Transition state analysis of adenosine nucleosidase from yellow lupin (Lupinus luteus)

Article abstract of DOI:10.1016/j.phytochem.2005.10.006

The transition state of adenosine nucleosidase (EC 3.2.2.7) isolated from yellow lupin (Lupinus luteus) was determined based upon a series of heavy atom kinetic isotope effects. Adenosine labeled with ¹³C, ²H, and ¹⁵N was

Full text of DOI:10.1016/j.phytochem.2005.10.006

PHYTOCHEMISTRY
Phytochemistry 67 (2006) 5–12
www.elsevier.com/locate/phytochem
Transition state analysis of adenosine nucleosidase from yellow
lupin (Lupinus luteus)
*
Carl Bates, Zachariah Kendrick, Nancy McDonald, Paul C. Kline
Department of Chemistry, Middle Tennessee State University (MTSU), Box 68, Murfreesboro, TN 37132, United States
Received 2 May 2005; received in revised form 24 August 2005
Available online 21 November 2005
Abstract
The transition state of adenosine nucleosidase (EC 3.2.2.7) isolated from yellow lupin (Lupinus luteus) was determined based upon a
series of heavy atom kinetic isotope effects. Adenosine labeled with 13C, 2H, and 15N was analyzed by liquid chromatography/electrospray
mass spectrometry to determine kinetic isotope effects. Values of 1.024 0.004, 1.121 0.005, 1.093 0.004, 0.993 0.006, and
1.028 0.005 were found for [10-13C], [10-2H], [20-2H], [50-2H], and [9-15N] adenosine, respectively. Using a bond order bond energy vibra-
tional analysis, a transition state consisting of a significantly broken C–N bond, formation of an oxocarbenium ion in the ribose ring, a
conformation of C3-exo for the ribose ring, and protonation of the heterocyclic base was proposed. This transition state was found to be
very similar to the transition state for nucleoside hydrolase, another purine metabolizing enzyme, isolated from Crithidia fasciculata.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Keywords: Yellow lupin; Lupinus luteus; Kinetic isotope effects; Transition state; Adenosine nucleosidase
NH2
1. Introduction
N
N
NH2
H2O
H
Adenosine nucleosidase (EC 3.2.2.7) is an enzyme of the
purine salvage pathway and has been implicated in a num-
ber of physiological roles including controlling the level of
cytokinins, synthesis of caffeine, and synthesis of S-adeno-
syl-L-methionine (Chen and Kristopeit, 1981; Koshiishi
et al., 2001; Edwards, 1996). It is a hydrolytic enzyme that
catalyzes the irreversible hydrolysis of various nucleosides
to ribose and the corresponding base.
HO
H
N
N
N
H
O
HO
H
N
(H,OH)
H
H
+
O
H
N
H
H
H
N
OH
OH
H
H
OH
OH
2
1
3
Adenosine nucleosidase activity has been detected in a
number of plants with the enzyme level varying widely from
one species toanother (Leszczynska et al., 1984). The enzyme
has been isolated from a number of plant sources including
Jerusalem artichoke shoots, tomato roots and leaves, spin-
ach beet leaves, barley leaves, tea leaves, wheat germ, coffee
leaves, malted barley, and yellow lupin (Le FlocÕh and
Lafleuriel, 1981; Burch and Stuchbury, 1986; Poulton and
Butt, 1976; Guranowski and Schneider, 1977; Imagawa
et al., 1979; Chen and Kristopeit, 1981; Campos et al.,
2005; Lee and Pyler, 1986; Abusamhadneh et al., 2000).
Abbreviations: LC/MS, liquid chromatography/mass spectrometry;
KIE, kinetic isotope effect; HMDS, hexamethyldisilazine; Tris, tris(hydr-
oxymethyl)aminomethane; SIM, selective ion monitoring; BEBOVIB,
bond energy bond order vibrational analysis; RNA, ribonucleic acid;
NAD+, nicotinamide dinucleotide; AMP, adenosine monophosphate;
IU-NH, inosine–uridine nucleoside hydrolase; IAG-NH, inosine–adeno-
sine–guanosine nucleoside hydrolase.
*
Corresponding author. Tel.: +1 615 898 5466; fax: +1 615 898 5182.
E-mail address: pkline@mtsu.edu (P.C. Kline).
0031-9422/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.phytochem.2005.10.006

6
C. Bates et al. / Phytochemistry 67 (2006) 5–12
Table 1
Adenosine nucleosidase from yellow lupin is a dimer
Kinetic isotope effects for adenosine nucleosidase from yellow lupin
Substrate
Isotope and type of effecta Kinetic isotope effectb
with a molecular weight of 72,000 (Abusamhadneh et al.,
2000). The enzyme displays broad substrate specificity
exhibiting activity with a variety of purine nucleosides
including adenosine 1, guanosine, inosine, 2-deoxyadeno-
sine, and 5-deoxyadenosine. In contrast to other plant pur-
ine hydrolases for which the pH optimum has been
reported to range from 4.0 to 6.5, the pH optimum for
the reaction catalyzed by yellow lupin adenosine nucleosi-
dase is 7.5 (Burch and Stuchbury, 1986; Imagawa et al.,
1979; Abusamhadneh et al., 2000).
[10-13C]adenosine 1 Primary
1.024 0.004
1.121 0.005
1.093 0.004
0.993 0.006
1.028 0.005
[10-2H]adenosine 1 a-Secondary
[20-2H]adenosine 1 b-Secondary
[50-2H]adenosine 1 d-Secondary
[9-2N]adenosine 1
Primary
a
A primary isotope effect has the label in the bond being broken. A
secondary isotope effect indicates the label is one (a), two (b), or four (d)
bonds away from the breaking bond.
b
Each isotope effect was the minimum of 12 individual determinations.
The kinetic isotope effects were calculated using Eq. 2 in Section 3.
Purine hydrolases, such as adenosine nucleosidase, have
been isolated from a number of other sources in addition to
plants including protozoan parasites, yeast, and Esche-
observed for [50-2H]adenosine. These KIEs were used to
determine the transition state of the enzyme-catalyzed pur-
ine hydrolysis reaction carried out by adenosine nucleosi-
dase. The isotope effects determined for this reaction are
listed in Table 1. A transition state based upon these KIEs
is proposed (Fig. 1B). The proposed transition state gave
the best agreement between the observed KIEs and a set
of predicted KIEs using a bond-order bond-energy analy-
sis. A cutoff model was used in calculating the transition
state using BEBOVIB. The atoms used in this model were
those in the ribosyl ring and imidazole ring but eliminated
the hydroxyl hydrogens along with the 3-hydroxyl group.
This model is similar to that used to calculate the transition
state for other purine hydrolase and is based on the criteria
of Stern and Wolfsberg (1966).
´
richia coli (Parkin et al., 1991a; Shi et al., 1999; Versees
et al., 2001; Pelle et al., 1998; Parkin, 1996; Miller et al.,
1984; Kurtz et al., 2002; Petersen and Moller, 2001). The
presence of these purine salvage enzymes has not been
detected in mammals, which rely on a different set of
enzymes to supply their nucleosides (Gopaul et al., 1996).
Nucleoside hydrolase (EC 3.2.2.1) isolated from Cri-
thidia fasciculata, an insect parasite, is the most extensively
studied of these hydrolytic enzymes. The enzyme has been
purified to homogeneity and the substrate specificity, pH
profile, and kinetic mechanism determined (Parkin et al.,
1991a). In addition the catalytically important amino acids
have been determined by site-directed mutagenesis and X-
ray crystallography (Gopaul et al., 1996; Degano et al.,
1996). Finally, the transition state has been determined
using heavy atom kinetic isotope effects, and a new class
of C-nucleoside analogs have been designed based upon
that transition state (Horenstein et al., 1991, 1993).
The transition state structure provides insight into the
multiple interactions that take place between the substrate
and enzyme in an enzyme-catalyzed reaction. In addition,
knowledge of the transition state allows the possibility of
the design of highly specific inhibitors (Schramm et al.,
1994). These inhibitors have the potential for a number
of uses including controlling the growth of a particular
plant or the analysis of a metabolic pathway. This study
presents the transition state of adenosine nucleosidase, a
nucleoside hydrolase isolated from yellow lupin, and com-
pares it to the transition state of the C. fasciculata enzyme.
The [10-13C]adenosine 1 kinetic isotope effect was
1.024 0.004. This primary KIE, along with the [9-15N]
KIE, reports on the extent to which the C1–N9 glycosidic
bond is broken in purine nucleosides. Based upon the BEB-
OVIB calculations, the bond order between C1–N9
required to match this isotope effect is reduced from 1.09
to 0.25. The bond orders reported in this paper are calcu-
lated using PaulingÕs equation relating bond order to bond
length
ijÞ=0:3
nij ¼ eðr ꢀr
;
ð1Þ
1
where nij is the Pauling bond order between the two atoms,
˚
r1 is the standard single bond length in A, and rij is the
˚
length in A between the two atoms. Using PaulingÕs bond
order equation this corresponds to an increase in bond
˚
length from 1.47 to 1.92 A for the C1–N9 bond.
The observed [10-13C]KIE is consistent with the forma-
tion of an oxocarbenium ion in the transition state similar
to that seen for nucleoside hydrolase from C. fasciculata
(IU-NH) (Horenstein et al., 1991). The [10-14C]KIE for
IU-NH is 1.044 0.004. To directly compare the two iso-
tope effects requires the use of the Swain–Schaad relation-
ship of 14C = 13CKIE1.9 (Melander and Saunders, 1980;
Swain et al., 1958). Based on this relationship the
[10-13C]KIE for yellow lupin adenosine nucleosidase is
equivalent to the [10-14C]KIE observed for IU-NH.
2. Results and discussion
2.1. Kinetic isotope effects
A family of heavy-atom kinetic isotope effects (KIE) for
adenosine nucleosidase from yellow lupin were determined
using [9-15N], [10-13C], [10-2H], [20-2H], and [50-2H]adeno-
sine 1 as substrates. These sites have been shown to be
important in the determination of the transition state for
other purine hydrolases. Substantial kinetic isotope effects
were observed for [9-15N], [10-13C], [10-2H] and [20-2H]aden-
osine 1, while no significant kinetic isotope effect was
The [20-2H]adenosine 1 KIE is 1.093 0.004. This b sec-
ondary effect is indicative of two important aspects of the
enzyme-catalyzed reaction. The first is the nature of the

C. Bates et al. / Phytochemistry 67 (2006) 5–12
7
H
NH2
C6
NH2
N7
C5
C4
C6
N7
N
H8 C8
N
C5
H8 C8
H5'
O5'
N9
C4
H5'
N3
O5'
N9
C5'
N3
H3'
C3'
C5'
1.47 Å
1.92 Å
H5"
O4'
1.40 Å
O4'
1.35 Å
H2'
C1'
H '
2 C1'
C4'
H4'
H5"
C4'
H4'
H3'
C2'
C2'
H1'
O3'
H1'
> 2.40 Å
C3'
O3'
O2'
O2'
nucleophile
B. Transition State
A. Ground State
Fig. 1. Ground and transition states for adenosine nucleosidase (the hydroxyl hydrogens are omitted for clarity). (A) The ground state of adenosine was
˚
based upon the crystal structure reported by Lai and Marsh followed by energy minimization using Spartan 5 and Gaussian 98W. The bond lengths (in A)
are shown for the most important points. The conformation of the ribose ring is C3-endo. (B) The transition state of adenosine 1 gave the best match
between the observed and BEBOVIB calculated kinetic isotope effects. The main features of the transition state are a substantially, but not completely
broken C1–N9 bond, very low bond order to the attacking nucleophile, a change in ribose conformation from C3-endo to C3-exo, a decrease in C1–O4
bond length as the oxocarbenium ion forms, and protonation of the heterocyclic base at N7.
transition state. This KIE is consistent with the formation
of an oxocarbenium ion predicted by the [10-13C]KIE. The
[20-2H]KIE also reports on the ring conformation. With
the increasing double bond nature of the C1–O4 bond,
the ribose ring may assume two possible conformations:
C3-exo or C3-endo (Horenstein et al., 1991; Berti and
Tanake, 2002). In the C3-exo conformation the dihedral
angle between C2–H2 and the empty p orbital is approxi-
mately 10ꢁ allowing hyperconjugation. in the C3-endo con-
formation the dihedral angle is much larger, reducing
hyperconjugation. Matching the large b secondary KIE
reported for this enzyme required that the sugar assume a
conformation in the transition state in which hyperconju-
gation is possible, thus requiring a shift from the C3-endo
conformation in the ground state to C3-exo conformation.
The large [20-2H]KIE seen for this enzyme is similar to the
[20-2H]KIEs reported for other purine hydrolases including
IU-NH. As noted by Berti all of the reported 2 KIEs for
this class of enzyme (with one exception) are greater than
1.1 for tritium KIEs and 1.065 for deuterium KIEs (Berti
and Tanake, 2002). The single exception noted by Berti
and Tanaka was for the RNA hydrolysis catalyzed by ricin
A-chain toxin in which the [20-2H]KIE was reported to be
1.012 0.005 (Chen et al., 2000). This small KIE was inter-
preted as indicating the ribose ring in the transition state
had assumed a C3-endo conformation.
(Berti and Tanake, 2002). The large effect is due to the
increased freedom around C1 as it undergoes rehybridiza-
tion in the formation of the oxocarbenium ion (Poirier
et al., 1994; Koerner et al., 2000; Matsson and Westaway,
1998).
Of all the KIEs measured in this study, the [10-2H] KIE
is the most difficult to interpret quantitatively. This KIE is
extremely sensitive to small changes in the C1–H1 bond
order. As pointed out by Berti, because of the closeness
of the oxocarbenium ion to H1 there are likely to be
non-covalent interactions involved in the stabilization of
the transition state that would affect the magnitude of
[10-2H]KIE.
With a [50-2H]adenosine 1 KIE of 0.993 0.005 for yel-
low lupin adenosine nucleosidase, there is no significant
KIE at this remote location. This is in contrast to the
results for other purine hydrolases such as IU-NH from
C. fasciculata (EC 3.2.2.1), NAD+ hydrolysis by diphtheria
toxin A-chain, and AMP nucleosidase from Azotobacter
vinelandii (EC 3.2.2.4), all of which exhibited significant
KIEs at the 5 position (Horenstein et al., 1991; Berti
et al., 1997; Parkin et al., 1991b). These results have been
interpreted to indicate that the hydroxyl group in the tran-
sition state is placed in such a position as to stabilize the
positive charge on the oxocarbenium ion. Support for this
interpretation was observed in the X-ray structure of IU-
NH co-crystallized with p-aminophenyliminoribitol in
which the hydroxymethyl group is positioned above the
positive charge on the 5-membered ring. In contrast the 5
KIE for adenosine nucleosidase was similar to the
[50-3H]KIE for the acid-catalyzed hydrolysis of AMP of
1.006 0.005 and the solvolysis of NAD+ of
1.000 0.003 (Mentch et al., 1987; Berti and Schramm,
1997). The lack of a KIE at the 5 position indicates that
in the reaction catalyzed by adenosine nucleosidase the
The [10-2H]adenosine 1 KIE for adenosine nucleosidase
is 1.121 0.005. The large a secondary [10-2H]KIE has also
been interpreted to be consistent with the formation of an
oxocarbenium ion in which the C–N bond is significantly
broken, but there is little bond order to the attacking nucle-
ophile (Poirier et al., 1994; Koerner et al., 2000; Matsson
and Westaway, 1998). The observed KIE for this enzyme
fits the observed KIEs for other purine hydrolases where
2
the H-KIE has been observed to range from 1.08 to 1.17

8
C. Bates et al. / Phytochemistry 67 (2006) 5–12
hydroxymethyl group plays no role. This interpretation is
also consistent with the substrate specificity of the enzyme.
5-Deoxyadenosine is a better substrate for the enzyme than
is adenosine indicating the hydroxymethyl group does not
take part in stabilizing the transition state (Abusamhadneh
et al., 2000).
mediate is stable enough for the leaving group to diffuse
away, or AN in which the intermediate is less stable
and there is not enough time for the leaving group to dif-
fuse away from the reaction site.
Several lines of evidence point to this reaction being an
ANDN reaction. The primary [10-13C]adenosine 1 KIE has
been used to distinguish the class of mechanism. ANDN
reactions yield significant [10-13C]adenosine 1 KIEs of
1.03–1.08 while the isotope effects are significantly smaller,
D
*
N
The [9-15N]adenosine
1
KIE is substantial at
1.028 0.005. This isotope effect along with the [10-13C]
isotope effect reflects the extent to which the C1–N9 glyco-
sidic bond is broken in the transition state. The magnitude
of this KIE indicates the C1–N9 bond is substantially but
not completely broken. A completely broken C1–N9 bond
would yield a [9-15N]adenosine 1 KIE of 1.04. To match
the observed isotope effect the C1–N9 bond order was
reduced to approximately 0.25 which is similar to the result
predicted by the [10-13C]adenosine 1 KIE. The extent to
which the C1–N9 bond is broken in the transition state
of adenosine nucleosidase is similar to the 0.225 0.095
bond order reported for IU-NH.
1.00–1.01, for the AN
D mechanism (Berti and Tanake,
N
*
2002). The observed [10-13C]adenosine 1 KIE for adenosine
nucleosidase is 1.024 0.004 consistent with an ANDN
mechanism.
Results from the BEBOVIB modeling also indicate the
reaction is bimolecular consistent with an ANDN mecha-
nism. While there is extremely low bond order between
the attacking nucleophile and the substrate, it was neces-
sary to include an interaction between the substrate and
the nucleophile to match the observed KIEs. Therefore,
while the reaction is not synchronized, it is still an ANDN
reaction, although a highly dissociative one. This mecha-
nism is the same as found for a number of N-riboside
hydrolases including E. coli nucleoside hydrolase, C. fascic-
ulata nucleoside hydrolase, AMP nucleosidase from A. vin-
elandii, and for the hydrolysis of NAD+ by diphtheria
toxin among others (Horenstein et al., 1991; Berti et al.,
1997; Mentch et al., 1987).
This kinetic isotope effect also reports on electronic rear-
rangements that may take place in the purine base in the
transition state. Matching the observed [9-15N]adenosine
1 KIE required the activation of the purine base by proton-
ation of N7. Protonation turns the relatively poor leaving
group of the heterocyclic base into a good leaving group.
In addition protonation aids in the breaking of the C1–
N9 bond by causing a decrease in the electron density
within the purine ring. Protonation of N7 is also consistent
with the mechanism proposed for the acid-catalyzed hydro-
lysis of purine nucleosides (Garrett and Mehta, 1972). Fur-
ther evidence of the requirement for N7 protonation may
be found in the substrate specificity of the enzyme (Abu-
samhadneh et al., 2000). Tubercidin (7-deazaadenosine),
a nucleoside analog in which N7 has been replaced with
a carbon, is not a substrate for this enzyme.
The majority of protozoan nucleoside hydrolases belong
to one of three groups based upon substrate specificity (Ver-
´
sees and Steyeart, 2003). One group consists of the base-spe-
cific E. coli IU-NH, while the second group consists of
´
purine-specific nucleoside hydrolases (Estupinan and Sch-
˜
ramm, 1994), and the third group consists of 6-oxopurine
specific nucleoside hydrolases. Of these three groups, aden-
osine nucleosidase from yellow lupin fits most closely with
the second group based on base specificity. Other examples
of this group include the nucleoside hydrolases from Try-
2.2. Transition state
´
panosoma vivax and Trypanosoma brucei brucei (Versees
Combining the results from the family of kinetic isotope
effects results in a transition state whose main features are:
(a) a highly dissociative transition state in which the C–N
glycosidic bond is extensively broken while there is yet little
bond formation between C1 and the attacking nucleophile;
(b) formation of an oxocarbenium ion between C1 and 04;
(c) a change in the ribose ring conformation from C3-endo
in the ground state to C3-exo in the transition state; and (d)
protonation of N7 of the heterocyclic base. This transition
state is similar to the reaction catalyzed by other purine
hydrolases.
et al., 2001; Parkin, 1996). However, further examination
indicates significant differences in substrate specificity of
these nucleoside hydrolases and yellow lupin adenosine
nucleosidase. The nucleoside hydrolases characterized to
date use ribonucleosides as substrates, while the 2, 3, or 5
deoxyribonucleosides are not substrates or are extremely
poor substrates. For example in the IAG-NH (EC 3.2.2.1)
from T. vivax the kcat/Km is reduced by 1700, 2.3 · 105,
and 7100 for 2-deoxyadenosine, 3-deoxyadenosine, and 5-
deoxyadenosine, respectively, compared to the kcat/Km for
adenosine. for IAG-NH from T. vivax all three hydroxy
´
In IUPAC nomenclature nucleophilic reactions can be
described as either ANDN or AN + DN where AN is a nucle-
ophilic addition and DN is a nucleophilic displacement
(Guthrie and Jencks, 1989). ANDN reactions are bimolecu-
lar SN2 reactions while the AN + DN reactions are unimo-
lecular SN1-type reactions. The SN1 reaction can be further
groups take part in the reaction (Versees et al., 2002). In
contrast adenosine nucleosidase uses both 2-deoxyadeno-
sine and 5-deoxyadenosine as substrates at a level compara-
ble to that of adenosine 1. In fact, 5-deoxyadenosine was the
best substrate tested. The nucleoside hydrolases character-
ized to date are metalloproteins with a unique topology
(Murzin et al., 1995; Orengo et al., 1997). In IU-NH, a cal-
cium ion in the active site is believed to coordinate to the 2
classified as either AN + DN or AN
D
in which
N
*
AN + DN describes a reaction in which the cationic inter-

C. Bates et al. / Phytochemistry 67 (2006) 5–12
9
and 5 hydroxyl group of the substrate and to the attacking
3.2. Synthesis of [9-15N]adenine (Sethi et al., 1982)
nucleophile (Degano et al., 1998). IU-NH has multiple con-
tacts between the enzyme and the ring deriving most of its
catalytic efficiency by distortion of the ribosyl ring (Degano
et al., 1996). A number of carboxylate residues hydrogen
bond to the ribose ring causing a distortion, which allows
the oxocarbenium ion to form. The question of how yellow
lupin adenosine nucleosidase forces the ribose ring into the
transition state conformation in the absence of the 2-OH
group remains to be answered.
Mazzella et al. (1996) have identified three ways in
which the purine hydrolases reach the transition state.
They are activation of the leaving group by protonation
of the purine base, stabilization of the oxocarbenium ion
by interactions between the enzyme and ribose of the sub-
strate, and ionization of the 2-hydroxyl group to stabilize
the oxocarbenium ion. Enzymes may rely on one of these
effects or use some combination of the above. For example
of the 17.7 kcal/mol decrease in the energy of activation for
the reaction catalyzed by IU-NH, the relative contributions
of protonation/activation of the leaving group and stabil-
ization of the oxocarbenium ion are 4.6 and 13.1 kcal/
mol, respectively (Gopaul et al., 1996).
It seems likely that protonation of the leaving group
provides the major contribution to the decrease in the
energy of activation for adenosine nucleosidase. This is
based upon substrate specificity since 2-deoxyadenosine is
a substrate for adenosine nucleosidase, but it is not for
IU-NH.
Based on a family of kinetic isotope effects a transition
state was proposed for the hydrolysis of purine nucleosides
by adenosine nucleosidase. The main features of the transi-
tion state are: (a) a highly dissociative transition state in
which the C–N glycosidic bond is extensively broken while
there is yet little bond formation between C1 and the attack-
ing nucleophile; (b) formation of an oxocarbenium ion
between C1 and O4; (c) a change in the ribose ring conforma-
tion from C3-endo in the ground state to C3-exo the transi-
tion state; and (d) protonation of N7 of the heterocyclic
base. Further work is being carried out on the structure of
the enzyme in an attempt to provide information on how
the transition state of the 2 deoxyribonucleosides is enforced.
5-Amino-4,6-dichloropyrimidine (1.75 g; 10.7 mmol)
was suspended in 35 mL absolute EtOH in a thick-wall
glass reaction tube. The suspension was cooled in an
EtOH-dry ice bath and [15N] ammonia (0.7 g; 38.8 mmol)
was bubbled through the solution. The reaction tube was
sealed and heated to 140 ꢁC in a Parr reaction bomb for
4 h. The reaction mixture was then allowed to cool to room
temperature, the solvent was removed under reduced pres-
sure, and the solid was recrystallized from EtOH (yield:
1260 mg; 8.7 mmol; 81%). The product, [4-15N, 5-14N]dia-
mino-6-chloropyrimidine, was authenticated by comparing
the melting point, ultraviolet spectrum, and retention time
on reverse-phase HPLC to an authentic sample of 4,5-dia-
mino-6-chloropyrimidine. The recrystallized product was
dissolved in 10 mL of 1:1 acetic anhydride-triethyl ortho-
formate and the solution refluxed for 3 h. The reaction
mixture was cooled and the solvent removed under reduced
pressure. The residue was dissolved in MeOH–H2O (1:1)
and purified by reversed-phase chromatography on an Isco
CombiFlash Graduate chromatography system equipped
with a Redi-Sep reversed-phase column. The column was
eluted with MeOH–H2O (1:1). Fractions containing
[9-15N]6-chloropurine were identified based upon their
ultraviolet spectrum and were pooled. The solvent was
removed under reduced pressure and dissolved in 15 mL
of anhydrous EtOH that was saturated with NH3. The
reaction mixture was sealed in a heavy-wall glass reaction
tube and heated to 150 ꢁC for 20 h. After the reaction mix-
ture had cooled, the solvent was evaporated under reduced
pressure and the residue recrystallized from EtOH. The
product [9-15N]adenine 2 was authenticated by comparing
the ultraviolet spectrum and behavior on a C18 reversed-
phase column (4.6 · 150 mm) eluted with 10 mM ammo-
nium acetate, pH 5.2, 2% methanol with that of an authen-
tic sample of adenine 2. The extent of isotopic enrichment
was determined by mass spectrometry (yield: 558 mg;
4.1 mmol; 38%).
[9-15N]Adenine 2 (250 mg; 1.9 mmol) was suspended in
5 mL anhydrous pyridine and benzoyl chloride (840 mg;
6 mmol) was added. The reaction mixture was refluxed
for 2 h and the excess pyridine was removed by distillation.
The residue was suspended in saturated NaHCO3 solution
and CHCl3 was added. The crude product precipitated,
and the crystals were recovered by filtration and washed
with H2O. The product was recrystallized from EtOH to
yield N6-benzoyl [9-15N]adenine 2 (354 mg, 1.48 mmol;
78%). The product was authenticated by comparison to a
standard sample of N6-benzoyladenine (Bullock et al.,
1957).
3. Experimental
3.1. Materials
Adenosine nucleosidase was isolated from yellow lupin
(Lupinus luteus) according to the procedure of Abusamhad-
neh et al. (2000) to yield an enzyme with a specific activity
of 6.3 lmol/min mg (135-fold purification 5.6% yield).
[10-13C], [20-2H] and [50-2H]adenosine 1 were obtained from
Omicron Biochemicals. [15N]Ammonia (98 at.%) was pur-
chased from Cambridge Isotope Laboratories. 5-Amino-
4,6-dichloropyrimidine and [1-2H]ribose were obtained
from Aldrich Chemical Co.
3.3. Synthesis of 1-O-acetyl 2,3,5-tri-O-benzoyl-b-D-
[1-2H]ribofuranoside (Kline and Serianni, 1990)
D-[1-2H]ribose (15.1 g; 100 mmol) was converted to a
mixture of methyl a and b [1-2H] ribofuranoside using

10
C. Bates et al. / Phytochemistry 67 (2006) 5–12
the procedure of Barker and Fletcher (1961). [1-2H]ribose
was dissolved in 50 mL of anhydrous MeOH with the sol-
vent removed under reduced pressure. This step was
repeated two times to remove residual H2O. The resulting
syrup was dissolved in 300 mL anhydrous MeOH and
cooled to 4 ꢁC. Concentrated sulfuric acid (1.5 mL) was
added and the mixture stirred for 14 h at 4 ꢁC. The reaction
mixture was passed through a column of Amberlite IRA-68
resin (40 g resin/mL of acid) in the OHꢀ form eluted with
anhydrous MeOH. The a and b anomers were separated by
chromatography on a Dowex 1 · 2 (200–400) column in
the OHꢀ form eluted with water. Fractions were assayed
by phenol-H2SO4 test (Chaplin, 1994). The anomers were
identified by 13C NMR spectroscopy and those fractions
containing the b anomer were concentrated under reduced
pressure and crystallized by adding a small seed crystal of
methyl b-D-[1-2H]ribofuranoside to the syrup. The crystal-
line sugar was dried under vacuum at 50 ꢁC (yield: 10.7 g;
6.48 mmol; 65%).
sure leaving a clear syrup. The last trace of HMDS was
removed by vacuum distillation at 50 ꢁC. The appropriate
1-O-acetyl 2,3,5-tri-O-benzoyl-b-D-ribofuranoside (2.0 g;
4.0 mmol) was dissolved in anhydrous CH2Cl2 (7 mL)
and trimethylsilyl trifluoromethanesulfonate added
(1.3 mL; 7 mmol). The solution containing the dissolved
sugar was added to the reflux apparatus containing the sily-
lated base and the reaction mixture refluxed under argon
for 5 h. After the condensation reaction was complete,
the reaction mixture was cooled to room temperature and
diluted with CH2Cl2 (20 mL). The solution was extracted
with ice-cold saturated aqueous NaHCO3 solution and
the organic phase dried over anhydrous Na2SO4. The sol-
vent was removed under reduced pressure at 30 ꢁC. Flash
chromatography of the residue on an Isco CombiFlash
Graduate chromatography system equipped with a Redi-
Sep reversed-phase column was eluted with 90/10
CH2Cl2/MeOH. The protecting groups were removed by
dissolving the sample in MeOH saturated with ammonia
and incubating at 25 ꢁC for 24 h. The solvent was removed
under reduced pressure at 30 ꢁC and the residue dissolved
in distilled H2O. The aqueous solution was extracted three
times with CHCl3 and Et2O. The H2O was removed under
reduced pressure with the resulting syrup chromatographed
on a Dowex 1 · 2 column in the OHꢀ form eluted with
H2O–MeOH (7:3) followed by H2O–MeOH (2:3). Labeled
adenosine-containing fractions were pooled and concen-
trated to yield 643 mg, 2.4 mmol, 60%. The identity of
the product was confirmed by 1H NMR spectroscopic
analysis and the extent of isotopic enrichment a determined
by LC/MS.
Dried methyl b-D-[1-2H]ribofuranoside (4.9 g; 30 mmol)
was dissolved in 20 mL anhydrous pyridine and 25 mL
CHCl3. The reaction flask was cooled in an ice bath and
17.5 mL benzoyl chloride (150 mmol) added dropwise with
stirring. The reaction mixture was incubated for 12 h at
4 ꢁC and at the end of this time excess benzoyl chloride
was decomposed by the addition of 3 g of crushed ice.
The reaction mixture was stirred for 2 h at room tempera-
ture and the organic phase extracted with 300 mL CHCl3
two times. The CHCl3 layer was dried with anhydrous
Na2SO4.
The organic phase was evaporated under reduced pres-
sure and dissolved in AcOH (7.5 mL) and Ac2O
(16.9 mL). The solution was cooled, and concentrated
H2SO4 (2.4 mL) was added slowly. The reaction mixture
was stored overnight at 4 ꢁC and the reaction stopped by
the addition of 24 g of crushed ice. The reaction mixture
was stirred for 30 min and extracted three times with CHCl3
(300 mL). The combined CHCl3 solution was washed with
ice-cold saturated NaHCO3 with the organic phase dried
(anhydrous Na2SO4). The solvent was removed under
reduced pressure and product recrystallized from isopropyl
alcohol. The product was authenticated by comparison to a
standard sample (yield: 11.3 g; 22.5 mmol; 75%).
3.5. Measurement of kinetic isotope effects
Heavy atom kinetic isotope effects were measured using
a modification of the competitive method developed for
stable isotopes (Parkin, 1991; Hunt et al., 2005). A reaction
mixture consisting of 50% specifically labeled and 50%
unlabeled adenosine 1 (1 mM total concentration) in
10 mM Tris, pH 7.2, was divided into two portions. The
reaction was initiated by the addition of adenosine nucleos-
idase and incubated at 37 ꢁC. The progress of the reaction
was monitored by reversed-phase HPLC on a ODS column
eluted with 95%, 10 mM ammonium acetate, pH 5.2; 5%
methanol. One reaction mixture was allowed to continue
to between 20% and 30% completion and the other allowed
to continue to 100% completion as determined by HPLC.
The reactions were stopped by the addition of 100 lL of
1 M HCl.
3.4. Synthesis of labeled ribonucleosides (Kline and Serianni,
1990)
Labeled adenosines were synthesized using a modifica-
tion of the method of Vorbruggen (Kline and Serianni,
¨
1990; Vorbruggen et al., 1981). A reflux apparatus was
¨
The isotopic composition of the appropriate product,
either ribose or adenine, depending on the location of the
label, was determined by chromatography on a Hewlett
Packard liquid chromatograph interfaced to an atmo-
spheric pressure interface-electrospray mass spectrometer
(LC/MS). A sample volume of 5 lL was injected onto a
Hypersil ODS column (100 · 2.1 mm) using water as the
mobile phase. The mass spectrometer was operated in the
assembled with 4.1 mmol of the appropriate base (N6-ben-
zoyl adenine or N6-benzoyl [9-15N]adenine) in the reaction
flask. After the apparatus was purged with argon, 25 mL
hexamethyldisilazane (HMDS) was added followed by
0.5 mL anhydrous pyridine. The suspension was refluxed
under argon until the base completely dissolved. Excess
HMDS was removed by distillation at atmospheric pres-

C. Bates et al. / Phytochemistry 67 (2006) 5–12
11
positive ion mode with a gas temperature of 350 ꢁC, drying
Acknowledgement
gas flow rate of 13.0 L/min, a nebulizer pressure of 25 psig,
and a fragmentor voltage of 40. Mass spectra for labels in
the ribose moiety were collected at 173 and 174 amu under
selected ion monitoring (SIM) conditions. For labels in the
adenine 2 moiety, mass spectra were collected at 158 and
159 amu under SIM conditions. The isotopic composition
was determined by extracting the appropriate ions from
the mass chromatogram and integrating the resulting
peaks.
The kinetic isotope effects were calculated by measure-
ment of the relative rates of hydrolysis of the nucleoside
containing a light (natural abundance) or heavy atom in
a given position. As measurement of kinetic isotope effects
using stable isotopes uses concentrations of labeled sub-
strate comparable to that of unlabeled substrate, the equa-
tions used to calculate the kinetic isotope effects must be
modified from those used for radio-labeled substrates
(Berti et al., 1997). The kinetic isotope effects were calcu-
lated using the following formula:
This work was supported by an American Chemical
Society Petroleum Research Fund Type B Grant (#
33688-B4).
References
Abusamhadneh, E., McDonald, N., Kline, P.C., 2000. Isolation and
characterization of adenosine nucleosidase from yellow lupin (Lupinus
luteus). Plant Sci. 153, 25–32.
Barker, R., Fletcher Jr., H.G., 1961. 2,3,5-Tri-O-benzyl-D-ribosyl and -L-
arabinosyl bromides. J. Org. Chem. 26, 4605–4609.
Berti, P., Blanke, S.R., Schramm, V.L., 1997. Transition state structure for
the hydrolysis of NAD+ catalyzed by diphtheria toxin. J. Am. Chem.
Soc. 119, 12079–12088.
Berti, P.J., Schramm, V.L., 1997. Transition state structure of the
solvolytic hydrolysis of NAD+. J. Am. Chem. Soc. 119, 12069–12078.
Berti, P.J., Tanake, K., 2002. Transition state analysis using multiple
kinetic isotope effects: mechanisms of enzymatic and non-enzymatic
glycoside hydrolysis and transfer. In: Tidwell, T.T., Richard, J.P.
(Eds.), Advances in Physical Organic Chemistry, vol. 37. Academic
Press, San Diego, pp. 239–314.
Bullock, M.W., Hand, J., Stokstad, E., 1957. Reduction of 6-aroylamin-
opurines with lithium aluminum hydride. J. Org. Chem. 22, 568–569.
Burch, L.R., Stuchbury, T., 1986. Purification and properties of adenosine
nucleosidases from tomato (Lycopersicon esculentum) roots and leaves.
J. Plant Physiol. 125, 267–273.
Campos, A., Rijo-Johansen, M., Carneiro, M., Fevereiro, P., 2005.
Purification and characterization of adenosine nucleosidase from
Coffea arabica young leaves. Phytochemistry 66, 147–151.
Chaplin, M.F., 1994. Monosaccharides. In: Chaplin, M.F., Kennedy, J.F.
(Eds.), Carbohydrate Analysis: A Practical Approach. IRL Press, New
York, pp. 2–3.
ln½ð1 ꢀ f Þð1 þ rꢀ1Þ=ð1 þ riꢀ1Þꢁ
ln½ð1 ꢀ f Þð1 þ0r0Þ=ð1 þ riÞꢁ
kinetic isotope effect ¼
;
ð2Þ
where f is the fraction of substrate hydrolyzed as deter-
mined by HPLC and ri and r0 are the ratio of peak inten-
sities, after correction for natural abundance, of
unlabeled/labeled products at the times the reaction was
stopped and started, respectively. The family of kinetic iso-
tope effects was used to determine the transition state as de-
scribed below.
Chen, C.M., Kristopeit, S.M., 1981. Metabolism of cytokinin: deribosyl-
ation of cytokinin ribonucleosides by adenosine nucleosidase from
wheat germ. Plant Physiol. 68, 1020–1023.
3.6. Transition-state modeling
Chen, X-Y., Berti, P.J., Schramm, V.L., 2000. Transition-state analysis for
depurination of DNA by ricin-A chain. J. Am. Chem. Soc. 122, 6527–
6534.
Degano, M., Almo, S., Sacchettini, J.C., Schramm, V.L., 1998. Trypan-
osomal nucleoside hydrolase. A novel mechanism from the structure
with a transition-state inhibitor. Biochemistry 18, 6277–6285.
Degano, M., Gopaul, D., Scapin, G., Schramm, V.L., Sacchetini, J., 1996.
Three-dimensional structure of the inosine–uridine nucleoside N-
ribohydrolase from Crithidia fasciculata. Biochemistry 35 (1996),
5971–5981.
An ad hoc bond-energy bond-order vibrational analysis
was used to match experimental intrinsic kinetic isotope
effects for [9-15N], [10-13C], [10-2H], [20-2H], and
[50-2H]adenosine 1 to proposed transition states using the
BEBOVIB-IV program from the Quantum Chemistry Pro-
gram Exchange (Sims et al., 1977; Berti and Tanake, 2002).
Due to the limitation on the maximum number of atoms
available in the BEBOVIB-IV program, a cutoff model
containing only relevant atoms was used (Sims and Lewis,
1984). The atoms omitted in the construction of the cutoff
model for transition state modeling were the 2, and 5
hydroxylic hydrogen atoms and the 3 hydroxyl group of
the ribose ring along with N1, C2, and the exocylic amino
group of the purine ring. Force constants for the various
vibrational modes were derived from reported values (Sims
et al., 1977; Sims and Lewis, 1984). The starting point for
the ground-state of adenosine was its crystal structure
(Lai and Marsh, 1972). This structure was then energy min-
imized using Spartan 5 and Gaussian 98w (Fig. 1). Transi-
tion-state structures were varied systematically to match
the observed kinetic isotope effects. The acceptable transi-
tion state gave a chemically reasonable structure that
matched the intrinsic kinetic isotope effects.
Edwards, R., 1996. S-adenosyl-L-methionine metabolism in alfalfa cell
cultures following treatment with fungal elicitors. Phytochemistry 43,
1163–1169.
´
Estupinan, B., Schramm, V.L., 1994. Guanosine–inosine-preferring nucle-
˜
oside N-glycohydrolase from Crithidia fasciculata. J. Biol. Chem. 37,
23066–23073.
Garrett, E.R., Mehta, P.J., 1972. Solvolysis of adenine nucleosides. I.
Effects of sugar and adenine substituents on acid solvolysis. J. Am.
Chem. Soc. 94, 8532–8541.
Gopaul, D., Meyer, S., Degano, M., Sacchettini, J., Schramm, V.L., 1996.
Inosine–uridine nucleoside hydrolase from Crithidia fasciculata.
Genetic characterization, crystallization, and identification of histidine
241 as a catalytic site residue. Biochemistry 35, 5963–5970.
Guranowski, A., Schneider, Z., 1977. Purification and characterization of
adenosine nucleosidase from barley leaves. Biochim. Biophys. Acta
482, 145–158.
Guthrie, R.D., Jencks, W.P., 1989. IUPAC recommendations for the
representation of reaction mechanisms. Acc. Chem. Res. 22, 343–349.

12
C. Bates et al. / Phytochemistry 67 (2006) 5–12
´
Horenstein, B.A., Parkin, D.W., Estupinan, B., Schramm, V.L., 1991.
˜
Mechanism from Isotope Effects. CRC Press, Boca Raton, FL, pp.
269–290.
Parkin, D.W., 1996. Purine-specific nucleoside N-ribohydrolase from
Trypanosoma brucei brucei. Purification, specificity, and kinetic mech-
anism. J. Biol. Chem. 271, 21713–21719.
Transition-state analysis of nucleoside hydrolase from Crithidia
fasciculata. Biochemistry 30, 10788–10795.
Horenstein, B.A., Zabinski, R.F., Schramm, V.L., 1993. A new class of C-
nucleoside analogues-1-(S)-aryl-1,4-dideoxy-1,4-imino-D-ribitols, tran-
sition state analogue inhibitors of nucleoside hydrolase. Tetrahedron
Lett. 34, 7213–7216.
Hunt, C., Gillani, N., Farone, A., Rezaei, M., Kline, P., 2005. Kinetic
isotope effects of nucleoside hydrolase from Escherichia coli. Biochim.
Biophys. Acta 1751, 140–149.
Imagawa, H., Yamano, H., Inoue, K., Takino, Y., 1979. Purification and
properties of adenosine nucleosidases from tea leaves. Agric. Biol.
Chem. 43, 2337–2342.
´
Parkin, D.W., Horenstein, B., Abdulah, D.R., Estupinan, B., Schramm,
˜
V.L., 1991a. Nucleoside hydrolase from Crithidia fasciculata. Meta-
bolic role, purification, specificity, and kinetic mechanism. J. Biol.
Chem. 266, 20658–20665.
Parkin, D.W., Mentch, F., Banks, G., Horenstein, B., Schramm, V.L.,
1991b. Transition-state analysis of a Vmax mutant AMP nucleosidase
by the application of heavy-atom kinetic isotope effects. Biochemistry
30, 4586–4594.
Kline, P.C., Serianni, A., 1990. 13C-enriched ribonucleosides: Synthesis
and application 13C–1H and 13C–13C spin-coupling constants to assess
furanose and N-glycoside conformations. J. Am. Chem. Soc. 112,
7373–7381.
Pelle, R., Schramm, V.L., Parkin, D.W., 1998. Molecular cloning and
expression of a purine-specific N-ribohydrolase from Trypanosoma
brucei brucei. Sequence, expression, and molecular analysis. J. Biol.
Chem. 273, 2118–2126.
Koerner, T., Westaway, K.C., Poirier, R.A., Wang, Y., 2000. An
unusually large secondary a-deuterium kinetic isotope effect in hydride
ion SN2 reactions. Can. J. Chem. 78, 1067–1072.
Koshiishi, C., Kato, A., Yama, S., Crozier, A., Ashihara, H., 2001. A new
caffeine biosynthetic pathway in tea leaves: utilization of adenosine
released from the S-adenosyl-L-methionine cycle. FEBS Lett. 499, 50–
54.
Kurtz, J., Exinger, F., Erbs, P., Jund, R., 2002. The URH1 uridine
ribohydrolase of Saccharomyces cerevisiae. Curr. Genet. 41, 132–141.
Lai, T.F., Marsh, R.E., 1972. The crystal structure of adenosine. Acta
Crystallogr., Sect. B 28, 1982–1989.
Lee, W.J., Pyler, R.E., 1986. Nucleic acid degrading enzymes of barley
mait. III. Adenosine nucleosidase from malted barley. Am. Soc. Brew.
Chem. J. 44, 86–90.
Petersen, C., Moller, L.B., 2001. The rihA, rihB, and rihC ribonucleoside
hydrolases of Escherichia coli. J. Biol. Chem. 276, 884–894.
Poirier, R.A., Wang, Y., Westaway, K.C., 1994. A theoretical study of the
relationship between secondary a-deuterium kinetic isotope effects and
the structure of SN2 transition states. J. Am. Chem. Soc. 116, 2526–
2533.
Poulton, J.E., Butt, V.S., 1976. Partial purification and properties of
adenosine nucleosidase from leaves of spinach beet (Beta vulgaris L.).
Planta 131, 179–185.
Schramm, V.L., Horenstein, B.A., Kline, P.C., 1994. Transition state
analysis and inhibitor design for enzymatic reactions. J. Biol. Chem.
269, 18259–18262.
Sethi, S., Gupta, S., Jenkins, E., Townsend, L., McCloskey, J., 1982. Mass
spectrometry of nucleic acid constituents. Electron ionization spectra
of selectively labeled adenines. J. Am. Chem. Soc. 104, 3349–3353.
Shi, W., Schramm, V.L., Almo, S., 1999. Nucleoside hydrolase from
Leishmania major. Cloning, expression, catalytic properties, transition
Le FlocÕh, F., Lafleuriel, J., 1981. The purine nucleosidases of Jerusalem
artichoke shoots. Phytochemistry 20, 2127–2129.
Leszczynska, D., Schneider, Z., Tomaszewski, M., Mackowiak, M., 1984.
Comparative studies on adenosine nucleosidase occurrence in plants.
Ann. Botany 54, 847–849.
˚
state inhibitors, and the 2.5 A crystal structure. J. Biol. Chem. 274,
21114–21120.
Matsson, O., Westaway, K.C., 1998. Secondary deuterium kinetic isotope
effects and transition state structure. In: Bethell, D. (Ed.), Advances in
Physical Organic Chemistry, vol. 31. Academic Press, San Diego, pp.
143–248.
Mazzella, L.J., Parkin, D.W., Tyler, P., Furneaux, R., Schramm, V.L.,
1996. Mechanistic diagnoses of N-ribohydrolases and purine nucleo-
side phosphorylase. J. Am. Chem. Soc. 118, 2111–2112.
Melander, L., Saunders Jr., W.H., 1980. Reactions Rates of Isotopic
Molecules. Wiley, New York, pp. 74–75.
Mentch, F., Parkin, D.W., Schramm, V.L., 1987. Transition-state
structure for N-glycoside hydrolysis of AMP by acid and by AMP
nucleosidase in the presence and absence of allosteric activator.
Biochemistry 26, 921–930.
Sims, L.B., Burton, G.W., Lewis, D.E., 1977Quantum Chemistry Program
Exchange, vol. 337. Indiana University, Bloomington, IN.
Sims, L.B., Lewis, D.E., 1984. Bond order methods for calculating isotope
effects in organic reactions. In: Buncel, E., Lee, C.C. (Eds.), Isotopes in
Organic Chemistry, vol. 6. Elsevier Science, New York, pp. 161–259.
Stern, M.J., Wolfsberg, M., 1966. Simplified procedure for the theoretical
calculation of isotope effects involving large molecules. J. Chem. Phys.
45, 4105–5124.
Swain, C.G., Stivers, E.C., Reuwer, J.F., Schaad, L.J., 1958. Use of
hydrogen isotope effects to identify the attacking nucleophile in the
enolization of ketones catalyzed by acetic acid. J. Am. Chem. Soc. 80,
5885–5893.
´
Versees, W., Decanniere, K., Pelle, R., Depoorter, J., Brosens, E., Parkin,
Miller, R.L., Sabouring, C., Krenitsky, T., Berens, R., Marr, J., 1984.
Nucleoside hydrolases from Trypanosoma cruzi. J. Biol. Chem. 259,
5073–5077.
D.W., Steyaert, J., 2001. Structure and function of a novel purine
specific nucleoside hydrolase from Trypanosoma vivax. J. Mol. Biol.
307, 1363–1379.
´
Murzin, A.G., Brenner, S.E., Hubbard, T., Chothia, C., 1995. SCOP: A
structural classification of proteins database for the investigation of
sequences and structures. J. Mol. Biol. 247, 536–540.
Orengo, C.A., Michie, A.D., Jones, S., Jones, D.T., Swindells, M.B.,
Thornton, J.M., 1997. CATH – a hierarchic classification of protein
domain structures. Structure 5, 1093–1108.
Versees, W., Decanniere, K., Van Holsbeke, E., Devroede, N., Steyaert, J.,
2002. Enzyme-substrate interactions in the purine-specific nucleoside
hydrolase from Trypanosoma vivax. J. Biol. Chem. 277, 15938–15946.
Verse´es, W., Steyeart, J., 2003. Catalysis by nucleoside hydrolases. Curr.
Opin. Struct. Biol. 13, 731–738.
Vorbruggen, H., Krolikiewicz, K., Bennua, B., 1981. Nucleoside synthesis
¨
Parkin, D.W., 1991. Methods for the determination of competitive and
noncompetitive kinetic isotope effects. In: Cook, P.F. (Ed.), Enzyme
with trimethylsilyl triflate and perchlorate as catalysts. Chem. Ber. 114,
1234–1255.