Introduction

CXCR3, a seven-transmembrane G-protein-coupled chemokine receptor, is expressed on a variety of cells, including activated effector and memory T cells, Th1 lymphocytes, B cells, NK cells and endothelial cells1, 2, 3. CXCR3 binds to multiple natural ligands, including monokine induced by interferon (IFN)γ (Mig or CXCL9), IFN γ-inducible protein 10 (IP-10 or CXCL10) and IFN-inducible T-cell α chemoattractant (I-TAC or CXCL11). By binding to these ligands, CXCR3 plays a role in many aspects of immune and inflammatory responses, particularly in orchestrating migration and activation of circulating leukocytes4. IP-10, which has a molecular mass of 10 kDa, is a member of the CXC superfamily4. It is expressed in predominance in Th1-mediated pathological inflammatory processes5, 6, 7. Blockage of IP-10 inhibits the recruitment of effector T cells to sites of inflammation following injury8, 9. On the other hand, IP-10 has been shown to inhibit angiogenesis and to have antitumor properties10, 11.

Sulfation of tyrosine in chemokine receptors is emerging as a posttranslational modification that contributes substantially to ligand binding. Tyrosine sulfation tends to occur in acidic regions of molecules, usually in regions containing multiple tyrosines12, 13. Chemokine receptors, including CCR5, CXCR4, CX3CR1 and CCR2b, are sulfated on their N-terminal tyrosines, and tyrosine sulfation is critical for ligand binding14, 15, 16, 17. For example, sulfated tyrosines of the amino terminus of CCR5 contribute to the binding of CCR5 to MIP-1α, MIP-1β, and HIV-gp120/CD4 complexes; in addition, they facilitate the entry of HIV-1 into cells expressing CCR5 and CD414. Sulfated tyrosines also contribute substantially to the formation of docking sites for C5a on the human C5aR18. Our previous work has shown that single sulfation of tyrosine 174 in the second extracellular loop (ECL2) of human C3aR forms a functional docking site for C3a19. Thus, tyrosine sulfation may represent the functional state of many 7-transmembrane segment (7TMS) receptors.

The roles of the extracellular domains of CXCR3 have previously been evaluated by analyzing the binding and function of chimeric CXCR3-CXCR1 receptors21. Some studies have shown that the N-terminal extracellular domains of CXCR3 play a role in the binding of IP-10, and that the second extracellular loop is important for receptor activation20, 21. Like many other chemokine receptors of the immune system, the N-terminus of CXCR3 contains two tyrosines flanked by an acidic amino acid, suggesting that they are candidates for sulfation12, 13.

In the present study, our data show that two tyrosines, 27 and 29, in the N-terminus of CXCR3 are sulfated, and that these two sulfated tyrosines synergistically play a pivotal role in the efficient association of CXCR3 with intact IP-10, as well as a role in calcium mobilization. CXCR3 variants lacking sulfated tyrosines and ECL2 do not efficiently mobilize calcium in response to IP-10. These data not only underscore the functional importance of sulfated tyrosine in the formation of docking sites for ligand binding but also show that the activation site is not located at the termini of G-protein-coupled receptors (GPCRs).

Materials and methods

Cells, plasmids, antibodies and peptides

HEK 293T and Cf2Th cells were obtained from the American Tissue Type Culture Collection (CRL115544 and CRL1430, respectively). Cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, penicillin, and streptomycin. An expression plasmid encoding the human CXCR3 fused with an N-terminal FLAG tag and a C-terminal hemagglutinin (HA) tag was generated by PCR from human lung cDNA, and subcloned into the pcDNA 3.1 expression vector (Invitrogen). CXCR3 variants with an N-terminal truncation (ΔNH2), a deletion of ECL2 (ΔECL2), or one or more tyrosines replaced by phenylalanine were prepared using the PCR-based QuikChange method (Stratagene) using Pfu polymerase. The entire coding sequence of all constructs was checked by sequencing. The anti-FLAG antibody and the anti-HA antibody were purchased from BD Pharmingen (San Diego, CA). IP-10 (PeproTech, Rehovot, Israel) was dissolved in water at 10 μg/μL and diluted into cell suspensions19. The expression plasmids for human endothelin receptors A and B were constructed as described above.

Using previously published methods, constructs for the cDNAs of human tyrosyl protein sulfotransferases (TPST) 1 and 2 were amplified by PCR from U87 human astroglioma cell cDNA and subcloned into pcDNA3.119, 22. Small hairpin RNA (shRNA) constructs targeting nucleotides 259-276 of TPST1 and nucleotides 73-94 of TPST2 were generated and subcloned into pBluescript under the control of the murine U6 promoter19, 23. All these constructs were kindly provided by Dr Hyeryun CHOE, Harvard Medical School, Boston, MA.

Transfection, metabolic labeling and immunoprecipitation of cxcr3, cxcr3 variants, endothelial receptor (etr) a and b

HEK293T cells were transfected with plasmids encoding CXCR3 or the CXCR3 variants using calcium phosphate19. One day later, cells were washed twice with phosphate-buffered saline (PBS), split 1:3, and labeled overnight with [35S]-cysteine and methionine ([35S]-Express, NEN) or [35S]-Na2SO3 (NEN). In some cases, cells were treated with 5 μg/mL tunicamycin (Sigma-Aldrich) to inhibit N-glycosylation 5 h prior to and during labeling14, 19. For immunoprecipitation, labeled cells were harvested and lysed in 1% N-dodecyl-β-D-maltoside (Anatrace) in PBS containing a protease inhibitor cocktail (Sigma-Aldrich and Roche Biochemicals) and 0.2 mmol/L PMSF (Sigma-Aldrich). Cell debris was removed by centrifugation at 18 000×g for 5 min at 4 oC, and the supernatants were immunoprecipitated using an anti-HA antibody covalently cross-linked to protein A Sepharose (Amersham Pharmacia Biotech). Immunoprecipitates were washed twice with 1% N-dodecyl-β-D-maltoside in PBS containing 0.5% SDS and once with PBS, then eluted with SDS sample buffer under reducing conditions by heating at 55 oC for 10 min. Samples were electrophoresed on 12% SDS-Tris/glycine polyacrylamide gels and analyzed by autoradiography.

O- and N-linked carbohydrates were enzymatically removed from immunoprecipitated CXCR3 under non-denaturing conditions using O-glycanase (Glyko) and PNGase F (New England Biolabs), respectively, as recommended by the manufacturers. The reactions were then mixed with SDS sample buffer and analyzed as above on 12% SDS gels19.

Flow cytometer analysis

Two days after transfection, HEK 293T cells were harvested in 5 mmol/L EDTA/PBS and 5 mmol/L MgCl2/PBS and washed with PBS. Cells (5×105) were incubated with 0.5 μg/mL of the anti-FLAG antibody for 30 min, followed by FITC-conjugated goat anti-mouse IgG (Santa Cruz Biotechnology) for 30 min. Finally cells were fixed in 2% paraformaldehyde and analyzed by fluorescence-activated cell sorting (FACS).

Sodium chlorate treatment

Transfected HEK 293T cells were cultured in medium containing 10 mmol/L sodium chlorate for at least 72 h24.

Binding of IP-10 to cells expressing CXCR3 and CXCR3 variant

Binding experiments were performed using HEK 293T cells transfected with wild-type CXCR3 or CXCR3 variants. Two days after transfection, cells were detached with 5 mmol/L EDTA in PBS, washed with DMEM, counted, and resuspended to a concentration of 2×106/mL in binding buffer [20 mmol/L HEPES (pH 7.4), 125 mmol/L NaCl, 1 mmol/L MgCl2, 1 mmol/L CaCl2, 5 mmol/L KCl, 0.5 mmol/L glucose, 0.2% BSA, 0.02% sodium azide]. An aliquot of cells was subjected to FACS analysis to verify that the receptors were expressed at comparable levels. For binding assays, an aliquot of 50 μL was incubated with 0.1 nmol/L [125I]-IP10 (NEN Life Science Products) and 0 to 100 nmol/L of unlabeled IP-10 for 30 min at 37 ºC in a final volume of 100 μL. Cells were centrifuged, washed once with binding buffer, and the bound C3a was determined by γ-counting.

For some experiments the degree of sulfation was modified by co-transfecting the CXCR3 with plasmids encoding TPST1 and 2 in order to increase sulfation, or with shRNA constructs directed against TPST1 and 2 in order to inhibit sulfation.

Calcium influx mediated by the cxcr3 and cxcr3 variants

Cf2Th cells were transfected with CXCR3 or CXCR3 variants using the Lipofectamine method (Invitrogen). One day later, the transfected Cf2Th cells were split into two parts. One aliquot of cells was subjected to flow cytometric analysis to quantify receptor expression as previously described19. The other aliquot was transferred to flasks and grown in DMEM for at least 24 h. Cells were harvested, counted, and incubated with the indicator dye Fura-2-AM (Molecular Probes) for 1 h at 37 ºC in 20 mmol/L HEPES (pH 7.4), 1 mmol/L CaCl2, 1 mmol/L MgCl2, 125 mmol/L NaCl, 5 mmol/L KCl, 0.5 mmol/L glucose, and 0.2% BSA. Cells were washed twice and resuspended in the same buffer to a concentration of 1×106/mL. Changes in intracellular calcium concentration in response to native IP-10 were determined fluorometrically at 37 oC by monitoring the emission at 510 nm and the excitation at 340 and 380 nm as a function of a time. Responses were quantified as the peak of the fluorescence ratio of 340/380 nm. Measurements were made using cells from at least three independent experiments19.

Statistical analysis

To calculate relative Kd values, binding data were analyzed using Prism software (GraphPad) running an algorithm for nonlinear regression of homologous one-site competition binding with ligand depletion25. Data are expressed as mean±SEM. Comparison between multiple groups was performed by ANOVA. A value of P<0.05 was considered a significant difference.

Results

CXCR3 is posttranslationally modified by N-glycosylation and tyrosine sulfation.

Sulfate can be incorporated into proteins at sites of N- or O-linked glycosylation or on tyrosines13. As shown in Figure 1A, immunoprecipitated CXCR3 is present in predominance as a form migrating with an apparent molecular weight of about 45 kDa. Because CXCR3 has two predicted sites of N-linked glycosylation that are also substrates for sulfation12, treating immunoprecipitated CXCR3 with PNGase F produces a mature form of CXCR3 of about 35 kDa. The predicted amino acid sequence includes a number of potential O-linked glycosylation sites in the N-terminal extracellular domain and in ECL2, and when these glycans are removed by treating immunoprecipitated CXCR3 with O-glycanase, the CXCR3 band does not shift on SDS-PAGE (Figure 1B). We conclude that wt CXCR3 is modified by N-linked, but not O-linked, glycosylation.

Figure 1
figure 1

Tyrosines 27 and 29 in the N-terminus of the CXCR3 are sulfated. (A, B) HEK 293T cells were transfected with plasmids encoding wild-type CXCR3 with N- and C-terminal epitope tags. CXCR3 immunoprecipitated from [35S]-cysteine- and methionine-labeled cells was treated with O-glycanase or PNGase F as indicated. (C) HEK 293T cells were transfected with plasmids encoding wild-type CXCR3 or the CXCR3 mutants YDF, FDY, and FDF. After 24 h the transfected cells were divided up and labeled with [35S]-cysteine and methionine or with [35S]-Na2SO3. Both sets of cells were treated with the N-glycosylation inhibitor tunicamycin 5 h before and during labeling. Twenty-four hours after labeling, cells were harvested, lysed and immunoprecipitated with an antibody against the C-terminal HA tag. Immunoprecipitates were analyzed under reducing conditions by electrophoresis on 12% SDS polyacrylamide gels. Gels were dried and exposed to X-ray film. Numbers at the left indicate the positions of molecular weight markers.

Full size image

Two tyrosines in the N-terminus of CXCR3 are adjacent to an acidic amino acid, which is a motif similar to that predicted for tyrosine sulfation. To test whether these tyrosines of CXCR3 are sulfated, we added tunicamycin to the cell medium before and during labeling to remove N-linked glycosylation. As shown in Figure 1C, cells expressing wild-type (wt) CXCR3 labeled with [35S]cysteine and methionine show a single major form of CXCR3 migrating at about 35 kDa (lane 4). As expected, only the mature band of CXCR3 is sulfated (Figure 1C, Lane 8). Figure 1C also shows that a variant of CXCR3 in which tyrosines 27 and 29 in the N-terminus were changed to phenylalanine did not incorporate detectable levels of [35S]-Na2SO3, even after extended exposure (Lane 5). Thus, we infer that tyrosines in the N-terminus of CXCR3 are sulfated and that tyrosine sulfation is independent of N-glycosylation elsewhere in CXCR3.

Sulfated tyrosines 27 and 29 at the N-terminus of CXCR3 are required for the protein to bind IP-10

Posttranslational modifications are known to affect receptor function, so we sought to identify the role of sulfation in the interactions between CXCR3 and IP-10. Sodium chlorate (NaClO3) is a selective nontoxic inhibitor of adenosine triphosphate (ATP) sulfurylase, which is required for formation of phosphoadenosine phosphosulfate (PAPS), the donor for sulfation reactions26. We used NaClO3 to treat cells expressing wt CXCR3.

As shown in Figure 2A, the binding affinity of CXCR3 for IP-10 was substantially lower in HEK 293T cells treated with 10 mmol/L NaClO3 than in the same cells without NaClO3 treatment. The Kd values were 1.4±0.2 nmol/L for untreated cells and 72.3±0.2 nmol/L for NaClO3-treated cells.

Figure 2
figure 2

Sulfated tyrosines 27 and 29 in the N-terminus of CXCR3 are necessary for IP-10 binding. (A) HEK 293T cells were transfected with the plasmid encoding wild-type CXCR3. One day later, cells were divided into two parts. One aliquot was treated with 10 mmol/L NaClO3 for 72 h, while the other aliquot was grown without NaClO3 treatment for 3 days. (B) HEK 293T cells were transfected with plasmids encoding wild-type CXCR3 or the CXCR3 variants ΔNH2 and ΔECL2. (C) HEK 293T cells were transfected with plasmids encoding wild-type CXCR3 or the CXCR3 variants YDF, FDY or FDF. Transfected cells were incubated with 0.1 nmol/L [125I]IP-10 and the indicated concentrations of unlabeled IP-10 as described in Materials and Methods. Cells were washed, and bound [125I]IP-10 was determined by γ-counting. Aliquots of the same transfections were analyzed for receptor expression levels by flow cytometry using the anti-FLAG tag antibody to verify that receptors were expressed at comparable levels. Data are expressed as % maximal binding.

Full size image

We next explored the binding site for IP-10 on CXCR3 in greater detail. As shown in Figure 2B, cells expressing wt CXCR3 showed affinity for IP-10 similar to that of cells expressing the ΔECL2 variant. In contrast, cells expressing ΔNH2 did not show detectable binding to IP-10. These data show that the N-terminus of CXCR3 is necessary for binding IP-10.

To further examine the importance of tyrosines 27 and 29 at the N-terminus for ligand binding, we studied the ability of IP-10 to associate with wt CXCR3 or with variants in which at least one of the two tyrosines in the N-terminus was mutated to phenylalanine. Transfected HEK 293T cells expressed similar levels of wt CXCR3 or variants, as indicated by the fluorescence intensity of anti-FLAG antibody staining. These cells were tested in competition binding experiments with [125I]IP-10 and increasing concentrations of unlabeled IP-10 as indicated. As shown in Figure 2C, the cells expressing CXCR3 mutants, FDY and YDF, showed lower binding affinity to [125I]IP-10 than did cells expressing wt CXCR3. However, only HEK 293T cells expressing the FDF variant failed to bind detectably to [125I]IP-10. These data indicate that sulfated tyrosines 27 and 29 of CXCR3 are important for the association of CXCR3 with IP-10.

Sulfated tyrosines 27 and 29 of CXCR3 are functionally important for calcium mobilization induced by IP-10

The model for ligand-receptor association has been well established in GPCRs, with the docking site at the N-terminus and activation site in the other parts of the receptor, including ECL2 and transmembrane domains18. As shown in Figure 2C, the binding site of IP-10 is at the N-terminus of CXCR3. It is speculated that binding of CXCR3 to IP-10 triggers intracellular signaling events, such as calcium mobilization.

Therefore, we investigated whether the sulfated tyrosines in CXCR3 affect calcium influx using the calcium mobilization assay. Since for most G-protein-coupled receptors, Cf2Th cells appear to express an intact Ca2+ signaling pathway19, we examined the ability of Cf2Th cells transfected with wt CXCR3 or the variants FDY, YDF, and FDF to mobilize calcium in response to IP-10. IP-10 at 200 nmol/L effectively induced effective calcium mobilization in cells expressing wt CXCR3, YDF, or FDY (Figure 3A and B), whereas no such induction was observed in cells expressing the FDF variant (Figure 3C). In a parallel experiment, cells expressing ΔECL2 failed to respond robustly to IP-10 stimulation (Figure 3D). Taken together, these data show that ECL2 domain contains the site of activation that allows CXCR3 to induce signal transduction.

Figure 3
figure 3

The CXCR3 variant FDF does not mobilize calcium in response to IP-10. Cf2Th cells were transfected with wild-type CXCR3 or with the CXCR3 variants YDF, FDF, or ΔECL2, as described in Materials and Methods. Cells were loaded with the calcium indicator dye Fura-2 and assessed for their ability to mobilize calcium in response to IP-10, which was added at the time points indicated by the arrows. Calcium flux was displayed as the fluorescence ratio of 340 nm to 380 nm. Receptor expression was determined by staining with the anti-FLAG-tag antibody. Mean fluorescence levels were 59.3 for wild-type CXCR3, 48.0 for YDF, 58.9 for FDF and 56.2 for ΔECL2; for mock-transfected cells, the value was 7.7. The data shown are representative of three independent experiments.

Full size image

Enzymatic regulation of the sulfation of tyrosines 27 and 29 of CXCR3

One possible explanation for a lack of IP-10 binding to the CXCR3 variant FDF is the absence of sulfate on tyrosines 27 and 29. An alternative explanation is that the expression of phenylalanine at this site alters receptor conformation due to the increase in hydrophobicity resulting from the absence of the hydroxyl group19. We addressed these possibilities by altering the extent of CXCR3 sulfation. In order to increase CXCR3 sulfation, HEK 293T cells were transfected with the plasmid encoding wt CXCR3 alone or in combination with the plasmids encoding TPST1 and 2. As shown in Figure 4A, cells co-transfected with the plasmids encoding wt CXCR3 and TPST 1 and 2 showed higher levels of sulfation (lane 4). In order to decrease CXCR3 sulfation, HEK 293T cells were transfected with the plasmid encoding wt CXCR3 or they were co-transfected with wt CXCR3 and shRNA constructs. Cells co-transfected with the plasmids encoding wt CXCR3 and shRNAs showed lower sulfation signal compared with cells transfected with wt CXCR3 (lane 5). Furthermore, we assayed whether TPST and shRNA affect the ability of cells expressing wt CXCR3 to bind to IP-10. As shown in Figure 4B, while CXCR3 expression levels were quite similar, TPST 1 and 2 increased mean maximal binding by approximately 30%. In contrast, the shRNA constructs decreased mean maximal binding by approximately 30%. Therefore, sulfated tyrosines at the N-terminus of CXCR3, not conformational change in the protein, are required for the protein to associate with IP-10.

Figure 4
figure 4

Tyrosine sulfation of CXCR3 is critical for binding affinity to IP-10. (A) HEK 293T cells were co-transfected with the CXCR3 plasmid alone or with plasmids encoding TPSTs or with shRNAs targeting TPSTs as indicated. Transfections were carried out in the presence of tunicamycin and labeled with [35S]cysteine and methionine or [35S]-Na2SO3. Receptors were immunoprecipitated and analyzed as described for Figure 1. (B) Cells from the same transfection were tested for maximal binding of [125I]IP-10 after verifying by flow cytometric analysis that expression levels of the receptors were comparable. Only transfected cultures showing similar levels of expression were used in the analysis. Results are expressed as mean±SEM of four independent experiments.

Full size image

ETRA and ETRB are not sulfated on tyrosines

We extended our study to examine whether tyrosine sulfation is a universal posttranslational modification that happens in other 7TMS. Unlike chemokine receptors, receptors for endothelin A and B have tyrosines adjacent to basic amino acids at ECL2. As demonstrated in Figure 5, sulfate moiety was not detectably incorporated in either ETRA or ETRB, indicating that ETRA and ETRB are not sulfated posttranslationally (Figure 5).

Figure 5
figure 5

ETRA and ETRB are not tyrosine-sulfated. HEK 293T cells were transfected with plasmids encoding endothelin receptors A and B carrying an HA tag at the C-terminus. One day later, transfected cells were labeled in the presence of tunicamycin as described in Figure 1.

Full size image

Discussion

We have shown here that tyrosines 27 and 29 in the N-terminus of CXCR3 are sulfated, and that these two sulfated tyrosines play an important role in binding IP-10. In this way, the sulfated tyrosines play a role in signal transduction and calcium mobilization in response to IP-10. This observation confirms the two-site model for ligand-receptor binding and activation of CXCR3.

Sulfation of tyrosines is a posttranslational modification occurring late in the trans-Golgi network and found on a wide spectrum of secreted proteins, as well as on extracellular regions of many membrane proteins of multicellular eukaryotes27. It has been estimated that up to 1% of all tyrosines in eukaryotic proteins may be sulfated28. Sulfated tyrosines at the N-termini of chemokine receptors play a critical role in helping these receptors bind their natural ligands. These include CCR5, CXCR4, CCR2b, and CX3CR1 as well as Duffy antigen receptor for binding chemokines14, 15, 16, 17, 29. Previous work has indicated that sulfated tyrosines in the N-terminus of chemotactic C5aR help to form the docking site for C5a, but not the activation site18. In addition, our work on the structure-function of human C3aR has shown evidence for a two-step model for C3aR function, in which the binding site is at tyrosine 174 in ECL2 and the activation site lies at some distance away from the binding site19.

The molecular cloning of CXCR3 places this molecule in the superfamily of G-protein-coupled receptors30, 31, 32. A splice variant of CXCR3 called CXCR3-B has recently been cloned. CXCR3-B has a longer N-terminal extracellular domain containing 52 amino acids, and it lacks a sulfatable tyrosine33. In contrast, CXCR3 has a highly anionic tyrosine-rich N-terminal segment. Tyrosines 27 and 29 lie adjacent to an acidic amino acid, which is essential for tyrosine sulfation, and the present study shows that both of these tyrosines are sulfated (Figure 2C). However, the two tyrosines are sulfated to the different extent. Single tyrosine-to-phenylalanine mutations at position 27 or 29 of CXCR3 cause the loss of sulfation signal seen in wt CXCR3, and these mutations greatly reduce the protein's binding affinity for IP-10. Only a mutant FDF, in which both tyrosines 27 and 29 are changed to phenylalanine, fails to detectably incorporate sulfate. Cells expressing FDF do not efficiently bind IP-10.

Our data support the notion that tyrosine sulfation sites are often clustered and that sulfation of one site may depend on the negatively charged sulfates already present at other sites34. For example, the tyrosines of CCR5 at positions 3–15 are sulfated in a stepwise manner; tyrosines 14 and 15 are sulfated first, followed by tyrosines 10 and 335. In the present study, to verify that the reduced binding of IP-10 by cells expressing FDF reflects the loss of sulfate and not a change in receptor conformation, we attempted to increase or decrease the extent of sulfation by co-transfecting cells with the expression plasmids for CXCR3 and TPSTs, or with CXCR3 and shRNAs targeting TPSTs. If sulfate were important for ligand binding, an increase or decrease in sulfation without a change in the level of receptor expression would lead to increased or decreased ligand binding19. Our data suggest that TPSTs increased tyrosine sulfation and the affinity of the receptor for IP-10. Conversely, when co-transfections were performed with shRNA constructs targeting TPSTs, both sulfation and binding were inhibited, without a significant change in receptor expression. The observation that CXCR3 binding was reduced only partially reflects enzymatic activity of TPSTs synthesized prior to transfection with shRNAs.

Tyrosine sulfation has been identified as a key mediator of protein-protein interactions involved in leukocyte adhesion, hemostasis and chemokine signaling26. Sulfate is a charged and highly polarizable moiety that is likely to contribute free energy to binding. Here we have shown not only that tyrosines 27 and 29 are sulfated but also that these sulfated moieties are critically important for binding native IP-10. Our data that double sulfated tyrosines contribute to ligand binding are inconsistent with the previous report showing that single mutation of tyrosine 27 or 29 of CXCR3 decreases its ability to bind its ligands, including IP-1036. There are several possible explanations for these inconsistencies. First of all, the previous authors report that the molecular weight of human CXCR3 is about 70 kDa, whereas our study shows that the molecular weight of human mature CXCR3 is about 45 kDa, consistent with the predicted amino acid composition. Secondly, tyrosines 27 and 29 are separated by one acidic amino acid, suggesting that both of these two tyrosines should be sulfated12, 35, 37. Thirdly, the previous authors were unable to exclude other factors that may have affected the binding of CXCR3 to IP-10, such as N-linked glycosylation. More importantly, those authors did not determine that the decreased binding affinity was due to a loss of sulfate or a change in receptor conformation. In contrast, we have used TPSTs and shRNAs to determine that sulfated tyrosines, not a conformational change in the receptor, affect the binding of receptor and ligand. Our observations in the present study clearly demonstrate that sulfated tyrosines 27 and 29 are functionally more important than other extracellular domains in binding IP-10.

Posttranslational modification occurs in chemokine and chemotactic receptors, including C3aR, C5aR, CCR5, CX3CR1, and CC2b15, 16, 17, 18, 19. We hypothesized that tyrosine sulfation might be a universal phenomenon in other 7TMS GPCRs, such as ETRA and ETRB38. However, we did not observe the sulfation of tyrosines in ETRA and ETRB. Thus tyrosine sulfation in 7TMS GPCRs should be further examined in order to determine its functional importance for this case of receptor-ligand interaction.

Activation of the G-protein-coupled receptor CXCR3 by IP-10 has been proposed to occur by a “two-step/two-site” mechanism20, 21. Traditionally, chemokine receptors are considered to possess two regions of interaction: a docking/binding domain rich in the acidic amino acid and tyrosine-rich N-terminus confer binding affinity; while an activation site, which includes the ECL2 and transmembrane helices, mediates the biological activity of the ligand18, 19, 39, 40. Our data imply that sulfated tyrosines 27 and 29 in the N-terminus of CXCR3 constitute the binding site of the receptor, which positions IP-10 appropriately for interaction with the activation site.

Many chemokine receptors have been suggested as therapeutic targets for inflammation41. Because the pathogenesis of inflammation involves the interaction of many receptors on various cells4, targeting an individual receptor may not generate a desirable outcome. Many chemokine receptors use sulfated tyrosine(s) to bind their ligands37. Thus, inhibition of tyrosine sulfation on these receptors may serve as a therapeutic target for inflammatory diseases.

Author contribution

Jin-ming GAO designed and performed the experiment and wrote the manuscript. Ruo-lan XIANG, Lei JIANG and Qi-ping FENG performed the experiments. Wen-hui LI designed the primers. Zi-jiang GUO and Qi SUN carried out immunoprecipitation and binding experiments. Zheng-pei ZENG and Fu-de FANG were actively involved throughout the entire study.