Differential gene expression in chronic wasting disease‐positive white‐tailed deer (Odocoileus virginianus)

Abstract Chronic wasting disease (CWD) is a transmissible spongiform encephalopathy (TSE) that affects cervid species throughout North America. We evaluated gene expression in white‐tailed deer collected by Illinois Department of Natural Resource wildlife managers during annual population reduction (e.g., sharpshooting) and disease monitoring efforts throughout the CWD‐endemic area of northcentral Illinois. We conducted comparative transcriptomic analysis of liver and retropharyngeal lymph node tissue samples between CWD‐positive (n = 5) and CWD‐not detected (n = 5) deer. A total of 74,479 transcripts were assembled, and 51,661 (69.36%) transcripts were found to have matched proteins in NCBI‐NR and UniProt. Our analysis of functional categories showed 40,308 transcripts were assigned to at least one Gene Ontology term and 37,853 transcripts were involved in at least one pathway. We identified a total of 59 differentially expressed genes (DEGs) in CWD‐positive deer, of which 36 and 23 were associated with liver and retropharyngeal lymph node tissues, respectively. Functions of DEGs lend support to previous relationships between misfolded PrP and cellular membranes (e.g., STXBP5), and internal cellular components. We identified several genes that suggest a link between CWD and retroviruses and identified the gene ADIPOQ that acts as a tumor necrosis factor (TNF) antagonist. This gene may lead to reduced production of TNF and impact disease progression and clinical symptoms associated with CWD (i.e., wasting syndrome). Use of candidate genes identified in this study suggests the activation of endogenous processes in CWD‐positive deer, which in turn may enable earlier detection of the disease.

Post-translational alterations from PrP to PrP Sc have been implicated as the causative factor leading to infection (Prusiner, 1989).
Chronic wasting disease (CWD) is of considerable interest and concern to wildlife managers throughout North America (Williams, Miller, Kreeger, Kahn, & Thorne, 2002). Due to the potential risk of transmission, CWD poses a possible threat to domestic species such as cattle (Basu et al., 2012) and swine (Moore et al., 2017), and cannot be ruled out as a potential risk to human health (Waddell et al., 2018). Risk of cross-species transmission is increased by the ability of prions to affect captive and free-ranging animals (Williams et al., 2002) as captive wildlife are more likely to come in contact with domestic species. Prevalence rates as high as 50% in free-ranging herds and 90% in captive herds have been documented (Haley & Hoover, 2015), though prions are difficult to diagnose in live cervids (Cheng et al., 2016) due to the logistical challenges of collecting diagnostic samples. Tonsil, lymphoid, and third eyelid tissues may be biopsied to confirm CWD infection in live animals; however, these tissue biopsies would require sedation or anesthesia in accordance with Institutional Animal Care and Use Committee (IACUC) protocols, and are impractical in free-ranging herds (Haley & Richt, 2017).
Clinical signs may help in visual diagnoses of positive animals, but are only apparent in final stages of disease progression (i.e., months to years after initial infection; Gilch et al., 2011;Williams, 2005), and may be the result of other chronic disease processes.
Little is known about prion transmission in native hosts (Saunders, Bartelt-Hunt, & Bartz, 2012). Some research suggests underlying mechanisms of PrP Sc formation may be dependent on the type of prion disease (i.e., infectious, sporadic, genetic; Harris, 1999).
Chronic wasting disease is transmitted primarily through direct contact between positive and susceptible animals via oral and mucosal membranes (Safar et al., 2008). Contact with prions in the environment via horizontal transmission (Saunders et al., 2012) and motherto-offspring vertical transmission (Nalls et al., 2013;Selariu et al., 2015) contribute to the rate of disease spread. Furthermore, prions are stable enough to withstand environmental changes such as ultraviolet radiation, freeze-thaw cycles, and bacterial and fungal enzymes (Gilch et al., 2011) and persist for at least 1 year (Kuznetsova, Cullingham, McKenzie, & Aiken, 2018;Wyckoff et al., 2016) in soil in the absence of CWD-positive deer (Johnson et al., 2006).
A single prior study identified gene expression changes in CWDpositive Rocky Mountain elk (Cervus elaphus) using microarray analysis and predetermined transcripts (Basu et al., 2012). This study provided evidence for the involvement of genes assigned to functional groups associated with biological regulation, metabolic process, and cellular process. Moreover, Basu et al. (2012) identified novel genes and numerous pathways that contributed to infection, including calcium signaling, apoptosis and cell death, immune cell trafficking, and inflammatory response. However, confinement to known transcripts is a disadvantage of traditional microarray studies. Studies addressing hypothesis-driven questions related to the role of specific genes in facilitating or reducing disease infection in white-tailed deer (Odocoileus virginianus; hereafter deer) are difficult to conduct given the limited availability of annotated deer genomes and transcripts currently available in the literature. Furthermore, researchers have evaluated the potential for CWD resistance in relation to sequence polymorphisms (Brandt et al., 2015;Kelly et al., 2008) and potential genetic risk factors (Matsumoto, Samuel, Bollinger, Pybus, & Coltman, 2013), but have not previously examined differentially expressed genes in CWD-infected and noninfected deer.
Next-generation sequencing (NGS) allows for discovery of novel transcripts in a more rapid and comprehensive method than other current technologies available (Mardis, 2008) at comparatively low costs (Metzker, 2010). Discovery of novel genes using NGS does not require a priori knowledge of genes that may be present, thus mitigating ascertainment bias. Thus, a need exists for NGS application (Basu et al., 2012) and discovery of novel transcripts to further gene expression studies in all TSE-impacted species. At the initiation of this research, transcriptome-level gene expression evaluation in free-ranging deer using ribonucleic acid (RNA)-sequencing technology had not previously been conducted.
Regardless of protein polymorphisms, liver tissues from CWDpositive and clinically affected deer are CWD immunohistochemistry (IHC)-negative, while retropharyngeal lymph nodes are CWD IHC-positive (Otero et al., 2019). Therefore, we sought to identify differentially expressed genes in liver, which produces proteins involved in the innate immune response (Gao, Jeong, & Tian, 2008), and retropharyngeal lymph nodes, which are sites of prion accumulation (Williams, 2005), from CWD IHC-positive and CWD IHC-not detected (hereafter CWDpositive and CWD-ND, respectively), free-ranging deer. Our study may contribute to an increased understanding of molecular mechanisms involved in the pathology and replication of CWD in cervid species.
To our knowledge, this is the first study evaluating gene expression in CWD using NGS to identify novel transcripts.

| Tissue extraction
From January to March 2015, liver, obex, and retropharyngeal lymph node samples were collected from 380 free-ranging adult (>1.5 years old) deer (Severinghaus, 1949) euthanized by Illinois Department of Natural Resources wildlife managers during annual population reduction and disease management efforts throughout the CWDendemic area of northcentral Illinois (Manjerovic, Green, Mateus-Pinilla, & Novakofski, 2014;Mateus-Pinilla, Weng, Ruiz, Shelton, & Novakofski, 2013). Following euthanasia, deer were transported to central processing locations within 6 hr of death, at which time tissues were rinsed using double-distilled water (ddH 2 O) and any blood removed prior to collection. At the time of necropsy, liver and retropharyngeal lymph node biopsy tissue samples were extracted using 6-mm Miltex surgical biopsy punches . For each animal, liver tissue samples were collected at the approximate center of the right anterior section of the right lobe. We randomly selected the right or left retropharyngeal lymph node for sampling, and collected tissue from the approximate center of the node.
Biopsy punches were placed into 1.5-ml centrifuge tubes and stored in 1.5 ml RNAlater (Thermo Fisher Scientific, Cat. No. AM7020) per the manufacturer's recommendations (Qiagen, Inc. We used test results to select CWD-positive and CWD-ND deer from our biopsy sample collection. We conducted RNA integrity analysis to determine suitable samples for sequencing. We paired CWD-positive suitable samples with randomly selected CWD-ND individuals of similar locations and age classes (i.e., adults). None of 380 deer sampled showed clinical signs; thus, we assumed that if they were CWD-positive, they were in similar stages of disease progression (Williams, 2005). The health status of sampled free-ranging deer was unknown as they had not been tested for other diseases (e.g., epizootic hemorrhagic disease, tuberculosis). From the 380 deer sampled, we selected 10 adult deer for this study. Inclusion criteria were based on integrity of RNA samples, age, sex, and having CWD test results from both obex and retropharyngeal lymph nodes.
We used tissue biopsy samples from 10 (5 CWD-positive [2 males and 3 females; treatment group] and 5 CWD-ND [3 males and 2 females; control group]) adult free-ranging deer for RNA-Seq analyses.
All five CWD-positive deer were IHC-positive in obex and retropharyngeal lymph nodes.

| RNA extraction and sequencing
Using a Qiagen RNeasy Mini Kit (Cat. No. 74104), we extracted RNA from each sample according to the manufacturer's instructions. We examined RNA integrity and quantity using a NanoDrop 1000 and a 2200 TapeStation system using RNA ScreenTape (Agilent, Cat. No. 5067-5576). A total of 5 μg RNA with an RNA integrity number

| De novo assembly
Raw sequencing reads were trimmed by Trimmomatic software (Bolger, Lohse, & Usadel, 2014) to remove low-quality sequencing reads before assembly. Reads with an average quality score below 15 in a 4 base pair sliding window, and reads with quality below 5 at the beginning and end were filtered. After trimming, cleaned reads were used for the reference transcriptome assembly based on Trinity version 2.06 with paired-end mode (Grabherr et al., 2011). Transcripts from liver and retropharyngeal lymph node tissues were separately assembled de novo. To obtain a comprehensive reference transcriptome, the two assemblies were merged and redundant transcripts were filtered by CD-HIT software with default parameters (Li & Godzik, 2006). To filter out misassembled transcripts and transcripts with low expression, raw sequenced reads were mapped to assembled reference transcriptomes using Bowtie 1.0.0 (Langmead, Trapnell, Pop, & Salzberg, 2009). Then, transcript abundance, fragments per kilobase per transcript per million mapped reads (FPKM), values were calculated using RNA-Seq by expectation maximization (RSEM) software (Li & Dewey, 2011), and transcripts with FPKM < 1 were filtered out (Li & Godzik, 2006). Filtered transcripts were used as the deer reference transcriptome for downstream analysis (Grabherr et al., 2011).

| Transcriptome annotation
Assembled transcriptomes were annotated using BLASTX against NCBI-NR and UniProt protein databases, with a cutoff E-value of <1e −6 .
We imported BLASTX results into BLAST2GO software (Conesa et al., 2005), and Gene Ontology (GO) terms, Enzyme Commission numbers, and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were annotated by BLAST2GO software. Protein-coding DNA sequence region was predicted using TransDecoder implemented in Trinity software (Haas et al., 2013). Sequences with a corresponding protein length greater than 100 were retained for further analysis.

| Differential gene expression analysis
Clean reads generated from liver and retropharyngeal lymph node tissues from CWD-positive and CWD-ND groups were mapped back to our assembled reference transcriptome separately, and fragments per kilobase of exon model per million fragments mapped values were calculated by RSEM software for each individual deer (Li & Dewey, 2011). The resulting data matrix that contained FPKM expression values for liver and retropharyngeal lymph node tissues of each individual was generated by "rsem-generate-data-matrix" script. This data matrix with FPKM values was imported into edgeR 2.14 (Robinson, McCarthy, & Smyth, 2010) to create a pairwise comparison between CWD-positive and CWD-ND deer, and identify differentially expressed genes (DEGs) with fold change > 2 2 (log fold change = log 2 [CWD-positive FPKM/CWD-ND FPKM]) and a pvalue < .001 for false discovery rate. We presented expressed genes with a false discovery rate ≤ 0.001. Differential expression analyses were conducted between liver samples from CWD-positive versus CWD-ND deer, and between retropharyngeal lymph node samples from CWD-positive versus CWD-ND deer. We defined an up-regulated gene as a gene that was differentially expressed in CWD-positive deer as compared to a CWD-ND deer, and a down-regulated gene as a gene that was differentially expressed in CWD-ND deer as compared to a CWD-positive deer based on log fold change and false discovery rate results. Genes were analyzed using Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes.
Gene Ontology enrichment analysis of the DEGs detected was conducted by DAVID function annotation tool with Fisher's exact test p-value ≤ .05 (Huang, Sherman, & Lempicki, 2008), to classify DEGs that were molecularly validated based on cellular components, biological processes, and molecular functions.

| Gene validation
Real-time quantitative PCR (qRT-PCR) was conducted to validate the DEGs identified by RNA-Seq. Two assays were designed for each region using PrimerQuest Tool from Integrated DNA Technologies, Inc. A total of 40 assays were used, with one assay repeated twice.
Samples were analyzed using a 2200 TapeStation (Thermo Scientific, Cat. No. 4368814) to verify removal of gDNA. Conversion of RNA to cDNA was accomplished using 1 μg of total RNA per reaction and a High Capacity cDNA Reverse Transcription Kit from Applied Biosystems. Complementary DNA was target-specific preamplified according to a gene expression preamp protocol (Fluidigm, Inc.). We used 12 amplification cycles in the thermal cycling step. Final products were diluted fivefold, and each sample was analyzed in three technical replicates and five biological replicates (i.e., 5 CWD-positive and 5 CWD-ND). BioMark reactions were set up as per Fluidigm's quick reference protocol (Fluidigm, Inc.). We performed qRT-PCR cycling and signal acquisition on the BioMark System and analyzed data using Fluidigm qRT-PCR analysis software (Spurgeon, Jones, & Ramakrishnan, 2008). We further classified DEGs as passing validation when all above parameters were met and amplification plots showed a clear exponential phase and saturation plateau and no residual primer dimers. We classified DEGs as failing validation when the previously mentioned parameters were not met, which led to no or aberrant amplification plots and melt curves. We classified DEGs as interpret with caution for a variety of reasons (e.g., remaining primer dimers, differing temperature peaks, or failed primers; Table 3) per the protocol of the Core Genomics Lab at the University of Chicago to aid in downstream biological interpretation and prioritization. Genomes pathways include purine metabolism, biosynthesis of antibiotics, pyrimidine metabolism, glycerophospholipid metabolism, and phosphatidylinositol signaling system (Table 3).

| RE SULTS
We identified 59 genes as differentially expressed in CWDpositive (as compared to CWD-ND) deer liver and retropharyngeal tissues (Table 4). Among these, 36 were found in liver tissue (16 up-regulated, 20 down-regulated) and 23 (12 up-regulated, 11 down-regulated) in retropharyngeal lymph node tissue; 29 genes have a known function when compared to UniProt and NCBI databases. Of 59 genes, 33 passed validation, 14 failed, and 12 should be interpreted with caution (Table 4). Function of genes that passed validation includes sodium channel proteins, endogenous retrovirus proteins, and cell death activators.
Differentially expressed genes associated with liver and retropharyngeal lymph node tissues included top functions assigned by Gene Ontology associated with cellular membranes, binding, apoptosis, metabolic processes, cellular processes, and catalytic activity (Table 5). Furthermore, we identified several DEGs (i.e.,  (2009). Naturally occurring PrP C is attached to the outer surface of the plasma membrane (Peters et al., 2003) and has been shown to be expressed during infection (Linden et al., 2008). Naturally occurring PrP has multiple binding partners involved in cytoskeletal processes (e.g., maintenance, cell growth; Zafar et al., 2011), and its function has been linked to copper homeostasis, oxidative stress, cell survival differentiation, cell signaling, and cell proliferation. Additionally, it has been associated with synaptic function, maintenance, or structure and a regulatory role at central and peripheral synapses (Westergard, Christensen, & Harris, 2007
Several ERV genes, including ERVK13-1 and ERVK-24, ERVK-25, and P11369 (Table 3), were identified in this study, suggesting a potential link between CWD and retroviruses. In addition, at least two DEGs are associated with ERV Gag, the protein responsible for synthesis of structural proteins necessary for the viral core. Proteins Pol and Env, which encode for reverse transcriptase, and proteins of the viral envelope, respectively, also were differentially expressed in CWDpositive deer (Coffin, Hughes, & Varmus, 1997). Combined with Pro, a virion protease, Gag, Env, and Pol create the backbone of replicating retroviruses (Petropoulos, 1997). Although Pro is not specifically CWD prions. Additionally, PrP has been suggested to influence retroviral activity as it may act as an antiretroviral, specifically in the spleen after immune stimulation (Lötscher et al., 2007). Prions and retroviral cells may be localized in the same cellular compartments, thus acting as cofactors in infection (Leblanc et al., 2006). TA B L E 4 Differentially expressed genes in chronic wasting disease-positive liver (LV) and retropharyngeal lymph node (RPLN) tissues from white-tailed deer collected in the chronic wasting disease-endemic area of northern Illinois during annual population reduction, winter 2015

CWD-ND FPKM d Annotation
Up-regulated in CWD-positive LV

| Association between CWD and immunerelated genes
Several differentially expressed genes identified in our study (i.e., ADIPOQ, CCL3; Table 3) are related to tumor necrosis factor (TNF), a cytokine that produces an immune response to help prevent the spread of infection. It induces fever and apoptotic cell death, and inhibits viral replication. Chronic exposure to TNF can lead to shocklike symptoms including a wasting syndrome (Chu, 2013). It also is important to maintaining follicular dendritic cell networks (Sallusto & Lanzavecchia, 1994). Kitamoto, Muramoto, Mohri, Doh-Ura, and Tateishi (1991) suggested follicular dendritic cells were important to the replication of prions in lymphoid tissues as early Prp Sc accumulates on these cells. Specifically, ADIPOQ, a TNF antagonist (Masaki et al., 2004), is up-regulated in liver tissues of positive deer.
A monokine, CCL3, is down-regulated in liver tissues of positive deer and is responsible for positive regulation of TNF production (Ramos et al., 2005). Such regulation of TNF suggests CWD-positive deer Note: We defined an up-regulated gene as a gene that was differentially expressed in CWD-positive white-tailed deer as compared to a CWD-ND (i.e., not detected) white-tailed deer, and a down-regulated gene as a gene that was differentially expressed in CWD-ND white-tailed deer as compared to a CWD-positive white-tailed deer based on log fold change (FC) and false discovery rate (FDR) results. a DEGs are those with fold change (log FC) > 2 2 and a p-value < .001 for false discovery rate ( are to be expected, as the liver is a part of the innate immune system (Gao et al., 2008).
In this study, IHC determined whether a deer was CWD-positive or CWD-ND; a CWD-positive deer was assumed to be far enough (i.e., not recently infected) in disease progression to exhibit an accumulation of prions in the RPLN tissue and obex. Although speculative, it is possible that prior to the IHC detectable stage of CWD infection, an initial increase in TNF production occurs in response to initial infection, which overtime becomes detrimental to deer whose response is to down-regulate TNF as identified in this study. There was no recorded evidence of declining physical condition in CWDpositive deer, which may be associated with long incubation periods (2-4 years) and absence of clinical symptoms during early stages of prion infection (Williams, 2005). Nevertheless, during later stages of infection chronic exposure to low concentrations of TNF (Wajant, Pfizenmaier, & Scheurich, 2003) may contribute to the wasting syndrome, depression, and cachexia associated with CWD (Chu, 2013).
Additionally, an up-regulated DEG in liver tissue of positive deer associated with interleukin-17 (IL-17; Table 3), is responsible for communication between cells, specifically as an inflammatory response in positive individuals (Huang, Zhang, & He, 2015). Furthermore, other studies have shown interleukin genes to be implicated in prion disease pathogenesis and the innate immune system (Bradford & Mabbott, 2012). Moreover, Meling, Skovgaard, Bårdsen, Heegaard, and Ulvund (2018)   partial CWD resistance-conferring PrP polymorphism at greater frequency than in natural settings) induced by captive breeding and evaluating transmission in a natural setting. Animals in captive facilities are exposed to higher concentrations of CWD prions over less space than their free-ranging counterparts. An increase in the number of CWD-positive animals in a smaller area may lead to higher infection rates in captive individuals and exposure to higher infectious doses of prions than their free-ranging conspecifics (Miller & Wild, 2004). A difference in prion concentration may affect gene expression and time of detection, thereby highlighting the importance of examining CWD gene expression in free-ranging naturally infected individuals. Consequently, a CWD-positive animal may not exhibit prion concentrations high enough for detection using traditional methods if tested too early in disease progression (Haley, Mathiason, Zabel, Telling, & Hoover, 2009). However, use of gene chips and in situ hybridization may enable researchers to select specific candidate genes as indicators of disease status (Lein, Zhao, & Gage, 2004). Gene expression analyses allow for the detection of genetic responses to stimuli before they are phenotypically visible (Klaper & Thomas, 2004), and use of DEGs identified in this study as candidate genes suggests the activation of endogenous process in CWD-infected deer that may advance the pathological process.

| Validation of unknown DEGs and potential functions
Genes that passed validation and were unannotated are candidates for further study. These genes could have implications for the transmission or replication of infectious prion proteins. Even genes that did not pass validation or that should be interpreted with caution may benefit from testing with additional primer sets. Any potential role of the DEGs discussed in this study should be examined in normal prion proteins. Normal PrP function is ambiguous, and DEGs identified in this study may further the understanding of PrP. While many normal prion protein functions have been described, underlying pathogenesis of TSEs is not well understood as amyloid deposits can be found in outwardly healthy individuals (Diack et al., 2016).
It also remains unclear whether conversion of PrP C to PrP Sc leads to a gain of function in PrP Sc -positive individuals (Collins, Lawson, & Masters, 2004) or a loss of function (Samaia & Brentani, 1998).
Additionally, DEGs associated with retroviruses warrant further investigation as they may be involved in CWD endocytic cell pathways related to CWD. Future studies should build upon CWD-associated DEGs identified in this study by examining DEGs in other tissues (i.e., brain stem, blood, rectoanal mucosa-associated lymph tissue) used in routine disease surveillance (Williams, 2005). Further genetic analyses at a transcriptome level could lead to a greater understanding of naturally occurring prion protein functions and thus aid in the understanding of disease-causing prion infection and formation mechanisms.

| CON CLUS IONS
Chronic wasting disease has been widely studied; however, many of the underlying mechanisms that influence transmission and disease spread in infected deer are not well understood. Our research highlights several areas for further investigation. Similar to Ersdal et al. (2009), our research suggests a change occurs in the plasma membrane of CWD-positive deer. Although not explicitly evaluated in this study, this could be due to coinfections with retrovirus or activation of endogenous retrovirus. Furthermore, as Gag, Env, and Pol proteins are differentially expressed, this suggests a link between endogenous retroviruses and CWD as previously presented by Leblanc et al. (2006). Additionally, further investigation of DEGs collected from a wider range of CWD tissues (i.e., obex, blood, tonsils, spleen) may provide greater insight into the mechanisms of disease progression. Investigation of DEGs we have presented may allow for the monitoring of specific genes and their expression, suggesting the activation of endogenous process in CWD-infected deer. anonymous reviewers for providing helpful comments on earlier drafts of our manuscript. We thank Illinois hunters for their willingness to participate in the management of CWD by donating tissue samples during the surveillance efforts at hunter harvest, and the cooperating landowners who allowed targeted reduction of deer on their property in CWD-infected areas.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
CNJ and PAS conceived and designed the research; EKT-L and PAS collected data; EKT-L, GL, and JW contributed reagents/materials and conducted analyses; EKT-L, GL, JTL, JW, PRZ, NEM-P, and CNJ wrote and substantially edited the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
RNA-Seq data are available at the NCBI SRA SRP158695.