N1,N10-Ethylene-bridged high-potential flavins: Synthesis, characterization, and reactivity

Article abstract of DOI:10.1016/S0040-4020(01)00313-1

N¹,N¹⁰-Ethyleneisoalloxazinium chloride and its 8-Cl-, 7-CF₃-, and 3-CH₃-7-CF₃-substituted analogs were synthesized for the purpose of exhibiting thermal reactivity with organic substrates. The new flavins were characterized spectroscopically and electrochemically, and were found to react with amines, thiols, and phenylhydrazine, the latter case exhibiting catalytic aerobic recycling. Reactions of aliphatic benzylic and cyclopropyl amines with the 7-CF₃ analog were also compared to their oxidations by tris(phenanthroline)iron(III). All reactions of the flavinium salts appear to occur through heterolytic rather than homolytic mechanisms.

Full text of DOI:10.1016/S0040-4020(01)00313-1

TETRAHEDRON
Pergamon
Tetrahedron 57 32001) 4507±4522
N1,N10-Ethylene-bridged high-potential ¯avins:
synthesis, characterization, and reactivity
Wen-Shan Li, Nanjing Zhang and Lawrence M. Sayrep
Department of Chemistry, Case Western Reserve University, Cleveland, OH 44106, USA
Received 3 August 2000; accepted 14 March 2001
AbstractÐN1,N10-Ethyleneisoalloxazinium chloride and its 8-Cl-, 7-CF3-, and 3-CH3-7-CF3-substituted analogs were synthesized for the
purpose of exhibiting thermal reactivity with organic substrates. The new ¯avins were characterized spectroscopically and electrochemically,
and were found to react with amines, thiols, and phenylhydrazine, the latter case exhibiting catalytic aerobic recycling. Reactions of aliphatic
benzylic and cyclopropyl amines with the 7-CF3 analog were also compared to their oxidations by tris3phenanthroline)iron3III). All reactions
of the ¯avinium salts appear to occur through heterolytic rather than homolytic mechanisms. q 2001 Elsevier Science Ltd. All rights
reserved.
1. Introduction
¯avins in solution, but that the potential may be raised
signi®cantly upon binding of good substrates.6 This could
result from a conformation change that engages hydrogen
bonding interactions that make the isoalloxazine ring a
better electron acceptor and/or by enforcing a somewhat
bent binding cavity that favors the non-planar conformation
of the two-electron reduced ¯avin as opposed to the planar
Flox form. Model studies demonstrating both approaches
have been published,7 though the use of such systems in
reaction chemistry with reducing substrates has been
limited.
Flavin-dependent oxidases such as the mitochondrial mono-
amine oxidase 3MAO; E.C.1.4.3.4) and d-amino acid oxi-
dase 3E.C.1.4.3.3) are of considerable importance to
mammalian biochemistry.1 These enzymes carry out
dehydrogenation of the organic substrate, accompanied by
reduction of the tightly- or covalently-bound ¯avin,
followed by O2-dependent reoxidation of the reduced ¯avin
with by-production of H2O2. The mode of transfer of
electrons between the ¯avin moiety and substrate in the
dehydrogenation half-reaction remains incompletely
de®ned, the mechanistic proposals being sequential e2/
H1/e2 transfer,2 hydrogen atom transfer,3 hydride transfer,
or two-electron transfer by way of a covalent intermediate
3addition±elimination).4 The electron-transfer pathway for
¯avin-mediated oxidation of amines has been arti®cially
modeled by photochemically driven electron transfer,5 but
normal ¯avins 3e.g. 3-methylumi¯avin) exhibit poor
ground-state reactivity toward amines 3or amino acids) on
account of the low reduction potential of the oxidized
¯avins 3,2 0.3 V vs. NHE). Thus it is necessary to
consider strategies that activate typical ¯avins and/or
choose analogs with built-in activation, to create a more
favorable reaction free energy. It has been determined in
the case of monoamine oxidase B that the ¯avin potential
in the enzyme active site is about the same as for simple
Our approach was to use a combination of two factors favor-
ing reduction of the Flox form: alkylation at N1 and attach-
ment of electron-withdrawing groups to the benzene
subnucleus. It is well known that electron-withdrawing
groups at position C7 and C8 as well as alkylation of either
N1 or N5 increase the ¯avin redox potential and reactivity
toward nucleophiles. In an effort to model a possible C4a
addition±elimination reaction for MAO, Mariano and
coworkers studied the N5-ethyl ¯avinium compound,4
shown by Bruice years ago to raise the one-electron reduc-
tion potential by 600 mV and to permit the generation of
stable C4a adducts.8 In the reaction with amines, the N5-ethyl
¯avinium compound generated C4a-amine adducts readily,
which however, underwent the desired elimination reaction
only upon heating in the presence of a base.4a Our choice of
examining N1-alkylation was based on our desire to avoid
a bias toward generation of metastable C4a adducts. The
decision to focus on the N1,N10-ethylene-bridged compound
rather than another 1,10-alkylated ¯avinium salt, such as
the corresponding N1,N3,N10-trimethyl analog, was based
on the report that the former is much more stable toward
rearrangement and/or degradation,9 possibly explaining
why it acts as a better redox catalyst.10
Keywords: ¯avin analogs; tris31,10-phenanthroline)iron3III); phenylhydra-
zine; monoamine oxidase.
p
Corresponding author. Tel.:11-216-368-3704; fax: 11-216 368-3006;
e-mail: LMS3@po.cwru.edu
Abbreviations: Flox: oxidized ¯avin; TPIP: tris31,10-phenanthroline)iron-
3III) perchlorate.
0040±4020/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved.
PII: S0040-4020301)00313-1

4508
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
Scheme 1. Key: 3i) ethanolamine, K2CO3, DMSO, 908C; 3ii) Pd/C, NH41HCO22, MeOH; 3iii) alloxan monohydrate, H3BO3, HOAc, 608C; 3iv) N-methyl-
alloxan monohydrate, H3BO3, HOAc, 608C; 3v) SOCl2, room temperature; 3vi) Sn/HCl 3aq).
Thus, the N1,N10-ethylene-bridged ¯avinium analogs 1a±d
were synthesized, along with the N3-methyl compound 1d
needed in order to evaluate the possibility of N3-deproto-
nation in 1c in the reaction with basic reductants, a situation
made likely by the combination of partial positive charge at
N1 and the C7±CF3 group. Details on the synthesis and
characterization of the analogs are provided along with
initial reactivity studies with thiols, hydrazines, and amines.
In the latter case, comparison of relative reactivity data to
that obtained for reaction of the same amines with the well-
known one-electron transfer oxidant FeIII3phen)3, suggests
that the ¯avin reactions utilize a mechanism other than
single-electron transfer. The following paper11 focuses on
the mechanism of the reaction of amines with the ¯avinium
compounds.
¯avin,12 in turn prepared by the classical condensation
reaction13 of alloxan with the appropriately substituted
N-32-hydroxyethyl)-ortho-phenylenediamine
compound
3Scheme 1). The syntheses started with the respective
2-bromonitrobenzene 2a±c, which was reacted with etha-
nolamine in the presence of K2CO3 as base to give the
2-nitroaniline derivatives 3a±c. Note that the substitution
of 2b was regioselective, as reported.14 The 2-nitroanilines
were puri®ed by chromatography and by recrystallization
from methanol to give orange crystalline solids. Catalytic
transfer hydrogenation of 3a and 3c in dry MeOH at 08C,
using ammonium formate and 10% palladium on activated
carbon as a catalyst, gave the diamine compounds 4a and
4c. In the case of the chloro analog 3b, catalytic hydrogena-
tion would result in dehalogenation, and thus its reduction
was accomplished with mossy tin in aqueous HCl.
The oxygen-sensitive diamine compounds 4a±c were
condensed anaerobically with alloxan monohydrate 3or
N-methylalloxan monohydrate for 1d) in acetic acid in
the presence of boric acid to give 10-32-hydroxyethyl)iso-
alloxazines 5a±d, which were subjected to reaction with
SOCl2 to afford the desired ¯avin compounds 1a±d. Obtain-
ing a good yield and purity in this ®nal reaction required
keeping the volume of SOCl2 as small as possible and keep-
ing the temperature at or below room temperature.
The ®nal ¯avinium salts 1a±d were characterized by 1H and
13C NMR spectroscopy, and formation of the cationic
N1,N10-ethylene bridge was con®rmed by the $1 p p m
down®eld shift of the four methylene protons relative to
the precursor ethanolamines 5a±d. Further veri®cation of
1a±c was achieved by isolation and full characterization 31H
and 13C NMR, HRMS) of the N5-acetyl derivatives 6a±c
3using zinc in acetic anhydride,12 Scheme 2) of the reduced
dihydro¯avins, which themselves were unstable to air
oxidation. Interestingly, whereas reductive acylation of 1a
and 1b required 1 h at room temperature or 30 min at re¯ux,
2. Results and discussion
2.1. Synthesis
Synthesis of the ¯avinium compounds 1a±d was achieved
by an intramolecular cyclization reaction effected by the
action of SOCl2 on the corresponding N10-32-hydroxyethyl)-

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
4509
Scheme 2.
reduction of 1c was complete in 5 min at room temperature,
consistent with an expected higher reduction potential for
this analog.
300 mV positive of DMI. Although obtaining reversible CV
waves for 1a and 1b required rapid scan rates, suggesting
instability of the resulting ¯avin radicals, the cyclic volt-
ammogram for the 7-CF3 analog 1c exhibited a reversible
one-electron redox wave even at scan rate below 50 mV/s.
2.2. Electrochemical properties
Despite the potential-raising effect of the N1,N10-ethylene
bridge, the increase in 200±250 mV observed is signi®-
cantly less than the 600 mV increase seen for N5-alkylation.
This distinction is probably a re¯ection of the fact that the
N1,N10-ethylene bridge resists the conformational change
from planar to bent that normally accompanies ¯avin reduc-
tion,15 whereas this is allowed in the case of N5-alkylation.
In this way, even the most electrophilic of the analogs
studied here 37-CF3 analog 1c) possesses only about half
of the increase in redox potential induced by N5-alkylation,8
and thus might be considered more biologically relevant.
Both cyclic voltammetry and the OSWV method were used
for electrochemical characterization of 1a±c. From the
measured one-electron redox potentials 3Table 1), it is
clear that all of these N1,N10 bridged ¯aviniums have much
higher potentials than the normal lumi¯avin analog N3,N10-
dimethylisoalloxazine 3DMI). Compound 1a exhibits a
potential about 200 mV positive of DMI, showing the effect
of the positive charge associated with the N1,N10-ethylene
bridge alone. In addition, the substitution of Cl at C8 and
CF3 at C7 increases the potential further, such that the most
highly oxidizing 7-CF3 analog 1c has a potential more than
Table 1. Redox potentials for ¯avinium salts 1a±c
2.3. UV±visible spectroscopic properties
Isoalloxazine
Cyclic voltammetry method
OSWV Method
The long wavelength absorption bands of ¯avinium salts
1a±1c and their reduced forms are listed in Table 2. A
conventional isoalloxazine derivative such as DMI is
known to have two characteristic absorption bands at ca.
329 and 437 nm for the oxidized form.16 The shorter-wave-
length absorption band is known to be sensitive to solvent
polarity, shifting to shorter wavelength in a hydrophobic
environment, whereas the longer wavelength absorption
band is fairly insensitive to solvent polarity.17 For the
¯avinium salts 1a±c, the shorter wavelength peak is red
shifted and the longer wavelength peak is blue shifted
compared with those bands exhibited by DMI. The parent
analog 1a exhibited lmax values at 356 and 403 nm at pH 6,
each about 20 nm blue-shifted from that reported for the
0
Ep,c 3mV)a
Eo 3mV)b
E 3mV)c
1a
1b
1c
DMId
2178
2124
294
2167
2103
272
2152
284
264
2413
2376
2396
a
Cathodic peak potential vs. SCE in water using 1 mM ¯avin, 0.167 M
phosphate 3pHˆ6), glassy carbon electrode, scan rateˆ50 mV/s.
Formal reduction potential 3Ep,a1Ep,c)/2 vs. SCE in water using 1 mM
¯avin, 0.167 M phosphate 3pHˆ6), glassy carbon electrode. Scan
rateˆ50 mV/s for 1c and DMI, 6000 mV/s for 1b, and 3000 mV/s for 1a.
Osteryoung square-wave voltammetry 3vs. SCE) in water using 1 mM
¯avin, 0.167 M phosphate 3pHˆ6), glassy carbon electrode.
N3,N10-dimethylisoalloxazine.
b
c
d
Table 2. Long wavelength absorption of ¯avinium salts 1a±c and their reduced formsa
Compound
Oxidized forma
max 3nm)
Reduced formb
max 3nm)
Solvent
l
1 3mM21 cm21
)
l
1 3mM21 cm21
)
1a
356
403
352
396
368
402
392
340
396
340
388
13.4
10.8
13.7
10.4
16.6
14.3
15.7
12.5
100
301
306
300
7.2
8.7
7.5
H2O
CH3CN
H2O
1b
1c
308
312
362
304
7.6
7.9
2.6
8.1
CH3CN
H2O
10.1
11.5
CH3CN
a
Stock solutions 35 mM) of 1a±c in DMSO were diluted 1:100 into either H2O±CF3COOH 50:1 3pH 6) or CH3CN.
Stock solutions 35 mM) of 1a±c in DMSO were diluted 1:100 into either H2O or CH3CN, followed by the addition of Na2S2O4 3®nal concentration 0.06 mM)
or NaBH4 3®nal concentration 0.03 mM), respectively.
b

4510
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
Scheme 3.
corresponding 7,8-dimethyl derivative.9 Compared to 1a,
whereas the 8-Cl substituent resulted in a slight red shift
of the most intense absorption, the 7-CF3 substituent
was associated with a blue shift of both absorption bands.
The 7-CF3 N3-methyl analog 1d exhibited spectral proper-
ties nearly identical to those of 1c, with lmax at 336 and
388 nm in CH3CN. The red shift seen for 1b probably
re¯ects resonance donation by chlorine, even though the
redox potential indicates that the 8-Cl substituent is overall
electron-withdrawing. In contrast to conventional iso-
alloxazines 3e.g., DMI), a change in solvent from H2O to
CH3CN resulted in little or no hypsochromic shift of either
band of 1a±c, a ®nding that must re¯ect the permanent
3delocalized) positive charge.
¯avins 1c in aqueous CH3CN with Na2S2O4, hydrazine, or
1,3-propanedithiol 3see below) followed by evaporation of
solvent anerobically and quenching with acetic anhydride
in degassed CF3COOH, led to the isolation of the same
N5-acetyl derivative 6c prepared above by the Zn/Ac2O
method. The N5-acetyl derivatives 6a±c exhibited absorp-
tions in DMSO near 260 nm 3stronger) and near 300 nm
3weaker), consistent with previous reports.20
2.4. Reduction of 1c by N-benzyl 5,6-dihydrophen-
anthridine
Since the reaction of phenylhydrazine with the 7-CF3 indi-
cated that reoxidation of the reduced ¯avin by O2 is slow
3see below), it was hoped that use of 1 equiv. of a mild
reducing agent might permit isolation of the reduced ¯avin,
at least in this case. Trial and error led to the ®nding that
reaction of 1c and 1d with N-benzyl-5,6-dihydrophenanthri-
dine 37) in acetonitrile generated the reduced forms 8c and
8d as light yellow solids that precipitated from solution
3Scheme 3).21 Although insuf®ciently soluble for 13C
Reduced ¯avins absorb strongly around 300 nm, with a
shoulder around 390±400 nm characteristic of the bent
geometry.18 The dihydro derivatives of 1a±c generated by
reduction in degassed 3anaerobic) solution, using Na2S2O4
in water or NaBH4 in CH3CN, displayed the expected
absorption around 300 nm, but lacked a discernible shoulder
around 400 nm, consistent with the conformational restric-
tion imposed by the 1,10-ethylene bridge. The shoulder at
,360 nm seen for 1c probably re¯ects the electronic effect
of the 7-CF3 substituent, since an absorption in this region is
consistently observed for dihydro¯avins containing other
strongly electron-withdrawing substituents.19 Despite the
clear generation of dihydro¯avins in solution, all attempts
to isolate them by treatment with traditional chemical reduc-
tants failed to yield material of suf®cient analytical purity
for full characterization. However, preparative reduction of
1
NMR study, 8c and 8d were characterized by H NMR
and by mass spectrometry, which to our knowledge, consti-
tutes the ®rst such achievement for a reduced N1,N10-dialkyl
dihydro¯avin.22
2.5. Reactions with phenylhydrazine
Alkyl- and arylhydrazines have been widely studied as
MAO inactivators23 owing to their antidepressant proper-
ties.24 Catalytic turnover of these compounds involves
Figure 1. Reaction of 1a with phenylhydrazine.

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
4511
Figure 2. Reaction of 1b with phenylhydrazine.
dehydrogenation to the corresponding azo compounds
RNyNH and ArNyNH, which decompose to N2 and the
respective hydrocarbon. Inactivation results from a com-
peting reaction resulting ultimately in the attachment of
catalase `scrubber' system, the lag period was eliminated.
These results suggest that 1a is being reduced by phenyl-
hydrazine in a manner that the resulting reduced ¯avin is
immediately reoxidized when O2 is present. The end of the
lag period coincides with the point where the dissolved O2 is
consumed.
25
the alkyl or aryl groupto the ¯avin.
The reactions of ¯avinium salts 1a±c with phenylhydrazine
in aqueous buffer were followed spectrophotometrically in a
closed cuvette 3monitoring 2DA of the long-wavelength
¯avin band, Figs. 1±3), such that if catalytic turnover
were to occur, involving reoxidation of the reduced ¯avin
by O2, the extent of turnover would be limited by the initial
amount of dissolved O2. Upon mixing of 1a with phenyl-
hydrazine, there was observed a lag period for the ®rst
1000 s, followed by reduction of [1a] over the next 300 s,
at which time the absorbance again plateaued 3Fig. 1).
However, the lag period was reduced by more than two-
fold if argon was bubbled through the solutions of the two
reactants. When the reaction was run in O2-free solution
afforded by the presence of a glucose/glucose oxidase/
The reaction of 8-Cl analog 1b with PhNHNH2 exhibited
basically the same behavior 3Fig. 2) as seen for 1a. The lag
seen without argon bubbling indicates that reoxidation of
the reduced ¯avin resulting from oxidation of PhNHNH2 by
1b is faster than its generation. However, for the reaction of
7-CF3 analog 1c, the decrease in absorbance did not exhibit
a lag period, and the increase in A340 observed following the
initial decrease was very slow 3Fig. 3), suggesting that the
reduced ¯avin in this case is reoxidized very slowly by O2.
Upon argon bubbling or use of the `O2 scrubber', the rate of
decrease in A340 was accelerated, consistent with an inter-
pretation that the initially observed A340 decrease without
argon bubbling represents a composite of simultaneous
Figure 3. Reaction of 1c with phenylhydrazine.

4512
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
Figure 4. Reaction of 1a with phenylhydrazine under controlled atmosphere conditions.
¯avin reduction 3faster) and reduced ¯avin reoxidation
3slower).
margin. The third cycle of O2 introduction exhibited a
further deterioration of the initial cycling seen. Our inter-
pretation of this data is that repetitive redox cycling of 1a
with PhNHNH2 results in eventual irreversible conversion
of 1a to a form that is inactive as a redox catalyst, but which
absorbs at 356 nm with an extinction similar to that of 1a.
Although no product analysis was performed on this system,
it is possible that the irreversible reaction re¯ects covalent
modi®cation of the ¯avin by the phenyl group of PhNHNH2
as occurs when MAO is irreversibly inactivated by
PhNHNH2.25
The nature of the reaction products was investigated in the
case of ¯avin 1c in CH3CN. Although the solubility of 1c in
this solvent was low, the reaction of 1.1 equiv. of PhNHNH2
resulted in a homogeneous solution that when quenched
with acetic anhydride, permitted isolation of the N5-acetyl
derivative 6c of the reduced ¯avin 8c. In addition, when the
reaction was followed in an NMR tube in CD3CN, the
formation of phenyldiazene 3PhNyNH) was con®rmed
through detection of its well-known decomposition product,
benzene.26
Overall, these results represent the ®rst report of catalytic
activity of a model ¯avin for the O2-dependent oxidation of
PhNHNH2. With use of an O2 scrubber, the initial rate of
decrease of the oxidized ¯avinium absorbance can be taken
to re¯ect the pseudo-®rst-order rates for reduction of each
¯avinium salt 1a±c by PhNHNH2, and these values are
listed in Table 3. As expected, the rate rank order re¯ects
the increase in oxidizing strength 1a,1b,1c, evident from
the measured redox potentials 3Table 1). At the same time
however, this rank order does not re¯ect the ef®ciency of the
¯avinium compound as a redox catalyst, since catalysis
depends on both the rate of reduction by PhNHNH2 and
the rate of reoxidation by O2. The reoxidation rate follows
1a.1b@1c, with the latter reduced 7-CF3 analog being
nearly resistant to O2-mediated reoxidation.
To further con®rm that the lag period observed for parent
¯avinium analog 1a re¯ects immediate O2-mediated reoxi-
dation of the reduced ¯avin, reaction of 1a with PhNHNH2
was studied under a condition where either argon or air
could be bubbled along the side of the cuvette at a ¯ow
rate of 5 mL/min. The reaction was initiated under argon
bubbling, and the A356 decreased steadily over the ®rst 6 min
3Fig. 4). At that point, air was bubbled for 2 s, resulting in
immediate return to the initial absorbance and then a
decrease over the next 10 min but to a lesser extent than
initially. Again air was bubbled for 2 s, but the decrease in
A356 following the initial rapid rise now exhibited a lag
period and then a slower decrease, again by a reduced
2.6. Reaction with thiols
Table 3. Pseudo-®rst-order rates for reaction of ¯avinium salts 1a±c with
PhNHNH2´HCla
It has been reported that ¯avins can effect the oxidation of
thiols to disul®des by an addition±elimination mechanism
proceeding via a C4a-mercapto adduct.27 It was thus of
interest to determine whether ¯avinium salts 1a±c could
also oxidize thiols such as thiophenol. In the case of the
8-chloro analog 1b, the reaction with thiophenol resulted
in a nearly isosbestic spectral change 3Fig. 5) indicative of
conversion of the ¯avin to a highly delocalized species 3for
a short time an intermediate can be detected at 410 nm). The
b
Flavin
Rate 3s21
)
1a
1b
1c
0.0102
0.032
0.104
a
Reaction conditions: 0.05 mM ¯avin, 0.67 mM PhNHNH2´HCl, 16 mM
glucose, 165 mM potassium phosphate buffer 3pH 6.01), 158C, containing
6.7 mg glucose oxidase and 1.0 mg catalase per mL.
Monitoring the decrease in the long wavelength Flox absorption.
b

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
4513
Figure 5. Time course of the spectral changes for the reaction of 8-chloro ¯avin 1b with thiophenol: [1b] 5£1025 M, [thiophenol] 5£1024 M, CH3CN, 258C.
Scheme 4.
intense product absorptions, lmax 3CH3CN) 450 nm
31ˆ34 mM21 cm21) and 282 nm 31ˆ17 mM21 cm21), are
consistent with the spectrum reported for 8-3arylmercapto)-
ribo¯avin [lmax 475 nm 31ˆ26 mM21 cm21) at pH 7],28
suggesting that the reaction involves nucleophilic displace-
ment of the 8-chloro substituent to give 8-3phenylmer-
capto)-N1,N10-ethyleneisoalloxazinium chloride 39). The
marked bathochromic shifts must re¯ect conjugative
electron release by the 8-thio substituent, as exempli®ed
by the paraquinoid resonance form 3Scheme 4).
Although the nuclear substitution undergone by 8-Cl analog
1b with thiols prevented an investigation of the ability of 1b
to catalyze thiol!disul®de oxidation, reactions of 1a and 1c
with thiols 3thiophenol or 1,3-propanedithiol) showed spon-
taneous bleaching of absorptions of the oxidized ¯avins,
concomitant with appearance of new absorptions at
303 nm characteristic of the reduced ¯avins. Furthermore,
NMR tube reactions of 1c with 1,3-propanedithiol in either
CD3CN or D2O exhibited rapid appearance of signals corre-
sponding to 1,3-dithietane. Additional evidence for thiol-
mediated reduction was obtained for 1c by isolation of the
N5-acetyl derivative 6c following quenching of the thiol
reaction with acetic anhydride. These ®ndings support the
ability of 1a and 1c to mediate oxidation of organic reduc-
tants, though no information on mechanism could be
obtained due to the rapidity of the reactions.
Table 4. Oxidation of benzylamines by ¯avins 1a, 1c, and 1da
Flavin
Amine
Yield of PhCHO 3%)b
1a
PhCH2NH2
25.0
5.0
,0.6
147.0
44.0
4.8
118.0
20.0
1.6
PhCH2NHCH3
PhCH2N3CH3)2
PhCH2NH2
PhCH2NHCH3
PhCH2N3CH3)2
PhCH2NH2
2.7. Oxidation of benzylamines
1c
Reactions of benzylamine and its N-methyl and N,N-
dimethyl analogs were explored with ¯avinium salts 1a
and 1c 3it was feared that the 8-Cl analog 1b would be
susceptible to nuclear substitution). In order to evaluate
relative reactivity as a function of both amine and ¯avin,
preparative-scale reactions were conducted with quanti-
tation of the oxidative debenzylation product PhCHO as
its 2,4-dinitrophenylhydrazone. The reactions were run in
stoppered ¯asks under argon to prevent evaporative loss of
1d
PhCH2NHCH3
PhCH2N3CH3)2
a
Conditions: [Flavin]:[Amine]:[Amine hydrochloride]ˆ10 mM:55 mM:
5 mM, 508C, 48 h under argon with exposure to air 35 min) after 6 and
12 h. PhCHO was removed by azeotropic distillation and quanti®ed as the
2,4-DNP derivative.
Yield based on ¯avins after correction for minor spontaneous autoxi-
dation.
b

4514
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
Figure 6. Time course of benzylamine oxidation by 1c. Reaction conditions: 55 mM PhCH2NH2/5 mM PhCH2NH2´Cl/508C/CH3CN/2 days in the dark under
argon with exposure to air for 5 min after 6 h and 12 h, in the presence 3V) and absence 3O) of 10 mM 1c. Yield of PhCHO based on ¯avin catalyst.
Scheme 5.
solvent, but O2 was admitted periodically to allow for cata-
lytic recycling. The background-corrected yields after 48 h
of reaction for ¯avins 1a, 1c, and 1d are listed in Table 4. It
is apparent that the 7-CF3-substituted ¯avins 1c and 1d are
stronger debenzylating agents than the unsubstituted parent
1a, and that the amine reactivity order is 18.28.38. In
addition, it should be noted that for 1c and 1d, the yield
of PhCHO based on ¯avin exceeds 100%, indicative of
some aerobic recycling. As can be seen from the time course
of PhCHO generation in the reaction of 1c with benzyl-
amine 3Fig. 6), PhCHO was produced rapidly in the ®rst
0.5 h and then more slowly over 24 h. The substantial slow-
ing of PhCHO production in excess of 1 equiv. based on
¯avin suggests that O2-dependent regeneration of oxidized
¯avin 1c occurs slowly, consistent with what was seen in its
reaction with phenylhydrazine.
rather than formation of an amine±¯avin adduct, which
would be sterically dif®cult for tertiary amines. Using
substoichiometric amounts of amines, the shift was shown
to be isosbestic and titratable 3data not shown). Evidence
that the spectral change represented a reversible equilibrium
was that the original spectrum of 1c could be reestablished
when the incubations of 1c with amines were quenched with
excess HCl or CH3COOH. Con®rmation of the occurrence
of N3-deprotonation29 was that no immediate spectral shifts
were observed when the same reactions were conducted
with N3-methyl analog 1d.
Subsequent to the rapid shift from 388 nm to 370 nm, irre-
versible reactions of the amines with 1c could be assessed
by measuring the rate of decay of the 370 nm band. The rate
constants obtained under pseudo-®rst order reaction con-
ditions are listed in Table 5. The apparent amine reactivity
towards 1c again follows the order 18.28.38, though the
nature of the reaction mechanism is not revealed by this
data, and may not be the same for both primary/secondary
and tertiary amines. In fact, whereas for primary and
secondary amines, the decrease in A370 was accompanied
by appearance of a band at 390 nm, the slower reaction of
the tertiary amines was accompanied by appearance of a
band at 376 nm. For the reactions with N3-methyl analog
1d, there was no initial spectral shift in the starting 388 nm
chromophore, but the same bands at 390 nm 3primary and
secondary amines) or 376 nm 3tertiary amines) were still
generated in the same relative rate order 3data not shown).
Structural interpretations of these spectral changes, accom-
plished using 13C-labeled ¯avins and 15N-labeled amines,
permitted ascribing the 390 nm band to the C10a-amine
adducts.11 Although unstable and previously identi®ed
tentatively only on the basis of spectral changes,30,31
C10a-nucleophile adducts in N1,N10-dimethyl ¯avinium
salts have been invoked as intermediates to rationalize
the isolation of spirohydantoin rearrangement products
The rates of reaction of 1c with benzylamine, its N-methyl
and N,N-dimethyl analogs, and tributylamine were followed
spectrophotometrically in amine/amine´HCl 310:1) buffered
CH3CN solution. In every case there was an immediate shift
of the oxidized ¯avin spectrum 3lmaxˆ388 nm) to one
exhibiting a lmax of 370 nm. Since this change was observed
even for tertiary amines, it appeared that the spectral shift
represented equilibrium deprotonation at N3 3Scheme 5)
Table 5. Pseudo-®rst-order rates for reaction of 7-CF3 ¯avin 1c with
benzylamines
a
Amines
Rate 3min21
)
PhCH2NH2
PhCH2NHCH3
PhCH2NMe2
n-Bu3N
0.59
0.085
0.075
0.014
a
From t1/2, monitoring reduction at 370 nm 3observed immediately upon
mixing amine with ¯avin). Reaction conditions: 0.05 mM 1c, 1 mM
amines/amine´HCl 310:1), CH3CN, 508C.

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
Table 6. Oxidation of benzylamines by 3phen)3Fe3III)a
4515
Amines
Products
Yield 3%) for [amine]ˆ28mMb
Yield 3%) for
[amine]ˆ560mMb
PhCH2NH2
PhCH2NHCH3
PhCHO
PhCHO
CH2O
PhCHO
CH2O
8.3
11.1
29.6
42.0
0.5
31.2
21.9
PhCH2N3CH3)2
20.6 323.0)c
0.5 36.5)c
a
Reaction conditions: 258C for 1 h in dry CH3CN under argon, [Fe3phen)33ClO4)3]ˆ56 mM.
Yield of the 2,4-dinitrophenylhydrazone based on 2 equiv. of Fe3phen)33ClO4)3.
Solid anhydrous K2CO3 35 equiv. based on Fe3III)) was added to the reaction ¯ask.
b
c
resulting from reaction of the ¯avinium salts with
water,30,32,33 alcohols,31,33 or amines33 under basic condi-
tions.
indicates that the ¯avin reactions do not followan electron-
transfer mechanism.
2.9. Oxidation of cyclopropyl-containing amines
2.8. Oxidation of benzylamines by tris31,10-phenan-
throline)iron3III) perchlorate
In the reaction with amines, the suggestion of an electron-
transfer oxidation mechanism in the case of TPIP but not in
the case of the ¯avinium salts, led us to investigate the
interaction of these two oxidants with cyclopropylamines,
which are known to undergo one-electron oxidative ring-
openings. TPIP-induced oxidation of N-cyclopropylbenzyl-
amine, N-cyclopropyl-a-methylbenzylamine, 31-phenyl-
cyclopropyl)methylamine, and the primary amine `control'
of the latter reactant, phenethylamine, was studied under
either the K2CO3-enhanced or excess-amine-enhanced
condition. Reactivity was evaluated by quanti®cation of
the 2,4-dinitrophenylhydrazine derivatives of the aldehyde
products, and the results are listed in Table 7. Oxidation of
the two cyclopropylamines resulted in products of benzylic
oxidation, but also products of cyclopropyl ring-opening, as
expected for generation of an N-cyclopropyl aminium
cation-radical intermediate. In contrast, the homologous
cyclopropylcarbinylamine exhibited only the ring-closed
product, as expected. The primary amine control of the latter
amine 3same carbon atom spacing) underwent reaction only
sluggishly.
Preparative-scale reactions of benzylamine and its N-methyl
and N,N-dimethyl analogs with the well-known outer-
sphere electron-transfer oxidant tris31,10-phenanthroline)-
iron3III) perchlorate 3TPIP) were conducted using an
amine:TPIP stoichiometry of 1:2 as well as 20:1. Higher
total levels of N-dealkylation were observed for the second-
ary and especially tertiary as opposed to primary amine in
both reaction conditions 3Table 6). Although an overall two-
electron oxidation of amine should theoretically consume 2
equivalents of Fe3III), the total yield of N-dealkylation
products based on Fe3III) increased 2±3-fold by increasing
the amine concentration 20-fold. In addition, variation in the
reaction conditions for one case 3third entry in Table 6)
revealed that addition of solid K2CO3 to the reaction ¯ask
resulted in a substantial increase in reaction yield, suggest-
ing that the difference in yield between the 1:2 and 20:1
conditions re¯ects increased basicity of the medium using
excess amine. It is interesting that the competition between
N-debenzylation and N-demethylation of the tertiary amine
varied markedly with the amine:TPIP stoichiometry, though
the reason for this is not clear. Nonetheless, the ®nding that
the order of amine reactivity, 38.28.18, is exactly opposite
to that found using ¯avins 1 as oxidants 3Tables 4 and 5),
A mechanism consistent with formation of the aldehyde
products arising from oxidation of the two cyclopropyl-
amines by TPIP is shown in Scheme 6. This mechanism
Table 7. Oxidation of cyclopropylamines by Fe3phen)33ClO4)3 3TPIP)
Conditionsa
Substrates
Productsb
Yield 3%)c
Amine:TPIP:3added base)
ratio
1:2:10 3K2CO3)
PhCHO
CH3CH2CHO
CH2yCHCHyO
10.0
3.6
,0.5
1:2:10 3K2CO3)
PhC3CH3)yO
CH3CH2CHO
CH2yCHCHyO
3.5
3.5
1.1
20:2
4.1
9.9
20:23D808C)
PhCH2CH2NH2
PhCH2CHO
a
Standard conditions: 0.56 mmol TPIP1amines, 258C, 1 h, 10 mL CH3CN, under Ar.
Products were isolated as 2,4-dinitrophenylhydrazine derivatives.
Yields are based on 2 equiv. of Fe3phen)33ClO4)3.
b
c

4516
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
Scheme 6.
suggests a partitioning of the initial aminium cation-radical
between benzylic deprotonation, leading to PhCHO or
PhC3yO)CH3, and ring opening. On the other hand, reaction
of the same two cyclopropylamines with ¯avin 1c gave no
evidence for ring-cleavage products, as determined by
monitoring the reactions in an NMR tube. Instead, the spec-
trum revealed 27% conversion of N-cyclopropylbenzyl-
amine to benzylidenecyclopropylamine, and a low level
3,1%) of the analogous conversion in the case of N-cyclo-
propyl-a-methylbenzylamine. In the latter case, a-branch-
ing may sterically retard whatever mechanism is responsible
for oxidation by 1c. Nonetheless, the absence of ring-
cleavage products supports the previous conclusion that
oxidation of amines by the ¯avinium salts 1 involves a
mechanism distinct from electron transfer.
in the reaction using N5-ethyl ¯avinium cation,4b where an
addition±elimination mechanism was proposed. It is well
known that cyclopropylamines can become the subject of
heterolytic4b,34,35 as well as homolytic ring-cleavage
processes. Thus, the result for 2-phenylcyclopropylamine
does not alter the conclusion reached above.
3. Conclusions and mechanistic proposals
This paper describes new high-potential N1,N10-ethylene-
bridged ¯avinium salts, which have been characterized
spectroscopically and electrochemically, and found to
induce thermal reactions with thiols and phenylhydrazine.
In the latter case, catalytic recycling of the parent analog 1a
was demonstrated, though catalysis by the higher-potential
analog 1c appears to be limited by slow O2-dependent
oxidation of the reduced ¯avin species. In addition, initial
reactivity studies on the reaction of benzylic and cyclo-
propyl-containing amines with the ¯avinium salts have
been compared to the reactions with the well-known
Interestingly, even the unsubstituted parent ¯avin 1a
effected oxidative cleavage of the highly activated
compound 2-phenylcyclopropylamine, leading to isolation
of the 2,4-dinitrophenylhydrazone of cinnamaldehyde in
45% yield. However, such conversion was also observed
Scheme 7.

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
4517
Scheme 8.
one-electron oxidant tris3phenanthroline)iron3III) perchlo-
rate 3TPIP). Whereas both the effect of alkyl substitution
on amine reactivity 338.28.18) and the observation of ring-
cleavage products from benzylic cyclopropylamines are
consistent with an electron-transfer oxidation mechanism
for TPIP, the opposite reactivity rank order and lack of
ring-cleavage products for benzylic cyclopropylamines
indicate that the ¯avin reactions do not involve electron
transfer. For the related N5-ethyl ¯avinium salts, an
addition±elimination mechanism via a C4a-adduct inter-
mediate was demonstrated for the observed oxidation of
amines.4 In the following paper, we provide evidence that
dehydrogenation of primary and secondary amines by
N1,N10-ethylene-bridged ¯avins 1 also involves mainly an
addition±elimination mechanism, but in this case pro-
ceeding via C10a adducts.11 For tertiary amines with strong
hydride donor capacity such as 7, a hydride transfer
mechanism appears to occur.21
recorded on a Varian Gemini 300 spectrometer. Chemical
shifts are reported relative to the standard resonance of the
residual protons in the deuterated solvents. High-resolution
mass spectra 3HRMS) were obtained on a Kratos MS25RFA
spectrometer using electron impact ionization 320±40 eV).
UV±visible spectra were recorded with a Perkin±Elmer
model Lambda 3B spectrophotometer ®tted with a water-
jacketed multiple cell holder for maintenance of constant
temperature. Spectral scan and kinetics data 3at constant
temperature) were obtained using PECSS software 3Perkin
Elmer Corporation, 1988). Thin-layer chromatography was
performed with Merck silica gel G plates. The solvents used
for kinetics determinations 3e.g., CH3CN was distilled from
P2O5) and reactive reagents such as thionyl chloride were
puri®ed by standard methods.36 The water used in this study
was deionized and doubly distilled. 3,10-Dimethyliso-
alloxazine 3DMI) was synthesized as described.16
4.2. Electrochemistry
The evidence for C10a nucleophile adducts formed from
N1,N10-dimethyl ¯avinium salts30±33 and for the N1,N10-
ethylene-bridged compounds 110 suggests that the oxi-
dations of phenylhydrazine and thiols by ¯avins 1 also
proceed via C10a adducts. Our proposed mechanisms are
shown in Schemes 7 and 8. For the reaction of PhNHNH2
with ¯avin 1a, addition of PhNHNH2 and subsequent b-
elimination affords reduced ¯avin and PhNyNH, which
decomposes to N2 and benzene 3Scheme 7). The reduced
¯avin is then aerobically recycled, allowing for catalytic
turnover. In the case of ¯avin 1c, the reduced ¯avin 8c is
only slowly reoxidized and can be trapped with acetic
anhydride to give 6c 3Scheme 7). For the reaction with
1,3-propanedithiol, Scheme 8 depicts the C10a-adduct
version of the C4a-adduct mechanism previously worked
out in detail in the case of normal ¯avins.27
Cyclic voltammetry and Osteryoung square-wave volt-
ammetry measurements in water were performed using a
BAS100 Electrochemical Analyzer with a three-electrode
system consisting of a glassy carbon working electrode
with sˆ6 mm, a platinum auxiliary electrode, and a Ag/
AgCl 3saturated KCl) reference electrode. The glassy
carbon electrode was polished with 0.3 and 0.05 mm
alumina powder, sonicated 3to remove polishing powder),
and washed with water. All of the solutions were prepared at
1 mM and contained 0.167 M potassium phosphate 3pH
6.01) as a supporting electrolyte. All measurements were
carried out under an atmosphere of nitrogen using N2-
purged solutions, and the potentials listed are given vs SCE.
4.3. Preparation of substituted 2-nitro-N-32-hydroxy-
ethyl)anilines 3a±c
Overall, the information revealed by studies such as
those described here and elsewhere,4,11,21 should provide
important information relevant to evaluating proposed non-
radical oxidation mechanisms mediated by ¯avoenzymes.
In a typical procedure, a solution of 4-bromo-3-nitrobenzo-
tri¯uoride 35.03 g, 18.5 mmol) and ethanolamine 33.39 g,
55.5 mmol) in 50 mL n-butyl alcohol containing solid
K2CO3 32.8 g, 20 mmol) was re¯uxed for 5±6 h with
constant stirring and then cooled. The reaction mixture
was ®ltered through a layer of Celite, and the ®ltrate was
concentrated under reduced pressure. The resulting suspen-
sion of orange crystalline solid was poured onto saturated
aqueous NaCl 350 mL) and extracted with EtOAc
32£100 mL). The organic extract was washed with saturated
4. Experimental
4.1. General experimental information
1H 3300 MHz) and 13C NMR 375 MHz) spectra were

4518
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
aqueous NaCl 33£20 mL), dried 3MgSO4), evaporated, and
puri®ed by silica gel ¯ash chromatography 32:1 CH2Cl2±
EtOAc eluant) to give 4-3tri¯uoromethyl)-2-nitro-N-32-
hydroxyethyl)aniline 33c) as an orange solid in 94% overall
yield: 1H NMR 3CDCl3) d 2.86 3br m, 1H, OH), 3.51 3q, 2H,
Jˆ5.2 Hz), 3.95 3t, 2H, Jˆ5.2 Hz), 6.96 3d, 1H, Jˆ9.0 Hz),
7.57 3dd, 1H, Jˆ9.0 and 2.2 Hz), 8.36 3s, 1H), 8.47 3br t, 1H,
NH); 13C NMR 3CDCl3) d 45.0, 60.5, 114.7, 117.5 3q, Jˆ
34.3 Hz, C±CF3), 123.6 3q, Jˆ270 Hz, CF3), 124.8, 130.8,
132.2, 147.0.
for 30 min at 58C, the reaction mixture was made basic with
50% aqueous NaOH and extracted with EtOAc 33£50 mL).
The organic layer was dried over MgSO4, ®ltered, and
concentrated under reduced pressure to provide 4b as a
1
white solid 31.46 g, 41%): H NMR 3DMSO-d6) d 3.16
3br, 2H), 3.69 3br, 2H), 4.58 3br, 2H, NH2), 4.68 3br, 1H),
4.82 3br, 1H), 6.49 3d, 2H, Jˆ8.4 Hz), 6.59 3d, 1H, Jˆ
8.4 Hz); 13C NMR 3DMSO-d6) d 45.8, 59.5, 109.2, 114.9,
115.9, 121.7, 133.9, 137.7.
4.5. Preparation of substituted 1,10-ethyleneisoalloxa-
zinium chlorides 1a±c
2-Nitro-N-32-hydroxyethyl)aniline 33a)37 was obtained in
the same manner in 54% yield based on 1-bromo-2-nitro-
1
benzene: H NMR 3CDCl3) d 2.08 3br t, 1H, OH), 3.53 3q,
To a solution of the corresponding N-3hydroxyethyl)-o-
phenylenediamine 4a±c 34.02 g, 18.3 mmol) in glacial
acetic acid 340 mL) at 508C that was thoroughly ¯ushed
with argon, was added alloxan 32.93 g, 18.3 mmol) and
boric acid 31.17 g, 19.0 mmol). After 1 h the solution was
cooled, and the yellow-orange solid formed was ®ltered,
washed with CH2Cl2 330 mL), and dried under vacuum.
Without delay, the yellow-orange solid was placed in a
round-bottom ¯ask that had been ¯ushed with N2, and thio-
nyl chloride 330 mL) was added gradually with constant
stirring. The reaction solution was stirred for 16 h at 508C
under N2 then cooled and ®ltered, and the solid was washed
with CH2Cl2 350 mL). The crude product was dissolved in a
minimum amount of 98% formic acid and re-precipitated
with ether.
2H, Jˆ5.4 Hz), 3.97 3t, 2H, Jˆ5.4 Hz), 6.67 3t, 1H,
Jˆ7.7 Hz), 6.90 3d, 1H, Jˆ8.7 Hz), 7.45 3t, 1H, Jˆ
7.7 Hz), 8.17 3d, 1H, Jˆ8.7 Hz), 8.24 3br m, 1H, NH). 13C
NMR 3CDCl3) d 45, 60.7, 113.9, 115.6, 126.9, 131.9, 136.4,
145.6.
5-Chloro-N-32-hydroxyethyl)-2-nitroaniline 33b)38 was
obtained in the same manner in 84% yield based on 2,4-
1
dichloronitrobenzene: H NMR 3DMSO-d6) d 3.41 32H),
3.63 32H), 5.00 3t, 1H, Jˆ5.2 Hz, OH), 6.68 3dd, 1H,
Jˆ8.7, 2.1 Hz), 7.14 3d, 1H, Jˆ2.1 Hz), 8.07 3d, 1H, Jˆ
8.7 Hz), 8.31 3t, 1H, Jˆ5.2 Hz, NH). 13C NMR 3DMSO-
d6) d 44.8, 58.9, 113.0, 115.2, 128.1, 129.8, 141.5, 145.9.
4.4. Preparation of substituted 2-332-hydroxyethyl)-
amino)anilines 4a±c
1,10-Ethyleneisoalloxazinium chloride 31a) was obtained in
61% yield. mp216±219 8C; 1H NMR 3CD3COOD±
CF3COOH, 1:6) d 4.94 3t, 2H, Jˆ8.9 Hz), 5.53 3t, 2H,
Jˆ8.9 Hz), 8.04±8.57 34H); 13C NMR 3CD3COOD±
CF3COOH, 1:6) d 45.9, 51.9, 117.2, 130.3, 131.9, 133.2,
134.2, 142.1, 142.2, 144.1, 146.7, 159.4. A sample of the
chloride salt 1a was converted to the perchlorate salt for
purposes of elemental analysis. Anal. Calcd for
C12H9ClN4O6: C, 42.31; H, 2.66; N, 16.45. Found: C,
42.13; H, 2.60; N, 16.35.
To a solution of 3c 35.71 g, 22.8 mmol) in 80 mL of dry
MeOH at 58C was added ammonium formate 37.2 g,
114.2 mmol) and 10% Pd on carbon 33.82 g). The reaction
was allowed to warm to 258C, where it was stirred for 1 h.
After this time, the reaction mixture was ®ltered through a
bed of Celite, and then the solids were collected and washed
with methanol. The ®ltrate was concentrated under reduced
pressure to afford a reddish residue, which was diluted with
saturated NaCl 3100 mL) and extracted twice with EtOAc
350 mL). The organic layers were combined, dried over
Na2SO4, ®ltered, and concentrated under reduced pressure
to afford 2-332-hydroxyethyl)amino)-5-3tri¯uoromethyl)-
8-Chloro-1,10-ethyleneisoalloxazinium chloride 31b) was
obtained in 56% yield. mp .3508C; 1H NMR
3CD3COOD±CF3COOH, 1:6) d 4.92 3t, 2H, Jˆ8.9 Hz),
5.49 3t, 2H, Jˆ8.9 Hz), 8.06 3d, 2H, Jˆ8.4 Hz), 8.50 3d,
1H, Jˆ8.4 Hz); 13C NMR 3CD3COOD±CF3COOH, 1:6) d
45.9, 51.9, 117.2, 130.9, 131.8, 134.2, 135.1, 140.6, 144.6,
146.6, 150.1, 159.1. Anal. Calcd for C12H8N4O2´H2O: C,
43.79; H, 3.06; N, 17.02. Found: C, 43.78; H, 2.56; N, 17.02.
1
aniline 34c) as a white powder in 80% yield: H NMR
3DMSO-d6) d 3.19 3t, 2H, Jˆ5.4 Hz), 3.65 3t, 2H, Jˆ
5.4 Hz), 4.78 3br m, 1H, OH), 4.92 3br, 2H, NH2), 4.99 3br
t, 1H, NH), 6.50 3d, 1H, Jˆ8.3 Hz), 6.84 3d, 2H, Jˆ8.3 Hz);
13C NMR 3DMSO-d6) d 45.6, 59.3, 108.0, 109.5, 114.6,
116.4 3q, Jˆ31.4 Hz, C±CF3), 125.5 3q, Jˆ270 Hz, CF3),
135.1, 139.1.
7-3Tri¯uoromethyl)-1,10-ethyleneisoalloxazinium chloride
31c) was obtained as a bright yellow powder in 67% yield.
mp193±196 8C; H NMR d 3CD3COOD±CF3COOH, 1:6)
1
2-332-Hydroxyethyl)amino)aniline 34a)39 was obtained from
3a in the same manner in 91% yield: 1H NMR 3DMSO-d6) d
3.08 3q, 2H, Jˆ5.5 Hz), 3.60 3q, 2H, Jˆ5.5 Hz), 4.34 3br t,
1H), 4.43 3s, 2H, NH2), 4.69 3t, 1H, Jˆ5.5 Hz, NH), 6.41±
6.56 34H); 13C NMR 3DMSO-d6) d 46.0, 59.6, 109.8, 114.2,
116.9, 117.7, 135.3, 136.2.
4.96 3t, 2H, Jˆ8.9 Hz), 5.58 3t, 2H, Jˆ8.9 Hz), 8.20 3d, 1H,
Jˆ8.8 Hz), 8.47 3d, 1H, Jˆ8.8 Hz), 8.81 3s, 1H); 13C NMR
3CD3COOD±CF3COOH, 1:6) d 46.2, 52.1, 122.6 3q, Jˆ
272.6 Hz, CF3), 118.9, 131.4, 131.7, 133.9, 135.4 3q, Jˆ
35.5 Hz, C CF3), 136.8, 141.1, 145.2, 146.4, 158.8.
4-Chloro-2-332-hydroxyethyl)amino)aniline
34b)
was
4.6. Preparation of N-methylalloxan
obtained according to a modi®cation of the published
method.40 To a solution of 3b 34.13 g, 19.1 mmol) and
mossy tin 36.80 g, 57.3 mmol) in water 360 mL) at 1008C
was added 35 mL conc. HCl dropwise. After being cooled
According to a modi®cation of established procedures,41
potassium chlorate 35 g, 40.8 mmol) was added slowly to
a 4.5N HCl solution 340 mL) of theobromine 3Acros

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
4519
Organics) 310 g, 55.5 mmol) over a period of 40 min at
508C. After the mixture was stirred for 10 min, a conc.
HCl 35 mL) solution of tin3II) chloride 36.5 g, 34.3 mmol)
was added dropwise over a period of 10 min at 08C. Then
the solution was cooled to 58C and the white solid was
separated rapidly by ®ltration, washed with cold water
followed by cold ether, and dried in vacuo to yield 6.1g
to give light yellow oils, which were precipitated on
addition of 5 mL of H2O. The solids were collected, washed
and dried to give light yellow solids.
1,10-Ethano-5-acetyl-1,5-dihydrolumi¯avin 36a). 54%
yield based on 1a: mp274±276 8C, dec.; UV 3DMSO)
256 nm 31 9600 M21 cm21), 294 nm 31 7937 M21 cm21);
1H NMR 3DMSO-d6) d 2.16 3s, 3H), 4.11 34H), 6.96±7.49
34H), 10.97 3s, 1H, NH); 13C NMR 3DMSO-d6) d 21.3, 43.3,
47.2, 91.6, 113.3, 122.7, 126.1, 126.4, 126.5, 136.6, 147.7,
152.3, 158.5, 171.7; EIMS m/z 3relative intensity) 284 3M1,
3.5), 242 3100), 171 359), 143 321), 129 37); HRMS 3EI) m/z
calcd for C14H12N4O3 284.0910, found 284.0909.
1
370%) of N-methylalloxantin. H NMR 3DMSO-d6) d 3.04
3s, 3H), 3.08 3s, 3H), 3.06 3s, 1H), 5.05 3s, 1H), 11.0 3s, 1H),
11.3 3s, 1H); 13C NMR 3DMSO-d6) d 26.8, 27.5, 69.5,
111.3, 148.8, 150.9, 168.5, 169.3. A suspension of the
N-methylalloxantin 36.0 g, 19.1 mmol) in water 39 mL)
was treated at 458C with concentrated nitric acid dropwise
until the solid dissolved. The solution was kept in a desic-
cator over sulfuric acid until the product crystallized. The
N-methylalloxan separated as ®ne white crystals 34.8 g,
7-3Tri¯uoromethyl)-1,10-ethano-5-acetyl-1,5-dihydrolumi-
¯avin 36c). 67% yield based on 1c: mp316±320 8C, dec.;
UV 3DMSO) lmax 258 nm 31 M21 cm21), 306 nm, 31
7571 M21 cm21); 1H NMR 3DMSO-d6) d 2.20 3s, 3H,
CH3), 3.95±4.14 34H, CH2CH2), 7.19±7.77 33H), 11.09 3s,
1H, NH); 13C NMR 3DMSO-d6) d 21.1, 43.4, 47.3, 91.7,
113.9, 122.5 3q, Jˆ36.3 Hz), 122.7, 123.6, 124.0 3q, Jˆ
272.6 Hz, CF3), 126.7, 140.1, 147.7, 151.9, 158.4, 171.9;
EIMS m/z 3relative intensity) 353 3[M11]1, 0.3), 352
3M1, 1.5), 324 30.7), 310 3100); HRMS 3EI) m/z calcd for
C15H11N4O3F3 352.0784, found 352.0781.
1
72%). mp150±151 8C; H NMR 3DMSO-d6) d 3.06 3s,
3H, ±CH3[hydrate form]), 3.12 3s, 3H, ±CH3[keto form]),
11.4 3s, 1H, ±NH[hydrate form]), 11.9 3s, 1H,± NH[keto
form]); 13C NMR 3DMSO-d6) 3signals from both hydrate
and keto forms) d 27.5, 85.2, 150.1, 155.8, 156.9, 165.7,
168.0, 168.9. Direct isolation of N-methylalloxan from
oxidation of theobromine 3bypassing N-methylalloxantin)
was unsuccessful.
4.7. Preparation of 3-methyl-7-3tri¯uoromethyl)-1,10-
ethyleneisoalloxazinium chloride 31d)
8-Chloro-1,10-ethano-5-acetyl-1,5-dihydrolumi¯avin 36b).
41% yield based on 1b; the above typical procedure led to
the production of the N5-acetyl derivative 36b) as a mixture
of two inseparable diastereomers in the ratio of 5:3.
mp304±310 8C, dec.; UV 3DMSO) lmax 261 31
The same procedure described above for 1a±c was followed
except for the used of N-methylalloxan instead of alloxan.
In this case the intermediate 3-methyl-7-3tri¯uoromethyl)-
10-32-hydroxyethyl)isoalloxazine generated prior to SOCl2-
mediated cyclization was characterized: mp234±235.5 8C;
1H NMR 3CD3COOD±CF3COOH, 1:6) d 3.51 3s, 3H), 4.42
3t, 2H, Jˆ4.7 Hz), 5.21 3t, 2H, Jˆ4.7 Hz), 8.38 32H), 8.66
3s, 1H); 13C NMR 3CD3COOD±CF3COOH, 1:6) d 29.1,
51.5, 60.1, 119.5, 123.0 3q, J[19F,13C]ˆ272.6 Hz, CF3),
131.3, 135.1, 133.7 3q, J[19F,13C]ˆ35.5 Hz, C-7), 134.0,
135.7, 139.0, 145.2, 151.6, 159.3. Following cyclization
and recrystallization, 1d was obtained in 60% yield: mp
186±1898C; lmax 3CH3CN) 336, 388 nm; 1H NMR
3CD3COOD±CF3COOH, 1:6) d 3.48 3s, 3H), 4.91 3t, 2H,
Jˆ9.1 Hz), 5.46 3t, 2H, Jˆ9.1 Hz), 8.11 3d, 1H, Jˆ8.8 Hz),
8.39 3d, 1H, Jˆ8.8 Hz), 8.73 3s, 1H); 13C NMR
3CD3COOD±CF3COOH, 1:6) d 29.5, 46.9, 51.7, 118.8,
131.4, 136.4, 122.8 3q, J[19F,13C]ˆ272.5 Hz, CF3), 131.8,
133.6, 134.9 3q, J[19F,13C]ˆ35.3 Hz, C-7), 141.1, 144.0,
146.8, 159.0; HRMS 3FAB) m/z calcd for C14H10N4O2F3
323.0756, found 323.0751. A sample of the chloride salt
1d was converted to the perchlorate salt for purposes of
elemental analysis. Anal. Calcd for C14H10ClF3N4O6´H2O:
C, 38.15; H, 2.74; N, 12.71. Found: C, 37.67; H, 2.63; N,
12.65.
1
17,800 M21 cm21), 310 nm 31 9013); H NMR 3DMSO-
d6) d 1.90 and 2.15 3s, 3H), 3.93±3.95 and 4.05±4.09
34H), 7.00±7.67 33H), 11.03 and 11.09 3s, 1H); 13C NMR
3DMSO-d6) d 20.9 and 21.2, 43.3, 46.8 and 47.3, 90.2 and
91.8, 113.3 and 113.8, 122.2 and 122.5, 123.6 and 125.3,
124.7 and 127.3, 130.6 and 130.9, 136.6 and 138.1, 147.2
and 147.6, 149.3 and 151.9, 157.9 and 158.4, 171.8 and
172.2; EIMS m/z 3relative intensity) 318 3M1, 0.9), 276
383), 242 39), 204 330), 177 326), 143 35); HRMS 3EI) m/z
calcd for C14H11N4O3Cl 318.0520, found 318.0522.
4.9. Reaction of 8-chloro-1,10-ethyleneisoalloxazinium
chloride 31b) with thiophenol
A solution of freshly distilled thiophenol 3500 mM) and 1b
350 mM) in CH3CN was monitored by repetitive scan
UV±Vis spectrophotometry in the range of 250±600 nm
at 258C.
4.10. Reduction of ¯avinium salt 1c under anaerobic
conditions by dithionite, 1,3-propanedithiol, or
hydrazine
4.8. Preparation of substituted 1,10-ethano-5-acetyl-1,5-
dihydrolumi¯avins 36a±c)
To a suspension of 1c 38.1 mg, 0.023 mmol) in 0.5 mL of
deaerated CH3CN was added under argon either Na2S2O4
348 mg, 0.023 mmol) in 80 mL of H2O, or 1,3-propane-
dithiol 32.4 mL, 0.023 mmol) or anhydrous hydrazine
30.73 mL, 0.023 mmol) in 80 mL of CH3CN. In either
case, an instantaneous color change occurred from green
to yellow. Without delay, the solutions were concentrated
in vacuo to give a yellow residue, to which was added
0.5 mL of deaerated CF3COOH and 0.6 mL of deaerated
According to a published procedure,12 the isoalloxazinium
salts 1a±c were allowed to react with Zn 30.27 g per mmol)
and 3CH3CO)2O 35.9 g per mmol) in CF3COOH solvent
with heating at re¯ux for 1 h 31a and 1b) or stirring at
room temperature for 5 min 31c and 1d). The inorganic
material was ®ltered off, and the ®ltrates were evaporated

4520
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
acetic anhydride under argon 3an instantaneous color
change occurred from yellow to red and then again to
yellow). After being stirred for 5 min at room temperature,
the reaction mixture was directly evaporated under reduced
pressure. The product was identical to N5-acetyl derivative
6c prepared above.
4.13. Kinetics of oxidation of benzylamines by 7-CF3
¯avin 1c
The rates of oxidation of benzylamine and its N-methyl and
N,N-dimethyl derivatives by 1c were determined by
following the decrease of the A370 3anion form of 1c) in a
closed cuvette. In a typical experiment, the cuvette
contained a solution of amine/amine´HCl 310:1) dissolved
in CH3CN. The reactions were initiated by adding 30 mL of
a 0.05 mM solution of 1c in DMSO to 2.97 mL of 1 mM
amine/amine´HCl 310:1), pre-equilibrated in a thermostatted
holder at 508C.
4.11. Preparation of substituted 1,10-ethano-1, 5-
dihydrolumi¯avins 8c and 8d
The reaction of ¯avinium salts 1c or 1d 335 mg) and N-ben-
zyl-5,6-dihydrophenanthridine21,42 335 mg) were dissolved
together in 1 mL of CH3CN±CH3OH 34:1) under argon.
Following complete dissolution, a bright yellow solid
emerged immediately. The solid was collected by centri-
fugation and washed with methanol to afford the reduced
¯avin.
4.14. Product analysis of the oxidation of benzylamines
A mixture of ¯avinium salt 1a, 1c or 1d 310 mM), benzyl-
amine 355 mM) and benzylamine hydrochloride 35 mM) in
dry CH3CN 320 mL) was stirred at 508C for 2 days in the
dark under argon with exposure to air for 5 min after 6 h and
again after 12 h. The reactions were quenched with 1N HCl.
Benzaldehyde was removed by azeotropic distillation and
quanti®ed as its 2,4-dinitrophenylhydrazine derivative.
7-Tri¯uoro-1,10-ethylene-1,5-dihydroisoalloxazine 38c). 1H
NMR 3DMSO-d6): 3.64 3t, 2H, Jˆ7.4 Hz), 3.99 3t, 2H,
Jˆ7.4 Hz), 6.33 3d, 1H, Jˆ8.0 Hz), 6.59 3s, 1H), 6.74 3d,
1H, Jˆ8.0 Hz), 7.01 3s, 1H), 10.60 3s, 1H). HRMS 3EI) m/z
calcd for C13H9F3N4O2 310.0678, found 310.0674.
4.15. Preparation of tris31,10-phenanthroline)iron3III)
perchlorate monohydrate 3TPIP)
7-Tri¯uoro-3-methyl-1,10-ethylene-1,5-dihydroisoalloxazine
1
38d). H NMR 3DMSO-d6): 3.06 3s, 3H), 3.64 3t, 2H, Jˆ
Fe3phen)33ClO4)3 was prepared by the method of Schilt and
Taylor.43 Chlorine gas was introduced into a solution of
Fe3NH4)23SO4)26´H2O 30.3 g, 0.76 mmol), phenanthroline
monohydrate 30.6 g, 3.0 mmol), H2SO4 30.2 mL) and H2O
3100 mL) until the color changed from red to light blue.
NaOH 30.12 g) and 70% HClO4 30.43 g) were mixed
together, the resulting concentrated NaClO4 solution was
added into the reaction solution, and the mixture was cooled
in an ice bath. The blue crystalline product was washed
three times with cold water and dried in vacuo at room
temperature to give TPIP 0.40 g 330%). The IR spectrum
was identical to that reported.43
7.3 Hz), 4.04 3t, 2H, Jˆ7.3 Hz), 6.33 3d, 1H, Jˆ7.8 Hz),
6.62 3s, 1H), 6.73 3d, 1H, Jˆ7.8 Hz), 7.02 3s, 1H). HRMS
3EI) m/z calcd for C14H11F3N4O2 324.0834, found
324.0844.
4.12. Kinetics of phenylhydrazine reduction of ¯avinium
salts 1a±c
The rates of reduction of ¯avinium salts by PhNHNH2 were
determined by measuring the decrease in the long wave-
length ¯avin absorption 31a, 356 nm; 1b, 368 nm; 1c,
356 nm). The cuvettes contained a solution of 0.67 mM
PhNHNH2´HCl dissolved in 165 mM potassium phosphate
buffer, and the reaction was initiated by adding 0.03 mL of a
0.05 mM solution of the ¯avinium salt in DMSO. Three
different conditions were employed: 3a) use of a tight-®tting
Te¯on stopper to retard continuous exposure to atmospheric
O2; 3b) purging of solutions with argon prior to initiation of
reaction, and continuous bubbling of argon 3and inter-
mittently air) from a needle inlet along the side of the
cuvette; 3c) use of an in situ `O2 scrubber' system, consisting
of glucose 316 mM), 20 mg of glucose oxidase, and 3 mg of
catalase per 3 mL reaction solution.
4.16. Cyclopropylamines
N-Cyclopropylbenzylamine was prepared44 by NaBH4
reduction of the Schiff base formed from benzaldehyde
and cyclopropylamine. N-Benzylidenecyclopropylamine:
1H NMR 3CDCl3) d 0.99 3m, 2H), 1.01 3m, 2H), 3.06 3m,
1H), 7.44 33H), 7.76 3m, 2H), 8.46 3s, 1H); 13C NMR
3CDCl3) d 9.06, 42.1, 127.7, 128.7, 130.2, 136.7, 158.4.
1
N-Cyclopropylbenzylamine: H NMR 3CDCl3) d 0.49 3m,
4H), 1.88 3s, 1H, NH), 2.21 3m, 1H), 3.90 3s, 2H), 7.38 35H);
13C NMR 3CDCl3) d 6.64, 30.2, 53.9, 126.9, 128.3, 128.5,
140.8. N-Cyclopropyl-1-phenethylamine45 was also pre-
pared in the same way, except with acetophenone. N-1-
Phenethylidenecyclopropylamine: 1H NMR 3CD3CN) d
0.84 3m, 2H), 0.94 3m, 2H), 2.32 3s, 3H), 3.17 3m, 1H),
7.37 33H), 7.78 3m, 2H); 13C NMR 3CD3CN) d 9.61, 15.8,
34.7, 127.2, 129.0, 129.9, 142.1, 163.5. N-Cyclopropyl-1-
phenethylamine: 1H NMR 3CDCl3) d 0.39 3m, 4H), 1.42 3d,
3H, Jˆ6.7 Hz), 1.83 3s, 1H, NH), 2.01 3m, 1H), 3.90 3q, 1H,
Jˆ6.7 Hz), 7.35 35H); 13C NMR 3CDCl3) d 6.58, 23.8, 29.0,
58.5, 126.7, 126.9, 128.4, 146.2. 1-3Phenylcyclopropyl)-
methylamine was prepared by LiAlH4 reduction of
1-phenylcyclopropanecarbonitrile by the reported method.46
In order to verify that the lag time seen for PhNHNH2
reduction of ¯avin 1a was due to oxidative recycling of
reduced ¯avin by the O2 content of the sample, we repeated
the experiment using periodic intentional bubbling of O2
following initial complete degassing. The cuvette reaction
content was established as above except that all stock solu-
tions were degassed by bubbling argon. The reaction was
monitored as usual at 356 nm under a controlled atmosphere
condition using bubbling of air into the cuvette for 2 s at a
¯ow rate of 5 mL/min at different time points 3360, 990 and
3480 s).

W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
4521
7. 3a) Shinkai, S.; Kawase, A.; Yamaguchi, T.; Manabe, O.;
Wada, Y.; Yoneda, F. J. Am. Chem. Soc. 1989, 111, 4928±
4935. 3b) Breinlinger, E.; Niemz, A.; Rotello, V. M. J. Am.
Chem. Soc. 1995, 117, 5379±5380. 3c) Cuello, A. O.;
McIntosh, C. M.; Rotello, V. M. J. Am. Chem. Soc. 2000,
122, 3517±3521. 3d) Hasford, J. J.; Kemnitzer, W.; Rizzo,
C. J. J. Org. Chem. 1997, 62, 5244±5245.
4.17. Product analysis of the oxidation of benzylamines
and cyclopropylamines by TPIP
To a solution of Fe3phen)33ClO4)3 356 mM) in dry CH3CN
310 mL) at 258C was added the amine 328 mM). After being
stirred for 1 h, the reaction mixture was acidi®ed with 1 mL
of 2,4-dinitrophenylhdyrazine solution 3144 mM in a 3:4:14
mixture of H2SO4, H2O, and EtOH) and concentrated under
reduced pressure. Flash chromatography 33£3 cm SiO2,
CH2Cl2 or EtOAc/CHCl3, 1:2) afforded the desired
2,4-dinitrophenylhydrazones. Quantitative aldehyde/ketone
analyses were carried out by dissolving the hydrazones and
a known weight of hexamethylbenzene in DMSO-d6 and
8. Naani Jr, E. J.; Sawyer, D. T.; Ball, S. S.; Bruice, T. C. J. Am.
Chem. Soc. 1981, 103, 2797±2802.
9. MuÈller, F.; Massey, V. J. Biol. Chem. 1969, 244, 4007±4016.
10. Tamagaki, S.; Sasaki, M.; Tagaki, W. Bull. Chem. Soc. Jpn.
1989, 62, 159±163.
11. Li, W.-S.; Sayre, L. M. Tetrahedron 2001, 57, 4523±4536.
12. Knappe, W. R. Chem. Ber. 1975, 108, 2422±2438.
13. Kuhn, R.; Weygand, F. Ber. Deut. Chem. Ges. 1935, 68,
1282±1288.
1
calculating the molar ratio using H NMR integrals. In
some cases, the 2,4-dinitrophenylhydrazones were
collected, concentrated, dried and weighed. The nature of
the hydrazone derivatives were con®rmed by their inde-
pendent synthesis. The 2,4-dinitrophenylhydrazone of
1-phenylcyclopropanecarboxaldehyde47 had 1H NMR
3DMSO-d6) d 1.33 3m, 2H), 1.44 3m, 2H), 7.31±7.39
35H), 7.62 3d, 1H, Jˆ9.4 Hz), 8.18 3s, 1H), 8.29 3d, 1H,
Jˆ9.4 Hz), 8.83 3s, 1H), 11.4 3s, 1H).
14. Huff, J. R.; King, S. W.; Saari, W. S. J. Org. Chem. 1982, 47,
582±585.
15. Hemmerich, P.; Veeger, C.; Wood, H. C. S. Angew. Chem.,
Int. Ed. Engl. 1965, 4, 671±687.
16. Yoneda, F.; Sakuma, Y.; Ichiba, M.; Shinomura, K. J. Am.
Chem. Soc. 1976, 98, 830±835.
17. Shinkai, S.; Harada, A.; Ishikawa, Y.; Manabe, O.; Yoneda, F.
J. Chem. Soc. Perkin. Trans. 2 1982, 125±133.
18. Hemmerich, P.; Ghisla, S.; Hartmann, U.; MuÈller, F. In
Flavins and Flavoproteins, Kamin, H., Ed.; University Park
Press: Baltimore, MD, 1971; pp 83±105.
4.18. Anaerobic reactions of 1c with hydrazines, amines,
and thiols monitored by H NMR spectroscopy
1
Stock solutions of 1c 340±80 mM ®nal concentration) in
degassed DMSO-d6 and of the hydrazine, amine, or thiol
in either degassed DMSO-d6 or CD3CN 340±80 mM ®nal
concentration) were added to an NMR tube containing
degassed DMSO-d6 or CD3CN under argon at 258C, and
the spectra were recorded at various time intervals. The
reaction of 1c with 1,3-propanedithiol was also followed
19. Raibekas, A. A.; Ramsey, A. J.; Jorns, M. S. Biochemistry
1993, 32, 4420±4429.
20. De Kok, A.; Veeger, C.; Hemmerich, P. In Flavins and Flavo-
proteins, Kamin, H., Ed.; University Park Press: Baltimore,
MD, 1971; pp 63±81.
21. Zhang, N., Sayre, L. M. In preparation.
22. The reduced compound N1,N10-ethylene-7,8-dimethyl-1,5-
dihydroisoalloxazine has been isolated but without characteri-
zation.9
1
by H NMR spectroscopy in D2O.
23. Patek, D. R.; Hellerman, L. J. Biol. Chem. 1974, 249, 2373±
2380.
Acknowledgements
24. Kennedy, S. H. J. Psychiatry Neurosci. 1997, 22, 127±131.
25. 3a) Kenney, W. C.; Nagy, J.; Salach, J. I.; Singer, T. P. In
Monoamine Oxidase: Structure, Function, and Altered Func-
tions, Singer, T. P., Von Korff, R. W., Murphy, D. L., Eds.;
Academic Press: New York, 1979; pp 25±37. 3b) Nagy, J.;
Kenney, W. C.; Singer, T. P. J. Biol. Chem. 1979, 254, 2684±
2688.
We are grateful for support of this research by the National
Institutes of Health through grant GM 48812. We also thank
Professor Fred Urbach and his groupfor use of their
electrochemistry equipment and for helpful advice.
References
26. Kosower, E. M. Acc. Chem. Res. 1971, 4, 193±198.
27. 3a) Loechler, E. L.; Hollocher, T. C. J. Am. Chem. Soc. 1975,
97, 3235±3237. 3b) Loechler, E. L.; Hollocher, T. C. J. Am.
Chem. Soc. 1980, 102, 7312±7321. 3c) Loechler, E. L.; Hollo-
cher, T. C. J. Am. Chem. Soc. 1980, 102, 7322±7327.
3d) Yokoe, I.; Bruice, T. C. J. Am. Chem. Soc. 1975, 97,
450±451. 3e) Yano, Y.; Nakazato, M.; Ohya, E. J. Chem.
Soc. Perkin Trans. 2 1985, 77±81. 3f) Nagata, T.; Fujimori,
K.; Oae, S. Heteroatom Chem. 1992, 3, 529±534.
28. Moore, E. G.; Ghisla, S.; Massey, V. J. Biol. Chem. 1979, 254,
8173±8178.
1. Singer, T. P. Chemistry and Biochemistry of Flavins, 3; CRC
Press: Boca Raton, FL, 1990.
2. Silverman, R. B. Acc. Chem. Res. 1995, 28, 335±342.
3. Walker, M. C.; Edmondson, D. E. Biochemistry 1994, 33,
7088±7098.
4. 3a) Kim, J.-M.; Bogdon, M. A.; Mariano, P. S. J. Am. Chem.
Soc. 1993, 115, 10591±10595. 3b) Kim, J.-M.; Hoegy, S. E.;
Mariano, P. S. J. Am. Chem. Soc. 1995, 117, 100±105.
3c) Hoegy, S. E.; Mariano, P. S. Tetrahedron 1997, 53,
5027±5046.
5. 3a) Simpson, J. T.; Krantz, A.; Lewis, F. D.; Kokel, B. J. Am.
Chem. Soc. 1982, 104, 7155±7161. 3b) Kim, J.-M.; Bogdon,
M. A.; Mariano, P. S. J. Am. Chem. Soc. 1991, 113, 9251±
9257. 3c) Kim, J.-M.; Cho, I.-S.; Mariano, P. S. J. Org. Chem.
1991, 56, 4943±4955.
29. The N3-H pKa for the 7,8-dimethyl analog of 1a has been
reported to be 5.9,9 so that the 7-CF3 analog 1c should be a
little lower.
30. Mager, H. I. X.; Addink, R. Tetrahedron Lett. 1979, 37, 3545±
3548.
6. Ramsay, R. R.; Sablin, S. O.; Singer, T. P. Progr. Brain Res.
1995, 106, 33±39.
31. Mager, H. I. X. Tetrahedron 1977, 33, 981±989.
32. Bruice, T. C.; Chan, T. W.; Taulane, J. P.; Yokoe, I.; Elliott,

4522
W.-S. Li et al. / Tetrahedron 57 )2001) 4507±4522
D. L.; Williams, R. F.; Novak, M. J. Am. Chem. Soc. 1977, 99,
6713±6720.
39. Reported previously: Karrer, P.; Schlittler, E.; Pfaehler, K.;
Benz, F. Helv. Chim. Acta. 1934, 17, 1516.
33. Dudley, K. H.; Hemmerich, P. J. Org. Chem. 1967, 32, 3049±
3054.
40. Natelson, S.; Gottfried, S. P. J. Am. Chem. Soc. 1939, 61,
1001.
41. Biltz, H. Ber. Deut. Chem. Ges. 1912, 45, 3659±3675.
42. Ostovic, D.; Lee, I.-S. H.; Roberts, R. M. G.; Kreevoy, M. M.
J. Org. Chem. 1983, 50, 4206±4211.
34. Sayre, L. M.; Singh, M. P.; Kokil, P. B.; Wang, F. J. Org.
Chem. 1991, 56, 1353±1355.
35. 3a) Gassman, P. G. Acc. Chem. Res. 1970, 3, 26±33.
3b) Hiyama, T.; Koide, H.; Nozaki, H. Bull. Chem. Soc. Jpn.
1975, 48, 2918±2921. 3c) Vaidyanathan, G.; Wilson, J. W.
J. Org. Chem. 1989, 54, 1815±1820.
43. Schilt, A. A.; Taylor, R. C. J. Inorg. Nucl. Chem. 1959, 9,
211±221.
44. Fuller, R. W.; Molloy, B. B.; Day, W. A.; Roush, B. W.;
Marsh, M. M. J. Med. Chem. 1973, 16, 101±106.
45. Silverman, R. B. Biochemistry 1984, 23, 5206±5213.
46. Kaiser, C.; Lester, B. M.; Zirkle, C. L.; Burger, A.; Davis,
C. S.; Delia, T. J.; Zirngibl, L. J. J. Med. Pharm. Chem.
1962, 5, 1243±1265.
36. Perrin, D. D.; Armarego, W. L. F.; Perin, D. R. Puri®cation of
Laboratory Chemicals, Pergamon Press: Elmsford, NY, 1966.
37. Reported previously: Threadgill, M. D. Synthetic Commun.
1985, 15, 1101±1105.
38. Reported previously: Huff, J. R.; King, S. W.; Saari, W. S. J.
Org. Chem. 1982, 47, 582±585.
47. Schuster, D. I.; Roberts, J. D. J. Org. Chem. 1962, 27, 51±53.