The RNeasy Midi Kit was purchased from Qiagen (Valencia, CA, USA). SYBR Premix EX Taq II and WST-1 were acquired from Takara Korea Biomedical Inc. (Seoul, Korea). Fetal bovine serum (FBS), Dulbecco’s modified Eagle’s medium (DMEM), and lipofectamine 2000 were purchased from Thermo Fisher Scientific (Waltham, CA, USA). Anti-ATG5 antibody was acquired from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Anti-LC3 antibody, paclitaxel, hydroxychloroquine, and rapamycin were purchased from Sigma-Aldrich (St. Louis, MO, USA). SBI-0206965 was obtained from Cell Signaling Technology (Danvers, MA, USA).

The development of Ras-NIH 3T3/Mdr cells, which show high levels of P-glycoprotein (P-gp) compared with their parental counterparts (Ras-NIH 3T3 cells), has been previously described [25]. The Ras-NIH 3T3/Mdr cells were maintained at 37°C in DMEM supplemented with 10% FCS. The Ras-NIH 3T3/Mdr cells were passaged at least three times in paclitaxel-free culture medium before use in assays. Paclitaxel was made up in dimethyl sulfoxide (DMSO) as a stock solution and freshly diluted in culture medium before each experiment. The final concentration of DMSO in all the experiments never exceeds 0.1%.

ATG5 CRISPR/Cas9 construct were obtained from ToolGen (Seoul, Korea). pEGFP-LC3 (Addgene #11546), pCI-neo-mAtg5 (Addgene #22956) and ptfLC3 (Addgene #21074) were obtained from Addgene (Cambridge, MA, USA). The cells were transiently transfected by Lipofectamine 2000 with an expression vector encoding pEGF-LC3 or pCI-neo-mAtg5. At 24 h post-transfection, cells were treated with paclitaxel.

ATG5 KO cell lines were generated with ATG5 CRISPR/Cas9 construct as previously described [9], with target single guide (sg) RNA sequence: 5΄-AAGATGTGCTTCGAGATGTGTGG-3΄. The expanded single cell clones were used for assessment of ATG5 gene status. The following primer sets were used to confirm ATG5 KO: forward primer, 5΄-GCTTCGAGATGTGTGGTTTG-3΄ and reverse primer, 5΄-CAGTGGTGTGCCTTCATATT-3΄. The PCR products were verified by agarose gel electrophoresis (2.0% [w/v] agarose) followed by staining with ethidium bromide.

The mRNA levels of four ABC transporters were measured by RT-qPCR. Briefly, cDNA weas used for qPCR containing primers specific for each ABC transporter. All primers were synthesized by Bioneer (Daejeon, Korea). The primer sequences used for the qPCR analysis are listed in Table 1. The qPCR was carried out with an Applied Biosystems 7300 Real-Time PCR System (Foster City, CA, USA). The qPCR data were evaluated by the 2^−ΔΔCt method [26], normalized by the expression of β-actin.

Autophagy was detected by conversion of LC3-I to LC3-II using immunoblot analysis. Immunoblotting was performed with monoclonal LC3 antibody as previously described [27]. Immunoblot images were captured using the Bio-Rad ChemiDoc XRS+ system (Hercules, CA, USA), and image analysis was done using the Bio-Rad Image Lab software version 5.2.1 (Bio-Rad Laboratories).

WST-1 assay was carried out to measure the cytotoxicity of paclitaxel on ATG5 KO and their parental MDR cells. The cells were seeded in quadruplicate wells of 96-well plates and were then treated with paclitaxel for 2 or 3 days. A volume of 10 μl of WST-1 was added to each well and incubated for 4 more hours at 37°C. The absorbance at 450 nm was measured using a SpectraMax 190 microplate reader (Molecular Devices, Sunnyvale, CA, USA).

Trypsinized cells were ethanol-fixed and then stained for total DNA with propidium iodide (PI) for 5 min. The DNA content was measured with the Gallios flow cytometer (Beckman Coulter, Inc., Brea, CA, USA). Data were acquired with the Kaluza analysis software (Beckman Coulter, Inc.).

Apoptotic assay was carried out using the FITC Annexin V Apoptosis Detection Kit (BD Biosciences Pharmingen, San Jose, CA, USA). Briefly, harvested cells (2 × 10⁶) were resuspended the cells in binding buffer and incubated with PI and FITC-conjugated Annexin-V for 10 min in the dark. After adding the binding buffer, the samples were immediately analyzed by the Gallios flow cytometer with the Kaluza analysis software (Beckman Coulter, Inc.).

We established ATG5 KO in Ras-NIH 3T3/Mdr cell lines using CRISPR/Cas9 technology, as we previously reported [27]. We designed a sgRNA targeting exon 2 of the ATG5 gene (Fig. 1A). RT-PCR analysis revealed a loss of ATG5 mRNA in one clone (clone 3) from six initial clones (Fig. 1B). Autophagy deficiency in clone 3 was also confirmed by blockage of conversion of soluble LC3-I to lipid bound LC3-II induced by rapamycin, which is a powerful autophagy inducer (Fig. 1C). The lipidation of LC3-I into LC3-II is a well-recognized hallmark to evaluate the degree of autophagy activity [28]. We further verified autophagy deficiency in ATG5 KO cells by transfecting a tandem mRFP-GFP-LC3 reporter (Fig. 1D). Yellow fluorescence, which represented co-localization of RFP fluorescence with GFP, indicates an autophagosome, and a RFP-only signal is considered an autolysosome [29]. In parental MDR cells (ATG5 wild-type [WT]), rapamycin induced the formation of autophagosomes and autolysosomes. However, the numbers of GFP-LC3 and RFP-LC3 puncta were remarkably reduced in ATG5 KO cells as compared with WT cells under both control and rapamycin-treated conditions.

As expected, Ras-NIH 3T3/Mdr cells showed stronger resistance to paclitaxel than Ras-NIH 3T3 cells (Fig. 2A). Interestingly, Ras-NIH 3T3/Mdr cells with ATG5 KO was re-sensitized to paclitaxel, suggesting that functional autophagy may be required for paclitaxel resistance in MDR cells. The involvement of autophagy in paclitaxel resistance was further confirmed by the addition of early and late autophagy inhibitors (Fig. 2B). SBI-0206965 has been shown to effectively inhibit autophagy at early-stage when targeting autophagy kinase ULK1 [30], whereas hydroxychloroquine (HCQ) has been known to block late-stage autophagic flux [31]. The influence of two autophagy inhibitors treatment on cell growth compared to control was measured over a 72-h time course. We found that treatment with either SBI-0206965 or HCQ could significantly sensitize Ras-NIH 3T3/Mdr cells to paclitaxel at concentrations as low as 300 nM. An early autophagy inhibitor SBI-0206965 was more effective in inhibiting cell survival than a late autophagy inhibitor HCQ. The ABC family of transporters has been known to be involved in transport-based classical MDR [32]. Among 49 different types of transporters [32], P-glycoprotein (P-gp/ABCB1), MDR protein 1 (MRP1/ABCC1), breast cancer resistant protein (BCRP/ABCG2), and MDR 10 (ABCC10/MRP7) transporters were selected because they are known to frequently drive chemo-sensitive cancers to MDR [33,34]. Thus, to investigate the effect of ATG5 KO on the expression of these transporters, the mRNA expression levels of four ABC transporters were determined using RT-qPCR analysis (Fig. 2C). Ras-NIH 3T3 cells were used as paclitaxel-sensitive control cells. There was no correlation between resistance to paclitaxel and expression of MRP1, MRP7 and BCRP proteins, regardless of whether they were either ATG5 KO cells or their parental MDR cells. Meanwhile, the expression level of P-gp, which is greatly increased in cells with acquired MDR, decreased significantly after the ATG5 KO. Interestingly, we also observed that AGT5 KO cell line derived from a non-resistant parent cells (Ras-NIH 3T3 ATG5 KO/Mdr), which was further manipulated to become MDR, showed a strong expression of P-gp comparable to that in Ras-NIH 3T3/Mdr cells. These results imply that the autophagy deficiency might not be the direct cause for downregulation of P-gp, as observed in Ras-NIH 3T3/Mdr ATG5 KO cells.

Next, we investigated the effect of ATG5 overexpression on paclitaxel resistance in ATG5 KO cells, previously shown to be sensitive to paclitaxel. The ATG5 overexpression was verified by western blot analysis (Fig. 3A). We observed that overexpression of ATG5 induced a shift from LC3-I to LC3-II in paclitaxel-treated ATG5 KO cells (Fig. 3B). Unexpectedly, ATG5 overexpression exerted little effect on paclitaxel resistance of ATG5 KO cells (Fig. 3C). These results raise the possibility that low level of resistance to paclitaxel in ATG5 KO cells may be related to other roles of ATG5 independent of its function in autophagy.

Flow cytometric analysis was performed to assess the cell cycle changes of ATG5 KO cells treated with paclitaxel. As shown in Fig. 4A, ATG5 KO caused a significant reduction in the cell population of cells in the S phase (18.572% in ATG5 KO versus 30.186% in their parental MDR cell populations) but an increase in the population of cells in the G₂/M phase (28.676% in ATG5 KO vs. 15.703% in their parental MDR cell populations). The number of cells in the G₁ phase remained almost unchanged (52.752% vs. 54.111% for ATG5 KO and their parental MDR cells, respectively). More interesting, paclitaxel caused a much stronger G₂/M arrest in ATG5 KO cells (84.065%) than in their parental MDR cells (36.999%). Overexpression of ATG5 in ATG5 KO cells displayed a cell cycle distribution similar to that of mock control cells regardless of paclitaxel treatment.

The data presented in Fig. 4B revealed that, in ATG5 KO cells, paclitaxel promoted necrosis, in addition to apoptosis. Population of necrotic cells increased significantly from 3.53% (in parental MDR cells) to 13.04% (in ATG5 KO cells). The proportion of late apoptotic cells also increased from 0.7% to 6.88% in ATG5 KO cells compared with their parental MDR cells after treatment with paclitaxel. When camptothecin was used as a positive control for apoptosis, there was little or no difference in apoptotic cells between ATG5 KO and their parental MDR cell populations. These data suggest that the difference in the necrotic cell population may contribute to the difference in sensitivity to paclitaxel between ATG5 KO and their parental MDR cells.