Loss of prion protein induces a primed state of type I interferon-responsive genes

The cellular prion protein (PrPC) has been extensively studied because of its pivotal role in prion diseases; however, its functions remain incompletely understood. A unique line of goats has been identified that carries a nonsense mutation that abolishes synthesis of PrPC. In these animals, the PrP-encoding mRNA is rapidly degraded. Goats without PrPC are valuable in re-addressing loss-of-function phenotypes observed in Prnp knockout mice. As PrPC has been ascribed various roles in immune cells, we analyzed transcriptomic responses to loss of PrPC in peripheral blood mononuclear cells (PBMCs) from normal goat kids (n = 8, PRNP+/+) and goat kids without PrPC (n = 8, PRNPTer/Ter) by mRNA sequencing. PBMCs normally express moderate levels of PrPC. The vast majority of genes were similarly expressed in the two groups. However, a curated list of 86 differentially expressed genes delineated the two genotypes. About 70% of these were classified as interferon-responsive genes. In goats without PrPC, the majority of type I interferon-responsive genes were in a primed, modestly upregulated state, with fold changes ranging from 1.4 to 3.7. Among these were ISG15, DDX58 (RIG-1), MX1, MX2, OAS1, OAS2 and DRAM1, all of which have important roles in pathogen defense, cell proliferation, apoptosis, immunomodulation and DNA damage response. Our data suggest that PrPC contributes to the fine-tuning of resting state PBMCs expression level of type I interferon-responsive genes. The molecular mechanism by which this is achieved will be an important topic for further research into PrPC physiology.


Introduction
The cellular prion protein (PrP C ) can misfold into disease-provoking conformers (PrP scrapie; PrP Sc ) that give rise to several neurodegenerative prion diseases, such as Creutzfeldt-Jakob disease (CJD) in humans, scrapie in sheep and goats, and bovine spongiform encephalopathy in a1111111111 a1111111111 a1111111111 a1111111111 a1111111111

RNA-seq data quality control
High quality RNA sequencing data (FASTQ) were derived from Beijing Genome Institute (BGI), with an average total reads of 58,806,319 per sample, average total mapped reads of 42,168,758, and average uniquely mapped reads of 38,253,898 per sample (S1 Fig). To validate the sequencing data, primers (S1 Table) were designed for 12 randomly selected differentially expressed genes (DEGs), using reverse transcription (RT) quantitative real-time PCR (qPCR) on the original RNA. As shown in Fig 1, qPCR analysis of mRNA levels correlated well with the RNA-seq analysis (r = 0.9616, p < 0.0001, Pearson correlation). Minor discrepancies could be due to sample variations, as RNA from only six goats per group were used for qPCR validation, compared with eight goats per group for RNA-seq analysis.
Lack of PrP C subtly alters the transcriptome in immune cells A high correlation was observed between averaged PRNP +/+ and PRNP Ter/Ter normalized gene expression data (r = 0.99, Pearson correlation). However, we found that not all PRNP +/+ and PRNP Ter/Ter goats could be clearly separated from each other, probably reflecting the phenotypic diversity of the goats (S2 Fig). Despite this, using edgeR [32] and a p-value cut-off < 0.05, 735 genes were differentially expressed between the two genotypes (S1 File). Further filtration of the gene list using cut offs for fold change (log2 FC ± 0.5) and mean number of reads (> 100 reads in one of the groups) generated a high-confidence gene list of 127 DEGs, of which 67 were upregulated and 60 were downregulated in the PRNP Ter/Ter genotype (S2 Table). Of note, as we have previously shown that the PBMC cell populations, mainly T cells, B cells and monocytes, are stable between the two genotypes compared in our study [26], the DEGs result from real genotype-associated shifts in gene expression, not shifts in the cell populations. Reassuringly, the PRNP gene was among the DEGs, with very few reads mapping to this locus in the mutant. The chromosomal distribution of the DEGs is found in S3 Fig. The PRNP gene is located on chromosome 13 in goats. Only 1 (SIGLEC1) of the 86 annotated DEGs also maps to Validation of RNA sequencing data with quantitative PCR. Validation of 12 randomly chosen, differentially expressed genes was performed with qPCR using the original RNA. Expression data from the two methods are presented as relative expression between PRNP Ter/Ter and PRNP +/+ animals (RNA-seq data n = 8, qPCR n = 6; r = 0.9616, p < 0.0001, Pearson correlation).
https://doi.org/10.1371/journal.pone.0179881.g001 chromosome 13. This gene is expressed at a low level and is irrelevant for the findings in our study.
Of the average total number of genes expressed in PBMCs from both genotypes, only 0.7 percent of the genes were altered upon loss of PrP C (Fig 2A). Using Ingenuity Pathway Analysis (IPA), of the 127 high-confidence DEGs, 86 genes were functionally annotated. Interestingly, 22 of these genes were categorized as "Viral infection" (p-value = 3.27x10 -5 ), and additional genes were related to other anti-virus-associated terms. The majority of these genes were upregulated in the PRNP Ter/Ter genotype compared with the PRNP +/+ genotype. Of the top canonical pathways, "Interferon signaling" was by far the most affected (p-value = 8.92x 10 -6 ). Due to these findings, we performed further analyses of the annotated DEGs using the Interferome database [33]. Strikingly, 60 of the 86 annotated DEGs were interferon-responsive genes ( Fig 2B). Of these, 42 were upregulated (red bar) and 18 downregulated (blue bar) in the PRNP Ter/Ter genotype. Fig 2C shows the inter-individual variation in gene expression of all samples represented in a heatmap, and hierarchical clustering analysis of the 60 interferonresponsive genes revealed a clustering of downregulated and upregulated genes between the PRNP +/+ and PRNP Ter/Ter genotypes.
Since the observed data could be due to altered expression levels of interferons or components in type I interferon signaling, we analyzed expression levels of a number of genes that could affect the expression of interferon-responsive genes. However, differences between the genotypes were not detected (Table 1), except for IFNB2-like, which was slightly downregulated in the PRNP Ter/Ter genotype (p-value = 0.025).

Introduction of PRNP inhibited MX2 gene expression in SH-SY5Y cells
To test whether PrP C could influence IFN-α responsiveness in a cell culture system with a different genetic makeup, we used human neuroblastoma SH-SY5Y cells, which normally express extremely low levels of PrP C . SH-SY5Y clones stably expressing human PrP C were generated (SH-SY5Y PrP high ) and assessed with regard to glycosylation and proteolytic processing to ensure physiological post-translational modification and trafficking of PrP C (S4 Fig). Eight clones stably expressing PrP C as well as untransfected SH-SY5Y cells were exposed to 3 U/ml IFN-α for 3h. One of the transfected clones showed aberrantly high MX2 gene expression levels and was excluded from the analysis. Of the seven clones included in the experiment, six displayed a significantly reduced response to IFNα, as assessed by the interferon-responsive gene MX2 expression levels, compared with the untransfected SH-SY5Y cells, using Dunnett's post hoc test for multiple comparisons (Fig 3) (n = 4, mean ± SEM). The levels of PrP C expression did not directly correlate with the degree of MX2 expression-level inhibition; however, this was not expected due to the complexity of the interferon signaling pathway, and the possible distance between PrP C interference and MX2 gene expression. On average, the clones showed a significantly inhibited response to IFNα (p-value = 0.0001) compared with the untransfected SH-SY5Y cells, using a two-way ANOVA.
Increased interferon-responsive gene expression in blood leukocytes devoid of PrP C after LPS challenge In an independent, parallel study [34,35], goats were challenged intravenously with lipopolysaccharide (LPS), thereby indirectly stimulating interferon pathways. RNA was extracted from circulating blood leukocytes, and gene expression of interferon-responsive genes was assessed by FLUIDIGM qPCR. As shown in Fig 4A, basal level expression (0h) of several interferonresponsive genes was slightly higher in the PRNP Ter/Ter (n = 13) genotype than in the PRNP +/+ (n = 12) genotype, albeit being significantly different for only IFI6 (p-value = 0.037). Moreover, Interferon-responsive genes dominate among the differentially expressed genes in goats lacking PrP C . Graphical presentation of (A) the total number and percentage of differentially expressed genes (DEGs) between the two genotypes, compared to the average total number of genes expressed in peripheral blood mononuclear cells from both genotypes, and (B) the total number of upregulated and STAT1 mRNA expression levels did not differ between the genotypes. One hour after LPS challenge, the mRNA expression level of interferon-responsive genes increased slightly and the difference between the two genotypes was more pronounced (Fig 4B), with three genes showing a statistically significant difference in expression level (ISG15 (p-value = 0.049), IFIT1 (pvalue = 0.02), and MX1 (p-value = 0.019), assessed by multiple t-tests).

Discussion
Similar to observations in transgenic mice [6], goats [36], and cattle [27] with knockout (KO) of PRNP, the PRNP Ter/Ter goats display no obvious loss-of-function phenotype [25,26]. Consequently, only subtle transcriptomic alterations were expected, corroborating data from KO mouse models [37][38][39][40][41]. Accordingly, this study revealed subtle expression differences affecting less than a percent of the expressed genes. However, analysis of the annotated DEGs using the Interferome database [33], identified a distinct expression profile, with 70 percent of the DEGs being classified as interferon responsive, of which several were among the top upregulated genes. Importantly, animals were age-matched and derived from the same research flock. The health status of this herd is frequently monitored and considered excellent. Prior to sampling, animals were assessed clinically by a veterinarian and found healthy, which was also confirmed by hematological analysis in an accompanying study [26]. Furthermore, we were unable to detect any differences in gene expression levels of neither interferons nor IFN signaling components. A flanking gene problem will also be present in the PRNP Ter/Ter goats; however, preliminary data indicate that this is very limited compared to inbred knockout mouse models. In the absence of alternative explanations, we consider the observed gene expression profile to be a true signature of PrP C loss-of-function. It is likely that this profile, which is evident at rest in the outbred and immunocompetent goats, might be even weaker or absent in inbred transgenic mice, housed in pathogen-depleted environments. It is, however, interesting to note that studies of prion disease in mice have revealed a gene expression profile similar to that observed in PrP C -deficient goats. Analysis of transcripts from mouse whole brain throughout the course of experimental CJD revealed an upregulation of several interferon-responsive genes, e.g. OAS, ISG15, and IRF-family members. Importantly, the upregulation of these genes occurred very early in the course of the disease, approximately 50 days before the onset of neuropathological signs and detection of PrP Sc [42]. Similar findings were recently reported in another study of prion-infected mice [43]. In a hamster model of scrapie, several interferon-responsive genes, including those encoding OAS and Mx protein, were upregulated during development of scrapie [44]. In addition, three interferon-responsive genes, assessed by qPCR studies, were moderately upregulated in a hamster model and different mouse models inoculated with scrapie strains [45]. Recently, transcriptomic data from cerebellar organotypic cultured slices infected with prions showed that a slight upregulation of several interferon-responsive genes was evident at 38 and 45 days post infection [46]. It is tempting to speculate that some of the observed gene expression alterations at very early stages of prion disease could, at least partly, reflect induced loss-of-PrP C function, and, thus, explain the similarity with the expression profile reported here. Further investigations are clearly needed to test this hypothesis.
Studies of human SH-SY5Y neuroblastoma cells transfected with human PRNP displayed a significantly dampened response (MX2 expression) to a low-level IFN-α stimulation, downregulated annotated DEGs. The fraction of upregulated (red) and downregulated (blue) interferonresponsive genes among the DEGs are also shown. (C) Hierarchical clustering of the interferon-responsive genes among the DEGs and expression data from all individual goats of both genotypes. Hierarchical clustering was performed using the ward algorithm on log2-normalized fold changes. compared with untransfected cells that are virtually devoid of PrP C . Furthermore, in an independent, parallel study involving older goat kids than those recruited for the RNA seq study, animals were challenged with LPS, which is a potent pro-inflammatory compound. In contrast to mice, which are relatively tolerant towards LPS, goats have a similar sensitivity as humans [34,35]. In line with data from the present RNA sequencing study, resting state expression levels of interferon-responsive genes in leukocytes were slightly elevated in the PRNP Ter/Ter genotype. Interestingly, the expression differences between the genotypes were increased one hour after LPS injection. Apparently, leukocytes without the expression of PrP C upregulated interferon-responsive genes more rapidly than their PrP C -expressing counterparts. The regulation of interferon-responsive genes expression level is multifaceted and tightly controlled at several levels [47,48], involving receptor downregulation, upregulation of a plethora of inhibitors as well as epigenetic modifications.  Taken together, our data suggest that PrP C contributes to dampening of type I interferon signaling at rest and that loss of PrP C induces a primed state of interferon-responsive genes. Accordingly, direct or indirect stimulation of type I IFN signaling, elicits a somewhat stronger immediate response when PrP C is absent. These data do not conflict with roles acclaimed to the prion protein. Indeed, they might strengthen previous observations and provide mechanistic hints of PrP C physiology.

Material and methods Animals
The animals (FOTS approval number ID 8058) included in the study were of the Norwegian Dairy Goat Breed obtained from a research herd of approximately 100 winter-fed goats at the Norwegian University of Life Sciences. Based on health surveillance through membership in the Goat health monitoring service and The Norwegian Association of Sheep and Goat Farmers and daily monitoring, the general health status of the herd is considered to be good. The entire flock was previously genotyped [25] concerning PRNP genotypes, and through selective breeding, goat kids with the two genotypes PRNP +/+ (n = 8; 4 female and 4 male) and PRNP Ter/ Ter (n = 8; 4 female and 4 male) were retrieved. Prior to inclusion in the experiment, all goat kids were examined clinically and found to be healthy.

Isolation of peripheral blood mononuclear cells
Blood was sampled from the jugular vein into EDTA tubes at 2-3 months of age. Peripheral blood mononuclear cells (PBMCs) were isolated by gradient centrifugation (Lymphoprep1, Axis-Shield, Dundee, Scotland) at 1760 x g without brake, and washed with PBS supplemented with EDTA (2 mM). Red blood cells were lysed by brief exposure to sterile water, and washed with PBS supplemented with EDTA (2 mM) prior to counting and trypan blue viability assessment using a Countess1 Automated Cell Counter (Life Technologies, Thermo Fisher Scientific, Waltham, MA).

Cell culture studies
Human neuroblastoma SH-SY5Y cells (Sigma-Aldrich, Merck, Kenilworth, NJ) were cultured in Eagle's Minimum Essential Medium and Ham's F12 (1:1) (Sigma-Aldrich) supplemented with 10% heat-inactivated fetal bovine serum (FBS), glutamine and antibiotics (1% streptomycin and penicillin) (all from Gibco, Thermo Fisher Scientific), and cultivated in T25 flasks at 37˚C with 5% (v/v) CO 2 at saturated humidity. SH-SY5Y cells were stably transfected with a plasmid construct, pCI-neo (Promega, Madison, WI) encoding human PRNP, using jetPRIME (Polyplus, Illkirch, France) according to the manufacturer's instructions. Transfected cells were grown under selection pressure of Geneticin (Thermo Fisher Scientific), and nine different single clones with variable levels of PrP C (SH-SY5Y PrP high ) were isolated (S4 Fig). Clone no. 8 showed an abnormal phenotype, and was excluded from the studies.

Isolation and sequencing of RNA
Total RNA was extracted using the Qiagen RNeasy mini plus kit (Qiagen, Germantown, MD) following the manufacturer's instructions. RNA concentration and purity was analyzed using NanoDrop-1000 Spectrophotometer (Thermo Fisher Scientific) or Epoch Microplate Spectrophotometer (BioTek Instruments Inc, Winooski, VT), and quality was assessed before RNA sequencing using RNA Nano Chips on an Agilent 2100 Bioanalyzer (both from Agilent Technologies, Santa Clara, CA). RNA was stored at -80˚C. Individual RNA samples of high quality (RIN ! 9.8) were sequenced by mRNA poly-A-tail, paired-end sequencing (Illumina HiSeq 2000) with 91 bp read-lengths (Beijing Genomics Institute (BGI), Hong Kong), retrieving a minimum depth of 5G clean data per sample. In detail, after the total RNA extraction and DNase I treatment, magnetic beads with Oligo (dT) were used to isolate mRNA. Mixed with the fragmentation buffer, the mRNA was fragmented into short fragments, and cDNA was synthesized using the mRNA fragments as templates. Short fragments were purified and resolved with EB buffer for end reparation and single nucleotide A (adenine) addition. The short fragments were connected with adapters. After agarose gel electrophoresis, the suitable fragments were selected for the PCR amplification as templates. During the QC steps, Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System were used in quantification and qualification of the sample library.
For the IFN-studies, RNA quality was assessed by TAE/formamide RNA gel electrophoresis. RNA samples were mixed with formamide (50% v/v, Sigma) and orange loading dye (New England Biolabs), denatured by heating for 5 min at 65˚C, put on ice, and loaded on 1% agarose gel containing 1xTAE buffer (0.04 M Tris-acetate, 1 mM EDTA) and visualized with SYBR™ Safe (Invitrogen, Thermo Fisher Scientific).

Analysis of RNA sequencing data
Reads were mapped to the goat genome assembly (CHIR_1.0) using SOAP2 [49]. Reads per gene were obtained using SOAP2 and the goat genome annotation (RefSeq, CHIR_1.0). Read counts were normalized to reads per kilobase per million mapped reads (RPKM) [50]. Testing for differentially expressed genes was performed using the function exactTest in edgeR [32].
Expression analysis by reverse transcription (RT) quantitative real-time PCR (qPCR) analysis cDNA was synthesized using SuperScript III Reverse Transcriptase, RNase Out, dNTP mix and Random Primers (all from Invitrogen, Thermo Fisher Scientific) at the following conditions: 5 min at 65˚C, >1 min on ice, 5 min at 25˚C, 1 h at 50˚C and 15 min at 70˚C.
For the RNA sequencing validation study, qPCR was conducted with LightCycler 480 Sybr Green I Master mix (Roche). cDNA corresponding to 2.5 ng RNA was used per reaction. The samples were run in duplicates in a total volume of 20 μl on a LightCycler 96 System (Roche). Conditions: 5 min at 95˚C; 40 cycles of 10 sec at 95˚C, 10 sec at 60˚C and 10 sec at 72˚C; and melting curve with 5 sec at 95˚C, 1 min at 65˚C and 97˚C. Relative expression levels were calculated using a standard curve generated from one randomly selected animal, run in triplicate, with GAPDH as a reference gene, and one randomly selected animal as a positive control. The average of six PRNP Ter/Ter animals was divided by the average of six PRNP +/+ animals, and compared relative to RNA sequencing data.
For the interferon-treatment studies using SH-SY5Y cells, qPCR was conducted with Light-Cycler 480 Sybr Green I Master mix (Roche). cDNA corresponding to 10 ng RNA was used per reaction. The samples were run in triplicate in a total volume of 10 μl on a LightCycler 96 System (Roche). Conditions: 5 min at 95˚C; 40 cycles of 10 sec at 95˚C, 10 sec at 60˚C and 10 sec at 72˚C; and melting curve with 5 sec at 95˚C, 1 min at 65˚C and 97˚C. Relative expression levels were calculated using the ΔΔCt method. ActB was used as a reference gene. An inter-run calibrator was included on every plate as a positive control. The qPCR-amplified sample was run on a 1% agarose gel, and visualized using SYBR™ Safe (Thermo Fisher Scientific).

LPS challenge and FLUIDIGM qPCR of whole blood leukocyte interferon-responsive genes
An intravenous LPS challenge was performed (0.1 μg/kg, Escherichia coli O26:B6) in 16 Norwegian dairy goats age 6-7 months (8 PRNP +/+ (female) and 8 PRNP Ter/Ter (7 female, 1 castrated male)) (FOTS approval number IDs 5827, 6903, and 7881), and 10 controls were included (5 of each genotype). In brief, blood samples were collected in PAX-gene blood RNA tubes before (0 h) and after LPS challenge (1 h). High quality RNA (RIN 9.0 ± 0.34) was extracted using the PAXgene Blood miRNA kit, and cDNA synthesis was performed in two replicates (QuantiTect Reverse Transcription Kit). The relative expression of ISGs in circulating leukocytes was assessed after qPCR on the Fluidigm Biomark HD platform and data analysis using GenEx5 software (MultiD, Sweden). The full study protocol, method description, and primer sequences can be found in [34,35].

Statistical analysis
Multiple t-tests or two-way ANOVA followed by Dunnett's post hoc test for multiple comparisons were used for statistical analysis of the data using Graph Pad Prism v. 6.07 (Graphpad, La Jolla, CA). For correlation analysis, the Pearson correlation coefficient was calculated. Mean values are presented ± SEM.

Ethics statement
The animal experiments were performed in compliance with ethical guidelines, and approved by the Norwegian Animal Research Authority (FOTS approval number IDs 8058, 5827, 6903, and 7881) with reference to the Norwegian regulation on animal experimentation (FOR-2015-06-18-761).