De Novo design and utilization of photolabile caged substrates as probes of hydrogen tunneling with horse liver alcohol dehydrogenase at sub-zero temperatures: A cautionary note

Article abstract of DOI:10.1016/S0045-2068(03)00018-X

In order to understand the influence of protein dynamics on enzyme catalysis and hydrogen tunneling, the horse liver alcohol dehydrogenase (HLADH) catalyzed oxidation of benzyl alcohol was studied at sub-zero temperatures. Previous work showed that wild t

Full text of DOI:10.1016/S0045-2068(03)00018-X

BIOORGANIC
CHEMISTRY
Bioorganic Chemistry 31 (2003) 172–190
www.elsevier.com/locate/bioorg
De Novo design and utilization of
photolabile caged substrates as probes
of hydrogen tunneling with horse liver
alcohol dehydrogenase at sub-zero
temperatures: a cautionary noteq
b,c,
Shiou-Chuan Tsaia and Judith P.Klinman
*
a
Department of Chemical Engineering, Stanford University, California 94305, USA
b
Department of Chemistry, University of California at Berkeley, CA 94720, USA
c
Department of Molecular and Cell Biology, University of California at Berkeley, CA, 94720, USA
Received 20 September 2002
Abstract
In order to understand the influence of protein dynamics on enzyme catalysis and hydrogen
tunneling, the horse liver alcohol dehydrogenase (HLADH) catalyzed oxidation of benzyl al-
cohol was studied at sub-zero temperatures.Previous work showed that wild type HLADH
has significant kinetic complexity down to )50 °C due to slow binding and loss of substrate
[S.-C. Tsai, J.P. Klinman, Biochemistry, 40 (2001) 2303]. A strategy was therefore undertaken
to reduce kinetic complexity at sub-zero temperatures, using a photolabile (caged) benzyl al-
cohol that prebinds to the enzyme and yields the active substrate upon photolysis.By com-
puter modeling, a series of caged alcohols were designed de novo, synthesized, and
characterized with regard to photolysis and binding properties.The o-nitrobenzyl ether 15,
with a unique long tail, was found to be most ideal.At sub-zero temperatures in 50% MeOH,
a two-phase kinetic trace and a rate enhancement by the use of 15 were observed.Despite the
elimination of substrate binding as a rate-limiting step, the use of caged benzyl alcohol does
not produce a measurable H/D kinetic isotope effect.Unexpectedly, the observed fast phase
corresponds to multiple enzyme turnovers, based on the stoichiometry of the substrate to en-
zyme.Possible side reactions and their effects, such as the re-oxidation of bound NADH and
q Supported by a grant (to JPK) from the National Science Foundation (MCB 9514126).
* Corresponding author.Fax: 1-510-643-6232/8369.
E-mail address: klinman@socrates.berkeley.edu (J.P. Klinman).
0045-2068/03/$ - see front matter Ó 2003 Elsevier Science (USA).All rights reserved.
doi:10.1016/S0045-2068(03)00018-X

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
173
the dissipation of photo-excitation energy, may offer an explanation for the observed multiple-
turnovers.The lack of observable deuterium isotope effects offers a cautionary note for the
application of caged substrates to isolate and study chemical steps of enzyme reactions,
particularly when NADH is involved in the reaction pathway.
Ó 2003 Elsevier Science (USA).All rights reserved.
1. Introduction
Extensive structural and functional studies have provided the foundation for our
understanding of enzyme catalysis [1].With the exception of protein conformational
changes, the behavior of enzymes has been described largely within a static frame-
work.However, there has been an increasing recognition of the importance of pro-
tein dynamics in enzyme catalysis, in particular, the contribution of protein
vibrational modes to bond cleavage processes (cf.[2]).
Much of the experimental evidence for the link between protein motion and catal-
ysis has come from studies of hydrogen transfer reactions [2–6].Although hydrogen
transfer reactions can, in principle, take place either by surmounting the reaction
barrier semi-classically, or tunneling through the reaction barrier quantum-mechan-
ically, the latter process has been found to be dominant in many enzyme catalyzed
C–H cleavage reactions (cf.[7] and refs therein).Since hydrogen tunneling is highly
dependent on the relative energies of reactant and product as well as the distance be-
tween donor and acceptor [2], modest changes in protein structure and dynamics can
be expected to impact, to a significant extent, the degree of tunneling.Specifically, if
the temperature is reduced to sub-zero, two opposing trends in hydrogen transfer
may emerge: first, if protein dynamics impact the height and width of the reaction
barrier, a reduction in protein motion that accompanies low temperature may lead
to a reduction or loss of tunneling.On the other hand, very low temperatures will
affect the Boltzmann distribution of molecules with sufficient energy to surmount
the reaction barrier, such that the only option for chemical reaction may become
barrier penetration [8,9].
In a previous study [9], using benzyl alcohol as the substrate to study both com-
petitive and non-competitive KIEs, we observed hydrogen tunneling for a mutant
form (F93W) of horse liver alcohol dehydrogenase (HLADH)1 with little kinetic
complexity down to )35 °C.At this temperature significant tunneling persisted, sug-
gesting either that the tunneling-enhancing protein motions persisted at )35 °C or
that barrier penetration through a more static barrier had begun to dominate the
reaction coordinate.It was of great interest to reduce the experimental temperature
below )35 °C to examine whether a transition between different types of tunneling
1
Abbreviations used: HLADH, horse liver alcohol dehydrogenase; F93W, mutant HLADH that differs
from the wild type enzyme by Phe93 ! Trp; YADH, yeast alcohol dehydrogenase; KIE, kinetic isotope
effect; HPLC, high performance liquid chromatography; kDa, kilodalton; MeOH, methanol; BzOH,
benzyl alcohol.

174
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
Scheme 1.Anticipated reaction pathway for interaction of HLADH with the caged reagent.Following
photo-dissociation of the caged reagent, alcohol (Alc) is oriented to face the active site zinc.Flipping of
bound alcohol is followed by the chemical (hydrogen transfer) step.Loss of bound aldehyde leads to
its rapid trapping by excess semicarbazide, ensuring reaction irreversibility.
behavior could be detected, as was recently reported for a thermophilic ADH [8].Be-
cause F93W HLADH is prone to precipitation below )35 °C, we turned to the study
of the more stable wild type HLADH, which is active down to )50 °C.However, ki-
netic isotope effects with the wild type enzyme indicate significant kinetic complexity
due to slow substrate binding/product release, which increases at lower temperatures
and masks the observation of tunneling [9].In order to study the behavior of
HLADH below )35 °C, a new approach to reduce kinetic complexity was necessary.
Light-based deprotection of masked functional groups of certain biomolecules
(‘‘caged’’ molecules) has been recognized as a simple, non-invasive technique to gen-
erate concentration gradients of substrates in a local environment [10].Irradiation of
the caged compounds with light removes the protecting group, with the subsequent
generation of the substrate activity.Herein, we report a unique strategy to study hy-
drogen tunneling at sub-zero temperature, utilizing photolabile (caged) substrates of
HLADH.As illustrated in Scheme 1, prebinding of caged substrate eliminates the
kinetic complexity arising from slow substrate binding at sub-zero temperatures,
as well as possible kinetic complexity arising from a more random binding mecha-
nism for cofactor and substrate in the presence of cryosolvents [9].An underlying
assumption of Scheme 1 was that the overall photolysis reaction would be less ther-
mally activated than subsequent chemistry [10], ensuring that the latter is kinetically
dominant at reduced temperatures.Additionally, it was expected that reorientation
of the photoreleased, bound substrate (Alc) would be rapid and facile.As we discuss
herein, an optimal caged substrate (15) that binds tightly to HLADH was designed
and synthesized.Detailed binding and kinetic studies indicate that this caged reagent
introduces several levels of kinetic complexity that were not originally anticipated.
This report offers a cautionary to the utilization of photolabile reagents to isolate
chemical steps in enzyme reactions.
2. Materials and methods
Reagent grade benzyl alcohol and benzoic acid were obtained from Fisher.Prior
to use, benzyl alcohol was purified by vacuum distillation.[Ring- 14CðUÞ]benzoic acid
(ICN) had a specific activity of 56–65 Ci/ mol.Liver alcohol dehydrogenase
(HLADH) was obtained from Boehringer–Mannheim as a suspension in a 20 mM

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
175
potassium phosphate (pH 7.0) and 10% ethanol solution. NMR spectra were mea-
sured in CDCl3 on either a Bruker 500 MHz or a Bruker 400 MHz NMR spectro-
photometer.
Benzyl (2-nitrobenzyl) ether 1 (Fig. 1B). To 1.0 g (6.62 mmol) of 2-nitrobenzyl al-
cohol in a 100-mL round-bottomed flask immersed in an ice-water bath was added
20 mL of dry THF followed by 320 mg (6.66 mmol) of powdered NaH (50% oil dis-
persion).After the gas evolusion subsided, 10 mL THF solution of benzyl bromide
(1.10 g, 6.43 mmol) was added. After 20 h, the reaction mixture was quenched by
the addition of 10 mL of water, followed by an extraction with ether (2 Â 20 mL).
The combined ether layer was washed with brine and then dried.Upon addition
of n-pentane (10 mL), the unreacted starting material 2-nitrobenzyl alcohol (50 mg,
5%) precipitated and was collected by filtration.The filtrate was a colorless solution,
which evaporated in air to give 0.54 g (25%) of compound 1 as white needles, m.p.
1
64–66 °C; H NMR d 8.05 (dd, J ¼ 8, 1 Hz, 1H), 7.85 (dd, J ¼ 8, 1 Hz, 1H), 7.63
(ddd, J ¼ 8, 1, 1 Hz, 1H), 7.42 (ddd, J ¼ 8, 1, 1 Hz, 1H), 7.40–7.29 (m, 5H), 4.95
(s, 2H), 4.66 (s, 2H); kmax 290 nm (e ¼ 4875cmÀ1 MÀ1); elemental analysis
C14H13NO3 calcd (%) C 79.53 H 6.20 N 6.63, found (%) C 78.89 H 6.25 N 7.00. Ben-
zyl (2,4-dinitrobenzyl) ether 2 and benzyl (2-nitro-3,5-dimethoxybenzyl) ether 3 were
attempted in a very similar fashion, but no products were obtained.
Fig.1.(A) Structures of ether-linked ( 1–3) and carbonate-linked (4–6) substrates.(B) Synthesis of 1.(C)
Synthesis of 4–6.

176
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
Benzyl (2-nitrobenzyl) carbonate 4 (Fig. 1C).Into 15 mL dichloromethane solu-
tion of triphosgene (656 mg, 2.21 mmol) cooled to )80 °C, a 10 mL dichloromethane
solution of dry pyridine (524 mg, 6.62 mmol) and 2-nitrobenzyl alcohol (1.0 g,
6.62 mmol) were added dropwise, and the reaction mixture was allowed to warm
to room temperature and stirred overnight.The resulting yellow reaction mixture
was then cooled to )80 °C, followed by the addition of a 10 mL dichloromethane so-
lution of dry pyridine (524 mg, 6.62 mmol) and benzyl alcohol (716 mg, 6.62 mmol).
The solution was then concentrated and washed with 10 mL dry THF, followed by
filtration.Purification by column chromatography (silica gel packed with 30% ben-
zene/n-hexane, eluted with 50% benzene/n-hexane) yielded 1.5 g (55%) of compound
1
4 as a white solid, m.p. 43–45 °C; H NMR d 8.12 (d, J ¼ 8 Hz, 1H), 7.64 (d,
J ¼ 1 Hz, 1H), 7.63 (s, 1H), 7.62–7.45 (m, 1H), 7.39–7.33 (m, 5H), 5.60 (s, 2H),
5.20 (s, 2H); elemental analysis C15H13NO5 calcd (%) C 62.71 H 4.56 N 4.87, found
(%) C 62.39 H 4.54 N 4.79. Benzyl (2,4-dinitrobenzyl) carbonate 5 and benzyl (2-ni-
tro-3,5-dimethoxybenzyl) carbonate 6 were synthesized in a similar fashion. 5, m.p.
1
51–53 °C; H NMR d 8.06 (d, J ¼ 9 Hz, 2H), 7.67 (t, J ¼ 8 Hz, 1H), 7.34–7.30 (m,
5H), 5.61 (s, 2H), 5.11 (s, 2H); elemental analysis C15H12N2O7 calcd (%) C 54.27
1
H 3.64 N 8.44, found (%) C 54.44 H 3.65 N 8.60. 6, m.p. 73–74 °C; H NMR d
7.71 (s, 1H), 7.38–7.34 (m, 5H), 6.70 (s, 1H), 5.59 (s, 2H), 5.20 (s, 2H), 3.93 (s,
3H), 3.85 (s, 2H); kmax 342 nm (e ¼ 3784cmÀ1 MÀ1); elemental analysis C17H17NO7
calcd (%) C 58.79 H 4.94 N 4.03, found (%) C 59.01 H 4.85 N 3.95.
3-Methoxy-4-(phenylmethoxy)benzaldehyde 7 (O-benzylvanillin) (Fig. 2A). A
mixture of vanillin (20 g, 0.128 mol), benzyl bromide (22 g, 0.128 mol), potassium
carbonate (8.8 g, 0.064 mol), and acetone (300 mL) were refluxed for 12 h and then
were added to water (300 mL).The precipitated product recrystallized from ethanol
to yield 31 g (100%) of compound 7 as colorless needles, m.p. 128–130 °C; 1H NMR
d 9.81 (S, 1H), 7.43–7.23 (m, 8H), 5.22 (s, 2H), 3.92 (s, 3H); mass (Mþ n=z) 242.3.
5-Methoxy-2-nitro-4-(phenylmethoxy)benzaldehyde 8 (O-benzyl-6-nitrovanillin).
Powdered O-benzylvanillin (31 g, 0.128 mol) was added to nitric acid (150 mL; d
1.46) and stirred at 0 °C.After one hour the solid product was filtered and recrystal-
lized from ethyl acetate to give 35 g (96%) of compound 8 as pale yellow needles,
1
m.p. 212–214 °C; H NMR d 10.39 (s, 1H), 7.65 (s, 1H), 7.44–7.35 (m, 6H), 5.25
(s, 2H), 3.91 (s, 3H); mass (Mþ n=z) 287.3.
4-Hydroxy-5-methoxy-2-nitrobenzaldehyde 9 (6-nitrovanillin).A mixture of pure
O-benzyl-6-nitrovanillin (35 g, 0.123 mol) and acetic acid (240 mL) was kept at
85 °C while 48% hydrobromic acid (75 mL) was added.The reaction mixture was
kept at 85 °C for 1 h.The solid product was filtered and recrystallized from hot eth-
1
anol to yield 14.5 g of compound 9 (60%) as yellow crystals, m.p. 208–210 °C; H
NMR d 10.39 (s, 1H), 7.70 (s, 1H), 7.46 (s, 1H), 4.05 (s, 3H); mass (Mþ n=z) 197.1.
4-(60-Hydroxyhexoxy)-5-methoxy-2-nitrobenzaldehyde 10.Compound 9 (5.0 g,
0.025 mol), KOH (1.425 g, 0.025 g) and 6-bromo-1-hexanol (4.6 g, 0.025 mol) were
added with 50 mL dry ethanol and refluxed overnight.The reaction mixture was fil-
tered, concentrated, extracted with ethyl acetate (2 Â 250 mL), washed with water
(2 Â 200 mL) and brine, and then dried.Liquid chromatography (silica gel packed
and eluted with 10% ethyl acetate/90% n-hexane) led to 2.5 g of compound 10

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
177
Fig.2.(A) Reactions leading to the non-radioactive caged substrate 15.(B) Reactions leading to the [ 14C]-
labeled protiated and deuterated caged substrate 15.
1
(30%) as yellow oil, H NMR d 10.41 (s, 1H), 7.56 (s, 1H), 7.38 (s, 1H), 4.12 (t,
J ¼ 7 Hz, 2H), 3.97 (s, 3H), 3.64 (t, J ¼ 7 Hz, 2H), 1.91–1.86 (m, 2H), 1.61–1.52
(m, 2H), 1.50–1.41 (m, 4H); mass (Mþ n=z) 297.3.
4-[60-(Methanesulfonyl)hexoxy]-5-methoxy-2-nitrobenzaldehyde 11.Compound
10 (2.5 g, 0.008 mol) and triethylamine (3.5 mL, d 0.726, 0.025 mol) were dissolved
in 100 mL dichloromethane under dry nitrogen, and mesityl chloride (1.6 mL,
d 1.48, 0.021 mol) was added by syringe. The reaction mixture was refluxed for
4 h, extracted with water (100 mL Â 2), washed by brine, and then dried.Upon con-
1
centration yielded 3.24 g of compound 11 (100%) as yellow oil, H NMR d 10.42 (s,
1H), 7.57 (s, 1H), 7.39 (s, 1H), 4.24 (t, J ¼ 7 Hz, 2H), 4.13(t, J ¼ 7 Hz, 2H), 3.99 (s,
3H), 2.99 (s, 3H), 1.92–1.88 (m, 2H), 1.80–1.77 (m, 2H), 1.52–1.51 (m, 4H); mass
(Mþ n=z) 375.4.
4-(60-Iodohexoxy)-5-methoxy-2-nitrobenzaldehyde 12.Compound
11 (3.24 g,
8.5 mmol) was dissolved in acetone 120 mL, and sodium iodide (3.7 g, 25 mmol)
was added.The clear yellow solution was refluxed overnight, followed by filtration
and concentration to yield 3.61 g of compound 12 (100%) as an orange solid, m.p.

178
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
68–70 °C; 1H NMR d 10.42 (s, 1H), 7.57 (s, 1H), 7.39 (s, 1H), 4.13 (t, J ¼ 6 Hz, 2H),
3.98 (s, 3H), 3.18 (t, J ¼ 7 Hz, 2H), 1.90–1.83 (m, 4H), 1.54–1.48 (m, 4H); mass (Mþ
n=z) 412.3.
6-(40-Formyl-20-methoxy-50-nitrophenoxy)hexyl triethylammonium iodide 13.Com-
pound 12 (3.6 g, 8.5 mmol) was dissolved in 50 mL acetonitrile, and 30 mL triethyl-
amine was added dropwise.The reaction mixture turned from orange to yellow,
and was refluxed for 4 h, followed by concentration and washing with ether
(3 Â 200 mL) to yield 3.66 g of compound 13 (90%) as yellow oil, 1H NMR d
10.43 (s, 1H), 7.60 (s, 1H), 7.40 (s, 1H), 4.17 (t, J ¼ 6 Hz, 2H), 4.01 (s, 3H), 3.50
(q, J ¼ 7 Hz, 6H), 3.38–3.34 (m, 2H), 2.01–1.94 (m, 2H), 1.81–1.65 (m, 2H),
1.45–1.39 (m, 4H), 1.21 (t, J ¼ 7 Hz, 9H); mass (Mþ n=z) 381.5.
6-[40-(Hydroxymethyl)-20-methoxy-50-nitrophenoxy]hexyl triethylammonium io-
dide 14.Compound 13 (3.6 g, 7 mmol) was dissolved with 50 mL dry methanol under
dry nitrogen, and added to 50 mL stirred ice-cold methanol solution of sodium bo-
rohydride (0.16 g, 4.3 mmol). The reaction was stirred at 0 °C for one hour.Acetone
(100 mL) was added to the resulting solid, followed by filtration and concentration to
yield 3.66 g of compound 14 (100%) as light yellow oil, 1H NMR d 7.57 (s, 1H), 7.39
(s, 1H), 5.05 (s, 2H), 4.25 (t, J ¼ 6 Hz, 2H), 3.95 (s, 3H), 3.50 (q, J ¼ 7 Hz, 6H), 3.39–
3.34 (m, 2H), 1.90–1.70 (m, 2H), 1.80–1.55 (m, 6H), 1.21 (t, J ¼ 7 Hz, 9H); mass (Mþ
n=z) 383.5.
6-{20-Methoxy-50-nitro-40-[(phenylmethoxy)methyl]phenoxy}hexyltriethylammo-
nium iodide 15 (Fig. 2A).Compound 14 (0.5 g, 1 mmol) and benzyl bromide (5 mL, d
1.4, largely in excess) were added via syringe to 10 mL dry THF, followed by the ad-
dition of sodium hydride (0.036 g, 1.5 mmol) and was refluxed overnight. The reac-
tion mixture was then quenched by the addition 100 lL MeOH, followed by
1
filtration and concentration to yield compound 15 (0.6 g, 100%) as yellow oil, H
NMR d 7.65 (s, 1H), 7.45–7.30 (m, 6H), 4.95 (s, 2H), 4.67 (s, 2H), 4.18 (t, J ¼ 6 Hz,
2H), 3.88 (s, 3H), 3.39 (q, J ¼ 7 Hz, 6H), 3.19–3.08 (m, 2H), 1.85–1.80 (m, 2H), 1.75–
1.65 (m, 2H), 1.65–1.50 (m, 2H), 1.50–1.40 (m, 2H), 1.29 (t, J ¼ 7 Hz, 9H); kmax
355 nm (e ¼ 3500cmÀ1 MÀ1); mass (Mþ n=z) 473.6. The overall yield from vanillin
to compound 15 is 15%.
[Ring-14C(U)]benzyl alcohol.Lithium aluminum hydride 30 mg (1 mmol) was
added to a stirred dry ether solution (5 mL) under nitrogen.One mCi of [ring-
14CðUފbenzoic acid (0.015 mmol, 65 mCi/mmol) in 1 mL toluene was added and
stirred for 4 h.The reaction mixture was quenched by adding 1 mL of ice water, fol-
lowed by the addition of 2 mL HCl to dissolve the resulting aluminum hydroxide.
The aqueous layer was extracted with ether (5 mL Â 4), and the combined ether layer
was washed with brine and then dried over MgSO4.Concentration in vacuo yielded
[ring-14C(U)]benzyl alcohol (950 lCi, yield 95%).The crude product can be further
purified by HPLC, eluting at 14 min on a reverse phase Rainin C18 column (25 cm
ꢀ
long, 100 A pore, 5 mm o.d.). An isocratic solvent was used (12% CH3OH, 12%
CH3CN, and 76% water), at a flow rate of 1 ml/min.Since no ring modification took
place and no non-radioactive starting material was introduced during the synthesis,
it was assumed that the specific activity of the product is close to the starting material
(65 Ci/mol).

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
179
[1,1-2H2]-[ring-14C(U)]benzyl alcohol.This was obtained quantitatively accord-
ing to the above procedure, except that a batch of [ring-14CðUÞ]benzoic acid with a
specific activity of 56 Ci/mol was used.Since no ring modification took place and no
non-radioactive starting material was introduced during the synthesis, it was as-
sumed that the specific activity of the product is close to the starting material
(56 Ci/mol).
[Ring-14C(U)]benzyl trichloroacetimidate 16.[Ring- 14CðUÞ]benzyl alcohol (100
lCi, 65 Ci/mol, 1.5 lmol) and 1 mL of dry ether stirred while sodium hydride
(1 mg, 41.6 lmol) was added.After 10 min, dry trichloroacetonitrile (14.mg,
9.25 lmol) was added to the reaction mixture.After 4 h, the reaction mixture was
quenched with 10 lL of MeOH and concentrated, followed by washing with
4 Â 2:5 mL of dry n-pentane.The solution was washed with 2 Â 1 mL of MeCN/
water (65/35), followed by brine and then dried.Upon removal of n-pentane by vac-
uum, the product 16 was obtained quantitatively as a colorless liquid.Compound 16
could be further purified by passing through a 10 cm glass column (packed with sil-
ica gel in 5% EtOAc/n-hexane, eluted with 5% EtOAc/n-hexane) or by HPLC (con-
dition described above); yield 80–90%.Since no ring modification took place and no
non-radioactive starting material was introduced during the synthesis, it was as-
sumed that the specific activity of the product is close to the starting material
(65 Ci/mol).
[Ring-14C(U)]-1,1-2H2-benzyl trichloroacetimidate.This was obtained according
to a similar procedure except that [Ring-14CðUÞ]-1,1-2H2-benzyl alcohol was used;
yield 80–90%.Since no ring modification took place and no non-radioactive starting
material was introduced during the synthesis, it was assumed that the specific activity
of the product is close to the starting material (56 Ci/mol).
[Ring-14C(U)]-caged substrate 15 (Fig. 2B).To exchange the anion from iodide
to ClOÀ4 , a solution of compound 14 (5 mg in 1 mL pure water) was passed through a
C18 sep-pak column (pre-washed with 2 Â 10 mL MeCN, 2 Â 10 mL water, and
3 Â 10 mL 0.1 M NaClO4 methanol solution) and eluted with 1 mL 0.1 M NaClO4
À
methanol solution.Upon concentration, the ClO salt of 14 was obtained quantita-
tively.A dry dichloromethane solution of the ClO salt of 14 (1 mg, 2.07 mmol),
4
À
4
non-radioactive benzyl trichloroacetimidate (0.52 mg, 2.06 mmol) and [ring-
14CðUÞ]benzyl trichloroacetimidate 16 (100 lCi, 1.5 lmol) was added.The final total
volume was 1.5 mL. The reaction was initiated by the addition of triflic acid (0.01 g,
0.05 mmol). The reaction mixture was neutralized by the addition of powdered so-
dium bicarbonate, followed by filtration and concentration to yield product (1 mg,
95%) as a yellow liquid.The product can be further purified on HPLC by elution
on a C-18 column with 0.1 M sodium cacodylate (pH 7.0) buffer in MeCN/Water
(65/35) and is eluted at 12 min.The concentration of 15 is determined by the extinc-
tion coefficient of 15 at kmax 355 nm (e ¼ 3500 cmÀ1MÀ1).Final specific activity of 15,
based on the radiation count (in dpm) in a sample with known concentration, was
54 mCi/mol.
[Ring-14C(U)]-1,1-2H2-caged substrate 15.This was obtained according to a sim-
ilar procedure, except that [ring-14CðUÞ]-1,1-2H2-benzyl trichloroacetimidate was
used; yield 85–95%.Final specific activity was 45 mCi/mol.

180
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
2.1. Computer modeling
The coordinate file of the X-ray crystal structure of the productive ternary com-
plex HLADH-NADþ-p-bromobenzyl alcohol was obtained from the laboratory of
Prof.B.Plapp [11].Computer modeling was performed on a Silicon graphics Indigo
computer with InsightII software.The caged substrates were docked visually by con-
straining the distance between the substrate and the enzyme residues to be greater
than the sums of van der Waals radii of the corresponding atoms.Fittings of the
caged substrates into the binding pocket were achieved by the overlapping of the
benzene ring of the caged substrates with the benzene ring of p-bromobenzyl alcohol
in the ternary complex structure, followed by bond-twisting around the two benzylic
CH2 group in the caged substrates to avoid any steric interaction.Benzyl alcohol
flipping was visualized first by overlapping the benzene ring of benzyl alcohol with
p-bromo benzyl alcohol in the original PDB file, followed by a rotation of the benzyl
alcohol around the center of the benzene ring.
2.2. Photolysis of the caged substrates
A quartz cuvet (1 mm light path) was purged with dry nitrogen and thermo-
stated by a cell holder with a circulation jacket.Into a quartz cuvet was injected
150 lL solution of compound 1, 6, or 15 (ca.03.4 mM) via a syringe.The irradi-
ation of compound 1, 6, or 15 was carried out by an Nd YAG Laser at 355 nm.
The photo-cleavage reaction was monitored by a Hewlett–Packard photo diode
1
array spectrometer and H NMR.A decrease in the concentration of the caged
substrate was measured by the decrease of their UV/Vis absorptions.The absorp-
tion wavelength and extinction coefficient for 1, 6, and 15 are 290 nm
(e ¼ 4875cmÀ1MÀ1), 342 nm (e ¼ 3784cmÀ1MÀ1), and 355 nm (e ¼ 3500cmÀ1
M
À1), respectively.This can be plotted against time and fitted with first order ki-
netics to obtain a photolysis rate for each of the caged substrates.Quantum yield
was measured by chemical actinometry, with standard photolysis of
a
K3FeðC2O4Þ (quantum yield f297 ¼ 1:24, f358 ¼ 1:25) [10].The photolysis rates
3
for both the caged substrates and K3FeðC2O4Þ were measured under identical
3
conditions.The quantum yields of the caged substrates were calculated according
to reference [10]:
vcage
vFe
/
¼ /Fe
Â
;
ð1Þ
cage
where /cage is the quantum yield for the photolysis of the caged substrates, /Fe is the
quantum yield for the photolysis of K3FeðC2O4Þ ; vcage is the photolysis rate for the
3
caged substrates, and vFe is the photolysis rate for K3FeðC2O4Þ [5].It was found that
3
the quantum yield of 15 is 0.24.
Binding study of the caged substrates.A CARY-118b UV/Vis spectrophotometer
was modified to work at sub-zero temperatures, as mentioned in previous work [9].
The binding kinetics was measured as described previously [9].

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
181
2.3. Design of low-temperature photolysis chamber
A chamber was constructed of polyacrylate plastic, which is not only a good ther-
mo-insulator but also lightweight and transparent.A round quartz cuvet (1 mm light
path) was installed in the chamber, as well as a thermostated cuvet holder matching
the size and shape of the cuvet.The thermostated jacket of the cuvet holder was at-
tached to the inlet and outlet of a Neslab ULT-80 cryogenic bath by two C-Flex tub-
ings. Two round quartz windows (2 in. o.d.) were installed on either side of the
chamber to define the light path for the YAG laser beam (5 ns pulse, average power
150 mW, 355 nm).The temperature in the cuvet was constantly monitored by a Fish-
er thermocouple with a type K probe (0.4 mm o.d.). To prevent moisture condensa-
tion on the cuvet, a stream of cold, dry nitrogen was blown into the bottom of the
cuvet holder via a C-Flex tubing, creating a dry nitrogen jacket around the cuvet.
The cold, dry nitrogen was cooled by passing through a homemade copper coil
(15 cm diameter, 100 turns).To prevent moisture condensation on the quartz win-
dows of the chamber, two separate streams of room temperature, dry nitrogen were
blown directly on the two quartz windows.The low-temperature chamber can be op-
erated within a very stable temperature range down to )70 °C.
2.4. Photolysis of 15 pre-bound to HLADH
One hundred and fifty microliters of the sample, containing 20 lM HLADH,
1 mM of 14C-labeled protiated or deuterated 15 (concentration determined by the ex-
tinction coefficient at 355 nm) and 10 mM of NAD, were placed into the cuvet.The
buffer contained 100 mM sodium cacodylate and 200 mM semicarbazide (pH* 7.0
[9]) in 30–50% MeOH.The sample was then irradiated with the YAG Laser in full
power for 0–15 s.The reaction for the photolysis of 15 was initiated by laser irradi-
ation.A control of 20 lM HLADH and 10 mM NAD were added to the cuvet at a
stable temperature, followed by the initiation of the reaction by the addition of 1 mM
14C-labeled benzyl alcohol (freshly distilled one day prior to the experiment) with a
pipetteman (both benzyl alcohol and the pipette tip were pre-equilibrated at the same
temperature as that in the cuvet) and simultaneous laser irradiation for 0–300 s.Each
sample (150 lL) contained 15,000–18,000 dpm of 14C-labeled 15 or benzyl alcohol.
Reactions of either 15 or benzyl alcohol were rapidly quenched with 20 lL of
2 mM HgCl2, using a pre-chilled pipette tip to mix the solution.The time points
at 5, 10, and 15 s were conducted with varying laser irradiation time at 5, 10, and
15 s, and the reaction was quenched with mercury salt immediately after laser irradi-
ation, while the time point at 30, 60, and 300 s was irradiated for 15 s and quenched
thereafter.Quenched samples were retrieved from the cuvet and frozen in dry ice.
Upon thawing for analysis, samples were centrifuged, injected to an HPLC column
ꢀ
(25 cm long, 100 A pore), and eluted at 1 mL/ min by either a Beckman 110B or a Shi-
matsu 10-SV HPLC system.The HPLC program was set to elute the sample isocrat-
ically (12:12:76, v/v, MeCN:MeOH:water) for 30 min, followed by a 10 min gradient
to 100% MeOH and 0.1 M NaClO4.The absorption of at least one wavelength (210,
280, or 350 nm) was monitored during the HPLC run.The eluant was collected with

182
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
a LKB-Bromma 2112 Redirac fractional collector in fractions of 1.5 mL in 20 mL
plastic vials; 12 mL of scintillation liquid (Ecolite from ACN) was then added to each
vial.The vials were counted on a LKB-Wallac 1209 Rackbeta liquid scintillation
counter for 300 s with 15 s of internal standard counting.Benzyl alcohol, benzalde-
hyde semicarbazone, and the caged substrate 15 eluted at 14–15, 28–31, and 40–
45 min, respectively.The percentage of conversion was calculated for each time point
by the ratio of radioactivity between the product benzaldehyde semicarbazone and
total counts in the sample.The time traces were plotted for percentage of conversion
versus time (Fig.3).
Control experiments were conducted and analyzed similar to the above reaction
condition.A control zero-time point was taken prior to the laser irradiation of 15
and analyzed as above; benzyl alcohol or its enzymatic products were not found
in these zero-time point samples.Also, during photolysis, temperatures were moni-
tored by a thermocouple, indicating no increase in temperatures.Further, to monitor
laser damage to the enzyme HLADH, a 150 lL sample of 20 lM HLADH and
10 mM NAD, with or without 1 mM non-radioactive 15, was irradiated with laser
for 60 s.The enzyme activity is then analyzed as previously described [12].HLADH
was found to retain full activity after laser irradiation, indicating no photo-damage
to the enzyme plus NAD.Additionally, 1 mM of benzyl alcohol was laser-irradiated
for 60 s in the presence of 10 mM NAD.This is followed by the addition of enzyme
HLADH to monitor the initial rate of NADH formation [12].The observed initial
rate is similar to the control without laser irradiation, indicating no photo-damage
to benzyl alcohol and NAD upon laser irradiation.Photolysis of 1 mM NADH in
the presence of 1 mM 15 was analyzed by the decrease of the absorbance at
340 nm (kmax of NADH).Finally, photolysis of 15 in the presence of buffer and
10 mM NAD produced only benzyl alcohol as analyzed by HPLC.
3. Results and discussion
3.1. Design principles
A number of important design principles of the caged substrate are related to its
binding affinity and orientation within HLADH.The X-ray crystal structure of
HLADH reveals that the residues in the binding pocket of HLADH are highly hy-
drophobic [11], such that incorporation of a hydrophobic component was expected
to lead to strong binding of caged substrates to the binding pocket.However, a sim-
ple hydrophobic caged substrate might bind to the enzyme in non-productive modes,
with the alcohol precursor pointing toward solvent rather than to the cofactor.Con-
straints were needed to ensure that the caged substrate would reside in a productive
binding mode that upon photolysis releases alcohol at the bottom of the cleft leading
to the active site.As will be obvious from the structures of caged reagents described
below, photo-released alcohol is required to flip by ca.180 ° in order to orient the al-
cohol functional group in a position suitable for catalysis (cf.Schemes 1 and 2).
Computer modeling indicated that there was sufficient space for this flipping to occur

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
183
Scheme 2.Illustration of the expected orientation of the final caged reagent ( 15) within the active site of
HLADH.The quaternary amine both increases the solubility of the caged reagent and places the alcohol
precursor near the active site of enzyme.
without significant steric interactions.One final constraint was that the caged sub-
strate have a significant UV/Vis absorption at a wavelength greater than 300 nm,
to place its kmax away from the range of protein absorbance.
3.2. Synthesis and properties of caged substrates 1–6
The simplest compound (1 in Fig.1A), contains the 2-nitrobenzyl moiety and the
substrate moiety.Initially, we attempted to synthesize 1 by the deprotonation of ben-
zyl alcohol, followed by its nucleophilic attack on 2-nitrobenzyl bromide.This reac-
tion did not produce 1, possibly due to the deprotonation of 2-nitrobenzyl bromide
to yield a resonance stabilized anion intermediate.It was found that reversing the
polarity of the components (deprotonation of 2-nitrobenzyl alcohol followed by cou-
pling with benzyl bromide) gave the desired cage compound 1 (Fig.1B).
The UV/Vis absorption spectrum of 1 shows the p ! pà absorption [7] at 295 nm
(e ¼ 4875cmÀ1MÀ1).The n ! pà transition is hidden within this absorption and is
the main excitation mode that results in the photo-cleavage reaction [13].As men-
tioned above, a compound with absorption above 300 nm was preferred to allow
for optimal photo-cleavage with minimal photo-damage of the protein.Different
functional groups were therefore added to 1 to generate 2 and 3; additionally, a series
of compounds using carbonate as the leaving group (4–6) was synthesized.The latter
were obtained by the use of a stoichiometric amount of triphosgene [14] instead of
phosgene gas (Fig.1C).Triphosgene has not been previously applied to the synthesis
of a mixed carbonate.
Among these caged substrates, 6 showed a significant n ! pà transition at 342 nm.
The photolysis of 6 by a YAG laser (355 nm) was completed in 10 s.However, the
highest concentration of 6 that could be dissolved in 30% methanol/water at 3 °C
was found to be 25 lM.This can be compared to a Ki ꢀ 75lM, determined for 6
in 30% MeOH at 0 °C.It was clear that a re-designed compound was needed with
increased solubility in relation to its dissociation constant.
3.3. A long-tailed caged substrate
To increase solubility while retaining the desirable features of 6, a caged sub-
strate was designed to contain a long hydrophobic tail with an ionic end.The

184
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
length of the hydrophobic tail was selected based on computer modeling; a six
carbon chain provides just enough length for the ionic group to protrude out of
the substrate binding pocket and into the bulk solvent (Scheme 2).The ionic
tetra-alkyl ammonium group provides better solubility, while the hydrophobic tail
increases the binding of the caged substrate within the hydrophobic substrate-bind-
ing pocket.Moreover, charge repulsion between the positively charged end of the
caged reagent and the positively charged zinc ion in the enzyme active site, as
well as steric hindrance from the bulky tetra-alkyl ammonium group, was expected
to greatly favor a productive binding mode.We believe this is the first time
that a caged substrate has been selectively modified for a specific enzyme active
site.
The synthesis of 15 (Figs.2A and B) began with commercially available vanillin,
which was protected as a benzyl ether (7 in Section 2) via a Williamson synthesis,
followed by a nitration at the 6-position (8 in Section 2) and a deprotection to re-
move the benzyl group that produces 6-nitrovanillin, 9.The attachment of a hy-
drophobic tail to 9 by a second Williamson synthesis was, however, problematic.
Compound 10 was eventually obtained by the deprotonation of compound 9 with
dry KOH, followed by coupling with a stoichiometric amount of dry 6-bromo-1-
hexanol.The alcohol functional group in compound 10 was activated with mesyl
chloride to yield the mesylate (11 in Section 2).This reacted easily with sodium io-
dide to form the iodide (12 in Section 2), followed by reaction with triethylamine
to generate 13.Reduction of the aldehyde to the alcohol ( 14), deprotonation with
NaH and reaction with an excess amount of benzyl bromide in THF produced the
caged substrate 15.Compound 15 is stable in aqueous solution and is not easily
photolyzable by daylight.For the synthesis of radiolabeled 15, the final coupling
step for the ether formation became a problem.Due to the change of reaction
scale, a coupling between ½14CŠ-labeled benzyl bromide and compound 14 failed
to produce radiolabeled 15.Many different reactions were attempted unsuccess-
fully, including reaction of ½14CŠ-labeled benzyl bromide with a silver salt prior
to coupling with 14 [15], as well as a Mitsunobu reaction that coupled 14 with
½14CŠ-labeled benzyl alcohol by reactions with triphenyl phosphine and diethylace-
tadiimide (DEAD) [16].Eventually, the ½14CŠ-labeled 15 was synthesized by an
acid-catalyzed coupling (trifluoroacetic acid, TfOH) between a stoichiometric
amount of the ClO4 salt of 14 and ½14CŠ-labeled benzyl trichloroacetimidate 16
(Fig.2B) [17].
The photolysis rate of 15 is solvent independent in water, methanol, acetonitrile,
or mixtures of these solvents. 15 were completely photolyzed to benzyl alcohol in 60 s
of laser irradiation at room temperature.The tetra-alkyl ammonium tail end of 15
greatly increases its solubility, which exceeds 10 mM.The ability of 15 to bind to
HLADH was assayed by a steady-state inhibition study, showing that 15 is a com-
petitive inhibitor of the substrate benzyl alcohol in 30% and 50% MeOH, with a Ki of
75 lM in 30% MeOH at 3 °C (Table 1).The increasing concentration of MeOH in-
creases both the KM of benzyl alcohol and the Ki of 15.We note that value of Ki,
relative to KM, goes down somewhat as the percentage of the cryosolvent increases
(Table 1).

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
185
Table 1
Interactions of HLADH with benzyl alcohol and 15 in the presence of increasing MeOHa
Solvent
kcatðsÀ1
0.65
Þ
KMðBzOHÞ ðlMÞ
KiðcageÞ ðlMÞ
75
Ki=KM
30% MeOH
50% MeOH
60% MeOH
70% MeOH
110
1500
1600
2100
0.68
0.44
0.55
0.54
0.024
0.027
0.021
630
880
1100
a All experiments were carried out in a 100 mM cacodylate, 200 mM semicarbazide, and 10 mM NAD
buffer (pH* 7.0) at 3 °C.
3.4. Rate enhancement for product formation from caged substrate 15 at 0 °C
An excess of ½14CŠ-labeled 15 or ½14CŠ-labeled benzyl alcohol (as the control) was
incubated with HLADH and NADþ; the mixture was then photolyzed with increas-
ing irradiation time periods in 50% MeOH at 0 °C (Fig.3A).Over the time course,
the percent of photolysis proceeds to completion (Fig.3A, —Â).The time points at
5, 10, and 15 s were conducted with varying laser irradiation time (for 5, 10, and 15 s)
and the reaction was quenched with mercury salt immediately after laser irradiation,
while the time points at 30, 60, and 300 s were irradiated for 15 s and quenched at the
appropriate time.The radiolabel ensured a sensitive measurement of the concentra-
tion of 15, benzyl alcohol and the benzaldehyde product.A two-phase kinetic trace
—
was observed with 15 pre-bound to HLADH (Fig.3A,
slow phase, the photolyzed 15 reacts with a similar rate as the oxidation of free ben-
).It appears that in the
d
d
—
zyl alcohol (Fig.3A, j j), suggesting that after 30 s, photolyzed 15 (i.e., benzyl
alcohol) reacts from the solution and behaves like free benzyl alcohol.However,
in the fast phase before 30 s, photolyzed 15 creates a burst of product formation, re-
sulting in a much faster rate than that of free benzyl alcohol oxidation.From the ob-
served rate acceleration, it appeared that the prebinding of 15 eliminated the kinetic
complexity of binding at 0 °C.We proceeded to conduct the same experiment at sub-
zero temperatures.
3.5. Lackof H/D KIE at )30 °C
At )30 °C, the photolysis proceeded to 80% completion (Fig.3B, —Â).A two-
phase kinetic trace was again observed upon photolysis of 15 (1 mM) in the presence
of HLADH (20 lM) and NAD (10 mM) in 50% MeOH.Either 1 mM of ½14CŠ-labeled
—
protiated (Fig.3B,
) or deuterated 15 (Fig.3B, M—M) was used for the mea-
d
d
surement of non-competitive kinetic isotope effects.As shown, the percent conver-
sion is much higher for the reaction of pre-bound 15 than the oxidation of benzyl
—
alcohol (Fig.3B, j j).Similar to the finding at 0 °C, the observation of rate ac-
celeration by 15 seemed to indicate that 15 had bound to HLADH prior to the pho-
tolysis and that the kinetic complexity from substrate binding was eliminated by the
use of the cage strategy.The rate of photolysis of 15 approximates the rate of oxi-
dation of the pre-bound substrate, resulting in 50% photolysis of 15 in 5–10 s.Most
significantly, the time traces of protiated and deuterated 15 are very similar, with the

186
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
Fig.3.(A) Photolysis of [ 14C]-15 or [14C]benzyl alcohol with HLADH at 0 °C in 50% MeOH.( d) % of
BzOH oxidation from 15; (j) % of BzOH oxidation from free BzOH; (Â) % of photolysis of 15.(B) Pho-
tolysis of protiated or deuterated [14C]-15 at )30 °C in 50% MeOH.( d) % of BzOH oxidation from pro-
tiated 15; (M) % of BzOH oxidation from the deuterated 15; (j) % of BzOH oxidation from free BzOH;
(Â) % of 15 photolysis in a separate experiment from above.The similarity in the time trace between pro-
tiated and deuterated [14C]-15 indicates a lack of non-competitive H/D KIE.For both (A) and (B), all ex-
periments were conducted in 150 lL of 200 mM semicarbazide and 100 mM sodium cacodylate (pH 7.0)
50% MeOH buffer with 20 lM HLADH, 10 mM NAD and 1 mM 15 or BzOH.The samples were irradi-
ated for 15 s, except for the samples that were quenched before 15 s, which were irradiated for the same
amount of time as the reaction.
deuterated substrate showing a slightly higher percent conversion than that of the
protiated substrate in the fast phase.In comparison, our previous study of non-com-
petitive H/D KIE at )30 °C, using protiated and deuterated benzyl alcohol to mea-
D
sure steady-state parameters, gave a ðkcat=KMÞ of 2.7 for the wild type enzyme and
D
ðkcat=KMÞ of 6.0 for the F93W mutant [4]. If the fast phase is, as anticipated, a sin-
gle-turnover event that isolates the hydrogen-transfer step as the major rate limiting
step, the non-competitive H/D KIE in the fast phase should be higher than the value
of ca. one measured in steady state, i.e., at each time point during the fast phase, the

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
187
protiated 15 should show a percent conversion at least 2.7 times higher than that of
the deuterated substrate.The similarity of the time traces between protiated and deu-
terated 15 indicates that steps other than the hydrogen transfer step have become
rate limiting in the reaction utilizing the caged substrate 15.Since the photolysis re-
action (Fig.3B, —Â) appears to react on a comparable timescale as the enzymatic
—
conversion (Fig.3B,
), it was possible that the turnover rate for bound 15 was
d
d
dictated by the photolysis reaction.This would have explained not only the lack of
an H/D KIE but also the absence of a significant temperature dependence of the time
traces (compare the percentage of conversion in Figs.3A and B), since a photo-ac-
tivated process is less temperature-dependent than a thermally-activated one [10].
3.6. Analysis of events during the fast phase
Further complication arises, however, from the results of Fig.3B when we con-
sider the stoichiometry of 15 versus HLADH.It can be estimated that under the con-
dition of 20 lM of HLADH and 1 mM of 15 (with a Ki of 1 mM in 50% MeOH), ca.
10 lM of 15 will prebind to HLADH.Upon photolysis, a maximum concentration
of 10 lM of pre-bound benzyl alcohol is expected to be converted to benzaldehyde
in a single enzyme turnover; therefore, the maximum percentage of conversion upon
photolysis of 15 for a single enzyme turnover should be 10 lM/1 mM ¼ 1%.Even if
the binding constant (Ki) were much smaller than 1 mM, such that a near-stoichiom-
etric binding between HLADH and 15 had occurred, the maximum conversion
would only be 20 lM/1 mM ¼ 2%.However, the fast phase shows a burst of >10%
conversion, which indicates multiple enzyme turnovers in the fast phase.
What can be the explanation for the observed multiple-turnovers in the first
phase, and is this related to an H/D KIE close to unity? First, extensive control ex-
periments eliminate a number of possible errors in the observations: (i) the concen-
tration of pure 15 (1 mM) was determined by its extinction coefficient immediately
before the experiments and should be accurate; (ii) a zero-time point was taken to
ensure that no 15 had photolyzed prior to laser irradiation; (iii) the amount of radio-
activity was kept constant (15,000–18,000 dpm) among samples; (iv) the total
amount of radioactivity post photolysis from HPLC fractions is consistent with
the amount prior to laser irradiation; and (v) the temperature of the photolysis cell,
monitored with a thermo-couple, did not increase during or after photolysis.We can
rule out any rate limiting step that occurs from free 15 as this would be expected to
eliminate the transient burst all together.The explanation for multiple turnovers in
the burst phase must lie, therefore, in the two major chemical events that took place,
namely the photolysis of 15 and the enzymatic oxidation of benzyl alcohol to benz-
aldehyde.
The photolysis mechanism of 2-nitrobenzyl ethers such as 15 is well-studied [10].
It has been proposed that nitroso- and benzyl-radicals are generated as one of the
photolysis intermediates [13].A plausible mechanism for the breakdown of 15 is
shown in Scheme 3 [13].Free radicals are able to inactivate yeast alcohol dehydro-
genase (YADH) through reaction with active site cysteines and histidines, resulting
in the loss of the active site zinc [18].Additionally, a radical can interact with surface

188
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
Scheme 3.Postulated mechanism for photolysis of caged reagents such as 15 [7], showing radical interme-
diates.Multiple turnovers could occur in the burst phase via oxidation of enzyme-bound NADH by the
bound photoproduct, nitrosoaldehyde (A) or by photoinduced radicals formed from a second molecule of
15 (B).
cysteines in HLADH, enhancing protein denaturation [19].However, the inactiva-
tion of YADH by the radical can be greatly reduced in the presence of NADH,
which absorbs the radical and acts as a ‘‘radical protectant’’ for the enzyme [20].
In fact, it was found that NADH reacts readily with the nitroso radical and is re-ox-
idized to NADþ [20,21].Since radical intermediates will be generated during the pho-
tolysis of 15, some of the above side reactions between the radical intermediate and
the enzyme/NADH may have taken place, namely, the inactivation of the enzyme
(regardless of the mechanism) or the re-oxidation of NADH.Through control exper-
iments, we found that laser irradiation on samples containing HLADH (20 lM) and
NADþ (10 mM), with or without 15 (1 mM), did not affect the specific activity of the
enzyme.This indicates that the enzyme was not damaged in the presence of radical
intermediate due to the photolysis of 15.Further laser irradiation on a sample con-
taining 14C-labeled benzyl alcohol (1 mM) and NAD (10 mM) did not produce de-
tectable amount of product benzaldehyde while prolonged laser irradiation of a
sample containing NADH (1 mM) and 15 (1 mM) resulted in a reduction of absorp-
tion at the kmax of NADH (340 nm).These control experiments indicate that during
the photolysis of 15, there is no detectable radical damage on HLADH, benzyl alco-
hol or NADþ, but that the amount of NADH is reduced.This suggests a re-oxida-
tion of NADH to NADþ by a radical intermediate or aldehyde produced by
photolysis (Scheme 3).
Mechanistic studies on HALDH-catalyzed benzyl alcohol oxidation have indi-
cated that the two major rate limiting steps in the forward direction are the hydro-
gen-transfer reaction ð23 sÀ1Þ and NADH dissociation from the enzyme ð5 sÀ1Þ [22].
Scheme 3 shows a mechanism whereby enzyme-bound NADH formed from benzyl
alcohol oxidation rebinds a second molecule of cage to form an intermediate analo-
gous to 2, but containing NADH instead of NADþ.Oxidation of NADH by the

S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
189
photoproduced radical would lead to enzyme-bound NADþ, which could enter into
a second productive reaction of cage (Scheme 3B).As an alternative, the nitrosoal-
dehyde photoproduct may stay bound to enzyme and facilitate the re-oxidation of
NADH into NADþ (Scheme 3A).Essentially, the enzyme serves to provide a high
local concentration of reactive radical or aldehyde that is capable of oxidizing en-
zyme-bound NADH back to NADþ, thus bypassing the slow NADH-release step.
The complexity of the processes in Scheme 3 could explain both the lack of an H/
D KIE and the observed multiple turnovers in the fast phase (Fig.3).
In closing, it is interesting to consider an alternative, though less likely explana-
tion, for the multiple turnovers of 15 in the burst phase.This could arise from
changes in the dynamic behavior of the protein that accompany the photochemistry,
giving faster turnover with either free 15 or free benzyl alcohol.This idea, discussed
by others in the context of ‘‘protein quakes’’ [23], is very interesting, implying that
photo-excitation of a bound chromophore can change the dynamic properties of
an enzyme-catalyzed reaction.The transfer of photo-excitation energy to the protein
cannot be 100% efficient, and heat will be generated in the energy-transfer process.
We monitored the reaction throughout the photolysis of 15 but did not detect any
raise of the temperature.Therefore, this explanation is likely only if an unknown en-
ergy-transfer process can channel the photo-excitation energy into an enzyme-cata-
lyzed reaction very efficiently or if heat is generated in a very local manner.We
consider the participation of photolysis products in NADH oxidation to be the more
likely cause of the observed multiple-turnovers, while noting the possibility that dis-
sipation of photo-excitation energy may increase protein flexibility and, to some ex-
tent, enzyme turnover.
Acknowledgments
We thank Dr.H.Yang for his help in building the photolysis apparatus, as well as
Prof.J.McCusker, Dr.N.Damrauer, and Dr.J.P.Kirby for their generous supply
of the YAG laser.
References
[1] A.Fersht, Structure and Mechanism in Protein Science: A Guide to Enzyme Catalysis and Protein
Folding, W.H. Freeman, New York, 1999.
[2] A.Kohen, JP. .Klinman, Chem.Biol.6 (1999) 191.
[3] M.J. Knapp, J.P. Klinman, Eur. J. Biochem. 269 (2002) 113.
[4] M.J. Sutcliffe, N.S. Scrutton, Eur. J. Biochem. 269 (2002) 3096.
[5] E.C. Slater, Eur. J. Biochem. 166 (1987) 89.
[6] P.K. Agarwal, S.R. Billeter, P.T. Rajagopalan, S.J. Benkovic, S. Hammes-Schiffer, Proc. Natl. Acad.
Sci.USA 99 (2002) 2794.
[7] M.J. Knapp, K. Rickert, J.P. Klinman, J. Am. Chem. Soc. 124 (2002) 3865.
[8] A.Kohen, R.Cannio, S.Bartolucci, JP. .Klinman, Nature 399 (1999) 496.
[9] S.Tsai, JP. .Klinman, Biochemistry 40 (2001) 2303.
[10] J.A. McCray, D.R. Trentham, Annu. Rev. Biophys. 18 (1989) 239.

190
S.-C. Tsai, J.P. Klinman / Bioorganic Chemistry 31 (2003) 172–190
[11] S.Ramaswamy, H.Eklnd, BV. .Plapp, Biochemistry 33 (1994) 5230.
[12] H. Eklund, B.V. Plapp, J.-P. Samama, C.-I. Branden, J. Biol. Chem. 257 (1982) 14349.
[13] H.Schupp, WK. .Wong, W.Schnabel, J.Photochem.36 (1987) 85.
[14] H.Eckert, B.Forster, Angew.Chem.Int.Ed.Engl.26 (1987) 894.
[15] R.D. Walkup, R.T. Cunnigham, Tetrahedron Lett. 28 (1987) 4019.
[16] D.L. Hughes, Org. React. 42 (1992) 335.
[17] U.Widmer, Synthesis 29 (1987) 568.
[18] S.Abelidis, JS. .Moore, A.Chakravarty, Int.J.Biol.Relat.Study Phys.Chem.Med.52 (1987) 413.
[19] P.Waring, A.Sjaarda, QH. .Lin, Biochem.Pharmacol.49 (1995) 1195.
[20] L.A. Videla, M. Salim-Hanna, E.A. Lissi, Biochem. Pharmacol. 44 (1992) 235.
[21] H.Fujii, J.Koscielniak, K.Kakinuma, LJ..Berliner, Free Radic.Res.21 (1994) 235.
[22] G.L. Shearer, K. Kim, K.M. Lee, C.K. Wang, B.V. Plapp, Bicohemistry 32 (1993) 11186.
[23] A. Ansari, J. Berendzen, S.F. Bowne, H. Frauenfelder, I.E.T. Iben, T.B. Sauke, E. Shyamsunder,
R.D. Young, Proc. Natl. Acad. Sci. USA 82 (1985) 5000.