T lymphocytes are highly motile and constantly reposition themselves between a free-floating vascular state, transient adhesion and migration in tissues. The regulation behind this unique dynamic behaviour remains unclear. Here we show that T cells have a cell surface mechanism for integrated regulation of motility and adhesion and that integrin ligands and CXCL12/SDF-1 influence motility and adhesion through this mechanism. Targeting cell surface-expressed low-density lipoprotein receptor-related protein 1 (LRP1) with an antibody, or blocking transport of LRP1 to the cell surface, perturbed the cell surface distribution of endogenous thrombospondin-1 (TSP-1) while inhibiting motility and potentiating cytoplasmic spreading on intercellular adhesion molecule 1 (ICAM-1) and fibronectin. Integrin ligands and CXCL12 stimulated motility and enhanced cell surface expression of LRP1, intact TSP-1 and a 130 000 MW TSP-1 fragment while preventing formation of a de-adhesion-coupled 110 000 MW TSP-1 fragment. The appearance of the 130 000 MW TSP-1 fragment was inhibited by the antibody that targeted LRP1 expression, inhibited motility and enhanced spreading. The TSP-1 binding site in the LRP1-associated protein, calreticulin, stimulated adhesion to ICAM-1 through intact TSP-1 and CD47. Shear flow enhanced cell surface expression of intact TSP-1. Hence, chemokines and integrin ligands up-regulate a dominant motogenic pathway through LRP1 and TSP-1 cleavage and activate an associated adhesion pathway through the LRP1–calreticulin complex, intact TSP-1 and CD47. This regulation of T-cell motility and adhesion makes pro-adhesive stimuli favour motile responses, which may explain why T cells prioritize movement before permanent adhesion.
T lymphocytes exist as immotile free-floating cells in the vascular system and as highly motile cells within tissues, where they migrate extensively, recognize antigens and position themselves relatively freely. An intriguing aspect of this behaviour is that the cells rarely stop and seem programmed not to adhere permanently. The high motility is an essential element of the immune surveillance against infectious agents and cancer cells throughout the organism[1-4] but unfortunately also promotes the infiltration of tissues that is a hallmark of autoimmune and allergic disorders as well as rejection of transplants and graft-versus-host disease. However, migration of T cells in tissues may also be beneficial as demonstrated by the fact that stimulation of recruitment of regulatory T (Treg) cells by the chemokine CXCL12/SDF-1 induces immunological tolerance and protection against experimental autoimmune encephalomyelitis.
In spite of the importance of T-cell motility both for beneficial and harmful processes its regulation is poorly understood. It is also unclear how the free-floating vascular state of these cells and particularly the transition between the free-floating state, transient adhesion and motility is regulated. Studies of T-cell motility and adhesion may deepen our insights into basic immune regulation and provide novel drug targets that can prevent lymphocyte infiltration of target organs and inflammatory disorders. T-cell motility is generally described as an effect of signalling from chemokine receptors or integrin-mediated adhesion to the intracellular motile machinery,[6, 7] but its regulation is probably more complex than this. Hence, highly motile T cells with activated integrins induced through chemokine receptors are non-adhesive whereas adhesion is up-regulated by flow.
Low-density lipoprotein receptor-related protein 1 (LRP1) is a multifunctional 600 000 molecular weight (MW) protein with a broad repertoire of ligand interactions.[9, 10] Thrombospondin-1 (TSP-1) is a 450 000 MW calcium-binding protein with binding sites for LRP1, calreticulin and integrin-associated protein (CD47)[11-13] that was previously implicated in motility and adhesion of T cells.[14, 15] Calreticulin is a 50 000 MW protein associated with LRP1 on the cell surface.[16, 17] LRP1 expression in T cells seems to predict unresponsiveness to anti-tumour necrosis factor therapy and LRP1 and endogenous TSP-1 direct a counter-adhesive and counter-proliferative motogenic cascade in T cells. Although this cascade seems to consist of a series of molecular interactions at the cell surface it is important to further clarify whether interactions at the cell surface level are critical for motility and adhesion. It is also of interest to investigate the impact of the free-floating vascular state and shear force on such interactions. To deepen the analysis of the importance of molecular interactions at the cell surface we examined the influence of various antibodies on the motility of human T lymphocytes in short-term experiments (5–30 min) based on the assumption that antibodies under these conditions selectively target cell surface elements. This led to the finding that ustekinumab, a human monoclonal antibody to the p40 protein subunit of interleukin-12 (IL-12) and IL-23, was a potent inhibitor of adhesion-independent T-cell motility while concomitantly affecting the cell surface expression of LRP1 and TSP-1. A number of other antibodies, including anti-integrin antibodies, did not affect T-cell motility. Notably, ustekinumab enhanced the adhesion of T cells on intercellular adhesion molecule 1 (ICAM-1) and fibronectin. Ustekinumab thus possesses unique functional properties through its combined effects both on motility and adhesion and provides strong evidence that the regulation of these functions is integrated at the cell surface level. Analysis of the specificity of ustekinumab indicated that it targeted LRP1 or an associated component distinct from TSP-1. We also made additional findings relevant to the understanding of how T-cell motility and adhesion are regulated. LRP1 and TSP-1 were found to be transported to the cell surface independent of each other and to subsequently interact. This interaction controlled the cell surface distribution and amount of TSP-1 within the plasma membrane. CXCL12 as well as contact with ICAM-1 or fibronectin substrata stimulated T-cell motility and up-regulated cell surface expression of LRP1 and TSP-1. Furthermore, both CXCL12 and these integrin ligands induced a prominent 130 000 MW TSP-1 fragment besides the up-regulation of intact TSP-1. The stimulatory effect of CXCL12 and integrin ligands on motility was inhibited by ustekinumab, which concomitantly enhanced cytoplasmic spreading on ICAM-1 and fibronectin, indicating that distinct motogenic factors control a common cell surface mechanism for regulation of motility and adhesion. Of particular interest, our studies further revealed that the pattern of expression of LRP1 and TSP-1 correlated with the state of the T cell as free-floating, motile, adherent or de-adherent. This points to the possibility that we have identified a mechanism that regulates the capacity of T cells to undergo transition between different states and hence reposition themselves within the organism.
Material and methods
Chemicals and antibodies
Rat tendon collagen type I, fibronectin and TSP-1 were purified and prepared as described elsewhere.[21-23] Poly-l-lysine (molecular mass 5300) was purchased from Miles-Yeda Ltd (Rehovoth, Israel). Dynabeads were from Dynal Biotech (Oslo, Norway). Brefeldin A was from Sigma (St Louis, MO). Wortmannin, colchicine and cycloheximide were from Sigma-Aldrich. Dynasore was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-CD3 (clone SK7, IgG1) and anti-CD4 (clone SK3, IgG1) were obtained from Becton Dickinson (Mountain View, CA). Mouse IgG, anti-CD8 (C8/144B, IgG1) and goat anti-mouse IgG were from Dacopatts A/S (Glostrup, Denmark). Anti-TSP-1 clone A6.1 (also called TSP-Ab-4, IgG1), clone C6.7 (also called TSP-Ab-3, IgG1) and anti-TSP-1 clone MBC 200.1 (also called TSP-Ab-9, IgG1) were from NEO-MARKERS (Fremont, CA). Anti-LRP1 (clone A2MRα2, IgG1) was obtained from Santa Cruz Biotechnology. Anti-calreticulin (clone FMC75), anti-IL-12 was obtained from R&D Systems Ltd, Europe (Abingdon, UK). Anti-calreticulin (clone FMC75), anti-IL-23 and anti-IL-12/IL-23p40 were from Biosite (Täby, Sweden). ICAM-1, CXCL12 and CCL5 were from R&D Systems Ltd. Anti-fibronectin [clone IST1, IgG1 was obtained from Sera lab (Loughborough, UK)]. Ustekinumab and infliximab were from Centocor B.V. (Leiden, the Netherlands) and efalizumab was from Industria Farmaceutica Serono S.p.A. (Rome, Italy). KRFYVVMWKK (4N1K) and KVFRWKYVMK (scrambled 4N1K) were synthesized by Tri pep (Novum Research Park, Huddinge, Sweden). RWI ESKHKS DFGKFVLSS (the TSP-1 binding site in calreticulin) and a scrambled control peptide (RSVWIKKELGSKDSFHSF) were synthesized by the Biomolecular Resource Facility (University of Lund, Lund, Sweden). Biotinylated peroxidase and avidin were from Vector Laboratories (Burlingame, CA).
Blood lymphocytes were purified using Lymphoprep and depleted of monocytes by treatment with carbonyl iron and magnetic removal of phagocytic cells. The use of lymphocytes from the blood of healthy individuals was approved by the local ethics committee. The cell preparations obtained consisted of 82–93% CD3-positive cells. Further enrichment of T cells was accomplished by depleting CD56-, CD19- and CD14-positive cells using beads coated with the corresponding antibodies. The birch (Bet v I) -specific T-cell clone AF 24 was obtained from Dr Jost van Nerven (ALK, Copenhagen, Denmark). AF24 was stimulated with anti-CD3 or specific antigen Betv G75 presented by HLA-identical B cells and cultured in the presence of IL-2 for 9–12 days before the experiments. Lymphocytes were cultured in RPMI-1640 (Gibco Ltd, Paisley, UK) supplemented with 2 mm l-glutamine, 0·16% sodium bicarbonate, 10 000 U/ml benzylpenicillin, 10 000 μg/ml streptomycin and 10% fetal calf serum or in serum-free AIM-V medium (Gibco Ltd). Human umbilical vein endothelial cells were isolated and cultured as described in medium 199 (Gibco Ltd) in 20% fetal calf serum without growth factor supplementation. The experiments were performed under serum-free conditions to exclude any interference of exogenous proteins and peptides. To maintain the lymphocytes in the free-floating state they were shaken on an IKAWERK KS 500 shaker at an agitation rate 150/min unless otherwise stated. To optimize the experimental conditions we also tested an STRG PLATFORM ROCKER and a Swelab MIXER 820 as well as a flow system created using a Pharmacia peristaltic pump and attaching tubes (Bergman-Labora AB, Danderyd, Sweden).
Small interfering RNA-mediated gene silencing
The expression of LRP1 was suppressed using the human T-cell Nucleofector kit (Lonza, Köln, Germany) and a Nucleofector device (Amaxa Biosystems, Köln, Germany) as previously described. Briefly, 5 × 106 T-enriched cells were resuspended in 100 μl of nucleofector solution and transfected with 500 nm final concentration of small interfering RNA (siRNA) using protocol U14. The siRNA consisted of LRP1 siRNA (human) (sense: AAGACUUGCAGCCCCAAGCAGtt; antisense: CUGCUUGGGGCUGCAAGUCUUtt) and control siRNA (sc-37007) from Santa Cruz Biotechnology and LRP1 SiRNASuppl (human) (sense: GCUGUGACAUGGACCAGUUtt; antisense: AACUGGUCCAUGUCACAGCgg) from Applied Biosystems (Foster City, CA). The degree of gene silencing and the influence of silencing on motility were determined 40 hr after introducing siRNAs.
The expression of various antigens was analysed in cells fixed in 2% paraformaldehyde at 4° for 20 min attached to glass slides coated with poly-l-lysine (10 μg/ml) at 4° over night. Antigen expression was detected with monoclonal antibodies and a complex of biotinylated peroxidase and avidin (Vector Laboratories). For detection of intracellular antigens cells were fixed in 2% paraformaldehyde and permeabilized by 0·1% saponin. The cells were examined in a Nikon Eclipse E1000M microscope (Nikon Instruments, Melville, NY). The intensity of the immunocytochemical staining was quantified using the image processing and analysis program imagej.
Biotinylation and immunoprecipitation
The surface membrane of intact lymphocytes was labelled with d-biotinyl-e-aminocaproic acid-N-hydroxysuccinimide ester (biotin-7-NHS) as described by the manufacturer (Roche Molecular Biochemicals, Mannheim, Germany). For immunoprecipitation, adherent cells were biotinylated, released by a cell scraper and then immunoprecipitated. Cells in suspension and de-adherent cells were biotinylated and immunoprecipitated. The reaction was stopped with 75 μl stop solution per tube after incubation for 15 min at room temperature and centrifuged at 300 g for 10 min. The supernatant was discarded and 5 ml cold PBS was added to each tube followed by centrifugation at 300 g for 10 min. The cells were lysed in 1 ml lysis buffer (50 mm core buffer, 150 mm NaCl, 0·1 mg/ml PMSF, 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1% Nonidet P-40 and 0·5% sodium deoxycholate) and incubated for 30 min on ice. After incubation for 15 min the cells were resuspended and centrifuged at 12 000 g for 10 min at 4° and the supernatants were transferred to clean Eppendorf tubes.
Immunoprecipitation was essentially carried out with protein G agarose beads as described (Roche). The supernatants were mixed with 1 μg antibody at 4° over night followed by centrifugation at 12 000 g at 4° for 20 seconds. Subsequently, the supernatants were discarded and the beads were resuspended in 1 ml washing buffer, and centrifuged again at 12 000 g at 4° for 20 seconds, the same procedure was repeated twice. After washing, 20 μl reducing buffer (2×, containing 0.15 g dithiothreitol in 5 ml 62.5 mM TRIS, 10% glycine and 2.5% SDS buffer) was mixed with the beads and heated at 95° for 4 min and subsequently centrifuged at 7000 g for 1 min to spin down the beads and the proteins were separated on SDS–PAGE gels. Proteins were transferred to the Hybond ECL membrane (GE Healthcare Biosciences, Uppsala, Sweden) and detected using the BMC chemiluminescence's blotting kit (Roche).
The samples were separated on SDS–PAGE gels and blotted onto a nitrocellulose membrane (Amersham), blocked over night with PBS, 4% BSA and 0·5% Tween. Filters were washed with PBS with 1·5% BSA and incubated with antibodies. ECL Western blotting detection reagents were used for detection with Hyperfilm TM (Amersham).
Collagen type 1 was diluted in serum-free RPMI-1640 and H2O (8/1/1), applied in plastic Petri dishes 1 ml/dish (30 mm; BD Biosciences, Franklin Lakes, NJ) and allowed to polymerize at room temperature. A total of 1·0 × 106 cells in AIM-V medium was added to each well with and without antibodies and allowed to migrate for different times. The cells were fixed in 2·5% glutaraldehyde for 10 min or in 2% paraformaldehyde for 20 min for immunocytochemistry and washed twice with PBS. Cell morphology and cell migration were evaluated in nine fixed positions in each well and at 50-μm intervals throughout the gel by the use of an inverted microscope (Nikon Eclipse TE300) and a digital depth meter (Heidenheim ND221). The results are given as mean number of infiltrating cells/field (× 20 objective) per infiltration depth (50 μm for the first two layers immediately beneath the gel surface and 100 μm for other layers further down). The infiltrating cells were identified in situ in the collagen gels using immunocytochemistry after fixation in paraformaldehyde. Migration was also analysed in a modified Boyden assay (transwell assay) using 8-μm nucleopore filters. The lower wells of 48-well Boyden chambers were filled with RPMI containing 1 mg/ml BSA whereupon the coated filters were placed in the chambers. The upper chambers were filled with 50 μl of 2 × 106 cells/ml in AIM-V medium with and without antibodies. Following incubation for 1 hr the number of cells in the lower chamber was counted in triplicate. Transendothelial migration was studied using a flow system created using a Pharmacia peristaltic pump and a flow chamber device consisting of an upper compartment containing circulating cells under shear stress separated from a lower compartment containing CCL5 by 8-μm human umbilical vein endothelial cell-coated inserts treated with tumour necrosis factor-α (TNF-α) 2 ng/ml for 12 hr before the experiment.[26, 27] The monolayer was overlaid with CXCL12 for 5 min and AF24 T cells were perfused into the chamber in Hanks' balanced salt solution (Gibco Ltd), 1 mm Ca2+, 1 mm Mg2+, 10 mm HEPES and 2 μg/ml BSA (Sigma) and allowed to accumulate on the endothelial monolayer for 3 min. Subsequently the flow was increased to 2 dyne/cm2 for 10 min. This procedure was repeated six times and transmigrated cells were counted.
To study cell adhesion, plastic Petri dishes (90 mm; Heger A/S, Svaddeveien, Norway) were coated with ICAM-1 (2 μg/ml), BSA (10 μg/ml) or fibronectin (10 μg/ml), and extensively washed before use. To analyse static adhesion the cells (10 000/position) in AIM-V medium were incubated on the substrates for different times, fixed in 2·4% cold glutaraldehyde for 10 min or in 2% paraformaldehyde for 20 min for immunocytochemistry and unbound cells were removed by gentle aspiration. To analyse adhesion of free-floating cells these were treated as described under Cells and allowed to adhere. The number of adherent cells per microscope field (20× objective) was counted. Cell adhesion was routinely, unless otherwise stated, evaluated in six fixed positions in triplicate. To examine de-adherent T lymphocytes with respect to surface expression of various components cells were allowed to adhere for 30 min, washed and incubated for 20 min. Subsequently, the medium containing cells that had detached during the 20-min period was removed. In addition, loosely adherent cells were washed away and pooled with the ones detached during the 20-min incubation. These cells were then analysed using biotinylation and immunoprecipitation.
AF24 T cells were mixed with Betv G75-pulsed B cells, allowed to settle in a Petri dish and incubated at 37° for 45 min. After incubation the cells were fixed in 2% paraformaldehyde and analysed using immunocytochemistry. T cells forming conjugates were quantified as a percentage of all T cells in the sample.
Staining intensity in immunocytochemistry experiments, number of migrating cells and adherent cells are presented as mean arbitrary units ± SD and the Mann–Whitney U-test was used to evaluate differences between groups. For determination of differences in cell proliferation paired Student's t-test was used. Two-tailed P values < 0·05 were considered statistically significant.
Ustekinumab arrests T-cell motility and targets LRP1 and TSP-1
The influence of various monoclonal antibodies on the motility of T cells from healthy individuals was examined under serum-free conditions to exclude any interference of exogenous proteins and peptides through determination of polarized shape and migration in three-dimensional type 1 collagen matrices. This is a well established model for analysis of lymphocyte motility and infiltration of tissues where the cells move independent of adhesive interactions with the substrate.[28-30] Ustekinumab abrogated the development of a motile morphology in virtually all T cells and inhibited their capacity to migrate into collagen gels (Fig. 1a,b). This effect of ustekinumab was concentration-dependent, maximal at or above 10 μg/ml and detectable at a concentration of 1 μg/ml but not at 0·1 μg/ml. Ustekinumab therefore inhibited motility in T cells without preference for any specific subset. In contrast, infliximab, a monoclonal antibody that binds to TNF-α, did not affect cell shape and migration. The majority of the cells inside the collagen were identified as CD3 positive (generally > 95%) using immunochemistry. Removal of ustekinumab restored motility completely within 2 hr, showing that the inhibitory effect was reversible (Fig. 1c). We have also examined the possible influence on T-lymphocyte motility of etanercept, a TNF-receptor II fusion protein, adalimumab a human monoclonal antibody against TNF-α, antibodies to integrin β2 (efalizumab) as well as mouse monoclonal antibodies to integrins β1 and β2 with negative results (not shown). This result was therefore consistent with earlier data by others showing that T-cell motility could not be blocked by anti-integrins β1, β2, β3 or αv antibodies.
The influence of ustekinumab on transendothelial migration of a birch allergen-specific T-cell clone (AF24) was examined using a flow system device that allows the transmigrated cells to be harvested (Fig. 1d). Confluent endothelial cells pre-treated with TNF were overlaid with CXCL12 10 ng/ml and the subendothelial compartment filled with medium containing CCL5 100 ng/ml. This combination of apical and subendothelial chemokines and shear stress has been shown to enhance transmigration, which was confirmed by the results in Fig. 1(d). Pre-incubation and infusion of AF24 with ustekinumab were found to inhibit transmigration to the lower compartment whereas infliximab did not affect transmigration (Fig. 1d) indicating that ustekinumab-induced inhibition of motility prevented transendothelial migration.
To further examine the influence of ustekinumab on T-cell motility we examined migration of cells from healthy individuals on ICAM-1 and fibronectin (Fig. 2a,b) and the capacity of AF24 T cells to form conjugates with antigen-presenting HLA-identical B cells in the presence of the drug under static conditions (Fig. 2c). Ustekinumab inhibited migration on ICAM-1 and fibronectin and decreased conjugate formation significantly whereas infliximab did not affect conjugate formation. This showed that ustekinumab inhibits T-cell motility and indicated that ustekinumab-induced inhibition of motility interferes with the formation of conjugates.
In addition to its effect on motility, ustekinumab at the same concentrations inhibiting motility markedly increased the cell surface expression of LRP1 and TSP-1 and simultaneously decreased the intracellular TSP-1 and LRP1 (Fig. 2d,e). Ustekinumab rendered 70–100% of all cells in separate experiments TSP-1- and LRP-positive on the surface (not shown). Ustekinumab did not increase the expression of CD4. Infliximab (Fig. 2d) and efalizumab (not shown) did not affect the expression of TSP-1 and LRP1. ELISAs to determine if ustekinumab reacted with TSP-1 yielded negative results (mean intensity 0·042 compared with the positive control 0·946, P < 0·01, and 0·049 in the negative control). The siRNA silencing of LRP1 (effects on the cells by silencing of LRP1 is described in section on Free-floating cells, expression of TSP-1 and LRP1 and adhesion, below) abrogated the reactivity of ustekinumab with the cells (mean intensity of immunostaining in arbitrary units 11 compared with 39 without silencing, P < 0·001). This indicated that ustekinumab targets LRP1 or an LRP1-associated component distinct from TSP-1.
Cell surface expression of LRP1 determines the effect of ustekinumab
To investigate the mechanism responsible for the increased cell surface expression of LRP1 and TSP-1 induced by ustekinumab we examined the possible influence of inhibitors of various cell functions (Fig. 3a,b). The protein synthesis inhibitor cycloheximide did not affect the stimulatory effect of ustekinumab on the cell surface expression of TSP-1 and LRP1, which argues against the possibility that the antibody increased the synthesis of LRP1 and TSP-1. We also examined whether ustekinumab inhibited T-cell motility through stimulation of exocytosis or inhibition of endocytosis of TSP-1 and LRP1. Dynamin is a ubiquitously expressed isoform of the dynamin family of large GTP-ases implicated in the regulation of exocytosis and endocytosis. Dynasore, a pharmacological inhibitor of dynamins,[33, 34] abrogated the increased cell surface expression of LRP1 induced by ustekinumab, indicating an important role of dynamin for T-cell expression of LRP1. In contrast, dynasore markedly increased the cell surface expression of TSP-1 and potentiated the stimulatory effect of ustekinumab on TSP-1 expression (gel data are shown in Fig. 3). Brefeldin A, which blocks transport from endoplasmic reticulum to the Golgi apparatus, inhibited the cell surface expression of TSP-1 and LRP1 (not shown) and the effect of ustekinumab on TSP-1 and LRP1 expression (Fig. 3a,b). Colchicine, a microtubule inhibitor that inhibits endocytosis,[36, 37] increased the cell surface expression of both TSP-1 and LRP1 (Fig. 3a,b). Dynasore inhibited whereas colchicine increased T-cell motility (Fig. 3c). Dynasore also inhibited the potentiating effect of colchicine on LRP1 expression as well as the colchicine-induced up-regulation of T-cell motility.
Dynasore provoked formation of a prominent cap-like accumulation of TSP-1 on the surface of many cells (Fig. 3d,e). Ustekinumab also increased the number of cells with a cap-like accumulation of cell surface TSP-1. Dynasore and ustekinumab therefore appear to have the same effect on the behaviour of TSP-1 on the cell surface. Infliximab, efalizumab and cycloheximide did not affect the pattern of cell surface expression of TSP-1.
The results in Fig. 3 indicated that LRP1 is transported to the cell surface where LRP1 promotes motility and internalizes TSP-1, whereas TSP-1 accumulates on the cell surface in the absence of LRP1. Figure 3 further indicated that ustekinumab inhibits this internalization. The cap-like localization of TSP-1 in the absence of LRP1 may reflect redistribution of TSP-1 within the plane of the plasma membrane or polar appearance and inhibition of subsequent spreading of TSP-1 over the cell surface. The cap-like localization of TSP-1 probably implies that LRP1 restricts the mobility of TSP-1 within the plasma membrane. LRP1 therefore seems to play a key role for T-cell motility and determines the behaviour of TSP-1 at the cell surface.
Ustekinumab inhibits the effect of CXCL12 on LRP1 and TSP-1
We have previously demonstrated that the chemokine CXCL12 induces cell surface expression of TSP-1 and LRP1 coupled to T-cell motility. In the light of the inhibitory effect of ustekinumab on spontaneous T-cell motility coupled to effects on the cell surface expression of LRP1 and TSP-1 we examined whether ustekinumab also affected CXCL12-induced motility and the influence of CXCL12 on TSP-1 and LRP1 expression. Ustekinumab was found to inhibit CXCL12-induced lymphocyte migration into collagen (Fig. 4a) and also inhibited the stimulatory effect of CCL5 on T-cell motility (not shown). Gel analysis of immunoprecipitated material from surface-biotinylated cells in Fig. 4(b) showed that CXCL12 enhanced the cell surface expression of LRP1 in comparison with control cells and induced surface expression of prominent intact 170 000 MW TSP-1 and a 130 000 MW TSP-1 band. Ustekinumab inhibited the CXCL12-induced prominent enhancement of the cell surface expression of intact TSP-1 and appearance of the 130 000 MW TSP-1 band (Fig. 4b). This supported the conclusion that ustekinumab targets LRP1 at the cell surface and inhibits LRP1-dependent processing of TSP-1.
Western blotting of whole cell material using another anti-TSP-1 antibody identified the CXCL12-induced cell surface-expressed 170 000 and 130 000 bands as TSP-1 (Fig. 4c). A comparison of cell surface TSP-1 (Fig. 4b) and whole cell TSP-1 in the absence of CXCL12 (Fig. 4c) demonstrated prominent intracellular 130 000 and 110 000 forms of TSP-1 but little cell surface TSP-1. It is also evident from Fig. 4(c) that CXCL12 prevented expression of the 110 000 MW TSP-1 band. Western blotting of whole cell material further showed that the 130 000 MW protein did not react with an antibody to the N-terminal domain of TSP-1 (Fig. 4d) and neither did the 110 000 MW band react with this antibody (not shown). This antibody identified a 40 000 MW band besides intact TSP-1 in CXCL12-exposed cells indicating that the 130 000 MW band appeared as a consequence of N-terminal cleavage of TSP-1 (Fig. 4d).
Dynasore abrogated the cell surface expression of LRP1 supporting the conclusion based on the results in Fig. 3 that LRP1 is transported to the cell surface through a dynamin-dependent process (Fig. 3b). Dynasore further induced cell surface expression of a prominent 80 000 MW band precipitable with anti-TSP-1 antibodies (Fig. 4b). This further supported the conclusion based on the results in Fig. 3 that the behaviour of TSP-1 is LRP1-dependent and points to the possibility that TSP-1 undergoes defective or alternative processing in the absence of LRP1.
Adhesion mimics CXCL12 in its effect on LRP1 and TSP-1
We next examined the influence of adhesion of cells to ICAM-1 and fibronectin on the cell surface expression of LRP1 and TSP-1. Adhesion to ICAM-1 and fibronectin was found to enhance expression of LRP1 and intact TSP-1 and induced a 130 000 MW TSP-1 band (Fig. 5a). This adhesion-induced TSP-1/LRP1 expression was indistinguishable from the TSP-1/LRP1 pattern induced by CXCL12 in non-adherent motile cells in Fig. 4. CD4, CD29 and actin were expressed to the same extent before and after adhesion. In contrast to adherent cells, which exhibited cell surface expression of LRP1 and 170 and 130 000 MW TSP-1 bands, de-adherent cells expressed a barely visible 130 000 MW band and a prominent 110 000 MW TSP-1 band and down-regulation of LRP1 at the cell surface (Fig. 5b).
Cells adherent to ICAM-1 (Fig. 5c,d) and fibronectin (not shown) exhibited a polarized cell shape. Ustekinumab prevented this polarization and the ustekinumab-treated cells showed circumferential apolar spreading along the entire cell circumference (Fig. 4c,d). CXCL12 increased the number of adherent cells compared with control cells (not shown) and cells with ustekinumab, and enhanced polarization (Fig. 5c,d). Cells adhering in the presence of both CXCL12 and ustekinumab exhibited apolar spreading similar to cells with ustekinumab only indicating that ustekinumab inhibited the polarization induced by CXCL12 (Fig. 5c,d). The induction of apolar spreading was readily reversible upon removal of ustekinumab (not shown). In contrast to ustekinumab, the control antibody infliximab did not affect the pattern of cytoplasmic spreading (Fig. 5c,d). The fact that ustekinumab inhibited polarized cell shape while enhancing cytoplasmic spreading indicated that this antibody targets a motogenic mechanism that is anti-adhesive.
Free-floating cells, expression of TSP-1 and LRP1, and adhesion
Flow rapidly up-regulates the capacity of T cells to adhere so it was of interest to examine the influence of the free-floating state on T-cell expression of TSP-1 and LRP1. Interestingly, free-floating cells exhibited up-regulation of intact TSP-1 on the cell surface in comparison with control cells but little LRP1 (Fig. 6). Subsequent adhesion of the free-floating cells to ICAM-1 or fibronectin-induced expression of LRP1 and a 130 000 MW TSP-1 band. Free-floating cells were therefore distinguishable from adherent cells and CXCL12-exposed cells with respect to cell surface expression of TSP-1 and LRP1.
To analyse the influence of LRP1 on adhesion we examined adhesion of T cells to ICAM-1 under static and free-floating conditions after silencing of LRP1 (Fig. 7a,b; see Supporting information, Fig. S1). Cells transfected with scrambled control siRNA adhered markedly better under free-floating than under static conditions. Transfection with LRP1 siRNA induced a significant up-regulation of adhesion to ICAM-1 even under static conditions. A separate LRP1 siRNA had the same effect, supporting this conclusion (not shown). The results in Fig. 7(a,b) showed that LRP1 is counter-adhesive and that flow overcomes this counteradhesive effect.
The enhancing effect of ustekinumab on adhesion in Fig. 5 and our previous finding that knockdown of TSP-1 reduces adhesion indicated that TSP-1 supports adhesion. To further examine the influence of TSP-1 on adhesion we examined adhesion of AF24 T cells to ICAM-1 under static conditions in the presence of different concentrations of exogenous TSP-1 (Fig. 7c). TSP-1 at a concentration of 0·01 μg/ml had a suggestive stimulatory effect on adhesion. At a concentration of 0·1 and 1 μg/ml TSP-1 clearly stimulated adhesion, whereas higher TSP-1 concentrations did not increase adhesion. A peptide mimetic of the CD47-binding C-terminal sequence in TSP-1, 4N1K, inhibited basic adhesion and the stimulatory effect of intact TSP-1 on adhesion whereas a scrambled control peptide did not inhibit adhesion (Fig. 7d). This indicated that intact TSP-1 promotes adhesion through CD47. 4N1K also reduced adhesion to ICAM-1 in the presence of CXCL12 and ustekinumab to enhance spreading as described in Fig 5(d) from 211 ± 32 cells per microscope field to 76 ± 22 (P < 0·01). The number of adherent cells per field with CXCL12 only was 79 ± 24, with ustekinumab only was 139 ± 32 (P < 0·05 versus CXCL12) and with CXCL12, ustekinumab and scrambled 4N1K 204 ± 40 (P > 0·05 versus CXCL12 + ustekinumab). This indicated that CXCL12 in the presence of ustekinumab stimulated adhesion to ICAM-1 through a mechanism that was dependent on TSP-1 and CD47.
The dependence of adhesion to ICAM-1 on TSP-1 and CD47 and the evidence that CXCL12 and ustekinumab stimulated adhesion to ICAM-1 through TSP-1 and CD47 suggested that TSP-1 transformed stimulation by CXCL12 and integrin ligands (Figs 4 and 5) into an adhesion signal. To elucidate this possibility we examined whether the TSP-1-binding site in calreticulin (CRT19-36), which is complexed to LRP1,[16, 17] and induces T-cell adhesion through the N-terminal of TSP-1 affected the influence of intact exogenous TSP-1 on adhesion to ICAM-1 after inhibition of the synthesis of endogenous TSP-1 using cycloheximide (see legend to Fig. 7e). After inhibition of TSP-1 synthesis CRT19-36 itself merely had a weak stimulatory effect on adhesion, whereas the level before inhibition of synthesis was four or five times higher than with a scrambled control peptide (not shown). However, CRT19-36 had a pronounced stimulatory effect on adhesion in the presence of exogenous TSP-1 that was inhibitable by 4N1K (Fig. 7e). This further strengthens the conclusion that intact TSP-1 stimulates adhesion through CD47 and indicates that calreticulin complexed to LRP1 triggers adhesion in response to stimuli that increase the expression of TSP-1 and LRP1 on the cell surface.
These results confirm and extend previous findings that T-cell motility is regulated by a counteradhesive motogenic mechanism directed by LRP1 and endogenous TSP-1. One important conclusion that can be drawn from the effects of ustekinumab is that motility and adhesion/spreading most likely are regulated at the cell surface level (Figs 1-3). A possible explanation for the inhibition of the LRP1- and TSP-1-directed motogenic mechanism by ustekinumab is that it reflects the exceptional broad ligand-binding capacity of LRP1, which recognizes > 30 structurally distinct ligands. Therefore, ustekinumab may either bind directly to LRP1 or to an LRP1-associated ligand. The effects of ustekinumab, dynasore and colchicine indicated that LRP1 and TSP-1 are transported to the cell surface as separate entities where they interact and then disappear through endocytosis (Fig 3). LRP1 seems to exert a profound control of the behaviour of TSP-1 within the plasma membrane as demonstrated by the cap-like localization and increased expression in the absence of LRP1 in dynasore-treated as well as in ustekinumab-treated cells (Fig. 3d,e).
Perhaps the most important conclusion that can be drawn from the present results is that CXCL12 and contact with ICAM-1 and fibronectin up-regulated a mechanism for integrated regulation of motility and adhesion by stimulating transport of TSP-1 and LRP1 to the cell surface. CXCL12 and integrin ligands induced surface expression of intact TSP-1 and a 130 000 MW TSP-1 fragment while preventing expression of 110 000 MW TSP-1 (Figs 4 and 5) so probably counterbalancing disappearance of TSP-1 and LRP1 through endocytosis. This effect was distinguishable from and more dramatic than that of IL-2, which stimulates LRP1-dependent motility through a persistent up-regulation of the synthesis of TSP-1. In contrast to motility induced by CXCL12 and integrin ligands, the constitutive motility correlated with a low cell surface expression of LRP1 and TSP-1, probably owing to less transport to the cell surface and concomitant LRP1-dependent endocytosis. The fact that ustekinumab inhibited motility and enhanced adhesion/spreading, while protecting intact TSP-1 and preventing formation of the 130 000 MW TSP-1 fragment, indicates that this fragment is motogenic, whereas intact TSP-1 promotes adhesion. This conclusion is supported by the previous findings that TSP-1 knockdown reduces adhesion and inhibition of motogenic signalling protects intact cell surface TSP-1 while enhancing adhesion. Additionally, exogenous TSP-1 stimulated T-cell adhesion, and a peptide mimetic of the CD47-binding sequence in TSP-1 prevented this effect, indicating that intact TSP-1 promotes adhesion through CD47 (Fig. 7). The free-floating state up-regulated the cell surface expression of intact TSP-1 (Fig. 6), which suggests that the stimulation of adhesion by shear flow depends on TSP-1. A role of intact TSP-1 and CD47 for adhesion signalling is further supported by the previous and present results showing that the C-terminal domain of TSP-1 promotes a pronounced adhesion and spreading process via CD47 in response to triggering signals delivered at the N-terminus by the TSP-1 binding site of calreticulin (Fig. 7). Because calreticulin is associated with LRP1 on the cell surface[16, 17] the LRP1/calreticulin complex probably delivers an adhesion signal to CD47 through intact TSP-1 that is abrogated by TSP-1 cleavage and formation of the 130 000 MW fragment. It is reasonable to assume that the strength of this adhesion signal depends on the amount of intact TSP-1 and that the association of CD47 to lymphocyte function-associated antigen-1 and very late antigen-4 controls ligand binding by these integrins by stabilizing the high-affinity state of the integrin–ligand binding site.
The TSP-1 interaction with CD47 that promotes adhesion may seem inconsistent with the stimulatory effect on motility of the C-terminal CD47 binding site in TSP-1 alone. However, only intact TSP-1 stimulates adhesion whereas the C-terminal CD47 binding site inhibits adhesion (Fig. 7). The balance between intact TSP-1 with capacity to trigger adhesion through CD47 and cleaved 130 000 MW TSP-1, which enhances motility through LRP1, therefore plays a critical role for the control of T-cell adhesion and motility. The inhibition of formation of a de-adhesion-coupled 110 000 MW TSP-1 fragment suggests that CXCL12 and integrin ligands enhance adhesion through this mechanism, which is a possible consequence of protease inhibition by high surface expression of intact TSP-1. De-adhesion further correlated with down-regulation of cell surface LRP1, which suggests that the formation of the 110 000 MW TSP-1 fragment represents an inhibitory step in the regulation of motility and adhesion coupled to the disappearance of LRP1.
The inhibitory effect on T-cell motility of LRP1 knockdown coupled to enhancement of adhesion was previously shown to be mimicked by an inhibitor of the Janus kinase–signal transducer and activator of transcription 1 (JAK/STAT) pathway. The present results showed that ustekinumab mimicked LRP1 silencing and inhibition of the JAK/STAT pathway in its effects on motility and adhesion as well as in the enhancing effect on the expression of intact TSP-1. This indicates that ustekinumab inhibits LRP1-dependent motogenic signalling through the JAK/STAT pathway and TSP-1 processing, and therefore promotes adhesion through TSP-1 via CD47. This is supported by the finding that induction of migration by plasminogen activator inhibitor-1 in smooth muscle cells and endothelial cells requires LRP1-dependent activation of JAK.[40-42] The effects of ustekinumab indicate that integrin-binding tissue components and CXCL12 up-regulate motility through LRP1-mediated counter-adhesive JAK signalling and stimulate adhesion through calreticulin-LRP1-triggered signalling through CD47 via intact TSP-1. Interestingly, extracellular signal-regulated kinase (ERK) activity seems to support T-cell adhesion as well as adhesion of non-lymphoid cells suggesting that CD47 and integrins trigger adhesion through ERK.
Our present and previous findings[14, 19] taken together indicate that CXCL12 and integrin ligands up-regulate a dominant motogenic mechanism, directed by LRP1-mediated JAK/STAT signalling and 130 000 MW TSP-1, and an associated LRP1-calreticulin-triggered adhesion process through intact TSP-1 and CD47, as depicted in the model in Fig. 8(a). The adhesion process is probably critically dependent on up-regulation of intact TSP-1 by shear flow, CXCL12 and integrin ligands. As further depicted in Fig. 8(b) this mechanism for integrated regulation of motility and adhesion is proposed to promote transendothelial migration. The fact that CXCL12 and integrin ligands trigger the same mechanism may account for the fact that integrins can bypass the need for chemokine signals to promote transendothelial migration and endows this mechanism with great flexibility. A particularly important feature of this mechanism is that it prioritizes motility while preventing permanent adhesion. This prioritization of motility is likely to play a key role for the capacity of T cells to constantly reposition themselves within the organism.
The CXCL12-induced and adhesion-induced up-regulation of the cell surface expression of LRP1 and TSP-1 and motility raises the possibility that the functional plasticity of T-cell subsets may reflect up-regulation or down-regulation of the motogenic LRP1/TSP-1 mechanism. Forkhead box P3 Treg cells for example represent a distinct lineage committed to suppressive functions via multiple mechanisms, but it is also conceivable that motility and adhesion determine the capacity of Treg cells to execute their function. This is supported by the evidence that CXCL12 maintains Treg cell function and that Treg cells exhibit enhanced migration.
The evidence that dynamin-dependent transport of LRP1 to the cell surface was necessary for motility is consistent with studies in non-lymphoid cells indicating that dynamin controls a kiss-and-run mechanism for fusion of post-Golgi vesicles with the cell surface. Dynamin has previously been shown to regulate T-cell activation by controlling actin polymerization at the immunological synapse, pointing to the possibility that synapse formation also involves LRP1. This contention is supported by the fact that ICAM-1, as shown here, activates the motogenic TSP-1/LRP1 mechanism and is a key structure on antigen-presenting cells involved in the formation of immunological synapses.
The accessibility on the cell surface together with the fact that it regulates infiltrative properties of T cells makes the motogenic mechanism described here a potential therapeutic target for tailored treatment of inflammatory conditions. Accordingly, this mechanism has multiple antigenic structures for monoclonal antibodies or hybrid molecules to interfere with T-cell infiltration of various organs during the course of autoimmune and allergic disorders as well as during rejection of foreign transplants and graft-versus-host disease. Ustekinumab has a beneficial effect in the treatment of psoriasis and Crohn's disease,[48, 49] which is consistent with the present findings. A different therapeutic approach taking advantage of the motogenic LRP1/TSP-1 pathway may be to treat patients with stimulators of this pathway so as to enhance the function of Treg cells.[5, 50]
In conclusion, these results indicate that T cells are programmed to reposition themselves within the organism through a cell surface mechanism for integrated regulation of motility and adhesion that prioritizes motility over permanent adhesion. This mechanism is controlled by chemokines, interactions with integrin ligands and shear flow. Such integration of distinct regulatory influences through a common mechanism may be of fundamental functional importance, and also provides a conceptual background for better understanding of other regulatory connections in T cells as well as non-lymphoid cell types including infiltrative properties of neoplastic cells. The motogenic mechanism may provide a useful background for development of functional T-cell markers and a versatile therapeutic target for treatment of T-cell-mediated conditions.
This study was supported by grants from the Edvard Welander Foundation and the Finsen Foundation.
The authors declare no conflict of interest.
TT analysed data and contributed to the design of the study and to the writing of the manuscript. EB performed research work. KGS designed the study, performed research work, analysed data and wrote the manuscript.