Cell fate decisions are driven through the integration of inductive signals and tissue-specific transcription factors (TFs), although the details on how this information converges in cis remain unclear. This study demonstrates that the five genetic components essential for cardiac specification in Drosophila, including the effectors of Wg and Dpp signaling, act as a collective unit to cooperatively regulate heart enhancer activity, both in vivo and in vitro. Their combinatorial binding does not require any specific motif orientation or spacing, suggesting an alternative mode of enhancer function whereby cooperative activity occurs with extensive motif flexibility. A fraction of enhancers co-occupied by cardiogenic TFs had unexpected activity in the neighboring visceral mesoderm but could be rendered active in heart through single-site mutations. Given that cardiac and visceral cells are both derived from the dorsal mesoderm, this 'dormant' TF binding signature may represent a molecular footprint of these cells' developmental lineage (Junion, 2012).

Dissecting transcriptional networks in the context of embryonic development is inherently difficult due to the multicellularity of the system and the fact that most essential developmental regulators have pleiotropic effects, acting in separate and sometimes interconnected networks. This study presents a comprehensive systematic dissection of the cis-regulatory properties leading to cardiac specification within the context of a developing embryo. The resulting compendium of TF binding signatures, in addition to extensive in vivo and in vitro analysis of enhancer activity, revealed a number of insights into the regulatory complexity of developmental programs (Junion, 2012).

Nkx (Tinman in Drosophila), GATA (Pannier in Drosophila), and T box factors (Doc in Drosophila) regulate each other's expression in both flies and mice, where they form a recursively wired transcriptional circuit that acts cooperatively at a genetic level to regulate heart development across a broad range of organisms. The data demonstrate that this cooperative regulation extends beyond the ability of these TFs to regulate each other's expression. All five cardiogenic TFs (including dTCF and pMad) converge as a collective unit on a very extensive set of mesodermal enhancer elements in vivo (Tin-bound regions) and also in vitro (in DmD8 cells). Importantly, this TF co-occupancy occurs in cis, rather than being mediated via crosslinking of DNA-looping interactions bringing together distant sites. Examining enhancer activity out of context, for example, in transgenic experiments and luciferase assays, revealed that the TF collective activity is preserved in situations in which these regions are removed from their native genomic 'looping' context (Junion, 2012).

In keeping with the conserved essential role of these factors for heart development, the integration of their activity at shared enhancer elements may also be conserved. Recent analyses of the mouse homologs of these TFs (with the exception of the inductive signals from Wg and Dpp signaling) in a cardiomyocyte cell line support this, revealing a signifcant overlap in their binding signatures (He, 2011; Schlesinger, 2011), although interestingly not in the collective 'all-or-none' fashion observed in Drosophila embryos. This difference may result from the partial overlap of the TFs examined, interspecies differences, or the inherent differences between the in vivo versus in vitro models. Examining enhancer output for a large number of regions indicates that this collective TF occupancy signature is generally predictive of enhancer activity in cardiac mesoderm or its neighboring cell population, the visceral mesoderm expression patterns that cannot be obtained from any one of these TFs alone (Junion, 2012).

There are currently two prevailing models of how enhancers function. The enhanceosome model suggests that TFs bind to enhancers in a cooperative manner directed by a specific arrangement of motifs, often having a very rigid motif grammar. An alternative, the billboard model, suggests that each TF (or submodule) is recruited independently via its own sequence motif, and therefore the motif spacing and relative orientation have little importance. The results of this study indicate that cardiogenic TFs are corecruited and activate enhancers in a cooperative manner, but this cooperativity occurs with little or no apparent motif grammar to such an extent that the motifs for some factors do not always need to be present. This is at odds with either the enhanceosome (cooperative binding; rigid grammar) or billboard (independent binding; little grammar) models and represents an alternative mode of enhancer activity, which was termed a 'TF collective' (cooperative binding; no grammar), and likely constitutes a common principle in other systems (Junion, 2012).

The data suggest that the TF collective operates via the cooperative recruitment of a large number of TFs (in this case, at least five), which is mediated by the presence of high-affinity TF motifs for a subset of factors initiating the recruitment of all TFs. The occupancy of any remaining factor(s) is most likely facilitated via protein-protein interactions or cooperativity at a higher level such as, for example, via the chromatin activators CBP/ p300, which interact with mammalian GATA and Mad homologs. This model allows for extensive motif turnover without any obvious effect on enhancer activity, consistent with what has been observed in vivo for the Drosophila spa enhancer and mouse heart enhancers (Junion, 2012).

Integrating the TF occupancy data for all seven major TFs involved in dorsal mesoderm specification (the five cardiogenic factors together with Biniou and Slp) revealed a very striking observation: the developmental history of cardiac cells is reflected in their TF occupancy patterns. Visceral mesoderm (VM) and cardiac mesoderm (CM) are both derived from precursor cells within the dorsal mesoderm. Once specified, these cell types express divergent sets of TFs: Slp, activated dTCF, Doc, and Pnr function in cardiac cells, whereas Biniou and Bagpipe are active in the VM. Despite these mutually exclusive expression patterns, the cardiogenic TFs are recruited to the same enhancers as VM TFs in the juxtaposed cardiac mesoderm. Moreover, dependent on the removal of a transcriptional repressor, these combined binding signatures have the capacity to drive expression in either cell type. This finding provides the exciting possibility that dormant TF occupancy could be used to trace the developmental origins of a cell lineage. It also explains why active repression in cis is required for correct lineage specification, which is a frequent observation from genetic studies. At the molecular level, it remains an open question why the VM-specific enhancers are occupied by the cardiac TF collective. It is hypothesized that this may occur through chromatin remodeling in the precursor cell population. An 'open' (accessible) chromatin state at these loci in dorsal mesoderm cells, which is most likely mediated or maintained by Tin binding prior to specification, could facilitate the occupancy of cell type-specific TFs in both CM and VM cells. Such early 'chromatin priming' of regulatory regions active at later stages has been observed during ES cell differentiation. The current data provide evidence that this also holds true for TF occupancy and not just chromatin marks. On a more speculative level, this developmental footprint of TF occupancy may reflect the evolutionary ancestry of these two organs. Visceral and cardiogenic tissues are derived from the splanchnic mesoderm in both flies and vertebrates. These complex VM-heart enhancers may represent evolutionary relics containing functional binding sites that reflect enhancer activity in an ancestral cell type (Junion, 2012).

Taken together, the collective TF occupancy on enhancers during dorsal mesoderm specification illustrates how the regulatory input of cooperative TFs is integrated in cis, in the absence of any strict motif grammar. This more flexible mode of cooperative cis regulation is expected to be present in many other complex developmental systems (Junion, 2012).