Published On: Thu, Apr 4th, 2019

Distinct gene-selective roles for a network of core promoter factors in Drosophila neural stem cell identity [RESEARCH ARTICLE]


Control of developmental gene expression is thought to rely on the combinatorial action of gene-specific transcription factors. The concerted action of activators and repressors is thought to ultimately converge on a highly conserved general transcription factor, TFIID. Based on studies in yeast and cultured mammalian cells, TFIID, composed of TBP and 13 TAFs, has been historically considered to be an essential yet passive player in gene regulation. However, the emergence in metazoans of both TBP and TAF paralogs, and of additional core promoter elements, has been proposed to contribute to the evolution of bilaterians by supporting more complex transcriptional programs (Duttke et al., 2014). Evidence from genetic and biochemical studies in a wide variety of model systems suggest that this diversity has indeed allowed multiple TAFs to take on cell- or tissue-specific functions. For example, the TAF9 paralog TAF9B regulates neuronal gene expression by associating with the SAGA/PCAF co-activator complex, whereas the TAF7 paralog TAF7L associates with TBP-related factor 2 (TRF2) to direct expression of a subset of post-meiotic genes during spermiogenesis (Herrera et al., 2014; Zhou et al., 2013). However, these examples are generally restricted to orphan TAFs, while prototypical TAFs are primarily present in TFIID and/or SAGA complexes.

In this study, using Drosophila NSCs as a model, we uncovered gene-selective functions for a subset of TAFs, NSC-TAFs, and some of these functions are shared with TRF2 whereas others are shared with their canonical binding partner, TBP. Our finding that NSC-TAFs did not regulate survival was unexpected, as deletion of Taf9 in chicken DT40 cells, of Taf4a in mouse embryos or TAF9 depletion in wing disc epithelial cells all result in increased apoptosis (Chen and Manley, 2000; Langer et al., 2016; Xie et al., 2014). However, it’s unclear whether TAFs are required for survival of embryonic stem cells (ESCs) as inducible depletion of TAF8 resulted in ESC cell death in one study whereas no cell death was detected upon knockdown of either TAF5 or TAF6 in a different study (Pijnappel et al., 2013; El-Saafin et al., 2018). In contrast to a report using murine ESCs (Pijnappel et al., 2013) in which a stable TAF5 knockdown ESC line prematurely differentiated without affecting the cell cycle, we show here that NSC-TAFs and TRF2 control stem cell identity in part, through direct regulation of the cell cycle. Depletion of NSC-TAFs or TRF2 by RNAi diminished the number of NSCs that incorporated the thymidine analog EdU, lowered the mitotic index, and rendered NSCs hypersensitive to cell cycle manipulation. Moreover, we identified DamID peaks at key cell cycle genes, including E2F1, CycE and string. We initially hypothesized that NSCs depleted for NSC-TAFs or TRF2 would both exhibit an extended G1 phase and be hypersensitive to manipulation of the G1/S transition. However, quantification of cell cycle phases using a FUCCI-based reporter showed that NSCs depleted for NSC-TAFs or TRF2 were in fact primarily in G2 and exhibited hypersensitivity to manipulation of both the G1/S and G2/M transitions. Intriguingly, a recent study showed that quiescent NSCs, which are known to extend a primary process, arrest primarily in G2 and are labeled by tribbles (trbl), which encodes a conserved pseudokinase (Chell and Brand, 2010; Otsuki and Brand, 2018). These phenotypes are remarkably similar to those observed upon depletion of NSC-TAFs or TRF2 and we note that the trbl locus is occupied by TBP, TRF2S and TAF5.

Because NSC-TAFs and TRF2 exhibit similar loss of function phenotypes and share at least 45 target genes in addition to the type I NSC marker ase, we proposed that they function together to direct expression of a subset of NSC-expressed genes (Fig. 7H). We tested this hypothesis by combining expression analysis using RNA-seq of FACS-purified NSCs and by determining the genomic binding sites for TBP, TRF2S and a representative NSC-TAF (TAF5), using Targeted DamID (TaDa). Our RNA-seq experiments revealed that many genes co-regulated by TBP and TAF9 are known or predicted to be important for NSC identity including the chromatin remodeler domino (Rust et al., 2018), the cell cycle genes E2F1, CycE (Fig. 7A) and string and the temporal identity factors Syp (Fig. 7B) and svp. However, the functional relevance of the NSC-TAF-TBP target genes remains to be determined. Similarly, TAF9-dependent genes that were unaffected upon TBP knockdown and that are bound by TAF5 are good candidates for mediating NSC-TAF’s function in self-renewal, such as the transcription factors Chinmo, Klumpfuss (Berger et al., 2012), HmgD and the polarity protein Insc (Fig. 7F). Lastly, while the function of the transcription factors (dati, hng3, HmgZ, mamo, Hr4, bi, jim) that are co-regulated by TRF2 and TAF9 and co-occupied by TRF2S and TAF5 are not well characterized, Dati (Fig. 7C), Jim and Mamo have recently been identified as important components of gene regulatory networks uncovered in a single-cell RNA-seq atlas of the adult Drosophila brain (Davie et al., 2018).

Because TRF2 neither binds the TATA box nor has sequence-specific DNA binding activity, how does the putative NSC-TAF-TRF2 complex recognize its target genes? Several lines of evidence suggest that DREF, which directly binds the DRE element, could be part of the DNA-targeting mechanism. First, depletion of DREF also results in fewer NSCs and smaller NSC lineages (Neumüller et al., 2011). Second, DREF is known to be part of the TRF2 complex (Hochheimer et al., 2002). Third, motif analyses with HOMER revealed that the DRE was over-represented in peaks identified by TaDa for all three fusion proteins. While a TRF2 complex larger than 500 kDa has been purified from embryonic nuclear extracts, none of the identified TRF2-binding proteins were TAFs (Hochheimer et al., 2002). However, a more recent identification of TRF2S-binding proteins from ovary lysates identified several TAFs (Andersen et al., 2017), raising the possibility that NSC-TAFs and TRF2 form a complex in vivo.

While depletion of TBP did not result in loss of Ase expression, nor diminish the number of NSCs, we found that TBP was essential for NSC cell cycle progression and directly regulated expression of many cell cycle genes. Intriguingly, knockdown of the RNA Pol II subunit RpII33 resulted in a more severe cell proliferation phenotype than TBP knockdown (Fig. 3A,B) but we note that a complication of the RNAi experiments is that TBP depletion reduced expression of UAS transgenes including the RNAi transgene itself whereas TRF2 or NSC-TAF depletion increased expression of UAS transgenes. However, given the lack of NSC loss in TBP-depleted brains, we were surprised to find that more than a third of NSC-expressed genes detected by our low-input RNA-seq were affected upon TBP knockdown. This is even more striking considering the fact that several NSC-TAFs (TAF1, TAF4, TAF8 and TAF9) were among the genes downregulated upon TBP depletion yet the NSC-TAF and TBP phenotypes are clearly different. Importantly, a study that sought to model the human neurological disorder SCA17 in Drosophila showed evidence that TBP is required for normal brain function, as removal of one copy of Tbp recapitulates some features of SCA17, such as impaired motility and age-dependent accumulation of vacuoles in the brain (Hsu et al., 2014).

We reasoned that the subset of TAFs that we identified is unique, as it is distinct from any previously described TAF complex. For example, NSC-TAFs partially overlap with a subset of TAFs that appear to co-regulate the size and composition of lipid droplets in the Drosophila fat body with TRF2 (Fan et al., 2017), yet that study did not identify lipid droplet functions for TAF2, TAF7 or TAF8, which are NSC-TAFs. Similarly, in mouse ESCs TFIID was proposed to be integral to the pluripotency circuitry (Pijnappel et al., 2013) yet depletion of two NSC-TAF orthologs (TAF7 or TAF8) did not affect ESCs identity.

Recent exome sequencing studies have produced compelling evidence that pathogenic variants in TAF1, TAF2, TAF8 and TAF13 are linked to intellectual disability and microcephaly (O’Rawe et al., 2015; Hellman-Aharony et al., 2013; El-Saafin et al., 2018; Tawamie et al., 2017). However, none of these variants have been modeled in vivo, in part due to a gap in our understanding of the function of TAFs during brain development. By demonstrating that TAFs are required for NSC cell cycle progression, NSC cell polarity, and act to prevent premature differentiation while not affecting survival, our work provides a foundation for future studies aimed at uncovering the causative variants in these disorders (Şentürk and Bellen, 2017).

About the Author

Leave a comment

XHTML: You can use these html tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>