Variations of the angiotensin II type 1 receptor gene are associated with extreme human longevity

Abstract

Longevity phenotype in humans results from the influence of environmental and genetic factors. Few gene polymorphisms have been identified so far with a modest effect on lifespan leaving room for the search of other players in the longevity game. It has been recently demonstrated that targeted disruption of the mouse homolog of the human angiotensin II type 1 receptor (AT₁R) gene (AGTR1) translates into marked prolongation of animal lifespan (Benigni et al., J Clin Invest 119(3):524–530, 2009). Based on the above study in mice, here we sought to search for AGTR1 variations associated to reduced AT₁ receptor protein levels and to prolonged lifespan in humans. AGTR1 was sequenced in 173 Italian centenarians and 376 younger controls. A novel non-synonymous mutation was detected in a centenarian. Two polymorphisms in AGTR1 promoter, rs422858 and rs275653, in complete linkage disequilibrium, were significantly associated with the ability to attain extreme old age. We then replicated the study of rs275653 in a large independent cohort of Japanese origin (598 centenarians and semi-supercentenarians, 422 younger controls) and indeed confirmed its association with exceptional old age. In combined analyses, rs275653 was associated to extreme longevity either at recessive model (P = 0.007, odds ratio (OR) 3.57) or at genotype level (P = 0.015). Significance was maintained after correcting for confounding factors. Fluorescence activated cell sorting analysis revealed that subjects homozygous for the minor allele of rs275653 had less AT₁R-positive peripheral blood polymorphonuclear cells. Moreover, rs275653 was associated to lower blood pressure in centenarians. These findings highlight the role of AGTR1 as a possible candidate among longevity-enabling genes.

Electronic supplementary material

The online version of this article (doi:10.1007/s11357-012-9408-8) contains supplementary material, which is available to authorized users.

Introduction

The phenotype of human longevity has many diverse determinants. Non-genetic factors such as health habits, socioeconomic conditions, diet and physical activity account for an important part; however, emerging evidence suggests that genetic factors contribute in a major way to variation in human lifespan particularly in populations known for exceptional survival (Sirtori et al. 2001; Bilal et al. 2008). Genetic determinants of longevity were initially reported in small, short-lived organisms including yeast, worm and flies, but a considerable number of these genes turned out to extend lifespan in mammals as well. Among plausible candidates for human longevity, polymorphisms in apolipoprotein E gene have been found to influence lifespan in multiple populations, probably through their association with lower risk of cardiovascular disease and dementia (Corder et al. 1996). More recently, findings in two independent populations documented that polymorphisms in the human forkhead box O3A (FOXO3A) gene, which encodes a key regulator of the insulin/insulin growth factor 1 signalling pathway, were associated with the ability to attain exceptional old age (Flachsbart et al. 2009; Willcox et al. 2008). However, as can be expected for a polygenic trait like longevity, variations of FOXO3A and of other genes in the insulin/insulin growth factor 1 signalling pathway found no or only modest effect on human lifespan (Flachsbart et al. 2009; Willcox et al. 2008; Pawlikowska et al. 2009). Thus, investigations on further candidates with substantial beneficial effects on human lifespan are needed to fully disclose the determinants of human ageing.

A recent study demonstrated that mice deficient for angiotensin II type 1A receptor (AT_1AR) (AT_1AR^−/−) developed less atherosclerosis and cardiac injury while ageing and had a significant prolongation of lifespan than wild-type littermates (Benigni et al. 2009). In rodents, two isoforms of AT₁R have been identified, AT_1AR and AT_1BR, while human cells express a single AT₁ receptor (AT₁R). Angiotensin II (AngII) has been involved in organ senescence, given its capacity to robustly stimulate mitochondrial production of reactive oxygen species (ROS) that are crucially implicated in age-related organ damage including heart, kidney and brain (Wilson 1990; de Cavanagh et al. 2007) and contribute to the plethora of clinical phenotypes associating with ageing (Min et al. 2009). Over time, exuberant ROS production exceeds the antioxidant defence of the mitochondrial and the nuclear enzymes, so that accumulating damage untimely destroys cells by apoptosis or necrosis (Benigni et al. 2010). The protection observed in AT_1AR^−/− mice from vascular damage and the associated extension of lifespan were interpreted as a consequence of mitochondrial integrity provided by the expression of survival genes that limited oxidative stress (Benigni et al. 2009). These findings prompted us to investigate the possible role of variations of AGTR1 in influencing human longevity, with the hypothesis that mutations causing reduced AT₁R expression levels might associate to prolonged lifespan in humans as observed in AT_1AR-deficient mice. Our studies identified two allelic variants in AGTR1 promoter which are significantly overrepresented in Italian centenarians. This association was subsequently replicated in an independent population of Japanese centenarians and controls.

Also, we demonstrate that the allelic variants in the promoter of AGTR1 are associated with reduced AT₁R receptor expression levels in peripheral blood neutrophils. In addition, screening of the whole coding region of AGTR1 in Italian cohort disclosed in one centenarian a novel mutation leading to amino acid substitution. Finally, we disclose an association between variants in AGTR1 promoter and diastolic blood pressure in centenarians.

Methods

Study population

A cohort of 173 Italian centenarians (between 99 and 106 years of age, mean 100.9 ± 1.7 years) living in Lombardy region, northern Italy, was recruited through years 2004–2009. The gender ratio was 83 % females to 17 % males. Subjects aged more than 99 years and 6 months were considered centenarians as they were in their 100th year of life. Centenarians were recruited through PRIN 2006 study. The younger control subjects (between 49 and 78 years of age, mean 65.9 ± 9.1 years, 43 % females and 57 % males) were Italians unrelated to centenarians and recruited in the same geographical area through COFIN 2000 study.

The Japanese replication sample consisted of 589 centenarians/semi-supercentenarians ageing 99–115 years (mean 104.5 ± 3 years). The gender ratio was 84 % females to 16 % males. Centenarians and semi-supercentenarians were recruited through year 2000 until now through four series of screening involving the Tokyo metropolitan area; the centenarians address books published by Ministry of Health, Labor and Welfare from 1997 to 2002; all the Japanese nursing homes; the centenarian address lists available at local prefectures; and finally through articles published by local newspapers. Younger controls (between 69 and 72 years of age, mean 70.1 ± 0.9 years, 52 % females and 48 % males) were 422 volunteers from Itami City and Asago City, Hyogo prefecture, Japan, who participated in SONIC Study for elderly subjects to investigate factors related to successful ageing in Japan. Written informed consent was obtained from all study participants.

DNA analysis of AGTR1

In the Italian group AGTR1 exons, their flanking regions, the 5′- and 3′-UTR regions (including 250 bp of the promoter upstream 5′-UTR) were amplified from genomic DNA and sequenced in the two directions on an ABI 3730 DNA analyser (Applied Biosystems). Primers (Online Resource 1) were designed according to AGTR1 genomic reference NC_000003, built 37.1, at National Center for Biotechnology Information. Variants are referred to transcript variant 4 (NM_031850) RNA reference sequence and numbering begins with the ATG start codon. For quality control purposes, 5 % random sample was sequenced twice. Moreover, subjects carrying very rare genotypes (<5 %) were confirmed by repeating PCR reactions, followed by resequencing of both DNA strands.

For the replication study in the Japanese group, SNP genotyping was performed by TaqMan SNP assay (Applied Biosystems). The genotyping conditions for TaqMan Assay were established by repeated genotyping of 48 controls.

Analysis of AT₁R presence in peripheral leukocytes

Fresh blood samples were processed for peripheral blood mononuclear cells (PBMC) and polymorphonuclear cells (PMNs) purification as described (Noris et al. 1999; Noris et al. 1996), labelled with an anti-human AT₁R antibody (Anti AT-1, N-10, Santa Cruz) followed by incubation with fluorescein isothiocyanate goat F(ab’)2 fragment anti-rabbit IgG (eBioscience). To identify the various leukocyte subsets, cells were stained with fluorochrome-conjugated mAb antibodies against CD45, CD16, CD14, CD3, CD4 (BD Bioscience and Biolegend). The samples were analysed by fluorescence activated cell sorting (FACS) analysis using FACSAria cytometer (BD Bioscience). Results were expressed as percentage of AT₁R positive cells (percent).

Statistical analysis

Quantitative variables were expressed as mean ± SEM. ANOVA and T test were used to compare continuous variables. Categorical variables were analysed by chi-square test. If n < 5, exact Fisher test statistics was used. In the presence of a category with “0” frequency, the statistical analyses were performed using Elrm software for R (Zamar et al. 2007). Markers deviating from the Hardy–Weinberg (HW) equilibrium in the control population (P < 0.05) were excluded from the analysis. Odds ratios (OR) and 95 % confidence intervals were analysed using SISA statistics.

Haploview software v4.1 (Barrett et al. 2005) was used for blocks definition, haplotypes estimation and allelic association tests. Nominal P values were reported and were corrected for multiple testing by permutation 100,000 times. The Bonferroni correction was applied to adjust for multiple testing when appropriate. Logistic regression analysis was performed using MedCalc software. Level of significance was set at a P value <0.05 assuming a two-tailed model.

Bioinformatic analysis

The sequence variants found in AGTR1 promoter were analysed for potential transcription factors binding sites using TESS (http://www.cbil.upenn.edu/cgi-bin/tess/tess). The effect of base substitution on protein stability was examined “in silico” using SIFT (Ng and Henikoff 2002; Kumar et al. 2009) and PolyPhen-2 (Sunyaev et al. 2001).

Results

Variations in AGTR1 and longevity in Italian group

In the Italian cohort, we sequenced exons and putative functional regions of AGTR1 to determine if any gene variant, which may alter the amino acidic sequence or transcriptional regulation, is differently represented in centenarians vs. younger controls. We identified eight not previously reported variants (Fig. 1 and Online Resource 2), including a non-synonymous variant (c.893A>G) leading to amino acid substitution at position 298 (p.Asn298Ser), which was found in a centenarian and not in controls. This change occurs in the seventh transmembrane domain of AT₁R and is predicted to be damaging in silico. In addition to the rare variants, we have also observed ten known SNPs during the sequencing of AGTR1, and therefore, we have evaluated their distribution in centenarians and in controls (Fig. 1; Table 1). Of the ten SNPs observed, two were not in HW equilibrium and were removed from further analyses. The rs422858 is a dinucleotide change (AG/CC), but for practical reasons here, the major allele AG was defined as “A” and the CC allele as “C” (Erdmann et al. 1999). Single marker analysis in the eight SNPs which passed the HW equilibrium test revealed significantly different rs422858 and rs275653 genotype distributions between centenarians and controls (P = 0.004 for both markers, Table 1). Rs422858 and rs275653, located in the promoter region of AGTR1 at −174/173 bp and at −113 bp, respectively, are in complete linkage disequilibrium. Under the recessive model, the minor alleles of rs422858 and rs275653 were significantly associated to extreme old age (unadjusted P = 0.003; Bonferroni-corrected P = 0.021; OR = 4.55) (Table 1). A significant difference was observed for rs422858 and rs275653 also under the dominant and the additive models, but this difference did not pass the Bonferroni correction for the dominant model.

Fig. 1.

Genomic map and LD plot of AGTR1 (not in scale). The filled boxes are exons, numbered 1 to 4; shaded boxes non-coding regions, black box coding region. a Over the genomic AGTR1 map the position of new rare substitutions found in the Italian population is indicated. Below are the known variations, SNP numbers and their location relative to each other. b The LD block structure was calculated using the solid spine method in Haploview. The LD plot on the locus is based on the measure of D’/LOD. SNPs numbered in italic were excluded from analyses because they did not respect the HW distribution

Table 1.

Association analysis of genotypes and alleles of AGTR1 SNPs with longevity in Italian cohort

			Genotype frequencies				Recessive model (mm vs. Mm + MM)			Dominant model (mm + Mm vs. MM)			Additive model MM vs. mm			Allele analysis
SNP	M/m		MM	Mm	mm	HWE P	OR	95% CI	P nom	OR	95% CI	P nom	OR	95% CI	P bgeno	MAF	OR	95% CI	P nom	P perm	MAF
rs422858	A/C	Ce	0.607	0.335	0.058	0.551	4.55	1.5–12.5	0.003 a	1.49	1.0–2.2	0.038	4.99	1.7–14.9	0.004 a	0.225	1.55	1.1–2.1	0.007	0.044	0.225
		Co	0.697	0.290	0.013											0.158					0.158
rs275653	A/G	Ce	0.607	0.335	0.058	0.551	4.55	1.5–12.5	0.003 a	1.49	1.0–2.2	0.038	4.99	1.7–14.9	0.004 a	0.225	1.55	1.1 2.1	0.007	0.044	0.225
		Co	0.697	0.290	0.013											0.158					0.158
rs2307085	C/A	Ce	0.965	0.035	0.000	1.000	nc			4.45	1.1–18	0.031	nc		0.022	0.017	4.39	1.1–17.7	0.023	0.149	0.017
		Co	0.992	0.008	0.000											0.004					0.004
rs2272656	A/G	Ce	0.434	0.497	0.069	0.010
		Co	0.389	0.507	0.105
rs5182	C/T	Ce	0.365	0.462	0.173	0.156	0.89	0.5–1.5	0.669	1.00	0.7–1.5	0.991	0.94	0.57–1.6	0.798	0.404	0.97	0.7–1.3	0.818	1.000	0.404
		Co	0.366	0.445	0.189											0.412					0.412
rs12721225	G/T	Ce	0.994	0.006	0.000	1.000	nc			2.75	0.3–23	0.331	nc		0.331	0.003	0.37	0.0–3.0	0.332	0.929	0.003
		Co	0.984	0.016	0.000											0.008					0.008
rs5183	A/G	Ce	0.974	0.026	0.000	0.930	nc		0.340	0.99	0.6–1.8	0.984	nc		0.873	0.050	0.27	0.5–1.7	0.716	1.000	0.050
		Co	0.974	0.025	0.001											0.056					0.056
rs5186	A/C	Ce	0.491	0.396	0.112	0.934	1.72	0.9–3.2	0.081	1.07	0.7–1.5	0.714	1.7	0.9–3.3	0.599	0.311	1.15	0.9–1.5	0.302	0.875	0.311
		Co	0.508	0.424	0.068											0.280					0.280
rs5189	G/T	Ce	0.918	0.082	0.000	0.761	nc		0.490	0.78	0.7–2.5	0.467	nc		0.571	0.041	0.75	0.4–1.4	0.379	0.945	0.041
		Co	0.897	0.097	0.005											0.054					0.054
rs380400	G/A	Ce	0.717	0.252	0.031	0.001
		Co	0.705	0.235	0.059

M major allele, m minor allele, Ce centenarians, Co controls, HWE Hardy–Weinberg equilibrium, P _nom nominal P, P _geno genotype P, P _perm P values after 100,000 permutation tests. Statistically significant P values are shown in bold

OR odds ratio, CI confidence interval

^aStatistically significant after Bonferroni correction (seven tests)

^b P value of chi-square test for genotype frequencies (2DF)

Rs2307085, located on exon 1, was nominally significantly associated to longevity at both genotypic (P = 0.022) and allele levels (P = 0.023) (Table 1). Rs422858 and rs275653 together with rs2307085 were found to be in a single Haplotype block (Fig. 1, Online Resource 3). In this block, the haplotype A-A-C was associated with a reduced lifespan (nominal P = 0.007, permuted P = 0.024); consistently, the minor C-G-C haplotype was nominally associated with longevity (P = 0.028) (Online Resource 3). Stepwise logistic regression analysis showed a significantly different distribution of the rs275653 (and the linked rs422858) genotypes among centenarians and controls (P = 0.0066), while no significance was observed for all the other SNPs. After adjusting for gender, rs275653 exhibited a stronger effect than without adjustment (P = 0.00095).

Logistic regression of the allele distributions of all markers confirmed the previous findings, with only the minor allele of rs275653 being significantly associated to longer lifespan (P = 0.0092; OR = 1.56) even after correction for gender (P = 0.00083). For each SNP examined, there was no significant difference in the allelic and genotypic distribution between males and females.

AGTR1 association replicated in Japanese centenarians and combined analyses

These results prompted us to validate the association data between variants of rs275653 and longevity in a Japanese population of 589 centenarians/semi-supercentenarians and in younger individuals whose age was comparable to that of Italian controls. We chose to study a population of different ethnic origin because the Japanese population is one of the most long-lived and includes a high number of centenarians. Data of 1000 Genomes project in Japanese population indicate that rs275653 and rs422858 are in complete LD as in Italian subjects, so we chose to genotype rs275653 because of the availability of a commercial TaqMan assay. In accordance to 1000 Genomes and HapMap data, in Japanese population the minor allele of rs275653 had a lower frequency (0.084) than in Italian controls (0.158). Case–control analysis of rs275653 in Japanese cohort revealed that the GG genotype was significantly associated to longevity under the recessive and additive models (P = 0.046 and P = 0.047, respectively, Table 2). Genotype distribution evidenced a trend (P = 0.077) while no difference was observed with the dominant model and at allele level (Table 2).

Table 2.

Association of rs275653 with longevity in single populations and in the combined group

		Genotype frequency				Recessive model (GG vs. AA + AG)			Dominant model (GG + AG vs. AA)			Additive model (AA vs. GG)			Allele
Population		AA	AG	GG	P	OR	95% CI	P	OR	95% CI	P	OR	95% CI	P	MAF	OR	95% CI	P
Italians	Ce	0.607	0.335	0.058	0.004	4.55	1.5–15.5	0.003	1.49	1.0–2.2	0.038	4.99	1.7–14.9	0.001	0.225	1.55	1.1–2.1	0.007
	Co	0.697	0.290	0.013											0.158
Japanese	Ce	0.829	0.160	0.012	0.077	6.82	1.02–∞	0.046 a	1.0	0.7–1.4	1.000	6.63	1.0–∞	0.047 a	0.092	1.01	0.8–1.5	0.555
	Co	0.832	0.168	0.000											0.084
Combined	Ce	0.778	0.199	0.022	0.015	3.57	1.33–10.0	0.007	0.94	0.7–1.2	0.639	3.51	1.3–9.6	0.009	0.122	1.03	0.8–1.3	0.797
	Co	0.768	0.226	0.006											0.119

Combined group consisted of 762 centenarians and 798 controls. Statistically significant P values are shown in bold

Ce centenarians, Co controls, MAF minor allele frequency, OR odds ratio, CI confidence interval

^aElrm test

Considering that, despite the different frequencies of the minor rs275653 G allele between Italian and Japanese cohorts, an enrichment of the GG genotype was evident in both groups of centenarians as compared to controls, we performed further analyses merging the two populations. In the combined analysis, association of rs275653 with longevity was observed using the recessive (P = 0.007) and the additive models (P = 0.009) (Table 2). Also, genotypic frequencies were significantly different between centenarians and controls in the overall population (P = 0.015). After adjusting for confounding factors, namely gender, antihypertensive treatment and population of origin, logistic regression gave P = 0.00034 and P = 0.00151 for the association between longevity and rs275653 considering either genotype distribution or the recessive model, respectively (Table 3).

Table 3.

Logistic regression analysis of rs275653 and longevity in the combined group

Variables included in the analysis	Overall
	OR	95% CI	P	Model fit
rs275653—genotype	2.04	1.38–3.02	0.00034	<0.00001
rs275653—recessive model	1.96	1.29–2.96	0.00151
rs275653—dominant model	–		nr
Gender (female vs. male)	2.76	1.94–3.91	<0.00001
Antihypertensive treatment	3.62	2.59–5.06	<0.00001
Population	–		nr

OR odds ratio, CI confidence interval, nr not retained in the model

FACS analysis of AT₁R on leukocyte populations

Given that the rs275653 SNP and the linked rs422858 are localized in the promoter region, we hypothesized that they might affect AGTR1 expression. We conducted in silico analyses to evaluate whether the allelic variants of rs275653 and rs422858 were associated with the presence/absence of DNA binding factors in AGTR1 promoter. Using Tess, we found that the region including the major allele variant A of rs422858 contains an E-box element (CANNTG), which can be the target of helix-loop-helix transcription factors, including Tal1 and NF-uEA. By contrast, the C allele, which is more frequent in our centenarians, causes the loss of the E-box site. No putative binding sites were predicted for allele variants of SNP rs275653 which, however, is in complete linkage disequilibrium with rs422858. Thus, we sought to investigate whether genetic variants in AGTR1 promoter could affect AT₁R receptor presence on peripheral blood leukocytes by FACS. Due to the low frequency of subjects homozygous for the minor alleles of rs422858 and rs275653, we could identify one Italian centenarian with this genotype alive at the time of laboratory determinations, so we decided to extend the observation of AT₁R presence to younger controls. Three groups of subjects, each including one centenarian and four younger controls and differing for their genotype at rs275653 (GG, AG and AA), were evaluated. The three groups of subjects were comparable for total cholesterol (GG 227 ± 48 mg/dL; AG 198 ± 30 mg/dL; AA 185 ± 24 mg/dL) and lipids. By FACS analysis, we found specific expression of AT₁R both on PMNs (57.5 ± 14 %) and on PBMC (17.3 ± 8.9 %), which include monocytes, B and T cells. A representative flow cytometric analysis of AT₁R on PMNs and on their CD16⁺ subset is shown in Fig. 2a, b. In PMNs of subjects homozygous for the minor rs275653 variant, significantly less cells positive for AT₁R (GG: mean 42.8 ± 8.8 %, median 43.5 %) were found as compared to heterozygous (AG: mean 68.9 ± 10.9 %, median 66.8 %, P = 0.0024) or homozygous individuals for the common allele (AA: mean 60.7 ± 10.4 %, median 61.0 %, P = 0.0185) (Fig. 2c). Analysis of AT₁R cell surface presence in the PMN CD16⁺ fraction showed significantly lower percentage of AT₁R-positive cells in subjects homozygous for the minor rs275653 allele (GG: mean 68.0 ± 8.1 %, median 64.4 %) in respect to heterozygous (AG: mean 87.0 ± 6.1 %, median 87.4 %, P = 0.0030) or homozygous subjects for the common allele (AA: mean 81.1 ± 4.5 %, median 80.5 %, P = 0.0134) (Fig. 2d). AT₁R protein levels in PBMCs showed a similar trend as in PMNs; however, differences among rs275653 genotype subgroups did not reach statistical significance (AT₁R-positive cells in PBMC according to genotype GG: mean 16.9 ± 5.7 %; AG: mean 20.5 ± 7.7 %; AA: mean 19.3 ± 7.1 %). Failure to find a statistical difference in the percentage of AT₁R-positive cells among genotypes is likely attributable to the high individual variability along with the small number of subjects analysed.

Fig. 2.

AT₁R expression in PMNs and in the CD16⁺ fraction of PMNs of subjects according to rs275653 genotype. Evaluation of AT₁R expression in controls was done twice to correct for inter-assay variability, and the means of the two evaluations were considered in the statistical analysis. Centenarians were analysed once. Representative flow cytometry analysis of AT1R expression on total PMN (a) and CD16+ (b) fractions of PMNs in subjects with different genotypes for rs275653 locus: homozygous for minor (GG) and major (AA) alleles and heterozygous (AG). On the left graphs, the fluorescence of negative controls is shown. c Percentage of cells positive for AT₁R surface expression in PMNs. Overall differences were evaluated by ANOVA (P = 0.0036); individuals with GG genotype had significantly lower expression of AT₁R as compared to those with AG (P = 0.0024) and AA (P = 0.0185) genotypes. No statistically significantly difference was observed between AC and AA genotypes. d AT₁R surface expression (percentage of cells positive ) in CD16⁺ fraction of PMNs. Overall there was a statistically significant difference among groups (P _ANOVA = 0.0016). AT₁R expression in the group of subjects homozygous for the minor variant of rs275653 was lower in respect to AG (P = 0.0030) and AA (P = 0.0134) groups. Each column shows the mean ± SEM

Association of rs275653 and blood pressure

High blood pressure is associated with severe complications, including heart, kidney disease and stroke, which reduce life expectancy in humans. The renin–angiotensin system is involved in blood pressure regulation and vasoconstriction to the extent that knocking out the AT₁R gene causes hypotension in mice (Oliverio et al. 1998). Based on this observation, we evaluated whether rs275653 was associated with changes in blood pressure.

Centenarians with GG genotype of rs275653 had lower blood pressure, either systolic (SBP), diastolic (DBP) or mean arterial pressure (MAP) with respect to centenarians with AA genotype. This difference was statistically significant for DBP under recessive (P = 0.040) and additive model (P = 0.031) as well as at allele analysis (P = 0.037) (Fig. 3). A significant difference was observed for SBP and MAP at allele analysis. In control subjects, systolic blood pressure levels were slightly reduced in subjects with rs275653 GG genotype; however, the difference did not reach a statistical significance (not shown).

Fig. 3.

Blood pressure in Italian and Japanese centenarians according to rs275653 genotype. Data are represented as mean ± SEM. a Diastolic blood pressure was significantly lower in centenarians with GG genotype as compared to those AA genotype (additive model, P = 0.031) or to centenarians with AA + AG genotypes (recessive model, P = 0.040). G allele was also associated to reduced DBP (P = 0.037). SBP and MAP (b and c, respectively) were statistically different in centenarians carrying the rs275653 G allele (P = 0.046 and P = 0.042, respectively)

Discussion

Here we have identified a region in the promoter of human AGTR1 associated to extreme longevity. Studies on a test group of Italian centenarians and controls provided evidence of a significant association between SNP variants of AGTR1 and human lifespan. These findings are sustained by data obtained in the cohort of centenarians and semi-supercentenarians of Asian descent, thus ensuring a validation on a genetically diverse population.

Although studies in short-lived organisms and in human isolated populations with exceptional lifespan have clearly shown that genes play a role in longevity, the specific genetic variants involved are still largely unknown. Evidences in mammals point to a role of genes involved in the major causes of death, i.e. cardiovascular disease, cancer and brain disease. The vasoactive hormone AngII has been involved in organ senescence, and in previous studies, we demonstrated that disruption of the mouse gene encoding for AT_1AR markedly prolonged animal lifespan (Benigni et al. 2009) and protected mice from vascular and cardiac injury.

Here we demonstrate that in an Italian cohort, the C variant of rs422858 and the linked G-variant of rs275653 are overrepresented in centenarians with an OR of 4.55 according to the recessive model, in line with what should be expected for a polygenic trait like longevity. Replication of rs275653 genotyping in the Japanese cohort, characterized by a large number of centenarians and semi-supercentenarians of homogeneous ethnic origin, confirmed an enrichment of the G allele in centenarians and its association to longevity according to the recessive model. Combined analyses sustained the association at genotype level. Data of European and Japanese populations from 1000 Genomes and from HapMap databases as well as our results on the Italian group demonstrate that rs275653 is in complete linkage disequilibrium with rs422858. The absence of the rs275653 GG genotype in our Japanese controls diverges from the data of the small cohort of Japanese subjects in 1000 Genomes database (rs275653 GG genotype: one over 90 subjects). However, comparison of the genotype prevalence in Japanese population of 1000 Genomes and in our Japanese controls failed to evidence any statistical difference both at genotype and at allele level.

In silico analyses indicate that the major allele variant of rs422858 belongs to an E-box element which is lost in the presence of the minor allele. Supportive evidence to our genetic data comes from ex vivo findings indicating that the longevity-associated variants have functional consequences on the expression of the corresponding receptor protein on the surface of circulating cells. Indeed, by FACS experiments, we demonstrated that PMNs of subjects homozygous for the minor rs275653 allele (and consequently for the C variant of rs422858) had a significantly lower percentage of AT₁R-positive cells with respect to subjects heterozygous and homozygous for the major allele. A similar difference was observed when the CD16⁺ PMN fraction was analysed. Consistently, previously reported transfection experiments in adrenal cortical and in vascular smooth muscle cells (Kumar et al. 2005) documented that reporter constructs containing the A rs422858 variant had higher promoter activity than constructs with the minor C variant, suggesting that this SNP could affect the levels of AT₁R gene transcription.

Our findings are in line with the hypothesis that SNP variations in the AGTR1 promoter can influence protein levels of the corresponding receptor and that a region including or linked to the AGTR1 promoter could play a role in modulating local RAS system. Since our sequencing-based screening study was confined to the proximal portion of AGTR1 promoter and to coding regions and intron/exon boundaries, the possibility that rs275653 and the linked rs422858 and rs2307085 are in linkage disequilibrium to variants in other regulatory elements in the promoter or in introns cannot be excluded.

It is likely that the rs275653 and rs422858 minor allele variants or other linked elements in the AGTR1 promoter dampen cell surface levels of AT₁R protein in PMNs as well as in their CD16⁺ subset. AT₁R levels in several cell types including neutrophils have been shown to be affected by hypercholesterolemia or dyslipidemia (Marino et al. 2007, 2009; Guasti et al. 2008). In the present study, blood cholesterol and lipids were not different among subjects with GG, GA or the AA genotypes of rs275653, thus ruling out that changes in blood lipid profile accounts for the differences in AT₁R expression.

Local AngII can be produced by tissues and cells, including circulating cells, due to the presence of the complete enzymatic machinery required to synthesize and transform angiotensinogen (Jurewicz et al. 2007; Rasini et al. 2006). Moreover, AngII and its precursors can increase, in a dosage-dependent manner, the mitogen-stimulated T and NK cell proliferation thus playing a role in the induction of inflammatory response and oxidative stress (Jurewicz et al. 2007). In PMNs of patients with peripheral artery disease, increased levels of AT₁R have been demonstrated, suggesting a role for AT₁R in the inflammatory processes which occur in atherosclerosis (Marino et al. 2009). AT₁R stimulation of human neutrophils by AngII increases O⁻₂ and ROS production (El Bekay et al. 2003; Paragh et al. 2002) through increased cytosolic Ca⁺⁺ levels, and such response is exacerbated in diabetes (Inukai et al. 2005) and hypertension (Alba et al. 2008). Based on the above evidence, the results of the present paper might be taken to suggest that a reduced number of AT₁R on PMNs of subjects carrying the minor rs275653 and rs422858 alleles translates into a reduced capability of such cells to generate ROS, thus creating anti-inflammatory state as to slow cell senescence. Further studies are needed to confirm this assumption and assess whether the variation observed in receptor expression only takes place in PMNs or is shared by other cell types.

Allelic variations in AGTR1 promoter, including the rs422858 major allele, have been associated with hypertension in Caucasian women (Kumar et al. 2005). Very recently, SNPs in AGTR1 promoter have been associated to the risk of hypertension also in Mexicans (Martinez-Rodriguez et al. 2011) and to salt sensitivity of blood pressure in a Chinese population (Gu et al. 2010). Other studies have found an association of another SNP, the rs5186 (A1166C) in the 3′-UTR region of AGTR1, with hypertension and the metabolic syndrome (Mottl et al. 2008; Palatini et al. 2009). The A1166C SNP affects AT₁R expression (Sethupathy et al. 2007) since the A variant is the target of the microRNA hsa-miR155. In untreated hypertensive young subjects, a correlation among protein expression levels and rs5186 genotype has been recently demonstrated (Ceolotto et al. 2011). In the Italian cohort, we failed to see any association of rs5186 with human longevity, suggesting that the mechanisms by which rs422858/rs275653 or linked factors affect lifespan are not limited to its effects on blood pressure. In both Italian and Japanese centenarians, however, we observed a statistically significant decreased systolic blood pressure in homozygous for GG rs275653 genotype suggesting that this polymorphism might influence blood pressure. In control subjects, we observed a reduction of SBP in subject homozygous for GG genotype; however, this difference was not statistically significant. It is likely that the wide range of age in the control population may have masked the association between the G allele of rs275653 and lower blood pressure.

Through AGTR1 exon sequencing of the Italian centenarians and younger controls, we discovered few novel variants, all but one in the non-coding regions. Interestingly, the only novel variant leading to amino acid substitution (c.893A>G, p.Asn298Ser) was found in a centenarian. This mutation is not in LD with promoter SNPs, and it is likely a private mutation. It occurs in the highly conserved NPxxY motif of most G-protein-coupled receptors, which is involved in the initiation of signal transduction following ligand binding (Rosenbaum et al. 2009). Site-directed mutagenesis studies have previously shown that Asn298Ser replacement reduced the number of binding sites and impaired the receptor signalling in respect to the wild-type receptor, suggesting that the change found in our centenarian could affect AT₁R activity (Nikiforovich et al. 2006). Further studies are needed to confirm this finding in humans. Our results indicate that genetic AGTR1 variants leading to reduced receptor expression may contribute to prolong human lifespan. These findings are of particular relevance as they disclose AT₁R as a peculiar pharmacological target for the prevention and treatment of disabilities and diseases that occur with increasing age.

Abbreviations

AT₁R

Angiotensin II type 1 receptor

AT_1AR

Angiotensin II type 1A receptor

AT_1BR

Angiotensin II type 1B receptor

AT_1AR^−/−

Mouse deficient for angiotensin II type 1A receptor

AGTR1

Human angiotensin II type 1 receptor gene

FOXO3A

Forkhead box O3A

ROS

Reactive oxygen species

AngII

Angiotensin II

NCBI

National Center of Biotechnology Information

PBMC

Peripheral blood mononuclear cells

PMN

Polymorphonuclear cells

OR

Odds ratio

FACS

Fluorescence activated cell sorting

HW

Hardy-Weinberg