Be careful with PC12. They are from pheocrhomocytoma and from rat origin. The others are neuroblastoma of human origin. SK-N-SH are the parental of the SH-SY-5Y. There are more extensive work in SH-SY-5Y and I think that they are the best characterized in terms of neuronal differentiation. IM32 are also quite neuronal, but I don't have experience with them.
IMR-32 is considered a human cholinergic neuroblastoma cell line; while retinoic acid induces human neuroblastoma SH-SY5Y to differentiate towards cholinergic neuronal phenotype. SK-N-SH seems to lack tyrosine hydroxylase and acetylcholine. PC12 cells are a rat pheochromocytoma cell line. I suggest to read the following paper because you can consider to use another neuroblasoma cell line such as LA-N-1 and LA-N-2:
We have had excellent success with SH-SY5Y and retinoic acid insofar as neurite extension is concerned. However, the markers expressed by this line after "differentiation" indicate that it does not become "normal neurons". PC12 is, as said before, not a neuroblastoma line, but it is probably the most commonly used culture model for cholinergic "neurons".
We have used all of these cells and find that SH-SY5Y cells are the best from several points of view, but they do not have a robust cholinergic phenotype even when differentiated. Both IMR32 and LAN-2 cells have cholinergic features, but LA-2 cells are much more difficult to grow. None of the cells are easy to transfect so it depends what you are trying to do with them and what you want as read-out measures.
Agree with the warning about PC-12 cells. They have their uses, but they are highly variable across laboratories, and even within cultures. They can phenotypic variations in the degree of neurite outgrowth in the presence and absence of retinoic acid, NGF or other factors. Subclones can have widely varying levels of neurotransmitters. So use with caution. On the other hand, SH-SY5Y cells, as mentioned, are a good line.
If you are not dependent on neuroblastoma cells, NTERA2 would be a good choice. These cells are a neural line from embryonal carcinoma / teratoma. Treatment with retinoic acid results in very good neurite outgowth.
We induced neuronal differentiation of SH-SY5Y cells by sequential treatment with retinoic acid and BDNF. In terms of neurite formation, it worked nicely. I have also used BrdU for cholinergic differentiation of mouse Neuro2a cell lines and it gave a favorable result.
Having used SH-SY5Y and NTERA2 cells the key is the growth and differentiation conditions. Over time after differentiation there can be changes in responsiveness and media supplements play an important role. I favour the SH-SY5Y's only because they are easier to handle and work for the compounds I'm interested in. Would suggest choice depends on end point measures. With regards transfection I've had little difficulty with SH-SY5Y's and stable cell line generation but transients are tempermental.