5-Methylcytosine and 5-Hydroxymethylcytosine in Scrapie-Infected Sheep and Mouse Brain Tissues

Scrapie is a neurodegenerative disorder belonging to the group of transmissible spongiform encephalopathies or prion diseases, which are caused by an infectious isoform of the innocuous cellular prion protein (PrPC) known as PrPSc. DNA methylation, one of the most studied epigenetic mechanisms, is essential for the proper functioning of the central nervous system. Recent findings point to possible involvement of DNA methylation in the pathogenesis of prion diseases, but there is still a lack of knowledge about the behavior of this epigenetic mechanism in such neurodegenerative disorders. Here, we evaluated by immunohistochemistry the 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC) levels in sheep and mouse brain tissues infected with scrapie. Expression analysis of different gene coding for epigenetic regulatory enzymes (DNMT1, DNMT3A, DNMT3B, HDAC1, HDAC2, TET1, and TET2) was also carried out. A decrease in 5mC levels was observed in scrapie-affected sheep and mice compared to healthy animals, whereas 5hmC displayed opposite patterns between the two models, demonstrating a decrease in 5hmC in scrapie-infected sheep and an increase in preclinical mice. 5mC correlated with prion-related lesions in mice and sheep, but 5hmC was associated with prion lesions only in sheep. Differences in the expression changes of epigenetic regulatory genes were found between both disease models, being differentially expressed Dnmt3b, Hdac1, and Tet1 in mice and HDAC2 in sheep. Our results support the evidence that DNA methylation in both forms, 5mC and 5hmC, and its associated epigenetic enzymes, take part in the neurodegenerative course of prion diseases.


Introduction
Epigenetics is a field of study focused on heritable changes in gene activity or function that are not associated with any change in the DNA sequence itself [1]. One of the most studied epigenetic modifications is DNA methylation, which consists of the covalent addition of a methyl group to the C-5 position of the nucleobase cytosine to form 5-methylcytosine (5mC) [2]. In mammals, this DNA methylation process occurs predominantly at cytosine residues within CpG dinucleotides [3]. Moreover, multiple forms of DNA methylation have been identified including 5mC, its hydroxylated derivative 5-hydroxymethylcytosine (5hmC), and its ensuing oxidation products 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC) [4]. Among these forms, 5mC and 5hmC are relatively stable and abundant in mammalian genomes [5], whereas 5fC and 5caC are rarer and can be transiently removed [6].
Different enzymes act as mediators of the DNA methylation process. DNA methyltransferases (DNMTs) are responsible for producing 5mC by covalently adding a methyl find possible associations between the brain levels of 5mC and 5hmC with the expression levels of genes encoding DNA methylation enzymes and with prion-related lesions. Figure 1 shows the PrP Sc deposition and vacuolation of the analyzed brain areas in Tg338 mice. The obex, mesencephalon, thalamus, and hypothalamus were the areas with higher PrP Sc deposits. Regarding vac vacuolation, obex, mesencephalon, hypothalamus, and septal area/striatum were the areas that showed higher levels of spongiosis. the course of these diseases and, to the best of our knowledge, no study has yet explored the role of 5hmC in prion pathology. The present study aimed to evaluate the 5mC and 5hmC brain profiles in a transgenic scrapie mouse model and sheep naturally infected with scrapie. Moreover, expression analysis of the genes coding different enzymes involved in the DNA methylation process was performed along with a correlation analysis in order to find possible associations between the brain levels of 5mC and 5hmC with the expression levels of genes encoding DNA methylation enzymes and with prion-related lesions. Figure 1 shows the PrP Sc deposition and vacuolation of the analyzed brain areas in Tg338 mice. The obex, mesencephalon, thalamus, and hypothalamus were the areas with higher PrP Sc deposits. Regarding vac vacuolation, obex, mesencephalon, hypothalamus, and septal area/striatum were the areas that showed higher levels of spongiosis. The scrapie characterization of the sheep used in this study was performed previously in an earlier publication from our group [34].

Gene Expression Profile in Tg338 Mice
The expression profile of seven genes described to be involved in epigenetic regulation (Dnmt1, Dnmt3a, Dnmt3b, Hdac1, Hdac2, Tet1, and Tet2) was analyzed by quantitative real-time PCR (RT-qPCR) in the thalamus of Tg338 mice, a transgenic scrapie mouse model expressing a transgenic VRQ allele of the ovine PRNP gene under the ovine PrP promoter [35,36]. The mice were divided into four groups of study: clinical and preclinical mice infected with scrapie and their corresponding controls, designated as clinical control and preclinical control, respectively.
In all the studied genes, preclinical mice showed lower expression levels compared to their controls. This decreased expression was statistically significant in Dnmt3b (p < 0.05) and Tet1 (p < 0.05), and showed a trend to signification (p = 0.1) in Dnmt1, Dnmt3a, Hdac2, and Tet2. Moreover, the expression of these genes in preclinical mice was also significantly lower than the one observed in clinical mice, the difference being statistically significant for Dnmt3b (p < 0.05) and Hdac1 (p < 0.05), and with a trend to signification for Dnmt1 (p = 0.1). On the contrary, no significant differences were observed between clinical mice and their controls ( Figure 2). Ct values of the RT-qPCR analysis are presented in Supplementary Table S1. levels of PrP Sc deposits. Data are shown as mean values ± SEM in the following brain areas: OBEX, cerebellum (CBL), mesencephalon (MES), hippocampus (HC), thalamus (T), hypothalamus (HT), parietal cortex (PC), septal area/striatum (SA), and frontal cortex (FC).
The scrapie characterization of the sheep used in this study was performed previously in an earlier publication from our group [34].

Gene Expression Profile in Tg338 Mice
The expression profile of seven genes described to be involved in epigenetic regulation (Dnmt1, Dnmt3a, Dnmt3b, Hdac1, Hdac2, Tet1, and Tet2) was analyzed by quantitative real-time PCR (RT-qPCR) in the thalamus of Tg338 mice, a transgenic scrapie mouse model expressing a transgenic VRQ allele of the ovine PRNP gene under the ovine PrP promoter [35,36]. The mice were divided into four groups of study: clinical and preclinical mice infected with scrapie and their corresponding controls, designated as clinical control and preclinical control, respectively.
In all the studied genes, preclinical mice showed lower expression levels compared to their controls. This decreased expression was statistically significant in Dnmt3b (p < 0.05) and Tet1 (p < 0.05), and showed a trend to signification (p = 0.1) in Dnmt1, Dnmt3a, Hdac2, and Tet2. Moreover, the expression of these genes in preclinical mice was also significantly lower than the one observed in clinical mice, the difference being statistically significant for Dnmt3b (p < 0.05) and Hdac1 (p < 0.05), and with a trend to signification for Dnmt1 (p = 0.1). On the contrary, no significant differences were observed between clinical mice and their controls ( Figure 2). Ct values of the RT-qPCR analysis are presented in Supplementary Table S1. In order to test if the lack of statistical significance was due to age differences within the group that could modify gene expression, we calculated the correlation between gene expression Ct values and age in controls. No significant Pearson correlation values were found (p > 0.05) for any of the analyzed genes. In order to test if the lack of statistical significance was due to age differences within the group that could modify gene expression, we calculated the correlation between gene expression Ct values and age in controls. No significant Pearson correlation values were found (p > 0.05) for any of the analyzed genes.

Gene Expression Profile in Sheep
In sheep, the expression profile of six genes (DNMT1, DNMT3A, DNMT3B, HDAC1, HDAC2, and TET1) was also analyzed by RT-qPCR in the thalamus region. The sheep were divided into three groups of study: preclinical scrapie-infected sheep, clinical scrapieinfected sheep, and healthy control sheep.
No significant differences were observed in the expression levels of DNMT1, DNMT3A, and DNMT3B genes between the different groups of animals. However, a trend to signification (p = 0.1) between control and preclinical sheep, and between control and clinical sheep was observed in TET1, whose expression levels, just like in mice, were lower in clinical and preclinical sheep compared to the control ones. In addition, downregulation of HDAC2 was observed in preclinical sheep compared to controls and clinical animals (p < 0.05). These expression trends in the HDAC2 gene were like those observed in preclinical Tg338 mice. A downregulation of HDAC1 expression was also observed in clinical scrapie sheep compared to controls (p = 0.1), in contrast with the increased expression displayed by the clinical Tg338 mice (Figure 3). Ct values of the RT-qPCR analysis are shown in Supplementary Table S2. Similar to mice, gene expression levels did not correlate with age in controls.
pie-infected sheep, and healthy control sheep.
No significant differences were observed in the expression levels of DNMT1, DNMT3A, and DNMT3B genes between the different groups of animals. However, a trend to signification (p = 0.1) between control and preclinical sheep, and between control and clinical sheep was observed in TET1, whose expression levels, just like in mice, were lower in clinical and preclinical sheep compared to the control ones. In addition, downregulation of HDAC2 was observed in preclinical sheep compared to controls and clinical animals (p < 0.05). These expression trends in the HDAC2 gene were like those observed in preclinical Tg338 mice. A downregulation of HDAC1 expression was also observed in clinical scrapie sheep compared to controls (p = 0.1), in contrast with the increased expression displayed by the clinical Tg338 mice (Figure 3). Ct values of the RT-qPCR analysis are shown in Supplementary Table S2. Similar to mice, gene expression levels did not correlate with age in controls.

5-Methylcytosine and 5-Hydroxymethylcytosine Brain Profiles of Tg338 Mice and Sheep
2.3.1. 5mC and 5hmC levels in Tg338 mice brains Immunohistochemistry was performed in CNS tissue sections from Tg338 mice for the detection of 5mC and 5hmC. Nine brain areas were studied in the four groups of mice: the frontal cortex, parietal cortex, thalamus, hypothalamus, hippocampus, mesencephalon, cerebellum, obex, and septal area/striatum.
In all the studied brain areas, 5mC and 5hmC displayed intranuclear staining in neurons and glial cells.
Regarding 5mC levels, clinical and preclinical mice displayed similar levels across all the studied brain areas (Figure 4). The same trend was observed between preclinical mice and their respective controls (Figure 4). The obex was the only area that showed a significant difference, namely a decrease in 5mC levels in clinical compared to control mice (p < 0.05) (Figures 4 and 5). To determine whether the lack of significance between the different groups was a consequence of differences in age at the sacrifice, we calculated Pearson's correlation between age and immunohistochemical parameters in the animals of  Immunohistochemistry was performed in CNS tissue sections from Tg338 mice for the detection of 5mC and 5hmC. Nine brain areas were studied in the four groups of mice: the frontal cortex, parietal cortex, thalamus, hypothalamus, hippocampus, mesencephalon, cerebellum, obex, and septal area/striatum.
In all the studied brain areas, 5mC and 5hmC displayed intranuclear staining in neurons and glial cells.
Regarding 5mC levels, clinical and preclinical mice displayed similar levels across all the studied brain areas (Figure 4). The same trend was observed between preclinical mice and their respective controls ( Figure 4). The obex was the only area that showed a significant difference, namely a decrease in 5mC levels in clinical compared to control mice (p < 0.05) (Figures 4 and 5). To determine whether the lack of significance between the different groups was a consequence of differences in age at the sacrifice, we calculated Pearson's correlation between age and immunohistochemical parameters in the animals of control groups, finding only a significant positive correlation (p < 0.05) in the cerebellum (r = 0.68) and hypothalamus (r = 0.70).
control groups, finding only a significant positive correlation (p < 0.05) in the cerebellum (r = 0.68) and hypothalamus (r = 0.70).  A different pattern was observed for 5hmC. Clinical mice showed similar levels to their controls in all studied brain areas ( Figure 6). In contrast, 5hmC levels were increased in preclinical mice compared to their controls and the clinical mice ( Figure 6). This increase was significant in some of the analyzed brain areas, including in the parietal cortex and cerebellum between preclinical and control mice (p < 0.05) and in the thalamus (p < 0.05), hypothalamus (p < 0.01), parietal cortex (p < 0.01), and cerebellum (p < 0.05) between preclinical and clinical mice (Figures 6 and 7). Hydroxymethylation did not display a significant correlation with age in any of the analyzed areas (Pearson's correlation p > 0.05).   A different pattern was observed for 5hmC. Clinical mice showed similar levels to their controls in all studied brain areas ( Figure 6). In contrast, 5hmC levels were increased in preclinical mice compared to their controls and the clinical mice ( Figure 6). This increase was significant in some of the analyzed brain areas, including in the parietal cortex and cerebellum between preclinical and control mice (p < 0.05) and in the thalamus (p < 0.05), hypothalamus (p < 0.01), parietal cortex (p < 0.01), and cerebellum (p < 0.05) between preclinical and clinical mice (Figures 6 and 7). Hydroxymethylation did not display a significant correlation with age in any of the analyzed areas (Pearson's correlation p > 0.05). A different pattern was observed for 5hmC. Clinical mice showed similar levels to their controls in all studied brain areas ( Figure 6). In contrast, 5hmC levels were increased in preclinical mice compared to their controls and the clinical mice ( Figure 6). This increase was significant in some of the analyzed brain areas, including in the parietal cortex and cerebellum between preclinical and control mice (p < 0.05) and in the thalamus (p < 0.05), hypothalamus (p < 0.01), parietal cortex (p < 0.01), and cerebellum (p < 0.05) between preclinical and clinical mice (Figures 6 and 7). Hydroxymethylation did not display a significant correlation with age in any of the analyzed areas (Pearson's correlation p > 0.05).

5mC and 5hmC Levels in Scrapie Sheep Brains
In sheep CNS tissue sections, immunohistochemistry was performed to detect 5mC and 5hmC in eight areas: the frontal cortex, basal ganglia cortex, basal ganglia, parietal cortex, thalamus, hippocampus, mesencephalon, and obex.
Both 5mC and 5hmC showed nuclear and perinuclear staining in neurons and glial cells. The levels of 5mC negatively correlated with age in the thalamus of control sheep (r = −0.95, p = 0.02). No significant correlations were found in any other areas for 5mC or 5hmC.
Both 5mC and 5hmC showed nuclear and perinuclear staining in neurons and glial cells. The levels of 5mC negatively correlated with age in the thalamus of control sheep (r = −0.95, p = 0.02). No significant correlations were found in any other areas for 5mC or 5hmC.
Interestingly, 5hmC displayed the same pattern as 5mC. Contrary to the changes found in mice, no differences were observed in the levels of 5hmC between clinical and preclinical sheep (Figure 10), and lower levels of 5hmC were found in both clinical and preclinical sheep in comparison to healthy animals ( Figure 10). This decrease in 5hmC levels was significant in the mesencephalon (p < 0.01), obex (p < 0.05), thalamus (p < 0.01), and parietal cortex (p < 0.01) of clinical animals (Figures 10 and 11), and also in the mesencephalon (p < 0.001), obex (p < 0.01), thalamus (p < 0.01), and parietal cortex (p < 0.01) of preclinical animals (Figures 10 and 11). Interestingly, 5hmC displayed the same pattern as 5mC. Contrary to the changes found in mice, no differences were observed in the levels of 5hmC between clinical and preclinical sheep (Figure 10), and lower levels of 5hmC were found in both clinical and preclinical sheep in comparison to healthy animals ( Figure 10). This decrease in 5hmC levels was significant in the mesencephalon (p < 0.01), obex (p < 0.05), thalamus (p < 0.01), and parietal cortex (p < 0.01) of clinical animals (Figures 10 and 11), and also in the mesencephalon (p < 0.001), obex (p < 0.01), thalamus (p < 0.01), and parietal cortex (p < 0.01) of preclinical animals (Figures 10 and 11).  Interestingly, 5hmC displayed the same pattern as 5mC. Contrary to the changes found in mice, no differences were observed in the levels of 5hmC between clinical and preclinical sheep (Figure 10), and lower levels of 5hmC were found in both clinical and preclinical sheep in comparison to healthy animals ( Figure 10). This decrease in 5hmC levels was significant in the mesencephalon (p < 0.01), obex (p < 0.05), thalamus (p < 0.01), and parietal cortex (p < 0.01) of clinical animals ( Figures 10 and 11), and also in the mesencephalon (p < 0.001), obex (p < 0.01), thalamus (p < 0.01), and parietal cortex (p < 0.01) of preclinical animals (Figures 10 and 11). The data are presented as mean values ± SEM. * Significant difference between control and clinical sheep (p < 0.05); ** significant difference between control and clinical sheep (p < 0.01); ++ significant difference between control and preclinical sheep (p < 0.01); +++ significant difference between control and preclinical sheep (p < 0.001). Figure 10. Comparison of 5hmC Image J scores in the three sheep groups: control, preclinical, and clinical. The analyzed brain areas were the obex, cerebellum (CBL), mesencephalon (MES), thalamus (T), parietal cortex (PC), basal ganglia (BG), basal ganglia cortex (BGC), and frontal cortex (FC). The data are presented as mean values ± SEM. * Significant difference between control and clinical sheep (p < 0.05); ** significant difference between control and clinical sheep (p < 0.01); ++ significant difference between control and preclinical sheep (p < 0.01); +++ significant difference between control and preclinical sheep (p < 0.001). Figure 11. Immunostaining patterns of 5hmC in the mesencephalon, obex, and thalamus of preclinical, clinical, and control sheep.

Correlation between 5mC and 5hmC Levels
Pearson's correlation was used to evaluate the relationship between 5mC and 5hmC. Comparing all the analyzed brain areas, 5mC and 5hmC displayed a weak but significant positive correlation in Tg338 mice (r = 0.1432, p = 0.0459) and a stronger correlation in sheep (r = 0.6136, p = 1.023 × 10 −12 ), as shown in Figure 12a,b, respectively.  Figure 11. Immunostaining patterns of 5hmC in the mesencephalon, obex, and thalamus of preclinical, clinical, and control sheep.

Correlation between 5mC and 5hmC Levels
Pearson's correlation was used to evaluate the relationship between 5mC and 5hmC. Comparing all the analyzed brain areas, 5mC and 5hmC displayed a weak but significant positive correlation in Tg338 mice (r = 0.1432, p = 0.0459) and a stronger correlation in sheep (r = 0.6136, p = 1.023 × 10 −12 ), as shown in Figure 12a,b, respectively.

Correlation between 5mC and 5hmC Levels
Pearson's correlation was used to evaluate the relationship between 5mC and 5hmC. Comparing all the analyzed brain areas, 5mC and 5hmC displayed a weak but significant positive correlation in Tg338 mice (r = 0.1432, p = 0.0459) and a stronger correlation in sheep (r = 0.6136, p = 1.023 × 10 −12 ), as shown in Figure 12a,b, respectively.

Correlation between the Levels of 5mC and 5hmC and the Expression Levels of Epigenetic Regulatory Genes in Tg338 Mice and Sheep
In the thalamus region of Tg338 mice and sheep brains, some associations were observed between 5mC and 5hmC levels, and several genes were implicated in epigenetic regulation using Pearson's correlation. Most of these associations were related to 5hmC levels.

Correlation of 5mC and 5hmC Levels with Prion-Related Lesions
Spearman's correlation revealed some associations between 5mC, 5hmC, and prionrelated lesions when comparing all the analyzed brain areas in the full set of animals, especially in sheep.

Discussion
Although recent evidence suggests an involvement of DNA methylation in prion disease pathology, the knowledge about its specific functions and roles is still limited. In the current study, we assessed the 5mC and 5hmC immunohistochemical brain profiles and the expression levels of genes encoding DNA methylation enzymes in a murine model of scrapie disease (Tg338 mice) and sheep naturally infected with scrapie. Moreover, we analyzed possible associations between the 5mC and 5hmC levels with prion-related lesions and with the expression levels of the DNA methylation-related genes.
The levels of 5mC have been studied in other neurodegenerative diseases, such as AD. In the brain of a triple-transgenic mouse model of AD (3xTg-AD) [37] and the brain of preclinical AD patients [38], a decrease in this epigenetic modification was reported. Similar results have been observed in our study, where scrapie-infected sheep and mice showed similar patterns. Clinical and preclinical animals displayed comparable brain levels of 5mC, which were significantly lower in the obex of clinical mice and the mesencephalon, thalamus, parietal cortex, and obex of clinical and preclinical sheep. Furthermore, a negative association of 5mC levels with prion-related lesions was observed in mice and especially in sheep, in which a negative correlation of 5mC with PrP Sc accumulation and vacuolation was found, although the statistical significance of this correlation was lost when considering only the set of scrapie animals. This decrease in 5mC immunostaining is not related with the loss of cells due to prion toxicity and the consequent loss of nuclei because immunoreactivity was normalized taking into account the number of nuclei in each area. As a result, the decrease must be due to a real decrease in immunostaining inside the nucleus of the different cell populations. Although experimental animals used here displayed a similar age, small age differences at animal sacrifice could possibly have an effect on DNA methylation. Within the control group, age variability was found significantly correlated with 5mC in the thalamus of the control sheep. Nevertheless, the reduction in 5mC was still significant in the scrapie thalamus. In control mice, 5mC correlated with age in hypothalamus and cerebellum. We cannot discard the possibility that significant differences in these two specific areas of CNS between control and scrapie mice could have been observed in more homogenous groups. If the decline observed in 5mC levels is a trigger of the disease or a consequence of the disease-related neurodegenerative lesions, it is still unclear and warrants further investigation.
Several reports have shown distinctive levels of 5hmC in Alzheimer's and Parkinson's diseases with different trends between studies. In preclinical and clinical AD patients, different brain regions have been shown to display higher levels of 5hmC [20,38,39]. This increment was also observed in 3xTg-AD [37] and amyloid precursor protein (APP)/presenilin 1 (PS1) [40] AD mouse models and in the cerebellum of PD patients [22,41]. However, other studies point to a decrease in 5hmC in the entorhinal cortex and cerebellum of AD patients [42] and the brain of 3xTg-AD [43] and APP/PS1 [44] mice. In the present study, contrarily to 5mC, 5hmC levels displayed opposite results between sheep and mice. An increase in 5hmC levels was found in preclinical mice, whereas in clinical and preclinical sheep, 5hmC levels decreased in different CNS areas. Differences between the experimental and the natural model may be due to differences in the stage of disease in preclinical animals, whereas a controlled scrapie inoculation in mice allows sacrificing animals in a true preclinical stage clearly distinct from the clinical phase; preclinical sheep can be detected in a late stage, closer to the onset of symptoms.
Although no association was found between prion lesions and 5hmC in mice, in sheep, both 5mC and 5hmC correlated negatively with PrP Sc deposits and vacuolation. Similarly, these correlations were not observed in the set of scrapie animals, suggesting that the correlation is linked to the course of the disease but not to the degree of the lesion. In the natural model, 5hmC follows the same decrease trend as 5mC. This positive correlation between 5mC and 5hmC has been reported in other studies of patients with autism [45] and AD [20]. As opposed to a negative correlation that could be related to the increment of 5hmC as a consequence of the decline in 5mC levels, the concurrent decrease in 5mC and 5hmC suggests that 5hmC could have a specific role in prion disease pathology and not only act as a demethylation intermediate of 5mC.
Differential expression of genes encoding epigenetic regulatory enzymes was also observed. Dnmt3b interacts with Hdac1 during the establishment of DNA methylation patterns [46]. Moreover, along with its role in de novo DNA methylation, Dnmt3b also functions as a DNA dehydroxymethylase, able to directly convert 5hmC to an unmethylated cytosine (C) [47]. In our study, preclinical mice displayed downregulation of Dnmt3b whereas, in clinical mice, Dnmt3b expression returns to normal levels. This early downregulation was not associated with a decrease in 5mC in the thalamus. In sheep, neither DNMT3B expression nor the levels of 5mC were modified significantly between preclinical and clinical animals. These expression changes are neither related to an increase in 5hmC, on the contrary, a significant decrease in 5hmC was observed in both, mice and sheep infected with scrapie. Modification in gene expression could be related with other biological processes. In the adult brain, Dnmt3b is required for neurogenesis, facilitating the neuronal maturation in the hippocampus of adult mice [48], and it also plays a role in regulating object-place recognition memory [49]. Hdac1 is needed as well for neuronal differentiation of murine hippocampal neural stem cells [50] and is involved in DNA damage repair pathways and cognitive function [51], including fear extinction learning [52]. This latter brain function is compromised in schizophrenia patients [52]. Upregulation of HDAC1 has been found in the prefrontal cortex of patients with schizophrenia [53] and overexpression of this enzyme seems to ameliorate the fear extinction learning cognitive function in mice [52]. The upregulation of Hdac1 observed in clinical mice could be a compensatory mechanism trying to maintain balanced levels of 5mC and 5hmC, and at the same time, counteract the neurodegenerative effects produced by scrapie disease.
A recent study in mouse embryos showed that Hdac1 and Hdac2 were essential for maintaining correct DNA methylation patterns during preimplantation development. Deficiency of Hdac1 and Hdac2 in these embryos caused an increase in both forms of DNA methylation, 5mC and 5hmC [54]. Therefore, the upregulation of HDAC2 observed in clinical scrapie sheep could contribute to the decrease in 5mC and 5hmC levels. Upregulation of HDAC2 has also been observed in the brain of patients with AD [55] and PD [56] and Swiss albino old mice, and HDAC2 overexpression has been correlated with reduced recognition memory [57,58]. Inhibition of HDAC2, conversely, slows down AD progression through ameliorating amyloid beta-induced neuronal impairments in APP/PS1 mice [59], enhancing mitochondrial respiration and reducing the levels of neurotoxic amyloid beta peptides in induced pluripotent stem cell-derived neurons [60]. These findings suggest that upregulation of HDAC2, in addition to contributing to the decline of 5mC and 5hmC, may also be entailed in aggravating the neurodegenerative process. Further studies are essential for delving into the specific role of this enzyme in prion diseases and studying its potential as a therapeutic target.
On the other hand, Tet1, which was downregulated in preclinical mice and preclinical and clinical sheep, is one of the enzymes capable of the conversion of 5mC to 5hmC, as well as its further oxidation to 5fC and 5caC intermediates [61]. The relationship between Tet1 expression and 5hmC levels is somewhat controversial. Decreased levels of TET1 accompanied by increased levels of 5hmC have been observed in chondrocytes from patients suffering osteoarthritis due to the minor conversion of 5hmC to 5fC or 5caC [62]. Cancer-related studies, however, show an association between reduced TET1 expression levels and a decrease in 5hmC [63,64]. In addition, this enzyme has important functions in the adult brain beyond its role in DNA demethylation, being necessary for a functional axonmyelinic interface and a successful myelin repair [65] and also critical in the regulation of the neuroinflammation process, associating a downregulation of Tet1 with aberrant activation of inflammatory response pathways [66]. Interestingly, neuroinflammation is increased during the course of scrapie disease [67][68][69][70]. In addition, Tet1 knock-out mice show a deficiency in adult neurogenesis, spatial learning, and memory [71]. The downregulation of Tet1 during scrapie disease could have different functional implications regarding the regulation of 5hmC levels, which are positively correlated with the levels of Tet1 in mice and decrease when the expression of this gene is downregulated and could also be involved in the impairment of the myelin repair and neuroinflammation pathways and the disruption of the neurogenesis process, contributing altogether to the disease progression.
In conclusion, our results show that DNA methylation and its regulating enzymes are involved in scrapie disease pathology, although there are differences in the epigenetic regulation between the natural model of the disease and the transgenic model. Both methylation forms, 5mC and especially 5hmC, were altered across different brain regions in the two models. 5mC diminished in infected mice and sheep and opposite trends were found for 5hmC, which increased in preclinical mice but diminished in preclinical and clinical sheep. The correlation found between 5mC and 5hmC suggests that the latter could have a relevant role in prion pathogenesis, not only as the intermediate derivative of 5mC. Differential expression of genes encoding epigenetic enzymes was also observed and related mainly to 5hmC patterns. Further studies are warranted to uncover the roles of 5mC and especially 5hmC in prion pathology. Oxidative bisulfite sequencing approaches would be necessary in order to quantify exactly the brain levels of 5hmC and 5mC and to identify specific genes with differential methylation or hydroxymethylation. More research is also indispensable to unravel the exact functions of the different enzymes of the epigenetic machinery in the disease progression and their potential for therapeutic interventions.

Tg338 mice
Tg338 mice (n = 24), which express a transgenic VRQ allele of the ovine PRNP gene under the ovine PrP promoter [35,36], were used in this study. The mice were divided into four groups: clinical mice (n = 6), preclinical mice (n = 6), clinical control (n = 6), and preclinical control (n = 6). Clinical and preclinical mice were intracerebrally inoculated with Tg338-adapted classical scrapie CNS homogenate, originally derived from ARQ/ARQ scrapie sheep, while preclinical and clinical controls were mock inoculated with noninoculated Tg338 CNS homogenate, using the procedure described earlier [72].
The animals included in the preclinical group were sacrificed before showing symptoms associated with scrapie and the clinical ones once they already showed these symptoms. At the same time, the control animals of each group were sacrificed at similar times. Animals in the preclinical groups were inoculated approximately 6 weeks later than the clinical groups in order to sacrifice all animals at similar ages and avoid a possible effect of age in the observed epigenetic modifications [73,74]. In an attempt to achieve homogeneous groups, controls and preclinical mice were sacrificed at different times corresponding to the days on which the animals in the clinical group were sacrificed, when each of them showed clear symptoms of prion disease. The characteristics of the Tg338 mice are summarized in Table 1. Table 1. Characteristics of the Tg338 mice whose samples were used in this study. ID = mouse identification code and DPI = days post-inoculation at which they were sacrificed.

ID Gender Age of Inoculation (Days) Age of Sacrifice (Days) DPI
Regarding ovine samples, after sheep were sacrificed, samples from eight areas of the CNS (the frontal cortex, basal ganglia cortex, basal ganglia, parietal cortex, thalamus, hippocampus, mesencephalon, and obex) were collected and fixed in 10% formaldehyde for further use in histopathological and immunohistochemical analyses. In addition, thalamus samples from each sheep were preserved in RNAlater TM solution (Thermo Fisher Scientific, Waltham, MA, USA) for gene expression analyses.

Neuropathological Evaluation of Tg338 Mice and Scrapie Sheep
Tg338 brains fixed previously in formalin were embedded in paraffin wax, cut into 4-µm-thick sections, and mounted on glass slides for evaluation of vacuolation, a prionrelated lesion, using hematoxylin and eosin (H-E) staining. PrP Sc deposition analysis was carried out using the paraffin-embedded tissue (PET) blot method previously described [75]. Briefly, paraffin-embedded sections were collected onto nitrocellulose membranes (0.45 µm pore size; Bio-Rad, Hercules, CA, USA) and digested with 250 µg/mL of proteinase K for 2 h at 56 • C. Proteins adhered to the membrane were denatured with 3 M guanidine thiocyanate (Sigma-Aldrich, St. Louis, MO, USA) and PrP Sc was detected using Sha31 mouse monoclonal antibody (SPI-Bio, Montigny le Bretonneux, France) at 1:8000 dilution for 1 h. Sections were then incubated for 1 h with an alkaline phosphatase-coupled goat anti-mouse antibody (Dako, St. Clara, CA, USA) at 1:500 dilution. Enzymatic activity was visualized using NBT/BCIP chromogen (Sigma-Aldrich, St. Louis, MO, USA). Vacuolation and PrP Sc deposition were semiquantitatively evaluated per area with a scale score, being 0 for lack of vacuoles/PrP Sc deposit and 5 for very intense vacuolation/PrP Sc deposit, following the standard method to assess these features [76]. The histopathological data of PrP Sc deposition and vacuolation are shown in Figure 1 and Supplementary Table S3. In scrapie sheep, paraffin-embedded 4 µm-thick sections from the eight previously mentioned areas were used for histopathological evaluation of vacuolation by H-E staining. PrP Sc was detected by immunohistochemistry using the monoclonal primary antibody L42 (R-Biopharm, Darmstadt, Germany) at 1:500 dilution for 30 min after formic acid treatment and proteinase K digestion, as previously described [77]. The histopathological evaluation of these sheep was performed in a previous study [34].

Gene Expression Analysis in Tg338 Mice
Seven genes, described to be involved in epigenetic regulation, were selected for the analysis of their expression profile in the thalamus of Tg338 mice: Dnmt1, Dnmt3a, Dnmt3b, Hdac1, Hdac2, Tet1, and Tet2. Table 2 shows TaqMan assays (Thermo Fisher Scientific) used for the amplification of the genes of interest. Table 2. References of the probes used for the RT-qPCR TaqMan amplification assay in Tg338 mice. ID = identification of the commercial company (Thermo Fisher Scientific, Waltham, MA, USA).

Dnmt1 Mm01151063_m1
Dnmt3a Mm00432881_m1 Dnmt3b Mm01240113_m1 Tet1 Mm01169087_m1 Tet2 Mm00524395_m1 Hdac1 Mm02745760_g1 Hdac2 Mm00515108_m1 Sdha Mm01352366_m1 H6pd Mm00557617_m1 Total RNA was isolated using the RNeasy Lipid Tissue Mini kit (Qiagen, Valencia, CA, USA). The quality and quantity of RNA were determined using a NanoDrop instrument (Thermo Fisher Scientific, Waltham, MA, USA). An amount of 200 ng of total RNA was used for retrotranscription with qScript TM cDNA Supermix (Quanta Biosciences TM , Gaithersburg, MD, USA), according to the manufacturer's instructions. The resulting complementary DNA (cDNA) was diluted 1:5 in water and gene expression was quantified by RT-qPCR using the TaqMan universal PCR master mix assays (Thermo Fisher Scientific) in a StepOne Plus Real-Time PCR instrument (Applied Biosystems, Waltham, MA, USA). All reactions were run in triplicate. The comparative quantification of the results was standardized by the 2 −∆∆Ct method [78], using the geometric mean of Sdha and H6pd housekeeping genes ( Table 2) as a normalizer. Student's t-test was applied to identify differences between groups, which were considered significant at p < 0.05.

Gene Expression Analysis in Sheep
The expression profile of six genes (DNMT1, DNMT3A, DNMT3B, HDAC1, HDAC2, and TET1) was assessed in sheep thalamus samples. The primers used for the RT-qPCR assay are listed in Table 3. Table 3. Sequences of primers used in the RT-qPCR assay in sheep. Fw = Forward primer and Rv = Reverse primer.

Gene
Primer Sequences Total RNA isolation and retrotranscription were performed following identical procedures to the ones used with murine samples. The resulting cDNA was diluted 1:5 in water and gene expression was quantified by RT-qPCR using the Fast SYBR TM Green Master Mix (Applied Biosystems, Thermo Fisher Scientific) in a StepOne Plus Real-Time PCR instrument (Applied Biosystems). Each PCR was performed in triplicate. The results were standardized by the 2 −∆∆Ct method [78], using the geometric mean of SDHA and G6PDH housekeeping genes [79] as a normalizer. To identify differences between groups, Student's t-test was performed, and significant differences were considered at p < 0.05.

Immunohistochemical Analysis of 5-Methylcytosine and 5-Hydroxymethylcytosine Levels in Tg338 Mice and Sheep
Immunohistochemistry for the detection of 5mC and 5hmC was performed in paraffinembedded CNS tissue sections from Tg338 mice and sheep. In Tg338 mice, nine brain areas were studied: the frontal cortex, parietal cortex, thalamus, hypothalamus, hippocampus, mesencephalon, cerebellum, obex, and septal area/striatum. In sheep, a total of eight brain areas were analyzed: the frontal cortex, basal ganglia cortex, basal ganglia, parietal cortex, thalamus, hippocampus, mesencephalon, and obex.
After deparaffination and rehydration, tissue sections were subjected to antigen retrieval with citrate buffer (pH 6.0) for 20 min at 85 • C in a PTLink (Dako). Afterward, en-dogenous peroxidase activity was blocked using a precast solution (Dako Agilent, Glostrup, Denmark). Sections were then incubated overnight at 4 • C with the primary antibodies 5mC (Thermo Fisher Scientific, AB_2787107) at 1:100 dilution in Tg338 mice and 1:75 in sheep and 5hmC (Thermo Fisher Scientific, AB_2610634) at 1:200 dilution in Tg338 mice and 1:75 in sheep. The omission of the primary antibody served as a background control for the nonspecific binding of the secondary antibody. Subsequently, the sections were incubated with an anti-mouse enzyme-conjugated EnVision polymer (Dako Agilent, Glostrup, Denmark) for 30 min at room temperature. Diaminobenzidine (DAB, Dako Agilent, Glostrup, Denmark) was used as the chromogen.
CNS sections were assessed and photographed using a Zeiss Axioskop 40 optical microscope (Zeiss, Jena, Germany). 5mC and 5hmC immunostaining levels were evaluated using Image J software and following the semi-quantification analysis method [80], in which for each image, the DAB staining intensity is normalized by the number of nuclei. Afterward, the normalized DAB staining intensity mean and its standard deviation were calculated for each group of animals (clinical, preclinical, and control) in both Tg338 mice and sheep. Student's t-test was applied to identify differences between scrapie-infected and control groups, which were considered significant at p < 0.05.

Correlation between 5mC and 5hmC Levels, Age, Gene Expression, and Prion-Associated Features
Pearson's and Spearman's correlations calculated with GraphPad Prism software v.5 were used to evaluate possible relationships between 5mC and 5hmC intensities and the expression levels of epigenetic regulatory genes in the thalamic region, and between 5mC and 5hmC levels and prion-related lesions in both Tg338 mice and sheep. Correlations were considered significant at p < 0.05. Correlations between the different methylation parameters and age were also calculated in control groups of mice and sheep in order to evaluate if age differences within groups could affect the obtained results.