Thermal C1-C5 diradical cyclization of enediynes

Article abstract of DOI:10.1021/ja803413f

Computational studies at the BLYP/6-31G(d) level (supplemented by BCCD(T)/cc-pVDZ calculations) suggest that in aryl-substituted 1,2-diethynylbenzenes, steric effects disfavor the thermal C¹-C⁶ diradical cyclization reaction (Bergman) and electronic effects favor the regiovariant C¹-C⁵ cyclization to the extent that the C¹-C⁵ process should become an important reaction pathway in the thermolyses of such compounds. Experimentally, thermolyses of 1,2-bis(2,4,6-trichlorophenylethynyl)benzene, a particularly favorable case, yields only products derived from C¹-C⁵ cyclization [specifically, 1-(2,4,6-trichlorobenzylidene)-2-(2,4,6-trichlorophenyl)-1H-indene and its hydrogenation product 3-(2,4,6-trichlorobenzyl)-2-(2,4,6-trichlorophenyl)-1H-indene], and even for the parent hydrocarbon 1,2-bis(phenylethynyl)benzene, the formation of C¹-C⁵ cyclization products is competitive with the major Bergman reaction. Although some C¹-C⁵ cyclization products are probably formed by transfer hydrogenation from 1,4-cyclohexadiene (commonly included in such reactions), thermolyses in the absence of 1,4-CHD as well as deuterium labeling studies confirm the existence of direct C¹-C⁵ diradical cyclizations for diaryl-substituted enediynes. Copyright

Full text of DOI:10.1021/ja803413f

Published on Web 09/18/2008
Thermal C1-C5 Diradical Cyclization of Enediynes
Chandrasekhar Vavilala,† Neal Byrne,‡ Christina M. Kraml,‡ Douglas M. Ho,§ and
Robert A. Pascal, Jr.*,†
Department of Chemistry, Princeton UniVersity, Princeton, New Jersey 08544, Lotus Separations, Princeton, New
Jersey 08544, and Department of Chemistry and Chemical Biology, HarVard UniVersity,
Cambridge, Massachusetts 02138
Received May 7, 2008; E-mail: snake@chemvax.princeton,edu
Table 1. ZPE-Corrected RBLYP/6-31G(d) Activation Energiesa
(kcal/mol) for the Cyclizations of Monosubstituted
1,2-Diethynylbenzenes
Thermal cycloaromatizations of enediynes (Bergman, C1-C6)1
and enyne-allenes (Myers-Saito, C2-C7)2 (Scheme 1) are of interest
because their diradical intermediates provide the mode of action
of natural enediyne antitumor antibiotics.3 In 1995 Schmittel found
an alternative diradical cyclization of enyne-allenes (C2-C6)4
leading to fulvene diradical intermediates,5 and these also have been
shown to cleave DNA in vitro.6
R
Ea (C1-C6) Ea (C1-C5)
a, b (Å)
Scheme 1. Thermal Cyclizations of Enediynes and Enyne-Allenes
H (7a)
phenyl (7b)
2,6-dichlorophenyl (7c)
2,6-dichloro-4-nitrophenyl (7d)
2,6-dichloro-4-aminophenyl (7e)
2,6-dimethylphenyl (7f)
24.6
28.7
30.8
31.8
30.7
30.5
37.2b
31.4
31.6
31.6
30.4
30.9
1.459, 1.406
1.462, 1.400
1.458, 1.395
1.459, 1.398
1.471, 1.406
a See note 13. b BS-UBLYP/6-31G(d), ref 12a.
effect, and not an electronic effect of the chlorines, is evident from
similar results found when a 2,6-dimethylphenyl group (7f) is
employed instead. Addition of either a nitro (7d) or an amino group
(7e) to the aryl ring makes the activation energies of the C1-C6
and C1-C5 cyclizations essentially equal at this level of theory.
Finally, a higher-level calibration of these calculations is given by
BCCD(T)/cc-pVDZ//RBLYP/6-31G(d) energies for the reactions
of 7c, yielding Ea values of 31.1 and 35.3 kcal/mol for the C1-C6
and C1-C5 cyclizations, respectively, suggesting that the RBLYP
calculations may somewhat underestimate the barrier to the C1-C5
process.13
The thermal C1-C5 diradical cyclization of enediynes, a regio-
variant of the well-known Bergman cyclization, is not well docu-
mented, if at all,7 but the C1-C5 cyclization of enediynes initiated
by either photoinduced electron transfer,8 electrophiles,9 radicals,10
or metals11 is known. For the parent enediyne 1, computational
studies by Schreiner found a much higher barrier for the C1-C5
diradical cyclization than for the C1-C6 pathway: 41.0 vs 25.2
kcal/mol at the BLYP/6-31G(d) level of theory (42.1 vs 27.1 kcal/
mol for BCCD(T)/cc-pVDZ single points at the BLYP geom-
etries).12 However, in the present work we show that this preference
can be reversed and the thermal C1-C5 diradical cyclization is an
important, and sometimes the major, reaction in diaryl-substituted
enediynes.
For an experimental test of the thermal C1-C5 cyclization, we
chose compound 8, a benzannulated enediyne with 2,4,6-trichlo-
rophenyl (TCP) groups at both alkyne termini. Not only is the
synthesis of 8 convenient (Sonogashira coupling of 1,2-diiodoben-
zene and 2,4,6-trichlorophenylacetylene14 gives 8 in 50% yield),
but also the addition of the second TCP group should further impede
the Bergman cyclization.
DSC measurements on neat enediyne 8 showed the onset of
reaction at about 250 °C. Thermolysis of 8 in toluene at 260 °C
in the presence of 1,4-cyclohexadiene gave indene derivatives
9 and 10 (Scheme 2) in 19% and 50% isolated yield, respectively,
after purification by supercritical fluid chromatography. The
To evaluate the feasibility of C1-C5 cyclization, we computa-
tionally screened several benzannulated enediynes substituted with
an aryl group at one alkyne terminus, and the results are summarized
in Table 1.13 From entries 7a and 7b, it is seen that the activation
energy for Bergman cyclization increases by 4 kcal/mol when the
hydrogen atom is replaced by a phenyl group (likely a steric effect),
but for the C1-C5 pathway it decreases by 6 kcal/mol. The latter
effect is due to stabilization of the emerging vinyl radical center in
the C1-C5 pathway, and is evident in the short phenyl-C6 bond
(bond b) in the transition state (TS). Such effects do not operate in
the Bergman pathway, but the C1-C6 cyclization remains strongly
preferred for 7b. However, when a 2,6-dichlorophenyl group is
substituted for phenyl (7c), increased steric effects (small in the
C1-C5 pathway) further slow the Bergman reaction, and the
competing TSs are brought to within 1 kcal/mol. That this is a steric
Scheme 2. Thermolysis of Enediyne 8
† Princeton University.
‡ Lotus Separations.
§ Harvard University.
9
10.1021/ja803413f CCC: $40.75
2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 13549–13551 13549

C O M M U N I C A T I O N S
Table 2. ZPE-Corrected Activation Energies (kcal/mol) for the
Thermal Cyclizations of 1,2-Bis(arylethynyl)benzenes
enediyne 11
enediyne 8
C1-C6
(TS 12) (TS 13) (TS 12c) (TS 13c)
C1-C5 C1-C6 C1-C5
level
BLYP/6-31G(d)//BLYP/6-31G(d)
BCCD(T)/cc-pVDZ//-BLYP/6-31G(d) 35.1
36.3
38.5
32.9
38.7
43.2
First, the computational data of Table 1 predict that the C1-C5
and C1-C6 cyclizations should become competitive upon addition
of an aryl group to even one alkyne terminus, due to both the
mesomeric stabilization of the C1-C5 TS and steric destabilization
of the C1-C6. Computational studies of the reactions of the
experimentally investigated enediynes 8 and 11 further strengthen
this conclusion (Table 2). At the BLYP/6-31G(d) level,13 the C1-C5
cyclization of hydrocarbon 11 is actually favored by 2.2 kcal/mol
over the C1-C6, and for the chlorinated compound 8, the C1-C5
cyclization is favored by a full 4.5 kcal/mol! Most of the latter
difference is due to steric destabilization of the Bergman cyclization
TS, as illustrated in Figure 2. Bulky terminal substituents tend to
inhibit the C1-C6 reaction,12a because these groups must be brought
together to flank the newly forming bond, but such steric effects
are smaller for C1-C5 cyclizations in which the substituents are
separated by one more atom. However, the Bergman cyclization
of compound 8 is particularly unfavorable, because the C1-C6 TS
structure contains two tight chlorine-chlorine contacts (3.59 Å
each; the sum of the van der Waals radii is 3.8 Å). In the C1-C5
TS structure the closest chlorine-chlorine contact distance is
3.92 Å.
Figure 1. X-ray structure of indene 10.
structures of 9 and 10 were determined from 1H NMR, 13C NMR,
and HRMS data, and the structure of 10 was established by X-ray
crystallography (Figure 1). The observed products 9 and 10
apparently arise from the C1-C5 cyclization of enediyne 8;
interestingly, no Bergman cyclization product was observed in
these reaction mixtures.15 Thermolyses at 210 °C returned
unchanged 8.
When thermolyses of 8 were conducted in toluene or toluene-
d8, at 260 °C in the absence of 1,4-CHD, the yields of 9 and 10
were greatly reduced,16 and again no Bergman products were
observed. Significantly, in the case of the toluene-d8 reaction,
GCMS analysis found substantial deuterium incorporation in
compound 9 (34% d2, 51% d1, 15% d0). These results are fully
consistent with C1-C5 cyclization via a diradical intermediate (14c,
Scheme 3) analogous to 2.
This is not the first thermal reaction reported to yield C1-C5
cyclization products. In one very special case, a relatively strained
macrocyclic enediyne underwent C1-C5 cyclization to an indenoin-
dene, possibly via a diradical intermediate.12a,17 Of greater relevance
is Matzger’s study7 of Bergman cyclizations of 1,2-bis(phenylethy-
nyl)benzene (11, Scheme 3), where he noted, “Remarkably, benzo-
fulvenes and their hydrogenation products were detected in the product
mixture in yields of up to 2% and ∼10%, respectively. Although direct
five-membered ring cyclization may be occurring, the majority of these
products likely arise by radical attack processes as evidenced by the
higher yields realized at lower temperatures in combination with higher
CHD concentrations.” In concert with the latter statement, Schreiner
has cautioned9c that “the thermal cyclization [of 11] yielding [14] is
unlikely because of the high barrier for cyclization (ca. 40 kcal/mol at
UBS-B3LYP/6-31G*)...” This is the central question: is it really
necessary to invoke radical attack processes for the formation of
indenes in such reactions, or is the direct formation of a C1-C5
cyclized, diradical intermediate an easily accessible pathway when
favored by both electronic and steric effects, as in diaryl-substituted
enediynes 8 and 11? Our present results argue strongly in favor of the
latter conclusion.
Figure 2. BLYP/6-31G(d)-calculated TS structures for the C1-C5 (left)
and C1-C6 (right) cyclizations of compound 8.
Once again, high-level calibration of these DFT calculations was
provided by BCCD(T)/cc-pVDZ single points for the reactions of
11 (Table 2). These placed the Bergman cyclization TS 2.2 kcal/
mol below the C1-C5 TS (rather than 2.2 kcal/mol above, as found
by the DFT methods). Extrapolating from these results, the C1-C5
cyclization should still be slightly favored for compound 8.
How do these computational studies match the available
experimental data? The formation of only C1-C5 cyclization
products upon thermolysis of 8 fits nicely with the DFT results
in Table 2. Equally significant is Matzger’s carefully tabulated
data7 for the thermolysis of 11, which show that although
Bergman cyclization gave rise to the major products in his
reaction mixtures, C1-C5 cyclization products were always
present in significant amounts. The C1-C6 to C1-C5 ratio was
typically ∼5:1, suggesting that in reality the C1-C6 TS is
favored by 1.8 kcal/mol (at 300 °C) over the C1-C5, a finding
that is in excellent agreement with the BCCD(T)/cc-pVDZ
calculations for the reactions of 11.
Scheme 3. Thermal Cyclization Paths for
1,2-Bis(arylethynyl)benzenes, and Other Related Compounds
However, Matzger (and a referee) are of the opinion that the
C1-C5 cyclization products in such reactions are due to transfer
9
13550 J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008

C O M M U N I C A T I O N S
hydrogenations involving 1,4-CHD, which is routinely added as a
hydrogen source in order to prevent the polymerization of diradicals.
(Indeed, yields of both C1-C6 and C1-C5 cyclization products
increase with increasing levels of 1,4-CHD.7) For this reason, we
sought evidence of C1-C5 cyclization under conditions where
transfer hydrogenation is unlikely.
Certainly, the formation of 9 upon thermolysis of 8 in the absence
of 1,4-CHD, as well as the incorporation of two deuterium atoms
in a substantial fraction of the molecules when 9 is produced in
toluene-d8, argue that these products, at least, do not result from
transfer hydrogenations.
References
(1) (a) Jones, R. R.; Bergman, R. G. J. Am. Chem. Soc. 1972, 94, 660–661.
(b) Bergman, R. G. Acc. Chem. Res. 1973, 6, 25–31.
(2) (a) Myers, A. G.; Kuo, E. Y.; Finney, N. S. J. Am. Chem. Soc. 1989, 111,
8057–8059. (b) Nagata, R.; Yamanaka, H.; Okazaki, E.; Saito, I. Tetra-
hedron Lett. 1989, 30, 4995–4998.
(3) (a) Nicolaou, K. C.; Dai, W.-M. Angew. Chem., Int. Ed. 1991, 30, 1387–
1416. (b) Borders, D. B.; Doyle, T. W. Enediyne Antibiotics as Antitumor
agents; Marcel Dekker: New York, 1995.
(4) Schmittel, M.; Strittmatter, M.; Kiau, S. Tetrahedron Lett. 1995, 36, 4975–
4978.
(5) (a) Engels, B.; Lennartz, C.; Hanrath, M.; Schmittel, M.; Strittmatter, M.
Angew. Chem., Int. Ed. 1998, 37, 1960–1963. (b) Schmittel, M.; Mahajan,
A. A.; Bucher, G.; Bats, J. W. J. Org. Chem. 2007, 72, 2166–2173.
(6) (a) Schmittel, M.; Kiau, S.; Siebert, T.; Strittmatter, M. Tetrahedron Lett.
1996, 37, 7691–7694. (b) Schmittel, M.; Maywald, M.; Strittmatter, M.
Synlett 1997, 165–166.
Even stronger evidence of a thermal C1-C5 cyclization is pro-
vided by thermolyses of compound 11 at much lower temperatures
than employed by Matzger. The activation energies of Table 2
suggest that the C1-C5 reaction might have a measurable rate even
below 200 °C, and indeed, thermolysis of 11 in toluene at 180 °C
for 17 h gave ca. 0.2% conversion to (Z)-1-benzylidene-2-phenyl-
1H-indene (16), 0.2% 2,3-diphenylnaphthalene (17), and, interest-
ingly, no detectable (E)-indene 18, as judged by GC-MS analysis
of the reaction mixtures with authentic standards of each compound
for comparison.18 The significance of this experiment is twofold.
First, a chain transfer hydrogenation at such a low temperature
without 1,4-CHD is highly unlikely. Second, the observed (Z)-
isomer 16 is the main product expected from a thermal C1-C5
cyclization, but the (E)-isomer 18 would be the principal product
from a transfer hydrogenation.19 However, when 1,4-CHD is
included in the thermolysis of 11 at 180°, compounds 16 (1.3%)
and 17 (0.2%) as well as 18 (0.2%) and the tetrahydro product 20
(0.3%) are observed, an indication that transfer hydrogenation may
now occur.19
The simplest interpretation of the aggregate computational and
experimental data is that enediynes 8 and 11 do undergo thermal
C1-C5 diradical cyclizations, although transfer hydrogenation likely
contributes to the formation of C1-C5 cyclization products in the
presence of 1,4-CHD.
The mechanistic features that result in the switch from a
C1-C6 to a C1-C5 pathwaysstabilization of the evolving
diradical intermediate in the C1-C5 pathway and a steric conflict
between substituents on the alkyne termini in the Bergman
TSsare precisely the features also responsible for the switch
fromtheMyers-SaitototheSchmittelcyclizationofenyne-allenes.4,5a
Thus, there is ample theoretical and experimental evidence that
C1-C5 diradical cyclizations can be observed in the thermal
reactions of enediynes.
(7) Lewis, K. D.; Matzger, A. J. J. Am. Chem. Soc. 2005, 127, 9968–9969.
(8) (a) Ramkumar, D.; Kalpana, M.; Varghese, B.; Sankararaman, S.; Jagadeesh,
M. N.; Chandrasekhar, J. J. Org. Chem. 1996, 61, 2247–2250. (b) Alabugin,
I. V.; Kovalenko, S. V. J. Am. Chem. Soc. 2002, 124, 9052–9053.
(9) (a) Whitlock, H. W., Jr.; Sandvick, P. E. J. Am. Chem. Soc. 1966, 88,
4525–4526. (b) Whitlock, H. W., Jr.; Sandvick, P. E.; Overman, L. E.;
Reichardt, P. B. J. Org. Chem. 1969, 34, 879–886. (c) Schreiner, P. R.;
Prall, M.; Lutz, V. Angew. Chem., Int. Ed. 2003, 42, 5757–5760.
(10) (a) Konig, B.; Pitsch, W.; Klein, M.; Vasold, R.; Prall, M.; Schreiner, P. R.
J. Org. Chem. 2001, 66, 1742–1746. (b) Kovalenko, S. V.; Peabody, S.;
Manoharan, M.; Clark, R. J.; Alabugin, I. V. Org. Lett. 2004, 6, 2457–
2460.
(11) Lee, C.-Y.; Wu, M.-J. Eur. J. Org. Chem. 2007, 3463–3467.
(12) (a) Prall, M; Wittkopp, A.; Schreiner, P. R. J. Phys. Chem. A 2001, 105,
9265–9274. (b) Stahl, F.; Moran, D.; Schleyer, P. R.; Prall, M.; Schreiner,
P. R. J. Org. Chem. 2002, 67, 1453–1461. (c) Schreiner, P. R.; Navarro-
Vazquez, A.; Prall, M. Acc. Chem. Res. 2005, 38, 29–37.
(13) At the RBLYP/6-31G(d) level of theory, most C1-C5 cyclization diyls (e.g.,
2) are not potential minima; however, the monoaryl diyls are shallow minima,
and thus computational TSs (Table 1) leading from enediynes 7b-f to the
C1-C5 diyls can be located by using TS search algorithms in Gaussian 03.
(We have used similar RBLYP calculations for the design of substrates for
diradical cyclizations; see: Semmelhack, M. F.; Wu, L.; Pascal, R. A., Jr.; Ho,
D. M. J. Am. Chem. Soc. 2003, 125, 10496–10497.) For the location of
both the C1-C5 and C1-C6 cyclization diaryl diyls, as well as the C1-C5
TS structures (Table 2), the BS-UBLYP/6-31G(d) method recommended
by Schreiner12a,c was employed. All reported DFT activation energies
include ZPE corrections from frequency calculations at the same level.
Activation energies obtained from BCCD(T)/cc-pVDZ single-point calcula-
tions at RBLYP/6-31G(d) or BS-UBLYP/6-31G(d) geometries (again
following Schreiner12a,c) include RBLYP or BS-UBLYP ZPE corrections,
respectively.
(14) Yoshimura, T. Polymer Bull. 1993, 31, 511–516.
(15) GC-MS analyses of the crude reaction mixtures found no compounds with
the formula C22H10Cl6 other than 9, and the characteristic naphthalene
resonances of a Bergman cyclization product were not observed in the 1
H
NMR spectra of the mixtures. Reduction of 9 to 10 is well precedented for
similar systems.7,8b
(16) C1-C5 cyclization products made up 15-35% of the volatile chlorinated
products, and 8 was entirely consumed, but because of polymer formation
in the absence of 1,4-CHD, the absolute yields were only 0.5-1.5%.
(17) Chakraborty, M.; Tessier, C. A.; Youngs, W. J. J. Org. Chem. 1999, 64,
2947–2949.
(18) 17 is commercially available, 16 (along with 17) was prepared by TPP-
sensitized photolysis of 11,8a 18 was prepared by Bu3SnH-induced radical
cyclization of 11,8b and a mixture of 16 and 18 (and many other products)
was generated by lithium naphthalide reduction of 11.9b
(19) At the BLYP/- and B3LYP/6-31G(d) levels, (E)-isomer 18 is more stable
than (Z)-isomer 16 by ∼1.5 kcal/mol; thus the excess of 16 reflects a
nonequilibrium mixture of products. Bu3SnH-induced cyclization of 11 gives
predominantly 18,8b and one would expect transfer hydrogenations to do
the same. Further, at the BS-UBLYP/6-31G(d) level, only the (Z)-isomer
of diyl 14 is a minimum, and at the UBLYP/6-31G(d) level, only the (E)-
isomer of H-atom addition product 19 is a minimum (although both have
wide C5-C6-Ar bond angles). To the extent that the structures of these
radicals are reflected in the products’ stereochemistry, the formation of
less stable 16 is an indicator of diradical cyclization.
Acknowledgment. This work was supported by National Science
Foundation Grant CHE-064879 and Petroleum Research Fund Grant
45801-AC4, which are gratefully acknowledged.
Supporting Information Available: Synthetic procedures, 1H and
13C NMR spectra for 8-10; a crystallographic information file (CIF)
for 10; GC-MS data for deuterium incorporation into 9; and an ASCII
text file containing calculated coordinates of enediynes 7b-f, 8, and
11, C1-C5 and C1-C6 cyclization transition states for 7b-f, and
cyclization transition states 12, 12c, 13, and 13c. This material is
available free of charge via the Internet at http://pubs.acs.org.
JA803413F
9
J. AM. CHEM. SOC. VOL. 130, NO. 41, 2008 13551