A Cell Surface ADP-ribosyltransferase Modulates T Cell Receptor Association and Signaling*

ART-1, a cell surface ADP-ribosyltransferase, is imbedded in the membrane by a glycosylphosphatidylinositol anchor. Function of this enzyme in mouse T lymphocytes is to transfer ADP-ribose groups from NAD to arginine residues, exposed on the extracellular domain of cell surface molecules. As a consequence, T cell responses are modulated. To explore the precise action of the enzyme, the T cell lymphoma EL-4 was transfected with the ART-1 gene, and its effects were examined. It is shown that ART-1 ADP-ribosylates distinct cell surface molecules, causing inhibition of T cell receptor signaling, concomitant to suppression of p56 lck kinase activation. These effects are explained by failure of T cell receptors and co-receptors to associate into a contiguous and functional receptor cluster.

Post-translational protein modifications constitute a central mechanism in cellular signal transmission. Although this has been long known in case of intracellular signaling pathways, it has not yet been documented for signal transmission on the outer cell membrane. That such regulatory mechanisms might exist is suggested by the presence of cell surface enzymes with specificity for peptides in the extracellular milieu. However, proof that these enzymes regulate cell surface receptor function by modifying them is still lacking. We recently described an ecto-enzyme on mouse cytotoxic T cells, named ART-1, that functions as an ADP-ribosyltransferase, attaching ADP-ribose groups to arginine residues of cell surface molecules (1). As a consequence, T cell functions such as cytolytic activity and cell proliferation are inhibited (1,2). Here we explore the function of this enzyme in a stable transfectant generated from lymphoma EL-4 and show that ADP-ribosylation of cell surface co-receptors inhibits formation of a functional T cell receptor (TCR) 1 cluster.
Fluorescent Confocal Microscopy-Cells were treated with 1 mM NAD for 3 h at 37°C and then incubated with 10 g/ml anti-CD3⑀mAb (500A2) on ice for 30 min. After washing, the cells were resuspended in PBS containing 10 g/ml rhodamine-labeled cross-linking secondary goat anti-hamster antibodies and incubated at 37°C for 15 min. After washing with ice-cold PBS containing 0.1% NaN 3 , cells were either stained with 10 g/ml FITC-labeled anti-CD45 mAb (30F11.1, PharMingen) diluted in PBS containing 0.1% NaN 3 for 30 min on ice or permeabilized with 1% saponine and stained with rabbit anti-human lck antibody (Upstate Biotechnology, Inc., Lake Placid, NY) diluted in PBS containing 0.1% NaN 3 for 30 min on ice, followed by incubation with FITC-labeled goat anti-rabbit antibody (Jackson, West Grove, PA). After fixation with 4% paraformaldehyde, cells were spread onto glass slides, mounted, and examined by confocal microscopy.

RESULTS AND DISCUSSION
To examine the function of ART-1, its gene was expressed in a mouse T cell lymphoma lacking this enzyme (6). EL-4 cells were chosen because they possess TCRs whose ligation induces IL-2 secretion. Fig. 1A shows that PMA and CIO stimulate secretion of large quantities of IL-2 in EL-4 as well as its ART-1 transfectant C17. Addition of ART-1 substrate, NAD, at concentrations as high as 1000 M exerts little effect on IL-2 secretion in either cell line. The high NAD concentration is physiologically relevant, because cell lysis caused by cytotoxic lymphocytes could result in local concentrations of this magnitude. The failure then of high NAD concentrations to inhibit cytokine secretion demonstrates that NAD is not toxic.
To examine whether ART-1 and its substrate have effects on * This work was supported by Public Health Service Grants CA37706 and AI43954. 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.
receptor-mediated cell activation, cells were incubated with anti-CD3 and anti-CD28 to induce receptor ligation and IL-2 secretion. It is seen in Fig. 1B that NAD has no effect in EL-4 cells, but there is a significant suppression of IL-2 secretion in C17 cells. To show that this effect depends on cell surface ADP-ribosyltransferase activity, C15, a transfectant expressing the ART-2/Rt6.1 gene, and therefore only one-fifteenth of cell surface enzyme activity of C17 was used: In C15 cells no suppressive effect of NAD on receptor-induced IL-2 secretion was demonstrable (data not shown). Therefore, ART-1 transfectants expressing high enzyme activity are sensitive to NADmediated regulation. The ART-1 enzyme effect likely occurs on the level of receptor-mediated cell activation, raising the question of whether the enzyme modifies CD3 and/or CD28 and thereby inhibits signal transmission. To find out, cells were incubated with 32 P-NAD, and cell lysates were analyzed for labeled proteins. It is seen that EL-4 lysates are not labeled, whereas C17 lysates contain several labeled molecules ( Fig.  2A). Using a panel of antibody for immunoprecipitation, strongly labeled molecules were identified as the CD11a and CD18 chains of LFA-1 in addition to CD44 and CD45. Weaker labeling is seen with CD27 and CD43 ( Fig. 2A). No labeling was demonstrable in other molecules expressed on EL-4 cells, i.e. CD29, CD40L, CD54, and CD62L (data not shown), nor was there labeling in CD28 (Fig. 2, A and B). Immunoprecipitation with anti-CD3⑀, which precipitates both CD3⑀ and CD3 chains and therefore the intact CD3 complex, did not reveal any labeled proteins (Fig. 2B and data not shown). Moreover, immunoprecipitations with anti-TCR␣, anti-TCR ␤, or anti-CD3 ␥,␦,⑀, using various antibodies all failed to uncover labeling in these chains (Fig. 2, A and B, and data not shown). Therefore, TCR-mediated cell activation is inhibited by ART-1 in the ap-parent absence of ADP-ribosylation of TCR/CD3 and CD28 chains.
Ligation of the CD3/TCR complex induces activation of receptor-associated protein phosphokinases, phosphorylation of phospholipase C-␥1, generation of inositol-phosphates, and mobilization of calcium (5-10). To examine whether these early events in TCR signaling are regulated by ART-1, the induction of inositol phosphates and activation of TCR-associated phosphokinases was assayed. It is seen in Fig. 1C that NAD has no effect on induced levels of inositol phosphates in EL-4 cells, but there is a very strong inhibition in C17 cells. To examine the activation of TCR-associated protein phosphokinases, autophosphorylation of p56 lck , p59 fyn , and ZAP70 was assayed (5,11,12). It is seen in Fig. 2 (C, D, and E) that in exponentially growing EL-4 and C17 cells not stimulated by receptor ligation, NAD has no effect on kinase activities, although there is some suppression of p56 lck autophosphorylation in C17 cells. When cells are activated by receptor ligation a significant increase in p56 lck and somewhat smaller increase in p59 fyn and ZAP70 autophosphorylation is seen (Fig. 2, C, D, and E). These increases are not affected by NAD in EL-4 cells. In C17 cells, however, NAD suppresses p56 lck autophosphorylation activity, induced by receptor ligation, to the level of nonstimulated controls (Fig. 2C). A similar result is seen with ZAP70 but not with p59 fyn (Fig. 2, D and E), hence ART-1 action involves an early step in receptor signaling, i.e. a reaction preceding the generation of inositol phosphates and p56 lck activation.
Circumstantial evidence suggests that activation of p56 lck is the consequence of cell surface receptor rearrangements in the contact zone between T cell and antigen presenting cell. This reorganization culminates in formation of a receptor cluster between the two cells, which serves to stabilize cell contact and facilitates contact between p56 lck and CD45, which activates p56 lck by dephosphorylation (13)(14)(15). Antibody-mediated cocapping and precipitation experiments have shown that many co-receptors are part of the TCR cluster, among those CD4, CD8, LFA-1, CD43, CD45, and CD44 (16 -18). The demonstration therefore ( Fig. 2A) that several of these molecules, i.e.

FIG. 1. NAD inhibits IL-2 secretion and generation of inositol phosphates in C17 cells induced by CD3/CD28 ligation. EL-4 and C17 cells were treated with NAD and then stimulated with PMA and CIO (A) or anti-CD3 and anti-CD28 (B).
After 20 h supernatants were collected and assayed for IL-2. C, cells were incubated with (ϩ) or without (Ϫ) NAD as indicated, then activated by anti-CD3 and anti-CD28, harvested, and assayed for inositol phosphates whose change over background levels (100%) is plotted .   FIG. 2. A and B, identification of ADP-ribosylated cell surface antigens on C17 cells. Intact cells were labeled with 32 P-NAD, and whole cell lysates were immunoprecipitated as indicated and analyzed by polyacrylamide gel electrophoresis. The precipitation of CD3⑀, CD3, and CD28 chains was verified by immunoblot (B). C-E, inhibition of kinase activation, induced by CD3 and CD28 cross-linking in NADtreated C17 cells. EL4 or C17 cells were incubated with (ϩ) or without (Ϫ) NAD as indicated, followed by immunoprecipitation of cell lysates with anti-lck (C), anti-fyn (D), or anti-ZAP70 (E) antibodies. The immunoprecipitates were assayed for kinase autophosphorylation activity. Autoradiographs are shown in the upper panels of C-E. Quantitation of kinase in precipitates was accomplished by immunoblot, shown in the lower panels of C-E. Nonactivated cells (Non-act.) were incubated in medium, and activated cells (Act.) were incubated with anti-CD3/ CD28. Increased kinase activities induced by anti-CD3/CD28 are reflected in increases in band intensity in the upper panels of the lck (C), fyn (D), and ZAP70 (E) precipitates.
CD45, LFA-1, CD43, and CD44 are ADP-ribosylated, raises the intriguing possibility that inhibition of TCR signaling is caused by interference of ADP-ribosylated co-receptors with formation of a functional signal transmitting TCR cluster.
This possibility was investigated by examining antibodyinduced cap formation on a single cell level as a model for T cell APC interaction. Cells were incubated with anti-CD3, followed by a rhodamine-conjugated second antibody and examined by fluorescent confocal microscopy. Results in Fig. 3A show that CD3 ligation in the absence of NAD induces very large caps in EL4 and C17 cells. To visualize the association of TCRs with p56 lck , permeabilized cells were reacted with anti-p56 lck and a fluorescein-conjugated secondary antibody. It is seen that p56 lck co-localizes with CD3 in the cap. Virtually identical results are seen in EL4 cells, incubated with NAD. In C17 cells, however, the presence of NAD inhibits formation of the cap, and only small aggregates of CD3 and p56 lck molecules are discernable. Therefore, in NAD-treated C17 cells there is inhibition of formation of a cap, containing CD3 and p56 lck molecules. These effects are also demonstrable, albeit less dramatically, under conditions in which receptor clustering is induced with one antibody only (Fig. 1), mediating receptor cross-linking via engagement with Fc receptors on adjacent cells.
Results in Fig. 2C had shown that NAD inhibits CD3/CD28mediated activation of the p56 lck kinase in C17 cells; therefore we predict that the interaction between p56 lck and CD45 is likely disturbed. The possibility that ADP-ribosylation of CD45 inhibits its phosphatase activity and is therefore responsible for the effect was found not to be the case (data not shown). An alternative explanation is that association of CD45 with the kinase is inhibited. It is seen in Fig. 3B that a CD3 cap that contains CD45 is formed in both EL-4 and C17 cells in the absence of NAD. However, in the presence of NAD cap formation is inhibited in C17 cells, and only small aggregates of CD3 and CD45 are discernable. Therefore, formation of a signal transmitting cap containing CD3, CD45, and p56 lck does not take place in NAD-treated cells expressing ART-1.
These results provide conclusive evidence that the inhibitory action of NAD in TCR-mediated activation of cells expressing ART-1 occurs at the level of transmembrane signaling. The inability of the TCR complex to transmit signals can now be explained by the failure of ADP-ribosylated co-receptors to aggregate with the TCR to form a functional signaling structure. Our demonstration that ADP-ribosylation of cell surface molecules causes inhibition of transmembrane signaling uncovers an entirely novel mechanism of cell regulation in which cell surface receptors rather than their ligands are modified by an ecto-enzyme. The modification alters the ability of receptors to associate and to form a functional cell surface cluster. Hence, ADP-ribosyltransferases represent a new family of cell surface enzymes able to regulate T lymphocyte function by modulating receptor association and transmembrane signaling. FIG. 3. A, co-localization of CD3 and p56 lck is inhibited in C17 but not in EL4 cells in the presence of NAD. EL4 and C17 cells were pretreated with NAD and then reacted with anti-CD3 antibody, followed by a secondary rhodamine-conjugated antibody. To visualize the localization of p56 lck , cells were permeabilized and then reacted with anti-lck antibody, followed by incubation with FITC-labeled secondary antibody. After fixation cells were examined by confocal microscopy. I, CD3 staining (rhodamine); II, lck staining (FITC); III, overlay of CD3 and lck staining; IV, phase contrast images of the selected field. B, co-localization of CD3 and CD45 is inhibited in C17 cells but not in EL4 cells in the presence of NAD. EL4 and C17 cells were pretreated with NAD and then reacted with anti-CD3 antibody, followed by a secondary rhodamine-conjugated antibody. To examine the localization of CD45, cells were reacted with FITC-labeled anti-CD45 antibody. I, CD3 staining (rhodamine); II, CD45 staining (FITC); III, overlay of CD3 and CD45 staining; IV, phase contrast images of the selected field.