|Home | About | Journals | Submit | Contact Us | Français|
We previously demonstrated that the naturally occurring splice variant stromal cell-derived factor 1γ/CXCL12γ is the most potent CXCL12 isoform in blocking X4 HIV-1, with weak chemotactic activity. A conserved BBXB domain (B for basic and X for any residue) located in the N terminus (24KHLK27) is found in all six isoforms of CXCL12. To determine whether the potent antiviral activity of CXCL12γ is due to the presence of the extra C-terminal BBXB domains, we mutated each domain individually as well as in combination. Although binding of CXCL12γ to heparan sulfate proteoglycan (HSPG) was 10-fold higher than that observed with CXCL12α, the results did not demonstrate a direct correlation between HSPG binding and the potent antiviral activity. CXCL12γ mutants lacking the conserved BBXB domain (designated γB1) showed increased binding to HSPG but reduced anti-HIV activity. In contrast, the mutants lacking the C-terminal second and/or third BBXB domain but retaining the conserved domain (designated B2, B3, and B23) showed decreased binding to HSPG but increased anti-HIV activity. The B2, B3, and B23 mutants were associated with enhanced CXCR4 binding, receptor internalization, and restored chemotaxis. Internalization of CXCR4 was more potent with CXCL12γ than with CXCL12α and was significantly reduced when the conserved BBXB domain was mutated. We concluded that the observed potent anti-HIV-1 activity of CXCL12γ is due to increased affinity for CXCR4 and to efficient receptor internalization.
Chemokines are small, structurally related chemoattractant cytokines characterized by conserved cysteine residues. Based on the positions of the first N-terminal cysteines, chemokines fall into four subfamilies. The CC and CXC subfamilies have been well characterized. The CC subfamily includes the following: regulated on activation, normal T-cell expressed and secreted (RANTES), monocyte chemoattractant protein 1 (MCP-1), and macrophage inflammatory peptides 1 (MIP-1). The prototype of the CXC subfamily is interleukin-8 (IL-8)/CXCL8. The C chemokine (lymphotactine) and CX3C chemokine (fractalkine) subfamilies were recently identified (reviewed in reference 30). The physiological activities of chemokines are mediated by the selective recognition and activation of chemokine receptors belonging to the seven-membrane-domain G-protein-coupled receptor superfamily (GPCRs). In addition, chemokines also bind to glycosaminoglycans (GAGs) through distinct binding sites. Chemokine binding to GAGs on cells, particularly endothelial cells, results in chemotactic chemokine gradients that allow correct presentation of chemokines to leukocytes, therefore enabling target cells to cross the endothelial barrier and migrate into tissues (reviewed in reference 10).
Stromal cell-derived factor 1 (SDF-1)/CXCL12 is a member of the CXC chemokine family and is a key regulator of B-cell lymphopoiesis, hematopoietic stem cell mobilization, and leukocyte migration (reviewed in reference 10). CXCL12 was originally thought to mediate these processes through the single receptor CXCR4 (9). However, later studies demonstrated that RDC-1/CXCR7 is also a receptor for CXCL12 (6, 11). CXCL12 has also been shown to block HIV-1 infection (5). There are two known human splice variants of CXCL12, referred to as CXCL12α and CXCL12β (27). The genomic structure of the CXCL12 gene revealed that human CXCL12α and CXCL12β are encoded by a single gene and arise by alternative splicing. The cDNAs corresponding to CXCL12α and CXCL12β encode proteins of 89 and 93 amino acids, respectively. A third splice variant, classified as CXCL12γ, has been identified in rats (14). The human equivalent of CXCL12γ was recently identified among other splice variants of CXCL12 (33). The novel human splice variants CXCL12γ, CXCL12, CXCL12δ, and CXCL12θ (also reported as CXCL12ϕ ) are expressed through alternative splicing events that result in different exons being added to the same first three exons. Therefore, all six splice variants of CXCL12 are identical in the first 88 amino acid residues from the amino terminus.
It has been demonstrated that CXCL12α and -β are expressed in numerous tissues, with the highest expression levels in the liver, pancreas, and spleen (33). The mRNA encoding CXCL12γ was detectable in the adult human heart but hardly detectable in several other tissues. On the other hand, CXCL12δ, -, and -θ could be detected in several human adult and fetal tissues, with the pancreas expressing the highest levels (33). Recent studies have demonstrated the tissue expression of CXCL12γ in the adult heart (24). We previously demonstrated that CXCL12γ is the most potent anti-HIV-1 inhibitor, with the weakest chemotactic activity and no detectable enhancing activity for hematopoietic progenitor cell survival or replating capacity (2). The first three exons present in the CXCL12γ splice variant are identical to those found in CXCL12α and CXCL12β. The fourth exon, however, contains a large number of basic residues that result in at least four additional BBXB domains in addition to the conserved 24KHLK27 domain (33). It is not known whether the additional BBXB domains in the C terminus of CXCL12γ result in higher affinity for heparan sulfate proteoglycan (HSPG) and whether differences in HSPG binding could explain the observed anti-HIV-1 potency or the low chemotactic activity.
The BBXB motif on RANTES has been suggested as the principal site for high-affinity binding to heparan sulfate. This binding controls receptor selectivity (22). It was previously demonstrated that a mutation of CXCL12α in the 24KHLK27 domain reduces the antiviral activity at least 50 percent without affecting the chemotactic activity (4, 29). It was proposed that chemokine binding to HSPG might concentrate the chemokine near the CXCR4 receptor or form a haptotactic chemokine gradient.
In this study, we analyzed the mechanism of the potent antiviral activity of CXCL12γ. We examined the role of the additional BBXB domains of CXCL12γ in the observed biological activities of CXCL12γ. Mutations in CXCL12γ were introduced to knock out the BBXB domains either individually or in combination. We analyzed receptor internalization and binding affinities of the mutant chemokines for CXCR4 and HSPG. The results demonstrate that the potent anti-HIV activity of CXCL12γ is due to its efficient binding and internalization of CXCR4. The results provide important insight into the structure-function relationship of CXCL12γ and suggest that determinants other than the BBXB domains are involved in the observed biological activities of CXCL12γ.
All cell lines were obtained from the American Type Culture Collection (Rockville, MD). The HeLa, TZM-bl, LM(tk−), and Sog9 cell lines were maintained in Dulbecco modified Eagle medium (DMEM; Quality Biologicals, Gaithersburg, MD) containing 10% fetal bovine serum (FBS) and antibiotics. The TZM-bl cell line, previously designated JC53-bl (clone 13), is a HeLa cell line that expresses endogenous CXCR4. The parental cell line (JC.53) stably expresses large amounts of CD4 and CCR5. The TZM-bl cell line is CD4+ CXCR4+ CCR5+ (21, 31) and was generated from JC.53 cells by introducing separate integrated copies of the luciferase and β-galactosidase (β-Gal) genes under the control of the HIV-1 promoter. The TZM-bl indicator cell line enables simple and quantitative analysis of HIV by use of either β-Gal or luciferase as a reporter. The CEM T lymphoblast cell line was maintained in RPMI 1640 containing 10% FBS and antibiotics. Human peripheral blood mononuclear cells (PBMCs) were prepared by Ficoll-Hypaque fractionation of cell concentrates obtained from healthy donors seronegative for hepatitis B virus and HIV. PBMCs were activated with 10 μg/ml phytohemagglutinin for 3 days in RPMI 1640 containing 10% FBS.
The 12G5 monoclonal anti-CXCR4 antibody conjugated to phycoerythrin (PE) was purchased from BD Biosciences Pharmingen (Franklin Lakes, NJ). AMD3100 was a gift from AnorMed Corp. (Langley, British Columbia, Canada). The CXCL12 expression vectors were previously constructed in our lab (2), using the pET32a expression plasmid from Novagen (Madison, WI). An antibody to heparin sulfate (HepSS-1) and its isotype were purchased from U.S. Biologicals. The HepSS-1 antibody recognizes an epitope in heparan sulfate glycosaminoglycan (HS-GAG). This epitope is closely related to the O-sulfated and N-acetylated glucosamine residue linked to glucuronic acid in HS-GAG. The antibody recognizes heparan sulfate in a variety of fresh and established human, mouse, monkey, rat, hamster, and chicken cell lines.
The pET32a expression plasmids containing the CXCL12α and CXCL12γ inserts were mutated according to protocol, using the Gene Tailor point mutagenesis system from Invitrogen. The expression plasmids, previously constructed in our lab (2), were methylated and used as templates in PCRs. The primers for the PCRs were designed in accordance with the protocol provided by Invitrogen (Carlsbad, CA). The mutations were generated by methylating the wild-type (wt) expression plasmids and using the methylated plasmids as templates for PCR. The primers were designed to mutate the basic residues of the BBXB domains into alanines. Upon transformation, the methylated wild-type DNA is degraded by the bacterial repair system, leaving only the mutated plasmid DNA. The PCR specifications were 20 cycles comprising 30 seconds of melting at 94°C, 30 seconds of annealing at 55°C, and 6 min of polymerizing at 72°C. The resulting PCR bands were ligated for 1 h using a rapid ligation system (Roche, Indianapolis, IN) and were transformed into DH5α TR1 competent bacteria. The plasmids were sequenced at the BFF sequencing core facility at Indiana University, Indianapolis, IN. We generated a variant with mutations of the conserved CXC cysteines of CXCL12γ to alanines (γΔC) as a negative control, since it is well established that CXCL8/interleukin-8 activity is dependent on the conserved cysteines (23). The CXCL12 mutants generated in this study are described in Fig. Fig.1A1A.
The wild-type and mutated CXCL12 splice variants were purified and expressed from Escherichia coli as previously described (2). Briefly, The BL21-Gold (DE3)pLysS strain (Stratagene, La Jolla, CA) was transformed with the pET32a recombinants. Single colonies were inoculated into 5 ml of LB medium with antibiotics and shaken overnight. For large-scale expression, the overnight cultures were inoculated into 1 liter of LB medium. The 1-liter culture was shaken until an optical density at 600 nm (OD600) of 0.2 was achieved. IPTG (isopropyl-β-d-thiogalactopyranoside) was added to induce expression, and the cultures were shaken for 4 h. After being shaken, the cultures were centrifuged and the pellets were stored at −80°C until purification.
Bacterial pellets for the CXCL12 splice variants were suspended in 5 ml lysis buffer per pellet and rotated at room temperature for 1 h. The lysates were then centrifuged, and the pellets containing inclusion bodies were denatured in buffer containing 6 M guanidine-HCl. The clarified supernatants were poured onto a Ni-nitrilotriacetic acid (Ni-NTA) histidine-binding column (Sigma, St. Louis, MO) equilibrated with 10 column volumes of denaturing binding buffer. Both ends of the column were immediately capped, and the columns were rotated at 4°C overnight. The denatured proteins were refolded on the column by stepwise removal of the denaturant as previously described (19). The columns were washed with decreasing levels of urea followed by increasing levels of imidazole to remove nonspecific bacterial contaminants. The proteins were eluted with 2 ml elution buffer, and the elution buffer was exchanged with phosphate-buffered saline (PBS) on a PD-10 desalting column. Protein concentrations were quantified by the Bradford assay (Bio-Rad, Hercules, CA) and analyzed for size and purity by SDS-PAGE. The proteins were also analyzed by Western blotting using a polyclonal antibody to the histidine tag that was present in all constructs (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). The relative concentrations of the purified proteins were converted to nanomolar values by using the predicted molecular weight of each recombinant.
Migration of CEM T lymphoblasts was measured as previously described (2). Briefly, CEM T lymphoblast cells were suspended at a density of 2 × 106 cells/ml in chemotaxis medium (Iscove's modified Dulbecco's medium [IMDM]) supplemented with 0.5% bovine serum albumin (BSA). The chemokines, with or without 100 nM AMD3100, were suspended in chemotaxis medium at the concentrations indicated in Fig. Fig.2.2. The suspended chemokines (in 0.6 ml) were first added to the wells of a 24-well plate, followed by a 5-μm-pore-size trans-well membrane for each well (Costar; Corning Incorporated, Acton, MA). The cells were added to the inside of the trans-well membrane at a volume of 0.1 ml (2 × 105 cells). The plates were then capped and incubated for 4 h at 37°C and 6% CO2. After the incubation period, the trans-well membranes were removed from the wells and the migrated cells counted by flow cytometry. Results are reported as percentages of cells that migrated compared to total cell counts in wells with no trans-well membranes.
HIV-1 Env-mediated fusion was performed by a vaccinia virus-based reporter assay as previously described (1). Briefly, the target CD4+ CXCR4+ CCR5+ TZM-bl cells (AIDS Reagent Program, Bethesda, MD) (12) were infected by a vaccinia virus encoding the bacteriophage T7 RNA polymerase. The target CD4− CXCR4− LM(tk−) and Sog9 cells were separately coinfected with vCB-3 (encoding CD4), vYF-4 (encoding CXCR4), and vTF7-3 (encoding T7 RNA polymerase). For each assay, effector HeLa cells were coinfected by PT7-lacZ vaccinia virus (carrying the lacZ reporter gene under the control of the T7 promoter) and vaccinia virus encoding either the X4 LAV Env, the R5 Ba-L Env, or the control Unc Env, an uncleaved HIV-1 Env that has its cleavage site mutated and therefore cannot engage in membrane fusion. The Unc Env was used to measure the nonspecific background in the fusion assay. The target cells were plated in a 96-well plate at 1 × 105 cells per well and treated for 1 h with the wt or mutated CXCL12 variants at increasing concentrations. After 1 h of incubation at 37°C and 6% CO2, the Env-expressing HeLa cells were mixed with the target cells in a 1:1 ratio. The cell mixtures were incubated for 2.5 h at 37°C and then lysed, and the substrate chlorophenol red-β-d-galactopyranoside (CPRG) was added. The extent of cell fusion was assayed by measuring the amount of β-galactosidase produced. When PBMCs were used as target cells, they were infected with PT7-lacZ vaccinia virus, while effector HeLa cells were coinfected with vaccinia virus encoding the T7 RNA polymerase and one of the previously listed HIV-1 Envs.
For CXCR4 competitive binding assays, the CD4+ CXCR4+ CEM T lymphocytes were used. Indirect binding experiments involved the use of fixed concentrations of CXCL12 mixed with escalating doses of a specific antagonist for CXCR4 (AMD3100). The cells were washed twice and suspended in cold binding buffer (50 mM HEPES, 5 mM MgCl2, 1 mM CaCl2, 150 mM NaCl, pH 7.4, 0.5% BSA) with a 100 nM concentration of each chemokine in the presence of increasing concentrations of AMD3100. The cells were incubated for 1 h at 4°C and washed once with cold binding buffer and once with PBS-1% BSA. To measure chemokine binding, the cells were suspended with 2 μg/ml fluorescein isothiocyanate (FITC)-conjugated anti-histidine-tag antibody or its isotype control (U.S. Biologicals) and incubated on ice for 1 h. The cells were then washed three times with PBS-1% BSA and suspended in PBS for fluorescence-activated cell sorter (FACS) analysis. The results were normalized to percent binding, with 100% being the geometric mean fluorescence intensity seen with no antagonist added. All assays were performed in triplicate and included a control with no CXCL12 or with a chemokine-like protein (MC148) to verify the background staining.
Direct binding experiments involved the use of 125I-labeled SDF-1α/CXCL12α (purchased from PerkinElmer, Boston, MA) and cold recombinant CXCL12 isoforms and mutants. CEM cells (1 million cells/tube, in triplicate), either untreated or treated with heparanase (10 mU/ml), were incubated with 0.25 nM 125I-labeled CXCL12α (specific activity of 2,200 Ci/mmol) and escalating concentrations of cold CXCL12 chemokines. The cell-chemokine mixtures were incubated at 4°C for 1 h and then washed three times with PBS-0.1% BSA. The cell pellet-associated counts were measured in a gamma counter. Binding was evaluated by calculating the % chemokine binding, where 100% binding was for CEM cells incubated with the 125I-labeled SDF-1α and no cold chemokine.
Sog9 cells lacking HSPG were originally derived from mouse LM(tk−) cells (7). Sog9 cells were previously isolated as chondroitin 4-O-sulfotransferase-1-deficient cells from LM(tk−) cells which were partially resistant to herpes simplex virus type 1 (HSV-1) infection and defective in the expression of heparan sulfate (HSPG) because of a splice site mutation in the EXT1 gene encoding the HS-synthesizing enzyme (7).
LM or Sog9 cells were harvested with 0.2 mM EDTA in PBS and washed three times with PBS supplemented with 1 mM CaCl2 and 0.1% BSA. The cells (3 × 106/ml) were then incubated for 1.5 h at 37°C and 5% CO2 in PBS-1 mM CaCl2-0.1% BSA in the presence of 10 mU/ml heparinase, 10 mU/ml chondroitinase, or PBS. Following incubation, the cells were washed with cold PBS that contained 1% BSA and suspended in 300 μl PBS-1% BSA containing a 300 nM concentration of the indicated CXCL12 variants. The cells were then incubated at 4°C for 1.5 h, washed with PBS-1% BSA, and stained for 1 h with an anti-His-tag antibody conjugated to FITC. The cells were analyzed by flow cytometry.
LM, Sog9, or CEM cells were harvested with 0.2 mM EDTA in PBS and washed three times with PBS supplemented with 1 mM CaCl2 and 0.1% BSA. The cells (3 × 106/ml) were then incubated for 1.5 h at 37°C and 5% CO2 in PBS-1 mM CaCl2-0.1% BSA in the presence of 10 mU/ml heparinase, 10 mU/ml chondroitinase, or PBS. Following incubation, the cells were washed and stained with a monoclonal antibody to heparin sulfate (U.S. Biologicals, Swampscott, MA) and a secondary FITC-conjugated anti-mouse antibody. The cells were analyzed by flow cytometry.
To measure internalization of CXCR4, we used a previously reported method, with some modifications (28). Briefly, the CXCR4+ CEM T cells were incubated for 30 min with 50 nM CXCL12 variants at 37°C. The cells were then centrifuged, washed once with 1 ml PBS, and then washed again with PBS containing 0.1% BSA. The cells were then incubated on ice with the anti-CXCR4 monoclonal antibody 12G5 conjugated with PE or the isotype control, suspended in PBS-1% BSA for 30 min, washed with PBS-1% BSA, PBS-0.1% BSA, and PBS consecutively, suspended in PBS only, and analyzed by flow cytometry. The same procedure was used to analyze the kinetics of CXCR4 internalization. For this analysis, the chemokine was incubated with the cells for 1 h, washed three times with IMDM-0.5% BSA, and then further incubated at 37°C for 2 and 3 h without chemokine. In some experiments, samples were removed every 30 min and surface expression of CXCR4 was analyzed by flow cytometry. Results are expressed in terms of percent geometric mean fluorescence intensity of CXCR4 staining in the absence of chemokine.
The amino acid sequences of CXCL12α and CXCL12γ, along with the mutations we generated, are illustrated in Fig. Fig.1A.1A. The BBXB mutations are denoted in the figure legends as follows: B1, 24KHLK27→24AHAA27 for both CXCL12α and CXCL12γ. The following mutations were constructed from CXCL12γ: B2, 77KKEK80→77AAEA80; B3, 85KRQK88→85AAQA88; and γΔC, 9CPC11→9APA11. Combinations of the mutations are denoted B12, B123, and B23 to indicate the locations of the mutated BBXB domains. DNA sequencing of the mutant plasmid DNAs confirmed the site mutations. Each chemokine purification procedure yielded approximately 0.5 to 2.0 mg protein. The purity of the proteins was verified on a Coomassie blue-stained SDS-PAGE gel (Fig. (Fig.1B).1B). To verify the identities of the affinity-purified chemokines, small aliquots of the purified proteins were fractionated by SDS-PAGE and blotted to a membrane, and the blot was probed with a polyclonal antibody to the histidine tag. The results of this analysis demonstrated that all purified proteins reacted with the anti-His antibodies and showed the expected molecular sizes of the mutant proteins (Fig. (Fig.1C).1C). Mutation of the large side chain amino acids to alanine resulted in a slight decrease in molecular weight that was observed in SDS-PAGE analyses. These experiments confirmed the purity and expected gel mobilities of the CXCL12 chemokine variants. In our previous study, we observed that the constructed His-tagged recombinant CXCL12α was indistinguishable from commercially purchased CXCL12α in terms of chemotaxis and anti-HIV activity, suggesting that the added C-terminal His tag did not significantly alter CXCL12 biological function (2).
To determine whether the purified recombinant CXCL12 proteins were biologically functional, we tested them in an in vitro chemotaxis assay, using the human CEM T-lymphoblastic cell line as target cells. The purified CXCL12α induced the expected cell migration that we previously reported (2). The efficiency of the induced migration was always comparable to that observed with commercial CXCL12α (2). Consistent with our previous data, the purified CXCL12γ induced a low level of cell migration at 100 nM (Fig. (Fig.2).2). We used the double-cysteine mutant CXCL12γΔC as a negative control, since it does not induce significant chemotaxis at any concentration (Fig. (Fig.22).
Consistent with previous data in the literature (4), mutating the conserved BBXB domain (B1) in either CXCL12α or CXCL12γ did not result in a significant effect on their chemotactic activities (Fig. (Fig.2).2). The chemotactic activities of the B2, B3, and B23 mutants were dramatically increased, to levels comparable to those observed with wild-type CXCL12α (Fig. (Fig.2).2). The significantly higher chemotactic activities of the CXCL12γ B2 and B3 mutants were restored only when the conserved BBXB domain (B1) at positions 24 to 27 remained unaltered. The CXCL12γ mutants that lost the second and/or third BBXB domain (B2, B3, or B23) induced significant levels of cell migration at 100 nM (Fig. (Fig.2).2). The chemotaxis activities of all affinity-purified CXCL12 proteins were eliminated by AMD3100 treatment, indicating that they were CXCR4 mediated (Fig. (Fig.2).2). The results demonstrated that the additional BBXB domains found at the novel carboxyl tail of CXCL12γ play an important role in the observed low level of chemotactic activity.
We previously reported that CXCL12γ exhibits at least a fivefold potency in inhibiting HIV-1 Env-mediated fusion (2). To determine the relative potencies of the C-terminal BBXB mutants of CXCL12γ, we tested their inhibitory activities in an HIV-1 Env-mediated fusion assay. A dose-response analysis was performed for each of the CXCL12 variants, which were preincubated with TZM-bl target cells (Fig. 3A and B) or human PBMCs (Fig. (Fig.3C).3C). The results show that the CXCL12γ mutants caused a dose-dependent inhibition of X4 Env fusion for both target cell types (Fig. 3A and C). The 50% inhibitory concentration (IC50) of each chemokine variant was calculated from at least three different experiments run in duplicate (Table (Table1).1). The R5 Ba-L Env was used as a negative control because it has been well established that CXCL12 does not block R5 Env-mediated fusion (2). None of the CXCL12 mutants showed any inhibition of R5 Env-mediated fusion (Fig. (Fig.3B3B).
We consistently observed that the B1 mutants of CXCL12α and CXCL12γ had significantly reduced inhibitory activities in the X4 fusion assay (Fig. 3A and C). In contrast, mutating the C-terminal BBXB domains consistently resulted in enhancement of the inhibitory effects of CXCL12γ. The calculated IC50 for the B1 mutant of CXCL12α or CXCL12γ was always higher than that of the wild-type chemokine (Table (Table1).1). The IC50 of the CXCL12γ B1 mutant was ~9-fold higher than that of wild-type CXCL12γ. In contrast, the IC50s of the B2, B3, and B23 mutants, which contained the conserved BBXB domain, were consistently lower than that of wild-type CXCL12γ (Table (Table1).1). Taken together, these results suggest that the conserved BBXB domain is critical for optimal antiviral activity for both CXCL12α and CXCL12γ and that the presence of the C-terminal BBXB domains seems to exert a positive effect on the observed potent antiviral activity of CXCL12γ.
To investigate chemokine binding and the role of HSPG, we used the HSPG-deficient mouse cell line Sog9 and its original HSPG+ parent LM cell line (7) as targets in the binding assays. Additionally, binding of the CXCL12 variants was examined with L cells treated with either chondroitinase or heparanase. These enzymes selectively degrade either chondroitin or heparan sulfate. The levels of bound chemokines were determined by FACS analysis, using a FITC-conjugated anti-His-tag antibody that detects the His tag present in all constructs. FACS analysis confirmed the deficient expression of HSPG in Sog9 cells (Fig. (Fig.4A)4A) and efficient HSPG expression in LM cells (Fig. (Fig.4B).4B). Expression of HSPG on LM cells was significantly reduced after heparanase treatment (Fig. (Fig.4C)4C) but not after chondroitinase treatment (Fig. (Fig.4D).4D). We also stained Sog9 and LM cells for CXCR4, and both cell lines were negative (data not shown). Consistent with previous data in the literature (4, 29), we observed a loss of HSPG binding of the CXCL12αB1 mutant. Binding of this mutant to HSPG was comparable to that observed with the HSPG-deficient Sog9 cells (Fig. (Fig.4E).4E). Binding of wt CXCL12γ was at least 10 times more efficient than that of wt CXCL12α (Fig. (Fig.4F).4F). All of the BBXB mutants had reduced binding to HSPG, except for the γB1 mutant that had lost the conserved BBXB domain (Fig. (Fig.4F).4F). The γB1 mutant chemokine consistently showed a significant increase in binding to LM cells. The binding of all chemokine variants was dramatically reduced after heparanase treatment but not after chondroitinase treatment (Fig. (Fig.4F).4F). These results indicate specific binding to HSPG. Among the single-site BBXB mutants, the most dramatic decrease in binding to LM cells was observed with the γB3 mutant. The results suggest a critical role for this BBXB site (B3) in HSPG binding. The lowest HSPG binding was observed with the triple BBXB mutant (B123), in which all three BBXB domains were mutated (Fig. (Fig.4F).4F). Binding to Sog9 cells served as the background, since these cells are defective in HSPG expression (Fig. (Fig.4F).4F). MC148 was used as another negative control, since it lacks BBXB domains and does not bind HSPG (Fig. (Fig.4F).4F). We previously described the construction and purification of an MC148 chemokine-like protein that is defective in chemotaxis assays (2). The results demonstrated efficient and specific binding of wt CXCL12γ to HSPG through its BBXB domains and suggest a critical role for the B3 domain.
To determine whether CXCL12 binds to CXCR4 or HSPG on CEM cells, we first verified expression of these surface molecules by FACS analysis. The results demonstrated efficient expression of CXCR4 (Fig. (Fig.5A)5A) and a lack of HSPG expression (Fig. (Fig.5B).5B). Analysis of CXCL12 binding to CXCR4 revealed that a higher concentration (~10-fold) of AMD3100 was required to displace the CXCL12γ chemokine on human CEM T cells than that used to displace CXCL12α (Fig. (Fig.5C5C and Table Table1).1). The AMD3100 concentration required to result in the reduction of 50% of binding by the γB1 and γB3 mutants was not significantly different from that for wild-type CXCL12γ. The results demonstrate that CXCL12γ and its mutants had significantly increased binding affinities for CXCR4 and that the loss of any BBXB domain in CXCL12γ had little effect on the observed binding.
To confirm the previous binding results, we performed a direct binding method using 125I-radiolabeled CXCL12α. The results of this direct binding analysis confirmed the data obtained with the indirect binding method. For example, higher concentrations of cold CXCL12α were required to result in 50% displacement of bound 125I-CXCL12α (Fig. (Fig.5D).5D). We consistently observed at least a 6 times higher binding affinity for CXCL12γ than for CXCL12α (Fig. (Fig.5D5D and Table Table2).2). Mutating the conserved BBXB domain reduced the binding efficiencies of both CXCL12α and CXCL12γ. However, eliminating the second and/or third BBXB domain of CXCL12γ increased the amount of chemokine binding to CEM cells (Fig. (Fig.5D5D and Table Table2).2). Treating CEM cells with heparanase had no effect on the binding affinities of the CXCL12 variants (Table (Table2).2). These results confirm the higher binding affinity of CXCL12γ for CXCR4 and demonstrate that binding of the CXCL12 variants to CEM cells occurs mainly through CXCR4.
To determine the efficiency of CXCR4 internalization induced by the CXCL12 variants, we analyzed the surface expression of CXCR4 on cells incubated with or without CXCL12 chemokine proteins. It has been demonstrated that CXCL12 does not block 12G5 binding to CXCR4 (28). The results of this analysis indicated that CXCL12γ induced a significantly higher level of CXCR4 internalization. The amount of CXCR4 surface expression on CEM cells incubated with CXCL12γ was significantly lower than that on cells incubated with CXCL12α (Fig. (Fig.6A).6A). The efficiency of CXCR4 internalization induced by the CXCL12γ BBXB mutants depended on whether or not the conserved domain was mutated. The second and/or third BBXB domain mutants were not significantly different from wild-type CXCL12γ in terms of the ability to induce CXCR4 internalization. The CXCL12γB1, CXCL12γB12, and CXCL12γB123 mutants induced less internalization of CXCR4 (Fig. (Fig.6A).6A). Incubation of CEM cells with the CXCL12γΔC mutant had no significant internalization effect on CXCR4 (Fig. (Fig.6A6A).
To determine the kinetics of internalization, we performed experiments involving incubation of the CXCL12 chemokine variants with CEM cells, washing, and further incubation without chemokine for different times. Following 3 h of incubation after washing away the chemokine, the cells preincubated with CXCL12α restored normal expression of CXCR4. In contrast, cells preincubated with CXCL12γ restored CXCR4 expression to 50% (Fig. (Fig.6B).6B). These results indicate that CXCL12γ induces more efficient and sustained internalization of CXCR4 than does CXCL12α.
This study aimed at investigating the mechanism of the potent anti-HIV activity of CXCL12γ. We hypothesized that the extra BBXB domains in CXCL12γ play a critical role in the observed potent anti-HIV activity. Our hypothesis was based on our recent findings demonstrating that CXCL12γ is at least 5 to 6 times more potent than CXCL12α in HIV blocking assays (2). The results demonstrated that mutating the C-terminal BBXB domains of CXCL12γ restored chemotaxis activity to levels comparable to those observed with CXCL12α. In contrast, mutating the conserved BBXB (24KHLK27) domain of CXCL12α resulted in significant decreases in CXCR4 binding and anti-HIV-1 effects but had no significant effect on its chemotactic activity. These observations are consistent with the data in the literature, except for the binding data (3, 29). The CXCL12αB1 mutant was previously reported to have the same binding affinity for CXCR4 as wt CXCL12α (29). This discrepancy might be explained by the different structures of our BBXB mutants. For example, Amara et al. (4), Fernandez et al. (29), Sadir et al. (25), and recently, Reuda et al. (24) utilized a BBXB mutant that had the conserved KHLK domain mutated to SSLS or SHLS, while ours was mutated to AHAA. Additionally, the studies by Laguri et al. and Reuda et al. (17, 24) utilized CXCL12γ constructs that included the sequence coding for the bovine rhodopsin C9 tag (TETSQVAPA) in frame at the C terminus of CXCL12γ (17, 24). Our CXCL12γ constructs contain a C-terminal His tag.
Decreased CXCR4 binding and internalization of CXCL12αB1 correlated with its lower antiviral activity. In most cases, higher-affinity binding to CXCR4 correlated with a higher efficiency of receptor internalization. An exception to this was the case for CXCL12γB1, whose binding to CXCR4 was not significantly different from that of wt CXCL12γ. It is possible that CXCL12γB1 binds to a different domain of CXCR4 that does not engage receptor activation and internalization. Alternatively, other properties of CXCL12γB1, such as structure and/or the ability to dimerize, might play a critical role in CXCR4 internalization. Previous studies suggested IL-8 dimerization as a mechanism for regulation of neutrophil adherence-dependent oxidant production (32).
Previous studies suggested that HSPG sequesters the chemokine by its interaction with the conserved BBXB domain close to the cell surface, creating a gradient effect (29). The results of this study suggest that in addition to the BBXB domain, other structural requirements might play a critical role in CXCL12 binding. We did not observe a direct correlation between binding to HSPG and the observed potent anti-HIV activity. However, such a correlation was true for CXCR4 binding and internalization. For example, binding to HSPG was significantly higher with CXCL12γB1 than with wt CXCL12γ, but CXCL12γB1 was significantly less potent in HIV blocking assays. Additionally, the γB3 mutant showed a significant reduction in HSPG binding but was the most potent HIV-1 inhibitor. HSPG binding does not seem to modify the antiviral activity of CXCL12 or promote internalization of CXCR4.
The restored chemotaxis of the C-terminal BBXB domain mutants suggests that the C-terminal tail of CXCL12γ acts as a suppressive domain that significantly impairs its chemotaxis activity. The results suggest that the extra BBXB domains in the C terminus of CXCL12γ contribute to an altered structure of the chemokine that impairs its chemotactic activity but enhances its anti-HIV activity. The finding that CXCL12γ exhibits significantly greater affinity for CXCR4 and significantly reduced chemotactic activity through CXCR4 introduces the intriguing idea that CXCL12γ may function as a natural CXCR4 antagonist, in a similar manner to that of the bicyclam AMD3100. AMD3100 has been shown to only bind CXCR4, without engaging receptor signaling (13, 15, 26). Since abundant CXCL12γ protein expression has been detected in cardiac muscle, valves, and large vessels (24), we hypothesize that CXCL12γ might function as a natural antagonist. This might represent a unique mechanism of self-regulation aimed at controlling potentially unfavorable signaling events in the heart myocytes by CXCL12α present in the bloodstream. Previous studies suggested that MCP-3 could act as an antagonist, since it binds CCR5 with a high affinity without inducing receptor internalization and has the ability to inhibit the functional response to MIP-1β (8).
Previous studies demonstrated that MCP-3 could compete efficiently for gp120 binding, but it was found to be a weak inhibitor of HIV infection, probably as a consequence of its inability to internalize CCR5 (8). Our study presents several lines of experimental evidence to suggest that the different antiviral activities of the BBXB mutants are associated with significant changes in the efficiency of CXCR4 internalization. First, internalization of CXCR4 was slightly impaired by CXCL12αB1 and significantly impaired by CXCL12γB1, the two mutants that lost the conserved BBXB domain. Second, wt CXCL12γ and CXCL12γB1 had similar affinities for CXCR4, but CXCL12γB1 had a significant loss in the ability to internalize CXCR4 and block HIV-1. Third, the CXCL12γ B2 and B3 mutants showed slight but significant increases in CXCR4 internalization that correlated with increased antiviral activity. Fourth, the kinetics of CXCR4 internalization indicated that CXCL12γ induced sustained CXCR4 internalization, in contrast to CXCL12α; CXCR4 expression was maintained at significantly lower levels with CXCL12γ 3 h after removing the chemokine. Finally, despite its efficient binding to CXCR4, the double-cysteine mutant CXCL12γΔC showed no significant antiviral effect due to its inability to induce CXCR4 internalization.
The mechanisms responsible for mammalian GPCR endocytosis remain largely undefined (reviewed in reference 18). Long-lasting CCR5 internalization in a subset of long-term nonprogressors has been suggested as a mechanism for the protective effect against disease progression (20). It is critical to understand CXCR4 internalization, since it controls the temporal and spatial aspects of G protein signaling. Identifying the specific mechanisms that regulate internalization of CXCR4 will provide important insight into the development of new strategies to manipulate receptor signaling and will provide novel targets for designing drugs that can be used in the prevention and treatment of a wide range of human diseases, including cardiovascular disease and cancer progression (10). It is possible that CXCL12γ stabilizes a distinct conformation of the CXCR4 receptor that may signal selectively to different G proteins or promote a distinct receptor conformation that facilitates better binding to β-arrestins. Alternatively, CXCL12γ may induce more efficient CXCR4 phosphorylation that enhances internalization. Recent studies demonstrated that CXCR4 dimerization and β-arrestin-mediated signaling account for the enhanced chemotaxis to CXCL12α in WHIM syndrome (16). The mechanisms that control CXCR4 endocytosis induced by CXCL12γ have yet to be elucidated fully.
We thank JoAnn Trejo for helpful discussions. The following reagents were obtained through the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH: TZM-bl cells, from John C. Kappes, Xiaoyun Wu, and Tranzyme Inc.; and AMD3100, from Anormed.
This study was supported by NIH grant RO1 A152019-01 to G.A. Q.J. was supported by a scholarship from the Chinese Scholarship Council, Beijing, China.
Published ahead of print on 16 December 2009.