Signaling through CD43 regulates CD4 T-cell trafficking

Cell-surface mucins, characterized by large, highly glycosylated extracellular domains, are important regulators of tumor progression as well as lymphocyte trafficking.¹  Overexpression and altered glycosylation of mucins on cancer cells, compared with normal cells, makes them ideal candidates for tumor-based vaccines.^2,3  CD43 is one such member of the mucin family,⁴  and is widely expressed on the surface of hematopoietic cells.⁵  Reduced expression and aberrant glycosylation of CD43 is found on the T cells of patients suffering from Wiskott-Aldrich syndrome,^6-8  as well as HIV infection.^9,10  In contrast, CD43 is overexpressed on lymphomas and leukemias, and while not normally expressed on nonhematopoietic cells, it is present on colon adenomas and carcinomas.^11,12 

In spite of its abundance on T cells, the precise function of CD43 remains elusive. T cells from CD43^−/− mice exhibit increased proliferation and adhesion in vitro,^13,14  and increased immune responses in vivo.¹⁵  Strikingly, CD43 is excluded from the immune synapse of activated T cells,¹⁶  and often localizes to a protein complex distal to the antigen-presenting cell (APC) contact site, referred to as the distal pole complex (DPC).^17,18  Several groups have reported that failure to exclude CD43 results in inhibition of IL-2 production.^19,20  Based on these data, CD43 has been characterized as a negative regulator of T-cell function.

The prevailing hypothesis to explain the mechanism of CD43-mediated negative regulation was that the CD43 extracellular domain acted as a barrier to cell-to-cell interactions.^13,21-23  However, we and others have shown that the cytoplasmic domain is not only necessary, but also sufficient for movement of CD43 away from the immunologic synapse, as well as CD43-mediated attenuation of T-cell proliferation and homotypic adhesion.^14,17,20  These data underscore a critical function for the CD43 cytoplasmic tail in regulating T-cell proliferation and adhesion.

Despite significant progress in elucidating the function of the CD43 cytoplasmic tail, there is no well-defined function for its extracellular domain. In the absence of an exclusive ligand, it seems likely that CD43 is actually a receptor for many proteins, both endogenous and exogenous, that bind to specific sugar moieties.^24-27  However, the physiologic relevance of these interactions in vivo remains to be elucidated.

The transmembrane and cytoplasmic regions of CD43 are highly conserved, supporting a role for CD43 in signal transduction.^28,29  The juxtamembrane region has a stretch of basic amino acids (KRR) that bind to the cytoskeletal linker proteins ezrin, radixin, and moesin (ERM).³⁰  This association is required for CD43 exclusion from the immunologic synapse of activated T cells,^17,19  as well as its localization to the cleavage furrow of dividing T cells.³¹  Interestingly, although CD43-mediated attenuation of T-cell proliferation requires an intact cytoplasmic tail, it is independent of CD43 association with ERM proteins.²⁰  This suggests that additional structural elements within the cytoplasmic tail are involved in this process.

The CD43 cytoplasmic tail has conserved serines and threonines that could be potentially phosphorylated. Indeed, human CD43 is phosphorylated in resting cells and hyperphosphorylated upon stimulation,^32-35  and ligation of CD43 with the mAb L10 results in activation of several signaling cascades.^34,36-38 

In this report, we investigate the physiologic role of CD43 phosphorylation in the regulation of T-cell function. We demonstrate that murine CD43 is phosphorylated constitutively and inducibly. We now identify Ser76 (numbering refers to position from start of cytoplasmic domain in murine CD43) as a novel site of basal phosphorylation. Surprisingly, mutation of Ser76 did not alter CD43-mediated attenuation of T-cell proliferation, or CD43 exclusion from the immune synapse. Using a competitive in vivo migration assay, we find that CD43^−/− T cells exhibit impaired trafficking to lymph nodes, but not the spleen. Reconstitution of CD43^−/− T cells with wild-type CD43 can restore trafficking to lymph nodes. However, mutant CD43 that cannot be phosphorylated at Ser76 fails to rescue the trafficking defect. Finally, we show that the cytoplasmic domain alone is not sufficient to rescue trafficking of CD43^−/− T cells, underlining an important function for the CD43 ectodomain in this process. These data suggest that CD43-mediated regulation of T-cell trafficking involves interaction of the CD43 ectodomain with endogenous ligands, which in turn may transduce a signal through phosphorylation at Ser76.

DBA/2 and BALB/c mice were from Charles Rivers Laboratories (Wilmington, MA) or the National Cancer Institute (Frederick, MD). DO.11.10 and DO.CD43^−/− mice¹⁷  were bred and housed in a specific pathogen–free condition at the University of Chicago Animal Resources Center (Chicago, IL).

Affinity-purified anti-CD28 (PV-1³⁹ ), anti-CD3 (145–2C11⁴⁰ ) and anti-mouse CD43 (S11⁴¹ ) were prepared in our laboratory. mAbs were purchased for CD4 (BD PharMingen, San Diego, CA), CD62L (BD PharMingen), KJ1–26 (CALTAG Laboratories, Burlingame, CA), CCR7 (eBiosciences, San Diego, CA) and human CD16 (BioLegend, San Diego, CA).

DO.11.10 and DO.CD43^−/− T lymphoblasts were prepared and assayed as previously described.^17,20 

CD43-FL and CD16–7-43 constructs were generated as previously described.¹⁷  The Ser72Ala, Ser76Ala, and Ser72AlaSer76Ala mutants were generated from CD43-FL using the Quickchange Site Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Empty vector–green fluorescent protein (EV-GFP) was made by cloning eGFP into the pBMN vector. CD43-FL was cloned into pEGFP-N1 (Clontech, Mountain View, CA) and subcloned into the pBMN vector for CD43-FL-GFP.

Flow cytometric analysis was performed with a FACS Calibur or LSRII (BD Biosciences, Mountain View, CA), and the data were analyzed with FlowJo software (Tree Star, Ashland, OR).

Competitive migration assays were performed in 1 of 3 different ways. (1) DO.11.10 and DO.CD43^−/− T lymphoblasts were centrifuged through a Percoll (Amersham Biosciences, Uppsala, Sweden) step gradient, and cells at the 50% to 75% interface were collected. Resting lymphoblasts (Figure 4A-C) or purified ex vivo T cells from DO.11.10 and DO.CD43^−/− mice (Figure 4F) were mixed in equal numbers. A total of 10⁷ cells were intravenously injected into naive BALB/c recipients. After approximately 16 hours, blood, spleen, and lymph nodes (axillary, brachial, and inguinal) were harvested and stained for CD4, CD43, and the transgenic T-cell receptor (TCR) DO.11.10 (using KJ1–26). The frequency of CD43⁻ (knock-out [KO]) and CD43⁺ (wild-type [WT]) cells was determined in the donor population (CD4⁺KJ1–26⁺). The ratio of CD43⁻ to CD43⁺ T cells was obtained by dividing the percentage of donor CD43⁻ cells by the percentage of donor CD43⁺ cells and corrected for the input ratio. (2) Resting T cells transduced with different constructs (Figures 4D, 6) were mixed in equal numbers and injected into mice intravenously. For CD43-FL-GFP– and EV-GFP–transduced T cells, donor CD4 T cells were identified by GFP, CD4, and CD43 expression. For CD43-FL and CD16–7-43, donor T cells were identified by CD4 and KJ1–26 staining. The frequency of donor-transduced CD43⁺ and transduced CD16⁺ cells was determined by staining for CD43 and CD16, respectively. (3) One set of naive BALB/c recipients was injected with resting CD43-FL transduced T cells, while another set was injected with resting CD43-Ser76Ala transduced T cells (Figure 5A). In each case, donor T cells were identified by KJ1–26 and CD4 staining. The frequency of donor-transduced (CD43⁺) cells was compared with that of nontransduced (CD43⁻) cells and is depicted as a ratio.

T cells were incubated in phosphate-free media supplemented with dialyzed FCS for 1 hour and then for 2 hours in media containing [³²P] orthophosphate (Amersham Biosciences) at 37 Bq/mL (1 mCi/mL). Cells were washed 5 times, stimulated with 50 ng/mL PMA, and lysed in buffer containing 50 mM Tris [pH 7.6], 150 mM NaCl, 0.5% Triton-X, 5 mM EDTA, phosphatase inhibitors (1 mM NaF, 1 mM Na₃VO₄), protease inhibitors (10 μg/mL aprotinin, 1 mM PMSF, 10 μg/mL leupeptin, and 10 μg/mL pepstatin) and phosphatase inhibitor cocktail (Sigma, St Louis, MO). Lysates were incubated overnight at 4°C with anti-CD43 coated Protein-G beads (Zymed, San Francisco, CA), boiled, separated by SDS-PAGE, transferred onto PVDF membranes, detected by autoradiography, and then analyzed for total CD43 levels by Western blotting using the LAS-3000 (Fujifilm, Stamford, CT).

CD43 was immunoprecipitated from 10⁷ EL-4 cells or DO11.10 lymphoblasts using the ProFound Mammalian Co-Immunoprecipitation Kit (Pierce, Rockford, IL), resolved by SDS-PAGE, stained with Coomassie R-250 (Pierce), excised, and washed extensively with 50:50 (acetonitrile/100 mM ammonium carbonate [pH 8.9]). Rehydrated gel pieces were trypsin digested (12.5 ng/μL [pH 8.9], 37°C; Promega, Madison, WI). Extracted peptides were injected onto a reversed-phase column (75-μm internal diameter Zorbax Stablebond [300 Å pore]) connected to an Agilent Technologies 1100 nano/capillary liquid chromatography/magnetic circular dichroism (LC/MSD) XCT system (Agilent Technologies, Santa Clara, CA). Samples were chromatographed at a flow rate of 300 nL/minute. The mass spectrometer was operated in positive ion mode with the trap set to data-dependent auto–tandem mass spectrometry (MS/MS) acquisition mode. Source conditions were as follows: capillary voltage (Vcap), −4500 V; drying gas flow, 4 L/minute; drying gas temperature, 250°C; and capillary exit voltage (CapEx), 65 V. Ions eluting from the LC column trigger the ion trap to isolate the ion and perform an MS/MS experiment scan after the MS full scan. Data files were searched using SpectrumMill (Agilent Technologies) and Mascot 1.9 (Matrix Sciences, Boston, MA) against the National Center for Biotechnology Information (NCBI) nonredundant (nr) database.⁴²  Putative phosphopeptides located by generating a neutral loss chromatogram (looking for predominant loss of 98, 49, or 32.6 mass units from any precursor ion) using the Agilent datasystem were then examined by an additional sequential product-ion spectra (MS³) LC/MS/MS experiment to confirm peptide sequence.

Immunofluorescence staining assessing CD43 localization in T-cell–A20 conjugates was performed as previously described.¹⁷  Briefly, transducted T cells were conjugated with OVA_323-339 pulsed A20 cells by spinning at 50g for 5 minutes followed by 5 minutes incubation at room temperature. Immunofluorescence staining with anti–CD43-FITC was performed as previously described.^17,20  After staining, cells were observed using a 63X Planapochromat objective and imaged using a Zeiss Axiovert 200 M (Roper, Tucson, AZ). Single images were processed by no-neighbor deconvolution using OpenLab 4.0.4 (Improvision, Lexington, MA) to remove out-of-focus haze. Conjugates that did not have CD43 at the T cell/APC interface were scored as excluded. At least 50 conjugates were scored in 3 independent experiments.

All statistics were done using a paired Student 2-tailed t test, except Figure 3D, which uses an unpaired Student 2-tailed t test. Error bars represent SEM.

While we have previously shown that the cytoplasmic domain of CD43 is both necessary and sufficient for CD43-mediated negative regulation of T-cell proliferation,^17,20  the specific structural elements remain undefined. CD43 has been shown to be phosphorylated constitutively and inducibly in human T cells.^32-35  To determine whether murine CD43 was also phosphorylated, we radiolabeled unstimulated ex vivo DO.11.10 T cells (Figure 1A), or resting DO.11.10 T-cell lymphoblasts (Figure 1B) with [³²P] orthophosphate and incubated them with or without PMA. We find that CD43 is constitutively phosphorylated in both ex vivo CD4 T cells and resting CD4 T lymphoblasts. Stimulation with PMA for 60 minutes increases the basal level of CD43 phosphorylation. Further, we did not observe any changes in total CD43 phosphorylation upon antigen stimulation (data not shown). This is consistent with a previous report that activation of human peripheral blood lymphocytes (PBLs) by OKT3 and IL-2 does not alter total CD43 phosphorylation.³³  Nevertheless, similar to human T cells, CD43 is constitutively and inducibly phosphorylated in murine CD4 T cells.

Although total CD43 phosphorylation was first demonstrated 15 years ago, the precise residues have not yet been mapped. The CD43 cytoplasmic tail has been shown to be highly homologous between mouse, rat, and human.^28,29  Using ClustalW⁴³  to align additional protein sequences available in the NCBI database, we now demonstrate that the CD43 cytoplasmic domain is highly conserved across a wide range of species, including cattle and dog (Figure 1C). Further, there is complete identity between human and chimpanzee CD43. Interestingly, in all 6 species examined, the cytoplasmic domain lacks any tyrosines, but has several serines and threonines that could be potentially phosphorylated. In fact, 7 serines and threonines (Thr8, Thr39, Ser57, Thr64, Thr65, Ser72, and Ser76 in murine CD43) are completely conserved across all the species examined. To identify potential phosphorylation sites, we performed MS/MS on purified CD43 from murine EL-4 thymoma cells. These data were later confirmed in resting DO.11.10 T lymphoblasts. Analysis of CD43 mass spectra for characteristic neutral losses related to phosphorylation resulted in the observation of a highly abundant phosphopeptide with mass-to-charge ratio (m/z) of 730.7 (Figure 2A). This corresponded to the triply charged ion from the peptide ⁷²SRQGSLVLEELKPGSGPNLK⁹¹ phosphorylated in 1 position. The MS/MS spectrum for m/z 730.7 is shown in Figure 2B. A prominent fragment ion at m/z 698, resulting from loss of H₃PO₄ from the precursor ion (m/z 730.7), was observed, verifying single phosphorylation of this peptide. This led us to conclude that the CD43 cytoplasmic domain is phosphorylated at 1 of 3 serines: Ser72, Ser76, or Ser86.

To determine which of the 3 serines was phosphorylated, the peptide corresponding to the ion at m/z 698 was selected for further analysis by MS³ (Figure 2C). Analysis of both b- and y-series ions in the spectra confirmed the sequence of the peptide. Predominance of normal y8-y10 sequence ions containing S86 demonstrated the complete absence of phosphorylation at this residue. Of interest are the b3* and b5* ions likely arising from β-elimination of the corresponding b3 and b5 ions. The almost complete absence of the b5 ion compared with the β-eliminated b5* ion strongly suggests that Ser72 or Ser76 are phosphorylated within the peptide (Figure 2D). However, relatively strong abundance of normal b3 ion compared with the β-eliminated b3* ion indicated that while Ser72 could be phosphorylated, Ser72 exists in both a nonphosphorylated and phosphorylated state (Figure 2E). Further, careful examination of the MS scan chromatogram revealed no trace of any diphosphorylated variant of this peptide. This suggests that either Ser72 or Ser76 is phosphorylated on a single CD43 molecule.

To determine whether Ser72 or Ser76 or both were phosphorylated, we generated mutant CD43 constructs containing alanine substitutions, either at position 72 (CD43-Ser72Ala) or 76 (CD43-Ser76Ala). These constructs or wild-type CD43 (CD43-FL) were transduced into DO.CD43^−/− T cells. Phosphorylation was assessed by incorporation of [³²P] orthophosphate. In both Ser72Ala and Ser76Ala mutants, CD43 phosphorylation is significantly reduced, showing that both Ser72 and Ser76 are phosphorylated (Figure 2F). Since Ser72 and Ser76 are in close proximity, it was possible that mutation at one site could potentially affect phosphorylation at the other site. We therefore assessed CD43 phosphorylation with a mutant CD43 construct containing alanine substitutions at both Ser72 and Ser76 (CD43-Ser72AlaSer76Ala; Figure 2G). Interestingly, mutation of both Ser72 and Ser76 reduced CD43 phosphorylation even more dramatically than mutation of either site alone, showing an additive effect of phosphorylation at Ser72 and Ser76. In conjunction with the absence of a diphosphorylated peptide in the CD43 mass spectra, these data suggest that even if both Ser76 and Ser72 are phosphorylated, they are likely to be phosphorylated on separate molecules. While these data do not preclude phosphorylation at other sites, they clearly demonstrate that Ser72 and Ser76 are predominant sites of phosphorylation within murine CD43 in resting T cells.

We next wanted to determine whether phosphorylation at Ser76 regulates CD43-mediated attenuation of T-cell proliferation. DO.CD43^−/− T cells were transduced with either CD43-FL or CD43-Ser76Ala (Figure 3A-B). In both cases, cells express similar levels of CD43 (Figure 3A; inset). Resting CFSE-labeled T cells were restimulated with OVA_323-339-pulsed A20 B cells. Proliferation was compared between CD43⁺ (transduced) and CD43⁻ (untransduced) CD4 T cells by CFSE dilution. As has been previously demonstrated,²⁰  and shown in Figure 3A, CD43^−/− T cells are hyperproliferative, and re-expression of CD43-FL decreases proliferation to wild-type levels. Interestingly, reconstitution with CD43-Ser76Ala is able to reduce proliferation similar to CD43-FL (Figure 3B). Surface expression of the activation markers CD69 and CD25 is similar between CD43-FL– and CD43-Ser76Ala–transduced cells upon restimulation (data not shown). Thus, phosphorylation at Ser76 is not required for CD43-mediated regulation of T-cell proliferation and activation.

CD43 exclusion from the immune synapse is mediated via interaction of its cytoplasmic domain with members of the ERM family.^17,19  In vitro binding studies have suggested that in addition to the KRR sequence, amino acids 62 to 78 within the cytoplasmic tail may also regulate association with ERM proteins.³⁰  These data prompted us to address whether phosphorylation at Ser76 affected CD43 exclusion from the immune synapse. Transduced T cells were conjugated with OVA_323-339-pulsed A20 cells, and assayed for CD43 localization.¹⁷  The CD43-Ser76Ala mutant is excluded from the immune synapse, and localizes to the distal pole similar to CD43-FL (Figure 3C). Furthermore, the number of conjugates that exclude CD43 is similar in cells expressing CD43-Ser76Ala or CD43-FL (Figure 3D). Thus, phosphorylation at Ser76 does not regulate CD43 exclusion from the immune synapse.

In addition to regulating T-cell functions upon activation, CD43 has also been shown to regulate T-cell trafficking. However, whether CD43 enhances or impairs T-cell migration remains controversial. One group has reported that CD43^−/− T cells show increased homing to secondary lymphoid organs, suggesting that CD43 negatively regulates T-cell trafficking.⁴⁴  In contrast, another group has demonstrated that blocking CD43 inhibits T-cell binding to high endothelial venules (HEVs), and homing to secondary lymphoid organs, arguing that CD43 positively regulates T-cell trafficking.⁴⁵  Thus, before we could investigate the role of phosphorylation at Ser76 in regulating T-cell trafficking, we needed to determine whether presence of CD43 enhanced or inhibited T-cell trafficking. We compared the ability of wild-type and CD43^−/− T cells to traffic to various tissues using a competitive in vivo trafficking assay. Rested, in vitro–expanded T cells were prepared from DO.11.10 (CD43^+/+) and DO.CD43^−/− (CD43^−/−) mice. CD43^+/+ and CD43^−/− T cells, easily distinguishable using CD43 staining (Figure 4A), were mixed in equal numbers and injected intravenously into naive BALB/c recipients. Donor T cells were distinguished from recipient T cells by staining for the transgenic DO.11.10 TCR. After 16 hours, blood, lymph nodes, and spleen were harvested, and analyzed for the frequency of donor CD43^+/+ and CD43^−/− CD4 T cells. We observed a similar frequency of donor CD43^−/− and CD43^+/+ T cells in the blood as well as in the spleen (Figure 4B bottom panels). Interestingly, in the lymph nodes there was a significant decrease in the frequency of CD43^−/− T cells compared with CD43^+/+ T cells. Hence, the ratio of CD43^−/− to CD43^+/+ T cells was significantly lower in the lymph nodes than in the blood and the spleen (Figure 4C).

To determine whether transduction of CD43^−/− T cells with wild-type CD43 would reconstitute normal T-cell trafficking to lymph nodes, CD43^−/− T cells were transduced with GFP-tagged wild-type CD43 (CD43-FL-GFP) or a control vector encoding GFP alone (EV-GFP). After transduction, resting T cells were harvested, mixed in equal numbers, and injected intravenously into naive BALB/c recipients. After 16 hours, donor T cells were enumerated in the blood, spleen, lung, and lymph nodes based on GFP and CD43 expression. Similar to Figure 4C, we find that the ratio of EV-transduced T cells to CD43-FL–transduced T cells is significantly lower in the lymph nodes compared with the blood (data not shown) and the spleen (Figure 4D). The lungs were also examined to determine if EV-transduced T cells exhibited selective tissue homing. However, there was no significant increase in the frequency of EV-transduced T cells in the lungs compared with the spleen, suggesting that the decreased lymph node homing of CD43^−/− T cells did not result from increased homing to tissues. We therefore concluded that presence of CD43 was specifically required for accumulation of T cells in lymph nodes, but not in lung or spleen. These findings confirm that transduction of CD43^−/− T cells can be used to study CD43-specific effects on T-cell trafficking, and together these data demonstrate that CD43 positively regulates trafficking of T cells to lymph nodes.

Homing of T cells into lymph nodes involves CD62L-mediated rolling and tethering on HEVs.⁴⁶  To test the hypothesis that CD43-mediated regulation of T-cell trafficking was secondary to an effect on CD62L expression, we examined CD62L levels on donor cells prior to injection. CD62L expression was similar between EV-GFP– and CD43-FL-GFP–transduced T cells (Figure 4E), suggesting that the differences in trafficking behavior were independent of CD62L expression. Nor did we observe any differences in expression of the integrins CD11a (LFA-1), β₁, and α₄β₇, or the activation markers CD25, CD69, and CD44 (data not shown).

Our results differ from those of Stockton et al, which showed that CD43 was a negative regulator of T-cell homing.⁴⁴  One difference between the studies is that while we used cells that had been previously activated, they used T cells ex vivo from the mouse. Thus, we repeated the in vivo trafficking assay described in Figure 4A-C using purified T cells ex vivo from wild-type DO.11.10 (CD43^+/+) or DO.CD43^−/− (CD43^−/−) mice. CD43^+/+ and CD43^−/− CD4 T cells were present in similar frequencies in the blood as well as the spleen (Figure 4F). Surprisingly, we found no difference in the frequency of CD43^+/+ and CD43^−/− T cells within the lymph nodes. Thus, ex vivo CD43^−/− CD4 T cells are not impaired in their ability to traffic to the lymph nodes. Together, our data suggest that CD43 positively regulates trafficking of memory-type CD4 T cells, but not naive CD4 T cells.

To determine whether phosphorylation at Ser76 was important for T-cell trafficking, we investigated the ability of CD43-Ser76Ala to restore trafficking of CD43^−/− T cells to lymph nodes. Using our previous staining scheme, we would not be able to distinguish CD43-FL–transduced cells from CD43-Ser76Ala–transduced cells if they were mixed together and injected (since both would be CD43⁺). Therefore, resting DO.CD43^−/− T cells transduced with either CD43-FL or CD43-Ser76Ala were injected separately into naive BALB/c recipients. In each case, the ratio of donor CD43⁺ (transduced) to CD43⁻ (untransduced) CD4 T cells was determined in different organs. For mice that received CD43-FL–transduced T cells, there is an increase in the ratio of donor CD43⁺ to CD43⁻ T cells in the lymph nodes compared with the blood (Figure 5A), confirming the data in Figure 4D that reconstitution with CD43-FL rescues the trafficking defect of CD43^−/− CD4 T cells. However, for mice that received CD43-Ser76Ala–transduced cells, the ratio of donor CD43⁺ to CD43⁻ cells in the lymph nodes was the same as that in the blood (Figure 5A), demonstrating that expression of the CD43-Ser76Ala mutant cannot restore trafficking of CD43^−/− T cells to the lymph nodes.

Entry of T cells into peripheral lymph nodes is regulated by expression of CD62L, the chemokine receptor CCR7, and the integrin LFA-1. To determine if phosphorylation at Ser76 regulates T-cell trafficking by altering surface expression of these molecules, we examined their levels on T cells transduced with CD43-FL (Figure 5B left panels) or CD43-Ser76Ala (Figure 5B right panels) prior to injection. T cells transduced with either construct express equivalent levels of CD43 (Figure 3A inset; data not shown). Most of the cells display a surface phenotype characteristic of circulating memory cells (CD44^hiCD62L^hi), whether they are transduced with CD43 (CD43⁺; Figure 5B bottom panels) or untransduced (CD43⁻; Figure 5B top panels). Similar levels of CD11a are found in all populations. We do observe a slight decrease in CCR7 expression in the CD43⁺ population in cells transduced with CD43-Ser76Ala compared with CD43-FL. However, there is no difference in the CD43⁻ and CD43⁺ populations in both cases, suggesting that differences in CCR7 expression does not account for the differences in trafficking behavior. Together, these data demonstrate that in contrast to T-cell activation and proliferation, phosphorylation at Ser76 plays an important role in CD43-mediated regulation of T-cell trafficking via a mechanism that is independent of expression of CD62L, CD11a, or CCR7.

While previous studies have highlighted an important role for the CD43 cytoplasmic domain in T-cell proliferation and CD43 exclusion,^14,17,20  the precise function of the CD43 ectodomain is yet to be determined. Several recent reports have shown interaction of CD43 with endogenous lectin-like molecules such as galectin-1 and E-selectin.^26,27,47  In light of these reports, and our data here showing that CD43 positively regulates T-cell trafficking, we hypothesized that interaction of the CD43 ectodomain with these or other ligands could regulate T-cell trafficking. To investigate this possibility, we used a previously published mutant CD43 construct CD16–7-43, wherein the CD43 extracellular and transmembrane domains have been replaced with those of human CD16 and CD7, respectively.²⁰  DO.CD43^−/− T cells were transduced with either CD43-FL or CD16–7-43. Resting transduced T cells were mixed in equal numbers, and injected intravenously into naive BALB/c recipients. The ratio of donor CD16⁺ to CD43⁺ CD4 T cells was determined in blood, spleen, and lymph nodes (Figure 6). The ratio of donor CD43⁺ to CD16⁺ CD4 T cells was similar in blood and spleen. However, there was a dramatic and significant decrease in the ratio of donor CD16⁺ to CD43⁺ CD4 T cells in the lymph nodes of recipient mice (Figure 6). Thus, in contrast to its function in regulating T-cell proliferation, the ectodomain of CD43 is required for T-cell trafficking to lymph nodes.

Over the years, a plethora of literature has suggested a signaling function for CD43, yet this study is the first to identify a specific phosphorylation site in CD43 that directly influences T-cell biology. Mass spectrometry and mutational analysis demonstrated that CD43 is specifically phosphorylated at Ser72 and Ser76, and this phosphorylation occurs in the absence of overt T-cell activation. Nevertheless, phosphorylation and dephosphorylation at these sites must be actively occurring since high levels of phosphorylation are detected in only 2 hours of radioactive orthophosphate labeling. These data suggest that either CD43 is interacting with ligands on other cells resulting in constitutive CD43 phosphorylation, or alternatively, basal CD43 phosphorylation is the result of activation of other receptors and intracellular signaling cascades.

Two important themes have emerged from our previous studies examining CD43 in T-cell immunity. First, the cellular localization of CD43 is highly regulated upon T-cell activation through interaction with ERM proteins, and CD43 localization to the distal pole after TCR stimulation forms a key component of CD43 functions.^16,17  Second, the ability of CD43 to negatively regulate T-cell activation and IL-2 production is mediated by signals through the intracellular domain, and is independent of its ectodomain.²⁰  Based on the dependence on the cytoplasmic tail for these CD43 functions, we hypothesized that they would be regulated by phosphorylation at Ser76. We were therefore surprised to find that phosphorylation at Ser76 was completely dispensable for regulation of T-cell proliferation and CD43 exclusion from the immune synapse. While Ser76 is undoubtedly a major site of phosphorylation within the CD43 cytoplasmic tail, our study and Piller et al show that other sites (especially Ser72) are phosphorylated (Figure 2; Piller et al³³ ). Thus, additional levels of regulation may exist within the CD43 cytoplasmic tail that contribute to CD43 movement and CD43 effects on T-cell proliferation, possibly through phosphorylation at alternate sites, including Ser72. Studies are under way to determine the role of phosphorylation at Ser72 in CD43-mediated regulation of T-cell functions.

We and others have defined CD43 as a negative regulatory molecule in T cells due to the observation that CD43^−/− T cells are hyperproliferative and produce more IL-2 than wild-type T cells. However, several groups have suggested that CD43 plays a positive role in T-cell migration during autoimmune and viral responses.^48-50  Although others have found contradictory evidence,^44,51  we have now clearly demonstrated that CD43 has a positive effect on T-cell trafficking to lymph nodes. It has been suggested that some of the phenotypes of CD43^−/− T cells may be secondary to background genetic differences between CD43 wild-type and CD43^−/− mice.⁵²  However, since transduction of CD43, but not a control vector, rescued the trafficking defect of CD43^−/− CD4 T cells, the potential role of background genes was eliminated.

We find that CD43 does not affect trafficking of ex vivo T cells; however, CD43 has a significant effect on the trafficking of rested, previously activated CD4 T cells. These cells are CD44^hiCD62L^hi, a surface phenotype characteristic of circulating memory cells. In fact, data from Swain's group have shown that when in vitro–generated CD4 effector T cells are further cultured in the absence of antigenic stimulation, they revert to a resting state, and display phenotypic and functional characteristics of in vivo generated memory cells.^53,54  Moreover, in adoptive transfer experiments, these cells have been shown to localize to the spleen and lymph nodes. Thus, it is possible that CD43 specifically regulates the trafficking of memory but not naive CD4 T cells.

The role of the CD43 ectodomain has remained elusive. Our past studies argued that the negative regulatory function of CD43 was mediated solely through the cytoplasmic tail, and we directly eliminated any role for the highly glycosylated ectodomain.²⁰  However, the recent emergence of literature on CD43 ligation by endogenous lectin-type molecules^24,26,27,55  has led us to reconsider the potential for direct signaling through CD43. Thus, our result that CD43-mediated regulation of T-cell trafficking to lymph nodes requires its ectodomain suggests a novel function for this domain of CD43.

A central question that emerges is the mechanism via which CD43 regulates T-cell trafficking into the lymph nodes. The decreased accumulation of CD43^−/− T cells in the lymph nodes could result from increased migration into nonlymphoid tissues, increased egress from the lymph nodes, or decreased entry into the lymph nodes. Since we do not observe increased frequency of CD43^−/− T cells compared with CD43^+/+ T cells in the lungs, it is unlikely that CD43^−/− T cells preferentially home to tissues. Another possibility is that CD43^−/− T cells are not retained in lymph nodes as well as CD43^+/+ T cells. S1P₁ regulates lymphocyte egress from lymph nodes,⁵⁶  and it is possible that CD43 may affect T-cell egress through a similar mechanism.

An important aspect of our findings is that CD43^−/− T cells are impaired in their ability to migrate to the lymph nodes, but not to the spleen. Entry into lymph nodes involves L-selectin–mediated tethering and rolling, followed by integrin–dependent firm arrest to the HEVs and subsequent transendothelial migration into lymph nodes.⁵⁷  In the spleen, which lack HEVs, lymphocytes exit the bloodstream in the marginal sinus and migrate into the white pulp. Unlike entry into lymph nodes, entry into spleen is not dependent on L-selectin, LFA-1, or α₄ integrins.⁵⁸  Thus, the hypothesis that we favor is that CD43 positively regulates T-cell entry into the lymph nodes. While our data show that CD43 does not influence the surface expression of CD62L or LFA-1, CD43 may regulate chemokine responsiveness and subsequent integrin activation. In fact, previous reports have shown that CD43 is an important regulator of β₁ and β₂ integrin–mediated adhesion. CD43^−/− T cells show increased binding to extracellular matrix ligands, and CD43^−/− CEM cells exhibit increased homotypic interactions.^13,21  Semmrich et al have shown that T cells expressing a constitutively active LFA-1 mutant show increased binding to ICAM-1 compared with wild-type T cells, yet are severely impaired in their ability to migrate due to impaired deadhesion at the trailing edge.⁵⁹  This gives rise to the exciting possibility that CD43, which is present in the uropod/trailing edge of migrating cells, regulates T-cell entry into lymph nodes by regulating integrin deactivation.

Our data that the CD43 ectodomain and Ser76 phosphorylation do not influence T-cell activation and proliferation but are required for regulation of trafficking provide a novel insight into CD43 function. First, CD43 has distinct functions in resting and activated T cells. Upon antigen stimulation, CD43 sets a threshold for T-cell activation and proliferation, while in resting T cells, CD43 sends tonic signals that regulate T-cell trafficking. Second, distinct structural domains of CD43 are involved in the regulation of these diverse functions. CD43-mediated attenuation of T-cell activation requires an intact cytoplasmic tail, but is independent of phosphorylation at Ser76 as well as its ectodomain. On the other hand, CD43-regulated T-cell trafficking requires both its extracellular and cytoplasmic domains. Taken together, our data reveal a complex and multifaceted role for CD43 regulation of T-cell function. The data we present here leads to the hypothesis that ligation of the CD43 ectodomain with endogenous lectinlike molecules results in phosphorylation at Ser76 within the CD43 cytoplasmic tail. This phosphorylation in turn sends downstream signals that result in regulation of T-cell trafficking. Our data suggest that CD43 functions as a signaling molecule, sensing extracellular cues and transducing intracellular signals that affect T-cell function.

Peptide synthesis, flow cytometry, mAb production, and digital light microscopy were conducted at the University of Chicago Cancer Research Center Core facilities (P30 CA 14599–28). Mass spectrometry was conducted at the University of Chicago Proteomics Core Laboratory.

This work was supported by National Institutes of Health (NIH) grants R01 AI44932 (to A.I.S.) and S06 GM08043 (to K.S.). J.L.C. is a postdoctoral fellow of the Cancer Research Institute.

National Institutes of Health

Contribution: P.D.M. and J.L.C. designed and performed research, analyzed data and wrote the manuscript. H.S.B., K.M.B. and A.B.S. helped design and perform experiments. K.S. and A.I.S. designed the research and analyzed data. P.D.M. and J.L.C. contributed equally to this manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Anne I. Sperling, Section of Pulmonary and Critical Care, Department of Medicine, University of Chicago, MC 6026, M624, 5841 S Maryland Ave, Chicago, IL 60637; e-mail: asperlin@uchicago.edu.