The association of angiotensin-converting enzyme with biomarkers for Alzheimer’s disease

Alzheimer’s disease (AD) is characterized by the accumulation of extracellular amyloid-beta (Aβ) plaques and intracellular neurofibrillary tangles (tau pathology), which are reflected by lower Aβ, and higher total tau (tau) and tau phosphorylated at threonine 181 (ptau) in cerebrospinal fluid (CSF) [1].

In the last decade, the role of the renin–angiotensin system (RAS) in the etiology of AD has received increasing attention. Inheritance of the I-allele – associated with lower plasma angiotensin-converting enzyme (ACE) levels [2] – was related to increased risk of AD [3, 4], although these findings have not been supported by recent genome-wide association studies [5, 6] and large haplotype studies [7]. Further, in vitro studies showed that ACE functions to degrade Aβ, and administration of ACE inhibitors promoted the accumulation of Aβ [8–10], while in vivo studies on various mouse models of AD showed indirect evidence that ACE can degrade Aβ (reviewed in [11]). Together, these data suggest that ACE is important for Aβ degradation, and hence low ACE activity can lead to increased Aβ-mediated neuronal damage, plaque accumulation, and risk of AD [12].

Few studies have examined the association between ACE and AD biomarkers in CSF [13, 14]. Results on the association between the I-allele, or haplotypes, of the ACE gene, which have been associated with lower ACE levels, and CSF Aβ were inconclusive [13, 14]. Further, one small-scale study found that administration of an ACE inhibitor did not influence AD biomarkers in CSF [15].

The current literature almost entirely comes from laboratory-based and genetic association studies. However, the laboratory-based studies are not comparable with the human disease state and the findings from genetic-association studies may be explained by linkage disequilibrium with the true risk factor [3]. Examining direct measures of ACE (such as serum or CSF ACE) in relation to CSF AD biomarkers might therefore be more clinically meaningful – especially since the widely prescribed antihypertensive ACE-inhibitors strongly inhibit ACE activity [16], thereby possibly not only reducing the unfavorable angiotensin (ANG)I to ANGII conversion, but also the favorable Aβ degradation.

The aim of the current study was to investigate the association of serum and CSF ACE protein levels and activity with CSF AD biomarkers – Aβ, tau and ptau – in a memory clinic cohort. It has increasingly been recognized that nearly all organs of the body have their own local paracrine-like RAS, with organ-specific functions [17]. The brain also has its own RAS system, acting largely independent of the peripheral RAS [17]. We therefore hypothesize that CSF and serum ACE reflect the activity in the brain RAS and peripheral RAS, respectively, and that CSF ACE activity and, to a lesser extent, serum ACE activity are associated with lower CSF Aβ, indicating worse Aβ-related pathology in the brain.

In the total population of 118 subjects, the mean age was 66 (standard deviation, range: 8, 50 to 79) years and 52% were female. Table 1 shows that patients in the AD group were older, more often female, and more often APOE-ϵ4 carriers than patients in the control group. They also had less hyperlipidemia and higher blood pressure levels. CSF Aβ levels were lower in AD patients, and CSF tau and ptau levels were higher in AD patients. Further, all ACE measures were lower in patients with AD compared with patients in the control group (Table 1). The Pearson correlation coefficient between CSF ACE protein level and CSF ACE activity was 0.26 (P = 0.005), between serum ACE protein level and serum ACE activity was 0.28 (P = 0.002), between CSF and serum ACE protein levels was 0.54 (P < 0.001), and between CSF and serum ACE activity levels was 0.07 (P = 0.494). The correlation between tau and Aβ was -0.43 (P < 0.001) and between ptau and Aβ was -0.41 (P = 0.0004). Further, APOE-ϵ4 carriers had significantly lower Aβ (Β = -0.32), and significantly higher tau (Β = 0.50) and ptau levels (Β = 0.31); P < 0.001 for all.

Cerebrospinal fluid ACE measures and cerebrospinal fluid biomarkers

In the total population, higher CSF ACE protein level was independently associated with higher CSF Aβ, tau and ptau levels (Table 2). When stratified by study group, the associations between higher CSF ACE protein levels and higher CSF Aβ, tau and ptau were present in the control group and the AD group, although significance was lost for CSF tau in the control group (Figures 1, 2 and 3).

	Model	Ln Aβ42	Ln tau	Ln ptau
CSF ACE protein levela	1	0.12 (0.05; 0.19)**	0.10 (-0.01; 0.22)	0.14 (0.05; 0.22)**
	2	0.09 (0.04; 0.15)**	0.15 (0.06; 0.25)**	0.17 (0.10; 0.25)**
	3	0.11 (0.05; 0.16)**	0.16 (0.05; 0.26)**	0.19 (0.10; 0.27)**
CSF ACE activityb	1	0.07 (-0.00; 0.14)	-0.06 (-0.18; 0.06)	0.02 (-0.07; 0.11)
	2	0.05 (-0.00; 0.11)	-0.03 (-0.13; 0.07)	0.04 (-0.04; 0.12)
	3	0.06 (0.00; 0.11)*	-0.04 (-0.14; 0.06)	0.03 (-0.05; 0.11)
Serum ACE protein levelc	1	0.09 (0.02; 0.17)*	-0.09 (-0.21; 0.03)	-0.02 (-0.11; 0.07)
	2	0.06 (-0.00; 0.12)	-0.03 (-0.13; 0.07)	0.02 (-0.06; 0.10)
	3	0.04 (-0.02; 0.11)	-0.04 (-0.15; 0.06)	0.02 (-0.06; 0.11)
Serum ACE activityd	1	0.06 (-0.01; 0.14)	-0.10 (-0.22; 0.01)	-0.02 (-0.10; 0.07)
	2	0.03 (-0.03; 0.09)	-0.05 (-0.15; 0.05)	0.02 (-0.06; 0.10)
	3	0.02 (-0.04; 0.08)	-0.06 (-0.17; 0.05)	0.01 (-0.08; 0.10)

Data presented as Β (95% confidence interval). Model 1: unadjusted. Model 2: adjusted for age, sex and study groups. Model 3: additionally adjusted for APOE-ϵ4 genotype, systolic blood pressure, diastolic blood pressure, body mass index, smoking, history of hypercholesterolemia and diabetes mellitus. Aβ, amyloid-beta; ACE, angiotensin-converting enzyme; CSF, cerebrospinal fluid; Ln, log-transformed; ptau, tau phosphorylated at threonine 181; RFU, relative fluorescence units; tau, total tau. *P < 0.05, **P < 0.01. ^aPer standard deviation increase (1.32 ng/ml). ^bPer standard deviation increase (59 RFU). ^cPer standard deviation increase (1.30 ng/ml). ^dPer standard deviation increase (133 RFU).

Higher CSF ACE activity was also marginally associated with higher CSF Aβ levels, but not with CSF tau or ptau levels (Table 2). When stratified by study groups, the association was particularly present in the AD group (Figure 1).

The main findings of this study are that lower CSF ACE protein levels, and to a lesser extent serum ACE protein and CSF ACE activity levels, were associated with lower CSF Aβ levels, indicating increased Aβ accumulation in the brain. Surprisingly, lower CSF ACE protein levels were associated with lower CSF tau and ptau levels, indicating less brain tau pathology. These associations were found in both control and AD patients and were independent of age, sex, APOE genotype, and vascular risk factors. Further, CSF and serum ACE protein and activity levels were lower in AD patients compared with controls.

To our knowledge, this is the first study that linked direct measures of ACE activity to the CSF AD biomarkers Aβ, tau, and ptau. Our results on the association of CSF ACE protein level and CSF Aβ are in line with studies that showed an association between the ACE gene and Aβ. Haplotypes of the ACE gene that were associated with lower ACE levels were related to lower CSF Aβ [13, 14], and another study showed that the I-allele (indicating lower ACE levels) was associated with slightly more Aβ load in the brain at autopsy [26]. The Rotterdam study examined the ACE genotype in relation to brain atrophy (which is associated with increased brain Aβ plaques) [1], and found that female I/I genotype carriers had smaller hippocampal and amygdalar volumes [27]. Further, we showed in the SMART-MR study that lower serum ACE levels were associated with more progression of cortical brain atrophy [28]. Together, these data suggest that lower ACE could lead to more Aβ accumulation in the brain, reflected by lower CSF Aβ, and support previous reports that higher ACE levels, through degradation of Aβ, may contribute to less accumulation of Aβ into senile plaques in the brain and subsequently be beneficial for the occurrence of brain atrophy.

Another novel finding was our observation that lower CSF ACE protein level was associated with lower CSF tau and ptau. A previous study relating the ACE genotype to tau load in the brain found no association [26]. The mechanisms explaining the relation between lower ACE and lower CSF tau levels remain largely unknown, although a role for ANGII could be suggested. A recent study showed that central infusion of ANGII in normal rat brains induced tau phosphorylation in a dose-dependent manner via activating glycogen synthase kinase 3β. Furthermore, this was reversed by co-administration of the ANG II type 1 receptor antagonist, losartan, which is commonly prescribed for the treatment of hypertension [29]. Our findings suggest that higher ACE levels which lead to higher ANGII levels can induce tau pathology in the brain, and in turn result in higher tau levels in the CSF. Coincidentally, and perhaps ironically, while such elevations in ACE may exacerbate tau-related pathology, these same elevations may also have a slightly beneficial mitigating effect in terms of reducing Aβ pathology. Yet no association was seen between CSF tau and ptau with CSF ACE activity, which we might expect to be a more biologically meaningful measure of the likely production of ANGII from its precursor ANGI [17].

CSF ACE levels were more strongly associated with AD biomarkers than serum ACE levels. This indicates that CSF ACE, as might be expected, could reflect the activity of the brain RAS, thereby showing stronger associations with brain pathology. Further, ACE protein levels were more strongly associated with AD biomarkers than ACE activity levels. Whereas protein levels indicate the concentration of ACE in body fluid, activity levels particularly indicate the ANGI to ANGII conversion capacity of the enzyme [22]. One might thus have expected that ACE activity levels in particular would be more indicative of and relevant to Aβ degradation, and would therefore be more strongly associated with CSF Aβ. At present we cannot explain this apparent inconsistency; it could be partly related to the ACE genotype although the genetic contribution to ACE levels has only been shown to be relatively modest [12], and unfortunately we did not have information on the ACE genotype in our population so we could not explore this. Another reason might be differential levels of post-translational modification in the various physiological compartments from which the samples were derived. ACE activity, and its proposed conversion of Aβ, has been shown previously to be modulated by post-translational modification [30]. In both CSF and serum there may therefore be differential levels of glycosylation that can explain the lower than expected correlations of activity with some of our outcome measures. Miners and colleagues previously showed a trend of divergence between ACE levels and activity in postmortem CSF, but this was not significant and the patterns observed may have been an artifact of postmortem delay where additional changes to the ACE may have occurred post mortem [22]. Yet here remains the possibility that in the brain, from which the levels of Aβ, tau and ptau emanate, levels of ACE activity are influenced in a more disease-specific manner by advancing pathology than occurs in the more peripheral compartments we have tested. Indeed, Miners and colleagues have shown in postmortem brain tissue from AD patients that their ACE activity and not their ACE levels are more markedly different than in the control group [22]. Similarly, according to our data there is some support that disease-specific modifications of ACE may be involved since the apparent discrepant findings between ACE levels and activity with respect to CSF Aβ were less marked and seemed to correlate more in the AD group (Figure 1B).

Among the strengths of the current study is the use of CSF ACE measures and CSF biomarkers, which gives the opportunity to examine during life the association between ACE and AD pathology, without potential confounding from postmortem delays. Second, we had paired CSF and serum samples for all of the patients. Finally, all patients were assessed in a standardized way and diagnosed according to commonly used criteria [18].

One of the limitations of our study is the relatively small study population. Yet the interesting results recommend replication in a larger study sample. Another limitation is the cross-sectional design that prohibits conclusions on cause or effect. Similarly, we cannot ascertain whether lower CSF ACE protein and activity levels cause brain Aβ pathology. However, because we found that the associations were also present in the control group, it is reasonable to assume that the downregulation of CSF ACE is not pathology driven, and that lower ACE levels might precede the pathology or alternatively are a general phenomenon associated with neurodegeneration. This hypothesis remains to be further investigated in a larger, longitudinal study.

This study has important clinical implications. If ACE in the brain does prevent Aβ accumulation in living humans, our data support the suggestions that taking ACE inhibitors as antihypertensive drugs, particularly those that can cross the blood–brain barrier, might compromise Aβ degradation [31]. Fortunately, while uncertainty remains on this issue, angiotensin-receptor blockers, which solely target ANG II effects whilst not interfering directly with ACE, could provide an immediate and readily available potential alternative anti-hypertensive therapy for patients at risk for AD [31–33].

We found in a memory clinic cohort that lower CSF ACE levels were associated with lower CSF Aβ levels, which suggests more accumulation of Aβ in the brain. This observation lends support to the hypothesis that ACE can degrade Aβ, thereby contributing to limiting its accumulation and possibly to the development or rate of progression of AD.