NK cell profiling in West Nile virus encephalitis reveals potential metabolic basis for functional inhibition

Natural killer (NK) cells are cytotoxic lymphocytes important for viral defense. West Nile virus (WNV) infection of the central nervous system (CNS) causes marked recruitment of bone marrow (BM)‐derived monocytes, T cells and NK cells, resulting in severe neuroinflammation and brain damage. Despite substantial numbers of NK cells in the CNS, their function and phenotype remain largely unexplored. Here, we demonstrate that NK cells mature from the BM to the brain, upregulate inhibitory receptors and show reduced cytokine production and degranulation, likely due to the increased expression of the inhibitory NK cell molecule, MHC‐I. Intriguingly, this correlated with a reduction in metabolism associated with cytotoxicity in brain‐infiltrating NK cells. Importantly, the degranulation and killing capability were restored in NK cells isolated from WNV‐infected tissue, suggesting that WNV‐induced NK cell inhibition occurs in the CNS. Overall, this work identifies a potential link between MHC‐I inhibition of NK cells and metabolic reduction of their cytotoxicity during infection.


INTRODUCTION
Natural killer (NK) cells are large granular cytotoxic lymphocytes that recognize and kill virus-infected and cancerous cells, providing a critical first line of defense. 1 Unlike cytotoxic T cells, NK cells do not require prior antigen exposure to mediate their effects.Instead, their effector functions, primarily cell killing and production of proinflammatory cytokines, are controlled by the summation and overall strength of activating and inhibitory signals transduced by germline-encoded receptors. 2][5][6] Further maturation steps predominately take place in the periphery where upregulation of certain molecules subdivide these cells into distinct stages of maturation.Tissue location, microenvironmental cues, host genetic composition, receptor repertoire and transcription factors contribute to the substantial phenotypic and functional diversity of NK cells. 1 Recently, studies have also established the critical link between the upregulation of specific metabolic pathways and effector functions in NK cells. 7,8he critical importance of NK cells to viral defense is demonstrated by the enhanced susceptibility of individuals to infection with naturally occurring NK deficiency. 9,10In mice, NK cells have been implicated in the control of cowpox, 11 murine cytomegalovirus (CMV), 12,13 vaccinia, 14 dengue, 15 influenza and vesicular stomatitis virus. 16Intriguingly, global inhibition of metabolism using either 2-deoxyglucose to block glucose metabolism or rapamycin to block a mammalian target of rapamycin 1 increased susceptibility to CMV mortality by inhibiting NK cell killing, 17 highlighting the importance of metabolism in supporting NK cytotoxicity.The expansion of specific NK subsets in various human infections has also been shown to contribute to viral control.For instance, in human CMV infection, "adaptive-like" NK cells expand to control viral infection. 18,19This population also expands in CMV-positive individuals infected with human immunodeficiency virus, hantavirus, chikungunya virus, hepatitis virus and SARS-CoV-2, impacting disease outcome. 20ntriguingly, unlike in CMV infection, there is no expansion of a detectable "clonal-like" population of human NK cells in response to West Nile Virus (WNV). 21est Nile virus is a mosquito-borne positive-strand RNA flavivirus and considered one of the most important causative agents of human viral encephalitis worldwide. 22he majority of WNV infections are asymptomatic; however, some individuals develop severe neuroinvasive disease including meningitis, encephalitis, acute flaccid paralysis, which can be fatal.Technically, NK cells should have anti-WNV properties. 23However, NK depletion has no effect on the morbidity or mortality of WNV-infected mice. 24This has previously been attributed to the upregulation of MHC-I on WNV-infected cells which transmits an inhibitory signal to NK cells through inhibitory Ly49 receptors, suppressing NK cell killing. 25,26his evasion mechanism is thought to have evolved to counteract NK activation via activating receptor NKp44 by WNV envelope protein binding. 27or a long time, the role of NK cells in flavivirus infection was poorly understood.Recent studies using more sophisticated techniques such as mass cytometry and single-cell RNA sequencing have shed light on the phenotypic and functional responses of human PBMCs to WNV infection. 21,28However, much of our knowledge is derived from NK cells expanded from PBMCs in culture and survival studies in mice with limited phenotypic or functional characterization.More extensive characterization of NK cells in a systemic model is required to determine whether these cells can be therapeutically modulated during WNV infection to enhance viral clearance.Therefore, in this study the phenotypic and functional heterogenicity of NK cells in the brain and BM was investigated over the course of WNV in mice to reveal their distinct immunometabolic, proliferative and cytokine status following lethal infection.

RESULTS AND DISCUSSION
][35] NK cells constitute 20% of this inflammatory infiltrate, with a substantial 80-fold increase in their numbers by day 7 p.i. (Figure 1a-c).The expansion of NK cells in the BM by day 5 p.i. (Figure 1a) likely contributes to their accumulation in the brain and is potentially a result of increased cytokine expression in the periphery, as the BM is not infected with WNV.Importantly, while these cells represent the third largest population in the encephalitic brain, NK cell depletion using a monoclonal antibody targeting NK1.1, had no effect on disease score, weight loss, viral load and the expression of relevant cytokines (Figure 1d-h).Using this method, 96-99% NK cells are depleted in the brain, spleen and BM without affecting other cell subsets (Supplementary figure 1).Supporting previous studies, 24 this suggests that these cells have no impact on viral immunopathology and are merely recruited to the CNS as a byproduct of the inflammatory cytokine milieu.
However, as NK cells have the capacity to kill virusinfected cells, we examined this in more detail by profiling these cells for the expression of maturation, inhibition and activation markers.The expression of maturation markers, CD11b and CD27, divides NK cells into four populations: double-negative immature NK cells, CD27 + proliferative NK cells, double-positive effector NK cells and CD11b + terminally mature NK cells. 36,37mmature and proliferative NK cells mainly express genes related to cell cycle and proliferation, while effector and terminally mature NK cells have cytotoxic and cytokineproducing functions. 36Importantly, the BM NK cell population was mostly composed of CD27 + proliferative cells with a small number of double-positive effector cells increasing in proportion with infection (Figure 2a, b).Similarly, the few NK cells (4 9 10 3 ) in the homeostatic brain were mostly CD27 + proliferative cells; however, from day 5 p.i. onwards double-positive effector cells comprised the major NK population in the brain.Interestingly, the double-positive effector and proliferative subsets were slightly reduced by day 7 p.i., with increasing proportions of CD11b + terminally mature cells (Figure 2c).Overall, the increasing maturation status of NK cells from the BM compared with brain suggest they could have the potential for cytolytic and cytokineproducing functions in the infected CNS.
As the activation status of NK cells is determined by the overall summation and strength of germ line encoded receptor signaling, we profiled NK cell subsets for relevant inhibitory and activation markers.NK cells were examined from day 5 p.i. in the brain as there is a limited number of infiltrating cells prior to this timepoint.Interestingly, increasing proportions of BM and brain NK cells expressed higher levels of the inhibitory receptor KLRG1 as they matured from proliferative to effector or terminally mature cells, but, as time progressed, reduced their expression of CD94 as time progressed (Figure 2d-f), which can be inhibitory or activating. 38Further, at day 7 p.i. a significantly smaller proportion of brain NK cells expressed the inhibitory receptor Ly49C, although the reduced number of terminally mature NK cells that expressed this receptor by day 7 did so at higher levels than day 5 terminally mature NK cells (Figure 2f).On the other hand, while increasing proportions of BM NK cells expressed  (d-g) clinical score (d), animal weights (e), percent of weight lost (f) and viral load (g), as determined by the number of plaque-forming units (PFU) using a virus plaque assay on brain tissue isolated at day 7 p.i. in mice treated with an isotype control or anti-NK1.1 monoclonal antibody.(h) Expression of Wnv and selected chemokines and pro-inflammatory genes at day 7 p.i. in the brain of NK-depleted and nondepleted mice.Data are presented as mean AE s.e.m. from one (g, h) or one of three (a-f) independent experiments with at least four mice per group.Ly49C with maturity, effector cells decreased their expression of this marker by day 7 (Figure 2e).While the precise mechanisms promoting the upregulation or downregulation of these markers is unclear, increased KLRG1 expression on NK cells has also been shown to occur in chronic hepatitis virus, 39 human immunodeficiency virus 40 and mouse CMV infection. 41xpression of this marker is inversely correlated with IFN-c production and NK cell function. 39,42Indeed, antibody blockade of KLRG1 promoted IFN-c production by NK cells in chronic hepatitis virus 39 and restored the capacity of NK cells to kill human immunodeficiency virus infected cells, 40 highlighting a potential therapeutic target in WNV infection to promote virus clearance.As the expression of these markers did not give a conclusive answer on the activation status of NK cells in the WNV brain, we next profiled their proliferative, cytolytic and cytokine response following WNV infection.
To assess the proliferative capacity of NK cells, mice were injected with bromodeoxyuridine (BrdU), a thymidine analogue that incorporates into synthesizing DNA.T cells were used as a positive control as these cells show the highest proliferative capacity after microglia in the WNV-infected brain as well as the highest expression of IFN-c. 31NK cells in the brain showed limited incorporation of BrdU, suggesting minimal proliferation (Figure 3a, b).This may not be surprising as the majority of NK cells in the brain show an effector phenotype.This was in contrast to BM NK cells which had the highest proliferative capacity among the NK cells and T cells in both compartments (Figure 3a).In terms of cytokine expression, NK cells expressed low amounts of IFN-c in the brain at 7 p.i. relative to T cells (Figure 3a, b).Notably, as IFN-c is expressed highly in the WNV-infected brain this was analyzed without an in vitro stimulation.In contrast, to quantify granzyme B and the degranulation marker CD107a, cells were incubated for 4 h at 37°C with brefeldin A. Interestingly, 32% of brain NK cells expressed granzyme B (Figure 3c, d).However, only 3% of these cells expressed the degranulation marker CD107a (Figure 3c, d), suggesting that while brain NK cells express cytolytic molecules, these are not being released. 21,27This could imply that brain NK cells have cytolytic capacity but remain inhibited within the infected brain. 25In contrast, BM NK cells expressed less granzyme B and more CD107a (Figure 3c, d).
In support of previous findings, we showed a substantial increase in the expression of MHC-I (H2-K b ) in the brain and BM over the course of WNV infection (Figure 4a-c). 25,27The expression of MHC-I was higher in the brain than the BM, increasing from day 3 p.i. on CD45 + brain cells and on CD45 À brain cells (astrocytes, oligodendrocytes, endothelial cells and neurons) from day 5 p.i. (Figure 4a-c).MHC-I is a "self" signal that strongly inhibits NK cell activation by binding to Ly49 inhibitory receptors. 43The decreased proportion of NK cells expressing Ly49C (Figure 2f) may be explained by the upregulation of Ly49C ligands such as MHC-I.Blocking the interaction between Ly49C and H2-K b on a model tumor cell enhanced NK-mediated cytotoxicity, 44 highlighting a potential target in WNV infection to evade MHC-I expression and to promote NK killing.
To investigate the relationship between NK cell killing and MHC-I upregulation in WNV infection, we cocultured fluorescent-activated cell-sorted brain and BM NK cells with RMA-S cells.These are a TAP-deficient lymphatic cell line unable to express MHC-I and thus are theoretically sensitive to NK cell killing. 45,46Importantly, both brain and BM NK cells were equally capable of killing RMA-S target cells in a Calcein release assay, as assessed through the release of the cell permeant dye by RMA-S into the supernatant, indicating direct cell lysis by NK cells (Figure 4d).The proportional increase in CD107a expression by NK cells with decreasing effectorto-target ratios supports the quantitative analysis of this assay (Figure 4e).This is because decreasing effector cell numbers increases the probability of exposure of any single NK cell to a target cell(s) thereby increasing the proportion of degranulating effector cells.
To confirm these findings, we used a flow cytometrybased assay using 7AAD to more accurately determine RMA-S viability.Interestingly, we demonstrated a 20% increase in the killing capability of BM NK cells relative to brain NK cells; however, this is likely explained by the 20% increase in the death of brain NK cells after culturing for 4 h (Figure 4f, g).This suggests equal killing capability across organs.Importantly, however, RMA-S killing increased in proportion to increasing NK numbers despite NK cell death.NK cell death is most likely explained by the lack of IL-2 added to culture, which is required to maintain cell viability; 47 however, IL-2 was omitted to preserve the brain-derived NK cell phenotype.Further, the increased death of brain NK cells may be due to the enzymatic dissociation of the brain which is not required for BM cell isolation.Importantly, this demonstrates that NK cells from WNV-infected mice are capable of killing target cells if MHC-I expression is absent and thus the interaction of this molecule with NK cells is bypassed.It has previously been demonstrated that human NK cells from PBMCs develop robust and polyfunctional responses to WNV in vitro. 21Further, co-culturing PBMCs with K562D2 stimulatory cells to expand a pure and active population of NK cells highly expressing activation receptors, inhibited WNV infection of Vero cells through both cytolytic and noncytolytic activities. 48Together, this implies that transfer of in vitro expanded NK cells during WNV infection could be used to combat infection, particularly if WNV-induced MHC-I expression and subsequent NK inhibition is reduced.
In order to better understand why the functional capacity of NK cells is seemingly not fulfilled in WNV encephalitis, we profiled the metabolic status of these cells as this is intrinsically linked to functional output. 7,80][51][52] Activated NK cells preferentially utilize glucose-driven glycolysis and oxidative phosphorylation (OXPHOS) for the production of energy in the form of ATP and an enhanced biosynthetic capacity to enable NK cell proliferation and effector functions. 53,54Not surprisingly, NK cells in the brain had a higher expression of glycolysis marker, GAPDH, OXPHOS-associated marker, IDH1, and amino acid transport, CD98, at day 5 p.i. and GAPDH and CD98 at day 7 p.i., compared with BM NK cells, potentially supporting migration and effector function in the CNS (Figure 5).However, GAPDH and IDH1 were significantly downregulated by brain NK cells by day 7 p.i. (Figure 5).This suggests that upon entry into the brain (about day 5 p.i.) these cells reduce their metabolic requirements as viral replication progresses, resulting in NK cell inhibition in the brain.The limited metabolic changes in the BM from day 5 to 7 p.i. may be related to the lower expression of inhibitory molecule MHC-I in the BM, relative to the brain (Figure 4c).Recently, a proof of concept study demonstrated the druggability of the glycolytic pathway during WNV infection, with pharmacological inhibitors of glycolysis reducing neuroinflammation in a mouse model of WNV infection. 55While this may be protective to prevent pathogenic immune responses coordinated by specific cell subsets such as brain-infiltrating macrophages which promote CNS damage, this can also inhibit protective virus clearance functions by NK cells.Cell-specific metabolic inhibitors are thus required to target pathogenic cells fueled by glycolysis.
Surprisingly, the amino acid transporter CD98 showed increased expression on brain NK cells in the same timeframe (Figure 5).Increases in nutrient receptors such as CD98 are known to accompany increases in OXPHOS and glycolysis with NK activation. 56,57Additionally, CD98 was upregulated on NK cells in CMV infection in a MyD88-dependent fashion and thought to contribute to NK-mediated viral clearance, as MyD88-deficient mice are unable to express CD98 and show higher viral loads. 58owever, other studies have shown that regulatory resident NK cells are more reliant on amino acids (requiring e.g.CD98) while cytotoxic NK cells increase glucose metabolism after cytokine stimulation. 59,60This is supported by the higher expression of CD98 and transferrin receptor CD71 on regulatory resident NK cells compared with peripheral blood NK cells, which have a higher expression of glucose transporter GLUT1. 60Overall, this may suggest that the downregulation in glycolysis and OXPHOS pathways and a corresponding increase in amino acid transporter CD98 may facilitate the generation of an enhanced regulatory NK cell phenotype with reduced cytotoxicity and killing in the brain.NK cell suppression by MHC-I may explain the downregulation of glycolysis and OXPHOS pathways by NK cells and their low expression of CD107a and IFN-c.Indeed, disruption of these metabolic pathways is associated with reduced IFN-c and granzyme expression by NK cells. 57,61More mechanistic studies are required to confirm the association between NK cell metabolism, MHC-I and functional output.Alternative techniques such as the extracellular flux assay may be useful to examine these pathways more broadly.However, the required cell sorting and in vitro stimulation make it unlikely that it will faithfully represent the in vivo status of these cells which is more precisely determined by flow cytometric assessment of protein expression.
Overall, this work provides detailed kinetic profiles of NK cells in the BM and brain of WNV-infected mice over the course of infection, defining their changes in maturation, inhibitory, proliferative, cytokine and immunometabolic status for the first time.We demonstrate that while NK cells mature upon migration from the BM to the brain, they upregulate inhibitory receptors, and evidently have a limited ability to produce IFN-c and degranulate.Furthermore, the reduced expression of GAPDH and IDH1 at day 7 p.i. suggests that these cells decrease their metabolic requirements upon entry into the brain, suggesting NK cell suppression.Isolation of these cells from WNV-infected tissues enabled the killing of MHC-I-deficient target cells, supporting a link between NK cell suppression and flavivirus-mediated upregulation of MHC-I.Future studies, however, are required to determine whether the identified pathways can be targeted therapeutically to promote virus clearance mechanisms.
The unique capacity of NK cells to provide rapid viral clearance without prior sensitization makes them an important therapeutic tool.More recent work has also identified the existence of human memory NK cells, allowing for the development of vaccination-based approaches. 62Currently, NK cell therapeutics are being explored for the treatment of human malignant diseases 63 and viral diseases, including coronavirus disease 2019. 64hus, better understanding of the mechanisms underlying NK cell inhibition in WNV infection may better inform the development of NK cell immunotherapeutics across a wide range of diseases.

METHODS WNV-infection of mice
Female 9-10-week-old C57BL/6 mice from the Animal Resource Centre (ARC, Western Australia, Australia) were kept in individually ventilated cages under specific pathogenfree conditions with access to food and water ad libitum.All experiments were performed in accordance with National Health and Medical Research Council's ethical guidelines with the animal ethics approval number K20/05-2016/976 and 2019/1696 approved by the University of Sydney Animal Ethics Committee.Mice were anesthetized with avertin (0.012 mL per g of body weight) prior to being infected intranasally with WNV (Sarafend) delivered in 10 lL of sterile phosphate-buffered saline (PBS) (as described previously 29 ).Mice were infected with 1.2 9 10 5 plaque forming units (PFU) of WNV a dose that is lethal in 100% (LD100) of mice.Mice were killed no later than day 7 p.i.

Detection and quantification of proliferating cells with BrdU
Mice were injected intraperitoneally with 1 mg of bromodeoxyuridine (BrdU; Sigma-Aldrich, USA) in 200 lL sterile PBS 3 h before sacrifice to detect proliferating cells.

Tissue processing for flow cytometry
All mice were anesthetized and transcardially perfused with ice cold PBS prior to tissue collection.Femurs were dissected out and flushed with cold PBS using a 30-gauge needle, while spleens were gently forced through a 70 lM nylon mesh sieve using a syringe plunger.Red blood cell lysis buffer (Invitrogen, USA) was used to lyse erythrocytes in single-cell suspensions of BM cells and splenocytes.Brains were processed into singlecell suspensions in PBS and DNase I (0.1 mg mL À1 , DN25, Sigma-Aldrich, USA) and collagenase type IV (1 mg mL À1 , C5138, Sigma-Aldrich, USA) using the gentleMACS dissociator (Miltenyi Biotec, DE).Subsequently, a 30%/80% Percoll gradient was used to isolate the cells from brain homogenates.After tissue processing, live cells were counted with trypan blue (0.4%) on a hemocytometer.Single-cell suspensions were incubated with purified anti-CD16/32 (Biolegend, USA) and UV-excitable LIVE/DEAD Blue (UVLD; Invitrogen, USA), Zombie NIR or Zombie UV TM Fixable Viability kit (Biolegend, USA) and subsequently stained with a cocktail of fluorescently-labeled antibodies (See Supplementary table 1).Cells were washed twice and fixed in fixation buffer (Biolegend, USA) or in Cytofix/Cytoperm (BD Biosciences, USA), if staining for intracellular markers, or True-Nuclear Transcription Factor Buffer Set (Biolegend, USA), if staining for metabolic markers.To determine the expression of CD107a and granzyme B, single-cell suspensions were incubated with brefeldin A (5 lg mL À1 , Enzo Life Sciences, USA) and fluorescently-labeled anti-CD107a for 4 h at 37°C and subsequently stained with anti-granzyme B after surface staining, fixation and permeabilization (Supplementary table 1).Anti-BrdU was stained intranuclearly, as described previously. 65Briefly, after cell surface staining and fixation, cells were incubated in Cytofix/Cytoperm (BD Biosciences, USA), Cytoperm Permeabilization Buffer Plus (BD Biosciences, USA) and DNase (DN25, 30 U/sample; Sigma-Aldrich, USA), prior to being stained with anti-BrdU.
Fluorescently-tagged antibodies were measured on a LSR-II fluorescence-activated cell sorter (FACS; Becton Dickinson, San Jose, CA) or 5 laser Aurora (Cytek Biosciences, USA).Acquired data were analyzed in FlowJo (BD Biosciences).Quality control gating including time, single cells, non-debris/cells and LIVE/DEAD staining was applied to exclude debris, doublets and dead cells.Cell subsets were identified using gating strategies shown in Spiteri et al. 2021 31 and Spiteri et al. 2023. 66Cell numbers were quantified using cell proportions exported from FlowJo and total live cell counts.Uniform Manifold Approximation and Projection (UMAP) was applied to CSV files, in RStudio using Spectre 67 with default settings, i.e. perplexity = 30, theta = 0.5 and iterations = 1000.

NK killing assay
The mutant lymphoma RMA-S cell line was kindly donated by Associate Professor Alex Sharland and Eric Son to assess NK cell killing ability.These cells are sensitive to NK cells as they are TAP-deficient and have a reduced cell surface expression of MHC-I molecules.NK cells were sorted from the brain and BM of WNV-infected mice on day 7 p.i.The "Tissue processing for flow cytometry" section above describes our tissue processing procedures and Supplementary table 1 lists the antibodies used for cell sorting.Cells were sorted into FBS on the 7-laser Influx cell sorter using the FACSDiva Programme (BD Biosciences).Sorted NK cells were then centrifuged, resuspended in RPMI (Lonza, CH) and 10% FBS, counted and plated in a 96 v-bottom well plate at effector to target ratios of 4:1, 2:1 and 1:1.All cells were 100% viable before being co-cultured.

Calcein release assay
One million RMA-S cells were stained with Calcein AM Viability dye (ThermoFisher Scientific, USA) in PBS (10 lM final concentration) for 30 min at 37°C.Cells were topped up with media, centrifuged and recounted before plating 1.1 9 10 4 cells per well in a total volume of 250 lL.RMA-S and NK cells were co-cultured for 4 h at 37°C before the supernatant was collected and analyzed on the TECAN Infinite M1000 Pro (excitation 485 nm, emission 530 nm, bandwidth 9 nm).RMA-S and NK cells were then stained with anti-CD107a for 30 min at 37°C and analyzed on the 5 laser Aurora (Cytek Biosciences).Spontaneous and maximum release was determined by quantifying the fluorescence of RMA-S cells alone and RMA-S cells treated with 2% triton, respectively (6 technical replicates per condition).Percent of killing was determined with the following formula: (experimental release-spontaneous release)/(maximum releasespontaneous release).Media and NK cells alone controls were also used to determine background fluorescence which was 14-fold below the maximum release value.

Flow cytometry 7AAD-based assay
Five million RMA-S cells were washed with PBS and stained with CellTrace TM Violet (ThermoFisher Scientific, USA) in PBS (0.15 lM final concentration) for 15 min at 37°C.Cells were topped up with media, centrifuged and recounted before plating 3 9 10 4 cells per well in a total volume of 200 lL.RMA-S and NK cells were co-cultured for 4 h at 37°C, stained with 7AAD (BD Biosciences, USA) and analyzed on the 5 laser Aurora (Cytek Biosciences).

Quantification of viral titer using a plaque assay
Virus-susceptible Baby Hamster Kidney fibroblast (BHK) cells were used to perform a virus plaque assay as described previously. 29BHK cells were seeded into a 12-well plate and infected with serial 10-fold dilutions of brain tissue homogenates for 1 h, at which time the homogenate was removed and the monolayer was overlaid with lowtemperature Agarose plug.Cells were incubated for a further 3 days before being fixed with 10% formalin (Sigma-Aldrich, USA) and stained with a 1% crystal violet solution (Sigma-Aldrich, USA).The plaque-forming units (PFU) per gram was calculated using the number of plaques, the inoculum volume and the dilution.

RNA extraction and real-time quantitative polymerase chain reaction
Brain tissue was homogenized in TRI Reagent (Sigma Aldrich, USA) using a tissue homogenizer (TissueLyser, Qiagen, DE).The High-Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific, USA) was used to generate cDNA and the Power SYBR TM Green PCR Master Mix (ThermoFisher Scientific, USA) was used to conduct qPCR, using primers all purchased from Sigma Aldrich (see Supplementary table 2), USA on the LightCycle â 480 Instrument II (Roche, CH), as described previously. 68Gene expression values were normalized to Rpl13a.

Statistical analysis
Statistical tests were carried out on data in GraphPad Prism 8.4.3 (GraphPad Software, La Jolla, CA), using one-way or two-way ANOVA to compare data with one or two variables, respectively.All datasets with one variable passed the Shapiro-Wilk normality test.Error bars are shown as standard error of the mean (s.e.m.).

Figure 1 .
Figure 1.NK cells do not contribute to viral immunopathology in WNV infection of the CNS.(a) Number of NK cells in the brain and BM following intranasal WNV infection.(b, c) Number (b) and UMAP plot (c) of CD45 + leukocyte subsets in the WNV-infected brain at day 5 and/or 7 p.i.(d-g) clinical score (d), animal weights (e), percent of weight lost (f) and viral load (g), as determined by the number of plaque-forming units (PFU) using a virus plaque assay on brain tissue isolated at day 7 p.i. in mice treated with an isotype control or anti-NK1.1 monoclonal antibody.(h) Expression of Wnv and selected chemokines and pro-inflammatory genes at day 7 p.i. in the brain of NK-depleted and nondepleted mice.Data are presented as mean AE s.e.m. from one (g, h) or one of three (a-f) independent experiments with at least four mice per group.

Figure 2 .
Figure 2. NK cells differentially express maturation, inhibitory and activating markers in the brain and BM of WNV-infected mice.(a) Flow cytometry dot plot showing the expression of CD27 and CD11b on BM and brain NK cells over the course of WNV infection.(b, c) Proportion of NK cell subsets according to CD27 and CD11b expression in the BM (b) and brain (c) over the course of infection.Statistics compare datapoints between timepoints within each subset and are relative to mock-infected mice.(d) Histograms showing the expression of CD11b, CD27, KLRG1, CD94 and Ly49C on total NK cells in the BM and brain.(e, f) Frequency of NK cell subsets in the BM (e) and brain (f) expressing KLRG1, CD94 and Ly49C and the median fluorescent intensity (MFI) of the positive population for each marker.Data are presented as mean AE s.e.m. from one of two independent experiments with at least three mice per group.

Figure 3 .
Figure 3. NK cells show a reduced degranulation ability in the WNV-infected CNS.(a) Flow cytometry dot plots showing the expression of IFN-c and BrdU on brain T cells and a fluorescence-minus-one control (FMO).(b) Proportion of NK and T cells from the brain and BM expressing IFN-c and incorporating BrdU at day 7 p.i. (c, d) Flow cytometry dot plot (c) and bar graph (d) showing the expression of CD107a and granzyme B in the BM and brain at day 7 p.i. Data are presented as mean AE s.e.m. from one of two (c, d) or from two (a, b) independent experiments with at least four mice per group.

Figure 4 .
Figure 4. NK cells isolated from WNV-infected tissue are capable of killing target cells upon bypassing MHC-I inhibition.(a) UMAP plot showing the expression of MHC-I (H-2Kb) on brain cells from the WNV-infected brain at day 7 p.i. (b, c) Expression of MHC-I on BM and brain cells over the course of WNV infection.(d) Killing capability of BM and brain NK cells, as determined using a Calcein release assay with RMA-S cells as targets.(e) Expression of CD107a on BM and brain NK cells after a 4 h co-culture with RMA-S cells.(f, g) Percent of dead RMA-S (f) and sorted NK cells (g) after a 4 h co-culture, as determined using 7AAD by flow cytometry.Data are presented as mean s.e.m. from one (d-g) or one of two (a-c) independent experiments with at least four mice per group.MC, monocyte-derived cell.

Figure 5 .
Figure 5. Brain NK cells reduce their metabolic requirements after infiltrating the WNV-infected CNS.Expression of GAPDH (glycolysisassociated), IDH1 (OXPHOS-associated) and CD98 (amino acid transporter) on NK cells from the BM and brain over the course of WNV infection.Data are presented as mean AE s.e.m. from one of two independent experiments with at least four mice per group.