Protein Kinase C ϵ Dependence of the Recovery from Down-regulation of S1P1G Protein-coupled Receptors of T Lymphocytes

Sphingosine 1-phosphate (S1P) from mononuclear phagocytes and platelets signals T cells predominantly through S1P1 G protein-coupled receptors (Rs) to enhance survival, stimulate and suppress migration, and inhibit other immunologically relevant responses. Cellular S1P1 Rs and their signaling functions are rapidly down-regulated by S1P, through a protein kinase C (PKC)-independent mechanism, but characteristics of cell-surface re-expression of down-regulated S1P1 Rs have not been elucidated. T cell chemotactic responses (CT) to 10 and 100 nm S1P and inhibition of T cell chemotaxis to chemokines (CI) by 1 and 3 μm S1P were suppressed after 1 h of preincubation with 100 nm S1P, but recovered fully after 12–24 h of exposure to S1P. Late recovery of down-regulated CT and CI, but not early down-regulation, was suppressed by PKC and PKCϵ-selective inhibitors and was absent in T cells from PKCϵ-null mice. The same PKCϵ inhibitors blocked S1P-evoked increases in T cell nuclear levels of c-Fos and phosphorylated c-Jun and JunD after 24 h, but not 1 h. A mixture of c-Fos plus c-Jun antisense oligonucleotides prevented late recovery of down-regulated CT and CI, without affecting S1P induction of down-regulation. Similarly, S1P-elicited threonine phosphorylation of S1P1 Rs was suppressed by a selective inhibitor of PKCϵ after 24 h, but not 1 h. Biochemical requisites for recovery of down-regulated S1P1 Rs thus differ from those for S1P induction of down-regulation.

T cells express predominantly S1P 1 and S1P 4 , of which S1P 1 transduces two distinct effects of S1P on T cell migration and also regulates other T cell functional responses (7,8). S1P is chemotactic for T cells at 0.001-0.1 M, enhances chemotactic responses to chemokines at 0.01-0.1 M, and suppresses T cell chemotaxis to numerous stimuli at 0.3-3 M. As T cell antigen receptor-dependent activation of T cells suppresses expression of S1P GPCRs and functional responses to S1P in parallel, the S1P-S1P 1 R axis is considered most important in controlling recruitment and stimulation of naïve and memory T cells by setting their response threshold to other stimuli.
One unanswered critical question about S1P 1 Rs of T cells and other types of cells is how they are maintained at a level of full functional expression in tissues and fluids where there are completely saturating micromolar concentrations of S1P. Epitope-tagged or fluorescent protein-containing recombinant S1P 1 Rs introduced by transfection into several different cell lines were down-regulated by S1P through phosphorylation, internalization, and translocation to a caveolar compartment by a G protein-coupled receptor kinase 2-dependent process (9,10). Rapid down-regulation and internalization, but not S1P 1 R binding and signaling functions, were facilitated by N-linked glycans of S1P 1 Rs (11). However, recovery and stabilization of cell-surface expression of S1P 1 Rs after down-regulation have not been examined previously. We now report that T cell S1P 1 R recovery from S1P-induced down-regulation requires protein kinase C ⑀ (PKC⑀) activity and AP-1 transcriptional complex, and involves PKC⑀-dependent late phosphorylation of the S1P 1 Rs.

EXPERIMENTAL PROCEDURES
Isolation, Transfection, Culture, and S1P Treatment of Cells-Mouse CD4 T cells were isolated from splenocytes of 6 to 8 week-old C57BL/6 female mice at a minimum purity of 97% using metallic beads bearing anti-CD4 monoclonal antibodies (MoAbs) and two cycles of magnetic retention chromatography (Miltenyi-Biotec, Auburn, CA), as described (8). HTC4 rat hepatoma cells, which lack any endogenous S1P or LPA Rs, were LipofectAMINE-transfected with a pcDNA3.1 (ϩ) construct (Invitrogen) encoding influenza hemagglutinin peptide (HA)-NH 2 -terminally tagged human S1P 1 Rs and selected by culture with 400 g/ml of geneticin. The level of expression of S1P 1 Rs was established by TaqMan real-time PCR (8). Suspensions of 0.5 ϫ 10 6 purified CD4 T cells per ml of RPMI 1640 with 5% charcoal-adsorbed FBS were transfected with 50 g/ml each of the c-Jun plus c-Fos antisense or corresponding sense phosphorothioated oligonucleotides (Biomol, Plymouth Meeting, PA), which had been preincubated with FuGene reagent (Roche Diagnostics), and further incubated for 16 h prior to exposure to S1P. Two-ml aliquots of some suspensions of purified CD4 T cells in RPMI 1640 with 50 g/ml fatty acid-free BSA (FAF-BSA) (Calbiochem) and of HTC4-S1P 1 (HA) cells in Dulbecco's modified Eagle's medium with 50 g/ml FAF-BSA were preincubated for up to 48 h in 6-well plates to which were added one or more biochemical inhibitors and/or 10 Ϫ7 M S1P at 12 h intervals or 1 h prior to assessment of S1P 1 R expression, phosphorylation, or functional signaling.
Quantification of Migration of Mouse CD4 T Cells-Migration of mouse purified CD4 T cells was analyzed in Transwell chambers (Costar, Cambridge, MA) with human type IV collagen (Sigma)-coated 5-m pore width polycarbonate filters and incubation for 4 h, as described previously (6). T cell suspensions were 1 ϫ 10 7 /ml in RPMI 1640 with 5% heated and charcoal-extracted fetal bovine serum, from which 0.1 ml portions of each were loaded into top compartments of chemotactic chambers. S1P, CCL21 (Exodus-2) and CCL5 (RANTES) (Peprotech, Inc., Rocky Hill, NJ) in 0.6 ml of the same buffer were the positive chemotactic stimuli. T cells that had migrated through the filter and into the lower compartment were counted and the chemotactic response (CT) expressed as a percentage of the total T cells initially added to the upper compartment. CT of T cells pretreated with 100 nM S1P and/or an oligonucleotide or pharmacological agent was expressed as a percentage of the concurrent CT of untreated T cells (100%). The inhibition of chemokine-evoked CT by 1 and 3 M S1P, termed CI, was expressed as percentage inhibition and compared with percentage inhibition of control T cells not preincubated with inhibitors and/or S1P or preincubated with control compounds. The significance of differences between migration altered with S1P and/or inhibitors and control CT or CI was calculated with a paired or two-sample Student's t test.
PKC⑀-Null Mice-Mice with a selective absence of PKC⑀, but normal tissue levels of other subtypes of PKC, were obtained from Dr. Robert Messing (Gallo Research Center, Emeryville, CA) (12).
Western Blot Analyses of S1P 1 GPCRs-Replicate suspensions of HTC4-S1P 1 (HA)-stable transfectants were incubated with various protein kinase inhibitors and phosphatase inhibitors before preincubation with S1P for 1 h or 12 and 24 h. Membranes of the transfectants then were prepared by homogenization in 50 mM Tris-HCl, 100 mM NaCl with 1 mM EDTA, 1 mM dithiothreitol, and 1% glycerol (pH 7.4) with HALT Protease Inhibitor Mixture (Pierce), recovered by centrifugation at 15,000 ϫ g for 20 min at 4°C, and solubilized in 0.1% Nonidet P-40. One hundred l of anti-HA rat MoAb-agarose matrix (Roche Molecular Biochemicals) was added to each preparation of solubilized membranes in 400 l of 50 mM Tris-HCl with 0.1% Nonidet P-40, incubated for 1 h at 4°C, washed twice with the same solubilizing buffer, resuspended in 40 l of electrophoresis loading solution, and heated for 3 min at 100°C. Immunoprecipitated proteins then were resolved by SDS-PAG electrophoresis, transferred, and stained as described previously (5) sequentially with 1 g/ml mouse anti-Thr(P) MoAb(H-2, Santa Cruz Biotechnology, Inc., Santa Cruz, CA) and anti-HA rat MoAb.
Nuclear Extraction and ELISA Quantification of c-Fos and Phosphorylated c-Jun/JunD-Replicate suspensions of 4 ϫ 10 6 mouse spleen CD4 T cells were preincubated either with various protein kinase inhibitors for 30 min before incubation with S1P for 1 h or with S1P for 24 h, which included re-additions at 12 and 1 h and where the protein kinase inhibitors were introduced at 23, 12, and 1 h. c-Fos and phosphorylated c-Jun and JunD ("c-Jun") were measured in nuclear extracts of the T cells by an ELISA, according to protocol instructions, where AP-1 complexes bind to fixed oligonucleotides containing the 12-Otetradecanoylphorbol-13-acetate-responsive element 5Ј-TGA(C/G)T-CA-3Ј and AP-1 constituents are quantified by binding of antibodies to accessible epitopes (Active Motif, Carlsbad, CA).

RESULTS AND DISCUSSION
The addition of S1P to S1P-deprived cells of many types results in down-regulation and internalization of S1P 1 Rs, but T cells freshly isolated from fully saturating ambient S1P levels of 100 -300 nM express a full complement of S1P 1 Rs (9, 10). To determine the course of recovery and persistence of T cell functional S1P 1 Rs, mouse CD4 T cells were isolated, preincubated in medium without S1P for 16 h, and then incubated for 48 h with 100 nM S1P present for the entire period or for shorter times prior to assessment of chemotaxis. In the absence of re-exposure to S1P, mean control chemotactic responses Ϯ S.D. to 10 Ϫ8 M and 10 Ϫ7 M S1P, respectively, were 17 Ϯ 4% and 24 Ϯ 6% for all intervals of preincubation as a group and mean control inhibition of chemotaxis to CCL21 by 10 Ϫ6 M and 3 ϫ 10 Ϫ6 M S1P as a group were 47 Ϯ 14% and 53 Ϯ 6%. A 1-h exposure, which down-regulates S1P 1 Rs, significantly decreased the chemotactic response of CD4 T cells to 10 Ϫ8 M and 10 Ϫ7 M S1P and concurrently reduced suppression of CD4 T cell chemotactic responses to CCL21 by 10 Ϫ6 M and 3 ϫ 10 Ϫ6 M S1P (Fig. 1). Similar decreases were observed for suppression of chemotactic responses to CCL5 by S1P. At 12 h and for up to 48 h of exposure to S1P, the S1P 1 R-mediated direct chemotactic effects and chemotactic inhibitory effects of S1P both returned to control levels similar to those for CD4 T cells not exposed to down-regulating levels of S1P.
The capacity of inhibitors of potentially relevant signal transducers to prevent recovery of function of down-regulated S1P 1 Rs also was assessed in chemotactic assays. After 24 h of Control chemotactic values (100% on the left ordinate, left one or two bars) without prior exposure to S1P ranged from 14 to 20% of the total number of CD4 T cells initially added to the top compartment of the chemotactic chambers at 1 and 24 h with 10 Ϫ8 M S1P and 18 to 26% with 10 Ϫ7 M S1P. The range for chemotaxis to 200 nM CCL21 was 34 -41% for 1 and 24 h preincubation without S1P, which represents 0% inhibition on the right ordinate for the one or two right-hand bars in each set. Control values for inhibition of chemotaxis to 200 nM CCL21 (right-hand ordinate) were 43-58% at 1 and 24 h for 10 Ϫ6 M S1P and 47-62% for 3 ϫ 10 Ϫ6 M S1P in the T cell compartment. A negative value for inhibition indicates enhancement. The statistical significance of effects of all inhibitors were calculated relative to the respective responses with no inhibitor (0) at 1 and 24 h, and the symbols are the same as described in the legend to Fig. 1. B, lack of recovery in CD4 T cells of PKC⑀-null mice. Each column and bar depicts the mean Ϯ S.D. of the results of two separate studies conducted in triplicate. Control chemotactic values (100% on the left ordinate) without prior exposure to S1P ranged from 16 to 25% and 11 to 18% of the total number of CD4 T cells initially added to the top compartment of the chemotactic chambers for wild-type and PKC⑀-null mice, respectively, with 10 Ϫ8 M S1P and 20 to 24% and 14 to 19% with 10 Ϫ7 M S1P. Chemotaxis to 200 nM CCL21 was 36 -40% and 34 -37% for wild-type and PKC⑀-null mice, respectively (0% for inhibition of chemotaxis to CCL21 on the right-hand ordinate). Control values for inhibition of chemotaxis to 200 nM CCL21 without S1P preincubation (right-hand ordinate) were 38 -42% at 1 and 24 h for wild-type mice and 49 -54% for PKC⑀-null mice with 10 Ϫ6 M S1P and 47-53% for wild-type mice and 50 -66% with 3 ϫ 10 Ϫ6 M S1P in the T cell compartment. Statistical significance was calculated for differences in corresponding values at the same time between wild-type and PKC⑀-null mice and symbols are the same as in Fig. 1. exposure of S1P-deprived CD4 T cells to S1P, the direct chemotactic effects and inhibitory actions on chemokine-evoked chemotaxis of 10 Ϫ8 and 10 Ϫ6 M S1P, respectively, had returned to control levels ( Fig. 2A, right frame). The PKC type-specific inhibitors calphostin C and Ro318220, and the PKC⑀ subtypeselective inhibitor m-PKC⑀V1-2 significantly suppressed the recovery of both functions of the S1P 1 Rs. In contrast, the PKA inhibitor KT5720 and the PKC␣/␤/␥ inhibitory peptide m-PKC␤C2-4 had no effect on recovery of down-regulated S1P 1 Rs (Fig. 2A). In addition, neither calphostin C nor m-PKC⑀V1-2 altered the S1P-elicited down-regulation of functional S1P 1 Rs after 1 h of exposure of CD4 T cells to S1P ( Fig.  2A, left frame). To confirm involvement of PKC⑀ in the recovery of down-regulated S1P 1 Rs, similar studies were conducted with CD4 T cells from selective PKC⑀-null mice. The patterns of down-regulation of functional S1P 1 Rs by a 1-h exposure to 100 nM S1P, reflected in reduced direct chemotactic responses and suppressed inhibition of chemotaxis to CCL21 relative to controls not preincubated with S1P, were identical in wild-type and PKC⑀-null mice (Fig. 2B, left two sets). In contrast, only the wild-type CD4 T cells had recovered S1P 1 R-mediated chemotactic and chemotactic inhibitory responses after 24 h of exposure to S1P (Fig. 2B, first of right two sets). PKC⑀-null mousederived CD4 T cells continued to show impaired direct chemotactic responses to S1P without suppression by S1P of CCL21-evoked chemotaxis after 24 h (Fig. 2B, second of right two sets), which was a pattern indistinguishable from that observed after 1 h (Fig. 2B, second of left two sets).
PKC⑀ in T cells has been linked to recruitment and functional activation of the AP-1 and N-FAT-1 transcription factors (13). Two approaches were used to examine independently the course and PKC⑀ dependence of activation of AP-1 by S1P and the involvement of components of AP-1 in recovery of downregulated functional S1P 1 Rs of CD4 T cells. First, the characteristics of S1P activation of c-Fos and c-Jun/JunD in CD4 T cells were investigated, as some growth effects of S1P involve engagement of the c-Fos promoter. S1P increased the nuclear contents of c-Fos and phosphorylated c-Jun/JunD by 1 h, and the levels were sustained for up to 24 h in the continued presence of a plasma concentration of 100 nM S1P (Fig. 3A). These increases were blocked significantly by the type-specific PKC inhibitor calphostin C and the PKC⑀-selective inhibitor m-PKC⑀V1-2 at 24 h, but not at 1 h, and not at either time point by the PKC␣/␤/␥ inhibitory peptide m-PKC␤C2-4 ( Fig.  3A) (14). Second, the involvement of AP-1 components in recovery of chemotactic signaling by down-regulated S1P 1 Rs of CD4 T cells was demonstrated by introducing c-Fos plus c-Jun antisense oligonucleotides into CD4 T cells, in amounts proven to FIG. 3. PKC⑀ dependence of the role of AP-1 in S1P effects on CD4 T cell migration. A, quantification by ELISAs of the involvement of PKC⑀ in S1P effects on expression of c-Fos and c-Jun/JunD in nuclear extracts of CD4 T cells. Each column and bar depicts the mean Ϯ S.D. of the results of two separate studies conducted in duplicate. All samples were incubated for 24 h with inhibitors and S1P added at 24, 12, and 1 h for the four "24-h" sets and at 1 h prior to termination for the four "1-h" sets. The c-Jun values include phosphorylated c-Jun and JunD. Mean net base-line values without S1P were 0.11 and 0.14 for c-Fos and c-Jun/JunD, respectively, at 1 h and 0.16 and 0.12 at 24 h. B, prevention of recovery of S1P 1 GPCR effects on migration of CD4 T cells by c-Fos ϩ c-Jun antisense oligonucleotides. Each column and bar depicts the mean Ϯ S.D. of the results of two separate studies conducted in triplicate. Control values (100% on the left ordinate) without S1P pretreatment ranged from 23 to 28% and 31 to 35% of the total number of CD4 T cells initially added to the top compartment of the chemotactic chambers for sense (S)-and antisense (AS)-treated sets, respectively, with 10 Ϫ7 M S1P (lefthand bar) and 23 to 29% and 24 to 27% for 200 nM CCL21 (middle bar). The range of control values without S1P pretreatment for inhibition of chemotaxis to 200 nM CCL21 was 37-45% and 39 -44% for sense-and antisense-treated sets, respectively, with 10 Ϫ6 M S1P in the T cell compartment (right-hand bar and right ordinate). Statistical significance was calculated for differences in corresponding values at the same time between sense-and antisense-treated sets, and symbols are the same as described in the legend to Fig. 1. decrease AP-1 sufficiently for functional alterations (15), prior to assessing S1P effects on chemotaxis. Suppression of both direct chemotactic effects of 10 Ϫ7 M S1P and of the inhibition by 10 Ϫ6 M S1P of CCL21-induced chemotaxis by 1 h of prior reexposure to S1P was not affected by the mixture of AP-1 antisense oligonucleotides nor the corresponding sense oligonucleotides (Fig. 3B, left-hand set). In contrast, recovery of both chemotactic stimulation by S1P and S1P inhibition of chemokine-elicited chemotaxis after 24 h was prevented by c-Fos plus c-Jun antisense oligonucleotides (Fig. 3B, set furthest to the right), whereas timely recovery was complete for CD4 T cells pretreated only with corresponding sense oligonucleotides. The migration response patterns were no different after 24 h than 1 h for the antisense oligonucleotide-pretreated CD4 T cells. S1P 1 (HA) Rs of S1P-deprived HTC4-transfectants re-exposed to 100 nM S1P were threonine-phosphorylated after 1 and 24 h (Fig. 4). Selective inhibition of PKC⑀ suppressed S1P-evoked threonine phosphorylation after 24 h, but not after 1 h. Protein phosphatase inhibition enhanced threonine phosphorylation of S1P 1 Rs slightly after 24 h but not after 1 h. Thus only late threonine phosphorylation of S1P 1 (HA) Rs appears to be PKC⑀-dependent and occurs in the same time period as reappearance of functional S1P 1 Rs. The PKC⑀ dependence of recovery and persistence of functional S1P 1 Rs in the presence of micromolar concentrations of S1P, which fully saturate the S1P 1 Rs, thus also may require S1P 1 R specific phosphorylation.
Elucidation of the different requisites for S1P and phorbol ester induction of rapid phosphorylation and internalization of epitope-tagged recombinant S1P 1 receptors in a line of hamster fibroblast transfectants revealed independent mechanisms (9). Immediate down-regulation evoked by S1P depended on 12 amino acids of the carboxyl terminus, was resistant to suppression by PKC inhibitors, and was mediated in part by G proteincoupled receptor kinase-2. In contrast, elicitation of immediate down-regulation by a phorbol ester was not dependent on the carboxyl terminus and was completely suppressed by PKC inhibitors. Dissociation of early S1P-evoked down-regulation of S1P 1 Rs from PKC activity was confirmed for T cells by the lack of effect of PKC inhibitors or of PKC⑀ genetic deletion on down-regulation of functional S1P 1 Rs (Fig. 2, A and B). Late recovery and persistence of S1P 1 Rs down-regulated by S1P were shown to depend on PKC⑀ by the suppressive effects of selective and broadly specific PKC inhibitors, PKC⑀ gene dele-tion, and antisense inhibition of portions of the S1P 1 -coupled signaling pathways (Figs. 2-4).
Coupling of expression of S1P 1 Rs to the activation of PKC⑀ was considered because of the known capacity of PKC⑀ to recruit components of the AP-1 transcription complex, which are implicated in S1P signaling (13). These observations have been extended to the demonstration of involvement of PKC⑀evoked AP-1 in late-phase recovery of down-regulated S1P 1 Rs (Fig. 3). A sustained rise in expression of the c-Fos and phosphorylated c-Jun/JunD components of AP-1 resulted in peak levels for all at 24 h after S1P stimulation (Fig. 3). Antisense suppression of these same components of AP-1 blocked PKC⑀dependent late recovery of S1P 1 Rs from down-regulation by S1P, without altering PKC⑀-independent S1P induction of early down-regulation. Although the site(s) of S1P-evoked and PKC⑀-dependent phosphorylation of S1P 1 Rs have not been established, threonine (Fig. 4), and presumably serine, are preferred phosphorylation targets consistent with the specificity of PKC. That PKC⑀-dependent phosphorylation of Edg-1, recruitment of AP-1 components, and recovery of down-regulated Edg-1 receptors follow a similar late time course suggests mechanistic relationships. However, the precise roles and interrelationships of late phosphorylation of S1P 1 Rs and AP-1 regulation of specific transcriptional events remain to be elucidated. Ongoing studies are examining effects of AP-1 antisense oligonucleotides on late phosphorylation of S1P 1 receptors, distinctive characteristics of S1P 1 R phosphorylation in PKC⑀-null mice, and a range of mutant Edg-1 Rs to determine which will not be PKC⑀-phosphorylated.
That distinct mechanisms govern S1P 1 R down-regulation and recovery has numerous implications for cell biology and immunology. S1P effects on S1P 1 Rs of T cells differ substantially from those of T cell antigen receptor activation, which persistently suppresses expression of all S1P GPCRs with a distinctive rank order of efficacy. It explains how S1P 1 Rs may be internalized by acute exposure to S1P, but avoid persistent down-regulation in the presence of plasma and lymph levels of 100 -300 nM S1P. It also suggests the possibility that agents may be developed which will distinguish and separately regulate down-regulation and recovery of S1P 1 Rs pharmacologically. FIG. 4. Western blot analyses of threonine phosphorylation of S1P 1 (Edg-1) GPCRs in HTC4-S1P 1 (HA) transfectants. The upper row shows immunolabeling of phosphothreonine (p-Thr) in the ϳ55-kDa immunoprecipitated HA epitope-tagged S1P 1 GPCRs, and the lower row shows the total amount of S1P 1 (HA) receptor. All samples were incubated for 24 h, but with inhibitors and S1P added at 24, 12, and 1 h for the three 24-h sets and at 1 h prior to termination for the three 1-h sets. N-Myristoyl-PKC⑀ V1-2 was used at 100 M and okadaic acid at 100 nM. The left-hand values are for molecular mass standards.