The mechanism of hydrolysis of aryl ether derivatives of 3-hydroxymethyltriazenes

Article abstract of DOI:10.1002/ejoc.200400582

1-Aryl-3-aryloxymethyl-3-methyltriazenes hydrolyse to the corresponding anilines and phenols by specific-acid-catalysed, general-acid-catalysed and pH-independent mechanisms. All compounds studied exhibit specific- and general acid catalysis, though for 5a general acid catalysis was not observed below a pH of approximately 4, while for compounds 5e,f, such catalysis was absent above a pH of approximately 5. The pH-independent pathway is observed only for those compounds, 5d-f, that contain good aryloxy nucleo-fugic groups. The specific-acid-catalysed pathway is supported by a solvent deuterium isotope effect (SDIE) of 0.64, consistent with a mechanism involving protonation of the substrate followed by rate-determining unimolecular decomposition of the protonated species. The k_H+ values gave rise to a Hammett p value of -0.93, reflecting the competing effect of the substituents on the protonation of the substrate and the cleavage of the aryl ether. Correlation of k _H⁺ with the pK_a of the phenol leaving group affords a β_1g of 0.3. Decomposi tion of the protonated intermediate proceeds via a triazenyliminium ion that can be trapped by methanol. The general-acid-catalysed process exhibits an SDIE of 1.43 and Hammett p values of 0.49, 0.84 and 1.0 for reactions catalysed by chloroacetic, formic and acetic acids, respectively. Correlation of K_A with the pK_a of the acid gave Bronsted a values that diminish from 0.6 for O-aryl systems that are poor nucleofuges (5a, b) to 0.2 for the best nucleofuge (5f), reflecting the different extents of proton transfer required to expel each phenol. Compounds containing powerful nucleofuges exhibit a pH-independent reaction that has an SDIE of 1.1, a Hammett p value of 3.4 and a Bronsted β_1g value of 1.4. These imply a mechanism involving displacement of the aryloxide leaving group to form a triazenyliminium ion intermediate that again was trapped as a methyl ether. Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim, Germany, 2005).

Full text of DOI:10.1002/ejoc.200400582

FULL PAPER
The Mechanism of Hydrolysis of Aryl Ether Derivatives of
3-Hydroxymethyltriazenes
Emília Carvalho,[a] Ana Paula Francisco,[a] Jim Iley,*[b] and Eduarda Rosa*[a]
Keywords: Prodrug / Hydrolysis / Kinetics
1-Aryl-3-aryloxymethyl-3-methyltriazenes hydrolyse to the
tion of the protonated intermediate proceeds via a triazenyl-
corresponding anilines and phenols by specific-acid-cata- iminium ion that can be trapped by methanol. The general-
lysed, general-acid-catalysed and pH-independent mecha-
nisms. All compounds studied exhibit specific- and general
acid catalysis, though for 5a general acid catalysis was not
observed below a pH of approximately 4, while for com-
pounds 5e,f, such catalysis was absent above a pH of approx-
imately 5. The pH-independent pathway is observed only for
those compounds, 5d–f, that contain good aryloxy nucleo-
fugic groups. The specific-acid-catalysed pathway is sup-
ported by a solvent deuterium isotope effect (SDIE) of 0.64,
consistent with a mechanism involving protonation of the
acid-catalysed process exhibits an SDIE of 1.43 and Hammett
ρ values of 0.49, 0.84 and 1.0 for reactions catalysed by chlo-
roacetic, formic and acetic acids, respectively. Correlation of
kA with the pKa of the acid gave Brønsted α values that di-
minish from 0.6 for O-aryl systems that are poor nucleofuges
(5a,b) to 0.2 for the best nucleofuge (5f), reflecting the dif-
ferent extents of proton transfer required to expel each phe-
nol. Compounds containing powerful nucleofuges exhibit a
pH-independent reaction that has an SDIE of 1.1, a Hammett
ρ value of 3.4 and a Brønsted βlg value of 1.4. These imply a
substrate followed by rate-determining unimolecular decom- mechanism involving displacement of the aryloxide leaving
position of the protonated species. The kH+ values gave rise
to a Hammett ρ value of –0.93, reflecting the competing effect
of the substituents on the protonation of the substrate and
the cleavage of the aryl ether. Correlation of kH+ with the pKa
group to form a triazenyliminium ion intermediate that again
was trapped as a methyl ether.
(© Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
of the phenol leaving group affords a βlg of 0.3. Decomposi- Germany, 2005)
Introduction
The synthetic and biological utility of the triazene group
is well-documented.[1] In particular, the biological action of
the anticancer 1-aryl-3,3-dimethyltriazenes 1, including a
clinically used drug for malignant melanoma, dacarbazine
(1a), is a consequence of their capacity to alkylate DNA.[2]
These compounds suffer metabolic oxidation by cyto-
chrome P450 enzymes to give hydroxymethyltriazenes 2,
which, by loss of formaldehyde, generate the cytotoxic
monomethyltriazenes 3 (Scheme 1).[3] These are known al-
kylating agents, capable of methylating DNA and RNA.[4,5]
Prodrugs of the 3-hydroxymethyl-3-methyltriazene 2 are
especially attractive as they bypass the requirement for
metabolic activation, an inefficient process in humans.[6]
Among the derivatives of 2 studied are esters,[7,8] particu-
larly more recently the “cascade release molecules” that
concomitantly target DNA and epidermal growth factor re-
ceptor tyrosine kinase,[9] and alkyl ethers.[10,11] Our investi-
gations of the hydrolysis of 3-alkyloxymethyl-3-alkyl-1-aryl-
[a] CECF, Faculdade de Farmácia,
Scheme 1. Metabolic pathway for dimethyltriazenes.
Avenida das Forças Armadas, 1649-019 Lisboa, Portugal
[b] Chemistry Department, The Open University,
Milton Keynes, MK7 6AA, U. K.
triazenes 4 showed these compounds to behave like acetals
of formaldehyde, hydrolysing in aqueous buffers by forma-
E-mail: j.n.iley@open.ac.uk
2056
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/ejoc.200400582
Eur. J. Org. Chem. 2005, 2056–2063

Mechanism of Hydrolysis of Aryl Ether Derivatives of 3-Hydroxymethyltriazenes
FULL PAPER
tion of a triazenyliminium ion intermediate by a specific-
acid-catalysed pathway.[11] However, in general, the alkoxy-
methyltriazene derivatives 4 are too stable toward hydrolysis
and metabolism[12] to be considered satisfactory prodrugs.
Acetals that contain a phenolic or other weakly basic
leaving group undergo hydrolysis more readily; for such
compounds, as well as the specific-acid-catalysed reaction
observed for the alkyl analogues, general acid catalysis and
pH-independent processes have been observed.[13] Thus,
aryl, rather than alkyl, ether derivatives 5 of the hydroxy-
methyltriazene metabolite would appear more attractive
candidates as prodrugs. Indeed, given the known cytotoxic
activity of 4-methoxyphenol towards melanogenic cells,[14]
compounds 5 offer the potential for developing mutual pro-
drug systems as antimelanoma agents.
(1)
The reactions are easily monitored using changes in the
UV spectra by following the loss of the starting material or
formation of the product. However, in solutions of pH Ͼ 7
the intermediate formation of the monomethyltriazene 3 is
observed; in these solutions the reactions were monitored
using the wavelength at which there is an isosbestic point
for the subsequent decomposition of 3 to the corresponding
aniline. Pseudo-first-order rate constants, kobs, for the hy-
drolysis of 5a–f were determined in HCl and NaOH solu-
tions and also in aqueous buffers at various pH values
(using several buffer concentrations for each pH). Some ex-
amples of these results for compounds 5a and 5f are shown
in Table 1, and these demonstrate that the dependence of
kobs upon proton and buffer concentration varies with the
pH, the nature of the buffer material and the structure of
the substrate. We shall discuss each in turn.
The aryloxymethyltriazenes 5 are known compounds.[15]
Compounds 5 are specific examples of 1-aryl-3,3-dialkyltri-
azenes; as a class, these compounds are generally stable ex-
cept when they are subjected to thermal or photochemical
degradation, or to the action of acid.[16] Under the latter
conditions, these compounds undergo protonation at the
N3 atom followed by scission of the N2–N3 bond to form a
secondary amine and a diazonium ion.[17] Indeed, this is a
reaction that has been used to release dialkylamines pro-
tected as a triazene functionality.[1] However, an investiga-
tion by Vaughan and coworkers[18] showed that aryloxy-
methyltriazenes behaved more like acetals than 3,3-dial-
kyltriazenes. Hydrolysis of these compounds was found to
be specific acid and buffer catalysed and that there is also a
spontaneous reaction for some compounds. Unfortunately,
these investigators did not study the reaction over a suffic-
iently wide pH range, nor did they examine the nature of
the buffer-catalysed process. Consequently, the reported
structure–activity relationships and solvent isotope effect
are composite values for all the reaction pathways involved.
Therefore, we have synthesised the aryloxymethyltriazenes
5a–f and have carried out an investigation of their hydroly-
sis in aqueous buffers to better understand the hydrolytic
mechanisms that operate for these compounds.
pH-Rate Profiles
Using the values of kobs in HCl and NaOH solutions,
together with the intercepts of plots of kobs vs. [buffer], the
pH-rate profiles shown in Figure 1 can be constructed.
Figure 1. pH-Rate profiles for the hydrolysis of compounds 5a–f:
filled triangle, 5a; circle, 5b; filled circle, 5c; square, 5d; triangle, 5e;
filled square, 5f.
Two different behaviours are observed. Compounds con-
taining electron-donating substituents on the aryl ether
ring, 5a,b, or with no substituent, 5c, display a specific-acid-
catalysed reaction below pH of about 7.5; above this pH
these compounds are stable, no decomposition being mea-
Results and Discussion
The aryloxymethyltriazenes 5 hydrolyze in aqueous solu- surable after 60 days. In contrast, compounds with electron-
tions to the corresponding anilines and phenols [Equa- attracting groups, 5d–f, exhibit the specific-acid-catalysed
tion (1)].
reaction together with a pH-independent process; the pH
Eur. J. Org. Chem. 2005, 2056–2063
www.eurjoc.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2057

E. Carvalho, A. P. Francisco, J. Iley, E. Rosa
Table 1. Pseudo-first-order rate constants, kobs, for the hydrolysis of 5a and 5f in aqueous buffers at 25 °C.
FULL PAPER
5a
5f
Buffer
HCl
[Buffer]/moldm–3
pH (pD) kobs/10–5 s–1
Buffer
HCl
[Buffer]/moldm–3
pH (pD) kobs/10–5 s–1
5×10–7
3.4×10–6
5.6×10–6
2.5×10–5
6.92×10–5
0.000145
0.000575
0.001
6.30
5.47
5.25
4.60
4.16
3.84
3.24
3.00
5.33
4.55
3.86
3.72
3.50
2.69
2.78
2.54
2.58
2.72
3.26
3.25
3.23
3.22
0.3
2.03
3.25
12.6
24.1
79.9
244
587
14.4
52.8
311
381
700
589
448
701
671
511
251
255
241
230
0.0005
0.001
0.005
0.01
0.03
0.1
3.30[a]
3.00[a]
2.30[a]
2.00[a]
1.51[a]
0.90[a]
0.30[a]
294
324
616
873
1620
1880
1750
0.5
DCl
1.17×10–5
7.14×10–5
0.00035
0.000474
0.0008
0.01
0.05
0.1
0.2
0.3
0.075
0.15
0.225
0.30
ClCH2CO2H
HCO2H
ClCH2CO2H
HCO2H
0.01
0.05
0.1
0.2
0.3
0.0375
0.075
0.15
0.3
2.99
2.87
2.83
2.84
2.88
3.29
3.30
3.29
3.27
3.25
341
404
507
618
681
402
433
552
649
804
0.4
CH3COOH
0.0375
0.15
0.225
0.30
0.01
0.05
0.10
0.20
0.30
0.0875
0.175
0.28
0.002
0.005
0.01
0.0025
0.01
0.0125
0.025
0.01
0.05
0.1
4.18
4.14
4.17
4.16
4.56
4.56
4.58
4.61
4.62
5.43
5.45
5.51
6.63
6.68
6.69
7.25
7.23
7.32
7.33
6.37
6.34
6.38
6.35
6.30
7.11
7.13
7.22
7.10
7.12
24.0
26.4
26.3
26.8
12.5
13.7
13.9
15.5
16.6
2.58
3.31
4.05
–
H2PO4
0.183
0.182
0.203
0.051
0.057
0.062
0.073
0.285
0.315
0.338
0.382
0.415
0.049
0.063
0.078
0.102
0.122
0.0125
0.025
0.05
0.1
6.53
6.54
6.58
6.63
6.71
377
346
422
325
400
0.2
imidazole
morpholine
piperazine
NaOH
0.05
0.1
0.2
0.025
0.1
0.15
0.0005
0.001
0.1
8.63
8.59
8.66
9.85
9.84
265
294
314
297
288
285
293
264
311
0.2
0.3
0.01
0.05
0.1
0.2
0.3
9.88
10.5[a]
11.5[a]
13[a]
[a] Calculated.
of incursion of the latter depends upon the substituent in sis (Table 2) were obtained from plots of kobs vs. [H+]. For
the aryl ether, but occurs around pH 3–4.
compound 5a, the reaction was also studied in DCl and
several buffers in D2O, allowing determination of the corre-
+
sponding kD value; the solvent deuterium isotope effect,
The Specific-Acid-Catalysed Reaction
kH+/kD+, is 0.64. A similar value, 0.62, has been reported
Observation of Figure 1 shows that all of the compounds for the 4-methoxyphenyl methyl acetal of benzaldehyde, a
suffer a proton-catalysed hydrolysis. Values of the second- functionality similar to the aryloxymethyltriazenes.[19] For
order rate constants, kH+, for the proton-catalysed hydroly- compound 5a, the temperature effect enabled a determi-
2058
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2005, 2056–2063

Mechanism of Hydrolysis of Aryl Ether Derivatives of 3-Hydroxymethyltriazenes
FULL PAPER
nation of 70 5 kJmol–1 for ΔH‡ and +4 10 JK–1 mol–1
for ΔS‡. These values are consistent with an A1 pathway.[20]
Table 2. Rate constants for the specific-acid-catalysed, kH+, and
spontaneous, k0, hydrolysis of 5a–f at 25 °C.
Com-
kH+/dm3mol–1s–1
k0/10–5 s–1
pound
5a
5.54, 3.8[a], 7.53[b], 10.8[c], 17.6[d]
8.68[e]
1.93
5b
5c
5d
5e
5f
1.44
1.12
0.787
0.397
16.4
23
285, 181[a], 349[b], 460[f], 763[g]
266[e]
Scheme 2. (a) Specific-acid, (b) general-acid and (c) spontaneous
pathways for the hydrolytic cleavage of aryloxymethyltriazenes.
[a] 293 K. [b] 300 K. [c] 304 K. [d] 309 K. [e] in D2O. [f] 303 K. [g]
308 K.
+
was studied in 0.1 m formate buffer, for which there is no
buffer catalysis for this compound, containing 50% meth-
anol. The reaction was monitored over time by HPLC and,
as well as the usual products of reaction, viz. the aniline
and phenol, the methoxymethyltriazene 4 (Ar = 4-CNC6H4,
R = Me) was also detected. This compound arises from the
trapping of the triazenyliminium ion shown in Scheme 2,
process a, and accounts for about 50% of the product yield
before itself decomposing (Figure 3).
Correlation of kH with σp (Figure 2) affords a ρ value
of –0.93 (r2 = 0.93) [correlation with σ– gives a ρ of –0.61
(r2 = 0.89)]. The sign of ρ reflects an increase in positive
charge in the phenol moiety as the transition state is
formed. However, the relatively low magnitude of ρ almost
certainly reflects the fact that it is a composite of the ρ for
substrate protonation and the ρ for subsequent decomposi-
tion. The solvent deuterium isotope effect is that expected
for a pre-equilibrium protonation of the substrate, consis-
tent with this interpretation. Thus, electron-donating sub-
stituents in the aryloxy ring will enhance protonation (nega-
tive ρ), but will retard the decomposition (positive ρ). The
observed value implies that the former effect predominates,
but the second effect lowers the magnitude of ρ.
The General-Acid-Catalysed Reaction
The data in Table 1 demonstrate that the hydrolyses of
compounds 5 are subject to buffer catalysis. For compound
5a, buffer catalysis was observed only above pH of about 4;
for compounds 5b,c, buffer catalysis was observed across
the pH range; for 5d–f, buffer catalysis was observed only
for chloroacetic, formic and acetic acid buffers.
Figure 4a shows the plots of kobs vs. [total buffer] for the
catalysis of 5a by acetic acid buffers, and Figure 4b shows
the corresponding plot of the slopes, kcat, from Figure 4a
against the mol fraction of the acid form of the buffer. The
latter clearly demonstrates that the buffer-catalysed reaction
Figure 2. Hammett plot for the specific-acid-catalysed hydrolysis of
compounds 5a–f.
Taken together, these data are in agreement with a two- involves only the general acid form of the buffer, and the
step, specific-acid-catalysed mechanism involving proton- bimolecular rate constant for general acid catalysis, kA, is
ation of the substrate followed by rate-determining unim- obtained from the intercept at a mol fraction of 1. For ex-
olecular decomposition of the protonated species perimental expediency, more generally values of kA were
(Scheme 2, process a). Of course, another mechanistic pos- determined from the slopes of plots of kobs vs. [general acid
sibility is the protonation of the triazene, followed by rate- form of the buffer]; when determinations of kA were made
determining formation of an O-aryl formaldehyde cation at two different pH values using the same buffer material
with concomitant expulsion of a monomethyltriazene. such plots were parallel confirming the general-acid nature
However, we can rule this out, and confirm the proposed of the buffer catalysis. The values of kA so obtained are
mechanism, by the following experiment. Compound 5a listed in Table 3.
Eur. J. Org. Chem. 2005, 2056–2063
www.eurjoc.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2059

E. Carvalho, A. P. Francisco, J. Iley, E. Rosa
FULL PAPER
Figure 3. Time courses for the reactions of (a) 5a and (b) 5f in 0.1
m pH 4.1 formic acid buffer containing 50% aqueous methanol:
filled diamond, 5; filled square, 4; diamond, corresponding phenol;
triangle, corresponding aniline.
Figure 4. Plots of (a) kobs vs. [total buffer] for the acetic acid cata-
lysed hydrolysis of 5a at 25 °C, and (b) the slopes of the lines in (a)
vs. mol fraction of the acid form of the buffer.
Substituent effects for the general-acid-catalysed reaction
were studied in chloroacetic, formic and acetic acid buffers.
Values of logkA were found to correlate better with σ– than
with σp, and the ρ values are 0.49, 0.86 and 1.0, respectively,
for the three acids (Table 3). The corresponding correlations
between logkA and the pKa of the phenol leaving group mett value for acetic acid catalysis quoted in Table 3 is ob-
afford βlg values of –0.23, –0.4 and –0.42 in the three acids, tained by omitting the rate constant for compound 5a,
respectively. These positive ρ values contrast with the nega- which appears more reactive than such a plot would imply.
tive value obtained for specific acid catalysis and imply that, A similar “up-turn” in reactivity was observed for electron-
as might be expected, general acid catalysis is associated donating substituents in the O-aryl acetals of benzalde-
with an increase of negative charge in the phenol moiety as hyde.[19] This effect has been ascribed to varying amounts
the transition state is formed. Moreover, the magnitude of of proton transfer and C–O bond breaking in the transition
ρ diminishes as the strength of the general acid increases. states; C–O bond breaking runs ahead of O–H bond forma-
Again, such an effect was also observed with the benzalde- tion for electron-withdrawing substituents, the reverse being
hyde O-aryl acetals.[19] We interpret this as implying there true for electron-donating substituents.[19]
is more extensive O–H bond making and less C–O bond
The kA values for compounds 5a–f also correlate with
breaking for a stronger acid than for a weaker acid. As the pKaЈ of the acid catalysts (i.e. the pKa values corrected
described below, the uncatalysed reaction that involves C– for ionic strength[21]) and the derived Brønsted α values are
O bond cleavage is highly sensitive to substituents and has listed in Table 3. These values are typical of general-acid-
a high positive value of ρ. Thus, the general-acid-catalysed catalysed pathways, and it is clear that the Brønsted α value
reaction involving more extensive C–O bond cleavage will diminishes as the nucleofugacity of the phenol leaving
have the more positive value of ρ. Interestingly, the Ham- group increases. Again, this behaviour parallels that of the
2060
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2005, 2056–2063

Mechanism of Hydrolysis of Aryl Ether Derivatives of 3-Hydroxymethyltriazenes
FULL PAPER
Table 3. Second-order rate constants, kA, for the general-acid-catalysed hydrolysis of 5a–f at 25 °C.
kA/10–3
m
–1 s–1
5d
Ј[a]
Buffer
pKa
Hammett ρ
5a
5b
5c
5e
5f
[b]
ClCH2CO2H
HCO2H
2.72
7.97
12.2
2.48
27.5
20.5
33.0
13.8
9.22
0.49
0.86
1.0
3.204[c]
3.59
22.8[c]
[b]
1.24
10.9
7.44
5.20[c]
3.85
4.074[c]
4.60
3.25[c]
0.65
MeCO2H
0.490
0.233[d]
0.753[e]
1.13[f]
0.865
3.73
5.084[c]
0.23[c]
2.36[d]
4.85[g]
6.72[e]
10.5[f]
[b]
[b]
[b]
pyridine
imidazole
phosphate
5.07
6.72
6.79
0.032
0.0046
0.049
[b]
[b]
[b]
0.0078
0.053
0.0063
Brønsted α
(r2)
0.71
(0.8)
0.58
(0.95)
0.74
(0.99)
0.46
(1)
0.38
(0.97)
0.21
(0.8)
[a] The pKa at 25 °C corrected for ionic strength effects; for work at other temperatures the pKa was further corrected accordingly. [b]
Buffer catalysis not observed. [c] In D2O. [d] 293.5 K. [e] 303.3 K. [f] 308.5 K. [g] 300.1 K.
O-aryl acetals.[19] The data imply that proton transfer from ure 1) demonstrates that this reaction extends across the pH
the acid catalyst to the aryl ether oxygen is 60–70% com- range up to a pH of about 13, with no incursion of a reac-
plete in the transition state for those compounds containing tion involving HO–. For these three compounds, a Ham-
relatively poor leaving groups, 5a-c, diminishing to 20–40% mett ρ value of 3.1 is obtained (Figure 5) when logkobs is
complete for the compounds containing the better leaving plotted against σ–; the equivalent Brønsted βlg value is 1.4.
groups. This is to be expected, as the better the leaving A Hammett ρ of 2.7 has been reported for 2-aryloxytetrahy-
group the smaller the extent of proton transfer required for dropyrans.[24] The magnitude and sign of ρ, and the fact
its expulsion. Consistent with this are the solvent kinetic that the correlation is with σ–, support a mechanism involv-
deuterium isotope effects for compounds 5a,e. For com- ing the unimolecular decomposition of the substrate, as de-
pound 5a the reaction was studied using acetic acid buffers picted in Scheme 2c, involving expulsion of a phenoxide ion
D
and a value for kAH/kA of 2.8 was determined; a value of and the formation of the corresponding triazenyliminium
2.14 has been reported for the acetic acid catalysed hydroly- ion. The high magnitudes of both ρ and βlg are the result
sis of benzaldehyde O-phenyl O-methyl acetal,[13] and a of the direct conjugation between the electron-attracting
value between 2.65 and 3.4, depending on the composition substituents on the phenol ring and the negative charge in
of the solvent, for the formic acid catalysed hydrolysis of 2- the transition state.
(4-nitrophenoxy)tetrahydropyran.[22] For 5e the reaction
was studied by using formic acid buffers, and the corre-
sponding isotope effect was calculated to be 1.4. Vaughan
and coworkers reported[18] a value of 1.6 Ͻ kH/kD Ͻ 2.28,
but, as mentioned earlier, this includes components from
both the proton and general acid pathways.
The effect of temperature on the buffer-catalysed path-
way was examined for compounds 5b and 5d in acetic acid
buffers (Table 3); for 5b the values of ΔH‡ and ΔS‡ are
70 5 kJmol–1 and –55 10 JK–1 mol–1, respectively, and
for 5d the corresponding values are 80 5 kJmol–1 and
–43 10 JK–1 mol–1. These values, together with the Ham-
mett ρ values, Brønsted α values and solvent isotope effects,
are consistent[23] with the A-SE2 mechanism shown in
Scheme 2b, in which the expulsion of the phenol leaving
group, to form a triazenyliminium ion, is aided by the gene-
ral acid. The nature of the buffer catalysis described here
contrasts with the role proposed by Vaughan,[18] who as-
sumed the buffer to be acting as a nucleophile.
Figure 5. Hammett plot for the spontaneous hydrolysis of 5d–f at
25 °C.
The spontaneous hydrolysis reaction was also studied for
5f in D2O solutions at pH 10–12 (Table 2), allowing the
calculation of a solvent isotope effect of 1.1. Values of 1.1
and 1.13 have been reported for the analogous spontaneous
hydrolysis of 2-(4-nitrophenoxy)tetrahydropyran[22] and O-
The pH-Independent Reaction
Above a pH of approximately 4, compounds 5d–f exhibit ethyl S-phenyl benzaldehyde acetal,[25] respectively; again
a pH-independent reaction. For 5f, the pH-rate profile (Fig- this is consistent with a process that involves unimolecular
Eur. J. Org. Chem. 2005, 2056–2063
www.eurjoc.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2061

E. Carvalho, A. P. Francisco, J. Iley, E. Rosa
FULL PAPER
ionisation and does not involve proton transfer,[26] such as phosphate buffer and a mean plasma half-life of 1.81 h.[27]
that proposed in Scheme 2c. To trap the triazenyliminium Indeed, a recently reported carbamate derivative of a mono-
ion formed by this spontaneous process, the reaction of 5f methyltriazene based on the O6-benzylguanine scaffold, a
was carried out in a mixed solvent system containing 50% compound that has in vitro antitumour activity comparable
methanol at a pH at which the spontaneous reaction ac- to temozolomide, has a hydrolysis half-life of 23 min.[28]
counted for 90% of the reaction. Loss of the starting mate- However, the most reactive of the current compounds, 5f,
rial, formation of the phenol and 4-cyanoaniline coprod- has a half-life of 1.5 min that is too short (leaving aside any
ucts, and formation of the methoxymethyltriazene trapped toxicological issues related to 4-nitrophenol) and com-
product, were monitored by HPLC. The data in Figure 3 pounds 5d,e have half-lives of around 25 min. The data we
show that the starting aryloxymethyltriazene can be entirely present in Figure 5 lead us to suggest that an aryloxy-
accounted for by the aniline, phenol and trapped triazene methyltriazene containing a biocompatible phenolic moiety
products. Under the conditions of the experiments, approxi- that has a Hammett σ– value of between 0.75 to 0.80 (or,
mately 50% of the iminium ion is trapped by the methanol what is equivalent, a pKa of approximately 8.2) should have
cosolvent. Significantly, no methoxymethyl aryl ether was a half-life in the region of 2 h making it a potential candi-
observed. Thus, compounds 5 undergo spontaneous hydrol- date for a triazene prodrug.
ysis to triazenyliminium and aryloxide ions rather than take
the alternative pathway of forming a triazenyl anion and an
Experimental Section
O-aryl formaldehyde cation. The variation of rate constant
with temperature affords ΔH‡ and ΔS‡ values of
69 5 kJmol–1 and –60 10 JK–1 mol–1, respectively. The
latter value is rather more negative than might be expected
for a spontaneous ionisation process, even accounting for
the presence of an organic cosolvent. However, a similar
value (–41 JK–1 mol–1) has been reported for the spontane-
ous hydrolysis of O-ethyl S-phenyl benzaldehyde acetal.[23]
It has also been reported that, for formaldehyde derivatives,
as we have here, the measured entropies of activation for
spontaneous reactions are some 60–80 JK–1 mol–1 more
negative than for the corresponding acid-catalysed pro-
cess;[24] the corresponding difference here is 64 JK–1 mol–1.
This is probably due to the differing involvement of the sol-
vent in the transition state; in the spontaneous process the
reactant is a neutral species that ionises to form a cation
and an anion, which will involve significant solvent organis-
ation in the transition state, whereas the acid-catalysed pro-
cess involves positively charged species throughout such
that solvent reorganisation in the transition state is likely to
be much less.
WARNING: All triazenes used in this study should be considered
as mutagenic and carcinogenic and appropriate care should be
taken to handle them safely.
Chemicals were reagent grade for the syntheses, LichroSolv
(Merck) grade for HPLC analyses, and AnalaR grade for the kin-
etic investigations. 1H NMR spectra were obtained as CDCl3 solu-
tions at 25 °C using a JEOL LA300 (300 MHz) spectrometer.
Chemical shifts, δ, are given in ppm, and proton–proton coupling
constants, J, are quoted in Hertz (Hz). Infrared spectra were re-
corded with a Perkin–Elmer 1310 spectrophotometer. Low resolu-
tion electrospray ionisation mass spectra were recorded with a VG
Quattro mass spectrometer. Elemental analyses were performed by
MEDAC Ltd., Brunel Science Centre, Surrey, TW20 0JZ, UK.
Substrates and Reagents: The 3-aryloxymethyl-3-methyl-1-(4-cya-
nophenyl)triazenes 5 were prepared as reported previously.[15,29] All
are new compounds:
5a: 83 mg, 33%; m.p. 84–85 °C; IR: ν˜ = 3054, 3007, 2954, 2221,
1600, 1504, 1443, 1034, 1013, 829 cm–1. H NMR: δ = 3.2 (s, 3 H,
1
3
NMe), 3.76 (s, 3 H, OMe), 5.65 (s, 2 H, NCH2O), 6.81–6.83 (d, J
3
= 9.03, 2 H, OAr), 6.92–6.94 (d, J = 9.03, 2 H, OAr), 7.44–7.46
3
3
(d, J = 8.55, 2 H, NAr), 7.61–7.63 (d, J = 8.55, 2 H, NAr) ppm;
m/z (%) = 297 [MH+]. C16H16N4O2 (296): calcd. C 64.87, H 5.41,
N 18.92; found C 65.0, H 5.50, N18.9.
Conclusion
5b: 145 mg, 64%; m.p. 76–78 °C; IR: ν˜ = 3055, 2969, 2220, 1597,
1490, 1436, 1351, 1190, 1004, 987, 833 cm–1. 1H NMR: δ = 3.31 (s,
3 H, NMe), 5.73 (s, 2 H, NCH2O), 6.98–7.00 (m, 3 H, OPh), 7.27–
Like alkoxymethyltriazenes, aryloxymethyltriazenes de-
compose by cleavage of the C–O aminal bond to form a
triazenyl iminium ion and a phenol. This distinguishes them
from other 1-aryl-3,3-dialkyltriazenes which undergo cleav-
age of the N2–N3 triazene bond.[16] The aryloxymethyltriaz-
enes are more reactive than their alkoxymethyl counterparts
and undergo general-acid-catalysed and spontaneous hy-
drolysis pathways alongside the reaction catalysed by H+.
In this regard, they parallel the chemistry of O-aryl acetals,
compounds to which they are structurally related. Unfortu-
nately, the compound that contains both triazene and 4-
3
7.33 (m, 2 H, OPh), 7.47–7.50 (d, J = 8.80, 2 H, NAr), 7.62–7.65
3
(d, J = 8.80, 2 H, NAr) ppm; m/z (%) = 267 [MH+]. C15H14N4O
(266): calcd. C 67.65, H 5.30, N 21.03; found C 67.6, H 5.3, N
20.8.
5c: 192 mg, 75%; m.p. 89–91 °C; IR: ν˜ = 3092, 3056, 2218, 1597,
1
1493, 1468, 1355, 1278, 978, 860, 820 cm–1. H NMR: δ = 3.30 (s,
3 H, NMe), 5.70 (s, 2 H, NCH2O), 6.92–6.95 (2 H, 3J = 8.98, OAr),
7.23–7.26 (2 H, 3J = 8.26 Hz, OAr), 7.47–7.48 (2 H, 3J = 8.62,
NAr), 7.63–7.66 (2 H, 3J = 8.62, NAr) ppm; m/z (%) = 301/303
[MH+]. C15H13N4OCl (300.5): calcd. C 59.91, H 4.36, N 18.63;
methoxyphenol anticancer agents, 5a, has a half-life of found C 59.9, H 4.30, N 18.45.
270 h at physiological pH, making it far too stable to be
5d: 37 mg; 13%; m.p. 120–122 °C; IR: ν˜ = 3064, 2223, 1664, 1600,
1576, 1474, 1360, 1236, 1206, 993 cm–1. 1H NMR: δ = 2.55 (s, 3
considered a suitable mutual prodrug when considered
alongside the clinically-used triazene prodrug temozolom-
H, Ac), 3.33 (s, 3 H, NMe), 5.80 (s, 2 H, NCH2O), 7.03–7.06 (2 H,
ide, a compound that has a half-life of 1.83 h in pH 7.4 3J = 8.98, OAr), 7.52–7.55 (2 H, J = 8.80, NAr), 7.64–7.67 (2 H,
3
2062
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.eurjoc.org
Eur. J. Org. Chem. 2005, 2056–2063

Mechanism of Hydrolysis of Aryl Ether Derivatives of 3-Hydroxymethyltriazenes
FULL PAPER
3
3J = 8.80, NAr), 7.92–7.95 (2 H, J = 8.98, OAr) ppm; m/z (%) =
[1] D. B. Kimball, M. M. Haley, Angew. Chem. Int. Ed. 2002, 41,
3338–3351.
[2] G. L. Cohen, C. I. Falkson, Drugs 1998, 55, 791–799.
[3] T. A. Connors, P. M. Goddard, K. Merai, W. C. J. Ross,
D. E. V. Wilman, Biochem. Pharmacol. 1976, 25, 241–246.
[4] J. L. Skibba, D. D. Beal, G. Ramirez, G. T. Bryan, Cancer Res.
1970, 30, 147–150.
309 [MH+]. C17H16N4O2 (308): calcd. C 66.22, H 5.23, N 18.17;
found C 66.25, H 4.71, N 18.0.
5e: 40 mg; 15%; m.p. 154–6 °C; IR: ν˜ = 3099, 3062, 2225, 1602,
1501, 1454, 1387, 1360, 1251, 994, 843 cm–1. 1H NMR: δ = 3.32 (s,
3 H, NMe), 5.79 (s, 2 H, NCH2O), 7.05–7.08 (2 H, 3J = 8.98, OAr),
7.52–7.55 (2 H, J = 8.77, NAr), 7.59–7.62 (2 H, J = 8.98, OAr),
7.65–7.68 (2 H, 3J = 8.80, NAr) ppm; m/z (%) = 292 [MH+].
C16H13N5O (291): calcd. C 65.97, H 4.50, N 24.03; found C 65.7,
H 4.2, N 23.6.
3
3
[5] F. W. Krüger, R. Preussman, N. Niepelt, Biochem. Pharmacol.
1971, 20, 529–533.
[6] C. J. Rutty, D. R. Newell, R. B. Vincent, G. Abel, P. M. God-
dard, S. J. Harland, A. H. Calvert, Brit. J. Cancer 1983, 48,
140.
[7] C. M. Hemens, K. Vaughan, J. Chem. Soc. Perkin Trans. 2
1986, 11–15.
[8] J. N. Iley, R. Moreira, E. Rosa, J. Chem. Soc. Perkin Trans. 2
1987, 1503–1508.
[9] R. Banerjee, Z. Rachid, J. McNamee, B. J. Jean-Claude, J. Med.
Chem. 2003, 46, 5546–5551.
[10] C. M. Hemens, H. W. Manning, K. Vaughan, R. J. LaFrance,
Y. Ta n g , Can. J. Chem. 1984, 62, 741–748.
[11] L. Fernandes, A. P. Francisco, J. Iley, E Rosa, J. Chem. Soc.
Perkin Trans. 2 1994, 2313–2317.
5f: m.p. 184–186 °C; IR: ν˜ = 3092, 3056, 2218, 1597, 1493, 1468,
1
1355, 1278, 978, 860, 820 cm–1. H NMR: δ = 3.33 (s, 3 H, NMe),
3
5.83 (s, 2 H, NCH2O), 7.07–7.10 (2 H, J = 9.16, OAr), 7.53–7.56
3
3
(2 H, J = 8.80, NAr), 7.66–7.69 (2 H, J = 8.80, NAr), 8.20–8.23
(2 H, 3J = 9.37, OAr) ppm; m/z (%) = 312 [MH+]. C15H13N5O3
(311): calcd. C 57.87, H 4.21, N 22.5; found C 57.7, H 4.0, N 22.1.
Kinetics: The decomposition of the triazenes 5 was followed by
monitoring the decrease in the UV absorbance of the substrate at
an appropriate wavelength with a Perkin–Elmer Lambda 2 spectro-
photometer. For solubility reasons, acetonitrile (10%) was used as
a cosolvent. The ionic strength of the reactions was maintained at
0.5 moldm–3 by the addition of NaClO4. In general, reaction solu-
tions were monitored continuously in cells thermostatted to
0.1 °C. Pseudo-first-order rate constants were obtained from
slopes of plots of ln(At–Aϱ) vs. time, where At and Aϱ are the ab-
sorbances at time t and infinity, respectively. Rate constants deter-
mined by this method were reproducible to 3%. For very slow
reactions, a noncontinuous method was used, in which the reaction
solution was kept in a thermostatted water bath and aliquots were
taken at timed intervals and their absorbances measured in the
spectrophotometer. For these reactions, only about 3 half-lives were
followed and an initial rate method was used to calculate the rate
constants. Rate constants determined by this method were repro-
ducible to 5%. The pH of the reaction solution was measured at
the commencement and conclusion of each experiment. In deuter-
ated solvents pD values were calculated from the expression pD =
pH + 0.4.[30]
[12] L. M. Cameron, R. J. LaFrance, C. M. Hemens, K. Vaughan,
R. Rajamaran, D. C. Chubb, P. M. Goddard, Anti-Cancer
Drug Des. 1985, 1, 27–36.
[13] E. Anderson, B. Capon, J. Chem. Soc. B 1969, 1033–1037.
[14] P. A. Riley (Ed.), Hydroxyanisole: Recent Advances in Anti-
melanoma Therapy, IRL Press, Oxford, 1984.
[15] M. P. Merrin, D. L. Hooper, R. J. LaFrance, R. Snooks, K.
Vaughan, Can. J. Chem. 1992, 70, 144–150.
[16] G. F. Kolar in Chemical Carcinogens (Ed.: C. E. Searle), ACS
Monograph 182, vol. 1, American Chemical Society, Philadel-
phia, 1984, ch. 14.
[17] F. Rakotondradany, C. I. Williams, M. A. Whitehead, B. J.
Jean-Claude, J. Mol. Struct. (THEOCHEM) 2001, 535, 217–
234.
[18] K. Vaughan, D. L. Hooper, M. P. Merrin, Can. J. Chem. 1992,
70, 2224–2233.
[19] B. Capon, K. Nimmo, J. Chem. Soc. Perkin Trans. 2 1975,
1113–1118.
[20] N. S. Isaacs, Physical Organic Chemistry, Longman, Harlow,
U. K., 1987.
[21] C. W. Davies, J. Chem. Soc. 1938, 2093–2098.
[22] T. H. Fife, L. H. Brod, J. Am. Chem. Soc. 1970, 92, 1681–1684.
[23] J. M. Williams, M. M. Kreevoy, Adv. Phys. Org. Chem. 1968,
6, 63–101.
Product Analysis: The UV spectra of the reaction solutions at the
conclusion of each experiment were identical with those of the cor-
responding anilines and phenols. In selected cases, the anilines and
the phenols were identified and quantified by HPLC, and in some
cases were isolated from large scale reactions. Hydrolyses of com-
pounds 5b and 5f were also carried out in a pH 4.1 formate buffer
containing 50% methanol. In these reactions, the formation of 1-
(4-cyanophenyl)-3-methoxymethyl-3-methyltriazene, 4 (Ar = 4-
NCC6H4, R = Me)[29] was observed by HPLC; quantification of
the aryl ether starting material and methyl ether product was
achieved with a system comprising a Lichrospher®100 5 μm RP-8
250 mm×4 mm column and acetonitrile/water (65:35) as eluent at
a flow rate of 1 mL min–1, while the corresponding products of
decomposition of the starting material, 4-cyanoaniline and phenol,
were quantified by an eluent system comprising acetonitrile/water
(30:70).
[24] G.-A. Craze, A. J. Kirby, J. Chem. Soc. Perkin Trans. 2 1978,
354–356.
[25] J. P. Ferraz, E. H. Cordes, J. Am. Chem. Soc. 1979, 101, 1488–
1491.
[26] E. K. Thornton, E. R. Thornton in Isotope Effects in Chemical
Reactions (Eds.: C. J. Collins, N. S. Bowman, R. Van Nos-
trand), New York, 1970, ch. 4.
[27] H. S. Friedman, T. Kerby, H. Calvert, Clin. Cancer Res. 2000,
6, 2585–2597.
[28] M. J. Wanner, M. Koch, G.-J. Koomen, J. Med. Chem. 2004,
47, 6875–6883.
[29] L. Fernandes, J. Iley, E. Rosa, J. Chem. Res. 1987, (S) 264–
265, (M) 2216–2229.
[30] R. G. Bates, Determination of pH, Wiley, New York, 1954.
Received: August 16, 2004
Eur. J. Org. Chem. 2005, 2056–2063
www.eurjoc.org
© 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2063