REVIEW ARTICLES

Bioactivation of Nitroglycerin bythe Mitochondrial AldehydeDehydrogenaseZhiqiang Chen and Jonathan S. Stamler*

The mitochondrial aldehyde dehydrogenase (ALDH2, mtALDH) was

recently found to catalyze the reduction of nitroglycerin (glyceryl

trinitrate [GTN]) to generate nitrite and 1,2-glyceryl dinitrate. The nitrite

generated within the mitochondria is metabolized further to generate

nitric oxide (NO)-based bioactivity, by reduction to NO and/or by

conversion to S-nitrosothiol, as revealed by a series of biochemical,

pharmacologic, and genetic studies. These studies also demonstrated

that mechanism-based inactivation of mtALDH is involved in the

development of GTN tolerance. In mice in which the mtALDH gene was

selectively deleted (mtALDH�/�), vascular responsiveness to low but not

to high GTN concentrations was eliminated, indicating the existence of

an additional mechanism of GTN biotransformation (bhigh KmQ path-

way). In addition, bioactivation of isosorbide dinitrate/mononitrate

vasodilators is independent of mtALDH. Induction of GTN tolerance in

vitro in aortae from normal mice selectively affected responsiveness to

low doses of GTN, and the remaining responsiveness to high doses of

GTN in mtALDH�/� vasculature did not exhibit tolerance. These

findings suggest strongly that the high Km pathway is not involved in

the development of GTN tolerance that is mechanism-based. Notably,

recent studies indicate that individuals of East Asian origin with the

common E487K mutation of mtALDH, which results in decreased

mtALDH activity, are significantly less responsive to GTN. These

observations in toto provide strong support for the conclusion that

mtALDH provides the necessary and sufficient enzymatic mechanism

for biotransformation of clinically relevant concentrations of GTN to

NO-based vasoactivity and indicate in addition that inactivation of

mtALDH plays a significant role in the development of mechanism-

based GTN tolerance. (Trends Cardiovasc Med 2006;16:259–265)D 2006, Elsevier Inc.

* Address correspondence to: Dr. Jonathan

S. Stamler, Department of Medicine, Box

2612, Duke University Medical Center, Dur-

ham, NC 27710, USA. Tel.: (+1) 919-684-6933;

fax: (+1) 919-684-6998;

e-mail: staml001@mc.duke.edu.

D 2006, Elsevier Inc. All rights reserved.

1050-1738/06/$-see front matter

Zhiqiang Chen is at the Department of Med-

icine, Duke University Medical Center, Dur-

ham, NC 27710, USA. Jonathan S. Stamler is

at the Department of Medicine, Duke Univer-

sity Medical Center, Durham, NC 27710, USA

and Department of Biochemistry, Duke Uni-

versity Medical Center, Durham, NC 27710,

USA.

TCM Vol. 16, No. 8, 2006

Nitroglycerin (glyceryl trinitrate, GTN)

has been used for well over a century to

treat angina pectoris, myocardial infarc-

tion, and heart failure, and continues to

remain a mainstay of therapy in the

management of these conditions. In

2001, GTN was prescribed for the treat-

ment of angina more than 2 million

times in the United States alone (Zaher

et al. 2004). In addition to its anti-

anginal benefits, GTN has been found

to induce ischemic preconditioning

(Leesar et al. 2001), a physiologic phe-

nomenon that can protect the heart from

lethal ischemia. Recently, GTN has also

been demonstrated to be beneficial in

noncardiovascular contexts, including

pain management (Lauretti et al. 2002,

Higa et al. 2004), treatment of chronic

anal fissure (Lindsey et al. 2004), pres-

ervation of organs for transplantation

(Loehe et al. 2004, Baxter et al. 2001),

and overall response and time to pro-

gression in patients with inoperable

small cell cancer (Yasuda et al. 2006).

However, chronic administration of

GTN is plagued by the development of

tolerance and may be associated with

pro-oxidant and atherogenic effects

(reflected in endothelial dysfunction

and perhaps matrix metalloproteinase

dysfunction; Death et al. 2002, Gori and

Parker 2002). A large-scale analysis indi-

cated significantly increased cardiovas-

cular morbidity in association with long-

term GTN use (Nakamura et al. 1999).

Despite its wide use in medicine, the

bioactivation of GTN has remained mys-

terious. It is generally assumed that

GTN is converted in smooth muscle cells

to nitric oxide (NO) or S-nitrosothiol

(SNO), which activates soluble guanylate

cyclase (sGC) and thus relaxes vascular

smooth muscle (Murad et al. 1978,

Ignarro et al. 1981). However, after dec-

ades of intensive investigation, themolec-

ular mechanisms of GTN transformation

to NO, which is closely associated with

the development of GTN tolerance

(a major clinical issue), have remained

undefined. Recent reviews have summar-

ized the ongoing controversies (Munzel

et al. 2005, Csont and Ferdinandy 2005,

259

Figure 1. Cellular metabolism of nitrogly-

cerin (GTN) yields two dinitrate products and

generates NO- and/or S-nitrosothiol-based

vasodilatory bioactivity. Bioactivation of

GTN in vascular tissue is associated predom-

inantly with production of 1,2-GDN rather

than 1,3-GDN, consistent with operation of

the mitochondrial aldehyde dehydrogenase as

a GTN reductase, as detailed in this review.

Figure 2. Preferential generation from GTN

(0.1 AM) of 1,2-GDN versus 1,3-GDN in

RAW264.7 cell lysate, as assessed by TLC

and scintillation counting (14C). See Chen

et al. (2002) for a detailed description of the

assay for GTN reductase activity.

Thatcher et al. 2004, Fung 2004, Gori

and Parker 2002, Bennett et al. 1994).

In the present review, we will summarize

recent developments that demonstrate a

central role for the mitochondrial alde-

hyde dehydrogenase (mtALDH) in cata-

lyzing GTN biotransformation.

� Mitochondrial Aldehyde

Dehydrogenase-Catalyzed GTN

Biotransformation. How Was It

Discovered?

Several candidate enzymes, including

glutathionine-S -transferases (GSTs),

cytochrome P450 reductase/cytochrome

P450, and xanthine oxidoreductase, have

been proposed to catalyze GTN biotrans-

formation. However, their possible roles

in generating NO bioactivity from GTN

remain controversial (Chen et al. 2002).

For example, although GSTs can metab-

olize GTN, the GST inhibitor sulfobro-

Figure 3. Multifunctional enzymatic mecha-

nisms of mtALDH. In acetaldehyde dehydro-

genation (shown at top), hydride transfer

results in the formation of NADH and an acyl

intermediate, followed by hydrolysis that

releases acetic acid. Esterase activity (shown

for the model ester substrate p-nitrophenyla-

cetate) entails acyl intermediate formation

followed by hydrolysis, without NADH for-

mation. Note that (as shown at bottom), if

mtALDH functioned as an esterase to catalyze

1,2-GDN formation from GTN, then inorganic

nitrate, rather than nitrite as observed, would

be produced.

Table 1. Glyceryl trinitrate concen-tration in human plasma

Route ofadministration

GTN concentration

ng/L* nM

Sublingal or oral 0.10–3.96 0.44–17.44

Transdermal 0.10–9.24 0.44–40.70

Intravenenous 0.40–120 1.76–528.6

* Values fom Hashimoto and Kobayashi

(2003).

260

mophthalein has no effect on GTN-

induced increases in cyclic guanosine

monophosphate (cGMP). Moreover, a

homozygous deletion of GSTA, which

has the highest GTN-metabolizing activ-

ity among GSTs in the vascular wall, has

no affect on GTN responsiveness in

humans (Haefeli et al. 1993). More gen-

erally, none of the candidate enzymes

preferentially catalyzes 1,2-GDN forma-

tion, which is the predominant dinitrate

product of vasculature GTN biotransfor-

mation (the minor dinitrate product is

1,3-GDN) (Figure 1) and related to devel-

opment of GTN tolerance.

The ideal strategy to identify a GTN-

bioactivating enzyme(s) entails the

purification of NO generating and/or

GTN-metabolizing activity using clini-

cally relevant GTN levels, which are

below 100 nM for most clinical applica-

tions (Table 1) (Hashimoto and Kobaya-

shi 2003). However, measuringNO at low

nanomolar concentrations is technically

difficult, and in fact most early studies

monitored degradation of GTN adminis-

tered at doses of 0.1–1 mM, which are at

least 1000-fold higher than levels of GTN

used clinically. Under these conditions,

metabolism of GTN obscures biotrans-

formation. Analytical techniques

adequate to detect picomole amounts of

GDNs, including silica gel thin-layer

chromatography (TLC)/liquid scintilla-

tion spectrometry using radioisotope-

labeled GTN, have been available for

decades. Using the TLC/liquid scintilla-

tion spectrometry technique to follow

1,2-GDN and 1,3-GDN formation from

physiologic amounts of 14C-labeled GTN

(0.1 AM), we screened animal tissues and

cell lines and found that mouse macro-

phage RAW264.7 cells, which can be

grown up in large numbers, metabolize

GTN to generate predominantly 1,2-GDN

(Figure 2). We developed a four-step

column chromatographic procedure to

purify this activity to homogeneity from

RAW264.7 cells and identified the

responsible enzyme by N-terminal se-

quencing as mtALDH (Chen et al. 2002).

Mitochondrial aldehyde dehydrogen-

ase is a major enzyme in ethanol metab-

olism, catalyzing aldehyde oxidation to

form acetic acid. In addition to this

dehydrogenase activity, mtALDH also

exhibits esterase activity (Feldman and

Weiner 1972). Mitochondrial aldehyde

dehydrogenase-catalyzed GTN reduction

to form 1,2-GDN (reductase activity) is a

previously unappreciated function

requiring supplementation of reductant

(see below), although it is well known

that GTN (and isosorbide dinitrate,

ISDN) can inhibit mtALDH activity in

vitro (alcohol-nitrovasodilator drug

interaction) (Mukerjee and Pietruszko

1994, Pietruszko et al. 1995).

� Characterization of mtALDH-

Catalyzed GTN Biotransformation

Mitochondrial aldehyde dehydrogenase

purified from bovine liver catalyzes pre-

dominantly 1,2-GDN formation (1,2-

GDN/1,3-GDN, ~8:1) from low levels of

GTN (1 AM). In the presence of NAD+,

the 1,2-GDN/1,3-GDN ratio increases to

TCM Vol. 16, No. 8, 2006

Figure 5. Alternate mechanisms of nitrite

formation from GTN by mtALDH. When

functioning as an aldehyde dehydrogenase

through an acyl intermediate (shown at top),

a water molecule is activated by Glu268 and

the resultant hydroxide ion will attack the

carbonyl carbon center to form acetic acid.

However, if the intermediate is a thionitrate,

the hydroxide may attack the sulfur of cys 302

to form sulfenic acid and release nitrite (T.

Hurley, personal communication). Alterna-

tively, attack of the thionitrate intermediate

by an adjacent cysteine thiol (cys 301 or cys

303) will also release nitrite but form a

disulfide at the active site.

more than 20:1 and the rate of GTN

metabolism increases ~10-fold. The opti-

mum pH of 9.0 is about the same as that

of aldehyde dehydrogenase activity

(Chen et al. 2002). In a subsequent

analysis of overexpressed human

mtALDH, Kollau et al. (2005) reported

that NAD+ addition resulted in a fourfold

increase in the rate of 1,2-GDN forma-

tion and an increase in the 1,2-GDN/1,3-

GDN ratio from 9.8 to 137.

Based on the reaction mechanism for

the esterase activity of mtALDH, the

analogous reaction with GTN would be

expected to yield an mtALDH-S-NO2

intermediate that would then undergo

hydrolysis to generate nitrate (Figure 3)

(Mukerjee and Pietruszko 1994). How-

ever, we detected stoichiometric forma-

tion from GTN of 1,2-GDN and nitrite.

Thus, the overall reaction can be

expressed as:

GTNþmtALDHredY1; 2� GDN

þ NO�2 þmtALDHox ð1Þ

Nitrite formation indicates that the

presumed mtALDH-SNO2 intermediate

is reduced and not hydrolyzed (as in the

case of dehydrogenation of aldehyde,

where the intermediate is mtALDH-

COCH3). Mechanism-based oxidation

and inactivation of mtALDH therefore

may involve formation of a disulfide

Figure 4. Dependence on thiol reductant of

mtALDH-catalyzed conversion of GTN (1 AM)

to 1,2-GDN in vitro. Purified bovine liver

mtALDH was assayed in buffer containing

100 AM NAD and additional co-factors, as

indicated. Degree of activation is expressed

relative to the maximal activation produced

by dithiothreitol (DTT). DHLP, dihydrolipoic

acid; TCEP, tris(2-carboxyethyl) phosphine;

h-ME, h-mercapto-ethanol; NAC, N-acetyl

cysteine; Cys, cysteine; Vc, ascorbic acid;

GSH, glutathione; NADH, nicotinamide

adenine diphosphate.

TCM Vol. 16, No. 8, 2006

that includes the active site Cys thiol

(Figure 4). Analysis of purified mtALDH

in vitro indicated that, to return to the

active state, the oxidized enzyme must

be reduced by exogenous thiols or other

reductants. We examined the reactiva-

tion by different reductants of mtALDH

isolated from bovine liver and found

that dithiothreitol is the most efficient

reductant tested, that lipoic acid is

more effective than 2-mercaptoethanol,

and that the non-thiol reductant tris[2-

carboxyethyl] phosphine is also an effi-

cient re-activator of oxidized mtALDH

(Figure 4). These results are consistent

with mechanism-based oxidative forma-

tion of a disulfide bond at the active site

of mtALDH, although other thiol oxida-

tion states such as sulfenic acid (–SOH),

which may form by attack of hydroxide

on the sulfur of cys 302 (Figure 5),

cannot be excluded (T. Hurley, personal

communication), and are also consis-

tent with previous suggestions that

active site disulfide formation underlies

the inhibition of mtALDH by NO and by

some ALDH inhibitors (DeMaster et al.

1997, Vallari and Pietruszko 1982, Shen

et al. 2000).

In human mtALDH, there are two

cysteine residues (cys 301 and cys 303)

adjacent to and flanking the active site

cys 302. The specific mechanism for

GTN reductase activity of mtALDH that

we have proposed entails formation of a

disulfide between one of these cys

residues and cys 302 (Chen et al.

2002), although formation of a disulfide

bond between adjacent cysteine resi-

dues is unfavorable unless one residue

is a selenocysteine (Zhong et al. 2000).

Mass spectrometric evidence supports

disulfide formation at the active site of

mtALDH in vitro but not in vivo (Shen

et al. 2000), and further analysis will be

required to discriminate definitively

between alternative reaction mecha-

nisms. In particular, alternative oxida-

tive derivatives such as RS(O)SR merit

consideration.

� Inhibition of mtALDH-Mediated

GTN Reductase Activity by ALDH

Inhibitors

Our finding that mtALDH functions as

a GTN reductase provides a new expla-

nation for the inhibitory effect of GTN

on aldehyde dehydrogenase activity

(alcohol-GTN drug interaction) and also

indicates that ALDH substrates and

inhibitors should suppress GTN reduc-

tase activity. We examined the effects of

the substrate acetaldehyde and of differ-

ent classes of ALDH inhibitors on GTN

bioactivation.

1,2-GDN formation by purified bovine

livermtALDHwas inhibited by the classic

substrate analog ALDH inhibitor, chloral

hydrate, and by the ALDH substrate,

acetaldehyde, with an IC50 of 99.5 and

815.5 AM, respectively (Chen et al. 2002).

The relatively high IC50 values as com-

pared with the reported K i (chloral

hydrate) and Km (acetaldehyde) for alde-

hyde dehydrogenase activity probably

reflect the absence of NAD+, which is

known to increase acetaldehyde binding

100-fold (Feldman and Weiner 1972). As

expected, cyanamide did not inhibit puri-

fied mtALDH because it requires bioacti-

vation by catalase (DeMaster et al. 1984).

Chloral hydrate and the specificmtALDH

inhibitor daidzin were subsequently

found to inhibit catalysis by human

mtALDH of 1,2-GDN formation (Sydow

et al. 2004, Kollau et al. 2005). It is of

interest that Daiber et al. (2004) have

reported that the ALDH inhibitor

benomyl inhibits the biotransformation

of GTN (as well as of pentaerythritol

tetranitrate and pentaerythritol trini-

trate), but not of ISDN and the isosorbide

261

mononitrates (ISMNs), consistent with

biotransformation of the isosorbide

nitrovasodilators by an mtALDH-inde-

pendent pathway (Chen et al. 2005).

It was also demonstrated that treat-

ment of aortic rings (obtained from

mouse, rat, or rabbit) with ALDH inhib-

itors, including chloral hydrate, cyana-

mide, benomyl, and daidzin, or with the

ALDH substrate acetaldehyde inhibited

GTN-induced relaxation (Chen et al.

2002, Kollau et al. 2005, Sydow et al.

2004, de la Lande et al. 2004). These

inhibitors do not attenuate sodium nitro-

prusside (SNP)-induced relaxation, indi-

cating that they exert their effect through

specific inhibition of vascular GTN

reductase activity, and inhibition of

GTN reductase and aldehyde dehydro-

genase activities was confirmed directly

by biochemical assay of aortic rings.

These results support the conclusion

that nitrite, generated by the GTN-reduc-

tase activity of mtALDH (Eq. (1)), is

further reduced within the mitochondria

to generate NO/SNO, and that this NO

bioactivity is largely responsible for the

increase in cGMP that mediates GTN-

induced vasorelaxation.

I n m tALDH kno c k ou t m i c e

(mtALDH�/�), the hypotensive effect of

intravenously infused GTN was elimina-

ted at low doses of GTN and substan-

tially decreased at high doses, in the

absence of effects on SNP-induced vaso-

relaxation (Chen et al. 2005). Inhibition

of mtALDH activity by the ALDH inhib-

itors chloral hydrate and cyanamide also

attenuated the decrease in blood pres-

sure induced by intravenous infusion of

GTN in dogs, rabbits, and rats (Chen

et al. 2002, Zhang et al. 2004), confirm-

ing a role for mtALDH in GTN biotrans-

formation in vivo. In addit ion,

pretreatment with cyanamide, adminis-

tered via the coronary artery, signifi-

cantly attenuated the temporary

increase in coronary blood flow that is

induced by GTN infusion, indicating

an action of mtALDH-derived NO vaso-

activity on the coronary vascular bed

in vivo.

� Mitochondrial Nitrite Reductase

Activity

The results discussed above suggest that

nitrite generated from GTN by mtALDH

within the mitochondria (Eq. (1)) is

262

further reduced or converted to NO

bioactivity and that this NO bioactivity

is exported from mitochondria to the

cytosol, where it activates sGC (sGC is

not detected in purified mitochondria by

immunoblotting; Chen and Stamler,

unpublished observation). To demon-

strate directly the generation and export

of NO bioactivity by mitochondria, we

used a reporter assay to test whether

addition of GTN and isolated mitochon-

dria results in the activation of sGC in rat

fibroblast cells (RFL-6 cells). This assay

is known to be very sensitive to NO, but

insensitive to GTN. Coincubation of

RFL-6 cells with mitochondria and vary-

ing concentrations of GTN resulted in

the dose-dependent generation of cGMP.

Generation of cGMP was blocked by the

addition of the ALDH inhibitor chloral

hydrate or of hemoglobin, an NO scav-

enger (Chen et al. 2005). These results,

as well as the finding of attenuated GTN-

induced production of cGMP by p0 cells

(Sydow et al. 2004), which are deficient

in mitochondrial function, demonstrate

the ability of intact mitochondria to

generate and export NO bioactivity.

It has been reported that mammalian

mitochondria express a nitrite reductase

activity and that this activity is associ-

ated with the cytochrome bc1 complex

(complex III) of the respiratory chain

(Kozlov et al. 1999). Nitrite might also be

transported to the intermembrane space,

where the high proton concentration

may facilitate its conversion to NO or

SNO. In addition, cytochrome c oxidase

(complex IV) may also reduce nitrite to

NO (Brudvig et al. 1980). The physio-

logic significance of the mitochondrial

nitrite reductase activity is not known,

and how it is regulated and how its

product, NO, affects mitochondrial func-

tion require further investigation.

Recent results of Nunez et al. (2005)

suggest that GTN-induced activation of

sGC does not involve free NO. In con-

trast, Kollau et al. (2005) have suggested

that mtALDH could generate NO directly

from GTN (without a nitrite intermedi-

ate). However, only high doses of GTN

were used, and no mechanism for the

three-electron reduction of GTN to NO

was provided. The form in which NO

bioactivity is conveyed from mitochon-

dria to cytosolic sGC remains unre-

solved. We note in this regard that

intramitochondrial SNOs are increased

by GTN (unpublished observation).

� Glyceryl Trinitrate Metabolism and

Bioactivity in mtALDH

Knockout Mice

Biochemistry

As expected, GTN biotransformation to

1,2-GDN is essentially eliminated in the

mitochondria of mtALDH�/� mice

(N90% decrease compared to wild-type

mitochondria), whereas 1,3-GDN pro-

duction is not affected. Accordingly,

mtALDH�/� mitochondria almost com-

pletely lose the ability to generate cGMP

from 1 AM GTN in the RFL-6 cell

reporter assay (Chen et al. 2005).

In intact aorta, biotransformation of

GTN to 1,2-GDN was eliminated at low

GTN concentrations (b1 AM) and signifi-

cantly attenuated at 1–10 AM GTN. The

remaining 1,2-GDN generating activity,

and the preferential generation of 1,3-

GDN at higher GTN concentrations

(N1 AM), indicates the existence of an

additional mechanism that operates at

relatively high GTN concentrations (high

Km pathway). Accordingly, cGMP pro-

duction after incubation of intact

mtALDH�/� aortic segments with low

concentrations of GTN (b1 AM) was

eliminated and was attenuated but not

eliminated at higher concentrations

(10 AM GTN) (Chen et al. 2005).

Physiologic Studies

We observed that the dose-response

curve for GTN-induced relaxation was

biphasic for wild-type mouse aorta, with

a distinct region of reduced responsive-

ness at ~1 AM GTN, consistent with

previous findings (Chen et al. 2005). In

contrast, the dose-response curve from

mtALDH�/� aorta appears monophasic,

with the most sensitive phase of the wild-

type curve (relaxation to GTN concen-

trations b1 AM) completely eliminated. In

contrast, SNP-induced vasorelaxation

was not affected in the knockout mouse

aorta, which indicates that loss of the

sensitive phase in the GTN dose-response

curve is due to the loss of mtALDH-

catalyzed generation of NO bioactivity

from GTN and not to attenuated respon-

siveness to NO-related vasoactivity (Chen

et al. 2005). In addition, hypotension

induced by GTN infusion in intact

mtALDH�/� mice was eliminated at low

GTN concentrations but not at high GTN

levels (Chen et al. 2005). We also deter-

mined that there is no change in vaso-

TCM Vol. 16, No. 8, 2006

relaxation by ISDN or the ISMNs of

mtALDH�/� aorta in vitro, indicating that

isosorbide nitrovasodilators are not bio-

activated by mtALDH. Interestingly,

Murphy et al. (2005) recently showed

that, in rat aorta, GTN but not ISDN can

inhibit mtALDH.

Mechanism-Based GTNTolerance

As summarized recently (Munzel et al.

2005), GTN tolerance is a complex

phenomenon. Impairment of GTN bio-

transformation, desensitization of sGC,

and inactivation of cGMP-dependent

kinase and myosin light-chain phos-

phorylation would all decrease vaso-

relaxation to GTN, that is, foster GTN

tolerance. In vivo, additional factors,

including oxidative stress and volume

redistribution, may play contributory

roles. Impairment of GTN biotransfor-

mation underlies mechanism-based or

classic tolerance. Mitochondrial alde-

hyde dehydrogenase inactivation, either

because of irreversible modification of

Figure 6. Tolerance to nitrovasodilators.

Mechanism-based tolerance to GTN (indi-

cated in red) is induced by oxidative inactiva-

tion of mtALDH, which can reflect depletion

of the endogenous reducing cofactor(s) that

subserves redox cycling of active site Cys thiol

(bthiol depletionQ) and/or the GTN-dependent

formation of higher oxidation state(s) of the

catalytic thiol. Increased mitochondrial pro-

duction of reactive oxygen species consequent

upon chronic administration of GTN (indi-

cated in green) may also play a role in

oxidative inactivation of mtALDH and/or its

reducing cofactor(s), and increased reac-

tive oxygen species production may also

contribute to GTN-dependent inactivation of

enzymes, other than mtALDH, which

subserve the bioactivation of other nitrovaso-

dilators such as ISMN and ISDN (cross-

tolerance). In addition, inhibitory effects on

downstream elements that transduce the

nitrovasodilator-derived and (S)NO-based

vasodilatory stimulus are possible.

TCM Vol. 16, No. 8, 2006

the active site cysteine residue (cys

302), or because of depletion of bthiolcofactorQ necessary for the GTN reduc-

tase catalytic cycle, provides a molec-

ular mechanism for mechanism-based

GTN tolerance. Mitochondrial aldehyde

dehydrogenase also becomes a conver-

gence point of the bsuperoxide hypoth-

esisQ and the bthiol depletion theoryQ ofnitrate tolerance. Although the identity

of the in situ bthiol reductantQ is not

known, the requirement for such a

reductant provides an explanation for

the bthiol depletion theoryQ of nitrate

tolerance. Superoxide (or derived per-

oxynitrite)-mediated mtALDH inactiva-

tion by thiol oxidation is also possible

(Figure 6). Daiber et al. (2005) reported

recently that GTN-stimulated superox-

ide production correlated well with

decreases in mtALDH activity. Chronic

treatment with GTN, but not occasional

doses of GTN, inhibits erythrocytic

ALDH activity, supporting a relation-

ship between diminished mtALDH

activity and the development of GTN

tolerance (Towell et al. 1985). Super-

oxide-induced inhibition of the enzymes

subserving bioactivation of other

organic nitrates, and/or of downstream

elements in the signaling pathway trans-

ducing vasorelaxation, may contribute

to cross-tolerance (Figure 6).

DiFabio et al. (2003) reported that

cyanamide and propionaldehyde inhibit

GTN-induced vasodilation, but also that

these inhibitors can further right-shift

the GTN-vasorelaxation dose-response

curve of tolerant aorta. Based on these

observations, they suggested that

mtALDH plays only a minor role in

GTN biotransformation. However, as

pointed out by de la Lande et al.

(2004), the model of tolerance used

may generate only mild to moderate

tolerance and the vasculature may retain

significant mtALDH activity [DiFabio

et al. (2003) did not provide measure-

ments of vascular mtALDH activity].

Regardless of the quantitative relation-

ship between the remaining mtALDH

activity and GTN bioactivation in toler-

ant vasculature, the results from

mtALDH�/� mice clearly demonstrate

that, in the absence of mtALDH activity,

ALDH inhibitors can no longer right-

shift the GTN-vasorelaxation dose-

response curve following the induction

of tolerance by GTN exposure in vitro

(Chen et al. 2005).

The nature of the oxidative modifica-

tion of the mtALDH active site cysteine

(cys 302) will be key in determining the

extent to which tolerance-related inacti-

vation of enzyme activity is reversible.

Covalent modification by disulfiram,

benomyl, cyanamide, or by nitrosylation

(Moon et al. 2005), disulfide formation

with an adjacent cysteine thiol (which

may be induced by NO), or other possi-

ble oxidation states of the cys 302 thiol

generated by superoxide will all lead to

inactivation of mtALDH. Whereas the

reversible modifications, including disul-

fide formation, could be reduced by the

as-yet-unidentified in situ thiol reduc-

tant, irreversible (or very slowly rever-

sible) modifications, such as –SO2H,

would result in prolonged mechanism-

based tolerance.

� Clinical Implications

Nitrovasodilator therapy has not pro-

vided survival benefits. It is possible that

mitochondrial damage and impaired

respiration (Needleman and Hunter

1966) that are consequent upon the

localization of GTN biotransformation

to mitochondria may provide at least a

partial explanation for the failure of

long-term GTN therapy to improve mor-

tality. Glyceryl trinitrate-related adverse

atherogenic effects and increased mor-

bidity also merit further investiga-

tion (Nakamura et al. 1999, Gori and

Parker 2002).

Pharmacogenetics

A polymorphism in the mtALDH gene,

the E487K mutation that is also known

as the oriental mutant, is prevalent in

East Asian populations. It has been

reported that 30% to 50% of Chinese,

Japanese, and Koreans (about 400–700

million individuals) carry this mutation,

either as a heterozygous or as a homo-

zygous carrier. The phenotype of this

mutation includes facial flushing after

ingestion of alcohol, which is a result of

lowered mtALDH activity that is caused

in turn by decreased NAD+ binding

(Larson et al. 2005). Thus, carriers

inefficiently metabolize the acetaldehyde

generated by alcohol oxidation. On the

basis of our findings, it may be predicted

that low mtALDH activity in E487K

carriers will be associated with reduced

vascular responsiveness to GTN. Two

263

recent clinical studies confirmed this

prediction (Mackenzie et al. 2005, Li

et al. 2006). Moreover, treatment of

normal subjects with disulfiram, an

ALDH inhibitor in clinical use, resulted

in decreased GTN responsiveness sim-

ilar to that seen in E487K mutation

carriers. Although large interindividual

variability in responsiveness to GTN has

been recognized over the more than a

century of GTN use (Haefeli et al. 1993),

a genetic signature (mtALDH polymor-

phism) directly linked to GTN respon-

siveness in humans has not been

reported previously. For patients with

the E487K mutation, altered dosages of

GTN or substitution of GTN by other

nitrates such as ISDN or ISMNs, which

are not bioactivated by mtALDH, should

be considered in clinical practice, and

clinical trials may be warranted.

Drug Interactions

A number of compounds in use clini-

cally, including the ALDH inhibitors

disulfiram, chloral hydrate, and the

analgesic acetaminophen, inhibit

mtALDH activity. Accordingly, co-

administration of these drugs with GTN

may be counter-indicated.

� Summary and Perspective

Recent biochemical, pharmacologic, and

genetics studies have demonstrated

clearly that mtALDH is the essential

enzyme in the biotransformation of

clinical levels of GTN, and that it plays

a significant role in the development of

GTN tolerance. These studies, and in

particular analyses in mtALDH knock-

out mice, also indicate that there is an

additional mechanism for GTN biotrans-

formation (high Km pathway), but that

this pathway probably plays no role in

the development of mechanism-based

GTN tolerance. In addition, mtALDH is

not involved in the bioactivation of the

isosorbide nitrovasodilators. Identifica-

tion of the enzyme(s) that subserves the

high Km pathway for GTN bioactivation,

and of the enzyme(s) that subserves

bioactivation of ISDN and the ISMNs,

remains an important task with signifi-

cant clinical implications.

Several important aspects of the

mtALDH-dependent pathway for GTN

biotransformation remain incompletely

characterized, including the identity of

264

the endogenous (presumably) thiol

reductant that participates in redox

cycling of mtALDH, the identity of the

mitochondrial nitrite reductase activity,

and the form and mechanism of export

of mitochondrially derived NO bioactiv-

ity. Clarification of these issues should

provide guidance in the effort to develop

clinical strategies to prevent the develop-

ment of, or reverse, nitrate tolerance

and, more generally, in the development

of strategies to minimize cellular dam-

age in the treatment of cardiovascular

diseases with nitrovasodilators (Ignarro

2002, Mayer 2003). Additionally, our

results discussed here suggest that a

clinical reassessment of the risk/benefit,

dosages, and responsiveness to GTN is

warranted and indicate the possibility

that GTN and ISDN/ISMN might pro-

vide independent clinical benefits.

Finally, the demonstrated ability of

mitochondria to generate and export NO

bioactivity from nitrite points to the

intriguing possibility that mitochondria

may serve as a source of endogenous

NO-based cellular signals that are inde-

pendent of NO synthase.

� Acknowledgments

We thank Dr. Douglas T. Hess for

valuable discussions and critical reading

and editing of the manuscript.

R

eferences

Baxter K, Howden BO, Jablonski P: 2001.

Heart preservation with Celsior solution

improved by the addition of nitroglycerine.

Transplantation 71:1380–1384.

Bennett BM, McDonald BJ, Nigam R, et al:

1994. Biotransformation of organic nitrates

and vascular smooth muscle cell function.

Trends Pharmacol Sci 15:245–249.

Brudvig GW, Stevens TH, Chan SI: 1980.

Reactions of nitric oxide with cytochrome

c oxidase. Biochemistry 19:5275–5285.

Chen Z, Zhang J, Stamler JS: 2002. Identi-

fication of the enzymatic mechanism of

nitroglycerin bioactivation. Proc Natl Acad

Sci U S A 99:8306–8311.

Chen Z, Foster MW, Zhang J, et al: 2005.

An essential role for mitochondrial alde-

hyde dehydrogenase in nitroglycerin bio-

activation. Proc Natl Acad Sci U S A 102:

12159–12164.

Csont T, Ferdinandy P: 2005. Cardioprotective

effects of glyceryl trinitrate: Beyond vascu-

lar nitrate tolerance. Pharmacol Ther

105:57–68.

Daiber A, Oelze M, Coldewey M, et al: 2004.

Oxidative stress and mitochondrial alde-

hyde dehydrogenase activity: A compari-

son of pentaerythritol tetranitrate with

other organic nitrates. Mol Pharmacol

66:1372–1382.

Daiber A, Oelze M, Sulyok S, et al: 2005.

Heterozygous deficiency of manganese su-

peroxide dismutase in mice (Mn-SOD+/�): A

novel approach to assess the role of oxida-

tive stress for the development of nitrate

tolerance. Mol Pharmacol 68:579–588.

Death AK, Nakhla S, McGrath KC, et al: 2002.

Nitroglycerin upregulates matrix metallo-

proteinase expression by human macro-

phages. J Am Coll Cardiol 39:1943–1950.

De la Lande IS, Stepien JM, Philpott AC, et al:

2004. Aldehyde dehydrogenase, nitric oxide

synthase and superoxide in ex vivo nitrate

tolerance in rat aorta. Eur J Pharmacol

496:141–149.

DeMaster EG, Shirota FN, Nagasawa HT:

1984. The metabolic activation of cyana-

mide to an inhibitor of aldehyde dehydro-

genase is catalyzed by catalase. Biochem

Biophys Res Commun 122:358–365.

DeMaster EG, Redfern B, Quast BJ, et al:

1997. Mechanism for the inhibition of

aldehyde dehydrogenase by nitric oxide.

Alcohol 14:181–189.

DiFabio J, Ji Y, Vasiliou V, et al: 2003. Role

of mitochondrial aldehyde dehydrogenase

in nitrate tolerance. Mol Pharmacol 64:

1109–1116.

Feldman RI, Weiner H: 1972. Horse liver

aldehyde dehydrogenase: II. Kinetics and

mechanistic implications of the dehydro-

genase and esterase activity. J Biol Chem

247:267–272.

Fung HL: 2004. Biochemical mechanism of

nitroglycerin action and tolerance: Is this

old mystery solved? Annu Rev Pharmacol

Toxicol 44:67–85.

Gori T, Parker JD: 2002. The puzzle of nitrate

tolerance: Pieces smaller than we thought?

Circulation 106:2404–2408.

Haefeli WE, Srivastava N, Kelsey KT, et al:

1993. Glutathione S-transferase polymor-

phism does not explain variation in nitro-

glycerin responsiveness. Clin Pharmacol

Ther 53:463–468.

Hashimoto S, Kobayashi A: 2003. Clinical

pharmacokinetics and pharmacodynamics

of glyceryl trinitrate and its metabolites.

Clin Pharmacokinet 42:205–221.

Higa M, Takasu N, Tamanaha T, et al: 2004.

Nitroglycerin spray rapidly improves pain

in a patient with chronic painful diabetic

neuropathy. Diabet Med 21:1053–1054.

Ignarro LJ: 2002. After 130 years, the molec-

ular mechanism of action of nitroglycerin

is revealed. Proc Natl Acad Sci U S A

99:7816–7817.

Ignarro LJ, Lippton H, Edwards JC, et al:

1981. Mechanism of vascular smooth

TCM Vol. 16, No. 8, 2006

muscle relaxation by organic nitrates,

nitrites, nitroprusside and nitric oxide:

Evidence for the involvement of S-nitro-

sothiols as active intermediates. J Pharma-

col Exp Ther 218:739–749.

Kollau A, Hofer A, Russwurm M, et al: 2005.

Contribution of aldehyde dehydrogenase to

mitochondrial bioactivation of nitrogly-

cerin: Evidence for the activation of puri-

fied soluble guanylate cyclase through

direct formation of nitric oxide. Biochem

J 385:769–777.

Kozlov AV, Staniek K, Nohl H: 1999. Nitrite

reductase activity is a novel function of

mammalian mitochondria. FEBS Lett

454:127–130.

Larson HN, Weiner H, Hurley TD: 2005.

Disruption of the coenzyme binding site

and dimer interface revealed in the crystal

structure of mitochondrial aldehyde dehy-

drogenase bAsianQ variant. J Biol Chem

280:30550–30556.

Lauretti GR, Perez MV, Reis MP, et al: 2002.

Double-blind evaluation of transdermal

nitroglycerine as adjuvant to oral morphine

for cancer pain management. J Clin Anesth

14:83–86.

Leesar MA, Stoddard MF, Dawn B, et al: 2001.

Delayed preconditioning-mimetic action

of nitroglycerin in patients undergoing

coronary angioplasty. Circulation 103:

2935–2941.

Li Y, Zhang D, Jin W, et al: 2006. Mitochon-

drial aldehyde dehydrogenase-2 (ALDH2)

Glu504Lys polymorphism contributes to

the variation in efficacy of sublingual nitro-

glycerin. J Clin Invest 116:506–511.

Lindsey I, Jones OM, Cunningham C, et al:

2004. Chronic anal fissure. Br J Surg

91:270–279.

Loehe F, Preissler G, Annecke F, et al: 2004.

Continuous infusion of nitroglycerin

improves pulmonary graft function of

non-heart-beating donor lungs. Transplan-

tation 77:1803–1808.

Mackenzie IS, Maki-Petaja KM, McEniery

CM, et al: 2005. Aldehyde dehydrogenase

2 plays a role in the bioactivation of nitro-

TCM Vol. 16, No. 8, 2006

glycerin in humans. Arterioscler Thromb

Vasc Biol 25:1891–1895.

Mayer B: 2003. Bioactivation of nitrogly-

cerin—a new piece in the puzzle. Angew

Chem 42:388–391.

Moon KH, Kim BJ, Song BJ: 2005. Inhibition

of mitochondrial aldehyde dehydrogenase

by nitric oxide-mediated S-nitrosylation.

FEBS Lett 579:6115–6120.

Mukerjee N, Pietruszko R: 1994. Inactivation

of human aldehyde dehydrogenase by

isosorbide dinitrate. J Biol Chem 269:

21664–21669.

Munzel T, Daiber A, Mulsch A: 2005. Explain-

ing the phenomenon of nitrate tolerance.

Circ Res 97:618–628.

Murad F, Mittal CK, Arnold WP, et al: 1978.

Guanylate cyclase: Activation by azide,

nitro compounds, nitric oxide, and

hydroxyl radical and inhibition by hemo-

globin and myoglobin. Adv Cyclic Nucleo-

tide Res 9:145–158.

Murphy TC, Arntzen R, Picklo MJ: 2005.

Nitrate-based vasodilators inhibit multiple

vascular aldehyde dehydrogenases. Cardio-

vasc Toxicol 5:321–332.

Nakamura Y,Moss AJ, BrownMW, et al: 1999.

Long-term nitrate use may be deleterious in

ischemic heart disease: A study using the

databases from two large-scale postinfarc-

tion studies. Am Heart J 138:577–585.

Needleman P, Hunter Jr FE: 1966. Effects of

organic nitrates on mitochondrial respira-

tion and swelling: Possible correlations

with the mechanism of pharmacologic

action. Mol Pharmacol 2:134–143.

Nunez C, Victor VM, Tur R: 2005. Discrep-

ancies between nitroglycerin and NO-

releasing drugs on mitochondrial oxygen

consumption, vasoactivity, and the release

of NO. Circ Res 97:1063–1069.

Pietruszko R, Mukerjee N, Blatter EE, et al:

1995. Nitrate esters as inhibitors and sub-

strates of aldehyde dehydrogenase. Adv Exp

Med Biol 372:25–34.

Shen ML, Lipsky JJ, Naylor S: 2000. Role of

disulfiram in the in vitro inhibition of rat

liver mitochondrial aldehyde dehydrogen-

ase. Biochem Pharmacol 60:947–953.

Sydow K, Daiber A, Oelze M, et al: 2004.

Central role of mitochondrial aldehyde

dehydrogenase and reactive oxygen species

in nitroglycerin tolerance and cross-toler-

ance. J Clin Invest 113:482–489.

Thatcher GR, Nicolescu AC, Bennettm BM,

et al: 2004. Nitrates and NO release: Con-

temporary aspects in biological and medic-

inal chemistry. Free Radic Biol Med 37:

1122–1143.

Towell J, Garthwaite T, Wang R: 1985.

Erythrocyte aldehyde dehydrogenase and

disulfiram-like side effects of hypoglyce-

mics and antianginals. Alcohol Clin Exp

Res 9:438–442.

Vallari RC, Pietruszko R: 1982. Human alde-

hyde dehydrogenase: Mechanism of inhib-

ition of disulfiram. Science 216:637–639.

Yasuda H, Yamada M, Nakayama K, et al:

2006. Randomized phase II trial comparing

nitroglycerin plus vinorelbine and cisplatin

with vinorelbine and cisplatin alone in

previously untreated stage IIIB/IV non-

small-cell lung cancer. J Clin Oncol

24:688–694.

Zaher C, Goldberg GA, Kadlubek P: 2004.

Estimating angina prevalence in a managed

care population. Am J Manag Care

10:S339–S346.

Zhang J, Chen Z, Cobb FR, et al: 2004. Role of

mitochondrial aldehyde dehydrogenase in

nitroglycerin-induced vasodilation of coro-

nary and systemic vessels: An intact canine

model. Circulation 110:750–755.

Zhong L, Arner ES, Holmgren A: 2000.

Structure and mechanism of mammalian

thioredoxin reductase: The active site is a

redox-active selenolthiol/selenenylsulfide

formed from the conserved cysteine–sele-

nocysteine sequence. Proc Natl Acad Sci

U S A 97:5854–5859.

PII S1050-1738(06)00077-6 TCM

265