Background

The neuropathological diagnosis of Parkinson’s disease (PD) is based on the loss of dopaminergic neurons in the substantia nigra, as well as by the presence of Lewy bodies (LB) and Lewy neurites in the substantia nigra and other brain regions [1]. A plethora of models have served as the foundation for research into PD pathogenesis. They range from yeast to primates and utilize nuclear gene expression based on inherited forms of PD, as well as dopaminergic neurotoxins [2].

Of these PD models, human cytoplasmic hybrids or “cybrids” are unique because platelet-derived mitochondrial DNA (mtDNA) from sporadic PD patients is expressed in mtDNA-free (Rho0) human neuroblastoma (SH-SY5Y) cells or other cell lines [3, 4]. Human cybrid cell lines provide an opportunity to study cellular consequences of the expression of mtDNA from sporadic PD patients.

Recent studies have clearly linked the consequences of mitochondrial dysfunction with sporadic and familial forms of PD [5, 6]. Our PD cybrid lines are characterized, in part, by abnormalities in oxygen utilization and mitochondrial electron transport chain (mtETC) function [3, 7]. In addition, PD cybrid lines spontaneously generate intracellular proteinaceous aggregates (cybrid Lewy bodies: CLB) that replicate the composition and ultrastructure of cortical LB [8]. Like LB in PD brain sections, CLB in our PD cybrid lines created from individual PD patients stain with eosin, Congo red, Thioflavin S, α-synuclein, and ubiquitin, as well as with markers for mitochondria, the proteasome and lysosomes [8]. Until recently, LB studies have been limited to the analysis of post-mortem tissues [9–11]. CLB formation in PD cybrids provides a unique opportunity to explore the influence of LB formation on cellular and mitochondrial function in a live cell model.

In an effort to better understand the relationship between CLB expression and mitochondrial and cellular dysfunction, we selected three different CLB-expressing PD cybrid cell lines that exhibit a range in oxygen consumption from severely compromised to near normal function. The three PD cybrid lines used in this paper were generated from platelets donated by patients at stage 2.0 Hoehn and Yahr Parkinson’s disease staging score (see Additional file 1). Other patient characteristics such as age, disease duration, L-dopa therapy and presence of dementia were also determined (see Additional file 1). Each of these three original PD cybrid lines (PD61Orig, PD63Orig, PD67Orig) was sub-cloned to enrich for cells expressing CLB by labeling the CLB in living cells with Congo red. Cell clusters expressing CLB from each PD line were selected, sub-cloned and expanded to generate cybrid lines PD61CLB, PD63CLB and PD67CLB (see Figure 1A).

Figure 1
figure 1

PD CLB cell lines were created from PD Orig cell lines. (A) PDCLB lines were created by selecting for cells containing large, circular, Congo red positive inclusions (CLB) as shown in this diagram. (B) PDOrig and PDCLB cybrid cells were stained with Congo red to visualize CLB (red) and MitoTrackerGreenFM to visualize mitochondria (green). The nucleus is indicated by white dashed line. Of the three PDOrig and PDCLB pairs, only PD63CLB had mitochondria that were morphologically different from PD63Orig. Scale bar = 5μm.

Full size image

Because other investigators have suggested that LB may be detrimental to neuronal function and survival [12–14], we anticipated that CLB-selected PD cybrid lines (PDCLB) would exhibit compromised function compared to the original PD cybrid lines (PDOrig). Contrary to our expectations, enrichment for CLB expression differentially affected each of the three PDOrig cybrid lines. Cellular and mitochondrial function improved in PD61CLB, worsened in PD63CLB and was unchanged in PD67CLB. Analysis of our results indicates that CLB expression in PDCLB cybrid lines did not correlate with the changes in cellular and mitochondrial function we detected. Rather, the change in function between PDOrig and PDCLB cybrid lines was determined by the presence or absence of mtDNA in nucleoids in PDOrig cells containing CLB.

Results and Discussion

CLB morphology and composition in PDOrig and PDCLBcybrid lines

Like LB in PD brain tissue, CLB in all PDOrig and PDCLB cybrid lines exhibited a consistent range in size (see Additional file 2) and stained uniformly with the histochemical dye Congo red (Figure 1B). Congo red binds to fibrillar α-synuclein as well as other misfolded, amyloidal (beta-pleated sheet folded) proteins [15]. While Congo red does not cross the blood brain barrier, it will cross living cell membranes and label intracellular amyloidal aggregates in vitro [16, 17]. Like LB in PD brain, CLB also labeled with antibodies to α-synuclein and polyubiquitin (see Additional file 3).

Using electron microscopy, CLB in all six PDOrig and PDCLB lines were structurally equivalent (Figure 2). The heterogeneous, dense granular appearance of CLB at the electron microscope level (EM) suggests that small protein aggregates contribute to the continuous formation of CLB (Figure 2). LB in PD brain tissue are also composed of aggregated, dense granular material [18]. CLB do not consistently contain straight filaments, consequently they more closely resemble cortical LB, rather than brainstem LB [8].

Figure 2
figure 2

Electron micrographs of CLB and mitochondria in PD Orig and PD CLB cybrid lines. CLB in PDOrig(A, C, E) and PDCLB(B, D, F) lines typically had an electron dense, compact core (core in PD63CLB is a double). The dense core is surrounded by a halo consisting of electron dense, heterogeneous aggregated material. Paired higher magnification images illustrate mitochondrial morphology in cells containing CLB. (A, B) The mitochondria in PD61Orig were enlarged with reduced numbers of cristae and a pale matrix. Mitochondria in PD61CLB were normal in appearance with a dense matrix and regular cristae. (C, D) Mitochondria in PD63CLB were swollen with reduced numbers of fragmented cristae and a little to no matrix while mitochondria in PD63Orig were swollen with fragmented cristae and a pale matrix. (E, F) Mitochondria in PD67Orig and PD67CLB were rod-like, dense, and featured organized cristae. N= nucleus. Scale bar = 1μm.

Full size image

Generation of PDCLBlines using Congo red to select CLB-expressing cells

In cultures of PD61Orig, PD63Orig and PD67Orig cybrid lines, cells with Congo red-labeled CLB were selected and propagated to generate sub-cloned lines enriched for cells with CLB (PD61CLB, PD63CLB and PD67CLB). PDCLB sub-cloned lines were propagated for 30 to 74 days, at which time numerous aliquots of each line were frozen for future use (see Figure 1A and Methods). Early in the sub-cloning process, CLB expression was increased (data not shown). However, assessment of Congo red-positive pixel intensity, pixel length or pixel area in cultures derived from current frozen stocks of PDCLB lines (at least 4-12 passages older) revealed no difference in the frequency of Congo red-positive staining among the PDOrig and PDCLB lines. Furthermore, there was no change in CLB expression frequency (see Additional file 2).

To understand why increased CLB expression was not sustained in PDCLB cybrid lines, it is important to think of CLB as aggresomes (as suggested by Olanow et al. [19]). Aggresomes are created in vitro by transiently inducing protein misfolding or the over-expression of proteins that are prone to misfolding [19]. Because the generation of misfolded and damaged proteins is continuous in PD cybrid lines, we consider CLB to be “professional aggresomes.” Like aggresomes, CLB form in the perinuclear region, contain punctate gamma tubulin staining and are composed of aggregated, damaged and misfolded proteins [8]. Some speculate that LB are “permanent trash dumps”, while others consider LB to be recycling centers [19]. Further studies will be necessary to address this important topic. One way to determine if CLB are trash dumps or recycling centers is to visualize changes in the distribution and expression of fluorescently labeled proteins in CLB using fluorescence recovery after photobleaching.

In PD substantia nigra, the expression level of LB appears to be constant (3-4%) irrespective of the duration of the disease. This observation is consistent with the idea that LB are constantly forming and being eliminated [20]. The frequency of CLB expression in PDCLB clones and parent PDOrig cybrid cell lines was also comparable to the frequency of LB found in PD patient brain [21].

There are several potential mechanisms that could contribute to CLB expression levels in PD cybrid lines. Rujano et al. [22] showed that aggresomes are asymmetrically distributed during somatic and stem cell mitosis. One daughter cell retains the aggresome while the other is free of damaged and misfolded proteins [22, 23]. The same asymmetric inheritance of aggresomes also happens when a cybrid cell with a CLB undergoes mitosis (data not shown). If CLB-free daughter cybrid cells proliferate more efficiently than cells with the burden of a CLB, the frequency of CLB-positive cells would decline during cell line expansion after initial sub-cloning. CLB-positive cells are not completely eliminated from cybrid lines because protein misfolding and aggregation is an ongoing process. Each PDOrig and PDCLB cybrid line (see Additional file 2) achieved a steady state level of CLB expression comparable to Rujano et al. [22].

Another potential mechanism that may play a role in the steady state level of CLB expression is cytoplasmic extrusion. Extracellular LB have been identified in PD brain sections using α-synuclein antibodies [24]. Doehner et al. [25] characterized the accumulation of granular Reelin/ β-amyloid deposits in mouse hippocampus. They detected Reelin-positive budd-like extrusions that they claim represent a protective reaction by postmitotic neurons with impaired protein degradation pathways. The extruded misfolded proteins, mitochondria, vacuoles and debris are then cleared by intrinsic glia [25]. Extracellular CLB have been seen in cultures of PD cybrid cell lines (See Additional file 4) that may be the result of cytoplasmic extrusion. The time-lapse studies of CLB-expressing cells needed to confirm this possibility are beyond the scope of this paper.

Mitochondrial morphology in PDOrig and PDCLBcybrid lines

Mitochondrial shape and changes in shape are intrinsically related to essential cellular functions such as mitochondrial membrane potential, ATP production, calcium signaling and ROS generation (reviewed in [26]). Consequently, the morphology of mitochondria either at the light or EM level provides insight into their functional capacity. Using light microscopy, we observed that mitochondria in PD61Orig, PD63Orig and PD67Orig cells containing CLB varied from elongate to short rod-like or globular in shape (Figure 1B). The mitochondrial morphology in the PDOrig lines was consistent with previous studies of PD cybrid cell lines [27].

Mitochondrial morphology at the light microscope level was qualitatively unchanged in PD61CLB and PD67CLB when compared to PD61Orig and PD67Orig, respectively (Figure 1B). However the mitochondria in PD63CLB were noticeably different from those in PD63Orig (Figure 1B). PD63CLB mitochondria were swollen, fragmented and globular, rather than rod-like.

The shift from rod-like mitochondria in PD63Orig to swollen, fragmented and globular mitochondria in PD63CLB is evidence of altered mitochondrial dynamics. Mitochondrial fragmentation can have many different causes [28]. Fragmented and dysfunctional, rather than elongated mitochondria, are more susceptible to mitophagy [29, 30]. Future studies will be necessary to reveal the specific cause of mitochondrial fragmentation in PD63CLB. Mitochondrial movement, especially in neuronal processes, is also influenced by mitochondrial shape (reviewed in [31]). Measurement of mitochondrial movement in the processes of differentiated PDOrig and PDCLB cybrid neurons is described below.

In light of these observations, we also processed fixed pellets of each PDOrig and PDCLB cell line for EM (Figure 2). Mitochondrial morphology at the EM level was qualitatively unchanged in PD67CLB when compared to PD67Orig (Figure 2E, F). The majority of the mitochondria in both PD67Orig as well as PD67CLB cells exhibited normal morphology with a rod-like shape, organized cristae and a dense matrix (Figure 2E, F). At the EM level, mitochondria in PD61Orig were enlarged (increased width) with a pale matrix and reduced numbers of cristae (Figure 2A, B). The mitochondrial morphology in PD61CLB was improved compared to PD61Orig with normal appearing rod-like mitochondria with a dense matrix and intact cristae (Figure 2A, B). The mitochondria in PD63Orig cells were swollen with a pale matrix and reduced and irregularly shaped cristae (Figure 2C, D). PD63CLB had primarily globular mitochondria with few cristae and a transparent matrix (Figure 2C, D).

Many critical mitochondrial functions are localized to cristae, such as the mtETC, iron/sulfur cluster biogenesis and the transport of mtDNA encoded proteins according to Zick et al. [32]. Therefore, the severe loss and disruption of cristae in PD63Orig and PD63CLB suggests that functions such as the mtETC and oxygen utilization should be dysfunctional. It has been estimated that 67% of all mitochondrial proteins are located in the matrix [33]. The matrix is the site for hundreds of enzymes, some of which participate in pyruvate and fatty acid metabolism and the citric acid cycle. Mitochondrial DNA enclosed in nucleoids, mitochondrial ribosomes and tRNAs are also localized in the matrix. The loss of matrix density in globular mitochondria in cybrid cells like PD63Orig and PD63CLB is indicative of a functionally disabled organelle that is a potential risk to the cell it occupies. Changes in mitochondrial morphology, such as the conversion of rod-like shapes to globular shapes can alter the cellular distribution of mitochondria. In complex cells like neurons, swollen and globular mitochondria can contribute to loss of synaptic function or cell death because these morphologically abnormal mitochondria cannot be transported into narrow caliber axons and dendrites [31, 34].

Mitochondrial oxygen consumption in PDOrig and PDCLBlines

Taking into account the abnormalities in mitochondrial morphology between some PDOrig and PDCLB cybrid lines, we measured oxygen consumption using a Seahorse Extracellular Flux Analyzer XF24 (Seahorse Bioscience) [35–37]. The three PDOrig lines that were selected for CLB cloning expressed a range of basal oxygen consumption values prior to cloning (Figure 3). If CLB expression is detrimental to cell function, then we anticipated that all three PDCLB lines would exhibit reduced oxygen consumption compared to PDOrig lines. Figure 3 shows the oxygen consumption rates (OCR) of confluent cultures of PDOrig and PDCLB lines at baseline and after sequential treatment with specific inhibitors (oligomycin to inhibit ATP synthase, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) to dissipate the proton gradient across the inner mitochondrial membrane, rotenone to inhibit complex I and antimycin A to inhibit complex III, see Methods). Use of these inhibitors permits the determination of key aspects of mitochondrial function including basal OCR, maximum capacity OCR, ATP-linked OCR, complex I-linked OCR and the non-mitochondrial (residual) OCR [36, 37]. Given the abnormal mitochondrial morphology shown above, it is important to determine if exposure to specific mitochondrial inhibitors during measurements of OCR cause any cell loss. For all three pairs, there was no difference between PDOrig and PDCLB lines in cell viability at the end of the experiment, as measured by live-dead counts (data not shown). The cell viability in these cell lines also did not significantly differ from the disease-free controls (n= three control lines).

Figure 3
figure 3

Respiration rates of PD Orig and PD CLB cybrid clones. (A, C, E) OCR was measured using the Seahorse XF24 analyzer for all three PD cybrid pairs and controls (not shown). Oligomycin, FCCP, rotenone, and antimycin A were added at the same time point for each experiment. in sequential to measure specific components of the mtETC (see Methods). (A, B) PD61CLB had significantly higher OCR but not residual (non-mitochondrial) respiration than PD61Orig (n=8). (C, D) PD63CLB had significantly lower OCR than PD63Orig (n=10). (E, F) OCR in PD67CLB did not differ significantly from PD67Orig (n=10). Two-way ANOVA with Bonferroni multiple comparisons; *, p<0.05; **, p<0.01; #, p<0.001.

Full size image

PD61CLB had significantly higher basal, maximal and complex I-linked OCR when compared to PD61Orig (Figure 3A, B) suggesting that sub-cloning of cells expressing CLB resulted in improved oxygen consumption in PD61CLB cells. This change was mtETC selective because there was no significant change in glycolysis (measured as extracellular acidification rate- ECAR, a surrogate for lactate production and aerobic glycolysis, data not shown) or in non-mitochondrial (residual) respiration. Significant improvements in basal, complex I-linked and maximum capacity OCR are also consistent with the improvement in mitochondrial ultrastructure in PD61CLB cells compared to PD61Orig (Figure 2B).

In contrast, PD63CLB had minimal levels of basal OCR and its response to mitochondrial inhibitors was significantly reduced when compared with PD63Orig (Figure 3C, D). There were also significant reductions in basal, maximum capacity, complex I- and ATP-linked OCR in PD63CLB compared to PD63Orig. However, there was no significant change in non-mitochondrial respiration. Therefore, this change in OCR in PD63CLB was selective for mtETC. There was also no compensatory up-regulation in ECAR in PD63CLB. This was surprising because other studies have shown that loss of complex I activity as a result of neurotoxicity induces a loss of OCR with a corresponding increase in ECAR [38]. This loss of mtETC function and OCR is consistent with the abnormal morphology of mitochondria in PD63CLB cells (Figure 2D).

Finally, PD67CLB exhibited basal, maximum capacity and complex I-linked OCR that was unchanged from PD67Orig (Figure 3E, F). There was also no change in ECAR (glycolysis) or non-mitochondrial respiration. These OCR values correlate with the consistent, normal morphology of mitochondria in PD67Orig and PD67CLB at the light and EM levels (Figures 1B, 2E, F).

Movement of mitochondria by axonal transport in PDOrig and PDCLBcybrid lines

One hypothesis for the dopaminergic neuron terminal degeneration seen in patients with sporadic PD is axonal transport failure [39]. Reduced axonal transport deprives the cell body of vital trophic factors and deprives axon terminals of synaptic vesicles and organelles like mitochondria [40]. Proper distribution of mitochondria to synapses is also crucial for synaptic homeostasis in response to changes in synaptic activity (reviewed by [41]). Based on studies of post-mortem sections of PD brain, Kanazawa et al. [42] concluded that LB and Lewy neurites are involved in altered axonal transport because LB can become Lewy neurites. Mitochondrial movement both anterograde and retrograde depends on motor proteins that utilize ATP [43]. Chu et al. [39] reported a decline in motor proteins early in sporadic PD brain that precedes other PD related changes like loss of dopamine or tyrosine hydroxylase. This loss of motor protein expression was also highest in nigral neurons containing α-synuclein inclusions.

We previously showed that axonal transport of mitochondria was significantly reduced in the tyrosine hydroxylase-containing processes of PD cybrids [44]. To study changes in axonal transport, PDOrig and PDCLB cybrid lines were differentiated into neuronal cells using low doses of staurosporine [45]. We measured the axonal transport velocity of fluorescently labeled mitochondria in individual cybrid neuron processes. In agreement with previous studies, mitochondrial velocity was reduced in all differentiated PDOrig cell lines compared to differentiated SH-SY5Y cells [44]. The velocity of mitochondrial movement in PD67CLB was not significantly different from PD67Orig (Figure 4). This outcome is consistent with the lack of change in mitochondrial morphology and OCR in PD67CLB after subcloning. Only PD61CLB exhibited a significant increase in mitochondrial velocity compared to PD61Orig (Figure 4). This result is consistent with other improvements in mitochondrial morphology and OCR (Figures 2B, 3A, B).

Figure 4
figure 4

Mitochondrial movement in differentiated in PD Orig and PD CLB cell lines. The graph shows the velocity of mitochondrial movement in processes of differentiated PD61Orig (n=4 cultures, 73 processes), PD61CLB (n=4 cultures, 94 processes), PD67Orig (n=4 cultures, 98 processes), PD67CLB (n=5 cultures, 108 processes), PD63Orig (n=8 cultures, 167 processes) and PD63CLB (n=4 cultures, 80 processes). The velocity of mitochondrial movement was significantly higher in PD61CLB when compared with PD61Orig determined by a t-test assuming equal variance. *p<0.03 1-tail and p<0.05 2-tail. The velocity of mitochondrial movement in SH-SY5Y neuronal processes was 0.220 microns per second +/- 0.01S.E.M. (n=3 cultures) and 0.236 microns per second +/-0.017S.E.M. In neuronal processes from three control cybrids (CNTL56, n=2 cultures, CNTL 64, n=3 cultures, CNTL68, n=3 cultures). Two montages illustrate the movement of mitochondria stained with MTRed during a 27sec period. Moving mitochondria in PD61Orig are marked with yellow and blue lines. Moving mitochondria in PD61CLB are marked with yellow, blue and red lines and a blue circle.

Full size image

Given the poor mitochondrial morphology and OCR in PD63CLB, it was a surprise that mitochondrial velocity was not comparably reduced (Figure 4). Whether induced by low dose staurosporine or retinoic acid, neuronal differentiation imposes increased demands for ATP and increased cellular stress. It is likely that cells with poor oxygen utilization and mitochondrial function are incapable of completing the process of differentiation. Therefore the most dysfunctional cells in PD63CLB should fail to differentiate. Additional file 5A depicts typical phase contrast images of PDOrig and PDCLB cybrid cells after differentiation. The cell density in a microscope field of differentiated PD63Orig and PD63CLB cybrid neurons was noticeably lower than the cell density for other cybrid lines. This information is also presented graphically in Additional file 5B. These data support the proposition that PD63orig cells with inadequate mitochondrial function were unable to differentiate and only those cells with sufficient mitochondrial function differentiated into neurons. Consequently the velocity of mitochondrial movement in differentiated PD63CLB neurons was comparable to other PD cybrid cell lines.

Nucleoid density in PDOrig and PDCLBcybrid cell lines

The sub-cloning of cybrid cells expressing CLB did not result in a uniform change in mitochondrial function among the PD cybrid cell line pairs. To establish if changes resulting from sub-cloning could be due to changes in mtDNA distribution, we first visualized nucleoids. Nucleoids are structures consisting of one or more mtDNA molecules and associated proteins like single-stranded DNA binding protein, Twinkle, mtDNA helicase and mitochondrial transcription factor A (TFAM) among others [46–48]. To visualize nucleoids, we used the DNA stain PicoGreen in combination with MitoTracker CMXRos (MTRed; Figure 5A, C, E) in live PDOrig and PDCLB cybrid cells. Nucleoid content was scored as “low/null” or “high” in cells from each of the PD cybrid pairs (Figure 5B, D, F, see Methods). Rho0 cells that lack mtDNA are devoid of PicoGreen staining and nucleoids (data not shown and [46]). PicoGreen staining is also independent of membrane potential or mtETC function [46]. PD61Orig contained cells that fell into the “low/null” category as well as cells with “high” numbers of nucleoids (Figure 5A, B). In contrast, PD61CLB contained significantly fewer cells in the “low/null” category and more cells in the “high” nucleoid category compared to PD61Orig (Figure 5B). This increase in cells with “high” numbers of nucleoids is consistent with previous data in this paper showing an improvement in mitochondrial function and morphology in PD61CLB. PD63CLB had significantly more cells that scored “low/null” and fewer cells in the “high” category than PD63Orig (Figure 5D). This was not surprising given the decline in PD63CLB mitochondrial function and morphology. Levels of nucleoid expression in PD67Orig and PD67CLB were comparable and consistent with previous data in this paper (Figure 5F).

Figure 5
figure 5

Nucleoid content in PD Orig and PD CLB cybrid lines. (A, C, E) Representative images of nucleoids in the PDOrig and PDCLB pairs using PicoGreen and MTRed (green: DNA, red: mitochondria). Yellow puncta are nucleoids in mitochondria. Cells that lack nucleoids are red with no yellow puncta (A). The nucleoids in PD63Orig and PD63CLB(C) were larger in diameter. (B, D, F) Cells were scored for nucleoid content as either “null/low” or “high” (see Methods) and graphed as a percentage of total cells, n= 5; #, p<0.001. (A, B) Consistent with other data, PD61CLB showed a significant increase in percent of cells that scored “high” and had fewer cells that scored “null/low” for nucleoid content compared to PD61Orig. (C,D) Conversely, PD63CLB showed an increase in cells that scored “null/low” and decrease in cells with “high” nucleoid density compared to PD63Orig. (E,F) There was no difference in nucleoid content between PD67CLB and PD67Orig. Two-way ANOVA with Bonferroni multiple comparisons, n = 5; #, p<0.001. Scale bar=5μm.

Full size image

In light of these results, we visualized nucleoids in individual PDOrig and PDCLB cells containing CLB using fluorescent markers: Congo red (CLB and small protein aggregates), PicoGreen (nucleoids) and MitoTracker Deep Red (mitochondria) as shown in Figure 6. Remarkably, we found that all CLB-positive PD61Orig cells contained numerous nucleoids (Figure 6A, top panel). The same result was true of PD67Orig (Figure 6C, top panel). However, the majority of CLB-expressing cells in PD63Orig did not contain nucleoids (Figure 6B, top panel). Taken together, these results indicate that the nucleoid content of the PDOrig cells containing CLB correlates with changes in mitochondrial quality and function detected in PDCLB cell lines. PD61CLB had better mitochondrial quality and function because it was sub-cloned from PD61Orig CLB-expressing cells containing numerous nucleoids. PD63CLB had reduced mitochondrial quality and function because it was sub-cloned from CLB-containing cells in PD63Orig with few nucleoids. CLB-containing cells in PD67Orig had numerous nucleoids and these cells yielded the PD67CLB cybrid line that also had cells with high numbers of nucleoids as well as adequate mitochondrial quality and function.

Figure 6
figure 6

Nucleoid content in cells containing CLB in PD Orig and PD CLB cell lines. (A,B,C) Representative images of cells containing CLB in PDOrig and PDCLB pairs. Cells were triple-stained with PicoGreen, Congo red, and MitoTracker Deep Red (green: DNA, red: CLB, purple: mitochondria). (A) CLB-containing cell in PD61Orig contained nucleoids that co-localized with mitochondria. PD61CLB (bottom panel) also featured CLB-containing cells with contained nucleoids. (B) CLB-containing cells from PD63Orig (top panel) and PD63CLB (bottom panel) did not contain nucleoids. (C) Cells containing CLB in PD67Orig (top panel) and PD67CLB (bottom panel) contained nucleoids. Scale bar=5μm-.

Full size image

Gene expression levels for mtETC genes in PDOrig and PDCLBcell lines

Since nucleoids contain mtDNA, we examined mtDNA copy number and expression in PDOrig and PDCLB lines. Previous studies have shown that complex I is damaged and functionally impaired in post-mortem PD cortex homogenates [49]. Analysis of PD cybrid cell lines (including the three PDOrig lines included in the paper) showed that complex I gene expression was reduced and showed a robust correlation with the changes in mtETC gene expression found in post-mortem PD cortex [50]. Enzymatic dysfunction related to complex I assembly is often associated with deficiencies in complexes III and IV because complex I assembly intermediates act like a scaffold for the assembly of other complexes in the mtETC [51]. We therefore measured gene expression and gene copy number for mitochondrial genes ND2 and ND4 (complex I), CO3 (complex IV) and 12s ribosomal RNA using quantitative real-time polymerase chain reaction (RT-qPCR) to create mitochondrial gene expression and copy number profiles (Figure 7).

Figure 7
figure 7

Quantitative real-time PCR analysis of gene profiles in PD Orig and PD CLB cell lines. (A) Mitochondrial gene expression for four mtDNA encoded genes (ND2, CO3, ND4, 12srRNA) measured using RT-qPCR from cDNA and compiled to create a gene expression profile. Starting quantities were normalized to the geometric mean for each cell line and graphed as the average fold change from the mean from three CNTL cybrid lines (56, 64, 68) for each gene in this profile. (B) Mitochondrial gene copy numbers for the same four genes was assayed from gDNA and expressed as described above. (C) Mitochondrial biogenesis gene expression (PGC-1α, NRF1, TFAM, TFB1M) was measured from cDNA and expressed as described above. Student’s t-test, with Welch’s correction in the case of non-equal variances; *, p<0.05; **, p<0.01; #, p<0.001.

Full size image

Figure 7A and B show that mitochondrial gene expression in PD61CLB increased nearly 3-fold and copy number increased more than 2-fold over PD61Orig. This observed improvement in mitochondrial gene expression and copy number in PD61CLB is consistent with the improved nucleoid content in PD61CLB cells. Furthermore, mtDNA copy number improved to control levels in PD61CLB while mtDNA expression levels were more than 2-fold higher than control levels. These results also suggest a potential mechanism for the improved mitochondrial morphology, mitochondrial respiration and mitochondrial movement observed in PD61CLB (Figures 1B, 2A, 2B, 3A, 3B, 4A) and supports our conclusion that mtDNA content in the cells containing CLB in PD61Orig was a driving factor in the functional improvements we observed in PD61CLB.

Mitochondrial gene expression was unchanged and remained reduced, as compared to control in both PD63Orig and PD63CLB (Figure 7A). There was a slight but significant increase (less than 2-fold) in mitochondrial gene copy number in PD63CLB (Figure 7B). This increase in gene copy number was surprising. We expected a decrease in mitochondrial gene copy number considering the reduced number of cells that scored “high” for nucleoid content (Figure 5C, D). Ashley et al. [46] suggested that fusion of nucleoids can occur as cells strive to maintain mtDNA copy number. Also, there is a linear relationship between mtDNA content and nucleoid volume [52]. Consequently, the large nucleoids in PD63CLB (Figure 5C) may harbor increased numbers of mtDNA. Additionally, PD63CLB had a decline in mitochondrial function and decrease in mitochondrial respiration, as compared with PD63Orig (Figures 1B, 2C, 2D; 3C, 3D, 4A). We speculate that this slight increase in mitochondrial gene copy number could be a compensatory mechanism. Since there was no change in mitochondrial gene expression, the increase in mtDNA copy number did not have downstream functional consequences. The poor cellular and mitochondrial function in PD63CLB reflects this outcome.

As expected, there was no change in mitochondrial gene expression or copy number between PD67Orig and PD67CLB (Figure 7A, B). Both of these cell lines exhibited similar mitochondrial morphology, mitochondrial respiration, mitochondrial movement and nucleoid content (Figures 1, 2, 3, 4). These results suggest that creation of PD67CLB from PD67Orig cells containing CLB did not substantially alter mtDNA genetic or phenotypic expression.

Biogenesis gene expression in PDOrig and PDCLBlines

Cellular regulation of mitochondrial biogenesis is critical for the maintenance of a functional pool of mitochondria in neurons [53]. In fact, the mitochondrial biogenesis pathway has emerged as a potential therapeutic target for PD [53, 54]. Peroxisome proliferator-activated receptor gamma co-activator 1-alpha (PGC-1α) is a transcriptional co-activator and serves as the master regulator of mitochondrial biogenesis (reviewed by [55]). A genome-wide analysis of PD patients and controls found that PGC-1α expression was reduced in PD patients [54]. Overexpression of PGC-1α in neurons was found to be protective in a neurotoxin mouse model of PD [56]. In cell culture, primary fibroblasts and cybrids generated using mtDNA from patients with mitochondrial diseases showed improved respiration after increased PGC1-α expression [57]. PPAR (peroxisome proliferator-activated receptor) agonists, such as bezafibrate (an agonist of PPARα), have also been shown to improve mitochondrial function in patient fibroblasts and myoblasts [58] and in a mouse model of mitochondrial disease [59].

To explore what role mitochondrial biogenesis plays in the mitochondrial changes found between the PDOrig and PDCLB lines, we measured the expression of four mitochondrial biogenesis genes including PGC-1α, nuclear respiratory factor 1 (NRF1), mitochondrial transcription factor B1 (TFB1M) and mitochondrial transcription factor A (TFAM), and used their expression levels to create a mitochondrial gene biogenesis profile. NRF1 is a DNA-binding protein that serves to positively regulate nuclear-encoded subunits of the mtETC [60]. In contrast, TFB1M and TFAM bind directly to mtDNA to initiate mitochondrial-encoded gene transcription [61, 62]. These four genes represent control of nuclear- and mitochondrial-encoded mtETC gene transcription, thereby creating a gene expression profile that can be used to quantitate cell-wide activation of mitochondrial biogenesis.

In PD61CLB, expression of mitochondrial biogenesis genes was significantly increased by approximately 1.5-fold compared to PD61Orig (Figure 7C). Improved biogenesis is consistent with the increased mitochondrial gene expression and mtDNA copy number described above (Figure 7A, B) and with the general improvement in cellular and mitochondrial function in PD61CLB compared to PD61Orig. This improvement in biogenesis may represent a shift in the population of cybrid cells in PD61CLB to include more cells with improved mtDNA gene expression and copy number, or it may represent the improved expression of mitochondrial genes within cells. PGC-1α enhances mtETC function and biogenesis by integrating cellular signals such as AMP/ATP ratios via the AMP-activated kinase (reviewed in [63]). Further research into this area would be beneficial for understanding the therapeutic potential of PGC-1α manipulation.

There was no difference in biogenesis gene expression between PD63Orig and PD63CLB (Figure 7C). This was not unexpected due to the decline in mitochondrial function in PD63CLB, compared to PD63Orig (Figures 3, 4). There was also no change in biogenesis gene expression between PD67Orig and PD67CLB (Figure 7C). This was also expected because there was no difference in mitochondrial gene expression, gene copy number, or mitochondrial function between these two cell lines.

It is important to remember that in the cybrid model, mtDNA from individual patients is expressed against a common background of nuclear genes from the SH-SY5Y parent cell line. The differences we observed in expression of nuclear mitochondrial biogenesis genes across different cybrid lines are ultimately a consequence of the presence of individual patients’ mtDNA. However, the differences in mitochondrial biogenesis signaling across PDCLB compared to PDOrig cell lines within each cybrid line are derived from selection of CLB-producing cybrids. We found that selection of cybrid cells for CLB expression had a differential effect on mitochondrial biogenesis. Mitochondrial biogenesis gene expression improved in PD61CLB, but was unchanged in PD63CLB and in PD67CLB. The molecular origins of these different consequences of CLB cloning remain unclear but suggest that the correlation between CLB expression and mitochondrial biogenesis signaling is not consistent.

Concluding Remarks

Due to the lack of PD models that spontaneously make LB, studies have been limited to cataloging contents or inferring mechanism of formation and cellular consequences of LB formation from post-mortem tissue (for example, [9–11]). While neuropathological studies have been very valuable, a live cell model would help investigate the biological significance of LB for PD pathogenesis. The discovery that PD cybrid cell lines spontaneously form and express CLB has provided us with a much needed live cell model of LB. CLB display the components found in LB in PD brain including eosinophilia, α-synuclein-, ubiquitin-, parkin- and Thioflavin S-staining [8]. Furthermore, CLB in PD cybrid lines are generated without the need for genetic over-expression of molecules like α-synuclein or inhibiting proteolytic processes such as proteosomal degradation or autophagy. PDOrig cybrid lines demonstrate that expression of the PD patient platelet mtDNA genes is responsible for the formation of CLB. By generating PDCLB cybrid cell lines from PDOrig cybrid cells expressing CLB, our aim was to create a model that focused on the cellular consequences of CLB formation.

The clinical significance of LB has been widely debated [64]. LB expression in the basolateral nucleus has been associated with visual hallucinations in PD [65]. Also, there are increased numbers of LB in demented versus non-demented PD brain sections from cortex, limbic structures and amygdala [66]. Several research groups have argued that LB are detrimental and contribute to neuronal degeneration in PD [12–14]. Harrower et al. [67], in contrast, proposed that LB mark a “struggling cell,” a concept of the LB that is consistent with the results presented in this paper. Harrower et al. [68] and others have suggested that LB are formed by neurons in a effort to maintain normal function in the face of an ongoing pathological process [11, 69–72]. The nature of the pathological process that results in LB formation has also been a subject of speculation. Zhou et al. [73] and Lin et al. [74] suggested that mitochondrial dysfunction precedes and drives LB pathology and neurological dysfunction in PD. The results presented in this paper support the proposal that mitochondrial dysfunction drives CLB pathology. Other alterations in cellular functions such as over-expression and aggregation of mutated α-synuclein or over-expression of wild type α-synuclein can also generate LB [73, 75].

If CLB are detrimental for cells, then all three PDCLB lines should have exhibited worsening of cellular and mitochondrial functions. This did occur in PD63CLB, but not in PD61CLB or PD67CLB. If CLB are an asset with beneficial functions, then cellular and mitochondrial function should improve. This was the outcome in PD61CLB but not in PD63CLB or PD67CLB. Sub-cloning CLB-expressing cells to produce PD67CLB did not substantially alter cellular or mitochondrial function. These data suggest that selection and sub-cloning of PDOrig lines for CLB expression did not drive the dysfunction in PDCLB cybrid lines. While α-synuclein aggregation does occur in PDOrig and PDCLB cybrids, levels of α-synuclein expression measured by RT-qPCR were not significantly different between PDOrig and PDCLB lines (see Additional file 6). Further studies are needed to determine what role LB play in PD. As proposed by Kanazawa et al. [42], LB and Lewy neurites may be more specifically involved in the disruption of axonal transport and the removal of damaged and misfolded proteins.

Our studies of PDOrig and PDCLB cybrid cell lines suggest that mtDNA is key to the expression of cellular and mitochondrial dysfunction, such as altered ETC activity and oxygen utilization, abnormal mitochondrial morphology, changes in axonal transport, etc. The PDCLB cell lines that featured significant changes in mitochondrial function and morphology also had distinct changes in mtDNA copy number and/or expression from the PDOrig lines. Our data supports the proposition by Esteves et al. [6] that mtDNA dysfunction is at least partly responsible for mtETC defects in sporadic PD. Exner et al. [5] further concluded that mitochondrial dysfunction is a “common denominator” in the pathogenesis of sporadic and familial PD.

One intriguing finding in our data was the identification of individual cells with CLB that appeared to lack nucleoids and functional mtDNA (see Figure 6B). Since formation of an aggresome or CLB is ATP-driven, it seems unlikely that a cell without mtDNA could generate a CLB. This idea leads to speculation that loss of mtDNA and nucleoids could be part of PD pathogenesis. CLB-bearing PD cybrid cells without mtDNA or nucleoids can survive in culture because of the supportive culture conditions. Nucleoid-free and mtDNA-free neurons are unlikely to survive in vivo unless they can derive sufficient support from surrounding glia. The composition of LB in neuropathological tissue from pre-PD substantia nigra suggests the convergence of multiple pathways such as mitochondrial dysfunction, oxidative stress, oxidative protein damage and altered post-translational modification play a role in PD disease progression [76]. This concept is supported by a recent publication that showed nuclear α-synuclein binds to the PGC-1α promoter in vivo and in vitro, and alters mtDNA copy number and function [77]. Our contribution to this ever-changing field has been to demonstrate the important role that mitochondrial quality, dynamics and function play in PD.

Methods

Cybrid cell lines

Cybrid cell lines were created from individual patients and controls as described previously [3, 78]. Cells were grown in growth media (GM) consisting of high glucose Dulbecco’s modified Eagle medium (DMEM, Gibco, Life Technologies) with 10% fetal bovine serum, antibiotic/antimycotic, 100μg/ml sodium pyruvate and 50μg/ml uridine, as described previously, to support the survival of cells with mitochondrial dysfunction [8, 79]. For imaging, cells were plated on 35mm poly-lysine coated dishes (MatTek Corp.). Cell lines were only kept in culture for a maximum of two months. Cell pellets from PDOrig and PDCLB pairs were always thawed simultaneously and grown under the same culture conditions.

Generation of sub-cloned cybrid lines based on CLB expression

Glass-bottomed 6-well plates (MatTek Corp.) were treated with 200μg/ml poly-l-lysine/H2O (m.w. 30,000-70,000, Sigma-Aldrich) at room temperature for ~40 min. Wells were rinsed twice with sterile water and plates were stored dry until they were loaded with cell suspension (typically 20,000 cells). Selected cybrid lines were harvested from T75 CellStar flasks (Greiner bio-one) with 0.05% trypsin diluted in phosphate buffered saline. GM was used to quench trypsin activity prior to re-plating of cells into glass-bottomed 6-well plates. Cells were incubated at 37°C with 5% CO2 for 24-48 hours until cells divided into 2-4 cell clusters. Following an overnight (~18-24 hours) treatment with 100μM Congo red (Sigma-Aldrich) made up in GM, dishes were rinsed twice with GM without phenol red (clearGM) and further stained with 80nM Mitofluor Green (Invitrogen) in clearGM for 20 minutes at 37°C. Wells were again rinsed with clearGM and labeled cells were visualized with epi-fluorescence (Olympus IX-70 microscope) using fluorescein isothiocyanate (FITC) filters (Mitofluor Green) and Texas Red filters (Congo red). Clusters of cells with large (3-5μM) Congo red stained spheres were marked with an inked objective marker (Olympus) on the underside of the coverslip well. After removing GM from the wells, 6mM sterile glass cloning rings were seated around the marked cell clusters using sterile silicone grease (both from Thermo Fisher). Each cell cluster was trypsinized (see above) and re-plated into one well of a 24-well plate. The expression of Congo red positive CLB was later reassessed and the wells with the largest and most numerous Congo red stained spheres were retained and combined (2-5 clones per well). The other clones were discarded. Combined clones were cultured in GM and passed into larger wells as they became confluent. Sub-cloned cybrid lines were expanded into T25 flasks (Greiner bio-one). Each cybrid line was harvested and re-plated into coverslip-bottom dishes and re-selected for CLB expression using Congo red and MitoFluor Green. These colonies were expanded into T75 flasks, at which time aliquots from each line were frozen for subsequent study.

Quantification of Congo red positive fluorescence

Cells were plated in 35mm dishes as described above and grown for 2-4 days until at least 75% confluent. Cells were stained with Congo red at 100μM for 24 hours. Dishes were then washed two times with clearGM with 25mM Hepes (Gibco, Life Technologies). Dishes were blinded for image collection and quantification. Images were acquired using an Olympus FV1000 confocal microscope (60X objective) at room temperature. Ten representative fields were taken at random per dish and analyzed using MetaMorph image analysis software (Molecular Devices). Studies were repeated with cells from a different passage. Images were set to a common inclusive threshold and pixels over 1μm were measured for total pixel area, pixel intensity and pixel length. Pixel values were normalized to number of cells in each image. To calculate CLB frequency, Congo red positive inclusions over 1μm in diameter were counted for each set of ten images per dish. Number of CLB per dish was normalized to number of cells counted per dish. Student’s t-tests were run to compare the original and sub-clone pairs (Prism, Graph Pad).

Electron Microscopy

Sub-confluent T75 flasks for each cybrid line were fixed with 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1M phosphate buffered saline, processed for EM, sectioned and stained by staff members of the Advanced Microscopy Facility at the University of Virginia, as previously described [27]. Stained sections were imaged on a Jeol JEM-1230 transmission electron microscope at the Virginia Commonwealth University Microscopy Facility.

Nucleoid imaging and quantitation

Live cells were grown in 35mm dishes, stained with Quant-It PicoGreen dsDNA and MitoTracker Red CMXRos (both from Molecular Probes/Life Technologies) and imaged in clearGM as described above. Dishes were blinded prior to imaging. Six images were acquired at random per dish using an Olympus FV300 confocal microscope. Cell count totals were acquired by counting the PicoGreen positive nuclei per frame. Nucleoids were defined as areas of PicoGreen and Mitotracker Red colocalization. Cells with less than five nucleoids per cell were considered “low/null”. All others were considered “high”. Only cells with both mitochondria and a nucleus in focus were counted. Two-way ANOVA with Bonferroni multiple comparisons were run to compare PDOrig and PDCLB cell lines at “low/null” versus “high” (Prism, Graph Pad). For analysis of nucleoids in cells with CLB, cells were co-stained with Congo red, MitoTracker Deep Red (50nM for 45 minutes, Molecular Probes/Life Technologies), and PicoGreen. Cells were imaged in clearGM on an Olympus FV1000 confocal microscope.

Cellular Respiration

Oxygen consumption was measured using the Seahorse Extracellular Flux Analyzer (Seahorse XF24, Seahorse Biosciences) according to manufacturer’s instructions. In brief, cells were plated in Seahorse XF24 culture plates and grown for 24 hours to form a confluent monolayer. One hour prior to each experiment, growth media was exchanged for unbuffered DMEM, pH 7.4. The following inhibitors were used to obtain a bioenergetic profile: oligomycin (1μM), FCCP (300nM), rotenone (100nM), and antimycin A (10μM). For all inhibitors, the pH was adjusted to 7.4 prior to the experiment. For each Seahorse experiment, three basal measurements of the oxygen consumption rate (OCR) were acquired and calculated by the Seahorse XF. Compounds were added in the order mentioned previously, with two measurements following each inhibitor. At the end of each experiment, OCR values were normalized to protein content (Micro BCA Kit, Pierce). OCR values are reported as means ± SEM, except for uncoupled respiration (FCCP), where the highest value was used. Statistics were calculated using two-way ANOVA with Bonferroni multiple comparisons post-hoc tests in Prism software (GraphPad, Prism).

RT-qPCR

RNA and DNA were extracted from three independent PDOrig and PDCLB cell pellets from sub-confluent T175 flasks using the All Prep RNA/DNA Kit (Qiagen). RNA and genomic DNA (gDNA) were quantified using the NanoDrop 2000 (ThermoScientific). Complimentary DNA (cDNA) was made from RNA using the iScript cDNA Synthesis Kit (Bio-Rad). Quantitative real-time PCR (RT-qPCR) was run on cDNA and gDNA samples to measure mitochondrial gene and mitochondrial biogenesis gene expression, as well as mitochondrial gene copy number. Glyceraldehyde 3-phosphate dehydrogenase, beta-actin and 18sRNA were used as endogenous reference genes. Primer and probe sequences can be found in the Additional file 7. Starting quantities were calculated by Bio-Rad CFX manager software based on cycle threshold of known standards (human fetal brain cDNA, human gDNA or human mtDNA). Samples and standards were run in quadruplicate. All values were normalized to the geometric mean of each sample from the endogenous reference genes. PDOrig and PDCLB data was expressed as a fold change from the pooled mean of three age-matched disease-free control cell lines (Controls 56, 64 and 68) for each gene (see Keeney et al. [7]). The fold changes for each gene from the mean of each individual PDOrig or PDCLB cell line were then averaged to represent the fold increase or decrease from control across each gene profile. Mitochondrial cDNA gene expression and gDNA copy number profiles were made up of ND2, CO3, ND4 and 12srRNA. The mitochondrial biogenesis gene expression profile was made up of PGC-1α, TFAM, NRF1 and TFB1M. Statistical analysis was done using Prism software (GraphPad) using student t-test. In the cases of unequal variances, Welch’s correction was performed.

Axonal transport of mitochondria in the processes of differentiated PDOrig and PDCLBcybrid neurons

Proliferating PDOrig and PDCLB cells were harvested from T75 flasks with 0.05% trypsin (Invitrogen, Life technologies) as previously described [44]. 40,000 cells in 2ml GM were added to each #0 glass bottomed 35mm dishes (MatTek Corp). After 18-24 hours, GM was removed and the differentiation media (DM) consisting of 500ml of Neurobasal with 10ml B27 supplements (Invitrogen, Life Technologies) plus glutamine (0.5mM), pyruvate, uridine and antibiotic-antimycotic, as previously described [45]. Staurosporine (4nM-8nM) dilutions were made fresh in DM and replaced every 2-3 days. Differentiation was completed on day 12 [45].

To measure mitochondrial movement, PDOrig and PDCLB cybrid neuronal cells were incubated with 50nM MitoTracker CMXRos (MTRed; Invitrogen, Life Technologies) for 10 min at 37°C. Time-lapse recordings were made using an Olympus IX70 microscope equipped with epifluorescence and Nomarski optics, a Lambda 10-2 filter wheel, a Photometrics CoolSnap HQ progressive scan CCD camera and a heater/controller to maintain cybrid cells at 37°C during image collection (World Precision Instruments, Inc). Image stacks and velocity measurements were collected using MetaMorph Imaging System (Molecular Devices). For standard recordings, images were collected every 3 seconds for 2 min. Mitochondrial movement was measured in PD61Orig (n=4 cultures, 73 processes), PD61CLB (n=4 cultures, 94 processes), PD67Orig (n=4 cultures, 98 processes), PD67CLB (n=5 cultures, 108 processes), PD63Orig (n=8 cultures, 167 processes) and PD63CLB (n=4 cultures, 80 processes). Unlike other studies, the velocity of all mitochondria in each cell process was tracked individually whether they moved or not. The average velocity for individual mitochondria was calculated using intervals where movement occurred. Neutral density filters are used to reduce illumination from the mercury lamp and minimize phototoxicity.