Efficient Inhibition of O-glycan biosynthesis using the hexosamine analog Ac5GalNTGc

There is a critical need to develop small molecule inhibitors of mucin-type O-linked glycosylation. The best known reagent currently is peracetylated benzyl-GalNAc, but it is only effective at millimolar concentrations. This manuscript demonstrates that Ac5GalNTGc, a peracetylated C-2 sulfhydryl substituted GalNAc, fulfills this unmet need. When added to cultured leukocytes, breast and prostate cells, Ac5GalNTGc increased cell surface VVA-binding by ~10-fold, indicating truncation of O-glycan biosynthesis. Cytometry, mass spectrometry and Western blot analysis of HL-60 promyelocytes demonstrate that 50-80μM Ac5GalNTGc prevented elaboration of 30-60% of the O-glycans beyond the Tn-antigen (GalNAcα1-Ser/Thr) stage. The effect of the compound on N-glycans and glycosphingolipids was small. Glycan inhibition induced by Ac5GalNTGc resulted in 50-80% reduction in leukocyte sialyl-Lewis-X expression, and L-/P-selectin mediated rolling under flow. Ac5GalNTGc was pharmacologically active in mouse. It reduced neutrophil infiltration to sites of inflammation by ~60%. Overall, Ac5GalNTGc may find diverse applications as a potent inhibitor of O-glycosylation.


INTRODUCTION
All mammalian cells display mucin-type O-linked GalNAc-type glycans as part of their glycocalyx. They play a number of key biological roles during development, tumorigenesis, cancer metastasis, leukocyte adhesion and inflammatory response (Chugh et al., 2018;Oliveira-Ferrer et al., 2017;Schnaar, 2016;Tran and Ten Hagen, 2013). The biosynthesis of mucin-type O-glycans is initiated by the transfer of GalNAc (N-Acetylgalactosamine) from its nucleotide-sugar donor (UDP-GalNAc) to Ser/Thr residues on the peptide backbone by a family of polypeptide GalNAc-transferases (ppGalNAcT, (Bennett et al., 2012)). There are 20 ppGalNAcTs in humans, and these are largely conserved in the animal kingdom. Many of these enzymes act on proline-rich peptides to result in -GalNAc-Ser/Thr (Tn-antigen) bearing glycopeptides. Other 'followup enzymes' (ppGalNAcT-T4, -T7, -T10, -T12, -T17) prefer to act on previously GalNAcylated glycopeptides to promote the formation of O-glycan clusters on mucinous proteins (Bennett et al., 2012). While the overlapping tissue expression patterns and substrate specificity of the ppGalNAcTs result in some functional redundancy, the second step of the O-glycan biosynthesis pathway is highly specific. Here, Galβ1→3 addition is mediated by a single, ubiquitously expressed enzyme called Core 1 β1,3Galactosyltransferase (C1GalT1) along with its chaperone protein Cosmc (Core 1 β1,3GalT-specific molecular chaperone) . The Galβ1,3GalNAcα-Ser/Thr core is then elaborated by additional glycosyltransferases (Brockhausen and Stanley, 2015).
The development of specific, small molecule inhibitors of mucin-type O-glycans would be beneficial for mechanistic studies and also translational applications. This would fill a need in the field since many of the current glycosylation pathway inhibitors only target either glycosidase function (e.g. castanospermine, thiamet-G, oseltamivir), N-glycan biosynthesis (swainsonine, deoxymannojirimycin, tunicamycin), or glycosphingolipid (GSL) biosynthesis (N-butyldeoxynojirimycin; D-threo-1-phenyl-2-palmitoylamino-3morpholino-l-propanol PPMP) (Gloster and Vocadlo, 2012;Hudak and Bertozzi, 2014). Attempts have been made to develop such O-glycosylation inhibitors with focus on GalNAc as these are uniquely part of Olinked glycans, although they also appear in dermatan sulfates, chondroitin sulfates and a subset of GSLs.
Such compounds are often peracetylated to enhance cell permeability. In this regard, studies using various GalNAc analogs suggest that C-6 modification of GalNAc may render the compound inactive as it is not activated by GalNAc-1-kinase in the salvage pathway (Pouilly et al., 2012). However, C-2 N-acyl modified GalNAc are tolerated by cells and incorporated into cell surface glycoconjugates to varying degrees depending on their chemical composition (Dube et al., 2006;Hang et al., 2003;Pouilly et al., 2012). Such compounds are however reported to function bio-orthogonally, without inhibition effects. Peracetylated 4F-GalNAc is another C-4 analog that acts as a glycosylation inhibitor at 50-100μM concentrations (Marathe et al., 2010). Like peracetylated 4F-GlcNAc, however, this compound is not directly incorporated into glycoconjugates (Barthel et al., 2011), and both compounds reduce depress UDP-HexNAc levels (Del Solar et al., 2020). Finally, the most common O-glycan inhibitor used currently is the decoy substrate GalNAc-O-Bn ('benzyl-α-GalNAc') that acts as an effective surrogate substrate when applied at high concentrations (2-4mM) (Alfalah et al., 1999;Huet et al., 1998;Kuan et al., 1989;Tsuiji et al., 2003). At lower dose (25-100μM), peracetylated GalNAc-O-Bn only acts as a primer that reports on the cellular carbohydrate biosynthesis pathways with minimal inhibitory function (Kudelka et al., 2016;Stolfa et al., 2016).
Previously, we reported a peracetylated N-thioglycolyl modified GalNAc analog ('Ac5GalNTGc'), which trimmed O-glycans globally, including on CD43 of Jurkat (Agarwal et al., 2013) and U937 cell lines (Dwivedi et al., 2018). In the current manuscript, we extended the characterization of this compound, with focus on its effect on leukocyte cell adhesion function, mechanisms of action and in vivo investigations of leukocyte recruitment. These studies contrast the function of Ac5GalNTGc with a panel of other peracetylated C-2 substituted GalNAc analogs, and also its peracetylated C-4 epimer, 'Ac5GlcNTGc' (Figure 1). These compounds are per-acetylated to enhance cell permeability. Once they enter cells from the culture medium, they are de-esterified in the cytosol and transported into the Golgi. Here, they either participate in biosynthetic processes or act as metabolic inhibitors. In such investigations, our data show that: i. Ac5GalNTGc effectively abolishes sialyl Lewis-X (sLe X ) epitope expression on human leukocytes when applied at 50 μM, and inhibits leukocyte rolling on L-and P-selectin ex vivo. Such binding is largely dependent on leukocytic O-glycans (Kieffer et al., 2001;Lo et al., 2013;Vestweber and Blanks, 1999).
Whereas the concentration in the extra-cellular milieu is in the ~50μM range, intra-cellular concentration is higher it ~1-1.5 mM (estimated in (Marathe et al., 2010)). ii. Addition of this compound to different cell types resulted in a dramatic upregulation of VVA-lectin binding, supporting the potential that Ac5GalNTGc acts by inhibiting core-1 glycan elaboration. iii. Although only a minor portion of the Ac5GalNTGc was incorporated into cellular glycoconjugates, upregulation of VVA-lectin binding correlated tightly with the extent of Ac5GalNTGc incorporation. Ac5GalNTGc did not alter cellular nucleotide-sugar levels or N-glycan biosynthesis. It had a quantitative effect on the relative abundance of GSLs. iv. Ac5GalNTGc was pharmacologically active in mice and it inhibited the extent of neutrophil recruitment to sites of inflammation in an acute peritonitis model. Overall, Ac5GalNTGc is a mucin-type O-glycosylation inhibitor. It is more potent compared to other molecules commonly used in literature, and thus could be broadly useful in diverse basic science and translational applications.

Ac5GalNTGc decreased cell surface sialyl Lewis-X expression: The effect of the panel of peracetylated
HexNAc analogs on cell surface carbohydrate expression was evaluated by culturing HL-60 promyelocytes with 50 μM of each of the analogs for 40h (Fig. 2). Among them, only Ac5GalNTGc significantly altered cell surface glycan structures. It doubled CD15/Le X expression ( Fig. 2A), reduced the expression of sLe X as determined using mAbs HECA-452 (Fig. 2B) and CSLEX-1 by 50-80% (Fig. 2C), and increased the VIM-2/CD65s epitope (Fig. 2D). Other compounds containing different C-2 substituents, and also the C-4 epimer Ac5GlcNTGc had no effect on glycan expression. A reduction in HECA-452 (CLA/sLe X ) expression, upon culture with Ac5GalNTGc, was also observed in western blots (Supplemental Fig. S1A). The increase in Le X expression is similar to prior work where CRISPR-Cas9 based inhibition of O-glycan biosynthesis in HL-60s uncovered sterically hidden CD15 epitopes on leukocyte N-glycans and glycolipids (Stolfa et al., 2016). In dosage studies, Ac5GalNTGc modified glycan structures at concentrations as low as 10 μM, with maximum efficacy at > 50 μM ( Fig. 2E-2H). None of the concentrations tested altered cell viability or growth rate over the first 40h, though treatment with 200 μM Ac5GalNTGc resulted in a longer lag-phase before resumption of growth following Ac5GalNTGc removal (Supplemental Fig. S1B-D). Overall, GalNAc with a thioglycolylamino-moiety at the C-2 position is a potent modifier of glycosylation. Based on these data, 80μM peracetylated Ac5GalNTGc was applied in all subsequent studies described below, unless stated otherwise.

Ac5GalNTGc truncates O-glycan biosynthesis:
A panel of lectins was applied to characterize changes in carbohydrate structures upon culture with Ac5GalNTGc (Fig. 3, Supplemental Fig. S2). Studies were performed both in the presence or absence of sialidase, as some lectins preferentially bind de-sialylated epitopes. Here, Ac5GalNTGc augmented VVA binding by 30-fold compared to vehicle treatment (Fig. 3A).
Similar observations were also made with another GalNAcα binding lectin, Soyabean Agglutinin SBA (Fig.   S2). In agreement with this, pronounced VVA binding to cells was observed in the fluorescent micrographs upon culture with Ac5GalNTGc (Fig. 3F). Compared to the sialidase treated [O]¯ cells (COSMC-knock out) which augmented VVA binding by 43-fold, Ac5GalNTGc was 13-fold effective (Fig. 3A). Thus, Ac5GalNTGc truncates 1/3 rd of the O-glycans at the Tn-antigen stage. Similar observations were also made in lectin blots, where Ac5GalNTGc dramatically elevated VVA binding to a variety of glycoproteins (Supplemental Fig.   S3A). Besides leukocytic cells, a similar increase in VVA-binding was also noted on other breast (T47D, ZR-75-1) and prostate (PC-3) cell lines, but not HEK293T kidney cells (Supplemental Fig. S3B). Consistent with the notion that O-glycans are specifically altered, a 30% reduction in the sialylated T-antigen was observed in HL-60s, upon using PNA (Fig. 3B, Fig. S2). In addition, Ac5GalNTGc decreased the N-acetyllactosamine structures reported by ECL by ~40% (Fig. 3C). α2,3-sialic acid measured using MAL-II (Fig.3D), complex N-glycans reported by PHA-L (Fig. 3E) and a panel of additional 15+ lectins that report on Man, Fuc, GlcNAc and Gal related epitopes (Fig. S2) remained largely unchanged. Overall, the data suggest a potent effect of Ac5GalNTGc on cell-surface O-glycans, with less effect on other types of glycoconjugates.

Ac5GalNTGc reduced leukocyte adhesion to L-and P-selectin:
Microfluidics based flow chamber studies determined if the reduced sLe X expression upon Ac5GalNTGc treatment attenuated leukocyte rolling on selectin-bearing substrates (Fig. 4). Here, Ac5GalNTGc treated HL-60s displayed 80% and 50% reduction in cell rolling on recombinant L-selectin (Fig. 4A) and on P-selectin bearing CHO-P cells (Fig. 4B), respectively. This reduction was not observed upon treatment with vehicle or GalNAc. In contrast to L-and P-selectin, Ac5GalNTGc did not alter E-selectin dependent leukocyte rolling on IL-1β stimulated HUVEC monolayers (Fig. 4C). P-selectin dependent leukocyte-platelet adhesion was also reduced by 40-50% upon cell culture with Ac5GalNTGc (Fig. 4D). Consistent with these functional data, Ac5GalNTGc reduced the apparent molecular mass of the major human L-/P-selectin ligand PSGL-1 by 15% (from 125 to 105KDa, Fig. 4E). It also reduced the mass of another common mucinous leukocyte glycoprotein, CD43. Together, the data demonstrate that O-glycan biosynthesis truncation by Ac5GalNTGc can reduce L-and P-selectin dependent leukocyte cell adhesion under shear, both in the context of leukocyte-endothelium and leukocyte-platelet binding.

Ac5GalNTGc is pharmacologically active and it reduces granulocyte migration to sites of inflammation:
To determine if Ac5GalNTGc is active in vivo, either mouse bone marrow cells (mBMCs) or neutrophils (mPMNs) were cultured with IL-3 (interleukin-3) and GCSF (granulocyte colony stimulating factor) ex vivo in the presence Ac5GalNTGc or control treatments for 40h (Fig. 5A). Similar to HL-60s, both Ac5GalNTGc treated mBMCs and mPMNs displayed a 6-fold increase in VVA binding at 40h (Fig. 5B).
Ac5GalNTGc also reduced P-selectin-IgG binding to neutrophils by 50-65% (Fig. 5D) and L-selectin binding by 60-80% (Fig. 5E). PSGL-1 expression remained unchanged (Fig. 5C). In another set of studies, mBMCs cultured with Ac5GalNTGc or vehicle were labeled with distinct fluorescent dyes (either red or green), mixed in equal proportion, and injected i.v. into recipient mice in a thioglycollate-induced model of acute peritonitis.
Twenty hours later, the peritoneal lavage was recovered, and the labeled neutrophils were enumerated using flow cytometry (Fig. 5F). Here, regardless of the labeling dye combination, Ac5GalNTGc reduced neutrophil migration into the peritoneum by ~50%, compared to vehicle control.
Next, to confirm the pharmacological activity of Ac5GalNTGc, 100mg/kg Ac5GalNTGc was infused into recipient mice once daily for 4-days before induction of peritonitis ( Fig. 5G). Here, also, Ac5GalNTGc reduced neutrophil extravasation into the peritoneal lavage by 65% (Fig. 5H). Peripheral blood counts and leukocyte differentials remained unchanged at day-4 (Table S1), and animals did not exhibit any signs of distress or abnormality. Significantly, the VVA binding to neutrophils in the peritoneal lavage was two-fold higher upon Ac5GalNTGc treatment (Fig, 5I). Together, the data confirm the in vivo metabolic activity of Ac5GalNTGc.

Ac5GalNTGc blunts T-antigen biosynthesis, with less effect on N-glycans and glycolipids:
We verified the effect of Ac5GalNTGc on overall O-glycosylation by feeding peracetylated GalNAc-OBn to cells cultured with Ac5GalNTGc or GalNAc (control), and, measuring extended glycans formed on this substrate ( Fig. 6A). All products formed were verified based on MS/MS and also LC retention time.
[NOG]¯ HL-60s fed with GalNAc-OBn served as negative controls, since they lack the O-glycan biosynthesis machinery.
Here, whereas 35% of the GalNAc-O-Bn was converted into extended O-glycans (disaccharides, trisaccharides etc.) in the GalNAc fed cells, this was reduced to 14% upon culture with Ac5GalNTGc and such biosynthesis was absent on [NOG]¯ cells. These data suggest ~60% (=21/35×100) inhibition of Oglycan elaboration upon Ac5GalNTGc treatment. The increased prevalence of exposed unmodified GalNAc upon culture with Ac5GalNTGc is consistent with the increased VVA binding noted previously (Fig. 3). Due to reduced GalNAc-O-Bn substrate extension, longer glycan chains including those containing the Tantigen (Galβ1-3GalNAc/core-1) and core-2 glycan (Galβ1-3[GlcNAc β1-6]GalNAc) related structures were reduced in the Ac5GalNTGc treated cells. Such reduction in O-glycan extension could account for the reduced molecular mass of PSGL-1 and CD43. Notably, consistent with cytometry measurements, sLe X epitope biosynthesis on core-2 O-glycans was reduced by ~70% in the Ac5GalNTGc treated cells. The direct incorporation of GalNTGc/HexNTGc (in non-acetylated or partially acetylated form) into carbohydrate products formed on the GalNAc-OBn substrate was not observed. In enzymology studies, a ~40% reduction in β1,3GalT activity was noted in the HL60s upon culture with Ac5GalNTGc, and this may partially explain the increased expression of the Tn-antigen epitope (Fig. 6B).
MALDI-TOF MS based glycomics profiling was undertaken to study the impact of Ac5GalNTGc on cellular N-glycans and GSL biosynthesis. Here, Ac5GalNTGc did not have any significant impact on Nglycan structures, with high molecular weight complex structures still being observed (Fig. 6C). In the case of GSLs, however, Ac5GalNTGc appeared to quantitatively affect the abundance of certain GSLs mainly containing fucose residues (m/z 1566, 1841, 2016, 2639, 3088 and 3262, Supplemental Fig. S4). Direct incorporation of HexNTGc into glycoconjugates was not detected. Overall, the data suggest that Ac5GalNTGc reduces the formation of the T-antigen on leukocyte O-glycans and it also has some impact on GSLs. The effect of the compound on N-glycan biosynthesis was negligible.

Minimal changes in sugar-nucleotide biosynthesis and low levels of Ac5GalNTGc derivative incorporation into cellular O-glycans, N-glycans and glycolipids:
Recent studies suggest that monosaccharide analogs, where selected hydroxyl groups are modified with different substituents, often act as metabolic inhibitors by altering cellular sugar-nucleotide compositions (Del Solar et al., 2020;Gloster and Vocadlo, 2012;van Wijk et al., 2015). However, this is not the mechanism of action of Ac5GalNTGc since it did not alter cellular sugar-nucleotide levels based on LC-MS/MS (Fig. 7A). Here, glucose based sugar-nucleotide standards were distinguished from corresponding galactose counterparts based on retention time since compounds containing Glc eluted first, i.e. UDP-Glc eluted prior to UDP-Gal, and UDP-GlcNAc before UDP-GalNAc (Del Solar et al., 2020). Based on these observations, the MS/MS fragmentation pattern of UDP-HexNTGc (Supplemental Fig. S5A) and the observed elution chromatogram ( Fig. S5B), our data suggest the possible conversion of Ac5GalNTGc into both UDP-GalNTGc and UDP-GlcNTGc in equal parts (Fig. 7B). While exact quantitation is not possible due to the absence of UDP-HexNTGc standards, MS ion counts of UDP-GalNTGc and UDP-GlcNTGc in cell lysates was 10-15 fold lower than that of UDP-GalNAc and UDP-GlcNAc suggesting only small amounts of UDP-HexNTGc formation.
The direct incorporation of Ac5GalNTGc and its derivatives into glycoconjugates was not observed in MS studies (Fig. 6), potentially due to their low abundance, below the instrument detection limit. Low levels of GalNTGc incorporation was also observed in O-and N-glycans in LC-MS/MS glycoproteomics investigations (data not shown). However, this could be readily detected using fluorescence methods. Here, consistent with a previous report (Agarwal et al., 2013), we observed a 10-15 fold increase in FITCmaleimide (5-FM) incorporation upon culture with Ac5GalNTGc, but not in controls containing Ac5GlcNTGc or Ac5GalNAc (Fig. 7C). 5-FM incorporation was 15-fold higher when the labeling reaction was performed in the presence of the reducing agent TCEP, suggesting that a majority of the cell-surface thiol groups introduced by Ac5GalNTGc were crosslinked via disulfide bridges under native conditions, i.e. they exist as GalNTGc-GalNTGc or GalNTGc-Cys moieties. Similar results were obtained for cell surface thiol measurement by flow cytometry wherein the two-step maleimide-PEG2-biotin Michael addition was followed by FITC-conjugated avidin staining (Supplemental Fig. S6A). Cell surface incorporation of GalNTGc derivatives was also observed using fluorescence microscopy ( Fig. 7D), but not in controls runs that used Ac5GlcNTGc or Ac5GalNAc (Fig. S6B). Ac5GalNTGc and its derivatives were maximally incorporated 24-48h post-treatment, with 5-FM signal being reduced to basal levels at 72-96h (Fig. 7E). Besides HL-60s, 5-FM was also incorporated into other human cell lines including breast (T47D, ZR-75-1), prostate (PC-3) and to a lesser degree into kidney (HEK293T) cells (Fig. S6C).
Studies were performed with a panel of CRISPR-Cas9 HL-60 knockouts that contain truncated Oglycans, N-glycans and/or GSLs, in order to determine the glycoconjugates that incorporate GalNTGcderivatives (Fig. 7F). This analysis suggests the increased prevalence of sulfhydryl groups in all families of cell-surface glycans, with incorporation being somewhat higher in GSLs (~40-50%) compared to N-glycans (~25-35%), followed by O-glycans (~20-25%). These estimates are based on the quantitative incorporation of 5-FM signal in the single and double knockout cell lines compared to wild-type HL60s. 5-FM signal was low/absent in the triple knockouts and thus this signal comes from thiol incorporation into glycans and not into other macromolecules. It could also be reduced upon protease digesting cell surface glycoproteins.
Consistent with the above, maleimide incorporation was observed in western blots of PSGL-1 with greater maleimide incorporation being noted for lower molecular mass glycoproteins that contain more truncated O-glycans (Fig. S7A). Additionally, Ac5GalNTGc may not be transformed into sialic acid using the pathway illustrated in Fig. S7B, since the removal of sialic acids by α2-3,6,8,9-neuraminidase did not reduce the measured 5-FM signal. Overall, the data are consistent with the notion that a portion of Ac5GalNTGc is converted into UDP-derivatives that are incorporated into cellular glycoproteins and possibly also GSLs.

DISCUSSION
Our results demonstrate that Ac5GalNTGc is a potent metabolic inhibitor of O-glycan biosynthesis in diverse cell types including mouse peripheral blood neutrophils, human promyelocytes, breast and prostate cancer cell lines. In all these cells, Ac5GalNTGc upregulated VVA binding suggesting that it may reduce core1 β1,3GalT1 (C1GalT1) activity. Upon quantitatively comparing VVA binding on Ac5GalNTGc treated cells with COSMC-knockouts that completely lack C1GalT1 activity, it is estimated that at least ~1/3 rd of the HL-60 O-glycans are truncated by Ac5GalNTGc as GalNAc bearing polypeptides with no extension. This conclusion is consistent with the ~20-30% decrease in molecular mass of mucinous proteins including PSGL-1 and CD43 on the leukocytes. Additionally, feeding peracetylated GalNTGc to these cells reduced Gal incorporation into benzyl-GalNAc by ~60% with respect to GalNAc control, with few extended O-glycan structures. N-glycan profiling using MALDI TOF MS confirmed minimal effect of Ac5GalNTGc on N-glycan biosynthesis. The compound, however, appeared to affect the abundance of some fucosylated GSLs via yet unidentified mechanisms. Finally, although, the current study did not examine glycosaminoglycan and O-GlcNAc type modifications, the results thus far suggest that the major impact of Ac5GalNTGc is on O-glycan biosynthesis with the compound reducing the elaboration of such entities by ~30-60%. Additionally, Ac5GalNTGc has greater potency compared to peracetylated GalNAc-O-Bn, a previously described O-glycan inhibitor which is added at 2-4mM into cell culture medium to reduce O-glycan biosynthesis (Alfalah et al., 1999;Huet et al., 1998;Kuan et al., 1989;Tsuiji et al., 2003). In contrast, the functional effect of Ac5GalNTGc was observed at concentrations as low as 10µM with maximum efficacy at 50-80µM. The lower usage dose and favorable pharmacological properties allow systemic usage of Ac5GalNTGc in murine models.
The functional effects of Ac5GalNTGc on O-glycan biosynthesis inhibition was assessed in a model of inflammation where leukocytes were recruited onto selectin-bearing substrates under shear flow. Here, Ac5GalNTGc dramatically reduced cell surface sLe X expression as measured using mAbs HECA-452 and CSLEX-1, and also using O-glycan mass spectrometry analysis. Such inhibition was prominent on mucinous proteins like the P-and L-selectin ligand PSGL-1, as the apparent mass of this glycoprotein was reduced by ~25% upon culture with Ac5GalNTGc. Ac5GalNTGc treatment also reduced leukocyte adhesion on L-and P-selectin, but not E-selectin, under hydrodynamic shear. This is consistent with a previous report that C1GalT1 in necessary for leukocytes recruitment via L-and P-selectin (Stolfa et al., 2016). Abolishing O-glycosylation, however, has only a minor effect on human leukocyte recruitment/tethering and rolling on E-selectin. Besides the effect on cell rolling, Ac5GalNTGc also reduced the extent of P-selectin-PSGL-1 dependent leukocyte-platelet adhesion under shear. Mouse neutrophils (and mBMCs) cultured ex vivo with Ac5GalNTGc for 40h also exhibited higher than normal levels of VVA binding, and reduced interaction with L-and P-selectin IgG fusion proteins. In the complex in vivo milieu during peritonitis, consistent with the inhibition effect of Ac5GalNTGc on L-/P-selectin dependent binding, neutrophil recruitment to sites of inflammation was reduced upon culture of cells with Ac5GalNTGc. Ac5GalNTGc also had excellent pharmacological properties, and it caused 60% reduction in leukocyte recruitment at sites of inflammation.
These data support the use of Ac5GalNTGc in vivo for anti-inflammatory therapy.
Studies were undertaken to determine the mechanism of Ac5GalNTGc action, with focus on β1,3GalT activity, since addition of this compound into cell culture medium both increased VVA binding and drastically reduced Gal incorporation into GalNAc-O-Bn substrate. β1,3GalT enzymatic activity was also partially reduced upon culture with Ac5GalNTGc, consistent with the notion that this compound may reduce core-1 glycan biosynthesis. In such investigations, we did not observe marked changes in the cellular nucleotide-sugar profile of HL60s culture with Ac5GalNTGc, that would indicate non-specific activity of the compound. This is unlike previous studies that used modified monosaccharides, which globally alter the cellular nucleotide-sugar profile (Del Solar et al., 2020;Rillahan et al., 2012;van Wijk et al., 2015). Here, culture of cells with per-acetylated 6F-GalNAc reduced the cellular UDP-GalNAc and UDP-GlcNAc pool by ~80-90% (van Wijk et al., 2015), 2F-Fuc also depressed cellular GDP-Fuc, and 3F-Neu5Ac similarly abolished CMP-Neu5Ac (Rillahan et al., 2012). In these studies, substantial amounts of UDP-(6F)GalNAc, GDP-(2F)Fuc and CMP-(3F)NeuAc were formed and this resulted in collateral reduction in corresponding unmodified nucleotide-sugars. Unlike this, the culturing of cells with Ac5GalNTGc resulted in relatively low levels of UDP-HexNTGc synthesis. Based on ion count data, both UDP-GalNTGc and UDP-GlcNTGc were formed via the salvage pathway, although the levels were 1/10-1/15 th that of UDP-GalNAc and UDP-GlcNAc.
Due to this, only low levels of GalNTGc and GlcNTGc were integrated into cellular glycolipids, N-glycans and O-glycans. This could be detected using sensitive fluorescence based methods (flow cytometry and microscopy), but not mass spectrometry. In this regard, the degree of GalNTGc integration into glycoproteins may be important for functional efficacy since cell systems with greater maleimide-FITC incorporation (e.g. HL-60, T47D, PC3) also displayed greater enhancement of VVA binding. HEK cells, on the other hand, exhibiting low maleimide incorporation and minimal change in VVA engagement.
Transcriptional analysis of these cells does not reveal any obvious differences in O-glycosylation related enzyme expression profiles to explain these observations (Supplemental Table S2), though nucleotidesugar analysis needs to be performed to quantify the efficiency of UDP-HexNTGc synthesis across these different cell types. Additionally, if Ac5GalNTGc reduces T-synthase activity, one possibility is that glycoproteins containing directly incorporated GalNTGc may bond directly with either C1GalT1 or its unique molecular chaperone COSMC, within the Golgi . Such molecular interactions may occur between the thiol residues of GalNTGc-derivatives and free Cys available on C1GalT1/COSMC. These may reduce T-synthase activity. In this regard, previous studies show that both C1GalT1 and COSMC are highly conserved proteins with 363 and 318 amino acid residues, respectively . They both contain 6 highly conserved Cys residues in their luminal/catalytic domain including a pair of vicinal Cys that may be targeted by GalNTGc containing glycoproteins. Alternatively, the steric repulsion of the bulky thiol groups may preclude binding of the GalNTGc-decorated polypeptide acceptors to C1GalT and resulting enzyme activity. Other factors that may contribute to Ac5GalNTGc inhibitory function include: i. potential roles for partially deacetylated GalNTGc or its derivative in regulating glycosylation; ii. inhibition of other enzyme activities besides C1GalT1/COSMC, particularly those related to ppGalNAcT function. Additional studies are needed to examine these hypothesis.
Overall, Ac5GalNTGc is a potent inhibitor of O-linked glycosylation. It fills an important gap in the field that lacks O-glycosylation inhibitors. The compound is pharmacologically active, it reduces the expression of sialofucosylated glycan epitopes on the leukocyte cell surface, inhibits L-and P-selectindependent molecular recognition under static and flow conditions, and displays the potential to have antiinflammatory properties. Besides basic science applications, the compound may find utility in translational studies where there is a need to trim mucinous glycoproteins for example during pulmonary disorders with excess mucin production, and in the context of investigations related to cancer metastasis and immunotherapy.

SIGNIFICANCE
Four common types of glycans are expressed on the surface of mammalian cells. These include Oand N-linked glycans on glycoproteins, glycosphingolipids (GSLs) and glycosaminoglycans (GAGs).
Currently, there are a number of ways to study N-glycan function using enzymes like PNGaseF/Peptide:Nglycosidase F that cleave these structures, and glycosidase inhibitors (e.g. kifunensine) that can truncate N-glycan biosynthesis. Small molecule inhibitors also exist for the study of GSLs (e.g. D-threo-1-phenyl-2-palmitoylamino-3-morpholino-l-propanol/PPMP), and lyases are commonly used to trim GAGs. Unlike these, few reagents are available to study mucin type O-glycosylation.       expression. F. mBMCs cultured with 80μM Ac 5 GalNTGc for 40h were mixed with VC at 1:1 ratio.
In mix 1, Ac 5 GalNTGc cells were labeled with CMTMR (Red) while VC was CMFDA (Green) labeled. Labels were swapped in Mix 2 (dot plot not shown). Mix 1 or 2 cells were tail-vein injected into recipient mice following thioglycollate injection i.p. Red:green ratio of Gr-1+ cells in the peritoneal lavage and bone marrow was measured at 20h. Ac 5 GalNTGc reduced neutrophil counts in peritoneum by 50% in both Mixes. G-I. Ac 5 GalNTGc (100mg/kg/day) or VC was injected daily into mice for 4 days prior to induction of peritonitis. Murine neutrophil (CD11b+, Gr-1/Ly-

LEAD CONTACT AND MATERIALS AVAILABILITY
Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Sriram Neelamegham (neel@buffalo.edu).

EXPERIMENTAL MODEL AND SUBJECT DETAILS
8-12 week-old C57BL/6 wild-type mice of either sex were used. Animals were randomized prior to experimentation. All animal studies were approved by the Roswell Park Cancer Institute Animal
All antibodies were mouse IgGs from BD Biosciences (San Jose, CA) unless otherwise mentioned.
These include anti-CD15/Lewis-X mAb HI98 (IgM), rat anti-Cutaneous Lymphocyte Antigen mAb HECA- Fluorescence microscopy: HL60 cells, with or without GalNAc analog treatment, were labeled with fluorescent VVA lectin as in the above cytometry studies. In some cases, the cells were α2-3,6,8,9neuraminidase treated prior to VVA-labeling. Following fixation using 0.5% paraformaldehyde, the cells were mounted using Prolong Gold Antifade Reagent (Invitrogen) following manufacturer's instructions.

SDS-PAGE and Western Blot:
Cells were lysed using RIPA buffer containing Halt TM protease inhibitor (Thermo-Pierce) for 30-45 min on ice. Following centrifugation at 14,000g, the supernatant was collected and boiled in Laemmli sample buffer containing β-mercaptoethanol. Lysates from ~0.3-1×10 6 cells were loaded in each well of either a standard 7.5% or 4-20% gradient gel (Thermo-Fisher). Following SDS-PAGE and transfer onto nitrocellulose, the membranes were probed using either mAb TB5, HECA-452 or L60 (1:1000 dilution), followed by 1:2500 HRP conjugated secondary Ab (Jackson Immuno, West Grove, PA) and enhanced chemiluminescence detection.

Microfluidic flow chamber based cell adhesion assay:
A custom flow chamber with dimensions of 0.4mm(W)  0.1mm(H)  1cm(L) was fabricated using polydimethylsiloxane (Buffone et al., 2013). This was vacuum-sealed on a tissue culture plastic Petri dish and mounted on the stage of an inverted Zeiss AxioObserver Z1 microscope. The flow chamber substrate was composed of CHO-P cell monolayer expressing P-selectin, IL-1β stimulated HUVEC monolayers expressing E-selectin, or L-selectin-Fc fusion protein that was incubated overnight at 25 μg/mL and subsequently blocked with 1% bovine serum albumin (BSA). 2×10 6 HL-60s/mL suspended in HEPES buffer were perfused over these substrates at 1 dyn/cm 2 .
Movies of the cell interactions were recorded using a pco.edge sCMOS camera (Kelheim, Germany) and data were analyzed as described previously .

DATA AND CODE AVAILABILITY
The data and reagents that support the findings of this study are available from the corresponding authors.