Design and proof-of-concept in vitro testing of a GPI-anchorless V127-derived therapeutic agent of low immunogenicity

Although the precise mechanism by which V127 confers protection is still being debated, there is broad agreement that the protective mutant acquires the same fold as native PrP^C, barring minor structural differences surrounding the mutated residue [18–21]. Similarly, when PrP^C is expressed without its N-glycans, a natural consequence of the removal of the GPI-anchor attachment sequence, it will still acquire its natural PrP-fold [22–24] (Fig 1A and 1B). These insights are notable for immunological considerations, as they predict that an anchorless PrPV¹²⁷ΔGPI therapeutic agent would have low immunogenicity. In contrast, rAAV capsids can be expected to elicit a moderate immune response in naïve mice. Whereas in humans prior exposure to natural AAVs represents a major consideration [25], for studies in mouse models this concern is small if no repeat administrations are planned. Consequently, so long as the design of the synthetic coding sequence for the transgenic expression of PrP^V127 avoided highly immunogenic CpG islets [26], the packaged therapeutic construct was expected to exhibit low immunogenicity, largely restricted to the DNA payload. To mitigate immunogenicity of the latter, we capitalized on redundancies in the genetic code and eliminated most occurrences of CpGs (S1A Fig).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Design of an all-in-one rAAV vector of low immunogenicity that can promote cross-correction.

(A) Genome organization of the human (Hs) PRNP gene. Zoom-in depicts the sequence organization of the ORF. Colors highlight coding segments as follows: beige, signal peptide for co-translational ER targeting; blue, octarepeats; brown, hydrophobic region; green, β-sheets; red, α-helices; grey, signal sequence for attachment of GPI-anchor. (B) The protein structure model identifies the approximate position of the G127V residue within the tertiary fold of PrP^C with a red circle. For ease of orientation, the image depicts a PrP^C molecule that is inserted into a membrane bilayer with a GPI-anchor. Grey shapes protruding from the red-colored globular domain represent stylized N-glycosylations. (C) Cartoon depicting in vitro paradigm used to assess if a GPI-anchorless MoPrP^V126ΔGPI expression product can lower the steady-state prion infection level of RML-inoculated CAD5 cells. The schematic of the cell culture dish was created in BioRender. Schmitt-ulms, G. (2026) https://BioRender.com/hmig0pj (D) Side-by-side comparison of cellular extracts from RML-infected CAD5 cells that were transfected with plasmids coding for a negative control protein (spEGFP), MoPrP^V126 or MoPrP^V126ΔGPI. Note the additional relative fast migrating band reactive to the HuMD13-directed PrP antibody, whose appearance is consistent with the expression of unglycosylated MoPrP^V126ΔGPI. (E) Proteinase K analysis of protein extracts shown in Panel C. The ectopic expression of MoPrP^V126 or MoPrP^V126ΔGPI lowered the intensity of Proteinase K-resistant PrP^Sc. (F) Relative quantitation of Proteinase K signals of RML-infected CAD5 cells transfected with plasmids coding for spEGFP, MoPrP^V126 or MoPrP^V126ΔGPI.

https://doi.org/10.1371/journal.ppat.1014124.g001

To our knowledge, it had not been reported whether the absence of a GPI-anchor interferes with the protective capacity of the V127 mutant. To begin to answer this question, we preceded the in vivo work with proof-of-concept in vitro experiments in a mouse cell line (CAD5). CAD5 cells had been derived from an immortalized CNS catecholaminergic cells (Cath.a) [27] and shown to be susceptibility to RML prions [28]. To implement this pilot experiment, the cells were RML prion inoculated over a span of three weeks, then passaged onto fresh plates and transfected with MoPrPV¹²⁶ΔGPI expression constructs (note that V126 in murine PrP is the ortholog of V127 in human PrP). To enhance expression the open reading frame was placed downstream of a CBh promoter and the coding sequence for the mouse PrP signal peptide for translocation into the endoplasmic reticulum (ER) was replaced with the human prolactin signal peptide (PRLss). Cell culture plates that were side-by-side transfected with the respective expression constructs coding for anchored MoPrP^V126 or EGFP carrying a signal peptide for ER translocation (spEGFP) served as positive and negative controls. Five days after transfection, cellular extracts were either directly analyzed by western blotting or first digested with Proteinase K (Fig 1C). The western blots of total cellular extracts showed the expected cell-bound expression of anchored, and therefore N-glycosylated, PrP^V126. In contrast, MoPrP^V126ΔGPI was observed as a faster migrating and sharper band, consistent with it being expressed in unglycosylated form.

The Proteinase K-digested cellular extracts of all treatment conditions gave rise to the protease-resistant band pattern expected for this RML-infected cell model, consistent with the presence of unglycosylated and single or double N-glycosylated PrP contributing to the Proteinase K-resistant material. However, the intensity of the respective PrP-reactive signals differed, with negative control (spEGFP) expressing cultures giving rise to the strongest protease-resistant bands, and MoPrP^V126ΔGPI-transfected cultures showing the lightest Porteinase K-resistant PrP signals (Fig 1D). The subsequent quantitation of PrP signals validated this visual impression but did not reach statistical significance in the direct comparison of the protective effect of MoPrP^V126 versus MoPrP^V126ΔGPI (Fig 1E). Together, this pilot data validated the notion that V126 can be protective also in the context of an anchorless expression constructed and hinted at cross-correction having contributed to the protection it conferred.

Retro-orbital injection of BvPrnpG127VΔGPI into prion-inoculated mice caused similar survival extension regardless of rAAV vector preparation method or capsid choice

Rather than working with wild-type mice, we selected for the in vivo work in this study the recently introduced bank vole knock-in (BvPrnp ki) model in which the endogenous mouse Prnp open reading frame (ORF) was replaced with the respective BvPrnp ORF encoding isoleucine at polymorphic codon 109 [29]. Two considerations guided this choice: 1) BvPrnp-expressing mice had been shown to develop disease faster than wild-type mice [30]. 2) In anticipation of future inoculation work with human prions, we hoped to capitalize on bank vole PrP’s universal prion acceptor characteristics, when we translate results from work with mouse prion inocula (this work) to future work with human prion inocula [31,32].

To maximize the expression of the heterologous BvPrnp^V127ΔGPI, the payload was designed as a self-complementary DNA, thereby avoiding the replication step that single-stranded AAV vectors must undertake upon transduction [33], a limiting biology dependent on the availability of host cell factors (S1B Fig). Specifically, the expression cassette of the construct was flanked by inverted terminal repeats (ITRs) derived from the AAV2 serotype, with one of the ITRs carrying a small deletion within the terminal repeat sequence (TRS) motif known to manifest in self-complementarity [34]. The expression of BvPrnp^V127ΔGPI was driven by the chimeric chicken β-actin promoter with intron (CBh) that had been shown to promote the steady expression of transgenes in mouse brains for extended periods [35].

While the study was underway, several advances were made in the rAAV field, including new purification methods and capsids with improved ability to cross the blood brain barrier (BBB). In line with the pilot nature of the study, we included comparisons of two of these developments: 1) an affinity capture-based purification method using AAVX matrices, which we compared to the commonly used iodixanol gradient centrifugation (S1 Fig) [36], and 2) the 9P31 (Voyager Therapeutics Inc.), which has been reported to achieve several-fold greater CNS penetrance than PHP.eB, a capsid widely for CNS targeting in mice after intravenous delivery [37].

Purity assessments of rAAV vectors that we prepared by AAVX affinity chromatography showed predominantly three bands derived from the three capsid isoforms VP1, VP2, and VP3 in the expected approximate ratio of 1:1:10 after SDS-PAGE separation (S1D Fig). To test the 9P31 vector, we assembled a self-complementary rAAV vector identical to the one described for the expression of BvPrnp^V127ΔGPI, except that the BvPrnp^V127ΔGPI ORF was replaced with a coding sequence for spEGFP (S1E Fig) [38]. In line with previous reports, the AAVX method led to cleaner preparations of the recombinant AAVs than iodixanol gradient centrifugations [36] and the 9P31 capsid led to greater CNS expression of the ectopic proteins than delivery in PHP.eB capsids [37,39].

As we were gearing up for the in vivo work, our breeding stock of BvPrnp ki mice produced relatively small litters, which forced us to replace plans of parallel inoculations with staggered inoculations. We hypothesized that the expression of V127 within the context of an anchorless BvPrnp expression construct (BvPrnp^V127ΔGPI) might lead to the longest survival extension due to enhanced cross correction. Consequently, we began the in vivo studies by comparing side-by-side mice that had been prion-inoculated on the same day, then were either left untransduced or were intravenously (retro-orbital) transduced 60 days later with rAAV vectors coding for spEGFP (negative controls) or BvPrnp^V127ΔGPI (Fig 2A). The latter cohort consisted of three sub-cohorts. Two of these were transduced with 9P31 vectors that had been purified either by iodixanol density centrifugation or AAVX affinity chromatography (S1C Fig). The third sub-cohort was transduced with rAAV vectors encapsulated in the PHP.eB capsid that had also been purified by AAVX affinity chromatography. Deferred were the analyses of mice transduced with 9P31 encapsulated vectors coding for anchored BvPrP^V127.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Heterologous expression of rAAV vector-delivered BvPrP^V127 ΔGPI causes survival extension in RML-infected BvPrnp ki mice.

(A) Timeline of pilot in vivo prion infection and treatment study in BvPrnp ki mice. Note that all endogenous and heterologous bank vole sequences employed in this work carried the I109 polymorphism. The schematic of the mouse was created in BioRender. Schmitt-ulms, G. (2026) https://BioRender.com/hmig0pj (B) Kaplan-Meier chart comparing the survival of negative control mice and treated mice. (C) Weight analysis chart depicts the relative weights of mice shown in Panel B from the time of 100 dpi to the time of sacrifice when the animals met humane prion disease endpoints. Note that weights of individual mice were normalized to their weight at 100 dpi to account for individual weight differences and sex-specific weight differences. (D) Nesting analysis chart showing the nesting score of mice shown in Panel B from 100 dpi to the end of the study. (E) Chart documenting that the alternative rAAV vector preparation methods based on AAVX affinity capture or iodixanol gradient centrifugation for purifying the therapeutic 9P31-BvPrnp^V127ΔGPI had an insignificant influence on survival extension. (F) No significant differences in the survival times afforded by the expression of BvPrnp^V127ΔGPI was observed whether the therapeutic payload was encapsulated in the PHP.eB or 9P31 capsid. In all graphs, 95% confidence intervals are indicated by shading around the curves. Asterisks shown in the survival charts indicate time-points when mice were prematurely sacrificed during the study to serve as controls in downstream biochemical analyses. Consequently, these three mice were excluded from the statistical analyses.

https://doi.org/10.1371/journal.ppat.1014124.g002

As expected, RML-prion inoculated BvPrnp ki mice began to show symptoms 130–150 days post-inoculation (dpi) and had to be sacrificed shortly thereafter when they reached prion disease endpoint (Fig 2B). Whether the prion-inoculated mice were left untransduced or transduced with the negative control 9P31-spEGFP vector made only a small difference to their survival, with transduced mice succumbing slightly sooner. In contrast, the cohort of mice that had been made to express the therapeutic BvPrnp^V127ΔGPI construct, survived approximately 50 days longer. One characteristic of the Kaplan-Meier curve for this cohort is a relatively broad spread of survival times from 150 to a maximum of 232 days, possibly reflecting variances associated with the administration of the therapeutic capsids (Fig 2B). Plotted results from monitoring body weights (Fig 2C) and nesting scores (Fig 2D) of the mice reflected both the survival extension as well as the more gradual decline in the BvPrnp^V127-ΔGPI cohort. When the latter results were deconvoluted based on how the 9P31-BvPrnp^V127ΔGPI capsids were prepared (Fig 2E) or whether the payload was encapsulated in 9P31 or PHP.eB, subtle shifts to longer survival time were observed when rAAV vectors had been AAVX purified and encapsulated in 9P31, but no significant differences emerged (Fig 2F). In particular, the finding that the increased ectopic expression conferred by 9P31 was not correlated with an increased survival extension was surprising as it points toward a ceiling efficacy in this paradigm.

Survival extension afforded by BvPrnp^V127ΔGPI is accompanied by reduced PrP^Sc accumulation but does not correlate linearly with steady-state expression levels of therapeutic payload

To begin to understand how the expression of BvPrnp^V127ΔGPI had delayed the disease, at 152 dpi, i.e., when the longest surviving prion-inoculated control mouse that had been transduced with the 9P31-spEGFP vector reached prion disease endpoint, a small number of BvPrnp ki mice, which had been treated with the 9P31 encapsulated vector coding for the BvPrnp^V127ΔGPI mutant, were also sacrificed, along with naïve age-matched control BvPrnp ki mice that had neither been prion-inoculated nor transduced. Both, these naïve mice, and the RML-inoculated mice treated with 9P31-BvPrnp^V127ΔGPI showed no overt signs of prion disease at the time of their sacrifice. Sagittal half brains of these mice were formalin-fixed, and the remaining half brains were homogenized and extracted, then processed for western blot analyses with or without prior digestion with Proteinase K. This analysis revealed the expected increase in total PrP levels, a result of the accumulation of PrP^Sc, in RML prion-inoculated mice (relative to uninoculated age-matched control mice) that had been transduced with 9P31-spEGFP negative control viruses (Fig 3A and 3B). In contrast, RML prion-inoculated mice that had been transduced with the therapeutic 9P31-BvPrnp^V127ΔGPI vector showed lower levels of PrP^Sc accumulation but were not devoid of Proteinase K-resistant PrP^Sc (Fig 3C and 3D).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Expression of heterologous BvPrnp^V127ΔGPI in RML-inoculated BvPrnp ki mice mitigates accumulation of PrP^Sc and increases total PrP signal intensities more than threefold.

(A) Western blot analysis of extracts from sagittal half brains of BvPrnp ki mice that were either from a cohort of naïve mice, or RML-inoculated mice that were mock-treated with a 9P31-spEGFP vector or treated with the 9P31-BvPrnp^V127ΔGPI vector. Note that all animals were sacrificed by 152 days, i.e., when the last surviving RML-inoculated, mock-treated cohort reached its prion disease endpoint. A western blot depicting steady-state beta actin (Actb) levels served as a loading control in this analysis. (B) Quantification of combined BvPrP and BvPrP^V127ΔGPI signal levels (normalized to Actb levels) revealed significant differences (Student’s t-test) in the three cohorts compared in Panel A. (C) Analysis of Proteinase K-resistant PrP^Sc in protein extracts analyzed in Panel A. (D) Bar graph quantifying band intensities of Proteinase K-resistant PrP^Sc shown in Panel A The data were normalized to PrP^Sc levels in the mock-treated cohort transduced with the 9P31-spEGFP vector. (E) Therapeutic vectors encapsulated in 9P31 caused robust expression of BvPrnp^V127ΔGPI that exceeded endogenous total PrP levels more than threefold. Western blot analysis of extracts from sagittal half brains of BvPrnp ki mice that were RML-inoculated but then were either mock-treated through transduction with the spEGFP vector or transduced with therapeutic vectors coding for BvPrnp^V127ΔGPI. For this analysis, brains from all mice were harvested after the animals reached their individual prion disease endpoint as shown by their unique dpi. To facilitate the recognition of the therapeutic BvPrnp^V127ΔGPI signal amongst a complex mixture of N-glycosylated endogenous BvPrP expression products, N-glycans were removed by PNGase F digestion prior to the western blot analysis. Note the pronounced signal of the BvPrnp^V127ΔGPI band, which migrated as expected a bit faster than the endogenous wild-type full-length BvPrP band on account of it lacking the GPI-anchor. The detection of Actb served again as loading control in this analysis. (F) The bar graph quantifies the intensities of combined BvPrP and BvPrnp^V127ΔGPI signal levels in the treatment cohorts depicted in Panel E. (G) Bar graph quantifying the intensities of C1 signals of treatment cohorts depicted in Panel E. Note the slight increase in both C1 and C2 levels in brain extracts of mice that had been transduced with 9P31 encapsulated rAAV vectors coding for BvPrnp^V127ΔGPI, possibly indicating that the ectopic expression of this protein contributed to the total intensity of these proteolytic fragments. (H) Side-by-side analysis of one brain extract from each of the four treatment cohorts analyzed in Panel C with or without digestion with Proteinase K followed by digestion with PNGase F. Note that the bands visible in lanes 6, 9, and 12 don’t appear to constitute doublet signals, which would be expected if the anchorless BvPrP^V127ΔGPI had contributed to the Proteinase K-resistant PrP^Sc. (I) Analysis of Proteinase K-resistant PrP^Sc levels in subset of brain extracts of mice analyzed in Panel E. This analysis focused on the mice that had been transduced with the 9P31-encapsulated and AAVX-purified rAAV vectors coding for BvPrnp^V127ΔGPI, which showed the highest ectopic expression of BvPrnp^V127ΔGPI. The western blot revealed no additional band whose migration would be consistent with the possible formation of Proteinase K-resistant BvPrnp^V127ΔGPI.

https://doi.org/10.1371/journal.ppat.1014124.g003

Once the remaining cohort of the 9P31-BvPrnp^V127ΔGPI-treated mice had reached the disease endpoint, we undertook additional western blot analyses, after processing their half brains as above. These experiments were guided by an interest in understanding why the sub-cohort, which had been treated with the PHP.eB-encapsulated payload coding for BvPrnp^V127ΔGPI had shown a similar survival extension as the mice whose identical therapeutic construct had been encapsulated in the 9P31 capsid, a counterintuitive result based on our prior observation that 9P31 vectors mediated an approximately seven-fold higher CNS transduction than PHP.eB vectors when payloads, virus preparation, and administration steps were identical [39]. Since we were particularly interested in determining the steady-state BvPrnp^V127ΔGPI expression levels that had been reached in the treated cohort, for this western blot analysis the total brain extracts were first digested with PNGase F so that PrP isoforms differing solely in their N-linked glycans would be reduced to a single band (Fig 3E and 3F). Intriguingly, this analysis corroborated the relative CNS transduction potencies of 9P31 and PHP.eB by showing that the BvPrnp^V127ΔGPI-derived western blot signals were considerably stronger when the payload had been encapsulated in 9P31 than the respective signal in brain extracts of mice that had been transduced with the corresponding PHP.eB vector. In fact, the western blot signal that we interpreted to represent BvPrnp^V127ΔGPI by its absence in non-transduced RML-inoculated mice or in mice that had been transduced with 9P31-spEGFP emerged as the strongest PrP antibody-reactive signal. The intensity of this band in 9P31-BvPrnp^V127ΔGPI mice exceeded more than threefold the respective signals for endogenous wild-type BvPrP. This result indicated that the transduction worked better than we had anticipated but it also indicated that the therapeutic potency of BvPrnp^V127ΔGPI, when measured based on the survival extension it conferred, did not correlate linearly to its expression level in BvPrnp ki mice. We also noted that C-terminal proteolytic bands of PrP^C were slightly elevated in cohorts transduced with 9P31-encapsulated virus particles coding for 9P31-BvPrnp^V127ΔGPI (Fig 3G), consistent with a scenario whereby the presence of the protective V127 variant may have allowed more of these fragments to accumulate before the mice reached the prion disease endpoint.

Finally, we sought to determine if the highly expressed BvPrP^V127ΔGPI contributed to the formation of Proteinase K resistant material in the treatment cohort after these mice had succumbed to prion disease. To reduce the complexity of signals expected if isoforms with varying N-glycan occupancy were present, we first digested representative brain extract from each of the four cohorts with Proteinase K, then removed N-glycans from the resolubilized PrP^Sc material by an additional digestion with PNGase F. The side-by-side analysis of the resultant PrP products next to undigested or only PNGase F digested samples revealed that mice, which expressed both the endogenous BvPrnp gene and the synthetic BvPrnp^V127ΔGPI construct, still only gave rise to a single PrP^Sc signal which ran at the same level as the respective band in mice that were not transduced (Fig 3H). Previous work with prion-inoculated transgenic mice that expressed both wild-type PrP and anchorless PrPΔGPI documented that Proteinase K-resistant unglycosylated bands derived from anchorless PrPΔGPI run considerably faster than the corresponding wild-type-derived bands and can easily be distinguished in a western blot analysis [11]. No such bands were observed even in mice characterized by the highest ectopic expression of BvPrP^V127ΔGPI (Fig 3I). These results suggested that BvPrP^V127ΔGPI did not contribute in RML-infected mice to the formation of Proteinase K-resistant PrP^Sc, even though the mice expressed relatively high levels of the protective variant and died at ~200 dpi of symptoms consistent with a late-stage prion disease diagnosis.

More than threefold increase in total PrP and 20-fold higher levels of unglycosylated PrP in mice transduced with BvPrnp^G127V-ΔGPI at gene levels equivalent to endogenous BvPrnp

Next, we sought to assess the extent to which the brain-wide delivery of the genetic payload of our gene therapy had occurred. To this end, we capitalized on nucleotide sequence differences in the endogenous BvPrnp ORF and the synthetic BvPrnp^V127ΔGPI-ORF. Specifically, we identified a stretch within the respective ORFs that could be amplified with the same PCR primers because it was flanked by identical nucleotide sequences yet could only be digested by one or another restriction enzyme, namely EagI for endogenous BvPrnp gene sequences and CsiI for BvPrnp^V127ΔGPI inserted into the genome following transduction (Fig 4A). By comparing the signal intensity of the fragments released by the respective restriction enzymes, this assay could inform about the relative amount of genomic DNA (gDNA) coding for BvPrP that was contributed by the endogenous and heterologous synthetic ORF. The analysis revealed that the relative amounts of genomic endogenous BvPrnp ORFs were approximately matched to those of the heterologous synthetic BvPrnp^V127ΔGPI ORFs in mice that had been transduced with the 9P31- BvPrnp^V127ΔGPI vectors (Fig 4B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. 3.6-fold increase in total PrP and 20-fold higher levels of unglycosylated PrP in mice transduced with BvPrnp^G127V-ΔGPI at gene levels equivalent to endogenous BvPrnp.

(A) Schematic of PCR and restriction enzyme (RE) digest analysis of genomic brain DNA preparations of mice for determining the relative abundance of endogenous and heterologous BvPrnp ORFs. (B) Transduction of mice with 9P31-BvPrnp^V127ΔGPI gave rise to the detection of RE cleavage products specific for heterologous BvPrnp^V127ΔGPI. Note that the signal intensities of RE digested PCR products in lanes 10 and 11 are approximately matched, indicating similar levels of BvPrnp and BvPrnp^V127ΔGPI ORFs in the genome. (C) Schematic showing sample workup scheme for global proteome analysis. The image of the sagittal half brain was created in BioRender. Schmitt-ulms, G. (2026) https://BioRender.com/hmig0pj. (D) MS2 spectrum of the tryptic peptide ESQAYYEGR contributed by endogenous and heterologous pools of BvPrP. (E) MS2 spectrum of the tryptic peptide GENFTETDVK comprising the second ‘NxT’ N-glycan acceptor consensus motif within BvPrP. (F-I) Quantitation of tryptic BvPrP peptides that are present in all BvPrP sequences analyzed (F), only observed in the sub pool of BvPrP that is non-glycosylated at GENFTETDVK. (G), only present in endogenous BvPrnp expression products (H), or only present in the heterologous BvPrnp^V127ΔGPI expression product (I). (J) Steady-state total BvPrP levels, comprising both endogenous and heterologous BvPrP expression products, revealed in global proteomic analysis based on 52% sequence coverage and the assignment of 151 PSMs to 12 unique BvPrP-derived peptides.

https://doi.org/10.1371/journal.ppat.1014124.g004

Although instructive in their own right, the DNA-based comparison cannot inform about steady-state protein levels of the endogenous BvPrP versus the ectopically expressed BvPrP^V127ΔGPI protein. This is because the method is agnostic to the influence of the promoters, with the expression of the BvPrnp knock-in being driven by the endogenous mouse Prnp promoter, and the ectopically expressed BvPrnpV127 relying on the CBh promoter. Moreover, the steady-state protein levels between the two proteins can also be expected to differ on account of the presence or absence of the GPI-anchor, which is known to influence the half-life of PrP. To dissect relative abundances of the endogenous and ectopic PrP in detail and deconvolute how the absence of the membrane anchor in BvPrP^V127ΔGPI impacts the ratio of N-glycosylated to glycosylated PrP in a half-brain, we generated samples for orthogonal mass spectrometry analyses from the same samples we had subjected to western blot analysis (Fig 3A).

To this end, the total brain extracts were fully denatured in the presence of urea, then reduced and alkylated, and finally trypsinized. Using 30-minute gradients, tryptic mixtures were separated on a reversed phase column. The effluents from this separation were online coupled by nanospray ionization to an Orbitrap Astral mass spectrometer, which was operated in data-independent acquisition (DIA) mode (Fig 4C). The depth of the proteome dataset that the DIA mode afforded allowed specific BvPrP-derived tryptic peptides to be compared in their relative quantities.

We focused these analyses initially on three specific BvPrP-derived tryptic peptides in the nine samples, namely 1) the peptide ESQAYYEGR present in both endogenous and heterologous PrP (Fig 4D), 2) the peptide GENFTETDVK, harboring one of the two N-linked glycan acceptor sites (NxT) present in the protein (Fig 4E), and 3) the respective BvPrP peptides that surround the G127 or V127 residues and therefore can be unequivocally assigned to endogenous BvPrP or the heterologous BvPrP^V127ΔGPI, respectively. These analyses detected the generic BvPrP peptide at lowest levels in the naïve mice, at slightly increased levels in RML-inoculated mice, and at 3.6fold higher levels in the 9P31-BvPrnp^V127ΔGPI-ORF-transduced mice (Fig 4F). The peptide harboring the unmodified N-glycan acceptor site was present at approximately 20-fold higher levels in the BvPrnp^V127ΔGPI-ORF-transduced mice. Considering the more than three-fold higher total BvPrP levels in these mice, this result suggests that the heterologous anchorless protein is approximately 6-fold less likely to carry the N-glycan at the respective acceptor site than the endogenous protein (Fig 4G). The BvPrP peptide comprising the wild-type BvPrP G127 residue was observed at about twice the level in the RML-inoculated samples (Fig 4H), consistent with the stabilization and overall increase in steady-state PrP levels that are commonly observed in prion-infected mice. Finally, as anticipated, the V127-comprising peptide was not detected in naïve or RML inoculated mice transduced with the control 9P31-spEGFP vector but was robustly detected in mice transduced with 9P31-BvPrnp^V127ΔGPI (Fig 4I).

Consistent with the western blot data and the genomic DNA analyses, the mass spectrometry-based quantitation indicated that the total steady-state BvPrP levels in mice which had been transduced with the 9P31-BvPrnp^V127ΔGPI-ORF vector were approximately 3.6-fold higher than the respective quantity in the naïve BvPrnp ki mice (Fig 4J).

However, these data also indicated that this increase at the protein level was not reflected at the gDNA level, where endogenous and heterologous BvPrnp ORFs were approximately matched in 9P31-BvPrnp^V127ΔGPI transduced brains.

BvPrnpV127ΔGPI reduced levels of Proteinase K-resistant PrP^Sc relative to anchored BvPrnp^V127

To begin to assess if cross-correction of an anchorless BvPrnp^V127ΔGPI expression construct can supersede the protection conveyed by the expression of anchored BvPrnp^V127, we next compared directly the survival curves of BvPrnp ki mice that had been transduced with 9P31-delivered virus vectors, whose therapeutic payload differed only in the inclusion or omission of the GPI-attachment sequence within the protective BvPrnp^V127 ORF. Consistent with the hypothesis that cross-correction may enhance the protective capacity of BvPrnp^V127ΔGPI transduced mice, the Kaplan-Meier curves of mice that had received retro-orbital injections of the 9P31-delivered vectors coding for anchored BvPrnp^V127 led to a shorter survival extension of ~25 days (Fig 5A). This finding was validated by western blot analyses of brain extracts of age-matched mice from the two treatment cohorts. Specifically, these analyses documented that levels of Proteinase K-resistant PrP^Sc were significantly higher (30%, p < 0.05) in mice that had been treated with the expression construct coding for anchored V127 (Fig 5B and 5C). As in our previous western blot analyses (Fig 3), no evidence of Proteinase K-resistant BvPrnp^V127ΔGPI was detected in these western blots when we analyzed samples of 9P31-BvPrnp^V127ΔGPI-transduced mice.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. BvPrnp^V127ΔGPI reduced levels of Proteinase K-resistant PrP^Sc relative to anchored BvPrnp^V127 yet gave rise to a distinct protein deposition phenotype.

(A) Comparison of survival extension in cohorts of BvPrnp ki mice following retro-orbital transduction of 9P31 encapsulated virus vectors coding for anchored BvPrP^V127 versus anchorless BvPrnp^V127ΔGPI. (B) Side-by-side comparison of steady state total PrP levels with and without prior Proteinase K digestion in end-stage prion disease BvPrnp^V127ΔGPI transduced with the respective therapeutic constructs. (C) Relative quantification of Proteinase K-resistant signals in the western blot shown in Panel B. (D) IHC characterization of PrP-reactive signals within the hippocampal formation of an RML-inoculated BvPrnp ki positive control mouse that was not transduced. (E) Negative control IHC image of a hippocampus from a BvPrnp ki mouse that was not prion-inoculated but transduced with a 9P31-EGFP virus. (F) BvPrnp ki mice following retro-orbital transduction of 9P31 encapsulated virus vectors coding for anchorless BvPrnp^V127ΔGPI at 196 dpi. The IHC images were counterstained with Hematoxylin-Eosin. Sizing bar is 100 µm.

https://doi.org/10.1371/journal.ppat.1014124.g005

When the brains of 9P31-BvPrnp^V127ΔGPI-transduced mice were assessed by immunohistochemistry (IHC), next to negative control mice that had been RML-inoculated and mock transduced (Fig 5D) or non-inoculated mice that were transduced with an spEGFP coding vector (Fig 5E), the pattern of protein deposition differed slightly. Specifically, a subset of deposits in the hippocampus of mice transduced with BvPrnp^V127ΔGPI tended to be larger than the corresponding hippocampal PrP deposits in mice transduced with spEGFP, which were smaller and more diffuse, as expected for the RML strain (Fig 5F). Further investigation of adjacent brain sections by IHC directed at Gfap or Iba1 validated the presence of astrocytosis and microgliosis in end-stage RML-infected mice but revealed no apparent difference in this regard in mice transduced with the 9P31-spEGFP or 9P31-BvPrnp^V127ΔGPI viral vectors (S2 Fig). A previous report described a predominant perivascular deposition phenotype in transgenic mice engineered to overexpress anchorless mouse PrnpΔGPI without the protective mutation [11]. No conspicuous association of deposits with blood vessels was observed in 9P31-BvPrnp^V127ΔGPI-transduced mice but this point warrants further investigation.

Expression of BvPrnp^V127ΔGPI slows perturbations to the proteome observed in prion-inoculated mice

Although a fair bit is known about how prion diseases affect specific proteins, and it has been reported that the disease affects global protein synthesis following activation of the unfolded protein response [40], systematic in-depth proteome comparisons of age-matched naïve and prion-inoculated end-state prion disease brains have, to our knowledge, not been reported. Moreover, despite reports of V127 conferring some resilience to structural dynamics in vitro [21,41], our understanding of how V127 confers its protection against prion diseases remains limited. Although it is apparent that the expression of V127 slows the accumulation of PrP^Sc, it is not known if its protective effect manifests in a mere slowing of prion disease-associated perturbations or involves more specific protective changes to the proteome. To fill these knowledge gaps, we further interrogated the mass spectrometry dataset of brain extracts from age-matched naïve BvPrnp ki mice, and the corresponding prion-inoculated and mock 9P31-spEGFP- versus 9P31-BvPrnp^V127ΔGPI-treated mice (Fig 4C).

A high degree of run-to-run proteomic sequence coverage afforded by the DIA acquisition mode obviated the need for isobaric labeling, thereby enabling MS2-based relative quantitation of consecutively analyzed samples. Taken together, this configuration allowed the deep unsupervised characterization of the mouse brain proteome, leading to the relative quantification of 4,874 proteins in all cohorts, comprised of three biological replicates for each of the three cohorts (Fig 6A). More than 800 of these proteins were identified based on peptide-to-spectrum matches (PSMs) that accounted for more than 50% coverage of their protein sequence, with approximately half of the protein identifications supported by ≥100 PSMs, and more than 4,800 protein identifications based on ≥10 PSMs. Even at cursory inspection, it was apparent that the three biological replicates for each cohort were more like one another than samples from other cohorts, an impression supported by results from an unsupervised hierarchical clustering analysis (see below).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Expression of BvPrnp^V127ΔGPI slows perturbations to the proteome observed in prion-inoculated mice.

(A) Global proteome summary diagram documenting benchmarks of the analysis, including the number of proteins whose steady-state levels in RML-inoculated and mock-treated brains were ≥ 33% altered relative to their average protein levels in naïve mouse brains. (B) Result from hierarchical clustering analysis of global proteome dataset. Note that the unbiased clustering analysis grouped the cohorts in accordance with the study design, i.e., it identified three groups of highly similar brain extracts. The global proteomic signature of the mice, which were RML-inoculated and transduced with 9P31-delivered BvPrnp^V127ΔGPI vectors, places them between the naïve and the RML-inoculated but mock-treated mouse cohort. (C) Zoom-in into the hierarchical clustering results focused on proteins whose steady-state levels are upregulated in RML-inoculated mice. The graph indicates highly consistent changes in the direction of perturbation for a given protein and corroborates that many proteins are to a lesser degree altered in their steady-state levels in the 9P31-BvPrnp^V127ΔGPI treatment cohort than the 9P31-spEGFP mock-treated cohort. (D) Short-listed results from KEGG pathway analyses undertaken with proteins whose levels were ≥ 33% up- or down-regulated.

https://doi.org/10.1371/journal.ppat.1014124.g006

When the relative abundances of individual proteins in all nine samples were computed by forming ratios of their combined ion intensities in each sample and their respective average ion intensities in the naïve BvPrnp ki cohort, it emerged that most proteins were not altered by more than 33% in their steady-state levels in the disease. Yet, taking the opposite perspective, 780 genes were ≥33% upregulated in their expression and 455 proteins were ≥33% down-regulated at end-stage prion disease, relative to age-matched naïve BvPrnp ki mice. BvPrnp^V127ΔGPI-treated mice fell between these two extremes and, importantly, exhibited no proteomic drifts relative to naïve mice that were not also observed in the RML-inoculated mice sacrificed at the humane prion disease endpoint. However, the abundance changes were less pronounced for these treated mice than what was observed for the respective protein abundances in the 9P31-spEGFP-transduced mice (Fig 6B and 6C). Taken together, these observations were consistent with the interpretation that the expression of BvPrnp^V127ΔGPI does not prolong survival by having a specific effect on a sub proteome which can compensate for prion disease-induced proteome perturbations.

We next undertook pathway analyses to investigate more systematically the changes to the proteome that manifested in RML-inoculated BvPrnp ki mice. To this end, we interrogated the KEGG pathway repository with identifiers of proteins that were either ≥33% upregulated or downregulated (Fig 6D). Because these analyses were secondary to the main objective to understand how mock-treated versus BvPrnp^V127ΔGPI-treated mice differ, the data cannot delineate the possible contributions of the rAAV transduction from proteomic shifts caused by prion disease alone. Consistent with expectations for prion disease-associated perturbations, these analyses revealed a highly significant reduction in the steady-state levels of proteins associated with synaptic biology. Interestingly, within the diverse types of synapses, glutamatergic synapses were the most significantly impacted (p = 8.0 E-37), with the GABAergic (2.3 E-19), adrenergic (1.5 E-13), dopaminergic (5.9 E-13), and cholinergic synapses (1.2 E-12), showing lesser impacts but still returning highly significant p-values. Also consistent with known perturbations of prion diseases to sleep-wake patterns, the KEGG pathway defining components of circadian entrainment was also one of the most downregulated (p = 2.6 E-18).

Several pathways were upregulated (≥33%) relative to age-matched naïve BvPrnp ki mice. Chief among them are the KEGG pathway defining components that facilitate ‘Protein processing in the endoplasmic reticulum’ (p = 4.3 E-12). Along with ER proteins, spliceosome proteins were next in this category in the order of relative significance (p = 3.5 E-11). As such, these analyses added granularity to the notions that prion diseases are characterized by marked deficiencies in synaptic biology and synaptic entrainment. The latter manifested alongside attempts of brain cells to delay their demise by increasing their ER processing capacity.

Proteomic signature of RML prion disease marked by astrocytosis, microgliosis, complement activation, and cellular calcium influx recognizable in asymptomatic mice

To get a better understanding of the biological basis for the survival extension, we next sought to answer the question whether the proteins whose steady-state levels are most increased in prion disease, can also be seen to show the same pattern 50 days ahead of their death in animals that had been transduced with 9P31-BvPrnp^V127ΔGPI vectors. We focused this analysis on gene products whose steady-state levels are most upregulated, as this would allow them to be positively identified. Consistent with the hierarchical clustering data, this analysis revealed that the treated animals not only showed the same proteins to be upregulated but the levels of their upregulation was mostly consistent and marked by a relatively high signal-to-noise, i.e., twofold to 25-fold steady-state protein levels relative to naïve mice, suggesting that the proteomic signature of these changes may become recognizable well ahead of the time point chosen for this pilot analysis (Fig 7). To understand the cellular context in which the proteins operate whose steady-state levels were most increased, we capitalized on the availability of an in-depth mouse brain single cell transcriptomics dataset [43], accessible through the Allen Brain Map, which can be interrogated with mouse gene lists through the ‘Transcriptomics Explorer’ algorithm. When we queried this algorithm with the 50 mouse genes whose protein products were most upregulated in RML-infected mouse brains, it returned almost exclusively assignments to brain cell types that are non-neuronal (Fig 7, heatmap data). Several of the expression products of the genes in the list, including GFAP and AIF, are known markers of astrocytosis and microgliosis, respectively, indicating the attempt of RML-infected brains to respond to injury.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Diagnostic signature of RML prion disease marked by astrocytosis, microgliosis, complement activation, and cellular calcium influx is already recognizable in asymptomatic mice.

Shortlist of proteins whose steady-state brain expression levels were most increased in RML-inoculated mice, relative to naïve control mice. Comparing the enrichment levels of individual proteins across the three treatment cohorts reveals that a proteomic signature similar to the 9P31-spEGFP transduced control mice (here shown as RML_1, RML_2, and RML_3) was already present in the 9P31-BvPrnp^V127ΔGPI transduced mice (here shown as RML + V127-ΔGPI_1, RML + V127-ΔGPI_2, and RML + V127-ΔGPI_3) approximately 50 days before they reached their humane prion disease endpoint. The list is dominated by proteins known to play a role in astrocytosis (highlighted in red color), microgliosis (orange), or the complement response (yellow). Also note the enrichment of proteins known 1) to carry calcium binding domains or to be functionally dependent on calcium (blue colored), 2) to be functionally linked to the reorganization of the actin cytoskeleton (brown), or 3) to form a CD44-CD109 complex (magenta). The column headed ‘Aging mice’, identifies proteins whose transcripts were recently shown to be increased in aging mice [42], thereby possibly discouraging their use as specific prion disease biomarkers. On the right, a partial heat map (from the Allen Brain Map) indicates the brain cell types that were previously reported to express most prominently the transcripts of genes shown [43].

https://doi.org/10.1371/journal.ppat.1014124.g007

Since many of the same biological processes are also activated in the ageing brain, we wondered how the shortlist of 50 proteins compares to a similar shortlist of 73 transcripts that were recently observed to be most consistently increased in ageing mice [42]. Remarkably, 16 of the 50 genes that were most upregulated in prion disease were also most robustly increased in ageing mice. Three additional characteristics of the top 50 gene products are notable and may warrant further investigation to evaluate their usefulness for generating a shortlist of prion disease-specific biomarkers, namely, 1) the list being significantly enriched (p = 6.7 E-3) in proteins that carry calcium binding domains or whose function is calcium-dependent (S100a4, S100a11, S100a6, Anxa4, Itgb2, Tgm1), 2) three of the four most upregulated proteins, S100a4, Flnc, and Pdlim4, being functionally linked to actin cytoskeleton organization, 3) the presence of CD44 and CD109 in the list, which are known to form functional interactions and are neither associated with non-neuronal cells nor being activated in aging mice. Taken together, this analysis revealed a signature of prion disease perturbations of potential use for monitoring disease progression that is strongly blunted by the gene therapy and therefore may also serve to gauge treatment success.

Lower steady-state levels of cell surface proteins residing in proximity to PrP^C cannot be accounted for by the loss of neurons that occurs late in the disease

In a prior report, we studied the molecular environment of PrP^C in a mouse brain using an in vivo crosslinking affinity capture paradigm and had published the top-ranked 40 proteins [44]. With the deep global proteome analyses at hand, we revisited this shortlist to determine if proteins in proximity to PrP^C await a specific fate late in the disease. All but one of the 40 proteins in the original paper were also identified in this global proteome analysis based on at least 10 unique peptides and a range of 185–1325 PSMs, thereby affording robust relative quantitation. Interestingly, at late-stage prion disease, none of these PrP^C candidate interactors was upregulated, many were approximately two-fold down regulated, and only a few were not affected in their steady-state levels. A common feature of the latter subgroup seems to be that they encounter the prion protein within its passage along the secretory pathway, whereas the downregulated proteins reside next to PrP at the cell surface. Many of these proteins belong to the subset of proteins that were most strongly and consistently downregulated in the entire dataset (Fig 8A). In fact, if proteins which we had reported to reside in proximity to PrP at the cell surface were captured in a KEGG pathway, its downregulation would emerge among all known KEGG pathways annotations as a highly significantly downregulated pathway (p = 2.37E-25), with only three cell surface proteins, namely contactin-1, the amyloid precursor protein, and the myelin associated glycoprotein, bucking this trend.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Lower steady-state levels of cell surface proteins residing in proximity to PrP^C cannot be accounted for by the loss of neurons that occurs late in the disease.

(A) Proteins reported to reside in proximity to PrP^C at the cell surface are downregulated in end-stage RML prion-inoculated mice. The list depicts previously reported proteins in an affinity-capture PrP interactome from in vivo crosslinked mouse brains [44]. The intensity of the red shading reflects the level of reduction of a given protein entry relative to its average steady-state levels in age-matched naïve control mice. Note the consistent red shading in proteins whose dominant subcellular compartment is the cell surface. (B) Although neurons are understood to die in late-stage prion disease, the steady-state levels of several neuronal markers (NeuN, nestin, or Tuj1) are not noticeably lowered in a sagittal half brain extracts generated at the prion disease end-stage, indicating that for a majority of brain neurons the disease merely causes the retraction of synapses but not their full disappearance. The table also shows well-known brain cell marker proteins that are preferentially expressed in astrocytes, microglia, or oligodendrocytes, indicating the activation of astrocytosis and microgliosis (recognizable by the blue shading that reflects upregulation), as well as consistent loss oligodendrocytes (recognizable by the consistent red shading of all oligodendrocyte marker proteins).

https://doi.org/10.1371/journal.ppat.1014124.g008

We next considered if a trivial explanation of neurons dying in late-stage prion diseases might account for this selective downregulation of cell surface proteins in proximity to PrP, since most of them are associated with neurons. To address this question, we extracted from the global proteome dataset the relative quantities of proteins, which are commonly associated with specific brain cell types. Specifically, this incomplete list was comprised of ten neuronal, eight astrocytic, four oligodendrocytic, and three microglia markers (Fig 8B). Within the ten neuronal markers, we observed that steady-state level of NeuN and Tuj1, proteins that are frequently used to quantify homeostatic and mature neurons, did not change in late-stage-disease. Neither did other neuronal markers which identify neuronal stem cells (Nestin) or specific subtypes of neurons (choline-O-acetyltransferase for cholinergic neurons and tyrosine 3-monooxygenase for dopaminergic or noradrenergic neurons). Only neuronal markers associated with synapses, including Synaptophysin or Synapsin, were observed to be downregulated. These results indicate that the mere neuronal death that occurs late in the disease does not account for the reduction in the steady-state levels of the previously reported interactors and candidate interactors of the prion protein.

Ethics statement

All animal procedures were based on guidelines by the Canadian Council on Animal Care and were authorized by the University Health Network (UHN) Animal Care Committee (Animal Use Protocol 6840). All personnel involved in animal care or surgical procedures received specialized training for their respective tasks to ensure the humane treatment of the animals.

Antibodies

Primary antibodies: Human monoclonal anti-prion F(ab’)2 antibody (epitope: residues 96–104 in mouse PrP) [60], clone HuM-D13, 1:5,000, generously provided by the laboratory of Dr. Emil F. Pai (University of Toronto, ON, Canada). Rabbit monoclonal anti-prion IgG antibody (epitope: 221SQA223 and Y225), clone EP1802Y, 1:10,000 (catalog number ab52604, Abcam, Cambridge, United Kingdom). Mouse monoclonal anti-PrP antibody, clone 9A2 1:500 (Wageningen Bioveterinary Research, Lelystad, Netherlands).

Mouse monoclonal anti-beta-actin (Actb) IgG2b antibody, clone BA3R, 1:40,000 (catalog number MA5–15739-HRP, Thermo Fisher Scientific, Waltham, MA, USA). Rabbit polyclonal anti-Gfap antibody, 1:4,000 (catalog number Z0334, Dako, Carpinteria, CA, USA). Rabbit anti-Iba1 antibody, 1:1,500 (catalog number 019–19741, Fujifilm Wako Pure Chemical Corporation).

Secondary antibodies: Goat anti-human IgG F(ab’)2 horseradish peroxidase (HRP) conjugated antibody, 1:5,000 (catalog number 31414, Thermo Fisher Scientific). Goat polyclonal anti-rabbit HRP secondary IgG antibody, 1:5,000 (catalog number 31460, Thermo Fisher Scientific).

Cloning

The synthetic BvPrnp^V127 and MoPrnp^V126 sequences avoiding CpG motifs were purchased from a local gene synthesis service (Bio Basic Inc, Markham, ON, Canada). For MoPrnp^V126, a Q5 Site-Directed Mutagenesis Kit (catalog number E0554S, New England Biolabs, Ipswitch, MA, USA) was deployed to exchange the mouse PrP signal peptide for ER translocation with the corresponding signal peptide from human prolactin (PRL ss). To this end, the following primers where designed using the NEBaseChanger online tool (https://nebasechanger.neb.com) to exchange AA1–22 with the human PRL ss: Forward: 5’ – cttctggtgtccaatctgttactgtgccagtctgtggctcctaaaaagagaccaaagcctg– 3’, Reverse: 5’ – cagcagcagacttcccttccaggggctgcccttgatgttcatgggatccgatatctagatg – 3’. The final product was then transformed into NEB Stable Competent E. coli (catalog number C3040H, New England Biolabs) and sequence verified.

The self-complementary therapeutic BvPrnp and MoPrnp vectors were built using Gibson assembly. The backbone of the recombinant AAV transfer plasmid with regulatory elements, designated as pscAAV-CBh-Null-WPRE3-enSV40pA, was a gift from the Michael J Fox Foundation (catalog number 194245, Addgene, Watertown, MA, USA). The transfer plasmid was opened with StuI and AgeI restriction enzymes. In parallel, the desired inserts where amplified using a PCR reaction mix composed of Q5 High-Fidelity 2x Master Mix (catalog number M0492S, New England Biolabs), template DNA, and the following primers:

1. BvPrnp^V127ΔGPI:
   1. Forward: 5’– tcaggttggaccggctagcaccggtgccaccATGGCCAACCTCAGCTACTG – 3’,
   2. Reverse: 5’ – gtaatccagaggttgattaggTCATCTGCCCTCATAGTAGGCCTGGG – 3’.
2. PRLssMoPrnp^V126ΔGPI
   1. Forward: 5’– ttggaccggctagcagccaccATGAACATCAAGGGCAGCCCCT– 3’,
   2. Reverse: 5’ – gtaatccagaggttgattaggtcaGGATCTTCTCCCGTCGTAATAGGC– 3’.
3. PRLssMoPrnp¹²⁶
   1. Forward: 5’– ttggaccggctagcagccaCCATGAACATCAAGGGCAGCC – 3’,
   2. Reverse: 5’ – TAATCCAGAGGTTGATTAGGtcatcccacaatcaggaagatgaggaag – 3’.

All forward primers included the insertion of a canonical Kozak sequence while the reverse primers of all ΔGPI constructs removed the GPI signal sequence and added a translation termination nonsense codon. PCR was performed for 30 cycles in two phases: 15 cycles with an annealing temperature matched to the primer region complementary to the insert, followed by 15 cycles at 72°C once the full-length primer sequence had been incorporated into the PCR product. Finally, the gel-purified open vector backbone and insert sequences were assembled with the help of the HiFi DNA Assembly Master Mix (catalog number E2621L, New England Biolabs) during a 1-hour incubation at 50°C and then transformed into NEB Stable Competent E. coli. The scAAV CBh spEGFP construct was generated as previously described [39].

Cell culture and rAAV vector production

Murine CAD5 cells were a gift from Charles Weissmann. The cells were cultured in Opti-MEM reduced serum medium (catalog number 31985070, Thermo Fisher Scientific), supplemented with 10% FBS (catalog number 12483020, Thermo Fisher Scientific) and 1x Glutamax (catalog number 35050061, Thermo Fisher Scientific). They were infected with 0.1% RML-infected mouse brain homogenate and then passaged 3–6 times to chronically infect them with prions. These RML-infected CAD5 wild-type cells were than transfected with our desired constructs using a 1:2 ratio of DNA to lipofectamine 2000 (catalog # 11668027, Thermo Fisher Scientific). Briefly, 60mm dishes were plated with 1e6 cells, and then transfected at 80% confluency in serum-free Opti-MEM media. The transfection mixture was then replaced 24 hours later with 3 mL of complete Opti-MEM media containing FBS. Additional media was added to cells over the course of 120 hours, after which they were lysed and processed for western blotting.

HEK293T cells were maintained in DMEM (catalog number 119650–92, Thermo Fisher Scientific) supplemented with 50 U/mL Pen/Strep (catalog number 15140122, Thermo Fisher Scientific), NEAA (catalog number 11140–050, Thermo Fisher Scientific) and 10% FBS (catalog number 12483020, Thermo Fisher Scientific). Cultures were grown at 37°C, in an atmosphere of 5% CO₂ and 95% relative humidity. The media were replaced every 48–72 hours and the cells were passaged at 90% confluency in using 0.25% Trypsin-EDTA (catalog number 15050065, Thermo Fisher Scientific). The rAAV vector production and purification followed a previously established protocol [36]. In brief, HEK293T cells were cultured in a HYPERFlask (1,720 cm² surface area) with 560 mL of growth medium (catalog number CLS10031–4EA, MilliporeSigma). Upon reaching 70–80% confluence, the cells were transfected with three plasmids: an AAV packaging plasmid containing Rep and Cap genes, a helper plasmid encoding adenovirus genes E2A, E4, and VA, and the rAAV transfer plasmid. Four days post-transfection, the cell culture medium was supplemented with 3 mL of Triton-X 100 (catalog number ×100, Sigma-Aldrich), 250 µl of RNAse A (catalog number EN0531, Thermo Fisher Scientific), 56 µL of Pluronic F-68 (catalog number 24040–032, Thermo Fisher Scientific), and 56 µL of Turbonuclease (catalog number T4330, Sigma-Aldrich). The mixture was agitated at 150 RPM and 37 °C for 1 hour. Afterward, the lysate was collected and the HYPERFlask was washed with 140 mL of Dulbecco’s phosphate-buffered saline (PBS) (catalog number 14190–144, Thermo Fisher Scientific). The combined lysate and PBS wash were then centrifuged at 4,000 × g (RCF) for 30 minutes to pellet the cell debris. Finally, the supernatant was filtered through a 0.45 µm PES Autofil bottle top filter (catalog number 1143-RLS, Foxx Life Sciences).

rAAV vector purification method

The purification of rAAV vectors made use of 50 µm POROS CaptureSelect AAVX resins (catalog number A36652, Thermo Fisher Scientific). The resin was washed with 20 column volumes (CVs) of Low Salt Wash Buffer (50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 0.01% Pluronic F-68 surfactant, 1% Triton-X100). The affinity capture step was initiated by the slow (0.3 mL/min) loading of rAAV vectors harvested from HEK293 supernatants onto the AAVX resins. Once the rAAV vectors were loaded, the resin was washed with 20 CVs of Low Salt Wash Buffer, 20 CVs of High Salt Wash Buffer (300 mM NaCl, 50 mM Tris/HCl, pH 7.4, 0.01% Pluronic F-68 surfactant), and 20 CVs of Low Salt Wash Buffer without Triton-X100. Elution was induced by pH drop (0.2 M glycine and 0.01% Pluronic F-68, pH 2-2.5) and led to the collection of 4 mL fractions, which were rapidly pH neutralized with 420 µL of neutralization buffer (1 M Tris/HCl, pH 8, 0.1% Pluronic F-68). Prior to use, rAAV preparations were filtered through 0.22 µM PES membranes using a Stericup Quick Release-GP filtration system (catalog number S2GPU11RE, MilliporeSigma Canada Ltd., Oakville, ON, Canada).

qPCR

Quantitative PCRs were undertaken to determine virus titers using a well-developed protocol (Challis et. al. 2019). Briefly, serial dilutions of DNA standards were generated using the respective transgene plasmids which were linearized with ScaI (catalog number R3122S, New England Biolabs). Viral preparations were digested with 50 U/mL of DNase I (catalog number EN0521, Thermo Fisher Scientific) in 2 mM CaCl², 10 mM Tris-HCl, 10 mM MgCl². Next, viral capsids were digested using Proteinase K (catalog number EO0491, Thermo Fisher Scientific) in 1M NaCl, 34 mM N-lauroylsarcosine. The final qPCR measurement was conducted using a LightCylcler 480 II (catalog number 05015278001, Roche Diagnostics, Indianapolis, IN, USA) with the help of the SYBR green master mix (catalog number 4367659, Thermo Fisher Scientific) as well as the following primers which target the CBh promoter: Forward: GTTACTCCCACAGGTGAGC and reverse: CCAACCAACCATCCCTTAAAC.

Animals

Animal studies were necessary to assess the in vivo efficacy of a rAAV vector delivered gene therapy that capitalizes on the protective V127 variant. C57BL/6J mice were obtained from the Princess Margaret Cancer Research Centre (University Health Network, Toronto, ON, Canada). Homozygous BvPrnp I109 ki mice (Background: C57BL/6J) were described before [61]. A maximum of five mice per cage were kept at artificial 12-hour day and night cycles, drinking water ad libitum, and given 18% protein chow as solid food source. The mice received daily checks to assess their health and appearance. Their cages were changed weekly. To immobilize mice to be inoculated, transduced with rAAV vectors, or euthanized, anesthesia was induced with inhaled 5% isoflurane. Subsequently, when survival of the animals was intended, isoflurane levels were maintained at 2%. The anesthetic depth was assessed by performing a toe pinch. Respiration was observed during surgery to ensure a regular respiratory pattern. Prion-inoculated mice were closely monitored for signs of distress or pain, and humane endpoints were predefined (>20% weight loss, abnormal posture, lethargy, and impaired ambulation). Mice meeting these criteria were euthanized. Nesting scores were assigned based on whether the mice built no nest (score = 0), built a flat nest (score = 1), or built a three dimensional nest (score = 2). The nesting material consisted of pulped cotton fiber supplied in sheets that break easily into nestlets (catalog number NES3600, Ancare Corp., Bellmore, NY, USA).

Intracerebral prion inoculations

Intracerebral prion inoculations were undertaken with 4–6-week-old mice. The inoculant was generated by infecting a C57BL/6J mouse with RML-prions, then sacrificing the animal when terminally ill with prion disease and homogenizing the brain to a final concentration of 3% brain extract in PBS (v/v). Following deep anesthesia, mice were free-hand injected into the right parietal lobe 20 µL of inoculant at a depth of 3 mm with a 29-gauge SafetyGlide insulin syringe (catalog number 305930, Becton Dickinson Canada, Mississauga, ON, Canada).

Retro-orbital injections

Ahead of this procedure, the mice were anesthetized as described in the ‘Animals’ section. Additionally, all mice received for this procedure one eye drop of 0.5% proparacaine hydrochloride ophthalmic solution as a local anesthetic before injection. The injections were administered into the right orbital sinus using a 28-gauge Micro-Fine IV insulin syringe (catalog number 329420, Becton Dickinson Canada). All animals received a single retro-orbital injection of 1 × 10¹² viral genomes (vg), diluted in PBS to a final volume of 100 µL. Six-week old female C57BL/6J mice were injected with 9P31-spEGFP to assess the CNS spread after retroorbital injection.

Mouse tissue collection

Mice were deeply anesthetized with isoflurane and euthanized by six-minute transcardiac perfusion with PBS. Next, the brains were carefully extracted, their meninges removed, and their brains split into two sagittal hemispheres. The right hemisphere was designated for biochemical analyses and kept at -80°C until homogenization. The left hemisphere was post-fixed for subsequent pathological analyses by immersion in 10 mL of neutral-buffered formalin (catalog number HT501128-4L, Sigma-Aldrich). If mouse hemispheres were assigned for cryo-sectioning, the post-fixation was limited to two hours at 4°C.

Cryo-sectioning of mouse brains for direct fluorescence detection

Mouse brains for cryo-sectioning were dissected at three weeks post-injection. After fixation, brains were cryoprotected by immersion in 30% sucrose in PBS at 4°C for up to 36 hours causing the brain tissue to sink. Brains were then incubated in a 1:1 mixture of 30% sucrose and Tissue-Tek O.C.T. Compound (catalog 25608–930, VWR, Radnor, PA, USA). Next, brains were positioned in a cryomold (catalog number 70182, Electron Microscopy Sciences, Hatfield, PA, USA) and embedded in O.C.T for another one hour at 4°C degrees. Finally, the brains were frozen by partially immersing the cryomolds in a liquid nitrogen-chilled 2-methylbutane bath for 2–3 mins until the O.C.T completely froze. The frozen blocks were briefly air-dried on dry ice to remove residual 2-methylbutane and stored at −80°C.

Before cryo-sectioning, brains were allowed to warm up in the Cryostat chamber (model HM525 NX, Thermo Fisher Scientific) to a temperature of −21°C during a two-hour acclimatization period. Finally, 16 µm sagittal sections were cut and collected using Kawamoto’s film method on a piece of adhesive Cryofilm (catalog number C-FUF303, Section-Lab Co. Ltd., Yokohama, Kanagawa Prefecture, Japan) [62]. Before imaging, sections were washed for five minutes in PBS and mounted with mounting media containing DAPI (catalog number ab104139, Abcam).

Immunohistochemical analyses of mouse brain sections

To prepare samples for immunohistochemical analyses, mouse brains were immersed for one hour in formic acid, then rinsed with water, and placed in a Leica Pearl for processing. After sectioning, the slides were dried overnight at 37°C. Ahead of the immunohistochemical staining, the slides were baked for an hour at 60°C to help the tissues to adhere better to the slides. Next the slides were dehydrated through a xylene and ethanol series. The immunodetection made use of the Mouse on Mouse (M.O.M) Elite Immunodetection Kit (catalog number PK-2200, Vector Laboratories, Newark, CA, USA) using the manufacturers instructions. As part of these instructions, the tissue sections underwent antigen retrieval with citrate, pH 6.0, and heat using a commercial Antigen Decloaker formulation (catalog number CB910M, Biocare Medical, Pacheco, CA, USA). Brain slices were incubated with a monoclonal mouse PrP antibody, 1:500 dilution (Wageningen Bioveterinary Research) for 30 min at 22 °C or with antibodies directed against Gfap, 1:4,000 (Dako), or Iba1, 1:1,500 (Fujifilm), overnight at 4 °C and images developed using the NovaRed system (Vector Laboratories Inc., Newark, CA, USA). Finally, brain tissue sections were counter-stained with hematoxylin and eosin, rapidly dehydrated, and cover-slipped for microscopy analyses.

Microscopy

Brain samples were imaged under #1.5 glass coverslips (catalog number 48393–060, VWR) on Fisherbrand Superfrost Plus microscope slides (catalog number 22-037-246, Thermo Fisher Scientific) using a Zeiss AXIO Observer 7 inverted LED fluorescence microscope (Carl Zeiss Canada Ltd., North York, ON, Canada). To reconstruct full sagittal brain views, individual fields of view were acquired sequentially across each section and digitally stitched into a composite image using ZEN Blue microscopy software (Carl Zeiss Canada Ltd.).

Homogenization and protein extraction

Mouse brain tissue was homogenized in 100 mM Tris-HCl (pH 8.3) using a Minilys homogenizer (catalog number P000673-MLYS0-A, Bertin Technologies, Rockville, MD, USA) and Zirconia beads (catalog number 11079110zx, Biospec, Bartlesville, OK, USA) with three sets of 30 second bead-beating pulses. The resulting 20% homogenates were aliquoted, and a protease inhibitor cocktail (catalog number 78425, Thermo Fisher Scientific) was added to all samples that were not designated for subsequent Proteinase K digestion. To extract proteins, a detergent stock solution giving rise to final concentrations of 0.5% deoxycholic acid (DOC) (catalog number DCA333.50, BioShop, Burlington, ON, Canada) and 0.5% NP-40 (catalog number NON505.100, BioShop) was added to the homogenate, followed by gentle vortexing and 30 minute incubation on ice. Next, insoluble debris were removed during consecutive five-minute spins at 500 × g and ten-minute spins at 5,000 × g (RCF). Total protein concentrations of the samples were determined using the Pierce BCA Protein Assay Kit (catalog number 23225, Thermo Fisher Scientific) and subsequently adjusted.

gDNA extraction and restriction digest

The gDNAs of study mice were extracted from brain homogenates with the Monarch Spin gDNA Extraction Kit (catalog number T3010S, New England Biolabs). For the subsequent PCR reaction, 50 ng of gDNA was amplified with Q5 Hot Start High-Fidelity 2X Master Mix (catalog number M0494S, New England Biolabs) and forward and reverse primers (final concentration 500 nM) that map to gDNA sequences shared by the endogenous and heterologous BvPrnp ORFs. Equal volumes of PCR product were then subject to a restriction enzyme (RE) digest with 1 µL of EagI (catalog number R3505S, New England Biolabs), CsiI (catalog number FD2114, Thermo Fisher Scientific), or both REs for cutting the endogenous or the heterologous amplicons or both, respectively. The RE digestion products were finally separated on a 1% agarose gel containing SYBR safe DNA gel stain (catalog number S33102, Thermo Fisher Scientific). The fluorescent image of the gel was captured with the ChemiDoc XRS+ System (Bio-Rad Laboratories, Hercules, CA, USA). Forward: AAGAAGCGGCCAAAG and reverse: TAGTAGGCCTGGGACTC.

PNGase F digest

To remove N-linked glycans by digestion with PNGase F (catalog number P0704S, New England Biolabs, Ipswich. MA, USA), we made use of supplier provided buffers. Specifically, equal amounts of proteins were denatured using 10 × Denaturing Buffer at 95˚C for 10 minutes. Next, NP-40, GlycoBuffer, and PNGase F enzyme were added to the denatured samples. To inhibit undesired proteolytic activity, 3 mM phenylmethanesulfonyl fluoride (PMSF) (catalog number PMS123.5, BioShop) was added to the reaction tubes. Next, samples were incubated for 2 h at 37°C with gentle shaking. Finally, an equal volume of 2 × LDS sample buffer containing 5% β-mercaptoethanol (BME) was added directly to the samples to stop the reaction and prepare the samples for western blot analyses.

Proteinase K digest

To assess the relative PrP^Sc content of brain extracts, 200 µg each of BCA-adjusted total proteins were treated with Proteinase K (catalog number 25530049, Thermo Fisher Scientific) at a final concentration of 50 µg/mL and a ratio of total protein to Proteinase K of 1:20. Next, samples were incubated for 45 min at 37°C under gentle shaking. The digestion was terminated by adding PMSF to a final concentration of 2 mM, followed by the addition of 2% Sarkosyl (catalog number SLS002.100, BioShop). Next, the digests were ultracentrifuged for 1 hour at 48,000 RPM and 4°C using an Optima TLX Ultracentrifuge (Beckman Coulter, Brea, CA, USA). Finally, the pelleted insoluble PrP^Sc was resuspended in 1x Bolt LDS sample buffer (catalog number B0007, Thermo Fisher Scientific), heat-denatured at 95°C for 10 minutes, then analyzed by western blotting. For cell culture experiments, the Proteinase K digestions were conducted on 500 µg of total lysate, with a ratio of 1:50 at a concentration of 50 µg/mL.

Western blotting

To detect total PrP within brain extract samples, proteins were denatured in 1x Bolt LDS sample buffer at a final concentration of 2 µg/µL. Next, samples were heated at 95 °C for 10 min, followed by briefly cooling on ice, immediately before gel loading. Proteins were separated by SDS-PAGE analysis on 10% Bolt Bis-Tris gels (catalog number NW00105BOX, Thermo Fisher Scientific). For PNGase F digested samples, 12% NuPage Bis-Tris gels (NP0342BOX, Thermo Fisher Scientific) were used. After the gel electrophoresis, proteins were transferred to 0.45 µm PVDF membranes (catalog number IPVH00010, Sigma-Aldrich, St. Louis, MO, USA). Following the blocking of the membrane in 5% skimmed milk (catalog number SKI400, BioShop Canada Inc) for 1 h at room temperature, membranes were incubated in primary antibodies overnight at 4°C with gentle rocking. Next, membranes were washed three times in 1x Tris-buffered saline containing 0.1% Tween-20 (TBST) (catalog number TWN508, BioShop) and incubated with the corresponding HRP-conjugated secondary antibodies for 1 h at room temperature. After washing the membranes trice again in 1x TBST, they were incubated for 1 min with Western Lightning Pro enhanced chemiluminescent (ECL) reagent (catalog number NEL120001EA, Revvity Health Sciences Inc., Mississauga, ON, Canada). Finally, membranes were exposed to autoradiography film (catalog number CLMS810, MedStore, Toronto, ON, Canada) and developed using a film developer.

Sample preparation for mass spectrometry

The protein concentration of brain extracts was adjusted to 4 µg/µL. To denature all proteins in the sample, including the PrP^Sc, 20 µg (in 5 µL) of total proteins were transferred to Protein LoBind (PLB, catalog number: PRE050LR-N, Diamed, Mississauga, ON, Canada) tubes and diluted with 9M urea (catalog number UR001.1, Bioshop) in a 1:2 ratio (v/v), achieving a 6 M urea concentration in the sample, followed by gentle vortexing and 30-minute incubation at room temperature to ensure complete denaturation. The pH of the samples was checked and, if below pH 8, increased to this pH using 1 M Triethylammonium Bicarbonate (TEAB) (catalog number 1861436, Thermo Fisher Scientific). Next the denatured proteins were reduced with 6.5 mM TCEP (catalog number TCE101, Bioshop) during a 30-minute incubation at 37°C, then alkylated for another 30 minutes in the dark at room temperature in the presence of 15 mM iodoacetamide (catalog number 1861445, Thermo Fisher Scientific).

Next, the proteins were subjected to solvent precipitation to maximize recovery for proteomic analyses [63]. To this end, samples were threefold diluted to prepare them for the precipitation. In parallel, silica beads of 9–13 μm mean particle diameter (catalog number 440345), which had been pre-cleaned and resuspended in acetonitrile (ACN) at a concentration of 70 µg/µL, were added at a 10:1 bead-to-protein ratio and the tubes were gently vortexed. Subsequently, 100% ACN was added to each sample at a 1:4 (v/v) ratio to achieve an 80% ACN concentration without pipette mixing, followed by gentle vortexing for 10 seconds. The precipitates were centrifuged for 5 minutes at 16,000 × g (RCF) and room temperature, before supernatants were carefully removed, using the tube hinge as a guide to avoid disturbing the pellet. To further remove non-protein contaminants, the pellets were washed three times with 80% ethanol, using at least twice the total precipitation volume for each wash, followed by 2-minute centrifugation at 16,000 × g (RCF) and room temperature.

After the final wash, the remaining supernatant was carefully removed, and the protein pellet resuspended by gently vortexing in 100 mM ABC with a 1:50 trypsin-to-protein ratio. To fully disrupt the pellet, the tubes were placed in a sonication bath for 2 minutes. To ensure complete digestion, the samples were then incubated at 37°C for 18 hours at 800 RPM in a thermomixer (Thermomixer Comfort). Following digestion, 1 µL of 10% formic acid (FA) (catalog number A117-50, Thermo Fisher Scientific) was added to each sample to adjust the pH to approximately pH 3. The beads and any insoluble debris were precipitated by 10-minute centrifugation at 16,000 × g (RCF) and room temperature, before transferring the peptide-enriched supernatants to a new tube. To further recover additional peptides from the silica beads, the centrifugation pellets were washed once with 2% ACN in 0.1% formic acid, and the resultant wash supernatants combined with the initial peptide-containing supernatants. The final peptide concentration was adjusted to 0.25 to 0.5 µg/µL in 1% acetonitrile with 0.1% formic acid.

Mass spectrometry data acquisition

Sample injected were three biological replicates from three treatment cohorts, with each sample composed of tryptic digests of 200 ng of total brain extract proteins. All data were acquired using a Vanquish Neo UHPLC system (Thermo Fisher Scientific) coupled to an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific) through an EASY-Spray source (Thermo Fisher Scientific). Trap-and-Elute injection was based on PepMap Neo Trap Cartridge (catalog number 174500, Thermo Fisher Scientific) to filter impurities from peptide mapping samples. During peptide separation, a 15 cm EASY-Spray HPLC Column (catalog number ES900, Thermo Fisher Scientific) was maintained at 50°C. The mobile phase A consisted of 0.1% formic acid in water, while mobile phase B consisted of 0.1% formic acid and 80% acetonitrile in water. The gradient was as follows: the mobile phase was initially held at 2% B, increased from 2% to 4% in 0.5 min at a flow rate of 0.7 µL/min, then further increased to 5% B from 0.5 to 0.6 min, increased to 7% B from 0.6 to 1 min, ramped to 22.5% B from 1 to 18 min, then increased to 35% B from 18 to 25.5 min, and further increased to 50% B from 25.5 to 27 min. The flow rate was set to 0.5 µL/min. The column was then washed with 99% B from 27 to 30 min at a flow rate of 0.7 µL/min. Data were acquired in data independent acquisition (DIA) mode with a normalized collision energy of 25% and a default charge state of +2. MS1 spectra were acquired in the embedded orbitrap mass analyzer at a resolving power of 240,000 every 0.6 s. MS2 spectra were acquired in the embedded astral analyzer, with precursor isolation windows of 2 Th spanning the range of 380 − 980 Th. The custom normalized AGC target was set as 500% for both MS1 and DIA. The MS1 mass range was the same as the MS/ MS precursor range. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD076487.

Processing of global proteomic data set

Protein identification and peptide peak integration made use of Proteome Discoverer (PD) software (Version 3.2, Thermo Fisher Scientific) with CHIMERYS. These analyses interrogated the Mus musculus database (TaxID = 10090, release = 407, supplemented with the entry for the major prion protein from bank vole (Accession number Q8VHV5). The database search was restricted to tryptic peptides, allowing up to two missed cleavages per peptide. The maximum fragment mass tolerance was set to 10 ppm. Carbamidomethylation at cysteine residues was specified as a fixed modification, while oxidation at methionine residues was included as a variable modification. The sample-to-sample normalization in PD was based on the ‘Total Peptide Amount’ computed for each sample. Prion peptide comparisons were conducted using Skyline (Version 24.1) [64], with tubulin alpha and tubulin beta used for sample normalization. The assignment of peptide transitions to BvPrP were manually reviewed to ensure accurate peak integration. Both the Proteome Discoverer and Skyline data were exported to Excel for further conditional formatting.

Cluster and KEGG analyses

Hierarchical clustering was undertaken with Cluster (Version 3.0, using Clustering Library Version 1.59) [65,66]. Prior to clustering, data columns were generated that inform about steady-state protein level ratios, computed by dividing cumulative MS1 ion intensities assigned to a given protein in a specific brain extract to the average MS1 ion intensities for the same protein in brain extracts from naïve mice. Subsequently, these steady-state protein level ratio data for all proteins, i.e., the rows within the Excel dataset, were clustered using centroid linkage clustering methods based on a ‘City-block distance’ metric and the relationship across the brain samples, i.e., the columns within the Excel dataset were inferred by hierarchical clustering using a Spearman Rank Correlation metric. Finally, hierarchical cluster analysis results were visualized in Java TreeView (Version 1.21) open source software [65].

KEGG pathway analyses were undertaken with the DAVID Bioinformatics suite (release DAVID 2021, Version 2023q4) of functional annotation tools made available by the National Institutes of Health [67,68]. Briefly, to initiate the analyses, lists of official gene symbols of genes whose expression gave rise to top- or bottom-ranked proteins in the global proteomic dataset were uploaded into the DAVID Analysis Wizard along with the selection of Mus musculus as the biological sample source. Next, a KEGG_PATHWAY analysis was undertaken from within the Annotation Summary Results page.

Assignment of proteins increased in their steady-state levels in prion disease to mouse brain cell types

The cell type assignment made use of previously reported mouse brain cell type annotations, which were originally obtained through single-cell transcriptomic analyses [43]. To interrogate these data with genes coding for proteins whose steady-state levels were most increased in RML-infected mice, which had reached the humane prion disease end-stage, we submitted the shortlist of genes to the online Transcriptomics Explorer algorithm (https://portal.brain-map.org/atlases-and-data/rnaseq). Separately, the ggplot package within RStudio was used to depict the non-neuronal portion of the brain cell type dendrogram, which this analysis revealed to comprise the genes whose levels were most profoundly increased in RML prion disease.

Statistical analyses

Western blot signal intensities from three biological replicates were quantified by densitometric analysis using ImageJ software and corresponding statistics were performed using Microsoft Excel. First, cohorts were subject to an F-test to assess the variance of the two compared groups. A Welch’s t-test or pooled t-test was subsequently chosen to assess the statistical significance of the differences between the averages of cohorts that were compared. Results were considered significant if p < 0.05 or denoted as non-significant with “ns” if p > 0.05. Asterisks were used to signify varying levels of significance: p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***).

Survival, weight, and nesting score charts were assembled in R using RStudio (Version 4.4.3). For all analyses, Excel raw data were uploaded into RStudio using the ‘readxl’ package. For the Kaplan-Meier analyses, the Greenwood formula computed a variance estimate, which was used by the Kaplan-Meier estimator (kmfit) to compute the 95% confidence interval. Plots were then generated with basic plotting functions embedded in ‘ggplot2’ and ‘ggfortify’ packages.

For generating the weight and nesting score charts, the geom_smooth() function within ggplot2 was used to depict the linear trend and the confidence interval using the Locally Estimated Scatterplot Smoothing ‘loess’ method.

We assumed a hypergeometric distribution when determining if it is statistically significant that 27 out of 30 cell surface proteins, which had been shown to reside in proximity to PrP in mouse brains [44], were downregulated ≥ 33% (out of 455 other proteins sharing this characteristic) in a global proteome dataset of 4874 proteins.