α_vβ₃-integrin-dependent activation of focal adhesion kinase mediates NF-κB activation and motogenic activity by HIV-1 Tat in endothelial cells

The Tat protein is a cationic 86-101 amino acid polypeptide that acts as the main transactivating factor of HIV-1 (Gatignol and Jeang, 2000). Direct and indirect evidence indicate that Tat is actively released by HIV-1-infected cells (Noonan and Albini, 2000). Extracellular Tat targets different types of uninfected cells, including endothelial cells (ECs) (Rusnati and Presta, 2002), causing a variety of biological effects possibly related to distinct AIDS-associated pathologies, including neuropathies (Dewhurst et al., 1996), immune suppression (Caputo et al., 1999; Noonan and Albini, 2000) and increased tumorigenesis in AIDS patients (Caputo et al., 1999). The mechanisms by which Tat exerts its pathological effects are manifold and mediated by various signalling receptors in target cells, including the vascular endothelial growth factor (VEGF) receptor 2 (KDR) (Albini et al., 1996), chemochine receptors (Albini et al., 1998), and integrins (Barillari et al., 1999a; Barillari et al., 1999b; Fiorelli et al., 1999; Mitola et al., 2000).

Integrins are a family of heterodimeric receptors that, unlike growth factor receptors, lack intrinsic tyrosine kinase activity. Yet, an early event during integrin signalling is the tyrosine phosphorylation of the non-receptor tyrosine kinase focal adhesion kinase (FAK) (Kumar, 1998; Schlaepfer et al., 1999) that, in turn, leads to the activation of the GTPase RhoA and/or pp60^src in different cell types (Palazzo et al., 2004; Sharma-Walia et al., 2004; Zhai et al., 2003). In ECs, this signal transduction pathway can be activated upon α_vβ₃ integrin engagement and leads to nuclear translocation of NF-κB and cell survival (Scatena et al., 1998). Accordingly, integrins regulate EC proliferation in vitro (Eliceiri, 2001) and angiogenesis in vivo (Varner and Cheresh, 1996).

In ECs, Tat interacts with α_vβ₃ and α₅β₁ integrins (Barillari et al., 1999a; Barillari et al., 1999b; Fiorelli et al., 1999; Mitola et al., 2000). This interaction is required by Tat to induce neovascularization in vivo and chemotactic and mitogenic activity in cultured ECs (Barillari et al., 1999a; Barillari et al., 1999b; Mitola et al., 2000). Also, integrins mediate the adhesion and spreading of ECs to substratum-immobilized Tat (Barillari et al., 1999a; Barillari et al., 1999b; Fiorelli et al., 1999; Mitola et al., 2000). However, the molecular bases and biological relevance of this process remain poorly elucidated. Relevant to this point, Tat accumulates in the extracellular matrix (ECM) as an immobilized protein (Chang et al., 1997) by interacting with heparan sulphate proteoglycans (HSPGs) (Chang et al., 1997; Tyagi et al., 2001). These findings suggest that the concentration of Tat can increase in the microenvironment, possibly representing a localized, persistent stimulus for EC adhesion and activation.

In this paper we demonstrate that HIV-Tat/α_vβ₃ interaction leads to activation of FAK, RhoA, NF-κB and pp60^src, which are required for the stimulation of motogenic activity in ECs.

Cyclo(-Arg-Gly-Asp-D-Phe-Val) (cRGDfV) and cyclo(-Arg-Ala-Asp-D-Phe-Val) (cRADfV) peptides were from Bachem (Beidendorf, Switzerland). Integrin α_vβ₃ was purified to homogeneity as described previously (Rusnati et al., 1997). Fibronectin (FN) was from Sigma (St Louis, MO, USA). Anti-α_vβ₃ monoclonal LM 609 antibody was from Chemicon International (Temecula, CA, USA). Anti-β3 antiserum was from G. Tarone (University of Turin, Italy). Anti-FAK antibodies A17 and C20, anti-phosphotyrosine antibody PY99, anti-Src polyclonal antibody SRC-2 and anti-RhoA antibody 26C4 were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The anti-FAK phosphospecific antibody sampler kit was from BioSource International (Camarillo, CA, USA). Anti-haemagglutinin (HA) antibody was from Upstate Biotechnology (Lake Placid, NY, USA). The cell-permeable NF-κB inhibitory peptide SN50 and its inactive analogue SN50M were from Biomol (Plymouth Meeting, PA, USA). The pp60^src inhibitor PP₂ (4-amino-5-(4-chlorophenyl)-7-(t-butyl) pyrazolo [3,4-d] pyrimidine), its inactive analogue PP₃, the anti-Src phospho-specific Tyr₄₁₆ antibody and the RhoA inhibitor exoenzyme C3 from Clostridium botulinum were from Calbiochem (La Jolla, CA, USA).

The 86 amino acid wild-type form of HIV-1 Tat and its mutants were expressed and purified from Escherichia coli as glutathione S-transferase (GST) fusion proteins as described previously (Rusnati et al., 1998).

Human α_vβ₃ integrin purification, cell-free α_vβ₃ integrin/GST-Tat interaction and cell adhesion assay with GM7373 ECs were performed as previously described (Rusnati et al., 1997).

A BIAcore X apparatus (BIAcore Inc, Piscataway, NJ, USA) was used. Surface plasmon resonance (SPR) was exploited to measure changes in refractive index caused by the ability of purified α_vβ₃ integrin to bind GST-Tat immobilized to a BIAcore sensorchip. For this, 50 μg/ml of GST-Tat were allowed to react with a flow cell of a CM5 sensorchip that was previously activated with 50 μl of a mixture of 0.2 M N-ethayl-N′-(3-diethylaminopropyl)carbodiimide hydrochloride and 0.05 M N-hydroxysuccinimide. These experimental conditions allowed the immobilization of 10281 resonance units (RU), corresponding to approximately 0.3 pmoles of GST-Tat. Similar results were obtained for the immobilization of BSA, here used as a negative control and for blank subtraction. Increasing concentrations of integrin α_vβ₃ in 10 mM Hepes, 150 mM NaCl, 3.4 mM EDTA, 0.005% surfactant P20, pH 7.4 were then injected over the BSA or GST-Tat surfaces for 4 minutes (to allow their association with immobilized proteins) and then washed until dissociation was observed. The SPR signal was expressed in terms of RU.

Transformed foetal bovine aortic endothelial GM7373 cells were obtained from the NIGMS Human Genetic Mutant Cell Repository (Institute for Medical Research, Camden, NJ, USA) (Grinspan et al., 1983). GM7373 cells were grown in Eagle's minimal essential medium containing 10% foetal calf serum (FCS) (Gibco, Grand Island, NY, USA), vitamins, essential and non essential amino acids.

BALB/c mouse aortic endothelial 22106 cells (MAECs) were from R. Auerbach (University of Wisconsin, Madison, WI, USA) and were grown in Dulbecco's minimal essential medium supplemented with 10% calf serum (Gibco).

Human umbilical vein ECs (HUVECs) were from Biowhittaker (Walkersville, MA, USA) and cultured in EGM-2 medium (Biowhittaker).

GM7373 cells were transfected with the expression vectors pCDNA 3.1 FRNK or pCDNA 3.1 FRNK-Ser1034 encoding for the FAK C-terminal domain (FRNK) or the inactive FRNK_L1034S mutant, respectively (Schlaepfer and Hunter, 1996), both tagged with a triple haemagglutinin (HA) (Sieg et al., 1999). To obtain stable transfectants, GM7373 cells were plated at 7×10⁵ cells/100 mm plates and were transfected with 8 μg of plasmid DNA using lipofectamin (Gibco) according to manufacturer's instructions. After 72 hours, 200 μg/ml of hygromycin-B (Gibco) were added to cell cultures. Hygromycin-B-resistant clones were isolated and tested for HA-tag expression by western blotting.

Tissue culture dishes (10 cm in diameter) containing subconfluent cultures of GM7373 or MAE cells were incubated at 37°C in serum-free medium for 24 hours. Then, cells were added with GST-Tat in absence or in presence of the different inhibitors and further incubated at 37°C for different periods of time. Alternatively, GM7373 cells maintained in serum-free medium for 24 hours were detached from culture dishes by 15 minutes incubation in PBS containing 15 mM EDTA and resuspended in medium containing 1% FCS and 25 mM Hepes, pH 7.5. Aliquots of 500,000 cells were then maintained in suspension or allowed to adhere to non-tissue culture dishes (60 mm in diameter) coated with GST-Tat or poly-L-lysine (PL) for 30 minutes at 37°C.

At the end of incubation, cells were lysed in RIPA modified lysis buffer [50 mM Tris-HCl pH 7.4 containing 150 mM NaCl, 1% Nonidet P-40, 0.25% sodium deoxycolate and a protease inhibitor mixture (1 mM phenylmethylsulphonylfluoride, 4 mM amino-n-caproic acid, 10 μg/ml leupeptin, 1 mM Na₃VO₄, 50 mM NaF)] and centrifuged (10 minutes at 15,000 g). Protein concentration was evaluated in the supernatants and 600 μg protein/sample were incubated for 3 hours at room temperature with anti-FAK rabbit polyclonal antibody A17 (1.5 μg/sample) and Ultra-link Immobilon Protein A (20 μl/sample) (Pierce, Rockford, IL, USA). At the end of incubation, samples were centrifuged (1 minute at 100 g) and pellets were washed extensively with RIPA modified lysis buffer, resuspended in reducing SDS-PAGE sample buffer and incubated for 5 minutes at 90°C. In some experiments, cells cultures were directly lysed in reducing SDS-PAGE sample buffer and incubated for 5 minutes at 90°C. Immunoprecipitates and total cell extracts were analysed on SDS-7% PAGE under reducing conditions followed by western blotting using anti-phosphotyrosine monoclonal antibody PY99 or the FAK phosphorylation site-specific antibodies, respectively. Parallel experiments with antibodies directed against the unphosphorylated form of FAK were performed for each experiment.

The extent of FAK phosphorylation was quantified by using the Image Pro-Plus analysis system (Media Cybernetics, Silver Spring, MD, USA). Briefly, the autoradiographs for total or phosphorylated FAK were digitized on a high resolution monitor and stored within the Pro-Plus analysis system's memory. The integrated densities of the bands were then evaluated and the values of those corresponding to phosphorylated FAK were normalized to total FAK protein levels.

GM7373 cells were allowed to adhere to the different substrata or were maintained in suspension exactly as described above. At the end of incubations, cells were lysed and immunoprecipitated with anti-Src antibody SRC-2 (1.5 μg/sample) as described above, resuspended in reducing SDS-PAGE sample buffer, incubated for 5 minutes at 90°C and analysed on SDS-12% PAGE under reducing conditions followed by western blotting using antibodies directed against the phosphorylated form of pp60^src.

GM7373 cells were allowed to adhere to the different substrata or were maintained in suspension exactly as described above. At the end of incubations, cells were lysed in Rho binding lysis buffer (50 mM Tris-HCl pH 7.4 containing 100 mM NaCl, 1% Nonidet P-40, 2 mM MgCl₂, 10% glycerol and the protease inhibitors mixture described above) and centrifuged (20 minutes at 16,000 g). Protein concentration was evaluated in the superrnatants and 300 μg protein/sample were incubated for 1 hour at 4°C with GST-Rho-GTPase-binding domain (30 μg/sample) immobilized to glutathione-Sepharose 4B (Sigma). Beads were then centrifuged (1 minute at 1000 g) and pellets were washed extensively with Rho binding lysis buffer, resuspended in reducing SDS-PAGE sample buffer, incubated for 5 minutes at 90°C, and analysed on SDS-15% PAGE under reducing conditions followed by western blotting using anti-RhoA antibodies.

GM7373 cell cultures (24 well plates) were incubated at 37°C in serum-free medium for 24 hours. Then, cells were added with GST-Tat in the absence or presence of the inhibitors under test and further incubated at 37°C for different periods of time. Alternatively, GM7373 cells maintained in serum-free medium for 24 hours were detached from culture dishes by 15 minutes incubation in PBS containing 15 mM EDTA and resuspended in medium containing 1% FCS and 25 mM Hepes, pH 7.5. Aliquots of 1×10⁶ cells were then allowed to adhere to non-tissue culture dishes (35 mm in diameter) coated with 20 μg/ml GST-Tat, FN, or PL for 3 hours at 37°C. At the end of incubation, the amount of activated NF-κB p50 or p65 subunit present in the cell extracts was determined using the TransAM NF-κB p50 or p65 Assay Kits (Active Motif, Carlsbad, CA, USA) according to the manufacturer's instructions (Renard et al., 2001).

GM7373 cells were allowed to reach confluence onto 3.5 cm polystyrene tissue culture plates or non-tissue culture plates coated with 20 μg/ml of GST-Tat, FN or PL in medium containing 10% FCS. After extensive washing, EC monolayers were wounded with a rubber policeman and incubated at 37°C with medium containing 0.4% FCS in the absence or presence of free GST-Tat (100 ng/ml) or of the different antagonists. Photomicrographs were taken under an inverted microscope (Olympus 1×51 microscope with a Camedia C-4040 digital camera, ×10/0.25; Olympus Biosystem, Munich, Germany) and wound repair was evaluated by measuring the area of the wound by computerized image analysis using the Image Pro-Plus analysis system.

Human α_vβ₃ integrin was assessed for its capacity to interact with GST-Tat in two different cell-free models. Firstly, BSA, FN, GST-Tat and GST devoid of the Tat moiety were immobilized onto non-tissue culture plastic and assessed for their capacity to bind α_vβ₃. As shown in Fig. 1A, α_vβ₃ binds to immobilized GST-Tat but not to FN, BSA or GST. Next, increasing concentrations of α_vβ₃ were injected over GST-Tat immobilized onto a BIAcore CM5 sensorchip and a set of sensorgrams was obtained (Fig. 1B). After subtraction of the blank, sensorgrams were used to calculate kinetic parameters: an association rate constant (k_on) equal to 1.16×10⁷ M^-1 s^-1 and a dissociation rate constant (k_off) equal to 3.78×10^-1 s^-1 characterized the interaction of α_vβ₃ with immobilized GST-Tat. Thus, GST-Tat/α_vβ₃ interaction occurs with a dissociation constant (K_d)=k_off/k_on equal to 32 nM. In some experiments, the association phase of GST-Tat/α_vβ₃ interaction was allowed to proceed to equilibrium and, after subtraction of the blank, the data were used to calculate an affinity value independent of the kinetics of binding. This analysis demonstrates that GST-Tat/α_vβ₃ interaction occurs with a K_d equal to 36 nM, a value consistent with our kinetics analysis. The association of α_Vβ₃ with GST-Tat is specific, since the integrin binds poorly to a BSA-coated sensorchip, with RU values similar to those obtained after the injection of buffer alone onto GST-Tat- or BSA-coated sensorchips (Fig. 1C).

Plastic coated with GST-Tat promotes adhesion of endothelial GM7373 cells with a maximum effect at 20 μg/ml, corresponding to about 96,000 adherent cells/cm² (Fig. 2A). Under the same experimental conditions, FN allows the adhesion of about 120,000 cells/cm², whereas no significant cell adhesion was observed on plastic coated with BSA or GST protein. Endothelial GM7373 cells express α_vβ₃ (Rusnati et al., 1997). Accordingly, the highly specific anti-α_vβ₃ monoclonal LM 609 antibody (Cheresh, 1987), but not irrelevant IgGs, inhibits cell adhesion to GST-Tat but not to FN (Fig. 2B). Integrin α_vβ₃ binds to both the RGD motif and the basic domain present in several adhesive proteins (Gehlsen et al., 1992; Mitola et al., 2000; Vogel et al., 1993). Accordingly, the Tat mutants GST-Tat 1e (characterized by the deletion of the amino acid sequence that contains the RGD sequence) and GST-Tat_{R49/52/53/55/56/57A} (in which the arginine residues 49, 52, 53, 55, 56, 57 within the basic domain were mutated to alanine residues) are characterized by a significantly lower cell-adhesive capacity with respect to the wild-type protein. In contrast, the mutant GST-Tat_Δ1-21 (containing a deletion of the amino acid sequence 1-21 that represents an absolute requirement for Tat transactivating activity) (Demarchi et al., 1996) retains a full cell-adhesive capacity (Fig. 2C). Also, GST-Tat loses its cell adhesive capacity after heat denaturation (Fig. 2C). These data indicate that the RGD motif, the basic domain, and a proper tridimensional conformation are required for Tat to interact with α_vβ₃ integrin present on the surface of ECs.

The cell adhesive capacity of GST-Tat is not restricted to GM7373 cells. Indeed, under the same experimental conditions, ECs of different origin, including HUVECs and MAECs, adhere to immobilized GST-Tat or FN but not to BSA (Fig. 2D).

Integrin-dependent EC adhesion can be considered the first step of a process that, together with EC migration and proliferation, leads to neovascularization (Folkman and Klagsbrun, 1987). EC activation following a mechanic wound of an EC monolayer resembles, at least in part, this process (Lauder et al., 1998). On this basis, we investigated the ability of ECs to repair a mechanically wounded monolayer (motogenic activity) when cells were allowed to adhere to immobilized GST-Tat. As shown in Fig. 2E,F, ECs adherent to immobilized GST-Tat show an improved capacity to repair the wounded monolayer when compared to cells seeded on FN or PL. The effect is inhibited by the α_vβ₃ antagonist cRGDfV but not by the control cRADfV peptide. At variance with substratum-immobilized GST-Tat, free GST-Tat does not induce motogenic activity when added to wounded GM7373 cell monolayers adherent to tissue culture plastic or to non-tissue-culture plastic coated with PL, FN or GST-Tat (Fig. 2E).

In conclusion, immobilized Tat induces adhesion and motogenic activity in ECs by interacting with α_vβ₃ integrin.

The activation of FAK, as well as of the downstream second messengers RhoA and pp60^src, are early integrin signalling events following EC adhesion and spreading (Kumar, 1998; Schlaepfer et al., 1999). To investigate whether these second messengers are activated following EC adhesion to immobilized Tat, GM7373 cells were seeded on plastic coated with GST-Tat or PL and incubated for 30 minutes at 37°C. GM7373 cells maintained in suspension for the same period of time were used as controls. As shown in Fig. 3A, FAK, RhoA, and pp60^src are activated in GM7373 cells adherent to immobilized GST-Tat but not in cells adherent to PL or maintained in suspension, indicating the specificity of the effect.

α_vβ₃ is expressed both at the basal and luminal aspects of ECs (Conforti et al., 1992) and Tat is present in the serum of AIDS patients (Westendorp et al., 1995). To assess whether the engagement of luminal α_vβ₃ by free Tat could also activate FAK in ECs, GM7373 cells adherent to tissue culture plastic were exposed to free GST-Tat. As shown in Fig. 3B,C, free GST-Tat induces FAK phosphorylation in GM7373 cells in a dose- and time-dependent manner (maximum effect being observed at 100 ng/ml and 30 minutes, respectively). Also, by using phosphospecific anti-FAK antibodies, we found that GST-Tat promotes the phosphorylation of FAK at position Y397 and Y577 in GM7373 cells, without affecting the levels of phosphorylation at Y407, Y576, Y861, and Y925 (Fig. 3D).

To demonstrate that FAK phosphorylation by free Tat is triggered via α_vβ₃ engagement, free GST-Tat was administered to GM7373 cells in the absence or in the presence of the anti-α_vβ₃ antibody LM609 or of cRGDfV cyclic peptide. As shown in Fig. 3E, both LM609 and cRGDfV inhibit FAK phosphorylation by free GST-Tat. However, no effect is exerted by an irrelevant antibody or by the cRADfV control peptide. Also, cRGDfV does not affect the basal levels of FAK phosphorylation in the absence of GST-Tat. Finally, free GST-Tat also triggers FAK phosphorylation in MAECs without affecting the level of unphosphorylated FAK protein (Fig. 3F).

To assess whether FAK connects α_vβ₃/Tat interaction to RhoA/pp60^src activation in ECs, we stably transfected GM7373 cells with the cDNA encoding for the FAK C-terminal domain (FRNK). FRNK exerts a dominant negative effect on FAK activation promoting its dephosphorylation at Tyr₃₉₇. This dominant-negative effect is abolished by the L1034S point mutation that prevents FRNK localization to focal contact sites (Sieg et al., 1999). After transfection with expression vectors harbouring the cDNA encoding for FRNK or FRNK_L1034S both tagged with a triple-HA, GM7373 transfectants were firstly screened for transgene expression by western blotting using specific anti-HA antibodies (data not shown). Only clones characterized by a similar expression of HA were utilized further. Among them, two clones expressing FRNK (named FRNK-9 and FRNK-18) and two clones expressing FRNK_L1034S (named FRNK_L1034S-4 and FRNK_L1034S-7) were used for further experiments. Parental GM7373 cells, FRNK-9 and FRNK_L1034S-7 clones (Fig. 4A), as well as FRNK-18 and FRNK_L1034S-4 (data not shown), are characterized by similar levels of expression of total FAK protein. Also, the different transfectants, but not parental GM7373 cells, express similar levels of FRNK or FRNK_L1034S proteins.

On these bases, EC adhesion and spreading and consequent activation of FAK, RhoA and pp60^src were evaluated in the different cell lines. As shown in Fig. 4B,C, FRNK or FRNK_L1034S overexpression does not hamper cell adhesion and spreading of ECs to immobilized Tat or FN. Nevertheless, Tat-adherent FRNK cells show a remarkable decrease in FAK phosphorylation when compared to parental or FRNK_L1034S transfected cells (Fig. 4D). Accordingly, FRNK cells adherent to immobilized Tat failed to activate RhoA but they retained the capacity to phosphorylate pp60^src (Fig. 4D).

Taken together, the data indicate that α_vβ₃-mediated adhesion of ECs to immobilized Tat triggers FAK phosphorylation, which, in turn, is required for the activation of RhoA, but not of pp60^src.

FAK phosphorylation triggers NF-κB activation (Funakoshi-Tago et al., 2003; Scatena et al., 1998) and NF-κB is activated by Tat in different cell types (Demarchi et al., 1996; Nath et al., 1999). On these bases, we assessed whether α_vβ₃-dependent EC adhesion to Tat and consequent FAK, RhoA and/or pp60^src activation triggers NF-κB activation. As shown in Fig. 5A, the p50 and p65 subunits of NF-κB are activated in GM7373 cells adherent to immobilized GST-Tat or FN [here used as a positive control (Scatena et al., 1998)] but not to PL nor in cells maintained in suspension.

GST-Tat also induces NF-κB activation in a time- and dose-dependent manner when administered as a free molecule to GM7373 cells (Fig. 5B,C). Tat-dependent NF-κB activation is inhibited by the NF-κB inhibitor SN50 but not by its inactive analogue SN50M (Fig. 5D). The capacity of cRGDfV, but not of cRADfV, to hamper NF-κB activation by Tat indicates the role of α_vβ₃ engagement in this biological response (Fig. 5D). Furthermore, Tat fails to activate NF-κB in FRNK but not in FRNK_L1034S transfectants (Fig. 5E). Finally, NF-κB activation is inhibited by the RhoA inhibitor exoenzyme C3 and by the pp60^src inhibitor PP₂ but not by its inactive control PP₃ (Fig. 5D). Taken together, the data indicate that NF-κB activation by Tat requires α_vβ₃/Tat interaction and FAK, RhoA and pp60^src activation.

In a final series of experiments we assessed the involvement of the second messengers activated by Tat in the induction of motogenic activity in ECs. As shown in Fig. 6A, FRNK overexpression hampers the capacity of immobilized Tat to induce motogenic activity in adherent ECs. No inhibition was instead exerted in this assay by overexpression of the inactive FRNK_L1034S mutant.

Also, the RhoA inhibitor exoenzyme C3 and the pp60^src inhibitor PP₂, but not its inactive control PP₃, prevented the migration of the wounded EC monolayer on immobilized Tat. A similar effect is also exerted by the NF-κB inhibitor SN50, but not by its inactive analogue SN50M (Fig. 6B). Thus, FAK, RhoA, pp60^src and NF-κB activation are all implicated in the motogenic activity triggered by immobilized Tat in ECs.

HIV-1 Tat interacts with α_vβ₃ integrin of ECs inducing their adhesion to the substratum, but little is known about the molecular bases and biological consequences of this process.

By exploiting real-time surface plasmon resonance, here we shown the high affinity (K_d=32-36 nM) and rapid kinetics (k_on and k_off=1.16×10⁷ M^-1 s^-1 and 3.78×10^-1 s^-1, respectively) of Tat/α_vβ₃ interaction. Interestingly, the binding of α_vβ₃ to classical adhesive proteins such as fibrinogen or vitronectin (VN) is characterized by similar affinity (27 nM and 64 nM for fibrinogen and VN, respectively) but slower dissociation rate (k_off=9.8×10^-4 s^-1 and 2.1×10^-4 s^-1 for fibrinogen and VN, respectively) (Takagi et al., 2002), demonstrating that these latter interactions are more stable than that of Tat. In vivo, the high stability of α_vβ₃/adhesive proteins interaction is required to provide a firm anchorage to adherent ECs. At variance with this, the rapid kinetics of Tat/α_vβ₃ interaction suggest a more `dynamic' adhesion, functional to migration of ECs adherent on substratum-immobilized Tat. Accordingly, here we reported that immobilized Tat stimulates a motogenic activity in adherent ECs.

To date, the process of angiogenesis has been viewed as the result of two distinct sets of inputs conveyed to ECs by free angiogenic growth factors interacting with their specific receptors of the luminal aspect of ECs, and by ECM components mainly interacting with integrins at the basal aspect of ECs. Here we observed that Tat acts simultaneously as an adhesive protein and as a free molecule, inducing ECs adhesion, signal transduction and motogenic activity. Accordingly, Tat triggers activation of intracellular second messengers both when presented to ECs as a substratum-immobilized or a free protein. However, we were not able to induce repair of a wounded monolayer of ECs adherent to tissue culture plastic by administering free Tat (Fig. 2E) at doses sufficient to induce second messengers activation in the same experimental conditions (see Figs 3 and 5) and comparable to the amount of Tat that remains coating the plastic (Rusnati et al., 1998). Accordingly, no significant differences were seen in wound repair when free Tat was administered to ECs adherent to Tat itself, FN or PL. These observations suggest that, in its substratum-immobilized form, Tat may be presented to ECs in a more appropriate way or may be more persistent in its stimulation. Alternatively, substratum-immobilized Tat, but not free Tat may alter cell-matrix interaction and cell traction, favouring EC motility.

Tat can be found immobilized in the ECM, where it associates with heparin-like HSPGs, and is protected from proteolitic degradation (Chang et al., 1997). Relevant to this point, a single heparin chain is able to bind up to six molecules of Tat (Rusnati et al., 1999) and Tat interaction with substratum-immobilized heparin is far more stable (k_off=2.7×10^-3 s^-1) (Rusnati et al., 2001) than that with α_vβ₃ (see above). These findings suggest the possibility that Tat accumulates in the ECM and increases its concentration in the microenvironment providing a firm substratum for EC adhesion and migration during angiogenesis in vivo. This possibility is further supported by the observation that the binding of Tat to HSPGs or to α_vβ₃ are not mutually exclusive, since free heparin does not inhibit Tat/α_vβ₃ interaction or EC adhesion to Tat (Urbinati et al., 2004). The capacity of substratum-immobilized Tat to induce signal transduction and motogenic activity in adherent ECs can find a physiological counterpart in vivo, where Tat released by HIV-positive cells in the lymph nodes may remain immobilized in the HSPGs-rich basal lamina of the capillaries during diffusion to the blood stream.

α_vβ₃ engagement by immobilized Tat induces FAK phosphorylation that, in turn, triggers the activation of different intracellular second messengers. In neurons, the residue Leu₁₀₃₄ of FAK acts as a docking site for p190RhoGEF, a Rho-specific GDP/GTP exchange factor (Zhai et al., 2003). Accordingly, we found that abrogation of FAK activity hampers the capacity of ECs to activate RhoA in response to their adhesion to immobilized Tat. Also, phosphorylation of FAK Tyr₃₉₇ creates a docking site for SH₂-containing src kinases, including pp60^src (Hsia et al., 2003). Here we found that, even though adhesion of ECs to plastic-immobilized Tat induces pp60^src phosphorylation, abrogation of FAK activity by FRNK overexpression does not affect the capacity of ECs to phosphorylate pp60^src in response to Tat. This lack of inhibition can be explained by the possible activation of pp60^src via a direct interaction with the cytoplasmic domain of β integrin subunit (Arias-Salgado et al., 2003). Alternatively, Tat can bind and activate tyrosine kinase VEGF receptor KDR, whose engagement can lead to a direct activation of src kinases (Chou et al., 2002). Further experiments are required to elucidate the complex interplay among integrin, KDR, FAK, RhoA and pp60^src in Tat-stimulated ECs.

Whatever their mutual interactions, α_vβ₃, FAK, RhoA and pp60^src activation are essential for motogenesis of Tat-adherent ECs. Interestingly, FAK abrogation inhibits the motogenic activity of Tat without affecting EC adhesion and spreading. However, we observed that FRNK overexpression inhibits Tat-induced proliferation of ECs (C.U., unpublished data), suggesting that, beside adhesion and spreading, the activation of α_vβ₃, FAK, RhoA and pp60^src are also implicated in the modulation of the mitogenic activity of Tat. Interestingly, Tat-induced FAK activation has been associated with migration and permeability of brain microvascular ECs (Avraham et al., 2004).

Our data demonstrate that NF-κB connects α_vβ₃, FAK, RhoA and pp60^src activation to the motogenic activity triggered by Tat in ECs. Tat has been shown to activate NF-κB in different cell types including astrocytes (Conant et al., 1998), macrophages (Kumar et al., 1999), T-cells (Li-Weber et al., 2000) and ECs (Cooper et al., 1996; Cota-Gomez et al., 2002; Pieper et al., 2002; Toborek et al., 2003), but the mechanism(s) of activation are not fully elucidated. Indeed, Tat can induce NF-κB activation both when administered extracellularly (Cota-Gomez et al., 2002) and when produced endogenously (Pieper et al., 2002). Here, we show that NF-κB is activated in ECs following extracellular interaction of Tat with α_vβ₃ and the consequent activation of FAK, RhoA and pp60^src. In ECs, NF-κB activation by Tat has, to date, been connected to the overexpression of surface adhesive molecules and/or activation of inflammatory pathways (Cooper et al., 1996; Cota-Gomez et al., 2002; Pieper et al., 2002; Toborek et al., 2003). The data here reported implicate NF-κB in the motogenic activity exerted by Tat in vitro, suggesting its involvement also in the neovascularization induced by Tat in vivo. Accordingly, adhesion of ECs to FN activates a NF-κB-dependent program of gene expression related to angiogenesis (Klein et al., 2002). These findings point to a possible general involvement of NF-κB in neovascularization triggered by `more classical' angiogenic growth factors. To date, the role of NF-κB in FGF2 biology is still being debated (Klein et al., 2002; Kroon et al., 2001; Messmer et al., 2000; Mohan et al., 2000; Pollet et al., 2003; Stoltz et al., 1996) and little data are available about NF-κB in VEGF-dependent angiogenesis. However, the possibility that NF-κB is involved in the biology of these two growth factors is also inferred by their capacity to accumulate in the ECM, thus inducing integrin-dependent EC adhesion and pro-angiogenic programs (Hutchings et al., 2003; Rusnati et al., 1997; Tanghetti et al., 2002).

Beside ECs, Tat induces substratum adhesion of Kaposi's sarcoma-derived spindle cells (Barillari et al., 1993), neurons (Cornaglia-Ferraris et al., 1995), myoblasts (Vogel et al., 1993), and epithelial cells (C.U. and C. Ravelli, unpublished), FAK phosphorylation in neurons (Milani et al., 1998) and NF-κB activation in various cell types (see above). Taken together, these observations suggest that Tat/α_vβ₃-dependent signal transduction pathway(s) are implicated in a broad array of biological effects connected with AIDS-associated pathologies, making them a suitable target for the design of anti-Tat strategies.

We wish to thank Dr. H. Lortat-Jacob for helpful discussion. This work was partially supported by grants from ISS (AIDS Project), from MIUR (Cofin 2003 and Fondi Locali per la ricerca) to M.R., from AIRC, MIUR (Centro di Eccellenza `IDET', Firb 2001, Cofin 2004), Fondazione Berlucchi and ISS (Oncotechnological Program) to M.P.