This is an observational case-control study to discover differential mRNA expression through a genome-wide transcriptome study (GWTS) using microarray technology and a real-time quantitative reverse transcription protein chain reaction (qRT-PCR) assay for validation of the results.
Subjects were selected from “CADAGENIA”, a registry in which patients with mutations in NOTCH3 have been consecutively recorded since 2017 from different parts of Spain, mostly Catalonia (Hospital Vall d’Hebron and Hospital del Mar, Barcelona). For matching purposes, control relatives (such as spouses or siblings) without a known NOTCH3 mutation were asked to enroll in the registry to avoid any potential bias due to differences between cases and controls, as well as other healthy volunteers.
Epidemiological data, blood analyses, cognition and neuroimaging profiles and skin biopsies were registered.
The inclusion criteria for cases for this differential expression study were: 1) age > 17 years, 2) having a cysteine-affecting NOTCH3 missense mutation (CNMM), and 3) having a skin biopsy available. The exclusion criteria were: 1) age < 18 years, 2) having a NOTCH3 mutation other than CNMM, and 3) not having a skin biopsy available.
The inclusion criteria for controls were: 1) age > 17 years, and 2) agreeing to have a skin biopsy. The exclusion criteria were: 1) age < 18 years, 2) having a known NOTCH3 mutation, and 3) not having a skin biopsy available.
As additional inclusion criteria for the GWTS, CADASIL patients and controls had to be matched with family members. For the qRT-PCR assay, this criterion was not needed.
Detailed clinical-epidemiological data were collected from each patient, including age; sex; vascular risk factors, such as hypertension defined as two measures on different days with blood pressure exceeding 140/90 mmHg or taking antihypertensive treatment; diabetes mellitus (DM), defined as basal glycemia in venous plasma ≥ 126 mg/dl, 2-h post-load plasma glycemia ≥ 200 mg/dl or HbA1c ≥ 6.5% or taking antidiabetic treatment; dyslipidemia; smoking habits; and type of mutation.
The cognitive profile was determined in patients with NOTCH3 mutations by means of a complete neuropsychological examination. The evaluated cognitive domains included: verbal memory, working memory, executive function, attention and information processing speed, motor speed and dexterity, and visuoconstructional skills.
For global cognition, the Montreal Cognitive Assessment (MOCA) was used as a screening test. Verbal memory was evaluated through the short-term total learning and delayed recall subtests from the Wechsler memory scale-III (WMS-III) word list. Working memory was determined by the forward and backward digits subtests from the Wechsler Adult Intelligence Scale (WAIS-III). EF was assessed by means of: phonetic (letters “P”, “M” and “R”) and semantic category (animals) verbal fluencies, the Stroop Color-Word test —number of words— and the Trail Making Test part B (TMT-B) —execution time—. Attention and IPS were evaluated through the Symbol Digit Modalities Test (SDMT), Stroop Word and Color tests —number of words— and the Trail Making Test part A (TMT-A) —execution time—. Motor speed and dexterity were rated by the Purdue Pegboard test, considering the dominant, non-dominant and both-hand trials. Visuoconstructional skills were evaluated by means of the block designs subtest from the WAIS-III.
Raw scores were adjusted into Z-scores by age and years of education following Spanish normative data (20–22). A higher adjusted Z-score indicates a better performance in all cases. We calculated cognitive domain indices by averaging the adjusted scores within each domain.
1.3. RNA EXTRACTION
A 6-mm skin punch biopsy was obtained for each participant in the study. The homogenization of the tissue was carried out with the TissueRuptor (Quiagen) and the RNA was extracted with an RNeasy® Plus Micro Kit (Quiagen), following the manufacturer’s instructions.
1.4. GENOME-WIDE TRANSCRIPTOME STUDY
From each sample, 10 ng of total RNA was used as the starting material. The quality of the isolated RNA was measured previously by capillary electrophoresis using a NanoChip (Bioanalyzer 2100, Agilent). Single-stranded cDNA suitable for labeling was generated from the total RNA using the GeneChip WT Pico Reagent Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. This kit makes it possible to generate robust expression profiles from as little as 100 pg of total RNA (10 cells). Purified sense-strand cDNA was fragmented, labeled and hybridized to the arrays using the GeneChip Hybridization, Wash and Stain Kit from the same manufacturer. After array scanning, raw data quality control was performed to check the overall performance of the processing.
For this assay, we selected the two most significantly differentially expressed mRNAs from the GWTS (p-value < 10− 3). As CADASIL is an arteriopathy that leads to brain hypoxemia, we wanted to select the genes from the GWTS that were related to neuronal ischemia in order to show a possible link and that belonged to the top fifteen most significant differentially expressed mRNAs. Therefore, we conducted a bibliographic search in PubMed with the term “(ischemi*[Title/Abstract]) AND gene[Title/Abstract]”
As previously reported (23), mRNA levels were measured by qRT-PCR using TaqMan® fluorogenic probes (see Supplemental Table I for those used in this study) on a 7500 Real-Time PCR System (Applied Biosystems, CA, USA). PPIA expression was used to normalize the results, as has been described previously (23).
qRT-PCR was performed using a standard TaqMan® PCR kit protocol consisting of 20 µl of PCR mix, including 5 µl of cDNA, 10 µl of 2x TaqMan® Universal PCR Master Mix (P/N: 4304437, Applied Biosystems, Foster city, CA, USA), 1 µl of TaqMan gene expression assay and 4 µl of water. Reactions were performed in two 384-well plates at 50ºC for two min and at 95 °C for 10 min, followed by 40 cycles at 95 °C for 15 s and 60 °C for one min. All reactions were run in triplicate and analyzed using the RQ App on Thermo Fisher Connect, following standard quality controls to assess the samples.
The results were a relative quantification (RQ) between the cycles of each sample relative to a single calibrator control sample.
1.6. STATISTICAL METHODS
Statistical and bioinformatics analyses were performed using custom scripts in R language, version 3.6.0 (R Core Team, 2019), with common Bioconductor packages. For the GWTS, after following a standard quality control, the Robust Multi-array Average (RMA) algorithm was used for pre-processing transcriptome data in order to perform background adjustment, normalization and summarization of the probe set expression values. Then, genes whose standard deviation (SD) was below the 65 percentiles of all the SD values, without a known Entrez Gene database identifier and without a valid annotation to the Gene Ontology database, were filtered out from the whole dataset and finally 6485 genes were considered for the statistical analysis. Selection of differentially expressed elements was based on a linear model analysis with empirical Bayes modification for the variance estimates. To deal with the false-discovery rate derived from multiple test comparisons, p-values were adjusted with the Benjamini and Hochberg method (24), considering genes with an adjusted p-value < 0.05 to be statistically significant.
The two most significant differentially expressed mRNAs from the GWTS (p-value < 10− 3) were evaluated in the replication cohort by qRT-PCR. Another two significant differentially expressed mRNAs from the top fifteen that were associated with ischemic neuronal death were also analyzed.
As the inclusion of outlying values could lead to erroneous interpretations (25), a box plot was performed for their identification. We used the “ggbetweenstats” function from the “ggstatsplot” package library. To know whether the outliers were statistically significant, and therefore that sample should be excluded, a Dixon's Q test was performed with the “dixon.test” function from the “outliers” package.
A p-value < 0.05 was considered statistically significant, after Bonferroni multivariable test correction, in the validation analysis.
To assess statistical significance, Fisher's Exact Test was used for categorical variables and a Mann-Whitney U test was used for numerical or ordinal variables. Pearson’s test was used to study the correlation between normal numeric variables.
1.7. EXPRESSION PROFILE
Brain expression of the mRNAs replicated in the qRT-PCR was studied using the GTEx Portal (https://gtexportal.org/home/) and expression by brain cell type was studied in the Single-nuclei Brain RNA-seq expression browser (http://ngi.pub/snuclRNA-seq/).