Abstract

   Hyperactivation of the phosphatidylinositol-3-kinase (PI3K) pathway is
   one of the most common events in human cancers. Several efforts have
   been made toward the identification of selective PI3K pathway
   inhibitors. However, the success of these molecules has been partially
   limited due to unexpected toxicities, the selection of potentially
   responsive patients, and intrinsic resistance to treatments. Metabolic
   alterations are intimately linked to drug resistance; altered metabolic
   pathways can help cancer cells adapt to continuous drug exposure and
   develop resistant phenotypes. Here we report the metabolic alterations
   underlying the non-small cell lung cancer (NSCLC) cell lines resistant
   to the usual PI3K-mTOR inhibitor BEZ235. In this study, we identified
   that an increased unsaturation degree of lipid species is associated
   with increased plasma membrane fluidity in cells with the resistant
   phenotype and that fatty acid desaturase FADS2 mediates the acquisition
   of chemoresistance. Therefore, new studies focused on reversing drug
   resistance based on membrane lipid modifications should consider the
   contribution of desaturase activity.

   Keywords: non-small-cell lung cancer, BEZ235, drug resistance,
   PI3K/Akt/mTOR pathway, lipid metabolism

1. Introduction

   The PI3K/Akt/mTOR pathway is an important signaling pathway regulating
   several essential cellular functions
   [[42]1,[43]2,[44]3,[45]4,[46]5,[47]6]. It is one of the most
   deregulated pathways in human cancer [[48]6,[49]7,[50]8,[51]9]. In
   particular, activating mutations of PI3K encoding genes (such as
   PIK3CA) or deletions in the negative regulators (such as PTEN) have
   been found in several human tumors, and this results in constitutive
   activation of the pathway, leading to aberrant growth, proliferation,
   and differentiation [[52]10]. Several efforts were made to try to
   specifically target the major components of the pathway. Indeed,
   inhibitors of PI3K, Akt, and mTOR have been extensively studied at the
   preclinical level, and some have been used in clinical settings for in
   different tumor types [[53]11,[54]12,[55]13]. At the preclinical level,
   very promising results were obtained with class-specific PI3K
   inhibitors, pan-PI3K inhibitors, Akt inhibitors, mTOR inhibitors, and
   dual PI3K-mTOR inhibitors, both in hematological and solid tumors
   [[56]11,[57]12]. However, in the clinic the benefits were below
   expectations and mainly restricted to a few tumor types
   [[58]9,[59]13,[60]14,[61]15,[62]16,[63]17]. At present, only two mTOR
   inhibitors (temsirolimus and everolimus) and three PI3K inhibitors
   (idelalisib, copanlisib, and alpelisib), both of which mainly target
   the delta isoform of class I PI3K, have been approved by the FDA and
   are on the market [[64]18,[65]19]. It should be noted that two of the
   three PI3K inhibitors have been approved only for the treatment of
   lymphomas.

   The disappointing clinical results might be partly due to unexpected
   toxicities, a lack of biomarkers able to select patients potentially
   responsive to these therapies, and intrinsic resistance to treatments
   [[66]20,[67]21]. To overcome these challenges, researchers are still
   investigating the pathway and drugs that can inhibit it.

   In this study, we explored the mechanisms associated with resistance to
   PI3K/Akt/mTOR inhibitors in non-small cell lung cancer (NSCLC) cells
   that were resistant to the usual PI3K/mTOR inhibitor BEZ235,
   underscoring biochemical pathways and proposing new interventions that
   may pave the way to overcome resistance. BEZ235 (also known as
   dactolisib) is a potent dual pan-class I (with a particular affinity
   for isoform α) PI3K/mTOR inhibitor belonging to the class of
   imidazoquinoline derivatives. It binds the ATP-binding pocket of the
   p110 subunit of PI3K and to the catalytic site of mTOR, so it is able
   to inhibit both mTORC complexes (mTORC1 and mTORC2), resulting in
   effective inhibition of tumor proliferation and growth
   [[68]22,[69]23,[70]24]. In particular, we investigated changes in
   metabolic pathways in BEZ235 resistant versus sensitive cell lines,
   with mTOR playing a pivotal role in cellular metabolism control
   [[71]25,[72]26,[73]27,[74]28]. It is well established that KRAS-mutated
   NSCLC cells display a distinct metabolic profile [[75]29,[76]30] and
   that KRAS mutations lead to the activation of the MAPK and
   PI3K/Akt7/mTOR pathways [[77]31]. Thus, we also evaluated the metabolic
   changes considering the G12C KRAS mutational status.

2. Methods

2.1. Generation of BEZ235 Resistant Clones and Drug Treatments

   NSCLC H1299 cells overexpressing wild-type (wt) or G12C mutant KRAS
   [[78]29,[79]30,[80]32,[81]33] were used to generate BEZ235-resistant
   clones.

   Wt or KRAS G12C mutant H1299 cells were initially treated with 25 nM
   BEZ235, and the surviving cells were re-plated and re-treated with
   increasing concentrations of the drug. The BEZ235 concentration was
   raised to 500 nM, and after 40 and 38 passages (for H1299 KRAS wt and
   H1299 G12C, respectively), cells were frozen and maintained in culture
   without further drug treatment.

   The activity of BEZ235 was evaluated using the MTS test. Cells
   (sensitive and resistant) were seeded in 96-well plates at the desired
   concentrations. Twenty-four hours after seeded cells were treated with
   increasing concentrations of BEZ235 and maintained at 37 °C for a
   further 72 h. MTS was added (10 µL/well) and absorbance at 490 nm was
   measured using the Glo Max plate reader (Promega, Madison, WI, USA).
   For each concentration, the experiment was repeated for a total of 6
   replicates, and at fixed time intervals (about 3–4 weeks) BEZ235
   activity was tested to verify that the resistant clones maintained
   their behavior.

   The response of sensitive and resistant clones to other drugs was
   evaluated using the same conditions as for BEZ235. The drugs analyzed
   were the pan-PI3K inhibitor BKM120, the isoform-specific PI3K
   inhibitors PIK75, TGX22, and CAL101, the mTOR inhibitors Torin-1 and
   Ink128, the dual-PI3K/mTOR inhibitor PF05212384, the Akt inhibitor
   ARQ-751, and the DNA-damaging agent doxorubicin. The combination
   experiments with BEZ235 and betullin (a sterol regulatory
   element-binding protein (SREBP) inhibitor) were performed by treating
   the cells with increasing concentrations of BEZ235 after maintaining
   the cells for seven days in culture with betullin, or by co-treating
   the cells with the two drugs. All the compounds used were purchased
   from TargetMol (Wellesley Hills, MA, USA).

2.2. Western Blot Analyses

   Proteins were extracted and visualized as reported in [[82]34].
   Separation of nuclear protein extracts from cytoplasmic proteins was
   performed by using a nuclear extraction kit (AbCam, Cambridge, UK)
   following manufacturer’s instructions. Immunoblotting was carried out
   with the following antibodies: anti-p70S6K(Thr389) #9206, anti-p70S6K
   #9202, anti-pAkt (Ser473) #4060, anti-Akt #9272, anti-S6(Ser235/236)
   ribosomal protein #2211, anti-S6 ribosomal protein 2217#,
   anti-4E-BP1(Thr37/46) #2855, anti-4E-BP1 #9644, anti-Acetyl-CoA
   carboxylase (ACC) #3662, anti- Stearoyl-CoA desaturase 1 (SCD1) #2794,
   and anti-Fatty Acid Synthase (FASN) #3189, provided by Cell Signalling
   Technology (Danvers, MA, USA). Anti-Erk #sc94, anti-Erk (Tyr204)
   #sc7383, anti-SREBP1 #sc-365513, anti-actin #sc-47778, and anti-lamin B
   #sc-374015 were obtained from Santa Cruz Biotechnology (Dallas, TX,
   USA).

2.3. Real-Time RT PCR

   Total RNA was extracted and reverse transcribed as reported in
   [[83]29]. The mixture was amplified by a 7900HT Sequence Detection
   System (Thermo Fisher Scientific, Waltham, MA USA). Actin was used as
   the internal control. Primers and TaqMan probes were purchased for all
   genes as ready-to-use solutions (Assay on Demand, Thermo Fisher
   Scientific, Waltham, MA USA). mRNA relative expression levels were
   calculated with the 2-ΔΔCt method. Wt for wt_bez and G12C for G12C_bez
   clones were set to 1. Two samples that showed at least a 2-fold
   difference in expression were considered differently expressed.

2.4. Metabolomics Sample Preparation

   H1299 cell clones were grown for 48 h in biological triplicate. At 48
   h, all four clones were assumed to have the same proliferation rate.
   All of the cell lines (three biological replicates/clone) were rapidly
   rinsed in a saline solution quenched with liquid nitrogen and stored at
   −80 °C until the analysis. Extraction was done as described in
   [[84]29].

2.5. Untargeted Metabolomics

   Flow injection analysis high resolution mass spectrometry (FIA-HRMS)
   was used for untargeted metabolomics of H1299 cell clones as previously
   described [[85]35]. All data were processed and analyzed in Matlab
   R2016a (The Mathworks, Natick, MA, USA) using an in-house developed
   script [[86]35].

2.6. Targeted Metabolomics

   A targeted quantitative approach using a combined direct-flow injection
   and liquid chromatography tandem MS/MS assay (AbsoluteIDQ^® p180 kit,
   Biocrates, Innsbruck, Austria) was applied, as previously published
   [[87]30].

2.7. Metabolic Pathway Analysis

   For biological interpretation of the metabolite dataset, we mapped the
   significant metabolites derived from both untargeted and targeted
   approaches into a biochemical network using MetaboAnalyst
   ([88]www.metaboanalyst.ca, accessed on 15 June 2022). Enrichment
   analysis (EA) tools were used to identify metabolic pathways most
   likely to be associated with the BEZ235 acquired resistance.

2.8. Free Fatty Acids (FFA)

   FFAs (palmitic and oleic acids) were measured in H1299 cells clones
   grown in culture for 48 h (1 × 106 cells) as detailed in [[89]34].

2.9. Free Cholesterol

   Intracellular free cholesterol levels were determined using a 1200
   Series HPLC system (Agilent Technologies, Santa Clara, CA, USA)
   interfaced to an API 5500 triple quadrupole mass spectrometer (Sciex,
   Thornhill, ON, Canada) as described in [[90]36].

2.10. Lipid Droplets Detection

   Cells were seeded at 15,000 cells/mL on glass coverslips. After 48 h,
   cells were fixed with 4% paraformaldehyde, and lipid droplets were
   detected by the Lipid Droplets Fluorescence Assay Kit (CaymanChem,
   AnnArbour, MI, USA), following the manufacturer’s instructions.
   Briefly, the kit is based on Nile Red staining, which is a fluorescent
   dye that is very sensitive and specific for neutral lipids, which
   constitute lipid droplets [[91]37]. Then, nuclei were counterstained
   with DAPI. Fluorescence was visualized at the confocal microscope
   (IX81, Olympus, Shinjuku, Japan).

2.11. Membrane Fluidity Assay

   Cells were seeded in black 96-well plates at a concentration of 50000
   cells/mL and after 48 h membrane fluidity was determined with a
   membrane fluidity kit (AbCam, Cambridge, UK) following the
   manufacturer’s instructions. The assay exploits a peculiar property of
   pyrenedecanoic acid (PDA), which, when added to cells, incorporates
   into the cell’s plasma membrane and forms homodimers, with the rate of
   formation proportional to the membrane fluidity. When dimers form, the
   emission spectrum of the pyrene probe shifts dramatically from 400 nm
   to 470 nm. Briefly, PDA was added to the cells, and they were incubated
   for 1 h at 25 °C in the dark. After unincorporated PDA was removed from
   the samples, fluorescence was read at 400 nm (monomer) and 470 nm
   (dimers) with the GloMax plate reader (Promega, Madison, WI, USA). The
   normalized excimer to monomer fluorescence ratio was then calculated,
   as a quantitative monitoring of the membrane fluidity.

2.12. Statistical Analyses

   Statistical analyses were performed using GraphPad Prism V8 (San Diego,
   CA, USA). The specific tests used in the experiments are reported in
   the figure legends. A p-value < 0.05 was considered statistically
   significant.

3. Results and Discussion

3.1. mTOR, Rather Than PI3K- Related Pathways, Are Mainly Involved in the
Resistance to BEZ235

   To investigate new mechanisms of acquired resistance to PI3K/Akt/mTOR
   axis inhibition, we generated BEZ235-resistant clones starting from our
   NSCLC H1299-derived clones, expressing wt or G12C-mutated KRAS
   [[92]29,[93]30,[94]32,[95]33]. Wt and G12C cells are similarly
   responsive to the dual PI3K and mTOR inhibitor BEZ235, with an IC50 of
   approximately 40-50 nM for both clones. For the induction of
   resistance, the cells were initially treated with 25 nM BEZ235, and the
   surviving cells were re-plated and retreated with the same
   concentration of the drug. After 2 passages, the concentration of
   BEZ235 was increased to 50 nM, then to 100 nM (after 6 passages) to
   arrive at 500 nM after 18 passages. This concentration was maintained
   for 20 additional passages. [96]Figure 1A reports the growth inhibitory
   activity of BEZ235 in parental wt and G12C cells and in resistant
   clones, wt_bez and G12C_bez, obtained after 40 and 38 passages in vitro
   with increasing concentrations of BEZ235, respectively. The activity of
   BEZ235 in resistant cells was determined after at least one further
   passage in vitro without the drug. The activity of BEZ235 was lower in
   both wt_bez and G12C_bez cells compared to their corresponding parental
   cells. For the wt cells, the IC50 of BEZ235 was 52.59 nM and became
   282.4 nM in the wt_bez cells (RI = 5.4). In the G12C cells, BEZ235 had
   an IC50 of 68.14 nM and in the G12C_bez cells of 398.5 nM (RI = 5.8).

Figure 1.

   [97]Figure 1
   [98]Open in a new tab

   (A) Dose-response curve of the clones treated with BEZ235 at different
   concentrations. The means and standard deviation (SD) of at least three
   independent experiments are plotted. Statistical analysis was carried
   out through the two-way ANOVA test and the Bonferroni post hoc test for
   multiple comparisons and is reported in [99]Supplemental Table S3. (B)
   Histograms representing the relative expression of MDR-1 mRNA levels.
   Actin was used as a housekeeping gene for normalization. The means of
   three replicates and SD are reported. (C) Histograms representing the
   relative expression of ABCG2 mRNA levels. Actin was used as a
   housekeeping gene for normalization. The means of three replicates and
   SD are reported. (D) Western blot analysis of MAPK and PI3K/mTOR
   pathway proteins in wt, wt_bez, G12C and G12C_bez clones at basal
   conditions. Ran was used as a loading control. (E–L) Dose-response
   curves of the clones treated with PI3K/Akt/mTOR inhibitors at
   increasing concentrations. On each graph, the specific drug with the
   corresponding target is reported. The means and standard deviation (SD)
   of at least three independent experiments are plotted.

   We first checked for the presence of overexpression of the multidrug
   resistance (MDR-1) and ATP-binding cassette, sub-family G, isoform 2
   (ABCG2) genes, both known to be involved in cell efflux of several
   drugs [[100]38,[101]39,[102]40], by measuring mRNA expression.
   [103]Figure 1B,C clearly show that BEZ235-resistant clones do express,
   at the mRNA level or even lower (in the case of MDR-1), the two genes
   compared to their respective parental cells, thus excluding increased
   expression of MDR-1 and ABCG2 as causes for the resistance. The similar
   viability of the four clones after treatment with increasing
   concentrations of doxorubicin, known to be one of the most studied
   targets of MDR-1 and ABCG2 [[104]41], confirmed that these extrusion
   pumps were not involved in the resistance ([105]Supplementary Figure
   S1A).

   We then decided to analyze at molecular level the expression and
   phosphorylation of proteins belonging to the pathway targeted by
   BEZ235. [106]Figure 1D reports a representative western blot, obtained
   from cell extracts taken from exponentially growing cells. Western blot
   analysis shows a greater phosphorylation of the KRAS downstream protein
   ERK in the mutated KRAS cell lines, G12C and G12C_bez. This increased
   phosphorylation is a consequence of the constitutive activation of the
   KRAS protein because of the presence of the G12C mutation in both
   clones. As far as proteins belonging to PI3K/mTOR pathways, significant
   alterations were observed in the activation of Akt protein, downstream
   PI3K, and p70, a direct target of mTOR, being higher in the BEZ235
   resistant clones. On the other hand, the activation levels of the
   proteins S6, activated by p70, and 4EBP1, directly inhibited by mTOR,
   were not different. We therefore hypothesize that the pathways aimed at
   activating protein synthesis (those of S6 and 4EBP1; [[107]42]) are not
   the main players in BEZ235 resistance mechanisms and that other
   pathways, probably regulated by p70S6K or Akt, might be involved.

   Being BEZ235 a dual PI3K/mTOR inhibitor, we tested whether a collateral
   resistance was observable using drugs acting at different levels in the
   PI3K/Akt/mTOR pathway, as well as which step of the axis was most
   involved in the resistance. [108]Figure 1E–L show the dose-response
   curves obtained from the cytotoxicity experiments performed on the
   clones treated with increasing concentrations of the drugs. Both
   resistant clones were equally sensitive (compared to the parental
   cells) to the pan-PI3K inhibitor BKM120 ([109]Figure 1E) and to the
   alpha isoform-specific PI3K inhibitors ([110]Figure 1F). In the case of
   treatment with CAL-101 ([111]Figure 1G), an inhibitor of PI3Kδ, all the
   clones were resistant. Treatment with TGX-221 ([112]Figure 1H), an
   inhibitor of PI3Kβ, induced a slight increase in resistance in
   resistant BEZ235 cell lines compared to the sensitive ones. The clones
   were also equally sensitive to the pan-Akt inhibitor ARQ751
   ([113]Figure 1I) and all insensitive to the mTORC1 inhibitor rapamycin
   ([114]Figure 1J), but were cross-resistant to the mTORC1 and mTORC2
   inhibitors torin-1 and INK-128 ([115]Figure 1K,L). Collectively, our
   data suggest that the resistance is more related to mTOR than PI3K.

3.2. BEZ235 Resistance Alters the Membrane Lipid Compartment

   mTOR has a role in controlling cell metabolism (e.g., regulation of
   glycolysis, pentose phosphate pathway, and de novo lipogenesis)
   [[116]43]. Furthermore, considering our previous findings about
   metabolic changes associated with the presence of a mutated KRAS in
   isogenic cells [[117]29,[118]30,[119]32], we used both targeted and
   untargeted metabolomics approaches to profile the metabolic asset of
   BEZ235 resistant clones in comparison to their respective parental cell
   lines. We identified the significant alteration of 164 and 133
   metabolites in wt_bez and G12C_bez resistant cells, respectively,
   relative to their sensitive counterparts ([120]Supplemental Table S1).
   Pathway enrichment analysis highlighted the deregulation of metabolites
   mainly involved in the urea cycle and ammonia recycling pathways
   ([121]Supplemental Table S2). This is consistent with the knowledge
   that changes in cellular metabolism not only result in tumor
   progression but also contribute to cancer cells’ resistance to drug
   treatments [[122]44]. Indeed, cancer cells may modulate glutamine
   metabolism and its interconnected ammonia recycling and urea cycle
   pathways, driving drug resistance [[123]45,[124]46,[125]47].

   Among the significantly different metabolites, we observed a strikingly
   higher proportion of lipid species in both resistant clones (105 out of
   164 in wt_bez and 92 out of 133 in G12C_bez) relative to the sensitive
   ones ([126]Figure 2A,B). The untargeted metabolomics approach indicated
   a widespread increment of phosphatidylethanolamines, ceramides, mono-
   and triglyceride species in both wt_bez and G12C resistant clones
   ([127]Supplemental Table S1). Focusing on the quantified lipid species
   in both resistant cells, we confirmed the increased levels of lysoPCs,
   PCs, SMs, and in their saturated (SFA), monounsaturated (MUFA), and
   polyunsaturated (PUFA) forms with no distinction in length (number of
   carbons from 16 to 42) and no relation to the KRAS mutational status
   ([128]Figure 2C,D and [129]Supplemental Table S1).

Figure 2.

   [130]Figure 2
   [131]Open in a new tab

   (A) Pie chart summarizing the metabolic class distribution of the
   statistically significant metabolites between the wt and wt_bez clones.
   (B) Pie chart summarizing the metabolic class distribution of the
   statistically significant metabolites between G12C and G12C_bez clones.
   (C) Sum of intracellular concentrations of lysophosphatidylcholines
   (lysoPC), phosphatidylcholines (PC), and sphingomyelins (SM) in wt,
   wt_bez, G12C, and G12C_bez clones. ** indicates statistically
   significant differences with p < 0.01 (two-way ANOVA and Tukey Kramer
   post-hoc test). (D) Sum and distribution of intracellular
   concentrations of saturated (SFA), monounsaturated (MUFA), and
   polyunsaturated (PUFA) lipid species. ** and *** indicate statistically
   significant differences with p < 0.01 and p < 0.001, respectively
   (two-way ANOVA and Tukey Kramer post-hoc test). (E) Representative
   images of lipid droplets detected by Nile Red fluorescent dye (green)
   in all the clones. DAPI (blue) was used to counterstain nuclei. The
   scale bar is 20.0 µM. (F) Intracellular concentration of free
   cholesterol. (G) Membrane fluidity expressed as relative fluorescence
   units (RFUs) of the ratio between PDA dimer and monomers, detected by a
   membrane fluidity kit (Abcam, see [132]Section 2). * and ** indicate
   statistically significant differences with p < 0.05 and p < 0.01,
   respectively (one-way ANOVA and Bonferroni post-hoc test). LysoPC,
   Lysophosphatidylcholine, PC, Phosphatidylcholine, PE,
   lysophosphatidylethanolamine, CE, Ceramide, SM, Sphingomyelin, MG,
   monoacylglyceride, and TG, Triglyceride.

   In addition to central cellular metabolic adaptation in cancer
   resistance, evidence shows that lipid metabolism is also involved in
   developing cancer resistance [[133]48]. Since we found that
   BEZ235-resistant cells had a higher abundance of lipid species relative
   to sensitive ones, we further investigated whether such an increase was
   due to either the storage and/or biosynthesis of these species in the
   cells. We first analyzed the amount of lipid droplets in sensitive and
   resistant clones. Lipid droplets are intracellular organelles present
   in nearly all cells. Their main function is to store oil-based reserves
   of metabolic energy and components of membrane lipids, which are
   crucial for survival [[134]49]. Using the Nile Red staining procedure,
   we did not find differences in lipid droplets between sensitive and
   resistant cells. In general, G12C cells (both sensitive and resistant)
   showed lower amounts of lipid droplets compared to wt KRAS expressing
   cells ([135]Figure 2E). Therefore, we can exclude the association
   between lipid droplets and BEZ235 resistance.

   Several studies suggest that altered lipid composition and physical
   properties of cell membranes contribute to the chemoresistance of
   cancers [[136]50]. The most common lipid alterations in drug-resistant
   cancer cells are the increased abundance of phospholipids,
   sphingolipids, and cholesterol content and the modulation of the ratio
   between saturated/unsaturated lipid content [[137]50,[138]51]. In our
   setting, we did not observe alterations in the level of free
   cholesterol between sensitive and resistant clones. Free cholesterol is
   primarily incorporated in the plasma membrane and is essential for the
   membranes’ physical integrity; indeed, its membrane enrichment up to
   20–40% results in a more rigid membrane bilayer [[139]52] ([140]Figure
   2F). Saturation status of lipids is also involved in membrane packing
   and fluidity; indeed, modification of the saturated/unsaturated lipid
   ratio results in an alteration of lipid packing in the membrane and
   consequently modulates membrane fluidity, which is a crucial phenomenon
   in cancer drug resistance [[141]44,[142]51]. In agreement with the
   increased amount of MUFA and PUFA, we found that both BEZ235-resistant
   cells displayed a statistically significant increase in plasma membrane
   fluidity compared to parental cells ([143]Figure 2G).

3.3. BEZ235 Resistance Is Linked to Increased Desaturase Expression and
Activity

   We then analyzed the involvement of SREBP1 in de novo lipogenesis and
   eventually in the acquisition of BEZ235 resistance in our models.
   SREBP-1 protein is a transcription factor master regulator of
   lipogenesis [[144]53,[145]54], which acts by activating the expression
   of more than 30 genes involved in the synthesis and uptake of
   cholesterol, fatty acids, triglycerides, and phospholipids
   [[146]55,[147]56,[148]57]. The aberrant activation of this pathway
   leads to the synthesis of fatty acids, which function as essential
   building blocks of membranes. SREBP-1 is synthesized as an inactive
   protein that is activated upon proteolytic cleavage prior to nuclear
   translocation [[149]58]. Furthermore, the nuclear accumulation of
   active SREBP-1 requires the activity of the mTORC1 complex [[150]59].
   We performed western blot analysis to assess SREBP-1 protein levels on
   both total ([151]Figure 3A) and nuclear ([152]Figure 3B) extracts from
   sensitive and resistant cells, but we did not observe any difference
   that could explain its role in BEZ235 resistance. Treatments with
   different schedules (co-treatment, pre-treatment for seven days) and
   concentrations (ranging from 1 μM to 10 μM, data not shown) of
   betullin, a SREBP1 inhibitor, were in fact not able to restore the
   sensitivity to BEZ235 both in wt_bez and G12C_bez clones. As such, it
   is unlikely that SREBP-1 might have a role in driving resistance in our
   model ([153]Supplementary Figure S1B,C). However, in mammalian cells,
   three different SREBP isoforms, SREBP-1a, SREBP-1c, and SREBP-2, have
   been identified, having only partially overlapping roles in the
   regulation of fatty acid and cholesterol synthesis [[154]60,[155]61].
   In particular, SREBP-1c is mainly involved in modulating fatty acid
   synthesis and de novo lipogenesis, while SREBP-1a takes part in two
   pathways, SREBP-1c and SREBP-2, specific to cholesterol metabolism
   [[156]62]. Therefore, we will analyze the different isoforms in future
   experiments to finally exclude the involvement of SREBP1 (and in
   particular SREBP1-c) in the resistance observed.

Figure 3.

   [157]Figure 3
   [158]Open in a new tab

   (A) Western blot analysis of SREBP1 protein expression in wt, wt_bez,
   G12C, and G12C_bez clones at basal conditions. Ran was used as a
   loading control. (B) Western blot analysis of SREBP1 protein expression
   in nuclear (left) and cytosolic (right) extracts of wt, wt_bez, G12C,
   and G12C_bez clones at basal conditions. Actin was used as a loading
   control. Lamin B was used as a nuclear extract marker. (C) Western blot
   analysis of ACC, FASN, and SCD1 protein expression in wt, wt_bez, G12C,
   and G12C_bez clones at basal conditions. Ran was used as a loading
   control. (D) Intracellular levels of Palmitic and Oleic acid in wt,
   wt_bez, G12C, and G12C_bez clones. (E) Histograms representing relative
   mRNA expression of the ELOVL2 and ELOVL5 genes. Actin was used as a
   housekeeping gene for normalization. The means of three replicates and
   SD are reported. (F) Histograms representing relative mRNA expression
   of the FADS1 and FADS2 genes. Actin was used as a housekeeping gene for
   normalization. The means of three replicates and SD are reported. (G)
   Estimated desaturase activity in the clones based on MUFA/SFA and
   PUFA/SFA ratio. **, ***, and **** highlight statistically significant
   differences with a p < 0.01, p < 0.001, and p < 0.0001, respectively
   (one-way ANOVA and Tukey Kramer post-hoc test).

   We also examined whether increased lipid abundance in resistant clones
   was due to changes in the key players of de novo lipogenesis by
   monitoring the expression and protein abundance of the lipogenic
   enzymes Acetyl-CoA carboxylase isoforms 1 and 2 (ACC1, ACC2, coded by
   the ACACA and ACACB genes, respectively), Fatty Acid Synthase (FASN),
   and Stearoyl-CoA desaturase 1 (SCD1) [[159]63]. As reported in
   [160]Figure 3C, we did not assist to an increase in their protein
   levels in BEZ235 resistant clones, but in G12C_bez cells rather to a
   decrease in the expression of SCD1. Accordingly, real-time PCR
   confirmed the absence of any increased expression of these enzymes.
   There was actually a trend toward a reduction in the expression of all
   the enzymes in BEZ235 clones compared to their parental counterparts
   ([161]Supplementary Figure S1D). Consistent with the absence of
   alteration in de novo lipogenesis, the BEZ235 resistant clones had
   comparable levels of the main products of this pathway
   [[162]64,[163]65], the palmitic (product of FASN) and oleic (product of
   SCD1) free fatty acids ([164]Figure 3D).

   The results obtained so far on the regulators, enzymes, and products
   related to the de novo lipogenesis process do not explain the increase
   in MUFA and PUFA observed in the clones that are stably resistant to
   BEZ235.

   The phospholipid fatty acid chain length and the number and position of
   double bonds can markedly influence membrane fluidity, permeability,
   and stability [[165]66]. Consequently, we evaluated the contribution of
   fatty acid desaturation and chain elongation in exerting the lipid
   modifications observed in our resistant clones by measuring the
   expression of fatty acid elongases 2 and 5 (ELOVL2 and ELOVL5) and
   fatty acid desaturases 1 and 2 (FADS1 e FADS2), responsible for the
   elongation and desaturation of fatty acid lipid chains
   [[166]67,[167]68].

   The analysis showed similar expression levels of the elongases ELOVL2
   and ELOVL5 and the desaturase FADS1 in all clones ([168]Figure 3E,F).
   Instead, a significant overexpression of the FADS2 enzyme is observed
   in both BEZ235-resistant clones compared to sensitive ones ([169]Figure
   3F). Accordingly, the ratio of MUFA/SFA and PUFA/SFA, an index of
   desaturase activity, was statistically significantly increased in
   BEZ235-resistant cells ([170]Figure 3G).

   FADS2 mainly plays a role in desaturation by introducing a double bond
   at the Δ6 position of the fatty acid chain, and this is the first
   rate-limiting enzyme for the conversion of upstream fatty acids into
   PUFAs [[171]69,[172]70]. PUFAs are critical for the regulation of
   membrane biophysical properties and the functions of transmembrane
   proteins and peripheral membrane proteins [[173]59,[174]65]. In
   accordance with our observations, the increased FADS2 expression is
   likely linked to the increased PUFA levels and membrane fluidity
   observed in the BEZ235 resistant clones.

4. Conclusions

   The PI3K pathway is the most frequently activated pathway in human
   cancers. Consequently, several compounds targeting the various nodes of
   this pathway have been developed. However, the majority of these
   compounds have been unsuccessful in patients due to high levels of
   toxicity as well as their inability to downregulate the pathway, which
   is partially mediated by the intrinsic adaptive response of tumors. The
   PI3K pathway is a key pathway regulating cell metabolism. Therefore, we
   focused on the characterization of the metabolic changes underpinning
   BEZ235 resistance in NSCLC cells. The observed general central cellular
   metabolic adaptation in BEZ235 resistant clones comes as no surprise,
   but we interestingly identified the central involvement of lipid
   modulation, in which the key contribution of FADS2-mediated fatty acid
   desaturation in increasing membrane fluidity is the peculiar feature of
   our BEZ235 resistance model. Understanding the role of the plasma
   membrane is pivotal to overcoming drug resistance. The interplay
   between the membrane fluidity and the efficacy of anticancer drugs is
   not restricted to the function of the membrane as a barrier to drug
   penetration but includes more complex mechanisms that the cell
   implements to maintain the homeostasis of the membrane’s physics to
   adapt to continued anticancer drug exposure. The precise molecular
   mechanisms linking BEZ235 resistance to the FADS2 enzyme, as well as
   how interfering with desaturase activity and possibly membrane fluidity
   may counteract BEZ235-acquired resistance, remain unknown.

Acknowledgments