Exploring the stereochemistry of CXCR4-peptide recognition and inhibiting HIV-1 entry with D-peptides derived from chemokines.

Chemokine receptor CXCR4 plays an important role in the immune system and the cellular entry of human immunodeficiency virus type 1 (HIV-1). To probe the stereospecificity of the CXCR4-ligand interface, d-amino acid peptides derived from natural chemokines, viral macrophage inflammatory protein II (vMIP-II) and stromal cell-derived factor-1alpha (SDF-1alpha), were synthesized and found to compete with (125)I-SDF-1alpha and monoclonal antibody 12G5 binding to CXCR4 with potency and selectivity comparable with or higher than their l-peptide counterparts. This was surprising because of the profoundly different side chain topologies between d- and l-enantiomers, which circular dichroism spectroscopy showed adopt mirror image conformations. Further direct binding experiments using d-peptide labeled with fluorescein (designated as FAM-DV1) demonstrated that d- and l-peptides shared similar or at least overlapping binding site(s) on the CXCR4 receptor. Structure-activity analyses of related peptide analogs of mixed chiralities or containing alanine replacements revealed specific residues at the N-terminal half of the peptides as key binding determinants. Acting as CXCR4 antagonists and with much higher biological stability than l-counterparts, the d-peptides showed significant activity in inhibiting the replication of CXCR4-dependent HIV-1 strains. These results show the remarkable stereochemical flexibility of the CXCR4-peptide interface. Further studies to understand the mechanism of this unusual feature of the CXCR4 binding surface might aid the development of novel CXCR4-binding molecules like the d-peptides that have high affinity and stability.

tion, CXCR4 together with another chemokine receptor CCR5 are two principal co-receptors for the cellular entry of the human immunodeficiency virus type 1 (HIV-1) 1 (4 -7). The stromal cell-derived factor-1 (SDF-1␣) is the only known natural ligand of CXCR4 and plays important roles in migration, proliferation, and differentiation of leukocytes (8,9). The viral macrophage inflammatory protein II (vMIP-II) encoded by human herpesvirus 8 (10) is an antagonistic chemokine ligand of CXCR4 (11,12). vMIP-II also interacts with other chemokine receptors such as CCR5 and CCR3 and inhibits HIV-1 entry mediated by these co-receptors.
CXCR4 and other chemokine receptors belong to the superfamily of seven transmembrane G-protein-coupled receptors (GPCRs) (13). These membrane proteins transmit signals from extracellular ligands to intracellular biological pathways via heterotrimeric G-proteins and have been a major class of therapeutic targets for a wide variety of human diseases (14). As such, characterizing the mechanism of biological recognition between these receptors and their ligands is essential for understanding the physiological or pathological processes that they mediate and devising novel strategies for clinical intervention. For CXCR4, studies have been carried out by a number of laboratories using chimeric chemokine receptors and site-specific mutants to study multiple domains of CXCR4 that are important for interacting with chemokine ligands and HIV-1 (15)(16)(17)(18)(19)(20)(21)(22)(23). However, because there is no high resolution crystal structure available for CXCR4 (or any other chemokine receptor) alone or complexed with ligands, the structural and biochemical basis of ligand binding and signaling through these important membrane receptors remains poorly understood.
To further define the structure-function relationship of the chemokine receptor-ligand interaction, theoretical computer modeling and site-directed mutagenesis were combined to predict plausible structural models for chemokine receptors and their complexes with ligands, such as interleukin-8 receptor ␤ (24) and CCR5 (25,26). Structural models of CXCR4 and its complex with ligands were also proposed (27,28). Complementary to modeling and mutational analyses of the receptors, chemical synthetic approach was used to dissect the structural and functional determinants of a number of chemokines and their binding to receptors (29). As to chemokines SDF-1␣ and vMIP-II, which recognize CXCR4, a large number of synthetic mutants and peptide analogs of SDF-1␣ were used to analyze the structure-activity relationship of different regions of SDF-1␣ in CXCR4 binding (30 -34).
For vMIP-II, we recently found that a synthetic 21-residue peptide derived from the N terminus of vMIP-II, designated as V1, is a potent antagonist of CXCR4 and inhibits HIV-1 replication in CXCR4 ϩ T-cell lines (35,36). Being highly amenable to chemical synthesis and modification, this V1 peptide prompted us to use chemically modified analogs of V1 as probes to study the molecular recognition of CXCR4-ligand complex. Because one important aspect of receptor-ligand interaction is the requirement of stereospecificity, we report here the synthesis and structure-function characterization of all-D-amino acid analogs of V1 peptide, designated DV1 peptides. Unexpectedly, these D-peptides display strong binding and antagonistic activity toward CXCR4, thus revealing that the peptide binding site on CXCR4 is tolerant of changes in chirality of ligands. Similar observations are also made for other D-peptides derived from the N terminus of SDF-1␣. These findings have important implications for understanding the mechanism of CXCR4-ligand interaction and designing novel inhibitory molecules. Furthermore, DV1 peptides are highly resistant to proteolytic degradation and show significant activity in blocking HIV-1 replication in CXCR4 ϩ cell lines, thus demonstrating their advantage over natural L-peptides for potential clinical application.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human chemokines SDF-1␣, MIP-1␤, and vMIP-II (R&D Systems, Minneapolis, MN) were lyophilized and dissolved as 1 or 2.5 g/l stock solutions in sterile phosphate-buffered saline (PBS) and stored at Ϫ20°C in aliquots. The radioiodinated MIP-1␤ and SDF-1␣ were purchased from PerkinElmer Life Sciences. The specific activity of 125 I-MIP-1␤ and 125 I-SDF-1␣ was 2200 Ci/mmol. Cell culture media and G418 were purchased from Invitrogen. The anti-CXCR4 monoclonal antibody (mAb) 12G5 (37) was purchased from BD PharMingen (San Diego, CA). 293 cells (gift from Dr. R. Doms, University of Pennsylvania) and Sup-T1 cells (provided by the National Institutes of Health AIDS Reagent Program) were maintained in Dulbecco's modified Eagle's medium plus 10% fetal bovine serum. The cell lines used for the HIV-1 replication assay were kindly provided by other laboratories: MT-4 cells by Dr. N. Yamamoto (Tokyo Medical and Dental University, Tokyo, Japan), chronically HIV-1-infected H9 cells (H9/ III B ) by the National Institutes of Health AIDS Research and Reference Reagent Program, and Sup-T1 cells by Dr. J. Hoxie (University of Pennsylvania). These cells were grown in 25 mM HEPES-buffered RPMI 1640 supplemented with 10% fetal bovine serum, 50 units/ml penicillin, 50 g/ml streptomycin, and 0.125 g/ml amphotericin B. HIV-1 was prepared from a culture supernatant of H9/III B cells.
CD Spectroscopy-The experiments were performed following our previously published procedure (24,40). 50 M solutions of peptides were prepared in 0.01 M sodium phosphate buffer. The spectra were recorded on an Aviv 62A DS spectrometer (Aviv Instruments Inc., Lakewood, NJ) with a 0.01-cm path length quartz cuvette and scanned by every 2 nm at room temperature.
Flow Cytometry-Following the procedure described in our recent publication (35), Sup T1 cells (2 ϫ 10 5 ) were incubated with an anti-CXCR4 monoclonal antibody (mAb) 12G5 (10 g/ml) and various concentrations of peptides for 40 min at 4°C, then with fluorescein isothiocyanate-conjugated goat anti-mouse IgG (Southern Biotechnology Associates, Inc., Birmingham, AL) for 40 min at 4°C. Finally the cells were fixed in 2% paraformaldehyde in PBS and analyzed on a FACScan flow cytometer (Coulter Epics Elite, Beckman Coulter, Hialeah, FL).
12G5 Competitive Binding to CXCR4 -Sup-T1 cells were incubated with anti-CXCR4 mAb 12G5 (10 g/ml) in the presence of different concentrations of the peptide at 4°C for 40 min. The cells were washed twice with the binding buffer and stained with fluorescein isothiocyanate-labeled secondary antibody at 4°C for another 40 min. Finally, the cells were washed two to three times and fixed with the fixing buffer. As a negative control, cells were stained only with secondary antibody. The final data points were taken as the mean fluorescence intensity of the peptide sample subtracted by that of the control. The IC 50 values of the peptides in competing with 12G5 binding to CXCR4 were calculated by using Prizm software.

125
I-SDF-1␣ Competitive Binding to CXCR4 -CEM-T4 cells were harvested and washed twice with PBS. Competition binding experiments were performed using a single concentration (0.2 nM) of 125 I-SDF-1␣ in the presence of increasing concentrations of unlabeled ligands in a final volume of 100 l of binding buffer (50 mM HEPES, pH 7.4, 1 mM CaCl 2 , 5 mM MgCl 2 , 0.1% bovine serum albumin) containing 2 ϫ 10 5 cells. Nonspecific binding was determined by the addition of 100 nM unlabeled SDF-1␣. Samples were incubated for 60 min at room temperature. The incubation was terminated by separating the cells from the binding buffer by centrifugation and washing once with 500 l of cold binding buffer. Bound ligands were quantitated by counting ␥ emissions.

125
I-MIP-1␤ Competitive Binding to CCR5-Following an experimental procedure similar to that described above, 293 cells transfected with CCR5 and 125 I-MIP-1␤ were used to determine the specific binding activity of peptides to CCR5. Intracellular Calcium Measurement-Sup T1 cells and CCR5 transfected 293 cells were used to measure the intracellular calcium influx.
[Ca 2ϩ ] i was measured using excitation at 340 and 380 nm on a fluorescence spectrometer (PerkinElmer LS50). Calibration was performed using 0.033% Triton X-100 for total fluorophore release and 1.66 mM EGTA to chelate free Ca 2ϩ . Intracellular Ca 2ϩ concentrations were calculated by using the fluorescence spectrometer measurement program.
Stability of Peptides in Human Serum-Pure human serum (kindly provided by Robert Korngold, Thomas Jefferson University, Philadelphia, PA) was diluted with PBS to 80% serum solution. DV1 and V1 peptides were dissolved in the serum solution with a concentration of 10 mM. Samples were collected at different time points of incubation at room temperature and subjected to HPLC analysis with the injection of 10-l peptide samples (Microsorb-MV C 18 5-m, 25 cm ϫ 4.6 mm, 80% CH 3 CN with 0.1% trifluoroacetic acid, UV 220 nm, 1 ml/min). The stability of the peptides was calculated based on the changes in the intensity of UV absorbance of the peptides.
Virus Neutralization Assay-Recombinant viruses were generated by cotransfecting the human 293 T cell line with pSVIIIenv plasmids expressing either HXBc2 or 89.6 envelope glycoproteins and pHXBH10⌬envCAT (41). Recombinant viruses were normalized for reverse transcriptase activity and used to infect the following target cells: (i) human PBMC activated with phytohemagglutinin (PHA) and interleukin-2 or (ii) canine thymocytes expressing human CD4 or human CD4 plus human CCR5 (42). The target cells were preincubated with the peptides at varying concentration for 1 h prior to coculture with the recombinant viruses. Infected cells were harvested 3 days later, and CAT activity was measured in cell lysates, as previously described (43).
HIV-1-induced Cytopathic Effect and Viral p24 Antigen Production-As described previously (44), MT-4 cells were seeded at 6 ϫ 10 4 cells/150 l/well in a 96-well microplate. HIV-1 was added at 50 TCID 50 / well in the absence or presence of peptide inhibitors. After 4 days of incubation at 37°C in a CO 2 incubator, the supernatant was examined for the level of viral p24 antigen using the enzyme immunoassay kit (Beckman Coulter). The cell viability was also determined by 3Ј-(4,5dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide method. The cytotoxic effect of peptides on MT-4 cells was assessed by incubating uninfected cells with peptides without the addition of HIV-1 under the same conditions as above.

D-Peptides Derived from vMIP-II N Terminus
Bind CXCR4 -In this study, we used synthetic all-D-amino acid peptides to explore the stereospecificity of the molecular recognition between the CXCR4 receptor and its ligands. Based on our previous finding that V1 peptide derived from the N terminus (residues 1-21) of viral chemokine vMIP-II antagonizes CXCR4 function and blocks HIV-1 entry via CXCR4 (35, 36), we synthesized DV1, an all-D-enantiomer of V1 peptide that has the identical amino acid sequence as V1 but the chirality of all residues changed to D-configuration (Table I). DV1 showed a circular dichroism (CD) spectrum that is the mirror image of V1 peptide (Fig. 1a). In addition, the spectra displayed negative and positive intensity bands near 220 nm for V1 and DV1, respectively, suggesting that both peptides adopted a random conformation in aqueous solution. If DV1 and V1 are mirror images of each other as suggested by CD, there is significant conformational difference between them, as highlighted by distinct amino acid side chain orientations in their structural models (Fig. 2).
DV1 peptide was tested for CXCR4 binding activity in a competitive binding assay using anti-CXCR4 mAb 12G5 following the procedure described previously (35). Surprisingly, despite the changes in chirality of all amino acids and thus extensive conformational difference from parent V1 peptide as described above, DV1 peptide showed strong binding affinity for CXCR4 (Fig. 3a). Furthermore, the CXCR4 binding affinity of DV1 peptide was much higher than that of V1 peptide, as shown by an IC 50 of 32 nM for DV1 as compared with that of 456 nM for V1 in competing with the specific CXCR4 binding by mAb 12G5 (Table I). Using another assay method, DV1 was shown to compete with 125 I-SDF-1␣ in CXCR4 binding with an IC 50 of 13 nM as compared with that of 218 nM V1 peptide (Fig.  3b). Because vMIP-II binds other receptors such as CCR5, we tested the activity of DV1 peptide in CCR5 binding using the 125 I-MIP-1␤ competitive binding method. In contrast to its strong CXCR4 interaction, DV1 peptide did not show any CCR5 binding even at a much higher concentration of 100 M (Fig.  3c). Consistently, a similar observation was made for V1 peptide (35). These results suggested that CXCR4 binding by both DV1 and V1 peptides appear to be receptor-selective.
To further demonstrate CXCR4 recognition by DV1 peptide, a FAM-labeled DV1 peptide was prepared for conducting direct ligand binding experiments on Sup T1 cells that naturally express CXCR4 (Fig. 4). The binding of FAM-DV1 on Sup T1 cells was mediated by CXCR4, as shown by the significant reduction of binding signal of FAM-DV1 upon the addition of SDF-1␣. This was consistent with the results of DV1 in competing with 125 I-SDF-1␣ in CXCR4 binding as described above and further supported the hypothesis that DV1 peptide recognizes CXCR4 receptor. To examine whether D-and L-peptides bind to similar or different site(s) on CXCR4, 100 M V1 peptide was added and found to block FAM-DV1 binding, completely similar to the effect of 100 M unlabeled DV1 peptide (Fig. 4). These results indicated that these peptides of opposite chiralities recognize similar or at least overlapping site(s) on the receptor. Again, this was consistent with the ability of both peptides in competing with 125 I-SDF-1␣ and mAb 12G5 in CXCR4 binding as described above. Competitive binding studies using FAM-DV1 peptide were also conducted on 293 cells transfected with wild type CXCR4. When 200 nM FAM-DV1 was used, there was significantly higher (ϳ50%) FAM-DV1 binding to CXCR4 transfected 293 cells than 293 cells, whereas the difference was much less for lower concentrations (100 and 200 nM) of FAM-DV1 (data not shown). This was probably a result of the high nonspecific binding of FAM-DV1 to 293 cells, which was also observed for 125 I-SDF-1␣. Consistent with the results obtained on Sup T1 cells, the binding of FAM-DV1 on b Those L-amino acid residues in the sequence are indicated by underlines, whereas the rest are D-amino acids.
c The position for alanine replacement is indicated by bold and underlined type CXCR4-transfected 293 cells was inhibited by the addition of SDF-1␣ (300 nM), DV1 (100 M), or V1 (100 M) (data not shown).
N-terminal Half of D-Peptides Is the Major Determinant of CXCR4 Binding-As described above, the chirality change in all amino acid residues of V1 peptide resulted in DV1 peptide with significantly increased CXCR4 affinity. To further explore the effect of chirality on receptor binding, peptides of hybrid chiralities were synthesized. By using Cys-11 as the central point of V1 peptide, we changed the chirality of residues before (residues 1-10) and after Cys-11 (residues 12-21) to D-configuration, respectively (Table I). Two peptides synthesized accordingly were therefore designated as V1-DCL and V1-LCD that were composed of mixed D-and L-amino acids. V1-DCL peptide has D-configuration for the N-terminal half and L-configuration for the C-terminal half, whereas V1-LCD peptide has the opposite. Both peptides were subjected to anti-CXCR4 mAb 12G5 competitive binding assay. V1-DCL displayed CXCR4 binding activity comparable with DV1, and a similar observation was made for V1-LCD and V1 (Table I and Fig. 5a). These results further confirmed the ability of CXCR4 in recognizing peptide ligands of opposite or mixed chiralities. In addition, they suggested that CXCR4 binding affinity of these peptides was mainly determined by the chirality of their N-terminal half with D-residues at this part interacting with CXCR4 receptor better than L-residues. Apparently, the first 10 residues of DV1 or V1 are the major determinant of activity. In contrast, chirality change in the C-terminal half had much less effect on CXCR4 binding activity.
To further address the role of the C-terminal half, we synthesized a truncated DV1 analog, designated as DV3, in which C-terminal residues (residues 11-21) of DV1 were removed. Although retaining CXCR4 binding, DV3 was less potent than DV1 (Table I and Fig. 5a). This is consistent with previous study of V3 peptide, a truncated V1 analog that also contains only first 10 residues and shows much decreased CXCR4 binding as compared with V1 (35). These results indicated that, although the N-terminal half (residues 1-10) is most important, the C-terminal half (residues 11-21) also plays a role in receptor binding.
Specific Residues in D-Peptides Important for CXCR4 Recognition-To further characterize the structure-activity relationship of DV1 peptide, eight mutant peptides containing alanine replacement at various positions of DV1 were synthesized (Table I). Consistent with the important role of the N-terminal half of DV1 peptide as discussed above, mutations in this part had more significant effect on receptor binding than mutations in the other part. Mutant peptides DV1-L1A, DV1-W5A, and DV1-R7A containing alanine substitutions at Leu-1, Tyr-5, and Arg-7, respectively, all displayed a substantial loss of CXCR4 binding with a decrease in IC 50 values of 34 -741-fold (Table I and Fig. 5b). In contrast, mutations in the C-terminal half (Cys-11, Cys-12, Gln-16, and Arg-18) led to little or less significant change in binding activity. To determine whether the difference in receptor affinity of these mutant peptides were the result of conformational changes induced by the mutations, we studied the solution conformations of DV1-L1A that showed the most dramatic change in CXCR4 binding by CD spectroscopy. DV1-L1A and DV1 displayed similar CD spectra, indicating no major conformational difference (Fig. 1b). As such, changes in CXCR4 binding of mutant peptides were more likely caused by the loss of specific side chains from the alanine replacement and thus should reflect the role of these side chains in receptor interactions.
Because both DV1 and V1 peptides recognize CXCR4 despite their very different side chain orientations caused by changes in chirality, this raised the question of whether these peptides may bind the receptor through nonspecific interactions. A diverse group of peptides and organic molecules with a high positive charge (ϩ8 or ϩ9) are known to bind CXCR4, presumably through electrostatic interactions with the negatively charged surface (Ϫ9) of CXCR4 (45)(46)(47)(48). It can be imagined that these interactions are likely less sensitive to changes in the stereochemistry of ligands so long as the overall positive charge is maintained. However, this should not be the case for DV1 peptide because it has an overall charge of only ϩ3.5, much lower than other highly positively charged CXCR4 ligands described above. In addition, mutant peptides DV1-K10A and DV1-R18A, in which positively charged side chains of Lys-10 and Arg-18 were removed, respectively, have an even lower charge of ϩ2.5, yet retain significant CXCR4 binding activity. Most importantly, as discussed above for mutants DV1-L1A and DV1-W5A, uncharged hydrophobic side chains (most notably the side chain of Leu-1) of DV1 peptide play an essential role in receptor recognition and are responsible for different receptor affinities of up to 741-fold. A similar conclusion was drawn in previous analysis of V1 and mutant peptides (36). Taken together, these data clearly demonstrated that, despite the seemingly contradictory notion of the nonspecificity of CXCR4 toward the stereochemistry of ligands, CXCR4 interacts with both DV1 and V1 peptides with a high degree of sensitivity for specific side chain groups in these peptides.
Given that the orientations of side chains in DV1 and V1 peptides are not the same because of their different D-and L-configurations (Fig. 2), one should expect that corresponding residues in DV1 and V1 peptides have distinct interactions with CXCR4 and thus contribute differently to receptor binding. To verify this, we compared the changes in receptor binding affinity caused by mutations in DV1 and V1 peptides (Table  II). The largest difference in residue contribution to receptor binding between DV1 and V1 peptides was found for those residues at the N-terminal half, e.g. Leu-1 was the most critical binding determinant and Lys-10 the least important one for DV1, whereas the corresponding Lys-10 was most critical for V1. By contrast, the difference in residue contribution to CXCR4 affinity between these two peptides was much smaller for the C-terminal half, which was consistent with the less important role of the C-terminal half in receptor recognition.
D-Peptides Derived from SDF-1␣ N Terminus Also Recognize CXCR4 -Having shown that the peptide binding surface of CXCR4 is tolerant of chirality changes in peptide ligands derived from the N terminus of vMIP-II, we wanted to test whether this may also apply to other chemokine-derived peptides. A peptide derived from the N terminus (residues 5-14) of SDF-1␣ has been shown previously to bind CXCR4 (31)(32)(33). With different size and sequence from DV1 and V1 peptides, this SDF-1␣ N-terminal peptide should allow us to test the generality of our hypothesis. Both D-and L-enantiomers of this peptide sequence were synthesized and designated as DS1 and S1 peptides, respectively (Table I). Their different conformations as the mirror image of each other were confirmed by CD spectra (data not shown). DS1 and S1 peptides showed similar binding affinity for CXCR4 in anti-CXCR4 mAb 12G5 competitive binding assay (Fig. 5c). The ability of both peptides to recognize CXCR4 was also verified by a different binding method using 125 I-SDF-1␣ (data not shown). These results further supported the notion that CXCR4 receptor can accommodate the changes in amino acid chirality of different peptide ligands.
D-Peptides Inhibit SDF-1␣ Signaling through CXCR4 -DV1 peptide was tested for activity in inducing signal via CXCR4 as measured by intracellular calcium influx in Sup T1 cells expressing the receptor. DV1 did not show any activity in triggering CXCR4 signal transduction but blocked SDF-1␣ signaling via CXCR4, thus indicating that it is an antagonist of CXCR4 function (Fig. 6). This is consistent with a previous observation that V1 peptide is also a CXCR4 antagonist (35). For those DV1 mutant peptides that had some binding activity for CXCR4 (Table I), their ability to induce signal or interfere with SDF-1␣ signaling via CXCR4 was also studied by using the same assay. None of the mutant peptides tested showed any signaling activity via CXCR4 and are thus receptor antagonists like the parent DV1 peptide (data not shown).
D-Peptides Are Highly Stable in Human Serum-The incorporation of D-amino acids into a peptide sequence should enhance its resistance to proteolytic degradation because unnatural residues are less likely to be recognized by natural enzymes. To test this, the biological stability of DV1 and V1 peptides was studied in human serum. The peptides were mixed with human serum and degraded products of peptides monitored by HPLC for a period of time. V1 peptide was rapidly degraded, and by 24 h the peptide peak on HPLC disappeared completely (data not shown). In contrast, DV1 peptide did not FIG. 2. Different amino acid side chain conformations between DV1 and V1 peptides. These models were randomly generated and based on the assumption that the peptides adopt a mirror image structure, as shown by CD spectra. Note that, whereas all amino acid side chains in these two peptides display different orientations because of their opposite chiralities, only the side chains of Leu-1 and Trp-5 that are shown by mutation studies to be critical for CXCR4 binding are chosen as representatives and highlighted in black. Two views of DV1 peptide are shown: one as the mirror image of V1 peptide with opposite peptide main chain and side chains and the other after 180°rotation with the same main chain direction but different side chain orientations.
show any degradation during the same period and even after 72 h of prolonged incubation with human serum. These results were consistent with the notion that DV1 peptide has much higher biological stability than V1 peptide, which should be of advantage for any potential clinical application.
D-Peptides Inhibit HIV-1 Replication in CXCR4 ϩ Cell Lines-Having shown that DV1 peptide is a potent CXCR4 antagonist and highly stable in biological conditions, we reasoned that it could inhibit the replication of CXCR4-dependent HIV-1 strains by blocking the entry of viruses via CXCR4 co-receptor. By measuring CAT activity in activated human PBMC infected with a recombinant HIV-1 containing the HXBc2 envelope glycoproteins, we found that DV1 peptide was a potent inhibitor of infection. By contrast, a control mutant peptide DV1-L1A, which contains a single alanine substitution at Leu-1 but substantially loses CXCR4 binding, did not show any activity (Fig. 7a). The DV1 peptide was completely inactive in control experiments measuring CAT activity in CCR5 ϩ CD4 ϩ Cf2Th cells infected with the dual tropic 89.6 HIV-1 strain (Fig.  7b). These data are in agreement with other results described   3. a and b, CXCR4 binding of peptides DV1 (Ⅺ) and V1 (छ) as well as controls vMIP-II (q) and SDF-1␣ (OE) as characterized by anti-CXCR4 mAb 12G5 competitive binding assay (a) and by 125 I-SDF-1␣ competitive binding assay (b). c, CCR5 binding of peptides DV1 (Ⅺ) and V1 (छ) as well as control vMIP-II (q) as characterized by 125 I-MIP-1␤ competitive binding assay.
above and demonstrate that DV1 peptide is a selective inhibitor of HIV-1 co-receptor function of CXCR4 but not CCR5. Note that two different cells (activated PBMC and CCR5 ϩ CD4 ϩ Cf2Th cells) were used to test the inhibition by the peptide of HIV-1 entry via CXCR4 and CCR5, respectively. Activated PBMC were used because they are the most relevant target cells for HIV-1 infection in vivo. Because activated PBMC express both CXCR4 and CCR5, we used an X4 virus, which utilizes only CXCR4 as a coreceptor. To demonstrate that entry of the virus is not inhibited by the peptide when CCR5 is used as a coreceptor, one needs to switch both target cells and virus. For this reason, we used Cf2Th cells that express CD4 and CCR5 and can be utilized by an R5 X4 virus that uses both CCR5 and CXCR4.
The ability of DV1 peptide to block HIV-1 replication was also confirmed in another assay measuring the production of viral p24 antigen in MT-4 cells after being infected with HIV-1 isolated from the supernatant of H9/III B cells. DV1 strongly . c, CXCR4 binding of peptides DS1 (Ⅺ) and S1 (q) derived from the N terminus of SDF-1␣. CXCR4 binding was determined by anti-CXCR4 mAb 12G5 competitive binding assay. inhibited p24 production as compared with control mutant DV1-L1A (Fig. 7c). Consistent with this observation, the cell viability assay showed that DV1 protected MT-4 cells from the cytopathic effects of HIV-1 infection (data not shown). It was noted that, when V1 peptide was used for comparison, it was active during the initial period and gradually lost effect over time, whereas DV1 peptide maintained its activity throughout the course of experiments. This is consistent with the higher biological stability of DV1 peptide as described above and supports the notion that peptides of D-amino acids have advantages for therapeutic application.

DISCUSSION
CXCR4 plays an essential role in many physiological functions as the receptor for chemokines such as SDF-1␣. In addition, CXCR4 is involved in the pathogenesis of HIV-1 infection by serving as the co-receptor for the cellular entry of T and dual tropic virus strains. CXCR4 belongs to the GPCR superfamily, which contains a large number of membrane proteins that are important therapeutic targets but whose structures and interactions with their ligands need to be further elucidated. In this study, we explored the mechanism of CXCR4-ligand recognition with synthetic chemokine-derived peptides containing Damino acids in an attempt to gain insights into the structure and function of CXCR4 and design novel inhibitors of HIV-1 entry. Through structure and function analyses of a series of D-peptides, D-and L-hybrid peptides, and mutant analogs derived from vMIP-II and SDF-1␣, we found that the CXCR4 binding surface is tolerant of changes in peptide stereochemistry. These peptides containing some or all D-amino acids display significantly different topologies on side chains from Lcounterparts and yet retain or even enhance their highly specific interactions with the receptor. This seems to be contrary to the notion that, in general, receptor-ligand interface is highly sensitive to changes in stereo configurations of ligands.
ALX40 -4C, a peptide of nine D-Arg residues, is known to bind CXCR4 (45). However, this peptide, like other highly positively charged molecules such as T-22 (46) and AMD3100 (47), most likely acts through overall charge-charge electrostatic interactions (48). This is different from the D-and Lpeptides reported here for which specific interactions of uncharged hydrophobic side chains of the peptides with the receptor are clearly important. D-Peptides of same L-peptide sequences such as DV1 described here are different from Dpeptides of reversed L-peptide sequences or so-called "retroinverso" peptides. Retro-inverso peptides can have the identical side chain topology but opposite main chain direction as compared with L-counterparts (49). In contrast, D-peptides such as DV1 retain the same main chain direction but adopt a distinct side chain topology (Fig. 2). Normally, such a change in the stereochemistry of side chains should disrupt a stereospecific ligand-receptor interface. There are only a limited number of cases reported in the literature where both D-and L-peptide enantiomers recognize the same receptors, such as L-and Dpeptide ligands for calmodulin (50,51), L-and D-peptides derived from type IV collagen that bind the ␣ 3 ␤ 1 integrin (52,53), and L-and D-peptide ligands for the co-chaperone DnaJ (Hsp40) (54). However, to our knowledge of current literature, CXCR4 binding by D-and L-peptide ligands derived from natural chemokines as reported here is unprecedented for chemokine receptors and membrane proteins of the GPCR family. Whether this is a unique property of CXCR4 or a similar phenomenon can be found in other chemokine receptors or GPCRs remains to be seen.
Our results provide surprising new insights into the biochemical nature of the CXCR4 binding surface and should prompt further studies of the mechanism for the sterical tolerance of CXCR4-peptide interactions. More studies are needed to characterize the detailed binding sites of D-and L-peptides on the receptor and structural basis of D-peptide binding. Such studies include site-directed mutations of potential receptor residues involved in the binding of these peptide ligands and molecular modeling of peptide-CXCR4 interaction using the recently proposed models of CXCR4 structure (27,28). In addition, it will be important to extend this present study of D-peptides to modifications of full-length chemokine proteins and examine the structure-function relationship of their interactions with the receptor. We are currently exploring these and other related issues. Finally, it is of interest to examine whether the observation reported here may have any implication for understanding CXCR4 interaction with HIV-1. One may speculate that the flexibility of the CXCR4-ligand interface might be a feature of CXCR4 that HIV-1 capitalizes on. This might allow sequence and/or conformational variation to occur in the V3 loop, one of the gp120 components thought to interact with CXCR4. If the resulting changes in the gp120 V3 loop are accommodated by the flexible binding surface of CXCR4 but not that of an antibody, the virus can then evade neutralizing antibodies although maintaining a high affinity for the coreceptor that is necessary for viral entry into the cell. These hypotheses as raised by the present study merit further investigations.
The understanding of the mechanism and structural basis for the remarkable property of the CXCR4 binding surface described here will have important implications for the design of novel ligands. Receptors and proteins mediating the HIV-1 entry process have become important targets for drug discovery, as agents blocking the virus from entering cells will serve as a new type of therapeutic that augments the arsenal of current drugs targeting the replication of the virus inside the cell. Small molecule inhibitors of HIV-1 entry via two principal co-receptors CXCR4 and CCR5 have been reported (32,(45)(46)(47)55). In addition, HIV-1 entry inhibitors have also been developed to target the HIV-1 envelope glycoprotein gp41 (56). In this study, by taking advantage of the stability of unnatural D-amino acids, DV1 peptide shows significant anti-HIV-1 activity in biological conditions and thus may be a desirable lead for potential therapeutic development. As shown by this D-peptide, the apparent lack of stereospecificity of the CXCR4 binding surface as observed here might be exploited for the design of other highly stable and potent ligand molecules with novel topological features.