NSCLC cells sustain phase separation of cytoplasmic membrane-less organelles to protect themselves against cisplatin treatment

Abstract

Cisplatin is the first platinum compound used for anticancer therapy, including non-small cell lung cancer (NSCLC). However, the clinical efficacy of cisplatin is strongly limited by cisplatin resistance. Hence, illuminating the mechanism of cisplatin resistance will aid in the development of therapeutic strategies that improve the sensitivity of cancer cells to cisplatin. Interestingly, membrane-less organelles, which are formed through biomolecular condensation in association with phase separation, have been recently linked with cancers. Here, we reveal a new molecular basis of cisplatin resistance in NSCLC, showing that cisplatin kills cancer cells by the alteration of cytoplasmic membrane-less organelles. Specifically, cisplatin treatment results in the disassembly of processing bodies (PBs) and the assembly of stress granule (SG)-like granules which are different from canonical SGs in NSCLC cells, but not cisplatin-resistant NSCLC cells. Moreover, alterations of PBs and noncanonical SG-like granules are associated with cisplatin-induced cancer cell death. Importantly, we found that disrupting PBs and canonical SGs with cycloheximide and FDA-approved pyrvinium helps cisplatin to kill cisplatin-resistant NSCLC cells. Taken together, our findings provide insight into the role of membrane-less organelle regulation in cisplatin resistance and offer an effective solution for overcoming cisplatin resistance in NSCLC.

Introduction

Chemotherapy is one of the most common strategies for the treating cancer, which accounts for approximately one-sixth of the global mortality each year [1]. In the clinic, cisplatin is one of the most effective chemotherapy drugs for the treatment of cancers, such as non-small cell lung cancer (NSCLC) [2, 3]. However, patients treated with cisplatin long-term inevitably develop drug resistance, which greatly reduces the therapeutic effect and results in the recurrence and metastasis of cancer [4]. Therefore, illuminating the mechanism of cisplatin resistance and finding alternative ways to improve the sensitivity of cancer cells to cisplatin are highly important.

Membrane-less organelles are intracellular biomolecular condensates (dynamic multi-molecular assemblies) formed by liquid–liquid phase separation of RNA and proteins, including nucleoli, nuclear speckles, Cajal bodies, stress granules (SGs), and processing bodies (PBs) [5]. Importantly, recent research has linked aberrant behaviors of membrane-less organelles with human diseases, including many types of cancer and neurodegenerative diseases [6]. Many anticancer drugs can affect membrane-less organelle formation and disturb their composition and basal function [7]. However, cancer cells may alter membrane-less organelle components and numbers to regulate cell function and allow cells to adapt to the microenvironment and resist drug treatment [8, 9]. There is growing evidence that SGs and PBs, two cytoplasmic membrane-less organelles, are strongly associated with tumorigenesis and drug resistance [10–12]. SGs and PBs share proteins and RNAs that are constantly shuttling between the two membrane-less organelles and a percentage of SGs and PBs “dock” under stress, which suggests that they are linked in function [11]. Functionally, they both regulate RNA fate including the decay and storage of mRNA. Although previous studies have shown that cisplatin alters SG composition and protein translation inhibition [13, 14], the contribution of the abnormal regulation of SGs or PBs to drug resistance in cancer therapy is unknown. Importantly, how to interfere with SG or PB dynamics to provide therapeutic strategies for cisplatin resistance is unknown.

Here, we report that cisplatin induces PB disassembly and SG-like granule formation in A549 cells, but not in A549/DDP cells, a drug-resistant NSCLC cell line. Importantly, pyrvinium, an FDA-approved anthelminthic drug, and CHX can increase cisplatin-induced cell death in drug-resistant cell lines via interfering PB and SG. Overall, our results link cisplatin-induced cytotoxicity with dysregulation of SG-like granules and PBs, and also provide a solution for overcoming cisplatin resistance in NSCLC.

Materials and methods

Plasmid constructs

The critical plasmids for this study, including EGFP-B23, and EGFP-p62 were described previously [15, 16]. EGFP-FBL was generated by PCR from a cDNA library and FBL was inserted into the pEGFP-N3 vector at EcoRI/SalI sites. EGFP-DAXX was generated by PCR from a cDNA library and DAXX was inserted into the pEGFP-C1 vector at the EcoRI/SalI sites. RFP-DCP1A was a gift from Dr. Benjamin Wolozin. EGFP-G3BP1 was a gift from Dr. Peiguo Yang. All the plasmids used in this study were verified by sequencing (GENEWIZ).

Cell culture, transfection, and chemicals

Human embryonic kidney 293 (HEK293) cells, HeLa cells, A549 cells, and A549/DDP cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, 11995500) or 1640 medium (UUBIO, U11001) supplemented with 10% fetal bovine serum (FBS; EallBio, 03.C16001DC), 100 U/ml penicillin, and 100 μg/ml streptomycin (UUBIO, U11005) at 37 °C in 5% CO₂ incubator. The plasmids were transfected into cells with HieffTrans^TM Liposomal Transfection Reagent (Yeasen, 40802ES02) in Opti-MEM (OMEM; Gibco, 31985-070) without serum. The EGFP-G3BP1-stable cell line was constructed using lentivirus carrying EGFP-G3BP1. The following drugs were used in our study: DMF (Sangon Biotech, A501361-0500), DMSO (BBI Life Science Corporation, A600163-0500), CHX (MedChemExpress, HY-12320), sodium arsenite (AS; Sigma, S7400), pyrvinium pamoate (pyrvinium; TargetMol, T7333), mitoxantrone (TargetMol, T6588), digitoxin (TargetMol, T0295), and cisplatin (MedChemExpress, HY-17394) dissolved in DMF.

Quantitative real-time PCR

Total RNA in HEK293 cells was isolated with RNA isolater total RNA extraction reagent (Vazyme, R401-01). cDNAs were obtained using HiScrip III RT SuperMix for qPCR (Vazyme, R323-01) and then, ChamQ Blue Universal SYBR qPCR Master Mix (Vazyme, Q312-02) was used to detect the expression of the target gene. The following primers were used: human DCP1A: 5′-CACCCCGGTGCTAATCACTC-3′ and 5′-GCTCAACGGGATTGTGTAGGTT-3′, human DDX6: 5′-GCCTACCTCATTCCCTTACT-3′ and 5′-CAACCTTTGCTACTCCTTTC-3′, human G3BP1: 5′-CGGGCGGGAATTTGTGAGA-3′ and 5′-TCTGTCCGTAGACTGCATCTG-3′, human GAPDH: 5′-AAATCCCATCACCATCTTCCAG-3′ and 5′-AGGGGCCATCCACAGTCTTCT-3′.

Immunoblot

The cells were treated with the indicated drugs and then lysed in cell lysis buffer (50 mM Tris–HCl (pH 7.6) with a protease inhibitor cocktail (Roche, 4693132001), 0.5% sodium deoxycholate, 1% NP-40, and 150 mM NaCl). Proteins were separated by 10% or 12% polyacrylamide gel electrophoresis (SDS–PAGE) and transferred onto polyvinylidene difluoride membrane (PVDF membrane; Millipore, IPVH00010). The following primary antibodies were used: anti-GAPDH (Proteintech, 60004-1-lg; 1:8000), anti-G3BP1 (Proteintech, 13057-2-AP; 1:4000), anti-DDX6 (OriGENE, TA804432S;1:600), and anti-GFP (SANTA CRUZ, sc-9996, 1:2000). The following secondary antibodies were used in our study: horseradish peroxidase-conjugated sheep anti-mouse and anti-rabbit antibodies (Jackson ImmunoResearch Laboratories). The proteins were visualized via an ECL detection kit (Vazyme, E411-04).

Immunofluorescence assay

For the immunofluorescence assay, the cells were fixed with 4% paraformaldehyde at room temperature for 10 min followed by permeabilization with 0.1% Triton X-100 for 10 min. The cells were incubated with primary antibody at 4 °C overnight after being washed with PBST for 5 min. Then the cells were incubated with secondary antibody at room temperature for 2 h. Finally, the immunostained or live cells were observed using a Nikon confocal microscope or a Nikon Ti2-E fluorescence microscope integrated with a pco.edge 4.2 bi sCMOS camera [17–19]. The following primary and secondary antibodies were used: anti-G3BP1 (Proteintech, 13057-2-AP; 1:2000), anti-G3BP1 (Santa Cruz, sc-365338; 1:250), anti-DCP1A (ABclonal, A7376; 1:500), anti-DDX6 (OriGENE, TA804432S; 1:200), anti-PABPC1 (Proteintech, 66809-1-lg; 1:500), anti-UBAP2L (Cell Signaling Technology, #40199; 1:500), anti-EDC4 (Proteintech, 17737-1-AP; 1:500), anti-LSM14A (Proteintech, 18336-1-AP; 1:500), anti-RPA194 (Santa Cruz, sc-48385; 1:500), anti-NCL (Proteintech, 10556-1-AP; 1:1000); Alexa Fluor 405 Goat anti-Rabbit IgG (Invitrogen, A-31556), Alexa Fluor 488-conjugated Affinipure Donkey Anti-mouse IgG (Proteintech, SA00006-1), Alexa Fluor 488-conjugated Affinipure Donkey Anti-rabbit IgG (Proteintech, SA00006-2), Alexa Fluor 594-conjugated Affinipure Donkey Anti-mouse IgG (Proteintech, SA00006-3), Alexa Fluor 594-conjugated Affinipure Donkey Anti-rabbit IgG (Proteintech, SA00006-4). Hoechst (Sigma, 23491-45-4) was incubated at room temperature for 20 min to label the nucleus.

PI staining assay and CCK8 assay

The cells were incubated with propidium iodide (PI, Beyotime, ST511) and Hoechst at 37 °C for 20 min after drug treatment, after which PI-stained cells (dead cells) and Hoechst-positive cells (total cells) were observed using fluorescence microscopy. The PI/Hoechst ratio was used to determine the percentage of surviving cells. For the Cell Counting Kit-8 (CCK-8, YEASEN, 40203ES76) assay, the cells were cultured in 96-well plates and treated with indicated drugs. Then, 10 µl of cell counting kit-8 solution with 90 µl of cell culture media was added to each well at 37 °C for 2 h. The absorbance value was detected using a spectrophotometer at a wavelength of 450 nm.

FRAP assay

The PBs in RFP-DCP1A-expressing cells or EGFP-G3BP1-positive condensates (SGs or SG-like granules) in stablely expressing cells were bleached using 594 nm or 488 nm lasers at 100% laser power. After photobleaching, time-lapse images were captured for ~5 min. The fluorescence intensity at each time point was normalized to the fluorescence intensity before photobleaching. Fluorescence recovery was quantified and analyzed using ImageJ and GraphPad Prism.

Statistical analysis

ImageJ was used to perform immunoblot densitometric analysis, image processing, and fluorescence intensity analysis. The data were used to generate charts using Prism 7.0 (GraphPad Software) software. P-values and means are indicated in the figure legends. The cartoon models were generated using Adobe Illustrator 2021. The 3D shapes shown in Fig. 2h, j were generated via imaris.

Fig. 2. Cisplatin affects the properties of PBs and SGs in cells.

a A549 cells were treated with 100 μM cisplatin for 24 h or 0.5 mM AS for 40 min, and then the cells were fixed and stained with anti-G3BP1 (green, SG marker) and anti-DDX6 (red, PB marker) antibodies. Scale bars, 10 µm. The normalized intensity of the corresponding line scan (white) was analyzed to show the relationship between DDX6 and G3BP1. b Schematic diagram of PBs (orange), classical SGs (blue) after AS treatment, and SG-like granules (green) after cisplatin treatment. c, d Quantification of the number and size of DDX6-labeled PBs and G3BP1-labeled SGs or SG-like granules per cell. Data are represented as means ± SD, ns, not significant, **P < 0.01, as analyzed by one-way ANOVA or t-test. e A549 cells were treated with 100 μM cisplatin for 24 h and then were fixed and stained with anti-PABPC1 or anti-UBAP2L, and anti-G3BP1 antibodies. Hoechst (blue) was used to indicate the nucleus. Scale bars, 10 µm. f A549 cells were treated and fixed as in (e) and stained with anti-EDC4 or anti-LSM14A, and anti-DDX6 antibodies. Hoechst (blue) was used to indicate the nucleus. Scale bars, 10 µm. g HEK293 cells were transfected with RFP-DCP1A and then were treated with DMF or 100 μM cisplatin for 24 h. FRAP analysis of PB properties and the FRAP recovery curves were quantified. Data are represented as means ± SEM, **P < 0.01, as analyzed by t-test. h Rendered 3D shapes and a schematic of PB after treatments in (g). i HEK293 cells stably expressing EGFP-tagged G3BP1 were treated with 0.5 mM AS for 30 min or 100 μM cisplatin for 24 h. FRAP analysis of G3BP1-labeled SG or SG-like granule properties and the FRAP recovery curves were quantified. Data are represented as means ± SEM, **P < 0.01, as analyzed by t-test. j Rendered 3D shapes and a schematic of G3BP1-labeled SG or SG-like granule after treatment as in (i).

Results

Cisplatin affects the phase separation of various membrane-less organelles in cells

Previous studies have shown that oxaliplatin and cisplatin, as platinum-based anticancer drugs, affect intracellular membrane-less organelles, such as nucleolus [20]. However, it is not fully understood whether cisplatin affects the assembly or disassembly of other membrane-less organelles in cells. To comprehensively investigate the effect of cisplatin on various membrane-less organelles, we first expressed EGFP-FBL, EGFP-B23 (nucleolar marker), EGFP-DAXX (PML body marker), RFP-DCP1A (PB marker), and EGFP-p62 (p62 body marker), along with labeled endogenous G3BP1 (SG marker), NCL, and RPA194 (nucleolar marker) to monitor the dynamics of membrane-less organelles following cisplatin treatment in HEK293 cells. Consistent with previous studies, cisplatin treatment led to nucleolar depolymerization (Fig. 1a and Supplementary Fig. S1a). In addition, cisplatin treatment decreased the number of DCP1A-labeled condensates and induced the formation of G3BP1-labeled condensates over time (Fig. 1c, d). In contrast with these, cisplatin did not affect the liquid–liquid phase separation of the PML body and p62 body (Fig. 1b, e). To determine whether the effects of cisplatin on these membrane-less organelles are associated with alterations in protein-levels, we analyzed the change of corresponding proteins. The data showed that cisplatin treatment decreased the levels of DAXX or FBL, but not the levels of B23 or p62, suggesting that the effects of cisplatin on these membrane-less organelles are not totally dependent on protein levels (Supplementary Fig. S1b). Taken together, these findings indicate that cisplatin affects the liquid–liquid phase separation of multiple membrane-less organelles in cells, including DCP1A- and G3BP1-positive condensates.

Fig. 1. Cisplatin affects phase separation of multiple membrane-less organelles.

a, b, d, e HEK293 cells were transfected with EGFP-FBL, EGFP-B23, EGFP-DAXX, RFP-DCP1A, or EGFP-p62. Then the cells were treated with 100 μM cisplatin (cis) for the indicated time. Fluorescence microscopy was used to image the cells. The percentage of dilute phase or dense phase of the indicated protein was quantified from 16 cells per group. Data are represented as means ± SEM, ns, not significant, **P < 0.01, as analyzed by two-way ANOVA. Scale bars, 10 µm. c HEK293 cells were treated as in (a) and then were fixed and stained with anti-G3BP1 antibody. Fluorescence microscopy was used to image the cells. The percentage of dilute or dense phase of indicated protein was quantified from 16 cells per group. Data are represented as means ± SEM, **P < 0.01, as analyzed by two-way ANOVA. Scale bars, 10 µm.

Cisplatin perturbs the properties of canonical PBs and SGs in cells

To determine the differences between cisplatin-related G3BP1 or DCP1A-positive condensates and canonical PBs/SGs, we analyzed multiple properties of G3BP1-positive condensates and DCP1A-positive condensates via the use of DDX6, another PB marker. The data revealed that treatment with arsenite (AS, a well-studied inducer of canonical SG formation) rapidly induced SG assembly and increased the number of PBs throughout the cytoplasm, occasionally revealing the “docking” position of SGs and PBs (Fig. 2a, b). In contrast to AS treatment, cisplatin treatment reduced the PB number but induced the formation of G3BP1-positive condensates which were smaller than the canonical SGs in both NSCLC A549 and HEK293 cells (Fig. 2c, d, Supplementary Fig. S2a, b). To test whether cisplatin-related G3BP1-positive condensates and PBs share some composition with canonical SGs and PBs, we labeled two proteins that exist in canonical SGs, PABPC1, and UBAP2L, and two proteins that exist in canonical PBs, EDC4 and LSM14A. The data showed that the cisplatin-related G3BP1-positive condensates and PBs had the same composition as canonical PBs and SGs (Fig. 2e, f). Therefore, we named the cisplatin-induced G3BP1-positive condensates SG-like granules. To further investigate the effects of cisplatin on the physical properties of PBs and SG-like granules, we performed fluorescence recovery after photobleaching (FRAP) analysis in RFP-DCP1A-expressing cells and EGFP-G3BP1-stable cell lines. Cisplatin treatment decreased the fluorescence recovery rate of DCP1A, suggesting that cisplatin induces the aging of PBs within cells, but the sphericity of the cisplatin-induced PBs was similar to that of the normal PBs (Fig. 2g, h). In addition, compared with AS-induced SGs, cisplatin-induced SG-like granules had greater dynamics and sphericity (Fig. 2i, j). Taken together, these data suggest that cisplatin-related PBs and SG-like granules are different from classical PBs and SGs.

To further examine the difference between cisplatin-related PBs/SG-like granules and canonical PBs/SGs, we monitored the recovery of PBs and SGs/SG-like granules after cisplatin or AS washout. The results of the immunofluorescence assay indicated that the number of AS-induced PBs recovered to normal levels and that SGs disappeared after AS washout (Fig. 3a–d). Unlike canonical PBs/SGs, the number of DDX6 and DCP1A-labeled PBs was difficult to increase, and G3BP1-labeled SG-like granules were present continuously after cisplatin removal, suggesting that the cisplatin-induced reduction in PBs and the formation of SG-like granules were persistent (Fig. 3d–h). Overall, these data suggest that cisplatin results in irreversible PB disassembly and SG-like granule formation.

Fig. 3. Cisplatin irreversibly affects the phase separation of PBs and SG-like granules.

a A549 cells were treated with 0.5 mM AS for 30 min and then were washed and cultured in fresh medium for 1 h. Next, the cells were fixed and stained with anti-G3BP1 (green, SG marker) and anti-DDX6 (red, PB marker) antibodies. Hoechst (blue) was used to indicate the nucleus. Scale bar, 10 µm. b, c The number of PBs or SGs per cell was counted. Data are represented as means ± SD, **P < 0.01, and one-way ANOVA. d Schematic diagram of PBs, SGs, and SG-like granules after cisplatin or AS treatment and washout. e, f A549 cells were treated with 100 μM cisplatin for 24 h and then the cells were washed and cultured in fresh medium for 12 or 24 h. Next, the cells were fixed and stained with anti-DDX6 (red) and anti-DCP1A (green) antibodies to indicate PBs. Hoechst (blue) was used to indicate the nucleus. Scale bar, 10 µm. The number of DDX6-labeled PBs per cell was counted. Data are represented as means ± SD, ns, not significant, **P < 0.01, and one-way ANOVA. g, h A549 cells were treated as in (e). Next, the cells were fixed and stained with anti-G3BP1 (red) to indicate SG-like granules. Hoechst (blue) was used to indicate the nucleus. Scale bar, 10 µm. The number of G3BP1-labeled SG-like granules per cell was counted. Data are represented as means ± SD, ns, not significant, **P < 0.01, as analyzed by one-way ANOVA.

Cisplatin treatment impairs the assembly of canonical PBs and SGs

Previous studies have shown that canonical PB and SG formation protects cancer cells against stress-induced cell death [12]. Moreover, the effects of cisplatin on PBs and SG-like granules were persistent, and we aimed to determine whether cisplatin-treated cells respond to subsequent stress. To this end, we examined PB and SG numbers in combination with cisplatin pretreatment and AS. The data revealed that cisplatin pretreatment abolished AS-induced SG formation and the increase of PB number in both A549 and HEK293 cells (Fig. 4a–e). Overall, cisplatin impairs canonical PB and SG assembly, which is beneficial for cancer cell survival.

Fig. 4. Cisplatin impairs the normal liquid-liquid phase separation of canonical PBs and SGs.

a Schematic diagram of PB, SG, and SG-like granule changes after cisplatin and AS treatment. b A549 cells were treated with 100 μM cisplatin for 24 h followed by 0.5 mM AS for 30 min and then were fixed and stained with anti-G3BP1 (green) to label SGs or SG-like granules, and anti-DDX6 (red) to label PBs. Scale bars, 10 µm. d HEK293 cells were transfected with RFP-DCP1A and treated as in (b) and then were fixed and stained with anti-G3BP1 (green) to label SGs or SG-like granules. Scale bars, 10 µm. c, e The number of DDX6 or DCP1A-labeled PBs and G3BP1-labeled SGs or SG-like granules per cell were quantified in (b) and (d). Data are represented as means ± SD, ns, not significant, *P < 0.05, **P < 0.01, as analyzed by one-way ANOVA.

Cisplatin-induced PB disassembly is prior to SG-like granule formation

To understand the relationship between cisplatin-induced PBs and SG-like granules, we first monitored their changes under cisplatin treatment over time. Under normal conditions, G3BP1 was uniformly distributed in the cytoplasm and DDX6-labeled PBs were persistent. Importantly, we observed that PBs decreased first, followed by SG-like granule formation in response to cisplatin treatment (Fig. 5a–d). Owing to the varying sensitivity of different cells to cisplatin, we examined the concentration and time-dependent effects of cisplatin on PBs and SG-like granules in two distinct cancer cell lines, NSCLC A549 cells and human cervical cancer HeLa cells. The number of PBs decreased, and SG-like granules appeared in both A549 and HeLa cells with increasing cisplatin concentration and time. However, HeLa cells were more sensitive to cisplatin treatment than A549 cells were (Fig. 5a–d and Supplementary Fig. S3a, b). To further test the time-dependent relationship between PBs and canonical SGs under different types of stress, we also monitored their formation over time under AS treatment in NSCLC A549 cells. Similar to cisplatin treatment, AS-induced SG formation lagged behind PB assembly (Fig. 5e–g). Taken together, these data suggest that cisplatin affects PBs before SG-like granules in a time- and concentration-dependent manner.

Fig. 5. Cisplatin-induced PB disassembly is prior to SG-like granule formation.

a A549 cells were treated with cisplatin for the indicated concentrations and time, and then the cells were fixed and stained with anti-G3BP1 (green) to label SG-like granules and anti-DDX6 (red) to label PBs. Scale bars, 10 µm. b Schematic diagram of PB and SG-like granule changes after cisplatin treatment at different time. The cartoon represents PBs as orange dots and SG-like granules as green dots. c, d The number of PBs per cell was counted and the percentage of cells with SG-like granules was quantified from three independent experiments. Data are represented as means ± SD, ns, not significant, **P < 0.01, and one-way ANOVA. e A549 cells were treated with 0.25 mM AS for indicated time, and then the cells were fixed and stained with anti-G3BP1 (green) to label SGs and anti-DDX6 (red) to label PBs. Scale bar, 10 µm. f, g The number of PBs per cell was counted per group and the percentage of cells with SGs was quantified from three independent experiments. Data are represented as means ± SD, **P < 0.01, and one-way ANOVA.

To elucidate the mechanism underlying the cisplatin-induced reduction in PBs and the formation of SG-like granules, we performed biochemical analysis to examine the SG-like granule and PB-related protein levels. The results revealed that the protein levels of DDX6 and DCP1A, PB-associated proteins, and G3BP1, SG-like granule-associated proteins, gradually decreased with increasing cisplatin concentration and treatment time in A549 cells (Supplementary Fig. S4a, b). In addition, these results were confirmed in HeLa cells (Supplementary Fig. S4c). To further determine whether the decrease in PB- and SG-like granule-associated protein levels is due to decreased transcription levels, we performed a quantitative real-time PCR assay, which suggested that cisplatin treatment significantly reduced the mRNA levels of PB- and SG-like granule-related genes in A549 and HEK293 cells (Supplementary Fig. S4d, e). Although cisplatin treatment decreased both the protein and mRNA levels associated with PBs and SG-like granules, SG-like granule assembly was observed, suggesting that SG-like granule formation is not entirely dependent on the transcriptional regulation of related genes and corresponding protein levels.

Cancer cells protect themselves from cisplatin treatment by inhibiting the formation of SG-like granules and sustaining the number of PBs

To explore whether cisplatin resistance is associated with PB and SG-like granule regulation, we compared the differences in PB and SG-like granule formation between NSCLC cells and cisplatin-resistant NSCLC cells in response to cisplatin treatment. The results suggested that cisplatin treatment had a weaker effect on PBs and SG-like granules in A549/DDP cells than in A549 cells (Fig. 6a–e). Similarly, cisplatin effectively killed A549 cells but was less toxic to A549/DDP cells suggesting that the development of cisplatin resistance was closely related to the regulation of PBs and SG-like granules within the cancer cells (Fig. 6f). CHX induces PB disassembly in the cytoplasm (Fig. 6c). As a result, we wanted to determine whether CHX pretreatment increased the sensitivity of cisplatin-resistant NSCLC cells to cisplatin by disturbing PBs and SG-like granules. As expected, CHX pretreatment resulted in cisplatin-induced SG-like granule formation and PB disassembly in A549/DDP cells (Fig. 6c–e). Importantly, CHX pretreatment significantly increased the cell death of cisplatin-treated A549/DDP cells (Fig. 6f). Overall, these data suggest that disturbing PBs and SG-like granules with CHX enhances the response of cisplatin-resistant NSCLC cells to cisplatin and improves the killing efficiency of cisplatin on cancer cells.

Fig. 6. Cancer cells protect themselves from the treatment of chemotherapy drugs by inhibiting the formation of SG-like granules and sustaining the PB number.

a Schematic of experimental design to examine the effect of CHX for cisplatin-induced PBs and SG-like granules. b, c A549 or A549/DDP cells were treated with DMF, 100 μM cisplatin for 24 h, 20 µg/mL CHX for 1 h, or 20 µg/mL CHX for 1 h, then washed out and followed by 100 μM cisplatin for 24 h. Then, the cells were fixed and stained with anti-G3BP1 (green) to label SG-like granules and anti-DDX6 (red) to label PBs. Hoechst (blue) was used for indicating the nucleus and fluorescence microscopy was used to image. Scale bars, 10 µm. d, e The number of PBs and SG-like granules was quantified per cell in each sample as means ± SD, ns, not significant, **P < 0.01, and one-way ANOVA. f A549 or A549/DDP cells were treated as in (b and c) for the indicated time and stained with PI and Hoechst for 20 min. And then the cell survivals were quantified over time with drug treatment. Data are collected from three independent experiments, mean ± SEM, **P < 0.01, and one-way ANOVA.

Combination treatment with pyrvinium overcomes cisplatin resistance via mediating PB disassembly and SG-like granule assembly in cisplatin-resistant NSCLC cells

To identify clinical drugs that improve the sensitivity of A549/DDP cells to cisplatin by disturbing PBs and SGs, we first tested the effects of three FDA-approved drugs (pyrvinium, mitoxantrone, and digitoxin) that have been reported to inhibit AS-induced SG formation [21], on canonical PBs and SGs under physiological conditions. The data suggested that pyrvinium treatment had a greater effect on reducing PB number under normal conditions than mitoxantrone and digitoxin did (Supplementary Fig. S5a, b). Moreover, pyrvinium pretreatment also decreased AS-induced PB and SG formation (Fig. 7a–c). Since PBs and SGs are dynamic membrane-less organelles that assemble and disassemble under stress and stress removal, we examined whether pyrvinium-related PB disassembly was recovered to physiological status upon stress removal. The results of immunofluorescence staining revealed that PBs were stable over 2 h after pyrvinium removal, suggesting that pyrvinium had a continuous effect on PBs (Fig. 7d). To further test the effect of pyrvinium on cisplatin-related PBs and SG-like granules, we treated A549/DDP cells with a low dose of pyrvinium followed by pyrvinium removal and cisplatin treatment. Compared with mitoxantrone and digitoxin, pyrvinium pretreatment better decreased the number of PBs and increased cisplatin-related SG-like granule formation in A549/DDP cells, suggesting that pyrvinium enhances the susceptibility of cisplatin-resistant NSCLC cells to cisplatin (Fig. 7e–g and Supplementary Fig. S5b, c). Finally, we tested whether pyrvinium pretreatment interfered with PBs and SG/SG-like granules to kill cisplatin-resistant NSCLC cells. As expected, the CCK8 assay revealed that compared with pyrvinium or cisplatin alone, a low dose of pyrvinium in combination with cisplatin could reduce the viability of A549/DDP cells (Fig. 7h). Taken together, these data suggest that pyrvinium synergizes with cisplatin to overcome drug resistance by disturbing cytoplasmic membrane-less organelles in tumor cells.

Fig. 7. Pyrvinium, an FDA-approved drug, enhances cisplatin cytotoxicity in cisplatin-resistant NSCLC cells by mediating PB disassembly and SG-like granule assembly.

a Schematic diagram of experimental design to test the effect of pyrvinium on canonical PBs and SGs. b A549/DDP cells were treated with 10 μM pyrvinium for 2 h followed by 1 mM AS for 1 h. Then, the cells were fixed and stained with anti-G3BP1 (green, SG marker) and anti-DDX6 (red, PB marker) antibodies. Scale bars, 10 µm. c The percentage of cells with SG-like granules and the number of PBs per cell were quantified after treatment as in (b). The data are represented as means ± SD, ns, not significant, **P < 0.01, and one-way ANOVA. d A549/DDP cells were treated with 10 μM pyrvinium for 2 h and then were cultured in a fresh medium for another 2 h. Next, the cells were fixed and stained with anti-G3BP1 (green, SG marker) and anti-DDX6 (red, PB marker) antibodies. Scale bar, 10 µm. The number of PBs was quantified per cell. The data are represented as means ± SD, ns, not significant, **P < 0.01, and one-way ANOVA. e Schematic diagram of experimental design to assess the effect of pyrvinium on cisplatin-associated PBs and SG-like granules. f A549/DDP cells were treated with 0.5 μM pyrvinium for 1 h, then washed out and followed by 100 μM cisplatin for 24 h. Then, the cells were fixed and stained with anti-G3BP1 (green) to label SGs/SG-like granules and anti-DDX6 (red) to label PBs. Scale bars, 10 µm. g The percentage of cells with SG-like granules and the number of PBs per cell were quantified after the treatment as in (f). The data are represented as means ± SD, ns, not significant, **P < 0.01, t-test. h Cell viability was detected by CCK-8 assay after the cells were treated as in (f). The data are represented as means ± SD, ns, not significant, **P < 0.01, and one-way ANOVA.

Discussion

Drug resistance is a common problem in cancer treatment. There are many explanations for resistance mechanisms, including drug efflux, factors that prevent drug-target interactions, pathways for removing damage from targets, and factors, and pathways involved in cellular responses to damage [22]. In this study, we provide mechanistic insights into phase separation and biomolecular condensation-mediated drug resistance in NSCLC. Our results show that NSCLC cells regulate the number of PBs and the production of noncanonical SG-like granules to protect themselves against treatment with chemotherapy drugs, thereby increasing cancer cell fitness (Fig. 8).

Fig. 8. A membrane-less organelle-centric view of cisplatin resistance in NSCLC.

Our study shows that cisplatin-mediated cytotoxicity is associated with PB and SG dysregulation. In NSCLC (A549) cells, cisplatin treatment induces PB disassembly and SG-like granule assembly to result in cell death. In contrast, cisplatin-resistant NSCLC (A549/DDP) cells protect themselves from cisplatin-induced cell death by maintaining cytoplasmic membrane-less organelles. Importantly, interfering PBs and SG-like granules with CHX and FDA-approved pyrvinium improves the sensitivity of cisplatin-resistant NSCLC cells to cisplatin.

Interestingly, we found that cisplatin-induced the disassembly of PBs and assembly of SG-like granules that were smaller than classical SGs, indicating changes in their properties and composition (Fig. 2). A hypothesis for how cisplatin induces the disassembly of PBs and assembly of SG-like granules is as follows: cisplatin directly targets PB/SG-related components, consistent with previous studies that show that cisplatin directly localizes to SG-like granules and interacts with the SG-associated proteins HSP and VCP and that SG-like granules exclude the translation initiation factors eIF3b, eIF4G, and mRNAs, which exist in classical SGs [13, 23].

Importantly, cisplatin impaired classical PB and SG assembly, which linked cisplatin-induced cytotoxicity with PB and SG dysregulation (Figs. 4 and 6). It is plausible that cisplatin treatment alters the cellular stress response machinery, making cells less capable of adapting to additional stressors. PB and SG formation in the cytoplasm is often considered the adaptive process of cells, which promotes cancer cell survival [12]. However, cisplatin impaired the ability of cells to respond to subsequent stress, such as the response induced by AS. In addition to cisplatin-induced SG-like granules, several kinds of granules differ from classical SGs and are independent of G3BP1, such as the antiviral response [24, 25]. However, whether G3BP1 is necessary for SG-like granule formation needs to be further explored.

Our study also provides a synergistic therapeutic strategy to overcome cisplatin resistance. Drug-resistant cells survived by inhibiting the formation of noncanonical SG-like granules and maintaining the number of PBs (Figs. 6 and 7). Moreover, PB disassembly occurred prior to SG-like granule assembly under cisplatin treatment, suggesting that the intervention of PB assembly is a more effective route to overcome drug resistance. Based on these findings, we revealed that disrupting PBs and SGs with CHX enhanced cisplatin-induced cytotoxicity. Importantly, pyrvinium, an FDA-approved drug, disrupted PB and SG assembly and synergized with cisplatin to kill cisplatin-resistant NSCLC cells (Figs. 6 and 7). The mechanisms by which pyrvinium inhibits PB and SG assembly may also result from the inhibition of Wnt signaling, which antagonizes SG assembly according to a previous study [26]. ‌In addition, pyrvinium also inhibits mitochondrial respiration and autophagy, which together contribute to cancer cell death [27–29]. Moreover, other drugs, including those based on macromolecules and nucleic acids, that target membrane-less organelles are worth studying in the context of drug resistance. Interestingly and importantly, the phase separation of cytoplasmic membrane-less organelles (such as canonical SGs and PBs) has been demonstrated to be associated with drug resistance and tumor cell fitness in patient-related tissues and cells [9, 30]. Our study suggests a potential contribution of phase separation of cytoplasmic membrane-less organelles to the clinical treatment of NSCLC with cisplatin (Fig. 8).

Author contributions

ZY, HFW, and ZLZ designed and supervised the study. NNL, LLR, DS, and BHL developed methods and performed experiments. NNL, LLR, DS, GQM, ZLZ, HFW, and ZY analyzed data. NNL, ZY, and HFW wrote, edited, and revised the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Nos. 32471048, 32371018, 82022022, and 82071274), the Undergraduate Training Program for Innovation and Entrepreneurship of Soochow University (2023xj038), a Project Funded by Science and Education Strengthening Health Program of Suzhou (ZDXM2024005), a Project Funded by Applied Basic Research (Medical and Health) and Science & Technology Innovation Program of Suzhou (SYWD2024028), a Project Funded by the Clinical Research Program of the WuJieping Medical Foundation (Grant no. 320.6750.19092-32), a Project Funded by Jiangsu Key Laboratory of Neuropsychiatric Diseases (BM2013003), a Key Project of Natural Science Foundation of Jiangsu Provincial Higher Education Institutions (23KJA310005), a Project Funded by the Interdisciplinary Basic Frontier Innovation Program of Suzhou Medical College of Soochow University (MP13202823), a Project Funded by the Suzhou International Joint Laboratory for Diagnosis and Treatment of Brain Diseases, and a Project Funded by the Priority Academic Program Development of the Jiangsu Higher Education Institutes (PAPD), a Key Project of Natural Science Foundation of Jiangsu Provincial Higher Education Institutions (23KJA310005), and the Postgraduate Research & Practice Innovation Program of Jiangsu Province (KYCX23_3279).

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41401-025-01551-5.