Deleterious ABCA7 mutations and transcript rescue mechanisms in early onset Alzheimer’s disease

Premature termination codon (PTC) mutations in the ATP-Binding Cassette, Sub-Family A, Member 7 gene (ABCA7) have recently been identified as intermediate-to-high penetrant risk factor for late-onset Alzheimer’s disease (LOAD). High variability, however, is observed in downstream ABCA7 mRNA and protein expression, disease penetrance, and onset age, indicative of unknown modifying factors. Here, we investigated the prevalence and disease penetrance of ABCA7 PTC mutations in a large early onset AD (EOAD)—control cohort, and examined the effect on transcript level with comprehensive third-generation long-read sequencing. We characterized the ABCA7 coding sequence with next-generation sequencing in 928 EOAD patients and 980 matched control individuals. With MetaSKAT rare variant association analysis, we observed a fivefold enrichment (p = 0.0004) of PTC mutations in EOAD patients (3%) versus controls (0.6%). Ten novel PTC mutations were only observed in patients, and PTC mutation carriers in general had an increased familial AD load. In addition, we observed nominal risk reducing trends for three common coding variants. Seven PTC mutations were further analyzed using targeted long-read cDNA sequencing on an Oxford Nanopore MinION platform. PTC-containing transcripts for each investigated PTC mutation were observed at varying proportion (5–41% of the total read count), implying incomplete nonsense-mediated mRNA decay (NMD). Furthermore, we distinguished and phased several previously unknown alternative splicing events (up to 30% of transcripts). In conjunction with PTC mutations, several of these novel ABCA7 isoforms have the potential to rescue deleterious PTC effects. In conclusion, ABCA7 PTC mutations play a substantial role in EOAD, warranting genetic screening of ABCA7 in genetically unexplained patients. Long-read cDNA sequencing revealed both varying degrees of NMD and transcript-modifying events, which may influence ABCA7 dosage, disease severity, and may create opportunities for therapeutic interventions in AD. Electronic supplementary material The online version of this article (doi:10.1007/s00401-017-1714-x) contains supplementary material, which is available to authorized users.


Supplemental texts
Neuropathological description S1: neuropathological examination in an ABCA7 PTC carrier Neuropathological examination was performed in patient EOD-P1, showing high-level AD neuropathological changes (A3B3C3) as well as a cerebral amyloid angiopathy. The severe global cerebral atrophy was most pronounced in the anterior temporal lobes, with marked neuronal loss and gliosis in the neocortex (especially the frontal, temporal and parietal cortices) and limbic system.
Amyloid pathology was frequent in the neocortex, hippocampus, amygdala, cingulum, striatum, and to a lesser extent in the thalamus, cerebellum and brainstem colliculi (CERAD neuritic plaque score of C and a Thal phase of 5, Figure 2a), in addition to abundant amyloid depositions in the wall of the cerebral and cerebellar leptomeningeal vessels ( Figure 2b). Tau-positive neurofibrillary tangles and neuropil threads were frequent in the neocortex of the frontal, temporal, parietal and occipital lobes, as well as the cingulum, hippocampus and parahippocampal region (Figure 2c), amygdala, nucleus basalis of Meynert, and to a lesser extent in the brainstem neurons, especially those in the substantia nigra, periaqueductal gray matter, locus caeruleus and raphe nuclei. No α-synuclein immunoreactivity was found.

Method S1: Western blotting
Fresh frozen brain tissue was available for carriers of ABCA7 PTC mutations c.67-1G>A and p.Leu1403fs.
Hippocampus was extracted from these two carriers as well as from fresh frozen brain tissue from three AD patients not carrying an ABCA7 PTC mutation. To quantify ABCA7 expression with Western blotting, we adapted the protocol described by Allen and colleagues [1]. Protein lysates were prepared for Western blot using a 0.1% triton lysis buffer (150 mM NaCl, 50 mM Tris pH 7.5, 0.1% Triton). The protein concentration was determined with a BCA assay (Pierce, Rockford, IL, USA), and equal amounts of protein (80µg) were separated on an NuPAGE 3%-8% Tris-Acetate gel and transferred to a PVDF membrane (Hybond P, Amersham Biosciences, Little Chalfont, UK ). Membranes were blocked in 5% milk in Phosphate-buffered saline with 0.1% Tween 20 (PBST) and probed overnight at 4°C with the ABCA7 primary antibody designed to epitope aa 2096-2146 (LS-C291064, LifeSpan BioSciences, Seattle, WA, USA; 1/500). Blots were incubated with rabbit IgG horseradish peroxidase-linked secondary antibody for 1 hour. Immunodetection was performed with the ECL-plus chemiluminescent detection system (Amersham Biosciences)). Equivalent sample loading was confirmed by probing with antiglyceraldehyde 3-phosphate dehydrogenase (GAPDH) antibody (GTX100118, Genetex, Irvine, CA, USA; 1/20.000). Protein bands were quantified using ImageQuantTL software, and relative amounts of ABCA7 protein were determined. We observed lower protein expression in PTC mutation carriers in comparison to non-carrier patients (figure S11), confirming a reduction of expression by PTC mutations. Both PTC mutation carriers had no (c.67-1G>A), or low (4%; p.Leu1403fs) potential transcript rescue mechanisms. While a clear difference is shown between carriers and non-carriers, variability in expression is also observed reflective of external modifiers of ABCA7 expression. Of note, this antibody targets the Cterminus of ABCA7, and therefore measures full-length protein. Truncated proteins may be formed, but cannot be observed.  PTC mutations studied on transcript level are denoted by their HGVS notation. Forward and reverse primers were used to amplify the cDNA region of interest. Mutations p.Trp1336* and p.Leu1403fs are in close proximity and the same, but barcoded PCR amplicon, was used. RNA source corresponds to the mutation carrying patient biomaterial. Read depth is the number of reads obtained at the mutation of interest. The amplicon size is denoted according to the canonical ABCA7 transcript (NM_019112) and may change due to alternative splicing.  Italian  m  CON  -33  72  ----EOD-C6  Italian  f  CON  -33  73  ----PTC mutations validated in early onset patients (AD) and control individuals (CON). AAD = age at death, AD = Alzheimer's disease, CON = control, DD = disease duration, FH = familial history, F = familial, PPA = primary progressive aphasia S = sporadic. Age refers to the onset age for patients and inclusion age for controls. Mutation nomenclature is provided according to Human Genome Variation Society (HGVS) on either the transcript or protein (if applicable) level. † This individual carried 2 PTC mutations that segregated on the same haplotype (see Figure S1). § For a description, see Neuropathological description S1. φ This SORL1 missense mutation of unknown importance was previously observed [3]. ϕ Note: EOD-P6.1 was not included in genetic association testing due to a pathogenic PSEN1 mutation. For individuals sharing the same PTC mutations we determined haplotypes based on SNPs and STR markers spanning ABCA7 and flanking regions as previously described [2]. Briefly, we selected 18 common SNPs (rs-numbers) that were covered within our targeting assay while passing Hardy Weinberg Equilibrium quality control (p>0.001). In addition, seven STR markers (D19S-notation) were genotyped with FAM-labeled multiplex PCR, supplemented with GeneScan™ 500 LIZ™ size standard (Thermo Fisher Scientific, Waltham, USA) after which capillary fragment analysis was performed on an ABI 3730 DNA Analyzer (Thermo Fisher Scientific, Waltham, USA). Heterozygosities of STR markers were obtained from the Marshfield Clinic database (www.marshfieldclinic.org) and MAF of SNPs was based on 1000 genomes data. ABCA7 is located in the 19p subtelomeric region, hence only two STR markers were selected on the telomeric side of ABCA7. The shared haplotype is shown per PTC variant. '-' = Samples did not share this marker. '*'This STR allele was absent in 1 individual carrying the respective mutation.  Individuals carrying a missense mutation with a Phred-scaled CADD score higher than 20 are shown. In addition, predicted effects of PolyPhen and Sift (incorporated into CADD) are shown. Age = onset age for patients (AD) and inclusion age for controls (CON). FH = Familial History. F = positive familial history (affected first degree relative). S = sporadic. † These individuals carried a "double deleterious missense" haplotype.      MinION sequencing ( Figure S3) demonstrated inframe exon 11 skipping as a potential rescue mechanism for PTC mutations located in exon 11 (e.g. p.Met370fs). The RNAseq samples depicted here, do not carry a PTC mutation in exon 11, and show a common occurrence of this novel ABCA7 isoform (red arrows). Exon 11 skipping is therefore not exclusive to exon 11 PTC mutation carriers, but can potentially rescue the negative effect of a PTC mutation in exon 11.   Using MinION cDNA sequencing we observed novel ABCA7 isoforms which comprised alternatively spliced exon 16 ( Figure S8). One particular splicing event encompassed the usage of a cryptic splice donor site 23bp upstream of the canonical splice donor site. In conjunction with a p.Glu709fs mutation (7bp deletion) this alternative splicing event can restore the transcript reading frame. Here, RNAseq analysis validates the existence of this isoform (red arrows). The upper RNAseq panel, corresponds to a p.Glu709fs carrier and depicts the limitations of RNAseq. No 7bp deletion is observed in exon 16, however, this is most likely a result of the low ABCA7 coverage in RNAseq. Furthermore it cannot be determined wether the alterenative splicing observed in this panel is in phase with the mutation. here -provide more clarity. Intron 41 retentions for instance are not spanned by the short reads, and it is therefore unknown whether these reads originate from a full or partial intron retention transcript. (e) An overview of the nucleotides (color code track) and ABCA7 gene layout (blue). The red lines denote a region that was analyzed at higher depth in panels e and f. (f) Zoomed in RNAseq reads confirm the c.5570+5G>C[C]-allele as the cause of 14bp intron retention. (g) The reference nucleotides, amino acids to which the reads in panel e align, and location of c.5570+5G>C (red arrow).

Supplemental tables
Figure S11: Hippocampal ABCA7 protein quantification in AD patients with or without an ABCA7 PTC mutation. (a) Western blotting (Method S1) was performed on hippocampal brain tissue from three AD patients without an ABCA7 PTC mutation (lanes 1, 2, and 3) and two carriers: c.67-1G>A (lane 4) and p.Leu1403fs (lane 5). A protein band is detected at approximately 234kDa, which corresponds to full length ABCA7 protein (ENSP00000414062). In addition, immunodetection of glyceraldehyde 3phosphate dehydrogenase (GAPDH) is shown below. (b) Relative quantity (RQ) of ABCA7 is normalized by GAPDH signal strength. Western blotting was repeated twice.