Fused in sarcoma (FUS) mutations on chromosome 16 have in recent years been recognised as a relatively rare, yet important cause of amyotrophic lateral sclerosis (ALS). In all, such mutations represent approximately 4 % of familial ALS cases (fALS) and less than 1 % of sporadic cases. The FUS protein is thought to have, amongst other functions, RNA binding properties similar to TDP-43, which itself is associated with the majority of ALS cases (both sporadic and familial) [19]. When investigated neuropathologically cases of ALS-FUS have been shown to exhibit FUS immunopositive inclusions in the anterior horn neurons of the spinal cord, the twelfth cranial nerve nucleus of the medulla, and/or within the neurons of the motor cortex [8–12]. The distribution and shape of the neuronal cytoplasmic inclusions (NCIs) have some similarities to those seen in ALS associated with deposition of TDP-43 (ALS-TDP). These NCIs can be filamentous-like and more globular, sometimes they are also associated with glial cytoplasmic inclusions (GCIs) and/or neurites. ALS associated with TDP-43 positive inclusions (ALS-TDP) and frontotemporal lobar degeneration with TDP-43 positive inclusions (FTLD-TDP) (a pathological process associated with frontotemporal dementia, semantic dementia and/or progressive non fluent aphasia) are believed to represent a clinicopathological spectrum, where some patients have clinical and pathological features of a frontotemporal dementia and ALS (FTLD-MND/ALS). ALS-FUS, however, appears to be distinct both clinically and pathologically from the frontotemporal lobar degeneration with FUS positive inclusions (FTLD-FUS) [13-15, 20]. Many cases of ALS-TDP, even when not associated with cognitive decline, still show TDP-43 immunopositive inclusions in the limbic regions such as hippocampus, amygdala or entorhinal cortex. Our study confirmed previous reports in that, apart from some inclusions in the basal ganglia and cerebellum, there were no cases with extramotor FUS positive inclusions, further confirming the distinction between ALS-FUS and FTLD-FUS. Indeed FTLD-FUS is not usually associated with mutations. Instead FTLD-FUS, unlike ALS-FUS, appears to exhibit inclusions that contain a number of so-called FET proteins including FUS, Ewing’s sarcoma protein (EWS) and TATA-binding protein associated factor 15 (TAF15) as well as Transportin 1 implying abnormalities in nuclear-cytoplasmic transport [13, 15, 21]. Additionally, in ALS-FUS mutations the arginine methylation status of FUS appears to differ from that in FTLD-FUS [14, 20, 22, 23]. Indeed the ALS-FUS mutations may instead result in the production of a more rigid protein which in turn appears to affect its RNA binding properties [19, 24].

Combined with previous work, this study suggests that in the early stages of ALS-FUS most FUS mutations preferentially exert their effects on the spinal motor neurons [11]. Whilst there have been studies attempting to correlate particular FUS mutations with clinical progression, there have been relatively few descriptions of pathology in ALS-FUS. Waibel et al. found that cases of ALS-FUS with truncating mutations were associated with a more aggressive phenotype than missense mutations [16]. This cannot be confirmed or refuted in this series because only one truncating mutation case is present, however, even 2 of the missense mutations had rapid disease progressions of 10 months or less. One pathological feature of particular interest is that all cases showed only mild to very mild loss of myelin in the lateral corticospinal tracts. This was in contrast to 4 of the cases demonstrating at least some UMN features clinically and one case (case 6) showing evidence of marked microglial activity in the motor cortex. Previous studies have also demonstrated this discrepancy; the clinical presentation in ALS-FUS is often of early proximal limb weakness sometimes with pelvic or scapular weakness, mild or late-onset bulbar features and no significant cognitive impairment and mild or no extrapyramidal abnormalities [10, 25]. Despite this, in Mackenzie et al.’s series the paucity of UMN features did not appear to correlate with the relatively severe FUS pathology in the motor cortex. Similarly Baumer et al. illustrated 2 cases with the p.P525L mutation, one of whom had no UMN features clinically yet marked degeneration of the corticospinal tracts [8]. This was further illustrated in a study by Riku et al. who attempted to distinguish the clinicopathological features of the lower motor syndrome of progressive muscular atrophy (PMA) with typical clinical ALS [26]. They found (albeit with relatively small numbers) that whilst cases of ALS-FUS accounted for 15 % of the PMA cases, all of the typical clinical ALS cases showed TDP-43 pathology. Nevertheless, in these 15 % (which were represented by 2 cases) there were still FUS positive inclusions seen in the motor cortex. In their series of 6 cases (4 with FALS) Hewitt et al. described minimal FUS pathology in the motor cortex [9]. However, there have been reported exceptions with one patient suffering from a locked-in state associated with a p.K510M mutation and abundant FUS pathology in the UMNs [27]. Tateishi et al. described a multiple system degeneration associated again with marked UMN pathology but this time in patients with a p.R521C mutation which is identical to our cases 3 and 4 and thus illustrates the degree of phenotypical variability [28]. Mouse models with FUS mutations tend to support the hypothesis that it is the LMNs which are particularly vulnerable to the structural changes that these mutations cause [29]. All of our cases exhibited some basophilic inclusions in the anterior horn neurons of the cord. Mackenzie et al. attempted to define specific groups of ALS-FUS by pathological patterns including the density of basophilic inclusions [10]. Those with numerous basophilic inclusions (including 2 cases with p.P525L mutations) tended to have a more rapid disease progression and appeared to have fewer GCIs, compared to cases with fewer basophilic inclusions and more glial inclusions who had a later onset, and slower progression (including 2 cases with p.R521C mutations). Our findings do not fit neatly into this distinct grouping. Whereas case 5 with a p.R521H mutation would certainly appear to partially conform to the above criteria with a disease progression of at least 36 months, and associated with few basophilic inclusions in the cord, there were, however, moderate numbers of GCIs. Similarly case 7 had few basophilic inclusions in the cord, and also a relatively slow progression but again moderate numbers of GCIs. An explanation for this discrepancy might be due to a much younger-aged cohort in Mackenzie et al.’s group, or alternatively in our group the reduced numbers of basophilic neurons may just have been a reflection of the severe neuronal loss present in the cord. Case 6 (with a p.P525L mutation) also had moderate-large numbers of glial inclusions, yet also had moderate numbers of basophilic inclusions but a rapid disease progression. However, there is an additional element in case 6 which may have affected this, in that this patient as well as having a p.P525L mutation also had a truncating p.Y374X TARDBP mutation on chromosome 1. The pathology in this case 6 was very unusual. Whilst there was moderate FUS pathology in the form of NCIs and GCIs in the spinal cord, the twelfth nerve nucleus, and mild pathology for FUS in the motor cortex, there was marked CD68 positivity in the motor cortex indicating increased microglial activity and associated with neuronal loss in this region. There was no TDP-43 pathology seen, but there was extensive p62 pathology, in the form of granular and synaptic-like positivity in the cytoplasm and cell membrane of neurons and glial cells together with extensive dot-like and neurite-like neuropil positivity. Furthermore, this was not only seen in the cord but also even more extensively in the motor cortex and neocortex where there was a more ramifying pattern to the staining in the neurons . This FUS mutation has been described previously but not with this associated p62 pathology [8, 10]. Daoud et al. have described the genetics of the p.Y374X TARDBP mutation and have predicted this to be a damaging variant, but as far as is known there have been no pathological descriptions [18]. P62 also known as sequestosome 1 is a ubiquitin-binding protein, important in protein degradation via the ubiquitin-proteasome system. As such it often can be demonstrated in association with, and sometimes co-localised with, abnormal proteins in specific neurodegenerative diseases such as tangles in Alzheimer’s disease and Lewy bodies in Parkinson’s disease. It is also seen in association with TDP-43 in FTLD-TDP and ALS associated with TDP-43. It is intriguing that despite the TARDBP mutation in case 6 there was no TDP-43 pathology. Whereas this patient died at the age of 23, the only known reported patient with the same TARDBP mutation died at the age of 63 so it may simply be that the FUS mutation is the more “aggressive” mutation and they act independently such that we are seeing the end stage of the FUS-ALS disease and the earliest (if any) stage of the ALS-TDP disease [18]. This, however, does not adequately explain the unusual p62 expression. Daoud et al. predicted that the actual TARDBP mutation produces a truncated TDP-43 protein due to a premature stop codon removing the last 41 amino acids [18]. It would therefore be logical that any inclusions could not be detected with the relatively specific antibody to pTDP-43. What is more difficult to explain is that the antibodies to the wild type (non-phosphorylated) TDP-43 also showed no inclusions and indeed no loss of normal nuclear TDP-43 immunopositivity. Homma et al. have described a missense TARDBP mutation with minimal TDP-43 pathology [30], and it may be that the TDP-43 pathology in our case is also so subtle it has been missed. It is also possible that the mutation resulted in unusual species of TDP-43 not detectable by conventional antibodies. Although not directly comparable with a truncating mutation it is interesting that mouse models that have undergone conditional gene depletion of TARDBP in the spinal cord neurons do not show TDP-43 positive inclusions (unlike missense mutation models) but do show an unusual pattern of polyubiquitinated proteins in the neurons but not in the form of the usual inclusions [31–33]. Therefore in this case the p62 expression may be showing 2 distinct patterns, one simply shadowing the FUS pathology and one corresponding to an unusual pattern of polyubiquitinated proteins similar to the TARDBP depletion model described. Whilst the p62 pathology may indicate damage or dysfunction in individual neurons or glial cells the relationship cannot be that straightforward since the intensity of such p62 immunopositivity in the neocortex does not appear to correlate with cognitive decline. In this regard it can perhaps be compared to the extensive p62 and ubiquitin immunopositivity seen in the cerebellum in cases of FTLD or ALS with the C9ORF72 repeat expansion. The immunoexpression here again does not appear to be related to neuronal loss or cerebellar signs, however, there is still a debate as to whether there is a toxic effect of the offending ubiquitinated dipeptide repeat proteins [34–36]. There also remains the theoretical possibility that there is a third mutation present (in the p62/sequestosome 1 (SQSTM1) gene) in this case which has given rise to the p62 pathology [37].

As far as is known this is the first pathological description of a patient with ALS-FUS associated with a p.K510E mutation. Syriani et al. described the clinical and genetic features of 2 cases with this mutation, one was 32 years old at death the other 43, and both had a rapid clinical course (8 and 10 months respectively), and neither demonstrated UMN signs [38]. Our case (case 2) did demonstrate some UMN and bulbar signs and a moderately rapid disease progression of 15 months. Pathologically what was interesting was the moderate to severe FUS involvement of the cord and medulla but again not the motor cortex. Also there were moderately numerous FUS positive NCIs and GCIs in the midbrain and basal ganglia.

The p.R514G missense mutation again has been detected previously and has been predicted to have a relatively mild phenotype [16], which is in keeping with our case 7, again to our knowledge the pathology associated with this mutation has not been reported. There were few basophilic inclusions and moderate numbers of FUS positive GCIs.