Age-Related Defects in Moesin/Ezrin Cytoskeletal Signals in Mouse CD4 T Cells¹

Cytoskeletal proteins participate in many aspects of T cell function, including formation of the immunological synapse and spreading of the T cell membrane (lamellipodia) adjacent to APC (1, 2, 3, 4). Recent models of immunosynapse formation suggest that the initiation and maintenance of the immunosynapse involves the interactions of surface molecules with the cytoskeleton (5, 6, 7). The ezrin-radixin-moesin (ERM)³ family of cytoskeletal proteins plays a key role in this remodeling, particularly with respect to the exclusion of glycoproteins (such as CD43) from the synapse area (4, 8, 9, 10, 11, 12, 13, 14). The ERM family is also critical to other aspects of T cell function, including cell adhesion, formation of lamellipodia, and membrane ruffling (14, 15). ERM action often involves linkages between cell surface molecules (including CD44, CD43, and P-glycoprotein or multidrug-resistant (MDR)-1 and adaptor proteins such as ezrin binding protein (EBP)50 with the plasma membrane (3, 16, 17, 18, 19, 20, 21, 22). The regulation of these interactions is thought to involve the phosphorylation of ERM (phospho-ERM), specifically on Thr⁵⁵⁸ of moesin or Thr⁵⁶⁷ of ezrin, leading to opening of the FERM domain in the ERM molecules and therefore to interaction with surface molecules and adaptor proteins (23, 24). In resting T cells from young mice, ERM proteins are usually phosphorylated (phospho-ERM), and dephosphorylation is induced by CD3 ligation and APC contact (6, 9). The level or status of ERM phosphorylation may be important in the breakdown of microvilli, spreading of the cell membrane over adjacent surfaces, formation of lamellipodia, and polarization of the cell, including asymmetric distribution of internal and surface macromolecules (6, 25, 26) during the formation of the immunological synapse (2, 6, 9, 12, 18, 21, 23, 24). During aging there is a decline in the proportion of CD4 T cells that can translocate surface molecules and signaling adaptor proteins to the area of T cell-APC contact (27, 28, 29, 30). Those T cells that are unable to form synapses are also unable to exclude CD43 from the area of APC contact. Aging also leads to defects in the association of TCR (CD3ζ) and CD43 to the cytoskeletal network of mouse CD4 T cells (28, 29). We hypothesized that age-related alteration in cytoskeletal reorganization might be responsible for the defects in translocation of proteins during immunosynapse formation. In one series of experiments, we treated CD4 T cells with an O-sialoglycoprotein endopeptidase (OSGE), a bacterial enzyme that cleaves surface proteins (including CD43) that contain O-linked glycans. We found that OSGE treatment of CD4 T cells from aged mice fully restores the translocation of signaling molecules, including the TCR, phospholipase Cγ, LAT (linker for activation of T cells), and Grb-2, to the immunosynapse (29, 31, 32). Treatment with OSGE did not, however, significantly repair the age-dependent decline in exclusion of CD43 from the area of T cell-APC contact, even in cells that formed a functional immunosynapse after OSGE treatment (29). Because exclusion of CD43 from the synapse is dependent on ERM function (8, 9, 33), we hypothesized that age-related defects in the ERM pathways might impair activation of aged T cells independent of the role of O-glycan containing surface proteins (29, 34). In this study, we document age-dependent defects in activation of ERM cytoskeletal proteins of CD4 T cells, and present data to suggest that some of the upstream signaling pathways of the Rho family of GTPases, known to play important roles in regulation of the ERM pathways in T cells, are also affected by aging.

Breeding pairs of the AND line of TCR transgenic mice, whose T cells respond to pigeon cytochrome c, were a generous gift from Drs. S. Swain and L. Haynes (Trudeau Institute, Saranac Lake, NY). Transgene-positive mice were aged in a specific pathogen-free colony at the University of Michigan (Ann Arbor, MI). Specific pathogen-free male (BALB/c × C57BL/6)F₁ (CB6F₁) mice were purchased from the National Institute of Aging contract colonies at the Charles River Breeding Laboratories and at Harlan Breeders. Mice were given free access to food and water. Sentinel animals were examined quarterly for serological evidence of viral infection; all such tests were negative during the course of these studies. Mice that were found to have splenomegaly or macroscopically visible tumors at the time of sacrifice were not used for experiments. CB6F₁ mice were used at age 6–8 mo (young) or 22–24 mo (old); AND mice were used at 6–8 or 14–16 mo of age.

Goat anti-CD43 (M19), rabbit anti-MDR, and mAb anti-CD45 (clone MB4B4) were purchased from Santa Cruz Biotechnology. Rabbit anti-RhoA, anti-Rac1, anti-phospho-ERM (anti-phospho-ezrin (Thr⁵⁶⁷)/phospho-moesin (Thr⁵⁵⁸)), anti-moesin, anti-LAT, and anti-phospho-LAT (Tyr¹⁹¹) came from Cell Signaling Technology. Rabbit anti-CD44 was purchased from Abcam. Monoclonal Ab anti-ezrin (clone 3C12) was purchased from Sigma-Aldrich. Monoclonal Ab anti-moesin (clone 38/87) was from NeoMarkers. Rabbit anti-EBP50 was obtained from Affinity BioReagents. The mAbs for flow cytometry and CD44 (clone IM7) were purchased from BD Biosciences. GST-Rhotekin and GST-Pak1 fusion proteins were acquired from Pierce. CD4⁺ T cells were obtained and stimulated from transgenic (35) or CB6F₁ (30) mice using the negative selection methods previously described. Flow cytometric analysis of a typical preparation showed it to be 90–95% positive for both CD3 and CD4.

Approximately 10 × 10⁶ CD4⁺ T cells were treated with OSGE and stimulated for 5 min by cross-linking CD3ε with CD4 and CD28 as described previously (29, 32); controls included cells that were not treated with OSGE and others that were not activated. The cells were lysed in 1 ml of 1% Brij-58, PBS (pH 7.4), 1 μg/ml aprotinin, 1 μg/ml leupeptin, 1 mM PMSF, and 1 mM sodium orthovanadate for 30 min on ice. The lysates were ultracentrifuged at 10,000 × g for 1 h at 4°C. The pellet (“Brij-58-resistant fraction”) was washed once with lysis buffer, centrifuged at 10,000 × g for 10 min at 4°C, resuspended in 50 μl of SDS sample buffer, and boiled for 15 min. The supernatant (“Brij-58-soluble fraction”) was used for total cell lysate analysis, for immunoprecipitation (for 6 h) with specific Abs precoupled to protein G-Sepharose as previously described (28, 35), or for the RhoA/Rac pulldown assay using 25 μg/sample of GST-Rhotekin or GST-Pak1 fusion protein as previously described (36). Samples were fractionated in polyacrylamide gels, transferred to polyvinylidene difluoride, and incubated with specific Abs as described; the bands were visualized by chemifluorescence and quantified as previously described (35).

Intracellular staining of T cells was performed according to a procedure recommended by BD Biosciences. Briefly, purified T cells were stained using Cy5-labeled anti-CD4 and PE-labeled CD44, then fixed and treated with methanol at −20°C. After two washes, the cells were stained with rabbit polyclonal Abs specific for the phosphorylated or unphosphorylated forms of the protein of interest, followed by goat anti-rabbit FITC. The cells were analyzed on a BD FACSCalibur, with gating for CD4 T cells.

Unless otherwise indicated, results are presented as the mean ± SEM. Statistical significance was assessed using a Mann-Whitney U test at a value for p = 0.05.

We and others have shown that some membrane-associated proteins can be resistant to Brij-58 detergent extraction under specific conditions (28, 29, 37, 38). We evaluated the effects of Brij extraction on ERM protein distribution in freshly isolated CD4 T cells. As shown in Fig. 1, CD44, CD45, MDR-1, EBP50, and the majority of the moesin and ezrin are solubilized by Brij-58 treatment. In contrast, CD3ζ (p21ζ) can be found distributed between both soluble and resistant fractions. Furthermore, TCR stimulation of CD4 T cells from young mice increases the amount of CD3ζ associated with the Brij-58-resistant fraction, but does not have any significant effect on distribution of CD3ζ in CD4 T cells from aged mice. In addition, CD43 is divided between the detergent-resistant and detergent-soluble fractions. TCR stimulation of CD4 T cells from young mice increases the amount of CD43 found in the detergent-resistant fraction, while decreasing the amount in the soluble fraction. However, stimulation does not have this effect on CD43 of CD4 T cells from old donors. The age-related change in CD3ζ distribution and the lack of response of CD43 in old donors are in good agreement with our previously published data (29). In contrast to its effects on CD43 and CD3ζ localization, TCR stimulation of freshly isolated CD4 T cells does not alter the extractability of molecules that are known or hypothesized to be excluded from the immunosynapse, including ERM, CD44, and CD45. These findings are in line with results of other studies on T cells (39, 40, 41, 42). The nature of this detergent resistance to extraction is not understood, but we hypothesize that cytoskeleton or cortical actin interaction with membrane proteins may play a role. Therefore alteration in the Brij-58 extraction may reflect age-related changes in the interactions between surface molecules and proteins of the cytoskeleton.

During the formation of the immunological synapse, the dephosphorylation and rephosphorylation of the ERM participates in the exclusion of surface molecules from the T cell-APC contact area (6, 7, 8, 9). This process can be partially mimicked by anti-CD3 stimulation, which triggers ERM dephosphorylation. As shown in Fig. 1 (p-ERM), anti-CD3 stimulation induces a decline in the ERM phosphorylation in CD4 T cells from young donors; however, ERM phosphorylation of CD4 T cells from old donors does not display such a change (for a more detailed study see Fig. 3). These results suggest that alterations in ERM phosphorylation may be in part responsible for the lack of CD43 exclusion in the course of T cell-APC contact.

During T cell-APC interaction there is an age-related decline in the number of CD4 T cells that can exclude CD43 from the T cell-APC contact area. The decline in CD43 exclusion can also be found in CD4 T cells from old donors that do form efficient immunosynapses (29). Treatment of old CD4 T cells with OSGE restores to youthful levels the fraction of cells that can form synapses, but does not increase the fraction that excludes CD43, suggesting that aged T cells might have a defect in cytoskeletal pathways required for CD43 translocation.

Because ERM/CD43 interaction plays a role in the exclusion of CD43 (6, 8, 9, 22), we postulated that age could alter CD43 association with ERM molecules. To test this hypothesis, we immunoprecipitated CD43 using a goat Ab that recognizes an internal domain of CD43 (M19) and analyzed the amount of moesin/ezrin associated with CD43. Fig. 2 shows a typical Western blot of moesin (using the mAb 38/87 that also recognizes ezrin) after CD43 has been precipitated from CD4 T cells varying in donor age, treatment with OSGE, and stimulation through the TCR. None of the experiments showed any significant amount of ezrin associated with CD43. The graph in Fig. 2 shows that unstimulated CD4 T cells from old donors have 2-fold higher levels of moesin associated with CD43 (p = 0.031) compared with young cells. As expected, stimulation of CD4 T cells from young donors leads to a significant decline in the amount of moesin associated with CD43 (p = 0.036) (6, 8, 10, 11). This decline is not seen in CD4 T cells from old donors. Western blot analysis of CD43 levels in the same lysates showed no significant effect of either age or stimulation (Fig. 2, top; note that OSGE treatment reduces the m.w. of the CD43 molecule). The age-related changes in CD43-moesin association and the lack of CD43 exclusion from the immunosynapse in CD4 T cells from old donors (29) suggest that age might alter pathways regulating the function of ERM proteins. Because OSGE treatment restores the function of T cells from aged mice, we evaluated the effects of OSGE on CD43-moesin association. As shown in Fig. 2, CD4 T cells from young donors show no significant effect of the OSGE treatment, consistent with the lack of effect of OSGE on immunosynapse formation of young CD4 T cells (29). In contrast, OSGE treatment of CD4 T cells from old donors, before stimulation, leads to a statistically significant decline in CD43-moesin association (p = 0.004) compared with stimulated aged cells not exposed to OSGE. OSGE treatment also leads to a small decrease in CD43-moesin association in resting CD4 T cells from old mice, but this effect does not reach statistical significance (p = 0.07). Taken together, these results suggest that OSGE treatment of T cells from aged mice can enhance some aspects of TCR signaling as related to the ERM pathway. OSGE treatment, however, does not fully restore the defects in CD43-moesin association, suggesting the presence of age-related defects in the ERM or in its signaling pathways that are independent of immunosynapse formation.

Because interactions between ERM and other proteins are regulated by phosphorylation, specifically Thr⁵⁵⁸ in moesin and Thr⁵⁶⁷ in ezrin (43, 44), we measured the levels of phospho-ERM in resting and stimulated CD4 T cells from young and old CB6F₁ donors, with and without OSGE treatment to restore synapse formation. The amount of phospho-ERM of each sample was assessed by Western blot analysis using an Ab that recognizes both phospho-Thr⁵⁵⁸ of moesin (p-moesin) and phospho-Thr⁵⁶⁷ of ezrin (p-ezrin). Fig. 3 shows a typical digital image of a phospho-moesin blot, with corresponding levels of total moesin (using mAb clone 37). Ezrin and phospho-ezrin are also present in the original images in Fig. 1, but at lower levels of detection in samples from old donors, making quantification difficult (data not shown). The graph in Fig. 3 presents the ratio of phospho-moesin to moesin from five independent experiments. Resting CD4 T cells from old mice have only half the level of moesin phosphorylation (p = 0.0025) seen in cells from young controls. Stimulation of CD4 T cells from young mice induces a significant dephosphorylation of the moesin (p = 0.003), in good agreement with published reports (6, 9, 16). In contrast, CD4 T cells from old donors do not show any decline in moesin phosphorylation after TCR stimulation. OSGE treatment enhances the dephosphorylation of moesin in CD4 T cells from old donors (p = 0.047), consistent with our previous observations that OSGE has a preferential effect on some aspects of early T cell activation in T cells from old mice (29, 32, 34), without a parallel effect on young CD4 T cells.

To examine whether age effects on moesin phosphorylation reflect the shift from naive to memory CD4 T cells with aging, we used flow cytometry to evaluate the ERM phosphorylation status using phospho-specific Abs, together with anti-CD44 as a marker to distinguish naive from memory cells. As controls, some aliquots were stained with anti-phospho-LAT (rabbit anti-phospho-Tyr¹⁹¹) and anti-LAT (rabbit polyclonal) Abs. Fig. 4 shows typical distributions of phospho-moesin and moesin in populations gated either for all CD4 T cells or for CD4 naive and memory subsets. Because the phospho-moesin Ab can also recognize phospho-ezrin, we refer to the signal in Fig. 4 as P-ERM. Both naive and memory CD4 T cells show an age-dependent decline in phospho-ERM, but little or no change in total moesin levels. A series of four experiments showed a significant decline with age in cellular levels of phospho-moesin (p = 0.0001) (Fig. 4), consistent with the immunoblotting data, but no change in the level of moesin itself. As in a previous report (30), we saw no significant age-dependent change in the level of LAT or phospho-LAT (Fig. 4), suggesting that the data on phospho-moesin does not reflect changes in fixation or other staining artifacts. We also obtained results similar to the LAT data using Abs to Lck and phospho-Tyr⁵⁰⁵ Lck (data not shown).

The decline with age in ERM phosphorylation in resting CD4 T cells could result in changes in the interaction of ERM proteins with T cell surface molecules and intracellular signaling proteins. To further test this hypothesis, we immunoprecipitated CD44, a surface glycoprotein known to be associated with ERM (22), and measured the amount of associated ezrin and moesin. As shown in Fig. 5 A, we found significant amounts of both ezrin and moesin associated with CD44; this is in contrast to the CD43 immunoprecipitates, in which only moesin was easily detectable. The image, showing results from two young and two old mice, suggests an age-dependent decline in the levels of both moesin and ezrin associated with CD44. Statistical analysis of four independent experiments showed a significant age-related decline in the amount of moesin associated with CD44 (p = 0.003). Ezrin showed a similar age-related decline, but the low levels of ezrin in resting CD4 T cells from old donors precluded its accurate quantification and statistical analysis.

ERM proteins can also interact with EBP50, an adaptor protein involved in the regulation of the Csk tyrosine kinase and TCR signaling (7, 45, 46, 47). Fig. 5,B shows a typical image of the moesin and ezrin levels in EBP50 immunoprecipitates from CD4 T cells of young and old donors. Analysis of four independent experiments showed a significant age-related decline in the amount of moesin associated with EBP50 (p = 0.0004). As in the case of CD44, the lower level of ezrin in samples of old mice precluded accurate quantification. In addition, we found no evidence for age-associated change in the total amount of T cell moesin (Figs. 4 and 5), ezrin (see Fig. 7), or EBP50 (data not shown) that could explain the age-related changes.

It has been reported (17, 48, 49, 50, 51, 52) that many aspects of cytoskeletal reorganization during TCR signaling, including the phosphorylation status of ERM proteins, are under the control of members of the Rho family of GTPases (RhoA, Rac, cdc42). Although details of the regulatory pathways are not yet well understood, it has been shown that RhoA and Rac may be upstream regulators of ERM phosphorylation (17, 25, 51, 52, 53, 54).

To determine whether age-related changes in GTPase activity of RhoA and Rac proteins could be implicated in the decline in ERM phosphorylation, we measured both the total RhoA protein level and the amount of active RhoA in CD4 T cells from young and old CB6F₁ mice. Fig. 6 shows typical experimental results for total RhoA and active RhoA in CB6F₁ CD4 T cells, as well as mean values for 12 young and six old donors from six separate experiments. We found no significant age-related change in the expression of total RhoA (Fig. 6,A). Using an assay for active RhoA based upon a GST-Rhotekin fusion protein construct that binds RhoA only when it is in its GTP-bound configuration, we found (Fig. 6,B) a significant age-related decline in the amount of active RhoA (p = 0.0001). To determine whether the shift from the naive to memory phenotype with age could explain the alteration in RhoA activity, we evaluated CD4 T cells from AND mice; flow cytometric analysis of CD44 expression (data not shown) indicated that ∼80% of the CD4 T cells in these mice were naive at the age of 15 mo. Fig. 6 C shows a significant decline in active RhoA in CD4 T cells from old AND mice (p = 0.028). These results confirm that, at least in the naive phenotype, there is an age effect on RhoA activity.

Using similar methods, we also evaluated the effect of aging on Rac levels and activity. Fig. 7 shows typical experimental results for total Rac and active Rac in CB6F₁ CD4 T cells, as well as mean values for seven young and six old donors from three separate experiments. As in the case of RhoA, we found no statistically significant difference in the level of Rac expression between young and old mice (see Fig. 7,A). We used a GST-Pak1 construct to measure Rac activity (Fig. 7,B), and noted an age-dependent increase in Rac function. Analysis of results from seven young and six old mice examined in three separate experiments revealed a 50% increase in Rac activity (p = 0.0012). To see whether these changes were due to a shift from a naive to a memory phenotype with age, we analyzed Rac activity in the AND transgenic mice from six young and three old donors and found a statistically significant 3-fold increase in active Rac (p = 0.023) without any change in total Rac expression (Fig. 7 C). These results confirm that naive CD4 cells, at least, show age effects in activity of both RhoA and Rac, within the family of Rho GTPases that regulate ERM cytoskeletal proteins.

The current model of immunosynapse formation suggests that the initial interactions between T cells and APCs induce formation of lamellipodia by relaxation of the T cell membrane and reorganization of the cytoskeletal network. These changes include increases in F-actin polymerization and translocation of TCR signaling molecules to the area of T cell contact with the APC, as well as exclusion of CD43 from the area of interactions. The experimental evidence shows that ERM proteins participate in the exclusion of CD43 and contribute to negative regulation of TCR signaling (6, 9, 16). These transitions seem to involve the dephosphorylation and rephosphorylation of ERM proteins and physical segregation of ERM, with its associated molecules including CD43 and CD44, away from the F-actin network in areas of microclusters and at the T cell APC interaction area (9, 16). The Brij-58 extraction experiments in Fig. 1 show that the ERM protein, as well as the ERM-associated molecules CD44, MDR-1, and EBP50, are largely detergent soluble. However, the distribution of CD43 between the detergent-soluble and detergent-resistant fraction (Fig. 1) suggests that extracellular or intracellular domains of CD43 may interact with other surface proteins or cytoskeleton proteins in an ERM-independent manner. The basis for CD43 resistance to detergent extraction is not known, but we hypothesize that this insoluble fraction may be associated with the F-actin cytoskeleton in freshly isolated CD4 T cells from young and aged mice.

If lamellipodia formation is dependent on the ERM (16), the age-related changes in ERM phosphorylation shown in Figs. 3 and 4 suggest a model in which age-related alterations in ERM signaling pathways are involved in the defects of lamellipodia formation found in CD4 T cells from old mice. In addition, the immunoprecipitation data shown in Fig. 5 suggest age-related alterations in the association of proteins with the ERM: CD4 T cells from old mice show significant declines in the association of CD44 and EBP50 with ERM proteins. Altered ERM association with CD44 and EBP50 might be attributable to changes in phosphorylation of ERM sites that trigger its open conformation. In good agreement with this hypothesis, we found no age-related alteration in the levels of moesin and ezrin proteins in CD4 T cells, but a significant decline in ERM phosphorylation detected both by immunoblotting and flow cytometry (see Figs. 3 and 4). A decline in the proportion of moesin and ezrin able to interact with CD44, EBP40, and other regulators could contribute to age-dependent declines in T cell activation. These changes appear to be independent of the naive to memory cell shift seen in aging mice.

Previous work on CD43 biochemistry in T cell lines and from structural studies would predict that age-related changes in moesin phosphorylation would lead to coordinated changes in moesin association with CD43, CD44, and EPB50. Our data, based on CD4 T cells freshly isolated from mice, are in good agreement with respect to CD44 and EBP50, but are at odds with the prediction with respect to CD43. There are, however, several aspects of ERM biology in freshly isolated CD4 cells that do not conform to previous reports on cell lines, including unexpectedly low levels of moesin, and virtual absence of ezrin, in CD43 immunoprecipitates (compared with CD44 immunoprecipitates). In addition, we note that CD43 is associated, partly, with the Brij-58 insoluble fraction, whereas CD44 and EPB50 are entirely Brij-soluble. We do not have a convincing molecular model that explains differences in CD43 coprecipitation from the data on CD44 and EPB50 coprecipitation, but these data do suggest that models that predict parallel behavior of CD43, CD44, and EBP50 may be incomplete with respect to freshly isolated T cells. It thus seems possible that an ERM-independent mechanism may contribute to the exclusion or retention of CD43 and other TCR-negative regulator molecules from the immunosynapse in CD4 T cells from old donors. Lastly, we can point to the previously reported age-related increase in glycosylation of CD43 (34), and can suggest that this age-dependent alteration might lead to increased association between CD43 and other moesin-associated surface proteins.

Studies using T cell lines and normal T cells from young donors are only gradually beginning to clarify relationships among kinases and phosphatases that control ERM phosphorylation. At present we do not know the molecular mechanism responsible for the age-related decline of ERM phosphorylation, but there is evidence that GTPases of the Rho family are among the upstream regulators of the ERM phosphorylation, with several studies showing that the Rho family of molecules, including Rac and RhoA, are critical for regulation of ERM signaling and lamellipodia formation (15, 17, 48, 50, 51, 52, 53, 54). It is possible that the age-related changes in activation of one or more Rho family GTPases may contribute to defects in ERM activation and changes in ERM phosphorylation. CD4 T cells from old mice show no significant change in the expression levels of RhoA and Rac, but do show a significant decline in the level of active RhoA (Fig. 6,B) and a significant increase in active Rac (Fig. 7,B). It is not likely that these changes are the result of the age-related shift from naive to memory cells because we found similar age-related declines in active RhoA and increases in active Rac using the pigeon cytochrome c-specific AND mouse strain, in which most of the CD4 T cells remain in a naive stage even late in life (55, 56). Studies using transfected cell lines suggest that increases in the activity of Rac are directly linked with a decline in ERM phosphorylation (53). We found this link to be the case for CD4 T cells from old donors (Figs. 3 and 7). This, as well as the decline in active RhoA, could well contribute to age-related deficits in lamellipodia formation and polarization of T cells and to the exclusion of CD43 and other proteins from the TCR immunosynapse. However, additional studies are needed to directly test whether GTPases are a key element in the age-related decline of CD4 T cell function.

We thank Lynn Winkelman, Maggie Vergara, Jessica Sewald, Bill Kohler, and Melissa Han for technical assistance.

The authors have no financial conflict of interest.