It seems something with the reference sequence generator has gone wrong (http://taletoolbox.genome-engineering.org/#).
Some nucleotides in the RVD linkers are not the same as they used to be so that the reference sequence does not correspond anymore to what is produced your 2012 Nature protocol.
The protein sequence is correct however, so there might be some inversion in part of the code or something?
Aligning the reference sequences for the same TALEN either generated with the old version of the tool or the newer one will illustrate my point.
It would be truly fantastic if you could fix that bug.
Thank you very much in advance!
Elphege