Silencing the Nucleocytoplasmic O-GlcNAc Transferase Reduces Proliferation, Adhesion, and Migration of Cancer and Fetal Human Colon Cell Lines

The post-translational modification of proteins by O-linked β-N-acetylglucosamine (O-GlcNAc) is regulated by a unique couple of enzymes. O-GlcNAc transferase (OGT) transfers the GlcNAc residue from UDP-GlcNAc, the final product of the hexosamine biosynthetic pathway (HBP), whereas O-GlcNAcase (OGA) removes it. This study and others show that OGT and O-GlcNAcylation levels are increased in cancer cell lines. In that context, we studied the effect of OGT silencing in the colon cancer cell lines HT29 and HCT116 and the primary colon cell line CCD841CoN. Herein, we report that OGT silencing diminished proliferation, in vitro cell survival and adhesion of primary and cancer cell lines. SiOGT dramatically decreased HT29 and CCD841CoN migration, CCD841CoN harboring high capabilities of migration in Boyden chamber system when compared to HT29 and HCT116. The expression levels of actin and tubulin were unaffected by OGT knockdown but siOGT seemed to disorganize microfilament, microtubule, and vinculin networks in CCD841CoN. While cancer cell lines harbor higher levels of OGT and O-GlcNAcylation to fulfill their proliferative and migratory properties, in agreement with their higher consumption of HBP main substrates glucose and glutamine, our data demonstrate that OGT expression is not only necessary for the biological properties of cancer cell lines but also for normal cells.

The post-translational modification of proteins by O-linked β-N-acetylglucosamine (O-GlcNAc) is regulated by a unique couple of enzymes. O-GlcNAc transferase (OGT) transfers the GlcNAc residue from UDP-GlcNAc, the final product of the hexosamine biosynthetic pathway (HBP), whereas O-GlcNAcase (OGA) removes it. This study and others show that OGT and O-GlcNAcylation levels are increased in cancer cell lines. In that context, we studied the effect of OGT silencing in the colon cancer cell lines HT29 and HCT116 and the primary colon cell line CCD841CoN. Herein, we report that OGT silencing diminished proliferation, in vitro cell survival and adhesion of primary and cancer cell lines. SiOGT dramatically decreased HT29 and CCD841CoN migration, CCD841CoN harboring high capabilities of migration in Boyden chamber system when compared to HT29 and HCT116. The expression levels of actin and tubulin were unaffected by OGT knockdown but siOGT seemed to disorganize microfilament, microtubule, and vinculin networks in CCD841CoN. While cancer cell lines harbor higher levels of OGT and O-GlcNAcylation to fulfill their proliferative and migratory properties, in agreement with their higher consumption of HBP main substrates glucose and glutamine, our data demonstrate that OGT expression is not only necessary for the biological properties of cancer cell lines but also for normal cells. The unique glycosyltransferase OGT transfers the GlcNAc moiety from UDP-GlcNAc to target proteins and the glycosidase OGA hydrolyzes the glycosidic bond; therefore, O-GlcNAcylation is versatile. Glucose and glutamine are the main substrates of the hexosamine biosynthetic pathway, which gives rise to UDP-GlcNAc. Moreover, the consumption of these two metabolites is increased in cancer cell lines. The flux of the hexosamine pathway is controlled by the rate-limiting enzyme GFAT. The level of UDP-GlcNAc, and consequently of O-GlcNAcylation, is tightly dependent upon the nutrient status of the cell. After its release by OGA, GlcNAc is reactivated by NAGK in the GlcNAc salvage pathway. HK, hexokinase; GPI, phosphoglucose isomerase; GFAT, glutamine:fructose-6-phosphate amidotransferase; GNPNAT1, glucosamine-6-phosphate acetyl transferase; AGM1, phospho-N-acetylglucosamine mutase; UAP, uridine di-phospho-N-acetylglucosamine pyrophosphorylase; NAGK, N-acetylglucosamine kinase; OGA  cycling is involved in many fundamental functions, including translation (1), transcription (2), cell signaling (3), and protein trafficking (4). Deregulation of O-GlcNAcylation dynamics actively participates in tumorigenesis and in the etiology of cancer (5). A synergy between unhealthy diet and cancer development is proposed because of the status of OGT's natural substrate, UDP-GlcNAc, which is positioned at the crossroad of metabolic pathways (6)(7)(8). Moreover, glucose transport and consumption is upregulated in cancer cells. This alteration of metabolism is called the "Warburg effect," Otto Warburg having devoted a large part of his research activities to "the origin of cancer cells" (9). The initial observation made by Cori and Cori (10) concluded that a hen's wing having Rous sarcoma produced more lactic acid than the normal wing. Therefore, it was suggested that a deficiency in glucose metabolism was responsible for carcinogenesis even if it is now accepted that it is rather a consequence. Interestingly, cancer cells not only increase glucose consumption but also use more glutamine (11). Glutamine is necessary for both non-essential amino-acids and for purine/ pyrimidine base synthesis in highly proliferative cells (11). In cancer cells, glucose is also used for production of amino-acids and lipids. A large part of the increased-glucose flux is diverted toward the pentose phosphate pathway (PPP) to generate deoxyribose needed for the synthesis of DNA. Interestingly, both glutamine and glucose are limiting substrates of the hexosamine biosynthetic pathway (HBP) (12). HBP results in UDP-GlcNAc synthesis subsequently used by OGT to modify proteins (6). Consequently, through O-GlcNAcylation, HBP activation is involved in cellular structure construction as well as regulation of metabolic flux, signaling pathways, cell homeostasis processes, and tumorigenesis. Increased O-GlcNAcylation levels have been reported in diverse kind of cancers: breast (5,13,14), lung (15), liver (16), prostate (17), chronic lymphocytic leukemia (18), colon (15,(19)(20)(21), and colitis-associated cancer patients (22). Regarding these recent observations, we wondered whether silencing OGT expression affects biological properties of colon cancer-derived cell lines (HT29 and HCT116) in comparison to a normal colon cell line (CCD841CoN, derived from a fetus).
Although O-GlcNAcylation and its cycling enzymes are more elevated in cancer cell lines, we found that both normal and cancer cell lines are impacted by siOGT. While the fetal colon cells are highly mobile, we demonstrate that OGT knockdown slows down their migration property in addition to a disturbance of the cytoskeleton. Together, these data show that OGT is crucial for the biological properties of cancer and non-cancer cell lines. However, we suggest that cancer cell lines are highly sensitive to metabolic deprivation and OGT knockdown, accordingly to their high-demand in nutrients, in particular glucose and glutamine.

MaTerials anD MeThODs
cell culture and Transfection of sirna HT29 and HCT116 cells were maintained in a Dulbecco's modified Eagle's medium (DMEM) and CCD841CoN cells were maintained in an Eagle's Minimum Essential Medium (EMEM) (Lonza) ( Table 1). The three cell lines were maintained in a medium supplemented with 10% (v/v) fetal calf serum, 2 mM l-glutamine, 5 IU/mL penicillin, and 50 μg mL −1 streptomycin at 37°C in a 5% (v/v) CO2-enriched humidified atmosphere. Cells were reverse-transfected with Lipofectamine RNAiMax (Life Technologies) according to manufacturer's instructions using 5 nM small interfering RNA targeting OGT (21)  cell adhesion assay
resUlTs hT29 and hcT116 cells exhibit higher levels of O-glcnacylation cycling enzymes compared to ccD841con Two colon cancer cell lines, HT29 and HCT116, respectively, derived from an adenocarcinoma and from a carcinoma ( Table 1) and the fetal colon cell line CCD841CoN were analyzed by Western blot to determine to the levels of O-GlcNAcylation, OGT, OGA, and Glutamine:Fructose-6-P amidotransferase (GFAT), the rate-limiting enzyme of the HBP (Figures 1A,B).
As shown by western blotting results, O-GlcNAcylated proteins levels, as well as OGA, OGT, and GFAT protein expression were much more elevated in cancer cell lines compared to the fetal cell line (Figure 1B), in accordance with our previous study (23). We next checked whether the three cell lines' O-GlcNAcylation levels were dependent upon glucose and glutamine, the two pivotal substrates of the HBP (Figure 1A). Thus, overnight depletion of either glucose or glutamine reduced O-GlcNAcylation processes, more particularly in HT29 and HCT116 cells (Figure 1C), which are high consumers of these two nutrients like most cancer cell lines (7)(8)(9)(10)(11). Moreover, we noticed that CCD841CoN were particularly sensitive to glucose starvation, leading to a high mortality rate. Therefore, we next assessed the impact of OGT knockdown on the biological properties of these cell lines.

OgT Knockdown Decreases cell Proliferation, reduces In Vitro cell survival, and impairs cell adhesion
The efficiency of siOGT was checked for the three cell lines. As expected, transfection of HCT116, HT29, and CCD841CoN with siOGT decreased the expression of the glycosyltransferase and drastically reduced O-GlcNAcylation levels (Figure 2).
We previously observed that the inhibitor of OGA, NButGT (24), accelerated the proliferation rate of MCF7 cells (23), whereas the potent OGT inhibitor, Ac-5SGlcNAc (25), slightly delayed cell proliferation (26). In the present study, we tested the impact of siOGT on the proliferation rate of CCD841CoN,  HT29, and HCT116 cells (Figures 3A-C, respectively). As expected and in agreement with our previous observations (26), OGT silencing decelerated the proliferation rate of the three cell lines, with an average 20% decrease for CCD841CoN and HT29, and 45% decrease for HCT116 ( Figure 3D). To go further, we tested the ability of colon cells to grow into a colony (in vitro cell survival assay) in response to siOGT (Figure 4A). Both for HT29 and HCT116 cells, the reduction of OGT expression dramatically decreased in vitro survival compared with siCtrltransfected cells (Figure 4B). The ability of a single cell to grow into a colony is characteristic of cancer cells. Accordingly, we were unable to assess the formation of colonies for the primary cell line CCD841CoN. These experiments showed that OGT and, consequently, O-GlcNAcylation are needed for colon cells growth.
We next performed cell adhesion assays for the indicated time periods (Figure 5). While siOGT decreased the adhesion of the three cell lines, we observed a higher impact of OGT silencing on the adhesion of HT29 and CCD841CoN cells. OGT is, therefore, necessary for adhesion of colon cells.

OgT Knockdown reduces Migration of colon cells
We wondered whether OGT is involved in the migration of colon cells. For this matter, we used two distinct approaches: the wound healing assay using culture-inserts μDish and the Transwell system assay, a more quantitative method. Wound healing was assessed 24 h after the culture-inserts were removed (Figure 6). While CCD841CoN were able to fill in the cell-free gap in 24 h, the gap separating the two edges of the migratory HT29 cells was tight at t = 24 h. However, HCT116 only slowly recovered the wound after 24 h. OGT knockdown impaired the closure of the wound for CCD841CoN and HT29 cells as no significant migration was measured after siOGT treatment in comparison to siControl. Nevertheless, even if the migration of the HCT116 cells was slow, siOGT also delayed significantly this process ( Figure 6B). Furthermore, these first observations were corroborated by the more quantitative Transwell assay ( Figure 7A). HCT116 cells did not display any significant migration across the filter pores even in the siControl conditions.
Regarding HT29, while these cells harbored a weak migration in Boyden chambers, a significant decrease was measured when OGT was silenced. Surprisingly, despite their size, CCD841CoN cells exhibited a high capacity of migration compared to HT29 and HCT116 in Boyden chambers. OGT silencing dramatically reduced their migratory capabilities, in accordance with wound healing assays. We hypothesized that a larger cytoskeleton network could confer the high-migratory properties of CCD841CoN cells of which the epithelial origin has not been clearly established (Table 1). First, we showed that actin, the major component of the microfilament network, was more heavily expressed in CCD841CoN cells compared to HT29 and HCT116 cells (Figure 7B). Interestingly, siOGT did not impact the expression of both actin and tubulin, another major component of the cytoskeleton network. Analysis of the cytoskeleton by confocal microscopy (Figures 7C-E) indicated that the cytoskeletal networks are more elaborated in CCD841CoN, which reinforces the doubt about the epithelial origin of this cell line according to ATCC. Surprisingly, OGT silencing did not impact the expression of actin, tubulin, and vinculin, a cytoskeletal protein associated with cell-cell and cell-matrix junctions. On the other hand, siOGT greatly affected the cytoskeletal networks and cell morphology in CCD841CoN cells. The cell shape appeared stocky and stunted, whereas the microfilament network, responsible for cell migration, was less extended.

DiscUssiOn
O-GlcNAcylation is fundamental for cell viability as demonstrated by the non-viability of OGT-depleted ES cells (27). Whereas three isoforms were described for OGT (nucleocytoplasmic, short, and mitochondrial isoforms), this enzyme is encoded by a single gene and downregulating its expression or inhibiting its catalytic activity affects major fundamental biological functions (3,26,27). Moreover, a deregulation in the O-GlcNAcylation cycling is involved in the etiology of diverse pathologies, such as type-2 diabetes, Alzheimer's, and cancers (6). Intriguingly, due to the correlation between UDP-GlcNAc, O-GlcNAcylation, and nutrients availability, diseases associated with unhealthy diet and metabolic disorders may be tightly linked to O-GlcNAcylation imbalance (28). It is, therefore, crucial to better understand through which metabolic mechanisms lifestyles negatively impact human health. Particularly, some cancers, among which colorectal cancers, are synergistically favored by metabolic problems and unhealthy lifestyles (29,30). Elevated-O-GlcNAcylation level was described in many cancers, even if the reasons and impacts of this increased level remain obscure. For example, breast cancer cells have increased-O-GlcNAcylation and OGT levels, and a reduction of OGT expression correlated to tumor growth decrease (5). High levels of OGT and its product, O-GlcNAcylation, were also reported in colon cancers and cancer cell lines (15,(19)(20)(21) and colitisassociated cancer patients (22) but very few initiatives to measure the effect of OGT knockdown were conducted. Among those few studies, OGA inhibition or shOGT exhibited no significant effect on HT29 cell invasion but enhanced and reduced, respectively, their anchorage-independent growth (15). Furthermore, OGT knockdown inhibited anchorage-independent growth (14), but contradictory results were also reported (19). This last study indicated that OGA silencing changed the anchorage-independent growth and the morphology of the primary colorectal cancer cell SW480 and of its metastatic clone SW620 (19). Whereas one previous study showed altered gene expression related to actin reorganization and cell migration in siOGA-transfected cells (19), we found that OGT silencing also modifies cell shape features of the fetal human colon cell line CCD841CoN. In our hands, siOGT did not modify the expression of actin but we suspected that actin-binding proteins are up-or downregulated when OGT is silenced. In that sense, OGT promotes breast cancer cells invasion in a cofilin-dependent manner. O-GlcNAcylation of cofilin at Ser108 localizes this actin-interacting protein to invadopodia (31). Also, under conditions of low OGT expression, actin may be itself affected by defective O-GlcNAcylation. In a previous work, we mapped a major O-GlcNAcylation site within the 318-324 region of β-actin expressed in Xenopus laevis oocytes (32) and later, one O-GlcNAcylation site was localized in the domain four of rat actin (33). Nevertheless, the function of actin O-GlcNAcylation remains to be established. In parallel, O-GlcNAcylation was also widely studied on tubulin, another major component of the cytoskeleton network. O-GlcNAcylation of α-tubulin reduces heterodimerization of α/β-tubulins and O-GlcNAcylated forms of tubulins are unable to polymerize into microtubule (34). Moreover, α-tubulin is heavily O-GlcNAcylated in primary colorectal cancer (20). These two independent studies tend to support our observations of a disorganization of microtubules in the primary colon cell line, while no significant effect of siOGT was found in the two colon cancer lines.
Beyond the effect of O-GlcNAcylation on structural proteins in a pathologic context, downregulation of OGT must interfere with the expression and/or the activity of regulatory proteins. A comparison between primary breast malignant tumors and benign tumors revealed the O-GlcNAcylation of crucial components of the "Warburg effect" only in cancer (14). One of the characteristics of cancer cells is the shift from an oxidative to a non-oxidative consumption of glucose. Oncogenic signaling pathways controlling the transcription factor hypoxia-inducible factor-1 (HIF1) alpha are responsible for this metabolic shift.
HIF1α stability is dependent upon O-GlcNAcylation level (35). GLUT1 expression, one of the HIF1α's target genes, is more heavily expressed when OGT is activated. Consequently, glucose transport into the cell is increased. Most of the glycolytic enzymes are modified by O-GlcNAcylation (32,36,37). Among those, the enzyme phosphofructokinase-1 (PFK1) (38) controls the entry of glucose into glycolysis. O-GlcNAcylation of PFK1 at Ser509 prevents the binding of the activator Fru-2,6-bis-phosphate. Consequently, this modification diverts the use of glucose to the PPP to produce pentoses and NADPH2, respectively, used for nucleic acids and lipids biosynthesis. This confers an advantage for cancer cell to increase their proliferation rate.
O-GlcNAc transferase is critical for normal cells and cancer cell homeostasis and adaptation to environment. Due to the plethora of OGT's targets, it is difficult to assign precisely the impact of OGT silencing. In light of the different elements exposed above, we suggest that a default of O-GlcNAcylation impacts on cell architecture as attested by the alteration of morphology observed in CCD841CoN cells and of metabolic routes. Moreover, knocking-down OGT also results in inactivation of mitogen signaling pathways as previously established (3,26,39,40).
Our observations indicate that OGT is crucial for the biological properties of normal colon-derived cells and colon cancer cell lines. However, colon cancer cells express higher amounts of OGT and O-GlcNAcylation than normal cells. Due to the addiction of cancer cells for glutamine and glucose (7)(8)(9)(10), the main substrates of HBP, it could be suspected that cancer cells were much more sensitive to changes in O-GlcNAcylation levels than normal cells while we found that both colon cancer and primary cell lines were affected by OGT silencing. OGT also interferes with cell migration, especially for the fetal cell line CCD841CoN, by reducing the size of the actin network, which participates in the alteration of the cell morphology. Along this study, we demonstrated that OGT impacts proliferation and migration of normal as well as cancer colon cells. This work highlights how O-GlcNAcylation, through the modulation of basic biological features, controls the properties of primary cells and also might initiate nutrientresponsive cancers, such as colorectal cancer.
aUThOr cOnTriBUTiOns AS designed, performed, and analyzed the experiments, and wrote the experimental procedures. SS and VD initiated the project, conceived, and performed experiments. SS edited the manuscript. SB and R-AT performed and analyzed the experiments. XB analyzed the experiments and managed the study at INSERM U908. IY-B designed, performed, and analyzed the