Introduction

Spastic paraplegia type 11 (SPG11; OMIM# 604360) is the most frequent form of autosomal recessive spastic paraplegias (HSP).1,2 It is mainly characterized by rapidly progressive spasticity and muscle weakness of early onset3 and a thin corpus callosum on brain magnetic resonance imaging (MRI); it is often associated with mental impairment and progressive cognitive decline.3 Nonetheless, cases of SPG11 without thin corpus callosum4,5 have been reported and, as more families are studied, the clinical spectrum is broadening.

The SPG11 gene encodes a ubiquitous, highly conserved protein named spatacsin.3 This protein is highly expressed in the central nervous system; recent immunocytochemistry assays showed that spatacsin is distributed throughout the cytoplasm and is probably involved in axonal transport.6 A possible involvement in DNA repair has also been suggested; in fact, a new HSP disease-causing gene was recently identified (KIAA0415, responsible for SPG48) and its product was found to be involved in DNA repair.7 Moreover, this protein is an interactor of two other HSP proteins, spatacsin and spastizin, whose overlapping brain expression patterns8 and clinical traits (SPG11 and SPG15) suggest an involvement of these three proteins in the same pathway.

Among HSP genes, SPG11 has a high mutation rate, and notwithstanding its size, it is actually the most frequently mutated gene in autosomal recessive spastic paraplegia with thin corpus callosum. To date, more than 100 different mutations1,2,3,4,5,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23 have been reported worldwide, in a total of 128 families with complex autosomal recessive spastic paraplegia.

In this study, we report 13 new families with SPG11 presenting novel or previously described mutations, as well as the molecular mechanism responsible for large deletions in this gene; we also discuss the high frequency of the large gene rearrangements identified in this gene.

Materials and Methods

Patients

We selected a group of 54 unrelated Portuguese patients with HSP to screen for SPG11 mutations: 19 cases with a suggestive pattern of autosomal recessive inheritance and 35 isolated cases. Twenty-one patients presented with pure HSP, whereas 33 had a complex phenotype, with mental impairment, neuropathy, and/or epilepsy in addition to the spasticity of the lower limbs. Some patients presented with a thin corpus callosum, white matter hyperintensity, or medullar atrophy on brain MRI. In addition, SPG11 mutation screening was also performed in two Dutch families.

Mutation screening

Peripheral blood was collected from patients and their relatives after written informed consent was obtained. Genomic DNA was extracted from peripheral blood leukocytes through salting out,24 and specific primers were designed for all coding regions (including splice site boundaries and intronic flanking regions) of the SPG11 gene (RefSeq NM_025137.3), using PrimerQuest. Each exon was amplified by PCR with HotStarTaq Master Mix (Qiagen, Hilden, Germany), directly sequenced with the Big Dye Terminator Kit v1.1 (Applied Biosystems, Carlsbad, CA), and loaded on an ABI 3130xl Genetic Analyzer (Applied Biosystems, Carlsbad, CA).

Multiplex ligation-dependent probe amplification

To detect large gene rearrangements (deletions or duplications), multiplex ligation-dependent probe amplification (MLPA) was performed using the Salsa MLPA kit P306-A1 (MRC-Holland, Amsterdam, The Netherlands), according to the manufacturer’s instructions. Fragments were analyzed on an ABI 3130xl Genetic Analyzer using 250-LIZ (Applied Biosystems, Carlsbad, CA), as a size standard, and GeneMarker v1.90 (SoftGenetics, State College, PA).

Long-range PCR and breakpoint analysis

Long-range PCR was performed on genomic DNA from probands to confirm the large SPG11 deletions detected by MLPA; breakpoints were determined by direct sequencing. The reaction used the Expand Long template PCR system (Roche Diagnostics, Manheim, Germany), according to the manufacturer’s instructions. Primers for the exons upstream and downstream of the deletion were used and the deletion breakpoints were afterwards narrowed down by primer walking. Finally, deletion junctions were amplified with specific primer sequences, followed by direct sequencing for breakpoint confirmation.

In silico analysis

All new mutations found were analyzed with Alamut v2.0 (Interactive Biosoftware, Rouen, France). The entire SPG11 gene (NCBI36:15:42641587:42743768:-1), including 600 bp upstream and downstream of the 5′ and 3′ untranslated regions, was analyzed using both RepeatMasker (http://www.repeatmasker.org) and CENSOR (http://www.ebi.ac.uk/Tools/censor) to identify interspersed repeats. Sequence identities of Alu repeat signatures of interest were evaluated with the National Center for Biotechnology Information (NCBI) BLASTN tool; the homologous genes of four primate species (Gorilla gorilla, Pan troglodytes, Pongo pygmaeus, and Macaca mulatta) were also analyzed with RepeatMasker and NCBI BLASTN for comparison with the human gene regarding the presence or absence of the particular Alu elements of interest. Finally, COMPASS25 was used to search for complex motifs, degenerated from the sequence CCNCCNTNNCCNC, in the human SPG11 gene.

Results

SPG11 mutations in patients with spastic paraplegia

Mutation screening and gene dosage analysis of SPG11 allowed the identification of disease-causing mutations in 14 of 54 unrelated HSP patients tested. Through SPG11 direct sequencing, we identified 6 point mutations (mostly small deletions) including one novel missense and a 4-bp deletion. MLPA allowed us to detect three new large gene rearrangements—two different single-exon deletions, as well as multiple exon deletions, and an interesting case of a heterozygous complex gene rearrangement. The main clinical features and the mutations of the diagnosed patients are summarized in Table 1; the presence of the identified mutations was confirmed in additional affected family members as well as its parental segregation.

Table 1 Clinical and imagiologic features of SPG11 patients identified
Full size table

The first patient studied (case 1) carried a novel homozygous 1-bp exchange, c.3809T>A in exon 22, resulting in a missense substitution of a valine at protein position 1270 for an aspartic acid ( Figure 1a ), a very conserved residue. Of interest, patients 8 and 11 were found to be homozygous for a novel large deletion encompassing exon 29. PCR of this exon did not result in fragment amplification, and MLPA confirmed the homozygous deletion. Although they share the same mutation, the two patients originate from different geographical regions. Patient 9 was found to be a compound heterozygous carrier of the c.733_734delAT, in exon 4, and a novel large deletion comprising exon 33 detected by MLPA. Patient 10 presented a more challenging case as she was found to be compound heterozygous for the c.6832_6833delAG deletion in exon 37 and a complex entire gene rearrangement comprising the duplication of exons 1 to 30 and 35 to 40 in combination with a deletion that encompasses exons 31 to 34 ( Figure 1b ). The duplicated area extended at least to the flanking genes CASC4 and B2M, also detected by the MLPA probe mix. This four-exon deletion was also present in two Dutch families, although not associated with this complex rearrangement. Last, we identified a novel frameshift mutation in patient 14, c.2996_2999delACAG, in exon 16, changing a histidine to a leucine at position 999, leading to an altered reading frame and a premature stop codon after 36 residues (p.His999LeufsX37) ( Figure 1c ). This patient had a complex form of spastic paraplegia, with neuropathy, epilepsy, and mental impairment, and a family history of the disease; however, after mutation screening and MLPA, we could find only this small deletion in the heterozygous state, thereby being unable to establish a diagnosis of SPG11.

Figure 1
figure 1

Pedigrees of three families of our cohort. (a) Family pedigree of proband 1, in whom we found a novel homozygous missense mutation (p.V1270D) (black arrow in the electropherogram, upper panel). Multiple sequence alignment of spatacsin with other species shows the conservation of this amino acid as well as the adjacent sequence. (b) Family of proband 10 (upper panel); electropherogram showing a heterozygous 2-bp deletion in exon 37 and a complex heterozygous gene rearrangement identified by multiplex ligation-dependent probe amplification, involving duplications and a deletion (lower panel). (c) Family of proband 14 (upper panel); electropherogram showing a heterozygous 4-bp deletion in exon 16 (lower panel). RPA, relative peak area.

Full size image

Breakpoint deletion determination and deletion mechanism

The breakpoints of the 1056-bp deletion (exon 29) were mapped by primer walking to 524 bp upstream and 317 bp downstream of exon 29 (c.4907-524_5121+317del) in both patients. This deletion ( Figure 2a ) leads to a frameshift, generating a premature stop codon (p.Gly1636AspfsX11). Through bioinformatic analysis, we were able to colocalize both 5′ and 3′ breakpoints within Alu elements, namely an AluSz in intron 28 and an AluY localized in intron 29. These elements are characterized by a high degree of homology (77% identity); moreover, there is a 24-bp region of microhomology adjacent to each breakpoint.

Figure 2
figure 2

Characterization of the spastic paraplegia type 11 (SPG11) intragenic deletions found in this study and schematic representation of its breakpoints. (a) Breakpoint deletion for exon 29 found in two Portuguese families; (b) representation of the exon 33 deletion found in one Portuguese family (showing contamination with wild-type allele in the sequencing product; nevertheless, sequence deconvolution confirms the deletion breakpoints); (c) exon 31 to 34 deletion breakpoint from two Dutch and one Portuguese patient; (d) breakpoint analysis of a previously reported SPG11 deletion shows the same rearrangement mechanism. An Alu-mediated mechanism appears to be responsible for all these deletions.

Full size image

Regarding the exon 33 deletion ( Figure 2b ), its breakpoints were mapped to 169 bp upstream of exon 33 and 833 bp upstream of exon 34 (c.6206-160_6344-833del), resulting in a loss of 1489 bp. This large rearrangement also produces a frameshift, giving rise to a premature stop codon (p.Gly2069AlafsX329). Similar to the previous rearrangement described, both 5′ and 3′ breakpoints also coincide with Alu elements (with the 5′ breakpoint within an AluSc8 in intron 32, and the 3’ breakpoint within an AluY in intron 33), with high degree of homology (86% identity) between them. Furthermore, this breakpoint matches a 19-bp microhomology domain.

The breakpoints for the four-exon deletion ( Figure 2c ) were also mapped (c.5867-3237_6478-450del) in the Portuguese and both Dutch families; likewise, it also results in a frameshift and a premature stop codon after 16 residues (p.T1956SfsX17). The boundaries of this 8323-bp deletion, again, overlap Alu elements (5′ within an AluY in intron 30 and the 3’ coinciding with an AluSx in intron 34), possessing an adjacent 42-bp region of microhomology and a high degree of identity (79%). All results found with RepeatMasker for interspersed repeats were confirmed with the CENSOR software; a total of 189 masked elements were found, representing 55% of the SPG11 gene. Also, COMPASS allowed the identification of 13 degenerate sequences from the CCNCCNTNNCCNC motif, distributed throughout the gene, and totaling 23 occurrences.

Discussion

Spatacsin mutation frequency in patients with spastic paraplegia

The SPG11 mutational spectrum of our cohort comprises mostly small deletions, together with three large-scale rearrangements (six different families), all distributed throughout the SPG11 gene.

Of interest, we found two recurrent mutations, c.529_533delATATT and c.733_734delAT, which together account for 41% of the mutated alleles. A founder effect has been suggested for a few mutations in SPG11 patients from the Mediterranean basin.5 In particular, a haplotype analysis has shown that both Portuguese and Brazilian patients with the c.529_533delATATT mutation share the same haplotype, pointing toward a common mutational origin.4 The 2-bp deletion in exon 4 (c.733_734delAT) is one of the two most frequent mutations worldwide 1,3,4,5,11,12,13,16,22, which might suggest either the presence of a mutational hotspot or a significantly old mutational event. Another interesting finding in this study is the frequency of large deletions, which appear in almost half of our index cases.

In addition, we have also found a novel missense mutation (V1270D) that replaces a highly conserved hydrophobic residue by a negatively charged hydrophilic amino acid ( Figure 1a ). All Web-based variant scoring algorithms used by Alamut (PolyPhen-2, SIFT, Align GVGD), as well as PANTHER,26 SNAP,27 PMut,28 and PhD-SNP,29 predicted this mutation to be pathogenic. This substitution is located adjacent to the third highly conserved α-helix3 and thus could have a major impact on protein folding.

The novel 4-bp deletion results in a premature termination codon in exon 16, and although we could not confirm SPG11 diagnosis in the patient carrying this deletion, we cannot fully exclude it either, since the second disease-causing mutation may be located in regions that our genetic screen does not cover: the promoter, the 3′ or 5′ untranslated regions (intronic flanking regions required for splicing were covered by our analysis). As a general rule, only premature termination codons (PTCs) located <50–55 nucleotides upstream, or downstream of the last exon–exon junction, manage to escape nonsense-mediated decay (NMD).30 Hence, the mutant transcript in question is probably targeted for degradation, because it is located closer to the 5′ end of the mRNA and upstream of several exon–exon junctions. Nevertheless, there are exceptions to this rule. Namely, it was recently shown that a PTC-causing mutation in the middle of the GABAA receptor GABRA1 escapes NMD;31 thus, we can neither confirm nor exclude SPG11 diagnosis.

In contrast, the large deletion found in patient 9, which causes exon 33 skipping, leads to a frameshift and a premature termination codon after 328 amino acids, meaning that, if translated, the protein would have almost a full-length size (2,397 amino acids), i.e., only 46 amino acids shorter than the wild type. Assuming a correct splicing of the remaining exons downstream of the deleted portion, and considering that the PTC is positioned after the last exon–exon junction, it is highly probable that this transcript escapes NMD, originating an almost full-length protein with a mutant C-terminal. This novel spatacsin C-terminus may behave differently from the normal protein, by abnormal protein interaction or trafficking, which could result in a more severe phenotype. In fact, this patient shows a phenotype more severe than average, with a classical age-at-onset (10 years), but presenting an unusually fast progression to wheelchair by the age of 18; this may be due to the presence/aberrant trafficking of the mutant protein.

The large duplicated region found in patient 10 (at least from B2M to CASC4, i.e.,~300 kb) does not seem to overly contribute to the patient’s phenotype. Here, the four-exon deletion (31–34), in combination with the 2 bp point mutation, appears to be responsible for the disease, resulting in the clinical signs compatible with an SPG11 phenotype. It appears that this complex rearrangement has arisen from an SPG11 allele missing exons 31–34, which was afterwards duplicated, encompassing a considerably large flanking region as MLPA has allowed us to determine.

There are several disorders associated with the 15q locus involving copy number variations that can comprise deletions, duplications, and even triplications—Angelman/Prader-Willi syndrome, 15q duplication syndrome, and 15q triplication32—which altogether illustrate the high degree of instability in this chromosomal region.

Alu-mediated SPG11 rearrangements

Regarding the single or multiple exon deletions found in our set of patients, the identification of the same microhomology domains in other Alu elements spread throughout the SPG11 gene ( Figure 3 ) gives rise to the hypothesis that different, yet similar, large-scale rearrangements may occur through recombination mediated by the various Alu elements.

Figure 3
figure 3

Evolutionary conservation of the Alu sequences involved in spastic paraplegia type 11 ( SPG11 ) deletions in nonhuman primates. The Alu elements found in this study to be involved in SPG11 deletions are boxed. Additional regions within the SPG11 gene showing these microhomology domains, as well as the evolutionary conservation of the different Alus in nonhuman primates, are also presented.

Full size image

Although the microhomology sequence involved in the deletion of exons 31–34 does not appear frequently in other Alu elements, as it happens in other cases, the same deletion has been found so far in five additional families of German and Dutch descent.4,10,22 This could suggest that the genomic architecture of this particular gene region is prone to the occurrence of this specific recombination, as we also found it in Portuguese patients.

Besides the deletions reported here, another has been mapped involving the loss of exons 37 to 39 in an Italian patient.11 Our analysis shows that this large gene rearrangement also overlaps with Alu elements and its breakpoints involve a microhomology sequence that is also present in five other Alu elements ( Figure 2d ).

For all SPG11 rearrangements mapped to date, the microhomology region adjacent to both breakpoints is located in the same relative position, closer to one or the other of the Alu extremities. The recombination event excises complementary parts of the Alu elements so that an Alu is left in the recombination spot and another leaves with the deleted portion; thus, these exchange genetic material with each other, allowing for an inter-/intrachromosomal form of nonallelic homologous recombination to take place ( Figure 2 ). Our findings indicate that, although each deletion appears to be unique, Alu sequences are determinant for the instability observed at the SPG11locus. The high frequency of Alu elements in SPG11 suggests a locus prone to novo rearrangements; however, among six patients carrying these large deletions, parental segregation excluded a de novo event in five of the families. Notwithstanding, no information is available regarding the presence of SPG11 rearrangements in the general healthy population.

Evolutionary conservation of Alus involved in human SPG11 deletions in nonhuman primates

To look at these Alu elements in the SPG11 gene from an evolutionary point of view, which could help to explain the genetic rearrangements we found, the SPG11 gene sequence of four nonhuman primates was analyzed ( Figure 3 ). Our results show that the Alu elements containing the microhomology regions that are involved in the large deletions found in SPG11 patients are also present in nonhuman primates and conserved across evolution. These repetitive elements appear to have increased in number, at the SPG11 locus, throughout evolution, with the exception of the Gorilla, though this may be due to the gene not being fully sequenced in this species. Apparent to the observer is also the increase in elements belonging to the newest Alu family, AluY.33 Altogether, the increasing number of Alus present in the gene ultimately favors recombinatory events that might lead to a loss of segments of the SPG11 gene through an intra-/interchromosomal mechanism of nonallelic homologous recombination.

Transposable elements are highly frequent in the human genome, constituting up to 40% of its sequence.34 The majority of these elements lost their transposition ability a long time ago, although some of them, in particular, LINE-1 and Alu, have been shown to maintain this capacity, thus contributing to recent evolutionary history of the primate genomes35 and also to the genomic instability found to be involved in cancer and several genetic disorders.36

Similar to what happens with the BRCA1 gene, often known to suffer Alu-mediated deletions,37,38 analysis of SPG11 also shows a high density of interspersed repeats, namely Alu elements, which together with the large size of this gene (~101 kb) favors the occurrence of various rearrangements. In addition, and similar to what seems to be happening with the SPG11 deletions, the Alu-mediated BRCA1 rearrangements arise from nonallelic homologous recombination.

Recombination hotspots are apparently unrelated to spatacsin rearrangements

Recently, a degenerate 13-bp motif (CCNCCNTNNCCNC) was associated with recombination hotspots and genome instability in humans.39 Moreover, this motif was present in breakpoint regions of disease-causing nonallelic homologous recombination hotspots40 and could, in theory, also be involved in SPG11 deletions. Through bioinformatic analysis, we found 13 degenerate sequences that occur 23 times over the gene’s length, although none of them is located near any of the deletions identified in SPG11 patients. Nevertheless, other recombination events mediated by clusters of this motif might occur, more specifically in introns 18 and 20, where these elements seem to cluster.

Conclusions

In this study, we report 13 new families with SPG11, carrying both novel and previously reported mutations. We describe a complex entire SPG11 rearrangement and show that large gene rearrangements are frequent among patients with SPG11. In addition, we present the molecular characterization of three large SPG11 deletions together with their mechanisms and show a high frequency for Alu- mediated deletions in this gene. Our analysis suggests that the high number of repeated elements in SPG11 together with the presence of recombination hotspots and the high intrinsic instability of the 15q locus all contribute toward making this genomic region prone to large gene rearrangements.

Our findings enlarge the amount of data relating repeated masked elements to neurodegenerative disorders and highlight the importance of these elements in human disease and in evolution.

Disclosure

The authors declared no conflict of interest.