https://orcid.org/0000-0002-0830-9403
Abstract
Amyotrophic lateral sclerosis and frontotemporal lobar degeneration are neurodegenerative diseases characterized by accumulation of insoluble aggregates of phosphorylated 43 kDa TAR DNA-binding protein (TDP-43) and linked with abnormal expansion of a hexanucleotide repeat in an intron of chromosome 9 open reading frame 72 (C9ORF72). However, the relationship between C9ORF72 mutations and TDP-43 aggregation remains unknown. Non-ATG-dependent translation of C9ORF72 repeats produces dipeptide repeat proteins, which form p62-positive aggregates in cerebral cortex and cerebellum of patients. Here, we show that the formation of poly-GA protein inclusions induced intracellular aggregation of endogenous and exogenous TDP-43 in cultured cells. Poly-GA aggregation preceded accumulation of phosphorylated TDP-43. These inclusions induced intracellular aggregation of phosphorylated TDP-43, but not tau or α-synuclein. Formation of phosphorylated TDP-43 aggregates depends on the number of poly-GA repeats. Detergent-insoluble fraction from cells co-expressing poly-GA and TDP-43 could function as seeds for further TDP-43 aggregation. These findings suggest a novel pathogenic mechanism that poly-GA protein aggregation directly promotes pathogenic changes of TDP-43 without the formation of nuclear RNA foci containing GGGGCC repeat expansion or loss-of-function of the C9ORF72 protein.
Introduction
Frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS) are closely related neurodegenerative disorders with overlapping neuropathological and genetic features. The most common genetic abnormality is expansion of a GGGGCC repeat in an intron of chromosome 9 open reading frame 72 (C9ORF72) (1,2). Both diseases are also characterized by protein inclusions, composed mainly of aggregates of TAR DNA-binding protein of 43 kDa (TDP-43), in the brains of patients (3,4).
While C9ORF72-repeat-expansion mutations are associated with formation of intracellular aggregates of phosphorylated TDP-43, these patients also have TDP-43-negative and p62-positive, neuronal cytoplasmic inclusions in the cerebellum, hippocampus and frontal cortex (5,6). C9ORF72 GGGGCC repeat expansions are translated in all possible reading frames through unconventional non-ATG-dependent mechanisms, resulting in production of dipeptide repeat (DPR) proteins, including poly-GA, poly-GP, poly-GR, poly-PA and poly-PR. These DPR species, especially poly-GA (7), are major components of TDP-43-negative inclusions (7–12), and have been proposed to cause disease through TDP-43-independent mechanisms. In support of this idea, poly-GR and poly-PR proteins are reported to cause neurodegeneration in Drosophila without aggregation of phosphorylated TDP-43 (13), and mammalian cell culture studies also indicate that poly-GA inclusions can cause neurotoxicity and neurodegeneration in the absence of TDP-43 aggregation (14–17). In addition, three groups have independently reported that GGGGCC repeat expansion in C9ORF72 disrupts nucleocytoplasmic transport in flies and yeast (18–20), providing a novel mechanism of poly-DPR protein neurotoxicity. Recently, Khosravi et al. reported that inhibition of nuclear import of TDP-43 by cytoplasmic poly-GA inclusions (21).
In contrast, Mackenzie et al. reported co-accumulation of phosphorylated TDP-43 and poly-GA proteins, with poly-GA aggregates present in the center of neurons, surrounded by accumulated TDP-43 in human diseased brains (22). This may imply that DPR pathology (accumulation of poly-DPR proteins) precedes TDP-43 accumulation, suggesting that formation of inclusions by poly-DPR proteins may promote intracellular TDP-43 aggregation, leading to neurodegeneration. Thus, at present it remains unclear whether poly-DRR protein accumulation and TDP-43 aggregation cause neurodegeneration through related or independent processes, or both, in ALS and FTLD patients with C9ORF72 mutation.
In the present study, we examined the relationships among expression of poly-DPR proteins and intracellular TDP-43 accumulation in vitro. Our findings indicate that poly-GA aggregation induces accumulation of endogenous and exogenous phosphorylated TDP-43. Aggregates including poly-GA and phosphorylated TDP-43 could function as seeds for further intracellular TDP-43 aggregation. Our results are consistent with the idea that poly-GA aggregation elicits pathogenic changes of TDP-43, leading to the onset of these diseases.
Results
Inclusions of poly-GA induce endogenous and exogenous phosphorylated TDP-43 aggregation in cultured cells
To determine whether accumulation of poly-DPR proteins influences formation of intracellular phosphorylated TDP-43 aggregates, we first developed a cellular model of poly-DPR protein aggregation. We generated synthetic constructs to express all reading frames of the GGGGCC repeat, encoding poly-GA, GP, GR, PA and PR proteins. These constructs, encompassing 100 repeats, contained a mixture of alternative codons to avoid GGGGCC repeats (16), and were subcloned into a pEGFP vector. SH-SY5Y cells were transiently transfected with GFP-tagged poly-DPRs 3 days prior to immunofluorescence analyses with anti-phosphorylated TDP-43 antibody, pS409/410. As shown in Figure 1, round inclusion-like structures positive for GFP were observed in cells expressing GFP-tagged poly-GA (GFP-GA), GFP-GR and GFP-PR. GFP-GA inclusions were localized mainly in the cytosol, while GFP-GR and GFP-PR inclusions were found in both cytosol and nucleus. Interestingly, in cells expressing GFP-GA, endogenous TDP-43 was phosphorylated and accumulated in the cytosol as shown in Figure 1. When either TDP-43 wild-type (WT) or mutant with deletion of the nuclear localization signal (78–84 residues: ΔNLS) was expressed together with GFP-DPR, phosphorylated TDP-43 was aggregated in cells co-expressing GFP-GA (Fig. 2). The number of phosphorylated TDP-43 aggregates in cells expressing GFP-GA and TDP-43 ΔNLS were greater than those in cells expressing GFP-GA and TDP-43 WT: the average number of phosphorylated TDP-43 aggregates was 4.3 ±1.1 (mean ± standard deviation) in an area of 80 ×80 μm for cells expressing GFP-GA and TDP-43 WT, while it was 6.7 ± 2.1 for cells expressing GFP-GA and TDP-43 ΔNLS (n = 6). We often observed phosphorylated TDP-43 aggregates surrounding GFP-GA inclusions (donut type) (Fig. 2). In cells expressing GFP-GA and TDP-43 WT, the average number of donut-type inclusions was 2.8 ± 0.89 in an area of 80 ×80 μm (n = 6), and thus the percentage of donut type was calculated to be 65.4%. In cells expressing GFP-GA and TDP-43 ΔNLS, the average number of donut-type inclusions was 4.5 ± 1.6 (n = 6), and thus the percentage of donut type was calculated to be 67.2%. In any case, significant cell death was not observed in cells expressing GFP-DPR.
Endogenous TDP-43 is phosphorylated and accumulated in cells containing poly-GA inclusions. Confocal laser microscopic analyses of SH-SY5Y cells transfected with GFP empty vector, GFP-tagged poly-GA (GFP-GA), GFP-GP, GFP-GR, GFP-PA or GFP-PR. These cells were immunostained with anti-phosphorylated TDP-43 (pS409/410) and counterstained with Hoechst 33342. Scale bars represent 20 μm. Note that there are endogenous phosphorylated TDP-43 aggregates positive for anti-pS409/410 antibody in only cells expressing GFP-GA.
Expression of poly-GA induces phosphorylated TDP-43 aggregation in cells expressing exogenous TDP-43. Confocal laser microscopic analyses of SH-SY5Y cells co-transfected with both GFP-tagged poly-GA (GFP-GA), GFP-GP or GFP-GR and TDP-43 WT or mutant lacking nuclear localization signal (ΔNLS). These cells were immunostained with anti-phosphorylated TDP-43 (pS409/410) and counterstained with Hoechst 33342. Scale bars represent 20 μm.
To biochemically confirm the expression of DPR proteins elicits intracellular TDP-43 aggregation, we next separated lysates of cells expressing each GFP-tagged DPR protein into Sarkosyl-soluble (Sar-sup) and Sar-insoluble (Sar-ppt) fractions, and analyzed these by immunoblotting using anti-GFP antibody. As shown in Figure 3A, GFP-GP and GFP-PA were detected in Sar-sup fractions as bands between 30–70 and 100–130 kDa, respectively. On the other hand, GFP-GA, GFP-GR and GFP-PR, which formed intracellular aggregates (Fig. 1), were hardly detected in Sar-sup fractions. Instead, these proteins were found in the stacking gel of Sar-ppt fractions, indicating that they form high molecular weight insoluble complexes. The most intense Sar-ppt band was found in cells expressing GFP-GA, suggesting that poly-GA is the most aggregation-prone DPR. We also analyzed these fractions by immunoblotting using anti-DPR antibodies. As shown in Supplementary Material, Figure S1, GFP-GP, GFP-PA and GFP-PR were detected in both of Sar-sup and Sar-ppt, while GFP-GA and GFP-GR were observed in the stacking gel of Sar-ppt fractions. When we probed fractions for phosphorylated TDP-43, we observed a band of phosphorylated TDP-43 (red arrowhead in Fig. 3A) in the Sar-ppt fraction of cells expressing GFP-GA, indicating that endogenous TDP-43 is phosphorylated and aggregated preferentially in cells expressing GFP-GA, in agreement with the results in Figure 1.
Biochemical evidence of phosphorylated TDP-43 aggregation in cells expressing poly-GA. (A) Immunoblot analyses of proteins extracted from cells expressing GFP empty vector, GFP-tagged poly-GA (GFP-GA), GFP-GP, GFP-GR, GFP-PA or GFP-PR. Proteins were extracted from cells with 1% Sarkosyl, and Sarkosyl-soluble (Sar-sup) and -insoluble (Sar-ppt) fractions were subjected to immunoblot analyses. Blots were probed using anti-GFP monoclonal antibody (anti-GFP), anti-TDP-43 monoclonal antibody (anti-TDP mono), anti-phosphorylated TDP-43 (anti-pS409/410) and anti-tubulin antibodies. Note that endogenous phosphorylated TDP-43 is aggregated in cells expressing GFP-GA (red arrow). (B) Immunoblot analyses of proteins extracted from cells co-expressing GFP empty vector, GFP-GA, GFP-GP, GFP-GR, GFP-PA or GFP-PR and TDP-43 WT (TDP WT). Blots were probed using anti-TDP mono, anti-S409/410 and anti-tubulin antibodies. Phosphorylated full-length TDP-43 (red arrowhead) and its CTFs (red arrows) were observed in Sarkosyl-insoluble (Sar-ppt) fraction of cells expressing GFP-GA. (C) Immunoblot analysis of proteins extracted from cells co-expressing GFP empty vector, GFP-GA, GFP-GP, GFP-GR, GFP-PA or GFP-PR and TDP-43 lacking NLS (TDP ΔNLS). Blots were probed using anti-S409/410 and anti-tubulin antibodies. Phosphorylated TDP-43 (red arrowhead) and its CTFs (red arrows) are observed in Sar-ppt fraction of cells expressing GFP-GA.
We also examined whether overexpressed TDP-43 is phosphorylated and aggregated by poly-GA expression in cells. SH-SY5Y cells were transiently transfected with plasmids encoding GFP-DPR, and either TDP-43 WT or ΔNLS, and cell lysates were fractionated into Sar-sup and Sar-ppt, followed by immunoblot analyses. Overexpression of TDP-43 WT and GFP-GA resulted in an increase in phosphorylated, insoluble full-length TDP-43 (red arrowhead in Fig. 3B) and C-terminal fragments (CTFs: red arrows in Fig. 3B) of TDP-43. Similarly, we found that full-length TDP-43 and CTFs are phosphorylated and aggregated in cells overexpressing the TDP-43 ΔNLS mutant with GFP-GA (Fig. 3C). The signal intensity of phosphorylated, insoluble TDP-43 was higher in extracts of ΔNLS-expressing cells compared with WT-expressing cells, suggesting that cytoplasmic poly-GA inclusions may cause preferential aggregation of cytoplasmic, rather than nuclear TDP-43, in accordance with the results in Figure 2. Taken together, our results suggest that formation of cytoplasmic poly-GA inclusions in SH-SY5Y cells triggers intracellular aggregation of phosphorylated TDP-43.
Poly-GA aggregation precedes accumulation of phosphorylated TDP-43
To see whether aggregation of TDP-43 is induced by poly-GA inclusions, we next investigated the temporal relationship between poly-GA inclusion formation and TDP-43 aggregation. We transiently transfected cells with plasmids expressing GFP-DPRs and analyzed lysates at various time points. As shown in Figure 4, GFP-GA proteins were present in Sar-ppt fraction 1 day after transfection, and remained in this fraction at all time points. In contrast, phosphorylated endogenous TDP-43 was hardly detected in the insoluble fraction until 2 days after transfection and was greatly increased at 3 days after transfection. Even when TDP-43 WT was overexpressed along with GFP-GA (Fig. 4), the signal intensity of phosphorylated full length TDP-43 and CTF increased, but phosphorylated TDP-43 proteins were not found in the insoluble fraction until 2 days after transfection. This indicates that aggregation of phosphorylated TDP-43 requires poly-GA expression and occurs after formation of poly-GA inclusions.
Time course of co-aggregation of poly-DPR and phosphorylated TDP-43. (A) Immunoblot analyses of proteins extracted from cells expressing GFP empty vector, GFP-tagged poly-GA (GFP-GA), GFP-GP or GFP-GR. Sarkosyl-soluble (Sar-sup) and -insoluble (Sar-ppt) fractions at Days 1–3 were prepared from cells and subjected to immunoblot analyses. Blots were probed using anti-GFP monoclonal antibody, anti-phosphorylated TDP-43 (pS409/410) and anti-tubulin antibodies. Endogenous phosphorylated TDP-43 is aggregated in cells expressing GFP-GA at Days 2 and 3 (red arrow). (B) Immunoblot analyses of proteins extracted from cells co-expressing GFP empty vector, GFP-GA, GFP-GP or GFP-GR and TDP-43 WT (TDP WT). Blots were probed using anti-S409/410 and anti-tubulin antibodies. Phosphorylated full-length TDP-43 (red arrowhead) and its CTF (red arrow) are observed in Sarkosyl-insoluble (Sar-ppt) fraction of cells expressing GFP-GA at Days 2 and 3.
Formation of phosphorylated TDP-43 aggregates depends on the number of poly-GA repeats
The number of C9ORF72 hexanucleotide repeats in the normal population ranges from 2 to 24. To determine the number of poly-GA repeats required for TDP-43 aggregation, we next constructed GFP-tagged expression vectors encoding various numbers of repeats (GA×20, 30, 40 and 50). SH-SY5Y cells were transfected with each vector, incubated for 3 days, and analyzed by immunofluorescence and immunoblotting. In cells expressing GA×0 (GFP alone) or GA×20, GFP signals were diffusely distributed throughout cells, and pS409/410 signals were not found (Fig. 5A). In cells expressing GA×30 or GA×40, some thread-like GFP-positive structures were observed in cytoplasm and were also negative for anti-pS409/410. In cells expressing 50 or 100 GA repeats, we clearly observed phosphorylated TDP-43 aggregates. Immunoblot analyses of cell extracts showed that poly-GA proteins with 30 or fewer repeats were predominantly found as monomers in the Sar-sup fraction, while those with 40 or more repeats were predominantly found in the Sar-ppt fraction, and failed to enter the resolving gel (Fig. 5B). This indicates that the ability of poly-GA proteins to form insoluble inclusions is highly dependent on repeat length. We also found that endogenous TDP-43 is phosphorylated and aggregated in cells expressing poly GA×50 and 100, demonstrating a correlation between extent of formation of phosphorylated TDP-43 aggregates and increased numbers of DPRs in poly-GA proteins. On the other hand, we tested whether TDP-43 directly associates with poly-GA by means of immunoprecipitation assay of cells expressing GFP, GFP-GA×30 or GFP-GA×100 in the presence of TDP-43 WT or ΔNLS. As shown in Supplementary Material, Figure S2, we found that TDP-43 WT and ΔNLS directly interact with soluble GA×30, but not with the aggregated form of GA×100.
Correlation between formation of phosphorylated TDP-43 aggregates and number of repeats of poly-GA. (A) Confocal laser microscopic analyses of cells transfected with GFP empty vector (GA×0), GFP-tagged poly-GA with 20 repeats (GA×20), GFP-GA with 30 repeats (GA×30), GFP-GA with 40 repeats (GA×40), GFP-GA with 50 repeats (GA×50) or GFP-GA with 100 repeats (GA×100). These cells were immunostained with anti-phosphorylated TDP-43 (pS409/410) and counterstained with Hoechst 33342. Scale bars represent 20 μm. Note that endogenous phosphorylated TDP-43 aggregates are observed in cells expressing poly-GA with 100 repeats (white arrowheads). (B) Immunoblot analyses of cells expressing various lengths of poly-GA repeats. Sarkosyl-soluble (Sar-sup) and -insoluble (Sar-ppt) fractions were prepared from cells and subjected to immunoblot analyses. Blots were probed using anti-GFP monoclonal antibody (anti-GFP), anti-phosphorylated TDP-43 (anti-pS409/410) and anti-tubulin antibodies. The band of endogenous phosphorylated TDP-43 is found in Sar-ppt of cells expressing GA×50 and GA×100.
Poly-GA inclusions do not facilitate intracellular accumulation of tau and α-synuclein
Since poly-GA expression induces TDP-43 aggregation, we examined whether it can also induce tau and α-synuclein aggregation in cultured cells. Cells were transfected with plasmids expressing GFP-GA, and either TDP-43, tau (3R1N, 4R1N), or α-synuclein, harvested 3 days later, and analyzed by immunoblotting. As expected, endogenous phosphorylated TDP-43 was detected in the Sar-ppt fraction of all cells expressing GFP-GA, and an increase in phosphorylated TDP-43 was observed in the Sar-ppt fraction of cells co-expressing TDP-43 WT or ΔNLS (Fig. 6A). However, neither phosphorylated α-synuclein (detected by anti-pS129) nor phosphorylated tau 3R1N or 4R1N (detected by anti-pS396) was found in the Sar-ppt fraction (Fig. 6A), suggesting that the ability of poly-GA to induce intracellular aggregate formation is likely to be specific for TDP-43.
Expression of poly-GA specifically facilitates intracellular aggregation of TDP-43. (A) Immunoblot analyses of cells transfected with empty vector (none), TDP-43 (WT or ΔNLS), α-synuclein (αS) or tau (3R1N or 4R1N) in the presence of poly-GA plasmid. Sarkosyl-soluble (Sar-sup) and -insoluble (Sar-ppt) fractions were prepared from cells and subjected to immunoblot analyses. Blots were probed using anti-GFP monoclonal antibody (anti-GFP), anti-phosphorylated TDP-43 (anti-pS409/410), anti-α-synuclein (anti-131–140), anti-phosphorylated α-synuclein (anti-pS129), anti-tau (anti-T46), anti-phosphorylated tau (anti-pS396), and anti-tubulin antibodies. *Non-specific bands. (B) Immunoblot analyses of proteins extracted from cells co-expressing GFP-GA and several deletion mutants of TDP-43. Blots were probed using anti-TDP-43 monoclonal (anti-TDP mono), anti-pS409/410 and anti-tubulin antibodies.
We next used several TDP-43 deletion mutations (23), Δ273–314, a deletion of the Gly-rich domain, Δ106–111, a deletion of RNP-2 of RRM-1, Δ145–152, a deletion of RNP-1 of RRM-1, Δ193–212, a deletion of RNP-2 of RRM-2, and Δ227–234, a deletion of RNP-1 of RRM-2, to map the region required for poly-GA-induced TDP-43 aggregation. Cells were co-transfected with plasmids expressing poly-GA and each mutant, and incubated for 3 days prior to harvesting. As shown in Fig. 6B, the amount of phosphorylated TDP-43 in Sar-ppt lysate fraction from cells expressing TDP-43 Δ193–212 was reduced compared with that in Sar-ppt fraction from cells expressing TDP-43 WT, suggesting that the region around RNP-2 in RRM-2 is important for aggregate formation.
Prion-like seeding activity of insoluble poly-GA and TDP-43 co-aggregates
We examined whether aggregates composed of poly-GA and phosphorylated TDP-43 have a prion-like seeding function. Sarkosyl-insoluble fraction (Sar-ppt) was prepared from cells expressing each poly-DPR and TDP-43 WT (see Fig. 3B). These fractions were used as seeds, and introduced into cells expressing TDP-43 WT or ΔNLS. As shown in Figure 7A, we found that only Sar-ppt seeds from cells expressing GFP-GA and TDP-43 WT (GA + WT seeds) induced aggregation of TDP-43 in cells expressing TDP-43 ΔNLS (red arrowhead in Fig. 7A). Sar-ppt seeds from cells expressing other GFP-DPRs and TDP-43 WT did not induce TDP-43 aggregation.
Insoluble co-aggregates of poly-GA and TDP-43 function as seeds for intracellular TDP-43 aggregation. (A) Immunoblot analyses of cells expressing TDP-43 WT (upper) or ΔNLS (lower) and treated with Sar-ppt seeds prepared from cells co-expressing each poly-DPR and TDP-43 WT. A band of phosphorylated TDP-43 was detected in Sar-ppt of cells expressing TDP-43 ΔNLS and treated with GA + WT seeds (red arrowhead). (B) Confocal microscopic analyses of cells expressing TDP-43 WT or ΔNLS alone, cells treated with GA + WT seeds alone, and cells expressing TDP-43 WT or ΔNLS and treated with GA + WT seeds. Note that phosphorylated TDP-43 aggregates are partially co-localized with GA + WT seeds in cells expressing TDP-43 WT or ΔNLS and treated with GA + WT seeds (white arrowheads).
To characterize TDP-43 aggregates induced from GA + WT seeds, we employed immunofluorescence analyses. Phosphorylated TDP-43 aggregates were not seen in cells expressing TDP-43 WT or ΔNLS in the absence of GA + WT seeds (Fig. 7B). In addition, GA + WT seeds were observed, but no phosphorylated TDP-43 aggregates were seen, in cells treated with GA + WT seeds alone. However, in cells expressing TDP-43 ΔNLS and treated with GA + WT seeds, we detected clear inclusions containing GFP-poly-GA and phosphorylated TDP-43. Furthermore, in cells expressing WT TDP-43 and treated with GA + WT seeds, we also detected aggregates positive for GFP and phosphorylated TDP-43, while in immunoblot analyses of these cells, aggregation of TDP-43 was not observed (upper panel in Fig. 7A). As shown in Figure 7B, the phosphorylated TDP-43 signal in cells expressing WT TDP-43 was weaker than the signal from cells expressing TDP-43 ΔNLS, suggesting that either TDP-43 aggregation occurs preferentially in the cytoplasm, or that the ΔNLS version of TDP-43 is more prone to aggregation. These results also suggest that the lack of signal in immunoblots of the Sar-ppt fraction from cells expressing TDP-43 WT and treated with GA + WT seeds (Fig. 7A) is owing to insufficient sensitivity of our immunoblots. These results indicate that aggregates containing insoluble poly-GA and phosphorylated TDP-43 can serve as seeds to induce further aggregation of TDP-43 in cultured cells.
Discussion
Although hexanucleotide repeat expansion mutations in C9ORF72 gene are the most common genetic cause of FTLD and ALS, the pathogenic mechanisms remain elusive. In general, three major mechanisms have been assumed: (i) formation of nuclear RNA foci containing GGGGCC repeat expansion, (ii) formation of protein inclusions composed of DPR proteins through non-canonical repeat-associated non-ATG translation and (iii) loss-of-function of the C9ORF72 protein. Given that the formation of aberrant protein aggregates is thought to be the major cause of neurodegeneration in most neurodegenerative diseases and that poly-GA inclusions are reported to be the major species of DPR accumulated in brains of patients with C9ORF72 mutations, we focused on the second possibility (ii) and tried to set up in vitro poly-GA expression systems.
Although several groups have already reported that poly-GA DPR proteins aggregate to form Ub and p62-positive inclusions in vitro (15–17), they have not observed intracellular phosphorylated TDP-43 aggregation. In this study, we show that the formation of poly-GA inclusions promotes aggregation of phosphorylated TDP-43 in cultured cells. Furthermore, poly-GA and TDP-43 were occasionally co-aggregated within the same inclusion bodies in cells, in accordance with the pathological observation that poly-GA and TDP-43 accumulate in the same hippocampal neurons in brains of patients with C9ORF72 mutations (22).
Formation of poly-GA inclusions induces apoptosis in cultured cells and primary neurons (14,15,17), and loss of cortical neurons and cerebellar Purkinje cells in brains of mice expressing DPRs (24). On the other hand, Swaminathan et al. observed in the vertebrate zebrafish model that GR DPR was associated with the greatest developmental lethality and morphological defects, and GA, with the least (25). Thus, the mechanism through which DPR proteins cause cytotoxicity has been controversial. Our present finding, that formation of poly-GA inclusions precedes accumulation of intracellular phosphorylated TDP-43, suggests that TDP-43 may mediate the cytotoxicity. Consistent with this idea, accumulation of DPR proteins within nerve cells occurs before TDP-43 accumulation and cell death in FTLD patients with C9ORF72 expansion mutations (26).
We showed here that GFP-GA (GA×100 repeats), but not GFP-GR (GR×100 repeats), induced endogenous phosphorylated TDP-43 aggregation in cultured cells. On the other hand, Saberi et al. recently revealed that poly-GR inclusions were co-localized with phosphorylated TDP-43 in dendrites in brains of patients (27). They did not report the length of GR repeats in the patients, but in general, the repeat length is thought to be 700–1600 units in FTLD/ALS patients (1,2). This suggested that poly-GR with 700–1600 repeat length may induce endogenous phosphorylated TDP-43 aggregation even in cultured cells.
We found that a central region of TDP-43 around the RNP-2 of RRM-2 (residues 193–212) is important for poly-GA-dependent aggregation of TDP-43. In addition, poly-GA causes aggregation of phosphorylated TDP-43, but not phosphorylated tau or α-synuclein. These results are consistent with previous studies that found no association between C9ORF72 repeat expansions and Alzheimer's or Parkinson's disease (28–30), and suggest that poly-GA specifically induces aggregation of phosphorylated TDP-43.
The number of GGGGCC repeats in the normal population ranges from 2 to 24 (1,2,31–33), while the number in FTLD/ALS patients is estimated to be 700–1600 (1,2). However, the minimal repeat size needed to cause FTLD/ALS has not been determined. Immunoblot analyses confirmed that phosphorylated TDP-43 becomes Sarkosyl-insoluble in cells expressing 50 or more GA repeats. Thus, we speculate that at least 50 GA repeats are required to cause poly-GA and TDP-43 inclusion formation and disease onset, but further investigation is necessary to confirm this.
Recent studies have shown that cell-cell transmission of aggregated proteins, including tau, α-synuclein and TDP-43, occurs in both cell culture and animal models (34–47). Therefore, it is thought that prion-like propagation of aberrant protein aggregates may be involved in the pathogenesis of neurodegenerative diseases. In cell culture and animal models, recombinant protein aggregates or detergent-insoluble proteins, prepared from diseased brains, are used as seeds and introduced into cultured cells or animal brains to observe propagation of protein aggregation. In the present study, we observed some phosphorylated TDP-43 aggregates surrounding GFP-GA inclusions (donut type), as shown in Figure 2. We also found that detergent-insoluble fraction prepared from cells containing poly-GA and phosphorylated TDP-43 inclusions could function as seeds for further intracellular TDP-43 aggregation. In inclusions formed from seeds, the GFP-positive seeds tended to be present in the center, surrounded by aggregated phosphorylated TDP-43 most likely generated endogenously by the recipient cell (Fig. 7B). These findings suggest that aggregated poly-GA functions as seeds or scaffolds for intracellular phosphorylated TDP-43 aggregation.
In summary, our results show that poly-GA protein aggregation promotes pathogenic changes of TDP-43, such as mislocalization and intracellular aggregation, leading to the onset of these diseases. Our cellular models should not only contribute to the understanding of pathogenic mechanisms of FTLD and ALS in patients with C9ORF72 mutation, but also may be useful tools for development of novel therapeutic strategies.
Materials and Methods
Antibodies
Monoclonal and polyclonal antibodies against a synthetic phospho-peptide of TDP-43 (anti-pS409/410) (48,49) and monoclonal anti-phosphorylated α-synuclein, pS129 (47,50) were used as described. Polyclonal anti-α-synuclein (anti-131–140) was prepared as described (51). Other antibodies and reagents were commercial products: anti-GFP monoclonal antibody, anti-tubulin-α antibody (Sigma), anti-phosphorylated tau, pS396 (Calbiochem), monoclonal anti-tau antibody T46 (Zymed Laboratories).
Antibodies raised against DPR were prepared as follows: these were raised against poly-(GA)8, poly-(GP)8, poly-(GR)8, poly-(PA)8 and poly-(PR)8 peptides with cysteine at N-terminus, conjugated to m-maleimidobenzoyl-N-hydrosuccinimide ester-activated thyroglobulin. The thyroglobulin–peptide complex (200 μg) emulsified in Freund’s complete adjuvant was injected subcutaneously into a New Zealand White rabbit, followed by 4 weekly injections of peptide complex emulsified in Freund’s incomplete adjuvant, starting after 2 weeks after the first immunization. The specificity of these antibodies for their native peptides was tested by enzyme-linked immunosorbent assay. These were employed at dilutions of 1:1000 for immunoblot analyses.
DNA constructs
Synthetic cDNAs for each DPR protein (poly-GA, PP, GR, PA and PR) with ATG start codon were synthesized (Life Technologies) as described (16) and subcloned into pEGFP-C1 (Clontech). Other vectors such as pcDNA3-TDP-43 WT and ΔNLS (52), pcDNA3-α-synuclein and pcDNA3-tau 3R1N and 4R1N (47) were used as previously reported.
Cell culture and transfection of expression plasmids
Human neuroblastoma SH-SY5Y cells obtained from ATCC were cultured in DMEM/F12 medium (Sigma) supplemented with 10% (v/v) fetal calf serum, penicillin–streptomycin–glutamine (Life Technologies), and MEM Non-Essential Amino Acids Solution (Life Technologies). The cells were maintained at 37°C under a humidified atmosphere of 5% (v/v) CO2 in air. They were grown to 50% confluence in six-well culture dishes for transient expression and then transfected with expression plasmids (usually 1 μg) using FuGENE6 (Promega) or XtreamGENE9 (Roche) according to the manufacturer's instructions. Under our conditions, the efficiency of transfection using pEGFP-C1 vector was 40–50%.
Immunofluorescence analyses
SH-SY5Y cells were grown on coverslips and transfected as described above. After incubation for the indicated times, cells were fixed with 4% paraformaldehyde, blocked with 5% (w/v) BSA/PBS, and stained with primary antibody at 1:1000 dilution at 37°C for 1 h. The cells were washed with 0.05% (v/v) Tween 20/PBS and further incubated with anti-rabbit IgG-conjugated Alexa-647 (1:1000) at 37°C for 1 h. Then the cells were incubated with Hoechst 33342 (Life Technologies) at 1:5000 dilution at 37°C for 15 min to counterstain nuclear DNA. The samples were analyzed using a LSM780 confocal laser microscope (Carl Zeiss).
Fractionation of cellular proteins and immunoblotting
SH-SY5Y cells grown in a six-well plate were transfected with several expression vectors. After incubation for 1–3 days, cells were harvested and lysed in 300 μl of homogenization buffer [10 mM Tris–HCl, pH 7.5 containing 0.8 M NaCl, 1 mM ethyleneglycol bis (β-aminoethyl ether)-N,N,N,N-tetraacetic acid (EGTA), 1 mM DTT and 1% N-lauroylsarcosine sodium salt (Sarkosyl)] by brief sonication. The lysates were centrifuged at 100 000g for 20 min at room temperature. The supernatant was recovered as Sarkosyl (Sar)-soluble fraction (Sar-sup). The pellet was suspended in 100 μl SDS-sample buffer and sonicated. The resulting samples were used as the Sar-insoluble fraction (Sar-ppt). Each sample was separated by SDS-PAGE and immunoblotted with the indicated antibodies, as described (45).
Introduction of protein aggregates as seeds into cultured cells
Cells co-expressing TDP-43 and GFP-tagged poly-DPRs were incubated for 3 days, then harvested, and Sarkosyl-insoluble fraction (Sar-ppt) was prepared as described above. For use as seeds, Sar-ppt was re-suspended in 100 μl PBS and sonicated for 10–20 s (TAITEC VP-050 ultrasonic homogenizer). The resulting suspension (10 μl) was mixed with 120 μl of Opti-MEM (Life Technologies) and 62.5 μl of Multifectam reagent (Promega). After incubation for 30 min at room temperature, 62.5 μl of Opti-MEM was added and incubation was continued for 5 min at room temperature. Then, the mixtures were added to cells expressing TDP-43, and incubation was continued for 6 h in a CO2 incubator. After incubation, the medium was replaced with fresh DMEM/F12 and culture was continued for the indicated period in each case. The cells were prepared for immunofluorescence and/or immunoblotting analyses as described above. Under our conditions, the efficiency of introduction of Sar-ppt seeds was ∼30%.
Supplementary Material
Supplementary Material is available at HMG online.
Acknowledgements
We thank Dr Junjiro Horiuchi for helpful comments and editing of the manuscript.
Conflict of Interest statement. None declared.
Funding
This work was supported by Ministry of Education, Culture, Sports, Science, and Technology Grants-in-Aid for Scientific Research (KAKENHI) Grants JP26117005 (to M.H.), Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (KAKENHI) Grant JP23228004 (to M.H.), a grant-in-aid for research on Brain Mapping by Integrated Neurotechnologies for Disease Studies (Brain/MINDS) from the Japan Agency for Medical Research and Development (AMED) JP14533254 (to M.H.), a grant from the Takeda Science Foundation (to T.N.), and a grant from the Brain Science Foundation (to T.N.).
