Divergent TCR-Initiated Calcium Signals Govern Recruitment vs. Activation of Human Alloreactive Effector Memory T cells by Endothelial Cells

Abstract

Early human allograft rejection can be initiated when circulating human host vs. graft antigen-specific CD8 and CD4 effector memory T cells (T_EM) directly recognize MHC Class I and II, respectively, expressed on the luminal surface by endothelium lining graft blood vessels. TCR engagement triggers both graft entry (TCR-driven transendothelial migration or TEM) and production of pro-inflammatory cytokines. Both TCR-driven TEM and cytokine expression are known to depend on T cell enzymes, myosin light chain kinase and calcineurin, respectively, that are activated by cytoplasmic calcium and calmodulin, but whether the sources of calcium that control these enzymes are the same or different is unknown. Using superantigen or anti-CD3 antibody presented by cultured human dermal microvascular cells to freshly isolated peripheral blood human T_EM under conditions of flow, models of alloantigen recognition in a vascularized graft, we tested the effects of pharmacological inhibitors of TCR-activated calcium signaling pathways on TCR-driven TEM and cytokine expression. We report that extracellular calcium entry via CRAC channels is the dominant contributor to cytokine expression, but paradoxically these same inhibitors potentiate TEM. Instead, calcium entry via TRPV1, L-Type Ca_v, and pannexin-1/P2X receptors appear to control TCR-driven TEM. These data reveal new therapeutic targets for immunosuppression.

Introduction

The initial recruitment of T cells to vascularized allogeneic organ transplants and the subsequent activation of T cell effector functions within the graft, such as cytokine synthesis and release, are both triggered by engagement of the T cell receptor for antigen (TCR) by non-self (allogeneic) antigens. At early times, this primarily involves direct recognition of intact non-self major histocompatibility complex (MHC) molecules with various bound peptides expressed on the surface of graft cells. In humans, microvascular endothelial cells (ECs) basally express both class I and II MHC molecules (MHC I and MHC II, respectively) in vivo, so that direct recognition of these molecules by host T cells may occur on the luminal surface of graft ECs. ECs are especially effective at mediating activation of effector memory T cells (T_EM) and, in humans, alloantigen presentation to circulating alloreactive T_EM is sufficient to trigger both T_EM recruitment and activation, initiating rejection without a need for priming additional alloreactive naïve T cells in secondary lymphoid organs or a requirement for graft-derived professional antigen presenting cells (“passenger leukocytes”) (1–5).

TCR-driven processes in humans differ from those that occur in commonly used rodent transplant models that typically lack alloreactive T_EM and in which ECs typically lack MHC II. Moreover, murine ECs, unlike human ECs, express B7 family co-stimulators, allowing activation of naïve CD8 T cells. To understand the process that occurs in humans, we have developed novel in vitro models (6, 7). T cell activation depends upon antigen-independent costimulation and cytokines or chemokines in addition to TCR engagement and these signals may vary depending upon the characteristics of the cell that is presenting the antigen as well as the local microenvironment. In our model, we use cultured human dermal microvascular ECs (HDMECs) placed in a flow chamber that provides levels of shear stress comparable to that occurring in post-capillary venules and we use peripheral blood T cells freshly isolated from healthy adult volunteer donors. We use HDMECs for these experiments as T cell recruitment typically takes place in the microvasculature and HDMECs provide a better model for such events than do large vessel ECs such as HUVECs (8). Since neither human naïve T cells nor central memory T cells respond to antigens on ECs by undergoing transendothelial migration (TEM), analyses of this process must focus upon T_EM. However, T_EM comprise only about 10% of circulating T cells and the signal is obscured when unfractionated T cells are used. To address this issue, we isolate these subsets from the PBMC fraction by negative selection. In adult humans, graft cells from any particular donor are typically recognized by fewer than 0.1% and 0.5% of allogeneic CD8 and CD4 T_EM, respectively. While this is much larger than the frequency of T_EM that recognize any particular microbe, it is generally too low to detect and study primary human T cell responses to alloantigens without expansion in culture, a process that alters T cell behaviors. Therefore, to model EC presentation of alloantigen to T_EM, we increase the percentage of cells that are activated through their TCR by loading a monolayer of cultured human ECs with a bacterial superantigen, toxic shock syndrome toxin 1 (TSST-1), that is able to activate all CD4 and CD8 T cells that express a TCR containing a Vβ2 exon encoded segment. Vβ2⁺ cells comprise about 5–10% of the T_EM population and can be detected with a mAb so that their responses can be compared to Vβ2^- cells that cannot recognize TSST-1 in the same assay. Presentation of TSST-1 to T_EM by an antigen presenting cell depends upon binding of the superantigen to MHC II, the expression of which is lost when human ECs are cultured. Therefore, we transduce cultured HDMECs to express CIITA and this is sufficient to re-induce expression of MHC II to levels comparable to that observed in vivo. Finally, the HDMEC monolayers in our model are pre-treated with TNF to induce synthesis of adhesion molecules and inflammatory chemokines that allow T_EM to firmly adhere to the HDMECs and undergo TEM independently of antigen recognition. TNF treatment also increases expression of certain co-stimulators on the HDMECs that boost TCR-mediated cytokine production. In the absence of TSST-1, both Vβ2⁺ and Vβ2^- T_EM will undergo rapid TEM in response to the inflammatory chemokines induced by the TNF-pretreatment of the HDMECs. Since the Vβ2^- population does not recognize the superantigen, we can analyze TCR and chemokine receptor (CR)-driven TEM in the same microscopic field. As an alternative, we have transduced HDMECs with CD32, an F_cγ receptor, and then load the EC with an anti-CD3 monoclonal antibody. This approach activates essentially all of the T cells through their TCR and can be used to validate results for TCR signals produced by superantigen, but does not allow comparison between TCR- and CR-driven responses.

In our previous investigations, we found that both TCR- and CR-dependent TEM require shear stress to impel bound T cells. This is delivered by flowing the medium over the HDMEC monolayer, generating force levels comparable to those that occur in the microvasculature (7). We have previously reported that chemokine receptor (CR)-driven and TCR-driven TEM do differ by several notable features. The first is kinetic: in the presence of venular levels of shear stress provided by flowing medium, T_EM responding to chemokine transmigrate faster, completing TEM within 10–15 minutes, while TCR-driven TEM takes 30–60 minutes. These data imply that blocking a TCR signal could actually accelerate TEM of Vβ2⁺ T_EM. The second difference is morphological. CR-driven T_EM spread and become flattened and establish a polarity consisting of a leading edge and a trailing uropod that contains cytolytic granules, mitochondria and the microtubule organizing center (MTOC). In contrast, T_EM undergoing TCR-driven TEM round up and quickly adopt an alternative polarity, with their MTOC repositioning between the T cell nucleus and EC apical surface and translocating mitochondria and granules to the same region. Over the next 15–30 minutes, TCR-triggered T_EM form a nuclear-free protrusion that extends through the EC monolayer at or near the intercellular junctions between adjacent ECs, eventually to be followed first by the MTOC and then the nucleus, i.e., as if the T cell were going in reverse compared to CR-driven chemotaxis or TEM. CD8 T_EM and CD4 T_EM TCR-TEM display similar morphological manifestations, but actually proceed through distinct pathways that utilize different cell surface receptors and either do (CD4) or do not (CD8) require granule exocytosis (9, 10). However, both CD4 and CD8 TCR-TEM require myosin activity mediated by calcium/calmodulin activation of myosin light chain kinase.

Once activated through their TCR, T_EM rapidly acquire and display effector functions, such as cytokine synthesis and release, detectable within a few hours. In the case of an allograft, this would allow alloreactive T_EM that had been recruited into the graft in response to a TCR signal to initiate the inflammatory response that underlies rejection. Since TCR signaling to activate cytokine expression is typically understood to occur through the Lck-ZAP70-LAT-Itk-Phospholipase C-γ (PLCγ)-IP₃R-STIM-CRAC-channel-calcium/calmodulin-calcineurin-NFAT-pathway (11), calcium/calmodulin provides a potential common node where TCR signaling may influence both TCR-TEM and cytokine expression. However, other TCR-activated calcium signals have been described, such as alternative second messengers NAADP and cADPR that bind to Ryanodine receptors on the ER that may release calcium stores and activate CRAC channels, or Two Pore channels on acidic lysosomes (which are prevalent in T_EM) which may provide another source of internal calcium mobilization, as well as other plasma membrane TCR-activated calcium channels, such as TRPV1, L-type calcium channels, and pannexin-1/P2X receptors (12–19). The goal of the present study was to determine if the calcium signals activated by TCR engagement that trigger TEM are the same or different from those used to initiate the synthesis of cytokines.

Our model allows us to detect three distinct and sequential steps in Vβ2⁺ CD4 and CD8 T_EM following engagement of the TCR by superantigen: (a) relocation of the MTOC to the region of the cell between the nucleus and the region of contact with the EC surface, detected by confocal immunofluorescence microscopy; (b). transmigration of the T cell between cells of the EC monolayer, resulting in the image of the cell changing from light to dark when imaged by phase contrast microscopy; and (c) synthesis of cytokines, notably IL-2 and IFN-γ, assessed by qRT-PCR for mRNA. As our model makes use of freshly isolated populations of circulating human T cells, we cannot use genetic manipulation and the principal tool used for dissecting these steps are pharmacological inhibitors. Here we report the surprising result that transmigration and cytokine synthesis appear to use different sources of calcium.

Materials and Methods

Cells and reagents

All human materials were obtained from de-identified blood or tissue donors under protocols approved by the Yale Human Investigation Committee. CIITA- and CD32-transduced human dermal microvascular ECs (CIITA and CD32 HDMECs, respectively) were generated using a retroviral vector and characterized as described (7). Prior to flow experiments, CIITA and CD32 HDMECs were incubated in the presence of 10 ng/ml recombinant human TNF (rhTNFα, R&D Systems) for 20–24 hours to upregulate adhesion molecules, and with 100 ng/ml recombinant TSST-1 (R-TT 606, Toxin Technology, Inc.) for 30 minutes on CIITA HDMECs to allow TCR-mediated activation of all T cells utilizing Vβ2 gene segment to form their TCR, approximately 5–10% of the circulating T cells in most donors. For siRNA knockdown of PECAM-1 (CD31), cells were transfected 72 h prior to flow with 10 nM siRNA (PECAM-1_1, Qiagen,) as described (20) (21).

Leukapheresis was performed on healthy volunteer adult donors and PBMCs were enriched by Ficoll-Hypaque density gradient centrifugation prior to cryopreservation of aliquotted cells. Total peripheral blood CD4 and CD8 T cells were isolated from the cryopreserved samples by positive selection with CD4 or CD8 Dynabeads magnetic beads and released with Detachabead (Dynal) according to the manufacturer’s protocol. CD4 and CD8 T_EM were enriched by depletion of naïve and central memory cells with anti-CCR7 mAb (BioLegend) and pan-mouse IgG beads. Approximately 80–90% (CD4) or 60–70% (CD8) of the initial T cell population as well as essentially all other leukocyte types were removed by these manipulations. The remaining T cells, highly enriched for T_EM, were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2 mM glutamine, and penicillin/streptomycin overnight prior to assays. Inhibitors and treatment conditions are listed in Table I.

Table I.

Inhibitors used in this study

Target	Reagent	Supplier	treatment	in flow medium
intracellular calcium	BAPTA-AM	Cayman	3–100 μM, 1 h	no
RyR	Dantrolene	Sigma	60 μM, 30 min	yes
RyR	Ryanodine	Tocris	400 μM, 30 min	no
RyR	8-Br-7-d-cADPR	BioLogic	100 μM, 30 min	no
TPC	trans-ned-19	Tocris	100 μM, 30 min	no
TPC	tetrandrine	Sigma	1–10 μM, 30 min	yes
Itk	BMS-509744	Calbiochem	10 μM, 30 min	no
Itk	PF 06465469	Axon Med Chem	10 μM, 30 min	no
CRAC	BTP2	Calbiochem	3 μM, 30 min	yes
CRAC	pyr6	Calbiochem	3 μM, 30 min	yes
Calcineurin	FK506	Enzo	0.1 μM, 2h	no
TRPV1	SB-366791	Enzo	0.1 μM, 30 min	yes
L-Type Cav	nifedipine	Sigma	30 μM, 30 min	yes
pannexin-1	trovafloxacin	Sigma	30 μM, 30 min	yes
pannexin-1	carbenoxolone	Sigma	10 μM, 30 min	yes
P2X	PPADS	Tocris	300 μM, 30 min	yes
P2X	Evans Blue	Sigma	30 μM, 30 min	yes

TEM assays

CIITA and CD32 HDMEC were grown to confluence on 10 μg/ml human plasma fibronectin-coated 35 mm coverglasses, treated with TNF and loaded with 100 ng/ml TSST-1 (for CIITA HDMEC) or 3 μg/ml anit-CD3 mAb Hit3a (for CD32 HDMEC) as described (7), washed twice with RPMI/10% FBS, and assembled with a parallel plate flow chamber apparatus (Glycotech) using the 0.01 inch height, 5 mm wide slit gasket provided by the manufacturer. On a 37°C heating surface, CCR7^low human CD4 or CD8 T_EM (1–2 × 10⁶ cells/500 μl) suspended in the same medium were loaded onto the EC monolayer at 0.75 dyne/cm² for 2 minutes, followed by washing with medium only at 1 dyne/cm² for 5, 30 or 50 minutes. Samples were then fixed with 3.7% formaldehyde in PBS, stained with anti-Vβ2TCR mAb (Beckman Coulter), followed by Alexafluor 488 or 546-conjugated goat or donkey anti-mouse IgG, mounted on slides using mounting medium containing DAPI (Prolong Gold, Invitrogen). Alternatively, samples were stained with FITC conjugated anti-Vβ2TCR mAb (Beckman Coulter), AlexaFluor 488-conjugated rabbit anti-FITC, and AlexaFluor 488-conjugated goat anti-rabbit IgG. Samples were then examined by microscopy with a Zeiss Axiovert 200M microscope. A FITC filter was used to detect FITC or Alexafluor 488-stained cells, a TRITC filter was used to detect Alexafluor 546-stained cells, and a DAPI filter used to detect DAPI-stained nuclei. Using a 40X/0.60 korr Ph2 objective, phase contrast optics were used to determine whether T cells were either on top or underneath the HDMEC monolayer. The percentage of transmigrated T cells were calculated for 100 cells per sample by analyzing five groups of 20 cells each, calculating the percentage for each group, and calculating the mean and s.e.m. for the groups. For total adhesion, T cells in ten fields using a 10X objective were counted for each sample. For antigen-induced adhesion, the percentage of Vβ2TCR+ cells from a total of more than 200 total cells counted was calculated for each sample. This percentage was divided by the percentage of Vβ2TCR+ cells of the input as determined by FACS to obtain the fold enrichment.

Cytokine expression assay

The procedure for TEM was followed as described above except, after flow for 30 or 50 min, samples were disassembled and placed in culture for a total of 3 h. Alternatively, 200,000 T cells were incubated with TNF-activated, TSST-1 preloaded CIITA HDMEC or Hit3a anti-CD3 mAb (BioLegend) preloaded CD32 HDMEC in C24 wells for 3 h. Samples were then processed for RNA preparation (RNeasy Mini Kit, Qiagen) and cDNA (High Capacity cDNA Reverse Transcriptase Kit, Applied Biosystems). Quantitative real time PCR using TaqMan probes for human IFN-γ, IL-2, and CD3e and TaqMan Gene Expression Master Mix (Applied Biosystems) were performed in a C1000 Touch Thermal Cycler CFX96 Real-Time System (Biorad).

Confocal Microscopy

To visualize MTOC of Vβ2⁺ T_EM transmigrating on TNF-activated CIITA HDMEC plus TSST-1, samples were stained with anti-Vβ2TCR mAb (Beckman Coulter), Alexa Fluor 546-conjugated donkey anti-mouse IgG (Invitrogen), permeabilized with 0.5% Triton/PBS, re-fixed with methanol/acetone (50/50), stained with rabbit anti-γ-tubulin (Sigma), Alexa Fluor 647-conjugated donkey anti-rabbit IgG and Alexa Fluor 488-conjugated phalloidin, and mounted on slides using mounting medium containing DAPI (Prolong Gold, Invitrogen).

Images of single T cells on the EC apical surface were captured with a Leica TCS SP5 Spectral Confocal Microscope, 405UV using a 63X oil immersion objective and sequential scanning with 405 Diode, argon and He/Ne laser excitation lines of 405 nm, 488 nm, 543 nm, and 633 nm. Six Z slices were captured encompassing the entire T cell starting from the EC interface.

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 post-test (Prism 6.0 for Macintosh). Statistical error is expressed as s.e.m. For experiments in which two groups were compared, a t-test was used. P values are designated as: *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001.

Results

Proteins whose functions are regulated by changes in the level of cytosolic calcium, such as calcineurin and myosin IIa, are thought to be required both for cytokine gene transcription and for TEM, respectively. To directly confirm a role for calcium levels, we prevented cytosolic calcium from interacting with its cellular targets by preloading T cells with BAPTA-AM. In the form of BAPTA-AM, this agent freely crosses the plasma membrane where cytosolic esterases cleave the acetoxymethyl groups from carboxylic acid side chains, trapping the now charged BAPTA moiety within the cytosol. BAPTA then chelates free calcium ion in the cytosol. Using our established model of TSST-1-dependent TCR signaling in primary resting T_EM by TNF-activated untransformed CIITA-transduced HDMECs, we tested the effects of BAPTA loading on TCR-driven TEM side by side with chemokine receptor (CR)-driven TEM under venular levels of shear stress. We further compared the effects of BAPTA on TCR-dependent cytokine expression using the same signal for activation, namely presentation of TSST-1 presented by TNF-activated CIITA-transduced HDMECs to peripheral blood CD4 or CD8 T_EM. In this system, IFN-γ and IL-2 mRNAs are upregulated more than a thousand fold within 3 h. As expected, BAPTA had a wide range of effects: it completely blocked cytokine expression, reduced both CR- and TCR-driven TEM, and affected adhesion. Interestingly, adhesion of Vβ2⁺ T_EM was less affected than adhesion of Vβ2^- T_EM (Figure 1), supporting observations that there is a calcium-independent effect on TCR-driven increases in integrin avidity (22).

Figure 1.

Intracellular calcium chelator BAPTA-AM reduces both TCR-driven TEM and cytokine production. A. Cytokine expression by T_EM either not interacting with EC (input) or treated with vehicle or BAPTA-AM (BAPTA) interacting with TNF-activated, TSST-1 preloaded HDMEC. Values represent those relative to the levels of cells treated with vehicle. B. TEM of T_EM across TNF-activated, TSST-1 preloaded HDMEC. TCR and CR refer to those T_EM that do or do not interact with TSST-1, respectively. C. The number of T_EM bound in the TEM assays. D. The percentage of TSST-1-specific T_EM cells bound. Graph shows fold increase over the percentage of the parent population. Graphs display mean and s.e.m. of data from 3 separate experiments using T cells from different donors.

Having established a role for intracellular calcium signals, we proceeded to test inhibitors of receptors on internal organelles that result in rises of cytosolic calcium. As pharmacological inhibitors of IP₃R are not available, we focused on ryanodine receptors. (RyR) expressed on the ER and on two pore channels (TPC) expressed on acidic granules, including the granzyme A⁺ granules that are abundant in T_EM, and that bind to TCR-signal generated second messengers such as cADPR and NAADP. To this end, we tested inhibitors of RyR (ryanodine, dantrolene, 8-Br-7-d-cADPR) and TPC (trans-ned-19, tetrandrine) (14, 23). While none of these inhibitors reduced cytokine expression, some increased TCR-TEM, but these findings were not consistent between compounds purported to inhibit the same channels, suggestive of off-target effects that preclude firm conclusions regarding the role of these channels in TCR responses (Supplemental Figures 1,2).

Since we could not directly assess IP₃R signaling, we instead tested additional targets in the Itk/PLCγ/CRAC channel pathway. Inhibitors of Itk, the enzyme that activates PLC-γ in the TCR signaling pathway, did not affect CR-driven TEM but, unexpectedly, increased TCR-driven TEM (Figure 2, Supplemental 3A). Inhibitors of STIM/ORAI CRAC channels (activated by IP₃ binding to IP₃R on the ER) fully recapitulated the effects of inhibiting Itk, i.e., they increased TCR-driven TEM at concentrations that severely curtailed cytokine expression (Figure 3, Supplemental Figure 3B). The increase in TEM of antigen-specific T cells treated with CRAC channel and Itk inhibitors could potentially result from a reversion to CR-driven TEM. Since TCR-driven TEM, but not CR-driven TEM, is characterized by an early movement of the MTOC into a position between the T cell nucleus and EC apical surface, we examined the effect of these inhibitors on MTOC localization. In the presence of these agents, MTOC relocation to the region between the T cell nucleus and EC apical surface, i.e. the TCR-driven TEM phenotype, still occurred (Figure 4). CR and TCR-driven TEM of CD4 T_EM (but not CD8 T_EM) can be also distinguished by their requirements for interaction with EC PECAM-1; only TCR-driven TEM requires PECAM-1 engagement. TCR-TEM of CD4 T_EM treated with Itk inhibitor is blocked by EC PECAM-1 knockdown (Figure 4), supporting the interpretation that Itk and CRAC channel inhibitors actually augment TCR-driven TEM, rather than blocking this process to allow CR-driven TEM.

Figure 2.

BMS-509744 and PF06465469, inhibitors of Tek family kinase Itk inhibits cytokine expression and augments TEM of T_EM in response to antigen presented by HDMEC. T cells were treated with PF 06465469 (PF, A,C) and BMS-509744 (BMS, A) and used in either cytokine (A,) or TEM (B,C) assays. TEM of cells specific for antigen (TCR) as well as cells not activated by antigen (CR) was measured. Graphs display mean and s.e.m. of data from 4 (A), 5 (B), 5 (C, CD4), 2 (C, CD8) separate experiments using T cells from different donors.

Figure 3.

BTP2 and pyr6, inhibitors of CRAC channels, reduce cytokine expression but augment TEM of T_EM in response to antigen presented by HDMECs. T cells were treated with BTP2 (A,B) and pyr6 (C,D) and used in either cytokine (A,C) or TEM (B,D) assays. TEM of cells specific for antigen (TCR) as well as cells not activated by antigen (CR) was measured. Graphs display mean and s.e.m. of data from 5 (A,B), 3 (C), 4 (D) separate experiments using T cells from different donors.

Figure 4.

Itk and CRAC channel inhibitors augment TCR-driven TEM. A. MTOC analysis of T cells treated with BMS-509744 (BMS) and CRAC channel inhibitors (CRAC) after 5 min flow. Graphs show percent of antigen-specific T cells with nuclei between the T cell nucleus and EC apical surface (% subnuclear MTOC). N > 110 for CD4 BMS, N=40 for CD8 BMS. N > 64 for CD4, N > 44 for CD8 CRAC.

B. TEM assay of CD4 T_EM treated with vehicle or BMS-509744 (BMS) on HDMEC treated with control or PECAM-1 siRNA. Graph displays mean and s.e.m. of data from 2 separate experiments with different donors.

The Itk/PLCγ/CRAC channel pathway is thought to be upstream of calcium/calmodulin activation of calcineurin-mediated NFAT signaling, and inhibition of calcineurin by FK506 showed, as expected, reduced cytokine expression. Calcineurin may also act at the immunological synapse to affect LFA-1 binding to ICAM-1 (24), which could affect TCR-mediated TEM. However, the same concentrations of FK506 had no effect on TEM (Figure 5). These results further suggest that the unanticipated augmentation of TCR TEM by inhibitors of the Itk/PLCγ/CRAC channel pathway does not involve calcineurin.

Figure 5.

Calcineurin inhibitor FK506 affects cytokine expression but not TEM of T_EM in response to antigen presented by HDMEC. T_EM were treated with FK506 and used in either cytokine (A) or TEM (B) assays. TEM of cells specific for antigen (TCR) as well as cells not activated by antigen (CR) was measured. Graphs display mean and s.e.m. of data from 6 (A) and 3 (B) separate experiments using T cells from different donors.

Our BAPTA experiments point to a positive role for cytosolic calcium in TEM but the unexpected effects observed in the Itk and CRAC channel inhibitor experiments suggested that other TCR-activated calcium channels must be involved, possible examples being TRPV1, L-Type Ca_v and pannexin-1/P2X receptors (15–18). Interestingly, TRPV1 inhibitor SB-366791 (15) and Ca_v1.1 inhibitor nifedipine reduced TCR-driven TEM by CD4 but not CD8 T_EM and cytokine expression was not affected in either T_EM subset (Figures 6, 7). P2X receptors are calcium channels that are activated by extracellular ATP. TCR signaling activates pannexin-1 hemichannels to release ATP into the extracellular space, thereby activating P2X receptors. Trovafloxacin and carbenoxolone, inhibitors of pannexin-1 (25), reduced both CD4 and CD8 T_EM TCR-driven TEM. Trovafloxacin, but not carbenoxolone, also reduced cytokine expression, with a bias towards IL-2 (Figure 8, Supplemental Figure 4). P2X receptor inhibitors Evans Blue and PPADS reduced TCR-TEM, but not cytokine expression (Figure 8). These data support a role for this pathway in TCR-TEM but suggest that effects of trovafloxacin on cytokine production are likely due to off-target effects.

Figure 6.

SB-366791 (SB), an inhibitor of TRPV1 controls TCR-driven TEM of CD4, but not CD8, T_EM. A. T cells were treated with SB-366791 (SB) and used in either cytokine (A) or TEM (B) assays. TEM of cells specific for antigen (TCR) as well as cells not activated by antigen (CR) was measured. Graphs display mean and s.e.m. of data from at least 4 separate experiments using T cells from different donors.

Figure 7.

Nifedipine, an L-type Ca_v channel blocker, controls TCR-driven TEM of CD4, but not CD8, T_EM. A. T cells were treated with nifedipine and used in either cytokine (A) or TEM (B) assays. TEM of cells specific for antigen (TCR) as well as cells not activated by antigen (CR) was measured. Graphs display mean and s.e.m. of data from 4 separate experiments using T cells from different donors.

Figure 8.

Pannexin-1/P2X receptors regulate TCR-driven TEM and selective cytokine expression. T cells were treated with trovafloxacin (trov, A,B), carbenoxolone (CBX, C,D), PPADS and Evans Blue (EB, E,F) and used in either cytokine (A,C,E) or TEM (B,D,F) assays. TEM of cells specific for antigen (TCR) as well as cells not activated by antigen (CR) was measured. Graphs display mean and s.e.m. of data from 5 (A), 4 (B,C), 3 (D), 5 (E) and 3 (F) separate experiments using T cells from different donors.

The signal provided to T cells by superantigen is thought to be a reasonable mimetic for that provided by MHC-peptide, but some differences in the signaling responses initiated by these TCR ligands are possible. An alternative widely used antigen mimetic is cross-linking of TCR-associated CD3 molecules. We therefore tested whether inhibitors of Itk, CRAC channels, and pannexin-1 would have a similar effect on T cells activated by anti-CD3 antibody as observed with TSST-1. To this end, we utilized HDMECs engineered to express FcRII (CD32) preloaded with anti-CD3 mAb in our TEM and cytokine expression assays. In agreement with the superantigen assays, Itk and CRAC channel inhibitors strongly reduced cytokine expression but not TEM, while trovafloxacin reduced TCR-TEM and, slightly, IL-2 (Figure 9).

Figure 9.

T cells treated with Itk inhibitor (PF), CRAC channel inhibitor (pyr6) and pannexin-1 inhibitor (trov) and used in cytokine and TEM assays on CD32 HDMEC preloaded with TCR-activating mAb. Graphs display mean and s.e.m. of data from 3 separate experiments using T cells from different donors.

Discussion

The aim of this study was to compare the TCR signaling pathways that lead to TEM and to cytokine expression of both CD4 and CD8 T_EM as they respond to antigen presented by ECs, thereby potentially revealing new targets for immunosuppressive therapies. Both processes depend upon a rise in cytosolic free calcium ion but our new results suggest that the pathways used to achieve this are different. We confirmed the expected result that cytokine synthesis is dependent upon Itk activation of PLC-γ followed by opening of CRAC channels likely via IP₃ and STIM1 signaling, resulting in calcium/calmodulin activation of calcineurin. In contrast, TCR-driven TEM of CD4 T_EM appears partly dependent upon TRPV1 and L-type Ca_v channels, while both CD4 and CD8 T_EM TCR-driven TEM are controlled by pannexin-1/P2X receptors. Although results from different pannexin-1/P2X receptor inhibitors were not consistent with regard to cytokine expression, we cannot rule out that this pathway also may make a contribution to IL-2 synthesis. Nevertheless, the likeliest explanation of the difference is that calcium ion is not uniformly increased throughout the cytosol, but rather is unevenly distributed, allowing for preferential interactions with specific molecular targets. The best candidates for these differences are the known role of calcineurin in cytokine transcription and myosin light chain kinase in transmigration, both of which are responsive to calcium/calmodulin.

The strengths of the current study are that it is based on the analysis of freshly isolated blood human T_EM interacting with untransformed human microvascular ECs. The principal limitations are that the observations arise from the use of a superantigen rather than alloantigen, are restricted to pharmacological agents and are based solely on in vitro experiments. We had previously shown that direct alloantigen responses, although comparatively infrequent vs. superantigen responses, appear to be similar (10). Unfortunately, it isn’t possible to routinely analyze sufficient events based on an unprimed allogeneic response. The use of a pharmacological approach, while potentially misleading due to off target effects, actually has the advantage of pointing to potential therapeutics. Current immunosuppressive regimens used in transplantation do not target and have not been effective at reducing T cell recruitment, a process that in mice appears to be initiated by intravascular presentation of alloantigen to naïve CD8 T cells (26). Thus, our data point to new potential approaches that can complement existing regimens. The limitation of our dependence upon in vitro experiments is also potentially significant, but it is important to note that rodent T cell interactions with ECs are significantly different from human interactions. Of particular relevance here are observations that human ECs can present antigens to CD4 T_EM whereas murine ECs appear to selectively activate CD4 regulatory T cells (27). Thus mouse transplant models have largely been restricted to analysis of CD8 T cell interactions with endothelium. Since there are significant differences between human CD4 and CD8 TCR-driven TEM (10), any therapy based solely on mouse data may not translate into human clinical settings.

In summary, here we show that TCR signals in T_EM initiated by alloantigen recognition of the surface of ECs and that lead to TEM or cytokine production diverge at the level of calcium signaling. Although both depend upon rises in cytosolic free calcium ion, various TCR-activated channels have different effects, consistent with the idea that the rise in cytosolic calcium may not be uniform and can affect distinct targets depending on local concentrations. Most current immunosuppressive regimens have focused on inhibition of cytokine synthesis or cytokine responses. Trovafloxacin is particularly interesting, as it impedes both CD4 and CD8 TCR-TEM. The only other compound identified to data that blocks both CD4 and CD8 TCR-driven TEM is an ICAM-1 blocking antibody (10). However, clinical trials of ICAM-1 blocking antibody in kidney transplantation were unsuccessful, although unforeseen side effects may have contributed to this outcome (28, 29). Perhaps pannexin-1 inhibitors could be developed as an alternative to block TCR-driven TEM of CD4 and CD8 T_EM cells. In any event, the experiments presented here reveal divergence of pathways downstream of TCR signaling that control TEM and cytokine production, opening the possibility of combining such agents to improve effectiveness.

Abbreviations:

CRAC

calcium release activated calcium

EC

endothelial cell

HDMEC

human dermal microvascular endothelial cell

IP₃

Inositol tris-phosphate

MTOC

microtubule organizing center

PLC

Phospholipase C

T_EM

effector memory T cell

TEM

transendothelial migration