Involvement of PrPC in kainate-induced excitotoxicity in several mouse strains

The cellular prion protein (PrPC) has been associated with a plethora of cellular functions ranging from cell cycle to neuroprotection. Mice lacking PrPC show an increased susceptibility to epileptic seizures; the protein, then, is neuroprotective. However, lack of experimental reproducibility has led to considering the possibility that other factors besides PrPC deletion, such as the genetic background of mice or the presence of so-called “Prnp flanking genes”, might contribute to the reported susceptibility. Here, we performed a comparative analysis of seizure-susceptibility using characterized Prnp+/+ and Prnp0/0 mice of B6129, B6.129, 129/Ola or FVB/N genetic backgrounds. Our study indicates that PrPC plays a role in neuroprotection in KA-treated cells and mice. For this function, PrPC should contain the aa32–93 region and needs to be linked to the membrane. In addition, some unidentified “Prnp-flanking genes” play a role parallel to PrPC in the KA-mediated responses in B6129 and B6.129 Prnp0/0 mice.

Although the role of the cellular form of the prion protein (PrP C ) in living organisms has been intensively studied, a clear consensus concerning the physiological functions of this protein is still elusive and controversial. To date, existing evidence implicates PrP C in numerous distinct cellular processes, including cell proliferation and differentiation 1,2 , copper homeostasis 3,4 , oxidative stress 5 and cell signaling 6,7 , among others.
The generation of different transgenic Prnp 0/0 mice (mixed B6129 Prnp Zrchl/Zrchl or co-isogenic 129/ Ola Prnp Edbg/Edbg backgrounds) in the early 1990s did not reveal any relevant phenotypic alteration of the mutant mice 8,9 . However, subsequent studies identified an abundance of phenotypic alterations (e.g., 10 ), including depressive-like behaviour 11 , cognitive deficits 12 , age-dependent behavioural abnormalities 13 , altered olfaction 14 , peripheral myelin deficits 15 , altered circadian rhythms 16 and an increased susceptibility to oxidative stress 5 and glutamate excitotoxicity [17][18][19] . Indeed, different laboratories have described in these strains and in congenic B6.129 Prnp Zrchl/Zrchl (B6129 mice backcrossed with C57BL/6 for several generations) an enhanced sensitivity to seizures after the administration of epileptogenic drugs such as kainic acid (KA), N-methyl-d-aspartic acid (NMDA), pilocarpine and pentylenetetrazol (PTZ), suggesting a neuroprotective role of the protein against excitotoxic insults (e.g., [17][18][19][20][21]. However, some studies suggest that PrP C is not involved in KA-mediated excitotoxicity and that the observed differences are likely egy to generate null mutations in mice is a powerful research tool to reveal in vivo the function of a single protein, as derived phenotypic alterations are usually attributed to the deleted gene. Nevertheless, behavioural alterations observed in null-mutant mice could result in some cases from the genetic background (see introduction, see also 26,49,50 ). In order to determine a possible influence of non-Prnp genes in the susceptibility to KA-induced seizures, we compared the epileptic response in B6129 (n = 20), 129/ Ola Prnp Edbg/Edbg (n = 9) and FVB/N Prnp 0/0 (n = 7) and wild type mice (B6129, n = 16; 129/Ola Prnp Edbg/Edbg (n = 11); FVB/N (n = 8) (Fig. 1a and Supplementary Tables 1-6). A multiple administration protocol, consisting of three intraperitoneal injections of KA (10 mg/kg body weight) at 30 min intervals, was used in this experiment. Seizure intensity was analysed during the 4 hours after the first KA injection and scored as indicated in Methods. For an easier comparative analysis, data from grades I to IV were grouped together.
After the behavioural study, mice were numbered and kept in separate boxes until histological studies. Percentages of the different strains of mice reaching each stage were represented (Fig. 1a). As indicated  Supplementary Tables 1-6), all mice achieved stages I-IV, developing hypoactivity and immobility shortly after the first injection. When comparing wild type vs Prnp-null mice, significant differences were observed in the percentage of mice reaching stages V and VI, with the B6129 Prnp Zrchl/Zrchl and 129/Ola Prnp Edbg/Edbg much more susceptible to seizures than their respective Prnp +/+ controls ( Fig. 1a and Supplementary Movies 1 and 2). In both groups, a high percentage of Prnp-null mice (100% and 75% respectively) developed loss of balance control and intermittent whole-body convulsions. More severe seizure activity, consisting of continuous seizures and/or 'popcorn' bouncing behaviour (stage VI) was also reported, leading to the death of 20% of the B6129 Prnp Zrchl/Zrchl mice tested. In contrast, no B6129 Prnp +/+ or 129/Ola Prnp +/+ mice reached stage VI or died during experiments.
Taken together, these results support the notion that the genetic background of the different Prnp 0/0 mice plays a role in KA susceptibility 22,23 , with the FVB/N background more prone to seizures than 129/ Ola or the mixed genetic background B6129 26,49 . The susceptibility of the FVB background 49,50 probably masks participation of PrP C at this specific KA-dosage.
Decreased seizures in B6.129 Prnp Zrchl/Zrchl compared to B6129 Prnp Zrchl/Zrchl mice. In a second set of experiments we used B6.129 Prnp Zrchl/Zrchl mice generated after crossing the original B6129 Prnp Zrchl/Zrchl mice with C57BL/6 mice over several generations. In these experiments, B6129 Prnp Zrchl/Zrchl mice were purchased directly from EMMA (Monterotondo, Italy) and they carried approximately ≈ 64% of C57BL/6 microsatellite markers (Charles River Laboratories) with ≈ 46% of non-C57BL/6 markers (129 in origin). We backcrossed these Prnp 0/0 mice with C57BL/6 mice over several generations (8)(9)(10) to reduce the non-C57BL/6 microsatellite markers to ≈ 6.5-7%. The presence of C57BL/6 or 129 markers in all phenotypes used in the present study was determined by the genetic testing service at Charles River Laboratories. In the test, 110 microsatellite markers at approximately 15 cM intervals were analyzed, spreading across the 19 autosomes and the X chromosome, which distinguishes among 129 microsatellite markers ranging from 92 to 94% of C57BL/6 (in B6.129 Prnp Zrchl/Zrchl mice). Thus B6129 and B6.129 Prnp Zrchl/Zrchl mice were treated with 8 mg/kg b.w. (B6129 n = 8 and B6.129 Prnp Zrchl/Zrchl n = 6) or 10 mg/kg b.w (B6129 n = 8 and B6.129 Prnp Zrchl/Zrchl n = 9) of KA and the number of seizures was determined during the first 30 minutes, from 30 to 60 minutes and from 60 to 180 minutes postinjection. In this experiment each mouse received a KA injection every 30 minutes as above. Data revealed a relevant number of seizures in B6129 Prnp Zrchl/Zrchl Prnp 0/0 mice compared with B6.129 Prnp Zrchl/Zrchl Prnp 0/0 (KA 8 mg/kg P = 0.0021; KA 10 mg/kg P = 0.0086; Mann-Whitney U test confidence interval (CI) = 95% ( Fig. 1d and Supplementary Movie 1). In addition, the differences between the numbers of seizures can be observed in both KA concentrations (8 and 10 mg/kg b.w). These data indicate that although PrP C plays a role in the increased susceptibility to KA as demonstrated above, the 129 associated genes also play a role in the observed results.

Chemical elimination of the GPI domain and of PrP C reduces neuroprotection to KA in
Prnp-transfected N2a cells. The GPI anchor in PrP C , as in other GPI-proteins, is not only necessary for the stability and attachment of the protein to the cell surface, but also for its association to specialized membrane microdomains (e.g., lipid rafts), its intracellular traffic and signal transduction events. In addition, the GPI group has been suggested as playing a role in prion disease toxicity, as transgenic mice expressing secreted forms of PrP C lacking its GPI-moiety showed no clinical symptoms despite accumulating PrP Sc in plaques 51 . Because in our microsatellite analysis the C57BL/10-PrP GPIless mice (Tg44 +/+ kindly provided by Dr. B. Chesebro) contained non-B10 regions in chromosome 3 and in chromosome 2 flanking the Prnp locus, we decided to check whether the degradation of the GPI binding domain of the PrP C leading to a decreased PrP C in the plasma membrane could overcome the neuroprotective function of PrP C (Fig. 3a-c). In addition, this approach would determine whether the neuroprotective effect against KA treatment of PrP C takes place at the membrane or intracellularly. Thus, N2a cells were treated either with Phospholipase C (PLC) enzyme or Glimepiride, a sulphonylurea approved for the treatment of diabetes mellitus, inducing the release of PrP C from the surface of prion-infected neuronal cells, which releases PrP C from the surface of neuronal cells 52 (see methods for details). Decreased levels of PrP C after treatments were detected with western blot (pcDNA = 0.39 ± 0.021; pcDNA-PrP C = 0.91 ± 0.025; pcDNA-PrP C + PLC = 0.45 ± 0.013; pcDNA-PrP C + Gli = 0.496 ± 0.014) ( Fig. 3a and Supplementary Fig. 2) or with immunocytochemistry in non-permeabilized cells (Fig. 3b) (pcDNA CTCF = 1.63 ± 0.09; pcDNA-PrP C CTCF = 6.44 ± 0.12. pcDNA-PrP C + Gli CTCF = 2.82 ± 0.20; pcDNA-PrP C + PLC CTCF = 2.063 ± 0.15. pcDNA vs pcDNA- Enhanced cell death in the CA3 hippocampal region ΔF35 and ΔC4 mice compared to B6129 mice. Prnp 0/0 Δ F35 (Δ F35) and Prnp 0/0 Δ C4 (Δ C4) mice ( Fig. 4a) were generated some years ago by nuclear injections of constructs into fertilized oocytes from B6129 Prnp Zrchl/Zrchl mice 53 . The number of copies of the construct was estimated as 25 for Δ C4 and 70 for Δ F35 mice. However, brain extracts obtained from these mice revealed similar levels of PrP− Δ C4 ( Prnp +/+ = 0.755 ± 0.075; PrP− Δ C4 = 0.81 ± 0.03; P = 0.4; CI = 95%, Mann-Whitney U test) to wild-type but lesser amounts of PrP− Δ F35 than wild-type ( Prnp +/+ = 0.91 ± 0.01; PrP− Δ F35 = 0,70 ± 0.021; P = 0.0048, CI = 95%, Mann-Whitney U test) ( Fig. 4b and Supplementary Fig. 3). Nevertheless only Δ F35 mice showed cerebellar degeneration at around 60 days of life 53 (Supplementary Fig. 4). Irrespective of the cell type, the expression of the truncated form may induce per se cell death in vivo 54 as well as in vitro 55 . Thus in the next experiments we treated these mice with 8 mg/kg b.w. of KA following the above-mentioned protocol. Behavioural results reported a similar evolution between Δ C4 ( n = 11) and Δ F35 ( n = 15)  Acute transfection of pcDNA-PrP ∆CD but not pcDNA-PrP ∆F35 protects N2a cells from KA excitotoxicity in vitro. In order to corroborate in vitro the participation of the OR in PrP C -mediated neuroprotection to KA, we performed a viability assay using N2a cells. Cells were transfected with vectors encoding either the full length of Prnp or two truncated forms lacking CD (PrP ∆CD , residues 95-133), which bridge the flexible amino proximal tail and the globular carboxy proximal domain, or else a longer deletion including the central domain (CD) plus the OR (PrP ∆F35 , residues 32-134) (Fig. 5a). After transfection, levels of PrP C and its truncated forms in transfected cells were checked with western blotting (Fig. 5b and Supplementary Fig. 5). In addition, the expression of the truncated forms was strictly modulated to avoid inducing cell death during the experiments. Transfected cells were treated overnight with 5 mM KA and further processed to WST-1 assays. Colorimetric WST-1 assay showed that PrP C and PrP ∆CD transfection increased cell culture viability after KA treatment, an effect that was not observed in PrP  pcDNA-PrP C vs pcDNA-PrP ∆F35 t = 5.42, mean diff. = 0.06, 95% CI of diff. = 0.021 to 0.09) (Fig. 5c). These results reinforce the idea of the OR domain participating in the described neuroprotective role of PrP C to excitotoxic damage in N2a cells.

Discussion
Role of PrP C in neuroprotection against KA. The diversity of phenotypic changes described in Prnp-knockout mice has hindered the study of the physiological function of PrP C10 . In fact the reported differences between Prnp +/+ and Prnp 0/0 mice have led to certain controversial results, especially in terms of electrophysiology and susceptibility to excytotoxic insults. Several pieces of evidence, including some from previous studies by our group using B6.129 Prnp Zrchl/Zrchl 2,18,21,40 , support the idea that Prnp-knockout mice are more susceptible to KA, NMDA and PTZ, exhibiting an enhanced epileptic response and neurotoxicity in the hippocampus when compared to wild type controls 17,19 . Contradictory results have been published by other groups, who described an elevated threshold for epileptiform activity in Prnp 0/0 hippocampal slices exposed to bicuculline, PTZ or zero-magnesium conditions 24 . Similar discrepancies have been found when analysing neurotransmission-associated parameters in mice devoid of PrP C34, 35,37,38 . Due to this lack of reproducibility between groups, the possibility that reported phenotypes could be attributed to external factors, such as different mice strains (e.g., 25 or 32 ) or experimental procedures (age of mice, KA concentration, treatments, etc.) must be strongly considered. Indeed, age-dependent loss of LTP in PrP C -null mice 56,57 has also been identified.
More recently, a comparative analysis between Prnp-knockout mouse strains has led to the proposal that the increased sensitivity to KA-induces seizures in Prnp-knockout mice is not associated with the absence of the protein but rather with inappropriate comparison with wild-type controls that differ in genetic background 22,23,58 . In this study, background influence in susceptibility to excitotoxic damage was corroborated by evaluating the onset of epileptic seizure of wild-type mice from B6129, B6.129,129/Ola and FVB/N strains. Our results show that FVB/N wild-type mice showed more seizures than B6129 and 129/Ola mice, corroborating previous studies 25,26,59 . In addition, 129/Ola wild-type showed increased expression of PrP C that may also reduce the number of grade VI seizures.
In our study we observed the relevant effects of intraperitoneal injection of KA in B6129 Prnp Zrchl/Zrchl Prnp 0/0 (with ≈ 46% 129 microsatellites), B6.129 Prnp Zrchl/Zrchl Prnp 0/0 (with ≈ 6-7% 129 microsatellites) and 129/Ola-Prnp 0/0 mice (with 100% 129 background) compared with the appropriate wild-type controls. Results indicate a decline in the epileptic seizures when low numbers of 129 microsatellites are present in B6.129 Prnp Zrchl/Zrchl Prnp 0/0 . These results point to polymorphisms of some of the "Prnp-flanking genes" as masking the participation of PrP C in neuroprotection, as reported in other processes 30 . Analysis of the 9 genes functions using gene ontology and Pubmed searches indicates that Prex1 60 , SIRPα 61 , Traf1 62 , Thbs1 63,64 , Rmdn3 65 , Tyro3 66,67 , Slc30a4 68 , Mertk 69 and B2m 70 are involved in neurotransmission, LTP and neuroprotection. Whether polymorphisms in these genes participate in these effects warrants further study. However, the phenotypic differences observed between 129/Ola Prnp +/+ and 129/Ola Prnp 0/0 mice could only be associated with Prnp-deletion, as no possible genetic variability exists between the two. These results differ from those recently published by Striebel and coworkers 23 who didn't observe PrP C -linked neuroprotection despite using the same mouse strain. These contradictory results may be due to KA-dosage or to minor differences in the experimental procedure. Furthermore, we must not forget that this neuroprotective property of PrP C is further supported by other in vivo models, such as transient knockdown of Prnp homologs in zebrafish 71 , and also some in vitro experimental approaches including PrP C downregulation and detachment of the plasma membrane in neuronal primary cultures and neuroblastoma cell lines ( 18-20 and present results).
In conclusion, our findings support the notion that PrP C is involved in neuroprotection to KA-induced seizures and excitotoxicity and that it actively participates in the increased epileptic response observed in mice devoid of PrP C . In parallel, as yet unknown factors associated with Prnp-flanking genes also affect KA susceptibility.

The neuroprotective function of PrP C depends on membrane anchoring. Our results indicate
that the neuroprotective function of PrP C against KA in N2a cells depends on membrane anchoring. The interaction of different regions of PrP C with plasma membrane through the GPI domain has been considered necessary to induce the clinical symptoms in GPI-negative transgenic mice (C57BL/10-PrP GPIless mice) 51,72 . In fact, the injection of antibodies directed to the α 1 and α 3 regions of the PrP C induces neurotoxic effects 73 . These results have also been corroborated in newly developed mice lacking the globular domain (FTgpi mice) 74 . In fact, these results corroborate previous observations 75 . Taken together, these studies suggest that the proximity of the flexible tail (N-terminal domain) to the plasma membrane triggers intracellular oxidative stress responses leading to cell death 74,76 . Under this scenario we can consider that our data reinforce the idea that the integrity of the N-terminal domain is mandatory for neuroprotection (see below) as well as the notion that membrane interaction is a necessary part of the neuroprotective function reported in vivo.

The neuroprotective function of PrP C is abolished in the absence of the OR of PrP C . Structural
analysis of PrP C architecture has determined different functionally relevant domains in this protein (see 77,78 for review). Besides the highly conserved hydrophobic domain, the flexible and unstructured N-terminus region includes a copper binding site consisting of four tandem repeats of the sequence PHGGGWGQ, which seems to be involved in the endocytic process of the protein 79 and copper homeostasis 3 . In fact a recent study indicates a relevant role of copper binding in the neuroprotective function of PrP C modulating NMDA receptor 48 . In our study we developed in vivo as well as in vitro tests to check the involvement of the OR domain as a mediator of PrP C neuroprotective function in a model of KA treatment. The behavioural and histological data presented here by Δ C4 and Δ F35 mice revealed increased cell death in the hippocampus of the KA injected mice compared to B6129 Prnp 0/0 (genetic background of these mice). In parallel, the CD is not involved in the neuroprotective functions of PrP C (at least in N2a cells) since their absence does not modify these properties when compared to full length PrP C , in contrast to the absence of the OR regions.
The overexpression of PrP− Δ F35 in a B6129 Prnp Zrchl/Zrchl Prnp 0/0 background leads to cerebellar neurodegeneration 53 , but not hippocampal degeneration 55 (Supplementary Fig. 4). In contrast, Δ C4 mice (with similar background) do not show cerebellar or hippocampal degeneration, but when subjected to controlled ischemia show significantly greater oxidative stress damage when compared to wild type mice 80 . This also happens when PrP− Δ F35 is overexpressed in HEK293 cells, leading to increased Caspase 3 activity in transfected cells and cell death 55 . Truncated forms of PrP C lacking the OR interfere with PrP C endocytosis via clathrin-coated vesicles and beta-cleavage of PrP C , respectively, thereby impairing the antioxidative functions of PrP C81,82 . Thus it is reasonable to consider that cells with an intrinsic deficit in oxidative stress homeostasis may also be more prone to KA treatment if the appropriate KA receptors are also expressed, as happens in the hippocampus [83][84][85] .
As indicated above, using antibody-mediated degeneration, Sonati et al., demonstrated that ligands directed to the α 1 and α 3 helices of the PrP C globular domain induce cerebellar cell death by activating oxidative stress that can be overcome by deletions in the OR region 73 . This also happens in FTgpi mice lacking the α 1-α 3 helix region of the PrP C 74 . In addition, mice lacking the Δ CD also reported white matter pathology and peripheral neuropathy 86 . Surprisingly, the reported degeneration of the Δ CD mice could be reversed by coexpression of PrP C lacking all octarepeats 86 . These data are in contrast to a recent study indicating that the antibody ICSM18 (recognizing aa143-153 of PrP C ) does not induce cell death 87 . In our experiments, we observed that cells transfected with PrP− Δ CD are able to overcome KA-mediated cell death as are those transfected with full length PrP C , in contrast to PrP− Δ F35 transfected N2a cells. Our in vitro experiments are different from the in vivo situation, since an effect of ligands in the CD regions is unlikely; rather, they suggest a parallel effect of KA excytotoxicity plus the homeostatic imbalance induced by the absence of the 32-93 region of the overexpressed PrP C . Despite the existing data, the precise mechanism underlying OR-dependent neuroprotection remains to be elucidated in the above-mentioned studies 73,87,88 .
In conclusion, our study dissects the effects of the intraperitoneal injection of various doses of KA in several Prnp mouse models and indicates that: i) PrP C plays a role in neuroprotection in KA-treated cells and mice; ii) for this role PrP C should be linked to the membrane; iii) polymorphisms of some unidentified "Prnp-flanking genes" play a parallel role to PrP C in the KA-mediated responses in B6129 and B6.129 Prnp Zrchl/Zrchl Prnp 0/0 mice; and iv) the absence of the aa32-93 region negatively affects the neuroprotective function of PrP C in KA-treated cells.

KA administration in mice and seizure analysis. Convulsive non-lethal seizures in mice were
induced by administration of KA in a multiple dose protocol. Fresh KA solution was prepared for each experiment. Animals were weighed and intraperitoneally injected with 8 or 10 mg/kg KA (b.w.) dissolved in 0.1 M PBS, pH 7.2. at 0 min, 30 min and 60 min. In parallel mice, 0.1 M PBS, pH 7.2 was injected as control (vehicle). Adult (2-3 months old) animals were used in all experiments except for B6129 Prnp Zrchl/Zrchl Prnp 0/0 PrP ∆C4 PrP ∆F35 and their respective controls, which were used at 5-7 weeks-old due to their described early lethality 53 .
After KA-injection mice were distributed in boxes (1-5 mouse/box) and the behaviour of the mice was recorded for 4 hours using a digital video camera (SONY DCR-HC30E Digital video camera). Seizure intensity was evaluated during the 4 hours after the first injection using the following criteria: grades I-II: hypoactivity and immobility, grades III-IV: hyperactivity and scratching, grade V: loss of balance control and intermittent whole-body convulsions and grade VI: continuous seizures and bouncing activity (commonly referred to as 'popcorn' behaviour). The characterization of seizure intensity was developed in two stages. Afterwards, a contingency table was generated by comparing both analyses and the final data were plotted. Grades I to IV were grouped together for better data representation, as all animals tested reached these four epileptic stages. The statistical analysis of the obtained data was performed using Mann-Whitney U non-parametric test using Prism 5.0c (Mac OsX, Grahpad). Data are presented as percentage or as mean ± standard error of the mean (S.E.M.). A value of ***P < 0.01 was considered statistically significant.
Densitometry and statistical processing of processed films. For quantification, developed films were scanned at 2400 × 2400 dpi (i800 MICROTEK high quality film scanner), and the densitometric analysis of the different PrP C bands was performed in each case using Quantity One Image Software Analysis (Biorad). Each densitometric value of PrP C and truncated form Δ C4, Δ F35 and Δ CD (0-255 gray scales) was normalized with the corresponding Tubulin densitometric values (0-255 gray scale). Three different experiments were used in each analysis unless specified. The statistical analysis of the obtained data was performed using Bonferroni post hoc test (Multiple comparison test) or Mann-Whitney U non-parametric test using Prism 5.0c (Mac OsX, Grahpad). Data are presented as mean ± standard error of the mean (S.E.M.). A value of **P < 0.05 was considered statistically significant.
Fluoro-Jade B staining. Mice were perfused with 4% paraformaldehyde dissolved in 0.1 phosphate buffer, pH 7.3 24 hours after the first KA injection, post-fixed overnight in the same fixative, and cryoprotected in 30% sucrose. 30 μ m-thick coronal brain sections were obtained in a freezing microtome (Leica, Wetzlar, Germany). Sections containing dorsal hippocampus (Bregma = − 1.2 to − 1.9 90 ) were rinsed for 2 h in 0.1 M Tris, pH 7.4, mounted and air dried at room temperature overnight. The next day, sections were pre-treated for 3 min in absolute ethanol, followed by 1 min in 70% ethanol and 1 min in distilled water. They were then oxidized in a solution of 0.06% KMnO4 for 15 min. After three rinses of 1 min each in distilled water, the sections were incubated for 30 min in a solution of 0.001% Fluoro-Jade B (Chemicon) containing 0.01% of DAPI (Sigma) in 0.1% acetic acid. The slides were rinsed in deionized water for 3 min each, dried overnight, cleared in xylene, cover-slipped with Eukitt (Merck, Darmstadt, Germany) and examined using an Olympus (Hamburg, Germany) BX61 epifluorescence microscope. The statistical analysis of the obtained data was performed using Bonferroni post hoc test (Multiple Scientific RepoRts | 5:11971 | DOi: 10.1038/srep11971 comparison test) using Prism 5.0c (Mac OsX, Grahpad). Data are presented as mean ± standard error of the mean (S.E.M.). A value of ***P < 0.01 was considered statistically significant.
Histology and immunofluorescence. For histology, mice were perfused with phosphate buffered 4% paraformaldehyde, pH 7.3 24 hours after the first KA injection, post-fixed overnight in the same fixative, and cryoprotected in 30% sucrose as above. A freezing microtome (Leica, Wetzlar, Germany) was used to obtain 30 μ m-thick coronal sections, which were rinsed in 0.1 M PBS before 1 hour's incubation at room temperature in 0.1 M PBS containing 0.2% gelatin, 10% normal goat serum, 0.2% glycine, and 0.2% Triton X-100. Sections were then incubated overnight at 4 °C with indicated primary antibodies. After washing in 0.1 M PBS containing 0.2% Triton X-100, sections were incubated with goat anti-rabbit Alexa Fluor 568-tagged secondary antibody (1:200 diluted; Molecular Probes, Eugene, OR, USA), washed in 0.1 M, PBS and mounted in Fluoromount (Vector Labs, Burlingame, CA, USA). Immunohistochemical controls, which included omission or substitution of primary anti-GFAP antibody by either normal rabbit serum prevented immunostaining. For quantification of GFAP-positive astrocytes in the stratum radiatum of the dorsal hippocampal region, immunoreacted sections (5 sections of each mouse, n = 5 mice per genotype) were photodocumented with an Olympus BX61 fluorescence microscope equipped with a cooled DP12L camera. Photomicrographs were obtained using a 40X objective with identical time exposure (100-150 ms) between preparations from each wild-type and respective knockout mouse. No modifications were applied to the obtained pictures. Numbers of GFAP-expressing cells were determined by counting positive cells in five frames (250 × 200 μ m) corresponding to the hippocampal CA1-3 regions of five mice of each genotype. Data were expressed as mean ± standard error of the mean (S.E.M). The statistical analysis of the obtained data was performed using Mann-Whitney U non-parametric test using Prism 5.0c (Mac OsX, Grahpad). A value of P < 0.01 was considered statistically significant.

RT-qPCR.
Total RNA from hippocampal samples obtained from treated (6 h after KA-administration) and non-treated mice was purified with the mirVana isolation kit (Ambion, Austin, TX, USA) and used to make the single-stranded cDNAs required as templates for the RT-qPCR amplification. Sets of primers used in this study were: for TNFα 5′ -AGCAAACCACCAAGTGGAGGA-3′ and 5′ -GCTGGCACCACTAGTTGGTTGT-3′ ; and for ILβ 5′ -TTGTGGCTGTGGAGAAGCTGT-3′ and 5′ -AACGTCACACACCAGCAGGTT-3′ . The reaction was performed with the Roche LightCycler 480 detector, using 2x SYBR Green Master Mix (Roche) as reagent, as indicated by the manufacturer. Amplification protocol consisted of a denaturation-activation cycle (95 °C for 10 min) followed by 40 cycles of denaturation-annealing-extension (95 °C, 15 sec; 60 °C, 40 sections; 72 °C, 5 sec; 98 °C, continuous). LightCycler 480 software was used for mRNA quantification. The data were analysed using the ΔΔCt method, which provides the target gene expression values as fold changes in the problem sample compared with a calibrator sample. Both problem and calibrator samples were normalized by the relative expression of a housekeeping gene (glyceraldehyde-3-phosphate dehydrogenase [GAPDH]). These analyses developed 3 different samples. The statistical analysis of the obtained data was performed using Mann-Whitney U non-parametric test using Prism 5.0c (Mac OsX, Grahpad). Data are presented as mean ± standard error of the mean (S.E.M.). A value of ***P < 0.01 was considered statistically significant.
Cell culture and treatments. The murine neuroblastoma cell line Neuro2a (N2a) expressing low levels of PrP C was grown at 37 °C, 5.5% CO2 in Dulbecco's modified Eagle's medium (DMEM) supplemented with 4.5 g/L glucose, 10% fetal bovine serum (FBS) and antibiotics (Invitrogen-Life Technologies, Barcelona, Spain). For vector transfection, cells were plated at 1 × 10 5 cells/well in a 24-well plate and transiently transfected the next day using Lipofectamine 2000 reagent in Optimem medium, as indicated by the manufacturer (Invitrogen-Life Technologies). Four hours after transfection, cells were washed and the medium was replaced with DMEM containing 10% FBS. PrP C expression after transfection was checked with western blot analysis. In a first set of experiments, N2a cell cultures were transiently transfected with pcDNA and pcDNA-PrP C and maintained in vitro at 37 °C, 5.5% CO 2 . Cells were deprived for 24 h and treated with Glimepiride 20 μ M and 0.2 u/ml PLC (final concentration). We used both treatments since some reports indicate that Glimepiride may act on ATP-dependent K+ channels that may also affect cell survival 91 . One hour later, 5 mM KA dissolved in 0.1 M PBS was added to the media. After treatment, cultures were rinsed twice in KA-free culture medium, and cell viability was determined with WST-1 viability assay (see below). In parallel, non-permeabilized cells were processed to PrP C detection by immunofluorescence using the 6H4 antibody and a goat anti-mouse Alexa Fluor 488-tagged secondary antibody (1:200 diluted; Molecular Probes). For fluorescence quantification of cell-bound Alexa Fluor 488, immunoreacted cultures (n = 10 per experimental group) were photodocumented with an Olympus BX61 + DP12L camera. Photomicrographs were obtained using a 20X objective with identical time exposure (250-300 ms) for preparations from each group. No modifications were applied to the obtained pictures. Fluorescence intensity was determined using ImageJ by measuring the corrected total cell fluorescence (CTCF) as: CTCF = integrated density -(area of selected N2a mesured cells x mean fluorescence of background). Data were expressed as mean ± standard error of the mean (S.E.M). The statistical analysis of the obtained data in these experiments was performed using Bonferroni post hoc test (Multiple comparison test) using Prism 5.0c (Mac OsX, Grahpad). A value of P < 0.05 was considered statistically significant. These experiments were repeated four times. In a second set of experiments, N2a cells were transfected with pcDNA3.1, pcDNA3.1-PrP C , pcDNA3.1-PrP ∆CD and pcDNA3.1-PrP ∆F35 . KA (5 mM) treatment was carried out on serum-deprived cells 24 h after transfection. Cell viability was determined using a commercially available WST-1-based assay. Cell cultures were incubated with WST-1 reagent for 2 hours. Then absorbance at 450 nm was measured in a multiwell plate reader (Merck ELISA System MIOS). Data were normalized with A450 in untreated controls. pcDNA3.1-PrP C , pcD-NA3.1-PrP ∆CD and pcDNA3.1-PrP ∆F35 were kind gifts from Prof. D. Harris (Boston University) and Prof. A. Aguzzi (University Hospital of Zurich). These experiments were repeated five times. The statistical analysis of the obtained data in these experiments was performed using Bonferroni post hoc test (Multiple comparison test) using Prism 5.0c (Mac OsX, Grahpad). Data are presented as mean ± standard error of the mean (S.E.M.). Values of **P < 0.05 and ***P < 0.01 were considered statistically significant.