Impact of a Demyelination-Inducing Central Nervous System Virus on Expression of Demyelination Genes in Type 2 Lymphoid Cells

We recently reported the role of type 2 innate lymphoid cells (ILC2s) in central nervous system (CNS) demyelination using a model of CNS demyelination involving recombinant herpes simplex virus 1 (HSV-1) that constitutively expresses mouse interleukin 2 (HSV-IL-2). In this investigation, we studied how ILC2s respond to HSV-IL-2 at the cellular level using cytokine and gene expression pro ﬁ ling. ILC2s infected with HSV-IL-2 expressed higher levels of granulocyte-macrophage colony-stimulating factor (GM-CSF), IL-5, IL-6, IL-13, IP-10, MIP-2, and RANTES, which include proin ﬂ ammatory cytokines, than did those infected with parental control virus. In contrast, T H 2 cytokines IL-4 and IL-9, which are typically expressed by ILC2s, were not induced upon HSV-IL-2 infection. Transcriptome sequencing (RNA-seq) analysis of HSV-IL-2 infected ILC2s showed signi ﬁ cant upregulation of over 350 genes and downregulation of 157 genes compared with parental virus-infected ILC2s. Gene Ontology (GO) term analysis indicated that genes related to “ mitosis ” and “ in ﬂ ammatory response ” were among the upregulated genes, suggesting that HSV-IL-2 infection drives the excessive proliferation and atypical in ﬂ ammatory response of ILC2s. This change in ILC2 activation state could underlie the pathology of demyelinating diseases. IMPORTANCE Innate lymphocytes have plasticity and can change functionality; type 2 innate lymphoid cells (ILC2s) can convert to ILC1 or ILC3 cells or change their activation state to produce IL-17 or IL-10 depending on environmental cues. In this study, we investigated the gene and cytokine pro ﬁ les of ILC2s, which play a major role in HSV-IL-2-induced CNS demyelination. ILC2s infected with HSV-IL-2 displayed a mas-sive remodeling of cellular state. Additionally, ILC2s infected with HSV-IL-2 differed from those infected with parental HSV in cellular and viral gene expression pro ﬁ les and in cytokine/chemokine induction, and they displayed enhanced activation and proin ﬂ ammatory responses. These changes in ILC2 activation state could underlie the pathology of demyelinating diseases. These results also highlight the possible importance of pathogens as environmental cues to modify innate lymphocyte functionalities.

associations, these concepts remain unproven (9)(10)(11). If an infectious agent is involved, it alone may not be sufficient to initiate MS pathology.
We recently showed that ILC2s play a role in initiating CNS demyelination induced by HSV-IL-2 infection (47), which led us to examine the direct response of type 2 ILCs to HSV-IL-2 infection ex vivo. Our approach included analysis of (i) the effect of HSV-IL-2 infection of ILC2s on their expression of cytokines and chemokines and (ii) the transcriptomes of ILC2s before and after HSV-1 infection. Our results indicate that HSV-1 infection and the presence of IL-2 induced an atypical and enhanced activation of ILC2s, which might promote the CNS demyelination pathology following ocular HSV-IL-2 infection in mice.
To further examine the similarities and differences in gene expression patterns induced by HSV-IL-2 and parental virus in ILC2s, we compared genes that had significant changes in expression after infection (Fig. 3). We found 507 genes upregulated in both HSV-IL-2-and parental virus-infected ILC2s (Fig. 3A). In addition, 549 and 130 genes were uniquely upregulated following infection with HSV-IL-2 and parental virus, respectively (Fig. 3A). We found 93 genes downregulated in both HSV-IL-2-and parental virus-infected ILC2s, while 265 and 152 genes were uniquely downregulated following infection with HSV-IL-2 and parental virus, respectively (Fig. 3B). Heat maps were used to identify genes that were either upregulated (red) or downregulated (blue) significantly upon HSV-IL-2-infection or parental virus infection (Fig. 3C). Genes, including  UL47, and UL52) (Fig. 4). Our results suggest that most, but not all, HSV-1 genes are expressed in infected ILC2s, and additional HSV-1 genes were expressed specifically in ILC2s infected with parental virus but not in HSV-IL-2-infected cells (57 genes versus 44 genes). HSV-1 encodes at least 80 genes (53)(54)(55), and a summary of HSV-1 genes expressed by both parental and HSV-IL-2 viruses and by parental but not HSV-IL-2 virus, as well as the genes that were not detected after infection with either virus, and their functions is shown in Table 2. Thus, in combination, HSV-1 and IL-2 suppressed expression of specific HSV-1 genes in infected ILC2s.

DISCUSSION
ILC2s are involved in immune responses, inflammation, metabolic homeostasis, and tissue remodeling (35,(37)(38)(39)(40)(41)(42)(43). Their dysregulation leads to diseases such as asthma and dermatitis, and we previously showed that ILC2s play an important role in HSV-IL-2-induced demyelination in the CNS (47). In this investigation, we studied how ILC2s respond to HSV-IL-2 infection at the cellular level using cytokine and gene expression profiling and found that cytokine expression in HSV-2-infected ILC2s was activated, increased, and atypical compared to conventional cytokine activation. The changes in ILC2 activation state may underlie the pathology of demyelinating diseases.
Recently, we have shown that IL-2 expressed by HSV-IL-2 binds to ILC2s (47). In this study, we have extended this finding using cytokine profiling to show that binding of virally expressed IL-2 to ILC2s is associated with higher expression of GM-CSF, IL-5, IL-6, IL-13, IP-10, MIP-2, and RANTES than seen in cells infected with parental virus or mock infected (Table 1). This is consistent with a study showing that cytokine activation of ILC2s induces downstream cytokines such as GM-CSF, IL-3, IL-6, IL-8, and IL-9 (56, 57). ILC2s have a cytokine profile similar to that of T H 2 cells (58, 59) and secrete IL-5 and IL-13 in response to IL-25, IL-33, and thymic stromal lymphopoietin (TSLP) (38). We were intrigued with our finding of GM-CSF upregulation because GM-CSF and IL-2 are more highly expressed in the blood of MS patients and treatment for MS reduces levels of these cytokines (60). Genetic evidence also implicates GM-CSF as a critical player in the mouse experimental autoimmune encephalomyelitis (EAE) model (61). These studies suggest that GM-CSF from ILC2s plays a pivotal role in enhancing inflammation, which  is associated with CNS demyelination (1,26,(62)(63)(64). Our results are consistent with these findings and suggest that ILC2s may be involved in the GM-CSF axis of demyelination pathology. RNA-seq analysis further defined features of the ex vivo ILC2 response against HSV-IL-2 infection. In our studies, expression of 44 HSV-1 genes was detected following HSV-IL-2 infection, while expression of 57 genes, and of over 75 HSV-1 genes, was detected in ILC2s infected with parental virus. We found no detectable expression of UL1, UL3, UL5, UL6, UL13, UL26, UL30, UL32, UL38, UL41, UL43, UL47, and UL52 genes in ILC2s infected with HSV-IL-2; this result could reflect the lower infectivity of HSV-IL-2 virus than of parental virus (65). Alternatively, it is possible that proinflammatory function of IL-2 could affect expression of these genes, either directly or indirectly as we have previously reported (63,64). However, the differences in the numbers of viral genes expressed in HSV-IL-2 and parental virus infections suggest that virally expressed IL-2 did not significantly enhance or block virus production. This study allowed us to compare parental HSV-1 and HSV-IL-2 data to differentiate the effects of IL-2 on the ILC2 antiviral and immune responses as opposed to viral replication.
GO term analysis indicated that HSV-1 infection, regardless of viral IL-2 expression, induced the expression of genes belonging to "response to interferon-beta," "response to virus," "herpes simplex infection," and "regulation of defense response" in ILC2s ( Fig.  1A and B and Fig. 2A and B). These GO terms included genes encoding proteins involved in anti-virus response, such as proteins of the IFN-induced protein with tetratricopeptide repeats (IFIT) family (66) or oligoadenylate synthase (OAS) (67). These proteins are known to inhibit viral gene expression and block virus replication. Induction of these genes in infected ILC2s indicates that an intrinsic immune response is the predominant antiviral response in HSV-1-infected ILC2s. This result is consistent with our previous study showing that ILC2s are refractory to HSV-1 replication (48).
RNA-seq analysis of ILC2s infected with parental virus or HSV-IL-2 revealed that IL-2 expressed by HSV-1 leads to various transcriptional changes (Fig. 1C, Fig. 2C, and Fig.  3). GO terms enriched in ILC2s infected with HSV-IL-2 include "mitotic cell cycles process," "meiotic nuclear division," and "inflammatory response," supporting the view that ILC2s proliferate due to enhanced activation and are prone to inflammation when IL-2 is produced. Genes belonging to "mitotic cell cycles process" or "meiotic nuclear division," such as Ccna1 (encoding cyclin A1) and Top2a (encoding topoisomerase IIa), indicate that ILC2s proliferate more upon infection with HSV-IL-2 than with parental virus. Genes belonging to "inflammatory response" include Ifngr1, Emilin2, Tnfaip3, Ecm1, and Hif1a. Since Tnfaip3 is known to reduce IL-13 and IL-5 production by ILC2s (68), its upregulation suggests that cytokine profiles may be modified.
Our data suggest that HSV-1 infection alters the activation state of ILC2 cells. ILC2s are known to express Th2 cytokines, including IL-4, IL-5, IL-9, and IL-13 (35)(36)(37). However, using a multiplex cytokine bead assay, we found that ILC2s infected with HSV-1, whether the parental strain or HSV-IL-2, did not express IL-4 or alter IL-9 expression. The lack of cytokine production may be caused by HSV-1 modifying the ILC2 immune response program through signaling events and/or by the lack of special environmental factors ex vivo, such as soluble factors and interactions with other cells. The induction of one such soluble factor, Hif1a, which is a transcription factor in ILC2s, is known to attenuate IL-33 receptor ST2 expression (69). Hif1a upregulation in CD8 1 T cells is known to enhance glycolysis and facilitate effector responses to prolong viral infection (70). Our assay also detected GM-CSF, suggesting a change in the ILC2 activation state following HSV-IL-2 infection because GM-CSF is typically expressed by ILC3s when stimulated with IL-23 and/or IL-1b (37,71). Isg15, which is known to play an important role in HSV-1 infection (72), was upregulated in both HSV-IL-2 and parental virus infections. Secretion of lsg15 protein induces IFN-g production from T cells and NK cells (73). These changes in transcription and cytokine profiles suggest a change in ILC2 activation state after infection.
Although IL-17A can be secreted from ILC2s (74) and plays important roles in MS in humans and in the EAE mouse model (61), it was not detected in our ILC2 gene expression profiling. It is possible that only a minor population of ILC2s infected with HSV-IL-2 secrete IL-17A. Alternatively, specific environmental cues, such as soluble factors or cell-cell interactions from pathogenic cells, might be necessary to induce IL-17A secretion from ILC2s.
We recently reported that Cxcl10 and Ccl5 expression was downregulated in the brains of HSV-IL-2-infected mice. These chemokines are associated with demyelinating pathology, which requires the presence of ILC2s (47). However, in our current ex vivo study, direct HSV-IL-2 infection of ILC2s upregulated expression of Cxcl10 and Ccl5 compared to that in mock-infected ILC2s. There are several possible reasons for this discrepancy. First, HSV-1 infection and IL-2 expression may induce other environmental factors that alter the expression profile of ILC2s in the brain in vivo but were not recapitulated in ILC2 cell culture. Second, the ILC2s used in our RNA-seq experiment were not derived from brain, and CNSderived ILC2s may respond with a different expression profile. Third, since the inflammatory environment driving demyelination is likely restricted to local pathogenic sites, cytokine expression in the whole brain may differ from that in the local environment. The finding that T H 17 cells form tertiary lymphoid follicles in the EAE model indicates that the inflammatory environment is localized (75).
ILC2 activation by cytokines does not normally lead to demyelinating disease, but ILC2 activation combined with viral infection and IL-2 is associated with demyelinating diseases. The current study suggests that activation of ILC2s by virally expressed IL-2 produces an atypical and enhanced response based on changes to its cytokine and gene expression profiles. This change in ILC2 activation state could be a major feature in the pathology of demyelinating diseases and provides direction for further research. High-dose IL-2 administration has been approved by the FDA to treat patients with late-stage metastatic melanoma or renal cell carcinoma, with an overall response rate of 16% (76). However, IL-2 therapy is associated with severe toxic side effects that include hypotension, vascular leak syndrome, liver dysfunction, and neurological disorders (77). Elevation of IL-2 is common in autoimmune patients; thus, our studies may have broad implications in terms of understanding the mechanisms by which elevated IL-2 interacts with environmental factors (in this case HSV-1) to promote autoimmunity.

MATERIALS AND METHODS
Cells and virus. Rabbit skin (RS) cells were generated in our laboratory, prepared, grown in minimal essential medium (MEM) plus 5% fetal bovine serum (FBS), and used as described previously (78). Plaque-purified HSV-1 recombinant virus expressing IL-2 (HSV-IL-2) and parental virus for HSV-IL-2 (dLAT2903) were grown in RS cell monolayers in MEM containing 5% fetal bovine serum, as we described previously (28,29,65). HSV-IL-2 expresses two copies of the murine IL-2 gene under the control of the latency-associated transcript (LAT) promoter of HSV-1 in place of LAT in a LAT-negative virus (dLAT2903) (65).
Luminex xMAP immunoassay (magnetic bead kit). Luminex assays were performed in the Immune Assessment Core at the University of California, Los Angeles (UCLA, CA) as we described previously (48,49). Briefly, mouse 32-plex magnetic cytokine/chemokine kits were purchased from EMD Millipore (Billerica, MA) and used according to the manufacturer's instructions. Isolated type 2 ILCs were infected with 10 PFU/cell of HSV-IL-2 or parental virus or were mock infected for 24 h. Media from infected and mock-infected cells were collected and 25-ml volumes of 1:2 diluted samples were mixed with 25 ml of magnetic beads and incubated overnight at 4°C with shaking. After incubation, the plates were washed twice with wash buffer in a Biotek ELx405 washer, and 25 ml of biotinylated detection antibody was added. The reaction mixture was incubated for 1 h at room temperature, and then streptavidin-phycoerythrin conjugate (25 ml) was added and the reaction mixture was incubated for another 30 min at room temperature. Following two washes, beads were resuspended in instrument sheath fluid buffer, and fluorescence was quantified using a Luminex 200 instrument (Luminex Corp., Austin, TX).
Library preparation and sequencing. Isolated type 2 ILCs were infected with 10 PFU/cell of HSV-IL-2 or parental virus or were mock infected for 24 h. Up to 10,000 infected or mock-infected cells per sample in triplicate were pelleted, washed, flash-frozen in liquid nitrogen, and then stored at -80°C. Total RNA, isolated using a SMART-Seq V4 ultralow RNA input kit for sequencing (TaKaRa Bio USA, Inc., Mountain View, CA), was used to generate double-stranded cDNA by reverse transcription for library preparation using the Nextera XT library preparation kit (Illumina, San Diego, CA). cDNA was quantitated using Qubit (Thermo Fisher Scientific). cDNA normalized to 80 pg/ml was fragmented, and sequencing primers were added simultaneously. A limiting-cycle PCR added index 1 (i7) adapters, index 2 (i5) adapters, and sequences required for cluster formation on the sequencing flow cell. Indexed libraries were pooled and washed, and the pooled library size was verified using a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) and quantified using Qubit. Libraries were sequenced using a NextSeq 500 (Illumina) with a 1 Â 75-bp read length and coverage of over 25 million reads/sample. Data analysis. The raw reads were aligned to the transcriptome using STAR (version 2.5.0) (81) and RSEM (version 1.2.25) (82) with default parameters, using a custom mouse CRCm38 transcriptome reference downloaded from http://www.gencodegenes.org, containing 92 ERCC sequences; all protein coding and long noncoding RNA genes were based on Mouse GENCODE M8 annotation. Expression counts for each gene (transcripts per million [TPM]) in all samples were normalized by sequencing depth. GO term analysis was performed using Metascape (https://metascape.org) (52). Data visualization was performed using the R statistical software (version 3.5.2).
Statistical analyses. Student's t test was performed using the computer program Prism (GraphPad, San Diego, CA). Results were considered statistically significant when the P value was ,0.05.