Inferred Causal Mechanisms of Persistent FMDV Infection in Cattle from Differential Gene Expression in the Nasopharyngeal Mucosa

Foot-and-mouth disease virus (FMDV) can persistently infect pharyngeal epithelia in ruminants but not in pigs. Our previous studies demonstrated that persistent FMDV infection in cattle was associated with under-expression of several chemokines that recruit immune cells. This report focuses on the analysis of differentially expressed genes (DEG) identified during the transitional phase of infection, defined as the period when animals diverge between becoming carriers or terminators. During this phase, Th17-stimulating cytokines (IL6 and IL23A) and Th17-recruiting chemokines (CCL14 and CCL20) were upregulated in animals that were still infected (transitional carriers) compared to those that had recently cleared infection (terminators), whereas chemokines recruiting neutrophils and CD8+ T effector cells (CCL3 and ELR+CXCLs) were downregulated. Upregulated Th17-specific receptor, CCR6, and Th17-associated genes, CD146, MIR155, and ThPOK, suggested increased Th17 cell activity in transitional carriers. However, a complex interplay of the Th17 regulatory axis was indicated by non-significant upregulation of IL17A and downregulation of IL17F, two hallmarks of TH17 activity. Other DEG suggested that transitional carriers had upregulated aryl hydrocarbon receptor (AHR), non-canonical NFκB signaling, and downregulated canonical NFκB signaling. The results described herein provide novel insights into the mechanisms of establishment of FMDV persistence. Additionally, the fact that ruminants, unlike pigs, produce a large amount of AHR ligands suggests a plausible explanation of why FMDV persists in ruminants, but not in pigs.


Introduction
Foot-and-mouth disease (FMD) is one of the most contagious and economically devastating viral diseases of livestock; the disease is caused by FMD virus (FMDV), a positivesense single-stranded RNA virus of the family Picornaviridae (genus Aphthovirus). Susceptible hosts include domestic and wild cloven-hoofed animals such as ruminants and pigs. Infection in cattle begins in the respiratory tract. During this primary infection, the virus replicates locally in the nasopharynx or lungs, depending on the route of exposure [1,2]. The infection subsequently spreads via systemic circulation (viremia) to secondary replication sites causing typical vesicles in the oral cavity, on the feet, and other sites of non-haired skin. Mortality is generally low in adult animals, but persistent infection can occur for long periods (30 days-5 years) in 50-80% of infected ruminants [3][4][5][6][7]. In contrast, persistent infection does not occur in pigs [8].
The site of persistent FMDV infection in cattle has been localized to the epithelial cells of the nasopharyngeal mucosa, including the dorsal soft palate and roof of the nasopharynx [6,[9][10][11]. Existing FMD vaccines do not prevent or cure persistent infection of pharyngeal epithelial cells [12]. Interestingly, in one set of experiments, FMDV in the oesophageal-pharyngeal fluid of persistently infected cattle was undetectable after dexamethasone treatment; however, virus levels returned to pretreatment levels after cessation of dexamethasone treatment [13]. Currently, the causal mechanisms of FMDV persistence are unknown.
To understand the mechanisms involved in FMDV persistence, a previous study applied an experimental, hypothesis-free functional genomics and bioinformatics approach to identify candidate mechanisms based on genes differentially expressed in tissues targeted and not targeted for persistent FMDV infection [14] and between the targeted tissues of carriers and non-carriers [15][16][17]. In that previous work, differential gene expression in nasopharyngeal tissues of carriers and non-carriers provided early evidence that type 1 regulatory T cells (Tr1) might play a role in persistent infection [15]. Further transcriptomic investigation using micro-dissected nasopharyngeal epithelia suggested that persistent FMDV infection was associated with compromised apoptosis and a reduced cellular immune response [16]. The continued analysis of the differentially expressed genes (DEG) in micro-dissected epithelia during persistent infection indicated that differential gene expression could affect the recruitment of neutrophils, antigen-experienced T cells and/or dendritic cells (DC), natural killer (NK) cell cytotoxicity, and the Th17 response in persistently FMDV-infected carriers [17]. The lung (a non-targeted tissue) was found to express significantly higher levels of TNF cytokines and their receptors than the pharyngeal tissues [14].
The current study provides further analysis of DEG from previously published data [16] collected during the transitional phase of infection that spans the period from acute to persistent infection. The main objective of this study was to infer potential causative factors and mechanisms of establishing FMDV persistent infection in cattle. Using a systems biology approach, we describe several hypothetical mechanisms for the establishment of persistent FMDV infection based on DEG in nasopharyngeal tissues, including contributory roles for aryl hydrocarbon receptor (AHR) ligands, leukocyte function, signaling pathways, and cytokines, chemokines, and their associated receptors.

Pathway and Gene Ontology Term Analysis
The probes with differential expression at FDR ≤ 0.1 showed that 1274 and 598 known genes were upregulated and downregulated, respectively, in nasopharyngeal epithelia of transitional carriers compared to terminators. The functional analysis of the upregulated DEG using DAVID tools detected significant enrichment in an immune-related gene ontology (GO) term (GO:0006955) and seven KEGG immune processes related to infection in T cells and epithelia, immune cell migration, phagocytosis, and four KEGG signaling pathways involved in immune regulation including (1) PI3K-Akt, (2) NFκB, (3) HIF-1, and (4) Wnt signaling pathways ( Table 1). The downregulated genes did not reveal any significant pattern in the same analyses.
Among these top five toxicity-inducing pathways, HIF1α and AHR signaling are mediated by two transcription factors that compete to form heterodimers with ARNT and play a critical role in regulating mucosal immunity [18,19]. NFκB signaling is crucial for the immune response [20]. On this basis, AHR, HIF1A, and NFκB signaling pathways were explored in more detail.  Figure 1. Top five upstream regulators (A) and top five biological processes/signaling pathways involved in toxicity (B) with the lowest likelihoods (i.e., p-value) of the associations/overlaps between the differentially expressed gene set (both up-and downregulated) and the pathways/biological processes by random chances in the Qiagen Ingenuity Pathway Analysis using the list containing ENTREZ numbers and up-and downregulated DEG. The dots in horizontal lines are the negative log transformation of p-values.

AHR and HIF1α Signaling
AHR and HIF1α compete to form heterodimers with aryl hydrocarbon receptor nuclear translocators (ARNTs). AHR was expressed at a significantly higher level (8.1-fold), and HIF1α expression was significantly downregulated by 1.8-fold in transitional carriers compared to terminators (Table 2). Among three ARNTs, ARNTL expression was the highest and was at a higher, although not significantly increased level (p = 0.05) in transi-

AHR and HIF1α Signaling
AHR and HIF1α compete to form heterodimers with aryl hydrocarbon receptor nuclear translocators (ARNTs). AHR was expressed at a significantly higher level (8.1-fold), and HIF1α expression was significantly downregulated by 1.8-fold in transitional carriers compared to terminators (Table 2). Among three ARNTs, ARNTL expression was the highest and was at a higher, although not significantly increased level (p = 0.05) in transitional carriers than terminators. The expression level of eight genes known to be induced by AHR signaling including B7H4 [21], CCL20 [22], CD8A [23,24], CD39 [25,26], CYP1B1 [19], IL6 [22,27], IL23A [28], and STAT3 [28] were significantly higher in transitional carriers than in terminators. A transcript variant of CD39 with a longer 3 non-coding sequence was also expressed at a higher level very close to significance (FDR = 0.11). Three AHR target genes, CYP1A1, CYP1A2, and IL33 [29], were also expressed at higher gene levels (p ≤ 0.05). Table 2. Average expression signal intensity (ESI), false discovery rates (FDR), and fold differences in AHR and HIF1A signaling-related genes that were differentially expressed between the nasopharynx epithelia of transitional carriers and terminators. HIF1 α expression is inducible by STAT3 and NFκB [30]. Three inhibitors of STATs (PIAS2, PIAS3, and PIAS4) were significantly upregulated at or close to significant levels in transitional carriers (DEG in NFκB signaling are listed in Table 3). PDK1 expression level, inducible by HIF1A [31,32], was 37.6 times lower in transitional carriers than in terminators, whereas three key inhibitors of HIF1α, HIF1AN, LIMD1, and VHL [30,33] were expressed at higher levels (p = 0.04, FDR= 0.05 and p = 0.03, respectively) in carriers than in terminators. In contrast to enhancing glucose uptake and glycolysis of HIF1A, an enzyme (ACSS3) that catalyzes the first reaction of fatty acid metabolism was upregulated 9.6-fold. HIF1A is activated by the AKT-mTOR signaling pathway via extracellular ATP and TCR signaling [34]. Although AKT1 and AKT2 were upregulated in transitional carriers, two inhibitory genes (PIK3IP1 and TSC1) of this signaling pathway [35,36] were also significantly upregulated. These results indicate reduced HIF1α signaling but increased AHR signaling in the epithelia of transitional carriers, which could impact mucosal immune response. Table 3. Mean expression signal intensities (ESI) and expression differences (fold) in genes of canonical and non-canonical NFκB signaling pathways between of transitional FMDV carriers and terminators.
On the other hand, the transcription factors, receptors, or receptor ligands in noncanonical NFκB signaling such as RELB, NFKB2/p100, CD27, LTB, LTBR, and TNFRSF8/ CD30 [20] were expressed at significantly higher levels in transitional carriers. Other genes involved in non-canonical NFκB signaling such as MAP3K14/NIK, CD40LG, RANKL, and TNFRSF1B [20] were also upregulated at p ≤ 0.05. Complement membrane attack complexes can activate non-canonical NFκB by forming an Akt+ NIK+ signalosome on Rab5+ endosomes [50]. These results indicate increased non-canonical NFκB signaling and suppressed canonical signaling in the epithelia of transitional carriers. This may promote immune tolerance by inducing tolerogenic DC and Treg cells and suppressing the Th17 response [20,41,51].

Wnt Signaling
The expression of five Wnts (WNT4, WNT5A, WNT7A, WNT10B, and WNT16) was significantly higher in transitional carriers than in terminators (Table 4), and WNT3 was at a higher level (p = 0.02). These results, together with the Wnt signaling pathway significantly overrepresented by DEG in the KEGG pathway analysis (Table 1), indicated increased Wnt signaling in carriers. This may indicate induction of tolerogenic DC that can inhibit Th17 and CD8+ cytotoxic T cell activity and promote Treg development, as described previously [52][53][54].

Myeloid Cell-Associated Factors
The expression of eight genes with immune inhibitory effects on macrophages or antigen-presenting cells (APC) (CD83, CD300D, EMR1, MFSD6, SIGLEC11, SIGLEC15, TIMD4, and TLR2) was significantly upregulated in transitional carriers compared to terminators (Table 8). Signaling through cell-membrane-associated CD83 appears to suppress functions in various immune cell populations [120], and soluble CD83 inhibits human monocyte differentiation into dendritic cells [121]. CD300 proteins are macrophage-specific receptors with regulatory effects [122,123]. CLEC1A dampens dendritic cell activation and downstream Th17 responses [124]. EMR1 mediates the induction of antigen-specific efferent regulatory T cells in peripheral tolerance [125]. MFSD6 is a mediator of MHC haplotypedependent but not MHC-unrestricted cytotoxicity of macrophages [126]. SIGLEC11 and SIGLEC15 are mainly expressed on macrophages and have an immunosuppressive effect on macrophages [48,127]. TIMD4, expressed only on APC including macrophages, mediates the removal of antigen-specific T cells during the contraction phase of the adaptive immune response [128,129]. TLR2 is a Toll-like receptor that can also induce immune tolerance [130][131][132][133][134]. On the other hand, MFSD6 recognizes certain MHC-I molecules and mediates MHC-I restricted killing by macrophages [126]. MFSD6 expression was 4.1 times lower in transitional carriers compared to terminators (Table 6). These results suggest increased activity of immunosuppressive macrophages and/or dendritic cells in transitional carriers.

Innate Immunity
Transitional carriers had a generally downregulated expression of defensin genes with two genes (DEFB1 and DEFB103A) significantly downregulated by 3.4-and 10.7-fold and one defensin gene (DEFB4B) significantly upregulated by 10.8-fold in transitional carriers (Table 8). DEFB103A is a broad-spectrum antimicrobial and has anti-picornavirus activity [135,136], which played a role in FMDV persistence. NID1 (a soluble NCR2 ligand with NK cell suppressing activity) [137] and MADD (an apoptosis-inhibiting gene) [138] were expressed at significantly higher levels (12.1-and 9.7-fold higher, respectively) in transitional carriers than those in terminators (Table 9). Table 9. Average expression signal intensities (ESI), false discovery rates (FDR), and fold differences of innate and humoral immunity-related genes differentially expressed between the nasopharynx epithelia of transitional carriers and terminators.

Discussion
Historically, it has been reported that approximately 50% of FMDV-infected ruminants remain persistently infected 28 days after infection [3][4][5]; however, experimental studies have shown that the proportion of carriers is often substantially higher [6,11].
Persistent infection does not occur in pigs [8], indicating the involvement of hostspecific factors in determining the divergence between FMDV carriers and terminators. The immune mechanisms inferred in this study are consistent with several hypothesized mechanisms identified in nasopharyngeal tissues during persistent infection [17], including (1) reduced recruitment of neutrophils and CD8+ T effector cells, (2) suppressed NK and macrophage cytotoxicity via downregulated MFSD6 and NID1, and (3) suppression of the Th17 response and canonical NFκB signaling pathway. Additionally, previous work demonstrated that expression of chemokines that recruit neutrophils and CD8+ T effector cells was reported to be significantly lower in pharyngeal tissue than in the lung, where primary, but not persistent, infection occurs [14,139], indirectly supporting the involvement of these chemokines in preventing FMDV persistent infection.
The differential expression of ELR+ CXCLs and CCL3 is consistent with the microscopic analyses of the nasopharynx of the animals included in this study, wherein there were reduced quantities of CD8+ T cells in the epithelia of transitional carriers compared to terminators [16]. These results suggested the importance of neutrophil and CD8+ T effector cell recruiting chemokines in FMDV clearance, given that CD8+ cytotoxic T cells kill infected cells and neutrophils can clear virus infection via phagocytosis and extracellular traps [140]. This also agrees in part with the finding from one study that dexamethasone injection inhibited FMDV production in the oesophageal-pharyngeal fluid of persistently infected cattle but did not cure the infection [13]. This is based on dexamethasone treatment causing neutrophilia, lymphopenia, and eosinopenia in ruminants [141][142][143].
IL-17RA signaling in the epithelium, activated by two Th17-specific cytokines, IL17A and IL17F, is required for neutrophil recruitment. During the transitional phase, the involvement of Th17 cells in the FMDV infection of pharyngeal epithelia was strongly supported by upregulated expression of Th17-promoting cytokines and chemokines (IL6, IL23A, CCL14, CCL20, and CCR6 listed in Tables 5 and 6) and Th17-associated genes (CD4, CD146, MIR155, TGFBR3, and ThPOK shown in Table 7) in transitional carriers, suggesting that the Th17 response was needed to clear FMDV. However, the expression of Th17 cytokines (IL17A, IL17F, and IL22) was not upregulated in transitional carriers, indicating that the activity of Th17 cells was suppressed, potentially as a result of upregulated Th17 suppressing genes CD39, CD69, IL16; LXRA, STAT5A, TIMD4, and TNFRSF6B ( Table 7). The suppression could also be mediated by several upregulated immune suppressive genes including ADCY4, BTLA, GIMAP5, and IL34 based on publications cited herein.
Effects of AHR signaling can also be mediated through dimerization with other transcription factors such as estrogen receptors [157][158][159], HIF1A [18,[160][161][162][163]; NFκB [164], PPAR [165,166]. Interestingly, estrogen receptor, ESR1, was detected as a top up-stream regulator, and DEG were over-represented in the PPAR signaling pathway in this study. HIF1A and AHR compete to form heterodimers with AHR nuclear translocators (ARNTs) and mutually inhibit each other. Reduced HIF1A signaling inhibits IL17 production in CD4+ T cells and cytotoxicity of CD8+ T cells contributing to T cell exhaustion in chronic infections [167]. AHR:RelA dimerization antagonizes the classical NFκB pathway, whereas AHR:RelB enhances non-canonical pathway signaling [164,168]. The DEG listed in Table 3 indicate higher non-canonical and lower canonical NFκB signaling in transitional carriers compared to terminators, according to the review article by Sun (2017), which could suppress the Th17 response according to IL17 signaling in the canonical NFκB pathway [169]. The non-canonical NFκB pathway plays an important role in promoting immune tolerance by inducing tolerogenic DC and Treg cells [20,41], suppressing the Th17 response [51]. Additionally, AHR signaling enhances Wnt signaling [103], and both Wnt and AHR signaling can induce tolerogenic DC [52,54].
Our results demonstrated upregulated AHR and its several target genes and downregulated HIF1A and its target genes in transitional carriers (listed in Table 2). AHR is a promiscuous xenobiotic receptor and ligand-dependent transcription factor that binds to various chemicals such as plant flavonoids, polyphenolics, and indoles as well as to pollutants such as synthetic polycyclic aromatic hydrocarbons and dioxin-like compounds [170]. Interestingly, some short chain fatty acids (SCFA), e.g., propionate and butyrate, are also AHR ligands [171] and can induce AHR expression [172,173] and increase cell response to AHR ligand stimulation [174]. Some of the AHR ligands are produced in the rumen as part of normal ruminant physiology, suggesting an interesting hypothetical mechanism to explain why ruminants, but not pigs are prone to persistent FMDV infections.
It is well-established that B-cell function is altered during the FMDV carrier state. Specifically, anti-FMDV IgA detection in secretions has been reported to be significantly higher in carriers than in non-carriers [175,176]. This indicates chronic stimulation of B cells and suggests that antibodies alone cannot clear the FMDV carrier state, but rather cell-mediated immunity is required. In the current study, B-cell induction was indicated by upregulation of CD21, CD19, and CD81 (the B cell co-receptor complex) and 16 immunoglobulin probes in transitional carriers (data not shown).
In summary, this work supports previous studies that indicated that the establishment and maintenance of the FMDV carrier state are associated with differential gene expression in the nasopharyngeal tissues, known to be the site of persistent infection. Specifically, pathway analysis of DEG suggested that several immune regulatory mechanisms are associated with FMDV persistence. DEG of cytokines, chemokines, and the receptors suggest an increased differentiation and migration of Th17 cells but reduced recruitment of neutrophils and CD8+ T effector cells to the infected tissues of transitional carriers compared to terminators. However, IL17A and IL17F expression were not increased in carriers, indicating complex regulation of the Th17 response. The Th17 response is known to play a key role in inducing the expression of neutrophil recruiting chemokines, which is regulated, in part, by AHR signaling. Upregulated AHR signaling in carriers was also supported by DEG in NFκB and Wnt signaling pathways. This leads us to speculate that AHR ligands produced in the rumen and their effect on various physiological functions might play a role in establishing FMDV persistent infection, which could also explain why FMDV persists in some ruminants but not in pigs.

Study Design and Gene Expression Data
The microarray data used in this study and the design of the animal experiments have been reported previously [16]. All data utilized in this study were derived from microarray-based gene expression profiles of micro-dissected nasopharyngeal epithelia from FMDV-infected cattle during the transitional phase of infection spanned from 12 to 21 dpi. The previous works defined the transitional phase as the period after acute infection but before the defined carrier phase; animals that remained infected during the transitional phase (transitional carriers) consistently progressed to becoming carriers [6].
The data were produced using a custom bovine gene expression 60-mer oligonucleotide microarray as described by Zhu et al. [14]. Microarrays and reagents were manufactured by Agilent Technologies (San Jose, CA, USA), and the lab procedures were conducted based on the protocols and equipment recommended by the manufacturer. For comparison of the gene expression levels between transitional carriers (animals that were still infected) and terminators (animals that had recently cleared infection) during the transitional phase of infection, microarray expression data from the micro-dissected nasopharyngeal epithelia of three animals from each cohort were compared, as previously reported by Stenfeldt et al. [16].

Statistical Analysis
R scripts implemented with the LIMMA package [177] were used to normalize and analyze the microarray data as previously described [15]. All signal intensities (averaged photons per pixel in the microarray images) used in the statistical analysis were Log2 transformed. Genes differentially expressed between transitional carriers and terminators with a false discovery rate (FDR) of 0.10 or smaller and an expression difference of at least 50% were considered statistically significant genes in this study. This FDR threshold increases the detection power (fewer false negatives/type II errors) with a false positive (type I error) rate of 0.10 in declared DEG, or one expected false positive in ten DEG, compared to FDR at 0.05, which balances type I and type II errors.

Pathway Analysis
The methods of pathway analysis of DEG have been described [17]. All bovine genes included in the microarray design were mapped to human reference genes using computer analysis via NCBI BLAST and/or manual annotation by aligning the microarray probe sequences on bovine genome sequences on the UCSC Genome Browser using BLAT (https://genome.ucsc.edu/index.html (accessed on 1 June 2022)). The list of upregulated and downregulated genes associated with each human Entrez Gene ID was analyzed with Ingenuity Pathway Analysis (IPA) (Qiagen, Germantown, MD, USA) and the NCBI Functional Annotation Bioinformatics Microarray Analysis program (DAVID Bioinformatics Resources version 6.8) to identify the biological pathways significantly over-represented among DEG. The biological functions of DEG were determined based on scientific publications (included as cited references) or on the NCBI Gene database (https://www.ncbi.nlm.nih.gov/gene/ (accessed on 1 June 2022)).

Biological Inferences
The biological inferences have been described [17], which were based on (i) reported biological functions of DEG, (ii) differential gene expression including averaged signal intensity and magnitudes (fold difference) of upregulated or downregulated expression, assuming that (1) genes with a higher signal intensity and larger differential expression have a more substantial biological role in their gene group and (2) upregulated expression enhances gene activities and vice versa. Differential expression of genes with cell-specific expression was also used to infer the differences in the number of the cells. Genes with no significant differential expression (FDR > 0.10) but known to play important roles in the relevant biological pathways/processes associated with other DEG were also used as references or supporting results for DEG-related mechanisms. Probabilities of differential expression at gene levels are listed as p-values along with the FDR. Expression levels of genes downregulated or upregulated in transitional carriers compared to terminators are shown as negative and positive values (fold changes), respectively. Immune regulatory mechanisms especially involved in mucosal immunity and its association with ruminant physiology were also taken into consideration in the formulation of the hypothesis.

Data Availability Statement:
The animal experiments of the microarray studies have been published, and the original raw data are available in the NCBI databases (accession number: GSE104058) (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc = GSE104058 (accessed on 1 June 2022)). The datasets generated for this study are located at: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? acc=GSE104058 (accessed on 1 June 2022).