The peptides, C17, C17-GSGC, CGSG-C17, and C21 were synthesized by Chinapeptides Corp. (Shanghai, China) with purity >90%. C52L and C34 are kind gifts from Dr. Shibo Jiang at the New York Blood Center (31). The peptide sequences are C17, KINVKQIAARLLPPLYS, C17-GSGC, KINVKQIAARLLPPLYSGSGC, CGSG-C17, CGSGKINVKQIAARLLPPLYS, C21, KINVKQIAARLLPPLYSLVFI, C52L, NHTTWMEWDREINNYTSLIHSLIEESQNLQEKNEQELLELDKWASLWNWFNIKIK, C34, WMEWDREINNYTSLIHSLIEESQNQQEKNEQELL. C17-GSGC and keyhole limpet hemocyanin (KLH) were conjugated to make C17-KLH by Chinapeptides Corp. All primers were synthesized by Sangon Biological Engineering Technology Inc. (Shanghai, China). The plasmids, including pNL4-3.LucR⁻E⁻, pHXB2-Env, pJRFL-Env, and pVSVG, were obtained from the NIH AIDS Research and Reference Reagent Program. The plasmids, including pSF162-Env, pCNE28-Env, and pCNE49-Env, and cell line Ghost-R5 were kindly provided by Dr. Linqi Zhang at Tsinghua University. Ghost-X4 cell is a kind gift from Dr. Zhiwei Wu at Nanjing University.

The antigenic region of CCR5 was predicted by three methods and the corresponding online servers, SVMTriP¹ (32), FBCpred² (33), and PAP³ (34).

Three 8-week-old female Balb/c mice were first injected with 100 μg of C17-KLH in complete Freund’s adjuvant (CFA) (Pierce, Rockford, IL, USA) into intraperitoneal cavity at day 0, followed by one boost with a mixture of 100 μg C17-KLH and incomplete Freund’s adjuvant (IFA) (Pierce, Rockford, IL, USA) on day 21, and two boosts with a mixture of 50 μg of C17-KLH and 100 μg of C17 peptide in IFA, on day 35 and day 49. Control group was immunized with PBS in CFA or IFA in priming or boost, respectively. The sera were harvested a week after the final immunization and stored at −20°C.

The immunogenicity of C17 was characterized by indirect ELISA. Microtiter plates were coated at 4°C overnight with 50 µl synthetic peptide C17, KLH-C17 or irrelevant peptide C34 from gp41 protein (all at 5 µg/ml) diluted in NaHCO₃ (pH 9.6). The wells were then blocked with 0.3% gelatin in PBS at 37°C for 2 h. After wash with PBS containing 0.05% Tween-20, mouse sera diluted in PBS at a fourfold dilution (starting from 1:100) were added, followed by incubation at 37°C for 2 h. The peroxidase-conjugated anti-mouse immunoglobulin antibodies (Dako, Denmark) were used as secondary antibody. After extensive washes, the substrate o-phenylenediamine was added and stopped by 2 M H₂SO₄ after 10 min. The sera titer was identified as the highest dilution of serum with the average values of 490 nm ≥ 0.2.

The total IgGs (anti-C17) were purified from mouse antisera using protein G affinity column (GE Healthcare) as the manual provided by the manufacturer. Briefly, 1 ml mouse serum was diluted with 4 ml PBS and passed through the affinity column. The column was washed with 20 ml PBS to remove non-specific antibodies. Total IgGs were eluted with low pH elution buffer (pH 2.5, 0.1 M glycine) and neutralized with high pH buffer (pH 8.0, 1 M Tris–HCl). The purified antibodies were then dialyzed against PBS by using a 30 kD ultra centrifugal filter device (Millipore, MA, USA). The antibody concentrations were quantified by measuring absorbance at 280 nm and stored at 4°C until use.

Ghost-R5 cells were incubated with anti-C17 total IgG or normal IgG (purified from negative mouse sera) at a concentration of 20 µg/ml at 37°C for 1 h. After wash with PBS (pH 7.4) containing 2% fetal bovine serum, cells were labeled by Alexa Fluor 647 goat anti-mouse IgG (H + L) (Life Technologies) on ice for 45 min. To demonstrate that the binding was caused by MPR-specific antibodies, anti-C17 was preincubated with 100 µM C17 peptide at 37°C for 1 h. An irrelevant peptide C34 was also used, and 2D7 was used as the positive control. To demonstrate that anti-C17 did not induce the internalization of CCR5, Ghost-R5 cells were preincubation with anti-C17 (100 µg/ml) or medium at 37°C for 2 h, followed by labeling CCR5 with PE-labeled anti-human CCR5 monoclonal antibody (2D7, BD Biosciences, USA). The unlabeled Ghost-R5 cells were used as the negative control. Human peripheral blood mononuclear cells (PBMCs) were also used. PBMCs were first activated with PMA (50 ng/ml, Sigma) and ionomycin (1 µM, Sigma) for 8 h. Activated PBMCs were incubated with anti-C17 (100 µg/ml) or normal IgG for 2 or 16 h, followed by labeling CCR5 with PE-labeled 2D7. The data were collected by BD Accuri C6 (BD Biosciences) and analyzed with FlowJo software.

The XTT assay was used to detect the cytotoxicity of MPR-derived peptides (C21, C17, C17-Chol, and Chol-C17) to Ghost cells as previously described (35). In brief, Ghost cells were plated in 96-well plates (100 μl/well, 1 × 10⁵/ml). Then equal volumes of the peptide at graded concentrations were added. After incubation at 37°C for 4 days, 50 µl of XTT solution (1 mg/ml, containing 0.02 µM phenazine methosulfate) was added. The absorbance at 450 nm was measured after 4 h.

The fluorochromes Fluo-4AM and FuraRed-AM (Life technologies) were added to HEK 293T cells transiently transfected with plasmids expressing wild-type (WT) CCR5 at a final concentration of 1 µM. The cells were washed once after incubation at 37°C for 20 min and resuspended in Hank’s balanced salt solution. Cells (10⁶) were incubated with anti-C17, normal IgG, or 2D7 (BD Pharmingen) at a concentration of 20 µg/ml, followed by stimulation with MIP-1β (1 µg/ml, R&D systems). The data were collected by BD Accuri C6 using time and events model. Calcium level was reflected by the ratio of Fluo-4/FuraRed fluorescence.

Alanine scan mutations were introduced with the Muta-direct™ site-directed mutagenesis kit (SBS Genetech, Beijing, China). In the CCR5-CCR3, CCR5-CCR2b, and CCR5-CXCR4 chimeras, the nucleotides coding for amino acids 22–38 of the hCCR5 receptor were replaced by nucleotides coding for amino acids 26–42 of the hCCR3, 34–50 of the hCCR2b, and 38–54 of the hCXCR4, respectively. Totally 3 µg of WT or mutated CCR5 plasmids were transfected into 293T cells in 6-well plate using transfection reagent VigoFect (Vigorous Biotechnology Beijing Co., Ltd., China) as instructions of the manufacturer. After 24 h, flow cytometry was performed to check the surface protein expression level. After washing with phosphate buffer solution (PBS, pH 7.4) containing 2% fetal bovine serum, the cells were labeled by FITC-conjugated mouse anti-human CCR5 monoclonal antibody (BD Biosciences, USA) at room temperature for 45 min. The data were collected by BD Accuri C6 and analyzed by CFlow Plus software.

Rink Amide HMBA resin was used for the solid-phase peptide synthesis (SPPS) with the loading of 0.23 mmol g⁻¹. The SPPS cycles were performed by microwave peptide synthesizer (Liberty, Discovery, CEM Corporation). The scale used for each peptide was 0.1 mmol. The coupling reagent for Fmoc-amino acid building blocks was HATU (1-[bis (dimethylamino) methylene]-1H-1,2,3-triazolo[4,5-b] pyridinium 3-oxid hexafluorophosphate), which was conducted under microwave irradiation for 10 min with a power of 20 W at 50°C. This process was repeated twice when coupled with Fmoc-Arg(pbf)-OH. The amino acids (5 equiv.) and HATU (5 equiv.), which were dissolved in DMF and DIEA (10 equiv.), were then dissolved in NMP. Those ingredients were added to the reacting system automatically. With the synthesis of C17-GSGC-Chol [Gly–Ser–Gly–Cys (GSGC) as a flexible linker], the first building block Fmoc-Asp (OH)-OChol as a side chain anchor was added manually. This building block (4 equiv.) was reacted with the resin in room temperature for 2 h, with HATU (4 equiv.), and DIEA (8 equiv.), which were dissolved in DMF/DCM (1:1) solvent. This side chain anchored reaction was followed by capping reagent treatment, which helped the unreacted amino groups capped by acetylation using HOBt (0.013 m) in Ac2O/DIEA/DMF (4.75:2.25:93.0 v/v/v), and the capping was carried out under microwave irradiation at 65°C for 30 s, and 30 s again with power of 40 W. Piperidine (20%)/HOBT (0.1 m) in DMF was used for Fmoc deprotection, which was under the condition of microwave irradiation at 75°C for 180 s, and 180 s again with a power of 48 W. For Chol-CGSG-C17 peptide, the last cycle with the building block Fmoc-Asp(OH)-OChol was coupled manually, which followed the same procedure with the Fmoc-Asp(OH)-OChol coupling in C17-GSGC-Chol peptide. After finishing all the coupling cycles, the resin was transferred to a flask, followed by a treatment with trifluoroacetic/triisopropylsilane/H₂O (95:2.5:2.5 v/v/v) cocktail for 2 h in room temperature to detach the peptide from the resin. The peptide detached from the resin was washed by diethylether three times then dried over vacuum. Those impure peptides were purified via preparative HPLC with a polar CN column with a flow rate of 7 ml/min. The pure peptide was dried by lyophilizer and was stored in −20°C.

Single-cycle pseudovirus infection assay was performed as previously described (36, 37). Briefly, pseudoviruses were packaged by transient transfection of HEK 293T cells with pNL4-3.LucR⁻E⁻ and plasmids expressing Env of HXB2 (B, X4), JRFL (B, R5), SF162 (B, R5), CNE28 (AE, R5), and CNE49 (BC, R5). Pseudovirus expressing vesicular stomatitis virus G (VSVG) protein served as a control. The titer of pseudoviruses was determined by twofold dilution in 1 × 10⁴ Ghost cells to final volume of 200 µl.

To measure capabilities of CCR5 mutants in supporting viral infection, Ghost-X4 cells were transfected with 60 µg of plasmids expressing WT or mutated CCR5 in 100 mm culture dish using VigoFect. Cells were harvested 36 h after transfection and distributed in 96-well plates with 1 × 10⁴ cells/well. After 12 h in 96-well plate, the cells were infected with 200 µl of JRFL or SF162 pseudovirus. The cells were then cultured at 37°C in 5% CO₂ for 48 h, and the relative luminescence units (RLUs) were measured by a luciferase kit (Promega, USA) and luminometer (Ultra 386, Tecan, USA). Results were normalized to the reporter activity measured with WT CCR5.

To evaluate the antiviral effects of peptides or antibodies, 1 × 10⁴ of Ghost cells were plated into 96-well plates 12 h before HIV-1 pseudovirus infection. For inhibitory activities of anti-C17, 100 µl of anti-C17 at a fourfold serial dilution from 5 µM was incubated with Ghost cells at 37°C for 30 min. Then 100 µl of JRFL, HXB2, CNE28, or CNE49 pseudovirus was added. To neutralize MPR-specific antibodies, anti-C17 was preincubated with 25 µM C17 peptide for 1 h. For inhibitory activities of MPR-derived peptides, 50 µl of MPR-derived peptides (C17, C17-Chol, or Chol-C17) at a twofold serial dilution from 50 µM was incubated with 1 × 10⁴ of Ghost cells at 37°C for 30 min. Then equal volume of JRFL, SF162, CNE28, CNE49, or VSVG pseudovirus was added to the mixture. For the time course experiment, 50 µl of JRFL pseudovirus was added to 100 µl of Ghost-R5 cells (0 min). Fifty microliters of C17-Chol peptide (final concentration 10 µM) or C52L peptide (final concentration 100 nM) were added to Ghost-R5 cells and JRFL virus mixture at time 0, 10, 20, 40, and 60 min. The cells were cultured at 37°C in 5% CO₂ and lysed 48 h later for RLU detection. RLUs were detected as the aforementioned methods.

Two tailed t-test (t-test) was used to analyze the difference between two groups. p Value was calculated by PRISM software (GraphPad, CA, USA), and a difference is regarded significant if p is less than 0.05.

The N-terminus and ECL2, illustrated in Figure 1A, has been demonstrated to be important for the entry of HIV-1, whereas the potential role of other domains on CCR5 has not been well documented. Antibody is a well-accepted tool to study the function of a specific region. A set of antibodies, including 2D7, have been used to demonstrate the crucial role of N-terminus and ECL2 in the cellular entry of HIV-1 (17). By using bioinformatic methods, including SVMTrip (32), FBCPred (33), and PAP (34), to identify potential antigenic regions on CCR5, we found a previously unnoticed region located between N-terminus and TM1 (Figure 1B). A 22 amino acid (aa)-length region (N-terminus 1 in Figure 1B) has been reported to be important for HIV-1 infection (9, 38), and the key region of N-terminus was subsequently shortened to 14 aa in length (N-terminus 2 in Figure 1B) (10). To minimize any potential spatial and functional overlap with previously well-characterized N-terminus, we selected a 17 aa length region that only shared one lysine (K22) residue with N-terminus 1. Because this region located closely to the membrane (Figure 1A, right panel), we named it as MPR, while the peptide derived from this region was designated as C17 (Figure 1B). The K22 was kept to improve the solubility of C17.

Direct blocking of MPR by antibodies was taken to study the role of MPR in HIV-1 infection. To generate antibodies against MPR, the C17 peptide was conjugated to KLH by a flexible linker GSGC. Balb/c mice were primed by intraperitoneal injections of 100 µg C17-KLH with CFA, followed by one booster with a mixture of 100 µg C17-KLH with IFA, and then two boosters with a mixture of 50 µg C17-KLH, 100 µg C17 peptide and IFA. This immunization strategy elicited relatively high IgG titers (geometric mean = 1:25,600) against both C17-KLH and C17 peptide, but not an irrelevant peptide C34 (Figure 1C). For further function studies, the total IgGs were purified from antisera by protein G affinity column to remove any non-specific inhibitory effect of serum components. These purified antibodies were designated as anti-C17 in following functional studies. As demonstrated by flow cytometry analysis, anti-C17 not only bound to peptides coated on an assay plate but also bound to cells expressing natural CCR5 proteins (Figure 1D). This binding could be blocked by preincubating anti-C17 with C17 peptide, confirming the specificity of the binding (Figure 1D). In contrast, preincubating anti-C17 with an irrelevant peptide C34 did not block the binding (Figure 1D).

Then, the antiviral effect of antibodies was evaluated by pseudovirus expressing envelop proteins from various HIV-1 strains. The strain name was used to represent individual pseudovirus hereafter unless indicated particularly. Anti-C17 could dose dependently inhibit the infection of CCR5-tropic JRFL virus at an IC₅₀ of 0.40 µM without any inhibition on CXCR4-tropic HXB2 virus (Figure 2A; Table 1). Furthermore, if anti-C17 was preincubated with C17 peptide to neutralize MPR-specific antibodies, the inhibitory capability of anti-C17 was completely lost at most antibody concentrations tested (Figure 2A). The impaired inhibitory capability was not caused by the direct blockage of viral infection by C17 peptide, because the peptide alone neither inhibited nor enhanced the infection of JRFL and HXB2 (Figure 2B). At the highest concentration (5 µM) of anti-C17, the blocking capability was still impaired significantly (p < 0.001) but not completely, probably due to the insufficient amount of C17 peptide to neutralize all C17 binding antibodies (Figure 2A). These results demonstrated that the inhibition was caused exclusively by antibodies targeting the MPR region. Moreover, the inhibition effect of anti-C17 was also confirmed in two clinic isolated strain CNE28 and CNE49 albeit with a higher IC₅₀ compared to JRFL (Table 1). Since some antibodies against ECL1 or N-terminus inhibited the viral infection by inducing the internalization of CCR5 (27, 39), the potential effect of anti-C17 on the receptor internalization was further evaluated. In sharp contrast to antibodies against ECL1, MPR-specific antibodies were less likely to induce the internalization of CCR5 receptor because incubating anti-C17 with Ghost-R5 cells at 37°C for 2 h did not decrease the amount of cell surface CCR5 labeled by PE-conjugated 2D7 antibody (Figure 2C), whose epitope located far from MPR (Figure 1A). Human PBMCs were enrolled to further confirm this result in cells with physiological CCR5 density. PBMCs were first activated and a small amount of cells became CCR5+ (Figure 2D). Incubating anti-C17 with these cells for either 2 or 16 h did not reduce the mean fluorescence intensity (MFI) of 2D7-labeled CCR5 (Figure 2E). Therefore, the inhibitory effect of anti-C17 was not ascribed to the internalization of CCR5. Next, the potential effect of anti-C17 on the physiological function of CCR5 was assessed by a calcium flux assay. As shown in Figure 2F, anti-C17 did not inhibit the calcium influx induced by MIP-1β, whereas 2D7 greatly abolished the function of CCR5. Taken together, the inhibition of HIV-1 infection by antibodies against MPR region suggested the important role of MPR in viral infection.

The antibody blocking experiment could not fully rule out the possibility that MPR-binding antibodies might interrupt the interaction between N-terminus and gp120 by the steric hindrance. To give another line of evidence showing MPR did play roles in HIV-1 infection, we replaced MPR of CCR5 with the same region from other three co-receptors (Figure 3A). The chimeras showed similar surface expressing level as WT CCR5, demonstrated by flow cytometry analysis after transient transfection (Figure 3B). Two commonly used CCR5-tropic strains, JRFL and SF162, were used to check whether these CCR5 chimeras could still support viral entry. As shown in Figure 3C, CCR5–CCR3 and CCR5–CCR2b chimeras, expressed in Ghost cells carrying human CD4 but not CCR5, showed significantly (p < 0.001) impaired capability to support JRFL infection, achieving less than 50% efficacy of WT CCR5. CCR5–CXCR4 also significantly (p < 0.01) impaired the viral infection, albeit to a lesser extent (Figure 3C). The infection of SF162 strain was more sensitive to the substitution of MPR, since all chimeras showed very poor capability to support the infection of SF162 (Figure 3D).

The alanine scanning mutagenesis was then performed to identify key residues participating in the viral infection. We found none of the single-residue mutations obviously (>125% or <75%) affected the expression level of CCR5 on the cell surface (Figure 3E). Then the Ghost cells carrying individual mutation were infected by JRFL or SF162 virus. As shown in Figure 3F, some of the mutations significantly enhanced or impaired the viral infection. For JRFL strain, I28A, L32A, and Y37A mutations impaired the viral infection by 30–50%, whereas I23A and N24A enhanced the infection by more than 50%. The importance of MPR was also confirmed in SF162 strain. Similar to JRFL, I23A, and N24A mutations consistently enhanced the viral infection by approximately 50%, while L32A mutation impaired the infection by 50% (Figure 3F). Slightly different from JRFL, two non-significant mutations for JRFL, K26A, and L36A, greatly impaired the viral infection by 50 or 75%, respectively, indicating SF162 strain was more sensitive to CCR5 MPR mutations (Figure 3F), consistent with the results from the MPR substitution experiment (Figure 3D). Although there was a slight difference between JRFL and SF162 in the alanine scanning mutagenesis, the key residues including I23, N24, and L32 which shared similar and obvious effects on both strains confirmed the role of CCR5 MPR in HIV-1 infection.

Next, we mapped all key residues (highlighted in red) that significantly affected the infection of either JRFL or SF162 in the structure of CCR5 (Figure 4) (7). Surprisingly, the residues I23, I28, L32, and L36 were all located outside the protein and formed a leucine zipper-like structure (Figure 4, middle panel). Clearly the function of this leucine zipper-like structure was not to mediate the dimerization of CCR5, since two key residues (highlighted in blue) for dimerization of CCR5 located on a different face of the protein (Figure 4, right panel) (40). This leucine-zipper-like structure was separated from the highly flexible N-terminus (green) by a disulfide bond (orange) and located away from ECL2 (yellow), which formed the binding site of natural CCR5 ligands with N-terminus (Figure 4, left panel). Moreover, the leucine zipper faced to the outer space of CCR5 rather than the binding pocket of Maraviroc, an antagonist of CCR5, indicating that this structure may not contribute to the binding of CCR5 ligand (Figure 4, left panel).

The altered capability of CCR5 chimeras or mutated CCR5 in supporting viral infection could be ascribed to the conformational change of other parts in CCR5 or the interruption of the interaction between MPR and viral proteins. If the interaction is the reason, the peptide derived from MPR may be able to bind to the corresponding domain on viral proteins and affect the viral infection. In fact, peptide inhibition has been extensively used to study the interaction between CCR5 and viral proteins (38, 41, 42).

Unexpectedly, C17 peptide showed no inhibitory effect (IC₅₀ > 50 µM) on all strains tested (Table 2), probably due to insufficient time for them to arrive at the target (43). One promising way to break this limitation was concentrating peptides at the place where the interaction happens. In this case, anchoring C17 to the cell membrane might enhance its inhibitory activity, since HIV-1 entry happened on the lipid raft and MPR was located closely to the membrane. Additionally, membrane anchoring could also constrain the peptide to the right orientation rather than approaching the target randomly (Figure 5A). Enhancing the membrane anchoring by extending the peptide to the transmembrane domain was not a good strategy, since C21 peptide containing additional four residues from TM1 showed obvious cytotoxicity which was very likely to give false positive results (Figure 5B). Therefore, we used an alternative strategy by conjugating cholesterol to C17. Cholesterol was an ideal membrane anchor to enrich the peptide on the cell surface via interacting with lipid rafts (43, 44). We designed two derivatives of C17 with cholesterol conjugated to the N- or C-terminal of the C17 peptide (Chol-C17 or C17-Chol) as shown in Figure 5A. C17-Chol was expected to mimic the physiological location and orientation of MPR, since it is the C-terminal of C17 located more closely to the cell membrane (Figure 5A). On the other hand, Chol-C17, whose C17 bound to membrane in an inappropriate orientation, was synthesized as a control to rule out any non-specific inhibitory effect caused by cholesterol conjugation. To synthesize cholesterol conjugated peptides, a linker (GSGC or CGSG) was added to the C-terminal or N-terminal of C17 to provide a conjugating site for cholesterol (Figure 5A, right panel). Both cholesterol-conjugated peptides showed no cytotoxicity at high (50 µM) or low (6.25 µM) concentration (Figure 5B).

Next, the inhibitory effects of two conjugates were measured. As expected, C17-Chol inhibited the infection of SF162 at a low micromolar level (IC₅₀ = 1.04 ± 0.10 µM) (Figure 5C; Table 2). In contrast, Chol-C17 showed no inhibitory effect, indicating that conjugating of cholesterol did not cause non-specific inhibition (Figure 5C). This result also demonstrated that the orientation of C17 was crucial for its functionality. Additional HIV-1 strains were also enrolled to evaluate the inhibitory capability of C17-Chol. In sharp contrast to C17 which showed no inhibitory effect (IC₅₀ > 50 µM) on all strains tested, C17-Chol inhibited the infection of another lab-adaptive strain JRFL (IC₅₀ = 1.38 ± 0.68 µM) and two clinic-isolated strain CNE28 (IC₅₀ = 0.55 ± 0.06 µM) and CNE49 (IC₅₀ = 0.42 ± 0.06 µM) (Table 2). Moreover, C17-Chol did not inhibit the infection of pseudovirus carrying VSVG protein whose entry did not depend on CCR5, further ruling out the possibility of non-specific inhibitory effects caused by C17-Chol (Table 2).

To further confirm that C17-Chol specifically interrupted the interaction between CCR5 and viral proteins, we next added C17-Chol at different time point after mixing JRFL virus with Ghost-CCR5 cells. As shown in Figure 5D, the inhibitory effect of C17-Chol was greatly impaired if being added as soon as 10 min after the mixing. In contrast, C52L, an inhibitor of gp41 fusion complex formation, could still inhibit viral entry at this time point, but completely lost the inhibitory effect 20 min after mixing virus and cells. These results revealed that C17-Chol targeted to a very early event in the viral infection, consistent with the fact that the interaction between CCR5 and gp120 happened before the formation of gp41 fusion complex (45, 46).