Integrative analysis of histone acetyltransferase KAT2A in human cancer

2.1 Human tissue source and ethical statement

This study was approved by the clinical trial ethical committee of the Shanghai Pulmonary Hospital, Tongji University School of Medicine. After the approval of the ethical committee, all methods and information collection in this study were performed according to relevant guidelines and regulations. This research complies with all relevant ethical principles. Samples were sent to Tongji University School of Medicine for further pathological analysis. All patient samples were collected following local regulations and after obtaining the informed consent approved by the institutional review board (IRB). This commitment to ethical guidelines ensures that the research respects the rights and well-being of the study participants. Details regarding the human tissues collected for this study are presented in Table 1.

RNA isolation (#74004, RNeasy Micro Kit, QIAGEN) and cDNA preparation (#QP057, SureScript™ First-Strand cDNA Synthesis Kit, GeneCopoeia) were conducted following the manufacturer’s instructions. The qPCR using PowerUp™ SYBR™ Green Master Mix (#A25742, Thermo Fisher, Carlsbad, CA, USA) was carried out on the Bio-Rad CFX96 system (#3600037, Bio-Rad Laboratories). Results were normalized by the GAPDH gene and the mRNA levels of KAT2A were calculated versus GAPDH (2 - Δ ⁢ Δ ⁢ C ⁢ t ). The primers used were: KAT2A Fwd (forward): GAGGTCATGCTGACCATCACTG, Rev (reverse): CAGTGAGTTGCCGATGACATGG; GAPDH Fwd: GTCTCCTCTGACTTCAACAGCG, Rev: ACCACCCTGTTGCTGTAGCCAA.

2.3 Western blot

Protein extraction, quantification, and Western blot analysis were conducted following the manufacturer’s instructions. Briefly, tissue protein was mechanically dissociated using Dounce Tissue Grinders (D8938, Sigma-Aldrich, USA) and disrupters (TissueLyser LT, Cat. No.: 85600, QIGEN) at 4 ∘ C. The homogenates were centrifuged at 15,000 rpm for 20 min at 4 ∘ C, and supernatants were removed and stored in aliquots at -80 ∘ C for downstream experiments. Total protein was quantified using the PierceTM BCA Protein Assay kit (#23225, Thermo Scientific, IL, USA). Equal amounts of total protein (50 μ g) were loaded into each protein lane. After electrophoresis, the gel was electrophoretically transferred to a polyvinylidene fluoride (PVDF) membrane at a constant current of 150 mA. Western blot analyses were performed using antibodies against KAT2A (ab153903, Abcam, rabbit, 1:1000) and GAPDH (#8884S, Cell Signaling Technology, rabbit, 1:1500). Blots were visualized using an enhanced chemiluminescence assay kit (#32106, Thermo Scientific, IL). Quantity One®1-D analysis software (Bio-Rad Laboratories, Inc) calculated the densitometric analysis from three independent experiments.

2.4 Immunohistochemistry (IHC)

Samples were fixed in formalin and embedded in paraffin. The staining process was performed following the manufacturer’s instructions. Briefly, after antigen retrieval, the sections were treated with 3% H 2 O 2 for 5 min and blocked with 3% BSA for 60 min at room temperature (RT). Sections were exposed to primary rabbit anti-KAT2A antibodies (1:300, ab217876, Abcam) overnight at 4 ∘ C, followed by secondary antibodies for 60 min at RT. Antibodies were labeled with horseradish peroxidase, and rabbit antibodies were labeled with alkaline phosphatase according to the instructions (ADI-950-100-0001, Enzo). Two investigators quantified the slides blindly.

2.5 Data source

UCSC Xena. The RNA sequencing (RNA-seq) expression profile and survival data regarding pan-cancer (33 cancer types) were downloaded from the University of California SANTA CRUZ database (UCSC Xena https://xenabrowser.net/) [22], as well as the clinical characteristics (e.g., sex, age, tumor stage, and survival time). Cancer names and abbreviations are presented in Table 2.

TISIDB database. TISIDB (http://cis.hku.hk/TISIDB /index.php) is an online website focused on the interactions between cancer and the immune system and contains multiple heterogeneous data types [23]. This study used TISIDB to construct the heatmap of KAT2A expression in different immune subtypes, the stages of different cancers from The Cancer Genome Atlas (TCGA), and the drugs targeting KAT2A.

cBioPortal. The cBioPortal (https://www.cbioportal. org/) is a widely used open-access website containing over 5,000 tumor samples from 20 cancer studies [24, 25]. The summarized results of KAT2A in the ‘Cancer Types Summary’ module were obtained for this study, and detailed KAT2A mutated information is presented in the diagram.

UALCAN. The UALCAN (http://ualcan.path.uab. edu/index.html) database is an interactive web portal for in-depth analyses of TCGA gene expression data [26]. Herein, the differences in the KAT2A expression between the gender of patients and cancer stages of individuals were compared using UALCAN.

STRING. The STRING database (http://string-db. org) is an online server that focuses on protein-protein interactions (PPIs) and functional protein networks [27]. The STRING database was used to explore KAT2A interaction networks.

Human protein atlas (HPA). The HPA (https://www. protei natlas.org/) is a database of proteins in human organs, tissues, and cells based on multiple omics approaches [28, 29].

HISTome2. The HISTome Infobase (http://www.actrec.gov.in/histome2/index.php) has information on histone proteins, modifiers for multiple organisms with drugs [30, 31], and was created by researchers from Advanced Center for Treatment, Research and Education in Cancer, Navi Mumbai. Potential HAT inhibitory drugs of KAT2A were downloaded from HISTome2.

Data from individual experiments are presented as means ± S.D., and differences were assessed by Student’s t-test, one-way or two-way analysis of variance (ANOVA) with Tukey post hoc test for multiple comparisons (GraphPad Prism Software Inc, San Diego, CA, USA). The Pearson correlation coefficient was used for pan-cancer and correlation analysis. R (version 4.0.2) was used for statistical analyses and image plots. Statistical significance was defined as 0.05, 0.01, 0.001.

3.1 Transcriptional level analysis of KAT2A among human cancers

To determine the KAT2A gene expression in different cancers, we analyzed 11,014 samples (9,675 primary tumors, 719 solid tissue normal, 394 metastatic tumors, 173 primary blood-derived cancers, 42 recurrent tumors, 10 additional new primary tumors, and one additional metastatic tumor) spanning 32 different human tumors (Tables 2 and 3). We explored KAT2A mRNA levels using the UCSC Xena database (http://xena.ucsc.edu/). KAT2A mRNA levels varied greatly among these cancer samples. CHOL and TGCT exhibited the highest KAT2A expression, while KICH presented a relatively low expression (Fig. 1A and B). Next, to examine the distribution of KAT2A transcripts in different sample types, we compared the expression of KAT2A in primary tumors, solid tissue normal, and metastatic groups. Using the available clinical information, we compared the primary tumors with their corresponding solid tissue normals in each of the 32 cancers. Since the number of normal samples was limited for some cancer types, we filtered the comparisons of differentially expressed genes (DEGs) using a cutoff value of $N>$ 50 and 0.000001 in the Student’s t -test. BRCA, KIRC, LIHC, LUAC, LUSC, PRAD, and THCA presented significantly different KAT2A expressions from solid tissue normal controls (Fig. 1C and D). Further, we selected BRCA, KIRC, LIHC, LUAC, LUSC, PRAD, and THCA and compared KAT2A levels for each sample using a Jitter plot (Fig. 1D-J). These results indicated that KAT2A expression significantly increased in multiple cancer types, including BRCA, KIRC, LIHC, LUAC, LUSC, PRAD, and THCA, compared to normal tissues.

OPEN IN VIEWER

Cancer incidence has a sexually dimorphic pattern, making the underlying mechanisms a crucial issue in cancer research and the development of precision medicine [32]. This sexual dimorphism might result from biological sex differences (e.g., hormone discrepancies and/or chromosomal homo/heterogeneity) [33, 34], as well as lifestyle and habit, which are also connected to the incidence of cancers [35]. Therefore, precision medicine based on gender is an important modality [36, 37]. Furthermore, the National Institutes of Health (NIH) emphasizes studying male and female vertebrate animals and humans. In this study, we investigated whether KAT2A expression has a sex-specific pattern. First, we conducted a side-by-side analysis of the KAT2A mRNA levels in males and females using samples of additional new primary tumor, additional metastatic tumor, metastatic tumor, primary blood-derived cancer, primary tumor, recurrent tumor, and solid tissue normal. However, we did not detect significant differences between males and females (Fig. 2A). Then, we evaluated the sex difference expression pattern of KAT2A in normal (Fig. 2B) and tumor tissues (Fig. 2C) separately and did not detect significant differences. To gain further insights into KAT2A expression, we selected the top 5 cancers according to the p -value (READ, p = 0.521; COAD, p = 0.528; ESCA, p = 0.540; MESO, p = 0.569; and THYM, p = 0.573) to confirm the accuracy of our analysis and compared KAT2A expression in males, females, and normal tissues using data from another database (http://ualcan.path.uab.edu/analysis.html). KAT2A expression increased in both males and females compared to normal tissues for READ (Normal-vs-Male, p = 1.624E-12; Normal-vs-Female, p = 1.624E-12) (Fig. 2D), COAD (Normal-vs-Male, p = 1.624E-12; Normal-vs-Female, p = 1.624E-12) (Fig. 2E), and ESCA (Normal-vs-Male, p = 1. 1.749E-12; Normal-vs-Female, p = 1.00729E-12) (Fig. 2F). However, KAT2A expression did not differ between males and females for READ ( p = 0.856; Fig. 2D), COAD ( p = 0.664; Fig. 2E), ESCA ( p = 0.947; Fig. 2F), MESO ( p = 0.232; Fig. 2G), and THYM ( p = 0.509; Fig. 2H). The results using data from UCSC Xena were consistent with the UALCAN database, despite the different sample numbers due to the varying sample resources. Hence, these results indicated that KAT2A gene expression did not have a sexually dimorphic pattern in normal and cancer samples.

OPEN IN VIEWER

OPEN IN VIEWER

OPEN IN VIEWER

To assess the clinical relevance of KAT2A expression in pan-cancer, we investigated its correlation with cancer pathological stages. KAT2A expression was significantly correlated with pathological stages for BLCA (Fig. 3A), BRCA (Fig. 3B), CESC (Fig. 3C), CHOL (Fig. 3D), COAD (Fig. 3E), ESCA (Fig. 3F), HNSC (Fig. 3J), KICH (Fig. 3H), KIRP (Fig. 3I), LIHC (Fig. 3G), LUAD (Fig. 3K), LUSC (Fig. 3L), READ (Fig. 3M), STAD (Fig. 3N), and THCA (Fig. 3O) ( 0.05 for Normal vs. all stages). However, KAT2A expression did not differ among stages in most tumors except for BLCA (Stage 2 vs. Stage 4, p = 0.0353) (Fig. 3B), HNSC (Stage 1 vs. Stage 2, p = 0.00464; Stage 1 vs. Stage 3, p = 0.000199; Stage 1 vs. Stage 4, p = 1. 556E-06) (Fig. 3J), KICH (Stage 1 vs. Stage 2, p = 0.00200; Stage 2 vs. Stage 3, p = 1. 921E-03) (Fig. 3H), KIRP (Stage1 vs. Stage 3, p = 0.00458), LIHC (Stage 1 vs. S tage3, p = 0.00261) (Fig. 3I), LUSC (Stage 1 vs. Stage 2, p = 1. 284E-03) (Fig. 3L), and READ (Stage 2 vs. Stage 4, p = 0.0467; Stage 3 vs. Stage 4, p = 4.061E-03) (Fig. 3M). Therefore, the expression of KAT2A was significantly associated with tumor stage in BLCA, HNSC, KICH, KIRP, LIHC, LUSC, and READ cancer samples.

3.4 PPI correlation analysis of KAT2A

KAT2A is a subunit of the HAT module, which belongs to two distinct macromolecular complexes: the Human SAGA complex [38] and ATAC [39]. Thus, KAT2A is involved in a network that correlates with pathological tumor development. Thus, we evaluated the connections among the genes in this network using the STRING online database. The PPI network presented 10 nodes (TADA3, CCDC101, TRRAP, SUPT3H, MYC, TADA2A, USP22, TADA2B, ATXN7, and TAF10) involved in the functional interactions with KAT2A (Fig. 4A). Moreover, eight genes – USP22 (Fig. 4B), CCDC101 (Fig. 4C), MYC (Fig. 4D), SUPT3H (Fig. 4E), TADA2A (Fig. 4F), TADA3 (Fig. 4G), TRRAP (Fig. 4I), and TAF10 (Fig. 4H) were positively correlated with KAT2A. The remaining two genes, TADA2B (Fig. 4J) and ATXN7 (Fig. 4K) were negatively correlated with KAT2A. The function of these KAT2A-related proteins is summarized in Table 4.

Genomic mutation is closely associated with tumorigenesis [40, 41]. Thus, to identify mutation sites in KAT2A genes, we assessed 10,967 samples from 10,953 patients. As the lollipop plot depicted (Fig. 5A, post-translational modification (PTM) sites were frequently found in the p300/CBP-associated factor (PCAF) domain (86-337), Acetyltransferase domain (547-628), and bromodomain (738-818). The PCAF, acetyltransferase, and bromodomain in KAT2A are highly conserved across species [7, 42]. The types and sites of KAT2A genetic alterations are presented in Fig. 5A. We detected eight phosphorylation sites, six acetylation sites, one ubiquitination site, two methylation sites, and three SUMOylation sites for all 837 amino acids. The mutation sites contained 114 missenses, 16 truncating, five inframe, and four fusion mutations; R783Q was the most frequent (Fig. 5A).

OPEN IN VIEWER

Next, we conducted a comparative analysis to determine the genetic alteration recurrence of the KAT2A gene in cancers. Then, we checked the genetic alterations of the KAT2A gene in cancer patients using the cBioPortal database. The frequency of KAT2A alteration significantly fluctuated across different cancers. The genetic alteration profiling of KAT2A showed that its mutation was one of the most important single factors for alteration in UCEC and SKCM. Also, we found the highest amplification frequencies of KAT2A in ESCA, STAD, PAAD, MESO, and PCPD (Fig. 5B).

Additionally, we analyzed the copy number alterations of the KAT2A gene among these cancers. We determined the highest amplification and the lowest deep deletion levels based on the batch normalized from Illumina HiSeq analysis, indicating a relatively rare frequency (Fig. 5C). Hence, various genetic alterations were implicated in KAT2A and might contribute to cancer development.

As a significant part of the complex microenvironment, the KAT2A-contained SAGA complex is significantly connected to the development and progression of various cancers [43, 44]. The quantity and activity status of tumor-infiltrating lymphocytes are important predictive criteria for cancer survival times [45, 46]. Cytotoxic CD8 + T cells (CTLs) are crucial in cancer initiation, progression, and metastasis [47, 48]. Herein, we explored the correlation between immune cell infiltration and KAT2A expression in various cancers based on CD8 + T cell levels using the TIMER database. We first compared KAT2A expression with the abundance of immunomodulators (Fig. 6A). KAT2A expression was significantly positively correlated with the infiltration of CD56-expressing cells (both bright and dim) in BLCA, CHOL, KIKC, KIRC, PAAD, PCPG, PRAD, READ, STAD, TGCT, and THCA. Furthermore, KAT2A expression was also negatively correlated with other infiltrated immune cells, such as cytotoxic CD8 + T cells (Fig. 6A). We selected BRCA, KIRC, LIHC, LUAC, LUSC, PRAD, and THCA to perform the correlation analysis between KAT2A expression and CD8 T cells abundances. KAT2A expression was negatively correlated with the abundances of active CD8 T cells in BRCA (rho = - 0.025), LIHC (rho = - 0.059), LUAD (rho = - 0.05), and LUSC (rho = - 0.152), and positively correlated with active CD8 + T cell abundances in KIRC (rho = 0.043), PRAD (rho = 0.035), and THCA (rho = 0.108) (Fig. 6B-H).

OPEN IN VIEWER

Our analysis using TCGA data showed that compared to the normal control tissues, BRCA, KIRC, LIHC, LUAC, LUSC, PRAD, and THCA tissues have significantly elevated KAT2A expression and that KAT2A expression is correlated with the infiltration of CD8 T cell abundances (Fig. 6A). Additionally, we explored the associations of KAT2A expression across immune and molecular subtypes (Fig. 7A). The top 5 KAT2A expression was PRAD, TGCT, STAD, BRCA, and LIHC. Then, we focused on these five cancer and assessed the expression of six immune and molecular subtypes. The main clinical and molecular characteristics were depicted as C1 (wound healing); C2 (IFN-gamma dominant); C3 (inflammatory); C4 (lymphocyte depleted); C5 (immunologically quiet); C6 (TGF-b dominant). KAT2A gene expression was similar in BRCA (Fig. 7B), PRAD (Fig. 7D), and STAD (Fig. 7E) within five immune and molecular subtypes. KAT2A expression decreased in C6 (LIHC) and C3/C4 (TGCT) in LIHC (Fig. 7C) and TGCT (Fig. 7F).

OPEN IN VIEWER

OPEN IN VIEWER

Furthermore, we assessed the associations of overall survival with KAT2A expression across human cancers. We determined the relationship between longer, shorter, and NS (no significant) overall survival rates with KAT2A expression among 30 cancer (Fig. 8A). Shorter overall survival rate was significantly correlated with elevated KAT2A expression in KIRC. Low KAT2A expression was significantly correlated with a longer overall survival rate in PAAD (Fig. 8A). We also analyzed the connections between overall survival rate and KAT2A expression in KIRC, PAAD, STAD, and LGG using Kaplan-Meier curves. A higher KAT2A expression indicates a shorter survival rate within 12 years ( p = 6.38E-11) in KIRC (Fig. 8B), while a higher KAT2A expression indicates longer survival in PAAD ( p = 0.0195) (Fig. 8C). However, KAT2A expression was not associated with the overall survival rate in STAD and LGG (Fig. 8D and E). These results indicated suggested that KAT2A expression was a great factor affecting the survival of KIRC and PAAD.

3.9 KAT2A expression in human normal and tumor tissues

Dysregulation of transcriptomes is closely associated with cancer, leading to alterations in specific genes or pathways that contribute to uncontrolled tumor growth [49]. Over 500 genes have been identified as strongly implicated in transforming normal cells into cancer cells [50]. However, tracking total mRNA levels is insufficient for the major feature of KAT2A expression and cancers. Translation and (or) translocation of mRNAs encoding proteins that govern cancer cell plasticity play a central role in tumor cell differentiation [51, 52]. Thus, to provide a comprehensive view of KAT2A expression and cancers, we examined the KAT2A protein and mRNA level patterns in human cancer tissues. KAT2A protein was mainly located intracellularly and expressed with low tissue specificity. To validate the KAT2A expression in these organs, we used colon, spinal cord, kidney, liver, pancreas, and stomach tumors and normal human tissue sections to analyze pathological KAT2A levels with IHC staining. In the tumor cores, colon, spinal cord, kidney, liver, pancreas, and stomach cancer cells and tumor stroma were positive for KAT2A (Fig. 9A, lower panel). Different tumor organs demonstrated a range of KAT2A intensity. The normal human colon, spinal cord, kidney, liver, pancreas, and stomach organs were also positive for KAT2A (Fig. 9A, upper panel). Next, we evaluated KAT2A levels among these organs. KAT2A mRNA levels greatly increased in tumor organs compared to normal ones by qPCR (Fig. 9B), 2-4 folds compared to their respective control tissues. Finally, we performed a Western blot assay with KAT2A and GAPDH antibodies and observed higher KAT2A protein levels in tumors than their respective control organs (Fig. 9C). Accordingly, the quantification data showed a similar higher KAT2A protein level using the GAPDH housekeeping gene for normalization (Fig. 9D). Therefore, KAT2A expression in tumors presented a comprehensive range compared to normal samples, consistent with the database analysis (Fig. 1).

OPEN IN VIEWER

Pathologically, dysregulation of epigenetic events can lead to cancer. Consequently, identifying drugs that can modify these epigenetic changes holds great clinical potential. As described, KAT2A-mediated lysine acetylation is a reversible post-translational modification involved in tumor genesis [53]. Small molecules or inhibitors that can modulate the HAT or its competence, such as HDACs (histone deacetylases), can be potential therapeutics for various diseases [54, 55]. Hence, to explore the potential mechanisms in which KAT2A participates in carcinogenesis, we used the TISIDB online tool to predict the drugs targeting KAT2A collected from the DrugBank database (Fig. 10A. DB01992 has a direct interaction with KAT2A and is targeted by other modules. DB01992 is a small molecule of Coenzyme A (Fig. 10B) known for its role in the synthesis and oxidation of fatty acids and pyruvate in the citric acid cycle. Furthermore, we searched the HISTome2 database for more information on HAT inhibitors. Gallic acid, Garcinol, Anacardic acid, Procyanidin, MB-3, CTK7A, Plumbagin, and Embelin were predicted to interact with KAT2A directly (Fig. 10C-J) [56, 57]. Overall, these results might accelerate drug understanding and future development.

OPEN IN VIEWER