Pat McGeer and colleagues at the University of British Columbia, Vancouver, report that tau is degraded by the protease thrombin. First author Testuaki Arai and colleagues incubated protease inhibitors with cytosolic extracts from human brain to see which inhibitors prevented proteolytic degradation of tau. (This work was also presented at the 34th annual meeting of the Society for Neuroscience, see SfN abstract 788.2.) When Arai used the thrombin-specific inhibitor PPACK (D-phenylalanyl-L-prolyl-arginyl chloromethyl ketone), he found that it completely repressed tau breakdown, whereas in the absence of the inhibitor, tau was degraded into short (< 25 KDa) fragments within hours.

To determine which peptide bonds are severed by the protease, Arai and colleagues incubated recombinant tau with purified thrombin and analyzed the resulting peptide fragments. This revealed that thrombin cuts tau in five locations (after Arg155, Arg209, Arg230, Lys340, and Lys257). The time course of proteolysis suggests that the Arg155-Gly156 bond is hydrolyzed first and that the remaining C-terminal end of tau is then cleaved from both the N- and C-terminal directions.

But what about phosphorylated tau, which is, after all, the most toxic form? Arai found that if tau is first phosphorylated with glycogen synthase kinase 3β (GSK-3β, which has been implicated in tau phosphorylation and may be a viable drug target; see ARF related news story), then the first thrombin cleavage, which normally takes place within seconds, proceeds much more slowly. For example, using Western blotting, Arai found that full-length unphosphorylated tau is completely eliminated after a three-hour incubation with thrombin, whereas about 50 percent of phosphorylated tau remains. Phosphorylation may also prevent the other four cleavages completely, because though smaller fragments of about 25, 23, and 16 KDa resulted from incubating unphosphorylated tau with thrombin, the authors failed to detect these peptides when they incubated phosphorylated tau with the protease. Physiologically, these observations may be relevant because when Arai used thrombin to digest paired-helical fragments of tau isolated from Alzheimer disease brain tissue, he found that digestion of the protein was negligible unless it was first incubated with alkaline phosphatase. This suggests that thrombin may have difficulty degrading excess phosphorylated tau in vivo, leaving it available for incorporation into neurofibrillary tangles.

Which brings us to the second paper from Cheng-Xin Gong and colleagues at the New York State Institute for Basic Research and Purdue University, Indiana. First author Fei Liu and colleagues report that protein phosphatase 5 (PP5) rivals protein phosphatase 2A (PP2A) in dephosphorylating tau.

In comparison to the hundreds of known kinases, there are only a handful of phosphatases and each of them must dephosphorylate a wide range of protein substrates. PP1, PP2A, and PP2B have all been shown to dephosphorylate tau in vitro, but PP2A may play a major role in vivo because in transgenic mice with reduced PP2A activity, tau becomes hyperphosphorylated and redistributed in CNS neurons (see Kins et al., 2001), while reduced expression of the phosphatase in the brain has been associated with AD (see, for example, Vogelsberg-Ragaglia et al., 2001). However, PP2A does not dephosphorylate all the sites on tau, suggesting that other phosphatases may be equally important in vivo.

When Liu and colleagues tested recombinant PP5 in a test tube assay, they found that it dephosphorylated tau about as well as PP2A does, but not necessarily at the same sites. There is evidence to suggest that PP2A dephosphorylates serines 262, 356, 396, and 404 most efficiently (see Sun et al., 2003), but Liu found that PP5 preferentially dephosphorylated tau at threonine 205, 212, and serine 409.

So are PP2A and PP5 cooperating to remove phosphate groups from tau in vivo? Perhaps. When Liu and colleagues measured PP5 activity in brain extracts from AD patients, they found it was reduced by about 20 percent. Expression levels were no different than in control samples, however, indicating that the enzymatic activity may be modulated in AD brain. The authors suggest that the phosphatase may be post-translationally modified or that an inhibitor may copurify with the enzyme to poison the phosphatase assay.

Indeed, protein partners are well known to modify protein phosphatase activities. Just recently, work from Bradley Denker and colleagues at Brigham and Women’s Hospital and Harvard Medical School showed that the G protein Gα12 interacts with the scaffolding subunit (Aα) of PP2A, enhancing its dephosphorylation of tau.

First author Deguang Zhu and colleagues found that Gα and Aα coimmunoprecipitate, and using immunofluorescence, the authors were also able to show that the proteins colocalize in primary neurons. In addition, the researchers found that adding Gα12 to PP2A in vitro elicited a threefold increase in phosphatase activity. The authors also found that, when expressed in COS cells that produce copious amounts of phosphorylated tau, Gα12 was found to stimulate dephosphorylation of tau by about 60 percent.

Ironically, another ligand of Gα12 is thrombin, suggesting that this G protein may regulate both the phosphatase and protease activities that are necessary to prevent accumulation of toxic tau.—Tom Fagan.

References:
Arai T, Guo J-P, McGeer PL. Proteolysis of non-phosphorylated and phosphorylated tau by thrombin. J. Biol. Chem In press 2004 November 12. Abstract

Liu F, Iqbal K, Grundke-Iqbal I, Rossie S, Gong C-X. Dephosphorylation of tau by protein phosphatase 5: Impairment in Alzheimer disease. J. Biol. Chem. In press 2004 November 15. Abstract

Zhu D, Kosik KS, Yanamadala V, Denker BM. Galpha 12 directly interacts with PP2A; Evidence for Galpha 12-stimulated PP2A phosphatase activity and dephosphorylation of microtubule associated protein, Tau. J Biol Chem. 2004 Nov 3; [Epub ahead of print] Abstract