As pgk pointed out to us, the acetic acid might be instrumental in removing the second equivalent of pyridine, the one that is ion-paired to the carboxylate group. However, at this point I would be happy for a relatively pure product in any form.
I recently rediscovered a description of the ethyl ester of our desired compound, which is compound (19) in this paper: Lubbers et al. "Design, synthesis, and structure–activity relationship studies of new phenolic DNA gyrase inhibitors" Bioorg Med Chem Letters (2007) 17:4708-14. I have not looked up one of the papers in reference 10 in this paper yet, which might be the actual description of how it was made. One advantage is that the ethyl ester of bromopyruvate is not as expensive a starting material as the free acid. One disadvantage (besides having to start over) is that we would need to work out a hydrolysis of the product that does not harm the pyridinium ketone functional group.
I looked up one paper in reference 10 within Lubbers' paper, the one from Tetrahedron (2004) 60:2937, by Wang et al. They make the ethyl ester of pyridinium pyruvate, but they move on, apparently without isolating it.