The hole is 8 kDa wide—the size of the amino-terminal stub that should be left behind when the 35-kDa piece is produced. That amino terminal peptide has not been observed. “If you cut TDP-43, you should get two parts,” said Xiao, who works with Janice Robertson at the University of Toronto, Canada. These scientists present a different theory. They propose that an alternative splice form of TDP-43 mRNA allows the ribosome to begin translation at a downstream start codon to make the carboxyl-terminal fragment of 35 kDa. Xiao cannot explain the provenance of the second 25-kDa C-terminal fragment, which could still be produced by caspase cleavage, he speculated.

In his poster, Xiao detailed the following scenario: TDP-43 has an alternatively spliced form that lacks 91 base pairs from exon 2. This deletion causes a frameshift, generating a premature stop codon and shortened open reading frame (ORF). It is unclear if this ORF is translated; however, the ribosome finds a second start codon farther down the transcript. This start codon normally encodes methionine 85 in full-length TDP-43, but in this alternate splice form acts as a second translational initiation site. That would then allow the ribosome to produce a protein of approximately 35 kDa. It is this product of alternative splicing and ribosomal scanning, not caspase cleavage, that accounts for the toxic C-terminal TDP-43 fragments, Xiao believes.

How did he get to this theory? While puzzling over the absence of the amino-terminal half of the supposedly cut-up protein, as well as a report that the fragment appears even in the absence of caspase 3 (Nishimoto et al., 2010), Xiao examined the mRNA sequences reported for TDP-43 in GenBank. He noticed the alternative splice form with the deletion and the premature stop codon. Xiao amplified and cloned this alternatively spliced mRNA from spinal cord samples donated by people who had amyotrophic lateral sclerosis (ALS). The splice variant was four times more prevalent in 12 ALS samples than in four control samples.

Xiao suspects that the first ORF of the alternatively spliced RNA is either not translated, or that the peptide product is rapidly degraded. To find evidence for the proposed C-terminal piece produced by the second start site, Xiao engineered an antibody specific for its amino end—MDETDASSA. He also generated an antibody to the amino end of the predicted caspase-cleaved 35kDA fragment—ASSAVKVKR. The antibodies clearly distinguished the fragments from each other, said Xiao, even though the two peptides share the ASSAVKVKR sequence. In patient tissue, the antibody to the translated fragment colocalized with a polyclonal full-length TDP-43 antibody, but the caspase-cleavage antibody did not. The researchers concluded that the cleavage-specific antibody does not recognize TDP-43 aggregates—and that the only fragments present were due to the alternative start site.

Next, Xiao transfected SH-SY5Y neuroblastoma cells and primary motor neurons with the code for the alternative start fragment. The cells formed toxic, ubiquitinated, insoluble aggregates in the cytoplasm; these recruited full-length TDP-43, matching what goes on in disease pathology. The scientists concluded that the 35-kDa TDP-43 fragment seen in people with ALS likely arises from alternative splicing and translation of an alternative open reading frame, not from caspase cleavage. But it remains possible that the 25-kDa fragment is a cleavage product, Xiao said.

The data are “convincing,” commented both Jean-Pierre Julien of Laval University in Québec City, Canada, and David Borchelt of the University of Florida in Gainesville, in e-mails to ARF. Both saw the data at the André-Delambre Foundation Symposium on ALS held 23-24 September 2011 in Québec City, Canada. “Their specific antibody is a very nice tool to study alternative splicing further,” emailed Mahmoud Kiaei of the University of Arkansas for Medical Sciences in Little Rock, who plans to collaborate with Xiao and Robertson.

Leonard Petrucelli of the Mayo Clinic in Jacksonville, Florida, noted that his and other groups have clearly seen a caspase-cleavage TDP-43 product (see ARF related news story; ARF news story; ARF story; and Dormann et al., 2009). If the enzyme cuts TDP-43 to make the 25-kDa piece, then it would also cut upstream to make the 35-kDa piece, he argued. As for the missing amino-terminal cleavage product, he said he has not specifically looked for it. The piece would be only 8 kDa and would run right off many standard gels. It could also be produced, but quickly degraded, he noted. But Petrucelli did not discount the possibility of alternative splicing: “Both mechanisms could exist,” he told ARF.—Amber Dance.