Transendothelial Migratory Pathways of Vδ1⁺TCRγδ⁺ and Vδ2⁺TCRγδ⁺ T Lymphocytes from Healthy Donors and Multiple Sclerosis Patients: Involvement of Phosphatidylinositol 3 Kinase and Calcium Calmodulin-Dependent Kinase II¹

It is well accepted that multiple sclerosis (MS)³ is an autoimmune disease characterized by demyelinating lesions and involvement of the CNS (1, 2). Several cell types and soluble mediators contribute to the onset and progression of demyelinization, including T lymphocytes (2, 3, 4); in particular, some previous reports point to an active role played by the γδ T cell subset in the pathogenesis of MS (5, 6, 7).

Two main fractions of γδ T cells have been described. One, expressing the TCR variable regions Vγ9 and Vδ2, represents the majority of peripheral blood γδ lymphocytes; Vδ1 T cells are instead resident within epithelial tissues, where they may provide a first line of defense against infectious agents and cancer (8, 9, 10). This small subset of γδ T cells represents a relevant proportion of the mucosal-associated lymphoid tissue, and it is unclear whether and how they recirculate. Notably, an increase of circulating Vδ1⁺ T lymphocytes has been detected in AIDS (11), suggesting that this subpopulation can also recirculate under certain pathological conditions.

In contrast, cells of the Vδ2 subset infiltrate chronic lesions and are detectable in the cerebrospinal fluid (CSF) in relapsing-remitting MS (7, 8, 12). To reach the site of lesion(s), γδ T lymphocytes must leave the bloodstream and migrate across the endothelial barrier. We have previously reported that, in the acute phase of disease, the Vδ2⁺ NKRP1a⁺ γδ T cell subset is expanded in the blood of MS patients; these cells display a high transendothelial migratory capability, apparently in the absence of chemotactic stimuli, and use the NK receptor protein 1a (NKRP1a; CD161) for this function (13).

In mice, γδ T cells recirculate preferentially through nonlymphoid tissues and show a particular tropism for skin and gut (14). The precise mechanism for this preferential migration is unclear; in general, lymphocyte extravasation is initiated by rolling on vascular endothelium, sustained through the engagement of specific glycoproteins, including P-selectin glycoprotein ligand 1 (PSGL1), by selectins, followed by integrin-induced adhesion to endothelial cells and leukocyte arrest (15, 16, 17). Then, lymphocytes must locomote to intercellular junctions between endothelial cells, where transmigration occurs (18). Among the adhesion molecules known to participate in extravasation, the platelet endothelial cell adhesion molecule 1 (PECAM1; CD31), which is expressed both at the endothelial cell borders and at the surface of leukocytes, creates a haptotactic gradient able to recruit circulating cells and to drive transmigration (19). To complete this process, leukocytes must reorganize their cytoskeletons and polarize toward endothelial cell junctions (20); however, little is known about the signaling events that promote successful transendothelial migration after lymphocyte arrest and recruitment.

Recently, we reported that PECAM1/CD31 is associated to the phosphatidylinositol 3 kinase (PI-3K) in human neutrophils (21). PI-3K was first described as a kinase that phosphorylates phosphoinositols at the D-3′ position of the inositol ring and produces novel inositides (22, 23). PI-3K can also be activated via LFA1, in the earliest steps of adhesion (23). Once activated, PI-3K induces phosphorylation and activation of the serine-threonine kinase Akt/protein kinase (PK)Bα (24). In contrast, in neutrophils, another kinase, the calcium calmodulin-dependent kinase II (CAMKII) has been shown to be crucial in mediating cytoskeletal rearrangement and cell polarization during locomotion (25).

In this paper, we show that Vδ1⁺ T lymphocytes do not bear NKRP1a and selectively express PECAM1, which is involved in their transendothelial migration, at variance with Vδ2⁺ T cells, which lack PECAM1 and use NKRP1a for transmigration. We also demonstrate that Vδ1 T lymphocytes use PI-3K, while Vδ2⁺ cells are CAMKII dependent, for transmigration. Interestingly, NKRP1a occupancy leads to the activation of CAMKII, whereas PECAM1 engagement triggers the PI3-K-induced activation of Akt/PKBα.

Blood samples were obtained from six healthy volunteers and from six patients with clinically active MS (Dipartimento di Neuroscienze Lancisi, Ospedale San Camillo, and Dipartimento di Scienze Neurologiche, Università Tor Vergata, Rome, Italy) according to protocols approved by the human experimentation committees of these two institutes. Patients were in the relapsing phase or in the first episode of disease, with abnormal magnetic resonance imaging brain scan; none had received immunosuppressive treatment for at least 3 mo before entering the study.

The anti-NKRP1a mAb 191B8 and the anti-CD31 mAb M89D3 (IgG2a), the anti-CD11a mAb 70H12 (IgG2a), the anti-Vδ1 mAb A13 (IgG1), and the anti-Vδ2 mAb BB3 (IgG1) were prepared as described (26, 27, 28, 29). The anti-HLA class I mAb W6/32 (IgG2a)-producing hybridoma was from the American Type Culture Collection (Manassas, VA). W6/32, 191B8, and M89D3 mAbs were purified from ascites fluid by affinity chromatography, and pepsin-digested F(ab′)₂ were prepared as described (26). The anti-CD3 mAb (UCHT-1, IgG1) was a kind gift from P. C. L. Beverly (Imperial Cancer Research Fund, London, U.K.). Purified anti-CD161 mAb and purified P-selectin-IgG fusion protein were from BD PharMingen (Milan, Italy). The affinity-purified FITC- or PE-conjugated goat anti-mouse (GAM) anti-isotype specific antiserum was from Southern Biotechnology Associates (Birmingham, AL). Purified GAM anti-Ig (H and L chains) was purchased from ICN Biomedicals (Aurora, OH). Recombinant IL-2 was from PeproTech (London, U.K.), rRIL-12 was from R&D Systems (Oxon, U.K.), and PHA was from Life Technologies (Grand Island, NY). All cells used in our experiments were cultured in RPMI 1640 medium (Biochrom, Berlin, Germany) supplemented with FCS (HyClone Laboratories, Logan, UT), human AB serum (BioWhittaker, Walkersville, MD), glutamine, and penicillin-streptomycin (Biochrom). The PI-3K inhibitors wortmannin and LY294002 were from Sigma-Aldrich (St. Louis, MO), and the CAMKII inhibitors KN62 and KN93 and the inactive compound KN92 were from Calbiochem-Merck (Darmstadt, Germany).

PBMC from healthy donors or MS patients were isolated by Ficoll-Hypaque (Biochrom) gradient. Adherent cells were depleted after adhesion to plastic petri dishes for 2 h at 37°C. Highly purified CD3⁺γδ⁺ Τ cells were obtained from PBMC following staining with anti-Vδ1 and anti-Vδ2 mAbs and cell sorting using a MoFlo cell sorter (Cytomation, Fort Collins, CO). To obtain clones and lines, cells were seeded at either 1 or 10 cells/well, respectively, in the presence of irradiated PBMC as feeder cells (10⁵ cells/well) in 96-well U-bottom microplates (Greiner, Nurtingen, Germany) and cultured in RPMI 1640 medium supplemented with 200 mM l-glutamine, 5% FCS, 5% AB serum, 1 μg/ml PHA, and 25 U/ml rIL-2. Cells were then expanded with rIL-2 and restimulated every 3 wk with PHA and irradiated feeder cells (5000 rad) according to standard procedure (13). In some experiments, cells were washed to remove rIL-2 and cultured in rIL-12-containing medium (5 ng/ml) for another 5 days.

Immunofluorescence staining of cultured cells was performed as described elsewhere (27). Briefly, aliquots of 10⁵ cells were stained with the corresponding mAb followed by FITC- or PE-conjugated anti-isotype-specific GAM serum (Southern Biotechnology Associates). Control aliquots were stained with isotype-matched irrelevant mAbs (BD Biosciences, Mountain View, CA) followed by FITC- or PE-GAM or with the fluorescent reagent alone. Samples were analyzed on a flow cytometer (FACSort; BD Biosciences) equipped with an argon ion laser exciting PE at 488 nm. Data were analyzed using Lysis II (version 1.1; BD Biosciences) and are expressed as log red fluorescence intensity (mean fluorescence intensity (MFI), arbitrary units (a.u.). Calibration was assessed with CALIBRITE particles (BD Biosciences) using the AutoCOMP computer program (version 2.1.2; BD Biosciences). In multicolor analysis, freshly isolated PBMC was stained with the following mAbs: anti-CD3 PE-Cy7, anti-CD31 PE (both from Caltag Laboratories, Burlingame, CA), and anti-CD161-allophycocyanin (BD Biosciences). Cells were analyzed with a MoFlo cytometer (Cytomation) and data were analyzed with FlowJo software (TreeStar, San Carlos, CA).

HUVECs were isolated and cultured as described (30) and used within four passages. Endothelial confluent monolayers were tested for their integrity before the migration assay as described (30), using Transwell cell culture chambers (polycarbonate filters, 3-μm pore size; Costar, Cambridge, MA). Vδ1⁺ or Vδ2⁺ T cells from healthy donors or MS patients were used after culture in rIL-2 or, when indicated, were recovered, washed twice, and cultured for another 6 days in the presence of rIL-12 (1 ng/ml). In some experiments, Vδ1⁺ or Vδ2⁺ T cell clones were preincubated for 30 min at 4°C with saturating amounts (5 μg/ml/10⁶ cells) of the W6/32, 191B8, or M89D3 mAbs and washed before the transmigration assay. In other experiments cells were exposed for 30 min to the PI-3K inhibitors wortmannin or LY294002, to the CAMKII inhibitors KN62 or KN93, or to the inactive compound KN92. To quantitatively express the results of transmigration assays, Vδ1⁺ or Vδ2⁺ T cell clones were labeled with ⁵¹Cr (sodium chromate; NEN, Boston, MA) and added to the upper compartment of the Transwell chamber, as described (26, 28). At different time points, migrated cells were recovered from the lower compartment of the chamber and lysed with 100 mM Tris-HCl (pH 7.4) containing 0.1% Triton X-100 (Sigma-Aldrich). The radioactivity of the samples was measured in a gamma counter. Results are expressed as the percentage of migrating cells calculated as described (26). Statistical analysis was performed using the Student t test and variance analysis.

PI-3K activity was tested indirectly by analyzing activation of the serine/threonine kinase Akt1/PKBα in cell lysates of Vδ1⁺ or Vδ2⁺ T cells with the commercial assay kit, using the specific substrate and [γ-³²P]ATP, after immunoprecipitation with the specific anti-Akt Ab (Upstate Biotechnology, Lake Placid, NY). Akt1/PKBα activity was tested upon ligation of PECAM1 obtained by incubation of the cells with 5 μg/ml of the specific mAb M89D3 followed by 10 μg/ml GAM Ig serum as described (21). As a control, cells were exposed to GAM Ig serum after treatment with an unrelated isotype-matched Ab (BD Biosciences). Some experiments were conducted in the presence of the PI-3K inhibitor LY294002. CAMKII activation in γδ cells was tested upon cross-linking of NKRP1a, PECAM1, or HLA-I obtained with the specific mAb 191B8, M89D3, or W6/32 (all IgG2a, 5 μg/ml) followed by GAM Ig serum. CAMKII activity was measured with the CAMKII assay kit, using the specific substrate and [γ-³²P]ATP, after immunoprecipitation with the specific anti-CAMKII Ab (Upstate Biotechnology) and chromatography. Results are expressed as cpm × 10⁻³ and are the mean ± SD of 10 independent experiments.

We have analyzed Vδ1⁺TCRγδ⁺ and Vδ2⁺TCRγδ⁺ bulk populations and clones for the expression of a panel of surface molecule involved in transendothelial migration, including CD161/NKRP1a, CD31/PECAM1, PSGL1, and CD11a/CD18 (LFA1) (13, 15, 18). Data depicted in Table I show that cloned Vδ1⁺ γδ T cells, while being mostly NKRP1a negative (41 of 43 clones tested were NKRP1a⁻), display high levels of PECAM1, which, in turn, is absent from Vδ2⁺ T cells (57 of 60 clones were PECAM1⁻). This evidence comes also from the ex vivo analysis of PBMC from MS patients (Fig. 1). Indeed, multicolor immunofluorescence showed that, in MS patients, the small fraction of Vδ1⁺ T lymphocytes was PECAM1⁺NKRP1a⁻ (Fig. 1, right panels). Conversely, Vδ2⁺ T cells were NKRP1a⁺ and expressed very low levels of PECAM1 (Fig. 1, left panels).

A similar clear-cut distribution has been observed in healthy donors (data not shown). However, when Vδ2⁺ γδ T cells from normal donors were analyzed at the clonal level, a small fraction (5%) was found to coexpress NKRP1a and PECAM1 (4 clones of 80 analyzed; data not shown). Furthermore, CD11a/LFA1 and CD162/PSGL1 are expressed at higher levels on Vδ2⁺ bulk populations and clones (Table I), while the expression of other adhesion molecules, such as β₁ integrins or ICAM-1, was comparable at the surface of the two γδ T cell subsets (data not shown). Altogether, these findings indicate that Vδ1⁺ and Vδ2⁺ T lymphocytes express differently two key molecules such as NKRP1a and PECAM1, reported to be involved in transendothelial migration.

We have reported that the circulating Vδ2⁺ γδ T cell subset uses NKRP1a to transmigrate across endothelial monolayers (13). In general, T cells expressing NKRP1a are capable of transendothelial locomotion in the absence of chemotactic stimuli (13, 26); this would imply that the adhesion molecule(s) involved could deliver a signal leading to leukocyte polarization (15, 16, 17, 18, 19, 20). Thus, we analyzed the signal transduction pathway associated with NKRP1a-driven γδ T cell migration, in particular CAMKII and PI-3K, which have been reported to be involved in regulating neutrophil motility (21, 25). To this purpose, Vδ2⁺ and Vδ1⁺ γδ T cell clones, derived from healthy donors and MS patients, were assayed for transmigration across HUVEC monolayers in the presence of specific inhibitors of either CAMKII or PI-3K.

Fig. 2 shows that transendothelial migration of Vδ2⁺ T lymphocytes (Fig. 2, A (MS patients) and C (healthy donors)) was faster than that of Vδ1⁺ T cells (Fig. 2, B and D). Of note, when Vδ2⁺ T cells were pretreated with the CAMKII inhibitors KN62 and KN93, but not with the inactive compound KN92, the number of migrating cells and the rate of transmigration were significantly (p < 0.05) decreased. The inhibition was dose dependent, as both KN62 and KN93 were effective from 10 to 1 μM, but not at 0.1 μM (Fig. 2,E). Consistent with this, all of Vδ2⁺ T cell clones, both from healthy donors and MS patients, display a superimposable behavior (Table II). Importantly, migration of Vδ2⁺ T cell clones was strongly inhibited (>70%) by the covering of NKRP1a with the F(ab′)₂ of the corresponding mAb (Table III). Conversely, Vδ1⁺ T cell lines (Fig. 2, B and D) and clones (Table II) were insensitive to CAMKII inhibitors; interestingly, the Vδ1⁺ clone MCL.6, which was found to express NKRP1a, displayed a higher rate of transendothelial migration and was inhibited by KN93 (Table II). In contrast, PI-3K blockers, such as Ly294002, had little or no effect on Vδ2⁺ T cell clones (Table II). Migration patterns of Vδ1⁺ and Vδ2⁺ T cell clones remained similar along the culture period (2 mo), independently of the proliferation rate of each cell line or clone analyzed (data not shown).

Because the culture of Vδ2⁺ T cells with rIL-12 leads to up-regulation of NKRP1a expression and enhancement of transendothelial migration (13), we asked whether also in this case CAMKII inhibitors could be effective; indeed, KN62 and KN93, at variance with the inactive KN92, impaired transendothelial migration of rIL-12 cultured Vδ2⁺ T lymphocytes of MS patients (Fig. 2 F) and normal donors (data not shown).

At variance with Vδ2⁺, Vδ1⁺ γδ T lymphocytes are mainly detected in peripheral tissues (8, 9, 10); however, there is no evidence, so far, that this subset is localized to CNS and/or in MS lesions. Nevertheless, we addressed the question of whether their transendothelial migration, detectable although reduced compared with that of Vδ2⁺ T cells, is driven by PECAM1 and dependent on a selected kinase. Indeed, it is well known that PECAM1 is responsible for haptotactic cell migration and that it is functionally associated with PI-3K in neutrophils (19, 21). Table III shows that the F(ab′)₂ of anti-PECAM1 mAb M89D3 was indeed able to strongly inhibit transendothelial migration of this cell subset; interestingly, the PI-3K blockers wortmannin and LY294002 exerted a dose-dependent inhibition of Vδ1⁺ T lymphocyte migration (Fig. 3, A and circles in C–F for MS patients; B and triangles in C–F for healthy donors), at variance with that found with CAMKII inhibitors (Fig. 2, B and D; see also Table II for γδ T cell clones). In contrast, migration of Vδ2⁺ T cell lines was also only partially reduced upon treatment with the PI-3K inhibitors LY294002 or wortmannin (Fig. 3, E and F). Cells incubated with DMSO (solvent of LY294002 and CAMKII inhibitors) did not have any effect on transendothelial migration (data not shown).

Thus, Vδ1⁺ and Vδ2⁺ T lymphocytes use different adhesion molecules and signaling pathways to migrate across vascular endothelium, in the absence of chemotactic stimuli, implying that tissue localization of the two γδ T cell subsets may be selectively regulated.

We further investigated whether the engagement of NKRP1a or PECAM1 induces the activation of CAMKII or Akt/PKBα (a substrate of PI-3K), respectively, on Vδ2⁺TCRγδ⁺ or Vδ1⁺TCRγδ⁺ T lymphocytes. As shown in Fig. 4,A, the engagement of NKRP1a by the use of specific mAb, at variance with an unrelated isotype-matched mAb, elicits the activation of CAMKII. This NKRP1a-mediated effect was found using Vδ2⁺TCRγδ⁺ bulk populations (data not shown) or clones from MS patients (Fig. 4) and healthy donors (data not shown), and it was inhibited by the specific CAMKII blocker KN93 (Fig. 4,A). In contrast, we found that triggering of PECAM1 on Vδ1⁺ T cells leads to activation of Akt/PKBα (Fig. 4,B), which is a hallmark of PI-3K engagement. Indeed, LY294002 exerted a strong inhibition of PECAM1-induced Akt/PKBα activation (Fig. 4,B). Importantly, the ligation of either PECAM1 or NKRP1a on Vδ2⁺TCRγδ⁺ or Vδ1⁺TCRγδ⁺ T lymphocytes did not lead to activation of CAMKII or PI-3K, respectively (Fig. 4). Taken together, these findings further support the notion that Vδ1⁺ and Vδ2⁺ may use different biochemical pathways to transmigrate through endothelium by the selective engagement of PECAM1 and NKRP1a, respectively.

In this paper we show that Vδ1⁺ γδ T lymphocytes selectively express PECAM1, which drives transendothelial migration of this cell subset, at variance with Vδ2⁺ T cells, which use NKRP1a for transmigration. The two molecules activate different signal transduction pathways: indeed, NKRP1a-mediated Vδ2⁺ T cell migration depends on the activation of CAMKII, while PECAM1-driven migration of Vδ1⁺ cells induces Akt/PKBα activation, which is a downstream target of PI-3K (24).

In MS, γδ T cells have been found in the lesions and in CSF (7, 8, 12), suggesting not only that they contribute to demyelinization but also that they egress from the bloodstream and reach the CNS to exert their pathogenic role. In agreement with this, we have previously found that circulating Vδ2⁺ T cell population expressing NKRP1a is strongly expanded in MS patients compared with healthy donors during the acute phase of the disease (13). Notably, this γδ⁺ T cell subset is able to spontaneously migrate across endothelial cell monolayers, and the NKRP1a adhesion molecule is needed for this process (13). Interestingly, cells of the Vδ1⁺ cell subset lack NKRP1a but express PECAM1, which is known to mediate haptotactic migration of leukocytes (19, 28). In both cases, cells must modify their shape and polarize toward the endothelial cell junction to complete transmigration.

Transendothelial migration of Vδ2⁺ and Vδ1⁺ T cell subsets depends on the activation of distinct kinases; indeed, the former is inhibited by blockers of CAMKII and the latter is inhibited by PI-3K inhibitors. In turn, the engagement of NKRP1a on Vδ2⁺ lymphocytes leads to activation of CAMKII, while PECAM1 oligomerization on Vδ1⁺ T cells triggers the activation of the PI-3K-dependent Akt/PKBα. Thus, the two subsets use different adhesion molecules and signaling pathways to transmigrate, suggesting that resident and circulating γδ T lymphocytes are equipped with distinct biochemical and molecular mechanisms to regulate their selective tissue localization. This might also account for the finding that in relapsing-remitting MS, γδ T cells found in the lesions differ from those in CSF (5, 6, 12, 31), supporting the hypothesis that the two γδ T cell subsets play distinct roles in the pathogenesis of the disease.

Both the above-mentioned kinases have been implied in the regulation of cell locomotion: in particular, CAMKII regulates vascular muscle cell migration (32) and neutrophil motility (25). It is conceivable that the shape of the migrating cell is modified by regulating calcium-dependent microtubule assembly (25, 33). Other studies report that chemotaxis of leukemic T cells induced by soluble VCAM1 requires activation of CAMKII and phosphorylation of vimentin (34), possibly contributing to cell polarization and initiating amoeboid locomotion. In turn, PI-3K seems to be coupled preferentially to haptotactic migration; indeed, keratinocytes and osteoblast-like cells use a PI-3K-dependent pathway to trigger cell movement driven by matrix components such as laminin (35, 36). This is of interest because in our experimental system PI-3K apparently regulates transendothelial migration of the Vδ1⁺ T cell subset, which is driven by a molecule, PECAM1, known to create a haptotactic gradient in endothelial cell junctions (19, 28). In this case, the mechanism underlying cell shape modulation seems to be focused on the reorganization of actin microfilaments, controlling filopodia formation and reducing focal adhesion (20, 37). In the last years, it has become evident that PECAM1 is important in signal transduction (38); in particular, we have reported that in neutrophils this adhesion molecule is associated with PI-3K and regulates integrin-mediated cell adhesion to endothelial cells and subendothelial matrix (21).

It is of interest that PI-3K is also coupled to chemokine receptor signal transduction. Recently, it has been reported that chemokine-induced high-affinity state of the β₂ integrin LFA1 is controlled by a PI-3K-dependent signaling pathway (39). This might account, at least in part, for the faster kinetics of transendothelial migration displayed by Vδ2⁺ T cells, which we showed to highly express LFA1. Indeed, a rapid integrin-driven lymphocyte arrest conceivably facilitates the interaction between other adhesion molecules and their endothelial ligand(s) involved in the progression of transmigration. Likewise, the presence of PSGL1 on this subset might contribute to start transendothelial motility, in keeping with other data (16).

The preferential usage of a PI-3K-dependent pathway by the resident subset of γδ T lymphocytes might also imply that they have been recruited to the tissues by soluble factors (e.g., chemokines) produced by T lymphocytes or other cell types, especially during acute inflammation (3, 40), as in the acute phase of MS. Finally, the identification of signaling pathways that differently regulate the homing of the two γδ T cell subset might be of interest to selectively block the recruitment, at the site of lesion, of cells involved in the pathogenesis of MS (4, 41).