There are a number of proteins and metabolites that worsen their autofluorescence upon fixation, among them lipofuscins, flavins, NADH/NADPH and porphyrin-containing proteins. Generally highly metabolic cell types have worse autofluorescence. Have you tried pre-bleaching the sample with strong UV light for 30 minutes prior to staining?

There are some quenchers out there. Sudan black, True Black (Biotium), sodium borohydride etc.

It is odd that you are getting so much autofluorescence with the 647 laser, though. Any stain/antibody conjugate that you add should be very bright compared to most sources of autofluorescence in the Cy5 range. Have you tried staining with anything?

Hi! The main problem is that due to the protocol first I "stain" (incubate with a polymer that has a label and must attack on the surface) on lived cells and after that is when I fix them, so It I do that cells would be dead and I could not perform the experiment.

What are your fixation conditions? Buffer composition? Time? Temperature?

2% PFA, RT, 20 minutes approx in PBS 

i was thinking to quench the signal with NaBH4, often used in this cases

Though PFA can lead to increased autofluorescence, it's very odd to get autofluorescence out that far.  I agree that NaBH4 washing would be the first approach, but it's harder to do with cells in suspension.

I agree, that's other issue that I must confront. I loose most of the cells in washes

If the cells are already fixed you could try spreading them out on a gelatin- or PLL-coated cover glass and hope that it's sticky enough to hold them through the wash.

Should I put the cells over the cover during the incubation with pfa or after the first wash? 

What kind of microscope you are using? If it is filter-based, then it is good to exclude wrong fillter in a filter rack and siimilar basics. 

I assume the image shown here is from negative control, i.e. one without the polymer with your dye? If it is not, then you should check that it does give similar signal. If it does not, then you might just have a lot of free fluorophores not attached to your polymer so that they dominate the staining. How stable is their bonding to your polymer under the conditions you subject them to? 

We use a super resolution microscopy (STORM) which is more sensitive to fluorescence that a normal fluorescence microscope. With the polymer, which is label with Rhodamine, I can see exactly the same profile. In fact at first I thought that I did a mistake with the negative control and it was not the real sample.