Positional Scanning Identifies the Molecular Determinants of a High Affinity Multi-Leucine Inhibitor for Furin and PACE4

Article abstract of DOI:10.1021/acs.jmedchem.6b01499

The proprotein convertase family of enzymes includes seven endoproteases with significant redundancy in their cleavage activity. We previously described the peptide Ac-LLLLRVK-Amba that displays potent inhibitory effects on both PACE4 and prostate cancer cell lines proliferation. Herein, the molecular determinants for PACE4 and furin inhibition were investigated by positional scanning using peptide libraries that substituted its leucine core with each natural amino acid. We determined that the incorporation of basic amino acids led to analogues with improved inhibitory potency toward both enzymes, whereas negatively charged residues significantly reduced it. All the remaining amino acids were in general well tolerated, with the exemption of the P6 position. However, not all of the potent PACE4 inhibitors displayed antiproliferative activity. The best analogues were obtained by the incorporation of the Ile residue at the P5 and P6 positions. These substitutions led to inhibitors with increased PACE4 selectivity and potent antiproliferative effects.

Full text of DOI:10.1021/acs.jmedchem.6b01499

Subscriber access provided by HACETTEPE UNIVERSITESI KUTUPHANESI
Article
Positional Scanning Identifies the Molecular Determinants of
a High Affinity Multi-Leucine Inhibitor for Furin and PACE4
Izabela Ma#uch, Christine Levesque, Anna Kwiatkowska, Frédéric Couture, Kévin
Ly, Roxane Desjardins, Witold A. Neugebauer, Adam Prahl, and Robert Day
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01499 • Publication Date (Web): 13 Mar 2017
Downloaded from http://pubs.acs.org on March 14, 2017
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.
Journal of Medicinal Chemistry 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 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Positional Scanning Identifies the Molecular Determinants of
a High Affinity Multi-Leucine Inhibitor for Furin and PACE4
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
♯
a
♯b,c
b,c*
b,c
Izabela Małuch , Christine Levesque Anna Kwiatkowska , Frédéric Couture ,
b,c
b,c
,
b
a
Kévin Ly , Roxane Desjardins Witold A. Neugebauer , Adam Prahl and Robert
b,c*
Day
a
Department of Organic Chemistry, Faculty of Chemistry, University of Gdańsk, 80ꢀ308
Gdańsk, Poland
b
Institut de Pharmacologie de Sherbrooke, Université de Sherbrooke, 3001 12e Avenue
Nord, Sherbrooke, J1H 5N4 Canada
c
Département de Chirurgie/Urologie, Centre Hospitalier Université de Sherbrooke, 3001
2e Avenue Nord, Sherbrooke, J1H 5N4 Canada Sherbrooke, Canada
1
♯
Authors contributed equally to this work: Izabela Małuch & Christine Levesque
* Corresponding Authors:
Anna Kwiatkowska (A.K.), telephone: 819ꢀ821ꢀ8000 # 70110, eꢀmail:
anna.kwiatkowska@usherbrooke.ca
Robert Day (R.D.), telephone: 819 821ꢀ8000 # 75428, eꢀmail:
robert.day@usherbrooke.ca
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
The proprotein convertase family of enzymes includes seven endoproteases with
significant redundancy in their cleavage activity. We previously described the peptide
AcꢀLLLLRVKꢀAmba that displays potent inhibitory effects on both PACE4 and prostate
cancer cell lines proliferation. Herein, the molecular determinants for PACE4 and furin
inhibition were investigated by positional scanning using peptide libraries that substituted
its leucine core with each natural amino acid. We determined that the incorporation of
basic amino acids led to analogs with improved inhibitory potency towards both
enzymes, whereas negatively charged residues significantly reduced it. All the remaining
amino acids were in general well tolerated, with the exemption of the P6 position.
However, not all of the potent PACE4 inhibitors displayed antiproliferative activity. The
best analogs were obtained by the incorporation of the Ile residue at the P5 and P6
positions. These substitutions led to inhibitors with increased PACE4 selectivity and
potent antiproliferative effects.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Page 3 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Introduction
The family of enzymes known as the proprotein convertases (PCs) includes seven kexinꢀ
like members, namely furin, PC1/3, PC2, PC4, PC5/6, PC7 and PACE4, which process
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
P4
P3
P2
P1
substrates that have the canonical consensus motif R ꢀX ꢀ(K/R) ꢀR (X represents any
1
ꢀ3
amino acid) . The enzymes PC5/6, PC7 and PACE4 are found within constitutive
secretory pathways and display a broad tissue distribution pattern, whereas furin is
1
, 4ꢀ6
ubiquitously expressed
. The intracellular localization and trafficking of each PC has
broad implications with regard to the final effects on precursor processing. Indeed, while
distinct in vivo functions can be ascribed to individual PCs, significant functional in vitro
7
ꢀ9
redundancy is also prevalent in this family of enzymes . The active site of PCs displays
3
a high level of homology . However, PCs have very different Cꢀterminal domains that
results in distinct intracellular localization and trafficking, and thus confers distinct
functions.
PCs have been implicated in the development and progression of various pathologies,
including neurodegenerative disorders, osteoarthritis, bacterial and viral infections and
1
, 2, 10
cancer
. For this reason, the exploration of PCs as druggable targets remains an area
of significant interest and holds the promise of great potential. To have a pharmacological
impact on such complex events, it is essential to better understand the molecular
determinants of PCs and their inhibitors.
The determination of the furin crystal structure and the development of PC homology
models have aided inhibitor development by providing critical information regarding the
3, 11ꢀ14
composition and disparities between the catalytic clefts
. These studies demonstrated
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
that subsites S1 to S4 are highly similar among all basic amino acid cleaving PCs, with
important differences appearing in the S5 and S6 subsites. It was therefore suggested that
3
selective inhibitors would require at least 6 residues . Recently, our research team
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
developed the first selective PACE4 inhibitor, the octapeptide with the following
15
structure AcꢀLLLLRVKRꢀNH . This inhibitor, named the MultiꢀLeu peptide
2
(
compound 1), displays a 20ꢀfold selectivity, with K s of 22 ± 6 nM for PACE4 and 430
i
1
5
±
10 nM for furin . This study confirmed that the S5ꢀS6 subsites, as well as the S7
subsite (and to a certain extent the S8 subsite), could be utilized to develop inhibitors that
distinguish between various PCs. Most notably, PACE4 is well inhibited by a peptide
with hydrophobic amino acids in these positions, while furin is not.
With regard to the potential clinical usefulness of such inhibitors, the peptide 1 exhibits
antiproliferative effects on the DU145 and LNCaP prostate cancer cell lines. Both cell
lines express high levels of PACE4, while another prostate cancer cell line, PC3, does not
express PACE4. PC3 cell proliferation is not affected by peptide 1. There are additional
lines of evidence demonstrating the important role of PACE4 in the progression of
1
6, 17
prostate cancer
. PACE4 is the only PC that is upregulated in malignant samples of
1
6, 18, 19
18, 19
prostate tissues, which was shown at both the mRNA
and protein levels
.
17
PACE4 also displays a unique role in cancer progression in vivo , supporting the
hypothesis that PACE4 inhibitors may be an important tool for prostate cancer therapy.
With this therapeutic indication in mind, it might be strategically important to develop
stable and high affinity PACE4 inhibitors with limited potency toward the other PCs.
These can be based on the backbone of peptide 1, with modifications using
ACS Paragon Plus Environment

Page 5 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
peptidomimetic strategies, which should result in compounds that will be optimal for in
vivo use.
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
To achieve this objective, structureꢀactivity relationship studies were performed to
2
0
evaluate potential modifications . Recent work has focused on the substitution of the
P1
Arg
residue of furin inhibitors with the unnatural arginine mimetic 4ꢀ
2
1ꢀ23
amidinobenzylamide (Amba), resulting in important gains in inhibitory potency
, and
demonstrating the favorable conformation of this mimetic for tight binding of the PC
14
active site . We have carried out similar work with our peptide 1 inhibitor, synthesizing
an Ambaꢀsubstituted version of this peptide (AcꢀLLLLRVKꢀAmba, peptide 2). We
confirmed an important increase (7ꢀfold) in inhibitory potency against PACE4 (Ki
20
PACE4: 3.1 ± 0.8 nM) . However, the introduction of the Amba moiety into inhibitor 1
also increased the inhibitory potency toward furin, which is likely due to the high affinity
of the Amba residue for furin subsite S1 (K furin: 4.3 ± 0.8 nM).
i
While the significant increase in potency is desirable, the lack of selectivity may
be problematic in regard to a therapeutic application of the inhibitor. Therefore, we
undertook a structureꢀactivity study using peptide 2 as a scaffold in order to explore
potential molecular determinants that could be substituted, while preserving potency and
improving selectivity. We designed a positional scan library substituting the P5ꢀP8
positions, while maintaining the important PC recognition motif in the P1ꢀP4 positions.
The library consists of 73 pure peptides (a reference analog (2) and 18 peptides for each
scan position, the P5 library: analogs 3 – 20, the P6 library: analogs 21 – 38, the P7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
library: analogs 39 – 56, the P8 library: analogs 57 – 74; Figure 1). Although our
primary indicator is the inhibition constant (K value), we also tested these peptides in
i
cellꢀbased assays using LNCaP and DU145 cell lines. Based on these results, we decided
further to evaluate the cell penetrating properties of the selected inhibitors.
Results
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Peptide synthesis
The first series of inhibitors (P5ꢀsubstituted peptides) was synthesized using 4ꢀFmocꢀ
hydrazinobenzoyl resin (Scheme 1). Briefly, the P8−P2 segments were simultaneously
assembled on the resin using standard Fmoc solidꢀphase peptide synthesis (SPPS). The
removal of the protected peptide from the resin was achieved in two steps, including
oxidation of the hydrazine moiety to diazene with copper (II) acetate followed by the
coupling of the protected Amba. The final compound was obtained after the sideꢀchain
deprotection with trifluoroacetic acid (TFA) and the liberation of the amidine group by
hydrogenation. Although this synthetic route afforded the desired compounds, the
laborious operations, difficulties in monitoring peptide cleavage from the resin and low
yields of the entire synthesis process (not higher than 20% after attempts to optimize the
oxidation process) motivated us to find another, more efficient synthetic route. The
remaining inhibitors were prepared by a combination of SPPS and solutionꢀphase
2
0
synthesis using 2ꢀchlorotritylꢀchloride resin as previously described (Scheme 2) . In
brief, the FmocꢀLys(Boc)ꢀOH was manually attached to the resin in the presence of N,Nꢀ
diisopropylethylamine (DIPEA) and then standard protocols for SPPS were employed to
synthesize the P8ꢀP2 segment. The protected peptide was cleaved under mild acidic
conditions followed by the coupling of protected 4ꢀamidinobenzylamine or 4ꢀ
ACS Paragon Plus Environment

Page 7 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
amidinobenzylamine·2HCl, and then final removal of all protecting groups. This
synthetic route proved to be robust, simple and reliable.
Fluorescent versions of the selected inhibitors were also obtained using this synthetic
route. In these cases, the P8ꢀP2 segment was Nꢀterminally modified with isothiocyanate
isomer I (FITC) group via βAla residue, which was used as a spacer. All peptides were
analyzed by liquid chromatography mass spectrometry with electrospray ionization
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
(
LCMS ESI ITꢀTOF) or highꢀresolution mass spectrometry (HRMS) and reversedꢀphase
highꢀperformance liquid chromatography (RPꢀHPLC). The purification was performed
using RPꢀHPLC and all analogs were obtained as lyophilized trifluoroacetic (TFA) salts.
The analytical data of all analogs are presented as supporting information in Table S1.
Enzymatic studies
To identify the molecular determinants for PC inhibition, a positional scanning library
was designed from the initial scaffold 2 (Figure 1). The library was screened against
PACE4 and furin and the K s were determined to compare the binding affinities of each
i
peptide in the library. All of the tested peptides showed an inhibitory effect at very low
concentrations comparable with the concentration of the enzyme used, therefore their Kis
2
4, 25
were calculated using the Morrison equation for reversible tightꢀbinding inhibition
.
As described previously, K s for peptide 2 are in the low nanomolar range for both
i
enzymes, with values of 3.1 ± 0.8 nM and 4.3 ± 0.8 nM for PACE4 and furin,
2
0
respectively .
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Position P5
The library screen comparing residues in the P5 position revealed a strong influence of
charge composition at this position (Figure 2A). Indeed, a major difference in PACE4
binding properties was observed for peptides containing positively charged residues as
analog 15 (K : 0.2 ± 0.1 nM), analog 10 (K : 2 ± 1 nM), analog 8 (K : 0.4 ± 0.3 nM)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
i
i
i
versus negatively charged residues as peptide 4 (K : 400 ± 90 nM), and peptide 5 (K : 46
i
i
±
9 nM), with a nearly 2000ꢀfold increase in K values between peptide 10 and analog 4.
i
The same trend was observed for furin, with K values between 0.7 ± 0.4 nM (analog 15)
i
and 230 ± 60 nM (analog 4). The high impact of charges within this position was also
2
2
previously described by Becker et al , whereas the addition of basic moieties in position
P5 increased the overall affinity of Ambaꢀcontaining peptides for furin.
With the exception of negatively charged amino acids, PACE4 tolerated a broad range of
amino acid substitutions at the P5 position, and uncharged residues presented K s in the
i
ꢀ9
1
0 M range. However, the Gly substituted peptide (analog 7) seemed unfavorable for
both PACE4 and furin, with K s values of 18 ± 6 nM and 12 ± 4 nM, respectively.
i
Because the Gly residue differs from other amino acid by its greater conformational
flexibility, these low values could indicate that the additional freedom in the peptide
chain destabilizes the molecular interactions between the peptide and active site cleft.
Furthermore, analog 9 with an Ile residue, seems to be unfavorable for furin (K : 29 ± 9
i
nM) and tolerated by PACE4 (K : 8 ± 3 nM).
i
ACS Paragon Plus Environment

Page 9 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
To identify possible substitutions that could result in a selective and potent inhibitor, a
selectivity index was calculated from the ratio between the furin and PACE4 inhibition
constants for each peptide (Figure 3A). As seen from these selectivity indexes, a few
residues offered an increase in selectivity when compared to Leu in the P5 position. Five
residues displayed at least a twoꢀfold selectivity for PACE4 over furin, namely His (2.3ꢀ
fold), Ile (3.4ꢀfold), Arg (3.5ꢀfold), Ser (2.0ꢀfold) and Trp (2.9ꢀfold).
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
Position P6
The evaluation of K values for the P6 peptide library demonstrated an attenuated
i
importance of positively charged amino acid residues for PACE4 in this position (Figure
2
B). Indeed, the inhibitory potency of analog 33 (K : 0.9 ± 0.5 nM), peptide 28 (K : 2.2 ±
i i
0
.6 nM) and compound 26 (K : 1.8 ± 0.4 nM) were less important in the P6 position than
i
for any other positions since 33 reached nearly subnanomolar inhibition, whereas 28 and
26 offered only nanomolar inhibition. The same trend was observed for furin, where only
inhibitor 33 (K : 0.50 ± 0.03 nM) demonstrated subnanomolar inhibition. Furthermore,
i
the K s obtained for the P6 library demonstrated an overall elevated inhibitory profile
i
compared to the P5 library. Whereas most peptides displayed K s below 10 nM toward
i
PACE4 in position P5, only 8 peptides from the P6 library displayed binding properties
ꢀ9
in the 10 M range, including inhibitors modified with basic residues (analogs 26, 28 and
33), Pro (peptide 31, K : 4 ± 1 nM), Val (peptide 36, K : 4 ± 2 nM), Gln (peptide 32, K :
i
i
i
4
.1 ± 0.7 nM), Tyr (peptide 38, K : 4.8 ± 0.2 nM) and Leu in the initial scaffold 2 (K : 3.1
i i
±
0.8 nM). Interestingly, we previously reported results from a partial combinatorial
library, where the presence of Leu in position P6 resulted in a strong affinity for PACE4,
15
with a K value in the same range as peptides with Lys and Arg at this position . This
i
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
result suggests that Leu residue in the P6 position greatly contributes to inhibitory
properties of peptide 1 towards PACE4.
The selectivity indexes were calculated for the P6 library screen against PACE4 and
furin, and indicated that only two peptides offered a selectivity of at least twoꢀfold toward
PACE4. Peptides containing Ile (analog 27) and His (analog 26) displayed selectivity
indices of 3.0 and 2.7, respectively (Figure 3).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Position P7
In the AcꢀLXLLRVKꢀAmba library, all peptides with exception of the negatively
charged analog 40 (K : 80 ± 20 nM) and analog 41 (K : 51 ±3 nM) displayed K s below
i
i
i
1
0 nM for PACE4 (Figure 2C). Although His containing peptides were among the best
PACE4 inhibitors at the other positions, compound 44 (K : 3 ± 2 nM) displayed average
i
inhibition at the P7 site, whereas other basic residues led to analogs 51 (K : 0.4 ± 0.4 nM)
i
and 46 (K : 0.5 ± 0.2 nM) with subnanomolar inhibition. The same trend was observed
i
for furin, with a K of 4.7 ± 0.3 nM for peptide 44 versus subnanomolar inhibition by
i
analog 51 (K : 0.5 ± 0.2 nM) and analog 46 (K : 0.6 ± 0.4 nM). Furthermore, large amino
i
i
acids containing aromatic moieties such as Trp (peptide 55, K PACE4: 8 ± 5 nM; furin:
i
1
3 ± 5 nM) and Phe (peptide 42, K PACE4: 8 ± 1 nM; furin: 7 ± 3 nM) showed slight
i
disadvantages compared to other residues. Therefore, flexibility constraints or steric
hindrance due to side chain composition could explain the disadvantage of the residues
Trp, Phe and His for both PACE4 and furin (Figure 2). However, inhibition for those
peptides remained in the low nanomolar range.
ACS Paragon Plus Environment

Page 11 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
The P7 library screen revealed very similar inhibitory profiles for PACE4 and furin.
Indeed, only slight differences were observed between the selectivity indexes calculated
for every substitution (Figure 3). This suggests that although position P7 appears to
participate in peptide affinity, subsite S7 could not be exploited for molecular
discrimination of both enzymes into a high affinity PACE4 inhibitor.
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
Position P8
The analysis of residues’ contribution to the P8 position in the AcꢀXLLLRVKꢀAmba
library demonstrated an attenuated influence of amino acid substitutions at this position
(Figure 2D). Indeed, although the basic residues: Lys (analog 64, K : 0.3 ± 0.2 nM), Arg
i
(
peptide 69, K : 0.6 ± 0.3 nM) and His (analog 62, K : 1.3 ± 0.6 nM) offered potent
i
i
inhibition toward PACE4, the K values obtained for other residues were in a narrow
i
range, from 1.5 nM for Ala (analog 57) to 12 nM for Asp (analog 58). Furthermore,
negative residues presented in peptides 59 (K : 8 ± 5 nM) and 58 (K : 12 ± 7 nM), did not
i
i
offer a substantial inhibitory disadvantage, as opposed to the results obtained for position
P5, P6 and P7. Thus, whereas basic residues contributed to inhibition at this position, the
P8 library screen demonstrated a minimal impact of residue substitutions on the initial
scaffold (2), suggesting that the P8 residue of the inhibitors would be outside the binding
cleft of PCs.
Consistent with this observation, no residue seems to offer discrimination between
PACE4 and furin as observed from the selectivity index. With the exception of peptide
5
7 (2.0ꢀfold selectivity) all of the selectivity indexes were below 2.0.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Proliferation studies
As described previously, the ML peptide and its analogs display antiproliferative effects
on DU145 and LNCaP prostate cancer cell lines, which both express the enzyme PACE4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
15, 17
at significant levels
. To evaluate the antiproliferative effects of each peptide in the P5
to P8 libraries, proliferation studies were performed at doses of 25, 50 and 100 µM on
DU145 cells (Figure 4). Peptides with antiproliferative effects on DU145 cells were
selected for further proliferation assays using peptide doses of 1, 10, 25, 50, 75 and 100
µM on DU145 and LNCaP cell lines, and the IC50 values were calculated from the
inhibition curves (Table 1). In position P5, IC s were calculated from selected peptides
5
0
containing analog 6 (DU145: 40 ± 10 µM; LNCaP: 50 ± 20 µM), analog 9 (DU145: 50 ±
0 µM; LNCaP: 28 ± 6 µM), analog 14 (DU145: 12 ± 1 µM; LNCaP: 26 ± 9 µM) and
1
analog 18 (DU145: 45 ± 7 µM; LNCaP: 60 ± 10 µM). While analogs 6, 9 and 18
exhibited similar antiproliferative effects on both DU145 and LNCaP cells, peptide 14
presented a 2ꢀfold improvement when compared to the control peptide 2. In the position
P6, only analogs containing either Phe (analog 24), Ile (analog 27) or Tyr (analog 38)
exhibited inhibitory effects on DU145 cell proliferation (Figure 4). Calculation of IC s
5
0
demonstrated that peptides with Phe (24, DU145: 38 ± 2 µM; LNCaP: 40 ± 10 µM) or Ile
27, DU145: 32 ± 4 µM; LNCaP: 80 ± 10 µM) were equally potent relative to the control
(
peptide 2, whereas substitution with Tyr (38, DU145: 50 ± 10 µM; LNCaP: 130 ± 40
µM) altered the inhibitory effects on both DU145 and LNCaP cells (Table 1). Screening
for position P7 revealed that only three peptides were capable of inhibiting DU145 cell
proliferation (Figure 4), and determination of the IC s for these peptides containing
5
0
either Phe (analog 42, DU145: 36 ± 7 µM; LNCaP: 40 ± 10 µM), Ile (analog 45, DU145:
ACS Paragon Plus Environment

Page 13 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
3
1 ± 5 µM; LNCaP: 40 ± 8 µM) or Trp (analog 55, DU145: 50 ± 10 µM; LNCaP: 130 ±
4
0 µM) substitutions demonstrated that all of the peptides displayed similar inhibitory
properties as the initial scaffold 2. In the P8 position (Figure 4), several peptides
displayed antiproliferative effects on DU145 and LNCaP cells. Whereas peptides
containing Phe (analog 60, DU145: 30 ± 10 µM; LNCaP: 41 ± 5 µM), Ile (analog 63,
DU145: 27 ± 5 µM; LNCaP: 40 ± 10 µM) and Trp (analog 73, DU145: 23 ± 4 µM;
LNCaP: 39 ± 3 µM) displayed similar inhibition properties to the control peptide 2 on
both DU145 and LNCaP cells, substitution with Asp (analog 58, DU145: 90 ± 4 µM;
LNCaP: 40 ± 10 µM), Glu (analog 59, DU145: 150 ± 20 µM; LNCaP: 120 ± 30 µM),
Gln (analog 68, DU145: 150 ± 40 µM LNCaP: >250 µM), Ser (analog 70, DU145: 100 ±
22 µM; LNCaP: >250 µM), Thr (analog 71, DU145: 140 ± 10 µM; LNCaP: 110 ± 20
µM), Val (analog 72, DU145: 75 ± 20 µM; LNCaP: 70 ± 20 µM) and Tyr (analog 74,
DU145: 23 ± 6 µM; LNCaP: 160 ± 20 µM) decreased inhibitory potency in the cellꢀ
based assay.
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
Of all the analogs tested in this experiment, only compound 14 showed increased
2
0
antiproliferative effects compared with the previously described control peptide 2 . For
every position tested, peptides substituted with Phe or Ile displayed similar
antiproliferative properties when compared to the lead compound. Interestingly,
according to Eisenberg’s hydrophobicity classification of amino acids, these two residues
26
are the most hydrophobic residues among the 20 natural amino acids . Furthermore,
peptides substituted with Val at the P5 position (analog 18), Trp at the P7 position (analog
5
5) and Trp at the P8 position (analog 73) also offered potent antiproliferative effects on
both DU145 and LNCaP cells and feature hydrophobic side chains. With the exception of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the polar Gln residue at the P5 position (analog 14), every peptide that displayed
inhibition in the low micromolar range on both prostate cancer cell lines was substituted
with highly hydrophobic residues. This observation suggests that the hydrophobic nature
of the P5ꢀP8 core of scaffold 2 is a critical feature to preserve antiproliferative effects,
and this information should be taken into account for further optimization of PACE4
inhibitors.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Cell penetration
Since our previous studies demonstrated the need for cell permeability of PACE4
inhibitors in order to elicit antiproliferative effects, we ought to assess ability of our new
compounds to penetrate cells. We therefore selected two analogs based on their in vitro
inhibitory activity, 9 and 15, and determined their cellular uptake. Both inhibitors display
potent activity against recombinant PACE4, however only analog 9 exhibits
antiproliferative effect on prostate cancer cell lines. To compare the quantity of the
internalized peptides, the FITCꢀlabeled derivatives of both analogs were prepared and
analyzed by flow cytometry after 1h incubation with DU145 cells. Falseꢀpositive signals
resulting from nonspecific interactions with membrane proteins were eliminated after
extensive trypsinization. Quantitative assessment of cell penetration of each peptide in
2
7
comparison with the initial scaffold (FITCꢀβAlaꢀLLLLRVKꢀAmba, compound 75) is
presented in Figure 5. The fluorescent version of peptide 15 (FITCꢀβAlaꢀLLLRRVKꢀ
Amba, compound 76) showed the lowest cellular uptake with the geometric mean
fluorescence intensity (GMFI) of 12.60 and 2.73 without and with trypsin wash,
ACS Paragon Plus Environment

Page 15 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
respectively. On the other hand, the fluorescent derivative of inhibitor 9 (FITCꢀβAlaꢀ
LLLIRVKꢀAmba, compound 77) exhibited improved cell permeability when compared
to the initial peptide scaffold (GMFI = 87.66 vs 46.61 without trypsin wash and GMFI =
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
9
9.26 vs 47.45 with trypsin wash). These data demonstrate a significant correlation
between the cellular uptake and the antiproliferative activity, indicating that the
hydrophobic region (P8ꢀP5) is important for the cell permeability and its disruption with
the positively charged amino acid led to an inactive analog 15.
Combined modifications
To investigate whether the selectivity profile of our inhibitors might be improved by the
combination of substitutions we design the analog containing two residues showing the
best selectivity profile (analog 78). As shown in Table 2, this new inhibitor turned out to
be slightly more potent for furin and completely unselective (K : 9.8 ± 4 nM for PACE4
i
and K : 8.1 ± 0.5 nM for furin). Although both substitutions individually provided an
i
increase in selectivity toward PACE4 (selectivity index for peptide 9: 3.4 and for peptide
2
7: 3.0), the combination of two Ile residues in the position P5 and P6 did not yield a
selective inhibitor (selectivity index for peptide 78: 0.8). However, it should be noted that
peptide 78 substituted with combined modification possesses potent antiproliferative
effects on both DU145 and LNCaP with IC of 20 ± 2 µM and 30 ± 10 µM, respectively
5
0
(Table 3).
P8
In addition, we prepared the DLeu ꢀsubstituted analog of this peptide (Acꢀ
[
DLeu]LIIRVKꢀAmba, analog 79), as it was demonstrated that the incorporation of the
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
DLeu into the structure of peptide 2 resulted in an analog Acꢀ[DLeu]LLLRVKꢀAmba,
2
7
known as compound C23 with 2ꢀfold selectivity towards PACE4 . However, also in this
case we did not observe any improvement in the selectivity profile and the resulting
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
analog possessed comparable inhibitory potency against PACE4 and furin (K : 14 ± 6 nM
i
for PACE4 and K : 12± 4 nM for furin, Table 2). On the other hand, peptide 79 similar to
i
its counterpart without the
DLeu residue exhibited potent antiproliferative activity on
both DU145 and LNCaP with IC of 37 ± 5 µM and 60 ± 30 µM, respectively (Table 3).
5
0
Discussion and conclusion
The development of PC inhibitors has been the focus of several studies since these
2
enzymes are implicated in various pathologies (review in Couture et al ). PACE4 is
involved in prostate cancer progression, and its inhibitors have demonstrated
1
5ꢀ17, 20
antiproliferative effects on prostate cancer cell lines
. However, the important role
of PCs in homeostasis suggests the relevance of developing selective inhibitors with
limited inhibitory properties toward other PCs. Our previous study demonstrated that
inhibitory selectivity toward PACE4 was achievable with short peptides, as compound 1
that displayed a 20ꢀfold binding affinity for PACE4 over furin (K PACE4: 22 ± 6 nM; K
i
i
1
5
furin: 430 ± 10 nM) . The addition of the Amba moiety into its structure (peptide 2)
resulted in a significant increase in inhibitory potency (K PACE4: 3.1 ± 0.8 nM) and an
i
20
important loss of selectivity (K furin: 4.3 ± 0.8 nM) . Despite the loss in selectivity,
i
peptide 2 appears as a promising lead compound for drug design because we observed
important improvements in stability along with an increase in antiproliferative
20
properties . Of important note, its derivative compound C23 (analog 80) was previously
ACS Paragon Plus Environment

Page 17 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
28
shown to inhibit tumor progression in a xenograft model of prostate cancer . A thorough
examination of this analog biodistribution showed adequate tumor delivery and extended
2
8
in vivo stability .
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
To optimize the selectivity of this drugꢀcandidate, we designed a positional scanning
library on the scaffold of peptide 2 to study the molecular determinants within this high
affinity PACE4 inhibitor. The library included the canonical recognition pattern
described for PCs, modified by the arginine mimetic Amba moiety and an Nꢀterminal
tetraꢀLeu sequence (Figure 1). Dahms et al determined the crystal structure of furin
bound to a short, nonspecific PhacꢀRVRꢀ4ꢀAmba peptide and demonstrated that the rigid
Amba moiety was deeply inserted into catalytic pocket, which resulted in tight enclosure
14
of this arginineꢀmimetic within the S1 subsite . The conformational advantage of the
Amba residue led to tightꢀbinding inhibitors, as previously described from enzyme
2
0ꢀ23, 29
kinetics assays
. In this study, all inhibitors tested presented a high affinity, with Kis
in the nanomolar or the picomolar range, consistent with these observations. However,
library screening demonstrated that, although the Amba residue encounters a tight
binding interaction between the PCs and peptide inhibitors, molecular determinants exist
for positions P5 and beyond because differences in binding affinities were observed.
For every position screened, the general trend indicated a preference for basic residues
Arg, Lys and His for both PACE4 and furin (Figure 2). Thus, this observation is in
accord with previous positional screening toward furin, which demonstrated that
30ꢀ32
positively charged amino acids were preferred for this enzyme
. Furthermore,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
substrate cleavage efficiency studies have previously shown that basic residues in
3
3, 34
positions P5 and P6 could increase cleavage efficiency by PCs
. Structural analyses
revealed that the PC catalytic cleft is fairly negatively charged, which explains the clear
preference for basic residues, as well as the unfavorable presence of the negatively
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
1ꢀ14
charged amino acid residues Asp and Glu
. With the exception of charged amino
acids, a distinction in binding preferences for large nonpolar aromatic residues was
observed for positions P6 and beyond. Indeed, the large amino acid Trp was among the
less potent substitution for positions P6 to P8, and Phe offered a slight disadvantage at
positions P7 and P8 for both PACE4 and furin. Finally, discrimination in binding
preferences could be observed on the basis of hydrophobicity. Whereas the very
hydrophobic residue Ile offered a disadvantage toward furin when introduced in positions
P5 and P6, this residue was well tolerated in the PACE4 binding cleft. For analogs
containing amino acids with shorter aliphatic side chains, Ala and Val, they displayed
comparable binding affinities for both enzymes. This result demonstrated that there are
differences in subsites S6 and S5 between PACE4 and furin.
It should be mentioned, that a previous screen performed on a partial combinatorial
library revealed that Leu at the P6 position provides the most selective inhibition towards
15
PACE4 over furin , whereas in the present study its Cβꢀbranched isomer – Ile showed
the highest selectivity. Unfortunately, with the positional scanning performed in the
present study, no peptide containing the Amba moiety displayed a 20ꢀfold selectivity as
1
5
previously observed for peptide 1 . This can be explained by the tight binding properties
of Amba, which stabilizes the active peptideꢀenzyme complex even with presence of
ACS Paragon Plus Environment

Page 19 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
unfavorable residues in positions P5 to P8. Indeed, despite the introduction of negatively
charged residues, K values remained in the nanomolar range for every position tested.
i
Although they lie outside the canonical recognition pattern for PCs, positions P5 to P7
appear to contribute to the peptides’ ability to bind to the furin and PACE4 binding clefts.
This study also suggests the limited participation of the position P8 to overall affinity of
the peptide 2. However, this position could be exploited to introduce Nꢀterminal
peptidomimetic modifications for further optimization of PACE4 inhibitors that could
enhance metabolic stability. Indeed, we previously observed that the introduction of the
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
unnatural amino acid,
DLeu, in the position P8 of peptide 1 did not alter its affinity
2
0
toward PACE4, while this substitution in position P5, P6 and P7 was not tolerated .
These results indicate that the P8 residue may be outside the catalytic cleft, thus
explaining the tolerance for substitution in this position.
To evaluate the contribution of each substitution on the antiproliferative properties of the
peptide 2, peptides from the P5ꢀP8 libraries were assayed in a proliferation assay using
DU145 cells. Only a few peptides in the P5 to P7 libraries demonstrated antiproliferative
properties, whereas ten peptides from the P8 library were selected for further analyses on
DU145 and LNCaP cells. The scan of the P5 to P8 library in the cellꢀbased assay
revealed that the in vitro affinity toward PACE4 is not directly informative of a peptide’s
antiproliferative effects. Indeed, the most potent PACE4 inhibitor tested in the enzyme
kinetics assay contained an Arg residue (analog 15, K PACE4: 0.2 ± 0.1 nM) displayed
i
no antiproliferative effects on DU145 cells (Figure 4). Of note, our previous work has
investigated the possibility of cytotoxicity coming from the peptide 2 and its analogs
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
using a membrane integrity assay, which measures the release of lactate dehydrogenase
in the media following cell lysis. These studies have shown that compound 2 (at the 100
ꢀM concentration) after 4 hours incubation with DU145 cell, led to less than 5% cell
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
death, suggesting that the calculated IC s values are the result of antiproliferative
5
0
2
0
properties rather than cytotoxic effects from this series of peptide . Using the FITCꢀ
labeled version of analog 15, we demonstrate that it has significantly reduced cellular
uptake when compared to analogs with hydrophobic Leu or Ile residue at this position
(Figure 5). These data strongly support our previous observations that inhibitor 1 or its
15
analogs have to target intracellular PACE4 to induce an antiproliferative response . The
incorporation of the charged side chain into the tetraꢀleucine core diminished its cell
penetration properties, possibly caused by disruption of the amphipathic structure of
inhibitor 2.
Amphipathic peptides are known to enter cells by an endocytosisꢀindependent
mechanism, likely through direct interactions with the membrane bilayer and membrane
3
5ꢀ37
disorganization, which leads to a broad cellular distribution of those peptides
.
Inhibitor 1 and its analogs display excellent cell penetration properties, and analyses of
cellular localization indicated a broad cellular presence of its FITCꢀlabeled analog when
1
5
incubated with cells . In this study, only peptides containing hydrophobic residues, with
the exception of an analog having Gln at the P5 position, displayed potent
antiproliferative effects on DU145 and LNCaP cells. Therefore, preservation of the
amphiphilic character should be considered upon designing new PACE4 inhibitors.
ACS Paragon Plus Environment

Page 21 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Despite the beneficial effect on the selectivity profile of Ileꢀsubstituted peptides (analog 9
and 27, 3ꢀfold improvement), their combination enhanced binding affinity for furin and
resulted in an unselective analog 78 (Table 2), indicating that binding sites of furin
exhibit some degree of cooperativity. This phenomenon has been already observed in a
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
3
8
wide range of proteases , including the proprotein convertase yeast homolog kexin,
where binding of a particular amino acid was shown to be influenced by the neighboring
39
residue within positions P1 to P4 . These results demonstrated that in case of our starting
inhibitor 2 only the single substitution Ile at the P5 or P6 position is well tolerated and
provides some degree of discrimination between furin and PACE4. This indicates that an
alternative approach such as targeting allosteric sites and exosites (alone or together with
the active sites) should be considered to optimize the selectivity profile of PACE4
inhibitors. This strategy has been proven successful to generate new anticoagulants, i.e.
highly selective thrombin inhibitors known as bivalent direct thrombin inhibitors
4
0
(
hirudinꢀderived peptides) , which bind simultaneously to the catalytic site and exosite 1
41
domain of thrombin . The importance of "adventitious binding points" has been also
investigated for proteinꢀbased PCs inhibitors. It has been shown that engineered variants
of eglin c exhibited enhanced selectivity towards target enzyme (e.g. up 41ꢀfold for furin
4
2
versus PC7 and 20ꢀfold for PC7 versus furin) . However, until now, to the best of our
knowledge, no exosites or allosteric sites have been identified for PCs.
In regards to the eventual use of PACE4 inhibitors as antiproliferative agents in prostate
cancer, it has not yet been determined if these compounds have secondary effects. In a
recent study, it was shown that PACE4 is an enzyme responsible for the activation of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
43
corin, also an enzyme which generates the natriuretic peptides in the heart . It was thus
speculated that PACE4 inhibition might results in cardiovascular side effects. However,
in our hands, the presently described PACE4 inhibitors do not produce any hypertensive
effects. Furthermore, PACE4 knockout mice are not hypertensive under normal
conditions (unpublished data).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
In conclusion, this paper demonstrates that molecular determinants for PACE4 and furin
exist within subsites S5 to S7. However, the influence of amino acid substitution
appeared attenuated with the presence of the Amba moiety in the P1 position, leading to
high affinity inhibitors of both PCs even in the presence of unfavorable residues in
positions P5 and beyond. Only, the incorporation of the Ile residue offered a 3ꢀfold
inhibitory selectivity toward PACE4 when introduced within position P5 or P6 of the
starting peptide. Furthermore, the presence of this highly hydrophobic amino acid residue
at these positions sustained the amphipathic nature of peptide 2 and the antiproliferative
effects observed toward prostate cancer cell lines. The findings described here could be
used to guide the design of new generation of PACE4 inhibitors with improved
selectivity and activity for potential therapeutic applications. As regard to therapeutic
applications of PACE4 inhibitors, a potent and stable yet less selective analog might be
suitable. Further optimization of therapeutic regimens or targeted delivery could offer
practical therapeutic index.
ACS Paragon Plus Environment

Page 23 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Experimental section
Peptide Synthesis
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
Fmoc amino acids, resins, coupling reagents and solvents were purchased from GL
Biochem, Merck Chemicals, Rapp Polymere and Sigma Aldrich. FmocꢀSPPS (Fmoc/tBu
strategy) was performed manually or on an automatic synthesizer Symphony (Protein
Technologies, USA). Protected 4ꢀamidinobenzylamine and 4ꢀamidinobenzylamine·2HCl
2
0,23, 44ꢀ46
were synthesized using protocols described in the literature
.
All P5ꢀsubstituted peptides were synthesized on a 4ꢀFmocꢀhydrazinobenzoyl AM
NovaGel resin (substitution 0.49 mmol/g), with the exception of the analog modified with
the Met residue. After removing the protecting group from the resin using a standard
procedure, the AcꢀP8ꢀP2 fragment of the peptide was synthesized using a Symphony
(Protein Technologies, USA) automatic synthesizer as described in Scheme 1. During the
synthesis, amino acid coupling was performed using 2.5 equivalents of amino acid, 2.5
equiv of Oꢀ(benzotriazolꢀ1ꢀyl)ꢀN,N,N′,N′ꢀtetramethyluronium tetrafluoroborate (TBTU),
2
.5 equiv of 1ꢀhydroxybenzotriazole (HOBt) as coupling reagents and 5 equiv of Nꢀ
methylmorpholine (NMM). Nꢀterminal acetylation was performed three times with acetic
anhydride/diisopropylethylamine (DIPEA)/dichloromethane (DCM) (1.5/1.5/7, v/v/v) for
1
0 min. Peptidyl resin oxidation and coupling of the protected 4ꢀamidinobenzylamine
were performed manually with 2.5 equiv of the compound, 0.7 equiv of Cu(OAc) and
2
47ꢀ49
1
.5 equiv of pyridine in anhydrous N,Nꢀdimethylformamide (DMF) for 24 h
. The
resin was drained, and DMF was removed in vacuo. The protecting groups were removed
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
from side chains using a standard mixture of trifluoroacetic acid (TFA)/triisopropylsilane
(
TIS)/ H O (95/2.5/2.5, v/v/v) for 2.5 h. The solvents were removed in vacuo and the
2
inhibitors were precipitated with ether, centrifuged and lyophilized from 30% acetonitrile
in water. The peptides were then dissolved in acetic acid/methanol (MeOH) (1/1, v/v),
diluted to a 1 M concentration, treated with catalyst (10% Pd/C) and hydrogenated using
a shakerꢀtype hydrogenation apparatus 3911 (Paar, USA) (1.5 h, 55 psi, RT). The catalyst
was filtered and washed with MeOH and solvents were removed in vacuo. The product
was diluted in 20% acetonitrile in water, lyophilized and purified by preparative HPLC
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(system Waters; purity >98%).
Compound 11, all peptides substituted at the P6, P7 and P8 positions, inhibitors with
multiple substitutions and FITCꢀmodified analogs were prepared using a combination of
SPPS and solutionꢀphase synthesis as shown in Scheme 2. The loading of the first amino
acid on the 2ꢀchlorotritylꢀchloride resin was prepared using 2 equiv of FmocꢀLys(Boc)ꢀ
OH, dissolved in dry DCM (10 mL per 1 g of resin) and rapid addition of 2 equiv of
DIPEA to the resin. After 1 h, MeOH was added (2ꢀ4 mL per 1 g of resin), and the
mixture was stirred for 15 min. The resin was drained and washed with DCM, DMF, and
DCM and dried in vacuo. Fmoc SPPS was continued manually or with an automatic
synthesizer using the same procedure as for the peptides presented above. For the FITCꢀ
labelled peptides, in the last step the resin was treated with a solution of 3 equiv of FITC
in pyridine/DCM (1:4 v/v) and agitated on the automated shaker for 16 h in room
temperature. The resin was filtrated and washed successively with acetonitrile (ACN),
DMF, isopropanol and DCM. Then, peptidyl resins were treated as their counterparts
ACS Paragon Plus Environment

Page 25 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
without the FITC moiety. Cleavage of fully protected peptides was obtained by treatment
with hexafluoroisopropanol (HFIP)/trifluoroethanol (TFE)/dichloromethane (DCM)
(3/2/5, v/v/v) for 2 h at RT. After solvent removal in vacuo, the analogs were precipitated
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
with cold diethyl ether, centrifuged and lyophilized from 50% tert-butanol in water.
Coupling of the the protected 4ꢀamidinobenzylamine or 4ꢀamidinobenzylamine ·2HCl
was carried in solution using 2 equiv of this compound, 2 equiv of 1ꢀcyanoꢀ2ꢀethoxyꢀ2ꢀ
oxoethylidenaminooxy)dimethylaminomorpholinoꢀcarbenium
hexafluorophosphate
(COMU) and 8 equiv of NMM or 6 equiv of DIPEA in anhydrous DMF overnight as
20
described previously . DMF was removed, and side group deprotection and
hydrogenation were performed as described above. The peptide was diluted in 20% ACN
in water, lyophilized and purified by reversed phase preparative HPLC to achieve a purity
of >98% using a Waters system (Kromasil 100 C8 column, 250 x 16 mm, 5 µm) with a
linear gradient of 80% ACN in water containing 0.1% of TFA from 13 to 43%, 13 to
5
0% or 20 to 65% for 90 min at a flow rate of 8 mL/min or a Shimadzu system (Jupiter 4
Proteo column, 90 Å, 250 x 10 mm) with a linear gradient of ACN from 20 to 60% for
ꢁ
50 min at a flow rate of 4 mL/min or a VARIAN ProStar system (Phenomenex Jupiter
C18 column, 300 Å, 250 × 21.20 mm, 5 ꢁm) with a linear gradient of ACN from 25 to
5
0% for 55 min a flow rate of 7 mL/min). All inhibitors were finally obtained as TFA
salts after lyophilization.
Inhibitor purity was determined using an Agilent analytical RPꢀHPLC system (VYDAC
C or AGILENT Elipse XDB C column) and a Shimadzu RPꢀHPLC system (Jupiter 4 ꢁ
18
18
Proteo 90 Å, 250 x 4.60 mm). The molecular masses of the synthesized compounds were
verified using LCMS ESI ITꢀTOF (Shimadzu, Japan) or HRMS (TripleTOF 5600,
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
ABSciex, USA) mass spectrometers. The physicochemical properties of all analogs are
presented as supporting information in Table S1. All tested compounds possess purity of at
least 98% (for FITCꢀpeptides at least 95%).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Enzymatic assays
The inhibition constants (K ) were determined using human recombinant PCs (PACE4
i
5
0
and furin) that were produced and purified as described previously . Competitive assays
were performed in 96ꢀwell microtiter plates for 60 min at 37°C with 100 µM of the
fluorogenic substrate pyrGluꢀArgꢀThrꢀLysꢀArgꢀ7ꢀamidoꢀ4ꢀmethylcoumarin (Bachem,
CA, USA) and different inhibitor concentrations (ranging from 10 µM to 1 nM). The
enzymes were added at a final concentration of 2 units per 100 µL in each well (PACE4:
2
0.18 nM; furin: 0.54 nM) and fluorescence was measured using a Gemini XS
spectrofluorometer equipped with SoftMaxPro5 (Molecular Devices, CA) (λ : 370 nm;
EX
λ : 460 nm; cutoff, 435 nm). K values were calculated from the Morrison equation for
EM
i
tight binding inhibitors using Prism 5.0 (GraphPad Software, CA, USA) as described
2
0
previously . The MichaelisꢀMenten constants (K ) for PACE4 and furin are 4.035 µM
M
and 5.04 µM, respectively. K s were determined in three independent experiments for
i
each peptide and the results presented in Figure 4 and Table 1 are the mean ± standard
deviation.
Assays with PACE4 were carried out in 20 mM BisTris pH 6.5, 1 mM CaCl and 1.8
2
mg/mL BSA buffer. Furin enzymatic assays were carried out in 100 mM HEPES, pH
7
.5, 1 mM CaCl , 1 mM βꢀmercaptoethanol, and 1.8 mg/mL BSA, except for His
2
containing peptides, which were performed in 20 mM BisTris pH 6.5, 1 mM CaCl , 1
2
ACS Paragon Plus Environment

Page 27 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
mM βꢀmercaptoethanol, and 1.8 mg/mL BSA. The inhibition assays of peptides
containing His residues were performed at pH 6.5 to ensure similar ionization of this
residue and to allow for proper comparison of the binding properties between PACE4 and
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
furin, given that the His side chain pK is 6.08.
a
Proliferation assays
Prostate cancer cell lines were obtained from ATCC and maintained at 37°C in a waterꢀ
saturated atmosphere with 5% CO in RPMI 1640 medium supplemented with either 5%
2
fetal bovine serum (FBS; Wisent Bioproducts, Canada) for DU145 or 10% FBS for
1
5,20
LNCaP. The MTT proliferation assay was performed as described previously
Briefly, DU145 cells were seeded at a density of 1500 cells/well and LNCaP cells were
seeded at a density of 2500 cells/well in a polyꢀ ꢀLysine coated plate. Twentyꢀfour hours
.
L
after cell plating, the inhibitors were added to the cells in fresh FBSꢀsupplemented media.
A first screen of the antiproliferative properties was performed using inhibitor doses of
2
5, 50 and 100 µM (Figure 4) with DU145 cells, and complete doseꢀresponse curves
were then determined with inhibitor doses of 0, 1, 10, 25, 50, 75, 100 and 250 µM on
DU145 and LNCaP cells (Table 1). The peptide inhibitors were incubated with the cells
for 72 hours and metabolic activity was evaluated by adding the MTT reagent to the
DU145 and LNCaP cells at final concentration of 1 mg/mL for 4 hours. The formazan
salt was solubilized with 100 ꢁL of 2ꢀpropanol HCl (24:1 N) and the total metabolic
activity was normalized relatively to vehicleꢀtreated cells (sterile bidistilled water). The
halfꢀmaximal inhibitory concentration (IC ) was calculated using Prism 5.0 (GraphPad
5
0
Software, USA). The proliferation assay results are the mean ± standard deviation from at
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 60
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
least 3 independent experiments (n=3).
Cell penetration
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The cell penetration assays were performed on DU145 cells using FITCꢀlabeled peptides.
Briefly, DU145 cells were incubated for 1 h in serumꢀfree RPMI media with 10 ꢁM of
the FITCꢀanalog and collected by trypsin:EDTA digestion followed by centrifugation.
Cells in suspension were then extensively washed twice with trypsin (0.05% v/v) for 5
min at 37 °C to remove nonspecific interactions with the membrane proteins. Then, cells
were incubated for 2 min with propidium iodine (10 ꢁg/mL) to exclude cells with altered
membrane and were immediately analyzed (10 000 events/sample) by fluorescence
activated cell sorting (FACS; Becton Dickinson; Mountain View, CA, USA). GeoMeans
were determined using CellQuest Software (Becton Dickinson).
Acknowledgements
The authors thank Xue Wen Yuan for assistance with peptide synthesis. We also thank
Dr. Leonid Volkov for his help with flow cytometry, and Dr. Hugo Gagnon
(PhenoSwitch Biosciences Inc.) for highꢀresolution mass spectrometry analysis. This
research was financed in part by the National Science Centre Poland decision no. DECꢀ
2
012/05/N/ST5/01080 (IM), the Canadian Cancer Society Research Institute (Impact
grant # 701590), the Fondation Mon Étoile, Prostate Cancer Canada and the Movember
Foundation (grants #2012ꢀ951 and #D2013ꢀ8 and TAG2014ꢀ02). FC holds a Banting and
Charles Best Canada Graduate Scholarships (grant #315690) from CIHR and Graduate
Studentship from Prostate Cancer Canada (Grant #GSꢀ2015ꢀ07).
ACS Paragon Plus Environment

Page 29 of 60
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Abbreviations
ACN, acetonitrile; Amba, 4ꢀamidinobenzylamide; COMU, 1ꢀcyanoꢀ2ꢀethoxyꢀ2ꢀ
oxoethylidenaminooxy)dimethylaminomorpholinoꢀcarbeniumꢀhexafluoro phosphate;
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
DCM, dichloromethane; DIPEA, diisopropylethylamine, DMF, N,Nꢀdimethylformamide;
DMSO, dimethyl sulfoxide; FACS, fluorescenceꢀactivated cell sorting; FBS: Foetal
Bovine Serum; FITC, fluorescein isothiocyanate; GMFI, geometric mean of fluorescence
intensity; HFIP, hexafluoroisopropanol; HOBt, 1ꢀhydroxybenzotriazole; HRMS, highꢀ
resolution mass spectrometry; IC , halfꢀmaximal inhibitory concentration; K , inhibition
5
0
i
constant; LCMS ESI ITꢀTOF, Liquid chromatography–mass spectrometry system with
electrospray ionization connective ion trap and time of flight analyzers; ML, multiꢀLeu
peptide inhibitor; NMM, Nꢀmethylmorpholine; PC, proprotein convertase; RPꢀHPLC,
reversedꢀphase highꢀperformance liquid chromatography; SPPS, Solidꢀphase peptide
synthesis; TBTU, Oꢀ(benzotriazolꢀ1ꢀyl)ꢀN,N,N′,N′ꢀtetramethyluronium tetrafluoroborate;
TFA, trifluoroacetic acid; TFE, trifluoroethanol; TIS, triisopropylsilane.
ACS Paragon Plus Environment