The cell lines used in this study were all offered by the Department of Pathology, West China Hospital. A panel of 5 breast cancer cell lines were used. And all cell lines were verified by short-tandem repeat (STR) analysis. The cancer cell lines consisted of all the main molecular subtypes of breast cancer, including three TNBC cell lines (MDA-MB-231, MDA-MB-468, BT-549), one Luminal cell line (MCF-7), and one HER2+ cell line (SKBR3). The SKBR3 cell line was cultured in McCoy’s 5A medium (Gibco) supplemented with 10% Fetal Bovine Serum (FBS) (Invitrogen Life Technologies) and 1% penicillin/streptomycin (Invitrogen Life Technologies), while the other cell lines were cultured in RPMI 1640 medium (Gibco) supplemented with 10% FBS and 1% penicillin/streptomycin. All cell lines were maintained by continuous passaging at 37°C with a humidified atmosphere of 5% CO₂.

To generate gene-edited cells, the PX330 plasmid was first linearized with BbsI (NEB) restriction enzyme. The DDX58-sgRNA oligos were then synthesized and cloned into the px330 plasmids using standard molecular cloning techniques. The DDX58-sgRNA sequences used in this study were sgRNA1: 5’GGATTATATCCGGAAGACCC3’ and sgRNA2: 5’ TCTCGGCGGAAAACAATTGA 3’. For transfection, MDA-MB-231 cells were plated at a density of 5x10⁵ cells per well in a six-well plate and allowed to attach overnight. The following day, the cells were transfected with 2 μg of sgRNA1-px330 and 2 µg of sgRNA2-px330 plasmid DNA using lipo3000 transfection reagent (Thermo Fisher Scientific) following the manufacturer’s protocol. After 48 hours, cells were dissociated into single cells and counted using a hemocytometer. To reduce the likelihood of multiple cells being present in a single well, the cells were serially diluted in cell medium to a final concentration of 0.5 cells per 100 μl. A multichannel pipette was then used to distribute 100 μl of diluted cells to each well of a 96-well plate. The 96-well plates were then examined under a microscope, and wells that might have been seeded with multiple cells were marked off after 3-4 days of culturing. The cells were then returned to the 37°C incubator and allowed to expand for 2-3 weeks before being transferred to 48- and 24-well plates. Genomic DNA was extracted from the cells, and PCR was performed to identify positive clones. Genomic DNA of cells was extracted, and polymerase chain reaction (PCR) was used to identify the positive clones. Finally, the positive clones were verified by first-generation sequencing and western blotting.

For siRNA transfection, approximately 3 x 10⁵ cells were seeded in six-well tissue culture plates. After 24 hours, when the cells reached a confluence of approximately 70%, they were transfected with control siRNA and DDX58-targeting siRNA using Lipofectamine 3000, following the manufacturer’s protocol. After 48 hours, the cells were lysed, and cell proteins were extracted to validate the knockdown effect of the siRNA. Validated cells were subsequently used for cell proliferation analysis.

All animal experiments were approved and performed in accordance with the West China Hospital of Sichuan University Biomedical Research Ethics Committee (2022-1056). And our study was reported in accordance with ARRIVE guidelines. All mice were raised in pathogen-free conditions. WT cells and DDX58-KO cells (1 × 10⁷ cells) were suspended in 100 μl of PBS and subcutaneously injected into the right flank of 4-week-old female BALB/c nude mice. Tumour volume was calculated using the formula (V =0.52 × Length × Wide²). After almost 3 weeks, mice were euthanized when the average tumour volume reached 1000 mm³ (Cervical dislocation after Sodium pentobarbital injection). Tumours were collected and cut into almost 17 mm³ cubes and inoculated subcutaneously into nude mice. Fifteen days after injection of the tumour mass, the average tumour volume reached 200 mm^3, and drug therapy started. Dox was dissolved in PBS and given by intraperitoneal (i.p.) injection at 5 mg/kg every three days for 12 days. Control mice received the same amount of PBS by i.p. injection. Tumour volume was monitored during this time. On the 12th day after the initiation of therapy, all mice were euthanized. The tumour growth inhibition rate (IR) was calculated using the formula IR (%) = (1 – TWt/TWc) x 100, where TWt and TWc are the mean tumour weights of the treated and control groups, respectively.

WT and DDX58-KO cells were seeded at 2x10⁵ cells per well in 12-well plates and treated with different concentrations of Dox. To serve as a positive control, poly(I:C) (10 μg/ml) (GLPBIO, GC14710) was added to the cells in the presence of Lipo3000. Six hours later, the cells were fixed with 100% methanol and blocked with 3% BSA in PBS (phosphate-buffered saline). A 1:100 dilution of anti-dsRNA antibody J2 monoclonal antibody (Nordic-MUbio,10010200) in 3% BSA in PBS was added into plates and incubated for 1 hr at room temperature (RT). Following three washes with 0.1% PBST (Triton-100 in PBS), the iFluor™ 488 anti-mouse antibody (HUABIO, HA1125) (1:100) was added to the plates and incubated for 1 hr at RT. Then, the plates were washed in 0.1% PBST three times, and diaminobenzidine tetrahydrochloride substrate (Sigma) was added for a 10-min incubation. Cells were visualized using an Imaging Universal microscope (Zeiss), and at least three fields per sample were counted.

Protein was extracted using RIPA lysis buffer (Beyotime, P0013c) and quantified with the BCA Protein Assay Kit (Thermo Fisher Scientific, 23250). The extracted proteins were separated on a 10% polyacrylamide gel (Bio-Rad) and transferred to a PVDF membrane for immunoblotting. Blocking was performed in 5% low-fat milk in Tris-buffered saline containing 0.1% Tween (TBST) for 1 hr at RT. After blocking, the membranes were incubated in primary antibody dilutions overnight at 4°C. The membranes were incubated with mouse anti-rabbit (1:2000) and mouse anti-mouse (1:2000) (Abcam) for 1 hour at 37°C. Detection was performed using ECL Plus western blotting detection reagents (Millipore, USA).

Briefly, cells were seeded into 96-well plates at a density of 3000 cells per well. After incubation for the indicated time points, 10 μL of CCK-8 solution (MedChemExpress, HY-K0301) was added to 100 μL of cell medium in each well, followed by a 2 hrs incubation at 37°C with 5% CO2. The absorbance value (OD) was then measured at 450 nm using a microplate reader.

After 24 hrs of Dox treatment, cells were harvested and stained with Annexin V-Alexa Fluor 647 (Biolegend, 640912) and 7AAD (Sigma–Aldrich, D9542) according to the manufacturer’s recommendation for apoptosis evaluation. Flow cytometry (CytoFLEX, C00445) was then used to analyse the stained samples, with each sample being tested in triplicate.

The Caspase 3/7 activity kit is a commercially available kit (KeyGEN BioTECH, KGAS037-100) used to measure the activity of caspase 3 and 7, which are key executioner caspases involved in apoptosis. The kit is designed to detect and quantify the activity of caspases 3 and 7 by measuring the enzymatic cleavage of a synthetic peptide substrate labelled with a fluorescent dye. The pictures were taken by Olympus microscope at low magnification (10x, objective), and at least three fields per sample were counted.

Data analysis and mapping were performed using GraphPad Prism 9, and all data are presented as mean ± standard deviation (SD). Significance was determined by the unpaired Student’s-test.

A Pan-cancer analysis was conducted to investigate DDX58 differential expression between tumour and normal tissues by UCSC database (https://xenabrowser.net/). We found that DDX58 expression was the highest expression level in breast cancer out of 34 types of cancer ( Figure 1A ). Of all the 34 cancer types, we also observed significant up-regulation of DDX58 expression in 19 tumours, and down-regulation in 9 tumours compared to their normal tissues ( Figure 1A ).

To further explore the potential clinical implications of DDX58 in breast cancer, we checked its prognosis significance in the Kaplan–Meier plotter dataset. Patients with low expression of DDX58 (DDX58 ^low, using median as threshold) had significantly shorter recurrence-free survival (RFS) (P=0.00078) and overall survival (OS) (P=0.01) ( Figure 1B ). This phenomenon was more obvious in TNBC subgroup patients, who have significantly shorter RFS (P=0.0096) and OS (P=0.032) in DDX58 ^low patients ( Figure 1C ). The prognosis of other BC subtypes was shown in Supplementary Figure S1 . DDX58 ^low patients only in Luminal A and HER2+ subtypes showed significantly shorter RFS (but no phenotype in OS) ( Supplementary Figure S1 ). To confirm the essential role of DDX58 in TNBC patients, we used a cohort of 196 female TNBC patients in our hospital, and semi-quantitative method for protein of immunohistochemistry (IHC) analysis was used. The baseline table of involved patients is shown in Supplementary Table S1 . Based on the modified Allred scoring system (37, 38), we divided the patients into two groups: 100/196 (51%) patients with high DDX58 expression (2≤score ≤ 3) and 96/196 (49%) patients with low DDX58 expression (0≤score<2) ( Figure 1D ). The data of our hospital again confirmed that DDX58 ^low patients in TNBC had worse prognosis for both RFS (P=0.0197) and OS (P=0.0066), which was consisting with the previous Kaplan–Meier plotter dataset results ( Figure 1E ).

To further investigate the effect of DDX58 on chemotherapy treatment efficacy in BC patients, we calculated the therapeutic effect of BC patients who underwent neoadjuvant therapy in four GEO datasets ( Figures 1F-I ). We found that the expression of DDX58 was significantly higher in patients with a pathological complete response (pCR) in 3/4 of all datasheets (GSE20194, GSE22093, and GSE163882) (*P< 0.05, **P < 0.01) ( Figures 1F-H ). Taken together, our results indicated that the low expression of DDX58 was significantly associated with worse therapeutic efficacy and poor clinical outcome. These data suggested that DDX58 might be a potential novel predictive prognostic biomarker for BC.

As DDX58 ^low patients associated with worse therapeutic efficacy and poor clinical especially in TNBC patients, we hypothesized that DDX58 play crucial roles in the process of chemotherapy. Firstly, we silenced DDX58 with specific siRNA in TNBC (MDA-MB-231), luminal (MCF-7) and HER2-positive (SKBR3) breast cancer cells. Then, we conducted cell viability assays after treatment of a panel of commonly used chemotherapy agents for BC, including microtubule stabilizer (paclitaxel), DNA topoisomerase IIα (TOP2A) inhibitor (doxorubicin; Dox), antimetabolite (5-fluorouracil; 5-FU) and DNA alkylating agent [cis-diammineplatinum(II); cisplatin].

Interestingly, when DDX58 expression was silenced in MDA-MB-231 cells, we observed significantly drug resistance after paclitaxel, Dox, or 5-FU treatment in DDX58-siRNA (DDX58-Knockdown) cells compared with the control cells (DDX58-scramble) ( Figures 2A-C ). This phenomenon was not observed in the group of cisplatin treatment ( Figure 2D ). Cisplatin bind to DNA and cause the DNA strands to crosslink, interfering the DNA replication process, subsequently inducing apoptosis, which is slightly different from the conventional paclitaxel, Dox, or 5-FU drugs (39, 40). However, for the other sub-types breast cancer cells, such as in MCF-7 cell and SKBR3, no dosage dependent manner chemo-drug resistance in DDX58 gene knockdown group was observed, neither in paclitaxel, Dox, 5-FU nor in cisplatin treatment ( Supplementary Figure S2 ).

In summary, all our data suggested that DDX58 gene knockdown led to multiple line chemo-drug resistance in TNBCs rather than other BC subtypes. This phenomenon also explained why only TNBC patients with DDX58^low had worse treatment response and poor prognosis, while the DDX58 ^low patients in other BC subtypes had no significant prognostic differences.

To further confirming the chemo-drug resistance effect in DDX58^low in TNBC cells, we first knocked out DDX58 gene in TNBC cells (MDA-MB-231) for the chemo-drug effect testing, and then we confirmed this in other TNBC cell lines by silencing DDX58. We knocked out this gene by the CRISPR/Cas9 system in MDA-MB-231 cells (DDX58-KO) ( Figure 3A ). Sanger sequencing and western blotting were used to confirm the knockout efficiency of this gene in DNA and protein level ( Figure 3A ). DDX58-KO and wildtype cells (WT) were treated with four types chemotherapy drugs mentioned above, and cell viability assays were carried out to detect drug efficiency. Our results demonstrated that DDX58-KO cells showed significant drug resistance under the treatment of three chemotherapy drugs including paclitaxel ( Figure 3B ), Dox ( Figure 3C ) and 5-FU ( Figure 3D ), except for cisplatin ( Figure 3E ). Cells were found to be the most resistant to Dox in strictly dose-dependent manner among all these drugs ( Figure 3C ). To verify this phenomenon in other TNBC cell lines, we silenced DDX58 gene by siRNA in BT-549 and MDA-MB-468 to confirm these results under treatment of Dox. Interestingly, Dox resistance was also observed in DDX58 gene silenced groups compared with in the WT group ( Figures 3F, G ). Thus, these results demonstrated that DDX58 gene deficiency caused chemo-drug resistance in multiple TNBC cell lines.

To reversely demonstrate the essential role of DDX58 gene in the process of chemo-drug treatment, we constructed a DDX58 stably overexpression cell line (DDX58-OE) and the control cell line (DDX58-OE-NC) in MDA-MB-231 cells. After four chemotherapy drug treatments, we observed that DDX58-OE cells were more sensitive to drugs than DDX58-OE-NC cells, especially in Dox treatment ( Supplementary Figure S3 ). Significantly fewer cells survived in the DDX58-OE group than the DDX58-OE-NC group within 48 hrs treatment of Dox ( Figure 3H ), and this phenomenon was also in a dose-dependent manner ( Figure 3I ). Very interestingly, survived cells tended to express fewer DDX58 protein level after a series of gradient Dox concentration treatment ( Figure 3J ). This data suggests that TNBC cells with high DDX58 expression are more sensitive to Dox treatment. In summary, using DDX58-KO/DDX58-KD and DDX58-OE in multiple TNBC cells, we demonstrated that cells with low DDX58 expression were resistant to multiple chemotherapy, e.g. paclitaxel, Dox, 5-FU, while cells with high DDX58 expression were sensitive to Dox treatment.

To further investigate the underlying mechanisms by which DDX58 affects chemotherapy resistance and prognosis in breast cancer, we conducted a functional exploration of DDX58 on the typical biological characteristics of cancer cells. Specifically, we constructed DDX58 gene knockdown (DDX58-shRNA) MDA-MB-231 cell lines. Firstly, we found that the number of monoclonal formations was inhibited in DDX58-shRNA compared to the control group ( Figure 4A ), and the similar results were further confirmed in DDX58-KO cells ( Figure 4B ), indicating that DDX58-shRNA and DDX58-KO led to impaired cancer cell growth and proliferation. The number of clones were calculated in every experimental group and the results was shown in Figure 4C . Secondly, our results indicated that the viabilities of DDX58-shRNA cells were significantly decreased, also the same results were found in DDX58-KO cells ( Figure 4D ). These above results further supporting the importance of DDX58 in maintaining cancer cell survival and proliferation.

Furthermore, we found that the migration of DDX58-shRNA cells enhanced ( Figure 4E ), and this phenomenon was further confirmed in DDX58-KO cells ( Figure 4F ). Then, we observed an enhancement in the invasion of DDX58-shRNA and DDX58-KO cells ( Figures 4G, H ). The statistical results of each experimental group were showed in Figures 4I, J . These results demonstrated that the migration and invasion of DDX58-deficient cells were enhanced. In conclusion, our data indicated that the depletion of DDX58 inhibited proliferation and promoted the migration and invasion in MDA-MB-231 cells.

DDX58, an RNA sensor, is maintained in an inactive conformation without recognizing dsRNAs (double-stranded RNA) (41). Upon encounter with the cytoplasmic dsRNAs, the receptor of DDX58 changes conformation and releases a pair of signalling domains (CARDs) that are selectively modified and interact with an adapter protein (MAVS), thereby leading to induction of Type I IFN, proinflammatory cytokines and apoptosis (41, 42). According to this, we drew the hypothesis diagram of associate signal pathway with Dox treatment in TNBC cells ( Figure 5A ). First, we checked the endogenous expression level of dsRNAs in MDA-MB-231 cells following 6 hrs Dox treatment. Poly(I:C) (10 μg/ml) was used as positive control. After 6 hrs Dox treatment, the results of immunofluorescence (IF) staining of dsRNAs revealed a significant dose-dependent increasing (8-fold) in MDA-MB-231 cells (Dox concentration: 1 µM) ( Figures 5B, C ). To further investigated the downstream signalling pathway of DDX58 involved after 6 hrs Dox treatment, we performed double fluorescence staining of MAVS and DDX58. The results revealed that the expression of MAVS was significantly induced ( Figures 5D, F ) and two proteins (DDX58, MAVS) showed the increased co-localization after Dox treatment ( Figures 5D , 4E ). Moreover, the data of RT-qPCR indicated that the expression of DDX58 and downstream Type I IFN signalling pathway genes significantly increased under 6 hrs Dox treatment, suggesting that Dox treatment could activate the DDX58-Type I IFN signalling pathway in MDA-MB-231 cells ( Figure 5G ). Taken together, our results confirms that Dox treatment elevates the level of endogenous dsRNAs, which then activates DDX58. The activated DDX58 interact with MAVS and the subsequent Type I IFN signalling pathway were induced.

Previous studies have reported the activation of DDX58 and Type I IFNs signalling, which cause damaged cells apoptosis (10, 12–16, 43–45). Our above data also confirmed that the DDX58 signalling and downstream Type I IFN signalling could be activated by Dox treatment in TNBC cells. These evidences indicated that the deficiency of DDX58 might decrease cell apoptosis induced by Dox treatment, and finally leading to chemotherapy resistance. To further explore specific molecular event, we performed apoptosis assays in DDX58-KO and WT cells. Specifically, we utilized a caspase 3/7 kit and flow cytometry measurements for apoptosis testing. Detection of caspase 3/7 is a commonly used method for assessing apoptotic activity in experimental cell populations. Our data indicated that caspase 3/7 activity in DDX58-KO cells did not change significantly with an increasing Dox dosage concentration, however, the activities of WT cells significantly increased in a Dox treatment dose-dependent manner ( Figures 6A-C ). Moreover, without Dox treatment, the results of flow cytometry revealed that there was no significant difference in early and later apoptosis between DDX58-KO and WT cells ( Figures 6D, F ). Under Dox treatment, the later phase apoptosis rate of DDX58-KO cells was significantly lower than that of WT cells ( Figures 6E, F ).

IRF3/7 (interferon regulatory factor 3/7) is transcriptionally induced in many cell types as a target of Type I IFN signalling pathway (9, 46, 47). Our western blotting data showed that the expression of IRF3/7 did not change significantly without Dox treatment ( Figure 6G ). Nevertheless, under the treatment of Dox, the expression of IRF3/7 and cleaved caspase 3 increased in WT cells, comparing to DDX58-KO cells ( Figure 6G ). These results further indicated that DDX58-Type I IFN signalling was activated by Dox. Apoptosis is also commonly reported to be the downstream of DDX58-Type I IFN signalling pathway. Therefore, our data indicated that knockout of DDX58 ultimately interrupted Dox-induced apoptosis, which was previously confirmed activated by DDX58-Type I IFN signalling pathway.

To further reveal the role of DDX58 in the process of Dox treatment in vivo. We established a cell line-derived xenograft (CDX) model in BALB/c nude mice using DDX58-KO and WT cells. The experimental design and Dox therapy regimen are depicted in Figure 7A . On days 24 and 27 post tumour implantation, the tumour volumes in the DDX58-KO group were significantly increased compared to those in the WT group (P< 0.05) ( Figure 7B ). This data indicates that DDX58 knockout can promote tumorigenesis in vivo. We next evaluated the impact of DDX58 in the process of Dox treatment based on these CDX model. After 12 days Dox treatment, the tumour volume of the DDX58-KO group was significantly greater than that of the WT group, with a 2.5-fold increasing ( Figure 7C ). The tumour growth curves of each nude mouse from the four groups (DDX58-KO, WT, DDX58-KO (Dox+), and WT (Dox+)) were shown in Figure 7D . Following 12 days of Dox treatment, tumours were harvested and photographed ( Figure 7E ). From these tumours, we found that the tumour growth inhibition rate (IR) of the DDX58-KO group (43%) decreased by 25% compared with that of the WT group (68%) ( Figure 7F ). These data, consistent with the results in vitro, revealed that knockout of DDX58 leads to Dox chemotherapy resistance in vivo.

To further investigate the molecular event of Dox resistance associated with DDX58 in vivo, we detected apoptosis-related markers in tumours ( Figure 7G ). The IF results demonstrated that there was a marked reduction in the percentage of TUNEL-positive cells in the DDX58-KO group under Dox treatment ( Figures 7G, H ). Conversely, the percentage of Ki67-positive cells in the DDX58-KO group increased significantly compared to that in the WT group within Dox dosing ( Figures 7G, I ). Furthermore, the expression of interferon-beta (IFN-β) (a marker of the Type I IFN signalling pathway) decreased significantly in the tumours of the DDX58-KO group compared to the WT group following Dox treatment ( Figures 7G, J ). We further used RT–qPCR to check the expression of marker genes of DDX58-Type I IFN signalling pathways with Dox treatment in vivo ( Figures 7K, L ). Among the groups without Dox treatment, we found that marker genes from DDX58-Type I IFN signalling pathway had no significant differences in DDX58-KO group compared to the WT group. However, under the treatment of Dox, the expression of all markers significantly decreased in the DDX58-KO group compared to WT group ( Figure 7L ). These results further confirmed that the DDX58 signalling pathway and downstream Type I IFN signalling pathway could be activated by Dox treatment in vivo.

Taken together, our findings provide evidence that DDX58 knockout promotes tumorigenesis and impedes the therapeutic efficacy of Dox in TNBC. The observed effects of DDX58 knockout highlight the potential role of DDX58 as a therapeutic target for TNBC treatment.