Synthesis and Evaluation of a Library of Trifunctional Scaffold-Derived Compounds as Modulators of the Insulin Receptor

Article abstract of DOI:10.1021/acscombsci.6b00132

We designed a combinatorial library of trifunctional scaffold-derived compounds, which were derivatized with 30 different in-house-made azides. The compounds were proposed to mimic insulin receptor (IR)-binding epitopes in the insulin molecule and bind to and activate this receptor. This work has enabled us to test our synthetic and biological methodology and to prove its robustness and reliability for the solid-phase synthesis and testing of combinatorial libraries of the trifunctional scaffold-derived compounds. Our effort resulted in the discovery of two compounds, which were able to weakly induce the autophosphorylation of IR and weakly bind to this receptor at a 0.1 mM concentration. Despite these modest biological results, which well document the well-known difficulty in modulating protein-protein interactions, this study represents a unique example of targeting the IR with a set of nonpeptide compounds that were specifically designed and synthesized for this purpose. We believe that this work can open new perspectives for the development of next-generation insulin mimetics based on the scaffold structure.

Full text of DOI:10.1021/acscombsci.6b00132

Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)
Article
Synthesis and Evaluation of a Library of Tri-Functional Scaffold-
Derived Compounds as Modulators of the Insulin Receptor
Benjamin Fabre, Jan Picha, Vaclav Vanek, Irena Selicharová, Martina Chrudinová,
Michaela Collinsova, Lenka Zakova, Miloš Bud#šínský, and Jiri Jiracek
ACS Comb. Sci., Just Accepted Manuscript • DOI: 10.1021/acscombsci.6b00132 • Publication Date (Web): 02 Nov 2016
Downloaded from http://pubs.acs.org on November 6, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Combinatorial Science is published by the American Chemical Society. 1155
Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 32
ACS Combinatorial Science
1
2
3
4
5
6
Synthesis and Evaluation of a Library of Tri-Functional Scaffold-
Derived Compounds as Modulators of the Insulin Receptor
7
8
9
Benjamin Fabre1, Jan Pícha1, Václav Vaněk, Irena Selicharová, Martina
Chrudinová, Michaela Collinsová, Lenka Žáková, Miloš Buděšínský and Jiří
Jiráček*
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Institute of Organic Chemistry and Biochemistry, Czech Academy of Sciences, Flemingovo n. 2,
16610 Praha 6, Czech Republic
1 These authors contributed equally to this work
Supporting Information
ABSTRACT: We designed a combinatorial library of triꢀfunctional scaffoldꢀderived
compounds, which were derivatized with 30 different inꢀhouseꢀmade azides. The compounds
were proposed to mimic insulin receptor (IR)ꢀbinding epitopes in the insulin molecule and bind
to and activate this receptor. This work has enabled us to test our synthetic and biological
methodology and to prove its robustness and reliability for the solidꢀphase synthesis and testing
of combinatorial libraries of the triꢀfunctional scaffoldꢀderived compounds. Our effort resulted in
the discovery of two compounds, which were able to weakly induce the autophosphorylation of
IR and weakly bind to this receptor at a 0.1 mM concentration. Despite these modest biological
results, which well document the wellꢀknown difficulty in modulating proteinꢀprotein
interactions, this study represents a unique example of targeting the IR with a set of nonꢀpeptide
compounds that were specifically designed and synthesized for this purpose. We believe that this
work can open new perspectives for the development of nextꢀgeneration insulin mimetics based
on the scaffold structure.
1
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 2 of 32
1
2
3
4
5
6
7
8
INTRODUCTION
Although promising, the modulation of proteinꢀprotein interactions (PPIs) has been recognized as
one of the most challenging tasks in drug discovery.1ꢀ4 Individual proteins involved in PPIs are, in
many cases, thought to be undruggable, as the interfacial surface area formed by interacting
proteins is usually large and solventꢀexposed, the interaction surfaces are often shallow and
featureless and the binding regions in proteinꢀprotein interactions are mostly nonꢀcontiguous.5,6
For these reasons, it is difficult to mimic these interfaces by using small organic molecules.
However, it can be assumed that molecules that mimic several different nonꢀcontiguous binding
hotspots at the target protein surface have better chances of being potent modulators of protein
functions than do compounds that mimic only one epitope.7ꢀ9 For this purpose, we have recently
developed a versatile triꢀfunctional scaffold with orthogonally protected arms that can be
selectively functionalized by different azides10 (I, Fig. 1).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Triꢀfunctional and orthogonally protected scaffold I (Ref.10). The three arms with free,
TESꢀprotected, and TIPSꢀprotected propargylamine moieties enable successive copper(I)ꢀ
catalyzed azide–alkyne cycloadditions (CuAACs). The spacer with carboxylic acid enables the
attachment to the resin, and the CF2 groups permit the quantitative 19F NMR analysis.
Decades of intensive research did not fully clarify the main questions about the interaction of
insulin with its transmembrane tyrosine kinase receptor, the insulin receptor (IR). Formidable
problems with the production of IR constructs for structural studies and the complexity of IRꢀ
insulin interactions hampered the progress in deciphering the molecular nature of this assembly.11
For more than a decade, our laboratory has been involved in the development of insulin
2
ACS Paragon Plus Environment

Page 3 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
analogues to study the interaction of insulin with IR.12ꢀ21 Recently, we contributed an important
scientific result,22 which provided the first insight into the nature of the insulin interaction with its
primary binding site (Site 1) on the IR. The study was later followed by the elucidation of the IRꢀ
bound conformation of the crucial Cꢀterminus of the insulin Bꢀchain.23,24 Although these studies
were extremely important in insulin research, they did not provide information about the insulin
interaction with its second binding site (Site 2) on the IR, which is thought to trigger IRꢀinduced
intracellular signaling. Nevertheless, extensive mutagenesis studies (reviewed in Ref.(25)) have
identified key amino acids (GlyA1, IleA2, ValA3, GluA4, TyrA19, AsnA21, GlyB8, SerB9,
LeuB11, ValB12, LeuB15, TyrB16, PheB24, PheB25 and TyrB26 for Site 1 and ThrA8, Ile A10,
SerA12, LeuA13, GluA17, HisB10, GluB13 and LeuB17 for Site 2), which participate in the
formation of binding epitopes (hotspots) in the insulin molecule that are responsible for binding
to both IR sites.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Considering the central position of insulin in antiꢀdiabetes therapy, there is a need for
alternative forms of this hormone that would be metabolically stable, i.e., suitable for oral
delivery, and less expensive to produce than insulin. Such medicines would have an impact on
the costꢀeffectiveness of the health systems. Therefore, nonꢀpeptide molecules, such as insulin
mimetics, which are able to bind efficiently IR and elicit the metabolic responses of native insulin
(insulin agonists), would be of major interest. There are only a few reports on the development of
nonꢀpeptide insulin mimetics. The majority of compounds with insulin agonistꢀlike properties
does not interact with the insulinꢀbinding domain of IR but does interact with different
intracellular targets (e.g., the Merck fungal metabolite Lꢀ783,281, Ref.26). Moreover, only a few
compounds target the extracellular domain of IR, although this occurs with only a relatively weak
affinity.27ꢀ30 The most advanced progress in the development of insulin mimetics resulted from
the application of phage display to generate peptides that are able to compete with insulin and
have high affinities. However, these artificial peptides have a similar molecular weight to insulin
and a similar stability. Nevertheless, the Novo Nordisk study31ꢀ33 proved that the idea of insulin
mimetics is realistic and promising. Moreover, a recent interesting study describes the discovery
of an agonistic aptamer that is able to bind the IR (the interaction site is unknown), stimulate IR
autophosphorylation and elicit the biological effects of native insulin.34
The incomplete structural information about the molecular nature of the insulinꢀIR
interaction and the dynamic complexity of this interaction impede the efficient use of
3
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 4 of 32
1
2
3
4
5
6
7
8
9
computational techniques to predict the structures of artificial IR binders. However, we believe
that molecules mimicking several different nonꢀcontiguous binding hotspots at the insulin surface
have better chances of being potent insulin agonists than do compounds that only mimic one
epitope. Moreover, we believe that a combinatorial approach for synthesizing and testing
thousands of different compounds is the best starting strategy for the discovery of the first IR
binders, which can be further optimized by molecular modeling and approaches analyzing the
structureꢀfunction relationship. We hypothesized that a library of compounds derived from the
structure of scaffold I could provide some new IR binders. Figure 2 explains the principle of our
strategy using an example of a few residues forming binding epitopes in insulin. Recently, we
showed10 that a model compound derived from scaffold I can adopt an extended conformation
and that the distances between the distal carbon atoms are similar to the size of insulin molecule
(20ꢀ30 Å). Based on these presumptions, in this study, we employed scaffold I to generate a
combinatorial library of 1,000 compounds using a pool of 30 different azides, which should
provide various chemical properties. The library of the scaffoldꢀIꢀderived compounds was tested
for binding to isoform A of the insulin receptor (IRꢀA) and the IGFꢀ1 receptor (IGFꢀ1R) and its
ability to activate these receptors.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4
ACS Paragon Plus Environment

Page 5 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. The proposed strategy for the development of insulin mimetics. (A) The molecular
surface of the active (i.e., receptorꢀbound) conformation of insulin is represented here by the
crystal structure of AsnB26ꢀinsulin, pdb code 4ung, from Ref.24 Examples of IRꢀbinding amino
acids (epitopes) are depicted in yellow (GlyA1, IleA2, ValA3, GluA4 and TyrA19), green
(PheB24 and PheB25) and blue (GlyB8, SerB9, HisB10, LeuB11, GluB13 and LeuB15). Only a
subset of IRꢀbinding amino acids is shown here and all IRꢀbinding amino acids of insulin are
listed in the main text. (B) The selected IRꢀbinding epitopes were extracted from the structure of
AsnB26ꢀinsulin and should be mimicked by the compounds derived from scaffold I, as shown in
(C).
RESULTS
Design of the Libraries. Scaffold I was synthesized using the procedure described in our
previous work(10) and employing several improvements (see the Supplementary Information).
The 30 azides used for the derivatization of scaffold I are shown in Table 1, and their
syntheses are described in the Supplementary Information. The synthesis of specific azides was
5
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 6 of 32
1
2
3
4
5
6
generally motivated by mimicking the character of IRꢀbinding amino acids in insulin (i.e.
lipophilic, aromatic, acidic or polar, see above).
We used a splitꢀandꢀmix strategy, which allows the simple generation of defined mixtures of
the compounds.35ꢀ37 For the coupling to the individual arms (A, B or C in Table 1) of scaffold I,
we divided the azides into three sets (AꢀC). Each set is composed of 10 azides with varying
physicochemical properties (aromaticity, charge, and lipophilicity) to achieve a greater chemical
diversity in the final compounds. Each of the azides is present in only one set (the azides from set
A were only inserted at position A of the scaffold, and so on).
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 1. Sets of Azides (A-C) Used for the CuAACs on the Individual Arms (A-C,
Respectively) of Scaffold I.
A
B
C
Azide set
1
2
3
4
6
ACS Paragon Plus Environment

Page 7 of 32
ACS Combinatorial Science
1
2
3
4
5
5
6
7
8
9
6
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7
8
9
10
Synthesis and Deconvolution of the Library. For the solidꢀphase synthesis of the library,
we used the PEGꢀbased ChemMatrix resin38,39 because it gave satisfactory results for CuAACs
with the CuSO4/sodium ascorbate system in tBuOH/H2O.10,40 We used the Ramage linker, which
can be cleaved by low concentrations of TFA (3ꢀ5%) to generate the terminal amides.41
The synthesis of the library is shown in Scheme 1. We prepared ten subꢀlibraries, labeled C1
to C10, to reflect the presence of a single and specific azide from set C in each of the subꢀ
libraries. Each subꢀlibrary contained 100 compounds, which arose from the combinations
between the azides from sets A and B. Thus, the total library was composed of 1,000 compounds,
and each of these compounds had three different azides on arms AꢀC. The chemistry, which was
described in detail in our previous report,10 was easy to implement; dry or inert conditions are not
required, all reactions are performed at room temperature and the synthesis of the library is
completed within 14 days.
7
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 8 of 32
1
2
3
4
5
6
7
8
9
After cleavage from the resin and subsequent Boc/tBu deprotection with trifluoroacetic acid
(TFA, steps f and g in Scheme 1), the ten mixtures were extracted with diethyl ether to remove
the nonꢀpolar impurities (particularly the triisopropylsilane (TIS) derivatives; TIS was used to
scavenge the possible carbocations).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We supposed that some residual copper may be present in the mixtures obtained after the
diethyl ether extraction and performed an ICP analysis of the subꢀlibraries C9 and C10 (these two
subꢀlibraries appeared particularly bluish). The analyses showed a copper content of
approximately 1% (w/w). We subsequently investigated different purification methods to remove
copper: separation on Cꢀ4 or Cꢀ8 SPE cartridges, treatment with a thiolꢀbased cation exchange
resin, fast purification by preparative RPꢀHPLC or treatment with gaseous hydrogen sulfide
(H2S) (for details see the SI). The treatment of the subꢀlibraries with hydrogen sulfide, followed
by filtration, was the most optimal, simple, short and efficient method. Briefly, the subꢀlibraries
were dissolved in ACN/H2O, and gaseous hydrogen sulfide was blown at the surface of these
solutions upon stirring. After precipitation of the copper sulfides (apparition of a darkꢀbrown
precipitate), the solutions were filtered and lyophilized. Elemental analyses showed that the
amount of copper in subꢀlibraries C9 and C10 dropped from 1% to 0.06% and 0.01% (w/w),
respectively.
8
ACS Paragon Plus Environment

Page 9 of 32
ACS Combinatorial Science
1
2
3
4
O
O
N
N
A1
H
TES
N
N
N
CF2
O
5
6
7
8
TES
TES
O
HN
(CF2)
(CF2)
I
OH
A1
b - azide
TIPS
TES
Resin
Ax
N
N
N
9
TES
TIPS
TIPS
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(CF2) COOH
split, a
(CF2)
mix
(10 reactors)
A10
N
N
N
TIPS
TIPS
TES
TES
(CF2)
(CF2)
A10
b - azide
TIPS
TIPS
Ax
Ax
N
N
N
N
N
N
N
N
TES
TES
By
N
B1
c - azide B1
(CF2)
(CF2)
Ax
N
N
N
N
N
TIPS
Ax
TIPS
N
By
mi
x
(CF2)
split
Ax
N
N
N
N
N
N
TIPS
N
N
N
B10
c - azide B10
(CF2)
(CF2)
TIPS
Ax
TIPS
Ax
Ax
N
N
N
N
N
N N
N
N
N
N
N
N
N
N
O
N
N
N
By
e - Azide C1
By
(CF2)
(CF2)
(CF2)
C1
f,
g,
h,
i
NH2
N
N
N
N
N
N
C1
C1
split, d
Ax
Ax
Ax
N
N
N
N
N
N N
N
N
N
N
N
N
N
N
O
N
N
N
By
By
By
e- Azide C10
(CF2)
(CF2)
(CF2)
C10
NH2
f,g,h,i
N
N
N
N
N N
C10
C10
10 sub-libraries Cn
Scheme 1. Synthesis of the subꢀlibraries C1 to C10. Reagents and conditions: (a) Ramage
ChemMatrix® (2 equivalents, 200 µmol/reactor), PyBroP (2 equivalents), DIPEA (4
equivalents), DMF, 16 h; (b) azide A (5 equivalents), sodium ascorbate (0.5 equivalent),
CuSO4.5H2O (0.1 equivalent), tBuOH/H2O, 2×16 h; (c) azide B (5 equivalents), sodium
ascorbate (10 equivalents), CuSO4.5H2O (5 equivalents), tBuOH/H2O, 16+5 h; (d) TBAF (5
9
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 10 of 32
1
2
3
4
5
6
7
8
9
equivalents), DMF, 3×1.5 h; (e) azide C (5 equivalents), sodium ascorbate (0.5 equivalent),
CuSO4.5H2O (0.1 equivalent), tBuOH/H2O, 2×16 h; (f) 5% TFA in DCM, 2×30 minutes; (g) 50
% TFA in DCM + 2% TIS, 2 h; (h) Diethyl ether extraction (20 mL); (i) treatment with gaseous
H2S in ACN/H2O (30 minutes) followed by filtration.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Next, based on the biological results (see below), we deconvoluted subꢀlibrary C4, as shown
in Scheme 2. We repeated the synthesis described in Scheme 1, but omitted the second splitꢀandꢀ
mix process and only introduced the azide C4 during the third CuAAC. Consequently, we
obtained ten subꢀlibraries, C4B1-C4B10, which contained 10 compounds each that only varied at
position A. As some subꢀlibraries maintained a darkꢀbrown coloration after treatment with
hydrogen sulfide and filtration, we performed the ICP analyses of subꢀlibraries C4B3 and C4B7
(these subꢀlibraries had a particularly dark color). The copper content in these mixtures was
0.17% and 0.34% (w/w), respectively, and was higher than subꢀlibraries C9 and C10.
Because the biological activity observed in C4 was not detected in subꢀlibraries C4B1ꢀ
C4B10 (see below), we considered physicalꢀchemical arguments, selected subꢀlibrary C4B7
(which contains compounds that had the lowest molecular weights and displayed good solubility
in water) to be deconvoluted and thus synthesized compounds C4B7A1-C4B7A10 (Scheme 2).
Figure S1 shows the superimposed RPꢀHPLC profiles of subꢀlibrary C4B7 and compounds
C4B7A1ꢀC4B7A10, which could be easily identified in subꢀlibrary C4B7. LCꢀMS analysis of
subꢀlibrary C4B7 also confirmed the presence of the expected compounds C4B7A1ꢀC4B7A10
(Figure S2). The crude compounds C4B7A1ꢀC4B7A10 were prepared with satisfactory HPLC
purities ranging from 46 % to 71% (see Table S1).
10
ACS Paragon Plus Environment

Page 11 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Scheme 2. Overview of the synthesis of subꢀlibraries C1-C10 and C4B1-C4B10 and compounds
C4B7A1-C4B7A10. The conditions are the same as the conditions described in Scheme 1.
Biological Evaluations of Sub-Libraries C1-C10, C4B1-C4B10 and Compounds
19
C4B7A1-C4B7A10. First, based on the concentrations determined by F NMR, we prepared 1
mM stock solutions of individual libraries in 100 mM Hepes/NaOH buffer, pH 7.5 (see
Experimental Methods). The concept and the methodology of the determination of the
concentration of scaffold Iꢀderived compounds were described in detail in our previous
papers.10,40
All libraries and compounds were tested for binding to both IRꢀA and the IGFꢀ1 receptor
(IGFꢀ1R). The resulting concentrations of the compounds in all experiments were 0.1 mM, unless
stated otherwise. Further, the ability of the libraries and compounds to stimulate the
autophosphorylation of both receptors was assessed. There were no significant differences
between the abilities of the two types of receptors (IRꢀA and IGFꢀ1R) to bind to the compounds
11
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 12 of 32
1
2
3
4
5
6
or in the abilities of the compounds to stimulate the receptors. For this reason, we only show the
data obtained for IRꢀA.
The results of the binding of subꢀlibraries C1ꢀC10 to IRꢀA are shown in Supplementary
Figure S3. Briefly, at a 0.1 mM concentration, none of the subꢀlibraries effectively inhibited the
binding of 125Iꢀinsulin to IRꢀA. However, two subꢀlibraries, C4 and C2, showed a weak but
significant ability to induce IRꢀA autophosphorylation (Figure 3).
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1.0
0.8
0.6
0.4
0.2
** ***
0.0
Figure 3. Stimulation of IRꢀA phosphorylation by 0.1 mM concentrations of subꢀlibraries C1-
C10. The experimental values were related to the activity of 10 nM human insulin. Asterisks
indicate that the autophosphorylation of the receptor induced by the library differs significantly
(** p < 0.01, *** p < 0.001) from the control (background).
In the next step, we tested the biological activities of subꢀlibraries C4B1ꢀC4B10, each of
which contain 10 compounds (Scheme 2). None of these subꢀlibraries significantly inhibited the
binding of 125Iꢀinsulin to IRꢀA (Figure S4) or induced IRꢀA autophosphorylation (Figure S5) at
0.1 mM. Finally, the same absence of biological activity was observed for compounds C4B7A1ꢀ
C4B7A10, which do not bind to IRꢀA (Figure S6) or activate IRꢀA (Figure S7).
Searching for Active Compounds in Sub-Library C4. We wanted to understand why the
weak but significant biological activity of subꢀlibrary C4 was not observed in subꢀlibraries
C4B1-C4B10. Therefore, we verified whether some active compound(s) could be isolated from
12
ACS Paragon Plus Environment

Page 13 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
subꢀlibrary C4. After several iterative purifications by RPꢀHPLC, we found two fractions (C4-1
and C4-2) that were able to activate IRꢀA (Figure 4A and 4B). These fractions were analyzed by
highꢀresolution mass spectrometry (HRMS), and the experimental HRMS signals were attributed
to compounds C4B2A10 and C4B3A6 (Figure 4A).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Simplified scheme for the separation of subꢀlibrary C4 by preparative RPꢀHPLC. (A)
Stimulation of IRꢀA phosphorylation by the fractions isolated from subꢀlibrary C4 using
preparative RPꢀHPLC. Four to five fractions were collected during each successive
chromatography step and tested for their ability to stimulate the cells. Phosphorylated receptors
were detected on western blots using an antiꢀphosphoꢀIGFꢀ1R βꢀsubunit (Tyr1131) / IR βꢀsubunit
(Tyr1146) antibody. The cells were stimulated with 0.1 mg/mL fractions for 10 min. Fractions
C4-1 and C4-2 were analyzed by HRMS and the major m/z signals detected were attributed to
compounds C4B2A10 and C4B3A6. (B) RPꢀHPLC chromatograms of the crude subꢀlibrary C4
and the two biologically active isolated fractions (C4-1 and C4-2).
Synthesis of the Focused Library. These results encouraged us to synthesize compounds
C4B2A10 and C4B3A6 and to design a small focused library of compounds that were
derivatized with the azides (A6, A10, B2 and B3) found in compounds C4B2A10 and C4B3A6.
We decided to preserve azide C4 at position C of the scaffold. Varying the two remaining
positions with four different azides would yield 16 different compounds (42). However, we only
prepared 10 compounds (including C4B2A10 and C4B3A6) to reduce the number of syntheses,
13
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 14 of 32
1
2
3
4
5
6
7
8
9
and the structures are shown in Table 2. The six remaining compounds (C4A10B2, C4A6B3,
C4A10BA6, C4A10B3, C4A6B2 and C4B2B3) were not prepared because they differ from the
compounds shown in Table 2 only in the swapping of azides at positions A and B. The
compounds shown in Table 2 were purified by RPꢀHPLC and obtained in satisfactory yields,
ranging from 22% to 64% (Table S1). The chromatograms of the crude compounds are shown in
supplementary Figures S8 and S9, and their purities ranged from 51ꢀ74% (Table S1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 2. Structures of the Compounds from the Focused Library.
Azide
Azide
Azide
Code
at position C
at position B
at position A
of the compound
C4B2A10
C4B3A6
C4A10A10
C4A6A10
C4B3A10
C4A6A6
C4B2A6
C4B2B2
14
ACS Paragon Plus Environment

Page 15 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
C4B3B2
C4B3B3
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The compounds were then tested for their ability to induce IRꢀA autophosphorylation.
However, none of the new compounds showed significantly stronger biological activity than did
the “original” C4 subꢀlibrary or fractions C4-1 and C4-2, and only compounds C4B2A10 and
C4B3A6 weakly activated IRꢀA at a concentration of 0.1 mM (Figure 5A). Finally, we checked
whether the purified compound C4B2A10 can compete with 125Iꢀinsulin in binding to IRꢀA. We
found that this compound weakly inhibits the binding of insulin to IRꢀA at a concentration of 0.1
mM (Figure 5B).
Figure 5. (A) Stimulation of IRꢀA authophosphorylation by compounds from the focused subꢀ
library at a concentration of 0.1 mM compared with the original C4 library and fractions C4-1
and C4-2 isolated from the C4 library. The experimental values are related to the activity of 10
nM human insulin. The control is the buffer in which the compounds were solubilized. (B)
Inhibition of the binding of 125Iꢀinsulin to IRꢀA by compound C4B2A10.
DISCUSSION
Synthesis. An important aspect of combinatorial synthesis is to establish a convenient
synthetic protocol that affords the desired compounds (often present in mixtures) with purities
15
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 16 of 32
1
2
3
4
5
6
7
8
and yields that are sufficient for reliable biological evaluation. The data shown in the Supporting
Information (e.g., Figures S1, S2, S8, and S9 and Table S1) show the suitability of the synthetic
protocol for generating large sets and mixtures of diverse compounds.
Another important condition for combinatorial synthesis is the versatility and robustness of
the synthetic method; i.e., the applied chemistry should not interfere with the different
functionalities present in the precursor compounds. Here, we reported the successful syntheses of
compounds with primary alcohol (azide B7), ether, carbamate (azide C4), vinylbenzene (azide
B3), catechol (azide A2), nitrile (azide A6), tertiary amine (azide A5), thioether amide (azide A8)
or sulfonamide (azide B2) moieties and obtained the desired compounds without sideꢀreactions.
The main impurities observed on LCꢀMS (not shown) resulted from the double azide addition
(both at the free and the TESꢀprotected alkyne) during the first CuAAC (e.g. the compound
C4A3A3 was also isolated together with C4B7A3). This is not surprising, as the TES group
could be cleaved under the conditions used for the CuAAC. Overall, these results confirm the
suitability of the method for combinatorial applications. However, we should also mention two
possible limitations of our methodology: the final TFA cleavage that precludes acidꢀlabile
functions, and the treatment with hydrogen sulfide is not suitable for disulfides.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The strategy of orthogonal TIPS/TES protection of terminal alkynes was first developed in
Aucagne’s group,(42) who selectively deprotected the TES group by using Ag(I) salts and these
conditions were recently applied by Liskamp’s group.43,44 However, under our experimental
conditions, silver deposits appeared on the resin, which disabled further synthesis. We solved this
problem by performing the CuAAC directly on the TESꢀprotected terminal alkyne and by using
an excess of copper(II) and sodium ascorbate.10,40 Under these conditions, the TES group is
selectively deprotected and the cycloaddition with the azide occurs in oneꢀpot. The mechanism of
the TES cleavage during high copper load CuAAC is unclear and is currently being investigated
in our laboratory. Consequently, the only difficult point during the synthesis was removing the
copper that accumulated in the resin, particularly during the oneꢀpot TESꢀdeprotection/CuAAC
that was performed with excess copper, which could hypothetically interfere with the bioassays.
It was not trivial to remove the copper because we needed to establish a general method that was
suitable for all of the compounds in our library, which have different functionalities, polarities
and charges. After investigating various alternative methods (see the Supporting Information), we
opted to treat the crude mixtures/compounds in solution with hydrogen sulfide after cleavage
16
ACS Paragon Plus Environment

Page 17 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
from the resin; this strategy led to the precipitation of copper that could be filtered from the
solution. This method allowed us to reduce the copper content in subꢀlibraries C1ꢀC10 to a
negligible level. However, the copper content in subꢀlibraries C4B1ꢀC4B10 and in compounds
C4B7A1ꢀC4B7A10 remained higher, despite the application of the same treatment (Table S1).
These variations can be explained by the particular functionality of some compounds; for
instance, the carboxylic acid moiety of C4B7A9 (copper content of approximately 7% w/w) may
interfere with the precipitation of copper sulfide. Thus, we experimentally verified that the
presence of copper sulfide in concentrations up to 300 µM did not positively or negatively
interfere with our cellular binding assays to avoid any ambiguity in interpreting the biological
assays (data not shown). We have not found other data in the literature on the copper contents
after solidꢀphase CuAAC (recently reviewed in Castro et al. in Ref.(45)) that we could compare
with our data. We are now also exploring alternative methods for TES deprotection.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Biological Activity. Only subꢀlibraries C4 and C2 significantly activated IRꢀA at 0.1 mM.
However, the degree of receptor autophosphorylation by C4 and C2 was much lower than the
activation induced by 10 nM human insulin (Figure 3); moreover, it was not convincingly
(possibly with the exception of weak inhibition by C4) accompanied by the inhibition of insulin
binding (Figure S3). This result could indicate that the compounds bind to another site on the
receptor than the insulin binding site or that the binding assay is less sensitive than our receptor
autophosphorylation assay. However, the identification of any IR activators would also be
interesting for further development of insulin agonists. Therefore, we selected subꢀlibrary C4 and
prepared subꢀlibraries C4B1ꢀC1B10, but none of these mixtures activated IRꢀA or inhibited
insulin binding (Figures S4 and S5).
Subꢀlibrary C4 contains 100 different compounds, and it can be expected that the weak
biological activity of C4 could originate either from a sum of the activities of several (or many)
weaker compounds or from the activity of a single active compound, which would be present in
the tested C4 mixture at a concentration of 0.001 mM. The presence of only a single active
compound should be revealed by the stronger bioactivity (compared to C4) of one of the C4B1ꢀ
C4B10 subꢀlibraries. However, this stronger bioactivity was not observed, indicating that the
bioactivity of C4 is instead composed of the contributions from several weak compounds. We
attempted to isolate active compound(s) from C4 by preparative HPLC to verify this hypothesis.
This iterative process yielded two active fractions, C4-1 and C4-2 (Figure 4), in which we
17
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 18 of 32
1
2
3
4
5
6
7
8
9
detected m/z signals for compounds C4B2A10 and C4B3A6. Based on this finding, we prepared
these two compounds and a small “focused” library of 8 compounds (Table 2), where we
maintained the C4 azide and varied the B2, B3, A6 and A10 azides (to verify whether some
combinations of these “active” azides could result in more potent compounds). We should note
that the original library contained only compounds with three different azides, whereas the
focused library contained compounds with the same azide at two different positions. The testing
of this “focused” library showed that only compounds C4B2A10 and C4B3A6 that were
identified by MS were able to activate IRꢀA, but their activities were weaker than the original
C4-1 and C4-2 fractions (Figure 5A). This result could mean that these fractions are not pure and
still contain some other active compounds, which is indeed probable (see Figure 4). Moreover,
we found that compound C4B2A10 is able to displace insulin from its IRꢀA receptor (Figure 5B),
albeit only weakly and at a concentration of 0.1 mM, which could indicate that the compound at
least partially binds to the insulin binding site on IR. We can only speculate why the activities of
compounds C4B2A10 and C4B3A6 were not manifested as significant activities in the C4B2 and
C4B3 mixtures. The reason may be the weak activities of C4B2A10 and C4B3A6, which are
only “visible” at 0.1 mM; if these compounds displayed some activity in the C4B2 and C4B3
mixtures (also tested at 0.1 mM), their concentration would be 0.01 mM and would be too low to
elicit IRꢀA phosphorylation and detect binding. This finding rather supports our hypothesis (see
above) that the apparent biological activity of the C4 mixture (containing 100 compounds) could
be due to a simultaneous additive effect of several or many compounds with weak activities
(including C4B2A10 and C4B3A6 representing probably more active compounds among them).
However, it is possible that some other phenomenon is behind the loss of activity. For example,
we cannot exclude the presence of some more active byꢀproduct in C4 (nonꢀidentified), which we
omitted during purification and which was not present in C4B1ꢀC4B7.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In summary, this is our first attempt to develop nonꢀpeptide insulin mimetics. We designed a
combinatorial library of triꢀfunctional compounds that were derivatized with 30 different azides,
which should mimic the IRꢀbinding epitopes in insulin and bind to and activate IR. Despite the
unsatisfactory biological results, this study represents a unique example of targeting IR with a set
of nonꢀpeptide compounds that were specifically designed and synthesized for this purpose.
Other published attempts at identifying insulin mimetics were based on molecular modeling
and/or screening of virtual or inꢀhouse libraries of compounds,26ꢀ30 phage display peptide
18
ACS Paragon Plus Environment

Page 19 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
libraries31ꢀ33 or oligonucleotide aptamers.34 The reason for the lack of significant biological
activity of our library of 1,000 compounds derived from scaffold I may be the selection of
inappropriate azides, the low diversity of the library or the “incompatibility” of the scaffold I
structure with the IR receptor. However, considering the high sensitivity of insulin receptor to
modifications of insulin structure, this result is not completely surprising and confirms the great
difficulty of the task. Insulin is a 51 amino acid hormone, and the deletion of only eight Cꢀ
terminal amino acids of the Bꢀchain leads to an approximately 2,500ꢀfold decrease in the binding
affinity from 0.2 nM to 500 nM (nonꢀpublished data from our laboratory). Moreover, some
structural modifications in the Bꢀchain, Cꢀterminal segment of insulin20 or even some single
amino acid modifications in insulin, e.g., at position B24 (Ref.17), B8 (Ref.19) or A2 (Ref.46) have
similar disruptive effects on insulin binding activity. However, we still believe that the idea of
nonꢀpeptide activators of the insulin receptor is realistic and we are encouraged by the result
reported by Schlein et al.27, who identified the small molecule thymolphtalein (MW 430) as an
agonist that displaced insulin from IR (Kd approximately 10ꢀ6 M). Moreover, we believe that the
selection of more convenient ligands (e.g., peptides or peptidomimetics) for modifying scaffold I
and/or a larger sized library could lead to more valuable biological results. Nevertheless, despite
the negative biological results, this work has enabled us to test our synthetic methodology and to
prove its robustness and reliability for the synthesis of combinatorial libraries on the solid phase
using our homeꢀmade library of azides with varying functionalities.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
EXPERIMENTAL METHODS
Chemistry
General Information. Unless stated otherwise, the reagents and solvents used in this study
were obtained from commercial suppliers and used without purification. The solvents were
evaporated at 55 °C and 2 kPa, and the products were dried over sodium hydroxide at r. t. and
100 Pa. Analytical RPꢀHPLC chromatography of the compounds was performed on a Waters
HPLC system (Waters 1525 Binary HPLC Pump and Waters 2787 Dual λ Absorbance Detector)
using a Nucleosil 120ꢀ5 C8 column (250 x 4.6 mm, 5 ꢁm, MachereyꢀNagel) at a flow rate of 1
mL/min. The compounds were detected at 218 and 254 nm. Preparative RPꢀHPLC
chromatography of the compounds was conducted on the same Waters HPLC system using a
Vydac 214TP101522 C4 column (250 × 22 mm, 10–15 ꢁm, Columbia, MD, USA) at a flow rate
19
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 20 of 32
1
2
3
4
5
6
7
8
9
of 9 mL/min. The detection was the same as for the analytical HPLC. MS spectra were obtained
on an LTQꢀOrbitrap XL FTMS mass spectrometer (Thermo Fisher, Bremen, Germany) in
electrospray ionization mode (Waters) or, in the case of HRMS (EI), on a GCT Premier (Waters).
19F NMR spectra were measured on a Bruker AVANCEꢀ500 instrument at 470.4 MHz. For the
quantitative analyses, the 19F proton decoupled spectra were measured with suppression of
heteronuclear NOE during a relaxation delay (3 s). A known amount of BrCF2COOEt was added
to the measured solution and used as an internal standard. The copper contents were determined
by ICP with a Spectro Arcos ICPꢀOES analyzer (plasma power: 1450 W; coolant flow Ar: 1.0
L/min; auxiliary flow: 0.8 L/min; nebulizer flow: 0.75 L/min; sample introduction system: Mod
Lichte; cyclonic spray chamber; matrix: 4% HNO3 (w/w); and internal standard Y (2 mg/kg)).
The syntheses of the 30 azides (shown in Table 1) used for the derivatization of scaffold I are
described in detail in the Supplementary Information.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Combinatorial Syntheses on the Solid Phase. Loading onto the resin: Ten fritted
polypropylene syringes (20 mL) were loaded with Ramage ChemMatrix® resin (385 mg each,
200 µmol, 0.52 mmol/g, Aldrich lot BCBN4288V) with free amino groups. The resins were
swelled in MeOH, DCM, and DMF (10 minutes each) using 10 mL of solvent/g of resin. The
resins were then washed with 3×5 mL of DMF (1 min each). The following freshly made
solutions were added to each resin: a 0.1 M solution of scaffold I in DMF (1 mL, 100 µmol), a
0.2 M solution of PyBROP in DMF (1 mL, 200 µmol), and a 0.4 M solution of DIPEA in DMF
(1 mL, 400 µmol). The 10 syringes were stirred (at 360° vertically) at room temperature for 16 h.
The resins were finally washed with 3×5 mL of DMF, MeOH, DCM, and DMF (1 min each).
First CuAAC: The resins in the ten syringes were washed with 3×5 mL of MeOH and
tBuOH/water (1:1, v/v) (1 min each). The following freshly made solutions were sequentially
added to the resins: i) a 0.5 M solution of azide A in tBuOH/water (1:1, v/v) (1 mL, 500 µmol),
ii) a 0.05 M solution of sodium ascorbate in tBuOH/water (1:1, v/v) (1 mL, 50 µmol), and iii) a
0.01 M solution of copper(II) sulfate pentahydrate in tBuOH/water (1:1, v/v) (1 mL, 10 µmol).
Two couplings were performed, each for 16 h at room temperature. Between couplings, the resins
were washed with 3×5 mL of tBuOH/water (1:1, v/v) (1 min each). The resins were finally
washed with 3x5 mL of tBuOH/water (1:1, v/v), DMF, H2O, MeOH, DCM, and DMF (1 min
each).
20
ACS Paragon Plus Environment

Page 21 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
Note: Aqueous solutions of sodium ascorbate (0.5 M) and copper(II) sulfate pentahydrate
(0.1 M) were first prepared and then diluted in tBuOH/water to prepare the sodium ascorbate and
copper(II) sulfate pentahydrate solutions in tBuOH/water (1:1, v/v).
Split and mix procedure: The resins were transferred into a 1 L beaker with DMF. The resin
was suspended in DMF and gently shaken on a moving plate for 30 minutes. The resin was
transferred into a tarred fritted Buchner funnel and the DMF was removed by suction. The resin
was then split into ten equal portions and transferred to ten fritted polypropylene syringes.
One-pot TES deprotection and second CuAAC: The resins were washed with 3×5 mL of
MeOH and tBuOH/water (1:1, v/v) (1 min each). The following freshly made solutions were
sequentially added to the resins: i) a 0.25 M solution of azide B in tBuOH/water (1:1, v/v) (2 mL,
500 µmol), ii) a 0.5 M solution of sodium ascorbate in tBuOH/water (1:1, v/v) (2 mL, 1000
µmol), and iii) a 0.25 M solution of copper(II) sulfate pentahydrate in tBuOH/water (1:1, v/v) (2
mL, 500 µmol). Two couplings were performed for 16 h and 5 h at room temperature. Between
couplings, the resins were washed with 3×5 mL of tBuOH/water (1:1, v/v), AcOH (to help
removing copper salts), H2O, DMF, DCM, MeOH and tBuOH/H2O (1:1, v/v) (1 min each). The
resins were finally washed with 3x5 mL of tBuOH/water (1:1, v/v), AcOH, H2O, MeOH, DCM
and DMF (1 min each).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Note: It is necessary to first prepare the aqueous solutions to dissolve the sodium ascorbate
and copper(II) sulfate pentahydrate in tBuOH/water (1:1, v/v). Hence, we first prepared aqueous
solutions of sodium ascorbate (1 M) and copper(II) sulfate pentahydrate (0.5 M), which were
then diluted with tBuOH.
TIPS deprotection: The resins were washed with 3×5 mL of DMF (1 min each). A solution
of TBAF (500 µL of a 1 M solution in DMF, 500 µmol) in DMF (3 mL) was added to each
reaction. Three treatments were performed, each for 1.5 h at room temperature. Between each
treatment, the resins were washed with 3×5 mL of DMF (1 min each). The resins were finally
washed with 3×5 mL of DMF, H2O, AcOH, H2O, MeOH, DCM and DMF (1 min each).
Third CuAAC: The resins were washed with 3×5 mL of MeOH and tBuOH/water (1:1, v/v)
(1 min each). The following freshly made solutions were sequentially added to the resins: i) a 0.5
M solution of azide C in tBuOH/water (1:1, v/v) (1 mL, 500 µmol), ii) a 0.05 M solution of
sodium ascorbate in tBuOH/water (1:1, v/v) (1 mL, 50 µmol), and iii) a 0.01 M solution of
copper(II) sulfate pentahydrate in tBuOH/water (1:1, v/v) (1 mL, 10 µmol). Two couplings were
21
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 22 of 32
1
2
3
4
5
6
7
8
9
performed, each for 16 h at room temperature. Between the two couplings, the resins were
washed with 3×5 mL of tBuOH/water (1:1, v/v) (1 min each). The resins were finally washed
with 3x5 mL of tBuOH /water (1:1, v/v), DMF, H2O, DMF, DCM, MeOH and DCM (1 min
each).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Cleavage from the resin: The resins were transferred to fritted reactors with DCM and dried
over NaOH in desiccators for 1 h. Each resin was then treated with a 5% TFA/DCM solution (10ꢀ
15 mL) for 30 minutes. The resins were washed with 3×5 mL of 5% TFA/DCM (1 min each).
The resulting solutions (filtrates) were transferred to tarred 100 mL flasks and evaporated. The
resins were treated a second time with a 5% TFA/DCM solution (10ꢀ15 mL, 30 minutes) and
washed with 3×5 mL of 5% TFA/DCM and AcOH (1 min each). All cleavage and washing
solutions (filtrates) for a given compound were transferred to the corresponding 100 ml flask and
evaporated to dryness. The yields (in mg) obtained for each subꢀlibrary or single compound are
shown in Table S2.
Boc/tBu deprotection: A 50 % TFA/DCM solution with 2% TIS (5 mL in total) was added to
the 100 mL flasks containing the subꢀlibraries or single compounds. The resulting mixtures were
stirred at room temperature for 2 h and evaporated to dryness.
Diethyl ether extraction: Diethyl ether (20 mL) was added to each 100 mL flask. The
resulting suspensions were sonicated for 5 minutes. The solutions were transferred to 50 mL
falcons and centrifuged (5 minutes, 7,200 rpm, 20 °C). The supernatants were discarded and the
remaining solids were transferred back to the corresponding 100 mL flask using ACN/H2O
solutions and lyophilized. We wished to verify that no desired product was extracted into the
diethyl ether during the extraction. We chose sublibrary C2 because it is the most lipophilic subꢀ
library. Therefore, we performed HPLC analyses of subꢀlibrary C2 before and after diethyl ether
extraction, as well as an MS analysis of the diethyl ether extract. These analyses revealed that no
major loss of the compounds occurred during the diethyl ether extraction (Figures S10 and S11).
Treatment with gaseous hydrogen sulfide: After the diethyl ether extraction, the subꢀlibraries
were dissolved in 30ꢀ50% ACN/H2O (3 mL). Gaseous hydrogen sulfide (H2S) was blown at the
surface of the solutions while stirring. The solutions were allowed to stand for 30 minutes and
slowly turned from light brown to black, which reflected the formation of the copper sulfide
precipitate. The mixtures were then filtered through a Rotilabo® PVDF filter (0.22 ꢁm),
lyophilized and weighed. The yields (in mg) are presented in Table S3. Sublibrary C1 (20.8 mg)
22
ACS Paragon Plus Environment

Page 23 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
was dissolved in 40% ACN/H2O (3 mL) and the solution was analyzed by HPLC to verify that no
compounds were lost during this process. Then, the solution was treated with hydrogen sulfide, as
described above. After filtration, the filtrate was again analyzed by HPLC and we verified that
the two HPLC chromatograms were identical (Figure S12).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
CAUTION: Hydrogen sulfide is a toxic gas that must only be manipulated in the fume hood.
19
Quantification by F NMR. The final subꢀlibraries were stored as frozen solutions in 40%
ACN/H2O at ꢀ80°C. For biological testing, a precise volume of each of these solutions was
transferred to an Eppendorf tube and lyophilized. The dry samples were transferred to NMR
tubes using d6DMSO (3×0.15 mL). A solution of commercial ethyl bromodifluoroacetate (100
µL of a 0.1 M solution in d6DMSO) was added to the NMR tube as an internal standard.10
Integration over the region corresponding to the CF2 signals and comparison to the standard CF2
signal allowed us to quantify the compounds containing the scaffold. Consequently, the
concentrations of the stock solutions in the scaffoldꢀderived compounds could be calculated. As
discussed above, the main impurities found in the single compounds are due to double
incorporation during the first CuAAC (the same azide clicked into the arms A and B). Although
we were not able to characterize such impurities in the library mixtures, they should be present.
As such impurities contain the CF2ꢀlabelled arm, the concentrations of the libraries in desired
compounds is probably slightly over evaluated.
Synthesis of the focused library. Compounds C4B2A10, C4B3A6, C4A10A10,
C4A6A10, C4B3A10, C4A6A6, C4B2A6, C4B2B2, C4B3B2 and C4B3B3 were synthesized on
the solid phase as described above (section 1.2), starting with 100 µmol of scaffold I. The RPꢀ
HPLC chromatograms of the crude compounds are shown in the Supporting Information (Figures
S8 and S9). The crude compounds were then purified by preparative RPꢀHPLC and the yields
over the whole process were calculated (Table S1). The purities of the HPLCꢀpurified
compounds were above 95 % in all cases.
Analyses of the libraries. Given the complexity of subꢀlibraries C1-C10, these subꢀlibraries
were not analyzed by MS or LCꢀMS. However, subꢀlibrary C4B7 was analyzed by LCꢀMS
(Figure S2), confirming the presence of the expected 10 compounds, C4B7A1-C4B7A10.
Compounds C4B7A1-C4B7A10 and the compounds from the focused library (see Table 2) were
analyzed by RPꢀHPLC (for the crude samples, see Figure S1) and MS/HRMS (for crude/pure
samples, see Table S1). In each library, the samples that maintained a dark coloration after
23
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 24 of 32
1
2
3
4
5
6
7
8
9
treatment with hydrogen sulfide were submitted to ICP analyses to determine their copper
contents. As all the crude mixtures of the compounds from the focused library (Table 2) visually
appeared to be free of copper, we randomly selected the crude compounds C4B2A10, C4B3A6
and C4B3B2, which gave some response in the IRꢀA phosphorylation assay, and determined their
copper contents.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Biology
Assays. The cell lines that were employed for the receptor binding studies and stimulation of
autophosphorylation were the human IMꢀ9 lymphocytes (for IRꢀA, purchased from ATCC) and
mouse embryonic fibroblasts with a targeted disruption of IGFꢀ1R that were stably transfected
with either human IRꢀA (Rꢀ/IRꢀA) or human IGFꢀ1R (R+39) and kindly provided by A. Belfiore
(University of Magna Graecia, Catarzano, Italy) and R. Baserga (Thomas Jefferson University,
Philadelphia, PA). The cells were grown and treated as previously described.21
Stock 1 mM solutions of the libraries and single compounds were prepared in 100 mM
Hepes/NaOH buffer, pH 7.5 and the concentrations were determined based on quantitative 19F
NMR analyses as described previously.10 The concentrations of the compounds from the focused
library (C4B2A10, C4B3A6, C4A10A10, C4A6A10, C4B3A10, C4A6A6, C4B2A6, C4B2B2,
C4B3B2 and C4B3B3) were prepared with the HPLC purified products based on their molecular
weights. The RPꢀHPLCꢀisolated fractions (C4-1 and C4-2) were dissolved in the same buffer at 1
19
mg/mL (w/v) (because not enough material was available for F NMR analysis), which should
roughly correspond to 1 mM solutions of the single compounds (molecular weights ranging from
800 Da to 1,200 Da). As needed, the solutions were heated and sonicated to support solubility.
All libraries and single compounds were tested for binding to human IRꢀA (in the
membranes of human IMꢀ9 lymphocytes) and to human IGFꢀ1R (in the membranes of mouse
embryonic fibroblasts) at a single concentration (0.1 mM or 0.1 mg/mL) and at least three times.
Eventually, 0.01 mM concentrations were checked. The binding assays were performed as
recently described by Krizkova et al.21
Mouse fibroblasts (Rꢀ/IRꢀA) and (R+39) were stimulated with the libraries, single compounds
and RPꢀHPLCꢀisolated fractions at concentrations of 0.1 mM or 0.1 mg/mL for 10 min in 24ꢀwell
plates. The cells were lysed and immunoblotted onto a PVDF membrane, as described by
Krizkova et al.21 The membranes were probed with an antiꢀphosphoꢀIGFꢀ1R βꢀsubunit (Tyr1131)
24
ACS Paragon Plus Environment

Page 25 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
/IR βꢀsubunit (Tyr1146) antibody (Cell Signaling Technology) to detect the extent of receptor
autophosphorylation. The cells in the control wells were treated with either buffer alone or with
10 nM insulin (Sigma). The series of samples were tested four times using different batches of
cells. The signal density generated by a sample was expressed as the contribution of
phosphorylation relative to the signal from 10 nM insulin in the same experiment. The
significance of an increase in the signal density generated by a particular sample (mean ±
standard error of the mean (SEM), n = 4) compared with the control (unstimulated cells) was
calculated using oneꢀway analysis of variance.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Determination of the impact of copper. Copper(II) sulfate pentahydrate (2.5 mg, 10 µmol)
was dissolved in 3 mL of 100 mM Hepes/NaOH buffer, pH 7.5. Gaseous hydrogen sulfide (H2S)
was blown at the surface of the solutions while stirring. The solution was allowed to stand for 30
minutes and completely turned black. After lyophilization, the black residue was redissolved
(suspension) in MQ water (virtual copper concentration of 3 mM) and diluted tenꢀfold for the
assays (final virtual copper concentration of 300 µM). This sample was tested alone or with
insulin (at 10 nM) and did not show false positive or negative results (not shown).
ASSOCIATED CONTENT
Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI:
Syntheses of azides A1ꢀC10 and the scaffold I. Description of methods to remove copper.
Figures showing RPꢀHPLC analyses of the subꢀlibrary C4B7 and compounds C4B7A1ꢀ
C4B7A10. Figure showing LCꢀMS analysis of C4B7. Table showing analytical data for
selected subꢀlibraries and compounds. Figures showing biological activities of subꢀlibraries
and single compounds. Figures showing HPLC traces of the compounds from the focused
library. Tables showing crude yields of subꢀlibraries C1-C10 and C4B1-C4B10 and
compounds C4B7A1ꢀC4B7A10. Figures showing HPLC traces and mass spectra of the C2
subꢀlibrary before and after the extraction with diethyl ether. Figure showing HPLC traces
of the subꢀlibrary C1 before and after treatment with hydrogen sulfide. Supplementary
references. (PDF).
AUTHOR INFORMATION
25
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 26 of 32
1
2
3
4
Corresponding Author
5
6
*Phone: +420 220183ꢀ441. Eꢀmail: jiracek@uochb.cas.cz
Authors’ Contributions
7
8
BJ and JP contributed equally to this work; BJ, JP, VV and IS conceived and performed the
experiments, analyzed the data and coꢀwrote the manuscript; MC, MC and LZ performed the
experiments and analyzed the data; MB measured and analyzed the NMR spectra; JJ conceived
the study, analyzed the data and coꢀwrote the manuscript. All authors have given their final
approval to the final version of the manuscript.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Funding
This work was supported by the Grant Agency of the Czech Republic (grant 14ꢀ17305S, to J.J.),
by the Academy of Sciences of the Czech Republic (Research Project RVO:6138963, support to
the Institute of Organic Chemistry and Biochemistry and postdoctoral stipend to B.F.), by the
Institute of Organic Chemistry and Biochemistry (IOCB) and the Czech Academy of Sciences
(postdoctoral stipend to B.F.) and partially by the Medical Research Council (Grant
MR/K000179/1).
Notes
The authors declare that there are no competing financial interests.
ACKNOWLEDGEMENTS
The authors would like to thank Dr. Stanislava Matějková for performing the ICP analyses.
ABBREVIATIONS
CuAAC, copper(I)ꢀcatalyzed azide–alkyne cycloaddition; ICP, inductively coupled plasma; IGFꢀ
1R, insulinꢀlike growth factor 1 receptor; IR, insulin receptor; IRꢀA, isoform A of the insulin
receptor.
REFERENCES
(1)
Milroy, L. G., Grossmann, T. N., Hennig, S., Brunsveld, L., and Ottmann, C. Modulators
of proteinꢀprotein interactions, Chem. Rev. 2014, 114, 4695ꢀ4748.
(2)
Wells, J. A., and McClendon, C. L. Reaching for highꢀhanging fruit in drug discovery at
proteinꢀprotein interfaces, Nature 2007, 450, 1001ꢀ1009.
26
ACS Paragon Plus Environment

Page 27 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
(3)
Robinson, J. A., DeMarco, S., Gombert, F., Moehle, K., and Obrecht, D. The design,
structures and therapeutic potential of protein epitope mimetics, Drug Discov. Today
2008, 13, 944ꢀ951.
(4)
(5)
(6)
(7)
(8)
(9)
Fletcher, S., and Hamilton, A. D. Protein surface recognition and proteomimetics: mimics
of protein surface structure and function, Curr. Opin. Chem. Biol. 2005, 9, 632ꢀ638.
Jones, S., and Thornton, J. M. Principles of proteinꢀprotein interactions, Proc. Natl. Acad.
Sci. U.S.A. 1996, 93, 13ꢀ20.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Lo Conte, L., Chothia, C., and Janin, J. The atomic structure of proteinꢀprotein
recognition sites, J. Mol. Biol. 1999, 285, 2177ꢀ2198.
Cunningham, B. C., and Wells, J. A. Highꢀresolution epitope mapping of hghꢀreceptor
interactions by alanineꢀscanning mutagenesis, Science 1989, 244, 1081ꢀ1085.
Moreira, I. S., Fernandes, P. A., and Ramos, M. J. Hot spotsꢀA review of the proteinꢀ
protein interface determinant aminoꢀacid residues, Proteins 2007, 68, 803ꢀ812.
Beal, D. M., and Jones, L. H. Molecular scaffolds using multiple orthogonal conjugations:
Applications in chemical biology and drug discovery, Angew. Chem. Int. Edit. 2012, 51,
6320ꢀ6326.
(10) Vanek, V., Picha, J., Fabre, B., Budesinsky, M., Lepsik, M., and Jiracek, J. The
development of a versatile trifunctional scaffold for biological applications, Eur. J. Org.
Chem. 2015, 3689ꢀ3701.
(11) Ward, C. W., Menting, J. G., and Lawrence, M. C. The insulin receptor changes
conformation in unforeseen ways on ligand binding: Sharpening the picture of insulin
receptor activation, Bioessays 2013, 35, 945ꢀ954.
(12) Zakova, L., Barth, T., Jiracek, J., Barthova, J., and Zorad, S. Shortened insulin analogues:
Marked changes in biological activity resulting from replacement of TyrB26 and Nꢀ
methylation of peptide bonds in the Cꢀterminus of the Bꢀchain, Biochemistry 2004, 43,
2323ꢀ2331.
(13) Zakova, L., Brynda, J., AuꢀAlvarez, O., Dodson, E. J., Dodson, G. G., Whittingham, J. L.,
and Brzozowski, A. M. Toward the insulinꢀIGFꢀI intermediate structures: Functional and
structural properties of the [Tyr(B25)NMePhe(B26)] insulin mutant, Biochemistry 2004,
43, 16293ꢀ16300.
27
ACS Paragon Plus Environment

ACS Combinatorial Science
Page 28 of 32
1
2
3
4
5
6
7
8
(14) Zakova, L., Kazdova, L., Hanclova, I., Protivinska, E., Sanda, M., Budesinsky, M., and
Jiracek, J. Insulin analogues with modifications at position B26. Divergence of binding
affinity and biological activity, Biochemistry 2008, 47, 5858ꢀ5868.
(15) Jiracek, J., Zakova, L., Antolíková, E., Watson, C. J., Turkenburg, J. P., Dodson, G. G.,
and Brzozowski, A. M. Implications for the active form of human insulin based on the
structural convergence of highly active hormone analogues., Proc. Natl. Acad. Sci. U.S.A.
2010, 107, 1966ꢀ1970.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(16) Antolikova, E., Zakova, L., Turkenburg, J. P., Watson, C. J., Hanclova, I., Sanda, M.,
Cooper, A., Kraus, T., Brzozowski, A. M., and Jiracek, J. Nonꢀequivalent role of interꢀ
and intramolecular hydrogen bonds in the insulin dimer interface, J. Biol. Chem. 2011,
286, 36968ꢀ36977.
(17) Zakova, L., Kletvikova, E., Veverka, V., Lepsik, M., Watson, C. J., Turkenburg, J. P.,
Jiracek, J., and Brzozowski, A. M. Structural Integrity of the B24 Site in Human Insulin Is
Important for Hormone Functionality, J. Biol. Chem. 2013, 288, 10230ꢀ10240.
(18) Krizkova, K., Veverka, V., Maletinska, L., Hexnerova, R., Brzozowski, A. M., Jiracek, J.,
and Zakova, L. Structural and functional study of the GlnB22ꢀinsulin mutant responsible
for maturityꢀonset diabetes of the young, Plos One 2014, 9, e112883.
(19) Kosinova, L., Veverka, V., Novotna, P., Collinsova, M., Urbanova, M., Moody, N. R.,
Turkenburg, J. P., Jiracek, J., Brzozowski, A. M., and Zakova, L. Insight into the
structural and biological relevance of the T/R transition of the Nꢀterminus of the Bꢀchain
in human insulin, Biochemistry 2014, 53, 3392ꢀ3402.
(20) Vikova, J., Collinsova, M., Kletvikova, E., Budesinsky, M., Kaplan, V., Zakova, L.,
Veverka, V., Hexnerova, R., Avino, R. J. T., Strakova, J., Selicharova, I., Vanek, V.,
Wright, D. W., Watson, C. J., Turkenburg, J. P., Brzozowski, A. M., and Jiracek, J.
Rational steering of insulin binding specificity by intraꢀchain chemical crosslinking,
Scientific Reports 2016, 6, 19431.
(21) Krizkova, K., Chrudinova, M., Povalova, A., Selicharova, I., Collinsova, M., Vanek, V.,
Brzozowski, A. M., Jiracek, J., and Zakova, L. Insulinꢀinsulinꢀlike growth factors hybrids
as molecular probes of hormone:receptor binding specificity, Biochemistry 2016, 55,
2903ꢀ2913.
28
ACS Paragon Plus Environment

Page 29 of 32
ACS Combinatorial Science
1
2
3
4
5
6
7
8
9
(22) Menting, J. G., Whittaker, J., Margetts, M. B., Whittaker, L. J., Kong, G. K. W., Smith, B.
J., Watson, C. J., Zakova, L., Kletvikova, E., Jiracek, J., Chan, S. J., Steiner, D. F.,
Dodson, G. G., Brzozowski, A. M., Weiss, M. A., Ward, C. W., and Lawrence, M. C.
How insulin engages its primary binding site on the insulin receptor, Nature 2013, 493,
241ꢀ245.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
(23) Menting, J. G., Yang, Y. W., Chan, S. J., Phillips, N. B., Smith, B. J., Whittaker, J.,
Wickramasinghe, N. P., Whittaker, L. J., Pandyarajan, V., Wan, Z. L., Yadav, S. P.,
Carroll, J. M., Strokes, N., Roberts, C. T., IsmailꢀBeigi, F., Milewski, W., Steiner, D. F.,
Chauhan, V. S., Ward, C. W., Weiss, M. A., and Lawrence, M. C. Protective hinge in
insulin opens to enable its receptor engagement, Proc. Natl. Acad. Sci. U.S.A. 2014, 111,
E3395ꢀE3404.
(24) Zakova, L., Kletvikova, E., Lepsik, M., Collinsova, M., Watson, C. J., Turkenburg, J. P.,
Jiracek, J., and Brzozowski, A. M. Human insulin analogues modified at the B26 site
reveal a hormone conformation that is undetected in the receptor complex, Acta
Crystallogr. D 2014, 70, 2765ꢀ2774.
(25) De Meyts, P. Insulin/receptor binding: The last piece of the puzzle? Bioessays 2015, 37,
389ꢀ397.
(26) Zhang, B., Salituro, G., Szalkowski, D., Li, Z. H., Zhang, Y., Royo, I., Vilella, D., Diez,
M. T., Pelaez, F., Ruby, C., Kendall, R. L., Mao, X. Z., Griffin, P., Calaycay, J., Zierath,
J. R., Heck, J. V., Smith, R. G., and Moller, D. E. Discovery of a small molecule insulin
mimetic with antidiabetic activity in mice, Science 1999, 284, 974ꢀ977.
(27) Schlein, M., Ludvigsen, S., Olsen, H. B., Andersen, A. S., Danielsen, G. M., and
Kaarsholm, N. C. Properties of small molecules affecting insulin receptor function,
Biochemistry 2001, 40, 13520ꢀ13528.
(28) Nakamura, T., Takahashi, H., Takahashi, M., Shimba, N., Suzuki, E., and Shimada, I.
Direct determination of the insulinꢀinsulin receptor interface using transferred crossꢀ
saturation experiments, J. Med. Chem. 2010, 53, 1917ꢀ1922.
(29) Tan, C., Wei, L. H., Ottensmeyer, F. P., Goldfine, I., Maddux, B. A., Yip, C. C., Batey, R.
A., and Kotra, L. P. Structureꢀbased de novo design of ligands using a threeꢀdimensional
model of the insulin receptor, Bioorg. Med. Chem. Lett. 2004, 14, 1407ꢀ1410.
29
ACS Paragon Plus Environment