Me and a teammate found this bug today while testing our key management tool.
His public key has expired on 2014-01-01 and so he added another year. His new
expire date was 2015-01-02 then and he uploaded the updated key to the keyserver
hkp://pgp.mit.edu.
Today we checked the key on the keyserver via the web-interface and gpg.exe
--search-key

As you can see on the screenshots, the expire date of the key seems to be
misinterpreted by gpg.exe, because the hkp output of the keyserver is correct.

You can validate this bug with any testkey you upload, that has changed its
expire date at least once.