Epigenetically silenced apoptosis-associated tyrosine kinase (AATK) facilitates a decreased expression of Cyclin D1 and WEE1, phosphorylates TP53 and reduces cell proliferation in a kinase-dependent manner

Abstract

Silencing of the Apoptosis associated Tyrosine Kinase gene (AATK) has been described in cancer. In our study, we specifically investigated the epigenetic inactivation of AATK in pancreatic adenocarcinoma, lower grade glioma, lung, breast, head, and neck cancer. The resulting loss of AATK correlates with impaired patient survival. Inhibition of DNA methyltransferases (DNMTs) reactivated AATK in glioblastoma and pancreatic cancer. In contrast, epigenetic targeting via the CRISPR/dCas9 system with either EZH2 or DNMT3A inhibited the expression of AATK. Via large-scale kinomic profiling and kinase assays, we demonstrate that AATK acts a Ser/Thr kinase that phosphorylates TP53 at Ser366. Furthermore, whole transcriptome analyses and mass spectrometry associate AATK expression with the GO term 'regulation of cell proliferation'. The kinase activity of AATK in comparison to the kinase-dead mutant mediates a decreased expression of the key cell cycle regulators Cyclin D1 and WEE1. Moreover, growth suppression through AATK relies on its kinase activity. In conclusion, the Ser/Thr kinase AATK represses growth and phosphorylates TP53. Furthermore, expression of AATK was correlated with a better patient survival for different cancer entities. This data suggests that AATK acts as an epigenetically inactivated tumor suppressor gene.

Fig. 1. Structure of AATK and hypermethylation of its CpG island promoter in cancer.

A AATK is characterized by a tyrosine kinase domain (NCBI tool for conserved domain search) [76]. B AATK is located on chromosome 17q25.3. A 537 bp CpG island (CGI) (green) overlaps the transcription start site [40]. The frequency of the individual CpG sites is depicted (black lines). For the methylation analyses via combined bisulfite restriction analysis (CoBRA) and pyrosequencing, a 237 bp PCR product is generated (gray bar). Two TaqI restriction sites for CoBRA (red), three CpGs for pyrosequencing (purple), and two CpGs (cg26717786 and cg15665342) for Illimuna450k array (blue) are marked. For epigenetic editing, four gRNAs (#1–#4) covering the CGI were generated (light green). C A panel of 14 cancer cell lines with three head and neck, two breast, two thyroid, one sarcoma, one ovary, two brain cancer cell lines, and an in vitro methylated (ivm) sample were analyzed via CoBRA. PCR products were mock (−) and TaqI (+) digested.

Fig. 2. AATK expression is epigenetically downregulated in cancer.

A Via Infinium Human Methylation 450 BeadChip (ilmnhm450K array) and visualized by the SMART app [37], the methylation levels of AATK at two CpG sites within the CpG island (cg1566342 and cg26717786) in normal tissues (n = 711) compared to tumor samples (n = 8893) from the TCGA project are plotted as beta-value (1 = 100% methylation) and the p-values are calculated by one-way ANOVA. B Correlation analysis of methylation (beta-value) and expression of AATK (log2). Methylation of two CpG sites (cg26717786 and cg1566342) is plotted for AATK expression (ENST00000417379.5) in various tumor samples (n = 8411) from the TCGA project. The data obtained from the SMART app [37] for all tumor entities are plotted together. The R-values of linear regression and the p-values of correlation are given by regression statics via Excel. C Epigenetic editing of the AATK promoter via the CRISPR/dCas9 system. The relative AATK expression is normalized to GAPDH and is displayed after co-transfection of four gRNAs targeting the AATK promoter with effector proteins, p300 core domain, EZH2, DNMT3A, or empty vector in HEK293T cells for 48 h. D Correlation analysis of AATK (205986_at) and DNMT3A (222640_at) or EZH2 (203358_s_at) expression in normal tissues (GSE7307). E AATK expression throughout tumor progression in breast cancer, head and neck squamous cell carcinoma, glioma, lung cancer (left: lung adenocarcinoma, right: lung squamous cell carcinoma) [44]. F Overall survival probability in correlation with expression of AATK was analyzed via km plotter in breast and stage four HNSCC [34], in colon adenocarcinoma with unknown KRAS mutation status (TCGA ID: COAD with date: 2000-01-01), glioma (GSE16011) and lung cancer (GSE3141) via R2 Genomics Analysis and Visualization Platform [32] and in pancreatic adenocarcinoma via GEPIA [45].

Fig. 3. Exogenous AATK expression inhibits cell growth and is associated with negative regulation of cell proliferation.

A Using the TREx293 cell system, growth curves of a control clone pool and an AATK expressing clone pool with doxycycline (induced) and without (uninduced) were generated. Cells were grown for eight days and counted every two days (displayed as mean of three independent experiments with SD, unpaired two-sided t-test). B Using the R2 Genomics Analysis and Visualization Platform [31], gene sets (public gene categories) negatively correlated to AATK expression in various cancer cell lines (GSE363133), breast cancer (tcgaBRCA1097), colon cancer (tcgaCOAD286), head and neck cancer (tcgaHNSC520), glioma (tcgaLGG516), lung cancer (GSE3141), as well as pancreatic cancer (tcgaPAAD178) were determined and are shown with color code. Solid: p < 0.001, striped: p < 0.01; dotted: p < 0.05. C Binding partners of AATK were detected in HEK293T cells after transfection of EYFP-tagged AATK and EYFP empty and subsequent pulldown with GFP-Trap and mass spectrometry. Enrichment of interaction partners of control samples was determined from three replicate experiments using the limma-based analysis package autonomics [77]. D Overrepresentation analysis of enriched interaction partners of AATK-EYFP vs. EYFP was performed against categories from the Kyoto Encyclopedia of Genes and Genomes [45] and Gene Ontology [46]. The contrast AATK-EYFP to EYFP displays the (sub)ontology GObp in the protein groups data sets. Data is facetted by regulation direction in the context of contrast. Significances were determined with a Fisher Exact Test.

Fig. 4. Expression of CCND1 and WEE1 is negatively regulated by AATK.

A Whole transcriptome analysis via affymetrix array of a control clone pool, an AATK wt (wild type) expressing clone pool and an AATK KD (kinase dead) expressing clone pool was performed. After eliminating lowly expressed (linear values ≤100) genes, logFC (≤−0.6, 0.6≥) with respect to AATK wt expressing sample and control clone were determined. Thereafter, the top 500 deregulated genes were subjected to GO term analysis via PANTHER [78]. All displayed GO terms are significantly enriched (Fisher’s exact test with FDR correction). B A panel of cell cycle regulatory genes differentially expressed after induction of AATK wt expression is displayed as relative expression to induced control clone and normalized to GAPDH (SD, unpaired two-sided t-test). C Various cell lines (HEK, U343, U251) were transfected with EYFP-tagged AATK wt, AATK KD, or EYFP empty. Relative expression to EYFP and normalized to β-Actin with SD (unpaired two-sided t-test). D A knockdown of AATK was performed for 48 h in various cell lines (MCF-7, HEK, and Sk-Mel13). Relative expression of various cell cycle regulatory genes normalized to β-Actin with SD (unpaired two-sided t-test). E Pearson correlation of AATK and CCND1 or WEE1 expression in breast cancer, colon adenocarcinoma, glioma, lung cancer (adeno- and squamous cell carcinoma), and pancreatic adenocarcinoma was analyzed via GEPIA [45]. TPM = transcripts per million.

Fig. 5. The Ser/Thr kinase AATK.

A Peptide-based kinase activity assays were performed with an AATK wt and AATK KD expressing clone pool with doxycycline for 24 h. Computational upstream kinase analysis uncovered the basal kinomic activity profile Ser/Thr kinases. Due to the differential pattern of peptide phosphorylation, an increased activity of certain upstream kinases was predicted and these are represented by a volcano plot. B The significantly over-activated upstream kinases that compare to the activity profile of AATK wt were subjected to STRING analysis (left) [50] and the significantly (FDR = false discovery rate) represented Reactome pathways are displayed (right). Bubbles represent kinases, lines highlight interactions. C Lysates of AATK wt or KD expressing clone pools with overexpression of HIPK2-GFP or GFP empty were subjected to immunoblotting. D Input (left) and in vitro kinase assay (right) with immunoprecipitated TP53-GFP added to immunoprecipitated FLAG (empty), AATK wt-FLAG or AATK KD-FLAG to determine the phosphorylation status of TP53 at Ser366. E Cell viability was determined using MTS assay. 24 h post induction of AATK wt or AATK KD expression. MTS absorbance was measured at 490 nm with reference at 650 nm hourly for 4 h. F Via the TREx293 cell system, growth curves of a control clone pool, an AATK wt and an AATK KD expressing clone pool were generated. Cells were grown 4 days and counted every day (displayed as mean of three independent experiments with SD, unpaired two-sided t-test).

Fig. 6. Expression of TP53 target genes is positively correlated with AATK expression in cancer.

Pearson correlation of AATK and BAX, BBC3 (PUMA) or CDKN1A (p21) expression in breast cancer, colon adenocarcinoma, lung cancer (adeno- and squamous cell carcinoma) and pancreatic adenocarcinoma was analyzed via GEPIA [45] (TPM = transcripts per million).