Human osteosarcoma cells used in our study were from ATCC, including 143B and HOS/MNNG. In a previous study, HOS has been demonstrated as MTAP-deleted osteosarcoma and its homologous derivatives, 143B and HOS/MNNG, inherit this feature as well (García-Castellano et al., 2002). All cell lines were stored at Shanghai Bone Tumor Institution (Shanghai, China). Osteosarcoma cells were maintained in DMEM (Wisent Inc., Canada) with 10% FBS (Wisent Inc., Canada) and 1% Penicillin-Streptomycin-Glutamine (Wisent Inc., Canada) in a humidified atmosphere of 5% CO2 and 95% air at 37°C. Cells were dissociated by TrypLE^TM Express (Gibco, Thermo Fisher Scientific Inc., United States) and passaged when they are 90–100% confluent. All cell lines were certificated by short tandem repeat (STR) analysis.

Cells were seeded in 96-well plates at a density of 4000–10000 cells per well and cultured overnight. Then, the drug was added to these 96-well plates and cultured for 72 h or more, and next the supernatant was removed gently. Cell proliferation was detected by the reagent CCK-8 (1:10, with serum-free medium, Dojindo Laboratories, Kumamoto, Japan) according to the manufacturer’s instructions. The absorbance at 450 nm was measured by using a microplate reader SpectraMax M3 (Molecular Devices, Sunnyvale, CA, United States) and Bio-Rad ChemiDoc MP Imaging System. All experiments were performed in triplicate.

Total protein was extracted from gathered cells by RIPA protein lysis buffer (Beyotime, Shanghai, China) on cold ice (4°C) in about 30 min. The protein then was collected and centrifuged (12000 rpm) at 4°C for 20 min. The supernatant was collected and the protein concentration of the lysates was measured by the Protein BCA Assay Kit (Bio-Rad, Hercules, CA, United States). For the Western blotting assay, 20 μg of protein mixed with 5 × SDS loading buffer was loaded per lane, separated with 10% SDS-PAGE, and transferred to a polyvinylidene fluoride (PVDF) membrane (MilliporeSigma, Burlington, United States). After blocked with 5% skimmed milk in TBST for 1 h at room temperature, the membrane was incubated at 4°C overnight with primary antibodies (1:1000, PRMT5 #79998, SDMA #13222, GAPDH #5174, Cell Signaling Technology Inc., Danvers, United States), and then, the membrane was washed three times for 10 min with TBS-T solution (TBS, 1 ×; Tween-20, 1:1000) and incubated for 1 h with the corresponding HRP-conjugated secondary antibodies (1:5000, Abgent, San Diego, United States). Chemiluminescent detection was performed using an ECL kit (Pierce Chemical, Rockford, IL, United States) and Bio-Rad ChemiDoc MP Imaging System (Bio-Rad, Hercules, CA, United States). All experiments were performed in triplicate.

For RT-PCR, EZ-press RNA purification kit (EZBioscience, Roseville, United States) was used to collect the purified total RNA in cultured cells. Then, a color-reverse transcription kit (EZBioscience, Roseville, United States) was applied to remove the genomic DNA (gDNA) and the reverse transcription of RNA to get the cDNA being performed. 2 × color SYBR green qPCR master mix (EZBioscience, Roseville, United States) was used for RT-PCR on QuanStudio six Flex RT-PCR systems (Thermo Fisher Scientific Inc., United States). The protocol was followed and the primers were typed as below:

GAPDH (glyceraldehyde-3-phosphate dehydrogenase):

Human-GAPDH-F GTC​TCC​TCT​GAC​TTC​AAC​AGC​G;

Human-GAPDH-R ACC​ACC​CTG​TTG​CTG​TAG​CCA​A.

MTAP (methylthioadenosine phosphorylase):

Human-MTAP-F GTC​ATA​GTG​ACC​ACA​GCT​TGT​GG;

Human-MTAP-R CCT​CTG​GCA​CAA​GAA​TGA​CTT​CC.

Gelatin microparticles (microrobots) were synthesized in a multi-channel fluidic module based on the shear stress of two-phase flow (Figure 3A). 0.5 g of gelatin [Cas (9000–70–8), Aladdin, Shanghai, China] was first dissolved in 10 ml of deionized (DI) water. Then, an additional 2 ml of Fe₃O₄ (10–30 nm, 25% in H₂O, Macklin, Shanghai, China), 1 ml EPZ015666 (10 uM, Cat. No.: HY-12727, MedChemExpress, United States) stained by rhodamine (Macklin, Shanghai, China), and 10 mg 2-hydroxy-4′-(2-hydroxyethoxy)-2-methylpropiophenone (Macklin, Shanghai, China) were added. The mixed solution and dimethyl silicone oil (Macklin, Shanghai, China) were injected into microfluidic chips as sample flow (disperse phase) and sheath flow (continuous phase) using syringe pumps (Harvard Apparatus, United States), respectively. Gelatin microparticles were generated and then irradiated using UV light source. After 10 min of irradiation, the case-hardened gelatin microparticles were obtained, collected by centrifugation, and then washed thrice with DI water to remove oil and any other residue.

The magnetic field generator consisted of a three-degree-of-freedom Helmholtz coil, multifunction data acquisition unit (DAQ, NI-PCI-6259), and three single-channel output power amplifiers. Based on controlling the current and the voltage of Helmholtz coil from the driving signal amplified by voltage amplifiers, an external rotating uniform magnetic field can be circularly generated in 3D space to manipulate the hydrogel microrobots. The Helmholtz coil was placed on the observation platform of the microscope to achieve real-time observation of swimming hydrogel microrobots. Magnetic dynamics and navigated locomotion of hydrogel microrobots.

The motional performance of the microrobots was investigated under an external magnetic field. A series of control experiments were carried out to elucidate the effect of different drive frequencies and the strength of the magnetic field on the velocity of microrobots. Then, the microrobots were navigated by the control strategy of three-dimensional rotating magnetic field generated by the three-degrees-of-freedom Helmholtz coil. In the x–z plane, the microrobot moved around the short axis (x-axis), applying a magnetic field given by H (t) = H ₀ [cos (ωt)e_x + sin (ωt)] e_z (H ₀ is the magnitude, ω is the angular frequency of the magnetic field, and t is the time). Here, e_x and e_z represent the unit vector along the x and z axes, respectively (hereafter, e_y represents that along the y axis). In the y–z plane, the microrobots moved around the long axis (y-axis), applying a magnetic field by H (t) = H ₀ [cos (ωt)e_y + sin (ωt)] e_z. By controlling the input of current, the direction of the rotating magnetic field can be changed to control the direction of the microrobot. Based on the dynamic control scheme above, a predefined trajectory could be walked by a magnetically controlled microrobot.

We designed an in vitro simulated experiment to detect the selective drug inhibition of the magnetic microrobots ( Figure 5A). We added the microrobots and carried the EPZ015666 into a cell culture dish (35 mm diameter) with a complete medium and then, the microrobots were given a directional magnetic field for 5 min, which was used to simulate the locomotion of microrobots in the blood. When the microrobots moved to one side of the dish, probably 1 min based on the strength of the magnetic field and before the drug released, the liquid close to this side, full of microrobots, was transferred to the 96-well plates cultured with HOS/MNNG cells, which was called “High-dose” group (high dose of microrobots). Meanwhile, the liquid of the other side which lacked microrobots was also transferred to the 96-well plates cultured with HOS/MNNG cells, which was called “Low-dose” group (low dose of microrobots). After treating for 72 h, a CCK-8 assay was given to detect the cell viability of both groups. Also, we used drug-free microrobots and the dissociative drug to repeat the experiment as control.

Transcriptomes of patients in The Cancer Genome Atlas (TCGA) and Therapeutically Applicable Research To Generate Effective Treatments (TARGET) databases were downloaded from the Genomic Data Commons (https://portal.gdc.cancer.gov/). The transcriptome of tumor cell lines was downloaded from the Cancer Cell Line Encyclopedia (CCLE) (https://sites.broadinstitute.org/ccle) database. All the data were processed by the same method (R version 4.1.2) and normalized to fragments per kilobase per million (FPKM) using R package edgeR (http://bioconductor.org). The survival rate of patients was analyzed by online tool GEPIA (http://gepia.cancer-pku.cn). Statistical analyses were performed using GraphPad PRISM 8 (GraphPad Software, Inc.) and some visualizations were performed using RStudio 1.4 (https://www.rstudio.com).

The enhancement or deficiency of gene function leads to cellular changes in various aspects and results in tumor generation or malignancy, respectively. The genomic mutation on the one hand, gives the tumor cell advantages in competitive growth, on the other hand, foreshadows hide vulnerabilities in tumor cells. Following this, some unexpected gene relationships have been found by CRISPR-cas9 screening and the synthetic lethality gene has been demonstrated as a vulnerability to a specific mutation in pan-cancer genomic analyses. Genomic mutation of MTAP deletion occurred commonly in pan-cancer in the TCGA database and there was nearly 10% MTAP deletion in sarcoma (Figure 1A). Methionine metabolism in tumor cells is disordered by MTAP deletion at chromosome 9p21.3. To further confirm the frequency of MTAP deletion in osteosarcoma, the pan-cancer cell lines in the CCLE database and the tumor tissue in the TARGET database were detected. Then, by comparing the two main genes located at chromosome 9p21.3, CDKN2A and MTAP, and two homologous cell lines of all osteosarcoma cell lines, 143B and HOS, the lowest copy number of CDKN2A and MTAP were shown in Figure 1B. In the TARGET database, 23.86% of osteosarcoma revealed genomic MTAP deletion, including the loss of heterozygosity (LoH) and homologous deletion (Figure 1C). The results have demonstrated that MTAP deletion happens frequently in osteosarcoma. As a result, the intracellular 5′-methylthioadenosine (MTA) cannot be metabolized because of the loss of MTAP function which leads to the tumor being dependent on PRMT5. Then, MTAP RNA expression in the hMSC (human mesenchymal stem cell) and common osteosarcoma cell lines (U2OS, SJSA-1, HOS, 143B, and HOS/MNNG) were further detected. The mRNA expression of U2OS and SJSA-1 was like hMSC, but HOS and its derivatives, 143B and HOS/MNNG, were loss of the mRNA expression of MTAP. Thus, the MTAP function was lost as indicated by the results (Figure 2A) and they, called MTAP-deleted osteosarcoma, might be sensitive to the PRMT5 inhibitor. PRMT5 inhibitor, EPZ01566, reduces the level of SDMA (symmetric di-methyl-arginine) modification and induces DNA damage to MTAP-deleted tumor cells and blocks tumor growth (Kalev et al., 2021). Therefore, the effect of PRMT5 inhibition on tumor growth was next detected. The cell viability of 143B and HOS/MNNG were influenced by EPZ015666, a PRMT5 inhibitor (Figure 2B). Considering whether it was related to the inhibition of the function of PRTM5 enzyme, we detected SDMA, the modification site of PRMT5, which was reduced by PRMT5 inhibition for 72 h (Figure 2C). To conclude, our results have demonstrated that PRMT5 inhibitor, EPZ015666, could block the growth of MTAP-deleted osteosarcoma and inhibits the level of SDMA modification in vitro. The growth inhibition of EPZ01566 to MTAP-deleted osteosarcomas is different on 143B and HOS/MNNG, while HOS/MNNG seems to be more sensitive to PRMT5 inhibition in vitro.

Theoretically, synthetic lethality drugs inhibit the growth of typical tumor cells without affecting normal cells. However, the synthetic lethality drug can also combine the target with normal cells (Wang et al., 2018). Although it slightly affects normal cells, the drugs arriving at the tumor location decrease and this gentle inhibition of tumor cells might fail. Also, side effects might appear with increasing the dose of such drugs. However, a drug delivery system might be useful for synthetic lethality drugs to reduce the side effects. Non-target drug delivery system based on magnetic-driven hydrogel microrobots could not only solve this problem but also avoid the obsession that no specific membrane protein is acknowledged to be the target of osteosarcoma. Microrobots, now being used for a multitude of applications, such as drug delivery, surgery, and diagnosis, can easily overcome the blood flow and approach the tumor loci actively, with the power of chemical or physical engine (Wu et al., 2021). Magnetic-driven microrobots are generated by moving charges and magnetic materials, and thus capable of moving to the scheduled destination by the navigation of magnetic fields (Mujtaba et al., 2021; Zhang et al., 2021; Zhou et al., 2021). In this study, a magnetic-driven hydrogel microrobot for synthetic lethality drug delivery system has been innovated, while the morphological characteristics of microrobots by electron microscopy have been detected as well. The diameter of magnetic-driven hydrogel microrobots was about 30 μm. The microrobots under bright field showed dark red (Figure 3B), which carried EPZ015666 stained by rhodamine and Fe₃O₄, and under dark field showed bright red fluorescence, counting about 200 microrobots per microliter (Figure 3C). This result revealed that the EPZ015666 was successfully loaded by microrobots. Then, the motion of the microrobots under magnetic field was detected. It has been observed that the velocity of microrobots was growing faster by enhancing the intensity and frequency of the magnetic field, and the highest velocity is 100 μm/s (Figures 3D,E). For predetermined locomotion experiments, the microrobots were driven by the predetermined direction of the magnetic field and tracked a square (Figure 4A and Supplementary Video S1) or letter H (Figure 4B and Supplementary Video S2). Our results have demonstrated that the microrobots carry the EPZ01566, stained by the rhodamine, and the movement of microrobots could be controlled by changing the magnetic field. Our drug delivery system could sustainedly release the drug into the complete medium for 15–20 min completely. The sustained-release microrobots are supposed to reduce the side-effect to normal cells and enhance the drug lethality in tumor cells (Chen et al., 2021). The selected inhibition of the magnetic-driven hydrogel microrobot drug delivery system was simulated in vitro (Figure 5A). Most of the microrobots were driven by the magnetic field to one side of the culture dish, a diameter equal to 35 mm for nearly 5 min. Then, by adding the liquid of this side (High-dose: high dose of microrobots) and the opposite side (Low-dose: low-dose of microrobots) in the 96-well plate culturing HOS/MNNG tumor cells, lower cell viability was detected in the high dose of microrobots (Figure 5B). In a simulated in vitro experiment, it has been indicated that the magnetic-driven microrobots could carry drugs and inhibit tumor growth. As a result, our magnetic-driven hydrogel microrobot drug delivery system can be used for the non-targeted therapy of MTAP-deleted osteosarcoma. But there is still a big gap between research and development and the clinical implication of treating osteosarcoma. The patients who unluckily suffered the osteosarcoma usually have a large tumor load. The locomotion of the microrobots in the solid sarcoma might be more difficult than in the blood cancer due to the difference in the medium. The microrobots overcoming the resistance to the blood flow and driven to the tumor site could be achievable, but the locomotion going further into the deeper site of the primary tumor seems to need further study. Our study will further verify whether the magnetic microrobots carried out by typical synthetic lethal drugs could inhibit the growth of osteosarcoma in vivo under the guidance of genomic features.