Bioluminescent CXCL12 fusion protein for cellular studies of CXCR4 and CXCR7

Introduction

Chemokines and chemokine receptors originally were identified as regulators of immune cell trafficking in normal physiology and inflammation. More recently, this family of proteins has been shown to regulate diseases including cancer, atherosclerosis, autoimmune diseases, and neuro-degenerative processes (1–4). As a result, chemokine receptors have emerged as new therapeutic targets, emphasizing the need to develop reagents for quantitative, high-throughput screening assays of chemokine binding and inhibition.

Conventional assays use radiolabeled chemokines to quantify binding to specific receptors and identify inhibitors of ligand-receptor interactions. While radioligand binding studies have been used successfully for these purposes, such assays generate radioactive waste that is cumbersome to dispose. As alternatives to radioactivity, investigators increasingly are using fluorescence or bioluminescence for cell-based assays in basic research and high-throughput screening. CXCL12 and other chemokines have been labeled with fluorescent dyes or proteins to detect ligand-receptor binding in intact cells (5). For example, Hatse et al. fused CXCL12 to Alexa Fluor 647 and used this reagent to detect binding and inhibition of CXCL12-CXCR4 interactions by flow cytometry (6). Similarly, binding of CXCL12 to receptor CXCR7 has been analyzed with a genetic fusion of CXCL12 to green fluorescent protein (GFP) (7).

While these studies show that fluorescent chemokines provide a viable replacement for radiolabeled chemokine binding assays in some experimental settings, bioluminescence assays have not been developed for chemokine-chemokine receptor binding studies. Relative to fluorescence-based techniques, bioluminescence assays with luciferase enzymes offer several potential advantages for quantifying accumulation of chemokines in intact cells (8). First, bioluminescence assays have substantially lower background signal than fluorescence, enhancing signal-to-noise ratios for chemokine binding. Improved sensitivity is particularly important for measuring ligand binding to chemokine and other seven-transmembrane receptors because these proteins may be expressed at modest levels in cells (9). Second, bioluminescent end points have a greater dynamic range of linear signal response than corresponding fluorescent probes. Finally, bioluminescence assays are affected less than fluorescence by colors and dyes in compounds and cell culture media. These advantages account for improved performance of bioluminescence relative to fluorescence in multiwell cell culture formats and highlight the need for a bioluminescent assay of chemokinechemokine receptor binding (10).

To develop a bioluminescent assay for binding of chemokines to receptors, we focused on chemokine CXCL12 and its two known receptors, CXCR4 and CXCR7. There is particular interest in developing specific inhibitors of CXCL12 binding to these two receptors both as chemical probes for investigating specific mechanisms of action and as potential therapeutic agents for treating diseases including cancer and HIV (11–13). We fused CXCL12 to the humanized form of luciferase from Gaussia princeps. Gaussia luciferase (GL) is the smallest known luciferase, containing only 185 amino acids. In cell-based assays, GL is ∼1000× brighter than firefly and Renilla luciferases (two other enzymes used commonly for bioluminescent assays), which increases the dynamic range for detection and quantification (14). Bioluminescence from GL is also independent of ATP, allowing the enzyme to function in both intracellular and extracellular compartments (15). While our studies focus on CXCL12, we expect this technique can be generalized to develop bioluminescent probes for other chemokines and other peptide signaling molecules.

Materials and methods

Cells

293T cells (Open Biosystems, Huntsville, AL, USA), and human breast cancer cell lines MDA-MB-231 and MCF-7 (ATCC, Manassas, VA, USA) were cultured at 37°C and 5% CO₂ in DMEM (Invitrogen, Carlsbad, CA, USA), 10% fetal bovine serum, 1% glutamine, and 0.1% penicillin/streptomycin/gentamicin (Invitrogen).

DNA constructs

Mouse CXCL12-α was amplified by PCR using plasmid CXCL12-degrakine (courtesy of Lishan Su, University of North Carolina, Chapel Hill, NC, USA) as a template. PCR primers were 5′-ATGCCTCGAGGCCACCATGGACGCCAAGGTCGTCG-3′ and 5′-GCATGAATTCCCCTTGTTTAAAGCTTTCTCCAGGTA-3′; restriction sites for XhoI and EcoRI, respectively, are underlined. The PCR product was ligated to the corresponding sites in plasmid EGFP-N1 (BD Biosciences, San Jose, CA, USA). GL was amplified using PCR primers 5′-ATGCGAATTCCGGCGGAGGTGGGTCCGGAGGCGGTGGGAGCGCCAAGCCCACCGAGAACAACGAAGACTTC-3′ and 5′-GCATGCGGCCGCTTAGCCTATGCCGCCCTGTGCGG-3′ with EcoRI and NotI restriction sites underlined (pGLuc-Basic, New England Biolabs, Ipswich, MA, USA). This strategy removes the 15–amino acid secretion signal from GL and inserts a (G₄S)₂ amino acid linker between CXCL12 and Gaussia (designated as CXCL12-GL). CXCL12-GL and the cytomegalovirus (CMV) promoter from EGFP-N1 were removed with restriction enzymes AseI and NotI and ligated to the blunted PacI site of FUPW (16). Unfused CXCL12 was inserted into the blunted EcoRI site of EGFP-N1 and then transferred to FUPW as described for CXCL12-GL. GL was excised from pGLuc-Basic with EcoRI and NotI and inserted into FUPW as described for CXCL12-GL. PCR products were verified by DNA sequencing.

Human CXCR7 regulated by a human EF-1α promoter was inserted into the blunted EcoRI site of lentiviral vector pSico (courtesy of Tyler Jacks, Massachusetts Institute of Technology, Cambridge, MA, USA) (17,18). Human CXCR4 fused to GFP (CXCR4-GFP) (Peter Hordijk, Landsteiner Laboratorium, Amsterdam, The Netherlands) and a human EF-1α promoter were inserted into pSico at blunted EcoRI and NotI sites, removing unfused GFP from the vector (19).

Lentiviruses and transduction

Replication-incompetent lentiviruses were prepared as described previously (20,21) and used to stably transduce 293T cells with CXCL12-GL, unfused CXCL12, or GL. Lentiviruses based on pSico viruses were used to stably transduce human MDA-MB-231 breast cancer cells with CXCR4-GFP, CXCR7, or GFP control. Resulting cells were designated as 231-CXCR4, 231-CXCR7, or 231-control, respectively.

Flow cytometry

Cell surface expression of CXCR4 and CXCR7 was determined as described previously (22) using monoclonal antibody 12G5 directly conjugated to phycoerythrin (PE) (480-nm excitation/578-nm emission) (BD Biosciences) to CXCR4 and 11G8 MAb (gift of ChemoCentryx, Mountain View, CA, USA) to CXCR7 (22). 11G8 binding was revealed with a goat anti-rabbit secondary antibody conjugated to DyLight 488 (493-nm excitation/518-nm emission) (Jackson ImmunoResearch, West Grove, PA, USA).

CXCL12-GL and GL supernatants

Confluent 293T cells were cultured overnight in DMEM containing 0.2% BSA (Probumin, Celliance, Norcross, GA, USA). Supernatants were centrifuged at 4000× g for 10 min (OmniSpin R; Sorvall, Thermo Fisher Scientific, Sugar Land, TX, USA) to remove cell debris prior to use in experiments.

Emission spectrum

Bioluminescence spectra from CXCL12-GL and GL were measured by diluting supernatants from 293T cells 1:100 in PBS at a final volume of 200 µL. After adding 20 µg coelenterazine substrate for GL (Cat. no. 07372, Fluka, St. Louis, MO, USA), light emission at room temperature was recorded with a PerkinElmer LS50B fluorimeter (model no. S2001; Waltham, MA, USA) using scanning intervals of 0.5 nm. Relative to cell culture experiments, a higher concentration of coelenterazine was used to increase the strength of bioluminescence signal and prevent depletion of substrate during the period of collection.

ELISA

CXCL12 levels in supernatants from 293T cells stably transduced with CXCL12-GL or CXCL12 were quantified by ELISA (Quantikine; R&D Systems, Minneapolis, MN, USA).

CXCL12-GL and GL stability

Effects of pH on bioluminescence from CXCL12-GL and GL were measured in solutions of PBS or 33-mM Tris-HCl with pH ranging 2.9–10.0. We combined 10 µL CXCL12-GL or GL supernatant with 190 µL buffer solution and incubated for 15 min at room temperature before measuring bioluminescence. The same procedure was used to test effects of various concentrations of NaCl solution in water (0–1 M) on CXCL12-GL and GL. To measure thermal stability of CXCL12-GL and GL, 60 µL each supernatant were incubated at 37–65°C in a PCR machine for 60 min. Bioluminescence was measured immediately after incubation, using 10 µL CXCL12-GL or GL combined with 190 µL PBS, pH 7.4.

Bioluminescence was quantified by adding coelenterazine (1 µg/µL solution in methanol) at a final concentration of 1 µg/µmL to solutions of CXCL12-GL or GL aliquoted in black-walled 96-well plates (Cat. no. 07-200-627; Costar, Thermo Fisher Scientific). GL activity was measured immediately after adding coelenterazine using an IVIS 100 system (Caliper, Hopkinton, MA, USA). Imaging was done with 5–20-s acquisition times and high-resolution binning. Data were quantified by region of interest analysis using Living Image 2.6 (Caliper).

Live cell imaging

Cells were plated in black-walled 96-well plates (1 × 10⁴ cells per well) using a Multidrop 384 dispensing system (Labsystems, Waltham, MA, USA), and experiments were performed the subsequent day. Cells were incubated with various dilutions of CXCL12-GL or GL supernatants as described in figure legends. For competition experiments, cells were incubated with CXCL12 (R&D Systems); CXCR4 antagonist AMD3100 (Sigma-Aldrich, St. Louis, MO, USA); or CXCR7 antagonists CCX733 or CCX754 (courtesy of ChemoCentryx) added 30 min before incubating cells with CXCL12-GL for 2 h at 37°C. Cells then were washed twice with PBS before measuring cell-associated bioluminescence. One milligram per milliliter coelenterazine in PBS was added to intact cells, and bioluminescence was measured immediately after adding substrate as described above.

Western blots

231-CXCR4, 231-CXCR7, and 231-GFP control cells were incubated overnight in DMEM medium containing 0.2% BSA. Cells then were treated with supernatants from 293T cells stably expressing CXCL12, CXCL12-GL, or GL for 10 min. 231-CXCR4 cells also were treated for 10 min with various concentrations of synthetic CXCL12 (R&D Systems) added to conditioned medium from 293T cells. Cells were lysed as described previously (16). Lysates were probed for AKT phosphorylated at serine 473 to assess activation of CXCR4 signaling, using a 1:500 dilution of primary antibody (Cell Signaling, Danvars, MA, USA). Primary antibodies were detected with a 1:2000 dilution of secondary anti-rabbit antibody conjugated to horseradish peroxidase (Sigma-Aldrich). Blots were developed with ECL Plus reagent (GE Healthcare Life Sciences, Piscataway, NJ, USA). Membranes then were stripped and re-probed for total levels of AKT using 1:500 and 1:2000 dilutions of primary (Cell Signaling) and secondary antibodies, respectively.

Statistics

Data were plotted as mean values with standard error of the mean (sem). Pairs of data were analyzed by two-tailed paired t-test to determine statistically significant differences (P < 0.05) (GraphPad Prism, La Jolla, CA, USA).

Results

CXCL12-GL fusion protein is secreted and activates CXCR4 signaling

Chemokine CXCL12 binds to CXCR4 and CXCR7, two receptors that are emerging as key regulators of cancer and other disease processes (23–25). To develop a bioluminescent chemokine probe for quantifying binding to CXCR4 and CXCR7, we fused the C-terminus of CXCL12 to the N-terminus of GL, removing the native secretion signal from GL. This design strategy maintains the secretion signal for CXCL12 and an unfused N-terminus of the chemokine, which is critical for binding to CXCR4 (26). Removing the secretion signal from GL also has been reported to increase bioluminescence from this enzyme (15).

To establish that CXCL12-GL is secreted from cells, we stably transduced 293T cells with a lentiviral vector that constitutively produces this protein. After culturing these cells overnight in serum-free medium, we collected supernatants and quantified amounts of CXCL12 and GL activity by ELISA and luminescence, respectively. In three independent preparations of CXCL12-GL, amounts of CXCL12 correlated directly with bioluminescence (Figure 1A). Based on measurements of bioluminescence, as little as 0.05 ng CXCL12 could be detected. These data establish bioluminescence as a sensitive, quantitative assay for levels of CXCL12-GL.

To establish that CXCL12-GL remained biologically active, we directly compared cell signaling in response to supernatants from 293T cells producing CXCL12-GL or unfused CXCL12. We used human MDA-MB-231 breast cancer cells stably transduced with CXCR4-GFP (231-CXCR4), CXCR7 (231-CXCR7) or GFP alone (231-control) for this analysis (Figure 1B). We note that 231-CXCR7 cells express relatively lower amounts of this chemokine on the cell surface than cells transduced with CXCR4, likely due to relatively greater intracellular expression of transduced and endogenous CXCR7 in cells (7,27).

We treated various 231 cell lines with ∼25 ng/mL CXCL12-GL or CXCL12 and determined activation of AKT, a known downstream effector of CXCL12-CXCR4 signaling (28). Relative to unfused CXCL12, cells treated with CXCL12-GL had comparable activation of AKT in 231-CXCR4 cells, as measured by phosphorylation at serine 473 (Figure 1C). In addition, Western blotting showed that treatment with CXCL12-GL produced a similar extent of AKT phosphorylation as that from comparable amounts of synthetic CXCL12. By comparison, neither CXCL12 nor CXCL12-GL altered phosphorylation of AKT in 231-control or 231-CXCR7 cells (data not shown). Results for 231-control and 231-CXCR7 cells are consistent with the lack of CXCR4 expression observed in 231-control cells by flow cytometry and failure of CXCL12-CXCR7 binding to activate AKT in most cell lines (7). Collectively, these studies show that CXCL12 remains functional when fused to GL.

Emission spectrum of CXCL12-GL

To determine effects of CXCL12 on bioluminescence from GL, we measured emission spectra of CXCL12-GL and GL at wavelengths ranging 300–700 nm. The emission spectrum from CXCL12-GL was comparable to that produced by unfused GL with a peak emission of ∼500 nm (Figure 2). The peak wavelength is slightly greater than the 480-nm peak wavelength reported previously for GL (14), potentially due to different assay conditions. Nevertheless, these data establish that fusion of CXCL12 to GL does not alter bioluminescence produced by the enzyme.

Temperature and pH stability of CXCL12-GL relative to GL

To further analyze and validate CXCL12-GL, we investigated effects of temperature, pH, and NaCl concentration on enzymatic activity of CXCL12-GL relative to unfused GL. Both CXCL12-GL and GL retained near peak bioluminescence after incubation for 1 h at temperatures ranging 37–50°C (Figure 3A). At higher temperatures, activities of both proteins decreased substantially, although loss of activity was relatively greater for CXCL12-GL at temperatures up to 65°C (P < 0.05). Reduced stability of CXCL12-GL at elevated temperatures may be due to denaturation of the CXCL12 component of the fusion protein with associated loss of activity of fused GL. Nevertheless, fusion to CXCL12 does not substantially impair thermal stability of GL.

GL is reported to be dependent upon sodium for peak activity (www.nanolight.com/proteins/Luciferase_Data.ppt#23). To quantify effects of CXCL12 on sodium dependence of GL, we incubated CXCL12-GL and GL in concentrations of NaCl ranging 0–1 M. For CXCL12-GL, peak bioluminescence increased from 0 mM to 200 mM NaCl and did not vary significantly at higher concentrations up to 1 M (Figure 3B). Similar to CXCL12-GL, bioluminescence from unfused GL also increased as NaCl concentrations increased to 200 mM. However, bioluminescence from GL decreased significantly relative to CXCL12-GL at ≥400 mM NaCl (P < 0.01).

GL also is reported to maintain activity across a wide range of pH (www.nanolight.com/proteins/Luciferase_Data.ppt#23). To establish effects of CXCL12 on pH stability of GL, we compared bioluminescence from CXCL12-GL and GL in either phosphate- or tris-buffered solutions of various pH. Relative to unfused GL, CXCL12-GL produced greater bioluminescence in acidic phosphate- or tris-buffered solutions (Figure 3, C and D). In phosphate buffer, CXCL12-GL had relatively constant activity from pH ∼4.0 to pH ∼11.0, while bioluminescence from GL dropped significantly at pH 6.0 and below (P < 0.01) (Figure 3C). Neither enzyme had detectable bioluminescence at pH 3.0 in phosphate buffer. GL had slightly greater activity in basic phosphate solutions, although the difference between GL and CXCL12-GL was significant only at pH 9.0 (P < 0.05). Both CXCL12-GL and GL had a narrower range of peak activities in tris-buffered solutions with the highest bioluminescence measured at pH 7.4 (Figure 3D). While differences between CXCL12-GL and GL also were less pronounced in tris solutions, CXCL12-GL had significantly greater activity at < pH 4.0 (P < 0.05 and < 0.01 for pH 4.0 and 2.9, respectively). Enhanced bioluminescence of CXCL12-GL under acidic conditions is likely due to the presence of additional basic amino acids in CXCL12 that buffer enzymatic activity of GL. Greater stability of both CXCL12-GL and GL across a wide range of pH in phosphate versus tris buffers may be the result of monovalent cations, which promote GL activity, in the phosphate buffer.

CXCL12-GL binding to chemokine receptors

To establish that CXCL12-GL can be used to quantify binding to known receptors CXCR4 and CXCR7, we incubated various 231 cell lines with ∼3 ng/mL CXCL12-GL or a comparable amount of GL based on bioluminescence. After 2 h of binding at 37°C, we washed cells with PBS and then measured bioluminescence from cell-associated CXCL12-GL or GL. 231-CXCR7 cells had significantly greater cell-associated CXCL12-GL than the other two cell lines (Figure 4A) (P < 0.01 and P < 0.005 when compared with 231-CXCR4 and 231-control cells, respectively), while accumulation of CXCL12-GL in 231-CXCR4 cells also was significantly higher than 231-control cells (P < 0.01). These data are consistent with CXCL12 having higher affinity for CXCR7 than CXCR4 (22,29), as well as the reported ability of CXCR7 to sequester this chemokine intracellularly (7). Although 231-control cells do not express cell surface CXCR4 or CXCR7 by flow cytometry, these cells do have detectable binding of CXCL12-GL. This binding may be due to the known association of positively-charged CXCL12 with heparin and other negatively-charged glycosaminoglycans present on the surface of cells (30). Binding of unfused GL to all cell types was comparable and barely detectable above background levels of bioluminescence, showing specificity of the assay.

To further validate CXCL12-GL as a probe for detecting chemokine receptors, we used increasing amounts of synthetic CXCL12 as a competitor for CXCL12-GL binding. Synthetic CXCL12 progressively blocked binding of 30 ng/mL CXCL12-GL to 231-CXCR7 cells, decreasing bioluminescence to levels in 231-control cells (Figure 4B). The apparent half maximal effective concentration (EC₅₀) was ∼60 ng/mL, which is higher than the value reported for binding of radioactive CXCL12 to CXCR7 at 4°C (31). This difference in binding may occur because our assay was performed at physiologic temperatures that permit endocytosis of CXCL12 bound to CXCR7 and recycling of receptors, as opposed to assays done at 4°C. Because Gaussia has substantially lower enzymatic activity at 4°C and the imaging instrument cannot be chilled to this temperature, we are unable to reproduce parameters of radiolabeled binding assays for CXCL12-CXCR7. Nevertheless, these data establish that CXCL12-GL can detect expression of CXCR4 and CXCR7 on intact cells under conditions compatible with high-throughput assays.

Pharmacologic inhibition of CXCL12-GL binding in living cells

Quantitative data provided by CXCL12-GL make it amenable as a screening probe for molecules that block receptor binding. As proof of principle for this application of CXCL12-GL, we tested small molecule inhibitors of CXCL12 binding to CXCR4 and CXCR7. We incubated 231-CXCR7 cells with CXCR7-selective inhibitors CCX733 or CCX754 for 30 min before adding ∼5 ng/mL CXCL12-GL for 2 h at 37°C. As a negative control, we treated cells with the CXCR4-specific agent AMD3100. Both CCX733 and CCX754 significantly reduced cell-associated CXCL12-GL in 231-CXCR7 cells with EC₅₀ values of ∼18 and 250 nM, respectively (Figure 5A). Particularly for CCX754, the calculated EC₅₀ value is higher than reported for competition binding assays performed with radiolabeled CXCL12 at 4°C. Similar to results for binding of CXCL12-GL to CXCR7, it is likely that these differences reflect assays performed at physiologic temperature rather than 4°C. By comparison, the CXCR4-selective inhibitor AMD3100 did not significantly affect binding of CXCL12-GL to 231-CXCR7 cells at concentrations up to 1 M, although AMD3100 did selectively block binding of CXCL12-GL to CXCR4 (data not shown).

We also quantified CXCL12-GL accumulation and inhibition in MCF-7 cells, a human breast cancer cell line that normally expresses CXCR7 (Figure 5B). We pretreated MCF-7 cells with increasing concentrations of CCX733 for 30 min before adding ∼5 ng/mL CXCL12-GL. CCX733 reduced accumulation of CXCL12-GL in MCF-7 cells in a dose-dependent manner (Figure 5C), establishing that the assay detects CXCR7 at levels present endogenously. Overall, these data demonstrate that CXCL12-GL can be used to identify molecules that specifically block binding of CXCL12 to CXCR7 relative to CXCR4.

We fused CXCL12 to GL as a bioluminescent probe for CXCR4 and CXCR7. CXCL12-GL is secreted from cells, binds to CXCR4 and CXCR7, and activates CXCR4-dependent signaling to a comparable extent as unfused or synthetic CXCL12. CXCL12-GL also retains bioluminescent properties of unfused GL. GL does not require ATP to produce bioluminescence, so CXCL12-GL can be detected and quantified in both intracellular and extracellular environments. Bioluminescence from CXCL12-GL correlates directly with amounts of chemokine measured by ELISA, providing a quantitative assay for chemokine binding and cellular accumulation. Our studies validate CXCL12-GL for studies of chemokine receptor binding and accumulation in intact cells, incorporating the known sensitivity of bioluminescence into multiwell plate assays for chemokines and chemokine receptors.

As compared with control cells, amounts of cell-associated CXCL12-GL were significantly greater in cells expressing CXCR7 and CXCR4 with CXCR7 promoting greater accumulation of this chemokine. Higher levels of CXCL12-GL in CXCR7-expressing cells may be due to a ∼5-fold greater binding affinity of CXCL12 for CXCR7 relative to CXCR4 (22). In addition, CXCR7 may internalize and sequester CXCL12 from the extracellular environment, as shown qualitatively by Boldajipour et al. using a CXCL12-GFP fusion protein (7). Our studies provide a quantitative assay to investigate differences in CXCL12 binding and accumulation between CXCR4 and CXCR7, identify key amino acids in CXCL12 that regulate receptor binding, and analyze proposed functions of these receptors.

We established that CXCL12-GL can be used to identify and characterize inhibitors of chemokine binding in intact cells. We note that EC₅₀ values for inhibition of CXCL12-GL binding to CXCR7 are higher than those reported previously for this series of CXCR7 inhibitors (22). It is likely that these variations are due to differences in experimental protocols. Values reported previously for CXCR7 inhibitors were from radioligand binding assays performed at 4°C, which measures only cell surface binding (since this temperature blocks endocytosis and trafficking of receptors). By comparison, our assays at 37°C quantify effects of inhibitors on CXCL12-CXCR7 binding under conditions that more closely reproduce in vivo physiology. Despite differences in calculated EC₅₀ values, our studies show that bioluminescent assays with CXCL12-GL can be used to identify selective inhibitors of CXCR7 versus CXCR4.

Our studies demonstrating that both CXCL12 and GL components of the fusion protein retain signaling and enzymatic functions, respectively, are consistent with recent publications showing feasibility of fusing GL to proteins of interest for in vitro and in vivo assays. GL normally is secreted, but the protein may be retained within cells by fusing it to actin (32). Ketteler et al. have used this strategy to develop cell-based assays for processes including apoptosis and autophagy by inter-posing defined amino acid recognition motifs for proteases between GL and actin (32). Cleavage of the GL-actin fusion protein releases GL for secretion, providing a quantitative assay for protease activity. GL also has been fused to a fluorescent protein to allow combined microscopic imaging and bioluminescence quantification of the secretory pathway in living cells, utilizing the endogenous secretion signal of GL to investigate effects of endoplasmic reticulum stress on this pathway (33). Venisnik et al. fused GL to an antibody fragment targeting carcinoembryonic antigen (CEA) (15). The GL-CEA fusion probe retained bioluminescence and antigen-specific targeting, allowing optical imaging of CEA-positive tumor xenografts in mice.

Recent studies have shown that GL and GL fusion proteins may be purified from bacterial or mammalian sources (15), and it also is possible to produce unfused GL using cell-free systems (34). Such techniques could be used if needed to generate more concentrated solutions of CXCL12-GL for cell-based assays. More concentrated solutions of CXCL12-GL may enable in vivo optical imaging of chemokine binding to CXCR4 or CXCR7 in mouse models of cancer or other diseases, although it will be essential to assess effects of injecting CXCL12-GL on the underlying process.

This research establishes that a bioluminescent chemokine-GL fusion protein, CXCL12-GL, can be used in multi-well plate assays to investigate chemokine receptor binding and quantify inhibitory effects of small molecules. This new reagent will enable functional studies of CXCR4 and CXCR7 in cell-based assays and facilitate testing of new compounds targeted to block binding of CXCL12 to one or both receptors. We envision that the strategy of creating bioluminescent chemokines, or lumikines, may become a general method for performing luminescence assays of chemokine-receptor binding. Lumikines are directly compatible with current high-throughput screening technologies, allowing known advantages of bioluminescent reporters to be incorporated into ongoing efforts to probe chemokine receptors and identify new modulators of chemokine-chemokine receptor binding.