The DNA damage response is involved in tumor recurrence and metastasis in lung adenocarcinoma patients
We examined the GSE30219 dataset from the NCBI GEO database to identify differences in biological function and signaling pathways in primary lung adenocarcinoma tissues versus recurrent tumor tissues. Tumor tissues from 85 patients with lung adenocarcinoma underwent whole transcriptome sequencing, which included 58 primary tissues and 27 tissues following tumor recurrence. Signaling pathways were annotated using a KEGG pathway analysis and revealed that the differentially expressed genes (DEGs) were also enriched in DNA damage checkpoint signaling, homologous recombination, and signal transduction in response to the DDR. This suggests that post-recurrence lung adenocarcinoma tissue is strongly correlated with the DDR (Fig. 1A, Fig. S1A). Combined with previous results from our group, we determined whether there was a correlation between DDR and tumor recurrence resulting from resistance to targeted therapy in lung adenocarcinoma.
EGFR signaling inhibition activates the DNA damage response in Osimertinib-resistant lung adenocarcinoma
To determine whether the DDR is involved in the development of acquired resistance of lung adenocarcinoma cells to Osimertinib, we exposed the lung adenocarcinoma cell lines, NCI-H1975 and PC9, to Osimertinib. NCI-H1975 is a lung adenocarcinoma cell line that contains L858R and T790M EGFR mutations and is sensitive to Osimertinib, whereas PC9 cells harbor a Del19 EGFR mutation and is also sensitive to Osimertinib. After prolonged stimulation with Osimertinib, the half maximal inhibitory concentration (IC50) value of Osimertinib were measured using a CCK-8 assay. The IC50 of the H1975 cell line was 1.78 µmol/L, whereas the IC50 of H1975/AR subline was 12.3 µmol/L (Fig. 1B). The IC50 of the PC9 cell line was 0.66 µmol/L, whereas the IC50 of the PC9/OR subline was 5.40 µmol/L (Fig. 1C), which indicated that Osimertinib acquired-resistant cell lines were established. After treating H1975 and H1975/AR cells with the same concentration of Osimertinib for different times, immunoblotting revealed that the expression of γ-H2AX in H1975 and PC9 cells was increased, whereas in H1975/AR and PC9/OR, the expression of γ-H2AX increased briefly following drug treatment, but was significantly reduced after 24 h (Fig. 1D-E). γ-H2AX is a phosphorylated histone protein, which is a marker of DNA DSBs[15]. A similar result was obtained using immunofluorescence to observe and count γ-H2AX-foci (Fig. 1F–G). A Comet assay was then used to further confirm these findings. DNA damage was evident after 24 h of Osimertinib treatment in H1975 cells, whereas in H1975/AR cells, DNA damage increased within 6 h and decreased after Osimertinib treatment for 24 h. H1975 cells exhibited more DNA damage compared with H1975/AR during the same time period (Fig. S1B–S1C). To identify the mechanism of DNA damage caused by Osimertinib and the differences in the degree of DNA damage between H1975 and H1975/AR cells, we hypothesized that changes in reactive oxygen species (ROS) levels may be involved in DNA damage caused by Osimertinib. Oxidative stress and damage to proteins, lipids, and DNA result from elevated ROS[16]. Under normal conditions, ROS level in H1975 cells are higher compared with that of H1975/AR (p < 0.001). Following Osimertinib treatment, the ROS levels in H1975 and H1975/AR were increased, whereas ROS levels in H1975 was higher compared with that of H1975/AR (p < 0.001). This indicates that the increase of ROS following Osimertinib treatment is lower in resistant versus sensitive cells (Fig. 1H), rendering these cells more resistant to oxidative stress and reducing DNA damage.
In eukaryotic organisms, DSBs perform DNA damage repair by two mechanisms: Non-homologous end joining (NHEJ) and homologous recombination (HR)[17, 18].To determine the variations in these repair pathways in the two pairs of cell lines following Osimertinib treatment, we measured KU70 and KU80, which are involved in the NHEJ pathway, and RAD51 and BRCA1 were are involved in the homologous recombination (HR) pathway by immunoblotting. In H1975 cells, Osimertinib inhibited the expression of genes involved in the HR and NHEJ repair pathways. BRCA1 and RAD51 expression decreased in H1975/AR, whereas KU70 and KU80 were markedly increased (Fig. 2A). We also observed a similar trend in PC9 and PC9/OR cell lines (Fig. S1D). We measured the expression of the aforementioned DNA repair genes in the two cell lines by qRT-PCR and found that the NHEJ pathway genes, 53BP1, KU70, and KU80, were highly expressed in H1975/AR, whereas genes involved in the HR repair pathway, RAD51, and BRCA1, were unchanged or decreased in expression. In addition, we found that the expression of ATM, an important upstream gene for DNA damage repair, was simultaneously increased (Fig. 2B). An NHEJ/HR assay was conducted to further validate the results presented above. Compared with the DNA damage group caused by I-SceI, Osimertinib treatment inhibited the NHEJ and HR pathways in H1975 cells (p < 0.01 and p < 0.01), whereas in H1975/AR cells, the HR pathway was inhibited, but the NHEJ pathway was abnormally increased (p < 0.05 and p < 0.05) (Fig. 2C). The results were consistent with that of immunoblotting, which indicate high expression of the upstream DNA damage repair signal in H1975/AR. The results suggest that one cause of acquired resistance to Osimertinib may be an abnormal increase in the NHEJ repair pathway.
Chromatin relaxation followed by condensation represent the initial effects of DSBs, which are necessary for both DSB damage and repair[7]. Therefore, we examined whether chromatin structure changes contribute to EGFR-TKI resistance. Chromatin relaxation analysis consistently demonstrated that chromatin degradation in H1975 cells was more evident compared with that of H1975/AR cells following the addition of micrococcal nuclease (MNase) after Osimertinib treatment (Fig. 2D–2E). These results indicate that the chromatin of H1975/AR is more condensed compared with that of H1975 cells following Osimertinib treatment, suggesting that cells may increase their resistance to oxidative stress to reduce DNA damage. In addition, chromatin condensation enhances the ATM-related upstream DDR signal, which enhances DNA repair.
EGFR oncogene inhibition reduces L3MBTL1 ubiquitination and stabilizes its expression in Osimertinib-resistant cells
MBT domain proteins are involved in the condensation of chromatin structure, but it is unclear how they relate to acquired Osimertinib resistance. We measured the MBT domain family member L3MBTL1 and two related homologs, L3MBTL2 and L3MBTL3, in H1975 and H1975/AR cells by qRT-PCR to determine whether the expression of MBT domain proteins changed in Osimertinib acquired-resistant cells. The results indicated that the mRNA levels of the MBT domain family in H1975/AR cells was significantly higher compared with that in H1975 cells, and L3MBTL1 increased the highest (4.2-fold vs. 3.1-fold, 1.6-fold change) (Fig. 3A). The L3MBTL1 protein was also highly expressed in H1975/AR cells (P < 0.001), and also in PC9/OR cells, the expression of L3MBTL1 was higher compared with that of PC9 cells (Fig. 3B).
Following Osimertinib therapy, we observed that L3MBTL1 gradually decreased in H1975 cells and increased in H1975/AR cells (Fig. S2A). L3MBTL1 was shown to be ubiquitinated after DNA strand breaks[13]. Therefore, we examined L3MBTL1 ubiquitination levels in H1975 or H1975/AR cells. The results indicated that the ubiquitination of L3MBTL1 was significantly increased in H1975 cells compared with that in the control group and slightly lower in H1975/AR cells following Osimertinib treatment (Fig. 3C, D). This suggests that L3MBTL1 is stabilized by lowering ubiquitination in H1975/AR cells. The ubiquitination of L3MBTL1 was inhibited in H1975 cells after treatment with the proteasome inhibitor MG132 in combination with Osimertinib (Fig. 3E–3F). The ubiquitination of L3MBTL1 in H1975/AR cells was increased following treatment with the intracellular protein synthesis inhibitor cycloheximide combined with Osimertinib (Fig. 3G–3H), which indicates that this process was likely mediated by the ubiquitin-proteasome pathway.
As an E3 ubiquitin ligase, RNF8 facilitates the recruitment of ATPase VCP/p97 to DNA damage sites following DSB, which increases the ubiquitination of L3MBTL1[13]. Immunoblotting revealed that the expression of RNF8 was significantly inhibited in H1975/AR cells following treatment with Osimertinib (Fig. 3I). We also performed immunoprecipitation assays with anti-RNF8 and found that RNF8 co-precipitates with EGFR in H1975 and H1975/AR cells (Fig. 3J). Accordingly, we deduced that in H1975/AR cells, a decline in RNF8 expression mediates the abnormal down-regulation of L3MBTL1 through proteasome-mediated ubiquitination, which results in L3MBTL1 expression in EGFR-TKI acquired-resistant cells.
L3MBTL1 is involved in the DNA damage process by regulating chromatin remodeling
Following L3MBTL1 knockdown in H1975/AR cells, we established gene expression profiles to determine the underlying mechanism. RNA-seq was used to identify DEGs between H1975/AR SH-NC and H1975/AR SH-L3MBTL1 cells. To avoid false-positives, three control experiments were run concurrently and a threshold minimum of a 2-fold difference was used as the exclusion criteria. We identified 1663 DEGs, of which 748 were increased and 915 were decreased (Fig. 4A). Gene Set Enrichment Analysis (GSEA) was used to analyze functional enrichment. Apoptosis, HR, mismatch repair, and nucleotide excision repair were among the DDR mechanisms associated with altered gene expression (Fig. 4B). Therefore, we hypothesized that L3MBTL1 mediates Osimertinib resistance through the DDR. We used siRNA to reduce the expression of L3MBTL. The efficiency of mRNA knockdown is shown in Fig. S2B, in which si-L3MBTL1-1 and si-L3MBTL1-3 were selected to reduce the possibility of off-target effects. Immunofluorescence staining revealed that compared with the control group, the foci formed by γ-H2AX and 53BP1 in H1975 cells overexpressing L3MBTL1 were significantly decreased. However, the formation of foci increased significantly after L3MBTL1 knockdown in H1975/AR cells and was more apparent following treatment with Osimertinib (Fig. 4C–4D). The results were confirmed by immunoblotting. In H1975/AR cells, the expression of γ-H2AX increased after knocking down L3MBTL1, and the expression of 53BP1, H4K20Me2, and P-P53 were also increased. Subsequently, we overexpressed L3MBTL1 in H1975 cells and discovered that the expression of γ-H2AX, 53BP1, H4K20Me2, and P-P53 were all decreased (Fig. 4E). In contrast, knocking down L3MBTL1 resulted in a reduction in DNA damage and apoptosis. We also verified these changes in PC9/OR by immunoblotting (Fig. S2C). These findings were further confirmed in a Comet assay of H1975/AR cells, in which the tail length increased following L3MBTL1 knockdown (0.9325 vs. 15.44, p < 0.001), whereas the tail length was further increased after L3MBTL1 knockdown and treatment with Osimertinib (15.44 vs. 28.70, p < 0.001) (Fig. 4F–4G). In addition, we found that si-L3MBTL1 in combination with Osimertinib induced apoptosis in H1975/AR cells as shown in Fig. S2D–S2E. An analysis of Annexin V-FITC/PI double-staining in H1975/AR cells revealed that apoptosis increased following L3MBTL1 knockdown (p < 0.001 and p < 0.05). These results demonstrate that L3MBTL1 enhances DNA damage and when combined with Osimertinib, DNA damage, genomic instability, and apoptosis are all increased.
The accumulation of DNA damage may be impacted by altered chromatin and chromosome structures. To determine whether L3MBTL1 is responsible for the abnormal compaction of chromatin structures in Osimertinib-resistant cells, Osimertinib was administered to H1975/AR SH-L3MBTL1 cells for 24h. A micrococcal nuclease digestion assay was then carried out on both cell lines to determine whether variations in L3MBTL1 expression alter chromatin conformation. Larger diffuse DNA fragments were observed in the chromatin of H1975/AR SH-L3MBTL1 cells. The DNA fragments were more degraded compared with that in the control group and after Osimertinib treatment for 24 h, the chromatin structure was looser (Fig. S2F–S2G). These results indicate that L3MBTL1 reduces the accumulation of DNA damage during Osimertinib treatment and increases resistance by condensing the chromatin structures in H1975/AR cells.
L3MBTL1 binds throughout the genome and plays a critical role in EGFR-TKI resistance
L3MBTL1 is considered a chromatin compaction factor[11]. We performed whole-genome ChIP-seq in H1975/AR cells to assess the distribution of L3MBTL1 throughout the genome. The results indicated the genome-wide distribution of L3MBTL1 (Fig. 5A). A total of 19,565 L3MBTL1 promoter peaks were counted, accounting for 4.8% of the transcription initiation points (TSS) using input as the background and Epic2 to call the IP peaks (Fig. 5B–5D). The peaks were subjected to gene association analysis and 4,782 peak-related genes were identified. A GO analysis indicated that the genes were enriched in cellular processes (e.g., cell cycle, cell metabolism, and cell stress response) and cell signal transduction (Fig. 5E). A KEGG analysis revealed that the genes associated with the L3MBTL1 promoter peaks were extensively enriched in EGFR and its downstream signaling pathways as well as some EGFR-TKI-related bypass pathways (Fig. 5F). These findings suggest that L3MBTL1 is an important player in the development of EGFR-TKI resistance in NSCLC cells.
The role of L3MBTL1 on the proliferation, invasion, and tumor sensitivity to Osimertinib acquired-resistant cells in vitro
To determine whether L3MBTL1 expression exerts a synergistic effect with Osimertinib in H1975/AR and PC9/OR cells, we divided the cells into four groups: a scrambled group, Osimertinib single-drug treatment group, si-L3MBTL1 group, and si-L3MBTL1 plus Osimertinib treatment group. The cells were transfected with siRNA for 48 h, whereas the control group was simultaneously transfected with the scrambled siRNA. Next, the Osimertinib group was treated with Osimertinib for 24 h. Colony formation assays revealed that cell proliferation was suppressed after knocking down L3MBTL1. In addition, cell proliferation was significantly reduced after knocking down L3MBTL1 in combination with Osimertinib (p < 0.01 and p < 0.001). Similar results were obtained in the L3MBTL1-1 and L3MBTL1-3 knockdown groups for H1975/AR cells (Fig. 6A–6B) and in the L3MBTL1-3 knockdown group containing PC9/OR cells (Fig. S3A). A Transwell assay was used to evaluate cell invasion. Knockdown of L3MBTL1 inhibited the invasion ability of H1975/AR cells, whereas treatment with Osimertinib in combination with L3MBTL1 knockdown inhibited invasion (p < 0.001) (Fig. 6C–6D). This suggests that L3MBTL1 inhibition combined with Osimertinib treatment significantly inhibits the proliferation and invasion capacity of H1975/AR cells.
To determine whether L3MBTL1 knockdown confers drug sensitivity in H1975/AR cells, we measured the IC50 values of cells treated with Osimertinib using a CCK-8 assay. After si-L3MBTL1-1 and si-L3MBTL1-3 knockdown of L3MBTL1, the IC50 values decreased by 31.9% and 50.9%, respectively, compared with the control group (p < 0.05 and p < 0.01) (Fig. 6E). We conducted the same test on H1975/AR cells following L3MBTL1 knockdown using lentivirus. The IC50 value of the L3MBTL1 shRNA knockdown group was reduced 72.3% compared with that of the control group (Fig. 6F), which supports the aforementioned findings. A potent and selective MBT inhibitor, UNC669, is more selective for L3MBTL1 compared with other MBT family homologs. When used alone, UNC669 had no effect on cell viability; however, when combined with Osimertinib, the IC50 of UNC669 (40 µM) combined with Osimertinib was reduced by 25.7% and 52.8%, respectively, compared with the control group (Fig. 6G). This suggests that UNC669 combined with Osimertinib may exert a synergistic effect on cell growth inhibition. The IC50 value was increased by 103.8% compared with the control group in H1975 cells overexpressing L3MBTL1 by CCK8 detection, indicating that the cells are resistant to Osimertinib (Fig. 6H). Knockdown of L3MBTL1 also markedly increased the sensitivity of PC9/OR cells to Osimertinib as the IC50 value was reduced by 44.2% (Fig. S3B). Taken together, our data supported the hypothesis that L3MBTL1 knockdown in NSCLC cells increases Osimertinib sensitivity, whereas L3MBTL1 overexpression confers resistance to Osimertinib.
L3MBTL1 regulates chromatin structure by combining with histone modification in the development of drug resistance
L3MBTL1 was previously show to maintain chromatin condensation by binding mono- and dimethylated histone H4K20 and compact nucleosome arrays[11]. Compared with H1975 cells, we found that H4K20Me1 and H4K20Me2 expression levels in H1975/AR cells were higher (Fig. 7A). Nonetheless, following treatment with Osimertinib for 24 h, the expression of H4K20Me1 did not change significantly, whereas the expression of H4K20Me2 gradually increased (Fig. 7B), which was a similar trend to that observed with L3MBTL1. Therefore, we hypothesized that H4K20Me2 plays an important role with L3MBTL1 in chromatin condensation in drug-resistant cells. We examined 53BP1, a DDR protein that competes with L3MBTL1 for binding to H4K20Me2[13]. Under normal circumstances, 53BP1 cannot bind to H4K20Me2 because L3MBTL1 competitively binds to it through a special MBT domain. Following DNA damage, L3MBTL1 is ubiquitinated, which exposes the spatial site of H4K20Me2. This facilitates 53BP1 binding to H4K20Me2 for recruitment to the site of DNA damage and further recruits other repair-related proteins to facilitate repair. Because of a variety of factors, 53BP1 may not be able to accumulate at the site of damage, which prevents it from recruiting downstream proteins for DNA repair[13]. To determine whether 53BP1 typically accumulates at DNA damage sites, we performed an immunofluorescence assay. The foci of 53BP1 and γ-H2AX, which represents sites of DNA damage, were observed by fluorescence microscopy after 24 h of Osimertinib treatment. The Pearson’s correlation coefficient was 0.25. Figure 7C shows that some of the 53BP1 foci did not aggregate to the DNA damage site (γ-H2AX) and protein overlap was decreased; therefore, we wondered if abnormal L3MBTL1 expression combined with abnormal binding of L3MBTL1 to H4K20Me2 would affect the aggregation of 53BP1. We carried out the same treatment on H1975/AR SH-L3MBTL1 cells and found that 53BP1 and γ-H2AX protein coincided with increased localization, and the Pearson correlation coefficient increased to 0.61 (p < 0.001) (Fig. 7C). This suggests that the proportion of 53BP1 accumulating at sites of DNA damage was increased and L3MBTL1 is a key factor that affects 53BP1 accumulation at DNA damage sites. We used siRNA to inhibit the expression of 53BP1 in H1975 and H1975/AR cells to further confirm the relationship between L3MBTL1 and H4K20Me2 and found that L3MBTL1 and H4K20Me2 expression levels were increased (Fig. 7E). Taken together, L3MBTL1 collaborates with H4K20Me2 to mediate the conformational changes in chromatin following Osimertinib treatment in H1975/AR cells.
The effect of L3MBTL1 on tumor growth and drug resistance in vivo
Tumor xenograft models of H1975/AR SH-NC and H1975/AR SH-L3MBTL1 cells were established in nude mice by subcutaneously injecting the cells into the right flank of the animals to determine the role of L3MBTL1 in Osimertinib resistance in vivo. When the tumor volume reached 150–200 mm3, the mice representing each cell type were treated with double-distilled water or Osimertinib (2.5 mg/kg/day) by gavage for 3 weeks (Fig. 8A). As shown in Fig. 8B–8C, tumor growth rate, volume, and tumor weight of the L3MBTL1 knockdown group were significantly lower compared with those of the control group (p < 0.01 and p < 0.01). The inhibitory effect of the combined Osimertinib treatment group was more obvious (L3MBTL1 knockdown group: p < 0.001 and p < 0.001; Osimertinib single treatment group p < 0.001 and p < 0.05). In paraffin-embedded tissue sections, the expression of Ki-67 in tumor tissues of the L3MBTL1 knockdown plus Osimertinib group was significantly lower compared with that of other groups, whereas the expression of γ-H2AX and 53BP1 increased (Fig. 8D). The results indicate that L3MBTL1 knockdown inhibits tumor growth and increases DNA damage.