There are basically two possibilities to do this.
1. Implicit solvent method. In ATK, you can assign a dielectric constant to the "vacuum" between atoms. This way you can introduce the effect of water via its dielectric constant without adding any actual water molecules. Naturally this is less accurate, but you save a lot of trouble (the whole MD part is not necessary any more, and the system is smaller to compute, since there are fewer atoms).
2. Explicit water molecules. Actually, given the dimensions of the system, you don't need that many water molecules, so this doesn't really add too much to the calculation time, except perhaps for the MD part, but you should be able to do that with less sophisticated methods like force-fields or similar.
I am aware of at least one ATK publication (since I am co-author!) with explicit water molecules in a two-probe system, and in fact one of the primary conclusions of the paper is that the water provides a very important transport pathway for the electrons. Now, the only problem is that the paper is still pending publication... It's accepted to Journal of Physical Chemistry, but it is not yet available from the journal website.
I may be able to send you a preprint, if you contact me by email and also state your affiliation etc.
