The study also provides evidence for tissue-specific BACE1 mRNA splicing, showing that the relative abundance of alternative transcripts varies in brain versus pancreas.

Previous studies have found BACE1 splicing variants in brain (Zohar et al., 2005; Tanahashi and Tabira, 2007), but no one had done a systematic, quantitative study of the isoforms. In the new work, Mowrer used PCR to measure levels of BACE1 transcripts in brain and pancreas. In both tissues, the major splice form is full-length BACE1. Smaller splice forms involving exons 3 and 4 appeared in both tissues, and Mowrer found small but reproducible and statistically significant differences in the relative number of these splice variants in brain versus pancreas. The results suggest that something in the cellular environment can regulate splicing. No differences were seen among different brain regions including cerebellum, frontal lobe, and hippocampus, but the researchers did find a previously unknown splice form of BACE1 that lacked exon 4.

The alternatively spliced region of BACE1 in exons 3 and 4 involves residues that lie close to the active site in the folded protein. Mowrer took the analysis one step further by probing the activity of BACE1 protein encoded by the different splice variants. She cloned each transcript and expressed them in HEK cells, then purified the proteins and tested them in an in-vitro assay for β-secretase activity. The activity of all four alternative splice variants was down by 80 percent or more compared to full-length BACE1.

This raises the possibility that if splicing can be modulated in cells, it could be a way to lower BACE activity. In support of this idea, Mowrer showed that when she treated APP-expressing HEK cells with antisense oligos to exon 3 and 5 splice sites to block normal splicing, the cells had higher levels of alternatively spliced isoforms encoding less active BACE species. This roughly halved the cells’ Aβ production.

“Our results serve as a proof of principle that it may be possible to pharmacologically tweak splicing. If we can do that, it should help reduce BACE1 activity, and Aβ production,” Wolfe told ARF. The antisense approach has some challenges, not the least of which is delivery of RNA oligos to the brain. At the same time, it has the potential advantage of only partially inhibiting BACE1, which might be important in preventing side effects due to the interference with the cleavage of non-Aβ substrates. For example, BACE1 also cleaves neuregulin and plays a role in myelination (see ARF related news story). A recent paper suggested that complete knockdown of BACE1 and defective neuregulin processing in mice can result in behavioral changes reminiscent of schizophrenia (see ARF related news story). Thus, incomplete inhibition of BACE1 may be a useful strategy to pursue for safely lowering Aβ, Wolfe said.

A possible role for alternatively spliced BACE1 species in vivo remains to be proven. As of yet, Wolfe says, he doesn’t have any evidence of splicing changes in AD, but his lab is continuing to look at that question. Finally, this study adds to the rapidly growing number of proposed ways by which BACE1 activity is controlled. There is translational inhibition by microRNA (see ARF related news story and ARF news story), by the stress response protein eIF2α (see ARF related Keystone story), there is transcriptional control via Cdk5 (see ARF related news story), and now splicing. Future research will sort out which of these regulatory mechanisms are most important in human AD, and which prove to be druggable.—Pat McCaffrey.

Reference:
Mowrer KR, Wolfe MS. Promotion of BACE1 mRNA alternative splicing reduces amyloid beta-peptide production. J Biol Chem. 2008 May 8. [Epub ahead of print] Abstract