Lower HRK levels and increased BCL-xL binding to BAK help therapy-treated melanoma cells avoid cell death

We decided to focus in melanoma since senescence and apoptotic protection play a key role in this type of cancer development and therapy outcome [18, 24, 28]. We used three melanoma cell lines; SK-MEL-103 (NRAS mutant), SK-MEL-28 (BRAFV600E mutant) [28], and the M16 (BRAFV600E mutant) that was derived from a patient continuously exposed to UV light, as previously described [29]. To induce senescence, we chronically treated these cell lines with a CDK4/6 inhibitor for 7 days [8] and, since this inhibitor does not uniformly induce senescence in all cancer cells [30], we confirmed the phenotype using different well-established markers: SA-β-galactosidase, p21, and p16 [31, 32]. As expected, we detected a significant increase of these markers in all three cell lines when treated with palbociclib, thus indicating therapy-induced senescence (TIS) (Fig. 1A, B). To assess senescence morphologically, we examined by transmission electron microscopy (TEM) for potential phenotypic changes. Overall, we detected elongated mitochondria as measured by a decrease in their circularity (Fig. 1C) and an increase in the number of lysosomes (Fig. 1D) in palbociclib-treated cells, as previously described [5, 33]. All these observations confirm that palbociclib effectively induces senescence in all three melanoma cell lines examined.

Senescent cells often present resistance to die by apoptosis [2]. Therefore, we next sought to study how TIS modulates apoptosis using BH3 profiling. As mentioned above, senescent cells present longer mitochondria, and we reasoned that this could affect MOMP and cytochrome c release measurements, which are required for BH3 profiling analyses [22–24]. Thus, we first evaluated if these selected senescent melanoma cells presented a larger mitochondrial content compared to non-senescent cells. Using TOM20 as a mitochondrial marker, we observed a significant increase in its expression in palbociclib-induced senescent SK-MEL-28, SK-MEL-103 and M16 cells compared to non-treated (proliferating) cells (Fig. 2A). We also evaluated cytochrome c staining by flow cytometry and, correlating with TEM, fluorescence microscopy and western blot, we detected a significant increase in its intensity after senescence induction (Fig. 2A). Therefore, for flow cytometry BH3 profiling analyses we adjusted the cytochrome c gating taking into consideration this variation in basal fluorescence (see the “Methods” section for more details). When comparing non-treated (proliferating) and palbociclib-treated melanoma cancer cells by BH3 profiling, we observed that these cell lines showed a different priming pattern when becoming senescent. If a given treatment or state sensitizes cells towards apoptosis, this would cause a leftward shift of the cytochrome c retained curve after BIM peptide exposure, since less amount of BIM is required to promote MOMP and cytochrome c release, indicating an increase in overall apoptotic priming (positive Δ% priming). In contrast, if a treatment or state renders cells more resistant to apoptosis, we would observe a shift of the curve to the right, indicating a decrease in priming (negative Δ% priming). Surprisingly, we observed that SK-MEL-103 cells treated with palbociclib became slightly primed for apoptosis, while the SK-MEL-28 and M16 cells clearly turn out to be less primed for apoptosis (Fig. 2B).

To further elucidate this distinct effect on apoptotic priming induced by palbociclib, we compared the expression of different pro-apoptotic members of the BCL-2 family of proteins between proliferating and senescent cells. For instance, in all cell lines we observed a significant reduction of the activator protein BIM (Fig. 2C), that could partially explain why senescent cells exert resistance to cell death. When further analyzing the SK-MEL-103 cells, we identified that the expression of the effector proteins BAX and BAK significantly increased after palbociclib treatment, correlating with the observed increase in Δ% priming detected by BH3 profiling with the BIM BH3 peptide (Fig. 2B). In fact, these results correlate with previous mRNA expression data results from the Serrano laboratory [34] (Supplementary Fig. 1). However, in SK-MEL-28 and M16 cells we observed that BAK remained constant or increased while BAX expression decreased, partially explaining the potential role of the latter regulating negative Δ% priming detected by BH3 profiling (Fig. 2C). Furthermore, we assessed tumors from SK-MEL-103 xenografts treated with palbociclib [35], which became senescent (Supplementary Fig. 2), and detected a decrease in BIM and an increase in BAK (and BAX) expression, correlating with our in vitro observations (Fig. 2D).

In summary, palbociclib promoted a decrease in the activator protein BIM, thus preventing apoptosis, and effectively inducing senescence, assessed by SA-β-galactosidase, p21 and p16. Paradoxically, one melanoma cell line became slightly primed for apoptosis (positive Δ% priming) while the other two showed protection towards this form of programmed cell death (negative Δ% priming) in part due to a distinct change in BAX expression after TIS.

Another hallmark of senescent cells is their pro-survival adaptation through anti-apoptotic BCL-2 family proteins, and non-surprisingly one of the most studied senolytics has been the promiscuous BCL-2/BCL-xL/BCL-W inhibitor navitoclax [11]. We sought to elucidate if TIS melanoma cells also presented a similar adaptation and, if that was the case, how they prevented apoptosis. Thus, we performed BH3 profiling using specific BH3 peptides BAD (for BCL-2 and BCL-xL), HRK (for BCL-xL) and MS1 (for MCL-1), as previously described [24, 25], on the same cell lines after senescence induction using palbociclib, and γ-irradiation (which also increased SA-β-galactosidase activity, p21 and p16 expression in all three cell lines as shown in Supplementary Fig. 3). In brief, a positive Δ% priming using these sensitizer BH3 peptides would point to a specific anti-apoptotic protein that senescent cells become addicted to. We found that SK-MEL-103, SK-MEL-28 and, to a lesser extent, M16 displayed a significant BCL-xL adaptation, as we detected an increase in the % of cytochrome c released with the HRK (and BAD, that binds to BCL-2 and BCL-xL) peptide (Fig. 3A). We also detected a minor increase in priming with the MS1 BH3 peptide, indicating a partial adaptation through MCL-1 but to a lesser extent than BCL-xL (Fig. 3A). We next explored anti-apoptotic inhibition potential to eliminate senescent cells. Using palbociclib and γ-irradiation, we generated senescent melanoma cells and tested BH3 mimetics’ senolytic activity. We exposed these cells to ABT-199 (venetoclax), the BCL-xL inhibitor A-1331852, the MCL-1 inhibitor S63845↗, ABT-263 (navitoclax) and the BCL-xL degrading PROTAC DT2216 [16]. As anticipated by our BH3 profiling results, we observed significant senolytic activity in these melanoma cell lines when targeting BCL-xL using A-1331852, navitoclax or DT2216 (Fig. 3B); and in SK-MEL-103 and SK-MEL-28 BCL-xL targeting therapies had a greater senolytic activity compared to M16. Furthermore, we detected a significant increase in cell death with the MCL-1 inhibitor S63845↗ in SK-MEL-28 suggesting an implication of this protein in apoptotic resistance in senescent cells harboring a BRAFV600 mutation [24]. These results indicate that TIS melanoma cells block apoptosis primarily through BCL-xL, and to a lesser extent using MCL-1. Consequently, inhibiting this anti-apoptotic protein with specific inhibitors - beyond navitoclax that also targets BCL-2 and BCL-W - such as the BH3 mimetic A-1331852 or a PROTAC like DT2216 that would not cause thrombocytopenia, due to the minor expression of E3 ligase in platelets [16], appears as an effective therapeutic strategy to remove melanoma senescent cells.

To elucidate how TIS melanoma cells become dependent on BCL-xL, we first analyzed the expression of this anti-apoptotic protein. We found in palbociclib-treated SK-MEL-103 and M16 cells that its expression increased. Surprisingly, SK-MEL-28 cells did not exert significant BCL-xL changes upon treatment, suggesting a potential role of another BCL-2 family member to regulate its availability (Fig. 4A–C). In this regard, we analyzed several BCL-2 family proteins, and found that the sensitizer HRK, which specifically binds to BCL-xL [36], was clearly downregulated in all three cell lines when induced into senescence (Fig. 4A–C). This downregulation observed in cell lines could be partially explained by a decrease in gene transcription (Fig. 4D). Furthermore, SK-MEL-103 xenografts treated with palbociclib also experienced a decrease in HRK protein expression (Fig. 4E). In other words, senescent inducers such as palbociclib and γ-irradiation decreased the expression of HRK, initially bound to BCL-xL, thus liberating this pro-survival protein to protect cells towards apoptosis. Similar results were observed in healthy donors comparing control skin with UV irradiated (Table 1), where the latter showed an increase trend in the senescence markers CDKN1A (p21) and CDKN2A (p16) (Supplementary Fig. 4), and a decrease trend in HRK mRNA expression (Fig. 4F) pointing to a potential BCL-xL adaptation in senescent skin from patients. Furthermore, knocking down HRK using siRNA in SK-MEL-28 and SK-MEL-103 promoted a similar effect regarding apoptotic priming and BCL-xL dependence as palbociclib-induced senescent cells (Supplementary Fig. 5A–C).

We previously showed that palbociclib treatment promoted a decrease in BIM, HRK and, in some cases, BAX expression, consequently preventing apoptosis. So, our next question was how BCL-xL inhibitors restored apoptotic cell death in senescent cells. The BCL-2 family of proteins represents a complex interactome regulated at multiple levels, including protein expression, posttranslational modifications, and dynamic binding affinities. To further elucidate the role of BCL-xL in TIS, we immunoprecipitated this anti-apoptotic protein with a high efficiency (Fig. 5A) and analyzed its binding to BIM, BAX and BAK in SK-MEL-103 and SK-MEL-28. As shown in Fig. 2, both cell lines exerted distinct BAX/BAK expression changes when senescence was induced with palbociclib. However, in our immunoprecipitation analyses we observed a clear increase in binding between BCL-xL and BAK in senescent cells compared to control cells, while it decreased or remained similar for the other two proteins (Fig. 5B). Similar results were also observed in SK-MEL-28 and SK-MEL-103 cells transfected with siRNA against HRK (Supplementary Fig. 5D). In summary, we describe for the first time that palbociclib-induced senescence promotes different changes in the BCL-2 family proteins’ expression that lead to an increase in BCL-xL:BAK binding that prevents apoptosis and explains why specifically BCL-xL targeted agents are effective senolytics.

Melanoma cell line SK-MEL-28 was purchased at ATCC (ATCC^® CRL-1772^™, ATCC, Manassas, VI, USA) and SK-MEL-103 cell line was provided by Dr. Marisol Soengas. Patient-derived cell line M16 was kindly provided by Dr. Susana Puig from Hospital Clínic de Barcelona. SK-MEL-28 cell line was cultured in RPMI 1640 medium (31870, Thermo Fisher, Gibco, Paisley, Scotland) supplemented with 10% heat-inactivated fetal bovine serum (10270, Thermo Fisher, Gibco), 1% of l-glutamine (25030, Thermo Fisher, Gibco), and 1% of penicillin and streptomycin (15140, Thermo Fisher, Gibco). SK-MEL-103 and M16 cells were cultured in DMEM high glucose medium (41965, Thermo Fisher, Gibco) supplemented with 10% heat-inactivated fetal bovine serum, 1% of penicillin and streptomycin and 1% of l-glutamine. All cells were tested for mycoplasma and maintained at 37 °C in a humidified atmosphere of 5% CO₂.

BH3 mimetics ABT-199 (venetoclax), A-1331852 (A-133), S63845↗ and ABT-263 (navitoclax) were purchased at Selleckchem (Munich, Germany) and the PROTAC BCL-xL degrader DT2216 was purchased at AbMole BioScience (Houston, Texas, USA). Treatments were performed directly in the culture media at the indicated concentrations.

For palbociclib-induced senescence, cells were seeded in a six-well plate and treated with palbociclib (S1116, Selleckchem) at 1 µM dose. After 3 days cells were replenished with fresh media and palbociclib. Seven days after the treatment, senescent cells were used for further analysis.

For irradiated-induced senescence, cells were seeded in a six-well plate and irradiated with a gamma irradiator MARK I 30 (J.L. Shepherd & Associates, San Fernando, CA, USA) from the Scientific and Technological Centers of the University of Barcelona at 10 Gy. After 3 days, the media was changed and cells were maintained in culture till day 7 after irradiation and used for further analysis.

Cells were stained using CellEvent™ Senescence Green Detection Kit (C10851↗, Thermo Fisher). Briefly, cells were washed with phosphate buffer saline (PBS), tripsinized, fixed with 4% formaldehyde during 10 min at room temperature, washed with PBS + 1% BSA and incubated with the working solution at 37 °C without CO₂ during 2 h. After incubation, cells were washed with PBS and then imaged using a spectral flow cytometry Cytek AURORA instrument (Cytek Biosciences, California, USA) from the Scientific and Technological Centers of the University of Barcelona.

BH3 profiling was developed at the Letai laboratory, and the experiments were performed as previously described [20, 22, 25]. After inducing senescence for 7 days cells were stained with the viability marker Zombie Violet (423113, BioLegend, Koblenz, Germany) and then washed with PBS and resuspended in MEB (150 mM mannitol, 10 mM hepes–KOH pH 7.5, 150 mM KCl, 1 mM EGTA, 1 mM EDTA, 0.1% BSA, 5 mM succinate) in a final volume of 25 μL. Afterwards, cells were exposed to the following peptides: 25 μL of BIM BH3 peptide (final concentration of 0.01, 0.03, 0.1, 0.3, 1, 3, and 10 μM), 25 μL of BAD BH3 peptide (final concentration of 10 μM), 25 μL of HRK BH3 peptide (final concentration of 100 μM), and 25 μL of MS1 BH3 peptide (final concentration of 10 μM) in MEB with 0.002% digitonin in a 96-well plate (3795, Corning, Madrid, Spain) for 1 h following fixation with formaldehyde. Cells were finally stained with cytochrome c antibody (Alexa Fluor® 647 anti-Cytochrome c—6H2.B4, 612310, BioLegend), p21 antibody (sc-6246 PE, Santa Cruz Biotechnology, Texas, USA) and p16 antibody (Alexa Fluor® 488 CDKN2A/p16INK4a, bs-4592R-A488, Bioss Antibodies, Woburn, Massachusetts, USA). Individual BH3 profiling analyses were performed using triplicates for DMSO, alamethecin (BML-A150-0005, Enzo Life Sciences, Lorrach, Germany), the different BIM BH3 concentrations used, BAD, HRK, and MS1 BH3 peptides. The expressed values stand for the average of three different readings performed with a high-throughput spectral flow cytometry Cytek AURORA instrument (Cytek Biosciences) from the Scientific and Technological Centers of the University of Barcelona. % of cytochrome c released is used to calculate Δ% priming which represents the difference between treated cells minus non-treated cells for a given peptide.

Due to the difference of mitochondrial content between proliferating and senescence cells and therefore a major cytochrome c retained signaling was detected in senescent cells versus control cells, we adjusted the gating for cytochrome c retention with the control condition of each experimental setting (proliferating and senescent). Analyses were performed using FlowJo software.

Cells were trypsinized and stained with fluorescent conjugates of Annexin V (Alexa Fluor® 647 Annexin V, 640912, BioLegend) and DAPI (62248, Thermo Fisher) and analyzed on a flow cytometry Gallios instrument (Beckman Coulter, Nyon, Switzerland) from the Scientific and Technological Centers of the University of Barcelona. Viable cells are Annexin V DAPI negative, and cell death is expressed as 100%-viable cells.

Cells were lysed using RIPA buffer (150 mM NaCl, 5 mM EDTA, 50 mM Tris–HCl pH = 8, 1% Triton X-100, 0.1% SDS, EDTA-free Protease Inhibitor Cocktail (4693159001 Roche, Mannkin, Germany)) for 30 min at 4 °C followed by a centrifugation at 16,100 × g for 10 min. Frozen tissue from SK-MEL-103 xenografts was homogenized in RIPA buffer using a tissue homogenizer and centrifuged at 16,100 × g for 10 min. The supernatant was stored at −20 °C for protein quantification performed using PierceTM BCA Protein Assay Kit (23227, Thermo Fisher).

Cells were lysed using Immunoprecipitation buffer (150 mM NaCl, 10 mM Hepes, 2 mM EDTA, 1% Triton, 1.5 mM MgCl₂, 10% glycerol, and EDTA-free Protease Inhibitor Cocktail (4693159001 Roche)) and centrifuged at 16,100 × g, 15 min at 4 °C. Supernatants were incubated with magnetic beads (Dynabeads 10003D, Thermo Fisher) conjugated to 5 µg of rabbit anti-BCL-xL antibody (CST2764, Cell Signaling, Leiden, The Netherlands) or 5 μg of rabbit IgG antibody (CST2729, Cell Signaling) at 4 °C overnight. After magnetization, a part of the supernatant was mixed with half volume of 4× SDS–PAGE sample buffer, heated at 96 °C for 5 min and stored at −80 °C as unbound fractions. The rest of the supernatant was discarded. The resulting pellet was washed and mixed with 60 µL 4× SDS–PAGE sample buffer and heated for 10 min at 70 °C followed by magnetization and collection of the supernatant for further immunoblotting analyses.

Proteins were separated by SDS–PAGE (Mini-Protean TGX Precast Gel 12%, 456–1045, Bio-Rad, Hercules, California, USA) and transferred to PVDF membranes (10600023, Amersham Hybond, Pittsburgh, PA, USA). Membranes were blocked with dry milk dissolved in Tris buffer saline with 1% Tween 20 (TBST) for 1 h and probed overnight at 4 °C with the primary antibodies of interest directed against: rabbit anti-BCL-xL (CST2764, Cell Signaling), rabbit anti-BIM (CST2933, Cell Signaling), rabbit anti-BAK (CST12105↗, Cell Signaling), rabbit anti-BAX (CST2772, Cell Signaling), rabbit anti-HRK (PRS3771, Sigma-Aldrich, San Luis, Missouri, USA), mouse anti-γ−tubulin (T6557, Sigma-Aldrich), rabbit anti-Tom20 (ab186734, Abcam, Cambridge, UK) followed by anti-rabbit or anti-mouse IgG HRP-linked secondary antibody (CST7074 or CST7076 Cell Signaling) in 3% BSA in TBST for 1 h at room temperature. Immunoblots were developed using Clarity ECL Western substrate (1705060, Bio-Rad), visualized with LAS4000 imager (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) and ImageJ was then used to measure the integrated optical density of bands.

For transmission electron microscopy (TEM) preparation and analysis of melanoma cells (proliferating and senescence) growing on coverslips, the cells were first washed with 0.1 M phosphate buffer (PB) to remove the excess of culture medium and fixed with a freshly made 3% glutaraldehyde solution in 0.1 M PB for 1 h at room temperature. Subsequently, the fixative was removed, and the samples were maintained at 4 °C in a fresh fixative solution until processing.

The post-fixation procedure was conducted using 1% OsO₄ for a period of 90 min at 4 °C. Then, samples were dehydrated in increasing ethanol solutions (in accordance with the standard procedures). Coverslips with cells were then inverted (with the cells facing downwards) and placed on top of BEEM® capsules filled with Spurr resin (Electron Microscopy Sciences, Hatfield, PA, USA). The polymerization of the resin was carried out at a temperature of 60 °C in a stove for a period of three days. At this point, the glass was removed by thermal contrast switch while the cell monolayer remained at the top of the polymerized block. A Leica ultramicrotome EM UC7 (Leica Microsystems, Wetzlar, Germany) was employed for sectioning. Ultra-thin sections (60–70 nm) were mounted on cooper grids and stained with 2% Uranyl-less solution for 10 min and with a lead-staining solution for 5 min. The sections were observed using a JEOL JEM-1010 transmission electron microscope (JEOL Ltd, Akishima, Tokyo, Japan) coupled with an Orius SC1000 CCD camera (model 832) (Gatan Inc., Pleasanton, CA, USA) at the Electron Microscopy Unit from the Scientific and Technological Centers of the University of Barcelona.

The circularity of mitochondria was analyzed using the follow statement: A circularity value of 1.0 indicates a perfect circle. As the value approaches zero, it indicates an increasingly elongated polygon. The ImageJ (Fiji) software incorporates the capacity to perform circularity calculations. Quantification was performed based on the analysis of 30–60 mitochondria analyzed from 17–20 cells, comprising both control and senescent cells, across three independent experiments.

To quantify and compare the number of lysosomes in control and senescent cells we used images captured always at the same magnification (×30.000), with each field representing an area of 25 μm². The total number of lysosomes counted ranged from 49 in the control to 183 in the senescent cells.

Cells were gently washed once with cold PBS and then fixed for 20 min with 4% paraformaldehyde in PBS. Cells were then washed three times with PBS and permeabilized for 10 min with 0.1% saponin in PBS. Blocking was performed for 5 min with 0.02% saponin and 1% BSA in PBS. For primary antibody incubation, cells were incubated with anti-Tom20 (ab186734, Abcam) diluted in 0.02% saponin and 0.1% BSA solution for 1 h at room temperature. After primary antibody incubation, cells were washed three times with PBS. Secondary antibody incubation was carried out by incubating cells with anti-Mouse IgG (H + L), Superclonal™ Recombinant Secondary Antibody Alexa Fluor™ 488 (A28175, Thermo Fisher) and anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor 555 (A21428, Thermo Fisher) for 45 min in 0.02% saponin and 0.1% BSA solution at RT. For DAPI staining, cells were incubated with diluted DAPI stock solution (10374168, Thermo Fisher) to a final concentration of 300 nM in PBS for 5 min. Following staining, cells were washed three times with PBS, rinsed in water and mounted in Mowiol^® 4–88 (17951–500, Polysciences, Warrington, PA, USA).

For knockdown experiments, cells were transfected with HRK siRNA SMARTpool (77L-HUMAN-XX-0005, Horizon Discovery, Cambridge, UK) or non-targeting siRNA pool (77D-001810-10-7505, Horizon Discovery) using Lipofectamine® RNAiMax (13778-075, Thermo Fisher). Briefly, 10⁵ cells per condition were seeded in 12-well plates. Lipofectamine® RNAiMax and siRNAs were diluted in Opti-MEM® medium (31985062, Thermo Fisher) according to the manufacturer’s instructions, and the siRNA-lipid complex was added to the cells. The following day, cells were replenished with fresh media and maintained in culture for 72 h for further analysis.

In the present study, GEO2R (www.ncbi.nlm.nih.gov/geo/geo2r/↗) was utilized to download the raw counts of the GSE246690↗ series to identify genes differentially expressed across experimental conditions of the SK-MEL-103 cell line (control and palbociclib) [34]. The data was analyzed using the edgeR (v.4.0.16) package from Bioconductor in the R environment (v.4.3.3) [40]. The sva (v. 3.50.0) package was used to remove batch effects and other unwanted variation [41]. Subsequently, differential expression analysis was conducted using the quasi-likelihood negative binomial generalized log-linear model (GLM) functions provided by the edgeR package. For visualization purposes Log2CPM expression values were converted to Z-scores and statistical significance was represented as Fold Discovery Rate (FDR) < 0.05.

Fourteen healthy subjects, comprising 10 female and 4 male, mean age 52.6 years (range 41–66) were recruited in Hospital Clínic de Barcelona (Table 1). Two distinct areas on the forearm were identified for biopsy: one area of photodamaged skin (L01) and one area of less sun-damaged (naturally protected) skin on the inner forearm (L02). These samples were obtained as part of another project led by Dr. Susana Puig following all ethical rules.

Subsequently, two 3-mm punch biopsies were taken on L01 and L02 areas. Each biopsy was included in RNAlater (RNAprotect Tissue Reagent, Qiagen, The Netherlands) and frozen to extract RNA and perform sequencing.

Total RNA was isolated using the RNeasy Mini kit (74104, Qiagen) following the manufacturer’s instructions. One microgram of RNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (4368814, Applied Biosystems, Thermo) with random hexamer primers. A 1:10 dilution of the resulting cDNA was used for quantitative real-time PCR (qPCR) on the Lightcycler 96 instrument (05815916001, Roche) with the Brilliant SYBR® Green QPCR Master Mix (600548, Stratagene, Sigma-Aldrich).

Data was normalized to the expression of the housekeeping gene RPL13, and the difference between treated and untreated cells was calculated as ΔΔCt. For visualization, fold changes were determined using the formula 2-ΔΔCt. The qPCR primers used for mRNA level analysis included HRK Rv-GCTTTCTCCAAGGACACAGGG, Fw-ACCTACTGGCCTTGGCTGTG, and RPL13 Rv- CTCGGGAGGGGTTGGTATTCATC, Fw- ATGGCGGAGGGGCAGGTTCT.

Tumor xenografts from SK-MEL-103 cells were generated as previously described [35]. Briefly, a suspension of 2×10⁵ SK-MEL-103 cells in PBS and matrigel was subcutaneously injected in the flank of athymic nude mice. Once tumors were visible, a group of mice received 100 mg/kg of palbociclib by oral gavage in 50 mM sodium lactate every day for 9 days, and the other group received vehicle. Tumor growth was monitored by caliper measurements and was then extracted and frozen for western blot analysis.

Data were analyzed using GraphPad Prism v.9.3.0 software and was represented as mean ± SEM of independent biological replicates. The statistical significance of the results was analyzed using Student’s t-tail test. *p < 0.05 and **p < 0.01 were considered significant.

Specifically, for RNA-Seq data analysis of patient samples, the R package LIMMA was used for data normalization, specifically employing the Variance Modeling at the Observational Level (VOOM). After normalization of expression data, the genes of interest for validation (CDKN2A, CDKN1A, and HRK) were selected. Global results of the RNA-sequencing analysis are under revision for their publication. Comparisons between the expression between L01 and L02 samples from the same patient were conducted using a paired Student’s t-test. Assumptions of normality and homogeneity of variances were checked and appropriately addressed. All analyses were performed using R version 3.3.0.

All data are available from the corresponding authors upon reasonable request.