Oligomerization of RANTES is required for CCR1-mediated arrest but not CCR5-mediated transmigration of leukocytes on inflamed endothelium

Chemokines control the selective migration of leukocytes in immune surveillance, inflammation, and atherogenesis.^1,2  This involves not only the directed migration of leukocytes along chemokine gradients within the extracellular matrix but also the arrest of circulating leukocytes on activated endothelium under flow. The immobilization of chemokines to endothelium via binding to glycosaminoglycans (GAGs) has been implicated in triggering leukocyte arrest, as well as transmigration (or chemorheotaxis) in flow.^3-7  Moreover, binding of chemokines to GAGs present on leukocytes may influence chemotactic responses.^8,9  The BBXB motif, in which B represents a basic amino acid, has been implicated as the principal heparin or heparan-sulfate binding site on several proteins.¹⁰  This motif is important for the chemokine-heparin interaction by mutagenesis studies (eg, ⁴⁴AANA⁴⁷-RANTES).^8,11-15  Single residue substitutions in this motif reduced GAG-binding and chemotaxis across endothelial monolayers, but not bare transwells, while retaining receptor affinity.¹⁶  The oligomerization of RANTES (regulated on activation, normal T cell expressed and secreted) may also influence its functional properties, for example, dimeric or tetrameric mutants retained G-protein–coupled receptor-mediated activities, but prolonged effects via protein tyrosine kinase activation were attenuated.^17-19  While GAGs mediate cell surface oligomerization of chemokines,¹⁷  the state of oligomerization may conversely modulate the capacity for GAG-binding depending on the chemokine.^20,21 

We have recently found that both GAG-binding and oligomerization of chemokines represent essential structural features for their ability to recruit leukocytes in vivo.²¹  However, the contribution of oligomerization to distinct steps of leukocyte recruitment has not been studied. RANTES immobilized on activated endothelium can trigger leukocyte arrest and transmigration, which are mediated by specialized roles of CC chemokine receptor 1 (CCR1) and CCR5, respectively.⁶  Here, we have investigated the functional effects of RANTES mutants with defects in oligomerization in physiologic assays of RANTES-triggered leukocyte recruitment on endothelium. The results demonstrate differential requirements for oligomerization of RANTES in CCR1-mediated arrest and CCR5-mediated transmigration.

Human microvascular endothelial cells (HMVECs; PromoCell, Heidelberg, Germany) and monocytic Mono Mac 6 cells were cultured as described.^5,6  HMVECs were stimulated with interleukin-1β (IL-1β; 10 ng/mL) for 24 hours. Monocytes and peripheral blood mononuclear cells (PBMCs) were isolated by NycoPrep 1.068 (Nycomed, Oslo, Norway) or Ficoll hypaque density gradient centrifugation, respectively.^5,6  CD45RO⁺CD4⁺ T lymphocytes were purified by negative selection using magnetic separation (Miltenyi Biotec, Bergisch Gladbach, Germany). Human IL-1β and RANTES were from PeproTech (Rocky Hill, NJ). The E26A, E66A, ⁴⁴AANA⁴⁷-RANTES mutants, and chemically modified Nme-Thr7-RANTES were produced as described.^13,21 

Monocyte adhesion experiments on purified intercellular adhesion molecule-1 (ICAM-1) were performed as described.²²  Laminar flow assays and chemotaxis were performed as described.^5,6,23  Confluent HMVECs activated with IL-1β (10 ng/mL) overnight were preincubated with RANTES or mutants at 20 nM for 30 minutes at 37°C. Surface enzyme-linked immunosorbent assay (ELISA; R&D Systems, Minneapolis, MN) was used to determine the heparan sulfate–specific immobilization of RANTES or mutants on HMVECs treated with or without heparitinase I (0.5 U/mL; Sigma, St Louis, MO) for 2 hours at 37°C. Leukocytes (5 × 10⁵/mL) pretreated with the small molecule antagonists BX471 (provided by Dr R. Horuk, Berlex Biosciences), TAK-779 (provided by Takeda Pharmaceuticals) at 1 μg/mL, or vehicle for 15 minutes⁶  in assay buffer were perfused at 1.5 dyne/cm². Arrest, spreading, and/or transmigration was analyzed in multiple high-power fields recorded by video microscopy.^5,6,23  Unstimulated HMVECs were grown on Transwell-filter inserts (5-μm pore size; Co-star, Cambridge, MA). PBMCs in RPMI-1640/0.5% bovine serum albumin were allowed to transmigrate across bare or HMVEC-coated filters toward RANTES, mutants, or buffer control in the lower chamber for 90 minutes at 37°C. Transmigrated monocytes were counted by flow cytometry. The ratio of RANTES-induced versus spontaneous migration was defined as chemotactic index.

RANTES immobilized on cytokine-activated endothelium triggers leukocyte recruitment in flow.⁶  We used this model to investigate the effect of RANTES mutants with impaired ability to oligomerize on leukocyte arrest. The structural characteristics of these mutants are summarized in Table 1. While the tetrameric E26A and the dimeric E66A have been described,^19,20  Nme-Thr7-RANTES is unable to form hydrogen bonds across the β-sheet dimer interface and was predicted to be monomeric by analogy to an IL-8 variant.²⁴  This property was confirmed by analytical ultracentrifugation, the inability of Nme-Thr7-RANTES to oligomerize on immobilized GAG, and a marginally reduced capacity to bind heparin, reflecting oligomerization on heparin at high concentrations of chemokine in this assay.²¹  Surface ELISA on activated HMVECs revealed an unaltered heparan sulfate–dependent immobilization of the oligomerization mutants compared with wild-type RANTES (Figure 1A), while that of the GAG-binding mutant ⁴⁴AANA⁴⁷-RANTES was abolished. In line with preserved chemotactic activities,²⁰  the ability of all oligomerization mutants to trigger adhesion on purified ICAM-1 in stasis was equivalent to wild-type RANTES, excluding an impairment in intrinsic receptor-binding or signaling properties (Figure 1B).

In support of published data,⁶  the immobilization of wild-type RANTES on activated HMVECs induced substantial arrest of isolated blood monocytes and CD45RO⁺CD4⁺ T cells in flow (Figure 1C-D). As shown by inhibition with the CCR1 antagonist BX471,⁶  RANTES-triggered arrest was mediated by CCR1 (Figure 1C-D). While all oligomerization mutants showed significantly reduced ability to induce shear-resistant monocyte arrest, the most marked defect was found with the monomeric form (Figure 1C). Similar results were obtained with CD45RO⁺CD4⁺ T cells for the dimeric E66A and monomeric Nme-Thr7-RANTES, while the tetrameric E26A was equivalent to wild-type RANTES (Figure 1D). The ability of E26A to trigger arrest of T cells but not monocytes may reflect lower levels of CCR1 on T cells,⁶  which may be sufficiently engaged by tetrameric RANTES. This is in agreement with results in an in vivo model of chemotaxis, in which RANTES displayed a structural requirement of a tetramer to recruit cells to the peritoneal cavity.²¹  These data infer that oligomerization of immobilized RANTES on HMVECs plays a critical role for triggering the arrest component of leukocyte recruitment in flow.

As opposed to arrest, RANTES-triggered leukocyte spreading and transmigration in flow, a phenomenon termed chemorheotaxis, is mediated by CCR5 rather than CCR1.^6,7  Inhibition with the CCR5 antagonist TAK-779 revealed that spreading/transmigration of monocytes and CD45RO⁺CD4⁺ T cells on activated endothelium in flow were dependent on CCR5 (Figure 1E-F). Notably, the oligomerization mutants retained the ability to trigger CCR5-mediated spreading/transmigration (Figure 1E-F). This demonstrates a dichotomy in the structural features of RANTES required for distinct functional properties in triggering arrest versus migration.

While GAG-binding mutants variably affect chemotaxis across bare filters or endothelium,^13,16,17  the ability of oligomerization mutants to induce transendothelial chemotaxis has not been investigated. In extension to recent data,²¹  the oligomerization mutants showed no significant alteration in their ability to induce monocyte chemotaxis across bare filters or endothelial monolayers, which was preferentially mediated by CCR5 but also by CCR1 (Figure 2B). In accordance with previous studies,^13,16 44AANA⁴⁷-RANTES showed a significantly reduced potential to trigger monocyte chemotaxis across bare filters or endothelium (Figure 2A-B). While monomeric Nme-Thr7-RANTES is fully active in filter assays,²¹  its optimal concentration differed by 10-fold from that of wild-type RANTES for the induction of transendothelial migration.

A further reduction in potency was observed for ⁴⁴AANA⁴⁷-RANTES, which induced transendothelial chemotaxis only at a concentration 100-fold higher than that for wild-type RANTES, without reaching maximal efficacy (Figure 2C). Although the ⁴⁴AANA⁴⁷ mutation interferes with both GAG-binding and CCR1 receptor affinity,¹³  the ability of ⁴⁴AANA⁴⁷-RANTES to trigger largely CCR5-dependent spreading/transmigration of leukocytes in flow was also abolished (Weber et al⁶  and data not shown), indicating a relevance of GAG-binding. Conversely, the slightly reduced potency of Nme-Thr7-RANTES may be due to marginal effects on GAG-binding.²¹ 

Our data provide the first molecular evidence that the degree of chemokine oligomerization differentially contributes to the efficacy of distinct steps of leukocyte recruitment, by establishing a functional dichotomy for the requirement of RANTES self-oligomerization in CCR1-mediated arrest but not CCR5-mediated spreading and transmigration. Since the motifs for CCR1 and GAG-binding overlap,¹³  shear-resistant arrest may require the immobilization and correct presentation of RANTES oligomers via GAGs for bridging between immobilized RANTES and CCR1 on rolling leukocytes. Our data corroborate that RANTES monomers suffice for chemotaxis, while GAG-binding is indispensable for transendothelial chemotaxis and for chemorheotaxis (ie, spreading/transmigration triggered in flow).^7,16  This may be due to interactions with GAGs on leukocytes^8,9  but may also support the formation of a haptotactic gradient for transmigration. Defining structural prerequisites for RANTES-triggered leukocyte recruitment harbors important implications for the design of anti-inflammatory compounds²⁵  precisely targeting the functionally relevant region.

We thank Dr R. Rossaint for continuous support.