Glycosylation-related Gene Expression in Prion Diseases

Several lines of evidence indicate that some glycoconjugates are efficient effectors of the cellular prion protein (PrPC) conversion into its pathogenic (PrPSc) isoform. To assess how glycoconjugate glycan moieties participate in the biogenesis of PrPSc, an exhaustive comparative analysis of the expression of about 200 glycosylation-related genes was performed on prion-infected or not, hypothalamus-derived GT1 cells by hybridization of DNA microarrays, semiquantitative RT-PCR, and biochemical assays. A significant up- (30-fold) and down- (17-fold) regulation of the expression of the ChGn1 and Chst8 genes, respectively, was observed in prion-infected cells. ChGn1 and Chst8 are involved in the initiation of the synthesis of chondroitin sulfate and in the 4-O-sulfation of non-reducing N-acetylgalactosamine residues, respectively. A possible role for a hyposulfated chondroitin in PrPSc accumulation was evidenced at the protein level and by determination of chondroitin and heparan sulfate amounts. Treatment of Sc-GT1 cells with a heparan mimetic (HM2602) induced an important reduction of the amount of PrPSc, associated with a total reversion of the transcription pattern of the N-acetylgalactosamine-4-O-sulfotransferase 8. It suggests a link between the genetic control of 4-O-sulfation and PrPSc accumulation.

. It also contains specific glycosaminoglycan binding sites (6), including three that bind heparan sulfate proteoglycans, which are specific components of the extracellular matrix (7). Glycans directly linked to PrP participate in the strain diversity, in cell to cell transfer of GPI-anchored PrP, allowing the transport of both normal and infectious protein and, as a consequence, the propagation of infection (8,9). Although the putative cofactors involved in the structural trans-conformation of PrP C into the pathogenic form are not yet identified, glycans and/or glycoconjugates are proper candidates. N-glycosylation of PrP, which was shown to interfere with PrP res accumulation, participates in the control of the accessibility of PrP determinants involved in its conversion, in relation to prion strain diversity and resistance (10 -12). The relative membrane mobility allowed by the GPI anchor, leads the prion protein to confine in "rafts" (sphingolipid and cholesterol rich semi-ordered membrane microdomains), a location which is essential for the conformational conversion of PrP C into PrP Sc (13)(14)(15). The ability of PrP to bind heparan sulfate proteoglycans has evidenced the importance of these glycoconjugates for the conversion process. Three regions of PrP, residues 23-52, 53-93, and 110 -128, were identified as being able to bind heparin and heparan sulfate (7). They can also be involved in the PrP binding to the LRP/LR laminin receptor that leads to PrP endocytosis, thus suggesting that proteoglycans can modulate the subcellular trafficking of PrP (16,17). Nevertheless the effect of heparan sulfates seems to be paradoxical as they can stimulate or inhibit PrP Sc formation. Depending on their level of sulfation, a variety of sulfated exogenous glycans such as dextran sulfate, pentosan polysulfate, and heparan mimetic can inhibit PrP Sc accumulation in cell cultures (18 -23). Inversely, heparan sulfate has been shown to be associated with cerebral prion amyloid plaques and with the more diffuse PrP res deposits that appear in early stages of prion diseases (24,25). Moreover, heparinase III sensitive-heparan sulfate proteoglycans, that are probably hyposulfated, can participate in the metabolism of prions and stimulate the cell-free conversion of PrP C into PrP Sc (26 -28). While the function of the complex formed by PrP and sulfated glycans has still to be determined, the molecular interactions between these two partners seem to have a pivotal role in prion diseases especially in the structural conformation of PrP.
In this study, to assess the genetic basis of the intervention of glycoconjugates in prion disorders, we used the derived hypothalamic neuronal GT1 cell line, which is a proper model to simulate PrP Sc accumulation that occurs at the late stages of the disease. Based on the use of a DNA microarray, we examined the expression of genes related to glycosylation and showed that the PrP Sc accumulation depends not only on ex-pression of genes involved in heparan and chondroitin synthesis but probably also on chondroitin sulfation.

EXPERIMENTAL PROCEDURES
Chemicals-Heparan mimetic (HM) was purchased from Professor D. Barritault (University of Paris XII). It was obtained by controlled chemical substitution of T40 dextran with defined amounts of carboxymethyl (CM), sulfate (S), and hydrophobic groups such as benzylamide (Bn) (29). The molecule used was HM2602 that contains 88% CM, 20% Bn, and 50% S substitutions per dextran unit (23).
Cell growth was performed at 37°C in Opti-Modified Eagle's Medium supplemented with 5% fetal calf serum, 5% fetal horse serum, 1% penicillin-streptomycin, and 1% sodium pyruvate, in the presence of 5% CO 2 . Cells were grown to confluence and subcultured every 3 days.
For cellular proliferation kinetics, 300,000 GT1 and ScGT1 cells were inoculated per plate with 15 ml of the above described growing medium. Cell proliferation was monitored every day until they reached the required confluence state. For harvesting the cells, they were washed with 5 ml of PBS, incubated 10 min with 2 ml of PBS-10 mM EDTA, recovered and counted after Trypan Blue staining.
Histochemical Procedures-C57Bl6 mice (8 weeks old; weight 20 g) were obtained from Harlan (Gannat, France). The animals received water and food ad libitum. Mice were sacrificed, the brain was recovered, immersed overnight in Carnoy's fluid fixative, and paraffin-embedded. Sagittal sections (5-m thick) were hydrated and pretreated with H 2 O 2 to block endogenous peroxidases and with 20% normal horse serum to reduce nonspecific staining. PBS containing 0.1% (v/v) Triton X-100 was used in all the steps. The slides were incubated with the primary antiserum diluted 1:50 for 15 min at room temperature. The labeled sites were detected with peroxidase-catalyzed signal amplification system (CSA, rabbit link, Ref. K 1498, Dako®), revealed with Novared (Dako®), and counterstaining with light hemalun. Controls were made by replacing the primary antibody with similarly diluted preimmune serum.
Heparan Mimetic Treatment and Analysis of PrP Sc Accumulation-GT1 and ScGT1 cells were treated for 6 days (two passages) with 10 g of HM2602 per ml of medium. As a control, a set of GT1 and ScGT1 cells were grown without HM for the same time. PrP Sc accumulation analysis was carried out as described previously (32).
RNA Extraction-Total RNAs were isolated using the RNeasy protocol (Qiagen Inc., Hilden, Germany). RNA concentration was measured using the Agilent device (Agilent 2100 Bioanalyzer).
DNA Microarray Analysis-DNA microarrays were prepared by ourselves on polylysine slides (CML, Menzel-glasses). They Labeled cDNA synthesis and microarray hybridizations were performed as described in the MICROMAX TM TSA TM labeling and detection kit (PerkinElmer Life Sciences). Equal amounts of biotin-labeled cDNA (corresponding to 10 g of total RNA from GT1 cells) and fluorescein-labeled cDNA (corresponding to 10 g of total RNA from ScGT1 cells) were hybridized on the slide. Various amounts (1-100 pg) of control RNAs (Arabidopsis RNAs, SpotReport-10 Array Validation System from Stratagene) were added to each batch of RNA samples for normalization. Hybridizations were carried out overnight at 65°C in a hybridization chamber (Corning).
After washings, biotin-labeled cDNA were revealed by streptavidinhorseradish peroxidase (HRP) and Cy5-tyramide. Fluorescein-labeled cDNAs were revealed using an anti-fluorescein-HRP antibody and Cy3tyramide. Cy3 and Cy5 fluorescence signals were measured using a GMS 418 Array Scanner (MWG). The raw data were analyzed using the Array-Pro Analyser software.
Hybridizations were repeated three times with biotin-labeled cDNA generated from RNAs from GT1 cells and fluorescein-labeled cDNA generated from RNAs from ScGT1 cells. Three other hybridizations were made with reversed Cy3 and Cy5 labeling.
Data were reported as the mean fold transcript level increase or decrease in the ScGT1 compared with GT1 cells. The fold change in relative transcript level (RTL) between ScGT1 and GT1 cells must be Ն2 or Յ2 for being taken into account.
Real-time PCR Analysis-Semiquantitative RT-PCR analyses were carried out in triplicate on the ABI Prism™ 5700 sequence detection system (PE Applied Biosystems). Reactions were performed in 25 l that includes 12.5 l of SYBR Green PCR Master Mix (Applied Biosystems). After each of the 40 cycles (made of 15 s of denaturation at 95°C and 1 min of annealing/extension at 60°C), change in SYBR Green dye fluorescence, allowed the determination of a threshold cycle (Ct). cDNAs used as template were obtained by reverse transcription of total RNA using 0.5 g of oligo-d(T) primers and the Superscript II RNase H reverse transcriptase (Invitrogen, Life Technologies). The final volume was of 20 l, the incubation was performed at 42°C for 50 min, and the inactivation at 70°C for 15 min.
Specific primers used are listed in Table I. The relative amount of PCR products between ScGT1 and GT1 cells was determined based on the C T value. Measurement of the level of TFIId transcription factor encoding mRNA was used to normalize the samples (33).
Glycosaminoglycans Quantification-GT1 and ScGT1 cell cultures, at three levels of confluence, were washed in phosphate-buffered saline and lysed in 400 l of K 2 HPO 4 100 mM pH 8.0, Triton X-100 0.5% buffer. The protein content of a 100-l aliquot was quantified (Micro BC assay, Amersham Biosciences), and the remaining 300 l of protein sample were digested at 56°C for 12 h with proteinase K (50 g/ml final). The enzyme was heat-inactivated at 90°C for 10 min, and the mixture was centrifuged (10 000 ϫ g, 10 min and 20°C) through an Ultrafree 0.2-m filter (Microcon).

4-O-GalNAc Hyposulfation in Prion Diseases
an adjusted 100-l aliquot of digested sample, shacked for 30 min, and centrifuged at 12 000 ϫ g for 10 min. After discarding the supernatant, 1 ml of a DMMB decomplexation solution (4 M guanidine hydrochloride in 10% propan-1-ol and acetate tri-hydrate buffer 50 mM pH 6.8) was added to the pellet. The mixture was shacked for 30 min and its absorbance was measured at 656 nm. The sulfated GAG amount was determined by comparison with a chondroitin sulfate solution curve. For chondroitin sulfate quantification, a 100-l aliquot of proteinase K-digested sample was mixed with 100 l of sodium nitrite (0.5%) and acetic acid (33%). The reaction was stopped by addition of 100 l of ammonium sulfamate (12.5%). Remaining chondroitin sulfate was quantified in 100 l of nitrous acid reaction mixture as described above.
For Western blotting, GT1 and ScGT1 cells were washed in PBS, lysed in 300 l of Tris-HCl, 50 mM pH 6.8, glycerol 10% (p/v), SDS 1% (v/v) buffer, and sonicated. The protein amount was determined according to the bicinchonic acid procedure (Sigma). 60 g of total protein were separated by electrophoresis on a 10% polyacrylamide gel. After transfer (200 mA, 1 h 30 min), the nitrocellulose membrane was incubated for 2 h in 10 ml of 1% blocking reagent (Western blocking reagent, Roche Applied Science). The incubation with the anti-GalNAc-4-ST1 antibodies (1:100 dilution) was carried out overnight at 4°C in 0.5% blocking solution. As a control, an anti-TFIId Western blotting was also performed (Anti-TfIId dilution: 1:750, Santa Cruz Biotechnology). Detections were performed using a goat anti-rabbit IgG HRP-coupled antibody (Dako). The chemiluminescent reaction (Roche Applied Science) was revealed by an Amersham Biosciences Hyperprocessor.

Expression Analysis of Glycosylation-related Genes in ScGT1
versus GT1 Cells-ScGT1 cells are a proper model to study PrP Sc accumulation. They were established on the basis of a derived hypothalamic neuronal cell line that expresses high amount of PrP C (eight times more than in the widely used neuroblastoma N2a cell line) and they persistently accumulate high levels of PrP res (31,35). The origin of these cells makes them the proper cellular model for studying the process of brain PrP Sc deposition that occurs in prion disorders (36).
Upon the 165 glycosylation-related genes whose relative transcript levels were estimated by hybridization on a DNA microarray, twelve presented a significant variation in ScGT1 cells as compared with GT1 cells (Table II). These genes belong to the glycosyltransferase (6 genes), glycosidase (3 genes) and carbohydrate sulfotransferase (3 genes (Table II). The CsGalNAcT-I enzyme is directly involved in glycosaminoglycan synthesis, particularly chondroitin sulfate (37,38). The GalNAc-4-ST1 enzyme transfers a SO 4 radical from 3Ј-phosphoadenosine-5Ј-phosphosulfate (PAPS) donor substrate to the hydroxyl at C4 position of a ␤1,4-linked N-acetylgalactosamine containing acceptor substrate. When the enzyme is free from the Golgi membrane or experimentally truncated, the SO 4 radical can be transferred on a ␤1,4-linked GalNAc residue belonging to a chondroitin glycan (39).
Results from the microarray were confirmed by semiquantitative RT-PCR (Fig. 1A). ChGn1 was found to be 30.2-fold overexpressed and Chst8 17.13-fold underexpressed in ScGT1 cells. Compared with the microarray data, the semiquantitative RT-PCR analysis shows for both genes even higher relative transcript level differences, indicating that DNA microarray technology underestimates changes in mRNA amounts, particularly for the most highly expressed genes.
To correlate these transcripts levels with protein amounts, specific antibodies toward both polypeptides were designed. An important and reproducible decrease in the amount of GalNAc-4-ST1 was found in ScGT1 cells as compared with GT1 cells (Fig. 1B).
Immunochemistry of N-acetylgalactosamine 4-O sulfotransferase 8 in Mouse Brain-To assess enzyme localization in mouse brain, search for GalNAc-4-ST1 was performed (Fig. 2) using the purified antibody directed against the N-terminal epitope of the enzyme located in its stem region (see "Experimental Procedures"). Both occipital cortex and lateral hypothalamus brain regions stained positively. In few cortical neurons, the presence of the enzyme was evidenced in pericellular space and cytoplasmic membrane ( Fig. 2A). GalNAc-4-ST1 was also detected as vesicular string in some neurites from other cells (Fig. 2B). A higher number of neurons from lateral hypothalamus showed a positive staining inside their cytoplasm and dendritic cell extensions (Fig. 2C). Altogether, this staining was in good agreement with the presence in neuronal cells of a significant non-Golgian enzyme fraction.
Measurement of Glycosaminoglycan and Chondroitin Sulfate Amounts-The amount of chondroitin sulfate, which depends on the balance between CsGalNAcT-I and chondroitinase activities, was quantified both in GT1 and ScGT1 cells. To this aim, total GAG and chondroitin sulfate amounts were measured at different levels of cell confluence (low confluence, proliferation phase, high confluence). The amount of total GAG did

4-O-GalNAc Hyposulfation in Prion Diseases
not significantly change according to the confluence stage (Fig.  3A). Interestingly, a clear difference in the chondroitin sulfate amount was evidenced between the two cell lines at low level of confluence (Fig. 3B). In GT1 cells, chondroitin sulfate amount progressively increased with cell growth (12.2 Ϯ 5.3 ng/g of protein at low confluence level corresponding to 37.6% of total GAG, 27.1 Ϯ 2.0 ng/g of protein at high confluence corresponding to 61% of total GAG). Conversely, high chondroitin sulfate content was measured in ScGT1 cells at low confluence (26.6 Ϯ 1.2 ng/g of protein, corresponding to 74% of total GAG) a content that remains identical at high confluence (27.7 Ϯ 2.0 ng/g of protein, corresponding to 67.7% of total GAG). The high chondroitin sulfate amount observed at low level of confluence correlates well with the higher ScGT1 cells growth rate compared with that of the GT1 cells (data not shown) and with the high level of ChGn transcript. Moreover, a lower amount of heparan sulfate compared with that of chondroitin sulfate was detected in GT1 and ScGT1 cells.

Relative Transcript Levels of Glycosylation-related Genes Involved in Heparan and Chondroitin Sulfate Synthesis-Both
CsGalNAcT-I transcript and chondroitin sulfate amounts evidenced a deep impairment of heparan and chondroitin sulfate synthesis pathways. To get more information about this phenomenon, the level of expression of all the known glycosylationrelated genes involved in this glycosaminoglycan synthesis was investigated. Common to the two pathways, tetrasaccharide linker is first synthesized by the enzymes encoded by Xylt1, Xylt2, b4galt7, b3galt6, b3gat3 genes (Fig. 4). Next, the enzymes encoded by Extl1/Extl3 and ChGn1 genes direct the heparan sulfate and chondroitin sulfate synthesis, respectively. At last, heparan and chondroitin sulfate chain elongation depends on the enzymes encoded by Ext1, Ext2, Extl1, Extl2, Extl3, or CsGalNacT2, CsGlcAT, ChPF, CSS genes, respectively (Fig. 4).
Most of the genes driving the heparan sulfate pathway (Ext1, Ext2, Extl2, Extl3) proved to be underexpressed in ScGT1 (Table III), which correlates well with the lower amount of heparan versus chondroitin sulfate previously observed (Fig. 3). Except for ChGn1 (RTL: ϩ30.2 Ϯ 7.02), the gene that initiates chondroitin synthesis on the tetrasaccharide linker, expression of all the other genes (CsGalNAcT2, CsGlcAT, ChPF) involved in the chondroitin pathway was down regulated or unchanged in ScGT1 cells (Table III).
Heparan Mimetic Both Reverses Chst8 Relative Transcript Levels and Induces Clearance of PrP Sc -In order to confirm the possible connection between the conversion of PrP C into PrP Sc and a Chst8-related GAG hyposulfation, heparan mimetic (HM) action on the regulation of this gene was assessed. Indeed, HM have been previously shown to inhibit efficiently PrP Sc accumulation in ScGT1 cells (22,23). The effect of such molecules that mimic endogen sulfated heparan and chondroitin structures on ChGn1 and Chst8 transcription levels was investigated. The efficient inhibition of PrP Sc accumulation after one treatment by HM at 10 g/ml was confirmed (Fig.  5A). Interestingly, no reversion of the ChGn1 relative transcription level was observed after two HM treatments (data not shown), a progressive down-regulation of the expression of Chst8 was detected, phenomenon which was complete after two HM treatments (Fig. 5B). No effect of HM treatment was evidenced in GT1 cells (Fig. 5B).

DISCUSSION
Glycosaminoglycans, specifically heparan sulfate, have been described as being important actors of prion diseases (6,7,16,28,40). Their function is still under investigation and the role of chondroitin sulfate in these disorders has to be more deeply examined. Heparan sulfates have controversial effects as they can stimulate or inhibit these disorders. By binding on specific PrP C sites, they are important partners in the interaction between PrP and the laminin receptor LRP/LR (7,16). Heparan sulfates have also been identified as critical factors in Alzheimer amyloidogenesis and they were found to be associated with PrP Sc insoluble aggregates and tissue amyloid prion deposits in the brain of infected animals (24,25,41,42). Furthermore, heparinase III-sensitive, possibly hyposulfated, heparans seem to be involved in PrP res metabolism (26). Pentosan polysulfate and usual heparan sulfate have a stimulating effect on the conversion of PrP C into PrP res . Inversely, other glycosaminoglycans, such as chondroitin and keratan sulfate, have no effect on this process (27,28,43,44). Interestingly, in scrapieinfected cell culture and in animal models, exogenous heparan and pentosan polysulfate, contrary to their stimulatory effect, are strong inhibitors of PrP Sc accumulation, probably through steric competition, while other glycosaminoglycans were neutral (21,(45)(46)(47)(48).
Correlation between PrP res and Glycosylation-related Genes-Using our DNA microarray technology, the correlation between the pathologic prion protein accumulation and the expression of numerous glycosylation-related genes was investigated in GT1 cells, cells that derived from the central nervous system. It confirmed the relevance of such a methodology to screen potential changes in glycosylation and demonstrated a profound modification of the expression of some glycosylation-related genes. Modification of the RNA transcript levels of genes involved in glycosaminoglycans synthesis, such as heparan and chondroitin sulfates, or in the 4-Osulfation of a non-reducing N-acetylgalactosamine residue was evidenced. Most of the genes involved in heparan sulfate synthesis were down-regulated (Table II), whereas GAG and chondroitin sulfate levels varied accordingly (Fig. 3). Only the ChGn1 gene was overexpressed. It encodes the chondroitin sulfate N-acetylgalactosaminyltransferase-I (CsGalNAcT-I), an enzyme involved in the initiation of chondroitin sulfate synthesis (37). Other genes encoding enzymes involved in chondroitin sulfate polymerization (CsGal-NAcT-2, CSS, CHPF and CsGlcAT) were found to be stably or underexpressed.
Chondroitin Sulfate Proteoglycans and Prion Diseases-The relative increase in the amount of chondroitin sulfate in ScGT1 cells (Fig. 3) suggested a balanced regulation of heparan and chondroitin sulfate synthesis (49) and also a chondroitin sulfate growth-promoting effect in these cells. The changes in the amounts of heparan and chondroitin sulfate in association with PrP res accumulation well correlate with several observations of such modulations after injury of the cerebral nervous system, in Alzheimer's and prion diseases (50 -52). Chondroitin sulfate proteoglycans added to cultures of rat hippocampal or cortical neurons have been described to rescue the cells from excitotoxic damage and to attenuate ␤-amyloid-induced neurodegeneration (53,54). An up-regulation of chondroitin sulfate proteoglycans after an injury of central nervous system, specifically around the region of the lesion, generally leads to an inhibitory effect toward axonal and neurite outgrowth (55)(56)(57). However, a correlation between a large amount of chondroitin sulfate proteoglycan in perineuronal nets surrounding neurons (extracellular matrix materials deposited around synaptic endings and in the space between neurons) and the protection toward the formation of Alzheimer's disease has been described (58 -60).
Importance of N-Acetylgalactosamine-4-O-sulfation-As opposed to the ChGn1 gene overexpression, a deep Chst8 underregulation was evidenced and correlated with a strong decrease of the level of GalNAc-4-ST1 (Fig. 1). The GalNAc-4-ST1 enzyme is a member of the HNK-1 sulfotransferase family that includes chondroitin-4-sulfotransferases 1, 2, and 3, dermatan-4-sulfotransferase-1, HNK-1 sulfotransferase (HNK-1 ST) and GalNAc-4-ST2 (61)(62)(63)(64)(65)(66)(67). It is a transmembrane type II enzyme localized in the Golgi apparatus, which was first described in the pituitary gland (39). It was shown to be the specific sulfotransferase that catalyzes the transfer of a sulfate group on the 4th carbon of a nonreducing terminal GalNAc of the glycoprotein luteinizing hormone (39,67). Sulfated GalNac is essential for luteinizing hormone (LH) neuroendocrine regulating circulatory half-life (68,69). GalNAc-4-SO 4 is of particular importance in endocrine regulation, since several studies described a role for PrP in these physiological functions. Indeed, an alteration of the regulation of the neuroendocrine pancreas has been observed after 139H scrapie strain infection (70). Moreover, a modification of the circadian rhythm in PrP knock-out mice, in Fatal Familial Insomnia patients and in scrapie-infected mice 2 supports this observation (71) (for review, see Ref. 72). As these neuroendocrine changes originate in the hypothalamus, from which the GT1 cell model derived, we can hypothesize that PrP C /PrP Sc endocrine changes is related to a specific cerebral localized hyposulfation of glycoprotein hormones.
An interesting feature of GalNAc-4-ST1 enzyme resides in the existence of a non-Golgian protein isoform, that accumulates in the cell culture medium (39). Our data on the enzyme immunolocalization confirm the existence of such a non-Golgian protein. Indeed, it was present both in the pericellular space and in the cytoplasmic membrane of neurites and dendritic extension and in the whole cytoplasm (Fig. 2). In contrast to the Golgi-linked enzyme, this non-Golgian isoform is able to transfer a SO 4 group on nonreducing GalNAc moieties of chondroitin and dermatan, suggesting a role for this enzyme in the regulation of chondroitin sulfation (39,67). The presence of GalNAc-4-ST1 in both fetal brain and in various regions of the central nervous system have suggested the presence on some glycoproteins of N-linked structures ending with ␤-1,4-linked GalNAc-4-SO 4 (39). Sulfated structures have been described on the carbonic anhydrase VI, the proopiomelanocortin, the Tamm-Horsfall glycoprotein, the urokinase, the sialoadhesin, and the tenascin R (TN-R) (73)(74)(75)(76)(77)(78)(79). The post-translational modifications of TN-R by three distinct sulfated oligosaccharides, contribute to diverse specific functions for TN-R. For example, the addition of O-linked chondroitin sulfate to TN-R contributes to the inhibition of cell adhesion and neurite outgrowth in vitro (80 -85). The GalNAc-4-SO 4 modified TN-R has been described to be predominantly synthesized by neurons and to be selectively localized in various brain regions, including the cortex, the hippocampus, the cerebellum and especially in perineuronal nets (79). Therefore, the expression of GalNAc-4-ST1 in a number of specific tissues and brain regions supports the idea that the tightly regulated synthesis of structures ending with ␤-1,4-linked GalNAc-4-SO 4 in a number of settings, are used in processes requiring specific recognition.   Exostosin-like 1 -0.6 Ϯ 1.0 Extl2 Exostosin-like 2 -5.6 Ϯ 1.0 Extl3 Exostosin-like 3 -9.0 Ϯ 4.0 ChGn1 Chondroitin ␤-1,4 N-acetylgalactosaminyltransferase1 ϩ 30.2 Ϯ 7.0 CsGalNacT2 Chondroitin sulfate N-acetylgalactosaminyltransferase-2 5.2 Ϯ 0.7 CsGlcAT Chondroitin sulfate ␤-1,3-glucuronyltransferase -3.8 Ϯ 0.3 ChPF Chondroitin polymerizing factor 0.6 Ϯ 0.2

4-O-GalNAc Hyposulfation in Prion Diseases
Possible Mechanisms for the Involvement of 4-O-Hyposulfated Chondroitin in the Conversion of PrP C into PrP Sc -The formation of PrP Sc seems to involve a direct molecular interaction between PrP C and an unknown PrP isoform acting as template (86). GAG could also participate in this interaction by helping to bring PrP C and template PrP close enough for interaction to occur (27). Interestingly, efficient sulfated GAG binding sites have been identified in the N-terminal region of PrP C . More generally, it was shown that polymerized GAGs increase PrP Sc , whereas small analogues such as heparan mimetic decrease it (26). Exogenous large chondroitin sulfates, but not small fragments, did not reduce the PrP Sc isoform accumulation in ScN2a cells (26). Ben-Zaken et al. (26) suggest that both heparan and chondroitin sulfate can serve as pro-prions cofactors with equal efficiency, but because chondroitin sulfate are poorly present in ScN2a cells, their enzymatic removal by chondroitinase ABC is undetectable in terms of PrP Sc production. Inversely, the chondroitin amount is higher than that of heparan sulfate in GT1 cells (Fig. 3) and consequently, in these cells, the efficacy of the large chondroitin sulfate GAG to bind PrP isoform could be at least similar to that of heparan sulfate GAG.
The finding that a high ChGn1 transcript level was found in parallel with a down-regulation of the Chst8 gene expression, underlines the possible connection between chondroitin synthesis and its sulfation level. Indeed, Kitagawa et al. (87) evidenced that specific sulfation, particularly 4-O-sulfation, can serve as a stop signal that precludes the chondroitin chain elongation. In this way, our data showing a possible ␤-1,4linked GalNAc-4-SO 4 hyposulfation, suggest the activation of an endogenous enzyme leading to the synthesis of large hyposulfated chondroitin. Hyposulfated GAG, mainly present in ScGT1 cells, would be highly effective in promoting PrP Sc accumulation by helping to bring PrP C and template PrP close enough for interaction to occur.

Why Does Heparan Mimetic Reverse Simultaneously Both PrP Sc Accumulation and Chst8 Transcript Levels in ScGT1
Cells-One possibility is that endogenous hyposulfated chondroitin derivatives could be components of transcription machinery that controls expression of the genes involved in chondroitin synthesis pathway. In addition to their high efficacy for PrP Sc clearance, free or PrP-bound heparan mimetic would thus compete with hyposulfated chondroitin derivatives, as regulators of the Chst8 gene transcription.