Detection of prion seeding activity in the olfactory mucosa of patients with Fatal Familial Insomnia

Fatal Familial Insomnia (FFI) is a genetic prion disease caused by a point mutation in the prion protein gene (PRNP) characterized by prominent thalamic atrophy, diffuse astrogliosis and moderate deposition of PrPSc in the brain. Here, for the first time, we demonstrate that the olfactory mucosa (OM) of patients with FFI contains trace amount of PrPSc detectable by PMCA and RT-QuIC. Quantitative PMCA analysis estimated a PrPSc concentration of about 1 × 10−14 g/ml. In contrast, PrPSc was not detected in OM samples from healthy controls and patients affected by other neurodegenerative disorders, including Alzheimer’s disease, Parkinson’s disease and frontotemporal dementia. These results indicate that the detection limit of these assays is in the order of a single PrPSc oligomer/molecule with a specificity of 100%.

Transmissible Spongiform Encephalopathies (TSEs) are a group of fatal and incurable neurodegenerative diseases that might present as sporadic, acquired or genetic conditions. The most common form of sporadic TSE is represented by the Creutzfeldt-Jakob disease (CJD) 1,2 . Acquired forms include variant CJD (vCJD) which is linked to the consumption of cattle infected with bovine spongiform encephalopathy (BSE) 3,4 . Genetic forms are associated to mutations and insertions in the PRNP gene encoding the cellular prion protein (PrP C ) 5 . Fatal Familial Insomnia (FFI) is caused by a point mutation at codon 178 of PRNP resulting in aspartic acid to asparagine substitution in coupling phase with methionine at polymorphic position 129 (D178N/129M) 6 . FFI is associated with sleep alterations, autonomic dysfunction, loss of weight and motor signs 7 . After symptoms onset, the disease usually leads to death within 2 years and no treatments are available at present 8 .
Although characterized by a great heterogeneity of clinical and neuropathological phenotypes, these diseases share a common infectious agent which is a conformationally altered form of the PrP C , known as PrP Sc , which is able to induce the conformational conversion of PrP C into new PrP Sc . Compared to PrP C , PrP Sc has higher content in β -sheet structures, is insoluble in mild detergents and partially resistant to protease degradation. Thus, after digestion with proteinase K (PK), PrP C is completely digested while PrP Sc shows a PK resistant core (27)(28)(29)(30) 9 . Mutations in the PRNP gene might favor the process of PrP C misfolding and aggregation thus favoring the formation of PrP Sc .
Both PrP C and PrP Sc can be variably glycosylated and give rise to diglycosylated, monoglycosylated and unglycosylated species 10 . After PK digestion, only the unglycosylated band can migrate at around 21 kDa (type 1) or 19 kDa (type 2) 1 . The relative abundance of each PrP glycoform (glycoform ratio) and the electrophoretic mobility of the unglycosylated form enable a biochemical classification of different prion strains. For instance, the sporadic forms of CJD are generally characterized by a predominance of the monoglycosylated PrP band, with Scientific RepoRts | 7:46269 | DOI: 10.1038/srep46269 unglycosylated PrP migrating at 21 kDa (sCJD-type1) or 19 kDa (sCJD-type2). BSE, vCJD and FFI are characterized by an abundance of the diglycosylated band with the unglycosylated one migrating at 19 kDa (type 2) 3,8,11 .
Recently, two innovative assays named Protein Misfolding Cyclic Amplification (PMCA) and Real Time Quaking-Induced Conversion (RT-QuIC) were generated to model the process of prion misfolding in vitro in an accelerated manner. PMCA 12 consists of cycles of incubation and sonication of samples containing small amount of PrP Sc in the presence of an excess of PrP C . This enables the exponential amplification of PrP Sc , and can begin the reaction with the equivalent of a single molecule of PrP Sc , which after amplification can give rise to billions of PrP Sc molecules 13 . Therefore, PMCA analysis allowed to detect extremely low levels of PrP Sc in urine 14 and blood 15,16 of patients with vCJD as well as in blood of sheep and primates experimentally infected with vCJD in preclinical stages 17 . RT-QuIC 18 alternates phases of incubation to phases of vigorous shaking where soluble recombinant prion proteins (recPrP) are used as substrate for detecting femtomolar amounts of PrP Sc . The reaction is monitored by Thioflavin-T (ThT) fluorescence and enabled detection of PrP seeds in the cerebrospinal fluid (CSF) of patients with sporadic CJD (sCJD) or genetic forms of prion disease (e.g. GSS and FFI) either in the symptomatic 19 or pre-symptomatic stage of the disease 20,21 . RT-QuIC has been further adapted to detect PrP Sc in other tissues, such as the olfactory mucosa (OM) of patients suffering from sporadic CJD 22 . Taken together, these findings suggest that RT-QuIC and PMCA have a huge potential to detect trace-amount of PrP Sc (≥ 1 femtogram) in peripheral tissues.
By taking advantage of recent observations 23-25 , we have optimized PMCA and RT-QuIC and extended their use for the analysis of OM samples collected from two patients with FFI, 16 patients with other neurodegenerative disorders including Alzheimer's disease (AD), Parkinson's disease (PD), Frontotemporal Dementia (FTD) and 10 healthy control (HC). Our results showed that both techniques were able to detect PrP Sc in FFI-OM samples in the order of a single molecule with 100% specificity.

Results
PMCA and RT-QuIC efficiently detect PrP Sc in serial dilutions of FFI brain homogenates. To assess whether PrP Sc associated to FFI (FFI-PrP Sc ) was able to amplify by means of PMCA, we have performed spiking experiment diluting FFI brain homogenate (from 10 −5 to 10 −12 dilution of the brain) into 10% healthy bank vole brain homogenates (BvBH) carrying methionine at position 109 of the prion protein (M109). After one round of amplification, we could efficiently detect up to a 10 −9 dilution while, after two rounds, we were able to amplify all dilutions (Fig. 1A). Similarly, dilutions of FFI brain homogenate were prepared in PBS and analyzed by means of RT-QuIC. The experiments were carried out using recombinant truncated form of the bank vole PrP (BvPrP(90-231)) as substrate and were seeded in triplicate with 2 μ L of each dilution (Fig. 1B). All reactions were performed at least three times by different operators. We could detect up to 10 −9 dilution of PrP Sc , while we did not efficiently detect any signal from higher dilutions. In particular, lower dilutions of brain homogenate induced a faster aggregation of BvPrP  (A) Ten % bank vole brain homogenates (BvBH) were spiked with serial dilution of FFI brain homogenate (from 10 −5 to 10 −12 ). The signal of PrP Sc was assessed by means of Western blotting, after PK digestion, with the 6D11 antibody. Mw refers to the molecular weight. PrP C refers to normal bank vole brain homogenate not digested with PK. (B) Serial dilutions of FFI brain homogenate were analyzed by means of RT-QuIC reactions using BvPrP (90-231) as substrate. Average ThT fluorescence were plotted against time. AD brain homogenate (dilution 10 −5 ) was used as negative control. BvPrP refers to unseeded recombinant protein.
Scientific RepoRts | 7:46269 | DOI: 10.1038/srep46269 replicates was higher than 10.000 AU and at least two out of three replicates crossed this value before 30 hours (see materials and methods for details). Samples that did not cross the threshold before 30 hours were considered negative. Brain homogenates of patient with AD and FTD were used as control and induced aggregation of BvPrP (90-231) after 30 hours, thus were considered negative (Fig. 1B).
PMCA and RT-QuIC can detect PrP Sc from samples of olfactory mucosa. OM samples (n = 28) collected from patients with FFI (n = 2), AD (n = 6), PD (n = 6), FTD (n = 4) and HC (n = 10) were blindly analyzed by means of PMCA and RT-QuIC. After three rounds of PMCA, we could detect PrP Sc only in FFI-OM samples ( Fig. 2A). All the other samples remained negative even after 7 rounds of amplification ( Supplementary Fig. 1). Similarly, through RT-QuIC analysis we could detect prion seeding activity in both FFI-OM samples, while PrP Sc was never detected in any of the OM belonging to the other patients analyzed (Fig. 2B). The final RT-QuIC reaction products were collected (at 30 hours) digested with PK and analyzed by means of Western blotting. As previously observed 24 , seeds collected from OM of patients with FFI, sporadic or iatrogenic forms of CJD were able to produce final RT-QuIC aggregates partially resistant to PK digestion and characterized by a molecular weight ranging from 14 to 18 kDa (Fig. 3A). On the contrary, OM samples collected from patients with PD, AD or healthy subjects did not produce any PK resistant aggregate. Brain homogenate of patients with FFI and iCJD were used as positive controls.

Estimation of PrP Sc concentration in the olfactory mucosa by means of quantitative PMCA (qPMCA).
To estimate the concentration of PrP Sc present in the olfactory mucosa of an FFI subject, we have performed a quantitative PMCA, as previously described 14 . First, we have carefully estimated the concentration of PrP Sc present in a FFI brain homogenate used to calibrate the PMCA reaction. The estimation was carried out by comparing the PrP Sc signal by Western blot with a known concentration of human recombinant PrP (Supplementary Fig. 2A). By this procedure we calculated that PrP Sc concentration in the 10% brain homogenate of this patient was approximately 141 ng/ml. Thereafter, we have performed serial dilutions of this brain homogenate and used them for PrP Sc detection after various rounds of PMCA. The results indicate that 1 round of PMCA enables to detect at around 10 −10 to 10 −11 g/ml of PrP Sc . Two or more rounds of PMCA allowed detection of 10 −13 -10 −14 g/ml of PrP Sc (Supplementary Fig. 2B). Olfactory mucosa of FFI patient resulted positive after 2 rounds of PMCA. According to our estimation the olfactory mucosa of this subject has a PrP Sc concentration of 1.41 × 10 −14 g/ml.
Biochemical characterization of PMCA reaction products. The biochemical profile of the amplified PrP Sc obtained from both FFI-OM samples was characterized by a predominance of the diglycosylated band of PrP with the unglycosylated one migrating at around 19 kDa 1 . As expected, PrP Sc obtained from FFI brain homogenate showed a similar biochemical profile that did not change after amplification by PMCA (Fig. 3B) 26 . This pattern was then compared to that of PrP Sc sCJD-type 1 or PrP Sc sCJD-type 2 before and after PMCA. Although, sCJD brain homogenates showed their characteristic PrP Sc biochemical profiles following PMCA the biochemical profile of both PrP Sc changed and acquired a unique feature characterized by the prevalence of the diglycosylated band with the unglycosylated fragment migrating at around 19 kDa. After three rounds of amplification PrP Sc could be detected in two samples (FFI-OM#1 and FFI-OM#2) that belong to subjects that were symptomatic at the time of OM collection. 10 −5 and 10 −8 refer to dilutions of FFI brain homogenate in PMCA substrate that were used as positive control for reaction. The signal of PrP Sc was assessed by means of Western blotting, after PK digestion, with the 6D11 antibody. Numbers in the right indicate the position of Mw. PrP C refers to normal bank vole brain homogenate that was not digested with PK. Dashed line indicates cropped images from separate gels. (B) RT-QuIC reactions were seeded with olfactory mucosa specimens of all patients. FFI brain homogenate (FFI-BH 10 −8 ) was used as positive control for the reaction. Two OM samples were found positive before the threshold set at 30 hours. Average ThT fluorescence were plotted against time.
Scientific RepoRts | 7:46269 | DOI: 10.1038/srep46269 Discussion RT-QuIC and PMCA efficiently demonstrated the presence of PrP Sc in urine 14 , CSF 19 , blood 15 and olfactory mucosa 22 of patients with prion diseases, at concentrations that are six orders of magnitude below the detection limits of the current diagnostic techniques, such as Western blot. RT-QuIC assay on OM samples obtained by nasal swabbing has been largely tested in patients with sporadic or genetic forms of CJD, showing a specificity of 100% and sensitivity of 97 and 75%, respectively, but never in patients with FFI or at risk of developing FFI.
In this work, we analyzed OM samples of 28 subjects 2 of whom carry the D178N mutation in the PRNP gene and already died with a clinical diagnosis of FFI. The other samples were collected from patients with different neurodegenerative disorders (including AD, PD and FTD) and healthy controls. RT-QuIC and PMCA analysis revealed the presence of PrP Sc in both FFI-OM samples that were collected 4 and 10 months after the disease onset. In both cases, three rounds of amplification were enough to enable detection of PrP Sc . Moreover, through quantitative PMCA we have estimated that the OM sample analyzed contained about 1 × 10 −14 g/ml of PrP Sc . This estimation should be carefully considered before being extended to other samples since it suffers from some limitations that might influence PMCA sensitivity. For instance, the efficiency of the nasal brushing itself and the dilution of samples before analysis might introduce technical variables to the assay. Even more important, the olfactory neurons are unique since they are replaced through life and the amount of PrP Sc collected during each nasal brushing might also depend on this phenomenon 27 . PrP Sc was not detected in patients affected by other neurodegenerative diseases. At this point, we wondered whether PMCA was able to faithfully amplify the PrP Sc originally detected in the OM of FFI patients. To this aim, we have performed a biochemical characterization of the amplified products and, after PK digestion, we observed a PrP banding profile characterized by a predominance of the diglycosylated species with the unglycosylated one migrating at around 19 kDa. This profile was compared to that of a PrP Sc collected from the brain homogenate of a patient who died for FFI 8 . Both profiles were similar and suggested that PMCA of OM samples might be able to maintain the biochemical feature of the PrP Sc typically associated to FFI. To confirm our observation, we decided to perform PMCA analysis in FFI brain homogenates and in two cases of sporadic sCJD-type 1 or sCJD-type 2 with PrP Sc biochemical profiles different to that of FFI. Following PMCA amplification, FFI-PrP Sc did not change its biochemical profile, while PrP Sc associated to each sCJD type converted to a PrP Sc glycotype characterized by a predominance of the diglycosylated band with the unglycosylated one migrating at 19 kDa (type 2). Therefore, we have realized that, even though our PMCA conditions enable PrP Sc amplification, the original biochemical features of the prion strain were not retained. Recent PMCA experiments 23,28 revealed that the glycoform ratio or the electrophoretic mobility of different prion strains Figure 3. Amplified OM-PrP Sc comparison with that of FFI or sCJD-type 1 and sCJD-type 2 brain homogenates. (A) Western blot of the final products of RT-QuIC reaction seeded with brain homogenates or olfactory mucosa from sCJD, iCJD, FFI, AD, PD patients and healthy controls (HC). Samples were digested with PK [50 μ g/mL] and immunoblot was probed with the C-terminal antibody SAF-84. Number in the right indicates the position of Mw. (B) Biochemical profile of amplified products (both brain homogenates and olfactory mucosa) was compared to that of FFI, sCJD-type 1 and sCJD-type 2 brain homogenates assessed before and after PMCA. FFI brain homogenate was concentrated 20-fold for PrP Sc detection. The signal of PrP Sc was assessed by means of Western blotting, after PK digestion, with the 6D11 antibody. Numbers in the right indicate the position of Mw: 21 kDa refers to type 1 PrP Sc , while 19 kDa refers to type 2 PrP Sc .
are not faithfully maintained after amplification using bank vole homogenates as substrate for the reaction. These results seem to be corroborated by in vivo experiments which demonstrated that intracerebral inoculation of different prion strain (including sCJD-type 1 and type 2) in bank voles 29 or transgenic mice expressing the bank vole prion protein 25 with methionine at codon 109 (Tg109M) gave rise to PrP Sc with electrophoretic mobilities similar to that of the original inoculum but with a glycoform ratio characterized by a predominance of the diglycosylated PrP band 30 . All these results suggest that, although bank voles can amplify different strains of prion, they might amplify PrP Sc with biochemical properties slightly different to that of the original prion strain. Certainly, additional factors other than PrP itself could play an important role in driving the fidelity of prion replication and further studies are necessary to clarify this intricate mechanism of amplification. PK digestion of RT-QuIC amplification products enabled detection of protease resistant recPrP useful for identifying reactions seeded with prion infected olfactory mucosa samples. However, differences in the Western blot banding profiles (between sCJD, iCJD or FFI) do not seem to have any correlation with distinct disease phenotypes.
Since the involvement of the olfactory system, including the olfactory bulb, in FFI has never been deeply investigated, we have hypothesized that PrP Sc could have spread from the central nervous system (CNS) to the OM through axonal transport along nerve fibers, as it happens for the sporadic forms of CJD 31 . Otherwise, some mutations of the PrP C gene, especially the D178N which is located in the hydrophobic core of the protein, are known to reduce the stability of the protein which is subsequently more prone to misfold 32,33 . This process might be favored in the nasal cavity, which is constantly exposed to chemical and physical stress 34 which predisposes PrP C to acquire less stable conformations, eventually leading to the formation of PrP Sc . If this were the case, the presence of PrP Sc in the OM is not directly linked to disease progression but instead represents a completely independent event.
Our results demonstrated that the OM collected from symptomatic FFI patients contain minimal amount of PrP Sc . Therefore, considering the extremely high levels of sensitivity and specificity of PMCA and RT-QuIC, we are planning to investigate whether PrP Sc might be detectable in subjects carrying the D178N mutation in the preclinical stage. There is only one study showing pre-symptomatic alterations in the thalamus of a D178N carrier who underwent to serial 18 FDG-PET that was characterized by a selective hypometabolism in the thalamus 13 months before the clinical onset. No correlations between thalamic dysfunction and presence of PrP Sc were analyzed 35 . Prospective longitudinal studies will enable to evaluate whether detection of PrP Sc in OM collected from subjects of a well characterized Italian FFI family, which is currently involved in a preventive clinical trial with doxycycline, might be used as quantifiable biomarker to monitor the effect of the drug on PrP replication and disease progression 33 .

Ethical approval. The study was approved by the institutional review board of Carlo Besta Neurological
Institute and performed according to the guidelines approved by the ethics committee. Written informed consent for participation in research was done in accordance with the Declaration of Helsinki   Olfactory mucosa brushing procedure. A step-by-step tutorial video of the nasal brushing procedure is available at: https://www.youtube.com/watch?v= wYb9W3u6uMY. This procedure was performed using flocked swabs (FLOQSwabs TM Copan Italia, Brescia, Italy), as previously described 22,36 .
Olfactory mucosa cells collections. Olfactory mucosa (OM) cells were collected from the mid-part of the inferior turbinate by brushing with a cotton swab, after treating the nasal mucosa of the patients with a topical anesthetic. Cotton swab was subsequently immersed in saline solution and olfactory cells collected from the brush by means of vortexing.

Olfactory mucosa cells preparation for PMCA analysis.
Olfactory mucosa cells were pelleted for 20 minutes at 800 × g at 4 °C. The supernatant was removed and approximately 2 μ l of the pellet was collected and transferred into a tube containing 25 μ l of PBS. The latter was sonicated (approximately 280 W) with a Microsonicator QSonica (Q700) until the pellet was dispersed and 10 μ L of this solution was added to 90 μ L of specific PMCA substrate and subjected to several rounds of amplification.
Substrate preparation for PMCA analysis. Bank voles carrying the M109 PrP genotype were sacrificed with carbon dioxide and immediately perfused using phosphate-buffered saline (PBS) plus 5 mM ethylendiaminetetraacetic acid (EDTA, Sigma). Brains were collected and homogenized in conversion buffer (PBS, 150 mM NaCl, 1% Triton X-100) at 10% (weight/volume). Heparin (100 μ g/mL, Sigma), digitonin (0.05%, Sigma) and 3 teflon beads were added to the homogenates to increase PMCA efficiency. To avoid contamination, perfusion and brain homogenates were carried out in a prion-free laboratory. PMCA procedure. PMCA was performed as previously reported 36  followed by a 30-second pulse of sonication set at a potency of 260-280 W. After one round of PMCA, an aliquot of the amplified material was diluted by a factor of 10 into fresh brain homogenate and an additional round was performed. Each sample was subjected to 7 rounds of PMCA. Considering our ultrasensitive level of amplification, all PMCA reactions were prepared under rigorous prion-free conditions. Overall, in order to avoid any contamination, appropriate number of negative controls were added in each PMCA assay. No contamination or de novo generation of prions were observed.

SDS-PAGE and Western blotting.
Ten microliters of final PMCA products was treated with 50 μ g/mL of PK (Invitrogen) for 1 hour at 37 °C under shaking (550 rpm). Digestion was stopped by the addition of LDS-PAGE loading buffer (Life Technologies), samples were heated at 100 °C for 10 minutes and loaded into 12% Bolt Bis-Tris Plus gels (Life Technologies). Proteins were separated by means of SDS-PAGE, transferred onto Polyvinylidene difluoride (PVDF, Millipore) membrane and incubated with 5% (wt/vol) dry nonfat milk in 0.05% (vol/vol) Tween-20 (prepared in Tris-HCl) for 1 h at room temperature with shaking. PVDF membranes were finally incubated with anti-PrP antibody 6D11 (epitopes 93-109, Covance) and developed with chemiluminescent system (ECL Prime, GE Healthcare Amersham).
Ten microliters of RT-QuIC reaction products was digested with 50 μ g/mL of PK at 37 °C for 30 minutes under shaking (550 rpm). Digestion was stopped by the addition of LDS-PAGE loading buffer, samples were heated at 100 °C for 10 minutes and loaded into 17% Bolt Bis-Tris Plus gels. Proteins were separated by means of SDS-PAGE, transferred onto Polyvinylidene difluoride (PVDF) membrane and incubated with 5% (wt/vol) dry nonfat milk in 0.05% (vol/vol) Tween-20 (prepared in Tris-HCl) for 1 h at room temperature with shaking. PVDF membranes were finally incubated with anti-PrP antibody SAF-84 (epitopes 160-170, Spi Bio) and developed with chemiluminescent system (ECL Prime). Olfactory mucosa cells preparation for RT-QuIC assay. Olfactory cells were collected from the brush by means of vortexing. The brush was then removed and the cells were pelleted for 20 minutes at 800 × g at 4 °C. The supernatant was removed, and approximately 2 μ l of the pellet was collected and transferred into a tube containing 25 μ l of PBS. The latter was sonicated (approximately 280 W) until the pellet was dispersed and 2 μ L of this solution was further diluted in 18 μ L of freshly prepared PBS. Two μ L of the final solution was added to each reaction well.

Real-Time Quaking Induced Conversion procedure. After purification, aliquots of the recombinant
BvPrP(90-231) were stored at − 80 °C in 10 mM phosphate buffer (pH 5.8). Before each test, the protein solution was allowed to thaw at room temperature and filtered through a 100 kDa Nanosep centrifugal device (Pall Corporation). The concentration of recPrP was determined by measuring the adsorbance at 280 nm. Samples were blindly analyzed at least three times (by different operators) in triplicate in a black 96-well optical flat bottom plate (Thermoscientific) and results showed a high degree of consistency.
The final reaction volume was 100 μ L and the reagents (Sigma) were concentrated as follow: 150 mM NaCl, 0.002% SDS, 1X PBS, 1 mM EDTA, 10 μ M ThT and 0.1 mg recPrP ml −1 . To avoid contamination, reaction mix was prepared in a prion-free laboratory. After the addition of 2 μ L of olfactory mucosa samples, the plate was sealed with a sealing film (Thermoscientific) and inserted into a FLUOstar OPTIMA microplate reader (BMG Labtech). The plate was shaken for 1 minute at 600 rpm (double orbital) and incubated for 1 minute at 55 °C. Fluorescence reading (480 nm) were taken every 15 minutes (30 flashes per well at 450 nm). Given the rapid response, a specific threshold was set to decrease the likelihood of false positives. A sample was considered positive if the mean of the highest two fluorescence values (AU) of the replicates was higher than 10.000 AU and at least two out of three replicates crossed the threshold that was set at 30 hours. This reaction cutoff was established because in all the experiments that were seeded with the brain homogenates of patients with Alzheimer's disease, no positive RT-QuIC reactions were observed until after 30 hours. In rare occasions, we could observe the aggregation of 1 replicate out of 3 before the threshold. In these cases, we have performed additional analysis of the samples in quadruplicate before considering it negative. The sample was confirmed negative if none or only one replicate (out of four) crossed the threshold before 30 hours. This allowed us to maximize the accuracy of the analysis thus reducing the risk of false positive or false negative results.