Activation of Aldehyde Dehydrogenase-2 Reduces Ischemic Damage to the Heart

See allHide authors and affiliations

Science  12 Sep 2008:
Vol. 321, Issue 5895, pp. 1493-1495
DOI: 10.1126/science.1158554


There is substantial interest in the development of drugs that limit the extent of ischemia-induced cardiac damage caused by myocardial infarction or by certain surgical procedures. Here, using an unbiased proteomic search, we identified mitochondrial aldehyde dehydrogenase 2 (ALDH2) as an enzyme whose activation correlates with reduced ischemic heart damage in rodent models. A high-throughput screen yielded a small-molecule activator of ALDH2 (Alda-1) that, when administered to rats before an ischemic event, reduced infarct size by 60%, most likely through its inhibitory effect on the formation of cytotoxic aldehydes. In vitro, Alda-1 was a particularly effective activator of ALDH2*2, an inactive mutant form of the enzyme that is found in 40% of East Asian populations. Thus, pharmacologic enhancement of ALDH2 activity may be useful for patients with wild-type or mutant ALDH2 who are subjected to cardiac ischemia, such as during coronary bypass surgery.

Cardiac ischemia is the leading cause of death. The discovery of a cardioprotective mechanism called preconditioning (induced by repetitive sublethal ischemic events) has triggered the search for pharmacological agents that mimic this effect (1, 2). Adenosine (3), ethanol (4), and selective activation of protein kinase Cϵ (PKCϵ)(4, 5) mimic ischemic preconditioning and reduce cardiac infarct size. Systematic searches for mediators of cardiac protection have identified a number of proteins whose levels or phosphorylation changes with cardioprotection (6, 7). However, whether the changes were critical for cardiac protection was not determined.

We used an unbiased proteomic approach in ischemic rat hearts treated with ethanol and a selective inhibitor and an activator of PKCϵ that we generated (4, 7, 8). We found that one protein whose phosphorylation status consistently correlated with cardioprotection was mitochondrial aldehyde dehydrogenase 2 (ALDH2) (Fig. 1 and fig. S1, A to D). Under normoxic conditions, ALDH2 appeared as four phosphoproteins after isoelectric focusing (IEF) for two-dimensional (2-D) SDS gel electrophoresis [note 1 in (5)]. After preconditioning by a brief exposure to ethanol (50 mM, 10 min) (9) or selective activation of PKCϵ by the isozyme-specific agonist peptide, ψϵRACK (receptor for activated C kinase) (1 μM, 10 min), which causes cardioprotection (8), there were only two (the more acidic) ALDH2 spots (Fig. 1A). The ethanol-induced shift in ALDH2 mobility was inhibited in the presence of the PKCϵ-selective antagonist peptide (ϵV1-2) (Fig. 1A), a treatment that we previously found to inhibit ethanol-induced cardiac protection (9). Therefore, ethanol-induced ALDH2 phosphorylation, which correlates with cardiac protection from ischemia, is dependent on PKCϵ activation.

Fig. 1.

(A) Ethanol and PKCϵ activation induce phosphorylation of mitochondrial ALDH2. Homogenates of rat hearts subjected to ischemia ex vivo were separated by IEF/SDS 2-D gel electrophoresis and probed with a mixture of phospho-serine and phospho-threonine antibodies. Using a Langendorff apparatus, hearts were perfused with oxygenated Krebs-Henseleit buffer alone as control, with 50 mM ethanol for 10 min, with 1 μM PKCϵ agonist (ψϵRACK) for 10 min [note 1 in (5)], or with 1 μM PKCϵ antagonist (ϵV1-2) for 5 min, followed by 10 min of perfusion together with 50 mM ethanol. The hearts were then subjected to a 30-min period of no-flow ischemia before homogenization. Treatment with ethanol and ψϵRACK induced a leftward shift of ALDH2 as compared with control, which was blocked with ϵV1-2 treatment. Blots were probed with antibodies against ALDH2 or against phospho-Ser and phospho-Thr (5). (B) ALDH2 activity correlates with cardiac protection from ischemic injury. Measurements of ALDH activities in normoxic and ischemic rat hearts treated with ethanol (EtOH, 50 mM), PKCϵ agonist (ψϵRACK), or PKCϵ antagonist (ϵV1-2) in the presence of ethanol using the Langendorf apparatus (5). Ischemic hearts were also treated with the ALDH2 inhibitor cyanamide (CYA) in the presence or absence of ethanol, PKCϵ agonist and antagonist, and the ALDH2 desensitizer, nitroglycerin (GTN). Shown is ALDH2 activity (μmol of NADH/min per mg protein) as a function of infarct size, measured by 2,3,5-triphenyltetrazolium chloride (TTC) staining from corresponding heart samples derived from the same studies as in Table 1. Linear regression yielded a high inverse correlation of R2 = 0.95.

How this mitochondrial enzyme was regulated by the cytosolic PKCϵ was not obvious. We first demonstrated that PKCϵ phosphorylates ALDH2 in vitro and that this phosphorylation results in a 38 ± 9% increase in ALDH2 catalytic activity [n = 6, P < 0.005 (fig. S2, A and B); notes 2 and 3 in (5)]. At least two phosphorylation sites were identified by mass spectroscopy, including Thr185 and Thr412 and possibly Ser279 [note 3 in (5)]. Further, coimmunoprecipitation of extracts from normoxic and ischemic hearts with antibodies against PKCϵ or ALDH2 confirmed the association of ALDH2 and PKCϵ in the mitochondrial fraction [fig. S3 and (5)]. Other subfractionation and biochemical studies have shown that PKCϵ regulates intramitochondrial proteins [e.g., (6)]. It is therefore likely that PKCϵ can enter the mitochondria and phosphorylate ALDH2 directly.

We next determined whether ALDH2 is activated in the intact heart following PKCϵ activation or ethanol treatment and whether there is a correlation between the activity of ALDH2 and infarct size under various treatment conditions. Ischemia alone did not affect ALDH2 activity (Table 1). However, ethanol treatment caused a 21% increase in ALDH2 activity relative to control and a 27% reduction in infarct size [P <0.05 (Table 1 and Fig. 1B)]. Treatment with the selective PKCϵ activator ψϵRACK (8) increased ALDH2 activity by 33% with a concomitant 49% reduction in infarct size; and inhibition of PKCϵ by the selective antagonist ϵV1-2 (7) abolished both the ethanol-induced increase in ALDH2 activity and the ethanol-induced cardiac protection from ischemia (Table 1). Further, in the presence of the ALDH inhibitor cyanamide (5 mM) (5, 10), ALDH2 activity was inhibited by 63% and infarct size increased by 49%, without causing cardiac damage under normoxic conditions; cyanamide also abolished ethanol- or ψϵRACK-induced protection and ALDH2 activation (Table 1 and Fig. 1B).

Table 1.

ALDH2 activity and infarct size in rat hearts subjected to ischemia and reperfusion, ex vivo. The experimental details are provided in (5) and in Fig. 1. EtOH, ethanol; ψϵR, ψϵRACK.

View this table:

Because cyanamide inhibits several ALDHs, we used another means to inhibit ALDH2. ALDH2 metabolizes nitroglycerin, which leads to generation of the vasodilator, nitric oxide. Yet prolonged treatment with nitroglycerin decreases ALDH2 activity (5, 11). We reasoned that if ALDH2 activity is critical for cardioprotection from ischemic damage, prolonged treatment with nitroglycerin should inhibit PKCϵ-dependent preconditioning. As expected, a 30-min treatment of nitroglycerin (GTN, 2 μM) in the ex vivo myocardial infarction model in rodents greatly inhibited ALDH2 activity and abolished ethanol- and PKCϵ-induced activation of ALDH2 (Table 1 and Fig. 1B), whereas the activity of another cardiac dehydrogenase remained unchanged (fig. S4, A and B) (5), which indicated that the changes in ALDH2 activity are probably specific. Concomitantly, GTN treatment increased ischemic cardiac damage from 45% in control to 59% and to 63 or 61% in the presence of ethanol or the PKCϵ activator (Table 1 and Fig. 1B). This effect was not due to nitric oxide generation; treatment with another nitric oxide–generating vasodilator, sodium nitroprusside (SNP, 10 μM), did not affect ALDH2 activity nor did it result in an increase in infarct size (Table 1). Therefore, there is an inverse correlation between ALDH2 activity and cardiac damage [R2 = 0.95 (Fig. 1B)], strongly suggesting that ALDH2 plays a pivotal positive role in mediating cardiac protection against ischemic injury. Creatine phosphokinase (CPK) release from the heart as an indicator of cardiac damage (4) yielded similar results [R2 = 0.97 (fig. S4, C and D)].

Nitroglycerin confers cardiac protection if the prolonged nitroglycerin treatment is terminated at least 1 hour before the ischemic event (12). Consistent with these findings, we found that 13 hours of nitroglycerin treatment (5 μg/min per kg of body weight, delivered by a patch) that was terminated 3 hours before the ischemic event decreased cardiac infarct size from 45 to 33% [GTN-off (Table 1)]. However, similar to our ex vivo data, when the nitroglycerin patch was left on, infarct size increased from 45 to 59% [GTN-on (Table 1)]. Therefore, sustained nitroglycerin treatment increased ischemic damage, probably by inducing ALDH2 inactivation [note 4 in (5)].

The inverse correlation between ALDH2 activity and cardioprotection against ischemic damage in rat [R2 = 0.95 (Fig. 1B)] (fig. S4D) does not prove that ALDH2 activation is sufficient to induce cardioprotection. We therefore searched for ALDH2 agonists using a high-throughput screen [note 1 in (5)] and identified N-(1,3-benzodioxol-5-ylmethyl)-2,6-dichlorobenzamide (Alda-1, Mr = 324) and similar analogs as ALDH2 activators (Fig. 2A and fig. S5A). We next determined whether Alda-1 activates ALDH2*2, a common East Asian mutant form that has only 1 to 5% of the catalytic activity of the wild-type ALDH2*1 form. [This E487K mutation (Glu at position 487 is replaced by Lys) is at the interface of the tetramer (13).] Alda-1 [median effective concentration (EC50) ≈ 20 μM] increased the activity of the mutant, ALDH2*2, 11-fold, the heterotetramer 2.2-fold (similar to the base levels of wild-type ALDH2), and the wild-type ALDH2*1/*1 homotetramers 2.1-fold over basal activity (Fig. 2A and fig. S5, C and D). Alda-1 had no effect on the activity of alcohol dehydrogenase 1 (5, 14), the cytosolic aldehyde dehydrogenase, ALDH1 (15), or the mitochondrial enzyme ALDH5 (5, 16) (fig. S5B [note 5 in (5)]).

Fig. 2.

(A) Alda-1 increases ALDH2 activity. Activation of wild-type and mutant versions of ALDH2 (homotetramers and heterotetramers) by Alda-1 (100 μM). Enzymatic activity of recombinant ALDH2 proteins (20 μg each) is presented as a percentage of control[n = 3; **P < 0.01 versus control; (5)]. (B) Alda-1 reduces cardiac damage in an ex vivo model of ischemia and reperfusion injury. (Top) Ex vivo cardiac ischemia model protocol. Myocardial infarct size, induced by 35 min of ischemia followed by 60 min of reperfusion after 10-min pretreatment with Alda-1 (20 μM) or vehicle control using Langendorff apparatus, as in Fig. 1, A and B [n = 6; P < 0.05; (5)]. Representative cross-sectional slices derived from a single heart stained by TTC without (control) and with Alda-1 treatment. Infarct area is indicated by the light pink color and marked with dotted lines. (C) Alda-1 reduces cardiac damage in an in vivo rat model of acute myocardial infarction. (Top) In vivo cardiac ischemia model protocol. Reduction of infarct size by injection of Alda-1 (8.5 mg/kg) before left anterior descending coronary artery (LAD) ligation (5) was also determined in vivo [n = 7; P < 0.01 (fig. S6, A and B)]. Shown is TTC staining of representative cross-sectional slices (seven rats per group).

We next used Alda-1 to determine whether direct ALDH2 activation was sufficient to induce cardioprotection. Rat hearts treated ex vivo with 20 μM Alda-1 before 35 min of ischemia followed by 60 min of reperfusion (as in Fig. 2B) had a 26 ± 6% smaller infarct (Fig. 2B) and 24 ± 7% less CPK release (n = 6; P < 0.05). Alda-1 also reduced infarct size in an in vivo rat model of acute myocardial infarction. After 35 min of ischemia and 60 min of reperfusion, infarct size of the left ventricular free wall was 43 ± 4% (n = 7) (5). Administration of 8.5 mg/kg Alda-1 into the left ventricle 5 min before ischemia decreased the myocardial infarction by 60 ± 4% [n = 7, P < 0.01 (Fig. 2C)] (fig. S6, A and B). Although low levels of noxious stimuli trigger cardioprotection (1, 2), Alda-1–induced cardioprotection was not associated with such a stress. JNK, a sensitive marker of cell stress, was not activated by Alda-1 treatment [fig. S7 and note 6 in (5)]. Therefore, activation of ALDH2 is sufficient to protect the heart from ischemia damage, in vivo.

4-Hydroxynonenal (4HNE) is a toxic aldehyde that accumulates during cardiac ischemia (17); thus, its removal by ALDH2 may be, at least in part, the mechanism by which ALDH2 activation protects the heart from ischemic damage. Furthermore, 4HNE itself can inactivate ALDH2 by forming protein adducts with the enzyme, which limits 4HNE removal (18). We confirmed that 4HNE induced rapid inactivation of ALDH2 in vitro and found that 4HNE-induced inactivation of ALDH2 was blocked by Alda-1 (Fig. 3A), which increased the detoxification of 4HNE (Fig. 3B). The molecular basis for Alda-1–induced ALDH2 protection is under investigation, but it is probably due to prevention of 4HNE adduct formation on ALDH2 (18).

Fig. 3.

(A) Effect of Alda-1 on 4HNE metabolism by ALDH2. In vitro metabolism of 4HNE (200 μM) by ALDH2 (arbitrary units) is lost within 1 min of incubation with the substrate (5), presumably because of 4HNE-induced ALDH2 inactivation (18). 4HNE-induced ALDH2 inactivation is blocked by Alda-1 (20 μM) (n = 3) as compared with vehicle control (n = 3). (B) The protection of ALDH2 from 4HNE-induced inactivation by Alda-1 correlates with a 34% reduction in 4HNE levels (n = 4; P < 0.05).

Although some of the pharmacological tools we used to regulate ALDH2 are relatively nonspecific, 12 different conditions demonstrate the correlative relation between ALDH2 activity and infarct size [R2 = 0.95 (Fig. 1B)]. Therefore, our data strongly suggest that ALDH2 activity is critical for cardioprotection from ischemia. In addition, ALDH2 contributes to ethanol metabolism, and ethanol was used to activate ALDH2. However, ethanol metabolism is unlikely to play a role in the ALDH2-mediated protection; ALDH2 activation also occurred in the absence of ethanol, as well as when we used the PKCϵ-selective activator or Alda-1. Finally, the importance of cytotoxic aldehydes, such as 4HNE, to overall ischemic injury has been previously suggested [note 7 in (5)] (17, 19, 20). It is possible that the major benefit of Alda-1 is to prevent the inactivation of cytoprotective ALDH2 by 4HNE, which would ensure continual detoxification of oxidative stress–induced cytotoxic aldehydes.

Our results raise the possibility that pharmacological enhancement of ALDH2 activity may be beneficial for patients subjected to cardiac ischemia (e.g., during coronary bypass surgery). The ability of Alda-1 to partially complement or restore mutant ALDH2*2 activity is noteworthy, as it is rare to find a small molecule that can specifically rescue a mutation in humans. Finally, our data from rodent models suggest that the prolonged use of nitroglycerin in East Asian carriers of Aldh2*2 who experience an ischemic event may need to be reconsidered and that these patients may benefit even more than carriers of the wild-type enzyme if treated with ALDH2 activators.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S7


References and Notes

Stay Connected to Science

Navigate This Article