In time of revision: of Wingless and morphogens
Alfonso Martinez Arias
The recent publication of the important work of C. Alexandre, LA. Baena and JP. Vincent on the molecular requirements for Wingless signalling in Drosophila (http://www.nature.com/nature/journal/vaop/ncurrent/full/nature12879.html) offers an opportunity to consider the relationship between ideas and facts in modern developmental biology.
The work reports the surprising finding that fruit flies whose only source of Wingless is a membrane tethered form of the protein are viable and, except for a few minor defects in growth, morphologically normal. This is a very surprising result and, although earlier work had shown some hints in this direction during wing development, there is much that invites thought and reflection in this incisive and beautiful piece of work. However, to understand this, we need a small amount of background.
Wingless is the Drosophila homologue of Wnt-1, a member of a major family of signalling molecules that play significant roles in development and disease. Wnt signalling became to prominence because of work in Drosophila which, in the 90s, was deep in the hunt for Morphogens, molecules postulated to be able to instruct the patterning of fields of cells. There are two definitions of morphogens –and as critical results begin to emerge, the differences between the two becomes more and more important. The original one by Alan Turing refers to molecules that can trigger the generation form. A second definition, more popular and the one most widely known, emerged in the 70s and is associated with Lewis Wolpert notion of Positional Information (PI). According to this a Morphogen is seen as a diffusible substance that specifies pattern in a field of cells in a concentration dependent manner i.e. cells can ‘interpret’ different concentrations in different manners and activate different genes. It is this second definition that has served as a guide to interpret much of the molecular genetics of developmental processes in Drosophila. Wingless signals through an elaborate molecular device of which only one element needs to be considered in this discussion: like all Wnt proteins, Wingless effects its signalling activity through the nuclear activity of ß-catenin (Armadillo in Drosophila) i.e. Wingless can signal to other cells, but in the cells that see Wingless, what matters is the state of Armadillo/ß-catenin.
In the 90s, the finding that many of the molecules uncovered by the genetic analysis of pattern formation were diffusible led to identify many of them as “morphogens”, in the wolpertian sense. Naturally one of these was Wingless and much of the effort to make it a Morphogen focused on its function during the development of the adult, in particular within a structure called the wing imaginal disc, that will give rise to the wing and thorax of the adult Drosophila. For most of the development of this structure (over 40 of the 96 hours that lasts its development and patterning), Wingless is expressed in a thin stripe bisecting the growing disc. As ill defined (timewise) removal of Wingless generated defects in the growth and patterning of the wing, the notion emerged that it was the diffusion of Wingless from the stripe that was responsible for the pattern. A number of experiments were designed to test this hypothesis, including testing the effects of diffusible Wingless in contrast to membrane tethered Wingless and to those of its effector Armadillo/ß-catenin. Key in these experiments, as it is to any test of a wolpertian Morphogen, was the identification of direct response targets to the signalling activity of Wingless. Following a tradition, three were identified that during the patterning of the wing disc could be interpreted in this light; from high to low response thresholds: senseless, Distalless and vestigial. The experiments were interpreted to suggest (I am being careful in how I phrase this: ‘interpreted to suggest’ i.e. there was a fair amount of wishful thinking here) that there was a functional gradient of Wingless in the wing disc and that indeed high levels of Wingless triggered senseless expression, intermediate did Distalless and low elicited vestigial. Furthermore, membrane tethered Wingless could only signal to adjacent cells and Armadillo/ß-catenin could only elicit a response in the cells where it was expressed. And everybody, or almost everybody bought into it. In characteristic style Nature, Cell and Science broadcasted the news: Wingless was –and by the way still is- a Morphogen. However, looking at the data some of us had problems with these readings, the design of the experiments and the interpretation of the results. NCS was not interested and some of this questioning can be found in other journals where results have, on average, a longer shelf life and more information (see Development or Developmental Biology) as opposed to ‘cool experiments’ for morphogene enthusiasts, perpetuated in reviews. Part of the reason for these doubts was the existence of another view of the function of Wingless in the wing that, even though ignored, had a firmer base on the experimental results.
In the alternative view, Wingless did not (and DOES NOT) act as a Morphogen –in the wolpertian sense. Furthermore, a large number of experiments suggested that there was not much of a relationship between the long range diffusion of Wingless and the patterning of the wing. This work has been summarized elsewhere and in more general terms (for some details see http://amapress.gen.cam.ac.uk/?p=1191 and references therein). The gist of it is that
- There are different phases of Wingless expression in the wing and that there is a need to correlate specific functions with this different phases. Particularly during a crucial phase in the growth of the wing disc, all cells appears to express low levels of wingless.
- That removal of the stripe of Wingless expression had little or no effect on the growth of the wing.
- That Wingless acted at different times by creating some sort of a memory (implemented by Vestigial and Distalless) that would be pass on to the next stage.
The new work of Alexandre et al is a very rigorous confirmation of these observations in the wing but goes beyond them, showing in a most convincing manner, that there is no requirement for Wingless long range diffusion at all during the development of Drosophila and providing details and important hints of how Wingless works during wing development. It also shows that this might be a theme for the way Wingless works in Drosophila. Using elegant novel engineering technology they substitute the wildtype copy of Wingless for a membrane tethered form, and show that a fly with this genotype is viable. From the point of view of morphology it is a very good looking fly, though it has some small defects on growth, physiology and reproduction; but, for all practical purposes -certainly those that concern Drosophila pattern formation buffs- it is good. The defects should make people think and opened many interesting questions about Wnt signalling. Significantly, the authors go on to show that most of the growth of the wing disc is associated with ubiquitous low levels of Wingless expression in all the cells of the disc during the early/mid part of imaginal development (second and early third larval instars to the experts) and that therefore all cells have access to the levels of Wingless that they need, when they need it. This is consistent with previous suggestions that to understand Wingless in the wing disc one has to take into consideration its different patterns of expression and focus on early events. Its extension to the rest of development and other tissues of the fly is extremely important and invites much thought.
These observations are, indeed, surprising. However, rather than rejecting them and looking for small holes in the experiment (there is no perfect experiment in Biology) maybe we should simply use them to.reconsider our views of Wnt signalling. After all, none of the people who are raising caveats about this work –and there are a few- fluttered an eye brow when thinking about Wingless as a Morphogen, disregarding experiments to the contrary and boosting weak results in journals of wide readership, to favour an idea which was only an idea (see http://amapress.gen.cam.ac.uk/?p=1191) . The new work is not without issues but they are minor and do not have to do with the techniques, the membrane tetheres Wingless or the experimental design (all of them fine), but rather with what the results tell us about Wingless and Wnt signalling. Importantly: the work does not imply that Wingless does not diffuse in Drosophila; it does!. The work does not say that diffusible Wingless might not play a role, in Drosophila or other organisms; it might!. All that the work says is that in Drosophila, during normal development, there is not a major need for Wingless to diffuse a long range to perform its function. One of the reasons why it is important to take stock of this, is because there are few experiments (any?) in vertebrates that show a requirement for ling range diffusion of Wnt proteins in pattern formation. The accolade of Wingless as a Morphogen in vertebrates is a shallow extrapolation of the statements about Wingless in Drosophila, mostly in NCS but also, by mimicry, in other journals. Let us hope that this work leads to a more careful and rigorous analysis of the function of these molecules (NB this does not mean that reviewers should ask for endless lists of experiments, it simply says that we should be a bit more critical of what we have and a bit more thoughtful in the interpretation of the results).
One of the reasons for the use of Wingless in Drosophila might have to do with the rapid and robust development of the organism. Computation of signals take time and the wing disc (as all development of the fruit fly) is under extreme temporal constrains that have led it to evolve rapid mechanisms built around deeply interlocked gene regulatory networks (behold the early segmentation cascade). Thus, in evolutionary time it might be easier to reengineer the system, on the basis of the Gene Regulatory Networks, than to change the properties of the molecules (in the case in hand here, the diffusibility of Wingless). Thus the situation in the wing.
The twilight of the Morphogen?
The notion of a Morphogen is very tightly linked to pattern formation and signalling and in the light of the XXI century, it might be good to look at its roots and maybe return to the Turing version instead of the wolpertian PI version. Many people would be happier. While it is becoming increasingly clear that signalling molecules can elicit concentration dependent responses, it is not at all clear what the role of these responses are in vivo. We find correlations between concentrations and differential patterning but as one often gets in reviewers comments: correlation is not causation. In these considerations we should not forget the issue of time integrals and their relationship to spatial concentration. It seems that we need to divorce ourselves from simple naïve notions based on qualitative models. The Wnt version might just be one.
In the end rather than hanging on to notions for which, let us remember, there was precious little evidence for, we should look at the new results and think about what they tell us about the system rather than about a molecule. There is much in the work of Alexandre et al about this and we should take it beyond the simple ‘Morphogen or no Morphogen’. A twist to some famous words of Jean Rostand come to mind: theories pass, the wing remains.