Developmental Regulation of Human Embryonic Stem Cell-Derived Neurons by Calcium Entry via Transient Receptor Potential (Trp) Channels

Abstract

Spontaneous calcium (Ca²⁺) transients in the developing nervous system can affect proliferation, migration, neuronal subtype specification and neurite outgrowth. Here we show that telencephalic human neuroepithelia (hNE) and post-mitotic neurons (PMNs) generated from embryonic stem cells display robust Ca²⁺ transients. Unlike previous reports in animal models, transients occurred by a Gd³⁺/La³⁺-sensitive, but thapsigargin- and Cd²⁺-insensitive mechanism, strongly suggestive of a role for transient receptor potential (Trp) channels. Furthermore, Ca²⁺ transients in PMNs exhibited an additional sensitivity to the TrpC antagonist SKF96365 as well as shRNA-mediated knock down of the TrpC1 subunit. Functionally, inhibition of Ca²⁺ transients in dividing hNE cells led to a significant reduction in proliferation, while either pharmacological inhibition or shRNA-mediated knockdown of the TrpC1 and TrpC4 subunits significantly reduced neurite extension in PMNs. Primary neurons cultured from fetal human cortex displayed nearly identical Ca²⁺ transients and pharmacological sensitivities to Trp channel antagonists. Together these data suggest that Trp channels present a novel mechanism for controlling Ca²⁺ transients in human neurons and may offer a target for regulating proliferation and neurite outgrowth when engineering cells for therapeutic transplantation.

Introduction

In animal models spontaneous calcium (Ca²⁺) oscillations play a significant role in nervous system development, involved in neural induction, proliferation, migration, axon guidance and growth cone morphology [1-5]. In Xenopus neurons, the frequency of Ca²⁺ “spikes” mediated by T-type Ca²⁺ channels controls neurotransmitter specificity, while slow Ca²⁺ “waves” appear critical for neurite outgrowth [1]. Alternatively, in rodent cortical progenitors, Ca²⁺ transients are mediated by release from intracellular stores [6], and simultaneous Ca²⁺ transients between ventricular zone progenitors coordinate cell cycle entry [5]. Precisely how Ca²⁺ controls these disparate processes remains unknown, but the source of Ca²⁺ appears to be critical as different Ca²⁺ entry mechanisms can activate specific downstream signaling cascades [7].

In addition to entry through voltage-gated calcium channels (VGCCs) and release from intracellular stores, members of the transient receptor potential (Trp) channel superfamily present an alternative mechanism for Ca²⁺ entry and regulate multiple processes in the developing and mature nervous system. Trp channels are a family of 28 non-selective cation channels, and all except TrpM4 and TrpM5 display varying degrees of calcium permeability [8]. Interestingly, members of the canonical Trp (TrpC) family are involved in store operated Ca²⁺ entry [9] which is thought to be an essential component in the establishment of intracellular Ca²⁺ fluctuations [10,11]. In the mature nervous system Trp channels play important roles in the processing of sensory information [12], fear-related learning and memory [13], and defects in particular channels underlie models of neurodegeneration such as cerebellar ataxia [14]. During early development members of the Trp channel families modulate neural progenitor proliferation [15] while at later stages, specific members of the TrpC family have been shown to both positively and negatively regulate neurite extension [16,17], likely due to the activation of Ca²⁺-dependent processes [18].

Whether and/or how Ca²⁺ transients affect human neuronal differentiation is not known. Human embryonic stem cells (hESCs) can be efficiently directed to proliferating neuroepithelial (hNE) cells and then to post-mitotic neurons [19], presenting an experimentally accessible tool to explore the regulatory mechanisms that underlie neuronal differentiation. Furthermore, the development of strategies to regulate hNE proliferation and PMN differentiation, including the regulation of intracellular Ca²⁺ concentrations, may allow us to better control the regenerative potential of stem cells. Therefore we sought to investigate the role and mechanism(s) of spontaneous increases in intracellular Ca²⁺ in hNE and PMNs.

Materials and Methods

Cell Culture

hESCs (H9 line, P16-30) were maintained and differentiated as described [19,20]. After 21 days of differentiation hNE clusters were treated with 0.5% Trypsin-EGTA (Invitrogen, Carlsbad, CA) for 5min, Trypsin Inhibitor (1mg/ml; Invitrogen) for 2min, and then triturated to single cells. Cells were centrifuged (1000rpm for 5min), and resuspended in neural differentiation media described elsewhere [19]. Cells were then plated onto glass coverslips at a density of 150,000 cells/gridded coverslip (Belco, Philadelphia, PA), or 75,000 cells/10 mm coverslip.

For primary cultures, 15-week-old human fetal brains were obtained from Clive Svendsen, UW-Madison in accordance with guidelines set by the National Institute of Health and the University of Wisconsin Madison Institutional Review Board for collection and use of such tissue. Cells were enzymatically dissociated with 0.5% Trypsin-EGTA, washed 3x with DMEM-F12+20%FBS, and resuspended in plating media described in [19]. After 5 days, cultures were fed with serum-free media that contained 4μm cytosine 1-β-d-arabinofuranoside (AraC) to inhibit glial growth [21]. Unless noted all chemicals were obtained from Sigma (St. Louis, MO).

Immunochemical Staining

Immunolabeling of hESC-derived neural progenitor cultures was performed according to previously established methods [20,22]. Primary antibodies used were: polyclonal β_III-tubulin (1:5000, Covance Research Products, Princeton, New Jersey), monoclonal β_III-tubulin (1:1000), monoclonal 3CB2 (1:1000, Developmental Hybridoma Studies Bank (DSHB; University of Iowa)), monoclonal Pax6 (0.5μg/ml, DSHB), polyclonal Sox2 (1:1000, R&D Systems), polyclonal FoxG1 (BF1; 1:1000, gift from Dr. Yoshiki Sasai; RIKEN, Kobe, Japan), polyclonal GFAP (1:5000, DAKO), monoclonal N-cadherin (1:5000, Santa Cruz Biotechnologies, Santa Cruz, CA), polyclonal nestin (1:300, Millipore, Billerica, MA), monoclonal vimentin (1:40, DSHB), and polyclonal BrdU (1:500, Accurate Chemicals, Westbury, NY). Secondary antibodies used were: Alexa-Fluor donkey anti-rabbit 488, Alexa-Fluor donkey anti-mouse 594, Alexa-Fluor donkey anti-goat 594 (Invitrogen), donkey anti-rat Cy3 (Jackson ImmunoResearch Laboratories, West Grove, PA), all at a concentration of 1:1000. Topro-3 (1:300) was used as a nuclear stain.

Immunoblotting

Western blots were performed essentially as described [23]. Membranes were probed with the indicated primary TRPC antibodies (Santa Cruz) and HRP-conjugated secondary antibodies (Pierce Biotechnology, Rockford, IL). Luminescent membranes were visualized on an Alpha Innotech FluorChem HD2 bioimaging system and quantified using spot densitometry analysis with AlphaEaseFC software (Alpha Innotech Corp., San Leandro, CA).

Proliferation/TUNEL Assays and Immunocytochemical Analysis

BrdU-treated cells were processed according to [24] using an anti-BrdU antibody (1:5000, Accurate Chemicals). TUNEL staining was performed according to the manufacturer's guidelines (Roche, Indianapolis, IN). Proliferation of GFP-expressing cells was assessed using a “Click-iT EdU” assay (Invitrogen) according to the manufacturer's instructions.

Acquisition and quantification of fluorescent intensities were determined using a Nikon confocal workstation (D-Eclipse C1) and EZ-C1 software (v3.5). Multiple coverslips (2-3) were prepared following the same experimental conditions and experiments were replicated three times to verify results. For quantification of hNE markers as well as BrdU and TUNEL, cells were considered positive if more than 20% of the nuclear area (defined as Topro-3⁺) contained pixels with above-threshold values. Neurite length measurements were performed using IPlab software v4.0 (BD Biosciences, Rockville, MD), on compressed z-stacks (7-10 images; 0.35μm/image) taken under 60x magnification. Length was determined by manually tracing primary neurites from the edge of the soma to the leading edge of the process. Only cells with intact nuclei were used for analysis.

Calcium Imaging

Fluo-4 and Fura-2 imaging were performed as essentially as described in [2,25], respectively. A detailed description of procedures can be found online in the supplemental materials section.

Quantitative PCR

RNA isolation and detection at day 12 and 21 was performed as described [26]. To calculate the fold change (FC) we used the following equations:

• FC=2^ΔC(t)

• ΔC(t)= (C(t)_{Target day21}-C(t)_{Reference day21})-(C(t)_{Target day12}-C(t)_{Reference day12})

PCR primers are listed in Supplemental Table 1.

RNA Inhibition

Short hairpin RNAs (shRNA) were provided by Dr. Mitchell Villereal (University of Chicago) and cloned into the LVTHM lentiviral vector provided by Dr. Didier Trono from Tronolab (http://tronolab.epfl.ch/). TrpC1 overexpression vector was provided by Dr. Mike Zhu (The Ohio State University) and TrpC4 overexpression vector was provided by Dr. James Putney (NIEHS). Lentiviral production was performed as described [27].

Results

Dissociated Human Neuroepithelia Spontaneously Reform Neural Tube-like Rosettes and Differentiate to Neurons

hESCs differentiate to forebrain neuroepithelial (hNE) within 10 days of culture without the addition of growth factors or morphogens [26]. By three weeks, spheres contain hNE and β_III-tubulin⁺ neurons [19], but analysis of these two populations is difficult due to the 3-dimensional structure. Therefore, we first asked whether hNE cells grown as a monolayer would retain their rosette-like morphology, a structure that is critical for maintaining cells in a progenitor state [28]. We found that 21 day-old dissociated hNE cells reformed neural tube-like rosettes within hours of plating (Figure 1A). Rosettes exhibited polarity similar to those observed at earlier time points [26], with strong N-Cadherin localization near the inner lumen (Figure 1B). At this time point β_III-tubulin⁺ cells were found surrounding rosettes (Figure 1A, arrows), never co-expressed the cell cycle marker Ki67 (Supplemental Figure 1A-B; n=3), and are referred to as post-mitotic neurons (PMNs).

Figure 1.

Dissociated hNE and PMNs reform rosettes and display spontaneous Ca Twenty-four hours after plating, hNE cells reformed rosette structures (DIC) while β_III-tubulin⁺ PMNs (arrows) were observed adjacent to rosettes. (B) N-Cadherin (red) expression was observed at the inner lumen of nestin⁺ cells (green) within rosettes. (C) Many cells within rosette structures expressed the radial glial marker 3CB2, whereas PMNs did not. (D) The majority of hNE cells, but not PMNs, expressed the neuroepithelial markers Sox2 and Pax6. FoxG1 was expressed in >90% of hNE and PMNs. (Ei.-iv.) Sequential images of cells loaded with Fluo-4; a subset displayed spontaneous increases in [Ca²⁺]i. (F) Overlay of cells imaged using Fluo-4 and stained for β_III-tubulin for post-hoc analysis (hNE cells=arrows; PMN=arrowhead). (G) Representative traces from images taken every second of cells shown in (E). Large [Ca²⁺]i fluctuations exhibited complex kinetics and variable amplitudes. (H) Representative traces of small-amplitude transients. (I) Mean exponential decay curve derived from small-amplitude transients (17 cells from 3 cultures). Scale bars represent 50μm. Error bars represent ± SEM.

Interestingly, robust expression of the radial glial markers 3CB2 and Vimentin was observed in hNE cells, most often associated with rosette structures (Figure 1C and data not shown). Expression of the NE markers Sox2 and Pax6 was evident in the majority of cells within rosettes as well as in cell aggregates without distinct morphology but rarely in PMNs (Figure 1D). Consistent with previous work, almost all cells at this time point expressed the forebrain marker FoxG1 (Figure 1D, [19]). Importantly, hNE progenitors are thought to remain exclusively neurogenic until 6-7 weeks, when some gliogenic precursors begin to give rise to astrocytes [19]. Thus, 21 day-old monolayer cultures of hNE cells give rise exclusively to PMNs, express forebrain markers as well as 3CB2 and Vimentin, reminiscent of developing radial glial cells that occupy the ventricular zone of mammalian cortices early in development [29-31].

Spontaneous Calcium Transients Exhibit Prolonged Durations in hNE and PMNs

To test for the presence of spontaneous Ca²⁺ transients we performed Fluo-4 Ca²⁺ imaging once per second for ten minutes. Figure 1Ei-iv shows a representative field of cells, some of which underwent Ca²⁺ transients during the imaging period. All transients observed appeared asynchronous and isolated to single cells, unlike some reports in mice that displayed coupled transients between adjacent cells [32]. Following imaging, cultures were stained for β_III-tubulin to distinguish between hNE and PMNs (Figure 1F). Figure 1G shows traces from cells in A, illustrating the variety of amplitudes and durations of Ca²⁺ transients, as well as their highly complex kinetics (Figure 1G_1-3). However, transients with smaller amplitudes exhibited more regular kinetics (Figure 1H). These traces were averaged and fit with a single exponential decay that resulted in a τ value of 22.2±3.6 seconds, and a mean duration of approximately sixty seconds (Figure 1I). Interestingly, this is almost six times longer than the mean duration of transients observed in mouse cortex [33], suggestive of a unique mechanism for their production. Using these data we calculated that sampling at ten second intervals would detect most Ca²⁺ transients.

Spontaneous Calcium Transients in Human NE and PMNs are Dependent on Extracellular Calcium

To determine the mechanism(s) underlying these transients we manipulated the extracellular milieu or applied standard pharmacological inhibitors. Removal of extracellular Ca²⁺ (n=4), or addition of the calcium chelators EGTA-AM (50uM; n=3) or BAPTA-AM (50uM; n=3), completely eliminated transients in both hNE and PMNs, while depletion of intracellular stores with thapsigargin had no effect (Figure 2B, C and data not shown; hNE: n=4, PMNs: n=4, p>0.05). However, robust increases in [Ca²⁺]_i were observed in both hNE and PMNs within minutes of thapsigargin addition, indicating that functional Ca²⁺ stores were present. Similar to findings in Xenopus neurons [1], high concentrations of Ni²⁺ (5mM), a non-specific inhibitor of VGCCs, almost completely abolished transients (Figure 2A-C; hNE: n=3, PMNs: n=3, p<0.001). Further, while the combined application of inhibitors of L-, P/Q-, N-, or T-type Ca²⁺ channels did not significantly reduce the incidence of transients in hNE (n=3, p>0.05), they did significantly reduce transient incidence in PMNs (Figure 2A-C; n=3, p<0.05). This effect appeared to be due to the inhibition of L-type Ca²⁺ channels as only the application of nifedipine significantly reduced transient incidence in PMNs (Figure 2C and data not shown; n=5, p<0.05). In addition, the sodium channel antagonist Tetrodotoxin (TTX) caused a small but significant reduction in Ca²⁺ transients in PMNs (Figure 2C; n=6, p<0.05). Thus, Ca²⁺ transients in hESC-derived neuronal cultures require Ca²⁺ entry from the extracellular environment and multiple Ca²⁺ entry mechanisms exist in PMNs whereas hNE cells may utilize a single mechanism.

Figure 2.

Spontaneous calcium transients are dependant on extracellular calcium but not release from intracellular stores (A) Representative traces of transient elevations of [Ca²⁺]i in untreated hNE cells (top), upon removal of extracellular Ca²⁺, application of 5mM Ni²⁺ (middle), or those treated with a cocktail of VGCC inhibitors (bottom). (B-C) Pooled data illustrating the change in the percentage of cells that underwent Ca²⁺ transients under various conditions. (C) PMNs were uniquely sensitive to nifedipine and TTX. Error bars represent ± SEM. *p<0.05, **p<0.01; student's T-tests.

Transient Receptor Potential (Trp) Channel Antagonists Differentially Inhibit Spontaneous Calcium Transients in hNE and PMNs

Because internal store release and VGCC activation did not account for the majority of spontaneous Ca²⁺ transients in hNE and PMNs (Figure 2B-C), we probed for additional mechanisms. Members of the transient receptor potential (Trp) channel family are non-selective cation channels that display varying degrees of calcium-permeability [8,12]. The majority of these channels are inhibited by trivalent metal ions Gadolinium (Gd³⁺) and Lanthanum (La³⁺). We found that application of 10μM Gd³⁺ or La³⁺ significantly attenuated transient incidence in hNE (Figure 3A; Control: n=7, Gd³⁺: n=5, p<0.01 and data not shown). Ca²⁺ transients remaining after Gd³⁺ treatment also had significantly smaller amplitudes (control: 189 ± 4% F/F_o, n=7, Gd³⁺: 139 ± 6% F/F_o, n=5, p<0.01). A concentration-response curve revealed the EC₅₀ of Gd³⁺ to be 3.8μM (Figure 3B). While application of more than 10μM Gd³⁺ (Figure 3B) almost completely abolished spontaneous Ca²⁺ transients, it tended to slowly precipitate in our culture media, making long-term incubations unfeasible (see below). Therefore, 10μM Gd³⁺ was used for the remainder of our experiments except where indicated.

Figure 3.

Spontaneous calcium transients in hNE cells and PMNs are differentially sensitive to antagonists of transient receptor potential (Trp) channels (A) Pooled data demonstrate that the percentage of hNE cells that displayed Ca²⁺ transients was significantly decreased upon application of the Trp channel antagonist Gd³⁺ (10μM). (B) Concentration-response curve for the effect of Gd³⁺ on the percentage of cells that underwent Ca²⁺ transients during a 15min. period. (C) Pooled data demonstrate that application of Gd³⁺ and SKF96365 significantly reduced the percentage of PMNs with Ca²⁺ transients. (D) Representative images displaying maximum Fluo-4 fluorescence intensity over a 15min. imaging period with and without Trp channel antagonists. (E-F) Representative traces of Ca²⁺ transients in hNE cells (E) and PMNs (F) that received control treatment, Gd³⁺, or SKF96365. (G) Representative confocal images of primary human cortical cultures of 15 gestational week-old tissue show differentiated astrocytes (GFAP; green) and PMNs (β_III-tubulin; red). (H) Representative traces of Ca²⁺ transients in human primary neurons with and without SKF96365 (5μM). (F) Pooled data demonstrate that Gd³⁺, SKF96365, and Nifedipine (5μM) decreased the percentage of cells with Ca2+ transients in human primary neurons. Scale bars represent 50μm.

Error bars represent ± SEM. *p<0.05, **p<0.01, ***p<0.001; student's T-tests.

To determine which specific Trp channel(s) mediate the observed Ca²⁺ transients we applied two antagonists believed to inhibit specific subfamilies. Ruthenium Red is used as an inhibitor of members of the TrpV family [34,35], while SKF96365 is used to inhibit TrpC family members [36,37]. Application of 5μM Ruthenium Red (n=3) or 5μM SKF96365 (n=5) had no significant effect on the percentage of hNE cells that displayed Ca²⁺ transients (Figure 3A, D arrowheads, E; p>0.05). Interestingly, while Ca²⁺ transients in PMNs displayed the same sensitivity to Gd³⁺ as hNE cells (compare Figures 5A and C; n=5, p<0.01), they were also significantly inhibited by SKF96365 (Figure 3C, D arrow, F; n=5, p<0.01), suggestive of a role for TrpC channels in PMNs (see below). Ruthenium Red, however, was without effect in PMNs (Figure 3C; n=3, p>0.05).

Figure 5.

TrpC channels contribute to neurite outgrowth in human neurons (A-B) Confocal images of hES-derived PMNs from control and SKF96365-treated cultures. (C) Pooled data demonstrate that SKF96365 treatments significantly reduced mean neurite length. (D-E) Confocal images of human primary cultures from control and SKF96365-treated cultures. (F) Pooled data show that neurite extension in human primary neurons was also significantly inhibited by SKF96365. (G) Pooled data from quantitative PCR experiments illustrate the relative amounts of mRNA for individual TrpC subunits in hNE cultures (day 12) and mixed hNE+PMN cultures (day 21). (H) Immunoblots of various TrpC proteins performed on samples from postnatal day 1 mouse brain (Brain), hNE cells (day 12), and PMNs+hNE (day 21). 40μg protein per lane. Scale bars represent 50μm. Error bars represent ± SEM. **p<0.01; student's T-tests.

To further characterize the Trp channel(s) and/or signaling pathways responsible for the induction of Ca²⁺ transients in hNE cells, we applied a host of pharmacological agents (Table 1). However, no treatment resulted in visible inhibition or augmentation of Ca²⁺ transients ruling out the involvement of connexin hemichannels, purinergic receptors, arachidonic acid (ARC) channels, ryanodine receptors, stretch-activated channels (SACs), nicotinic acetylcholine receptors (nAChRs), GABA/Glutamate receptors and other VGCCs, all of which can cause direct or indirect Ca²⁺ influx. Thus, for hNE cells we were unable to identify specific channels responsible for triggering Ca²⁺ transients.

Table 1.

Pharmacoligcal treatments with no observable effects on the incidence of calcium transients in hESC-derived neuronal cultures.

Antagonist	Concentration	Target	Reference(s)
2-APB	100μM	IP3 receptors TrpC5/TrpM2	[64-66]
Amiloride	1mM	TrpA1/ TrpP3/ TrpML1	[50,67,68]
AP5/CNQX	25μM/10μM	NMDA/AMPA receptors	[69,70]
Cadmium	10μM	VGCCs	[71]
d-Tubocurarine/α- Bungarotoxin	5 &500μM/100nM	Nicotinic Acetylcholine Receptors (nAChRs)	[72,73]
Flufenamic Acid	100μM	TrpM2/ TrpM4/ TrpM5/ TrpP3/ TrpC5/ TrpC6	[64,74-76]
Gentamicin/GsMTx4	100μM/5μM	TrpA1 & Stretch activated channels (SACs)	[67,77,78]
Picrotoxin/CGP 46381	50uM ea.	GABAA-C receptors	[79,80]
Ryanodine	50μM	Ryanodine Receptors	[81]
Spermine	500μM	TrpM4/ TrpM5/ TrpM7	[76,82]
Suramin	100μM	Purinergic Receptors	[5]
Agonist	Concentration	Target	Reference(s)
Arachidonic Acid	10μM	ARC Channels/ Trp/ TrpL/ TrpV4	[83-85]
Caffeine/Carbachol	10mM/10-100μM	Ryanodine receptors	[81]
OAG	100μM	TrpC3/ TrpC6	[86,87]
Sphingosine	50μM	TrpM3	[88]

To determine whether the SKF96365-sensitive Ca²⁺ transients in PMNs were an artifact of the differentiation of neurons from hESCs, we cultured human embryonic cortical neurons from a 15 week-old fetus. Figure 3G shows fetal cultures after one week, which were mainly comprised of PMNs (red) and astrocytes (green). Fluo-4 Ca²⁺ imaging revealed the presence of spontaneous transients which displayed a similar sensitivity to Trp antagonists when compared to hESC-derived neurons (Figure 3H, I; control: 225 cells from 9 coverslips, Gd³⁺: 153 cells from 6 coverslips, p<0.001; SKF96365: 313 cells from 8 coverslips, p<0.001). In addition, nifedipine significantly reduced the incidence of Ca²⁺ transients in human primary cultures while thapsigargin was without effect (Figure 3I; nifedipine: 175 cells from 5 coverslips, p<0.05, Thapsigargin: 179 cells from 4 coverslips, p>0.05). SKF96365 appeared to have no effect on spontaneous Ca²⁺ fluctuations in glial cells which were similar to those observed in primary rodent cells [38]. Together, these data show that spontaneous Ca²⁺ transients in hESC-derived neurons are nearly identical to those taken directly from human embryonic tissue and that neither the presence of glia or treatment with AraC modified their pharmacological sensitivity.

Inhibition of Trp Channels Significantly Attenuates Proliferation in hNE cells

In rodent and chick neural progenitor cells, Ca²⁺ has been shown to act at different stages of the cell cycle to promote cellular proliferation [32,39]. In human cells, application of 10μM Gd³⁺ significantly reduced the number of cells that incorporated BrdU during a 12-hour treatment (Figure 4A-B; 1606 cells from 3 coverslips, p<0.001). Similar data were obtained using the cell-cycle marker Ki67 (Supplemental Figure 1), suggesting that Gd³⁺ was not simply interfering with BrdU incorporation.

Figure 4.

Trp channel inhibition significantly reduces proliferation of hNE cells (A) Representative confocal images showing BrdU incorporation after 12-hour control, Gd³⁺ (10μM), or Ni²⁺ (500μM) treatments. (B) Pooled data demonstrate that 10μM Gd³⁺ and 500μM Ni²⁺ significantly decreased BrdU incorporation. (C) Representative confocal images of TUNEL-stained cells that were untreated or treated with 10μM Gd³⁺ or 500μM Ni²⁺. (D) 5μM SKF96365 and 500μM Ni²⁺ significantly increased cell death. Scale bars represent 50μm. Error bars represent ± SEM. **p<0.01, ***p<0.001; student's T-tests.

The Gd³⁺ effect appeared to be specific, as Trp antagonists that did not affect Ca²⁺ transients (1μM Gd³⁺, SKF96365, and Ruthenium Red (Figure 3A-B)), also did not affect the number of hNE that incorporated BrdU (Figure 4B; 1μM Gd³⁺: 1084 cells from 3 coverslips, SKF96365: 1185 cells from 3 coverslips, Ruthenium Red: 1338 cells from 3 coverslips, p>0.05). TUNEL staining revealed no significant difference between cells treated with 10μM Gd³⁺ and untreated controls (Figure 4C, D; Control: 1150 cells from 3 coverslips, Gd³⁺: 1599 cells from 3 coverslips, p>0.05), suggesting that this was not due to increased cell death. Ni²⁺ was used at two different concentrations to control for non-specific effects of long-term incubations of extracellular metals. 10μM Ni²⁺ had no effect on BrdU incorporation (1485 cells from 3 coverslips, p>0.05) or TUNEL staining (1261 cells from 2 coverslips, p>0.05). However, 500μM Ni²⁺ caused a small but significant decrease in BrdU incorporation (Figure 4B; 1220 cells from 2 coverslips, p<0.01), although this concentration also led to a 4-fold increase in cell death (Figure 4D, E; 1139 cells from 2 coverslips, p<0.001). This suggests that high concentrations of Ni²⁺, like those that are used to inhibit Ca²⁺ transients ([40], Figure 3), likely have non-specific toxic effects that result in changes in BrdU incorporation.

Blockade of TrpCs Reduces Neurite Outgrowth from Human Neurons

In PMNs, TrpCs have been shown to serve many functions during development including the control of growth cone guidance and neurite extension [16,41]. We found that incubation of hESC-derived cultures for 12 hours with SKF96365 significantly reduced mean neurite length compared to controls (Figure 5A-C; Control: n=5, SKF96365: n=5, p<0.001). In contrast, while nifedipine significantly reduced Ca²⁺ transient incidence (Figure 2C), it did not affect neurite length (Figure 5C; n=5, p>0.05). Furthermore, this was likely not due to a non-specific effect of SKF96365 on cell health (Figure 4D; 1140 cells from 3 coverslips, p<0.01), as 500μM Ni²⁺, which also caused a significant increase in cell death (Figure 4D), did not affect neurite length (Supplemental Figure 2). These results were verified in human primary cultures, where 12-hour treatments with SKF96365 reduced mean neurite length by approximately 50% compared with untreated or nifedipine-treated neurons (Figure 5D-F; Control: 123 cells from 3 coverslips, SKF96365: 182 cells from 3 coverslips, p<0.001; nifedipine: 95 cells from 3 coverslips, p>0.05).

To determine which TrpCs may be involved, we examined transcript expression levels in both hNE cells (day 12) and after PMNs began to differentiate (day 21, [19,26]). Quantitative RTPCR (qPCR) revealed mRNA expression of TrpC1, C3, and C4 in both populations, and low levels of TrpC6 transcripts were detected at day 21 (Figure 5G). In contrast, western blots revealed no expression of TrpC1 and moderate levels of TrpC4 protein at day 12. However, both subunits were strongly expressed at day 21 (Figure 5H), indicating an upregulation of both proteins in PMNs. While mRNA was detected for other TrpCs (e.g. TrpC3 and TrpC6) at various time points during differentiation (Figure 5G), little or no protein expression was detected (Figure 5H). Together these data demonstrate that TrpC-dependent regulation of neurite outgrowth in human neurons appears to be reproduced in hESC-derived neurons and that TrpC1 and TrpC4 are the primary subunits expressed in early human forebrain neurons.

TrpC1 and TrpC4 Regulate Neurite Extension in hESC-derived Neurons

To test the roles of specific TrpC subunits, targeted reduction of either TrpC1 or TrpC4 was performed using shRNA [42] incorporated into a lentiviral vector. In TrpC1 or TrpC4-transfected HEK293 cells, qPCR demonstrated an 11.3±1.2 (n=3) fold decrease of TrpC1 mRNA and a 9.0±0.6 (n=3) fold decrease of TrpC4 mRNA in cells that received the respective shRNA. A scrambled shRNA was without effect. Western blots also demonstrated a significant attenuation of targeted TrpC isoforms without concomitant loss of non-specific subunits (Figure 6A; TrpC1 shRNA: 20±2.9%, n=4, p<0.001; TrpC4 shRNA: 17.4±4.7%, n=4, p<0.001). Compared to nontransfected cells, a scrambled shRNA control construct was without effect (TrpC1: 97.9±1.6%, n=3, p>0.05; TrpC4 99.4±0.9%, n=3, p>0.05).

Figure 6.

TrpC1 and TrpC4 contribute to neurite outgrowth in hESC-derived neurons (A) Western blots performed on TrpC-transfected HEK cells that also received shRNA constructs directed toward either TrpC1 or TrpC4 demonstrate specific knock down of each construct while a scrambled shRNA construct had no effect (p>0.05). 10μg protein per lane (TrpC rows); 40μg protein per lane (Actin). (B) Representative confocal images of control shRNA GFP expression in hNE cells (arrowheads) and PMNs (arrows) 10 days after viral infection. (C) Confocal images of hNE cells expressing control, TrpC1, or TrpC4 shRNA (green) that incorporated EdU during a 12-hour exposure. Pooled data revealed no significant differences between groups (p>0.05). (D) Confocal images of PMNs (β_III-Tubulin⁺) that expressed control, TrpC1 or TrpC4 shRNA (green). Pooled data demonstrate a significant decrease in mean neurite length for cells that expressed either TrpC1 or TrpC4 shRNA, while the control shRNA was without effect. Scale bars represent 50μm. Error bars represent ± SEM. *p<0.05; student's T-tests.

The addition of shRNA-containing lentiviruses to hESC-derived forebrain cells revealed robust infection of hNE cells within rosette structures (Figure 6B, arrowheads), with reduced efficiency in β_III-Tubulin⁺ PMNs (Figure 6B, arrows). Proliferation analyses revealed no significant difference in EdU incorporation (see methods) between control-infected cells and those that did not incorporate virus (Figure 6C, p>0.05). Cells that expressed either TrpC1 or TrpC4 shRNA also showed no significant difference in proliferation rates compared to controls (Figure 6C, p>0.05). These findings were not surprising as TrpC1 protein did not appear to be expressed in hNE cells, and TrpC4 has not been previously implicated in regulating cell proliferation. Importantly, these data suggest that neither the virus itself nor shRNA hairpins caused alterations in cell viability.

Interestingly, significant reductions in mean neurite length were detected in PMNs that received either TrpC1 or TrpC4 shRNA, (Figure 6D; Control: 120.9±11.4μm; TrpC1 shRNA: 93.5±14.7μm; TrpC4: 80.2±10.2μm, n=3, p<0.05), but not in cells that received control shRNA (Figure 6D; 122.5±7.8μm, n=3, p>0.05). The modest reduction in neurite length in the TrpC1 RNAi condition appeared to be due to large reductions in a subset of cells (Figure 6D, arrowheads), whereas others appeared to be unaffected (Figure 6D, arrows).

To determine whether these differences were associated with changes in intracellular Ca²⁺ dynamics we performed calcium imaging on GFP⁺ cells using Fura-2. Spontaneous Ca²⁺ transients observed using Fura-2 were similar in duration to those measured with Fluo-4, but had diminished amplitudes (Supplemental Figure 3), likely due to differences in Ca²⁺ affinity inherent to the dyes. No significant differences were found between cells infected with control shRNA and unstransfected cells (Supplemental Figure 3B; Untransfected: 20.3±2.0%, 425 cells from 4 coverslips; Control shRNA: 23.3±4.5%, 154 cells from 3 coverslips, p>0.05). However, the percentage of cells that displayed transients was significantly reduced in the population that expressed TrpC1 shRNA (Supplemental Figure 3A, B; 11.3±3.3%, 279 cells from 4 coverslips, p<0.05), but not in cells that expressed TrpC4 shRNA (17.0±4.5%, 172 cells from 3 coverslips, p>0.05).

The fact that both TrpC1 and TrpC4 shRNA independently reduced neurite outgrowth, while only TrpC1 shRNA reduced global calcium transients, suggested that whole cell Ca²⁺ signals are likely not the most critical factor for human neurite extension. Previous studies have shown that local Ca²⁺ signals within growth cones are critical for morphology, steering, and extension/retraction (reviewed in [43]). Consistent with this we found that 1hr. treatment of cells with BAPTA, which is able to chelate Ca²⁺ proximal to the plasma membrane [44,45], significantly reduced neurite outgrowth compared to untreated or EGTA-treated controls (control: 101.6±4.7μm; EGTA: 106.2±5.9μm; BAPTA: 80.0±4.9μm, n=4 ea., p<0.05). Furthermore, this occurred despite the fact that both BAPTA and EGTA completely blocked global calcium transients (above). Taken together, these data suggest that TrpC1 plays a role in both local and global Ca²⁺ signals in human neurons, but that only local Ca²⁺ signals via TrpC1 and TrpC4 channels can regulate neurite extension.

Discussion

Here we show that hNE and PMNs undergo spontaneous Ca²⁺ transients mediated by Gd³⁺/La³⁺-sensitive channels, suggestive of a role for transient receptor potential (Trp) channels. Ca²⁺ transients displayed relatively long durations, similar to Ca²⁺ “waves” observed in animal models [40]. While blockade of Ca²⁺ transients in hNE cells led to a significant reduction in progenitor cell proliferation, they were inhibited only by broad spectrum antagonists and their precise mechanism remains unknown. In contrast, PMNs from hESCs or primary fetal tissue exhibited L-type Ca²⁺ channel- and TrpC1-dependent Ca²⁺ transients. Furthermore, pharmacological inhibition or downregulation of TrpC1 and TrpC4 proteins caused significant reductions in neurite outgrowth. Collectively, our data demonstrate that Trp channels play a critical role in the generation of spontaneous Ca²⁺ transients in human neurons and their progenitors, and that Trp channels regulate multiple aspects of human neural development.

In the nervous system Trp channels are critically important for somatosensory transduction of a wide array of stimuli [8,12], fear-related learning and memory [13], visual information processing [46,47], and the survival of multiple cerebellar neurons [14,48]. However, while the importance of spontaneous whole cell Ca²⁺ fluctuations in CNS development has been demonstrated in animal models, to our knowledge this is the first report of a Trp channel-dependent mechanism. For instance, neural induction in Xenopus and Pleurodeles requires Ca²⁺ entry through L-type VGCCs, the expression of which correlates with the development of spontaneous Ca²⁺ oscillations [49]. Calcium fluctuations found in the developing ventricular zone (VZ) of the mouse cortex are dependent upon IP₃-mediated Ca²⁺ release from intracellular stores [33]. Furthermore, Gu and Spitzer (1994) described two distinct types of Ca²⁺ transients in the developing Xenopus spinal cord. Short duration Ca²⁺ “spikes” mediated by T-type Ca²⁺ channels regulate neurotransmitter specification, while long duration Ca²⁺ “waves”, which are blocked by mM concentrations of Ni²⁺, appear necessary for normal neurite extension [40]. However, the induction mechanism for Ca²⁺ “waves” has remained uncharacterized. It is interesting to speculate that the Trp channel-dependent Ca²⁺ transients we observed in human cells are the same as Ca²⁺ “waves” observed in Xenopus neurons as their pharmacological sensitivity appears similar.

Although evidence in neurons is limited, Trp channel-dependent Ca²⁺ oscillations have been observed in many non-excitable cells. In rabbit vascular smooth muscle, phenylephrine- or endothelin-1-induced Ca²⁺ oscillations occur by an SKF96365-sensitive mechanism [50,51]. In a stimulated T-lymphocyte line, expression of dominant negative TrpM4 subunits converted oscillatory Ca²⁺ signals to sustained increases in [Ca²⁺]_i [52]. Similarly, an immortalized B-cell line lacking TrpC1 demonstrated significantly reduced Ca²⁺ oscillations and capacitative Ca²⁺ entry in response to antibody stimulation [53]. Thus, it appears that Ca²⁺ transients in human forebrain cells may occur by similar mechanisms that exist in non-excitable tissues. However, while knockdown of TrpC1 did significantly reduce transient incidence and remaining transient amplitude in human neurons, TrpC1 expression did not account for all spontaneous Ca²⁺ influx. Therefore, further investigation is required to identify other channels that mediate Ca²⁺ transients in hESC-derived neurons.

Our results suggest that hNE cells and PMNs utilize disparate channels to control various developmental processes. This is highlighted by the temporal expression pattern and physiological role of TrpC1, which represents a critical discrepancy between our data and previous reports. A study by Fiorio Pla et al. (2005) demonstrated that bFGF-mediated increases in Ca²⁺ and cellular proliferation were significantly reduced in TrpC1 knockout animals. However, here we show that TrpC1 is not expressed in hESC-derived forebrain progenitor cells, but is upregulated upon neuronal differentiation. Further, TrpC1 knock down did not affect proliferation suggesting that TrpC1 does not regulate proliferation of our cultures. This discrepancy may reflect a difference in gestational age. Fiorio-pla et al. used neural stem cells derived from the E13 rat cortex, a stage at which progenitors from both the VZ and sub-ventricular zone (SVZ) are present. Our hNE cells may represent an earlier developmental time point, before the formation of the SVZ, when VZ progenitors predominate [54,55]. Interestingly, we found that blockade of Ca²⁺ transients in hNE cells with Gd³⁺/La³⁺ did significantly reduce proliferation that was independent of TrpC channels, suggestive of a role for other Trp channel family members controlling neural progenitor cell proliferation.

Lastly, our data suggest that TrpC1 and TrpC4 act as molecular targets for controlling neurite elongation in human neurons. A recent report demonstrated that TrpC4 was upregulated under conditions of axonal regeneration and that shRNA-mediated inhibition of TrpC4 in dorsal root ganglion neurons reduced neurite outgrowth [17]. Current studies also point to stretch-activated receptors as key players in neurite outgrowth as their inhibition led to the acceleration of neurite extension [2]. In one study, TrpC1 was recently shown to be activated directly by membrane stretch [56] and TrpC1 knock down in Xenopus spinal neurons eliminated growth cone turning in response to netrin-1 [37]. In human cells, only TrpC1 and TrpC4 subunits appeared to be expressed in differentiating neurons. Pharmacological or shRNA-mediated inhibition of either TrpC1 or C4, but not inhibition of L-Type Ca²⁺ channels, significantly inhibited neurite extension. Surprisingly, only inhibition of TrpC1 led to reductions in global Ca²⁺ transients. This suggests that TrpC1 may act independently of TrpC4 in the cell soma to partially regulate global spontaneous Ca²⁺ transients, perhaps through interactions with non-TrpC family members [57]. However, both TrpC1 and TrpC4 may be acting at the growth cone to regulate small local Ca²⁺ transients that control neurite outgrowth, possibly as a single heteromeric channel [58]. This idea is supported by the fact that only BAPTA, a Ca²⁺ chelator with fast kinetics and high affinity could reduce both calcium transients and neurite outgrowth, while EGTA (having slower kinetics and lower affinity) could reduce global calcium transients but not neurite outgrowth. It should be noted however, that other Ca²⁺ regulatory mechanisms such at glutamate receptors [59,60] and IP₃ receptors [61] may work in concert with TrpC channels in growth cones to regulate neurite extension as TrpC knock down did not fully inhibit outgrowth.

In addition to the basic insights of the Ca²⁺-dependent regulatory mechanisms that exist during human neural development, our data suggest new methods for engineering cells for therapeutic transplantation. Although significant advancements have been made [62], tumor formation and excess proliferation remain critical concerns for stem cell-based replacement therapies [63]. Here we found that Ca²⁺ transient inhibition by Trp channel antagonists resulted in a significant reduction of BrdU incorporation in progenitor cells with no observable effect on PMNs. Thus, inhibition of Trp channels may limit unwanted proliferation of transplantable cells. Secondly, the ability to guide neurites to their appropriate targets or modulate neurite outgrowth rates may help to accelerate synaptic integration and thus, the desired effects of neural transplants. Here we show that TrpC family members regulate neurite length in human neurons. Therefore, modulating the activity of specific members of the TrpC family may aid in the creation of safe and effective cell-replacement therapies for neurodegenerative disorders.