Time-resolved infrared spectroscopy of intermediates and products from photolysis of 1-(2-nitrophenyl)ethyl phosphates: Reaction of the 2- nitrosoacetophenone byproduct with thiols

Article abstract of DOI:10.1021/ja964430u

Rapid scan Fourier transform infrared (FTIR) spectroscopy and time- resolved single wavelength infrared (IR) spectroscopy have been applied to study the mechanism of photochemical release of adenosine 5'-triphosphate (ATP) from its P³-[1-(2-nitrophenyl)ethyl] ester (caged ATP). Bands arising from phosphate and non-phosphate vibrations characteristic of the aci-nitro anion intermediate and from free ATP and hyproducts of the nitrophenylethyl group have been assigned using ¹³C, ¹⁵N, and ¹⁸O isotopomers of caged ATP. Monitoring several of these bands using time-resolved single frequency IR spectroscopy confirms that release of ATP occurs in a single exponential process synchronous with the decay of the aci-nitro anion intermediate. Spectral characteristics of the reaction products arising from the l-(2- nitrophenyl)ethyl group in the absence and presence of dithiothreitol (DTT) have been determined. The major final byproduct from photolysis conducted in the presence of DTT is 3-methylanthranil. The mechanism of formation of this compound from 2-nitrosoacetophenone has been investigated in detail by a combination of spectroscopic, kinetic, and chemical methods and reconciled with earlier data. The byproduct species likely to be present on the time scale of most biological experiments using caged compounds is 2- hydroxylaminoacetophenone as a mixture of ring-chain tautomers.

Full text of DOI:10.1021/ja964430u

J. Am. Chem. Soc. 1997, 119, 4149-4159
4149
Time-Resolved Infrared Spectroscopy of Intermediates and
Products from Photolysis of 1-(2-Nitrophenyl)ethyl Phosphates:
Reaction of the 2-Nitrosoacetophenone Byproduct with Thiols
Andreas Barth,† John E. T. Corrie,*,‡ Michael J. Gradwell,‡ Yashusi Maeda,†
Werner Ma1ntele,† Tanja Meier,† and David R. Trentham‡
Contribution from the Institut fu¨r Theoretische und Physikalische Chemie, UniVersita¨t Erlangen,
Egerlandstrasse 3, 91058 Erlangen, Germany, and National Institute for Medical Research,
The Ridgeway, Mill Hill, London NW7 1AA, UK
ReceiVed December 27, 1996X
Abstract: Rapid scan Fourier transform infrared (FTIR) spectroscopy and time-resolved single wavelength infrared
(IR) spectroscopy have been applied to study the mechanism of photochemical release of adenosine 5′-triphosphate
(ATP) from its P3-[1-(2-nitrophenyl)ethyl] ester (caged ATP). Bands arising from phosphate and non-phosphate
vibrations characteristic of the aci-nitro anion intermediate and from free ATP and byproducts of the nitrophenylethyl
group have been assigned using 13C, 15N, and 18O isotopomers of caged ATP. Monitoring several of these bands
using time-resolved single frequency IR spectroscopy confirms that release of ATP occurs in a single exponential
process synchronous with the decay of the aci-nitro anion intermediate. Spectral characteristics of the reaction
products arising from the 1-(2-nitrophenyl)ethyl group in the absence and presence of dithiothreitol (DTT) have
been determined. The major final byproduct from photolysis conducted in the presence of DTT is 3-methylanthranil.
The mechanism of formation of this compound from 2-nitrosoacetophenone has been investigated in detail by a
combination of spectroscopic, kinetic, and chemical methods and reconciled with earlier data. The byproduct species
likely to be present on the time scale of most biological experiments using caged compounds is 2-hydroxyl-
aminoacetophenone as a mixture of ring-chain tautomers.
Introduction
Scheme 1
Over the last decade flash photolysis of photolabile but
biologically inert molecules (“caged” compounds) to release
biological effectors has had significant impact on the analysis
of rapid physiological processes. By their nature such processes
demand kinetic analysis, and the rate and mechanism of the
photorelease are an integral part of this. Therefore it is desirable
to have a direct, noninferential method for measurement of
photorelease rates from caged compounds. We recently de-
scribed the use of rapid scan Fourier transform infrared (FTIR)
spectroscopy and time-resolved single wavelength infrared
spectroscopy to determine the rate of ATP release upon
photolysis of caged ATP, the P3-[1-(2-nitrophenyl)ethyl] ester
of adenosine triphosphate 1a.1 The spectroscopic data in the
earlier paper were principally concerned with interpretation of
changes in phosphate vibrational modes among the starting
material 1a, the intermediate aci-nitro anion 2, and the final
free ATP (see Scheme 1). The 18O isotopomers 1b and 1c were
used to help make secure assignments of phosphate-associated
bands in the intermediate 2 and the ATP end product. Based
on these assignments, time-resolved single wavelength IR
spectroscopy enabled direct measurement of the rate of free ATP
formation. Several matters remained unanswered, including a
lack of specific assignments for non-phosphate vibrational bands
associated with the aci-nitro intermediate 2. In addition we had
been unable to make definitive assignments of particular
phosphate bands, especially in the region near 1260 cm-1, and
there was a discrepancy between the rate of ATP formation
measured by IR and that expected from previous measure-
ments2,3 that were based on transient UV spectroscopy of the
aci-nitro chromophore.
The present paper gives near-complete band assignments
through the use of further isotopomers of caged ATP which
contain 15N in the nitro group (1d) or 13C in the benzylic carbon
(1e), together with corresponding isotopomers of caged methyl
phosphate 4a and the [â,âγ-18O3]caged ATP isotopomer 5.
Spectra of caged methyl phosphate are not described in detail
but are used to clarify particular assignments. Discrepancies
in the rates of ATP formation measured by different techniques
have been resolved and rationalized by a combination of
transient UV spectroscopy of solutions of caged ATP photolyzed
under the conditions of the IR measurements, together with
consideration of the pH of solutions used for measurements.
We have also used IR spectroscopy to aid identification of
* Author to whom correspondence should be sent.
† Universita¨t Erlangen.
‡ National Institute for Medical Research.
(2) Walker, J. W.; Reid, G. P.; McCray, J. A.; Trentham, D. R. J. Am.
Chem. Soc. 1988, 110, 7170-7177.
(3) McCray, J. A.; Trentham, D. R. Annu. ReV. Biophys. Biophys. Chem.
1989, 18, 239-270.
X Abstract published in AdVance ACS Abstracts, April 15, 1997.
(1) Barth, A.; Hauser, K.; Ma¨ntele, W.; Corrie, J. E. T.; Trentham, D.
R. J. Am. Chem. Soc. 1995, 117, 10311-10316.
S0002-7863(96)04430-7 CCC: $14.00 © 1997 American Chemical Society

4150 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Barth et al.
Figure 1. Infrared difference spectra of caged ATP photolysis at pH 8.5, 1 °C recorded 4-60 ms after the photolysis flash. Solid line, compound
1a; dashed line, compound 1d; dotted line, compound 1e. Labels refer to peaks in the solid line spectrum.
byproducts which arise from the cage group after photolysis
and especially to unravel the multistep process which occurs
when photolysis is performed in the presence of a thiol. Thiols
have been used since the first descriptions of caged compounds
in biology4 to remove 2-nitrosoacetophenone 3 formed as the
initial byproduct (Scheme 1), but the underlying chemistry had
not been defined. Since byproducts are likely to be formed in
millimolar concentrations in parallel with release of the biologi-
cal effector species, it is desirable to have information about
their nature.
downshifted by 28 and 17 cm-1 respectively. The negative band
from the symmetric [15N]nitro vibration overlaps with a positive
band at 1318 cm-1, and its true position is probably nearer to
1325 cm-1 where it appears without overlap in the spectrum
recorded after 2.7 s (data not shown). The true downshift is
therefore ∼22 cm-1, consistent with the reported shift of 24
cm-1 for the νs vibration of [15N]nitrobenzene.5 Three positive
bands at 1465, 1330, and 1179 cm-1 are downshifted 12-15
cm-1 by 15N substitution to 1451, 1318, and 1164 cm-1
respectively. Similar bands and shifts are observed for the aci-
nitro intermediate from caged methyl phosphate 4a and its 15
,
N
Results and Discussion
The IR difference spectra shown in the figures follow the
same convention as previously,1 i.e., negative bands are associ-
ated with absorption by the starting caged compound and
positive bands represent absorption by the species present at
the time of acquisition of the spectrum. Difference spectra of
Figures 1 and 2 were acquired at pH 8.5 and 1 °C as described
previously.1 The unphotolyzed caged compound served as the
reference spectrum, and spectra recorded in the first 60 ms after
the photolysis flash represent almost entirely the transition from
caged compound to the aci-nitro intermediate 2, whereas spectra
accumulated from 2.7 to 23 s after photolysis represent >99%
of the transition from caged to free ATP. As discussed below
and depending upon the experimental conditions, chemistry
involving the byproducts may not be complete until more time
has elapsed.
Transition of Caged ATP to the aci-Nitro Intermediate.
Figure 1 shows difference spectra in the range 900-1550 cm-1
for the absorption of the aci-nitro anion 2 and relevant
isotopomers minus absorption of starting material for caged ATP
1a and isotopomers 1d and 1e. Figure 2a shows corresponding
aci-nitro anion spectra for 1a and the â,âγ-18O3 isotopomer 5.
Related spectra for the γ-18O isotopomers 1b and 1c were
reported previously.1
(a) 15N Isotopomer. For the 15N isotopomer 1d, large
spectral shifts arising from the isotopic substitution are only
observed in the region above 1160 cm-1. As expected, the
negative bands at 1527 and 1347 cm-1 from the antisymmetric
and symmetric NO2 vibrations of caged ATP are strongly
isotopomer 4b. The isotopic shifts of the three bands are smaller
than those for the nitro bands of the starting compounds and
suggest that the corresponding vibrations are less localized in
the aci-nitro intermediates. We refer at other points in the text
to the high degree of coupling which is apparent among
vibrational modes of these intermediates. Proposed assignments
are that the 1465 cm-1 band corresponds to the stretching
vibration of the aci-nitro CdN bond, while the 1330 and 1179
cm-1 bands derive from antisymmetric and symmetric N-O
(4) Kaplan, J. H.; Forbush, B.; Hoffman, J. F. Biochemistry 1978, 17,
1929-1935.
(5) Da¨hne, S.; Stanko, H. Spectrochim. Acta 1962, 18, 561-567.

Photolysis of 1-(2-Nitrophenyl)ethyl Phosphates
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4151
Figure 2. Infrared difference spectra of caged ATP photolysis at pH 8.5, 1 °C. Solid line, compound 1a; dashed line, compound 5. (a) Spectra
obtained 4-60 ms after the photolysis flash. (b) Spectra obtained 2.7-23 s after the photolysis flash. The labels refer to peaks in the solid line
spectrum.
stretching vibrations of the nitronate. The positions of the latter
two bands are at the upper end of the range of values given by
Feuer et al.6 for alkylnitronate salts, i.e., 1205-1316 and 1040-
1175 cm-1, for the antisymmetric and symmetric stretches. The
same authors reported the CdN stretching frequency for these
nitronates in the range 1587-1605 cm-1, compared to our value
of 1465 cm-1. Evidently bond orders in the aci-nitro intermedi-
ate differ substantially from those indicated by structure 2,
implying that other canonical forms make significant contribu-
tions to the resonance hybrid. Small downshifts of 1-3 cm-1
are observed for bands that appear in the spectrum deriving from
photolysis of the nonisotopic compound 1a at 1379, 1245, 1123,
1031, and 1014 cm-1. These shifts suggest that the nitrogen
atom of the nitro group is involved to a minor extent in the
corresponding vibrations.
at 1260 cm-1 were subjects of some conjecture in our earlier
paper,1 where it was proposed that the band at 1260 cm-1
-
involved PO2 vibrations but the band at 1245 cm-1 did not.
In the spectrum of the intermediate from caged methyl phosphate
4a, a comparable band appears at 1244 cm-1 (shifted to 1242
cm-1 in 18O isotopomer 4d) and is downshifted to 1233 cm-1
in the 13C isotopomer 4c. The shift observed upon 13C
substitution in the benzylic carbon of both caged compounds,
together with the lack of a comparable shift upon 18O substitu-
tion of the bridging oxygen, indicates that the band involves a
carbon-carbon vibration in the CdC(Me)O system of the aci-
nitro intermediate 2. A shift of 11 cm-1 is smaller than expected
(∼25 cm-1) for an isolated C-C oscillator and indicates that
other parts of the molecule are also involved, as is supported
by the small 18O and 15N shifts described above. The 13C shift
also confirms the presence of two separate vibrational modes
(1260 and 1245 cm-1) in this region of the spectrum.
(b) 13C Isotopomer. The largest effect in the spectrum of
the intermediate from 13C isotopomer 1e is a downshift of the
band at 1245 to 1234 cm-1. This band and its adjacent shoulder
The only other substantial isotope effect for the 13C isoto-
pomer is in the positive band at 992 cm-1, which shifts to 985
cm-1. This is consistent with the previous interpretation that
(6) Feuer, H.; Savides, C.; Rao, C. N. R. Spectrochim. Acta 1963, 19,
431-434.

4152 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Barth et al.
absorptions in this region are associated with coupled C-O-
(PO3)3-C backbone modes perturbed by formation of the aci-
nitro intermediate. Small downshifts (2-4 cm-1) are evident
for several bands below 1380 cm-1, which again indicates
extensive coupling of vibrations involving the benzylic carbon
with other vibrational modes. An unassigned positive band at
1379 cm-1, which was hardly affected by 15N substitution, is
downshifted by 2 cm-1 while the positive 1330 cm-1 band,
in caged ATP 1a by an overlapping νas negative band at 1251
cm-1 involving the R- and â-phosphates.1 When this band is
absent as in caged methyl phosphate 4a or downshifted by â-18O
substitution (see below), the absorption intensity in the region
-
1245-1260 cm-1 is substantially enhanced. These νas PO2
bands are overlaid by the CdC(Me)O band at 1245 cm-1 as
discussed above and by a positive band at 1209 cm-1 which
probably involves the R-phosphate, since it is unaffected by â-
or γ-18O labeling.
-
tentatively assigned above to νas of the dNO2 system is
downshifted by 4 cm-1. Both bands have the same shifts in
spectra from photolysis of [13C]caged methyl phosphate 4c and
must be associated with the aci-nitro portion of the intermediate
2. It appears reasonable in view of the extended conjugation
The νs PO2- stretching modes around 1100 cm-1 are affected
by â-18O substitution. Most notably the intense positive band
at 1123 cm-1 in the spectrum of the aci-nitro intermediate
derived from the unlabeled compound 1a splits into two bands
at 1127 and 1110 cm-1 in the difference spectrum arising from
photolysis of compound 5. This double peak probably does
not represent two different vibrational modes but instead arises
from overlap of a negative band near 1120 cm-1 with a positive
band in the region 1110-1130 cm-1. The negative band can
of the aci-nitro moiety that there should be small effects of 13
C
substitution on many of the vibrational modes. More surprising
are the respective 4 and 2 cm-1 downshifts of the strong positive
bands at 1123 and 1092 cm-1 and the 2 cm-1 downshift of the
negative band at 1076 cm-1. We previously1 assigned these
bands to symmetrical PO2- modes on the basis of our own 18
O
-
be assigned to a νs PO2 vibration of caged ATP, seen as a
results and those of Takeuchi et al.,7 who assigned the highest
wavenumber vibration to an in-phase mode and the two lower
wavenumber bands to out-of-phase modes. Since the vibrations
of the entire triphosphate chain are coupled, the effect of 13C
substitution in 1e could be expected to be felt by all three of
the observed symmetrical PO2- modes. Similarly the negative
band at 1059 cm-1, previously assigned1 to a νs PO2- vibration
predominantly involving the γ-phosphate, shows a downshift
of 4 cm-1 with 13C substitution. These small effects on
band in the absorbance spectrum (not shown) at 1131 cm-1 for
1a and 1b, at 1123 cm-1 for 1c, and at 1115 cm-1 for 5. Upon
formation of the aci-nitro intermediate 2, the vibration appears
to intensify and shift slightly, as indicated by the positive band
in the range 1120-1130 cm-1 in the difference spectra for
isotopomers 1a-c and 5. Only for 5 do the positive and
negative bands overlap in such a way that the difference
spectrum shows two separate positive peaks. The mode appears
to be more delocalized in the aci-nitro intermediate than in caged
ATP and to involve vibrations other than pure νs PO2- modes,
since its position is less affected by â- or γ-18O3 substitution
than in caged ATP but is sensitive to 13C and 15N substitution.
Takeuchi et al.7 suggested that there was extensive mixing of
phosphate modes are similar in magnitude to the effects of 15
N
substitution on the phosphate bands at 1123, 1031, and 1014
cm-1 (see above). However it is notable that the shifts caused
by the two isotopic substitutions affect different subsets of
phosphate modes.
(c) â,âγ-18O3 Isotopomer. As expected from our previous
results,1 the spectrum of aci-nitro intermediate formation from
â,âγ-18O3 isotopomer 5 showed no isotopic shifts in the range
1550-1300 cm-1 (Figure 2a). In common with the spectrum
of aci-nitro intermediate formation from γ-isotopomer 1c, the
positive band seen for the unlabeled intermediate at 1260 cm-1
-
the νs PO2 vibrations among the three phosphate groups of
ATP at pH 3, where protonation of the terminal phosphate
simulates the effect of alkylation in caged ATP. Our results
indicate that coupling is also possible with groups remote from
the triphosphate. Similar effects are observed in the spectrum
of aci-nitro formation from caged methyl phosphate 4a, where
the band at 1126 cm-1 shows downshifts of 3 cm-1 for 15N
substitution (4b), 4 cm-1 for 13C substitution (4c), and 2 cm-1
for 18O1 substitution (4d).
is downshifted and overlaps with the vibration at 1245 cm-1
,
which does not involve the PO2- group (see above), to give an
intense positive band with a maximum at 1246 cm-1 (cf. 1242
cm-1 for 1c).1 Since this 1260 cm-1 band is similarly affected
by both â- and γ-18O substitution, it is evident that at least some
The positions of the positive band at 1092 cm-1 and negative
bands at 1076 and 1059 cm-1 are essentially unaffected in 5,
although a small new shoulder appears at 1047 cm-1. These
-
of the νas PO2 modes in the aci-nitro intermediate 2 are
coupled. The negative band at 1219 cm-1 was previously
-
bands were previously1 assigned to νs PO2 modes following
-
assigned1 to a νas PO2 vibration of caged ATP to which the
Takeuchi et al.7 and because they were shifted upon γ-18O3
substitution. The lack of a significant isotope effect upon â-18O3
γ-phosphate contributes and as expected remains unaffected by
â-labeling in compound 5, while a shoulder at ∼1205 cm-1
suggests the presence of a weak positive band. For the γ-labeled
compound 1c, where the negative 1219 cm-1 band is down-
shifted to ∼1180 cm-1, it is notable that a well-resolved positive
band is present at 1209 cm-1, while both 1a and 1b show an
inflection at this point in the profile of the 1219 cm-1 negative
band.
To summarize this spectral region for the aci-nitro intermedi-
ate 2, it appears that the νas PO2- vibration of the γ-phosphate
is upshifted in frequency when the aci-nitro intermediate forms
to give the positive band at 1260 cm-1 and the negative band
at 1219 cm-1, both of which are sensitive to γ-18O3 substitution.
In caged ATP itself this γ-PO2- vibration is not coupled to the
â-PO2- vibrations but does couple in the aci-nitro intermediate.
The intensity of the absorption around 1260 cm-1 is diminished
-
substitution shows that the νs vibrations of the â-PO2 group
do not contribute.
The positive band at 1031 cm-1, which is downshifted to
1020 cm-1 in the γ-18O3 isotopomer 1c, is unaffected in the
â,âγ-18O3 isotopomer 5 but the remaining band profile below
1020 cm-1 is characterized by splitting of the sharp bands at
1014 (negative), 992 (positive), and 970 cm-1 (negative) to give
broadened bands at 1019/1014, 995/984, and 967/957 cm-1
,
respectively. This pattern is more complex than the generalized
downshift that was observed for the γ-18O3 isotopomer 1c but
is in line with the previous assignment to delocalized C-O-
(PO3)3-C backbone modes.1
The information gained from isotopomers 1d, 1e, and 5 is
consistent with the previous interpretation1 of the difference
spectra for formation of the aci-nitro intermediate 2. In
particular the band at 1245 cm-1, which contributed intensity
to the previous1 single wavelength kinetic monitoring at 1251
(7) Takeuchi, H.; Murata, H.; Harada, I. J. Am. Chem. Soc. 1988, 110,
392-397.

Photolysis of 1-(2-Nitrophenyl)ethyl Phosphates
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4153
cm-1, has been shown to involve a vibrational mode associated
with the aci-nitro portion of the intermediate 2.
bands at 1076 and 1059 cm-1 present in the difference spectrum
of aci-nitro formation are unaffected in the â-isotopomer 5.
Shoulders remain at these positions in the end spectrum (Figure
2b) but with diminished intensity. The 1059 cm-1 negative band
was previously assigned to a νs PO2- mode of caged ATP with
a strong contribution of the γ-phosphate,1 that is perturbed in
the aci-nitro intermediate and absent from the final free ATP.
The diminished intensity for the â-isotopomer 5 is evidently
caused by a further component of the intensity in the end
spectrum that shifts from ∼1070 cm-1 for unlabeled caged ATP
1a to 1047 cm-1 for 5. For both isotopomers the negative band
intensifies at these positions as the aci-nitro intermediate reacts
to give the final products (cf. Figure 1c of Ref. 1). The
additional negative band at 1047 cm-1 evidently represents a
νs PO2- mode with a large contribution from the â-phosphate,
that is little perturbed by formation of the aci-nitro intermediate
but is sensitive to unmasking of the terminal phosphate. In
summary, our observations of the νs PO2- modes for caged ATP
are that the high wavenumber modes are less localized on
individual phosphate groups than the low wavenumber modes.
These data are consistent with Takeuchi’s findings for ATP
itself.7
Overall Transition from Caged To Free ATP at pH 8.5
in the Absence of a Thiol. This section is principally concerned
with analysis of changes in phosphate modes when ATP is
released from the aci-nitro intermediate 2. Experiments under
different conditions have enabled characterization of the byprod-
ucts arising from the 1-(2-nitrophenyl)ethyl group and are
described in later sections. Figure 2b shows difference spectra
of the photolyzed solution accumulated in the time range 2.7-
23 s after the photolysis flash for caged ATP 1a and its â,âγ-
18O3 isotopomer 5. Corresponding spectra for the γ-18O
isotopomers 1b and 1c have been published previously.1 End
spectra for the 15N and 13C isotopomers 1d and 1e are not shown,
since with a few exceptions there were only very minor
perturbations of weak bands in the region 1800-1250 cm-1
.
The exceptions include the large shifts for the negative bands
arising from the nitro group in 1d already described above.
Changes in the intensity and position of the nitro bands during
the transition from the aci-nitro intermediate to ATP and
byproducts are discussed in the following section on single
wavelength kinetic measurements. The remaining exceptions
are a shift of the carbonyl absorption of the 2-nitrosoaceto-
phenone byproduct for the 13C compound 1e (see below) and a
weak positive band at 1265 cm-1 which is evident in the end
spectrum of photolysis of caged methyl phosphate 4a and shifts
to 1249 cm-1 with 13C substitution. This band is also present
in the caged ATP spectra but is obscured by the large negative
band at 1251 cm-1. However it is visible at 1268 cm-1 in the
spectrum of the â,âγ-18O3 isotopomer 5 because of the down-
shift of the negative band (see below). The magnitude of the
13C-induced shift is compatible with a localized C-C vibration
and a reasonable assignment is to the Ar-C bond of the
byproduct 3.9
Finally, at frequencies below 1030 cm-1 there are a number
of changes caused by â-labeling. This spectroscopic region
involves PR-O-Pâ-O-Pγ vibrational modes, often coupled
to νs modes of the PO2- or PO32- groups. The intensity of the
negative band at 1019 cm-1 in the spectrum from 1a is reduced
in the spectrum from 5, probably because of the downshift of
a negative band from this position to a region close to the
inflection observed at 998 cm-1. The negative band at 938 cm-1
is downshifted to 920 cm-1. The 1019 and 938 cm-1 bands
were previously assigned1 to backbone C-O and P-O vibra-
tions of caged ATP, possibly with a contribution from the
nonbridging P-O vibrations, and the present data support this.
Two positive bands at 989 and 917 cm-1 that appear on
formation of free ATP are affected to differing extents by
â-labeling. The former is weakly downshifted by 3 cm-1, while
the latter is strongly downshifted with a maximum apparently
below the 900 cm-1 limit of our spectral range. These findings
indicate that the vibrational modes corresponding to the higher
frequency difference bands at 1019 and 989 cm-1 have only a
-
In the region of the νas PO2 vibrations, the strong negative
band at 1251 cm-1 in the spectrum of 1a, which is downshifted
by only 2 cm-1 in the γ-18O3 isotopomer 1c,1 is shifted to 1238
cm-1 in the â-isotopomer 5. A new positive band appears for
5 at 1177 cm-1, and there is a marked inflection point at ∼1220
cm-1 which is absent in the spectrum of 1a. These results
confirm our previous interpretation of the νas PO2- vibrations,1
namely that the negative 1251 cm-1 band arises from an
alteration in the vibrational modes of the R- and â-PO2- groups
consequent on the change from caged to free ATP, now
supported by the shift to 1238 cm-1 in 5. In the region around
-
small contribution of νs â-PO2 and Pâ-O-Pγ vibrations,
whereas the lower frequency bands at 938 and 917 cm-1 have
a significant contribution from these modes, in agreement with
Takeuchi’s original assignment.7
-
1220 cm-1 a positive band for the νas mode of the R- and â-PO2
Single Wavelength Kinetic Measurements. As mentioned
in the Introduction, the rate measured by time-resolved single
frequency IR spectroscopy in our previous work1 for photolytic
release of ATP was unexpectedly fast (218 ( 33 s-1 at pH 7
and 22 °C) compared to an estimated value3 of 50 s-1 under
the experimental conditions. We have been concerned to
examine the basis of this discrepancy.
groups of free ATP overlaps with a negative band from the νas
mode of the disappearing γ-PO2 group of caged ATP. The
-
result of the overlap for the unlabeled compound 1a is to cancel
absorption in this region. For the â-isotopomer 5 these
absorptions no longer fully cancel because of the downshift of
the positive band to 1177 cm-1
.
Absorption by antisymmetric modes1,7 of the newly formed
The kinetic measurements in the earlier paper were made at
frequencies (1119 and 1251 cm-1) that reported directly on the
rate of free ATP formation. In light of the additional informa-
tion obtained from the new isotopomers in the present work,
we extended these kinetic measurements to other bands, some
of which have now been shown to be characteristic of the aci-
nitro portion of the overall intermediate 2. The new measure-
ments were made at 10 °C, and results are shown in Table 1.
Figure 3b shows a typical kinetic trace to illustrate the
“instantaneous” transition from caged ATP to the aci-nitro
intermediate 2 and subsequent slower breakdown of the latter
to free ATP. At all IR frequencies studied the slow phase was
monoexponential, and the measured rate constants agree well
2-
γ-PO3 group of ATP dominates the spectrum around 1120
cm-1, and the position of the absorption envelope is not shifted
by â-labeling. However, the fine structure of the band profile
-
is significantly altered because of overlapping νs PO2 modes
of caged and free ATP. As also seen in the spectrum of aci-
nitro formation, a minimum appears at 1121 cm-1 due to a νs
-
PO2 mode of 5, shifted from its position at 1131 cm-1 in
unlabeled compound 1a. As described above, the negative
(8) Baraba´s, K.; Keszthelyi, L. Acta Biochim. Biophys. Acad. Sci. Hung.
1984, 19, 305-309.
(9) Colthup, N. B.; Daly, L. H.; Wiberley, S. E. Introduction to Infrared
and Raman Spectroscopy, 3rd ed.; Academic Press: San Diego, 1990; p
296.

4154 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Barth et al.
Table 1. Transient Kinetics of the Photochemical Cleavage of
As well as these IR measurements, the transient absorbance
change at 420 nm was recorded under identical conditions in
the same experimental cell (see Experimental Section for details)
and showed the expected instantaneous increase in absorbance
upon flash irradiation, followed by an exponential decay at 52
( 3 s-1 (pH 7.0, 10 °C), that is concordant with the IR
measurements (Table 1). Thus the data conclusively equate the
decay rates of the aci-nitro intermediate 2 measured by different
techniques with the rate of release of the ATP product.
Nevertheless we were concerned to resolve the discrepancy
between our earlier IR measurement1 of the ATP release rate
(mean 218 s-1 at pH 7.0, 22 °C) and the values now reported.
As one approach we used transient UV-visible spectroscopy
in a larger optical cell2 to obtain a reference decay rate of the
aci-nitro intermediate 2 for comparison with the IR kinetic
measurements. Because of the longer optical path length in
this UV-visible experiment, it was necessary to use a solution
with a much lower concentration of caged ATP. The solution
had a sufficiently large volume to ensure accurate pH determi-
nation and was supplemented with ADP to mimic the ionic
strength contribution made in the IR solutions by caged ATP.
The measured rate of aci-nitro decay had a mean value of 92
s-1 at pH 7.0 and 20 °C, i.e., approximately 2.4-fold slower
than the rate previously determined1 by IR measurements under
similar conditions. We believe the likely source of the
discrepancy is a variation of pH: the difficulties for accurate
pH control when mixing concentrated solutions of caged ATP
and buffers in the small (1-2 µL) volumes used for the IR
measurements are illustrated by the systematic difference
between the rates measured in our earlier and present studies.
Furthermore the buffering of the IR solutions may be insufficient
to compensate for the protons released by photolysis of the high
concentration of caged ATP. Thus while the IR measurements
are able to provide an excellent linkage of aci-nitro decay and
product release rates, for accurate estimation of the aci-nitro
decay rate it is desirable to work in dilute, well buffered
solutions.
Caged ATPa
wavenumber,b
cm-1
step
exponential k (pH 7.0), k (pH 7.9),
signd s-1 s-1
signc
1060
1119
1251
1330
1347
1465
1527
negative
positive
positive
positive
negative
positive
negative
e
e
e
positive
negative
negative
positive
negative
positive
53 ( 3
57 ( 2
61 ( 8
f
62 ( 9
47 ( 10
7.4 ( 0.4
7.5 ( 0.5
7.7 ( 0.8
f
8.5 ( 1.9
9.0 ( 1.8
a Measurements at 10 °C; data are given as mean ( SD for n ) 4.
The rate constant k corresponds to that shown in Scheme 1. b Spectral
resolution 20 cm-1 except for bands at 1330 and 1347 cm-1, where
spectral resolution was 10 cm-1 c Sign of the “instantaneous” absorb-
.
ance change consequent on photolytic formation of the aci-nitro
intermediate 2. d Sign of the exponential change in absorbance as the
aci-nitro intermediate 2 decays. e Step change in absorbance only.
f Absorbance change subsequent to initial step too small to determine
accurately.
In the final part of this section, we determined the temperature
dependence of the decay rate of the aci-nitro intermediate at
pH 7.0 over the range 2-40 °C, using measurements at three
IR frequencies (1119, 1251, and 1330 cm-1), of which the first
and last are specific for release of ATP and for aci-nitro decay
respectively, while the middle value reports a combination of
these. Measurements at each frequency showed identical
temperature dependence, and the results are shown as an
Arrhenius plot (Figure 3a), from which the measured activation
energy is 62 ( 2 kJ mol-1. This value agrees with work by
Baraba´s and Keszthelyi,8 who reported an activation energy in
the range 52-67 kJ mol-1 for data from UV measurements in
10 mM buffer over the pH range 6-8.
Characterization of Byproducts from the 1-(2-Nitro-
phenyl)ethyl Group at pD 6.0. (a) In the Absence of a Thiol.
For reasons outlined in section (b) below, experiments to
characterize the photolysis products that arise from the 1-(2-
nitrophenyl)ethyl group were performed under mildly acidic
conditions at 35 °C. These experiments were performed in D2O
buffer solutions in order that the spectral region 1550-1700
cm-1 was not obscured by solvent absorption: changing
between H2O and D2O did not otherwise affect the spectra.
Spectroscopic features were essentially identical over the pH
range 6.0-7.5, but at the lower pH the time courses of their
evolution were more accessible to observation. Figure 4a shows
difference spectra of photolysis of caged ATP 1a performed in
the absence of a thiol. Acquisition of spectra began 4 ms after
the photolysis flash, by which time at pH 6 and 35 °C decay of
Figure 3. (a) Arrhenius plot of the kinetics of selected IR bands
following flash photolysis of caged ATP, pH 7.0. b, 1119 cm-1; 0,
1251 cm-1; 4, 1330 cm-1, (b) Kinetic IR absorption recording of caged
ATP photolysis measured at 1251 cm-1 at pH 7.9 and 10 °C. The flash
artefact has been subtracted.
at the different frequencies. Further discussion of results at
individual frequencies is not required, except to comment on
the data for the negative 1527 cm-1 band, which arises from
the nitro group of caged ATP that disappears upon conversion
to the aci-nitro intermediate. We noted previously1 that in the
transition from the intermediate to the end products the intensity
of the negative band diminishes and the minimum shifts by 2
.
to 1525 cm-1 Comparable effects are seen for the 15N
isotopomer 1d, where the observed shift is from 1500 to 1499
cm-1. The effect is less clear for the symmetric NO2 vibration
because adjacent positive bands may distort the band profile.
With this caveat, the observed absorption minimum shifts from
1347 to 1344 cm-1 between the intermediate and final spectra
(see Figure 1 of ref 1). These observations, together with the
kinetics reported here for the 1527 cm-1 band, confirm our
previous report. Nevertheless, changes of intensity and fre-
quency of the nitro bands as the aci-nitro intermediate decays
are inconsistent with the mechanism shown in Scheme 1, where
the nitro group is transformed within the time of the light pulse.
Further investigation will be required to probe this phenomenon.

Photolysis of 1-(2-Nitrophenyl)ethyl Phosphates
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4155
(shifted to 1393 and 1348 cm-1 by 15N substitution) that are
close to positions of 1409 and 1389-1397 cm-1 reported for
cis-nitroso dimers.10 The disappearance of the monomer band
at 1425 cm-1 and its replacement by the dimer band at the same
position could only be observed in the 15N compound 1d, for
which the two bands were shifted to different extents. Upon
longer incubation at 35 °C (spectrum acquired at 42-62 s; not
shown) a positive band appears at 1269 cm-1 (1243 cm-1 with
15N substitution), close to the position (∼1257 cm-1) at which
trans-nitroso dimers absorb.10 The growth of the 1269 cm-1
band was accompanied by a decrease in the absorption bands
of the cis-dimer. Evidently the chemistry of the photolysis
byproduct at neutral or weakly acidic pH and in the absence of
a thiol is essentially that expected11,12 for the known byproduct,2
2-nitrosoacetophenone 3. At pH 8.5 none of the bands
characteristic of the nitrosoketone was observed, except for a
carbonyl absorption at 1686 cm-1. Evidently other reactions
supervene at higher pH, but we have not pursued this matter as
it is of little relevance to uses of caged compounds in biology.
(b) In the Presence of a Thiol. The most striking differences
between the IR spectra recorded after photolysis at 35 °C with
and without dithiothreitol (DTT) present are in the region 1600-
1700 cm-1 (compare Figure 4a,b). In the presence of DTT the
first spectrum, recorded 4-60 ms after the photolysis flash,
shows almost no absorption in this region, but in the second
spectrum, recorded 0.3-3.6 s after the flash, a strong band
appears at 1641 cm-1. The intensity of this band diminishes
upon further incubation, and its maximum shifts to 1645 cm-1
,
with a shoulder at higher frequency (third spectrum, at 120-
140 s). The intensity of the negative band at 1528 cm-1 in the
third spectrum is diminished by the appearance of an over-
lapping positive band (see below). These changes between the
second and third spectra were paralleled in the region 1450-
1480 cm-1, where two medium intensity bands at 1451 and 1475
cm-1 in the second spectrum were replaced by a single stronger
band at 1466 cm-1 in the final spectrum. The bands at 1475
(second spectrum) and 1466 cm-1 (third spectrum) are down-
shifted by 2 and 3 cm-1, respectively, upon 15N substitution
(isotopomer 1d). The most pronounced effect of 13C substitution
(isotopomer 1e) is a downshift of 32 cm-1 for the band at 1641
cm-1 in the second spectrum. In the final spectrum the bands
Figure 4. (a) Infrared difference spectra of caged ATP photolysis in
the absence of dithiothreitol (DTT) at pD 6.0, 35 °C. Spectra were
obtained 4-60 ms (solid line) and 1.3-24 s (dotted line) after the
photolysis flash. Labels refer to peaks in the solid line spectrum. (b)
Infrared difference spectra of caged ATP photolysis in the presence of
200 mM DTT at pD 6.0, 35 °C. Spectra were obtained 4-60 ms (solid
line), 0.3-3.6 s (dashed line), and 120-140 s (dotted line) after the
photolysis flash. Labels refer to peaks in the dashed line spectrum.
at 1645 and 1466 cm-1 are downshifted by 6 and 5 cm-1
respectively.
,
To gain further information about this reaction sequence, we
monitored the UV spectral changes which took place when
2-nitrosoacetophenone was treated with excess DTT in a 1 cm
path length cuvette, where times to add reagents and record
spectra were each 20-30 s. Therefore the first intermediate
seen by IR spectroscopy is missed in these experiments. The
results are given in Figure 5a, which shows that the spectrum
of the nitrosoketone is promptly converted to a much less intense
spectrum (λmax 343 nm) upon addition of the thiol. Upon longer
incubation (ca. 1 h at pH 7, 35 °C for completion) this spectrum
was transformed to a final spectrum with a maximum at 316
nm and several smaller peaks between 265 and 280 nm. Single
wavelength kinetic monitoring at 316 nm under the same
conditions (Figure 5b) showed a half-time of 9 min for the
development of the final spectrum. The rate of this absorbance
change was unaffected by doubling the thiol concentration but
increased by an order of magnitude at pH 6.0 (data not shown).
We note that at the <1 mM concentrations used here and in
related kinetic experiments below, the nitrosoketone is entirely
the aci-nitro intermediate is >99% complete. Thus the spectra
represent only reactions of the 2-nitrosoacetophenone byproduct
3 and its cis- and trans-dimers. They were acquired principally
as part of our investigation of the end product formed in the
presence of a thiol [see section (b) below], but they are discussed
briefly here as they also reveal features of the chemistry of
2-nitrosoacetophenone. As would be expected, there were no
significant changes in phosphate modes as ATP release is
already complete. Therefore this spectral region is omitted from
Figure 4.
The spectra in Figure 4a were acquired in the time frames
4-60 ms and 1-3.6 s after the photolysis flash, and the changes
between the two are a consequence of dimerization of the nitroso
group of 3. The first spectrum shows positive bands at 1501
and 1425 cm-1 (shifted to 1485 and 1417 cm-1 respectively by
15N substitution). The band at 1501 cm-1, characteristic for
the nitroso monomer,10 decayed with an estimated 0.17 s half-
time as the nitroso dimer formed. The carbonyl absorption
moves from 1684 to 1686 cm-1 as the dimerization proceeds
(1647 and 1649 cm-1, respectively upon 13C substitution). The
spectrum after dimerization has bands at 1425 and 1385 cm-1
(11) Azoulay, M.; Fischer, E. J. Chem. Soc., Perkin Trans. 2 1982, 637-
642.
(12) Zuman, P.; Shah, B. Chem. ReV. 1994, 94, 1621-1641 and
references therein.
(10) Reference 9, p 353.

4156 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Barth et al.
Figure 5. (a) Ultraviolet spectra showing the reaction of 2-nitrosoacetophenone (0.118 mM) with DTT (5 mM) at pH 7.0, 35 °C. Solid line,
2-nitrosoacetophenone; dashed line, 2-nitrosoacetophenone within 30 s after addition of excess DTT; dotted line, spectrum after 1 h. (b) Time-
dependent absorbance at 316 nm. The arrow shows the time of mixing excess DTT with the 2-nitrosoacetophenone solution.
Scheme 2
1
in the monomeric form, as shown by H NMR spectra run at
varying concentrations (see Experimental Section). The dimer
is undetectable at 2 mM concentration in either CDCl3 or
CD3OD solution. Also the UV spectrum (Figure 5a) shows the
double peak typical of monomeric aromatic nitroso com-
pounds.11 Therefore the spectral changes seen here and in
previous2 kinetic experiments are not influenced by the mono-
mer-dimer equilibrium of 2-nitrosoacetophenone.
Previous kinetic work2 based on absorbance measurements
at 380 nm had found two steps in the reaction of DTT with
2-nitrosoacetophenone. The final, slow step identified in the
present work was beyond the time scale of the earlier experi-
ments. At pH 7.0, 21 °C the first process had a second-order
rate constant of 3.5 × 103 M-1 s-1. The second process was
Furthermore, the band at 1641 cm-1 in the IR spectrum at 0.3-
independent of DTT concentration, with an observed first-order
3.6 s (Figure 4b, dashed line) is consistent with an internally
rate constant of 45 s-1. The product of the second step had a
hydrogen bonded carbonyl as would be expected in 8 [cf. 1636
UV spectrum with a maximum at 373 nm (measured in ethanol
cm-1 for 2-aminoacetophenone as a saturated solution in D2O
by difference spectroscopy against the starting nitrosoketone).2
(data not shown)]. The initial steps of Scheme 2 are in
When the experiment of Figure 5a was repeated in ethanol
agreement with previous studies of thiol reactions with aromatic
solution, a similar absorbance maximum was observed at 380
nitroso compounds12 and in particular with a report that
nm instead of the 343 nm value found in aqueous solution.
4-nitrosoacetophenone is reduced by thiols to the corresponding
The results of the previous kinetic measurements and the
present IR and UV spectroscopic studies have been interpreted
for the case where DTT is the thiol in Scheme 2. The sequence
is initiated by nucleophilic addition of a thiolate anion to the
nitroso group of 3 to give the intermediate 6, which is inferred
to exist predominantly in its cyclic form (see below). By
contrast, the product of the second step appears to exist as a
hydroxylamine.14
The overall transition from 6 to 7/8 involves an irreversible
reduction of the N-S bond followed by a ring-chain tautom-
erism, and we wished to establish which of these steps
corresponded to the 45 s-1 process seen previously.2 The best
studied analogy for the intramolecular reduction of 6 is from
the extensive studies by Whitesides et al. of thiol-disulfide
mixture of the closed and open forms 7 and 8, respectively.
interchange reactions: with DTT the second (intramolecular)
The long-wavelength absorption band in the intermediate UV
step to cleave the mixed disulfide is faster than the initial
spectrum at 343 nm is unlikely to arise from 7 but would be
interchange15a because of the high effective concentration15b of
consistent with the presence of a proportion of the open-chain
the intramolecular thiol group of DTT. Thus the reduction of
form 8 [cf. 2-aminoacetophenone13 has λmax(EtOH) 359 nm].
(13) Weast, R. C. Handbook of Chemistry and Physics, 50th ed.;
Chemical Rubber Co.: Cleveland, OH, 1969; p C-93.
(14) Diepold, C. H.; Eyer, P.; Kampffmeyer, H.; Reinhardt, K. AdV. Exp.
Med. Biol. 1982, 136B, 1173-1181.

Photolysis of 1-(2-Nitrophenyl)ethyl Phosphates
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4157
Table 2. Rate Constants for Ring-Chain Tautomerism of
2-Hydroxylaminoacetophenone 7/8 at 15 °C
pH
k3, s-1
k-3, s-1
7.5
8.0
8.5
1.32
4.5
15.0
0.33
1.15
4.7
a See Experimental Section for details of solution conditions. b For
definitions of k3 and k-3 see Scheme 2.
6 might be expected to be a rapid step. By contrast, rates of
ring-chain tautomerism within one set of related compounds
can be modulated by substituent effects to span at least three
orders of magnitude16 and prediction of the 7 H 8 interchange
rates was uncertain. We measured these two rate constants
using a saturation-transfer NMR technique17 for solutions of
the ring-chain tautomers 7 and 8 prepared by treating a
choloroform/methanol solution of 2-nitrosoacetophenone 3 with
aqueous solutions of DTT buffered over the pH range 7.5-
8.5. The spectra showed two singlets with an intensity ratio of
∼1:4 at δ 1.83 and 2.63, respectively. The latter overlapped
with a multiplet in the range δ 2.6-2.75 arising from DTT but
could readily be deconvoluted (see Experimental Section).
Continuous irradiation of the upfield singlet caused a marked
decrease in the peak height of the singlet at δ 2.63, thereby
confirming that the two signals arose from species in chemical
exchange. The signal at δ 1.83 corresponds to the cyclic
tautomer 7 and the signal at δ 2.63 to the open form 8. Details
of the saturation-transfer experiments are given in the Experi-
mental Section and the derived rate constants for ring opening
(k3) and closing (k-3) are given in Table 2. Figure 6a shows
data for the measurements at pH 8.0.
Assuming that Scheme 2 correctly interprets the overall
process, the observed rate of the 380 nm transient, kobs, is given
by (k3 + k-3). Extrapolation from the data of Table 2 suggests
that at pH 7.0 and 15 °C, kobs would be ∼0.56 s-1, i.e., much
slower than the value of 45 s-1 measured at pH 7.0, 21 °C.2
Since the NMR data were obtained under different conditions
of solvent, temperature, and ionic strength to the original flash
photolysis measurements, we repeated the latter under conditions
identical to one set of the NMR experiments (pH 8.0, 15 °C).
2-Nitrosoacetophenone was generated rapidly by flash photolysis
of 1-(2-nitrophenyl)ethyl phosphate (caged Pi) for which the
decay of the aci-nitro anion has a half-time ∼0.1 ms under these
conditions.2 The observed rate constant of the 380 nm transient
was 14 s-1 in protiated solvent (H2O-CH3OH) and 8.6 s-1 in
deuterated solvent (D2O-CH3OD; Figure 6b). These rate
constants were unaffected by changing the DTT concentration
from 100 to 300 mM, and the value in the deuterated solvent
agrees well with that of 5.65 s-1 measured by NMR. These
data provide strong evidence that the rate-determining step of
the 380 nm transient is the ring to chain tautomerism 7 f 8,
i.e., k2 . (k3 + k-3). There are insufficient data to determine
whether the approximately 2-fold decrease in rate in the
deuterated solvent arises from small pH discrepancies between
the two solutions or from a kinetic isotope effect.
Figure 6. (a) Height of the NMR signal at δ 2.63 as a function of the
saturation time at δ 1.83 for a pH 8.0 solution of 2-hydroxylamino-
acetophenone 7/8 at 15 °C (see Experimental Section for details of
solution composition). Peak heights are normalized to the height of
the signal without irradiation. (b) Absorbance transient at 380 nm
following flash photolysis of 1-(2-nitrophenyl)ethyl phosphate in the
presence of 100 mM DTT at pH 8.0, 15 °C (deuterated solvent, see
Experimental Section for details of solution composition). The arrow
on the abscissa marks the time of the laser flash. The optical path length
was 4 mm.
of consecutive reactions (analogous to 6 f 7 H 8) with
comparable rates for each step may explain the complex thiol
dependence for the rate of formation of 8 when a monofunc-
tional thiol was used.2 For DTT (Scheme 2) the initial adduct
6 appears not to be present in significant concentration. Thus
the IR spectrum recorded in the 4-60 ms time frame principally
represents the reduced compound 7. Subsequent establishment
of the 7 H 8 equilibrium is represented in the spectrum acquired
at 0.3-3.6 s. The initial formation of 7 implies that its precursor
6 is also in the cyclic form shown and it is of interest why 6
exists in the cyclic form, whereas for 7/8 the equilibrium favors
the open form. The open form 8 is stabilized by its strong
internal hydrogen bond. By contrast, in an open tautomer of
6, there would be steric interactions between the substituents
on the nitrogen and the o-acetyl group. These interactions would
be relieved by cyclization.
The final step of Scheme 2 proposes an acid-catalyzed
dehydration of 7/8 to generate 3-methylanthranil 9 (3-methyl-
2,1-benzisoxazole) as the end product responsible for the IR
bands at 1645 and 1466 cm-1 (Figure 4b) and the UV absorption
at 316 nm (Figure 5a). The final UV spectrum closely matches
that of 3-methylanthranil,18 and the absorbance at 316 nm
corresponds to its formation in 74% yield. A strong band at
1521 cm-1 in the IR spectrum of 3-methylanthranil (see
Experimental Section) is obscured by the negative band at 1528
cm-1 in the final difference spectrum arising from photolysis
of the unlabeled compound 1a but is observable at 1525 cm-1
in the corresponding spectrum from the 15N isotopomer 1d
where the negative band is strongly downshifted. The small
isotopic shifts observed in the IR spectrum of 3-methylanthranil
(see above) indicate that the bands do not arise from localized
vibrational modes. By comparison, the IR spectrum of
With a monovalent thiol, the previous analogy with thiol/
disulfide interchange reactions15 suggests that reduction of the
analogue of 6 to give 7 may be much slower. The possibility
(15) (a) Whitesides, G. M.; Lilburb, J. E.; Szajewski, R. P. J. Org. Chem.
1977, 42, 332-338. (b) Burns, J. A.; Whitesides, G. M. J. Am. Chem. Soc.
1990, 112, 6296-6303.
(16) Valters, R. E.; Flitsch, W. Ring-Chain Tautomerism; Plenum
Press: New York, 1985; Chapter 2.
(17) Cayley, P. J.; Albrand, J. P.; Feeney, J.; Roberts, G. C. K.; Piper,
E. A.; Burgen, A. S. V. Biochemistry 1979, 18, 3886-3895. Hyde, E. I.;
Birdsall, B.; Roberts, G. C. K.; Feeney, J.; Burgen, A. S. V. Biochemistry
1980, 19, 3738-3746.
(18) Yates, P.; Hand, E. S. J. Am. Chem. Soc. 1969, 91, 4749-4760.

4158 J. Am. Chem. Soc., Vol. 119, No. 18, 1997
Barth et al.
[13C1]benzene has been analyzed in detail and shows similarly
small shifts for almost all bands.19
photocleavage of o-nitrobenzyl derivatives has played a pivotal
role in the development of caged compounds for rapid release
of biological effectors. The results presented above demonstrate
the utility of time-resolved infrared spectroscopy as a tool for
mechanistic and product investigations of the photolysis of caged
compounds. Provided that the necessary stable isotopomers of
a compound under study can be synthesized to enable definitive
identification of relevant IR bands, the method is capable of
unequivocal and direct determination of the rate of product
release from its caged precursor. Notably it should enable
detailed investigation of caged compounds that show biphasic
decay of putative aci-nitro anion intermediates when observed
by transient UV spectroscopy.25 The capacity of the method
to aid identification of photolysis products has also been
demonstrated in this work and in a further recent application
where we have confirmed an unexpected decarboxylation which
occurs during photolysis of sulfonamide-protected derivatives
of amino acids.26
The data presented above in support of Scheme 2 are all
spectroscopic, and we wished to obtain some additional chemical
evidence. To confirm the intermediacy of the hydroxylamine
8, a solution of 2-nitrosoacetophenone in chloroform (with added
triethylamine to maintain a basic environment and suppress
dehydration to the anthranil) was treated with excess DTT. The
major product isolated was the azoxybenzene 10, formation of
which provides evidence for the presence of hydroxylamine 8,
since it would be generated by condensation of 8 with the
starting nitrosoketone. The result also suggests that the reaction
kinetics in chloroform differ from those in aqueous solution,
since in the latter case the nitrosoketone would mostly be
consumed by the first reaction with thiol before a significant
concentration of the hydroxylamine could be formed.
Experimental Procedures
Materials. Caged ATP 1a was prepared by esterification of ATP
with 1-(2-nitrophenyl)diazoethane as previously described,2 and the
â,âγ-18O3 analogue 5 was prepared similarly from [â,âγ-18O3]ATP,27
for which the overall isotopic enrichment was 90%. The 15N and 13C
analogues of caged ATP (1d, 1e) and of caged methyl phosphate (4b,
4c) were prepared as described,28 and [18O]caged methyl phosphate 4d
was prepared similarly from [18O]caged phosphate.29 For FTIR
measurements at pH 8.5, solutions contained the caged compound (85-
120 mM) in N,N-bis(2-hydroxyethyl)glycine-KOH (Bicine) buffer (200
mM, pH 8.5). For measurements at pD 6.0 to study the byproducts,
aliquots (1 µL each) of solutions in H2O of the caged compound (85-
120 mM) and 2-(N-morpholino)ethanesulfonic acid-KOH (MES)
buffer (200 mM, pH 6.0) plus DTT (200 mM) where required were
evaporated on the IR window under a nitrogen stream and reconstituted
in D2O (1 µL). For the single-frequency kinetic measurements,
solutions contained caged ATP (90 mM) in N-(2-acetamido)iminodi-
acetic acid-KOH (ADA) buffer (800 mM, pH 7.0) or Bicine buffer
(800 mM, pH 7.9). Silica gel (Merck 9385) was used for flash
chromatography.
Spectroscopy. Details of the FT and dispersive IR spectrometers
have been given previously.1 For measurement of transient signals at
420 nm in the same cell as used for the FTIR and single-frequency
experiments, the sample was placed in a spectrometer fitted with the
same xenon flash tube as used in the IR spectrometers, a monochro-
mator (H20, Jobin Yvon, Longjumeau, France), and a photodiode
detector and preamplifier built in W.M.’s laboratory. The sample
composition was the same as for the IR measurements. Separate
measurement of the UV transient with a dilute solution of caged ATP
was performed as previously described,2 for a solution containing caged
ATP (0.5 mM), ADP (90 mM), and K-ADA buffer (800 mM, pH 7.0).
2-Nitrosoacetophenone 3. This compound was prepared as de-
scribed.30 The 1H NMR spectrum (2 mM in CDCl3) had δ (400 MHz)
2.75 (s, 3H), 7.00 (d, 1H), 7.60 (t, 1H,), 7.71 (d, 1H) and 7.78 (t, 1H).
At 100 mM in CDCl3 all these signals were still present but were
complemented by a further set of signals from the dimer at δ 2.68 (s,
6H), 7.66 (t, 3H), 7.78 (t, 3H), 7.90 (d, 1H), and 7.91 (d, 1H). The
intensity of the latter set of signals was reduced in a spectrum at 15
mM concentration (1:11 molar ratio of dimer:monomer) compared to
a spectrum at 100 mM (1:3 dimer:monomer).
To confirm the formation of 3-methylanthranil, caged ATP
solutions at pH 7 that contained excess DTT were photolyzed
using a single 50 ns pulse of 347 nm light from a frequency-
doubled ruby laser,2,20 and the products were analyzed quanti-
tatively by reverse-phase and anion-exchange HPLC. Single
flash photolysis of 9.8 mM caged ATP released 1.0 mM ATP
(anion-exchange HPLC). Reverse-phase HPLC showed a major
hydrophobic component which coeluted with authentic 3-
methylanthranil and corresponded to formation of the compound
at a concentration of 0.74 mM, i.e., 74% of the released ATP
concentration. Several unidentified minor peaks presumably
account for the remaining materials balance. In keeping with
the time course of formation of 3-methylanthranil shown in
Figure 5b, it was notable that the height of the HPLC peak
corresponding to this compound did not stabilize until the
photolyzed solution had been allowed to stand for ∼1 h after
irradiation. Chromatograms run at earlier times showed a peak
with a shorter retention time, that disappeared as the 3-methyl-
anthranil peak stabilized. This peak was present in chromato-
grams run immediately after mixing 2-nitrosoacetophenone with
DTT and evidently corresponds to the equilibrium mixture of
7 and 8. Reduction of 2-nitrosoacetophenone to 3-methyl-
anthranil has previously been reported by electrochemistry under
acidic conditions,21 with triphenylphosphine22 and with bisulfite.23
Nevertheless, on the time scale of most biological experiments
that use caged compounds, the major byproduct species present
must be the tautomeric forms 7 and 8 of 2-hydroxylamino-
acetophenone, although when a monofunctional thiol is used
significant concentrations of the initial thiol adduct with
2-nitrosoacetophenone may also be present.
Conclusion. Photochemical rearrangements and photo-
chromism of ortho-substituted nitro compounds have been of
fundamental interest for almost 100 years.24 In particular, the
(19) Brodersen, S.; Christoffersen, J.; Bak, B.; Nielsen, J. T. Spectrochim.
Acta 1965, 21, 2077-2084.
(25) Corrie, J. E. T. J. Chem. Soc., Perkin Trans. 1 1993, 2161-2166.
Peng, L; Goeldner, M. J. Org. Chem. 1996, 61, 185-191. Ellis-Davies, G.
C. R.; Kaplan, J. H.; Barsotti, R. J. Biophys. J. 1996, 70, 1006-1016.
(26) Corrie, J. E. T.; Papageorgiou, G. J. Chem. Soc., Perkin Trans. 1
1996, 1583-1592.
(27) Cohn, M.; Hu, A. J. Am. Chem. Soc. 1980, 102, 913-916.
(28) Corrie, J. E. T. J. Labelled Compd. Radiopharm. 1996, 38, 403-
410.
(29) Corrie, J. E. T.; Reid, G. P. J. Labelled Compd. Radiopharm. 1995,
36, 289-300.
(30) Shabarov, Y. S.; Mochalov, S. S.; Stepanova, I. P. Dokl. Akad. Nauk
SSSR 1969, 189, 1028-1030.
(20) McCray, J. A.; Herbette, L.; Kihara, T.; Trentham, D. R. Proc. Natl.
Acad. Sci. U.S.A. 1980, 77, 7237-7241.
(21) Le Guyader, M. C. R. Seances Acad. Sci., Ser. C 1966, 262, 1383-
1386.
(22) Shabarov, Y. S.; Mochalov, S. S.; Fedotov, A. N.; Kalashnikov, V.
V. Khim. Geterotsikl. Soedin. 1974, 9, 1195-1197.
(23) Fedotov, A. N.; Mochalov, S. S.; Shabarov, Y. S. Zh. Prikl. Khim.
1977, 50, 1775-1777.
(24) Wettermark, G.; Black, E.; Dogliotti, L. Photochem. Photobiol. 1965,
4, 229-239. DeMayo, P.; Reid, S. T. Quart. ReV. (London) 1961, 15, 393-
417. Morrison, H.; Migdalof, B. H. J. Am. Chem. Soc. 1965, 30, 3996.

Photolysis of 1-(2-Nitrophenyl)ethyl Phosphates
J. Am. Chem. Soc., Vol. 119, No. 18, 1997 4159
3-Methylanthranil 9. Prepared as described22 by reduction of
2-nitrosoacetophenone with triphenylphosphine to give a colorless oil:
bp (Kugelrohr oven temperature) 190 °C (2 mmHg); UV [25 mM Na
phosphate, pH 7.0-EtOH (9:1)] λmax nm (ꢀ, M-1 cm-1) 257 (1270),
261 (1360), 266 (1640), 271 (1500), 278 (1900), 316 (6100); IR (film)
1643 (vs), 1566 (m), 1521 (s), 1463 (s), 1441 (m), 1401 (m), 1377
(w), 1284 (w), 1220 (m), 1163 (m), 1144 (m), 1081 (w), 1032 (w),
of the measured line height for the signal at δ 2.63 versus saturation
time were fitted to a single exponential using the Kaleidagraph
software33 running on an Apple PowerMacintosh, and the lifetime of 8
was calculated as previously described.17
The relative concentrations of 7 and 8 were estimated from the areas
of Lorentzian-shaped peaks fitted to the signals at δ 1.83 and 2.63
using the line deconvolution methodology within VNMR 5.1. The
lifetime of 7 and the corresponding rate constants were calculated from
these data.17
1
987 (m) cm-1; H NMR (CDCl3, 90 MHz) δ 2.78 (s, 3H), 6.82-7.57
(m, 4H).
To study the formation of 2-hydroxylaminoacetophenone by transient
absorption spectroscopy at 380 nm following flash photolysis (Figure
6b), a solution containing 1-(2-nitrophenyl)ethyl phosphate29 (2 mM),
DTT (200 or 600 mM), and Tricine (20 mM) was diluted with an equal
volume of methanol. For corresponding experiments in deuterated
solvent, the buffer and DTT were exchanged with D2O as described
above, and the final D2O solution was diluted with CH3OD. Solutions
were equilibrated at 15 °C and photolyzed with a 1 µs pulse of 320
nm laser light, and the absorption at 380 nm was recorded as previously
described.2
Identification and Quantification of 3-Methylanthranil from
Photolysis of Caged ATP in the Presence of DTT. An aliquot (20
µL) of a solution of caged ATP 1a (9.8 mM) and DTT (20 mM) in Na
phosphate buffer (25 mM, pH 7.0) was placed in a 25 µL cell at 19 °C
and irradiated with a single 50 ns pulse of 347 nm light from a
frequency-doubled ruby laser (Laser Applications Inc., Winter Park,
FL). The average energy of the light flash was 93 mJ (range 70-110
mJ). Six irradiated aliquots were combined and analyzed by anion
exchange and reverse-phase HPLC. HPLC mobile phase flow rates
were 1.5 mL min-1, and compositions were as follows: for anion
exchange [Whatman Partisphere SAX column, Cat. no. 4621-0505;
detection at 254 nm, 0.25 M (NH4)H2PO4 (adjusted to pH 5.5 with 2
M NaOH) plus 13% MeOH (v/v)]; for reverse-phase [Merck Lichro-
sphere RP8 column, Cat. No. 50832; detection at 313 nm, 25 mM NaH2-
PO4 (adjusted to pH 6.5 with 2 M NaOH)-MeOH (3:2 v/v)]. Retention
times (min) were as follows: (anion exchange), caged ATP 3.4, ATP
14.0; (reverse-phase), 3-methylanthranil 11.7. The different analytes
caused no mutual interference on their respective columns.
2,2′-Diacetylazoxybenzene 10. Dithiothreitol (75 mg, 5 mmol) was
added in one portion to a stirred solution of 2-nitrosoacetophenone (250
mg, 1.67 mmol) and triethylamine (505 mg, 5 mmol) in chloroform
(50 mL). After 10 min, the solution was washed with water, dried,
and concentrated. The residue was flash chromatographed [EtOAc-
hexanes (20:80)] to give 10: mp 115-116.5 °C (from EtOAc-
hexanes); UV (EtOH) λmax nm (ꢀ M-1 cm-1) 243sh (13 100), 313
(5700); IR (Nujol) 1690, 1685, 1595, 1265, 765 cm-1; 1H NMR (CDCl3)
δ 2.56 (s, 3H), 2.59 (s, 3H), 7.27-7.83 (m, 7H), 8.04-8.19 (m, 1H).
Anal. Calcd for C16H14N2O3: C, 68.07; H, 5.00; N, 9.92. Found: C,
67.86; H, 4.95; N, 9.89.
Reaction of 2-Nitrosoacetophenone with DTT Studied by UV
Spectroscopy. A solution of 2-nitrosoacetophenone (1 mL, 1.18 ×
10-4 M) in 10 mM Na phosphate, pH 7.0-EtOH (9:1) was equilibrated
at 35 °C in a 1 cm path length quartz cuvette placed in a Beckman
DU70 spectrophotometer, and the absorbance spectrum was recorded.
An aliquot (25 µL) of a solution of DTT (200 mM) in 10 mM Na
phosphate, pH 7.0 was added, and the spectrum was recorded within
30 s. Incubation was continued for 1 h, and a further spectrum was
recorded. In a subsequent experiment, this protocol was repeated, but
absorbance at 316 nm was monitored continuously.
Ring-Chain Tautomerism of 2-Hydroxylaminoacetophenone.
Solutions of 20 mM Tricine were adjusted with 1 M NaOH to pH values
of 7.5, 8.0, and 8.5, lyophilized, and redissolved in D2O to restore the
original concentration. Each solution was made to 5 mM with DTT.
Aliquots (30 µL) of a CDCl3 solution of 2-nitrosoacetophenone (100
mM) were diluted with CD3OD (470 µL), and these solutions were
mixed with aliquots (500 µL) of the buffered DTT solutions. Each
solution for NMR was prepared immediately before the measurements
began and equilibrated at 15 °C in the probe of a Bruker AM 400WB
NMR spectrometer. Spectral data were processed and analyzed using
the Varian VNMR 5.1 software package31 running on an Axil
Sparcstation 10/40 clone. The experimental protocol was designed to
correct for the slow dehydration of the 7/8 mixture to 3-methylanthranil.
Thus for each sample a series of interleaved FIDs was collected with
irradiation of the signal at δ 1.83 for each of 20 saturation times in the
range 0.1-2.6 s, applied in randomized order. Two further FIDs were
interleaved, again in randomized order, either with continuous saturation
or with the decoupler gated off, corresponding to infinite and zero
saturation times, respectively. The sequence was repeated to give a
total of eight transients for each saturation time. A delay >5 × T1
was inserted prior to each saturation pulse to eliminate fast-pulsing
artifacts. T1 was estimated using the inversion recovery method.32 Plots
Concentrations of ATP and 3-methylanthranil in the photolyzed
solution were determined by comparison of chromatographic peak
heights with those from reference solutions of the authentic compounds.
In the reverse-phase chromatogram of the photolyzed solutions, minor
peaks with retention times 3.6, 7.0, and 8.4 min were present in addition
to the main peak for 3-methylanthranil. The first of these peaks
disappeared as the peak for 3-methylanthranil stabilized.
Acknowledgment. We are grateful to Dr. M. R. Webb for
the gift of [â,âγ-18O3]ATP. We thank the MRC Biomedical
NMR Centre for access to facilities, Dr. J. Feeney for valuable
discussions on the saturation-transfer experiment, and Dr. V.
R. N. Munasinghe for help with NMR spectroscopy. This work
was supported by an ARC grant from the Deutscher Akade-
mischer Austauschdienst and the British Council, and by a
fellowship from the Ciba-Geigy Foundation (to Y.M.).
(31) VNMR Version 5.1; Varian Associates, Palo Alto, CA, 1995.
(32) Harris, R. K. Nuclear Magnetic Resonance Spectroscopy; Pitman,
London 1983; p 81.
JA964430U
(33) Kaleidagraph; Synergy Software, Reading, PA, 1994.