"Straight or Split: Signals to Transcription".

Marcel van den Heuvel

MRC Functional Genetics Unit, Department of Human Anatomy and Genetics, University of Oxford, South Parks Road, Oxford OX1 3QX, UK

e-mail: marcel.vandenheuvel@human-anatomy.oxford.ac.uk

Since the discovery of substances in serum media that are able to drive cells into proliferation and/or differentiation, investigators have tried to understand how such signalling molecules can influence cells to change their behaviour. The complex nature of the responses to signals, and the equally complex signalling pathway leading to those responses, have made life difficult for the researcher. However, recent evidence obtained in genetic developmental systems indicates that a multiplicity of downstream events can be accomplished by regulation of the activity of just one transcription factor.

To general surprise, about two decades ago, biochemical studies [1-3] coincided with investigations in the regulation of embryonic development. Morphological studies on embryos predicted that substances existed that were able to drive extensive cell-fate changes during embryogenesis [4, 5]. The advent of molecular biology in genetic studies and sophisticated biochemistry made it possible to finally isolate the substances that were so magically delivering patterning in the embryo [6, 7]. Intriguingly, it appeared that the molecules defined in serum media during tissue-culture experiments were exactly the same as those delivering pattern in embryos [8]. It perhaps is not much of a surprise now, but at that time the perception that cancer research and developmental biology could overlap remained to be discovered.

Although today several classes of molecule are recognized as being involved in these processes, I shall concentrate here on discussing signalling proteins. During development, the expression of most of such secreted signalling proteins is tightly regulated, both spatially and temporally. Indeed, ectopic expression of these factors can lead to detrimental effects, indicating their strong influence on the cell's behaviour in vivo. In addition to the effects seen in early development, multiple later roles for such molecules have now also been indicated in, for example, neuronal differentiation [9]. However, in keeping with the tissue culture experiments, these factors can drive cells into proliferation or into differentiation depending on the cells' background and the presence of other signals. All experiments thus indicate a highly complex pattern of responses to these signals, and how this is accomplished is still not understood.

Information on how these signalling proteins lead to downstream effects has accumulated through studies in both cell-culture biochemistry and whole-animal gene manipulation. These experiments indicate that these
signals lead, via complex pathways in the responding cell, to multiple transcriptional read-outs. However, new evidence generated on the basis of the genetics available in Drosophila suggests that perhaps the situation is actually less complex than thought [10]. The work has focused on the signalling pathway downstream of the Hedgehog (hh) signal. As for many other signals, Hh and its pathway perform many roles during animal development and are also causative agents in human disease. Despite the apparent complexity of events regulated by Hh, the results obtained indicate that only a single transcription factor is required downstream and, importantly, this protein regulates both the repression and activation of targets. In the absence of signal, a repressor form of the zinc-finger transcription factor Cubitus interruptus (Ci) is produced that is required to repress both various target genes and hh expression itself  [11]. This repressor form is produced by cleavage of the protein whereby a transcriptional activator domain is removed, leaving just the zinc-finger DNA-binding domain and a repressor domain [12] (see Fig. 1).

Figure 1: Drawing showing the various forms of Cubitus interruptus protein (Ci) present as transcriptional regulators of hedgehog (hh) target genes. In the absence of Hh protein, Ci is cleaved and, in the process, the carboxy-terminal activator domain is removed. CiRepr is formed and represses transcription of target genes. In the presence of low concentrations of Hh, the cleavage is inhibited and full-length Ci protein, containing the activator domain, accumulates in the cytoplasm. It follows that transcription of target genes is activated. In the presence of high concentrations of Hh, another activating step drives full-length Ci into the nucleus, from where it is rapidly exported and degraded. An additional set of target genes is transcribed at high Hh concentration because of higher levels of Ci activity in the nucleus.

As such, this protein acts 'virtually' downstream of hh, to repress hh expression in cells not exposed to the signal. Thus the pathway ensures that the expression of the signal remains switched off appropriately. Hh, like other signalling proteins, is thought to act in a concentration-dependent manner. In the presence of low concentrations of Hh, Ci cleavage is inhibited, thus modulating the amount of repressor. The resulting full-length Ci protein is a transcriptional activator. However, most of this protein remains in the cytoplasm owing to the action of the Suppressor of fused protein, which prevents full-length Ci from entering the nucleus. In the cells exposed to the highest concentrations of Hh, Ci protein becomes difficult to detect [13]. This is thought to be the result of the rapid nuclear export of the protein. The virtual absence of Ci protein from both cytoplasm and nucleus in these cells is thus explained by maximal entry into the nucleus, after which it is rapidly exported and degraded. Several lines of evidence confirm that it is indeed the same DNA-binding domain that performs all the functions of Ci [14, 15]. Most of what is seen in the Hh pathway in flies is replicated in the vertebrate situation. However, the situation might be slightly more complex in vertebrates because of the presence of three homologues of ci, the Gli genes. Perhaps it is Gli2, which can act as both repressor and activator, that is most similar to Ci in flies, whereas Gli1 and Gli3 seem to act as activator and repressor, respectively [16, 17]. The dualistic behaviour of the genes encoding Ci/Gli is also evident from the phenotypes caused by mutations in the genes encoding individual Gli proteins [18]. Obviously the regulation of Ci and Gli proteins requires more components in the pathway than has been described above (see, for instance, refs [19, 20]).

Is Ci alone in being manipulated to inhibit its downstream targets in the absence of its stimulator? Perhaps not. Recent evidence suggests that in some well-known signalling pathways the control of expression of
downstream targets is regulated by one and the same transcription factor both positively and negatively. For the transcription factor that mediates the activity of the Wingless/Wnt family of signal molecules, LEF (lymphoid enhancer factor)/TCF (ternary complex factor)/Pangolin, a slightly different situation, but one that results in similar repressor activity, has been indicated. In the absence of signal, this transcription factor is associated with a general repressor protein [21]. Similarly, in the Notch/Delta signalling pathway the Suppressor of Hairless transcription factor can both repress and activate target genes [22]. Even in the well-known signal family TGF-b (transforming growth factor-b)/BMP (bone morphogenetic protein)/activin, the SMAD factors acting downstream can activate or be found associated with a repressor protein and thus repress transcription [23]. However, it is not clear whether this is taking place in the absence of signal or in the presence of specific signals.

We therefore come to the conclusion that both downstream transcriptional repression and activation could be mediated by the same factor. One now has to ask oneself if perhaps the complexity of the cellular responses to signalling proteins might be caused in part by the accumulation of two effects, loss of repression and activation, rather than by the action of multiple transcription factors. Perhaps this has not been asked before because of the experimental approaches that have been taken to analyse these pathways. On the one hand there was the pure genetic approach that often led to straightforward linear pathways; on the other hand was pure biochemical work, mostly done on cells in tissue culture, that often led to complex 'network of arrows' schemes. As usual the truth might lie in the middle and a combination of both approaches seems most useful.

However, despite perhaps a simpler transcriptional output downstream of signalling proteins, other cellular consequences than transcriptional are likely events in signalling pathways. One important issue is the cell's
architecture. In the Wingless/Wnt signalling pathway, a known component of the cytoskeleton is essential for nuclear activation. b-Catenin normally acts as a linker between the cytoskeleton and the extracellular matrix
downstream of the calcium-dependent cadherin molecules but is recruited by the Wingless/Wnt signalling pathway. In certain circumstances this might reduce its capability to function in the cytoskeletal background. Similarly a protein is required in the hh pathway, at least in flies, that binds the cytoskeleton. However, for Costal-2 the connection seems perhaps more passive; in addition, no clear vertebrate homologue that resembles costal-2 has been identified. Another factor linked to the cell's architecture is the influence of signalling pathways on the plasma membrane. Possibly the best-known examples of downstream effectors influencing the plasma membrane and associated proteins are Ptdins(4,5)P₂ and Ptdins(3,4,5)P₃. However, recent evidence suggests a role for a low-density lipoprotein receptor in the Wingless/Wnt signalling pathway [24]. Despite the finding of a ligand protein for these receptors [25], exactly how this can influence the pathway is for the moment unclear. It is interesting to note in this light that the Patched protein, the receptor of Hh, is localized to patches in the plasma membrane that are as yet uncharacterized.

Although we can perhaps understand more of the transcriptional output downstream of a single signalling protein, a cell in a dish or in an animal will always be exposed to a multiplicity of influences. In addition, its response to all of these is also dependent on what the cell (or its progenitor) has previously been exposed to. Even more complex backgrounds, where inhibitors of signals are present as well as crosstalk between signals, have all been shown. Together these influences will lead to a subtle but definite shift towards a particular goal, be it differentiation into a particular lineage or mitosis. It is therefore perhaps not surprising that there is not such a clear 'point of no return' as previously thought, and that what were thought to be fixed lineages of cells can be switched to participate in other differentiation pathways or to proliferate [26].

In all, the researcher into signalling pathways should not despair: only a limited number of signalling proteins seem to be present in the animal kingdom, and these and their pathways are highly conserved. If, as shown for Hh signalling, the complexity does not lie in its transcriptional activity but in its response, the novel technologies making use of whole-genome information and mining gene expression data will eventually lead to some kind of understanding of the signals and their actions.

References:

1. Massagué, J. Cell 49, 437-438 (1987).

2. Gospodarowicz, D., Ferrara, N., Schweigerer, L. & Neufeld, G. Endocr. Rev. 8, 95-114 (1987).

3. Nusse, R. Trends Genet. 2, 244-247 (1988).

4. Spemann, H. & Mangold, H. Archiv. Mikr. Ent. Mech. 100, 599-638 (1923).

5. Nieuwkoop, P. D. Wilh. Roux Arch. 162, 341-373 (1969).

6. Slack, J. M. W., Darlington, B. G., Heath, J. K. & Godsave, S. F. Nature 326, 197-200 (1987).

7. Smith, J. Development 99, 3-14 (1987).

8. Rijsewijk, F., Schuermann, M, Wagenaar, E., Parren, P., Weigel, D. & Nusse, R. Cell 50, 649-657 (1987).

9. Hall, A. C., Lucas, F. R. & Salinas, P. C. Cell 100, 525-535 (2000).

10. Methot, N. & Basler, K. Development 128, 733-742 (2001).

11. Dominguez, M., Brunner, M., Hafen, E. & Basler, K. Science 272, 1621-1625 (1996).

12. Aza, B. P., Ramirez, W. F., Laget, M. P., Schwartz, C. & Kornberg, T. B. Cell 89, 1043-1053 (1997).

13. Ohlmeyer, J. T. & Kalderon, D. Nature 396, 749-753 (1998).

14. Alexandre, C., Jacinto, A. & Ingham, P. W. Genes Dev. 10, 2003-2013 (1996).

15. Muller, B. & Basler, K. Development 127, 2999-3007 (2000).

16. Aza_Blanc, P., Lin, H. Y., Ruiz I Altaba, A. & Kornberg, T. B. Development 127, 4293-4301 (2000).

17. Wang, B., Fallon, J. F. & Beachy, P. A. Cell 100, 423-434 (2000).

18. Biesecker, L. G. Strike three for GLI3. Nature Genet. 17, 259-260 (1997).

19. van den Heuvel, M. Signal transduction pathways in development: Hedgehog proteins and their receptors. In Encyclopedia of Life Sciences (http://www.els.net) (2000).

20. Capdevila, J. Hedgehog signalling. In Encyclopedia of Life Sciences (http://www.els.net) (2000).

21. Cavallo, R. A. et al. Nature 395, 604-608 (1998).

22. Morel, V. & Schweisguth, F. Genes Dev. 14, 377-388 (2000).

23. Wotton, D., Lo, R. S., Lee, S. & Massagué, J. Cell 97, 29-39 (1999).

24. Wehrli, M. et al. Nature 407, 527-530 (2000).

25. Mao, B. et al. Nature 411, 321-325 (2001).

26. Krause, D. S. et al. Cell 105, 369-377 (2001). 

1. "Mated Models of Gene Regulation in Eukaryotes".

2. "Selective Control of DNA Helix Openings during Gene Regulation".