29 November 2005. This is the third installment of a four-part news series about the role of the microtubule-associated protein tau from the 35th Annual Conference of the Society for Neuroscience, held November 12 to 16 in Washington, D.C. See also Introduction and Part 1, Part 2, and Part 4.
On the Chopping Block
What with Aβ-degrading enzymes up to almost a half-dozen candidates by now (neprilysin, IDE, ECE, ACE, MMP-9; see ARF related SFN story), one would think that an equally impressive phalanx of enzymes is at hand to slice up tau. Alas, tau degradation remains shrouded in mystery. Activated calpain is known to cleave tau in pathological situations, and some researchers have proposed that tau cleavage by caspases promotes neuropathology. The ubiquitin-proteasome pathway has been implicated in pathological situations (see Yang presentation below), but whether it disposes of tau as a normal, or neuroprotective measure remains questionable (David et al., 2002; Feuillette et al., 2005). At the SfN conference, a little-noticed talk at the very end of an afternoon session, when most of the audience had already slipped away, began to fill this gap with a well-substantiated story about an obscure enzyme that relentlessly chops away at tau from the protein’s amino terminal until it is completely degraded.
George Jackson, of the University of California, Los Angeles, presented some of the data gathered in a three-way collaboration among the labs of Daniel Geschwind, also of UCLA; Jackson; and Skip Binder at Northwestern University in Chicago. The story began in Geschwind’s lab, where scientists looked for gene expression differences between cerebellum and cortex of the frontotemporal dementia mouse model P301L. They acted on the hunch that the cerebellum resists tau pathology and neurodegeneration by means of some sort of protective gene expression. This microarray analysis found not only expected genes, such as the apoptosis break bcl2 and the neuroprotective VEGF, but also unexpected ones. This group included PSA, which stands for puromycin-sensitive aminopeptidase. Aminopeptidases are enzymes known to be able to generate free amino acids, but are poorly studied in the human brain. The scientists validated PSA upregulation with Northern blots and in situ hybridization, as well as tissue microarray analysis of normal human and FTD brain where, again, PSA expression was decreased relative to normal brain in vulnerable areas, such as frontal cortex, but increased in resistant areas, such as cerebellum.
Jackson presented in more detail his work validating the function of PSA in a fly model of neurodegeneration. In brief, flies expressing 4R tau (chosen because FTDP mutations lead to a relative increase in 4R tau) in the 800 ommatidia of their compound eyes show a “rough eye” phenotype that allows for ready detection of neurodegeneration. Cross-sections of affected compound eyes reveal a morphology in disarray and cell loss. Genetic tinkering changed that phenotype in ways that jibed with the idea of PSA being a neuroprotective modulator of tau, Jackson reported. That is, coexpressing an inactive mutant form of PSA worsened the 4R tau phenotype, whereas overexpressing active PSA remedied it. PSA suppresses both mutant and wild-type tau phenotypes in flies, and it leads to a decline in tau levels. Binder’s lab then cloned and purified recombinant human tau and incubated it with PSA. The peptidase degraded tau, chewing it up without a trace, and even appeared to show a relative preference for mutant tau over wild-type tau. Puromycin, as well as the peptidase inhibitor bestatin, counteracted this degradation.
This work suggests that certain brain areas upregulate PSA in the presence of mutant tau, and that the enzyme then takes tau apart one amino acid at a time. It is known that PSA expression tends to decline with age, Jackson noted, adding that PSA may make a new therapeutic target. The researchers don’t yet know why PSA appears to be relatively specific for tau, or why it degrades FTDP-mutant tau more readily than wild-type. Interestingly, the PSA gene resides on chromosome 17, near the FTD locus.
Tau Proteomics for a Future Biomarker
The PSA story started out with a gene expression difference, but for other questions it’s necessary to zero in on post-translational changes, which DNA microarrays do not capture. A case in point is the issue of how a slew of changes to existing tau protein occur in relation to disease. A large literature on hyperphosphorylation of tau paired helical filaments, and a smaller one on tau ubiquitination, have made clear that these changes occur and are important. Yet their precise place in the pathogenic cascade has not been determined. It is not clear, for example, when these changes happen relative to the appearance of degradation-resistant forms of tau. One limitation has been that most studies look at some phosphorylation sites but not others, making it difficult to organize and compare existing knowledge quantitatively. Austin Yang at the University of Southern California took a proteomic stab at measuring tau modifications in human tissue more comprehensively. Yang’s team took autopsy samples from early AD brain provided by his collaborator Peter Davies at Albert Einstein College of Medicine in the Bronx, New York. Then the scientists immunopurified soluble pre-tangle tau, digested it with proteases, and ran the peptides through liquid chromatography and high-throughput mass spectrometry. This procedure yielded a complete fingerprint of all tau phosphorylation sites, as well as three amino acids on tau where ubiquitin attaches and elongates poly-ubiquitin chains, Yang said. The data indicate that the formation of ubiquitin chains depends on the phosphorylation state of soluble tau, and that the ubiquitin-proteasome plays a role in the subsequent step of forming protease-resistant tangles. On this issue, Frank LaFerla’s group at the University of California, Irvine, also reported at the conference that blocking the proteasome intensified tau pathology, though these scientists propose that an intraneuronal buildup of Aβ oligomers (see upcoming SfN news summaries) inhibits the proteasome.
Yang’s kind of proteomics analysis is at an early stage, only now moving into exploratory tests with cerebrospinal fluid. However, it is a much richer indicator of pathological tau modifications than the CSF phospho-tau antibody tests currently under development as a diagnostic marker, and could one day replace them, Yang hopes.—Gabrielle Strobel.
See also Introduction and Part 1, Part 2, and Part 4 of this series.