nability of glibenclamide to reach therapeutic levels in the brain whereas the partial restoration of the LOWR argues that it may involve pathways outside the BBB, such as primary afferent neurones. Administration of glibenclamide via into the lateral ventricle of nV59M and control littermates via an osmotic mini-pump had no effect on anaesthesia sensitivity. The most likely reason for this is that glibenclamide did not reach a high enough level in the brain, as we were unable to measure any quantifiable drug concentration when glibenclamide was administered this way. Unfortunately, we could not increase the drug concentration in the mini-pump solution because it was not soluble at higher concentrations. Thus, we could not determine whether the drug might have been effective at higher doses. Bolus administration of a suprapharmacological dose of glibenclamide led to INK1117 detectable levels of the drug in the CSF and brain, but this was too transient and too short to enable the animal to recover and be tested for anaesthesia sensitivity. Consequently, we cannot distinguish if a lack of a change in anaesthesia sensitivity is due to too low a drug dose, or because of the absence of KATP channel activity during development produced irreversible changes in neuronal function. Fig 8. Effect of intracranioventricular glibenclamide on the sensitivity to isoflurane. Time taken for the LORR before and one week after ICV infusion of vehicle or glibenclamide in nV59M and control mice. Time taken for the LOWR before and one week after ICV infusion PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19755095 of vehicle or glibenclamide in nV59M and control mice. Representative Nissl-stained coronal section of mouse brain showing the site of the ICV cannula.. Data are meanSEM. n.s. not statistically significant. doi:10.1371/journal.pone.0134476.g008 14 / 18 Glibenclamide Administration Fails to Reach Effective Levels in Brain Implications for therapy While it is difficult to directly compare our results and data from patients treated with sulphonylureas, because glibenclamide was administered via different routes and there are pharmacokinetic differences between species, our data show that if the PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/19755912 plasma concentration is high enough, glibenclamide may reach the brain and CSF. However, this is only 
achieved with plasma glibenclamide concentrations at which the drug also blocks other types of ion channels, and thus it may have off-target effects in peripheral tissues. It therefore seems unlikely that such high drug doses would be tolerated in patients. The subcutaneous glibenclamide concentration we administered in rats was ~17 times the maximum oral dose taken by patients with neonatal diabetes. Plasma levels in ND patients are also 330-fold lower. Thus it seems likely that glibenclamide levels in the brain will be very low in patients with neonatal diabetes. Although 10ng/ml glibenclamide blocks wild-type KATP channels by ~80%, it is much less effective on mutant KATP channels. Furthermore, it is the absolute magnitude of the KATP current that matters functionallythis is difficult to quantify as it will also depend on channel block by intracellularly generated ATP. Thus it is not a simple matter to predict KATP current magnitude in vivo. Nevertheless, our data show that even when plasma glibenclamide levels are very high, no effect of the drug is observed on anaesthesia sensitivity. This may be important if, like the mouse model, DEND patients experience reduced sensitivity to inhalation anaesthesia as they may be at r