Caco-2 cells, derived from a colon adenocarcinoma, were obtained from DSMZ and were cultured from passage 32 to 39 in DMEM containing 4.5 g/L glucose (Gibco) supplemented with 2 mM L-glutamine, 1 mM sodium pyruvate, 1% (v/v) non-essential amino acids (NEAA), 1% (v/v) penicillin/streptomycin and 10% (v/v) heat-inactivated FBS in an atmosphere of 90% air/10% CO₂ (v/v) at 37°C. Cells were negatively tested for mycoplasma contamination prior to the study using the MycoAlert detection kit (Lonza). 6-, 48- and 96-well plates were used for cell growth and each well was systematically coated with 1% of sterile type-I collagen (Sigma, C8919) for 1 h at 37°C prior to seeding. The culture medium was changed every 2–3 days. All the fatty acids used in the experiments were purchased from Larodan: stearic acid (STE, 10-1800), oleic acid (OLE, 10-1801), alpha-linolenic acid (ALA, 10-1803), docosahexaenoic acid (DHA, 10-2206), punicic acid (PunA, 10-1875), jacaric acid (JA, 10-1876) and alpha-eleostearic acid (αESA, 10-1871). The fatty acids used to treat the cells were conjugated to albumin in a 4:1 ratio (w/w) and dissolved in phosphate buffer saline before being stored at −80°C in aliquots. Ferrostatin-1 (fer-1, Sigma Aldrich, SML0583) and RSL3 (Selleck Chemicals, S8155) were dissolved in DMSO. Cell viability assessment was performed using PrestoBlue™ cell viability reagent (Fisher Scientific, 12083745).

For cell differentiation related viability assays, cells were cultured in 48-well plates and seeded at a density of 60,000 cells/well. The timeline for cell differentiation started 24 h after seeding. From day 0 to day 21 of cell differentiation, cells were treated every 2 or 3 days with 7 different fatty acids, varying in degree of unsaturation, at a concentration of 100 µM. Three additional conditions combining fer-1 at 10 µM and one of the 3 CLnAs at 100 µM were used for the assay. For dose-response related cell viability assays, cells were cultured in 96-well plates at a seeding density of 27,000 cells/well to preserve the surface ratio with cells grown in 48-well plates. Cells were treated at day 2 of cell differentiation with increasing concentrations of PunA, JA or αESA or with the canonical ferroptotic inhibitor RSL3. For both experiments, after 72 h of treatment, cell viability was assessed with PrestoBlue™ reagent, which assesses the amount of living cells based on their capability to reduce non fluorescent blue resazurin contained in the reagent into resorufin, an intensively red-fluorescent dye according to manufacturer’s instructions.

Cells were seeded at a density of 15,000 cells/well in 96-well black/clear bottom plates (Greiner Bio One, 655090). Cells were treated with the 7 different fatty acids at a concentration of 100 μM, the 3 CLnAs being applied on cells either with or without fer-1 10 µM. Lipid peroxidation was measured after 12 h by adding 5 µM of the lipid peroxidation sensor C11-BODIPY 581/591 (4,4-difluoro-5-(4-phenyl-1,3-butadienyl)-4bora-3a,4a-diaza-s-indacene-3-undecanoic acid, Fisher Scientific, D3861). The probe contains 2 double bonds, which are oxidized in the presence of lipid ROS, thereby inducing a shift in fluorescence from red (excitation = 580 nm, emission = 620 nm) to green (excitation = 500 nm, emission = 540 nm) that was measured using the fluorometer Infinite Mpro200 plate reader (Tecan). Relative fluorescence, obtained upon division of the green fluorescence by the red fluorescence was used to measure the intracellular peroxidation level. The value obtained was then further divided by the value of the control condition to obtain a fold change expression.

Cells were seeded at a density of 480,000 cells/well in a 6-well plate. Cells were incubated with 100 µM of each of the 7 fatty acids for 24 h. The dose was selected based on the results from dose-response curves and the timing was chosen in order to prevent cell death from occurring at the time of cell collection. After 24 h, cells were washed twice with ice-cold PBS before being lysed in a RIPA buffer supplemented with a protease inhibitor cocktail for 5 min. The cell lysates were then collected and centrifuged at 16,000 rcf for 10 min. Finally, the supernatant was retrieved and stored at −80°C until bicinchoninic acid protein assay (Thermo Fisher Scientific, 23225) and Western blot were performed. Loadings were prepared by mixing a defined volume of sample to achieve 20 µg of proteins with an equal volume of Laemmli buffer. Samples were denatured for 5 min at 95°C before being loaded on a 10% SDS-PAGE gel. After 1 h migration at 120 V, the gel was transferred in a Mini TransBlot electrophoretic Transfer Cell using a PVDF membrane (VWR, NEF1002001PK), previously activated in methanol, for 45 min at a constant intensity of 350 mA. Membranes were then blocked in 5% skimmed milk-containing TTBS for 1 h at room temperature (RT) under constant shaking. Afterwards, the membranes were incubated at 4°C overnight under constant shaking with the following primary antibodies, diluted in 5% skimmed milk-containing TTBS: anti-ACSL4 (Abcam, ab110007, 1:1000), anti-Cyclin A2 (Cell Signaling Technology, 67955, 1:1000), anti-DPP4 (Cell Signaling Technology, 67138, 1:1000), anti-GPX4 (Abcam, ab125066, 1:1000), anti-β-Actin (Merck, A5441, 1:10000), anti-VCP (Cell Signaling Technology, 2648, 1:1000). The next day, membranes were washed 3 times in TTBS and were incubated with horseradish peroxidase-conjugated secondary antibodies (Cell Signaling Technology, anti-rabbit 7074, anti-mouse 7076) in 5% skimmed milk-containing TTBS for 1 h at RT. Finally, the secondary antibody was discarded, membranes were washed 3 times in TTBS and the chemiluminescent signal was read after addition of Amersham ECL detection reagent (VWR, RPN2134) with a chemiluminescence Amershan Imager 600 (Cytiva).

GraphPad Prism 9.1 software was used to perform statistical analysis by using either one-way or two-way ANOVA with Sidak’s multiple comparison tests. Statistical significance to the control or to another treatment is determined as follows: *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001; ****p ≤ 0.0001, with p being the p-value adjusted for multiple comparisons.

The viability of Caco-2 cells was assessed after 2 days of cell culture (referred to as day 2) when treated with increasing doses of three different CLnAs, namely, PunA, JA or αESA, or of the ferroptosis inducer RSL3 (Figure 1). All three CLnAs were cytotoxic to Caco-2 cells, with lethal doses LD50 of 23.99 ± 5.42, 24.75 ± 5.34 and 43.34 ± 8.68 µM for PunA, JA and αESA, respectively (Figures 1A–C). Interestingly, the addition of oleic acid (OLE), a monounsaturated fatty acid with anti-ferroptotic potential (Magtanong et al., 2019), did not significantly impact JA cytotoxicity on Caco-2 cells (Supplementary Figure S1), supporting its strong effect as pro-ferroptotic fatty acid. Similarly to CLnAs, RSL3 was also cytotoxic to Caco-2 cells, with an LD50 of 65.59 ± 6.95 µM (Figure 1D). Of note, Caco-2 cells at day 2 appeared to be relatively resistant to RSL3-driven ferroptosis when compared to other colorectal cancer cell lines (Doll et al., 2017; Xie et al., 2017; Doll et al., 2019). It should however be emphasized that the latter studies were conducted on cancer cells at high proliferation rates (i.e., corresponding to the day 0 in the present study) (Xie et al., 2017), while our results were obtained on Caco-2 cells already engaged in the process of cell differentiation (i.e., day 2) and thus at an already reduced proliferation rate.

Caco-2 cells were cultured over 21 days in order to differentiate into enterocytes and were exposed to different fatty acids with various degrees of unsaturation and chain lengths every second or third day. Over the course of their differentiation, Caco-2 cells progressively developed resistance to the cytotoxicity of all three CLnAs (Figure 2). When exposed to 100 µM PunA, the viability of Caco-2 cells (that was null upon treatment at day 2) amounted to 40% at day 6% and 65% at days 9 to finally reach 85% at day 21, thereby recovering a viability close to the control condition (Figure 2A). Similar trends were observed for the two other CLnAs. The relative viability of Caco-2 cells treated with JA rose from 0% at day 2 to about 70% at days 9 and to more than 85% at day 21 (Figure 2B) while αESA exhibited the strongest and fastest recovery in cell viability according to the time of treatment, going from 8% at day 2%–80% at day 9, and reaching more than 95% at day 21 (Figure 2C). Importantly, the addition of ferrostatin-1 (fer-1), a ferroptosis inhibitor, prevented the cytotoxicity of CLnAs at all days of cell culture (Figures 2A–C), thereby confirming ferroptosis as the cell death pathway induced by CLnAs in Caco-2 cells. In comparison, the viability of Caco-2 cells exposed to 100 µM RSL3 remained impacted until 17 days of cell differentiation, since it rose from 0% at day 2 to about 80% at day 17, thus at a later time of differentiation compared to CLnAs (Figure 2D). Surprisingly, the addition of fer-1 did not prevent RSL3 cytotoxicity, probably due to the large dose differences of the two compounds (i.e. 100 μM of RSL3 and only 10 µM of fer-1). This suggests that targeting GPX4 as a major antioxidant system is more detrimental to differentiated cells than influencing cellular lipid composition with CLnAs. Interestingly, none of the other tested fatty acids, namely, STE, OLE, ALA and DHA, was cytotoxic to Caco-2 cells at any time of cell culture (Figure 2D). These results are in line with previous reports on the effect of these fatty acids on cancer cells (Magtanong et al., 2019; Beatty et al., 2021; Dierge et al., 2021) and support the specific pro-ferroptotic potential of CLnAs. These data also suggest that along their differentiation, enterocyte-like Caco-2 cells progressively develop ferroptosis resistance mechanisms that prevent CLnA cytotoxicity.

We next investigated the effect of CLnAs on lipid peroxidation levels in differentiating Caco-2 cells by using the C11 BODIPY assay (Dierge et al., 2021; Vermonden et al., 2021). We cultured Caco-2 cells for either 2, 9 or 16 days and exposed cells for 12 h to the following conditions: RSL3, STE, OLE, ALA, DHA or a CLnA, either alone or combined with fer-1. At day 2, PunA, JA and αESA induced respectively a significant 4.9-, 7.3- and 4.1-fold increase in lipid peroxidation compared to the control (Figure 3A). These increases in lipid peroxidation were inhibited by the addition of fer-1, further confirming that CLnAs induce ferroptosis through lipid peroxidation in undifferentiated Caco-2 cells. None of the other tested fatty acids impacted the level of lipid peroxidation compared to the control (Figure 3A). Of note, RSL3 at the dose tested (i.e. 1 μM) did not induce changes in lipid peroxidation in Caco-2 cells at day 2 (Figure 3A). More importantly, CLnA-induced lipid peroxidation was completely abolished in Caco-2 cells at day 9 compared to day 2 (Figure 3B). As a result, at day 9, the level of lipid peroxidation in Caco-2 cells treated with a CLnA was not significantly different from the amounts measured at day 2 in the presence of fer-1. None of the other tested fatty acids induced a change in lipid peroxidation between day 2 and day 9 of cell culture (Figure 3B; Supplementary Figure S2). Intracellular lipid peroxidation was slightly but significantly increased in Caco-2 cells treated with either CLnA on day 16 as compared to day 9 (Figure 3B). This difference between days 9 and 16 may be due to the establishment of senescence in long-term cultured Caco-2 cells, which is linked to an overall accumulation of intracellular ROS (Sies and Jones, 2020) and a possible increase in CLnA-induced peroxidation. This hypothesis is supported by a visible increasing trend in the abundance of the p21 senescence marker protein (McHugh and Gil, 2018) in Caco-2 cells at day 16 of their differentiation compared to day 2 (Supplementary Figure S3). Still, altogether our findings support that differentiating enterocytes progressively acquire resistance to CLnA-induced lipid peroxidation and subsequent ferroptosis.

Our results indicate that Caco-2 cells progressively develop CLnA-induced ferroptosis resistance throughout their differentiation in enterocytes. We therefore assumed that resistance to CLnA cytotoxicity would correlate positively with differentiation markers and negatively with proliferation markers. We performed immunoblots on Caco-2 cells cultured for 2, 9 or 16 days. We selected cyclin A, a major regulator of cell cycle and division (Deshpande et al., 2005; Natoli et al., 2011), as a marker of cell proliferation, and dipeptidyl peptidase 4 (DPP4), which is an enzyme expressed at the membrane of differentiated enterocytes (Yoshioka et al., 1991), as a marker of cell differentiation. The abundance of cyclin A significantly dropped at days 9 and 16 of culture of Caco-2 cells, with a 2.5-fold decrease from day 9 compared to day 2 (Figure 4A). On the contrary, DPP4 level in Caco-2 cells increased after 9 days of differentiation and was further upregulated after 16 days of differentiation, with a 2.2-fold rise in abundance at day 16 compared to day 2 (Figure 4B). Taken together, these data confirm that the onset of CLnA-induced ferroptosis resistance coincides with the shift of Caco-2 cells from a proliferative to a differentiated state.

In recent years, multiple studies have highlighted the major role of GPX4 as ferroptosis suppressor, as demonstrated by the induction of ferroptosis when GPX4 is blocked with specific pharmacological inhibitors or by genetic deletion (Seiler et al., 2008; Yang et al., 2014; Hangauer et al., 2017; Viswanathan et al., 2017; Zou et al., 2019). The main role of GPX4 is to protect cells against the deleterious accumulation of lipid hydroperoxides (Seiler et al., 2008; Yang et al., 2014). Given its major role in ferroptosis, we investigated whether GPX4 is involved in the development of the resistance of differentiating Caco-2 cells to CLnA cytotoxicity. Caco-2 cells showed a progressive increase in GPX4 abundance throughout their differentiation, with a significant 2.4-fold rise at day 16 compared with day 2 (Figure 4C). This increase was similar when Caco-2 cells were exposed to PunA or ALA, showing a 2.6-fold rise in GPX4 abundance at day 16 compared to day 2 (Figure 4C). In fact, when treated with any of the tested fatty acids, GPX4 levels remained unchanged compared to the control, both at day 2 (Figure 4D) and at day 16 of cell differentiation (Figure 4E), indicating that the observed upregulation of GPX4 was not induced in response to CLnAs. Interestingly, while the viability of Caco-2 cells treated with a CLnA did not significantly increase between days 9 and 16 (Figure 2A), GPX4 abundance continuously rose over cell differentiation (Figure 4E), suggesting that lipid peroxide detoxification is already fully insured with a sub-maximal amount of GPX4 in Caco-2 cells. More importantly, Caco-2 cells at day 16 showed again CLnA sensitivity when co-treated with JA and a GPX4 inhibitor (i.e., RSL3) (Supplementary Figure S4A), further emphasizing the major role of GPX4 in protecting Caco-2 cells against CLnA cytotoxicity. Of note, we also assessed changes in the expression of another ferroptosis marker, ACSL4, which preferentially activates PUFAs in their CoA form to incorporate them in phospholipids and which was found to be a ferroptosis initiator (Doll et al., 2017). However, ACSL4 levels remained unchanged between days 2, 9 and 16 of Caco-2 culture (Supplementary Figure S4B), suggesting that ACSL4 is not involved in the development of Caco-2 resistance to CLnA-induced ferroptosis. On the whole, our results indicate that while proliferating Caco-2 cells are sensitive to ferroptosis induced by CLnAs, Caco-2 cells progressively increase their GPX4 protein levels throughout their differentiation in enterocytes, thereby acquiring the detoxification mechanism necessary to prevent ferroptosis.