Integrative pan-cancer analysis of MEK1 aberrations and the potential clinical implications

Alterations of mitogen-activated protein kinase kinase 1 (MEK1) are commonly associated with tumorigenesis, and MEK1 is thought to be a suitable targeted therapy for various cancers. However, abnormal MEK1 alterations and their relevant clinical implications are unknown. Our research comprehensively analyzed the MEK1 alteration spectrum and provided novel insight for targeted therapies. There were 7694 samples covering 32 types of cancer from The Cancer Genome Atlas (TCGA) database. They were used to conduct an integrative analysis of MEK1 expression, alterations, functional impacts and clinical significance. There was a dramatic difference in the alteration frequency and distribution and clinical implications in 32 types of cancer from the TCGA. Skin cutaneous melanoma (SKCM) has the most alterations and has therapeutic targets located in the protein kinase domain, and the growing expression of SKCM is positively related to patient prognosis. MEK1 expression in lung adenocarcinoma (LUAD), kidney renal papillary cell carcinoma (KIRP), esophageal carcinoma (ESCA) and liver hepatocellular carcinoma (LIHC) is decreased, which is associated with better prognosis, while MEK1 expression in thymoma (THYM), stomach adenocarcinoma (STAD), kidney renal clear cell carcinoma (KIRC), testicular germ cell tumors (TGCTs) and head and neck squamous cell carcinoma (HNSC) is increased, which is associated with better prognosis. Mesothelioma (MESO) has the second highest alterations but has no therapy targets. This study provided a great and detailed interpretation of MEK1 expression, alterations and clinical implications in 32 types of cancer and reminded us to fill the gap in MEK1 research from a new perspective.

www.nature.com/scientificreports/ resulting in a series of tumorigenesis and alloplasia. It was previously found that a diverse set of MEK1 gene mutations are associated with various somatic tumors such as melanoma, histiocytic neoplasms, colorectal cancer and lung cancer 10 . Accordingly, MEK1 inhibitors have been deemed to be an attractive strategy for the treatment of numerous cancers due to their crucial functions. Many MEK1 inhibitors have been naturally created and extensively used to treat a wide range of cancers due to their preclinical and clinical potentialities 11 . Méndez-Martínez et al. found that MEK1 inhibitors for the treatment of metastatic BRAF-mutant cutaneous melanoma and NRAS mutant melanoma were ground-breaking therapeutic regimens, respectively 12 . Kim et al. found that MEK1 inhibitors contributed to the treatment of NSCLC and confirmed their anticancer activity 13 . Additionally, MEK1 inhibitors also had a positive influence in advanced thyroid cancer (THCA), neurofibromatosis type 1 (NF1), BRAF-mutant pediatric low-grade gliomas, colon cancer and histiocytic neoplasms 10 . Thus, many tumors are related to MEK1, which should receive more attention and be explored deeper. Surprisingly, there are a lot of MEK1 inhibitors that have been approved for the treatment of various cancers by the Food and Drug Administration (FDA). For instance, the combination of highly selective inhibitors of MEK1/2, including trametinib, cobimetinib and binimetinib, and BRAF inhibitors have been served as the FDA-approved strategies for the treatment of BRAF V600E-mutant melanoma 14 . On April 10, 2020, selumetinib received FDA approval for pediatric patients with neurofibromatosis type 1, and they must be at least 2 years old, symptomatic and unable to surgery 15 . Moreover, Researches on other MEK1 inhibitors are also performing, such as PD-0325901, TAK-733 and Refametinib 11 .
However, prior studies have failed to identify the comprehensive and acknowledged profiling and significance of MEK1 due to a limited number of samples and/or limited types of cancer. We conducted this research using bioinformatics tools such as The TCGA and Kaplan-Meier analysis. Systematic mutation, copy number variants (CNVs) and expression were obtained across 32 different types of cancer from the TCGA. Then, we take advantage of Kaplan-Meier curves to analyze the clinical significance and promising insights according to the corresponding alterations.

Materials and methods
Data acquisition from various bioinformation databases. MEK1 expression in 53 normal tissues was acquired using genotype-tissue expression (GTEx), whose primary functions include the collection of the genetic impact on transcriptome and gene regulation mechanism, and MEK1 expression data were log10 transformed 16 . Then, cBioPortal was applied to obtain MEK1 mRNA expression data. cBioPortal is very useful for the exploration, visualization, and analysis of cancer genomics and clinical data 17 . We further utilized GEPIA, which is a practical and comprehensive website contributing to gene expression profiling and interactive analysis, to compare the transcriptional levels between 32 types of cancer and matched normal tissues 18 . Moreover, MEK1 protein expression was extracted from the first reverse phase protein array (RPPA) dataset containing more than 8000 samples of 32 types of cancer from TCGA 19 .
To determine MEK1 in terms of mutational and clinical significance, we utilized cBioPortral again to obtain these related data. The mutation data included indels, single-nucleotide variants (SNVs) and CNVs that contained amplification and deep deletion. The log ratio value of CNVs was set as follows: − 2/ − 1 = deletion, 0 = diploid, 1 = gain, and 2 = amplification. Then, the exploration of the clinical significance was mainly based on Kaplan-Meier curves. Kaplan-Meier curves are a method to predict the future and a convenient way to evaluate the effect of genes on patient survival 20,21 . Survival indexes such as overall survival (OS) and progression-free survival (PFS) were primarily used to analyze the association with MEK1 alterations. The 95% confidence intervals and the hazard ratios were plotted as forest plots. Data analysis. IBM SPSS Statistics v26.0.0.5 (https:// www. ibm. com/ cn-zh/ produ cts/ spss-stati stics) was used for data analysis. Student's t-tests, Cox regression analysis and linear regressions were appropriately applied. A p value less than 0.05 was considered significant. All cancer samples were statistically analyzed. A single sample at the genetic level was classified into two major categories: altered and unaltered. The Onco Query Language (OQL) was mainly applied.

MEK1 expression across 32 types of cancer.
Prior studies have substantiated that RAS/MAPK pathway overactivation promotes overexpression, leading to changes in MEK1 expression in various tumors 17 , but these results were far from sufficient. Studies on MEK1 expression remain a daunting challenge. Thus, we utilized GTEx to obtain an integrative analysis of MEK1 expression in pancancer and 53 normal tissues, and then we compared abnormal MEK1 expression with normal. Supplementary Figure S1A shows that MEK1 was differentially and markedly expressed in different normal tissues. The top three expression levels were as follows: cell EBV-transformed lymphocytes, brain-cerebellar hemisphere and brain-frontal cortex (BA9), respectively. Conversely, MEK1 was expressed at low levels in the kidney cortex, left ventricle of the heart and pancreas. Then, the MEK1 mRNA expression levels were profiled in pancancer. The spectrum of MEK1 mRNA expression was dramatically variable, indicating that MEK1 harbored some distinct traits to be expressed widespread, such as diffuse large B cell lymphoma (DLBC), acute myeloid leukemia (LAML) and uterine carcinosarcoma (UCS); or to be expressed narrowly, such as KIRC, THCA and pancreatic carcinoma (PAAD). The generation of this difference may be associated with multiple subtypes in one cancer and diverse gene phenotypes (Fig. 1A). To determine the actual changes in MEK1 expression in pancancer, we made a comparison between tumors and matched normal tissues. Obviously, these expression changes were markedly differential. As shown in the Supplementary Figure S1B  www.nature.com/scientificreports/ Because of the correlation between mRNA expression and protein expression 22 , we further analyzed the MEK1 protein expression in pancancer and these data obtained from the Cancer Proteome Atlas (TCPA) 19 . The results presented a striking similarity to MEK1 mRNA. Likewise, the profile of MEK1 protein expression was wide, and there were significant differences in different tumors (Fig. 1B). Furthermore, we observed that there was a high positive correlation between MEK1 mRNA and protein expression in pancancer (r = 0.6703, p < 0.0001). These observations implied that MEK1 gene expression was the key to driving and regulating protein expression and had a tremendous impact on tumor functions (Fig. 1C).

The specific mutation frequency and distribution of MEK1 in pancancer.
Increasing evidence has demonstrated that mutation plays a promising role in regulation of gene expression and activation in mammalian cells 23,24 . Hence, we next wanted to evaluated the frequency and distribution of MEK1 mutation in pan-cancer. The 1.4% (108/7867) MEK1 total mutation frequency for patients was obtained from cBioPortral, and the range in the samples was from 36 (CHOL) to 1084 (BRCA) ( Table 1). Across the 32 types of cancer, the SKCM mutation frequency was the highest (6.7%) and was more than twice as high as the second highest of CHOL (2.8%). The three highest frequencies were UCEC, CESC, and colorectal adenocarcinoma (COEADRE); and the frequencies were 2.6%, 2.4% and 2.2%, respectively. Conversely, TCTA and KIRC had low MEK1 mutation frequencies; and GBM, LGG, SARC, testicular germ cell tumor (TGCT), PRAD, PAAD and KIPR had almost no MEK1 mutation frequencies ( Fig. 2A). These results may have some limitations in the sample and the range, but we believe the limitations may be solved perfectly because the importance attached to MEK1 and the accumulation of samples gradually increased.
In the following work, we analyzed functional domains in which MEK1 mutations were located. The functional domains of MEK1 were classified as protein kinase domains (68-361 aa) and other domains using the Pfam database. The protein kinase domain is involved in cell metabolism though regulating cell signaling and activating other proteins by adding phosphate groups to protein substrates 19 . As Fig. 2B and Supplementary  Table S1 show, in general, the mutations in the protein kinase domain (68-361 aa) were twice as many as others across the 32 types of cancer. The cancers whose mutations were most commonly distributed in the protein kinase domain were SKCM, UCEC, LUAD, STAD, LUSC, HNSC and BRCA. Importantly, STAD and LUAC mutations were all in the protein kinase domain. The opposite was that the mutations in CESC and COADREAD were distributed primarily in other domains whose functions were learned little, and their roles and profiles must continue to be explored.
Based on the classification of MEK1 functional mutation domains, our research was extended to further explore the defined mutation sites. Regarding the functional effects on protein coding, 108 MEK1 mutations were divided into three categories: missense mutations, truncating mutations and in-frame mutations. The most frequent mutation site, far more than the mutation of any other site, was located at 124 aa belonging to the protein kinase domain (Supplementary Figure S2A). Furthermore, the mutation of MEK1 in P124S/L almost exclusively occurred in SKCM (Supplementary Figure S2B). This finding implied that P124S/L was a mutational hotspot, having great potential to become a new therapeutic target for clinical treatment. As elaborated by Diamond et al., cobimetinib was significantly effective for the treatment of SKCM harboring the P124L mutation 25 . However, there was no FDA approval, and the mechanism of P124S/L mutation leading to tumors was thought to be unknown. www.nature.com/scientificreports/ Next, we wanted to explore the biological significances of MEK1 mutations in tumorigenesis.108 MEK1 mutations were divided into three categories based on the oncogenic effects and predictive significance. They were unknown (48 mutations), oncogenic (33 mutations), and likely oncogenic (27 mutations). The majority of these mutations were unknown, indicating that it is a major challenge to determine their functions and that it is necessary for us to make greater efforts. Of note, the top four are SKCM (18/33), COADREAD (5/33), STAD (3/33) and LUAD (2/33) in oncogenic. Interestingly, in likely oncogenic, the top four are also SKCM (6/27), COADREAD (4/27), LUAD (4/27) and STAD (2/27) (Fig. 3A). Then, we analyzed the oncogenic effect separately according to the type of cancer. The oncogenic class accounted for the largest proportion in SKCM (18/30), COADREEAD (5/13) and STAD (3/7). The unknown class made up a large majority of UCEC (13/14), CESC (4/7), LUSC (4/5), HNSC (3/5), BRCA (4/5), ESCA (2/3) and BLCA (2/2) (Fig. 3B).

MEK1 CNVs across 32 types of cancer.
In addition to mutations, CNVs can also affect the expression and activation of cancer-associated genes, involving in the occurrence and development of many tumors 27,28 . MEK1 CNV frequency varied greatly in different tumors. There were almost amplification and deep deletion. The top three total MEK1 CVN frequencies were MESO (3.5%), KICH (1.5%) and SARC (1.2%). The bottom three were LUSC (0.2%), PRAD (0.2%) and LUAD (0.18%). Importantly, SARC and UCES had both amplifica- LGG and PAAD (Fig. 6A). Surprisingly, we observed that the MEK1 CNV frequency varied with the mutation sites. More than 50% of mutations in the protein kinase domain and other domains were accompanied by shallow MEK1 deletion and copy gain. The difference was that shallow deletion was dominant in the protein kinase domain while copy gain was dominant in other domains (Fig. 6B).

MEK1 alterations and patient survival.
The correlation between MEK1 mRNA expression and survival time was discussed to determine the clinical interpretation of MEK1 expression. According to the actual clinical Our results revealed that high MEK1 expression was a risk factor for patient OS in AUAD and LIHC. However, it was a protective factor for patient OS in melanoma, THYM, STAD, KIRP and ESCA (Fig. 7A,B). Next, the results of MEK1 expression and patient PFS are shown in Fig. 8A. High MEK1 expression was positively correlated with patient PFS in KIRC, STAD, TGCT and HNSC. Interestingly, a negative association between high expression and patient PFS was not observed.

Discussion
In this research, we analyzed MEK1 in depth across 32 types of cancer from the following aspects: MEK1 expression, alterations, functional impact and clinical significance. Melanoma accounted for the largest proportion of alterations, and mutations were more common. Furthermore, MEK1 expression increased significantly, and mutations were mainly distributed in the protein kinase domain in SKCM. Overall, mutations occurred more often in the protein kinase domain, which was more likely to have therapeutic targets. Other types of cancer with high MEK1 alterations included UCEC, CHOL and COADREAD. There were several similarities such as the alteration frequency was approximately 2.8%, mutations were dominant and MEK1 expression decreased slightly compared with normal tissues. Notably, MESO had the second highest alterations in which amplification was   This clinical trial showed that 3-year overall and relapse-free survival rate in combination-therapy group were 86% and 58%, which are both significantly higher than that in placebo group 31 . Patients with NF1 often have a high burden of genetic disorder owing to improved tumor predisposition syndrome and limited management options. This disease is a genetically heterogeneous disorder and leads to multiple manifestations and complications, such as pain, function damage, disfigurement and malignant transformation 32,33 . NF1 patients lack neurofibromin which leads to some dysregulated signaling molecules, including RAS, RAF, MEK1/2 and ERK 34 . Nowadays, MEK1 inhibitors can be used to treat the NF1 patients. Previous clinical researches indicated that treatment with selumetinib could obviously shrank the quantities of plexiform neurofibromas in patients with NF1. Moreover, selumetinib treatment could prevent the worsening of spinal canal distortion, reducing the need for surgical interventions and enhancing the clinical benefit 35 .
Histiocytic sarcoma is a rare, aggressive, and poorly understood hematopoietic neoplasm, and the majority of patients diagnosed live only 6 months 36 . One study found that histiocytic neoplasms might drive the RAF-MEK-ERK pathway and result in the overexpression of MEK1 37 . Naturally, trametinib was used to treat histiocytic neoplasms and preliminarily proved to be effective 38 . And a previous clinical research confirmed that   40 . Our findings showed that CORDREAD had high MEK1 expression, which was little associated with prognosis because these samples possibly had no MEK1 F53L mutation.
MEK1 is regulated by upstream factors and undergo a series of corresponding implications. Several seminal studies over the past decades have delineated upstream factors of MEK1 correlate with a greater risk of cancers, such as BRAF. It is well known that prooncogenic BRAF V600E-mutation could drive the malignant biological   42,43 . These abnormal upstream signalings could eventually participate in tumorigenesis by regulating the cellular MEK1.
In this article, we discussed the expression, mutation frequency and CNVs of MEK1 in pan-cancers. However, the interrelationship between these three issues remains to be expected. Wee et al. gave a comprehensive analysis to identify the relationship between CNVs and gene expression in pan-cancers. This group reported that numerous oncogenes with frequent copy number of gain (CNG) were concentrated in the RAS-MAPK pathways 44 . The neuroendocrine carcinoma (NEC) patients harboring BRAF V600E mutations had 46.7% mutation frequency, which was significantly related to high levels of RAF, MEK and advanced tumor stages 29 .With the development of detection methods and scientific understanding, the findings in this study would represent a novel therapeutic strategy for patients with MEK1-related cancers.

Conclusions
Overall, our study identified MEK1 characterizations across 32 TCGA types of cancer from four aspects including MEK1 expression, alterations, functional impact and clinical implications. Some alterations are closely related to tumorigenesis while others are strongly associated with targeted therapy. In addition, some kinds of cancer with high alteration frequency are involved in better prognosis, but others are just the opposite. Furthermore, some cancers with high expression or target sites can be treated with targeted drugs. Genomic profiling may provide significant insight into cancer treatment for using targeted drugs. Collectively, our results implicated the biological significances and applications of MEK1 in a variety of cancers. Future researches should be used to explore the potential regulation of MEK1 and evaluate the possible applications of MEK1 inhibitors in pan-cancer, which might provide promising therapeutic strategies and choices for cancer patients with MEK1 dysregulation.