Dipeptide repeat proteins are present in the p62 positive inclusions in patients with frontotemporal lobar degeneration and motor neurone disease associated with expansions in C9ORF72

Frontotemporal Lobar Degeneration (FTLD) is a clinical, pathological and genetically heterogeneous condition. The major clinical syndromes principally involve personality and behavioural change (behavioural variant frontotemporal dementia, or bvFTD) or language alterations of a fluent (semantic dementia) or non-fluent (progressive non-fluent aphasia) nature [1]. All three syndromes can be accompanied by Motor Neurone Disease (MND) though the combination of FTD and MND is most common. Histologically, around half of cases have tau-based pathology, half have TDP-43-based pathology, and about 5% have FUS-based pathology [2, 3]. Importantly, around 40% of all cases have a strong family history of similar disease, irrespective of clinical or histological subtype [1].

By 2007, causative mutations had been identified in tau (MAPT) [4] and progranulin (GRN) [5, 6]. Nonetheless, it was well known at that time that a number of large, independent FTD + MND kindreds demonstrated linkage to chromosome 9 in the region 9p21.2-p13.3 [7‐10]. Subsequently, three GWAS studies for MND [11‐13], and one for FTLD [14] identified a susceptibility locus within this linked region, with strongest association coming from a 80 kb haplotype block containing 3 genes, MOBKL2B, IFNK and C9ORF72. It has now been shown that this at least some of this association is due to the presence of a large hexanucleotide (GGGGCC) in C9ORF72 gene in patients with both FTLD and MND [15, 16]. The expansion occurs in the first intron or the promoter region of the gene, depending upon the transcript isoform in question, and can number up to as many as 1500 repeats. The expansion is found in about one in every twelve patients with FTLD, but this varies depending on geographical region with the expansion being rare in Asia [16].

Pathologically, most FTLD cases with the expansion [15‐20], like many non-mutational cases of FTLD [2, 21, 22], show inclusion bodies within neurones (NCI) and glial cells of the cerebral cortex and hippocampus that contain the nuclear transcription factor, TDP-43, and are said bear a TDP-43 histological subtype termed FTLD-TDP type B (according to [23]), compatible with a clinical diagnosis of FTD and MND. Others, however, show a TDP-43 histological type characterised by the presence of many short neurites (DN) along with NCI within the outer layers of the cerebral cortex, and termed FTLD-TDP type A [23]. Nonetheless, irrespective of TDP-43 histological type, expansion carriers also show a unique pathology within the hippocampus [17] and cerebellum [17‐19, 24] characterised by NCI that are TDP-43 negative, but immunoreactive to p62 protein.

P62 is a marker for proteins destined for proteasomal degradation though the precise target protein within such NCI remains uncertain, and the pathogenetic mechanism underlying the hexanucleotide expansion remains unclear. Nonetheless, this is likely to result from one, or a combination, of three possible mechanisms: i) haploinsufficiency leading to loss of C9orf72 protein [15, 16], ii) the expansion might form nuclear foci of toxic RNA and sequester other RNA-binding proteins such as Pur Alpha (Pur α) [25], or iii) RAN (repeat associated non ATG-initiated) translation of the expanded repeat region may lead to the 'inappropriate’ formation of dipeptide repeat proteins (DPR) which may aggregate and confer neurotoxicity [26, 27].

Here, we show that DPR are at least one of the target protein(s) within the TDP-43 negative, p62-positive NCI in cases of FTLD associated with hexanucleotide (GGGGCC) expansions, and that such peptides are not associated with other histological forms, or genetic subtypes, of FTLD.

Screening the 84 FTLD and 23 MND cases with p62 revealed 16 cases, 13 with FTLD (cases #1-13) and 3 with MND (cases #14-16) (Table 1) showing p62 positive, TDP-43 negative NCI within granule cells of the cerebellum and other cerebellar cell types, and in granule cells of the dentate gyrus, and pyramidal cells of CA4, CA3 and CA2 regions of the hippocampus, as described previously [17‐19, 24]. Twelve of the 13 FTLD cases showed TDP-43 proteinopathy, classifiable [23] as either type A (6 cases) or type B (6 cases): the other case had tauopathy compatible with corticobasal degeneration (CBD) (Table 1). Eight of the FTLD cases had presented with bvFTD clinically, 4 with FTD + MND and one (with CBD pathology) with a combination of FTD and corticobasal syndrome. All MND cases bore typical TDP-43 pathological changes in motor neurones of brain stem nuclei and spinal cord (where this was available for analysis).

Frozen tissue was available for 12 of the 16 cases showing p62 pathological changes in cerebellum and hippocampus. These included 9 cases with FTLD (cases #1-3, 7, 9-13) and all 3 cases with MND (cases #14-16) (Table 1). All 12 of these cases demonstrated a pathological expansion in C9ORF72 by Southern blotting. Expansion size ranged from ~5 kb (~450 repeats) to in excess of 23 kb (over 3600 repeats) (Figure 1). No expansions were detected in cases where no p62 pathological changes were observed. There was no correlation between repeat size and age of disease onset or duration (see Additional file 1: Figure S1).

In the present study we have observed p62 positive, TDP-43 negative NCI within cerebellar granule cells, and granule cells of the dentate gyrus and pyramidal cells of the hippocampus (see [17‐19, 24]) in 13 of 84 (15%) cases of FTLD and in 3 of 23 (13%) cases of MND. Interestingly, all cases with expansion showed a FTD, FTD + MND or MND clinical phenotype, and 15/16 bore appropriate TDP-43 protein pathological changes. However, one expansion carrier with clinical FTD (case #13 in Table 1) was of FTLD-tau with CBD pathology. This case was described by us previously [20] and is important as it is only the second case of CBD to be described with expansion in C9ORF72 (see [32] for details of the other case), though there was no post-mortem confirmation of CBD in this latter instance. The present case demonstrated an expansion in C9ORF72 both on repeat-primed PCR (not shown), and on Southern blot (see lane #8 in Figure 1). Although the case showed no TDP-43 pathology, DPR were present in cells of the cerebellum and hippocampus (see Table 2), though curiously those NCI in granule cells of the cerebellum were p62-immunopositive, while those in hippocampus dentate gyrus and CA4 regions were not positive for p62.

None of the other cases with other histological or genetic forms of FTLD showed p62 positive, TDP-43 negative NCI within either cerebellar granule cells or within the hippocampus. The pathological findings described here are therefore broadly consistent with those previously reported by others in unselected series of cases of FTLD and/or MND [17‐19, 24].

All of the 12 p62 positive cases where frozen brain tissues were available for analysis showed an expansion in C9ORF72 by Southern blotting, and are therefore consistent with the suggestion [17, 18] that this type of p62 pathology is pathognomic for cases of FTLD and/or MND associated with C9ORF72 mutation. Thus, it can be presumed that an expansion was also present in the other four FTLD cases with relevant p62 pathology where no southern blotting (or repeat primed PCR) was possible due to lack of fresh frozen brain tissue. The lack of association between expansion length and age of disease onset or duration presumably reflects that a minimum size of expansion is required for disease. Present data suggests this might be around 500 repeats, and expansions beyond this are not additionally detrimental.

The exact target protein within the p62 immunoreactive NCI remains uncertain. Here, we performed immunohistochemistry employing antibodies to DPR putatively produced in the brain through non-ATG initiated (RAN) translation of the expansion itself. Although double immunofluorescence labelling was not performed in the present study, microscopic observations on consecutive serial sections stained with each DPR antibody suggested that most, if not all, of the p62-positive NCI within Purkinje and dentate nucleus cells of the cerebellum, and neurones of CA2, CA3 and CA4 of the hippocampus are similarly immunoreactive to poly-GA, poly-GP and poly-GR antibodies (and to a much lesser extent, poly-AP antibody), whereas only a (variable) subset of p62-positive NCI within granule cells of the cerebellum and dentate gyrus appeared immunoreactive to these antibodies. These microscopic impressions were supported by statistical analysis showing good correlations between ratings of inclusion body frequencies as assessed using poly-GA, poly-GP and poly-GR antibodies. Other studies [26, 27] where double immunofluorescence labelling was indeed performed substantiate the present observations.

None of the NCI was immunoreactive to poly-PR antibody, though this antibody did immunolabel chromatin granules, especially in Purkinje cells and CA4 neurones, in some C9ORF72 expansion bearers. Although this kind of immunostaining was not seen in any of the control subjects in such cells, it was not consistently present in all expansion bearers, and so the relevance of this (to pathogenesis) remains uncertain. None of the DPR antibodies were immunoreactive to any other kinds of inclusions (TDP or tau) associated with other histological or genetic forms of FTLD, or the NII containing CAG repeats seen in HD. These findings are consistent with those recently reported by Mori and colleagues [27]. The pattern of DPR immunolabelling in the cerebellum and hippocampus in C9ORF72 expansion bearers mirrored that reported by Ash et al. [26], and this appeared identical to that shown by our own poly GP antibody, and that prepared by M Hasegawa. C9RANT antibody was raised to a panel of dipeptide repeat immunogens [26], but judging from the similarities between its pattern of immunolabelling and that of our own poly-GP antibody, it might be considered that its specificity, or at least, it avidity is greatest for the poly-GP component of the immunogen mix. The finding of strong and consistent staining with poly-GA, poly-GP and poly-GR antibodies suggests that most of the aberrant translation relates to sense transcripts, though the slight, but variable immunostaining with poly-AP antibody implies some antisense translation might also occur. However, it remains possible that the relative paucity of immunostaining seen here with antibodies to antisense transcripts reflects poor antigen avidity on the part of the antibody rather than indicating an absence of the relevant polypeptide with DPR inclusions per se.

Nevertheless, it appears that DPR may not be the only proteins present within these structures since the p62 positive inclusions have also been shown to contain hnRNPA3 [33], and others reported the inclusions in granule cells of the dentate gyrus and cerebellum (at least) to be immunoreactive to ubiquilin-2 [34]. There is little or no DPR immunostaining of TDP-43 in hippocampus or cerebral cortex in either expansion bearers, or patients with other histological forms of FTLD [29], consistent with those observations of NCI based on p62 immunostaining [17‐19, 24]. Furthermore there was no immunostaining of NII containing expanded poly-Q stretches in patients with HD (see also [27]). Collectively, these findings reinforce the suggestion that p62-, DPR-positive NCI are pathognomic for FTLD (and MND) associated with C9ORF72 expansions.

It remains uncertain as to how these NCI containing DPR may relate to disease pathogenesis and other fundamental aspects of disease pathology such as TDP-43 positive NCI and neurites. While NCI containing DPR are located in clinically and pathologically relevant regions, such as the hippocampus and adjacent temporal cortex, they are actually present in cells distant from the major TDP-43 changes (layers V and VI in cerebral cortex and areas CA2/3/4 in hippocampus), although it is acknowledged that cells in the dentate gyrus of the hippocampus may contain one or other, but rarely both, types of inclusion [19, 27]. Those cells containing DPR did not show any other outward signs of neurodegeneration, whether these being small granule cells of dentate gyrus or cerebellum, or larger pyramidal cells of hippocampal CA regions, cerebral cortex or Purkinje neurones of the cerebellum. Likewise, there appeared to be no obvious loss (thinning out) of granule cells in either region or depletion of cells from hippocampal pyramidal or cerebellar Purkinje cell layers. Moreover, such changes in the cerebellum appear to be without functional (clinical) consequence, although there have been reports of cerebellar atrophy in expansion bearers [35, 36].

Interestingly, there appeared to be no qualitative or quantitative differences between either the number, or the pattern, of DPR stained NCI between FTLD-TDP type A and type B cases, in either hippocampus or cerebellum. This observation begs the question as to the role of C9ORF72 expansions in determining histological phenotype and their relationship to TDP-43 pathology. It is possible that variability in repeat size may have a role in this, but we were unable to show any obvious differences in repeat size between Type A and type B cases by Southern blot. Alternatively, it is possible that the expansion confers an 'additive’ effect, through induction of a process marked by p62 pathology, which is superimposed upon a 'background TDP-43 proteinopathy derived through a similar mechanism to that seen in non-expansion bearers sharing the same TDP-43 histological phenotype. If that were so, then it might be presumed that patients bearing the mutation suffer a 'double whammy’, and that the expansion plays no real part in determining the TDP-43 proteinopathy and the basic underlying disorder. However, if this 'chance scenario’ were true, then it might be anticipated that C9ORF72 repeat expansions might commonly occur in association with other disorders, though there is little, and conflicting, evidence for this in, for example, Alzheimer’s disease [37‐43] or Parkinson’s disease and other parkinsonian syndromes such as Corticobasal syndrome [44‐47]. Indeed, it is possible that in at least some of these instances where an expansion has been reported in patients with conditions other than FTLD or MND, the underlying condition may still be FTLD, though presenting in an atypical way [39].