New tricks for DIX

The 85-residue DIX protein domain is rather mysterious. Although it's present in a family of eukaryotic signalling proteins, including dishevelled (Dvl), its structure and function have remained unclear. In Nature, however, Michael Overduin and colleagues now reveal that DIX has a role in segregating Dvl into two different intracellular pools, and that, by doing so, it controls a critical divergence point in the Wnt signalling pathway.

Dvl signals downstream of Wnt receptors in the Wnt signalling pathway. Membrane-targeted Dvl stabilizes β-catenin and thus Wnt signalling, which promotes cell proliferation and embryonic axis formation.

In chinese hamster ovary (CHO) cells, Overduin and co-workers saw that endogenous dishevelled-2 (Dvl2) exists in two pools — one associated with actin stress fibres and another associated with cytoplasmic vesicles. To investigate whether the DIX domain is involved in Dvl targeting, the authors overexpressed the Dvl2 DIX domain and full-length Dvl2 in CHO cells, and, in both cases, they found that Dvl2 co-localized with actin and cytoplasmic vesicles. However, when they overexpressed a Dvl2 construct that lacked the DIX domain (Dvl2ΔDIX), they found that it remained free in the cytosol.

Using NMR techniques, the authors showed that the Dvl2 DIX domain has two α-helices (α1 and α2), and helical (α3) and extended structural elements, although their three-dimensional arrangement remains unknown. Furthermore, they identified an actin-binding element — YFFKSM⁶⁰ — next to α2 of the Dvl2 DIX domain, as well as a phospholipid-interaction loop — VKEEIS⁷² — between the actin-binding element and α3. They found that mutating lysine 58 in the actin-binding motif to alanine (K58A) abolished actin binding, and, in addition, showed that mutating lysine 68 and glutamate 69 of the phospholipid-binding region to alanine (K68A/E69A) disrupted vesicle associations, without disrupting actin binding.

So, how is this relevant to Wnt signalling? The authors compared the levels of β-catenin in CHO cells that were overexpressing wild-type Dvl2, Dvl2ΔDIX and Dvl2(K68A/E69A), and found that they were significantly lower in the latter two cases. This result is consistent with the inability of Dvl2ΔDIX and Dvl2(K68A/E69A) to be targeted to membranes, and hence to stabilize β-catenin. When they compared the levels of β-catenin in CHO cells overexpressing wild-type Dvl2 and Dvl2(K58A), they found no difference, which indicates that the actin-binding property of Dvl2 is not required for Wnt signalling.

In Xenopus, signalling by β-catenin activates genes that are needed for dorsal axis formation. When the authors studied Dvl2(K58A)-mutant Xenopus axis induction, they found that the dorsal axis duplication frequency was increased, which indicates that the actin-bound pool of Dvl2 is usually sequestered from downstream Wnt/β-catenin signalling. By contrast, they showed that axis duplication was not induced in Dvl2(K68A/E69A)-mutant Xenopus, which indicates that vesicle targeting of Dvl2 is needed for downstream Wnt/β-catenin signalling. Overduin and colleagues have therefore revealed that the DIX domain is a new signalling module that can target proteins to actin stress fibres and cytoplasmic vesicles to control Wnt signalling.