I agree with Christoph, that it's one thing to show colocalisation and another to demonstrate a functional interaction. That said, we currently use HA tag (YPYDVPDYA) to tag a family of potassium channels that are in low abundance in the plasma membrane. The tag is relatively small and pretty straightforward to introduce by mutagenesis. We have placed the tag in an external facing loop, where it hasn't disturbed trafficking of the target proteins. It is well-recognised by a commercially available monoclonal antibody and we've been able to use confocal and epifluorescence microscopy to look at colocalisation. In the past, I have had success tagging membrane proteins with both 6xHis and Strep tags, placing them in C-terminal and internal locations. The tags were recognised by commercial antibodies; however the internal location did disturb some important protein-protein interactions, so there were functional consequences (some of them helpful for elucidating structural interactions, in fact). I think you will always have to bear in mind that tags can alter trafficking and function, so you may need to have more than one attempt.

Halo Tags from Promega work very well and you can get different colour dyes that are either cell permeable or non-cell permeable.

I agree with Christoph, that it's one thing to show colocalisation and another to demonstrate a functional interaction. That said, we currently use HA tag (YPYDVPDYA) to tag a family of potassium channels that are in low abundance in the plasma membrane. The tag is relatively small and pretty straightforward to introduce by mutagenesis. We have placed the tag in an external facing loop, where it hasn't disturbed trafficking of the target proteins. It is well-recognised by a commercially available monoclonal antibody and we've been able to use confocal and epifluorescence microscopy to look at colocalisation. In the past, I have had success tagging membrane proteins with both 6xHis and Strep tags, placing them in C-terminal and internal locations. The tags were recognised by commercial antibodies; however the internal location did disturb some important protein-protein interactions, so there were functional consequences (some of them helpful for elucidating structural interactions, in fact). I think you will always have to bear in mind that tags can alter trafficking and function, so you may need to have more than one attempt.

FLAG and Myc tags are extremely negatively charged and might affect the function of you protein in the membrane because lipids can also be negative charged. Less dominant are the positively charged His-tags. Even better is a non-charged tag, Maybe HA-tag or Strep tag

You may be interested in tagging the extracellular part of your protein of interest with the alpha-bungarotoxin binding site and using different alpha-bungarotoxin (BTX) conjugates (different AlexaFluor available from Molecular Probes) to visualize either total protein expression, cell surface localization, internalization, or membrane insertion of your membrane protein. This tool is very useful for live cell imaging as it does not require secondary antibody to visualize the tagged-protein. This method was developed by the lab of Richard Huganir (The Johns Hopkins University). Here is a PubMed link to a publication describing the method where you can find links to the free full text: http://www.ncbi.nlm.nih.gov/pubmed/15563595

We used V5-tag to a lot and it is very good for other IF or WB detections. Attention should be paid to the location of the tag insertion. If not suitable to add at termini, you can put into other region of protein, the location where it would be on surface and less interruptive to the secondary structure.

We've used the V5, 6xHis, c-myc, and HA tags noted already, as well as some fusion fluorophores (GFP, mCherry, dsRed) to look at a variety of transmembrane proteins; groups have also used tetracysteine tagging for transmembrane protein localization, to good effect.

The problem is, the tag's effects on localization (not to mention function) is often dependent on the nature of your protein. Your best bet is to see what groups working with similar proteins to those you want to study use for their own work, and start from there. We've noted incorrect localization due to tag (eg, failure to undergo proper trafficking) as well as altered distribution (particularly with clumping in the case of the fluorophore fusion, but also with some of the tags under some conditions). So, your best bet is to start with what seems to work from the literature for your protein family, and go from there. (Also, how will you know if distribution is altered if you add a tag? Controls will be essential!)

Another responder noted the difference between co-localization and interaction. Transmembrane proteins have been visualized effectively using BiFC. So, this might be an alternative, perhaps more informative, approach to address your need.

I don't know what kind of protein expression you are getting but most membrane proteins express a levels well below 1 mg/L so GFP tags have been used because expression can be monitored and optimized even if the starting yield is low because GFP fluorescence is very sensitive and is able to detect low concentrations. The tag can subsequently be cleaved. It is thought that expression can be hampered if the first 50 residues don't fold during expression on the the ribosome. I thing with a GFP on the N-terminal you have a better chance of expressing the membrane protein.

For clarity, the protein is an ER protein, not a plasma membrane protein. Thanks for all the great responses though!

you can use GFP as a marker, there are commercially available antibodies that work fine, and it has the advantage that you can observed the protein of interest with any fluorescence microscopy technique.

We found that HA tag makes the best tag especially for an external epitope due to the availability of good monoclonal antibodies and low background in cells like HEK or MDCK. FLAG tags work well at the N-terminal, but often we use triple tags to increase the sensitivity of detection. As a first go, i would always tag at the C-terminus using HA, Myc or FLAG - YFP is good especially in live cells, but is weaker). HA is always our #1 choice.

Just to add my two cents. Whatever tag you use, its success will depend on it not intefering with the protein function and its transport machinery. So it is a balancing act of clever engineering of the chimeric protein and use of right tag(s). A search of literature to see what others have used successfully in the past for similar experimentation will be my starting point. I hope this helps.

A lot of possibilities have been mentioned already to study colocalization (which ideally is done by confocal LSM microscopy including deconvolution). If you want to go not only for colocalization but close interaction between two (membran) proteins, the application of FRET microscopy is a good choice. I did this by co-transfecting my membrane protein of interest with C-terminal CFP and YFP (of course those "tags" are much bigger than HA, FLAG, etc. but they enable you to do live cell imaging). Special microscope equipment to do FLIM-FRET measurements is available from PicoQuant (http://www.picoquant.com/) for example.

some small tag may be candidate. I used GFP as a tag and found that amount of green fusion protein existed in cytoplasm when GFP was fused at c-terminal of our membrane protein.

I have used fusions with green fluorescent protein (GFP) and red fluorescent protein (RFP) in order to see co-localization in yeast of two different proteins by fluorescence microscopy. It seemed to work fine, however, in my hands, the position of the fluorescent protein is an empirical issue, thus, one of the proteins "preferred" to have the GFP at the N-terminus whereas the other didn't express unless the RFP was at the C-terminus. By the way, there are also yellow and blue fluorescent proteins available, I'm not sure if all of them have specific antibodies, I guess they have