Cleared or not, those images were so obviously doctored that it's hard to believe a professor who read the manuscript could miss it.
Yes it's worrying. As an author you are publicly responsible for the content - you should be confident of the results and claims. I'm not saying you should distrust your coworkers as a rule, but you should at least double check the data to make sure you are convinced by them.
The amazing thing is that manipulated images such as this, and NMR spectra in supporting information for example, do sometimes slip through even when the manipulation is so blindingly obvious that it is quite clear the PI (and the reviewers) never checked it. If you're going to fake data, at least learn how to use photoshop.
I suspect that the "publish or die" culture is at least partly responsible for this (both manipulation and hasty proof reading, not to mention yield inflation, cherry picking etc.).