The ω-3 Polyunsaturated Fatty Acid Docosahexaenoic Acid Enhances NK-Cell Antitumor Effector Functions Open Access

Adequate intake of ω-3 polyunsaturated fatty acids (PUFA), which cannot be synthesized in vivo in mammals, is critical for human health, and PUFA deficiency is closely associated with increased risk of various chronic diseases, such as cardiovascular diseases and cognitive disorders (1, 2). Due to these health benefits, the antitumor effects of ω-3 PUFAs have also been extensively studied (3). Many clinical trials have been conducted to evaluate the effectiveness of ω-3 PUFA supplementation in affecting cancer-associated symptoms, such as inflammation, cachexia, neuropathy and quality of life, in patients who suffer from weight loss, fatigue, inflammation, or postoperative complications (4). However, the detailed mechanisms underlying the benefits of ω-3 PUFA–mediated protection against cancer are still largely unknown.

Animal studies have demonstrated that ω-3 PUFAs, including docosahexaenoic acid (DHA) and eicosatetraenoic acid (EPA), exert antitumor effects by directly inducing cancer cell death and suppressing metastasis, either in combination with conventional anticancer therapies or alone (5). For example, DHA/EPA-enriched phosphatidylcholine suppresses tumor growth and metastasis by accelerating cancer cell apoptosis and increasing extracellular matrix (ECM) degradation (6). In addition, a recent study reported that excess ω-3 PUFAs undergo peroxidation and subsequently induce ferroptosis in acidic cancer cells (7). However, most of the previous studies on the antitumor effects of DHA have used immunodeficient mice, such as BALB/c-nu mice.

An appropriate immune response is critical for the repression of tumor development and progression. T cells, B cells, and natural killer (NK) cells are the main effectors of tumor rejection and act by secreting cytokines or antibodies as well as performing direct tumor cell lysis; thus, modulating the numbers and functions of these cells can contribute to cancer treatment (8, 9). Although alterations in the immune system caused by ω-3 PUFA supplementation have been described for many years, the effects of ω-3 PUFAs on the immune response are unclear, suggesting that many studies on ω-3 PUFAs have produced inconsistent conclusions on the effects of ω-3 PUFAs on the immune system in different disease models (10). These results suggest that ω-3 PUFAs play multifaceted roles in different immune cells and diseases. Thus, whether the antitumor activity of ω-3 PUFAs also results from their regulation of the immune response warrants further in-depth study. Collectively, two issues remain to be resolved: one is DHA can have an immunomodulatory role during cancer development and the other is whether the preventive effects of DHA in cancer are dependent on immunomodulation.

To date, reports of the effects of ω-3 PUFAs on NK-cell number and function are contradictory. Dietary supplementation with EPA-rich oil (4.05 g/day) does not alter the number of circulating NK cells in humans (11), while a more recent study showed that dietary supplementation with ω-3 PUFAs (1 g/day) lowered the percentage of NK cells in the peripheral blood (12). In mouse models, ω-3 PUFA–rich fish oil was found to impair spleen but not lung NK-cell activity in mice infected with influenza (13), whereas other studies have shown that dietary DHA promotes the activation of splenic NK cells (5, 14). Recently, we found that DHA feeding resulted in a reduced frequency of NK cells in the blood of mice but an enhanced capacity for IFNγ production by NK cells and enhanced NK-cell cytotoxicity during murine cytomegalovirus (MCMV) infection (15). NK cell–derived IFNγ is a pivotal factor that sustains tumor dormancy, thereby preventing tumor metastasis to the liver and lung (16, 17). These findings prompted us to explore whether DHA affects the antitumor effects of NK cells in vivo.

In this study, we showed that DHA supplementation resulted in a decrease in the number of lung metastatic colonies formed by B16F10 melanoma cells and that this was mainly dependent on NK cells. Moreover, DHA enhanced NK-cell effector functions by improving metabolic status and mitochondrial activity via proliferator-activated receptor gamma coactivator-1α (PGC-1α).

C57BL/6J-Rag2^em3Lutzy/J (Rag2^−/−) and B6N.129(FVB)-Ppargc1a (PGC-1α)^tm2.1Brsp/J (PGC1α^fl/fl) mice were purchased from The Jackson Laboratory (Sacramento); Ncr1-Cre (C57BL/6-Ncr1tm1(iCre)/Bcgen) mice were a kind gift from Beijing Biocytogen (Beijing, China); and 8-week-old wild-type (WT) C57BL/6J mice (approximately 22–25 g) were purchased from Hunan Sja Laboratory Animal Co., Ltd. The mice were divided into two groups as described previously (15). The DHA supplementation group was fed a diet containing DHA (2.48% DHA, D201124, Dyets Biotechnology Ltd), and the control group was fed a control diet [D200208, Dyets Biotechnology (Wu Xi) Ltd.; Supplementary Table S1]. DHA oil was purchased from MedChemExpress or kindly provided by Solutex GC, SL. PGC1α^fl/fl mice were crossed with Ncr1-Cre mice to generate NK cell–specific deletion of PGC-1α (PGC1α^fl/fl/Ncr1-Cre⁺, referred to as PGC1α^ΔNK mice), and PGC1α^fl/fl mice were used as controls for PGC1α^ΔNK mice. Both the control and PGC1α^ΔNK mice were fed a diet containing DHA. All mice were housed under specific pathogen-free conditions at the Hunan Children's Hospital Animal Facility on a 12-hour light/dark schedule with free access to food and water. All animal procedures and protocols were approved by the Animal Ethics Committee of Hunan Children's Hospital and followed the guidelines of the Institutional Animal Care and Use Committees of Hunan Children's Hospital (Changsha, Hunan, China). The ethics committee–approval code was HCHDWLL-2021-07. All mice were sacrificed by cervical dislocation under anesthesia with 2% pentobarbital sodium.

B16F10 melanoma cells and YAC-1 cells were kind gifts from Ying Wan at Third Military Medical University (Chongqing, China) received in 2019. All cell lines were negative for mycoplasma after testing with a Mycoplasma PCR Detection Kit (Sigma-Aldrich). None of the cell lines used in this paper are listed in the database of commonly misidentified cell lines maintained by ICLAC. None of the cell lines were reauthenticated within the past year. B16F10 cells were maintained in DMEM (Gibco) containing 10% heat-inactivated fetal bovine serum (FBS; Gibco) and 1% penicillin–streptomycin solution (HyClone). YAC-1 cells were cultured in RPMI-1640 medium (Gibco) containing 10% FBS and 1% penicillin–streptomycin solution. All the cells were cultured at 37°C with 5% CO₂ for two or three generations prior to their utilization in experiments.

A mouse B16F10 melanoma tumor model was generated as described previously (17). WT, Rag2^−/−, or PGC1α^ΔNK mice were fed with DHA-enriched diet-fed or control diet for 3 weeks and then i.v. injected with 0.5×10⁵ B16F10 tumor cells. For the NK cell–depletion study, DHA-enriched diet–fed WT mice were administered an NK cell–specific neutralizing antibody (anti-NK1.1, PK136; Bio X Cell) to deplete NK cells or an isotype-matched mouse IgG2α monoclonal antibody control (Bio X Cell) by i.v. injection (200 μg per mouse) on days −4 and −1 before i.v. injection of B16F10 melanoma cells. After i.v. injection of 5 × 10⁵ B16F10 melanoma cells, the appropriate antibody (250 μg per mouse) was i.p. injected on days 7 and 15. During each experiment, 4 to 5 additional WT mice were i.v. injected with B16F10 tumor cells and utilized to monitor the progression of tumor metastasis every two days, starting from day 10 following tumor injection.

Single-cell suspensions were prepared from the bone marrow (BM; cells were derived from the femurs and tibias of two hind legs), spleen, peripheral lymph nodes (pLN), liver, and lung as previously described (15, 18). The BM, spleen, pLNs, thymus, lung, and liver tissues were ground and passed through a 70-μm stainless steel mesh and pellets were collected after centrifugation. The BM and spleen mononuclear cells were separated from the pellets by lysing erythrocytes with Red Blood Cell Lysis Buffer (Thermo Fisher Scientific). The obtained lung and liver cells were resuspended in 20% and 40% Percoll (GE Healthcare) in RPMI-1640 medium supplemented with 5% FBS, respectively, and then centrifuged (2,000 rpm, 4°C, 5 minutes). The cells isolated from each tissue sample were counted with an automated cell counter (Countstar IC1000).

Eight-week-old male C57BL/6J mice were sacrificed by cervical dislocation. Single-cell suspensions of the spleen were prepared as described above. NK cells were negatively selected with a mouse NK-cell isolation kit (MojoSort Mouse NK-Cell Isolation Kit, BioLegend) and further purified through sorting with a FACSAria III cell sorter (BD Biosciences). The antibodies used for the sorting were anti-CD3 (17A2, BioLegend) and anti-NK1.1 (PK136, BioLegend). NK cells were sorted as CD3⁻ and NK1.1⁺ and the purity of the NK cells was ≥ 98.0%.

Purified mouse NK cells were cultured with SCGM (CellGenix) supplemented with 10% FBS in a 96-well U-bottom plate for future experiments. For the flow cytometry analysis of IFNγ, CD107a, MitoTracker, TMRM, CD71, CD98, and 2-NBDG, metabolic flux analysis, immunoblotting and qRT-PCR of target genes, purified mouse NK cells were stimulated with IL15 (50 ng/mL; BioLegend) or IL2 (1,000 IU/mL; BioLegend) plus IL15 (10 ng/mL) in the presence of DHA (5 μmol/L, 10 μmol/L, or 20 μmol/L) for 10 days, unless otherwise indicated. To explore the role of the PGC-1α pathway in mediating DHA's effect on NK cells, DHA (10 μmol/L), DHA (10 μmol/L) plus SR-18292 (20 μmol/L; Selleck Chemicals LLC), SR-18292 (20 μmol/L) alone or an equal volume of DMSO was added to the cultures in the presence of IL15 (50 ng/mL) for 48 hours or 10 days.

All antibodies used for staining and subsequent flow-cytometric analysis are listed in Supplementary Table S2. Standard protocols were followed for flow cytometry, as previously described (15, 19). All flow cytometry experiments were carried out on a BD LSRFortessa cell analyzer, and the data were analyzed with FlowJo software (V10.8.1). Briefly, to detect surface markers, cells were stained with antibodies in staining buffer [phosphate-buffered saline (PBS) containing 2% mouse serum (Thermo Fisher Scientific), 2% horse serum (Thermo Fisher Scientific), and anti-CD16/CD32 blocking antibodies (2.4G2, #553141, BD Biosciences)] in the dark for 15 minutes at room temperature.

For intracellular IFNγ staining, cells were stimulated with eBioscience Cell Stimulation Cocktail (81 nmol/L PMA and 1.34 nmol/L ionomycin, eBioscience) plus BD GolgiPlug and GolgiStop protein transport inhibitor (BD Biosciences) for 4 hours or were cocultured with YAC-1 cells at a 5:1 ratio in the presence of PMA/ionomycin plus BD GolgiPlug and the GolgiStop protein transport inhibitor for 6 hours. Then, the cells were stained with surface markers, followed by fixation and permeabilization using a fixation/permeabilization solution kit (BD Biosciences).

For other intracellular proteins, including PGC-1α, granzyme B, and perforin, cells were stained with surface marker–specific antibodies, permeabilized with reagents from the Foxp3/Transcription Factor Staining Buffer Set Kit (eBioscience), and then stained with the relevant primary antibodies or isotype-matched control antibodies.

To evaluate CD107a expression on splenocytes, lung lymphocytes and purified mouse NK cells cultured in vitro, the cells under investigation were cocultured with YAC-1 cells at a 5:1 (for splenocytes or lung lymphocytes) or 3:1 (for purified NK cells) ratio in the presence of anti-CD107a (BD Biosciences) and protein transport inhibitors (GolgiPlug and GolgiStop). After 6 hours of coculture at 37°C, the percentages of CD107a⁺ NK cells were determined via flow cytometry. To determine the glucose uptake capacity of NK cells, cells were cultured in prewarmed (37°C) RPMI-1640 medium (Life Technologies) containing 100 μmol/L 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino)-2-deoxyglucose (2-NBDG, a fluorescent glucose analog; Invitrogen) for 10 minutes at 37°C in the dark.

To determine mitochondrial activity, splenic cells were cultured (37°C, 30 minutes) in prewarmed (37°C) RPMI-1640 medium containing 20 nmol/L MitoTracker Green FM (Invitrogen) or tetramethylrhodamine methyl ester (TMRM; Invitrogen) in the dark. Then, the cells were washed and stained for surface markers.

For lung lymphocyte isolation, fresh mouse lung tissues were cut into pieces and then digested for 45 minutes at 37°C in Hank's solution containing 10% FBS, 1 mg/mL collagenase I (Sigma-Aldrich), 1 mg/mL collagenase II (Gibco), and 25 μg/mL DNase I (Sigma-Aldrich). After digestion, the cells were resuspended in 20% Percoll in PBS and the pellets were collected after centrifugation (450 × g, room temperature, 10 minutes). Finally, the cells were suspended at a density of 1,000 cells/μL in PBS, and about 10,000 cells were loaded onto a Chromium Controller instrument after completion of a cell suspension preparation procedure using GemCode Gel Bead and Chip (10× Genomics) following the manufacturer's recommendations. Libraries were prepared using 10× Genomics Library Kits and sequenced on an Illumina NovaSeq 6000 to generate 150-bp paired-end reads according to the manufacturer's instructions (Berry Genomics).

The sequencing data were demultiplexed, aligned against the mouse reference genome mm10, and quantified using CellRanger (version 6.0.1, 10× Genomics). Quality control filtering, variable gene selection, dimensionality reduction, and cell clustering were performed using the R package Seurat (version 4.1.0; ref. 20). For in-depth analysis of cell populations, only cells with <7.5% mitochondrial genes were retained. To avoid batch effects, the “FindIntegrationAnchors” function was applied to integrate the samples. Normalization, scaling, and variable gene selection were performed using the “NormalizedData,” “FindVariableFeatures,” and “ScaleData” functions in Seurat with standard settings (21). In addition, the R package “scDblFinder” was used to identify doublets. After principal component analysis, dimensionality reduction was performed using the uniform manifold approximation and projection (UMAP) algorithm. Clusters were then identified using the “FindNeighbors” and “FindClusters” functions in Seurat, with resolution parameters of 0.3. Next, differentially expressed genes (DEG) were identified using the Wilcoxon test implemented in the “FindAllMarkers” function. Genes with an average fold difference of at least 0.25 (log scale) between the cells in the tested cluster and the remaining cells and an adjusted P < 0.05 were considered significant. Gene set variation analysis (GSVA), a nonparametric and unsupervised software algorithm, was used to perform the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis with the R package GSVA (22).

Mitochondrial stress test kits (Agilent) were used for metabolic flux analysis. Purified splenic NK cells (purity ≥ 98.0%) from WT mice were treated with DMSO or 10 μmol/L DHA in the presence of 1,000 IU IL2 plus 10 ng/mL IL15 or 50 ng/mL IL15 alone for 10 days. Then, a total of 4 × 10⁵ purified NK cells were plated per well on polylysine-coated Seahorse plates in XF media [25 mmol/L glucose (Agilent), 2 mmol/L glutamine (Agilent), and 1 mmol/L pyruvate (Agilent)]. The basal oxygen consumption rate (OCR) in NK cells was measured three times with an XF-96 Extracellular Flux Analyzer (Agilent Technologies, Seahorse Bioscience). Afterward, 2 mmol/L oligomycin (Agilent), 1.5 mmol/L FCCP (Agilent), and 1 mmol/L rotenone (Agilent) with 1 mmol/L antimycin (Agilent) were sequentially added. The basal OCR and maximum respiration were calculated according to the manufacturers’ instructions. The OCR was normalized to the cell protein concentration in each experiment.

Splenic NK cells sorted from WT mice were initially treated with DMSO or 10 μmol/L DHA for 48 hours in the presence of 1,000 IU IL2 plus 10 ng/mL IL15 or 50 ng/mL IL15 alone. Total RNA was extracted using TRIzol (Invitrogen) and reverse transcribed into cDNA using the EvoM-MLV Reverse transcription kit [Accurate Biotechnology (Hunan) Co., Ltd] following standard protocols described previously (23). qRT-PCR was performed with 2 × SYBR Green pro taq HS Premix [Accurate Biotechnology (Hunan) Co., Ltd.] and run on a Roche LightCycler 480 II. The PCR reaction mixture was adjusted to 10 μL with 5 μL 2 × SYBR Green pro taq HS Premix, 0.2 μmol/L primers, 1 μL cDNA templates (about 4 ng cDNA), and 3.6 μL RNase free water. The reactions were run as follows: 3 minutes at 95°C, 40 cycles consisting of 10 seconds at 95°C, and 30 seconds at 60°C. Three replicates were run for each sample. The cycle threshold (CT) values of a target gene and internal control (Gapdh) were obtained from each sample. The relative mRNA expression levels of each gene were calculated by normalizing the relative cycle threshold value to the control group after normalization to the internal control (^∆∆CT method). The cycle threshold (Ct) values were normalized to those for internal controls. The primer sequences for qRT-PCR are shown in Supplementary Table S3.

To detect DHA uptake in immune cell subsets and tumor cells, DHA was labeled with fluorescein isothiocyanate (FITC; MedChemExpress; referred to as FITC-DHA) by MedChemExpress. After labeling with a CellTrace Far Red Cell Proliferation Kit (Thermo Fisher Scientific), 2× 10⁵ B16F10 cells were cocultured with 2 × 10⁶ splenocytes from WT C57BL/6J mice in the presence of 10 μmol/L FITC-DHA for 6 or 24 hours. DHA uptake by CD19⁺ B cells (CD3⁻NK1.1⁻CD19⁺CD4⁻CD8⁻), total CD3⁺ T cells (CD3⁺NK1.1⁻CD19⁻), CD4⁺ T cells (CD3⁺NK1.1⁻CD19⁻CD4⁺CD8⁻), CD8⁺ T cells (CD3⁺NK1.1⁻CD19⁻CD4⁻CD8⁺), NK cells (CD3⁻NK1.1⁺CD19⁻CD4⁻CD8⁻), and B16F10 cells was determined by flow cytometry.

Immunoblotting analysis was performed as described previously (19). After treatment with DMSO (as a control) or DHA in the presence of 50 ng/mL IL15 for 48 hours, purified splenic NK cells (purity≥ 98%) were directly lysed with 2× Laemmli buffer (Bio-Rad) supplemented with 2.5% β-mercaptoethanol and boiled for 10 minutes. The proteins were separated by SDS‒PAGE, transferred onto PVDF membranes (GE Healthcare), blocked with 5% TBST-diluted skim milk for 1 hour at room temperature and incubated with primary antibodies in 5% BSA at 4°C for approximately 12 to 18 hours, followed by incubation and detection with Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific, 32209) in the Invitrogen iBright FL1500 imaging system (Thermo Fisher Scientific). The primary antibodies used were specific for phospho-Akt (Ser473) (p-AKT^Ser473), p-S6 (Ser235/236) (p-S6 ^Ser235/236), c-Myc and β-Actin. Anti-rabbit HRP (ZSGB-BIO, ZB-2301) and anti-mouse HRP (ZSGB-BIO, ZB-2301) were used as secondary antibodies.

The Schmiedel data set from the human protein atlas website (https://www.proteinatlas.org/) was used to analyze mRNA expression levels of FABP, FAT, and FATP family members among NK cells, CD4⁺ T cells, CD8⁺ T cells, and B cells. The transcriptomic data were generated from 13 immune cell types isolated from 91 healthy subjects and reported as transcripts per kilobase of exon model per million mapped reads (TPM; ref. 24). The analysis was done with GraphPad Prism (GraphPad Software 8.0). Two-tailed unpaired Student t test was used to compare the differences between the two groups. Data are represented as median ± interquartile range, as indicated.

Statistical analysis was carried out using SPSS 23.0. For in vivo experiments, the sample size was 3 to 15 per group and two or four independent experiments were performed for each experiment. If the data were normally distributed, an unpaired two-tailed Student t test was used to analyze the differences between the two independent groups. Otherwise, a two-tailed Mann–Whitney U test was used to analyze the differences. For in vitro experiments, the sample size was 4 to 8 per group, two or four independent experiments were performed for each experiment and paired t tests were used for analysis. The data in this study are expressed as the median and interquartile range. Differences with a P ≤ 0.05 were considered significant. All the graphs were generated with GraphPad Prism 8.0 (GraphPad Software, Inc.).

The scRNA-seq data reported in this paper have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: CRA015282) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa. All other data are available in the article and its supplementary data files or from the corresponding author upon reasonable request.

Due to the disagreement about the effects of DHA on the immune system, we first examined the effect of DHA on the immune system in a steady state. WT C57BL/6J mice were fed with a control or special DHA-enriched diet for 3 weeks, after which the frequencies and numbers of T cells, B cells, and NK cells were assessed by flow cytometry. The data showed that the frequencies and numbers of CD3⁺ T cells (gated as CD45⁺CD3⁺CD19⁻ cells), including CD4⁺ and CD8⁺ T-cell subsets and B cells (gated as CD45⁺CD3⁻CD19⁺ cells) in the BM, spleen, pLN, liver and lung tissues were not affected by DHA after 3 weeks of DHA-enriched diet feeding (Fig. 1A and B; Supplementary Fig. S1A). However, the frequencies of NK cells (gated as CD45⁺CD3⁻CD19⁻NK1.1⁺NKp46⁺ cells) in DHA-treated mice were significantly decreased in the BM, spleen, pLNs, and lung but not in the liver tissues. The total number of NK cells was significantly reduced in the BM. Three weeks of DHA feeding resulted in slightly increased proportions of naïve CD62L⁺CD44⁻CD4⁺ T and CD8⁺ T cells and immature NK cells (CD27⁺CD11b⁻) in some organs (Fig. 1C; Supplementary Fig. S1B and S1C and S1D).

As the time required to produce a mature lymphocyte starting from hematopoietic stem cells is 6 to 8 weeks (25), we also fed mice control or DHA-enriched diets for 8 weeks. The data showed that the total pool of NK cells was slightly reduced after 8 weeks of DHA-enriched diet feeding, especially in the lung, similar to that observed after 3 weeks of DHA-enriched diet feeding (Supplementary Fig. S2A and S2B). The frequency and number of B cells were significantly lower in the BM and spleen. Although the frequencies of total CD3⁺ T cells and CD4⁺ and CD8⁺ T-cell subsets in some organs were significantly greater after DHA treatment, their total numbers did not change except in the liver. The changes in the frequencies of T cells and T-cell subsets might be due to the reduced pools of NK cells and B cells after DHA treatment. After 8 weeks of DHA feeding, the changes in naïve T cells and NK-cell maturation were similar to those observed after 3 weeks of DHA feeding (Supplementary Fig. S2C and S2D).

The expression of IFNγ in NK cells was unchanged after culture in the presence of only PMA/ionomycin (Fig. 1D) but significantly increased after coculture with YAC-1 cells in the presence of PMA/ionomycin for 6 hours ex vivo (Fig. 1E). The expression of the cytotoxic molecule CD107a in NK cells was comparable between control and DHA-fed mice when cocultured with YAC-1 cells for 6 hours ex vivo (Fig. 1F). The expression of perforin and granzyme B in NK cells was unchanged in the spleen of DHA-enriched diet–fed mice compared with those of control diet–fed mice (Fig. 1G).

Overall, these results suggested that DHA supplementation resulted in a slight decrease in the pools of NK cells and B cells, an increase in naïve T and immature NK cells, and a slight increase in IFNγ production when cocultured with tumor cells ex vivo.

One previous study using immunocompetent WT mice demonstrated that DHA supplementation immediately after Lewis lung cancer cell implantation inhibited tumor growth and lung colonization in vivo. The authors reported that the mechanism involved the direct inhibition of tumor cell survival and ECM degradation (6, 7); however, changes in immune cell subsets after tumor cell implantation were not reported. To investigate the effects of preventive administration of DHA-enriched diet on the immune response in the context of tumor burden, we first treated WT C57BL/6J mice with control or DHA-enriched diets for 3 weeks and then challenged the mice with i.v. injection of B16F10 melanoma cells (Fig. 2A). Three weeks after the i.v. injection of B16F10 melanoma cells, we found that the mice treated with 3 weeks of DHA supplementation had significantly fewer lung tumor nodules than did the control diet–treated mice (Fig. 2B). To assess whether continuous DHA feeding is obligatory for the antitumor effect of DHA, we tested withdrawing DHA-enriched diet at the time of the i.v. injection of B16F10 melanoma cells. The results showed that there was a decreasing trend but no significant difference in lung tumor cell colonization in DHA-enriched diet–fed mice compared with control diet–fed mice (Supplementary Fig. S3A and S3B). This finding indicates that the direct effects of DHA on tumor cells likely contribute to its antitumor effect.

Then, we determined the percentages and numbers of T cells, B cells and NK cells in various tissues of the mice after tumor cell injection. Our data showed that the proportions and numbers of NK cells, CD3⁺ T cells, CD4⁺ T cells, CD8⁺ T cells, and B cells in various organs were comparable between control diet– and DHA-enriched diet–fed mice after tumor cell injection (Fig. 2C; Supplementary Fig. S4A). With respect to NK-cell maturation, although the number of cells in the mature NK-cell subset (CD27⁻CD11b⁺) decreased in the spleen and increased in the liver in DHA-enriched diet–fed mice, no difference was found in the lung between control diet– and DHA-enriched diet–fed mice (Fig. 2D). The percentages of naïve CD62L⁺CD44⁻ cells among both CD4⁺ and CD8⁺ T cells were comparable in both the spleen and lung tissues between control diet– and DHA-enriched diet–fed mice after tumor cell injection (Supplementary Fig. S4B and S4C).

As NK cell–derived IFNγ plays critical roles in controlling tumor growth (16), we also determined the capacity of NK cells to stimulate IFNγ production in both the spleen and lungs after tumor cell injection. The data showed that the ratio of IFNγ⁺ NK cells and geometric mean fluorescence intensity (gMFI) of IFNγ in total NK cells from either the spleen or the lungs of DHA-enriched diet–fed mice increased after 6 hours of ex vivo PMA/ionomycin stimulation, although the gMFIs of IFNγ in IFNγ⁺ NK cells were not changed (Fig. 2E; Supplementary Fig. S4D). With respect to NK-cell cytotoxicity, the ratio of CD107a⁺ NK cells was also significantly greater in the lung tissues but not in the spleens of mice fed with DHA when cocultured with YAC-1 tumor cells for 6 hours (Fig. 2F). When ex vivo cocultured with YAC-1 in the presence of PMA/ionomycin, the ratio of IFNγ⁺ NK cells and the gMFIs of IFNγ in total NK cells from the spleens of DHA-enriched diet–fed mice were also greater than those from control diet–fed mice, although the gMFIs of IFNγ in IFNγ⁺ NK cells were also not changed (Fig. 2G; Supplementary Fig. S4E). These findings agree with our previous finding that DHA-enriched diet–fed mice exhibit an enhanced capacity for IFNγ production by NK cells during MCMV infection (15). The gMFIs of IFNγ in CD4⁺ T cells and CD8⁺ T cells were not significantly different (Supplementary Fig. S4F and S4G).

Compared with those of control diet–fed mice, the lung tissues of DHA-enriched diet–fed mice presented upregulated protein levels of granzyme B, indicated by the gMFI, in NK cells, although the change was not significant (Fig. 2H). The protein level of perforin in NK cells in both the spleen and the lungs was comparable between control diet– and DHA-enriched diet–fed mice (Fig. 2H).

Overall, after DHA-enriched–diet feeding, NK-cell numbers in various tissues were reduced under steady-state conditions but were comparable and there were no obvious effects on T cells or B cells after tumor challenge, whereas IFNγ and CD107a production by NK cells was significantly increased in the tumor microenvironment (TME) of lung tissue–bearing metastatic melanoma mice. These findings suggest that NK cells play critical roles in the antitumor effects of DHA. However, due to the reduced antitumor effect observed when withdrawing the DHA-enriched diet at the time of tumor cell injection, it is apparent that there are also direct effects of DHA on tumor cells that contribute to its antitumor effects.

To comprehensively understand the immunomodulatory effect of DHA supplementation after tumor implantation, we next performed scRNA-seq analysis of lung CD45⁺ cells from control or DHA-enriched diet–fed mice that were challenged with an i.v. injection of B16F10 melanoma cells using a 10× Genomics platform. A total of 22,380 cells (control: 11,019 cells; DHA: 11,361 cells) were included in the analysis after initial quality control checks. Subsequently, these cells were divided into 13 subsets and annotated clusters based on the expression of known markers, and the DEGs across these subsets between the control and DHA groups were also calculated (Fig. 3A and B; Supplementary Tables S4 and S5; ref. 26). The relative proportions of most of the T-cell, B-cell, and NK-cell subsets among the CD45⁺ cells in the lungs were comparable between the control diet– and DHA-enriched diet–fed mice, which was consistent with the flow cytometry results (Fig. 3C).

To investigate the signaling pathways involved in the immunomodulatory effect of DHA in each subset, we performed GSVA. After evaluating the activities of 186 signaling pathways in the KEGG database and then comparing the differences in all 13 subsets between control diet– and DHA-enriched diet–fed mice, only 6 subsets, NK cells, naïve CD4⁺ T cells, naïve CD8⁺ T cells, activated CD8⁺ T cells, B cells, and neutrophils, exhibited differences in pathways between the control diet– and DHA-enriched diet–fed mice (Fig. 3D). Compared with those in control diet–fed mice, most KEGG pathways related to immune cell activation or effector functions were downregulated in B cells, naïve CD4⁺ T cells, naïve CD8⁺ T cells, activated CD8⁺ T cells and neutrophils from DHA-enriched diet–fed mice. It was observed that ribosome, proteasome, and protein export pathway activity was upregulated in B cells, naïve CD4⁺ T cells, naïve CD8⁺ T cells, and activated CD8⁺ T cells. GSVA revealed that the KEGG pathway related to NK cell–mediated cytotoxicity was the most significant pathway among the upregulated pathways; and Alzheimer's disease and Huntington's disease pathways that share a substantial number of gene sets with the oxidative phosphorylation (OXPHOS) pathway were also significantly upregulated in NK cells from DHA-enriched diet–fed mice (27). The Notch signaling pathway was the only downregulated KEGG pathway in DHA-enriched diet–fed mice (Fig. 3D).

We further subclustered NK cells into 4 subsets based on cluster-specific marker genes, NK1, NK2, NK3 and ILC1s (Supplementary Fig. S5A and Supplementary Table S6). NK1, NK2, and NK3 cells highly expressed Eomes and Tbx21 (which encodes the T-bet protein), while ILC1s did not express Eomes; moreover, both Tbx21 and Il7r (which encodes the CD127 protein) were expressed. NK3 cells expressed relatively higher levels of the late mature NK-cell markers Itgam (encoding the CD11b protein) and Klrg1, as well as the NK-cell cytotoxicity–related genes Ifng, Gzmb, Gzma, and Prf1 (Supplementary Fig. S5B). The percentage of NK3 cells among NK cells was dramatically greater in the lungs from DHA-enriched diet–fed mice than in those from control diet–fed mice (Supplementary Fig. S5C). Moreover, the expression levels of Prf1, Gzma, Gzmb and Ifng were relatively greater in the lungs of the DHA-enriched diet–fed mice bearing tumor cells than in those of the control diet–fed mice (Supplementary Fig. S5D and Supplementary Table S7).

Mitochondria, the essential hubs of metabolic activity, are critical for OXPHOS (28). In general, healthy mitochondria generate a proper membrane potential for the movement of substrates from the cytosol to the mitochondrial matrix for OXPHOS (29). NK-cell mitochondrial activity is impaired in the TME (30). Therefore, we next determined the overall mitochondrial content and mitochondrial membrane potential of lung NK cells 3 weeks after B16F10 melanoma cell injection. We found that lung NK cells from DHA-enriched diet–fed mice with B16F10 melanoma lung tumor growth had a significantly increased overall mitochondrial content, as indicated by flow-cytometric labeling with MitoTracker and a significantly increased mitochondrial membrane potential, as indicated by flow-cytometric labeling with TMRM (Fig. 3E and F). Consistent with this, 6 DEGs are related to mitochondrial function in NK cells between DHA-enriched diet–fed and control diet–fed mice, including Ndufa13, Cox8a, Ndufs4, Cox6c, Tomm7, and Selenow (Supplementary Table S5).

The above findings indicating that DHA treatment specifically enhanced NK-cell effector functions and mitochondrial activity prompted us to explore whether there was a discrepancy in DHA absorption between NK cells and other cells. As such, splenocytes from eight-week-old male WT C57BL/6J mice were cocultured with B16F10 cells for 6 or 24 hours in the presence of FITC-labeled DHA. The data showed that the ratio of FITC-DHA⁺ among all cells was much higher in both B16F10 and NK cells, followed by B cells, CD4⁺ T cells, and CD8⁺ T cells after 6-hour coculture (Supplementary Fig. S6A). After 24-hour coculture, the gMFIs of FITC-labeled DHA were highest in NK cells compared with B cells, T cells, and B16F10 cells, although the ratios of FITC-DHA showed no difference among different cell subsets (Supplementary Fig. S6B and S6C).

Because DHA is transported by a number of membrane-associated and cytoplasmic proteins, including plasma membrane fatty acid–binding protein (FABP), fatty acid translocase (FAT), and fatty acid–transport protein (FATP; refs. 31, 32), we next checked the mRNA levels of FABP, FAT, and FATP family members among NK cells, CD4⁺ T cells, CD8⁺ T cells, and B cells based on the Schmiedel data set from the human protein atlas website (https://www.proteinatlas.org/). As shown, compared with T cells and B cells, the mRNA levels of SLC27A3, FABP3, and SLC27A1 were much higher in NK cells, which might be responsible for the favorable absorption of DHA by NK cells. The expression of other molecules, such as FABP1-2, FABP4-12, SLC27A2, and FAT, was very low and undetectable among NK cells, CD4⁺ T cells, CD8⁺ T cells, and B cells (Supplementary Fig. S6D).

Collectively, our data suggested that the DHA-enriched diet specifically enhanced NK-cell mitochondrial OXPHOS activity in the TME in lung tissues from tumor-bearing mice and that NK cells absorbed DHA relatively more than other cells did.

To further explore the essential role of NK cells in mediating the antitumor effect of DHA, we first used Rag2^−/− mice, which are deficient in functional T and B cells, to establish a B16F10 melanoma lung tumor growth model after 3 weeks of DHA-enriched or control diet feeding. Our data showed that like the WT mice, compared with the control diet, DHA supplementation inhibited lung colonization by the tumor cells (Fig. 4A). Consistent with the findings in WT mice, the frequency and gMFI of IFNγ in NK cells were significantly greater than those in control mice, and the gMFIs of perforin and granzyme B in NK cells were not different between DHA-enriched diet–fed Rag2^−/− mice and control diet–fed mice in the context of tumor burden (Fig. 4B–D).

To further confirm the essential role of NK cells in mediating the antitumor effect of DHA, we depleted NK cells in WT mice by injecting PK136 antibodies prior to B16F10 melanoma cell injection (Fig. 4E). Flow-cytometric analysis confirmed that NK cells were completely depleted in various tissues (Supplementary Fig. S7). Compared with DHA-enriched diet–fed mice treated with isotype control IgG antibody injection, DHA-enriched diet–fed mice treated with PK136 antibody injection had significantly greater numbers of metastatic nodules in lung tissue and these numbers were even comparable with those of control diet–fed mice (Fig. 4F).

All these in vivo data demonstrated that enhanced NK-cell effector functions are essential for mediating the suppressive effect of DHA supplementation on B16F10 melanoma lung tumor growth.

Our above-described in vivo and scRNA-seq data indicated that enhanced OXPHOS and NK cell–mediated cytotoxicity played critical roles in mediating the antitumor effect of DHA. To explore whether DHA directly affects NK-cell IFNγ production and direct cytotoxicity in vitro, we treated sorted splenic NK cells (purity ≥ 98%) with DMSO (as a control) or different concentrations of DHA in the presence of 1,000 IU IL2 plus 10 ng/mL IL15 or 50 ng/mL IL15 alone, because IL2 and IL15 are important cytokines for NK-cell proliferation, activation, survival and homeostasis (33–35). The data showed that compared with DMSO, the addition of DHA increased IFNγ production in a concentration-dependent manner (Fig. 5A). When cocultured with target cells or YAC-1 cells, the percentage of cells expressing CD107a was greater in NK cells treated with DHA than in NK cells treated with cytokines alone (Fig. 5B).

We measured the mitochondrial mass and membrane potential and found that DHA significantly enhanced the mitochondrial content (indicated by MitoTracker; Fig. 5C) and slightly but not significantly increased the mitochondrial membrane potential (indicated by TMRM; Fig. 5D) in the presence of IL15 alone, consistent with the above in vivo data. The ratio of TMRM/MitoTracker was slightly reduced in DHA-treated NK cells in the presence of 1,000 IU IL2 plus 10 ng/mL IL15 or 50 ng/mL IL15 for 10 days (Fig. 5E). In addition, the expression levels of the transferrin receptor CD71 and the amino acid transporter CD98, both of which control cellular access to nutrients (27), were also determined. The protein level of CD71 increased in DHA-treated NK cells in the presence of IL2 plus IL15 or IL15 alone (Fig. 5F), whereas the protein level of CD98 increased on the surface of DHA-treated NK cells only in the presence of IL2 plus IL15, not in the presence of IL15 alone (Fig. 5G). Glucose uptake, which was indicated by the fluorescent glucose analogue 2-NBDG (36), was also increased in DHA-treated NK cells in the presence of IL2 plus IL15 or IL15 alone (Supplementary Fig. S8A). Then, we determined the mitochondrial OXPHOS activity by metabolic flux analysis and found that both the basal OCR and maximum respiration were significantly greater in DHA-treated NK cells than in control NK cells, irrespective of the presence of IL2 plus IL15 or IL15 alone (Fig. 5H). The enhanced OXPHOS activity by DHA treatment may explain why there is no significant change of TMRM after DHA treatment, as OXPHOS contributes to only a part of the whole mitochondrial membrane potential (37).

To detect the mRNA expression levels of key genes related to OXPHOS pathways, purified splenic NK cells sorted (purity ≥ 98.0%) from WT mice were treated with DMSO or 10 μmol/L DHA for 48 hours in the presence of 1000 IU IL2 plus 10 ng/mL IL15 or 50 ng/mL IL15 alone. The data showed that the RNA expression of mt-cytb, mt-Co1, mt-Co2, mt-Co3, mt-Atp6, and mt-Nd1, which are involved in OXPHOS (38, 39), was significantly increased in the DHA-treated NK cells incubated with IL2 plus IL15 or IL15 alone, except that the expression of mt-Co2 showed an increasing but nonsignificant trend in the DHA-treated NK cells incubated with IL15 alone (Fig. 5I). These results further suggested that DHA promoted NK-cell mitochondrial activity.

Previous studies have demonstrated that mitochondrial activity is regulated by several metabolic pathways, such as the PGC-1α, mTOR, and c-Myc pathways (37, 40). Our initial studies revealed that the protein level of PGC-1α was significantly greater in DHA-treated NK cells than in control NK cells in the presence of IL2 plus IL15 or IL15 alone for 10 days in vitro, whereas no differences were found in the protein levels of p-S6^Ser235/236 (indicator for mTOR complex 1 activity), p-Akt^Ser473 (indicator for mTOR complex 2 activity), or c-Myc (Fig. 6A and B). Our data are consistent with a previous study in which DHA promoted PGC-1α expression, which enhanced mitochondrial biogenesis in myoblasts (41). We next used the PGC-1α signaling inhibitor SR-18292 in the presence of DHA. Purified splenic NK cells sorted (purity ≥ 98.0%) from WT mice were treated with DMSO, 10 μmol/L DHA or DHA (10 μmol/L) plus SR-18292 (20 μmol/L) in the presence of 50 ng/mL IL15 alone. Our results showed that IFNγ production, mitochondrial mass, membrane potential, and glucose uptake were increased by DHA, and this was reversed by SR-18292 (Fig. 6C–F; Supplementary Fig. S8B). However, the increase in CD107a expression induced by DHA was only slightly inhibited by SR-18292 (Fig. 6D). Overall, these findings suggest that DHA directly enhances the mitochondrial mass and effector functions of NK cells by regulating the PGC-1α pathway in vitro.

One recent study demonstrated that PGC-1α plays critical roles in mediating the effects of NK cells on B16F10 cells in vivo (42). To confirm the critical role of PGC-1α in the mediation of DHA-related antitumor effects by NK cells in vivo, we next fed littermate controls and PGC-1α^ΔNK mice with DHA-enriched diet for 3 weeks and then challenged the mice with an i.v. injection of B16F10 melanoma cells. We found that the PGC-1α^ΔNK mice had more lung tumor nodules in their lungs than did their littermate controls (Fig. 6G).

Overall, these findings suggest that DHA directly enhances the mitochondrial metabolism and effector functions of NK cells by regulating PGC-1α expression.

To date, the most frequently referenced anticancer activities of ω-3 PUFAs are the direct activation of apoptotic pathways and an increase in lipid peroxidation, which can lead to cytotoxicity in tumor cells (43). In addition, the beneficial effects of ω-3 PUFAs have been reported to result from their binding to G protein–coupled receptors that facilitate the clearance of cancer cell debris via macrophage phagocytosis (44). In the present study, we found that DHA treatment increased IFNγ levels and direct cytotoxicity in NK cells both in vivo and in vitro. By using scRNA-seq, Rag2^−/− mice, and WT mice with NK-cell depletion, we revealed that DHA could suppress lung tumor growth by promoting NK-cell effector functions. The finding that DHA withdrawal at the time of B16F10 cell injection had a weaker antitumor effect than continuous DHA feeding suggested that DHA also directly affects tumor cell growth and even enhances the TME for NK-cell action.

Cellular metabolism is critical for the appropriate functioning of NK cells and mitochondria are the central hubs of metabolic signals, especially OXPHOS (45). Increased mitochondrial activity and OXPHOS are important for NK-cell IFNγ production and cytotoxicity during activation (27). Our scRNA-seq results showed that DHA supplementation could increase the level of NK cell–mediated cytotoxicity as well as OXPHOS in NK cells. Furthermore, DHA supplementation increased the mitochondrial mass and mitochondrial membrane potential in NK cells both in vivo and in vitro. Metabolic flux analysis also revealed that DHA treatment enhanced NK-cell OXPHOS activity, as indicated by the increase in the basal OCR and maximum respiration after DHA treatment. This evidence suggested that DHA-enhanced mitochondrial activity mediates the increase in NK-cell effector functions.

ω-3 PUFAs can induce the expression of transcription factors that regulate mitochondrial biogenesis, including PGC-1α and nuclear respiratory factor-1 (NRF1; ref. 46). DHA directly promotes mitochondrial biogenesis and induces the expression of genes encoding several key mitochondrial enzymes (47). Our data showed that DHA enhanced the expression of PGC-1α, along with its target genes mt-Co1, mt-Co2, mt-Co3, mt-Cytb, mt-Atp6, and mt-Nd1, in NK cells, consistent with previous findings (41, 48). One previous study revealed that PGC-1α is also responsible for OXPHOS activity in NK cells (42). Treatment of NK cells with SR-18292 reduced the effect of DHA on IFNγ secretion, mitochondrial mass, membrane potential, and 2-NBDG in NK cells. By using PGC-1α^ΔNK mice in which PGC1-1α was conditionally deleted from NK cells, we found that the protective effects of DHA against tumor cells were abolished in PGC1-1α–deficient NK cells. Taken together, these data reveal that DHA may act as a direct agonist to promote the expression of mitochondrial genes and enhance mitochondrial OXPHOS activity in NK cells via PGC-1α.

Why DHA preferentially activates NK cells but not T cells, B cells, or other immune cells is intriguing. By using FITC-labeled DHA to treat splenocytes and B16F10 cells in vitro, we found that NK cells exhibited more favorable absorption of DHA than did other cells (e.g., T cells, B cells, and tumor cells). By using the Schmiedel data set from the human protein atlas website, we found that NK cells showed high expression of some membrane-associated proteins, including SLC27A3, FABP3, and SLC27A1. Collectively, our data suggest that a relatively higher expression of some membrane-associated proteins in NK cells might be responsible for the favorable absorption of DHA by NK cells.

Our scRNA-seq data showed that several subsets of DHA-fed mice, including B cells, naïve CD4⁺ T cells, naïve CD8⁺ T cells, and activated CD8⁺ T cells, had enhanced ribosome, proteasome, and protein export activity. The upregulated activity of these pathways indicated that these cells were healthy and exhibited good maintenance and quality control (49). However, the activation or effector functions of these subsets were downregulated in DHA diet–fed mice. Previous studies have reported that the incorporation of ω-3 PUFAs into the membrane of CD4⁺ T cells induces changes in the membrane domains, which might blunt the secretion of cytokines (50, 51). In addition, several DHA-derived specialized proresolving mediators, such as RvD1, inhibit the secretion of the cytokines TNFα, IFNγ, IL17, and IL2 by CD4⁺ T and CD8⁺ T cells but enhance NK-cell cytotoxicity (52–54). Thus, we believe that DHA has a widespread influence on cell biological processes and that the output of cell function is a comprehensive manifestation of various biological events. Understanding the underlying mechanisms may improve the therapeutic effects of ω-3 PUFA usage on cancer.

In summary, we have shown that DHA suppresses lung tumor growth of B16F10 melanoma cells by enhancing NK-cell IFNγ production and cytotoxicity, possibly through the improvement of mitochondrial activity via the PGC-1α pathway. Our findings broaden the current knowledge on the capacity of DHA supplementation to protect against tumor growth from the perspective of immunomodulation.

No disclosures were reported.

S. Wu: Data curation, methodology, writing–original draft, writing–review and editing. H. Peng: Data curation, methodology. S. Li: data curation, methodology. L. Huang: Data curation. X. Wang: Data curation. Y. Li: Data curation. Y. Liu: Methodology. P. Xiong: Visualization. Q. Yang: Methodology. K. Tian: Methodology. W. Wu: Methodology. R. Pu: Methodology. X. Lu: Investigation. Z. Xiao: Investigation. J. Yang: Investigation. Z. Zhong: Investigation. Y. Gao: Supervision, writing–original draft, writing–review and editing. Ya. Deng: Conceptualization, supervision, writing–original draft, writing–review and editing. Yo. Deng: Conceptualization, supervision, writing–original draft, writing–review and editing.

This study was supported by funding from the National Key Research and Development Project (No. 2020YFA0113500 to Yo. Deng and No. 2019YFA0111200 to Y. Gao), the National Natural Science Foundation of China (82304511 to S. Wu, No. 81922068 to Yo. Deng, No. 81703521 to Ya. Deng, and No. 81900055 to H. Peng) and the Scientific Research Project of the Hunan Provincial Health Commission (No. B202302078468 to S. Wu). This study was also supported by funding from the Natural Science Foundation of Hunan Province (No. 2023JJ30320 to Ya. Deng; 2023JJ40349 to S. Wu; No. 202122JJ40197 to P. Xiong; and No. 2021JJ40274 to H. Peng). This work was also supported by funding from the Science Foundation of Hunan Children's Hospital (2019 and 2021). This study was supported by grants from the Chongqing Science and Technology Commission of China (cstc2021jcyj-jqX0006 to Yo. Deng) and Guangdong Association of Clinical Trials (GACT)/Chinese Thoracic Oncology Group (CTONG), Guangdong Provincial Key Lab of Translational Medicine in Lung Cancer (CTONG-YC20210104 to Z. Zhong). We wish to thank Pro. Li Liping for strong spiritual support. The authors would like to thank Solutex GC, SL, for providing the OMEGATEX 0076-TG DHA oil.