KIAA1522 potentiates TNFα-NFκB signaling to antagonize platinum-based chemotherapy in lung adenocarcinoma

The human lung adenocarcinoma cell lines A549 and NCI-H1299 were acquired from the American Type Culture Collection (ATCC, Manassas VA, USA). The murine lung cancer cell line 889-DTC (889) was a kindly provided by professor Winslow [25, 26]. HEK293T cells were acquired from the American Type Culture Collection (ATCC, Manassas, VA, USA). Cell lines were maintained at 37 °C in 5% CO2 in Dulbecco’s modified Eagle medium supplemented with 10% fetal bovine serum.

Recombinant human TNFα (AF-300-01A) was purchased from PeproTech. Cisplatin (HY-17394) was from MedChemExpress, QNZ (S4902) and Cycloheximide (CHX) was from Selleck. The primary antibodies used in this work are as follow: KIAA1522 (WB, 1:1000; IHC, 1:200; HPA032050, Sigma ImmunoChemicals, St Louis, MO, USA), Phospho-NF-kB p65 (Ser536) (WB, 1:1000; CST, 3033); NF-κB p65 (WB, 1:1000; CST, 8242), Phospho-IkBα (Ser32) (WB, 1:1000; CST, 2859), IkBα (WB, 1:1000; CST, 4814), Phospho-IKKα/β (Ser176/180) (WB, 1:1000; IHC, 1:150; CST, 2697), IKKα (WB, 1:1000, CST, 11930), TNFR1 (WB, 1:1000, Proteintech, 21,574–1-AP), TNFR2 (WB, 1:1000; IHC, 1:50, Proteintech, 19,272–1-AP), GAPDH (WB, 1:1000, Santa Cruz, sc-25,778), β-actin (WB, 1:2000, Santa Cruz, sc-47,778), Ki67 (IHC, 1:200, GeneTex, GTX16667), cleaved caspase-3 (IHC, 1: 500, CST, 9664).

C-terminal Flag-tagged KIAA1522 were cloned into pLVX-IRES-ZsGreen plasmid. The forward shRNA sequences for targeting human KIAA1522 are as follow: sh-1: CCGGCCACGGCCTTCGATACTTATGCTCGAGCATAAGTATCGAAGGCCGTGGTTTTTG; sh-2: CCGGCCTACCTGTCGAAGTTGATTCCTCGAGGAATCAACTTCGACAGGTAGGTTTTTG. shRNA sequences were cloned into pLKO.1 plasmid (Sigma-Aldrich, Missouri, USA). The sgRNAs for targeting mouse kiaa1522 gene are as follow: sg-1: GCCGAGAGTGACAACCGTCA; sg-2: AGTGGGAGACCTCCTCATCT. The sgRNA sequences were cloned into lentiCRISPRv2 plasmid which was a gift from Feng Zhang (Addgene plasmid # 52961) [27] The lentiCRISPR-EGFP sgRNA1 plasmid (Addgene plasmid #51760) from Feng Zhang was used as sgRNA control [28]. HEK293t cells were used for lentivirus packaging. Lentiviral vectors expressing target genes were co-transfected with lentiviral packaging plasmids psPAX2 (Addgene plasmid, 12,260) and pMD2.G (Addgene plasmid, 12,259) with Lipofectamine® 3000 (Invitrogen). The medium containing lentivirus was harvested at 48 h and 72 h after transfection and used to infect cultured cell lines. Transduced cells were isolated by puromycin selection or FASC sorting.

The sgRNA sequences for AAV-mediated in vivo kiaa1522 editing are GGAAGTCAGGAAGGCGACGG and GCCGAGAGTGACAACCGTCA. Each sgRNA was connected next to a U6 promoter and serially cloned into the pAAV-U6-gRNA v2.0-CMV-NLS-Cre-3FLAG-P2A-EGFP-WPRE vector (OBiO Technology (Shanghai) Corp., Ltd), which express a nuclear localization signal fused Cre protein. The sgkiaa1522-expressing AAV was generated by co-transfection of HEK293T cells with AAV expression vector, AAV helper plasmid and AAV Rep/Cap plasmid. The cells were lysed by a freeze-thaw procedure at 72 h after transfection. The viral particles were purified by iodixanol step-gradient ultracentrifugation and concentrated by a molecular-mass-cutoff ultrafiltration device. a non-sense sequence GCACTACCAGAGCTAACTCA was also cloned into the pAAV vector to generated the sg-control virus. The B6-Gt (ROSA)26Sor^{tm1(CAG-LSL-cas9,-tdTomato)/}Nju (LSL-Cas9) mice (from Nanjing University Model Animal Resource Platform) were crossed by Kras^(LSL-G12D) mice (from Shanghai Model Organisms Center, Inc.) to generate the conditionally Cas9/ Kras^G12D expressing mice. 8–12-week-old mice were anesthetized by intraperitoneally injection of pentobarbital sodium, and then delivered intratracheally by the recombinant AAV particles (2 × 10¹¹/mouse). For in vivo cisplatin treatment, the mice were treated intraperitoneally by cisplatin (3.5 mg/kg B.W.) once a week for one month.

For in vivo tumorigenesis assays, 5 × 10⁵ 889 cells were subcutaneously injected into C57BL/6 mice (5–6 weeks age) for 1 month. 5 × 10⁶ A549 cells were subcutaneously injected into BALB/c nude mice (5–6 weeks age) for 2 months. The control and KIAA1522-overexpressed A549 cells were subcutaneously injected into C57BL/6 mice. Two weeks later, the mice were subjected to drug treatment: cisplatin (7 mg/kg B.W.) once a week and QNZ (1 mg/kg B.W.) twice a week via intraperitoneal injection. The tumor volumes were measured by the formula: (length×width²)/2.

To assess the inhibitory effects of small-molecular reagents in vitro, 1 × 10⁴ cells were seeded in 12-well plates. 72 h later the cells were treated with vehicle, cisplatin (10–20 μM), QNZ (500 nM), or combination for 48 h. Then the cells were fixed by methanol and stained with crystal violet (Beyotime, C0121).

Cells were seeded at 2000 cells in 200 μL DMEM per well in 96-well culture plates. At the indicated time points, 20 μl Cell Counting Kit-8 reagent (Beyotime, C0039) was added to each well and incubated at 37 °C for 2–4 h. The absorbance values (OD 450 nm) were measured using a spectrophotometer. The absorbance values at 600 nm were used as references.

889 cells with or without TNFα (10 ng/ml) treatment were transiently transfected with NFκB luciferase reporter plasmid (1 μg), and 20 ng of the Renilla luciferase plasmids. 48 h post-transfection, the firefly and Renilla luciferase activities were monitored using the Dual-Luciferase Reporter Assay System (Promega). NFκB signaling activity is determined by the ratio of firefly to Renilla luciferase activity.

Cells were lysed by RIPA buffer (Thermo Fisher Scientific, 89,901) with protease inhibitors cocktail (Roche Diagnostics, 05892970001) and phosphatase inhibitor cocktail (Roche Diagnostics, 04906845001). The lysates were clarified by centrifugation at 13000 g for 30 min at 4 °C. Protein concentrations were determined by BCA protein assay kit (Thermo Fisher Scientific, 23,225) followed by boiled with loading buffer. Protein samples (50–150 μg) were separated through SDS-PAGE, then transferred to nitrocellulose filter membrane (Pall Corporation) blocked and incubated with the primary antibodies. After washing with TBST, the blots were incubated with Anti-rabbit IgG HRP-linked antibody (CST, 7074) and Anti-mouse IgG HRP-linked antibody (CST, 7076), then visualized by the SuperSignal West Dura Extended Duration Substrate (Thermo Fisher Scientific, 34,076).

Co-immunoprecipitation (Co-IP) was performed using Protein G–agarose suspension (Millipore, 16–266). The cells were lysed by IP lysis buffer (Beyotime Institute of Biotechnology, P0013), and then incubated by 50 μl of Protein G–agarose suspension for 3 h at 4 °C on a rocking platform to reduce non-specific binding. After removing the beads, the supernatant was supplemented with the primary antibodies followed by incubation for another 3 h at 4 °C. A total of 100 μl of Protein G–agarose was then mixed to each sample, and the incubation was continued overnight on a rocking platform. The immunoprecipitates were collected by centrifugation and washed three times with the TBS. The agarose was boiled with loading buffer and subjected to western blot analysis.

The tissue microarrays and slides were deparaffinized, rehydrated, immersed in 3% hydrogen peroxide solution for 15 min, heated in citrate buffer (pH 6.0) for 25 min at 95 °C, and cooled for at least 60 min at room temperature. Between each incubation step, the slides were washed three times with PBS (pH 7.4). After blocked with 10% normal goat serum for 30 min at 37 °C and washed, the slides were incubated overnight at 4 °C with primary antibodies against target proteins and visualized using the PV-9000 Polymer Detection System (GBI, USA) following the manufacturer’s instructions or GTVisionTMIII Detection System/Mo&Rb (GeneTech, GK500710). After washing with PBS, the slides were counterstained with hematoxylin.

Protein expression levels were determined on the basis of staining intensity and the percentage of immunoreactive cells. Staining intensity was rated as 0 (negative), 1 (weakly positive), 2 (moderately positive), and 3 (strongly positive). The percentage of immunoreactive cells was graded as 0 (0%), 0.5 (1–10%), 1 (11–20%), 2 (21–50%), 3 (51–80%), or 4 (81–100%). The average of tumor cell staining intensity score multiplied by the percentage of positive cells score represented the final score of one sample. The prognostic value of certain protein was evaluated by univariate or multivariate cox regression analyses in different subtypes of patients using R packages survival and survminer. The Nomogram survival predictive model was constructed using RMS package. The optimal cutoff value of IHC staining scores were estimated by R package maxstat.

Total RNA was extracted by TRIZOL Reagent (Life technologies, CA, USA) according to the manufacturer’s instructions, then checked for RIN numbers to inspect RNA integrity by Agilent Bioanalyzer 2100 (Agilent technologies, CA, USA). Qualified total RNA was further purified using RNAClean XP Kit (Beckman Coulter, Inc. CA, USA) and RNase-Free DNase Set (QIAGEN, GmBH, Germany).

Library construction and sequencing were performed at the Shanghai Biotechnology Corp. RNA libraries were prepared for sequencing using VAHTS Universal V6 RNA-seq Library Prep Kit for Illumina (Vazyme, Nanjing, China) before submitted to Illumina Hiseq 2500 system. Clean data was generated by trimming the adaptor and filtering rRNAs using Seqtk (https://​github.​com/​lh3/​seqtk). Then the clean reads were mapped to mouse reference genome (GRCm38) with Hisat2 (version: 2.0.4) to generate the BAM files for each sample. The uniquely mapped fragments of genes were counted by Stringtie (version:1.3.0). Gene expression was evaluated by normalized Fragments Per Kilobase of exon model per Million mapped reads (FPKM) using TMM (trimmed mean of M values) methods. The raw data of RNA sequencing were deposited in Gene Expression Omnibus (GEO accession number: GSE146072).

The bioinformatic analyses were completed through R programming language (version 3.6.2) in RStudio software. In the processed RNA-seq data, the genes with more than 500 total counts were subjected to the edgeR analysis to estimate the fold change of each gene between groups. Under the criterion that FDR q-value≤0.05, and log2FC ≥ 1 or ≤ − 1. The pathway enrichment analysis and the gene-sets enrichment analysis (GSEA) were performed using ClusterProfiler package [29]. The gene expression matrix was transformed to an enrichment score matrix by GSVA package [30] for the comparison of different genesets between groups. The genes significantly down-regulated by KIAA1522-depletion in 889 cells were defined as the KIAA1522 positive regulated genes which were constructed to be a geneset by GSEABase package and subjected to the following data mining analyses.

We downloaded FPKM-normalized RNA-seq data of TCGA-LUAD and TCGA-LUSC datasets together with the associated sample information using TCGAbiolinks package [31]. The FPKM values were transformed to TPM values for the following calculation. The microarray-derived transcriptome datasets from GEO database were downloaded by GEOmirror package and the sample information were acquired by GEOquery package. The lung datasets include GSE3141, GSE8894, GSE13213, GSE11969, GSE37745, GSE31210, GSE30219, GSE50081, GSE43580 and GSE14814. For the GEO datasets including more than one histological type, the lung adenocarcinoma samples were selected for further usage. The single cell sequencing dataset E-MTAB-6149 [32] was downloaded from the ArrayExpress website. The data was processed by Seurat package [33] for tSNE dimension reduction and clustering. GSVA program was used to determine KIAA1522 signature score and the scores of genesets associated with TNFα-NFκB signaling and cisplatin resistance. Correlation analyses were performed by cor.test function, and visualized by ggpubr package. Datasets containing survival information were used to perform survival analyses and cox regression analyses by survival and survminer package. Cutoff values for the tested factors were estimated by maxstat package. Meta-analyses were performed using metafor package in the fixed-effects model and the visualization was achieved by either forestplot or ggplot2 package. Alluvial diagram was drawn by alluvial package.

Platinum-based adjuvant chemotherapy is the mainstay of post-surgery management for non-small cell lung carcinoma (NSCLC). However, a group of patients experienced drug insensitivity. To identify key genes driving and predicting chemoresistance. We initially screened our previously identified candidate NSCLC biomarkers [21, 34] for their prognostic value in the patients receiving platinum-based chemotherapy. After preliminary immunochemical assays in a few paired NSCLC and non-tumoral tissues, there were twenty-two candidate expression-altered proteins that were tested in a total of 598 primary NSCLC samples plus 500 adjacent non-tumoral lung tissues dissected from 598 patients, showing extremely high expression of these proteins in tumoral specimens (Supplementary Fig. 1A). Univariate cox analyses were used to examine their association with chemotherapeutic outcome, illustrating that the predictive effectiveness of KIAA1522 were more predominant than other proteins (Fig. 1a). Multivariate cox analysis demonstrated that KIAA1522 was an independent prognostic factor in NSCLC, in either patients receiving chemotherapy or all NSCLC patients (Supplementary Fig. 1B). Similarly, in a nomogram predictive model of NSCLC patients, expression levels of KIAA1522 contributed to lower survival rates (Supplementary Fig. 1C-D).

Immunohistochemistry assays in paired tumoral and non-tumoral tissues showed that KIAA1522 was significantly elevated in both adenocarcinoma and squamous cell carcinoma (Fig. 1b-c). To distinguish the chemo-sensitive predictive roles of KIAA1522 in different histological subtypes. We performed multivariate cox analyses in adenocarcinoma and squamous cell carcinoma separately. The results revealed that KIAA1522 displayed significant effect in adenocarcinoma patients but not in squamous cell carcinoma (Fig. 1d). In our cohort of NSCLC patients, there is no difference in survival rate between histological types in both chemo-naive and chemotherapy-experienced patients (Supplementary Fig. 1E). The platinum-based chemotherapy seems to improve the survival probability, but hardly separate the Kaplan–Meier curves significantly (Supplementary Fig. 1F). Remarkably, in both ADC and SCC, incorporating the expression levels of KIAA1522 successfully stratified one group of lung adenocarcinoma patients with the most favorable outcomes, who express lower level of KIAA1522 and underwent platinum-based chemotherapy (Fig. 1e). When the NSCLC patients were grouped by histological types and KIAA1522 levels, chemotherapy in KIAA1522-low expression group of adenocarcinomas was more efficient than other groups (Fig. 1f). In TCGA lung cancer datasets, we found that the expression of KIAA1522 elevated in both adenocarcinomas and squamous cell carcinomas (Supplementary Fig. 2A), but only in the adenocarcinoma datasets, KIAA1522 expression connected with poor prognosis (Supplementary Fig. 2B). Above all, the clinical analyses highlight the predictive value of KIAA1522 in both overall survival and survival after platinum-based chemotherapy. Meanwhile, the results suggest the tumor-promoting and chemo-resisting roles of KIAA1522 in lung adenocarcinomas.

We employed an oncogenic Kras-induced lung adenocarcinoma model to study whether and to what extend KIAA1522 influence tumorigenesis and chemoresistance. The kiaa1522 specific sgRNAs were integrated together with a Cre expressing cascade into recombinant adeno-associated virus (AAV). Then, the rAAVs were delivered intratracheally to the LSL-Cas9/LSL-Kras^G12D mice (Fig. 2a). The mice delivered by control sgRNA and kiaa1522 sgRNA were divided into two groups treated by vehicle and cisplatin respectively (Fig. 2a). After the mice were sacrificed, the lungs were weighted (Fig. 2b), then subjected to Bouin′s staining or HE staining (Fig. 2c). The tumoral lungs in the kiaa1522 sgRNA-introduced mice had reduced lung weights compared to control mice. Upon cisplatin management, the lung weights were further lost in the kiaa1522-edited mice (Fig. 2b). When comparing the tumor burden between groups, we found that down-regulation of KIAA1522 dramatically attenuated Kras^G12D-induced lung adenocarcinoma in situ. More importantly, the depleting of KIAA1522 synergistically improved the efficiency of cisplatin in shrinking lung tumors (Fig. 2c-e, Supplementary Fig. 2C). The following survival analysis also substantiated the benefits of KIAA1522 down-regulation which not only elevated the survival rate of tumorigenic mice but also made them more sensitive to cisplatin therapy (Fig. 2f). Coincidently, the combination of KIAA1522 depletion and cisplatin treatment considerably induced apoptosis in vivo, monitored by immunohistochemical staining of cleaved-caspase 3 (Fig. 2g).

We also down-regulated KIAA1522 expression in the established lung adenocarcinoma cell lines and analyzed by their tumorigenic capability and cisplatin response. The results showed that loss of KIAA1522 severely impaired subcutaneous tumors (Fig. 3a-b), and reduced the amount of Ki67 positive cells in vivo (Fig. 3c). Moreover, down-regulation of KIAA1522 reduced the IC50 value in response to cisplatin (Fig. 3d). The KIAA1522 depleted cells were more sensitive to cisplatin treatment in vitro (Fig. 3e). Oppositely, overexpression of KIAA1522 rendered the cells chemoresistance to cisplatin (Fig. 3f).

The above results strongly indicate that KIAA1522 is required for both the tumorigenesis and resistance to cisplatin in lung adenocarcinoma, these facts coincide with the clinical relevance of KIAA1522 in lung adenocarcinoma patients.

To understand the molecular basis of KIAA1522-mediated acquirement of chemoresistance to cisplatin, we performed transcriptional profiling assays to detect dysregulated genes upon KIAA1522 depletion in 889 lung adenocarcinoma cells (Fig. 4a). Enrichment analysis of KIAA1522 positive regulated genes using HALLMARK signatures showed that the geneset HALLMARK_ TNFA_ SIGNALING_ VIA_ NFKB pathway was on the top of enriched pathways (Fig. 4b), suggesting that KIAA1522 may enhance TNFα-NFκB signaling pathway. This conclusion was further strengthened by gene set enrichment analysis (GSEA) and gene set variation analysis (GSVA) analysis. GSEA and GSVA results both showed that TNFα-NFκB signaling associated genesets were down-regulated in the KIAA1522 depletion cells (Fig. 4c-d). Similar to TNFα-NFκB signatures, The GSVA scores of cisplatin resistance signatures and cisplatin-responsive signatures were also down-regulated in the 889 cells expressing sg-kiaa1522. While the gene signatures negatively related to cisplatin resistance were up-regulated in KIAA1522 down-regulated cells (Fig. 4e).

In the TCGA-LUAD dataset, when considering KIAA1522 expression (KIAA1522 mRNA) and TNFα/NFκB status together, the patients with low KIAA1522 and TNFα signal levels or low KIAA1522 and NFκB activity had the best outcomes (Supplementary Fig. 3A; Fig. 4f). In addition, we also observed the correlation of KIAA1522 expression with both TNF signaling (Supplementary Fig. 3B), and NFκB activity (Fig. 4g) in the GSEA assays. TNFα-NFκB signaling has been reported to cause cisplatin insensitivity in several types of malignancies [13]. We found that inhibition of KIAA1522 reduced a set of genes down-stream of NFκB signaling, including anti-apoptotic genes confering cisplatin resistance, such as Xiap, Bcl2 and Birc5/survivin [15] (Supplementary Fig. 3C). So, the hyperactivation of TNFα-NFκB signaling may be a key molecular event responsible for KIAA1522-induced counteractivity to cisplatin.

We next explore whether KIAA1522 promotes TNFα-NFκB signaling transduction in the cultured cells. We found that down-regulation of KIAA1522 by shRNA or Cas9/sgRNA substantially inhibited the basal levels of phosphorylated NF-κB p65/RelA (Fig. 5a, Supplementary Fig. 4A). In the presence of exogenous TNFα to stimulate NFκB signaling in the serum-deprived cells, loss of KIAA1522 remarkably alleviated NFκB signaling activity monitored by luciferase reporter (Supplementary Fig. 4B). Western blotting assays showed that TNFα was less effective to up-regulated the phosphorylation of p65, IKKα/β and IκB in the KIAA1522-downregulated cells (Fig. 5b-c, Supplementary Fig. 4C-D), whereas forced expression of KIAA1522 yielded an opposite effect, which expedited the activation of NFκB signaling (Fig. 5d, Supplementary Fig. 4E). Moreover, we found that depletion of KIAA1522 decreased the abundance of the Tumor Necrosis Factor receptors, TNFR1 and TNFR2 (Fig. 5e, Supplementary Fig. 4F-G). Overexpression of KIAA1522 significantly increased TNFR2 in parallel with the activation of p65, but not TNFR1 (Fig. 5f, Supplementary Fig. 4H). So, the up-regulation of TNFR2 may contribute to the KIAA1522-mediated TNFα-NFκB activation. The AAV-mediated deletion of KIAA1522 in Kras^G12D-induced murine lung cancer also repressed TNFR2 and NFκB signaling, as indicated by phosphorylated IKKα/β (Fig. 5g). The results verified the regulation of TNFα-NFκB pathway in vivo. Next, we also explore the potential mechanisms underlying KIAA1522 regulating TNFR2, we found that depletion of KIAA1522 did not decrease the mRNA expression of KIAA1522 (Supplementary Fig. 5A), while KIAA1522 interacted with TNFR2 (Supplementary Fig. 5B) and contributed to protein stability of TNFR2 (Supplementary Fig. 5C).

To further underpin the links between KIAA1522 and TNFα-NFκB signaling and cisplatin responsiveness by clinical evidences, we performed integrative analyses using transcriptome profiles of lung adenocarcinoma patients from multiple cohorts, including single cell transcriptome studies. Considering the lake of KIAA1522 probes in several platforms, we defined a KIAA1522_ POS_ REG_ GENES geneset by KIAA1522 positive regulated genes in 889 cells (Fig. 4a) to calculate KIAA1522 signature score representing KIAA1522 expression amount, in bulk- or single cell-level transcriptome assays.

In the lung cancer single cell transcriptome dataset E-MTAB-6149, we clustered the cells by Seurat software (Supplementary Fig. 6A) and filtered tumor cells by the expression of EpCAM (Supplementary Fig. 6B). Then the clusters were further classified by the expression of ADC marker NAPSA/ Napsin A and SCC marker TP63 / P63. Some groups of cells expressing NAPSA and negative for TP63 were identified as lung adenocarcinoma cells (Fig. 6a). Before in-depth analyses, we examined the reasonability of using the GSVA scores derived from KIAA1522 downstream signatures. Firstly, in the sorted adenocarcinoma cells, we observed that the KIAA1522 signature scores were much higher in the KIAA1522 mRNA positive cells than that in KIAA1522 negative cells, suggesting the consistency of KIAA1522 mRNA levels and KIAA1522 signature scores (Fig. 6b). Secondly, we analyzed the clinical relevance of cisplatin resistance signatures and found that the two selected signatures were both correlated with poor survival in the lung adenocarcinoma patients after cisplatin-based therapy (Supplementary Fig. 6C). indicating these GSVA scores recapitulated physically resistance to cisplatin. Pearson correlation analyses indicated that the KIAA1522 signature scores were correlated with the TNFα signaling, NFκB activity and cisplatin resistance associated signatures (Supplementary Fig. 6D). The 3D plot clearly exhibited that the single cells with high levels of KIAA1522 signature scores were hyperactivated in TNFα-NFκB signaling and highly resistant to cisplatin simultaneously (Fig. 6c). Besides the representative genesets, we also showed that the KIAA1522 scores were highly consistent with the GSVA scores of a handfuls of TNFα-NFκB associated signatures (Fig. 6d).

In the bulk RNA sequencing dataset TCGA-LUAD, we also observed a similar distribution of the scores of all the signatures tested including the signatures related to TNFα-NFκB signaling, cisplatin resistance and KIAA1522 signature scores (Supplementary Fig. 7A). To strengthen this result, systematic correlation analyses showed that the positive correlations were universal in multi-central datasets (Fig. 6e). The correlations of KIAA1522 signature scores with TNFα-NFκB- cisplatin resistance signatures were independent of oncogenic KRAS and TP53 mutations (Supplementary Fig. 5B). Furthermore, meta-analysis was performed to review the prognostic effects of KIAA1522 signature score in independent datasets with survival information. The results showed that high level of KIAA1522 signature score predicted poor outcome in most of the datasets (Fig. 6f). The fixed effects model estimated hazard ratio was 1.918 (1.723–2.113) with a P-value< 0.001, this is in agreement with the conclusion detected in our cohorts of patients. More importantly, there are 39 cases in the dataset GSE14814 received platinum-based adjuvant chemotherapy. In this specific cohort, lower level of KIAA1522 scores also implicated good therapeutic results (Fig. 6g). When considering the combined effects of KIAA1522 score and NFκB status, the result showed that the groups of patients with both low levels of KIAA1522 and NFκB activity had the best outcomes (Fig. 6h-i).

To test whether hyperactivation of NFκB signaling is critical for the KIAA1522-induced chemoresistance, we used a NFκB inhibitor QNZ to block NFκB signaling activity (Fig. 7a). QNZ ameliorated the irresponsiveness of the exogenous KIAA1522 expressed cells (Fig. 7b-c). Coherently, in the presence of QNZ, either the control cells or the KIAA1522-depleted cells yielded similar effects in response to cisplatin (Fig. 7d-e). These results substantiated that NFκB signaling is engaged in KIAA1522 mediated cisplatin resistance, and enlightened us to cooperatively use of NFκB inhibitor and cisplatin to circumvent the treatment-refractory nature of lung adenocarcinomas expressed high level KIAA1522. To exam this speculation, we performed the in vivo pharmacology experiment. The results revealed that although forced expression of KIAA1522 did not accelerate tumorigenesis, but it desensitized inhibitory effects of cisplatin in vivo. The KIAA1522-induced cisplatin resistance, like the in vitro observations, dampened by NFκB inhibition (Fig. 7f). Collectively, synergic usage of NFκB inhibitor and cisplatin counteracted the cisplatin-refractory phenotype of KIAA1522 overexpressed lung adenocarcinoma cells.