Background
Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disease characterized by progressive degeneration of the upper and lower motor neurons, leading to muscular weakness and atrophy [
1,
2]. ALS typically occurs in individuals between ages 42 and 65 and invariably leads to death due to respiratory failure three to four years after disease onset [
1,
2]. Growing evidence shows that genetic risk factors, environmental exposure, and aging are implicated in the pathogenesis of ALS [
3,
4]. Approximately 10% of ALS cases are familial ALS (FALS), and the remaining cases are classified as sporadic ALS (SALS). In recent years, with the advent of next-generation sequencing technology, more than 40 genes have been discovered to be involved in the pathogenesis of ALS [
3,
5].
Despite the increase in genetic insights and the characterization of numerous ALS-associated genes, the biological mechanisms underlying how ALS genes contribute to the ALS phenotype remain to be elucidated. Various lines of genetic, pathological and neurobiological evidence have identified axon transport deficits as essential pathogenic mechanisms for ALS [
6]. First, many ALS-associated genes (such as
PFN1,
TUBA4A,
DCTN1,
ALS2,
NEFH,
KIF5A, and
SPAST) are known to regulate cytoskeletal dynamics and function as well as intracellular transport events [
3,
6‐
9]. Second, animal studies have shown that defects in axonal transport and distal axonal damage are key pathogenic features in ALS animal models and precede the ALS-like symptoms in these animals [
10‐
15]. Third, neuropathological analysis has indicated that some ALS patients exhibit denervation and reinnervation changes in muscles but normal-appearing motor neurons, suggesting that axonal damage occurs earlier than the initial apparent symptoms in ALS, that is, before clinical manifestations of disease [
16]. Moreover, axonal cytoskeletal disorganization and defective transportation have been reported in a series of other neurodegenerative diseases, such as spastic paraplegia (SPG), Charcot-Marie-Tooth disease, and Parkinson’s disease [
17,
18], highlighting axonal transport as a potential convergent mechanism of pathological neurodegeneration.
KIF1A encodes a kinesin-3 molecular motor [
19] that transports both synaptic vesicle precursors (SVPs) carrying synaptic vesicle proteins (such as synaptophysin, VAMP2, and RAB3A) [
20‐
22] and dense core vesicles (DCVs) carrying neuropeptides and neurotrophic factors [
23‐
25].
KIF1A deficiency in mice leads to remarkable neuronal degeneration and death [
26], indicating a pathogenic association with neurodegenerative conditions. Moreover,
KIF1A heterozygous mutations have been linked to a variety of neurodegenerative and neurodevelopmental diseases [
27]. Currently, mutations in
KIF1A are listed as associated with three disorders in the OMIM database: hereditary sensory and autonomic neuropathy type 2 (HSAN2) with a recessive pattern; mental retardation type 9 (MRD9) with dominant inheritance; and SPG30 with either dominant or recessive inheritance. HSAN and SPG are known to share pathogenic genes with ALS. For example,
KIF5A and
SPG11 are characterized as common causative genes of both ALS and SPG [
28‐
31], and the
SPTLC1 gene, which causes HSAN, has recently been identified as a causative gene of juvenile ALS [
32,
33]. In addition, SPG overlaps with ALS in terms of clinical symptoms and genetic risk factors. Thus, we hypothesize that the
KIF1A-related mutations are a genetic risk of ALS pathogenesis.
In this study, we performed whole-exome sequencing (WES) in 941 patients with ALS, validated the candidate variants in an additional publicly available cohort of 4366 patients with ALS. We identified rare damage variants (RDVs) in the KIF1A gene associated with ALS, and investigated the functional effects of these variants on axonal transport.
Methods
Patients and clinical analysis
In this study, 941 patients with ALS, including 55 FALS and 886 SALS patients, were enrolled for mutation screening of candidate genes. This ALS cohort included 753 ALS patients whom we had reported on previously [
2]. Among these patients, pathogenic nucleotide repeat expansion mutations in
C9ORF72 and
ATXN2 were excluded by PCR and repeat-primer PCR analysis. A comprehensive battery of clinical data including age, sex, family history and clinical features such as age at onset (AAO), site of onset, disease duration, and ALS Functional Rating Scale–Revised score were collected from all participants. To perform burden analysis, we selected WES data from 6708 East Asian individuals without any neurological disease in the gnomAD database v2 as the control for our ALS Chinese cohort. In addition, burden analysis was also performed in the Project MinE cohort (including 4366 ALS patients and 1832 controls) [
34,
35]. In addition,
KIF1A variants were screened in an independent ALS cohort from a publicly available dataset (ALSdb, New York City, New York (URL:
http://alsdb.org) [Mar, 2022]), which consists of 3317 SALS patients who have undergone WES [
36].
WES analysis
All patients with ALS underwent WES using a previously described method [
2,
37]. Variants that failed to meet our quality control requirements (coverage depth < 10, allele balance < 0.25, and Phred quality score < 20) were excluded. RDV was included for further analysis if it was (1) located in the
KIF1A gene; (2) rare, i.e., for heterozygous variants, a minor allele frequency (MRD) less than 0.1% in the 1000 Genomes Project, Exome Aggregation Consortium (ExAC) and gnomAD; for homozygous or compound heterozygous variants, a minor allele frequency (MAF) less than 1% in the above public databases; (3) a nonsynonymous, indel, or putative splice site variant; and (4) pathogenic. For SALS, the pathogenicity was predicted by ReVe, a pathogenicity-computation method [
38], with an ReVe value more than 0.7; for FALS, the pathogenicity was identified based on co-segregation results and prediction of ReVe. Then, the variants identified by WES were validated by Sanger sequencing.
Plasmids and reagents
To construct the pGW1-KIF1A-3xFLAG plasmid,
KIF1A cDNA was obtained by RT-PCR from mRNA extracted from human embryonic stem cells and cloned into the pGW1 vector [
39] between the EcoRI and HindIII sites. The FLAG tag sequence was added to the reverse primer during PCR. To generate the pGW1-KIF1A-Venus or the pGW1-RFP-RAB3A construct, the cDNAs of KIF1A and RAB3A were obtained by RT-PCR and cloned into the pGW1 vector between the EcoRI and HindIII sites by using the Gibson assembly method. The pCMV-3xFLAG-RAB3A plasmid was constructed by PCR amplifying RAB3A from pGW1-RFP-RAB3A and subcloning the PCR fragment into the 3xFLAG-pCMV vector (Sigma, St. Louis, MO; E7658) between the EcoRI and BamHI sites. KIF1A point variants were designed using the NEBase Changer website. All clones were sequenced, and the expression of correctly sized proteins was confirmed by Western blot. FLAG-VAMP2 and RFP-synaptophysin plasmids were as described previously [
40,
41]. The primer sequences used for constructing these plasmids are listed in Additional file
1: Table S1.
The following antibodies were used: DYKDDDDK Tag (CST, Danvers, MA; 8146), GFP (Invitrogen, Waltham, MA; A11122), actin (CST, 4970S), tubulin (TransGen, Beijing, China, P10513), GAPDH (Abcam, Cambridge, United Kingdom, ab8245), RFP (Rockland, Limerick, PA; 600-401-379), Alexa Fluor 555 donkey anti-mouse IgG (H + L) (Invitrogen, A21422), goat anti-rabbit IgG-HRP (Absin, Shanghai, China, abs20040ss), and goat anti-mouse IgG-HRP (Absin, abs20039ss).
Coimmunoprecipitation
HEK293T cells were cultured in DMEM (Gibco, Waltham, MA; 11995500) supplemented with 10% fetal bovine serum (ProCell, Wuhan, China, 164210-50) and 100 U/ml penicillin–streptomycin solution (Thermo Fisher, Waltham, MA; 15140148). The cells were transfected using a GenJet instrument (SignaGen, Frederick, MD; SL100488) according to the manufacturer’s instructions. Forty-eight hours after transfection, cells were lysed in coimmunoprecipitation buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EGTA, 1% Triton X-100 and proteinase inhibitors) at 4 °C for 30 min and then centrifuged at 13,000 rpm at 4 °C for 30 min to separate the soluble and insoluble fractions. Approximately 10%–20% of the supernatant was used as the input control, and the remainder was used for immunoprecipitation. The input was mixed with 2 × Laemmli sample buffer (60 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 5% β-mercaptoethanol, 0.01% bromophenol blue) and boiled at 95 °C for 10 min. The remainder of the supernatant was incubated with anti-FLAG M2 beads (Sigma, A2220) for 4 h at 4 °C. The beads were washed with coimmunoprecipitation buffer three times, and then the protein complexes were eluted with 2% SDS.
Western blot
Cell lysates or immunoprecipitated proteins were separated by SDS-PAGE and then transferred to nitrocellulose filter membranes. After blocking with 5% nonfat milk in TBST (10 mM Tris, 70 mM NaCl, pH 7.6, 0.1% Tween-20), the membranes were incubated with primary antibodies overnight at 4 °C, washed with TBST three times for 15 min each, and then incubated with the corresponding HRP-conjugated secondary antibodies for one hour at room temperature. Quantification of blots was performed using ImageJ software (NIH, Bethesda, MA).
Neuronal culture and transfection
Primary mouse cortical neuron cultures were prepared from E17.5 C57/B6L embryos as previously described [
40]. Cortical tissues were dissected from embryos in precooled Hanks' Balanced Salt Solution (HBSS) and digested in prewarmed HBSS containing 0.025% trypsin (Gibco, 15090-046) at 37 °C for 20 min. Neurons were plated on poly-
D-lysine (Sigma, P1024)-coated coverslips (200,000 cells/ml for 12-well plates, 500,000 cells/ml for 6-well plates). The cultures were maintained at 37 °C in 5% CO
2 in neurobasal medium (Gibco, 21103049) containing 1 × B27 (Gibco, 17504044), 1 × GlutaMAX (Thermo Fisher, 35050061), and 100 U/ml penicillin–streptomycin solution (Thermo Fisher, 15140148). After seven days in vitro (DIV), the cultured neurons were transfected with plasmids using Lipofectamine 2000 Transfection Reagent (Thermo Fisher, 11668019).
Immunofluorescence and image analysis
At four days after transfection, the cultured neurons were fixed with 4% paraformaldehyde (Polysciences, Warrington, PA; 04018-1) and 4% sucrose in PBS for 15 min at room temperature, followed by three washes with PBS. The cells were then incubated with primary antibody overnight at 4 °C, washed three times in PBS, and then incubated with secondary antibodies for 1 h at room temperature. Subsequently, the cells were washed three times in PBS and mounted in VECTASHIELD Mounting Media (Vector Labs, Newark, CA; H-1000). Z-stack images were acquired with a Zeiss LSM880 confocal microscope. Quantification of colocalized RAB3A/KIF1A or VAMP2/KIF1A was performed with Imaris software (Oxford Instruments, Abingdon, United Kingdom).
Statistical analysis
Statistical analysis was performed with SPSS 25.0. Descriptive statistics (mean ± SD) were calculated for continuous variables. One-way ANOVA and Dunn’s test were used to compare continuous variable data. For comparison of categorical variables, Fisher’s exact test was applied. Association analysis of the RDVs was performed across the entire KIF1A gene and certain regions using Fisher's exact test, which was performed on an allelic basis. A P level of 0.05 was defined as the threshold of statistical significance.
Discussion
Here, we provide clinical, genetic, and cell biological data to define a new
KIF1A-associated phenotype of ALS. We identified eight variants in the
KIF1A gene from 10 of 941 patients with ALS. The frequency of candidate RDVs in
KIF1A suggested that
KIF1A might be a common causative or risk-conferring gene for ALS, especially when the variants are located in the C-terminal half of the protein, a cargo-binding region. However, the significant association of
KIF1A gene with ALS was not replicated in the Project Mine ALS cohort. This may be caused by difference in genetic background, which is well recognized in ALS. For example, the expansion mutation in
C9orf72 is the most common genetic cause of ALS in Europe, but not in Asia. In addition, mutations in
KIF5A, identified as a novel ALS gene in a European ancestry genome-wide analysis study, were rare in our ALS patients [
2,
29].
In the present study, the ALS patients with RDVs in the
KIF1A gene showed high clinical heterogeneity. The AAO ranged from 40 to 67 years, and the survival time ranged from more than 9 to more than 84 months. We noted several interesting genotype–phenotype correlation phenomena. For example, a same variant, R370C [
42] or P1587L, is associated with different clinical manifestations. We speculate that the different phenotypes in these patients with P1587L might be caused by the difference in disease duration. That is, it is possible that patient P6 might show cognitive impairment with a longer survival time. Interestingly, we found that ALS patients carrying RDVs in
KIF1A tended to show sensory disturbance. Consistent with this observation, mutations in
KIF1A have also been reported to cause HSAN, which affects sensory neurons. Etiologically, this could be explained by the various biological functions of the KIF1A protein. KIF1A is required for transportation of the neurotrophin receptor TrKA [
24]. TrkA protein is implicated in the sensory pathway [
43], and disruption of KIF1A-mediated axon transport of TrKA leads to sensory neuron loss [
23]. Therefore, we speculate that the ALS-related variants in
KIF1A might have diverse functional impacts that result in sensory disturbance in ALS patients.
Notably, we also identified one patient who carried RDV in other ALS pathogenic genes. In addition to p.R370C in KIF1A, patient P1 (A148) also carried a c.C312G; p.Y104X variant in the CHCHD10 gene. This patient showed upper limb and neck weakness and atrophy six months after onset. More detailed clinical information was not available due to the loss of follow-up. The frequency of the c.C312G; p.Y104X variant in the public databases was less than 0.01% and functional prediction suggested pathogenic or uncertain significance in ClinVar. In this patient, the RDV in KIF1A and RDVs in other causative genes may both contribute to the pathogenesis of ALS. These results are in line with the oligogenic genetic model of ALS.
Our present results, along with other findings, suggest that
KIF1A might be a shared causative gene for SPG, HSAN, and ALS. Some other common causative genes for these different types of disease have been reported. For example, pathogenic mutations in either
KIF5A or
SPG11 lead to SPG and ALS [
28‐
31], and mutations in the
SPTLC1 gene lead to HSAN and ALS [
32,
33,
44]. Interestingly, the
KIF1A-associated HSAN and MRD phenotypes have also been recognized to be mild phenotypes of complex SPG [
27]. Therefore, we hypothesize that HSAN, SPG, and ALS may represent a continuum of phenotypes associated with variants in the
KIF1A gene. As is the case for
NOTCH2NLC, C9ORF72, and
PRRT2 genes, the clinical spectrum of
KIF1A-related disease varies greatly [
37,
45,
46], giving rise to a large number of “
KIF1A-related disorders”. The high clinical heterogeneity of “
KIF1A-related disorders” may be explained by multiple etiological mechanisms, such as the multiple biological functions of the KIF1A protein, potential modifying genes, and epigenetic and environmental factors.
Consistent with previous studies of KIF5A [
29,
47], we found that most variants within the KIF1A C-terminal cargo-binding domain are related to ALS, while the majority of missense variants at the N-terminal motor domain are related to SPG. The variety of clinical phenotypes may be caused by the different changes in function caused by variants in different regions of KIF1A. Our results highlight the importance of the cargo-binding activity of the kinesin motor in ALS pathogenesis. Precise regulation of motor-cargo interactions is critical for cargo recruitment, transport efficiency and specificity, and thus for modulating the temporal and spatial distribution of the cargos. Thus far, it is still unclear how cargo recognition, binding, transport, and detachment affect ALS symptoms. In this study, the biochemical and cellular biological examinations revealed that the ALS-related
KIF1A variants modulate motor-cargo interactions and alter the colocalization of SVPs with the KIF1A motor in neurons. These findings suggest that the disease-associated variants indeed regulate the physiological role of kinesin motors. As KIF1A transports many types of cargo, including SVPs, DCVs, and lysosomes, alterations in the interaction with any cargo might contribute to ALS symptoms. Thus, further comprehensive investigations into the binding and transportation of a wide range of cargos are needed to dissect the pathogenic mechanisms of ALS-associated variants.
In this study, our results revealed that the ALS-related
KIF1A variants might promote binding to cargos, implicating a gain of function for
KIF1A variants. Previous results have identified relationships between diseases and gain-of-function mutations in genes involved in axonal transport, such as SPG30 and
KIF1A, congenital fibrosis of the extraocular muscle type 1 and
KIF21A, and autosomal dominant lower extremity-predominant spinal muscular atrophy-2A and
BICD2 [
48‐
51]. In addition, a missense mutation in the
Dync1h1 (cytoplasmic dynein heavy chain 1) gene causes progressive motor neuron degeneration in mice [
52] and enhances binding to the dynein light intermediate and intermediate chains [
53]. We show here that the ALS-related
KIF1A variants located at the C-terminal increase SVP binding and recruit more SVPs, suggesting that these variants are gain-of-function rather than loss-of-function mutations. Together, these studies indicate that the balance of tightly regulated axon transport events is essential for neuron physiology and survival. It is likely that the abnormal accumulation of SVPs or other KIF1A cargos caused by ALS-related
KIF1A missense variants is deleterious and neurotoxic, eventually resulting in neurodegeneration.
Autoinhibition is a common mechanism for regulation of Kinesin motor activity, such as KIF1A and KIF5A. Recently, ALS-related mutations in KIF5A have been reported to result in a lack of autoinhibition, leading to a toxic gain of function [
54]. KIF5A is autoinhibited by a direct intramolecular interaction between its C-terminal isoleucine-alanine-lysine (IAK) motif and the N-terminal motor domain. ALS-related mutation within the KIF5A C-terminal region disrupts the autoinhibition and leads to a constitutively activated motor. In contrast, current studies have suggested that the CC2-FHA-CC3 region serve as a regulatory region of KIF1A to modulate autoinhibition and dimerization of molecular motor [
55‐
57]. As most of the ALS-associated
KIF1A variants identified in this study locate at the C-terminal CC4-PH region that is supposed to bind cargo, we did not test the effect of these mutations on autoinhibition. However, it is possible that further genetic studies may identify more ALS-related mutations that regulate motor autoinhibition. The putative mechanism involving autoinhibition and its relation to ALS pathogenesis needs more investigations.
In this study, we tested the cargo-binding capacity of six ALS-related
KIF1A point mutations. Results indicated a variability in the effect of these mutations on cargo binding, likely because that these mutations locate at different domains of KIF1A motor. Several studies have dissected the role of different KIF1A domains in cargo binding. For example, Hummel et al. showed that CC4 and PH regions are required for KIF1A-cargo interaction [
57]; Stucchi et al. reported that the stalk domain of KIF1A (amino acids 657–1105) interacts with Liprin-a and TANC2 to help DCV trafficking [
24]. However, the exact binding regions for different cargos remain largely unclear. We found that R1100C and R1356Q, which locate between regions CC4 and PH, alter the interaction with RAB3A and VAMP2. In addition, P1587L and A1643V that locate at the PH domain also alter the binding to VAMP2, while moderately affecting binding to RAB3A (
P = 0.0570, P1587L;
P = 0.3055, A1643V). These findings support the notion that CC4 and PH, as well as the region between them, interact with cargos. Moreover, we also noted that several variants, such as R370C, are located in the CC1 region. It has been reported that the CC1 domain does not bind to cargos but regulates dimerization [
58,
59]. Consistently, we found that the R370C variant did not exhibit altered binding to SVP cargo proteins in co-immunoprecipitation experiments. The R370C variant might contribute to ALS pathogenesis through a mechanism distinct from that of other variants.
Our study had certain limitations. First, principal component analysis was not performed on the entire cohort to demonstrate the similar ethnic background between patients and controls. Second, the size of the cohort was relatively small compared to other studies of genome-wide variants in ALS. In addition, the P-values identified in this study would not reach statistical significance if in a genome-wide study. However, burden analysis in genome-wide studies did not always identify all known ALS causative or risk genes as ALS-associated genes. In a word, it requires more studies to validate the association of KIF1A with ALS. Third, due to the late onset of ALS, many family members of the patients in our study were unavailable for co-segregation analysis. Therefore, we were not able to determine whether these variants were de novo variants. Fourth, the sensory disturbance was evaluated only based on subjective description, and quantitative sensory tests were not performed; thus, this relationship of variants in KIF1A and sensory disturbance should be verified in the future. Last, in the present study, we conducted only a preliminary functional assay of cargo binding in cultured cells. Careful examination of whether these ALS-related KIF1A variants lead to neurodegenerative conditions and neuron loss in animal models will help to further clarify the pathogenic consequences and their relevance to ALS.