Angiotensin-(1–7) Is a Modulator of the Human Renin-Angiotensin System
Abstract—The renin-angiotensin system is important for cardiovascular homeostasis. Currently, therapies for different cardiovascular diseases are based on inhibition of angiotensin-converting enzyme (ACE) or angiotensin II receptor blockade. Inhibition of ACE blocks metabolism of angiotensin-(1–7) to angiotensin-(1–5) and can lead to elevation of angiotensin-(1–7) levels in plasma and tissue. In animal models, angiotensin-(1–7) itself causes or enhances vasodilation and inhibits vascular contractions to angiotensin II. The function of angiotensin-(1–5) is unknown. We investigated whether angiotensin-(1–7) and angiotensin-(1–5) inhibit ACE or antagonize angiotensin-induced vasoconstrictions in humans. ACE activity in plasma and atrial tissue was inhibited by angiotensin-(1–7) up to 100%, with an IC50 of 3.0 and 4.0 μmol/L, respectively. In human internal mammary arteries, contractions induced by angiotensin I and II and the non–ACE-specific substrate [Pro11,D-Ala12]-angiotensin I were antagonized by angiotensin-(1–7) (10−5 mol/L) in a noncompetitive way, with a 60% inhibition of the maximal response to angiotensin II. Contractions to ACE-specific substrate [Pro10]-angiotensin I were also inhibited, an effect only partly accounted for by antagonism of angiotensin II. Angiotensin-(1–5) inhibited plasma ACE activity with a potency equal to that of angiotensin I but had no effect on arterial contractions. In conclusion, angiotensin-(1–7) blocks angiotensin II–induced vasoconstriction and inhibits ACE in human cardiovascular tissues. Angiotensin-(1–5) only inhibits ACE. These results show that angiotensin-(1–7) may be an important modulator of the human renin-angiotensin system.
- angiotensin I
- angiotensin II
- angiotensin-converting enzyme
The renin-angiotensin system (RAS) is considered one of the most important regulatory systems for cardiovascular homeostasis. Angiotensin II (Ang II), which is cleaved from angiotensin I (Ang I) by angiotensin-converting enzyme (ACE), is a vasoconstrictor and growth stimulator when acting on the Ang II type 1 receptor (AT1). Consequently, ACE inhibitors and AT1 blockers decrease blood pressure and have beneficial effects in cardiovascular and renal disease.1
The mechanisms by which ACE inhibitors exert their effects are not entirely understood. Several studies suggest that potentiation of bradykinin through diverse mechanisms plays a pivotal role.1 2 3 4 5 However, recently it has been proposed that angiotensin-(1–7) [Ang-(1–7)] may also mediate some ACE inhibitor–related effects. It has been shown in animal and cell culture studies that ACE inhibition can increase plasma Ang-(1–7) levels.3 6 7 Animal and cell culture studies suggest that Ang-(1–7) itself inhibits ACE activity, antagonizes AT1 receptors, enhances bradykinin-induced effects, stimulates prostanoid release, and stimulates nitric oxide (NO) through an Ang-(1–7) receptor.8 9 10 11 12 13 14 Thus, angiotensin-(1–7) can act as a vasodilator through kinins and NO in vessels shown to contract in response to Ang II, can have antihypertensive actions in rats, and inhibits cultured rat vascular smooth muscle cell (VSMC) growth.15 16 17 This has led to the concept that Ang-(1–7) may be an endogenous counterplayer of the RAS through a wide variety of mechanisms.18
Although the counterregulatory actions of Ang-(1–7) have been well documented in animal models, no extensive studies have addressed the effects of Ang-(1–7) on human cardiovascular function. A single study by Kono et al19 showed that Ang-(1–7) causes a modest, transient rise in blood pressure in normotensive subjects. In patients with essential hypertension treated for 6 months with captopril, plasma Ang-(1–7) levels increase shortly after oral intake of captopril.20 This increase in Ang-(1–7) levels may be caused by inhibition of conversion of Ang-(1–7) to angiotensin-(1–5) [Ang-(1–5)] by ACE.7 21 Plasma Ang-(1–7) levels correlated inversely with diastolic blood pressure in these patients.20 These results suggest that Ang-(1–7) might be a clinically relevant counterplayer of the RAS. Because Ang-(1–7) levels are increased by ACE inhibition, whereas Ang-(1–5)7 21 levels are decreased, a role for Ang-(1–5) as a counterregulator of the RAS would be unexpected. To our knowledge, the direct effects of Ang-(1–7) and Ang-(1–5) on human vascular function have not been studied. Therefore, we sought to study the effects of Ang-(1–7) and Ang-(1–5) on plasma or tissue ACE activity and contractile function in human arteries.
Acquisition of Human Arteries, Plasma, and Atrial Tissue
Internal mammary arteries (IMA) were obtained from 24 patients and the right atrial appendix was obtained from 1 patient who underwent coronary artery bypass grafting surgery between February 1997 and March 1998 in the University Hospital of Groningen, Netherlands. Human plasma was obtained from 1 healthy donor by the Northern Center of Blood Transfusion, Groningen. Studies with excess grafting arteries and right atrial appendices are approved by the local ethics review committee, and informed consent was obtained in all cases.
Measurement of ACE Activity
ACE activity in plasma (35 times diluted) and atrial tissue (1 mg per assay) was measured according to the method of Cushman and Cheung,22 using 10 minutes of incubation with 7 mmol/L of hippuryl-l-histidyl-l-leucine (HHL) at 37°C. The effect of different doses of lisinopril or Ang-(1–7) was measured in quadruplicate. In plasma, the experiment was repeated twice, first with different doses of Ang-(1–7) and Ang I, and the second time with different doses of Ang-(1–7) and Ang-(1–5). Lineweaver-Burk analysis was done with varying doses of Ang-(1–7) (10−8 to 10−5 mol/L) or lisinopril (10−11 to 10−8 mol/L) and HHL (0.5 to 4.0 mmol/L).
Preparation of Arterial Rings for Organ Bath Experiments
IMA were prepared for isotonic (1.4 g) organ bath experiment as described before.23 In short, IMA rings were kept in a carbogen-buffered, aerated (95% O2/5% CO2) Krebs’ solution, containing the following (mmol/L): NaCl 120.4, KCl 5.9, CaCl2 2.5, MgCl2 1.2, NaH2PO4 1.2, glucose 11.5, and NaHCO3 25.0, at 37°C. Before the onset of all dose-response experiments, the rings were allowed to equilibrate for ≈1 hour and were rinsed at least 4 times, after which responsiveness was tested with the use of 10−5 mol/L phenylephrine (PE). Finally, PE was washed out until the rings were relaxed and stabilized. If necessary, the test was repeated.
Effect of Ang-(1–7) and Ang-(1–5) on Angiotensin-Induced Contractions
After PE was washed out to relax and stabilize the rings, the rings were incubated for 20 minutes with 10−4 mol/L NG-monomethyl-l-arginine (L-NMMA) to inhibit endothelium-dependent counterregulation against contractions to angiotensins. In the Ang-(1–7)–treated rings, 10−5 mol/L Ang-(1–7) was administered 10 minutes after addition of L-NMMA. After the preincubation period, increasing cumulative doses (10−10 to 10−6 mol/L) of Ang II, Ang I, or [Pro10]-angiotensin I ([Pro10]Ang I), a substrate solely converted by ACE,24 were added. Finally, 10−5 mol/L PE (PE-post) and, after stabilization, 60 mmol/L KCl were added. In a second set of experiments, dose-response curves (10−9 to 10−5 mol/L) to [Pro11, D-Ala12]-angiotensin I ([Pro11, D-Ala12]Ang I), a substrate solely converted by non-ACE pathways,25 were performed as described for the first set. In a third set of experiments, the effect of Ang-(1–5) on Ang II–induced contractions was studied in exactly the same way as for Ang-(1–7). In a fourth set of experiments, we tested whether Ang-(1–7) (10−9 to 10−5 mol/L) had an effect in rings precontracted with 10−5 mol/L PE (PE-pre) in the presence of 10−4 mol/L L-NMMA.
Mean ACE activity after treatment was expressed as a percentage of the mean maximal ACE activity in the presence of vehicle. The IC50 was calculated with the logistic curve-fitting procedure of Sigma Plot (Jandel Corporation).
In IMA rings, treatment [Ang-(1–7) or Ang-(1–5)] and control (untreated rings) experiments were performed within parallel rings of each single patient. Per patient and treatment, experiments were performed at least in duplicate. Per patient, the means of the absolute responses were calculated first and then expressed as a percentage of the response to PE-post or PE-pre. ANOVA for repeated measures was used to calculate probability value. The Greenhouse-Geisser adjustment was made for multisample asphericity. Contractions to PE-post were compared by t test. The EC50 for Ang II or [Pro11, D-Ala12]Ang I was estimated per patient by logistic curve fitting, with the use of Mann-Whitney rank sum tests to assess significance of differences. One case in the Ang-(1–7)–treated Ang II group was due to the absence of contractions to Ang II. For all statistical tests, SPSS/PC statistical software was used. All probability values mentioned are absolute when 0.001<P<0.05. P>0.05 was considered not significant.
All compounds for Krebs’ solution, NaOH, and K2HPO4 were from Merck. Ang-(1–7) and Ang-(1–5) were purchased from Bachem. [Pro10]Ang I was produced by Bio-Tez. [Pro11, D-Ala12]Ang I was produced by Isogen bv. All other chemicals were purchased from Sigma Chemical Co. All chemicals were dissolved in distilled H2O.
Table 1⇓ shows baselines characteristics of 23 patients in whom excess grafting material was used for assessing vascular function. The data of 1 patient were excluded because of an incomplete clinical data set. The patient from whom the atrial appendix was obtained was an 81-year-old man who received no preoperative medication and did not smoke.
Effect of Ang-(1–7) and Ang-(1–5) on Plasma and Tissue ACE Activity
Ang-(1–7), Ang I, and Ang-(1–5) inhibited the conversion of HHL in human plasma. However, Ang-(1–7) was more potent than Ang I or Ang-(1–5), whereas Ang I and Ang-(1–5) were equipotent (Figure 1A⇓). As a positive control, we used lisinopril to indicate that the assay was suitable for measuring inhibition of ACE activity. IC50 values were as follows: Ang-(1–7), 3.0×10−6 mol/L; lisinopril, 7.4×10−10 mol/L. In a homogenate of human right atrial appendix, Ang-(1–7) and lisinopril inhibited ACE, with IC50 values of 4.1×10−6 and 3.3×10−9 mol/L. A Lineweaver-Burk plot show that Ang-(1–7) inhibits plasma ACE activity in a noncompetitive/mixed way (Figure 1B⇓), but competitive component seems to dominate. Lisinopril inhibits ACE competitively (Figure 1C⇓), and it is therefore unlikely that the effect of Ang-(1–7) is confounded by accumulation of the His-Leu product.
Effect of Ang-(1–7) and Ang-(1–5) on Contractions Induced by Angiotensins in Human IMA
Ang-(1–7) 10−5 mol/L significantly inhibited dose-dependent responses to Ang II (Figure 2A⇓). Ang-(1–7) inhibited the maximal responses by 60% but did not cause a significant shift in EC50 concentration of Ang II (Table 2⇓). Ang-(1–5) did not change Ang II–induced responses (Figure 2E⇓).
Ang-(1–7) significantly inhibited contractions to Ang I and [Pro10]Ang I (Figure 2B⇑ and 2C⇑, respectively). The [Pro10]Ang I curve appears to be shifted more to the right by Ang-(1–7) in comparison to Ang I and Ang II. [Pro11, D-Ala12]Ang I was not significantly inhibited (Figure 2D⇑). EC50 values for [Pro11, D-Ala12]Ang I and Ang II were not changed (Table 2⇑).
Neither PE-post responses nor KCl responses were significantly different between control and Ang-(1–7)–treated rings (Table 2⇑). Ang-(1–7) did not have an effect in PE-precontracted IMA in the presence of L-NMMA (n=5; data not shown).
The present study shows that Ang-(1–7) antagonizes Ang II–induced contractions in human arteries and inhibits human plasma, atrial, and arterial ACE. To our knowledge, this is the first study that shows a direct effect of Ang-(1–7) on human vascular tissue. The data suggest that Ang-(1–7) is a modulator of the human cardiovascular RAS. Whereas Ang-(1–7) modulates the human RAS, its metabolite Ang-(1–5) does not affect Ang II–induced vasoconstriction and inhibits plasma ACE activity only at very high doses.
The blockade of Ang II–induced contractions by Ang-(1–7) in our experiments could be accomplished in different ways: (1) by antagonism of AT1 receptors, (2) by release of NO or other vasorelaxing factors (such as prostaglandins, eg, PGI2), or (3) by intracellular mechanisms. Although we did not directly address this question in this report, some clues concerning the mechanisms are given by our data. Firstly, we assessed whether Ang-(1–7) specifically inhibits Ang II–induced vasoconstriction. Recent studies have shown that Ang-(1–7) causes a vasodilating response in canine and porcine coronary arteries, rat aorta, and rabbit renal arterioles.10 12 15 26 Recent studies suggest that this action of Ang-(1–7) is mediated by release of NO through a yet undiscovered angiotensin receptor, situated at the endothelium or media.10 12 13 Release of NO could cause nonspecific inhibition of vasoconstriction. However, in our experiments we added L-NMMA, and no dilator effects of Ang-(1–7) were seen in precontracted rings. Furthermore, it was shown that Ang-(1–7) inhibits Ang II–induced but not PE- or KCl-induced contractions in human arteries treated with L-NMMA. Thus, a nonspecific antagonism of vasoconstriction through NO-independent relaxation can be excluded as well. Therefore, we suggest that in our experiments, Ang-(1–7) specifically antagonized Ang II–induced vasoconstriction through a noncompetitive blockade of AT1 receptors, as also suggested by studies in rabbit arteries.11 Alternatively, cross talk between AT1 and other angiotensin receptors may antagonize Ang II–induced vasoconstriction in human arteries. Further study is needed to find the receptors involved in specific counterregulation of the RAS by Ang-(1–7).
Apart from inhibition of Ang II–induced responses, the study addressed the question of whether Ang-(1–7) modulates the human RAS through ACE inhibition. ACE inhibition by Ang-(1–7) was confirmed by a biochemical assay in human atrial tissue and plasma. Additional proof was found by functional measurements in arteries with the use of [Pro10]Ang I. Although it is not demonstrated by a full dose-response curve, the rightward shift in the dose-response curve to [Pro10]Ang I is evidently present. Because [Pro10]Ang I is converted to Ang II solely by ACE,24 this shift is probably due to inhibition of ACE. In accordance, inhibition of ACE by Ang-(1–7) seems largely competitive. Ang I can still be converted through alternative pathways,27 28 as shown by the responses to Ang I and [Pro11, D-Ala12]Ang I. Therefore, Ang-(1–7) seems to primarily inhibit the ACE pathway. We conclude that Ang-(1–7) could modulate the human RAS not only through antagonism of Ang II but also through inhibition of ACE.
Our study shows that Ang-(1–7) inhibits human ACE in a noncompetitive/mixed way. This is a new insight, because it was assumed until now that Ang-(1–7) is a competitive inhibitor of human ACE.9 21 The mixed inhibition may be caused by the presence of 2 active sites in ACE, to both of which Ang-(1–7) can bind with equal affinity.9 Speculatively, binding to one site could inactivate the other, either by a conformational change of the enzyme or by a steric interference between the 2 ligands. If degradation of HHL predominantly takes place through the C-terminus, Ang-(1–7) would be a noncompetitive/mixed inhibitor by binding to the N-terminus.
Because Ang-(1–7) inhibits the RAS in human cardiovascular tissues, it may be a factor with important physiological implications. The relevance of Ang-(1–7) as a physiological modulator will depend on the quantity of the peptide produced in vivo. In this in vitro study, effects of Ang-(1–7) were observed at high doses compared with the plasma levels of the compound.3 20 Such a discrepancy has also been noted for Ang I and Ang II.3 20 A practical approach may be to compare the ratio of angiotensins when measured in vivo and in vitro. Thus, the higher is the Ang-(1–7)/Ang II or Ang-(1–7)/Ang I ratio, the stronger Ang-(1–7) antagonizes Ang II or the conversion of Ang I. In normotensive rats and renin-treated dogs, chronic ACE inhibition increases the plasma Ang-(1–7)/Ang II ratio to the extent that an effect of Ang(1–7) can be expected.3 6 Interestingly, in patients with essential hypertension, the plasma Ang-(1–7)/Ang II ratio decreases on long-term treatment with captopril, whereas arterial pressure is not lowered.20 These studies suggest that the Ang-(1–7)/Ang II ratio is important for an effect of ACE inhibition on blood pressure and is related to the etiology of the hypertension. Similarly, the Ang-(1–7)/Ang I ratio can be discussed to estimate the role of Ang-(1–7) inhibition of Ang I conversion through ACE. Our present data show that Ang-(1–7) competes better than Ang I for the HHL conversion site. When we consider the Ang-(1–7)/Ang I ratios measured in vivo, Ang-(1–7) could modulate Ang I conversion by ACE in untreated normotensive rats and hypertensive humans.3 20 Whereas a role for Ang-(1–7) as an ACE inhibitor may be predictable, the contribution of its metabolite Ang-(1–5) is uncertain. Although our results show that Ang-(1–5) may be a competitor for Ang I conversion, no data are available that could predict its role in vivo. In future studies, in vivo Ang-(1–5) levels should be measured.
Apart from dose-related issues, comparisons between in vivo and in vitro effects of Ang-(1–7) must be made to estimate its role under physiological conditions. Our results show that Ang-(1–7) inhibits ACE and antagonizes Ang II–induced contractions. In contrast, Kono et al19 (1986) show that intravenous infusion of Ang-(1–7) causes a small, biphasic increment of blood pressure in normotensive humans. Similar effects were seen in anesthetized rats.29 However, Ang-(1–7) did not seem to elicit a direct effect on blood pressure in other studies with normotensive and hypertensive rats, although this may have been caused by differences in experimental setup.8 30 In other models, Ang-(1–7) caused vasodilation of various isolated arteries and had a depressor effect in the pithed and the spontaneously hypertensive rat.16 18 Speculatively, in in vivo studies, counterregulation mechanisms against Ang-(1–7) may be present in untreated animals, or hypertension may induce a state of increased sensitivity for Ang-(1–7). Both mechanisms could create the discrepancies between in vivo and in vitro studies for normotensive models and reflect the distinction between a systemic and a local RAS. We conclude that systemic effects of Ang-(1–7) may differ from local effects and depend on the model that is used. Very recently, it was shown that spontaneous and transgenic hypertensive rats have an increased sensitivity for intravenously administered Ang-(1–7) compared with normotensive rats.7 Therefore, more specific studies in normotensive and hypertensive humans should address the effects of Ang-(1–7) on blood pressure and vascular reactivity to assess its role in humans under physiological conditions during cardiovascular disease. The value of such studies was demonstrated by Ferrario et al,31 who found an inverse correlation between urinary Ang-(1–7) concentrations and blood pressure in normotensive and hypertensive humans. It was proposed that the inverse relation between Ang-(1–7) and blood pressure may provide a marker for forms of essential hypertension.31
In this study, we have shown for the first time that Ang-(1–7) is an antagonist of vascular responses to Ang II in human arteries and an inhibitor of ACE in human plasma, cardiac tissue, and vascular tissue. Therefore, Ang-(1–7) could have an important modulating effect on the human RAS.
This study was supported by the Netherlands Organization for Scientific Research (grant 950-10-642). The authors wish to thank Katja Wolters for her indispensable support in the biochemical measurement of ACE activity and Dr Hans L. Hillege for his advice on the statistical methods.
- Received December 7, 1998.
- Revision received January 5, 1999.
- Accepted March 17, 1999.
Gavras H, Gavras I. Endothelial Function in Cardiovascular Disease: The Role of Bradykinin. London, England: Science Press; 1996.
Berkenboom G, Brékine D, Unger P, Grosfils K, Staroukine M, Fontaine J. Chronic angiotensin-converting enzyme inhibition and endothelial function of rat aorta. Hypertension. 1995;26:738–743.
Campbell DJ, Kladis A, Duncan AM. Effects of converting enzyme inhibitors on angiotensin and bradykinin peptides. Hypertension. 1994;23:439–449.
Gohlke P, Scholkens BA, Kuwer I, Bartenbach S, Schnell A, Unger T, Linz W. Angiotensin converting enzyme inhibition improves cardiac function: role of bradykinin. Hypertension. 1994;23:411–418.
Minshall RD, Tan F, Nakamura F, Rabito SF, Becker RP, Marcic B, Erdös EG. Potentiation of the actions of bradykinin by angiotensin-converting enzyme inhibitors: the role of expressed human bradykinin B2 receptors and angiotensin I-converting enzyme in CHO cells. Circ Res. 1997;81:848–856.
Kohara K, Brosnihan KB, Chappell MC, Khosla MC, Ferrario CM. Angiotensin-(1–7): a member of circulating angiotensin peptides. Hypertension. 1991;17:131–138.
Yamada K, Iyer SN, Chappell MC, Ganten D, Ferrario CM. Converting enzyme determines plasma clearance of angiotensin-(1–7). Hypertension. 1998;32:496–502.
Paula RD, Lima CV, Khosla MC, Santos RAS. Angiotensin-(1–7) potentiates the hypotensive effect of bradykinin in conscious rats. Hypertension. 1995;26:1154–1159.
Deddish PA, Marcic B, Jackman HL, Wang H-Z, Skidgel RA, Erdös EG. N-domain-specific substrate and C-domain inhibitors of angiotensin-converting enzyme: angiotensin-(1–7) and keto-ACE. Hypertension. 1998;31:912–917.
Tallant EA, Lu X, Weiss RB, Chappell MC, Ferrario CM. Bovine aortic endothelial cells contain an angiotensin-(1–7) receptor. Hypertension. 1997;29:388–393.
Muthalif MM, Benter IF, Uddin MR, Harper JL, Malik KU. Signal transduction mechanisms involved in angiotensin-(1–7)-stimulated arachidonic acid release and prostanoid synthesis in rabbit aortic smooth muscle cells. J Pharmacol Exp Ther. 1998;284:388–398.
Brosnihan KB, Li P, Ferrario CM. Angiotensin-(1–7) dilates canine coronary arteries through kinins and nitric oxide. Hypertension. 1996;27:523–528.
Benter IF, Ferrario CM, Morris M, Diz I. Antihypertensive actions of angiotensin-(1–7) in spontaneously hypertensive rats. Am J Physiol. 1995;269:38:H313–H319.
Freeman EJ, Chisolm G, Ferrario CM, Tallant EA. Angiotensin-(1–7) inhibits vascular smooth muscle cell growth. Hypertension. 1996;28:104–108.
Ferrario CM, Chappell MC, Tallant EA, Brosnihan KB, Diz I. Counterregulatory actions of angiotensin-(1–7). Hypertension. 1997;30:535–541.
Chappell MC, Pirro NT, Sykes A, Ferrario CM. Metabolism of angiotensin-(1–7) by angiotensin-converting enzyme. Hypertension. 1998;31:362–367.
Cushman DW, Cheung HS. Spectrophotometric assay and properties of the angiotensin-converting enzyme of rabbit lung. Biochem Pharmacol. 1971;20:1637–1648.
Buikema H, Grandjean JG, van den Broek S, van Gilst WH, Lie KI, Wesseling H. Differences in vasomotor control between human gastroepiploic and left internal mammary artery. Circulation. 1992;86:II-205–II-209.
Kinoshita A, Urata H, Bumpus FM, Husain A. Measurement of angiotensin I converting enzyme inhibition in the heart. Circ Res. 1993;73:51–60.
Ren Y, Garvin JL, Carretero OA. Vasodilator action of angiotensin 1–7 (Ang 1–7) in the isolated afferent arteriole of the rabbit kidney. J Am Soc Nephrol. 1998;9:345A. Abstract.
Urata H, Kinoshita A, Misono KS, Bumpus FM, Husain A. Identification of a highly specific chymase as the major angiotensin II-forming enzyme in the human heart. J Biol Chem. 1990;265:22348–22357.
Roks AJM, Buikema H, Pinto YM, van Gilst WH. The renin-angiotensin system and vascular function: the role of angiotensin II, angiotensin-converting enzyme, and alternative conversion of angiotensin I. Heart Vessels. 1997;suppl 12:119–124.
Abbas A, Gorelik G, Carbini LA, Scicli AG. Angiotensin-(1–7) induces bradykinin-mediated hypotensive responses in anesthetized rats. Hypertension. 1997;30:217–221.
Lima CV, Paula RD, Resende FL, Khosla MC, Santos RAS. Potentiation of the hypotensive effect of bradykinin by short-term infusion of angiotensin-(1–7) in normotensive and hypertensive rats. Hypertension. 1997;30:542–548.
Ferrario CM, Martell N, Yunis C, Flack JM, Chappell MC, Brosnihan KB, Dean RH, Fernandez C, Novikov SV, Pinillas C, Luque M. Characterization of angiotensin-(1–7) in the urine of normal and essential hypertensive subjects. Am J Hypertens. 1998;11:137–146.