A novel method for the determination of isokinetic ratios and its application in the synthesis of two new positional scanning libraries

Article abstract of DOI:10.1021/co300060s

A novel method for the direct evaluation of the equimolarity of the compounds contained in a mixture is presented. We applied the method toward calculating isokinetic ratios for the reaction between the amine termini of a resin bound peptide fragment and

Full text of DOI:10.1021/co300060s

Research Article
pubs.acs.org/acscombsci
A Novel Method for the Determination of Isokinetic Ratios and Its
Application in the Synthesis of Two New Positional Scanning
Libraries
Marc A. Giulianotti,† Ginamarie Debevec,† Radleigh G. Santos,† Laura E. Maida,† Wenteng Chen,†,‡
Lili Ou,†,‡ Yongping Yu,†,‡ Colette T. Dooley,† and Richard A. Houghten*,†
†Torrey Pines Institute for Molecular Studies, 11350 SW Village Parkway, Port St. Lucie, Florida 34987, United States
‡College of Pharmaceutical Science, Zhejiang University, Hangzhou, P.R. China 310058
S
* Supporting Information
ABSTRACT: A novel method for the direct evaluation of the
equimolarity of the compounds contained in a mixture is
presented. We applied the method toward calculating
isokinetic ratios for the reaction between the amine termini
of a resin bound peptide fragment and a sulfonyl chloride to
produce equal molar mixtures of sulfonamides. The results of
this study and the application of the method to the synthesis of
two new positional scanning synthetic combinatorial libraries
(PS-SCL) are discussed.
KEYWORDS: positional scanning library, isokinetic ratios, combinatorial libraries, harmonic mean, piperazine, sulfonamide, guanidine
amino acids, carboxylic acids, and aldehydes.9−13 These
methods include the use of amino acid analysis, as well as
INTRODUCTION
■
Combinatorial chemistry is a valuable tool in the rapid
binary and iterative HPLC analysis, and MALDI-TOF. In this
production of compound libraries for biological evaluation.1,2
publication, we introduce a novel method for the comparison of
Specifically, positional scanning synthetic combinatorial
LCMS UV spectra, which allows direct relative evaluation of
libraries (PS-SCL) enable the systematic screening of
the equimolarity of the compounds contained in that mixture;
thousands to trillions of compounds through the use of only
further, this method is applicable in cases more complex than
hundreds of samples.2,3 This exponential reduction in sample
would be feasible by simply using a reference compound. We
size enables many endeavors, such as the evaluation of nearly 4
applied it to the reaction between the amine termini of a resin
million small molecules through a direct in vivo screening to
bound peptide fragment and a sulfonyl chloride to produce a
identify two novel potent, low-liability antinociceptive com-
pounds4 and the multiclonal evaluation of trillions of 10-mer
sulfonamide. The isokinetic ratio was determined and the effect
of the on-resin peptide fragment was evaluated. Two novel PS-
SCLs were synthesized based on the calculated ratios.
peptides to identify immunoprevalent antigens recognized by
human CD4+ T cells.5 None of these efforts could have been
Furthermore the two PS-SCLs were each screened in two
currently conducted without the use of PS-SCL.
opioid (mu and kappa) binding assays in part to determine the
A key component of PS-SCL involves the ability to produce
synthetic integrity.
near equimolar amounts of each individual compound in each
mixture sample. Two methods for accomplishing equimolar
The two novel PS-SCLs were designed around two separate
mixtures are the use of physical mixtures (“split and mix”)6−8 or
scaffolds (1) a piperazine linked to a sulfonamide (Scheme 1
chemical mixtures.9 While the use of physical mixtures has the
advantage of assuring near equimolar incorporation of each
individual compound in the mixture samples, it is limited in
that as the number of R groups increases around the scaffold
Structure 7) and (2) a cyclic guanidine linked to a sulfonamide
(Scheme 1 Structure 8). A number of approved pharmaceut-
icals contain piperazines, such as imatinib (cancer treatment),
quetiapine (antipsychotic), and eszopiclone (treatment of
insomnia), sulfonamides, such as sulfamethoxazole (antibiotic),
the corresponding number of split and mixes required increases
xipamide (hypertension and edema), and darunavir (HIV), and
exponentially, quickly rendering the technique less valuable.
The use of chemical mixtures where “isokinetic mixtures” are
guanidines, such as guanadrel (antihypertensive) and guanethi-
dine (antihypertensive). The synthesis (Scheme 1) utilizes our
applied to produce equimolar mixtures does not have the same
libraries from libraries approach,14 where multiple libraries are
limitations of the split and mix method; however, to utilize
chemical mixtures in the production of PS-SCL, one needs to
determine the isokinetic ratios for the reagents used.
Several methods have been reported for determining the
isokinetic ratio for a variety of different reagent classes, such as
Received: June 1, 2012
Revised: August 21, 2012
Published: August 23, 2012
© 2012 American Chemical Society
503
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
Scheme 1. (a) Boc-AA/DIC/HOBt (6ex) in DMF, (b) (1) 55%TFA/DCM, (2) 5%DIEA/DCM, (c) Sulfonyl Chloride/DIEA
(8ex) in DMF, (d) HF 0 °C 1.5 h, (e) (1) Borane−THF 60 °C 96 h, (2) Piperidine 60 °C 24 h, (f) Bromoacetic Acid/DIC/
DIEA (5ex/5ex/2.5ex), (g) HF 0 °C 7 h, (h) Cyanogen Bromide/DMF (5ex); (A) mBHA Resin-(L-Phenylalanine)-(L-
Phenylalanine), (B) mBHA Resin-(Equal Molar Mixture of Amino Acids; Ls, Ds, and Unusuals)-(Equal Molar Mixture of Amino
Acids; Ls, Ds, and Unusuals), (C) Mixture Sample Defined with L-Phenylalanine (R1), L-Phenylalanine (R2), and a Mixture of
Different Sulfonly Chlorides at R3, (D) Compound Derived from Using L-Phenylalanine (R1), L-Phenylalanine (R2), and Any
One Sulfonyl Chloride (R3), (D-45) Compound Derived from Using L-Phenylalanine (R1), L-Phenylalanine (R2), and 4-tert-
Butylbenzene-1-sulfonyl Chloride (R3)
synthesized from the same core in this case the peptide linked
sulfonamide (Scheme 1 Structure 3). In fact, we could have
made separate libraries or libraries from libraries of both 3 and
4. Additionally we could expand the diversity of our collection
by utilizing different cyclization conditions on 5 to produce
additional “libraries from libraries”, such as cyclic urea linked
sulfonamides and cyclic thiourea linked sulfonamides utilizing
were synthesized and analyzed by LCMS (see Table S1 in
Supporting Information). Of the 104 compounds, 40 tested the
robustness of the R3 position, 20 for 7, and 20 for 8 (see Table
1). Of these 40, 30 (15 for 7 and 15 for 8) were shown to have
final UV purities greater than >75%. Therefore the 15 sulfonyl
chlorides that provided these 30 compounds (with >75%
purity) were selected to determine isokinetic ratios. It may be
noted that compounds in the R3 position that contained
fluorine substitution yielded low product purity. This may be
due to fluorine reacting with the piperazine during the
reduction step (Scheme 1 step e2); however, this side reaction
was not explored, and these sulfonyl chlorides were eliminated
from further study.
Having tested the robustness of the reaction conditions to
different reagents, it is then necessary to determine a correction
factor for each compound to account for differences in UV
absorbance. Applying the tea bag method 16, 15 individual
compounds where Boc-L-Phenylalanine was utilized in the R1
and R2 position were synthesized according to scheme 1. The
15 sulfonyl chlorides of interest were coupled to this sequence
separately (Scheme 1 compound 3) providing 15 separate
compounds that differ at only the R3 position (Scheme 1
carbonyldiimidazole and thiocarbonyldiimidazole to name just
15
a couple
.
Determination of Proper Reagent Ratio. We have
previously calculated, utilized, and reported the reagent ratios
for amino acids competitively coupled to solid supported free
amines 9. These ratios are utilized in the generation of Library 7
and 8. However, until now we had not determined the proper
reagent ratios for sulfonyl chlorides reacting with solid
supported free amines. To synthesize an equimolar mixture
of sulfonamides, ratios must be determined to adjust for the fact
that steric and electronic hindrance will vary among the
different sulfonyl chlorides used in the mixture.
Initially, to test the robustness of different amino acids and
sulfonyl chlorides to the reaction conditions described in
Scheme 1, 104 individual compounds (52 of 7 and 52 of 8)
504
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
obtained from using 4-tert-butylbenzene-1-sulfonyl chloride at
the R3 position (Scheme 1 compound D-45), with a sufficiently
distinct retention time was selected as the control to which all
other compounds were compared. Each compound was then
physically mixed with the control to create an equal molar
mixture. These samples were analyzed using HPLC-MS. The
integration parameters were set to allow the best integration
possible for the UV peaks 214 nm, and all data sets were
analyzed using these same integration parameters. Since the
concentration of each compound is known to be equal, any
differences in peak area can be attributed to differences in molar
absorbance. A correction factor is therefore calculated for each
compound as the ratio of the control compound’s peak area to
the peak area of that compound:
Table 1. 40 Different Individual Compounds (20 of Type 7
and 20 of Type 8) Were Made to Test the Robustness of the
a
R3 Position
Areacontrol
Arean
CF =
n
This definition insures that by multiplying the peak area of
the nth compound by its corresponding control factor, an area
equal to peak area of the control is obtained. This process was
performed in duplicate to confirm the correction factors and to
minimize experimental error. See Table 2.
The rate at which each sulfonyl chloride reacts will vary
based on steric and electronic hindrance. To synthesize an
equimolar mixture, a ratio of each component of the sulfonyl
chloride mixture must be calculated in order to compensate for
the reaction rate differences. Each test mixture C was
synthesized using a 1:1 mixture of the test sulfonyl chloride
and the control sulfonyl chloride (Scheme 1 step c), each at a 4-
fold excess over the resin (for a combined 8-fold excess total).
The resulting on-resin synthetic mixtures were then cleaved
from the solid support and analyzed by HPLC-MS using the
same method and integration parameters as were used to
calculate the correction factor. The UV 214 nm peak area under
the curve is used to calculate the reaction rate ratio:
Areacontrol
BRn =
CF Arean
n
The ratios obtained are thus the rate at which the nth sulfonyl
chloride reacts relative to the control compound, with the
differences in absorptivity accounted for via multiplication by
the control factor. See Table 2.
Differences in Reaction Rate Caused by Multiple
Simultaneous Reactions. The above calculations were done
as individual comparisons between a given sulfonyl chloride
and the control sulfonyl chloride. To determine if the presence
of additional (more than two at a time) sulfonyl chlorides
simultaneously would change the reaction rate of each sulfonyl
chloride, these ratios were then tested in the presence of
multiple sulfonyl chlorides including the control. A mixture of
all the sulfonyl chlorides would yield a final mixture that would
not allow for the analysis of individual peaks due to some final
compounds D (i.e., D-52 rt 17.89 min and D-47 rt 17.91 min)
having similar retention times and/or same molecular weights,
and therefore individual changes in reaction rates would not be
determinable. Because of this, multiple sample sets were set up
to maximize the amount of compounds present that would still
be distinguishable by HPLC-MS analysis. The control
compound, D-45, was present in every sample set. Each
designed mixture of sulfonyl chlorides was competitively
coupled to L-phenylalanine, L-phenylalanine−MBHA resin
(Scheme 1 Compound A) using an 8-fold excess (meaning
a
The boxed functionalities are the functionalities that provided the
desired compounds in purities sufficient enough to be included in the
final library syntheses.
compound D). L-Phenylalanine in the R1 and R2 position was
chosen because of its strong UV absorbance properties at 214
nm and nonreactive side chain. All individual compounds were
analyzed using reverse phase HPLC-MS (See Table 2). The UV
component of the HPLC-MS is used in the study in order to
quantify the relative ratios while the MS component is used to
aid in the identification of the compounds. A compound,
505
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
Table 2. Calculated Values for the Various Synthesized Compounds with L-Phenylalanine in the First Two Positions and a
Sulfonyl Chloride in the Third Position
molecular
weight
retention
time
UV absorptivity correction factor
relative to control
ratio relative to final reagent ratio relative
sulfonyl chloride
compound
control
to control
thiophene-2-sulfonyl chloride
benzenesulfonyl chloride
D-34
D-35
D-36
458.10
451.90
494.05
17.00
17.23
19.50
1.30
0.89
0.81
4.08
0.60
0.37
3.15
0.67
0.45
2,4,6-trimethylbenzene-1-sulfonyl
chloride
naphthalence-2-sulfonyl chloride
4-chlorobenzene-1-sulfonyl chloride
4-bromobenzene-1-sulfonyl chloride
4-iodobenzene-1-sulfonyl chloride
D-37
D-39
D-40
D-41
D-42
502.00
486.10
531.90
578.00
519.90
18.66
18.38
18.58
18.83
18.91
0.51
0.97
1.03
1.10
1.02
0.26
0.49
0.51
0.48
0.41
0.52
0.51
0.50
0.44
0.40
4-(trifluoromethyl)benzene-1-
sulfonyl chloride
4-methoxybenzene-1-sulfonyl
chloride
D-44
D-45
481.95
508.15
17.31
20.01
1.06
1.00
1.11
1.00
1.04
1.00
4-tert-butylbenzene-1-sulfonyl
chloride (control)
2-chlorobenzene-1-sulfonyl chloride
2-bromobenzene-1-sulfonyl chloride
Biphenyl-4-sulfonyl chloride
D-47
D-48
D-50
D-51
D-52
485.80
532.00
528.15
466.20
465.95
17.91
18.09
19.46
17.96
17.89
0.88
0.87
0.84
0.78
0.96
0.77
0.69
0.72
0.63
0.70
0.87
0.79
0.85
0.81
0.73
3-methylbenzene-1-sulfonyl chloride
4-methylbenzene-1-sulfonyl chloride
Figure 1. Example of two UV spectra (red and green), which are not directly comparable because of differences in attenuation and concentration,
and the result (blue) of finding the optimal overlap between them. The difference in the spectra is now apparent and quantifiable.
the total excess of the sulfonyl chlorides) over the resin. The
experiment was done so that every compound was analyzed in
duplicate. The HPLC-MS 214 nm UV peaks were analyzed
using the same integration parameters that were used to obtain
the ratios in the previous experiments. The correction factor for
each compound was used to compare the amount of that
compound to the amount of the control compound. No
compound deviates in amount by more than 20% from that of
the control (see Figure 1). This is well within experimental
error and shows that there is no substantial effect on the
reaction rates of the individual sulfonyl chlorides from the
presence of other sulfonyl chlorides.
Differences in Reaction Rate Caused by Different On-
Resin Amino Acids. It has been demonstrated that differences
in the on-resin amino acid onto which the reagents couple will
cause differences in the reaction rates of the individual
reagents.9,17 To explore the extent to which this is the case,
we have developed a novel technique to quantify the overall
difference between two mixtures containing multiple com-
pounds. In general, direct comparison between two different
HPLC-MS 214 nm UV spectra is complicated by differences in
concentration and attenuation from run to run. Further, use of
a reference compound cannot quantify differences in peak
shape that may result when multiple compounds with similar
retention times are present, and can only be used to adjust
relative peak heights. However, working under the assumption
that two identical samples will have geometrically similar 214
nm UV spectra, it is posited that through appropriate shifting
and scaling two 214 nm UV spectra may be compared directly
for similarity. In particular, given two samples, first the HPLC-
MS was used to determine the initial retention time of the first
peak of interest, ti, and the final retention time of the final peak
of interest, tf, of both samples. Then, the baseline was assumed
to be linear and was removed from both samples via the
equation:
506
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
Figure 2. 214 nm UV spectra of stock solution spiked with specific compounds, as compared to the stock solution alone, after optimal overlap
correction.
Figure 3. Errors are between 214 nm UV spectra of stock solution spiked with specific compounds, as compared to the stock solution alone, after
optimal overlap correction. The relationship between the percentage errors and amount of compounded added is clearly linear, with all R2 values
exceeding 0.99. Steeper slopes correspond to higher UV absorptivity.
where integration of the difference term D(t) corresponds to
UV(tf ) − UV(ti)
UVwithoutbaseline(t) = U(t) −
(t − ti)
the area between the two spectra after overlapping has
occurred, and therefore quantifies differences in the relative
makeups of the two samples. To this end, given two 214 nm
tf − ti
− UV(ti)
UV spectra, a variant of the Accelerated Random Search
18
It was assumed that for two identical samples there exist
parameters A, b, h, and k such that
Algorithm
can be used to find the optimal overlap
parameters A, b, h, and k which minimize the area between
the two spectra; this minimum area serves as a measure of the
“difference” between the two samples (See Figure 1). The
Accelerated Random Search Algorithm is an optimization
algorithm shown in ref 18 to have excellent rapid convergence
properties while maintaining consistency and requiring minimal
parameter manipulation to achieve optimal results.
To test the resolution of the above method a synthetic
mixture of 15 compounds (type C scheme 1) (termed “stock
solution” from here on out) was made using L-phenylalanine
(R1), L-phenylalanine (R2), and X3 (where X3 represents a
mixture of the 15 sulfonyl chlorides of interest using the
prescribed ratios calculated as in Table 2). Multiple aliquots of
this stock solution were made and spiked with one of three
sample1
withoutbaseline
sample2
withoutbaseline
UV
(t) = A·UV
(b·(t − h)) + k
In other words, it was assumed that there exists appropriate
vertical and horizontal scaling parameters (A and b,
respectively) and vertical and horizontal shifting parameters
(k and h, respectively) so that the UV spectra of identical
samples can be made to “overlap” one another. Furthermore,
for two samples that are not identical it may be thought to have
the relationship
sample1
withoutbaseline
sample2
withoutbaseline
UV
(t) = A·UV
(b·(t − h)) + k
+ D(t)
507
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
different individual compounds (D-34, D-37, D-45) at one of
three different percentage increases, 25%, 50%, and 100%. The
samples were then analyzed using the HPLC-MS 214 nm UV
spectra, with each sample compared to the unspiked stock
solution. Optimal overlapping was performed as described
above, and the area between the resulting curves was calculated
(see Figures 2 and 3, and Table 3). The spectra after
and unusuals) in the R1 and R2 positions (Scheme 1
Compound B). The two reaction vessels were set up. In the
first reaction vessel two “tea bags”16 were added, one containing
compound A and the other containing compound B. In the
second reaction vessel, two tea bags both containing compound
A were added. To each vessel an equal molar mixture of the 15
sulfonyl chlorides was added. It was added so as to provide only
a 1 equivalent amount of sulfonyl chloride to available free
amine (note this differs from the 8-fold excess used above). The
assumption is that in the presence of Compound B, if
Compound B has an average reaction rate to any of the
sulfonyl chlorides that differs drastically from that of
Compound A, it will disrupt the equimolarity of the reagents
reacting with Compound A and cause an unequal distribution
of products. The final mixture products (of the form
Compound C Scheme 1) from Compound A reacted in the
presence of Compound B and in the absence of Compound B
were analyzed using HPLC-MS 214 nm UV, and their spectra
were compared both to each other and to the aforementioned
stock solution (done using the corrected ratios with an 8-fold
excess) (see Figure 4).
The largest difference between spectra for the stock solution
(using the 8-fold excess) and the spectra from the mixture
products derived from compound A using the 1-fold excess in
the presence and absence of compound B occurs at an
approximate retention time of 17.0 min. This corresponds to
compound D-34, L-phenylalanine (R1), L-phenylalanine (R2),
thiophene-2-sulfonyl chloride (R3); this difference is unsurpris-
ing since D-34 has such a slow reaction rate (Table 2). Most
importantly, the spectra for both the 1-fold excess compound C
samples do not differ substantially from one another, with an
overall difference of 7.01%. There are slight differences between
the spectra at all retention times, but the largest difference
occurs at an approximate retention time of 18.85 min. This
peak corresponds to compound D-41, L-phenylalanine (R1), L-
phenylalanine (R2), 4-iodobenzene-1-sulfonyl chloride (R3),
and compound D-42, L-phenylalanine (R1), L-phenylalanine
(R1), 4-(trifluoromethyl)benzene-1-sulfonyl chloride (R3), both
of which have similar correction factors to the control D-45, L-
phenylalanine (R1), L-phenylalanine (R2), 4-tert-butylbenzene-
1-sulfonyl chloride (R3) (see Table 2). In the spiking
experiment described early, it was determined that a mixture
which contained 2-fold the amount of D-45 corresponded to a
Table 3. Percentage Error between 214 nm UV Spectra of
Stock Solution Spiked with Specific Compounds, As
Compared to the Stock Solution Alone
test sample
stock solution
comparison to stock solution (%)
0.61
2.22
3.48
5.97
5.09
8.24
16.79
2.77
4.62
8.22
stock solution +25% D-34
stock solution +50% D-34
stock solution +100% D-34
stock solution +25% D-37
stock solution +50% D-37
stock solution +100% D-37
stock solution +25% D-45
stock solution +50% D-45
stock solution +100% D-45
overlapping clearly show differences even when only 25%
additional compound is added, and the area difference between
the curves scales linearly with the amount of compound added.
Unsurprisingly, the compound with the lowest correction factor
(and therefore highest UV absorptivity), D-37 (correction
factor of 0.51, Table 2), also had the greatest slope; the
compound with the highest correction factor (and therefore
lowest UV absorptivity), D-34, had the lowest slope.
With the ability to compare UV spectra directly, the
following experiment was devised to determine the impact of
differing on-resin amino acids to the reaction rates of the
sulfonyl chlorides. Ideally, in order for the above ratios to
produce as equimolar a mixture as possible, the group of all on-
resin functionalities should react at approximately the same
average rate as the specific functionality used to calculate the
ratios (L-phenylalanine). Two different on-resin starting
materials were prepared; one with L-phenylalanine used in
the R1 and R2 positions (Scheme 1 Compound A) and one
with an equal molar mixture of 66 amino acids (including L, D,
Figure 4. 214 nm UV spectra after optimal overlap correction of 8-fold stock solution and 1-fold reactions. The differences between the samples are
not substantial.
508
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
Figure 5. Results of a Monte Carlo simulation demonstrating the robustness of mixtures to errors in equimolarity when tested in simple competitive
binding assays.
difference in spectra of 8.22%, therefore, we posit that no single
compound molarity in the mixture obtained from reacting
compound A in the presence of compound B differs more than
2-fold from that of the mixture obtained from reacting
compound A in the absence of compound B. In fact, since
the 7.01% error is distributed among multiple peaks, and
multiple compounds are present in certain peaks, it is likely that
the difference in relative constituency is far less than 2-fold.
This means that the average reaction rate of all on-resin
functionalities is only marginally different from the reaction
rates used to calculate the ratios for this specific reaction.
Robustness of Combinatorial Libraries to Errors in
Equimolarity. Although the possibility of differing reaction
rates resulting in a 2-fold error in equimolarity may seem to
preclude usefulness for this library for meaningful screening
data, this is in fact not the case. To demonstrate this,
hypothetical mixtures containing various percentages of 10 nM
IC50 active compounds and 1000 nM IC50 inactive compounds
were considered in association with a simple competitive
binding assay. The IC50 of such a mixture as a function of its
percentage of active compounds may be calculated directly
using the harmonic mean 19. To determine the effect of n-fold
error in the equimolarity of compounds in this setting, for each
percentage of active compounds a Monte Carlo simulation was
performed, generating a uniform error in relative abundance of
at most n-fold, for n = 2, 5, and 10. The upper and lower
bounds of the middle 95% of simulations is plotted in Figure 5
as a function of the percentage of active compounds; as is
evident, even a 10-fold maximal error in equimolarity did not
result in even a 2-fold error in mixture IC50 at any percentage of
active compounds. The maximal errors in equimolarity
presented herein are therefore well within the acceptable
range for use in screening.
compound 100-fold better than the current best, but rather that
the mixture contains some compounds that bind well, and
others that do not bind at all. In our experience, a 100-fold
difference in activity among the thousands of compounds in the
library is a reasonable assumption.
Application to the Design of Two New Positional
Scanning Libraries. The correction factors obtained for the
sulfonyl chlorides were then applied to synthesize two
heterocyclic positional scanning libraries. A linear sulfonamide
precursor is synthesized using the “tea bag” method. Two
different cyclizations are applied to form either a 2-piperazine
sulfonamide library or a 2-guanidine sulfonamide library (See
scheme 1 structure 7 and 8). As mentioned earlier and reported
in Table S1 in the Supporting Information, 104 individual
compounds were synthesized and analyzed to test the
robustness of different functionalities at the three diversity
positions. The results of the sulfonyl chlorides analyzed and the
resulting 15 chosen for ratio determination are shown in Table
1. These 15 sulfonyl chlorides were all used in the synthesis of
the two new positional scanning libraries (7 and 8). In Table S1
in the Supporting Information we also report the results
obtained from testing different functionalities at the R1 and R2
positions. From these results 34 amino acids (L, D, and unusual)
were selected for the libraries. Each library therefore contains
17 340 compounds (34 × 34 × 15). A list of the amino acids
used in the libraries as well as their associated ratios can be
found in the Supporting Information (Table S2).
Evaluation of Two New Positional Scanning Libraries
in Mu and Kappa Binding Assays. The two new libraries 7
and 8 were both screened in two different opioid binding
assays, a mu binding assay and a kappa binding assay
(Complete results can be found in the Supporting Information,
See Table S3 Supporting Information). The screening was
done in part to confirm the integrity of the synthesis of the
libraries. The harmonic mean offers an excellent method of
using such positional scan data to assess the integrity of the
synthesis process 19. In particular, since each position
The values 10 nM and 1,000 nM are representative of the
harmonic mean of the “active” individual compounds versus the
“inactive” individual compounds within the mixture. This does
not mean to imply that the mixture will always contain a
509
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
subdivides the entire library, if the individual compounds are
equimolar in all mixtures then the harmonic mean of the IC50
values of the mixtures in each position should be the same, and
that value should also be the IC50 of the all X mixture. On the
other hand, if there is substantial difference in the constituents
of the mixtures by position, then the harmonic means at each
position may be drastically different. In Table 4, we see the
CONCLUSION
■
We have demonstrated the utility of a new method for
determining the isokinetic ratios of reactant species. We then
applied it in order to calculate the ratios for mixtures of sulfonyl
chlorides reacted with on-resin amines. In doing so, we also
demonstrated a new method to access the effect of the on-resin
amines toward the relative reaction rates of mixtures of sulfonyl
chlorides and shown that our calculated ratios are not
substantially effected by the on-resin amines. These new
methodologies can be applied to study a wide variety of
isokinetic ratios and the only instrument of analysis needed is
an HPLC.
Table 4. Harmonic Mean Values for Library 7 and 8 in Mu
and Kappa Binding Assays
experimental IC50, untested extrapolated from %Inh (nM)
library 7
library 8
Additionally the determination of ratios for the formation of
a sulfonamide bond enables us to expand the diversity of
current positional scanning libraries. The two specific libraries
described in this publication have been added to our current
collection of positional scanning libraries that are available for
use through collaborative efforts.
mu
kappa
mu
kappa
R1
11147
8864
9508
8134
10971
9019
9345
7416
4444
4992
5249
6884
15357
15792
13343
14337
R2
R3
all X
EXPERIMENTAL PROCEDURES
■
harmonic means of the IC50 values of the mixtures at each
position, for both targets and both libraries. For mixtures that
did not show substantial activity at high doses, IC50 values were
extrapolated using single-point percent inhibition. The All X
was also treated in this manner. As is evident, the resulting
harmonic means do not vary substantially, with a maximal
deviation of 1.55-fold and most values quite a bit more similar.
This is well within the resolution of the assay, and we can
therefore be further confident that there is not substantial
difference in the constituencies of the mixtures at the different
positions.
Discussion Regarding Use of Detection Methods in
Determining Isokinetic Ratios. It should be noted that we
initially assessed the validity of utilizing the MS portion of the
HPLC-MS in determining the ratios for the reaction studied in
this manuscript. However it quickly became apparent that
correction factors would be needed for MS as equal molar
mixtures of two compounds did not produce equal ratios of
ions and the MS signal does not scale linearly with molarity.
Additionally we determined that the MS was less consistent
than the UV component at providing reproducible data. For
example if we injected the same equal molar mixture of two
components into the HPLC-MS over multiple runs the run to
run variation in UV absorption was significantly lower than the
variation in the MS ionization. Therefore the MS did not
provide any advantage over the UV component in determining
our isokinetic ratios.
At the time of this study, we did not have reasonable access
to an NMR for use in determining the ratios of this reaction.
This in of itself is one advantage of the method we report in our
manuscript as a researcher would theoretically only need use of
an HPLC to determine relative ratios. The MS component is
only used as confirmation of compound identity. Since an
HPLC is more affordable and likely more available to a wider
number of researchers than either HPLC-MS or NMR we felt
our reported approach would be more applicable. It should be
noted that we now do have reasonable access to an NMR and
are currently working out a method for calculating ratios with
the NMR. However at this time we are finding that one limiting
factor is the nonconsistent confounding influence of the solvent
peak which strongly biases attempts for direct overlap. We are
hoping to resolve this problem and soon report our findings on
the use of NMR for the determination of isokinetic ratios.
General Procedure for Individual Compounds 5 (As
Outlined in Scheme 1). A tea bag filled with mBHA resin is
neutralized with 3 washes of 5% DIEA in DCM. The tea bag
filled with neutralized resin is added to a reaction vessel that has
been allowed to preactivate for 15 min with a Boc-AA [6 equiv,
0.1 M dimethylformamide (DMF)], DIC (6 equiv, 0.1 M
DMF), and HOBT (6 equiv, 0.1 M DMF) for 2 h at room
temperature, followed by washes DMF (2×) and dichloro-
methane, DCM (2×). The Boc was removed using 55%
trifluoroacetic acid in DCM for 30 min, followed by washes
with DCM (2×), IPA (2×), and DCM (3×). Then the resin
was neutralized using 5% DIEA in DCM (3×) followed by
washes of DCM (3×). This was repeated so that a compound 2
was obtained. Sulfonyl chloride (8 equiv, 0.1 M anhydrous
DCM) was coupled to 2 using DIEA (8 equiv) overnight
producing compound 3. Each coupling reaction was monitored
by the use of the ninhydrin test to ensure the coupling was
complete. The tea bag containing 3 was added to a reaction
vessel containing a 40-fold excess of BH3−THF and heated at
60 °C for 96 h after which the solution was poured off and the
resin filled tea bag was treated with piperidine at 60 °C for 24 h.
General Procedure for Individual Compounds 7. A tea
bag containing compound 5 is added to a reaction vessel
containing bromoacetic acid (5 equiv, 0.1 M anhydrous DCM),
DIC (5 equiv), and DIEA (2.5 equiv) and shaken overnight.
The solution is poured off and excess reagents are washed off
by successive washes with DMF and DCM. The resulting
products 6a and 6b are then added to a reaction vessel
containing a 40 fold excess of BH3−THF and heated at 60 °C
for 96 h, after which the solution was poured off and the resin
filled tea bag was treated with piperidine at 60 °C for 24 h. The
resin is allowed to dry prior to cleavage with HF/anisole at 0
°C for 7hrs. The final product 7 is extracted from the resin
using 95% acetic acid frozen and lyophilized. After which it is
dissolved in a solution of 50% acetonitrile and 50% water,
frozen, and lyophilized before the dry powder is then analyzed.
(Note: It should be noted that several other methods for
transforming compounds of type 5 to compounds of type 7 were
attempted including treatment with oxalyldiimidazole followed by
reduction, as well as several different conditions involving the use of
1,2-dibromoethane, to circumvent the need for the second reduction.
However none of the tested conditions produced the desired test
510
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
compounds in the yields and quality that the method described here
did. The results of the 1,2-dibromoethane experiments are the
subject matter of previous publication20 and will also be addressed
in future publications.) H1NMR of 7−33 N-((S)-1-((S)-2-
benzylpiperazin-1-yl)-3-(4-hydroxyphenyl)propan-2-yl)-
benzenesulfonamide (500 MHz, DMSO-d6): δ 8.30 (1H,s),
7.67 (2H, d, J = 5.0 Hz), 7.54 (1H, t) 7.45 (2H, t), 7.27- 7.02
(10H, m), 3.33 (2H, s), 2.78 (1H, q), 2.64 −2.53 (6H, m), 2.39
(2H, s), 2.31 (2H, q), 2.24 (1H, q), 1.23 (1H, s).
ASSOCIATED CONTENT
* Supporting Information
Data for library members. This material is available free of
charge via the Internet at http://pubs.acs.org.
■
S
AUTHOR INFORMATION
Corresponding Author
*Phone: (772)-345-4800. Fax: (772)-345-3649. E-mail:
rhoughten@tpims.org.
■
General Procedure for Individual Compounds 8. A tea
bag containing compound 5 is added to a reaction vessel
containing cyanogen bromide (5 equiv, 0.1 M anhydrous
DMF) and shaken overnight. The solution is poured off and
excess reagents are washed off by successive washes with DMF
and DCM. The resin is allowed to dry prior to cleavage with
HF/anisole at 0 °C for 1.5 h. The final product 8 is extracted
from the resin using 95% acetic acid frozen and lyophilized.
After which it is dissolved in a solution of 50% acetonitrile and
50% water, frozen, and lyophilized before the dry powder is
then analyzed. H1NMR of 8−40 N-((S)-1-((S)-5-benzyl-2-
iminoimidazolidin-1-yl)-3-phenylpropan-2-yl)-4-bromobenze-
nesulfonamide (500 MHz, DMSO-d6): δ 8.45 (1H, s), 7.50
(1H, d, J = 1.7 Hz), 7.49 (4H, d, J = 1.8 Hz), 7.37−7.22 (5H,
m), 7.09−7.01 (6H, m), 3.68 (2H, quin), 3.32 (1H, t), 3.22
(1H, t), 3.22 (1H,q), 2.96−2.93 (2H, m) 2.67−2.62 (3H, m).
General Procedure for μ Receptor Binding Assay. Rat
cortices were homogenized using 50 mM Tris, pH 7.4, and
centrifuged at 16 500 rpm for 10 min. The pellets were
resuspended in fresh buffer and incubated at 37 °C for 30 min.
Following incubation, the suspensions were centrifuged as
before, the resulting pellets resuspended in 100 volumes of
buffer A plus 2 mg/mL bovine serum albumin (membrane
buffer) and the suspensions combined. Each assay tube
contained 0.5 mL of membrane suspension, 2 nM [3H]-
DAMGO, and 0.02 mg/mL mixture in a total volume of 0.65
mL. Assay tubes were incubated for 1 h at 25 °C. The reaction
was terminated by filtration through GF/B filters, soaked in 5
mg/mL bovine serum albumin, 50 mM Tris, pH 7.4, on a
Tomtec Mach II Harvester 96. The filters were subsequently
washed with 6 mL of assay buffer. Unlabeled DAMGO was
used as a competitor to generate a standard curve and
determine nonspecific binding. Bound radioactivity was
counted on a Wallac Betaplate Liquid Scintillation Counter.
General Procedure for κ Receptor Binding Assay.
Guinea pig cortices and cerebella were homogenized using 50
mM Tris, pH 7.4, 10 mM MgCl2−6H2O, 200 μM PMSF (assay
buffer) and centrifuged at 16 500 rpm for 10 min. The pellets
were resuspended in fresh buffer and incubated at 37 °C for 30
min. Following incubation, the suspensions were centrifuged as
before, the resulting pellets resuspended in 100 volumes of
buffer A plus 2 mg/mL bovine serum albumin (membrane
buffer) and the suspensions combined. Each assay tube
contained 0.5 mL of membrane suspension, 2 nM [3H]U69
593, and 0.02 mg/mL mixture in a total volume of 0.65 mL.
Assay tubes were incubated for 2 h at 25 °C. The reaction was
terminated by filtration through GF/B filters, soaked in 5 mg/
mL bovine serum albumin, 50 mM Tris, pH 7.4, on a Tomtec
Mach II Harvester 96. The filters were subsequently washed
with 6 mL of assay buffer. Unlabeled U50,488 was used as a
competitor to generate a standard curve and determine
nonspecific binding. Bound radioactivity was counted on a
Wallac Betaplate Liquid Scintillation Counter.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The authors are grateful for the support of the National
Institute on Drug Abuse (DA031370) and the State of Florida,
Executive Office of the Governor’s Office of Tourism, Trade,
and Economic Development. The authors would also like to
thank Jon Appel for his editorial assistance and excellent
critiques.
REFERENCES
■
(1) Dolle, R. E.; Bourdonnex, B. L.; Worm, K.; Morales, G. A.;
Thomas, C. J.; Zhang, W. Comprehensive Survey of Chemical
Libraries for Drug Discovery and Chemical Biology: 2009. J. Comb.
Chem. 2010, 12, 765−806.
(2) Houghten, R. A.; Pinilla, C.; Giulianotti, M. A.; Appel, J. R.;
Dooley, C. T.; Ostresh, J. M.; Yu, Y.; Maggiora, G. M.; Medina-Franco,
J. L.; Brunner, D.; Schneider, J. Strategies for the Use of Mixture-Based
Synthetic Combinatorial Libraries: Scaffold Ranking, Direct Testing In
Vivo, and Enhanced Deconvolution by Computational Methods. J.
Comb. Chem. 2008, 10, 3−19.
(3) Pinilla, C.; Appel, J. R.; Borras, E.; Houghten, R. A. Advances in
̀
the Use of Synthetic Combinatorial Chemistry: Mixture-Based
Libraries. Nat. Med. 2003, 9, 118−122.
(4) Reilley, K. J.; Giulianotti, M.; Dooley, C. T.; Nefzi, A.;
McLaughlin, J. P.; Houghten, R. H. Identification of Two Novel,
Potent, Low-Liability Antinociceptive Compounds from the Direct In
Vivo Screening of a Large Mixture-Based Combinatorial Library. AAPS
J. 2010, 12, 318−329.
(5) Judkowski, V.; Bunying, A.; Ge, F.; Appel, J. R.; Law, K.; Sharma,
A.; Raja-Gabaglia, C.; Norori, P.; Santos, R. G.; Giulianotti, M. A.;
Slifka, M. K.; Douek, D. C.; Graham, B. S.; Pinilla, C. GM-CSF
Production Allows the Identification of Immunoprevalent Antigens
Recognized by Human CD4 + T Cells Following Smallpox
Vaccination. PLoS One 2011, 6 (9), No. e24091.
(6) Houghten, R. A.; Pinilla, C.; Blondelle, S. E.; Appel, J. R.; Dooley,
C. T.; Cuervo, J. H. Generation and Use of Synthetic Peptide
Combinatorial Libraries for Basic Research and Drug Discovery.
Nature 1991, 354, 84−86.
(7) Lam, K. S.; Salmon, S. E.; Hersh, E. M.; Hruby, V. J.; Kazmierski,
W. M.; Knapp, R. J. A New Type of Synthetic Peptide Library for
Identifying Ligand-Binding Activity. Nature 1991, 354, 82−84.
(8) Furka, A.; Sebestyen, F.; Asgedom, M.; Dibο, G. General Method
for Rapid Synthesis of Multicomponent Peptide Mixtures. Int. J.
Peptide Protein Res. 1991, 37, 487−493.
́
̇
(9) Ostresh, J. M.; Winkle, J. H.; Hamashin, V. T.; Houghten, R. A.
Peptide Libraries: Determination of Relative Reaction Rates of
Protected Amino Acids in Competitive Couplings. Biopolymers 1994,
34, 1681−1689.
(10) Ljungdahl, T.; Veide-Vilg, J.; Wallner, F.; Tamas, M. J.; Grøtli,
M. Positional Scanning Peptide Libraries for Kinase Substrate
Specificity Determinations: Straightforward and Reproducible Syn-
thesis Using Pentafluorophenyl Esters. J. Comb. Chem. 2010, 12, 733−
742.
(11) Acharya, A. N; Ostresh, J. M.; Houghten, R. A. Determination of
Isokinetic Ratios Necessary for Equimolar Incorporation of Carboxylic
511
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512

ACS Combinatorial Science
Research Article
Acids in the Solid-Phase Synthesis of Mixture-Based Combinatorial
Libraries. Biopolymers 2002, 65, 32−39.
(12) Dombi, K. L; Steiner, U. E.; Richert, C. Rapidly Measuring
Reactivities of Carboxylic Acids to Generate Equireactive Building
Block Mixtures: A Spectrometric Assay. J. Comb. Chem. 2003, 5, 45−
60.
(13) Ostresh, J. M.; Schoner, C.; Giulianotti, M.; Kurth, M.; Dorner,
̈
B.; Houghten, R. Combinatorial Libraries: Equimolar Incorporation of
Benzaldehyde Mixtures in Reductive Alkylation Reactions. Pept. Proc.
Am. Pept. Symp. 2002, 5, 57−58.
(14) Nefzi, A.; Ostresh, J. M.; Yu, J.; Houghten, R. A. Combinatorial
Chemistry: Libraries from Libraries, the Art of the Diversity-Oriented
Transformation of Resin-Bound Peptides and Chiral Polyamides to
Low Molecular Weight Acyclic and Heterocyclic Compounds. J. Org.
Chem. 2004, 69, 3603−3609.
(15) Nefzi, A.; Ostresh, J. M.; Meyer, J.-P.; Houghten, R. A. Solid
Phase Synthesis of Heterocyclic Compounds From Linear Peptides:
Cyclic Ureas and Thioureas. Tetrahedron Lett. 1997, 38, 931−914.
(16) Houghten, R. A. General Method for the Rapid Solid-Phase
Synthesis of Large Numbers of Peptides: Specificity of Antigen−
Antibody Interaction at the Level of Individual Amino Acids. Proc.
Natl. Acad. Sci. U.S.A. 1985, 82, 5131−5135.
(17) Boutin, J. A.; Gesson, I.; Menlin, J.-M.; Bertin, S.; Lambert, P.-
H.; Volland, J.-P.; Fauchere, J.-L. Limitations of the Coupling of
̇
Amino Acid Mixtures for the Preparation of Equimolar Peptide
Libraries. Mol. Diversity 1997, 3, 43−60.
́
(18) Appel, M. J.; LaBarre, R; Radulovic, D. On Accelerated Random
Search, SIAM. J. Optim. 2003, 14, 708−731.
(19) Santos, R. G.; Giulianotti, M. A.; Dooley, C. T.; Pinilla, C.;
Appel, J. R.; Houghten, R. A. Use and Implications of the Harmonic
Mean Model on Mixtures for Basic Research and Drug Discovery. ACS
Comb. Sci. 2011, 13, 337−344.
(20) Nefzi, A.; Mimna, R. A.; Houghten, R. A. Parallel Solid-Phase
Synthesis of Disubstituted 1,6-Piperazine-2-ones. J. Comb. Chem. 2002,
4, 542−545.
512
dx.doi.org/10.1021/co300060s | ACS Comb. Sci. 2012, 14, 503−512