Cutting Edge: Human β Defensin 3—A Novel Antagonist of the HIV-1 Coreceptor CXCR4¹

Recent findings are shedding new light into the importance of host defense peptides in immunity. With the discoveries in mucosal epithelium of human β-defensins (hBDs),³ ∼4- to 5-kDa cationic, amphipathic peptides with antimicrobial and immunosurveillance properties (1, 2, 3), there is reason to propose that these innate defense peptides may function not only to protect the host against microbial pathogens at the mucosal barrier, but also to marshal adaptive host immune responses to both prevent and clear infection. Our previous studies demonstrated that HIV-1 induces hBD-2 and -3 expression in human oral epithelial cells and that these peptides block HIV-1 replication via a direct interaction with virions and through modulation of CXCR4, the coreceptor used by HIV-1 X4 isolates to infect target cells (4). The present study explored the mechanism of the interaction between hBD-3 and CXCR4. We found that this peptide could compete with stromal-derived factor 1 (SDF-1), the CXCR4 ligand, for binding to cells and also could induce internalization of CXCR4 in immunocompetent cells without promoting calcium mobilization, activation of ERK-1/2 MAPK, or chemotaxis. This activity was specific for CXCR4 since there was no inhibition of calcium mobilization as a result of RANTES, fMLP, ATP, or carbachol binding to their respective receptors.

CEM X4/R5 cells (human T cell line expressing CXCR4 and CCR5) and THP-1 cells (human monocytic leukemia cell line) were cultured and maintained as described previously (4). To induce fMLP receptor expression, THP-1 cells were differentiated in RPMI 1640 medium supplemented with 300 μM dibutyryl cyclic AMP (Sigma-Aldrich) for 2 days.

CD4⁺ cells were isolated from human PBMCs (Ficoll-Paque sedimentation; Amersham Biosciences) by negative selection using MACS and the CD4⁺ T cell isolation kit II (Miltenyi Biotec). The resulting purity of CD4⁺ cells was >90%, as determined by flow cytometry.

rhBD-3 was produced by cloning hBD-3 cDNA into pET-30c (from J. Härder and J. Schröder, Kiel University, Germany) (Novagen) as described previously (5). LPS presence was deemed negligible by Charles River Laboratories, i.e., <5 EU/mg rhBD-3 and representing <0.01 ng/ml in our assay system.

CEM X4/R5 cells were washed and resuspended in balanced salt solution (BSS) containing 140 mM NaCl, 5 mM KCl, 1.5 mM CaCl₂, 1 mM MgCl₂, 10 mM d-glucose, 1 mg/ml BSA, and 20 mM NaHEPES (pH 7.5). One million cells in a final volume of 300 μl were placed in each tube along with 0.1 nM ¹²⁵I-labeled SDF-1 or RANTES (PerkinElmer NEN). Unlabeled SDF-1 or RANTES (1 μg/ml for each) or various concentrations of hBD-3 were added simultaneously. Only ¹²⁵I labeled SDF-1 or ¹²⁵I-labeled RANTES was added to control tubes. Cells were incubated at 37°C for 30 min, or on ice for 4 h, and then washed three times with BSS. Cellular binding of ¹²⁵I-labeled SDF-1 or ¹²⁵I-labeled RANTES was determined by gamma counter readings. All experiments were performed in triplicate.

CEM and THP-1 cells were washed in PBS and resuspended at a concentration of ∼10⁶ cells/ml in BSS. The cell suspension was supplemented with 1 μM fura 2-AM (Molecular Probes) and incubated at 37°C for 45 min. Cells were pelleted, washed once, and resuspended in fresh BSS to a density of 10⁶ cells/ml. Cytosolic Ca²⁺ levels in 1.5 ml of stirred cell suspensions were assayed fluorometrically at 37°C using equipment and calibration protocols that have been described previously (6). The cells were treated with hBD-3, SDF-1 (R&D Systems), RANTES (R&D Systems), carbachol, ATP, or fMLP (all from Sigma-Aldrich) that were added as 0.5- to 15-μl aliquots from concentrated stocks. Cells were then permeabilized with 50 μg/ml digitonin (Sigma-Aldrich) to facilitate calibration of fura-2 fluorescence as a function of extracellular Ca²⁺ levels.

CEM X4/R5 cells were collected, washed twice with PBS, resuspended in medium containing 0.5% FBS, with or without 20 μg/ml rhBD-3, followed by incubation at 37°C in 5% CO₂ for 2 h. Cells were treated with FACS/Perm (BD Pharmingen) at room temperature, 10 min, and then washed three times with PBS. Cells were stained with PE-labeled anti-CXCR4 mAb (clone 12G5; BD Pharmingen) at room temperature, 90 min, followed by washing three times with PBS. Samples were observed using a dual scanning confocal microscope system (Zeiss LSM 510; Oberkochem) and analyzed with the Zeiss LSM 5 Image Browser.

CEM X4/R5 cells or CD4⁺ PBMCs were resuspended at a concentration of 10⁷ cells/ml in BSS. Aliquots (500 μl) were pretreated with either 10 nM SDF-1, hBD-3 (20 μg/ml; 3.9 μM), hBD-3 before SDF-1, carbachol (10 μM), or phorbol myristate acetate (100 nM PMA; Sigma-Aldrich) for 3 min. Reactions were terminated by transfer to an ice bath and centrifugation to pellet the cells. Following aspiration and ice-cold PBS washing, the pelleted cells were routinely disrupted in 200 μl of lysis buffer containing 25 mM HEPES (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1 mM β-glycerol phosphate, 2 mM sodium pyrophosphate, 5 mM EDTA, 5 mM EGTA, 1 mM sodium orthovanadate, 2 mM PMSF, 0.1% aprotinin, and 10 μg/ml leupeptin. After 20 min on ice, the lysate was centrifuged at 10,000 × g for 10 min at 4°C; 150 μl of the supernatant was denatured by addition of 4× Laemmli sample buffer (7) and boiling for 5 min. The extracted proteins were separated by SDS-PAGE on 12% gels (Bio-Rad) and transferred to 0.45-μm Immobilon-P polyvinylidene difluoride membranes (Millipore). The blots were serially probed with anti-phospho-ERK1/2 (Thr²⁰²/Tyr²⁰⁴) mouse mAbs and then anti-total ERK1/2 rabbit polyclonal Abs (Cell Signaling Technology). Bound primary Abs were visualized with appropriate peroxidase-coupled secondary Abs (Santa Cruz Biotechnology) using Enhanced Luminol Chemiluminescence Reagent (Amersham Biosciences).

Chemotaxis assays were performed as described by Chernock et al. (8). Briefly, Jurkat cells (clone E6–1; human T cell line; ATCC TIB-152) or activated primary T lymphoblasts (from Dr. A. Levine, Department of Gastroenterology, Case Western Reserve University, Cleveland, OH; prepared as previously described (9)) were washed and resuspended in RPMI 1640 with 0.5% FBS at a concentration of 1 × 10⁷ cells/ml. A 100-μl aliquot of each cell type was added into respective Transwell inserts of chemotaxis chambers (Transwell Permeable Supports; Corning), where the lower compartment contained 600 μl of RPMI 1640 and 0.5% FBS. SDF-1 (100 ng/ml) and hBD-3 (0–40 μg/ml) were added into each lower compartment except in control chambers. After incubation at 37°C for 3 h, inserts were removed, and cells that migrated into the lower chamber were counted.

We previously reported that interaction of hBDs with the CEM lymphoid cell line resulted in decreased surface CXCR4 expression and concomitant internalization of the defensin peptides (4). Given these observations, we surmised that CXCR4 receptors were also internalized in response to hBD exposure. Herein, we show that hBD-3 treatment of CEM cells results in CXCR4 internalization as detected by confocal microscopic analysis of permeabilized cells stained with anti-CXCR4 Abs (Fig. 1,B). This is particularly apparent in confocal views of cells appearing in the Z axis (arrows), where hBD-3 treated cells show dense internal staining. Untreated cells demonstrate CXCR4 surface localization (Fig. 1 A). These results are consistent with our previous FACS findings showing a significant reduction in CXCR4 surface expression when unstimulated human PBMCs or CEM cells were incubated with hBD-3 (4).

To elucidate further the interaction of hBD-3 with CXCR4, we conducted a competition assay to assess the ability of hBD-3 to inhibit the binding of the CXCR4 ligand SDF-1 to CEM cells expressing both CXCR4 and CCR5. hBD-3 inhibited the binding of ¹²⁵I-labeled SDF-1 (Fig. 2,A) but not the binding of ¹²⁵I RANTES (Fig. 2,C) in a dose-dependent manner (Fig. 2,A). When the experiment was conducted at 4°C to inhibit internalization of CXCR4, inhibition of ¹²⁵I-SDF-1 binding was also seen (Fig. 2 B), indicating that hBD-3 and SDF-1 compete for binding to common or proximate cell surface sites.

Since hBD-3 promoted internalization of CXCR4 and blocked SDF-1 binding, we anticipated that hBD-3 would also elicit calcium mobilization by activating the G protein-effector cascades regulated by agonist-occupied CXCR4. To our surprise, at concentrations ranging from 2.5 to 20 μg/ml (0.48–3.9 μM), hBD-3 did not induce a calcium mobilization response in CEM cells but rather blocked the calcium flux induced by SDF-1 in a dose-dependent manner (Fig. 3, A–C) as soon as 1 min after application (data not shown). Importantly, hBD-3 could not block the calcium mobilization induced by carbachol (Fig. 3, A–C), indicating that this defensin had no effect on the M1-muscarinic receptors also expressed in CEM cells. This result prompted us to test other ligands to assess hBD-3 specificity in inhibiting calcium signaling through other G protein-coupled receptors (GPCRs). We found that hBD-3 had no effect on RANTES-induced calcium signaling in CEM cells (Fig. 3,D) or human PBMCs (data not shown), or on calcium signaling elicited by fMLP (Fig. 3,E) or ATP (data not shown) in the THP-1 monocytic cell line. Similar to its actions in CEM cells, hBD-3 inhibited SDF-1-induced calcium mobilization in these THP-1 cells (Fig. 3,E) and in peripheral blood human T lymphocytes (Fig. 3, F and G). Thus, hBD-3 can repress SDF-1-dependent calcium signaling in both lymphoid and myeloid cell lines and in primary human lymphocytes.

In addition to mobilizing calcium, signal transduction through SDF-1/CXCR4 binding coordinates activation of MAPK pathways, such as ERK (ERK-1/2) (10, 11). Therefore, to define further the antagonism of SDF-1/CXCR4 interactions by hBD-3, we assessed the ability of hBD-3 to inhibit SDF-1-induced phosphorylation of the ERK-1/2 MAPK. We found that hBD-3 did not induce ERK-1 or ERK-2 phosphorylation but abrogated SDF-1-dependent ERK1/2 activation in both CEM X4/R5 cells and in freshly isolated CD4⁺ PBMCs (Fig. 4).

Finally, since SDF-1/CXCR4-dependent lymphocyte locomotion is well established (12) and since our experiments up to this point indicated that hBD-3 acts antagonistically toward CXCR4, we surmised that it would block SDF-1-induced T cell chemotaxis. Our chemotaxis assays demonstrated that hBD3 did indeed reduce SDF-1-dependent chemotaxis in a dose-dependent manner in both Jurkat cells and activated human T cells (Fig. 5, A and B).

The experiments reported here demonstrate that rhBD-3 induces internalization of CXCR4 without promoting calcium mobilization or activation of ERK-1/2 MAPK. There is at least some specificity for CXCR4 because hBD-3 does not inhibit calcium mobilization as a result of RANTES, fMLP, ATP, or carbachol binding to their receptors. Since these ligands induce calcium flux via either a Gi (SDF-1, RANTES, and fMLP) or Gq (carbachol and ATP)-coupled receptor pathway (10, 11) and since hBD-3 also blocks SDF-1 binding, the data suggest that hBD-3-mediated inhibition of calcium flux occurs at the level of the CXCR4 receptor and not at the G protein level.

The interaction of agonistic ligands with GPCRs induces conformational changes that coordinately elicit activation of the downstream G protein-dependent signaling cascades and also promote desensitization/internalization of the activated receptor. The canonical model for GPCR internalization involves phosphorylation of the receptors, association of arrestin-family adapters with the phosphorylated receptors, targeting of the GPCR-arrestin complexes to clathrin-coated pits, and, finally, dynamin-dependent internalization of the complexes. Previous studies have indicated that SDF-1 induces internalization of CXCR4 via these classical arrestin- and dynamin-dependent mechanisms (13, 14). Although antagonistic ligands for most GPCRs do not elicit either coupling to G proteins (by definition) or desensitization/internalization, a small but increasing number of GPCRs have been shown to internalize upon binding antagonists. These include the CCK receptor for cholecystokinin (15), the 5HT_2A receptor for serotonin (16), the NPY₁ receptor for neuropeptide Y (17), and the CRF₁ receptor for corticotrophin releasing factor (18). Our results indicate that hBD-3 also induces CXCR4 internalization despite its actions as an antagonist of SDF-1 binding to, and activation of, CXCR4. The mechanistic relationships among ligand binding, functional activation, and CXCR4 internalization remain incompletely defined, however. Early structure-function analyses of SDF-1 interaction with CXCR4 indicated that N-terminal residues 12 through 17 of SDF-1 are required for selective and high-affinity binding of the chemokine to the N-terminal tail of CXCR4, whereas the extreme N-terminal amino acids (Lys-Pro) of SDF-1 are required for functional activation of CXCR4 as a consequence of interaction with the receptor’s second extracellular loop (19, 20). Cai et al. (21) reported recently that rSDF-1 lacking the C-terminal α-helical domain acts as an antagonist of SDF-1-induced CXCR4 signal transduction but can induce CXCR4 internalization. Other SDF-1-derived peptides can induce potent signal transduction through activation of CXCR4 but elicit little internalization of the receptor (22). Taken together, these data indicate that CXCR4 may conceivably assume multiple conformations that can independently regulate G protein coupling vs receptor internalization.

SDF-1/CXCR4 interactions have been implicated in a number of vital biological functions, including hemopoiesis, neurogenesis, angiogenesis, cardiogenesis, and immune cell trafficking (12, 23, 24, 25). Our findings indicate that hBD-3 may regulate some of these activities at selected sites such as those proximate to mucosal surfaces where the concentration of hBDs is high (26, 27). These findings also indicate that calcium influx and ERK1/2 phosphorylation are not necessary to induce CXCR4 internalization and thus implicate the likely importance of other as yet undescribed mechanisms in trafficking of CXCR4. This novel finding also may lead to a better understanding of how to inhibit the SDF-1/CXCR4 axis by use of endogenous agents such as hBD-3 or its derivatives and also may provide a new direction for drug design.

The authors have no financial conflict of interest.