The P-selectin glycoprotein ligand-1 (PSGL-1) is involved in the initial contact of leukocytes with activated endothelium, and its adhesive function is regulated through its proteolytic processing. We have found that the metalloprotease ADAM8 is both associated with PSGL-1 through the ezrin–radixin–moesin actin-binding proteins and able to cause the proteolytic cleavage of this adhesion receptor. Accordingly, ADAM8 knockdown increases PSGL-1 expression, and functional assays show that ADAM8 is able to reduce leukocyte rolling on P-selectin and hence on activated endothelial cells. We conclude that ADAM8 modulates the expression and function of PSGL-1.
ADAM family is a group of transmembrane glycoproteins that possess proteolytic and signaling properties and are implicated in both cell adhesion and cell fusion processes 1. ADAMs are usually activated by furin or other convertases as well as auto-catalytically, in the case of ADAM8 2. It has been described that ADAM8 cleaves important cell surface proteins 3, 4, cytokines and growth factors 5. ADAM8 is overexpressed under several pathological conditions involving inflammation and remodeling of the extracellular matrix, including malignant diseases and asthma 6–8.
P-selectin glycoprotein ligand-1 (PSGL-1), through its interaction with P-, E- and L-selectins, mediates the tethering and rolling of leukocytes on endothelial cells prior to their extravasation 9, 10, triggers the activation of transcription factors like cFos 11 in leukocytes and induces the generation of tolerogenic DCs which promote the differentiation of Treg cells 12. Although it was described that PSGL-1 was a substrate of the proteases BACE1 and ADAM10 13, neither the physiological context of cleavage nor the mechanism responsible for its shedding have been identified so far. In this work, we demonstrate that in leukocytes ADAM8 associates with PSGL-1 through ezrin–radixin–moesin (ERM) proteins, and that this interaction modulates the expression and function of this adhesion receptor.
Results and discussion
Association of PSGL-1 with ADAM8
To identify intracellular molecules able to associate with PSGL-1, we performed a proteomic analysis of HL-60 cell lysates pulled down with the cytoplasmic tail of PSGL-1 fused to GST (GST-PSGL-1cyt) 11. This analysis revealed the presence of a 98-kDa protein that corresponded to ADAM8 (data not shown). Additional pull-down experiments performed with fragments of the cytoplasmic tail of PSGL-1 fused to GST, detected an additional protein of 75–80 kDa, which likely corresponded to a cleaved form of ADAM8 (Fig. 1A, left panel). To map the region of PSGL-1 involved in this interaction, pull-down experiments from lysates of HL-60 cells were performed with different fragments of the cytoplasmic tail of PSGL-1 fused to GST. We found that the 18 juxtamembrane amino acids of PSGL-1 were sufficient to mediate its interaction with ADAM8 (Fig. 1A, right panel).
We next assessed whether endogenous ADAM8 and PSGL-1 were associated. ADAM8 was first immunoprecipitated from HL-60 cell lysates, and WB analysis of the immunoprecipitated molecules showed that PSGL-1 was present in the ADAM8 immunoprecipitate (Fig. 1B). Likewise, reciprocal analysis revealed the presence of ADAM8 in the PSGL-1 immunoprecipitate (Fig. 1C), thus confirming the association of these two endogenous molecules in the HL-60 cell line. To assess whether the PSGL-1/ADAM8 association also occurred in primary myeloid cells, we performed experiments using human neutrophils. PSGL-1 was found in the ADAM8 immunoprecipitate from resting neutrophils (Fig. 1D, left panel) and the PSGL-1/ADAM8 association increased after neutrophil activation with PMA (Fig. 1D, right panel), indicating that PSGL-1 and ADAM8 were also associated in neutrophils.
ERM proteins link PSGL-1 and ADAM8
Since we had observed the presence of a cluster of basic residues in the juxtamembrane cytoplasmic tail of ADAM8 (Fig. 1E, upper panel underlined residues) that has been reported to act as binding sites for ERM proteins 11, we performed immunoprecipitation (IP)/WB experiments to assess whether ADAM8 could also associate with the ERMs and found the presence of ERM proteins in the ADAM8 immunoprecipitates (Fig. 1E, lower panel). Additional assays showed that in ERM-knocked down cells PSGL-1 failed to associate with ADAM8 (Fig. 1F). Therefore, these results strongly suggest that ERM proteins link PSGL-1 and ADAM8. In the activated state, the ERM molecules can interact with other ERM molecules. Our data suggest that two or more molecules of ezrin or moesin, that are interacting with each other, could also be interacting with different membrane receptors, acting as connectors between membrane proteins, as it could be the case for ADAM8 and PSGL-1. This interaction would allow their localization at specific protrusive membrane regions, favoring their binding to ligands 14 and facilitating the shedding of adhesion receptors during the rolling phase.
Proteolytic cleavage of PSGL-1 by ADAM8
The reports that PSGL-1 can be processed by a protease 15, 16 and that PSGL-1 could be a candidate substrate for ADAM8 17, in addition to our finding of PSGL-1/ADAM8 association, prompted us to investigate the potential involvement of ADAM8 in the cleavage of PSGL-1. Experiments with neutrophils incubated with activated sADAM8 showed a significant diminution in the membrane expression of PSGL-1 (Fig. 2A, left panel), while the expression of CD11a and CD44 did not change (Fig. 2A, right panel). Accordingly, when PSGL-1 and ADAM8 were co-expressed by J77 T cells, a significant reduction of PSGL-1 expression was detected (Fig. 2B). In contrast, a mutated form of ADAM8, which lacks catalytic activity 3, did not affect the amount of PSGL-1 in the cell surface. Accordingly, higher amount of soluble PSGL-1 (sPSGL-1) was detected in the supernatant of the cells transfected with active ADAM8, whereas transfection with the inactive form of ADAM8 did not change the amount of sPSGL-1 released by cells transfected with the empty plasmid (Fig. 2C). Furthermore, the silencing of ADAM8 in HL-60 cells with siRNA increased the amount of PSGL-1 and L-selectin on these cells (Fig. 2D). To further demonstrate that ADAM8 is able to cleave PSGL-1, we incubated human recombinant PSGL-1 with thermolysin-activated sADAM8 and found that PSGL-1 is processed by ADAM8 in a cell-free system (Fig. 2E). Altogether, these data demonstrate not only that ADAM8 is associated with PSGL-1 but also that ADAM8 is able to proteolytically modulate the expression of this adhesion receptor.
ADAM8 can regulate PSGL-1 rolling function
We next determined the functional consequences of PSGL-1 processing by ADAM8, by testing the effect of sADAM8 on the rolling of CEM cells on P-selectin and on the rolling of THP1 on HUVEC. As expected, the pre-incubation with activated sADAM8 significantly reduced the ability of the CEM cells to roll on P-selectin (Fig. 3A) and the ability of THP1 cells to roll on TNF-α-HUVEC activated (Fig. 3B). Moreover, the number of rolling HL-60 cells on histamine-activated HUVEC increased two-fold upon knockdown of ADAM8 expression (Fig. 3C). This result strongly indicates that ADAM8 controls the function of this adhesion receptor.
It has been reported that ADAM8 is involved in the pathogenesis of inflammatory diseases such as experimental autoimmune arthritis or airway hypersensitivity reactions 6, 7, 18. Our previous work showed that ADAM8 promoted the shedding of L-selectin 4. In this report, we describe a role of ADAM8 in the shedding of PSGL-1 and its effect on leukocyte rolling on P-selectin and on activated endothelial cells. PSGL-1 mediates the initial contact of leukocytes with the activated endothelium through P-selectin. In addition, L-selectin mediates their secondary rolling on the extravasating leukocytes. Therefore, it is tempting to postulate that during the inflammatory phenomenon ADAM8 could modulate the extravasation of leukocytes through the shedding of both L-selectin and PSGL-1. This effect on leukocyte extravasation could account for the role of ADAM8 in the pathogenesis of inflammatory diseases. Hence, we propose that ADAM8 could be a potential therapeutic target in inflammatory conditions through the modulation of its expression and/or function.
In this work, we show that PSGL-1 and ADAM8 associate in leukocytes and that the ERMs act as connector proteins. We propose that the activation triggered during leukocyte contact with the activated endothelium would activate the proteolytic activity of ADAM8 which then would be able to process both PSGL-1 and L-selectin, and hence influence the leukocyte infiltration during inflammation. We also propose that the observed consequences of a deregulated ADAM8 in inflammatory diseases, as allergic asthma, would be explained through its association with PSGL-1, which probably regulates the expression and activation of this adhesion receptor during leukocyte extravasation.
Materials and methods
Antibodies, reagents, plasmids and cells
ADAM8 mAb (clone 143338), human recombinant ADAM8 ectodomain, human recombinant P-selectin and human recombinant chimeric Fc-PSGL-1 were from R&D Systems (Minneapolis, MN, USA). Rabbit anti-human ADAM8 polyclonal antibody (clone H-50) and ADAM8 siRNA were from Santa Cruz Biotechnology (CA, USA). Anti-PSGL-1 KPL-1 mAbs were from BD Pharmingen (San José, CA, USA). The rabbit anti-ezrin 90/3 polyclonal antibody was generously provided by Dr. Heinz Furthmayr. The mAbs CD44 and CD11a were generated in our group. The expression vector pcDNA3 containing full-length ADAM8 and its catalytically inactive form (H604A, H608A) have been previously described 3. GST-PSGL-1 constructs were described elsewhere 11. Ezrin and moesin siRNA primers: EZRIN sense: 5′-UCCACUAUGUGGAUAAUAA-3′, ezrin antisense: 5′-UUAUUAUCCACAUAGUGGA-3′; moesin sense: 5′-AGAUCGAGGAACAGACUAA-3′, moesin antisense: 5′-UUAGUCUGUUCCUCGAUC-3′ were purchased from Sigma-Aldrich Chemical (St. Louis, MO, USA). sPSGL-1 7 ELISA kit was from Bender Medsystem (Vienna, Austria).
Human cell lines and HUVEC were cultured as described elsewhere. Human neutrophils were isolated and cultured as described 4.
Immunoprecipitation, western blot and pull down of proteins with GST-fused proteins
HL-60 and neutrophils, resting and activated with PMA (15 min at 37°C), were processed for immunoprecipitation and western blot as described elsewhere. Pull down of proteins from HL-60 cell extracts was performed as previously described 11.
All nucleofection assays were performed following Amaxa nucleofector instructions. J77 cells were nucleofected with 5 μg of either full-length or mutated pcDNA3-ADAM8 with Amaxa nucleofector. For RNA silencing, HL-60 cells (2×106 cells) were nucleofected with 1.6 μg of ADAM8 siRNA, ERM siRNA 19 or scramble siRNA. ADAM8 and ezrin/moesin expression was checked by immunoprecipitation and WB. PSGL-1 expression was analyzed by ELISA, WB and flow cytometry 24 h after transfection with ADAM-8 siRNA and 72 h after transfection with ERM siRNA.
Soluble ADAM8 proteolytic activity
Human neutrophils suspended in HBSS were pre-incubated with either thermolysin (0.75 μg/mL), recombinant non-activated sADAM8 or thermolysin-activated sADAM8 (20 min at 37°C). Surface expression of PSGL-1, CD11a and CD44 was analyzed by flow cytometry. ELISA assay was performed following manufacturer' instructions. PSGL-1-Fc chimera was incubated for 2 h 30 min a 37°C with thermolysin-activated ADAM-8 (1:1 proportion) and analyzed by Coomassie staining.
Plates were incubated overnight at 4°C with 10 μg/mL human rP-selectin and blocked for 1 h at 37°C with 2% BSA. HUVEC were seeded in plates previously incubated with 20 μg/mL fibronectin (Sigma-Aldrich Chemical) for 1 h at 37°C, grown to confluence and stimulated for 6 h with 20 ng/mL TNF-α or for 10 min with 100 μM histamine. THP1 and CEM cells were pre-incubated 2 h at 37°C with 0.75 μg/mL thermolysin, 2.8 mg of non-activated ADAM8 or thermolysin-activated (30 min at 37°C) ADAM8 and then incubated for 10 min at 4°C in the presence or absence of 50 μg/mL of anti-PSGL-1 Ab. HL-60 was previously nucleofected with siRNA ADAM-8 and incubated for 24 h at 37°C. Treated cells were injected at 2 (CEM and THP1 cells) and 0.75 (HL60 cells) dynes/cm2, in the flow chamber (Glycotech, MD, USA) and assayed for rolling experiment as described elsewhere.
Results were expressed as arithmetic mean±standard deviation (SD) or standard error (SE) of the mean. Wilcoxon signed rank test was used to determine significant differences between means.
This work was supported by grants from the Spanish Ministry of Health (Fondo de Investigaciones Sanitarias) to A. Urzainqui (FIS-PI080894) and to F. Díaz González (FIS 09/02209), by grants from Ministerio de Ciencia e Innovación to F. S. M. (SAF2008-02635) and to F. M. (SAF2008-02251) and by the Redes RIER, RTICC (RD06/0020/1037) and RECAVA del Instituto de Salud Carlos III and RTICCR.
Conflict of interest: The authors declare no financial or commercial conflict of interest.