Selective linkage of mitochondrial enzymes to intracellular calcium stores differs between human induced pluripotent stem cells, neural stem cells and neurons

Abstract

Mitochondria and releasable endoplasmic reticulum (ER) calcium modulate neuronal calcium signaling, and both change in Alzheimer’s disease (AD). The releasable calcium stores in the ER are exaggerated in fibroblasts from AD patients and in multiple models of AD. The activity of the alpha-ketoglutarate dehydrogenase complex (KGDHC), a key mitochondrial enzyme complex, is diminished in brains from AD patients, and can be plausibly linked to plaques and tangles. Our previous studies in cell lines and mouse neurons demonstrate that reductions in KGDHC increase the ER releasable calcium stores. The goal of these studies was to test whether the relationship was true in human iPSC-derived neurons. Inhibition of KGDHC for one or 24 hours increased the ER releasable calcium store in human neurons by 69% and 144%, respectively. The effect was mitochondrial enzyme specific because inhibiting the pyruvate dehydrogenase complex, another key mitochondrial enzyme complex, diminished the ER releasable calcium stores. The link of KGDHC to ER releasable calcium stores was cell type specific as the interaction was not present in iPSC or neural stem cells. Thus, these studies in human neurons verify a link between KGDHC and releasable ER calcium stores, and support the use of human neurons to examine mechanisms and potential therapies for AD.

Graphical Abstract

Introduction

The physiological and morphological connections between mitochondria and the endoplasmic reticulum (ER) are critical in the regulation of neuronal calcium signaling and thus neuronal function (Csordás et al. 1999; Hajnóczky et al. 1994; Rizzuto et al. 1993; Rizzuto et al. 1998; Karagas & Venkatachalam 2019). The IP₃ receptors (IP₃R) on the ER appear to be in close proximity to the mitochondria (Seitaj et al. 2018). Considerable evidence suggests that these interactions may be important in Alzheimer’s disease (AD).

Altered calcium homeostasis in AD has been known for decades (Gibson & Peterson 1987). Exaggeration of IP₃ releasable calcium stores in response to bradykinin or bombesin (BRCS) with AD is a highly reproducible change that was first shown in fibroblasts from patients with AD (Ito et al. 1994). Multiple groups replicated the finding in multiple models including cell lines as well as neurons and fibroblasts from AD mouse models (Stutzmann et al. 2004; Stutzmann et al. 2006; Leissring et al. 2000). The observed changes in cytosolic calcium are a result of bradykinin activating IP3 receptors and not a consequence of off target bradykinin stimulation in the cell (Stutzmann 2005; Foskett et al. 2007). Some studies suggest the change in fibroblasts is diagnostic for AD (Hirashima et al. 1996). The mechanism and its implications for AD and therapies is an active area of investigation (Popugaeva et al. 2017). The mechanism is not established, although it is known that abnormal proteins resulting from presenilin mutations interact with the IP₃ receptor Ca²⁺ release channel and alter its stimulatory effect in response to IP₃ (Cheung et al. 2008; Cheung et al. 2010). These calcium changes are likely pathologically important, and modulating Ca²⁺ release is a reasonable therapeutic target (Schrank et al. 2019). However, the best model to use for these investigations is not clear. These experiments tested whether human neurons are an appropriate model.

Reduced metabolism always accompanies AD. Since first reported 40 years ago, hundreds of papers show that the decline in fluorodeoxyglucose -positron emission (FDGPET tomography) with AD correlates to the loss of cognitive function. The decline in FDG-PET is one of the biomarkers of AD (Jack et al. 2018). Our data is consistent with a decline in KGDH activity underlying the reduction in glucose metabolism (Gibson et al. 2010). We have shown that reducing alpha-ketoglutarate dehydrogenase complex (KGDHC) either genetically (Dumont et al. 2009) or with thiamine deficiency (Karuppagounder et al. 2009) stimulates plaque formation. Thus, the reduction in KGDHC models the mild impairment of metabolism that always accompany AD. In non-human tissues, inhibition of the KGDHC exaggerates BRCS in an AD-like manner (Gibson et al. 2012). Metabolism in humans and mice is very different, so these experiments examined the interactions in human neurons.

Thus, the current experiments tested the interactions of KGDHC with BRCS in human neurons. In addition, the specificity of the response was tested by comparing the consequences of diminished KGDHC and pyruvate dehydrogenase complex (PDHC) activities in neurons. The experiments also tested the specificity of the KGDHC/ER link at different stages of development (iPSC, NSC). The results reveal that the consequences differ between the two enzymes and between human iPSC, NSC and neurons. The results support the hypothesis that treating the mitochondrial deficit as defined by a reduction in KGDHC activity in neurons may ameliorate calcium abnormalities in AD.

Materials and Methods.

The study was not pre-registered, however a presubmission version was posted on bioRxiv under the following link: https://urldefense.proofpoint.com/v2/url?u=https-3A__www.biorxiv.org_content_10.1101_2020.06.20.162040v1.abstract&d=DwIFaQ&c=lb62iw4YL4RFalcE2hQUQealT9-RXrryqt9KZX2qu2s&r=F3eXLduzE_luUk3Kfzj4Oyo3SejC6s6lxAikICdnbqc&m=4O5Ja3ske3Eu3OO1YqLKQ2X4AhPRxkHRrGHuaPfcoqM&s=KWXaeegFGsfL6UUgQBTum-ezLviBTTXpCAKg7550AKQ&e=

Culture of iPSC, NSC and iPSC-derived neurons

The authors have no way to identify the human from whom these lines were derived. Thus, institutional approval for their use is not required. We selected iPSC that were generated from fibroblasts (AG6842) obtained from the Coriell Institute (Camden, NJ). The fibroblasts were from a 75-year-old male donor, who was from a family bearing A246E PSEN1 mutation, but he did not have the mutation and was APO ε3/ε3. Fibroblasts were reprogrammed using four high-titer retroviral constructs prepared by the Harvard Gene Therapy Core Facility that encoded human Oct4, KLF4, SOX2 and c-Myc, respectively (Sproul et al. 2014). The iPSC colonies were initially selected by morphology, passaged several times to remove transformed cells and expanded before characterization. After iPSC were expanded to multi-well format, they were characterized using a variety of quality control assays, including: karyotype and fingerprinting for cell-line genetics, alkaline-phosphatase (AP) enzymatic activity for reprogramming process, immunostaining for pluripotency markers, qPCR for endogenous pluripotent markers and viral transgene silencing, as well as APOE and PSEN1 genotyping (Sproul et al. 2014) .

Heterogeneity is always a concern with iPSC. The description of the sources of variability has been thoroughly reviewed (Volpato & Webber 2020). The field has not developed standard criteria to address this concern in terms of multiple clones from one donor, multiple conversions from one donor, multiple initiating tissues or how many donors would be required to account for genetic heterogeneity.

The iPSC line in question was generated without cloning to include all cells and avoid selection of non-typical cells. Our conditions produced very homogenous populations. All the cells differentiated to neurons responded to potassium depolarization (see results).

While the homogeneity of this iPSC line is imperfect, we have used these cells much like we and others have used SH-SY5Y, N2A or PC12 cells, which are also heterogeneous. This iPSC line would be available for others to use as well. Additionally, we used commercially available reagents so others could reproduce the finding.

The neural stem cells were never used beyond passage 8. iPSC were cultured on dishes and plates with pre-coating (one hour at 37 °C) of Geltrex matrix solution (1:100 dilution) and the cells were cultured with StemFlex Medium for one or two days (ThermoFisher Scientific, Grand Island NY). iPSC were harvested with StemPro Accutase (ThermoFisher Scientific) for one minute at 37 °C and then cultured with StemFlex medium containing ROCK Inhibitor, Stemolecule Thiazovivin (1 μM; Reprocell, Beltsville, MD) for the first 24 hours of culture. After 24 hours, the medium was replaced with StemFlex Medium without ROCK Inhibitor, and medium was then exchanged daily. iPSC were cryopreserved in Synth-a-Freeze cryopreservation medium (ThermoFisher Scientific) and experiments were only performed using iPSC from passage 30 to 35.

NSC were induced from iPSC with a seven-day incubation with PSC Neural Induction Medium (NIM, Neurobasal Medium and Neural Induction Supplement; ThermoFisher Scientific) on Geltrex pre-coated plates. Media was changed daily. After induction, the NSC were harvested with StemPro Accutase for 5 min at 37 °C and cultured with Neural Expansion Medium (NEM, Neurobasal Medium, Advanced DMEM/F-12 and Neural Induction Supplement; ThermoFisher Scientific) containing Rock inhibitor Y27632 (5 μM, Sigma) for the first 48 hours of culture. NSC were cryopreserved in Synth-a-Freeze cryopreservation medium and neuronal differentiation was only preformed using passage five NSC.

For neuronal differentiation, the dishes or plates were pre-coated with poly-L-ornithine (100 μg/mL) at 37 °C for two hours and laminin (3 μg/mL) at 37 °C overnight. iPSC-derived neurons were differentiated from NSC (passage 5) for 28 days with neuronal differentiation medium (Neurobasal Medium, 1 × B-27 supplement with antioxidant, 1 × GlutaMAX supplement, 2 × CultureOne supplement, 200 μM ascorbic acid, 1 × Anti-Anti, 20 ng/mL BDNF, 20 ng/mL GDNF; ThermoFisher Scientific) containing Rock Inhibitor Y27632 (5 μM) for the first 48 hours of differentiation at the seeding density of 2.5 × 10⁴ cell/cm². The media was exchanged partially by removing half of the medium and replacing it with Neuronal Differentiation Medium twice weekly. Cells were typically used for neuronal markers characterization and BRCS after four weeks in culture. Differentiation was done at least 15 times. This protocol from ThermoFisher Scientific produces cortical neurons without a preference for neuronal type.

Immunostaining of cell type specific markers

iPSC, NSC and iPSC-derived neurons were cultured on Delta T dishes (Bioptechs, Butler, PA) at different seeding densities and culture times (iPSC and NSC, 2 × 10⁵ cell/dish for two days; iPSC-derived neurons, 1 × 10⁵ cell/dish for four weeks). After culture, the cells were fixed with Image-iT Fixative Solution (4% formaldehyde; ThermoFisher Scientific) at room temperature for 15 min. After fixation, the cells were rinsed with DPBS (ThermoFisher Scientific) and blocked with blocking buffer (0.1% Triton X-100, 1% BSA in DPBS) at room temperature for one hour. After blocking, the cells were incubated with primary antibodies in blocking buffer at 4 °C overnight. The next day, the cells were rinsed with DPBS and incubated with secondary antibodies in blocking buffer at room temperature for one hour, following by staining with DAPI solution (ThermoFisher Scientific) at room temperature for 15 min. After DAPI staining, the dishes were mounted with ProLong Gold Antifade Mountant (ThermoFisher Scientific). The images were taken using a 20× objective under a Nikon 80i epifluorescence microscope. The table shows the details of primary and secondary antibodies.

Primary Antibody	Vendor	RRID	Cat #	Dilution/dose
OCT4	ThermoFisher Scientific	AB_2534182	A13998	1:400
SSEA4	ThermoFisher Scientific	AB_2533506	414000	10 μg/mL
Nestin	BD Biosciences	AB_396354	611658	1:1000
SOX2	ThermoFisher Scientific	AB_2533841	8-1400	3 μg/mL
SOX1	R&D systems	AB_2239897	AF3369	1:100
PAX6	ThermoFisher Scientific	AB_2533534	42-6600	1:200
MAP2	ThermoFisher Scientific	AB_2533001	13-1500	2 μg/mL
Synaptophysin	EMD Millipore	AB_94786	MAB329	1:500
Neurofilament	ThermoFisher Scientific	AB_10984147	MA5-14981	1:100
Secondary Antibody	Vendor	RRID	Cat #	Dilution/dose
Alexa Fluor 488 donkey anti-mouse	ThermoFisher Scientific	AB_141607	A21202	1:2000
Alexa Fluor 488 donkey anti-goat	ThermoFisher Scientific	AB_2534102	A11055	1:2000
Alexa Fluor 594 donkey anti-rabbit	ThermoFisher Scientific	AB_141637	A21207	1:2000
Alexa Fluor 594 donkey anti-goat	ThermoFisher Scientific	AB_2534105	A11058	1:2000

Histochemical assay of KGDHC and PDHC

To estimate relative KGDHC and PDHC activity in intact cells, histochemical assays of these two enzymes were performed based on our pervious study (Park et al. 2000). iPSC, NSC and iPSC-derived neurons were seeded on 24-well plates at a seeding density of 5 × 10⁴ cells/well for different culture times (iPSC and NSC for 2 days, and iPSC-derived neurons for 28 days) in 0.5 mL of growth medium (details in culture section). On the day of each experiment, the medium was aspirated, and the cells were washed once with balanced salt solution (BSS) [140 mM NaCl, 5 mM KCl, 2.5 mM CaCl₂, 1 mM MgCl₂, 5 mM glucose and 10 mM HEPES (pH 7.4)]. The cells were then treated with SP (trisodium succinyl phosphonate, 100, 200 and 500 μM) or dimethyl acetyl phosphonate (DMAP) (5 or 100 μM) in 0.5 mL of BSS for 60 min at 37 °C. At the end of the treatment, the buffer was aspirated and the well was washed with 200 μL of Hank’s Balanced Salt Solution (HBSS) containing 0.05% (v/v) Triton X-100. Wells were incubated with 200 μL of either complete assay mixture or assay mixture without substrates as a negative control. Samples were incubated with the assay mixture for 40 min. After incubation, the treatment medium was aspirated and the cells were washed with Ca²⁺- and Mg²⁺-free HBSS. The dark blue formazan product was solubilized with 10% (w/v) SDS in 0.01 N HCl overnight in a CO₂ incubator at 37 °C. The absorbance was read at 570 nm with a Spectra Max 250 model plate reader (Molecular Devices, Sunnyvale, CA).

Measurement of K⁺ depolarization

Cells were loaded with 2 μM fura-2 AM (ThermoFisher Scientific) in BSS for one hour at 37 °C and rinsed twice with BSS (Gibson et al. 2002). Then, 2 mL of BSS was added to each dish and [Ca²⁺] was monitored on the stage of an inverted Olympus IX70 microscope at 37 °C with a Delta Scan System (Photon Technology International, Edison, NJ). Excitation wavelengths were alternated between 350 and 378 nm (band pass 4 nm) and emission was monitored at 510 nm with a Hamamatsu C2400 SIT camera (Hamamatsu, Hamamatsu City, Japan) at 5-second intervals. Basal [Ca²⁺] was measured for one minute. 50 mM KCl (Sigma) was added to depolarize voltage gated calcium channels to allow entry of extracellular calcium, which increases cytosolic calcium. The signal was measured for another 5 minutes. Each value was the average of 32 images taken within 5 seconds. Standard images of fura-2 solutions with minimum and maximum [Ca²⁺] were taken at the end of each day’s experiment to calculate the intracellular calcium concentrations. No blinding was performed throughout the experiments. (Huang et al. 1991).

The calcium max (Rmax) is 1M Ca2+ and calcium min (Rmin) is 0.1 mM Ca²⁺. The concentration of chelator, EGTA is 1 mM. The calcium concentrations have been calculated by our acquisition software, ImageMaster (PTI) with the following equation: R: 350/378

$$[\text{Ca}2+]=\text{Kd}\mathrm{x}(\mathrm{R}\text{-}\mathrm{R}\text{min})∕(\mathrm{R}\max \text{-}\mathrm{R})$$

Measurement of BRCS

Cells were loaded with 2 μM fura-2-AM in BSS for one hour at 37 °C and rinsed twice with Ca²⁺-free BSS (140 mM NaCl, 5 mM KCl, 1.5 mM MgCl₂, 5 mM glucose, 10 mM HEPES, 0.1 mM CaCl₂,1mM EGTA, pH 7.4) (Gibson et al., 2002). Then, 2 mL of Ca²⁺-free BSS was added to each dish and [Ca²⁺] was monitored with the Delta Scan System. Basal [Ca²⁺] was measured for one minute. Bradykinin (100, 200 and 500 nM; Sigma) was added to release calcium from internal calcium stores and the signal was measured for another 5 minutes. Each value was the average of 32 images taken within 5 seconds. Standard images of fura-2 solutions with minimum and maximum [Ca²⁺] were taken at the end of each day’s experiment to calculate the intracellular calcium concentrations. The values of area [Ca²⁺]_i are an integration of [Ca²⁺]_i within the certain response time after adding bradykinin. No blinding was performed throughout the experiments. (Huang et al. 2010).

Statistics

Since these are not animal or human experiments, randomization was not used. In all experiments, controls and experimentals were intentionally alternated to avoid bias. No blinding procedure was performed. Sample size was determined by knowing the variance with each of these methods, but no formal calculation of sample size was made. Variance in these cell experiments was very small so box plots were not used.

Microscoft EXCEL (RRID: SCR_016137) was used to prepare graphics and IBM SPSS Statistics 25 (RRID:SCR_002865) was used to test the statistical significance for one-way ANOVA with Student Newman Keul’s test. The number of samples for each experiment, the number of independent experiments, and the precise statistical treatment are in the legends for each figure. No test for outliers, assessment of the normality, or sample calculation were performed for all analysis.

Results

Generation and characterization of iPSC, NSC and iPSC-derived neurons

All of the cell types were confirmed with their cell-specific markers (iPSC: OCT4 and SSEA4; NSC: Nestin, SOX2, SOX1 and PAX6, iPSC derived neurons: MAP2, neurofilament and synaptophysin) (Figure 1). The identity of the iPSC-derived neurons was confirmed by depolarization of voltage gated calcium channels by 50 mM KCl. K⁺ depolarization increased intracellular calcium response in iPSC-derived neurons, but not in either iPSC or NSC (Figure 2). Thus, both the antibody marker studies and the depolarization indicated that our robust neuronal differentiation protocol yielded highly homogenous neurons.

Figure 1.

Immunocytochemical characterization of iPSC, NSC and iPSC derived neurons.

NSC were differentiated from iPSC (passage 30-35) after 7 days in culture. NSC were cultured for five passages before initiating neuronal differentiation. Cells were differentiated for four weeks. Cell specific markers were confirmed in cells at the three stages of differentiation. All makers followed the appropriate time course.

Cells were cultured on Delta T dishes at different seeding densities and culture times (iPSC, 2 × 10⁵ cell/dish for two days; NSC, 2 × 10⁵ cell/dish for two days; neurons, 1 × 10⁵ cell/dish for four weeks). Markers were tested by immunocytochemistry (iPSC: OCT4 and SSEA4; NSC: Nestin, SOX2, SOX1 and Pax6, iPSC derived neurons: MAP2, neurofilament and synaptophysin). No primary antibody controls were used. Rather, we carefully documented the presence of the markers in different cell types (e.g. confirm absence of NSC markers in neurons).

Figure 2.

Characterization of voltage gated calcium channels in iPSC-derived cells

Functional characterization of the neurons was done by assessing the responses of [Ca²⁺]_i to depolarizing concentrations of K⁺ (KCl, 50 mM) in iPSC, NSC and iPSC derived neurons. Cells were loaded with Fura-2-AM (2 μM) in BSS for 60 min. After loading, the cells were rinsed with BSS and [Ca²⁺]_i was measured for one min of basal response and for another 5 min after adding KCl (50 mM): (A) The temporal response of [Ca²⁺]_i after adding KCl at 1 min. (B) the percentage of cells responding to KCl in the three types of cells. Values are means ± SEM (n=106, 92 and 72 cells in iPSC, NSC, and iPSC-derived neurons, respectively) from three independent cell culture preparations and depolarization experiments. The error bars were too small to show. *** indicates values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

BRCS in iPSC, NSC and iPSC-derived neurons

The dose response to bradykinin determined whether BRCS varied with differentiation. Cells at each stage of differentiation had a dose dependent response to bradykinin. The magnitude of the response in iPSC varied with the days of culture. The responses of calcium may be altered by cell-cell interactions, including cell distribution, cell number, as well as the extracellular area. iPSC cultured for 36 hours form embryoid bodies (EB) formation and this formation may reduce the response of internal calcium. The patterns of change in iPSC (Figure 3) and neurons (Figure 5) were very similar. The [Ca²⁺]_i peaked and then went back to the basal [Ca²⁺]_i within two minutes after bradykinin addition. However, the NSC responded differently than the other cells. The [Ca²⁺]_i did not return to basal [Ca²⁺]_i even after 5 min of bradykinin addition (Figure 4).

Figure 3.

BRCS in iPSC

iPSC were cultured on Delta T dishes for one (A) or two (B) days. The cells were loaded with Fura-2 AM (2 μM) in BSS for 60 min for dye loading. The media were changed to calcium free BSS and the calcium measurements were initiated. Bradykinin (100, 200 or 500 nM) was added after 1 min of basal [Ca²⁺]_i measurement. The left panel shows the tracings and the right panel shows the integration of the [Ca²⁺]_i peak over the 5 min interval after bradykinin addition. Values are means ± SEM (n=355 and 250 cells in one and two days of culture, respectively) from two independent cell culture preparations and calcium experiments. Different letters indicate values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Figure 5.

BRCS in iPSC-derived neurons

NSC (passage 5) were seeded on Delta T dishes. After four weeks of differentiation, the cells were loaded with Fura-2 AM (2 μM) in BSS for 60 min. After loading, the media were changed to calcium free BSS, and the calcium measurements were initiated. Bradykinin (100, 200 or 500 nM) was added after 1 min of basal [Ca²⁺]_i measurement. Panel A shows the tracings and panel B shows the integration of the [Ca²⁺]_i peak over the 5 min interval after bradykinin addition. Values are means ± SEM (n=435 cells) from three independent cell culture preparations and calcium experiments. Different letters indicate values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Figure 4.

BRCS in NSC

NSC (passage 5) were cultured on Delta T dishes for two days. After two days of culture, the cells were loaded with Fura-2-AM (2 μM) in BSS for 60 min (Gibson et al. 2002). After loading, the media were changed to calcium free BSS. The calcium measurements were initiated and bradykinin (100, 200 or 500 nM) was added after 1 min of basal [Ca²⁺]_i measurement. Panel A shows the tracings and panel B shows the integration of the [Ca²⁺]_i peak over the 5 min interval after bradykinin addition. Values are means ± SEM (n=490 cells) from two independent cell culture preparations and calcium experiments. Different letters indicate values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

The consequences of reducing KGDHC on BRCS in iPSC, NSC and iPSC-derived neurons

Cells were treated with the specific KGDHC inhibitor, trisodium succinyl phosphonate (SP, 500 μM), before adding bradykinin. First, the inhibitory effect of SP on in situ KGDHC was confirmed with a histochemistry assay. In iPSC, 500 μM SP inhibited KGDHC activity by 67% after a one-hour treatment and by 75% after 24 hours of treatment (Figure 6). In NSC, SP inhibited KGDHC activity by 89% after one hour of treatment and by 94% after 24 hours of treatment (Figure 7). In iPSC-derived neurons, 500 μM SP inhibited KGDHC activity by 87% after one hour of treatment and by 83% after 24 hours of treatment (Figure 8). Reducing KGDHC activity with SP for one hour and 24 hours did not alter BRCS in iPSC (Figure 6) nor NSC (Figure 7). However, reducing KGDHC activity in iPSC-derived neurons exaggerated BRCS by 69% and 144% after a one-hour or 24-hour treatment, respectively (Figure 8).

Figure 6.

Inhibiting KGDHC did not alter BRCS in iPSC.

iPSC were cultured in 24-well plates at a seeding density of 1 × 10⁵ cells/well for two days. After two days of culture, cellular KGDHC activity was measured by histochemistry (Panel A). iPSC were cultured on Delta T dishes for one or two days. After culture, the cells were pre-treated with trisodium succinyl phosphonate (SP, 500 μM) for one hour (in BSS) or 24 hours (in completed media). After 1 hour of Fura-2-AM loading, the media was changed to calcium free BSS, the calcium measurements were initiated and bradykinin (500 nM) was added after 1 min of basal [Ca²⁺]_i measurement. (Panel B). Values in Panel A are means ± SEM (n= 6 wells for each groups) from two independent cell culture preparations and histochemistry experiment. Values in Panel B and C are means ± SEM (n=261 and 272 cells for one hour and 24 hours of treatment, respectively) from two independent cell culture preparations and calcium experiments. The letter indicates values not statistically significant (p>0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Figure 7.

Reducing KGDHC did not alter BRCS in NSC

NSC were cultured on 24-well plates at the seeding density of 1 × 10⁵ cell/well for two days. After two days of culture, KGDHC activity was measured by histochemistry (Panel A). NSC were cultured on Delta T dishes for two days. After culture, the cells were pre-treated with trisodium succinyl phosphonate (SP, 500 μM) for one hour (in BSS) or for 24 hours (in completed media). After 1 hour of Fura-2-AM loading, the media were changed to calcium free BSS, and the calcium measurements were initiated and bradykinin (500 nM) was added after 1 min of basal [Ca²⁺]_i measurement (Panel B). Values in Panel A are means ± SEM normalized with cell number in each well (n= 18683, 21359, 10837, and 18672 NSC cells in control for one hour, SP for one hour, control for 24 hours, SP for 24 hours, respectively) from two independent cell culture preparations and histochemistry experiments. Values in Panel B and C are means ± SEM (n=208 and 205 cells for one hour and 24 hours of treatment, respectively) from two independent cell culture and calcium experiments. The letter indicates values not statistically significant (p>0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Figure 8.

Reducing KGDHC exaggerated BRCS in neurons

iPSC-derived neurons were cultured on 24-well plates at the seeding density of 5 × 10⁴ cell/well for four weeks. After four weeks of culture, KGDHC activity was measured by histochemistry (Panel A). Neurons were cultured on Delta T dishes for four weeks. After culture, the cells were pre-treated with SP (500 μM) for one hour (in BSS) or 24 hours (in completed media). After 1 hour of Fura-2-AM loading, the media was changed to calcium free BSS, and the calcium measurements were initiated and bradykinin (500 nM) was added after 1 min of basal [Ca²⁺]_i measurement (Panel B). Values in Panel A are means ± SEM normalized with cell number in each well (n= 908, 941, 964, and 999 neurons in control for one hour, SP for one hour, control for 24 hours, SP for 24 hours, respectively) from two independent cell culture preparations and histochemistry experiments. Values in Panel B and C are means ± SEM (n=627 and 603 cells for one hour and 24 hours of treatment, respectively) from two independent cell culture and calcium experiments. Different letters indicate values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Consequences of reducing PDHC on BRCS in iPSC-derived neurons

iPSC-Derived neurons were treated with the specific PDHC inhibitor, dimethyl acetyl phosphonate (DMAP, 5 or 100 μM) for one hour (Park et al. 2000). First, we confirmed the inhibitory effect of DMAP with histochemistry. In these neurons, 5 μM or 100 μM DMAP inhibited in situ PDHC activity by 75% and 76%, respectively (Figure 9A). 5 μM or 100 μM DMAP inhibited BRCS by 44.8% and 46.4%, respectively (Figure 9).

Figure 9.

Reducing PDHC diminished BRCS in iPSC derived neurons

iPSC-derived neurons were cultured on 24-well plates at the seeding density of 5 × 10⁴ cells/well for four weeks. After four weeks of culture, the activity of PDHC was measured by histochemistry (Panel A). Neurons were cultured on Delta T dishes for four weeks. After culture, the cells were pre-treated with DMAP (5, 100 μM) for one hour (in BSS). After 1 hour of Fura-2-AM loading, the media was changed to calcium free BSS, the calcium measurements were initiated, and bradykinin (500 nM) was added after 1 min of basal [Ca²⁺]_i measurement.(Panel B). Values in Panel A are means ± SEM normalized with cell number in each well (n= 1496, 1657, and 1672 neurons in control, 5 μM DMAP, and 100 μM DMAP, respectively) from two independent cell culture preparations and histochemistry experiments. Values in Panel B and C are means ± SEM (n= 840 cells) from two independent cell culture and calcium experiments. Different letters indicate values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Consequences of reducing PDHC or KGDHC on resting cytosolic free calcium

The BRCS change in iPSC-derived neurons did not appear to be related to altered resting cytosolic free calcium, since changes in cytosolic free calcium were minimal when reducing KGDHC or PDHC in the absence IP₃ agonists (Figure 10).

Figure 10.

Effects of reducing KGDHC or PDHC on basal calcium in iPSC, NSC, and iPSC derived neurons

Basal [Ca²⁺]_i in iPSC (A), NSC (B), and iPSC derived neurons (C)(D) were calculated from the same experiments in Figure 6-9. Values are means ± SEM. Different letters indicate values vary significantly (p<0.05) from the other groups by ANOVA followed by Student Newman Keul’s test.

Discussion

The exaggerated release of calcium from ER by activation of either ryanodine receptors (Stutzmann et al. 2004; Stutzmann et al. 2006) (Flucher et al. 1993) or IP₃ receptors (Ito et al. 1994) (Chakroborty & Stutzmann 2014; Cheung et al. 2008; Del Prete et al. 2014) (Popugaeva & Bezprozvanny 2013) (Shilling et al. 2014) is among the most reproducible abnormalities that support a critical role of calcium in AD (Peterson C 1985). FAD-causing mutant PSEN1 (M146L) and PSEN2 (N141I) interact with the IP₃R Ca²⁺ release channel and exert profound stimulatory effects on its gating activity in response to saturating and sub-optimal levels of IP₃ (Cheung et al. 2008). In addition, the membrane hyperpolarization associated with ER-calcium release is enhanced in mutant PSEN1 mice and reduces spiking activity and responsivity to synaptic inputs (Stutzmann et al. 2006; Oddo et al. 2003; Stutzmann 2007). The calcium dysregulations seen in AD models are not merely accelerated or amplified signaling changes, inevitable in old age, but rather, are novel and pathogenic changes to fundamental calcium signaling patterns (Stutzmann 2007). An increased number of ryanodine receptors appears important in mice with mutated presenilin; increased ROS expression in neurons contributes to increased calcium release, synaptic decline and increased RyR2 expression seen in the hippocampus and cortex. This increased IP₃ receptor-evoked calcium release may involve the increase of stromal interaction molecule 1 (STIM1) expression and the decrease of STIM2 expression in the soma. These STIMs translocate to plasma membrane–ER contact sites, where they bind to Orai1 channels and open a Ca²⁺ influx pathway to slowly replenish ER Ca²⁺ stores using extracellular Ca²⁺ (Mustaly-Kalimi et al. 2018; Schrank et al. 2019).

The consequences of the increased cytosolic calcium following release from ER calcium stores on the mitochondria is not clear. The increased calcium leads to premature and severe defects in synaptic plasticity, behavior and cognitive function (Lacampagne et al. 2017). Mitochondria serve as an acute calcium buffer, so the exaggerated BRCS likely increases mitochondrial calcium concentration. In heart, increased intracellular Ca²⁺ concentration results in excess uptake of Ca²⁺ by the mitochondria and increased respiratory rate by the TCA cycle. However, this also leads to mitochondrial calcium overload as well as metabolic dysfunction (Schrank et al. 2019). Prolonged increases of Ca²⁺ release open the mitochondrial permeability transition pore (PTP) to release apoptotic signaling molecules (e.g., cytochrome C) (Schrank et al. 2019; Mustaly-Kalimi et al. 2018).

Reductions in KGDHC activity similar to those in AD may underlie the calcium changes described above or may exaggerate the calcium abnormality. These studies demonstrate that, in human neurons, AD related reductions in KGDHC can cause AD like changes in calcium. Our previous studies demonstrated this pattern in mouse neurons and cell lines (Huang et al. 2003; Gibson et al. 2012; Huang et al. 2014). Many mechanisms are plausible, and they may interact with one another. Reducing KGDHC alters the mitochondrial membrane potential and calcium buffering, but the same concentration of succinyl-phosphonate that was used in these experiments did not alter the mitochondrial membrane potential in our previous experiment (Banerjee et al. 2016). Diminishing KGDHC increased BRCS under conditions that increased the mitochondrial NAD⁺/NADH ratio and elevated cytochrome c release (Gibson et al. 2012; Chen et al. 2017). Reductions in KGDHC diminish GTP (Kiss et al. 2013), and GTP is known to modulate ER-mitochondrial interactions (Hajnóczky et al. 1994). Diminished KGDHC activity causes release of mitochondrial proteins (Banerjee et al. 2016). KGDHC can control oxidant production and select oxidants can cause AD-like changes in BRCS (Huang et al. 2010). KGDHC controls the post-translational modification succinylation (Yang et al. 2019), and post-translational modifications of the ryanodine receptor by phosphorylation, oxidation and nitrosylation are present in brains of AD patients, and in two murine models of AD (Lacampagne et al. 2017). Oxidant induced modification of protein thiol groups may underlie the AD-related exaggeration of the BRCS. Oxidants can alter BRCS by activating the ryanodine receptor by trans-nitrosation of thiol containing proteins (Huang et al. 2005; Arnelle & Stamler 1995).

The coupling of the mitochondria to releasable ER calcium in neurons was enzyme selective. Inhibiting PDHC diminished BRCS, in contrast to the exaggerated BRCS following KGDHC inhibition. No other studies have examined the consequences of inhibiting PDHC on BRCS. Inhibiting PDHC blocks acetyl-coA entry into the tricarboxylic acid cycle, which depletes the substrates of the tricarboxylic acid cycle. On the other hand, blocking KGDHC may lead to accumulation of substrates, such as alpha-ketoglutarate, that leave the mitochondria and serve as signaling molecules throughout the cell.

These are the first studies to compare the patterns of BRCS in iPSC, NSC and neurons, as well as their response to inhibition of KGDHC. Previous studies only showed the cytosolic calcium changes in iPSC-derived neurons for the study of AD (Prè et al. 2014; Duan et al. 2014), motor or dopaminergic neurons (Hartfield et al. 2014; Dafinca et al. 2016) or other neurodegenerative diseases (Hartfield et al. 2014; Schöndorf et al. 2014; Rabenstein et al. 2017; Prè et al. 2014). Although cells at all stages of development had IP₃ releasable stores, the patterns were markedly different in NSC compared to iPSC and iPSC-derived neurons. There are no other data on these stores in NSC, so any discussion of mechanism would be purely speculative.

The experiments in this manuscript demonstrate that the mitochondrial link to BRCS varies between various differentiation stages. Reducing KGDHC in human iPSC-derived neurons exaggerated BRCS, but had no impact in NSC or iPSC. This difference may relate to the fact that mitochondria have a very different role in iPSC, NSC and neurons. iPSC and NCS primarily use glycolysis for energy, whereas neurons depend on mitochondria. BRCS are sensitive to select free radicals, and free radical production is much higher when cells use mitochondria for energy (Khacho et al. 2019; Intlekofer & Finley 2019). Thus, differences in ROS may underlie the differences in the sensitivity to inhibition of PDHC or KGDHC.

These results demonstrate in human neurons that reducing KGDHC enhances release of calcium from the ER. The effect is specific to KGDHC since inhibiting PDHC inhibits the release. Furthermore, the coupling of KGDHC to the ER varies with the differentiation state. The results provide evidence in iPSC-derived human neurons that diminishing the activity of KGDHC causes AD-like changes in release of calcium from the endoplasmic reticulum.

Abbreviations used:

AP

alkaline-phosphatase

APOE4

epsilon 4 allele of Apolipoprotein E gene

AD

Alzheimer’s disease

t-BHP

tert-butyl-hydroperoxide

BRCS

bombesin or bradykinin releasable calcium stores

BSS

balanced salt solution

CESP

carboxyethyl succinyl phosphonate

DMAP

dimethyl acetyl phosphonate

ER

endoplasmic reticulum

FDGPET

fluorodeoxyglucose (FDG)-positron emission tomography (PET)

HBSS

Hank’s Balanced Salt Solution

IP₃

inositol 1,4,5-trisphosphate

IP₃R

inositol 1,4,5-trisphosphate receptors

iPSC

induced pluripotent stem cells

KGDHC

alpha-ketoglutarate dehydrogenase complex

NEM

neural expansion media

NIM

neural induction media

NSC

neural stem cells

PDHC

pyruvate dehydrogenase complex

PSEN1

presenilin 1

PSEN-2

presenilin-2

PTP

permeability transition pore

RRID

research resource identifier (see scicrunch.org)

RyR

ryanodine receptors

SNAP

S-nitroso-N-acetylpenicillamine

SP

trisodium succinyl phosphonate

STIM

stromal interaction molecule

TCA

tricarboxylic acid cycle