M6A associated TSUC7 inhibition contributed to Erlotinib resistance in lung adenocarcinoma through a notch signaling activation dependent way

Background The small tyrosine kinase inhibitors (TKIs) subversively altered the lung cancer treatments, but patients will inevitably face the therapy resistance and disease recurrence. We aim to explore the potential roles of non-coding RNAs in sensitizing the TKIs effects. Methods: Multiple cellular and molecular detections were applied to confirm the mechanistic regulations and intracellular connections. Results We explored the specific gene features of candidates in association with resistance, and found that m6A controlled the stemness of EMT features through METTL3 and YTHDF2. The miR-146a/Notch signaling was sustained highly activated in a m6A dependent manner, and the m6A regulator of YTHDF2 suppressed TUSC7, both of which contributed to the resistant features. Functionally, the sponge type of TUSC7 regulation of miR-146a inhibited Notch signaling functions, and affected the cancer progression and stem cells’ renewal in Erlotinib resistant PC9 cells (PC9ER) and Erlotinib resistant HCC827 cells (HCC827ER) cells. The Notch signaling functions manipulated the cMYC and DICER inner cytoplasm, and the absence of either cMYC or DICER1 lead to TUSC7 and miR-146a decreasing respectively, formed the closed circle to maintain the balance. Conclusion PC9ER and HCC827ER cells harbored much more stem-like cells, and the resistance could be reversed by Notch signaling inactivation. The intrinsic miR-146 and TUSC7 levels are monitored by m6A effectors, the alternation of either miR-146 or TUSC7 expression could lead to the circling loop to sustain the new homeostasis. Further in clinics, the combined delivery of TKIs and Notch specific inhibitory non-coding RNAs will pave the way for yielding the susceptibility to targeted therapy in lung cancer. Supplementary Information The online version contains supplementary material available at 10.1186/s13046-021-02137-9.


Background
Lung cancer is the most dangerous cancer type worldwide, topping the cancer related mortality [1]. The lung cancer incidence tends to be stable, and even decreased slightly in western world [1,2], but situations on cancer prevention were severe in developing China [3][4][5]. Exploring more sensitive screening strategy, improving the radical operation methods, or developing more effective adjuvant therapeutic agents was so urgent than ever [6,7]. Lung adenocarcinoma consists of lots of therapy targets, and the EGFR Tyrosin Kinase related Inhibitors (TKIs) have been widely and effectively applied in clinical treatments for a decade, shortening the suffering process [8,9]. However, inevitably, the targeted therapies must face the recurrence, and receive the second or third line of chemo-radiotherapy [10][11][12], and to identify more novel and effective therapeutic molecules and agents will be helpful and promising.
Traditional non-coding genes were always referred to the miRNAs when researching the post-translational regulations, and the lncRNAs (long non-coding RNAs) and circRNAs (circular RNAs) were later revealed for supplementing the affection of RNA family of non-coding members [13][14][15][16][17]. Individually, they could act as crucial modulator toward to certain downstream genes in many ways [18,19]. The N-6-methyladenosine (m6A) regulation of RNAs attracted lots attention due to its specific and strong modification ability of epigenetic functions [20][21][22]. The m6A controller of writers, readers, and erasers could modulate the mRNA stability and translation, to mediate downstream effects [23,24]. However, the roles of m6A in controlling the non-coding RNAs biogenesis were still not clear. To further identify the candidates to improve the TKIs treatments sensitization, we tentatively explored the supporting role of TUSC7 in cancer suppression, trying to establish the m6A correlated lncRNA functions in modulating the TKIs therapies resistance.

Cell lines and cell culture
The human lung adenocarcinoma cells lines (PC9 and HCC827), and human embryonic cell line (HEK-293 T) were purchased from ATCC (American Type Culture Collection, VA, USA) or the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). The Erlotinibresistant cell lines (PC9/ER and HCC827/ER) were established by chronic exposure to increasing concentrations of drugs. The ectopic expression of TUSC or miR-146 in cells were constructed and restored as pervious study elucidated [25]. All cell lines were cultured in DMEM medium supplemented with 10% FBS (Gibco), 1% penicillin and 1% streptomycin, and incubated at 37 °C in a humidified atmosphere with 5% CO 2 .

Quantitative real-time PCR and western blot
Total RNA was extracted from cells using Trizol, according to the manufacturer's protocol, and reverse-transcribed into cDNA by using SYBR RT-PCR kit (Takara, JAPAN). Real-time quantitative PCR (RT-qPCR) was performed with SYBR Premix ExTaqTM II Kit (Takara, JAPAN). The sequences of the primers for PCR were synthesized by Sangon Company (Shanghai, China) and were listed in Supplemental Materials and Methods. The relative expression of mRNA and miRNA were calculated by using the formula:2 -ΔΔCΤ . For western blot analysis, the total protein from cell extracts was harvested using RIPA buffer contained protease inhibitors. The protein extracts were fractionated by 10% SDS-PAGE, transferred onto a nitrocellulose membrane, and then incubated with primary antibodies at 4 °C overnights, followed by HRP-conjugated secondary antibody (1:5000, #7074, Cell Signaling Technology) and visualized by using ECL Blotting Detection Reagents (Merck Millipore). The primary antibodies were as follows: anti-Notch1 (1:500; Rabbit mAb, #3608, Cell Signaling Technology), anti-Notch2

RNA-immunoprecipitation
The cells (5 × 10 6 )/ml were harvested and washed twice with ice-cold 1 × PBS buffer. Collected cell pellet was lysed for 15 min on ice by RIP buffer for 30 min, and pretreated with a 1:10 dilution in NT2 buffer. The cell lysate was further centrifuged at 15,000×g for 15 min at 4 °C, followed by treated with magnetic beads conjugated to human anti-ATG3 antibody (1:50) or the control IgG for 18 h at 4 °C and further washed twice with cold NT2 Buffer. Magnetic beads subsequently mixed with the diluted lysates (10 μl beads/ml lysate). The mixture was re-suspended in 100 μl NT2 Buffer containing 30 μg proteinase K to digest the protein. Co-purified RNA was extracted by the TRIzol reagent and used in subsequent qRT-PCR assay.

M6A methylation quantification assay
The m6A methylation status of cells detected using the m6A RNA Methylation Quantification Kit (Epigentek, Cat#P-9005-113) according to the manufacturer's instructions. In brief, 200 ng of total RNA was used as an input respectively. Then RNA samples were captured and detected by spectrophotometer (Bio Tek Instruments, Inc. US) at 450 nm. The level of m6A methylation was calculated according to the manufacturer's instructions.
As for LC-MS/MS assay, mRNA was purified from the total RNA using via oligo dT magnetic beads. Then 200 ng mRNAs were incubated with 0.5 U nuclease P1 in reaction system at 42 °C for about 1 h. After that, mRNAs were incubated with 3 μL of 1 M NH4HCO3 and 1 μL of 1 U/μL alkaline phosphatase at 37 °C for 2 h. And mRNAs were diluted and filtered. After which, a C18 column were used to separate mRNAs [26][27][28][29]. Then mRNAs were analyzed by an Agilent (6410 QQQ) triple-quadrupole LC mass spectrometer. Calibration curves were used to calculate the Ratio of m6A to A.

Nude mouse xenograft model
A total of 21 4-week-old female BALB/cA-nu nude mice were purchased from Beijing Huafukang Biosciences (Beijing, China), then we maintained them in specific pathogen-free conditions. Control vector, TUSC7 knockout, FLI-06 treated H1975 cells (1*10 7 ) cells were suspended in 100 μL of serum-free DMEM medium (Hyclone, USA), mixed with matrix gel (Corning, USA), and then were injected subcutaneously. The changes in the tumor size were recorded every 3 or 5 days. We calculated the tumor volume using the following formula: V = 1/2 × l × w2 (l is the longer axis, 2 is the shorter axis). All mice were sacrificed 28 days after the injection of cells. The dissected tumor samples were immersed in 4% paraformaldehyde (BioSharp, China) and embedded in paraffin.

Statistical analysis
Statistical analysis was carried out by using Graph Pad Prism 6 and SPSS 20.0 software (SPSS Inc., Chicago, IL, USA). All numerical data were expressed as mean ± standard deviation (SD). Experiments were carried out with three or more replicates. Two or more groups were assessed by using Student's t test or ANOVA individually. P < 0.05 was considered to be statistically significant.

Notch signaling activation in lung adenocarcinoma pointed to poor survival expectance
The important members of Notch signaling [30,31] were screened for expression patterns with using Pan-Cancer Atlas of TCGA data base. The total samples of 507 patients were collected, and the heat maps indicated the universal overexpression of notch signaling participants ( Fig. 1A-B), and EGFR was correlated with aberrant Notch expression. Deep analysis from TCGA (Nat Genet 2016 data base) indicated the grouped enrichment of Notch signaling factors (Fig. 1C), and the changing was consistency in groups harboring most irregulated Notch functions (Fig. 1D).

Erlotinib resistant cells harboring notch signaling activation and TUSC7 inhibition
Lung adenocarcinoma cell lines of PC9 and HCC827 were selected for their characteristics of specific EGFR mutant status, and the signatures of PC9 and HCC827 cells referring to erlotinib treatment were shown in Figure S1. The concentration of 0.2 μM Erlotinib was chosen as the function candidate, which decreased the Notch1 ( Figure S2A) and Notch2 ( Figure S2B) mRNA levels in PC9 and HCC827 cells. The Erlotinib treatment also increased the TUSC7 expression level significantly (Figure S2C), together with EGFR level slightly decreased ( Figure S2D). Blotting results showed that the Erlotinib treatment decreased the Notch signaling factors in PC9 and HCC827 cells ( Figure S2E). We constructed Erlotinib resistant lung cancer cells with advancing concentration gradient ( Figure S3), and the differentially expressed lncRNAs between Erlotinib sensitive and resistant cells were primarily detected by gene panel selection ( Figure S2F). The differentially expressed lncRNAs were pasted for GO analysis of functional identification ( Figure S2G). The detailed information showed that TUSC7 of both PC9ER ( Figure S2H) and HCC827ER ( Figure S2I) decreased greatly, and did not react to Erlotinib treatment, comparing to the Erlotinib sensitive cells. Last but not least Erlotinib failed to inhibit the Notch signaling functions in PC9ER and HCC827ER cells ( Figure S2H-I).

Notch signaling inhibition was required for TUSC7 alleviating of the Erlotinib resistance
The stem-like cells were accused for treatment resistance, and to study the stem cells' renewal ability in contributing to Erlotinib resistance, the ALDH1 phenotype and spheres formation nature were applied. The Erlotinib resistant PC9ER and HCC827ER cells consisted of more stem-like cells ( Fig. 2A-B), and Erlotinib treatment failed to decrease the stem cells number (Fig. 2C). Higher stem cells ratio indicated resistant status, and resulted in group resistance. The Notch inhibitors were chosen as FLI-06 (inhibitor-1, ab120633, Abcam), and γ-Secretase inhibitor (inhibitor-2, ab146170, Abcam). Both PC9 and HCC827 cells responded to Notch signaling inhibitors, with stem-like cells ratios decreasing significantly (Fig. 2D-E). The lowered concentration of Notch signaling inhibitors sensitized resistant cells to Erlotinib treatment (Fig. 2F, Figure S4A-B), indicating the synergistic effects of TKI agents and Notch signaling inhibitors.
To study the TUSC7 functions, Lentiviral based TUSC7 expression vesical was introduced into PC9ER and HCC827ER cells. Erlotinib alone inhibited the Notch signaling slightly, but TUSC7 effectively enhanced the Erlotinib functions in PC9ER (Fig. 2G) and HCC827ER cells (Fig. 2H), and stimulated the suppressive functions of Erlotinib in both PC9ER ( Figure S4C) and HCC827ER cells ( Figure S4D). Moreover, the addition of lowered concentration of Notch signaling inhibitor strengthened TUSC7 functions (Fig. 2G-H).
Next, we found that combined TUSC7 and Erlotinib decreased the stem cells ratio greatly in both PC9ER (Fig. 2I) and HCC827ER cells (Fig. 2J). The stem cells' renewal suppression evaluation did not show significant differences between TUSC7 alone and the combination of TUSC7 and FLI-06 ( Fig. 2K-L, Figure S4E-G). The combined TUSC7 and Erlotinib decreased the stem cells associated signatures, decreasing EMT markers in PC9ER cells ( Figure S5A) and HCC827ER cells ( Figure S5B).

TUSC7 sensitization of Erlotinib through miR-146a/notch signaling inhibition was dependent on NUMB restoration
To reveal the transduction mechanisms, bioinformatic screening of the possible connections between TUSC7 and its binding partners was conducted. We noticed that miR-146a shared common sequences with the untranslated regions of TUSC7 (Fig. 3A), and miR-146a mimics decreased the Luc-activity of TUSC7 significantly (Fig. 3B) in 293 T cells, which was also confirmed in PC9ER (Fig. 3C) cells and HCC827ER cells (Fig. 3D).
To testify the binding probabilities between TUSC7 and its downstream partner, biotin labeled sense and anti-sense RNAs of TUSC7 were used for RNA pull down detection, and the connection between TUSC7 and NUMB in resistant lung adenocarcinoma cells was identified (Fig. 3E). Further, RNA immunoprecipitation revealed that TUSC7 was enriched with NUMB expression in PC9ER cells (Fig. 3F, left) and HCC827ER cells (Fig. 3F right). Informatic screening of the potential miRNAs' targets suggested that miR-146a may bind to NUMB (Fig. 3G), and NUMB decreased greatly in cancer group (Fig. 3H). The alignment of miR-146a and the 3'UTR of NUMB was constructed through using the enhanced green fluorescent protein (EGFP) reporter assay. The wild-type 3'UTR sequence and the mutant 3'UTR sequence of NUMB were cloned downstream from the EGFP-coding sequence respectively, to construct the reporter plasmid and the mutant vector. The co-transfection of miR-146a mimics and the wide-type reporter plasmid strongly reduced the EGFP intensity (Fig. 3I, left), but not happened in mutant-type reporter plasmid (Fig. 3I, right). On the contrary, TUSC7 did not reduce the EGFP activity of the NUMB (Fig. 3J). MiR-146a decreased the NUMB expression, which could be rescued by TUSC7, and the TUSC7 inhibition (TUSC7in) also decreased the NUMB at the protein level (Fig. 3K).

m6A in resistant cells contributed to TUSC7 inhibition and miR-146a overexpression
The pluripotency status of resistant PC9ER and HCC827ER cells contributed to specific miR-146a and TUSC7 patterns, and to characterize the roles of m6A in therapy resistance, we investigated the variations of m6A levels, and identified that the m6A levels of total RNAs from resistant cells were statistically more abundant than sensitive original cells by using LC/MS (Fig. 4A). To further characterize the roles of m6A in generating the resistance, we used siRNAs to tentatively test the m6A related processers in controlling of TUSC7 and miR-146a. METTL3 affected the miR-146a level (Fig. 4B), and YTHDF2 affected TUSC7 level (Fig. 4C),and results were all confirmed by using the lentiviral based METTL3/YTHDF2 knock-down systems ( Fig. 4D-E). Dysregulated METTL3 (Fig. 4F) and YTHDF2 (Fig. 4G) affected the EMT and pluripotency features through controlling m6A status. Moreover, decreasing either METTL3 or YTHDF2 significantly altered Snail and EMT factors expressions. m6A at Snai1 mRNA was greatly increased in resistant cells [26,37], and we found the METTL3 inhibition decreased m6A at Snai1 (Fig. 4H-I), which later failed to activate the miR-146a promoter (Fig. 4J) [38]. In the other hand, the m6A at TUSC7 level increased in resistant cells (Fig. 4K), and the recognition of TUSC7 m6A peak by YTHDF2 degraded TUSC7 [39,40]. The Me-RIP (methylated RNA immunoprecipitation) assay confirmed that the high abundance of m6A modification in cells with YTHDF2 inhibition (Fig. 4L). All the results revealed that m6A determined the expression levels of miR-146a and TUSC7 in resistant cells, sustaining the balanced status.

TUSC7 formed feedback loop with miR-146/notch signaling and lead the Erlotinib re-sensitization
The unique m6A at different RNA sequences determined intrinsic RNA expressions, cellular homeostasis may be disturbed when manipulating certain lncRNA candidate. CMYC was proved previously to promote the translation of lncRNAs [41][42][43], and DICER1 was critical for miR-NAs maturation [44][45][46]. Results showed that in PC9ER and HCC827ER cells, Notch signaling inhibition lead to cMYC promoter activity decreasing, and TUSC7 inhibition strongly increased the activity (Fig. 5A). Inhibition of (See figure on next page.) Fig. 2 Notch inhibition decreased the self-renewal ability of Erlotinib resistant cells and re-sensitized the resistant cells to Erlotinib. A The addition of Erlotinib decreased the ALDH1A1 positive cells of PC9 and HCC827 cells significantly, but did not affect the ratios of Erlotinib resistant PC9ER and HCC827ER cells. B The addition of Erlotinib decreased the spheres number of PC9 and HCC827 cells significantly, but did not affect the number of Erlotinib resistant PC9ER and HCC827ER cells. C Representative images of ALDEFLUOR isolation were detailed exhibited. Two kinds of Notch signaling inhibitors, FLI-06 (inhibitor-1), and γ-Secretase inhibitor (inhibitor-2) were used. 200 nM of inhibitor-1 (D) decreased the self-renewal ability of multiple kinds of lung cancer cells, and 50 nM of inhibitor-2 (E) decreased the self-renewal ability of multiple kinds of lung cancer cells. F Notch signaling inhibition decreased the stem cells' ratio of the Erlotinib resistant cells significantly, and further, the much-lowered concentration of Notch signaling inhibitor-1, the 20 nM of FLI-06 sensitized both PC9ER and HCC827ER cells to Erlotinib treatment greatly. Erlotinib alone inhibited the Notch signaling slightly, and lowered concentration of FLI-06 mildly inhibited the Notch signaling, but effectively enhanced the Erlotinib functions in PC9ER (Fig. 3G) and HCC827ER cells (Fig. 3H). Combined TUSC7 and Erlotinib decreased the stem cells ratio greatly in both PC9ER and HCC827ER cells (Fig. 3I-J). MiR-146a conducted NUMB degradation was blocked by TUSC7 in a sponge combination manner. A The predicated connection sites between TUSC7 and its binding partners showed that miR-146a shared the common untranslated regions. B Recombined miR-146a mimics decreased the Luc-activity of TUSC7 in 293 T cells. MiR-146a decreased the Luc-activity of TUSC7 in PC9ER cells (C) and HCC827ER cells (D). E Blotting results referring to RNA pull-down test showed the connection between TUSC7 and NUMB in PC9ER and HCC827ER cells. F RNA immunoprecipitation revealed that TUSC7 was enriched with NUMB expression in PC9ER cells (Left) and HCC827ER cells (Right), and the IgG was set as the immunoprecipitation control, the MALAT1 was set as the primer control. G Informatic screening of the potential miRNAs' targets suggested that miR-146a may bind to NUMB. H The expression level of NUMB with 526 cancer and 59 normal samples in LUAD, and the expressions with 501 cancer and 49 normal samples in LUSC were evaluated, and NUMB decreased greatly in cancer group. I The co-transfection of miR-146a mimics and the wide-type reporter plasmid strongly reduced the EGFP intensity, and miR-146a mimics reduced nearly 40% of the TUSC7 luciferase intensity, but not happened in mutant-type reporter plasmid. J TUSC7 alone did not reduce the EGFP activity of NUMB. K MiR-146a decreased the NUMB expression, which could be rescued by TUSC7, and the TUSC7 inhibition (TUSC7-in) also decreased the NUMB level Li et al. J Exp Clin Cancer Res (2021) 40:325 TUSC7 increased miR-146a expression, and then stimulated the DICER1 activity (Fig. 5B). The scheme image represented the cMYC promoter detection modes (Fig. 5C), and the mutation blocked the EGFR/Notch signaling feedback loop on lncRNA transcription induction. Western blotting further confirmed that in PC9ER and HCC827ER cells, miR-146 and TUSC7 acted the opposite way to promote the DICER1/ CCND1 expression (Fig. 5D). TUSC7 and miR-146a was sustained by cMYC (Fig. 5E, left) and DICER1 (Fig. 5E, right) respectively. The regulative signaling pathways were drafted and illustrated for detailed exhibition (Fig. 5F), the dysregulated TUSC7 and miR-146a formed the feedback loop with their downstream effectors to sustain the new homeostasis.
In vivo study confirmed the effective inhibition of TUSC7 exhibited on tumor growth, and the Notch signaling inactivation by using FLI-06 also suppressed the in vivo tumor expansion (Fig. 5G), further proved its suppressive functions.

Conclusion
Lung cancer treatments have been refined greatly, with constantly and novelly emerging components, and the small molecular compounds improved the life quality and expected response to therapies in patients carrying sensitive EGFR mutants [47][48][49]. Surgery manners have been evolved continuously with little operation wound and duration, however, the improvement has brought little progress on overall survival [50][51][52]. The 1st generation of small tyrosine kinase inhibitors subversively altered the lung cancer treatments, and the "small step" has made the "major progress", increasing the PFS with little suffering.
Reality is cruel. No matter how ideal results the TKIs treatment has brought, patients with advanced lung cancer will inevitably face the therapy resistance and disease recurrence [53][54][55]. To improve the TKIs sensitization and to overcome resistance will be the headline goal in the current emergency [56]. Many factors were thought to be correlated with 1st generation of TKI functions' achieving, and certain genes and compounds sensitized the treatments by oncogenic signaling repression, but detailed mechanistic regulations have not been illustrated clearly ever [10,57].
We have been focusing on noncoding RNAs related cancer emergence, progression, stemness features and therapy response for decades, and have revealed some important factors that may contribute to [44,45,58]. Previously, we identified the miR-146 functions in lung adenocarcinoma with suppressive affections on stem cells' renewal, and the tentative exploration of TKIs treatment improvements will facilitate its practical application.
In this study, we identified the inhibitive roles of TUSC7 in lung cancer progression, and after acquiring the Erlotinib resistant cells, gene panel was used for massive assessing of the dysregulated non-coding RNAs. We first explored the features of candidate genes in association with resistance, and the activated miR-146a/Notch signaling was sustained in resistant stem cells in a m6A dependent manner of and METTL3/Snail cascade. M6A related YTHDF2 mediation of suppressed TUSC7 also contributed to resistant features. Functionally, m6A controlled the stemness of EMT features through METTL3 and YTHDF2 in resistant PC9ER and HCC827ER cells.
In detail, TUSC7 sensitized the Erlotinib effects, and decreased the stem cells ratio through Notch signaling inhibition. Bench study showed that, TUSC7 sponged to miR-146 and then released NUMB to control Notch signaling, the latter of which was critical for maintaining cancer stem cells (CSCs) pool. DICER1 and cMYC activity was critical for sustaining the non-coding RNAs maturation [41,43,46,59,60], the sponge style of TUSC7 regulation toward to miR-146 released the controlling of NUMB expression in PC9ER and HCC827ER cells, which manipulated the DICER and cMYC cascade inner cytoplasm [43,46,61,62]. The absence of either cMYC or DICER1 will lead to TUSC7 and miR-146 decreasing respectively, formed the closed circle to maintain the new balance.
To be concluded, the lncRNA of TUSC7 affected the cancer progression and stem cells renewal, and TUSC7 suppression of Notch signaling determines the Erlotinib treatment response. PC9ER and HCC827ER cells harbors much more stem-like groups, which dominated in therapy response, and their resistance could be reversed by Notch signaling inactivation. Interestingly and importantly, the intrinsic miR-146 and TUSC7 levels are monitored and sustained by m6A effectors, and disturbing the miR-146 and TUSC7 expression patterns will push themselves to form the circling loop to sustain the new homeostasis. Further in clinics, the combined using of TKIs and Notch specific inhibitory non-coding RNAs will pave the way for yielding the susceptibility to targeted therapy in lung cancer. A Notch signaling inhibition resulted in cMYC promoter activity decreasing, and TUSC7 inhibition strongly increased the activity. B Inhibition of TUSC7 increased miR-146a expression, and then stimulated the DICER1 activity. C The scheme image represented the cMYC promoter detection modes. D Western blotting confirmed that in PC9ER and HCC827ER cells, miR-146 and TUSC7 acted the opposite way to promote the DICER1/CCND1 expression. E TUSC7 and miR-146a was sustained by cMYC (Left) and DICER1 (Right) respectively. F The regulative signaling pathways were drafted and illustrated for detailed exhibition. G In vivo study confirmed the effective inhibition of TUSC7 exhibited on tumor growth, and the Notch signaling inactivation by using FLI-06 also suppressed the in vivo tumor expansion