Transition state structure for the hydrolysis of NAD+ catalyzed by diphtheria toxin

Article abstract of DOI:10.1021/ja971317a

Diphtheria toxin (DTA) uses NAD⁺ as an ADP-ribose donor to catalyze the ADP-ribosylation of eukaryotic elongation factor 2. This inhibits protein biosynthesis and ultimately leads to cell death. In the absence of its physiological acceptor, DTA catalyzes the slow hydrolysis of NAD⁺ to ADP- ribose and nicotinamide, a reaction that can be exploited to measure kinetic isotope effects (KIEs) of isotopically labeled NAD⁺s. Competitive KIEs were measured by the radiolabel method for NAD⁺ molecules labeled at the following positions: 1-¹⁵N = 1.030 ± 0.004, 1'-¹⁴C = 1.034 ± 0.004, (1- ¹⁵N, 1'-¹⁴C) = 1.062 ± 0.010, 1'-³H = 1.200 ± 0.005, 2'-³H = 1.142 ± 0.005, 4'-³H = 0.990 ± 0.002, 5'-³H = 1.032 ± 0.004, 4'-¹⁸O = 0.986 ± 0.003. The ring oxygen, 4'-¹⁸O, KIE was also measured by whole molecule mass spectrometry (0.991 ± 0.003) and found to be within experimental error of that measured by the radiolabel technique, giving an overall average of 0.988 ± 0.003. The transition state structure of NAD⁺ hydrolysis was determined using a structure interpolation method to generate trial transition state structures and bond-energy/bond-order vibrational analysis to predict the KIEs of the trial structures. The predicted KIEs matched the experimental ones for a concerted, highly oxocarbenium ion-like transition state. The residual bond order to the leaving group was 0.02 (bond length = 2.65 A?), while the bond order to the approaching nucleophile was 0.03 (2.46 A?). This is an A(N)D(N) mechanism, with both leaving group and nucleophilic participation in the reaction coordinate. Fitting the transition state structure into the active site cleft of the X-ray crystallographic structure of DTA highlighted the mechanisms of enzymatic stabilization of the transition state. Desolvation of the nicotinamide ring, stabilization of the oxocarbenium ion by apposition of the side chain carboxylate of Glu148 with the anomeric carbon of the ribosyl moiety, and the placement of the substrate phosphate near the positively charged side chain of His21 are all consistent with the transition state features from KIE analysis.

Full text of DOI:10.1021/ja971317a

J. Am. Chem. Soc. 1997, 119, 12079-12088
12079
Transition State Structure for the Hydrolysis of NAD+
Catalyzed by Diphtheria Toxin†
Paul J. Berti,‡ Steven R. Blanke,§ and Vern L. Schramm*,‡
Contribution from the Department of Biochemistry, Albert Einstein College of Medicine,
1300 Morris Park AVenue, Bronx, New York 10461, and Department of Microbiology and
Molecular Genetics, HarVard Medical School, 200 Longwood AVenue,
Boston, Massachusetts 02115
ReceiVed April 25, 1997X
Abstract: Diphtheria toxin (DTA) uses NAD+ as an ADP-ribose donor to catalyze the ADP-ribosylation of eukaryotic
elongation factor 2. This inhibits protein biosynthesis and ultimately leads to cell death. In the absence of its
physiological acceptor, DTA catalyzes the slow hydrolysis of NAD+ to ADP-ribose and nicotinamide, a reaction
that can be exploited to measure kinetic isotope effects (KIEs) of isotopically labeled NAD+s. Competitive KIEs
were measured by the radiolabel method for NAD+ molecules labeled at the following positions: 1-15N ) 1.030 (
0.004, 1′-14C ) 1.034 ( 0.004, (1-15N,1′-14C) ) 1.062 ( 0.010, 1′-3H ) 1.200 ( 0.005, 2′-3H ) 1.142 ( 0.005,
4′-3H ) 0.990 ( 0.002, 5′-3H ) 1.032 ( 0.004, 4′-18O ) 0.986 ( 0.003. The ring oxygen, 4′-18O, KIE was also
measured by whole molecule mass spectrometry (0.991 ( 0.003) and found to be within experimental error of that
measured by the radiolabel technique, giving an overall average of 0.988 ( 0.003. The transition state structure of
NAD+ hydrolysis was determined using a structure interpolation method to generate trial transition state structures
and bond-energy/bond-order vibrational analysis to predict the KIEs of the trial structures. The predicted KIEs
matched the experimental ones for a concerted, highly oxocarbenium ion-like transition state. The residual bond
order to the leaving group was 0.02 (bond length ) 2.65 Å), while the bond order to the approaching nucleophile
was 0.03 (2.46 Å). This is an ANDN mechanism, with both leaving group and nucleophilic participation in the
reaction coordinate. Fitting the transition state structure into the active site cleft of the X-ray crystallographic structure
of DTA highlighted the mechanisms of enzymatic stabilization of the transition state. Desolvation of the nicotinamide
ring, stabilization of the oxocarbenium ion by apposition of the side chain carboxylate of Glu148 with the anomeric
carbon of the ribosyl moiety, and the placement of the substrate phosphate near the positively charged side chain of
His21 are all consistent with the transition state features from KIE analysis.
Introduction
demonstrate that the industrialized world is not exempt from
such diseases. Each of these bacteria exert their toxicity through
an exotoxin that uses NAD+ as an ADP-ribose donor to ADP-
ribosylate a host protein. Diphtheria toxin ADP-ribosylates
eukaryotic elongation factor 2 (EF-2)15 in vivo at diphthamide,
a posttranslationally modified histidine residue, thereby prevent-
ing protein biosynthesis and killing the cell.16 Cholera and
pertussis toxins ADP-ribosylate different G-proteins at arginine
and cysteine residues, respectively. In addition to the potential
Bacteria that use ADP-ribosylating enzymes to exert their
toxicity continue to be a major cause of morbidity and mortality
around the world. There have been recent, large-scale epidemics
of diphtheria in southern Africa,1 eastern Europe, and the former
Soviet Union2 (>48 000 cases) and of cholera in South America3
(950 000 cases), Africa4,5 (12 000 deaths), and India6 (>150 000
cases). Recent outbreaks of whooping cough in the U.S.,7-9
Canada,10 Australia,11 the U.K.,12 Switzerland,13 and Spain14
(8) Christie, C. D.; Marx, M. L.; Marchant, C. D.; Reising, S. F. New
Engl. J. Med. 1994, 331, 16-21.
† This work was supported by NIH Research Grant AI34342 and a
postdoctoral fellowship from the Natural Sciences and Engineering Research
Council (Canada) to P.J.B.
(9) Rosenthal, S.; Strebel, P.; Cassiday, P.; Sanden, G.; Brusuelas, K.;
Wharton, M. J. Infect. Dis. 1995, 171, 1650-1652.
‡ Albert Einstein College of Medicine.
(10) De Serres, G.; Boulianne, N.; Douville Fradet, M.; Duval, B. Canada
Communicable Dis. Rep. 1995, 21, 45-48.
§ Harvard Medical School. Present address: Department of Biochemical
and Biophysical Sciences, HSC - 430, University of Houston, Houston,
TX 77204.
(11) MacIntyre, R.; Hogg, G. Aust. J. Public Health 1994, 18, 21-24.
(12) Syedabubakar, S. N.; Matthews, R. C.; Preston, N. W.; Owen, D.;
Hillier, V. Epidemiol. Infect. 1995, 115, 101-113.
* Corresponding author: phone, (718) 430-2813; fax, (718) 430-8565;
e-mail, vern@aecom.yu.edu.
(13) Matter, H. C.; Schmidt-Schlapfer, G.; Zimmermann, H. Schweiz.
Med. Wochenschr. 1996, 126, 1423-1432.
X Abstract published in AdVance ACS Abstracts, November 15, 1997.
(1) van Geldermalsen, A. A.; Wenning, U. Ann. Trop. Paediatr. 1993,
13, 13-19.
(14) Montiano Jorge, J. I.; Lazcoz Huarte, A.; Amiama Perez de
Villarreal, C.; Mendia Gutierrez, M.; Benito Fernandez, J.; Santiago
Burruchaga, M.; Pocheville Guruceta, I.; Gutierrez Villamayor, C.; Gaz-
telurrutia Abaitua, L. An. Esp. Pediatr. 1992, 37, 184-186.
(15) Abbreviations: DTA, recombinant diphtheria toxin A chain; EF-2,
eukaryotic elongation factor 2; HPLC, high-pressure liquid chromatography;
IPTG, isopropyl-â-D-thiogalactopyranoside; KIE, kinetic isotope effect;
NAD+, nicotinamide adenine dinucleotide (oxidized form); NaAD+,
nicotinic acid adenine dinucleotide; NaMN+, nicotinic acid mononucleotide;
nLG and nNu, bond orders to the leaving group and to the incipient
nucleophile; nX-Y, bond order between atoms X and Y; rX-Y, bond length
between atoms X and Y; SDS-PAGE, sodium dodecyl sulfate polyacryla-
mide gel electrophoresis; TS, transition state.
(2) Hardy, I. R.; Dittmann, S.; Sutter, R. W. Lancet 1996, 347, 1739-
1744.
(3) Koo, D.; Traverso, H.; Libel, M.; Drasbek, C.; Tauxe, R.; Brandling-
Bennett, D. Bull. Pan Am. Health Org. 1996, 30, 134-143.
(4) Swerdlow, D. L.; Levine, O.; Toole, M. J.; Waldman, R. J.; Tauxe,
R. V. Lancet 1994, 344, 1302-1303.
(5) Siddique, A. K.; Salam, A.; Islam, M. S.; Akram, K.; Majumdar, R.
N.; Zaman, K.; Fronczak, N.; Laston, S. Lancet 1995, 345, 359-361.
(6) Popovic, T.; Fields, P. I.; Olsvik, O.; Wells, J. G.; Evins, G. M.;
Cameron, D. N.; Farmer, J. J. r.; Bopp, C. A.; Wachsmuth, K.; Sack, R.
B.; et al. J. Infect. Dis. 1995, 171, 122-127.
(7) Starke, J. R.; Tan, T. Q.; Chacko, M. R.; Cleary, T. G.; Connelly, K.
K.; Kline, M. W. Texas Med. 1994, 90, 35-45.
(16) Wilson, B. A.; Collier, R. J. Curr. Top. Microbiol. Immunol. 1992,
175, 27-41.
S0002-7863(97)01317-6 CCC: $14.00 © 1997 American Chemical Society

12080 J. Am. Chem. Soc., Vol. 119, No. 50, 1997
Berti et al.
Amersham Life Sciences (Arlington Heights, IL). Except for 18O-
labeled NAD+, isotopically labeled NAD+’s were synthesized as in
Rising and Schramm.31
18O-NAD+’s. [5-18O]Glucose. Labeled glucose was synthesized
from the ketone precursor (the generous gift of Michael Sinnott of
University of Illinois at Chicago) similarly to the method described
previously,32 except that the material was not recrystallized after LiAlH4
reduction. [5-18O]glucose was purified by sequential phosphorylation
with hexokinase, anion-exchange chromatography, dephosphorylation
with glucose 6-phosphatase, and repurification under the same condi-
tions. Hexokinase and glucose 6-phosphatase were dialyzed against
50 mM potassium phosphate (pH 7.5) before use to remove unlabeled
glucose. [5-18O]Glucose (55 µmol in 2 mL) was phosphorylated to
glucose 6-phosphate with 2 u of hexokinase, 60 mM ATP, and 60 mM
MgCl2 in 50 mM potassium phosphate (pH 7.5) for 2 h at room
temperature. The reaction mixture was reduced to approximately half-
volume under vacuum, applied to a 10 mL column of DEAE-Sephadex
A-25 which had been equilibrated with H2O, then washed with 10 ×
1 mL of H2O. [5-18O]Glucose 6-phosphate was eluted with 10 × 1
mL of 1 M NH4OAc. The fractions containing product were identified
using the reducing sugar assay,33 pooled, frozen, and lyophilized twice.
The [5-18O]glucose 6-phosphate was dephosphorlyated with 0.1 u of
glucose 6-phosphatase in 50 mM potassium phosphate (pH 7.5) for 2
h at room temperature. Anion exchange chromatography was per-
formed as above, with [5-18O]glucose eluting in the H2O fractions.
18O-NAD+. [4′-18O]NAD+ and [4′N-18O,8A-14C]NAD+ 34 were syn-
thesized as described previously,31 with several modifications. The
synthesis of nicotinic acid adenine dinucleotide (NaAD+) was done in
two steps to allow incorporation of the 14C-label from [8-14C]ATP. In
the first step, nicotinic acid mononucleotide (NaMN+) was synthe-
sized,31 except that hexokinase, glucose 6-phosphate dehydrogenase,
6-phosphogluconate dehydrogenase, and phosphoriboisomerase were
combined and dialyzed against 50 mM potassium phosphate (pH 7.5)
to minimize addition of unlabeled intermediates from the enzyme
preparations. The NH3 needed for the glutamate dehydrogenase reaction
is normally supplied from the above enzyme preparations, which are
(NH4)2SO4 suspensions; in this reaction, NH3HCO3 was added to 1.5
mM. After 9 h, the reaction was heated at 100 °C for 2 min to stop
the enzyme reactions. Fructose (2 mM), 0.2 u of hexokinase, and 4 u
of myokinase were added to convert ATP to AMP. NaMN+ was
isolated by C18 reverse phase HPLC chromatography on a 7.8 × 300
mm column in 50 mM NH4OAc (pH 5.0) then lyophilized. NaAD+
was synthesized in two parallel reactions. To the lyophilized NaMN+
was added 1 mL of 4 mM ATP, 4 mM MgCl2, 50 mM KCl, 50 mM
potassium phosphate (pH 7.5), and 0.25 u of NAD+ pyrophosphorylase,
plus 10 µCi of [8-14C]ATP in one reaction mixture. The reactions were
followed by HPLC for 15 h. The product NaAD+ was purified, and
NAD+ was synthesized as described previously.31 Starting with water
with an 18O content of 60% to make labeled glucose, the [4′-18O]NAD+
contained 37.1% 18O-label, as determined by mass spectrometry (see
below). 4′-18O KIEs measured by the radiolabel technique were
corrected for the extent of 18O-labeling.
Figure 1. NAD+ molecule with positions of isotopic labels (3H, 14C,
15N, 18O) in bold type.
for treating bacterial disease, inhibitors of enzymes that cleave
N-ribosidic bonds (e.g., diphtheria,17,18 pseudomonas exotoxin,19
ricin,20 and gelonin21 ) have the potential to be used with
immunotoxins in the treatment of cancer, to rescue normal cells
from nonspecific killing by the immunotoxins.22,23 Enzymes
catalyzing other ADP-ribosylation reactions of great biological
interest have recently been described, such as poly(ADP-ribose)
synthase, involved in DNA repair,24 and cyclic-ADP-ribose
synthase,25 involved in intracellular signaling, plus other mono-
ADP-ribosyl transferases.26 Determination of the transition state
structures for diphtheria toxin-catalyzed reactions will be
applicable to other biologically important systems.
In the absence of its physiological target, diphtheria toxin,
like cholera and pertussis toxins, catalyzes a slow hydrolysis
of NAD+ to ADP-ribose and nicotinamide.27 Although the
reaction has no physiological significance, it can be exploited
to measure the kinetic isotope effects (KIEs) of labeled NAD+
substrates (Figure 1). From KIEs determined at many labeled
positions in the NAD+ molecule, it is possible to determine the
transition state structure of the reaction using bond-energy/bond-
order vibrational analysis. Knowing the transition state structure
will permit further characterization of the mechanisms by which
the enzyme stabilizes the enzymatic transition state complex to
promote catalysis. The structure of the enzyme-stabilized
transition state provides a target structure for the design of
transition state analogues as inhibitors. Transition state analyses
of NAD+ hydrolysis catalyzed by cholera28 and pertussis29 toxins
have been reported. In the accompanying article,30 the transition
state structure of solvolytic hydrolysis of NAD+ is reported and
a new, structure interpolation method of transition state structure
determination is described.
Materials and Methods
Materials. Hexokinase, myokinase, and glucose 6-phosphatase were
from Sigma Chemical Co. (St. Louis, MO); [8-14C]ATP was from
(17) Li, B. Y. R. S. J. Biol. Chem. 1994, 269, 2652-2658.
(18) Murphy, J. R.; vanderSpek, J. C. Sem. Cancer Biol. 1995, 6, 259-
267.
Recombinant Diphtheria Toxin A-Chain (DTA). Recombinant
DNA Procedures. An entirely synthetic gene encoding the catalytic
domain of diphtheria toxin was used for expression of protein (S.R.B.
and R. J. Collier, unpublished results). The synthetic gene was designed
by altering the codon usage of the corynephage â-gene to reflect the
bias exhibited by highly expressed proteins in Escherichia coli. The
gene was divided into smaller, evenly spaced fragments by engineering
unique restriction sites throughout the open reading frame. The DTA
synthetic gene was cloned into pET-15b (Novagen, Inc.), replacing the
NcoI-BamHI fragment, and transformed into the E. coli strain BL21
(DE3a) for expression of the proteins under transcriptional control of
the T-7 promoter.
(19) Kreitman, R. J.; Pastan, I. Sem. Cancer Biol. 1995, 6, 297-306.
(20) Roy, D. C.; Ouellet, S.; Le Houillier, C.; Ariniello, P. D.; Perreault,
C.; Lambert, J. M. J. Nat. Cancer Inst. 1996, 88, 1136-1145.
(21) French, R. R.; Penney, C. A.; Browning, A. C.; Stirpe, F.; George,
A. J.; Glennie, M. J. Br. J. Cancer 1995, 71, 986-994.
(22) Gould, B. J.; Borowitz, M. J.; Groves, E. S.; Carter, P. W.; Anthony,
D.; Weiner, L. M.; Frankel, A. E. J. Nat. Cancer Inst. 1989, 81, 775-781.
(23) Pai, L. H.; Pastan, I. In Biologic Therapy of Cancer, 2nd ed.; DeVita,
V. T., Jr., Hellman, S., Rosenberg, S. A., Eds.; Lippincott: Philadelphia,
PA, 1995; pp 521-533.
(24) Sugimura, T.; Miwa, M. Mol. Cell. Biochem. 1994, 138, 5-12.
(25) Lee, H. C.; Galione, A.; Walseth, T. F. Vitam. Horm. 1994, 48,
199-257.
(26) See: ADP-ribosylation reactions; Poirier, G. G., Moreau, P., Eds.;
Springer-Verlag: New York, 1992.
Fermentation and Harvest of E. coli. A 50 mL culture of L-broth
(100 µg/mL ampicillin) was inoculated with a single colony from an
(27) Kandel, J.; Collier, R. J.; Chung, D. W. J. Biol. Chem. 1974, 249,
2088-2097.
(31) Rising, K. A.; Schramm, V. L. J. Am. Chem. Soc. 1994, 116, 6531-
6536.
(32) Bennet, A. J.; Sinnott, M. L. J. Am. Chem. Soc. 1986, 108, 7287-
7294.
(33) Parkin, D. W.; Horenstein, B. A.; Abdulah, D. R.; Estupinan, B.;
Schramm, V. L. J. Biol. Chem. 1991, 266, 20658-20665.
(28) Rising, K. A.; Schramm, V. L. J. Am. Chem. Soc. 1997, 119, 27-
37.
(29) Scheuring, J.; Schramm, V. L. Biochemistry 1997, 36, 4526-4534.
(30) Berti, P. J.; Schramm, V. L. J. Am. Chem. Soc. 1997, 119, 12069-
12078.

Transition State of DTA-Catalyzed NAD+ Hydrolysis
J. Am. Chem. Soc., Vol. 119, No. 50, 1997 12081
LB-amp plate freshly streaked with BL21(DE3a) transformed with pET-
15b-DTA. This inoculum was grown to an OD600 ) 0.5 and then placed
at 4 °C overnight. On the second day, 2 L baffled flasks, each
containing 500 mL of L-broth (100 µg/mL ampicillin) prewarmed to
37 °C, were inoculated with 5 mL of the 50 mL overnight culture. The
flasks were agitated on a rotary shaker at 37 °C. The cultures were
induced at an OD600 ) 1.0 with 1-2 mM isopropyl-â-D-thiogalacto-
pyranoside (IPTG). The cells were harvested after 1-2 h, immediately
chilled to 4 °C, and centrifuged at 3000g for 10 min at 4 °C. The
pellets were resuspended in 5 mL of ice cold sonication buffer (50
mM Na2HPO4 (pH 8.0), 100 mM KCl, 0.1% Tween-20, 1.0 mM
(phenylmethyl)sulfonyl fluoride, and 20 mM â-mercaptoethanol). The
resuspended pellets were either frozen at -70 °C or sonicated
immediately.
Purification of Mutant Toxins. The combined pellets were
sonicated three times on ice for 30-45 s, using a Sonifier Cell Disruptor
350 (Branson Sonic Power Co.) with the power control at 5 and the
duty cycle at 40%. The extracts were chilled on ice for at least 1 min
between each sonication cycle. Cellular debris was pelleted by
centrifuging at 5000g for 30 min at 4 °C. The supernatants were
collected and chilled while the pellets were resuspended in 5-10 mL
of sonication buffer and sonicated again as described above.
The synthetic gene encoding DTA was designed with a polyhistidine
N-terminal fusion peptide, facilitating purification by nickel chelate
affinity chromatography. The crude extracts were clarified immediately
before chromatography by centrifuging at 20000g for 15 min The
extracts were loaded directly onto a 3-5 mL nickel chelate affinity
column (Qiagen) that had been preequilibrated with 10 bed volumes
of sonication buffer. The column was washed with 5 bed volumes of
sonication buffer plus 20 mM imidazole, followed by 2 bed volumes
of 25 mM Tris‚HCl (pH 8.0) plus 20 mM imidazole. The purified
proteins were eluted with 5 bed volumes of 25 mM Tris‚HCl (pH 8.0)
plus 50 mM imidazole. The column fractions were analyzed by SDS-
PAGE, and the purified protein was pooled.
Proteolytic Removal of the Polyhistidine Fusion Peptide. The
synthetic gene was designed with a modified thrombin recognition
sequence (Leu-Val-Pro-Arg-Gly-Ala) linking the C-terminus of the
polyhistidine fusion peptide to the N-terminus of DTA. This design
ensured that cleavage of the Arg-Gly peptide bond with thrombin
results in production of DTA with an authentic amino-terminal sequence
(Gly-Ala). Typically, 1 mg of DTA was incubated with 2-5 u of
human thrombin (Novagen) for 1-5 days at 20 °C in 20 mM Tris‚-
HCl (pH 8.4), 150 mM NaCl, 2.5 mM CaCl2, and 2 mM DTT. The
intact DTA was purified using a single anion-exchange FPLC chro-
matography fractionation step (FPLC: Mono-Q, Pharmacia).
Alternatively, the polyhistidine fusion peptide was removed by
limited trypsin digestion, also yielding an authentic amino terminus.
Generally, 1.5 mg of protein in 20 mM Tris‚HCl (pH 7.4) and 10 mM
â-mercaptoethanol was incubated with 7.5 µg of bovine trypsin (Sigma)
for 60 min on ice. The entire incubation mixture was loaded on a 1.0
mL nickel chelation affinity column to bind uncut DTA as well as the
cleaved polyhistidine peptide. The flow-through fractions were col-
lected and combined with two 1.0 mL washes of 20 mM Tris‚HCl
(pH 7.4) and 10 mM â-mercaptoethanol. This protein was further
purified using anion exchange chromatography (FPLC Mono Q from
Pharmacia). Side fractions were pooled, desalted, and repurified using
anion exchange chromatography.
Commitment to Catalysis. The isotope trapping method was used
to detect any commitment to catalysis. DTA (76 µM final concentration
in 10 µL) was mixed with NAD+ (36-1000 µM) containing 2.4 ×
105 cpm [4-3H]NAD+ in 50 mM potassium phosphate (pH 6.0) for 5
s, then an excess of unlabeled NAD+ (100 µL of 10 mM) was added
and the reaction allowed to proceed at 37 °C for 5kcat
.
[4-3H]-
Nicotinamide released in the reaction was measured as described,28 and
the commitment factor calculated by plotting 1/{[NAD+]trapped/[DTA-
]
total} versus 1/[NAD+]0. The y-intercept represents the fraction of
NAD+ that would be trapped at infinite substrate concentration.
Radiolabel Technique. KIEs were measured by the competitive
method with radiolabeled substrates by the same method as described
previously,28 including the complete reaction with NAD+ glycohydro-
lase and purification and liquid scintillation counting of the product
ADP-ribose. Hydrolysis of labeled NAD+’s by DTA was performed
as above with 3 µM DTA for 3 h. The competitive radiolabel method
entails measuring the relative rates of hydrolysis of an NAD+
3
isotopically labeled with H or 14C at the position of interest and one
labeled at a remote site with the other radionuclide, which acts as a
reporter for the natural isotope at the position of interest. The KIE is
3
determined from the ratios of 14C/3H (for a H KIE) in the partial and
complete reactions, with 14C acting as a reporter on 1H at the position
of interest. Observed KIEs calculated this way are corrected to 0%
reaction to allow for depletion of the faster reacting isotope.36 The
remote label for determination of the 1′-14C, 4′-18O, and 1-15N KIEs
was 4′-3H. Since the measured 4′-3H KIE ) 0.990 ( 0.002 was
significantly different from unity, the heavy atom KIEs were corrected
by multiplying by the measured 4′-3H KIE. All the 3H KIEs were
measured using 5′-14C as the remote label, which was shown to have
no measurable KIE for cholera toxin or solvolytic hydrolysis of NAD+.28
The 4′-18O and 1-15N KIEs were measured with doubly labeled NAD+’s,
[4′-18ON,8-14CA] and [1-15N,5′-14C], with the remote radionuclide acting
as a reporter for the stable heavy isotope.
Mass Spectrometry. The 4′-18O KIE was also measured by the
competitive method using whole-molecule positive-ion electrospray
mass spectrometry on residual NAD+ after reaction to approximately
50% completion. Anderson and co-workers have shown that KIEs can
be determined by whole-molecule mass spectrometry with precision
similar to the radiolabel method.37,38 [4′-18O]NAD+ (200 µM total
concentration, 37.1% 4′-18O) with 6 µM DTA in 100 µL of 50 mM
potassium phosphate (pH 6.0) was reacted at 37 °C for approximately
3 h, until the reaction was 50% complete. The extent of the reaction
was determined by HPLC on a C18 reverse-phase column, 4 × 300
mm, in 50 mM ammonium acetate (pH 5.0). The relative peak areas
(A260) of equimolar ADP-ribose and NAD+ under these conditions are
0.76(( 0.03):1. The remaining NAD+ was purified from the reaction
mixture under the same HPLC conditions and >99% of the peak was
collected and lyophilized overnight. The sample was redissolved in 2
mL of water and lyophilized again to remove volatile salts. The sample
was dissolved in 100 µL of 0.1% aqueous trifluoroacetic acid for mass
spectrometry. The substrate NAD+ (0% reaction) sample was prepared
in the same way, except without DTA. Mass spectrometry of [4′-18O]-
NAD+ without repurification gave identical results (not shown),
demonstrating that the HPLC purification did not cause fractionation
of the labeled NAD+.
Electrospray mass spectra of NAD+ were collected in single ion
monitoring mode at m/z ) 664.1 and 666.1, with a dwell time of 50
ms at each m/z. The former peak corresponds to unlabeled NAD+,
the latter to [4′-18O]NAD+ plus molecules doubly substituted with
KIE Measurement and Enzyme Kinetics. Kinetic Constants and
Steady State Conditions. Hydrolysis of labeled NAD+’s by DTA was
performed with 100 µM total NAD+ in 50 mM potassium phosphate
(pH 6.0). The kinetic constants of hydrolysis were measured as
described previously.28 Because the enzyme concentration was a
significant fraction of the substrate concentration, Selwyn’s test35 was
used to ensure that steady state conditions still obtain. If steady state
conditions hold, reactions performed at constant [NAD+] and varied
[DTA] will give constant [product] for a given [DTA] × time; this
was true up to the highest [DTA] tested, 16 µM (data not shown).
2
natural abundance H, 13C, 15N, and natural abundance 18O. Samples
were infused at 2 µL/min for g10 min per run, with three runs per
KIE measurement.
Unlike KIEs measured by liquid scintillation counting, the molar
amount of labeled material is not negligible compared with unlabeled
material, so different equations are used to calculate the KIE. Also, it
is necessary to account for the natural abundance isotopes in the peak
(36) Parkin, D. W. In Enzyme mechanism from isotope effects; Cook, P.
F., Ed.; CRC Press Inc.: Boca Raton, FL, 1991; pp 269-290.
(37) Gawlita, E.; Paneth, P.; Anderson, V. E. Biochemistry 1995, 34,
6050-6058.
(38) Bahnson, B. J.; Anderson, V. E. Biochemistry 1991, 30, 5894-
5906.
(34) In this substrate (only), one of the labels occurs in the adenosyl
portion of the NAD+ molecule, 8-14C of the adenine ring. This is indicated
by the subscripted “A” in the label, to distinguish it from the 5′-18O label
in the nicotinamidyl portion, “N”.
(35) Selwyn, M. J. Biochim. Biophys. Acta 1965, 105, 193-195.

12082 J. Am. Chem. Soc., Vol. 119, No. 50, 1997
Berti et al.
Figure 2. Reaction coordinate diagram. Positions in reaction coordinate
space are illustrated by plotting the leaving group bond order (nLG) on
the ordinate and nucleophile bond order on the abscissa (nNu). Thus,
the reaction proceeds from the substrates in the lower left corner to
products in the upper right. A classical DN + AN (SN1) reaction
mechanism involves formation of an oxocarbenium ion intermediate,
with complete loss of the leaving group, before the nucleophile bond
order starts to increase. A concerted, synchronous ANDN (SN2)
mechanism follows the dotted diagonal line, where increase in nNu at
Figure 3. Distortional KIEs as modeled with EtOH. The bond angle,
θ ) C2-C1-O1, is analogous to the bond angle C4′-C5′-O5′
of NAD+ and the 1-3H KIE analogous to the 5′-3H KIE.
5′-3H KIE. Two sources for the large 5′-3H KIE were considered:
distortion of the C4′-C5′-O5′ bond angle and protonation of the
R-phosphate. The distortional KIE was examined using ethanol as a
model compound (Figure 3). The reactant model was ethanol optimized
using an hybrid density functional theory minimization using Becke’s
exchange functional42 and Perdew and Wang’s correlation functional43
with the 3-21+G** basis set using the program Gaussian 94.44 The
effects of changes in the C2-C1-O1 bond angle were analyzed by
fixing it at 20° greater or less than the fully optimized structure at 106.3°
and reoptimizing the rest of the ethanol molecule. Structures at values
of C2-C1-O1 between the extremes were calculated by varying
the internal coordinates (bond lengths, bond angles, and torsional angles)
between the optimal structure and the constrained structures with the
program CTBi. 1-3H KIEs were calculated for each structure using
BEBOVIB, with bond stretching and bond angle bending force
constants taken from the AMBER force field of Kollman’s group45
after conversion to appropriate units. The torsional force constants
were 0.01 mdyn‚Å/rad2 for the C-O bond and 0.02 mdyn‚Å/rad2 for
C-C.46
each point matches exactly the loss of nLG
.
at m/z ) 666.1, which is 0.060 times the abundance of the isotopically
unsubstituted peak at m/z ) 664.1. The equation of Bigeleisen and
Wolfsberg39 is
18k ) {ln(1 - [18P]/[18S]0)}/{ln(1 - [16P]/[16S]0)}
where [18P] is the concentration of 18O-labeled product and [18S]0 is
the initial concentration of 18O-labeled substrate. This can be rearranged
to
18k ) ln[(1 - f)(1 + r0-1)/(1 + ri-1)]/ln[(1 - f)(1 + r0)/(1 + ri)]
where f () [P]/[S]0) is the fractional extent of reaction as determined
The effects of phosphate protonation were examined by optimizing
ethyl phosphate in the neutral and monoanion forms at the same level
of theory as ethanol. 1-3H KIEs were calculated from the fractionation
factors derived by the program QUIVER47 from the optimized
structures.
by the HPLC assay, and ri and r0 are the ratio of peak intensities (I664.1
,
I
666.1) at times t ) i and 0, corrected for the natural abundance of m/z
) 666.1, which was calculated from the natural abundances of elemental
13C, 15N, H, and 18O:
2
rx ) 1/[(I664.1/I666.1) - 0.060]
Results
Since the majority of atoms of NAD+ are in isotopically insensitive
positions, it was assumed that this ratio did not vary in the course of
the reaction. In deriving this equation, we do not assume that [18S]/
([18S] + [16S]) ≈ [18S]/[16S].40
Kinetic Constants. Hydrolysis of NAD+ by DTA was
conducted in 50 mM potassium phosphate buffer (pH 6.0) at
37 °C. At this pH, the nonenzymatic, base-catalyzed hydrolysis
which occurs above pH 7 was minimized. Under these
conditions, DTA hydrolyzes NAD+ with kinetic constants of
Transition State Structure Determination. Bond-Energy/Bond-
Order Vibrational Analysis. The transition state structure was
determined by the structure interpolation method as described in the
accompanying report. Briefly, trial transition state structures are
generated automatically by the program CTBi30 for selected values of
kcat ) 3.0 (( 0.3) × 10-3 s-1, kcat/KM ) 10.9 ( 0.4 M-1‚s-1
,
and KM ) 2.7 (( 0.3) × 10-4 M (Figure 4). KM closely
approximates Kd, since there is no external commitment to
catalysis (see below). The observed KM is greater than reported
nLG,TS and nNu,TS as a function of the “oxocarbenium ion character” of
the ribosyl moiety, and their KIEs are predicted using the program
BEBOVIB.41 Reaction coordinates for concerted (with leaving group
departure) and unimolecular attack of the nucleophile were investigated.
Unimolecular departure of the leaving group is not a reasonable
mechanism given the crystal structure but was investigated anyway.
The predicted transition state structure was selected by plotting, for
each isotopic label, the areas on a More-O’Ferrall-type reaction
coordinate diagram (Figure 2) where the measured and predicted KIEs
matched. The point where all the KIEs matched identifies the transition
state structure.
(42) Becke, A. D. Phys. ReV. A 1988, 38, 3098-3100.
(43) Perdew, J. P.; Wang, Y. Phys. ReV. B 1992, 45, 13244.
(44) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.;
Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G.
A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.;
Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.;
Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; Head-
Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94, ReVision C.2; Gaussian,
Inc.: Pittsburgh, PA, 1995.
(45) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K.
M., Jr.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.;
Kollman, P. A. J. Am. Chem. Soc. 1995, 117, 5179-5197.
(46) Eyster, J. M.; Prohofsky, E. W. Spectrochim. Acta 1974, 30A, 2041-
2046.
(39) Bigeleisen, J.; Wolfsberg, M. AdV. Chem. Phys. 1958, 1, 15-76.
(40) Melander, L.; Saunders, W. H., Jr. Reaction rates of isotopic
molecules; John Wiley & Sons: New York, 1980; p 98.
(41) Sims, L. B.; Burton, G. W.; Lewis, D. E. BEBOVIB-IV, QCPE No.
337; Quantum Chemistry Program Exchange, Department of Chemistry,
University of Indiana: Bloomington, IN, 1977.
(47) Saunders, M.; Laidig, K. E.; Wolfsberg, M. J. Am. Chem. Soc. 1989,
111, 8989-8994.

Transition State of DTA-Catalyzed NAD+ Hydrolysis
J. Am. Chem. Soc., Vol. 119, No. 50, 1997 12083
Table 1. Measured and Predicted KIEs of DTA Catalyzed NAD+
Hydrolysis
measd KIE
(( 95%
confidence
interval)
predicted
KIE at
transition
label of interest
1-15Nc
KIE type
na
stateb
primary, leaving 1.030 ( 0.004
group
primary
4
1.029
1.034
1′-14Cd
1.034 ( 0.004
4
2
4
3
3
4
3
1-15N,1′-14Ce
1′-3Hf
double primary 1.062 ( 0.010
R-secondary
â-secondary
γ-secondary
δ-secondary
R-secondary
1.200 ( 0.005
1.142 ( 0.005
0.990 ( 0.002
1.032 ( 0.004
0.991 ( 0.003
1.186
1.141
2′-3Hf
4′-3Hf
-
g
5′-3Hf
-
h
4′-18O, mass
spectrometryi
4′-18O,
-
0.986 ( 0.003
3
-
radiolabeledj
4′-18O, combined
0.988 ( 0.003
6
0.985
a Number of independent, quadruplicate determinations (triplicate
Figure 4. Hanes plot of kinetic constants for DTA-catalyzed NAD+
hydrolysis in 50 mM potassium phosphate (pH 6.0) at 37 °C.
b
for mass spectrometry). nLG,TS ) 0.02 and nNu,TS ) 0.03. c Labeled
substrate: [1-15N,5′-14C]NAD+. Remotely labeled substrate: [4′-
3H]NAD+. d Remotely labeled substrate: [4′-3H]NAD+. e Labeled sub-
strate: [1-15N,1′-14C]NAD+. Remotely labeled substrate: [4′-3H]NAD+.
f Remotely labeled substrate: [5′-14C]NAD+. g 4′-3H KIE was not
modeled quantitatively. h 5′-3H KIE could be matched by distorting the
i
C4′-C5′-O5′ bond angle or protonating the R-phosphate. Measured
by positive-ion whole-molecule mass spectrometry. Labeled substrate:
[4′-18O]NAD+. j Labeled substrate: [4′N-18O,8A-14C]NAD+. Remotely
labeled substrate: [4′-3H]NAD+.
ribosylation of EF-2. The lower limit on the rate of product
release is set by kcat ) 20 s-1 for the ADP-ribosylation of EF-
2.48 Even if product release is fully rate-limiting in that reaction
(for which there is no evidence for or against), that would be 7
× 103-fold faster than kcat ) 3.0 × 10-3 s-1 for the hydrolysis
reaction, further evidence that the chemical step is rate-limiting.
Measured KIEs. The KIEs measured in this study were
determined as described previously28 using the competitive
method with radiolabeled substrates (Table 1). The 4′-18O KIE
was also determined by mass spectrometry. The KIEs measured
by each technique were within experimental error of each other
(Table 1, and see the Discussion).
Figure 5. Commitment to catalysis. The ordinate intercept, 358, is
the reciprocal of the fraction of NAD+ that would react with DTA,
without dissociation, at infinite concentration of NAD+. This corre-
sponds to a commitment factor of 0.0028.
previously at pH 8.0, 2.0 (( 0.5) × 10-5 M,48 though kcat/KM
Transition State Structure. The transition state structure
that matched the experimental KIEs had high oxocarbenium ion
character, with nLG,TS ) 0.02 and nNu,TS ) 0.03. This is a
) 12.5 M-1‚s-1 is similar.
Commitment to Catalysis. KIEs in kcat/KM report on the
rate-limiting step (or steps if more than one is significantly rate-
limiting) up to the first irreversible step of the reaction.
Substrate molecules that bind to the enzyme and form products
without dissociating from the Michaelis complex cause com-
mitment to catalysis and decrease the experimentally measured
KIEs. For full commitment, only binding isotope effects on
formation of the Michaelis complex would be observed. The
KIEs of the chemical step would be kinetically unobservable.
The extent of commitment to catalysis was quantitated using
an isotope trapping experiment.49 The results show that, at
infinite substrate concentration, less than 0.3% of substrate
would proceed forward to the chemical step without dissociation
(Figure 5). Thus, the extent of forward commitment to catalysis
is negligible.
50
concerted ANDN (SN2) reaction, with both the leaving group
and nucleophile participating in the reaction coordinate. No
combination of nLG and nNu was found for DN + AN (SN1) or
DN*AN (unimolecular attack of the nucleophile on the oxocar-
benium ion) mechanisms that matched the experimental KIEs.
The predicted transition state structure was identified from a
contour plot of the area in reaction coordinate space in which
the predicted KIEs for each label matched the measured one
(Figure 6). The point where the contours converge indicates
the transition state structure, at nLG ) 0.02 and nNu ) 0.03.
The predicted KIEs for 4′-3H and 5′-3H were small throughout
reaction coordinate space and are therefore not shown. This is
consistent with the results from the solvolytic hydrolysis of
NAD+, which shows that for a similar transition state structure
(nLG,TS ) 0.02, nNu,TS ) 0.005) the experimental and predicted
KIEs at H4′ and H5′ are negligible for the intrinsic chemical
step. Thus the observed remote KIEs for the DTA reaction arise
from the enzymatic stabilization of the transition state (see the
Discussion).
In principle, product release could be rate-limiting and
therefore reflected in the experimental KIEs. If it were, a burst
of product formation would be observed under pre-steady-state
conditions. In fact, the rate of product appearance is constant
for reaction times as short as one-fifth of a turnover (not shown),
demonstrating that product release is not rate-limiting. Further
evidence comes from the kinetic constants for the ADP-
4′-3H KIE. The enzymatic 4′-3H KIE, 0.990 ( 0.002, was
significantly more inverse than for the solvolytic hydrolysis
(0.997 ( 0.001). Together with the 5′-3H KIE, these effects
(48) Blanke, S. R.; Huang, K.; Collier, R. J. Biochemistry 1994, 33,
15494-15500.
(49) Rose, I. W. Methods Enzymol. 1980, 64, 47-59.
(50) Guthrie, R. D.; Jencks, W. P. Acc. Chem. Res. 1989, 22, 343-349.

12084 J. Am. Chem. Soc., Vol. 119, No. 50, 1997
Berti et al.
The radiolabel technique is an inherently indirect measure,
using liquid scintillation counting of radionuclides as reporters
on 18O and 16O, whereas mass spectrometry directly measures
the isotopically pure 18O- and 16O-containing NAD+ molecules
(after compensating for the natural abundance of m/z ) 666.1).
With the radiolabel technique, isotope ratios were measured in
the products of the reaction after isolation on ion exchange
columns, while the mass spectrometry KIEs were measured from
residual substrate after isolation by reverse phase HPLC. Given
these differences in the measurement techniques, the identity
of the measured KIEs within experimental error demonstrates
the reliability of the measurement methods.
Description of KIEs. The predicted KIEs for the proposed
transition state structure match all measured KIEs within the
95% confidence interval and are all within 0.002 of the exact
measurement, with the sole exception of the 1′-3H KIE. Given
the similar 1′-3H KIEs measured for NAD+ solvolysis (1.194)30
and hydrolysis by cholera (1.186)28 and pertussis (1.207)29
toxins, and the similarity to other N-riboglycoside hydrolyses,52
the measured KIE is also likely to be correct. Computational
prediction of the 1′-3H KIE is difficult since the range of the
predicted KIEs within reaction coordinate space varies from 0.60
to 1.36 and was larger than for any other label. Within this
range, the difference between the predicted (1.186) and mea-
sured (1.200) KIEs at the transition state is small. The transition
state structure derived with the structure interpolation method
is consistent with a qualitative evaluation of the measured KIEs,
with previously reported kinetic and structural studies of
DTA (see below), and with other enzymatic N-riboside
hydrolyses.28,29,53-58
[1-15N,1′-14C] KIE. The double primary, [1-15N,1′-14C] KIE
) 1.062 ( 0.010 (n ) 2), was measured and found to be, within
experimental error, equal to the product of the individual KIEs,
1.065. This relationship was also observed for the solvolytic
hydrolysis of NAD+, where the measured [1-15N,1′-14C] KIE
) 1.034 ( 0.002 (n ) 6) was equal to the product of the
individual KIEs, 1.036. That the measured double KIE equals
the product of the individual ones is a necessary consequence
(though not diagnostic) of a mechanism where a single step is
rate-limiting. The equality of those values demonstrates the
accuracy and reliability of the KIE measurements.
Figure 6. Match of predicted versus experimental KIEs. For each
isotopic label, the colored area represents the match of the predicted
with the experimental KIEs as a function of nLG and nNu in this More-
O’Ferrall-type reaction coordinate diagram. The light shading represents
the 95% confidence interval of each measured KIE, with dark shading
representing the exact measured KIE (approximately ( 0.001): blue,
1-15N; red, 1′-14C; gray, 1′-3H; green, 2′-3H; orange, 4′-18O. The
predicted transition state structure is indicated by *. Only the top left
(dissociative) corner of reaction coordinate space is shown.
demonstrate that the enzyme is using binding energy remote
from the catalytic site to promote catalysis. No attempt was
made to quantitatively model the enzymatic 4′-3H KIE. A
modest change in any of the bond angles between C4′ and O4′,
C3′, or C5′ could cause a KIE of this relatively small magnitude.
5′-3H KIE. The experimental 5′-3H KIE of 1.032 ( 0.004
matched the predicted KIEs the ethanol model at C2-C1-
O1 angles -14° and +15° from the equilibrium value of 106°,
that is, at 92° and 121°, respectively (Figure 3). The calculated
secondary 1-3H KIE for the protonation of ethyl phosphate was
1.049, also of sufficient magnitude to account for the experi-
mental result.
Other KIEs. The combination of the relatively large 1-15N
KIE, indicating large loss of C1′-N1 bond order, and small
1′-14C KIE, indicating low bond order to the approaching
nucleophile, is diagnostic of a dissociative transition state. This
is supported by the secondary KIEs. The large 1′-3H ) 1.200
( 0.005 is consistent with a dissociative transition state.
Rehybridization of C1′ toward sp2 plus a decrease in the steric
crowding around H1′ led to a decrease in the out-of-plane
bending forces, which results in a large, normal KIE.59 The
large 2′-3H KIE ) 1.142 is diagnostic of hyperconjugation,
which is evidence for the ribose ring conformation in the
transition state being 3′-exo. Hyperconjugation, as distinct from
inductive effects, is highly sensitive to the angle between the
Discussion
Meaning of KIEs. KIEs reflect differences in the energy of
each isotope between two states. The competitive method
measures KIEs on kcat/KM (or V/K),51 where the relevant species
are the reactant (free enzyme and free substrate in solution) and
the enzymatic transition state complex. This is true both for
radiolabeled substrates where the molar concentrations of the
radiolabels are small fractions of the total NAD+ concentration
and for the stable, nontrace 18O-labeled NAD+.
Mass Spectrometric Measurement of KIE. Because of the
indirect nature of KIE measurements by the dual-radiolabel
technique and the difficulty in checking the results by other
means, as well as the susceptibility to interference and the small
size of the effects being measured, the accuracy and precision
of measured KIEs is always a source of concern. These
concerns are compounded when dual remote labels must be used
to measure KIEs of stable isotopes such as 15N and 18O. An
important validation of the dual remote label method was
obtained when the 4′-18O KIE measured by mass spectrometry
was within experimental error of the one measured with dual
remote radiolabels: 0.986 ( 0.003 using the radiolabel tech-
nique versus 0.991 ( 0.003 by mass spectrometry (Table 1).
(52) Schramm, V. L. In Enzyme mechanism from isotope effects; Cook,
P. F., Ed.; CRC Press Inc.: Boca Raton, FL, 1991; pp 367-388.
(53) Kline, P. C.; Schramm, V. L. Biochemistry 1995, 34, 1153-1162.
(54) Kline, P. C.; Schramm, V. L. Biochemistry 1993, 32, 13212-13219.
(55) Horenstein, B. A.; Parkin, D. W.; Estupinan, B.; Schramm, V. L.
Biochemistry 1991, 30, 10788-10795.
(56) Mentch, F.; Parkin, D. W.; Schramm, V. L. Biochemistry 1987, 26,
921-930.
(57) Parkin, D. W.; Schramm, V. L. Biochemistry 1987, 26, 913-920.
(58) Parkin, D. W.; Leung, H. B.; Schramm, V. L. J. Biol. Chem. 1984,
259, 9411-9417.
(59) Poirier, R. A.; Wang, Y.; Westaway, K. C. J. Am. Chem. Soc. 1994,
116, 2526-2533.
(51) Northrop, D. B. Annu. ReV. Biochem. 1981, 50, 103-131.

Transition State of DTA-Catalyzed NAD+ Hydrolysis
J. Am. Chem. Soc., Vol. 119, No. 50, 1997 12085
C2′-H2′ bond and the scissile C1′-N1 bond,52,60 being
maximal when the bonds are eclipsed or anti and zero when
the dihedral angle is 90°. The planarity of the sp2-like O4′ and
C1′ atoms forces ring atoms C4′-O4′-C1′-C2′ to be coplanar.
The only conformational flexibility within the ribose ring is at
C3′, which can be either 3′-exo or 3′-endo. In the 3′-exo
conformation, the C2′-H2′ and C1′-N1 bonds are in nearly
perfect eclipse, with a dihedral angle of 0.1°. In the 3′-endo
conformation, the angle is 59.8° (not shown). The magnitude
of the measured 2′-3H KIE is near the limit expected for a
glycosidic ring system61 and similar to previously measured 2′-
3H KIEs for ribosyl N-glycosides.28,29,52,55 The inverse 4′-18O
KIE occurs because loss of C1′-N1 bond order is compensated
for in the transition state by an increase in the C1′-O4′ bond
order. This increased bond order to the ring oxygen atom
restricts its vibrational environment, resulting in an inverse KIE.
The transition state structures of N-riboside hydrolyses for
other enzymes derived in this lab28,29,53-58 have consistently
indicated highly oxocarbenium ion-like transition state struc-
tures, including NAD+ hydrolysis catalyzed by cholera and
pertussis toxins. Indeed, all evidence indicates that glycoside
hydrolyses in general, enzymatic and nonenzymatic, proceed
through oxocarbenium ion-like transition state structures.62-75
4′-3H and 5′-3H KIEs. In the solvolytic hydrolysis of NAD+,
the measured KIEs at the remote sites are unmeasurably small
(5′-3H KIE ) 1.000 ( 0.003) or negligible (4′-3H KIE ) 0.997
( 0.001). Thus, the significant experimental measured KIEs
for the enzymatic reaction (1.032 ( 0.004 and 0.990 ( 0.002,
respectively) are not intrinsic to the chemical features of the
transition state. They reflect the influence of the enzymatic
binding energy at sites remote from the catalytic site to promote
catalysis that results in structural changes at C4′ and C5′. A
small, inverse 4′-3H and significant, normal 5′-3H KIE are
becoming ubiquitous features of enzymatic N-riboglycoside
environment of the R-phosphate, or some combination of these
two effects. Protonation of ethyl phosphate gave a 1-3H KIE
(1.049) larger than the experimental 5′-3H KIE. This KIE arises
from inductive electron withdrawal toward the phosphate leading
to weaker C-H bonds. This result demonstrates that a change
in the electrostatic environment of the R-phosphate of NAD+
at the transition state, such as protonation or ion-pair formation
by interaction with His21 of DTA, could account for the
experimental KIE. Either increasing or decreasing the C2-
C1-O1 angle in ethanol results in normal KIEs, with the
predicted KIEs matching the experimental 5′-3H KIE at angles
of 92° and 121°. The calculated KIEs arise from a decrease in
the C-C-H, O-C-H, and H-C-H bending force constants
caused by the distortion of C1 of the model ethanol.55 It is not
yet possible to distinguish between these two possibilities as
the source of the experimental 5′-3H KIE.
Transition State Structure. The following structural changes
can be proposed as the NAD+ proceeds from the reactant
(nLG,react ) 0.77) to the transition state (nLG,TS ) 0.02, nNu,TS
)
0.03): The loss of bond order to the leaving group localizes
the positive charge from the ground state onto the ribose ring.
The loss of leaving group bond order, ∆nLG ) -0.75, is
compensated for by increases in bond orders within the ribose
ring, ∆nC1′-O4′ ) +0.63, ∆nC1′-C2′ ) +0.22, and ∆nC1′-H1′
)
+0.04, plus the appearance of the nucleophile, ∆nNu ) +0.03.
The sum of bond orders to C1′, ∑nC1′, is greater in the transition
state (3.78) than in the reactant molecule (3.62). As discussed
in the accompanying report, this change is likely due to the
increased ability of C1′ to form π-bonding interactions with
O4′ and C2′ when rehybridized toward sp2. This has also been
observed in the transition state structures of cholera28 and
pertussis29 toxin reactions. Another change that occurs is the
flattening of the ribose ring. With the sp2 character of C1′,
C2′, and O4′ increasing at the transition state, the atoms C4′-
O4′-C2′-C1′ become coplanar. C3′ is the only ring atom that
is not part of the plane; it projects below the plane in a 3′-exo
conformation. Flattening of the ribose ring and the sp2
rehybridization of C1′ cause the nicotinamide ring and the
phosphate to move closer together, above the plane of the ribose
ring.
3
hydrolysis.28,29,54-56 The remote H KIEs were analyzed in a
separate step from the KIEs adjacent to the breaking bonds,
since these effects are not inherent to the intrinsic chemical step
of the reaction.
Ab initio calculations using small molecule models showed
that the experimental 5′-3H KIE could arise from distortion of
the C4′-C5′-O5′ bond angle, by a change in the electrostatic
The charge distribution of NAD+ changes substantially
between the reactant and transition states (Figure 7). In the
reactant, the nitrogen of the nicotinamide ring bears a formal
positive charge of 1+. At the electrostatic potential surface,
the positive charge (red) is delocalized throughout the nicoti-
namide ring and into C1′ and H1′ of ribose. In the transition
state, the nicotinamide ring becomes neutral and the positive
charge transfers to the ribosyl ring. The electrostatic surface
of the transition state is similar to that for a ribosyl oxocarbe-
nium ion and nicotinamide which are not interacting with each
other.
(60) Sunko, D. E.; Szele, I.; Hehre, W. J. J. Am. Chem. Soc. 1977, 99,
5000-5004.
(61) Ashwell, M.; Guo, X.; Sinnott, M. L. J. Am. Chem. Soc. 1992, 114,
10158-10166.
(62) Sinnott, M. L. Chem. ReV. 1990, 90, 1171-1202.
(63) Padmaperuma, B.; Sinnott, M. L. Carbohydr. Res. 1993, 250, 79-
86.
(64) Tull, D.; Withers, S. G. Biochemistry 1994, 33, 6363-6370.
(65) Tanaka, Y.; Tao, W.; Blanchard, J. S.; Hehre, E. J. J. Biol. Chem.
1994, 269, 32306-32312.
(66) Wang, Q. P.; Graham, R. W.; Trimbur, D.; Warren, R. A. J.; Withers,
S. G. J. Am. Chem. Soc. 1994, 116, 11594-11595.
(67) Braun, C.; Brayer, G. D.; Withers, S. G. J. Biol. Chem. 1995, 270,
26778-26781.
(68) McIntosh, L. P.; Hand, G.; Johnson, P. E.; Joshi, M. D.; Korner,
M.; Plesniak, L. A.; Ziser, L.; Wakarchuk, W. W.; Withers, S. G.
Biochemistry 1996, 35, 9958-9966.
(69) Sinnott, M. L.; Jencks, W. P. J. Am. Chem. Soc. 1980, 102, 2026-
2032.
(70) Kneir, B. L.; Jencks, W. P. J. Am. Chem. Soc. 1980, 102, 6789-
6798.
(71) Amyes, T. L.; Jencks, W. P. J. Am. Chem. Soc. 1989, 111, 7888-
7900.
(72) Banait, N. S.; Jencks, W. P. J. Am. Chem. Soc. 1991, 113, 7951-
7958.
(73) Banait, N. S.; Jencks, W. P. J. Am. Chem. Soc. 1991, 113, 7958-
7963.
(74) Cherian, X. M.; Van Arman, S. A.; Czarnick, A. W. J. Am. Chem.
Soc. 1990, 112, 4490-4498.
(75) Huang, X.; Surry, C.; Hiebert, T.; Bennet, A. J. J. Am. Chem. Soc.
1995, 117, 10614-10621.
Comparison with Other Transition State Structures. The
transition state structures for NAD+ hydrolysis by cholera toxin
(nLG,TS ) 0.09, nNu,TS ) 0.005)28 and pertussis toxin (nLG,TS
)
0.1, nNu,TS ) 0.001)29 catalyzed hydrolysis of NAD+ have been
reported. The reported leaving group bond orders were
somewhat higher and the nucleophile bond orders somewhat
lower than reported here for DTA. In those studies, the increase
in C-N bond orders within the nicotinamide ring in the
transition state was not included in the model, resulting in
increased residual leaving group bond order in the transition
state. The increase in predicted nLG’s leads, in turn, to low
predicted nNu’s to match the ribosyl KIEs. Using the structure
interpolation method, the predicted transition state bond orders
for these reactions become nLG,TS ) 0.02 and nNu,TS ) 0.02 for

12086 J. Am. Chem. Soc., Vol. 119, No. 50, 1997
Berti et al.
Figure 7. Structures of reactant, transition state, and the hypothetical oxocarbenium ion, with independent, noninteracting nicotinamide and water
molecules. (a) Atoms present in the cutoff models used in BEBOVIB calculations are colored by element; all others are colored gray. Changes in
bond order between the reactant and the transition state are indicated. All atoms within two bonds of the isotopically labeled positions were included
in KIE calculations, yielding “highly proper” cutoff models.89 (b) Electrostatic potentials projected onto the molecular surfaces of the full molecules
in the same orientation as in part a, with red representing positiVe electrostatic potential and blue representing negatiVe potential. Calculations were
done at the RHF/6-31G** level. Molecular surfaces are at an electron density of 0.002 e/bohr3, and the electrostatic potential spectrum is from -0.1
(blue) to 0.1 hartree/e (red).
the cholera toxin reaction and nLG,TS ) 0.05, nNu,TS ) 0.001
for the pertussis toxin reaction. These values are similar to the
nLG,TS ) 0.02 and nNu,TS ) 0.03 for the DTA reaction. These
are small changes in the predicted transition state structures that
reflect refinement in the modeling procedure. In all cases, the
transition state structures for toxin-catalyzed hydrolysis of
NAD+ are highly oxocarbenium ion-like.
exponential functions of bond order; therefore, this difference
of 0.54 Å in bond length to the water nucleophile represents a
real difference in the transition state structures. At the predicted
transition state for the DTA reaction, the predicted KIEs are
within the 95% confidence interval of all measured KIEs except
the 1′-3H. The small difference of 1.200 and 1.186 between
experimental and predicted KIEs at this position are easily
explained by the well-documented difficulty in predicting
R-secondary KIEs at this position.59,76
The DTA transition state structure is also similar to the
transition state for the pH-independent solvolytic hydrolysis of
NAD+, where nLG,TS ) 0.02 and nNu,TS ) 0.005. The difference
in nNu, 0.025, between the enzymatic and solvolytic transition
states represents a 6-fold difference in bond order. KIEs are
Catalytic Mechanism and the Structure of DTA. In
considering the catalytic mechanism of DTA, it is worth noting
(76) Westaway, K. C. Tetrahedron Lett. 1975, 48, 4229-4232.

Transition State of DTA-Catalyzed NAD+ Hydrolysis
J. Am. Chem. Soc., Vol. 119, No. 50, 1997 12087
that the toxin catalyzes NAD+ hydrolysis much more weakly
than the ADP-ribosylation reaction. With kcat ) 3.0 (( 0.3) ×
10-3 s-1, compared with k ) 5 × 10-7 s-1 for the uncatalyzed
reaction at 37 °C,77 the rate enhancement is only 104-fold. The
biologically relevant ADP-ribosylation of EF-2 exhibits a kcat
) 20 s-1, a 4 × 107-fold rate enhancement.48 Much of the
catalytic power brought to bear during the ADP-ribosylation
reaction is evidently not operative in the hydrolytic reaction.
NAD+ Binding. The X-ray crystallographic structures of
diphtheria toxin include ones with an endogenous inhibitor,
ApUp, bound in the active site,78 with substrate NAD+ 79 and
with the active site unoccupied.80 The nicotinamide ring of
substrate NAD+ binds in a deep hydrophobic pocket on the
enzyme, becoming desolvated. The low dielectric environment
lowers the energy difference between the reactant and transition
states by increasing the energy of the positively charged
nicotinamide ring in the reactant relative to the neutral form at
the transition state, a case of ground state destabilization. This
deep pocket would make release of the product nicotinamide
difficult before the ADP-ribosylation reaction was complete,
consistent with the observed inversion of configuration of the
anomeric carbon in the reaction with EF-2.81 The side chain
of Glu148, which has been shown by photoaffinity labeling82
and mutagenesis83 to be crucial for catalysis, is located where
it can stabilize the formation of the oxocarbenium ion-like
transition state (see below). His21 is located near one of the
phosphate groups of NAD+, where it could assist electrostati-
cally in NAD+ binding. There are, however, some features of
the X-ray crystallographic structure with bound NAD+ that
appear unfavorable for catalysis.
“Anticatalytic” Structural Features. Several (apparently)
favorable interactions between DTA and bound NAD+ in the
crystal structure appear to be “anticatalytic”; that is, they appear
to stabilize the reactant in preference to the transition state. The
carboxamide nitrogen of the nicotinamide ring and one of the
R-phosphate nonbridging oxygens are positioned to form an
intramolecular hydrogen bond. This bond would need to be
broken during C1′-N1 bond cleavage, making it an anticatalytic
interaction. It would not be possible for the N-H‚‚‚OdP
hydrogen bond to be a “fulcrum” about which the nicotinamide
ring rotates as the C1′-N1 bond breaks because of the presence
of an additional hydrogen bond from the carboxamide nitrogen
to Gly22 CdO, which would need to be broken instead.
The closest contact the carboxylate of Glu148 makes with
NAD+ is through the C5 atom of the nicotinamide ring (3.04
Å) and O2′ of the ribose ring (3.39 Å). The nicotinamide ring
is positively charged in the reactant (Figure 7b) but becomes
electroneutral in the transition state. Interaction of the reactant
with a negatively charged carboxylate stabilizes the reactant
relative to the transition state. Thus, a favorable interaction
between Glu148 and the nicotinamide ring would be anticata-
lytic. In the crystal structure, the distance to C1′, the site of
accumulation of positive charge at the transition state, is 3.96
Å, and the interaction is at an oblique angle, not ideally suited
to forming a strong interaction (see below).
The presence of these “anticatalytic” features in the DTA/
NAD+ crystal structure implies (a) that these favorable interac-
tions must be broken, at some energetic cost, to attain the
transition state and/or (b) that, although the general mode and
orientation of binding of NAD+ are the same as in the productive
Michaelis complex, some of the details are different.
Binding of the Transition State. Docking the transition state
structure with the nicotinamide ring superimposed on that of
the crystallographically determined NAD+ in the active site cleft
results in the ribosyl ring projecting into solution, with no
contacts to DTA. While the sides of the nicotinamide binding
pocket are well defined by the side chains of Tyr54 and Tyr65,
the top of the pocket is less well defined, with contacts
contributed by Tyr20, Gly22, Ser55, Thr56, Ala62, and Phe140.
Only minor changes in protein conformation are necessary to
accommodate binding the nicotinamide ring deeper in the
pocket. In this way, the positively charged ribosyl ring at the
transition state is in contact with the stabilizing negative charge
of Glu148.
The transition state structure could be fitted into the active
site of DTA, without modifying the protein structure, by
changing the angle between the nicotinamide ring and the ribose
ring and by varying the torsional angles about the C4′-C5′ and
C5′-O5′ bonds. Torsional rotations have low energy and have
negligible effects on KIEs. Similarly, because of the low bond
order to the leaving group, nLG,TS ) 0.02, a change in the angle
between the nicotinamide ring and the ribosyl moiety also would
have a negligible effect on the KIEs. The nicotinamide ring
was positioned as deeply as possible in the binding pocket. The
ribosyl portion was positioned with rC1′-N1 ) 2.65 Å, with C1′
near the Glu148 carboxylate and with the position of the
phosphate group adjusted to avoid steric clashes while remaining
near the His21 side chain.
This docking experiment is a crude model of the enzyme/
transition state interactions since no attempt was made to account
for the unknown changes in protein structure that occur to form
the transition state with the physiological second substrate, EF-
2. However, with the transition state structure fitted in the active
site and the electrostatic potential of the enzyme projected onto
its solvent accessible surface, the electrostatic stabilization of
the transition state becomes clear (Figure 8). The negative (blue
dots) electrostatic potential of Glu148 is juxtaposed directly with
the positive (solid red) electrostatic potential of the transition
state at C1′. The positive electrostatic potential of His21 (red
dots) complements the negative potential of the transition state’s
phosphate group (solid blue). With this orientation of the ribose
ring, a perpendicular approach of the nucleophilic water brings
it into close contact with Glu148. In this geometry, Glu148
can act as a general base catalyst or can electrostatically polarize
the H-O bond of water to promote nucleophilic attack. Also,
with this orientation, it becomes possible to form a C-H‚‚‚O
hydrogen bond between H1′ of NAD+ and a carboxylate oxygen
of Glu148. C-H‚‚‚O hydrogen bonds are weak, but biologically
significant, interactions.84,85
These features of the catalytic site suggest strategies for
designing transition state inhibitors of diphtheria toxin. The
transition state structure of the ribosyl group is similar to that
for the hydrolysis of inosine catalyzed by nucleoside hydrolase
from Crithidia fasiculata,55 with a highly dissociative, oxocar-
benium ion-like transition state. For that enzyme, the electro-
static and geometric characters of the transition state were
exploited to design and synthesize transition state analogue
(77) Johnson, R. W.; Marschner, T. M.; Oppenheimer, N. J. J. Am. Chem.
Soc. 1988, 110, 2257-2263.
(78) Choe, S.; Bennett, M. J.; Fujii, G.; Curmi, P. M.; Kantardjieff, K.
A.; Collier, R. J.; Eisenberg, D. Nature 1992, 357, 216-222.
(79) Bell, C. E.; Eisenberg, D. Biochemistry 1996, 35, 1137-1149.
(80) Bell, C. E.; Eisenberg, D. Biochemistry 1997, 36, 481-488.
(81) Oppenheimer, N. J.; Bodley, J. W. J. Biol. Chem. 1981, 256, 8579-
8581.
(82) Carroll, S. F.; McCloskey, J. A.; Crain, P. F.; Oppenheimer, N. J.;
Marschner, T. M.; Collier, R. J. Proc. Natl. Acad. Sci. U.S.A. 1985, 82,
7237-7241.
(83) Wilson, B. A.; Reich, K. A.; Weinstein, B. R.; Collier, R. J.
Biochemistry 1990, 29, 8643-8651.
(84) Desiraju, G. R. Acc. Chem. Res. 1996, 29, 441-449.
(85) Desiraju, G. R. Acc. Chem. Res. 1991, 24, 290-296.

12088 J. Am. Chem. Soc., Vol. 119, No. 50, 1997
Berti et al.
Figure 8. Stereodiagram of the transition state structure fitted into active site of DTA. Projection of the electrostatic potentials onto the molecular
surface shows the complementarity of the contact site for the transition state. The ball-and-stick figure of the transition state on the lower right
shows the orientation of the transition state in the active site. The ring oxygen, O4′, is labeled on the molecular surface. The region of positive
potential (solid red) associated with C1′ of the TS is apposed with the negative potential (blue dots) of the carboxylate of Glu148. The negative
electrostatic potential of the phosphate oxygens (solid blue) is near the positive potential (red dots) of His21 (not visible behind the phosphate). The
proximity of the nucleophile water to the carboxylate of Glu148 raises the possibility that it promotes catalysis as a general base catalyst or by
polarizing the H-O bond to enhance the nucleophilicity of the oxygen. The electrostatic potentials were calculated using the Delphi module of the
program Insight II (Biosym Technologies, San Diego, CA). Charges were of 1+ were assigned to Lys, Arg, and His21 of DTA, 0.5+ to other His
residues, and 1- to Glu and Asp. Point charges on the atoms of the transition state structure were assigned from the natural population analysis90
charges calculated from the wave function as in Figure 7. The molecular surfaces were calculated as the Connolly surfaces,91 but with the atomic
radii reduced by a factor of 0.8 so that the surface approximates a smoothed van der Waals surface. The contact site of DTA includes all residues
that contribute to the molecular surface surrounding the transition state: Tyr20, His21, Tyr54, Ser55, Thr56, Tyr65, Phe140, Glu148. Secondary
structural elements of DTA, as calculated by the Kabsch and Sander criteria,92 are shown, with R-helices in purple and â-strands in blue. The
transition state structure from Figure 7 was fitted into the active site cleft by allowing the bond angles between the nicotinamide and ribosyl
moieties to vary, as well as the torsional angles about C4′-C5′ and C5′-O5′.
inhibitors with Ki’s to 2 nM.86-88 Preliminary studies show
that phenyliminoribitol and p-nitrophenyl amidrazone are modest
inhibitors of DTA, with Ki’s near 100 µM (data not shown).
Given the extremely high specificity of DTA for the nicotina-
mide ring and adenosyl portion of the molecule, work is
underway to incorporate these features into inhibitor structures.
Work is also underway to characterize the transition state
structure for the physiological reaction catalyzed by diphtheria
toxin, the ADP-ribosylation of EF-2.
and nucleophile participating in the reaction coordinate. The
4′-3H and 5′-3H KIEs indicate the use of binding energy remote
from the scissile bond to promote catalysis, possibly through a
change in the electrostatic environment of the R-phosphate and/
or distortion of the C4′-C5′-O5′ bond angle. KIEs were
measured by the competitive radiolabel technique, with the [4′-
18O] KIE also measured by whole-molecule positive ion mass
spectrometry.
The transition state structure docked into the crystal structure
of DTA suggests features of the catalytic mechanism. Desol-
vation of the nicotinamide ring in a hydrophobic pocket,
electrostatic stabilization of the positive charge on the oxocar-
benium ion-like transition state by the carboxylate group of
Glu148, electrostatic stabilization of the pyrophosphoryl group
by His21, and activation of the attacking water nucleophile can
be proposed from the complex of DTA with the experimentally
determined transition state structure.
Conclusion
The transition state structure of DTA-catalyzed hydrolysis
of NAD+ was determined by using the structure interpolation
method of bond-energy/bond-order vibrational analysis of
multiple KIEs. The transition state is oxocarbenium ion-like,
with low residual bond order to the leaving group, nC1′-N1
)
0.02, and low bond order to the approaching nucleophilic water,
nC1′-O ) 0.03. The reaction mechanism is ANDN (i.e.,
concerted) and highly asynchronous, with both the leaving group
Acknowledgment. The authors thank Mr. Edward Nieves
and Dr. Ruth H. Angeletti for their invaluable help in developing
the method of mass spectrometric determination of KIEs. We
also thank Dr. R. John Collier for supporting the production of
DTA and for his cooperation and encouragement for this study.
Dr. Michael Sinnott generously provided the ketone precursor
used in the synthesis of [5-18O]glucose.
(86) Horenstein, B. A.; Schramm, V. L. Biochemistry 1993, 32, 7089-
7097.
(87) Horenstein, B. A.; Zabinski, R. F.; Schramm, V. L. Tetrahedron
Lett. 1993, 34, 7213-7216.
(88) Horenstein, B. A.; Schramm, V. L. Biochemistry 1993, 32, 9917-
9925.
(89) Sims, L. B.; Lewis, D. E. In Isotope effects: Recent deVelopments
in theory and experiment; Buncel, E., Lee, C. C., Eds.; Elsevier: New York,
1984; Vol. 6, pp 161-259.
Supporting Information Available: Summary of KIE
calculations for the DTA-catalyzed hydrolysis transition state,
plus the BEBOVIB input (9 pages). See any current masthead
page for ordering and Internet access instructions
(90) Reed, A. E.; Curtiss, L. A.; Weinhold, F. Chem. ReV. 1988, 88,
899-926.
(91) Connolly, M. L. Science 1983, 221, 709-713.
(92) Kabsch, W.; Sander, C. Biopolymers 1983, 22, 2577-2637.
JA971317A