T cell activation is a complex process dependent on three sequential signals, the first comprises antigen/MHC recognition by the TCR/CD3 complex; the second involves the generation of a costimulatory signal induced by CD28 after CD80/CD86 recognition in antigen presenting cells (APC); and the third consists of the effect of cytokines that maintain cell survival, expansion, and polarization like IL-2, IFN-γ and IL-4 (1–3). This process begins with the immunological synapse formation, the contact interface between the T cell and the APC, where diverse proteins converge to promote stimulation, adhesion, communication, and signaling, many of which are highly dependent on the glycosidic portion of glycoproteins (4–6). A proper activation is highly dependent on costimulation since this second signal enhances the one induced by antigen recognition (7–9) and CD28 remains as the canonical costimulatory molecule in T cells (10). For example, although LFA-1 and CD2 have been proposed as alternative costimulatory molecules, the stimulus induced through these molecules is not strong enough to endure activation and allow adequate T cell effector functions. Other molecules like ICOS and OX40 appear once cell activation is complete and their costimulation provides survival and maintenance signals (11–14).

Earlier reports from our group showed that mouse and human CD4⁺ T cells stimulated in vitro with anti-CD3 and the Amaranthus leucocarpus lectin (ALL) activate and proliferate similarly to anti-CD3/CD28 activated cells (15, 16), suggesting that other molecules can generate initial costimulatory signals and, potentially, the existence of an alternate T cell co-activation mechanism. ALL identifies O-glycoproteins with a Galβ1-3GalNAc-O-Ser/Thr sequence, even in the presence of sialic acid residues in their structure (17). In mouse thymus, it recognizes CD4⁺ T cells in the medulla and the cortico-medullary junction, suggesting that the glycoproteins identified by ALL (gpALL) are expressed until T cells mature (18, 19).

Additionally, one of the molecules recognized by ALL in circulating mouse CD4⁺ T cells is a 70-kDa membrane O-glycoprotein that belongs to the ERM family (ezrin, radixin, and moesin) with high sequence homology to moesin (20). Moesin is mainly expressed in endothelial cells, platelets, T lymphocytes and to a lesser extent in some types of epithelial cells, where ezrin predominates (21, 22). It has been classically described as a non-glycosylated cytosolic protein involved in cell polarity maintenance and the interaction between membrane proteins with the cytoskeleton (23, 24); however, a few studies report its presence on the cell surface (25, 26). Moesin is a 67.8-kDa protein, composed of 577 amino acids, and its structure is comprised of a FERM or amino-terminal domain, a central or alpha-helical domain, and a moesin or carboxy-terminal domain (27); its cytosolic activity is regulated by phosphorylation of the conserved amino acid Thr⁵⁵⁸, which induces the protein to adopt an active conformation (28, 29). This protein is determinant for T cell homeostasis and function since moesin deficient mice show a reduced number of T cells in the periphery and lymph nodes (30, 31), and other studies have shown that suppressing moesin expression with siRNA hampers IL-2 production by T cells after in vitro stimulation (32).

Expanding our knowledge of costimulatory molecules in T cells is of great interest because they have become crucial targets for the treatment of different illnesses like cancer and autoimmune diseases (33, 34), turning any new molecule or mechanism involved in T cell activation into a potential clinical target. Therefore, in this study, we aimed to determine gpALL and moesin expression dynamics in CD4⁺ and CD8⁺ T cells after in vitro activation and if the 70-kDa glycoprotein detected by ALL is an O-glycosylated form of moesin. Finally, we evaluated if in vitro stimulation with anti-CD3 and anti-moesin could induce activation and proliferation of CD4⁺ and CD8⁺ T lymphocytes.

We first performed a comparative analysis of gpALL expression kinetics by flow cytometry in CD4⁺ T and CD8⁺ T lymphocytes stimulated for 24 h, 48 h, and 72 h with anti-CD3 and anti-CD28 mAbs ( Figure 1 ). Results show that 30% of unstimulated CD4⁺ T lymphocytes are ALL⁺, this percentage doubles 24 h post-activation and reaches its maximum after 48 h, when 90% of CD4⁺ cells are recognized by ALL ( Figures 1A, C ). In contrast, 88% of not activated CD8⁺ lymphocytes already express gpALL and peak positivity occurs at 48h post-activation where 98.8% of cells are ALL⁺ ( Figures 1B, C ); in both cases, a decrease of ALL binding is observed at 72 h ( Figures 1A, B, D ). Despite the difference in cell percentage positivity, gpALL expression was equivalent in both unstimulated T cell subsets, and in both cases, peak expression was reached at 48 h post-activation; gpALL expression increased 4.7 times in CD4⁺ cells ( Figures 1A, D ) and 4.3 times in CD8⁺ cells ( Figures 1B, D ). Although gpALL expression increases in both cell subsets as a consequence of activation, almost 90% of unstimulated CD8⁺ cells are ALL⁺; showing that gpALL is constitutively expressed in this cell subset but inducible in CD4⁺ cells ( Figure 1 ).

To further analyze these differences, we evaluated the O-glycoprotein profile recognized by ALL in unstimulated and 48 h stimulated cells with anti-CD3/CD28 by lectin blot. These results ( Figures 2A, C ) revealed that in both CD4⁺ and CD8⁺ T cells the number of glycoproteins and the intensity of the bands recognized by ALL increased after 48 h of activation compared to the gpALL expressed in unstimulated cells ( Figures 2A–C ). ALL recognized O-glycoproteins with molecular weights that vary between 20 and 150-kDa in both stimulated and unstimulated cells; after stimulation, over twice the bands were detected in CD4⁺ cells (Unstimulated 7 vs Stimulated 15), while 5 additional bands were detected in CD8⁺ cells (unstimulated 11 vs Stimulated 16) ( Figure 2A ). Quantitative analysis of the glycoproteins recognized by ALL ( Figure 2C ), showed that the mean gpALL expression level increased 1.1 times after activation in CD4⁺ T cells and 0.8 times in CD8⁺ T cells compared to unstimulated samples. Within these, the 70-kDa protein ( Figure 2A , arrow) is of particular interest for us because its expression has been previously reported in T cells from the thymus, peripheral blood, lymph nodes, and spleen (18). Herein we found an increased expression of the 70-kDa protein in CD4⁺ and CD8⁺ T cells after 48 h stimulation (1.2 and 0.9 times, respectively. Figure 2A , arrows, and 2D ).

This 70-kDa protein has been identified as an ERM family member with high sequence homology to moesin (20), thus, we evaluated the membrane surface expression kinetics of moesin by flow cytometry. As can be seen in Figures 3A–C , some unstimulated CD4⁺ and CD8⁺ T cells were moesin⁺ (19.2 and 10.25%, respectively), these percentages increased up to 54.8% in CD4⁺ cells and 50.2% in CD8⁺ cells 48 h after stimulation. Consequently, membrane moesin expression increased slightly in both cell subsets after stimulation, up to 1.8 times in CD4⁺ T cells at 72 h and 1.4 times in CD8⁺ T at 24 h ( Figures 3A, B, D ). These results demonstrate that moesin is expressed in the cell membrane and the expression kinetics is equivalent in CD4⁺ and CD8⁺ cells after activation.

Arenas et al. also showed that this moesin-like protein co-localized with gpALL on the cell surface of CD4⁺ lymphocytes after activation (20). Thus, to demonstrate that an O-glycosylated form of moesin (O-moesin) is expressed in both CD4⁺ and CD8⁺ cell subsets, we performed precipitation and cross-recognition assays using an anti-moesin mAb and ALL in protein extracts from 48 h activated CD4⁺ and CD8⁺ T lymphocytes. First, we immunoprecipitated the proteins with anti-moesin, and performed a lectin-blot using ALL and a western-blot developing with anti-moesin ( Figure 4A ), results showed that ALL recognizes 5 proteins with equivalent molecular weights in both cell subsets (150, 100, 80, 70 and 60-kDa) and an additional 45-kDa protein in CD4⁺ cells only. The 70-kDa band was the most prominent in both cell subsets and corresponds to the only band observed in the anti-moesin western blot. In parallel, ALL precipitated proteins were analyzed in a western blot developed with anti-moesin, where a single 70-kDa band is revealed in both cell subsets and a lectin blot with ALL ( Figure 4B ). The latter showed the presence of 5 bands in CD4⁺ T lymphocytes and 4 bands in CD8⁺ T, including the 70-kDa O-glycoprotein. These results show that ALL recognizes an O-moesin expressed in CD4⁺ and CD8⁺ T lymphocytes, the detection of additional molecules in the lectin blots, suggests that these molecules could have been isolated as part of a larger molecular complex.

Up to date, no reports describe the presence of a glycosylated moesin form. Prediction of possible O-glycosylation sites using NetOGlyc 4.0 server ( Supplementary Figure 2 ) showed that among the 577 amino acids that integrate moesin, only 3.8% of them have > 50% probability to be Ser/Thr O-glycosylated: 1 at the FERM domain, 13 at the alpha-helical domain, and 8 at the moesin domain. Only 6 sites have > 80% probability of being O-glycosylated, within this group Thr⁴⁶⁹ has the higher probability of modification (95%) but Thr⁵⁵⁸ (84%), located at the moesin domain, is particularly interesting because it has been reported as a fundamental site for the activation and inhibition of intracellular moesin (23).

Given that ALL can provide a costimulatory signal for T cells similar to the one provided by anti-CD28 in the presence of anti-CD3 in vitro and that we have demonstrated herein that O-moesin is recognized by this lectin, we aimed to determine if we could accomplish a similar activation profile after moesin costimulation. To this end, purified T lymphocytes were cultured in the presence of anti-CD3 and anti-moesin for 72 h, and different activation parameters were evaluated: CD69 and CD25 expression, cell proliferation, and IL-2 production.

Figure 5A shows a representative CD69 and CD25 expression analysis in both T cell subsets 72 h after activation. Analysis of CD4⁺ cells stimulated with CD3/moesin showed that 3.87% had a very early activated phenotype (CD69⁺CD25^-), 10.5% had a mid-activation phenotype (CD69⁺CD25⁺) and 11.4% showed a late activation phenotype (CD69^-CD25⁺). Among CD3/Moesin activated CD8⁺ T cells 24.6% where at the earliest activation point (CD69⁺CD25^-), 42% were mid-activated (CD69⁺CD25⁺) and 3.3% were at the latest activation stage (CD69^-CD25⁺). In both cases, the comparison with control cells activated with CD3/ALL and CD3/CD28 showed a similar pattern to those observed in cells activated with CD3/moesin ( Figures 5A, B ); no differences were found when CD69 and CD25 expression was analyzed 6 h post activation ( Supplementary Figure 3 ). These experiments demonstrate that moesin is a costimulator as effective as CD28.

Cell proliferation analysis ( Figure 6 ) revealed that after CD3/moesin stimulation CD4⁺ cells divided 4 times and CD8⁺ cells divided 5 times. Statistical analysis of the percentage of divided cells ( Figure 6B ) showed that 3.1% of CD4⁺ and 18.13% of CD8⁺ cells from the original population were able to divide at least once (37). A similar proliferation pattern was observed in control cells stimulated with CD3/ALL and CD3/CD28, although proliferation was slightly more efficient in the latter condition. This data demonstrates that CD3/moesin activated T cells show a classic mitogen-induced cell proliferation pattern, including the faster division rate observed in CD8⁺ cells. Finally, we evaluated IL-2 production ( Figure 6C ) in CD3/moesin stimulated T cells, where 91.33 pg/ml were detected, 3.7 times more than unstimulated cells (25.33 pg/ml), slightly but not significantly more than CD3/ALL stimulated cells (77.26 pg/ml) and less than CD3/CD28 cells (133.9 pg/ml).

These results demonstrate that costimulation through moesin is strong enough to initiate, maintain, and propagate the necessary signaling that leads to T cell activation and proliferation, outlining CD3/moesin as an alternate cell activation path. Moreover, given the similar activation and proliferation patterns observed between cells stimulated with anti-CD3 and anti-moesin or anti-CD3 and ALL, along with the results from IP experiments, it is tempting to speculate that the costimulatory signal is provided by the O-glycosylated form of moesin.

This work aimed to determine if the 70-kDa protein recognized by the Amaranthus Leucocarpus Lectin (ALL) in the T cell membrane is an O-glycosylated form of moesin and its potential ability to work as a CD4⁺ and CD8⁺ costimulatory molecule. Previous studies had described the presence of gpALL in mouse peripheral CD4⁺ and CD8⁺ T lymphocytes (18) however, the expression dynamics of these molecules after activation was unknown. Our results show that in vitro stimulation with anti-CD3/CD28 leads to an increase in ALL⁺ cells within the CD4⁺ and CD8⁺ subsets along with an increased expression of gpALL that peaks 48 h post-activation. This is consistent with the glycoprotein profile recognized through lectin blot by ALL and agrees with previous reports describing the remodeling of different glycosylation profiles after T cell activation (5, 38). Although gpALL expression increases similarly in both cell subsets after activation, it must be noted that the proportion of ALL⁺ cells is very different in freshly obtained CD4⁺ and CD8⁺ lymphocytes, which is approximately 30% and 90% respectively. This observation leads to wonder if the proportion of gpALL expressing non-activated CD4⁺ and CD8⁺ T cells is related to different activation or regulatory processes in each subset.

Among the bands recognized by ALL, the 70-kDa protein has been of particular interest. This molecule doubles its expression in CD4⁺ and CD8⁺ T cells after 48 h post-activation, which concurs with previous findings in mouse splenocytes (18), thymus (39), and macrophages (40). Further analysis of this protein showed that it is a member of the ERM family with 41% homology to an unnamed protein related to moesin (20), thus, given that its identity had not been confirmed we performed several experiments to address this issue. Although the moesin sequence contains no apparent transmembrane domain (41), it has been detected in the membrane of different cell lines (26) and the periphery of several hematopoietic cells (42). Kinetic expression analysis of cell surface moesin in CD4⁺ and CD8⁺ T cells after activation with anti-CD3/CD28 revealed that a similar percentage of moesin⁺ cells is observed within both unstimulated T cell subsets and that a similar expression dynamic is observed after activation. These observations contrast with the flow cytometric gpALL analysis, where a differential expression is observed between T cell subsets, however, we must consider that ALL recognizes several other molecules besides the 70-kDa protein and moesin expression experiments were limited by the analysis of the single protein, all of which accounts for the observed differences. While performing these experiments it caught our attention that we were only able to detect moesin by flow cytometry after cells had been fixed, it has been described that PFA alters cell surface mechanical properties due to the induction of covalent crosslinking between molecules (43), reduces cell mass density and destructs membrane integrity by dissolving some membrane lipids (44). We think that the membrane remodeling caused by PFA fixation allows the anti-moesin mAb to bind its otherwise hidden target, it is tempting to speculate that moesin is embedded in the cell membrane and that the O-GalNAcylation detected by ALL projects outwards. Further analysis on this structural subject will be determinant to understand moesin’s function.

The existence of the O-glycosylated moesin (O-moesin) was confirmed with IP experiments given that we were able to detect moesin after precipitation with ALL and vice versa. Interestingly, the lectin blot from the anti-moesin precipitated sample shows that besides the presence of the 70-kDa band, other proteins are detected by ALL, suggesting that moesin could have been precipitated as part of a molecular complex where other molecules share the same O-glycosylation. Moreover, in the lectin blot from the ALL-precipitated sample a dense 45-kDa band in CD4⁺ and CD8⁺ subsets and a 100-kDa band in CD8⁺ cells were also detected, suggesting that besides O-moesin, other O-glycosylated proteins are heavily expressed on T cells after activation. This agrees with the flow cytometry experiments where we can see a higher cell percentage positivity to ALL than moesin, particularly in the CD8⁺ subset.

Since moesin has a molecular weight of 67.8-kDa (27) and the O-moesin detected herein has a molecular weight of 70-kDa, we hypothesize that this difference corresponds to the glycosidic proportions of the O-glycoprotein, but this remains to be demonstrated by instrumental techniques in future works. Nevertheless, the bioinformatics analysis indicated that Thr⁵⁵⁸ is a site with a very high probability of O-glycosylation. This position has been described as fundamental for moesin function regulation since phosphorylation at this amino acid induces the protein to adopt an active/extended conformation where it can interact with other proteins (28, 29). Phosphorylation/O-glycosylation interplay has been described as a molecular switch that can regulate protein function and location (45); in fact, after T cell activation moesin is phosphorylated, and once activated it removes CD43 from the immunological synapse allowing proper activation (46). We consider that moesin Thr⁵⁸⁸ could be a target site for regulation through an O-glycosylation/phosphorylation interplay which could mediate the function and location of the protein in the cytosol or the cell membrane. Under this hypothesis, other interesting questions arise, like if O-moesin is synthesized de novo and directed to the cell membrane, or if the glycan addition to the protein occurs at the cytosol redirecting its translocation towards the cell membrane. The latter would be a rare example where the O-glycosylation of the T antigen type is synthesized in a place other than the endoplasmic reticulum (47); the answers to these questions are beyond the scope of this work but are proposed as thought-provoking perspectives.

Finally, we demonstrate that moesin can induce a costimulatory signal equivalent to the one provided by CD28 during in vitro activation with anti-CD3, considering that anti-CD3/moesin stimulated T cells were able to express activation molecules, produce IL-2 and proliferate. Although these responses were slightly lower than the ones observed after CD28 costimulation, the proliferation and activation patterns were the same between cells costimulated with ALL and anti-moesin. These results are in agreement with previous experiments reporting ALL costimulatory capacity in mouse and human T cells (15, 16) and that O-moesin is located in lipid rafts, cell membrane structures fundamental for the formation of the immunological synapse (20). Our observations reinforce other works describing moesin as a fundamental molecule for T cell homeostasis, maturation, and function: moesin^-/- mice exhibit T cells with reduced activation capacity and IL-2 production; lymphopenia as a consequence of a diminished incapacity to egress from the thymus, CD8⁺ Treg cells from these animals show decreased proliferation (30–32) and an X-linked moesin-associated immunodeficiency where diminished T cell proliferation is observed has been described in humans (48). Moesin also promotes actin polymerization during blast formation and is essential for the formation of the immunological synapse and T cell activation (32, 46, 49).

Altogether, our results demonstrate that moesin provides a, previously undescribed, costimulatory signal in CD4⁺ and CD8⁺ T cells that along with CD3 crosslinking can induce activation and proliferation; whether the signal is provided by the O-glycosylated form of moesin, if the activated cells can be polarized, perform cytolysis or become memory cells, and which is the natural ligand for this receptor, remain to be determined. However, our results show that T cells can be activated through a CD28 independent costimulatory pathway, which opens the door to increase our understanding of T cell biology and to investigate new potential immunomodulatory therapeutic targets.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Comité Interno para el Cuidado y Uso de Animales de Laboratorio (CICUAL), División de Investigación. Facultad de Medicina, UNAM.

WG-H, EZ, and ET designed the research study. WG-H performed the experiments. WG-H, FC-S, and ET analyzed the data. ET, RS, WG-H, and EZ wrote the article. All the authors discussed the results, reviewed the manuscript, and approved the submitted version.

This work was supported by grants IN218719, IN204717, IN213818, IN218919, and IN200321 from PAPIIT (DGAPA, UNAM, Mexico) and by grant 251116 from CONACYT (Mexico). WGH is the recipient of a Ph.D. fellowship from CONACYT (Registry 742472). WG-H is part of Biochemical Sciences PhD Program of the Universidad Nacional Autónoma de México and of the Université Paris-Est Créteil, France.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.