I use Axure on daily basis. Initially, I used to cover every interaction in the wireframe. Of course it gets 'clickable' then (well, maybe some elements excepted, that need to be described because wireframing these would be overworking or lead to inadequate perception). But a side effect is that on many levels (Client, graphic designers, programmers) these actions remain hidden due to not enough envolvement into plunging in to understand the wireframe completely, put it against the user flow etc.
So, I decided to go the descriptive way. Axure allows you to add item descriptions, but these need clicking to reveal them. For programmers it is pretty clear, but for Client - not really, there is still not enough involvement to click. So, instead of adding actions or labels to the wireframes, I just sketched layouts in Axure, then added an orange box with white letters (so that it looks like a label). Several times I also put all the descriptive boxes in a hidden dynamic panel and added a "click to reveal descriptions".
But there is one major problem with this approach - to wireframe screens with multiple states, one needs to multiply the views in the structure, to cover multiple states. This results in an oversized structure and "Oh my God, it's so big and soooo complicated." from the Client. So, it's not a perfect solution neither. (Something like Photoshop's layer comps would be handy here... it could be covered by buttons switching interface to different states, anyway).
I then started combining both, but now I'm about to change the approach again, covering the interactions wherever it is possible and adding visible description boxes everywhere else. I also want to use some post-WIMP icons to show which elements on the wireframe are clickable, scrollable, draggable etc. just to draw attention of the Client reviewing wireframe.
My question is:
How can I deal with situations when wireframes are just superficially reviewed by Client?
Of course it refers only to situations when Client is left alone with the interpretation, in some cases we go through the wireframe, but I would like these to get self-explanatory in the meaning of interacting with them.