Previously, we established a molecular classification by dividing MM into two subtypes with distinct prognosis on the basis of a gene module co-expressed with MCL-1 (6). The survival analysis of all patients with anthracycline-based therapy suggested that the MCL1-M low group had a markedly better overall survival (OS) compared to the MCL1-M high group. We performed a Gene Ontology (GO) analysis with differentially expressed genes in the low and high MCL1-M groups and found that the upregulated genes in the low MM MCL1-M group were significantly enriched in immune-related biological processes, especially in the type I interferon response ( Figure 1A ), In contrast, the upregulated genes in the high MM MCL1-M were most significantly enriched in the cell cycle, DNA replication, and DNA repair ( Figure 1B ). Given that doxorubicin is a common treatment for MM (18–20), we wanted to assess the IC50 of this chemotherapy drug in two MCL1-M subtypes. We trained the predictive model on the data set from the GDSC cell line by ridge regression (21, 22). We estimated the IC50 of doxorubicin for each sample in GSE19784 and GSE24080 based on the predictive model. We observed a significant difference in doxorubicin sensitivity between low- and high-risk MCL1-M MM ( Figures 1C, D ). Obviously, low MCL1-M MM had significantly lower IC50 values for doxorubicin than those in high MCL1-M MM, indicating that patients in the low-risk MCL1-M group were more sensitive to doxorubicin. These results indicate that the MCL1-M is a unique genomic classifier that enables identifying the low-risk group with a unique biological process of activation of the interferon response, and thus predicting sensitivity to doxorubicin.

To further explore the relationship between interferon response and doxorubicin sensitivity, we first assessed the effects of doxorubicin treatment on 5 MM cell lines for 24 or 48 hours. Subsequent cell viability assays revealed that different cell lines had different survival rates at the same concentration and treatment time ( Figure 2A ). According to the cell survival curve, the inhibitory effects of doxorubicin on myeloma cell lines were time-dependent and dose-dependent, in which H929 was drug-sensitive cell line and OPM-2 was drug-insensitive cell line. A previous study showed that anthracyclines stimulate malignant cells to rapidly produce type I IFNs rapidly (16). By binding to type I IFN receptors on neoplastic cells, type I IFNs trigger autocrine and paracrine circuits that result in the release of chemokines (16). Tumors lacking type I IFN receptors did not respond to doxorubicin (16). We then examined the proliferation of myeloma cells at different concentrations of IFN-α treated for different times. We found that IFN-α significantly increased the percentage of apoptosis in the five MM cell lines, in which H929 was a drug-sensitive cell line and OPM-2 was drug-insensitive cell line ( Figure 2B ). Together, these results showed that doxorubicin-sensitive MM cell lines responded to IFN-α. We combined the chemotherapy drug with IFN-α to treat sensitive and resistant cell lines ( Figure 2C ). Cell viability analysis revealed that the proliferation of drug-sensitive H929 cells was significantly inhibited by the combination with IFN-α, while the growth of drug-insensitive cells was not affected ( Figures 2D, E ). Subsequently, we treated myeloma cells with doxorubicin(0.8μM), IFN-α(2000U/ml), doxorubicin(0.4μM) + IFN-α(1000U/ml) for 24 hours and 48 hours, respectively. Our flow cytometric apoptosis assay showed that the combination of drugs with IFN-α significantly increased the rate of apoptosis in drug sensitive cells but did not affect apoptosis in drug-insensitive cells ( Figures 2F, G ), suggesting that doxorubicin-sensitive MM cell lines responded to IFN-α and combined therapy could further promote apoptosis of myeloma cells in vitro.

To elucidate the mechanism underlying the anti-myeloma effects of anthracyclines, we performed RNA sequencing (RNA-seq) using H929 cells from the doxorubicin sensitive cell line that were treated with doxorubicin (0.4 μM) for 48 hours. RNA-seq analysis showed that after doxorubicin treatment a series of genes associated with the interferon response were upregulated in H929 cells ( Figure 3A ). The relevant gene names and expressions have been included in the Supplementary Table 1 .The significant impact of doxorubicin on IFN-regulated genes was further confirmed by GSEA ( Figure 3B ). These results suggest that type I IFN signaling may be associated with the anti-myeloma effect of doxorubicin. Using qRT-PCR, we confirmed that the expression of genes related to the interferon response was indeed up-regulated in H929 after doxorubicin treatment ( Figure 3C ). In contrast, these genes were not significantly upregulated in the drug-insensitive cell line OPM-2 after doxorubicin treatment ( Figure 3C ). These findings indicate that the type I IFN response in MM cells triggered by doxorubicin exerts an important antitumor effect and loss of this response can result in resistance to chemotherapy.

Based on preclinical findings, the DNA damage repair process could trigger interferon-related immune responses (17, 23, 24). It is well known that anthracyclines could inevitably damage nuclear DNA (25–28). In our study, down-regulated differential expression genes in H929 after doxorubicin treatment were significantly enriched in DNA damage and DNA repair-related pathways ( Figure 3D ). This result indicated that DNA was seriously damaged and repair function was impaired in drug-sensitive cell lines. To investigate whether the degree of DNA damage was different in MM cells with different doxorubicin sensitivity, immunofluorescent staining of γ-H2AX was performed. Phospho-H2AX or γ-H2AX- is a marker of double-stranded DNA breaks (29, 30). Obviously, we observed higher fluorescence signals for γ-H2AX in H929 cells compared to the drug-insensitive cell line OPM-2 ( Figure 3E ), indicating that DNA damage was more severe in drug-sensitive cell lines, whereas drug-insensitive cell lines had strong DNA repair ability in the face of damage events. To confirm whether DNA damage caused by anthracyclines produces dsDNA (double-stranded DNA) that activates the interferon response in drug-sensitive cell lines. We stained myeloma cells with a dsDNA-specific antibody after anthracycline treatment and found that the accumulation of dsDNA in the cytoplasm was more evident in H929 cells than in OPM-2 cells ( Figure 3F ). These results confirmed that drug-insensitive cell lines have a better ability to repair damaged DNA in time, thus avoiding the interferon response.

Multiple myeloma cell lines MM.1S and OPM-2 were purchased from the Cell Bank (Chinese Academy of Sciences, Beijing, China). RPMI-8226, H929, and U-266 cells were kindly provided by Dr. Yang Yuan (Guizhou Provincial People’s Hospital). Cell lines were cultured in RPMI1640 medium (GIBCO; Thermo Fisher Scientific) supplemented with 10% FBS (GIBCO; Thermo Fisher Scientific), 100 U/mL of penicillin, and 100 mg/mL of streptomycin (Sigma-Aldrich). Doxorubicin was purchased from Selleckchem and resuspended in DMSO. Interferon-α (IFN-α, SRP4595, Sigma) was reconstituted in water. The Trizol RNA extraction reagent was purchased from Thermo Fisher Scientific. Anti-γ-H2AX (ab81299) was purchased from Abcam (Cambridge, UK). Anti-dsDNA (SC-58749) was purchased from Santa Cruz Biotechnology.

To evaluate the antiproliferative effects of doxorubicin and IFN-α, MM cell lines (5×10³ to 1×10⁴ cells/well in 96-well plate) were treated with a single drug or a combination of the two drugs or with DMSO for 24 or 48 hours. Cell viability was evaluated using a Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) according to the manufacturer’s instructions. The CI value was evaluated by CompuSyn software.

Apoptosis was evaluated by Annexin-V/PI staining and flow cytometric analysis using an ApoScreen Annexin V Apoptosis Kit (Southern Biotech, Birmingham, AL, USA) according to the manufacturer’s instructions. Data were analyzed using FlowJo software version 10.

Total RNA from multiple myeloma cells was extracted with TRIzol (Thermo Fisher Scientific) following the manufacturer’s instructions. Subsequently, reverse transcription of 1 µg RNA into cDNA was carried out using a Hiscript II Q RT SuperMix reagent kit (Vazyme, China). The RT-qPCR assay was then performed with this mixture using SYBR Green PCR Master Mix (Roche Diagnostics, Basel, Switzerland) in a 96-well PCR plate (Nest Biotechnology, Wuxi, China).

The cells were seeded in 6-well plates at a density of 2×10⁵ cells/well and cultured at 37°C for 24 hours. Doxorubicin at different concentrations was added to the 6-well plates for 24 hours. After treatment, cells were collected and incubated in poly-L-lysine-coated slides for 1 h at 37°C. The cells were then fixed in 4% paraformaldehyde for 20 min at room temperature, permeabilized with 0.5% Triton X-100 for 15 min and blocked with 1% BSA dissolved in PBS for 1 h. Cells were then incubated with anti-γ-H2AX or a primary antibody against dsDNA overnight at 4°C, followed by goat anti-mouse IgG conjugated secondary antibodies for 1 h at room temperature in the dark. The nuclei were stained with 0.1 μg/mL DAPI for 5 min. The cells were viewed using a fluorescent microscope.

H929 cells were cultured for 48 hours in the presence or absence of doxorubicin. RNA was extracted and used for library construction and then submitted to the Illumina Novaseq 6000 platform for sequencing. Differentially expressed genes (DEG) before and after drug treatment were identified using the DESeq2 package (version 1.36.0) (40). GO and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were employed to distinguish the different biologic functions and pathways between DEGs. Gene set enrichment analysis was performed using the Molecular Signature Database (MSigDB) (41). The R package ‘OncoPredict’ of 402 drugs was used to predict in vivo drug responses in patients with high and low MCL1-M MM (21).

The statistical analysis was performed by using GraphPad Prism 9.0 (San Diego, CA, USA). The significance of differences was determined by two-tailed paired Student’s t-test. Data shown in the study were obtained from at least three independent experiments and probability values of p<0.05 were considered to be statistically significant.