Centre de Biophysique Moléculaire - UPR 4301

The present study investigated the potential of metabolic glycoengineering followed by bioorthogonal click chemistry for introducing into cell-surface glycans different immunomodulating molecules. Mouse tumor models EG7 and MC38-OVA were treated with Ac4Ga1NAz and Ac4ManNAz followed by ligation of immunostimulants to modified cell surface glycans of the living cells through bioorthogonal click chemistry. The presence of covalently bound oligosaccharide and oligonucleotide immunostimulants could be clearly established. The activation of a reporter macrophage cell line was determined. Depending on the tumor cell line, covalently and noncovalently bound CpG activated the macrophages by between 67 and 100% over controls. EG7 cells with covalently attached immunostimulants and control injected subcutaneously into C57BL/6 mice. All tumor cells subjected to the complete treatment with control molecules formed tumors like nontreated cells confirming cell viability. However, when CpG oligonucleotide was linked to cell-surface glycans, tumor growth was slowed significantly (60% reduction, n = 10, by covalently bound CpG compared to noncovalently bound CpG, n = 10). When mice that had not developed large tumors were challenged with unmodified EG7 cells, no new tumors developed, suggesting protection through the immune system.

The solid-phase chemical assembly of a protein through iterative chemoselective ligation of unprotected peptide segments can be followed with chemical and/or enzymatic transformations of the resulting immobilized protein, the latter steps thus benefitting from the advantages provided by the solid support. We demonstrate here the usefulness of this strategy for the chemo-enzymatic synthesis of glycoprotein analogues. A linker was specifically designed for application to the synthesis of O-glycoproteins : this new linker is readily cleaved under mild aqueous conditions compatible with very sensitive glycosidic bonds, but is remarkably stable under a wide range of chemical and biochemical conditions. It was utilized for solid-supported N-to-C peptidomimetic triazole ligation followed by enzymatic glycosylation, ultimately leading to a very large MUC1-derived glycoprotein containing 160 amino acid residues, 24 α-GalNAc moieties linked to Ser and Thr, and 3 triazoles as peptide bond mimetics.

Metabolic chemical reporters (MCRs) of glycosylation are analogues of monosaccharides that contain bioorthogonal functionalities and enable the direct visualization and identification of glycoproteins from living cells. Each MCR was initially thought to report on specific types of glycosylation. We and others have demonstrated that several MCRs are metabolically transformed and enter multiple glycosylation pathways. Therefore, the development of selective MCRs remains a key unmet goal. We demonstrate here that 6-azido-6-deoxy-N-acetyl-glucosamine (6AzGlcNAc) is a specific MCR for O-GlcNAcylated proteins. Biochemical analysis and comparative proteomics with 6AzGlcNAc, N-azidoacetyl-glucosamine (GlcNAz), and N-azidoacetyl-galactosamine (GalNAz) revealed that 6AzGlcNAc exclusively labels intracellular proteins, while GlcNAz and GalNAz are incorporated into a combination of intracellular and extracellular/lumenal glycoproteins. Notably, 6AzGlcNAc cannot be biosynthetically transformed into the corresponding UDP sugar-donor by the canonical salvage-pathway that requires phosphorylation at the 6-hydroxyl. In vitro experiments showed that 6AzGlcNAc can bypass this roadblock through direct phosphorylation of its 1-hydroxyl by the enzyme phosphoacetylglucosamine mutase (AGM1). Taken together, 6AzGlcNAc enables the specific analysis of O-GlcNAcylated proteins, and these results suggest that specific MCRs for other types of glycosylation can be developed. Additionally, our data demonstrate that cells are equipped with a somewhat unappreciated metabolic flexibility with important implications for the biosynthesis of natural and unnatural carbohydrates.

A good correlation between the expression of mucin1 (MUC1) and T antigen was found in breast cancer tumors and breast cancer cell lines, especially after treatment with neuraminidase. The association between the appearance of T antigen and the overexpression of MUC1 was further confirmed by transfecting MDA-MB-231 cells and murine 4T1 mammary carcinoma cells with cDNA for MUC1 and using an RNAi approach to inhibit the expression of MUC1 gene in T47D cells. Furthermore, we discovered that in 4T1 cells which express the sialyl LeX antigen, overexpression of MUC1 caused not only appearance of T antigen, but also loss of the sialyl LeX structure. As the observed changes in O-glycan synthesis can be associated with changes in the expression of specific glycosyltransferases, core 1 β1,3-galactosyltransferase, core 2 β1,6-N-acetylglucosaminyltransferase (C2GnT1) and β-galactoside α2,3-sialyltransferase (ST3Gal I), we studied their expression in parental, vector-transfected and MUC1-transfected MDA-MB-231 and 4T1 cells as well as T47D cells transduced with small hairpin RNA targeted MUC1 mRNA. It was found that the expression of C2GnT1 and ST3Gal I is highly decreased in MUC1-expressing MDA-MB-231 and 4T1 cells and increased in T47D cells with suppressed expression of MUC1. Therefore, we found that changes in the structure of O-linked oligosaccharides, resulting in the occurrence of T antigen, are at least partially associated with MUC1 overexpression which down-regulates the expression of C2GnT1 and ST3Gal I. We showed also that the overexpression of MUC1 in 4T1 cells changes their adhesive properties, as MUC1-expressing cells do not adhere to E-selectin, but bind galectin-3.

Changes in glycosylation are correlated to disease and associated with differentiation processes. Experimental tools are needed to investigate the physiological implications of these changes either by labeling of the modified glycans or by blocking their biosynthesis. N-acetylgalactosamine (GalNAc) is a monosaccharide widely encountered in glycolipids, proteoglycans and glycoproteins ; once taken up by cells it can be converted through a salvage pathway to UDP-GalNAc which is further used by glycosyltransferases to build glycans. In order to find new reporter molecules able to integrate into cellular glycans, synthetic analogs of GalNAc were prepared and tested as substrates of both enzymes acting sequentially in the GalNAc salvage pathway, the galactokinase 2 (GK2) and the uridylpyrophosphorylase AGX1. Detailed in vitro assays identified the GalNAc analogs which can be transformed into sugar nucleotides and revealed several bottlenecks in the pathway : a modification on C6 is not tolerated by GK2 ; AGX1 can use all products of GK2 although with various efficiencies ; all analogs transformed into UDP-GalNAc analogs except those with alterations on C4 are substrates for the polypeptide GalNAc transferase T1. Besides, all analogs which could be incorporated in vitro into O-glycans were also integrated into cellular O-glycans as attested by their detectetion on the cell surface of CHO-ldlD cells. Altogether our results show that GalNAc analogs can help to better define structural requirements of the donor substrates for the enzymes involved in GalNAc metabolism and those that are incorporated into cells will prove valuable for the development of novel diagnostic and therapeutic tools.

GalNAc is the initial sugar of mucin-type O-glycans, and is a component of several tumor antigens. The aim of this work was to determine whether synthetic GalNAc analogs could be taken up from the medium and incorporated into complex cellular O-glycans. The cell line employed was CHO ldlD, which can only use GalNAc and Gal present in the medium for the synthesis of its glycans. All GalNAc analogs with modified N-acyl groups (N-formyl, N-propionyl, N-glycolyl, N-azidoacetyl, N-bromoacetyl, and N-chloroacetyl) were incorporated into cellular O-glycans, although to different extents. The GalNAc analogs linked to Ser or Thr could be extended by the β3-galactosyltransferase glycoprotein-N-acetylgalactosamine 3β-galactosyl transferase 1 in vitro and in vivo and by α6-sialyltransferase α-N-acetylgalactosaminide-α-2,6-sialyltransferase 1. At the surface of CHO ldlD cells, all analogs were incorporated into sialylated O-glycan structures like those present on wild-type CHO cells, indicating that the GalNAc analogs do not change the overall structure of core-1 O-glycans. In addition, this study shows that the unnatural synthetic GalNAc analogs can be incorporated into human tumor cells, and that a tumor antigen modified by an analog can be readily detected by a specific antiserum. GalNAc analogs are therefore potential targets for tumor immunotherapy.

In cancer, mucins are aberrantly O-glycosylated, and consequently, they express tumor-associated antigens such as the Tn determinant (alpha-GalNAc-O-Ser/Thr). As compared with normal tissues, they also exhibit a different pattern of expression. In particular, MUC6, which is normally expressed only in gastric tissues, has been detected in intestinal, pulmonary, colorectal, and breast carcinomas. Recently, we have shown that the MCF7 breast cancer cell line expresses MUC6-Tn glycoproteins in vivo.

The glycopeptide construct shown was synthesized using a convergent strategy based on oxime chemical ligation. It is composed of a universal T-helper and three tumor-related epitopes from the human mucin MUC1, an non-glycosylated repeat unit, and two units glycosylated with the Tn and TF epitopes, respectively. In association with a mild adjuvant suitable for human therapy, this construct elicited a strong specific immune response in mice, directed against natural tumor-associated structures.

UDP-GalNAc has been synthesised with high yield from GalNAc, UTP and ATP using recombinant human GalNAc kinase GK2 and UDP-GalNAc pyrophosphorylase AGX1. Both enzymes have been prepared in one step from I L cultures of transformed Escherichia coli and I he UDP-GalNAc produced has been purified by a simple procedure. The method described is a rapid ;and efficient means to produce UDP-GalNAc as well as analogues like UDP-N-azidoacetylgalactosamine (UDP-GalNAz). (c) 2005 Elsevier Ltd. All rights reserved.

We have identified a homozygous G > A substitution in the donor splice site of intron 6 (IVS6 + 1G > A) of the cytidine monophosphate (CMP)-sialic acid transporter gene of Lec2 cells as the mutation responsible for their asialo phenotype. These cells were used in complementation studies to test the activity of the 2 CMP-sialic acid transporter cDNA alleles of a patient devoid of sialyl-Le(x) expression on polymorphonuclear cells. No complementation was obtained with either of the 2 patient alleles, whereas full restoration of the sialylated phenotype was obtained in the Lec2 cells transfected with the corresponding human wild-type transcript. The inactivation of one patient allele by a double microdeletion inducing a premature stop codon at position 327 and a splice mutation of the other allele inducing a 130-base pair (bp) deletion and a premature stop codon at position 684 are proposed to be the causal defects of this disease.

Background : Following systemic administration, polyplexes must cross the endothelium barrier to deliver genes to the target cells underneath. To design an efficient gene delivery system into lung epithelium, we evaluated capture and transfection efficiencies of DNA complexed with either Jet-PEI (TM) (PEI-polyplexes) or histidylated polylysine (His-polyplexes) in human lung microvascular endothelial cells (HLMEC) and tracheal epithelial cells.

A series of three O-methylated UDP-GalNAc analogues have been synthesised using a divergent strategy from a 3,6-di-O-pivaloyl GlcNAc derivative. The biological activity of these probes toward polypeptide-alpha-GalNAc-transferase T1 has been investigated. This study shows that this glycosyltransferase exhibits a very high substrate specificity. (C) 2003 Elsevier Science Ltd. All rights reserved.

Our goal was to develop mimics of MUC1, highly immunogenic to induce an efficient immune response against the tumor-associated form of MUC1, and sufficiently different from the natural antigen to bypass the tolerance barrier in humans. With the aim of obtaining a well-defined peptide construct as a means of evoking the precise immune responses required in immunotherapy, we synthesized artificial mimics of the MUC1 protein composed of two MUC1 repeat units of inverse orientation and a universal T-helper epitope. To synthesize these heteromeric peptide constructs, we followed a convergent approach using chemoselective ligation based on oxime chemistry. A stem peptide was first synthesized bearing two orthogonally masked aldehydes. After successive depfotection, two oxime bonds can be specifically generated. The proposed strategy proved to be concise and robust, and allowed the synthesis of the tri-branched protein in a very satisfactory yield. The different constructs were tested for their ability to generate antibodies able to recognize the MUC1 protein.

Several oncogenic proteins are known to influence cellular glycosylation. In particular, transfection of codon 12 point mutated H-Ras increases CMP-Neu5Ac : Galbeta1,4GlcNAc alpha2,6-sialyltransferase I (ST6Gal I) activity in rodent fibroblasts. Given that Ras mediates its effects through at least three secondary effector pathways (Raf, RalGEFs and PI3K) and that transcriptional control of mouse ST6Gal I is achieved by the selective use of multiple promoters, we attempted to identify which of these parameters are involved in linking the Ras signal to ST6Gal I gene transcription in mouse fibroblasts. Transformation by human K-Ras or H-Ras (S12 and V12 point mutations, respectively) results in a 10-fold increase in ST6Gal I mRNA, but no alteration in the expression of related sialyltransferases.

UDP-GalNAc:polypeptide alphaN-acetylgalactosaminyltransferases (ppGaNTases) transfer GalNAc from UDP-GalNAc to Ser or Thr. Structural features underlying their enzymatic activity and their specificity are still unidentified. In order to get some insight into the donor substrate recognition, we used a molecular modelling approach on a portion of the catalytic site of the bovine ppGaNTase-T1. Fold recognition methods identified as appropriate templates the bovine alpha1,3galactosyltransferase and the human alpha1,3N-acetylgalactosaminyltransferase. A model of the ppGaNTase-T1 nucleotide-sugar binding site was built into which the UDP-GalNAc and the Mn2+ cation were docked.

Transgenic mice expressing the Simian virus 40 large T antigen under the control of the liver-specific human antithrombin-III promoter all develop well-differentiated hepatocellular carcinoma. During tumour development serum ceruloplasmin (Cp) increases gradually until it reaches 30 times control levels in all transgenic mice at 6 months of age. The accumulation of Cp in the serum is due to the increased transcription of the Cp gene as well as to the increase in Cp mRNA stability in the livers of the transgenic mice. One-half of the overproduced Cp is charged with copper and Cp-associated serum oxidase activity increases in parallel with the holo-Cp concentration. Through its ferroxidase activity Cp is involved prominently in iron metabolism. Analysis of copper and iron in serum and liver revealed increased copper levels in the serum of tumour-bearing animals and which increased in parallel with Cp concentration ; the amounts of copper in the liver were unchanged. In contrast, serum iron remained constant during tumour development whereas the iron concentration in the livers of the transgenic mice decreased.

Transgenic mice expressing the Simian virus 40 large T antigen under the control of the liver-specific human antithrombin-III promoter all develop well-differentiated hepatocellular carcinoma. During tumour development serum ceruloplasmin (Cp) increases gradually until it reaches 30 times control levels in all transgenic mice at 6 months of age. The accumulation of Cp in the serum is due to the increased transcription of the Cp gene as well as to the increase in Cp mRNA stability in the livers of the transgenic mice. One-half of the overproduced Cp is charged with copper and Cp-associated serum oxidase activity increases in parallel with the holo-Cp concentration. Through its ferroxidase activity Cp is involved prominently in iron metabolism. Analysis of copper and iron in serum and liver revealed increased copper levels in the serum of tumour-bearing animals and which increased in parallel with Cp concentration ; the amounts of copper in the liver were unchanged. In contrast, serum iron remained constant during tumour development whereas the iron concentration in the livers of the transgenic mice decreased.

MR60 is an intracellular membrane protein which has been shown to act as a mannoside specific lectin and to be identical to ERGIC-53, a protein characteristic of the endoplasmic reticulum-Golgi apparatus-intermediate compartment, acting as a shuttle. According to its primary sequence, this MR60/ERGIC-53 protein contains a luminal domain including the carbohydrate recognition domain, a stem, a transmembrane segment and a cytosolic domain. The endogenous MR60/ERGIC-53 protein is spontaneously oligomeric, (dimers and hexamers), In this paper, we study the relationship between the oligomerization state and the sugar binding capacity by using recombinant proteins.

Liver cancer is one of the most frequent and lethal malignancies worldwide, Early detection is hampered by the absence of reliable markers, Mice transgenic for the SV40 large T antigen under the control of a liver-specific promoter spontaneously develop well-differentiated hepatocellular carcinomas between 8 to 10 weeks of age.

By use of threading methods, the C-terminal region of uridine diphospho-N-acetyl-D-galactosamine:polypeptide N-acetylgalactosaminyltransferases (ppGalNAc-transferases) was predicted to have the same fold as the lectin-domain of the plant cytotoxins ricin and abrin-a, for which crystal structure are available. The sequence are very low. Nevertheless, the amino acids in the hydrophobic core essential for the structure stability and the cysteine residues are conserved. In addition, the amino-acids involved in carbohydrate binding are conserved in ppGalNAc-transferases. The extra C-terminal domain of these enzymes is therefore a putative glycan-binding domain. A model of the lectinlike domain of human ppGalNAc-transferase T1 was built using knowledge based methods. Geometry optimization of the complex with galactose allowed prediction that this domain could bind this monosaccharide. However, the interaction seems to be rather weak, and at the moment there is no evidence that ppGalNAc-transferases displays a lectin activity in vivo.