T cell receptor signaling co-regulates multiple Golgi genes to enhance N-glycan branching.

T cell receptor (TCR) signaling enhances (cid:1) 1,6GlcNAc-branching in N -glycans, a phenotype that promotes growth arrest and inhibits autoimmunity by increasing surface retention of cytotoxic T lymphocyte antigen-4 (CTLA-4) via inter-actions with galectins. N -Acetylglucosaminyltransferase V ( MGAT5 ) mediates (cid:1) 1,6GlcNAc-branching by transferring N -acetylglucosamine (GlcNAc) from UDP-GlcNAc to N -gly-can substrates produced by the sequential action of Golgi (cid:2) 1,2-mannosidase I (MIa,b,c), MGAT1 , (cid:2) 1,2-mannosidase II (MII, IIx), and MGAT2 . Here we report that TCR signaling enhances mRNA levels of MIa,b,c and MII,IIx in parallel with MGAT5 , whereas limiting levels of MGAT1 and MGAT2 . Blocking the increase in MI or MII enzyme activity induced by TCR signaling with deoxymannojirimycin or swainsonine, respectively, limits (cid:1) 1,6GlcNAc-branching, suggesting that enhanced MI and MII activity are both required for this phenotype. MGAT1 and MGAT2 have an (cid:1) 250- and (cid:1) 20-fold higher affinity for UDP-GlcNAc

T cell activation and differentiation induce a number of functionally important changes in cell surface N-glycosylation (1)(2)(3)(4). For example, activation signaling leads to replacement of ␣2,6-linked sialic acid with ␣1,3-linked galactose on the termini of N-acetyllactosamine (Gal␤1,4GlcNAc) units. N-Acetyllactosamine is the ligand for the galectin family of carbohydratebinding proteins, and the addition of terminal ␣2,6-linked sialic acid by the ST6Gal1 transferase inhibits galectin binding (4 -7). T cell differentiation leads to enhanced expression of ST6Gal1 in T H 2 relative to T H 1/T H 17 cells, thereby reducing sensitivity to galectin-1-induced apoptosis (4). Galectins bind surface glycoproteins in proportion to N-acetyllactosamine content, forming a molecular lattice that regulates distribution of glycoproteins to membrane microdomains (i.e. lipid rafts) and inhibits their loss to endocytosis (1, 8 -13). Binding avidity of galectins for glycoproteins is regulated by the number of attached N-glycans (i.e. N ϭ occupied N-X-(S/T) sites, X P), an encoded feature of protein sequences, as well as the degree of Golgi-mediated GlcNAc branching, which together determine N-acetyllactosamine content in glycoproteins (14). The number of N-X-(S/T) sites varies widely between different glycoproteins, with growth-promoting membrane glycoproteins (e.g. T cell receptor (TCR)) 2 generally displaying large numbers of N-X-(S/T) sites (i.e. n Ͼ 5), whereas growth inhibitory receptors (e.g. cytotoxic T lymphocyte antigen-4 (CTLA-4)) have few N-glycans (n Յ 4) (14). The large difference in galectin avidity for high and low multiplicity receptors allows Golgi-mediated changes in GlcNAc branching to differentially control surface retention (i.e. endocytosis rates) of these receptors and therefore transitions between growth and arrest signaling (14).
In resting T cells, where endocytosis is minimal, galectin-3 binds the TCR and the tyrosine phosphatase CD45 but partitions them to different membrane compartments by opposing F-actin targeting (12). Galectin binding prevents spontaneous TCR oligomerization in the absence of antigen, thereby blocking recruitment of multiple adaptor proteins and CD4-Lck to TCR, F-actin-mediated transfer to GM1-enriched lipid microdomains (GEMs), and activation signaling by Lck. In parallel, galectin partitions CD45 to GEMs by counteracting F-actin-mediated exclusion, dampening Lck activation. Upon encounter with antigen, GEMs cluster at the immune synapse formed at the contact site between T cells and antigen-presenting cells, with the galectin lattice reducing TCR/CD4-Lck and increasing CD45 concentration to dampen T cell activation and T H 1 differentiation (1,2,12). After activation, membrane turnover increases in T cell blasts, markedly limiting surface levels of the growth inhibitor CTLA-4 via constitutive endocytosis (14). GlcNAc branching increases in blasting T cells, enhancing CTLA-4 affinity for galectins and opposing surface loss to endocytosis (14). Thus, GlcNAc branching in N-glycans negatively regulates T cell growth early by raising T cell activation thresholds and later by enhancing CTLA-4-mediated growth arrest.
Autoimmunity is a complex trait that is influenced by environmental inputs. In this regard, N-glycan GlcNAc branching is regulated by the nutrient environment and metabolism via hexosamine pathway production of UDP-GlcNAc, the sugar nucleotide donor for Mgat1, Mgat2, Mgat4, and Mgat5 (14,16). Overexpression of Mgat1, an enzyme that has an ϳ250-fold lower K m for UDP-GlcNAc than Mgat5, paradoxically reduces ␤1,6GlcNAc branching by limiting UDP-GlcNAc supply to Mgat5 (14). Metabolic supplements to the hexosamine pathway suppress TCR signaling, T cell activation/proliferation, T H 1 differentiation, CTLA-4 endocytosis, and autoimmunity (EAE and autoimmune diabetes). TCR signaling results in sequential increases in glucose flux, UDP-GlcNAc production, ␤1,6Glc-NAc branching, and finally, CTLA-4 surface expression.

JOURNAL OF BIOLOGICAL CHEMISTRY 32455
MAN2A2 are both expressed, and loss of the former is reported to have little effect on GlcNAc branching or T cell growth (21). Effects of targeted deletion of other Golgi ␣-mannosidase genes in T cells have not been explored (22,23). Here we report that TCR signaling enhances expression of all five Golgi ␣-mannosidase genes in parallel with Mgat5, an activity required for optimal production of ␤1,6GlcNAc-branched N-glycans. In contrast, TCR signaling has little or negative effect on Mgat1 and Mgat2 expression, presumably to increase supply of UDP-GlcNAc to Mgat4 and Mgat5 and promote GlcNAc branching. Our data suggest that TCR signaling differentially regulates multiple Golgi enzymes at the mRNA levels to enhance GlcNAc branching in T cell blasts, and subsequently, growth arrest by CTLA-4.

EXPERIMENTAL PROCEDURES
Cell Culture, FACS, and ELISA-Human Jurkat T cells and peripheral blood mononuclear cells (PBMCs) were cultured in RPMI 1640 medium, supplemented with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin, and 50 M ␤-mercaptoethanol. For FACS, cells were collected and stained with fluorescent labeled Phaseolus vulgaris leukoagglutinin (L-PHA-FITC, Vector Laboratories) and anti-CD4 antibody as described previously (16). Splenocytes from myelin basic protein (MBP) immunized and non-immunized PL/J mice were stained with anti-CD4 (RM4 -5), anti-CD25 (PC61), and anti-CD69 (H1.2F3) from eBioscience and L-PHA-FITC from Vector Laboratories. All incubations and washes were performed on ice. Analyses were done with a FACScan flow cytometer using the CellQuest program (BD Biosciences). ELISA for IL-2 was performed on supernatant from PBMCs stimulated for 2 days as per the manufacturer's instructions (eBioscience).
To measure mRNA stability, cDNA was derived from cells treated for 0, 3, 6, 9, and 12 h with the transcriptional inhibitor actinomycin D (5 g/ml, Sigma). C t values were compared with mock-treated cells at corresponding time points, with mRNA half-life determined using a standard curve generated by 2-fold serial dilutions for each primer (24).
Phylogenetic Analysis-10 kb of 5Ј-untranslated region upstream sequences for various N-glycan-processing genes were  /ml). The -fold difference was calculated by the ⌬⌬C t method using glyceraldehyde-3-phosphate dehydrogenase as an endogenous control, with colorimetric results displaying comparisons with unstimulated mock control cells at each time point (i.e. 0, 3, 6, 9, 12, 24, 48, and 72 h). C, 10 kb of 5Ј-untranslated region nucleotide sequences of N-glycan pathway genes were retrieved from the Geneatlas data base and analyzed by the BioEdit software. The phylogenetic graph and distance were generated by the Neighbor-Joining/UPGMA method Version 3.62a2.1. N/A, not applicable.

Regulation of Golgi Genes by TCR Signaling
retrieved from the Geneatlas data base and phylogenetically analyzed using the Neighbor-Joining/UPGMA (unweighted pair group method with arithmetic mean) method (Version 3.62a2.1) (25). A phylogenetic distance value relative to MGAT5 generated by this software is also reported in Fig. 2C.
EAE Model of Inflammatory, Autoimmune Disease-EAE was induced by subcutaneous immunization of wild-type PL/J mice on days 0 and 15 with 100 g of bovine MBP (Sigma) emulsified in complete Freund's adjuvant containing 4 mg/ml heat-inactivated Mycobacterium tuberculosis (H37RA; Difco) distributed over two spots on the hind flank. Splenocytes were harvested at day 40 and analyzed by FACS. All procedures and protocols with mice were approved by the Institutional Animal Care and Use Committee of the University of California, Irvine, CA.

Basal mRNA Levels of N-Glycan Pathway Genes in Jurkat T Cells-
We assessed relative mRNA levels of multiple N-glycan pathway genes in resting Jurkat T cells by TaqMan quantitative real-time PCR. We examined MI genes (MAN1A1, MAN1A2, and MAN1C1), MII genes (MAN2A1 and MAN2A2), and MGAT1, MGAT2, and MGAT5 (Fig. 1A). Basal mRNA levels of MAN1C1, MGAT1, and MGAT2 were ϳ25-500-fold lower than the other tested genes (Fig. 1B). MI is encoded by three genes, and these data suggest that MAN1A1 and MAN1A2 are the major contributors to MI enzyme activity in Jurkat T cells. In contrast, MGAT1 and MGAT2 are the only genes encoding N-acetylglucosaminyltransferases I and II, respectively, and are absolutely required for GlcNAc branching (26 -28). Monitoring the decay of mRNA by quantitative real-time PCR in Jurkat T cells treated with the transcriptional inhibitor actinomycin D (24) suggests that the mRNA half-life of MGAT1 (t1 ⁄ 2 ϭ 7.7 h) is approximately twice that of MGAT5 (t1 ⁄ 2 ϭ 3.7 h), MAN1A1 (t1 ⁄ 2 ϭ 3.3 h), and actin (t1 ⁄ 2 ϭ 4.0 h) (Fig. 1C). This suggests that the large difference in mRNA levels of MGAT1 versus MGAT5 arises predominantly from differences in transcriptional control rather than mRNA stability. TCR Signaling Differentially Regulates MGAT5 versus MGAT1 and MGAT2 mRNA-GlcNAc branching is significantly enhanced by TCR signaling, a phenotype that limits cell surface loss of CTLA-4 by endocytosis (1,2,14). FACS analysis with leukoagglutinin (L-PHA), a plant lectin recognizing ␤1,6GlcNAc-branched N-glycans produced by Mgat5, provides a useful marker of cell surface GlcNAc branching (29). Consistent with previous observations in mouse ex vivo T cells (1, 2), L-PHA binding and MGAT5 mRNA levels in Jurkat T cells increase in proportion to TCR signal strength induced by  NOVEMBER 20, 2009 • VOLUME 284 • NUMBER 47

JOURNAL OF BIOLOGICAL CHEMISTRY 32457
anti-CD3 antibody (Fig. 2, A and B). The addition of anti-CD28 antibody, which provides a co-stimulatory signal that maximizes T cell activation, further enhanced L-PHA binding ( Fig.  2A). The expression of MGAT5 mRNA is regulated by the RAS-RAF-MAPK signaling pathway (30,31), which is activated by TCR signaling. At lower levels of anti-CD3 stimulation, MGAT5 mRNA expression is up-regulated ϳ3-fold starting from 12 h and reaches ϳ8-fold after 48 h. Notably, stronger anti-CD3 stimulation shifts peak expression to earlier time points (Fig. 2B) followed by relatively rapid decline to near baseline. MGAT5 mRNA has a short half-life (ϳ3.7 h, Fig. 1C), suggesting that strong TCR signaling induces a rapid up-and down-regulation of MGAT5 transcription. Similar kinetics were observed for TCR signaling-mediated changes in IL-2 mRNA, and to a lesser extent, IFN␥ (Fig. 2B). A correlation plot comparing all time points at both levels of anti-CD3 stimulation revealed a strong statistical correlation between MGAT5 and IL-2 (R 2 ϭ 0.915, p Ͻ 0.0001, Fig. 3A) but not IFN␥ (R 2 ϭ 0.118, p ϭ 0.27, Fig. 3A). In contrast, IL-4 mRNA levels were reduced at all time points, including when MGAT5 levels were maximally enhanced by TCR signaling (Fig. 2B). However, peak levels of MGAT5 mRNA were associated with the smallest declines in IL-4 mRNA and vice versa; therefore a strong correlation was still observed between these two genes (R 2 ϭ 0.917, p Ͻ 0.0001, Fig. 3A).
MGAT1 and MGAT2 mRNA levels displayed limited increases at low levels of TCR signaling with enhancement occurring significantly earlier than MGAT5 (Fig.  2B). Remarkably, high levels of stimulation reduced MGAT1 and MGAT2 mRNA levels by up to ϳ2-fold at time points when MGAT5 levels were enhanced (Fig.  2B). Indeed, there was no correlation between mRNA levels of MGAT5 and MGAT1 or MGAT2 (Fig. 3B). Phylogenic comparison of the 5Ј 10-kb promoter regions suggested that the distance between MGAT5 and MGAT1 or MGAT2 (Fig. 2C) is consistent with RAS-RAF-MAPK responsiveness of MGAT5 (30,31) and predominantly "housekeeping" promoter elements in MGAT1 and MGAT2 (32,33).
TCR Signaling Up-regulates MI and MII in Parallel with MGAT5 to Promote GlcNAc Branching-In contrast to MGAT1 and MGAT2, mRNA levels of the three MI genes (MAN1A1, MAN1A2, and MAN1C1) and two MII genes (MAN2A1 and MAN2A2) increased in parallel with MGAT5 at both low and high levels of anti-CD3 stimulation (Fig. 2B). Statistical analysis of mRNA levels under both conditions showed that the expression levels of MGAT5 transcripts strongly correlated with MAN2A1 (R 2 ϭ 0.9260, p Ͻ 0.0001), MAN2A2 (R 2 ϭ 0.9260, p Ͻ 0.0001), MAN1A1 (R 2 ϭ 0.8965, p Ͻ 0.0001), MAN1A2 (R 2 ϭ 0.8328, p Ͻ 0.0001), and MAN1C1 (R 2 ϭ 0.8148, p Ͻ 0.0001) (Fig.  3, C and D). At high dose TCR stimulation, peak levels of these genes coincided with down-regulation of MGAT1 and MGAT2 mRNA levels (Fig. 2B). Phylogenetic analysis of 10 kb upstream of the transcription start site revealed the greatest similarity to MGAT5 was MAN2A1 and MAN1A1 (Fig. 2C).

Regulation of Golgi Genes by TCR Signaling
Analysis of PBMCs from five healthy human subjects revealed similar results. Anti-CD3 ϩ anti-CD28 stimulation similarly enhanced L-PHA binding in CD4 ϩ T cells from all five subjects (Fig. 4A) and correlated with IL-2 secretion (Fig. 4B). This was associated with peak increases in mRNA of MGAT5, MAN2A1, MAN2A2, and MAN1A1 at 12 h after stimulation (Fig. 4, C-F). To confirm that T cell activation in vivo was associated with increased GlcNAc branching, we compared ex vivo mouse T cells from a control mouse with a mouse clinically affected by EAE (an animal model for the human inflammatory disease multiple sclerosis). As expected, FACS analysis of EAE splenocytes revealed increased expression of the T cell activation markers CD69 and CD25 as well as increased L-PHA binding in CD4 ϩ T cells (Fig. 5A).
Together, these data suggest that the Mgat5 and Golgi ␣-mannosidase enzymes are synchronously up-regulated at the transcriptional level by TCR signaling to enhance GlcNAc branching in activated T cells. To confirm that increases in MI and MII enzyme activity are required for up-regulation of Glc-NAc branching, we utilized the MI inhibitor deoxymannojiri-  NOVEMBER 20, 2009 • VOLUME 284 • NUMBER 47 JOURNAL OF BIOLOGICAL CHEMISTRY 32459 mycin (DMN) and the MII inhibitor swainsonine (SW) (Fig.  1A) (34). As expected, co-incubation of non-stimulated Jurkat T cells with increasing concentrations of DMN or SW both markedly reduced GlcNAc branching, as measured by L-PHA, while concurrently increasing high mannose structures, as revealed by concanavalin A staining (Fig. 5B). Remarkably, incubation of activated Jurkat T cells with concentrations of DMN and SW that have no effect on L-PHA binding in resting cells prevented phorbol 12-myristate 13-acetate (PMA)/ionomycin (Ionom)-mediated increases in L-PHA binding (Fig. 5C). The minimal concentrations of DMN and SW appear to block the increase in MI and MII activity induced by activation signaling, thereby preventing increased flux of N-glycan acceptors to Mgat5 and enhanced ␤1,6GlcNAc-branched N-glycans.

DISCUSSION
GlcNAc branching of N-glycans attached to cell surface glycoproteins such as TCR, CD45, and CTLA-4 titrates binding to galectins, forming a molecular lattice that negatively regulates T cell growth and susceptibility to autoimmunity. TCR signaling up-regulates MGAT5 gene expression and metabolic supply of UDP-GlcNAc to the Golgi to promote GlcNAc branching and CTLA-4 retention at the cell surface. Here we find that TCR signaling differentially regulates mRNA expression of multiple genes upstream of Mgat5 in the Golgi to enhance Glc-NAc branching. MGAT5 mRNA expression strongly correlates with the three Golgi MI and the two Golgi MII genes. Limiting TCR signaling-mediated increases in MI or MII enzyme activity confirms that this phenotype is required to enhance GlcNAc branching. In contrast, TCR signaling-mediated changes in mRNA levels in MGAT1 and MGAT2 do not correlate with MGAT5/MI/MII genes, being reduced when MGAT5 expression is maximally increased. MAGT1, MAGT2, and MAGT5 act sequentially, utilize UDP-GlcNAc as substrate, and are absolutely required for biosynthesis of ␤1,6GlcNAc-branched N-glycans. However, MGAT1 and MGAT2 have 200-and 20-fold lower K m for UDP-GlcNAc than MGAT5, suggesting unequal competition for substrate in the medial Golgi (14). Indeed, overexpression of MGAT1 reduces GlcNAc branching by reducing supply of UDP-GlcNAc to downstream GlcNAc transferases (14). We conclude that positive regulation of Glc-NAc-branched N-glycans by TCR signaling, and by extension, CTLA-4 surface retention and inhibition of autoimmunity, require coordinated up-regulation of MI, MII, and Mgat5 activities and UDP-GlcNAc biosynthesis, coupled with limited changes/suppression of Mgat1 and Mgat2.
MAN2A1-deficient resting T cells from C57BL/6 mice are reported to display little change in GlcNAc-branched N-glycans, suggesting that MAN2A2 supplies sufficient MII activity to compensate for this loss (21). Whether this is also true in activated T cells requires further investigation; however, it is consistent with our conclusion that up-regulation of total MII activity, via increases in both MAN2A1 and MAN2A2 mRNA, is the physiologically relevant change. Similarly, experimentally blocking up-regulation of a single MI gene in activated T cells is unlikely to be biologically significant; rather, enhanced MI activity distributed over increases in the three genes is likely the relevant phenotype required to increase GlcNAc branching in activated T cells.
Enhanced GlcNAc branching in activated T cells promotes cell surface retention of CTLA-4 (14), a critical negative regulator of autoimmunity (35). Coordinated up-regulation of MGAT5, MAN1A1, MAN1A2, MAN1C1, MAN2A1, and MAN2A2 mRNA downstream of TCR signaling suggests a common transcriptional regulator(s) activated by the phosphatidylinositol 3-kinase/ERK pathway. Defining the molecular mechanism for this co-regulation should provide new critical regulators of GlcNAc branching and expand the hexosamine/ Golgi gene network regulating CTLA-4 surface retention. Genetic variations in network genes are prime candidates for regulation of human autoimmunity.
Increasing the expression of MGAT1 reduces GlcNAc branching by reducing availability of UDP-GlcNAc to Mgat4 and Mgat5, suggesting that limiting Mgat1 activity is necessary for optimal increases in GlcNAc branching. MGAT1 and MGAT2 are simple genes with two and one exons, respectively, whereas MGAT5 has 16 exons with long intervening introns. It is tempting to speculate that these differences evolved to permit continuous and consistent production of glycoproteins with a minimum affinity for galectin (i.e. mono-and biantennary N-glycans by Mgat1 and Mgat2) and limited but tightly regulated production of high affinity galectin ligands by Mgat5. As exemplified in T cells, our data suggest that coordinated expression of multiple N-glycan pathway genes and UDP-Glc-NAc production by phosphatidylinositol 3-kinase/ERK signaling is necessary for optimal regulation of GlcNAc branching, and when disturbed, may lead to diseases such as autoimmunity, cancer, and type 2 diabetes (1, 36 -38).