Cell adhesion and migration properties of beta 2-integrin negative polymorphonuclear granulocytes on defined extracellular matrix molecules. Relevance for leukocyte extravasation.

Regulated adhesion of leukocytes to the extracellular matrix is essential for transmigration of blood vessels and subsequent migration into the stroma of inflamed tissues. Although beta(2)-integrins play an indisputable role in adhesion of polymorphonuclear granulocytes (PMN) to endothelium, we show here that beta(1)- and beta(3)-integrins but not beta(2)-integrin are essential for the adhesion to and migration on extracellular matrix molecules of the endothelial cell basement membrane and subjacent interstitial matrix. Mouse wild type and beta(2)-integrin null PMN and the progranulocytic cell line 32DC13 were employed in in vitro adhesion and migration assays using extracellular matrix molecules expressed at sites of extravasation in vivo, in particular the endothelial cell laminins 8 and 10. Wild type and beta(2)-integrin null PMN showed the same pattern of ECM binding, indicating that beta(2)-integrins do not mediate specific adhesion of PMN to the extracellular matrix molecules tested; binding was observed to the interstitial matrix molecules, fibronectin and vitronectin, via integrins alpha(5)beta(1) and alpha(v)beta(3), respectively; to laminin 10 via alpha(6)beta(1); but not to laminins 1, 2, and 8, collagen type I and IV, perlecan, or tenascin-C. PMN binding to laminins 1, 2, and 8 could not be induced despite surface expression of functionally active integrin alpha(6)beta(1), a major laminin receptor, demonstrating that expression of alpha(6)beta(1) alone is insufficient for ligand binding and suggesting the involvement of accessory factors. Nevertheless, laminins 1, 8, and 10 supported PMN migration, indicating that differential cellular signaling via laminins is independent of the extent of adhesion. The data demonstrate that adhesive and nonadhesive interactions with components of the endothelial cell basement membrane and subjacent interstitium play decisive roles in controlling PMN movement into sites of inflammation and illustrate that beta(2)-integrins are not essential for such interactions.

One of the main functions of endothelial cells is to prevent leukocyte emigration from the blood vessel into the underlying tissues, and only in cases of inflammation is this barrier function physiologically removed. During such extravasation processes emigrating cells not only have to rapidly traverse the tight endothelial cell monolayer but also the basement membrane of the blood vessel endothelium and migrate into the underlying interstitial extracellular matrix. Considerable effort has been made to understand the initial steps in this extravasation process, i.e. rolling along the blood vessel endothelium and firm adhesion, revealing the function of specific cell adhesion molecules of the selectin, immunoglobulin, and ␤ 2 -and ␣ 4 -integrin families (1). However, little is known of the subsequent interaction of leukocytes with the underlying extracellular matrix (ECM) 1 proteins of the blood vessel basement membrane and of the interstitium. A fact frequently ignored is that blood vessel basement membranes are biochemically unique and that growth factors and cytokines that affect other aspects of endothelial cell physiology also alter ECM expression (2)(3)(4)(5)(6)(7).
One of the major components of basement membranes is the laminin family of molecules, heterotrimers composed of ␣, ␤, and ␥ chains. To date 5␣, 4␤, and 3␥ laminin chains (2, 8 -10) have been identified that may combine to form at least 12 different isoforms that have tissue-specific and developmentally regulated expression patterns (2,3,(11)(12)(13)(14). The data suggest that laminins play significant roles in adhesion, migration, and differentiation of several cell types. Further, the occurrence of diseases involving laminin gene defects, such as epidermolysis bullosa and congenital muscular dystrophies and the generation of mice lacking particular laminin chains, suggests that this family of basement membrane molecules play crucial roles in vivo (reviewed in Ref. 15). Nevertheless, the information available on the functional significance of individual isoforms remains limited.
Endothelial cells express two laminin isoforms, depending on their tissue of origin and state of growth or activation (2,16). Laminin 8 (composed of laminin ␣ 4 , ␤ 1 , and ␥ 1 chains) is expressed by all endothelial cells regardless of their stage of development, and its expression is strongly up-regulated by cytokines and growth factors that play a role in inflammatory events, such as interleukin-1 or tumor necrosis factor-␣ (TNF-␣) (2). 2 Laminin 10 (composed of laminin ␣ 5 , ␤ 1 , and ␥ 1 chains) is detectable primarily in basement membranes of capillaries and venules commencing 3-4 weeks after birth (3,18). In contrast to laminin 8, endothelial cell expression of laminin 10 is up-regulated only by strong proinflammatory signals and, in addition, angiostatic agents such as progesterone and its derivatives (19). 2 Other extracellular matrix molecules are also differentially expressed by endothelium, varying with the endothelium type and/or activation state. Several endothelial cell-specific proteoglycans and extracellular molecules such as BM40 (SPARC), fibrinogen, thrombospondin, and fibronectin have been reported to be regulated by proinflammatory cytokines (20 -23). The data speak for a dynamic endothelial cell extracellular matrix that presents different molecular information depending on the type of endothelium and/or physiological situation.
Among the first cells present at a site of inflammation are polymorphonuclear granulocytes (PMN) and monocytes (24). Depending on the type of inflammation, it is thought that these cells provide the necessary signals for subsequent extravasation of lymphocytes into the site of inflammation. It cannot be disputed that the ␤ 2 -integrins are essential for adhesion of PMN, monocytes, and lymphocytes to endothelium and also for traversing the endothelial cell monolayer; however, the molecular interactions necessary for the subsequent transmigration of the basement membrane and entry into the subjacent interstitial extracellular matrix remain largely undefined (25,26). We have previously shown that mouse PMN, progranulocytic cells, and monocytes express functionally active ␤ 1 -and ␤ 3integrins on their surface that are employed for specific interactions with ECM molecules of basement membranes and the interstitium (27). Integrins of the ␤ 2 class, in contrast, were shown to mediate a general stickiness to many ECM molecules and even to bovine serum albumin and plastic (27)(28)(29). The aim of the present study is to define the role of cell-matrix interactions in PMN migration into extravascular tissues by using ␤ 2 -integrin null PMN. Comparisons have been made with mouse wild type PMN and the progranulocytic cell line 32DC13. In contrast to previous studies, the extracellular matrix molecules employed here in in vitro adhesion and migration assays reflect those expressed at sites of extravasation in vivo, with emphasis on the recently identified endothelial cell laminin isoforms, laminins 8 and 10.
Flow Cytometry Analysis-Flow cytometry analysis was carried out as previously described (27).
Isolation and Characterization of Laminins 8 and 10 -Mouse laminin 8 (composed of ␣ 4 , ␤ 1 , and ␥ 1 chains) was isolated from the conditioned media of 3T3 fibroblasts and the MC3T3-G2/PA6 preadipocyte cell line using a combination of an ion exchange chromatography (POROS 20 HQ column) and immunoaffinity chromatography (CNBr-Sepharose) with a rat mAb to laminin ␥ 1 (3E10). 2 Laminin 10 was isolated from human placenta by affinity chromatography using mouse anti-human ␥ 1 (D18) and, subsequently, mouse anti-human laminin ␣ 5 (4C7) antibodies. 2 Mouse anti-human laminin ␤ 2 (C4) antibody was employed to immunoabsorb laminin ␤ 2 -containing complexes where necessary. The purity of the laminin 8 and 10 preparations was assessed by immunoblotting and enzyme-linked immunosorbent assay using laminin chain-specific antibodies as described previously (11). For comparison, a commercial source of human laminin 10/11 (Life Technologies, Inc.) was also employed in adhesion and migration assays.
Attachment Assays-In vitro cell attachment assays were performed as previously described (27) using colorimetric analysis of lysosomal hexosaminidase (A 405 ) to quantitate the number of adherent cells. Because leukocytes are activated by chemokines and cytokines are released at sites of inflammation in vivo, attachment assays were performed with nonactivated and activated cells. Activation was achieved either with a physiological activator, fMLP (10 ng/ml), or with the general leukocyte activator, PMA (100 ng/ml), that bypasses ligandreceptor interactions and simulates a strong activation signal (27). The effect of divalent cations was tested by performing attachment assays in the presence of 1 mM EDTA. Inhibition studies assessed the effects of linear and cyclic RGD peptides or antibodies against integrin subunits and extracellular matrix proteins on cell-matrix interactions and involved a 30-min preincubation of cells prior to addition to protein-coated microtiter plates with varying concentrations of antibodies (10 -300 g/ml) or peptides (10 -100 M for cyclic and linear RGD peptides). Experiments were carried out using a single concentration of extracellular matrix protein (30 g/ml) at which cell binding was saturated, in the presence of varying concentrations of inhibitory antibodies or peptides. The experimental procedure was otherwise as described previously (27). The percentage of cells that bound specifically to the coating substrate was determined as follows: ((A 405 of total bound cells Ϫ A 405 of BSA-bound cells)/A 405 of 50000 applied) ϫ 100 ϭ % Specific Binding.
Migration Assays-Transmigration assays were performed using Costar Transwells (polycarbonate filter, 5-m pore size). Membranes were coated overnight at 4°C with 10 -15 g/ml of laminins 1, 8, or 10 prepared in our laboratory, commercial laminin 10/11, fibronectin, or vitronectin and subsequently blocked with 1 mg/ml BSA for 1 h at 37°C. Only ␤ 2 -integrin null PMN were used in these experiments (for reasons outlined under "Results"). 1 ϫ 10 5 cells in attachment buffer (Dulbecco's modified Eagle's medium, 0.5% BSA, pH 7.5) were added to the upper chamber, and 10 Ϫ7 M fMLP, a physiological chemotactic factor for PMN, was added simultaneously to the bottom chamber. Control experiments were performed in the absence of fMLP. Cells were incubated at 37°C for 2 h, subsequently collected, and then counted microscopically.
Statistical Analyses-Cell adhesion and migration assays were carried out at least six times with triplicates performed in each experiment. Student's t test was used to test statistical significance of differences in Bmax (maximal adhesion) values measured with activated and nonactivated PMN; differences in mean rates of migration across all substrates was compared pairwise using the Student Newman Keuls method (i.e comparisons were made between fMLP-induced transmi-gration across laminin 10 and 8, laminin 10 and 1, laminin10 and fibronectin, laminin 10 and vitronectin, and laminin 1 and 8.

Laminin 8 and Laminin 10 Purification
Using ion exchange and affinity chromatography employing laminin chain-specific mAbs produced in our laboratory, intact mouse laminin 8 and human laminin 10 were purified from conditioned media of mouse 3T3 and MC3T3-G2/PA6 cells, and human placenta, respectively. Laminin 8 was prepared using a rat mAb specific for mouse laminin ␥ 1 (3E10) in affinity chromatography (Fig. 1, lane 3), whereas human laminin 10 was isolated using mouse mAbs to the human laminin ␥ 1 (D18) and ␣ 5 (4C7) chains (Ref. 34 and Fig. 1, lane 7). Immunoblot analysis of the isolated laminin isoforms revealed the presence of only laminin ␣ 4 , ␤ 1 , and ␥ 1 chains in the laminin 8 preparation, whereas the human laminin 10 preparation contained predominantly laminin ␣ 5 , ␤ 1 , and ␥ 1 chains (laminin 10) and trace amounts of laminin ␤2 chain (laminin 11) as detected by immunoblots and enzyme-linked immunosorbent assay (data not shown). For comparison, a commercially available human laminin 10/11 (Life Technologies, Inc.) preparation isolated from pepsin-digested placenta by affinity chromatography using the 4C7 mAb and widely used in in vitro assays is also shown (Fig.  1, lane 5). In contrast to laminin 10 prepared in our laboratory, the intact 400-kDa laminin ␣ 5 chain (3) was not detectable in the commercial preparation of laminin 10/11, and several proteolytic fragments were present (Fig. 1, lanes 5 and 6). Immunoblots of commercial laminin 10/11 revealed the laminin ␣ 5 chain as a major 200 -220-kDa band plus minor bands of 125 and 97 kDa (Fig. 1, lane 6). This commercial preparation of laminin 10/11 was found to be significantly more adhesive than that prepared in our laboratory. Most experiments were therefore carried out with the laminin 10 preparation from our laboratory, which better reflects the laminin isoform found in blood vessel basement membranes.

Integrin Expression on ␤ 2 -Integrin Null PMN
Flow cytometric analysis revealed no difference in the surface expression of ␤ 1 -and ␤ 3 -integrins between wild type and ␤ 2 -integrin null PMN (Table I) and confirmed the absence of the ␤ 2 -integrin from the surface of the latter cells (Fig. 2). Moderate levels of integrins ␣ 6 ␤ 1 , ␣ 5 ␤ 1 , and ␣ 4 ␤ 1 occurred on the cell surface of both wild type and ␤ 2 -integrin null PMN, which did not change upon activation with PMA (Fig. 2); ␣ 2 ␤ 1 or ␣ 3 ␤ 1 were confirmed undetectable on either cell type either before (45) or after activation (Fig. 2) despite the fact that the antibodies employed are functional in flow cytometric analysis on other cell types (37,46). Similar results are obtained with 32DC13 cells, as previously described (27). Integrin ␤ 3 was expressed on wild type (Table I), ␤ 2 -integrin null PMN ( Fig. 2 and Table I), and 32DC13 cells (Table I). Activation with PMA markedly increased integrin ␤ 2 expression in wild type but not in ␤ 2 -integrin null PMN ( Fig. 2 and Table I). Treatment with PMA strongly reduced L-selectin expression on ␤ 2 -integrin null PMN ( Fig. 2 and Table I), wild type PMN, and 32DC13 (27), indicating effective activation of the cells. Treatment of cells with fMLP had no effect on the surface expression of integrins (Table I) (27).

Specific Cell Adhesion to ECM Molecules
We have previously reported that ␤ 2 -integrins confer a general stickiness to PMN resulting in strong adhesion to many substrates, including BSA and plastic (27), and also extracellular matrix molecules, such as collagen types I and IV (47,48). This strong adhesion makes it impossible to identify non-␤ 2integrin-mediated adhesive processes. Ablation of the ␤ 2 -integrin significantly reduced this "background" binding from 50 - 60% in activated wild type PMN to below 10% in the ␤ 2integrin null PMN (Fig. 3), confirming the promiscuous nature of ␤ 2 -integrin on PMN. The absence of this background binding in the case of ␤ 2 -integrin null PMN permitted clear identification of adhesive and nonadhesive substrates.
To best assess the role of PMN-matrix interactions in the extravasation process, the following ECM molecules were employed in all in vitro adhesion assays: (i) typical endothelial cell basement membrane components: laminins 8 and 10, collagen type IV, and perlecan, (ii) basement membrane components that do not occur in endothelial cell basement membranes: laminins 1 and 2, and (iii) interstitial matrix molecules that are up-regulated in the interstitial matrix surrounding blood vessels during inflammation: collagen type I, tenascin-C, fibronectin, and vitronectin.
In the nonactivated or the fMLP-treated state, wild type and ␤ 2 -integrin null PMN showed the same qualitative pattern of specific binding to the interstitial matrix molecules, fibronectin and vitronectin, and the basement membrane molecule laminin 10 (Table II). No specific binding to laminins 1, 2, and 8, collagen types I and IV, tenascin-C, or perlecan was measured (Table II). Coatings of the basement membrane heparan sul- Control is the binding of fluorescently labeled second layer antibody plus appropriate isotype control antibody. The vertical axis shows the cell number, and the horizontal axis shows the log fluorescence intensity. Integrin receptors investigated included ␤ 1 , ␣ 3 , ␣ 5 , ␣ 6 , ␤ 3 , ␤ 2 , ␣ 2 , and ␣ 4 (latter two are not shown). L-selectin expression was used as a measure of PMN activation. fate proteoglycan, perlecan, or of the basement membrane collagen type IV repelled cells from coated surfaces, as previously reported for U937 and human erythroleukemic cells lines (49) and for wild type PMN (27).
Concentration-dependent binding of nonactivated and PMAactivated ␤ 2 -integrin null PMN to fibronectin (Fig. 4A), vitronectin (Fig. 4B) and laminin 10 ( Fig. 5A) is shown and revealed statistically significant increased maximal binding values (B max ) after PMA stimulation (p Ͻ 0.005 for all three substrates, Student's t test). However, no significant difference was observed in the approximate binding affinities (1/B max ) for these three substrates before and after PMA activation (data not shown). Wild type PMN binding to fibronectin, vitronectin, and laminin 10 was slightly reduced after PMA activation (Table  II), as previously reported for fibronectin and vitronectin (27).
The progranulocytic cell line 32DC13 showed significant binding only to fibronectin (35%) and laminin 10 (32%) in the nonactivated or fMLP-treated state (Table II). Upon PMA activation the extent of binding to fibronectin and laminin 10 increased significantly, and an additional binding to vitronectin was induced (Table II). In contrast to fully mature PMN, PMA activation of 32DC13 also induced low level and low affinity binding to laminins 1 (18%), 2 (10%), and 8 (24%) ( Table II). Binding to all three laminin isoforms was statistically significantly above background binding to BSA. No binding of 32DC13 to collagen types I and IV, perlecan, or tenascin-C was detectable before or after activation (Table II).

Integrin Receptors Involved in ECM Binding
To determine the cellular receptors mediating adhesion to the extracellular substrates, inhibition studies were carried out using nonactivated, fMLP-treated, and PMA-activated wild type or ␤ 2 -integrin null PMN and 32DC13.
Fibronectin-Cell binding to fibronectin can be mediated by ␣ 5 ␤ 1 , by ␣ 4 ␤ 1 , or by the ␣ v series integrins (50). ␣ 4 ␤ 1 binding to fibronectin is not RGD-dependent, whereas both ␣ 5 ␤ 1 and the ␣ v series integrins bind to the RGD sequence in fibronectin. However, ␣ v -mediated interactions with fibronectin are very sensitive to inhibition by linear RGD peptides (Ͻ10 M), and ␣ 5 ␤ 1 -dependent binding is inhibited only at high concentrations of linear RGD (Ͼ100 M) (27,43). Because functionblocking antibodies exist only for mouse ␣ 4 -and ␤ 1 -integrins but not ␣ 5 -and ␣ v -integrins, we exploited the differential sensitivity to linear RGD peptides to characterize the interaction of the mouse cells with fibronectin. Specific binding of wild type and ␤ 2 -integrin null PMN and 32DC13 to fibronectin was measured in the presence of anti-mouse ␣ 4 mAb (R1-2), anti-␤ 1 polyclonal antibody (Ha2/5), a chick antibody against the RGDcell binding site in fibronectin that is specifically recognized by the ␣ 5 ␤ 1 integrin, linear RGD, and cyclic RGD peptides (EMD66203 and EMD69601).
Binding of wild type and ␤ 2 -integrin null PMN and 32DC13 to fibronectin was significantly inhibited only by high concentrations of linear RGD (Ͼ100 M) (27), the anti-␤ 1 -integrin antibody (Ha2/5), and the antibody against the RGD-fibronectin cell binding site. The chicken antibody inhibited specific binding to fibronectin only at concentrations of 300 g/ml; however, similar concentrations of purified preimmune chick IgG did not inhibit binding to fibronectin. Cyclic RGD peptides (EMD6603/EMD69601) at concentrations up to 100 M and the anti-integrin ␣ 4 antibody (R1-2) (10-50 mg/ml) had no effect on binding, suggesting ␣ 5 ␤ 1 -mediated adhesion to fibronectin. An example of the results is shown for ␤ 2 -integrin null PMN in Fig. 4C.
Vitronectin-Binding to vitronectin is mediated by the ␣ v series integrins that interact with high affinity to the RGD sequence in this molecule. Cyclic RGD peptides have been described that specifically inhibit murine ␣ v Ϫintegrin interactions with vitronectin (27,43,51). Further, ␣ v -mediated interactions are very sensitive to inhibition by linear RGD peptides that abolish cell binding at concentrations of Ͻ50 g/ml (27,43,51). The effects of 10 -100 M cyclic RGDfV peptide (EMD66203), control cyclic R␤ADfV peptide (EMD69601), linear RGD, and antibodies against ␤ 1 -(Ha2/5) and ␤ 3 -integrins (2C9G2) were therefore tested for their ability to inhibit cell adhesion to vitronectin. The cyclic RGDfV peptide (EMD66203) and linear RGD significantly inhibited binding of all three cell types to vitronectin at concentrations of 10 M. The absence of effect of the ␤ 1 -integrin antibody (up to 100 g/ml) and the inhibition of binding by the 10 g/ml ␤ 3 -integrin antibody suggested ␣ v ␤ 3 -mediated adhesion to vitronectin. An example of the data for ␤ 2 -integrin null PMN is shown in Fig. 4D.
Integrin ␣ 6 ␤ 1 is considered to be a specific receptor for laminin 1 (8). However, despite the constitutive expression of integrin ␣ 6 ␤ 1 on the cell surface of wild type and ␤ 2 -integrin null PMN, binding to laminins 1, 2, or 8 could not be induced either by cellular activation by PMA or maximal activation of integrins with 10 mM Mn 2ϩ . An example of these data is shown for ␤ 2 -integrin null PMN on laminin 1 in Fig. 5C. The absence of cell binding was not related to the ability of the different laminins to bind plastic or due to degradation of the laminins, because 32DC13 bound to laminins 1 and 8 in an ␣ 6 ␤ 1 -dependent manner (data not shown). This was investigated further by control experiments performed with the human fibrosarcoma cell line, HT1080, which showed extensive binding (90% specific adhesion) to laminin 1 via integrin ␣ 6 ␤ 1 (Fig. 5D) (57, 58) and non-integrin ␣ 6 ␤ 1 -dependent binding to laminin 10 (Fig. a Statistically significant differences in B max values measured for nonactivated and PMA-activated cells (Student's t test, p Ͻ 0.05). b Cells repelled from coated surfaces as revealed by video microscopy.

FIG. 4. Specific cell adhesion of nonactivated (ࡗ) and PMA-activated (Ⅺ) ␤ 2 -integrin null PMN to increasing molar concentrations of fibronectin (A) and vitronectin (B) and of PMA-activated ␤ 2 -integrin null PMN to 30 nM fibronectin (C) or 30 nM vitronectin (D) in the presence or absence (Control) of added inhibitors.
Inhibitors employed included linear RGD peptide, 10 M control cyclic RbADfV (69601), or ␣ v -specific cyclic RGDfV (66203) peptide, anti-␤3 integrin (2C9G2), a chicken antibody against the RGD-fibronectin cell binding site that is recognized by ␣ 5 integrin and associated control chicken IgG, hamster anti-rat ␤ 1 -integrin (Ha2/5) and associated control hamster IgG, anti-␣ 4 -integrin (R1-2), or 10 mM EDTA. Data shown were the same for wild type PMN, and the same pattern of results as in C and D was found for nonactivated ␤ 2 -integrin null PMN. Data shown are for one representative experiment where the mean specific adhesion is calculated from at least triplicate values per concentration Ϯ S.D. 5D). This suggests that other factors or accessory molecules on the surfaces of cells are involved in regulating ␣ 6 ␤ 1 -dependent recognition of different laminin isoforms.

Migration Assays
Transmigration across laminin-, fibronectin-, or vitronectincoated filters in response to a chemotactic gradient of fMLP permitted quantification of rates of migration. Only ␤ 2 -integrin null PMN could be used in these experiments, because the high background binding of wild type PMN masked differences between substrates, and 32DC13 could not be induced to migrate. Significantly more transmigration was measured in the presence of fMLP on all substrates, except commercial laminin 10/11 and BSA (Fig. 6), both of which supported only minimal rates of migration. In the presence of a fMLP gradient, ␤ 2integrin null PMN migrated most efficiently and at similar rates across laminin 10-(ϳ200 cells/min), fibronectin-(220 cell/min), and vitronectin-coated filters (219 cells/min). Student Newman Keuls post hoc test revealed no statistically significant difference in the rates of migration across these three substrates, whereas significantly lower rates of migration (p Ͻ 0.005) were measured across laminin 8-(65 cells/min) and laminin 1-coated (40 cells/min) filters (Fig. 6). Rates of PMN transmigration across laminin 1-or 8-coated filters were not significantly different. PMN transmigration of laminin 10-and 8-coated filters was significantly reduced by anti-integrin ␤ 1 (Ha2/5) and ␣ 6 (GoH3) at concentrations of 10 g/ml (Fig. 6), implicating the ␣ 6 ␤ 1 integrin in the transmigration process. However, on laminin 1 or anti-integrin ␤ 1 or ␣ 6 antibodies had no statistically significant effect (Fig. 6). DISCUSSION PMN interaction with extracellular matrix molecules of the endothelial cell basement membrane and the underlying interstitium is essential for their migration into inflamed tissues. Although attempts have been made to understand this process using in vitro assays similar to those employed in the present study, in the past little consideration has been given to whether the extracellular matrix molecules employed in such studies occur at sites of inflammation, either in endothelial cell basement membranes or in the interstitial matrix (3,16,59,60).
A further controversy in the published data is the relative contribution of ␤ 1 -integrins to PMN extravasation processes as compared with ␤ 2 -integrins. Although the expression of functionally active ␤ 1 -integrin on the PMN surface is gaining acceptance (27, 28, 61), it still remains difficult to investigate ␤ 1 -integrin function on the PMN without the complication of ␤ 2 -integrin-mediated background adhesion. In the present study we have attempted to more precisely define PMN-ECM interactions in the extravasation process (1) by eliminating ␤ 2 -integrin-mediated interactions through the use of PMN isolated from ␤ 2 -integrin null mice (26), and (2) by employing basement membrane and interstitial matrix molecules known to occur around blood vessels at sites of inflammation in in vitro adhesion and migration assays. Our choice of extracellular matrix molecules for in vitro studies was based on occurrence in endothelial cell basement membranes (laminins 8 and 10, collagen type IV, and perlecan) and in the interstitium of inflamed tissues (tenascin-C, fibronectin, and vitronectin), as compared with ECM molecules that do not occur in association with blood vessels (laminins 1 and 2).
Laminins are difficult to isolate in pure forms because of their large size and tendency to self-aggregate. As a consequence, commercial preparations of laminins are frequently prepared using proteolytic digestion and subsequent affinity chromatography, resulting in truncations of laminin chains and digestion products. Our analysis of such a commercial laminin 10/11 preparation clearly demonstrated the presence of severely truncated laminin ␣ 5 chain and several digestion products. In view of the fact that the laminin ␣ chains carry the cell-binding domains (62), such commercial preparations should be used with caution. Proteolytic digestion may result in the exposure of epitopes that are masked in vivo and that may have different properties to the intact native molecule. Using chain-specific mAbs, we have purified biologically active laminins 8 and 10, the only two laminin isoforms to date known to occur in blood vessel basement membranes. The laminin 10 preparation, isolated without proteolytic digestion and by affinity chromatography using anti-laminin ␥ 1 and ␣ 5 chain mAbs, was less adhesive than commercial laminin 10/11 (data not shown) and supported high rates of PMN transmigration. In our hands, the commercial preparation of laminin 10/11 did not support PMN transmigration; a similar result has been recently reported for human monocytes (63) and is discussed in detail below. These differences may be due to the absence of a full-length laminin ␣ 5 chain in the commercial laminin 10/11 and/or the exposure of additional binding sites caused by proteolytic digestion. Alternatively, these functional differences may be related to the fact that the commercial laminin is a mixture of laminin 10 (␣ 5 ␤ 1 ␥ 1 ) and 11 (␣ 5 ␤ 2 ␥ 1 ), whereas the laminin 10 preparation prepared in the present study contained principally laminin 10 and only trace amounts of laminin 11. Studies on the functional differences between laminin 10 and 11 are currently in progress.
Interestingly, the pattern of specific ECM binding of wild type and ␤ 2 -integrin null PMN did not differ significantly, indicating that the ␤ 2 -integrins do not mediate specific binding to any of the extracellular matrix molecules tested here. Transmigration experiments also clearly showed that ␤ 2 -integrin is not essential for migration of PMN across extracellular matrix substrates. Our data show that components of the nonendothelial cell basement membranes, such as laminin 1 and 2, do not support adhesion of mature PMN regardless of the cellular activation state. However, components of endothelial cell basement membranes and of the interstitial matrix in inflamed tissues provide both adhesive and nonadhesive substrates that probably control how and when PMN migrate into inflamed tissues and whether they become arrested at such sites. The data also demonstrate that despite the absence of strong adhesion, integrins can still signal specific information to cells. Even though PMN bound strongly to laminin 10, fibronectin, and vitronectin and not at all to laminins 1 and 8, all substrates supported fMLP-induced transmigration, albeit at significant reduced rates in the case of the latter two substrates. Monocytes have recently been shown to transmigrate across laminin 8-coated but not laminin 10-coated filters (63). The difference from our data is very likely to be related to the source of laminin 10/11 employed because Pedraza et al. (63) used the commercial laminin 10/11, which we also found to be a poor substrate in transmigration studies. It cannot be excluded, however, that the migratory behavior of monocytes differs from that of PMN, as implied by the fact that monocytes show significant rates of migration across laminin-coated filters even in the absence of a chemotactic gradient (63), whereas PMN do not. It is interesting to note that the rate of monocyte transmigration across recombinant laminin 8, composed of human laminin ␣ 4 and ␥ 1 and mouse laminin ␤ 1 chains (53), employed in the monocyte study was comparable with that measured in the present study using native mouse laminin 8 and PMN. Whether these two laminin 8 preparations support cell adhesion equally is difficult to assess because Pedraza et al. (63) used toluidine blue staining of adherent cells as a measure of relative adhesion rather than the quantitative cell adhesion assays used here.
The laminin cell adhesion and migration data presented here demonstrate that different laminin isoforms probably all bind and signal through ␣ 6 ␤ 1 , and it may be that the number of laminin molecules bound, the differential strength of adhesion, and/or the involvement of accessory molecules determines the cellular response. A similar observation has been made with mouse melanoma cells that migrate on laminin 1 using ␣ 6 ␤ 1 without strong adhesion (64) and Jurkat T-cells, which do not FIG. 6. Transmigration data for ␤ 2integrin null PMN. Number of cells transmigrated/min Ϯ S.E. across Transwell filters (5-m pore size) coated with 15 g/ml laminins 10, 8, and 1 prepared in our laboratory, commercial laminin 10/11 (LM 10/11), fibronectin (FN), vitronectin (VN), or BSA in the presence (open bars) or absence (gray bars) of fMLP in the lower chamber. 1 ϫ 10 5 cells were added per filter and allowed to transmigrate for 2 h at 37°C; the total number of cells was then counted in the lower chamber. Control is transmigration in the absence of added inhibitors; anti-␣ 6 is 5 g/ml anti-integrin ␣ 6 (GoH3); anti-␤ 1 is 10 g/ml hamster anti-integrin ␤ 1 (Ha2/ 5); and IgG is 10 g/ml control hamster. All values represent the means of at least six separate experiments Ϯ S.E. bind to laminin 1 but respond to interactions with this molecule with distinct intracellular Ca 2ϩ signals (65). Whether differential intracellular signals are induced in PMN by adhesive and nonadhesive substrates is currently under investigation. The fact that ␤ 2 -integrin null PMN can bind to and migrate on laminin substrates using ␤ 1 -integrins also demonstrates that ␤ 2 -integrin is not essential for this function. The absence of PMN in inflamed tissues in the ␤ 2 -integrin null mouse (26) and in LAD-1 patients (66) is, therefore, exclusively due to the inability of these cells to penetrate the endothelial monolayer and not the underlying basement membrane.
Integrin ␣ 6 ␤ 1 has been described as a specific receptor for laminin 1 (reviewed in Ref. 8) and laminin 8 (52,53,63) on several cell types, including HT1080 and the progranulocytic cell line 32DC13 as shown here. Nevertheless, despite ubiquitous expression of significant levels of this receptor on the PMN surface, binding to laminin 1 or 8 could not be induced even under maximal cell (PMA) or integrin (Mn 2ϩ ) activation conditions. This suggests that the ability to distinguish between laminin isoforms is dependent upon other factors/accessory molecules that affect ␣ 6 ␤ 1 binding activity. Indeed, ␣ 6 ␤ 1 (and ␣ 3 ␤ 1 ) has been shown to be associated with the tetraspan family of molecules, which have been implicated in cell adhesion and migration (reviewed in Refs. 67 and 68) and which occur on PMN cell surfaces (69,70). Whether such integrinassociated molecules are involved in the ␣ 6 ␤ 1 -laminin interactions in the case of PMN is currently under investigation.
Our data confirm the promiscuous nature of ␤ 2 -integrins, which, when activated, mediate nonspecific binding to several substrates. Such high affinity, broad specificity adhesion may be advantageous for the adhesion to and migration across the endothelial monolayer but not for the regulated adhesion to ECM molecules necessary for migration across the basement membrane and into the interstitium. The data presented here suggest that the differential extent and strength of PMN adhesion to different ECM molecules provides the balance of adhesive and nonadhesive forces necessary for penetration of basement membranes at sites of inflammation and that such interactions are mediated by the ␤ 1 -and ␤ 3 -integrins but not ␤ 2 -integrins. It has been shown that ␤ 1 -integrins are up-regulated on PMN after transmigration of endothelial cell monolayers both in vitro and in vivo (17,28). It is therefore possible that PMN surface integrin expression and/or activity is altered during transmigration of the endothelial cell monolayer, such that ␤ 1 -and ␤ 3 -integrin activity predominates in the subsequent ECM transmigration steps.