The issue with this is that while taking fluorescence, there is background fluorescence of the compound still not washed out after PBS washing.

There are certain drugs that bind to their targets quite strongly and even washing for several times would not completely get rid of it.

If you do flow cytometry staining, you could use different fluorochromes for your targets (what you actually would like to see) and that way you could possibly get rid of the drugs fluorescence (at least for the analysis). You just have to make sure that the fluorescence wavelengths do not overlap.

If you want to stain them in the attached state, maybe a similar approach would also work so that you use different (not overlapping) colours. Maybe it's even interesting to see where the drug binds in your cells and where your protein of interest is ;-)

Good luck

We come across a collaborator telling us a somewhat similar problem.

If you are doing immunostaining, then the logical way out is to change the secondary antibody's fluorophore to a deep red or NIR one.

if you are doing small molecule (dye) staining, then I suggest you use confocal imaging, at relatively low magnification like 40x (oil lens). With confocal, you can restrict your excitation and emission regions, so as to exclude or minimize the nonspecific fluorescence (which cannot be avoided in epifluorescence). 

You can include a "drug only but no dye" (control) to see how much autofluorescence the drug controbibutes. 

When doing low enough magnification in multiple locations, you can collect enough cells and use ImageJ to extract the fluorescence. After subtracting the background, you can have a fairer and more confident estimate of the effects. 

Alternatively, as noted above by Miriam, you can try FACS. 

***

I also agree with Miriam that if the drug indeed is fluorescent and binds to some cell components, it gives away hints on how it operates. It's not necessarily a bad thing!

Thanks Nai-Kei-Wong