Type I IFN stimulates lymph node stromal cells from adult and old mice during a West Nile virus infection

Abstract Advanced age is a significant risk factor during viral infection due to an age‐associated decline in the immune response. Older individuals are especially susceptible to severe neuroinvasive disease after West Nile virus (WNV) infection. Previous studies have characterized age‐associated defects in hematopoietic immune cells during WNV infection that culminate in diminished antiviral immunity. Situated amongst immune cells in the draining lymph node (DLN) are structural networks of nonhematopoietic lymph node stromal cells (LNSCs). LNSCs are comprised of numerous, diverse subsets, with critical roles in the coordination of robust immune responses. The contributions of LNSCs to WNV immunity and immune senescence are unclear. Here, we examine LNSC responses to WNV within adult and old DLNs. Acute WNV infection triggered cellular infiltration and LNSC expansion in adults. Comparatively, aged DLNs exhibited diminished leukocyte accumulation, delayed LNSC expansion, and altered fibroblast and endothelial cell subset composition, signified by fewer LECs. We established an ex vivo culture system to probe LNSC function. Adult and old LNSCs both recognized an ongoing viral infection primarily through type I IFN signaling. Gene expression signatures were similar between adult and old LNSCs. Aged LNSCs were found to constitutively upregulate immediate early response genes. Collectively, these data suggest LNSCs uniquely respond to WNV infection. We are the first to report age‐associated differences in LNSCs on the population and gene expression level during WNV infection. These changes may compromise antiviral immunity, leading to increased WNV disease in older individuals.


| INTRODUC TI ON
West Nile virus (WNV) is a positive sense RNA virus that causes seasonal outbreaks across several continents and continues to threaten public health. WNV is the leading cause of mosquitoborne disease in the United States, however, most infections are asymptomatic (McDonald et al., 2021). In some cases (<1%), WNV can infect the central nervous system (CNS), leading to severe neuroinvasive disease and encephalitis. Following the bite of an infected mosquito, WNV enters the draining lymph node (DLN), from which it replicates, enters the bloodstream, disseminates into peripheral organs, and eventually reaches the CNS . Early innate cues induce inflammatory cytokines that coordinate robust adaptive immune responses, all of which are critical for the containment of WNV disease Samuel & Diamond, 2005). This coordination is centralized in the DLN where hematopoietic immune cells interact with neighboring stromal cells in lymphoid tissue to facilitate rapid yet controlled adaptive immune responses.
LNSCs contribute to anti-viral immunity by integrating inflammatory signals to initiate activation and stimulate adaptive immune responses. Innate signaling pathways, like type I interferon and Toll-like receptors, are stimulated in FRCs and endothelial cell subsets (Gregory et al., 2017;Krishnamurty & Turley, 2020;Malhotra et al., 2012), and an absence of LNSC subsets can lead to attenuated cell-mediated and humoral immune responses during viral infection (Cremasco et al., 2014). LNSC function can also be disrupted by a viral infection, leading to altered lymphoid architecture and, later, impairments in lymphocyte accumulation and organization (Fiorentini et al., 2011;Steele et al., 2009). Infection of FRCs by Ebola (EBOV) and lymphocytic choriomeningitis (LCMV) viruses display shrunken morphology and impaired antigen drainage, respectively (Fiorentini et al., 2011;Steele et al., 2009). However, the mechanism by which LNSCs recognize and respond to an ongoing WNV infection has not been elucidated.
Furthermore, severe neuroinvasive WNV disease strongly correlates with age in humans, despite similar levels of infection rates across different age groups (Centers for Disease Control and Prevention (CDC), 2019). This disparity is due in part to an age-dependent decline in innate and adaptive immunity known as immune senescence (Nikolich-Žugich, 2018). The mechanisms behind immune senescence are multifactorial. Understanding the mechanisms which lead to this age-dependent susceptibility is critical to develop effective prophylactic and therapeutic strategies.
Animal studies of WNV infection recapitulate the age-dependent decline in antiviral immunity and increased susceptibility. Aged mice infected with WNV display higher viral burdens across multiple organs and increased disease severity (Brien et al., 2009;Funk et al., 2021;Richner et al., 2015). Defects in T cell activation and differentiation, as well as diminished long-lived plasma cell development, have all been attributed to increased disease severity in the aged setting (Brien et al., 2009;Richner et al., 2015). A growing body of evidence suggests that the stromal cells of the lymph node can also contribute to this immune senescence. In both old mice and old non-human primates, the lymphoid tissue microenvironment is disorganized and exhibits a greater level of fibrosis (Kwok et al., 2022;Thompson et al., 2017). Previously, we attributed reduced trafficking of naïve T cells into and within the DLN to defects in the microenvironment of the aged lymph node (Richner et al., 2015). The cellular mechanism leading to these age-dependent LNSC defects and how they contribute to the disintegration of the aged immune response, especially during WNV infection, is unclear.
In this study, we examined how LNSCs recognize and respond to an acute WNV infection in both adult and aged settings. Using a mouse model of WNV, we identified age-dependent differences in LNSC population dynamics following an infection. LNSCs from adult and old mice are activated upon WNV infection, primarily through type I IFN signaling. Activation-induced similar antiviral gene expression patterns in adult and old LNSC cultures. However, LNSCs from old mice also constitutively upregulated several immediate early response (IER) genes associated with immune suppression. Together these data suggest that LNSCs from adult and old hosts generate an antiviral response to WNV infection through a predominantly IFN-driven mechanism of activation. Age-associated changes in LNSC composition may contribute to the blunted adaptive immune responses and increased severity of WNV infections in older individuals.

| Aged lymph nodes display alterations in LNSC subsets following WNV infection
To investigate the age-dependent changes in LNSC subsets in a West Nile virus (WNV) infection, we first assessed changes in total cellular dynamics in the draining lymph node (DLN) throughout the acute course of an infection. The rear footpads of adult (8-10 weeks of age) and old (18 months of age) C57BL/6J mice were infected subcutaneously with 1 × 10 3 FFU WNV-Kunjin and popliteal DLNs harvested at 2, 4, 6, and 10 days post-infection (dpi), or from uninfected mice (0 dpi). In a previous study, we demonstrated that WNV-Kunjin, a less pathogenic variant of WNV, elicits similar lymphocyte trafficking and acute immune responses in the DLN, compared to the more pathogenic strain, WNV-NY99 (Richner et al., 2015). Following inoculation of adult mice with WNV-Kunjin, total cell numbers in the DLN rapidly increased, going from 2.3 × 10 6 total cells in the uninfected DLN to 11.9 × 10 6 total cells by 2 days post-infection ( Figure 1a). The total number of cells remained elevated in the adult mice at days 2, 4, and 6 post-infection before decreasing from day 6 to 10 as the infection resolved. On the contrary, old DLNs began with 1.4 × 10 6 total cells and increased to only 5.7 × 10 6 total cells by 2 days post-infection. Total cellularity remained more than 2-fold lower in the old DLN at days 4 and 6 post-infection.
In addition to measuring total cellularity, we also examined differences in major LNSC subsets in these adult and old infected DLNs. To quantify LNSCs in WNV-infected DLNs, we enzymatically digested popliteal lymph nodes after WNV infection and stained single cell suspensions for CD45, Podoplanin (Pdpn), and CD31 to identify the major LNSC subsets: FRCs, LECs, and BECs. At resting state, the stromal subset found in the highest frequency were the FRCs, followed by LECs and then BECs (Figure 1b), similar to the frequencies reported by others (Fletcher et al., 2011;Malhotra et al., 2012).
In the adult mice, the numbers of CD45 − LNSCs in the DLN peaked at day 6 post-infection. LNSC accumulation in the DLN of the old mice had blunted and delayed kinetics, with peak accumulation at 10 dpi ( Figure 1c). This delayed accumulation of LNSCs in the aged mice correlated with delayed proliferation as signified by Ki67 + staining, a marker for cell division. Adult LNSC proliferation peaked at days 4 and 6 post-infection with Ki67+ staining in greater than 50% of CD45 − cells. Old LNSCs displayed a delayed pattern of proliferation, peaking at day 6, but was significantly reduced compared to adult at 4 dpi ( Figure 1d). The dominant proliferating LNSC populations were the FRCs and LECs, with the BECs showing less Ki67 staining over the acute infection course ( Figure S1). The population of FRCs, BECs, and DNCs in the adult mice displayed the same pattern of cellular dynamics, with a peak at day 6 ( Figure 1e). The FRC population in the old mice remained stable from 0 to 6 days postinfection, then increased by 10 dpi. The number of BECs peaked in adult DLNs at 6 dpi, while old LNs displayed a 1.7-fold reduction in this subset. Intriguingly, the number of LECs was consistently lower in the old DLN compared to the adult DLN (p < 0.05 at days 4 and 6 post-infection). At 4 dpi, old DLNs displayed a significant 6.3-fold F I G U R E 1 Stromal cell dynamics are altered in aged lymph nodes following WNV infection. Adult (8-10 weeks) and old (18 months) C57BL/6J mice were subcutaneously infected with 1 × 10 3 FFU of WNV. Draining popliteal LNs were harvested from infected mice at days 2, 4, 6, and 10 after infection and digested for flow cytometry. (a) Numbers of total cells per DLN. (b) Representative dot plots show the gating strategy used to identify LNSCs in the lymph nodes of adult and old mice; percentages on plots represent frequency of cells within the total CD45 − population (left plot). CD45 − LNSCs were divided into stromal subsets based on Pdpn and CD31 staining (right plot): FRC, LEC, BEC, and DNC. Numbers of (c) total CD45 − LNSCs and (e) FRC, LEC, BEC, and DNC stromal subsets from WNV-infected DLNs of adult and old mice. (d) The frequency of proliferating Ki-67 + CD45 − LNSCs from WNV-infected DLNs was quantified with flow cytometry. (f) Ratios of FRC:LEC:BECs from adult and old DLNs at day 4 following WNV infection. The results are averaged from 4 independent experiments with 3-18 mice per timepoint. Data are expressed as the mean ± the standard error of the mean (SEM). Statistically significant differences between adult and old groups at each timepoint are denoted by asterisks (*p < 0.05; ***p < 0.001; ****p < 0.0001; Multiple unpaired T tests).

| LNSCs support WNV replication and are potent producers of inflammatory cytokines
Being the structural network woven throughout the DLN, LNSCs are in the direct presence of viral antigen during an infection. In addition, FRCs and LECs can become infected themselves and serve as a reservoir for viral progeny (Fiorentini et al., 2011;Mueller et al., 2007;Steele et al., 2009). Thus, we first characterized WNV replication in LNSC cultures. To establish lymph node FRC cultures, we iso- To model LNSC stimulation and define the mechanisms by which LNSCs become activated during WNV infection, we adapted an in vitro culture system that contained the major LNSC subsets (Fletcher et al., 2011). Following the digestion of pooled skin-draining lymph nodes from adult mice, we plated cells on gelatin-coated plates.
We then sought to characterize LNSC activation after a WNV infection. Previous microarray data demonstrated that inflamed FRCs, LECs, and BECs highly expressed Cxcl9 (Malhotra et al., 2012).
CXCL9 is an inflammatory chemokine that drives immune cell migration into and within the DLN and strongly polarizes T cells toward an antiviral T H 1 lineage (Antonelli et al., 2010;Groom et al., 2012).
To test Cxcl9 as an activation readout, LNSC cultures were infected with WNV for 20 h. indicating that WNV can stimulate LNSCs.

| WNV-induced type I IFN activates LNSCs
Within the DLN during infection, LNSCs are exposed to viral antigen and secreted pro-inflammatory cytokines (Krishnamurty & Turley, 2020;Malhotra et al., 2012;Richner et al., 2015). To mimic this inflamed microenvironment, we infected murine bone marrow- Indeed, delivery of poly (I:C) with an anti-IFNAR1 antibody, which blocks type I IFN signaling, completely inhibited Cxcl9 upregulation. These data indicate that type I IFN is a dominant driver of LNSC activation in the context of a viral infection.
To further define the mechanism of LNSC activation, we depleted individual factors from the BMDC-WNV supernatants with monoclonal antibodies. Cxcl9 can be induced by IFNƔ (Antonelli et al., 2010).
When LNSC cultures were pre-treated with anti-IFNGR antibody to block type II IFN signaling and stimulated with infected supernatant, the result was a relatively modest two-fold reduction in Cxcl9   blocked. This data revealed that activation of adult and old LNSCs occurred in the same type I IFN-driven mechanism, and these signaling pathways are stable in advanced age.
We further investigated aged LNSC function by assessing the global gene expression pattern in stimulated aged LNSCs. RNA from adult and old LNSCs that were untreated or stimulated overnight with BMDC-WNV supernatant was collected for RNA sequencing.
Using 2-factor multi-group analysis, analogous to two-way ANOVA, we calculated the effect of stimuli treatment or age on differential gene expression. The interaction term, which captures DEGs where the effect of treatment depended on age, was also calculated. We observed that stimuli treatment significantly drove the variability seen in LNSC gene expression (6422 DEGs, FDR <0.05; Table S2).
Age and treatment did not interact to affect gene expression (0 DEGs, FDR <0.05; Figure S5, Table S2). Activated LNSCs from old mice showed regulation of 1996 DEGs (1045 upregulated and 951 downregulated) compared to untreated old LNSCs (Table S2). The total number of DEGs in activated, old LNSCs was much lower than those in activated adult LNSCs, as shown in Figure 3c  Several of these cytokines, especially IL6 and GM-CSF, comprise the senescence-associated secretory phenotype (SASP) typically seen in aged cells (Robbins et al., 2021). This data shows that cytokine production and secretion in activated, old LNSCs also remains intact.
To determine if age had any significant impact on LNSC gene expression, we performed pairwise comparisons between untreated adults versus old LNSCs or activated adults versus old LNSCs. These comparisons revealed the age-dependent regulation of 20 and 25 DEGs in untreated and BMDC-WNV-stimulated groups, respectively (Table S3) Relative fold gene expression of Nr4a1, Zfp36, and Fos at baseline in adult and old LNSCs, measured by qRT-PCR. The results are averaged from two independent experiments with LNSCs isolated from 2-3 mice per experiment. Data are expressed as the mean ± SEM. Statistical significance is denoted by asterisks, comparing differences between adult and old groups (*p < 0.05; **p < 0.01; Unpaired t test). (c) Adult or old mice were infected with 1 × 10 3 FFU WNV and draining popliteal lymph nodes were harvested at 2 days post-infection. Lymph nodes were cyro-sectioned and stained for B220 (blue) and ZFP36 (green). Scale bar = 100 μm. have observed previously in the aged setting (Richner et al., 2015).
Our results counter findings with the influenza virus. Adult and old mice had similar total cell numbers in the lung-draining mediastinal lymph node following an influenza infection, but old mice had reduced numbers of FRCs and LECs (Masters et al., 2019). We postulate that these differences are due to lymph node location as well as pathogen-specific properties. Mucosal antigen exposure within mediastinal lymph nodes throughout life may lead to differences in LNSC priming and expansion, compared to more isolated skin DLNs.  (Krishnamurty & Turley, 2020;Rodda et al., 2018).
Therefore, further studies utilizing single-cell RNA sequencing will be required to precisely determine virally induced gene expression patterns in the heterogenous LNSCs.
Other findings have identified numerous functional defects in aged LNSCs. Aged LECs suffer from delayed proliferation and increased vessel permeability upon viral challenge (Masters et al., 2019;Zolla et al., 2015). The expansion potential and B/T cell zone architecture of the DLN are shunted due to defects in aged FRCs (Masters et al., 2019;Textor et al., 2016). These aged FRCs formed disrupted reticular cell networks and produced less T cell chemokines CCL19 and CCL21 following viral infection (Masters et al., 2019;Sonar et al., 2022). Aged BECs are defined by altered expansion kinetics and cellular morphology, the latter of which impedes lymphocyte entry (Masters et al., 2019). To investigate the contribution of aged LNSCs to immune senescence in our model system, we compared adult and old LNSC cultures following stimulation. Counter to our initial hypothesis, LNSCs from adult and old mice robustly induced cytokine expression at equivalent levels.
Further, we did not observe any differences in the gene expression Despite the overall similarity between the adult and old LNSC cultures, we found that aged LNSCs upregulated immediate early response (IER) genes in both stimulated and unstimulated conditions.
Of the upregulated IER genes, Nr4a1, Zfp36, and Fos are also associated with immune suppression. Nr4a1 encodes for Nur77, a transcription factor with roles in inflammation and T cell responses (Li et al., 2015;Liu et al., 2019). Old Nur77-deficient mice suffered from systemic inflammation with severe immune cell infiltration and IL6, TNFα production (Li et al., 2015). Zfp36 encodes for tristetraprolin (TTP), a RNA binding protein that destabilizes target transcripts, especially proinflammatory Tnfα and Il6 (Moore et al., 2018). FOS is an AP-1 transcription factor subunit (Bahrami & Drabløs, 2016), and lack of Fos expression in vitro and in vivo led to increased IL6 and TNFα production following LPS treatment (Ray et al., 2006). Despite the documented roles of these IER genes in inflammation, these genes

| Virus and cells
The WNV-Kunjin strain (CH16532) was generously provided by Michael Diamond at Washington University in St. Louis. Viral stocks were propagated in Vero-E6 cells and titers were determined by focus-forming assay (FFA) as previously described (Brien et al., 2013), using an anti-WNV E16 monoclonal antibody (anti-WNV E16 mAb) (Nybakken et al., 2005).

| Mice and WNV infection
Adult (

| Antibodies
The fluorochrome or biotin-conjugated primary antibodies used for flow cytometry and immunofluorescence microscopy were

| Lymph node digestion for LNSC analysis
The LN digestion protocol described below was adapted from a previous protocol (Fletcher et al., 2011).

| LNSC enrichment and cell culture
Following digestion, LN cells were resuspended in 10 mL cell media and plated on 10 cm dishes coated with 0.2% gelatin (Sigma).

| Ex vivo stimulation
Bone marrow-derived dendritic cells were plated in 12-well plates at 10 5 cells/well and, following attachment overnight, were infected  Filtered reads were aligned to the mouse reference genome mm10 from Ensembl using STAR (Dobin et al., 2013). Genomic features were quantified against Ensembl annotations for mm10 using fea-tureCounts (Liao et al., 2014). Gene expression was normalized to counts per million units and averaged across sample groups, followed by principal component analysis to assess sample clustering.
Differential gene analysis was performed in edgeR using a 2-factor multi-group analysis to determine the effect of treatment and/or age (Robinson et al., 2010). As well, pairwise comparisons were performed between pairs of age and treatment groups, for example, "Adult Control versus Old Control" or "Old BMDC-WNV versus Old Control." The gene expression profiles from adult and old LNSCs following ex vivo stimulation were analyzed for significant enrichment of functional pathways using the ToppFun application within the ToppGene Suite (Chen et al., 2009). Upregulated differentially expressed genes (FDR < 0.05) with a log2-fold change >0 were used for pathway enrichment analysis. Results were filtered to only REACTOME pathway terms with FDR < 0.05 (Jassal et al., 2020).

| Quantitative cytokine analysis
Supernatants from stimulated adult and old LNSC cultures, or BMDC WNV supernatant alone were collected and stored at −80°C.

| Immunohistochemistry and fluorescence microscopy
Popliteal draining lymph nodes from adult and old mice were harvested 2 days post-infection with WNV-Kunjin. DLNs were frozen in O.C.T. blocks and 8μm-thick sections were cut using a cryostat.

| Data analysis
All data, excluding bioinformatic analyses, were analyzed using tively. The graphical abstract was generated using BioRe nder.com.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare they have no conflicts of interest.

S U PP O RTI N G I N FO R M ATI O N
Additional supporting information can be found online in the Supporting Information section at the end of this article.