Both of the terms you're mentioning are odd ways of compressing down what is going on in Kant. Odd enough that I wasn't sure Kant had stated either of them in that way.
"transcendental knowledge" is knowledge that is not of objects but rather about the apparatus of knowledge. In other words, for him, it is the metaphysics of knowing, i.e. it is knowing the categories themselves and the forms of sensibility that condition our knowledge of objects (http://en.wikipedia.org/wiki/Transcendence_%28philosophy%29 http://www.iep.utm.edu/kantmeta/ ).
"a priori knowledge" is confusingly knowledge that is achieved under the conditions of the a priori (rather than knowledge that is about the a priori) and thus refers to the way we know objects under categories and through the forms of sensibility (Cf. http://plato.stanford.edu/entries/apriori/)
Transcendental knowledge is not subsumed under the analytic / synthetic distinction because it is not really subject to the conditions of understanding objects. (Here, I use really as a technical term). However, for us, transcendental knowledge is an ideal, so its possible that the mode through which we encounter it requires us to synthesize features together to grasp what it is (though never perfectly).