Abstract
Background:
Frontotemporal lobar degeneration (FTLD) designates a group of neurodegenerative diseases with remarkable clinical, pathological, and genetic heterogeneity. Mutations in progranulin gene (GRN) are among the most common causes of familial FTLD. The GRN C157KfsX97 mutation is the most frequent mutation occurring in Southern Italy and has been already described in a previous work.
Objective:
In this study, we reported on additional cases carrying the same mutation and performed a genetic study on the whole cohort, aiming at demonstrating the existence of a founder effect and estimating the age of this mutation.
Methods/Results:
Based on the haplotype sharing analysis, a founder effect was highly probable, while the age of the mutation, estimated by means of DMLE+ software, resulted in a range between 52 and 82 generations, with the highest frequency at about 62 generations, 1,550 years ago.
Conclusion:
This is the first study that reports the age estimation of the most recent common ancestor for the GRN C157KfsX97 mutation recurring in Southern Italy. Mutation dating in a geographically restricted population may be useful in order to plan genetic counseling and screening programs in the field of public health.
INTRODUCTION
Frontotemporal lobar degeneration (FTLD) comprises a group of neurodegenerative diseases heterogeneous from a clinical, pathological, and genetic point of view. Frontal behavioral and cognitive changes characterize the behavioral variant of frontotemporal dementia (bvFTD), whereas language disorders predominate in primary progressive aphasia, further distinguished into a nonfluent/agrammatic variant and a semantic variant. Frontotemporal dementia with motor neuron disease, progressive supranuclear palsy syndrome, and corticobasal syndrome (CBS) are also included in the FTLD spectrum [1]. Family history is present in 30–40% of FTLD patients, and a pathogenic mutation with autosomal dominant transmission can nowadays be identified in the majority of them [2].
Three major genes are responsible of genetic FTLD: microtubule-associated protein tau (MAPT) gene, whose mutations drive tau pathology (FTLD-TAU), progranulin (GRN), and chromosome 9 open reading frame 72 (C9orf72) genes, which are associated with TDP-43 pathology (FTLD-TDP) [1]. Additional less common causative genes have also been described [3].
GRN mutations are among the most recurrent causes of FTLD [4, 5], accounting for up to 12% of all FTLD cases and 26% of familial cases [6]. About 200 GRN variants have been identified so far [7, 8], and clear pathogenicity has been proven for several null mutations determining haploinsufficiency with partial loss of progranulin [9–11]. Incomplete age-related penetrance and significant inter- and intra-familial phenotypic variability are distinguishing features for most of them [1, 13].
In a previous study, we reported the GRN C157KfsX97 mutation in seven apparently unrelated patients coming from Southern Italy, hypothesizing the existence of a likely founder effect [14].
We recently identified the same mutation in four additional patients native to the same geographic area, and we aimed at understanding whether the GRN C157KfsX97 mutation arose several times, or rather a unique founder could explain its recurrence in Southern Italy. In the present study, we performed an extended haplotype sharing analysis to confirm the hypothesis of a founder effect for this mutation and estimated the age of the most recent common ancestor.
MATERIALS AND METHODS
Participants
All 11 patients came from Southern Italy and were recruited at: 1) Neurological Division of the University of Campania “L. Vanvitelli”, Naples, Italy (patients 1–3; 5–9); 2) Department of Neurology of Toulouse and Hôpital Pitié-Salpêtrière, Paris, France (patient 4); 3) Division of Neurology V - Neuropathology, Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy (patients 10 and 11). They all underwent a complete diagnostic assessment for cognitive disorders, except for cases 3 and 11, who only were referred for molecular and genetic evaluation. Plasma progranulin dosage with an ELISA kit (Human Progranulin ELISA kit, Adipogen Inc.) and subsequent GRN analysis were performed for all subjects, except patients 4 and 11, whose plasma samples were not available. Clinical, instrumental, and biochemical data of the cases 1–7 were previously reported [14]. Since then, we identified four more apparently unrelated patients coming from Southern Italy and carrying the GRN C157KfsX97 null mutation, reported below.
Case 8
A 60-year-old man presented at age 55 with an insidious onset of a neuropsychiatric disorder mainly characterized by jealousy delusions and depressed mood. Over the next few years, he developed progressive language disturbances, with interrupted speech, poor fluency, and word-finding difficulties. Neurological examination showed nonfluent aphasia and frontal release signs. Mini-Mental State Examination (MMSE) score was 24.2/30. Brain MRI evidenced mild left temporo-insular atrophy, whereas left fronto-temporal and temporo-parieto-occipital hypoperfusion was detected on 99mTc-SPECT brain scan. Cerebrospinal fluid (CSF) biomarkers revealed only slight increase of total tau (450 pg/ml; normal values -n.v.- < 400 pg/ml) [15]. Plasma progranulin dosage was 31.2 ng/ml (n.v. > 61.55 ng/ml) [16]. A diagnosis of “Primary Progressive Aphasia (PPA) with additional neuropsychiatric disturbances” was made. A family history of cognitive and behavioral impairment was present: his father died at age 70 affected by dementia, a brother died at age 45 affected by an unspecified behavioral disorder, and a paternal cousin was diagnosed with obsessive-compulsive disorder.
Case 9
A 67-year-old man insidiously developed a frontal syndrome with marked echolalia and palilalia at age 65. In the following years, behavioral symptoms as irritability, hyperorality, and visual hallucinations were reported. Neurological examination showed bilateral (L > R) akinetic-rigid parkinsonian syndrome with freezing of gait and frontal release signs. MMSE score was 10.3/30 and Frontal Assessment Battery (FAB) score was 5/18. Brain MRI showed marked cortico-subcortical frontal atrophy, prevalent on the right side. Plasma progranulin was 29.4 ng/ml. He received a diagnosis of bvFTD with parkinsonism. Family history was positive for neurodegenerative disorders: a brother was affected by unspecified dementia and two maternal cousins suffered from a parkinsonian syndrome.
Case 10
The patient was a woman with a medical history of dysthyroidism, hypertension, and uterine cancer treated with hysterectomy. At age 65 she presented depression and slowly progressive alteration of speech. A few months later, neurological examination showed dysarthria and dysgraphia, associated with impairment of language and executive functions. Brain 18FDG-PET showed hypometabolism of frontal, temporal, and parietal left cerebral cortices, more evident in frontal cortex. One year after disease onset, MRI showed diffuse cortical atrophy and scattered T2 hyperintense lesions involving the periventricular and subcortical white matter. MMSE and FAB scores were 23/30 and 10/18, respectively. An extensive neuropsychological evaluation showed alterations of phonemic fluency, attention, executive functions, and constructional apraxia. CSF Aβ42 and T181 phosphorylated tau were within reference ranges, while total tau was slightly increased (497 pg/ml; n.v. < 400). Plasma progranulin was 35.8 ng/ml. A sister who died at age 54 and a living brother were also affected by unspecified dementia. Both parents died at age 70 of stroke.
Case 11
This woman received a diagnosis of familial FTLD at age 73 in other hospital, and unfortunately no additional details were available.
To estimate the age of the mutation, we recruited 48 healthy controls from Southern Italy, all assessed by formal neuropsychological evaluation to exclude any cognitive impairment. Family history of neurodegenerative diseases was also carefully ruled out.
Median age at onset of patients was 59 (range 52–73 years) and median age at blood collection of controls was 62 (range 52–71 years).
Genetic analysis
Blood samples were collected from patients and controls after obtaining informed consent, as approved by the local ethical committee (For Italian samples: Ethical Committee of the University of Campania “L. Vanvitelli”, Naples, Italy; 108/2017. For French samples: Ethics committee of “Assistance Publique – Hôpitaux de Paris” Ile de France VI). DNA was extracted from lymphocytes using standard procedures. All GRN coding exons, flanking intronic sequences, and 5’ and 3’ untranslated regions were sequenced as previously described [14].
For the haplotype sharing study, we analyzed 8 short tandem repeat (STR) markers spanning a region of 7.29 Mb flanking GRN on chromosome 17q21. D17S1787, D17S1801, D17S951, D17S1861, D17S934, D17S950, D17S931, and D17S1827 were amplified with a fluorescently labeled primer and analyzed on an ABI3130xl DNA analyzer (Life Technologies/Thermo Fisher Scientific). Alleles were scored using the GeneScan software (Life Technologies/Thermo Fisher Scientific).
Estimation of the mutation age
The data available for the estimation consisted of the genotypes from 11 patients and 48 controls, including 8 STR markers and the GRN C157KfsX97 mutation. Estimation of the most recent common ancestor age was performed with the DMLE+ software [17] following a previous protocol [18]. Estimations with ESTIAGE were impossible because the software is not available anymore. Briefly, haplotype frequencies were calculated from genotypes in cases and controls using Arlequin 3.11 [19]. The population growth rate of Campania (r) was estimated using the equation T1 = T0e(g r) where T1 is the estimated population of the region from the latest 2019 census (5.8 million inhabitants), T0 is the estimated population in 1861, date of the first census during the Italian national reunification (2.4 million) and “g” is the number of generations between T0 and T1 (g = 6, with one generation considered 25 years long). The resulting growth rate for Campania is r = 0.148. DMLE+ reports the results of multiple iterations (2 million) of Bayesian inference of mutation age based on the observed linkage disequilibrium at the provided genetic markers, correcting the results based on the provided control haplotypes representing the general population. The distribution allows to define a 95% confidence interval (CI) for the mutation age.
RESULTS
Genetic analysis
GRN sequencing on our four patients disclosed the C157KfsX97 mutation previously reported in the other seven patients of our cohort [14].
The haplotype sharing study (Table 1) disclosed a common haplotype with eight consecutive STR markers (from D17S1787 to D17S1827, spanning 7.29 Mb) shared in seven patients (cases 1–6 and 10). Patients 8, 9, and 11 shared seven, six, and five consecutive markers, respectively, of the common haplotype. Patient 7 showed the shortest shared haplotype, which was anyway 2.7 Mb long (from D17S1787 to D17S951); three additional consecutive shared markers (from D17S934 to D17S931) were present, although within a possible recombination region.
Haplotype sharing analysis
Shared short tandem repeats (STR) are in bold. The alleles differing from the shared one are in red. aMap position from UCSC Genome Browser on Human Feb. 2009 (GRCh37/hg19) Assembly.
Estimation of the mutation age
Using the DMLE+ software to estimate the age of the most recent common founder, we obtained an estimated age of 62 generations (95% CI: 52–82 generations). The plot shows the frequency at which each number of generations was output from the iterations (Fig. 1). The green interval shows the 95% CI of the distribution.

Age estimation of the most recent common founder of the GRN C157KfsX97 mutation (DMLE 2+ software). The frequency at which each number of generations was output from the iterations is shown. The green bars constitute the 95% confidence interval of the distribution.
DISCUSSION
An increasing number of GRN mutations have been reported since the identification of this gene as one of the causes of FTLD [8, 20]. Most of the pathogenic mutations are frame-shift, splicing, or STOP mutations resulting in null alleles which lead to a condition of haploinsufficiency [1]. Some mutations show a high frequency, worldwide or in more restricted geographical areas. This raises the issue of discriminating between multiple and independent mutation occurrences, which would suggest a mutation hot spot within the GRN locus, and a single occurrence, which would be rather in favor of a unique founder. The most common GRN mutations worldwide, IVS0 + 5G > C and R493X, appear to have had one or at most two original founders, respectively [5, 22]. The same situation has been described for the most frequent GRN mutation in Italy, the T272fs (alias L271LfsX10): although two different studies did not actually provide comparable results in terms of mutation dating (probably due to different number of families enrolled), they demonstrated the existence of one or at most two founders [18, 23].
Thus, we decided to explore the presence of a common founder for the GRN C157KfsX97 mutation, present in a cluster of apparently unrelated patients originating from Southern Italy.
It is worth mentioning that Southern Italy is a territory which has contributed several mutations in dementia genes. The PSEN1 M146L mutation, detected in at least 15 families worldwide, was in fact demonstrated by haplotype analysis to have a common founder originating in Calabria region [24]. The APP A713T mutation, described in several families, has also been hypothesized to have its founder in the same Southern region [25].
A first report on seven patients carrying the GRN C157KfsX97 mutation [14] suggested a common founder, due to a widely shared haplotype. The investigation of additional four patients in this study seems to confirm a common haplotype: eight consecutive STR markers were shared in seven patients (cases 1–6, 10), at least five STR markers were shared in other three patients (cases 8, 9, 11) and three markers in one patient (case 7) (Table 1). For patients 8, 9, and 11, a likely recombination event distal to the mutation site on the common chromosome caused the inheritance of different STR markers. For patient 7, a more complex recombination pattern must be hypothesized or, alternatively, a different founder haplotype. Nevertheless, this latter scenario appears less probable because of the sharing of additional three STR markers (from D17S934 to D17S931) spanning a long region (2.7 Mb).
To date other mutations in the GRN gene have been reported in the same geographical area, but only concerning cases belonging to single families [26–29] or few families [30]. The GRN C157KfsX97 represents therefore the most frequent mutation in Southern Italy [14]. On this basis, we considered it important to study the potential founder effect of this mutation, thereby defining for the first time the existence of a common ancestor for an FTLD-causing mutation in Southern Italy.
The wide clinical variability of our patient cohort ([14], this study), which showed different clinical phenotypes encompassing CBS, bvFTD, and PPA, as well as some heterogeneity in the age of onset, may suggest the presence of potential genetic modifiers. However, the finding of the haplotype sharing and the unique founder suggests that these genetic factors should be searched for in other genomic regions, independent from the haplotype region. Common variants in GRN and transmembrane protein 106B gene (TMEM106B) have already been reported to influence FTLD risk and age of onset [3, 32], and very recently a large genome-wide association study showed the GDNF family receptor alpha-2 gene (GFRA2) as an additional FTLD risk factor/disease modifier [33].
Our findings are consistent with previous reports that no mutation hot spot could be described to date in GRN locus, based on the founder analyses carried out so far [5, 22]. Conversely, similar studies on mutations in other FTLD-related genes led to different results. The P301L, one of the most frequent mutations in MAPT gene [34], was carried on multiple independent haplotypes, indicating multiple occurrences and featuring this codon as a mutation hot spot [35].
To estimate the age of the most recent common founder we used the DMLE+method, which is based on genetic data and takes into account the estimated population growth rate. The analysis suggests that the most recent common founder dates back 62 generations. The estimated 62 generations age is the center of the distribution of the estimations obtained during the permutations. The analysis shows some limitations highlighted by the size of the confidence interval, that falls in the 52–82 generations range for 95% confidence, or 750 years. This large confidence interval has its root causes to the number of permutations and the sample size. Although increasing the permutations can indeed increase the detail of the distribution, the key point is the sample size: in order to achieve more precise estimations, the analysis must be run on more patients, increasing the recruitment requirements. However, this is of course a complex issue, due to the rarity of GRN-associated FTLD.
At present, 62 generations constitute the best estimation that can be achieved with the available techniques and data. Considering 25 years for a generation, it indicates 1,550 years ago. This approximately corresponds to the period of the fall of the Western Roman Empire (476 AD), an age in which Campania region became a “melting pot” between the local inhabitants of Latin and Greek origin and the invaders coming from Northern Europe.
Mutation dating in a geographically restricted population may be useful in order to plan genetic counseling and screening programs in the field of public health. Mutation carriers, both affected or presymptomatic, may be offered participation in clinical trials for new drugs possibly increasing progranulin expression, which might be an effective way to disease treatment or modification.
