Experiments involved sex- and age-matched inbred C57BL/6 mice from Harlan Laboratories, Germany, maintained in a pathogen-free environment. The mice were a minimum of 7 weeks old at the start of the experiments. The FV complex, consisting of B-tropic Friend murine leukemia helper virus and polycythemia-inducing spleen focus-forming virus, was injected. The FV stock was prepared as a 15% spleen cell homogenate from susceptible BALB/c mice infected for 14 days with 4,000 SFFU of FV. Mice received intravenous injections of 0.1 ml phosphate-buffered saline (PBS) containing 40,000 SFFU of FV. Notably, the virus stock did not contain lactate dehydrogenase-elevating virus. All experiments were approved and conducted in compliance with the State Agency for Nature, Environment, and Consumer Protection (LANUV).

Spleens were homogenized and washed with PBS. Splenocytes were plated onto 96 well plates and stained for 15 min at room temperature in the dark. Cells were washed and acquired at multi-parameter flow cytometer or fixed for further intracellular (Fixation/Permeabilization Solution Kit, BD Biosciences) or intranuclear stainings (Foxp3 Transcription Factor Fixation/Permeabilization kit, ThermoFisher). For cytokine stainings, splenocytes were restimulated with ionomycin (500 ng/ml), phorbol myristate acetate (PMA; 25 ng/ml), monensin (1X, BioLegend), and brefeldin A (2 μg/ml) diluted in RPMI buffer at 37°C for 3 h. Multi-parameter flow cytometry was done with the following antibodies: CD3 (clone 145-2C11, Thermo Fisher), CD36 (HM36, BioLegend), CD49b (DX5, Thermo Fisher), CD69 (H1.2F3, BioLegend), GzmB (GB11, Becton Dickinson), IFN-γ (XMG1.2, Thermo Fisher), NK1.1 (PK136, Becton Dickinson), pS6 (D57.2.2E, Cell Signaling). HCS LipidTOX Green (ThermoFisher) was stained 1:1000 after fixation of cells before acquisition.

RPMI1640 with 10% Fetal Bovine Serum and 100 U/ml Penicilin/100 µg/ml Streptomycin was prewarmed and splenocytes were added in duplicate on 96 well plates. Inhibitors were added (final concentrations: 2-DG 100 mM, oligomycin 1 µM, etomoxir 5 µM, 40 µM) and incubated for 30 min at 37°C. Puromycin was added (final concentration 10 µg/ml, Sigma) for 20 min. Cells were washed with cold PBS and FC block (BD Biosciences) was performed at 4°C for 5 min. Cell surface was stained at 4°C for 30 min in the dark and cells were washed. Cells were fixed with Fixation/Permeabilization Solution Kit (BD Biosciences). Anti-puromycin (Sigma, clone 12D10) staining was performed in perm-buffer at 4°C for 1 h. Cells were measured at BD Canto II. Glucose dependency and fatty acid capacity was calculated by using the geoMFI: (no inhibitor – 2-DG)/(no inhibitor – 2-DG Oligo). Mitochondrial dependency and glycolytic capacity was calculated by using the geoMFI: (no inhibitor – oligomycin)/(no inhibitor – 2-DG Oligo).

Splenocytes were plated on 96 well plates. Bodipy FL C16 (4,4-Difluoro-5,7-Dimethyl-4-Bora-3a,4a-Diaza-s-Indacen-3-Hexadecanoacid, Invitrogen) was diluted in supplemented RPMI1640 in 12,5 µM (final concentration) and added to wells. Incubation at 37°C for 30 min in a humidified atmosphere with 5% CO₂. Cells were stained and measured at BD Canto II.

NK cells were extracted from spleens using the MojoSort Mouse NK cell Isolation Kit from BioLegend, following the manufacturer’s instructions. YAC-1 target cells, labeled with carboxyfluorescein succinimidyl ester (CFSE, 2.5 μM), were co-incubated with NK cells at an E:T ratio of 25:1 in prewarmed 1xDMEM (Sigma D2429, supplemented with 3.7 g/L sodium bicarbonate, 10% FCS, 50 μM β-mercaptoethanol). The co-incubation occurred in 96-well U-bottom plates at 37°C in a humidified 5% CO2 atmosphere. After 4 hours, cells were washed and subsequently stained with a fixable viability dye. Fluorescence was acquired at BD Canto II. Killing was calculated with (dead tumor cells)/(all tumor cells) x 100.

Splenocytes were lysed for erythrocytes and plated in 96 well plates. Cells were cultured for two days in total with IL-15 (12,5 ng/ml) and when indicated, etomoxir (5 µM, 40 µM) were added to culture. IL-2 (20 ng/ml) and IL-12 (10 ng/ml) were added for the last 24 hours of IL-15 incubation. Cells were stained and acquired at flow cytometer.

NK cells were isolated using MojoSort Mouse NK cell Isolation Kit from BioLegend and stained with Tag-It Violet (BioLegend, 25 µM final). YAC-1 cells were labeled with CFSE from BioLegend (final concentration 5 µM) and used in an E:T ratio of 1:2. Etomoxir (40 µM) was added to co-culture. Cells were centrifuged for 1 min at 20 g and co-incubated for 1 min and 60 min at 37°C. After gentle mixing, ice-cold PFA (0.5%) was added and cells were measured at flow cytometer. Calculation was performed by dividing the conjugates (CFSE⁺Violet⁺) through all NK cells (Violet⁺), multiplied by 100.

Bead isolated NK cells (MojoSort, BioLegend) were added in the upper compartment of transwells (pore size 3 µM, Costar). Enriched samples contained around 77% NK cells. YAC-1 cells were seeded in the lower compartment. Cells were seeded in an E:T ratio of 1:2 in 1xDMEM for 4 hours at 37°C. Etomoxir (40 µM) was added in both compartments, if indicated. Cells were stained and NK cells were analyzed and flow cytometer.

NK cells were isolated using magnetic beads from BioLegend and frozen in DNA/RNA Shield from Zymo Research. Enriched samples contained at least 77% NK cells. RNA was isolated with RNeasy Micro Kit from Qiagen. cDNA was synthesized with SCRIPT cDNA Synthesis Kit from Jena Biosciences. The real-time PCR was conducted in duplicate using the innuMIX qPCR DSGreen Standard (Innuscreen) and the Rotor-Gene Q (Qiagen). The quantitative mRNA levels were normalized to β-actin mRNA expression levels. The following oligonucleotide sequences were used: β-actin 5’-aaatcgtgcgtgacatcaaa-3’, 5’‐caagaaggaaggctggaaaa-3; PPARα 5’ CGGGAAAGACCAGCAACAAC 3’, 5’ TGGCAGCAGTGGAAGAATCG 3’; PPARγ 5′- GTACTGTCGGTTTCAGAAGTGCC -3′, 5′- ATCTCCGCCAACAGCTTCTCCT 3′; SCARB1 5′- ACACCCGAATCCTCGCTGGAAT -3′, 5′- CCGTTGGCAAACAGAGTATCGG -3′; SCARB2 5′- TAGCCAACACCTCCGAAAACGC -3′, 5′- CGAACTTCTCGTCGGCTTGGTA -3′; FABP5 5′-GACGACTGTGTTCTCTTGTAACC-3′, 5′- TGTTATCGTGCTCTCCTTCCCG-3′; CD36 5′-GGACATTGAGATTCTTTTCCTCTG-3′, 5′-GCAAAGGCATTGGCTGGAAGAAC -3′; CPT1α 5′- GGCATAAACGCAGAGCATTCCTG-3′, 5′-CAGTGTCCATCCTCTGAGTAGC-3′;

Caprylic acid (FA 8:0), myristic acid (FA 14:0), palmitic acid (FA 16:0), palmitoleic acid (FA 16:1 Δ9), stearic acid (FA 18:0), oleic acid (FA 18:1 Δ9), linoleic acid (FA 18:2 Δ9,12), and linolenic acid (FA 18:3 Δ9,12,15) were purchased from Sigma-Aldrich (Traufkirchen, Germany). Decanoic acid (FA 10:0), lauric acid (FA 12:0), arachidic acid (FA 20:0), arachidonic acid (FA 20:4 Δ5,8,11,14), eicosapentaenoic acid (FA 20:5 Δ5,8,11,14,17), docosanoic acid (FA 22:0), docosapentaenoic acid (FA 22:5 Δ7,10,13,16,19), docosahexaenoic acid (FA 22:6 Δ4,7,10,13,16,19), lignoceric acid (FA 24:0), hexacosanoic (FA 26:0) acid as well as the stable isotope labled internal standards ²H₂-decanoic acid (²H₂-FA 10:0), ²H₂-pentadecylic acid (²H₂-FA 15:0), ²H₄-stearic acid (²H₄-FA 18:0), ²H₈-arachidonic acid (²H₈-FA 20:4 Δ5,8,11,14), and ²H₄-lignoceric acid (²H₄-FA 24:0) were purchased from Cayman Chemical Company (Hamburg, Germany). 2,3,4,5,6-Pentafluorobenzyl bromide (for GC derivatization; ≥98.5%) and N,N‐diisopropylethylamine (≥99.5%) were purchased from Sigma-Aldrich (Traufkirchen, Germany). Dichloromethane (≥99.8%) was purchased from Thermo Fisher Scientific (Schwerte, Germany), methyl tert-butyl ether (MTBE; LC-MS grade), and acetic acid (LC-MS grade) were purchased from Merck (Darmstadt, Germany), methanol (LC-MS grade) was purchased from Avantor (Darmstadt, Germany). Potassium hydroxide (pro analysi) was purchased from Bernd Kraft GmbH (Duisburg, Germany). Ultrapure water with a resistivity of 18.2 M Ω/cm was desalted and filtered by a Sartorius Stedim water purification system (Sartorius, Goettingen, Germany). For lipid extraction, a 5 µL serum sample was combined with a 100 µL methanol solution containing internal standards, including ²H₂-decanoic acid (²H₂-FA 10:0), ²H₂-pentadecylic acid (²H₂-FA 15:0), ²H₄-stearic acid (²H₄-FA 18:0), ²H₈-arachidonic acid (²H₈-FA 20:4 Δ5,8,11,14), and ²H₄-lignoceric acid (²H₄-FA 24:0), each at a concentration of 2,000 nM. To hydrolysis complex lipids, 60 µL of 10 M potassium hydroxide was added to the mixture. The samples were homogenized in an ultrasonic bath for 5 minutes at 0°C and then incubated at 60°C for 30 minutes to ensure complete hydrolysis. The solution was neutralized by adding 70 µL of 50% acetic acid (v/v) while keeping the samples on ice. Afterwards, the samples were subjected to lipid extraction based on the method outlined by Matyash et al. (26), with some modifications. This involved the addition of 300 µL of methanol, followed by vortexing for 1.5 minutes. Subsequently, 600 µL of methyl tert-butyl ether (MTBE) was added and the mixture was vortexed for another 1.5 minutes. Phase separation was induced by the addition of 300 µL, and the mixture was again vortexed for 1.5 minutes. The upper phase, containing the fatty acids, was then collected. The aqueous phase was washed with 300 µL of MTBE and combined with the previously collected MTBE phases. Finally, the samples were dried using a vacuum evaporator at 30°C. The dried lipid extracts were reconstituted in 20 µL of a solution of 10% diisopropylethylamine in dichloromethane (1/9; v/v) with the addition of 10% 2,3,4,5,6 pentafluorobenzylbromide in dichloromethane (1/9; v/v). After the incubation at 50°C for 1 hour, the solvent was evaporated under vacuum at 30°C. The derivatized fatty acids were dissolved in 100 µL of methanol for subsequent analysis. Fatty acid analysis was conducted following the procedure described by Görs et al. (27). An Agilent 7890B GC, equipped with a liquid autoinjector and a DB-5 column (30 m x 250 μm x 0.25 μm) (Agilent Technologies, Waldbronn, Germany), was used for gas chromatographic analysis. A 1 µL sample was injected with a split ratio of 1:10 and a septum purge flow of 1.5 mL/min at 320°C. Helium was used as the carrier gas at a flow rate of 1.3 mL/min. The temperature gradient was as follows: initial 100°C, increased at 15°C/min to 160°C, followed by a linear increase of 5°C/min to 320°C, which was held for 5 minutes. The GC was coupled with an APCI source to a 6495 triple Quad mass spectrometer (Agilent Technologies, Waldbronn, Germany). The APCI used a corona current of 35 µA, a source gas flow of 11 L/min, a source gas temperature of 270°C, a nitrogen auxiliary gas flow of 1 L/min, and a capillary voltage of 2000 V. Negative ionization was used, and ions were monitored in pseudo SRM mode. Therefore, the m/z ratios of the [M-PFB]- ions of the corresponding fatty acid was set as m/z for Q1 and Q3, with no fragmentation energy applied. The cycle time was 300 ms.

GraphPad Prism version 8 was employed for graph preparation and statistical analysis. Statistical differences between two groups (naïve, 7 dpi) were assessed using either the Mann–Whitney test (nonparametric) or unpaired t-test (parametric). The analysis of multiple groups (naïve, FV, treated) utilized the nonparametric Kruskal–Wallis one-way analysis of variance (ANOVA) on ranks and Dunn’s multiple-comparison test (nonparametric), or ordinary one-way ANOVA with Tukey multiple-comparison test (parametric). Correlating variables, linear regression and Goodness of fit were determined.

C57BL/6 mice were infected with 40,000 Spleen Focus-Forming Units (SFFU) of FV and examined at 7 dpi. Splenic NK cells from both naïve and FV-infected mice were analyzed for activation by measuring CD69. As shown in Figure 1A , NK cell activation is approximately 2.7 times higher in FV-infected compared to naïve animals. In addition to the augmented activation of NK cells during acute infection, the percent of GzmB expressing NK cells was higher upon FV infection ( Figure 1B ). Nevertheless, the expression of GzmB on NK cells was not changed ( Figure 1C ). Increased cytotoxic activity of NK cells requires metabolic reprogramming. Thus, we determined the mTOR activity in NK cells, which is a master regulator that senses nutritional and environmental signals and promotes cellular growth. Here we analyzed phospho-S6 (pS6), a marker for the mTOR signaling pathway, over time in FV infection ( Figures 1D, 1E ). We found the peak of pS6 expression at 7 dpi, which was already reduced at 14 dpi ( Figure 1E ). These findings demonstrate that NK cells experience increased activation and undergo metabolic reprogramming during the acute phase of FV infection.

Activated immune cells have higher energy demands compared to quiescent cells. Energy is provided in form of ATP to cells. We have generated an approximation of ATP demand in NK cells by using the SCENITH method (28). This method works based on the premise that the most energy demanding process in the cell is protein synthesis. Puromycin is added to the cell and is then incorporated into newly synthesized proteins. Puromycin incorporation was significantly higher in NK cells during acute infection compared to NK cells from naïve mice demonstrating that NK cells responding to acute infection have increased rates of protein synthesis and an increased demand for ATP to support the protein translation ( Figures 2A, B ). If in a parallel sample puromycin plus a mitochondrial inhibitor or glycolysis inhibitor is added, it will end in less puromycin incorporation because ATP is now limited due to the inhibition of mitochondrial activity or glycolysis. In this way you can gauge how a cell is making its ATP, whether it is through glycolysis, glucose-fueled mitochondrial OXPHOS or metabolism of FAs. To interrogate how the NK cells were generating their ATP, we first treated splenocytes with inhibitors to inhibit glycolysis (2-deoxyglucose, 2-DG) or inhibition of glycolysis and mitochondrial OXPHOS (oligomycin) together (2-DG+oligomycin), which provides a negative control of cells with extremely limited capacity to make ATP ( Figures 2C, D ). There was a strong reduction in puromycin incorporation in cells provided the inhibitor of glycolysis 2-DG indicating reduced ATP pools ( Figure 2C ). In the presence of 2-DG, the ability of the cell to make ATP through glycolysis or glucose-fueled mitochondrial OXPHOS is prevented. Under these conditions the remaining ATP is most likely generated through the mitochondrial oxidation of amino acids (AAO) or fatty acids (FAO). When cells are provided with puromycin plus both 2-DG and oligomycin, the processes of AAO and FAO are also blocked. Therefore, a substantial difference in the puromycin incorporation between the 2-DG condition and the 2-DG plus oligomycin condition indicates that the cells are engaging in significant levels of either AAO or FAO ( Figure 2D ). The fuel utilization in NK cells from naïve mice and those with acute FV infection was interrogated. Interestingly, glucose dependent ATP production (2-DG sensitive puromycin incorporation) was reduced in NK cells from acute infection with a concomitant increase in β-oxidation and/or amino acid oxidation in NK cells from infected animals, demonstrating the ability of NK cells to switch from glycolysis to FAs or amino acids as fuels ( Figure 2D ). Second, treatment of cells with oligomycin, which blocks the ATP production through mitochondrial OXPHOS, quantifies how much translation depends on OXPHOS. Then the difference between inhibiting OXPHOS alone and inhibition of OXPHOS plus glucose-fueled glycolysis (2-DG plus oligomycin) provides information about ATP-dependent protein synthesis that can be maintained in the absence of any mitochondrial generated ATP. This reflects the glycolytic capacity because cells increase glycolysis to the maximum possible rate in the absence of OXPHOS. Taken together, these data reveal two interesting insights. Firstly, NK cells from mice during acute infection have a significant increase in the glycolytic machinery allowing for a higher maximal glycolytic rate when required, such as if OXPHOS were inhibited in vivo due to hypoxia. Secondly, under conditions when mitochondrial OXPHOS is possible, NK cells show a switch in fuel preference away from glucose towards FA and/or amino acids during acute FV infection ( Figure 2D ).

Therefore, we next considered whether FAs are important fuels for NK cells during acute retroviral infection. Short- and medium-chain fatty acids with a chain length less than 12 carbons can freely diffuse into the mitochondrial matrix. However, the transport and subsequent oxidation of long-chain fatty acids with a chain length between 14 to 18 carbons are regulated by the carnitine palmitoyltransferase 1 (CPT-1) system (29). We treated splenocytes with low doses of etomoxir (5 μM, 40 μM), which is an irreversible inhibitor of CPT-1, and inhibited the transport of long-chain fatty acids into mitochondria. In previous studies it was demonstrated that high concentrations of etomoxir (up to 200 μM) had an influence on the proliferation of cells independent of β-oxidation (10). We have shown that the proliferation of NK cells was not altered by treating splenocytes with 5 µM or 40 µM etomoxir ( Supplementary Figure 2 ). When low-dose of etomoxir was used to inhibit CPT-1, a decrease in puromycin incorporation was observed, though not to the level seen with 2-DG plus oligomycin. These data argue that inhibition of β-oxidation diminishes ATP levels in the cells and that FAs are an important fuel, along with glucose, used by NK cells to make ATP during acute retroviral infection ( Figures 2E, F ). Interestingly, etomoxir treatment significantly decreased the puromycin expression of NK cells from FV-infected animals almost to similar levels as 2-DG or oligomycin single treatment ( Supplementary Figure 1A ). This shows that FAs are major contributors to energy generation upon acute retroviral infection. NK cells from uninfected animals only slightly decreased the puromycin expression after etomoxir treatment. Moreover, NK cells from naïve animals strongly decreased the puromycin expression after single 2-DG or oligomycin treatment, demonstrating their strong dependence on glycolysis and OXPHOS independent of β-oxidation. To proof the influence of etomoxir on the NK cells energy generation, we analyzed the influence of low-dose etomoxir on the oxygen consumption rate (OCR) ( Supplementary Figure 1B ). Similar to SCENITH data, we did not observe a reduction of OCR after low-dose etomoxir treatment in NK cells from naïve animals, but observed a significant decrease of OCR after etomoxir treatment in NK cells isolated from FV-infected animals. Taken together, these data demonstrate that NK cells increase their ability to utilize FA as source of energy during acute infection.

CD36 is expressed by a variety of myeloid immune cells, B cells as well as NK cells (17, 30, 31) and acts to facilitate transport of long-chain FAs into cells (32). Interestingly, we show here that a subset of NK cells expresses a low level of CD36 and that CD36 expression is induced during acute infection with FV ( Figures 3A, B ). Moreover, we divided NK cells in a CD36^high and CD36^low population ( Figure 3C ) and analyzed the activation, proliferation, cytotoxicity and metabolic markers ( Figure 3D ) in CD36^high and CD36^low NK cells from FV-infected mice. CD36^high NK cells strongly increased their proliferation (KI-67), which peaked at 7 dpi ( Supplementary Figure 5A ) and activation (CD69). Also, cytotoxicity markers, such as TRAIL and GzmB, were highly enriched in CD36^high NK cells. CD36^high NK cells produced more IFN-γ and increased the expression of several metabolic markers such as the metabolic master regulator cMyc, the transferrin receptor CD71, CD98, which is important for amino acid transport ( Figure 3D ). Interestingly, CD36^high NK cells had a higher puromycin expression compared to CD36^low NK cells from acute FV infection ( Supplementary Figure 5B ). To confirm that NK cells increase the uptake of FAs in line with increased expression of CD36, we analyzed the ability of NK cells to take up FAs. Therefore, we incubated NK cells with Bodipy FL C16 and analyzed the FAs uptake over time in FV infection. We found that the maximum uptake of FAs by NK cells was at 7 dpi, which decreased until 28 dpi ( Figure 3E ). Moreover, the population of activated NK cells increased the uptake of FAs ( Supplementary Figures 3A, B ). Interestingly, there was a positive correlation between the FA uptake and activation of NK cells ( Supplementary Figure 3C ). Also, CD36⁺ NK cells have an increased incorporation of a fluorescent tagged C16 FA, Bodipy FL C16, during acute retrovirus infection compared to NK cells from naïve mice ( Supplementary Figure 3D ). Next, we analyzed the mRNA isolated from NK cells of uninfected and FV-infected animals for molecules involved in FA uptake and oxidation (CD36, CPT1α, PPARα, PPARγ, FABP5, SCARB1, SCARB2) ( Figure 3F ). Because the enrichment of NK cells was approximately 77%, we analyzed the other cell subsets that were present in the isolated fraction. We found almost similar percentages of macrophages, granulocytes and dendritic cells (DCs) in cells isolated from naïve or FV-infected animals ( Supplementary Figure 4 ). We found a significant increase in the expression of FA transporter CD36 as well as 20-fold augmented CPT-1 (target for etomoxir) mRNA level in splenic NK cells isolated from acute FV-infected mice compared to NK cells from naïve mice. SCARB genes encode e.g. for scavenger receptor class B type 1 (SCARB1) or type 2 (SCARB2) and belong to the CD36-like superfamily. Proteins encoded by SCARBs are essential for cholesterol homeostasis and binding of high-density lipoprotein cholesterol. There was a 1.5-(SCARB1) and 1.4-fold (SCARB2) increase in the mRNA levels in NK cells of FV-infected mice. The transcription factor PPARα is regulated by free fatty acids and promotes uptake, use, and catabolism of FA by upregulation of genes involved in FA transport, fatty acid binding and activation, and β-oxidation (33). We found a significant increase in PPARα expression in NK cells from acute FV infection. PPARγ is a ligand-inducible transcription factor and regulates the expression of genes that are involved in lipid transport and metabolism such as CD36 (34). During acute FV infection, NK cells PPARγ mRNA level was significantly increased compared to naïve NK cells. Next, we analyzed the expression of small intracellular lipid chaperone Fatty acid-binding protein 5 (FABP5), which is important for FA uptake and transportation. In line with the above-mentioned genes, we found an upregulation of FABP5 in NK cells after FV infection. We hypothesized that the increased uptake of FAs takes place in organs in which FV highly replicates. Thus, we analyzed the uptake of FAs in the spleen (high virus replication) and blood (low virus replication) ( Supplementary Figure 3E ). Indeed, we found a decreased uptake of FAs in the blood. These data collectively show that NK cells increase molecules involved in FA uptake and catabolism during acute retrovirus infection.

Neutral lipids are important for energy storage and energy production in cells (35). We analyzed the staining of LipidTOX, which is a stain for neutral lipid droplets ( Figure 4A ). We found similar levels of LipidTOX in NK cells from naïve and FV-infected mice. Next, we were interested whether NK cells that take up FAs through CD36 accumulate FAs. Thus, we analyzed LipidTOX in CD36^high and CD36^low NK cells and found a significant increase in LipidTOX in CD36^high NK cells ( Figure 4B ). When we inhibited the uptake of FAs through CD36 by incubating the cells with sulfo-N-succinimidyl oleate (SSO), we found a significant decrease of LipidTOX staining level ( Supplementary Figure 6A ). Our hypothesis is that CD36^high NK cells increase uptake of FAs and so we wanted to exclude any differences in the overall availability of FAs in naïve mice versus FV-infected mice. For this reason, we analyzed the free and esterified FAs in the plasma of the blood. No significant differences were found in the composition of free and esterified FAs from naïve or retrovirus-infected mice ( Figure 4C ). Linoleic acid (18:2) followed by palmitic acid (16:0) and oleic acid (18:1) were present in high concentrations in the plasma of mice. Taken together, the data argues that during FV infection NK cells have an increased capacity to take up and use the FAs available in the serum.

We have found a decreased protein translation, assuming a decreased generation of ATP, after blocking the β-oxidation ( Figures 2E, F ) and augmented mRNA levels of molecules involved in β-oxidation ( Figure 3F ). Since long-chain FAs cannot freely enter the mitochondrial matrix, they depend on CPT-1 for transport. Hence, we inhibited the β-oxidation by using low concentration of etomoxir (5 µM), which stops the transport of FA into the mitochondria. Etomoxir-treated NK cells showed similar levels of activation, cytokine-production as well as GzmB expression compared to untreated NK cells ( Figures 5A–C ). Nevertheless, we hypothesized that etomoxir has a detrimental effect on the killing capacity of NK cells. Thus, we analyzed the killing capacity of NK cells treated with etomoxir in vitro ( Figures 5D–G ). We first analyzed the influence of etomoxir on target cells (YAC-1) and found no effects on the viability of tumor target cells ( Figure 5D ). Next, we analyzed the influence of etomoxir on the viability of NK cells. We did not detect a decrease in viability after etomoxir treatment compared to untreated NK cells ( Figure 5E ). Next, we analyzed the degranulation of NK cells and did not observe any effects of etomoxir treatment on NK cell degranulation ( Figure 5F ). Interestingly, CD36^high NK cells exhibited a higher degranulation compared to CD36^low NK cells upon acute infection ( Supplementary Figure 5C ). Nevertheless, we found a decreased target cell killing by NK cells treated with etomoxir. Here, NK cells from infected mice showed a more than 2-fold reduction in killing capacity. Besides treatment with etomoxir, the blockade of CD36 by SSO resulted in a significant reduction of NK cell killing ( Supplementary Figure 6B ). Collectively, these data demonstrate the importance of β-oxidation for NK cells killing abilities.

The process of killing a target cell is a multi-step process (36). First, NK cells have to migrate and recognize target cells. Second, after adhesion an immunological synapse has to be formed. Third, target cell can be killed by release of cytotoxic substances into this gap. Because we have seen defects in target cell killing after etomoxir treatment of NK cells ( Figure 5G ) but no differences in NK cells GzmB expression ( Figure 5C ), we asked if we find alterations in conjugate formation or migration of NK cells. So, we went one step back and analyzed the conjugate formation of NK cells with target cells (YAC-1). NK cells isolated from retrovirus-infected mice exhibited a significant higher conjugate formation compared to NK cells from naïve mice ( Figure 6A ), but treatment with etomoxir did not alter the conjugate formation in both analyzed experimental groups ( Figure 6B ). Then we went one further step back and analyzed the migratory capacity of NK cells. Migration of NK cells was almost zero when incubated without attractants. Thus, we used YAC-1 cells and plated them in the compartment below. We found a significant reduction in the migration of NK cells after etomoxir treatment in NK cells isolated from FV-infected mice ( Figure 6C ). We analyzed the degranulation of NK cells in the inserts (no migration) and wells (migrated) and did not detect differences between etomoxir-treated or untreated groups ( Figure 6D ). Although we found similar percentages of CD36^high and CD36^low subsets in the insert and well ( Supplementary Figure 5D ), we detected an increased degranulation of migrated CD36^high NK cells ( Supplementary Figures 5E, F ). Interestingly, the migration of CD36^high NK cells was strongly decreased after etomoxir treatment ( Supplementary Figure 5G ). Together, this data show that the conjugate formation is not impaired but the migration is diminished after inhibiting the β-oxidation in NK cells from FV infection.