Alterations in gut microbiota linked to provenance, sex, and chronic wasting disease in white-tailed deer (Odocoileus virginianus)

Chronic wasting disease (CWD) is a fatal, contagious, neurodegenerative prion disease affecting both free-ranging and captive cervid species. CWD is spread via direct or indirect contact or oral ingestion of prions. In the gastrointestinal tract, prions enter the body through microfold cells (M-cells), and the abundance of these cells can be influenced by the gut microbiota. To explore potential links between the gut microbiota and CWD, we collected fecal samples from farmed and free-ranging white-tailed deer (Odocoileus virginianus) around the Midwest, USA. Farmed deer originated from farms that were depopulated due to CWD. Free-ranging deer were sampled during annual deer harvests. All farmed deer were tested for CWD via ELISA and IHC, and we used 16S rRNA gene sequencing to characterize the gut microbiota. We report significant differences in gut microbiota by provenance (Farm 1, Farm 2, Free-ranging), sex, and CWD status. CWD-positive deer from Farm 1 and 2 had increased abundances of Akkermansia, Lachnospireacea UCG-010, and RF39 taxa. Overall, differences by provenance and sex appear to be driven by diet, while differences by CWD status may be linked to CWD pathogenesis.

CWD sample collection and testing. CWD testing on farmed deer was conducted per USDA CWD Program Standards and under the purview of USDA Animal and Plant Health Inspection Service (APHIS). Testing was performed using standard operating protocols approved and audited by the USDA and documented in the quality system of the Wisconsin Veterinary Diagnostic Laboratory (WVDL). The WVDL is a National Animal Health Laboratory Network (NAHLN) Level 1 laboratory and is accredited by the American Association of Veterinary Laboratory Diagnosticians. The heads were removed from all farmed deer greater than one year of age and the obex region of the brainstem and medial retropharyngeal lymph nodes (MRLN) were collected following USDA APHIS guidelines 37 . Regulatory surveillance samples were shipped to the National Veterinary Services Laboratory (NVSL) for CWD Immunohistochemistry (IHC). Enzyme-linked immunosorbent assay (ELISA) testing was performed at WVDL. IHC and ELISA-based testing for the abnormal prion protein were performed on the dorsal motor nucleus of the vagus nerve in the obex and MRLNs. For IHC testing, tissues were preserved in 10% neutral buffered formalin, embedded in paraffin, sectioned at 5 µm, mounted on slides, and examined using IHC with monoclonal antibody (Mab) F99/97.6.1 38 . All testing was performed using USDA-APHIS-NAHLN approved protocols (IHC: SOP-PS-0002.02; ELISA: SOP-PS-0007.01). Animals were considered CWD-positive if any one of the tissues examined contained detectable PrP CWD . Animals in which tissues did not contain detectable PrP CWD were considered CWD non-detect animals. CWD testing was performed under USDA regulatory guidelines and not for the purposes of this study. CWD test results generated under USDA purview were shared with investigators on this study for the purpose of assessing deer gut microbiota in relation to CWD status.
DNA extraction, amplification, and sequencing. DNA  Sequence processing and analysis. A total of 231 samples were submitted for sequencing. Raw, pairedend reads were processed and denoised in QIIME2 v. 2020.2 41 . Taxonomy was assigned using the SILVA 132 99% amplicon sequence variants (ASVs) database from the 515F/806R classifier 42,43 , and samples were filtered at a sequencing depth of 10,000 features. This resulted in the retention of 229 samples with the loss of 2 samples-one CWD non-detect male deer from Farm 1 and one CWD non-detect female deer from the free-ranging population. After filtering, 5,803,410 reads from 229 samples were used for analysis (average of 25,342 reads per sample). Reads per sample ranged from 10,049 to 92,179 reads. Alpha (Shannon Diversity Index) and beta diversity were analyzed using QIIME 2 41 . Beta diversity indices were compared using permutational multivariate analysis of variances (PERMANOVA) between weighted and unweighted Unifrac distance matrices. P-values were corrected for multiple comparisons using the Benjamini-Hochberg FDR correction, and values less than 0.05 were considered significant. An analysis of composition of microbes (ANCOM) was used to determine differentially abundant taxa between groups after filtering out taxa that had fewer than 10 reads and taxa that occurred in fewer than two deer. ANCOMs generate W values, which represent the number of times the null hypothesis is rejected in pairwise comparisons of microbial species ratios between groups. For example, a W value of 80 would indicate the null hypothesis was rejected 80 times when comparing microbial species ratios between different groups. We performed ANCOMs at both the L7 and amplicon sequence variant (ASV) levels. The L7 level is roughly equivalent to a species level while an ASV is roughly equivalent to a bacterial strain and may differ from another ASV by as few as one nucleotide 44 . Multiple ASVs may be classified as a single L7 level taxa. However, deeper genome sequencing is necessary for true species and strain differentiation. The single CWD-positive female was not included in statistical analyses comparing CWD-positive and CWD non-detect animals to reduce any confounding introduced by sex. Age was compared using a Kruskal-Wallis test between CWD-positive and CWD non-detect deer by farm after testing the data for normality using a Shapiro Wilk Test (R Studio, version 1.3.1093. Sequencing data are available at NCBI Bioproject PRJNA688284.
To identify microbial taxa that were differentially abundant between farmed and free-ranging deer, we combined all deer from Farm 1 and 2-excluding CWD-positive deer-and compared these against the free-ranging deer. Through an ANCOM at the L7 (roughly species) level, we identified 82 taxa that were differentially abundant (Supp. Table 1). Twenty-six of these taxa were in the order Bacteroidales (phylum Bacteroidetes) and seven of these were in the family Prevotellaceae. The vast majority of the Bacteroidales taxa (22 of 26) were significantly increased in the farmed deer. On the other hand, free-ranging deer had significantly greater abundances of taxa (25 of 38) in the Clostridiales order (phylum Firmicutes), all of which were in the Ruminococcaceae and www.nature.com/scientificreports/ Lachnospiraceae families. Based on these results, we decided to compare log Firmicutes:Bacteroidetes (F:B) ratios for farmed and free-ranging deer. Log F:B ratios are associated with dietary energy harvest and higher ratios indicate greater energy extraction [45][46][47] . We found significantly higher F:B ratios in the free-ranging deer as compared to the farmed deer (Log F:B ratios (mean ± SE), Free-ranging: 0.39 ± 0.03; Farmed: 0.08 ± 0.02; Kruskal-Wallis p < 0.0001).
We also discovered significant differences in microbial composition but not diversity by sex on Farm 1 and in free-ranging deer (CWD non-detect deer only; Farm 1: Unweighted UniFrac PERMANOVA p = 0.008, Shannon Diversity Index p = 0.34, Weighted UniFrac PERMANOVA p = 0.003; Free-ranging: Unweighted UniFrac PERMANOVA p = 0.018, Shannon Diversity Index p = 0.53, Weighted UniFrac PERMANOVA p = 0.066; Fig. 2a, c, d Supp. Fig. 2a, c). No significant differences in microbial composition or diversity were detected by sex on Farm 2 (CWD non-detect deer only; Farm 2: Unweighted UniFrac PERMANOVA p = 0.179, Shannon Diversity Index p = 0.15, Weighted UniFrac PERMANOVA p = 0.115; Fig. 2b, d Supp. Fig. 2b). There were also no differentially abundant microbial taxa detected by sex on Farm 2. However, on Farm 1, we identified a single taxa that was significantly increased in males. This was an uncultured bacterium from the order Bacteroidales, family RF16 (ANCOM, L7-roughly species level, W = 626). In the free-ranging deer population, there were multiple differentially abundant taxa by sex, with the two most differentially abundant including a microbe in the genera Oscillibacter and a microbe in the family Lachnospiraceae, genera GCA-900066575. Both of these taxa were significantly increased in males (Supp. Table 2). Microbial composition and diversity by CWD status. Based on the microbial composition differences observed by sex and the fact that there was only one CWD-positive female in the entire data set, we opted to ana-   www.nature.com/scientificreports/ in the order Gastranaerophilales also increased in non-detect males (ANCOM W = 34); and a microbe in the family Lachnospiraceae UCG-10 increased in CWD-positive males (ANCOM W = 28; Fig. 4c) (Supp. Table 3). On Farm 2, at the L7 level, two microbes were found to be differentially abundant. Both were increased in CWD-positive males and included a unidentified rumen bacterium from the class Bacilli, order RF39 (ANCOM W = 132; Fig. 4b) and a microbe from the family Lachnospiraceae UCG-10 (ANCOM W = 115; Fig. 4c). On Farm 2, multiple ASVs were also differentially abundant (Supp . Table 4), the top three of which, all increased in CWD-positive males, were an ASV from the class Bacilli (formerly Mollicutes), order RF39 (ANCOM W = 132; Fig. 4e), an ASV from the family Lachnospiraceae UCG-10 (ANCOM W = 176; Fig. 4f), and an ASV from the genera Akkermansia (ANCOM W = 90; Fig. 4g). On Farm 1, only one ASV in the Akkermansia genera was found to be differentially abundant and increased in CWD-positive males (ANCOM W = 1958; Fig. 4h). This Akkermansia ASV differed from the Akkermansia ASV on Farm 2, suggesting species or strain level variance in Akkermansia by farm; although, deeper sequencing is required to asses this. Akkermansia taxa at the L7 level were not differentially abundant on either farm (Fig. 4d).

Discussion
In this study, we used 16S rRNA gene sequencing to compare the gut microbiota of farmed and free-ranging white-tailed deer (Odocoileus virginianus). We hypothesized that deer gut microbiota would differ by provenance (Farm 1, Farm 2, and Free-ranging) and disease status (CWD-positive, CWD non-detect). Indeed, microbial composition and diversity did vary with provenance. Moreover, composition but not diversity varied with sex (Farm 1 and Free-ranging only) and with CWD status (Farm 1 and Farm 2).

Drivers of microbial community composition by provenance. Multiple factors could contribute
to the gut microbial differences we observed based on provenance, including diet, spatial proximity 48 , host genetics 49 , and biogeography 50 . Although, diet likely has the strongest influence on gut microbial composition and diversity as compared to the other factors [51][52][53] . The free-ranging deer in this study had diets that primarily consisted of browse, small plants, shrubs, grasses and occasional agricultural, landscaping, and garden plants 54 . The farmed deer had access to pastures and were also fed a variety of commercial deer feeds, grains, hay, and supplemental items, including peanuts, roasted soybeans, and dandelions. As diets differed between farmed and free-ranging deer, it was not surprising that farmed and free-ranging deer had significantly different microbial communities or that there were significant differences in gut microbiota between the two farms with different feeding regimens. Multiple previous studies have also reported gut microbial differences between wild and captive animals [55][56][57] , including ruminants 36,58 .
We predicted that based on differing diets, host genetics, and biogeography, we would observe distinct microbial signatures in deer from each location (Farm 1, Farm 2, Free-ranging) (Fig. 1a). We further predicted that within locations, farmed deer would have more similar (less distant) microbiota due to more regulated diets and more limited "home ranges" as compared to free-ranging deer (Fig. 1c, Farm 1 to Farm 1, Farm 2 to Farm 2, Free-ranging to Free-ranging). Finally, we hypothesized that the greatest differences in microbial communities would be observed between farmed and free-ranging deer since farmed deer generally share more similar diets (formulated commercial feeds, grains, hay, pasture) than free-ranging deer (Fig. 1c, Farm 1 to Farm 2, Farm 1 to Free-ranging, Farm 2 to Free-ranging). Our results supported each of these predictions. Although we cannot parse the individual effects of diet, spatial proximity, host genetics, and biogeography in this data set, the differentially abundant taxa identified between farmed and free-ranging deer strongly support a role for diet as a key driver of the microbial community differences we observed. Free-ranging deer consume a plant and fiber-rich diet full of shrubs and browse, while farmed deer consume a starchier diet of grains and commercial feed in addition to pasture and hay. Microbial taxa in the Lachnospiraceae and Ruminococcaceae families were increased in abundance in the free-ranging deer, while Bacteroidales taxa, like Prevotellaceae, were increased in the farmed deer (Supp. Table 1). Lachnospiraceae and Ruminococcaceae taxa are associated with plant-rich diets, and these taxa metabolize plant materials such as cellulose and hemiceullulose 57,59,60 . Bacteroidales and Prevotellaceae are more commonly associated with starch consumption, and in ruminants, Bacteroidales, including Prevotella, increase in animals on concentrate/grain diets 57,61-63 .
Increased Firmicutes:Bacteroidetes ratios in free-ranging deer further suggests increased energy extraction and fermentation efficiency. In humans, increased F:B ratios are associated with obesity; in farmed ruminants, increased ratios are positively correlated with average daily gain 45,46 . In foregut-fermenting primates (which have ruminant-like digestion), wild primates exhibited higher F:B ratios than captive primates 47 . This was attributed to the need for the wild primates to maximize energy extraction from "low-quality" food items such as fibrous plants, bark, and seeds, while captive primates, with "high quality" diets rich in soluble carbohydrates, were less dependent on efficient energy harvest 47 . Similarly, free-ranging deer gut microbiota may maximize energy extraction from a fibrous browse diet, while the grain-rich diets of farmed deer reduces the need for fermentation efficiency and creates a niche for microbial taxa capable of metabolizing soluble starches and sugars.
Microbial community structure by sex. Interestingly, we also identified microbial composition differences by sex on Farm 1 and in free-ranging deer. No differences were observed on Farm 2, which also had the smallest sample size (n = 18 males, 12 females), limiting our power to detect these differences. Microbial community alterations associated with sex could be attributed to differential feeding / diets by sex or hormonal influences on the gut microbiome. We received anecdotal reports of differential feeding by sex on Farm 1 based on deer breeding and growth requirements. While we did not characterize the diet of free-ranging deer in this study, a previous study on white-tailed deer reported that, in winter, female deer in the Midwest consumed more grass (higher quality feed) and less browse than male deer 64 . Our samples were also collected from free- www.nature.com/scientificreports/ ranging deer in the Midwest during winter. Differentially abundant taxa observed between sexes (Bacteroidales, Lachnospiraceae, Oscillibacter-all increased in males) also supported a role for diet in microbial community differences by sex. Notably, Oscillibacter species increase in humans on diets high in resistant starch and low in carbohydrates 65 , which is consistent with the browse-rich winter diet of free-ranging male deer 64 . Breeding hormones have also been linked to gut microbial changes by sex in wild animals 66,67 . However, a 2017 study on white-tailed deer observed no differences in microbial composition between sexes 52 ; although, sampling season differed from this study. Our samples were collected January through March, which corresponds to estrous cycling or pregnancy in females and post-rut (declining testosterone levels) in males 68 .
Chronic wasting disease and the gut microbiota. On both farms, we observed significant differences in microbial composition in CWD-positive deer as compared to non-detect deer. Twenty-five of the 26 total CWD-positive deer across both farms were male (Table 1). Previous studies in wild deer have reported that CWD prevalence is two times higher in males, and that males have a threefold greater risk of CWD infection as compared to females 69 . These differences in infection risk and prevalence by sex are thought to be linked to increased CWD transmission amongst male social groups outside of breeding season 69 . Alternately, models of CWD outbreaks in captive deer predict that density-dependence and indirect transmission 70 play an important role in CWD spread. Although penning arrangements on Farm 2 are unknown, on Farm 1, male deer were penned with females during rut (fall) and then separated into bachelor herds for the remainder of the year. As such, transmission through male social groups and indirect, density-dependent transmission (in bachelor pens) could have played a role in the predominantly male infections observed in farmed deer. Because of this skew by sex, we opted to analyze only males in relation to CWD status. This within-farm, male-only analysis mitigated potential gut microbial confounders, including sex, diet, and biogeography. Differentially abundant microbial taxa common across both farms and increased in CWD-positive animals included: two different microbes in the class Bacilli, order RF39 (formerly Mollicutes RF39)-one increased on Farm 1 and one increased on Farm 2; a microbe in the family Lachnospiraceae UCG-10; and two different ASVs in the Akkermansia genera-one increased on Farm 1, and one increased on Farm 2 (Fig. 4). The fact that these three taxa (RF39, Lachnospiraceae UCG-10, Akkermansia) emerged as CWD-associated on two independently run farms over 100 miles apart is intriguing and merits further attention. In a previous study, RF39 was found to be increased in a mouse model of the relapse-remitting form of multiple sclerosis (MS) 71 , which is a disease that shares many features with prion diseases, including CJD 72 .
Besides RF39, we also observed an increase in Lachnospiraceae UCG-010 in CWD-positive animals on both farms at the L7 level (Fig. 4b, c). Lachnospiraceae taxa have been reported in other studies on wild and captive deer gut microbiota 73,74 . Lachnospiraceae has also been noted in association with neurologic diseases. However, it is decreased, rather than increased, in studies on Parkinson's disease (PD) 75,76 , Alzheimer's Disease (AD), and amyotrophic lateral sclerosis (ALS) 28,77 . Moreover, multiple studies highlight the ability of Lachnospiraceae species to produce butyrate which helps maintain the epithelial barrier 78,79 . However, Lachnospiraceae family taxa have also been associated with type 2 diabetes 79 and intestinal inflammation 80 .
Like Lachnospiraceae, Akkermansia is commonly associated with health 81,82 and has been shown to reduce pathological alterations (amyloid beta-protein accumulation) and cognitive impairments in one mouse model of AD 83 . However, more recent evidence has promoted caution in defining Akkermansia as a "good bug" 84 . In fact, the mucin-degrading Akkermansia is reportedly increased in multiple neurologic diseases, including PD, multiple sclerosis, and AD 32,[85][86][87][88] . Akkermansia has additionally been associated with fasting and malnutrition, as it can utilize host mucin as its sole energy source while other microbes rely on dietary substrates consumed by the host 57,89 . Importantly, recent work has positively correlated Akkermansia muciniphilia abundance with M-cell density and function in the gut 90 , an intriguing finding as M-cell density is linked with increased susceptibility to orally acquired prion diseases 6 .
How and why these three taxa (RF39, Lachnospiracea UCG-010, Akkermansia) are associated with CWD are the next important questions to answer. Do these taxa contribute to a gut environment that is more permissive to orally-ingested prions? Can Akkermansia directly or indirectly stimulate M cell density and function? For example, Akkermansia degrades host mucin, thinning the protective mucus barrier that lines the gut. In concert with a pro-inflammatory Lachnispiraceae species, these microbes could create an inflammatory environment that induces colonic M-cells 11,14,15,91 , potentially enhancing prion uptake 6,92 . Gut inflammation has also been linked to the progression of neurodegenerative disease including AD and PD [30][31][32] . Alternately, are these taxa increased as a result of prion disease? Early clinical signs of CWD can include behavioral and locomotive changes followed by eventual wasting and weight loss 1 . Subtle behavioral changes could conceivably alter diet and drive dietary differences in the gut microbiota between deer with and without CWD. Akkermansia relies exclusively on host mucin as a nutrient source; therefore, Akkermansia could feasibly increase in relative abundance in a host that is consuming less food; although, anorexia is typically only associated with terminal stage disease, and most of these deer were pre-clinical. Finally, could these taxa be providing protective effects in the presence of a prion disease? Lachnospireacea and Akkermansia are associated with many health benefits, and increased relative abundances of these species are associated with protection against metabolic diseases and reduced pathological changes in AD 28,79,83,84,93 . Bacilli (e.g. RF39-formerly in phylum Tenericutes, now in Firmicutes) have also been posited to play a protective role in the gut as they are decreased in relative abundance in the presence of colitis 94 .
This study represents the first investigation, to our knowledge, of white-tailed deer gut microbiota in relation to CWD. We acknowledge several limitations to the present study. First, while our results suggest that differential diets are the major driver of microbial community differences by provenance and sex, we cannot explicitly rule out the potential effects of spatial proximity, host genetics, or biogeography. Polymorphisms in the deer PRNP gene can significantly influence susceptibility to CWD 95  www.nature.com/scientificreports/ interactions were not evaluated in this study. Second, as farmed and free-ranging deer had markedly different microbial communities, we cannot be certain that microbial composition differences observed in farmed deer based on CWD-status are generalizable to free-ranging deer. Third, microbial composition is not representative of microbial function 96 , and future studies using shotgun metagenomics and metabolomics are warranted to capture function. Fourth, while Akkermansia, RF39, and Lachnospiraceae UCG-010 are associated with CWD, further work is needed to clarify if these differences preceded or succeeded disease. Fifth, Farm 1 DNA samples underwent ethanol precipitation to improve DNA purity, which could have altered the microbial community of these samples in comparison to Farm 2 and free-ranging deer samples. However, inter-individual differences in gut microbiota largely overwhelm differences introduced by extraction method, including ethanol precipitation 97,98 . Finally, fecal samples from Farm 1 and free-ranging deer underwent library preparation and sequencing on an Illumina MiSeq at The Ohio State University Molecular and Cellular Imaging Center, while fecal samples from Farm 2 underwent library preparation and sequencing on an Illumina MiSeq at Argonne National Laboratory. While differences between laboratories and sequencing facilities can lead to differing results in microbiome studies 99 , we limited these effects by using the same methodology and kits (Qiagen PowerFecal) for all DNA extractions, the same region and primers for sequencing (V4 -515F and 806R), and all sequencing data was combined and underwent sequence processing and taxonomy assignment together. Further, results by sex and CWD status were analyzed independently for each location (Farm 1, Farm 2, Free-ranging).
In conclusion, we report differences in gut microbiota in white-tailed deer by provenance (Farm 1, Farm 2, Free-ranging), sex, and CWD status. Differences by provenance and sex are likely driven by diet, while differences by CWD status are more challenging to interpret and include increased abundances of Akkermansia, Lachnospireacea UCG-010, and RF39 taxa in CWD-positive deer. Priorities for future research include determining how these taxa may be associated with CWD susceptibility or pathogenesis, characterizing the gut microbiota of free-ranging cervids with CWD, examining the gut microbiota within a host before and after CWD infection, and assessing M-cell presence and abundance in CWD-positive and CWD non-detect animals to elucidate potential relationships between gut microbiota, M-cells, and chronic wasting disease. www.nature.com/scientificreports/