Not sure if you got this working but I know when I was doing westerns with channels/abs I wasn't used to we would see weird sizes all the time and we decided it was due to the glycosylation state of the channel. Could it be something like in this paper (fig2) where they should that the alpha subunit of the nAchR is heavily glycosylated? You could try the tricks they use...
Sometimes proteins run differently on a gel than the predicted MW due to post-translational modifications such as glycosylation - but this shouldn't change the MW by too much. As Julia said it could be a dimer if it is double the weight. Could be also that your samples have a splice variant of that gene. Saying that, that band on the blot doesn't look too specific - if you altered your ab concentrations / block to clean up the blot you might flind that band disappears. First step is a positive control then if you want to be sure you could cut out the band from the gel and sequence it