Abstract
Human microvascular endothelial cells (ECs) constitutively express MHC class II in peripheral tissues, the function of which remains unknown. In vitro assays have established that the recognition of EC MHC class II can affect cytokine expression, proliferation, and delayed transendothelial migration of allogeneic memory, but not naive, CD4+ T cells. Previously, we have shown that effector memory CD4+ T cells will rapidly transmigrate in response to the inflammatory chemokine IFN-γ-inducible protein-10 (IP-10) in a process contingent upon the application of venular levels of shear stress. Using two models that provide polyclonal TCR signaling by ECs in this flow system, we show that TCR engagement antagonizes the rapid chemokine-dependent transmigration of memory CD4+ T cells. Inhibitor studies suggest that TCR signaling downstream of Src family tyrosine kinase(s) but upstream of calcineurin activation causes memory CD4+ T cell arrest on the EC surface, preventing the transendothelial migration response to IP-10.
The only established function of class I and class II MHC molecules is to present bound peptides in a form that may be recognized by TCRs for Ags. It is thus of considerable interest that in uninflamed peripheral tissues of large mammals, including humans, microvascular endothelial cells (ECs)3 display both class I and II MHC molecules at substantially higher levels than any other resident cell type with the possible exception of dendritic cells (1, 2, 3). Although human ECs down-regulate MHC molecule expression in culture, losing class II molecule expression completely, expression is readily restored by exposure to IFN-γ (4, 5). IFN-γ also up-regulates certain costimulators such as ligands for ICOS, OX40 and 4-1BB that, along with constitutively expressed lymphocyte function-associated Ag-3 (LFA-3) (CD58), allow cultured human ECs to effectively activate cytokine synthesis and proliferation of allogeneic memory T cells (defined by the expression of CD45RO) in an MHC-specific manner (6, 7, 8, 9, 10). Unlike professional APCs, human ECs do not typically express B7 molecules (CD80 or CD86), and this may contribute to the explanation of why ECs are inefficient at activating naive T cells (defined by expression of CD45RA) (11, 12, 13).
Despite the well-established functionality of MHC molecules on cultured human ECs, the in vivo function(s) of MHC molecule expression on ECs is unknown. Two nonexclusive possibilities are that ECs display MHC molecules bearing self-derived peptides to provide low affinity interactions with the TCRs of circulating memory cells as a survival signal maintaining the memory pool or that ECs present nonself-derived (e.g., microbial) peptides (or in the case of transplantation, nonself MHC molecules) to promote the local recruitment and activation of Ag-specific memory T cell populations. In support of the latter idea, the presentation of Ag by cultured human ECs has been shown to stimulate delayed transendothelial migration (TEM) of resting CD4+ memory T cells over a period of 18–24 h (14).
ECs have a special relationship with memory T cells, displaying adhesion molecules such as ICAM-1 (CD54) and VCAM-1 (CD104) that effectively engage counter-receptors up-regulated on memory subsets, i.e., LFA-1 (CD11a/CD18) or VLA-4 (CD49d/CD29), respectively (15). We have recently shown that monolayers of cultured human ECs that express elevated levels of ICAM-1 and/or VCAM-1, either as a result of TNF-mediated activation or by means of retroviral transduction, are able to capture memory T cells in a flow chamber and, in the combined presence of the inflammatory chemokine IFN-γ-inducible protein-10 (IP-10) (CXCL10) and venular levels of shear stress (1 dyne/cm2), cause rapid TEM of effector memory (EM) CD4+ T cells (defined as CD45RO+ but CCR7low and L-selectinlow) (16). Neither naive T cells nor central memory T cells rapidly transmigrate in response to IP-10 despite the expression of CXCR3 on a substantial portion of the latter cell population (16, 17). However, both central memory and EM cells do respond to the homeostatic chemokine stromal cell-derived factor (SDF)-1α, indicating that central memory cells can transmigrate (18). In the present study we examine the effect of TCR engagement and signaling on the capture and TEM of CD4+ memory T cells in the presence and absence of chemokines. Contrary to expectation, we find that TCR signaling, acting through a Src family kinase, causes memory CD4+ T cell arrest on the EC apical surface and prevents the TEM response to IP-10.
Materials and Methods
Cells
HUVECs were isolated and cultured as described (1619). To construct the retroviral vector containing CD32, the extracellular and transmembrane fragment of wild-type FcγR type IIB (aa 1–242 of 311) was isolated from cDNA extracted from human PBMCs by PCR using the primers GGACAGTGCTGGGATGACTATGGA (forward) and GCTATTACCTGCAGTAGATCAAGG (reverse). The translation start (ATG) and stop (TGA) codons are underlined, and the reverse primer contained two inserted stop codons. The 760-bp cDNA insertion was subcloned into the pCDNA vector using the Topo TA cloning kit (Invitrogen Life Technologies). The correct orientation and the DNA sequence of the insertion were verified by sequencing. The DNA insertion was excised with HindIII and NotI and subcloned into the LZRSpBMNZ retroviral vector. The expression of CD32 or MHC class II in transduced HUVECs were determined by FACS analysis (FACSort; BD Biosciences) after immunostaining with FITC-conjugated mAbs to CD32 (BD Biosciences) or HLA-DR (Immunotech) or with isotype-matched control Abs.
CD4+ T cells were isolated by positive selection with magnetic beads and released with DetachaBead (Dynal Biotech) from PBMCs prepared by a Ficoll gradient of blood collected from healthy donors. Memory (CD4+CD45RA−) T cells were isolated by depletion of CD45RA+ cells from CD4+10) for 30 min before the flow assay.
t-butyl)pyrazolo[3,4-d]pyrimidine (PP2; Calbiochem), cyclosporine, or vehicle (DMSO) for 30 min before the flow assay. For the costimulator blocking experiments, ECs were incubated in the presence of blocking Abs to 41BB ligand (41BBL), ICOS ligand (ICOSL), LFA-3, OX40 ligand (OX40L), or control IgG as described (Transendothelial migration assays under conditions of venular shear stress
+CD45RA− or CD4+CD45RA CCR7low T cells (106 cells/500 μl) suspended in the same medium were loaded onto the HUVEC monolayer at 0.75 dyne/cm2 for 2 min, followed by medium only at 1 dyne/cm2 for 15, 30, 45, or 60 min. Samples were then fixed with 2% formaldehyde in PBS, stained with Abs as appropriate (anti-Vβ2 mAb (Immunotech) followed by Alexa Fluor 594- or 488-conjugated donkey anti-mouse IgG (Molecular Probes) for CIITA HUVEC/TSST-1 samples; anti-NFAT-1 mAb (BD Bioscences) precomplexed with Alexa Fluor 488- or 647-conjugated anti-mouse IgG Fab (Molecular Probes); anti-phospho-c-Jun rabbit polyclonal Ab (Cell Signaling) followed by Alexa Fluor 594-conjugated donkey anti-rabbit IgG; or PE-conjugated anti-CXCR3 (BD Biosciences)), mounted on slides using mounting medium containing 4′,6′-diamidino-2-phenylindole (DAPI; Molecular Probes), and examined by microscopy. A FITC filter was used to detect calcein-labeled or Alexa Fluor 488-stained cells, a TRITC filter to detect Alexa Fluor 594-stained cells, a Cy5 filter to detect Alexa Fluor-647-stained cells, and a DAPI filter to detect nuclei. Phase contrast optics were used to determine whether CD4+ T cells were either on top or underneath the HUVEC monolayer. The total number of captured CD4+ T cells (i.e., T cells adhering to the apical EC surface and the transmigrated T cells) and the number of transmigrated CD4+ T cells were counted in each ×40 (original magnification) field; 10 fields were analyzed for each sample. To measure capture of CD4+ T cells on CD32 HUVEC preloaded with IgG or OKT3, calcein-labeled cells in six ×10 (original magnification) fields (covering almost all the sample) were counted per sample.
To measure CXCR3 expression, one million memory CD4+ cells were added to TNF-treated IgG- or OKT3-preloaded CD32 HUVECs (grown in 6-well dishes) for 15 min, the unbound CD4+ cells were removed, and samples were trypsinized and stained with CXCR3-FITC and CD3-PE and analyzed by FACS. Alternatively, cells bound to TNF-activated CIITA HUVECs with or without TSST-1 were stained with Vβ2-FITC, CXCR3-PE, and CD4-PC5 and then analyzed by FACS.
Statistics
For experiments in which more than two groups were compared, statistical significance was determined by one-way ANOVA using a 95% confidence interval and the Tukey posttest (Prism 4.0 for Macintosh). Statistical error is expressed as SEM. For experiments in which two groups were compared, a t test was used.
Results
We have previously observed that EM CD4+ T cells will adhere to ECs expressing ICAM-1 or VCAM-1 and transmigrate in response to surface-bound IP-10 under conditions of venular shear stress (16). In this study we used this model to examine the consequences of EC Ag presentation. Human EC expressing HLA-DR, e.g., in response to IFN-γ treatment or CIITA transduction, will stimulate cytokine secretion and proliferation of alloreactive CD4+ memory T cells (10). However, these cells are present at too low a frequency in peripheral blood to detect using the flow system, and there is no simple method for their identification and enrichment except by activation. Therefore, we developed two new models that provide polyclonal TCR signaling by ECs in human T cell populations. In the first model, HUVECs were retrovirally transduced to express CD32 (CD32 HUVECs) (Fig. 1⇓A), an FcγR that binds and displays IgG Abs, and then preloaded the transduced cells with the anti-CD3 mAb OKT3. In a second model, HUVECs were engineered to express MHC class II by retroviral transduction of CIITA (CIITA HUVEC) (Fig. 1⇓A), and then preloaded with the class II-binding superantigen TSST-1. Although OKT3 mAb would be expected to bind to every TCR, the superantigen TSST-1 binds specifically to T cells expressing a TCR encoded using the Vβ2 gene segment, which constitutes 5–10% of peripheral blood T cells in healthy adults (20). Both models result in measurable activation of CD4+ memory T cells in our flow system, as evidenced by rapid, signal-dependent translocation of NFAT-1 to the nucleus (Fig. 1⇓, B and C). NFAT-1 translocation was unaffected by the presence of a chemokine. Similar results were seen for activation of AP-1, measured by the presence of phospho-c-Jun in the nucleus (Fig. 1⇓D).
Models of polyclonal TCR signaling by ECs. A, Surface expression on HUVECs of molecules capable of displaying polyclonal activators. HUVECs engineered by retroviral transduction of CD32 (left panel) or CIITA (right panel) were stained with CD32-FITC and HLA-DR-FITC, respectively (thick lines), isotype controls (thin lines), or left unstained (dotted line). B, Immunofluorescence microscopy of lymphocytes after 15 min of shear stress on CD32 HUVECs preloaded with IgG (top panels) or anti-CD3 mAb clone OKT3 (bottom panels) stained with anti-mouse IgG Fab Alexa Fluor 488 precomplexed with anti-NFAT-1 mAb and mounted in DAPI-containing medium. The arrow indicates a T cell in which NFAT-1 has translocated into the nucleus. C, Immunofluorescence microscopy of lymphocytes after 15 min of shear stress on CIITA HUVECs with TSST-1 and IP-10 that were stained with Vβ2 TCR mAb and donkey anti-mouse IgG Alexa Fluor 594 followed by anti-mouse IgG Fab Alexa Fluor 488 precomplexed with anti-NFAT-1 mAb and mounted in DAPI-containing medium. The arrow indicates a Vβ2 TCR positive cell in which NFAT-1 has translocated into the nucleus. D, Activation of AP-1 is seen in TCR-engaged EM CD4 lymphocytes. Shown is immunofluorescence microscopy of lymphocytes after 30 min of shear stress on CIITA HUVECs with TSST-1 (top panel) and IP-10 (bottom panel) that were stained with mAb to Vβ2 followed by DAM-488 and rabbit polyclonal to phospho-c-Jun followed by DAR-594 and anti-mouse IgG Fab Alexa Fluor 647 precomplexed with anti-NFAT-1 mAb and mounted in DAPI-containing medium. Arrows point to T cell nuclei staining by phospho-c-Jun.
In TEM assays under conditions of venular shear stress, the display of OKT3 by TNF-activated CD32-HUVECs blocked IP-10-driven transmigration of memory CD4+ cells (Fig. 2⇓A) but increased the number of cells captured consistent with an effect of TCR signaling in integrin avidity (21, 22) (Fig. 3⇓A); TNF treatment, but neither OKT3 nor IP-10, was necessary for the binding of T cells to CD32 HUVEC under conditions of venular shear stress, i.e., the EC monolayers failed to capture any T cells in the absence of TNF treatment (not shown). Similarly, display of TSST-1 by TNF-activated CIITA HUVECs increased the capture (Fig. 3⇓B) and blocked the IP-10-driven transmigration of Vβ2 TCR-positive memory CD4 cells but had no effect on Vβ2 TCR-negative cells over a range of IP-10 concentrations and for up to 1 h of flow (Fig. 2⇓, B, E, and F); TNF activation, but neither TSST-1 nor IP-10, was also required for T cell binding under conditions of shear stress in these assays (not shown). Some T cell preparations showed a measurable level of chemokine-independent transmigration. Interestingly, TSST-1 signals also reduced the TEM of Vβ2 TCR-positive cells below that of Vβ2 TCR-negative cells in the absence of IP-10 (Fig. 2⇓E).
TCR engagement inhibits the TEM of memory CD4+ lymphocytes under conditions of shear stress. A, TEM assay of memory CD4+ T cells with TNF-activated CD32 HUVECs incubated with the anti-human CD3 mAb clone OKT3 or isotype controls either overlaid or not with IP-10 (3 μg/ml). The percentages of transmigrated cells are shown, derived from the combined data of three separate experiments, 10 fields per sample per experiment. B, TNF-activated CIITA HUVECs were incubated or not with TSST-1 and processed as in A except that after fixation the samples were stained with anti-Vβ2 TCR mAb and donkey anti-mouse IgG Alexa Fluor 594 and the number of Vβ2 TCR-negative cells (upper graph) and Vβ2 TCR-positive cells (lower graph), transmigrated or not, were counted per field, Ten fields per sample. The combined data of three separate experiments are presented. For the Vβ2 TCR-negative cells, the percentage of TEM was calculated for each field (p > 0.05 for comparison of nontreated (n.t.) vs TSST alone and IP-10 alone vs TSST plus IP-10 (TSST, IP-10); p < 0.001 for all other comparisons). For the Vβ2 TCR-positive cells, the total number from 10 fields was used to calculate the percentage of TEM (p < 0.001 for comparison of IP-10 to all other conditions, and p > 0.05 for all other comparisons). C, TNF-activated CIITA HUVECs were incubated with TSST-1, overlaid or not with SDF-1α, and processed as in B. The percentage of migration was calculated from the total of 10 fields per sample. Data (mean percentage of TEM ± SEM) from five separate experiments are shown (Vβ2-negative vs Vβ2-negative plus SDF (Vβ2−, SDF), p < 0.01; Vβ2-negative plus SDF (Vβ2−, SDF) vs Vβ2-positive alone and Vβ2-negative plus SDF vs Vβ2-positive plus SDF (Vβ2+, SDF), p < 0.001; p > 0.05 for all other comparisons). D, CIITA HUVECs treated with TNF, VEGF (20 ng/ml) and IFN-γ (1000 U/ml) for 20 h were overlaid with TSST-1 and processed as described above. Percentages of TEM were calculated for groups of 20 Vβ2 TCR-positive and -negative cells (mean ± SEM of four groups per sample presented; p value of t test = 0.0005). E, TCR-mediated inhibition of TEM is independent of the dose of IP-10. Shown is a TEM assay of EM CD4+ cells on TNF-activated CIITA HUVECs incubated with TSST-1 and overlaid or not with IP-10 at the doses indicated. The percentages of TEM were calculated for groups of 20 Vβ2 TCR-positive and -negative cells (means ± SEM of four groups per sample are presented). F, TCR-mediated inhibition of TEM persists for 60 min. Thirty- and 60-minute TEM assays of EM CD4+ cells on TNF-activated CIITA HUVECs incubated with TSST-1 and overlaid or not with IP-10 or SDF. The percentages of TEM were calculated for groups of 20 Vβ2 TCR-positive and -negative cells. Means ± SEM of five groups per sample are presented (p > 0.05 for comparison of all Vβ2-positive groups to each other; p < 0.001 for comparison of Vβ2-negative, SDF or Vβ2-negative, IP-10 at 30 and 60 min vs all Vβ2-positive, Vβ2-negative, and nontreated (n.t.) groups except for Vβ2-negative nontreated at 60 min vs Vβ2-negative IP-10 at 60 min (p < 0.01).
Ag presentation by EC increases the number of Ag-specific T cells captured under conditions of shear stress. A, CD4 memory cells captured by CD32 HUVECs preloaded with IgG or OKT3. The number of cells per ×10 (original magnification) field were counted, six fields per sample. Graph shows combined mean and SEM from triplicates. B, CIITA HUVECs preloaded or not with TSST-1. The percentages of Vβ2 TCR positive cells per field were calculated. Shown are combined data from three separate experiments.
The effects of TCR signals were not specific for IP-10 responses. Similar results were seen for TNF-treated CIITA HUVECs preloaded with TSST-1 overlaid with SDF-1α (Fig. 2⇑, C and F) as well as for IFN-γ-, VEGF-, and TNF-treated (treatments that induce the expression of several CXCR3 ligands, including IP-10, by HUVEC (23)) CIITA HUVECs preloaded with TSST-1 (Fig. 2⇑D).
The inhibition of chemokine-dependent transmigration observed in both models suggests that TCR signals were responsible for inhibition of the response to IP-10. To explore this possibility further, we treated lymphocytes with PP2, a Src family kinase inhibitor known to block TCR signaling in human lymphocytes (24). In these assays, PP2 blocked NFAT-1 translocation to the nucleus in Vβ2 TCR-positive cells and in cells on OKT3-preloaded CD32 HUVECs (Fig. 4⇓A–C, lower panels). We also found that, while PP2 caused a nonspecific reduction of transmigration, it released Vβ2 TCR-positive CD4+ T cells from the inhibition caused by TCR signals from TSST-1-preloaded CIITA HUVECs (as well as VEGF and IFN-γ treated HUVECs), allowing a response to IP-10 as well as endogenous EC chemokines (Fig. 4⇓, A and B). PP2 also released the inhibition of transmigration in response to IP-10 on OKT3-preloaded CD32 HUVECs (Fig. 4⇓C). We also blocked TCR-mediated NFAT-1 translocation using cyclosporine. In this instance, we did not reverse the antagonism of TEM caused by TCR engagement (Fig. 5⇓), suggesting that the block in transmigration is not dependent on calcineurin, the target of cyclosporine. Finally, we examined whether costimulator blockade would relieve the TCR-based inhibition of TEM. We observed no effect of blocking Abs to 41BBL, ICOSL, LFA-3, and OX40L (Fig. 6⇓), reagents that reduce T cell activation by ECs (10).
Effects of TCR signaling inhibitor on TCR-mediated inhibition of chemokine-driven TEM under shear stress. A, CD4+CD45RA− lymphocytes were either treated with PP2 or vehicle and used in the 15-min TEM assay with TNF-activated CIITA HUVECs incubated with TSST-1 and overlaid or not with IP-10 and processed as in Fig. 2⇑B. The ratio of the mean percentage of transmigration in response to IP-10 compared with the absence of IP-10 was calculated. Shown are the mean and SEM of these ratios from four separate experiments (upper panel; ∗, p < 0.05). The lower panel shows immunofluorescence microscopy of PP2-treated sample stained as in Fig. 1⇑C. B, PP2-treated EM CD4 lymphocytes were used in the 15-min TEM assay with TNF-, IFN-γ-, and VEGF-treated CIITA HUVECs as in Fig. 2⇑D. The graph shows mean ± SEM from the percentage of cells transmigrated from five groups of 20 cells each analyzed (p = 0.81). Right panels show immunofluorescence microscopy of sample stained as in Fig. 1⇑A. C, PP2-treated EM CD4 lymphocytes were used in the 15-min TEM assay with TNF-treated CD32 HUVECs preloaded with control mAb or OKT3 mAb and overlaid or not with IP-10. Percentages of TEM were calculated for 100 cells per sample. In the upper panel, mean ± SEM values from three separate experiments are shown (p > 0.05 for IgG alone vs OKT3 alone and IgG plus IP-10 (IgG, IP10) vs OKT3 plus IP-10 (OKT3, IP10); all other comparisons are p < 0.001). In the lower panel immunofluorescence microscopy of PP2-treated sample stained as in Fig. 1⇑B is shown.
Inhibition of TEM by TCR signaling is upstream of calcineurin. Memory CD4+ lymphocytes were treated with vehicle or cyclosporine and used in the 15-min TEM assay with TNF-activated CIITA HUVECs incubated with TSST-1 and overlaid or not with IP-10 and processed as in Fig. 2⇑B. Upper panel shows mean ± SEM values of data from two separate experiments (p < 0.01 for comparison of Vβ2-cells, treated or not with cyclosporine (CsA), responding to IP-10 (columns 2 and 4) vs their controls (columns 1 and 3, counting from the left) or to the Vβ2-positive cells (not) responding to IP10 (columns 6 and 8), whether treated with cyclosporine or not (p > 0.05 for comparison of all Vβ2-positive groups to each other. Lower panel shows immunofluorescence microscopy of cyclosporine-treated sample stained as in Fig. 1⇑C.
EC costimulatory molecules 41BBL, ICOSL, LFA-3, and OX40L are not involved in the TCR signaling that inhibits TEM. Memory CD4+ lymphocytes were used in the 15-min TEM assay with TNF-activated CIITA HUVECs incubated with TSST-1 and control IgG or blocking Abs to 41BBL, ICOSL, LFA-3, and OX40L and overlaid or not with IP-10 and processed as in Fig. 2⇑B.
To examine whether the receptor for IP-10 was down-regulated in response to TCR engagement, we stained CD4+ cells on TNF-treated CD32 HUVECs preloaded with IgG or OKT3, as well as TNF-treated CIITA HUVEC with or without TSST-1, for CXCR3. CXCR3 was not down-regulated on cells after 15 min of polyclonal stimulation (Fig. 7⇓).
CXCR3 is not down-regulated on TCR-engaged memory CD4+ lymphocytes. A, Memory CD4+ lymphocytes were incubated for 15 min with TNF-activated CD32 HUVEC preloaded with IgG or OKT3. Attached cells were collected and stained for CXCR3. B, Memory CD4+ lymphocytes were incubated for 15 min with TNF-activated CIITA HUVECs preloaded or not with TSST-1. Attached cells were stained for Vβ2 and CXCR3.
Discussion
The results presented here of TCR-mediated inhibition of TEM seem to contradict some prior reports. Specifically, in long-term (multihour) assays, resting human CD4+ memory T cells specific for HLA-DR alloantigen were found to migrate across IFN-γ-treated HUVECs at 2–4 times higher frequency than that of CD4+ memory T cells specific for a third party alloantigen (14). However, transmigration of activated alloantigen-specific CD4+ T cell clones across MHC class II-expressing ECs was either enhanced or blocked, depending on whether the T cell clones were B7-dependent or -independent in proliferation assays, respectively, suggesting a dependence on signal strength (14, 25). Moreover, these studies were performed in the absence of shear stress and in the absence of up-regulated EC adhesion molecules by TNF. A final important difference is that we studied the effect of TCR signals on the rapid (15 min) chemokine-dependent TEM of resting CD4+ memory T cells across TNF-activated ECs under conditions of venular shear stress, not the slower response seen in static systems. Collectively, these observations suggest the hypothesis that TCR signals and chemokine signals initiate distinct sets of T cell responses, i.e., an Ag-dependent adhesive process that may occur in the absence of chemokine and an Ag-independent recruitment of effector T cells that requires cytokine-induced EC activation as well as chemokine display.
Mouse ECs are not capable of presenting class II to CD4+ T cells but can present class I MHC to CD8+ T cells (26), and studies in mice have also provided in vitro and in vivo evidence that Ag presentation by ECs increases the recruitment of specific CD8+ T cells into tissue. Using an activated insulin-specific mouse CD8+ T cell clone, Ag presentation by pancreatic ECs was shown to be essential for homing to the pancreatic islets, and adhesion under shear stress was increased to aortic ECs presenting Ag in vitro (27). The lack of a requirement for cytokine-induced expression of adhesion molecules in this latter system as opposed to that of HUVECs may be that cultured mouse aortic ECs constitutively express substantial levels of VCAM-1 in vitro (28). In a mouse model using activated HY-specific polyclonal CD8 cells, however, adhesion was no different either in static in vitro assays or in vivo. In the in vivo assays, cognate recognition of ECs specifically enhanced diapedesis of these CD8+ T cells across IFN-γ-treated cremaster muscle venules within 20 min (29). It is currently unclear which of the assorted possible explanations for the discrepant results between a model using activated mouse CD8+ T cell clones and lines in vivo and one using resting human CD4+ memory cells under shear stress in vitro is valid. However, these studies do reinforce the view that TCR signals are relevant for the recruitment of Ag-specific T cells to peripheral sites of inflammation in rodents as well as humans, although rodent ECs typically express only MHC class I molecules and selectively activate only CD8+ T cells (26, 30). Unlike mouse ECs, human ECs do present Ags to EM CD4+ T cells. It may be that in humans Ag presentation by EC MHC class II to CD4+ cells may increase their specific binding and, possibly, their subsequent transmigration through a specific route.
In summary, the engagement of TCRs with Ag presented by TNF-activated ECs under conditions of shear stress elicits the translocation of NFAT-1 to the nucleus within 15 min, indicating that TCR signaling is activated in these cells. The concomitant increase in phospho-c-Jun in the nucleus suggests that the T cells are being activated rather than anergized. The increase in the number of Vβ2 TCR-positive cells on CIITA HUVECs with TSST-1 and on CD32 HUVECs with OKT3 indicates that TCR signaling could positively influence the recruitment of Ag-specific cells, at least in the sense that more are captured. However, this signaling does not substitute for or augment diapedetic chemokine signals. On the contrary, we have found that TCR signaling through Src family tyrosine kinases is capable of antagonizing rapid chemokine-mediated TEM of resting memory CD4+ T cells. Nevertheless, it is possible that Ag-specific T cells may, at later times, transmigrate by a chemokine-independent mechanism. Preliminary observations suggest that TCR engagement initiates the protrusion of a specialized structure from the captured T cells, and we are currently investigating the nature and function of this structure. Because the TCR signaling induced by the “Ag-presenting” EC models in this study was capable of quickly activating T cell transcription factors, one could speculate that this TCR signaling would have phenotypic and functional consequences on the recruited T cells and would add to the list of such changes occurring in lymphocytes after TEM (31).
Acknowledgments
We thank William Sessa for use of the immunofluorescence microscope; Nicholas Torpey for the CIITA retrovirus producing cells; Martin Kluger for the use of the flow chamber apparatus; and Louise Benson, Gwendoline Davis, and Lisa Gras for excellent assistance in cell culture.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵1 This work is funded by National Institute of Health Grant P01-HL070295 (to J.S.P.).
↵2 Address correspondence and reprint requests to Dr. Jordan S. Pober, Yale University School of Medicine, Boyer Center for Molecular Medicine, 295 Congress Avenue, New Haven, CT 06520. E-mail address: jordan.pober{at}yale.edu
↵3 Abbreviations used in this paper: EC, endothelial cell; 41BBL, 41BB ligand; DAPI, 4′,6′-diamidino-2-phenylindole; EM, effector memory; ICOSL, ICOS ligand; IP-10, IFN-γ-inducible protein-10; OX40L, OX40 ligand; PP2, 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine; SDF, stromal cell-derived factor; TEM, transendothelial migration; TSST-1, toxic shock syndrome toxin-1.
- Received June 12, 2006.
- Accepted December 21, 2006.
- Copyright © 2007 by The American Association of Immunologists