Gradients of fibroblast growth factors often induce cells to adopt different fates. A study in zebrafish embryos reveals another, unexpected role when the factors are trapped in small spaces by a special arrangement of cells. See Letter p.120
Building complex, multicellular organs during embryonic development is not just about making different cell types, it is about getting the right cells in the right place. For a cell to have some sense of where it is, it must integrate diffusible signals released by its neighbours. On page 120 of this issue, Durdu et al.1 provide evidence for a surprising new way by which diffusible signals such as fibroblast growth factors (FGFs) are controlled — by trapping them in small, closed extracellular spaces called microlumina, from which they have access to only a discrete collection of cells.
Exactly how signalling molecules provide enough spatial information to build complex organisms is still obscure, but studies of the principles of signalling generally split into two types. Those of the upstream part of signalling ask how the movement of diffusible molecules — sometimes called morphogens — is controlled to form appropriate spatial gradients2,3, for example by 'sticky' molecules in the extracellular matrix4. Studies of the downstream part ask how these spatial distributions are used by receiving cells to control cellular 'decisions'. A well-characterized example of the upstream part is the FGF family of secreted proteins5. These often form coherent spatial gradients within which different levels of signalling divide the responding cell population into sub-groups with different fates (Fig. 1a).
Durdu and colleagues took a fresh look at FGFs in their study of the development of the zebrafish lateral line — a sensory organ that lies along either side of all fishes, allowing them to sense vibrations in the water. In this developmental process, about 100 cells (called the lateral-line primordium) start near the head-end of the embryo and, over a two-day period, collectively migrate along the entire length of the developing body under the skin towards the tail6. During this journey, subgroups of cells cluster together within the primordium. These are called rosettes, because the cells adopt a radial arrangement in which each cell has an extension towards an apparent central common connection point (Fig. 1b). As the primordium migrates along the body, it drops off these rosettes one by one at regular intervals. Each rosette goes on to develop into a discrete mechanosensory organ.
The authors knew that manipulating FGFs can affect the spacing of dropped organs7, but not whether this was through a general effect on primordium velocity. They therefore quantified time-lapse movies of developing zebrafish embryos in which Fgf3 levels had either been upregulated by overexpression or repressed by drug inhibition. In both cases, they saw that the migratory velocity of the primordium was unaltered, which means that Fgf3 was affecting the drop-off frequency instead.
Having established a clear link between FGF signalling and rosette drop-off, Durdu et al. next explored where the signalling occurs. Fluorescence imaging of Fgf3 attached to green fluorescent protein suggested that it was localized into small, concentrated volumes at the apical centre of each rosette. Correlative microscopy (which combines fluorescence microscopy with electron microscopy) then revealed a striking cell-membrane arrangement: at the apical centre of each rosette was a microlumen formed by the cell membranes of all the cells of that rosette (Fig. 1c).
The researchers again used time-lapse imaging to show that the moment when Fgf3 starts to accumulate in a microlumen correlates with the time when that rosette begins to slow down in preparation for dropping out of the primordium. This pointed towards the intriguing possibility that FGF signalling is used on a very local basis to control the behaviour of just the 20 or so cells of one rosette. Durdu and co-workers went on to use all the advantages of the zebrafish system — ease of genetic modification and micromanipulation, and its suitability for high-quality time-lapse imaging — to test the idea.
They modified a single rosette so that one of its cells had increased Fgf3 levels (using either single-cell transplantation or a stochastic inducible genetic system), and observed that just this rosette was forced to drop out early from the primordium. On average, neither the rosettes before nor after it were prematurely dropped. To perform the opposite experiment, they punctured microlumina with a laser, thereby letting Fgf3 leak out. Satisfyingly, they observed the expected delay in rosette drop-off, again without affecting the previous or subsequent rosettes.
Several questions are not addressed in the study: for example, how the microlumina form in the first place; how levels of FGF expression are controlled; and, perhaps most directly relevant to the authors' findings, how FGF signalling accelerates rosette drop-off. But the strength of Durdu and colleagues' experiments is that single rosettes were manipulated in vivo, thus providing evidence that the microlumen can indeed restrict FGF signalling to the cells of just one rosette.
In this system, FGFs do not adopt one of their conventional upstream roles, in which a coherent swathe of different signalling levels splits a responding population of cells. Instead, the microlumen forces FGFs to take on a more downstream role: coordinating the response to a morphogenetic event, and ensuring that all cells of the rosette respond while none of the neighbours do. It is an intriguing case of multicellular architecture feeding back to control molecular signalling directly.
Because FGF concentrations accumulate only when the microlumen is topologically complete, the factors also provide a temporal checkpoint to the process. It thus unites a group of cells both temporally and spatially in a coordinated all-or-nothing response. This is an interesting, and slightly surprising, way to use a highly diffusible signalling molecule, but may turn out to be a widely employed mechanism in nature.