Depending on what you want to do. In example they work quite well for the N- or C-terminal modification of proteins. Could you specify your question please?

I want to ligate two domains of a protein; one domain labeled, one domain unlabeled and ligate them together. I would like to produce both domains recombinantly.

This should not be to hard to Do. Most inteins will leave behind a short linker sequence, which will remain in the protein. There are inteins as the acel terl intein or the ssp dnab m86 which are able to ligate to fragments in a quantitative manner. The solubility of the two inteins mentioned is quite good as well.

You need to be aware of the sequence dependence. In contrast to expressed protein ligation there is a sequence dependence on the terminal residues from the N-terminal part as well. If you need to exactly produce a specific sequence, and cannot change that it may become difficult or impossible. If you are lucky with the sequence it will work well...If you link two domains you can likely adjust the sequence in the linker part so that it matches the sequence of the native substrate, and then it should work!

Thanks JM and OZ. So far I see that the dependence at the C-term is CFN. I am hoping a SGV sequence will work as that will be close to what I need. The N-terminal limitations are not clear. Are there any published results on the N-term requirement for each of these inteins?

I could use EPL but from what I am reading this protocol is much slower than the split intein protocol and it leaves a Cys residue also.

I am not an expert on this, but I believe that SGV instead of CFN will not work. If this is the linker, why does the sequence play a role at all? You will likely have much more success with the native sequence!

Using the M86 intein the linker could work I guess. This intein leaves behind a serine residue (the +1 position). VG as +2 and +3 could work, although there is no information regarding these two positions so far (afaik).