ACSL3 knockdown impairs extracellularly derived FAs activation and reduces PDAC cell proliferation
ACSL3 is overexpressed in human and mouse pancreatic cancer tissue compared to healthy pancreas [10]. To find out to which extent the expression of ACSL3 in PDAC cell lines agrees with patient samples, we assessed the mRNA and protein levels of ACSL3 in a panel of 6 KRAS mutant human PDAC cell lines, namely MiaPaCa-2, Hs766T, CFPAC1, SU86.86, AsPC-1 and PANC-1. We found that all PDAC cell lines evidenced higher ACSL3 mRNA and protein levels compared to immortalized human pancreatic duct epithelial cells (HPDE), which derive from healthy pancreas (Fig. 1A and B) [16].
We previously provided evidence that ACSL3 mediates the activation and retention of extracellularly derived FAs in KRAS mutant lung cancer cells [9]. Accordingly, RNAi-mediated knockdown of ACSL3 in HPDE-iKRASG12D cells, which are engineered to carry a doxycycline–inducible oncogenic KRASG12D expression (thereafter named HPDEK) (Fig. 1C) [12], leads to a reduction of extracellularly derived FAs activation (Fig. 1D). This is evidenced by the reduced retention of extracellularly supplied FAs labelled with the fluorescent FA BODIPY 500/510 C4,9 (Bodipy-FA), which mimics a long-chain FA (Fig. 1D). Accordingly, staining of lipid droplets (LDs) with LipidTOX, a LD marker, and analysis by flow cytometry showed a reduction in the deposition of lipids in LDs upon ACSL3 knockdown in PDAC cell lines, confirming a decrease in lipid retention (Fig. 1E).
We then aimed to evaluate whether suppression of ACSL3 affects human pancreatic cancer cell proliferation. To this end, we knocked down ACSL3 with 2 different shRNAs and measured the relative cell number 72 h later (Fig. 1F). We found 50% reduction of relative cell number compared to the control in five out of six cell lines (Hs766T, CFPAC1, SU86.86, AsPC-1 and PANC-1) and a 90% reduction in MiaPaCa-2 cells. Moreover, immunoblot for poly ADP-ribose polymerase (PARP) revealed a marked increase in cleaved PARP, a marker of cell death, only in MiaPaCa-2 cells, while for all the other cell lines the increase was mild or undetectable (Fig. 1G). This disparate effect in cell death upon ACSL3 knockdown was also confirmed by measuring caspase-3 activity upon ACSL3 knockdown in 2 representative cell lines MiaPaCa-2 and PANC-1 (Fig. 1H).
In order to assess whether ACSL3 is an exclusive vulnerability of KRAS mutant cells, we performed cell proliferation assays in HPDE and HPDEK cells. Interestingly, ACSL3 knockdown in HPDE cells showed a modest reduction in cell number, while the number of KRASG12D expressing HPDEK cells was drastically reduced (Fig. 1I). The anti-proliferative effect in HPDEK cells was also associated with an increase in caspase-3 activity, while no effect was observed in HPDE cells (Fig. 1J).
Our results indicate that in pancreatic cancer cells ACSL3 loss-of-function impairs extracellularly derived FAs cellular retention, reduces cell proliferation and variably induces cell death.
Restriction of extracellularly derived FAs recapitulates the impact of ACSL3 suppression on PDAC cell proliferation
The reduced proliferation of PDAC cells upon ACSL3 suppression could be caused by an impaired cellular retention of extracellularly derived FAs. To directly test this possibility, we assessed the impact of extracellular lipid restriction in KRAS mutant cancer cells. For this purpose, we incubated PDAC cell lines with media containing normal or lipid depleted serum. We used 2 different human KRAS mutant PDAC cell lines, AsPC-1 and PANC-1, as well as HPDE and HPDEK cells. Similar to ACSL3 suppression, switch from normal to lipid depleted media caused a reduction in cell proliferation of AsPC-1, PANC-1 and HPDEK cells lines, but not of HPDE (lacking mutant KRAS) cells (Fig. 2A and B). Notably, serum lipid depletion did not affect the protein levels of ACSL3 or other ACSL isoenzymes, excluding the possibility that the reduced cell proliferation could be caused by changes in ACSL isoforms (Supplementary Fig. 1A and 1B).
Treatment of AsPC-1 and PANC-1 cells with media containing delipidated serum caused a mild increase in caspase-3 activity (1.2-fold) compared to the normal serum-treated cells (Fig. 2C). Furthermore, assessment of apoptosis by flow cytometry revealed no change in the percentage of Annexin V/propidium iodide (ANN/PI)-positive cells, confirming that lipid depletion is not sufficient to trigger a detectable apoptotic cell death (Supplementary Fig. 1C). Notably, the treatment with delipidated serum induced a 2-fold increase in caspase-3 activity of HPDEK cells (Fig. 2C), and a similar change in the percentage of ANN/PI-positive cells, indicating increased sensitivity to serum lipid depletion compared to PDAC cell lines (Supplementary Fig. 1D).
Taken together, these results suggest that the extracellular lipid depletion reduces cell proliferation in all KRAS mutant cancer cell lines, while causing only a moderate increase in cell death in PDAC cells.
ACSL3 suppression or serum lipid depletion increase the autophagic flux of PDAC cells
The lack of an obvious cell death induction in all PDAC cell lines tested upon ACSL3 suppression or serum lipid depletion suggests the insurgence of possible mechanisms that provide a source of FAs to replenish the missing lipids. A potential candidate mechanism that could recycle cellular components, in order to provide lipids, is autophagy [17, 18]. Indeed, there is strong evidence indicating that autophagy acts in regulating the turnover of lipids from intracellular lipid stores (macrolipophagy) and consequently, inhibition of autophagy increases lipid storage in lipid droplets (LDs) [17]. To assess whether the autophagic flux status impacts LDs turnover in pancreatic cancer cells, we inhibited autophagy with HCQ, an inhibitor of autophagosome–lysosomal fusion and measured the abundance of LDs by staining of HPDEK cells with BODIPY 493/503. We found that inhibition of autophagy increased the accumulation of lipids in LDs, suggesting that autophagy acts to maintain a high LD turnover in pancreatic cancer cells under basal conditions (Fig. 3A).
To evaluate whether ACSL3 suppression leads to an increase in autophagic flux, we knocked down ACSL3 with 2 different shRNAs in HPDEK cells and transduced them with a plasmid expressing mCherry-EGFP-LC3B, which is a tandem reporter for autophagosomes. Due to the low pH of lysosomes, the fluorescence of EGFP is quenched whereas the mCherry remains stable. Therefore, an increase in autophagic flux is manifested as quenching of the EGFP fluorescence and the autophagic index is determined by the mCherry/EGFP ratio [19, 20]. Confocal microscopy evidenced a significant induction of the autophagic index upon ACSL3 knockdown that was similar to serum starvation, the positive control for autophagy induction (Fig. 3B and C). Furthermore, imaging flow cytometry analysis upon treatment of mCherry-GFP-LC3B transduced HPDEK cells with Triacsin C, an ACSL3 inhibitor, revealed a reduction of EGFP-positive cells indicative of increased autophagy (Fig. 3D). These results were also confirmed by assessing the autophagy markers LC3B-II and p62. Specifically, immunoblot analysis revealed a reduction in p62 (indicative of increased p62 degradation) and an increase in LC3B-II upon ACSL3 knockdown, indicating increased autophagic flux (Supplementary Fig. 2A). Moreover, LC3B-II increased upon ACSL3 knockdown also in MiaPaCa-2 and PANC-1 cell lines (Supplementary Fig. 2B).
Next, we assessed the induction of autophagy upon serum lipid depletion. Immunoblot analysis evidenced a marked induction of autophagy in AsPC-1 and PANC-1 cell lines as shown by an increase in LC3B-II protein level (Fig. 3E). Since an increase in LC3B-II could also be caused by a block in autophagosomal maturation [21], we inhibited autophagy with HCQ, which blocks autophagosome–lysosomal fusion. In the presence of HCQ, serum lipid depletion caused a further increase in LC3B-II compared to the single treatments (serum lipid depletion alone and HCQ alone), confirming increased autophagic flux, rather than a block in autophagosomal maturation (Fig. 3E).
To prove that it is the depletion of extracellularly derived lipid substrates of ACSL3 that trigger autophagy upon serum lipid depletion, we assessed whether the addition of oleic acid (a lipid substrate of ACSL3) to the cell culture media could rescue serum lipid depletion-induced autophagy. We found that the addition of oleic acid rescued serum lipid depletion-induced autophagy in the presence of ACSL3, but not upon ACSL3 knockdown (Fig. 3F). These data confirm that it is the depletion of extracellularly derived FAs that trigger autophagy and that ACSL3 is important for their activation and retention.
Taken together, our results suggest that the reduction of extracellularly derived lipids, either by serum lipid depletion or ACSL3 inhibition, triggers an increase in autophagic flux, possibly to replenish cancer cells with the necessary lipids.
Acsl3 deletion enhances autophagy in KPC tumors
In order to assess whether the depletion of extracellularly derived lipids, mediated by ACSL3 suppression, triggers autophagy also in vivo, we generated mice carrying a transgene allowing tamoxifen-dependent expression of Cre recombinase under the pancreas-specific Pdx-1 promoter (Pdx1-CreERT2), a Cre-activatable KrasG12D allele (LSL-KrasG12D/+), homozygous for a Cre-conditional p53 knockout allele (p53floxflox), and either wild type or homozygous for an Acsl3 knockout allele (Acsl3+/+ or Acsl3−/−, respectively), to generate 2 experimental groups: LSL-KrasG12D/+;p53flox/flox;Pdx1-CreERT2;Acsl3+/+ and LSL-KrasG12D/+; p53flox/flox;Pdx1-CreERT2;Acsl3−/− (hereafter named KPC;Acsl3+/+ and KPC;Acsl3−/−, respectively). Tamoxifen administration to these mice drives the Cre-mediated recombination and results in the excision of the loxP-flanked stop codon (LSL), thereby leading to conditional expression of KrasG12D and deletion of p53 in the pancreas of the mice. The relevant aspect of the KPC mouse model is the spontaneous generation of tumors which faithfully recapitulate the whole spectrum of lesion grades already described in humans [22]. As previously reported, the KPC;Acsl3−/− mice were born according to the expected Mendelian ratio and without obvious macroscopic defects during development or adult life compared to the KPC;Acsl3+/+ mice [10].
Next, we assessed whether KPC;Acsl3−/− tumors exhibited higher basal autophagy levels compared to KPC;Acsl3+/+. Therefore, we performed immunoblot analysis to assess the degradation of p62 on pancreatic cancer tissue from KPC;Acsl3+/+ and KPC;Acsl3−/− mice. Interestingly, we found reduced p62, indicative of an increased autophagic flux (Fig. 4A). Immunofluorescence (IF) staining of LC3B and p62 on paraffin-embedded tissue from KPC;Acsl3+/+ and KPC;Acsl3−/− mice followed by computer-assisted quantification revealed a LC3B-positive punctae increase and p62 decrease in PanINs and PDAC tumors of KPC;Acsl3−/− mice (Fig. 4B and C). Interestingly, the number of LC3B-positive punctae increased progressively from healthy to PanIN and PDAC lesions (Fig. 4C), confirming previous observations that human PDAC exhibits high levels of basal autophagy [23,24,25]. Even though not exhaustive in determining flux variations over time, studies involving human patient material with similar LC3B/p62 patterns are interpreted as bona fide evidence of increased autophagy [21, 26].
Lipid depletion sensitizes PDAC cells to autophagy inhibition in vitro
Our data link lipid depletion (by serum lipid depletion or ACSL3 knockdown) with reduced proliferation, differential induction of cell death, and concomitant increase in autophagic flux, suggesting that autophagy may play a pro-survival role in this context. Therefore, we aimed to assess whether autophagy inhibition could further enhance the anti-proliferative activity of ACSL3 inhibition. For this purpose, we treated HPDEK cells with increasing concentrations of Triacsin C and/or HCQ and measured cell proliferation. We found that combining ACSL3 with autophagy inhibition cooperated in suppressing cell proliferation in HPDEK cells compared to the single treatments (Fig. 5A).
Next, we employed the HPDEK cells previously transduced with a shRNA against ACSL3 and treated them with HCQ or bafilomycin A1 to inhibit autophagy [27]. We found that ACSL3 knockdown cooperated with HCQ or bafilomycin A1 in suppressing the proliferation of HPDEK cells (Fig. 5B and C). Moreover, the combination of HCQ with ACSL3 knockdown significantly enhanced the caspase-3 activity compared to the single treatments, indicative of enhanced caspase-dependent cell death (Fig. 5D). These results were further confirmed in PANC-1 cells because treatment with HCQ or bafilomycin A1 sensitized PANC-1 cells to the anti-proliferative effect of ACSL3 knockdown (Fig. 5E and F). Notably, the combination of HCQ with ACSL3 knockdown led to a striking induction of caspase-3 activity compared to the single treatments (Fig. 5G). 4-aminoquinolones such as HCQ in addition to inhibiting autophagy have effects also on other cellular processes (i.e. mitochondrial function) [28]. However, since the second autophagy inhibitor, bafilomycin A1, provided similar results we conclude that the cooperative effects that we observed can be ascribed to autophagy inhibition.
Consistent with our data upon ACSL3 knockdown, serum lipid depletion combined with HCQ cooperated in suppressing the proliferation of HPDEK cells (Fig. 5H). Notably, ACSL3 knockdown in cells treated with lipid depleted serum (with or without HCQ treatment) did not enhance further the anti-proliferative effect, confirming that the impact of ACSL3 on cell proliferation is solely dependent on the activation of extracellularly derived lipids (Fig. 5H).
Lastly, serum lipid depletion combined with HCQ treatment enhanced apoptosis compared to the single treatments in both PANC-1 and HPDEK cells (Fig. 5I and Supplementary Fig. 3A).
Taken together, these results suggest that the enhanced autophagy upon extracellular lipid depletion has a pro-survival role, and that concurrent lipid depletion and autophagy inhibition suppresses PDAC proliferation and enhances cell death more effectively than lipid depletion alone.
We next addressed the mechanistic basis of lipid depletion-induced increase in autophagic flux. Similar to serum starvation, lipid depletion is expected to drain cells from essential building blocks. Nutrient starvation is a potent inducer of autophagy through a mechanism involving suppression of the mechanistic target of rapamycin complex 1 (mTORC1) [29]. Therefore, we speculated that extracellular lipid restriction, either by serum lipid depletion or ACSL3 inhibition, induces autophagy by inhibiting the mTORC1 signaling pathway. To test this possibility, we assessed the activation status of a downstream target of mTORC1, S6 kinase. Immunoblot analyses evidenced that upon ACSL3 knockdown (Fig. 5J) or shift to lipid depleted serum (Fig. 5K), the phosphorylated form of S6 appeared reduced, suggesting suppression of the mTORC1 signaling pathway.
ACSL3 knockdown sensitizes PDAC to autophagy blockade in vivo
Based on our in vitro findings, we decided to test whether ACSL3 inhibition and autophagy blockade can cooperate in suppressing tumor formation in vivo. To this end, we transduced the human patient-derived PANC-1 cells with a shRNA against ACSL3 and we generated subcutaneous xenograft implants in immunodeficient mice. When the xenografts reached an average size of 100 mm3 the mice were treated with HCQ and the size of the xenograft tumors was measured with a caliper over time. Notably, the combination treatment was more effective in suppressing tumor progression compared to the vehicle control or single treatments (Fig. 6A). Moreover, the combination therapy significantly increased the survival of mice compared to the single treatments (Fig. 6B).
Taken together, our results indicate that the combination of autophagy and ACSL3 inhibition in vivo greatly improves the efficacy and survival outcome of the single treatments providing the rationale for future implementation of this combination strategy.