The zebra®sh Nodal signal Squint functions as a morphogen

Secreted morphogens induce distinct cellular responses in a concentration-dependent manner and act directly at a distance. The existence of morphogens during mesoderm induction and patterning in vertebrates has been highly controversial, and it remains unknown whether endogenous mesoderm inducers act directly as morphogens, function locally or act through relay mechanisms. Here we test the morphogen properties of Cyclops and SquintÐtwo Nodal-related transforming growth factor-b signals required for mesoderm formation and patterning in zebra®sh. Whereas different levels of both Squint and Cyclops can induce different downstream genes, we ®nd that only Squint can function directly at a distance. These results indicate that Squint acts as a secreted morphogen that does not require a relay mechanism. The level of Nodal signalling is a key factor in determining the cell fate induced in responding cells. We ®rst tested whether Cyclops (Cyc) and Squint (Sqt) can act on responding cells at a distance, and whether the activation of different downstream genes is dependent on concentration and distance. To provide a local source of Sqt or Cyc, sqt or cyc RNA was injected, together with the lineage tracer ̄uorescein-dextran, into a single cell of 128±256-cell zebra®sh (Danio rerio) embryos (Fig. 1a). The injected embryos were ®xed 3 h later before the onset of gastrulation (50% epiboly), and analysed by in situ RNA hybridization with antisense probes for no tail/Brachyury (ntl), goosecoid (gsc), bhikhari (bik) and ̄oating head ( ̄h)Ðknown targets of Nodal signalling. We found that ̄h, ntl and bik were expressed by cells 4±6 ( ̄h) and 6±8 (ntl, bik) cell diameters away from Sqt-producing cells (Fig. 1c, e, m). In contrast, gsc was detected only in cells expressing sqt and their immediate neighbours (Fig. 1d). This pattern recapitulates the expression of these genes in the zebra®sh organizer, where high levels of Nodal signalling are required to activate gsc, and lower levels are suf®cient to induce expression of ̄h, ntl and bik. When we reduced the levels of sqt tenfold, ntl was induced only close to the source and gsc was not induced (Fig. 1f, g). Expression of different levels of cyc resulted in a similar concentration-dependent induction of ntl and gsc. High levels induce the expression of both genes, whereas lower levels induce only ntl (Fig. 1i, j; and data not shown); unexpectedly, however, we never observed ntl or bik expression beyond two cell diameters from cyc-expressing cells, although the level of cyc was high enough to induce gsc expression (Fig. 1i, j, l). Notably, increasing the injected cyc RNA to 60 pg did not change the ectopic ntl domain, but induced ectopic gsc expression in all cells injected with RNA (data not shown). This indicates that more active Cyc protein is made, although western analysis with a haemagglutinin A (HA)-tagged form of Cyc revealed no signi®cant difference in HA±Cyc protein production on injection of 6 pg or 60 pg (data not shown). This result indicates that post-transcriptional mechanisms such as translational ef®ciency or protein stability might contribute to the range of action of cyc. The expression of bik in cyc, sqt and cyc;sqt double mutants is consistent with different ranges for endogenous Cyc and Sqt (Fig. 1o±s). In cyc mutants, Sqt is suf®cient to activate bik 6±8 cells from the margin (Fig. 1q), whereas in sqt mutants, Cyc induces bik only locally (Fig. 1p). The amino-terminal prodomain of transforming growth factor (TGF)-b signals is thought to interact with extracellular matrix and so may be responsible for the different range of action of Cyc and Sqt. We found, however, that the prodomains do not change the range of chimaeric Cyc±Sqt and Sqt±Cyc proteins (Fig. 1h, k), indicating that differences in the