SEARCH

SEARCH BY CITATION

Keywords:

  • homeodomain;
  • Hox;
  • Meis;
  • PBC;
  • transcription

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References

Editor's suggested further reading in BioEssays ftz Evolution: Findings, hypotheses and speculations (response to DOI 10.1002/bies.201100019) Abstract

On the border of the homeotic function: Re-evaluating the controversial role of cofactor-recruiting motifs Abstract

Control of DNA replication: A new facet of Hox proteins? Abstract

Classification of sequence signatures: a guide to Hox protein function Abstract


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References

One of the most magnificent spectacles that nature offers is the tremendous diversity of living animals. Such morphological diversity results in a large part from the activity of particular classes of DNA-binding proteins that are collectively referred to as differential transcription factors (TFs) 1. These TFs act by controlling the expression of unique sets of downstream target genes during embryogenesis 2. Most TFs recognize short DNA-binding sites that are often also similar among members of the same family 3. Just how proteins with overlapping DNA-binding specificities carry out distinct functions in vivo is a fundamental and unresolved question in biology.

The Hox family of TFs is one of the best examples of this paradox. These TFs are organized in paralog groups whose numbers differ between species, but whose functions have been evolutionarily conserved for assigning distinct cell fates along the anterior-posterior (AP) axis of all bilaterian embryos 4. Despite their specific functions in vivo, classic and high throughput assays have established that Hox proteins preferentially recognize similar AT-rich nucleotide sequences as monomers in vitro 3. These converging observations strongly suggest that each Hox protein may require the presence of additional DNA-binding partners to recognize their correct set of target genes in vivo. To date, only one class of such cofactors, the PBC proteins, has been characterized. These proteins form cooperative DNA-binding complexes with Hox members of all except the most posterior paralog groups 5, 6. Accordingly, PBC proteins are required for several Hox-controlled events during embryonic development 7. Importantly, the interaction with PBC cofactors has been shown to be sufficient for specifying distinct DNA-binding preferences between different Hox members 8–10. Thus, the role of the PBC proteins reveals what seems like another paradox where transcriptional specificity results from interactions with a single cofactor that is common to several Hox members.

Here we discuss recent data showing that the Hox/PBC paradox is partly explained by the existence of flexible and potentially specific interaction modes. In particular, we reconsider this tangram-like molecular strategy in light of genome-wide binding properties of Hox proteins. We propose a model where flexible Hox/PBC interaction modes are differently interpreted depending on the instructive role of the target cis-regulatory sequences. This model illustrates how plasticity in protein-protein interactions could generally constitute a cost-effective molecular strategy for specifying the activity of transcription factors in vivo.

Role of the PBC-class cofactors for Hox specific activities

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References

The role of PBC proteins as direct Hox interacting partners was first established by electromobility shift assays showing that different Hox/PBC complexes could preferentially recognize different nucleotide sequences 11. It was later demonstrated that the identity of the two central nucleotides of the Hox/PBC binding site is critical for the specific regulation of Hox target enhancers in vivo 12, 13. Although this rule does not apply systematically 14, 15, these results highlighted the potentially strong instructive role of the DNA-binding site in dictating the choice of the interacting Hox/PBC complex.

More recent studies have revealed that the specific regulation of fork head (fkh) by Sex combs reduced (Scr) occurs through the recognition of a specific DNA structure that is itself dependent on the nucleotide sequence 16. Using a high throughput approach based on systematic evolution of ligands by exponential enrichment and massively parallel sequencing (SELEX-seq) to define preferential DNA-binding sites, it was shown that this mode of recognition also applied to all other Drosophila Hox/PBC complexes 10. This approach distinguished three Hox/PBC specificity groups that preferentially recognize distinct DNA minor groove topographies. Importantly, the role of paralog-specific residues in this recognition mode is dependent on the presence of the PBC cofactor. So far, PBC proteins are the only class of cofactors known to modulate Hox DNA-binding specificities.

Besides their role in Hox DNA-binding, PBC cofactors can also intervene in dictating a correct Hox regulatory potential. This mode of regulation has long been postulated in the “activity regulation” model, although this was only supported by one characterized regulatory example 17. Nevertheless, paralog-specific residues of Hox proteins are often oriented away from the DNA 18, suggesting that protein-protein interactions are important for Hox transcriptional specificity. Various Hox-PBC interaction modes could thus affect the availability of those paralog-specific residues for particular co-repressors and co-activators, potentially impinging on their regulatory potential. This mode of action is likely to apply to Ubx proteins of arthropods that display various levels of repressive activities depending on the interaction mode used to recruit the PBC cofactor 19. Similarly, the functional switch of Deformed (Dfd) into a Scr-like protein was shown to depend on mutations affecting the regulatory activity of a N-terminally located motif 20.

Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References

Hox proteins can recruit the PBC cofactors through flexible interaction modes

The Hox/PBC interaction mode was initially described as relying on a six-residue peptide called the hexapeptide (HX) and this interaction occurred at various distances upstream of the homeodomain (HD) 18. Because of its conservation in most Hox proteins, the HX has long been considered as a generic and obligatory protein interface for Hox/PBC complex formation. This assumption has recently been challenged by the discovery that other Hox protein motifs could also convey PBC-recruiting activities 21, 22. It was further observed that the formation of vertebrate and invertebrate Hox/PBC complexes could occur in the absence of the HX in vitro and in vivo 22. This unexpected dispensability of the HX motif for heterodimeric complex formation was often revealed in the presence of a third partner, the Meis protein. This protein is thought to indirectly participate in most of the Hox/PBC functions through the formation of PBC/Meis complexes that are initially required for the nuclear translocation of the PBC cofactor 23.

The influence of Meis on Hox/PBC complex formation was shown to vary according to the topology of the DNA-binding sites 22, illustrating the allosteric effect of the DNA on the overall Hox/PBC/Meis complex conformation. The role of Meis also highlights the fact that the association between Hox and PBC proteins is not rigid, being strongly influenced by surrounding protein partners. This could explain the ability of the Hox/PBC complexes to adapt their interaction modes to various promoter environments.

Flexibility in interaction modes also appears to be an appropriate molecular strategy for both distinguishing and diversifying the activity of different Hox/PBC complexes. Accordingly, it was found that the regulation of specific target genes of the Drosophila Scr, Ultrabithorax (Ubx) and AbdominalA (AbdA) proteins occurs through three different interaction modes with the PBC cofactor 24. In addition, the AbdA protein has been shown to use various combinations of at least three different protein motifs for accomplishing PBC-dependent functions in vivo 25. Interestingly, the capacity of AbdA to diversify its interaction mode with PBC coincides with the observation that this Hox protein regulates the larger set of target genes in the Drosophila embryo 26.

The genome-wide binding profile of Hox proteins contrasts with their specific transcriptional activities

Our comparison of ChIP and transcriptome analyses performed with the Drosophila Dfd and Ubx proteins 26, 27 demonstrates that a small proportion of target genes contain a corresponding binding peak of the Hox protein in their vicinity (Fig. 1A and B). This result was not expected given that the two Hox proteins bind in large excess with regard to the expected number of their downstream target genes (Fig. 1A and B). Even more surprisingly, we observed similar or even higher rates of Ubx binding in the proximity of putative Dfd target genes, and vice versa (Fig. 1A and B). This last observation not only applies to the full set of downstream genes, which are a mixture of direct and indirect regulatory events, but is also confirmed by analyses restricted to the specific putative direct Hox target genes (Fig. 1C). Thus, a substantial fraction of Hox genomic binding could be nonfunctional, highlighting the importance of post-DNA-binding mechanisms for controlling Hox activities in vivo.

thumbnail image

Figure 1. Comparison between genome-wide binding and transcriptome analyses of Dfd and Ubx in the Drosophila embryo. A: Comparison between the ChIP (chromatin immuno precipitation) of Dfd (in blue, 27) and the transcriptome (in green, 26) of Dfd (upper case) or Ubx (lower case) in the Drosophila embryo. B: Comparison between the ChIP of Ubx (in blue, 27) and the transcriptome (in green, 26) of Ubx (upper case) or Dfd (lower case) in the Drosophila embryo. A small proportion of Dfd (35) and Ubx (59) downstream genes show a corresponding binding peak of the Hox protein in their vicinity (by considering 10,000 bases upstream and downstream of the binding pic) 26, 27. These genes are considered as direct (or primary) target genes. C: Ubx binding in the vicinity of the 35 direct target genes of Dfd. A majority of these target genes (25) do not contain a corresponding Ubx binding peak. A significant proportion of Dfd target genes (6) are recognized by Ubx although they are not found in the Ubx transcriptome 26. In this context, the Ubx binding has no regulatory effect and is thus considered as nonfunctional. Finally, a small number of binding peaks occur in the proximity of genes that are indeed regulated by Ubx, in a similar (purple) or opposite (green) manner to Dfd. In those few cases, the genomic binding fragments are systematically different between the two Hox proteins 27, 30.

Download figure to PowerPoint

Integrating flexible Hox/PBC interaction modes to the nature of the target cis-regulatory sequence

In this last section we will speculate about how various interaction modes with PBC could differently influence the regulatory activity of Hox proteins depending on the nature of the target enhancer. In particular, we propose that the regulation of paralog-specific, semi-paralog-specific, and general Hox target genes may rely on the existence of instructive, semi-instructive and non-instructive DNA-binding sites for Hox/PBC complexes, respectively (Fig. 2).

thumbnail image

Figure 2. Coupling the tangram-like association mode of Hox/PBC complexes to the nature of the target cis-regulatory sequence. In this model, the influence of Hox/PBC interaction modes on the overall transcriptional activity varies according to the instructive, semi-instructive or non-instructive character of the DNA-binding sites. This balance is illustrated by the graded gray colors above and below the scheme. Various Hox/PBC/Meis interaction modes are schematized by the tangram-like model on the left, and by the corresponding colored Hox proteins associated with PBC and Meis on the different types of target enhancers. The Meis partner is systematically schematized to better illustrate the role of additional cofactors in Hox-PBC interaction flexibility. PBC and Meis are symbolized by dark and light gray-colored triangles, respectively. For instructive binding sites, selectivity can occur at the DNA-binding and/or post-DNA-binding level (role of the promoter environment), as discussed in the main text. Semi-instructive binding sites will also dictate a unique transcriptional output (activation or repression), but compared to the instructive ones, they are able to accommodate various interaction modes. These various interaction modes could potentially influence the level of the Hox regulatory activity (as symbolized by the different size of the colored arrows). Non-instructive binding sites tolerate both various interaction modes and transcriptional outputs. Examples of known target enhancers are provided in each case.

Download figure to PowerPoint

Instructive binding sites are specific to a unique Hox/PBC complex for correct regulation (Fig. 2). Selection can occur at the DNA-binding level, as observed for the CG11339 enhancer 14 of the Drosophila labial (Lab) protein. However, most of the specific Hox target enhancers characterized so far are recognized by at least two different Hox/PBC complexes in vitro 28. Still, the regulatory activity remains specific to a unique Hox/PBC complex when assessed in vivo. This is the case for the auto-regulatory element of Dfd, which allows the cooperative DNA-binding of Dfd/PBC and Scr/PBC complexes 20. Similarly, a chimeric Dfd protein that is able to bind the fkh element as efficiently as Scr with the PBC cofactor is still not able to activate the enhancer in vivo 20. These results suggest that instructive binding sites could also discriminate Hox/PBC complexes at the post-DNA binding level by tolerating a unique interaction mode consistent with the surrounding promoter environment. This mechanism could also explain the significant proportion of nonfunctional binding events of Hox proteins across the genome.

Semi-instructive binding sites tolerate more flexibility since they are present in enhancers regulated by several Hox proteins (Fig. 2). At these sites, different Hox/PBC complexes can assemble through different interaction modes, possibly with the help of the Meis protein. Although these different interaction modes could be responsible for various levels of activity (as discussed above), the final transcriptional output (repression or activation) will always be the same, illustrating the instructive role of the enhancer on the Hox/PBC regulatory potential. One example of semi-instructive binding sites is found in the enhancer of the Distalless (Dll) target gene 29, which can be repressed by different Hox proteins associated through at least three different interaction modes with the PBC cofactor. Other enhancers, such as those of the teashirt (tsh) target gene, also fulfil the definition of semi-instructive binding sites 21.

Finally, non-instructive binding sites tolerate the binding and regulatory activity of several, if not all, Hox/PBC complexes (Fig. 2). These are distinct from semi-instructive sites since they allow different transcriptional responses, depending on the Hox protein considered. To date, such sites exist only on non-natural enhancers, including the artificially modified fkh-consensus enhancer. This latter allows both activation and repression depending on the associated Hox/PBC/Meis complex 13.

Our model relies on few characterized target cis-regulatory sequences. Moreover, the extent to which Hox proteins are using PBC for their correct activity remains to be explored. Thus, in the future it will be critical to characterize supplementary target cis-regulatory sequences of Hox/PBC complexes. This could be achieved by combining transcriptome and ChIP analyses for different Hox proteins in wild type versus PBC or Meis mutant contexts.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References

How Hox proteins regulate their correct set of target genes in so many different developmental contexts is a puzzling issue that remains even more intriguing given that only one class of cofactors is so far known to modulate their DNA binding properties. Here we have discussed how the recruitment of this single class of common cofactors, the PBC proteins, could be sufficient for ensuring many different Hox functions in vivo. This role appears to rely on an amazingly high level of interaction flexibility with Hox proteins, allowing Hox/PBC complexes to adopt different conformations that are themselves differently interpreted depending on the instructive role of the target DNA-binding site. The overall conformation of Hox/PBC complexes is influenced by the positioning of Hox, PBC, and other cofactors such as the Meis proteins. This exemplifies how the architecture of the DNA-binding site of the enhancer could impact Hox/PBC activities. We postulate that in addition to better sensing the promoter environment, plasticity in Hox-PBC interaction modes allows Hox proteins to be part of many different enhancesomes without affecting their overall assembly.

Finally, the existence of various interaction modes with PBC could also ensure the robustness of Hox transcriptional programs during development and evolution. It remains to be determined whether Hox proteins could use flexible interaction modes with other types of cofactors and whether such a molecular strategy could apply to other families of differential TFs as well.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References

We thank J. Burden for proofreading the manuscript. Research in the author's laboratory is supported by the CNRS, Ecole Normale Supérieure (ENS) of Lyon, and grants from ARC (Association de Recherche contre le Cancer) and FRM (Fondation pour la Recherche Médicale).

The authors have declared no conflict of interest.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Role of the PBC-class cofactors for Hox specific activities
  5. Reconciling the role of the PBC cofactor with the widespread DNA binding of Hox proteins in vivo
  6. Conclusion
  7. Acknowledgements
  8. References