Peanut Agglutinin High Phenotype of Activated CD8 (cid:1) T Cells Results from de Novo Synthesis of CD45 Glycans*

Following activation in the periphery, murine CD8 (cid:1) T cells exhibit a characteristic increased binding of peanut agglutinin (PNA), reflecting an increased expression of hyposialylated O -linked glycans (Gal (cid:2) 1–3GalNAc (cid:3) - O -Thr/Ser) on the cell surface. In this report, we show that the majority of the PNA receptors expressed on activated CD8 (cid:1) T cells are carried by CD45. Other glycoproteins ( e.g. CD8) and the glycolipid asialo-GM1 also carry PNA receptors, although to a much lesser extent. Analysis of enzymes involved in the sialylation/de-sialylation pathways showed that generation of PNA receptors in activated CD8 (cid:1) T cells is not due to up-regulation of endogenous sialidases. Instead, our results indicate that the PNA high phenotype results from de novo synthesis of CD45 carrying reduced sialylated core 1 O- glycans. 1 differentially subpopulations results from hyposialylation of core 1 O-glycans of CD45. activation stimuli. To study the effect of a sialidase inhib- itor, resting and activated cells were cultured for 36 h with and without Reagents and Flow Cytometry— and a PE-labeled incubated in of anti-asialo-GM1 g streptavidin-PE or PNA-FITC or asialo-GM1-PE, respectively,

Following activation in the periphery, murine CD8 ؉ T cells exhibit a characteristic increased binding of peanut agglutinin (PNA), reflecting an increased expression of hyposialylated O-linked glycans (Gal␤1-3GalNAc␣-O-Thr/Ser) on the cell surface. In this report, we show that the majority of the PNA receptors expressed on activated CD8 ؉ T cells are carried by CD45. Other glycoproteins (e.g. CD8) and the glycolipid asialo-GM1 also carry PNA receptors, although to a much lesser extent. Analysis of enzymes involved in the sialylation/de-sialylation pathways showed that generation of PNA receptors in activated CD8 ؉ T cells is not due to up-regulation of endogenous sialidases. Instead, our results indicate that the PNA high phenotype results from de novo synthesis of CD45 carrying reduced sialylated core 1 O-glycans.
Nearly 30 years ago PNA 1 was observed to recognize differentially subpopulations of lymphocytes in the thymus, and it is now routinely used as a marker of intrathymic T cell development ( Fig. 1) (1)(2)(3). Immature cortical thymocytes express abundant nonsialylated core 1 O-glycans (Gal␤1-3GalNAc␣-Ser/Thr), which are the preferred ligands of PNA, giving rise to their characteristic PNA high phenotype. In contrast, mature medullar thymocytes express increased levels of a sialyltransferase, ST3Gal I, that sialylates the core 1 O-glycans producing a structure not recognized by PNA, Sia␣2-3Gal␤1-3GalNAc␣-Ser/Thr (4 -7). As a result, the mature cells exhibit the PNA low phenotype that is maintained upon their transition into the periphery.
The functional significance of the glycosylation change detected by PNA has been underscored by analysis of ST3Gal I null mice in which T cells are constitutively PNA high (7). These mice exhibit a striking reduction of naive CD8 ϩ T cells due to increased apoptosis in the periphery. A major carrier of PNA receptors in CD8 ϩ thymocytes is the CD8 glycoprotein itself (6,8), and recent reports (9 -11) suggest that the sialylation status of core 1 O-glycans of CD8 during thymocyte maturation regulates its function. Indeed, CD8 expressed on immature thymocytes (PNA high ) binds more efficiently to the MHC class I than that expressed on mature cells (PNA low ). Consistent with these findings, the constitutively PNA high mature CD8 ϩ T cells from ST3Gal IϪ/Ϫ mice also show increased binding to MHC class I tetramer relative to wild type PNA low cells (9). Thus, the glycosylation change detected by PNA appears to modulate CD8 function during selection and maturation of CD8 ϩ T cells in the thymus.
In the periphery, activation of T lymphocytes leads to the re-expression of a PNA high phenotype (7,12,13), which is the basis for the use of PNA as a marker of germinal centers in spleen, lymph nodes, and Peyer's patches. By using a transgenic lymphocytic choriomeningitis virus infection model, Galvan et al. (13) have shown that CD8 ϩ T cells are converted to the PNA high phenotype. As these CD8 ϩ T cells become memory cells they exhibit an intermediate level of PNA binding (PNA int ). The increase in hyposialylated core 1 O-glycans may directly impact CD8 function, as discussed above, and can potentially modulate the functions of other cell surface glycoproteins. Indeed, Weiss and colleagues (14) have shown that de-sialylation of the O-linked glycans of CD45 can influence its dimerization and phosphatase activity. Changes in the sialylation status of O-linked glycans have also been postulated to modulate the interactions of lymphocytes with glycan-binding proteins (15)(16)(17).
The conversion of resting PNA low CD8 ϩ T cells to activated PNA high cells clearly results from decreased sialylation of core 1 O-glycans because peripheral CD8 ϩ T cells in the ST3Gal I null mice are PNA high (7). However, the mechanism for this change is not known. Several reports have suggested that sialidase(s) could generate PNA receptors by de-sialylating core 1 O-glycans (13,18,19). Consistent with this hypothesis, there is a clear increase in sialidase activity following activation of T cells (13, 18 -20). However, no direct causal link between the increase of sialidase activity and increase in PNA receptors has been established. Conversely, in a study of CD4 ϩ T cell differentiation following activation in vitro, Grabie et al. (21) demonstrated that the phenotypes of PNA high Th1 and PNA low Th2 cells correlated with low and high expression of ST3Gal I sialyltransferase, respectively. Thus, in principle, the conversion from PNA low to PNA high phenotype following activation of CD8 ϩ T cells could be accomplished either by the action of a sialidase or down-regulated expression of a sialyltransferase.
In this report we have investigated the nature of the PNA receptors on activated CD8 ϩ T cells, and we have assessed the role of sialidases and sialyltransferases in their generation. Although multiple glycoproteins and glycolipids contribute to the increase in PNA receptors, CD45 was found to be the primary PNA receptor. Most surprisingly, the PNA receptors were not generated from pre-existing CD45 by the action of sialidases but instead were carried primarily by CD45 generated following activation of CD8 ϩ T cells. The results suggest that the majority of PNA receptors on activated CD8 ϩ T cells results from hyposialylation of newly synthesized core 1 Oglycans of CD45.

MATERIALS AND METHODS
Mice-C57Bl/6 mice (6 -10 weeks old) were obtained from The Scripps Research Institute Custom Breeding core. SM/J mice (6 -10 weeks old) were purchased from The Jackson Laboratory (Bar Harbor, MA). GM2/GD2 synthaseϪ/Ϫ mice were kindly provided by Dr. R. Schnaar at The Johns Hopkins University School of Medicine (22).
Preparation and Culture of Splenocytes-To obtain single cell suspensions, spleens were ground between two frosted glass slides and passed through a 200-m nylon mesh (Polysciences, Warrington, PA). Erythrocytes were lysed in 150 mM ammonium chloride, 10 mM potassium carbonate, and 0.1 mM EDTA (pH 7.2) for 5 min at room temperature, and the resulting splenocytes were resuspended in RPMI 1640 media, 5% fetal bovine serum, 50 M ␤-mercaptoethanol. Cells were activated for 24, 48, or 72 h with immobilized anti-CD3 (15 g/ml in sodium bicarbonate (pH 9)) (Pharmingen), in RPMI media supplemented with 40 ng/ml IL-2 and 10 ng/ml IL-4 (R & D Systems, Minneapolis, MN). Resting cells were cultured for the same periods of time but without the activation stimuli. To study the effect of a sialidase inhibitor, resting and activated cells were cultured for 36 h with and without 2 mM 2,3-dehydro-2-deoxy-N-acetylneuraminic acid (Neu5Ac2en) (Sigma).
Staining Reagents and Flow Cytometry-FITC-labeled PNA (EY Laboratories, San Mateo, CA) and biotin-conjugated anti-asialo-GM1 (Seikagaku America, East Falmouth, MA) were used at a concentration of 2 and 5 g/ml, respectively. FITC-and PE-labeled mouse anti-CD8 (clone 53-6.7) (Pharmingen) were used at 5 g/ml. For fluorescence staining, cells were incubated in aliquots of 5 ϫ 10 5 cells in 100 l of PBS containing 10 mg/ml BSA with PNA-FITC for 30 min on ice. Staining with anti-asialo-GM1 was performed on ice, followed by 1 g of streptavidin-PE (Jackson ImmunoResearch Laboratories, West Grove, PA). Anti-CD8-PE or anti-CD8-FITC antibodies were co-incubated with PNA-FITC or asialo-GM1-PE, respectively, for dual color flow cytometry analysis. All samples were washed twice with 1 ml of the same buffer. Cells were gated based on viability and binding to anti-CD8 antibody. Flow cytometry data was acquired on a FACSCalibur cytometer (BD Biosciences) and analyzed using the Cellquest software system.
Isolation of CD8 ϩ T Cells-CD8 ϩ T cells were purified from freshly prepared and activated splenocytes using the MidiMACS system (Miltenyi Biotec, Auburn, CA) according to the manufacturer's instructions. Briefly, cells were resuspended in PBS containing 5 mg/ml BSA and 2 mM EDTA and incubated with anti-CD8 microbeads for 15 min at 8°C. After washing, cells were resuspended in the same buffer and applied to the column. The purity of CD8 ϩ T cell fractions used was Ն90% as judged by flow cytometry. Surface Biotin Labeling, PNA Precipitation, CD45 Immunoprecipitation, and Western Blot-Cell surface biotin labeling was performed as described previously (9). Briefly, cells were washed twice in PBS (pH 8), containing 1 mM MgCl 2 and 0.1 mM CaCl 2 , resuspended in buffer containing EZ-link sulfo-NHS-LC-LC-Biotin (Pierce) at a concentration of 0.5 mg/ml per 25 ϫ 10 6 cells, and rotated at room temperature for 30 min. Labeling was terminated by washing the cells twice in PBS.
For lectin precipitation, cells with or without biotinylation were lysed in PBS buffer (pH 7.2) containing 1% Triton X-100 on ice for 20 min, and the clarified supernatants were incubated with PNA-conjugated agarose (Sigma) for 3 h at 4°C. PNA-agarose beads were washed three times in the lysis buffer and once in 50 mM Tris (pH 7.5) and 150 mM NaCl. Precipitates were eluted with 200 mM lactose and resolved on SDS-PAGE gels. For CD45 immunoprecipitation, cells with or without biotinylation were lysed as described above, and the clarified supernatant was incubated with CD45 antibody (4 g of antibody for 100 g of lysates) for 2 h at 4°C. Protein G-agarose was added to the mixture and incubated for 1 h at 4°C. After incubation, the beads were washed as described above, and the precipitates were eluted using sample buffer. Proteins were separated on SDS-PAGE gels and transferred to a nitrocellulose membrane (Invitrogen). After blocking, blots containing biotinylated and nonbiotinylated proteins were incubated with 5 g/ml horseradish peroxidase (HRP)-conjugated PNA and HRP-conjugated streptavidin, respectively. Proteins were visualized by chemiluminescence (PerkinElmer Life Sciences).
Expression of Full-length Murine Neu3 in COS-7 Cells-The entire coding region of murine Neu3 was obtained by RT-PCR using primers from the published sequence (23). The primers sequences used are as follows: 5Ј-ATGGAGGAAGTCCCACCCTAC-3Ј (sense); 5Ј-CTTTAGTC-GCTACTAGGGCTG-3Ј (antisense). The PCR products were cloned into pcDNA3.1/His C vector (Invitrogen). Effectene Transfection reagent (Qiagen, Valencia, CA) was used to transfect the construct into COS-7 cells that were cultured in Dulbecco's modified Eagle's medium supplemented with penicillin/streptomycin and 10% fetal bovine serum. Cells were harvested 48 h post-transfection, washed in PBS, and lysed by sonication in 12 mM CaCl 2 (pH 6. substrate. A sample containing 1 milliunit of sialidase from Vibrio cholerae (Roche Applied Science) instead of cell homogenate was used as positive control. Samples were incubated for 60 min at 37°C, and reactions were terminated by the addition of 100 l of sodium bicarbonate (pH 10). The fluorescent product was measured on a Packard FluoroCount TM spectrofluorometer with excitation at 360 nm and emission at 460 nm, using 4-methylumbelliferone (Sigma) to obtain a calibration curve.
ST3Gal I Enzyme Assay-Fresh and activated CD8 ϩ T cells were lysed in a buffer containing 0.1 M NaCl, 50 mM sodium cacodylate (pH 6.5), 1.5% Triton CF-54 with protease inhibitor mixture (Calbiochem), spun at 6000 rpm for 5 min, and supernatants assayed for protein concentration using the BCA kit (Pierce). Enzymatic assays were carried out in duplicate using 240 g of total protein. Controls without exogenous acceptor were used to determine endogenous ST3Gal I activity. Positive controls were performed using purified human ST3Gal I. Reaction mixtures contained 10 g of anti-freeze glycoprotein (AFGP), 0.19 Ci of CMP-NeuAc (304 mCi/mmol), and cell lysates in a total volume of 100 l and were performed for 16 h at 37°C. Sialylated products were isolated by gel filtration on a Sephadex G-25 column (Sigma), eluted with 0.1 M NaCl, and quantified in a Packard scintillation counter.

Increased Expression of PNA-binding Sites following in Vitro
Activation of Peripheral CD8 ϩ T Cells-To investigate the molecular nature of PNA receptors and the mechanism of their generation on CD8 ϩ T cells, conditions for in vitro activation were established to recapitulate the conversion from the PNA low to the PNA high phenotype observed in vivo (24). Splenocytes were isolated and activated with immobilized anti-CD3 and IL-2/IL-4 as described under "Materials and Methods." The binding of PNA-FITC to CD8 ϩ T cells was analyzed after 24, 48, and 72 h of cell activation by flow cytometry. As demonstrated in Fig. 2, resting CD8 ϩ T cells showed low levels of PNA binding (PNA low ) in the three time points analyzed (Fig. 2, broken lines). After 24 h of activation there was an increase in PNA binding (PNA high ) in the entire population of CD8 ϩ T cells, which became more pronounced at 72 h (Fig. 2, solid  lines). Unless otherwise indicated, all subsequent experiments were performed on cells activated under these conditions for 72 h.
CD45 Is the Major Glycoprotein Carrier of PNA Receptors-Several groups have investigated PNA receptors on thymocytes using two different but related approaches. One is direct staining of blots using PNA (7,25), and the other is indirect staining of surface-labeled proteins precipitated with PNA microbeads (6, 8, 9, 26 -28). These two approaches yield somewhat different results with respect to the relative contribution of CD8 as a major carrier of PNA receptors. To determine the optimal conditions for evaluating PNA receptors on peripheral CD8 ϩ T cells, the two basic approaches were compared directly as shown in Fig. 3. We carried out PNA-agarose precipitation from lysates of both biotinylated and nonbiotinylated thymocytes, and we performed Western blot analysis using either streptavidin-HRP to detect the biotinylated proteins or PNA-HRP to directly stain PNA receptors. All bands were precipitated through specific interaction with PNA because no bands were observed in control precipitations in the presence of lactose (not shown) or when lactose was present during blotting with PNA ( Fig. 3, lanes 5 and 6). As shown in Fig. 3, when the blot was probed with streptavidin-HRP, multiple protein bands were detected in the biotinylated sample, with CD8a and CD8b being the predominant bands (lane 2). However, when the blot was probed with PNA-HRP, the CD8 bands became minor, and the rest of the bands had similar intensity (Fig. 3, lanes 3 and  4). Some PNA-reactive proteins migrated at a higher position in lane 4 (Fig. 3) due to the biotin attachment that increases the molecular mass. The reason that CD8 was detected very intensely on the streptavidin-HRP blot is likely because both CD8a and CD8b have a very high lysine content, with more than 10% lysine residues in their protein sequences. Based on these results we concluded that direct staining with PNA was the preferred method for assessing the relative level of PNA receptors on CD8 ϩ T cell glycoproteins.
To determine whether the same proteins carried PNA receptors on activated peripheral CD8 ϩ T cells, we conducted PNAagarose precipitation of fresh and activated splenic CD8 ϩ T cells followed by PNA-HRP blot. As shown in Fig. 4A, no PNAspecific bands were detected in resting CD8 ϩ T cells. The single band seen at 30 kDa is nonspecific because it was not competed away in the presence of lactose (Fig. 4A, lanes 3 and 4). In contrast, activated cells contained two PNA-reactive bands, a major band of 200 kDa and a minor band of 65 kDa (Fig. 4A, lane 2). The molecular mass of the top band suggested that it was CD45, a major T cell surface glycoprotein. To test this, immunoprecipitation of cell lysates with an anti-CD45 antibody was performed followed by a PNA-HRP blot. The results shown in Fig. 4A, lane 5, indicate that this band is indeed CD45 and that this glycoprotein is the major carrier of PNA-binding sites in activated CD8 ϩ T cells.
In order to investigate whether there are other minor components containing PNA-binding sites, we carried out PNA precipitation of biotinylated cell lysates followed by streptavidin-HRP blot (Fig. 4B). Our results clearly indicated that a few proteins are PNA-reactive in the resting CD8 ϩ T cells, and their intensity increased upon activation. By immunoprecipitation, we identified three of the proteins to be CD45, CD8a, and CD8b (data not shown, arrows in Fig. 4B indicate the positions). Although there are few other proteins containing PNA-binding sites, CD45 shows the most dramatic increase following cell activation.
Relative to the PNA receptor on glycoproteins, the major difference in the ganglioside structure is that the Gal␤1-3GalNAc sequence is in ␤-linkage to a sugar instead of in ␣-linkage to threonine. In order to determine whether PNA binds to GM1 and asialo-GM1, the gangliosides were chromatographed on HPTLC plates and stained with PNA (Fig. 5A). Whereas PNA bound to asialo-GM1, no binding was observed to GM1, most likely due to the presence of the internal sialic acid.
Because PNA recognizes asialo-GM1, we next questioned whether or not the expression of asialo-GM1 changed upon activation of CD8 ϩ T cells. The binding of anti-asialo-GM1 to CD8 ϩ T cells was analyzed by flow cytometry after 24, 48, and 72 h of activation. As shown in Fig. 5B, after 24 h of activation there was an increase in the expression of asialo-GM1 of ϳ50% of CD8 ϩ T cells. The percentage of CD8 ϩ T cells expressing asialo-GM1 increased after 48 and 72 h of activation; however, the difference in staining intensity between resting (Fig. 5B, broken lines) and activated cells (Fig. 5B, solid lines) was not as large as that observed with PNA (Fig. 2). Because asialo-GM1 can be recognized by PNA and its expression increases upon activation of CD8 ϩ T cells, it is likely that asialo-GM1 is a PNA receptor in activated CD8 ϩ T cells.
In order to directly test the extent of contribution of asialo-GM1 to the increased expression of PNA-binding sites on activated CD8 ϩ T cells, we compared the binding of PNA to CD8 ϩ T cells isolated from wild type mice and a GM2/GD2Ϫ/Ϫ mouse strain. The GM2/GD2 synthaseϪ/Ϫ mouse strain does not express complex gangliosides, including asialo-GM1, because it is missing a key GalNAc transferase required for their biosynthesis (22). Our results showed that despite the absence of asialo-GM1, binding of PNA to activated CD8 ϩ T cells from these mice was not significantly different from that observed with cells from wild type mice (Fig. 5C). Therefore, changes in the expression of the ganglioside asialo-GM1 cannot account per se for the conversion to the PNA high phenotype upon activation of CD8 ϩ T cells. It is clear that asialo-GM1 can be detected by PNA, and the expression of asialo-GM1 increased in the activated CD8 ϩ T cells, but the total contribution of asialo-GM1 to PNA binding is small. Taken together with the results from Fig. 4, these data suggest that CD45 is the main PNA receptor in activated CD8 ϩ T cells.
Changes in Expression of Sialyltransferases and Sialidases Potentially Implicated in Generating PNA-binding Sites-As described above, changes in the expression of either sialidases or sialyltransferases could account for the increase in PNA receptors following cell activation (7, 13, 18 -21). The expression of PNA-binding sites, i.e. core 1 O-glycans, as depicted in Fig. 6, can be regulated by at least one of three sialyltransferases that synthesize the NeuAc␣2-3Gal␤1-3GalNAc sequence, ST3Gal I, ST3Gal II, or ST3Gal IV. Additionally, it can be generated by a sialidase that removes sialic acid and exposes the Gal␤1-3GalNAc sequence.
Of the three sialyltransferases listed above, only ST3Gal I was detected in resting CD8 ϩ T cells using a microarray analysis carried out in our laboratory. 2 ST3Gal I was previously implicated in generating PNA low phenotypes in mature thymocytes and in CD4 ϩ Th2 cells (5,7,21). In order to investigate whether the PNA high phenotype of activated CD8 ϩ T cells might be regulated by ST3Gal I, transcript levels of this enzyme were analyzed by real time RT-PCR. As shown in Fig. 7A, the mRNA levels of ST3Gal I were lower in activated CD8 ϩ T cells when compared with resting cells, although the difference was not statistically significant (p ϭ 0.17).
To evaluate if the lowered expression level of ST3Gal I mRNA

FIG. 4. CD45 is the major PNA-reactive glycoprotein on activated CD8 ؉ T cells.
A, freshly isolated and activated CD8 ϩ T cells (activated with immobilized anti-CD3 and media supplemented with IL/2/IL4) were harvested and lysed in Triton X-100 buffer (1 ϫ 10 8 cells/ml for fresh samples and 2 ϫ 10 7 cells/ml for activated samples to compensate for lower protein content of fresh cells). Lysates were subjected to PNA-agarose precipitation or anti-CD45 immunoprecipitation, and the precipitated proteins were resolved by SDS-PAGE, transferred to nitrocellulose, and blotted with PNA-HRP in the presence or absence of lactose. B, fresh or activated CD8 ϩ T cells were isolated, biotinylated, and lysed, and the clarified supernatant was used for PNA-agarose precipitation. The precipitates were resolved on SDS-PAGE, transferred to nitrocellulose, blotted with SA-HRP, and detected by chemiluminescence as described under "Materials and Methods." was accompanied by a decreased enzymatic activity, fresh and activated CD8 ϩ T cells were used in assays to determine sialyltransferase activity using AFGP, which contains the Gal␤1-3GalNAc sequence used by ST3Gal I as a substrate. The specific activity of ST3Gal I in activated CD8 ϩ T cells was reduced by  30% (p ϭ 0.008; Fig. 7B). Altogether, these results suggest that the decrease in expression and activity of ST3Gal I could account for the PNA high phenotype of activated CD8 ϩ T cells.
As mentioned above, a sialidase could also be involved in generating the PNA high phenotype. Four sialidase genes have been reported in mouse, Neu1-4 (23,(31)(32)(33). Of these, Neu4 is not expressed in CD8 ϩ T cells (33), and Neu2 was shown to be a cytoplasmic enzyme (34) and is not accessible to cell surface proteins. Accordingly, the expression levels of Neu1 and Neu3 were evaluated by real time RT-PCR as shown in Fig. 7. The expression levels of Neu1 were found to be lower in activated CD8 ϩ T cells, when compared with those of resting cells (Fig.  7A). In contrast, the expression levels of the sialidase Neu3 were 3.2-fold higher in activated CD8 ϩ T cells, when compared with their resting counterpart (Fig. 7A), raising the prospect that Neu3 could be involved in the increase of PNA binding to activated CD8 ϩ T cells.
Neither Neu1 nor Neu3 Are Responsible for the Generation of PNA Receptors-Although Neu1 mRNA levels decreased in activated CD8 ϩ T cells, the decrease is not statistically significant (p ϭ 0.06), and we wished to obtain independent support for its possible involvement, or lack thereof, in the generation of PNA receptors. For this reason we compared the binding of PNA to CD8 ϩ T cells isolated from wild type mice with those isolated from the naturally occurring mouse strain SM/J. These mice, due to a mutation in Neu1, have reduced Neu1 sialidase activity of less than 10% that of wild type mice (35). PNA binding to T cells isolated from these two strains of mice was analyzed by flow cytometry, and the results are shown in Fig.  8. No significant differences were found between CD8 ϩ T cells isolated from SM/J mice and those from the wild type strain regarding their ability to generate PNA-binding sites. Therefore, we concluded that Neu1 is not required for the increased PNA binding following activation of CD8 ϩ T cells.
Results of the expression analysis suggest that Neu3 could account for the increased sialidase activity of activated CD8 ϩ T cells reported previously (13,18,19). Neu3 has also been reported to be a plasma membrane enzyme, topologically well suited for the generation of PNA receptors. On the other hand, it seems unlikely that Neu3 could account for the generation of , and Neu3 genes are differentially expressed in fresh and activated CD8 ϩ T cells. CD8 ϩ T cells were isolated from C57Bl/6 mice and activated as described under the "Materials and Methods" for 72 h, and quantitative analyses of ST3Gal I, Neu1, and Neu3 transcript levels were conducted by real time PCR. Fold change between fresh and activated samples was calculated after normalization to the housekeeping gene ribosomal phosphoprotein and represents the means Ϯ S.D. of 3-4 replicates. Positive and negative values indicate higher expression in the activated and fresh samples, respectively. B, ST3Gal I activity of fresh and activated CD8 ϩ T cells isolated and cultured as described under "Materials and Methods." Fresh and activated CD8 ϩ T cells were lysed in Triton CF-54 buffer, and the clarified supernatants were used for the activity assay. Reaction mixtures contained 10 g of AFGP as the acceptor and 0.19 Ci of CMP-NeuAc as the donor in a total volume of 100 l. Sialylated products were isolated by gel filtration on a column of Sephadex G-25 in 0.1 M NaCl and were quantified in a scintillation counter. The activity is expressed as picomoles of product per mg of lysate proteins and represents means Ϯ S.D. of three replicates. PNA receptors based on its reported specificity for ganglioside substrates. Indeed, no activity has been detected toward glycoprotein and very low activity toward oligosaccharide substrates (36,37).
To investigate further the involvement of Neu3 in generating PNA receptors, we sought to use the sialidase inhibitor Neu5Ac2en because it was shown previously (38) to inhibit Neu3 activity in cell culture systems. Control studies were performed to test the ability of Neu5Ac2en to inhibit Neu3 sialidase activity. Enzymatic assays were carried out using Neu3-expressing COS-7 cells and the substrate 4-MU-NANA, in the presence and absence of the inhibitor, as described under "Materials and Methods." As shown in Fig. 9A, Neu5Ac2en is a potent inhibitor of Neu3 sialidase activity, revealed by complete loss of Neu3 activity when 2 mM Neu5Ac2en was added to the reaction mixture.
To evaluate the contribution of the Neu3 sialidase in generating PNA receptors, we included 2 mM Neu5Ac2en in the media during activation of spleen cells. Subsequent analysis by flow cytometry showed that the increase of PNA binding to activated CD8 ϩ T cells was virtually unaffected by the presence of Neu5Ac2en (Fig. 9B). These results indicate that Neu3 does not contribute to the conversion from PNA low to PNA high following cell activation.
PNA Receptors on CD45 Are Generated de Novo following Activation of CD8 ϩ T Cells-Our results suggest that CD45 accounts for the majority of PNA receptors on activated CD8 ϩ T cells. However, the combined data on sialidases and sialyltransferases do not point to any one enzyme being clearly responsible for generation of these receptors on CD45. On balance, it appears that sialidases cannot account for the appearance of PNA receptors. If this is true, then PNA receptors are not likely to be generated from pre-existing glycans but rather result from PNA-positive glycans produced following activation of T cells. Therefore, we sought to determine whether PNA receptors were carried exclusively on newly synthesized CD45 or were generated in part from CD45 pre-existing at the time of activation. We reasoned that pre-existing CD45 could be biotinylated prior to activation, which could then be distinguished from newly synthesized CD45 following activation by using streptavidin-based reagents. Accordingly, fresh splenocytes were biotinylated and cultured with and without the activation stimuli. PNA precipitation was carried out with biotinylated, resting, and activated CD8 ϩ T cells, and precipitates were resolved on SDS-PAGE followed by PNA-HRP blot. As observed previously, CD45 was found only following activation (Fig. 10, lane 2). In order to determine whether there is PNA-reactive biotinylated CD45 in resting and activated CD8 ϩ T cells, cell lysates were precipitated with PNAagarose and subjected to streptavidin-HRP blotting. As shown in Fig. 10 (lanes 3 and 4), no biotinylated CD45 was detected in either resting or activated cells after precipitation with PNAagarose. In contrast, anti-CD45 immunoprecipitation led to the detection of biotinylated CD45 in both resting and activated cells (Fig. 10, lanes 5 and 6). Therefore, these results indicated that PNA-reactive CD45 is not biotinylated, and it is synthesized de novo. Overall these results show that cell activation is accompanied by de novo synthesis of CD45 and that this newly synthesized CD45 is the main carrier of PNA-binding sites in activated CD8 ϩ T cells. Furthermore, these results support the role of ST3Gal I in modulating the expression of PNA-binding sites on CD45. DISCUSSION Differential staining of lymphocytes by PNA is widely used as a convenient marker of their differentiation and activation (1-3, 7, 13). These staining patterns result from the interconversion of the glycan ligand of PNA (Gal␤1-3GalNAc␣-O-Thr) and its sialylated form (Sia␣2-3Gal␤1-3GalNAc␣-O-Thr), which is not recognized by PNA (4 -7). In addition to the utility of PNA as a histochemical marker, several reports have provided evidence that the underlying glycosylation changes detected by PNA have functional implications for T cell biology (16,39), including modulation of the functions of key regulatory molecules such as CD8 (9 -11) and CD45 (14). In this report we have investigated the conversion of peripheral splenic CD8 ϩ T cells from PNA low to PNA high following activation in vitro, both with respect to the molecules that carry PNA receptors and the mechanism for their appearance.
A striking finding is that CD45 carries the majority of the PNA receptors on activated CD8 ϩ T cells. Indeed, CD45 is the only band readily detected following PNA precipitation and FIG. 9. Neu5Ac2en inhibits the activity of Neu3 expressed in COS-7 cells but has no significant effect on PNA binding to activated CD8 ؉ T cells. A, COS-7 cells transiently transfected with recombinant Neu3 were harvested 48 h post-transfection and then were lysed and assayed for sialidase activity, as described under "Materials and Methods." Activity is expressed as relative fluorescent intensity. B, splenocytes isolated from C57Bl/6 mice were cultured as described under "Materials and Methods." Resting and activated cells, cultured in the absence and presence of 2 mM Neu5Ac2en, were double stained with FITC-PNA (x axis) and PE-anti-CD8 (y axis). direct staining of Western blots with labeled PNA (Fig. 4A). In marked contrast, PNA blots of thymocytes reveal multiple bands of similar intensity in the 65-200-kDa range (Fig. 3). Similarly, PNA blots of the constitutively PNA high T cells from ST3Gal I null mice (7), or CD8 ϩ T cells treated with sialidase, reveal CD43 as a predominant band. 3 Thus, during activation of peripheral CD8 ϩ T cells, CD45 disproportionately acquires PNA receptors relative to other glycoproteins like CD43 that carry equivalent or greater "potential" PNA receptors blocked by sialic acids.
Detailed analysis of the O-linked glycans of CD43 before and after activation of human T cells by Piller et al. (40) is consistent with the lack of PNA receptors on CD43 of activated murine T cells reported here. The O-glycans of CD43 from resting cells shifted from predominantly core 1 (Gal␤1-3GalNAc␣Thr) to the more complex core 2-type glycans (Gal␤1-3[Gal␤1-4GlcNAc]GalNAc␣Thr) on activated cells. Both of these neutral cores are PNA ligands because substitution of the GalNAc moiety at the 6 position does not block PNA binding (7). However, the terminal galactose residues on both core 1 and core 2 were found to be fully sialylated (40), which precludes binding of PNA. Thus, as observed for murine CD8 ϩ T cells in this report, CD43 is predicted to be a minor contributor of PNA receptors on peripheral activated human T cells as well.
Whereas CD45 is the primary PNA receptor, by using more sensitive methods of analysis other glycoconjugates were found to carry increased levels of PNA receptors on activated CD8 ϩ T cells relative to resting cells. Indeed, PNA precipitation of biotinylated cell lysates followed by streptavidin blot detected several glycoproteins that bound to PNA in resting cells, which increased 2-3-fold following activation (Fig. 4B). These include CD8 and the predominant protein at 65-75 kDa. This method of detection amplifies the detection of CD8 due to its high lysine content resulting in a high level of biotinylation (Fig. 3) (8). Notably, by using this method, CD45 still showed the most dramatic increase in PNA receptors (Fig. 4B).
In addition to glycoproteins, the ganglioside asialo-GM1, containing the terminal Gal␤1-3GalNAc sequence, is recognized as a ligand by PNA (Fig. 5A), and the expression of asialo-GM1 on CD8 ϩ T cells increases following cell activation (Fig. 5B). However, CD8 ϩ T cells isolated from GM2/GD2 synthase null mice, which cannot express asialo-GM1 (22), acquired similar levels of PNA receptors as CD8 ϩ T cells from wild type mice. Taken together, these results suggest that whereas asialo-GM1 binds PNA and clearly increases on activated CD8 ϩ T cells, it does not constitute a major fraction of the total PNA receptors.
In principle, the conversion from PNA low to PNA high following activation of CD8 ϩ T cells can be achieved in two ways as follows: 1) de novo synthesis of hyposialylated O-linked glycans recognized by PNA, or 2) "uncovering" PNA receptors on sialylated O-linked glycans through the action of a suitable sialidase (see Fig. 6). Several reports (13, 18 -20) have suggested that sialidases are responsible based on the correlation between increased sialidase activity and acquisition of hyposialylated glycans and PNA receptors in activated T cells. Although this is an attractive hypothesis, there has been no direct demonstration that the sialidase activity detected can account for the generation of PNA ligands.
In preliminary experiments we confirmed that activated CD8 ϩ T cells exhibited an increase in sialidase activity as detected by the substrate 4-MU-NANA. 4 Of the four known mammalian sialidases (Neu1-Neu4), Neu2 is a cytoplasmic enzyme (34), and Neu4 is not found in lymphocytes (33), leaving Neu1 and Neu3 as candidates for the increased sialidase activity in activated lymphocytes. Neu1 was eliminated as a candidate because it exhibited decreased expression in activated CD8 ϩ T cells (Fig. 7), and there was no noticeable difference in acquisition of PNA receptors following activation of CD8 ϩ T cells from wild type and Neu1-deficient SM/J mice (Fig.  8). Whereas the plasma membrane sialidase, Neu3, exhibited increased expression in activated CD8 ϩ T cells, a potent sialidase inhibitor of Neu3 had no effect on the acquisition of PNA receptors (Fig. 9). Taken together, these results suggested that none of the four known murine sialidases can account for the increased PNA receptors in activated CD8 ϩ T cells.
To test further for the possible involvement of a sialidase in generating PNA receptors, we biotinylated cell surface proteins of splenocytes prior to activation as a tag for "pre-existing" CD45. Following activation, CD8 ϩ T cells were purified, and CD45 precipitated by PNA was analyzed. No biotinylated CD45 could be detected, indicating that all PNA receptors carried by CD45 were a result of de novo synthesis and not due to the action of sialidases. Thus, the appearance of PNA receptors on CD45 appears to result from an altered biosynthesis pathway resulting in hyposialylated O-linked glycans.
The most likely cause of hyposialylated O-linked glycans (PNA ligands) is alteration of the activity of sialyltransferase ST3Gal I. The regulated expression of ST3Gal I is well documented to account for the PNA high and PNA low phenotype of immature and mature thymic T cells, respectively (5, 7), and to account for the difference in PNA reactivity of CD4 Th1 cells (PNA high ) relative to CD4 Th2 cells (PNA low ) (21). We observed a small but statistically significant decrease in ST3Gal I specific activity (ϳ30%), which could result in hyposialylation of O-linked glycans if the amount of enzyme is rate-limiting, especially in the context of the increased production of glycoproteins in an activated T cell. However, other factors could contribute or cause the hyposialylation. For example, it is possible that there is a limitation of production or transport of the donor substrate CMP-sialic acid. Recently, a conserved mammalian oligomeric Golgi-localized protein complex, the COG complex, has been implicated in organizing the structure and 3 Q. Yan and J. Paulson, unpublished observations. 4 M. Amada and J. Paulson, unpublished results.
FIG. 10. PNA receptors on CD45 are generated de novo following activation of CD8 ؉ T cells. Splenocytes were isolated and biotinylated as described under "Materials and Methods." Biotinylated splenocytes were incubated with or without anti-CD3/IL2/IL4 for 72 h, and CD8 ϩ T cells were purified. Purified cells were then lysed in Triton X-100 buffer, and the supernatant was used for PNA-agarose precipitation or anti-CD45 immunoprecipitation. The eluants were resolved on SDS-PAGE and blotted with PNA-HRP (lanes 1 and 2) or streptavidin-HRP (SA-HRP, lanes [3][4][5][6]. To compensate for the dilution of biotinylated proteins in the dividing activated cell samples, five times more protein was loaded in lanes 2, 4, and 6 than in lanes 1, 3, and 5. activity of the Golgi apparatus and localization of Golgi membrane proteins (41,42). Thus it is possible that activation affects the COG complex resulting in mis-localization of ST3Gal I, effectively reducing its activity where it is needed.
Several reports have suggested that the hyposialylated core 1 O-glycans detected by PNA in activated CD8 ϩ T cells may have functional significance. The constitutively PNA high CD8 ϩ T cells of the ST3Gal I-deficient mouse undergo normal development in the thymus but undergo rapid apoptosis in the periphery by a mechanism that is still unclear (7). Moreover, a recent report by Starr et al. (39) suggests that biochemical de-sialylation or genetic deficiency in ST3Gal I transferase may augment CD8 ϩ T cell sensitivity by enhancing immunological synapse formation to cognate antigens. Recent studies on the effects of O-glycosylation of CD8 in thymocytes showed that CD8 expressed on immature PNA high double positive thymocytes bound more efficiently to MHC class I molecules than that expressed on mature PNA low thymocytes, an interaction shown to be related to the hyposialylation of core 1 O-glycans detected by PNA (9 -11). However, although hyposialylated glycans of CD8 could be detected on activated cells (Fig. 4B), the increase over that present on resting cells was modest and, compared with sialidase-treated cells, represented a minor portion of the total CD8 glycans (data not shown). Thus, if sialic acid plays a role in regulating CD8 affinity for MHC in activated CD8 ϩ T cells, it would play a minor role relative to the role played in immature (PNA high ) and mature (PNA low ) thymocytes.
The major carrier of PNA receptors on activated CD8 ϩ T cells is CD45. The primary function of CD45 in T cells is to dephosphorylate the negative regulatory tyrosine of the Src family protein tyrosine kinases involved in the initiation of the immune response (43). Upon activation of CD8 ϩ T cells, CD45 is converted from the RABC isoform to the RO isoform, which eliminates the majority of the O-glycans located on the peptide sequence encoded by the alternatively spliced ABC exons. Weiss and co-workers (14,44) found that the RO form dimerizes more efficiently than the larger isoforms containing Oglycans, and that dimerization down-regulates phosphatase activity of this important signaling molecule. They further showed that sialylation and O-glycosylation can reduce dimerization of the isoforms containing the ABC exons (14). We found that after 72 h of activation different isoforms exist and so does the RABC form (data not shown). Based on our results, we hypothesize that upon activation and in addition to alternative splicing that produced the RO form, the hyposialylation of the O-linked glycans of CD45 may in the interim promote dimerization of de novo synthesized CD45 to down-regulate CD45 phosphatase activity. Further investigation into the putative involvement of O-glycan sialylation in modulating of T cell immune response is required.