Relationship between single nucleotide polymorphisms in the 3′UTR of amyloid precursor protein and risk of Alzheimer’s disease and its mechanism

Background and objective: Deregulation of the expression of amyloid precursor protein (APP) can lead to the development of Alzheimer’s disease (AD). Recent studies have shown that many single nucleotide polymorphisms (SNPs) in the 3′ untranslated region (UTR) of APP are associated with the development of AD. Since microRNAs (miRNAs) are involved in the regulation of APP expression, we believe that the APP 3′UTR polymorphism may affect the regulation of APP expression in miRNAs. Results: The levels of miR-101-3p, miR-153-3p, miR-144-3p, miR-381-3p, and miR-383-5p in plasma of patients with AD were significantly lower than those in the control group. The APP-534G/A site A allele was a protective factor for AD risk (adjusted odds ratio (OR) = 0.700, 95% confidence interval (95% CI): 0.573–0.840, P<0.001). The APP-369C/G site variation was not associated with AD risk. The APP-118C/A site A allele was a protective factor for AD (adjusted OR = 0.762, 95% CI: 0.639–0.897, P=0.001). The APP-534G/A site mutation affects the regulation of APP protein expression by miR-101-3p, miR-144-3p, miR-153-3p, and miR-381-3p, and the mutation of the APP-118C/A site affects miR-101-3p, miR-144-3p, miR-153-3p, and miR-383-5p regulation of APP expression. Conclusion: APP 3′UTR polymorphisms can affect the regulation of APP expression by miRNAs and thus affect the occurrence of AD.

miR-144-3p, miR-153-3p, and miR-381-3p, are all APP 3 UTR. Vilardo et al. [11] showed that miR-101-3p is a negative regulator of APP expression and affects the accumulation of amyloid β, suggesting that miR-101-3p may play a role in neuropathological disorders. Liang et al. [12] identified the binding site of miR-153-3p on APP 3 UTR by luciferase assay, and confirmed that miR-153-3p inhibited APP expression by experiments. Overexpression of miR-144-3p is ubiquitous in primate brain and AD patients, and miR-144-3p acts as a negative regulator of a disintegrin and metalloprotease 10 (ADAM10) in the pathogenesis of AD [13]. MiR-185 is capable of targeting DNA methyltransferase 1 and modulating DNA methylation in human gliomas [14], but it is not clear whether it can regulate APP expression. MiR-381-3p regulates stromal cell-derived factor-1 (SDF-1) SDF-1/CXC chemokine receptor type 4 (CXCR-4) CXCR4 signaling pathway through leucine-rich to repair of nerve damage in acute cerebral ischemia after cerebral lymphatic obstruction [15]. However, whether miR-381-3p is involved in the regulation of APP expression is currently lacking evidence. MiR-383-5p may play a key role in focal cerebral ischemia by regulating post-transcriptional levels of peroxisome proliferator-activated receptor γ (PPARγ) expression [16], whether it has function during the pathogenesis of AD is not clear.
There is increasing evidence that genetic variation in sites that bind to miRNAs may significantly increase the risk of neurodegenerative diseases. For example, Wang et al. [17] showed that the single nucleotide polymorphisms (SNPs) of the 3 UTR of the fibroblast growth factor 20 (FGF20) gene may affect its binding to miR-433 and lead to an increased risk of PD [18]. The role of miRNAs in the development of AD has been supported by a large body of research, and a variety of miRNAs have also been used as markers for the diagnosis of AD [19][20][21][22]. Simultaneously, a number of AD-specific gene mutations have been identified in the 3 UTR of APP and BACE1, for example, Brouwers et al. [23] showed that APP-369C>G and -534G>A increased the APP promoter activity by nearly two-fold. Theuns et al. [24] identified three mutations (-118C→A, -369C→G and -534G→A) in European patients with AD, showing a nearly two-fold increase in neuronal specificity of APP transcriptional activity in vitro.
In the present study, we analyzed whether the expression of APP is regulated by miRNAs such as miR-101-3p, miR-144-3p, miR-153-3p, and miR-381-3p in a Chinese Han patient population with AD, and whether the expression regulation was related to the polymorphisms at the APP 3 UTR-534G/A, -369C/G, and -118C/A loci.

General data
A total of 385 patients with AD treated in our hospital from March 2015 to January 2018 were enrolled in the present study, including 200 males and 185 females aged 60-85 years with a mean age of 73.4 + − 8.9 years. All AD patients were sporadic. The diagnostic criteria for AD were based on the diagnostic criteria for neurological communication disorders and strokes, the Association of Alzheimer's Disease and Related Diseases (NINCDS-ADRDA) [25], and the Alzheimer's Disease Diagnostic Practice Standard (OCDAD). Exclusion criteria were: a diagnosis of cognitive impairment caused by vascular dementia, Lewy body dementia, frontotemporal dementia, Parkinson's disease dementia, and other diseases excluding AD [26]. For the control group, 385 physical examinees without AD were recruited and matched to the age and sex of the patients; these participants all had a Mini-Mental State Examination (MMSE) score higher than 24, and the group consisted of 201 males and 184 females, aged 62-81, with an average age of 72.5 + − 7.8 years. All subjects in the study signed informed consent and complied with the Declaration of Helsinki.

Cell culture
Human HeLa cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (Invitrogen, Carlsbad, CA, U.S.A.). One day before transfection, HEK293 cells were inoculated into a 24-well plate with 1.0 × 10 5 cells per well, and HeLa cells were inoculated into a six-well plate with 20% confluency. According to the manufacturer's instructions, Lipofectamine 2000 (Invitrogen) was used for transfection. Table 1 Primers of the miRNA RT-qPCR   miRNAs  Primer sequence   miR-101-3p  F:5 -TGGGCTACAGTACTGTGATA-3   R:5 -TGCGTGTCGTGGAGTC-3 Table 1. The C t value of the detected sample was analyzed using EcoStudy software v4.0 (Illumina Inc, San Diego, CA, U.S.A.). All samples were standardized by U6 small nuclear RNA (U6 snRNA) as a control. The expression of miRNAs was assessed by a 2 − C T analysis.

Enzyme-linked immunosorbent assay detection
Cells were lysed 24 h after transfection, and total cellular proteins were extracted using the EpiQuik Whole Cell Extraction Kit (Epigentek, Farmingdale, NY, U.S.A.) and the expression of APP protein were detected by the Amyloid Precursor Protein Human ELISA Kit (Life Technologies Corporation, Carlsbad, CA, U.S.A., Catalog # KHB0051), with three replicate wells per sample.

Genotyping
Genomic DNA was extracted from the blood cells of all subjects, and PCR amplification was carried out using the primers in Table 2. The PCR conditions were: 95 • C, 5 min; 94 • C, 20 s; 56 • C, 30 s; 72 • C, 30 s; 35 cycles: 72 • C, 5 min; storage at 4 • C. After the PCR was completed, the PCR amplification products were purified and sequenced by Shanghai Shenggong Bioengineering Technology Co., Ltd.; the sequencing results were analyzed using Vector NTI 11.5.1 (Invitrogen).

Statistical analysis
Continuous variables were expressed by mean + S.D., and independent-sample t tests were used for statistical analyses. Categorical variables were expressed in percentage [n(%)] and analyzed using Chi-squared tests. Whether the genotype frequency was consistent with the Hardy-Weinberg equilibrium (HWE) was also assessed with Chi-squared  tests. The correlation between SNPs and AD was determined based on the distribution of allele frequencies and genetic models (additive, dominant, and recessive models), and odds ratios (OR) and 95% confidence intervals (CI) were used in an unconditional logistic regression analysis, corrected for age, sex, education, and other factors. SPSS20.0 software (IBM, Chicago, IL, U.S.A.) was used for all statistical analyses. Multivariate dimensionality reduction analysis (MDR) was used to analyze the interaction among -534G/A, -369C/G, and -118C/A loci. All tests were two-tailed, and P<0.05 indicated a statistically significant difference.

General data
Age, sex, education level, MMSE scores, and clinical dementia rating (CDR) scores of the AD group and the control group are shown in Table 3. There was no significant difference in age, sex, or education (P>0.05). The MMSE and CDR scores of the AD group were lower than the scores of the control group (P<0.05).

Bioinformatics prediction of miRNA target sites in APP 3 -UTR
To identify new miRNA target sites in human APP 3 -UTR, we used TargetScan (version 7.2) to predict the interaction between miRNAs and APP 3 UTR. The potential targets of miR-101-3p, miR-144-3p, miR-153-3p, miR-185, miR-381-3p, miR-383-5p, and other miRNAs are located in the APP 3 UTR, and the sequences that miRNAs may target are shown in Table 4.

Analysis of plasma miRNA expression levels in the AD and the control group
The levels of plasma miRNAs in 385 patients with AD and 385 healthy subjects were detected by RT-qPCR. The results show that the levels of miR-101-3p, miR-153-3p, miR-144-3p, miR-381-3p, and miR-383-5p were significantly lower in patients with AD than in the control group. Differences between groups were statistically significant P<0.001, <0.001, <0.001, <0.001, <0.001, <0.001) (Figure 1).

Association between APP gene polymorphism and AD risk
The genotypic analysis of APP gene-534G/A, -369C/G, and -118C/A loci showed that those three loci genotype frequencies are consistent with the HWE (P>0.05). Based on the GG genotype of the APP gene-534G/A locus, GA and AA genotypes were protective factors for AD (adjusted OR = 0.715, 95% CI: 0.571-0.878, P=0.001; adjusted OR =  ; with the G allele as a reference, the A allele was a protective factor for AD risk (adjusted OR = 0.700, 95% CI: 0.573-0.840, P<0.001). The risk of AD for CG, GG, and G allele carriers was not significant when compared with the CC genotype as a reference and the C allele at the -369C/G locus of the APP gene (P>0.05); in the dominant and the recessive models, AD risk was likewise not significant (P>0.05) (

Analysis of endogenous regulation of APP expression by miRNAs
The results of the enzyme-linked immunosorbent assay (ELISA) kit showed that in HeLa cells transfected with APP-534G/A locus GG, GA, and AA genotype plasmids, the highest level of APP protein was detected in the cells transfected with miR-101-3p, miR-144-3p, miR-153-3p, and miR-381-3p, followed by the cells transfected with the GA genotype, and the lowest level detected for the AA genotype (P<0.001). There was no difference in APP protein expression among genotypes carrying the APP gene-534G/A locus in the cells transfected with miR-185 and miR-383 (P>0.05) ( Figure 3A). In HeLa cells transfected with APP-369C/G CC, CG, and GG genotype plasmids, there was no difference in the expression of APP protein in the cells transfected with miR-101, miR-144, miR-153, miR-185, miR-381, and miR-383 (P>0.05). ( Figure 3B). In HeLa cells transfected with APP-118C/A CC, CA, and AA genotype plasmids, the highest level of APP protein was detected in cells transfected with miR-101-3p, miR-144-3p, miR-153-3p, and miR-381-3p, followed by the CA genotype. The lowest expression was detected for the AA genotype, and the difference was statistically significant (P<0.001). However, there was no significant difference in APP protein levels among genotypes carrying the APP-118C/A locus (P>0.05) detected in the cells transfected with miR-185 and miR-381 ( Figure 3C).

MDR analysis of interaction among SNPs at APP-534G/A, -369C/G, -118C/A loci
The interaction between SNPs at APP-534G/A, -369C/G, -118C/A loci was analyzed by MDR. The results showed that the interaction between-534G/A locus and-369C/G locus was the strongest, the interaction between-369C/G, -118C/A locus was the second, and the interaction between-534G/A locus and-118C/A locus was the smallest ( Figure  4).

Discussion
In recent years, the molecular mechanism and early diagnosis of AD occurrence and development have become the focus of research, especially the biomarkers for specific early warning signs [27,28]. Studies have found that free miRNAs in plasma can be used as biomarkers for a variety of diseases, such as Parkinson's disease [29], tumors [30],  and AD [31]. Some researchers have found that miRNAs in peripheral blood play an important role in the diagnosis of AD [19][20][21]. The current study shows that the levels of free miR-101-3p, miR-153-3p, miR-144-3p, miR-381-3p, and miR-383-5p in the plasma of patients with AD are lower than the levels in healthy subjects, making them potential molecular markers of AD. In addition, bioinformatics research based on target prediction found that all of these miRNAs target the APP 3 UTR region; whether mutations in this region affect the efficiency of miRNAs targeting regulation is a key issue worthy of study. AD is one of the most common forms of dementia, and its neuropathological markers include Aβ and τ accumulation, which manifest as senile plaques and neurofibrillary tangles or dystrophic neuritis [32]. Aβ is derived from its parental molecule Aβ precursor protein (APP), which is usually cleaved by β-secretase (BACE1) from the Aβ N-terminus to form soluble Aβ [33]. The human APP gene is located on chromosome 21, and alternative splicing produces 8-11 kinds of APP protein isoforms with different protein amino acid lengths [34]. APP overexpression is closely related to the occurrence of AD [35], and can therefore be used to reduce the production of AD by repressing the expression of APP and Aβ in sporadic AD. APP 3 UTR is a key region for miRNAs to target the regulation of APP expression. According to earlier research, genetic polymorphisms of this site are related to the efficiency of miRNAs targeting regulation [36]. Using bioinformatics to predict miRNAs targeting the expression of the APP gene, we found six kinds of miRNAs, namely miR-101-3p, miR-153-3p, miR-144-3p, miR-381-3p, and miR-383-5p. Studies have shown that miR-101-3p markedly reduces the expression of reporter genes under the control of APP 3 -UTR in human cell culture [37]. Specific target protectants that block the interaction between miR-101-3p and its functional target site within the APP 3 -UTR enhanced the level of APP expression in HeLa. Thus, endogenous miR-101-3p regulates the expression of APP in human cells through specific sites located within its 3 -UTR. The results of the present study show that miR-101-3p is down-regulated in AD, consistent with the results of Hebert et al. [38]. miR-153-3p is associated with the development of many human diseases, such as non-small cell lung cancer [39], liver cancer [40], and others. The initial detection of miR-153-3p involvement in neurodegenerative diseases has been associated with α-synuclein regulation in PD, a protein that is also involved in both familial and sporadic AD pathology. The results of the current study show that the expression of miR-153-3p is also down-regulated in patients with AD, consistent with the results of Junn et al. [41]. The miR-144-3p gene, located on chromosome 10, was originally identified as an anti-cancer gene and plays a crucial role in tumor cell proliferation, apoptosis, invasion, and metastasis [42]. Up-regulation of miR-144-3p expression is often accompanied by persistent cognitive decline [43]. miR-144-3p can also be targeted to bind to APP 3 UTR, and studies have shown that miR-144-3p overexpression markedly inhibits APP protein expression [44], consistent with the results of the current study. miR-381-3p is located in a cluster within the 14q32.31 chromosomal region. It has been shown that miR-381-3p is associated with the development of various diseases, and its down-regulation can inhibit lung cancer [45], breast cancer [46], and colorectal cancer [47]. miR-381-3p is therefore considered to be a potential tumor suppressor. miR-383-5p is also a disease-associated miRNA, and its ectopic expression has been shown to be involved in cell growth, apoptosis, and expression of apoptosis-related proteins [48]. At present, although bioinformatics are used to predict the binding of miR-381-3p and miR-383-5p to APP 3 UTR, there are only few studies on the correlations between miR-381-3p, miR-383-5p, and AD. The results of the current study show that the levels of miR-381-3p and miR-383-5p in the plasma of patients with AD are down-regulated. We believe that these decreased expression levels of miR-381-3p and miR-383-5p may lead to an increase in APP expression and soluble Aβ levels and to the onset of AD; further research into this issue is however needed.
At the same time, we also found that genetic variations of the -534G/A and -118C/A sites in the APP 3 UTR region affected the binding of miRNAs to APP 3 UTR and thus the regulation of APP expression by miRNAs. The -534G/A locus variation affects the binding of miR-101-3p, miR-144-3p, miR-153-3p, miR-381-3p, and APP 3 UTR. The -534G/A locus A allele is a protective factor for AD, but in the present study, the APP protein level in transfected A allele cells was significantly higher than the G allele. We speculate that there may be multiple miRNAs involved in the regulation of APP protein expression in vivo. The co-regulation of multiple miRNAs in individuals carrying the A allele down-regulates APP expression. When miR-101-3p, miR-144-3p, miR-153-3p, and miR-381-3p were transfected alone, APP expression was up-regulated. The specific mechanism needs further research.While the -118C/A site affects miR-101-3p, miR-144-3p, miR-153-3p, and miR-383-5p binding to APP 3 UTR. Similarly, the -118C/A site A allele is a protective factor for AD. However, APP expression in cells of the A allele was increased after transfection of miR-101-3p, miR-144-3p, miR-153-3p and miR-383-5p, respectively. We believe that there may be other miRNAs involved in the regulation of APP expression in the body, and the specific mechanism needs further study. In addition, the results of the current study suggest that the genetic variation at the -369C/G locus is unrelated to the occurrence of AD, and that the regulatory effect of miRNAs on APP protein expression levels is not related to the genetic variation at the -369C/G locus. Although Theuns et al. found that the -369C/G locus is associated with APP transcriptional activity, which is inconsistent with the results of the current study, we believe that this may be related to inconsistencies in the genetic background of different ethnic groups.
The present study has several shortcomings that need to be mentioned. First, due to limitations of the sample collection, we could not obtain AD brain samples to analyze the correlation between APP expression levels and miR-NAs and the relationships between different genetic variations of APP 3 UTR. Second, the number of rare mutation carriers is small, which may have affected the statistical analyses.