Membrane-type 1-Matrix Metalloproteinase Regulates Intracellular Adhesion Molecule-1 (ICAM-1)-mediated Monocyte Transmigration^*

The migration of leukocytes through the endothelium to sites of infection or inflammation is a key process for the maintenance of physiological defense mechanisms. When this process is dis-regulated and becomes chronic, inflammatory diseases such as arteriosclerosis and arthritis manifest. The steps in leukocyte transmigration (TM)⁵ are initiated though activation of the endothelial cells (ECs) by cytokines such as TNF-α, interleukin-1, and interleukin-6. The selectins initiate the rolling and tethering of leukocytes to the endothelium. This step permits the engagement between β₂ and β₁ integrins with intercellular cell adhesion molecule (ICAM-1) and vascular cell adhesion molecule (VCAM-1) to allow firm adhesion and spreading of leukocytes. ECs express low levels of ICAM-1 and VCAM-1, but cytokine stimulation elevates the expression of these receptors on ECs. The migration of leukocytes through the EC barrier involves platelet endothelial cell adhesion molecule-1 and junctional adhesion molecules. Finally, proteolytic degradation of the basement membrane extracellular matrix by metalloproteinases (MPs) in particular is required to promote extravasation (reviewed in Refs. 1 and 2).

Although ICAM-1, VCAM-1, and platelet endothelial cell adhesion molecule-1 play an important role in in vivo and in vitro experimental models of TM (1–4), there is little data on the mechanistic role of the individual adhesion molecules. ICAM-1 comprises of five immunoglobulin-like motifs on the extracellular surface, followed by a stem, a transmembrane domain, and a short cytoplasmic tail (5). The cytoplasmic tail contains three tyrosine residues, two of which become phosphorylated at positions 485 and 474 upon ligation to modulate ICAM-1 function (6, 7). The role of these residues in leukocyte-endothelial migration has not yet been defined.

Proteolysis is an important step during and after transmigration (8, 9), as degradation of the basement membrane and matrix of the media (of larger vessels) or stroma is required. The zinc-dependent MPs belonging to the metzincin family possess a highly conserved catalytic domain, yet have an enormous range of functions and substrates. MPs can be divided into major families: matrix MPs (MMPs) and the family called A Disintegrin And Metalloproteinases (ADAMs). The majority of the MMPs are secreted proteins, but there are six membrane-bound members belonging to a subclass of membrane-type MMP (MT-MMPs). These proteinases degrade extracellular matrices and promote the shedding of cell surface proteins (10, 11). The MT-MMPs also promote cell migration, metastasis, and angiogenesis (12). In particular, the shedding of adhesion molecules CD44 and syndecan by MT1-MMP has been linked to cell migration (13, 14).

The ADAMs are primarily involved in ectodomain shedding of wide variety of cell surface molecules (10, 11). ADAMs regulate adhesion via interaction with integrins through their Disintegrin/Cysteine-Rich domains (15) and are increasingly being associated with aspects of vascular pathology (16, 17). Our previous study identified ADAM-17 (or TACE, TNF-α converting enzyme) as the protease responsible for mediating the regulated cleavage of ICAM-1 in 293 cells induced through phorbol 12-myristate 13-acetate (18).

The secreted MMPs are regulated and effectively inhibited by each of the four known endogenous tissue inhibitors of MPs (TIMPs). TIMP-2 and -3 are very effective against MT-1, -2, -3, and -5 MMPs, whereas TIMP-1 is a very weak inhibitor of these MT-MMPs (19). The glycosylphosphatidylinositol-anchored members MT4 and 6-MMP are inhibited by all four TIMPs. TIMPs are also selective against ADAMs, for example, ADAM-10 is inhibited by TIMP-1 and -3, whereas ADAM-17 is inhibited by TIMP-3 only. To date, TIMP-2 has not been shown to inhibit any of the ADAMs (20–22).

Although previous studies have indicated a role for members of the MMP family in leukocyte TM (23, 24), the role of ICAM-1cytoplasmic sequence and the proteolytic mechanism regulating leukocyte TM has been unclear. Our results for the first time indicate a clear role for the cytoplasmic sequence of ICAM-1 in regulating TM. The studies indicate that MT1-MMP mediates TM and interacts with ICAM-1. MT1-MMP and ICAM-1 co-localize at the membrane ruffles of ECs. This interaction also results in the ectodomain cleavage of ICAM-1.

EXPERIMENTAL PROCEDURES

Reagents—Monoclonal antibodies (mAb) against CD11a, -11b, and -18 were from BioLegend (San Diego, CA). The polyclonal antibody against the N terminus of active MT1-MMP, used for inhibition of transmigration, was obtained from Cedarlane Laboratories Ltd. (Ontario, Canada). TIMP-1, TIMP-2, and TIMP-3 polyclonal antibodies were from Chemicon International Inc. (Temecula, CA). LB-2 (anti-CD54), a mAb directed against the ectodomain domain of ICAM-1, and a VCAM-1 mAb was purchased from BD Biosciences Immunocytometry Systems (San Jose, CA). R98, a polyclonal antibody directed against the cytoplasmic tail of ICAM-1 was developed in our laboratory (18, 25), as was R98017 against the ICAM-1 ectodomain. TNF-α was obtained from PeproTech (Rocky Hill, NJ). The ECL plus system was from Amersham Biosciences. The enzyme-linked immunosorbent assay kit for the detection of human soluble ICAM-1 was purchased from Diaclone (Cell Sciences Inc., Canton, MA).

Cells—Human aortic ECs purchased from Clonetics (San Diego, CA) were grown in DMEM/F-12 supplemented with 20% fetal calf serum (fetal calf serum) and endothelial growth medium-2 with EGM-2 supplements (7, 25). Human dermal microvascular endothelial cells (Clonetics) were grown in EGM-2 supplemented with EGM™-2 SingleQuots (Clonetics). THP-1 cells were maintained in DMEM/F-12 supplemented with 5% fetal calf serum and 1.0 mm l-glutamine at 37 °C, 5% CO₂, and 100% humidity. NautCell™ 293 (Microbix Biosciences Inc., Toronto, Canada) and HeLa (European Collection of Cell Cultures) were cultured in DMEM, 10% fetal calf serum, 1 mm l-glutamine, and penicillin/streptomycin at 5% CO₂, 37 °C, 100% humidity. Human embryonic kidney (HEK) 293 cells stably expressing ICAM-1 (293_ICAM-1) were described previously (6, 18, 25). Monocytes were isolated from blood drawn from healthy human volunteers, using a Percoll gradient technique (31). The adhesion of monocytes to ECs, performed using the Rose Bengal staining method, was described previously (18, 25).

Retroviral Vectors of ICAM-1—Using high fidelity Pfu Turbo DNA polymerase (Stratagene) full-length ICAM-1 cDNA was amplified by PCR using ICAM-1-pcDNA3.1 as the template (6, 18, 25). Using the sense primer, 5′-taaagcggccgccatggctcccagcagc-3′, and the antisense primer, 5′-ccgactggatccctgtcccgggataggttca-3′, a Not-1 site at the 5′-end and a BamH1 site at the 3′-end were created, respectively. A 3′-GFP-blasticidin fusion gene was amplified using the template pTracer-CMV/Bsd (Invitrogen) to create Age1 and Pml1 restriction sites at the 5′- and 3′-ends, respectively. Amplified ICAM-1 was cloned into the retroviral vector pUC-MMP-IC-eGFP (26) at NotI-BamH1 sites with expression to be driven by an internal promoter. The GFP-blasticidin fusion gene was cloned in the same vector upstream of ICAM-1 at the Age1-Pml1 sites to be driven by the viral long terminal repeat promoter. DNA sequencing verified the sequence of the full-length ICAM-1. A pUC-MMP-IC-eGFP plasmid clone coding for the full-length open reading frame of ICAM-1WT was used as a template to generate the ICAM-1 mutants to replace tyrosine residues at positions 474, 476, and 485 with alanine or to delete a cytoplasmic region (ICAM-1TR) with the QuikChange site-directed mutagenesis kit (Stratagene) (7). Retrovirus was produced in 293-T cells by triple transfection: (a) the packaging plasmid that expresses gag and pol genes (pcG-Ori-Gag-Pol), (b) the plasmid encoding the envelope protein (pcG-vsv-G) (26), and (c) the vector plasmid (pUC-MMP-IC-eGFP) expressing ICAM-1 or its mutants, using Lipofectamine. Viral supernatants, harvested 48 h after transfection, were used to infect EC in the presence of Polybrene (8 μg ml^-1). The expression of ICAM-1 and its mutants was analyzed by flow cytometry (fluorescence-activated cell sorting) and by Western blot analysis (6, 18, 25) (Fig. 2). The ICAM-1-transfected ECs are henceforth termed as “EC_ICAM-1.”

MT1-MMP-expressing Adenovirus—Adenovirus was prepared using the AdMax system from Microbix Biosystems, Inc. The MT1-MMP cDNA isolated from human ECs was cloned into the vector pDC516 and sequenced. MT1-MMP cDNA was PCR-amplified to create HindIII and Sal1 restriction sites at 5′- and 3′-ends, respectively. The vector, pDC516, and the MT1-MMP inserts were double digested separately with HindIII and Sal1 and ligated together. The Ad5 ΔE1/E3 adenoviral genome vector pBHGfrt(ΔE1,E3)FLP and MT1-MMP-pDC515 were co-transfected into NautCell™ 293s at a ratio of 5:1 using Fugene6 (Roche Diagnostics GmbH) in suspension in serum-free medium for 10 min prior to seeding in T25 flasks. Once the cytopathic effect was complete, individual expressing clones were isolated as described by Nicklin and Baker from the crude recombinations (27). MT1-MMP expression was checked by Western blot with sheep anti-human polyclonal antibody N175/6 (28) of infected HeLa lysates. The absence of replication-competent virus was established by infection of HeLa and monitoring for cytopathic effect. High titer stocks were prepared by infecting 12 T175 flasks of NautCell™ 293s and purifying recombinant adenovirus from cell lysates by CsCl gradient ultracentrifugation. Titer was estimated by end-point dilution in 293 cells, and consistency of virus quality was determined by comparison with the titer estimated by BCA assay.

The cDNA encoding E240A MT1-MMP in pCDNA3.1 zeo+ was provided by Christian Roghi (Cambridge Research Institute, Cancer Research UK, Cambridge, UK). The cDNA, subcloned by standard PCR methods incorporating SalI and NheI sites at the 3′- and 5′-ends of the primers, was ligated in pDC515 and then sequenced to check for errors before co-transfection with the adenoviral genome vector in Naut™ 293 cells as described for wild-type MT1-MMP.

Transmigration Assay—EC_ICAM-1 (4 × 10⁴) were seeded on polycarbonate microporous membranes of 6.5-mm diameter and 5-μm pore size in Transwell chambers (Costar, Cambridge, MA) precoated with human fibronectin (10 μg ml^-1)at37°C for 2 h. 100 μl of EC medium was added to the top chamber, and 0.6 ml was added to the lower chamber. After 3–4 days of culture, the integrity of the EC monolayer was assessed by microscopic observation and by measuring the permeability of the monolayer using fluorescein isothiocyanate-albumin or horseradish peroxidase. Monocytes were freshly isolated using a Percoll gradient. Prior to the TM assay, the media in the Transwell was replaced with 100 μl of monocytes (2 × 10⁵ cells) on the EC monolayer. The inserts were transferred to a new 24-well plate (lower chambers) containing 0.6 ml of fresh RPMI plus supplemented with 5% fetal calf serum and Ca²⁺/Mg²⁺ (1 mm)at 37 °C in 5% CO₂. After 2 h the media in the lower chamber, containing migrated monocytes, were collected. Monocytes attached to the bottom of the Transwells were detached using 5 mm EDTA in DMEM (15 min at 4 °C) and then counted under a light microscope.

In select experiments, the EC_ICAM-1 monolayer or the monocytes were pretreated with the test or the control antibodies (20 μg ml^-1) for 30 min. The antibodies were maintained throughout the assay. Also, EC_ICAM-1 cells were preincubated at 37 °C with the pharmacological inhibitors for 1 h prior to the addition of monocytes.

siRNA Transfection—MT1-MMP siRNA, Oligo-2, was purchased from Dharmacon (Lafayette, CO) and was transfected into EC_ICAM-1 using GenePorter Transfection Reagent (Gene Therapy Systems, San Diego, CA). EC_ICAM-1 cells were seeded on fibronectin-coated Transwells for 48 h. For transfection, 100 nm siRNA was allowed to complex with 4 μl of transfection reagent in DMEM for 45 min. The complex was overlaid on the EC monolayer in 100 μl at 37 °C. After 15 min, 0.6 ml of serum containing growth media was added to the bottom chamber for 18 h (18). EC mRNA were analyzed using TaqMan QT-reverse transcription-PCR 48 h after transfection with commercially available primer-probe sets (Applied Biosystems Inc., UK) for MT1, -2, -3, and -4-MMP to demonstrate specificity. Cell lysates were analyzed by Western blot to confirm depletion of MT1-MMP at the protein level.

Adenoviral Infection—High titer adenovirus expressing TIMPs, MT1-MMP, and the RAd60 control virus were produced as described (32, 33 and “Experimental Procedures”). EC_ICAM-1, seeded on Transwells, was infected at the required multiplicity of infection (m.o.i.) in 100 μl of media. TM assay was performed 18 h later, and then fresh media was added.

In Vitro Collagenase Assay—HeLa were seeded overnight at a density of 3 × 10⁵ cells well^-1 of a 6-well dish in DMEM, 10% fetal calf serum, glutamine, penicillin, streptomycin, 37 °C, and 5% CO₂. The medium was changed for a 1-ml volume containing no adenovirus or control adenovirus (RAd60), wild-type MT1-MMP, or E240A MT1-MMP expressing virus for 2 h and at an m.o.i. of 100 pfu cell^-1 before the medium was replaced with 3 ml fresh medium and cultured for 48 h. Cells were then scraped into 150 μl of collagenase assay buffer (20 mm Tris-HCl, pH 7.8, 10 mm CaCl₂, 0.05% Brij-35 (v/v), 0.025% sodium azide (w/v), Complete™ EDTA-free mixture inhibitor (Roche Applied Science) and then lysed by passing through a 20-gauge needle three times. Cell lysates were then incubated on ice with 20 μg ml^-1 of the anti-MT1-MMP antibody (directed against a peptide corresponding to the N terminus of active MT1-MMP) or nonspecific rabbit IgG for 1 h. The lysates were then mixed with DQ-collagen (Molecular Probes) at a final concentration of 100 μg ml^-1 in wells of a 96-well plate and incubated overnight in the dark at 37 °C. Fluorescence was measured on a TECAN spectrafluor plus microtiter plate reader at an excitation of 485 nm and emission of 595 nm. Infections were performed in duplicate, lysates were pooled, and samples were then read in duplicate on the 96-well plate.

Expression of His-tagged MT1-MMP—A His₆ tag was added to MT1-MMP using the template MT1-MMP cDNA. MT1-MMP(His) cDNA was PCR-amplified with high fidelity DNA polymerase Pfu Turbo (Stratagene) using designed primers. The resulting PCR product created HindIII at the 5′-end and EcoR1 and His₆ at the 3′-end for inserting into pcDNA3.1. HEK 293 cells were transfected with MT1-MMP(His)-pcDNA3.1 using Lipofectamine (18). After 48 h, the cells were lysed, and MT1-MMP(His) was purified using the MagneHis™ Protein Purification System (Promega, Madison, WI) as per the manufacturer's instructions.

ICAM-1: MT1-MMP Binding Assay—50 μg ml^-1 of purified His-tagged MT1-MMP was diluted 1:1 with PBS. 96-well plates were coated with 100 μl of His-tagged MT1-MMP at 4 °C for 16 h. Wells coated with bovine serum albumin served as negative control. The wells were washed twice with wash buffer (PBS with 0.05% Tween-20, pH 7.2) and blocked with 5% bovine serum albumin. Bovine serum albumin was removed, and the plates were dried for 30 min at room temperature and then incubated with 100 μl of 293_ICAM-1 cell lysate (1 × 10⁶ cell ml^-1 lysate) for 2 h at room temperature. The plates, washed three times, were incubated with a biotinylated ICAM-1 polyclonal antibody for 2 h. The plates were then incubated with streptavidin-horseradish peroxidase conjugate for 20 min and washed. The tetramethylbenzidine substrate was added for color development. The reaction was stopped with 1 n H₂SO₄, and the plates were read at 450 nm using a plate reader.

Immunolocalization of MT1-MMP and ICAM-1—Human aortic endothelial cells stably infected with ICAM-1WT, ICAM-1Y485A, or ICAM-1TR were seeded on 10-mm coverslips and 12-well plates in EGM-2 medium for 48 h prior to fixing in 4% para-formaldehyde in PBS for 20 min at room temperature. The coverslips were then incubated in PBS, 100 mm glycine, pH 7.5, for 5 min before incubation in PBS, 0.01% Triton X-100 for 5 min. The coverslips were then incubated with mouse anti-ICAM-1 (LB-2) and sheep anti-MT1-MMP (N-175/6) at 10 μg ml^-1 each in PBS for 1 h at room temperature. After incubation with the primary antibodies the coverslips were washed three times for 5 min in PBS before incubating with anti-mouse Texas Red (Jackson Laboratories) and anti-sheep Alexa488 secondary (Molecular Probes) antibodies at 1:200 and 1:500 in PBS for 1 h at room temperature. After washing for three times 5 min in PBS and briefly in Milli-Q water, the coverslips were mounted in Vectashield (Vecta Laboratories). Images were taken on a Leica TCS SP5 confocal microscope using a 63× oil objective and sequential scans to detect each fluorochrome.

RESULTS

To evaluate the selective role of ICAM-1 in leukocyte TM, ECs were transfected to express ICAM-1 using a viral vector. ICAM-1 expression was controlled to mimic levels that are achieved by TNF-α stimulation (Fig. 1A). VCAM-1 levels were negligible in ICAM-1-transfected ECs. However, ICAM-1 and VCAM-1 were highly expressed when non-transfected ECs were treated with TNF-α (Fig. 1A). Surface expression of ICAM-1 and VCAM-1 was confirmed by fluorescence-activated cell sorting analysis.

To examine the capacity of the transfected ICAM-1-expressing ECs (EC_ICAM-1) to regulate monocyte TM, the cells were grown on Transwells for 2 days to develop a monolayer and secrete extracellular matrix proteins. The integrity of the monolayer was monitored using fluorescein isothiocyanate-labeled albumin or horseradish peroxidase. A negligible amount of albumin (<1%) passed through the EC monolayer (data not shown). Human blood monocytes (2 × 10⁵) were layered on EC, and 2 h later the cells that transmigrated through the ECs into the lower chamber were counted. An average of 6.6 × 10⁴ monocytes transmigrated through EC_ICAM-1, whereas only 1.5 × 10⁴ monocytes migrated through ECs not expressing ICAM-1 (Fig. 1B). Interestingly, the numbers of monocytes migrating through EC_ICAM-1 and TNF-α-stimulated ECs (non-transfected) were very similar. The specific monocyte TM was ∼25.5% of the monocytes applied. In subsequent results, we have subtracted the number of monocyte transmigrating through non-ICAM-1 EC from the number transmigrating through EC_ICAM-1 and expressed this value as 100% TM. Henceforth, TM through EC_ICAM-1 cells is expressed as % ICAM-1-dependent TM in all subsequent results.

The ICAM-1-blocking mAb LB-2 (18, 29) blocked monocyte TM by >75% (Fig. 1C). Antibodies against the α-subunits of the β₂ integrin, anti-α_L (CD11a) and anti-α_M (CD11b), effectively blocked TM. The anti-β₂ (CD18) mAb completely abolished monocyte migration (Fig. 1D). Together, the above results establish the specificity of our assay and demonstrate that retrovirally transfected ICAM-1-expressing ECs mediate leukocyte TM.

Because the cytoplasmic tail of ICAM-1 has been reported to become phosphorylated and regulate ICAM-1 function (6, 7), we examined whether deletion of the cytoplasmic tail and tyrosine substitutions at positions 474, 476, and 485 would affect TM. The ICAM-1 expression for each mutant was equalized, by titrating each virus to produce ICAM-1 levels closely matching that produced in the presence of TNF-α (15 ng ml^-1 and 16 h). The Western blots of the cell lysates probed with an ectodomain specific antibody (R-94017) (6, 7) show equal expression of the mutations to that noted with WT. An antibody specific to the cytoplasmic tail (R98) confirmed the absence of the cytoplasmic sequence in ICAM-1TR (Fig. 2A). However, TM through ICAM-1TR was drastically reduced to ∼10% of ICAM-1WT. Similarly, ICAM-1Y485A also showed a dramatic reduction (down to 16.3% of control) in the capacity to support TM. Although ICAM-1Y474A reduced TM by 55%, ICAM-1Y476A did not affect TM (Fig. 2B). These results indicate that the cytoplasmic tail and the specific tyrosine residues within it play a role in regulating TM. To establish that ICAM-1 cytoplasmic mutations did not affect their adhesive functions, monocyte adhesion to the mutant cells was compared with ICAM-1WT. The time course of monocytic cells adhesion to EC expressing ICAM-1Y485A was unaltered and was comparable to WT at early time points (1 h). The adhesion was CD18-dependent in each case (supplemental Fig. S1, A and B). At 2- to 4-h time points, the adhesion to ICAM-1Y485A cells was marginally lower than ICAM-1WT.

The results shown in Figs. 1 and 2 were obtained using human aortic ECs. Because transendothelial leukocyte migration occurs in vivo primarily through the microvessels, which branch off the larger blood vessels, we used human dermal microvessel ECs to verify the above results. Almost identical results as shown in Figs. 1 and 2 were obtained with microvessel ECs. However, with microvessel ECs, the monocyte TM was consistently 16–20% higher than that observed with aortic ECs (data not shown).

We sought to identify the proteinase(s) regulating the ICAM-1-dependent TM. The MMP/ADAM family of MPs breakdown extracellular matrix proteins and regulate the migration of a variety of cell types (8, 9). Inhibitors of MPs (TNF-α protease inhibitor-1 and MMP inhibitor-III) blocked ICAM-1-dependent monocyte TM (Fig. 3A). Because these hydroxamic acid-based inhibitors inhibit MMPs, ADAMs, and ADAM-TS family members, we utilized TIMPs to obtain a more specific profile of inhibition so as to better understand the identity of the MP involved. TIMP-1, -2, and -3 were virally transfected into EC_ICAM-1. TIMPs were transfected at an m.o.i. of 25–100 pfu cell^-1, or with a control virus, RAd60, containing the CMV promoter alone. The specific expression of TIMPs, 48 h after transfection, is shown in Fig. 3B. Monocyte TM, through EC_ICAM-1 expressing various TIMPs, indicated that TIMP-2 and TIMP-3 blocked TM, whereas TIMP-1 had no effect (Fig. 3C), and the effect was dose-dependent. At an m.o.i. of 100 pfu cell^-1, TIMP-3 inhibited TM by 70%, whereas TIMP-2 was 62% effective. The kinetics of monocyte adhesion to ICAM-1Y485A and ICAM-1WT transfected with TIMPs was found to be very similar (supplement Fig. S1C).

Physiologically, leukocyte TM occurs when the endothelium is activated by inflammatory cytokines such as TNF-α. We next examined the effect of TIMPs on TM in TNF-α-stimulated ECs (Fig. 3D). The profile of inhibition was very similar to that observed in Fig. 3C. Although at 100 pfu cell^-1, TIMP-2 and -3 blocked only 40 and 55% of TM, respectively. Nevertheless, this result is reassuring and indicates that TIMP-2- and -3-sensitive proteases regulate TM. Monocytic cells were also induced to express TIMPs through adenoviral transfection to examine if the cellular context of TIMP expression affected TM across EC_ICAM-1. The transmigration of TIMP-2 and -3 overexpressing monocytic cells was also inhibited, although to a lesser degree than when TIMP-2 and -3 were expressed in ECs alone (Fig. 3E). The TM profile in the presence of TIMPs suggests that MT-MMPs, rather than secreted MMPs or ADAMs, mediate monocyte TM.

Through elimination, we surmise that one or more of the remaining MT1-, MT2-, MT3-, and MT5-MMPs regulate TM. To identify the specific MT-MMP, we focused on MT1-MMP, because this protease exhibits broad substrate specificity against numerous extracellular matrix proteins and is implicated in tumor metastasis, cell migration, and angiogenesis (12); however, its role in monocyte TM through ECs is only now becoming evident (24). We initially utilized antibodies against MT1-MMP (Fig. 4A). Anti-MT1-MMP blocked TM by ∼50%, whereas a control isotype-matched antibody had no effect. The function-blocking capacity of the anti-MT1-MMP antibody was assessed for its ability to neutralize collagenase activity of HeLa cells expressing MT1-MMP. The anti-MMP antibody selectively inhibited collagenase activity of MT1-MMP by ∼45% (Fig. 4B), whereas in cells expressing mutant MT1-MMP (MT1-EA) the inhibition was much less. The blocking of TM by the function blocking anti-MMP antibody provides evidence for the role of MT1-MMP in TM, and this was further verified by MT1-MMP knockdown utilizing siRNA. MT1-MMP siRNA knockdown in ECs was firstly verified by TaqMan reverse transcription-PCR (Fig. 4C) and Western blot (Fig. 4D). We then introduced the siRNA in ECs and in monocytic cells to evaluate its effect on each cell type. TM was abolished almost equally, irrespective of the cells in which the siRNA was introduced (Fig. 4E).

MT1-MMP is present on both ECs and monocytic cells. Monocytic cell stimulation with phorbol 12-myristate 13-acetate or dibutyryl-cAMP promotes a phenotypic change from monocyte to macrophage and increases expression of MT1-MMP (Fig. 5A) (30). We examined whether elevated MT1-MMP levels in ECs or monocytes would augment TM. MT1-MMP overexpressed in ECs by adenoviral transfection consisted of multiple forms corresponding in mass to the pre-pro, pro, active, and the 45-kDa forms lacking the catalytic domain, which indicates elevated MT1-MMP activity (31) as detected by Western blot. Unlike HT1080s (31), aortic ECs appear to produce a single degraded form on overexpression (Fig. 5B). N-terminal sequencing has shown different cell lines produce different truncated MT1-MMP products (32, 33). Adenoviral transfection of monocytes shows little accumulation of the pro and active forms of MT1-MMP in comparison and a clear difference in processing to the inactive form (Fig. 5C). Adenoviral overexpression of MT1-MMP in ECs dramatically augmented TM (Fig. 5D). Even at a low m.o.i. of 12.5 pfu cell^-1, MT1-MMP increased TM >2-fold. At the same level of transfection in monocytic cells, the increase in TM was marginally less, although total expression levels were approximately equal (Fig. 5D). This difference in TM may be due to a lack of stability of active MT1-MMP in monocytes indicated by the significant increase in degraded forms. Nevertheless, these data in conjunction with the TIMP transfection data (Fig. 3) suggest perhaps MT1-MMP on EC may play a prominent role in TM. Again, using microvessel EC, we noted that TIMPs 2 and 3, but not TIMP-1 blocked TM and overexpression of MT1-MMP in the microvessel EC also augmented TM (data not shown).

Leukocyte TM is reduced in ECs expressing ICAM-1 lacking the cytoplasmic sequence and tyrosine substitutions (Fig. 2). To examine if the introduction of MT1-MMP in cells lacking the capacity to support TM would rescue the phenotype, ECs expressing ICAM1TR and ICAM-1Y485A were transfected with MT1-MMP adenovirus and compared with a control viral vector (Fig. 6). The results indicate that ICAM-1Y485A cells were rescued and gained TM function by 85% at a level of infectivity of 25 pfu cell^-1. MT1-MMP transfection in ICAM-1TR cells also rescued TM function but to a lesser extent than ICAM-1Y485A cells. The kinetics of monocyte adhesion and TM on ICAM-1Y485A cells was found to be comparable to that of ICAM-1WT cells when transfected with MT1-MMP and with various TIMPs (supplemental Fig. S1, D and E). Taken together, these results indicate a role for MT1-MMP in ICAM-1-dependent leukocyte TM.

We hypothesized that MT1-MMP directly regulates TM function by locating proximal to and interacting with ICAM-1. The results in Fig. 7A provide indication of ICAM-1 and MT1-MMP interaction in a binding assay. In this assay, purified MT1-MMP was bound to plastic dishes, and lysates from 293 cells expressing ICAM-1WT, and its mutants, were allowed to bind MT1-MMP. Equal expression of ICAM-1 and the mutants was verified (6, 25). The bound ICAM-1 was detected with an antibody detection system. Low levels of ICAM-1 were found to bind bovine serum albumin (data not shown), whereas the binding of ICAM-1 to MT1-MMP was found to be high and concentration-dependent. However, very low amounts of ICAM-1TR or ICAM-1Y485A bound to MT1-MMP. These results provide evidence for ICAM-1 binding to MT1-MMP and suggest the importance of the ICAM-1 cytoplasmic sequence in mediating interaction with MT1-MMP.

To further confirm our observation that MT1-MMP and ICAM-1 associate using the in vitro binding assay, we used confocal microscopy to determine if co-localization at the cellular level could be observed. ICAM-1WT and endogenous MT1-MMP only significantly co-localized at the ruffled edge of ECs (Fig. 7B). Otherwise ICAM-1 was predominantly cell surface-localized and MT1-MMP-localized in a perinuclear compartment, which has been previously identified to be the trans-Golgi network and endosomes (data not shown) (34). In ICAM-1TR cells, ICAM-1 staining at membrane ruffles was not observed to the same extent as ICAM-1WT, and in ICAM-1Y485A ruffle staining was not observed. In both instances, staining of endogenous MT1-MMP at membrane ruffles was not observed (Fig. 7B), indicating localization of MT1-MMP at membrane ruffles is driven by ICAM-1 expression and is dependent on its intracellular domain.

We have presented evidence that MT1-MMP interacts with ICAM-1 and that these two proteins co-localize at the cell surface. If MT1-MMP and ICAM-1 are in fact co-localized then there is the possibility that MT1-MMP could exert its enzymatic activity upon ICAM-1. We tested this possibility and examined whether ICAM-1 becomes cleaved in the presence of MT1-MMP. EC_ICAM-1 cells were transfected with TIMP-1, -2, and -3, and 24 h later the media was harvested and soluble ICAM-1 levels were quantified (18). ICAM-1 cleavage in EC_ICAM-1 was inhibited by TIMP-2 and -3 but not TIMP-1 (Fig. 8A) thus suggesting a role for MT-MMPs in the ectodomain cleavage of ICAM-1. Overexpression of MT1-MMP augmented ICAM-1 shedding (Fig. 8B). This result suggests that ICAM-1 is a substrate of MT1-MMP.

DISCUSSION

Our studies indicate the following: 1) ICAM-1, an adhesion receptor on ECs, independently mediates monocyte TM; 2) ICAM-1 cytoplasmic domain and the tyrosines at position 474 and 485 regulate TM; 3) MT1-MMP is the key protease mediating TM; 4) MT1-MMP and ICAM-1 interact and co-localize at the cell membrane; and 5) ICAM-1 is a substrate for MT1-MMP.

ICAM-1-dependent monocyte TM levels were close to that achieved with TNF-α stimulated ECs (Fig. 1A). In these studies, the specificity was demonstrated utilizing antibodies against ICAM-1 and its counter-receptors, lymphocyte function associated antigen-1 and macrophage-1 antigen. The profile of inhibition (Fig. 3), that is, sensitivity to TIMP-2 and TIMP-3 and insensitivity to TIMP-1, suggested that MT-MMPs are the likely candidates mediating TM. Because MT1-MMP is involved in pericellular proteolysis and in extracellular matrix remodeling, tumor metastasis, and angiogenesis (12), we examined the role of MT1-MMP in TM. Overexpression of MT1-MMP in EC_ICAM-1 dramatically augmented TM (Fig. 5), whereas siRNA directed against MT1-MMP blocked TM by 70%. The result was verified using function blocking antibody against MT1-MMP (Fig. 4). Because TIMP-2 and -3 when transfected into either ECs or monocytic cells blocked TM, this indicated that MT1-MMP on both cell types has the potential to regulate TM. Studies with siRNA directed against MT1-MMP (Fig. 4) also pointed toward the same conclusion. The requirement of MT1-MMP on both ECs and monocytic cells could suggest a distinct function for the protease on each cell type in the TM process.

TIMP-1 inhibits MT4- and MT6-MMPs, and also ADAM10. Similarly ADAM17 is only TIMP-3 sensitive (20), whereas ADAM8 and 9 are insensitive to either of the TIMPs (22). Therefore, these proteases are unlikely to play a major role in monocyte TM. The synthetic MP and TIMP inhibition we see with monocyte transmigration also bears resemblance to the study of Faveeuw et al. (23), which described inhibition of lymphocyte transmigration with a hydroxamate inhibitor but not TIMP-1 in vivo. In this case the authors attribute the lack of TIMP-1 inhibition of transmigration due to lack of inhibition of L-selectin shedding by ADAM17 (TACE). The lack of TIMP-1 inhibition, in our system, indicates a role for MT-MMPs in TM. We have shown that ADAM17 mediates PKC-dependent ICAM-1 shedding in ECs stimulated with TNF-α (18). However, this mechanism of regulation of ICAM-1 by TACE may not be important in TM, as indicated by the TIMP inhibition profile. We have also found that TACE knockdown in ECs with siRNA has no effect on transmigration in EC_ICAM-1 cells and neither does adenoviral overexpression of TACE.⁶ We are therefore beginning to gain a better understanding of the role of specific MPs in TM.

Although the extracellular portion of ICAM-1 mediates leukocyte adhesion, the intracellular cytoplasmic sequence regulates TM (Fig. 2). Deletion of the 28 amino acids of the cytoplasmic sequence abolished monocyte TM. TM was also abrogated in ICAM-1Y485A cells, whereas in ICAM-1Y474A TM diminished by 54% but, at the proximal substitution Y476A, TM was unaffected (Fig. 2). These results suggest a key role for the cytoplasmic sequence in TM. We have previously shown that Tyr-485 and Tyr-474, but not Tyr-476, become phosphorylated upon ICAM-1 ligation (6) and that this phosphorylation mediates interaction with the Src-homology domain 2 containing phosphatase-2 (SHP-2). The cytoplasmic tail and the phosphoresidues regulate the ectodomain cleavage of ICAM-1 and EC survival (7). Therefore, monocyte TM through ICAM-1 ligation may be dependent on a variety of signals generated through the cytoplasmic interactions, and these could include mitogen-activated protein kinase (6, 7), Src, pCas (35), and effectors such as the small GTPase, Rho (36). The cytoplasmic tail of ICAM-1 directly associates with SHP-2 (6) and the actin-binding protein, α-actinin (37). The requirement of Rho in leukocyte TM has been indicated (36) as it is for ICAM-1-dependent TM, although in this case using Chinese hamster ovary cells (38). Because the cytoplasmic sequence interacts directly and indirectly with such a wide variety of proteins, it is no surprise that we were only partially able to rescue the TM function when MT1-MMP was introduced in ICAM-1TR cells. However, ICAM-1Y485A cells largely regained TM function in the presence of MT1-MMP, even under low level of MT1-MMP viral infection. This positive result suggests that, because the defect in ICAM-1Y485A cells is mild, it can be reverted. However, because ICAM-1Y485A cells also exhibit aberration in other ICAM-1 intracellular functions (6, 7), this gain of TM function result needs to be taken cautiously until further studies such as relocation of ICAM-1 and MT1-MMP within specific microdomains of the cell membrane are performed in ICAM-1WT and mutant cells.

MT1-MMP mediates the ectodomain shedding of CD44, an adhesion receptor that binds to hyaluronan. Cleavage of CD44 renders motility to the cells. MT1-MMP has been shown to be physically associated with CD44. This interaction is through the extracellular hemopexin domain of MT1-MMP (39). MT1-MMP also cleaves other adhesion molecules such as syndecan-1 and the α_v subunit of the integrin α_vβ₃ that are involved in cell migration mediated through MT1-MMP (13, 40). The mechanistic role of MT1-MMP in ICAM-1-dependent TM remains unclear. Our results indicate that ICAM-1 is a substrate for MT1-MMP. It is therefore possible that MT1-MMP-mediated cleavage of ICAM-1 may co-ordinate the TM process. Src and ERK signaling, an intrinsic part of ICAM-1-mediated adhesion and transmigration (29, 35), are also implicated in MT1-MMP function (41, 42) through its intracellular domain. Because the ICAM-1 cytoplasmic tyrosine mutations result in defective ERK signaling (7) and reduced TM (Fig. 2B), it is possible that ERK signaling could modulate MT1-MMP function in TM. The implication of MT1-MMP in leukocyte trafficking opens avenues for utilizing MT1-MMP as a therapeutic target for intervention and blockade of leukocyte migration in inflammatory diseases such as arthritis and atherosclerosis. In this aspect, it is interesting to note that significant lesion regression was observed when TIMP-2, an endogenous inhibitor of MT1-MMP, was overexpressed in atherosclerosis-prone apoE^-/- mice (43).