Signal transduction

Grabbing phosphoproteins

Article metrics

The control of complex cellular events such as DNA repair, actin rearrangement, programmed cell death and transcriptional regulation requires acute orchestration of many signalling molecules, including kinases, phosphatases, proteases, guanine-nucleotide-binding (G) proteins and transcription factors. These molecules must relocate to specific sites in the cell, and their activities need to be regulated in a timely manner, often involving interactions with other proteins. Several papers1,2,3, the latest by Durocher et al. in Molecular Cell1, have revealed a new twist on the mechanism by which protein phosphorylation regulates the assembly of such complexes.

The structural basis for the reversible assembly of signalling complexes came with the discovery that the Src homology-2 (SH2) domains of various signalling proteins bind directly to phosphorylated tyrosine (‘phosphotyrosine’) residues on growth-factor receptors or adaptor molecules. Auto-phosphorylation by these growth-factor receptor tyrosine kinases creates binding sites for proteins that contain an SH2 domain; conversely, dephosphorylation disassembles the complex.

Over the past decade this concept has provided the framework for advances in tyrosine-kinase signalling and, until recently, it was thought to be unique to tyrosine-kinase signalling pathways in higher eukaryotes. Phosphorylation of proteins on serine and threonine, by contrast, was thought to regulate signalling by inducing conformational changes in the molecules involved, rather than by creating binding sites for other proteins.

The new papers1,2,3 dispel this theory, revealing families of protein domains that directly bind to phosphoserine or phosphothreonine residues in the same way as the SH2 domain interacts with phosphotyrosine (Fig. 1). The first clue was the discovery4,5 that the so-called 14-3-3 proteins bind directly to phosphoserine in a specific sequence context. Second, the conserved tryptophan–tryptophan (WW) domain of a protein called Pin1 was found2 to bind phosphoserine within a phosphoserine– proline motif. And now, Durocher et al.1 reveal that a family of ‘Forkhead-associated’ (FHA) domains, found in a variety of signalling proteins conserved from prokaryotes to mammals, can bind directly to phosphothreonine peptides.

Figure 1: An expanding family of phosphoserine/phosphothreonine-binding proteins.

The Src homology-2 (SH2) domain was the first motif shown to bind a phosphorylated amino acid — in this case, phosphotyrosine. Durocher et al.1 now show that the two Forkhead-associated (FHA) domains in the Rad53 protein are phosphothreonine-binding modules, and Li et al.3 have found that the FHA domain of the KAPP protein interacts with several protein kinases in a phosphorylation-dependent manner. Lu et al.2 have shown that the WW domain of Pin1 can also bind phosphoserine/threonine. The family of 14-3-3 proteins is also known to bind to distinct phosphoserine/threonine motifs4,5.

So it seems that modular domains allowing signalling complexes to assemble through phosphoserine and phosphothreonine residues are more common than expected. These domains are thought to have evolved independently from unrelated sequences. The FHA domains were first identified by a database screen centred on a conserved region in members of the Forkhead-type transcription factor family6. Found mainly in protein kinases, phosphatases and transcription factors, FHA domains consist of roughly 75 amino acids divided into three or four highly conserved sequence blocks. They were identified in proteins from yeast, plants, mammals and bacteria, and in many uncharacterized open reading frames.

But the function of FHA domains remained a mystery. A year before these domains were identified as distinct signalling modules, John Walker's laboratory identified a 239-amino-acid stretch within the kinase-associated protein phosphatase (KAPP) of the tiny weed Arabidopsis thaliana that bound a serine/threonine-receptor-like kinase, RLK5. This interaction depended on phosphorylation of RLK5 on serine or threonine residues7. Sequence comparisons revealed an FHA domain in the middle of this kinase-interacting domain, and earlier this year Walker and colleagues reported3 that this FHA domain is critical for the phosphorylation-dependent binding of KAPP to several protein kinases.

A second clue that FHA domains might be phosphoserine/threonine binding modules emerged from studies by David Stern and colleagues. They were searching for proteins that bind to Rad53, a protein kinase involved in controlling cell-cycle checkpoints in the yeast Saccharomyces cerevisiae8. Rad53 contains two FHA domains flanking a central protein kinase domain, and Stern's group mapped the interaction of Rad53 with another DNA-damage-control protein, Rad9, to the second of these FHA domains. Like the interactions between KAPP and RLK5, the Rad53–Rad9 coupling was specific for the phosphorylated form of Rad9.

Durocher et al.1 now report that, in fact, both of Rad53's FHA domains can bind to Rad9. They show the first direct interaction between the amino-terminal FHA domain of Rad53 and short peptides containing phosphothreonine. Then, using immobilized peptides, they demonstrate that Rad53's carboxy-terminal FHA domain, along with the FHA domains from proteins in Arabidopsis, humans, yeast and Mycobacterium tuberculosis, also binds sequences that contain phosphothreonine. Although the exact motifs that FHA domains recognize — and the serine/threonine kinases that generate these motifs — are not known, all the data point to the FHA domains as 'SH2-domain equivalents' in the world of phosphoserine/threonine signalling.

But there are other phosphoprotein-binding domains. Some WW domains, for example, bind to phosphoserine. They consist of around 35–40 amino acids folded into a three-stranded β-sheet, and bind to a variety of proline-rich proteins. One such protein, Pin1, regulates entry to and exit from mitosis, presumably by isomerizing phosphoserine–proline bonds9. Earlier this year, Lu et al.2 showed that the WW domains from Pin1, the yeast Pin1 homologue Ess1 and the ubiquitin ligase NEDD4, act independently as phosphoserine/threonine-binding modules.

The most highly conserved phosphoserine/threonine-binding proteins identified to date are members of the 14-3-3 protein family4,5. In contrast to modular signalling elements (such as FHA and WW domains), which are usually interspersed with other domains in signalling proteins, 14-3-3 molecules are themselves functional dimeric proteins. There are seven different 14-3-3 isotypes in mammalian cells, and at least ten in Arabidopsis, allowing them to act in tissue- and organelle-specific ways. When 14-3-3 proteins bind their phosphoprotein prey, they are thought to restrict the bound proteins to the cytosol, either by preventing them from entering the nucleus or by speeding their passage out of it.

Other families of phosphoprotein- binding domains probably remain to be discovered. Those that have been identified so far are unrelated in primary sequence and three-dimensional structure, so must have evolved independently. In some cases — such as phosphotyrosine-binding and WW domains — only one sub-group of the family has phosphopeptide-binding specificity. Perhaps sub-groups of other, previously defined modular domains have evolved phospho-amino-acid specificity. For example, there is evidence that WD40 repeats within F-box proteins mediate the phosphoserine/threonine-dependent binding and ubiquitin-mediated degradation of cyclin-dependent kinase inhibitors10,11. The challenge for the future will be to identify the function of each phosphoserine/threonine-binding protein in specific signalling cascades.


  1. 1

    Durocher, D., Henckel, J., Fersht, A. R. & Jackson, S. P. Mol. Cell 4,387–394 (1999).

  2. 2

    Lu, P. J., Zhou, X. Z., Shen, M. & Lu, K. P. Science 283, 1325–1328 (1999).

  3. 3

    Li, J., Smith, G. P. & Walker, J. C. Proc. Natl Acad. Sci. USA 96, 7821–7826 (1999).

  4. 4

    Muslin, A. J., Tanner, J. W., Allen, P. M. & Shaw, A. S. Cell 84, 889–897 (1996).

  5. 5

    Yaffe, M. B. et al. Cell 91, 961–971 (1997).

  6. 6

    Hofmann, K. & Bucher, P. Trends Biochem. Sci. 20, 347–349 (1995).

  7. 7

    Stone, J. M., Collinge, M. A., Smith, R. D., Horn, M. A. & Walker, J. C. Science 266, 793–795 (1994).

  8. 8

    Sun, Z., Hsiao, J., Fay, D. S. & Stern, D. F. Science 281, 272–274 (1998).

  9. 9

    Yaffe, M. B. et al. Science 278, 1957–1960 (1997).

  10. 10

    Skowyra, D., Craig, K. L., Tyers, M., Elledge, S. J. & Harper, J. W. Cell 91, 209–219 (1997).

  11. 11

    Henchoz, S. et al. Genes Dev. 11, 3046–3060 (1997).

Download references

Author information

Correspondence to Michael B. Yaffe or Lewis C. Cantley.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.