Genetic analysis of familial alzheimer’s disease, primary lateral sclerosis and paroxysmal kinesigenic dyskinesia: a tool to uncover common mechanistic points
- Autores: Jacek Szymanski
- Directores de la Tesis: Jordi Perez Tur (dir. tes.)
- Lectura: En la Universitat de València ( España ) en 2021
- Idioma: español
- Tribunal Calificador de la Tesis: José Manuel Fuentes Rodríguez (presid.), María Salome Sirerol Piquer (secret.), Angel Constantino Pérez Sempere (voc.)
- Programa de doctorado: Programa de Doctorado en Neurociencias por la Universitat de València (Estudi General)
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
Introduction Neurological diseases can have a wide spectrum of phenotypes, causing disruptions in daily life activities and, often, a progressive disease ending in loss of life. Molecular mechanisms underlying many of these diseases can have common pathways and depending on the gene in question, its dysfunction may lead to a variety of conditions.
Alzheimer's disease (AD) is the most common neurodegenerative dementia in the World with an estimate of 50 million people living with dementia worldwide. Symptoms of AD include difficulty remembering names or recent events, apathy, depression, disorientation, confusion, difficulty speaking, walking and swallowing. The canonical division of AD into early onset (EOAD) and late onset (LOAD) is set at 65 years of age. Both are characterised by intracellular hyperphosphorylated tau protein aggregates called neurofibrillary tangles (NFTs) and extracellular senile plaques composed primarily of clumps of amyloid-β (Aβ) peptide. The AD pathology starts at the entorhinal cortex, a region of the medial temporal lobe and causes neuron death leading to atrophy of brain tissue. AD is a multifactorial disease with a dichotomous pattern of inheritance with approximately 70 % of the causes being genetic and the rest environmental. Mutations in genes: APP, PSEN1 and PSEN2, are the most common causes of EOAD and allele ε4 of APOE is a well-established risk factor of LOAD. Three main pathways encapsulate most of the AD genetic risk factors: the immune response, lipid metabolism and endocytosis. Among the genes involved in immune response, CR1 has been associated with AD through genome-wide association studies (GWAS) and is an important part of the complement system, and significant for this work. A particular CR1 isoform, CR1*2, is expressed at lower levels than isoform CR1*1 and it is speculated it affects lower Aβ clearance and dysregulation of the complement system.
Primary lateral sclerosis (PLS) is considered as part of the amyotrophic lateral sclerosis (ALS) pathological spectrum. ALS, although considered a rare disease, is the most common motor neuron disease (MND). PLS is characterised by spinobulbar spasticity caused by upper motor neurons (UMNs) degeneration, while ALS is characterized by degeneration of both UMNs and lower motor neurons (LMNs) at spinal and bulbar level, causing limb paralysis, dysarthria, dysphagia and fatal respiratory failure. Clinical features of PLS include spasticity, slight weakness in the lower limbs, adult-onset, progressive course, duration of longer than 4 years and pseudobulbar symptoms. Behavioural and cognitive deficits may occur with ALS, ranging from mild, moderate to frontotemporal dementia (FTD) which was also seen in patients with PLS. The genetic basis of PLS is not well understood although mutations in genes SPG7 and TBK1 were reported in patients affected by familial PLS. Mutations in SOD1 gene, encoding an antioxidant enzyme, and in the protein encoded by C9orf72 gene are the most common causes of ALS. ALS pathogenesis may be caused by glutamate excitotoxicity, mitochondrial dysfunction, impaired structure or transport in axons and oxidative stress.
Paroxysmal kinesigenic dyskinesia (PKD), the most common paroxysmal movement disorder, is characterized by a range of involuntary movements triggered by sudden motion. With onset in childhood or adolescence its clinical features include recurrent attacks involving chorea, athetosis, dystonic postures or ballismus. Severity of these attacks typically decreases with age. Mutations in gene PRRT2 are thus far the only cause of this disorder. PRRT2 is known to interact with proteins SNAP-25, SYT1 and SYT2 in the presynaptic membrane of neurons, which are involved in signalling in nerve cells. The lack of PRRT2 often caused through nonsense-mediated mRNA decay (NMD) pathway, due to premature termination codons (PTCs) in the transcript, is said to be the common molecular mechanism involved in haploinsufficiency causing PKD.
Objectives The main objective of this work is to comprehend the etiopathogenesis of three neurological diseases affecting three distinct Spanish families and to study the genetic and molecular mechanisms affecting these diseases.
Materials and methods DNA samples were collected from members of three Spanish families: UGM037 (11 individuals) affected by AD, UGM471 (7 individuals) affected by PLS and UGM478 (7 individuals) affected by PKD. A control population of healthy individuals and individuals with AD was also available and all the samples were collected with approval of the corresponding institutional review boards of the corresponding hospitals with signed informed consent from patients. The bacterial strain used for all necessary manipulations was Escherichia coli DH5α and the human cell line was SH-SY5Y.
Whole-exome sequencing (WES) was used as the tool to identify variants within the exome of the patients for further investigation. Through base-calling and image analysis, read alignment and SNP calling the data could be transformed into a workable database of variants. Filtering and prioritization followed by Sanger sequencing validation of the results was used to identify the most interesting variants, possibly involved in causing the disease. Public databases such as Collaborative Spanish Variant Server (CSVS) or Genome Aggregation Database (gnomAD) were used for variant assessment. The variant frequency was also verified through allele-specific PCR (ASPCR) performed on the control populations.
After identifying the most interesting variants, plasmid manipulations were performed to obtain specific cDNAs, corresponding to sequences of the genes with these variants, in specific expression vectors. Site-directed mutagenesis was used for introduction of single nucleotide variants (SNV), FLAG epitope and enzyme restriction sites. Subcloning was used to transfer specific cDNAs to corresponding expression vectors. In silico analysis was performed using HOPE, I-Mutant 2.0 and ConSurf web services. That way the influence of a specific variant on protein function and stability, as well as whether a specific residue is conserved, could be established.
In the case of genes involved in AD, genotyping of APOE and CR1 was important and performed using PCRs. In the case of APOE, the PCR was followed by enzyme restriction and analysis of band distribution on an agarose gel. In the case of CR1, the PCR was followed by Sanger sequencing or analysis of presence or absence of bands on an agarose gel.
SH-SY5Y cells were maintained in a complete growth medium and Lipofectamine 2000 Transfection Reagent was used for transfecting them with plasmids corresponding to specific experiments. One experiment was to compare the mRNA and protein levels of specific genes affected by the identified variants. In another, the same comparison was made after inhibition of the NMD pathway, using NMDI14. In order to measure mRNA levels, the mRNA was extracted from transfected SH-SY5Y cells after approximately 48h of incubation and by reverse transcription cDNAs were obtained. These were used in quantitative PCR (qPCR) with specifically designed primers to amplify the reverse transcribed transcript with appropriate controls. In order to measure protein levels, the proteins were extracted from transfected SH-SY5Y cells after approximately 48h of incubation and used for Western Blot (WB) analysis. Housekeeping gene actin was used as control.
For functional analysis of ADPRH protein and its variants, the proteins were extracted from transfected SH-SY5Y cells and subjected to purification using GST-tagged protein affinity column. After sample dialysis and concentration the proteins could be used for an activity assay. This involved cholera toxin (CT) ADP-ribosylation of substrate and further ADP-ribose cleavage by ADPRH and its variants. The samples were then loaded onto an ultraperformance liquid chromatography (UPLC) column which allowed for separation, identification, and quantification of components based on data from a detector measuring absorbance at 260 nm. Also co-immunoprecipitation was used with protein extracts from SH-SY5Y cells co-transfected with ADPRH and ADPRHL1 in order to ascertain whether protein-protein interaction takes place.
Statistical analysis was performed using the student's t test and one-way or two-way ANOVA to test differences between group means, with a post hoc Tukey multiple comparisons of means test. R software was used to perform the tests.
Results and discussion The filtration, prioritization and Sanger sequencing validation identified variant rs764542666 in gene CR1 encoding a PTC c.C406T p.R136* (CR1R136*) as the likely cause of AD in family UGM037. The clinical data suggested a LOAD in the family, which would be more in line with risk factors rather than causative mutations. The family members did possess the APOE ε4 allele, however remarkably, genotyping revealed a healthy member of the family with a ε4/ε4 APOE genotype at age 88. The CR1R136* variant segregated with the disease in the pedigree, not affecting any of the non-AD members. As CR1*2 isoform of CR1, which is expressed at lower levels than CR1*1, was previously associated with higher risk of LOAD, a possible NMD of CR1R136* transcript could cause haploinsufficiency and a similar effect. Study of mRNA and protein levels revealed the CR1R136* to be expressed at lower levels than the wild-type (CR1WT). These levels would then significantly increase after treatment with NMD inhibitor, suggesting involvement of this pathway in CR1R136* transcript destruction. Genotyping of CR1*2 isoform and rs3818361 in CR1 in samples from UGM037 family, showed that none of them had CR1*2 isoform and rs3818361 did not segregate with the disease, not being present in any of the samples from patients. Both of these were genotyped as they were previously described as associated with AD. CR1*2 isoform, as mentioned before, is expressed at lower levels than CR1*1 thus possibly affecting lower Aβ clearance and dysregulation of the complement system. The rs3818361 was genotyped to verify whether it may be the real culprit behind the disease, being in linkage disequilibrium with CR1R136*. Children of the patients and the one healthy family member, were not diagnosed with AD, however their DNA samples were collected at ages between 50 and 57. They may be at risk of developing the disease at a later age, especially as all except for one of them had one copy of the APOE ε4 allele. Also three out of seven of them had the CR1R136* variant. Although in vitro overexpression of a protein in a cellular model has its limitations, the molecular mechanism behind AD in family UGM037 seems to be haploinsufficiency caused by NMD pathway destruction of CR1R136* transcript. Further functional studies on the effects of CR1R136* would be recommended and a future follow up with the younger members of the family. To the best of my knowledge, CR1R136* (rs764542666) would be the first known AD causative mutation in gene CR1.
Available clinical data supported the diagnosis of PLS in UGM471 family members. Through WES data filtration, prioritization and Sanger sequencing validation two variants were identified which could possibly cause the symptoms in the family. One was a novel variant c.G884C p.R295P in ADPRH (ADPRHR295P) and the other a previously described mutation c.T497C p.L166P in PSEN1 (rs63750265; PSEN1L166P). The latter was discovered within the UGM471 family genome recently, therefore they will be discussed chronologically, from discovery.
ADPRH is a ubiquitous protein found in cytoplasm of both mice and humans. ADPRH hydrolyses the N-glycosidic bond between arginine and ADP-ribose, cleaving ADP-ribose from substrate in the ADP-ribosylation cycle. The reaction occurs to be specific to mono-ADP-ribosylated substrates, due to the protein’s structure. ADP-ribosylation cycle is very important for cell regulation. The ADPRHR295P variant was predicted to be deleterious by SIFT and Polyphen. It was not found in any public database and Sanger sequencing confirmed its segregation with the disease in UGM471 family pedigree. However, two healthy individuals from this family also carried ADPRHR295P. ASPCR frequency assessment of ADPRHR295P in control population showed none of the 196 non-PLS individuals were carriers. The arginine in position 295 in ADPRH was found to be a conserved residue and a change to a proline could affect the function or stability of the protein significantly. Reassessment of WES data showed a paralog of ADPRH, ADPRHL1 to be affected by an ADPRHL1L294R variant, also predicted to be deleterious. This variant segregated with the disease in the family, not being present in the healthy individuals. ADPRH together with ADPRHL1 are speculated to be involved in actin filaments assembly and modulation of actin polymerisation may be involved in disruption of the nucleocytoplasmic transport which is important in ALS pathogenesis. Thus the ADPRHR295P variant may have a low penetrance, causing the disease only in certain members of the family, or may be accompanied by ADPRHL1L294R variant, to affect the patients. Co-immunoprecipitation did not show an interaction between the two proteins, however they may work in unison without a direct interaction. Activity assay for ADPRHR295P, where cholera toxin (CT) was used to ADP-ribosylate a substrate and wild-type ADPRH (ADPRHWT) and its variants (ADPRHR295P and ADPRHD55A/D56A) were used to cleave ADP-ribose, showed that ADPRHR295P had a similar activity efficiency as ADPRHWT. The ADPRHD55A/D56A variant where the essential active site residues were changed, did not show hydrolase activity. Although this assay showed no diminishing of the activity of variant ADPRHR295P, it did not take into account the possible destabilising effect of the variant. The RT-qPCR assay to measure mRNA levels of this variant showed that they did not vary from the wild-type, however, the protein levels measured with WB proved to be significantly lower than the ones for ADPRHWT or for other variants examined (ADPRHR295Q, ADPRHD55A/D56A). Only very rare truncating variant ADPRHR295* which has been previously described, proved to have very low mRNA levels and no protein could be detected. This suggest it may be affected by the NMD pathway. While it is difficult to say whether variant ADPRHR295P affects the disease in family UGM471 it is important to know its strong effect on the protein stability for the further study of little studied mono-ADP-ribosylation.
Mutation c.T497C p.L166P in PSEN1 (rs63750265) was recently discovered in family UGM471. Although mutations in PSEN1 are commonly causing EOAD, there have been reports associating PSEN1 with PLS and ALS. PSEN1A431E and PSEN1L381V are two examples of mutations associated with PLS and AD, causing an early onset at around 40 years of age. Mutation PSEN1L166P is known to cause AD and spastic paraparesis at an early age, with the first case found being a 15 year old girl. Other mutations at position L166 in PSEN1 were also found to be associated with AD (PSEN1L166V, PSEN1L166R (rs63750265), PSEN1L166H (rs63750265), and PSEN1L166del (rs63751458)). Most of them were associated with cognitive impairment, although some were associated with motor symptoms. PSEN1L166P mutation causes partial loss of γ-secretase cleavage function and increases the Aβ42/Aβ40 ratio by reducing the Aβ40 levels. Also PSEN1 functions as endoplasmic reticulum (ER) Ca2+ leak channels, and the PSEN1L166P mutation disrupts that function. Recent discovery of the PSEN1L166P mutation in the UGM471 family did not allow for a more thorough investigation of its effects, but it was confirmed to segregate with the disease in the pedigree. The aggressive nature of this mutation, the early age of onset and the motor symptoms, strongly suggest it is PSEN1L166P which causes the disease in the family. While it is not associated with ALS nor PLS, the patients may have been also affected by other environmental or genetic factors to produce this phenotype. At the same time reports of PLS patients with mutations in PSEN1 exist. Either the effect of the PSEN1 mutations is heterogeneous enough to cause these different disorders, the disorders are much more related due to the molecular mechanisms that cause them, or the effect of PSEN1 mutation is affected by interaction with other proteins. Certainly, common neurodegeneration-linked genes should be looked at when identifying a possible cause of a disease, regardless of which disease they are associated with. Patients with PLS, ALS or spastic paraparesis should be investigated for PSEN1 mutations.
In PKD, the filtration, prioritization and Sanger sequencing validation identified a novel variant c.C316T p.Q106* in PRRT2 (PRRT2Q106*) as the likely cause of the disease in family UGM478. Clinical data for the family support the diagnosis of PKD. The assessment of the WES results lead to a strong suspection of PRRT2Q106* as being the causative mutation of the disease. Mutations in PRRT2 are the only known cause of PKD thus far, and often they are truncating mutations leading to haploinsufficiency due to NMD pathway destruction of PTC carrying transcript. PRRT2Q106* was not found in any database therefore the only frequency data was obtained through our ASPCR assay for a control population of 192 samples. None of them carried this variant, yet it segregated with the disease in the pedigree, not being present in any of the non-PKD family members. In accordance with previous results, significantly lower mRNA levels for the variants with PTCs (PRRT2Q106*, PRRT2Q163*, PRRT2Q250*), in comparison with the wild-type (PRRT2WT), were found in a cellular model with PRRT2 and its variants overexpressed. PRRT2Q163* and PRRT2Q250* were chosen as controls as previously described variants in PRRT2. Inhibition of the NMD pathway by treatment of transfected SH-SY5Y cells with NMDI14, increased the mRNA levels for PRRT2Q106* and PRRT2Q163* and showed an increasing trend for PRRT2Q250*. This suggests that the NMD pathway may in fact be the culprit behind mRNA decay prompted by PTCs in the variants studied. Interestingly protein levels of the novel variant PRRT2Q106* were undetectable with WB before and after NMDI14 treatment, while the other variants studied (PRRT2Q163* and PRRT2Q250*) had significantly lower protein levels than PRRT2WT before treatment and increased levels after. This may be due to the proline-rich regions being important for protein stability, as PRRT2Q106* has a PTC before that region and PRRT2Q163* and PRRT2Q250* in the middle and after it. These results suggest the novel variant PRRT2Q106* is probably the cause of PKD in the UGM478 Spanish family. The molecular mechanisms responsible for the affliction may be the NMD pathway causing decay of the transcript leading to haploinsufficiency. Lack of PRRT2 in turn causes hyperexcitability through dysregulated neurotransmitter release and hyperactivity of Na+ channels.
Common or related pathological molecular mechanisms may affect neurological disorders, traditionally considered as unrelated, in the intricate network of the nervous system. In this work I have outlined some of such putative mechanisms. Single nucleotide variants which may affect different phenotypes through partial loss-of-function due to protein destabilisation or haploinsufficiency due to NMD.
The estimated possible number of human haploinsufficient genes is 12,443 out of approximately 22,000. While the total number of human genes is a matter of debate and further study, their estimation indicates that we can expect a great number of genes where protein level dose effect may be essential. Haploinsufficiency is important in neurological disorders. Recently C9orf72 was found to be haploinsufficient in ALS/FTD due to the GGGGCC repeat expansion.
In this work, I am postulating that CR1 and PRRT2 are haploinsufficient in Spanish families with AD and PKD respectively. While it is established that most mutations in PRRT2 lead to loss-of-function and haploinsufficiency, to my knowledge there is no such reports on CR1. Haploinsufficiency, therefore, emerges as a common factor between these and other neurological diseases. Furthermore, the molecular mechanism behind the CR1 and PRRT2-related haploinsufficiency seems to be the NMD elicited by SNVs encoding PTCs, demonstrating a common molecular mechanism in distinct neurological diseases.
Loss-of-function is strictly related to haploinsufficiency which is a dominant phenotype in organisms heterozygous for such alleles. Although variant PSEN1L166P (rs63750265) was not found to cause haploinsufficiency, it affects a partial loss of γ-secretase cleavage and ER Ca2+ leak channel function. The variant ADPRHR295P, identified in the same family, is shown to significantly destabilize the protein, drastically affecting its levels. This in turn may impede its function. Whether ADPRHR295P variant in hererozygosis is in fact deleterious remains to be seen, depending on its tolerance to decreased protein dose. However, loss-of-function, whether full or partial, encompasses the underlying molecular mechanisms of the SNVs described in this work, which contribute to independent neurological diseases.
Discussing the molecular mechanisms in AD and PLS in the two Spanish families, and the involvement of CR1R136* (rs764542666) and PSEN1L166P (rs63750265) variants in disease pathogenesis, it is important not to omit other possibly contributing factors. While CR1R136* may be a causative mutation in the family, its members had a high incidence of APOE ε4. In families with APP mutations, the incidence of APOE ε4 was related to an earlier age of onset, while the incidence of APOE ε2, with a later age of onset, with regards to APOE ε3. Interestingly, PSEN1E318G variant is related to an increased risk of AD, dependent on APOE ε4. While otherwise PSEN1E318G was considered non-pathogenic, its interaction with APOE ε4 increased Aβ deposition, causing a faster cognitive decline and neurodegeneration. Thus, while carrying CR1R136* variant may be sufficient to develop AD, it is also probable that the members of the Spanish family studied, were affected solely by the APOE ε4 risk factor or a combination of the two.
Similarly, other factors, whether environmental or genetic, may affect symptoms developed by the family with PLS. Although PSEN1L166P seems to be responsible for the phenotype experienced by the patients, their symptoms differ from the more canonical AD features related to this variant. This divergent disease expression may be due to PSEN1 gene pleiotropy, however, it may also be due to other contributing factors. Environmental factors have been found to play an important role in ALS and they cannot be disregarded in a familial disease. Here, I propose a genetic factor which may contribute to the dissimilar symptoms experienced by the members of this family. Novel variant ADPRHR295P may have an effect on the disease development, destabilising actin filaments in the presence of ADPRHL1L294R variant, prompting a phenotype closer to PLS, together with the aggressive PSEN1L166P mutation. The complexity of neurological diseases comes, in part, from a cumulative nature of defects that cause them, and thus it is always essential to search for other factor which may add to the observed phenotype. Further studies into the effects ADPRH variants may have on neurological diseases are needed as it contributes to the still poorly understood, but very important mono-ADP-ribosylation.
The general limitations of the in vitro overexpression of a protein in a cellular model apply in the entire study. Differences between the cellular model and the corresponding cells in the organism, problems with establishing appropriate microenvironment, such as interactions with other cells, or the fact that the protein is artificially overexpressed in naturally unavailable amounts. Further functional studies may be needed for all the described variants.
To conclude, variant rs764542666 in gene CR1 encoding a PTC c.C406T p.R136* is the likely cause of AD in a Spanish family UGM037, based on WES and genetic expression study. NMD pathway provoked haploinsufficiency of CR1 is the probable molecular mechanism behind the disease. Variant rs764542666 is probably the first known AD causative mutation in CR1, encouraging research into the rare truncating variants in this gene.
Mutant rs63750265 in gene PSEN1 encoding a missense mutation c.T497C p.L166P is the likely cause of PLS in a Spanish family UGM471, based on WES study, segregation analysis and previous knowledge, raising questions on pleiotropic effects of the mutation. The molecular mechanisms behind mutant rs63750265 causing the disease in family UGM471 are probably loss of γ-secretase cleavage function, increase of Aβ42/Aβ40 ratio and impairment of ER Ca2+ leak channel function. Novel variant in gene ADPRH encoding a missense variant c.G884C p.R295P strongly destabilizes the protein while not affecting its function, shedding light on the study of mono-ADP-ribosylation.
Novel variant in gene PRRT2 encoding a PTC c.C316T p.Q106* is the likely cause of PKD in a Spanish family UGM037, based on WES and genetic expression study. This work supports the hypothesis of NMD pathway provoking haploinsufficiency of PRRT2 as the molecular mechanism behind PKD.
