Two Novel Mutations in the First Transmembrane Domain of Presenilin1 Cause Young-Onset Alzheimer’s Disease

INTRODUCTION

The presenilin-1 protein (PS1) is the catalytic unit of γ-secretase implicated in the cleavage of amyloid-β protein precursor (AβPP) to produce amyloid-β (Aβ), variants of which are thought to be critical in the development of amyloid plaques and in triggering other downstream effects in Alzheimer’s disease (AD) including neurofibrillary tangle (NFT) formation and synaptic and neuronal loss. Most mutations causing familial autosomal dominant AD (ADAD) are in the PSEN1 gene and are clustered in the 9 transmembrane domains (TMDs) of the PS1 protein. It is hypothesized that these mutations cause disease by increasing the relative or absolute amount of longer cleavage products of AβPP including the 42-amino acid length version of Aβ (Aβ₄₂). At least 219 different pathogenic mutations have been described in PSEN1, some of which appear in multiple families as the result of founder effects and others as “private” mutations appearing in singlefamilies.

Small molecules have been found to modulate Aβ₄₂ and Aβ₄₀ production by binding to γ-secretase (gamma-secretase modulators). Specifically TMD-1 of PS1 might be a key regulatory site [1]. Compounds that lower or raise Aβ₄₂ production have been found to bind to TMD-1 of PS1. Single amino acid substitutions in TMD-1 have been linked to Aβ₄₂ accumulation and autosomal dominant AD (http://www.molgen.ua.ac.be/admutations/). Here we present the clinical and biomarker findings associated with two novel pathogenic mutations causing amino acid changes in TMD-1 of PS1 and the neuropathological findings associated with one.

MATERIALS AND METHODS

In the current report, we describe the clinical and imaging features of two cases of young-onset AD found to have mutations in the region of the PSEN1 gene coding for the TMD-1 domain of PS1.

Sequencing of the coding region of PSEN1 was performed using standard Sanger sequencing in a research laboratory and then confirmed independently from separate blood draws in a CLIA-approved laboratory. The population frequency of variants identified was sought by querying the 1000 Genomes, Exome Variant Server, ExAC, and AD&FTD databases. The effects of the mutations on protein function were estimated using SIFT (http://sift.jcvi.org) and PolyPhen modeling methods (http://genetics.bwh.harvard.edu/pph2/).

To quantitate production of the intracellular domains of AβPP and Notch cleavage associated with the P88L mutation, recombinant retrovirus encoding wild-type (wt) or P88L mutant PS1 was infected into #1210/DKO cells expressing gal4-fused γ-substrates and cognate luciferase reporters [2]. Using this reporter cell, we were able to analyze the generation of AβPP intracellular domain (AICD) and Notch intracellular domain (NICD), both of which are generated by γ-secretase activity, as the luciferase activity. The γ-secretase inhibitor, (3,5-Difluorophenylacetyl)-L-alanyl-L-2-phenylglycine tert-butyl ester (DAPT) [3] was synthesized as previously described [4] and dissolved in dimethyl sulfoxide. All reporter activities were standardized by averaging the luciferase activity of wt PS1 expressing cells. Expression of PS1 and reconstitution of the γ-secretase complex was confirmed by immunoblot analysis using anti-PS1NT (a kind gift from Dr. G. Thinakaran (University of Chicago)) [5]and anti-Nicastrin N1660 (Sigma), respectively. The monoclonal antibody anti-α-tubulin AA4.3 developed by Dr. C. Walsh was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD, National Institutes of Health, and maintained by The University of Iowa, Department of Biology, Iowa City, IA.

The brain autopsy, macroscopic and microscopic examinations, and immunohistochemistry on Case 2 were performed in standard fashion established in the University of California Davis Alzheimer’s Disease Center in accordance with Institutional Review Board (IRB) approved protocols, as described previously[6, 7].

Extensive tissue samples were obtained throughout the brain and examined with H&E and Bielschowsky’s silver stains, as well as PHF-1, Alz50, Aβ, and α-synuclein immunostains. The semi-quantitative assessments of AD pathology and Lewy body (LB) pathology were conducted according to consensus guidelines [7, 8]. The histopathology and neuropathology findings were reviewed by an experienced neuropathologist (L-W. J.).

RESULTS

Case 1

The index patient is a Caucasian woman with 16 years of education and unremarkable medicalhistory until she began showing symptoms consistent with myoclonus beginning in her 20 s. Working as a middle school teacher, she began having cognitive decline around age 41 characterized by word-finding difficulties, misplacing things, and getting lost in familiar locations. She was soon unable to work and between age 41 and 44 her cognition worsened and she also developed worsened myoclonus and axial rigidity with bradykinesia and shuffling gait consistent with Parkinsonism. At age 44 she had a single seizure while on buproprion which did not recur after this medication was discontinued. When seen at age 45, she had a Clinical Dementia Rating Scale score of 3 (representing severe dementia) and had excessive emotional lability consistent with pseudobulbar affect, bradykinesia, moderate hypomimia, apraxia, rigidity and spasticity with hyperreflexia of all extremities, and myoclonus of the arms. She could walk independently only with difficulty and had postural instability. She had significant depression as well as anxiety. Her Mini-Mental Status Examination (MMSE) score at that time was 2/30.

The patient’s 63-year-old mother was without cognitive or other symptoms of neurodegenerative disease but her father’s history and that of his family were unknown, there being no contact with the father since he was in his 20 s. Her full sister, who was 4 years younger, also had myoclonus followed by cognitive decline.

The index patient had an exhaustive work-up for inflammatory and other reversible causes of cognitive impairment which was unrevealing. MRI at age 44 demonstrated significant generalized cortical atrophy without microhemorrhages (Fig. 1). FDG-PET showed hypo-metabolism in the putamina and temporo-parietal regions (Fig. 2). Cerebrospinal fluid markers performed in a commerciallaboratory showed decreased Aβ₄₂ (126 pg/mL) and increased total tau (409 pg/mL) levels with a ratio (AT index = 0.17) consistent with AD (levels greater than 1.0 seen in non-AD dementias). P-tau was also elevated at 69 pg/mL (upper limit of normal 61 pg/mL) also consistent with this diagnosis.

OPEN IN VIEWER

Fig.1

Axial and coronal FLAIR MRI images of Case 1 at age 44. Note the diffuse atrophy with ventriculomegaly and relative loss of the heads of the caudate nuclei.

OPEN IN VIEWER

Fig.2

FDG-PET scan of case 1 at age 44, MMSE = 2/30. Note the profound hypometabolism particularly in the temporo-parietal cortex and putamina with relatively preserved metabolism in the occipital lobe and sensorimotor cortex.

PSEN1 sequencing was performed for the patient, her symptomatic younger sister, and her asymptomatic mother. A heterozygous nucleotide change from C to T at nucleotide 263 (NM_000021.3:c.263C>T) causing the replacement of proline by leucine at codon 88 (NP_000012.1:P88L) in the first transmembrane portion of the protein was present in the patient and her sister, but not in their mother. This variant was not found in the 1000 Genomes, Exome Variant Server, ExAC, or AD&FTD databases (accession date 11/15/2016). The effect of the mutation on protein function was estimated using SIFT and PolyPhen modeling methods. The P88L mutation in TMD-1 of PS is predicted to be ‘tolerated’ by SIFT but ‘probably damaging’ by PolyPhen-2. Recently, it was reported that partial loss-of-function of the γ-secretase activity has been associated with increased Aβ₄₂ ratio in ADAD-linked PS1 mutations [9]. Coincidentally, we had previously analyzed the effect of P88L mutation on the Aβ production as an artificial loss-of-function mutant in TMD-1 [1]. We reported that P88L mutation caused a significant increase in the Aβ₄₂ (₄₃)/Aβ₄₀ ratio that was higher than that caused by L166P mutant PS1, one of the most potent pathogenic PS1 mutations [10]. Consistent with this, we also found that the P88L mutation caused a significant reduction in the generation of AICD and NICD greater than that caused by the L166P mutant in cultured cells (Fig. 3). The P88L mutation was therefore demonstrated to negatively affect γ-secretase activity.

OPEN IN VIEWER

Fig.3

Production of the intracellular domains of AβPP and Notch by PS1 with the P88L and L166P mutations in mouse embryonic fibroblasts. #1210/DKO cells were infected transiently with recombinant retrovirus encoding the P88L and L166P PS1 mutants. Cells were then harvested and analyzed by immunoblotting using antibodies specific to nicastrin (N1660), PS1 (anti-PS1NT), or tubulin (AA4.3) (A). Relative levels of intracellular domains of AβPP (B) and Notch (C) in the infected #1210/DKO cells are measured as relative luciferase activity (n = 3; mean±standard error of the mean [SEM]). Values for 10 μM DAPT are shown as a control (white bar). Statistical analyses were performed by Tukey’s test. ^***p < 0.001 compared with wild-type PS1 (wt).

Case 2

This patient’s symptoms began at age 39 with gradually progressive amnesia severe enough to prompt a leave of absence from his job as an attorney. When first seen at age 40 his symptoms were limited to complaints of forgetfulness, particularly for recent important events. There was no clear history of dementia in the family, although a paternal grandfather died at a young age. His mother had no cognitive symptoms at the last contact at age 74. His father died of a myocardial infarction at age 56; there had been an unclear history of “work difficulties” and early retirement. A paternal aunt died at age 61 and two maternal aunts and one uncle lived through their 70 s without dementia. The patient had three living brothers; at last contact they were aged 29, 43, and 46 with no cognitive symptoms. His past medical history was remarkable for two episodes of minor head injury with loss of consciousness. His neurological examination was unremarkable. A neuropsychological examination revealed a MMSE score of 30, though with poor performance on tests of memory, abstraction, and problem solving ability on more in-depth neuropsychological assessment. Laboratory evaluation was normal, including an MRI. FDG-PET scan revealed hypometabolism in the temporal and parietal lobes.

Over the course of the next several years, his amnesia became gradually more dense and his ability to function independently became compromised. His neurological examination was remarkable only for brisk deep tendon reflexes. Four years after presentation he had a generalized seizure and was placed on phenytoin. He gradually developed more profound dementia symptoms including apathy, withdrawal, and delusions. He was treated with donepezil without benefit. His last clinic visit occurred at age 48, when he scored 17 on the MMSE. He was subsequently placed in a nursing facility, treated with neuroleptics, developed sepsis, and expired at age 49.

At autopsy, the brain weighed 1250 g with moderate frontal and temporal, and mild parietal atrophy. The hippocampus was moderately atrophic (Fig. 4). Multiple regions, notably neocortical areas, medial temporal structures, and basal nucleus show moderate to severe neuronal loss and gliosis. Virtually every surviving neuron in hippocampal CA1 contained NFT. Unusually large neuritic plaques were found in hippocampus and subiculum. The thalamus was mildly atrophic and gliotic. There was mild neuronal loss with neuromelanin-laden macrophages in the substantia nigra, but no LBs were identified. The locus ceruleus neurons contained NFTs but no LBs. The cerebellum showed a patchy loss of Purkinje cells and Bergmann glia hyperplasia. Semi-quantitative assessment of AD lesions revealed extensive neuritic plaques and NFTs (Fig. 5), reaching an “ABC” score of A3 (Thal phase 4 out of 5), B3 (Braak and Braak stage VI out of VI), and C3 (CERAD frequent neuritic plaques). According to the National Institute on Aging-Alzheimer’s Association guidelines [7], this high level of AD neuropathologic constituted an adequate explanation of the patient’s clinical dementia. There was also wide-spread Vonsattel grade 3 (out of 4) cerebral amyloid angiopathy [11]. The α-synuclein immunohistochemistry did not provide evidence for LB pathology in brainstem orneocortex.

OPEN IN VIEWER

Fig.4

Coronal section of the brain of Case 2 at the level of lateral geniculate body, which shows moderate hippocampal atrophy.

OPEN IN VIEWER

Fig.5

Case 2. A) Aβ immunohistochemical detection of amyloid deposition in left superior frontal cortex. Scale bar = 500 μm. B) PHF1 immunochemical detection of neurofibrillary pathology in left superior frontal cortex. Scale bar = 100 μm.

DNA was extracted for genetic testing from frozen brain tissue. Sequencing of the PSEN1 gene revealed a G to C transversion at nucleotide position 265 which is predicted to cause a valine to leucine substitution at codon 89 (V89L). It was not found in the ExAC, EVS, 1000 genomes or AD&FTD databases (accession date 11/15/2016) and is predicted to be deleterious by SIFT and probably damaging by PolyPhen-2.

DISCUSSION

In this report, we describe two mutations in TMD-1of the PSEN1 gene that are probably pathogenic for autosomal dominant AD of young onset by the proposed criteria of Guerreiro et al. [12]. The woman with the P88L mutation presented at age 41 with early myoclonus, followed by memory problems, a seizure, Parkinsonism, spasticity, ataxia, apraxia, and dystonia. The man with the V89L mutation began having memory problems at age 39, followed by a seizure and findings of corticospinal tract dysfunction. Neurological symptoms and signs tend to be more prominent in ADAD than sporadic AD, particularly in association with PSEN1 mutations [13] and when the age of onset is particularly young [14]. Myoclonus and seizures are common [15]. Spastic paraparesis, extrapyramidal signs, and cerebellar ataxia have also been described, though occur less frequently and in association with specific mutations. Significant phenotypic heterogeneity is observed within families but to a greater extent between families [16, 17]. This is highlighted by the similar age and clinical presentation of the two sisters harboring the P88L mutation. The V89L mutation in PSEN1 arising from a different nucleotide transversion has already been described in a Spanish family [18]. In the reported cases, the onset of symptoms was somewhat later (in the late 40 s and 50 s) and in two of the three cases symptoms suggestive of behavioral variant frontotemporal dementia were described (obsessive and disinhibited behavior). The neuropathological changes associated with the two different V89L mutations were grossly similar, with typical AD changes associated with cerebral amyloid angiopathy but without LB formation. The different phenotypes arising from the same mutation may reflect environmental differences or differences in the genetic background of these distinct families.

Though pathogenic mutations in PSEN1 have been described throughout the gene and specifically in those regions coding for the transmembrane portions of the protein, TMD-1 is particularly conserved relative to PSEN2 [19] supporting a more critical role in PS1 function. In their review of 168 patients with PSEN1 mutations, Ryan et al. found a cluster of five mutations with particularly young age of onset (less than 40 years) in the hydrophilic loop adjacent to TMD-1 [15]. Notably, we had already examined and reported the effects of the P88L mutation in PS1 on AβPP cleavage in vitro prior to this discovery of the mutation in humans [1]. We explored the role of TMD1 and found a Aβ₄₂/Aβ₄₀ ratio that was 10-fold higher than the P88C mutation, and 5-fold higher than the better known clinically aggressive L166P mutation. In addition, longer Aβ species (Aβ₄₅ and Aβ₄₆) were also overproduced. In this study, we further revealed that P88L mutation caused a significant reduction in the production of intracellular domain of AβPP as well as Notch in a cell-based reporter assay, suggesting that P88L mutation impairs both endopeptidase- and carboxypeptidase-like activities of γ-secretase [20]. This is consistent with recent findings that reduced presenilin proteolytic function leads to increased production of longer Aβ species [2, 21]. In light of our previous findings in this cell-based assay, our confirmation of a clinically aggressive disease resulting from this mutation and at the adjacent codon suggests a critical role of the TMD1 domain in γ-secretase activity.