6. HNF4α isoforms and their specific interactome
Despite the above findings, few studies have assessed the landscape of HNF4α isoforms’ specific interactome and their impact on gene expression. Only three studies have addressed these aspects [31, 32, 65]. The pioneering study performed by Daigo et al. used two independent HNF4α antibodies, one raised against the F domain and another against the A/B domain, to pulldown native endogenous HNF4α isoforms from the human hepatic HepG2 cell line. [65]. Semi-quantitative proteomic analysis revealed that HNF4α could interact with a wide variety of protein complexes, some associated with the DNA-dependent protein kinase catalytic subunit, histone acetyltransferase complexes, alternative splicing, chromatin remodeling, and nucleosome remodeling. These authors also identified the formation of heterodimers between HNF4α and HNF4γ and further explored the impact of this novel interaction on the transcriptional regulation of genes involved in liver metabolism [65]. With the use of large-scale approaches, two additional studies indicated that each of these HNF4α isoforms could influence a variety of genes with different patterns of expression and that for most tested isoforms, a positive modulatory effect was observed on their target genes with some exceptions. For example, the study from Ko et al . [32] reported that non-canonical isoforms of the P1b group negatively modulate gene expression. In contrast, Lambert et al. [31] showed that only the P2a-α9 isoform acted likewise. These discrepancies may result from the nature of these analyses, where Lambert et al . investigated the entire transcriptome in their cellular system. In contrast, Koet al. focused on a cluster of genes restricted to inflammation and immune response. Nevertheless, the study from Lambert et al.raised the possibility that the P1b subgroup of isoforms may act as negative regulators of the expression of several genes through interaction with other transcription factors [31]. Another significant difference between these two studies is the cellular system. The different or contradictory results reported could be related to the biological models and the overall design used in each study.
Lambert et al. chose to stably single-insert a doxycycline-inducible copy of each HNF4α isoform into an intestinal epithelial cell line (HCT116) devoid of endogenous HNF4α expression. This approach presented the advantage of comparing the transcriptome in the absence and presence of any given isoform in an endogenous range level that would be expected under this biological context. On the other hand, Ko et al . used a hepatocyte cell line (HuH7) that endogenously expressed HNF4α and transiently transfected each of the isoforms in different combinatory setups. This approach presents the disadvantage of producing possible supraphysiological levels of each of these isoforms and possible interference of endogenous HNF4α isoforms naturally produced in this cell line. However, both studies recognize a complex variability of isoforms in the N- and C-terminal regions and agree that this could be why the non-canonical isoforms of the P1b-α4 to α6 subgroup harbor different affinity with the DR1 motif, the classical and functional response element for HNF4α. Additionally, Ko et al. noted that canonical P1b group isoforms could bind and activate transcription, even when the binding sites in their DR1 motif 1 and 2 are mutated [32].
The study by Lambert et al. used powerful and complementary quantitative proteomic approaches to precisely identify the interactome of each HNF4α isoform in the context of HCT116 cells [31]. The authors observed that HNF4α isoforms interacted with a joint group of 69 proteins, for which some were previously identified and discussed above. For example, interactions with transcriptional activating complexes CBP/p300 and repressing complexes NCOR1 and 2 were observed (Figure 2). Other and newly identified large complexes were identified to interact with most individual HNF4α isoforms [31]. For example, the Switch/Sucrose-Nonfermentable (SWI/SNF) family of chromatin remodeling complexes, including the subunits AT-rich interaction domain (ARID)1A, ARID1B, ARID2, bromodomain containing 7 (BRD7), double PHD fingers 2 (DPF2), polybromo 1 (PBRM1), SWI/SNF related, matrix associated, actin dependent regulator of chromatin (SMARC)A4, SMARCC1, SMARCC2, SMARCD2, and SMARCE1 were all identified in this context (Figure 2). These complexes can activate or inhibit transcription in the presence of activators or repressors, and alterations in some of their units have been associated with different types of cancer and neurological disorders [66-68]. In addition, interactions between several HNF4α isoforms and subunits of the mediator complex, including mediator complex (MED)1, MED14, and MED15, were identified (Figure 2) [31]. Interestingly, Daigo et al. previously identified MED16 and MED24 as interactors of HNF4α [65], and some of these interactions were also observed to take place using GST-pulldown and co-immunoprecipitations assays [69, 70]. Given this complex’s importance in regulating the expression of many genes, mutations in some of these subunits are often associated with different diseases, including cancer [71, 72]. Interactions between HNF4α isoforms and enhancer of polycomb homolog 1 (EPC1), BRD8, and E1A binding protein P400 (EP400), all subunits of the histone H4/ nucleosome acetyltransferase complex (NuA4/Tip60), were also observed [31]. This complex involves transcriptional regulation, chromatin modification, cell migration and invasion, mitosis, and genomic instability (10). In addition, some of these subunits are also associated with colorectal cancer progression [73].
Finally, additional interactions involving some of the HNF4α P1 and or P2 isoforms with members of the nucleosome remodeling and deacetylase (NuRD) complex, including metastasis associated (MTA)1, MTA2, GATA zinc finger domain containing (GATAD)2A, and GATAD2B have also been observed (Figure 2) [31]. Likewise to SWI/SNF complexes, NuRD is involved in regulating transcriptional events, genome integrity, and cell cycle progression. Defects in the activity of the complex have been linked to defects in embryonic development, premature aging, oncogenesis, and cancer progression [74]. Given the importance of the complex in tumor diseases, some of the members of this complex, such as MTA1, have been proposed as potential therapeutic targets [75].