The Need for an Alternative to Radicals as the Cause of Fragmentation of a Thiamin-Derived Breslow Intermediate

Article abstract of DOI:10.1002/anie.201702240

Mandelylthiamin (1) is a conjugate of benzoylformate and thiamin that loses CO₂ to form the classic Breslow intermediate (2), whose expected fate is formation of the thiamin conjugate of benzaldehyde (3). Surprisingly, it was observed that 2 decomposes to 4 and 5 and rearranges to 6 in competition with the expected protonation to give 3. Recent reports propose that the alternatives to protonation arise from homolysis followed by radical-centered processes. It is now found, instead, that the spectroscopic observations cited in support of the proposed radical pathways are likely to be the result of other events. An alternative explanation is that ionization of the enolic hydroxy group of 2 and resultant electronic reorganization leads to C?C bond cleavage and non-radical intermediates that readily form 4, 5, and 6.

Full text of DOI:10.1002/anie.201702240

Angewandte
Communications
Chemie
International Edition: DOI: 10.1002/anie.201702240
German Edition: DOI: 10.1002/ange.201702240
Breslow Intermediates
The Need for an Alternative to Radicals as the Cause of Fragmentation
of a Thiamin-Derived Breslow Intermediate
Michael Bielecki and Ronald Kluger*
Abstract: Mandelylthiamin (1) is a conjugate of benzoylfor-
mate and thiamin that loses CO2 to form the classic Breslow
intermediate (2), whose expected fate is formation of the
thiamin conjugate of benzaldehyde (3). Surprisingly, it was
observed that 2 decomposes to 4 and 5 and rearranges to 6 in
Scheme 1. Decarboxylation of 1 in neutral aqueous solutions produces
competition with the expected protonation to give 3. Recent
reports propose that the alternatives to protonation arise from
homolysis followed by radical-centered processes. It is now
found, instead, that the spectroscopic observations cited in
support of the proposed radical pathways are likely to be the
result of other events. An alternative explanation is that
ionization of the enolic hydroxy group of 2 and resultant
Breslow intermediate 2 followed by protonation to the thiamin–
benzaldehyde adduct 3.
À
electronic reorganization leads to C C bond cleavage and non-
radical intermediates that readily form 4, 5, and 6.
Scheme 2. Combining thiamin with benzaldehyde and triethylamine as
base in refluxing methanol results in products 4, 5, and 6 via
a fragmentation and rearrangement of 2, respectively.
B
reslowꢀs insightful studies showed that the formation of
acyl carbanion equivalents in decarboxylases result from
addition of a nucleophilic carbene derived from enzyme-
bound thiamin diphosphate to 2-ketoacids.[1] This led to
generalizations in which the addition of thiamin-related N-
heterocyclic carbenes (NHC) to aldehydes lead to syntheti-
cally useful acyl carbanion synthons.[2]
Mandelylthiamin (1, Scheme 1), the conjugate of thiamin
and benzoylformate, loses CO2 to produce the expected
Breslow intermediate 2. Protonation gives isolable 2a-(1-
hydroxybenzyl)thiamin (3).[3]
An attempt by Oka to achieve a thiamin-catalyzed benzoin
condensation of benzaldehyde via formation of 2 unexpectedly
gave 4, 5, and 6 (Scheme 2).[4] Studies of the reactions of 2
revealed its spontaneous fragmentation to Okaꢀs products 4 and
5 where the absorbance of 5 at 328 nm is characteristic,
providing a spectroscopic handle for kinetic studies.[5]
computations indicate that the proposed radicals are ener-
getically accessible. Nonetheless, it is surprising to encounter
formation of radicals under the conditions in which we see
similar products from 2.
McIntosh and co-workers generate their Breslow inter-
mediate in methanol by combining benzaldehyde with the
NHC precursor and DBU in the presence of oxygen. Under
these conditions, other reactions can produce radicals: 1) air-
oxidation of benzaldehyde;[8] 2) reaction of benzaldehyde
with the Breslow intermediate;[9] 3) reaction of benzaldehyde
with the conjugate acid of the intermediate;[9] 4) air-oxidation
of the intermediate;[10] and 5) air-oxidation of benzoin.[11]
These potential side reactions and lack of controls permit
neither confident assignment of a signal to the proposed
radical pair nor a basis for quantitation. Attempts to detect or
McIntosh and co-workers recently reported that N-
heterocyclic carbenes react with benzaldehyde to give prod-
ucts that parallel the outcomes in Scheme 2.[6] They ascribe
the results to reactions of radicals from homolysis of the
Breslow intermediate observed in EPR spectra and compu-
tations. They propose that radicals lead to the products from
1 in Scheme 2, both the rearrangement and fragmentation
forming via a radical pair from homolysis of the Breslow
intermediate (Scheme 3).[7] Recombination and disproportio-
nation lead to both sets of products. Simulated EPR spectra of
expected radicals are consistent with observed spectra and
[*] M. Bielecki, Prof. R. Kluger
Department of Chemistry, University of Toronto
Toronto, Ontario, M5S 3H6 (Canada)
E-mail: r.kluger@utoronto.ca
Scheme 3. A radical mechanism, based on McIntosh and co-workers’
proposal, for the formation of 4, 5, and 6. The Breslow intermediate 2
undergoes a spontaneous decomposition to a radical pair. The pair
undergoes disproportionation to form 4 and 5 and recombination to
give 6.
Supporting information for this article can be found under:
https://doi.org/10.1002/anie.201702240.
Angew. Chem. Int. Ed. 2017, 56, 1 – 4
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1
These are not the final page numbers!

Angewandte
Communications
Chemie
isolate TEMPO adducts that would be expected from radicals
also gave negative results.[7]
We conducted analogous studies with 5,5-dimethyl-1-
pyrroline N-oxide (DMPO). As with 2-SPBN-Na, there is no
EPR signal, while exposure to oxygen produces a signal. Given
the oxidative reactivity of 4-hydroxy-TEMPO towards 2, this
results from reduction of aminoxyl radicals by 2 to the EPR-
silent hydroxylamines. Exposure to oxygen regenerates the
radicals.[17] By integrating the EPR signal with 1.0 mm 4-
hydroxy-TEMPO solution as a standard, we estimate that
0.8 nmol of a radical species is trapped. Assuming fragmenta-
tion to be a radical process occurring at 10 mol%, this
corresponds to only 0.05 mol% interception by the spin trap.
This very low extent of trapping may arise from radical pair
recombination being faster than desolvation or from minor side
reactions. Nucleophilic additions to both nitroso and nitrone
spin traps may lead to false-positive results.[18] Hydroxylamines
that form from addition react with atmospheric oxygen to give
EPR-active aminoxyl radicals. Based on the nucleophilic
properties of Breslow intermediates, such an event can account
for a signal, which is not relevant to the fragmentation
process.[19] Finally, the small amount of benzaldehyde that
forms would be air-oxidized and then produce radicals.
It is well-established that 1 decarboxylates in neutral
aqueous buffers to form 2, which rapidly undergoes fragmen-
tation to 4 and 5 (and potentially other products).[5] Thus, we
prepared 1[12] and analyzed the products of its decarboxyla-
1
tion in phosphate buffers in D2O by H NMR spectroscopy.
We added genuine samples of potential products and
compared their NMR signals with those from the reaction
in solution.[13] We confirmed the formation of 3, 4, 5, and 6 in
relative molar concentrations of 100:30:30:3. We had not
previously noted formation of 6, as previous studies were
concerned with the rate of the decarboxylation process. We
also note the formation of small amounts (2–3 mol%) of
thiamin and benzoic acid, which could result from hydrolysis
of 2-benzoylthiamin upon oxidation of 2.[14] However, in the
absence of oxygen, the products do not include benzoic acid.
In the presence of oxygen, benzaldehyde, produced from 3,
reacts with oxygen to form benzoic acid.
We attempted to trap intermediate 7a, which would form
from a radical pathway, by adding the water-soluble radical
trap 4-hydroxy-2,2,6,6-tetramethylpiperidine 1-oxyl (4-hy-
droxy-TEMPO) to the reaction mixture. UV/Vis spectroscopy
revealed formation of none of the characteristic band at
328 nm from 5.[5b] Lower concentrations of 4-hydroxy-
TEMPO significantly decrease that band. Analogous reac-
tions were conducted in D2O and studied by 1H NMR
spectroscopy. These reveal only two major products: thiamin
and benzoic acid. This is consistent with oxidation of 2 by 4-
hydroxy-TEMPO as reported by Studer and co-workers, who
propose a single-electron transfer (SET): TEMPO or 4-
hydroxy-TEMPO, oxidizes the Breslow intermediate to an
acyl derivative in 2:1 stoichiometry, respectively.[15] The
aminoxyl radical[16] itself is reduced to the corresponding
hydroxylamine. In the present case, oxidation of the inter-
mediate produces 2-benzoylthiamin, which reacts rapidly with
water to produce thiamin and benzoic acid. Both the UV and
1H NMR spectra indicate that fragmentation products do not
form when the decarboxylation of 1 occurs in the presence of
mm amounts of 4-hydroxyTEMPO. The rate constant for the
fragmentation reaction of 2 at 258C is about 104 sÀ1.[5a] Since
we can detect the presence of fragmentation products at about
1 mol%, the rate constant for reaction of 4-hydroxy-TEMPO
with 2 must be at least 106 sÀ1. This suggests why McIntosh and
co-workers do not isolate TEMPO-radical conjugates.[7]
We also investigated the fragmentation of 2 using nitrone
spin traps and EPR. With N-tert-butyl-a-(2-sufophenyl)ni-
trone sodium salt (2-SPBN-Na) as a water-soluble spin trap,
EPR established a limit of detection of 100 nm with 4-
hydroxy-TEMPO standards. The samples contained 10 mm
1 and 20 mm 2-SPBN-Na in pH 7.0 phosphate buffer. All of
the EPR experiments where conducted in the absence of
oxygen and light. UV spectroscopy reveals that 10 mol%
1 undergoes fragmentation. We recorded the initial EPR
spectrum within 20 min and another after 5 h and a reference
spectrum in the absence of 1. There is no EPR signal over 5 h.
However, exposure of the reaction mixture after 5 h to
oxygen produces a signal. The control did not produce a signal
when exposed to oxygen.
If a radical pair were responsible for the EPR signal, our
spin-trapping results indicate that radical disproportionation
and recombination would have to be faster than desolvation.
From our product studies, the radical disproportionation/
recombination ratio (kdisp/krec), from the relative amounts of 4
and 5 to 6, would have to be about 10:1. Studies on resonance-
stabilized radical pair recombination and disproportionation
reactions show that spin delocalization greatly favors recom-
bination over disproportionation. Typically, kdisp/krec is below
0.1.[20,21] Although steric factors may affect recombination,
values of kdisp/krec > 1 are not accessible.[21] Both 7a and 7b
exhibit delocalization of electron spin into aromatic rings.
Moreover, the fragmentation, which is a b-elimination, would
À
require homolysis of an RO H bond, a process that is not
normally accessible.[22] Therefore, in a radical process more of
the rearrangement product (6) would form than would 4 and
5. Under these circumstances, kdisp/krec ꢀ 10 is inconsistent
with reactivity patterns of radical pairs.[20a]
Thus, decarboxylation of 1 in aqueous solutions produces
2 from which the rearrangement product 6 and fragmentation
products 4 and 5 are clearly formed. Therefore, the reactivity
of
2 parallels the reactivity of Breslow intermediates
described by McIntosh and co-workers. Unlike generation
of a Breslow intermediate from a mixture containing a thia-
min-like carbene precursor, benzaldehyde, and base, decar-
boxylation of 1 avoids side reactions that lead to radicals. Our
spin-trapping experiments produce only a weak EPR signal,
corresponding to 0.05 mol% of the total amount of radicals
that would form if fragmentation were a radical process. The
rapid reduction of aminoxyl radicals by the Breslow inter-
mediate poses an experimental challenge where oxygen
would rescue the reduced spin-trap adducts. However, this
will also lead to oxidation of hydroxylamines that would have
been formed by nucleophilic additions to the spin trap, giving
an irrelevant EPR signal. The product distribution is also
inconsistent with a radical pair mechanism, where the
recombination product (6) would exceed the fragmentation
products (4 and 5).
2
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 4
These are not the final page numbers!

Angewandte
Communications
Chemie
[1] R. Breslow, J. Am. Chem. Soc. 1957, 79, 1762 – 1763.
[2] D. M. Flanigan, F. Romanov-Michailidis, N. A. White, T. Rovis,
Chem. Rev. 2015, 115, 9307 – 9387.
[3] R. Kluger, Q. Hu, J. Am. Chem. Soc. 2002, 124, 14858 – 14859.
[4] Y. Oka, S. Kishimoto, Chem. Pharm. Bull. 1970, 18, 534 – 541.
[5] a) R. Kluger, I. F. Moore, J. Am. Chem. Soc. 2000, 122, 6145 –
6150; b) R. Kluger, J. F. Lam, C. S. Kim, Bioorg. Chem. 1993, 21,
275 – 283.
[6] S. Alwarsh, K. Ayinuola, S. S. Dormi, M. C. McIntosh, Org. Lett.
2013, 15, 3 – 5.
[7] S. Alwarsh, Y. Xu, S. Y. Qian, M. C. McIntosh, Angew. Chem.
Int. Ed. 2016, 55, 355 – 358; Angew. Chem. 2016, 128, 363 – 366.
[8] M. Sankar, E. Nowicka, E. Carter, D. M. Murphy, D. W. Knight,
D. Bethell, G. J. Hutchings, Nat. Commun. 2014, 5, 3332.
[9] J. Rehbein, S.-M. Ruser, J. Phan, Chem. Sci. 2015, 6, 6013 – 6018.
[10] a) O. Bortolini, C. Chiappe, M. Fogagnolo, A. Massi, C. S.
Pomelli, J. Org. Chem. 2017, 82, 302 – 312; b) S. De Sarkar, A.
Biswas, R. C. Samanta, A. Studer, Chem. Eur. J. 2013, 19, 4664 –
4678.
Scheme 4. An anionic mechanism for the fragmentation and rear-
rangement reactions of the Breslow intermediate.
A more likely mechanism for fragmentation and rear-
rangement involves a reactive carbanion (4’; Scheme 4). We
propose that enol 2 ionizes to form a zwitterion (2’) by
transfer of the proton from its hydroxyl group substituent,
[11] E. G. Delany, C.-L. Fagan, S. Gundala, K. Zeitler, S. J. Connon,
Chem. Commun. 2013, 49, 6513 – 6515.
À
[12] M. Bielecki, G. W. Howe, R. Kluger, Bioorg. Chem. 2015, 62,
124 – 129.
followed by C N bond-breaking. The product (5) derives
stability from re-aromatization of the thiazole ring and drives
elimination of a charge-stabilized carbanion. The ionization
of the OH group of 2 closely resembles that of the enol form
of mandelic acid with its trio of electronegative atom
substituents acidifying the hydroxyl group with pKA = 6.[23]
The incorporation of a single deuterium into the CH3 group of
4 upon fragmentation in D2O is also consistent with the
formation of the carbanion precursor and with the formation
of 6 from the addition of 4’ to 5.[24] D2O cannot serve as
a source of deuterium atoms in a radical mechanism, owing to
the very high energy of hydroxyl radicals.[25]
[13] A slight turbidity was observed as the reaction approached
completion, which is due to the low solubility of 5 and 6. To
assure complete dissolution of all products, a few drops of
MeOD were added before acquiring 1H NMR spectra.
[14] a) G. E. Lienhard, J. Am. Chem. Soc. 1966, 88, 5642 – 5649;
b) F. G. White, L. L. Ingraham, J. Am. Chem. Soc. 1962, 84,
3109 – 3111.
[15] J. Guin, S. De Sarkar, S. Grimme, A. Studer, Angew. Chem. Int.
Ed. 2008, 47, 8727 – 8730; Angew. Chem. 2008, 120, 8855 – 8858.
[16] IUPAC recommends using “aminoxyl radical/s” rather than
“nitroxide radical/s” or “nitroxides”: Pure Appl. Chem. 1995, 67,
1307 – 1375, see p 1352..
In conclusion, our results indicate that the hypothetical
formation of radicals does not account for the products of
fragmentation from Breslow intermediate 2. It is also
important to note that enzymes produce Breslow intermedi-
ates that do not undergo fragmentation or rearrangement.[26]
Yet, the potential fragmentation rate constant is larger than
the enzymic kcat. This is consistent with enzymes avoiding
delocalization of the C2a carbanion as an incidental conse-
quence in their evolved efficiency.[27]
[17] a) M. Sentjurc, R. P. Mason, Free Radical Biol. Med. 1992, 13,
151 – 160; b) V. Nothig-Laslo, A. M. Bobst, Croat. Chem. Acta
1991, 64, 1 – 8.
[18] a) A. R. Forrester, S. P. Hepburn, J. Chem. Soc. C 1971, 701 –
703; b) F. Leinisch, J. Jiang, E. F. DeRose, V. V. Khramtsov, R. P.
Mason, Free Radical Biol. Med. 2013, 65, 1497 – 1505; c) M.
Triquigneaux, B. Tuccio, R. Lauricella, L. Charles, J. Am. Soc.
Mass Spectrom. 2009, 20, 2013 – 2020.
[19] M. N. Hopkinson, C. Richter, M. Schedler, F. Glorius, Nature
2014, 510, 485 – 496.
[20] a) R. Klein, R. D. Kelley, J. Phys. Chem. 1975, 79, 1780 – 1784;
b) S. F. Nelsen, P. D. Bartlett, J. Am. Chem. Soc. 1966, 88, 137 –
143.
Acknowledgements
[21] K. R. Kopecky, M. Y. Yeung, Can. J. Chem. 1988, 66, 374 – 376.
[22] a) R. S. Davidson, Q. Rev. Chem. Soc. 1967, 21, 249 – 258;
b) M. S. Kharasch, J. L. Rowe, W. H. Urry, J. Org. Chem. 1951,
16, 905 – 913.
We thank Derek Pratt (Ottawa) for critical discussions and
NSERC Canada for support through a Discovery Grant.
[23] Y. Chiang, A. J. Kresge, V. V. Popik, N. P. Schepp, J. Am. Chem.
Soc. 1997, 119, 10203 – 10212.
[24] G. Ikeda, R. Kluger, Can. J. Chem. 2005, 83, 1277 – 1280.
[25] H. Shaw, H. D. Perlmutter, C. Gu, S. D. Arco, T. O. Quibuyen, J.
Org. Chem. 1997, 62, 236 – 237.
[26] R. Kluger, K. Tittmann, Chem. Rev. 2008, 108, 1797 – 1833.
[27] S. Lꢁdtke, P. Neumann, K. M. Erixon, F. Leeper, R. Kluger, R.
Ficner, K. Tittmann, Nat. Chem. 2013, 5, 762 – 767.
Conflict of interest
The authors declare no conflict of interest.
À
Keywords: Breslow intermediates · C C bond cleavage ·
fragmentation · reaction mechanisms · thiamin
Manuscript received: March 1, 2017
Final Article published: && &&, &&&&
Angew. Chem. Int. Ed. 2017, 56, 1 – 4
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
3
These are not the final page numbers!

Angewandte
Communications
Chemie
Communications
Breslow Intermediates
M. Bielecki, R. Kluger*
Perhaps not so radical: A Breslow inter-
mediate surprisingly gives a mixture of
products (blue). It has been proposed
that these form via radicals (red), but it is
now suggested that a non-radical process
prevails (green).
&&&& — &&&&
The Need for an Alternative to Radicals as
the Cause of Fragmentation of a Thiamin-
Derived Breslow Intermediate
4
www.angewandte.org
ꢀ 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2017, 56, 1 – 4
These are not the final page numbers!