXCR7 levels affects CXCR4 binding to β-arrestin 2.
*Units for this equation are (nM/s) N !" : Avogadro's number Table B3: CXCR7 + Cells Differential equations describing the change in the concentration of species in CXCR7 + cells over time.
All equations from Tables S4B and S4C apply, except for the following: