Dissecting GATA1 Protein Interactions in Normal and Malignant Human Erythroblasts

Acute erythroid leukemia (AEL) is characterized by uncontrolled accumulation of transformed erythroblasts. Previous analysis of murine and human AEL revealed aberrant regulation of the master regulator GATA1, which controls terminal erythroid differentiation in multi-protein complexes acting as activators or repressors of gene expression. Although most malignant erythroblasts constitutively express abundant GATA1 protein, terminal erythroid differentiation is impaired. Notably, overexpression of GATA1 significantly induced partial or complete terminal erythroid differentiation of the human AEL cell line K562 or immortalised HUDEP2 human erythroblasts, respectively. These observations led us to hypothesize that blocked terminal erythroid differentiation in AEL might be the consequence of titratable dose-dependent aberrant GATA1 protein interactions. We comparatively analysed nuclear extracts from three human AEL cell lines (F36P, K562, KMOE2) and primary cells from an AEL patient. In addition, we analysed HUDEP2 and primary human erythroblasts (hEBST) from healthy donors that retain the potential for complete in vitro terminal erythroid differentiation. We quantified protein expression using a tandem mass tag (TMT) based approach (n=3/cell type) and we compared putative GATA1 interactions by immunoprecipitation (IP) followed by liquid chromatography mass spectrometry (MS) (n=3/cell type). Quantitative proteomics identified 6774 commonly expressed proteins in AEL and "normal" erythroblasts with a high reproducibility (mean coefficients of variation <10%) for all six different cell types. Unsupervised hierarchical clustering displayed a clear separation of the AEL cells from the "normal erythroblasts" (hEBST, HUDEP2). 386 proteins were higher expressed in the AEL group (logFC>=2; q<0.05), whereas 623 were more abundant in normal erythroblasts (logFC>=2; q<0.05). IP-MS analysis of nuclear lysates from the AEL cell lines, the AEL primary sample, HUDEP2 and hEBST resulted in a matrix containing 1616 proteins from which 126 proteins seem to significantly differentially interact with GATA1. 54 proteins were more enriched in the AEL group, whereas 72 proteins were more enriched in "normal" erythroblasts (q<0.5). Principal component analysis (PCA) showed for all cell lines a similar clustering pattern, accounting for 24% and 32% of the variance. Pulled-down proteins in hEBST and HUDEP2 clustered together and were closer to F36P and KMOE, than LAM49 and K562. Notably, we found significant enrichment (validated by immunoblotting) of the SKI protooncogene in AEL cells (logFC=1.82; q=0.013), a finding which not only confirmed previous findings in murine AEL models (MEL cells, erythroblasts from Nsd1 ^-/-mice) but also speaks for the functionality of our approach. Similarly, the LRPPRC leucine-rich PPR-motif-containing protein overexpressed in several cancers, as well the lactate dehydrogenases A and B (LDHA, LDHB) were significantly enriched in malignant erythroblasts (logFC>2; q<0.05). Furthermore, the ZEB2 zinc finger E-box-binding homeobox 2 protein, was significantly enriched in AEL cells (logFC=2.02; q=0.005). In contrast, the hematopoietic master transcription factor Runt-related transcription factor 1 (RUNX1) (logFC=2.48; q=0.0018) as well as DNA binding protein Ikaros (IKZF1) (logFC=1.71; q=0.13) were significantly enriched (validated by immunoblotting) in HUDEP2 and hEBST. Moreover, the MCM6 DNA binding mini-chromosome maintenance complex component 6 critical for proper DNA replication was enriched in normal erythroblasts (logFC=1.99; q=0.0001). Interestingly, one of the most strongly enriched (and validated by immunoblotting) proteins in normal erythroblasts was the nuclear pore complex protein NUP155 (logFC=6.1; q=0.0000001). Integration of the quantitative proteomics and the IP-MS analysis identified 118 proteins differentially expressed and differentially pulled-down by GATA1-IP, of which 49 were enriched in malignant and 69 proteins in normal erythroblasts (q<0.5). This shows that we reproducibly identified proteins that are differentially associated with GATA1 which are also differentially expressed in AEL cells versus normal erythroblasts. A targeted CRISPR/Cas9 screen is under way to identify GATA1-interacting proteins responsible for impaired erythroid differentiation of AEL cells.