2.2.1. Protein-Protein Interactions
More and more cases are emerging where the interaction of proteins is regulated by protein phosphorylation, subsequently resulting in the regulation of cellular function. Two different mechanisms can account for phosphorylation-dependent protein interactions. First, interacting proteins could directly bind to the phosphorylation site via phosphopeptide binding sequences, or second, a "hidden" binding site could be indirectly exposed due to a conformational change resulting from phosphorylation.
Analogous to tyrosine-phosphorylated proteins, whose binding partners contain SH2 and PTB domains as specific phosphotyrosine binding modules, serine/ threonine-phosphopeptide binding domains also exist. An example is the so-called 14-3-3 proteins that form dimers and contain one phosphopeptide binding site per monomer. They can therefore induce "homodi-
merization" by binding identical proteins to each 143-3 subunit. This is likely to play a role in the regulation of the protein kinase raf. Alternatively, they may form complexes of distinct phospho-proteins. As 143-3 proteins seem less target specific than protein kinases are substrate specific, the actual phosphorylation reaction is the main specificity determinant for 14-3-3 mediated interactions.
WW domains have recently been identified as modular phospho-serine/threonine peptide binding sites. WW domains are short regions of about 30 residues present in several proteins that were thought to bind proline-rich sequences in interacting proteins; it now appears that they may be more important for the regulated binding to phosphorylation sites. The opposite scenario seems to be the case for protein-protein interactions by the co-called PDZ domains that are abolished by serine/threonine phosphorylation of their binding sites.
The following examples illustrate how the regulation of protein-protein interactions through phosphorylation can be crucial for cells, even without affecting the catalytic activity of phosphoproteins. The first example comes from the yeast DNA damage response pathway. In response to DNA damage the noncatalytic Rad9p protein becomes phosphorylated. The major function of this phosphorylation event is to induce binding to the protein kinase Rad53p. This interaction depends on yet another modular domain, the so-called FHA domain located in the C-terminus of Rad53p. The FHA domain is essential for the activation of Rad53p by a postulated protein kinase. If this interaction is abolished, yeast are impaired in their ability to induce transcription of repair enzymes and to arrest the cell cycle until the damage is repaired, resulting in a severe loss of viability. Phosphorylation of Rad9p is crucial for the survival of the cell. A similar example is found in the differentiation of striated muscle. Here, the small noncatalytic protein telethonin is phosphor-ylated by the protein kinase domain located in the C-terminus of the large structural protein titin (>30,000 residues) and telethonin can then associates with the N-terminal region of titin as a crucial event in the formation of myofibrils.
As the subcellular localization of soluble proteins is generally controlled via a set sequence of proteinprotein interactions, the regulation of the subcellular localization of proteins by phosphorylation is in some way just another facet of the regulation of protein interactions by phosphorylation. Most examples of
phosphorylation-regulated subcellular localization relate to the regulation of the nuclear localization of proteins. Phosphorylation can affect the nuclear localization of proteins two ways. Some proteins are preferentially localized to the nucleus following phos-phorylation, whereas other proteins are excluded from the nucleus if they are phosphorylated. The classical mechanism for proteins to be imported into the nucleus—unless they are small enough to freely traffic through nuclear pores—is to bind to the soluble import factor importin-a via a basic nuclear localization signal (NLS). In many cases the interaction with importin-a can be abolished by phosphorylation in, or in close proximity to, the NLS by various protein kinases, resulting in their cytoplasmic retention. Phos-phorylation can also prevent the nuclear export of proteins, for example, in the case of cyclin B1 that accumulates in the nucleus after it is phosphorylated at the G2/M transition of the cell cycle. An interesting variation of the regulation of nuclear localization by phosphorylation has recently been identified for the Cdc25 phosphatase that is crucial for the initiation of mitosis by dephosphorylating the inhibitory Tyr15 in the mitotic Cdc2 cyclin dependent protein kinase. Cdc25 phosphorylated on Ser216 binds to 14-3-3 proteins that serve in this case as "attachable" nuclear export signals. Cdc25 is therefore retained in the cytoplasm during interphase, preventing the access to its substrate (Fig. 7). There are several protein kinases that can phosphorylate Cdc25 on this residue, and the checkpoint kinases Chk1 and Chk2/Cds1 may be the most relevant physiologically for this process. When the cell senses that all requirements for an accurate mitosis are fulfilled, activity of these kinases is reduced, and the unphosphorylated Cdc25 can enter the nucleus to signal the onset of mitosis.
Phosphorylation on serine or threonine residues can be crucial for regulating a protein's fate, and again, this can work two ways. A prominent example of a protein whose stability is enhanced in response to phosphorylation is the tumor suppressor p53. One function of p53 is to act as a transcription factor for p27, an inhibitor of mitotic cyclin-dependent kinases. Phospho-p53 accumulates in the cell, whereas dephospho-p53 is more rapidly destroyed. A candidate protein kinase regulating p53 stability is the ATM protein that functions as an effector in mammalian DNA damage and oxidative stress pathways. As a result, cells lacking ATM have reduced levels of p53, and humans lacking ATM suffer from a syndrome called ataxia-telangiectasia that comprises increased susceptibility to tumor development, increased sensitivity to ionizing radiation, and neuronal degeneration. It is currently not known if there is a specific mechanism for the destruction of unphosphorylated p53, or if the latter is simply less stable.
Mechanisms for a plethora of proteins that are degraded in a phosphorylation-dependent manner have recently been elucidated. Such proteins mostly function in the cell cycle and are recognized by so-called F-box proteins only when they are phosphory-lated. The F-box proteins, of which more than 400 are currently known, associate with their targets via leucine-rich repeat or WD40 modular domains, and interact through the F-box motif with Skpl in the so-called SCF ubiquitin ligase complex. In this way, the phosphorylated proteins become ubiquinated, earmarking them for degradation via the 26S proteasome. In some sense, this phosphorylation-dependent protein degradation adds a component of irreversibility to the primarily reversible nature of protein phosphorylation.
Was this article helpful?