Decreased Sirtuin Deacetylase Activity in LRRK2 G2019S iPSC-Derived Dopaminergic Neurons.

Mitochondrial changes have long been implicated in the pathogenesis of Parkinson's disease (PD). The glycine to serine mutation (G2019S) in leucine-rich repeat kinase 2 (LRRK2) is the most common genetic cause for PD and has been shown to impair mitochondrial function and morphology in multiple model systems. We analyzed mitochondrial function in LRRK2 G2019S induced pluripotent stem cell (iPSC)-derived neurons to determine whether the G2019S mutation elicits similar mitochondrial deficits among central and peripheral nervous system neuron subtypes. LRRK2 G2019S iPSC-derived dopaminergic neuron cultures displayed unique abnormalities in mitochondrial distribution and trafficking, which corresponded to reduced sirtuin deacetylase activity and nicotinamide adenine dinucleotide levels despite increased sirtuin levels. These data indicate that mitochondrial deficits in the context of LRRK2 G2019S are not a global phenomenon and point to distinct sirtuin and bioenergetic deficiencies intrinsic to dopaminergic neurons, which may underlie dopaminergic neuron loss in PD.

Introduction

Mutations in leucine-rich repeat kinase 2 (LRRK2) are associated with both familial and sporadic Parkinson's disease (PD), and exhibit clinical symptoms and pathology typical of sporadic PD (Gatto et al., 2013). The G2019S mutation is the most common and confers hyper-kinase activity, but it is unclear how the gain of function in kinase activity results in PD pathogenesis. The precise biological function of LRRK2 remains largely unknown, but increasing evidence suggests that mutations in LRRK2 contribute to mitochondrial dysfunction and oxidative stress (Wang et al., 2008, Ng et al., 2009, Saha et al., 2009, Mortiboys et al., 2010).

In vivo and in vitro studies examining the consequence of LRRK2 G2019S on mitochondrial health show altered mitochondrial morphology, increased fragmentation, elevated reactive oxygen species (ROS) production, and decreased respiration (Ryan et al., 2015, Yue et al., 2015). Induced pluripotent stem cell (iPSC) derived neurons generated from homozygous and heterozygous LRRK2 G2019S patients showed mitochondrial damage (Sanders et al., 2014), deficits in basal and maximal respiration, an increase in mitochondrial trafficking (Cooper et al., 2012), and increased susceptibility to H2O2, 6-OHDA, and rotenone (Nguyen et al., 2011, Reinhardt et al., 2013). Factors implicated in dopaminergic vulnerability include higher metabolic activity, increased oxidative stress due to dopamine oxidation, and calcium buffering defects, all of which may converge on mitochondrial malfunction.

We used LRRK2 G2019S iPSC-derived dopaminergic, glutamatergic, and sensory neurons to explore functionally relevant mitochondrial parameters by which neurons may be susceptible to disease. We focused on addressing two key questions. First, we asked if the LRRK2 G2019S mutation causes consistent mitochondrial changes across multiple neuronal subtypes. We found that mitochondrial respiration deficits were observed in LRRK2 G2019S iPSC-derived dopaminergic and glutamatergic neuron cultures, whereas no mitochondrial defects were observed in LRRK2 G2019S iPSC-derived peripheral sensory neuron cultures, suggesting a shared CNS weakness. However, LRRK2 G2019S iPSC-derived dopaminergic neurons displayed additional mitochondrial distribution and trafficking abnormalities, indicating a unique midbrain dopaminergic phenotype. Second, we asked what cellular mechanisms may contribute to the unique mitochondrial phenotypes observed in dopaminergic neuron cultures. Accumulating evidence suggests that protein acetylation is a key regulatory mechanism in mitochondrial function. Sirtuins are protein lysine deacetylases that are localized to the nucleus, cytosol, and mitochondria, where they serve a variety of anti-aging and metabolic roles (Chalkiadaki and Guarente, 2012, Guarente, 2013, Herskovits and Guarente, 2013, Buler et al., 2016). Humans encode seven distinct sirtuins that harbor various roles in mitochondrial biogenesis, movement, and energetics, and collectively may play a role in the defects observed in PD dopaminergic neurons. Surprisingly, we found that LRRK2 G2019S iPSC-derived dopaminergic neurons exhibit increased sirtuin levels, but decreased deacetylase activity compared with control cells. Because sirtuins require nicotinamide adenine dinucleotide (NAD+) as a co-substrate to catalyze deacetylation, we assessed NAD+ levels and observed a decreased NAD+ pool in LRRK2 G2019S dopaminergic neurons compared with healthy dopaminergic neurons. The decrease in NAD+ correlated with elevated acetylation of sirtuin substrates p53, α-tubulin, and SOD2. Together, these data suggest that LRRK2 G2019S confers cell-type specific bioenergetic defects, and that dopaminergic neurons may be more significantly impacted in PD due to low endogenous NAD+ levels and reduced sirtuin deacetylase activity.

Results

LRRK2 G2019S iPSC-Derived Dopaminergic Neurons Display Altered Mitochondrial Content and Distribution

To examine mitochondrial health across neuronal subtypes expressing the LRRK2 G2019S mutation, we utilized human iPSCs derived from three independent LRRK2 G2019S patients and three unaffected control individuals (Schwab and Ebert, 2015) and differentiated the cells toward midbrain dopaminergic neurons, forebrain glutamatergic neurons, and peripheral sensory neurons. Importantly, we have previously shown that neurons from these LRRK2 G2019S iPSC lines have consistently elevated expression of phosphorylated LRRK2 at serine 935, but similar levels of total LRRK2 protein compared with the control iPSC neurons (Schwab and Ebert, 2015).

We first examined overall mitochondrial content and distribution across the three neuronal subtypes. Cells were immunostained for the mitochondrial membrane protein TOM20 and co-labeled with tyrosine hydroxylase (TH), βIII-tubulin (TUJ1), or peripherin to mark dopaminergic, glutamatergic, and sensory neurons, respectively (Figures 1A, 1D, 1G, and S1A). Immunofluorescence (Figure S1B) and western blot analysis (Figure S1C) for the NMDA receptor 2B subunit (NR2B) were used to further confirm glutamatergic neuron identity. Consistent with our previous report (Schwab and Ebert, 2015), there was no difference in differentiation efficiency across the individual cell lines (Figures 1B, 1E, and 1H), and differentiation efficiencies were consistent with other reports in the literature (Chambers et al., 2009, Chambers et al., 2012, Kriks et al., 2011). Using immunofluorescence intensity (Figures 1C, 1F, and 1I) and western blot analysis (Figure 1J), mitochondrial content was reduced in LRRK2 G2019S iPSC-derived dopaminergic neurons compared with controls, but not in the other neuronal subtypes. We next measured mitochondrial distribution along the neurite and divided the data points into four equal quartiles to assess differences in proximal versus distal distribution. Because we and others have shown that LRRK2 G2019S iPSC-derived dopaminergic neurons exhibit shortened neurites (Cooper et al., 2012, Sanchez-Danes et al., 2012, Reinhardt et al., 2013, Schwab and Ebert, 2015), quartiles were determined relative to each neuronal subtype and not compared across neuron types. LRRK2 G2019S iPSC-derived dopaminergic neurons displayed significantly reduced mitochondrial distribution along the distal neurite length (Figure 1K). In contrast to previous reports using mouse cortical neurons expressing LRRK2 G2019S (Cherra et al., 2013), neither mitochondrial content nor distribution was diminished in LRRK2 G2019S glutamatergic and sensory neurons compared with control neurons (Figures 1F, 1I, 1L, and 1M), suggesting neuron specific mitochondrial properties in human LRRK2 G2019S conditions.

Figure 1

LRRK2 G2019S iPSC-Derived Dopaminergic Neurons Display Altered Mitochondrial Content and Distribution

Representative images of each individual control line (1, 2, and 3) and each individual LRRK2 G2019S iPSC line (het, 1, and 2) immunostained for (A) tyrosine hydroxylase (TH, red), (D) βIII-tubulin (TUJ1, green), and (G) peripherin (red) to mark dopaminergic, glutamatergic, and sensory neurons, respectively. Nuclei labeled with Hoechst (blue). Scale bars, 50 μm. Quantification of differentiation efficiency (B, E, and H), mitochondrial content (C, F, and I), TOM20 protein (J), and mitochondrial distribution (K, L, and M) compared by one-way repeated measures ANOVA with Tukey's post-hoc test. n.s., not significant. ∗∗p < 0.01 in (C) indicates that all three LRRK2 samples are significantly reduced compared with all three control samples. ∗p < 0.05 in (I) indicates that control 1 is significantly reduced compared with LRRK2 1 and 2. ∗p < 0.05 in (J) indicates that all three LRRK2 samples are significantly reduced compared with all three control samples. ∗p < 0.05 in (K) indicates that LRRK2 het is different from all three controls (0%–25%) and all three LRRK2 lines are different from all three control lines for 25%–50%, 50%–75%, and 75%–100%. ∗p < 0.05 in (L) indicates that LRRK2 1 is significantly different from all samples at 0%–25%. ∗p < 0.05 in (M) indicates that control 2 and 3 are different from LRRK2 het and LRRK2 1 at 0%–25% and 25%–50%, and that LRRK2 1 is different from all three controls at 50%–75% and 75%–100%. n = 4 independent experiments. All error bars are SEM. See also Figure S1.

LRRK2 G2019S iPSC-Derived Dopaminergic Neurons Display Increased Mitochondrial Velocity and Motility

Neurons depend on regulated mitochondrial trafficking to match energy demand and to control the clearance of damaged mitochondria (Ashrafi and Schwarz, 2013). We utilized live-cell imaging to investigate if LRRK2 G2019S induced changes in mitochondrial trafficking and mobility within iPSC-derived dopaminergic, glutamatergic, and sensory neurons. We observed an increase in mitochondrial velocity in LRRK2 G2019S iPSC-derived dopaminergic neurons compared with control dopaminergic neurons (Figure 2A). Interestingly, neither LRRK2 G2019S glutamatergic nor sensory neurons were altered compared with their respective control neurons (Figure 2A). Similar to a previous study (Cooper et al., 2012), we also saw an increase in the percentage of mobile mitochondria in dopaminergic neurons, but this effect was not observed in LRRK2 G2019S glutamatergic or sensory neurons (Figure 2B). Finally, we found that LRRK2 G2019S iPSC-derived dopaminergic neurons display increased retrograde mitochondrial velocity (Figure 2C), which correlates with diminished mitochondrial content in the distal neurite (Figure 1K). Together, these data highlight intrinsic dopaminergic neuron mitochondrial trafficking deficits.

Figure 2

LRRK2 G2019S iPSC-Derived Dopaminergic Neurons Display Increased Mitochondrial Velocity and Motility

Measurements of (A) mitochondrial velocity (μm/s), (B) percent mobile mitochondria, and (C) direction of mitochondrial movement are shown for each individual iPSC line. There was a significant difference for each of the three LRRK2 lines compared with each of the control lines in the dopaminergic neuron cultures, but not in either the glutamatergic or sensory neuron cultures. n.s., not significant. ∗p < 0.05 and ∗∗p < 0.01 by one-way repeated measures ANOVA with Tukey's post-hoc test. n = 4 independent experiments. All error bars are SEM. See also Figure S4.

Mitochondrial Respiration Is Decreased in LRRK2 G2019S iPSC-Derived Dopaminergic and Glutamatergic Neurons

We next asked if mitochondrial bioenergetics was affected in LRRK2 G2019S iPSC-derived neurons. Control and LRRK2 G2019S iPSCs were differentiated into dopaminergic, glutamatergic, and sensory neurons to measure oxygen consumption rate (OCR) (Dranka et al., 2011, Zhang et al., 2012, Patitucci and Ebert, 2016). Basal glycolytic rates were not changed in any of the LRRK2 G2019S neurons compared with controls (data not shown). However, consistent with previous studies (Mortiboys et al., 2010, Mortiboys et al., 2015, Cooper et al., 2012, Papkovskaia et al., 2012), we observed a marked decrease in OCR for ATP-linked (Figure 3A), maximal (Figure 3B), and spare respiration (Figure 3C) in LRRK2 G2019S dopaminergic neuron cultures compared with control dopaminergic neuron cultures. Interestingly, we also observed a decrease in these same parameters for LRRK2 G2019S forebrain glutamatergic neuron cultures compared with control, but not in LRRK2 G2019S sensory neuron cultures (Figures 3A–3C). Together, these data suggest that LRRK2 G2019S may preferentially alter mitochondrial respiration rates of central neurons and spare peripheral neurons.

Figure 3

Mitochondrial Respiration Is Decreased in LRRK2 G2019S iPSC-Derived Dopaminergic and Glutamatergic Cultures

All three LRRK2 G2019S iPSC-derived dopaminergic and glutamatergic cultures display diminished (A) ATP-linked, (B) maximal, and (C) spare respiration compared with all three of the respective control cultures. LRRK2 G2019S iPSC-derived sensory neurons are unchanged compared with controls. ∗p < 0.05 by one-way repeated measures ANOVA with Tukey's post-hoc test. n = 4 independent experiments. All error bars are SEM. See also Figures S2, S3, and S4.

Consistent with the observed decrease in bioenergetics, LRRK2 G2019S dopaminergic cultures displayed diminished ATP and ADP levels compared with control neurons, whereas LRRK2 G2019S glutamatergic cultures displayed changes only in ATP levels (Figures S2A and S2B). No change was observed in sensory neuron cultures (Figures S2A and S2B). To further assess the reduction in ATP levels, we examined components of the electron transport chain by western blot using the OXPHOS antibody cocktail. Complex I (NDUFB8) and IV (COX II) were undetectable in all neuron cultures, and complex V (ATP5A) and II (SDHB) levels were unchanged (Figures S3A–S3C). However, the levels of complex III (UQCRC2) were significantly decreased in both dopaminergic and glutamatergic neuron cultures from LRRK2 G2019S iPSCs compared with control (Figures S3D and S3E). Therefore, it is possible that decreased expression or increased degradation of complex III in LRRK2 G2019S iPSC-derived dopaminergic and glutamatergic neuron cultures alters proton transport, thereby diminishing the proton gradient necessary for ATP production.

LRRK2 Kinase Inhibition Using GSK2578215A Does Not Restore Mitochondrial Dysfunction

Since the G2019S mutation confers hyper-kinase activity (West et al., 2005, Greggio et al., 2006, Jaleel et al., 2007, Luzon-Toro et al., 2007), we tested whether LRRK2 kinase inhibition using GSK2578215A (Reith et al., 2012) could rescue the observed mitochondrial defects in LRRK2 G2019S iPSC-derived neurons. We have previously shown that treatment with 1 μM GSK2578215A reduced expression of phosphorylated LRRK2 in LRRK2 G2019S iPSC-derived neurons (Schwab and Ebert, 2015). Therefore, we treated dopaminergic, glutamatergic, and sensory neuron cultures with GSK2578215A (1 μM) for 1 week prior to endpoint analysis. However, LRRK2 kinase inhibition only showed an effect in the LRRK2 G2019S Het line for mitochondrial velocity, but there was no effect in any LRRK2 G2019S line for ATP-linked, maximal, and spare respiration in dopaminergic neurons. GSK2578215A did not have a positive or negative effect on glutamatergic and sensory neurons (Figures S4A and S4B). These results suggest that increased kinase activity is not directly contributing to mitochondrial malfunction in this model system.

LRRK2 G2019S Dopaminergic Neuron Cultures Show Increased Sirtuin Expression but Decreased Deacetylase Activity

Sirtuins are a family of NAD+-dependent protein deacetylases that regulate many cellular processes, and changes in sirtuin expression levels may contribute to altered mitochondrial production, metabolism, and movement (Chalkiadaki and Guarente, 2012, Guarente, 2013, Herskovits and Guarente, 2013). Sirtuin-1 (SIRT1) has been shown to localize to the cytoplasm and nucleus where it plays a role in mitochondrial biogenesis (Tang, 2017). Western blot analysis of SIRT1 revealed a significant increase in LRRK2 G2019S iPSC-derived dopaminergic neurons compared with control cells in which SIRT1 levels were below the detection limit (Figures 4A and 4B). SIRT1 deacetylates and activates PGC1α (Rodgers et al., 2005, Liu et al., 2008, Canto et al., 2009, Wilson et al., 2010), which is an essential metabolic regulatory transcription factor. As expected based on the decreases in mitochondrial respiration and ATP pool, the levels of PGC1α were markedly decreased in iPSC-derived LRRK2 G2019S dopaminergic cultures (Figures 4C and 4D). In contrast, there was no difference in SIRT1 or PGC1α expression in LRRK2 G2019S glutamatergic cultures compared with control cultures (Figures S5A–S5C). Acetylated p53 is a known direct target of SIRT1 and showed a trend toward increased expression in LRRK2 G2019S iPSC-derived dopaminergic neurons compared with controls (Figures 4E and 4F), indicating reduced SIRT1 activity.

Figure 4

LRRK2 G2019S Dopaminergic Neuron Cultures Show Altered Levels of Sirtuin-1, PGC1α, and Acetylated p53

(A and B) Western blot analysis revealed a significant increase in (A and B) SIRT1 in all three LRRK2 G2019S iPSC-derived dopaminergic cultures compared with the three controls. REVERT was used as a loading control.

(C and D) In contrast, PGC1α expression levels were significantly reduced in all three LRRK2 G2019S iPSC-derived dopaminergic cultures compared with controls. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a loading control.

(E and F) Levels of acetylated-p53 were increased in LRRK2 G2019S iPSC-derived dopaminergic cultures, although this did not reach significance. REVERT was used as a loading control.

n.s., not significant. ∗p < 0.05 by one-way ANOVA with Tukey's post-hoc test. n = 3 independent experiments. All error bars are SEM. See also Figure S5.

Sirtuin-2 (SIRT2) is known to mediate microtubule-based cellular trafficking through its effect on tubulin acetylation (North et al., 2003). Considering the significant alterations in mitochondrial movement in LRRK2 G2019S iPSC-derived dopaminergic neurons (Figure 2), we tested whether SIRT2 expression and/or activity were likewise disrupted. Similar to SIRT1, SIRT2 protein levels were significantly increased in LRRK2 G2019S iPSC-derived dopaminergic neurons compared with control (Figures 5A and 5B), but deacetylase activity was impaired as noted by the significant increase in the levels of acetylated α-tubulin (Figures 5A and 5C).

Figure 5

LRRK2 G2019S Dopaminergic Neuron Cultures Show Increased Sirtuin-2 and Acetylated Tubulin

Western blot analysis revealed a significant increase in (A and B) SIRT2 and (A and C) acetylated tubulin expression in all three LRRK2 G2019S iPSC-derived dopaminergic cultures compared with controls. REVERT was used as a loading control for both targets. ∗p < 0.05 by one-way ANOVA with Tukey's post-hoc test. n = 3 independent experiments. All error bars are SEM.

Next, we tested sirtuin-3 (SIRT3) levels and activity as SIRT3 is the primary mitochondrial deacetylase shown to regulate mitochondrial bioenergetics and ATP generation (Weir et al., 2013). Immunoblot analysis of SIRT3 expression levels revealed significantly increased expression in LRRK2 G2019S dopaminergic cultures compared with control (Figures 6A and 6B). There was a trend toward a decrease in SIRT3 activity based on higher levels of acetylated SOD2 (Figures 6C and 6D). Although the changes associated with SOD2 did not reach significance, the trend is consistent with an overall diminished deacetylase activity in PD dopaminergic neurons. There was no difference in SIRT3 levels between control and LRRK2 G2019S glutamatergic neurons (Figure S6). To determine whether increased SIRT levels were associated with excessive kinase activity, we tested whether GSK2578215A was sufficient to lower SIRT levels. However, similar to a lack of effect on mitochondrial dysfunction, we did not observe changes in SIRT1, SIRT2, or SIRT3 expression levels (data not shown).

Figure 6

LRRK2 G2019S Dopaminergic Neuron Cultures Show Increased Sirtuin-3 and Acetylated SOD2

Western blot analysis revealed a significant increase in (A and B) SIRT3 expression and (C and D) a trend toward increased acetylated SOD2 expression in LRRK2 G2019S iPSC-derived dopaminergic cultures compared with control. SIRT3 was normalized to GAPDH, and SOD2 was normalized to REVERT. n.s., not significant. ∗p < 0.05 by one-way ANOVA with Tukey's post-hoc test. n = 3 independent experiments. All error bars are SEM. See also Figure S6.

Since sirtuins require NAD+ as a co-substrate to catalyze deacetylation, we hypothesized that low sirtuin activity in LRRK2 G2019S iPSC-derived dopaminergic neurons was due to low NAD+ levels. Indeed, NAD+ levels were significantly decreased in LRRK2 G2019S dopaminergic neurons compared with controls, a deficiency that was not observed in the other neuronal subtypes (Figure 7). Taken together, these data suggest that diminished NAD+ levels limit sirtuin deacetylase activity in LRRK2 G2019S iPSC-derived dopaminergic neurons, and may underlie the mitochondrial malfunction observed in this system and contribute to selective dopaminergic neuron loss in PD.

Figure 7

LRRK2 G2019S Dopaminergic Neurons Exhibit Significantly Reduced NAD+ Levels

HPLC analysis revealed that all three LRRK2 G2019S expressing dopaminergic neurons have significantly reduced NAD+ levels compared with controls. There was no difference between control and LRRK2 G2019S-expressing neurons for either glutamatergic or sensory neuron cultures. n.s., not significant. ∗p < 0.05 compared with control dopaminergic neurons by one-way repeated measures ANOVA with Tukey's post-hoc test. n = 3 independent experiments. All error bars are SEM.

Discussion

Evidence from in vitro and in vivo studies suggest that mitochondrial malfunction is a common characteristic in PD. Consistent with these reports, we show that LRRK2 G2019S iPSC-derived neurons display mitochondrial abnormalities, including altered content, distribution, trafficking, and respiration, which appear to be independent of increased kinase activity; interestingly, these mitochondrial defects manifest differently depending upon neuron subtype, with dopaminergic neuron cultures exhibiting the most profound changes compared with glutamatergic or sensory neuron cultures.

Regulation of mitochondrial distribution is essential to meet metabolic requirements and to remove aged and damaged mitochondria. The decreased content and reduced distal distribution of mitochondria within dopaminergic neurons may be due to damaged mitochondria being removed by autophagy or an inability to sufficiently replenish the neurites with healthy mitochondria, which has been reported for LRRK2 G2019S expressing mouse cortical neurons and SH-SY5Y cells (Cherra et al., 2013). This result is in contrast to our data for LRRK2 G2019S iPSC-derived glutamatergic neuron cultures. The discrepancy may be due to inherent differences between the model systems. In addition, the default forebrain telencephalic neuron patterning utilized here (Ebert et al., 2013, Kim et al., 2014) may not fully recapitulate the maturity and cell-type specificity found within the mouse cortex. Additional studies using specific iPSC cortical patterning techniques may be necessary to address the discrepancy (Mariani et al., 2012, Shi et al., 2012).

The distribution of mitochondria within neurons is dependent on efficient and regulated mitochondrial trafficking. The transport and function of mitochondria are linked as mitochondria supply ATP energy to motor proteins to transport them along the cytoskeleton to areas of high energy demand (Schwarz, 2013). Thus, abnormalities in mitochondrial trafficking may lead to abnormal mitochondrial distribution and protein transport and ultimately result in cellular dysfunction. We show that LRRK2 G2019S can affect mitochondrial trafficking based on increased retrograde mitochondrial velocity and mobile mitochondria in LRRK2 G2019S iPSC-derived dopaminergic neurons. These data correlate well with diminished mitochondrial content in the distal neurite. The trafficking results presented here are in contrast to a study using PD sporadic mitochondrial NT2 cybrids that showed a decline in overall mitochondrial velocity and a decrease in the percent mobile mitochondria (Esteves et al., 2014), but do support the findings in SH-SY5Y cultures (Cronin-Furman et al., 2013), and in LRRK2 G2019S and PINK1 Q456X expressing neurons (Cooper et al., 2012). The increase in percent mobile mitochondria within LRRK2 G2019S dopaminergic neurons suggests that the balance between motile and stationary mitochondria is perturbed. Anchored mitochondria provide local energy sources within neurons and are crucial to ensure that metabolically active areas are supplied with ATP, especially at presynaptic terminals. Moreover, stationary mitochondria are also required for axonal branching and maintenance (Kang et al., 2008). Therefore, disruption in stationary mitochondria may explain the deficient neurite elongation and branching that we and others have observed in LRRK2 G2019S-expressing neurons (MacLeod et al., 2006, Plowey et al., 2008, Cherra et al., 2010, Ramonet et al., 2011, Cooper et al., 2012, Sanchez-Danes et al., 2012, Reinhardt et al., 2013, Schwab and Ebert, 2015). However, additional studies are needed to establish a specific link between LRRK2 G2019S and mitochondrial motility.

The LRRK2 kinase inhibition results suggest that increased kinase activity due to the G2019S LRRK2 mutation is not directly contributing to the mitochondrial and SIRT abnormalities in this iPSC-based system. Although GSK2578245A has been shown to be selective for LRRK2 kinase inhibition (Reith et al., 2012), and we previously observed beneficial effects on morphology and calcium response in LRRK2 G2019S iPSC-derived sensory neurons (Schwab and Ebert, 2015), it is possible that a different kinase inhibitor or treatment paradigm would be more effective.

A number of studies have tested the effect of modifying SIRT1, SIRT2, or SIRT3 in PD models, but the results have been mixed. For example, increasing levels of SIRT1 or SIRT3 protect dopaminergic neurons from toxin-induced cell death (Tang, 2017). In contrast, most data indicate that inhibiting SIRT2 expression is neuroprotective in PD models (Outeiro et al., 2007, Godena et al., 2014, Liu et al., 2014, Chen et al., 2015, Di Fruscia et al., 2015), although another study suggested otherwise (Patel and Chu, 2014). More research is needed to determine the specific contributions of the various sirtuins to PD, but the divergent effects of sirtuin upregulation and inhibition may be due to the different targets of the individual sirtuins. Importantly, previous studies have not examined the levels of NAD+ necessary for sirtuin activity in human dopaminergic neurons.

Recent studies have shown that PGC1α and SIRT1 play key roles in cell metabolism and mitochondrial biogenesis (Rodgers et al., 2005, Revollo and Li, 2013). PGC1α is active in the deacetylated form, which is achieved by the deacetylase function of SIRT1, and plays a protective role in ROS defense. Here we show lower expression levels of active PGC1α in LRRK2 G2019S iPSC-derived dopaminergic neuron cultures compared with glutamatergic neuron cultures. We also found a strong trend toward increased acetylated p53 (∼2- to 3-fold) in LRRK2 G0291S iPSC-derived dopaminergic neurons compared with controls, a direct target of SIRT1, which may contribute to the vulnerability of dopaminergic neurons to oxidative stress in PD. To support this idea, acetylation of p53 at Lys382, the residue recognized by the antibody used here, is directly associated with activation of apoptosis and ROS production (Yamaguchi et al., 2009). Similarly, a small-molecule activator of PGC1α resulted in enhanced resistance against oxidative stress in human dopaminergic neurons (Makela et al., 2016).

SIRT2 is localized to the cytoplasm and, together with HDAC6, is largely responsible for deacetylation of α-tubulin (North et al., 2003). Studies have shown that increasing acetylated α-tubulin, through inhibition of HDAC6, increases mitochondrial movement in hippocampal neurons (Chen et al., 2010). Our studies are consistent with this notion as we find significantly increased mitochondrial velocity and retrograde trafficking in the LRRK2 G2019S iPSC-derived dopaminergic neurons. Mutant LRRK2 has been shown to associate with deacetylated tubulin and disrupt vesicle trafficking (Godena et al., 2014). However, our data suggest that the altered mitochondrial trafficking may be independent of mutant LRRK2 as neither LRRK2 G2019S-expressing glutamatergic neurons nor sensory neurons exhibit altered mitochondrial trafficking, and the LRRK2 kinase inhibitor GSK2578215A did not improve trafficking defects in dopaminergic neurons. It is important to note that our data are in contrast to a recent report showing that NAD+ levels and SIRT2 deacetylase activity were increased in PD patient-derived cybrid cell lines (Esteves et al., 2017). This discrepancy could be due to differences between the cybrid and iPSC model systems, but another plausible explanation could be due to variations in NAD+ levels among different cell types, as demonstrated by the variable levels in dopaminergic, glutamatergic, and sensory neurons.

As the main mitochondrial deacetylase, SIRT3 has also been shown to regulate many aspects of mitochondrial function, including metabolism, ATP generation, and limiting oxidative stress (Shi et al., 2005, Hallows et al., 2006, Lombard et al., 2007). Mechanistically, SIRT3 has been directly linked to reducing ROS production through activation of SOD2 (Qiu et al., 2010, Someya et al., 2010, Tao et al., 2010). In PD, SIRT3 has been implicated as a causative factor in dopaminergic neuron loss in MPTP- and rotenone-treated neurons (Zhang et al., 2016a, Zhang et al., 2016b). Although we observe increased SIRT3 protein levels in LRRK2 G2019S iPSC-derived dopaminergic neurons, as was the case for SIRT1 and SIRT2, increased expression did not correlate with increased activity, which is likely due to lower NAD+ levels in LRRK2 G2019S iPSC-derived dopaminergic cultures.

Our data demonstrate unique, and potentially LRRK2 kinase-independent, changes in sirtuin activity and NAD+ levels in human dopaminergic neurons expressing the LRRK2 G2019S PD-associated mutation and may underlie pathological mechanisms of dopaminergic neuron loss in PD. The results of these studies also provide further insight into the role of sirtuins in dopaminergic neurons and stress the importance of considering NAD+ levels in conjunction with sirtuin activity when designing therapeutic intervention for the treatment of PD.

Experimental Procedures

Cell Culture

Human iPSCs were obtained from publically available samples derived from two homozygous LRRK2 G2019S patients (ND35367∗C, ND40018∗C Coriell Institute) and one heterozygous LRRK2 G2019S patient (ND40019∗C Coriell Institute). Three previously characterized unaffected control lines were used (GM003814 Coriell Institute, GM02183 Coriell Institute, iPSK3; Ebert et al., 2009, Si-Tayeb et al., 2010, HD iPSC Consortium, 2012). A table describing the demographic, reprogramming, and source information for the control and PD iPSC lines is provided in our previously published work (Schwab and Ebert, 2015). Karyotypically normal iPSCs, used between passages 5 and 10, were grown in feeder-free conditions on Matrigel substrate in StemMACS iPS-Brew (Miltenyi). Karyotype G-banding was performed by Cell Line Genetics (Madison, WI). Neural progenitor cells (EZ spheres) were generated and maintained as described previously (Ebert et al., 2013). The use of iPSCs was approved by the Medical College of Wisconsin's Institutional Review Board (PRO25822) and the Human Stem Cell Research Oversight Committee.

Neural Differentiation

EZ spheres were differentiated into dopaminergic neurons using fibroblast growth factor-8, purmorphamine, and growth factors as described previously (Ebert et al., 2013). Telencephalic excitatory projection neurons were derived from EZ spheres as described previously (Ebert et al., 2013). Sensory neurons were derived from monolayer iPSCs as described previously (Chambers et al., 2012). For LRRK2 kinase inhibition experiments, 1 μM GSK2578215A (Tocris) was added at every feeding starting at 1 week prior to endpoint analysis. DMSO (1 μM) was used as the vehicle control. All analyses were performed at 5 weeks of total differentiation for dopaminergic neurons and 4 weeks of total differentiation for glutamatergic and sensory neurons.

Immunocytochemistry

Plated cells were fixed in 4% paraformaldehyde in PBS (pH 7.4) for 20 min at room temperature. Nonspecific labeling was blocked and the cells permeabilized prior to primary antibody incubation. Cells were subsequently labeled with the appropriate fluorescently tagged secondary antibodies. Hoechst nuclear dye was used to label nuclei. Antibodies are listed in Table S1.

Western Blot

Whole-cell lysates were isolated from dopaminergic and glutamatergic neuron cultures using 1× 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) Cell Extract buffer with protease inhibitors (Cell Signaling Technology). Twenty micrograms of protein was run on 10% Tris-HCl polyacrylamide gels (Bio-Rad), transferred to polyvinylidene fluoride membrane (Millipore), and probed following standard chemiluminescent methods. Alternatively, membranes were probed following the LI-COR fluorescent western blot protocol and scanned using the Odyssey Infrared Imaging System. Protein quantifications were normalized using glyceraldehyde-3-phosphate dehydrogenase or REVERT Total Protein Stain. Antibodies are listed in Table S1.

Mitochondrial Content and Distribution

Images were acquired from five random fields per coverslip and then analyzed using region of interest and line scan tools included in NIS Elements software. For region of interest measurements, the entire neuron was automatically traced and then TOM20 intensity was measured over this area. For line scan measurements, neurites projecting from a cell body were randomly selected and then measured for TOM20 intensity along the neurite length. Each neurite measurement was then divided into quartiles with respect to each neurite's overall length. A minimum of 250 neurons were analyzed from each line. Only TH+ (dopaminergic), TUJ1+ (glutamatergic), and peripherin+ (sensory neurons) were analyzed.

Mitochondrial Trafficking

Plated cells were loaded with 25 nM MitoTracker Green FM (Thermo Scientific; M7514) for 15 min, washed, and imaged on a Nikon fluorescent microscope. Videos were acquired from at least 5 random fields per coverslip, and a minimum of 250 neurons were analyzed and calculated using NIS Element kymograph software. Only the mitochondria found in neurites projecting from clearly identifiable cell bodies were used to determine anterograde and retrograde mitochondrial movement.

Mitochondrial Bioenergetics

iPSC lines were seeded onto Seahorse Bioscience 96-well microplates and patterned toward dopaminergic, glutamatergic, and sensory neurons. The plates were analyzed using a standard method for measuring OCR using the XFe96 Extracellular Flux Analyzer (Dranka et al., 2011, Patitucci and Ebert, 2016).

Nucleotide Measurements

ATP, ADP, and NAD+ were analyzed by high-performance liquid chromatography (HPLC) following perchloric acid precipitation, as described previously (Perez et al., 2010). Solvent A (125 μL; 0.1 M potassium phosphate and 4 mM tetrabutylammonium bisulfate [pH 6.0], diluted 64:36 in water [v/v]) was added to the supernatants (100 μL). Protein concentrations were determined by Bradford assay using 1× CHAPS buffer with PMSF and 1× DTT (Cell Signaling Technology). HPLC analysis of nucleotides was performed on a Supelco C-18 column using solvent A and solvent B (0.1 M potassium phosphate and 4 mM tetrabutylammonium bisulfate [pH 6.0], diluted 64:36 in methanol ]v/v]) with a flow rate of 1 mL/min. The column was equilibrated with solvent A, and the compounds were eluted during a linear increase in the level of solvent B to 50% between 1.5 and 5.5 min, followed by an increase to 65% over the next 7.5 min. ATP, ADP, and NAD+ peaks were measured for each sample, compared with the standards, and expressed in nmol per mg of protein.

Statistical Analysis

Data are from a minimum of three independent experiments, each with a minimum of three biological replicates. Samples were blinded prior to analysis. With consultation from the Department of Biostatistics at the Medical College of Wisconsin, statistical analysis was performed with GraphPad Prism software using one-way ANOVA, one-way repeated measures ANOVA, or two-way repeated measures ANOVA with Tukey post-hoc analysis. Results are presented as mean ± SEM and considered statistically significant at p < 0.05.

Author Contributions

Conception and Design, A.J.S. and A.D.E.; Collection and/or Assembly of Data, A.J.S., S.L.S., M.R.M., K.A.B., and A.D.E.; Data Analysis and Interpretation, A.J.S., S.L.S., K.A.B., and A.D.E.; Provision of Study Materials, J.A.C. and A.D.E.; Manuscript Writing and/or Editing, A.J.S., S.L.S., K.A.B., J.A.C., and A.D.E. All authors provided input for the final submitted manuscript.