Introduction

Spastic paraplegia is characterized by an abnormal gait caused by lesions of the axons of the upper motor neurons in the cortico-spinal (pyramidal) tract. Spastic stiffness in the legs, brisk reflexes, and bilateral extensor plantar reflexes are the clinical hallmarks of the disease. Hereditary spastic paraplegias (HSP) were described in the nineteenth century by Strümpell and Lorrain and are clinically and genetically heterogeneous [1, 2]. Clinically, HSP are categorized as pure or complicated. Pyramidal signs including proximal muscle weakness and/or bladder dysfunction characterize pure HSP, but patients may also have decreased vibration sense at the ankles, pes cavus or scoliosis. In complicated forms of HSP, other neurological signs are present, such as cerebellar signs, peripheral neuropathy, cognitive impairment, and extra-neurological signs such as gastroesophageal reflux, retinal degeneration, optic atrophy, and deafness. The modes of inheritance are autosomal dominant, autosomal recessive (AR-HSP), and X-linked. More than 30 loci have been associated with HSP and are designated in the chronological order of identification after the initials SPG (spastic paraplegia gene) [3].

In 1959, Kjellin described an autosomal recessive form of HSP with onset in young adulthood associated with mental retardation, further intellectual deterioration, hand amyotrophy, and macular degeneration [4]. This disorder was later shown to be linked in two families to a novel locus on chromosome 14q, designated SPG15, in a 19-cM interval flanked by markers D14S1038 and D14S61 [5].

We report the clinical phenotype of three new families with HSP significantly linked to SPG15 and the results of molecular studies that reduced the candidate interval.

Materials and methods

Patients

Twenty families with complex (n = 14) or pure (n = 6) AR-HSP were selected. They included 147 individuals, 64 of whom were affected. All individuals were examined by a neurologist (AL, AH, NBi, IL, MT, ZA, AD, MY, ABe, or ABr). Local ethics committees of the involved institutions approved the study, and the DNA was extracted from blood samples using a standard protocol and after written and informed consent.

Three consanguineous families of Arab origin from Algeria, Morocco, and Israel (families FSP353, FSP444, and FSP671, respectively; Fig. 1) were included in a genome-wide scan. The 17 other families, 10 of them consanguineous, originated from Algeria (n = 6), France (n = 2), Portugal (n = 2), Morocco (n = 2), Tunisia (n = 2), Saudi Arabia (n = 1), Lebanon (n = 1), and Czech Republic (n = 1). All families had previously been excluded for mutations in the SPG7 gene either by direct sequencing of the gene in the index cases [6] or by the absence of shared haplotypes at three flanking markers in affected subjects (data not shown) or, at least, no homozygosity in consanguineous affected patients. In addition, family FSP671, which had a thin corpus callosum in addition to spastic paraplegia, was not linked to SPG11 or SPG21 [7].

Fig. 1
figure 1

Pedigrees and haplotype reconstruction using several analyzed informative markers. The code numbers of all sampled individuals (asterisk) are given close to the symbols. Black circles (women) and squares (men) indicate affected members. The homozygous haplotype, in which the mutation is presumed to be found, is flanked by black boxes. Arrowheads indicate the position of probable recombination events. Distances between markers and map positions in chromosome 14 are indicated in Mbases according to the Ensembl database and Illumina (#)

Genotyping

Microsatellite markers were amplified with fluorescent primers and the fragments resolved on an ABI-3730 sequencer (Applied Biosystems, Foster City, CA, USA). Genotypes were determined with the GeneMapper 3.5 software (Applied Biosystems), and haplotypes were reconstructed manually.

The genome-wide scan was performed by standard techniques using 400 microsatellites (ABI-Prism linkage mapping set v2, Applied Biosystems) in families FSP353 and FSP444, and 6,000 SNPs (Linkage_IV_B_Genetic_Map_v2_RevD, Illumina) in family FSP671. Pairwise and multipoint linkage analyses were performed using Allegro 1.2c [8], Merlin 1.0.0 [9] and Fastlink 4.0 [10]. The disease was considered an autosomal recessive trait (95% penetrance), with a disease allele frequency of 0.0001 and equal recombination fractions for men and women. Genetic distances were those of the Marshfield Centre for Medical Genetics (http://research.marshfieldclinic.org/genetics), and map positions were verified on the human genome sequence draft (http://www.ensembl.org).

Candidate gene analysis

All coding exons and splice junctions, including at least 50 base pairs (bp) on each side, as well as ~450 bp in the 5′- and 3′-UTRs of the GPHN (Gene-ID10243, 23 exons) and SLC8A3 (Gene-ID6547, 7 exons) genes were sequenced in an affected individual from each of the three linked families identified in this study by BigDye v3.1 chemistry on an ABI-3730 automated sequencer according to the manufacturer’s recommendations (Applied Biosystems). Sequences were analyzed using Seqscape 2.5 (Applied Biosystems).

Results

Before the genome scan in the three most informative families, linkage to several known loci for AR-HSP (SPG5, SPG21, SPG24, SPG28, and SPG30) was excluded using appropriate polymorphic markers because the patients were not genetically identical by descent (data not shown).

Fine mapping of SPG15

The genome-wide scan performed using microsatellites in families FSP353 and FSP444 provided evidence of linkage on chromosome 14q in a region overlapping the SPG15 locus, with multipoint logarithm of the odds (LOD) scores of 3.5 and 2.4, respectively.

Analysis of 18 additional microsatellite markers in family FSP353 generated a significant pairwise LOD score >3 (Table 1) at Θ = 0.00 for marker D14S588 (Z = 3.14) and highly positive values of 2.98 at markers D15S1046, D14S119, and of 2.95 at markers D14S125 and D14S1065. A maximal and significant multipoint LOD score of 3.5 was reached in the D14S1012D14S1002 interval (Fig. 2). Haplotype reconstructions showed that all markers were homozygous in affected patients in the interval flanked by markers D14S1012 and D14S1002 because of a recombination between D14S1012 and D14S1026 in individual 10, and between D14S258 and D14S1002 in individuals 4 and 6 (Fig. 1). The telomeric boundary would be placed at marker D14S1029 by taking into account the young unaffected patient 11, aged 18, who carries the distal part of the linked haplotype.

Table 1 Pairwise LOD scores for a series of 17 informative markers from chromosome 14q in the three SPG15-linked families
Fig. 2
figure 2

Multipoint linkage analysis for the three families linked to SPG15. LOD scores are plotted according to the genetic map of chromosome 14q. Distances between markers are indicated in centimorgans according to the Marshfield database. SNPs rs912377 and rs8688 were used only for family FSP671 calculations using genetic distances deduced from the physical distances (1 cM~1 Mb)

Similarly, the use of 15 additional markers in family FSP444 generated positive two-point LOD scores at ten consecutive markers between D14S981 and D14S1002 (Table 1) with a maximal multipoint LOD score of 2.4 in this interval (Fig. 2). Haplotype reconstructions were in accordance with the linkage results and showed that all markers in this interval were homozygous in affected patients. The interval was flanked by two recombinations observed in patient 9 between D14S981 and D14S125 and between D14S258 and D14S1002.

For family FSP671, the genome-wide scan provided evidence of linkage at 17 consecutive SNPs on chromosome 14q with a multipoint LOD score of 4.3 (data not shown). Further analysis of 18 highly informative microsatellite markers in this region confirmed linkage with pairwise LOD scores >3 (Table 1) at Θ = 0.00 for markers D14S1046 (Z = 3.71), D14S981 (Z = 3.02), D14S1069 (Z = 3.29), D14S588 (Z = 3.31), and D14S258 (Z = 3.48). By taking into account microsatellites and SNPs, a maximal and significant multipoint LOD score of 4.3 was obtained in the rs912377rs8688 interval (Fig. 2) in accordance with haplotype reconstructions (Fig. 1) showing that all markers in this interval were homozygous in affected patients and were flanked by two recombinations observed between D14S258 and rs8688 in patient 4 and between rs912377 and D14S1012 in patient 5.

Therefore, the minimal region shared by all affected patients in the three SPG15-linked families is of 5.3 Mb between D14S981 and rs8688, corresponding to ~4.3 to 8.6 cM according to the Decode and Marshfield maps, respectively. This reduces significantly the previously published SPG15 interval of 16.7 Mb (19.5 cM) [5]. Six other possible locations with multipoint LOD scores between 1.1 and 1.9 were detected on chromosomes 2, 5, 8, 10, 20, and 21 in family FSP353 but were excluded when 25 additional markers were used (data not shown). In family FSP444, the analysis of five other markers allowed us to exclude two locations with multipoint LOD scores >1 on chromosomes 2 and 4 (data not shown). Apart from SPG15, no other region in the genome was positively linked to the disease in family FSP671.

Direct sequencing of the coding exons and UTRs of GPHN and SLC8A3, two genes from the candidate interval, did not reveal any new polymorphisms, deletions, or mutations in the index cases of the three families.

Frequency of SPG15

Seventeen additional families with AR-HSP were screened for linkage to SPG15 with a series of ten markers between D14S981 and D14S1002. In all families, however, co-segregation of this region with the disease was excluded on the basis of haplotype reconstructions showing that affected sibships were not identical by descent for the analyzed markers (data not shown).

Clinical presentation of patients linked to SPG15

Patients from all three families presented with an early onset spastic paraplegia associated with cognitive deterioration and additional neurological symptoms that varied among patients and families (Table 2).

Table 2 Clinical characteristics of SPG15 patients

The clinical description of family FSP671 has already been reported [7]. Briefly, the phenotype included spastic paraplegia associated with saccadic pursuit, cognitive impairment, and thin corpus callosum with white matter abnormalities in the two patients who had magnetic resonance imaging (MRI) studies. Patients FSP671-4 and FSP671-5 initially manifested gait disturbances during adolescence (13 to 19 years), and patient FSP671-10 was reported to have mild mental retardation since early childhood. In addition, axonal neuropathy and extrapyramidal signs were noted in one patient each.

In patients from family FSP353, spasticity was associated with axonal neuropathy in the two patients who underwent electromyographic studies. Cerebellar ataxia and mental deterioration were also found in one patient. Age at onset was similar and ranged from 10 to 12 years. The progression of the disease was moderate and running, became difficult by 6 years of evolution, and impossible after 20 years. Family FSP444 also had a complicated form of spastic paraplegia, with cerebellar signs (n = 2) and intellectual deterioration (n = 1). Ages at onset ranged from 12 to 16. The progression of the disease was severe in one patient who needed bilateral support to walk and one patient who was bedridden after 15 years of progression. The third patient had difficulty walking after 2 years of progression.

Brain MRI or computed tomography (CT) could not be performed in families FSP353 and FSP444, who lived in a remote area, and it was therefore not possible to verify whether they also had a thin corpus callosum like family FSP671 [7]. Similarly, ocular examination for maculopathy was not possible in these two families. However, decreased visual acuity was not detected on neurological examination. Formal ocular examinations in family FSP671 and an electroretinogram in one patient were unremarkable.

Discussion

In a group of 20 kindreds with AR-HSP, we identified three new SPG15 families, suggesting a frequency of ~15%. Furthermore, haplotype reconstruction in our SPG15 families reduced the candidate interval from 16.7 [5] to 5.3 Mb flanked by polymorphic markers D14S981 and rs8688. This region contains ~50 candidate genes, including GPHN, encoding Gephyrin, a microtubule-associated protein involved in membrane protein-cytoskeleton interactions thought to anchor glycine receptors to subsynaptic microtubules. SLC8A3, a sodium/calcium exchanger that is highly expressed in brain, muscles, and retina and plays a role in the neuromuscular junction, was also considered a good candidate. No mutations, deletions, or new polymorphisms were detected in either gene by direct sequencing of their coding exons and UTRs. Mutations in introns, uncharacterized exons or duplications have not been excluded, however, but they are unlikely because three patients of different origins and probably with different mutations were screened.

The SPG15 locus was identified in families with Kjellin syndrome. It associates spastic paraplegia, distal amyotrophy in the hands, mental retardation, oligophrenia, and central retinal degeneration with an onset occurring during infancy or adolescence [4]. Signs of cerebellar dysfunction, psychosis, epilepsy, sphincter incontinence, and extra pyramidal signs have also been reported [11]. Webb et al. [11] also identified diffuse brain atrophy on cerebral MRI affecting the cerebral hemispheres, corpus callosum, and brainstem.

The clinical phenotype in our patients is very similar to the patients studied by Webb et al. [11]. Gait disturbance began in the second decade of life, caused by progressive spastic paraplegia. More variable features include cognitive impairment, signs of cerebellar dysfunction, distal amyotrophy, and peripheral neuropathy. Interestingly, thin corpus callosum on brain MRI in family FSP671 [7] also confirms previous reports [11], and cerebral white matter abnormalities in this family add a new imaging feature characterizing this disorder.

The association of cerebellar signs and mental retardation or deterioration with spastic paraplegia has already been reported in several forms of HSP, including SPG11 [12]. In contrast, the association of macular degeneration with HSP was considered to be characteristic of Kjellin syndrome. Interestingly, none of our SPG15 patients complained of decreased visual acuity, and formal ocular examination in family FSP671 did not reveal retinal changes. Given the late onset of maculopathy in the patients reported by Webb et al. [11], it is possible that our patients will develop macular degeneration in the future. Our study underlines the variability of SPG15, which is not always associated with the full picture of Kjellin syndrome. However, spastic paraplegia with cognitive impairment and mild cerebellar signs seem to be common to all SPG15 families.

In conclusion, we have identified three new families linked to the SPG15 locus, estimated the frequency of this genetic entity to ~15%, and significantly reduced the number of candidate genes, two of which were excluded. The clinical features varied among patients and families, associating either cerebellar signs or distal amyotrophy with cognitive impairment, thin corpus callosum, and white matter changes, but not maculopathy in contrast to the Kjellin syndrome. Our data confirms the wide phenotypic heterogeneity of HSP and of SPG15 in particular.