Overcompensation! Do you have a granulocyte marker in another fluorochrome? Do you want to check your compensation values in the real world of your fully stained sample? There's a good story that tells: always put in a bivariate plot all stains one vs another and let them talk!
Hi, just saw your question. Overcompensation is not the only way to get this result. It could be a baseline restore issue. I presume this was run using DiVA? Have a look at the uncompensated Cy5. Does the negativity go away? If so, it's a comp issue. If not, you may have created an artefact on the machine. This wouldn't be a big surprise since you're using nanoparticles. I used to use qdots to reproduce the effect when we were looking at it a few years ago. Also recommend that you look at Cy5 vs. time. It's possible it's just some end-of-run weirdness from running the sample dry.