Human lung cancer PC9 cells were provided by the Chinese Academy of Medical Sciences (China). Brain metastatic subpopulations (PC9BrM1, PC9BrM2, and PC9BrM3) were built by infusing PC9 cells in immunodeficient mice through left ventricular injection and separating metastatic cells from collected brain metastases (Liu et al., 2019). Highly brain metastatic lung cancer PC9BrM3 cells were obtained through repeated injection separation expansion cycling three additional times, and maintained in RPMI 1640 containing 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 U/mL streptomycin at 37°C in a humid environment with 5% CO₂.

I verified that all human cell lines have been authenticated through STR (or SNP) profiling within the past 3 years, and all experiments were conducted using mycoplasma-free cells.

Antibodies against CD146 (661531lg) GAPDH (60041lg), BCL2 (601781lg) and BAX (505992lg) were provided by Proteintech (China). Antibodies targeting γH2AX (#9718), cleaved caspase 3 (#9664), cleaved PARP (#5625), phosphoNFκBp65 (#3033), NFκB (#8242), phosphoIkBα(#2859)and IkBα(#4814)were from Cell Signaling Technology.

Cell viability was monitored with CCK8 kit (CCK8; K1018, ApexBio, United States) following the manufacturer suggestions. In a 96well plate, cells were seeded at 5,000/well for overnight incubation. Then, the treatments were removed and replaced by specific drugs for 72 h. After 72 h of culturing, the culture medium was enriched with 10% CCK8 and subsequently incubated for an additional 2 h. The obtained data (optical density) were assessed using a Multiskan™ FC Microplate Reader from Thermo Fisher Scientific at a wavelength of 450 nm.

For the colony formation assay, cells underwent seeding into a 6well plate at 1,000/well, followed by incubation at 37°C and 5% CO₂ with the medium refreshed every other day. Following a 2 weeks incubation period, 4% paraformaldehyde was used for fixation, with subsequent staining with 4% crystal violet for 30 min.

The migration assay used Transwell chambers (Corning, New York, United Kingdom). Tumor cells (1 × 10⁴/well) were plated into chambers and coincubated with 200 µL of serumfree medium with or without pemetrexed. Totally 72 h later, fixation was carried out with 4% paraformaldehyde with subsequent crystal violet staining. Photomicrographs were obtained. Experimental data were statistically assessed with ImageJ.

Total protein extraction from harvested cells was carried out with RIPA cell lysis buffer that contained protease (Meilunbio, China) and phosphatase (Sigma, United Kingdom) inhibitor cocktails. Protein concentration was assessed with the BCA Protein Assay Kit (Thermo Fisher). Proteins were then separated by SDSPAGE and electrotransferred onto a nitrocellulose membrane with a miniProtean electrophoresis system (Millipore, United States). After a 2 h blocking with Trisbuffered saline supplemented with 0.1% Tween20 (TBST) and 5% skimmed milk, the membranes underwent successive incubations with primary and secondary antibodies. The ECL kit (Tanon, China) was utilized for detection. ImageJ was utilized for quantifying protein bands (National Institutes of Health, United Kingdom). All experiments are conducted by at least three independent researchers.

CD146targeting shRNA and the respective negative control shRNA (shNC) were built in a lentiviral vector provided by GeneChem (China). The target sequence of the CD146 shRNA was 5′GAG​CGA​ACT​TGT​AGT​TGA​A3′. 293T cells were infected with shRNAexpressing lentivirus utilizing Lipofectamine 2000 (Invitrogen, United States) per the manufacturer’s protocol.

Infection with the shRNAexpressing lentivirus and establishment of stably infected single clones were carried out as directed by the manufacturer. Immunoblot was carried out to assess knockdown efficiency. Transfectionpositive cells were screened with complete medium containing puromycin (1 μg/mL) and cultured continuously to ensure stable gene expression in cells. CD146 overexpression lentivirus (OECD146) was purchased from the GeneChem Company.

For apoptosis assay, cells after transfection with CD146 shRNA and control shRNA (10⁶ cells) underwent culture in 6well plates and treatment with PEM (80 nM) for 72 h prior to cell collection. According to the manufacturer’s instructions, cell staining was carried out with AnnexinVAPC and 7aminoactinomycin D (7AAD) (ECKA218, Elabscience). After staining, flow cytometry (BD Accuri TM C6 cytometer) was employed for analysis using Flowjo.

For cell cycle assays, flow cytometry was carried out with a cell cycle kit (no. C1052; Beyotime, China) per the manufacturer’s protocol, and data analysis utilized the ModFit software.

Tumor cells underwent seeding in a 48 well plate at 10,000 per well for quantitating γH2AX. Pemetrexed was administered for 48 h, followed by fixation with 4% paraformaldehyde (KeyGEN BioTECH) for 20 min with addition of 0.25% Triton X100 for permeabilization for 810 min. Blocking was performed with 3% BSA (Sigma) at ambient for 60 min. Primary antibody targeting γH2AX (1:400) was added for overnight incubation at 4°C. Then, cells were stained with Alexa Fluor 594 (1:100; Proteintech) for 1 h followed by DAPI (1:1,000, KeyGEN Biotech) counterstaining for 10 min at ambient shielded from light. A fluorescence microscope (Nikon TE200, Japan) was utilized for image acquisition, and ImageJ was utilized for quantitation.

Athymic female BALB-c-nu mice, aged 4–6 weeks, were purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd. and were kept in pathogen-free conditions. The highly brain metastatic subpopulation PC9BrM3 with negative control shRNA (shNC) or CD146 silencing (shCD146) (5 × 10⁶ in 100 µL PBS) was administered by subcutaneous (s.c.) injection to the dorsal left flank of 4 weekold BALB/c nude mice (n = 10/group). Each group was further subdivided into the control and dosing groups (n = 5). Interventional treatment was performed after the tumor reaches 100,150 mm³. Control mice were administered DMSO by intraperitoneal injection, and the dosing group was administered 100 mg/kg pemetrexed by intraperitoneal injection once weekly. Tumor volumes were measured at 3 day intervals. The mice were euthanized on the 24th day of treatment, and the tumors were excised for further experiments. The euthanasia was performed via intraperitoneal injection of sodium pentobarbital (200 mg/kg), with death confirmed by respiratory and cardiac arrest, as well as pupil dilation. Humane endpoints such as tumor ulceration, tissue swelling, and cachexia were monitored, and none of the mice in the study exhibited these conditions.

All animals were handled per the recommendations of the Dalian Medical University Animal Research and Care Committee, and studies were performed based on the Dalian Medical University Animal Experimentation Guidelines.

Immunohistochemistry (IHC) was carried out to detect γH2AX, BAX, cleaved caspase 3 in mouse subcutaneous tumor tissues. Paraffinembedded tumor tissue sections underwent deparaffinization, hydration, and antigen retrieval. After treatment with 3% hydrogen peroxidemethanol for 20 min, blocking was performed with 5% goat serum albumin for 30 min. Then, incubation was carried out with antibodies targeting mouse BAX (1:100), cleaved caspase 3 (1:100) and γH2AX (1:400) overnight at 4°C. Next, the corresponding secondary antibody was added for a 20 min incubation at ambient. Then, staining was performed with the diaminobenzidine kit. Positive cells were detected microscopically and ImageJ was utilized for analysis.

Quantitative data are mean ± standard deviation (SD) from at least 3 replicates. GraphPad Prism 8.0.2 (GraphPad Software, United States) was used for data analysis. Group pairs and multiple groups were compared by the t-test and one way ANOVA (analysis of variance) with post hoc Tukey’s test. An effect was considered statistically significant with a p-value below 0.05.

We used the human lung cancer cell line PC9 and injected it intracardially into immunodeficient mice to isolatebrain metastatic subpopulations. From these, we isolated PC9 clones with higher metastatic potential. After two rounds of in vivo screening and in vitro expansion, we obtained a series of brain metastatic subpopulations derived from the parental PC9 cells (Figure 1A). We previously showed that PC9BrM3 cells have high propensity to metastasize to the brain (Liu et al., 2019). Additionally, these cells have consistent genetic homology, which rules out heterogeneity between individual cases, and are an ideal model for drug sensitivity/resistances studies.

To compare parental PC9 cells and highly brain metastatic PC9BrM3 cells, these cells were administered pemetrexed. Cell viability and colony formation assays revealed obvious pemetrexed drug resistance of the highly brain metastatic PC9BrM3 compared with the parental PC9 group (Figures 1B–E).

CD146 is a multifunctional molecule produced by multiple cancers. To assess CD146 expression in brain metastatic NSCLC cells, its protein amounts were quantitated. Western blot showed that CD146 was markedly upregulated in brain metastatic cells in comparison with parent cells (Figures 1F, G).

Because CD146 is highly expressed in BM, we evaluated the potential link between upregulated protein expression and acquired pemetrexed resistance in brain metastatic cells. shRNA was employed to knockdown CD146, and PC9BrM3 cell proliferation was examined after pemetrexed administration. Cell viability assay showed that inhibition of CD146 with shRNA in BM cells increased PEM sensitivity (Figures 2A, B), whereas CD146 overexpression promoted PC9 cell resistance to PEM (Figures 2C, D). In addition, the proliferative ability of PC9BrM3 cells was altered after CD146 knockdown or overexpression, including after pemetrexed administration (Figure 2E). Conversely, in the colony formation assay, CD146 knockdown significantly decreased PEM resistance in BM cells. Correspondingly, CD146 overexpression in PC9 cells increased PEM resistance (Figure 2F). In addition, the migration assay demonstrated silencing or overexpression of CD146 affected the migratory ability of PC9BrM3 cells, and pemetrexed sensitivity was also significantly affected (Figure 2G). Taken together, these results suggested CD146 expression is positively associated with resistance to PEM in BM cells.

Previous studies have shown that pemetrexed can mediate DNA damage in NSCLC (Ding et al., 2020). Therefore, the effect of CD146 on DNA damage in brain metastases from lung cancer was examined in presence of pemetrexed. The results of immunofluorescence indicated γH2AX (a biomarker of DNA double strand breaks) foci were significantly more abundant in BM cells with CD146 knockdown in the presence of pemetrexed compared with control cells; meanwhile, the accumulation of γH2AX foci was reduced in PC9 cells with CD146 overexpression (Figure 3A). Similarly, immunoblot showed that CD146 downregulation enhanced DNA damage induced by pemetrexed in BM cells, and CD146 upregulation in PC9 cells with low CD146 expression attenuated this effect (Figure 3B). It has been shown that pemetrexed induces cell cycle arrest in NSCLC (Chen et al., 2014). According to FCM plots, we found that in the absence of pemetrexed, the expression level of CD146 did not overly affect cell cycle progression in brain metastases; however, after pemetrexed administration, the proportion of brain metastases with CD146 knockdown in the S phase was significantly increased compared with controls, while PC9 cells overexpressing CD146 attenuated the effect of pemetrexed (Figure 3C). This suggested that CD146 is required in enhanced DNA damage repair induced after treatment with pemetrexed. More interestingly, silencing or inhibition of CD146 in brain metastatic cells in the presence of pemetrexed downregulated cyclin A2 and CDK2 (cell cycle related proteins), whereas in PC9 cells, CD146 overexpression promoted their expression (Figure 3D). The latter findings suggested CD146 is critical for maintaining cell cycle progression after pemetrexed treatment.

We previously demonstrated CD146 plays a critical regulatory role in pemetrexed dependent DNA and cell cycle arrest in lung cancer brain metastases. Therefore, we hypothesized that CD146 also mediates pemetrexed resistance in brain metastases through reduced apoptosis. Therefore, the effect of CD146 on apoptosis was assessed in lung cancer brain metastases by flow cytometry. It was clear that the apoptotic rate of brain metastases treated with pemetrexed after knocking down CD146 was significantly higher compared to control cells, with increased apoptotic cell ratio (Figure 4A); conversely, the apoptotic rate of PC9 cells overexpressing CD146 under pemetrexed treatment was higher than that of control cells (Figure 4B).

Moreover, the subsequent changes in protein amounts confirmed this finding. After treatment with pemetrexed, the expression levels of cleaved PARP, cleaved caspase 3 and BAX were markedly elevated in brain metastatic cells with CD146 knockdown compared with the control group while BCL2 was downregulated (Figure 4C); Conversely, after exposure to pemetrexed, the levels of cleaved PARP, cleaved caspase 3, and BAX were significantly reduced in PC9 cells overexpressing CD146 compared to the control group, whereas BCL2 exhibited an increase (Figure 4D). Echoing flow cytometry results, binding of pemetrexed to CD146 shRNA yielded the highest levels of cleaved PARP, cleaved caspase 3 and BAX, indicating that CD146 is resistant to pemetrexed mediated apoptosis in lung cancer brain metastases.

CD146 is known to induce the progression of malignant melanoma via the NFκB pathway (Ruma et al., 2016). We investigated whether CD146 induces pemetrexed resistance through this signaling pathway. First, immunoblot demonstrated pp65 and pIKBα were downregulated in brain metastases with knockdown CD146 relative to the control group, while pp65 and pIKBα were upregulated in PC9 cells overexpressing CD146 relative to the control group (Figures 5A, B). This indicates that CD146 could regulate NFκB signaling in lung cancer brain metastatic cells. More interestingly, the NFκB pathway was downregulated overall when pemetrexed exerted its effects, and pp65 and pIKBα were downregulated more significantly in brain metastatic cells with CD146 knockdown relative to controls. Similarly, pp65 and pIKBα levels in PC9 cells overexpressing CD146 did not decrease as much as in controls in the presence of pemetrexed (Figures 5A, B). Then, we used the NFκB inhibitor APDC to determine the appropriate concentration needed to artificially inhibit NFκB expression in PC9BrM3 cells (Figure 5C). Through the CCK8 assay, we found that inhibit of the NFκB pathway significantly increased the sensitivity of brain metastatic cells to pemetrexed, and interestingly, CD146’s effect on pemetrexed resistance was greatly reduced (Figures 5D, E). Flow cytometry analysis revealed that, compared to the individual addition of pemetrexed, the combination with the inhibitor APDC led to an increased apoptosis rate in BrM3 cells with knocked down CD146, as well as in PC9 cells with CD146 overexpression. This suggests that blocking the NFκB signaling pathway significantly enhances the sensitivity of brain metastatic cells to pemetrexed, while overexpression of CD146 attenuates this effect (Figures 5F, G). Based on these results, we speculate that CD146 could mediate pemetrexed resistance in brain metastatic cells by regulating NF-ΚB dependent resistance to apoptosis.

To further validate our in vitro findings, we established BrM3NC cell derived xenograft tumors and BrM3shCD146 cell derived xenografts in nude mice and applied them in combination with pemetrexed to assess pemetrexed resistance in vivo. We found their combined effect more significantly inhibited the growth of subcutaneous tumors in comparison with pemetrexed alone or knockdown CD146 alone; additionally, subcutaneous tumors were extracted from nude mice and weighed at the end of the study. As a result, tumors in the BrM3shCD146 cell group treated with pemetrexed had significantly lower weights compared with other experimental and control groups (Figures 6A–C). Immunoassays on subcutaneous tumors showed apoptosis related signature molecules, including BAX and cleaved caspase 3, and the DNA damage signature molecule γH2AX showed elevated levels in the BrM3shCD146 group in comparison with the BrM3NC group treated with pemetrexed (Figure 6D). These results suggested that CD146 may mediate pemetrexed resistance in NSCLC brain metastatic cell.