Prion potentiation after life-long dormancy in mice devoid of PrP

Abstract Prions are neurotropic pathogens composed of misfolded assemblies of the host-encoded prion protein PrPC which replicate by recruitment and conversion of further PrPC by an autocatalytic seeding polymerization process. While it has long been shown that mouse-adapted prions cannot replicate and are rapidly cleared in transgenic PrP0/0 mice invalidated for PrPC, these experiments have not been done with other prions, including from natural resources, and more sensitive methods to detect prion biological activity. Using transgenic mice expressing human PrP to bioassay prion infectivity and RT-QuIC cell-free assay to measure prion seeding activity, we report that prions responsible for the most prevalent form of sporadic Creutzfeldt–Jakob disease in human (MM1-sCJD) can persist indefinitely in the brain of intra-cerebrally inoculated PrP0/0 mice. While low levels of seeding activity were measured by RT-QuIC in the brain of the challenged PrP0/0 mice, the bio-indicator humanized mice succumbed at a high attack rate, suggesting relatively high levels of persistent infectivity. Remarkably, these humanized mice succumbed with delayed kinetics as compared to MM1-sCJD prions directly inoculated at low doses, including the limiting one. Yet, the disease that did occur in the humanized mice on primary and subsequent back-passage from PrP0/0 mice shared the neuropathological and molecular characteristics of MM1-sCJD prions, suggesting no apparent strain evolution during lifelong dormancy in PrP0/0 brain. Thus, MM1-sCJD prions can persist for the entire life in PrP0/0 brain with potential disease potentiation on retrotransmission to susceptible hosts. These findings highlight the capacity of prions to persist and rejuvenate in non-replicative environments, interrogate on the type of prion assemblies at work and alert on the risk of indefinite prion persistence with PrP-lowering therapeutic strategies.


Introduction
Prions are proteinaceous pathogens that infect the CNS and cause fatal neurodegenerative diseases in human and farmed or wild animals. 1 Neurons are the primary target cells for prion replication, but supporting glial cells, notably astrocytes can also be infected. Both prion replicative information and pathogenicity (e.g. which CNS regions will be impacted) are encoded in the structure of PrP Sc assemblies, a misfolded conformer of the ubiquitously expressed, host-encoded prion protein PrP C . 2 In infected species, prions self-replicate by a mechanism in which PrP Sc templates PrP C conformational conversion and polymerization. Multiple strains of prions are recognized phenotypically in defined hosts, due to structurally distinct PrP Sc conformers. [3][4][5][6] There is clear evidence that a single strain is composed of structurally heterogeneous PrP Sc assemblies or substrains with markedly distinct, biochemical/biophysical properties and biological activity. 7 PrP Sc assembly diversification may occur during the pathogenesis and participate to neurotoxicity and adaptation. [8][9][10][11][12] At the molecular level, PrP Sc assemblies are formed from elementary bricks of PrP. 13 These sub-elements, termed suPrP are of low-size and exhibit strong resistance to denaturant treatments such as urea. 13 PrP Sc assemblies are in dynamic equilibrium with suPrP. Such dynamics may contribute to PrP Sc assembly diversification process. 12 Quantifying prion concentration in a test sample has for long relied on time-consuming bioassays in animals. Prion infectivity titre can be obtained either by end-point titration of the sample in the reporter animals or by using incubation time values as a measure of a titre, once the prion dose-response curve has been established. 14 Alternatively, and in an accelerated way, cell-free assays estimate prion concentration by measuring prion selfconverting activity. In the protein misfolding cyclic amplification assay 15 and the real-time quaking-induced conversion (RT-QuIC) assay, 16,17 the test sample is mixed with a substrate containing PrP C or recombinant, monomeric PrP, respectively, and submitted to cycles of sonication (protein misfolding cyclic amplification) or shaking (RT-QuIC) and quiescent incubation. If the sample contains PrP Sc seeds, PrP C and recombinant PrP will be converted into PrP Sc or into amyloid aggregates, respectively. In the RT-QuIC assay, the presence of amyloid assemblies is followed in real-time by incorporation of thioflavin T, an amyloid-sensitive fluorescent dye. Both tests usually detect sub-infectious doses of prions-thus exhibit greater sensitivities than those of the animal bioassays-and have a wide range of fundamental and applied applications, including prion inactivation studies and diagnostics. [18][19][20] Prions can persist in the environment (soil or aqueous) for years. [21][22][23][24] Within tissues and in particular within the brain of challenged animals that are not permissive to prions, there is a limited amount of information on prion fate. Seminal experiments demonstrated that PrP 0/0 mice, in which the gene encoding PrP C (Prnp) has been disrupted, are absolutely resistant to infection by mouseadapted prions, derived from sheep scrapie [25][26][27] or human Creutzfeldt-Jakob disease (CJD). 28 Measurements of residual prion infectivity in the brain of the PrP 0/0 challenged animals by bioassay in indicator wild-type mice revealed that these prions were rapidly eliminated within weeks post-challenge. 25,26 Resurgent bursts of infectivity were occasionally measured in bio-indicator mice 20-30 weeks after challenge. The authors interpreted these results as the presence of residual infectivity or 'inadvertent cross-contamination'. 25,28,29 Whether other prions, including from those natural resources would exhibit similar clearance rates in PrP 0/0 mice remains unknown.
Here, we re-investigate the issue of prion persistence in PrP 0/0 mouse brain using human prions responsible for the most prevalent form of sporadic Creutzfeldt-Jakob disease (MM1-sCJD sub-type), an animal bioassay using transgenic mice overexpressing human PrP and the highly sensitive RT-QuIC assay. We show that MM1-sCJD prions can persist for the entire life in PrP 0/0 mouse brain with disease potentiation on retrotransmission to humanized mice.

Methods
Prion-infected samples MM1-sCJD Fr2 sample (frontal cortex extract) 30,31 was provided by our collaborators (A.P.L. and I.Q.) within the frame of the French National Neuropathology Network for CJD, based on availability of autopsyretained frozen brain material and informed consent from the relatives of patients for autopsy and research use, according to French regulations (L.1232-1 to L.1232-3, Code Santé Publique). MM1-sCJD UK1 sample, a WHO reference material (frontal cortex extract), 30

Animal experiments
All the experiments involving animals were carried out in strict accordance with EU directive 2010/63 and were approved by INRAE Local Ethics Committee (Comethea; permit numbers 12-034 and 15-056). PrP 0/0 mice were the so-called Zurich 1 line. 25,33 The human PrP tg650 line has previously been described. 30 This line is homozygous with about 6-fold overexpression of human PrP C (Met129 allele) in the brain. These mice do not develop any abnormal phenotype or neurological signs with aging and have a normal life span around 2-2.5 years. They do not develop any spontaneous prion disease upon inoculation with uninfected brain material. 30,34 Only PrP 0/0 and tg650 females were used; they were 6-8 weeks old at the time of inoculation. All mice were group housed by 3-5 in polypropylene cages in a standard temperature-and humidity-controlled biosafety laboratory 3 animal facility with a 12-h light-dark rhythm, unlimited access to food and water and enrichment (igloos, wood toys, nests). Cages, food, enrichment and water were sterilized before use.

Mouse bioassays
To avoid any cross-contamination, a strict protocol was followed, based on the use of disposable equipment and preparation of all inocula in a class II microbiological cabinet. MM1-sCJD Fr2 was prepared at 10% w/v brain homogenate in 5% w/v glucose with a Precellys (Ozyme, Montigny-le-Bretonneux, France). MM1-sCJD UK1 was directly provided at 10% w/v homogenate in 5% glucose.
Two groups of individually identified PrP 0/0 mice (9 mice per group) were intracerebrally inoculated with 20 ml of UK1 or Fr2 brain homogenate, using a 27-gauge disposable syringe needle inserted into the right parietal lobe. Animals were anaesthetized with 3% isoflurane during the procedure and disposed on a heating pad until they fully recovered. They were monitored daily for general health. They were euthanized at defined time points or at end-life by cervical column disruption. Their brains were carefully collected with separate, disposable tools, homogenized at 20% w/v in 5% glucose and stored at À80 C until further use. For bioassay, 20 ml of the solution were intracerebrally reinoculated at 10% w/v to groups of individually identified tg650 mice (5-7 mice per group). The inoculation procedure was the same as above. Animals were supervised daily for the appearance of neurological signs associated with the development of a prion disease. Animals at terminal stage of disease or at end-life were euthanized by cervical column disruption. Terminal stage criteria for MM1-sCJD strain in tg650 were defined as severe kyphosis, severe ataxia, lethargy, inability to reach food or water and irreversible dorsal decubitus. Brains were collected and homogenized at 20% w/v (for immunoblotting) or directly frozen on dry ice (for histoblotting) before storage at À80 C or fixed by immersion in neutral-buffered 10% formalin (for lesion profiling). The same procedure was followed for sub-passaging in tg650 mice.
For titration of MM1-sCJD or tg650-passaged MM1-sCJD brain infectivity, 30 groups of individually identified tg650 mice (4-6 mice per group) were inoculated intracerebrally (20 ml) with serial 10-fold dilutions of brain homogenates prepared in 5% w/v glucose solution containing 5% w/v bovine serum albumin, using the same procedure as above. Animals inoculated with the initial dose at 10% were assigned an infectious dose (ID) of 10 À1 . The mice were monitored daily, euthanized at terminal stage or at end-life. Their brains were carefully collected with separate, disposable tools, homogenized at 20% w/v in 5% glucose and stored at À80 C for immunoblot analyses. Infectivity titres (ID 50 ; dose that infects half the challenged animals) were calculated by the SpearmanÀKä rber method. 35 For comparison with the RT-QuIC assay, the titres were expressed as ID 50 per ml of 10% (w/v) brain homogenate.

Western blot
PrP res was extracted from 20% brain homogenates with the Bio-Rad TeSeE detection kit, as previously described. 30,31 Briefly, 200 ll aliquots were digested with proteinase K (200 lg/ml final concentration in buffer A) for 10 min at 37 C before precipitation with buffer B and centrifugation at 28 000 Â g for 5 min. Pellets were resuspended in Laemmli sample buffer, denatured, run on 12% Bis-Tris Criterion gels (Bio-Rad, Marne la Vallée, France), electrotransferred onto nitrocellulose membranes, and probed with 0.1 lg/ml biotinylated anti-PrP monoclonal antibody Sha31 antibody (human PrP epitope at residues 145-152), 36 followed by streptavidin conjugated to horseradish peroxidase. Immunoreactivity was visualized by chemiluminescence (Pierce ECL, Thermo Scientific, Montigny le Bretonneux, France). The size and relative amounts of PrP res glycoforms were determined using Image Lab software after acquisition of chemiluminescent signals with the Chemidoc digital imager (Bio-Rad, Marne la Vallée, France).

Vacuolar lesion profiles
Haematoxylin-eosin-stained paraffin-embedded brain tissue sections were used to establish standardized vacuolar lesion profiles in mice, as previously described. 38,39 Analyses were performed on 3-5 brains per passage.

RT-QuIC
RT-QuIC amplifications were performed as previously described. 17,40 Briefly, 2 ml of 10% brain homogenates (i.e. 10 À1 ID) were serially diluted in 20 mM sodium phosphate buffer pH 7.4, 130 mM NaCl, 0.1% SDS and 1Â N2 supplement (Thermo Fisher, France). Then, 2 ml of each dilution were loaded in individual wells of a black 96-well optical bottom plate containing 98 ml of 20 mM sodium phosphate buffer pH 7.4, 300 mM NaCl, 10 mM thioflavin T, 1 mM EDTA and 100 mg/ml of purified recombinant human PrP, Met129 allele. 41 The plate was sealed using Nunc Amplification Tape (Nalgene Nunc International, France), placed in a Xenius XM spectrofluorometer (Safas, Monaco) and incubated for 48-60 h at 47 C. Until the end of the measurements, cycles of 1-min orbital shaking (600 rpm) and 1-min rest were applied, and the fluorescence was recorded every 30 min. Experiments were performed in triplicates or pentaplicates. Each curve was fitted with the following equation: where Y is the fluorescence intensity and X the time. The following parameters were then calculated from the fit: fluorescence intensity maximum, slope at the inflexion point and lag time (estimated by extending the tangent at the inflexion point to the initial baseline Ymin; see Supplementary  Fig. 6).
The thresholds used to determine RT-QuIC positivity were obtained from unseeded reactions in which a fluorescence increase was observed (see Supplementary Fig. 6). The means 6 SEM values obtained were 28.8 6 1.5 h for the lag time, 32.5 6 2.8 for the fluorescence intensity maximum and 3 6 1 h À1 for the slope at the inflexion point. If one of these parameters from a RT-QuIC reaction differed from these thresholds, the RT-QuIC reaction was considered positive.
Seeding activity titre (SD 50 ; seeding dose giving thioflavin T positivity in 50% of the replicates) was estimated by the Spearman-Kä rber method. 35 When <100% of the RT-QuIC reactions seeded with the first dilution scored positive or when no dilution scored 100% positive, a trimmed variant of the Spearman-Kä rber method was applied. 42 For comparison with the bioassay, the values were expressed as SD 50 per ml of 10% (w/v) brain homogenate.

Statistical analysis
GraphPad Prism 9.0 software (GraphPad, La Jolla, CA, USA) was used to establish the Kaplan-Meier curves plotting the percentage of mice without prion disease against the incubation time. This software was also used to draw the RT-QuIC graphs and vacuolar profiles.

Data availability
All relevant data are within the manuscript and its supporting information files. Data are fully available without restriction.

Results
Lifelong persistence of CJD infectivity in PrP 0/0 mice To challenge PrP 0/0 mice, we used as inocula two unrelated brain homogenates from MM1-sCJD, one from the United Kingdom (UK1, a WHO reference material) and one from France (Fr2). We previously reported that UK1 and Fr2 homogenates were fully pathogenic for human PrP (Met129) tg650 mice, resulting in minimal disease durations of $150-160 days (Fig. 1). 30 We intracerebrally challenged PrP 0/0 mice (Zurich 1 line) 25,33 with high dose of UK1 and Fr2 [20 ml at 10% (w/v)]. The two infections were performed independently. As expected, none of the inoculated PrP 0/0 mice developed a neurological disease nor accumulated disease-specific, proteinase K-resistant PrP Sc (PrP res ) (Figs 1 and 2A; Supplementary Fig. 1).
To document the presence of residual infectivity in sCJD-inoculated PrP 0/0 mice, we intracerebrally inoculated a cohort of bio-indicator tg650 mice with brain material collected from PrP 0/0 mice euthanized healthy approximately a year post-inoculation, i.e., at mid-life (mid) and at end-life (late, 450 to 700 days post-inoculation). Whichever the time of collection, tg650 retrotransmission with PrP 0/0 -passaged UK1 and Fr2 resulted in a stereotyped prion disease in a major proportion of mice, as based on the appearance of neurological signs and accumulation of PrP res in the brain (Figs 1 and 2B and C). With mid brains, three out of four retrotransmissions were positive, with 57%, 80% and 87% attack rates. With late brains, two out of two retrotransmissions were  The electrophoretic pattern of PrP res purified from the brain of all positive tg650 mice inoculated with mid and late samples closely resembled that observed in the brain of tg650 mice on direct inoculation of MM1-sCJD, with predominance of monoglycosylated and unglycosylated PrP res and a 21 kDa migration pattern for unglycosylated PrP res (Fig. 2B and C), a stereotyped pattern referred to as T1 PrP res . 43,44 This strongly suggested that genuine MM1-sCJD prions persisted in PrP 0/0 brains. Serial passage of PrP 0/0 -retropassaged prions was performed in tg650 mice to further compare their strain properties with those of the initial and well-characterized MM1-sCJD prions. This was done by standard strain typing method comparing disease duration, PrP res electrophoretic pattern and neuroanatomical distribution of PrP res and of vacuoles. On sub-passaging, the mean survivals decreased. They established for the most advanced set of transmission (UK1) to $160 days (Fig. 1), a mean survival time typical of MM1-sCJD prions directly serially passaged in tg650 mice. 30 Immunoblotting showed that T1 PrP res accumulated in the brain on serial passage ( Fig. 2B and C). The neuroanatomical distribution of PrP res is strain-specific. 45 Direct and PrP 0/0 -intermediate transmission of MM1-sCJD prions to tg650 mice led to a similar distribution pattern of PrP res , from the primary retrotransmission onwards, as studied by histoblotting on antero-posterior coronal brain sections ( Supplementary Fig. 2). In particular, PrP res from mid and late brains accumulated specifically in certain thalamic nuclei ( Fig. 2D; Supplementary Fig. 2). This thalamic tropism is pathognomonic of MM1-sCJD prions in tg650 mice. 30,46,47 Other brain areas scored consistently PrP res -positive such as the cingulate cortex, the cingulum, the septum, the basal forebrain, the colliculi and the pons ( Supplementary Fig. 2). In the posterior thalamic nuclei and in the basal forebrain There were variable levels of PrP res deposition amongst the analysed brains, which may be due to different incubation times between the mice (Supplementary Fig. 2).
Strain-specific vacuolar lesion profiles were established by histological examination. 38,39 The vacuolation was relatively limited, and there was some variability in the intensity on serial transmission of mid and late brains, as on direct transmission of MM1-sCJD prions. In both groups, the profiles were relatively similar, with most intense areas of vacuolation in the thalamus, the superior colliculus, the frontal cortex and the mesencephalic tegmentum (Fig. 2E).
Collectively, these data show life-long persistence of MM1-sCJD prions in PrP 0/0 mouse brain. The PrP 0/0 -remnant seeds seem to retain the strain memory of the parental prions, suggesting no drastic evolution of the strain structural determinant despite a 1-2-year dormancy in PrP 0/0 brains. These remnant seeds were efficient to induce disease back in tg650 mice with respect to attack rate, suggesting substantial levels of replicating activity.

PrP 0/0 -dormant prions and low dose of MM1-sCJD prions show discrepant virulence in human PrP mice
End-point titration-based bioassay in animals allows determining prion infectivity in tissues. 14 Such an assay allows correlating the dose of infectious material with the proportion of affected animals and their mean survival times. 48 To establish such a correlation for mid and late PrP 0/0 brains, we titrated UK1 MM1-sCJD CNS material in tg650 mice by limiting dilution. The same material after one passage in tg650 mice (tg650-UK1) was previously titrated and is shown for comparison. 30 The two Kaplan-Meier curves describing the survival percentage as a function of time and dose are depicted in Fig. 3. Both titrations provided a consistent picture. As expected, the disease incubation period increased and the probability of infection became smaller with dose decrease. Inoculation of material up to the 10 À5 dilution resulted in 100% lethality. At the 10 À6 dilution, lethality was below 100%. The dose at the disease limit established at the 10 À7 dilution. At this dose, 2/6 and 4/6 animals were PrP res -positive (UK1: 351; 381 days; tg650-UK1: 342 6 37 days). For each dose including the limiting one, individual incubation periods were below the limit value of 400 days for all but one mouse that was euthanized at 451 days post-inoculation (Fig. 3). The Spearman-Kä rber method allowed calculating the dose that infects half the challenged animals (ID 50 ). UK1 and tg650-UK1 had values of 10 8.3 and 10 8.7 ID 50 /ml of 10% (w/v) tg650 brain, respectively (Table 1). Fr2 was not titrated by limiting dilution in tg650 mice. However, Fr2 has similar incubation time as UK1 and tg650-UK1 in tg650 mice 30 and they share similar seeding activity by RT-QuIC free assay (see below; Table 1). We thus considered UK1 and tg650-UK1 titrations as valid for Fr2.
Remarkably, the disease incidence and/or the incubation periods of the tg650 mice that got sick upon inoculation with mid and late PrP 0/0 brains were considerably greater than inferred from UK1 and tg650-UK1 titrations (Fig. 3). Disease occurred at 57 to 87% and 100% attack rates with the positive mid and late brains, respectively. For UK1 mid and late brains, twelve out of thirteen PrP res positive mice had individual incubation periods over the limit value of 400 days (Fig. 3A). Among them, nine had incubations periods around or over 500 days. For Fr2 mid and late brains, the situation was more balanced, seven out of the fifteen mice had incubation periods over 400 days. The other eight mice had incubation periods in the range of the 10 À6 /10 À7 dilution (Fig. 3B).
The collective titration (by the incubation period bioassay) of tissues containing low amounts of MM1-sCJD prions in our laboratory showed individual incubation periods rarely over 400 days post-inoculation ( Supplementary Fig. 3). This is consistent with the endpoint titrations (Fig. 3) and further suggests that at low or limiting dose, the disease duration of MM1-sCJD prions in tg650 mice is below 400 days. This 400-day limit is consistent with the fact that in all our end-point titrations performed so far (in the homotypic PrP context), the incubation period fold increase between the incubation duration at the lowest and at the limiting dilution is 2.17 6 0.32. 8 Applying this value to MM1-sCJD titration (mean incubation duration between 150 and 160 days) 30,47 would result in a theoretical incubation duration value at the limiting dose between $280 and $400 days. This 400-day limit is also consistent with MM1-sCJD end-point titrations with other transgenic mouse lines expressing human PrP. 49,50 Thus, the efficient transmission observed with three mid and late brains coupled with long incubation periods appears discrepant with respect to the virulence of low doses of MM1-sCJD prions in tg650 mice.

Low seeding activity of PrP 0/0 -dormant prions
To provide further quantitative estimates of prion concentration in PrP 0/0 brains, we examined the remnant MM1-sCJD seeding activity in mid and late PrP 0/0 brains relative to MM1-sCJD seeding activity in the brains of clinically sick tg650 mice inoculated with UK1 and Fr2. Two more PrP 0/0 brains were analysed compared to the bioassay, at 152 days (Fr2) and 552 days (UK1) postinfection. Ten-fold dilutions of PrP 0/0 or tg650 brains inoculated with UK1 and Fr2 were mixed with human recombinant PrP and submitted to the RT-QuIC assay. 17,40 Representative reactions are shown in Fig.4A and the results are summarized in Fig. 4B. All the individual data are shown as Supplementary Fig. 5. There was no increase in thioflavin T fluorescence up to 20-25 hours when human recombinant PrP was mixed with  , (B)). Ten-fold dilutions, ranging from 10 À1 to 10 À8 , as indicated, were intracerebrally inoculated to reporter tg650 mice. The 10 À1 dilution corresponds to the inoculation of 20 ml 10% (w/v) per mice. Kaplan-Meier curves plot the percentage of mice without prion disease (survival) against the incubation time (days post-inoculation). The different colours and symbols describe the dilutions inoculated. For comparison, grey symbols/dash lines refer to tg650 mice inoculated with mid and late brains from PrP 0/0 mice inoculated with UK1 (A) and Fr2 (B). Survival is expressed as mean 6 SEM days; in parenthesis number of diseased, PrP res -positive mice/number of inoculated mice. For nonresponder groups of tg650 mice, the range of survival time is given. serial dilutions of aged uninfected tg650 brain (Supplementary Fig. 4; Fig. 4B), suggesting no spontaneous conversion during this period. Starting from 10 2diluted 10% brain material mixed (1:50 dilution) with the recombinant PrP and thioflavin T containing buffer for the RT-QuIC reaction, brains from terminally sick tg650 mice inoculated with UK1 and Fr2 showed 100% positive replicates down to the 10 À7 dilution. The limiting dilution was achieved at the 10 À8 dilution. These end-point titrations allowed calculating the median seeding dose (SD 50 ) per millilitre of tg650 mouse brain homogenate by the Spearman-Kä rber method. It established at 10 10.6 and 10 10.2 SD 50 per mL of 10% (w/v) tg650 brain for UK1 and Fr2, respectively ( Fig. 4B; Table 1). Mid and late PrP 0/0 brains exhibited variable, yet measurable and consistently low seeding activity by RT-QuIC. In short, four out of eight PrP 0/0 brains achieved 100% positive replicates up to the 10 À3 dilution. At this dilution, the remaining four brains had 60-80% of positive replicates. The limiting dilution was For each brain tested, the average Spearman-Kärber estimates of the SD 50 /ml of 10% (w/v) brain homogenate are indicated, as well as the extrapolated ID 50 /ml and per inoculated mouse (20 ml), as inferred from Table 1. For Fr2, extrapolation is presented as grey values, as no direct measure of the ID 50 was available. NBH ¼ Normal brain homogenate.
Correlating the Spearman-Kä rber calculations of the SD 50 and ID 50 per ml 10% brain (Table 1) allowed us to extrapolate the theoretical ID 50 in PrP 0/0 brains for UK1 experiment. Except for one mid brain, the values would be below 1 ID 50 as 20 ml are inoculated in bio-indicator tg650 mouse (Fig. 4B). A similar extrapolation was made for Fr2 (although no titration was done), based on the similar correlation between SD 50 and ID 50 found for both UK1 and tg650-UK1 (see above and Table 1). This would provide theoretical ID 50 values for Fr2 mid and late brains below 1 up to 2.0 ID 50 (Fig 4B). Collectively, the RT-QuIC analyses indicate low remnant seeding activities in mid and late PrP 0/0 brains. They allow to extrapolate that the amount of infectivity inoculated to bioindicator tg650 mice would be below <1 or in the range of 1-2 ID 50 for all but one brain. Such values would not permit disease at the attack rate we observed, except for the second mid UK1 brain (Fig. 1). Thus, the efficacy at which mid and late brains infect tg650 mice appears discrepant relative to their seeding activity as measured by RT-QuIC.

Kinetics of PrP 0/0 -dormant prions versus MM1-sCJD prions
Beyond quantitating the seed concentrations in a test sample, the RT-QuIC can compare the seeding activity of closely related prion samples, based on the kinetic parameters of recombinant PrP polymerization reaction. We compared here the lag phase, slope at the inflexion point and final fluorescence intensity values ( Supplementary  Fig. 6A) of mid and late PrP 0/0 brains with those of tg650 brains infected with UK1 and Fr2. The lag time tended to last longer ( Supplementary Fig. 6B), the slope at the inflexion time ( Supplementary Fig. 6C) and the final fluorescence intensity (Supplementary Fig. 6D) tended to be lowered in PrP 0/0 brains at equivalent dilutions of brain homogenate. Yet, the brain matrix/seeding particles concentration was not the same between PrP 0/0passaged versus tg650-passaged MM1-sCJD and might contribute to the small differences observed, due to potential inhibitory effect of the brain matrix on the seeding of recombinant PrP polymerization reactions. 17 This suggests that the PrP 0/0 dormancy has not significantly altered/impacted the converting activity of the seeds.
Finally, the SD 50 concentrations over time in PrP 0/0 mice allowed us to calculate PrP Sc half-life in the PrP 0/0 brain. The values established to 25 days for UK1 and 9 days for Fr2 ( Supplementary Fig. S7). At the time of intracerebral inoculation, up to 99% of prion infectivity may escape the brain due to the spill over of surplus inoculum. 51 Applying this reduction factor to the inoculum SD 50 value at the time of injection would increase the half-life values to 50 and 15 days, respectively.

Discussion
In this study, we first asked how long prions persist in PrP 0/0 brain and remain infectious. We found that MM1 prions responsible for the most common form of sporadic CJD in humans could persist in the brain of PrP 0/0 mice, for their entire life, as shown by bioassay in human PrP transgenic mice and by measuring their seeding activity by RT-QuIC. These data considerably extend the period for which prions were previously found to persist in PrP 0/0 brains. 25,28 Prion resistance to inactivation is strain-dependent 52 and MM1-sCJD prions may be more difficult to degrade than the laboratory mouse prions used in these studies. The mouse lines used as bio-indicators for back-passage also differed; we used transgenic mice overexpressing human PrP instead of conventional mice (CD-1 or ddY mice). 25,26 Whether overexpression confers higher susceptibility (not simply shortened incubation times) remains to be determined. 34,53,54 Other major differences in the protocols could account for the differences observed between these earlier studies and ours, including notably the fact that brain pools were bioassayed whereas we tested individual PrP 0/0 brains (no dilution effect, one brain was negative here) and that these brain pools were heated for 20 min at 80 C, thus likely reducing the total amount of infectivity.
Measuring by RT-QuIC prion seeding activity over the mouse lifespan provided a unique opportunity to estimate PrP Sc half-life in PrP 0/0 brains. 55 It ranges from 9 to 25 days (15-50 days if inoculum escape from the brain is taken into account), depending on the MM1-sCJD brain analysed. These values are much higher than the 1.5-3 days found upon PrP C expression stoppage in infected cells or mouse brain 56,57 and suggest that total prion clearance in PrP 0/0 brain may be difficult.
Pathogen clearance from neurons is usually non-cytolytic and requires immunologically specific processes. 58 As prions are formed from abnormal conformations of the host-encoded prion protein, they are immunologically self-tolerated after CNS or extraneural infection of PrP Cexpressing individuals. To our knowledge, there are no reports of an antibody response mounted in intracerebrally inoculated PrP 0/0 mice. Early stage of prion infection may stimulate or involve brain innate immune response, through varying molecules and signalling pathways, including Toll-like receptors, 59 anti-inflammatory cytokines, 60 interferon-related pathways 61 and complement. 62 Such pathways may activate neuron-supporting glial cells such as microglial cells to remove toxic material. 63,64 Whether these pathways are activated in prioninoculated PrP 0/0 mice and/or are fully functional in the absence of the PrP gene remains unsure. 65,66 It may therefore be difficult to eliminate prion from brain territories, particularly in our experimental setup where we used high doses to inoculate the mice.
Second, we investigated in details PrP 0/0 -dormant prions desilencing on back passage to human PrP mice and asked whether their replicative capacities would be altered. We revealed that the probability of reinfection was higher than inferred from their low SD 50 or ID 50 concentrations. This efficacy was not palpable in terms of disease incubation periods which were overall aberrantly prolonged but in terms of disease attack rate. Remarkably, the most efficient brains with respect to attack rate were the late brains collected at 500-600 days post-inoculation. The underlying processes associated with the 'rejuvenation' of PrP 0/0 dormant prions must accommodate three intricate observations: (i) the PrP Sc assemblies partly escaped total clearance, suggestive of colonization of inaccessible territories and/or existence of conformations allowing 'absolute' resistance to catabolism; (ii) on contact with convertible PrP C back into tg650 mice, the dormant assemblies exhibited a replicative advantage compared to freshly diluted counterparts with similar activity, not in terms of fastness to disease, neither in terms of faster converting activity as shown by the RT-QuIC kinetics but in terms of disease incidence. Thus, the dormant assemblies were not the best catalysts but were the best initiators of the disease; (iii) MM1-sCJD strain properties (in tg650 mice) appeared preserved in dormant prions. In other words, these observations imply that the dormant assemblies cannot be considered as hidden, diluted material stored in inaccessible reservoir(s) that simply re-enters the conversion process unchanged. Several hypotheses could be formulated. The first one is a strain evolution during dormancy to a restrained degree of magnitude compatible with PrP 0/0derived prions retaining the parental prion strain phenotype in tg650 mice. A second one is the removal (or adsorption as with soil-bound prions?), 23 during dormancy, of molecules (e.g. carbohydrates, lipids) 67-69 serving as structural backbone to maintain PrP Sc infectious/virulence properties. Local conformational change in PrP Sc due to removal/addition of molecules may indeed change the pattern of infection. 70 The third one which is linked to recent compelling evidence that prion assemblies are not a continuum of assemblies of different size with the same core structure. 7,8,12,13,71,72 Synergies between these subassemblies are key to prion replication, diversification, and adaptation. A simple urea-induced disassembling process 7,13 or dilution process 8 can alter certain sub-assemblies with respect to their conformation, impacting directly prion biological activity. 7 Similar phenomenon could occur during life-long dormancy in PrP 0/0 brain with persistence in the brain of only certain sub-assemblies and elimination of other, that would overall impact the disease pathogenesis.
This last hypothesis would be consistent with our RT-QuIC experiments. In routine use, the RT-QuIC assay exhibits high analytical sensitivity and strong correlation with infectivity bioassay in measuring prion concentration. 17 Here, the RT-QuIC was able to detect PrP 0/0 -dormant prions but the measured SD 50 values underestimated their bona fide infectiousness. The RT-QuIC assay generates recombinant, thioflavin T positive, PrP amyloid polymorphs that are off-pathway to prion infectivity, at least with wild-type PrP sequence. 73,74 PrP 0/ 0 life-long dormancy may have specifically preserved the PrP Sc assemblies responsible for prion infectivity and destroyed part of the PrP Sc assemblies responsible for RT-QuIC seeding activity, resulting in aberrantly low SD 50 values.
Multiple lines of evidence support the view that other neurodegenerative diseases such as Alzheimer's, Parkinson's and Huntington's diseases involve similar mechanisms of misfolding and aggregation of hostencoded polypeptides through a seeded protein polymerization process. [75][76][77][78][79] The proteopathic seeds involved in the propagation of these diseases can also persist lifelong. 80,81 The extreme longevity of prion and prion-like seeds strongly advocate for stringent measures to mitigate accidental or iatrogenic transmission of these diseases by contaminated non-disposable surgical instruments or biologics. From a therapeutic standpoint, the question of whether prions or prion-like proteins can be totally eliminated from the brain is crucial. In prion diseases but also in other neurodegenerative diseases, 82 therapeutic strategies aimed at lowering/dosing the production of the disease-causing protein are emerging, using chemical compound targeting the unfolded protein response, 83 anti-PrP antibodies 84,85 or antisense oligonucleotides. 86,87 Our experiments would suggest that the brains of the treated individuals may remain potentially contagious and that replication would restart once the treatment is lifted. Early intervention before infectivity peaked in the brain and therefore our capacity to diagnose the disease as early as possible to avoid accumulation of too high levels of infectivity is key to the success of these highly promising therapies. Alternatively, multiple treatments/ interventions using the above-mentioned strategies and pharmacotherapies for promoting intra-cerebral clearance 88 may turn out to be the most viable approach to ensure total prion clearance.

Supplementary material
Supplementary material is available at Brain Communications online.