β‑Arrestin-Mediated Activation of Chemokine Receptor CCR7 by Human β‑Defensin‑3 Drives Oral Squamous Cell Carcinoma Progression

Abstract

Human β-defensin-3 (hBD3), an epithelial-derived antimicrobial peptide, was recently identified as a regulator of oral squamous cell carcinoma (OSCC) progression through its interaction with the chemokine receptor CCR7a key mediator of immune cell trafficking and tumor metastasis. In this study, we aimed to clarify the structural and mechanistic basis of hBD3-induced CCR7 activation by employing molecular docking combined with atomic-level molecular dynamics simulations. Binding of hBD3 induced marked conformational rearrangements in CCR7, enhancing β-arrestin-2 recruitment and initiating noncanonical intracellular signaling cascades. Comparative studies with the endogenous CCR7 ligands CCL19 and CCL21 revealed that hBD3 exhibited dual functionality: intact disulfide bonds confer agonistic properties by promoting receptor activation, whereas the disruption of these bonds converts the peptide into an antagonist. Moreover, distinct phosphorylation signatures within the serine/threonine clusters of the C-terminal tail of CCR7 were found to govern downstream signaling outcomes. Collectively, these results reveal a mechanism reliant on both disulfide bond stability and phosphorylation patterns through which hBD3 differentially regulates CCR7 function, providing novel insights into ligand-specific receptor modulation and potential therapeutic strategies for CCR7-mediated malignancies.

Introduction

G-protein-coupled receptors (GPCRs) constitute the largest family of membrane receptors in eukaryotes, orchestrating a wide range of physiological processes, including neurotransmission, immune regulation, and chemotaxis. , These receptors share a conserved framework of seven transmembrane (TM) helices and translate extracellular cues into intracellular responses through three principal modes: G-protein-dependent signaling, β-arrestin-dependent signaling, and integrated responses involving both. Ligand binding induces conformational shifts that enable GDP–GTP exchange in the Gα subunit, initiating downstream effector activation and second messenger production. In parallel, receptor phosphorylation by G-protein-coupled receptor kinases promotes β-arrestin binding, which not only drives receptor desensitization and internalization but also triggers distinct and sustained signaling cascades. β-Arrestin-directed pathways contribute to inflammation control, cell survival, and transcriptional regulation and have been increasingly linked to cancer progression. Within the GPCR superfamily, chemokine receptors form a specialized group responsible for directing immune cell trafficking. Among these, CCR7 is pivotal for lymphocyte homing and immune surveillance and is activated by two endogenous ligands, CCL19 and CCL21. Although CCL19 and CCL216 target the same receptor, they elicit divergent signaling outcomes: CCL19 activates both G-protein and β-arrestin pathways and drives receptor internalization, while CCL21 predominantly engages G-protein signaling with limited arrestin recruitment, thereby sustaining chemotactic responses. , This ligand-biased signaling fine-tunes immune responses and carries important implications for tumor biology. The CCL19/CCL21–CCR7 axis plays a crucial role in cancer progression by shaping immune cell migration and tumor dissemination. CCR7 is frequently overexpressed in oral squamous cell carcinoma (OSCC), where it promotes lymphatic dissemination and is associated with poor clinical outcomes. Similarly, in breast cancer, CCR7 expression is linked to enhanced lymphatic invasion and poor prognosis, as tumor cells exploit CCL21-mediated chemotaxis to colonize lymphatic tissues and establish distant metastases. Consequently, therapeutic strategies targeting CCR7 and its ligands hold promise for curbing tumor progression and improving clinical management.

Recent studies have shown that CCR7 activation extends beyond the classical chemokines. Human β-defensin-3 (hBD3), a cationic antimicrobial peptide secreted by epithelial and immune cells, has been identified as a noncanonical CCR7 ligand with strong immunomodulatory and tumor-promoting properties. The pro-tumorigenic effects of hBD3, mediated through CCR7, indicate that hBD3 enhances tumorigenesis by driving immune cell recruitment and shaping a microenvironment conducive to cancer cell survival and spread. hBD3 is highly expressed in OSCC, particularly in squamous cell carcinoma of the head and neck, and supports a pro-inflammatory tumor microenvironment by recruiting dendritic cells and tumor-associated macrophages. Notably, hBD3 induces CCR7 upregulation independently of the classical G-protein-coupled signaling cascade, , implying that hBD3 may activate alternative signaling pathways, although the precise mechanisms remain unclear. Additionally, hBD3 interacts with CXCR4 as a competitive antagonist, blocking stromal-derived factor 1-mediated responses such as calcium mobilization and ERK1/2 phosphorylation. This antagonistic effect requires intact disulfide bonds in hBD3; partial or complete disruption of these bonds diminishes its antagonistic activity and shifts the peptide toward agonist behavior.

Arrestins are versatile regulators of GPCRs, initially recognized for their role in desensitizing activated receptors. However, subsequent research has shown that arrestins also promote receptor endocytosis and participate in downstream signaling pathways. In GPCRs, phosphorylation sites are concentrated in the third intracellular loop (ICL3) and the C-terminal tail, which serve as the primary binding regions for arrestin. These sites typically form clusters of serine (Ser) and threonine (Thr) residues, termed S/T clusters, which are essential for stabilizing GPCR–arrestin interactions. The number and arrangement of these sites differ among GPCRs, reflecting receptor-specific regulatory mechanisms. Importantly, the role of phosphorylation sites depends not only on their abundance but also on their precise location. Distinct phosphorylation patterns modulate both the affinity and functional outcomes of GPCR–arrestin binding. Depending on the specific combination of sites, some enhance arrestin recruitment and activation, whereas others inhibit these events. Studies on the vasopressin V2 receptor demonstrated that activated β-arrestin1 undergoes marked conformational rearrangements compared with its inactive form. A defining feature is a 20° rotational twist between the N- and C-domains around a central axis, indicating a substantial interdomain rearrangement. This structural shift is accompanied by notable adjustments in the N-domain and central loops, which are crucial for β-arrestin activation and receptor association. The major structural elements in this process include the finger loop, middle loop, and lariat loop, all of which undergo structural changes upon activation. In the inactive state, these loops display flexibility; however, upon activation, they adopt ordered conformations that facilitate efficient receptor engagement.

Despite growing evidence of the role of hBD3 in CCR7 signaling, the molecular basis of this interaction remains only partly understood. hBD3 can upregulate CCR7 expression through NF-κB and promote receptor internalization. Nonetheless, whether hBD3 directly activates CCR7and how this compares structurally and functionally with canonical ligandsremains unresolved. Receptor phosphorylation patterns, especially within the S/T clusters of the CCR7 C-terminal tail, are known to regulate arrestin engagement and downstream signaling; nevertheless, their role in hBD3-mediated pathways has not been fully elucidated. To address these questions, atomic-level molecular dynamics (MD) simulations coupled with essential dynamics sampling (EDS) were employed to investigate how the disulfide bonds and phosphorylation patterns of hBD3 influence its activity and conformational changes. Our findings revealed that wild-type hBD3 (hBD3wt) promotes CCR7 activation, whereas mutant hBD3 (hBD3mu) tends to stabilize CCR7 in an inactive state. The phosphorylation patterns of CCR7 also played a key role in arrestin recruitment. Notably, CCL19 promoted transient receptor activation while sterically hindering G-protein coupling, whereas hBD3 favored prolonged internalization and β-arrestin-biased signaling. Elucidating this noncanonical activation mode provides new insights into the role of chemokine receptor signaling in cancer progression and highlights CCR7 as a promising therapeutic target in CCR7-mediated malignancies.

Materials and Methods

Preparation of Ligand–CCR7 Complexes

The active and inactive conformations of human CCR7 were obtained from the GPCR database (https://gpcrdb.org/) and refined by removing the first 35 disordered N-terminal residues to improve structural stability. Although individual structures of CCR7 (GPCRdb), CCL19 (PDB: 2MP1), and CCL21 (PDB: 2L4N) are available, no experimentally determined complex structures exist. To address this, homology models of CCL19–CCR7 and CCL21–CCR7 complexes were constructed following the homology modeling method proposed by Meiler et al. using the CXCL8–CXCR2 complex (PDB: 6LFO) as a template, which shares 59.6% sequence identity with CCR7 and 36–39% sequence identity with the chemokines (Figure S1). The structure of wild-type human β-defensin 3 (hBD3wt; PDB: 1KJ6) was docked to CCR7 using HADDOCK2.4. Regarding the initial binding structure of hBD3, due to the lack of the structure of hBD3 bound to CCR7, the docking setup was designed based on previously reported findings on hBD3–CXCR4 interactions. Since CCR7, like CXCR4, is a class A GPCR and belongs to a chemokine receptor that shares similar topological and structural features, a comparable binding interface centered around the orthosteric pocket and extracellular regions was considered for CCR7. Among the generated docking poses, Cluster 2 was selected as the representative model based on its lowest Z-score, large buried surface area (2640 Ų), high Fraction of Common Contacts, low interface RMSD, and favorable energy (Figure S8). Thus, the combination of structural, energetic, and contact-based evaluations provides strong support for the selection of Cluster 2 as the initial hBD3wt–CCR7 binding structure for subsequent MD simulations. A mutant version (hBD3mu) was generated by substituting the cysteines forming the disulfide bonds (C11–C40, C18–C33, and C23–C41) with alanines (Figure S2).

Preparation of CCL19/hBD3–CCR7−β-Arrestin-2 Tricomplex Structures and Phosphorylation of CCR7

For the construction of CCL19/hBD3–CCR7−β-arrestin-2 (βarr2) tricomplexes, active CCR7–ligand structures were aligned with the phosphorylated μ-opioid receptor−βarr2 complex (pp-μOR−βarr2) reported by Mafi et al. The final 46 residues of μOR, essential for βarr2 recruitment, were substituted with the CCR7 C-terminal region using MOE2022.02 software, thereby enabling CCR7 to mimic the arrestin-binding function of μOR. The merged model was structurally refined to eliminate steric clashes. Phosphorylation of serine and threonine residues in the C-terminal tail of CCR7 was modeled using the CHARMM–GUI Web server under two schemes: (i) distal phosphorylation at T372–T375 and S377 and (ii) full phosphorylation at S356, S357, S364, S365, T372–T375, and S377. This design was informed by findings from the study by Kohout et al., who highlighted distal phosphorylation as the key determinant of CCL19-mediated receptor phosphorylation.

MD Simulations and EDS

MD simulations for all systems were performed using the CHARMM36 all-atom force field in GROMACS 2020. A homogeneous 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine bilayer was generated via the CHARMM–GUI web server. Two receptor setups were considered: one bound to βarr2 and the other without βarr2. Each system was solvated with TIP3P water molecules, and sodium and chloride ions were added to neutralize the system at 0.15 M. Simulations were run in the NPT ensemble, maintaining a temperature of 310 K and a pressure of 1 bar using a velocity-rescaling thermostat. Semi-isotropic pressure coupling was applied with a coupling constant of 0.1 ps and a compressibility of 4.5 × 10^–5 bar^–1 in both the xy-plane and z-dimension. Long-range electrostatic interactions were computed using the particle-mesh Ewald method with a Fourier spacing of 0.12 nm and fourth-order interpolation. The equilibration process was conducted in two stages. − First, the system temperature was gradually increased from 0 to 100 K, then to 200 K, and finally to 310 K, with each step lasting 500 ps. During this stage, complex structures were fully restrained using a force constant of 1000 kJ mol^–1 nm^–2. In the second stage, restraints at 310 K were progressively reduced from 1000 kJ mol^–1 nm^–2 to 500, 300, and finally 100 kJ mol^–1 nm^–2. Each step was run for 2.0 ns. After equilibration, all restraints were removed and production simulations were performed.

To explore broader conformational landscapes, EDS was employed. EDS introduces flooding potentials derived from quasiharmonic approximations of collective motions, destabilizing local energy minima to accelerate conformational transitions while preserving native pathways. The flooding procedure proceeds in two steps: first, the free-energy surface of the system is approximated quasi-harmonically; second, a multivariate flooding potential (V_fl) is constructed from this approximation, raising the base of the educt energy well while leaving higher-energy regions, particularly the surrounding barriers, largely unaffected. A summary of all simulation systems is provided in Table S1.

Water Network Analysis

We employed the C-Graphs tool to investigate water-mediated signal transduction pathways. , This analytical approach was used to map hydrogen bond networks connecting the ligand-binding pocket to the arrestin-binding interface through the MD trajectory analysis. Stable, persistent water-mediated connections were identified using the following strict criteria: a minimum hydrogen bond occupancy threshold of 70% and a maximum bond length of 3.5 Å. These criteria ensured that only the most significant and long-lasting water-mediated interactions were retained while transient or weak associations were excluded.

Protein Correlation Network and Community Analysis

Dynamic correlation network analysis was performed using the Bio3D R package to characterize residue interaction communities within the protein system. Correlated atomic fluctuations were quantified by using a dynamic cross-correlation matrix. In the resulting network, residues were represented as nodes with edges weighted by the correlation values between residue pairs. Community detection analysis was then performed to identify clusters of residues exhibiting highly correlated motions, offering insights into functional residue interactions and communication pathways within the receptor.

Energy Landscape Calculations

The free-energy landscape (FEL) is a key framework for assessing protein conformational stability by mapping the distribution of conformational states according to their energy levels. In this study, the FELalso known as the potential of mean forcewas examined with an emphasis on activation-related features. For ligand-bound receptor systems, the analysis focused on the xQxP motif and the tyrosine toggle, whereas for tricomplex systems it addressed arrestin interdomain reorientation of the C-edge loop along with the receptor tyrosine toggle. Free energy was calculated by using the following equations:

For ligand-bound receptor systems:

$$G(\mathrm{xQxP},\mathrm{tyrosine\; toggle})=-k_{\mathrm{B}}T\ln P(\mathrm{xQxP},\mathrm{tyrosine\; toggle})$$

For tricomplex systems:

$$G(\mathrm{C‐edge\; loop},\mathrm{tyrosine\; toggle})=-k_{\mathrm{B}}T\ln P(\mathrm{C‐edge\; loop},\mathrm{tyrosine\; toggle})$$

where P, T, and k _B denote the probability distribution function, absolute temperature, and Boltzmann constant, respectively. The xQxP, C-edge loop rotation angle, and tyrosine toggle switches were tracked over time to derive the probability distributions calculated from the trajectories of each system at 310 K. FEL was calculated using the gmx_sham module of the GROMACS software package.

Additional Analyses

The distances between selected atoms and conserved motifs were calculated by using the gmx_distance tool in the GROMACS 2020.4 package. The rotation of the α-helix in the N-domain of arrestin was analyzed by using gmx_helixorient, while the rotation of the C-edge in the C-domain was quantified with gmx_angle using three reference points: the Cα atom of the CCR7 core (T124), the C-edge loop (R327), and the arrestin core (A122). In this study, root-mean-square deviation (RMSD) was used to evaluate overall conformational variations of the protein Cα atoms. RMSD values were calculated by using the gmx_mpi rms module of GROMACS 2020.4. Root mean square fluctuation (RMSF) is a key metric for assessing the flexibility of specific atoms or residues within a molecular system during a simulation. In this study, the reference structure was the minimized and equilibrated initial receptor. RMSFs of the Cα atoms of arrestin were computed by using the gmx_mpi rmsf module in the GROMACS 2020.4 package.

Results

Comparative Analysis of Ligand Interaction Sites on CCR7

Ligand-binding sites on CCR7 were categorized as CRS1 (N-terminus), CRS1.5 (the interface between the N-terminus and TM domain, or near the ECLs), and CRS2 (orthosteric pocket) (Figure a, b). Microsecond-scale simulations showed that both CCL21 and hBD3wt engage all three regions, including deep insertion into CRS2, consistent with agonist-like activity (Figure c). By contrast, CCL19 primarily interacted with CRS1 and CRS1.5, with minimal interaction at the orthosteric site (Figure c), indicative of a more superficial binding mode. The presence of disulfide bonds in hBD3wt stabilized its interactions across CRS1 to CRS2. This binding pattern is consistent with previous findings on hBD3wt interactions with the CXCR4 receptor, which also involved multiple receptor regions, with the N-terminal and extracellular loop (ECL) regions of CXCR4 being identified as the primary binding sites for hBD3 (Figure d). The selected docking region is consistent with the experimentally supported activation mechanisms observed in related GPCR systems. For an initial binding structure comparison between hBD3wt and two distinct receptors, the RMSD value of the superposition is 5.81 Å. By contrast, disruption of disulfide bonds in hBD3mu increased peptide flexibility, broadening electrostatic contacts with CRS1.5 and CRS2 residues (E205^ECL2, E215^ECL2, E218^5.33, E299^7.26), while reducing hydrogen bonding with the CCR7 N-terminus (Figure c, e), a hallmark of its antagonistic profile. Distance maps confirmed that most ligands maintain interactions with the N-terminus, ECLs, and core TM regions of CCR7, particularly those of TM6 and TM7 (Table S2). The RMSD profiles for all systems over the course of the simulations are shown in Figures S3–S4. The receptor Cα RMSD exhibited a stable plateau throughout the production window, indicating that the receptor Cα conformation had converged (Figure S3). In the active CCR7 conformation, the CCL21–CCR7 system displayed Cα RMSD values of 0.5–0.6 nm, while the hBD3wt–CCR7active system exhibited slightly higher values ranging from 0.65 to 1.0 nm (Figure S3A). The elevated RMSD observed for hBD3wt relative to that of CCL21 suggests larger structural fluctuations and more pronounced conformational shifts in the receptor. Likewise, the ligand RMSD relative to the receptor binding pocket stabilized after approximately 800 ns across all simulations, demonstrating that ligand positioning had reached equilibrium (Figure S4).

1.

Structural and interaction profiles of different ligands bound to CCR7. (a) Side view of the CCR7 structural model. (b) Visualization of the chemokine recognition sites (CRS): CRS1 (light red), CRS1.5 (orange), and CRS2 (dark yellow). (c) Binding modes and principal interactions of selected ligands with CCR7: CCR7–CCL21 (pink–red), CCR7–hBD3wt (yellow–orange), CCR7–hBD3mu (lime–green), and CCR7–CCL19 (cyan–blue). (d) Structural overlay of the final frames of CCR7–hBD3wt (yellow–orange) and CXCR4–hBD3wt (purple–magenta). (e) Combined distance map of distinct ligands bound to CCR7. The X-axis and Y-axis represent CCR7 residue numbers and different ligands, respectively. The minimum-distance heatmap is shown on the right, with 0 nm indicated in red, gradually transitioning to blue as residue-pair distances reach or exceed 1 nm.

RMSD analyses were conducted to evaluate the effect of disulfide bond disruption in hBD3 (Figure S7). The RMSD of hBD3wt remained consistently stable at ∼0.5 nm, reflecting the preserved structural integrity maintained by its disulfide bonds. By contrast, hBD3mu, lacking these bonds, displayed markedly higher RMSD values that fluctuated between 1.0 and 2.0 nm, indicative of enhanced structural flexibility. The steep increase in the RMSD for hBD3mu within the first 200 ns indicates a rapid conformational rearrangement upon disulfide bond loss. For the mutant hBD3 bound to CCR7, the initial binding structure was the same as that of wild hBD3 bound to CCR7, but the two types of hBD3 were obviously different after long-term MD simulations. Superposition analysis further corroborated these dynamics, showing that hBD3mu adopts a more expanded and less ordered structure compared with hBD3wt (Figure S6). A comparison of binding-site orientation with mutant hBD3 bound to active CXCR4 (Figure S11) further highlighted receptor-specific behaviors, where the mutant form exhibits an antagonist-like interaction in CCR7 but may off-target to CXCR4 in a manner consistent with an agonist-like response. Together, these results underscore the critical role of disulfide bonds in maintaining the compact, stable structure of hBD3wt, whereas their absence in hBD3mu leads to pronounced structural deviations and increased dynamic flexibility throughout the simulation. Meanwhile, in all arrestin-bound tricomplex systems, the Cα RMSD profiles of the whole protein complexes stabilized between ∼0.7 and 1.0 nm (Figure S5), indicating that the receptor–arrestin–ligand assemblies remained structurally stable throughout the simulations. These results confirm that the tricomplex configurations were well equilibrated and suitable for further conformational and energetic analyses.

Conformational Dynamics in the Process of Arrestin Activation

Arrestins adopt a conserved fold comprising two concave β-sheet-rich domains (N- and C-domains) connected by a central crest containing the finger, middle, lariat, and C loops , (Figure a, b). In the arrestin-bound tricomplex systems, all models displayed comparable RMSD values of CCR7 ranging from 0.6 to 0.8 nm (Figure S3B). Binding to a phosphorylated GPCR C-terminus induces structural rearrangements that convert arrestin from its basal state to its active receptor-bound state. This transition involves interdomain rotation and crest reorganization, allowing arrestin to engage the receptor core while sterically blocking G-protein coupling. In CCR7–arrestin complexes, binding is mediated by four major interfaces. First, the arrestin finger loop adopts an α-helical conformation, interacting with TM3, TM5, and ICL2 of CCR7, consistent with its role as a key activation sensor (Figure c). Second, disruption of the polar corestabilized in the inactive state by three negatively charged residues (D26, D290, D297) and two positively charged residues (R169, R393)is facilitated by lariat loop rearrangement. In the hBD3–CCR7fullP−βarr2 complex, K288 further stabilizes the interaction by forming a hydrogen bond with pS356 and pS357 of CCR7 (Figure d). Third, the arrestin back loop residue L311 establishes a stabilizing contact with the ICL3 residue R254 of CCR7 (Figure e). Fourth, the middle loop interacts with ICL2, contributing to the receptor core engagement (Figure f). RMSF analysis additionally confirms pronounced flexibility within central crest loops, highlighting dynamic rearrangements during arrestin activation (Figure g).

2.

Global conformational changes of arrestin upon activation. (a) Functional domains of arrestin. (b) Superposition of the inactive βarr2 crystal structure (PDB: 3P2D, gray) with the final molecular dynamics (MD)-derived active βarr2 conformation, emphasizing the central crest region. (c) The finger loop (green) adopts an α-helical conformation upon activation, engaging the cavity between TM3 and TM5 of CCR7. (d) A conserved lysine (K288) in the lariat loop (cyan) directly interacts with the phosphorylated C-terminal tail of CCR7. (e) The back loop of βarr2 (blue) interacts with the ICL3 of CCR7. (f) The middle loop of βarr2 (magenta) interacts with ICL2 of CCR7. (g) Root mean square fluctuation (RMSF) plots of four arrestin-bound tricomplex systems, highlighting βarr2 functional loops undergoing major conformational changes: the finger loop (light green), lariat loop (light cyan), back loop (light blue), and middle loop (light magenta).

Distinct CCR7 Activation Mechanisms Mediated by Ligand-Specific Molecular Switch Dynamics

MD simulations of ligand–CCR7–arrestin tricomplex systems revealed that CCR7 activation is governed by coordinated transitions across key microswitches (Figure a). Notably, the canonical ionic lock formed between R3.50 and D/E6.30, which was originally described in rhodopsin and β-adrenergic receptors, is not universally conserved among class A GPCRs. In CCR7, position 6.30 is occupied by an arginine rather than a negatively charged residue, which precludes the formation of the classical R3.50–D/E6.30 interaction. Consistently, analysis of the distance between R3.50 and the nearest acidic residue, E6.29, revealed no significant differences between the active and inactive receptor models (Figures S15 and S16). This indicates that CCR7 activation is not governed by a canonical ionic-lock mechanism. Instead, the outward displacement of TM6 on the cytoplasmic sidea hallmark of GPCR activationwas clearly observed in the CCR7^CCL21 and CCR7^hBD3wt complexes, where TM3–TM6 distances closely matched the active reference value of 1.732 nm (Figure b), followed by the disruption of salt bridge formed between TM3 DRY motif (D153^3.49) and TM6 (K260^6.32), with the distance extending ∼7–10 Å compared to the inactive reference value (1.569 nm) (Figure S14). This outward shift also influenced TM6–TM7 spacing, with active systems showing greater separation consistent with cytoplasmic opening. These conformational changes were further stabilized by interactions within the conserved tyrosine toggle (Y243^5.58–Y326^7.53), which showed enhanced contact in the agonist-bound complexes. By contrast, CCR7^hBD3mu preserved intact salt bridge formation, TM3–TM6, TM6–TM7, and tyrosine toggle geometries, with peak distances corresponding to the inactive reference values of 1.396, 0.778, and 1.903 nm, respectively (Figure b).

3.

Molecular switch motifs in the active structure of chemokine receptor CCR7. (a) Locations of conserved microswitch residues across the seven transmembrane (TM) helices of CCR7, annotated using the Ballesteros–Weinstein numbering system. (b) Ligand-specific effects on molecular switches in receptor–ligand complexes without arrestin: TM3–TM6 (Cα of A157^3.53–K260^6.32), TM6–TM7 (Cα of K260^6.32–I329^7.56), tyrosine toggle (OH of Y243^5.58–Y326^7.53), xQxP (Cα of F232^5.47–Q276^6.48), HL1 (Cα of M144^3.43–F272^6.44), and HL2 (Cα of M144^3.43–V268^6.40). (c) Corresponding ligand-dependent effects on the same molecular switches in arrestin-bound tricomplex systems.

These helical rearrangements were coupled to reorganization of the xQxP motif at TM5–TM6, which exhibited shorter inter-residue distances in all active systems, facilitating the helix movements required for receptor activation. Consistently, water-mediated disruption of the hydrophobic layers (HL1: M144^3.43–F272^6.44 and HL2: M144^3.43–V268^6.40) was prominent in agonist-bound and phosphorylated CCR7 complexes, reflecting water permeation during activation (Figure b and c, Figure S3). Conversely, the CCR7^hBD3mu system maintained intact hydrophobic cores, stabilizing the receptor in an inactive conformation. Together, these findings indicate that CCL21 and hBD3wt promote CCR7 activation through coordinated switching of TM motifs and rearrangement of core packing, whereas hBD3mu fails to induce these transitions and instead favors an antagonist-like inactive conformation (Figure b).

Phosphorylation of the CCR7 C-terminal tail further modulated these microswitch transitions. In tricomplexes, particularly those bound to CCL19, distal or full phosphorylation promoted the outward TM6 displacement, tyrosine toggling, xQxP motif engagement, and hydrophobic layer disruption. Notably, in hBD3-bound tricomplexes, the CCR7fullP_βarr2 complex exhibited transitions between active and inactive conformations, as reflected by tyrosine toggle and HL2 distances that lay between the two states. Similarly, the CCR7distalP_βarr2 complex showed transitions involving TM3–TM6 spacing, tyrosine toggling, and xQxP motif engagement, indicating a dynamic equilibrium between the active and inactive states in hBD3-induced signaling under specific receptor phosphorylation patterns (Figure c).

Ligand-Specific and Phosphorylation-Dependent Reorientation of Arrestin Domains Drives Distinct CCR7 Regulatory Responses

Distinct conformational reorientations of arrestin domains were elicited by different ligands and CCR7 phosphorylation patterns, leading to divergent regulatory outcomes. The CCR7distalP_βarr2 complex exhibited a pronounced counterclockwise rotation of the α-helix-I within the N-domain of β-arrestin (Figure a), a structural rearrangement associated with receptor internalization. By contrast, the CCR7fullP_βarr2 complex exhibited only a modest α-helix-I displacement, indicating a reduced internalization response. This underscores the importance of phosphorylation site quality rather than quantity, consistent with earlier findings.

4.

Arrestin interdomain twist and rearrangement. (a) Orientation of α-helix-I in the N-domain of arrestin across multiple CCR7–arrestin simulations. Colored traces represent different systems or replicates; the central region shows distributions relative to the reference angle of the inactive state, and the right panel illustrates helix positioning and rotational movements associated with activation. (b) Dynamic changes in the interdomain twist angle between the N- and C-domains of arrestin are shown on the left for each system. The central panel shows distributions relative to the active-state reference (∼25°), and the right schematic depicts the C-domain rotational transition accompanying arrestin activation and signaling.

In the CCR7distalP_βarr2 complex, no substantial α-helix-I shift was detected. Instead, the C-edge loop underwent a notable reorientation (Figure b), suggesting enhanced arrestin binding affinity and a stronger tendency toward receptor desensitization with limited internalization. The α-helix-I orientation was slightly more negative in CCR7fullP_βarr2 than in CCR7distalP_βarr2 , suggesting a mild bias toward receptor internalization, although not as pronounced as in CCR7distalP_βarr2 . These results indicate that hBD3 promotes a β-arrestin-biased pathway, enhancing receptor internalization through distinct arrestin conformations, while CCL19 acts as a balanced agonist, favoring transient desensitization. Together, these findings highlight that both ligand identity and phosphorylation context precisely regulate CCR7 outcomes through arrestin structural dynamics.

Continuous Water-Mediated Networks Enable Signal Transduction in Activated CCR7

Internal water molecules play a crucial role in GPCR activation by mediating the hydrogen bond networks that drive conformational transitions. In CCR7, dynamic hydrogen bond analysis revealed a continuous water-mediated network in the agonist-bound systems CCR7^CCL21 and CCR7^hBD3wt, linking key residues F275^6.47, R318^7.45, and N322^7.49 (Figure a). This network enabled penetration of water through the disrupted hydrophobic core, thereby supporting receptor activation. However, the antagonist system CCR7^hBD3mu lacked this continuous connectivity (Figure a); its intact hydrophobic barrier diverted water flow and restricted its passage toward the intracellular region. Additionally, a unique hydrogen bond between Y136^3.32 and Y279^6.51, forming the tyrosine toggle, was observed exclusively in CCR7^hBD3wt, promoting localized water clustering in the receptor cavity.

5.

Structural and water-mediated interaction analysis of CCR7. (a) Ligand-bound states. (b) Ligand- and arrestin-bound states. In each system, the water-mediated hydrogen bond network is projected such that the x-axis represents the x–y plane of the protein, and the y-axis corresponds to the z-axis. Nodes indicate amino acid residues capable of forming hydrogen bonds, and edges represent water-mediated interactions between them. Below each structural panel, the corresponding network graph highlights connections spanning ligand-binding and intracellular regions, revealing the extent and organization of hydrogen bond networks in active versus inactive receptor states. Key residues associated with ligand and arrestin binding are indicated, enabling a direct comparison of hydrogen bonding patterns and interaction sites across distinct conformational and functional states of CCR7.

In arrestin-bound tricomplexes, a conserved water-mediated network involving Q276^6.48, N280^6.52, R318^7.45, and N322^7.49 was consistently maintained across all systems (Figure b), underscoring its role in sustaining water flux and stabilizing the active state. Together, these results show that hydrophobic core disruption permits the formation of a persistent water channel, reinforcing the active conformation and facilitating downstream signaling.

Binding Interface between the Phosphorylated CCR7 C-Terminal Tail and Arrestin

The CCR7 C-terminal phosphorylation pattern modulates arrestin binding by differentially engaging key basic residues within defined pockets (Figure a). In the distal phosphorylation pattern (pT372–pT375, pS377), interactions were concentrated primarily in pocket C. In the CCR7distalP_βarr2 complex, pS377 formed contacts with arrestin residues R3 and R99, while pT374 and pT375 interacted with K6 and R3, respectively. By contrast, the CCR7distalP_βarr2 complex showed pT372 interactions with R3, R99, and K103, and pT373 interactions with R99, R102, and K103, all within the α-helix-I region of pocket C (Figure b).

6.

Binding interfaces between the the phosphorylated CCR7 C-terminal and the arrestin N-domain. (a) Electrostatic surface representation of arrestin within the CCR7–arrestin complex. Charge distribution was calculated using the Poisson–Boltzmann equation, with blue and red denoting regions of positive and negative electrostatic potential, respectively. Phosphorylated CCR7 C-terminal residues forming electrostatic interactions with arrestin are labeled in black, while the three polar core regions of the arrestin N-domain, enriched in positively charged residues, are labeled in white. (b) Heatmap of hydrogen bond interactions between CCR7 and arrestin, with arrestin as the donor and CCR7 as the acceptor. Hydrogen bonds were defined by a donor–acceptor distance cutoff of 3.5 Å and an angle cutoff of 120°.

Under full phosphorylation (pS356, pS357, pS364, pS365, pT372–pT375, and pS377), a broader binding interface was observed. In the CCR7fullP_βarr2 system, pS357 and pT373 engaged arrestin K7 and R3, respectively, within pocket C. Additionally, the CCR7fullP_βarr2 complex showed interactions in pocket A, where pS356 contacted K156 and R157, while pS364 and pS365 bound to K7 in pocket C (Figure b). These patterns suggest that although both ligands utilize pocket C for interaction, full phosphorylation in the CCL19 complex extends interactions into pocket A, potentially enhancing the specificity of arrestin binding. This observation aligns with previous reports that α-helix-I acts as a critical region for recognizing phosphorylated GPCR tails.

Discussion

Dynamical Network Analysis Reveals Ligand-Dependent CCR7 Activation and β-Arrestin Signaling Mechanisms

This study investigated how different ligands and phosphorylation patterns shape CCR7 signaling, with a particular focus on interactions with βarr2. By integrating MD simulations and structural analyses, we provide mechanistic insights into how hBD3 variants and endogenous chemokines (CCL19 and CCL21) differentially regulate CCR7 activation and downstream signaling outcomes. − Although the first 35 N-terminal residues of CCR7 were truncated in our simulations due to their intrinsic disorder and lack of experimental structural data, the transmembrane core and extracellular loops (especially ECL2), including functionally critical residues such as W114 ^2.60 at the top of TM2, were fully retained. This tryptophan is conserved in over 80% of chemokine receptors but in less than 15% of nonchemokine class A GPCRs, and it has been experimentally validated to mediate initial chemokine engagement in receptors CCR5 and CXCR4. − Experimental evidence further supports the dominant role of ECL2 and upper transmembrane residues in ligand-specific recognition. For example, Jørgensen et al. demonstrated that mutations of N-terminal residues K50 and R54 impaired signaling for both CCL19 and CCL21, highlighting their contribution to the early recognition step. In contrast, mutation of the centrally located R209 in ECL2 selectively attenuated CCL19-mediated signaling while enhancing CCL21 responses, indicating that ECL2 and upper-TM residues, rather than the flexible distal N-terminus, are the primary determinants of ligand-specific activation.

Another related study also demonstrated that chemokine receptors can utilize their extracellular domains in a ligand-specific manner. For example, in CXCR2, the ligands CXCL1, CXCL7, and CXCL8 rely on different extracellular regions for binding; while the N-terminus is crucial for CXCL1 and CXCL8, it is dispensable for CXCL7. Similarly, CCR5–CCR2 chimeric receptor experiments revealed that swapping all extracellular domains between the two receptors is both necessary and sufficient for ligand recognition, underlining the differential contribution of these external regions across receptors. Collectively, these data indicate that although the N-terminus facilitates the initial docking of CCL19/CCL21, the subsequent stable engagement and signaling mainly depend on ECL2 and upper-TM residues, which were preserved and accurately modeled in our simulations. The truncation of the unresolved, flexible N-terminal residues therefore does not significantly alter the transmembrane or ECL interface responsible for the ligand-specific recognition and activation of CCR7.

Microswitch and community network analyses further revealed ligand-dependent activation states and interaction dynamics. Agonists, including hBD3wt, CCL19, and CCL21, induced varying degrees of CCR7 activation, whereas the antagonist hBD3mu stabilized the receptor in an inactive conformation. Community network analysis revealed that CCL21–CCR7 and hBD3wt–CCR7 complexes shared similar TM clustering: TM3, TM5, and TM6 were grouped into one node, while TM1, TM2, and TM7 formed another, indicative of an active receptor conformation. Notably, the hBD3wt–CCR7 complex involved 19 nodes, suggesting a more complex interaction network than that of CCL21–CCR7. By contrast, in the hBD3mu–CCR7 complex, TM6 remained clustered with TM7, likely restricting outward TM6 movement and preventing activation (Figure S12, Table S3).

In tricomplex systems, the presence of arrestin increased the number of nodes but reduced the size of clusters compared with arrestin-free systems, reflecting more discrete and localized interaction networks upon arrestin engagement (Figure S13, Table S4). Microswitch analysis further supported ligand-specific signaling outcomes. In CCL19–CCR7−βarr2 systems, all microswitches remained fully active, consistent with balanced signaling and desensitization through βarr2 recruitment, facilitated by phosphosite interactions at arrestin pockets A and C (Figure ). By contrast, hBD3–CCR7−βarr2 complexes displayed partial microswitch transitions and exclusive engagement with pocket C, suggesting arrestin-biased signaling (Figures , ). Importantly, counterclockwise rotation of α-helix-I in hBD3-bound systems may facilitate spatial accommodation for IκBα binding, consistent with the reported role of βarr2 in NF-κB activation. Conversely, CCL21–CCR7 systems exhibited intermediate microswitch behavior, aligning with their characterization as weak agonists that favor G-protein-biased signaling (Figure ).

Free-Energy Landscape Analysis of CCR7 Complex Systems in Various States

GPCRs adopt a broad spectrum of conformational states that govern downstream signaling upon ligand binding. To delineate state transitions and activation pathways of CCR7, FEL analyses were performed. For ligand-bound receptor systems, FELs were projected onto the xQxP motif engagement and tyrosine toggle distances (Figure a). The CCR7^hBD3wt system occupied the global minimum, representing the most active conformation with short tyrosine toggle and xQxP motif distances. By contrast, CCR7^CCL21 resided in a distinct low-energy basin, characterized by a slightly larger tyrosine toggle distance but more compact xQxP engagement than CCR7^hBD3wt, indicative of a partially active conformation. The CCR7^hBD3mu mutant was located in a higher-energy region with an increased separation of both switches, suggesting impaired activation capacity.

7.

Energy landscapes of CCR7 molecular switches and arrestin interdomain reorientation. (a) State transitions from inactive to active states in ligand-bound CCR7 systems. The black star indicates the antagonist-bound inactive state, the brown star denotes the weak agonist–bound partially active state, and the green star represents the fully active state. (b) State transitions coupling CCR7 microswitch dynamics with arrestin C-edge loop rotation angles in ligand- and arrestin-bound CCR7 systems, alongside the comparative data from the μOR−βarr2 complex. Energy values are expressed in kcal/mol.

In the tricomplex systems, FELs were evaluated using the tyrosine toggle distance and arrestin C-edge loop rotation (Figure b). The CCR7fullP_βarr2 system exhibited a broad conformational landscape with minimal tyrosine toggle distance but little C-edge loop rotation, indicating structural flexibility and transitions between intermediate states. In contrast, both CCR7distalP_βarr2 and CCR7distalP_βarr2 systems shifted toward regions with intermediate tyrosine toggle distances, increased C-edge loop rotation angles, and the lowest energy, reflecting a more dynamic and potentially favorable state for arrestin recruitment. Notably, the CCR7fullP_βarr2 system remained confined to a distinct high-energy region with minimal C-edge loop rotation, suggesting a reduced suitability for arrestin recruitment. Comparable strategies have also been widely applied in other GPCR–arrestin studies. For instance, in μOR−βarr2 complexes, the potential of mean force (PMF) was calculated using umbrella sampling along a reaction coordinate defined by the z-axis distance between the receptor and arrestin centers of mass. This energetic framework supported the interpretation of the structural transitions occurring during arrestin recruitment, in which rotation of the C-edge loop by approximately ∼ 20° ± 4° facilitates formation of the fully engaged complex and is accompanied by a substantial reduction in free energy to approximately ∼ −8 kcal/mol. ,

Mechanistic Exploration of CCR7 Binding Interfaces and Activation Dynamics

MD simulations revealed distinct ligand–receptor interaction patterns among CCL19, CCL21, and hBD3wt, highlighting differences in binding preferences and activation mechanisms. CCL21 and hBD3wt demonstrated extended interactions into the CRS2 region, whereas CCL19 primarily engaged the CRS1 and CRS1.5 domains (Figure ). These findings are consistent with earlier reports emphasizing the role of CCL19 and CCL21 core domains in driving their differential activation profiles of CCR7. Notably, CCL21but not CCL19appears to rely on interactions within the major binding pocket, specifically involving residues N290^6.62, Y308^7.35, I293^6.65, and C298^7.25. Conversely, both chemokines exhibited prominent interactions with extracellular loop residues and the proximal N-terminal region, underscoring their shared interaction sites outside the core binding pocket (Table S2).

Dual Functional Roles of hBD3 Across Distinct Chemokine Receptors

This study further predicted the binding conformations of CCL19, hBD3wt, and hBD3mu to CCR7, as illustrated by the distance map in Figure e. Notably, the binding site of hBD3mu overlaps with those of CCL19 and hBD3wt on CXCR4, consistent with previous findings showing that hBD3 competes with CXCL12 for CXCR4 binding. Remarkably, hBD3wt functions as a CXCR4 antagonist, while cysteine mutations convert it to an agonist. Contrarily, the present study reveals an opposite functional pattern at CCR7: hBD3wt acts as an agonist, while the cysteine-mutated form (hBD3mu) functions as an antagonist. This opposing behavior may arise from differential interactions at the receptor N-terminus, as hBD3mu fails to form stable contacts with the CCR7 N-terminal region despite both hBD3 variants accessing the CRS2 site. Sustained engagement of hBD3wt with the CCR7 N-terminus thus appears critical for receptor activation, underscoring the importance of specific receptor–ligand contacts in shaping functional outcomes.

Furthermore, the structural basis for these interactions is consistent with principles established across chemokine–receptor systems in which electrostatic complementarity plays a central role. In the CXCL12–CXCR4 complex, positively charged chemokine residues (K1, R8, R12) are inserted into negatively charged pockets of the receptor (D10, E14, E15, E179, D181, D182, D187, D193, D262, E288), highlighting the conserved contribution of electrostatic interactions to binding specificity. , Comparable charge complementarity has also been reported for CXCL8–CXCR1 and other chemokine–GPCR pairs. Consistent with these observations, electrostatic potential surface analysis of CCR7 revealed that its N-terminal tail and core domain are enriched in negatively charged regions, facilitating interactions with the positively charged surfaces of hBD3wt (Figures S9, S10). Such electrostatic complementarity likely underlies the sustained engagement of hBD3wt with the CCR7 N-terminus and provides a mechanistic explanation for its agonistic activity at this receptor.

There are some limitations of this study. The simulations were restricted to the formation of the trimeric complex of the ligand, CCR7, and arrestin without extending to downstream interactions between arrestin and NF-κB, as direct structural evidence for arrestin–NF-κB coupling is still lacking. Future studies will be needed to elucidate the molecular mechanisms by which arrestin may promote NF-κB activation and regulate the signaling pathways involved in tumor progression.

Conclusions

hBD3 is strongly overexpressed in OSCC, where it has been implicated in modulating immune responses and enhancing tumor progression. In this study, we examined how hBD3 variants together with endogenous chemokines, CCL19 and CCL21, influence CCR7 signaling dynamics. This signaling axis is increasingly recognized for its roles in cancer cell migration and immune cell trafficking. Through MD simulations and structural analyses, we showed that hBD3wt and CCL19 act as potent agonists of CCR7, promoting β-arrestin-biased signaling and sustained receptor internalization. By contrast, hBD3mu exhibited antagonist-like behavior, failing to activate CCR7 and competitively disrupting agonist binding through overlapping interface engagement without forming a continuous hydrogen bond network. Compared to hBD3 bound to CXCR4, our findings confirmed dual functional roles of hBD3 in distinct chemokine receptors, CCR7 and CXCR4. , CCR7 phosphorylation patterns further modulated arrestin recruitment, with full C-terminal phosphorylation supporting broader interactions across arrestin pockets, particularly in the CCL19 system, which is also consistent with previous findings. These results suggest that ligand-specific phosphorylation signatures contribute to biased signaling outcomes. Consistently, the comparison with the prior study of μOR−βarr2 supports the same pattern of arrestin engagement, aligning with the conformational behavior observed in our simulations. The distinct interaction profile of CCL21, compared with hBD3wt and CCL19, supports its characterization as a weaker agonist, reinforcing the idea that subtle variations in ligand–receptor engagement can markedly influence downstream signaling in arrestin recruitment and NF-κB activation (Figure ). Taken together, this study elucidates the molecular determinants of CCR7 activation and biased signaling in the context of hBD3 overexpression in OSCC. These insights provide a framework for the development of selective CCR7 modulators with potential applications in cancer immunotherapy and the treatment of inflammatory diseases.

8.

Proposed mechanism of hBD3-driven CCR7 activation leading to β-arrestin-2-mediated NF-κB signaling in oral squamous cell carcinoma.

The data that support the findings of this study are available in the Supporting Information of this article.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.5c01298.

• Summary of all simulation systems, residue pairs forming hydrogen bonds, information on the community residue of nodes in the system interaction network, homology modeling results, RMSD profiles of various systems over time, docking results of hBD3 to CCR7, surface charge distribution for different systems, and comparison of microswitches among different CCR7 systems (PDF)

$.

N.P.K. and C.-C.C. contributed equally to this work. CRediT: Novelyn Putri Kinanti conceptualization, data curation, formal analysis, investigation, methodology, software, validation, visualization, writing - original draft; Chun-Chun Chang data curation, formal analysis, funding acquisition, investigation, supervision, writing - review & editing; Xing-Yan Lai data curation, formal analysis, investigation, methodology, software, visualization; Hao-Jen Hsu conceptualization, funding acquisition, investigation, project administration, resources, supervision, validation, writing - original draft, writing - review & editing.

This research was funded by the National Science and Technology Council (NSTC) of Taiwan under grant numbers MOST 111-2221-E-303-003, NSTC 113-2221-E-320-001, and NSTC 112-2221-E-303-001-MY2. This research was also funded by Buddhist Tzu Chi Medical Foundation (Grant Number: TCMMP114-03-01). This research was funded by a collaborative grant from Buddhist Tzu Chi Medical Foundation, Tzu Chi University, and Academia Sinica (Grant Number: TCAS-114-01)

The authors declare no competing financial interest.