Keywords

Introduction

Frontotemporal lobar degeneration (FTLD) is a collective term for a disease group characterized by progressive neurodegeneration limited to frontal and temporal lobes. FTLD is clinically divided into three types: frontotemporal dementia (FTD), semantic dementia (SD) and progressive nonfluent aphasia (PNFA) (Neary et al. 1998). This classification is based on clinical manifestations that reflect differences in the degenerative brain region. They do not reflect specific neuropathological characteristics. FTLD can be subdivided into three neuropathological groups, depending on the presence of inclusion bodies or a certain protein component (McKhann et al. 2001). The first group, exhibiting “tauopathy”, has tau-positive inclusion bodies. This group includes Pick’s disease, corticobasal degeneration (CBD), progressive supranuclear palsy (PSP) and frontotemporal dementia with parkinsonism linked to chromosome 17 (FTDP-17). The second group might be called FTLD-U since it is similar to FTLD but has ubiquitin-positive tau-negative neurocytoplasmic inclusions (Mackenzie et al. 2006a). FTLD is divided into two types, FTLD with motor neuron disease: (FTLD-MND) and FTLD with MND type inclusions but without MND. The third group consist of FTLD without tau- or ubiquitin-positive inclusions, and this group has been considered a dementia lacking distinctive histology (DLDH). However, most cases of the third group consist of FTLD-U with inclusions which are identified using high-sensitivity ubiquitin immunostaining. Rare cases with tau-negative, cytoplasmic and nuclear ubiquitin-positive inclusions have also been found (Mackenzie et al. 2006c).

Some 35–50% of FTLD patients have a family history of dementia and the causative gene loci have been identified on chromosomes 3, 9 and 17. Microtubule-associated protein tau (tau, MAPT), valosin-containing protein (VCP) and charged multivesicular body protein 2b (CHMP2B) have been identified as causative of FTLD. The identification of the tau gene mutation on chromosome 17q21 reminds us of the importance of tau in neurodegenerative disease research (Hutton et al. 1998). However, a considerable number of familial FTLD-U cases linked on chromosome 17q21 with tau-negative, cytoplasmic and nuclear ubiquitin-positive inclusions have been found.

In 2006, Cruts and Baker identified a granulin (GRN) mutation in FTLD-U patients (Baker et al. 2006; Cruts et al. 2006). Since then, more than 190 GRN mutations have been reported including substitutions, insertions and deletions (Tables 1, 2 and 3 and Alzheimer Disease & Frontotemporal Dementia Mutation Database, http://www.molgen.ua.ac.be/FTDMutations/) (Cruts et al. 2012). The common pathological mechanism in these mutations was proposed to arise from haploinsufficiency. Symptoms of haploinsufficiency appear after inactivation of one allele of the causative gene in a dominantly-inherited disease (Wilkie 1994). With GRN mutation, a mutated form of mRNA is degraded by nonsense-mediated decay (NMD) which is likely to create a null (no expression) allele. It is thought that the functional form of the PGRN protein decreases with disease onset.

Table 1 GRN mutations (pathogenic)
Full size table
Table 2 GRN mutations (pathogenic nature unknown)
Full size table
Table 3 GRN mutations (not pathogenic)
Full size table

A TAR-DNA binding protein of 43 kDa (TDP-43), the main component of ubiquitin-positive inclusions, was observed in FTLD and ALS patients and was identified in 2006 by Arai et al. and Neumann et al. (Arai et al. 2006; Neumann et al. 2006). The tau-negative, ubiquitin-positive inclusions that were seen in GRN mutation brains were also identified as containing TDP-43.

GRN Mutations and Pathological Mechanisms

Baker and colleagues examined more than 80 candidate genes within the 3.53-cM (6.19-Mb) critical region clarified by haplotype analysis of Canadian tau-negative FTD families (Baker et al. 2006). They identified an insertion mutation of four base pairs (CTGC) in exon 1 of the GRN gene (g.90_91insCTGC) [g: genomic DNA, ins: insertion]. The numbering is relative to the reverse complement of GenBank accession number AC003043.1, starting at adenine (A) of Met 1. This mutation causes a frame shift at codon 31 that induces a premature termination codon after a read through of 34 amino acids (p.Cys31LeufsX34) [p: protein, fs: frame shifts, X: termination codon]. The p.Cys31LeufsX34 mutation was absent in 550 North American control individuals. They sequenced the GRN gene in affected families in Canada, the USA, UK, Netherlands and Scandinavia and identified an additional eight GRN mutations in nine families. These mutations were as follows: four nonsense mutations: g.1087C>T (p.Gln125X), g.2609G>A (p.Trp386X), g.2923C>T (p.Arg418X), g.3073C>T (p.Gln468X); two flame shift mutations: g.1102_1105delCAGT (p.Gln130SerfsX124), g.2597delC (pThr382SerfsX29); one splicing site mutation: IVS8-1G (p.Val279GlyfsX4); and mutation in start codon: g.2T>C (p.Met1?). Next, they extracted RNA from the lymphoblasts of cases with mutations g.90_91insCTGC (p.Cys31LeufsX34) and c.2923C>T (p.Arg418X), performed quantitative RT-PCR analysis and found that the expression of GRN mRNA was reduced by approximately 50%. They performed sequencing of GRN mRNAs and found that most of them encoded wild type GRN, whereas the mutated type of GRN was rarely detected. These results suggested that the mutated mRNA was degraded by NMD. NMD degrades mRNA with a premature termination codon (PTC) which arise from a splicing error or mutation, and thereby prevent production of an abnormal protein (Maquat 2004).

When the lymphoblasts from patients were treated with the NMD inhibitor, cycloheximide, the mutated mRNA was increased. Immunoblotting analysis revealed that the amount of wild type PGRN protein had decreased compared with the controls and mutated PGRN protein was barely detected. They also detected a significant reduction in the amount of mutated mRNA in the brains of patients with the g.2T>C mutation. They suggested that translation of the protein did not occur because the Kozak sequence was disrupted by the g.2T>C mutation.

Cruts and colleagues also identified five novel GRN mutations, IVS0+5G>C (now termed IVS1+5G>C), g.3G>A (p.Met1?), g.1094_1095delCT (p.Pro127ArgfsX2), g.1872G>A (p.Ala237TrpfsX4), and g.1087C>T (p.Gln125X). IVS1+5G>C indicates a point mutation in the intron 1 splice donor site causing intron 1 retention, resulting in nuclear mRNA degradation (Cruts et al. 2006). Sequence analysis of GRN in 103 Belgian FTD patients identified this mutation in the eight probands belonging to different branches of the Belgian founder family. An in silico analysis of the IVS1+5G>C mutation predicted an intense decline in the binding efficiency of the U1 snRNP complex.

Next, they analyzed full length GRN cDNA from the brains and lymphoblasts of two probands, abnormal transcripts. According to the polymorphism (rs5848) in the 3′ untranslated region of the GRN gene, probands were judged C/T heterogeneous (the T-allele is the disease haplotype). However, on sequence analysis of cDNA from their lymphoblasts or brain tissue, only the C allele was observed. These results suggested a complete disappearance of mutated GRN mRNA. Immunoblot analysis using an extract from the lymphoblasts of a proband showed PGRN protein reduction. They confirmed loss of mRNA and wild type PGRN protein reduction in the cases of the g.1087C>T (p.Gln125X) mutation. Subsequently, Gass and colleagues performed systematic screening for the GRN gene in 378 FTLD and 48 ALS cases at the Mayo Clinic and identified 23 GRN mutations in 39 FTLD cases.

Twenty of these twenty-three mutations (4 nonsense mutations, 12 frame shift mutations and 4 splicing donor site mutations) predicted production of PTC and mutated mRNA degradation by NMD. They also identified novel mutations in the splicing donor site of exon 1 (IVS1+1G>A) as well as a missense mutation (g.26C>A (p.Ala9Asp)). In this study, no mutation was identified in ALS cases. RT-PCR analysis of a brain with an IVS1+1G>A mutation revealed two bands corresponding to mutated GRN mRNA and wild type GRN mRNA, respectively. These results suggested that the IVS1+1G>A mutation did not cause degradation of mutated mRNA by NMD. Initiation of NMD first required a translation process, so that it has been speculated that any IVS1+G>A mutation would escape NMD because no translation would start without the Kozak sequence. The g.26C>A (p.Ala9Asp) mutation was identified as singular missense mutation in this study, the 9th alanine in exon 1 of GRN being replaced by aspartic acid. The 9th alanine corresponds to the hydrophobic core of the signal peptide. Mutated mRNA was reduced in the g.26C>A (p.Ala9Asp) brain by an unknown mechanism. If a mutated allele was translated in this case, it would produce a mutated PGRN protein lacking binding capability to the signal recognition motif and could not be transported to the endoplasmic reticulum. Since 2006, many novel GRN mutations have been found and are listed in Tables 1, 2 and 3.

PGRN Protein Is Not a Component of Ubiquitin-Positive Inclusion Bodies

Immunohistochemical staining using antibodies for all regions of PGRN protein showed that some of the neurons and activated microglia were positive. Ubiquitin-positive neuronal cytoplasmic inclusions (NCI) and neuronal intranuclear inclusions (NII) were negative with PGRN antibodies (Baker et al. 2006; Cruts et al. 2006). These results indicated that PGRN accumulation did not occur during development of the FTD pathology caused by the GRN mutation. PGRN-positive neuron and activated microglia were also observed in the brains of normal elderly individuals and Alzheimer’s disease (AD) cases.

Clinico-Pathological Characterization of GRN Mutation Carriers

Incidence Rate

In the Belgian study, Cruts and colleagues found GRN mutations in 10.7% (11 out of 103) of the FTD cases overall and in 25.6% (11 out of 43) of familial FTD cases (Cruts et al. 2006). MAPT mutation frequencies were 2.9% (3 out of 103) in the non-familial FTD and 7% (3 out of 43) in the familial FTD cases. These results indicated that GRN mutations are approximately a 3.5 times more frequent cause of FTD in Belgian patients. GRN mutation data of Gass and colleagues showed mutations in 10.5% (39 out of 378) of FTD and 25.6% (32 out of 144) of familial FTD cases. However, they pointed out that there was some bias in their cases because the Mayo Clinic treated many familial FTLD patients or FTLD patients with a definitive pathological diagnosis.

To exclude this kind of clinical bias, 167 non-selective FTLD cases were collected between 1990 and 2006 in five different Alzheimer’s disease research centers and analyzed. The frequency of the GRN mutation was 48%. It was noted that the frequencies of the GRN and MAPT mutations were almost the same; the frequency of the MAPT mutation was 44% in the same series of brains. Further investigation of this similarity will be needed. Of 649 dementia cases collected in Minnesota between 1987 and 2006 as part of a dementia research project, 15 were diagnosed with FTLD. Three patients were identified with the GRN mutation. The frequency of the GRN mutation in the dementia patients overall was calculated to be 0.5%.

Pickering-Brown and colleagues reported that the frequency of the GRN mutation was 7.3% (14 out of 192 FTLD patients) (Pickering-Brown et al. 2008) whereas Le Ber and colleagues reported the frequency to be 6.4% (32 out of 502 FTD patients) (Le Ber et al. 2008). The frequency of GRN mutations in probands was 5.7% (20 of 352) in fvFTD, 4.4% (3 of 68) in primary progressive aphasia (PPA) and 3.3% (1 of 30) in corticobasal syndrome (CBS). The authors also mentioned that no mutations were found in the 52 probands with FTD-MND. Yu et al. found the frequency of the GRN mutation to be 6.9% (30 of 434) (Yu et al. 2010).

Age of Onset

The age of onset of FTLD in Belgian patients with the IVS1+5G>C mutation was 45–70 years (average 63.4 ± 6.8) (Cruts et al. 2006). This mutation was identified in a few asymptomatic individuals; one who had died at 41 years of age, two who had died within the normal age of onset at ages (44 and 54 years) and the one who died at 81. Gass et al. found that the age of onset was 48–83 years (average 59.0 ± 7.0) among GRN mutation carriers over all (Gass et al. 2006). Other studies demonstrated that the average age of FTLD onset was 59.0 ± 5 (Pickering-Brown et al. 2008); 59.4 ± 9.4 in FTD, 62.0 ± 7.9 in FTD-MND, 63.8 ± 8.5 in PPA and 61.8 ± 9.7 in CBS (Le Ber et al. 2008). In another study, the average age of onset was 57.7 years, which was calculated from the onset age of 31 GRN mutation-positive patients from 28 different families (Yu et al. 2010). Leverenz et al. investigated two families with the GRN c.709-2A>G mutation (now termed g.1871A>G (p.Ala237TrpfsX6)) (Leverenz et al. 2007). In family 1, the mean age of onset was 55.6 ± 8.9 years (range = 35–69), the mean age at death was 65.5 ± 6.8 years (range = 56–78) and the mean duration was 9.8 ± 5.5 years (range = 4–22). In family 2, the mean age of onset was 61.0 ± 6.6 years (range = 50–67), the mean age at death was 68.6 ± 6.0 years (range = 57–73) and the mean duration was 6.8 ± 0.4 years (range = 6–7) (Leverenz et al. 2007).

Clinicopathological Images of FTLD

Patients with the GRN IVS1+5G>C mutation show non-fluent aphasia (Cruts et al. 2006). Gass et al. indicated that FTLD patients with the GRN mutation often exhibited dysphasia and this was rarely accompanied by motor neuron dysfunction (Gass et al. 2006). Pathological analysis of GRN IVS1+5G>C patients revealed the presence of neuronal cytoplasmic inclusions (NCIs). Neuronal intranuclear inclusions (NIIs) were also observed in all cases. These observations corresponded with previous reports in which NIIs were commonly detected in familial FTLD patients without motor neuron dysfunction (Mackenzie and Feldman 2003; Woulfe et al. 2001). These results suggested that NIIs would be a pathological marker of PGRN mutation cases. However, NIIs were also found in sporadic FTLD cases or FTLD patients with motor neuron dysfunction, indicating that more investigation will be needed (Mackenzie et al. 2006a). Investigating the clinical response to the GRN mutation, Gass et al. found that the most common diagnosis was FTD followed by PA. Other diagnoses were CBD, AD with convulsions and motor dysfunction (PD, parkinsonism and FTD-MND) (Gass et al. 2006). Snowden and colleagues reported that in a single pedigree of the g.3073C>T (p.Gln468X) mutation, patients showed symptoms of FTD and PA (Snowden et al. 2006). Masellis et al. reported that a patient with the GRN IVS7+1G>A (p.Val200GlyfsX18) mutation exhibited CBD-like symptoms (Masellis et al. 2006). Pickering-Brown et al. reported that in patients with the GRN mutation, 57% were diagnosed as FTD, 36% as PNFA and 7% as apraxia and parkinsonism (Pickering-Brown et al. 2008). Le Ber et al. reported that 63% of patients with the GRN mutation were diagnosed as fvFTD with other clinical patterns being PPA (16%), CBS (6%) and Lewy body disease (LBD) (6%) (Le Ber et al. 2008). They also found that 9% of patients had other diagnoses including AD and parkinsonism (Le Ber et al. 2008). The most common diagnosis was FTD including PPA and CBS. Other clinical phenotypes such as AD, AD+PD and LBD were observed (Yu et al. 2010).

Immunohistochemical analyses for phosphorylated TDP-43 revealed a considerable number of neuronal cytoplasmic inclusions and dystrophic neurites in GRN mutation cases (Fig. 1). In FTLD-TDP, TDP-43 pathology falls within four histological subtypes (types A-D) based on the predominant type of TDP-43-positive structures exhibited (Mackenzie et al. 2011). Type A is characterized by numerous short dystrophic neurites (DNs) and crescentic or oval shaped neuronal cytoplasmic inclusions (NCIs). Cases of FTLD-TDP with a GRN mutation invariably display type A pathology (Cairns et al. 2007; Josephs et al. 2007; Mackenzie et al. 2006b).

Fig. 1
figure 1

Immunohistochemical staining of the temporal lobe of a GRN mutation case with antibody to phosphorylated TDP-43. Numerous neuronal cytoplasmic inclusions (arrows) and dystrophic neurites (arrowheads) were stained with anti-TDP-43-pS409/410 antibody and the section was counterstained with hematoxylin. Scale bar = 100 μm

Full size image

PGRN Protein Levels in GRN Mutation Carriers

Plasma PGRN protein levels were measured in FTLD patients with the g.1975_1978delCTCA (p.Leu271LeufsX10) mutation or the g.2473 C>T (p.Gln341Arg) mutation and in unaffected individuals with the g.1975_1978delCTCA (p.Leu271LeufsX10) mutation, and in all cases were found to have significantly reduced expression of PGRN (Ghidoni et al. 2008). Plasma PGRN was proposed as a useful biomarker. Sleegers et al. reported that serum PGRN levels were reduced in both affected and unaffected carriers of the PGRN null mutation (IVS1+5G>C) compared with their noncarrier relatives (Sleegers et al. 2009). The authors also measured serum PGRN levels in carriers of the g.1129T>C (p.Cys139Arg) and g.3542C>T (p.Arg564Cys) mutations, and found them to be significantly lower than in controls, but greater than in null mutation carriers. They concluded that the serum PGRN level is a reliable biomarker for diagnosis of FTLD caused by a PGRN null mutation (Sleegers et al. 2009).

Plasma PGRN levels were measured in PGRN loss-of-function mutation carriers, FTLD patients without GRN mutations or symptomatic/asymptomatic GRN mutation carriers (Finch et al. 2009). Pathogenic GRN loss-of-function mutations such as g.26C>A (p.Ala9Asp), g.1098_1101delTAGT (p.Gln130SfsX125), g.2273_2274insTG (p.Trp304LeufsX58fs), g.2450delC (p.Gly333ValfsX28), g.3240C>T (p.Arg493X) and g.3175A>G (p.Ala472_Gln584del) resulted in significantly reduced plasma PGRN levels. Missense mutations (g.2422G>A (p.Ala324Thr)), g.2968C>T (p.Arg433Trp), g.3012C>T (p.His447His) and g.3586G>A (p.Pro578Pro) were associated with plasma PGRN levels equal to those of the controls, but g.55C>T (p.Arg19Trp) and g.1129T>C (p.Cys139Arg) cases showed plasma PGRN levels below the level of detection in controls. These results suggested that g.55C>T (p.Arg19Trp) and g.1129T>C (p.Cys139Arg) mutations might induce a partial loss of PGRN function (Finch et al. 2009). Plasma PGRN levels were also lower than those in carriers of the PGRN g.1A>G (p.Met1), g.1129T>C (p.Cys139Arg), p.Ala89ValfsX41 and p.Ala303AlafsX57 mutations (Gomez-Tortosa et al. 2013).

Mean plasma PGRN levels within the FTLD group were significantly lower in patients with GRN mutations than in those with C9ORF72 expansions, or those without mutation (Gibbons et al. 2015). Meeter and colleagues recently reported that PGRN levels in the plasma and CSF of patients with a loss-of-function GRN mutation (g.366delC (p.Ser82ValfsX174), g.1087C>T (p.Gln125X), g.1102_1105delCAGT (p.Gln130SerfsX125) and g.2902_2903delGT (p.Val411SerfsX2)) and presymptomatic loss-of-function GRN mutation carriers were lower than those of healthy controls (Meeter et al. 2016).

It has been reported that the homozygous carriers of the T-allele of rs5848 have an elevated risk of developing FTD. TT genotype carriers had lower serum PGRN levels than CT or CC carriers (Hsiung et al. 2011). The rs5848 T-allele is known to be a miRNA-659 binding site and rs5848 may enhance translational inhibition of GRN and alter the risk of FTD and other dementias (Hsiung et al. 2011).

The Effect of GRN Mutation and Its Influence on PGRN

The effects of GRN mutation and its influence on PGRN function are as follows.

  1. 1.

    Mutations that introduce a premature termination codon (PTC) induce nonsense-mediated mRNA decay machinery.

  2. 2.

    Mutations in the intron 1 splice-donor site such as IVS1+3A>T and IVS1+5G>C may generate intron 1 read-through mRNA. Such aberrant mRNAs may not be capable of normal transport through the nuclear pore complex, so that they may remain in the nuclear area where they are then liable to be degraded by the nuclear mRNA degradation system.

  3. 3.

    Complete gene deletion such as found in delGRN (Gijselinck et al. 2008) or g.-95_3490del in French patients (Rovelet-Lecrux et al. 2008) may lead to no PGRN at all.

  4. 4.

    Missense mutations in the signal peptide may induce mislocalization of PGRN and insufficient translocation to the endoplasmic reticulum (ER).

  5. 5.

    Missense mutations in other areas may also cause problems. If the mutations exist in the consensus sequence of PGRN, they may be pathological because aberrant protein folding may occur in the ER and reduce PGRN secretion to the extracellular lumen. However, the pathological nature of almost all of them is unknown. The other missense mutations are considered to be benign.

GRN Mutation: Multiple Proteinopathy?

GRN mutations were initially found in tau-negative patients (Baker et al. 2006) (Cruts et al. 2006), but recent findings indicate that these mutations are associated with other neurodegenerative disorders with tau pathology, including AD and CBD. Leverenz et al. found that families with the GRN g.1871A>G (p.Ala237TrpfsX6) mutation had variable clinical presentations such as PD, AD, HD, depression and schizophrenia (Leverenz et al. 2007). Immunohistochemical analyses revealed that six of seven cases had evidence of distinctive tau pathology and two of the seven cases also had α-synuclein pathology (Leverenz et al. 2007).

A reduction in progranulin in tau transgenic mice was associated with an increasing tau accumulation (Hosokawa et al. 2015). A reduction in progranulin in APP transgenic mice was associated with a decrease in Aβ accumulation (Takahashi et al. 2017; Hosokawa et al. 2018).

Human GRN mutation cases were investigated histochemically and biochemically by Hosokawa and colleagues. Results showed a neuronal and glial tau accumulation in 12 of 13 GRN mutation cases (Hosokawa et al. 2017). Tau staining revealed neuronal pretangle forms and glial tau in both astrocytes and oligodendrocytes. Furthermore, phosphorylated α-synuclein-positive structures were also found in oligodendrocytes as well as in the neuropil. Immunoblot analysis of fresh frozen brain tissues revealed that tau and α-synuclein were present in the sarkosyl-insoluble fraction and were composed of three- and four-repeat tau isoforms, resembling those found in AD. These data suggested that PGRN reduction might be the cause of multiple proteinopathies due to the accelerating accumulation of abnormal proteins. These might include TDP-43 proteinopathy, tauopathy and α-synucleinopathy (Hosokawa et al. 2017).

Very recently, Sieben and their colleagues reported that a family with a GRN loss-of-function mutation (IVS1+5G>C) had tau and α-synuclein pathology (Sieben et al. 2018). Of nine members of this family, all were tau-positive and one case had extensive Lewy body pathology. No Aβ pathology or mild accumulation was observed (Sieben et al. 2018).

Recent findings have suggested that different clinical phenotypes may occur in carries of the same GRN mutation and additional tau or α-synuclein accumulation may be observed. It has been also reported that PGRN deficiency causes lysosomal dysfunction (Tanaka et al. 2017). Lysosomal dysfunction may reduce protein degradation in brain cells, allowing aggregation-prone neurodegenerative disease-related proteins to deposit more easily (Hosokawa et al. 2017).