Differential CCR7 Targeting in Dendritic Cells by Three Naturally Occurring CC-Chemokines

Abstract

The CCR7 ligands CCL19 and CCL21 are increasingly recognized as functionally different (biased). Using mature human dendritic cells (DCs), we show that CCL19 is more potent than CCL21 in inducing 3D chemotaxis. Intriguingly, CCL21 induces prolonged and more efficient ERK1/2 activation compared with CCL19 and a C-terminal truncated (tailless) CCL21 in DCs. In contrast, tailless-CCL21 displays increased potency in DC chemotaxis compared with native CCL21. Using a CCL21-specific antibody, we show that CCL21, but not tailless-CCL21, accumulates at the cell surface. In addition, removal of sialic acid from the cell surface by neuraminidase treatment impairs ERK1/2 activation by CCL21, but not by CCL19 or tailless-CCL21. Using standard laboratory cell lines, we observe low potency of both CCL21 and tailless-CCL21 in G protein activation and β-arrestin recruitment compared with CCL19, indicating that the tail itself does not improve receptor interaction. Chemokines interact with their receptors in a stepwise manner with ultimate docking of their N-terminus into the main binding pocket. Employing site-directed mutagenesis we identify residues in this pocket of selective CCL21 importance. We also identify a molecular switch in the top of TM7 important for keeping CCR7 in an inactive conformation (Tyr312), as introduction of the chemokine receptor-conserved Glu (or Ala) induces high constitutive activity. Summarized, we show that the interaction of the tail of CCL21 with polysialic acid is needed for strong ERK signaling, whereas it impairs CCL21-mediated chemotaxis and has no impact on receptor docking consistent with the current model of chemokine:receptor interaction. This indicates that future selective pharmacological targeting of CCL19 versus CCL21 should focus on a differential targeting of the main receptor pocket, while selective targeting of tailless-CCL21 versus CCL21 and CCL19 requires targeting of the glycosaminoglycan (GAG) interaction.

Keywords: CCL19; CCL21; CCR7; ERK; biased signaling; dendritic cell; tailless-CCL21.

Figure 1

3D chemotaxis of human dendritic cells (DCs) in response to CCL19 and CCL21. Human DCs naturally expressing CCR7 were used for the 3D chemotaxis assay. In this assay, a collagen filled channel with the cells is in contact with a source and sink reservoir on either side, causing the cells to experience a linear gradient as chemokine gradually diffuses from source to sink. (A) Column diagrams showing the directional migration of DCs in response to either 10 nM (left) or 100 nM (right) CCL19 or CCL21 source concentrations. The values are calculated as chemotactic index (CI) in the MATLAB software. (B) Spider diagrams depicting the migration pattern in response to the indicated concentrations of the chemokines. Statistical significance was calculated using unpaired t test. NS, not significant; *P < 0.05 (n = 3).

Figure 2

ERK1/2 activation in human dendritic cells in response to CCL19 and CCL21. Human DCs with natural CCR7 expression were used for the ERK1/2 activation. The percentage of phosphorylated ERK1/2 was calculated as described by the manufacturer (Meso Scale Discovery, MD, USA) as % Phosphoprotein = [(2 × Phospho-signal)/(Phospho-signal + Total signal)] × 100. The pERK1/2 data are normalized to buffer and displayed as fold ERK1/2 activation over buffer control. (A–C) The effect of 1 nM (A), 10 nM (B), or 100 nM (C) of CCL19 and CCL21. (D) The effect of PTX (10 μg/ml) on ERK1/2 activation by CCL19 and CCL21. (E) Timely effect of 10 nM and (F) 100 nM CCL19 and CCL21 on DC ERK1/2 phosphorylation. (G) The effect of neuraminidase (NA) treatment of DCs on ERK1/2 activation by CCL21 and (H) by CCL19. Statistical significance was calculated using unpaired t test. NS, not significant; *P < 0.05, **P < 0.01 (n = 3–6).

Figure 3

Response of the three naturally occurring chemokines (CCL19, CCL21, and tailless-CCL21) on 3D chemotaxis of human dendritic cells. (A) Column diagrams showing the directional migration of DCs in response to 10 nM and 100 nM of CCL19, CCL21, or tailless-CCL21 source concentrations. The values are calculated as chemotactic index (CI) in the MATLAB software. (B) Column diagrams showing the pERK in response to 100 nM CCL19, CCL21, or tailless-CCL21 in neuraminidase treated DCs as percentage of the signal of the corresponding chemokine in non-neuraminidase treated cells. Data are normalized, and all non-treated responses set to 100%. Statistical significance was calculated using unpaired t test. **P < 0.01, *P < 0.05, NS, not significant (n = 3).

Figure 4

CCL21 binds to the surface of human DCs and forms discrete puncta; a feature not matched by tailless-CCL21. Fluorescence microscopy pictures of DCs incubated (A) with100 nM CCL21, (B) 100 nM tailless-CCL21, or (C) in the absence of ligand obtained on LSM 780 confocal microscope using 63× oil-objective (ligands were stained with Alexa 488 and the cell nucleus visualized with Hoecst DNA staining). (D) Zoom in on DC incubated with 100 nM CCL21 to visualize that anti-CCL21 staining followed the villi-like surface of non-adherent DCs. (E) Visualization of two different types of CCL21 puncta. (F) The anti-CCL21 antibody also recognized tailless-CCL21.

Figure 5

CCR7 G protein activation, β-arrestin recruitment, and internalization induced by CCL19, CCL21, and tailless-CCL21. (A) Dose–response curves of CCL19 (circles), CCL21 (squares), and tailless-CCL21 (triangles) obtained in PI-turnover assay measured in HEK293 cells cotransfected with CCR7 and the chimeric G protein G_qi4myr. Statistical significant difference from CCL19 was calculated by unpaired Student’s t-test. *P < 0.05, **P < 0.01, and NS, not significant. (B) β-arrestin recruitment measured in U2OS cells stably transfected with β-arrestin 2 linked to an enzyme acceptor and transiently transfected with enzyme donor-fused CCR7 with increasing concentrations of CCL19 (circles), CCL21 (squares), and tailless-CCL21 (triangles). Statistical significance between CCL19 and CCL21 values was calculated using unpaired t test. **P < 0.01. (C) β-Arrestin 2-dependent internalization measured in U2OS cells coexpressing untagged CCR7 DNA, β-arrestin 2 fused to the enzyme acceptor part of β-gal, and endosomes tagged with PK, the enzyme donor. Dose–response curves for CCL19 (circles) and CCL21 (squares) are shown. Statistical significance was calculated using unpaired t test (n = 3–4).

Figure 6

Structure and alignment of CCL19 and CCL21. (A) NMR solution structure of CCL21 (PDB reference 2L4N) and CCL19 (PDB reference 2MP1). Cartoon structures are aligned in PyMOL to demonstrate the secondary structures likely to be found in both proteins (top). Exposed charges of either structure are shown on the surface presentation (bottom) with positive charges in blue and negative charges in red. Pale blue and pale red represents surface-exposed nitrogen- and oxygen atoms, respectively. The lower line divides the chemokine core domain from the highly flexible N-terminus, whereas the upper line delimits the regions that most closely match the mark of a GAG-binding domain, i.e., a dense cluster of exposed positive charges. The light blue ellipse on CCL21 represents the large unsolved C-terminal tail, of which most is also removed in tailless-CCL21. (B) Amino acid sequence alignment of CCL19 and CCL21 using the MAFFT multiple-aligner plug-in of Geneious Pro 6.1.7 software (Biomatters Ltd., Auckland, New Zealand). The secondary structures are shown with symbols, using arrows for beta-strands and a cylinder for the alpha-helix. Identical residues are black, similar residues are gray, and positively and negatively charged residues are blue and red, respectively. The unsolved C-terminus of CCL21 is outlined in a light blue box.

Figure 7

Site-directed mutagenesis scan of CCR7. (A–E) Dose–response curves of CCL19 (black squares) and CCL21 (white squares) on CCR7 WT (dotted line) and mutants obtained in PI-turnover assay measured in HEK293 cells cotransfected with CCR7 constructs and the chimeric G protein G_qi4myr. (A) Mutational analyses of position 312 (Y312E, upper panel and Y312A, lower panel). (B) Mutation of position 309 (D309A) and (C) position 137 (K137A). (D) Mutation of position 193 (E193A, upper panel and E193D, lower panel). (E) Mutation of position 133 (F133A, upper panel and F133H, lower panel). (F) Effects of mutation K130A, K137A, and Q227A at 100 nM chemokine concentrations. Top (G) and side (H) view of the chemokine binding pocket of CCR7 based on homology-modeling from CCR5 (PDB reference 4MBS) (36) with helices indicated by roman letters. (G) Y312 is located at the border between the major (left) and minor (right) binding pocket, while D309 is located at the top of the minor binding pocket. A view into the major binding pocket presents the relative positions of F133, K137, and E193 (H). Middle section: serpentine model, upper panel, and helical wheel, lower panel of CCR7 with indication of the included mutations. Statistical significance in signaling through WT CCR7 and mutants with either CCL19 or CCL21 was calculated using unpaired t test. **P < 0.01, *P < 0.05 (n = 3–9).

Figure 8

Overview of differences in signaling induced by CCL19, CCL21, and tailless-CCL21. Schematic illustration of the relative effect of CCL19, CCL21, and tailless-CCL21 on diverse cellular effects presented in Figures 1–6. Thickness of arrows signifies effect, i.e., thick arrow, high activity; thin arrow, low activity.

Figure 9

Model for diverse mechanisms of CCL19, CCL21, and tailless-CCL21 in regulation of CCR7 activation. As revealed by the structure comparison (Figure 5), CCL21 harbors structural motifs indicating a unique ability to strongly interact with GAGs, also supported by earlier reports. Increased GAG binding builds a local CCL21 reservoir at the cell surface, with CCL21 present in an inhibited, non-inhibited, or even facilitated form dependent on the GAG carrying it. In contrast, CCL19 only displays weak GAG interaction and readily diffuses away from the cell if not immediately bound to CCR7. GAG-bound CCL21 on the contrary, may either (i) interact directly with CCR7 if bound to polysialic acid, (ii) interact with CCR7 after GAG detachment caused by protease induced tail removal, with tailless-CCL21 probably being immediately ready for receptor engagement since release is expected to occur at the DC surface and thus in close proximity to CCR7 molecules ready to capture the chemokine, or (iii) interact with CCR7 after transfer from a GAG that presents CCL21 in its inhibited state (CS-B), to a GAG that allows CCR7 engagement (polysialic acid), with CS-B here acting as a dormant reservoir. CCL19 does not form a local reservoir, and once bound to CCR7, it gets quickly internalized and degraded, which is not the case for CCL21. Our model thus predicts that CCL19 and CCL21 induce differential CCR7 activation, with CCL19 creating a short-lived (temporary) signal, and CCL21 displaying a weaker, but more persistent CCR7 activation profile.