Splicing of intron 3 of human BACE requires the flanking introns 2 and 4.

Regulation of proteolytic cleavage of the amyloid precursor protein by the aspartic protease BACE may occur by alternative splicing and the generation of enzymatically inactive forms. In fact, the presence of exonic donor and acceptor sites for intron 3 generates the two deficient variants BACE457 and BACE476. In HEK293 cells, when introns are inserted separately in the BACE cDNA, we found that whilst introns 2 and 4 are efficiently spliced out, intron 3 is not removed. On the other hand, splicing to wild-type BACE is restored when intron 3 is flanked by the two other introns. The presence of all three introns also leads to alternative splicing of intron 3 and the generation of BACE476. In contrast, BACE457 expression takes place only after mutating the donor splice site of intron 3, indicating that additional regulatory elements are necessary for the use of the splicing site within exon 4. Overall, our data demonstrate that a complex splicing of intron 3 regulates the maturation of the BACE mRNA. This appears orchestrated by domains present in the exons and introns flanking intron 3. Excessive BACE activity is a risk factor for Alzheimer's disease, therefore this complex regulation might guarantee low neuronal BACE activity and disease prevention.

Clustering transmembrane-agrin induces filopodia-like processes on axons and dendrites.

The transmembrane form of agrin (TM-agrin) is primarily expressed in the CNS, particularly on neurites. To analyze its function, we clustered TM-agrin on neurons using anti-agrin antibodies. On axons from the chick CNS and PNS as well as on axons and dendrites from mouse hippocampal neurons anti-agrin antibodies induced the dose- and time-dependent formation of numerous filopodia-like processes. The processes appeared within minutes after antibody addition and contained a complex cytoskeleton. Formation of processes required calcium, could be inhibited by cytochalasine D, but was not influenced by staurosporine, heparin or pervanadate. Time-lapse video microscopy revealed that the processes were dynamic and extended laterally along the entire length of the neuron. The lateral processes had growth cones at their tips that initially adhered to the substrate, but subsequently collapsed and were retracted. These data provide the first evidence for a specific role of TM-agrin in shaping the cytoskeleton of neurites in the developing nervous system.

Isoform pattern and AChR aggregation activity of agrin expressed by embryonic chick retinal ganglion neurons.

Several isoforms of chick agrin, which differ in their activity to aggregate AChRs at the neuromuscular junction, are generated by alternative splicing at splice site B. We analyzed the isoform pattern and the functional properties of agrin in a defined population of CNS neurons. At all developmental stages retinal ganglion cells purified by immunopanning expressed the agrin B0, B11, and B19 isoforms. Single-cell RT-PCR of individual retinal ganglion cells revealed simultaneous expression of B0 and B11 isoforms in about half of the neurons analyzed. Despite the expression of agrin isoforms active in AChR aggregation, ganglion cells did not aggregate AChRs when cocultured with myotubes. Addition of exogenous agrin to myotube-ganglion cell cocultures indicated that AChR aggregation is inhibited. These results demonstrate that a defined population of CNS neurons can simultaneously express several agrin isoforms and that the AChR aggregation activity of agrin might be regulated not only by alternative splicing but also on the protein level.