RNA-binding protein KHSRP promotes tumor growth and metastasis in non-small cell lung cancer

The human NSCLC cell lines A549, NCI-H1299, NCI-H838, NCI-H358, and NCI-H292 and human embryonic kidney HEK-293 T cells were obtained from American Type Culture Collection (ATCC, Manassas, VA). All cell lines were cultured according to ATCC protocols in medium supplemented with 10% fetal bovine serum (Biowest, South America origin), 100 u/ml penicillin (Sigma-Aldrich, UA), and 100 μg/ml streptomycin (Sigma-Aldrich, UA) at 37 °C in a humidified atmosphere containing 5% CO2.

The nuclear proteins of NCI-H1299 and NCI-H358 cells were extracted and quantified by BCA protein quantitative method. iTRAQ labeling was conducted using an iTRAQ Reagent 4-Plex kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s protocol. Briefly, 100 μg lysate of each sample was reduced with tris- (2-carboxyethyl) phosphine and alkylated with methyl methanethiosulfonate and then digested overnight at 37 °C with trypsin (mass spectrometry grade, Promega, Madison, WI). The ratio of trypsin to protein was 1:20. The iTRAQ labeled samples were then combined according to the specified set and transferred into a new EP tube, desalted with Oasis HLB cartridges (Waters, Milford, MA) and dried in a vacuum centrifuge (Concentrator Plus, Eppendorf, Germany) [7].

A NanoLC system (NanoLC-2D Ultra, Eksigent, Dublin, CA) equipped with a Triple TOF 5600 mass spectrometer (AB SCIEX, USA) was used for analysis. Peptides were trapped on a NanoLC pre-column (Chromxp C18-LC-3 μm, size 0.35 × 0.5 mm, Eksigent, Dublin, CA) and then eluted onto an analytical column (C18-CL-120, size 0.075 × 150 mm, Eksigent, Dublin, CA). The NanoLC gradient was 5–35% Buffer B (98% ACN, 2% H2O, 0.1% FA) over 120 min at a flow rate of 300 nL/min. Full-scan MS was performed in positive ion mode with a nano-ion spray voltage of 2.2 kV. Survey scans were acquired from 350 to 1500 (m/z) with up to 40 precursors selected for MS/MS (m/z 100–1500). The collision energy (CE) for collision-induced dissociation was automatically controlled using an Information-Dependent Acquisition CE parameter script to achieve optimum fragmentation efficiency. The mass spectrometer was calibrated using beta galactosidase tryptic peptides.

Protein identification and iTRAQ quantitation were performed with ProteinPilot 4.5 software (AB SCIEX, USA). A strict unused confidence cutoff > 1.3 and more than two peptides were used as the qualification criteria. The peptide confidence level was 95%. Proteins with a fold change larger than 1.5 or less than 0.67 with a Student’s t-test p-value <0.05 were selected as differentially expressed proteins. The up-regulated and down-regulated proteins were analyzed by DAVID Bioinformatics Resources 6.7 (http://​david.​abcc.​ncifcrf.​gov). The up-regulated and down-regulated proteins were analyzed from the three GO terms of biological processes (BP), cellular component (CC) and molecular functions (MF), respectively.

Samples were analyzed on the mass spectrometer in two phases: data-dependent acquisition (DDA) was followed by SWATH acquisition on the same sample with the same gradient conditions used. The detailed parameter setting and library generation methods could be referred to the reference [8]. We selected proteins with at least two peptides identified for relative quantitation analysis. With three technical replicates for each sample, analysis of relative quantitation was performed by t-test analysis. The protein levels with a p-value less than 0.05 were considered significantly different in our experiment.

Total RNA was extracted from the lung cancer cells and tissues using TRIzol Reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s protocol. cDNA was synthesized by random primers and the PrimeScript RT Reagent Kit (Takara, Dalian, China). The primer sequences for real-time PCR are shown in Additional file 1: Table S1. Real-time polymerase chain reaction (qPCR) was performed using SYBR Premix Ex Tag (Takara, Dalian, China). The PCR conditions were as follows: 95 °C for 15 s followed by 40 cycles of 95 °C for 5 s and 60 °C for 30 s. The PCR primers used are listed in Table S1. β-actin was used as the internal control.

Cell and tissue proteins were extracted using T-PER® Protein Extraction Reagent (Thermo Scientific, USA) with a phosphatase inhibitor cocktail (Roche Applied Science, Switzerland) and a proteinase inhibitor cocktail (Roche Applied Science, Switzerland). The protein concentrations were determined using the Pierce™ BCA Protein Assay Kit (Thermo Scientific, USA). Proteins were separated by SDS-PAGE gels and transferred to nitrocellulose (NC) filter membranes (Millipore, Massachusetts, USA) or polyvinylidene fluoride (PVDF) membranes (Millipore, Massachusetts, USA). The membranes were incubated with primary antibodies overnight at 4 °C and probed with secondary antibodies at room temperature for 1~2 h. The following antibodies were used: anti-KHSRP (Abgent, 1:500), anti-HNRNPC (Sigma-Aldrich, 1:1000), anti-IFN-α (Santa Cruz, 1:500), anti-JAK1 (Santa Cruz, 1:500), anti-JAK2 (Santa Cruz, 1:500), anti-p-Stat1 (Sigma-Aldrich, 1:1000), anti-Stat1 (Sigma-Aldrich, 1:1000), anti-p-Stat3 (Sigma-Aldrich, 1:1000), anti-p-Akt (CST, 1:1000), anti-p-Erk (CST, 1:1000) and anti-α-Tubulin (CST, 1:1000).

Small-interfering RNA (siRNA) oligonucleotides for KHSRP and HNRNPC were designed and synthesized by RiboBio (Guangzhou, China). The primer sequences for the siRNAs are shown in Additional file 1: Table S2. Transient transfection was performed using Lipofectamine 2000 Reagent (Invitrogen, Carlsbad, USA) according to the manufacturer’s instructions. After transfection for 48 h, the cells were used for functional assays, including migration, invasion, RNA extraction, and Western blotting.

Cells were seeded in 96-well plates at 1 × 10³ cells per well and cultured in a final volume of 100 μl of culture medium supplemented with 10% FBS. According to the manufacturer’s instructions, 10 μl of Cell Counting Kit-8 (CCK-8, Dojindo, Kumamoto, Japan) was added to each well, and the mixture was incubated at 37 °C for 2 h. The absorbance was measured at 450 nm.

Cell migration and invasion assays were performed in 24-well plates with 8-μm-pore size chamber inserts (BD Biosciences, New Jersey, USA). For the migration assays, 5 × 10⁴ cells in 200 μl of serum-free culture medium were seeded into each well of the upper chamber with the noncoated membrane, and 800 μl of medium supplemented with 10% FBS was added to the lower chamber. For invasion assays, 1 × 10⁵ cells in 200 μl of serum-free culture medium were seeded into each well of the upper chamber with the Matrigel-coated membrane, and 800 μl of medium supplemented with 10% fetal bovine serum (FBS) was added to the lower chamber. The cells that migrated through the membrane were fixed with 100% methanol, stained with 0.1% crystal violet for 30 min, imaged and counted under a light microscope (Olympus, Japan).

For the wound-healing assay, cells were seeded into 24-well plates and grown to approximately 90% confluence. The monolayers were scratched with a 200 μl sterile pipette tip. The cells were rinsed with phosphate-buffered saline and cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) (or Roswell Park Memorial Institute 1640 medium, RPMI 1640) supplemented with 1% fetal bovine serum. The wounded monolayers were photographed at 0 h and 24 h after the wounds were made.

All animal experiments were approved by the Animal Ethics Committee of the Shanghai Cancer Institute. Six- to eight-week-old female BALB/c-nu/nu mice were bred by Shanghai Cancer Institute (Shanghai, China) and housed in specific pathogen-free (SPF) conditions in a laboratory animal facility.

For the in vivo xenograft assays, 3 × 10⁶ A549 cells stably expressing shKHSRP or the negative control and 2 × 10⁶ NCI-H292 cells stably expressing KHSRP or the lentiviral vector were separately subcutaneously inoculated into the dorsal right flanks of the nude mice (6 per group). The tumor size was measured two times every week. The tumor volume (V) was measured by calipers and calculated according to the following formula: (length × width × width)/2. After eight or 10 weeks, the mice were sacrificed, and the tumors were harvested at necropsy and fixed in 10% neutral PB-buffered formalin. The fixed tumors were stained with hematoxylin and eosin (H & E).

For the in vivo metastasis assays, 3 × 10⁶ A549 cells stably expressing shKHSRP or the negative control (8 per group) and 2 × 10⁶ NCI-H292 cells stably expressing KHSRP or the lentiviral vector (10 per group) were separately injected into the lateral tail veins of nude mice. After eight or 10 weeks, the mice were sacrificed, and their lungs were harvested at necropsy and fixed in 10% neutral PB-buffered formalin. The fixed lung tissues were stained with H & E and analyzed for the presence of metastasis.

3 × 10⁶ A549 cells stably expressing shKHSRP or the negative control (8 per group) were separately injected into the lateral tail veins of nude mice. After eight or 10 weeks, the mice were sacrificed. We used a Berthold LB983 NightOwl System (Berthold, Bad Wildbad, Germany) to monitor the tumour lung metastasis. For ex vivo biofluorescence imaging (ex vivo BFI), mice lungs were excised and placed in the chamber of the NightOwl LB 983 Molecular Light Imager and imaged.

HEK-293 T cells with Flag- and HA-tagged-KHSRP or control were routine cultured. When the cells fullness reached more than 90%, the cells were scraped off directly with a cell scraper using immunoprecipitation lysis buffer (Beyotime Institute of Biotechnology, Shanghai, China) with protease and protein phosphatase inhibitors (Roche Applied Science, Switzerland). Three milligrams of protein were incubated with 30 μl of protein A/G magnetic beads (Millipore, Massachusetts, USA) for 2 h. The beads were removed, and 12 μl of the primary antibody (Flag, HA, KHSRP or HNRNPC) or isotype IgG was added to the supernatant at 4 °C overnight with gentle mixing on a rocking platform to capture the fusion proteins. Then, 40 μl of protein A/G beads was added to each immunoprecipitation mixture for 4 h. The magnetic beads were collected by placing the tube in the appropriate magnetic separator. The beads were washed three times with cooled IP lysis buffer to remove the nonspecifically bound proteins. The bound fusion proteins were eluted from the beads and denatured by boiling for Western blot analysis. The bound fusion proteins were separated by SDS-PAGE and stained with a Silver Staining Kit (Beyotime Biotechnology, Shanghai, China). The specific bands were identified, and the peptides were digested with trypsin. The peptides were then analyzed on an Orbitrap Fusion mass spectrometer (Thermo Fisher, Waltham, USA). Protein identification was performed using Protein Pilot 4.5 (AB Sciex, Texas, USA) [9].

Fresh human non-small cell lung cancer tissues and matched adjacent noncancerous tissues for real-time PCR and Western blot analyses were collected from the Department of Lung Cancer at Shanghai Chest Hospital affiliated with Shanghai Jiao-tong University and the Department of Cancer at Huashan Hospital affiliated with Fudan University, Shanghai, China between 2011 and 2018. During the operation, human surgical specimens were immediately frozen in liquid nitrogen and stored at −80 °C for further investigation [10]. All of the tissue specimens for this study were obtained with patient informed consent. The study was approved by the Ethics Committee of Shanghai Jiao-tong University and the Ethics Committee of Fudan University.

A tissue microarray containing 75 paired NSCLC tissues and matched adjacent noncancerous tissues was purchased from Shanghai Biochip Co., Ltd. (Shanghai, China). Immunohistochemical staining was performed to detect the expression of KHSRP and HNRNPC in NSCLC tissues and matched noncancerous tissues. The average gray value of the image was used as a quantitative evaluation of the expression level using Image-Pro Plus 6.0 software.

To discover nuclear proteins potentially associated with cancer metastasis, we used human lung cancer high metastatic NCI-H1299 cells and low metastatic NCI-H358 cells as experimental materials in combination with iTRAQ and SWATH™, two proteomics methods, to analyze and identify differentially expressed nuclear proteins. In the iTRAQ experiments, a total of 4324/4353 proteins (global FDR < 1%) were identified using Protein Pilot 4.5 in two technical replicates. An unused protein score > 1.3 and peptides ≥2 were used as criteria, and 3987/3896 proteins were identified; 3859/3796 proteins were quantified. We identified 162 upregulated proteins and 157 downregulated differentially expressed nucleoproteins (Fig. 1a). In the SWATH™ experiments, a total of 1129 proteins (global FDR < 1%) were identified, and 867 proteins were quantified with peptides ≥2 and P < 0.05 in three technical replicates. We identified 180 upregulated and 263 downregulated differentially expressed nucleoproteins. We observed a similar distribution pattern at the protein level in the two approaches, and most quantified proteins were distributed close to 0 and within a ratio range of ±2 Log₂ (Additional file 1: Figure S1A). The consistency and correlation between the quantitative iTRAQ and SWATH™ methods were good (R² = 0.7975) (Fig. 1b). Through bioinformatics analyses, the experimental data for the differentially expressed nuclear proteins obtained by a pair of high and low metastatic cells and two protein quantitative methods were integrated. After the intersection, 116 lung cancer metastasis-related differentially expressed nucleoproteins were identified, including 52 upregulated proteins and 64 downregulated proteins (Fig. 1c, d; Additional file 1: Table S3, Additional file 1: Table S4). Gene Ontology (GO) analysis revealed that these proteins are mainly involved in DNA metabolism, mRNA metabolism, RNA cleavage, DNA processes, RNA processes and other functions. The molecular functions mainly included nucleotide binding, ribonucleotide binding, ATP binding, etc. (Additional file 1: Figure S1B, C, D). The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://​proteomecentral.​proteomexchange.​org) via the iProX partner repository with the dataset identifier PXD015861.

According to the literature reports, we selected the upregulated proteins which had been reported in metastasis but not in NSCLC metastasis or reported in cancer but not in metastasis or not reported at all in cancer. Finally, 9 potential NSCLC metastasis-associated proteins, including SMC5, PSIP1, KHSRP, VRK1, GTF21, PTRF, SNW1, VASP, DDX21, were selected to further validate the mass spectrometry results (Additional file 1: Figure S2A). In order to observe the expression of these proteins, the expression levels of candidate proteins were determined by qRT-PCR and Western blot in NCI-H1299, A549, NCI-H358 and NCI-H292 cell lines. The results showed that KHSRP, PSIP1 and VASP were higher expressed in the highly metastatic potential NSCLC cells (NCI-H1299 and A549) than in the low metastatic potential cells (NCI-H358 and NCI-H292) (Additional file 1: Figure S2B, C). To investigate whether KHSRP, PSIP1 and VASP were involved in human NSCLC metastasis progress, siRNA was used to knock down the expression of these genes in A549 cell line (Additional file 1: Figure S3A), and then the in vitro invasion and migration abilities of cells transfected with siRNAs were evaluated using transwell assays (Additional file 1: Figure S3B). We observed that knockdown of KHSRP genes could significantly inhibit the migratory and invasive abilities of NSCLC cells. These results suggested that the KHSRP might be crucial in the metastasis of NSCLC cells.

To investigate the functional roles of KHSRP in NSCLC cells, we first assessed the migration and invasion abilities of the four human lung cancer cell lines. The result showed that A549 and NCI-H1299 cell lines had higher potential to metastasis than NCI-H358 and NCI-H292 cell lines (Fig. 2a) [11]. We detected the expression levels of KHSRP in a panel of 4 NSCLC cell lines (A549, NCI-H1299, NCI-H838 and NCI-H292) by real-time PCR and Western blotting (Fig. 2b). The A549 cell line showed high endogenous KHSRP expression, while the NCI-H292 cell line showed low endogenous KHSRP expression, indicating that KHSRP expression was much higher in the highly metastatic NSCLC cell lines than in the NSCLC cell lines with low metastatic potential. Therefore, we selected A549 cells for KHSRP knockdown and NCI-H292 cells for KHSRP overexpression.

We established stable models of KHSRP knockdown in A549 cell lines and stable models of KHSRP overexpression in the NCI-H292 cell line (Fig. 2c). The CCK-8 assay showed that KHSRP knockdown substantially reduced the proliferation ability of A549 cells, whereas the overexpression of KHSRP significantly enhanced the proliferation ability of NCI-H292 cells (Fig. 2d). A wound-healing assay showed that the knockdown of KHSRP decreased A549 cell migration at the edge of exposed regions, whereas the overexpression of KHSRP enhanced NCI-H292 cell migration at the edge of exposed regions (Fig. 2e). Furthermore, the migration and Matrigel invasion assays in vitro were performed with A549 and NCI-H292 cells. The results showed that the knockdown of KHSRP in A549 cells significantly inhibited the cell migration and invasion potential, whereas the overexpression of KHSRP in NCI-H292 cells markedly promoted the cell migration (P < 0.001, Fig. 2f) and invasion potential (P < 0.001, Fig. 2g). These results indicated that the knockdown of KHSRP significantly inhibited the proliferation, migration, and invasion, whereas the overexpression of KHSRP significantly promoted the proliferation, migration, and invasion of NSCLC cells in vitro.

To assess whether KHSRP could impact tumorigenic capacity in vivo, a xenograft tumor mouse model was established by subcutaneously injecting A549-NC, A549-shKHSRP, NCI-H292-Vector and NCI-H292-KHSRP cells into the right dorsal flanks of nude mice. After 25 days, the tumor volume of the KHSRP interference group was smaller than that of the negative control group (P < 0.01), and the tumor volume of the KHSRP overexpression group was significantly larger than that of the control group (P < 0.01). At the end of the experiments, the xenograft tumors were isolated, and their weights were measured. The tumor weights of the KHSRP stable interference group were smaller than those of the negative control group (P < 0.05); in addition, the tumor weights of the overexpression group were higher than those of the control group (P < 0.01), and this was consistent with the results of the tumor volume change (Fig. 3a, b).

To evaluate whether KHSRP could promote tumor metastatic capacity in vivo, we injected A549-NC and A549-shKHSRP cells into the lateral tail veins of nude mice (eight mice per group). After 8 weeks, the in vivo imaging of animals showed that the intensity and area of luciferase expression in the interference group were significantly lower than those in the control group (Fig. 3c); at the same time, the knockdown of KHSRP significantly decreased the number of in vivo metastatic lung nodules compared to that in the negative control group (n = 8, P < 0.05). In contrast, the stable overexpression of KHSRP significantly increased the number of metastatic lung nodules (n = 10, P < 0.05). Metastatic nodules on the surfaces of mouse lungs were confirmed by H & E staining (Fig. 3d). Therefore, our data demonstrate that high KHSRP expression enhanced tumor growth and metastasis in vivo, which was consistent with our in vitro findings.

To elucidate the molecular mechanism by which KHSRP promotes metastasis in NSCLC cells, we next sought to identify proteins that interact with KHSRP in NSCLC cells. We performed tandem affinity purification (TAP) followed by mass spectrometry using Flag- and HA-tagged-KHSRP or control in HEK-293 T cells (Fig. 4a). Mass spectrometry identified 20 potential KHSRP-interacting proteins based on the criteria of unique peptides >2 and p < 0.05 with triple repeat appearances (Fig. 4b, c). The CO-IP experiment showed that the interaction between HNRNPC and KHSRP could be observed by immunoprecipitation in HEK-293 T cells with overexpression of KHSRP. On the contrary, the interaction between KHSRP and HNRNPC could also be observed by immunoprecipitation in HEK-293 T cells with overexpression of HNRNPC. The Co-IP experiments demonstrated that KHSRP and HNRNPC could coprecipitate with each other in HEK-293 T cells (Fig. 4d). Furthermore, the Co-IP experiments using endogenous proteins also demonstrated that KHSRP and HNRNPC could interact with each other (Fig. 4e). The immunofluorescence staining results showed that the KHSRP protein was colocalized and coexpressed with HNRNPC in the nucleus (Fig. 4f). Taken together, our results were consistent with the tandem affinity purification-mass spectrometry (TAP-MS) results. In order to further clarify the regulatory relationship between KHSRP and HNRNPC, we used Western blotting method to verify their expressions. Western blotting results showed that KHSRP could regulate the expression of HNRNPC, while HNRNPC could not regulate the expression of KHSRP, indicating that HNRNPC was a downstream protein of KHSRP (Fig. 4g). To verify whether KHSRP could stabilize the HNRNPC protein, we examined the effect of both KHSRP depletion and overexpression on the stability of endogenous HNRNPC protein in the presence of the protein synthesis inhibitor cycloheximide (CHX). The half-life of the KHSRP protein was not significantly different in A549 KHSRP knockdown cells and NCI-H292 KHSRP overexpression cells compared to that in the control cells (Fig. 4h). These data collectively suggest that KHSRP could specifically interact with HNRNPC to form a complex in the nucleus of NSCLC cells.

First, we detected the expression of HNRNPC in 4 cell lines using real-time PCR and Western blotting. The expression trend of HNRNPC was consistent with that of KHSRP (Fig. 5a). To investigate the functional roles of HNRNPC in lung cancer progression, the transient knockdown of HNRNPC using siRNAs was established, and the interference efficiencies of the three interference fragments were over 50%. Then, we established stable models of HNRNPC knockdown in the A549 and NCI-H1299 cell lines as well as stable models of HNRNPC overexpression in the NCI-H292 cell line (Fig. 5b). CCK-8 and clone formation assays showed that HNRNPC knockdown substantially reduced the proliferation ability of A549 and NCI-H1299 cells, whereas the overexpression of HNRNPC significantly enhanced the proliferation ability of NCI-H292 cells (Fig. 5c, d). Moreover, the migration and Matrigel invasion assays showed that the knockdown of HNRNPC in A549 and NCI-H1299 cells significantly inhibited the cell migration (P < 0.05, Fig. 5e) and invasion potential (P < 0.05, Fig. 5f), whereas the overexpression of KHSRP in NCI-H292 cells markedly increased the cell migration (P < 0.001, Fig. 5e) and invasion potential (P < 0.001, Fig. 5f). Similarly, wound-healing assays showed the same role for HNRNPC as that shown by the migration and Matrigel invasion assays (P < 0.05, Fig. 5g). Overall, the knockdown of HNRNPC significantly inhibited the proliferation, migration, and invasion of NSCLC cells, whereas the overexpression of HNRNPC markedly promoted the proliferation, migration, and invasion of NSCLC cells in vitro.

Currently, the identities of KHSRP-associated signaling molecules that are responsible for mediating human lung cancer cell metastasis are unclear. To further elucidate the molecular mechanism of KHSRP in regulating NSCLC metastasis, a human protein array was utilized to compare the relative levels of 18 key molecules related to tumor signaling pathways between A549 cells with or without KHSRP knockdown. The intensities of the spots on the array were quantified by Image J analysis. Spots with statistical significance were selected to further validate the expression differences by Western blotting. The results show that 9 key molecules were altered, including ERK1/2, STAT1, STAT3, Akt (Thr308), Akt (Ser473), S6, PRAS40, p38 MAPK and GSK3β (Fig. 6a), of which only STAT1 could be validated in A549 cells. The expression of key molecules in signaling pathway was detected by knockdown of KHSRP in A549 cells and knockdown of HNRNPC in NCI-H1299 cells. The results showed that the key molecules of IFN-α, JAK1, JAK2, p-STAT1 were changed obviously, but p-STAT3, p-Akt, p-Erk had no change. The results indicated that KHSRP could interact with HNRNPC and HNRNPC may promote the metastasis of lung cancer through IFN-α-JAK-p-STAT1 signaling pathway (Fig. 6b).

To determine whether the expression of KHSRP and HNRNPC in NSCLC is related to the prognosis of patients, we performed Western blotting analysis of 36 pairs of cancerous and noncancerous fresh tissues from NSCLC patients. The expression levels of KHSRP and HNRNPC in the 36 NSCLC tissue specimens were higher than those in adjacent noncancerous tissues (Additional file 1: Figure S4). Similar results were observed from the TCGA database (Additional file 1: Figure S5A). Moreover, high KHSRP or HNRNPC protein expression was associated with an advanced TNM stage (Fig. 7a). However, we detected the expression of KHSRP in 73 pairs of NSCLC and adjacent noncancerous lung tissues using real-time PCR. The mRNA expression levels of KHSRP in the 73 NSCLC tissue specimens were higher than those in adjacent noncancerous tissues, but there was not significantly associated with the TNM stage, lymph node and distant site metastasis (P > 0.05) (data not shown). In addition, the Oncomine database (www.​oncomine.​com) showed that KHSRP and HNRNPC were increased in particular types of solid tumors, including breast, colorectal and sarcoma (Additional file 1: Figure S5B). UALCAN (http://​ualcan.​path.​uab.​edu/​) showed different expression patterns of KHSRP and HNRNPC in a variety of tumors (Additional file 1: Figure S5C). Then, we performed immunohistochemistry analysis of 75 pairs of cancerous and noncancerous tissues from NSCLC patients. Statistically, the overall expression of KHSRP and HNRNPC was much higher in cancerous tissues than in the adjacent noncancerous tissues (P < 0.001). In 75 NSCLC specimens, high expression of KHSRP was found in 52 cases of NSCLC, and low expression of KHSRP was found in 23 cases of NSCLC; high expression of HNRNPC was found in 57 cases of NSCLC, and low expression of HNRNPC was found in 18 cases of NSCLC. The protein expression level of KHSRP was significantly associated with the TNM stage (P < 0.05) and lymph node and distant site metastasis (P < 0.05) but was not associated with sex, age or tumor size (Table 1). There was no correlation between HNRNPC expression and tumor stage or metastasis (P > 0.05) (Table 2, Fig. 7b, Additional file 1: Figure S6). Importantly, KHSRP protein expression was positively correlated with that of HNRNPC, suggesting a potential KHSRP-HNRNPC pathway in lung cancer tissues (R = 0.481, P < 0.01) (Fig. 7c). The ROC curves illustrated that the areas under the curve of the KHSRP- and HNRNPC-based predictions were 0.889 and 0.970, respectively, suggesting that they could both potentially be applied for the prediction of patient survival (Fig. 7d). To further explore the role of KHSRP and HNRNPC in predicting cancer prognosis, we analyzed the KHSRP and HNRNPC mRNA expression and the corresponding clinical data from the publicly available GEO database (GSE30219, n = 293; GSE102287, n = 34). Kaplan-Meier survival analysis showed that higher expression levels of both KHSRP and HNRNPC were strongly associated with a shorter survival time for lung cancer patients (Fig. 7e, Additional file 1: Figure S7A). Interestingly, the shortest survival time was observed in the group with the highest expression of both KHSRP and HNRNPC (Fig. 7f). In addition, the Kaplan-Meier Plotter database (www.​kmplot.​com) showed that the Kaplan-Meier survival analysis result was the same as the above result (Additional file 1: Figure S7B). Taken together, these findings indicate that KHSRP plays a critical role in lung cancer development and metastasis and might be a potential prognostic biomarker for this disease.