Photoionization of organic phosphates by 193 nm laser light in aqueous solution: Rapid intramolecular H-transfer to the primarily formed phosphate radical. A model for ionization-induced chain-breakage in DNA?

Article abstract of DOI:10.1021/ja974393m

In aqueous solution, 193 nm (6.4 eV) photolysis of inorganic and organic phosphates such as ribose-5-phosphate leads to ionization with formation of the corresponding oxygen-centered phosphate radicals, O₃PO. These (oxidizing) radicals function as traps with respect to hydrogens attached to α-, β-, or, possibly, γ-carbons, whereby in the case of the β-hydrogens a six-membered transition state for transfer of the hydrogen to the phosphate oxygen is possible, leading to high rate constants (up to >5 x 10⁷ s^-1) for H-transfer in these unimolecular reactions. In the case of (deoxy)ribosephosphates the six-membered transition state is possible for transfer of the hydrogen at C4 to the phosphate group at C5 as well as at C3. In DNA, the resulting C4'-radical will undergo a rapid β-elimination of the phosphate-ester group, this step representing the DNA chain break. The apparently easy H-transfer from a carbon to a phosphate radical, by which these radicals are 'repaired', is why phosphate radicals are not observed in irradiated DNA. Insofar as hereby the C4'-radical is formed, the mechanism of DNA chain breakage is the same for the 'direct' and the 'indirect' effect.

Full text of DOI:10.1021/ja974393m

3928
J. Am. Chem. Soc. 1998, 120, 3928-3934
Photoionization of Organic Phosphates by 193 nm Laser Light in
Aqueous Solution: Rapid Intramolecular H-Transfer to the Primarily
Formed Phosphate Radical. A Model for Ionization-Induced
Chain-Breakage in DNA?
Steen Steenken* and Ludmila Goldbergerova
Contribution from the Max-Planck-Institut fu¨r Strahlenchemie, D-45413 Mu¨lheim, Germany
ReceiVed December 29, 1997
Abstract: In aqueous solution, 193 nm (6.4 eV) photolysis of inorganic and organic phosphates such as ribose-
5-phosphate leads to ionization with formation of the corresponding oxygen-centered phosphate radicals,
O3PO•. These (oxidizing) radicals function as traps with respect to hydrogens attached to R-, â-, or, possibly,
γ-carbons, whereby in the case of the â-hydrogens a six-membered transition state for transfer of the hydrogen
to the phosphate oxygen is possible, leading to high rate constants (up to >5 × 107 s-1) for H-transfer in these
unimolecular reactions. In the case of (deoxy)ribosephosphates the six-membered transition state is possible
for transfer of the hydrogen at C4 to the phosphate group at C5 as well as at C3. In DNA, the resulting
C4′-radical will undergo a rapid â-elimination of the phosphate-ester group, this step representing the DNA
chain break. The apparently easy H-transfer from a carbon to a phosphate radical, by which these radicals are
“repaired”, is why phosphate radicals are not observed in irradiated DNA. Insofar as hereby the C4′-radical
is formed, the mechanism of DNA chain breakage is the same for the “direct” and the “indirect” effect.
Introduction
acterized.6 In many studies, very little evidence for sugar
(deoxyribose) and no evidence for phosphate-centered radicals
was obtained, although, on the basis of the relatively high mass-
fraction (0.6) of deoxyribose and phosphate in DNA, these
radicals are expected to be formed in appreciable yields by the
unselective ionizing radiation. It has therefore been suggested
that a “spin-transfer” takes place leading, finally, to a base-
centered radical.1,7 To study a (potential) spin-transfer from
phosphate to the deoxyribose moiety and to obtain information
about a chain breakage mechanism starting out from a directly
ionized phosphate ester function, the present experiments were
performed.
Despite more than a generation of radiation chemists’ and
biologists’ efforts,1 the understanding of the mechanistic details
of the radiation or radical-induced chain breakage of DNA is
still incomplete. This is particularly true with respect to the
“direct effect”,2 which appears to be at least as important as
the “indirect effect”, which involves the mediation by the •OH
radical (produced by irradiation of the solvent, H2O). Concern-
ing the indirect effect, a mechanism has been proposed (the
C4′-mechanism)3 which explains well all experimental observa-
tions and which has recently been supported by experiments
on DNA-model compounds (oligomers).4 In contrast, the
mechanism by which the direct effect operates, i.e., leads to
the lethal chain breaks, is not at all well understood. On the
basis of EPR-experiments,1a,c,d,g the radiation damage is localized
mainly on the bases, the reductiVe equivalent on the pyrimidines
(initially at cytosine (C), at higher temperatures at thymine
(T))1a,c,d,g,5 and the oxidatiVe equivalent preferentially on guanine
(G), whose (deprotonated) radical cation has been well char-
Results and Discussion
1. UV-Absorption Spectra of Phosphates. In Figure 1 are
presented the absorption spectra in aqueous solution of the
phosphate dianion, HPO42-, and of the organic phosphates,
n-butyl phosphate and triethyl phosphate, and in Figure 2 those
of 2-hydroxyethyl phosphate (ethylene glycolphosphate) and
ribose-5-phosphate.
(1) For reviews on DNA, see, e.g., (a) Bernhard, W. A. AdV. Radiat.
Biol. 1981, 9, 199. (b) Steenken, S. Chem. ReV. 1989, 89, 503. (c) Close,
D. M. Magn. Res. ReV. 1991, 15, 241. (d) Hu¨ttermann, J. In Radical Ionic
Systems: Properties in Condensed Phases; Lund, A.; Shiotani, M. Klu-
wer: Dordrecht 1991; p 435. (e) Steenken, S. Free Rad. Res. Commun.
1992, 16, 349. (f) O’Neill, P.; Fielden, E. M. AdV. Radiat. Bio. 1993, 17,
53. (g) Becker, D.; Sevilla, M. D. AdV. Radiat. Bio. 1993, 17, 121. (e)
Steenken, S. Biol. Chem. 1997, 378, 1293.
2-
With HPO4 and the monoalkyl phosphates, the ꢀ at 193
nm is between 100 and 180 M-1 cm-1. The absorption band
of the phosphate function, which extends into the region <187
nm, where measurements with our equipment were not possible,
has been assigned to charge transfer to solvent (CTTS).8
(2) See, e.g.; Le Sech, C.; Frohlich, H.; Saint-Marc, C.; Charlier, M.
Radiat. Res. 1996, 145, 632.
(5) Barnes, J. P.; Bernhard, W. A. J. Phys. Chem. 1995, 99, 11248.
(6) (a) In aqueous solution: Candeias, L. P.; Steenken, S. J. Am. Chem.
Soc. 1989, 111, 1094. Steenken, S.; Jovanovic, S. V. J. Am. Chem. Soc.
1997, 119, 617. (b) In the solid state: Hole, E. O.; Nelson, W. H.; Sagstuen,
E.; Close, D. M. Radiation Res. 1992, 129, 119.
(7) In the case of thymidine-6-yl radical, the feasibility of the reVerse
reaction has recently been studied with the result that its rate is low (<2
s-1): Barvian, M. R.; Barkley, R. M.; Greenberg, M. M. J. Am. Chem.
Soc. 1995, 117, 4894.
(3) Dizdaroglu, M.; von Sonntag, C.; Schulte-Frohlinde, D. J. Am. Chem.
Soc. 1975, 97, 2277. For reviews, see: Schulte-Frohlinde, D. In Mechanisms
of DNA Damage and Repair; Simic, M. G., Grossman, L., Upton, A. C.,
Eds.; Plenum: New York, 1980; p 19. Schulte-Frohlinde, D. In Radiation
Protection and Anticarcinogens; Nygaard, O. F., Simic, M. G., Eds.;
Academic: New York, 1983; p 53.
(4) For a review, see: Giese, B.; Beyrich-Graf, X.; Erdmann, P.; Petretta,
M.; Schwitter, U. Chem. Biol. 1995, 2, 367.
(8) Halman, M. Top. Phosph. Chem. 1967, 4, 49.
S0002-7863(97)04393-X CCC: $15.00 © 1998 American Chemical Society
Published on Web 04/11/1998

Photoionization of Organic Phosphates
J. Am. Chem. Soc., Vol. 120, No. 16, 1998 3929
2-
2-
Figure 1. UV spectra of HPO4 (filled circles), n-BuOPO3 (open
circles) and (EtO)3PO (triangles) in aqueous solution pH 7-8.
Figure 4. Spectra recorded at 125 ns (open circles), 0.8 µs (closed
circles), and 8 µs (triangles) on 248 nm photolysis of an oxygenated
25 mM solution of Li4P2O8 in water, pH 9.2.
•-
circles) on removal of e- by O2, whereby O2 is formed
aq
whose absorption (λmax ) 241 nm)12 adds on top of that of
•-
HPO4 (Figure 3, filled circles).
The formation of HPO4•- was checked by producing it inde-
pendently (see Figure 4), i.e., by 248 nm photolysis of the per-
oxide, -HO3P-O-O-PO3H-, reaction 1:
hν
-HO3P-O-O-PO3H-
8 2HPO4•- Φ ) 0.8 13 (1)
248 nm
From a comparison of Figure 3, filled circles, with Figure 4
it is apparent that the phosphate radical, HPO4•-, is the product
of the ionization of HPO42-. From a quantitative inspection of
Figure 3 it is also clear that photoionization cannot be the only
photoreaction of HPO42- induced by the 193 nm light. On the
basis of the optical density (filled circles) at 515 nm, measured
using an oxygenated (to scavenge e-aq) aqueous solution of
HPO42-, compared with the optical density at 700 nm (open
2-
Figure 2. Absorption spectra of of HOCH2CH2OPO3 (circles) and
of ribose-5-phosphate (triangles) in water, pH 8.
circles), due to e-aq, and taking account of the extinction
•-
coefficients of HPO4 at 515 and of e- at 720 nm (19000
aq
M-1cm-1),9 the photochemical yield of HPO4 compared to
•-
that of e- turned out to be 224%. Therefore, there must be
aq
an additional source of HPO4•-, other than ionization (eq 2a)
of HPO42-. A reasonable path to HPO4•- would be homolysis
of the O-H bond, eq 2b:
Figure 3. Absorption spectra recorded at 0.1 µs after 193 nm photolysis
of 20 mM Na2HPO4 in deoxygenated (open circles) and at 0.8 µs in
oxygenated (filled circles) aqueous solution. The latter spectrum has
been magnified by the factor 5.
2. Laser Flash Photolysis (LFP) of Phosphates. Inorganic
Phosphate and Simple Alkyl Phosphates. On photolysis of
the phosphates with 20 ns pulses of 193 nm laser light from a
ArF-excimer laser, the hydrated electron, e-aq, was formed, as
evidenced by its characteristic9 absorption spectrum with λmax
at ≈700 nm and its reactivity with e-aq-scavengers.10 E.g., in
Figure 3 is shown the spectrum observed on 193 nm photolysis
The photon energy of 193 nm light (6.4 eV ) 148 kcal/mol)
is sufficient to break an O-H bond (110-120 kcal/mol). The
formation of the phosphate radicals via 2a amounts to 45% and
to 55% via 2b. The pKa-value of HPO4•- has been determined11
to be 8.9.
Similar experiments were performed with a series of simple
organic phosphates. In Figure 5, a and b, are shown the spectra
obtained on 193 nm photolysis of an aqueous solution of methyl
2-
of a 20 mM solution of HPO4 in the absence (open circles)
and presence of O2 (filled circles). In the absence of O2, the
broad band extending from ca. 400 nm to g700 nm is that of
e-aq, whereas the weaker band at ≈225 nm is due to the
phosphate radical, HPO4•-, whose main absorption band (at 515
(11) Maruthamuthu, P.; Neta, P. J. Phys. Chem. 1978, 82, 710.
(12) Czapski, G.; Dorfman, L. J. Phys. Chem. 1964, 68, 1169.
(13) Quantum yield for decomposition of the peroxide at pH 9.2,
measured by comparison with that (0.9, cf. Faria, J. L.; Steenken, S. J.
Phys. Chem. 1992, 96, 10869) for decomposition of S2O82- and taking (from
nm with ꢀ ) 1550 M-1cm-1 11
) becomes visible (Figure 3, filled
•-
ref 11) ꢀ(HPO4
)
515 nm ) 1550 M-1 cm-1. The value 0.8 was also obtained
(9) Hug, G. E. Nat. Stand. Ref. Data Ser. Nat. Bur. Stand. 1981, 69, 6.
(10) Buxton, G. V.; Greenstock. C. L.; Helman, W. P. J. Phys. Chem.
Ref. Data 1988, 17, 513.
if the photoionization in water of I- with 248 nm light (Φ ) 0.29; Jortner,
J.; Levine, R.; Ottolenghi, M. Stein, G. J. Phys. Chem. 1961, 65, 1232, see
also ref 14b) was used for actinometry.

3930 J. Am. Chem. Soc., Vol. 120, No. 16, 1998
Steenken and GoldbergeroVa
Figure 6. Absorption spectra recorded at 350 ns after the 193 nm
laser pulse on photolyzing oxygenated aqueous solutions at pH 8 of
methyl phosphate (filled circles), ethyl phosphate (open circles), n-butyl
phosphate (open triangles), and diisopropyl phosphate (filled triangles).
In the inset is shown the (first-order) decay of EtOPO3•-, produced
by photolysis of an oxygenated 8 mM solution of ethyl phosphate at
pH 8.
Figure 5. (a) Time-resolved spectra recorded on 193 nm photolysis
2-
of a deoxygenated aqueous solution of 12 mM MeOPO3 at pH 8.
Filled circles: 275 ns, open circles: 775 ns, filled triangles: 2.75 µs,
open triangles: 7.75 µs after the flash. Inset: Time-dependence of the
absorption at 540 nm and at 720 nm. (b) Same solution as in (a) but
containing 1 mM O2 to scavenge e-aq. Filled circles: 350 ns, open
circles: 1.55 µs, filled triangles: 5.5 µs, open triangles: 15.5 µs after
the flash.
Figure 7. Time-dependent absorption spectra observed on 193 nm
photolysis of a deoxygenated 10 mM aqueous solution of ribose-5-
phosphate, recorded at 150 ns (filled circles), 775 ns (open circles),
1.5 µs (filled triangles), and 7.75 µs (open triangles) after the pulse
phosphate, CH3OPO32-, in the absence (a) and in the presence
of oxygen (b). In Figure 5a, the presence of e- is clearly
aq
(≈30 mJ). Inset: yield of e- as a function of laser power/mJ.
aq
seen, as is a peak at ≈240 nm, indicative of a phosphate-cen-
tered radical. If the electron is removed by reaction with O2,
the presence of a species absorbing at ≈520 nm is evident
(Figure 5b).
Table 1. Quantum Yields for the 193 nm Photoionization of
Phosphates in Aqueous Solution
a
phosphate
φ(e-
)
aq
On the basis of the similarity of its spectrum with that of
HPO4•-, the species is identified as the methyl phosphate radical,
CH3OPO3•-. As seen in the inset in Figure 5b, the species
decays with a first half-life of ca. 2 µs (which is much shorter
than that of the inorganic phosphate radical, see inset in Figure
4) in a reaction which is predominantly second order, but which
has a first-order component, the rate of which turns out to be
2-
HPO4
0.53b
monomethyl phosphatec
monoethyl phosphatec
mono-n-butyl phosphatec
ethyleneglycolphosphatec
glycerol-1-phosphatec
glycerol-2-phosphatec
ribose-5-phosphatec
0.61
0.65
0.58
0.43
0.61
0.55
0.46 (0.47)b
0.42
2-deoxyribose-5-phosphatec
smaller than 105 s-1
.
H2PO4
0.33b
-
On photolysis of ethyl phosphate under the same conditions,
there also was a band at ca. 520 nm, which is assigned to the
phosphate radical, EtOPO3•-, but its intensity at 350 ns after
the pulse was only about half that in the case of methyl
ribose-5-phosphate monoanion
diethyl phosphate
diisopropyl phosphate
diethyleneglycolphosphate
triethyl phosphate
0.24b
0.11 (pH 6)
0.35
0.29
•-
0.085
phosphate (see Figure 6). It was found that the radical EtOPO3
had a much shorter lifetime than MeOPO3 or HPO4
•-
•-
a Error (10%. b From ref 15. c Dianion. d pH ≈ 8.
aqueous solution as an actinometer with φ(e-aq) ) 0.46,14 the
quantum yields for photoionization as listed in Table 1 were
obtained.
As is evident from Table 1, for the mono-alkyl phosphates
the quantum yields are in the range 0.42-0.65, which is similar
,
decaying with first-order kinetics (see inset in Figure 6) with k
) 7.9 × 105 s-1 as measured at 455-560 nm.
When n-butyl phosphate was photolyzed, the 520 nm band
was hardly discernible. A similar observation was made in the
case of diisopropyl phosphate (see Figure 6).
In the 193 nm ionization of the phosphates, the yields of e-
aq
(14) (a) Dainton, F. S.; Fowles, P. Proc. R. Soc. 1965, 287, 312. See
also: (b) Jortner, J.; Ottolenghi, M.; Stein, G. J. Phys. Chem. 1964, 68,
247. However, there is also a more recent value for photoionization of Cl-
(Φ ) 0.41): Iwata, A.; Nakashima, N.; Kusaba, M.; Izawa, Y.; Yamanaka,
C. Chem. Phys. Lett. 1993, 207, 137.
were in all cases proportional to the laser light intensity (see,
e.g., inset in Figure 7), so the process leading to the production
of e- (ionization) is monophotonic. From the linear e-
-
aq
aq
yield vs laser-power plots, taking the photoionization of Cl- in

Photoionization of Organic Phosphates
J. Am. Chem. Soc., Vol. 120, No. 16, 1998 3931
Table 2. Products and Their Quantum Yieldsafrom the 193 nm
Photolysis of HOCH2CH2OPO3 at pH 8b
2-
product
φ/Ar
φ/O2
e-
H2
0.43
0.09
0.69
aq
e 0.005
0.46
0.02
inorganic phosphate
acetaldehyde
0.70
glycolaldehyde
formaldehyde
succinaldehyde
0.008
0.005
≈0.02
0.13
0.04
traces
a Error limits (10%. b The power per pulse was 15-16 mJ.
to that (0.53) for the phosphate dianion, HPO42-. There is
obviously only a very small influence of the nature of the
organic group on the quantum yield, which indicates that the
Figure 8. Dependence of the quantum yields for formation in
deoxygenated solution of inorganic phosphate (filled circles) and
acetaldehyde (open circles) on the light intensity per pulse, measured
as mJ/pulse.
photogenerated e- stems from the phosphate and not from
aq
the alkyl function. For the di-alkyl phosphates, the ionization
quantum yields appear to be somewhat smaller (average ) 0.25
( 0.12), and an even smaller number is observed for tri-ethyl
phosphate (φ ) 0.085). Since alkylation of the phosphate anion
corresponds to protonation and since protonation reduces the
the experiments described in Figure 8, the total dose was kept
constant at 1200 mJ).
This dose-rate dependence is indicative of a chain reaction
occurring. On this basis, the results so far presented are ex-
plained in terms of the following reactions:
electron density of an anion, the observed decrease of φ(e-
)
aq
with increasing alkylation reflects the corresponding decrease16
in the ionization potential. Concerning DNA, in this molecule
are, of course, present dialkyl phosphate groups () monoanions)
which should have higher ionization potentials than the monoalkyl
phosphate dianions discussed in this study.
hν
2-
HOCH2CH2OPO3
8 HOCH2CH2OPO3•- + e-
,
aq
193 nm
φ ) 0.43 (3)
HOCH2CH2OPO3•- f HOC•HCH2OPO32- + H+ (4)
(Deoxy)ribose-5-phosphate and 2-Hydroxyethyl Phos-
phate. In Figure 2 is contained the absorption spectrum of
ribose-5-phosphate in aqueous solution at pH 8 (natural pH).
Visible is the phosphate band with λmax e 187 nm. On
irradiation into this band with 193 nm light, a strong signal due
k g 5 × 107 s-1
HOC•HCH2OPO32- f OCHC•H2 + HPO4
2-
(5)
to e- was seen (Figure 7), whose half-life is ca. 0.3 µs. As
aq
shown in the inset, the formation of e-aq is monophotonic. From
k g 2.3 × 106 s-1
the slope of this dependence, compared with that from Cl-,
φ(e-
)
) 0.46. Based on the ∆OD at ≈ 515
OCHC•H2 + HOCH2CH2OPO3 slow8
2-
aq ribose-5-phosphate
nm in oxygenated solution, there was no evidence for the
presence of a phosphate radical, which means that its lifetime
is e 20 ns () pulse length of laser).
CH3CHO + HOC•HCH2OPO3 (6)
2-
The first step (eq 3) is the photoionization of the phosphate
function in EGP, leading to the phosphate radical where the
unpaired spin is probably strongly localized on an oxygen
atom.17 This oxyl-radical must be very short-lived, since, even
immediately after the 20 ns pulse, spectroscopic evidence for
it (expected at ≈ 520 nm) was not found. As the reason for
this, we suggest a very rapid intramolecular H-atom shift, from
the â-carbon to the phosphato oxygen (eqs 4 and 7). On the
basis of the nonobservability of the -OPO3•- signal at 520 nm,
the rate constant for this reaction must be g1/20 ns ) 5 × 107
s-1. As the product of this 1,5 H-shift, which can proceed
through a six-membered ring, a carbon-centered R-hydroxy-â-
With 2-hydroxyethyl phosphate, 193 nm photoionization
occurs with a quantum yield of 0.43, which is similar to that
(0.46) for ribose-5-phosphate. Despite this relatively high quan-
tum yield as reflected by a strong signal of e-aq, there was no
evidence, based on the signal at ≈515 nm, for the presence of
a phosphate radical. This means that the phosphate radical
necessarily produced in the ionization step must have a lifetime
of <20 ns.
3. Product Analysis Results. HOCH2CH2OPO32-. For
HOCH2CH2OPO32-, it was possible to perform a quantitative
product analysis study, the results of which are presented in
Table 2 and Figure 8.
As seen from Table 2, in the absence of O2, the main
productssother than e-aq and H2sare inorganic phosphate and
acetaldehyde, whose yields are equal. The small yield of
glycolaldehyde probably indicates traces of oxygen in the
solution. This is concluded from the fact that the yield of
glycolaldehyde increases by the factor 16 in going from an Ar-
to an O2-saturated solution. The effect of varying the dose/
pulse was also investigated, the results of which are shown in
Figure 8. From Figure 8 it is evident that the yield of inorganic
phosphate and of acetaldehyde depends on the intensity per pulse
of the 193 nm light, increasing with decreasing dose/pulse (in
(7)
phosphato radical is formed (eq 4). Radicals of this type have
(15) Candeias, L. P.; Steenken, S. J. Am. Chem. Soc. 1992, 114, 699.
(16) See Tasaki, K. Yang, X.; Urano, S.; Fetzer, S.; LeBreton, P. R. J.
Am. Chem. Soc. 1990, 112, 538.
(17) For ESR spectra of phosphate radicals see, e.g.: Subramanian, S.;
Symons, M. C. R.; Wardale, H. W. J. Chem. Soc. (A) 1970, 1239.

3932 J. Am. Chem. Soc., Vol. 120, No. 16, 1998
Steenken and GoldbergeroVa
Concerning the possibility of intramolecular H-transfer from
a C to a phosphate oxyl radical, it may be relevant that the
•-
analogous intermolecular reaction, i.e., that between HPO4
/
•2-
PO4 and ribose or deoxyribose, does in fact exist, the rate
constants measured at pH 924 being 9 × 107 and 7.5 × 107
M-1 s-1 25
.
Support for the intramolecular transfer of an H• from a C
to a phosphate oxygen (eq 4/7) comes from ESR-measure-
ments of γ-irradiated silver diethyl phosphate, where the
-
•CH2CH2O(CH3CH2O)PO2 radical was seen and interpreted
as resulting from H-transfer to the initially formed radical
• 26
oxidation product (EtO)2PO2 . Intramolecular H-transfers have
been found also in other alkyl phosphates.27 Of relevance is
also the fact28 that in the mass spectra of trialkyl phosphates,
the positive molecular ion decomposes very rapidly such that
its contribution to the recorded spectrum is almost zero. In the
case of tributyl phosphate, this has been interpreted as resulting
from an intramolecular H-transfer from a carbon to the oxygen
of the P-oxyl group.29
•
Figure 9. Absorption spectrum measured on reaction of OH with
1 mM EGP at pH 8 (a). Comparison with absorption spectrum of
•CH2CHO (b; from ref 20).
been shown18 to undergo a rapid, heterolytic elimination of
inorganic phosphate to yield â-oxoalkyl radicals which are
oxidizing in character.19 In this reaction (eq 5), the formyl
methyl radical is produced which has been shown20 to be able
to abstract an H-atom from R-hydroxyalkanes. If this hap-
pens (eq 6), a chain reaction is started, the chain carrier being
•CH2CHO, the products of the chain reaction are acetaldehyde
and inorganic phosphate, in equal yields, as experimentally
observed (see Table 2).
As evident, the H-transfer mechanism eq 7 involves a six-
membered transition state, which is a way possible only for
alkyl phosphates with R g Et. For R ) CH3, the transition
state necessarily involves the less favorable five-membered ring,
and this, together with the higher C-H bond energy of a pri-
mary vs a secondary alkyl group, is probably the reason why
the rate constant for H-transfer in methyl phosphate (e 105 s-1
)
is lower than that with ethyl- (7.9 × 105 s-1) or the higher
phospates (>5 × 107 s-1). In the case of 2-hydroxyethyl
phosphate, H-abstraction from the 2-position should be strongly
enhanced by the activating effect of the OH-group. In agree-
ment with this is the fact that k(abstraction) is in fact larger
•
In the presence of oxygen, the chain carrier, CH2CHO, is
scavenged, as is visible from the dramatic decrease in the yield
of CH3CHO and the increase in the yield of glycolaldehyde.
Under this condition, the yield of inorganic phosphate should
be equal to that of the primarily produced (by the ionization)
phosphate radicals, and this is in fact the case, as seen from
Table 2 (φ(phosphate)oxygen ) 0.46 as compared to φ(e-aq) )
0.43).
than 5 × 107 s-1
.
Ribose-5-phosphate. Product analysis experiments were also
performed after 193 nm photolysis of ribose-5′-phosphate at pH
8. It was found that in aqueous deoxygenated solution (Ar or
N2O-saturated) the quantum yield of inorganic phosphate was
0.2 which dropped to 0.08 when O2 (1 mM) was admitted. Of
organic compounds in deoxygenated solution, the main pho-
tolysis products were 5-deoxypentos-4-ulose, pentos-4-ulose,
and pentodialdose. In the presence of O2 (1 mM), the yield of
5-deoxypentos-4-ulose dropped to below the detection limit,
whereas those of pentos-4-ulose and pentodialdose remained
approximately the same.
To check the formation of the formylmethyl radical from
ethyleneglycolphosphate, pulse radiolysis experiments were
performed with an N2O-saturated aqueous solution of 1 mM
HOCH2CH2OPO32-. Under this condition, the OH-radicals
produced by the pulse are expected to abstract an H-atom from
a carbon atom of the substrate, preferably from the â-carbon
relative to the phosphate group. This carbon is activated by
the OH-group it carries. The resulting R-hydroxy-â-phosphato
radical is the same as that from the intramolecular H-atom shift
to the phosphato radical (eq 4/7) and is expected to rapidly
undergo the heterolytic â-phosphate elimination reaction 5
yielding the formylmethyl radical. In Figure 9 the measured
spectrum is compared with that of the authentic21 formylmethyl
radical: The similarities between the two spectra with respect
to band position and epsilon are satisfactory.22 The rate constant
These results can be explained (Schemes 1 and 2) by
photoionization of the phosphate group (φ ) 0.46) followed
by intramolecular H-abstraction by the ionization-produced
phosphato-radical from C4 (main reaction), and from C5.
H-abstraction from C4 is via a six-membered transition state,
as shown in Scheme 1.
2-
measured for the reaction of OH• with HOCH2CH2OPO3 is
The resulting C4-radical then eliminates a phosphate anion
2.3 × 109 M-1s-1, similar to that (2.4 × 109 M-1s-1 21
)
for
in a heterolytic fashion. This is the famous C4′-mechanism3,4
reaction with ethyleneglycol. From the fact that there was no
(21) Bansal, K. M.; Gra¨tzel, M.; Henglein, A.; Janata, E. J. Phys. Chem.
1973, 77, 7, 16.
•
delayed formation of absorption due to CH2CHO in a 1 mM
solution of HOCH2CH2OPO32- at pH 8 it is concluded that the
rate constant for (the intramolecular) elimination of phosphate
(22) The fact that in the HOCH2CH2OPO32- case the absorption at <280
nm is stronger than that of the formylmethyl radical may be due to the
additional formation of the R-phosphatoalkyl radical from the reaction of
from the â-phosphato radical (eq 5) is g2.3 × 109 M-1 s-1
×
HOCH2CH2OPO3 with OH•.
2-
10-3 M ) 2.3 × 106 s-1, a value in agreement with analogous
(23) See, e.g., ref 18 and Behrens, G.; Koltzenburg, G.; Schulte-Frohlinde,
cases.23
D. Z. Naturforsch. 1982, 37c, 1205.
(24) Due to pKa(HPO4•-) ) 8.9,10 the rate constant refers mainly to the
•2-
(18) Samuni, A.; Neta, P. J. Phys. Chem. 1973, 77, 1425. Steenken, S.;
Behrens, G.; Schulte-Frohlinde, D. Int. J. Radiat. Biol. 1974, 25, 205. See,
also: Steenken, S. In Free Radicals: Chemistry, Pathology and Medicine;
Rice-Evans, C., Dormondy, T., Eds.; Richelieu Press: London, 1988; p
51.
less reactive form, i.e., PO4
.
(25) Nakashima, M.; Hayon, E. J. Phys. Chem. 1970, 74, 3290.
(26) Bernhard, W. A.; Ezra, F. S. J. Phys. Chem. 1974, 78, 958.
(27) Haase, K. D.; Schulte-Frohlinde, D.; Kourim, P.; Vacek, K. Int. J.
Radiat. Phys. Chem. 1973, 5, 351.
(19) Steenken, S. J. Phys. Chem. 1979, 83, 595.
(20) von Sonntag, C.; Thoms, E. Z. Naturforsch. B 1970, 25, 1405.
Burchill, C. E.; Perron, K. M. Can. J. Chem. 1971, 49, 2382.
(28) McLafferty, F. W. Anal. Chem. 1956, 28, 306; Quayle, A. In
AdVances in Mass Spectrometry; Pergamon: London, 1959; p 365.
(29) .Wilkinson, R. W.; Williams, T. F. J. Chem. Soc. 1961, 4098.

Photoionization of Organic Phosphates
J. Am. Chem. Soc., Vol. 120, No. 16, 1998 3933
Scheme 1
Scheme 2
of DNA-chain breakage. The C4′-radical can be produced also
by H-abstraction via a six-membered transition state if the
phosphate group in the 3′-position is ionized. On this basis
the C4′-radical is quite a likely radical to be formed (and
therefore a chain break induced) on ionization of a phosphate
group in DNA.30 Obviously, by this reaction the phosphate
radical is “repaired”. In this connection it is relevant that
phosphate radicals, in contrast to sugar radicals, have never been
seen in γ-irradiated nucleotides.6b,31 If irradiated with densely
ionizing particles, evidence has been found also in DNA for
sugar but not for phosphate radicals.32 Furthermore, Scheme
1 (and 2) provides a mechanistic basis for the observation33 that
the frank strand breaks induced by <200 nm UV-irradiation of
plasmid DNA result from excitation of the sugar-phosphate
backbone. The strand breaks induced in model DNA com-
pounds by monochromatic soft X-rays34 can be explained on
the same basis.35
To explain the formation of the product pentodialdose, it is
necessary to assume abstraction by the phosphate radical of the
H at C5, a reaction which involves a five-membered ring,
followed by oxidation, addition, and elimination, as shown in
Scheme 2.
(30) To which extent these concepts, which are based on simple, low-
molecular weight model compounds, whose intramolecular motions are not
restricted, can be applied to the rather stiff DNA polymer remains to be
seen. It is obvious that model calculations on H-transfer to phosphate radicals
at C3′ or C5′ from “adjacent” deoxyribose C-H groups would be very
helpful.
(32) Becker, D.; Razskazovskii, Y.; Sevilla, M. D. Radiat. Res. 1996,
146, 361.
(33) Gut, I. G.; Farmer, R.; Huang, R. C.; Kochevar, I. E. Photochem.
Photobiol. 1993, 58, 313.
(31) Sevilla, M. D.; Becker, D. Electron Spin Resonance, A Specialist
Periodical Report 1994, 14, 130.
(34) Ito, T.; Saito, M.; Kobayashi, K. Int. J. Radiat. Biol. 1992, 62, 129.
(35) The authors, however, propose a different mechanism.

3934 J. Am. Chem. Soc., Vol. 120, No. 16, 1998
Steenken and GoldbergeroVa
In principle, it is also possible that a phosphate radical is
repaired intramolecularly by electron transfer from a base, e.g.,
G, as indicated:
possible for H-transfer from C4 to the phosphate group at C5
as well as at C3. In DNA, the resulting C4′-radical will undergo
a rapid â-elimination of the phosphate-ester group, this step
representing the DNA chain break. The apparently easy
H-transfer from a carbon to a phosphate radical, by which these
radicals are repaired, is probably why phosphate radicals are
not observed in irradiated DNA. A consequence of the
involvement of the C4′-radical is that the mechanism of DNA
chain breakage is the same for the direct and the indirect effect.
Experimental Section
Inorganic phosphate (analytical grade) was from Merck. Ribose-
and 2-deoxy-5-ribosephosphate were from Fluka or Sigma. Dialkyl
phosphates were prepared from the corresponding trialkyl phosphates
(from Fluka) by partial hydrolysis with aqueous NaOH, following a
procedure as described for diethyl phosphate.37 Ethyleneglycolphos-
phate was synthesized from 2-chloroethanol and polyphosphoric acid
according to ref. 38. The water was purified with a Millipore-Milli-Q
system. The optical densities of the solutions at 193 nm were ca. 0.4-
1/cm. The solutions were deaerated with argon or saturated with
oxygen and flowed through the 2 mm (in the direction of the laser
light) by 4 mm (in the direction of the analyzing light) Suprasil quartz
cell with flow rates of ca. 3 mL/min, using pulse rates of 0.4 Hz. Unless
otherwise indicated, the pH was 8, and the temperature was 20 ( 1
°C. The experiments were carried out with an argon fluoride excimer
laser (Lambda Physik EMG103MSC) that delivered unfocused 20 ns
pulses with a power of 3-50 mJ (measured at the position of the cell
with a Gentech ED-200 power meter). The optical signals were
digitized with Tektronix 7612 and 7912 transient recorders interfaced
with a DEC LSI11/73+ computer, which also controlled the other
functions of the apparatus and on-line preanalyzed the data. Final data
analysis was done with a Microvax II connected to the LSI.
To test this idea, applied, however, to a bimolecular reaction,
the photochemically produced (via eq 1) phosphate radical was
reacted with 2′-deoxyguanosine, monitoring the absorption at
525 nm. At pH 7, the rate constant for this reaction, in which
the one-electron oxidized guanine radical6a was formed, was
measured to be 8.7 × 108 M-1 s-1, which is considerably higher
than that (≈1 × 108 M-1 s-1, see above) for reaction with
•-
d-ribose, which indicates that the reaction mechanism of HPO4
with the “aromatic” nucleoside may be different from that with
the aliphatic ribose-unit. However, the yield of the G-centered
•-
radical from the reaction of HPO4 was 2.4-times lower than
that ()100%)6a from the reaction of SO4•-. This is indicative
of a higher tendency of HPO4•-, as compared to SO4•-, for H-
relative to electron transfer.
Summary and Conclusions
In aqueous solution, 193 nm photolysis of inorganic and
organic phosphates such as ribose-5-phosphate leads to ioniza-
tion with formation of the corresponding oxygen-centered
phosphate radicals, O3PO•.36 These radicals abstract hydrogens
attached to R-, â-, or, possibly, γ-carbons, whereby in the case
of the â-hydrogens a six-membered transition state for transfer
of the hydrogen from the carbon to the phosphate oxygen is
Quantum yield measurements of 193 nm photoionization are based
on e-aq yields from argon-saturated aqueous solutions of NaCl (Merck)
of the same OD at 193 nm as those of the phosphate solutions. The
quantum yield for the production of e- from Cl- is assumed to be
aq
the same as that with 185 nm light, reported as 0.46.14 For the product
analysis measurements, the Farkas-actinometer (5 M ethanol in water,
formation of H2 (measured by GC)) was used, taking Φ(H2) ) 0.4.39
possible, leading to high rate constants (up to >5 × 107 s-1
)
JA974393M
for H-transfer in these unimolecular reactions. In the case of
(deoxy)ribosephosphates the six-membered transition state is
(37) McIvor, R. A.; McCarthy, G. D.; Grant, G. A. Can. J. Chem. 1956,
34, 1819.
(36) In the case of DNA, neutral phosphato radicals are expected from
the DNA monoanionic phosphate ester function. These neutral radicals
should be considerably more oxidizing (i.e., even better H-abstractors) than
the phosphato radical anions discussed in this paper.
(38) Cherbuliez, E.; Probst, H.; Rabinowitz, J. HelV. Chim. Acta 1958,
41, 1693.
(39) For a discussion of this actinometer, see: von Sonntag, C.;
Schuchmann, H. P. AdV. Photochem. 1977, 10, 59.