Profiling whole-tissue metabolic reprogramming during cutaneous poxvirus infection and clearance

ABSTRACT Vaccinia virus (VACV) infection induces prominent changes in host cell metabolism. Little is known about the global metabolic reprogramming that takes place in whole tissue during viral infection. Here, we performed a longitudinal metabolomics study in VACV-infected mouse skin to investigate metabolic changes in the tissue during infection. We assessed metabolites in homogenized skin of the ear pinnae over time in the presence or absence of antigen-specific T cells using untargeted mass spectrometry. VACV infection induced several significant metabolic changes in the tissue, including in the levels of nucleic acid metabolites (reflecting the impact of viral replication on the skin metabolome). Furthermore, monocyte- and antiviral T cell-produced metabolites, such as itaconic acid and glutamine, were significantly increased following infection, highlighting the immune response’s contribution to the global skin metabolome. Additional RNA-Seq of infected skin tissue recapitulated metabolic changes identified by metabolomics analysis. Overall, our study reveals the metabolic balance of viral replication and the antiviral immune response in the skin, elucidating metabolic pathways that could contribute to cutaneous poxvirus control in vivo. IMPORTANCE Human poxvirus infections have caused significant public health burdens both historically and recently during the unprecedented global Mpox virus outbreak. Although vaccinia virus (VACV) infection of mice is a commonly used model to explore the anti-poxvirus immune response, little is known about the metabolic changes that occur in vivo during infection. We hypothesized that the metabolome of VACV-infected skin would reflect the increased energetic requirements of both virus-infected cells and immune cells recruited to sites of infection. Therefore, we profiled whole VACV-infected skin using untargeted mass spectrometry to define the metabolome during infection, complementing these experiments with flow cytometry and transcriptomics. We identified specific metabolites, including nucleotides, itaconic acid, and glutamine, that were differentially expressed during VACV infection. Together, this study offers insight into both virus-specific and immune-mediated metabolic pathways that could contribute to the clearance of cutaneous poxvirus infection.

through person-to-person contact, resulting in a global public health emergency (4).The MPXV outbreak highlighted gaps in our knowledge of poxvirus pathogenesis since the eradication of smallpox.Most human poxvirus infections occurred before the advent of sophisticated immunological and metabolic profiling methods.Thus, we still have much to learn about poxvirus immune control and metabolic regulation in the tissue environment.
Viruses induce metabolic alterations to infected host cells.Most virus-induced metabolic changes have been defined using reductionist systems including the in vitro infection of cultured cells.The overall metabolic changes induced at the tissue level during viral infection or how immune-mediated viral clearance further perturbs or resolves metabolic deviations from uninfected skin warrants continued investigation.Nonetheless, several unique host cell metabolic requirements for vaccinia virus (VACV) replication in vitro have been previously detailed (5,6).Viral replication consumes host cell macromolecules such as nucleotides, phospholipids, and amino acids to synthe size new viral genomes and assemble progeny virions.De novo fatty acid synthesis is needed for VACV viral replication in vitro, as chemical antagonists of fatty acid synthe sis dramatically impair virion production (5).Additionally, a metabolomics screen of VACV-infected human fibroblasts identified changes in glutamine levels, suggesting that viral replication is dependent on glutamine metabolism (6).In further support, VACV encodes a protein (C16) that enhances glutamine metabolism, which is thought to increase the synthesis of metabolic precursors needed for viral replication (7).Studies have also demonstrated that VACV induces a pseudo-hypoxic state in the host and can inhibit host DNA, mRNA, and protein synthesis to favor viral replication utilizing cellular processing machineries (8,9).Thus, VACV alters host cellular metabolism in numerous ways to promote successful viral replication.
In addition to metabolic changes occurring in virus-infected cells, viral infection also modifies the tissue's cellular composition.The rapid recruitment of innate and adap tive cells into the tissue following infection significantly increases metabolic demands.Infected tissues must carefully balance the metabolic requirements of the epithelium along with a robust cellular immune infiltrate to promote viral clearance, limit immunemediated damage, and heal wounded tissues.Here, we investigated the metabolic changes induced in the skin during VACV infection using a previously established mouse infection model (10,11).In this model, VACV is inoculated into the mouse ear skin, where the virus replicates productively in epidermal keratinocytes (10).In wild-type mice, skin infection remains localized and is controlled by a robust cell-mediated immune response in the tissue that includes both monocytes and cytotoxic CD8 + T cells (10,11).Monocytes are recruited to the skin after infection and a portion of monocytes become VACV infected.Antigen (Ag)-specific CD8 + T cells kill virus-infected monocytes to help limit infection (10).In wild-type animals, viral titers also reveal viral replication in the skin is followed by the gradual clearance of cutaneous VACV infection (10).
Due to their importance in antiviral and anti-cancer immunity, CD8 + T cell metabolism has been extensively investigated in the past decade.Cytotoxic T cells have unique energy requirements because of their rapid division, migration, and effector activity.As the overall energy demand increases in response to inflammation and Ag recognition, activated T cells engage in aerobic glycolysis, increase mitochondrial function, and alter mitochondrial morphology (12)(13)(14).Besides increased glucose uptake, amino acids and lipids are crucial metabolites for T cell differentiation and function (15)(16)(17).Glutamine, the most abundant nonessential amino acid in serum, has been implicated in several aspects of CD8 + T cell function and can be metabolized through glutaminolysis to fuel the tricarboxylic acid cycle (18).Furthermore, inhibition of glutamine metabolism reduces T cell production of inflammatory cytokines including interferon-gamma (IFN-γ) and interleukin-17 (IL-17) (19).Host defense against cytomegalovirus and tuberculosis infection depends on glutamine metabolism (19,20), though an in vivo role during the anti-VACV host response has not yet been documented.
In addition to the metabolic requirements of cytotoxic CD8 + T cells, macrophages and monocytes also engage in specific metabolic reprogramming to support function.Lipopolysaccharide (LPS)-activated macrophages produce large quantities of itaconate, and this metabolite has been identified in Mycobacterium tuberculosis-infected mouse lungs (21)(22)(23).In the context of viral infection, mice lacking itaconate-metabolizing enzyme present with increased Zika virus replication and decreased survival (24).Thus, the resolution of infection can be critically dependent upon leukocyte-produced metabolites.
The development of untargeted metabolomics analyses through mass spectrome try affords an unprecedented opportunity to perform a global survey of distinct, virus-induced metabolic changes in whole tissues.Here, we performed whole-tissue metabolomics of the skin to characterize the metabolic signature of T cell-mediated clearance of VACV infection.We identified numerous metabolic changes induced by VACV infection that were independent of T cells, along with changes likely associated with Ag-specific T cell-mediated viral clearance.Concurrently, we performed whole-tis sue transcriptomics to better understand the broad metabolic pathways that were significantly altered at the mRNA level.Together, our analyses of the whole-tissue metabolome and transcriptome post-VACV infection provide a deconstructed analysis of the metabolic reprogramming in a virus-infected tissue undergoing immune control.

Mice
Rag1 -/-(Taconic stock number 146) and C57BL/6-Tg(TcraTcrb)1100Mjb (OT-I) T cell receptor (TCR) transgenic mice (Taconic stock number 175) were obtained from the Taconic Farms and National Institute of Allergy and Infectious Diseases (NIAID) Intramu ral Research Repository.Stock Tg (CAG-DsRed*MST) 1Nagy/J mice (Jackson Laboratory stock #5441) were crossed with OT-I TCR transgenic mice and bred to homozygosity to create dsRed OT-I mice.Six-to 12-week-old female mice were used in all experiments.All animals were housed under controlled conditions of humidity, temperature, and light (12-hour light/12-hour dark cycles), specific pathogen-free conditions (including surveillance and negativity for murine norovirus, mouse parvovirus, and mouse hepatitis virus).Standard rodent chow and water supplied were available ad libitum.

Viruses and infections
Mice were anesthetized using isoflurane with 4% induction and 2% oxygen flow for 3 minutes.Mice were then removed from the isoflurane chamber and epicutaneously infected in the ear pinnae by five gentle pokes with a bifurcated needle dipped in VACV (with working titers of ~2×10 8 pfu/mL) as previously described (10).VACV-NP-S-eGFP contains a nuclear localizing protein (influenza virus nucleoprotein, NP), the cognate Ag recognized by OT-I CD8 + T cells (SIINFEKL, S), and enhanced green fluorescent protein (eGFP).VACV-NP-eGFP is an identical virus lacking SIINFEKL.For these studies, one ear was infected with VACV-NP-S-eGFP (NPS) and the contralateral ear with VACV-NP-eGFP (NPE) as previously described (11).

Adoptive transfers
Ag-specific CD8 + T cells were purified from the lymph nodes and spleen of female dsRed OT-I mice using an EasySep mouse CD8 + T cell negative selection kit (StemCell).After purification, cells were tested for purity and activation by flow cytometry to achieve at least 85% purity with greater than 98% CD69 -population (CD69 APC clone: H1.2F3, Biolegend 104518). 1 × 10 4 purified dsRed OT-I CD8 + T cells were transferred intrave nously into Rag1 -/-recipients 1 day prior to infection for flow cytometry and metabolo mics experiments.2.5 × 10 5 dsRed OT-I CD8 + T cells were transferred for RNA-Seq analysis.

Viral titers via plaque assay
Ears were harvested at the indicated day post-infection (dpi), then fine chopped and digested by collagenase I for 1 hour at 37°C.Digested tissue suspensions were collected after vigorous pipetting.Samples were kept frozen and processed in a single batch for plaque assay where they were freeze-thawed three times, serially diluted, and plated on 143b thymidine kinase deficient (TK-) cells.TK-cells were incubated for 48 hours, then stained with crystal violet for plaque counting.

Metabolomics
Ears (n = 3 per group) were harvested and weighed on days 5, 6, 7, 8, and 10 post-infec tion and snap frozen at −80°C.Frozen tissues from all time points were homogenized in a single batch using metal bead homogenization tubes (MP Biomedicals) with high performance liquid chromatography (HPLC)-grade water normalized to tissue weight (50 mg tissue:1 mL H 2 O).After homogenization, 800 µL pre-chilled HPLC-grade methanol was added to 200 µL of tissue homogenate for methanol extraction.Samples were placed in a −20°C freezer to facilitate protein precipitation.The methanolic extracts were centrifuged at 14,000 rcf at 4°C for 10 min and the supernatant was vacuum concentrated using a SpeedVac.
Untargeted metabolomics was then performed using ultra-high performance liquid chromatography-high-resolution tandem mass spectrometry (UHPLC-HRMS/MS).Detailed methodology is provided in Supplemental Methods.All chemicals identified and referenced annotations are listed in Tables S1 and S2.Additional information on specific chemicals referenced in main figures are provided in Table S3.In brief, sam ples were analyzed using an ultra-high performance liquid chromatograph (Vanquish, Thermo Scientific) coupled to a high-resolution mass spectrometer (Orbitrap Fusion Tribrid, Thermo Scientific).LC-MS and LC-MS/MS data were acquired.LC-MS data were collected from individual samples (n = 1 injection), system blanks (injection of solvent used to resolubilize samples), and a pooled quality control (QC).The pooled QC was injected multiple times at different volumes and used in data processing.LC-MS/MS data, used to annotate features, were collected using the AcquireX (Thermo Scientific) deep scan methodology in which the pooled QC was injected multiple times (n = 7).
Chromatographic separation was carried out on a 2.1 × 100 mm, 100 Å, 2.6 µm, F5 analytical column (Phenomenex) with corresponding guard cartridge.The column was maintained at 30°C during separation with solvent pre-heater.Gradient elution was performed after an initial period of isocratic elution using water with 0.1% acetic acid vol/vol (A) and acetonitrile with 0.1% acetic acid vol/vol (B).Separation was performed as follows: 0% B from 0 to 2.0 min, 0% to 100% B from 2.0 to 10.5 min, 100% B from 10.5 to 12.0 min, 100% to 0% B from 12.0 to 13.0 min, 0% B from 13.0 to 20.0 min.The flow rate was 0.5 mL/min.
MS and MS/MS data were collected with an EASY-IC.MS data were acquired at 120,000 resolution from m/z 100-1,000 with an radio frequency (RF) lens of 60% and maximum injection time of 50 ms.MS/MS data were acquired at 30,000 resolution using an isolation width of 1.5 (m/z), stepped-assisted higher energy collisional dissociation (HCD) (energy steps were 20, 35, and 60, and a maximum injection time of 54 ms).The inclusion list was generated and updated via AcquireX with a low and high mass tolerance of 5 ppm.Dynamic exclusions were performed with the following parameters: exclude after n = 3 times, if occurs within 15 s, exclusion duration of 6 s, a low mass tolerance of 5 ppm, a high mass tolerance of 5 ppm, and excluding isotopes.

Untargeted metabolomics data processing
Compound Discoverer 3.3.0.550 (Thermo Scientific) was used to process.rawfiles to generate descriptors of each feature (e.g., m/z, retention time), annotation information (e.g., MS/MS database match), and peak area.We processed the output from Compound Discoverer using in-house R scripts via Jupyter Notebooks in order to format the data, putatively annotate chemical features, assess signal response in pooled QC samples, assess signal variance in the pooled QC versus individual samples (i.e., dispersion ratio), and perform statistical analyses (multivariate and univariate).Features were putatively annotated based on MS/MS spectral matching to in-house reference libraries, publicly available databases, and authentic chemical standards where possible.Representative metabolites have been affirmed by MS/MS matching, m/z mass error, and retention time similarity to authentic chemical standard.The affirmation of itaconic acid performed for this study in the positive ion mode to level 1 confidence is shown in Fig. S3.

Whole ear transcriptomics
Ears were collected on the indicated day post-infection and immediately placed in buffer RLT (Qiagen).Tissues were homogenized, and RNA was isolated using Lysing Matrix S (1/8″) metal beads (MPBiomedicals) and a FastPrep-24 Instrument (MPBiomedicals).RNA was then purified using a Qiagen RNAEasy Mini Kit (Qiagen) according to the manufac turer's protocol.An on-column DNAse digestion was performed prior to RNA elution.RNA concentration and purity were accessed using an Agilent TapeStation 4200.Samples were sequenced on a NovoSeq 6000 S1 using Illumina Stranded Total RNA Prep, Ligation with Ribo-Zero Plus, and paired-end sequencing.The samples had 142 to 216 million pass filter reads with more than 90% of bases above the quality score of Q30.
Raw fastq files were trimmed for quality and adapter contamination with Cutadapt v1.18 (25).Trimmed reads were aligned to the mouse GRCm38.p6reference genome and GENCODE release 21 reference transcriptome using STAR v2.7.0f run in two-pass mode (26).RNA-Seq by Expectation-Maximization (RSEM) v. 1.3.1 (27) was used to compute the expected counts which were further converted to counts per million (CPM) by eVITTA (28).Genes that were expressed with at least 1 CPM in a minimum of three samples were included in the principal component analysis (PCA) and differential expression testing by edgeR (v.3.40.2) (29).Differential expression analysis compared samples in different conditions including mice receiving OT-I T cells with NPS infection vs no OT-I transfer with NPS, OT-I with NPE vs transferred OT-I with NPS infection, OT-I with NPE vs naïve tissue (control), and transferred OT-I with NPS infection vs naïve tissue (control).False discovery rate (FDR) cut-off value of 0.05 was used to detect differentially expressed genes for downstream analysis.PCA projections and volcano plots were generated using ggpubr (v.0.6.0) and ggplot2 (v.3.4.2) packages in R (v.4.2.2), respectively.
Pre-ranked gene set enrichment analysis (GSEA) was performed by ranking genes based on the P-value and log 2 fold change statistics from the edgeR analysis.Gene Ontology (GO) biological process (BP) GSEA was performed using (30) and Kytoto Encyclopedia of Genes and Genomes (KEGG), and Reactome (RA) pathways GSEA were performed using eVITTA (28).Significant enrichments were those with an FDR q-value <0.05.Pathway enrichment map visualizations for glutamate metabolism were performed using Cytoscape (v.3.9.1)(31).

Statistical analyses
Statistical significance for individual metabolites was calculated using a univariate t-test with false discovery rate correction or one-way analysis of variance (ANOVA) with Dunnett's multiple comparisons.Statistical analyses and graphs were performed and generated using GraphPad Prism 9, R, or as otherwise indicated in the supplemental Materials and Methods section.PCAs and volcano plots were generated in R using prcomp and ggplot packages.Data used to generate PCAs are provided in Table S1.For volcano plots, p-values and log 2 fold change between groups calculated based on mass spec peak area of all features were used for illustrative graphs.

T cells and monocytes are recruited to VACV-infected skin
We have previously established and characterized an immunocompetent mouse model of epicutaneous poxvirus infection with VACV (10,11,32).Here, we adapted this model to analyze metabolic changes occurring in VACV-infected skin during T cell-mediated viral clearance.To control T cell killing conditions, we adoptively transferred 1 × 10 4 OT-I CD8 + T cells (TCR transgenic T cells recognizing SIINFEKL) into Rag1 -/-recipients (lacking endogenous adaptive immune cells including T and B cells).In this system, only the adoptively transferred OT-I CD8 + T cells can kill SIINFEKL-expressing VACV-infected monocytes.Therefore, to compare tissue in which CD8 + T cells could kill infected cells to those without T cell killing, we infected one ear of the mouse with VACV-NP-S-eGFP (expressing SIINFEKL, abbreviated NPS throughout the text) and the other with VACV-NP-eGFP (lacking SIINFEKL, abbreviated NPE).In this model, NPS infection activates OT-I CD8 + T cells, which then traffic into both ears regardless of SIINFEKL expression; however, no OT-I CD8 + T cell-mediated killing occurs in NPE-infected ears (11).We harvested whole ears at 5, 6, 7, 8, and 10 dpi (Fig. 1A).Flow cytometric analysis of single-cell suspensions of VACV-infected skin revealed detectable numbers of infected monocytes and adoptively transferred OT-I CD8 + T cells by 5 dpi and throughout the five study time points (Fig. 1B through E).By 7 dpi, Ag-specific CD8 + T cells continued to accumulate in the skin and accounted for ~10%-20% of the total lymphocytes (Fig. 1B and C, top panels).As T cells entered and killed infected cells, NPS-infected monocytes (determined by virus-encoded GFP expression) decreased in number until they were negligible by 10 dpi (Fig. 1B, bottom panel, and Fig. 1D).In the contralateral ear infected with NPE (lacking cognate Ag), VACV-infected monocytes remained high in number throughout day 10 even though both ears had similar number of cytotoxic CD8 + T cells in the tissue (Fig. 1C, bottom panel, and Fig. 1E).Despite clearance of infected monocytes by 10 dpi in NPS-infected tissues by OT-I CD8 + T cells, infectious virus from whole-tissue digestion was still detectable at this time point (Fig. 1F).However, VACV viral titers gradually increased over this time period in NPE-infected tissue (without cytotoxic T cell restriction) (Fig. 1G).Together these data show that VACV infection of the skin induces T cell recruitment that clears virus-infected monocytes and reduces overall VACV viral titers in an Ag-dependent manner by 10 dpi.

VACV infection induces metabolic changes in the skin
Using this system of infection with two different VACVs, we next sought to examine the metabolic profile of VACV-infected skin in the presence and absence of cytotoxic CD8 + T cell-mediated clearance of viral infection.We performed whole-tissue untargeted metabolomics on VACV-infected skin from our infection model (NPS or NPE infection of Rag1 -/-mice) with and without adoptively transferred OT-I CD8 + T cells (Fig. 2A).Thus, we included five different sample groups: uninfected naïve mice as controls, NPS infection with transferred OT-I CD8 + T cells, NPE infection with OT-Is, NPS infection without OT-Is, and NPE infection without OT-Is.This allowed us to examine metabolic changes induced by T cell entry alone and those associated with viral clearance.Metabolites extracted from whole ear pinna homogenates underwent UHPLC-HRMS/MS in positive and  annotated in the negative ionization mode.All features (annotated and unannotated) are listed in Tables S1 and S2.
To understand global metabolic changes in the infected skin, we first used PCA to compare the five sample groups based on all features (Fig. 2B).PCA plots composed from features (annotated and unannotated) in both positive and negative ion modes revealed a clear separation of the uninfected control group (orange dots) from the infected samples.Unexpectedly, all post-infection tissue samples displayed very limited separation despite experimental conditions (e.g., the presence or absence of T cells and cognate Ag).Infection progression influenced the tissue metabolome as samples grouped together based on time post-infection.Additionally, the metabolic profile of infection with cognate Ag and Ag-specific T cells was similar to other infected groups and highly different from naïve control tissue.This was also the case when a single time point, 8 dpi, was examined (Fig. 2C), and for all sampled time points (Fig. S2).Thus, VACV infection progression and the ensuing innate immune response were overriding contributors to the overall metabolome of the skin regardless of whether Ag-specific T cells were present in the tissue.

Metabolic changes align with time post-infection
Although the five experimental groups did not show distinct separation in PCA, a fraction of the total features detected by untargeted metabolomics displayed a gradually increasing or decreasing trajectory (P < 0.05 comparing each time point post-infection to the reference time point of 5 dpi) (Fig. 3).We detected increases in several nucleo tides and associated molecules involved in DNA synthesis during infection including: deoxycytidine-5´-monophosphate, adenosine monophosphate, 2´deoxycytidine, and guanosine.The changes in nucleotide species over time may reflect the rapid and robust synthesis of the VACV DNA genome that requires large amounts of DNA precursors.We also observed a decreasing trend in several metabolites such as uridine, creatine, and 2´-deoxyguanosine-5´-diphosphate.Interestingly, VACV encodes a uracil DNA glycosy lase and a deoxyuridine triphosphatase to prevent uracil incorporation into new DNA genomes (33,34).Together, our data show that the global metabolome of VACV-infected skin tissue could be largely influenced by time or infection progression.

Itaconic acid is produced in VACV-infected skin
We next sought to assess the magnitude of metabolic changes in the tissue after VACV infection.We compared all features (both annotated and unannotated) from uninfected control tissue to infected samples either without (Fig. 4A) or with (Fig. 4B) OT-I CD8 + T cells.In both samples, itaconic acid was identified as a highly upregulated metabolite with the most significant P-value compared to uninfected skin.Itaconic acid is produced by myeloid cells and is upregulated during both bacterial and viral infections (35,36).Itaconic acid was identified via retention time matching, measured accurate mass (m/z), and MS/MS spectral information.Potential isobaric metabolites, citraconic acid, and mesaconic acid were excluded based on retention time (Fig. S3).Recently, itaconic acid has been the focus of intense study due to its immunomodulatory functions (21).Therefore, we closely examined itaconic acid levels over time in our samples with different conditions.By 5 dpi, our earliest time point, itaconic acid was significantly upregulated compared to uninfected tissues (P < 0.0001) (Fig. 4C).Over the next 5 days of infection, itaconic acid remained highly upregulated ((P < 0.0001 for each time point post-infection).Analysis of samples undergoing T cell killing (or not) also did not reveal any clear differences in itaconic acid levels (Fig. 4D).Thus, although T cells kill infected monocytes, elimination of this infected monocyte subpopulation does not appear to be enough to shift the whole-tissue metabolome away from itaconate production.Notably, however, there are still numerous uninfected myeloid cells and other leukocytes present in the ear tissue over the five study time points (Fig. S4).

RNA sequencing of whole skin reveals altered metabolic pathways during infection
To complement our metabolomics analyses, we performed RNA sequencing of uninfec ted whole ear tissue as well as infected ear tissue with and without transferred OT-I CD8 + T cells at 6 dpi to identify transcriptomic changes associated with VACV infection.In comparison to the metabolome PCAs (Fig. 2), a PCA based on gene expression revealed more distinct separation of the five sample groups.Uninfected control tissues and tissues undergoing T cell-mediated killing clustered separately from groups lacking killing (Fig. 5A).We identified 285 genes that were significantly upregulated (log 2 |FC| ≥ 1.5, FDR ≤ 0.05) and 31 genes that were significantly downregulated (log 2 |FC| ≤ −1.5, FDR ≤ 0.05) in NPS-infected tissues with OT-I CD8 + T cells compared to NPS-infected tissues without OT-I CD8 + T cells.Several T cell-specific genes were significantly upregulated including Cd8a, Cd8b1, Il2rb, Eomes, and Ifng (Fig. 5B).Additionally, Gimap3, which codes for a GTPase located on the outer membrane of the mitochondria, was significantly upregulated in the tissue with cytotoxic T cells (Fig. 5B) (37).
The nature of these changes in expression was interrogated using a pre-ranked GSEA to determine enrichment in GO BP terms and KEGG and RA pathways.GO terms related to adaptive immune response and T cell-mediated cell killing were upregulated in NPSinfected tissues with OT-I CD8 + T cells compared to those without OT-I CD8 + T cells (Fig. 5C).However, GO terms related to mitochondrial RNA processing and mitochondrial gene expression were upregulated in NPS-infected ears without OT-I CD8 + T cells or cytotoxic T cell-mediated control of VACV infection (Fig. 5C).Pathways related to arachidonic acid metabolism (adj.P-value = 0.018), pyruvate metabolism (adj.P-value = 0.04), mitochondrial translation, elongation, termination (adj.P-value = 0.04), and tricarboxylic acid cycle (TCA) and respiratory electron transport (adj.P-value = 0.045) were significantly upregulated in NPS tissues without OT-I CD8 + T cells (thus due to transcriptional changes in other cells during infection) (Table S4).

Antiviral T cells alter specific metabolite levels in the skin
Despite the infection-driven global metabolic changes observed through metabolomics and transcriptomics analyses in whole skin tissue, we detected significant differences in glutamic acid that varied with the adoptive transfer of T cells (Fig. 6A).Glutamic acid was significantly reduced in NPS-infected tissue with OT-I CD8 + T cells compared to those without T cells (P = 0.036), indicating the possible consumption or utilization of glutamic acid by cytotoxic OT-I CD8 + T cells during Ag-specific killing (Fig. 6A).As a product of glutamine metabolism, glutathione was significantly increased in all sample groups compared to uninfected control samples (Fig. 6B), indicating both infection and immune responses could contribute to the production of glutathione.
Given the identification of glutamic acid and glutathione metabolic changes, we explored differentially expressed genes involved in the glutamine and glutathione metabolism pathways comparing mRNA levels from NPS-infected tissues with OT-I CD8 + T cells to those without T cells and to uninfected control tissues (Fig. 6C through E).A volcano plot revealed the upregulation of Gls2, encoding glutaminase that catalyzes the breakdown of glutamine to glutamate and ammonia, in no OT-I-NPS tissues, consistent with the higher level of glutamic acid detected by metabolomics (Fig. 6A and C).Slc38A1, encoding a key glutamine transporter, was also significantly upregulated in NPS-infected tissue with OT-I CD8 + T cells (Fig. 6C).Likewise, numerous glutathione metabolismassociated genes were differentially expressed in NPS-infected tissue with T cells compared to uninfected control tissues (Fig. 6D).In particular, the expression of Gsr, encoding glutathione-disulfide reductase that is responsible for glutathione recycling, was significantly increased in NPS-infected tissue with T cells (Fig. 6D).Furthermore, mRNA encoding several glutathione S-transferases (GSTA genes) were significantly lower in NPS-infected tissue with T cells compared to control samples (Fig. 6D and E).
Collectively, our data show that glutamine and glutathione metabolism, identified by both metabolomics and transcriptomics, are differentially regulated during T cellmediated VACV clearance in the skin.

DISCUSSION
With the tremendous recent advances in untargeted metabolomic profiling technolo gies, it is now possible to analyze complex tissues, such as the skin, to better understand metabolic alterations during infection or disease.Previous studies have characterized the skin metabolome of dermatological disorders and diseases such as psoriasis, atopic dermatitis, and melanoma (38,39).The identification of specific amino acid and lipid metabolites in these inflammatory skin disorders have demonstrated the potential value of using metabolic biomarkers as diagnostic and prognostic tools.However, studies of the whole-tissue metabolome during cutaneous viral infections have been lacking.Here, we used whole-tissue mass spectrometric metabolomic profiling and RNA sequencing to identify metabolic pathways altered in the skin during cutaneous VACV infection.We observed changes that were likely due to the metabolic requirements of the virus as well as due to myeloid and T cells recruited into the skin to help control viral replication.
Untargeted metabolomic profiling revealed several unique features of poxvirusinfected skin.The global metabolomic signature of VACV-infected skin tissue was clearly distinct from uninfected skin, with several metabolites, including itaconic acid, upregula ted after infection.While we expected CD8 + T cell-mediated killing of VACV-infected cells to shift the metabolomic profile in the tissue, we did not uncover significant differences in the levels of most metabolites in experimental groups with and without OT-I CD8 + T cells.Instead, progression of infection and the innate immune response overwhelmingly shaped the skin metabolome during infection.Levels of many metabolites in nucleic acid metabolism such as adenosine monophosphate, 2´-deoxycytidine, guanosine, and uridine were altered in a time-dependent manner.VACV replicates to create roughly 10,000 new DNA genomes of approximately 200 kb each per infected cell, incurring considerable nucleotide consumption (40,41).Furthermore, VACV encodes several viral DNA-modifying enzymes, including uracil-DNA glycosylase and thymidine kinases, that catalyze DNA repair and the rapid generation of DNA precursor pools (42,43).Finally, in addition to viral replication, VACV inhibits and interferes with host DNA synthesis, mRNA processing, and protein translation, which may result in the diminishing levels of metabolites such as uridine (9).Therefore, VACV possesses many known mechanisms that could account for the perturbed nucleic acid metabolism we observed in infected skin.
One of the most upregulated metabolites in our metabolomics data was itaconic acid.Itaconic acid is produced by myeloid cells, including monocytes and macrophages, and is the product of the mitochondrial enzyme ACOD1 (21,36,44).Cutaneous VACV infection induces a large influx of monocytes into the skin (10), and Acod1 mRNA was highly upregulated in RNA sequencing data from whole skin.In vitro, the TCA cycle intermediate aconitate is increased in VACV-infected fibroblasts (45), suggesting that infected monocytes might also have access to increased precursors for conversion to itaconic acid.Although further metabolic profiling will need to be performed on monocytes isolated from the infected skin, our data suggest that monocytes respond to the inflammatory environment during VACV infection through the production of itaconic acid.Interestingly, this pathway has recently been shown to exert multiple immunomo dulatory functions during antiviral immune responses.For instance, itaconic acid can inhibit M2 macrophage polarization in vivo (46).Itaconic acid regulates production of the key antiviral cytokines type I interferons (IFN-I) (47).As a result, itaconic acid may also exhibit antiviral activity that is independent of its role in IFN-I production (21,48).Lastly, itaconic acid is also thought to be anti-inflammatory (21), and skin inflammation, such as that occurring during atopic dermatitis, promotes VACV replication (49,50).Thus, our untargeted metabolomics approach identified an anti-inflammatory metabolite that could potentially dampen VACV replication or spread in the skin.
Additionally, select metabolites in the glutamine and glutathione metabolism pathways were significantly altered in VACV-infected skin.Glutamic acid was significantly decreased in NPS-infected skin (with Ag-specific OT-I CD8 + T cells) compared to those infected with NPE (lacking T cell-mediated killing of infected cells).Glutamic acid, or its ionic form glutamate, is the product of glutamine hydrolysis catalyzed by the enzyme glutaminase.From our RNA sequencing analysis, we observed the upregulation of the glutaminase-encoding gene, Gls2, in NPS-infected tissues without OT-I CD8 + T cells compared to those with OT-Is, suggesting the possible increase of glutamine hydrolysis in infected tissues lacking T cell-mediated killing.This is consistent with previous in vitro studies demonstrating that VACV uses glutamine (and not glucose) for viral replication (5-7, 51), which would be higher in tissues lacking killing (with higher viral titers).Furthermore, glutamate accumulates in VACV-infected cells due to glutamine metabo FIG 6 (Continued) genes in the glutamate metabolism pathway identified in the KEGG database are colored with gene names labeled in boxes.
(D) Volcano plot of differentially expressed genes in the glutathione metabolism pathway in NPS-infected skin with OT-I CD8 + T cells compared to uninfected control tissue.Gene expression changes reaching statistical significance in the glutathione metabolism pathways are color coded and labeled.(E) Metabolic pathway map illustrating glutamine/glutathione metabolism and associated protein mediators using Cytoscape.Significantly expressed genes encoding for proteins in the glutamine and glutathione metabolism pathways corresponding to genes identified in (D) are colored.Uninfected control group is shown in yellow and OT-I NPS-infected skin in purple, corresponding to the colors in the volcano plots.lism and nucleotide biosynthesis, which may reflect our finding of higher glutamic acid levels in NPE-infected tissues that lack Ag-specific T cells.
Several observations suggest that T cells also directly influence glutamine levels in the tissue.The increase of glutamine transporter SLC38A1 in NPS-infected tissue with OT-I CD8 + T cells may indicate the transport and utilization of glutamine during Ag-specific viral elimination.Previous studies have identified glutamine as a key metabolite that fuels T cell activation and function (17,52).T cells are highly sensitive to glutamine levels in vitro, and the precursors or products of glutamine metabolism are unable to replace glutamine (53).In addition to glutamine metabolism, glutathione was increased in all experimental groups compared to uninfected controls.Activation of T cells and viral infections are frequently associated with increased production of reactive oxygen species (54,55).Glutathione is an endogenous cellular antioxidant that can ameliorate oxidative damage (56,57).Furthermore, a previous study showed that ablation of T cell glutathione production decreases antiviral immunity against murine lymphocytic choriomeningitis virus infection (56).In the context of metabolic pathways, glutathione synthesis utilizes glutamate as precursors, further highlighting the involvement of glutamine and glutathione metabolism during the T cell response to cutaneous VACV infection.
Several in vitro studies have examined the metabolic changes that occur in VACVinfected cells, allowing a direct comparison to our observations in vivo.Viral growth factor (VGF) is a VACV-encoded secreted epidermal growth factor (EGF) homolog that enhances viral spread and cellular motility (45,(58)(59)(60).As VGF-induced cellular proliferation imposes increased energetic requirements which are fueled through the TCA cycle (5,61), VGF is a key regulator of host metabolism in infected cells (45).Future studies using a VGF-deletion virus will be needed to understand the effect of VGF in vivo on host cell metabolism, but the decreased in vivo replication of this virus will also need to be considered in relation to metabolic alterations (58).Unfortunately, we did not identify metabolic components of the TCA cycle in our analyses, so we are unable to quantify the contribution of increased TCA cycle metabolism by VACV-infected, VGF-producing cells.
There were several limitations to our study.First, the discriminatory power of untargeted metabolomics for stereoisomers (and sometimes constitutional isomers) is limited; therefore, annotated metabolites and metabolic pathways should be valida ted using more specific analytical methods.Second, we did not employ mouse models deficient in the metabolic pathways identified to ascertain the biological role of differentially expressed metabolites during VACV infection.As metabolites were extracted from whole tissue, we were unable to distinguish the levels of extracellular metabolites and intracellular metabolites to determine consumption or synthesis.Lastly, the post-infection time points used in this study were limited to 10 dpi.Later time points after infection clearance would provide better understanding of the metabolic state of tissue recovery.
In this study, we explored the dynamic balance of viral metabolic needs and immune cell requirements in poxvirus-infected skin using whole-tissue metabolomics and transcriptomics.Our data revealed that viral infection and immune response progression were the main factors shaping global metabolic changes in the skin, rather than the ability of T cells to clear virus-infected cells.The complexity of the whole-tissue environment and cellular dynamics will likely mask subtle metabolic differences that might be uncovered with single-cell metabolome analyses.Nonetheless, our wholetissue analysis provides a framework for better understanding the global metabolic changes that take place in the skin during acute viral infection.

FIG 1
FIG 1 OT-I CD8 + T cell killing kinetics of VACV-infected monocytes in the skin.(A)Experimental design to study the killing kinetics of adoptively transferred OT-I CD8 + T cells in VACV-infected mouse ears.(B and C) Representative flow plots from 7 dpi (top panels) and 10 dpi (bottom panels) showing OT-I CD8 + T cells (left panels) and VACV-infected monocytes (right panels) overlaid on total live CD45 + lymphocytes in (B) NPS-and (C) NPE-infected ears.Percentages of indicated cell populations are included in the representative plots.(D and E) Numbers of OT-I CD8 + T cells (red) and infected monocytes (green) following (D) NPS and (E) NPE infection in ears over five time points post-infection.Dots represent the mean value of all samples (n = 8).Error bars show the SEM.Results are from two independent time-course experiments with four animals per time point in each experiment.Statistical analysis was performed using one-way ANOVA and Dunnett's multiple comparison between each time point post-infection to 5 dpi as the reference group (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).(F and G) Viral titers determined by plaque assay for all samples over the five time points post-infection for (F) NPS-and (G) NPE-infected tissues.Error bars show the SEM.Results are from two independent experiment with four animals per time point in each experiment.Statistical analysis was performed using one-way ANOVA and Dunnett's multiple comparison between each time point using

FIG 2
FIG 2 Metabolomics of VACV-infected skin reveals infection-induced changes.(A)Experimental design for whole-tissue metabolomics.Ears from five experimental groups were harvested on 5, 6, 7, 8, and 10 dpi and kept frozen until methanol extraction.Methanol-extracted samples were analyzed by untargeted metabolomics.(B) PCA plots of features in positive and negative ion mode for all samples.Each data point represents an individual sample (n = 3 per condition per time point).Color indicates experimental groups.Circle size reflects the time point of harvest.(C) PCA plots of features in positive and negative ion modes for 8 dpi samples.Dots show individual samples.Colors represent experimental groups.

FIG 3
FIG 3 Metabolites in the whole skin reflect infection progression.Selected metabolites with temporal trajectory changes over the infection time course are shown.Representative violin graphs showing 10 metabolites with increased levels over time and five metabolites with reduced levels over time.Dots represent individual samples combined from the five experimental groups.Statistical analysis was performed using one-way ANOVA and Dunnett's multiple comparison between each time point post-infection to the uninfected group (D0) (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

FIG 4 11 FIG 5
FIG 4 Itaconic acid is produced in VACV-infected skin.(A and B) Volcano plot showing significantly different metabolites between (A) NPS-infected tissue without transferred OT-I CD8 + T cells and uninfected control tissues, and (B) NPS-infected tissue with transferred OT-I CD8 + T cells and uninfected control samples.Metabolites with adjusted P-value of less than 0.05 and a fold change greater than 1 are colored as indicated on the graph.Itaconic acid is labeled.(C and D) Violin plots showing itaconic acid levels separated by (C) time point (three biological replicates for each sample group on each dpi) and (D) experimental groups (n = 3 biological replicates per group at each time point).Dots represent individual samples.Statistical analysis was performed using one-way ANOVA and Dunnett's multiple comparison between each time point or experimental group to the D0 uninfected group (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Full 13 FIG 6
FIG 6 Select metabolites change in the skin during T cell-mediated viral clearance.(A)Box plot showing detected levels of glutamic acid.Each data point represents an individual sample.The shapes of the data points indicate the collection time point.Error bars show the minimal to maximal range of the data.Statistical significance was calculated by a multiple t-test with Benjamin-Hochberg false discovery rate correction.(B) Box plot illustrating glutathione levels detected in each experimental group.The shapes of the data points indicate collection time point.Statistical significance was calculated using one-way ANOVA and Dunnett's multiple comparison between each experimental group to the uninfected group (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).(C) Volcano plot showing differentially expressed genes from RNA-Seq data in the glutamate metabolism pathway in NPS-infected tissue with and without OT-I CD8 + T cells.Significantly altered (Continued on next page)