Introduction
Alzheimer’s disease (AD) is the most common cause of dementia [
1] and has emerged as an important societal issue and a global priority [
2]. In the absence of clinically meaningful disease-modifying treatments, the number of adults with dementia worldwide has been projected to more than triple by 2050 relative to 2010 levels [
3‐
5]. Much of the increase is projected to occur in low and middle-income countries (LMICs) [
6]. Countries in Latin America (LatAm) will experience the largest dementia increase and impact. However, genetics of early-onset AD remain markedly under-explored in LatAm populations [
7‐
9].
Genetic determinants of AD include pathogenic variants in the
amyloid precursor protein (
APP),
presenilin 1 (
PSEN1), or
presenilin 2 (
PSEN2) genes leading to dominantly inherited AD (DIAD) [
10,
11], and susceptibility loci harboring alleles that modify the risk of developing the disease [
12,
13]. Despite its low frequency relative to sporadic AD, families living with dominantly inherited Alzheimer disease (DIAD) mutations face a higher burden of disease due to the significant early onset.
To date, more than twenty-four dominantly inherited Alzheimer disease (DIAD) pathogenic variants have been reported in Latin American (LatAm) countries, including twenty-one
PSEN1, two
PSEN2, and one
APP variant, with unique characteristics including the presence of common ancestors, evidence of a high grade of admixture and ancestry background (e.g., African, Western European, Asia, and Native American) and presence in large extended families following regional distribution usually related to founder effects [
11]. Yet, the number and distribution of DIAD pathogenic variants in LatAm remain underestimated [
9,
11,
14]. In 2019, the Dominantly Inherited Alzheimer Network (DIAN), an international network designed to follow families with mutations in
APP,
PSEN1, and
PSEN2 that cause DIAD, launched the DIAN-LatAm initiative to identify new DIAD families from LatAm countries and offer research opportunities and experimental therapies to prevent, delay or treat AD. The DIAN-LatAm network includes six performance sites in Mexico (Guadalajara and Mexico City), Colombia (Medellín), Brazil (Sao Paulo), and Argentina (Buenos Aires, and Salta). Since its inception, the DIAN-LatAm and DIAN Expanded Registry (DIAN EXR;
https://dian.wustl.edu/our-research/registry/) have collaborated to identify new families with AD mutations and provide them information on research opportunities, creating a cohort-ready population that may accelerate study recruitment and completion. In a partnership between DIAN and LatAm countries (e.g., Mexico, Costa Rica, Brazil, Colombia, Chile, and Argentina, among others), we aimed to identify new pathogenic variants of familial Alzheimer’s in Latin America and refine the pathogenicity criteria. In this effort, we have screened 34 families at risk for DIAD (families with a history of early-onset dementia in two or more generations), identifying 15 families with known pathogenic variants [
11] and five with variants of unknown significance. Two families were negative for DIAD variants or other genes associated with autosomal dominant cause of dementia (e.g., MAPT, GRN, TARDBP, FUS), probands on those families were
APOE4 carriers; 11 families are undergoing additional genetic counseling for future genetic testing. In this manuscript, we prioritize novel variant descriptions over newly-identified carriers of previously identified variants. Frequency, distribution by country, and clinical characteristics of known pathogenic variants have been described elsewhere [
11].
Here, we describe five new families from LatAM (four from Brazil and one from Mexico) with multiple generations of early-onset AD (<65 years). At the time of enrollment, whether the previously uncharacterized variant is the cause of disease in these families remained unknown.
To address this gap, we took a multi-pronged approach to determine the pathogenicity of these five PSEN1 variants by performing clinical-cognitive, genetic, and cell-based analyses.
Discussion
We identified one novel frameshift deletion in
PSEN1 and four novel missense variants in
PSEN1 in LatAm families from Brazil and Mexico. These novel variants were not present in the public sequencing databases including population-matched individuals or population-matched exomes (as of Dec 2021). Four of these variants (p.Val103_Ser104delinsGly, p.Lys395Ile, p.Pro264Se, p.Ala275Thr) showed evidence of elevated levels of Aβ42 and Aβ42/Aβ40 ratios, similar to known pathogenic
PSEN1 variants and were classified as likely pathogenic. Additionally, two
PSEN1 variants (p.Val103_Ser104delinsGly and p.Lys395Ile) were shown to segregate with disease in each family, which is the gold standard for pathogenicity. We did not have sufficient evidence to classify
PSEN1 p.Ile414Thr as pathogenic. According to the ACMG/CAP guidelines [
31],
PSEN1 p.Ile414Thr would have been classified as likely pathogenic. The position Ile414 and nucleotide c.1241 are highly conserved across species and the amino-acid change is predicted to be deleterious using
in silico programs (BayesDel_addAF, DANN, DEOGEN2, EIGEN, FATHMM-MKL, LIST-S2, M-CAP, MVP, MutationAssessor, MutationTaster, PrimateAI, and SIFT). The variant is absent from the population databases, including 250 Mexican Mestizo individuals supporting that it is not a common variant in the Mexican population. However, in vitro assay results showed this variant did not alter Aβ levels in a manner consistent with pathogenic mutations, and segregation studies were not possible for this family. Future studies should explore this variant in more detail.
Pathogenic presenilin mutations are enriched in the transmembrane domain. Interestingly, the variants of unknown significance were also enriched in the transmembrane domain and are highly conserved between PSEN1 and PSEN2. The location and conservation of these variants support a prediction of likely pathogenicity based on in silico analyses alone [
23].
PSEN1 p.Val103_Ser104delinsGly, p.Pro264Ser, and p.Ala275Thr have not been previously reported. However, these sites have other mutations reported: p.V103G, p.Pro264Leu, p.Ala275Ser, and p.Ala275Val. PSEN1 p.V103G and p.Ala275Ser have a clinical presentation of AD but the pathogenicity has not yet been confirmed, whereas PSEN1 p.Pro264Leu and p.Ala275Val have been classified as pathogenic and likely pathogenic, respectively [
28,
29,
32‐
34]. Pathogenic mutations have not been described at sites impacted by PSEN1 p.Lys395Ile and p.Ile414Thr. Notably, PSEN1 p.Pro264Ser and p.Lys395Ile occur near the aspartate residues that regulate the catalytic activity of PSEN1. Pathogenic mutations PSEN1 p.Pro267Ser and p.Pr0355Ser also occur near the aspartate residues and have been shown to alter γ-secretase activity [
35‐
37]. Thus, these variants of unknown significance may impact the Aβ42/Aβ40 ratio via disrupted catalytic activity of PSEN1 and the γ-secretase complex. PSEN1 p.Ala396Thr has previously been reported to lead to increased Aβ40 and Aβ42 production without a change in the Aβ42/Aβ40 ratio in mouse neuroblastoma cells, while an independent cell-free assay suggests that PSEN1 p.Ala396Thr leads to reduced Aβ40 yielding an increase in Aβ42/Aβ40; thus, which the mechanism of the effect remains to be resolved the disruption in Aβ processing is consistent [
21,
25,
38].
PSEN1 pathogenic variant carriers had a variable clinical presentation, including memory impairment, behavioral changes, language, and extrapyramidal symptoms consistent with what we know about PSEN1 mutations. Our study found differences in age-at-symptomatic onset (range, 38–55 years), rate of progression, and clinical presentation for each variant and within families (Table
2). Heterogeneity in age-at-symptomatic onset and clinical presentation in DIAD populations has been highlighted in previous studies [
39,
40]. Differences in the mutation (missense vs. in-frame splice) or location (codon <200 vs.>200) may impact the pathologic function of PSEN1, contributing to differences in biochemica l[
41] and phenotypic expression of disease. However, as suggested by Ryan et al., family mutation alone does not explain the large proportion of the observed heterogeneity [
39,
40,
42,
43], suggesting that other environmental or genetic factors modify the effect of
PSEN1 mutations.
Several known DIAD variants from Latin America show relevant evidence of geographical cluster and funders’ effects. We could not identify such evidence in the novel variants described in this study, suggesting a possible de novo origin in these families.
Our findings are limited by (1) the absence of AD biomarkers and (2) the lack of large genetic databases with sequencing data on LatAm populations.
Due to limitations in local resources, AD biomarkers were not collected in any of the family members. The newly developed blood-based biomarker (e.g., plasma Aβ42 and Aβ42/Aβ40 ratios and plasma phosphorylated tau proteins [ptau181, ptau217, ptau205]) [
44‐
46] may represent an alternative for future studies and is especially relevant for underserved communities. The development and activation of new DIAN sites in LatAm (Argentina, Colombia, Brazil, Mexico) will facilitate future biomarkers collection in this population in DIAD families from the region. Newly identified families have been invited to join ongoing observational studies and clinical trials in DIAN sites.
To determine whether the novel variants represented rare or common polymorphisms, we examined population-based sequencing databases. This analysis is limited by the lack of diverse representation in these databases that are largely populated by individuals of European-American descent [
47], raising the possibility that these variants could be seen with greater frequency in African or Native American populations which would reduce their likelihood of being pathogenic. Population stratification is particularly important when considering rare variants. It is important to highlight that genetic testing and counseling in this study was restricted to families with two or more generations with a history of early-onset dementia. Large population-based and more inclusive studies aimed at identifying novel genetic variants in Latin America remain critical but unfortunately limited by funding. Finally, although genetic counseling was offered to family members, attempts to maintain confidentiality raised specific challenges to document family history and discuss further genetic testing among the family; therefore, segregation was not available for all the families.
Our study provides further insights into the genetics of AD in LatAm by identifying five novel mutations in the PSEN1 gene. This report is a step forward to improving the inclusion/engagement of LatAm families in research. Genetic counseling and testing are now available for at-risk, cognitively healthy individuals who may opt to learn their genetic status, decide about reproductive choices, and enroll in clinical trials. Family discovery is of great relevance for the region, as new initiatives are underway to extend clinical trials and observational studies to families living with DIAD. Broader inclusion of families with DIAD from across the world is likely to increase our understanding of the pathogenic variants that contribute to AD, with the potential to inform determinants and modifiers of AD pathogenicity, improve the care of patients and families with DIAD, and advance the development and evaluation of putative AD-modifying therapeutics.
Conclusions
Based on our findings, including clinical presentation, imaging, genetic, segregations studies, and cell-based analysis, we propose that PSEN1 p.Val103_Ser104delinsGly, p.Lys395Ile, p.Pro264Se, and p.Ala275Thr are likely pathogenic variants resulting in DIAD, whereas PSEN1 p.Ile414Thr is likely a risk factor.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.