|Home | About | Journals | Submit | Contact Us | Français|
ACE is a candidate gene for Alzheimer's disease (AD) and associations have been reported between ACE variants and plasma ACE levels, AD risk, AD age at onset of disease and cerebrospinal fluid levels of Aβ. Despite evidence that ACE can degrade Aβ, the relationships between ACE variants and the levels of different types of Aβ in the brain are not known. We have investigated the relationship between AD-associated ACE variants, for which the associations with brain activity of ACE were previously analysed, and brain homogenate levels of soluble, insoluble and oligomeric Aβ. Reported AD risk variants in the ACE indel (rs1799752) and its ‘proxy’ rs4343 were significantly associated with soluble Aβ level in AD only (p=0.001), as was rs1800764 but less so (p=0.014). In contrast, insoluble Aβ was associated with ACE indel and rs4343 variants in controls only (p < 0.01). No associations were found for oligomeric Aβ. These data indicate a complex relationship between ACE and Aβ that differs between AD and control brains.
Angiotensin converting enzyme (ACE) plays a key role in the renin-angiotensin system (RAS) pathway. ACE catalyses the formation of the vasoconstrictor angiotensin II (Ang II) from angiotensin I and is also responsible for the cleavage and inactivation of the vasodilator bradykinin, resulting in vasopression. The actions of the RAS have been extensively studied in the periphery, particularly the role of Ang II in hypertension. However, it is now recognised that nearly all organs of the body have their own local paracrine-like RAS, with organ-specific actions . The actions of Ang II within the central nervous system are of increasing interest in the context of Alzheimer's disease (AD). Ang II inhibits the release of acetylcholine (ACh) and has a pro-inflammatory effect (for review see ).
The level and activity of ACE within the cerebral cortex are generally elevated in AD [3–6]. This would be expected to increase the production of Ang II, the proinflammatory, anticholinergic and vasopressor actions of which could all exacerbate cognitive dysfunction in AD. That elevated ACE activity may be deleterious in AD is supported by recent population-based observational studies. Here RAS-acting anti-hypertensives such as ACE inhibitors (ACE-Is) and angiotensin receptor blockers (ARBs), which inhibit Ang II signalling, were associated with reductions in the incidence and rate of cognitive decline in MCI and AD [7–15].
The ACE gene (ACE) has featured for a considerably long period as one of the top susceptibility genes for AD [16–18], according to the ongoing meta-analysis database of AD candidate genes listed on Alzgene (http://www.Alzgene.org). ACE indel and related haplotypes associated with AD risk have reduced plasma ACE [19–22] whereas protective genotypes have elevated ACE . We have recently observed reduced ACE activity in CSF from AD patients  in agreement with previous work [24–25].
Several studies have shown ACE to be capable of degrading Aβ in vitro [26–30]. Variations in ACE were associated with differences in CSF Aβ level [16, 31] and some animal data suggested that Aβ level was increased by administration of ACE inhibitors [32–33]. It is not currently known whether ACE inhibitor use in humans influences Aβ accumulation. Few studies have tested whether AD-associated ‘risk’ ACE variants are associated with Aβ level in the brain [34–35] which may also shed further light on possible in vivo interactions and, to our knowledge, none has examined this association in relation to the different forms of Aβ.
Prompted by our previous observation of associations between ACE variants and ACE protein level and activity in CSF and human brain tissue , we have now explored the relationship between ACE variants and different forms of Aβ in greater detail, in mid-frontal cortex from AD and control brains.
For this study, we analysed data that we had obtained in recent studies of Aβ species in the context of AD [36–38]. Measurements of total soluble and insoluble Aβ and of oligomeric Aβ1-42 had been made on samples of mid-frontal cortex (Brodmann area 6). In samples from this region we had also previously measured ACE levels and activity. The tissue was from the South West Dementia Brain Bank, University of Bristol, and the studies had Research Ethics Committee approval. The AD cases (N=72, 64% female, 79.58 ± 9.0 years at death, 45.01 ± 24.0 hours post-mortem delay) had ‘probable’ or ‘definite’ AD according to the criteria of the Consortium to Establish a Registry for Alzheimer's Disease (CERAD) . Controls (N=33, 49% female, 79.21 ± 10.9 years at death, 45.42 ± 42.7 hours post-mortem delay) were defined by the absence of AD or other neuropathological abnormalities. The age, gender and postmortem delay were matched as closely as possible between the AD and control groups (see above). As previously described for this cohort, information on prior use of ACE-inhibitors (ACE-Is) was available for only a small number of subject, and ACE level and activity were not related to the past history of ACE-I use in these cases .
ACE genotype data and the genotypic and allelic frequencies of ACE frequencies in this cohort were previously reported . The four SNPs analysed were examined in several other studies in the context of hypertension and ACE plasma level [19, 40] and AD [16, 31, 41–43]. The most common ACE variant , the intron 16 indel (rs1799752) has close linkage disequilibrium (LD) with rs4343 (in exon 17 of ACE) (r2 =0.91) , whilst rs4291 and rs1800764 are located in the ACE promoter and 5′ untranslated region (UTR) of the gene and were previously shown to be associated with AD.
The measurement of total soluble and insoluble Aβ in the cases in this study was reported previously , as was the measurement of oligomeric Aβ1-42. Soluble and guanidine-HCl extracted (insoluble) fractions were analysed by sandwich ELISA in which monoclonal anti-Aβ (4G8 clone, raised against amino acids 18-22; Millipore, Watford, UK) was used for the capture step and biotinylated anti-human β-amyloid monoclonal antibody (10H3 clone) (Thermo Fisher Scientific, Northumberland, UK) for the detection step. Measurement of oligomeric Aβ1-42 in the soluble fraction was by sandwich ELISA using a rabbit polyclonal Aβ1-42 antibody (Millipore, Watford, UK) for the capture step and monoclonal mouse anti-oligomeric Aβ antibody (clone 7A1a, New England Rare Reagents, ME, USA) for the detection step [38, 44].
Statistical Package for Social Science software (12.0.1) was used to examine the relationship between ACE genotype and Aβ load by analysis of variance (ANOVA), adjusting for multiple testing by the Bonferroni method. Values of p < 0.05 were considered significant.
The level of total soluble Aβ was significantly influenced by ACE indel polymorphism in the combined AD and control cohorts (ANOVA p = 0.002) and in AD cases alone (ANOVA p = 0.001) but not in controls alone (Table 1). Post-hoc Bonferroni testing showed significantly higher soluble Aβ in individuals with ACE II genotype than with ID or DD genotypes in the combined and AD cohorts (Table 1). No significant difference in soluble Aβ was observed between ID and II genotypes, although there was a suggestion of a gene-dose effect. Soluble Aβ level was significantly associated with ACE SNPs rs4343 and rs1800764 in the combined and AD cohorts but not in controls alone and not with rs4291 although similar patterns were apparent (Table 2). There was no association between ACE genotype and oligomeric Aβ1-42 level in any of the cohorts (Table 1 and and22).
Total insoluble (guanidine-extracted) Aβ level did not differ according to the ACE indel (Table 1) or any of the other ACE SNPs (Table 2) in the AD or combined AD and control cohorts. In contrast, in the control cohort alone, insoluble Aβ load differed significantly according to ACE indel (ANOVA p = 0.009) but not in the allele dose dependent manner observed for soluble Aβ in AD. Instead, in controls insoluble Aβ was elevated solely in cases heterozygous (ID) for the ACE indel compared to both homozygous groups (Table 1). For ACE SNP rs4343, a proxy SNP of the indel because of its close linkage disequilibrium (LD) , insoluble Aβ was similarly significantly different (ANOVA p = 0.006, post-hoc AG>GG p=0.020; AG>AA p=0.080) following the same pattern of association as for the indel and consistent with the allelic distributions expected by their LD (Table 2).
These data reveal a complex relationship between ACE variants and the levels of different species of Aβ in AD. Individuals carrying the ACE II ‘risk’ genotype for AD  have lower plasma  and CSF  ACE protein compared to samples homozygous for the ‘protective’ D allele and heterozygotes with reportedly intermediate risk . These ACE genotypes have also previously been associated with variation in the CSF level of Aβ1-42 in AD [16, 31].
The current data show that the AD risk II genotype and related alleles in nearby SNPs are also associated with a significantly higher concentration of total soluble Aβ in AD. This may reflect more efficient cleavage by ACE of soluble than insoluble forms of Aβ and is consistent with the hypothesis that higher-risk ACE variants increase the risk of AD through reduced ACE-mediated degradation of insoluble Aβ. However, the relationship between ACE genotype and ACE activity is complicated and differs for plasma  and CSF  from that in brain tissue. ACE activity and, to a lesser extent, ACE protein concentration, are elevated not reduced in AD [3–6], and we previously found that ACE variation has no major influence on ACE activity in brain tissue .
Increases in ACE protein level and activity have been demonstrated in human SH-SY5Y neuro-blastoma cells following their exposure to oli-gomeric and fibrillar Aβ1-42 , in keeping with observations in brain tissue of a positive association between parenchymal Aβ and ACE activity . These findings suggest that ACE activity may increase as a protective response to accumulation of Aβ, possibly irrespective of the underlying genetic background, but again with differing levels of efficiency in the degradation of soluble and insoluble Aβ. The biological implications of this increase in terms of the influence on Aβ level are unclear. Studies in mouse models of AD have also yielded inconclusive data: in some administration of ACE-inhibitors did not result in an increase in Aβ load , in others [32–33] involving longer periods of treatment, increased Aβ accumulation was observed. Further studies are needed to clarify the long-term effects of ACE-inhibitors on Aβ (see  for review).
Epidemiological data indicate that elevated blood pressure in mid-life is associated with increased risk of dementia later in life (reviewed in ), although some people with prior hypertension are no longer hypertensive by the time dementia develops . Elevated ACE activity in AD, possibly in response to Aβ deposition, could have an influence on the risk of developing AD or its exacerbation over this sort of timescale, perhaps mediated via localised central hypertensive actions of Ang II. Hypertension has already been shown to alter blood-brain barrier (BBB) permeability and increase Aβ deposition in mouse models [47–48]. A partly self-perpetuating cycle could therefore be envisaged in which a slow increase in the amount of Aβ upregulates brain ACE as a means of attenuating Aβ accumulation but also causes increased production of Ang II, which could exacerbate hypertension and with that affect cerebral blood flow, increase inflammation and inhibit ACh release. We have proposed a similar model to account for the upregulation of another putative Aβ-degrading enzyme in AD, endothelin converting enzyme 2 (ECE-2) . ECE-2 is responsible for the production of another vasoconstrictor peptide ET-1 (i.e. in addition to Ang II). A further example of this sort of feedback loop is provided by the vasoactive Aβ-degrading enzyme, neprilysin, the production of which was also shown to be increased in neuronal, microglial and cerebrovascular cells by exposure to Aβ and which can produce a vasoconstrictor effect through its degradation of the vasodilator bra-dykinin  and in human APP transgenic mice .
The observed increases to the activity of brain ACE in response to Aβ may also influence the recently reported association between recurrent ICH and ACE . Recurrent ICH is often a manifestation of Aβ-related cerebral amyloid angiopathy . Elevated risk of recurrent ICH was associated with the AD-protective indel D allele. Healthy controls possessing the TT genotype of another ACE variant (rs4311) associated with recurrent ICH, had significantly higher peripheral (serum) levels of ACE . It is possible that elevation of ACE in response to escalating Aβ levels (in this case perhaps related to cerebral amyloid angiopathy) may increase the likelihood of ICH through hypertensive effects .
The present data provide further evidence implicating ACE in the pathogenesis of AD. We suggest that this association may relate to the divergent roles of ACE in degradation of Aβ mediating a short-term neuroprotective action and the production of Ang II which may lead to longer term more wide-ranging deleterious consequences (i.e. hypertension, damage to the blood-brain barrier, reduced CBF, Aβ deposition, inflammation and reduced cholinergic activity).
This study was made possible with equipment grant support from both the Bristol Research into Alzheimer's and Care of the Elderly (BRACE) and Alzheimer's Research Trust (ART) charities.