The neuropathology associated with repeat expansions in the C9ORF72 gene
- First Online:
- Cite this article as:
- Mackenzie, I.R.A., Frick, P. & Neumann, M. Acta Neuropathol (2014) 127: 347. doi:10.1007/s00401-013-1232-4
- 1.7k Views
An abnormal expansion of a GGGGCC hexanucleotide repeat in a non-coding region of the chromosome 9 open reading frame 72 gene (C9ORF72) is the most common genetic abnormality in familial and sporadic FTLD and ALS and the cause in most families where both, FTLD and ALS, are inherited. Pathologically, C9ORF72 expansion cases show a combination of FTLD-TDP and classical ALS with abnormal accumulation of TDP-43 into neuronal and oligodendroglial inclusions consistently seen in the frontal and temporal cortex, hippocampus and pyramidal motor system. In addition, a highly specific feature in C9ORF72 expansion cases is the presence of ubiquitin and p62 positive, but TDP-43 negative neuronal cytoplasmic and intranuclear inclusions. These TDP-43 negative inclusions contain dipeptide-repeat (DPR) proteins generated by unconventional repeat-associated translation of C9ORF72 transcripts with the expanded repeats and are most abundant in the cerebellum, hippocampus and all neocortex regions. Another consistent pathological feature associated with the production of C9ORF72 transcripts with expanded repeats is the formation of nuclear RNA foci that are frequently observed in the frontal cortex, hippocampus and cerebellum. Here, we summarize the complexity and heterogeneity of the neuropathology associated with the C9ORF72 expansion. We discuss implications of the data to the current classification of FTLD and critically review current insights from clinico-pathological correlative studies regarding the fundamental questions as to what processes are required and sufficient to trigger neurodegeneration in C9ORF72 disease pathogenesis.
Frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS) are related clinical syndromes with overlapping molecular pathogenesis. FTLD is the second most common cause of dementia in the presenile age group (<65 years) and accounts for 5–15 % of all dementias. It is characterized by progressive deterioration in behaviour, personality and/or language, with relative preservation of memory, due to a predominant atrophy of the frontal and temporal lobes. A family history of dementia is present in 25–50 % of cases, often with an autosomal dominant pattern of inheritance. ALS is the most common form of motor neuron disease in which a predominant loss of motor neurons from the brain and spinal cord leads to fatal paralysis and death, usually within 1–5 years. Most cases are sporadic; however, ~5 % are familial. A clinical overlap between ALS and behavioural variant frontotemporal dementia (bvFTD) has long been recognized with up to 50 % of ALS patients showing some executive function deficits [32, 57] and up to 15 % of bvFTD patients presenting with motor neuron dysfunction [12, 31].
The molecular basis for this clinical overlap was provided by the discoveries of common neuropathological and genetic abnormalities in ALS and FTLD strongly suggesting overlapping disease mechanisms. In 2006, the DNA/RNA binding protein TDP-43 was identified as the accumulating disease protein in the majority of ALS cases and to be the most common pathological subtype of FTLD , now referred to as FTLD with TDP-43 pathology (FTLD-TDP) . The spectrum of TDP-43 proteinopathies was characterized to include sporadic and genetic cases with mutations in the genes GRN, VCP, TARDBP as well as families with a genetic linkage to a region on chromosome 9p21 [13, 36, 50, 51]. The identity of the chromosome 9 gene remained elusive until 2011, when two groups discovered the defect as being an abnormal expansion of a GGGGCC (G4C2) hexanucleotide repeat in a non-coding region of the chromosome 9 open reading frame 72 gene (C9ORF72) [19, 56]. This mutation is now known to be the most common genetic abnormality in familial and sporadic FTLD and ALS and the cause in most families where both, FTLD and ALS, are inherited [24, 55].
C9orf72 is an uncharacterized protein and the underlying mechanisms of how the intronic repeat expansion mutation contributes to neurodegeneration is unclear. Based on knowledge from other repeat expansion diseases such as spinocerebellar ataxia type 8 and myotonic dystrophy type 1, three potential pathogenic mechanism are suggested: (1) haploinsufficiency (2) toxic RNA gain of function by sequestration of RNA binding proteins to transcripts with expanded repeats and (3) toxicity of products generated by unconventional repeat-associated non-ATG (RAN) translation of transcripts with expanded repeats [6, 15, 65].
The aim of this review was to summarize the neuropathological features associated with the C9ORF72 expansion and to discuss insights from postmortem studies in dissecting the role of different processes in C9ORF72 disease pathogenesis.
Molecular pathology in C9ORF72 expansion cases
TDP-43 pathology in C9ORF72 expansion cases
TDP-43 is a highly conserved, predominantly nuclear DNA/RNA binding protein that can shuttle between the nucleus and cytoplasm with several well-described functions in RNA regulation including control of splicing, mRNA transport and stability, although its function in the brain is not well understood [11, 17, 29]. All TDP-43 proteinopathies, including those with C9ORF72 expansions, are associated with characteristic pathological features of TDP-43. TDP-43 inclusion body formation is consistently associated with a substantial reduction in the normal physiological diffuse nuclear staining. Moreover, pathological forms of TDP-43 show evidence of abnormal processing, being hyperphosphorylated, ubiquitinated and N-terminally truncated (Fig. 1l) [13, 50, 52]. These findings implicate potential roles for both, loss-of-function and gain-of-function mechanisms of TDP-43 associated cell death. However, the molecular mechanisms underlying TDP-43 proteinopathies are complex and currently not completely understood, as discussed in detail elsewhere [17, 25, 29, 39].
The TDP-43 pathology in C9ORF72 expansion cases consists of compact or granular neuronal cytoplasmic inclusions (NCI), diffuse neuronal cytoplasmic staining (“pre-inclusions”), dystrophic neurites (DN), (oligodendro)glial cytoplasmic inclusions (GCI) and variable presence of neuronal intranuclear inclusions (NII) (Fig. 1a–k). Besides the consistently affected frontal and temporal cortex, hippocampus and pyramidal motor system, a wide range of other neuroanatomical regions such as amygdala, striatum, thalamus, substantia nigra, midbrain tectum and tegmentum and inferior olives often show TDP-43 pathology [26, 49]. While there is overlap in the neuroanatomical distribution of TDP-43 pathology among C9ORF72 expansion cases with different clinical phenotypes (FTLD, FTLD/ALS or ALS), the extent of TDP-43 pathology differs significantly in key anatomical regions among cases with distinct clinical phenotypes and most importantly shows strong correlations with the degree of degenerative changes in the respective CNS regions [26, 41].
Specifically, C9ORF72 expansion cases with clinical ALS show pathology indistinguishable from typical sporadic ALS, with predominant degeneration and TDP-43-positive NCI and pre-inclusions of variable morphology in upper motor neurons (Fig. 1d) and lower brainstem and spinal cord motor neurons (Fig. 1e, f), often together with variable amounts of GCI pathology in these regions (Fig. 1g). The severity of degeneration and TDP-43 pathology in the pyramidal motor system is significantly higher in cases with clinical ALS compared with those with FTLD. The extramotor cortex, hippocampus and subcortical regions are usually mildly or not affected [41, 49, 61].
C9ORF72 expansion carriers with a clinical FTLD or FTLD/ALS phenotype show a combination of FTLD-TDP and ALS-TDP with more severe degeneration and amounts of TDP-43 pathology in the frontal and temporal neocortex compared with those with pure ALS [26, 41, 49]. The subcortical regions consistently show moderate to abundant TDP-43 pathology in FTLD and FTLD/ALS. Cases with clinically pure FTLD show significantly less degeneration and TDP-43 pathology in lower motor neurons compared with those with FTLD/ALS or pure ALS. However, even in the absence of clinical motor neuron dysfunction at least some TDP-43 positive NCI are usually present.
The pattern of neocortical TDP-43 in the majority of C9ORF72 cases with sufficient cortical pathology is consistent with FTLD-TDP type B pathology with compact NCI in all cortical layers and relatively few DN and NII (Fig. 1a). In addition, neurons with diffuse granular cytoplasmic TDP-43 reactivity (“pre-inclusions”) (Fig. 1b) [9, 13, 26, 41, 61]. This is in good accordance with the fact that among all FTLD-TDP cases (genetic and non-genetic), type B is the pathological subtype most common in patients with a mixed phenotype of bvFTD and ALS [34, 40]. However, some heterogeneity exists with a significant number of C9ORF72 cases showing concomitant or pure type A pathology that is characterized by compact small NCI and numerous DN in the superficial neocortical layers as well as rare lentiform NII (Fig. 1c) [41, 43, 49, 60]. Among all FTLD-TDP cases, this pattern is commonly associated with bvFTD and non-fluent aphasia and seen in all familial cases with GRN mutations [13, 34, 35]. Moreover, single cases with C9ORF72 expansions are also reported with FTLD-TDP subtype C pathology [27, 49].
The underlying cause and significance of this heterogeneity in the associated pattern of FTLD-TDP remains to be determined. No association was found between C9ORF72 repeat length and TDP-43 pattern , suggesting that the pathological heterogeneity is not a reflection of some length variation in the expansion. Type A pathology seems to be more common in those with clinically pure FTLD, while those with FTLD and ALS usually have type B pathology [26, 43, 49]. Notably, a strong association between the presence of dual type A and B pathology and advanced age and longer disease duration was observed . Together with the fact that concomitant TDP-43 pathology in Alzheimer’s disease and many other common neurodegenerative conditions usually shows type A pathology [3, 64], and an association with the less common allele of the rs5848 polymorphism in the GRN gene , it is suggestive that type B might be the primary pathology associated with the C9ORF72 expansion and that additional type A pathology develops in a genetically susceptible subset of patients with advancing age .
A final consideration raised by the reported heterogeneity in FTLD-TDP patterns is that in some cases with the C9ORF72 expansion it may be difficult to accurately subtype the FTLD-TDP pathology based on the current criteria. Following the discovery of TDP-43 as the pathological protein in most FTLD-U cases , several studies suggested that the same criteria originally developed to subtype FTLD-U pathology using ubiquitin-immunohistochemistry [34, 58] could be directly applied to TDP-43 immunohistochemistry without significant modification [13, 18]. While this may be true for most cases of FTLD-TDP, those with type B pathology, particularly when associated with the C9ORF72 expansion, present a problem. With ubiquitin immunohistochemistry, type B cases are characterized by the presence of numerous NCI in all cortical layers with a relative paucity of DN [34, 40, 58]. However, we now know that many of the NCI in cases with the C9ORF72 expansion are composed of DPR proteins (see below) and these are not reactive for TDP-43. Moreover, the numerous delicate neurites and “pre-inclusions” concentrated in cortical layer 2 demonstrated by TDP-43 immunohistochemistry in many cases of type B, including those with the C9ORF72 expansion, are not seen with ubiquitin immunohistochemistry [13, 18]. Hence, some cases with the C9ORF72 expansion and type B ubiquitin-immunoreactive pathology could easily be mistaken for type A with TDP-43 immunohistochemistry.
In summary, the consistent presence of TDP-43 pathology in C9ORF72 expansion carriers and the strong association of TDP-43 burden and distribution with clinical phenotypes and degeneration provide strong evidence for a central role of TDP-43 dysmetabolism and accumulation in the neurodegenerative process.
Dipeptide-repeat protein pathology in C9ORF72 expansion cases
In addition to TDP-43 aggregation, a unique and highly characteristic feature of patients with expansions in C9ORF72 is the presence of NCI in the cerebellar granule cell layer, hippocampal pyramidal neurons and neocortex that stain positive for proteins of the ubiquitin proteasome system (UPS), such as ubiquitin, ubiquilins and p62, but are negative for TDP-43 [1, 2, 10, 16, 26, 42, 49, 54, 61, 63].
Dipeptide-repeat pathology is restricted to neurons as demonstrated by double-label immunofluorescence [4, 41] and consists of NCI, NII and DN. The most detailed neuropathological data of DPR pathology in a wide spectrum of neuroanatomical regions are available with antibodies against poly-GA , poly-GP  and poly-GR  which show a similar and homogeneous neuroanatomical distribution of DPR pathology in C9ORF72 expansion cases, irrespective of the clinical phenotype.
Specifically, in the neocortex, numerous NCI are present in layers II–VI ranging from granular dot-like in small non-pyramidal neurons to ring and star-shaped inclusions in pyramidal neurons. The extent of DPR-positive inclusions is similar in all neocortical regions, including those that are vulnerable in ALS and FTLD, such as frontal cortex (Fig. 2c) and primary motor cortex, as well as usually non-affected regions such as occipital cortex (Fig. 2d). Small, delicate DN are consistently seen, however, with a more variable extent. In the hippocampus, NII and star-shaped NCI are always abundant in the dentate granule cells (Fig. 2e) and in the pyramidal layer, primarily the CA3/4 region (Fig. 2f). Massive DPR pathology is present in the cerebellum with small granular, dot-like NCI and NII in the granule cells (Fig. 2g) and molecular layer, occasional NCI in Purkinje cells and numerous DN in the molecular layer. NCI and DN are less abundant in subcortical regions such as striatum and thalamus and absent or extremely rare with only single NCI per section in the motor nuclei of the brainstem and spinal cord (Fig. 2h). Sometimes neurons with diffuse cytoplasmic staining (“pre-inclusions”) can be seen.
While there is no doubt that DPR pathology is a highly specific pathological feature in C9ORF72 expansion carriers, its pathomechanistic relevance is less clear. The highly consistent pattern of DPR pathology among cases, regardless of the clinical phenotype, and the lack of association of DPR pathology with degenerative changes are not supportive of a direct causative role of aggregated DPR proteins in C9ORF72-mediated neurodegeneration . The observed dissociation between DPR inclusion body load and neurodegeneration might rather be interpreted as inclusion body formation being a potentially protective response to cope with soluble toxic protein species, as also discussed in other neurodegenerative diseases such as Huntington’s disease  or that the production of DPR proteins is not a relevant factor in C9ORF72 disease pathogenesis.
RNA foci pathology in C9ORF72 expansion cases
Another mechanism proposed in repeat expansions disorders (e.g. fragile X-associated tremor ataxia syndrome and myotonic dystrophy) is RNA toxicity caused by aggregation of transcripts with the expanded repeat into nuclear RNA foci and subsequent sequestration of RNA binding proteins, thereby interfering with their physiological function and resulting in major RNA processing alterations . The presence of RNA foci in C9ORF72 expansion cases was first reported by DeJesus-Hernandez et al.  in neurons of the frontal cortex and spinal cord in human brain tissue. While one study was unable to replicate this finding , improved FISH protocols have now confirmed the presence of RNA foci composed of both, sense and antisense transcripts, as a consistent feature in brain tissue [22, 28, 44] as well as in lymphoblasts, fibroblasts and induced pluripotent stem cell-differentiated neurons from C9ORF72 expansion cases [2, 21, 28].
Despite the abundance of RNA foci formation in C9ORF72 expansion cases, their pathomechanistic relevance remains unclear due to the fact that clinico-pathological correlation studies are limited. One study on eight cases with clinical FTLD bearing expansions in C9ORF72 noted a significant inverse correlation between sense RNA foci burden in neurons in frontal cortex with age at onset and age at death . However, this needs to be confirmed in larger series. In addition, no correlation between RNA foci burden and neurodegeneration has been reported so far, and detailed studies comparing the distribution pattern of RNA foci pathology in cases with different clinical phenotypes are required.
Moreover, the identities of sequestered and perhaps pathomechanistically relevant proteins in RNA foci in human brains are currently unknown. A growing list of potential candidates such as hnRNP A3 , hnRNP-H , Pur-alpha  and ADARB2  have been reported from in vitro binding studies and await now further characterization.
Alterations of C9orf72 expression
Evidence for a potential role of haploinsufficiency in disease pathogenesis is provided by the finding of reduced RNA levels of at least some, and possibly all, of the three major C9ORF72 transcripts in blood, lymphoblasts and fibroblasts, as well as postmortem brain tissue in C9ORF72 expansion carriers [5, 19, 24, 56] and supported by the fact that downregulation of the zebrafish orthologue of C9ORF72 leads to altered morphology of motor neuron axons and locomotor deficits . Reduced C9ORF72 transcription is triggered by epigenetic changes including abnormal binding of trimethylated histones (H3K9, H3K27, H3K79, H4K20) to the expanded repeats . Aberrant CpG methylation near the expanded repeat was reported in one study with an inverse correlation of the degree of methylation with disease onset , but this was not detectable in a second study .
However, the current lack of specific antibodies against the functionally uncharacterized C9orf72 is a severe limitation that has prevented reliable studies on C9orf72 protein expression in humans. Thus, no reliable information is available on the cell types expressing C9orf72, their neuroanatomical distribution, the physiological cellular distribution of C9orf72 as well as potential alterations in the disease process (reduced expression, abnormal distribution or accumulation) and their association with TDP-43, DPR and RNA foci pathology.
Insights into disease mechanisms from postmortem studies
Importantly, the only neuropathological change with strong clinico-pathological correlations and association with neurodegeneration in C9ORF72 expansion cases is TDP-43 pathology. This strongly implies that alterations due to expansions in C9ORF72 resulting in TDP-43 accumulation and dysmetabolism as a secondary downstream effect likely plays a central role in the neurodegenerative process in C9ORF72 pathogenesis. However, the link between C9ORF72 expansion and TDP-43 aggregation remains unclear. TDP-43 pathology is only rarely associated with DPR pathology in the same neuron [41, 47] and the neuroanatomic distribution of TDP-43 and DPR pathology only partially overlaps; therefore, a mechanism where DPR protein formation triggers TDP-43 accumulation seems unlikely. Likewise, the poor association of TDP-43 pathology with RNA foci formation is also not supportive of a strong direct link between these processes .
Implications of complex neuropathology associated with C9ORF72 expansions to current FTLD classification
The unique features and heterogeneity of the neuropathology in FTLD cases with the C9ORF72 expansion may complicate diagnosis and challenges certain aspects of the existing classification system for FTLD. According to the current recommendations for the nomenclature and nosology of neuropathologic subtypes of FTLD, cases are grouped into broad categories based on the identifiable protein abnormality that is most characteristic and presumed to be most closely related to pathogenesis [37, 38]. As a result, the vast majority of FTLD cases can be assigned to one of three major molecular groups, FTLD-TDP, FTLD-tau or FTLD-FET/FUS [38, 55]. The recognition and characterization of TDP-43 and DPR protein aggregates in cases with the C9ORF72 expansion raises questions about the most appropriate classification. One possibility that has been suggested is to reclassify these cases as FTLD-DPR . This would appropriately recognize the DPR pathology as a more specific feature and would also be supported by reports of rare cases of C9ORF72 expansion carriers found to have DPR inclusions without significant TDP-43 pathology [43, 47, 60]. However, this would also imply that the DPR pathology is more closely related to disease pathogenesis; a position that is contradicted by the clinicopathological study describing that the anatomical pattern of neurodegeneration and the phenotype correlated strongly with TDP-43 pathology but not at all with DPR pathology . Moreover, creation of this new category would result in the confusing situation where cases with identical patterns of TDP-43 pathology would receive different classification, depending on the presence or absence of the mutation (FTLD-DPR or FTLD-TDP, respectively). Therefore, until the relative importance of TDP-43 and DPR pathologies are determined, it would seem prudent to continue to include all cases with significant TDP-43 pathology as FTLD-TDP, but recognize that those with the C9ORF72 expansion have an additional pathological feature that is of diagnostic utility but of uncertain pathogenic significance.
As discussed above in detail (see TDP-43 pathology in C9ORF72 expansion cases), the variability in FTLD-TDP subtypes being associated with the C9ORF72 expansion requires further attention and highlights a more general need to re-evaluate the criteria used to subtype FTLD-TDP in light of recent advances in the molecular neuropathology of FTLD.
Finally, a number of studies have reported C9ORF72 expansion cases with significant tau-based pathology, usually in combination with FTLD-TDP [7, 26]. The morphology, abundance and anatomical distribution of the tau-immunoreactive inclusions exceed what might be expected for age and resembles either Alzheimer’s disease or FTLD-tau [7, 26]. Moreover, a single C9ORF72 expansion case has been reported to be associated with FTLD-tau and pathological features of corticobasal degeneration in the absence of significant TDP-43 pathology . While the link between C9ORF72 expansions and tau pathology in a subset of at-risk patients requires further investigation, the co-existence of two major classes of FTLD-related molecular neuropathology may additionally complicate diagnosis and classification.
Thus, specific issues that require further study and that may need to be addressed in future revisions to the nomenclature include (1) appropriate classification for conditions with more than one protein abnormality as a characteristic feature, (2) re-evaluation of criteria for distinguishing FTLD-TDP subtypes based on TDP-43, (3) possible co-occurrence of two FTLD-TDP subtypes in a single case and (4) possible co-occurrence of two major molecular classes of FTLD in a single case.
The C9ORF72 expansion is associated with a complex and multifaceted neuropathology. Clinico-pathological correlative studies have provided important insights into potentially relevant disease mechanisms; however, one also has to be aware of the limitations of studies investigating patients with end-stage disease. Thus, formulated hypotheses based on neuropathological features now need to be addressed in cellular and animals models to further elucidate the relevance of the distinct processes in C9ORF72 disease pathogenesis.
This work was supported by grants from the German Helmholtz Association (VH-VI-510, and W2/3 program, MN), the German Federal Ministry of Education and Research (01GI0704, MN), the Canadian Institutes of Health Research (74580, IRM), the Pacific Alzheimer’s Research Foundation (C06-01, IRM).