IP3 accumulation and/or inositol depletion: two downstream lithium's effects that may mediate its behavioral and cellular changes

Lithium is the prototype mood stabilizer but its mechanism is still unresolved. Two hypotheses dominate—the consequences of lithium's inhibition of inositol monophosphatase at therapeutically relevant concentrations (the ‘inositol depletion' hypothesis), and of glycogen-synthase kinase-3. To further elaborate the inositol depletion hypothesis that did not decisively determine whether inositol depletion per se, or phosphoinositols accumulation induces the beneficial effects, we utilized knockout mice of either of two inositol metabolism-related genes—IMPA1 or SMIT1, both mimic several lithium's behavioral and biochemical effects. We assessed in vivo, under non-agonist-stimulated conditions, 3H-inositol incorporation into brain phosphoinositols and phosphoinositides in wild-type, lithium-treated, IMPA1 and SMIT1 knockout mice. Lithium treatment increased frontal cortex and hippocampal phosphoinositols labeling by several fold, but decreased phosphoinositides labeling in the frontal cortex of the wild-type mice of the IMPA1 colony strain by ~50%. Inositol metabolites were differently affected by IMPA1 and SMIT1 knockout. Inositoltrisphosphate administered intracerebroventricularly affected bipolar-related behaviors and autophagy markers in a lithium-like manner. Namely, IP3 but not IP1 reduced the immobility time of wild-type mice in the forced swim test model of antidepressant action by 30%, an effect that was reversed by an antagonist of all three IP3 receptors; amphetamine-induced hyperlocomotion of wild-type mice (distance traveled) was 35% reduced by IP3 administration; IP3 administration increased hippocampal messenger RNA levels of Beclin-1 (required for autophagy execution) and hippocampal and frontal cortex protein levels ratio of Beclin-1/p62 by about threefold (p62 is degraded by autophagy). To conclude, lithium affects the phosphatidylinositol signaling system in two ways: depleting inositol, consequently decreasing phosphoinositides; elevating inositol monophosphate levels followed by phosphoinositols accumulation. Each or both may mediate lithium-induced behavior.


Fig. S1: The effect of postmortem interval (A), Li dose (B) and extent of IMPA1 KO (C) on 3 Hphosphoinositols accumulation
A. Dissected brains of WT mice were left at room temperature for 0, 30, and 60 minutes prior to the procedure for assaying phosphoinositols accumulation. PMI upto one hour did not affect phosphoinositols accumulation. injected ICV (as above for 'Acute Li administration'). About half of the animals of each group received a subconvulsive dose of 100 mg/kg pilocarpine (Sigma-Aldrich, St. Louis, MO, dissolved, 10 mg/ml in saline) subcutaneously (s.c.) 15 minutes prior to sacrifice and brain extraction. Saline was injected in the control animals. Acute pilocarpine administration increased phosphoinositols labeling in whole brain and in combined cortex and hippocampus (3). We now investigated (a) whether different brain regions, i.e. frontal-cortex and hippocampus, strongly implicated in mood regulation (4), respond in a similar way to pilocarpine-induced phosphoinositols accumulation; (b) whether Li also hypersensitizes this neurochemical effect of pilocarpine; and (c) whether IMPA1-KO mice mimic the effect of Li treatment in their response to pilocarpine. None of the animals showed seizures prior to the time of sacrifice. In the frontal cortex both Li-treated and IMPA1-KO mice showed a nonsignificant ~40% trend towards increased phosphoinositols accumulation (Supplementary Fig. 2A).
Pilocarpine further increased phosphoinositols accumulation (280% as compared with WT mice and 210% as compared with Li-treated mice) with no effect on the IMPA1-KOs (Supplementary Fig. 2A).
In the hippocampus Li treatment and IMPA1-KO increased phosphoinositols labeling as compared with WT mice (200% and 70% increase, respectively, Supplementary Fig. 2B) while pilocarpine had no effect ( Supplementary Fig. 2B) suggesting brain region differential response to agonist stimulation (5). Alternatively, the different results in the frontal-cortex and hippocampus might stem from a different distribution of muscarinic receptors in these brain regions (6). Despite the similarity between IMPA1-KO mice and Li-treated mice in the hypersensitivity to pilocarpine-induced seizures (13) the IMPA1-KO mice did not mimic Li treatment regarding phosphoinositols accumulation in response to pilocarpine stimulation. Namely, pilocarpine did not induce further phosphoinositols accumulation neither in the fontal cortex nor in the hippocampus. It is notable that a similar quantitative difference between Litreated mice and IMPA1-KO mice in their response to a sub-convulsive dose (100 mg/Kg) of pilocarpine was also observed at the behavioral level. All Li-treated mice seized following pilocarpine administration whereas only 60% of the IMPA1-KO mice seized (7). according to the manufacturer's instructions with slight modifications as we have previously described (8,9) to avoid spontaneous seizures induced by the liposome suspensions. Reconstituted liposomes were mixed 1:5 with either aCSF or IP 3 followed by a short sonication for 5 sec using a Polytron cell homogenizer (Heat System Ultrasonic, Newtown, CT, USA) at 50% power capacity. Mice were tested in the FST 45 min following IP 3 administration (Supplementary Fig. S1A). Based on the results of the dose-response experiment we selected the 150 µg IP 3 dose for later experiments.
We also carried out a dose-response experiment for the effect of 5, 10 and 30 pmoles of xestospongin-C in the FST. As previously reported (10), 30 pmoles xestospongin-C administered ICV in liposomes 45 min prior to the test reduced the immobility time ( Supplementary Fig. S3B). Based on the results we selected the 10 pmole dose for later experiments because it was the highest dose that did not have an effect in the FST.  Fig. 4). For the open field test mice were placed individually in monitor boxes (37.5x37.5x45cm) and their behavior digitally reordered.
Distance moved was analyzed using the Noldus EthoVision system (Wageningen, the Netherlands).
Neither IP 3 nor IP 1 affected motor activity in the open field test (Supplementary Fig. 4A, B). For the beam walking test mice were placed on a beam at the end of which there was food and their home cage. We measured how long it took the mice to get to the food. The test ended when the mice reached the food or if they fell from the beam. Neither IP 3 nor IP 1 altered the balance of the mice in the beam walking test (Supplementary Fig. 4C, D). For the hanging-wire test each animal was placed on a metal grid that was gently shaken to cause a grabbing reflex, and slowly turned upside down. Latency to fall was counted during a session of 1 min, as previously described (11). IP 3 but not IP 1 decreased their strength in this paradigm ( Supplementary Fig. 4E, F). Hence, non-specific behavioral effects of IP 3 were ruled out because of the lack of the compound's effect in the open-field and the beam walking tests. The significant effect of IP 3 in the hanging wire test was disregarded since it was counterintuitive with respect to the FST results in which the mice were more, rather than less, mobile as compared with the control group.