An in vivo model system to study the impact of Li on PLC-induced PIP₂ turnover

To study the turnover of PIP₂ after receptor-activated PLC turnover, we created a cell line which heterologously expressed two components. To stimulate high levels of PLC activity, we expressed the human muscarinic acetylcholine receptor (m1AchR) encoded by the gene CHRM1; previous studies have shown that activation of heterologously expressed m1AchR results in high rates of PIP₂ hydrolysis leading to IP₃ production (Parker et al, 1991; Raghu et al, 1997). To monitor the levels of PIP₂ at the plasma membrane, we expressed the PH domain of PLC-δ fused to GFP (PH-PLCδ::GFP). This probe binds to PIP₂ and in imaging experiments; its relative distribution between the plasma membrane and the cytosol has been used by us and others to monitor plasma membrane PIP₂ levels (Várnai & Balla, 1998; Chakrabarti et al, 2015). The PH-PLCδ-GFP probe is a sensitive reporter for PIP₂ but it also binds to IP₃; there is another probe –Tubby c-terminal domain that is also employed as a sensor for PIP₂. However, the Tubby c-terminal domain has exhibited lower sensitivity to report on changes of PIP₂ during PLC activation, although being more specific in its affinity towards PIP₂ (Szentpetery et al, 2009). Furthermore, a very recent study has noted that in contrast to PH-PLCδ-GFP probe, the Tubby domain binds selectively to certain domains of the plasma membrane at membrane contact sites making it not an optimal detector of PIP₂ levels across the plasma membrane (Thallmair et al, 2023). Due to these reasons, we preferred to use the PH-PLCδ-GFP probe as the sensor for PIP₂.

A cell line expressing both m1AchR and PH-PLCδ::GFP can therefore be used to stimulate PLC and monitor the turnover of PIP₂ at the plasma membrane. We generated stable HEK293T cell lines expressing m1AchR (control-M1) and PH-PLCδ::GFP was transiently expressed in those cells (Fig 1B); when stimulated with the m1AchR agonist carbamoylcholine (carbachol: Cch), the levels of PH-PLCδ::GFP at the plasma membrane drop after which they gradually recover back towards pre-stimulation levels (Fig 1C). The changes in plasma membrane PIP₂ was quantified as the ratio of fluorescence intensity of the PH-PLCδ::GFP probe at the plasma membrane to its fluorescence intensity in the underlying cytosol in the same part of the cell. As a positive control for our assay to monitor PIP₂ regeneration after PLC activation, we used shRNA to deplete Nir2; Nir2 (PITPNM1/RdgBαI) encodes for a phosphatidylinositol transfer protein required for the resynthesis of PIP₂ after PLC stimulation (Kim et al, 2013, 2016; Chang & Liou, 2015). Under our experimental conditions, we found that depletion of Nir2 in our HEK293T reporter line (Fig S1A) resulted in a decreased rate in the recovery of plasma membrane PIP₂ levels after stimulation with carbachol (Figs 1D and E and S1B).

To measure the effect of Li treatment on PIP₂ resynthesis, we incubated the reporter line with 1 mM Li; this concentration of LiCl was used in our experiments because the therapeutic range for Li medication in BD patients was 0.75–1.5 mEq/litre (Francis et al, 2004). After Li treatment, we stimulated the cells with carbachol and compared the dynamics of plasma membrane PIP₂ levels with that in non-Li–treated cells. We found that in Li-treated cells, the rate with which PIP₂ levels at the plasma membrane recovered to pre-stimulation levels was significantly slower than in untreated controls (Fig 1F and G). These findings strongly suggest that treatment with Li results in a reduction in the rate of PIP₂ resynthesis after PLC stimulation.

In an alternative approach, we quantified total cellular PIP₂ mass using liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) at three distinct time points—corresponding to the basal state, one-minute post-stimulation of PLC via carbachol addition (hydrolysis of PIP₂) state, and post-stimulation (recovery of PIP₂) state. From these experiments, the time for recovery of PIP₂ (to pre-stimulation levels) was standardized at 15 min post stimulation with carbachol (Fig 1H). Using these conditions, we compared the total PIP₂ mass between untreated cells and those treated with Li. We found that the mass of PIP₂ during the recovery phase was lower in Li-treated cells compared with controls (Figs 1I and S1C); likewise, the level of PIP in Li-treated cells was also lower during the recovery phase (Figs 1J and S1D). Thus, Li treatment lowers the recovery of PIP and PIP₂ levels after PLC activation. PIP refers collectively to all the isoforms of PIP: PI3P, PI4P, and PI5P. These cannot be individually distinguished by mass alone because their mass is identical. However, preexisting literature has shown that PI4P is the most abundant isoform of PIP present in cells; its level is ∼50-fold higher than that of PI5P (Rameh et al, 1997). Hence, we can infer that the change in the total mass of PIP observed by LC-MS/MS is mainly reflective of the levels of PI4P.

Chronic Li treatment reduces PLC-dependent intracellular Ca²⁺ mobilization

A key outcome of agonist-triggered, PLC-mediated PIP₂ hydrolysis is intracellular Ca²⁺ signalling (Berridge, 2009). Because we noted a reduction in the turnover of PIP₂ after Li treatment, we tested the impact of Li treatment on intracellular Ca²⁺ levels [Ca²⁺]_i. Using the control-M1 cell line, we monitored [Ca²⁺]_i under both resting conditions and stimulation with carbachol, an agonist of m1AchR. Cells were incubated in 1 mM Li before the experiment: basal [Ca²⁺]_i were no different between treated and untreated cells (Fig 2A). Stimulation of untreated cells with carbachol evoked a rise in [Ca²⁺]_i which then decayed back to resting levels (Fig 2B). In cells treated with Li, the carbachol-induced rise in [Ca²⁺]_i was reduced compared with that seen in untreated cells (Fig 2C). Receptor-activated PLC-mediated Ca²⁺ signalling is a biphasic process with two components, an initial release of Ca²⁺ from intracellular stores followed by store-operated influx of Ca²⁺ into the cell from the extracellular medium (Prakriya & Lewis, 2015). We performed experiments to separately monitor each of these components. During the initial phase of the assay, carbachol stimulation was performed in the presence of zero extracellular Ca²⁺, thus selectively monitoring the release of Ca²⁺ from intracellular stores (Fig 2D and E). After this, the extracellular solution was supplemented with Ca²⁺ and the influx of Ca²⁺ into the cytoplasm was monitored (Fig 2D and F). This analysis revealed that in Li-treated cells, both intracellular Ca²⁺ release (Fig 2E) and the subsequent store operated Ca²⁺ influx into cells (Fig 2F) was reduced compared with Li-untreated controls. This reduction in receptor-activated release of Ca²⁺ from intracellular stores was not because of a reduction in the size of intracellular Ca²⁺ stores as emptying of stores by the application of thapsigargin resulted in equivalent rises in [Ca²⁺]_i in Li-treated cells and untreated controls (Fig 2G and H).

According to the “inositol depletion hypothesis” (Berridge et al, 1989), Li can restrict the available cytoplasmic inositol pool available for phosphatidylinositol (PI) (and thence PIP₂) resynthesis by inhibiting IMPase—thereby, lowering down the synthesis of the pool of PIP₂ that is required for PLC activity. This cytoplasmic inositol pool can also be replenished by transport of inositol from the extracellular medium via plasma membrane inositol transporters (summarized in Balla [2013]). This implies that if a deficit in the available pool of cytosolic inositol is responsible for the attenuated calcium signalling in Li-treated cells, this might be rescued by supplementation of the extracellular medium with inositol. The standard concentration of inositol in DMEM High Glucose (DMEM; Life Technologies) is 40 μM (7 mg/litre). We grew HEK293T cells in an inositol-rich DMEM media; inositol concentration was ca. 28 mg/litre; which is similar to the inositol concentration in the cerebrospinal fluid (Nixon, 1955; Swahn, 1985; Shetty et al, 1996). For our inositol-rich DMEM, 117 μl of 100 μM of inositol (18 mg/ml) was added to 100 ml of DMEM High Glucose , supplemented with 10% FBS—this led to the final effective concentration of inositol in the media ∼155 μM (28 mg/litre). This media were referred to as the inositol-rich media and used for the inositol supplementation experiments. These cells were subjected to Li treatment and agonist induced Ca²⁺ influx was quantified. As controls in this experiment, we used Li-treated control-M1 cells grown in DMEM not supplemented with inositol. As in previous experiments, we found that when cells were grown in DMEM, Li treatment resulted in a reduction of agonist-induced Ca²⁺ influx (Fig 2I and J). However, when cells grown in inositol-supplemented DMEM were treated with Li, the Li-induced reduction in Ca²⁺ influx was rescued compared with that seen in cells grown under equivalent Li treatment but in DMEM not supplemented with inositol (Fig 2I and J), compared with Li non-treated cells. We also observed that the decreased rate of PIP₂ resynthesis post PLC stimulation in Li-treated cells was rescued by supplementation of inositol in the medium (Fig S2D).

IMPA1 is required for agonist-mediated calcium signalling and PIP₂ resynthesis

IMPase, the enzyme that dephosphorylates inositol monophosphate to generate myo-inositol has been proposed as a direct target of the action of Li, inhibiting its activity (Hallcher & Sherman, 1980; Saudek et al, 1996). There are two genes encoding IMPase in the human genome IMPA1 and IMPA2 of which IMPA1 is the predominant isoform expressed in HEK293T cells. To test the requirement of IMPA1, we generated a HEK293T cell line in which IMPA1 was deleted by CRISPR/Cas9 genome editing (Fig S3A); m1AchR was expressed in these cells by lentiviral transduction; these cells will be denoted as IMPA1^−/−M1. Western blot analysis of IMPA1^−/−M1 showed the complete absence of the IMPA1 protein (Fig 3A), IMPA2 transcript levels were not altered (Fig S3B) and the levels of m1AchR protein were comparable between control–M1 cells and IMPA1^−/−M1 cells (Fig S3C and D). We compared the response of IMPA1^−/−M1 cells with control–M1 cells. When stimulated with carbachol, IMPA1^−/−M1 cells showed a reduced Ca²⁺ influx response that could be rescued by reconstitution of the cells with the IMPA1 cDNA (Fig 3A–C). Further analysis under conditions of zero extracellular Ca²⁺ revealed that the release of Ca²⁺ from intracellular stores was reduced in in IMPA1^−/−M1 cells compared with control cells (Fig 3D and E) with an intact IMPA1 gene. We also noted a decrease in the subsequent store operated calcium entry (SOCE) in these IMPA1^−/− cells with respect to the control–M1 cells (Fig S3E). Lastly, we found that after carbachol stimulation, the rate at which PIP₂ levels were restored at the plasma membrane was slower in IMPA1^−/−M1 and this could be rescued by reconstitution with a WT IMPA1 transgene (Fig 3F).

IMPA1 is required for the effect of Li on receptor-activated Ca²⁺ influx and PIP₂ resynthesis

We tested the effect of Li treatment on IMPA1^−/−M1 compared with untreated IMPA1^−/−M1 cells. We analysed receptor-activated Ca²⁺ influx after Li treatment; this was not altered compared with untreated IMPA1^−/−M1 cells (Fig 3G and H); in the same experiment, treatment of control–M1 cells with Li resulted in a big reduction in agonist-activated Ca²⁺ influx. We also analyzed the two phases of receptor activated Ca²⁺ influx and found that whereas release of [Ca²⁺]_i from stores was not affected by Li treatment in IMPA^−/−M1 cells, the reduction of SOCE previously noted on Li treatment in control–M1 cells was still present (Fig 3I–K). We also found that the rate of regeneration of PIP₂ at the plasma membrane was unchanged in IMPA1^−/−M1 cells subjected to Li treatment, compared with untreated IMPA1^−/−M1 cells (Figs 3L and S3F and G). These observations suggest that the ability of Li to inhibit Ca²⁺ influx and slow PIP₂ resynthesis after PLC activation requires an intact IMPA1.

Li reduces excitability in human cortical neurons

To test the relevance of our findings to human cortical neurons, changes in whose activity presumably underlies the therapeutic effect of Li in bipolar disorder patients, we generated human forebrain cortical neurons from iPSC in vitro (Sharma et al, 2020). We differentiated forebrain cortical neurons from neural stem cells (NSC) derived from an Indian control iPSC line, D149 (Fig 4A) (Iyer et al, 2018). The generated NSCs expressed previously described markers characteristic of NSC such as Nestin, SOX1, SOX2, and PAX6 (Fig 4B) and when differentiated, expressed the neuronal marker MAP2; a proportion of cells in the culture expressed the glial marker GFAP (Fig 4C). When these NSCs were differentiated into cortical neurons, they exhibited spontaneous [Ca²⁺]_i elevations, referred to as transients, that increase as a function of age in vitro (Sharma et al, 2020; Akhtar et al, 2022). We measured somatic [Ca²⁺]_i transients from such neuronal cultures at 45 d in vitro (DIV45) and quantified their frequency. We observed that the frequency of [Ca²⁺]_i transients was decreased after Li treatment (1 mM) for 10 d (Fig 4D and E). We also measured changes in [Ca²⁺]_i after stimulation with carbachol in DIV45 cortical neurons treated with Li. In our cultures, carbachol stimulation results in a rise in [Ca²⁺]_i that peaks before gradually decaying towards the baseline; in Li-treated cultures, the peak of this response was reduced (Fig 4F and G). We also performed experiments to measure Ca²⁺ mobilization from internal stores; these revealed that carbachol-activated release of Ca²⁺ in Li-treated neurons was reduced compared with those that were untreated (Fig 4H–J). By contrast, there was only a modest difference in thapsigargin-mediated rise in [Ca²⁺]_i between Li-treated and -untreated cortical neurons (Fig S4A–C). These observations suggest that both neuronal excitability and PLC-mediated Ca²⁺ signalling is attenuated in human cortical neurons after Li treatment.

Inhibition of GSK-3β does not phenocopy the effect of Li on human cortical neurons

It has been proposed that the inhibition of GSK-3β by Li underlies its therapeutic effects in BPAD. To test this hypothesis, we treated HEK293T cells with the well-established pharmacological inhibitor of this enzyme- CHIR99021 (An et al, 2014). Treatment of control-M1 cells with 10 μM CHIR99021 for 24 h resulted in a robust elevation of β-catenin levels indicating potent inhibition of GSK-3β under these conditions (Fig 5A). Under these conditions, we found that GSK-3β inhibition did not result in reduced agonist-induced Ca²⁺ influx (Fig 5B and C). Likewise, in human iPSC-derived forebrain cortical neurons, the frequency of spontaneous Ca²⁺ transients (Fig 5D and E) and carbachol-activated Ca²⁺ influx was not altered by GSK-3β inhibition (Fig 5F and G). Lastly, after PLC activation and PIP₂ depletion, the rate at which PIP₂ levels at the plasma membrane were restored was not impacted by treatment of cells with the GSK-3β inhibitor (Fig 5H).

Li and GSK-3β inhibition induce distinct transcriptional responses

We also examined the transcriptional response of human iPSC derived forebrain cortical neurons to treatment with 10 μM CHIR99021 for 24 h. RNA-seq analysis revealed large changes in gene expression with ca. 5,000 genes each being up- or down-regulated (Fig S5C). GO analysis revealed a broad range of pathways including cardiomyopathy, amino acid metabolism, and extracellular matrix interactions (Fig 6F). Strikingly, there was no enrichment of pathways related to synaptic transmission or calcium signalling; the only brain-related pathway that was enriched was axon guidance and the identity of the genes picked up suggested ECM function. We overlapped the set of genes up- or down-regulated in Li and GSK-3β inhibition transcriptomes to assess the extent of overlap between the two gene sets and found that, both in the domain of up- and down-regulated genes (Figs 6G and H and S5D and E), there was at best a 3% overlap in terms of altered genes. These findings suggest that at the level of transcriptional changes, the mechanism of action of Li and GSK-3β is distinctive.

Mechanisms of Li-induced transcriptional changes

To understand the mechanisms underlying Li-induced transcriptome changes, we analysed the likely transcriptional control mechanisms. For this, we used the htFtarget database that has information on all human transcription factors (TFs) and the genes they control, tissue-specific TF-target information, data on TFs for noncoding RNA, and information on co-regulation for a target gene and its TFs (Zhang et al, 2020). TFs controlling genes differentially expressed in Li-treated neurons (DIV45) were identified from the htFtarget database. Out of 650 up-regulated genes in Li-treated neurons, 587 were identified as associated with one or more TFs in this database. Likewise, out of 525 down-regulated genes, 333 genes were linked to a TF. 485 and 323 unique TFs were identified that control the up-regulated and down-regulated genes, respectively. We then further looked for transcription factors that control differentially expressed genes that are involved in calcium signalling (Table S1-Sheet 3) (Fig S5A). Of the up-regulated genes, only RYR receptor was associated with three different transcription factors (CTCF, SPI1, and USF1) that could control its expression. On the other hand, several genes involved in calcium signalling that were down-regulated were controlled by transcription factors. Expression of CACNA1C, CACNA1D, CAMK2B, CAMK4, CHRM3, GRIN1, ITPR2, RYR3, SCN9A, and SLC2A13 genes is regulated by transcription factors. Of these ITPR2 and CAMK2B are regulated by 10 different TFs, whereas CACNA1D, CHRM3, RYR3, SCN9A and SLC2A13 are regulated by a single transcription factor (Fig S5B). This suggests that some of the genes that are differentially expressed in Li-treated neurons are under transcriptional control which can be a direct or indirect effect of Li treatment.

DNA constructs

For PIP₂ measurement, PH-PLCδ::GFP probe was used, which was a gift from Tamas Balla (plasmid # 51407; Addgene). The cDNA for human muscarinic acetylcholine receptor m1AchR was obtained from cDNA Resource Center (MAR0100000) and was cloned via Gibson assembly into a lentiviral pHR transfer backbone (having Puromycin resistance cassette) for lentiviral generation. The Nir2 shRNA was obtained from GE Dharmacon (RHS4430-200186441: Clone id V2LHS_62836).

Mammalian cell culture, treatment, and transfection

HEK293T cells were maintained under standard conditions at (37°C, with 5% CO₂) in DMEM High Glucose (DMEM; Life Technologies), supplemented with 10% FBS. For Li treatment, the cells were grown and maintained in these media, supplemented with 1 mM LiCl (Sigma-Aldrich) for 5 d (unless mentioned otherwise). The cells were tested for mycoplasma, before the experiments. MycoAlert (Lonza) was used per the manufacturer’s protocol to test the spent media for mycoplasma contamination. Transfection with the plasmid expressing PH-PLCδ::GFP probe was done 24 h before the imaging experiments using polyethylenimine solution (1 mg/ml; PEI; Polysciences Inc.) when the cells were around 70% confluent.

Lentiviral transduction to generate HEK293T cell line stably expressing muscarinic acetylcholine receptor m1AchR

Recombinant lentiviral particles were packaged in HEK293T cells grown in DMEM High Glucose media, supplemented with 10% FBS. The HEK293T cells cultured in the six-well plates were transfected with a solution consisting of 1.2 μg of pHR-IRES plasmid encoding muscarinic acetylcholine receptor CHRM1 cDNA, along with 1.2 μg of each of the helper plasmids (psPAX and pMD2.G) in 500 μl of Opti-MEM (Invitrogen) and 110 μl polyethylenimine (1 mg/ml; PEI). The medium was changed after 12 h and the virus was harvested at 72 h after transfection. Then, HEK293T cells were transduced with these recombinant viral particles and selected via puromycin treatment (1.5 μg/ml) for 1 wk to obtain the HEK293T cell line, expressing m1AchR. Western blot confirming the expression of m1AchR is shown in Fig S3C and D.

Generation of NSCs and differentiation and maintenance of neuronal culture

NSCs were generated as previously described (Mukherjee et al, 2019), with slight modifications. Human iPSCs were differentiated to embryoid bodies in E6 medium (Gibco). After that, primary neural rosettes that were formed, were manually picked, passaged, and eventually plated as NSC monolayer in Neural Expansion Medium (NEM; DMEM/F12 with 2 mM Glutamax, supplemented with N2, B27 without vitamin A, 20 ng/ml bFGF, 100 μM nonessential amino acids, 2 μg/ml heparin sulphate, 100 U/ml penicillin–streptomycin) (all components procured from Life Technologies). The NSCs were subjected to CD133+ selection as previously described via FACS (Peh et al, 2009) to eliminate cells of other lineages. The sorted NSCs were further characterized by immunofluorescence for the specific markers. NSCs were grown on Matrigel-coated tissue dishes in NEM at 37°C and 5% CO₂.

For differentiation, NSCs were seeded onto Matrigel- Growth Factor Reduced (Corning) coated cover-slip dishes or 35 mm dishes and expanded in NEM till the cells reached 70% confluency. NEM was then withdrawn and replaced with Neural Differentiation Medium (NDM; DMEM/F12 with 2 mM Glutamax, supplemented with N2, B27 containing vitamin A, 100 μM nonessential amino acids, 100 U/ml penicillin–streptomycin, 10 ng/ml BDNF, 10 ng/ml GDNF, 10 ng/ml IGF-1, 50 μg/ml ascorbic acid [Sigma-Aldrich] and 2 μM DAPT [Sigma-Aldrich]). DAPT was supplemented for the first 14 d. Media were changed every 2 d during the differentiation process. All the experiments were done DIV45 post differentiation (The day of addition of NDM was DIV0). For chronic Li treatment, the neuronal cultures were maintained in NDM supplemented with 1 mM LiCl for the last 10 d (DIV35–DIV45).

Live cell imaging for the PIP₂ probe

To analyse the dynamics of PIP₂ at the plasma membrane by following the change in translocation of the PH-PLCδ::GFP probe, live cell imaging was performed as stated in the study by Várnai and Balla (1998), with minor modifications. 20,000 cells were seeded in Poly-L Ornithine -coated cover-slip dishes and grown for 96 h with or without Li. 24 h before imaging, the cells were transfected with PH-PLCδ::GFP plasmid using polyethylenimine as per the manufacturer’s protocols. Carbamoylcholine chloride mediated changes in the membrane localization of PH-PLCδ::GFP probe was examined by a time-lapse method using an Olympus FV 3000 confocal microscope. The cells were kept in pre-warmed Kreb’s Ringer Buffer (120 mM NaCl, 4.7 mM KCl, 1.2 mM CaCl₂, 0.7 mM MgSO₄, 10 mM glucose, 10 mM HEPES; pH-7.4) and at 37°C while imaging. Changes in the PIP₂ membrane localization were quantified by taking a fluorescence intensity ratio of the plasma membrane (F_PM) and the cytosol (F_Cyt) across the time-lapse and normalizing it to the first value.

Calcium imaging for carbamoylcholine stimulation

For studying the calcium physiology, HEK293T-M1 cells were seeded in Poly-L Ornithine coated cover-slip dishes, whereas iPSC-derived cortical neurons (DIV45) were grown on Matrigel-coated cover-slip dishes. Before imaging, the cells were washed twice with HBSS (10 mM HEPES, 118 mM NaCl, 4.96 mM KCl, 1.18 mM MgSO₄, 1.18 mM KH₂PO₄, 10 mM Glucose, 2 mM CaCl₂; pH-7.4) and then loaded with the ratiometric Ca²⁺ indicator Fura-2 AM (4 μM) (Acetoxymethylester; Invitrogen) along with 0.002% Pluronic F-127 for 30 min. The excess dye was removed by washing the cells thrice with HBSS and cells were incubated for an additional 20 min to equilibrate the intracellular dye concentration and allow de-esterification. Imaging was performed for 10 min using a 40X objective (N.A. 0.75) of wide-field fluorescence microscope Olympus IX-83 for HEK293T cells, whereas a 20X objective (N.A. 0.50) was used for neuronal cultures. Free- and Ca²⁺-bound Fura-2 fluorescence intensities (excitation 340/380 nm, emission 510 nm) were recorded at every 5-s intervals using an EM-CCD camera (Evolve 512 Delta; Photometrics). Basal cytosolic Ca²⁺ was recorded in HBSS buffer for the initial 18 frames. Carbamoylcholine chloride (20 and 50 μM, respectively, for HEK and neuronal cultures, as determined in preliminary concentration–response experiments) (C4382; Sigma-Aldrich) was added to the cells for activating PLC and thereby, inducing Ca²⁺ release from the endoplasmic reticulum store. After recording for 42 frames, 20 μM Ionomycin (Calbiochem) was added and recorded for 24 frames for recording R_max. 24 more frames were recorded, post addition of 4 mM EGTA (containing 0.01% Triton-X-100) for chelation of Ca²⁺ ions and determination of R_min.

Calcium imaging for SOCE

For the SOCE experiment, the cells were kept in “zero Ca²⁺ HBSS” (10 mM HEPES, 118 mM NaCl, 4.96 mM KCl, 1.18 mM MgSO₄, 1.18 mM KH₂PO₄, 10 mM glucose; pH-7.4) just before the imaging. After recording the initial 12 frames for basal cytosolic Ca²⁺, an optimal concentration of thapsigargin (Tg) (10 μM) (Invitrogen) was added for the store depletion and 60 frames were recorded. Then, 2 mM CaCl₂ was then added to the extracellular buffer to induce SOCE—60 frames were recorded post induction of SOCE. Subsequently, 20 μM Ionomycin (Calbiochem) was added and recorded for 24 frames; then 4 mM EGTA (containing 0.01% Triton-X-100) was added and 24 frames were recorded. For studying PLC-activated calcium mobilization from stores and SOCE, carbamoylcholine chloride (20 μM and 50 μM, respectively, for HEK and neuronal cultures) was added instead of thapsigargin.

Analysis of calcium recordings

The CellSens Olympus software was used for the analysis. A dynamic region of interest (ROI) was drawn for each cell which exhibited a rise in [Ca²⁺]_i (as seen by increase in fluorescence post-stimulation) to track the fluorescence changes over time. The emission intensities corresponding to 340- and 380-nm excitations were used to calculate the 340/380 ratio for each cell across all time points. The [Ca²⁺]_i was estimated using the Grynkiewicz equation (Grynkiewicz et al, 1985) as follows:[Ca2+]i(nM)=Kd×SF×(R−Rmin/Rmax−R)where R_min and R_max refer to the minimum and maximum 340/380 ratio, respectively. 224 nM was taken as K_d for Fura-2 AM in human cells and SF (scaling factor) was calculated for each cell by dividing the fluorescence emission intensity at Ca²⁺ free form with the fluorescence emission intensity at Ca²⁺ bound form of the dye after excitation at 380 nm. The final Ca²⁺ traces were plotted using GraphPad Prism 9.0.

Generation of the IMPA1 deletion in cell lines

For generation of HEK293T-IMPA1^−/− cell line, the initial region of the coding sequence (exon1) of the IMPA1 gene was targeted by a pair of sgRNAs and deleted via CRISPR-Cas9 technology, then these cells were transduced to stably express Muscarinic receptor M1. The sgRNAs targeting the start of the coding sequence of IMPA1 gene were designed via the crispr.mit.edu. The sequences of the sgRNAs are as follows:

5′ GGAGAACCACCTTGTTGGCCAAGCTGGAATGTAGTGGCGTTTTAGAGCTAGAAATAGCAAGTT 3′

5′ GGAGAACCACCTTGTTGGGTAATATGGTACAGACACACGTTTTAGAGCTAGAAATAGCAAGTT 3′

These sgRNAs were cloned individually into humanized pgRNA plasmids (pgRNA-humanized was a gift from Stanley Qi [plasmid # 44248; Addgene]). The cells were co-transfected with the three plasmids (both the pgRNA-humanized plasmids encoding the sgRNAs and pLentiCas9 blast encoding for the Cas9 protein endonuclease from Streptococcus pyogenes) and selected with puromycin (2 μg/ml) for 48 h. Then the genomic DNA was isolated, and a Surveyor PCR was performed to check the occurrence of the deletion. Once confirmed that the sgRNAs were able to target the exon 1 of IMPA1 gene, the transfected cells were seeded in 96 wells as single cells and gradually expanded. After selection via Surveyor PCR using Surveyor primers (primers flanking the region ought to be deleted), a HEK293T cell line was selected which had showed only the shorter amplicon (amplicon after deletion) in the Surveyor PCR. This HEK293T cell line was then expanded and the expression of IMPA1 was checked by Western blotting using the anti-IMPA1 antibody. No corresponding bands for IMPA1 were observed, indicating that IMPA1 was knocked out in this cell line. This HEK293T-IMPA1^−/− cell line did not show any morphological difference with respect to HEK293T cells and did not require inositol supplementation in the media for maintenance.

Western blotting

For estimating the expression of the IMPA1, cells were suspended in appropriate volume of lysis buffer and kept on ice for lysis; 10 μl of it was used for Bradford protein assay (Bio-Rad) for quantification. Post quantification, an equal amount of samples was heated at 95°C with Laemmli loading buffer for 5 min, then loaded and separated on 10% SDS–PAGE gel. For estimating the expression of the muscarinic acetylcholine receptor m1AchR, cells were pelleted and directly lysed by adding Laemmli’s loading buffer and repeated syringing with insulin syringes (31 gauge). Then the samples were heated at 95°C for 5 min and loaded onto a pre-cast polyacrylamide 4–12% gradient gel (Bolt, Invitrogen). The proteins were then transferred onto a nitrocellulose membrane and then blocked with 10% Blotto (Santa Cruz Biotechnology) diluted in Tris Buffer Saline containing 0.1% Tween-20 (0.1% TBST) for 40 min. Subsequently, the membrane was incubated with the rabbit anti-IMPA1 antibody (1:10,000 diluted in 5% BSA in 0.1% TBST) (ab184165; Abcam) or goat anti-CHRM1 antibody (1 μg/ml, diluted in 5% BSA in 0.1% TBST) (ab77098; Abcam) overnight at 4°C. For loading control, the blot was incubated with mouse anti-β-tubulin antibody (1:4,000 diluted in 5% BSA in 0.1% TBST) (DSHB Hybridoma Product E7). The blots were then washed thrice with 0.1% TBST and then incubated with the appropriate HRP-conjugated secondary antibodies (1:10,000 dilutions; Jackson Laboratories, Inc.) for 1.5 h. Post incubation with the secondary antibodies, the blot was washed thrice with 0.1% TBST and developed using Clarity Western ECL substrate (Bio-Rad) on a GEImageQuant LAS 4000 system. The ImageJ software was used for the densitometric analysis of the blots. Firstly, the background intensities were subtracted from the images using an average of mean background intensities of a few ROIs adjacent to the band of interest. Then ROIs were drawn around the bands of interest and the integrated density of each ROI was extracted from the image. Finally, the ratio of the integrated density of the IMPase to that of the loading control (β-Tubulin) was calculated.

Calcium imaging for measuring transients

For studying the effect of Li on the excitability of cortical neurons, spontaneous calcium events (calcium transients) were measured in D149 cortical neuronal cultures at DIV45 for both untreated and Li-treated neurons, according to previously published lab protocols (Sharma et al, 2020). The neuronal cultures were washed twice for 10 min with HBSS and then loaded with 4 μM of Fluo-4 AM (F14201; Molecular probes) in HBSS having 0.002% Pluronic F-127 for 30 min at RT. Imaging was performed for a time-span of 10 min using a 20X objective (N.A. 0.50) with a time interval of 1 s at 488 nm illumination. The baseline measurement was recorded for a period of 4 min to visualize the calcium transients. Then 10 μM tetrodotoxin (Hellobio) was added to abolish calcium transients and recorded for another 4 min. A dynamic ROI was manually drawn around each neuronal soma; individual Ca²⁺ traces were then obtained using the CellSens Dimensions software. The fluorescence intensity value from each neuronal soma was normalized to the first baseline signal and the individual traces were plotted using GraphPad Prism 9.0. The frequency of calcium transients was measured manually from first-derivative filter traces.

Post-hoc immunofluorescence for neuronal characterization

Post calcium imaging, the neuronal cultures were washed thoroughly a few times with HBSS buffer and then fixed with 4% PFA dissolved in PBS for 20 min. Post fixation, the cultures were washed twice with PBS for 5 min each and then permeabilized with 0.1% Triton-X (dissolved in PBS). The cultures were then blocked with 5% BSA diluted in PBS, for 1 h at RT. Subsequently, the neurons were incubated with the primary antibodies diluted in 5% BSA in PBS overnight at 4°C. Next day, the cultures were subjected to three washes with PBS (10 min each) to remove nonspecifically bound primary antibody. Respective secondary antibodies at a dilution of 1:300 in 5% BSA in PBS were used for incubation for 2 h. DAPI was used as a nuclear marker at a dilution of 1:1,000. The primary and secondary antibodies used for immunocytochemistry are listed in Table S2.

RNA isolation, cDNA synthesis and real-time quantitative PCR

The total RNA was extracted from HEK293T cells or mature neuronal cultures, in multiple biological replicates using TRIzol (Ambion, Life Technologies) following the manufacturer’s protocol and thereby, quantified via NanoDrop 1,000 spectrophotometer (Thermo Fisher Scientific). For cDNA synthesis, 1 μg of the RNA from each replicate was treated with DNase I (amplification grade, Thermo Fisher Scientific) and then incubated with Superscript II reverse transcriptase (Invitrogen) along with random hexamers and dNTPs. A no reverse transcription control sample was also included for each sample. Real-time quantitative PCR analysis was performed on an Applied Biosystems 7500 fast qRT-PCR system using diluted cDNA samples and primers against genes of interest and GAPDH as internal controls. The C_t values obtained for different genes were normalized to those of GAPDH from the same sample. The relative expression levels were calculated using ΔC_t method, whereas the fold change was calculated using ΔΔC_t method. The primers used for qRT–PCR are listed in the Table S3.

LC-MS–based PIP and PIP₂ measurements

To validate whether the PH-PLCδ-GFP probe-based imaging of PIP₂ in the HEK293T cells reflected the changes in PIP₂ level, mass spectrometry measurements of PIP₂ from the cells were carried out following existing laboratory protocols (Sharma et al, 2019). Reverse phase liquid chromatography was coupled with high-sensitivity mass spectrometry (LC-MS) and a multiple reaction monitoring (MRM) method was employed to detect PIP₂ levels from the whole cells.

RNA sequencing and transcriptomic analysis

RNA extracted from mature neurons (DIV45 with and without Li treatment) and mature neurons (DIV45 with and without CHIR99021 treatment) as mentioned above was submitted for paired-end sequencing. Quality check and sequencing was carried out at the NCBS Sanger sequencing facility using Illumina Hiseq 2500 system. Single-stranded RNA sequencing libraries were prepared for sequencing using NEBNext Ultra II Directional RNA Library Prep with Sample Purification Beads (Catalog no-E7765L). Paired-end reads, (125 base pair for each neuronal sample and 100 base pair for each NSC samples) was obtained after sequencing. Post adapter trimming, FastQC was used to check the quality of reads. The reads were then aligned to human genome (GRCh38) using hisat2-build and hisat2-align module. The module for single-strand sequencing was used in the Hisat-2 protocol. The output generated in the bam format, was further used to obtain the counts of each of the individual gene using the HT-Seq pipeline. The differential expression of genes was then calculated using DeSeq2. GO annotation analysis was done using the DAVID tool.

Sampling and statistical analysis

Each experiment was performed unblinded on different biological groups with multiple biological replicates. No statistical analysis was done a priori to determine the sample sizes. For the calcium imaging experiments, quantification was from a minimum of 100 cells for all genotypes from multiple dishes of each type. For the PIP₂ probe experiments, ∼50 cells were taken for the quantitative analysis. Statistical significance was computed using Student’s unpaired t test to check for differences in means between samples of different genotypes. Two-way ANOVA with post hoc Tukey’s pairwise comparison was used for the grouped analysis. All statistical analyses were performed on Graph Pad Prism (version 9) and schematic representations were created with biorender.com.