Effects of aromatic thiols on thiol-disulfide interchange reactions that occur during protein folding

Article abstract of DOI:10.1021/jo015600a

The folding of disulfide containing proteins from denatured protein to native protein involves numerous thiol-disulfide interchange reactions. Many of these reactions include a redox buffer, which is a mixture of a thiol (RSH) and the corresponding disulfide (RSSR). The relationship between the structure of RSH and its efficacy in folding proteins in vitro has been investigated only to a limited extent. Reported herein are the effects of aliphatic and especially aromatic thiols on reactions that occur during protein folding. Aromatic thiols may be particularly efficacious as their thiol pK_a values and reactivities match those of the in vivo catalyst, protein disulfide isomerase (PDI). This investigation correlates the thiol pK_a values of aromatic thiols with their reactivities toward small molecule disulfides and the protein insulin. The thiol pK_a values of nine para-substituted aromatic thiols were measured; a Hammett plot constructed using σ_p^- values yielded ρ = -1.6 ± 0.1. The reactivities of aromatic and aliphatic thiols with 2-pyridyldithioethanol (2-PDE), a small molecule disulfide, were determined. A plot of reactivity versus pK_a of the aromatic thiols had a slope (β) of 0.9. The ability of these thiols to reduce (unfold) the protein insulin correlates strongly with their ability to reduce 2-PDE. Since the reduction of protein disulfides occurs during protein folding to remove mismatched disulfides, aromatic thiols with high pK_a values are expected to increase the rate not only of protein unfolding but protein folding as well.

Full text of DOI:10.1021/jo015600a

4244
J . Org. Chem. 2001, 66, 4244-4249
Effects of Ar om a tic Th iols on Th iol-Disu lfid e In ter ch a n ge
Rea ction s Th a t Occu r d u r in g P r otein F old in g
Todd V. DeCollo and Watson J . Lees*
Department of Chemistry, Center for Science and Technology, Syracuse University,
Syracuse, New York 13244
wjlees@syr.edu
Received February 28, 2001 (Revised Manuscript Received April 25, 2001)
The folding of disulfide containing proteins from denatured protein to native protein involves
numerous thiol-disulfide interchange reactions. Many of these reactions include a redox buffer,
which is a mixture of a thiol (RSH) and the corresponding disulfide (RSSR). The relationship between
the structure of RSH and its efficacy in folding proteins in vitro has been investigated only to a
limited extent. Reported herein are the effects of aliphatic and especially aromatic thiols on reactions
that occur during protein folding. Aromatic thiols may be particularly efficacious as their thiol pKa
values and reactivities match those of the in vivo catalyst, protein disulfide isomerase (PDI). This
investigation correlates the thiol pKa values of aromatic thiols with their reactivities toward small
molecule disulfides and the protein insulin. The thiol pKa values of nine para-substituted aromatic
-
thiols were measured; a Hammett plot constructed using σp values yielded F ) -1.6 ( 0.1. The
reactivities of aromatic and aliphatic thiols with 2-pyridyldithioethanol (2-PDE), a small molecule
disulfide, were determined. A plot of reactivity versus pKa of the aromatic thiols had a slope (â) of
0.9. The ability of these thiols to reduce (unfold) the protein insulin correlates strongly with their
ability to reduce 2-PDE. Since the reduction of protein disulfides occurs during protein folding to
remove mismatched disulfides, aromatic thiols with high pKa values are expected to increase the
rate not only of protein unfolding but protein folding as well.
Sch em e 1. F old in g of a P r otein w ith Mism a tch ed
Disu lfid e Bon d s
In tr od u ction
The production of proteins in bacteria has been es-
sential for the elucidation of protein structure and
function. Unfortunately, in many cases when proteins are
overexpressed in bacteria, they are isolated as aggregates
of inactive protein known as inclusion bodies. To obtain
active protein from the inclusion bodies, the inactive
protein is denatured and then folded in vitro.1-4 For
proteins containing disulfide bonds, the folding rate
usually is limited by a thiol-disulfide interchange reac-
tion. This reaction occurs within the protein itself or
between the protein and the redox buffer, which is a
mixture of a thiol (RSH) and the corresponding disulfide
(RSSR, Scheme 1).5 Despite intense examination of
protein folding, only a few studies have investigated the
relationship between the structure of the redox buffer and
the efficiency of protein folding. The standard redox
buffers have been mercaptoethanol or cysteine deriva-
tives such as glutathione.
each active site is essential for catalysis, has a pKa value
of 6.7, and is exceptionally reactive with disulfides.7-11
Glutathione, mercaptoethanol, and almost all other
aliphatic thiols have thiol pKa values above 7. In com-
parison, aromatic thiols have thiol pKa values between 4
and 7 and are more reactive toward aromatic disulfides
than aliphatic thiols of similar pKa values.12-15 Conse-
(6) Edman, J . C.; Ellis, L.; Blacher, R. W.; Roth, R. A.; Rutter, W. J .
Nature (London) 1985, 317, 267-70.
(7) Darby, N.; Creighton, T. E. Biochemistry 1995, 34, 16770-80.
(8) Hawkins, H. C.; Freedman, R. B. Biochem. J . 1991, 275, 335-
9.
We sought to design small molecule thiols as redox
buffers for protein folding. The design was based on the
thiol pKa value and inherent nucleophilicity of protein
disulfide isomerase (PDI), nature’s solution to the protein-
folding problem. PDI has two similar active sites which
both contain two thiols in a CXXC motif.6 One thiol in
(9) Walker, K. W.; Lyles, M. M.; Gilbert, H. F. Biochemistry 1996,
35, 1972-80.
(10) LaMantia, M.; Lennarz, W. J . Cell (Cambridge, MA) 1993, 74,
899-908.
(11) Laboissiere, M. C. A.; Sturley, S. L.; Raines, R. T. J . Biol. Chem.
1995, 270, 28006-9.
(1) Woycechowsky, K. J .; Raines, R. T. Curr. Opin. Chem. Biol. 2000,
4, 533-39.
(2) De Bernardez Clark, E. Curr. Opin. Biotechnol. 1998, 9, 157-
(12) Wilson, J . M.; Bayer, R. J .; Hupe, D. J . J . Am. Chem. Soc. 1977,
99, 7922-6.
(13) Szajewski, R. P.; Whitesides, G. M. J . Am. Chem. Soc. 1980,
102, 2011-26.
63.
(3) Chaudhuri, J . B. Ann. N. Y. Acad. Sci. 1994, 721, 374-85.
(4) Rudolph, R.; Lilie, H. FASEB J . 1996, 10, 49-56.
(5) Wedemeyer, W. J .; Welker, E.; Narayan, M.; Scheraga, H. A.
Biochemistry 2000, 39, 4207-16.
(14) Whitesides, G. M.; Lilburn, J . E.; Szajewski, R. P. J . Org. Chem.
1977, 42, 332-8.
(15) De Maria, P.; Fini, A.; Hall, F. M. J . Chem. Soc., Perkin Trans.
2 1977, 149-51.
10.1021/jo015600a CCC: $20.00 © 2001 American Chemical Society
Published on Web 05/22/2001

Thiol-Disulfide Interchange Reactions
J . Org. Chem., Vol. 66, No. 12, 2001 4245
Sch em e 2
Sch em e 3
quently, the thiol pKa and reactivity of aromatic thiols
more closely match those of PDI. The correlation between
aromatic thiol pKa value and reactivity in reactions that
take place during protein folding is the focus of our
research. Previously, the only reaction of aromatic thiols
in aqueous solution investigated was that with Ellman’s
reagent (5,5′-dithiobis(2-nitrobenzoic acid)), an aromatic
disulfide.12,14 Results from these studies indicated en-
hanced reactivity of aromatic thiols relative to aliphatic
thiols which was ascribed to either a “hydrophobic
interaction between aryl groups on the attacking and
central thiols” or to “hard-soft acid-base behavior”.12 If
hydrophobic interaction between aryl groups is required
for enhanced reactivity then aromatic thiols will not
exhibit similar enhanced reactivity during protein folding
reactions. In addition, aryl disulfides are conceivably poor
models for a protein disulfide. Preliminary results from
another laboratory have reported no clear correlation
between reaction rates of thiols in general with Ellman’s
reagent and effectiveness as reducing agents for pro-
teins.14 Thus, it is unclear if results obtained with
Ellman’s reagent will be similar to results obtained with
protein disulfides or even other small molecule disulfides
that do not contain two aromatic groups. Therefore,
initial investigations featured the reactions of small
molecule aromatic and aliphatic thiols with small mol-
ecule disulfides and protein disulfides to determine if
aromatic thiols have enhanced reactivity with a broad
range of disulfides. In addition, to allow the rationale
design of aromatic thiols we investigated the relationship
between thiol pKa and reactivity with small molecule
disulfides and protein disulfides. The two-step reduction
of protein disulfides to dithiols is an important component
of the folding process as it removes intermediates with
mismatched disulfide bonds.5
Sch em e 4
Ta ble 1. p Ka Va lu es of P a r a -Su bstitu ted Ar om a tic
Th iols
method of determining
thiol pKa values
substitutent
CO2H
CH2CO2H
CN
COCH3
CH2OH
N(CH2CH2OH)2
UV-vis
titration
lit.24,25
5.80
1
2
3
4
5
6
7
8
9
6.0
6.6
4.9
5.0
6.4
5.33
7.0
6.9
OH
NO2
H
4.5
6.6
4.71, 4.50
6.61, 6.43
lithium aluminum hydride. Compound 6 was obtained
via a thiocyanate intermediate, which was subsequently
reduced to the desired aromatic thiol (Scheme 4).18,19
p Ka Deter m in a tion s. The pKa values of the aromatic
thiols were determined either by measuring the change
in the UV-vis spectrum as a function of pH20,21 or by
acid-base titration.22,23 Although the UV-vis method
was preferred, it was not applicable to all compounds,
as it required the spectra of the thiol and thiolate to be
readily distinguishable. The error in the pKa values at
95% confidence was less than (0.2 for all compounds
measured by the UV-vis method (Table 1). For com-
pounds 6 and 7, thiol and thiolate spectra were not
sufficiently distinguishable requiring recourse to acid-
base titration (Table 1).
A method to predict the pKa values of aromatic thiols
was sought, as this would allow the rational design of
thiols. A Hammett plot comparing σp- of the substituent
with the observed pKa values provided a slope of -1.62,
F ) -1.6 ( 0.1 (Figure 1).26 A previous Hammett plot
yielded F ) -1.8 ( 0.1, but it was based upon nonstand-
We report the synthesis of numerous water-soluble
aromatic thiols and their impact upon reactions that take
place during protein folding. The pKa values of several
aromatic thiols were determined for the first time, and
a Hammett plot was constructed to create a predictive
model for aromatic thiol pKa values. The rate constants
of aromatic thiols with small molecule nonbis(aromatic)-
disulfides were determined and correlated with thiol pKa
values. In addition, the relative rates of reducing insulin
with aromatic or aliphatic thiols or PDI were determined.
The effect of thiol pKa on reactivity and the relative
reaction rates of PDI and small molecule thiols are
discussed.
(17) Kaji, A.; Araki, Y.; Miyazaki, K. Bull. Chem. Soc. J pn. 1971,
44, 1393-9.
Resu lts a n d Discu ssion
(18) Brewster, R. Q.; Schroeder, W. Organic Syntheses; Wiley: New
York, 1943; Collect. Vol. II, pp 574-6.
Syn th esis. A series of para-substituted aromatic thiols
was synthesized with electron-donating or electron-
withdrawing substituents (1-6, Scheme 2). Thiols 1-4
were synthesized via a Newman-Kwart reaction (Scheme
3).16,17 Compound 5 was prepared by reduction of 1 with
(19) O’Connor, G. L.; Nace, H. R. J . Org. Chem. 1953, 18, 2118-22.
(20) Lees, W. J .; Singh, R.; Whitesides, G. M. J . Org. Chem. 1991,
56, 7328-31.
(21) Benesch, R. E.; Benesch, R. J . Am. Chem. Soc. 1955, 77, 5877-
81.
(22) Houk, J .; Singh, R.; Whitesides, G. M. Methods Enzymol. 1987,
143, 129-40.
(23) Whitesides, G. M.; Houk, J .; Patterson, M. A. K. J . Org. Chem.
1983, 48, 112-15.
(16) Newman, M. S.; Karnes, H. A. J . Org. Chem. 1966, 31, 3980-
4.

4246 J . Org. Chem., Vol. 66, No. 12, 2001
DeCollo and Lees
F igu r e 1. Hammett plot of para-substituted benzenethiol pKa
-
values versus σp-. The σp value for a dimethylamino sub-
F igu r e 2. Determination of the second-order rate constant,
kobsd, for the reduction of 2-PDE by compound 6. The slope
corresponds to kobsd. The values of [A]0, the initial concentration
of 2-PDE, and [A] are calculated from the change in absorbance
at 343 nm. The 2-pyridylthiolate released in the reaction
absorbs at 343 nm (ꢀ ) 7060 mol-1 L-1 cm-1). The standard
deviation of kobsd is (10%.
stituent was used for the diethanolamino substituent. No
suitable σp- value could be found for the -CH2COO- substitu-
ent.
Sch em e 5. Red u ction of 2-P DE by Th iola tes
(RS-) a t p H 7 (F igu r e 2)
Ta ble 2. Ra te Con sta n ts for th e Red u ction of 2-P DE by
Th iols a t p H 7 (Sch em e 5)
obsd
thiol
pKa of thiol k1
(M-1 min-1
)
k1 (M-1 min-1
)
6
7
2
5
1
7.0
6.9
6.6
6.4
6.0
5.7
7.3
8.7
14 × 104
13 × 104
7.7 × 104
5.6 × 104
3.2 × 104
2.9 × 104
1.3 × 104
0.24 × 104
29 × 104
23 × 104
11 × 104
7.0 × 104
3.5 × 104
3.0 × 104
3.8 × 104
12 × 104
ard σp values instead of σp-, thus limiting its predictive
ability for our purposes.24 Now if the σp value of a
-
substituent is known, the pKa value of the corresponding
para-substituted aromatic thiol is predictable.24,25
Kin etics of th e Red u ction of Sm a ll Molecu le
Disu lfid es by Th iols. The only thiol-disulfide inter-
change reaction of aromatic thiols investigated previously
was that with Ellman’s reagent.12,14 For comparison, we
initially sought to perform similar experiments with our
aromatic thiols, some of which were expected to be
considerably more reactive than those previously studied
at pH 7. Preliminary results indicated that the assay was
not practical without the use of a stop-flow apparatus
due to exceptionally fast reactions. Interpretation of the
data is also complicated because Ellman’s reagent and
the mixed disulfide between Ellman’s reagent and an
aromatic thiol have similar reactivities.12,14 However, the
results did confirm the enhanced reactivity of aromatic
thiols relative to aliphatic thiols of similar pKa value, at
least with Ellman’s reagent. Therefore, to obtain quan-
titative results we could either advance to a different
disulfide instead of Ellman’s reagent or lower the pH
significantly from physiological pH, which would slow the
reaction. Since we were ultimately interested in reactions
with proteins at close to physiological pH we chose
to advance to a less reactive disulfide, 2-pyridyldithio-
ethanol (2-PDE) (Scheme 5).
2-HOC6H4SH
F3CCH2SH
glutathione
thiol, thereby simplifying data interpretation relative to
Ellman’s reagent. Unlike Ellman’s reagent, which has a
double negative charge at pH 7, 2-PDE is neutral at pH
7, thus minimizing charge effects between thiol and
disulfide. In addition, 2-PDE will not allow π-stacking
between the nucleophilic aromatic thiolate and the group
attached to the most electrophilic sulfur of the disulfide.
The most electrophilic sulfur is attached to an aliphatic
group for 2-PDE and an aromatic group for Ellman’s
reagent. The extinction coefficient at λmax of the thiolate
-1
of 2-PDE (ꢀ343 ) 7060 mol
L-1 cm-1) is half that of
Ellman’s reagent (ꢀ412 ) 13 700 mol-1 L-1 cm-1), but this
is more than compensated for by the enhanced accuracy
enabled by the slower reaction rate of 2-PDE.27
obsd
The 2-PDE assay provides k1
for the reaction
between 2-PDE and a thiol at pH 7 (Table 2, Figure 2).
If the pH of the reaction is changed then k1
obsd
may also
change. A rate constant independent of pH, k1, can be
calculated; it corresponds to the reaction of pure thiolate
with disulfide and thus reflects the intrinsic reactivity
of the thiolate, independent of pH effects. Under the
reaction conditions, the thiol form of the compound is a
very poor nucleophile and contributes negligibly to the
reaction. To calculate k1, the proportion of the compound
in the thiolate form is determined from the pH of the
The disulfide 2-PDE provides an almost ideal substrate
for determining reaction rate constants of aromatic thiols
with disulfides. We have found the reaction rate con-
stants for 1 (Scheme 2), a slowly reacting aromatic thiol,
and trifluoroethanethiol with 2-PDE at pH 7 are approx-
imately 4% and 2% of their reaction rate constants with
Ellman’s reagent, respectively.12,14 The mixed disulfide
between aromatic thiol and mercaptoethanol (10, Scheme
5) is expected to react with aromatic thiol at a much
slower rate than the reaction of 2-PDE with aromatic
solution and the pKa of the thiol; (1/(1 + 10pK -pH). The
a
measured rate constant, k1obsd, is divided by the propor-
tion of the compound in the thiolate form to obtain the
reaction rate constant of the thiolate k1 ) k1obsd(1 +
10pK -pH). The reaction rate constants of aromatic thiols
a
with 2-PDE increases with increasing thiol pKa values,
as is the case with Ellman’s reagent. A plot of log k1
versus para-substituted aromatic thiol pKa is linear with
(24) De Maria, P.; Fini, A.; Hall, F. M. J . Chem. Soc., Perkin Trans.
2 1973, 1969-71.
(25) J encks, W. P.; Salvesen, K. J . Am. Chem. Soc. 1971, 93, 4433-
6.
(26) Exner, O. Correlation Analysis of Chemical Data; Plenum
Press: New York, 1988.
(27) Grimshaw, C. E.; Whistler, R. L.; Cleland, W. W. J . Am. Chem.
Soc. 1979, 101, 1521-32.

Thiol-Disulfide Interchange Reactions
J . Org. Chem., Vol. 66, No. 12, 2001 4247
a positive slope, â ) 0.9 ( 0.1 (r2 ) 0.99). The slope is
greater than the â values of 0.2-0.7 previously reported
for thiol-disulfide interchange reactions.12,14 The greater
slope may be the result of aromatic thiolates reacting
with a nonbis(aromatic) disulfide, a substrate type that
has not been investigated previously.
The pKa value of the thiol with the greatest k1
obsd
at a
given pH in a thiol-disulfide interchange reaction can
be calculated from â and the pH of the solution.14 Two
competing factors affect the reaction rates of thiols. The
first one is the portion of thiol in the reactive thiolate
form, which is determined by the thiol pKa value and the
pH of the solution. The greater the proportion of com-
pound in the thiolate form the lower the pKa of the thiol.
The second one is the nucleophilicity of the thiolate,
which for a structurally similar set of compounds in-
creases with the pKa of the conjugate acid (thiol) and can
be determined from the thiol pKa value and â, the
proportionality constant. The more reactive the thiolate
the greater the pKa of the conjugate acid (thiol). The
optimum pKa value of the thiol can be determined from
F igu r e 3. Following the reduction of insulin by compound 6
and DTT (squares), and an analogue of compound 6 and DTT
(circles) by observing the light scattered at 650 nm. The rate
of reduction is determined from the slope of the linear
portion.31 The curve observed for the reduction of insulin by
an analogue of compound 6 and DTT was almost identical to
that of DTT only (not shown).
Sch em e 6. Red u ction of Na tive In su lin by
Ar om a tic Th iol + DTT a t p H 6.5
14
a
the following equation: (1 - â)/â ) 10pH-pK
.
With a â
value of 0.9 (2-PDE), the optimum thiol pKa value is
approximately 1 unit greater than the pH of solution.
With a â value of 0.5 (reported for Ellman’s reagent),12
it is equivalent to the pH of solution. Thus, depending
on whether the aromatic thiol is reacting with 2-PDE or
Ellman’s reagent its optimum pKa value is quite different.
Kin et ics of R ed u ct ion of P r ot ein Disu lfid es b y
Th iols. The first assay attempted was the reduction of
the protein papain-S-S-Me to papain-SH, the enzymati-
cally active form.28-30 Preliminary results using a coupled
assay demonstrated that the reaction rates of aromatic
thiols with papain-S-S-Me were considerably faster than
aliphatic thiols with papain-S-S-Me at pH 7. Unfortu-
nately, the reaction with aromatic thiols was so rapid at
pH 7 that no quantitative data could be obtained.
Therefore, from the papain assay we can conclude that
aromatic thiols react faster with protein disulfides than
aliphatic thiols at pH 7.
Increases in the rate of reduction of the protein insulin
in the presence of dithiothreitol (DTT) by aromatic thiols
provided quantitative data.31 Native insulin has three
disulfide bonds, two of which connect the R chain to the
â chain. Cleavage of these disulfide bonds leads to the
formation of two separate peptides (Scheme 6). When
insulin is reduced at pH 6.5, the liberated â-chain slowly
reaches a critical concentration at which it precipitates
out of solution, forming a cloudy white suspension which
scatters 650 nm light. The rate of reduction of insulin
was followed by measuring the rate of increase in light
scattering with time (Figure 3).
The relative rate was measured to account for minor
variations between batches of native insulin. The control
reaction demonstrated that other functional groups in the
molecule did not change the rate of reduction of insulin
by DTT. DTT, a strongly reducing dithiol, was added to
all reactions to ensure the equilibrium was shifted toward
reduced insulin. DTT will reduced the aromatic disulfides
formed along with reduced insulin back to the aromatic
thiol, thus shifting the equilibrium toward reduced
insulin. In all three reactions, the light scattered at 650
nm initially changed very little, but as the precipitate
formed, the scattered light increased linearly with time
(Figure 3). Holmgren has shown the rate of reduction of
insulin is proportional to the slope of the linear portion
and to the lag time before precipitation occurs.31 We
found the relative ratios of the slopes to be more
reproducible than the relative ratios of the lag times.
These results were qualitatively confirmed by tricine-
sodium dodecyl sulfate-polyacrylamide gel electrophoresis
(SDS-PAGE).32
The addition of aromatic thiols to a mixture of insulin
and DTT increased the rate of reduction of insulin but
the addition of aliphatic thiols did not, within experi-
mental error. The rate of reduction of insulin increased
with increasing pKa of the aromatic thiol (Table 3). Given
the â value, 0.9, for 2-PDE the optimum pKa value of the
aromatic thiol is predicted to be approximately 1 unit
greater than the pH of the solution, the data are
consistent with this prediction.14 The trend in relative
reaction rates also correlates with the reactivity of the
The rate of reduction of native insulin by a mixture of
the thiol of interest (1 equiv of thiol groups relative to
thiols in reduced insulin) and DTT (2 equiv) was com-
pared to the rate of reduction of native insulin by DTT
(2 equiv) alone. A control reaction containing native
insulin, DTT (2 equiv), and a compound structurally
similar to the thiol of interest but with the thiol group
replaced by either a hydroxyl group or hydrogen was also
prepared. All three reactions were run simultaneously.
(28) Singh, R.; Whitesides, G. M. J . Org. Chem. 1991, 56, 2332-7.
(29) Singh, R.; Blattler, W. A.; Collinson, A. R. Anal. Biochem. 1993,
213, 49-56.
(30) Mole, J . E.; Horton, H. R. Biochemistry 1973, 12, 816-22.
(31) Holmgren, A. J . Biol. Chem. 1979, 254, 9627-32.
(32) Schaegger, H.; Von J agow, G. Anal. Biochem. 1987, 166, 368-
79.

4248 J . Org. Chem., Vol. 66, No. 12, 2001
DeCollo and Lees
Ta ble 3. Rela tive Ra tes of In su lin Red u ction by Th iol +
Sch em e 7
DTT (0.8 m M) a t p H 6.5
rate relative to
thiol (mass, concentration)
pKa of thiol
DTT only
1 (123 µg, 0.7 mM)
6.0
6.4
6.6
7.0
7.0
7.3
8.3
8.7
6.7
6.7
1.6
1.9
2.0
3.3
8.1
1.1
0.9
1.0
3.2
1.8
5 (102 µg, 0.7 mM)
equilibrium is shifted away from reduced insulin and
toward the aromatic thiolate (Scheme 7).
2 (134 µg, 0.7 mM)
6 (190 µg, 0.7 mM)
6 (380 µg, 1.5 mM)
In conclusion, we have demonstrated a strong correla-
tion between the σp- value of a substituent, the pKa value
of the corresponding para-substituted aromatic thiol, the
reactivity of the corresponding para-substituted aromatic
thiol with small molecule disulfides, and the ability of
the corresponding para-substituted aromatic thiol to
increase the rate of protein unfolding. The proportionality
constant linking aromatic thiol pKa to reactivity (log k1)
was considerably greater for 2-PDE (â ) 0.9) and protein
disulfides than that previously reported for Ellman’s
reagent (â ) 0.5). Indicating, in contrast to the results
with Ellman’s reagent, that the most reactive aromatic
thiol for protein unfolding will have a pKa value greater
than the pH of the solution.14 The enhanced reactivity,
at approximately neutral pH, of aromatic thiols with
disulfides relative to aliphatic thiols was shown to occur
not only for Ellman’s reagent but also for nonbis-
(aromatic) small molecule disulfides and, most impor-
tantly, protein disulfides, thus clearly demonstrating that
the enhanced reactivity of aromatic thiols is not due to
hydrophobic interactions between aryl groups on the
attacking and central thiol. Furthermore, equivalent
rates of protein unfolding were observed when PDI or
10 times the weight of aromatic thiol 6 as PDI was added.
PDI, the in vivo protein folding catalyst, is rarely used
for protein unfolding or folding, as it is expensive and
can be difficult to separate from other proteins. Since the
reduction of protein disulfides occurs during the folding
of many proteins,5 aromatic thiols with high pKa values
such as 6 are expected to increase the rate of protein
folding as well.
CF3CH2SH (93 µg, 0.7 mM)
cysteine (98 µg, 0.7 mM)
glutathione (245 µg, 0.7 mM)
PDI (20 µg, 0.0003 mM)
PDI (10 µg, 0.0001 mM)
Ta ble 4. Rela tive Rea ction Ra tes of Th iols
relative rates
2-PDE at pH 6.5
thiol
pKa
k1/(1 + 10pKa-6.5
)
PDE
insulina
6
2
5
1
7.0
6.6
6.4
6.0
7.3
8.3
7.0 × 104
4.9 × 104
3.9 × 104
2.6 × 104
0.52 × 104
0.075 × 104
3.3b
2.3
1.9
1.3
0.25
0.04
3.3
2.0
1.9
1.6
1.1
0.9
CF3CH2SH
cysteine
a
The reaction rates of a mixture of thiol and DTT with insulin
are relative to DTT only; the minimum value is approximately 1.
Set to 3.3 to allow a direct comparison with the reaction rates of
b
insulin.
Ta ble 5. Red u ction of In su lin by Ar om a tic Th iols in th e
Absen ce or P r esen ce of DTT
rate relative to
thiol
pKa of thiol
DTT only
5 (0.7 mM)
5 (0.7 mM) + DTT (0.8 mM)
6 (0.7 mM)
6.4
6.4
7.0
7.0
0.6
1.8
3.4
3.4
6 (0.7 mM) + DTT (0.8 mM)
aromatic thiols with 2-PDE, k1, after an adjustment to
account for the pH of the reaction, pH ) 6.5 (Table 4).
To achieve the same rate of reduction of insulin in the
presence of DTT, 10 times and 5 times the weight of
aromatic thiol 6 is needed to match PDI (57 kDa) and
Thioredoxin (12 kDa),31 respectively. On a mole basis,
these numbers increase to 1000 and 200, per active site,
respectively. However, the expense of using thioredoxin
or PDI is significant and sometimes prohibitive. Also,
thioredoxin and PDI can be difficult to separate from the
protein of interest, while small molecules such as 6 are
not. Therefore, on a weight basis, aromatic thiol 6
compares favorably with PDI and thioredoxin, nature’s
protein refolding/unfolding catalysts.To allow a more
direct comparison with the 2-PDE results, the aromatic
thiols were tested for their ability to reduce insulin in
the absence of DTT. Again, three reactions were run
simultaneously. The first reaction measured the rate of
reduction of insulin by the aromatic thiol (1 equiv) alone,
the second measured the rate of reduction of insulin by
DTT (2 equiv) alone, and the third reaction measured
the rate of reduction of insulin by a mixture of aromatic
thiol and DTT (Table 5). Although thiols 1, 5, and 6 were
tested, only 5 and 6 yielded measurable rates. The
reaction rate for compound 5 with insulin was 20% that
of compound 6; the same comparison for 2-PDE was 60%.
The discrepancy is likely the result of a shift in the
equilibrium. As the pKa of the aromatic thiol (ArSH)
decreases, the concentration of aromatic thiolate (ArS-)
increases at the expense of aromatic thiol, and the
Exp er im en ta l Section
Gen er a l Rem a r k s. NMR spectra were recorded at 300
MHz (1H) and at 75 MHz (13C) on a Bruker spectrophotometer.
Chemical shifts were indirectly referenced to TMS via solvent
signals (CDCl3, 7.26 ppm for 1H and 77.00 ppm for 13C; CD3-
OD, 49.00 ppm for 13C). Thin-layer chromatography (TLC) was
conducted on Aldrich general-purpose silica gel on polyester
plates with UV indicator. Silica gel chromatography was
performed with E. M. Science silica gel (230-400 mesh). Dry
THF was obtained by distillation from sodium metal in the
presence of benzophenone. Routine drying of organic solutions
was carried out with anhydrous magnesium sulfate. All
reactions performed under inert atmospheric conditions were
carried out under Ar. All reagents purchased, including
compounds 7-9, were used without purification unless other-
wise noted. UV-vis spectra were recorded on a Cary 1 UV-
vis spectrophotometer. All proteins were purchased from
Sigma. E & R Microanalytical Laboratory Inc. performed
elemental analysis.
4-[N,N-Bis(2-h yd r oxyet h yl)a m in o]b en zen et h iocya n -
a te.18 A 250-mL flask equipped with a stirring bar was charged
with N-phenyldiethanolamine (10.0 g, 0.055 mol), ammonium
thiocyanate (8.4 g, 0.110 mol), and 35 mL of glacial acetic acid.
After the mixture was cooled to 10 °C, a solution of bromine
(8.8 g, 0.055 mol) in glacial acetic acid (20 mL) was added
dropwise over 30 min while keeping the temperature below
20 °C. The mixture was then stirred for an additional 10 min
at room temperature and poured onto 600 mL of H2O. After
filtration, the filtrate was extracted with ethyl acetate (3 ×

Thiol-Disulfide Interchange Reactions
J . Org. Chem., Vol. 66, No. 12, 2001 4249
100 mL), washed with saturated NaHCO3 (100 mL), dried over
MgSO4, and concentrated in vacuo to yield 11.20 g (86%) of
4-[N,N-bis(2-hydroxyethyl)amino]benzenethiocyanate: 1H NMR
(300 MHz, CDCl3) δ 7.39 (d, J ) 9.0 Hz, 2 H, Ar H), 6.66 (d, J
) 9.0 Hz, 2 H, Ar H), 3.78 (t, J ) 4.9 Hz, 4 H, NCH2CH2OH),
3.55 (t, J ) 4.9 Hz, 4 H, NCH2CH2OH); 13C NMR (75 MHz,
CDCl3) δ 149.6, 134.6, 113.6, 112.6, 107.2, 60.1, 54.9.
(2-PDE) (approximately 18 µM) was prepared in a fashion
similar to the thiol solution. The concentration of the 2-PDE
was determined by mixing 0.500 mL of 2-PDE solution with
concentrated 4-mercaptobenzenemethanol (0.100 mL of a 1.25
mM solution) and 0.400 mL of pH 7.0 buffer and measuring
the absorbance at 343 nm (ꢀ343 ) 7060 mol-1 L-1 cm-1 for the
thiolate of 2-mercaptopyridine). This absorbance was compared
to the absorbance of a standard containing concentrated
4-mercaptobenzenemethanol (0.100 mL of a 1.25 mM solution)
and 0.900 mL of pH 7.0 buffer. In a 1 mL cuvette, appropriate
amounts of the thiol and 2-PDE solutions were mixed together
to form an equimolar solution and the rate of the 2-thiopyri-
dine anion formation was measured over 20 min by observing
the increase in absorbance at 343 nm. The initial concentration
of PDE, [A]0, will approximately equal the concentration of
2-mercaptopyridine at the end of the reaction, [A]0 ) (Absfinal
- Absinitial)/7060, where Absfinal is the absorbance after the
reaction has gone to completion and Absinitial is the absorbance
prior to the beginning of the reaction. Using linear regression,
2,2′-[(4-Mer ca p top h en yl)im in o]biseth a n ol (6).19 A solu-
tion of 4-[N,N-bis(2-hydroxyethyl)amino]benzenethiocyanate
(5.0 g, 26.4 mmol) in 30 mL of THF was added dropwise over
1 h to a 250-mL flask containing lithium aluminum hydride
(2.5 g, 65.9 mmol), 20 mL of THF, and a stirring bar. After 48
h, the reaction was quenched by the dropwise addition of 1
mL of H2O. The mixture was then acidified to pH 1 with 1 N
HCl and washed with diethyl ether (50 mL). The aqueous layer
was adjusted to pH 6 and extracted with ethyl acetate (3 ×
50 mL). The combined organic layers were dried over MgSO4
and concentrated in vacuo. Recrystallization from benzene/
heptane (1:1) yielded 1.20 g (27%) of 4-[N,N-bis(2-hydroxy-
ethyl)amino]benzenethiol as a light yellow powder: mp 134
°C (lit. 136 °C);33 1H NMR (300 MHz, CDCl3) δ 7.21 (d, J )
8.8 Hz, 2 H, Ar H), 6.57 (d, J ) 8.9 Hz, 2 H, Ar H), 3.78 (t, J
) 4.9 Hz, 4 H, N-CH2-CH2-OH), 3.51 (t, J ) 4.9 Hz, 4 H,
N-CH2-CH2-OH), 3.30 (s, 1 H, SH); 13C NMR (75 MHz,
CDCl3) δ 146.9, 133.1, 114.7, 113.4, 60.6, 55.1. Anal. Calcd for
the slope of a plot of ((Absobsd - Absinitial)/7060)/(((Absfinal
Absinitial)/7060)((Absfinal - Absobsd)/7060)) versus time was
determined. The slope corresponds to kobsd
-
.
Kin etics of Red u ction of In su lin by Ar om a tic Th iols.31
All measurements were performed at room temperature and
in duplicate. Insulin (1.0 mL of a 10 mg/mL stock solution)31
was diluted with 8.0 mL of aqueous buffer (pH 6.5, 0.10 M in
potassium phosphate, 2 mM EDTA). This cloudy solution was
made clear by adjusting to pH 3.0 with the addition of 1.0 M
HCl and rapidly titrating the solution back to pH 8.0 with 1.0
M NaOH. The final volume of the dilute insulin (1.0 mg/mL,
0.167 mM) solution was adjusted to 10.0 mL with water. A
representative kinetic assay is described; others followed a
similar procedure. Three cuvettes were prepared containing
1.00 mL of dilute insulin each. Dithiothreitol (40 µL of a 25.0
mM solution in buffer (pH 6.5, 0.10 M in potassium phosphate,
2 mM EDTA)) was then added to each cuvette just prior to
the addition of the last component. At 1 min intervals 6 (160
µL of a 5.0 mM solution in buffer) was added to cuvette 1, a
structurally similar nonthiol containing compound, N-phenyl-
diethanolamine (160 µL of a 5.0 mM solution in buffer) was
added to cuvette 2, and 160 µL of buffer was added to cuvette
3. The absorbance of each cuvette at 650 nm was then
measured every 3 min for 3 h. The slopes were determined
from a plot of absorbance versus time. The ratio of the slopes
of the data obtained from cuvettes 1 and 3 provided the relative
rate of reduction of 7 compared to dithiothreitol (DTT).
C
10H15NO2S: C, 56.31; H, 7.09; N, 6.57; S, 15.03. Found: C,
56.59; H, 6.87; N, 6.51; S, 14.74.
Deter m in a tion of p Ka Va lu es (UV-Vis Meth od ).20,21
Initially, 24 buffers of varying pH were prepared (50 mM):
glycine, pH 2.5, 3.0, 3.3; 2,2-dimethylsuccinate, pH 3.7, 4.0,
4.3, 4.7, 5.0, 5.3, 5.7, 6.0, 6.3, 6.7; Tris, pH 7.0, 7.3, 7.7, 8.0,
8.3, 8.7; glycine, pH 9.0, 9.5, 10.0, 10.5; H3PO4, pH 11.0. A
100 mM ethanolic solution of each aromatic thiol (100 µL) was
diluted 90-fold with selected buffers. The buffers were chosen
so that the pH of the buffer would be within 1.5 units of the
pKa of the thiol tested. The absorbances of these diluted
solutions were measured at the λmax of the corresponding
thiolate. This absorbance data was then plotted against pH.
The plot was compared with plots derived from theory. The
pKa value was determined from the best-fit curve. Additional
titratable functional groups had no observable effects upon the
results with the exception of the p-hydroxy group.
Titr a tion Meth od .22,23 Deoxygenated distilled water (10
mL) was allowed to equilibrate under Ar in a 50 mL three-
neck round-bottom flask equipped with a stirring bar. A pH
meter probe was inserted in the neck of the flask. The
appropriate amount of the thiol was added to the flask to make
a 0.010 M thiol solution. The thiol solution was titrated against
0.151 M carbonate-free KOH by adding 350 µL of KOH,
waiting 1 min., recording the pH, then repeating the proce-
dure. After 20 points (7 mL total KOH added) were measured,
a titration curve was created. The plot was compared with
plots derived from theory. The pKa value was determined from
the best-fit curve. All titrations were performed in duplicate.
Kin etics of th e Red u ction of 2-P yr id yld ith ioeth a n ol
by Th iols.27 A thiol solution (approximately 18 µM) was
prepared in deoxygenated buffer (pH 7.0, 0.100 M sodium
phosphate buffer, 2.0 mM EDTA). The concentrations of the
Ack n ow led gm en t . Financial support from the
National Science Foundation (CHE9975076), the
Petroleum Research Fund (34902-G4), the Camille
and Henry Dreyfus Foundation (New Faculty Award),
and Syracuse University is gratefully acknowledged. We
thank Dr. Rajeeva Singh of Immunogen and Dr. Roger
Hahn of Syracuse University for helpful comments. We
also thank Dr. Rajeeva Singh for providing 2-PDE.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures and spectral data for the synthesis of compounds 1-5.
Experimental procedure and the results for the SDS-PAGE
experiment. This material is available free of charge via the
Internet at http://pubs.acs.org.
thiol solution was determined using Ellman’s reagent (ꢀ412
)
13 700 mol-1 L-1 cm-1). A solution of 2-pyridyldithioethanol
(33) Valu, K. K.; Gourdie, T. A.; Boritzki, T. J .; Gravatt, G. L.;
Baguley, B. C.; Wilson, W. R.; Wakelin, L. P. G.; Woodgate, P. D.;
Denny, W. A. J . Med. Chem. 1990, 33, 3014-19.
J O015600A