Eighteen patients with kidney tumor were included in this study. Kidney specimens were obtained. These patients were treated by nephrectomy for renal cancer in our center between January and May 2021. We were able to establish primary cultures of clear cell renal cancer in 8 of them. The clinical and pathological data of those 8 patients are reported in Supplementary Table 1 of the Supplementary Data .

Once the nephrectomy was performed, the surgical specimen was transported in a fresh buffered salt solution to the department of Pathology for macroscopic examination. After anatomopathological analysis, a 1 cm³ non-necrotic fragment of the tumor was selected and stored at 4°C in 50 mL of PBS (phosphate-buffered saline, Sigma Aldrich, St. Louis, MO, USA) until use. Following this, mechanical digestion was performed under sterile conditions using two cold scalpels. The resulting millimeter-sized pieces were incubated at 37°C for 90 min under agitation in the presence of DMEM 25mM glucose, 1mM pyruvate, 1% penicillin–streptomycin (Sigma Aldrich), 0.2 Wünsch units/mL liberase (Roche, Basel, Switzerland) and 0.1 mg/mL DNase (Roche). The solution obtained was sieved through a 70 μm filter (SPL Life Sciences, Pocheon, Republic of Korea). Suspended cells were pelleted for 5 min at 300× g. Next, cells were subcultured in culture medium 1 containing 33% of Mix AmnioMAX™ (AmnioMAX™medium + AmnioMAX™ supplement) (Thermo Fisher Scientific, Waltham, MA, USA), 10% of Fetal Bovine Serum (FBS) (Thermo Fisher Scientific), 1% of RGEM1™ (LONZA, Basel, Switzerland), 2mM of Glutamine (Sigma-Aldrich), 1% of penicillin and streptomycin (Sigma-Aldrich) and 54% of DMEM 25mM of glucose (Sigma-Aldrich) prewarmed to 37°C and transferred to 25 cm2 culture flasks (SPL Life Sciences). After overnight incubation (37°C, 5% CO2), culture medium was renewed, and the unattached cells were removed. This procedure was repeated every 24 h until the flasks were free of debris and unattached cells. Thereafter, the culture medium was changed every 2 to 3 days until confluence. At 70–80% confluence, cells were washed with PBS and detached using trypsin (20 μL/cm2 trypsin-EDTA 1X (trypsin 0.05%; EDTA 0.02%), Sigma-Aldrich) for 5 min. Cells were then pelleted by 5 min centrifugation at 500× g and resuspended in 2 mL of culture medium 2 containing 33% of Mix AmnioMAX™ (Thermo Fisher Scientific), 10% of FBS (Thermo Fisher Scientific), 2mM of Glutamine (Sigma-Aldrich), 56% of DMEM 25mM of glucose (Sigma-Aldrich). Cell count was performed by trypan blue exclusion and Neubauer hemocytometer. Some of the cells were frozen at −80°C, and the rest were transferred to a 75 cm2 flask with culture medium 2. Once 80% confluence was obtained, the cells were transferred to a T175 cm2 flask for amplification. Culture medium was optimized for survival after mechanical and chemical digestion and to avoid culture infections during tissue handling. It contained growth factor supplements for use with renal epithelial cell basal medium (1% REGM™ (Renal Epithelial Cell Growth Medium SingleQuots™ Kit, LONZA, Bâle, Switzerland)) and 1% antibiotic (penicillin and streptomycin, Sigma-Aldrich). After the first passage, antibiotics were no longer needed and could constitute an avoidable bias for cell growth.

The ACHN cells (kindly provided by the laboratory of translational genomics at the National Cancer Institute) and Renca cells were cultured at 37C° in a humidified atmosphere (95% air; 5% CO2) in standard DMEM high glucose media (25mM) (PAN biotech, Aidenbach, Germany), supplemented with 10% Fetal Bovine Serum (FBS) and 2mM glutamine (PAN biotech). The medium was changed every 2 days. The doses of low glucose 2.8mM (LG) and ketone bodies (KB) were chosen in agreement with concentrations used in previous in vitro studies (22). For the KB treatment, 5 mM of acetoacetate and β-D-hydroxybutyrate (Sigma Aldrich, Lyon, France) were added to the low glucose media (2.8mM) supplemented with 10% SVF and 2mM glutamine, and cells were collected after 6 days of culture for further analysis.

To assess cell viability in response to glucose starvation and KB exposure, we used an automated live cell imaging (IncuCyte ZOOM system, Sartorius) in which cell density and shape were observed by photonic microscopy while cells were maintained under standard medium or exposed to KB as described above. Real-time live-cell images were taken every 2 hours and for analysis of cell viability, well confluence (%) was automatically calculated during 144h by using the Basic Analyzer segmentation mask (IncuCyte ZOOM software 2015A). Cell viability was analyzed by Crystal Violet staining as described elsewhere (23). Absorbance was measured using a microplate reader at 590 nm (CLARIOstar®, BMG Labtech).

Citrate synthase activity, considered as a marker of mitochondrial mass was measured by adding cells to a reaction mix (DTNB 1mM, oxaloacetic acid 10 mM, acetyl coA 0.3 mM, triton X100 0.1%) and the appearing rate of CoA-SH was measured at 37°C at 412nm on an UVmc2 spectrophotometer (SAFAS, Monaco) as described elsewhere (24).

Mitochondrial oxygen consumption measurements were performed at 37°C and atmospheric pressure using a high-resolution oxygraph (O2K, Oroboros instrument, Innsbruck, Austria). Respiration rates on intact cells (4x10⁶ cells) were measured in standard medium or low glucose with added ketone bodies (KB) as described elsewhere (24). Oxygen consumption rate (OCR) and Extracellular acidification rate (ECAR) were also measured with an XF96 extracellular flux analyzer (Agilent Technologies, Santa Clara, CA) to determine the bioenergetic profile of intact cells (30 × 10³ cells/well) after being exposed to KB or HG as recommended by the manufacturer. Three independent replicates of each oxygen consumption measurement were generated, and results were normalized according to cell concentration.

Glucose, pyruvate and lactate concentrations were measured by spectrophotometry with a Xenius apparatus (SAFAS, Monaco) according to the manufacturer (Abcam, Paris France). The cytosolic NAD(H)-redox state was determined using the metabolite indicator method, based on the lactate dehydrogenase reaction, by measuring pyruvate to lactate ratio in cell culture.

To determine the mtDNA copy number in human cells, quantitative PCR (qPCR) was performed by SYBR Green incorporation using the Chromo4 System (Biorad) following the manufacturer recommendations. Selected primers were respectively, MT-ND4 (Forward: actctcactgcccaagaact and Reverse: gtgtgaggcgtattatacca) and MT-COX1 (Forward: tacgttgtagcccacttccact and Reverse: agtaacgtcggggcattccg) to quantify the mtDNA copy number, and GAPDH (Forward: ccctgtccagttaatttc and Reverse:caccctttagggagaaaaa) and AIBI (Forward: ggagtttcctggacaaatga) and Reverse:aggactggcgtttatgtctt) to quantify the nuclear DNA in ACHN cells. The PCR reactions were performed as follows: initial denaturing at 95°C for 10 min, and 35 cycles at 95°C for 30 s, 60°C for 1 min. A melting curve was analyzed in order to check the specificity of the PCR products.

Total RNA was extracted using RNeasy Micro kit following the recommendations of the manufacturer (Qiagen, Hilden, Germany). A Bioanalyzer 2100, with a RNA6000 Nano kit (Agilent Technologies, Santa Clara, CA, U.S.A.) was used to assess RNA quality. The RNAs was reverse transcribed using Superscript II reverse transcriptase (Invitrogen, Waltham, Ma). Expression of selected genes including Nuclear Respiratory Factor 1 (NRF1), Nuclear Respiratory factor 2 (NRF2), Peroxisome Proliferator-Activated Receptor Gamma Coactivator 1-Alpha (PGC1α) and Transcription factor A (TFAM) and PDL-1 or CD274 was analyzed by SYBR Green fluorescent dye incorporation using the Chromo4 System (Biorad) as described elsewhere (25). Specific gene expression was calculated using the 2^-ΔΔCT method using Actin B for normalization. Primer sequences are available upon request.

Twenty females (two experiments, 5 mice per group and per cage) eight-week-old CD-1 nude mice (Charles River’s laboratory, France) were maintained in a temperature-controlled room with food and water provided ad libitum. 2 × 10⁶ ACHN cells were subcutaneously grafted in the flank of nude mice. Mice were randomized into two groups, either fed with a standard diet (control group) or with a 2:1 high fat ketogenic diet (ketogenic group).

Standard diet (ssniff Spezialdiäten GmbH, Soest, Germany) contained 16.1% proteins, 7.1% lipids and 57.2% glucids, resulting in a 0.09:1 ketogenic ratio. KD, TCM completed (ssniff Spezialdiäten GmbH, Soest, Germany) contained 13.8% proteins, 56.1% lipids (31.1% TCL, 25% TCM-C10) and 12.5% glucids, resulting in a 2.1:1 ketogenic KD diet was provided by the Vitaflo company (Vitaflo International Ltd, Liverpool, New England). Tumor progression was monitored weekly by ultrasounds for 8 weeks (Vevo 770^® , VisualSonics, Toronto, Canada). Tumor volume was calculated using the empirical ellipsoid formula: V (mm3) = 4/3 π x (l/2) x (w/2) x (d/2), with l for length, w for width and d for depth. The tumor growth (growth percentage) was calculated according to the (V - Vi)/Vi formula, V standing for the measured volume and Vi for the initial volume. Glycemic blood and blood ketone levels were monitored by standard devices (Freestyle Optium Neo, Abbott Laboratories, Illinois, USA) after tail section, twice a week. Eight weeks after xenografts, mice were sacrificed by cervical dislocation after isoflurane anesthesia. Tumors were removed from animals, measured and weighed, then quickly frozen and stored at -80°C.

For the second assay, twenty 7–8-week-old female BALB/c mice (Janvier Labs, Le Genest Saint Isle, France) (two experiments, 5 mice per group and per cage) were inoculated with 5×10⁵ Renca tumor cells (ATCC, CRL-2947) subcutaneously at day 0 and then injected with the indicated treatment. Isotype control (LTF‐2 clone) and anti-PDL1 antibody (10F.9G2 clone, Bio-XCell, West Lebanon, NH, USA) or vehicle (PBS) were conducted by intraperitoneal injection (200µg/mouse) every 3 days. Ketogenic or standard diet was administered ad libitum. Tumor volumes were measured and calculated with the equation: volume=length×width²/2. Survival was performed using human endpoints, mice were sacrificed when an endpoint was reached to comply with ethical regulations. This means that experiments were terminated when mice reached defined criteria (tumor size length>17mm; weight loss > 20%, or ulceration at the tumor site). Twelve days after tumoral cell injection, animals were pooled and randomly divided into 2 designated experimental groups one with ketogenic diet and the second with standard diet. Each group was subdivided in 3 groups, one received anti-PDL1 mAb, the second isotype control and the third PBS (vehicle).

All tissue samples were formalin-fixed, paraffin-embedded and cut into 4-μm-thick sections. Slides were stained with hematoxylin, eosin and saffron (HES). Immunohistochemical staining was carried out with rabbit monoclonal anti-PD-L1 antibody (clone NAT-105, Leica,1:200 dilution, Germany) and anti-mib1 antibody (clone ab15580, Abcam, 1:500 dilution, UK), on a Bond-III automated immune stainer (Leica Microsystem, Nussloch, Germany). After heat antigen retrieval during 20 min in ER2 pH 9 buffer, sections were incubated at room temperature for 20 min with the primary mouse monoclonal anti-PDL-1 antibody (ab205921, abcam) at a 1:500 dilution and were revealed with the Leica Bond Refine detection kit for 10 min. The specimens were then counterstained with hematoxylin (both from Leica Microsystems, Newcastle, U.K).

Statistical analysis was conducted using Graphpad software (Prism version 8.0.1). Differences between groups were evaluated by the Mann-Whitney statistical test. The asterisk (*, ** and ***) indicates significant differences (respectively p < 0.05, p < 0.01 and p < 0.0001). Analysis of tumor growth between the two experimental groups was performed using the Student’s t test. Kaplan-Meier curves and corresponding Gehan-Breslow-Wilcoxon tests were used to evaluate the statistical differences between groups for the survival studies. A p-value <0.05 was considered to be statistically significant.

The ACHN cells were cultured in low glucose medium enriched in KB or standard medium during 6 days. At day 6, growth proliferation of ACHN cells was significantly reduced by 11% in the presence of low glucose in addition to ketone bodies compared to HG medium ( Figure 1A ). Similar results of growth proliferation were reduced in LG with added KB treated primary kidney cancer cells ( Supplementary Figure ). To better understand the reduction of growth cell proliferation under LG with added KB, metabolites in cell culture supernatants were quantified. Glucose consumption was significantly increased by 77% in ACHN cells under low glucose in addition to ketone bodies media, also associated with a significant reduction of 86% of lactate production ( Figure 1B ). The reduction of lactate was confirmed with 24% decrease of the [lactate]/[glucose] ratio ( Figure 1C ). The cytosolic NADH/NAD+ ratio was estimated by using the lactate/pyruvate (L/P) ratio at equilibrium and was significantly reduced (90%) for ACHN cells exposed to low glucose coupled with ketone bodies compared to high glucose medium, suggesting a switch from glycolysis to oxidative metabolism ( Figure 1D ). The level of ECAR of ACHN cells was significantly reduced by 54% in low glucose with added ketone bodies treated cells compared to high glucose cells ( Figure 1E ).

To confirm our hypothesis, mitochondrial respiration in intact ACHN cells was measured. The basal respiration was significantly increased in ACHN by 1.6 fold change (61%) compared to untreated cells after 6 days of LG in addition to KB exposure ( Figure 2A ). Furthermore, the part of OXPHOS respiration used to produce ATP was significantly increased with a 2 fold change in cells grown in LG with added KB compared to HG cells ( Figure 2B ). However, caution is needed in extrapolating OXPHOS respiration to ATP synthesis through OXPHOS in tumor cells (26). Similarly, the maximal capacity of respiration was increased by 1.6 fold change ( Figure 2C ). In addition, LG with added KB exposure significantly increased mtDNA copy number in ACHN cells with a 2.6 fold change (159%) compared to untreated cells after 6 days of low glucose coupled with ketone bodies exposure, supporting the induction of mitochondrial biogenesis ( Figure 2D ). In parallel citrate synthase (CS) activity considered as a key marker of mitochondrial mass, was significantly increased by 98% in treated cells compared to control cells ( Figure 2E ).

The expression of genes involved in mitochondrial biogenesis including, NRF1, NRF2, PGC1a and TFAM were assessed in treated cells ( Figure 2F ). After 6 days of low glucose in addition to ketone bodies exposure gene expression of selected mitochondrial genes were significantly induced by NRF1 by 2.1, NRF2 by 2.2, PGC1a by 2.9, and TFAM by 14.1 fold changes strongly suggesting an increase of mitochondrial biogenesis. Moreover, LG with added KB significantly improved the mitochondrial network by reducing fragmented mitochondria by 29% and increasing mitochondrial lengths by 225% with the appearance of tubular structures demonstrating the beneficial effect of LG coupled with KB on mutant cells ( Figures 2G, H ).

We further analyzed the consequences of KD on the growth proliferation of ACHN cells grafted mice. Animals were subdivided into two groups, one group with a normal diet (ND) and one group with a 2:1 KD ( Figure 3A ). Tumor growth was regularly measured once a week in both mouse groups. The mean tumor growth was 930 ± 50% in the control group vs 190 ± 70% in the KD group after 8 weeks of KB exposure (p<0.001) ( Figure 3B ). The mean blood ketone level was 0.78 (± 0.16) mmol/L in the control group, and 1.12 (± 0.16) mmol/L in the ketogenic group (p=0.001) ( Supplementary Data ). The mean blood glucose level was 7.8mM in the control group and 6.8mM in the ketogenic group (p=0.01) ( Supplementary Data ). In mice, ketosis was quickly reached without weight change between both groups ( Supplementary Data ).

Since immunotherapies are now widely used in kidney cancer therapies, we looked at the influence of KD on the PD-1 immune-checkpoint inhibitor. In the KD treated mouse group, we confirmed the over-expression of PD-L1 by immunochemistry on the surface of tumor cells ( Figure 3C ). The Ki67 tumor proliferation rate was not significantly different between the two conditions (around 80%) ( Figure 3C ). Gene expression of CD274, which encodes the immune inhibitory receptor ligand of PD-L1 was increased by 2 fold change (99%) in mouse tumors in the KD group compared to ND group ( Figure 3D ). This increase was further confirmed by qPCR in ACHN cells in vitro revealing that PD-L1 was significantly 4 fold change increase after 6 days of exposed to low glucose environment in addition to ketone bodies ( Figure 3E ).

Next we used Renca cells, a murine cell line of renal carcinoma exposed for 6 days with low glucose in addition to ketone bodies. Number of Renca cells was significantly lower in LG with added KB condition than in HG condition with a significant reduction of 83% viability as revealed by crystal violet staining ( Figures 4A, B ). Mitochondrial respiration rates of Renca cells exposed for 6 days to low glucose coupled with ketone bodies compared to HG cells were measured. The basal respiration was significant induced with a 1.9 fold change (88%), ATP-linked respiration with a 2.2 fold change (121%) and the maximal respiration capacity was significantly increased in low glucose, KB treated cells than in HG ( Figures 4C–E, G ) confirming the significant improvement of oxidative mitochondrial metabolism already seen with the ACHN LG with added KB treated cells. Glucose consumption was significantly reduced by 83% in Renca cells under low glucose, KB treatment, also associated with a significant reduction of 91% of lactate production ( Figure 4G ). The reduction of lactate was confirmed with 54% decrease of the [lactate]/[glucose] ratio ( Figure 4H ). The level of acidification of Renca cells was significantly reduced by 63% in low glucose coupled with KB treated cells ( Figure 4I ).

To confirm the effect of KB on tumoral cells, syngeneic mice were treated with KD and/or anti-PD-L1 immunotherapy 12 days ( Figures 5A ) after been grafted with Renca cells. Tumors from mice under KD treated or not presented a reduced volume ( Figure 5A ). Eighteen days after tumor graft, mice under KD and treated with anti-PDL1 mAb presented a significantly reduced tumor volume compared to mice under KD treated with isotype control ( Figures 5B, C ). Interestingly, treatment with anti-PDL1 mAb combined with KD was significantly efficient in one mouse under KD treated with anti-PDL1 mAb which survived while all mice from the other groups died. ( Figure 5D ).