Retracted results

Investigators developing new mouse models of autism, especially those working on SHANK3, have been busy over the past few years. No fewer than five different lines of SHANK3 mutant mice have been established, a substantial effort that reflects in part SHANK3’s importance in the brain, and in part its complex structure.

Each of these mouse lines has unique features, which is why it was unsettling for those in the field to see on 17 January a retraction of a paper describing one of these mutants. Paul Worley and his colleagues originally published the paper in Cell in 2011.

This particular SHANK3 mutant represented an attempt to mimic the mutation that Thomas Bourgeron and his colleagues reported in an individual with autism. In that case, an alteration in exon 21 of SHANK3 disrupts the protein’s binding to another protein, called HOMER.

Given the apparent importance of this protein-protein interaction, the mutant was of particular interest.

The results bore this out. The animals have weakened signaling in certain brain cells and less interest in other mice than controls do. Interestingly, the mutation seems to act as a ‘dominant negative,’ meaning that the mutant copy interferes with the normal copy of SHANK3.

This was the first evidence for such a SHANK3 mutation, and the paper made a case that it somehow results in increased tagging of the normal copy by ubiquitin — a molecule that marks a protein for destruction.

An independent team in 2009 implicated proteins involved in this tagging in autism risk. Given those findings, this result opened up the possibility that SHANK3 might be an important target of this family of proteins.

The retraction seems to call at least some of these claims into doubt. The researchers wrote that figures displaying some of the ubiquitin assays “were improperly assembled.” They went on to say, however, that the errors “did not affect the quantitative analysis of polyubiquitination.”

This explanation left some room for interpretation, and may have left some readers wondering about the status of the mouse and the validity of the major findings. To clarify, I contacted Worley, who provided some reassurance. Most importantly, despite the formal retraction of the entire paper, the mouse does indeed carry the mutation that was reported and is available through The Jackson Laboratory for other investigators to work on.

Worley says that they have verified that their SHANK3 mutation is associated with reduced levels of the normal copy, although it will take some work to nail down the underlying mechanism. New experiments are under way to address the role of ubiquitination in SHANK3 biology, which Worley says he expects to include in a corrected manuscript that will soon be submitted for publication.

A more detailed explanation accompanying the retraction would have been helpful, as it wasn’t obvious in my reading that a full retraction was warranted (as opposed to an erratum or an addendum to the paper).

All told, though, this is the way things are supposed to work, with researchers stepping forward to alert the community when published results are in question. That the mouse remains an important model for autism is clearly also good news. We’ll have a better sense of how to move forward once the corrected version of the paper is published.