Neuronal Ndst1 depletion accelerates prion protein clearance and slows neurodegeneration in prion infection

Select prion diseases are characterized by widespread cerebral plaque-like deposits of amyloid fibrils enriched in heparan sulfate (HS), a abundant extracellular matrix component. HS facilitates fibril formation in vitro, yet how HS impacts fibrillar plaque growth within the brain is unclear. Here we found that prion-bound HS chains are highly sulfated, and that the sulfation is essential for accelerating prion conversion in vitro. Using conditional knockout mice to deplete the HS sulfation enzyme, Ndst1 (N-deacetylase / N-sulfotransferase) from neurons or astrocytes, we investigated how reducing HS sulfation impacts survival and prion aggregate distribution during a prion infection. Neuronal Ndst1-depleted mice survived longer and showed fewer and smaller parenchymal plaques, shorter fibrils, and increased vascular amyloid, consistent with enhanced aggregate transit toward perivascular drainage channels. The prolonged survival was strain-dependent, affecting mice infected with extracellular, plaque-forming, but not membrane bound, prions. Live PET imaging revealed rapid clearance of recombinant prion protein monomers into the CSF of neuronal Ndst1- deficient mice, neuronal, further suggesting that HS sulfate groups hinder transit of extracellular prion protein monomers. Our results directly show how a host cofactor slows the spread of prion protein through the extracellular space and identify an enzyme to target to facilitate aggregate clearance.

distribution during a prion infection. Neuronal Ndst1-depleted mice survived longer and 48 showed fewer and smaller parenchymal plaques, shorter fibrils, and increased vascular 49 amyloid, consistent with enhanced aggregate transit toward perivascular drainage channels. 50 The prolonged survival was strain-dependent, affecting mice infected with extracellular, 51 plaque-forming, but not membrane bound, prion strains. Live PET imaging revealed rapid 52 clearance of prion protein monomers into the CSF in mice expressing unsulfated HS, further 53 suggesting that HS sulfate groups hinder transit of extracellular prion monomers. Our results 54 directly show how a host cofactor slows the spread of prion protein through the extracellular 55 space and identify an enzyme target to facilitate aggregate clearance. 56

Author summary 57
Prions cause a rapidly progressive neurologic disease and death with no curative treatment 58 available. Prion aggregates accumulate exponentially in the brain in affected individuals 59 triggering neuronal loss and neuroinflammation. Yet the additional molecules that facilitate 60 aggregation are largely unknown, and their identification may lead to new therapeutic targets. 61 We have found that prions in the brain preferentially bind to a highly sulfated endogenous 62 polysaccharide, known as heparan sulfate (HS). Here we use genetically modified mice that 63 express poorly sulfated neuron-derived HS, and infect mice with different prions strains. We 64 find that the mice infected with a plaque-forming prion strain show a prolonged survival and 65 fewer plaques compared to the controls. We also found that the prion protein was efficiently 66

Introduction 71
The spread of aberrant protein aggregates through the brain is a key pathogenic mechanism 72 in Alzheimer's, Parkinson's, and prion disease 1-6 . Potential mechanisms for aggregate spread 73 include bulk transport within the extracellular space (ECS) [7][8][9][10][11] . The ECS harbors a dynamic 74 reservoir of interstitial fluid (ISF) that flows toward perivascular clearance pathways 11 . 75 Evidence suggests that bulk ISF flow is enhanced during sleep, fostering the efflux of 76 metabolic waste and proteins, such as amyloid-β (Aβ), from the brain [12][13][14] . Thus, alterations in 77 the structure or composition of perivascular clearance pathways 15 or the extracellular matrix 78 (ECM) 16 during aging or neurodegenerative disease could hinder the clearance of peptides 79 and oligomers 17 . 80 Prion diseases are rapidly progressive neurodegenerative disorders in which prion protein 81 aggregates (PrP Sc ) deposit on cell membranes or as plaques embedded in the brain ECM, 82 depending on the prion conformation [18][19][20][21]  glucosamine residues, also depicted as disaccharides) attached to one or more protein 86 cores 26,27 . A large body of evidence implicates HS in fibril formation and in the endocytosis of 87 protein aggregates in vitro [28][29][30][31][32] . Moreover, polyanions administered intraventricularly increase 88 survival time in experimental rodent models and in patients, potentially by blocking prion 89 binding to endogenous HS [33][34][35][36][37][38][39][40][41][42][43][44] . Transgenic expression of mammalian heparanase, an enzyme 90 that degrades HS, delays prion onset and progression 45 . 91 HS chains bind and concentrate proteins, such as growth factors and cytokines, and the level 92 and pattern of sulfation determine the binding affinity 26,27 . Here, we show that brain-derived 93 prions selectively bind highly sulfated HS chains. To understand how the sulfation level 94 modifies prion aggregate spread and disease progression, we use conditional Ndst1 knock-95 out mice to reduce the sulfation of either neuron or astrocyte generated HS, and challenge 96 mice with diverse prion strains. We found that mice with reduced neuronal HS sulfation 97 suggesting no change in the PrP Sc conformational properties (S6A Fig). Thus, reducing HS 188 sulfation prolongs survival and reduces parenchymal plaques and plaque size in the brain, 189 without detectably altering PrP Sc secondary or tertiary conformation. 190 To further test PrP Sc fibril conformation using biochemical analyses, we measured the 191 biochemical properties of PrP Sc , including the electrophoretic mobility of the proteinase K (PK)-192 resistant core, glycoprofile, stability in chaotropes, and aggregate solubility (S6B- D Fig and  193   Fig 3A-B). Notably the only difference was in PrP Sc solubility, in which there was an increase 194 (25%) in the SynCre + brains (Fig 3A-B) suggesting that the aggregate population was 195 generally smaller. To further assess aggregate size, we next purified prion fibrils from brain 196 homogenate and measured the length of isolated fibrils on electron microscopy grids. We 197 found that fibrils were approximately 20% shorter (on average) in SynCre+ brains (Fig 3C-D), 198 consistent with a change in aggregate size as suggested by the higher PrP Sc solubility. 199 The increased PrP Sc clearance into the cerebrospinal fluid (CSF) may lead to enhanced 200 deposition in the spinal cord. Thus, we next compared the PrP Sc seeding activity in the spinal 201 cord from SynCre-and SynCre+ using real-time quaking-induced conversion (RT-QuIC) (S6 202   Table). However, there were no differences in the proportion of mice with seeding activity in 203 spinal cord (SynCre-: 7 of 11 mice, SynCre+: 14 of 16 mice; Fisher's exact test; p = 0. 19), 204 suggesting that reducing Nhs sulfation does not lead to increased incidence of PrP Sc 205 deposition in the spinal cord. 206

Decreasing neuronal HS sulfation enhances PrP C clearance into the spinal cord 207
Given that i) HS is a significant component of the brain ECM and ii) decreasing nHS sulfation 208 reduces parenchymal plaque load and prolongs survival in prion-infected mice (Fig 2B-H  In summary, our study demonstrates a previously unrecognized role for highly sulfated HS 245 chains advancing prion disease progression. We found evidence that HS and PrP bind by 246 electrostatic interaction, with PrP Sc and sulfated HS engaging within the brain parenchyma, 247 enhancing fibril elongation, parenchymal plaque formation, and markedly accelerating 248 disease. Interestingly, this effect was cell source and prion strain dependent, as reducing 249 neuronal, but not astrocytic HS, prolonged survival, and only impacted strains with a GPI-250 anchorless component. Finally, we systematically identified the more highly sulfated HS 251 molecules concentrated within prion aggregates from mice and human brain by LC/MS, further 252 supporting NDST1 as a therapeutic target. 253

Discussion 254
PrP Sc and Aβ plaques are enriched in HS 23,25,46,60 , yet whether HS facilitates fibril formation 255 in vivo has been unclear. Here we establish that neuron-generated HS accelerates prion 256 propagation in a strain-dependent manner. We show the importance and specificity of HS 257 sulfate groups in binding PrP, as the HS molecules enriched within prion aggregates were 258 more highly 6-O-, 2-O-, or N-sulfated compared to brain lysate, and sulfation was key to 259 amplifying prion conversion. Notably, depleting nHS sulfation reduced parenchymal plaque Previous studies report heparin and HS promote fibril formation in vitro 50,61 , and prion-infected 267 mice and vCJD patients treated with HS mimetics survive longer [33][34][35][39][40][41][42][43][44] . Using genetic 268 models, we and others recently found that shortening HS chains reduces plaque numbers as 269 well as extends survival and improves behavior in prion-infected and Alzheimer's mouse 270 models, respectively 46, 62 , implicating HS in both prion and Aβ plaque formation in vivo. Our 271 studies now demonstrate the highly sulfated nature of HS bound to prions using LC/MS, and 272 the feasibility of manipulating HS biosynthetic pathways to enhance PrP clearance, increasing 273 the resistance of mice to extracellular plaque formation. Notably, the few parenchymal plaques 274 present in mice expressing less sulfated nHS were small and the isolated fibrils shorter and 275 aggregates more soluble, together providing strong evidence to support that sulfated HS 276 recruits prion subfibrillar aggregates, facilitating fibril elongation and plaque formation in the 277 brain parenchyma. Whether other reported co-factors, such as lipids or nucleic acids [63][64][65] , also 278 promote PrP conversion in vivo is unclear and would be important to address in future studies. 279 Nevertheless, manipulating HS polymerase or sulfotransferase expression to reduce HS 280 length or sulfation, respectively, may be a viable therapeutic strategy to promote the clearance 281 of prions, Aβ, and other extracellular aggregates into the CSF. 282 Reducing nHS sulfation did not affect the disease progression of Ndst1 f/f SynCre +/mice 283 infected with GPI-anchored prions (RML), despite PrP C and PrP Sc being embedded within an 284 extensive meshwork of cell surface HSPGs. In vitro, HS reportedly promotes the endocytosis 285 of PrP Sc , as well as tau and α-synuclein 28-31, 66, 67 , yet reducing nHS sulfation did not detectably 286 alter RML prion propagation, astrocyte reactivity, or neurotoxicity. Although we found that 287 heparin binds with high affinity and promotes GPI-anchored prion conversion in vitro, minimal 288 HS was bound to GPI-anchored prions isolated from brain 46 , suggesting that the PrP Sc 289 membrane location limits access to HS, and HS may not enhance neurotoxicity in vivo. Given 290 that GPI-anchored prions concentrate within lipid rafts, the membrane curvature or phosphate 291 head groups, lipid raft size, or HS location and spatial orientation may hinder HS-PrP binding 292 and constrain fibril elongation for membrane-bound prions, abrogating any major scaffolding 293 effect by HS 68 . In contrast, extracellular GPI-anchorless PrP C and PrP Sc are unconstrained in 294 the parenchyma and may more readily bind extracellular HS chains, consistent with our 295 previous finding of abundant HS bound to extracellular prions 46 . That said, we cannot exclude 296 that HS promotes the endocytosis, neuronal toxicity, or propagation of different prion 297 conformers than used here, as prions propagating in different brain regions may access HS 298 chains with a sulfation code better suited for binding. 299 Notably, only reducing nHS sulfation impacted disease progression; reducing astrocytic HS 300 sulfation had no effect. Although both neurons and astrocytes produce and secrete HS 57,58,69 , 301 PrP may selectively bind nHS due to the level and pattern of sulfation, or astrocytes may 302 simply synthesize less HS. Alternatively, considering that only neuronal PrP C is essential for 303 PrP-linked toxicity 70 , it is conceivable that membrane-bound HSPGs such as syndecans or 304 glypicans, facilitate extracellular PrP Sc (GPI-anchorless) binding to PrP C or neuronal surface 305 receptor complexes in trans, potentiating neurotoxic signaling pathways. In this case, reducing 306 nHS sulfation may decrease PrP Sc interactions with cell surface receptors. Future studies may 307 further elucidate the mechanism underlying the prolonged survival. 308 Our studies support a broader role of HS and the ECM in retaining proteins and promoting 309 aggregation in the aging brain, with implications beyond prion disease. Given that HS also 310 binds lipids and chemokines, this work suggests that increases in HS levels or sulfation may 311 slow protein efflux and increase neuroinflammation, particularly in the presence of protein 312 aggregates. Similar to PrP, the N-terminus of Aβ also binds to HS 71 and HS sulfate groups are 313 required for Aβ binding 72 . Given the similarities of our prion disease model to AD, future studies 314 to (i) characterize how HS levels and composition change with age and disease in the human 315 brain, (ii) determine whether HS enhances neuroinflammation, (iii) define the structure of HS 316 concentrated within Aβ plaques, and (iv) devise strategies to enhance the clearance of Aβ in 317 AD models by reducing HS sulfation would be of high priority. 318 We demonstrate a significant increase in the lifespan of prion-infected, nHS-depleted mice, 319 which suggests that HS biosynthetic enzymes, such as Ndst1, may be therapeutic targets for 320 select neurodegenerative diseases. Reducing HS sulfation in early disease would be expected 321 to enhance aggregate transport toward perivascular clearance pathways. However, a caveat 322 is a possible increase in meningeal vascular amyloid, as observed in prion-infected mice with 323 reduced HS sulfation as shown here or with shortened HS chains 46 . Given that these results 324 highlight the importance of HS sulfation in plaque formation, future work may define and inhibit 325 PrP Sc and Aβ binding to vessel-associated molecules to further promote prion clearance into 326 the CSF. 327 In conclusion, our data provide evidence that PrP selectively binds highly sulfated neuronal 328 HS, thereby identifying HS as the first integral in vivo co-factor in prion fibril formation and 329 plaque assembly. Reducing HS sulfation decreased the parenchymal plaque burden and 330 prolonged survival, suggesting that in vivo, PrP and HS electrostatic interactions are critical 331 for binding. We also directly demonstrate how sulfated HS slows PrP C clearance from the ISF 332 in real time, indicating that GPI-anchorless PrP C transits through the brain by bulk flow, and 333 that clearance may be increased by manipulating HS chemical properties. Importantly, our 334 data strongly supports the pursuit of therapeutic strategies targeting HS biosynthetic enzymes 335 to facilitate protein aggregate clearance. 336

Prion transmission studies in mice 338
Ndst1 f/f mice 73 were bred to mice that express the Cre-recombinase under the neuron and astrocyte 339 specific promoters, synapsin1 and glial fibrillary acidic protein (SynCre and GFAPCre), respectively, 340 and to tga20 mice, which overexpress mouse PrP C56 . Homozygosity for tga20 was determined by 341 anti-rabbit or anti-mouse; OmniMap system; Ventana) was incubated on the sections for 12 minutes 376 at 37 °C. The primary antibody was visualized using DAB as a chromogen followed by hematoxylin 377 as a counterstain. Slides were rinsed, dehydrated through alcohol and xylene and cover slipped. The PrP and HS dual immunolabelling was performed as previously described but with minor 390 changes 46 . Briefly, tissue sections were deparaffinized, and epitope exposure was performed using 391 formic acid, PK, and heated citrate buffer (pH 6). Sections were blocked and incubated with anti-392 profile for the individual animal in a specific brain area and was depicted in the 'radar plots'. Two 408 investigators blinded to animal identification performed the histological analyses. 409

Quantitative analysis of astrocytic and microglial inflammation 410
To measure astrocytic gliosis and microglial activation in Ndst1 f/f SynCre, Ndst1 f/f tga20 +/+ SynCre, 411 Ndst1 f/f GFAPCre and Ndst1 f/f tga20 +/+ GFAPCre mice, slides containing cerebral cortex, corpus 412 callosum, hippocampus, thalamus, hypothalamus, and cerebellum were imaged using the Olympus 413 EX41 microscope with DP Controller. Images were converted to grayscale, and FIJI (an ImageJ 414 based image processing software) was used to measure the total brain area and quantify astrocyte 415 and microglia reactivity using the "Measure" function. Astrocyte and microglia were demarcated 416 using the "Find the edges" function and particle analysis was used to measure the area occupied. 417 The total area covered by astrocyte and microglia was divided by the total area for each brain 418 region. 419

Western blot and glycoprofile analyses 420
For ME7 strain in WT mice, PrP Sc was concentrated from 10% brain homogenate in phosphate 421 buffered saline (PBS) (w/v) by performing sodium phosphotungstic acid (NaPTA) precipitation prior 422 to western blotting 74 . Briefly, 20 μl of 10% brain homogenate in an equal volume of 4% sarkosyl in 423 PBS was nuclease digested (benzonase TM , Sigma) followed by digestion with 20 μg/ml PK at 37 °C  Scientific) and visualized on a Fuji LAS 4000 imager. Quantification of PrP Sc glycoforms was 433 performed using Multigauge V3 software (Fujifilm). For mCWD, 100 μl of 10% brain homogenate 434 was concentrated using NaPTA as described above and digested with 100 μg/ml PK at 37 °C for 435 45 minutes. 436

Conformation stability assay 437
To measure prion strain stability in guanidine chloride (GdnHCl), 10% brain homogenates were 438 RML, ME7, and mCWD prions were used to seed mouse PrP C (no heparin reactions). The PrP C 467 was newly prepared for each independent experiment. The prion seeds were derived from brain 468 homogenate that was pooled from mice inoculated with the same prion strain. The brain 469 22 homogenate samples pooled to generate the seeds were consistent between the experiments. 470 Prion-infected brain homogenate (10% w/v) was added to PrP C -expressing RK13 cell lysate (1:10, 471 PrP Sc : PrP C by volume) and subjected to repeated 5 seconds sonication pulses (S4000, QSonica) 472 with 10 minutes of incubation between each pulse, over a total period of 24 hours. Sonication power 473 was maintained at 50-60% and samples were continuously rotated in a water bath at 37 °C. 474 Samples were then digested with 200 µg/ml PK for 30 minutes at 37 °C and analyzed by western 475 blot using the anti-PrP monoclonal antibody 3F4 76

Purification of PrP Sc for mass spectrometry and electron microscopy 508
To analyze HS bound to PrP Sc , PrP Sc was first purified from mouse brains as previously 509 described 46,77 . Briefly, one ml of 10% brain homogenate was mixed with an equal volume of TEN(D) 510 buffer (5% sarkosyl in 50 mM Tris-HCl, 5 mM EDTA, 665 mM NaCl, 0.2 mM dithiothreitol, pH 8.0), 511 containing complete TM protease inhibitors (Roche). Samples were incubated on ice for 1 hour and 512 centrifuged at 18,000 x g for 30 minutes at 4 °C. All but 100 μl of supernatant was removed, and the 513 pellet was resuspended in 100 μl of residual supernatant and diluted to 1 ml with 10% sarkosyl 514 TEN(D). Each supernatant and pellet was incubated for 30 minutes on ice and then centrifuged at 515 18,000 x g for 30 minutes at 4 °C. Supernatants were recovered while pellets were held on ice. 516 Supernatants were added separately into ultracentrifuge tubes with 10% sarcosyl TEN(D) buffer 517 containing protease inhibitors and centrifuged at 150,000 x g for 2.5 hours at 4 °C. Supernatants 518 were discarded while pellets were rinsed with 100 μl of 0.25 M NaCl in TEN(D) buffer with 1% 519 sulfobetaine (SB 3-14) and protease inhibitors and then combined and centrifuged at 200,000 x g 520 for 2 hours at 20 °C. The supernatant was discarded, and pellet was washed and then resuspended 521 in 200 μl of ice cold TMS buffer (10 mM Tris-HCl, 5 mM MgCl2, 100 mM NaCl, pH 7.0) with protease 522 inhibitors. Samples were incubated on ice overnight at 4 °C. Using syringe and blunt needles, 523 samples were homogenized and then incubated with 25 units/ml nuclease (benzonase TM , Sigma-524 Aldrich) and 50 mM MgCl2 for 30 minutes at 37 °C at 120 x g followed by a digestion with 1 mg/ml 525 PK (final concentration) for 1 hour at 37 °C at 120 x g. PK digestion was stopped by incubating 526 samples with 2 mM PMSF on ice for 15 minutes. Samples were incubated with 2 mM EDTA for 527 15 minutes at 37 °C at 120 x g. NaCl (0.25 M final) was added to all tubes followed by an equal 528 volume of 2% SB 3-14 buffer. For the sucrose gradient, a layer of 0.5 M sucrose, 100 mM NaCl, 529 10 mM Tris-HCl, and 0.5% SB 3-14, pH 7.4 was added to ultracentrifuge tubes. Samples were then 530 carefully transferred, and the tubes topped with TMS buffer. Samples were centrifuged at 200,000 x 531 g for 2 hours at 20 °C. The pellet was rinsed with 0.5% SB 3-14 in PBS. Pellets were resuspended 532 in 50 μl of 0.5% SB 3-14 in PBS and stored at −80 °C. Gel electrophoresis and silver staining were 533 performed to assess the purity of PrP Sc . 534

Heparan sulfate purification and analysis by mass spectrometry 535
Heparan sulfate (HS) was extracted from whole brain homogenates and purified by anion exchange 536 chromatography, as previously described 46  Negative stain electron microscopy 544 400 mesh lacey carbon grids (Ted Pella) were glow discharged, placed on a 7 μl droplet of purified 545 PrP Sc sample, and incubated for 20 minutes in a humidified chamber. Grids were then blotted on 546 filter paper, immersed briefly in Nano-W stain (Nanoprobes) and blotted again before incubating 547 on a droplet of Nano-W for 1 minute. Grids were then blotted dry and imaged on a FEI Tecnai 548 TF20 (200kV, FEG) with a 4k x 4k CMOS-based Tietz TemCam-F416 camera. Fibrils purified 549 from terminally ill mCWD-infected Ndst1 f/f tga20 +/+ SynCre-and SynCre+ brains were distributed 550 on the electron microscopy grids as single filaments as well as clusters of fibrils. Fibril lengths 551 were assessed by three blinded investigators, and lengths recorded using Image J or iMOD 552 software packages when both ends of a non-overlapping filament were readily visualized.

PrP C mutant generation and heparin sepharose chromatography 554
To identify the HS-binding domains, three clusters of lysine-and arginine-residues and one cluster 555 of asparagine residues were exchanged for alanine within mouse Prnp in a pcDNA3.1 vector by 556 and PrP C in all eluates were analyzed for PrP C level by immunoblot using POM19 antibody. At least 572 three experimental replicates were performed for each PrP C construct. 573

RT-QuIC analysis of mCWD-inoculated mice 574
The RT-QuIC reaction mix was composed of 10 mM phosphate buffer (pH 7. that PrP remained radiolabeled, Zr89-PrP C at different dilutions was loaded in 10% Bis-Tris gel 608 (Invitrogen) and electrophoresed four days after radiolabeling. Zr89 was added to the protein ladder 609 (Precision Plus Protein Standard, Dual Color, Bio-Rad) at the expected size for recPrP, 25 kDa, 610 and the gel was scanned in a phosphoimager (Typhoon). 611

Stereotaxic injection of conjugated PrP 612
Mice (14-16 weeks old) (n = 3 -4 mice/genotype/experiment) were anesthetized with isoflurane. 613 Mice were weighed and placed in a three-point stereotaxic apparatus (Stoelting). A 2 cm midline 614 incision was made in the skin over the sagittal suture to expose bregma, and a burr hole was drilled 615 in the left parietal bone (0.62 mm caudal, -1.75 mm lateral to bregma) using an Ideal Micro-drill. A 616 22-gauge needle (Hamilton) was inserted to a depth of 3.5 mm and 1 µl of Zr89-PrP C was injected 617 at a rate of 75 nl/minute over 15 minutes using a Quintessential Stereotaxic Injector (Stoelting). The 618 needle remained in the injection site for 10 minutes post-injection prior to removal from the brain to 619 prevent backflow. Animals were next removed from the stereotaxic device and placed on the PET 620 scanner (G.E. Vista), and radioactivity measurements were collected as dynamic scans using list 621 mode over 30 minutes. Scans were repeated 20 hours later. PET images were reconstructed using 622 Vista DR (G.E. Healthcare) software and the area and volume covered by radioactivity as well as 623 the signal intensity were assessed with FIJI (an ImageJ based image processing software). 624

Statistics 625
Log-rank (Mantel-Cox) tests were performed to assess survival differences between groups. A 626 Student's t-test (two-tailed, unpaired) with Bonferroni's post test was used to determine the 627 statistical significance between the Ndst1 f/f SynCre +/and SynCre -/and Ndst1 f/f GFAPCre +/and 628 GFAPCre -/mouse groups for the PrP C level of expression, lesion profiles, activated microglia, PrP Sc 629 glycoprofiles, PrP Sc conformation stability and PrP Sc fibril structure. One-way ANOVA with Tukey's 630 post test was performed to determine statistical significance in the levels of prion conversion by 631 PMCA. Two-way ANOVA with Bonferroni's post test was used to compare the composition of HS 632 associated with different prion strains, the number of vascular versus parenchymal plaques, the 633 binding affinity of PrP C mutants with heparin, the levels of PrP in the spinal cord, and the area 634 covered by high PET intensity signal (signal > 100 µCi). Unpaired two-tailed t-test with Bonferroni's 635 post test was used to compare the average sulfate groups per disaccharide as well as the plaque 636 length and the mCWD fibril length and solubility in Ndst1 f/f SynCre +/and SynCre -/brain. The 637 proportion of Ndst1 f/f SynCre +/versus SynCre -/mice with prion seeding in spinal cord was compared 638 using non-parametric Fisher's exact tests. For all analyses, p < 0.05 was considered significant. 639