I think that you have several options for a co-culture system; however, I think that using the transwell setup may be difficult and ultimately, the system you use will depend ultimately on what you are looking for. Using the transwell system it may be possible to use media conditioned before hand through culturing different combinations of cells, then removing that media spinning it down, and placing that in the bottom well of the transwell. The problem with this is that I have found the gradients of signaling molecules to be somewhat transient. Another option would be to plate two cell lines on the bottom with the cell line of main interest on the bottom; here there are many permutations that allow you to further analyze/ validate initial findings, but with these permutations the setup becomes cumbersome and costly. Finally, I would suggest that you consider using a 3D culture system in which cells are encapsulated in sodium alginate or methyl cellulose in this way the cells are really only in contact through soluble factors as in the transwell system. In this system you can make the spheroids different sizes and shapes so as to distinguish which cells are in which 3D culture. Furthermore, studies show that this is a slightly more accurate in vitro model. The downside to this last method is that there are very few if any well studied functional assays, but these methods are pretty good for looking at protein changes that may or may not occur as a result of co-culture. Regardless of which way you choose, this type of experiment is pretty tricky to get to work.