Dravet Syndrome is an intractable form of childhood epilepsy associated with deleterious mutations in SCN1A, the gene encoding neuronal sodium channel Nav1.1. Earlier studies using human induced pluripotent stem cells (iPSCs) have produced mixed results regarding the importance of Nav1.1 in human inhibitory versus excitatory neurons. We studied a Nav1.1 mutation (p.S1328P) identified in a pair of twins with Dravet Syndrome and generated iPSC-derived neurons from these patients. Characterization of the mutant channel revealed a decrease in current amplitude and hypersensitivity to steady-state inactivation. We then differentiated Dravet-Syndrome and control iPSCs into telencephalic excitatory neurons or medial ganglionic eminence (MGE)-like inhibitory neurons. Dravet inhibitory neurons showed deficits in sodium currents and action potential firing, which were rescued by a Nav1.1 transgene, whereas Dravet excitatory neurons were normal. Our study identifies biophysical impairments underlying a deleterious Nav1.1 mutation and supports the hypothesis that Dravet Syndrome arises from defective inhibitory neurons.https://doi.org/10.7554/eLife.13073.001
Dravet Syndrome, initially described by Dr. Charlotte Dravet in the 1970s, represents one of the most severe forms of intractable childhood epilepsy (Dravet et al., 2011). Seizures in Dravet Syndrome appear early in infancy, progress throughout childhood, and are often accompanied by developmental delays in language, motor function, learning and social skills (Dravet, 2011). Heterozygous loss-of-function mutations in the SCN1A gene are the most common genetic basis for Dravet Syndrome (Catterall et al., 2010; Escayg and Goldin, 2010). SCN1A encodes the voltage-gated sodium channel Nav1.1, which contributes the fast depolarization of neuronal membranes during an action potential. Loss of one copy of Nav1.1 might be expected to reduce neuronal excitability, but paradoxically the genetic perturbation causes epilepsy. Initial insights into the neurobiological function of Nav1.1 came from studies of a mouse model of Dravet Syndrome, where the loss of one copy of Scn1a impaired the excitability of hippocampal inhibitory interneurons but not excitatory neurons (Yu et al., 2006). The disproportionate contribution of Nav1.1 to the excitability of inhibitory interneurons in the hippocampus, cortex, cerebellum and thalamus was confirmed by additional studies (Cheah et al., 2012; Dutton et al., 2012; Kalume et al., 2007; Martin et al., 2010; Ogiwara et al., 2007, 2013; Tai et al., 2014; Tsai et al., 2015), but the effect of the Nav1.1 mutation in the mouse is dependent on the genetic background with seizures occurring in the C57/B6 background but absent in the 129 background (Mistry et al., 2014; Ogiwara et al., 2007; Rubinstein et al., 2015; Tsai et al., 2015; Yu et al., 2006).
The effect of the genetic background on the mouse phenotype has highlighted the importance of characterizing the cellular effects of Dravet mutations in a human genetic context. Until recently, experiments on human neurons required the use of tissues obtained following surgeries. However, the development of induced pluripotent stem cell (iPSC) technology (Takahashi et al., 2007) has made it possible to generate neurons from patients with disease-causing mutations. Several research groups have used iPSCs to study the pathophysiology of Dravet Syndrome. Two studies demonstrated impairment of action potential firing in patient iPSC-derived inhibitory neurons without examining the excitatory neurons (Higurashi et al., 2013; Liu et al., 2016). Two other studies reported elevated excitability and increased sodium currents in patient-derived excitatory neurons alone (Jiao et al., 2013) or both excitatory neurons and inhibitory neurons (Liu et al., 2013). The precise reason for differences between the studies is unclear; however, differences in the generation of specific neuronal cell types likely contribute to the discrepancy. A side-by-side comparison of inhibitory neurons and excitatory neurons derived from Dravet and control iPSCs, including detailed cellular characterization of the neuronal cell types would help to resolve some of the controversy surrounding these studies.
In this study, we investigated the p.S1328P mutation in Nav1.1 identified in a pair of twins with Dravet syndrome. We characterized the effect of this mutation on the biophysical properties of the channel in human neurons and used patient-derived iPSCs to study the consequence of this mutation on the activity of inhibitory and excitatory telencephalic neurons. We found that the disease-associated mutation impairs both functional expression and channel gating of Nav1.1. We also found that inhibitory neurons from the Dravet patients have reduced sodium currents and a pronounced defect in action potential firing particularly when stimulated by large injections of current. In contrast, both the sodium currents and the excitability of excitatory neurons from Dravet patients were indistinguishable from those of control subjects. The reduced excitability of Dravet inhibitory neurons could be rescued by ectopic expression of Nav1.1, while reducing Nav1.1 levels by RNAi in wild-type inhibitory neurons could phenocopy the Dravet cells. Together these findings suggest that the p.S1328P mutation reduces the function of the Nav1.1 channel, which selectively impairs the activity of inhibitory neurons in humans.
We obtained fibroblasts from two twins affected with Dravet Syndrome. Sequencing of Nav1.1 coding sequence from these patients revealed a serine to proline mutation at position 1328 (Figure 1A; Reference mRNA: GenBank AB093548.1; reference protein: GenBank BAC21101.1). The Serine-1328 residue is located in a voltage-sensing transmembrane segment of the protein, and it is conserved in human and mouse Nav1.1 α subunits. This mutation has previously been described in other individuals with Dravet Syndrome (SCN1A Infobase: http://www.scn1a.info/; SCN1A variant database: http://www.molgen.ua.ac.be/scn1amutations/) but the functional effects of the mutation are not known. To determine whether the Nav1.1-p.S1328P mutation alters the function of the channel we expressed wild type and mutant Nav1.1 channels in human neurons. To distinguish the exogenous from the endogenous channels we introduced a second mutation, p.F383S, that makes the channel resistant to tetrodotoxin (TTX). This allowed us to block the activity of the endogenous channels with TTX (Figure 1—figure supplement 1) (Cestèle et al., 2008, 2013) to measure the currents carried by the expressed channels in isolation. Expressing Nav1.1 constructs in neurons allowed us to examine the function of the mutant channel in cells that express the correct auxiliary subunits and other molecular partners that regulate channel gating. We made two lentiviruses: mCherry-T2A-Nav1.1-F383S that expresses the engineered control protein and mCherry from a bisystronic message (Donnelly et al., 2001; Szymczak et al., 2004), and mCherry-T2A-Nav1.1-F383S-S1328P, which expresses the mutant channel and mCherry. Both viruses contained a Doxycline (Dox)-regulated promoter that allowed us to control the expression of the constructs. We transduced the lentiviral vectors into human neurons derived from the human ESC line H9 that expresses the transcription factor from the NgN2 (Figure 1B). NgN2-expression generates highly homogeneous and rapidly maturing human excitatory neurons in easily scalable quantity (Zhang et al., 2013).
We first used Western blotting to measure the expression levels of control and mutant channels. We found that Nav1.1-F383S and Nav1.1-F383S-S1328P were expressed at similar levels in NgN2 neurons following treatment with either 0.1 µg/ml or 2 µg/ml Dox in the culture media (Figure 1C), suggesting that the mutation does not affect the expression of the channel. We then used whole-cell patch clamp to measure the TTX-resistant sodium currents. We found that Nav1.1-F383S-S1328P produces currents that are significantly smaller than those of control channels when the cells are treated with both 0.1 µg/ml and 2 µg/ml Dox (Figure 1D and Table 1). This suggests that the S1328P either affects the number of channels on the cell surface or the activity of the Nav1.1 channel. The trend that the current observed in the presence of 2 µg/ml Dox was larger than that observed in the presence of 0.1 µg/ml Dox in the cells expressing the mutant channels but not the control channels suggests that there is a defect in trafficking the channel to the cell surfaces (Figure 1D and Table 1). We also examined the biophysical properties of the Nav1.1-F383S-S1328P channels that do make it to the cell surface. The activation voltage (V1/2) was hyperpolarized by around 5 mV, while the steady-state inactivation voltage (V1/2’) showed a hyperpolarizing shift of around 15 mV (Figure 1E and Table 2). This increased sensitivity for activation is likely to enhance the activity of the channel in neurons slightly, however, the much larger increase in steady-state inactivation offsets this effect and probably results in a net loss of function by reducing the fraction of channels that are available to open under prolonged current injections.
We next generated six iPSC lines from the Dravet Syndrome twins carrying the heterozygous p.S1328P mutation in Nav1.1 (Figure 1A). We converted skin fibroblasts from the patients into iPSCs using episomal vectors expressing SOX2, OCT3/4, L-MYC, KLF4, LIN28, and p53-shRNA as described previously (Okita et al., 2011). We used RT-PCR to verify that the reprogramming vectors were lost in most iPSC cell lines after the 15th passage (Figure 1—figure supplement 2). The iPSC lines expressed the pluripotency markers NANOG, TRA-2-49, OCT3/4 SOX2, and LIN28 (Figure 1—figure supplements 2 and 3). Four of the iPSC lines responded well to neural differentiation, which was evident by the formation of neural rosettes (Figure 1—figure supplement 3). We examined the genomic integrity of the iPSC lines at early passages (passage six to eight) by identifying copy number variants (CNVs) and compared them with CNVs in the source skin fibroblasts. The iPSCs had between three and sixteen novel CNVs, all of which were less than 1.5 Mb in size and none of which occurred in well annotated pathogenic regions (Supplementary file 1). The number of novel CNVs in each cell line was below the median number observed in a previous SNP array based characterization of reprogramming associated CNVs (Hussein et al., 2011). We also generated six iPSC lines derived from 4 control subjects, 2 males and 2 females, none of whom had a diagnosis of epilepsy. In addition to these iPSCs lines we also included the human embryonic stem cell line H9 (ESC-H9). The use of the control and Dravet cell lines in various experiments is summarized in Supplementary file 2.
We next differentiated iPSCs /ESCs into telencephalic excitatory neurons and inhibitory neurons in parallel. The excitatory neurons were generated using a protocol we have previously described (Paşca et al., 2011) (Figure 2A top and Figure 2—figure supplement 1). Briefly, iPSC/ESC colonies were dislodged from culture vessels by dispase and grown in suspension for five days in the presence of the BMP pathway inhibitor Dorsomorphin (5 µM) and the TGFβ pathway inhibitor SB431542 (10 µM) to form embryoid bodies (EBs). The EBs were plated on poly-L-ornithine and laminin coated dishes and then treated with neuronal induction media containing recombinant human FGF2 (20 ng/ml) and EGF (20 ng/ml) for ten days to induce neural rosettes. The rosettes were then isolated mechanically and expanded in suspension for seven days before plating onto glass coverslips for post-mitotic neuronal differentiation. Finally the neurons were dissociated and plated onto a monolayer of rat astrocytes for functional maturation over fifty to seventy days. This protocol generates neural stem cells and progenitors that express FOXG1 and PAX6, two markers that are characteristic of telencephalic identity (Figure 2C, top) and cortical neurons that express layer-specific cortical markers including FOXP1, SATB2, CTIP2 and ETV1 (Paşca et al., 2011).
To generate telencephalic inhibitory neurons we also started with EB formation using the dual SMAD inhibition protocol (Figure 2—figure supplement 1). The EBs were exposed to a combination of patterning molecules (Figure 2A, bottom) to generate neural stem cells and progenitors with medial ganglionic eminence (MGE) identity. The MGE is a key ventral telencephalic structure that gives rise to inhibitory neurons. We sequentially applied a low-dose (10 nM, between Day 3 and Day 8) and high-dose (100 nM, between Day 8 and Day 15) of Smoothened agonist SAG (CAS 364590-63-6), to activate Hedgehog signaling for MGE differentiation (Danjo et al., 2011). On Day 7, we treated the cell culture for one day with retinoic acid (RA), which promotes the differentiation of MGE cells partly by enhancing Sonic Hedgehog expression (Goulburn et al., 2012), and exposed the cells to FGF2 (10 ng/ml) and IGF1 (10 ng/ml) for the next 8 days to promote the proliferation of nascent MGE cells (Maroof et al., 2010). Between Day 15 and Day 35, we cultured the cells in neuronal media supplemented with trophic factors including BDNF (20 ng/ml), NT-3 (20 ng/ml), GDNF (10 ng/ml), cyclic AMP (0.5 mM), BMP4 (20 ng/ml, only between Day 15 and Day 25) and IGF1 (10 ng/ml, only between Day 25 and Day 35). On Day 35, the nascent neurons were dissociated into single cells and plated onto a confluent layer of rat cortical astrocytes for fifty to seventy days, as was described for the excitatory neurons.
The neural stem cells and progenitors that were generated using the ventral telencephalon patterning protocol co-expressed transcription factors NKX2.1, FOXG1 and OLIG2 (Figure 2B and C, Figure 2—figure supplement 2A and B), which together characterize the MGE region in the human embryonic brain (Maroof et al., 2013; Nicholas et al., 2013). The cells also expressed LHX6 (Figure 2B), an MGE-specific transcription factor acting downstream of NKX2.1, as well as transcription factors GSH2 and DLX2 (Figure 2B) that cooperate in the development of ventral telencephalic structures including MGE (Wonders and Anderson, 2006). In contrast, PAX6, a transcription factor characteristic of dorsal telencephalic progenitors that form excitatory neurons (Toresson et al., 2000; Yun et al., 2001), was downregulated in the ventral patterning protocol (Figure 2B and C). Finally the protocol led to minimal expression of the transcription factors COUP-TFII and SP8 (Figure 2B and Figure 2—figure supplement 2C). COUP-TFII is known to be expressed in the caudal ganglionic eminence (CGE), the dorsal portion of lateral ganglionic eminence (dLGE), and the dorsal portion of MGE (dMGE). SP8 is largely co-expressed with COUP-TFII in the dorsal CGE (dCGE) and dLGE but absent from MGE (Cai et al., 2013; Ma et al., 2013). The strong induction of NKX2.1 and mild induction of COUP-TFII and SP8 suggest that a large fraction of the neural stem cells and progenitors made using this protocol have an MGE identity, while a small fraction of the cells have dLGE and/or CGE identity.
The post-mitotic neurons (Day 15 to Day 90) generated from the ventral telencephalic protocol have many features of cortical inhibitory neurons. Among all neurons that were identified by MAP2 staining, about 30% were co-stained by an anti-GABA antibody (Figure 3A), suggesting that they produce the inhibitory neurotransmitter GABA. Among the inhibitory (GABA+) neurons, approximately 50% co-expressed Calretinin, and 2%-8% expressed Somatostatin, and the rest expressed neither marker (Figure 3B and C). We did not identify any cells that expressed Parvalbumin, possibly because these cells emerge during postnatal development and require both cell-intrinsic and non-intrinsic factors for specification, migration and maturation (Okaty et al., 2009; Sultan et al., 2013; Wonders and Anderson, 2006). The expression of Calretinin and low expression of Somatostatin are similar to an earlier report that describes development of telecephalic interneurons using an MGE-biased differentiation protocol (Nicholas et al., 2013). However, an alternative scenario could be that the Calretinin-expressing neurons in our current study arise from a population of CGE-like progenitors, as in the adult mammalian cortex where Calretinin expression is mostly restricted to CGE-lineage interneurons (Wonders and Anderson, 2006). To distinguish between the two possibilities, we performed double immunostaining for NKX2.1 and Calretinin in differentiating cells before plating them to rat astrocyte feeders. We observed co-expression of NKX2.1 and Calretinin in many cells with progenitor-like morphology and also in a population of cells showing neuronal morphology (Figure 3—figure supplement 1A). Furthermore, expression of SP8, an aforementioned transcription factor found in CGE-derived progenitors and neurons, was barely detected in our iPSC-derived inhibitory neurons (Figure 3—figure supplement 1B and C). These results are consistent with findings from the differentiating neural rosettes (Figure 2—figure supplement 2C) and support the idea that the Calretinin-expressing inhibitory neurons in our study originate from MGE-like progenitors.
Both the excitatory and inhibitory neuronal differentiation protocols generate mixed populations of cells only some of which are mature neurons. We used fluorescent indicators to select the mature neurons for electrophysiological characterization. To identify excitatory neurons we used a lentivirus that expresses EGFP under the control of the CaMKII promoter (Shcheglovitov et al., 2013) (Figure 2—figure supplement 1 and Figure 3D). To identify inhibitory neurons we used a lentivirus in which GFP is expressed under the control of an enhancer sequence that is found between the mouse Dlx1 and Dlx2 genes (Dlxi1/2b-GFP, Figure 2—figure supplement 1 and Figure 3D). The mouse Dlxi1/2b enhancer is homologous to its counterpart in the human genome and labels MGE-like progenitor cells differentiated from mouse ES cells (Chen et al., 2013). Approximately 65% of the Dlxi1/2b-GFP labeled neurons were co-stained by an anti-GABA antibody, indicating they are inhibitory neurons (Figure 3D and E). To further identify the cells that were being patch clamped we collected the cell soma after every recording and used single-cell RT-PCR to measure the expression of GAD65 and GAD67. About 70% of the Dlxi1/2b-GFP neurons expressed at least one of the two markers, consistent with the results of the anti-GABA staining and independently of whether the neurons were derived from control or Dravet iPSCs (Figure 3F). Furthermore, by single-cell RT-PCR only a small percentage of these GAD67/65-positive inhibitory neurons (0.9% of the control sample group and 5.4% of the Dravet group) expressed 5HT3aR, a serotonin receptor and specific marker for CGE-derived inhibitory interneurons, while more than 60% (for both the control and Dravet groups) expressed Calbindin, which is commonly found in MGE-derived interneurons (Figure 3—figure supplement 2). These results are consistent with other characteristics of the neural differentiation protocol and indicate that the cellular composition of the inhibitory neuron pool is dominated by MGE-like rather than CGE-like lineages.
Prior to studying the differences between inhibitory neurons from patients and controls we characterized basic physiological properties of inhibitory neurons from control iPSCs. The Inhibitory neurons fired trains of action potentials upon a 1000-millisecond current injection, and the fraction of cells firing continuous trains increased as the cells matured at least until differentiation day 90 (Figure 3G). At this later stage, 73.3% of the inhibitory neurons showed an accommodating pattern of action potential firing; 3.3% had a non-accommodating pattern; a 5% showed a stuttering pattern; and the remaining 18.4% fired at most a few or single action potentials (Figure 3—figure supplement 3). We also observed spontaneous inhibitory synaptic currents that were blocked by Gabazine indicating that the cells have functional GABAergic synapses (Figure 3H). To determine whether the iPSC-derived inhibitory neurons could integrate into a neural circuit and form functional inhibitory synapses, we infected the cells with a lentivirus expressing Channelrodopsin 2 fused to YFP under the control of the human synapsin-1 promoter (pLenti-Syn1-ChR2(H134R)-YFP) and engrafted them into rat hippocampal organotypic slices. After 8 weeks on the slices, multipolar, ChR2-YFP positive neurons could be observed in the hippocampus (Figure 3I; Figure 3—figure supplement 4A) containing presynaptic-bouton like structures (Figure 3—figure supplement 4B). We used whole-cell patch clamp to measure the currents in the transduced cells and found that illumination with 470 nm light could reliably activate ChR2 and elicit repetitive action potentials (Figure 3—figure supplement 4C). Patch clamping of rat neurons surrounding the axons of ChR2-YFP-cells revealed light-driven Gabazine sensitive postsynaptic currents (Figure 3J; Figure 3—figure supplement 4D and E) indicating that the engrafted human inhibitory neurons formed synapses with the rat neurons. Taken together these experiments suggest that the protocol for specifying inhibitory neurons from iPSCs generated cells that expressed the ventral telecephalic and terminal identity markers, fired repetitive action potentials and formed functional GABAergic synapses.
We next characterized the physiological properties of inhibitory and excitatory neurons from patients with Dravet syndrome and compared them to cells from individuals without the disease. Because Dravet Syndrome is caused by the loss of one copy of the Nav1.1 channel we first assessed voltage-dependent Na+ currents (INa) in Dravet and control cells. Using whole cell voltage clamp recording (Figure 4—figure supplement 1A) under conditions that eliminate contaminating calcium and potassium currents, we observed voltage-dependent and tetrodotoxin (TTX) sensitive sodium currents (Figure 4A and B). When we measured INa in CaMKII-GFP-labeled telencephalic excitatory neurons, there was no significant difference between control and Dravet excitatory neurons in the amplitude (Figure 4C and D; Table 3), voltage dependence for activation (Figure 4D and Table 3), or steady-state inactivation profile (Figure 4—figure supplement 1B, Figure 4—figure supplement 2A and Table 4). In contrast, we observed a significant reduction in the INa amplitude in Dlxi1/2b-GFP-labeled inhibitory neurons from individuals with Dravet Syndrome relative to controls (p=0.034 by t-test; Figure 4E and F; Table 3). There were no significant differences between control and Dravet inhibitory neurons in the voltage dependence for INa activation (Figure 4F; Table 3) or the steady-state inactivation (Figure 4—figure supplement 2B; Table 4).
Then we assessed the capacity of Dravet and control neurons to fire action potentials in response to injections of a depolarizing current. Both control and Dravet excitatory neurons were able to fire trains of action potentials when injected with up to 60 pA current (Figure 5A). There was no statistical difference between control and Dravet excitatory neurons in the maximum frequency of action potential firing (fmax, Figure 5B) or in the quantitative relationship between current injection and the number of action potentials that a cell fired (the dynamic output, Figure 5C and D). This is consistent with our previous observations that the Nav1.1 mutation has little or no effect on the sodium current in excitatory neurons.
In contrast when we measured action potential firing in inhibitory neurons we observed a large difference between control and Dravet samples. There was a 40% reduction in the maximal firing frequency of inhibitory Dravet neurons compared to those in controls (Figure 5E and F; p = 0.0154 by t-test). At high current injection levels Dravet inhibitory neurons often fired only a few action potentials before entering depolarization block, in contrast to the sustained trains of action potentials observed in the control neurons. Therefore, the input-output relationship of Dravet inhibitory neurons was bell shaped with reduced firing observed at high levels of stimulation (Figure 5G and H; between Dravet and control inhibitory neurons p<0.0001 by two way ANOVA, and p<0.05 by post hoc Sidak’s multiple comparisons at the two highest stimulation levels). This observation strongly suggests that the loss of one copy of the Nav1.1 channels impairs the ability of inhibitory neurons from Dravet patients to fire at high frequencies.
The simplest explanation for the selective impact of a loss-of-function Nav1.1 mutant on inhibitory neurons in Dravet patients is that the Nav1.1 is preferentially expressed in these cells. We compared the expression of Nav1.1 mRNA between inhibitory neurons and excitatory neurons derived from human iPSCs. When normalized to the expression of NaVβ3 mRNA, which encodes a sodium channel auxiliary subunit uniformly present among subtypes of mouse cortical neurons (Sugino et al., 2006), we observed a 1.6–5.6 fold enrichment of Nav1.1 mRNA in cultures of inhibitory neurons relative to excitatory neurons derived from multiple control iPSC lines, Dravet iPSC lines and the hESC line H9 (Figure 6A). Using MAPT mRNA as an alternative reference for normalization, we obtained a consistent result for the control iPSC line 8858–3 (Figure 6—figure supplement 1). The data suggest that Nav1.1 expression is enriched in inhibitory neurons, and the enrichment can be observed within the time frame of human iPSC in vitro differentiation that is much shorter compared to the natural time course of human brain development preceding disease onset. To examine the functional contribution of enriched Nav1.1 expression, we developed lentiviral expression constructs carrying Nav1.1-tRFP-shRNAs (one scramble shRNA sequence and two Nav1.1-targeting shRNA sequences in separate vectors; Figure 6B) and a doxycycline-inducible mCherry-T2A-Nav1.1-cDNA (mCherry-T2A tagged wild-type Nav1.1, Figure 6C), which served to decrease or increase Nav1.1 expression in neurons, respectively. We first verified that both shRNAs reduced the expression of Nav1.1 mRNA by around 60% (Figure 6—figure supplement 2; p<0.005 by Sidak’s multiple comparisons following ANOVA). We then transduced the human neurons grown on rat astrocytes and measured INa currents and membrane excitability in neurons that co-express the lentiviral tracers (tRFP or mCherry) and the cell-type reporters (CaMKII-GFP or Dlxi1/2b-GFP). We found that suppressing Nav1.1 expression using the lenti-shRNAs reduced the INa amplitude (Figure 6D middle) in Dlxi1/2b-GFP+ neurons derived from the control hESC line H9, but not in the CaMKII-GFP+ excitatory neurons (Figure 6D left). Conversely, the Nav1.1 lenti-cDNA rescued the INa amplitude in the Dlxi1/2b-GFP+ neurons derived from Dravet iPSCs (Figure 6D right). In parallel to the effects on INa, the Nav1.1-shRNAs impaired action potential firing in control Dlxi1/2b-GFP+ neurons (Figure 6E), whereas the cDNA rescued action potential firing in the Dravet Dlxi1/2b-GFP+ neurons (Figure 6F). These results indicate that Nav1.1 is a major component of INa in inhibitory neurons and plays an important role in controlling the excitability of these cells.
Dravet Syndrome is a prototypical form of epileptic encephalopathy caused by loss of function mutations in the Nav1.1 channel. We investigated a Nav1.1 mutation in a pair of twins with Dravet Syndrome and found that it resulted in a decrease in the current carried by the channel but not in a complete loss of function. We studied the effects of this mutation on excitatory and inhibitory interneurons by taking great care to generate and characterize human ESC/iPSC-derived telencephalic excitatory neurons and inhibitory neurons in parallel. We found that the p.S1328P mutation affected INa and action potential firing specifically in patient-derived inhibitory neurons without affecting excitatory neurons. Furthermore, we determined that Nav1.1 is expressed at higher levels in human telencephalic inhibitory neurons than in excitatory neurons, and that it makes a disproportionate contribution to the excitability of these cells. This is consistent with previous findings in mouse models of Dravet syndrome that correlated Nav1.1 expression and function in inhibitory interneurons (Cheah et al., 2012; Dutton et al., 2012; Ogiwara et al., 2013; Ogiwara et al., 2007; Tai et al., 2014) but differs from some of the previous findings in iPSC models (Higurashi et al., 2013; Jiao et al., 2013; Liu et al., 2016; Liu et al., 2013). Specifically, two previous papers have suggested that there is an increase in INa and neuronal excitability in Dravet neurons (Jiao et al., 2013; Liu et al., 2013) rather than a selective decrease in INa and excitability in inhibitory neurons as we have observed here. Two other studies (Higurashi et al., 2013; Liu et al., 2016) broadly agree with our findings, although the authors only characterized inhibitory neurons, and the developmental origin and trajectory of the inhibitory neurons were not studied.
It is not clear why there is a discrepancy between the various iPSC studies. One possibility is that the different groups studied different sets of mutations which could differentially affect Nav1.1 expression and function. We found that p.S1328P affects the trafficking and gating of Nav1.1 expressed heterologously in neurons. This correlated with the decreased sodium current and excitability in patient-derived inhibitory neurons.
A second possibility underlying the discrepancy between studies is that experimental conditions that were used to measure sodium channel activity and neuronal excitability were different between our work and earlier papers. For example, in two earlier studies the sodium currents were quantified as the inward component of the whole cell current, without pharmacologically eliminating voltage-dependent potassium currents and calcium currents (Jiao et al., 2013) or with calcium currents eliminated only (Liu et al., 2016). This could confound the results as the measurements of current amplitude, kinetics and voltage dependence could be contaminated by membrane conductance to the non-sodium ions.
A third possibility is that the method we used to generate neurons in this paper produced physiological properties different than those studied in earlier papers. Although in all four prior studies neural differentiation began with the formation of embryoid bodies (EBs), the earlier papers do not provide sufficient information on specific steps of neural differentiation or cell lineage markers for us to determine if the cells are similar to ours. Our neurons resemble inhibitory and excitatory neurons found in the developing human telencephalon, based on the expression of cell-type specific markers in both neural progenitors and neurons. The critical question is whether iPSC-derived neurons in a culture dish fully mimic the neurons in patient brains. While we can’t answer this question definitively, it is encouraging that the neurons in our experiments have similar phenotypes compared to those observed in vivo in multiple mouse model of Dravet Syndrome.
Why does loss of Nav1.1 have a greater effect on the excitability of inhibitory neurons relative to excitatory neurons? One possibility might be that excitatory neurons are more effective than inhibitory neurons at upregulating other sodium channels in response to loss of one copy of Nav1.1. Our data suggests that this is not the case. We observed that human Nav1.1 mRNA was present at higher levels in ESC/iPSC-derived inhibitory neurons than excitatory neurons, and that bi-directional manipulation of Nav1.1 expression level using lentiviral constructs correlated with INa currents and membrane excitability in the inhibitory neurons. In addition, studies in rodents suggest that cortical inhibitory interneurons express both Nav1.1 and Nav1.6 in their axon initial segments while cortical excitatory neurons express only Nav1.6 channels in this cellular compartment (Lorincz and Nusser, 2008; Ogiwara et al., 2007). As the axon initial segment plays a key role in the initiation of action potentials, this difference could account for the disproportionate effect of loss of Nav1.1 on inhibitory neuron firing in Dravet Syndrome.
All procedures and experiments involving primary fibroblasts and iPSCs from Dravet Syndrome patients were carried out at Stanford University after informed consent and under protocols approved by the IRB and SCRO of Stanford University. A clinical skin biopsy procedure was carried out to obtain primary fibroblasts from a pair of twins diagnosed with Dravet Syndrome. The primary fibroblasts were expanded and cryopreserved at the Stanford cytogenetics lab. The primary fibroblast culture medium contains DMEM (base medium, Gibco), FBS (10%, Gibco), NEAA (1%, Gibco), Sodium pyruvate (1%, Gibco), GlutaMax (1%, Gibco), PenStrep (100 units/mL working concentration, Gibco), and β-mercaptoethanol (0.1 mM, Sigma Aldrich). A set of episomal expression vectors encoding SOX2, OCT3/4, KLF4, LIN28, L-MYC and P53-shRNA (Okita et al., 2011) were delivered to the fibroblasts using a Basic Primary Fibroblasts Nucleofector Kit (Lonza). The fibroblasts were allowed to recover and then seeded onto mouse DR4 MEF feeders (prepared by Stanford University Transgenic Mouse Facility). The co-culture was maintained for 24 hr in the primary fibroblast medium and later on in standard human ES medium containing DMEM/F12 (base medium, Gibco), KnockOut Serum Replacement (20%, Gibco), GlutaMax (0.5%, Gibco), NEAA (1%, Gibco), Penstrep (100 units/mL working concentration, Gibco), β-mercaptoethanol (0.1 mM, Sigma Aldrich), and human recombinant FGF2 (10 ng/ml, R&D systems). After 3 weeks, emerging iPS cell colonies were isolated by manual picking and expanded separately as cell lines, which are named in de-identified codes. The established iPSC lines were mycoplasma-free based on tests using the Mycoalert Mycoplasma Detection Kit (Lonza).
Immunocytochemistry using antibodies against the pluripotency markers NANOG (nuclear localization, antibody: R&D Systems AF1997) and TRA-2-49 (cytoplasmic localization, antibody: DSHB Hybridoma Product TRA-2-49/6E, originally developed by P. W. Andrews) were conducted as described before (Yazawa et al., 2011). Quantitative RT-PCR was performed to examine the silencing of episomal reprogramming factors and the establishment of corresponding endogenous factors, using PCR primers described in Supplementary file 3.
For SNP array assessment of copy number variants (CNVs), cultures of skin fibroblasts and iPSCs were dissociated with Accutase (Gibco), washed three times in PBS, pelleted, and preserved at -80°C. The cell samples were shipped to a commercial service (CapitalBio Corp., Beijing, China) for CNV analysis using the Affymetrix Genome-wide Human SNP Array 6.0. The SNP array contains more than 946,000 probes to detect both known and novel CNVs. Data analysis was based on a copy number polymorphism (CNP) calling algorithm developed by the Broad Institute.
Control iPSC lines NH2-6, 6593–7 and 6593–8 were described in previous studies (Pasca et al., 2011; Yazawa et al., 2011). Control iPSC lines 8858-C and 8858–3 were generated under informed consent and protocols approved by IRB and SCRO at Stanford University, using skin fibroblasts from healthy donor 8858 and identical reprogramming methods as described for the Dravet Syndrome iPSCs. Basic characterization of these two iPSC lines is described in Figure 1—figure supplement 4. Control iPSC line 8402–2 WT9 is a subclone of iPSC line WT-GM08402A-SeV-hiPSC 1, which was originally generated at Novartis Institutes for BioMedical Research, using GW08402 fibroblasts from the Coriell Institute for Medical Research and Sendai-virus-based reprogramming (Fusaki, et al., 2009). WT-GM08402A-SeV-hiPSC 1 passed tests for pluripotency gene expression and viral vector removal. In our current study, we verified pluripotency marker expression and confirmed a normal karyotype in the subclone 8402–2 WT9 (Figure 1—figure supplement 4). All control iPSC lines were mycoplasma-free based on tests by the Mycoalert Mycoplasma Detection Kit (Lonza).
An overview of the procedure is presented in Figure 2A. First, human ES/iPS cell colonies were detached from DR4 MEF feeders with Dispase and grown as suspension culture for five days to form embryoid bodies (EBs). The EB medium was the standard human ES medium supplemented with SMAD inhibitors Dorsomorphin (5 µM, Sigma Aldrich) and SB431542 (10 µM, Tocris) but without FGF2. Next, EBs were plated for the induction of neural rosettes over 10 days (Figure 2B). The neural induction media contained Neurobasal-A (Base medium, Gibco), B-27 without Vitamin-A (2%, Gibco), GlutaMax (1%, Gibco), and Penstrep (100 units/mL working concentration, Gibco). Human recombinant FGF2 (20 ng/ml for dorsal telencephalon induction, 10 ng/ml for ventral telencephalon-MGE induction; R&D systems), EGF (20 ng/ml, EMD Millipore), IGF1 (10 ng/ml, EMD Millipore) and small molecules SAG (10 nM and 100 nM, Enzo Life Sciences, Farmingdale, New York) and retinoic acid (RA) (10 µM, Sigma Aldrich) were added to the neural induction media in specific combinations (Figure 2A). Neural rosettes were isolated by mechanical picking and re-plated for neuronal differentiation. For rosettes harvested from the dorsal telencephalon patterning scheme, an extra expansion step was included, where the neural progenitors were allowed to proliferate in suspended neurospheres over seven days (Figure 2—figure supplement 1). The neuronal differentiation was divided into an initial stage (20 days) and a secondary, astrocyte- supported stage (~2 months), where the immature human neurons were co-cultured with a rat cortical astrocyte monolayer. The neuronal differentiation media contained Neurobasal-A (Base medium, Gibco), B-27 without Vitamin-A (2%, Gibco), GlutaMax (1%, Gibco), and Penstrep (100 units/mL working concentration, Gibco), recombinant human BDNF (10 ng/ml, PeproTech, Rocky Hill, New Jersey) and recombinant human NT-3 (10 ng/ml, PeproTech). Additional neurotrophic or signaling molecules were used for the initial stage of inhibitory neuron differentiation. Between Differentiation Day 15 and 25 (Figure 2—figure supplement 1), we used BDNF (10 ng/ml, extra, PeproTech) and NT-3 (10 ng/ml, extra, PeproTech), GDNF (10 ng/ml, R&D systems), cyclic AMP (0.5 mM, Sigma Aldrich) and BMP4 (20 ng/ml, R&D systems); between Differentiation Day 25 and 35, we used BDNF (10 ng/ml, extra, PeproTech) and NT-3 (10 ng/ml, extra, PeproTech), GDNF (10 ng/ml, R&D systems), cyclic AMP (0.5 mM, Sigma Aldrich) and IGF1 (10 ng/ml, EMD Millipore). The rat cortical astrocytes were prepared from newborn to P2 rat pups, using a procedure modified from a previous publication (McCarthy and de Vellis, 1980), according to a lab protocol approved by the administrative panel on laboratory animal care (APLAC) at Stanford University. At the transition to astrocyte-supported neuronal differentiation, human neurons were dissociated with Accutase and plated at the density of 10,000 cells/cm2.
This procedure was carried out according to a lab protocol approved by the administrative panel on laboratory animal care (APLAC) at Stanford University. The culture of rat hippocampal slices was prepared and maintained according to a standard method (Hanson, et al., 2010). In brief, bilateral hippocampi from P6 Sprague Dawley rat pups were isolated and sliced to the thickness of 400 microns. The slices were laid on top of the PTFE membrane of a Millipore millicell culture insert, which was placed in a media filled dish to form an air-liquid interface. CaMKII-ChR2-YFP labeled human inhibitory neurons were concentrated to 35,000 cells/µl, and 1.5 µl (~50,000 cells) were added to the top surface of each slice. The slices were maintained in 5% CO2, at 37°C for 3 days and then at 34°C for two months. The culture media comprises minimum essential medium (MEM, Gibco)/HBSS (Gibco) (2/1 mixed as the base medium), Penstrep (100 units/mL working concentration, Gibco), HEPES (12.5 mM diluted from 1M stock solution, Gibco), heat-inactivated horse serum (25%, Gibco). To drive activity of ChR2, 470-nm light from an LED was pulsed through the 40x immersion lens, under the control of programmed digitizer output. Patch clamp recordings of the grafted human neurons or the host rat neurons were carried out at room temperature and in continuous perfusion of artificial cerebrospinal fluid (ACSF) bubbled with 95% O2, 5% CO2. The GABAA receptor competitive antagonist Gabazine (SR 95,531 hydrobromide, Tocris) was dissolved in DMSO and used at a final concentration of 10 µM.
The CaMKII-GFP vector contains a 1.3 kb promoter of the mouse α-calcium/calmodulin-dependent protein kinase II (Dittgen et al., 2004). The vector was generated during a previous study in the lab (Shcheglovitov et al., 2013). The Dlxi1/2b-GFP vector contains the mouse Dlxi1/2b enhancer, a β-globin minimal promoter, and the EGFP coding sequence in tandem. The vector is a variant of the Dlxi1/2b-β-globin-mCherry vector described previously (Chen et al., 2013). The CMV-tRFP-Nav1.1-shRNA vectors were made by replacing the SnaBI-tGFP-XhoI fragment of the pGIPZ-Nav1.1-shRNA vectors (Open Biosystems) by the SnaBI-tRFP-XhoI fragment from pLemir-NS (Open Biosystems). The original pGIPZ-Nav1.1-shRNA vectors were: vector # RHS4346, which carries the scrambled control shRNA; vectorV3LHS_360126, which carries Nav1.1-shRNA-1; vector V3LHS_360129, which carries Nav1.1-shRNA-2. The TRE-mCherry-T2A-hNav1.1-cDNA vector was made by a two-step cloning procedure. First, a mCherry-Gly-Ser-Gly-T2A-NotI fragment was recombined with the BamHI-SmaI-linearized pTight-ASCL1-N174 plasmid (a kind gift from the Crabtree lab at Stanford University) using the CloneEZ PCR cloning kit (GenSript, Piscataway, New Jersey). The resulting plasmid pTight- mCherry-T2A was linearized by NotI and recombined with the human Nav1.1 wild-type cDNA (encoding the 1997-aa isoform of Nav1.1, under an MTA from GlaxoSmithKline to the Dolmetsch lab at Stanford University), or with Nav1.1-F383S and Nav1.1-F383S-S1328P cDNAs (generated at Novartis) also using the CloneEZ PCR cloning kit. The final construct has these functional elements in tandem: Tetracycline Response Element (TRE), mCherry, T2A (mediating 'self-cleavage' during translation), and human Nav1.1 cDNA.
Lentiviruses encoding CaMKII-GFP, TRE-mCherry-T2A-Nav1.1-cDNA (wild type), or EF1α-rtTA were prepared at the Stanford University Neuroscience Gene Vector Core, using helper plasmids psPAX2 and pMD2.G. Viral titers (infectious units/ml) were determined by infecting HEK-293 cells and quantitating the amount of integrated provirus by Q-PCR using WPRE-specific probes and primers. Transduction of differentiating human neurons was carried out 4 days before re-plating to rat cortical astrocytes. The CaMKII-GFP virus was used at MOI = 1 (1 infectious unit/cell). The TRE-mCherry-T2A-Nav1.1-cDNA virus and EF1α-rtTA virus were used together at MOI = 15 (each virus). To activate TRE and drive mCheery-T2A-Nav1.1-expression, Doxycycline was added to the neuronal culture at 0.1 µg/ml, seven days after viral transduction. mCherry expression was detected after another seven days (Figure 6C).
A second group of lentiviruses encoding Dlxi1/2b-GFP or CMV-tRFP-Nav1.1-shRNAs (scrambled, shRNA1, or shRNA2) were prepared by transfecting HEK293T cells and harvesting viral supernatants 36 hr after transfection. The helper plasmids used for Dlxi1/2b-GFP were pVSV-g, pRSVr and pMDLg-pRRE (Chen et al., 2013). The helper plasmids for CMV-Nav1.1-shRNAs were pCMV-dR8.2 dvpr and pCMV-VSV-g. The Dlxi1/2b-GFP virus was further concentrated by ultracentrifugation, while the CMV-tRFP-Nav1.1-shRNAs viruses were used in the form of fresh supernatants. Tests of transduction efficiency showed that the amount of viruses produced from 1 cm2 of 80% confluent HEK293T cells can be used to transduce 400,000 differentiating human neurons.
A third group of lentiviruses encoding TRE-mCherry-T2A-Nav1.1-F383S, TRE-mCherry-T2A-Nav1.1-F383S-S1328P, or EF1α-rtTA were prepared by transfecting HEK293N cells and harvesting viral supernatants 36 hr after transfection. The helper plasmids used were psPAX2 and pMD2.G. Viral supernatants were concentrated using Lenti-X Concentrator (Clontech, Mountain View, California) and viral titers were determined with a Lenti-X qRT-PCR Titration Kit (Clontech).
H9 hESCs with stably integrated doxycycline-inducible Ngn2 and the tet-transactivator were maintained in mTESR medium on matrigel. One day prior to differentiation, cells were plated at 80,000 cells per well of a 6-well dish on hESC-qualified matrigel (Corning). The next day (day 0), medium was changed to differentiation medium (DMEM/F12, glutamax, N2 supplement, non-essential amino acids, penicillin/streptomycin, 10ng/mL BDNF, 10ng/mL NT3, and 2 µg/mL doxycycline [Dox]). On day two, the medium was changed; this time and thereafter the medium included B-27 supplement. On day three, cells were dissociated with accutase for 5 min, centrifuged for 5 min at 300 g, resuspended, and either 1 × 106 cells were seeded per well of a 6-well plate coated with matrigel or 1 × 105 cells were seeded per well of a 24-well plate containing matrigel-coated coverslips pre-seeded with rat astrocytes (Lonza). On day four, the media was changed and the dox concentration was reduced to 1 µg/mL. Partial media changes were performed on days six and eight to reduce the dox concentration to 0.5 µg/mL and 0.1 µg/mL, respectively. On day nine, cells were co-infected with equal titers of lentiviruses encoding EF1α-rtTA (3.0× 108 copies/cm2 of culture area) and equal titers of lentiviruses (1.5 × 108 copies/cm2 of culture area) encoding mCherry-T2A-Nav1.1-F383S or mCherry-T2A-Nav1.1-F383S-S1328P, and the dox concentration was raised to 2 µg /mL in half the wells. 50% media changes were done every other day with 2 µg /mL or 0.1 µg /mL dox until day 17, when the cells were lysed for western blotting or subjected to electrophysiology analysis.
All specimens were fixed at room temperature in 4% paraformaldehyde in PBS. The blocking buffer contained 10% normal goat serum, 3% bovine serum albumin, and 0.25% Triton-X 100 in PBS. The primary antibodies are listed in Supplementary file 4. The secondary antibodies were goat anti-rabbit/rat/mouse IgG conjugated with Alexa-488, −568 or −647 (Invitrogen). Images of cultured neurons were captured using a Zeiss M1 Axioscope; images of rat hippocampal slices were captured using a Zeiss spinning disc confocal microscope, equipped with 488/594/640 lasers (Perkin Elmer).
Total RNA from samples of interest was purified using the RNeasy kit (Qiagen) and the RNase-free DNase Set (Qiagen). After normalization for concentration, RNA was reverse-transcribed into cDNA using the SuperScriptIII First-strand Synthesis SuperMix and random hexamer primers (Invitrogen). Real-time PCR for the majority of the cDNA samples was performed using the FastStart Universal SYBR Green Master (Rox) mix (Roche) on a Mastercycler realplex machine (Eppendorf) or a ViiA 7 machine (Applied Biosystems). A subset of the RNA samples were obtained at a later stage of the study, and the resulting cDNAs were analyzed by a 96.96 Dynamic array (Fluidigm, South San Francisco, California) and the EvaGreen detection method as described previously (Paşca et al., 2011). This subset includes cDNAs for the neuronal cultures from the control iPSC line 8858–3 and Dravet iPSC line 6358–2, as well as cDNAs for HEK293T cells used to assess Nav1.1-shRNAs. All real-time PCR data were quantified relative to a reference gene by the delta-delta Ct method. All PCR primers are listed in Supplementary file 3.
At the end of a patch clamp recording, the cell soma was collected by a strong suction into the glass pipette, which contained 5 µL of internal solution. The pipette content was expelled into a 0.2 ml PCR tube, which has been prefilled with 9 μl of a pre-amplification mix, and immediately frozen on dry ice. The pre-amplification mix contains multiplexed primers for genes of interest (55 nM, each oligo) and components of the CellsDirect One-Step qRT-PCR kit (Invitrogen). Cell samples were then processed on a PCR thermocycler, during which cellular RNAs were converted into cDNAs, and the cDNAs of genes of interest were pre-amplified selectively by gene-specific primers. Pre-amplified samples were treated with ExoSAP-IT (Affymetrix) to remove residual primers and were diluted four fold with water. Realtime PCR reactions were performed on a 96.96 Dynamic array (Fluidigm) using the EvaGreen detection method, as previously described (Paşca et al., 2011). An amplicon dissociation step (melting curve analysis) was included to determine PCR specificity. Data were collected and analyzed using Fluidigm Real-Time PCR Analysis software. A reaction with its threshold cycle (Ct) below 30 and a verified melting curve indicates the presence of expression for the gene of interest. All PCR primers are listed in Supplementary file 3.
HEK293T cells were transduced with freshly produced lentiviral supernatants expressing CMV-tRFP-Nav1.1-shRNAs (scrambled, shRNA1, and shRNA2). The viral transduction was carried out twice on two sequential days, with 6 hr of incubation in viral supernatants each day. Four days after the second transduction, the tRFP reporter was visible in the majority of cells for all three lentiviruses (Figure 6—figure supplement 2A), indicating sufficient and similar levels of expression among the three shRNA constructs. At this point the cells were transfected with a human Nav1.1 expression plasmid pCAG-hNav1.1-IRES-EGFP. Three days later, total RNA was isolated and quantitative RT-PCR was performed on a 96.96 Dynamic array (Fluidigm) by the EvaGreen detection method, as previously described (Paşca et al., 2011). The pCAG-hNav1.1-IRES-EGFP plasmid was generated by inserting human Nav1.1 cDNA (provided by GlaxoSmithKline and under MTA) into the NotI site of the pCAGIG vector (Addgene Cat. 11159).
Patch clamp recordings were performed at room temperature using an EPC10 amplifier (HEKA, Germany), and signals were sampled at 10 kHz with a 3 kHz Bessel filter. The voltage-dependent sodium currents (INa) were recorded in the voltage clamp mode with P/8 leak subtraction. Traces acquired under a series resistance greater than 25 MΩ were discarded from analysis. Voltage protocols were applied to evaluate voltage-dependent activation and steady-state inactivation. Protocols details are described in Figure 4—figure supplement 1. The external solution for INa contained: 50 mM NaCl, 94 mM TEA-Cl, 10 mM HEPES, 1 mM CaCl2, 2 mM MgCl2, 0.3 mM CdCl2, 1 mM BaCl2, 20 mM Glucose, and pH was titrated to 7.4. The internal solution contained 24 mM CsCl, 110 mM Cs-Methanesulfonate, 4 mM Mg-ATP, 0.3 mM Na2-GTP, 10 mM Na2-Phospocreatine, 10 mM EGTA, 10 mM HEPES, and pH was titrated to 7.3. The predicted liquid junction potential is -12 mV. Data reflecting voltage-dependent activation (i.e. the I-V relationship) were fitted with an equation in the form of I = g*(V-VR)/(1+exp(-0.03937*z*(V-V1/2))), where I is the amplitude of the current peak, g is a factor reflecting the number of active channels in the cell recorded, V is the membrane potential under which the current develops, z is the apparent gating charge, and V1/2 is the half-maximal activation potential. Data reflecting steady-state inactivation were first normalized to the peak current under minimal inactivation (the first sweep) and then fitted with an equation in the form of I = 1/(1+exp((V- V1/2')/a')), where I is the normalized amplitude of the current peak, V is the voltage value of the inactivating pre-pulse, V1/2' is the half-maximal inactivating voltage and a' is a slope factor. Data describing the normalized channel conductance (g/gmax) of Nav1.1-F383S and Nav1.1-F383S-S1328P were fitted with g/gmax = 1/(1+exp((V-V1/2)/a)) (for voltage-dependent activation) and g/gmax = 1/(1+exp((V- V1/2’)/a’)) (for steady-state inactivation), where V1/2 and V1/2’ are half-effective membrane potentials, and a and a’ are slope factors. Neuronal action potentials were recorded in the current clamp mode. Resting membrane potential was recorded immediately after the pipette breaks in. Then neurons were held at –65 mV, and sustained current injections (duration, 1 s; increment, 10 pA) were applied to evoke action potentials. The external solution contained: 140 mM NaCl, 2.5 mM KCl, 2.5 mM CaCl2, 2 mM MgCl2, 1 mM NaH2PO4, 10 mM HEPES, 20 mM Glucose, and pH was titrated to 7.4. The internal solution contained 120 mM K-Gluconate, 20 mM KCl, 4 mM NaCl, 4 mM Mg-ATP, 0.3 mM Na2-GTP, 10 mM Na2-Phospocreatine, 0.5 mM EGTA, 10 mM HEPES, and pH was titrated to 7.25. The predicted liquid junction potential is -14.4 mV. Tetrodotoxin citrate (TTX) was solubilized in water as 10 mM stocks and used at 0.8 µM or 1 µM in external solution. Gabazine (SR 95,531 hydrobromide, Tocris) was solubilized in DMSO as 10 mM stocks and used at 10 µM in external solution. Data were initially analyzed with the Patchmaster software (HEKA). Further analysis was performed using IgorPRO, Microsoft Excel, and GraphPad Prism. The iPSC lines involved in each electrophysiology experiment are described in Supplementary file 2.
Human NgN2-induced neurons were maintained in 6-well plates until day 17. Membrane proteins were extracted using the Mem-PER Plus kit (ThermoFisher Scientific), with all processing steps carried out on ice or at 4ºC and the permeabilization and solubilization buffers supplemented with Halt protease and phosphatase inhibitor cocktail (ThermoFisher Scientific). Solubilized membrane protein samples were quantified with Pierce BCA Protein Assay kit (Thermofisher Scientific). Quantity-normalized samples were supplemented with the 4x NuPAGE LDS sample buffer (ThermoFisher Scientific) and 2% (v/v, final concentration) β-mercaptoethanol and incubated at 70°C for 5 min. The samples (12.5 µg of total membrane protein) were then loaded to a Novex 3–8% Tris-Acetate gel (ThermoFisher Scientific), together with Chameleon Duo pre-stained protein ladder (LI-COR, Lincoln, Nebraska). After gel electrophoresis, proteins were transferred to a nitrocellulose membrane using the iBlot2 system (ThermoFisher Scientific). The high-molecular-weight (above 160 kDa) half of the membrane was blotted with a Nav1.1 antibody (AB5204/RRID AB_91751, EMD Millipore; used at 1:200), and the low-molecular-weight (below 125 kDa) half was blotted with an antibody for Na/K-ATPase (ab 76020/RRID AB_1310695, Abcam, Cambridge, Massachusetts), a ubiquitous membrane protein serving as the sample loading control. Detection of both proteins was carried out with a HRP conjugated secondary antibody (W401B, used at 1:5000, Promega, Madison, Wisconsin) and the SuperSignal West Femto Chemiluminescent Substrate (ThermoFisher Scientific) on a ChemiDoc MP imaging system (BioRad).
Nonfunctional NaV1.1 familial hemiplegic migraine mutant transformed into gain of function by partial rescue of folding defectsProceedings of the National Academy of Sciences of the United States of America 110:17546–17551.https://doi.org/10.1073/pnas.1309827110
Specific deletion of NaV1.1 sodium channels in inhibitory interneurons causes seizures and premature death in a mouse model of Dravet syndromeProceedings of the National Academy of Sciences of the United States of America 109:14646–14651.https://doi.org/10.1073/pnas.1211591109
Lentivirus-based genetic manipulations of cortical neurons and their optical and electrophysiological monitoring in vivoProceedings of the National Academy of Sciences of the United States of America 101:18206–18211.https://doi.org/10.1073/pnas.0407976101
Preferential inactivation of Scn1a in parvalbumin interneurons increases seizure susceptibilityNeurobiology of Disease 49:211–220.https://doi.org/10.1016/j.nbd.2012.08.012
Efficient induction of transgene-free human pluripotent stem cells using a vector based on Sendai virus, an RNA virus that does not integrate into the host genomeProceedings of the Japan Academy, Series B 85:348–362.https://doi.org/10.2183/pjab.85.348
Hippocampal Slice CulturesIn: L. C Doering, editors. Protocols for Neural Cell Culture (4th ed). New York, NY: Springer. pp. 299–311.
Modeling Dravet syndrome using induced pluripotent stem cells (iPSCs) and directly converted neuronsHuman Molecular Genetics 22:4241–4252.https://doi.org/10.1093/hmg/ddt275
The largest group of superficial neocortical GABAergic interneurons expresses ionotropic serotonin receptorsJournal of Neuroscience 30:16796–17604.https://doi.org/10.1523/JNEUROSCI.1869-10.2010
Dravet syndrome patient-derived neurons suggest a novel epilepsy mechanismAnnals of Neurology 74:128–139.https://doi.org/10.1002/ana.23897
Cell-type-dependent molecular composition of the axon initial segmentJournal of Neuroscience 28:14329–14340.https://doi.org/10.1523/JNEUROSCI.4833-08.2008
Subcortical origins of human and monkey neocortical interneuronsNature Neuroscience 16:1588–1597.https://doi.org/10.1038/nn.3536
Prospective isolation of cortical interneuron precursors from mouse embryonic stem cellsJournal of Neuroscience 30:4667–4675.https://doi.org/10.1523/JNEUROSCI.4255-09.2010
Altered function of the SCN1A voltage-gated sodium channel leads to gamma-aminobutyric acid-ergic (GABAergic) interneuron abnormalitiesJournal of Biological Chemistry 285:9823–9834.https://doi.org/10.1074/jbc.M109.078568
Preparation of separate astroglial and oligodendroglial cell cultures from rat cerebral tissueJournal of Cell Biology 85:890–902.https://doi.org/10.1083/jcb.85.3.890
Three groups of interneurons account for nearly 100% of neocortical GABAergic neuronsDevelopmental Neurobiology 71:45–61.https://doi.org/10.1002/dneu.20853
Molecular taxonomy of major neuronal classes in the adult mouse forebrainNature Neuroscience 9:99–107.https://doi.org/10.1038/nn1618
Production and organization of neocortical interneuronsFrontiers in Cellular Neuroscience 7:221.https://doi.org/10.3389/fncel.2013.00221
Impaired excitability of somatostatin- and parvalbumin-expressing cortical interneurons in a mouse model of Dravet syndromeProceedings of the National Academy of Sciences of the United States of America 111:E3139–E3148.https://doi.org/10.1073/pnas.1411131111
Genetic control of dorsal-ventral identity in the telencephalon: opposing roles for Pax6 and Gsh2Development 127:4361–4371.
The origin and specification of cortical interneuronsNature Reviews Neuroscience 7:687–696.https://doi.org/10.1038/nrn1954
Gsh2 and Pax6 play complementary roles in dorsoventral patterning of the mammalian telencephalonDevelopment 128:193–205.
William CatterallReviewing Editor; University of Washington, United States
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your work entitled "Deficit in inhibitory neuron activity and amelioration by Cannabidiol in a human cellular model of Dravet Syndrome" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by William A. Catterall as the Reviewing Editor and Gary Westbrook as the Senior Editor. One of the two reviewers involved in the review of your submission has agreed to reveal their identity: Reviewer, Scott Baraban.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Dravet Syndrome is an intractable genetic epilepsy syndrome with substantial premature death and severe co-morbidities, which are caused by mutations in the SCN1A gene encoding Nav1.1 voltage-gated brain sodium channel. Induced pluripotent stem cell methods are used here to determine the functional impact of a Dravet Syndrome mutation. Previous published work on iPSC-derived neurons has given inconsistent results. The highlights of this work are: 1. well-developed and well-defined iPSC-derived neurons; 2. evidence that a Dravet Syndrome mutation impairs the sodium currents and excitability of inhibitory neurons; 3. evidence that excitatory pyramidal cells are unaffected; and 4. evidence that the nonpsychotropic marijuana component cannabidiol inhibits excitability of excitatory neurons and increases excitability of inhibitory neurons.
Summary and essential revisions:
The recommendation of the reviewers and guest editor is for major revisions and resubmission.
1) The work on cannabidiol is considered to be premature and tangential to the main thrust of this manuscript, and therefore we recommend that it be developed as a separate manuscript for submission elsewhere.
2) The reviewers and guest editor are also concerned that, as now written, the manuscript seems relevant to only a single Dravet Syndrome mutation out of nearly one thousand. However, the significance of the work could be substantially enhanced by demonstrating that this mutation is a complete, or nearly complete, loss-of-function mutation, because the majority of Dravet Syndrome mutations are thought to be complete loss-of-function. This study could therefore serve as a model for this large set of disease mutations. This question could be addressed by introducing a single-residue mutation to block tetrodotoxin inhibition, expressing the mutant and an appropriate control in your iPSC-derived neurons, blocking endogenous sodium channels with tetrodotoxin, and determining the functionality of the toxin-resistant Dravet Syndrome mutant. Several groups have used this approach successfully in other contexts.
3) An additional point of interest is to determine whether this mutant is expressed as a protein at all, in order to assess whether dominant negative effects of the mutant on wild-type sodium channels are possible through association with a limiting supply of auxiliary subunits or through other means.
4) Interpretation of these interesting findings, in the context of an intact network, requires a more complete understanding of the precise interneuron identity. Although this may seem a bit trivial at first glance, it is critical to how the two major findings presented here are interpreted. We have learned much about interneuron diversity and interneuron origins since the initial 2006 Catterall et al. identification of a defect in firing associated with a GABA-expressing bipolar neuron from Nav1.1 mutant mice. Rubenstein, Fishell, Marin and many others have carefully classified GABA-expressing neurons into a wide variety of sub-populations with distinct embryonic origins. From this literature, and looking at the data presented here – putative interneurons with a telencephalic identity express GABA or GAD (30-50%), and of these 50% express calretinin with a very small percentage expressing SST and none expressing PV. This would suggest that the iPS protocol used enriches for GABA-expressing CR+ interneurons and would be more in line with a Caudal (not Medial) Ganglionic Eminence origin designation; current-clamp firing properties shown in Figures 3G and 5E are also consistent with a CR+ Regular-Spiking Non-Pyramidal cell designation. Interneurons derived from the MGE are primarily parvalbumin+ (~65%) and somatostatin+ (~35%) sub-types and it is rare to have CR+ interneurons derived from an MGE lineage (Anderson et al. Cereb. Cortex 2002; Xu et al. J. Neurosci. 2004; Butt et al. Neuron 2005; Fishell et al. Novartis Found. Symp. 2007; Gelman et al. Jaspers 2012). It would be helpful to know if the interneurons shown here also express 5HT3a receptors (see below), VIP or CoupTFII, as this would further help in distinguishing interneuron sub-populations.
5) If the telencephalic interneurons shown here are primarily CGE-derived CR+ interneurons, an interneuron cell population that can target other interneurons (Freund and Buzsaki, Hippocampus 1996; Gonchar and Burkhalter, Cereb. Cortex 1999; Chamberland et al. Front. Cell Neurosci. 2010; Urban et al. Acta Biol. Hung. 2002), then the interpretation that DS-derived interneurons have reduced sodium current and reduced firing would functionally result in a loss of inhibitory tone onto inhibitory neurons or an overall disinhibition placed in the context of a compete network environment. This would not be a reasonable explanation for epilepsy in DS, as it would be associated with increased network excitability and likely an increased propensity toward seizures. Nor would this be entirely consistent with mouse data from Catterall and other groups who have demonstrated that Cre-mediated inactivation of Nav1.1 in SST+ or PV+ interneuron sub-populations (which primarily target excitatory neurons) can result in epileptic phenotypes.
6) Firing properties depicted are not entirely consistent with those expected for mature interneurons or principal cells. First, the representative traces in Figure 5 do not match very well with the quantitative input-output plots. The excitatory neuron appears to exhibit spike frequency accommodation, as expected in the plot but not the sample traces. Whereas the control interneuron, which would not be expected to show SFA does appear to accommodate. Second, the spike firing rates shown for interneurons and excitatory neurons appear equivalent. But mature interneurons should have much higher firing frequencies.https://doi.org/10.7554/eLife.13073.041
- Yishan Sun
- Ricardo E Dolmetsch
- Thomas Portmann
- Yishan Sun
- John LR Rubenstein
- Yishan Sun
- John LR Rubenstein
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
The authors are grateful to the family who participated in this study. We thank Jennifer M Perry (clinical research associate) for her assistance in enrolling the family. Philip Manos, Dr. Arnaud Lacoste, and Dr. YounKyoung Lee kindly shared control iPSC line WT-GM08402A-SeV-hiPSC 1, which we used to generate the subclone 8402-2 WT9. WT-GM08402A-SeV-hiPSC 1 was derived from the fibroblast cell line GM08402 from the NIGMS Human Genetic Cell Repository at the Coriell Institute for Medical Research. We thank Maria L Fabian and Dr. Ben A Barres for providing rat cortical astrocytes, Dr. Lief Fenno and Dr. Karl Deisseroth for the pLenti-Syn1-ChR2(H134R)-YFP virus, and Dr. Michael Lochrie and the Stanford University Neuroscience Gene Vector core for generating and titering other lentiviral vectors. The wild-type human Nav1.1 cDNA described in Figure 6 was acquired from GlaxoSmithKline (GSK) and used in the Dolmetsch lab at Stanford University under an MTA between GSK and Stanford. The authors acknowledge Dr. Massimo Mantegazza, Dr. Andrew Powell, Dr. Mike Bird, Dr. Jane Lewis, and Dr. Jaysen Rajkomar for their help with material transfer. The Nav1.1 cDNAs encoding p.F383S and p.F383S-p.S1328P mutations was synthesized de novo at Genewiz and Novartis with help from Dr. Paul Groot-Kormelink. We also thank Dr. Alex Shcheglovitov and Dr. Jason Bant for their advice on electrophysiology techniques, as well as Dr. Masayuki Yazawa, Dr. Ajamete Kaykas, Dr. Shengda Lin, Jesse Mull, Marie Sondey, Erica Seigneur, David M Lipton, Le An, Alex Sosa, Samantha Madala, Kimberly Hsu, and members of the Dr. David Prince’s lab and Dr. John Huguenard’s lab for scientific discussions and assistance with various lab techniques. This work was supported by private funds from the Hsiao family (to YS), Stanford University School of Medicine Dean’s postdoctoral fellowship (to YS), the Epilepsy Foundation postdoctoral fellowship (to YS), the Swiss National Science Foundation postdoctoral fellowship (PBSKP3-123434, to TP), Grants MH049428, RB2-1602 from California Institute for regenerative Medicine (CIRM) to JLRR, Grant RT2-01906 from CIRM to RED.
Human subjects: All experiments involving primary fibroblasts and iPSCs from the Dravet Syndrome patients and healthy subjects were carried out at Stanford University after informed consent and under study protocols (No. 12481, No. 232, and No. 327 ) approved by the institutional review board (IRB) and Stem Cell Research Oversight (SCRO) of Stanford University. For drafting and revising the manuscript, authors currently affiliated with Novartis (Y.S., C.G., K.A.W., W.G.,K.C.,and R.E.D.) analyzed iPSC-related data that had been previously generated at Stanford University. They also incorporated new data from a control iPSC line (8402-2 WT9) that was generated at Novartis from fibroblast cell line GW08402, in accordance with the conditions of the NIGMS Repository Samples, governed by the Coriell IRB in accordance with DHHS regulations, as outlined in the MTA between Coriell and Novartis.
Animal experimentation: In this study primary cultures of rat cortical neurons, rat cortical astrocytes, and rat hippocampal organotypic slices were prepared in strict accordance with lab protocols granted to the Novartis neuroscience department (No. 14 NSC 024), the lab of Dr. Ben A. Barres (No. 10726), and the lab of Dr. Daniel V. Madison (No. 10565), respectively, under the governance of the institutional animal care and use committee (IACUC) of Novartis and the Administrative Panel on Laboratory Animal Care (APLAC) of Stanford University.
- William Catterall, Reviewing Editor, University of Washington, United States
- Received: November 15, 2015
- Accepted: June 30, 2016
- Version of Record published: July 26, 2016 (version 1)
© 2016, Sun et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.