Abstract

Rapid neurodegeneration distinguishes prion disease from other neurodegenerative disorders. Notably, normal prion protein (PrPC) is essential for prion-induced rapid neurodegeneration, but the underlying mechanism remains unknown. Here, we show that the unstructured N-terminal region of PrPC induces rapid and lethal neurodegeneration in mice, accompanied by the hallmark of prion disease, spongiosis. The neurotoxic N-terminal PrP is soluble, associates peripherally with lipid membranes, and induces neurotoxicity only when a critical threshold is exceeded. Both the N-terminally localized KKRPKP sequence and octarepeats contribute to neurotoxicity, with KKRPKP being essential. Without it, the N-terminal PrP is innocuous but exacerbates either neurodegeneration caused by N-terminal PrP or neurodegeneration in prion disease induced by intracerebral prion inoculation in mice. Our findings establish that soluble N-terminal PrP causes rapid neurodegeneration in prion disease and is a target for intervention.

INTRODUCTION

Prion diseases, also known as transmissible spongiform encephalopathies, are rapidly progressive and fatal neurodegenerative disorders well-known for their unconventional transmission agent, prions (1). PrPSc, the self-propagating prion conformer of the host-encoded normal prion protein (PrPC), is sufficient to transmit the disease through the seeded conversion of PrPC into PrPSc (2–5). This prion concept has now been greatly expanded, with substantial evidence indicating that the seeded aggregation of pathogenic proteins is a common feature among neurodegenerative disorders associated with amyloid deposition (6–8). One less understood aspect of prion diseases is their unusually rapid neurodegeneration (9, 10), which results in lethality in transmission studies and may explain why prion disease was long considered the only transmissible neurodegenerative disorder.

Seminal studies have demonstrated that the rapid neurodegeneration in prion disease is not directly caused by the aggregated PrPSc; rather, the neuronal expression of PrPC is essential (11, 12), suggesting that PrPSc induces alterations in PrPC, and these altered proteins cause neurotoxicity. However, the mechanism by which the innocuous PrPC becomes highly neurotoxic remains unclear. In vitro studies using organotypic brain slice cultures, electrophysiology, and primary neuronal cultures indicate that the flexible N-terminal domain of PrP [PrP(N)] is detrimental to neurons (13, 14). But whether PrP(N) is similarly neurotoxic in vivo and its relationship to prion disease remain entirely unknown, primarily due to the unstructured nature of PrP(N), which complicates its expression in vivo. Our recent discovery that fusing a nanobody (Nb) significantly enhances the in vivo production of PrP(N) (15) enables us to directly study its biological effects in animals.

RESULTS

PrP(N) causes rapid and lethal neurodegeneration

To determine the in vivo effects of PrP(N), we fused PrP1-110, the natural proteolytic cleavage product known as N1 (16), with a Nb against green fluorescent protein (17). The same Nb directly fused with PrP signal peptide and the full-length PrP (PrP1-254) were used as controls (Fig. 1A). These plasmids were packaged into recombinant adeno-associated virus (rAAV) and delivered to newborn wild-type (WT) mice via intracerebroventricular (ICV) injection. Immunoblot analysis revealed that the expression of PrP1-110-Nb was predominantly localized within the central nervous system (fig. S1A). The levels of PrP1-110-Nb were modest compared to endogenous PrP and significantly lower than those of the Nb control (Fig. 1B and fig. S1, B and C). While all control mice remained healthy, the expression of PrP1-110-Nb resulted in rapid lethal neurodegeneration in WT mice (Fig. 1C). In the terminal stage, these mice developed rapid weight loss, rough hair, kyphosis, lethargy, and severely impaired mobility (fig. S1, D and E), with death occurring 2 to 8 days after the onset of these symptoms. Histopathological analyses revealed that the diseased brains exhibited spongiosis, astrogliosis, and microgliosis (Fig. 1D and fig. S1F), which are the pathological hallmarks of prion disease. When PrP1-110-Nb was expressed in PrP-null mice, the expression pattern was found to be similar to that observed in WT mice (fig. S2A). The levels of PrP1-110-Nb were higher than those in WT mice, although still modest compared to the endogenous PrP in WT mice or the Nb control (Fig. 1E and fig. S2, B and C). Consistent with the findings in WT mice, PrP1-110-Nb induced lethal neurodegeneration in PrP-null mice. The clinical manifestations were similar to those of WT mice, although the progression appeared to be more rapid, occurring within ~2 to 7 days (Fig. 1F and fig. S2D). Spongiosis, astrogliosis, and microgliosis were also observed in the diseased brains (fig. S2E). These findings demonstrate that the PrP1-110-Nb–induced neurotoxicity is independent of endogenous PrP.

Fig. 1. The N-terminal region of PrP is sufficient to cause rapid neurodegeneration with spongiosis and gliosis.

(A) Schematic diagram illustrating PrP, PrP1-110-Nb, and Nb control. CC1, positively charged cluster 1; CC2, positively charged cluster 2. (B) Immunoblot analysis of endogenous PrP, rAAV-mediated expression of PrP1-254, and PrP1-110-Nb in WT mice using the 6D11 anti-PrP antibody. Square bracket indicates endogenous PrP and arrowhead points to the position of PrP1-110-Nb. (C) Survival curves of WT mice expressing PrP1-110-Nb, PrP1-254, or Nb control. ***P = 0.0006; ****P < 0.0001. (D) Representative images of histopathological analyses with hematoxylin and eosin (H&E) staining, and immunohistochemical staining using anti-GFAP and anti-Iba1 antibodies as indicated (scale bars, 50 μm). (E) Immunoblot analysis of rAAV-mediated expression of PrP1-254 and PrP1-110-Nb in Prnp knockout (KO) mice and endogenous PrP in WT mice using the 6D11 anti-PrP antibody. One-way analysis of variance (ANOVA) revealed no statistically significant differences (n = 3, except for PrP1-110-Nb where n = 4. P = 0.984 for the comparison between WT and PrP1-254, P = 0.225 between WT and PrP1-110-Nb, and P = 0.286 between PrP1-254 and PrP1-110-Nb). (F) Survival curves of Prnp KO mice expressing PrP1-110-Nb, Nb control, or PrP1-254. ****P < 0.0001. (G) Survival curves of the LSL-PrP1-110-Nb+/−/Prnp-iCre+/− mice (KI-PrP1-110-Nb+/−) and the control LSL-Nb+/−/Prnp-iCre+/− mice (KI-Nb+/−). ****P < 0.0001. (H) Survival curve of induced LSL-PrP1-110-Nb+/−/Rosa26-CreERT2+/− mice (n = 8). (I) The expression level of PrP1-110-Nb in tamoxifen-induced LSL-PrP1-110-Nb+/−/Rosa26-CreERT2+/− mice was compared to that in WT mice received ICV injection of rAAV expressing PrP1-110-Nb (N1-Nb), LSL-PrP1-110-Nb+/+ mice received ICV injection of a rAAV expressing Cre under the control of the CAG promoter (Cre), and LSL-PrP1-110-Nb+/−/Prnp-iCre+/− mice. PrP1-110-Nb was detected with immunoblot analysis with an anti-VHH antibody. Uninduced LSL-PrP1-110-Nb+/−/Rosa26-CreERT2+/− mice were used as a control.

Because ICV rAAV injection does not replicate the endogenous PrP expression pattern in vivo, we used CRISPR-Cas9 technology to generate three mouse lines. Two of these lines contain lox-stop-lox-PrP1-110-Nb and the control lox-stop-lox-Nb inserted at the H11 site (designated as LSL-PrP1-110-Nb and LSL-Nb mice, respectively). The third line has the codon-improved iCre recombinase replacing the open reading frame of PrP (designated as Prnp-iCre mice). Crossing LSL-PrP1-110-Nb and Prnp-iCre lines resulted in PrP1-110-Nb+/−/iCre+/−/PrP+/− mice, which expressed higher levels of PrP1-110-Nb+/− compared to those receiving the ICV rAAV injection, and developed fatal neurodegeneration with a mean survival time of 15.3 ± 0.2 days (Fig. 1G and fig. S3, A and B). The progression of the disease was rapid, with a complete loss of the ability to stand or move occurring within 1 to 2 days from onset. In contrast, no abnormalities were observed with the control Nb+/−/iCre+/−/PrP+/− mice.

Due to the high neurotoxicity of PrP1-110-Nb, all mice succumbed to fatal neurodegeneration at a relatively young age. To rule out the possibility that neurotoxicity was specifically associated with young mice, we crossed LSL-PrP1-110-Nb mice with Rosa26-CreERT2 mice and induced transgene expression at 8 weeks of age. The level of transgene expression in these mice was significantly lower, yet still sufficient to induce 50% lethality (Fig. 1, H and I, and fig. S3C). The disease progressed rapidly, and diseased mice reached terminal stage with severe motor deficits within just 1 to 2 days. These results confirm that PrP1-110-Nb is sufficient to cause rapid and lethal neurodegeneration.

PrP(N) needs to exceed a toxic threshold to cause rapid neurodegeneration

The differences in PrP1-110-Nb–induced neurodegeneration in WT and PrP-null mice (comparing Fig. 1, C and F) may be attributed to variations in transgene dosage and/or the neuroprotective effect of endogenous PrP. To test these possibilities, we prepared a second batch of rAAV that expressed higher levels of PrP1-110-Nb, which effectively abolished the differences in survival time (fig. S4, A and B), although a statistically insignificant prolongation of survival in WT mice was still evident.

Using the second batch of rAAV, a twofold dilution resulted in 50% lethality in WT mice and 100% lethality in PrP-null mice. A fivefold dilution led to 50% lethality in PrP-null mice, while WT mice remained asymptomatic. A 10-fold dilution completely abolished toxicity (Fig. 2A). The difference in PrP1-110-Nb levels appeared to be small, particularly between no dilution (100% lethality) and twofold dilution (50% lethality) in WT mice (Fig. 2B). The unaffected mice in the groups of mice with 50% lethality remained healthy and showed no abnormalities. The PrP1-110-Nb level was only slightly elevated in mice that developed fatal neurodegeneration (Fig. 2, C and D). Spongiosis and gliosis were observed exclusively in the mice that developed neurodegeneration (fig. S4C).

Fig. 2. Neurotoxicity of PrP1-110-Nb occurs only when its level exceeds a toxic threshold.

(A) Survival curves of Prnp KO and WT mice received ICV injection of rAAV diluted by a factor of 2 (left), 5 (middle), and 10 (right). **P = 0.01; ***P = 0.0008. (B) Immunoblot analysis of PrP1-110-Nb in indicated mice using an anti-VHH antibody. Statistical differences were determined by one-way ANOVA. The sample size was n = 3, except for the Prnp KO mice expressing PrP1-110-Nb at a dilution factor of 10, where n = 5. The P values were 0.0098 for the comparison between 1:2 and 1:5 dilutions, and 0.0015 for the comparison between 1:2 and 1:10 dilutions in Prnp KO mice. In WT mice, the P value was 0.0254 for the comparison between 1 and 1:5 dilution and 0.0149 for the comparison between 1 and 1:10 dilution. (C and D) Levels of PrP1-110-Nb in WT mice received ICV injection of twofold diluted rAAV (50% lethality), and Prnp KO mice received ICV injection of fivefold diluted rAAV (50% lethality) were determined by slot blot analysis using an anti-VHH antibody. “Dead” indicates mice that developed fatal neurodegeneration, while “Alive” indicates mice that did not develop any abnormalities. Each column was loaded with one mouse brain homogenate at decreased amounts as indicated. Statistical analysis was performed using unpaired Student’s t test [for WT mice, n = 2 (Dead) and n = 8 (Alive), P = 0.0111; for Prnp KO mice, n = 4 (Dead) and n = 8 (Alive), P = 0.0275].

Collectively, these results suggest that, while endogenous PrPC exerts a mild neuroprotective effect, an increased dosage of PrP1-110-Nb effectively overrides this protection. When the level of PrP1-110-Nb exceeds the toxic threshold, it triggers rapid neurodegeneration. This aspect of PrP(N)-induced neurotoxicity recapitulates the phenomena observed in prion diseases (18, 19).

PrP1-90 is sufficient to cause rapid neurotoxicity

To refine the region responsible for rapid neurotoxicity, we generated PrP1-90-Nb rAAV, the naturally occurring N2 fragment of PrP that lacks the CC2 region (Fig. 3A) (16). The transgene expression levels following ICV injection were comparable to or lower than those of PrP1-110-Nb (fig. S5A), yet this was sufficient to elicit rapid neurodegeneration. The presence of endogenous PrP exhibited a modest neuroprotective effect, extending survival time from 16.3 ± 0.4 days to 33.7 ± 8.4 days (Fig. 3A). The pathological changes were similar in WT and PrP-null mice, with both showing spongiosis and gliosis (Fig. 3B). We also created mice with lox-stop-lox-PrP1-90-Nb inserted at the H11 site (designated as LSL-PrP1-90-Nb mice). Crossing these mice with Emx1-cre mice, which express Cre recombinase in neurons, resulted in rapid neurodegeneration (Fig. 3, C and D, and fig. S5B). These results revealed that PrP1-90, recognized as the N2 fragment of PrP, is sufficient to cause rapid neurodegeneration.

Fig. 3. The neurotoxic PrP(N) is soluble and peripherally associated with lipid membranes.

(A) Survival curves of WT and Prnp KO mice received ICV injection of rAAV expressing PrP1-90-Nb. ****P < 0.0001. (B) Representative images of histopathological analyses of mice expressing rAAV-mediated PrP1-90-Nb expression (scale bars, 50 μm). (C) Survival curve of EMX1-cre+/−/LSL-PrP1-90-Nb+/− mice (n = 20). (D) Representative images of histopathological analyses of mice described in (C) (scale bars, 50 μm). (E) WT and Prnp KO mouse brain lysates were separated into soluble (S) and insoluble (I) fractions and the indicated proteins were detected by immunoblot analysis using an anti-VHH antibody. (F) Percentage of soluble proteins. Statistical analysis was performed with one-way ANOVA. The sample size was n = 3, except for endogenous PrP, where n = 5. The P values observed were 0.0161 for the comparison between PrP1-110-Nb and Nb, 0.0378 between PrP1-110-Nb and PrP1-90-Nb in Prnp KO mice, and 0.0210 between PrP1-90-Nb and endogenous PrP. (G) Top: Schematic diagram of PrP1-135-Nb. Bottom left: The solubility of PrP1-135-Nb was determined by immunoblot analysis using an anti-VHH antibody. Bottom right: Survival curves of WT and Prnp KO mice received ICV injection of rAAV expressing PrP1-135-Nb. (H) The migration of PrP1-110-Nb and PrP1-90-Nb in the gradient was detected by immunoblot analysis using an anti-VHH antibody. Arrowheads indicate full-length PrP1-90-Nb. (I) The migration of endogenous PrP and PrP1-110-Nb in the gradient, with or without the extraction with 0.5 M NaHCO3, pH 11 as indicated, was detected by immunoblot analysis using the 6D11 anti-PrP antibody. In (H) and (I), the numbers on the top indicated 12 fractions collected from the top to the bottom of the density gradient.

The neurotoxic PrP(N) is soluble, extracellular, and peripherally associated with membranes

Prion disease is characterized by the conversion of soluble, proteinase K (PK)–sensitive PrPC to aggregated, PK-resistant PrPSc (1). However, the physical state of neurotoxic PrP remains unclear, which is crucial for elucidating the neurotoxic mechanism. We analyzed the solubility of neurotoxic PrP(N) and found that, in the presence of mild detergents, most of Nb-fused PrP(N) was as soluble as endogenous PrP and the soluble Nb control (Fig. 3, E and F), indicating that soluble PrP(N) induced neurotoxicity. Consistent with this conclusion, PrP1-135-Nb, which includes the hydrophobic domain of PrP, significantly increased insolubility (Fig. 3G and fig. S6A). Although the level of PrP1-135-Nb was higher than that of PrP1-110-Nb (fig. S6B), its ability to induce rapid neurodegeneration was reduced (Fig. 3G).

Because rapid neurodegeneration and PK-resistant PrPSc are both characteristics of prion disease, we analyzed PK-resistance of PrP(N). A ~14-kDa PK-resistant band was detected in the Nb control (fig. S7A), consistent with the fact that Nb is a tightly folded protein domain (20). However, an additional band of ~15 kDa was observed in mouse brains expressing PrP1-110-Nb. This PK-resistant band was not present in brains expressing PrP1-90-Nb and was not detected by the 8B4 antibody, which recognizes an epitope at the very N terminus of PrP (fig. S7B). Therefore, it is likely formed by the 20 extra amino acids in PrP1-110-Nb (residues 90 to 110). Given that the PK-sensitive PrP1-90-Nb is highly neurotoxic, we conclude that this PK-resistance is irrelevant to its neurotoxicity. There is no evidence of conversion of endogenous PrPC to PK-resistant PrPSc in these mice.

Despite the assistance of the Nb fusion, a small portion of PrP1-110-Nb or PrP1-90-Nb may still fail to enter the secretory pathway, and substantial accumulation of full-length PrP in the cytosol is known to be neurotoxic (21–23). To rule out the possibility that the rapid neurotoxicity observed here was due to this portion of PrP1-110-Nb, we created rAAV that expresses PrP23-110-Nb, which lacks the signal peptide (fig. S8A). Although it was expressed at levels comparable to those of PrP1-110-Nb, PrP23-110-Nb failed to cause neurodegeneration (fig. S8, B to D). Together, these results led us to conclude that rapid neurodegeneration is caused by soluble, PK-sensitive PrP(N) in the extracellular space, which is consistent with the localization of glycosylphosphatidylinositol (GPI)–anchored PrPC on the outer leaflet of the plasma membrane.

To address the question of how extracellular PrP(N) can cause rapid neurodegeneration, we conducted a density gradient analysis in which membrane-associated proteins migrated to the top (24) to determine whether the neurotoxic PrP(N) is associated with the membrane. As expected, the GPI-anchored endogenous PrP migrated to the top, while most of the Nb control remained at the bottom (fig. S9A). A significant portion of PrP1-110-Nb or PrP1-90-Nb migrated to the top (Fig. 3H and fig. S9B), indicating that they were associated with membranes. Some degradation was observed with the portion of PrP1-90-Nb that remained at the bottom, consistent with the notion that unbound soluble PrP1-90-Nb is more susceptible to degradation. When the membrane was subjected to extraction with alkaline sodium bicarbonate solution, almost all PrP(N) was removed from the membranes and remained at the bottom of the gradient, whereas GPI-anchored endogenous PrP still migrated to the top (Fig. 3I and fig. S9C). These results show that neurotoxic PrP(N) is peripherally associated with cell membranes.

CC1 is essential, while OR modulates the rapid neurotoxicity induced by PrP(N)

The neurotoxic PrP(N) contains two distinct structural features, the positively charged 23KKRPKP28 sequence (CC1 region) and the octarepeat (OR) region. To determine the role of CC1, we generated mutants that either deleted CC1 (N1ΔCC1-Nb) or substituted positively charged residues in CC1 with methionine (N1MCC1-Nb). These mutations completely abolished the ability of PrP1-110-Nb to induce rapid neurodegeneration (Fig. 4, A and B). The protein levels of N1ΔCC1-Nb and N1MCC1-Nb were comparable to or exceeded those of PrP1-110-Nb (fig. S10), indicating the absence of neurotoxicity is not due to a lower dosage of N1ΔCC1-Nb or N1MCC1-Nb. Consistent with this conclusion, the expression of N2ΔCC1-Nb via ICV rAAV injection in newborn mice also failed to cause any abnormalities (Fig. 4C). These findings reveal that CC1 is essential for the rapid neurodegeneration caused by PrP(N).

Fig. 4. CC1 is essential, while OR contributes the neurotoxicity caused by PrP(N).

(A and B) Survival curves of WT and Prnp KO mice received ICV injection of rAAV expressing PrP1-110-Nb with either deleted (N1ΔCC1-Nb) or mutated (N1ΜCC1-Nb) CC1 as indicated. (C) Survival curve of Prnp KO mice received ICV injection of rAAV expressing N2ΔCC1-Nb (3 × 1010 gc). (D) Survival curves of Prnp KO mice received ICV injection of PrP1-90-Nb and N2ΔOR-Nb at indicated dosages. (E) Left: Schematic diagram illustrating N1ΔCC1-Nb, PrP1-90-Nb, and Nb control. Right: Survival curves of Prnp KO mice received ICV injection of rAAVs expressing PrP1-90-Nb (3 × 109 gc) plus Nb control (3 × 1010 gc) or PrP1-90-Nb (3 × 109 gc) plus N1ΔCC1-Nb (3 × 1010 gc). (F) Survival curves of Prnp KO mice received ICV injection of rAAVs expressing PrP1-90-Nb (6 × 109 gc) plus Nb control (3 × 1010 gc) or PrP1-90-Nb (6 × 109 gc) plus N2ΔCC1-Nb (3 × 1010 gc).

Given that the OR, the other structural feature of the neurotoxic PrP(N), has been implicated in the prion toxicity (13, 25), we hypothesized that the OR also contributes to the neurotoxicity caused by PrP(N). ICV injection of rAAV expressing N2ΔOR-Nb did not result in a complete loss of toxicity; rather, it caused significantly milder neurotoxicity compared to similar levels of PrP1-90-Nb (Fig. 4D and fig. S11A). Considering the variability among different rAAV preparations, we produced three additional batches of rAAV-N2ΔOR-Nb and all of them resulted in milder lethality (ranging from 12.5 to 62.5%) compared to similar levels of PrP1-90-Nb (fig. S11, B to F). Spongiosis was observed only in N2ΔOR-Nb mice that developed neurodegeneration, while the brains of unaffected N2ΔOR-Nb expressing mice appeared normal (fig. S11G). Given that the levels of N2ΔOR-Nb were comparable to those of PrP1-90-Nb that produced close to 100% lethality (fig. S11, A, C, and F), the milder neurodegeneration associated with N2ΔOR-Nb suggests that, while not essential, the OR region does contribute to the neurotoxicity caused by PrP(N).

Innocuous ΔCC1 PrP(N) accelerates neurodegeneration caused by PrP(N)

Because the N1ΔCC1-Nb, which removes only six amino acids at the very N terminus, completely eliminates neurotoxicity, we investigated whether co-expressing N1ΔCC1-Nb could interfere with neurotoxicity induced by PrP(N). Unexpectedly, instead of interference, the coexpression of N1ΔCC1-Nb with PrP1-90-Nb significantly accelerated neurodegeneration, with a more pronounced effect observed at lower levels of the neurotoxic PrP1-90-Nb (Fig. 4E and fig. S12, A and B). Similarly, coexpression of the innocuous N2ΔCC1-Nb with PrP1-90-Nb also resulted in enhanced neurodegeneration (Fig. 4F). Collectively, these results reveal that the benign PrP(N)ΔCC1 exacerbates the neurotoxicity caused by PrP(N).

PrP(N) causes rapid neurodegeneration in prion disease

The relevance of PrP(N)-induced neurotoxicity to prion disease is a critical question, and we took two approaches to address it. First, we conducted mass spectrometry (MS) analysis to compare the proteomic changes associated with PrP1-110-Nb–induced neurotoxicity to those observed in prion diseases (table S1). When compared with mice inoculated with bovine serum albumin (BSA) as a control, a total of 641 significantly changed proteins were identified in the brains of mice with terminal prion disease (P < 0.05, |fold change| > 1.5), of which 371 were up-regulated and 270 were down-regulated (fig. S13A and table S2). Using the same MS approach conducted concurrently, we identified 791 significantly changed proteins in the brains of mice at the terminal stage of PrP1-110-Nb–induced neurotoxicity compared to those in mice expressing the control Nb (577 up-regulated and 214 down-regulated) (fig. S13A and table S2).

Despite the notable differences between the two conditions, including the absence of PrPSc or aggregated PrP in PrP(N)-induced neurotoxicity, distinctions between prion infection versus rAAV-mediated PrP(N) expression, and different ages of animals, we observed that over 24.3% of the proteomic changes associated with PrP1-110-Nb–induced neurotoxicity overlapped with those in prion disease (Fig. 5A). More than 95.3% of these changes occurred in the same direction (Fig. 5B and table S3), supporting the similarity in the underlying neurotoxic processes. The commonly down-regulated proteins are proteins involved in protein phosphorylation, cytoskeleton dynamics, and protease activity, located in the synapses and dendritic spines, thereby affecting the synaptic function (fig. S13B). This finding is consistent with the well-documented synaptic damage observed in preclinical and clinical stages of prion disease (26, 27). The commonly up-regulated proteins (table S3) are primarily associated with immune response and endocytosis (fig. S13C), which aligns well with the severe gliosis in neurodegeneration. Collectively, proteomic analysis supports a strong correlation between the rapid neurodegeneration induced by PrP(N) and neurodegeneration in prion disease.

Fig. 5. PrP(N) contributes to the rapid neurotoxicity in prion disease.

(A) The intersection of significantly changed proteins (P < 0.05, |fold change| > 1.5) in terminal prion disease compared to BSA-inoculated mice, and those in the PrP1-110-Nb group compared to the control Nb group. (B) The heatmap illustrates the trend of significantly changed proteins at the point of intersection. (C and D) Survival curves of WT mice received ICV injection of rAAV expressing Nb (3 × 1010 gc) or N1ΔCC1-Nb (3 × 1010 gc) and then intracerebrally inoculated with brain homogenates prepared from a mouse succumbed to terminal prion disease caused by rPrPres17kDa recombinant prion (C) or RML prion (D).

Second, we took advantage of the fact that N1ΔCC1-Nb accelerates PrP(N)-induced neurotoxicity and tested whether it could exacerbate neurodegeneration in prion disease. WT mice receiving ICV injections of rAAV expressing N1ΔCC1-Nb or Nb control resulted in comparable levels of expression in the brain (fig. S14A). At 5 to 6 weeks of age, we performed intracerebral inoculation using brain homogenate prepared from a mouse that succumbed to terminal prion disease caused by recombinant prion (rPrPres17kDa) (28, 29). Compared to mice expressing the Nb control, the expression of N1ΔCC1-Nb significantly accelerated weight loss and reduced the survival time of prion-infected mice (Fig. 5C and fig. S14B), while the production of PK-resistant PrPSc and the associated neuropathology remained unchanged (fig. S14, C and D). Because the levels of endogenous PrP were identical in mice expressing N1ΔCC1-Nb or Nb control (fig. S14E), we concluded that the acceleration of neurodegeneration is due to the presence of N1ΔCC1-Nb. To rule out the possibility that this observation was specific to this particular prion strain, we repeated the same experiment with the classic Rocky Mountain Laboratory (RML) prion. Consistently, the presence of N1ΔCC1-Nb significantly accelerated weight loss and reduced the survival time of prion-infected mice, and this effect was observed in both female and male mice (Fig. 5D and fig. S15, A to E). The amount and banding pattern of PrPSc, however, were unaltered (fig. S15F). The fact that N1ΔCC1-Nb exacerbates neurodegeneration caused by both PrP(N) and prion infection establishes the role of PrP(N) in the rapid neurodegeneration in prion disease. Collectively, our findings support that PrP(N) causes rapid neurodegeneration in prion diseases.

DISCUSSION

Although the requirement of normal PrPC in the neurodegeneration of prion disease is well established (11, 12), the mechanism by which the host-encoded, highly expressed PrPC becomes neurotoxic remains unknown. Previous in vitro and ex vivo studies have suggested that PrP(N) is neurotoxic (13, 14); however, this remains a subject of debate, as some studies have suggested that PrP(N) may actually have neuroprotective effects (30, 31). To date, the neurotoxic potential of PrP(N) in vivo and its role in the rapid neurodegeneration of prion disease remain unknown. Our study has demonstrated that PrP(N) is highly neurotoxic in vivo, leading to rapid neurodegeneration with the classical neuropathological hallmarks of prion disease. The notion that PrP(N) causes rapid neurodegeneration in prion disease is further supported by several findings of our study. First, the identification of a toxic threshold for PrP(N)-induced neurotoxicity is consistent with previous observations in prion disease (18, 19), which showed that, following prion infection, prion infectivity quickly reaches a plateau, while the clinical onset of prion disease occurs only when the total disease-related PrP, but not the classic PK-resistant PrPSc, reaches a toxic threshold. Second, the substantial overlap in proteomic changes between PrP(N)-induced neurotoxicity and prion disease, despite the obvious differences, supports the similarities in neurotoxic processes. Third, N1ΔCC1-Nb, which is harmless in the brain on its own, similarly exacerbates PrP(N)-induced neurotoxicity and the neurodegeneration in prion disease, strongly indicating that the same neurotoxic mechanism is responsible for these two neurodegenerative conditions. Collectively, these lines of evidence support the conclusion that PrP(N) causes the rapid neurodegeneration in prion disease.

The origin of neurotoxic PrP(N) in prion disease remains uncertain. One possibility is that, in prion disease, processing activity may increase, leading to elevated production of N1 or N2 fragments that results in neurotoxicity. However, despite the detection of these fragments in normal animals (32), there is now no evidence showing a significant increase in their levels during prion disease. A more plausible explanation is that the interaction between PrPSc and PrPC exposes the normally hidden PrP(N) (33), allowing it to interact with cell surface molecules to cause rapid neurotoxicity (Fig. 6). This model is consistent with previous findings that PrP(N) normally binds to the folded C-terminal domain of full-length PrP (14, 33, 34) and that certain PrPC-binding antibodies induce rapid neurodegeneration (13, 35–37), presumably by disrupting this intramolecular interaction.

Fig. 6. Model of rapid neurotoxicity in prion disease.

The interaction between PrPSc and PrPC exposes PrP(N), enabling it to bind to an unidentified cell surface molecule (X). This binding initiates a toxic signaling cascade that results in rapid neurotoxicity. The figure was created using BioRender. Yan, R. (2025) https://BioRender.com/kc051x5.

Consistent with previous in vitro studies using brain slices (13, 38) and cultured cells (14, 39), we found that CC1 is essential for PrP(N)-induced neurotoxicity. The neurotoxicity of N2ΔOR-Nb, although significantly reduced compared to N2-Nb (Fig. 4D), suggests that CC1 serves as the primary driver while OR acts as the modulator of neurotoxicity. This observation that OR modulates but is not essential for neurotoxicity aligns well with previous findings from brain slice cultures (38) but differs from observations that PrP(N)-induced spontaneous currents do not require OR (14). The contribution of OR to PrP(N)-induced neurotoxicity is further supported by the toxicity enhancement observed with PrP(N)ΔCC1, which contains OR as its only distinct structural feature shared with the neurotoxic PrP1-90-Nb. Based on previously reported OR-OR interactions (40) and the recently identified OR-mediated liquid-liquid phase separation (41), it is plausible that OR-mediated interaction between the nontoxic PrP(N)ΔCC1 and the neurotoxic molecules, PrP1-90-Nb or endogenous PrPC in prion disease, may enhance the neurotoxic executor CC1’s ability to bind its target on membranes more effectively. Alternatively, this interaction may simply prevent the degradation of the freely accessible N terminus of PrP. These two possibilities are not mutually exclusive and could both contribute to enhancement of PrP(N)’s neurotoxicity.

An interesting finding of our study is that neurotoxic PrP(N) is soluble, contrasting with the notion that aggregated proteins are the direct cause of neurodegeneration, yet consistent with observations that neurodegeneration in prion disease requires the neuronal expression of normal PrPC, which is soluble on the cell membrane. A popular theory explaining PrP(N)-induced neurotoxicity involves the insertion of PrP(N) into the membrane to form a current-conducting channel (42); however, this process would require hydrophobic PrP-membrane interactions and PrP aggregation. Our finding that neurotoxic PrP(N) is soluble and only peripherally associated with membranes is incompatible with this theory. A more plausible mechanism is the binding of PrP(N) to membrane surface molecules resulting in the disturbance of physiological neuronal activity, which presents a more accessible target for disease intervention.

PrP is essential for prion disease because it is the substrate for conversion to the pathogenic PrPSc form, and it is responsible for rapid neurotoxicity, a distinguishing feature of prion disease. It is also important to recognize that the excessive accumulation of PrPSc aggregates can be detrimental as well. Thus, we propose two types of neurotoxicity associated with prion disease: a rapid neurodegenerative process attributed to PrP(N) and a slower neurodegenerative process resulting from the excessive accumulation of PrPSc aggregates. The latter bears resemblance to the prolonged disease processes observed in other neurodegenerative disorders with amyloid accumulation. These two types of neurotoxicity explain previous observations that prion infection in mice expressing PrP lacking residues 23 to 88 or 23 to 31 leads to fatal neurodegeneration, but the survival time is substantially prolonged compared to prion infection in mice expressing similar or lower levels of WT PrP (43, 44).

In addition to its essential role in prion disease, PrP also binds to oligomers formed by various amyloidogenic proteins, resulting in neurotoxic effects (45, 46). Intriguingly, one of the amyloid oligomer binding domains of PrP partially overlaps with the region responsible for the release of PrP(N) (45–48). Thus, further investigation into the neurotoxic pathway of PrP(N) may have implications that extend beyond prion disease.

MATERIALS AND METHODS

Experimental design

The objective of this study was to investigate the biological role of the N-terminal domain of PrP in vivo and its relevance to prion disease. To achieve this, we expressed N-terminal PrP fragments fused with a Nb, which enhanced their expression and permitted an evaluation of their effects on neuronal and animal survival, as well as the resulting histopathological changes. Following the characterization of the phenotypes, we conducted studies to identify the toxic domains, assess their properties related to neurotoxicity, and compare the proteomic alterations with those observed in prion disease. In the animal studies, subjects were randomly assigned to minimize bias, and an adequate number of experimental samples were included to ensure robust statistical analyses.

Plasmid construction

Plasmids PrP1-110-Nb and PrP1-254 were generated in our previous study (15). The primers and templates to construct other expression plasmids are listed in table S4. The first 25 residues of murine PrP, which contain the 22–amino acid signal peptide, were added to the Nb control plasmid to ensure that it entered the secretory pathway. All polymerase chain reaction (PCR) amplification products were ligated with Bam HI (New England Biolabs, catalog no. R3136S)– and Eco RI (New England Biolabs, catalog no. R3101S)–digested pAAV-CAG-tdTomato plasmid. Sequencing, primer synthesis, and template DNA synthesis were carried out at Azenta Life Sciences (Suzhou, China). PCR polymerase (catalog no. AP221-11Kit), ligase (catalog no. CU201-02), and competent cells (catalog no. CD201-02) were purchased from TransGen Biotech Co. Ltd. (Beijing, China). The plasmid CAG-Cre was provided by the Vector Core at Chinese Institute for Brain Research, Beijing (CIBR). The plasmids mentioned above were packaged into rAAV (PhP.eB) by the Vector Core at CIBR.

Animals

All mouse experiments and procedures were conducted in strictly accordance with the Regulations for Laboratory Animal Welfare and Ethics Management of CIBR (approval number CIBR-IACUC-039). WT mice were obtained from the Laboratory Animal Resource Center of CIBR. Prnp knockout mice, previously generated for our earlier study (23), were bred in the CIBR animal vivarium.

Knock-in mice LSL-PrP1-110-Nb, LSL-PrP1-90-Nb, LSL-Nb (all inserted at the H11 locus), and Prnp-iCre (inserted at the Prnp locus) were developed by the Genetic Manipulation Core of CIBR. To create these mice, the transgenes LSL-PrP1-110-Nb, LSL-PrP1-90-Nb, and LSL-Nb were placed under the control of the CAG promoter, with a lox-stop-lox cassette inserted in front of each transgene to inhibit expression. All mice exhibited no abnormalities. To generate Prnp-iCre mice, the codon-improved Cre recombinase (iCre) was knocked into the Prnp locus, replacing the Prnp gene.

EMX1-Cre transgenic mice (The Jackson Laboratory 005628) express Cre recombinase in ~88% of the neurons in the neocortex and hippocampus, as well as in the glial cells of the pallium (49). EMX1-Cre+/−/LSL-PrP1-90-Nb+/− mice were generated by crossing the LSL-PrP1-90-Nb+/+ mice with EMX1-Cre+/+ transgenic mice.

The Rosa26-CreERT2+/− mouse was generously gifted by Y. Rao at CIBR. After crossing with LSL-PrP1-110-Nb+/+ mice, the induction of Cre recombinase was performed at 8 weeks of age through intraperitoneal injection of tamoxifen (Sigma-Aldrich, catalog no. T5648-5G) at a dose of 10 μg/g per day for 7 days.

For prion inoculation, 5- to 6-week-old mice were anesthetized, and their top fur was shaved. Following sterilization, 1% brain homogenate prepared from a mouse that succumbed to terminal prion disease was injected into the recipient mouse’s brain using a 1-ml syringe. The injection was performed slowly, administering 30 μl of inoculum per mouse at a vertical depth of 1 mm on the right side of the sagittal suture where the parietal bones meet and ~1 mm anterior to the lambda suture where the sagittal suture intersects with the occipital bone. The injection took around 1 min, and the needle was kept in place for an additional minute after the injection. After inoculation, the mice were closely monitored for signs of disease progression, behavioral changes, and other relevant parameters. Once the terminal stage was reached, tissue samples were collected for analysis.

For inoculation with prion17kDa, only female mice were used. The inoculum was 1% brain homogenate prepared from a mouse succumbed to terminal prion disease induced by intracerebral inoculation of rPrPres17kDa recombinant prion (28). Control mice were inoculated with BSA following the same protocol. For RML prion inoculation, both sexes were used. The inoculum was 1% brain homogenate prepared from a mouse succumbed to terminal prion disease induced by intracerebral inoculation of RML prion.

ICV injection and tissue collection

Newborn pups (postnatal day 1) were anesthetized on ice for 15 s, after which 3 × 1010 genome copies (gc) of virus mixed with fast green were injected bilaterally into the lateral ventricles using a glass microelectrode (RWD Life Science, catalog no. B-15086-10F). The injection process was precisely carried out by using an injection pump (LSP Syringe Pump, catalog no. LSP01-1B). The injection site was at a point two-fifths along the line connecting the Lambda and the eye, with a vertical depth of ~2 mm. Following the injection, the needle was held in place for 10 s before being gently withdrawn. After the procedure, pups were returned to their home cage. Except for those specified in the figure legend, all injections were carried out with 3 × 1010 gc of virus.

Mice were euthanized at the terminal stage or at specified time points. For tissue collection, the mice were perfused with precooled phosphate-buffered saline (PBS) and then dissected. The mouse brains were cut sagittally, with one-half fixed in 10% neutral buffered formalin (Beijing Solarbio Science and Technology Co. Ltd., catalog no. G2161) for histopathological analyses and the other half stored at −80°C for biochemical analyses. Other indicated tissues were also collected for biochemical analyses.

Biochemical analyses

Tissue lysate preparation

Frozen mouse brain or other tissues were weighed to prepare a 10% (w/v) homogenate in PBS using a homogenizer (Next Advance, BT24M). To prepare lysates for analysis, a stock solution was added to the brain or tissue homogenate to achieve a final concentration of 0.5% Triton X-100 and 0.5% sodium deoxycholate. The mixture was then sonicated for 5 min in a water-bath sonicator (Qsonica Q700; amplitude, 10; pulse on, 10 s; pulse off, 10 s), and protein concentration was measured using the bicinchoninic acid (BCA) assay (Thermo Fisher Scientific, catalog no. 23227).

WB analysis

Western blot (WB) analysis was conducted as previously described (50). The primary antibodies used in the WB analyses included the 6D11 antibody (BioLegend, catalog no. 808003; 1:2000), variable domain of a heavy chain only antibody (VHH) antibody (GenScript Biotech Corporation, catalog no. A01860; 1:2000), 8B4 antibody (Santa Cruz Biotechnology, catalog no. sc-47729; 1:100), SAF32 antibody (Cayman, catalog no. 189720, 1:200), 3F10 antibody (a gift from Y.-S. Kim at Hallym University; 1:10,000), and anti–β-actin antibody (Sigma-Aldrich, catalog no. A1978; 1:2000). The secondary antibodies used for WB were horseradish peroxidase (HRP)–conjugated Affinipure goat anti-rabbit immunoglobulin G (IgG; H+L) (Proteintech, catalog no. SA00001-2; 1:5000) and HRP-conjugated Affinipure goat anti-mouse IgG (H+L) (Proteintech, catalog no. SA00001-1; 1:5000). Blots were developed using an enhanced chemiluminescence (ECL) Kit (Tanon, catalog no. 180-5001) and imaged with an iBright FL1000 imaging system (Thermo Fisher Scientific). For fig. S7B, which presents the analysis of PK-resistant bands, identical samples were loaded in triplicate and separated by prestained molecular weight markers. Following the transfer, the membrane was blocked with 5% skim milk (Solarbio, catalog no. D8340) for 1 hour and then cut vertically along the marker lanes to isolate each replicate. Each membrane strip was then incubated with specific antibodies as indicated in the figure legend. After incubation, the strips were reassembled along the original cut lines and subjected to ECL imaging. This approach enabled a direct comparison of antibody-specific band detection, allowing for the identification of specific epitopes present within the PK-resistant fragments.

Slot blot

After determining the protein concentration using the BCA assay, the appropriate volume of brain lysate was added to a prewashed slot blot apparatus (Bio-Rad, Bio-Dot SF Cell). Following sample filtration, the membrane was washed three times with tris-buffered saline and allowed to air-dry. The nitrocellulose membrane was then blocked with a 5% (w/v) milk solution at room temperature for 1 hour. Subsequent steps were performed in accordance with the previously described WB protocol (50).

Solubility analysis

Protein (250 μg; in 100 μl) was centrifuged at 1000 rpm for 5 min at 4°C. The supernatant was transferred to an ultracentrifuge tube containing a preloaded 50-μl 10% sucrose cushion. After centrifugation for 4 hours in an ultracentrifuge (Thermo Fisher Scientific, MX 150 Plus) at 100,000g at 4°C, the supernatant was collected and mixed with an appropriate volume of 5× SDS sample buffer [250 mM tris-HCl (pH 6.8), 12% SDS, 10% β-mercaptoethanol, 50% glycerol, and 0.04% bromophenol blue]. The pellet was washed with 100 μl of PBS, after which 50 μl of 10% sucrose was added to the bottom of the tube. The sample was then centrifuged at 100,000g for 3 hours at 4°C. The pellet was resuspended in 1× SDS sample buffer in the same volume as the supernatant and then sonicated for 5 min in a water bath sonicator (Qsonica Q700; amplitude, 40; pulse on, 10 s; pulse off, 10 s). After incubation at 95°C for 10 min, the samples were separated by 14% SDS–polyacrylamide gel electrophoresis and subjected to WB analysis.

PK digestion

Stock PK solution was added to 100 μl of proteins (5 μg/μl) to achieve a final concentration of 10 to 30 μg/ml as indicated in the figure legends. Digestion was carried out at 37°C for 1 hour and was terminated by adding phenylmethylsulfonyl fluoride (Biosharp, catalog no. BL507A) to reach a final concentration of 2 mM, followed by incubation on ice for 5 min. PK-resistant proteins were detected by WB analysis.

Lipid membrane interaction

Stock sucrose and EDTA solutions were added to 100 μl of 10% brain homogenate to achieve final concentrations of 8% and 1 mM, respectively. The mixture was incubated on ice for 10 min and then centrifuged at 1000 rpm at 4°C for 5 min. The supernatant was mixed with stock iodixanol solution to reach a final concentration of 36%, and 800 μl of the well-mixed solution was loaded into the bottom of an ultracentrifuge tube. Subsequently, 1.2 ml of 31% iodixanol solution and 200 μl of 5% iodixanol solution were added sequentially to the top. The gradient was then centrifuged at 4°C and 199,000g for 3 hours. Twelve 180-μl fractions were collected from top to bottom, and an appropriate amount of each fraction was used for WB analysis.

NaHCO3 extraction

Stock sucrose and EDTA solutions were added to 10% mouse brain homogenate to achieve final concentrations of 8% and 1 mM, respectively. After incubation on ice for 10 min, the homogenate was centrifuged at 1000 rpm for 5 min at 4°C to remove unbroken cells and nuclei. The supernatant was transferred to an ultracentrifuge tube and ultracentrifuged at 346,000g for 30 min at 4°C. The pellet was resuspended in 400 μl of 0.5 M NaHCO3 (pH 11) solution and then mixed with 600 μl of 60% iodixanol to achieve a final concentration of 36%. Density gradient centrifugation was performed as described above.

Histopathological analyses

The formalin-fixed tissue was dehydrated using an ethanol gradient, embedded in paraffin and cut into 5-μm-thick sections with a microtome (Leica, HistoCore BIOCUT). Hematoxylin and eosin (H&E) staining was carried out using a kit purchased from Beijing Solarbio Science and Technology Co. Ltd. (catalog no. G1120). For immunohistochemical staining, sections were rehydrated through a series of graded ethanol solutions. Antigen retrieval was performed by boiling the sections in a solution containing sodium citrate dihydrate (2.9 g/liter) and anhydrous citric acid (0.19 g/liter) for 10 min. The samples were then allowed to cool on the bench until they reached room temperature. After blocking