I used a CMV-ßGal vector to normalize expression of luciferase reporter gene constructs back in the 90s. Unfortunately, CMV-ßGal expression was quite variable in unrelated experiments. Finally we replaced CMV-ßGal by an EF1-alpha-ßGal construct which gave similar results in independent experiments. A colleague told me that they also had problems when they tried to use the CMV promoter to drive the expression of resistance genes in targeting vectors and they've shifted to the human and murine PGK promoters .

I showed in a 2001 paper in JBC (276:31978-85) that the CMV promoter functions poorly in cultured hippocampal and sympathetic neurons, whether the promoter is transduced via adenoviral transfer or is in the context of chromatin. Interestingly, we found that the CMV promoter can be strongly induced in neurons by stimuli, such as neuronal activity, that activate the transcription factor CREB: this is mediated via 5 CREB binding sites in the promoter. Often, people tell me they have good success with CMV when they transfect neurons with traditional techniques (lipofectamine, Ca2+ phosphate); however, this likely stems from activation of the promoter when the cell membrane is disrupted during transfection. Indeed, if one depolarizes hippocampal cultures days after transfection, they will induce expression from the CMV promoter in many cells that otherwise would appear non-transfected, revealing that the transfection efficiency is much higher than thought. Also, CMV might function well when delivered by certain viruses due to cross-talk from viral genes activating CMV. Finally, we showed in a subsequent paper (Mol Cell Neurosci. (2004)26(1):50-62) that ongoing synaptic activity in culture produces robust expression from CMV through the CREB binding sites. Thus, expression seen in neurons in vivo may be related to the level of activity, and not simply whether or not a particular neuron is infected. The bottom line is that, under most circumstances, the CMV promoter should be avoided when working with neurons, something that is, unfortunately, often not recognised.