Prion diseases, or transmissible spongiform encephalopathies, are a group of neurodegenerative diseases, which share similar pathogenic mechanisms; the prion protein (PrP) undergoes posttranslational modification and is converted from an endogenous (PrP^C) to a pathogenic isoform (PrP^Sc) [1–3]. This conversion leads to the changes of chemical and physical properties, including insolubility in nondenaturing detergents, partial resistance to protease digestion, and a high tendency to form aggregates and amyloid fibrils [4–6]. Prion diseases are associated with PrP^Sc accumulation in the brain, specifically of PrP amyloid fibrils. Currently, the mechanisms underlying this conversion are incompletely defined [7, 8]. The invariable lethality of prion diseases necessitates the identification of compounds that prevent misfolding of PrP^C, replication, or accumulation of PrP^Sc [9, 10].

Current knowledge on prion diseases has revealed potential therapeutic strategies, including the prevention of the conversion to PrP aggregates, induction of degradation of these aggregates, destabilization the PrP^Sc structure, or interference with PrP^C/PrP^Sc cellular uptake [11]. Therapeutic trials have been conducted using a wide variety of potential prion diseases treatments, and numerous compounds have been extensively investigated to demonstrate antiprion activity in cell culture models [12–14]. However, no current treatment has sufficient activity to halt disease progression in infected animal models. The therapeutic use of these compounds is restricted by their intrinsic cytotoxicity and pharmacokinetic properties of drug delivery, including absorption and circulation, as well as by their limited ability to pass through the bloodbrain barrier (BBB), resulting in failed efficacy in vivo [15, 16]. Moreover, the few compounds that have been tested in humans at the postsymptomatic stage of the disease, including flupirtine [17], quinacrine [18], vidarabine [19], pentosan polysulfate [20], chlorpromazine [21, 22] and doxycycline [23], do not significantly alter the clinical course of the disease. Limitations may include oral administration, nonspecific binding to PrP^C, high toxicity even at low doses, partial effects on clinical symptoms that do not extend to increased survival, and transient activity in the brain depending on the disease course [24].

We previously described the discovery of small antiprion molecules obtained from a combined structure and ligandbased virtual screening system in silico. This approach allows the effective screening of antiprion compounds that are structurally different from known active compounds, as it reduces both time and cost. Moreover, during docking into PrP^C hotspot regions, blocking of conversion, stability, and crossing the BBB can be predicted. Various in vitro tests have demonstrated that BMD4229 is particularly outstanding, as it blocks conversion or reduces the protease K (PK)resistant PrP^Sc fragments (PrP^res) [25]. PrP^Scs mean infectious PrP aggregates and are generated by conversion of endogenous PrP^C in infected cells and mammals. PrP^ress mean PKresistant and have pathological core forms of amino acids 90–231 of PrP^Sc, and also are generated from the in vitro conversion of recombinant PrP, like a real timequaking induced conversion (RTQuIC) assay. These aggregates can be detected by western blotting and inhibited by antiPrP compounds [26]. Additionally, the structural mechanisms of BMD4229 inhibition have been studied [27]. In the present study, we extended previous research by designing the derivatives of BMD4229 using in silico feedback, specifically with chemical modifications based on docking pose of BMD4229; the aim was to elucidate the improved antiprion activity compared to that achieved by a previous procedure [25]. Here, we screened the antiprion activities of the derivatives in a PrP^Scinfected cell model and by using an RTQuIC assay. Of them, the most effective derivatives were compared with BMD4229 in a PrP^Scinfected mouse model to identify their toxicities and antiprion activities. And virtual binding sites were additionally selected. The findings have potential relevance to prion diseases treatment, thereby contributing to decreasing the lethality of prion diseases.

Compounds

The derivatives having a core structure of BMD4229 (Nphenylbenzo[ d]oxazole5sulfonamide) were designed and docked into the 2D structure of PrP^C using the ChemDraw Ultra as reported before [25]. As the final outcome, 14 derivatives with high binding energy were selected, named as BMD422901 to BMD422914, and their structures were optimized. BMD422901 to 2909 were synthesized in the Enamine (Monmouth Junction, NJ, USA), BMD422910 to 2912 were synthesized in the Akos Consulting & Solutions GmbH (Steinen, Germany) and BMD422913 and BMD422914 were synthesized in the Interchem (Paramus, NJ, USA) and ChemDiv(San Diego, CA, USA), respectively. For in vivo study, BMD4229 and BMD422910 were additionally synthesized in the 4Chem Laboratory (Gyeonggido, Republic of Korea). Their purity was determined by HPLC or proton NMR to be >95%. The chemical structures of 14 derivatives and BMD4229 were described in Fig. 1. Quinacrine dihydrochloride was purchased from SigmaAldrich (Saint Louis, MO, USA; Q3251). All tested compounds were dissolved in DMSO, diluted to produce a 50 mM stock solution, and stored at −20°C for in vitro and in vivo testing. The stock compounds were resuspended in 0.03 N NaOH in PBS just before use.

Inhibition of PrP^res in scrapieinfected cells

Prioninfected murine neuroblastoma (ScN2a) cells, derived from the neuroblastoma2a cell line of the American Type Culture Collection are stably infected with PrP^Sc (Rocky Mountain Laboratory [RML] strain); the cells were generously provided by Dr. Ryu at Hanyang University in Republic of Korea. Cell culture was performed as described previously [25]. Briefly, cells were seeded in T25 cm² flasks (TPP, Z707481) at 2×10⁵ cells, and preincubated in DMEM (Gibco, Big Carbin, OK, USA) (12430) containing 10% fetal bovine serum, 1% penicillinstreptomycin (Gibco, 15140), and 2 mM Lglutamine (Gibco, 25030) for 24 h under 5% CO₂ at 37°C. The media were replaced with fresh media containing the test compounds. After incubation for 3 days, the media were again replaced with fresh media containing the test compounds, and cells were incubated for 3 days more. Finally, the cells were harvested after incubation with the compounds for 6 days. The stock solutions of 14 derivatives were diluted in DMEM to a final concentration of 20 μM. Control cell cultures were treated with DMSO only (0.1% v/v) and quinacrine (1 μM) as activity controls. Secondary cell testing was dosedependently performed at concentrations of 1.2, 2.5, 5, and 10 μM. Each compound was tested in duplicate, in three independent experiments.

The cells were rinsed once in PBS and then incubated with 0.1% (vol/vol) trypsinEDTA (Gibco, 25300) for 1 min at room temperature. The detached cells were centrifuged at 4000×g for 10 min at 4°C and rinsed once with PBS. Cells were lysed in icecold lysis buffer (10 mM EDTA, 10 mM Tris, pH 8.0, 100 mM NaCl, 0.5% [wt/vol] Nonidet P40, and 0.5% [wt/vol] sodium deoxycholate), with two freezethaw cycles in liquid nitrogen, and then sonicated at an amplitude of 30%. The total protein concentration of cell lysates was adjusted to 200 μg/mL.

Cell lysates were digested with PK (20 μg/mL) (Merck, Kenilworth, NJ, USA) (70663) for 60 min at 37°C. The reaction was stopped with 2 mM Pefabloc (Roche, 11429876001), and the lysates were centrifuged for 90 min at 20,000×g at 4°C. Pellets were resuspended in SDS sample buffer (125 mM TrisHCl, pH 6.8, 5% [vol/vol] glycerol, 6 mM EDTA, 5% [wt/vol] SDS, 0.04% [vol/vol] bromophenol blue, and 12.5% [vol/vol] βmercaptoethanol).

Protein mixtures were separated by SDSpolyacrylamide gel electrophoresis for 2 h at 150 V, at 4°C, using 12% gels. Proteins were then transferred to a PVDF membrane using the iBlot system (Invitrogen, Carlsbad, CA, USA) for 7 min. The membrane was blocked for 1 h at room temperature in 5% (wt/vol) skim milk in PBS containing 0.1% (vol/vol) Tween 20 (PBST), and then incubated overnight at 4°C with the monoclonal PrP antibody 6H4 (1:2500; Prionics, Zurich, Switzerland; PRN01011), diluted in 10% (vol/vol) blocking buffer. After washing with PBST, the membrane was incubated for 1 h at room temperature with HRPconjugated antimouse IgG (1:5,000; DAKO, Santa Clara, CA, USA; P0260). The signals were visualized using ECL (EBP1073; Elpis Biotech, Daejeon, Republic of Korea), and the protein bands were scanned using Image Scanner III (GE, Munich, Germany). The relative band densities are shown as the volume intensity/ mm², relative to the βactin (1:5000; Cell Signaling technology, Beverly, MA, USA; 4970) band density.

Total PrP measurement

Total PrP in lysates of cells treated with desired compounds was measured using the ELISA kit for PRION (PRNP; Cloud-Clone Corp., Katy, Texas, USA) according to the manufacturer’s instructions. The absorbance at 450 nm was measured using a plate reader (VICTOR3, PerkinElmer, MA, USA). All samples were analyzed in triplicate.

Inhibition of PrP^res in RTQuIC

Human recombinant PrP (rPrP, residues 23–231; accession no. K02234) was produced, and the RTQuIC reaction was performed as described in our previous report, as a modified method of Wilham et al. and Cramm et al. [28–30]. Briefly, Escherichia coli transfected with vector with human PrP sequence was grown in Luria broth medium with kanamycin and chloramphenicol. Protein was expressed by Overnight Express Autoinduction system (Novagen, Darmstadt, Germany; 71300). The rPrP was purified from inclusion bodies using NiNTA Superflow resin (Qiagen, Hilden, Germany; 30450). The protein was refolded under gradient condition, eluted with 10 mM phosphate buffer (pH 5.8). Protein purity was measured by SDSPAGE, mass spectrometry. The final composition of the reaction buffer was as follows: 162 mM phosphate buffer (pH 6.9), 170 mM NaCl, 1 mM EDTA, 10 μM Thioflavin T (ThT), and 0.1 mg rPrP. Ninetysix microliters of buffer were dispensed into each well of a clearbottomed black 96well microplate (Nunc, NY, USA; 265301).

Two microliters of diluted sporadic Creutzfeldt Jacob disease (sCJD) brain homogenate from the National Institute for Biological Standards and Control (NIBSC; reference code NHBX0/0001) were seeded into each well. Two microliters of each compound (BMD422908, 2909, 2910, and 2914; 1.2, 2.5, 5, 10, and 20 μM) and DMSO were added to designed wells. Compounds were solubilized in DMSO. The plate was sealed and incubated for 72 h at 44°C, with intermittent shaking using a BMG Labtech Optima Fluostar plate reader. Each reaction was performed in quadruplicate. A positive response was indicated if the mean of the two highest readings at 72 h was greater than twofold, compared to the negative control. In a previous study, to exclude the possibility that compounds were affecting the ThT fluorescence dye rather than substrate rPrP or PrP amyloid, RTQuIC products were collected from wells after the last reaction, and PrP^res was detected using western blotting as mentioned above. Additionally, a dyeindependent conversion assay which is called Multimer Detection System (MDS) was performed to further confirm the nature of ThT. As a result, ThTcompound interference is not observed in optical (data not shown). This result was consistent with that of our previous experiment [30].

Inhibition of PrP^res in mice

In vivo toxicity tests were conducted in 6weeksold male and female wildtype mice (Institute of Cancer Research, London, UK). The mice were assigned to fourteen groups (n=5/group). Three male and three female groups received each 4, 20, and 100 mg/kg/d of BMD4229. Also, three male and three female groups received each 4, 20, and 100 mg/kg/d of BMD422910. One male and one female group received each DMSO with 0.03 N NaOH in PBS as a control. The compounds were daily administered intraperitoneally using a 1mL syringe and a 26gauge, 1.27 to 2.54cm (1/2 to 1 inch) needle with a short bevel for 7 days. The injection was made in the lateral aspect of the lower left quadrant [31, 32]. The volume injected into each mouse was 10 mL/kg. During and after daily compound administration, mice were closely observed for side effects by monitoring body weight and behavioral changes, such as drinking and eating patterns [33, 34]. After 7 days, the following parameters were tested; clinical signs, body weights, food consumption, organ weights, hematological parameters, and clinical chemistry. The hematological parameters included the total erythrocyte count, hemoglobin, hematocrit, mean corpuscular volume, mean corpuscular hemoglobin (MCH), MCH concentration, platelet count, and total leukocyte count. The clinical chemistry included assays for alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, glucose, blood urea nitrogen, creatinine, total cholesterol, triglycerides, total protein, albumin, and the A/G ratio.

For in vivo infectivity tests, 8weeksold female wildtype C57BL/6 mice were assigned to five infected groups (n=50/compound group, n=15/DMSO group) and one uninfected control group (n=10/group). The five infected groups were inoculated intracerebrally (i.c.) with 20 μL of 1% mouseadapted scrapie (ME7) brain homogenate. At 30 days postinfection (DPI), the compounds and DMSO were weekly administered intraperitoneally until 200 DPI. Two infected groups received each 5 and 20 mg/kg/w of BMD4229. The other two infected groups received each 5 and 20 mg/kg/w of BMD422910. The remaining one infected group and the one uninfected group received DMSO with 0.03 N NaOH in PBS. Health status was inspected daily, and body weights were recorded weekly over a period of 200 days. Seventeen to nineteen mice were euthanized at both designed 100 and 170 DPI for detection of PrP^res. For survival test, eight to ten mice were euthanized after the onset of prion diseasesassociated clinical symptoms (abnormal tail tonus, hindlimb paralysis, or weight loss) per group, and the same number of uninfected mice were treated with DMSO and observed parallelly [35]. Actually, one to three mice died for unknown reasons per group. Experimental design for both in vivo toxicity and infectivity were summarized in Table 1.

Brains and spleens of the mice euthanized at 100 and 170 DPI were removed, weighed, homogenized in buffer (PBS; pH 7.4) containing 1 mM EDTA, 5 M NaCl, Triton X100, and protease inhibitor, using a Precellys tissue homogenizer with glass beads to make 1% homogenate, and then used for western blotting. A total of 300 μL of plasma was prepared from 700 μL of whole blood, which was collected from the heart. From this sample, 30 μL of plasma mixed with 10 μL of sample buffer was used for western blotting. Mice that died from causes other than prion diseases were excluded from the data set [36]. The protein bands were scanned using Image Scanner III (GE). Scatter plots of individual data were generated using the ggplot2 software package with R software package (version 3.4.3).

Survival durations and hazardous factor analysis

Survival durations were analyzed by KaplanMeier survival analysis, and hazardous factors were then generated using MedCalc Statistical Software version 17.6 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2017) [37]. Significance was indicated by p<0.05.

Molecular docking

Molecular docking calculations were performed using the CDOCKER program to identify binding mode of novel anti-prion compound, which uses a CHARMm forcefield and grid-based molecular dynamics docking method. To determine the binding pose of compounds to the hotspot binding site of PrP^C, the NMR structure of PrP^C (1AG2) was retrieved from the Protein Data Bank. The hotspot regions of PrP^C with key residues (Asn159, Gln160, Lys194, and Glu196) was used to define the binding site of PrP^C. The best docking pose for each compound was selected based on the best scoring conformations from LigScore2, binding patterns, and visual inspection. The predicted structures of PrP^C-compound complexes were visualized using the PyMOL program.

Statistical analyses

Each experiment was repeated three times. One-way analysis of variance with Tukey-Kramer post hoc tests was conducted using Microsoft Excel and SAS (version 9.3). Differences were considered statistically significant when p<0.05.

Derivatives

To investigate the interaction pattern of BMD42-29 on the hotspot regions of PrP^C with key residues, we conducted molecular docking experiments. N-phenylbenzo[d]oxazole-5-sulfon-amide group of BMD42-29 interacted with Asn159 and Gln160 of PrP^C and docked into a cavity surrounded by Pro 158, Asn159 and Gln160, which indicated that benzene sulfonamide group likely contributes to the tight binding with PrP^C. Consequently, we decided to use N-phenylbenzo[d]oxazole-5-sulfonamide group of BMD42-29 as the starting point for further compound modifications. From the interaction patterns of the hotspot region, we expected that the substitution of a imidazo[1,2-α]pyrimidine group would allow compounds with other substituents at this position to bind the PrP^C and increase the binding specificity. In order to identify BMD42-29 derivatives of the hotspot region binding interaction with core scaffolds, we performed substructure search screening from integrated database. Finally, the 14 derivatives were identified by docking simulations and evaluated for experimental validation.

Antiprion activity of derivatives in a cell model

Inhibition of PrP^res amplification was observed in ScN2a cells treated with 14 derivatives (Fig. 2). As previously reported, no cytotoxicity was observed in compounds derived in silico below 20 μM, which was consistent with the derivatives used in the present study (data not shown) [25]. Therefore, 20 μM was regarded as an optimal dose and was administered to cells (Fig. 2A). Like that treated with DMSO, BMD422901, BMD422905, and BMD42 2913 treatment did not inhibit PrP^res amplification in cells. We observed a partial reduction of PrP^res levels <50% in BMD42 2902, BMD422903, and BMD422907, whereas BMD422904, BMD422906, BMD422911, and BMD422912 reduced a partial of PrP^res levels >50%. The BMD422908, BMD422909, BMD422910, and BMD422914 reduced PrP^res levels by >90%, similarly to what is observed with quinacrine. To identify dosedependent activities, four effective derivatives (BMD422908, BMD422909, BMD422910, and BMD422914) were serially diluted to 1.2, 2.5, 5, and 10 μM and applied to cells (Fig. 2B). Doseresponse inhibition was identified by all four effective derivatives.

The four effective derivatives reduced PrP^res levels even at 1.2 μM. The IC⁵⁰ values were 1.8 μM for BMD422908, 5 μM for BMD422909, <1.2 μM for BMD422910, and 2.1 μM for BMD422914. More than 90% inhibition was observed at 5 and 10 μM BMD422910 and at 10 μM BMD422914.Given doses of the derivatives with similar activity, BMD422910 was the most effective derivative compound. Semiquantitative intensities for Fig. 2A and 2B bands are provided in Fig. 2C and 2D, respectively. The expression of total PrP was constant in conditions of various compounds and doses, except for quinacrine (Fig. 2E and 2F). This suggested that BMD422910 as well as other derivatives only affected PrP^res at low concentrations without change of total PrP.

Antiprion activity of derivatives in RTQuIC

Blocking activity of PrP^C conversion was observed by RTQuIC using BMD422908, BMD422909, BMD422910, and BMD422914 with the same doses (Fig. 3). Increased and decreased fluorescence indicate that PrP aggregation is generated and blocked, respectively. The fluorescence was increased in DMSO controls of all experiments at 5~11 h, indicating that PrP^C conversion was rapidly generated without any disruptions. In case of BMD422908, fluorescence is not elevated at 20 μM until 62 h, whereas it was elevated in most concentrations at 18~22 h, indicating that there was no blocking activity at most doses of BMD422908. Also, the higher the dose, the later the amplification began, indicating that showed the partial blocking activity dosedependently (Fig. 3A).

In case of BMD422909, fluorescence was increased at 8~18 h in all doses similar to that of the DMSO control, indicating that PrP^C conversion was not blocked by BMD422909 and even it did not show the partial blocking activity (Fig. 3B). In case of BMD422910, the baseline level of fluorescence was maintained at 5, 10, and 20 μM until the last reading (72 h), except that the fluorescence was increased at around 62 h in wells treated with 1.2 and 2.5 μM (Fig. 3C). This result suggested that PrP^C conversion was completely blocked until the last reaction, and also the concentration required for blocking activity was relatively low in BMD422910. In case of BMD422914, the baseline fluorescence was maintained at 10 and 20 μM until the last reading (72 h), except that the fluorescence was partially elevated at 1.2, 2.5, and 5 μM (Fig. 3D). Given that the threshold concentration of BMD422910 showing completely blocking activity was 5 μM and the partial blocking activity was also much more effective than that of other derivatives, BMD422910 was the most effective antiprion activity.

Toxicological assessments of BMD42-29 and BMD42-2910 in uninfected mice

Based on cell and RT-QuIC assays, BMD42-2910 was selected for further in vivo studies among the derivatives. Therefore, BMD42-29 as a main body and BMD42-2910 were administered intraperitoneally to normal ICR mice to conduct toxicological tests. We did not observe any significant differences in daily observation such as clinical signs, body weights, and food consumption, organ weight, hematological parameters and clinical chemistry in mice treated with BMD42-29 and BMD42-2910 for a week, even though weights of liver and spleen and the albumin/globulin (A/ G) ratios were slightly changed than in control in some cases (data not shown). Therefore, we concluded that those compounds were not toxic in vivo at concentrations ≤100 mg/kg/d.

Antiprion activity of BMD4229 and BMD422910 in PrP^Scinfected mice

Considering both the optimal inhibitory concentration of 20 μM (10 mg/kg) in cells and RTQuIC and the maximum of 100 mg/kg in in vivo toxicological tests, the predicted optimal and nontoxic doses were 5 and 20 mg/kg and used to assess antiprion activity in vivo. The mice were infected intracerebrally at first and administered intraperitoneally from 30 DPI once a week. Fig. 4A showed representative immunoblots for PrP^res obtained from brain, spleen, and plasma samples from PrP^Scinfected mice treated with BMD4229 or BMD422910 at concentrations of 5 and 20 mg/kg. Compared to DMSOtreated mice, PrP^res was reduced in brains of mice treated with 5 and 20 mg/kg of BMD422910 and 5 mg/ kg of BMD4229. No antiprion activity was observed in brains treated with 20 mg/kg BMD4229. Antiprion activity was much more effective at 170 than 100 DPI in brains of mice treated with BMD422910. PrP^res in spleen was not reduced by BMD422910; in contrast, BMD4229 showed antiprion activity in spleen at 170 DPI. However, PrP^res in plasma was not affected.

Each individual immunoblot data were displayed as scatter plots (Fig. 4B). There were individual differences in all brains at 100 DPI. However, most data were distributed without individual differences in the brains at 170 DPI, except for 5 mg/kg of BMD4229 at 170 DPI. Although PrP^res of brain was reduced in 5 mg/kg BMD4229, 5 mg/kg BMD422910, and 20 mg/kg BMD422910 at 100 DPI as shown in Fig. 4A, the individual differences may not support the conclusion that they represented antiprion activity. Therefore, BMD4229 and BMD422910 partially reduced PrP^res of brain at 100 DPI. By contrast, PrP^res of brain was almost completely reduced in 5 and 20 mg/kg BMD422910 at 170 DPI. It was consistent with their individual data without the differences. Hence, the findings in brains that 5 and 20 mg/kg of BMD422910 showed antiprion activity at 170 DPI were reliable. As was the case in brains, the individual differences of PrP^res of pleen was hard to conclude that they represented antiprion activity. It was consistently ineffective in plasma. But in part, the antiprion activity was higher in spleen than in plasma, considering degree of variation between individuals.

Furthermore, survival rates were measured in PrP^Scinfected mice treated with BMD4229 and BMD422910. KaplanMeier survival analysis demonstrated that mice treated with 5 and 20 mg/kg BMD422910 survived longer than mice treated with 5 and 20 mg/kg BMD4229 or DMSO (Fig. 4C). Survival durations were increased by approximately 7 and 30 days in mice treated with BMD4229 and BMD422910, respectively. Mean hazard ratios with 95% confidence intervals indicated that it was reduced in treatment of BMD4229 and BMD422910, as compared with that of DMSO (Table 2).

The docking conformation of BMD4229 and BMD422910 to PrP^C at specific amino acids

The predicted binding modes of BMD4229, BMD422910, and BMD422904 on PrP^C hotspot binding site were shown in Fig. 5. We identified that the docked pose of BMD422910 has a similar conformation to that of PrP^CBMD4229 complexes. After applying the LigScore2 scoring functions for filtering, the best poses were selected based on exhibiting good interactions with hotspot residues (Asn159, Gln160, Lys194, and Glu196). The BMD4229 and BMD422910 interacted strongly with Asn159, Gln160 and Lys194 or Glu196 by forming hydrogen bonding interactions, as well as intermolecular πstacking interactions through its benzoxazole ring (Fig. 5A and 4B). However, BMD422904 failed to forms hydrogen bonds with Lys194 or Glu196 because of its relatively smaller substituted ring size (Fig. 5C). These results demonstrated the central role of Asn159, Gln160, Lys194 and Glu196 in the binding affinity of antiprion compounds to PrP^C. These interaction patterns of BMD422910 suggested that possible chemical modifications at the top hydrophobic pocket can lead to increase the binding affinity of the compounds. The remaining active compounds did not show binding scores high enough to allow the simulation of their docking to PrP^C.

Existing antiprion compounds include those described in library databases, FDA approved drugs for treating other diseases, natural products, antibodies, peptides and derivatives [12]. The inhibitory effects of these compounds on PrP^res are reliably identified in PrP^Scinfected cell models; however, therapeutic effects are not often observed in PrP^Scinfected mice, and clinical trials have been unsuccessful to date [13]. An advantage of exploring antiprion compounds in silico is that it enables simulated binding of the compounds to PrP^C in virtual cellular environments. Antiprion activity can therefore be expected in compounds generated using in silico systems [38, 39].

In a previous study, BMD4229 and BMD4235 were the most effective antiprion compounds tested [25]. BMD4229 displayed optimal properties, exhibiting inhibition of PrP^res amplification and stronger binding affinity than other previously reported antiprion compounds [40, 41]. However, it did not produce a marked reduction in PrP^C levels, possibly indicating that it stabilizes PrP^C and inhibits its pathological conformational change to PrP^res. In contrast, BMD4235 produced the highest antiprion activity in both PrP^res and PrP^C levels, although it had lower binding affinity, which suggests that BMD4235 acts by either stabilizing or eliminating PrP^C and may control PrP^C via nonspecific mechanisms. Hence, we found the derivatives of BMD4229 based on its remarkable inhibition for PrP^res amplification and strong binding affinity, with the aim to determine more effective compounds via in vivo testing.

Infectious forms of animal prions can be propagated in cell culture, notably in ScN2a cell lines [42, 43]. Various immunological methods are available to measure prion load in these cell lines, and therefore provide a valuable means of evaluating the antiprion activity of potential compounds. We performed RTQuIC by ThT fluorescence, as well as western blotting of cell lysates using a PrP antibody. RTQuIC is a recent innovation that allows realtime detection and monitoring of rPrP conversion to amyloid form [28]. RTQuIC has been successfully used to detect small amounts of PrP^Sc in brain tissue, cerebrospinal fluid, blood, saliva, and nasal fluid from human, cervid, ovine, bovine, and rodent species. PrP amyloid resulting from the conversion of rPrP to its aggregated form is indicated by increased ThT fluorescence, which directly indicates betasheet structure formation in protein aggregates [44].

Inhibitory effects of blocking PrP^C conversion were confirmed both in our cell model and by RTQuIC in vitro. Antiprion effects of BMD422910 and BMD422914 were also observed in cells, but differed from those observed by RTQuIC, indicating that BMD422910 blocks PrP^C conversion more effectively than BMD422914. Additionally, BMD422908 and BMD422909treated cells produced minimal PrP^res, whereas they were unable to prevent conversion in our RTQuIC assay. Consistent results would be ideal; however, consistency may be unnecessary, because antiprion strategies vary according to the mechanism of action. The RTQuIC data clearly confirmed the blocking of rPrP conversion; however, the ScN2a cell data are ambiguous because PrP^res amounts cannot be monitored continuously; therefore, the complete absence of PrP^res cannot be guaranteed and the mechanism of action is unknown for ScN2a cell screening. We thus infer that BMD422908 and BMD422909 may exhibit their antiprion activity via the degradation of PrP^res, rather than blocking of PrP^C, since RTQuIC indicated low activity.

PrP^res was dramatically reduced at 170 DPI in the brains of PrP^Scinfected mice treated with BMD422910, and this effect was sustained. Individual differences were not statistically significant; thus we could not conclude whether BMD4229 has antiprion activity in vivo. We attempted to identify antiprion activity of the compounds at 70 DPI, to model the early state of the disease; however, load of PrP^res was not detectable in DMSO control group, and therefore this timepoint was excluded from analyses (data not shown).

Administration of BMD422910 into the abdominal cavity is relatively convenient, requires a small amount of the compound, and has a higher absorption rate than oral administration because it bypasses the digestive system [45]. In our study, administration into the abdominal cavity resulted in high activity in the brain, as supported by several lines of evidence. First, the compound might be small enough to cross the BBB in order to directly affect PrP^Sc in the brain. Second, the activity of the compound could be markedly observed in brain areas with extensive PrP^Sc distribution. However, it was not easily distinguished in spleen or plasma, although the compound had suitable effects. This is likely because PrP^Sc in the spleen and plasma has relatively low infectivity and pathogenicity and is present at low concentration [46]. Therefore, high expression of PrP^Sc is likely necessary to generate the observed outcomes in the brain.

It has been proposed that the hydrogen bonds from strand S1 and helix B may prevent conversion of PrP^C to PrP^Sc [47]. BMD4229 and BMD422910 had strong hydrogen bonds at Asn159, Gln160, Lys194 and Glu196 within PrP^C and interacted with conserved PrP^C hotspot residues, indicating the importance of hydrogen bonds interactions and the hydrophobic environment; this was predicted by the pharmacophore model. We found that BMD422904 shows partial antiprion activity in our PrP^Scinfected cell model, and also binds PrP^C at Asn159 with a hydrogen bond. Although only one binding site was involved in the docking simulation, antiprion activity was feasible. Therefore, we infer that compound binding to Asn159 of PrP^C may be effective. In a previous study, we reported that the hotspot regions in β1 (Asn159), α2 (Val189, Tyr192, and Lys194), and the α1~α2 loop (Glu196), which are involved in the pathogenic conversion process, are obstructed by binding to the antiprion compounds. In particular, the α1~α2 salt bridge between Arg156 and Glu196 of the misfolded PrP^C returns to the PrP^C state in the presence of BMD4229 [27]; this finding is consistent with other reports [39, 47]. BMD422910 has hydrogen bonding interactions at Glu196, which are not absent in BMD4229. Hence, the high antiprion activity of BMD422910 may result from binding to Glu196 of PrP^C in the binding mode.

Future research should first assess whether PrP^Sc strains differ between cell (RML strain) and mouse (ME7 strain) models. As previously reported, prion studies often produce incompatible results, depending on which PrP^Sc strain is used [48]. Therefore, other PrP^Sc strains should be used in future studies, to enable generalization of BMD422910 activities. Second, BMD422910 should be tested using a more specific nonclinical trial in humanized models. For example, we plan to use PrP^Scinfected human neuronal cells and humanized transgenic mice to study potential humanhost effects. Third, cell signaling and cytokines should be analyzed in cells and in the brains of mice treated with BMD422910, to study the specific action of involved pathways. Finally, mutant PrPs at 159, 194 and 196 amino acid should be introduced in RTQuIC assays in order to determine if these positions are important for their binding or inhibition.

In conclusion, we discovered compounds that inhibit the conversion of PrP^C to PrP^res and validated the compounds using in vitro, and in vivo approaches. BMD422910, a derivative of BMD4229, was selected based on the results of a previous study, which indicated that it reduces PrP^res generation sufficiently in mouse cells and the brain. Furthermore, the predicted binding mode docking into PrP^C thereby indicated the existence of specific binding sites with hydrogen bonding. Therefore, the presently discovered inhibitor is a promising novel antiprion compound with potential to reduce the fatality of prion diseases.

This research was supported by a research grant [2014-NG52003-00] and funding from the Korea Centers for Disease Control and Prevention (4800-4838-303). RT-QuIC assay using sCJD brain homogenate and in vivo infection were conducted inside the Biosafety Level 3 facility of Korea Centers for Disease Control and Prevention (KCDC-11-3-06).