Background
Hereditary spastic paraplegia (HSP) refers to a group of rare disorders in which progressively spastic gait is the major and often primary symptom. The pathological correlate of this phenotype is a distal-to-proximal degeneration of upper motoneuron axons in the corticospinal tract [
1]. While HSPs have long been categorized according to age at onset, mode of inheritance, and presence/absence of additional symptoms, a genetic classification scheme is currently taking over. To date, more than 70 spastic paraplegia gene loci (SPGs) have been reported [
2,
3].
HSP type SPG8 is an autosomal dominant form which is associated with a comparatively young age at onset and a pure, but rather severe phenotype [
4,
5]. In 2007,
KIAA0196 mutations were identified in six SPG8 pedigrees [
5], but only six more families have been published since [
6‐
10].
KIAA0196 codes for strumpellin, a 1159-residue protein which contains a short central spectrin repeat but otherwise seems to lack recognizable homology domains [
5]. Strumpellin is a member of the multiprotein WASH regulatory complex (SHRC) [
11,
12]. This complex associates with retromer, another multi-protein complex, and regulates the tubular extension of early endosomes [
11,
13‐
16]. It may thereby facilitate cargo sorting for endosome-to-Golgi retrieval, for membrane receptor recycling and/or for targeting to the lysosomal degradative pathway [
11,
17,
18]. Distinct additional roles in autophagy have been proposed more recently [
19‐
21].
Eight unique
KIAA0196 missense mutations have been associated with HSP so far [
5‐
10]. By affecting residues 226, 471, 583, 591, 619, 620, 626, and 696, they seem to cluster in the protein’s central part. Interestingly, an overlapping central region is also affected by a genomic deletion of exons 11–15 (encoding residues 470–672) [
8]. Functional assays have been performed for some of the missense variants, but did not reveal any alterations regarding subcellular localization, interaction potential, SHRC assembly, retromer binding, and endosomal tubulation [
12,
22,
23]. In contrast, RNAi-mediated knockdown of strumpellin was found to have strong effects in cell lines [
11,
14,
22,
23] and in zebrafish embryos including abnormal development of spinal cord motoneurons [
5,
22]. Collectively, these findings have been interpreted in light of the mutational mechanism relevant for SPG8: they were suggested to argue against a dominant negative effect of mutant strumpellin alleles on the wild-type allele, but to, instead, indicate loss-of-function-mediated haploinsufficiency [
8,
22,
23]. Against this background, the apparent absence of classical loss-of-function mutations (i.e. non-sense, frame-shift, splice-site, whole gene deletions) in SPG8 patients was attributed to a lack of appropriate tools (e.g. for detecting deleted alleles) and/or compensation by the non-inactivated allele [
23].
In the present study we used genetic and in vivo approaches to further elucidate the potential mechanisms by which mutations in KIAA0196 cause HSP. Our findings strongly question the current haploinsufficiency hypothesis. As they also provide additional evidence against relevance of a dominant negative effect of mutant on wild-type strumpellin, we discuss alternatives and provide a conceptual basis for experimental testing.
Discussion
Our study aimed at testing the current haploinsufficiency hypothesis for the autosomal dominant HSP type SPG8. As a first pertinent approach, we generated a murine
E430025E21Rik allele that lacks an out-of-frame exon. As we found virtually no corresponding mRNA
in vivo, probably because of nonsense-mediated decay [
40], this alteration effectively generated a knockout allele. Contrary to what has been suggested [
23], the heterozygous presence of such an inactivating mutation neither resulted in compensatory up-regulation of mRNA expression from the wild-type allele nor in substantially enhanced stability of the strumpellin protein. Heterozygous presence of a knockout allele therefore entails a reduction of the amount of strumpellin.
Near complete reduction of strumpellin levels by RNAi-mediated knockdown has been shown to severely destabilize other SHRC subunits [
11,
12]. We show that the ~35 % reduction in strumpellin levels observed upon heterozygous knockout results in a similar degree of reduction for Fam21 levels. This is consistent with the proposed general dependence of SHRC subunit stability on proper assembly of the complex [
11‐
13,
17].
Reduced abundance of SHRC in heterozygous strumpellin knockout cells likely results in correspondingly reduced overall SHRC activity. The physiological role of SHRC is not yet completely understood, but recent studies suggested an involvement in maintaining the structure/complexity of the endo- and lysosomal compartments [
17], in inhibiting autophagy [
19], but also in promoting autophagy [
20,
21]. The effects of blocking SHRC function in these studies were rather severe and included collapsing of organelle compartments and aggregate formation. We did, however, not observe any evidence for abnormal endosomes, lysosomes, or autophagosomes/autolysosomes in heterozygous strumpellin knockout MAFs. We also found no increase in endosomal tubulation, i.e. a major consequence of (near) complete reduction of the major SHRC subunit
Wash1 [
11,
23]. This suggests that, in contrast to complete SHRC inactivation, moderately reduced SHRC abundance/activity is enough to maintain grossly normal function of the corresponding organelle systems.
There were also no phenotypic consequences associated with a reduction of murine strumpellin/SHRC
in vivo. One may argue that axon length and maximum age would render mice unsuited models for HSPs in the first place. Numerous recent studies, however, have shown impressive and clearly disease-relevant phenotypes in murine models for progressive axonopathies. These include knockouts of two genes that have been shown to cause pure dominant HSP by haploinsufficiency: for both
SPAST (spastin protein, SPG4) and
REEP1 (REEP1 protein, SPG31) there is progressive gait impairment, the severity of which negatively correlates with the amount of wild-type protein available [
26,
41,
42]. While reducing the amounts of either spastin or REEP1 is thus pathogenic, this does not seem to apply to strumpellin/SHRC.
Bona fide
KIAA0196 knockout mutations have thus far not been found in human HSP patients. There are, however, only few pertinent screens reported, and tools for detecting deleted alleles have been lacking. A second part of our study thus addressed the potential presence of inactivating large deletions. The application of a homemade MLPA assay to 240 HSP index patients failed to reveal this class of mutations. Again, the comparison to the haploinsufficiency HSP genes
SPAST and
REEP1 is of interest, as both their mutation and variation spectra turn out to be very different from those of
KIAA0196 (Table
2). The genetic data available therefore further indicate that strumpellin knockout alleles are well tolerated when present heterozygously. The generalization of this, i.e. that heterozygous absence of any SHRC member is tolerated, is suggested by the fact that the DECIPHER database [
http://decipher.sanger.ac.uk/] lists several heterozygous whole gene deletions for each SHRC member, but none of the corresponding cases is tagged with a clinical movement phenotype. Taken together, our
in vivo and genetic data as well as currently available database entries do not support a relevance of haploinsufficiency for SPG8.
The major alternative to haploinsufficiency in dominantly inherited disease is a dominant negative effect of mutant on wild-type protein, thereby effectively leading to complete loss-of-function. However, the embryonic lethality of our homozygous knockout mice, which parallels early embryonic death of homozygous
Wash1 knockouts [
17,
19], is not compatible with complete loss-of-function in otherwise healthy HSP patients. Against this background, we note a recent report of a recessive
KIAA0196 mutation affecting the very 3′ part of the coding sequence: the mutation, which is predicted to maintain only residual activity, does not confer an upper motoneuron phenotype, but results in cardiac, cerebellar, and craniofacial abnormalities [
43]. These symptoms probably correspond to what is observed upon (incomplete) strumpellin knockdown in zebrafish [
22].
Given that neither haploinsufficiency nor a dominant negative effect seem to mediate pathogenicity of HSP-associated strumpellin alterations, what does? We propose to consider a toxic gain-of-function effect caused by the accumulation of misfolded protein. Along this line, the interaction of strumpellin with the aggregation-prone VCP protein, and its presence in several neurodegeneration-associated intracellular deposits is noteworthy [
22]. The fact that no aggregation has been observed upon overexpression of mutant strumpellin [
23] may simply reflect the large differences in the time scale (several hours in cell culture vs. early adulthood clinical consequences in patients). Fully resolving the mutational mechanism in SPG8 will probably require appropriate knockin
in vivo models.
Acknowledgement
We thank Heike Kiesewetter, Annett Büschel, Kerstin Stein, Katrin Schorr, Ina Ingrisch, and Bettina Rudolph for technical assistance, and Matthew N. Seaman and C. Kaether for providing the FAM21 antiserum and the LC3-RFP plasmid, respectively. The study was funded by the German Society of Clinical Chemistry and Laboratory Medicine (NA2014 to A.J.), the IZKF of Jena University Hospital (RS2012 to A.J.), the Gentechnologiestiftung (to C.B.), the Deutsche Forschungsgemeinschaft (HU800/10-1 to C.H.), the BMBF (0315581B and 01GQ0923), the IZKF of Tübingen University (grant 1970-0-0 to R.S.), the European Union (FP7-E112009DD, FP7-E04006DD, FP7-2012-305121, PIOF-GA-2012-326681, and 01GM1408B), the program Investissements d’avenir (ANR-10-IAIHU-06), the Fondation Roger de Spoelberch (R12123DD), and the German HSP patient organisation.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AJ and MK generated and phenotyped the mouse line, and performed protein level analyses. NJ and CF performed qPCR. RS, SKli, SKle, JK, GS, LS, and AB saw and diagnosed the patients, collected genomic DNA, and provided corresponding clinical information. AJ and CB performed MLPA and ex vivo experiments. CH and CB designed the study. AJ, CH, and CB drafted the manuscript. All authors read and approved the final manuscript.