Little is known about the role of RNA-binding proteins in brain development or disease, but accumulating evidence indicates their involvement in neurological disorders. In fact, we found that the RNA-binding protein Pum1 is crucial for neurological function in both mice and humans. Pum1-haploinsufficient mice develop ataxia at 5 weeks of age and show Purkinje cell degeneration at 10 weeks. Pum1 knockout mice are sicker: they are born at a lower mendelian ratio, are smaller than wild-type, have early seizures, and more severe cerebellar degeneration. We then found that mutations in human PUM1 also cause two very different diseases that parallel what we observed in mice: a mild, adult-onset pure cerebellar ataxia in patients with a mutation that reduces PUM1 levels by ~25%, and an early-onset disorder that causes several cognitive and physical delays, smaller size, motor incoordination, and seizures in patients with PUM1 mutations that reduce its levels ~40-60%. But how do the specific mutations alter PUM1 function, aside from making it less stable? The most obvious place to look is at PUM1 targets. The only published neuronal targets are ATXN1 and E2F3, and their abundance is increased by similar amounts (~50%) in both the adult-onset ataxia and early-onset cases. The ataxia might be explained by elevated abundance of cerebellar ATXN1, but the broader phenotype of the developmental disorder must involve dysregulation of other PUM1 targets. There is, however, more of a puzzle here than is first apparent. The mildest mutation, T1035S, which reduces PUM1 levels by only 25%, is in homology domain (HD); of the mutations that produce the severe, early onset phenotype, R1139W is in HD8, and R1147W is just outside this domain. Why, then, do R1139W and R1147W produce equally severe phenotypes, when only the former abolishes PUM1's repressor activity? And why is T1035S so mild, when it also abolishes PUM1's repressor activity? We propose that the milder disease results from target dysregulation, whereas more severe disease results when levels of PUM1 fall below a certain point (perhaps 30-40%), because its interacting partners either cannot form their normal complexes or the complexes fall apart quickly, causing loss of function of those interactors (and loss or gain of function of downstream targets). To test this two-part hypothesis, we will: 1) map the Pum1 targetome in the mouse brain as well as that of Pum2, its homolog (there may be regulatory overlap between the two proteins); 2) identify PUM1 protein interactors, and 3) study the cross-talk between Pum1 and Pum2 in mice. In sum, our recent discoveries not only define two new neurological diseases, they demonstrate that understanding the post- transcriptional regulation of disease-related proteins, like the PUF family, can lead to the identification of new candidate disease genes.
Our recent discoveries not only define two new neurological diseases, they demonstrate that understanding the post-transcriptional regulation of disease-related proteins can lead to the identification of new candidate disease genes. PUM1 and PUM2 targets and interactors can now provide a wealth of new candidate disease genes (after we validate them in vivo). Finally, the idea that modest changes in protein levels can cause different clinical manifestations should prompt re-evaluation of hypomorphic mutations formerly considered nonpathogenic.
