Photochemical rearrangement of enediynes: Is a 'photo-Bergman' cyclization a possibility?

Article abstract of DOI:10.1021/ja9722943

The enediynes 1 and 2 were photolyzed in 2-propanol to yield products akin to those analogous to a thermal Bergman cyclization mechanism. In addition, products resulting from photoreduction of one of the triple bonds of the enediyne functionality are also

Full text of DOI:10.1021/ja9722943

J. Am. Chem. Soc. 1998, 120, 1835-1841
1835
Photochemical Rearrangement of Enediynes: Is a “Photo-Bergman”
Cyclization a Possibility?
Ariella Evenzahav and Nicholas J. Turro*
Contribution from the Chemistry Department, Columbia UniVersity, New York, New York 10027
ReceiVed July 10, 1997
Abstract: The enediynes 1 and 2 were photolyzed in 2-propanol to yield products akin to those analogous to
a thermal Bergman cyclization mechanism. In addition, products resulting from photoreduction of one of the
triple bonds of the enediyne functionality are also formed. The product distributions and yields were found
to be dependent on the substituents on the triple bonds and on the nature of the double bond; in particular,
phenyl substituents on the triple bonds eliminated the photoreduction products, and overall yields were higher
when the double bond of the enediyne was of aromatic character. Triplet sensitization studies and laser flash
photolysis experiments point toward radical mechanisms taking place during formation of both classes of
products, with the photoreduction products forming from the excited triplet state and the cyclized naphthyl
products forming from either the singlet or the triplet states. For the cyclization reaction, a substituted 1,4-
dehydronaphthalene biradical species, structurally identical to a hypothetical Bergman biradical, is suggested
to be the most likely intermediate.
Introduction
understood. However, despite the fact that both double and
triple bonds display rich photoreactivities, comparatively very
few reports on the photochemistry of enediynes exist in the
literature. The first related study was conducted in 1967,18,19
whereupon diynes were shown to yield intermolecular adducts.
The first intramolecular photochemical rearrangement of ene-
diynes to yield a cyclized product was discovered by accident
in 1968;20 however, no further activity in this field took place
until almost two decades later, when enediynes were shown
during photosensitization to undergo cis,trans-isomerization
The chemistry of enediynes has been an area of avid scientific
interest during the past two decades, as a result of its application
in the design of a class of potent antitumor antibiotics.1-8 For
example, the drugs calicheamicin, dynemicin, and esperamicin,
each of which possesses an enediyne unit, have been extensively
studied, and have been shown to produce a highly reactive
aromatic 1,4-biradical intermediate, which is responsible for the
biological activity. Such intermediates were first characterized
by Bergman,9 who trapped the resulting p-benzyne biradical
upon thermolysis of an acyclic enediyne. Since then the thermal
rearrangement, commonly referred to as the Bergman rear-
rangement, has been thoroughly explored both experimentally
and theoretically,10-17 and both the thermodynamic and kinetic
properties of the thermal cyclization are becoming better
21,22,
around the double bond,4 as well as to cleave DNA strands,
24 although the mechanism of neither process has been eluci-
dated.
In 1994 we reported the photosensitized reaction of an
aromatic enediyne to afford a cyclized naphthyl product,
possibly via a 1,4-dehydronaphthalene intermediate akin to that
23
(1) Nicolaou, K. C.; Liu, A.; Zeng, Z.; McComb, S. J. Am. Chem. Soc.
1992, 114, 9279-9284.
(2) Nicolaou, K. C.; Dai, W.; Wendeborn, S. V.; Smith, A. L.; Tirisawa,
Y.; Maligres, P.; Hwang, C. Angew. Chem., Int. Ed. Engl. 1991, 30, 1032-
1036.
(3) Nicolaou, K. C.; Smith, A. L. Acc. Chem. Res. 1992, 25, 497-503.
(4) Kagan, J.; Wang, X.; Chen, X.; Lau, K. Y.; Batac, I. V.; Tuveson,
R. W.; Hudson, J. B. J. Photochem. Photobiol. B: Biol. 1993, 21, 135-
142.
expected from a thermal Bergman rearrangement (Scheme
1). Following our report, in 1996 Funk and co-workers24
reported the photochemistry of a series of aromatic enediyne
molecules, which were found to yield a Bergman-type cyclized
product in the presence of 1,4-cyclohexadiene as a hydrogen
donor; moreover, water-soluble analogues of the enediynes
studied were found to be effective in binding to DNA and in
(5) Nicolaou, K. C.; Ogawa, Y.; Zuccarello, G.; Kataoka, H. J. Am. Chem.
Soc. 1988, 110, 7247-7248.
(15) Gleiter, R.; Kratz, D. Angew. Chem., Int. Ed. Engl. 1993, 32, 842-
845.
(16) Jones, R. R.; Bergman, R. G. J. Am. Chem. Soc. 1972, 94, 660-
661.
(17) Lindh, R.; Perrson, B. J. J. Am. Chem. Soc. 1994, 116, 4963.
(18) Ipaktschi, J.; Staab, H. A. Tetrahedron Lett. 1967, 45, 4403-4408.
(19) Bossenbroek, B.; Shechter, H. J. Am. Chem. Soc. 1967, 89, 7111-
7112.
(20) Campbell, I. D.; Ellington, G. J. Chem. Soc. (C) 1968, 1968, 2120-
2121.
(21) Shiraki, T.; Sugiura, Y. Biochemistry 1990, 29, 9795-9798.
(22) Uesawa, Y.; Kuwahara, J.; Sugiura, Y. Biochem. Biophys. Res.
Commun. 1989, 164, 903-911.
(23) Turro, N. J.; Evenzahav, A.; Nicolaou, K. C. Tetrahedron Lett. 1994,
35, 8089.
(24) Funk, R. L.; Young, E. R. R.; Williams, R. M.; Flanagan, M. F.;
Cecil, T. L. J. Am. Chem. Soc. 1996, 118, 3291-3292.
(6) Nicolaou, K. C. Angew. Chem., Int. Ed. Engl. 1993, 32, 1377-1500.
(7) Wender, P. A.; Zercher, C. K.; Beckham, S.; Haubold, E.-M. J. Org.
Chem. 1993, 58, 5867-5869.
(8) Golik, J.; Clardy, J.; Dubay, G.; Groenewold, G.; Kawaguchi, H.;
Konishi, M.; Krishnan, B.; Okhuma, H.; Saitoh, K. J. Am. Chem. Soc. 1987,
109, 3461-3462.
(9) Bergman, R. G. Acc. Chem. Res. 1973, 6, 25-30.
(10) Yoshida, K.; Minami, Y.; Azuma, R.; Saeki, M.; Otani, T.
Tetrahedron Lett. 1993, 34, 2637-2640.
(11) Kraka, E.; Cremer, D. J. Am. Chem. Soc. 1994, 116, 4929-4936.
(12) Lockhart, T. P.; Bergman, R. G. J. Am. Chem. Soc. 1981, 103,
4091-4096.
(13) Lott, W. B.; Evans, T. J.; Grissom, C. B. J. Chem. Soc., Perkin
Trans. 2 1994, 2583-2586.
(14) Myers, A. G.; Dragovich, P. S.; Kuo, E. Y. J. Am. Chem. Soc. 1992,
114, 9369-9386.
S0002-7863(97)02294-4 CCC: $15.00 © 1998 American Chemical Society
Published on Web 02/05/1998

1836 J. Am. Chem. Soc., Vol. 120, No. 8, 1998
EVenzahaV and Turro
Scheme 1
cleaving single DNA strands upon irradiation. Suprisingly, no
other example of a direct photochemical reaction of enediynes
has been reported. The mechanistic study of photosensitized
electron transfer rearrangements of an aromatic enediyne has
also been recently reported,25 but these reactions have been
shown to yield different products than those expected from a
hypothetical thermal case analogous to the Bergman rearrange-
ment.
whereas 1 yields the Bergman-type product and additionally
products arising from triple bond photoreductions. Furthermore,
we find that not only the types of products but also the yields
of cyclization are dependent on the structure of the parent
compound. From the results of flash photolysis experiments,
we rationalize the formation of the cyclized products as arising
from a 1,4-aromatic biradical in analogy to a thermal mecha-
nism.
The induction of a Bergman cyclization of enediynes via a
photochemical mechanism would be of great interest for a
number of reasons. First, the observation of a photochemical
reaction of enediynes to yield intermediates and products akin
to those expected from a thermal Bergman rearrangement would
provide yet another example of unsaturated organic molecules
which undergo rearrangements via both thermal and photo-
chemical mechanisms, such as the Diels-Alder reaction.
Moreover, a photoinduced Bergman cyclization could have
potential application in the design and activity of new enediyne-
based antitumor drugs; whereas the current antibiotics are
activated thermally at physiological temperature, a photolytic
mechanism would allow a time-controlled activation of the drug
by selective exposure to cell-permeable photons. With the rapid
advances in the field of photodynamic therapy, such a possibility
is becoming increasingly feasible and attractive. Furthermore,
a photochemical enediyne cyclization could also exhibit potential
in the design and preparation of new polymeric materials.
Enediyne polymers have been shown to possess rather striking
spectroscopic properties very different from those displayed by
the monomers themselves.15 Photosensitized cyclization of such
polymer matrices can lead to the formation of new polymers
containing a high degree of aromaticity and distinct properties
without the need for high temperatures or strong reagents.
To examine whether a photoinduced reaction of enediynes
is possible, and whether the reaction proceeds in a fashion
comparable to the thermal case, the enediynes 1 and 2 were
prepared and their photochemistry explored. The structures of
these compounds possess the attractive features of (i) facile
synthesis from readily available starting materials, (ii) elimina-
tion of cis,trans-isomerization of the double bond which is
locked into a cis conformation by confinement in a ring, (iii)
chromophores which would enhance light absorption and allow
for “tracking” via spectroscopic methods, and (iv) thermal
stability at slightly elevated temperatures which occur during
irradiation with light. Furthermore, the variation of the nature
of the double bond as well as the substituents on the triple bonds
of the enediynes would elucidate the effect of the structure on
the photochemical reactivity.
Experimental Section
The enediynes 1 and 2 were synthesized via a copper coupling
reaction of the appropriate dihaloalkene with an excess of 2 equiv of
the alkyne;1 in particular, these consisted of 1,2-diiodobenzene and
1-propyne for 1, and 1,2-diiodobenzene and phenylacetylene for 2. Both
123 and 225 have been previously prepared and characterized. All
products were purified by flash column chromatography on silica gel
with hexane as eluent; in some cases, multiple column purifications
were run until over 99% purity by GC was achieved. 1: 1H NMR
(200 MHz, CDCl3, 25 °C, TMS): δ 1.05 (t, 6H); 1.5 (m, 4H), 2.42 (t,
4H), 7.18 (m, 2H), 7.35 (m, 2H); GCMS (70 eV EI, 1 peak) m/z 210;
high-resolution MS (70 eV EI) m/z 210.1402 (estimated m/z 210.1409;
error -3.3 ppm); UV/vis max (hexane) (M-1 cm-1): ꢀ292 ∼ 480, ꢀ300
∼ 280, ꢀ308 ∼ 80, and ꢀ313 ∼ 30. 2: 1H NMR (200 MHz, CDCl3, 25
°C, TMS): δ 7.3 (m, 10H), 7.58 (m, 4H); GCMS (70 eV EI, 1 peak)
m/z 278; high-resolution MS (70 eV EI) m/z 278.1103 (estimated m/z
278.1096; error +2.7 ppm); UV/vis max (hexane) (M-1 cm-1): ꢀ260
55500, ꢀ272 ∼ 81500, ꢀ300 ∼ 22800, ꢀ308 ∼ 24700, ꢀ330 ∼ 10700.
∼
Photolysis experiments were carried out with either a Rayonette
Photochemical Reactor, equipped with eight 3000 Å lamps, or a
medium-pressure Hanovia mercury lamp. In the latter case, a potassium
chromate filter solution was employed to filter out wavelengths below
313 nm. Analytic irradiations were performed in test tubes and a
“merry-go-round” apparatus was employed to ensure equal average light
absorption by all samples. Irradiation runs contained pentadecane as
an internal standard and were analyzed by gas chromatography on a
Hewlett-Packard 5890 equipped with a SE-30 column and a flame
ionization detector. For all irradiations, the solutions’ optical density
at the wavelength of irradiation was above 1. Samples were deoxy-
genated by purging with argon for 10-15 min. Preparative scale
irradiations were performed for identification of the products. In this
case, the products were separated by column chromatography and
characterized by their mass and NMR spectra, and the GC retention
times of the separated products were compared to those of the analytic
runs. A Varian 200 MHz NMR spectrophotometer and a Hewlett-
Packard 5890 GC Series II, equipped with a HP5972 mass spectral
detector, were used.
UV/vis absorption spectra were recorded on a Hewlett-Packard
8452A diode array spectrophotometer. Fluorescence spectra were
recorded on an LS-5 or an IAS Fluoromax-2 spectrophotometer.
Quantum yields of fluorescence for 1 and 2 were determined by
comparison to a solution of 9,10-diphenylanthracene by matching the
optical densities at the excitation wavelength and comparison of the
integrated area for the two spectra. In this manner, fluorescence
quantum yields of approximately 0.04 and 0.4 were estimated for 1
and 2, respectively. Phosphorescence spectra were measured on an
LS-50 spectrophotometer. The energies of the excited singlet and triplet
states of compounds 1 and 2 were determined from the (0,0) band of
We report herein the successful photochemical induction of
the reaction of enediynes 1 and 2. In particular, we find that 2
yields the naphthyl derivative 2P, expected from a hypothetical
thermal Bergman cyclization, as the only product of photolysis,
(25) Ramkumar, D.; Kalpana, M.; Varghese, B.; Sankararaman, S. J.
Org. Chem. 1996, 61, 2247-2250.

Photochemical Rearrangement of Enediynes
J. Am. Chem. Soc., Vol. 120, No. 8, 1998 1837
Scheme 2
Table 1. Conversion and Product Yields of the Photochemical
Reaction of 1 Sensitized by Triplet Sensitizers Xanthone and
Acetophenone
sensitizer conv., %a ED1P, % ED1Pb, % ED1Pc, %b M.B., %c
the low-temperature fluorescence and phosphorescence spectra, and
were found to be 92 and 67 kcal/mol for 1 and 81 and 56 kcal/mol for
2, respectively. For the low-temperature spectroscopies, the cell
compartment of the instrument was fitted with a Dewar modified with
a quartz cell bottom, to allow the incident light to pass through. The
samples were contained in the Dewar filled with liquid nitrogen in
quartz 5 mm NMR tubes. Background spectra of the Dewar setup
lacking the sample were recorded during each experiment for reference.
Flash photolysis experiments were performed at room temperature
by using the 308 nm output of a MPB Technologies AQX-150 XeCl
excimer laser (ca. 2 mJ/pulse, 20 ns fwhm) for excitation of the
enediynes, or the 355 nm output of a Continuum Surelite I Nd:YAG
laser (ca. 8 mJ/pulse, 20 ns fwhm) for excitation of xanthone in the
triplet sensitization studies. Samples were prepared with an optical
density of 0.3 at the excitation wavelength, and were excited in 1 × 1
Suprasil quartz cells, after deoxygenating by purging with argon.
Oxygen saturation of the solutions for quenching measurements was
achieved by bubbling the solutions with oxygen. For the sensitization
and quenching experiments, argon saturated static samples were
employed, and a fresh solution was used for each quencher concentra-
tion. Typically 5-10 pulses were employed for each decay trace.
Transient absorption spectra employed a flow system so as to ensure
that a fresh volume of sample was irradiated with each pulse.
All solvents used were of spectroscopic grade, and the absence of
absorbing species at the wavelength of irradiation was ensured by UV/
vis spectroscopy.
none
Xd
15
76
83
24
2
3
29
2
8
26:9
18:1.5
3:0
90
15
21
APe
a Conversion of 1. b Yields given correspond to percent 1Pc (cis):
percent 1Pc (trans). c Mass balance. d Xanthone. e Acetophenone.
Scheme 3
Results
enediyne, conversion is increased compared to direct photolysis;
however, the mass balance and yields of the products are
substantially decreased, as shown in Table 1.
Steady-state photolysis of a 5 mM solution of 1 in 2-propanol
affords the products 1Pa, 1Pb, and cis- and trans-1Pc in a ratio
of approximately 2:4:2:1 (Scheme 2). When the reaction is
carried out to 50% conversion of 1, a yield of 25% for the
cyclized 1Pa and a mass balance of 80% are achieved; at higher
conversions it is likely that intermolecular reactions yielding
high molecular weight products not discernible by GC must be
formed, as evidenced by a rapid decrease in the mass balance.
In the presence of 1,4-cyclohexadiene, photolysis of 1 yields
1Pa as well as a product whose mass spectrum is consistent
with a reduction product containing an equivalent of the
hydrogen donor. Moreover, although 1 photochemically con-
verts in ethanol, sec-butanol, and tert-butanol, formation of 1Pa
and other products detectable by GC are only observed during
photoreaction in the first alcohol. When the reaction of 1 is
triplet sensitized by the irradiation of xanthone or acetophenone,
both of whose triplet energies are higher than that of the
Photolysis of enediyne 2 in 2-propanol, interestingly, yields
the cyclized 2P as the only product observed by GC charac-
terization with high mass balance (ca. 75%; Scheme 3); reduced
products such as those observed from photoreaction of 1 are
not formed to any detectable extent. In the presence of 1,4-
cyclohexadiene, however, although photochemical conversion
of the enediyne does take place, no product is observed by GC,
suggesting the formation of an uncharacterized polymeric
material. In the case of triplet sensitization of 2 by excitation
and subsequent energy transfer from xanthone, no conversion
of the enediyne is observed.
In the laser flash photolysis experiments, excitation of a
deaerated solution of 1 in hexane leads to the production of a
single transient with a maximum at ca. 320 nm, shown in Figure
1, which decays with a lifetime of t ∼ 20 ms. The decay,

1838 J. Am. Chem. Soc., Vol. 120, No. 8, 1998
EVenzahaV and Turro
the transient of 1, the transient of 2 is also quenched by
molecular oxygen, with a rate constant of 1 × 109 M-1 s-1
.
Again, 2 is an effective quencher of the triplet transient of
xanthone; the decay of the latter, monitored at 635 nm, is
quenched by 2 with a rate constant of ∼5 × 109 M-1 s-1, while
a concurrent growth is observed at 370 nm, near the maximum
of the transient of 2.
Discussion
The photolysis studies described in the Results section provide
proof that the enediynes 1 and 2 can react photochemically to
produce the same cyclized product that would be formed if a
thermal Bergman rearrangement were to take place. The
questions which arise pertain to the mechanism of the reaction;
in particular: (i) What is the intermediate(s) responsible for the
formation of the cyclized products, and how does the mechanism
compare to that of the thermal case? (ii) What is the mechanism
that leads to the formation of the photoreduced products and
why are the latter produced only from 1? (iii) Are the transients
observed by laser flash photolysis present on the photoreaction
pathway of the enediynes, and what is their nature? We will
attempt to answer these questions by looking at the data obtained
from both steady-state and time-resolved experiments and by
referring to similar systems which have been extensively covered
in the literature.
Figure 1. Transient absorption spectrum of 1 in hexane, about 500 ns
after the laser pulse.
A priori there are a number of different possible mechanisms
which could plausibly account for the formation of the products
observed. These include the intermediacy of either radical,
ionic, or radical ionic intermediates, and could entail either a
concerted or a stepwise mechanistic pathway. On the basis of
this assumption, one can write out a multitude of mechanistic
schemes which would lead to the formed products. Further-
more, it is also possible that the cyclized products observed
could result from secondary processes, whereupon a stable
product absorbs a second photon to further rearrange to the
observed compounds, although in this study such a mechanism
would be highly unlikely, since isolation and subsequent
photolysis of the primary photoproducts do not react further to
the cyclized material.
Figure 2. Transient absorption spectra of 1 in 2-propanol immediatly
after the laser pulse (a) and approximately 10 µs after the pulse (b).
The decay of 1 at 350 nm is shown in the inset.
A single step concerted mechanism to yield an intermediate
that would then undergo hydrogen abstraction fails to account
for the different types of products observed for 1, since clearly
1Pa arises from a cyclization step that is absent in the pathway
to 1Pb and 1Pc. An ionic mechanism is considered unlikely,
mainly due to the reactivity of the enediynes in nonprotic media
such as 1,4-cyclohexadiene, which acts as a hydrogen donor.
Moreover, if the reaction were to be of ionic nature, one would
expect an isopropoxy adduct from photolysis of 1 in 2-propanol;
instead, the R-isopropyl adduct 1Pb is formed. Hence it is much
more likely that a radical mechanism is at play, at least in the
formation of the acetylene reduction products 1Pb and 1Pc.
There is, additionally, a precedence for radical reactivity of
acetylenes.26 tert-Butylphenylacetylene, for example, when
photolyzed in 2-propanol, has been shown to produce the
respective cis and trans alkene from two hydrogen abstraction
steps, as well as the alkene formed by addition of an equivalent
of 2-propanol across the triple bond, just as has been observed
for photoreaction of 1 yielding cis and trans 1Pc and 1Pb,
respectively.
Figure 3. Transient absorption spectrum of 3 in acetonitrile ap-
proximately 500 ns after the laser pulse. The decay of 3 at 390 nm is
shown in the inset.
monitored at 320 nm, is quenched by molecular oxygen with a
rate constant of 1.3 × 109 M-1 s-1. The transient spectrum of
1 in 2-propanol, however, in which the enediyne is reactive,
contains an additional absorption band in the region of 370 nm,
which decays with a much longer lifetime, as evidenced by
observation at ∼15 ms after the laser pulse (Figure 2). A similar
transient is obtained when 1 is photolyzed in hexane in the
presence of 1,4-cyclohexadiene in which the enediyne is
reactive, as mentioned before. Furthermore, 1 quenches the
excited-state triplet transient of xanthone, with a rate constant
of 9 × 109 M-1 s-1
.
Flash photolysis of 2 yields a single transient species (Figure
3), which does not exhibit the solvent dependence of 1. This
species has an absorption spectrum similar in structure to that
of 1 in hexane, as shown in Figure 3, with the maximum red
shifted to approximately 390 nm. Monitoring the decay of the
transient in acetonitrile yields a lifetime of ∼35 ms; in
2-propanol, this lifetime is reduced to 15 ms. As observed for
As far as the formation of an intermediate involved in the
production of the cyclized products 1Pa and 2P, there are two
plausible scenarios, based on the time-resolved data obtained
(26) Perreten, J.; Chihal, D. M.; Griffin, G. W.; Bhacca, N. S. J. Org.
Chem. 1976, 41, 3931.

Photochemical Rearrangement of Enediynes
J. Am. Chem. Soc., Vol. 120, No. 8, 1998 1839
Scheme 4
in this study and on literature precedents of similar systems.
The first involves the intermediacy of a radical cation species.
In fact, it has been reported that photolysis of 2 in the presence
of suitable electron acceptors leads to the production of
substituted indene compounds via an enediyne-derived radical
cation.25 Moreover, extensive literature exists on the absorption
spectra of aromatic radical cations formed by direct excita-
tion;27,28 these spectra are structurally very similar to the ones
obtained for 1 and 2, particularly in terms of the visible
absorption band. This scenario, however, seems to be rather
unlikely under the conditions of the system studied herein. The
reaction of the enediynes to yield the cyclized substituted
naphthalenes proceeds under steady-state conditions without the
presence of electron acceptors, and clearly the naphthalene
products observed for both 1 and 2 are structurally different
from the indenes produced in the aforementioned study.
Furthermore, if the transient spectra of 1 and 2 were to reflect
the presence of a radical cation, then one would expect to
observe a particular behavior characteristic of such species, such
as quenching of the decays by nucleophiles such as acetate or
amines,29,30 and lifetime elongation through addition of electron
traps such as carbon tetrachloride. Although we attempted to
quench the transient decays of 1 and 2 with a variety of amines
including DABCO, triethylamine, and n-butylamine, as well as
acetate, and we added various concentrations of carbon tetra-
chloride, we did not observe in any one case any significant
quenching or lifetime elongation, respectively. Moreover, the
transients were efficiently quenched by oxygen at close to the
diffusion-controlled limit, which would not be expected if the
transient were attributed to a radical cation.29
pounds described herein, have been shown to undergo cycliza-
tion reactions via radical intermediates, from both the excited-
state singlet and triplet states. Additionally, the presence of a
radical transient would explain the hydrogen abstraction steps
during photolysis of the enediynes and would justify the
observed quenching of the transient decay by oxygen.
It is worthwhile, at this point, to consider the photoreactivity
of the diyne 3, studied by Zimmerman and co-workers,34 which
possesses a striking similarity to the photoreactivity of 1. In
particular, photolysis of 3 in 2-propanol produced the reduced
3b and the cyclopentadiene 3a. The mechanism postulated by
Zimmerman involved excitation of one triple bond in the excited
triplet state, which then could either be reduced by two
consecutive hydrogen abstraction steps from two individual
2-propanol molecules or, alternatively, could cyclize to yield a
1,4-cyclopentadienyl biradical, which would then abstract
hydrogens to produce 3a (Scheme 4). At first glance, com-
parison of the products obtained from 1 to those from 3 would
suggest that the underlying mechanism should be similar for
both systems. There are, however, the following differences
which need to be considered: (i) whereas 1 does display similar
reactivity to 3, 3 produces only the cyclized material and no
photoreduction products; and (ii) triplet sensitization of 3 with
xanthone produces 3a and 3b in roughly the same ratio as direct
photolysis of 3, whereas in the case of triplet sensitization of 1
the ratio of the photoreduction products 1Pb and 1Pc to cyclized
product 1Pa is much larger than that obtained during direct
irradiationsfurthermore, triplet sensitization of 2 yields to no
product formation.
To account for these differences, we turn to a qualitative
argument relating the structures of the observed enediynes to
the expected excited state responsible for their reaction. The
quantum yield of fluorescence for a given aromatic molecule
can be correlated to its structural rigidity; we expect the phenyl
substituents on the triple bonds of 2 to force the molecule into
a more rigid conformation than 1;36 this hypothesis is supported
by the significantly larger fluorescence quantum yield of 2
compared to that of 1. A lower fluorescence quantum yield
A second scenario would entail the formation of a radical
species that would lead to the cyclized naphthyl products.
32
Enynes31, and diynes,33,-35 structural relatives of the com-
(27) Liu, A.-D.; Sauer, M. C.; Jonah, C. D.; Trifunac, A. D. J. Phys.
Chem. 1992, 96, 9293-9298.
(28) Liu, A.; Sauer, M. C.; Loffredo, D. M.; Trifunac, A. D. J.
Photochem. Photobiol. 1992, 67, 197-208.
(29) Johnston, L. J.; Schepp, N. P. J. Am. Chem. Soc. 1993, 115, 6564-
6571.
(30) Workentin, M. S.; Johnston, L. J.; Wayner, D. D. M.; Parker, V.
D. J. Am. Chem. Soc. 1994, 116, 8279-8287.
(34) Zimmerman, H. E.; Pincock, J. A. J. Am. Chem. Soc. 1973, 95,
3246.
(35) Hayashi, T.; Mataga, N.; Inoue, T.; Kaneda, T.; Irie, M.; Misumi,
S. J. Am. Chem. Soc. 1977, 99, 523.
(36) Turro, N. J. Modern Molecular Photochemistry; The Benjamin/
Cummings Publishing Co., Inc.: Menlo Park, CA, 1978; Chapter 5, p 111
and Chapter 6, p 171.
(31) Tinnemans, A. H. A.; Laarhoven, W. H. J. Chem. Soc., Perkin Trans.
2 1976, 1111.
(32) Zheng, M.; DiRico, K. J.; Kirchhoff, M. M.; Phillips, K. M.; Cuff,
L. M.; Johnson, R. P. J. Am. Chem. Soc. 1993, 115, 12167-12168.
(33) Johnson, R. P. Chem. ReV. (Washington, D.C.) 1989, 89, 1111-
1124 and references therein.

1840 J. Am. Chem. Soc., Vol. 120, No. 8, 1998
EVenzahaV and Turro
would, in turn, imply that the probability is greater of other
processes for deactivation of the excited singlet state taking
place, such as intersystem crossing to the triplet state or reactions
from the singlet state. If we were to assume that the photore-
duction products 1Pb and 1Pc arise from the triplet state whereas
the cyclized 1Pa arises from the singlet excited state, this would
explain why 2 does not form the equivalent photoreduction
products, if its intersystem crossing efficiency is lower than that
of 1.
Scheme 5
The involvement of two separate excited states in the
formation of the two classes of products can be further justified
by the behavior of the transients produced by photolysis of the
two enediynes 1 and 2. In particular, 1 produces an additional
transient absorption band that is lacking in the spectrum derived
from 2. Taking the fact that 1 also produces under photolysis
conditions the photoreduction products which are lacking in the
reaction of 2, the common spectrum to the radical intermediate
leading to 1Pa and 2P may be correlated with the transient
unique to 1 to the products 1Pb and 1Pc. The time-resolved
triplet sensitization experiments would then suggest that the
radical leading to the Bergman-type product could also be
formed from the triplet state, although during direct excitation
the electronic state most probably involved would be the excited
singlet state.
It remains to elucidate the structure of the radical intermediate
leading to formation of 1Pa and 2P. Since the cyclized
Bergman-type product and the uncyclized photoreduction
products for the reaction of 1 seem to be produced from two
distinct transients, it is very likely that the former transient
observed be formed after the cyclization step, structurally
resembling the inert product; the most possible structures would
then be either a 1-naphthyl radical or a 1,4-naphthyl biradical.
Both aromatic radicals38 and biradicals37 have been investigated
in a time-resolved fashion in the literature. Specifically, the
study of the 9,10-anthracenyl biradical reported by Schottelius
et al.37 may be pertinent to our system. Although the absorption
bands of the transient spectrum of the latter species differs from
that observed for 1 and 2, specifically in lacking the visible
absorption band, the calculated rate constants for hydrogen
abstraction reactions are very similar. In particular, Schottelius
reports bimolecular rates of hydrogen abstraction from aceto-
Scheme 6
nitrile and 2-propanol to be 1.1 × 103 and 6.5 × 103 M-1 s-1
,
respectively. From the decay lifetimes of the transient of 2 we
obtain pseudo-first-order rate constants of 7 × 104 and 3 × 104
M-1 s-1 for those two solvents; adjusting for the molarity of
the solvents, which are 19 and 13 M for acetonitrile and
2-propanol, respectively, we arrive at bimolecular rate constants
of 2 and 4 × 103 M-1 s-1. These estimations would actually
correspond to upper limits of the hydrogen abstraction rate, since
they assume that all molecules of 2 react by hydrogen abstrac-
tion, whereas it is highly probable that a portion of 2 reacts
intermolecularly with itself. Furthermore, not only do these
values fall close to those calculated for the anthracenyl biradical,
but they are also much lower than rates of hydrogen abstraction
reported for aromatic monoradicals.38
We propose the most probable mechanism taking place during
photoreaction of 1 and 2 as shown in Schemes 5 and 6,
respectively. Both enediynes are photochemically excited to
the singlet state. Reaction from the singlet state leads to
formation of the Bergman-type cyclized products, most probably
via a naphthyl biradical, which abstracts two hydrogens from
an isopropyl molecule to yield the final 1Pa and 2P. The
excited 1 can undergo intersystem crossing to the triplet state,
which is responsible for the acetylene photoreduction products
observed. It is also possible that the triplet state of the enediynes
can also produce the cyclized products 1Pa and 2P, as indicated
in the last equation of each scheme.
Before closing, let us comment on the observed effect of the
structure of the enediynes as related to their reactivity,39
specifically the effect of the triple bond substituents on the
product formation of the enediynes. Comparison of the product
distribution of 1 and 2 indicates that the presence of aromatic
rings on the triple bonds drives the reaction toward cyclization
and eliminates the formation of photoreduction products. This
may be correlated to the hypothesized intersystem crossing
efficiency of the two subsets; in particular, as mentioned before,
the phenyl substituents on 2 may cause increased rigidity in
the overall structure through steric hindrance or even possibly
some π-stacking interactions, absent in 1, which would lead to
a less efficient triplet state crossing from the singlet. Assuming
that the singlet state is responsible for the production of the
(37) Schottelius, M. J.; Chen, P. J. Am. Chem. Soc. 1996, 118, 4896-
4903.
(38) Scaiano, J. C.; Stewart, L. C. J. Am. Chem. Soc. 1983, 105, 3609-
3613.

Photochemical Rearrangement of Enediynes
J. Am. Chem. Soc., Vol. 120, No. 8, 1998 1841
naphthyl products but not the photoreduction products, one
would expect the latter to be formed mainly in the triplet
photoreaction of 1.
be expected from a thermal Bergman rearrangement. Both
steady-state and time-resolved investigations strongly suggest
that the cyclized products obtained for each of the enediyne
analogues are formed via a mechanism that produces a 1,4-
naphthyl biradical as its intermediate. The photochemical
reaction differs from a thermal analogue mainly in that the
former seems to arise as a result of excitation of an acetylenic
unit rather than of a conjugative effect of the entire enediyne
functionality.
Conclusions
We report the photosensitized reaction of a series of enediynes
which yield a cyclization product identical to that which would
(39) Photolysis of the compounds 1,2-bis(1-pentynyl)cyclopentene (3)
and 1,2-bis(1-phenylacetylenyl)cyclopentene (4) under the same conditions
as those reported for 1 and 2 yielded a similar product distribution,
specifically 5,6-dipropylindan and 1-(1-pentenyl)-2-(1-pentynyl)cyclopen-
tene for 3 and 5,6-diphenylindan for 4 (preliminary results). Although the
same reactivity trends are observed for these compounds, with the phenyl
subsituted 4 being more reactive than 3, the overall product yields of 1 and
2, whose double bond is part of an aromatic framework, are much higher
than those for 3 and 4, where the double bond is isolated in a cyclopentenyl
ring. In the latter set, the orbitals of the double bond would be conjugated
only with the triple bonds on either side, whereas in the former set there
would be a competing conjugative effect with the aromatic framework of
the phenyl ring in which it is contained. This observation would suggest
that the photoreactivity of the enediyne compounds studied is a result of
excitation of the individual acetylenic units and does not rely on a
conjugative effect of the enediyne functionality.
Acknowledgment. The authors wish to thank Prof. K. C.
Nicolaou for his initial contribution to the project. Helpful
suggestions from Prof. L. I. Johnston and Dr. I. Gould are also
acknowledged. This work was supported financially by Merck,
Upjohn, and the NSF.
JA9722943