Abstract

Background

   Lung tumors and normal lung tissues show large differences in
   epigenetic modification which can affect the chromosome structure and
   expression of genes. However, the epigenetic reprogramming in lung
   adenocarcinoma remains unclear.

Methods and Results

   With the bioinformatics analysis, we found that some activated
   super-enhancers (SEs) only appear in lung adenocarcinoma cells, and 781
   abnormal activated super-enhancers (AASEs) were found. Not only are the
   traditional oncogenes found to be activated by AASEs, such as MET and
   SLC2A1, but also some new genes were activated by AASEs, which probably
   contributes to the carcinogenic process in lung cancer. The enrichment
   analysis of the genes activated by AASEs shows that the glycolysis
   process and cell proliferation were enhanced and the apoptotic process
   was negatively regulated. Two AASEs were separately knockout by
   CRISPR/Cas9 in A549, PC-9, and H1299 cell lines and the expression of
   target genes decreased. The motif of CTCF, SMARCA1, SOX4, FOXM1, IRF3,
   IRF7, and STAT2 was enriched in AASEs, supporting that the chromosome
   structure changed and these transcription factors would be the master
   regulators on the formation of AASEs.

Conclusion

   This study provided comprehensive insight into the mechanisms of SEs,
   as well as a potential therapeutic target for lung cancer.

   Keywords: lung adenocarcinoma, epigenetic, super-enhancer,
   transcription factors

Introduction

   Lung cancer is the most common cause of cancer-related deaths
   worldwide, and lung adenocarcinoma is the most frequent pathological
   type in lung cancer.[32]^1^,[33]^2 It is important to investigate the
   underlying mechanisms of lung tumorigenesis and tumor progression. Lung
   tumor cells and normal lung cells have very different characteristics,
   including the histological features, expression of genes, and
   epigenetic modification of genome.[34]^3^,[35]^4 Super-Enhancers (SEs)
   are regions of genome comprising multiple individual enhancers which
   are the key regulatory elements to control tissue-specific
   transcription. SEs have been identified by locating genomic regions
   that are highly enriched in the H3K27ac ChIP-Seq signal.[36]^5 Compared
   to typical enhancers, SEs are larger, and exhibit a higher
   transcription factor and histone modification density. An array of
   transcription factor proteins bind on SEs and drive transcription of
   genes involved in cell identity.[37]^6^–[38]^8 In cancer cells,
   abnormal SEs enhance the expression of critical oncogenes such as
   CACNA1H, MYC, LMO1, and RARA.[39]^9^–[40]^12 Cancer cells generate new
   SEs nearoncogenes that are involved in tumor pathogenesis, and the
   absence of these SEs would reduce by at least 50% the survival rate of
   cancer cells.[41]^9 Accordingly, targeting the SEs and transcription
   factors on it will be the approaches to cancer diagnosis and clinical
   therapeutics, including small-molecule inhibitors against
   super-enhancers binding proteins and gene therapy strategies.[42]^13

   In the study of SEs in lung cancer, we found hundreds of emerging SEs
   in lung adenocarcinoma. Abnormal activation of these super-enhancers
   enhance the expression of oncogenes which would be the cause of lung
   adenocarcinoma pathogenesis. With the analysis and mining of
   bioinformation, we got all AASEs and their associated oncogenes in lung
   adenocarcinoma. The activated biological processes and pathways will
   show us the major characteristic and new therapeutic target of lung
   adenocarcinoma. Furthermore, we predicted the transcription factors
   binding on AASEs, which could be the pathogenic elements and help us
   understand pathogenesis in lung cancer.

Materials and Methods

Cell Culture

   The human A549, PC-9, and H1299 cell lines were obtained from ATCC
   (American Type Culture Collection). A549 were cultured in Dulbecco’s
   Modified Eagle’s Medium (DMEM, Gibco, Grand Island, NY, USA) containing
   10% fetal bovine serum (FBS, HyColne, UT, USA) at 37°C in a humidified
   atmosphere containing 5% CO[2]. PC-9 and H1299 were cultured in Roswell
   Park Memorial Institute (RPMI) 1640 (Gibco) containing 10% FBS at 37°C
   in a humidified atmosphere containing 5% CO[2].

Genome Editing

   For the knockout of AASEs in A549, two expression cassettes encoding
   the sgRNA sequences (sgRNA1-MET-AASE CTCCGTTAGAGGCGTGTCAG,
   sgRNA2-MET-AASE TACTCTTAGTATCCTGACTA; sgRNA1-EGFR-AASE
   TCATCTTCGATGACTCTGCT, sgRNA1-EGFR-AASE CTCACATGTGTCTATTCGGG) flanking
   the deletion region were cloned into a plasmid that expresses a
   codon-optimized version of Cas9. Lipofectamine 3000 reagent
   (Invitrogen, Carlsbad, CA, USA) was used to transfect the plasmid into
   the A549 cells. The transfected A549 cells were selected with puromycin
   for 2 days. The genotyping primers are as follows: MET-AASE:
   5ʹ-AAAGCAGTGGGCATATGGGA-3ʹ (forward), 5ʹ-GTGCTCTAATGCAGGTTGGG-3ʹ
   (reverse1), 5ʹ-ACCTACTTAGCCTACTCA-3ʹ (reverse2); EGFR-AASE:
   5ʹ-TGAGTAGGCTAAGTAGGT-3ʹ (forward), 5ʹ-AAGGCACAGCCCGTGAAAT-3ʹ
   (reverse1), 5ʹ-CACATTCATGCCACAGAA-3ʹ (reverse2).

Quantitative RT-PCR

   Total RNA was purified from A549 with RNeasy Mini Kit (QIAGEN), and
   total RNA (1 µg) was reversely transcribed into cDNA and analyzed by
   qPCR. The primers for β-Actin are 5ʹ-GCCAACACAGTGCTGTCT-3ʹ (forward)
   and 5ʹ-AGGAGCAATGATCTTGATCTT-3ʹ (reverse). The primers for MET are
   5ʹ-TGCACAGTTGGTCCTGCCATGA-3ʹ (forward) and 5ʹ-CAGCCATAGGACCGTATTTCGG-3ʹ
   (reverse). The primers for EGFR are 5ʹ-AACACCCTGGTCTGGAAGTACG-3ʹ
   (forward) and 5ʹ-TCGTTGGACAGCCTTCAAGACC-3ʹ (reverse). The levels of MET
   and EGFR mRNA were normalized to those of β-Actin. Changes in mRNA
   expression were calculated according to the 2^−ΔΔCT method (CT, cycle
   threshold).

Analysis of Gene Expression Profile

   The gene expression profiles of lung cancers and matched adjacent
   non-malignant lung were downloaded from the GEO database (accession:
   [43]GSE75037 and [44]GSE32863) ([45]https://www.ncbi.nlm.nih.gov/geo/).
   [46]GSE75037 and [47]GSE32863 were based on Illumina WG6-V3 expression
   arrays, and [48]GSE75037 included 83 lung adenocarcinomas and 83
   matched adjacent non-malignant lung tissues, [49]GSE32863 included 58
   lung adenocarcinomas and 58 matched adjacent non-malignant lung
   tissues. Differentially expressed genes between tumors and
   non-malignant lung tissues were identified using two-group comparisons,
   and P<0.05 was set as the cutoff criterion.

The Analysis of Super Enhancer and ChIP-Seq Data

   The super-enhancer regions of the lung cancer cell line (A549) and
   normal lung tissue (lung_30y). AASEs were found by calculating the
   super-enhancer regions which only occur in A549 compared to normal lung
   tissue though BEDTools.[50]^30 The super-enhancer corresponded ChIP-seq
   data of H3K27ac were obtained from the Encyclopedia of DNA Elements
   (ENCODE).[51]^31 The ChIP-Seq data were displayed using the UCSC Genome
   Browser ([52]http://genome.ucsc.edu/). All analyses were performed
   using human (build hg19, GRCh37) RefSeq annotations downloaded from the
   UCSC genome browser.

Functional and Pathway Enrichment Analysis

   Panther ([53]http://www.pantherdb.org/) was used to class the genes
   based on Gene Ontology (GO) molecular function. DAVID
   ([54]https://david.ncifcrf.gov/home.jsp, version 6.8) was used for the
   KEGG pathway and GO analysis for the activated genes. EASE Score, a
   modified Fisher Exact P-value, is used to measure the gene-enrichment
   in annotation terms. P-value<0.05 to be considered strongly enriched in
   the annotation categories.

Construction of Protein–Protein Interaction Network

   To further explore the relationships between genes activated by AASEs,
   the protein–protein interaction network was mapped by (STRING) database
   ([55]http://string-db.org/) with default parameter and visualized
   though Cytoscape software ([56]https://cytoscape.org/).

Analysis of Motif Enrichment

   The AME tool in MEME suite is used to identify motifs that are
   relatively enriched in the sequences of AASEs with default
   parameter.[57]^32 The motif database is HOCOMOCO Human (v11
   CORE).[58]^33

Survival Analysis

   Kaplan-Meier Plotter is a web tool that predicts the prognostic values
   of genes in lung cancer patients ([59]http://kmplot.com/analysis/). The
   patients with lung cancer were divided into two groups according to the
   particular gene expression level (high vs low expression). Based on
   these categories, overall survival (OS) analysis of the two patient
   groups was compared by the tool. The HR with 95% CIs and log-rank
   P-value were calculated and shown.

Results

Abnormal Activation of Super-Enhancers in Lung Adenocarcinoma

   Lung tissue and lung tumor share the same genome, their histone
   modifications such as H3K27ac H3K4me1 are distinct.[60]^4 In the
   research of epigenetic processes in lung cancer, we found some regions
   of chromosome were overactivated in lung cancer. These regions were
   largely marked with H3K27ac which is an active enhancer marker in
   genome and clusters of these enhancers formed new super-enhancers in
   lung cancers ([61]Figure 1A). The super-enhancers data of lung normal
   tissue and lung cancer cell lines were obtained from the SEdb
   database.[62]^14 After analysis, we find 781 emerging super-enhancer
   regions in lung cancer line (A549) ([63]Table S1). We can find the
   abnormally activated super-enhancers (AASEs) emerged in lung cancer and
   the nearby genes, such as ENO1, MET, EGFR, SLC2A1, and TKT, were
   over-expression in lung cancer tissues ([64]Figure 1B). These genes are
   widely recognized for their importance in cancer.

Figure 1.

   [65]Figure 1
   [66]Open in a new tab

   Abnormal activation of super-enhancers in lung cancer. (A) A
   hypothetical model of how SEs are abnormally activated in lung cancer
   and induce the overexpression of genes distantly (enhancer) or closely
   (promoter). (B) Examples of the abnormal activation of super enhancers
   and the involved genes are overexpressed in lung cancer tissues. The
   expression of genes came from [67]GSE32863 (58 lung adenocarcinoma and
   58 adjacent non-tumor lung tissues).

AASEs Cause Oncogenesis and Promote Tumor Development

   The closest active genes of SEs were obtained from the SEdb
   database.[68]^14 Combining the gene expression data of [69]GSE75037 and
   [70]GSE32863 which were tested from lung adenocarcinoma tumors and
   their matched histologically normal adjacent lung tissue samples, we
   screened 261 genes which were activated by AASEs and over-expressed in
   lung cancer ([71]Figure 2A and [72]B, [73]Table S2). The major
   molecular functions of these genes are binding and catalytic activity
   ([74]Figure 2C). From the results of the protein–protein interaction
   network, we could find the the HSPA4, HSPB1, EGFR, MAPK6, and MET were
   in the center of the network ([75]Figure 2D). From the results of the
   GO biological process and KEGG pathway enrichment analysis, we found
   these genes were significantly enriched in the glycolysis process,
   metabolic pathway, negative regulation of apoptotic process, cell
   proliferation, TGF-β signaling pathway, and VEGF signaling pathway
   ([76]Figure 3A and [77]B). The over-activated of these genes formed the
   major characteristic in lung cancer cells.

Figure 2.

   [78]Figure 2
   [79]Open in a new tab

   Genes activated by abnormal SEs in lung cancer. (A) Screening the genes
   activated by SE which is over-expressed in both two lung cancer
   databases ([80]GSE75037 and [81]GSE32863). (B) Heatmaps show the
   expression of screened genes in (A). (C) Molecular function of
   over-expressed genes activated by AASEs. (D) The protein–protein
   interaction network of the genes activated by AASEs.

Figure 3.

   [82]Figure 3
   [83]Open in a new tab

   AASEs enhance the carcinogenicity through up-regulation of the involved
   gene. (A) Enriched Gene Ontology (GO) biological process of
   over-expressed genes activated by AASEs. (B) Enriched KEGG pathway of
   over-expressed genes activated by abnormal AASEs.

Identify the Activation of AASEs in Lung Adenocarcinoma

   To determine whether AASEs could activate the transcription of nearby
   genes, we knockout the AASEs of MET and EGFR in A549, PC-9, and H1299
   cell lines though CRISPR/Cas9-based genome engineering ([84]Figure 4A).
   The pool of cells after transfection and screening were collected to
   identity the activation of AASEs near MET and EGFR. The results showed
   the deletion of AASEs decreased the expression of the MET and EGFR mRNA
   in A549, PC-9, and H1299 cell lines ([85]Figure 4B).

Figure 4.

   [86]Figure 4
   [87]Open in a new tab

   Identify the activation of AASEs though CRISPR/Cas9-based genome
   engineering. (A) Knockout the AASEs of MET and EGFR in A549 cell line
   (MET-E-KO and EGFR-E-KO). Scissors sketch shows the regions that
   knockout through CRISPR/Cas9 (left). PCR-based genotyping of MET-E-KO
   and EGFR-E-KO (right). (B) The RNA expressions of MET and EGFR were
   decreased when their AASEs were deleted (*P<0.05).

Master Transcription Factors Activate AASEs

   Master transcription factors could bind on the SEs and determine the
   cell type-specific.[88]^6 To find which master transcription factors
   bound on AASEs, we analyzed the motifs of transcription factors which
   were enriched in the sequences of AASEs ([89]Figure 5A). The most
   enriched transcription factor is CTCF, which is thought to be in
   regulating the 3D structure of chromatin and forming enhancer-promoter
   loops ([90]Figure 5B).[91]^15^,[92]^16 The other enriched transcription
   factors are SMARCA1, SOX4, FOXM1, IRF3, IRF7, and STAT2, and these
   factors function in regulating the transcription of genes. Consistent
   with the mRNA expression data of lung tumor tissues and adjacent
   non-tumor tissues, SMARCA1, SOX4, FOXM1, IRF3, IRF7, and STAT2 are
   overexpressed in lung cancer and correlated with the poor prognosis of
   lung cancer patients, but not CTCF ([93]Figure 5C and [94]D).

Figure 5.

   [95]Figure 5
   [96]Open in a new tab

   The master transcription factors contribute to the AASEs in lung
   cancer. (A) Motif of enriched transcription factors in AASEs. (B) The
   enrichment P-value of transcription factors in AASEs. (C) The
   overexpression of master transcription factors in lung cancer. (D)
   Log-rank (Mantel Cox) survival test of lung cancer patients based on
   the levels of IRF7, STAT2, FOXM1, IRF3, SOX4, and SMARCA1 (low
   expression n=963, high expression n=963).

Discussion

   Our results identified the AASEs that contribute to the pathogenesis in
   lung cancers. AASEs promote the initiation and progression of lung
   cancer by enhancing the expression of genes such as ENO1, MET, EGFR,
   SLC2A1, and TKT ([97]Figure 1B). New SEs appeared around these genes in
   lung cancer cells, but not in normal lung tissues. The genes we show in
   Figure represent the well-known oncogenes: ENO1 is up-regulated in
   several tumors including breast, lung, prostate, and pancreas, and
   plays an important role in the Warburg effect in cancer cells.[98]^17
   Another Warburg effect gene SLC2A1, also known as GLUT1, plays a key
   role in glucose uptake in many cell types including cancer cells. High
   expression of SLC2A1 will promote growth and proliferation in cancer
   cells.[99]^18 MET is a proto-oncogene and plays a role in cellular
   survival, embryogenesis, and cellular migration and invasion.[100]^19
   Overexpression of MET is also associated with multiple human
   cancers.[101]^20 EGFR (Epidermal growth factor receptor) has been
   reported to be implicated in the pathogenesis of many human
   malignancies and promote the metastasis of cancer.[102]^21 TKT is a
   thiamine-dependent enzyme involved in glycolysis and the pentose
   phosphate pathway which is essential in cancer energy
   metabolism.[103]^22 Furthermore, with the visualization of AASEs
   regions and ChIP-Seq signal data of H3K27ac, we could find a new
   activated mechanism of the oncogenes and more genes involved in cancer
   pathogenesis. We choose two AASEs and knockout them in three lung
   adenocarcinoma cell lines, the expression of target genes MET and EGFR
   fell by half. MET and EGFR were well-known oncogenes, and their AASEs
   regions will be the therapeutic target in lung cancer.

   To investigate the function and pathway affected by AASEs, enrichment
   analysis of the genes activated by AASEs was undertaken. The results
   showed the energy-related processes and pathways appeared frequently.
   Increased glucose uptake and aerobic glycolysis are the major
   characteristic in cancer cells.[104]^23 The energy metabolism
   associated genes such as ENO1, SLC2A1, ENO3, G6PD, and TKT were
   activated by AASEs and over-expressed in lung cancer cells. Meanwhile,
   cell proliferation was positively regulated and the apoptotic process
   was negatively regulated in lung cancer cells.

   In the results of enriched transcription factors on AASEs, we found
   CTCF to have the minimum P-value. CTCF (CCCTC-Binding Factor) is a
   multifunctional protein in genome regulation and gene
   expression.[105]^24^,[106]^25 CTCF can bind and remodel the
   three-dimensional structure of the genome which promotes the formation
   of long-range interaction of chromosome.[107]^26 A number of studies
   have demonstrated the co-localization of CTCF and cohesion on
   chromosomes, suggesting they can steady the chromatin loops between
   enhancers and promoters and also promote the binding of transcription
   factors at enhancers.[108]^27^–[109]^29 Therefore, the enrichment of
   CTCF on AASEs means large changes of chromosome structure happened on
   AASEs and help the formation of the connections between SEs and
   promoter of genes. The transcription factors such as IRF7, SOX4, STAT2,
   and FOXM1 will promote the activation of AASEs and regulate the
   transcription of downstream genes through long-range interaction. High
   expression of these transcription factors means poor survival rates in
   lung cancer patients. They can be biomarkers and therapeutic targets in
   lung cancer. However, further studies are required to find the
   mechanism that these transcription factors are activating AASEs by in
   lung cancer.

Acknowledgments