Abstract

INTRODUCTION

Amyotrophic lateral sclerosis (ALS) is an adult-onset motor neuron disease (MND) in which the loss of spinal cord motor neurons leads to a relatively rapid onset of paralysis and death. So far the only effective treatment is benzothiazole riluzole, which only modestly extends patient survival (1). A common pathological feature of almost all cases of ALS is the formation of neuronal cytoplasmic inclusions (NCIs) and dystrophic neurites containing the normally nuclear TAR-DNA binding protein of 43 kDa (TDP-43) (2). A second RNA binding protein that forms intraneuronal inclusions and is causative for a subset of ALS is the fused in sarcoma (FUS) protein (3-11). Detailed analysis of the proteins that help form and maintain pathological inclusions aids the understanding of the underlying cellular mechanisms that lead to degeneration of the motor neurons.

The pathologic accumulation of TDP-43 and FUS to NCIs is common to several neurodegenerative disorders. In frontotemporal degeneration (FTD), which shows clinical and pathological overlap with ALS (12, 13), TDP-43 forms NCIs in a number of regions within the brain (2, 14, 15). Another example of pathological inclusions found in ALS and FTD are basophilic inclusions. Basophilic inclusions are negative for TDP-43, but are positive for FUS and a number of other RNA binding proteins (3-5, 9, 16-19). TDP-43 and FUS are proteins involved in a range of RNA biogenesis processes, including the transport of RNA transcripts into cytoplasmic stress granules (20, 21). Stress granules are cytosolic structures that store RNA in a translationally repressed manner upon exposure to stress (22). It has been proposed that the localization of TDP-43 and FUS to cytoplasmic stress granules may lead to the formation of TDP-43 and FUS NCIs in ALS (19, 22, 23). Several reports indicate that some but not all stress granule markers colocalize to TDP-43 NCIs and to basophilic inclusions in spinal cord neurons in ALS (18, 19, 24).

We reported that modulation of stress granule-associated pathways can modulate TDP-43 toxicity in Drosophila, yeast and mammalian cells and that Poly A binding protein 1 (PABP-1) forms NCIs in ALS motor neurons; like ataxin 2, a gene associated with ALS/FTD and a stress granule protein, PABP-1 can modulate TDP-43 toxicity in Drosophila (25, 26). PABP-1 pathology has been noted in TDP-43 NCIs in the motor neurons of ALS patients of undocumented genetic background and in basophilic inclusions in ALS motor neurons (18, 27, 28). Given that we have shown that PABP-1 function is a regulator of TDP-43 toxicity in in vivo model systems, we now translate these studies into the human setting and have thoroughly examined the pathology of PABP-1 in the spinal cord of human ALS. We demonstrate that PABP-1 pathology is a common feature of multiple RNA-binding protein-associated inclusions in various genetic subtypes of ALS patients and, notably, it is twice as prevalent in patients harboring a GGGGCC repeat expansion in C9orf72.

MATERIALS AND METHODS

RESULTS

Study Subjects Clinical Characteristics and Diagnosis

We examined spinal cord tissue from a total of 25 subjects (Table 1). Ten of the patients had no known mutations in the coding regions of TARDBP, UBQLN2, ATXN2 and C9orf72 and all had TDP-43 pathology in the motor neurons of the spinal cord (32, 33). The median age of onset for ALS without an identified mutation was 66.3 years; median disease duration was 4.9 years; 5 were male and 5 were female. Four patients had an intermediate polyglutamine (polyQ) expansion (27-33 CAG repeats) in ATXN2, with median age of onset of 63.5 years and median disease duration of 1 years; 2 were male and 2 were female. Nine patients had a C9orf72 repeat expansion; the median age of onset was 62.9 years and median disease duration was 2.9 years; 7 were male and 2 were female. Two patients had BIBD with ALS; 1 was male and 1 was female; the median age of onset was 70 years; the median disease duration was 5 years.

PABP-1 forms pathological accumulations in ALS motor neurons with TDP-43 pathology To provide detailed insight into the PABP-1 pathology in motor neurons of ALS patients, we determined whether PABP-1 pathology overlapped with pathological TDP-43 inclusions. Serial mirror sections from the spinal cord of 5 ALS patients were stained with either an antibody that detects only the pathological form of TDP-43 (phosphorylated TDP-43 [pTDP-43]) or an antibody to PABP-1. TDP-43 pathology was assessed as pre-inclusion, round-like or skein-like (Fig. 1a, c). Motor neurons with pTDP-43 accumulations were scored for the presence of PABP-1 accumulations on the mirror section. This analysis showed that PABP-1 occurred in motor neurons that contained pTDP-43 inclusions (Fig. 1A-D), with 40% of pTDP-43-containing motor neurons also showing PABP-1 pathology (Table 2).

Open in a separate window

Figure 1

TAR DNA binding protein of 43 kDa (TDP-43) pathology and poly-A binding protein-1 (PABP-1) pathology occur in the same motor neurons. Mirror sections from the spinal cord of amyotrophic lateral sclerosis (ALS) patients were stained for either phosphorylated TDP-43 (pTDP-43) or PABP-1. (A) TDP-43 pathology in ALS (case 3) motor neurons (arrowheads); 1 motor neuron contains a round-like inclusion (arrow). The mirror section was stained for PABP-1 and the same motor neurons were identified (arrowheads). (B) The same motor neuron that contains a round-like pTDP-43 inclusion contains a round-like PABP-1 inclusion (arrow). (C) TDP-43 skein-like accumulations were identified in the motor neurons of the spinal cord of ALS case 5 (arrows). (D) The same motor neurons in mirror sections were immunoreactive for PABP-1 (arrows). (E) Motor neurons of the spinal cord of a control patient without ALS (arrows) show an even cytoplasmic localization pattern for PABP-1, and a general lack of PABP-1 in the nucleus (arrowheads). Scale bar: 30 μm.

Table 2

Quantification of Poly A Binding Protein 1 and TAR DNA Binding Protein of 43 kDa Inclusions in Mirror Sections of ALS Patient Spinal Cord

Case #	Number of MNs with pTDP-43 No. of Inclusions	Number of MNs with pTDP-43 and PABP-1 No. of Inclusions	Percentage of MNs with pTDP-43 and PABP-1 Inclusions
2	7	2	29
3	21	8	38
5	13	2	15
6	24	7	29
12	8	7	88
		Mean	40

Open in a separate window

MNs, motor neurons; PABP-1, poly A binding protein 1; pTDP-43, phosphorylated TAR DNA binding protein of 43 kDa.

PABP-1 Colocalizes to TDP-43 Inclusions in ALS Motor Neurons

To determine whether PABP-1 pathology was present within the pTDP-43 inclusions, spinal cord tissue of 10 ALS patients without an associated mutation (ALS-no mut) were co-stained for PABP-1 and pTDP-43 using double label immunofluorescence. This approach demonstrated that PABP-1 did not colocalize with pre-inclusions of pTDP-43 (Fig. 2A-C), but that PABP-1 colocalized with pTDP-43 skein-like accumulations (Fig. 2D-F), and round-like accumulations (Fig. 2G-I). Not all of the cases examined had pTDP-43 pathology in the sections used for immunofluorescence; those cases were excluded from further analysis (Table 3). Of the ALS-no mut cases in which TDP-43 pathology was observed, PABP-1 was identified in 36% of pTDP-43 inclusions (Fig. 4), a value similar to that of the smaller subset of patients analyzed by serial section immunostaining (Table 2).

Open in a separate window

Figure 2

Poly-A binding protein-1 (PABP-1) colocalizes to mature TAR DNA binding protein of 43 kDa (TDP-43) inclusions but not to pre-inclusions. (A-C) ALS spinal cord tissue was co-labeled for phosphorylated TDP-43 (pTDP-43) and PABP-1. pTDP-43 localized to cytoplasmic pre-inclusions in motor neurons of case 3 (arrow). *Asterisk indicates autofluorescent lipofuscin (A). PABP-1 localization was not enriched in pTDP-43 pre-inclusions (arrow) (B, C). (D-F) pTDP-43 localized to skein-like inclusions (arrow) of case 3 (D). PABP-1 colocalized with pTDP-43 in skein-like inclusions (arrows) (E, F). (G-I) pTDP-43 localized to round-like inclusions in motor neurons (arrow) of case 6 (G). PABP-1 colocalized with pTDP-43 in round-like inclusions (arrows) (H, I). Scale bar: 50 μm.

Open in a separate window

Figure 4

Percentage of poly-A binding protein-1(PABP-1) colocalization with TAR DNA binding protein of 43 kDa (TDP-43) inclusions. Pathological accumulations containing both PABP-1 and pTDP-43 occurred at a frequency of 36% in amyotrophic lateral sclerosis (ALS) patients with no mutation (ALS-no mut), 47% in ATXN2-ALS patients, and 64% in C9-ALS patients. A Fisher exact test was performed and significance after a Bonferroni correction was set to 0.0166667; ns: not significant, *, p < 0.0057. n = total number of inclusions counted across all cases.

Table 3

Quantification of Poly-A Binding Protein-1 Colocalization with TAR DNA Binding Protein of 43 kDa Inclusions

		Pathology of Round Inclusions			Pathology of Skein Inclusions
Case #	Mutation	Total pTDP-43	Number of pTDP-43 & PABP-1	Percentage pTDP-43 & PABP-1	Total pTDP-43	Number of pTDP-43 & PABP-1	Percentage pTDP-43 & PABP-1
2	-	3	2	67	0	0	0
3	-	0	0	0	15	8	53
4	-	0	0	0	1	0	0
5	-	1	0	0	0	0	0
6	-	1	1	100	0	0	0
8	-	1	1	100	9	5	56
9	-	0	0	0	19	0	0
10	-	7	4	57	2	0	0
11	ATXN2 (22/32)	0	0	0	4	1	25
12	ATXN2 (22/27)	0	0	0	1	0	0
13	ATXN2 (22/31)	0	0	0	3	3	100
14	ATXN2 (22/29)	0	0	0	4	3	75
16	C9orf72	2	1	50	0	0	0
17	C9orf72	1	1	100	1	0	0
18	C9orf72	0	0	0	4	2	50
19	C9orf72	2	1	50	0	0	0
20	C9orf72	0	0	0	2	1	50
21	C9orf72	0	0	0	2	2	100
22	C9orf72	0	0	0	2	2	100
23	C9orf72	0	0	0	28	18	64

Open in a separate window

-: no mutation in: TARDBP, UBIQLN2, ATXN2, C9orf72; ATXN2; intermediate polyQ expansion in ataxin 2, pathological repeat length is indicated in brackets; C9orf72: GGGGCC hexanucleotide repeat expansion; PABP-1: poly A binding protein 1; pTDP-43: TAR DNA binding protein of 43 kDa.

We found no evidence of PABP-1 pathology that was independent of phosphorylated TDP-43 inclusions (data not shown), indicating that PABP-1 is a feature of mature inclusions. Additionally PABP-1 cannot bind to the cleaved form of TDP-43 and as a result does not localize to phosphorylated TDP-43 inclusions in the motor cortex, where pathological TDP-43 is mainly cleaved (27). We confirmed the absence of PABP-1 pathology in the motor cortex in 6 cases with phosphorylated TDP-43 inclusions (data not shown). The absence of PABP-1 pathology in the motor cortex suggests that the localization of PABP-1 to phosphorylated TDP-43 inclusions in the motor neurons of the spinal cord is selective and is not the result of a non-specific RNA-binding capacity of PABP-1 to pathogenic inclusions in the CNS.

PABP-1 Colocalization to pTDP-43 Occurs in ALS Patients Harboring an Intermediate PolyQ Repeat Expansion in ATXN2

Intermediate polyQ expansions of 27-33 repeats in ATXN2 are a risk factor for ALS, ATXN2-ALS (26, 34-40); recent data indicate familial association of expansions with disease (41, 42). Additionally, evidence suggests that modulation of TDP-43 toxicity by ataxin 2 is mediated through PABP (25, 26). Therefore to assess whether PABP-1 pathology was also a feature of ATXN2-ALS, we examined PABP-1 colocalization to TDP-43 inclusions in such patients (Tables 1, ​,3).3). Only pTDP-43 skein-like accumulations were observed in tissue from the 5 ALS patient samples examined (Table 3). This is consistent with previous findings suggesting that skein-like inclusions are the predominant pathology in ATXN2-ALS spinal cord motor neurons (43). From the 4 cases, 11 skein-like accumulations were observed in total (Table 3), with PABP-1 colocalizing with 47% of TDP-43 skein-like inclusions (Figs. 3A-F, ​,44).

Open in a separate window

Figure 3

Poly-A binding protein-1 (PABP-1) colocalization with TAR DNA binding protein of 43 kDa (TDP-43) inclusions occurs in the motor neurons of ATXN2-ALS and C9-ALS. (A-F) PABP-1 colocalized with phosphorylated TDP-43 (pTDP-43) in skein-like accumulations in ATXN2-ALS patients (arrows and arrowheads). (G-L) PABP-1 colocalizes with pTDP-43 inclusions in C9-ALS both in skein-like accumulations, arrows (G-I), and in round-like accumulations, arrows (J-L). Scale bars: 20 μm.

PABP-1 Colocalizes to Pathological FUS Inclusions in Motor Neurons

FUS is an RNA binding protein that like TDP-43 forms NCIs in a number of neurodegenerative diseases (3-11). One such disease in which FUS forms NCIs in spinal cord motor neurons is BIBD, a progressive neurodegenerative disease with variable clinical phenotypes ranging from ALS to FTD (3, 11). Similar to TDP-43, FUS localizes to stress granules (27, 45). We examined 2 cases of ALS with BIBD for potential colocalization of FUS and PABP-1 in motor neuron NCIs (Table 1). We found evidence of 2 types of FUS inclusions: round basophilic inclusions (BIs) in case 25 (Fig. 5A) and large NCIs in case 24 (Fig. 5B). In case 25, PABP-1 colocalized with FUS to the BI (Fig 5A-C). Three large NCIs immunoreactive for FUS were also observed but only 1 contained PABP-1. In case 24, only large NCIs were observed; 2 out of 14 co-stained with PABP-1 (Fig. 5F-H). These data suggest that PABP-1 is a protein that is involved in the pathology of both TDP-43 and FUS.

Open in a separate window

Figure 5

Poly-A binding protein-1 (PABP-1) colocalizes with pathological fused in sarcoma (FUS) in spinal cord motor neurons of patients with FUS pathology. (A-C) PABP-1 colocalizes with round-like FUS inclusions in a motor neuron from spinal cord tissue of patient with basophilic inclusion body disease (BIBD), case 25. (D-F) Large round FUS filamentous inclusions were observed in spinal cord motor neurons of BIBD, case 24 (D). The large round FUS inclusions were typically negative for PABP-1 (arrows), although PABP-1 occasionally colocalized with FUS in the large round inclusions in patient with BIBD, case 24. (G-I) Two out of 14 FUS-positive large round inclusions were also positive for PABP-1) in this case. Scale bar: 20 μm.

DISCUSSION

PABP-1 is a marker of stress granules, accumulations that occur in cells that contain proteins and RNAs. Under stress conditions, cells reprogram their transcriptome to generate proteins that can aid in situations of stress. Because proteins associated with ALS are RNA-binding proteins that can localize to stress granules, there is suggestion that the inclusions in ALS are stress granule remnants. Previously, we reported that the stress granule marker PABP-1 is important for mediating the toxicity of TDP-43 in model systems, and we identified PABP-1 inclusions in ALS spinal cord (25). To extend the functional relationship between TDP-43 and PABP-1 further, we examined the relationship between PABP-1 inclusions with phosphorylated TDP-43 in the spinal cord of ALS patients. Our studies indicate that PABP-1 can be found in TDP-43 inclusions in tissue of ALS patients with no known mutations (ALS-no mut), those with an intermediate polyQ repeat expansion in ATXN2 (ATXN2-ALS) and a GGGGCC-hexanucleotide repeat expansion in C9orf72 (C9-ALS), and in patients with FUS pathology. Thus PABP-1 is a protein that pathologically associates with inclusions in ALS of multiple genetic associations.

Interestingly, PABP-1 does not localize to TDP-43 pre-inclusions; rather, PABP-1 colocalizes with mature TDP-43 inclusions, in both skein-like and round-like inclusions of ALS-no mut patients. PABP-1 also colocalized with TDP-43 inclusions in ATXN2-ALS, in which only skein-like inclusions were observed. PABP-1, however, colocalized to both round-like and skein-like inclusions in C9-ALS spinal cord motor neurons. Thus, PABP-1 is a common pathogenic protein that colocalizes to NCIs in a number of different subtypes of ALS. Quantification of PABP-1 colocalization with TDP-43 inclusions indicates that PABP-1 is present in TDP-43 inclusions more frequently in the spinal cord of C9-ALS. C9-ALS patients have an earlier age of onset and a more rapid disease progression but no difference in TDP-43 pathology in spinal cord motor neurons has been observed (32, 44). Our data suggest that there is a difference in NCI pathology in spinal cord motor neurons of C9-ALS patients; intriguingly, this difference does not reside with pathogenic TDP-43, but may reside with the pathological proteins that colocalize to the TDP-43 inclusions.

Modulation of stress granule components modulates TDP-43 toxicity in Drosophila (25). Among these proteins, ataxin 2 and the fly polyA binding protein (PABP) show striking interactions with TDP-43; reduction of PABP mitigates TDP-43 toxicity in the nervous system whereas the expression of human ataxin 2 in the nervous system enhances TDP-43 toxicity. The interaction between ataxin-2 and TDP-43 is dependent upon the PABP binding domain of ataxin 2 (25). In human spinal cord motor neurons, PABP-1 may act to promote formation of TDP-43 inclusions, stabilize the TDP-43 inclusions or be involved in the maturation of TDP-43 inclusions. Decreasing the levels of PABP-1 may reduce the toxic effect of TDP-43 NCIs in spinal cord motor neurons by slowing the rate of formation, stabilization or maturation of pathogenic NCI. Alternatively, reduced PABP-1 levels may prevent PABP-1-associated proteins, such as ataxin 2, from colocalizing to TDP-43 inclusions, thereby reducing the overall toxicity that the NCIs exert on the spinal cord motor neurons (25, 26).

Taken together, these studies highlight that combining functional studies in model systems with neuropathological data can give important insight into pathological mechanisms that may be perturbed in patients. Additionally, given the mechanistic overlap between TDP-43 and another causative gene for ALS, FUS, and the finding reported here that PABP-1 localizes to FUS inclusions in the motor neurons of the spinal cord tissue of 2 cases of BIBD with ALS underscore the broad relevance of interaction data from model systems to guide neuropathological studies. Our findings highlight the pathological association of PABP-1 in ALS to association in cases with a range of genetic backgrounds and in situations of FUS pathology.

ACKNOWLEDGMENTS

REFERENCES