Fast fragmentation and slow protonation: A buffer-dependent isotope effect in reactions of N-methyl hydroxy(benzylthiamine) analyzed by the Keeffe-Jencks equations

Article abstract of DOI:10.1002/poc.764

The fragmentation of 2-(1-hydroxybenzyl)thiamine in neutral solution (to cleave the pyrimidine and thiazolium) has been shown to compete very effectively with elimination of benzaldehyde to produce thiamine in neutral solution. The fragmentation is believed to involve protonation competing with C-N bond cleavage in the C2α conjugate base as a rate-determining step. We report that proton removal from C2α of N1′-methyl-2-(1-hydroxybenzyl) thiamine (MHBnT) is rate-limiting in low concentrations of pH 6 phosphate buffer: reprotonation competes with the subsequent fragmentation step (cleaving the pyrimidine-thiazolium bridge derived from thiamine) at higher buffer concentrations. Comparison of the observed rates of reaction of protio and C2α-deutero MHBnT reveals a non-linear variation of the kinetic isotope effect that fits precisely to a ratio derived from the Keeffe-Jencks rate law formulation for El_CB reactions. The fragmentation step is clearly distinct from the proton removal step and the isotope sensitivity is limited to the initial step. The variation of the isotope effect is a result of changes due to differing contributions from the hydroxide and buffer-catalyzed reaction mechanisms. Copyright

Full text of DOI:10.1002/poc.764

JOURNAL OF PHYSICAL ORGANIC CHEMISTRY
J. Phys. Org. Chem. 2004; 17: 507–510
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/poc.764
Fast fragmentation and slow protonation:
a buffer-dependent isotope effect in reactions
of N-methyl hydroxy(benzylthiamine) analyzed
by the Keeffe--Jencks equationsy
Glenn Ikeda and Ronald Kluger*
Davenport Chemistry Laboratory, Department of Chemistry, University of Toronto, Toronto, Ontario, Canada M5S 3H6
Received 24 June 2003; revised 22 September 2003; accepted 22 September 2003
ABSTRACT: The fragmentation of 2-(1-hydroxybenzyl)thiamine in neutral solution (to cleave the pyrimidine and
thiazolium) has been shown to compete very effectively with elimination of benzaldehyde to produce thiamine
in neutral solution. The fragmentation is believed to involve protonation competing with C—N bond cleavage in
the C2ꢀ conjugate base as a rate-determining step. We report that proton removal from C2ꢀ of N10-methyl-2-
(1-hydroxybenzyl)thiamine (MHBnT) is rate-limiting in low concentrations of pH 6 phosphate buffer: reprotonation
competes with the subsequent fragmentation step (cleaving the pyrimidine–thiazolium bridge derived from thiamine)
at higher buffer concentrations. Comparison of the observed rates of reaction of protio and C2ꢀ-deutero MHBnT
reveals a non-linear variation of the kinetic isotope effect that fits precisely to a ratio derived from the Keeffe–Jencks
rate law formulation for E1CB reactions. The fragmentation step is clearly distinct from the proton removal step and
the isotope sensitivity is limited to the initial step. The variation of the isotope effect is a result of changes due
to differing contributions from the hydroxide and buffer-catalyzed reaction mechanisms. Copyright # 2004
John Wiley & Sons, Ltd.
KEYWORDS: fragmentation; protonation; isotope effect; thiamine; Keeffe–Jencks equations
INTRODUCTION
The reaction is catalyzed by hydroxide and other anio-
nic Brønsted bases, promoting the loss of the proton from
C2ꢀ. At pH 7, the rate of conversion of HBnT to the
fragmentation products is about 1000 times greater than
the rate of formation of thiamine and benzaldehyde. The
proportion of fragmentation relative to elimination reflects
the extent to which HBnT is protonated at N10.7 Jencks
noted that such a result does not require that the kinetically
active species be that which involves the thermodynamic
siteofprotonation. Hesuggeststhatproducingacompound
with an alkyl group in place of the proton can simplify the
problem: ‘Model Compounds. The problem with the pro-
ton is that it is mobile, and its position on one or another
atom in the transition state cannot be decided from the rate
law of a reaction. Now if one substitutes a methyl group for
the proton, the position of a group which differs only
slightly from the proton in its polar charge is known and
from the behavior of the model compound the behavior of
the corresponding protonic compound may be inferred.’8
Thus, the N10 methylpyrimidinium species (MHBnT),
prepared from HBnT and dimethyl sulfate, fragments
without competition from the elimination reaction (even
in alkaline solutions where HBnT releases benzaldehyde).
The rate is consistent with that observed for the proto-
nated species.7 The N10-benzylpyrimidinium compound
(BHBnT) also follows the same reaction patterns.9
The base-catalyzed addition of thiamine to benzaldehdye
produces 2-(1-hydroxybenzyl)thiamine (HBnT), an inter-
mediate in the thiamine-catalyzed benzoin condensation.
The reaction requires loss of the proton from C2 of the
thiazolium ring, generating the ylide.1,2 This elusive
species was definitively studied by Washabaugh and
Jencks.3–5 In the benzoin condensation, the C2ꢀ proton
of HBnT is transferred to a base, with the resulting anion
adding to the carbonyl carbon of a second benzaldehyde.
In alkaline solution (pH > 9) in the absence of benzalde-
hyde, HBnT reverts to thiamine and benzaldehyde.
However, in neutral and acidic solutions HBnT does
not revert to thiamine and benzaldehyde but instead
fragments to the alternative set of products, dimethyla-
minopyrimidine (DMAP) and phenyl thiazole ketone
(PTK)6 (Scheme 1).
*Correspondence to: R. Kluger, Davenport Chemistry Laboratory,
Department of Chemistry, University of Toronto, Toronto, Ontario,
Canada M5S 3H6.
E-mail: rkluger@chem.utoronto.ca
Contract/grant sponsor: Natural Sciences and Engineering Council of
Canada.
ySelected paper part of a special issue entitled ‘Biological Applications
of Physical Organic Chemistry dedicated to Prof. William P. Jencks’.
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2004; 17: 507–510

508
G. IKEDA AND R. KLUGER
according to the method described by Zoltewicz.14
MHBnT was recovered as a white solid in 18% overall
yield. 1H NMR (300 MHz, DMSO-d6): ꢁ 9.3 (s, 1H), 8.5
(s, 1H), 7.7 (s, 1H), 7.4–7.3 (m, 5H), 6.8 (s, 1H), 6.3
(s, 1H), 5.3 (s, 2H), 3.8 (m, 2H), 3.5 (s, 3H), 3.1 (m, 2H),
2.5 (s, 3H), 2.3 (s, 3H). ESI-MS (high resolution)
[C20H25N4O2S]2þ: calcd 385.1692; found 385.1679.
For 2ꢀ-deutero-MHBnT, benzaldehyde-1-d was used
with a similar work-up substituting deuterium chloride
and deuterium oxide in place of hydrogen chloride and
water. The incorporation of deuterium was confirmed by
the lack of the C2ꢀ proton signal in the proton NMR
spectrum and the exact mass and fragmentation pattern
observed by electron ionization mass spectrometry. Over-
Scheme 1
Extensive kinetic analysis of phosphate-catalyzed frag-
mentationofHBnTanditsalkylatedanaloguesshowedthat
the rate-determining step in the fragmentation of proto-
nated or alkylated HBnT changes from removal of the
proton from C2ꢀ at low buffer concentrations towards
fragmentation of the conjugate base at higher buffer con-
centrations.10 The resulting dependence of the observed
rate on buffer concentration was analyzed by the methods
used by Keeffe and Jencks for E1CB reactions.11,12 This
yielded the surprising result that the rate constant for the
fragmentation step is very large, competitive with that for
protonation of the carbanion (which can be drawn as an
enamine). A mechanistic scheme that is consistent with the
observed kinetics is shown in Scheme 2.
1
all yield 26%. H NMR (300 MHz, DMSO-d6): ꢁ 9.3 (s,
1H), 8.5 (s, 1H), 7.7 (s, 1H), 7.4–7.3 (m, 5H), 6.8 (s, 1H),
5.4 (s, 2H), 3.8 (m, 2H), 3.6 (s, 3H), 3.1 (m, 2H), 2.5 (s,
3H), 2.3 (s, 3H). ESI-MS (high resolution)
[C20H24DN4O2S]2þ: calcd 386.1754; found 386.1755.
Kinetic measurements
This mechanism requires that there be a primary (H/D)
kinetic isotope effect under all conditions. Thus, we have
investigated the kinetic isotope effect as a function of buf-
fer concentration in the fragmentation of MHBnTand have
found that the results are consistent with this expectation.
The fragmentation of MHBnT was followed by monitor-
ing the increase in absorbance at 328 nm (ꢂmax for PTK).
Data were collected on an interfaced computer and
analyzed with a curve-fitting program. After establishing
that the observed kinetics fit a first-order rate law, we used
the method of initial rates to obtain additional rate data.
In the case of the deuterated reactant, this is necessary
because the isotope exchange competes with the overall
reaction, converting the remaining deutero-MHBnT to
the unlabeled form in direct competition with fragmenta-
tion. Reactions were followed to conversion of 2% of the
reactant, transferred to vials and incubated at 40 ꢀC in a
water-bath. After 10 half-lives the final absorbance was
measured to determine the total PTK that was formed (to
provide accurate concentration measurements in each
sample).
EXPERIMENTAL
Potassium phosphate buffer solutions were kept in a
jacketed beaker maintained at 40 ꢀC. A pH electrode
was standardized against reference solutions at the
same temperature. Ionic strength was maintained at 1.0
by the addition of potassium chloride. The dependence of
observed rates on buffer concentration was measured in
solutions at pH 6.1, where [H2PO4ꢁ] ¼ 2[HPO42ꢁ].
Synthesis of MHBnT and C2a-deutero-MHBnT
HBnT was synthesized by the procedure of Doughty
et al.13 followed by methylation by dimethyl sulfate
RESULTS
The variation in the rate of fragmentation of MHBnT in
0.01–0.4 M potassium phosphate buffer at pH 6.1 is
concave downwards, consistent with a change in rate-
determining step with increasing buffer concentration.
Although the rate does not become independent of buffer
concentration in this range, the change is sufficient to
provide accurate data analysis to determine the rate
constants and ratios that generate the curvature.
Application of the steady-state approximation to
Scheme 2 gives the equation
ðkB½Bꢂ þ kOH½OHꢁꢂÞkf
kobs
¼
ð1Þ
kBH½BHꢂ þ kH O þ kf
2
Scheme 2
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2004; 17: 507–510

FAST FRAGMENTATION AND SLOW PROTONATION
509
Figure 2. Observed KIE on fragmentation as a function of
hydrogen phosphate buffer concentration
Figure 1. Dependence of the observed first-order rate
coefficient on buffer concentration for MHBnT (*) and
MHBnT-C2ꢀ-d (~) at 40 ꢀC, I ¼ 1.0. Curves are fitted to
the Keeffe--Jencks equation for a mechanism with a buffer-
dependent change in rate-determining step
which describes the relationship between the data points
in Fig. 2. From this relationship, it is apparent that the
isotope effect is on the hydroxide-or buffer-catalyzed
proton removal step (Table 2). At low buffer concentra-
tion, the isotope effect on proton removal due to hydro-
xide is observed. As buffer is increased, the isotope effect
on buffer catalysis becomes the dominant contributor.
Exchange-out of the deuterium from C2ꢀ in H2O must be
considerably faster than deuteration of the protio com-
pound in D2O. The solvent isotope effect on the overall
fragmentation is inverse because the intermediate is more
rapidly protonated in H2O.
which describes a rectangular hyperbola. The data points
in Fig. 1 are fitted to this equation, which is based on a
mechanism with the rate-determining step changing from
proton transfer to fragmentation with increasing buffer
concentration. From this plot, the following values were
obtained by fitting the data graphically accordꢁin1g to
the Keeffe–Jencks procedure:12 k ¼ 7.7 ꢃ 10ꢁ5 s and
1
1
k0 ¼ 1.1 ꢃ 10ꢁ5 sꢁ1 for MHBnT; k ¼ 2.2 ꢃ 10ꢁ5 sꢁ1 and
k0 ¼ 2.4 ꢃ 10ꢁ6 sꢁ1 for MHBnT-C2ꢀ-d.
Further application allows the determination of impor-
tant rate constants and ratios (Table 1) using the extra-
DISCUSSION
polated values for k0 ( ¼ kobs at [Bꢁ] ¼ 0) and k ( ¼ kobs
The values for kBꢁ indicate that the rate of fragmentation
is largely dependent on the basic component of the buffer.
The fitted values that give kBH/kf establish that fragmen-
tation competes with the kinetically significant rate
of protonation of the C2ꢀ conjugate base. In the case
of N10-benzyl-2-(ꢁ1-1hydroxybenzyl)thiamine (BHBnT),9
1
where the rate is independent of buffer concentration).
The ratio of the curves generated from the data in Fig. 1
yields the kinetic isotope effect as a function of buffer
concentration (Fig. 2). The ratio of the two rate equations
for the protio- and deutero-substrates yields a function
that describes the buffer dependence of the isotope effect.
This relationship can be simplified by the assumption that
both the second step (kf) and reprotonation of the enam-
ine (by water or buffer) are isotope-independent as they
are subsequent to removal of the proton or deuteron. This
leads to the equation
kBH/kf ¼ 56 ꢄ 4 M
. Therefore, fragmentation may
compete more effectively with protonation in MHBnT
than in BHBnT. This would cause saturation of buffer
catalysis to require a higher effective concentration for
MHBnT, as we report here. The result is consistent with
protonation of MHBnT being slower than protonation of
BHBnT, fragmentation being faster in MHBnT, or a
combination of both.
kBH½Bꢂ þ kOHH½OHꢁꢂ
The rate constant for buffer-catalyzed proton removal
is about the same in the two compounds (based on kBꢁ). If
we make the reasonable assumption that the pKa for
kobs
¼
ð2Þ
kBD½Bꢂ þ kODH½OHꢁꢂ
Table 1. Derived rate constants and ratios for fragmenta-
tion of MHBnT (H/D) and BHBnT
Table 2. Derived isotope effects on rate constants for the
fragmentation of MHBnT (H/D)
ꢁ1
Substrate
kB (Mꢁ1 sꢁ1
)
kBH/kf (M
)
ꢁ
Ratio (H/D)
Isotope effect
ꢁ3
ꢁ4
ꢁ3
MHBnT
MHBnT-C2ꢀ-d
BHBnT
(1.1 ꢄ 0.1) ꢃ 10
(1.9 ꢄ 0.1) ꢃ 10
(7.4 ꢄ 0.5) ꢃ 10
6.6 ꢄ 0.3
4.1 ꢄ 0.2
56 ꢄ 4
k0
kB
4.4
5.8
ꢁ
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2004; 17: 507–510

510
G. IKEDA AND R. KLUGER
proton loss at C2ꢀ is independent of the group on N10,
then the rate constant for protonation of the intermediate
is different in the two intermediates. Alternatively, the
observation could result from the fragmentation step or
steps being faster in MHBnT. At this point we must note
the possibilities rather than choose among them.
the buffer-catalyzed process. However, MHBnT has two
localized positive charges that might affect the geometry
of the encounter with the phosphate ion with its multiple
negative charges. The solvation of hydroxide will also
be very different so that the isotope effect will need to
be determined in more examples in order to establish if
these comparisons are general.
Estimating the pKa for the two compounds leads to the
values of kBH and kf. We estimate the pKa of the N0-
alkylated HBnT to be 14.9. This is derived from the
measured pKa of 15.7 for the related methoxybenzyl
methylthiazolium salt,15 taking into account the inductive
effect of the N10-alkylated pyrimidine ring and the
methoxy substituent at C2ꢀ.16 MHBnT and BHBnT
are likely to have similar pKas for their respective C2ꢀ
carbon acids as there is significant distance between N10
and C2ꢀ. Assuming similar pKas for both compounds
gives kBH ¼ 2.1 ꢃ 106 Mꢁ1 sꢁ1 and kf ¼ 3.8 ꢃ 104 sꢁ1 for
BHBnT versus kBH ¼ 3.2 ꢃ 105 Mꢁ1 sꢁ1 and kf ¼ 4.8 ꢃ
104 sꢁ1 for MHBnT. Therefore, it becomes apparent that
the smaller kBH/kf ratio measured for MHBnT is probably
a consequence of reprotonation being slower and frag-
mentation being slightly faster. Whereas kBH differs by an
order of magnitude, kf is similar for both compounds.
This suggests that the fragmentation step is not driven by
interactions with the pyrimidine substituents. Thus, dif-
ferences in kBH would contribute a larger perturbation to
the differing rate constant ratios (kBH/kf).
Decarboxylation of the conjugate of thiamine
diphosphate and benzoylformate in the mechanism of
benzoylformate-decarboxylase generates the conjugate
base at C2ꢀ of the diphosphate of HBnT. The enzyme
appears to function without fragmentation of the cofactor.
Our results show that fragmentation competes very effec-
tively with protonation. Since the decarboxylation should
lead to an intermediate that is relatively long-lived, the
enzyme does not rely on the addition of a proton from a
Brønsted acid to avoid the destructive reaction.19,20
Acknowledgments
We thank the Natural Sciences and Engineering Council
of Canada for support through a Discovery Grant. This
paper is dedicated to Dr Bill Jencks, an inspiring and
brilliant scientist. By following his rules we discover the
questions as well as the answers.
The KIEs observed in this study are large, consistent
with proton removal being involved in the rate-limiting
step at low buffer concentrations. The magnitude of the
KIEs are in the range of Jordan and co-workers’ reported
values of 4–6 for the hydroxide-catalyzed removal of a
proton from substituted 2-(1-methoxybenzyl)-3,4-di-
methylthiazol-3-ium salts.15,17 The theoretical maximum
for a PKIE involving a carbon–hydrogen bond being
broken in the rate-determining step at room temperature
is ꢅ7.18 A more reactant-like or product-like transition
state decreases this value. Therefore we conclude that
within this buffer range, a carbon–hydrogen bond is being
broken in the transition state of the rate-determining step,
with the proton being centrally located between the
Brønsted base and substrate. In the absence of buffer,
the KIE of 4.4 involves rate-determining transfer of a
proton or deuteron to hydroxide. An increase in the
isotope effect is observed where the introduction of buffer
begins to increase the partitioning between the deuterio
and protio substrates. This is a consequence of the term
due to buffer catalysis (kBꢁ) being subject to a larger
isotope effect.
REFERENCES
1. Breslow R. J. Am. Chem. Soc. 1957; 79: 1762–1763.
2. Breslow R. Chem. Ind. (London) 1957; 26: 893–894.
3. Washabaugh MW, Jencks WP. Biochemistry 1988; 27: 5044–
5053.
4. Washabaugh MW, Jencks WP. J. Am. Chem. Soc. 1989; 111: 674.
5. Washabaugh MW, Jencks WP. J. Am. Chem. Soc. 1989; 111:
683.
6. Kluger R, Lam JF, Kim C-S. Bioorg. Chem. 1993; 21: 275–283.
7. Kluger R, Lam JF, Pezacki JP, Yang C-M. J. Am. Chem. Soc.
1995; 117: 11383–11389.
8. Jencks WP. Catalysis in Chemistry and Enzymology. McGraw-
Hill: New York, 1969.
9. Kluger R, Moore IF. J. Am. Chem. Soc. 2000; 122: 6145–6150.
10. Moore IF, Kluger R. J. Am. Chem. Soc. 2002; 124: 1669–1673.
11. Keeffe JR, Jencks WP. J. Am. Chem. Soc. 1981; 103: 2457–2459.
12. Keeffe JR, Jencks WP. J. Am. Chem. Soc. 1983; 105: 265–279.
13. Doughty MB, Risinger GE, Jungk SJ. Bioorg. Chem. 1987; 15:
15–30.
14. Zoltewicz JA. Synthesis 1980; 3: 218–219.
15. Barletta GL, Zou Y, Huskey WP, Jordan F. J. Am. Chem. Soc.
1997; 119: 2356–2362.
16. Moore IF. PhD Thesis. University of Toronto, 2003.
17. Barletta G, Huskey WP, Jordan F. J. Am. Chem. Soc. 1992; 114:
7607–7608.
18. Westheimer FH. Chem. Rev. 1961; 61: 265–273.
19. Sergienko EA, Wang J, Polovnikova L, Hasson MS,
McLeish MJ, Kenyon GL, Jordan F. Biochemistry 2000; 39:
13862–13869.
The isotope effect on kBꢁ is 5.8, approaching the upper
limit for the conditions, whereas that on k0 is only 4.4. A
simple comparison of the magnitudes of the isotope
effects on k0 and kBꢁ might indicate that the proton is
more symmetrically positioned in the transition state of
20. Polovnikova ES, McLeish MJ, Sergienko EA, Burgner JT,
Anderson NL, Bera AK, Jordan F, Kenyon GL, Hasson MS.
Biochemistry 2003; 42: 1820–1830.
Copyright # 2004 John Wiley & Sons, Ltd.
J. Phys. Org. Chem. 2004; 17: 507–510