Treatment of Microglia with Anti-PrP Antibodies Induces Neuronal Allergenicity

Background: Previous reports identied proteins associated with ‘apoptosis’ following cross-linking PrP C with motif-specic anti-PrP antibodies in vivo and in vitro. The molecular mechanisms underlying this IgG-mediated neurotoxicity and the role of the activated proteins in the apoptotic pathways leading to neuronal death has not been properly dened. Previous reports implicated a number of proteins, including apolipoprotein E, cytoplasmic phospholipase A2, prostaglandin and calpain with anti-PrP antibody-mediated ‘apoptosis’, however, these proteins are also known to play an important role in allergy. In this study, we investigated whether cross-linking PrP C with anti-PrP antibodies stimulates a neuronal allergenic response. Methods: Initially, we predicted the allergenicity of the epitope sequences associated with ‘neurotoxic’ anti-PrP antibodies using allergenicity prediction servers. We then investigated whether anti-PrP antibody treatment of neuronal (N2a) and microglia (N11) cell lines leads to a neuronal allergenic response. Results: We found that both tail- and globular-epitopes were allergenic. Specically, binding regions that contain epitopes for ‘neurotoxic’ antibodies such as ICSM18 (146-159), ICSM35 (91-110), POM 1 (138-147), POM 2 (57-88) and POM 3 (95-100) lead to activation of allergenic related proteins. Following direct application of anti-PrP C antibodies on N2a cells, mass spectrometry analysis identied 4 neuronal allergenic-related proteins when compared with untreated cells. Furthermore, mass spectrometry analysis identied 8 neuronal allergenic-related proteins following cross-linking N11 cells with anti-PrP C antibodies prior to co-culture with N2a cells, when compared with untreated cells. Of importance, we showed that the allergenic effects triggered by the anti-PrP antibodies were more potent when antibody-treated microglia were co-cultured with the neuroblastoma cell line. Furthermore, in both direct and co-culture with antibody-treated microglia, we demonstrate that the allergenic proteome was part of the PrP C -interactome. Conclusions: This study showed for the rst time that anti-PrP antibody binding to PrP C triggers a neuronal allergenic response (we termed ‘IgG-Mediated Neuronal Allergenic Toxicity’) and highlights the important role of microglia in triggering IgG-mediated neuronal allergenic toxicity. Moreover, this study provides an important impetus for including allergenic assessment of therapeutic antibodies for neurodegenerative to derive safe and targeted biotherapeutics. In order to verify whether the potential allergenic effect is caused by molecules released from N11 cells following treatment with anti-PrP antibodies, the N11 cells were plated and cultured on tissue culture inserts (Nunc™ Polycarbonate Cell Culture Inserts, 0.4-micron pore size) in 24 well plate at 200,000 cells/well for 48 hours. The N11 cells were treated daily with 3μg of different anti-PrP antibodies as above. The tissue culture inserts containing antibody-treated N11cells were transferred to 24 well tissue culture plate containing conuent N2A cells and left for 3 days. Finally, the N2a cells were removed from the wells and centrifuged at 800 rpm for 5 minutes and lysed with NP-40 lysis buffer and AEBSF protease inhibitor before storing at −80°C until further use. Co-culture of N2a cells with anti-PrP antibody-treated N11 cells led to the identication of 2346 proteins (only p < 0.05) after LC-MS analysis when compared with co-culture of N2a cells with untreated-N11 cells. Out of the 2346 proteins, only the differentially expressed proteins were considered using maximum fold change ≥ 10, at least 2 identied unique peptides and a condence score ≥ 40. The stringent parameters used here led to the identication of 113 proteins (Additional le 2: Table S2). The 113 proteins were assessed for allergenicity using AllergGAtlas database and 8 proteins were conrmed to be allergenic (Table 5), including IF rod domain-containing protein (VIM), peroxiredoxin-1 (PRDX1), Legumain (LGMN), cytoskeletal beta-actin (ACTB), V(D)J recombination-activating protein 1 (RAG1), L-lactate dehydrogenase (LDHA), Receptor-type tyrosine-protein phosphatase C (PTPRC), and TIR domain-containing protein (TLR3). Among the identied 8 allergenic-related proteins, 7 (VIM, LGMN, ACTB, RAG1, LDHA, TLR3, PTPRC) showed the highest mean for DMT when compared with N2a cultured with untreated-N11 cells (Table 5). On the other hand, PRDX1 was found to be downregulated (Table 5).

activating 4 allergenic-related proteins. Co-culture of N2a with antibody treated N11 led to a more extensive alteration of the proteome and identi ed 8 allergenic-related proteins. This study demonstrates and for the rst time that cross-linking PrP C with anti-PrP antibodies leads to a neuronal allergic reaction and also highlights the crucial role played by microglia in this IgG-Mediated Neuronal Allergenic Toxicity.

Methods
The overall methods of the in silico study, in vitro experimental setup, western blotting, immuno uorescence study and liquid-chromatography massspectrophotometry (LC-MS) analysis are illustrated in in Fig. 1.

Prediction and Validation of the Three-dimensional Structure of the Human Major Prion Protein
The I-TASSER server (https://zhanglab.ccmb.med.umich.edu/I-TASSER/) [44] was used to predict the three-dimensional (3D) structure of the full-length (253 amino acid) huPrP. I-TASSER is a hierarchical template-based protein three-dimensional structure prediction server which uses the query protein sequence and predicts the structure through multiple threading alignments, iterative structure assembly simulations, and comparative functional modelling approaches using state-of-art algorithms [44]. The best protein model can be selected based on the template modelling score (TM-score), con dence score (C-score), and root mean square deviation (RMSD) score [44]. The predicted structure was visualized using the PyMOL v2.3 (The PyMOL Molecular Graphics System, Version 2.3 Schrödinger, LLC.).
The predicted 3D structure was primarily evaluated based on the Ramachandran plot [45] in PROCHECK [46] that shows mainly most favoured region, additionally allowed regions, and the disallowed regions of the protein structure that was predicted by the PDBsum server (http://www.ebi.ac.uk/pdbsum) [46]. The protein 3D model was further assessed by the ProSA (https://prosa.services.came.sbg.ac.at/prosa.php), a web server that can recognize the errors of the theoretical protein model and calculates the overall quality of the protein 3D model [47]. The SAVES v5.0 server (https://servicesn.mbi.ucla.edu/SAVES/) was used for the Verify3D score of the protein model [48]. Verify3D determines the compatibility of an atomic model (3D) with its own amino acid sequence (1D) by assigning a structural class based on its location and environment (alpha, beta, loop, polar, nonpolar) and comparing the results to good structures [48].

Prediction of Linear and conformational B Cell epitopes in the Human Major Prion Protein
The linear and conformational or discontinuous B Cell epitopes in the huPrP 3D structure were predicted using the Ellipro server (http//tools.iedb.org/ellipro/) [49]. Ellipro is a web-server which uses the geometrical properties of the protein structure in combination with MODELLER program and residue clustering algorithm for the prediction of the B Cell epitopes in the protein region protruding from the protein's globular surface [49]. We used the default parameters (minimum score 0.5 and maximum distance 0.6Å) of the Ellipro server for the prediction of both linear and conformational B Cell epitopes.

Prediction of the Toxicity and Allergenicity of the Linear B Cell epitopes
We used ToxinPred server (http://crdd.osdd.net/raghava/toxinpred/) [50] for the prediction of toxic/non-toxic nature of the linear B Cell epitopes identi ed the Ellipro server as described above. We used both the support vector machine (SVM) and quantitative matrix (QM) method in ToxinPred server for non-toxic epitope selection. The ToxinPred server has been developed based on the QM and machine learning technique using different properties of the peptides for the prediction of toxicity or non-toxicity of the peptides with 93.92% and 88.00% accuracy in SVM and QM methods, respectively. This server can also be used to identify the most toxic regions in the protein sequence [50].

Treatment of Mouse Neuroblastoma and Microglia Cell Lines with Anti-PrP Antibodies
We used a mouse neuroblastoma (N2a) (American Type Culture Collection, ATCC, USA) [53] and a mouse microglia (N11) [54] cell line to investigate the allergenic effect of anti-PrP antibodies. The N2a cells were used to assess allergenicity following direct application of anti-PrP antibodies (DAT). The N11 cells, initially treated with anti-PrP antibodies, were used to assess their allergenic effects on N2a cells following direct co-culture (DMT) or after separating the antibody-treated N11 and N2a cells by a tissue culture insert (IMT). Both N2a and N11 cells were grown in Dulbecco's Modi ed Eagle Medium (DMEM) (Thermo Fisher Scienti c, Australia), 10% fetal bovine serum (FBS) (Gibco, Fisher Scienti c, Australia), and 1% Penicillin-streptomycin (Sigma, USA) at 37 0 C in 5% CO 2 .
Direct Antibody Treatment (DAT): N2a cells were plated on 24 tissue culture well plates (Falcon, country) at 200,000 cells/well for 48 hours in tissue culture medium and kept in an incubator at 37 0 C and 5% CO 2 until optimum growth and adhesion to the surface of the plates were observed. The medium was changed daily. After 48 hours, 3μg of different anti-PrP antibodies, including ICSM18 [40], ICSM35 [40] , POM1 [43], POM2 [43], POM3 [43], SAF32 [42], or SAF70 [41] were added daily to the N2a cultures for 3 days. The cells were then removed from the plates and centrifuged at 800 rpm for 5 minutes. The cells were lysed with NP-40 lysis buffer (150mM NaCl, 1.0% Nonidet P-40 and Triton X-100, 50 mM Tris-Cl, adjust PH to 7.4) with addition of AEBSF protease inhibitor (Sigma, USA) and stored at −80°C until further use.

Direct Microglia Treatment (DMT):
The N11 cells were plated and cultured on a Petri dish at 200,000 cells/well for 48 hours in culture medium and incubated at 37 0 C in 5% CO 2 . N11 cells were then treated with 3μg of different anti-PrP antibodies as above daily for 3 days. The antibody-treated N11 cells were centrifuged at 800 rpm for 5 minutes before co-culturing with con uent N2a cells for 3 days. Finally, the N2a/antibody-treated N11 co-culture was centrifuged at 800 rpm for 5 minutes and the pellet was lysed with NP-40 lysis buffer with addition of AEBSF protease inhibitor then stored at −80°C until further use. In another experiment, the antibodytreated N11 cells were centrifuged at 800 rpm for 5 minutes and the pellet was lysed with NP-40 lysis buffer with addition of AEBSF protease inhibitor then stored at −80°C until further use.

Indirect Microglia Treatment (IMT):
In order to verify whether the potential allergenic effect is caused by molecules released from N11 cells following treatment with anti-PrP antibodies, the N11 cells were plated and cultured on tissue culture inserts (Nunc™ Polycarbonate Cell Culture Inserts, 0.4-micron pore size) in 24 well plate at 200,000 cells/well for 48 hours. The N11 cells were treated daily with 3μg of different anti-PrP antibodies as above. The tissue culture inserts containing antibody-treated N11cells were transferred to 24 well tissue culture plate containing con uent N2A cells and left for 3 days. Finally, the N2a cells were removed from the wells and centrifuged at 800 rpm for 5 minutes and lysed with NP-40 lysis buffer and AEBSF protease inhibitor before storing at −80°C until further use.
Western Blot Analysis 30µl of cell lysate (300 µg/mL) derived from antibody-treated cells was mixed with an equal volume of 1x Laemmli buffer (Bio-Rad, CA, USA). The solution was vortexed then heated for 5 min to 95 0 C. The solution was left to cool down before loading 30μl of sample into 12% SDS-PAGE gel (Bio-Rad, CA, USA) and run at 200 Volt for 5 min then 1h 30 min at 100V in running buffer (Bio-Rad, CA, USA). Following transfer at 18V for 2h 30 min in transfer buffer (Bio-Rad, CA, USA), the membranes were blocked using 2% bovine serum albumin (BSA) (Sigma-Aldrich, USA) followed by human TrueStain FC x Tm blocker (Biolegend, San Diego, USA) (5µl/blot). The blots were rinsed with TBST and 0.5 µg/ml of primary antibody mouse anti-human CD64 (FcγRI) (Biolegend, San Diego, USA), mouse anti-human CD16 (FcγRIII) (Biolegend, San Diego, USA), monoclonal mouse anti-phosphoserine (Sigma-Aldrich, USA), and monoclonal mouse antiphosphotyrosine (Sigma-Aldrich, USA) were added for overnight incubation before washing with 0.1% TBST buffer. The secondary antibody goat anti-mouse IgG (Fab speci c) (1:80000) (Sigma-Aldrich, USA) was then added for 1hour at room temperature. The blot was washed using 0.1% TBST then visualized using the Clarity Western ECL Substrate (Bio-Rad, CA, USA) in iBright™ CL1000 Imaging System (Thermo Fisher Scienti c).

Immuno uorescence Studies
Cover slips were sterilized by immersing in 70% ethanol followed by washing in 100% ethanol, rinsing in autoclaved water, and nally washing with RPMI. The coverslips were then coated with 0.1% gelatin-coating solution in ddH20. For direct antibody treatment (DAT), The N2a cells were plated on the 0.1% gelatincoated coverslip placed in a 24 well plates (Falcon, country) at 200,000 cells/well for 48 hours in tissue culture medium and kept in at 37 0 C and 5% CO 2 until optimum growth and adhesion to the surface of the coverslip were observed. The medium was changed daily. After 48 hours, 3μg of different anti-PrP antibodies were added daily to the N2a cultures for 3 days. For direct microglia treatment, the N11 cells were plated and cultured on a Petri dish at 200,000 cells/well for 48 hours in culture medium and incubated at 37 0 C in 5% CO 2 . N11 cells were then treated with 3μg of different anti-PrP antibodies as above.
The antibody-treated N11 cells were centrifuged at 800 rpm for 5 minutes before co-culturing with con uent N2a cells on a 0.1% gelatin-coated coverslip in a 24 well plates for 3 days. Mass spectrometry was performed using a Waters SYNAPT G2-Si (HDMS) spectrometer tted with a nano electrospray ionization source and operating in positive ion mode. Mass accuracy was maintained by infusing at 0.5 μL/min a lock spray solution of 1 pg/μL leucine encephalin in 50% aqueous acetonitrile, plus 0.1% formic acid, calibrated against a sodium iodide solution. The capillary voltage was maintained at 3 kV, cone voltage at 30 V, source offset at 30 V, ion block temperature 80 0 C, gas (N2) ows: purge gas 20 L/hr., cone gas 20 L/hr. MassLynx Mass Spectrometry Software (Waters Corporation, USA) was used to process the data. Each sample was run for three times in the LC-MS system and nally the collected data were run against the mouse proteome using Uniprot database and analysed using Progenesis QI software (Waters Corporation, USA).

Functional Analysis and Protein-Protein Interaction Prediction
Functional analysis of the nal dataset of DAT, DMT and IMT was performed using DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/) [59,60]. DAVID is an enriched online uni ed biological knowledge base and analytic tools which thoroughly extract the biological meaning from the expansive gene or protein list [59]. The interaction analysis among all the identi ed genes or proteins were achieved using STRING v11.0 (https://string-db.org/). This online platform is used as functional protein association networks which provides an in-depth assessment and integration of protein-protein interactions including both direct and indirect associations [61].

Identi cation of the Allergy Related Genes
The nal dataset from LC-MS was checked to nd out whether there is any allergy related genes or not in our identi ed gene list. We used AllerGAtlas 1.0 (http://biokb.ncpsb.org/AlleRGatlas/), a human allergy-related genes database which has been developed based on the 1195 well-annotated human allergyrelated genes, determined by text-mining and manual curation [62]. The objectives of developing this AllerGAtlas database was to look on the pathogenesis and epidemiology of individual cases, novel diagnostic and prognostic biomarker, individual treatment responses and precision medicine [62].

Classi cation of Gene and Gene Enrichment Analysis
The identi ed gene dataset was submitted to the online-based PANTHER classi cation system v14.0 (http://www.pantherdb.org/) for the classi cation of the identi ed genes based on the biological process, cellular components, molecular function, protein class, and signaling pathways [63]. This is a wide-ranging system that helps assess and analyse extensive genome-wide experimental data [63]. In addition, the gene enrichment analysis was conducted on the identi ed nal dataset of DAT, DMT and IMT using FunRich software v 3.1.3 [64]. FunRich is a stand-alone software tool used mainly for functional enrichment and interaction network analysis of genes and proteins [65].

Statistical analyses
Statistical analyses were assessed by Student's t-test, Chi-Squared test or Anova test. The results were considered signi cant at p<0.05. However, Bonferroni corrected p-value was used in FunRich analysis.

Results
Modelling of the Human Major Prion Protein Three-dimensional Structure Following assessment with the I-TASSER server [44], ve 3D models of the huPrP were predicted, but only the best protein model ( Fig. 2A) was selected based on the con dence score (C-score = -4.13), estimated template modelling score (TM-score = 0.28 ± 0.09), and estimated root mean square deviation (RMSD = 16.0 ± 3.1 Å) score. The Ramachandran plot analysis of huPrP by PDBsum server [46] showed 68.4% residues in most favoured region, 24.7% in allowed regions, 3.7% in generously allowed regions, and 3.2% residues in the disallowed regions for the predicted prion protein structure (Fig. 2B). The ProSA server [47] assesses the overall and local model quality of the predicted protein structure. The local model quality is shown by knowledge-based energy score, in which the positive values represent the erroneous or problematic regions of the input structure (Fig. 2C). The sequence position in the negative values of the knowledge-based energy con rms the good quality of our predicted protein structure. Moreover, ProSA also provides the z-score for overall model quality prediction and was − 5.74, indicating that it is located in the acceptable area [47] (Additional le 1: Figure S1A). The Verify 3D determine the compatibility of an atomic model (3D) with its own amino acid sequence (1D) by assigning a structural class based on its location and environment (alpha, beta, loop, polar, non-polar etc.) and compares the results to good structures [48]. For our protein structure, we found 80.63% of the residues have averaged 3D-1D score ≥ 0.2, well within the acceptable limit to be considered a good quality protein structure (Additional le 1: Figure S1B and Figure S1C).

Identi cation of Allergenic B Cell Epitopes in the Three-dimensional Structure of the Human Major Prion Protein
We predicted the linear B cell epitopes from the huPrP 3D structure and found 10 B cell linear epitopes ranging from 4 to 26 amino acids long with a protrusion index (PI) of 0.503 to 0.798 (Table 1) [66]. The full-length huPrP is divided into three major parts, including the exible tail (FT) region , which comprises the octa-peptide repeats (OR) region  and the globular domain (GD) regions (124-230). We found four epitopes (epitope L7: 24-49, epitope L9: 56-66 and epitope L10: 76-86, epitope L2:  in the FT region, two of them located in the OR region. The GD contained three epitopes (epitope L4: 137-153, epitope L3: 167-174 and epitope L6: 189-204). On the contrary, the short epitope 1-4 (epitope L8) is located in the N-terminal region (signal peptide region) and comparatively two longer epitopes L1:222-238 (both in GD and non-structured region) and L5:245-253 (non-structured region) are located in the C-terminal region of the full length huPrP ( Table 1). The position of the B cell linear epitopes on the protein structure are illustrated in (Fig. 3A & 3B). We also predicted 9 B cell conformational epitopes, shown in Table 2. The length of the B cell conformational epitopes ranged between 4 and 36 amino acid residues. The protrusion index value for the B cell conformational epitopes ranged between 0.55 and 0.883.  Table 2 Predicted conformational B cell epitopes from the three-dimensional structure of the human major prion protein.

Epitope
No.

Treatment of Mouse Neuroblastoma Cell lines with Anti-PrP Antibodies Leads to Neuronal Allergenicity
Treatment of N2a cells with anti-PrP antibodies ICSM18, ICSM35, POM1 and SAF70 led to the identi cation of 211 proteins (p < 0.05) after LC-MS analysis when compared with untreated N2a cells. Out of the 211 proteins, only the differentially expressed proteins were considered using maximum fold change ≥ 10, at least 2 identi ed unique peptides and a con dence score ≥ 40. The stringent parameters used here led to the identi cation of 26 proteins (Additional le 2: Table S1). Of note, the stringent parameters used to identify proteins associated with anti-PrP treatment are unusually high and would allow elimination of 'false-negatives' post LC-MS analysis. The 26 proteins were then assessed for allergenicity using AllergGAtlas database (http://biokb.ncpsb.org/AlleRGatlas/) [62] and 4 allergy related proteins, including beta-actin (ACTB), fatty acid-binding protein 5 (FABP5), protocadherin 11 (PCDH11X), and myomegalin (PDE4DIP). Among the 4 allergenic-related proteins, ACTB, PCDH11X and PDE4DIP were upregulated but FABP5 was found to be down regulated when compared to untreated control ( Table 3). The functional annotation of the 4 allergenic-related proteins through DAVID bioinformatics resources [59,60] showed that ACTB is associated with platelet aggregation and cellular response to electrical stimulus; PCDH11X is involved in the negative regulation of phosphatase activity; FABP5 was found to be associated with phosphatidylcholine biosynthetic process and transport while PDE4DIP was found to be involved in cellular protein complex assembly. Protein-protein interaction of the identi ed 4 allergenic-related proteins with prion protein (PRNP) showed that PrP networks with ACTB via Co lin-1 (CFL1), while no direct interaction was observed for FABP5, PCDH11X and PDE4DIP (Fig. 5A). Finally, analysis of individual anti-PrP antibody treatments revealed that PDE4DIP was present after DAT with ICSM18 and POM1 treatment, however, DAT with ICSM35 and SAF70 was not found to be associated with allergy related proteins (Table 4).

Table 4 Identi cation of antibody-speci c allergenic proteins following direct antibody treatment (DAT). (√) Present and (-) Absent
Gene ID Accession ICSM Antibodies SAF Antibody POM Antibody

Gene Ontology (GO) Analysis of Allergy Related Proteins Associated with Direct Antibody Treatment
We performed Gene Ontology (GO) analysis using the gene classi cation server, PANTHER classi cation system (v.14.0) [63], and the protein gene enrichment software, FunRich (Functional Enrichment analysis tool-Version 3.1.3) [64] (reference list "rodent database") for the analysis of cellular components, molecular function, biological process, and the signalling pathway. PANTHER identi ed ACTB and PCDH11X allergenic-related genes involved in cellular components, biological processes and signalling pathways. The cellular component analysis of the 4 allergenic-related proteins showed that both ACTB and PCDH11X are associated with cell and membrane while ACTB is involved in membrane-enclosed lumen, organelle, protein-containing complex, and supramolecular complex (Fig. 5B). The molecular function of ACTB gene was found to be associated with binding and structural molecule activity (data not shown). The biological process of PCDH11X gene was found to be associated with biological adhesion, whereas ACTB was found to be involved in biogenesis, cellular process, developmental process, localization, locomotion, and multicellular organismal process (Fig. 5C). Signalling pathway analysis identi ed ACTB and PCDH11X association with cadherin signaling and Wnt signaling pathways. In addition, signaling pathway analysis of ACTB was found to be involved in Alzheimer disease-presenilin pathway, cytoskeletal regulation by Rho GTPase, Huntington disease, integrin signalling pathway, nicotinic acetylcholine receptor signaling pathway, and in ammation mediated by chemokine and cytokine signaling pathway (Fig. 5D). Co-Culture of Anti-PrP Antibody Treated-Microglia with Mouse Neuroblastoma Cell Line Leads to Neuronal Allergenicity Co-culture of N2a cells with anti-PrP antibody-treated N11 cells led to the identi cation of 2346 proteins (only p < 0.05) after LC-MS analysis when compared with co-culture of N2a cells with untreated-N11 cells. Out of the 2346 proteins, only the differentially expressed proteins were considered using maximum fold change ≥ 10, at least 2 identi ed unique peptides and a con dence score ≥ 40. The stringent parameters used here led to the identi cation of 113 proteins (Additional le 2: Table S2). The 113 proteins were assessed for allergenicity using AllergGAtlas database and 8 proteins were con rmed to be allergenic (Table 5) (Table 5). On the other hand, PRDX1 was found to be downregulated (Table 5).  Fig. 7). It was previously shown that overexpression of PrP C itself activates the NADPH oxidase (NOS) for reactive oxygen species (ROS) production that initiates the co lin activation and nally induce co lin-actin rods in hippocampal neurons [68].

A study by Esue et al. demonstrated a direct interaction between actin and vimentin laments mediated by the tail domain of vimentin [69]. Kristiansen et al.
showed that mild proteasome impairment in prion-infected cells leads to the formation of aggresomes that contain VIM, HSP70, ubiquitin as well as proteasome subunits [70]. Of interest, PTPRC was found to be upregulated in mice brain following inoculation with prions [71]. A study by Ramljak et al. showed that there is a direct interaction between PrP C and lactate dehydrogenase (LDHA) and revealed that LDHA expression is increased under hypoxic conditions [72]. Protein-protein interaction also showed that both RAG1 (node 6 in Fig. 7) and TLR3 (node 5 in Fig. 7) indirectly interact with ACTB (node 1 in Fig. 7) via PTPRC (node 3 in Fig. 7) while PRDX1 (node 7 in Fig. 7) indirectly interacts with ACTB via LDHA (node 4 in Fig. 7) (Fig. 7). A study by Wagner and co-workers showed that PRDX6 was upregulated in scrapie-infected mice and neuronal cell lines [73]. However, LGMN (node 8 in Fig. 7) did not interact with any of the identi ed allergenic-related proteins as well as with prion protein (node 9 in Fig. 7).
Among the identi ed allergenic-related proteins, VIM was found to be involved in the progression of allergic diseases via in ammasome [74,75] and VIM-P38MAPK complex facilitates mast cell activation via FcϵRI/CCR1 activation [76]. LDHA was identi ed as a potential marker in allergic alveolitis, airway in ammation, allergic encephalomyelitis, asthma disease [77][78][79][80]. PRDX1 was found as a negative regulator of in ammation [81], Th2-type airway in ammation, and allergen-related hyperresponsiveness [82]. PTPRC was found to be associated with asthma related phenotypes in a microarray analysis [83].
LGMN was found to be involved in allergic reaction by potentiating antigen processing [84]. A study by Sehra et al. showed that RAG1-de cient mice exhibited reduced mast cell in ltration when it was used as a chronic model of allergic in ammation [85].
TLR3 activation in an established experimental allergic asthma mice model increased the release of proin ammatory cytokines and mucus production which was also associated with the increased production of interleukin 17 (IL-17A) by natural killer (NK) cells [86].
In order to verify whether the 8 allergenic-related proteins were speci cally stimulated in neurons (and not in both neurons and microglia) following co-culture with antibody-treated microglia, we compared the proteome of the anti-PrP antibody-treated microglia without co-culture with neurons and found that anti-PrP antibody-treated microglia only did not display any common allergy-related proteins with DMT (Additional le 2: Table S3) indicating that our 8 identi ed allergy-related proteins were speci cally activated in neurons.

Gene Ontology analysis of Allergy Related Proteins Associated with Direct Microglia Treatment
PANTHER analysis of the allergy related protein after DMT showed that the allergenic-related proteins are involved in cellular components, molecular functions, biological processes, signaling pathways, and protein classes (Fig. 8). Cellular components were classi ed into different groups in PANTHER analysis including cell (PRDX1, TLR3, ACTB) and membrane (TLR3, ACTB), membrane-enclosed lumen (ACTB), protein-containing complex (ACTB), and supramolecular complex (ACTB) (Fig. 8A). The molecular function was further classi ed into 3 different groups where it was observed that ACTB were shown to be involved in both structural molecule activity and binding activity and LGMN, PRDX1 and PTPRC are associated with catalytic activity (Fig. 8B). The biological process analysis showed that PRDX1 and TLR3 are involved in biological regulation, immune system process, and response to stimulus whereas PTPRC, LGMN, PRDX1, TLR3, and ACTB were found to be associated with cellular process (Fig. 8C). On the other hand, ACTB, was shown to be involved in biogenesis, developmental process, localization, locomotion, and multicellular organismal processes (Fig. 8C). The analysis also showed that PTPRC, LGMN, and PRDX1 are associated with metabolic processes while TLR3 is involved in signaling and multi-organism processes (Fig. 8C). The signaling pathway analysis is divided into 11 groups whereas PTPRC is involved in both B cell and T cell activation and JAK/STAT signaling pathway (Fig. 8D). On the other hand, ACTB is found to be associated with Alzheimer disease-presenilin pathway, cadherin signaling pathway, cytoskeletal regulation by Rho GTPase, Huntington disease, in ammation mediated by chemokine and cytokine signaling pathway, integrin signalling pathway, nicotinic acetylcholine receptor signaling pathway, and wnt signaling pathway (Fig. 8D). Contactless Co-Culture of Anti-PrP Antibody Treated-Microglia and Mouse Neuroblastoma Cell Lines Leads to Neuronal Allergy.
Contactless co-culture of anti-PrP antibody treated-microglia N11 and N2a cells was designed to verify whether the allergenic effects caused by DMT were due to a direct cognate interaction of N2a and N11 or via indirect release of microglial factors which in turn might have led to allergenicity. N11 cells were initially treated with anti-PrP antibodies, including ICSM18, ICSM35, POM1, POM2, POM3, SAF32 or SAF70 on tissue culture inserts before placing the inserts containing antibody-treated microglia on tissue culture plate containing untreated N2a cells (IMT). IMT resulted in an initial dataset of 292 proteins (p < 0.05) after LC-MS analysis. Differentially expressed proteins (p < 0.05) were considered with a maximum fold change ≥ 10 and at least 2 identi ed unique peptides and a con dence score ≥ 10 and identi ed a total of 11 proteins (Additional le 2: Table S4). Out of the 11 proteins, AllergGAtlas database identi ed Integrin beta-4 (ITGB4) (upregulated, p = 0.034, maximum fold change 45, con dence score 33.6, peptide 6, unique peptide 3) as being allergenic.
The protein-protein interaction analysis showed that ITGB4 indirectly interacts with PRNP via ITGB6 and NCAM 1 (Additional le 1: Figure S2A). Santuccione and co-workers showed that activation of p59fyn was achieved via PrPC recruitment NCAM to lipid rafts [87]. Ghodrati et al. also showed that NCAM directly interacts with PrP and identi ed the transforming growth factor β and integrin signaling as prion interactors via gene ontology analysis [88].
Liu and co-workers revealed that ITGB4 is involved in airway hyper-responsiveness and lung in ammation in allergic asthma. ITGB4-de cient mice displayed increased in ltration of lymphocyte, neutrophil, and eosinophil as well as expression of IL-4, IL-13, and IL-13A in lung tissue [89]. Tang et al. showed that ITGB4-de ciency causes spontaneous exaggerated lung in ammation in early life [90]. A study by Yuan and co-workers showed that p53 pathway activation in ITGB4-de ciency prompts the senescence of airway epithelial cells [91]. Yuan and co-workers also demonstrated that the lack of ITGB4 is responsible for increased Th2 responses in allergic asthma by down-regulation of CCL17 and EGFR pathway in airway epithelial cells [92].

Discussion
Hypersensitivity reactions are triggered by the immune response. Little is known about the so-called IgG-mediated neuronal hypersensitivity, however, a body of new emerging studies suggest that hypersensitivity is an important feature in response to IgG immunotherapy or disease-associated auto-antibodies [93][94][95]. Fcγ receptors (FcγRs) are known to mediate protective immune functions via binding of IgG molecules to the Fc domain in addition to modulating the adaptive immune response. FcγRs have been implicated in hypersensitivity reactions; for instance, Fc γ-chain-de cient mice were protected against a number of autoimmune disorders (reviewed in [67]), suggesting an important role for FcγRI in hypersensitivity reactions. Furthermore, functional polymorphism in FCγRs genes was shown to play an important role in the pathogenesis of allergy [96]. A number of in vitro and in vivo studies have demonstrated the presence of FcγRs in neurons. Previous reports implicated IgG in inducing a neuronal hypersensitivity reaction [94]. Of importance, Fuller et al. highlighted the importance of increased expression and ligation of FcγRs in the CNS as a result of administration of therapeutic antibodies or by endogenous IgG which resulted in vascular damage and exacerbation of neurodegeneration [97]. Furthermore, experimental treatment with anti-PrP antibodies directed against PrP C led to neuronal apoptosis based on non-molecular microscopic assessments [23,24,26]. Upon further analysis, some of these studies also revealed that anti-PrP antibodies induced activation of allergenic-related proteins identi ed by the AllerGAtlas database [21,27,28]. We therefore sought molecular con rmation of a neuronal hypersensitivity/allergenic process associated with anti-PrP antibody treatment of neuroblastoma and microglia. Initially, we performed in silico analysis to predict the most antigenic epitopes from the huPrP 3D structure and to verify whether some of the predicted motifs overlap with those recognized by the reported 'neurotoxic' anti-PrP antibodies such as ICSM and POM antibodies [40,43]. The in silico analysis revealed a set of antigenic B cell linear epitopes located on the exible tail (FT) region: L2 (89-111), L7 (24-49), L9 (56)(57)(58)(59)(60)(61)(62)(63)(64)(65)(66), and L10 (76-86) of which L9 and L10 are on the octa-peptide repeats (OR) in the FT region. Interestingly, epitopes L2 and L9/L10 were mapped to the 'neurotoxic' antibodies ICSM35 [26,40], POM3 and POM2 [23,24,27,43]. We also con rmed that L2, L9 and L10 were toxic following assessment with the ToxinPred server [50] by quantitative matrix based method (QM method). The insilico analysis also revealed a set of antigenic B cell linear epitopes located on the globular (GD) region: L4 (137-153), L3 (167-174), L6 (189-204), L1 (222-238) and L5 (245-253), of which L4 was mapped to the 'neurotoxic' antibodies ICSM18 [24,26,40] and POM1 [24,27,43]. The ToxinPred server also identi ed L4 as being toxic. While performing in silico analysis to assess antigenicity and toxicity of epitopes located on the huPrP 3D structure, we also noticed that some antigenic epitopes were predicted to be allergenic by AllergenFP [52] and AllerTop [51]  L10 (76)(77)(78)(79)(80)(81)(82)(83)(84)(85)(86) with AllerTop server. Of interest, L4 was mapped to the neurotoxic antibodies POM1 and ICSM18 and L2 was mapped to ICSM35 and POM3 while L9 and L10 were both mapped to POM2. POM1, a similar antibody to ICSM18 and mapped to the L4 epitope was previously shown to induce neurotoxicity via calpain [27]. Our LC-MS data also shows that calpain 1 is activated by ICSM18 treatment, but on the contrary led to inhibition of calpain 3 (data not shown), probably via negative feedback following cross-linking PrP C . In addition to its newly characterized role in antibody-induced neuronal apoptosis, calpains have a well-established role in allergy [98][99][100]. For instance, a study by Wu et al. showed that calpain 1 contributes to mast cell degranulation [98]. Furthermore, inhibition of mGluRs, known to regulate histamine [34], abolished the anti-PrP antibody toxic effects [28]. Taken together and in addition to the allergenic-related proteins associated with ICSM35 treatment reported by Tayebi and colleagues [21], this provides su cient evidence to investigate the allergenic pathways potentially induced by treatment with anti-PrP antibodies which we refer to as "IgG-Mediated Neuronal Allergenic Toxicity".
To that end, we treated mouse neuroblastoma (N2a) and microglia (N11) cell lines with anti-PrP antibodies then assessed the neuronal allergenic proteome following mass spectrometry analysis as well as expression of neuronal FcγRI and FcγRIII. Initially, we applied the anti-PrP antibodies directly (DAT) on N2a cells and showed that SAF70 and POM1 but not ICSM18 led to a high molecular band associated with both FcγRI and FcγRIII but not in untreated control, indicating that antibody-treated N2a cells are positively primed to regulate an immune response [101]. DMT with anti-PrP antibodies does not appear to affect expression of both CD64 and CD16, however, IMT with anti-PrP antibodies led to increased CD64 and CD16 expression with ICSM18 and POM1. These results indicate that despite a possible role for FcγRs in inducing allergenicity via antibody treatment of microglia following binding of the Fc portion, it is more probable that this was caused by cross-linking of PrP C with anti-PrP antibodies via binding with the Fab region. A report by Lunnon et al. demonstrated increased activation of microglial FcγRs in a mouse model of prion diseases, suggesting that this increase is directly linked to PrP alteration [102]. The FcγRs activation pro le as shown by western blotting contrasted with the outcome observed following LC-MS analysis. Failure of DMT to alter FcγRs activation was not re ected in the pro le of the allergenic proteome as this identi ed 8 upregulated allergenic proteins, further proving that this allergenic reaction is caused following cross-linking of PrP C and not via binding to the Fc portion of the antibody. This dichotomy is also re ected by the DAT where FcγRI and III was activated but LC-MS identi ed only 4 proteins associated with allergy. Amongst the 4 proteins, ACTB was shown to interact with PrP C via co lin 1 by STRING.
Walsh and co-workers demonstrated that overexpression of PrP C activates the NADPH oxidase (NOS) for reactive oxygen species (ROS) production which initiates co lin activation and nally induce co lin-actin rods in hippocampal neurons [68]. Moreover, DAVID analysis showed that ACTB was involved in platelet aggregation. Blood platelets play an active and essential role in allergic in ammation and pathogenesis of the allergic diseases [103,104]. The cytoplasmic FABP5 is found in adipocytes and was previously shown to be with allergic asthma where FABP5 is upregulated in sputum of allergic asthmatics and had positive correlation with the vascular endothelial growth factor (VEGF) [105]. Lee et al. also observed an association between FABP5 and atopic march or allergic march where knockdown of FABP5 dramatically reduced IL-17A in T cells from atopic march patients [106]. Another study by Shum et al. revealed that FABP controls allergic airway in ammation [107]. Ge et al. showed that FABP4 is involved in the regulation of eosinophil recruitment and plays a proin ammatory role during allergic asthma development [108]. Of note, FABP5 is an intracellular lipid carrier protein involved in the regulation of in ammation and its inhibition decreases levels of prostaglandin E 2 and proin ammatory cytokines [109]. PCDH11X and PCDH1 are cell adhesion molecules and belong to the cadherin protein family that are highly expressed within the CNS [110][111][112]. PCDH1 was shown to be associated with childhood asthma, bronchial hyperresponsiveness as well as eczema and other atopic phenotypes [113][114][115]. Variation in the PCDH11X gene is associated with late-onset Alzheimer's disease [116,117]. Finally, PDE4 plays an essential regulatory role in immune and in ammatory cells [118]. The involvement of PDE4 in allergy and asthma has been widely studied in human blood leukocytes and a positive regulatory role of the IL-4 has also been established [118][119][120][121]. In addition, inhibition of PDE4 leads to reduction of hyperresponsiveness, airway in ammation as well as phosphodiesterase activity [122][123][124].
The PANTHER analysis showed that DAT-associated allergenic-related proteins are associated with Wnt and integrin, in ammation mediated by chemokine and cytokine and nicotinic acetylcholine receptor signaling pathway. The Wnt/β-catenin signaling pathway is involved in airway remodeling in chronic asthma and enhances the development of allergic airway disease [125][126][127]. The recruitment of eosinophils is a prominent feature of asthma and integrin is involved in the regulation of extravasation of eosinophils [128,129]. The Nicotinic acetylcholine receptors modulate the synaptic and cellular functions in the brain and are important for the regulation of cytokine release [130]. Impairment of the α7-nicotinic acetylcholine receptor leads to high production of cytokine that enhances the possibility of tissue damage [131,132]. FunRich analysis of DAT-associated allergenic-related proteins highlighted the association between regulation of prostaglandin biosynthetic process and allergic lung in ammation [133,134]. Similarly, cytochalasin B was shown to be associated with asthma through enhancing release of platelet-activating factor (PAF)-induced histamine [135]. During the course of prion disease, mouse microglia are highly activated and express TGF-β and PGE 2 , both known mediators of allergy [136,137]. Moreover, anti-PrP therapeutic antibodies also led to strong microglial activation associated with neuronal loss [25]. Due to the important role played by microglia in the exacerbation of neuropathology in the prion and other related disorders, we sought to verify whether co-culture of N2a with anti-PrP antibody-treated N11 (DMT) leads to a neuronal allergenic reaction. We have identi ed 8 allergenic-related proteins following DMT, 7 of which were upregulated with the exception of PRDX1 that was found to be downregulated. The protein-protein interaction analysis showed that VIM, interacts with ACTB. VIM is an intermediate lament protein which plays an important role in stabilizing intracellular architecture and a regulatory role in the NLRP3 in ammasome where IL-1β and caspase-1 were decreased in VIM de cient macrophage cells [74].
Of note, IL-1β and NLRP3 in ammasome are involved in the progression of allergic diseases [75]. A study by Toda et al. revealed that VIM-P38MAPK complex facilitates mast cell activation via FcϵRI/CCR1 activation [76]. Amongst the DMT associated allergenic-related proteins, LDHA was found to interact with PRDX1 by STRING. A study by Ramljak and co-workers revealed that PrP C and LDHA are direct interactors and that LDHA expression increased under hypoxic conditions [72]. Plasma LDHA was found to be a potential marker for cryptogenic brosing alveolitis and extrinsic allergic alveolitis [77]. In three separate study, serum LDHA was also found to be a potential marker for airway in ammation [78], atopic dermatitis [138] and experimental allergic encephalomyelitis [79]. Al Obaidi and co-workers showed that sputum LDHA is a potential and accurate marker in asthma disease [80]. Taken together, the above studies clearly highlight the important role played by LDHA in allergy. Interestingly, PRDX6, a protein that directly networks with our downregulated PRDX1, was shown to be upregulated in scrapie-infected mice and neuronal cell lines and controls expression of PrP C and PrP Sc in neuronal cells [73]. PRDX1 is a ubiquitous antioxidant enzyme known to act as a negative regulator for protection against in ammation [81]. Inouue et al. showed that PRDX1 protects against allergenrelated hyperresponsiveness and Th2-type airway in ammation and involved in the inhibition of allergen-speci c T-cell proliferation through immunological synapse [82]. In protein-protein interaction analysis PTPRC directly interacted with ACTB, TRL3, RAG1 and RAG2. PTPRC was upregulated in brain of mice following infection of prion [71]. PTPRC is associated with asthma related phenotypes [83] and its ligation enhances the frequency of constitutive apoptosis in human eosinophils [139]. LGMN helps to destroy the ASNase activity (degraded asparaginase produced by Escherichia coli) leading to an allergic reaction by potentiating antigen processing [84]. RAG1 plays an important regulatory role in the reorganization and recombination of T cell receptor (TCR) and immunoglobulin (Ig) genes [140]. A study by Sehra et al. showed that Rag1-de cient mice exhibited reduced mast cell in ltration when it was used as a chronic model of allergic in ammation [85]. Moreover, Rag1-de cient zebra sh showed that the immunity expression and apoptosis associated genes are increased and showed greater prevalence of cell cycle arrest and oxidative stress [141]. Of interest, rag1-mice were shown to be highly resistant to oral challenge with prions [142]. Another allergenic-protein identi ed following DMT is TLR3 that directly interact with PTPRC. TLR3 is a membrane protein which act as a pathogen recognition receptor and is expressed in the CNS and other cell types [143]. TLR3 activation by poly (inosinic-cytidylic) acid in an established experimental allergic asthma mice model increased the release of proin ammatory cytokines and mucus production which was also associated with increased production of IL-17A by NK cells [86]. Starkhammar and colleagues showed that combined stimulation of TLR3 and TLR4 causes airway hyperresponsiveness which is increased during an ongoing allergic in ammation [144]. ACTB, which was found in both DAT and DMT, is associated with platelets aggregation [145] and might have a role in allergic diseases [103,104].
Analysis of individual anti-PrP antibody treatment identi ed that the highest allergenic effect, as assessed by the number of allergenic-related proteins, was associated with the GD and FT targeting antibodies ICSM18/SAF70 and ICSM35 which shared 5 common proteins. ICSM18 and 35 were produced in PrP-null against truncated hrPrP 91-231 while SAF70 was raised in hamsters using SAF preparation. However, both ICSM18 and SAF70 bind to a similar epitope on the GD but ICSM35's epitope is located on the FT domain. ICSM35 and SAF70 are of the same Ig isotype (IgG2b). It remains a challenge to pinpoint which of these antibody characteristics led to activation of the common proteins, however, there is strong indication that inherent antibody properties (e.g. epitope; isotype etc.) trigger similar allergenic pathways. POM2 and SAF32, two antibodies that bind to an epitope on the octa-repeat activated 6 proteins separately where 5 proteins were common and the protein TLR3 and PTPRC was found to be activated by POM2 and SAF32, respectively. SAF32, similar to SAF70 was raised in hamsters using SAF preparation and is an IgG2b, while POM2 was raised in PrP-null mice against full-length mrPrP  and is an IgG1. In this case, the common octa-repeat epitope appears to be playing a key role in triggering similar allergenic pathways by these 2 antibodies, however, all molecular aspects should also be considered, including antibody a nity for instance. It is noteworthy that among the 5 common proteins shared by POM2 and SAF32, 4 proteins were also common with ICSM18, ICSM35 and SAF70 treatments, possibly re ecting the involvement of several antibody properties. Finally, POM1 and POM3, raised in PrP-null mice against full-length mrPrP  and of IgG1 isotype with binding motifs located on GD and FT domains respectively activated only 4 proteins with very little commonality with the other antibody treatments.
The PANTHER analysis of the 8 allergy-related proteins identi ed pathways that were common between DAT and DMT. However, DMT also includes JAK/STAT signalling pathway, B cell and T cell activation, and toll receptor signaling pathway. The JAK/STAT signaling pathway was found to be associated with the chronic in ammatory skin disease atopic dermatitis [146]. Toll-like receptors are also involved in allergy pathogenesis [147], asthma [148], and allergic rhinitis [149]. Functional gene enrichment analysis identi ed positive regulation of protein tyrosine phosphatase activity besides the other immune response related function such as positive regulation of T cell mediated immunity, regulation of humoral immune response, regulation of interleukin-8 production, and positive regulation of antigen receptor-mediated signaling pathway. Protein tyrosine phosphatases is involved in the regulation of allergic asthma where the protein tyrosine phosphatase inhibition in allergen-challenge phase or allergen-sensitization phase helps decrease the development of asthma which correlated with increased T helper 1 (Th1) response [150]. Finally, IMT activated one allergy-related protein. ITGB4 expression reduces antigen presentation and regulates airway in ammation reaction in allergic asthma [151]. Yuan and co-workers demonstrated that lack of ITGB4 is responsible for increasing of Th2 responses in allergic asthma by down-regulation of CCL17 and EGFR pathway in airway epithelial cells [92]. ITGB4 de ciency was also found to be associated with airway in ammation where ITGB4-de cient mice showed increase of microglia and pro-in ammatory cytokines, TNF-α, IL-6, and IL-1β in the hippocampus and prefrontal cortex [152]. The effect by IMT, albeit limited as it led to activation to a single allergenic protein, might have been triggered by release of cytokines such as IL-1, Il-6, and TNF-α antibody-treated N11 [54] and possibly induced a neuronal allergenic reaction.

Conclusion
Antibody-mediated therapy for prions attracted intense debate and controversy as some of the reported were contradictory partly because these relied on a somewhat super cial assessment using microscopy and also due to the failure of investigating the ne molecular events caused by cross-lining PrP C with anti-PrP antibodies. Luckily, this controversy related to prion antibody treatment did not lead to fatalities in humans affected with CJD. However, Alzheimer's disease trials have led the death of individuals administered with therapeutic antibodies. This study led to unique discovery showing that anti-PrP antibodies led to neuronal allergenicity via different pathways but also highlights the key role played by microglia in causing the allergenic reaction. This study also emphasizes the need to include a screening 'allergenicity' step during development of therapeutic antibodies to avoid potential side-effects. Ethics approval and consent to participate Not applicable Consent for publication