The different tagging strategies have something to do with protein folding. Depending on the tertiary structure of your protein of interest, you should always think of cloning both constructs, C- and N-terminally tagged. This saves time and increases the likelikood of a successful job.

Imagine, your C-terminus will be fold inside your protein. In this case, you will never get a GFP signal.  In addition, when handling proteins which should be secreted, a tag at the N-terminus should be avoided.

The different tagging strategies have something to do with protein folding. Depending on the tertiary structure of your protein of interest, you should always think of cloning both constructs, C- and N-terminally tagged. This saves time and increases the likelikood of a successful job.

Imagine, your C-terminus will be fold inside your protein. In this case, you will never get a GFP signal.  In addition, when handling proteins which should be secreted, a tag at the N-terminus should be avoided.

Location of tag depends on the end use of the constructs/protein and the folding nature of your protein. Some things to consider when tagging a protein, I believe, are:

1. Why do you need the tag: If the purpose is to see protein expression, what method are you using to assay expression. I frequently use both Westerns and flow and using N- or C- terminal tagging never affected the assay. If you need GFP as a reporter for translation you should consider IRES dependent GFP expression. This will retain GP in the cytoplasm and so your protein is not hindered by GFP, i.e. for subsequent assays. Some tags like MBP are used to facilitate protein folding, if this is the case, clone your construct with N-terminal tag.

2. What is the end goal for the expressed protein? if you need the protein to assay binding, you should consider steric hindrance by the tag. It might interfere with your actual experiment

3. Is your protein secreted, cytoplasmic or membrane bound? Depending on the localization of your protein  you could See if you need the tag for the final goal of the experiment. This way you could put a protease cleavage site to remove the tag after expression.

Hope this is helpful

Often, the goal of GFP tagging is to visualize the normal localization of a functional protein.  The location of the tag can impact the localization and/or function of the protein of interest. For example, proteins with n-terminal nuclear localization signals will have altered localization if the GFP is fused to the n-terminus of the protein.

Just to add another example - if your protein is processed by cleavage, your tag might come off if it's on the side that gets cleaved off. You can also use this to track cleavage of the protein if that is your goal. For example, my group have tagged a particular enzyme on the N-terminal side with HA and on the C-terminal side with EGFP. Depending on where the protein is going in the cell (Golgi/ER, plasma membrane), either the N-terminal or C-terminal side is removed and this differential tagging allows us to verify correct protein processing. Hope this helps a bit.

In addition to the structural/signaling reasons already noted, C-terminal fusion might be preferred in some cases since translation rates can be influenced by 5' mRNA secondary structure. So if you're tagging the chromosomal copy of a gene, C-terminal fusion might avoid altering expression levels.

Related to Maurice Van Steensel's comment, N- and C-termini might be located in different intracellular compartments with different pH, etc that could influence fluorescence.

A very good discussion

GFP tagging is to normal localization of protein

If all constructs looks the same, you might believe to look at the real thing!

GFP at N and C can make a difference depending upon the folding of your protein, sometimes the GFP gets buried inside the protein and is not available to be detected by the antibody. Both N or C may work or sometimes only one, so it is good to check both.

GFP is a great tool to assess dynamically by videomicroscopy the localization of your protein of interest. It can also be used when antibodies are not avalaible to determine where a protein is targeted into the cell. The drawback is that the folding and thereby the function and/or the localization of the protein of interest might be altered. That being said, the observation of two different localization for a protein tagged in NT or CT might not necessarily argue that one is false. (an exemple is provided in this paper: "Apoptotic activity of a nuclear form of mitogaligin, a cell death protein. Biochemical and Biophysical Research Communications 01/2009; 378(4):816-20".).

Another consideration to be taken into account when tagging a protein with GFP is that GFP fluorescence decreases with pH. This can be important when tagging proteins that compose the membrane of the lysosomes. In that case GFP should be added to the extremity of the protein exposed to the cytoplasm and not to the lysosomal lumen. Noteworthy, this property of GFP fluorescence can be exploited to quantify lysosomal function, or autophagy for exemple.

I wish you the best Daniel,