I have a protein that I would like to mislocalize to the plasma membrane in early embryos. Ideally, I would like to have as little of the protein as possible in the cytosol or on intracellular membranes. Can anyone suggest a good tag or protein fusion for doing this? The best I know of is the rat PLCdelta PH domain, but since that domain just binds peripherally to PIP2, I worry that there is likely a sizable pool of protein in the cytosol. I also thought about a CAAX box, which is appealing because the lipid anchor should be pretty much exclusively found in membranes - but from the couple of papers I looked at, it appears that there is a lot of GFP::CAAX on intracellular membranes.
There are a few transmembrane options available. In the original GRASP paper from the Bargmann lab, they fused half of a split GFP to CD4 and used that as a membrane marker. I don’t know if they’ve made a full-length GFP or mCherry version. One advantage of CD4 is that it’s monomeric. Based on work from Liqun Luo, I’ve developed a mouse CD8 fusion to GFP or mCherry. I had to mess around with the linker a little bit, but the protein expresses very well and localized to the plasma membrane. I haven’t looked in single cells to determine if there’s significant intracellular localization. Of note is that mCD8 is an obligate homodimer. That means it should express well and be stable, but you wouldn’t want to go making fusion proteins to it and try to co-express with another version. One advantage of CD4 and CD8 is that there are mouse monoclonals to the extracellular domain that you can use if you’re doing immunostaining or western blots. I’ve used this construct for quantitative fluorescence ratio analysis (live imaging of GFP fusions and mCD8-mCherry), and the construct expresses very well in neurons and muscles. The mCherry version has some punctate structures that aren’t nearly as evident with the GFP version, but your mileage may vary. Because it’s a transmembrane protein, it’s going to be much dimmer than you’re used to, but it’s perfectly bright from a high-copy transgene. Expression via bombardment or MosSCI will make it much dimmer. Let me know if you’re interested, and I’ll see about how to get it to you.
There is a transmembrane GFP protein in the Fire vector kit that has been used, but it doesn’t work nearly as well as mCD8.
That said, the PH domain should be largely on the plasma membrane and not on intracellular vesicles and will be bright. But if you’re looking at processes that could be affected by Phospholipase C signaling and PIP2 hydrolysis, you might consider mCD8 instead.
Dan, you are right to be concerned about CAAX sequences. Few have been characterized well in elegans, and in mammals many cause targeting to multiple subcellular compartments. Plus, those compartments can be altered with context.