PRKDC Induces Chemoresistance in Osteosarcoma by Recruiting GDE2 to Stabilize GNAS and Activate AKT

Osteosarcoma is the most common form of primary bone malignancy worldwide, mostly occurring in adolescents and children (1, 2). The mainstay of osteosarcoma treatment is surgery and combined neoadjuvant chemotherapy, in which the backbone of first-line chemotherapy comprises doxorubicin (DOX), cisplatin, methotrexate, and ifosfamide. Despite increased understanding of the tumor microenvironment (3–5) and molecular pattern (6–8), the prognosis for osteosarcoma has not improved over the last 2 decades (9). The development of drug resistance is the most common driver of unfavorable outcomes (10). In patients with drug-resistant disease, additional doses or drugs could not markedly improve prognosis because of the occurrence of local recurrence and metastasis (11). Therefore, an urgent clinical need to reveal the underlying mechanisms of drug resistance and exploit new strategies to increase drug sensitivity is observed.

DOX, a member of the anthracycline family, suppresses tumor growth by interacting with DNA and DNA topoisomerase II (12). The chemoresistance of DOX involves various molecular mechanisms, including drug efflux (13), enhancement of DNA repair (14), metabolic alteration (15), and activation of kinase-related pathways. It is well documented that the PI3K/AKT pathway mediates resistance to DOX (16, 17). DOX induced AKT activation and PI3K/AKT inhibitors, which dramatically enhances its cytotoxicity. Several molecules have been demonstrated to regulate the PI3K/AKT pathway to mediate DOX sensitivity. UDP-glucose ceramide glucosyltransferase, a key enzyme for the synthesis of glycosylated sphingolipids, activates AKT by altering the composition of glycosphingolipid-enriched microdomains (18). LRP1B directly binds to NCSTN and regulates its protein expression, subsequently modulating the PI3K/AKT pathway and affecting the DOX resistance (19). Thus, further exploration of the regulatory network of the PI3K/AKT pathway could provide novel targets for the reversion of DOX resistance in osteosarcoma.

PRKDC encodes the catalytic subunit of DNA-dependent protein kinase (DNA-PK), which is a serine/threonine–protein kinase that plays a critical role in cancer biology. Evidence has shown that PRKDC is overexpressed and phosphorylated in various solid tumors, leading it to be proposed as a potential therapeutic target (20, 21). More recently, PRKDC has been identified as a critical modulator of tumor sensitivity to radiotherapy and DNA-damaging agents (22–24). PRKDC also participates in the innate immune response, and targeting PRKDC could improve checkpoint blockade immunotherapy (25). Despite the crucial role of PRKDC in the cellular processes and drug resistance of cancers, the underlying mechanism remains unclear.

Glycerophosphodiester phosphodiesterase 2 (GDE2), also known as glycerophosphodiester phosphodiesterase domain containing 5, is a six-transmembrane protein that engages in neuronal differentiation (26). Matas-Rico and colleagues found that GDE2 promoted neuroblastoma differentiation and was associated with poor prognosis (27). Since then, GDE2 has been illustrated as a critical driver for various tumors, including colorectal cancer (28), nasopharyngeal carcinoma (29), and gastric cancer (30). Furthermore, GDE2 was reported to mediate 5-Fu and cisplatin resistance (28, 29), but the regulatory mechanism remains unclear and needs further exploration. Guanine nucleotide-binding protein-G(S) subunit alpha (GNAS) encoding the alpha-subunit of the stimulatory G protein (Gsα) was originally discovered to regulate neurotransmitters and many hormones (31). Recently, GNAS mutants were reported to drive tumorigenesis of pancreatic cancer by modulating PKA-mediated SIK suppression (32). Moreover, the activating mutants of GNAS have been demonstrated to be linked to a subgroup of inflammatory liver tumors with STAT3 activation (33). GNAS has also been shown to be genetically activated in a small subset of small cell lung cancer, in which it promotes tumor progression via the PKA/PP2A axis (34). However, whether GNAS is associated with chemoresistance in cancer remains unclear.

In this study, we identified PRKDC as a critical driver of doxorubicin (DOX) sensitivity in osteosarcoma through a CRISPR screen. In terms of the underlying mechanism, our results revealed that PRKDC was highly activated in DOX-treated cells and chemoresistant osteosarcoma. We demonstrated that PRKDC could activate the PI3K/AKT pathway by recruiting GDE2 to stabilize GNAS, which ultimately resulted in DOX resistance in osteosarcoma. These findings, for the first time, suggest that the PRKDC/GDE2/GNAS/AKT axis has a critical role in regulating DOX sensitivity in osteosarcoma and provides novel targets for enhancing DOX sensitivity.

Osteosarcoma samples were collected from The Second Xiangya Hospital, Central South University. Written informed consent was acquired from all patients before surgery. The studies were conducted in accordance with recognized ethical guidelines (Declaration of Helsinki) and approved by the Ethics Committees of The Second Xiangya Hospital (Approval No. 2023-Z0477).

We collected specimens of needle biopsy and surgical resection from patients diagnosed with osteosarcoma before and after standard MAP chemotherapy (consisting of methotrexate, DOX, and cisplatin). The treatment efficacy was assessed by CT or MRI according to RECIST 1.1 (35). Specimens from patients with progressive disease and partial response were collected for further analysis. We defined patients who achieved a partial response as chemotherapy sensitivity and patients with progressive disease as chemoresistance.

The 143B, HOS, U2OS, MG63, A549, PC-9, HepG2, Hep3B, 22Rv1, C4-2, and HEK293T cell lines were purchased from Procell Life Science & Technology Co., Ltd. The 143B and HEK293T cells were cultured in DMEM medium (Procell Life Science & Technology); the HOS, MG63, HepG2, and Hep3B cells were cultured in MEM (Procell Life Science & Technology); the A549, PC-9, 22Rv1, and C4-2 cells were cultured in RPMI1640 medium (Procell Life Science & Technology); and the U2OS cells were cultured in McCoy’s 5A medium. All culture medium was supplemented with 10% fetal bovine serum (FBS; NEWZERUM Ltd.). All cells were incubated at 37°C in 5% CO₂. Cells were tested for Mycoplasma contamination every 3 months using the TransDetect PCR Mycoplasma Detection Kit (TransGen Biotech). All cell lines were identified using short tandem repeat profiling.

Total RNA was extracted from osteosarcoma cells using a standard TRIzol (Thermo Fisher Scientific) protocol. Nano-Drop 2000 (Thermo Fisher Scientific) was used to measure the concentration and quality of RNA. RNA was reverse-transcribed and amplified using Hifair III Reverse Transcriptase (Yeasen Biotechnology Co., Ltd.). RT-qPCR reactions were performed on Quantstudio5 (Applied Biosystems) using SYBR Green Master Mix (#AG11746, Accurate Biology). The relative gene expression levels were calculated by the 2^−ΔC_t method after normalizing to GAPDH levels. The sequences of all primers are provided in Supplementary Table S1.

Cells were lysed with RIPA buffer (#P0013, Beyotime), and the protein concentration was measured using a Micro BCA Protein Assay Kit (Sigma-Aldrich). Subsequently, the proteins were separated by SDS-PAGE and electroblotted onto PVDF membranes. The membrane was blocked with 5% skimmed milk and incubated with the corresponding primary antibody at 4°C overnight. After washing with TBST three times, the membrane was incubated with a secondary antibody for 1 hour at room temperature. The membrane was then washed three times with TBST, and enhanced chemiluminescence (ECL) was added dropwise onto the PVDF membrane. The protein signal was detected by Bio-Rad Image Lab.

The following antibodies were used: DNA-PKcs (#ab32566, abcam, 1:5,000 dilution); DNA-PKcs S2056 (#ab124918, abcam, 1:5,000 dilution); pAKT-S473 (#4060S, Cell Signaling Technology, 1:1,000 dilution); pAKT-T308 (#13038, Cell Signaling Technology, 1:1,000 dilution); ENO1 (#11204–1-AP, Proteintech, 1:2,000 dilution); GDE2 (#25703-1-AP, Proteintech, 1:2,000 dilution); GNAS (#10150–2-AP, Proteintech, 1:1,000 dilution); beta-actin (#66009–1-Ig, Proteintech, 1:5,000 dilution); GAPDH (#60004-1-Ig, Proteintech, 1:5,000 dilution); BAX (#50599-2-Ig, Proteintech, 1:1,000 dilution); BCL2 (#68103-1-Ig, Proteintech, 1:1,000 dilution); HRP conjugated goat-anti-mouse antibody (#511103, Zenbio, 1:10,000 dilution); and HRP conjugated goat-anti-rabbit antibody (#511203, Zenbio, 1:10,000 dilution).

For the colony formation assay, osteosarcoma cells were seeded into six-well plates at a concentration of 1000 cells per well. The cells were then cultured for 2 weeks until a colony was clearly formed. The cells were washed with PBS three times and stained with 0.1% crystal violet. The number of colonies was counted using ImageJ software.

The CCK-8 assay was conducted as previously reported (36). In brief, osteosarcoma cells were seeded into 96-well plates at a concentration of 2000 cells per well and incubated overnight. Corresponding drugs were added into each well and incubated for the indicated time. Next, 10 μL of sterile CCK-8 was added to each well and incubated at 37°C for 1 hour before measuring the absorbance at 450 nm.

The siRNAs or plasmids were incubated with Lipofectamine 2000 (Thermo Fisher Scientific) in serum-free Opti-MEM medium for 15 min. Then, the mixture was carefully added to plates or dishes. After 6 hours of transfection, the Opti-MEM medium was replaced with complete DMEM. Cells were collected 72 hours after transfection for follow-up experiments. The siRNA sequences of PRKSC, GDE2, and GNAS are listed in Supplementary Table S1.

To establish a stable CRISPR/Cas9 knockout cell line, sgRNA plasmids were co-transfected with package plasmids (pMD2.G and psPAX2) into HEK293T cells using lipo2000 (Thermo Fisher Scientific). After 48 hours, the supernatant was collected and concentrated using Amicon Ultra Centrifugal Filters (Millipore). For viral transduction, the virus was added to the culture medium with Polybrene (Beijing Solarbio Science & Technology). After 48 hours of transduction, the cells were selected with 2 μg/mL puromycin for 7 days. LentiCRISPR v2 (#52961, Addgene) was provided by Feng Zhang's laboratory. The PRKDC sgRNA sequences were obtained from our customized Human Brunello kinome pooled library (SYNBIO Technologies) and are provided in Supplementary Table S2.

The human kinome CRISPR knockout library was customized by SYNBIO Technologies. This library targets 507 human kinases. For each gene, 10 corresponding sgRNAs were designed and synthesized. A total of 164 sgRNAs that did not target any gene were used as controls. The sequences of all sgRNAs are provided in Supplementary Table S2. The workflow of the CRISPR screen is illustrated in Fig. 1A. In brief, cells were transduced with the kinome CRISPR knockout library at a low MOI of 0.3 and coverage of 500×. The transduced cells were then selected with 2 μg/mL puromycin for 7 days to remove the negative cells. The mutant cell pool was treated with DOX (0.1 µmol/L) or DMSO for another 7 days. At the end of treatment, 3 × 10⁶ cells were collected from each group for DNA extraction. The sgRNAs were amplified using NEBNext High-Fidelity 2× PCR Master Mix and sequenced by Genergy Biotechnology. The results were analyzed using the MAGeCK v0.5.7 algorithm (37).

Paraffin-embedded sections of clinical and mouse samples were used for IHC staining following a previously reported protocol. The clinical samples were stained with DNA-PKcs S2056 (#ab124918, Abcam), and the mouse samples were stained with Ki-67 (#ab16667, Abcam) and cleaved caspase-3 (#9661, Cell Signaling Technology). The IHC scores were independently assessed by two senior pathologists using a previously reported scoring method (38).

The Annexin V-FITC/propidium iodide (PI) assay was used to measure the apoptosis rate according to the manufacturer’s instructions. Briefly, cells were stained with Annexin V-FITC and PI from the Annexin V-PE/PI Apoptosis Detection Kit (#40302ES60, YEASEN) for 10 minutes before detecting using a Cytek NL-CLC. For Ki-67 detection, cells were fixed and permeated using the eBioscience Foxp3/Transcription Factor Staining Buffer Set (Invitrogen) before staining with Ki-67 antibodies (Biolegend) for 30 min. To detect cell surface markers, cells were directly incubated with APC anti-CD11b (#101211, Biolegend), APC/Fire anti-CD45 (103153, Biolegend), and PE anti-CD3 antibodies (#100205, Biolegend) for 30 min at 4°C. After washing three times, the cells were subjected to flow cytometry analysis. FlowJo software (FlowJo) was used to analyze the data.

The comet assay was conducted as previously reported (39). In brief, cells were digested and mixed with low-melting point agarose before spreading the cell suspension onto a microscope slide pre-coated with a thin layer of agarose. The cells were then lysed and subjected to electrophoresis in an alkaline buffer. After neutralization, DNA was stained using PI (#C2041S-5, Beyotime), and the DNA fragments were visualized under a fluorescence microscope.

Total proteins were extracted using IP buffer and incubated with protein A + G beads (#P2029, Beyotime) and IgG (#A7007, Beyotime) or primary antibody for 24 hours. The beads were collected and washed with PBS three times. After boiling with an SDS-loading buffer for 10 minutes, the eluted protein was analyzed by gel electrophoresis and liquid chromatography–tandem mass spectrometry (LC–MS/MS).

All animal experiments were approved by the Institutional Animal Care and Use Committee of The Second Xiangya Hospital, Central South University under Approval No. 20230301. BALB/c nude mice (4–5 weeks old) were purchased from Hunan SJA Laboratory Animal Co., Ltd., and housed under pathogen-free conditions. Human osteosarcoma 143B cells (5 × 10⁶) were subcutaneously injected into the right posterior flank of mice to establish a subcutaneous xenograft tumor. Mice were randomly allocated into different groups once the tumor volume reached 50 to 100 mm³ and treated with the indicated drugs (DOX, 2 mg/kg, twice a week, intraperitoneally; AZD7648, 70 mg/kg, once a day, orally). The tumor volume was calculated using the formula of π/6 × length × width². At the end of the experiments, the mice were euthanized, and the tumors were excised and weighed.

To analyze the safety of drug combination treatment, BALB/c mice (4–5 weeks old) were randomly allocated into different groups and administered DOX, AZD7648, or their combination (DOX, 2 mg/kg, twice a week, intraperitoneally; AZD7648, 70 mg/kg, once a day, orally). At the end of treatment, the bone marrow, serum, liver, lung, heart, spleen, and kidney samples were collected for further analysis. Specifically, the bone marrow was digested into individual cells and stained for flow cytometry analysis. Serum markers of liver, kidney, and heart damage were detected using a relative assay kit (AST: #C010-2-1, Nanjing Jiancheng Bioengineering Institute, ALT: #C009-2-1, Nanjing Jiancheng Bioengineering Institute, LDH: #A020-2-2, Nanjing Jiancheng Bioengineering Institute, CK-MB: E-EL-M0355, Elabscience). Organic injury to the liver, lung, heart, spleen, and kidney was assessed by pathological analysis.

Patient-derived organoid models were derived from patients with osteosarcoma in our cancer center, following a previously described protocol (40). Briefly, fresh specimens were washed and fragmented into smaller sections. The minced tissues were further subjected to digestion. After filtration and centrifugation, the sediment was resuspended in a complete organoid culture medium. These cell clusters were further cultivated within a 96-well low-adsorption plate using a 3D-shaker system. The cell viability within organoids was assessed using the CellTiter-Glo 3D Cell Viability Assay Kit (Promega).

The Chou–Talalay combinatorial index (CI) was also used to evaluate drug synergies between DOX and AZD7648. The CI value was calculated using the following formula: CI = DA/da + DB/db, in which da and db are the IC₅₀ doses of the single drugs, and DA and DB refer to the concentration of each drug in combination treatment in which the combination effect reached IC₅₀.

GraphPad Prism 9 was used for the statistical analyses. The experimental data are presented as the mean ± SD. Unpaired two-sided Student t test or one-way ANOVA was used for comparison between groups. Statistical significance was defined as P-values < 0.05.

The datasets used and/or analyzed during the current study are available from the corresponding authors on reasonable request. The data generated in this study are publicly available in Gene Expression Omnibus at GSE268415.

To explore the key kinases involved in DOX sensitivity in osteosarcoma, we conducted a kinome-wide CRISPR screen in 143B cells (Fig. 1A). We hypothesized that the knockout of a DOX-resistant gene could enhance the efficacy of DOX in tumor suppression. Under DOX treatment, cells carrying sgRNAs targeting core sensitive genes will be negatively selected. After selection, more than 500× coverage of cells was collected and sequenced, and 100% of the sgRNA sequences were retained in the DMSO-treated samples, which ensured the effectiveness of our screen. As a result, we identified six kinases that were significantly negatively enriched (log₂FC > 0.5; P < 0.05) in DOX-treated cells compared with the vehicle control, including PRKDC, CSNK2A1, PIK3R4, MAPK3, ILK, and ICK (Fig. 1B), in which PRKDC was the most negatively selected gene, with all targeting sgRNAs decreased in the DOX treatment group (Fig. 1C). Thus, the loss of PRKDC could sensitize osteosarcoma cells to DOX treatment.

We next detected the level of DNA-PKcs protein and DNA-PKcs Ser2056 phosphorylation to explore the response of PRKDC to DOX treatment. We demonstrated an elevated expression of DNA-PKcs Ser2056 phosphorylation, but not DNA-PKcs protein (Fig. 1D), indicating that DOX treatment activated PRKDC in osteosarcoma. We also detected the level of DNA-PKcs Ser2056 phosphorylation in paired pre- and post-chemotherapy osteosarcoma tissues by IHC, and the results showed that DNA-PKcs Ser2056 phosphorylation was significantly increased in post-chemotherapy tissues (Fig. 1E and F). Meanwhile, we assessed the chemotherapeutic outcome of patients with osteosarcoma from our cancer center according to RECIST 1.1. The DNA-PKcs Ser2056 phosphorylation level was evaluated in tumor tissues from 15 pairs of patients with a status of progressive disease or partial response. We found that DNA-PKcs Ser2056 was upregulated in tissues from the progressive disease group (Fig. 1E and F). These findings suggest that PRKDC plays a critical role in the response to DOX treatment and may promote the development of DOX resistance.

To validate the impact of PRKDC on DOX sensitivity in osteosarcoma, we constructed a sgRNA vector targeting PRKDC and generated stable PRKDC knockout cells. Western blotting confirmed that both sgRNAs were able to eliminate the expression of DNA-PKcs protein in the 143B, HOS, U2OS, and MG63 cell lines (Fig. 2A). The PRKDC knockout showed no obvious impact on cell growth (Supplementary Fig. S1A), whereas PRKDC knockout significantly suppressed cell proliferation and induced cell apoptosis under the treatment of DOX (Fig. 2A–D; Supplementary Fig. S1B). Hence, the PRKDC knockout could significantly enhance the sensitivity of osteosarcoma to DOX treatment. We also explored the sensitivity of PRKDC knockout cells to other chemotherapy drugs for osteosarcoma, including cisplatin, methotrexate, and sorafenib. We found that the PRKDC knockout did not affect the sensitivity of osteosarcoma to these drugs (Supplementary Fig. S1C), indicating that the PRKDC knockout specifically increased the DOX sensitivity. Subsequently, we examined other chemotherapy drugs that induce DNA double-strand breaks, including etoposide and teniposide. The results showed that the PRKDC knockout also significantly increased the sensitivity of osteosarcoma to them (Supplementary Fig. S1D).

We next confirmed these observations in an in vivo model. To this end, 143B cells with or without PRKDC knockout were inoculated in the right posterior flank of the BALB/c nude mice. When the tumor volume reached 50 to 100 mm³, the mice were treated with DOX or vehicle control for 14 days. The result confirmed that the depletion of PRKDC significantly increased the sensitivity of the osteosarcoma model mice to DOX treatment (Fig. 2E–H). In addition, the group with the PRKDC knockout and DOX treatment showed significantly reduced Ki-67-positive cells and increased cleaved caspase-3-positive cells in osteosarcoma tissues compared with the other groups (Supplementary Fig. S1E). These findings suggest that PRKDC is an important driver of DOX sensitivity in osteosarcoma.

AZD7648 is a small molecule inhibitor specific for PRKDC inhibition. We next tested whether AZD7648 could work synergistically with DOX in the suppression of osteosarcoma. The Chou–Talalay method was used to evaluate the synergistic effect. The results revealed that AZD7648 and DOX synergistically inhibited the growth of four osteosarcoma cell lines at multiple concentrations (Fig. 3A), indicating that AZD7648 is a sensitizer for DOX in osteosarcoma. Similarly, we found that AZD7648 treatment did not significantly increase the sensitivity of osteosarcoma cells to cisplatin and methotrexate (Supplementary Fig. S2A and S2B). Moreover, the combination treatment of AZD7648 and DOX significantly impeded cell proliferation and induced cell apoptosis compared with single drug treatment (Fig. 3B and C; Supplementary Fig. S2C).

Next, we performed in vivo validation in a subcutaneous xenograft mouse model. When the tumor reached 50 to 100 mm³, the mice were randomly divided into four groups and treated with vehicle, DOX, AZD7648, or the combination of DOX and AZD7648. DOX treatment alone suppressed tumor growth, whereas the combination of DOX and AZD7648 completely halted osteosarcoma growth in vivo (Fig. 3D–G). The tumor tissues from the combination treatment group showed a significantly decreased expression of Ki-67 and increased levels of cleaved caspase 3 (Supplementary Fig. S2D). To further validate the efficacy of this combination, we established two cases of osteosarcoma organoids (Fig. 3H), which were treated with vehicle, DOX alone, AZD7648 alone, or a combination of DOX and AZD7648. Remarkably, the combination therapy demonstrated significant inhibition of the growth of osteosarcoma organoids (Fig. 3I). We next explored the impact of PRKDC inhibition on DOX across various solid tumors, including lung, liver, and prostate cancers. Intriguingly, our findings revealed that PRKDC inhibition markedly heightened the responsiveness of these tumors to DOX (Supplementary Fig. S3A–S3C). Notably, the combined administration of DOX and AZD7648 exhibited synergistic suppression of these tumors (Supplementary Fig. S3D–S3F).

To assess the safety profile of the DOX and AZD7648 combination, we conducted in vivo toxicity tests in BALB/c mice. The mice were randomly divided into four groups and subjected to corresponding drug treatments for 14 days. At the end of the experiment, the bone marrow, blood plasma, heart, liver, kidney, lung, and spleen samples were collected for further analysis (Supplementary Fig. S4A). The results revealed that DOX treatment alone induced bone marrow suppression (Supplementary Fig. S4B), nephrotoxicity (Supplementary Fig. S4C), and cardiotoxicity (Supplementary Fig. S5A), whereas it had no significant impact on the spleen (Supplementary Fig. S5B), lung (Supplementary Fig. S5C), and liver (Supplementary Fig. S5D). Importantly, compared with DOX treatment alone, the combination of DOX and AZD7648 did not remarkably exacerbate these toxic effects. Taken together, these findings demonstrate that the combination treatment achieved both satisfactory efficacy and safety.

To further reveal the underlying mechanisms of PRKDC-mediated DOX sensitivity in osteosarcoma, HOS cells were transfected with PRKDC siRNAs or treated with AZD7648, and then subjected to transcriptome analysis. The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis and the gene set enrichment analysis indicated that both siRNA and AZD7648 treatment downregulated the PI3K/AKT signaling pathway (Fig. 4A–F). We also identified 964 common differentially expressed genes in both the PRKDC-knockdown and AZD7648-treated cells (Fig. 4G) and confirmed that they were also enriched in the PI3K/AKT pathway (Fig. 4G). Consistent with the transcriptome analysis, Western blot analysis revealed that the AKT phosphorylation levels were decreased in the PRKDC-knockdown cells compared with the control (Fig. 4H).

To elucidate the mechanism by which PRKDC regulates the PI3K/AKT pathway, the PRKDC interactors were purified using IP and identified using LC–MS/MS. As a result, we found several PRKDC-associated proteins, among which, GDE2 and ENO1 were the top two potential PRKDC interactors (Fig. 4I). However, coimmunoprecipitation (co-IP) analysis in the 143B, HOS, U2OS, and MG63 cells confirmed that only GDE2 binds to the PRKDC protein (Fig. 4J). Echoed with PRKDC regulation, the GDE2 knockdown reduced the AKT phosphorylation levels in osteosarcoma cells (Fig. 4K), whereas the GDE2 overexpression elevated AKT phosphorylation levels (Fig. 4L). Knockdown of GDE2 alone had no obvious impact on osteosarcoma cell proliferation and apoptosis (Supplementary Fig. S6A and S6B), but significantly restrained the growth of osteosarcoma cells under the treatment of DOX (Fig. 4M–R). These findings demonstrate that PRKDC could regulate DOX sensitivity in osteosarcoma by binding with GDE2 to activate the PI3K/AKT pathway.

To assess whether GDE2 was responsible for PRKDC-mediated DOX sensitivity, we treated GDE2-knockdown cells with AZD7648, DOX, or their combination. We found that GDE2 knockdown eliminated the superior efficacy of combination treatment compared with DOX treatment alone (Fig. 5A). The results of the in vivo experiment showed that the therapeutic effect of DOX was not impacted by AZD7648 in GDE2-knockdown tumors (Fig. 5B–F). We further demonstrated that GDE2 knockdown diminished the effect of PRKDC on the alterations in AKT phosphorylation (Fig. 5G). Similarly, the effect of the PRKDC inhibitor AZD7648 on AKT phosphorylation was also removed by GDE2 knockdown (Fig. 5H). Taken together, these data suggest that the effect of PRKDC on DOX sensitivity and AKT phosphorylation is mediated by GDE2. However, the protein level of GDE2 was not significantly affected by PRKDC knockdown or AZD7648 treatment (Fig. 5I and J), indicating that PRKDC recruited but did not regulate the stability of GDE2.

We further investigated the specific mechanism by which GDE2 regulates the PI3K/AKT pathway. We compared the level of GDE2-binding protein between cells with and without PRKDC knockdown. The co-IP and LC–MS/MS analyses revealed that the binding of GNAS to GDE2 in PRKDC-knockdown cells was significantly reduced compared with that in the relative control (Fig. 6A). Previous studies reported that the activated mutation of GNAS was associated with the progression of osteosarcoma (41, 42), and that the overexpression of GNAS promoted activation of the PI3K/AKT pathway (43).

The co-IP assay confirmed that GNAS and GDE2 interacted with each other in osteosarcoma cells (Fig. 6B). Therefore, we hypothesized that GDE2 binds with GNAS to modulate the PI3K/AKT pathway. The knockdown of GDE2 significantly decreased the expression of the GNAS protein but did not significantly alter its mRNA expression (Fig. 6C and D). In addition, treatment with AZD7648 or PRKDC knockdown downregulated the protein, but not the mRNA level of GNAS in osteosarcoma cells (Fig. 6E–G). More importantly, we demonstrated that GDE2 knockdown eradicated the effect of PRKDC knockdown or AZD7648 treatment on the GNAS protein expression (Fig. 6H and I). The co-IP assay using the anti-GDE2 antibody further demonstrated that PRKDC knockdown or AZD7648 treatment significantly reduced the binding of GNAS to GDE2 (Fig. 6J and K). These data indicated that PRKDC could stabilize the GNAS protein by recruiting GDE2.

In addition, GNAS knockdown remarkably reduced AKT phosphorylation in osteosarcoma cells (Fig. 6L) and eliminated the effect of GDE2 knockdown on AKT phosphorylation (Fig. 6M). Consistently, the effect of PRKDC knockdown or AZD7648 treatment on AKT phosphorylation was also diminished by GNAS knockdown (Fig. 6N and O). These findings suggest that PRKDC regulates AKT phosphorylation and DOX sensitivity by recruiting GDE2 to stabilize the GNAS expression. As evidence has shown that AKT activation directly participates in DNA damage response and repair, we sought to investigate whether AKT regulates DNA damage in osteosarcoma cells (44). To this end, we analyzed the levels of DNA damage in osteosarcoma cells with AKT inhibition. The results indicated that AKT inhibition significantly increased DOX-induced DNA damage (Supplementary Fig. S6C), demonstrating that the PRKDC/GDE2/GNAS axis regulates DOX sensitivity by modulating the AKT-mediated DNA damage response.

A previous study reported that PRKDC regulates the nonhomologous end-joining (NHEJ) process to mediate DNA repair (45). Therefore, we investigated whether impaired NHEJ repair due to PRKDC inhibition contributes to increased DOX sensitivity in osteosarcoma. Our findings revealed that the GDE2/GNAS overexpression largely attenuated the effect of PRKDC inhibition on DOX sensitivity, suggesting that PRKDC primarily modulates DOX sensitivity through the GDE2/GNAS axis in osteosarcoma (Supplementary Fig. S7A and S7B). We also examined the interplay of PRKDC/GDE2/GNAS with the BCL2 family, which may account for the increase in DOX-induced cell death. Interestingly, our results indicated that the knockdown of PRKDC/GDE2/GNAS had no discernible impact on the expression of BCL-2 and BAX (Supplementary Fig. S7C–S7E).

Despite the advances in tumor biology over the past decades, the therapeutic method for osteosarcoma remains limited, especially for patients with drug resistance or metastasis. Kinases are often activated in cancers and contribute to tumor progression and drug resistance (46, 47). In this study, we conducted a kinome-wide CRISPR screen and identified PRKDC as a critical determinant of DOX sensitivity in osteosarcoma. The knockout of PRKDC remarkably increased the sensitivity of osteosarcoma to DOX. Previous studies have also reported that inhibition of PRKDC enhances the efficacy of DOX in cancer treatment. Fok and colleagues found that the PRKDC inhibitor AZD7648 increased DOX sensitivity in breast cancer (23). Similarly, the PRKDC inhibitor NU7441 sensitized breast cancer cells to radiation and DOX therapy (24). However, these studies did not illustrate the underlying mechanisms of PRKDC-mediated DOX sensitivity.

Through a transcriptome analysis, we observed that the PI3K/AKT pathway was downregulated upon PRKDC knockdown or AZD7648 treatment. Consistently, previous findings have also demonstrated that PRKDC could regulate the activation of the PI3K/AKT pathway (48). Inhibition of the PI3K/AKT signaling pathway has been shown to increase DOX sensitivity in breast cancer (49). Thus, we hypothesized that PRKDC-mediated DOX sensitivity was attributed to its ability to regulate the PI3K/AKT signaling pathway. We further studied the molecular mechanisms that collectively control the PI3K/AKT pathway and the sensitivity of osteosarcoma to DOX. The results showed that PRKDC could recruit GDE2 in an expression-independent manner. GDE2 is a chemoresistance-related gene that was previously reported to regulate 5-Fu resistance in colorectal cancer and confer cisplatin resistance in nasopharyngeal carcinoma (28, 29). We found that GDE2 knockdown improved DOX sensitivity in osteosarcoma by regulating the AKT pathway activation. More specifically, GDE2 activates AKT by binding to GNAS and stabilizes its expression. Collectively, these findings expound on a novel mechanism for PRKDC-mediated DOX sensitivity.

While the involvement of PRKDC and AKT in DOX resistance has been reported in other tumors, the detailed mechanistic axis remains incompletely elucidated, particularly in osteosarcoma, in which relevant studies are lacking. Significantly, our study revealed that in osteosarcoma, PRKDC predominantly affected DOX sensitivity through modulation of the GDE2/GNAS/AKT axis, rather than by directly regulating NHEJ repair (50). This discovery offers novel insights into the influence of PRKDC on DOX sensitivity, which will assist with the development of related therapeutics in the future.

Drug combination constitutes a fundamental approach for osteosarcoma therapy because it can overcome compensatory mechanisms and increase therapeutic efficacy (51, 52). We systematically explored drug combination treatment strategies for osteosarcoma in our previous study (53). Herein, we demonstrated that knockout of PRKDC obviously increased DOX sensitivity and that the PRKDC inhibitor AZD7648 could significantly synergize with DOX in the suppression of osteosarcoma growth. A Phase I/IIa clinical trial is underway to evaluate the therapeutic efficacy of the combination of AZD7648 and pegylated liposomal DOX in patients with advanced cancers (NCT03907969). We propose that the combination of AZD7648 and DOX represents a strategy for osteosarcoma therapy with excellent potential, although clinical trials are still needed to evaluate its effectiveness and safety. In addition, there is currently no inhibitor for GDE2 and GNAS. Identifying these inhibitors through high-throughput screening may provide new therapeutic alternatives for unmet clinical applications. However, this may present a challenge because GDE2 and GNAS are ubiquitous and may not be feasible therapeutic targets.

In summary, our results suggest the critical role of PRKDC in DOX sensitivity of osteosarcoma. Mechanistically, PRKDC could recruit GDE2 to stabilize the GNAS expression and activate the PI3K/AKT signaling pathway. The PRKDC/GDE2/GNAS/AKT axis could serve as a potential target for enhancing DOX sensitivity in osteosarcoma, and the corresponding combinational therapy strategies are well worth developing.

No disclosures were reported.

W. Zhang: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, and writing–review and editing. W. Li: Conceptualization, data curation, formal analysis, validation, investigation, visualization, methodology, and writing–review and editing. C. Yin: Resources, investigation, and methodology. C. Feng: Resources, investigation, and methodology. B. Liu: Investigation and methodology. H. Xu: Investigation and methodology. X. Jin: Conceptualization, data curation, supervision, investigation, and writing–review and editing. C. Tu: Conceptualization, resources, data curation, supervision, funding acquisition, project administration, and writing–review and editing. Z. Li: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, project administration, and writing–review and editing.

This work was supported by the National Natural Foundation of China (grant no. 82272664 to C. Tu; grant no. 82172500 to Z. Li), National Key Research and Development Program of China (grant no. 2023YFC2507603 to Z. Li), Science and Technology Innovation Program of Hunan Province (grant no. 2023RC3085 to C. Tu), Hunan Provincial Natural Science Foundation of China (grant no. 2022JJ30843 to C. Tu), Science and Technology Development Fund Guided by Central Government (grant no. 2021Szvup169 to Z. Li), Hunan Provincial Administration of Traditional Chinese Medicine Project (grant no. D2022117 to C. Tu), Scientific Research Program of Hunan Provincial Health Commission (grant no. B202304077077 to C. Tu), and Hunan Provincial Health High-level Talent Scientific Research Project (grant no. R2023054 to C. Tu). We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.