Optimization strategy of single-digit nanomolar cross-class inhibitors of mammalian and protozoa cysteine proteases

B i o o r g an i c C h e m i s tr y 1 0 1 (2 020 )1 04039

Contents lists available at ScienceDirect

Bioorganic Chemistry

journal homepage: www.elsevier.com/locate/bioorg

Optimization strategy of single-digit nanomolar cross-class inhibitors of

mammalian and protozoa cysteine proteases

Lorenzo Cianni^a,1, Fernanda dos Reis Rocho^a,1, Fabiana Rosini^a, Vinícius Bonatto^a,

⁎

Jean F.R. Ribeiro^a, Jerônimo Lameira^a,b, Andrei Leitão^a, Anwar Shamim^a,

,

⁎

Carlos A. Montanari^a,

^aMedicinal and Biological Chemistry Group, Institute of Chemistry of São Carlos, University of São Paulo, Avenue Trabalhador Sancarlense, 400, 23566-590 São Carlos,

SP, Brazil

^bOn Leave from Drug Designing and Development Laboratory, Federal University of Pará, Rua Augusto Correa S/N, Belém, PA, Brazil

A R T I C L E I N F O

A B S T R A C T

Keywords:

Cysteine proteases (CPs) are involved in a myriad of actions that include not only protein degradation, but also

play an essential biological role in infectious and systemic diseases such as cancer. CPs also act as biomarkers

and can be reached by active-based probes for diagnostic and mechanistic purposes that are critical in health and

disease. In this paper, we present the modulation of a CP panel of parasites and mammals (Trypanosoma cruzi

cruzain, LmCPB, CatK, CatL and CatS), whose inhibition by nitrile peptidomimetics allowed the identiﬁcation of

speciﬁcity and selectivity for a given CP. The activity cliﬀs identiﬁed at the CP inhibition level are useful for

retrieving trends through multiple structure–activity relationships. For two of the cruzain inhibitors (10g and

4e), both enthalpy and entropy are favourable to Gibbs binding energy, thus overcoming enthalpy–entropy

compensation (EEC). Group contribution of individual molecular modiﬁcation through changes in enthalpy and

entropy results in a separate partition on the relative diﬀerences of Gibbs binding energy (ΔΔG). Overall, this

study highlights the role of CPs in polypharmacology and multi-target screening, which represents an imperative

trend in the actual drug discovery eﬀort.

Cysteine protease inhibitors

Structure activity relationships

Molecular dynamics simulation

Matched molecular pair analysis

Isothermal titration calorimetry

Diﬀerential scanning calorimetry

1. Introduction

Therefore, its inhibition interrupts several cellular processes leading to

the death of the parasite [3].

Currently, druggable cysteine proteases (CPs) are fundamental

biological targets for the discovery of novel therapeutics for many

human diseases: from parasitic diseases to cancer [1–4]. They are one of

the most abundant classes of enzymes, being involved in many biolo-

gical processes, such as cell-cycle, signaling, and cell death [2]. CatK is

the only cathepsin expressed at high levels in osteoclasts. Thus, in-

hibitors have been developed for the treatment of bone diseases such as

osteoporosis [5]. While CatL and CatS upregulations have been re-

ported for many human cancers and autoimmune diseases, several in-

hibitors of these two enzymes are nowadays in clinical phase as new

drug candidates [1,2].

LmCPB is one of the three families of CPs genes expressed in the

genus Leishmania, along with the class A (LmCPA) and the class C

(LmCPC) of cysteine proteases. The class B cysteine protease (CPB) is

also part of the papain family and has essential involvement in the

parasite life cycle in L. mexicana [4]. Accordingly, this enzyme is under

investigation as a potential biological target for the treatment of dis-

eases triggered by L. mexicana infection. Recently, we have disclosed an

extrathermodynamic relationship to model the inhibition constants K_i,

converted to the pK_iscale, between LmCPB and Cz. This quantitative

model is useful for extracting chemical information that allows the

prediction of structure–activity relationships (SARs) between the two

enzymes [7].

Cruzain (Cz), also assigned as cruzipain (the full-length native en-

zyme), is the major cysteine protease expressed in the parasite

Trypanosoma cruzi, which is the etiological agent of Chagas Disease

(CD) [6] Cz is present in all stages of the parasite life cycle, inducing

damage to the host tissue and facilitating the parasite invasion.

In general, reversible inhibitors of mammalian CPs are mostly

considered for the development of new drug candidates. A similar trend

is observable for inhibitors of non-human cysteine proteases. The

reason, in both cases, is to reduce oﬀ-target eﬀects and thereby impair

^⁎Corresponding authors.

E-mail addresses: shamim.hej@gmail.com (A. Shamim), Carlos.Montanari@usp.br (C.A. Montanari).

¹these authors contributed equally to the work.

https://doi.org/10.1016/j.bioorg.2020.104039

Received 9 February 2020; Received in revised form 19 June 2020; Accepted 19 June 2020

A v a i l a b l e o n l i n e 2 4 J un e 2 020

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 1. 2D Structure representation of Odanacatib, Neq0659 (4d) and Neq0820 (4e) as high-aﬃnity cysteine protease inhibitors.

undesirable reactions with endogenous nucleophiles and cytotoxicity

[1].

heating steps. Finally, we performed 100 ns of MD simulation using the

NPT ensemble for each ligand-CatS non-covalent complex and ligand-

Cz covalent complexes. To analyze all the MD trajectories, we have used

CPPTRAJ [21] available in AmberTools software package. The ﬂuc-

tuation graphs were produced through Gnuplot 5.0 [22]. We used

Chimera (UCSF) [23] to visualize the output of the simulations and

generate the Figures related to the computational results.

We have recently reported that peptidic and non-peptidic nitrile

based Cz inhibitors are eﬃcient trypanocidal agents [7–10]. Speciﬁ-

cally, we have conveyed on two new Odanacatib [11] derivatives

(Neq0659 and Neq0820, numbered in this work as 4d and 4e) as low

nanomolar Cz inhibitors, Fig. 1. We have also shown that the dipeptidyl

nitrile scaﬀold tolerates S3-P 3 interactions that can eﬃciently increase

the aﬃnity for the desired CP [12,13]. Thereafter, we have performed

structural optimization, synthesis and structure–activity relationships

(SARs) on a panel of CPs (Cz, LmCPB, CatK, CatL and CatS) for nitrile-

based inhibitors by the deconstruction of our published inhibitors

(Neq0659 and Neq0820, numbed in this work as 4d and 4e). To better

understand the reasons for cross-class inhibition of the new inhibitors

described here, we performed molecular dynamics simulations (MDS),

isothermal titration calorimetry (ITC) and diﬀerential scanning calori-

metry (DSC) studies on Cz inhibitors as CP surrogates.

2.2. Synthesis

All chemicals were purchased as reagent grade and used without

further puriﬁcation, unless otherwise noted. N,N-Dimethylformamide

(DMF) was dried over 3 Å activated molecular sieves for 72 h. All non-

aqueous reactions were carried out under argon atmosphere in oven-

dried glassware. Solvents used in high performance chromatography

(HPLC) were supplied by Tedia and used without further puriﬁcation

Thin layer chromatography was performed on Fluka Analytical

Sigma-Aldrich silica gel matrix, pre-coated plates with ﬂuorescent in-

dicator 254 nm and/or staining solutions. Flash column chromato-

graphy was performed on silica gel (pore size 60 Å, 70–230 mesh).

¹H and ¹³C NMR spectra were recorded on HP – 400 and 500 MHz

instruments in CDCl₃or DMSO‑d₆. Chemical shifts are referenced to the

residual solvent peak and J values are given in Hz. The following

multiplicity abbreviations are used: (s) singlet, (d) doublet, (dd) doublet

of doublets, (ddd) doublet of doublet of doublets, (dt) doublet of triplet,

(t) triplet, (q) quartet, (m) multiplet, and (br) broad.

2. Methods

2.1. Molecular dynamics simulations

Herein, the initial coordinates for molecular dynamics simulations

were taken from CatS crystal structure with PDB code 3OVX [14] and

Cz crystal structure with PDB code 3I06 [15]. Note that ligands were

placed in the active site of CatS and Cz using the coordinate information

obtained from the crystal structure of a dipeptidyl nitrile in complex

with Cz [9]. The partial charges for ligands were obtained using AM1-

BCC [16] and the ligands were parameterized with the GAFF [16] force

ﬁeld. Each system was solvated in a truncated octahedron TIP3P water

box [17]. Counter ions were added to maintain the electro-neutrality of

each ligand–protein complex system. It is also important to point out

that the standard protonation state at pH = 7 was assigned to all io-

nizable residues, where the protonation states of all the residues of

protein were carefully deﬁned according to the PROPKA.[17] The

AMBER 16 suite of programs [18] together with the Amber ﬀ14SB [19]

force ﬁeld were used to perform MD simulations, where the SHAKE

algorithm [20] was used to maintain all the bonds at their equilibrium

distances. Initially, the hydrogen atoms, water molecules, and ions were

minimized using 10,000 cycles of steepest descent and conjugate gra-

dient algorithms. Then, the whole system was heated through several

Characterization and separation of compounds were carried out

with a HPLC system. The analytical HPLC system consisted of a

Shimadzu LC (Kyoto, Japan) equipped with a LC-20AT pump, a LC-

20AD pump, a SIL-20A HT autosampler, a DGU-20A5 degasser, a CBM-

20A, SPD-M20A DAD detector and an FRC-10A fraction collector. Data

acquisition was performed using LCsolution software version 1.26 SP5.

The LC system was coupled to an AmaZon SL ion trap mass spectro-

meter (Bruker Daltonics, Bremen, Germany) equipped with an elec-

trospray ionization (ESI) interface. Data acquisition was performed

with Bruker Daltonics Data Analysis software (version 4.2.383.1).

Spectra for all compounds and further details can be found in the

supplementary information.

Solvents were ﬁltered through a 0.45 μm Merck-Millipore ﬁlter

before use and degassed in an ultrasonic bath. In the established HPLC

protocol, chiral analysis and separation were carried out at 32 °C

(column oven) where not otherwise speciﬁed, using analytical and

2

L. Cianni, et al.

BioorganicChemistry101(2020)104039

semi-preparative cellulose-2 Phenomenex column (Analytical: 5 μm,

250 mm × 4.6 mm I.D, semi-preparative: 5 μm, 250 mm × 10 mm I.D)

or Diacel column (IC-chiralpak: 5 μm, 250 mm × 4.6 mm) via isocratic

elution with a ﬂow rate of 0.5 (analytical) and 2.36 mL min⁻¹(semi-

preparative). The most common mobile phase composition was acet-

onitrile–water (50:50) (v/v). Volumes of 10 μL (analytical) and 1000 μL

(semi-preparative) were injected. Quantiﬁcation was carried out at

200–800 nm and the chromatographic run time varied according to the

sample.

puriﬁed by HPLC equipped with a chiral column.

2.2.4. Synthesis and characterization

Synthesis and characterization of compounds 3a-3c, 4d and 4e are

reported in our previous work [8].

(S)-N-((S)-1-cyano-2-phenylethyl)-3-phenyl-2-(((S)-2,2,2-triﬂuoro-1-

phenylethyl)amino)propanamide 3d. Compound 3d was synthesized

according to procedure GP1 followed by GP2 as a white solid (48.7%

yield over 3 steps); mp. 99.0–99.7 °C. Purity (HPLC) > 99%. ¹H NMR

(500 MHz, CDCl₃) δ 7.47 – 7.28 (m, 9H), 7.22 (d, J = 7.1 Hz, 2H), 7.18

– 7.11 (m, 4H), 5.02 (dt, J = 8.7, 6.5 Hz, 1H), 3.93 (q, J = 6.5 Hz, 1H),

3.43 (d, J = 5.4 Hz, 1H), 3.00 (dddd, J = 27.3, 20.7, 13.9, 6.0 Hz, 4H),

2.15 (s, 1H). ¹³C NMR (125 MHz, CDCl₃) δ 171.72, 135.91, 133.55,

132.46, 129.65, 129.36, 129.35, 129.06, 129.04, 129.00, 128.58,

Speciﬁc rotations ([α]^T= 100 α/lc, in deg mL g⁻¹dm⁻¹, but re-

ported herein in degrees) were observed at the wavelength 589 nm, the

D line of a sodium lamp. T was set to be 25 °C. Samples were weighted

using a precision balance (Sartorius, Model CPA26P) and dissolved in

methanol (HPLC grade, PanReac). Rotations were measured using a

Digital Polarimeter (P2000, Jasco): α = observed rotation in degrees;

l = cell path length of 0.1 decimeter; c = concentration in g 100 mL⁻¹

.

128.03, 127.53, 124.89 (q, J = 282.2 Hz), 117.59, 63.30 (q,

Values were calculated using 5 measurements for each compound.

Melting points were determined by a Quimica Micro MQAPF-302 ap-

paratus and are uncorrected.

J = 28.7 Hz), 60.04, 40.90, 38.61, 37.46. IR (ﬁlm, cm⁻¹) 3330.92,

3064.19, 3031.24, 2929.21, 2858.11, 2858.11, 2244.41, 1678.51,

1603.65, 1495.33, 1455.18, 1369.25, 1347.12, 1265.44, 1166.91,

23

1119.58, 1029.88, 918.63, 848.58, 746.17, 700.45. [α]_D

2.2.1. General procedure for the synthesis of amines from ketones and

aminoacids (GP1)

(Chloroform) −468.33.

To

a

solution of ^L-phenylalanine/L-leucine methylester hydro-

(S)-2-(((S)-1-(3-bromophenyl)-2.2.2-triﬂuoroethyl)amino)-N-((S)-1-

cyano-2-phenylethyl)-4-methylpentanamide 3e. Compound 3e was

synthesized according to procedure GP1 followed by GP2 as a white

solid (50% yield over 3 steps); mp. 171.9–172.8 °C. Purity (HPLC) 97%.

¹H NMR (500 MHz, CDCl₃) δ 7.53 (ddd, J = 8.0, 1.8, 0.9 Hz, 1H), 7.43

(s, 1H), 7.39 – 7.32 (m, 3H), 7.27 – 7.16 (m, 4H), 6.96 (d, J = 5.7 Hz.

1H), 5.04 (dd, J = 15.5, 6.8 Hz, 1H), 3.93 (q, J = 6.9 Hz, 1H), 3.39 (dd,

J = 8.3, 5.0 Hz, 1H), 3.11 – 2.96 (m, 2H), 1.69 (dt, J = 12.9, 6.5 Hz,

1H), 1.50 (ddd, J = 13.7, 8.6, 5.0 Hz, 1H), 1.36 (ddd, J = 14.2, 8.7,

5.8 Hz, 1H), 0.92 (dd, J = 6.3, 5.8 Hz, 6H). ¹³C NMR (126 MHz, CDCl₃)

δ 172.77, 135.50, 133.62, 132.72, 131.22, 130.66, 129.30, 129.01,

128.04, 127.02, 126.34, 123.54, 122.96, 117.72, 63.16, 62.87, 62.58,

chloride (1 mmol, 1 equiv.) in dry methanol (11 mL), under argon at-

́

mosphere, potassium carbonate (1.2 mmol, 1.2 equiv.) and 3/4-bromo-

2.2.2-triﬂuoroacetophenone (1.1 mmol, 1.1 equiv.) were added. The

mixture was stirred at 50 °C for 18 h, then it was ﬁltered to remove

potassium carbonate and the solvent was evaporated to get white or

slightly yellow gummy solid. The residue was used for the next step

without further puriﬁcation. The imine intermediate in a round bottom

ﬂask under argon was solubilized in MeOH then ACN was added. The

temperature was decreased to −40 °C before adding freshly synthesized

Zn(BH)₄drop wise. The reaction was stirred for 3 h at −40 °C. After

that. water was added drop wise until the bubbling stops. 2 M HCl was

added until pH = 2. Then the mixture was extracted with ethyl acetate

(3 × 15 mL). The combined organic phases were dried over Na₂SO₄,

ﬁltered and evaporated under reduced pressure. The product was then

puriﬁed by ﬂash silica column chromatography using 20–30% ethyl

acetate in hexane to get a white solid (55–60% yield over 2 steps).

62.29, 59.55, 42.33, 40.87, 38.70, 24.78, 22.93, 21.75. IR (ﬁlm, cm⁻¹

)

3319.50, 3030.41, 2959.80, 2928.35, 2872.20, 2249.01, 1672.25,

1568.38, 1519.56, 1477.32, 1469.41, 1356.40, 1318.68, 1268.36,

1233.84, 1196.91, 1179.41, 1168.11, 1155.61, 1124.17, 1075.59,

23

921.49, 897.93, 875.32, 793.05, 718.47, 701.90. [α]_D(Chloroform)

−158.73.

2.2.2. General procedure for the synthesis of peptide bond (GP2)

The amine synthesized via GP1 (0.7 mmol, 1.0 equiv.). HATU

(0.875 mmol, 1.25 equiv.) and (S)-2-amino-3-phenylpropanenitrile

(0.79 mmol, 1.1 equiv.) were sequentially introduced into a 25 mL

round bottomed reaction ﬂask. provided with magnetic stirrer and

under argon atmosphere. Then, dry DMF (7 mL) and N.N-diisopropy-

lethylamine (1.75 mmol, 2.5 equiv.) were added. After 22 h at room

temperature, ethyl acetate (30 mL) was added to the reaction and the

organic layer was washed with saturated aqueous NaHCO₃(3 × 20 mL)

and Brine (3 × 20 mL). The organic layer was dried over Na₂SO₄,

ﬁltered and dried under reduced pressure. Puriﬁcation by ﬂash column

chromatography on silica (ethyl acetate: n-hexane 2:8 v/v) gave the

peptide as a white solid (60–70% yield).

(S)-2-(((S)-1-([1.1′-biphenyl]-3-yl)-2.2.2-triﬂuoroethyl) amino)-N-((S)-1-

cyano-2-phenylethyl)-4-methylpentanamide 4a. Compound 4a was

synthesized according to procedure GP1 followed by GP2 and GP3 as

a white gummy solid (31.1% yield over 4 steps). Purity (HPLC) 97%. ¹H

NMR (500 MHz, CDCl₃) δ 7.66 – 7.55 (m, 3H), 7.52 – 7.43 (m, 3H),

7.39 (t, J = 7.4 Hz, 1H), 7.34 – 7.25 (m, 3H), 7.23 (d, J = 7.6 Hz, 1H),

7.18 (d, J = 6.5 Hz, 2H), 7.06 (d, J = 8.7 Hz, 1H), 5.02 (dt, J = 8.7,

6.8 Hz, 1H), 4.06 (dd, J = 9.6, 4.6 Hz, 1H), 3.46 – 3.34 (m, 1H), 3.06 –

2.93 (m, 2H), 2.06 (s, 1H), 1.79 – 1.68 (m, 1H), 1.53 (ddd, J = 13.7,

8.8, 4.8 Hz, 1H), 1.38 (ddd, J = 14.2, 8.8, 5.6 Hz, 1H), 0.93 (dd,

J = 12.3, 6.6 Hz, 6H). ¹³C NMR (126 MHz, CDCl₃) δ 173.37, 142.10,

140.31, 134.28, 133.68, 129.52, 129.27, 129.01, 128.95, 128.89,

128.77, 128.27, 127.95, 127.71, 127.24, 127.06, 127.00, 126.52,

124.27, 122.02, 117.79, 63.76, 63.53, 63.30, 63.08, 59.26, 42.55,

40.91, 38.77, 24.83, 23.04, 21.78. IR (ﬁlm, cm⁻¹) 3326.75, 3063.09,

3032.71, 2957.93, 2931.54, 2871.00, 2244.44, 1664.38, 1600.29,

2.2.3. General procedure for the synthesis of CeC bond via suzuki reaction

(GP3)

A stream of argon was passed through a suspension of aryl bromide

3e (0.43 mmol, 1.0 equiv.) synthesized via GP1 followed by GP2, the

proper aryl/heteroaryl boronic acid (0.56 mmol, 1.3 equiv.), 2 M

Na₂CO₃(1.52 mL) and DMF (5 mL) for 10 min. PdCl₂(dppf) (5 mol %)

was then added and the reaction was warmed to 80 °C and stirred under

argon. After 3 h, H₂O (15 mL) was added and the aqueous phase was

extracted with ethyl acetate (3 × 15 mL). The combined organic phases

were dried over Na₂SO₄and puriﬁed by ﬂash column chromatography

using a gradient of ethyl acetate/n-hexane (3:7 to 8:2) to give a light

yellow or white solid with a 50–75% yield. Final compounds were

1496.12, 1480.88, 1455.80, 1367.40, 1345.77, 1261.58, 1154.37,

23

1116.89, 851.91, 801.88, 757.65, 699.22. [α]_D

−173.00.

(Chloroform)

Synthesis of (S)-N-((S)-1-cyano-2-phenylethyl)-4-methyl-2-(((S)-2.2.2-

triﬂuoro-1-(3′.4′.5′-trimethoxy-[1.1′-biphenyl]-3-yl)ethyl)amino)

pentanamide 4b. Compound 4b was synthesized according to procedure

GP1 followed by GP2 and GP3 as a white solid (33.1% yield over 4

steps); mp. 50.9–51.6 °C. Purity (HPLC) 97%. ¹H NMR (500 MHz,

3

L. Cianni, et al.

BioorganicChemistry101(2020)104039

CDCl₃) δ 7.57 (d, J = 8.2 Hz, 1H), 7.43 (dd, J = 12.5, 4.6 Hz, 2H), 7.34

– 7.26 (m, 3H), 7.23 (d, J = 7.7 Hz, 1H), 7.16 (d, J = 6.3 Hz, 2H), 7.04

(d, J = 8.8 Hz, 1H), 6.76 (s, 2H), 5.00 (dt, J = 8.7, 6.9 Hz, 1H), 4.05

(dd, J = 8.7, 4.4 Hz, 1H), 3.92 (d, J = 10.9 Hz, 9H). 3.45 – 3.39 (m,

1H), 2.93 (dd, J = 7.7, 4.9 Hz, 2H), 2.10 (s, 1H), 1.74 (ddd, J = 12.7,

8.6, 6.5 Hz, 1H). 1.51 (ddd, J = 13.7, 8.8, 4.8 Hz, 1H). 1.38 (ddd,

J = 9.9, 8.9, 5.6 Hz, 1H), 0.94 (t, J = 6.6 Hz, 6H). ¹³C NMR (126 MHz,

DMSO‑d₆) δ 174.91, 153.60, 140.69, 137.61, 136.08, 135.63, 135.45,

129.50, 128.90, 127.73, 127.20, 119.06, 104.69, 62.99 (q,

J = 20.0 Hz), 60.63, 59.88, 56.40, 42.80, 41.85, 37.44, 24.58, 23.00,

22.45. IR (ﬁlm, cm⁻¹) 3324.80, 2956.12, 2928.74, 2870.00, 2854.12,

1681.95, 1579.17, 1710.87, 1485.75, 1464.15, 1406.05, 1346.16,

Table 2

K

_Mand V_maxobtained for the ﬁve CPs through the Michaelis-Menten equation.

The experimental K_Mresults retrieved from BRENDA [25] website are: Cz, 1 to

3 µM; LmCPB, 5 to 17 µM; CatL, 0.7 to 6 µM; CatS, 14 to 42 µM; CatK, 1.5 to

4.6 µM.

Enzyme

K_M(µM)

V_max(µmol g⁻¹

k_cat(s⁻¹

)

k_cat/K_M(M⁻¹

s⁻¹

)

s⁻¹

)

Cruzain 1.8

0.2

0.9

0.2

0.5

7.4

81

0.6

8.5

0.8

1.5

0.7

0.2

2.1

0.3

0.1

0.06

0.01

0.22

0.02

0.04

11.1 × 10⁴

31.3 × 10⁴

9.1 × 10⁴

0.3 × 10⁴

LmCPB

CatL

6.7

3.3

37

11

CatS

4.6

CatK

1.7

0.07 2.2

0.01 3.5 × 10⁴

23

1251.30, 1156.50, 1122.58, 1005.52, 796.58, 701.20. [α]_D

(Chloroform) −192.31.

2.3.2. Determination of the inhibition constant (K_i)

The enzymatic activity was evaluated by ﬂuorimetric assays (Biotek

Synergy HT), monitoring the hydrolysis rate of the ﬂuorogenic sub-

strate with ﬂuorescence emission at 460 nm (excitation at 355 nm) at

37 °C. The same protocol described in Table 1 was used for the enzyme

activation and concentration steps. The concentration of the substrates

used in the assays was ﬁxed and kept equal to the K_Mvalue, so

[S] = K_M.

Synthesis of (2S)‐N‐(1‐cyano‐2‐phenylethyl)‐4‐methyl‐2‐{[(1S)‐2.2.2‐triﬂuoro

‐1‐[3‐(pyrimidin‐5‐yl)phenyl]ethyl]amino}pentanamide 4c. Compound 4c

was synthesized according to procedure GP1 followed by GP2 and GP3

as a white solid (36.1% yield over 4 steps); mp. 50.1–50.9 °C. Purity

(HPLC) 97%. ¹H NMR (500 MHz, CDCl₃) δ 9.23 (s, 1H), 8.93 (s, 2H), 7.61

– 7.55 (m, 1H), 7.52 (t, J = 7.7 Hz, 1H), 7.43 (s, 1H), 7.38 – 7.25 (m, 4H),

7.22 – 7.18 (m, 2H), 7.11 (d, J = 9.0 Hz, 1H), 5.07 (dt, J = 9.0, 6.6 Hz.

1H), 4.05 (q, J = 7.1 Hz, 1H), 3.42 (dd, J = 8.9, 4.8 Hz, 1H), 3.00 (d,

J = 6.6 Hz, 2H), 1.81 – 1.68 (m, 1H), 1.51 (ddd, J = 13.7, 8.8, 4.8 Hz,

1H), 1.39 (ddd, J = 14.2, 8.9, 5.7 Hz, 1H), 0.94 (dd, J = 6.6, 4.2 Hz, 6H);

¹³C NMR(126 MHz, CDCl₃) δ 173.21, 157.63, 154.99, 135.31, 135.12,

133.66, 130.20, 129.35, 129.30, 129.01, 128.99, 128.95, 128.69, 128.11,

127.96, 126.96, 126.37, 125.24 (q, J = 283.1 Hz), 124.12, 117.82, 63.37

(q, J = 28.8 Hz), 59.73, 42.77, 40.74, 38.83, 24.82, 23.04, 21.72; IR (ﬁlm,

cm⁻¹) 3324.95, 3032.18, 2957.88, 2933.10, 2870.90, 2243.47, 1681.52,

1555.68, 1496.55, 1455.64, 1408.70, 1367.80, 1348.16, 1262.88,

Visual inspection and a pre-reading of the plate wells were carried

out to check for possible precipitations and background ﬂuorescence,

respectively. None of the compounds showed a signiﬁcant ﬂuorescence

signal at around 460 nm, the emission wavelength used to monitor the

kinetics of the reaction. Thus, the potential eﬀects of the inner-ﬁlter did

not need to be considered in our experiments. Data analysis and ma-

nipulation were performed using Sigma Plot10.

Each experiment was performed in triplicate for each compound.

The initial rates of substrate hydrolysis under the ﬁrst order reaction

were calculated using the Gen5TM Biotek software. The apparent in-

hibition constant K_i′ was determined by non-linear regression ﬁt using

the equation V_s= V_o/(1 + [I]/K_i′), where V_sis the steady-state rate, V_o

is the rate in the absence of inhibitor, and [I] is the inhibitor con-

centration. The true inhibition constant K_iwas calculated by the cor-

rection of K_i′ according to K_i= K_i′/(1 + [S]/K_M), where [S] is the

substrate concentration and K_Mis the Michaelis constant. A control

measurement with the Neq0570 reversible covalent fast-binding in-

hibitor was performed for each conﬁguration plate [13].

23

1155.91, 1119.45, 1029.72, 798.06, 725.42, 700.93; [α]_D(Chloroform)

−355.00.

Synthesis and characterization of compounds 10a and 10b and 10c-

10k are reported in our previous works [9] and [7], respectively.

2.3. Enzymatic inhibition studies

Expression, puriﬁcation of Cruzain and LmCPB are reported in our

recent works [13,24]. CatL, CatK and CatS were purchased by Enzo Life

Science®.

2.4. Isothermal titration calorimetry (ITC)

2.3.1. Determination of the Michaelis constant (K_M)

Dialysis and concentration were performed over 30 to 60 min period

on Amicon ultra 10 kDa membranes (Merck Millipore). The inhibitor

was placed in the syringe and the protein in the cell. 0.001% Triton-X

100 was added to the solution of the syringe and the cell to prevent the

protein from aggregation. All thermodynamic parameters were calcu-

lated by Microcal PEAQ-ITC Analysis Software and experiments were

performed at least in duplicate. The parameters were calculated using

an appropriate one site ﬁtting to model a cruzain one binding site. More

details are reported in our previous work [13].

The enzymatic kinetics assays were performed in ﬂuorimeter

equipment of the Biotek Synergy HT system, using a black 96-well

microplate. (Corning®) at 37 °C. The excitation wavelength used was

360 nm and emission 460 nm. The enzymes were activated in an assay

buﬀer (Table 1), containing 7 mM DTT (Sigma-Aldrich), and 0.014% of

Triton X-100 (Sigma-Aldrich). The activation time and temperature

vary between the enzymes. The concentrations used for the enzymes are

shown in Table 1. The reaction rate was followed by 5 min, at which the

hydrolysis rate of the ﬂuorogenic substrate was measured. A ﬁnal

concentration of 5% (10 µL) of DMSO (Sigma-Aldrich) was used in each

well. Eight diﬀerent substrate concentrations were prepared with a 0.5

dilution factor. The assays were performed in triplicates and the results

are shown in Table 2.

2.5. Diﬀerential scanning calorimetry (DSC)

The thermal stability assay was carried out for cruzain. The T_m,

Table 1

Parameters used in kinetic assays for diﬀerent cysteine proteases.

Enzyme

[Enzyme] (nM)

Substrate

Assay Buﬀer

Activation Time

Temperature of Activation

Cruzain

LmCPB

CatK

0.15

7.0

Z-FR-AMC

Z-LR-AMC

Z-FR-AMC

100 mM sodium acetate pH 5.5

100 mM sodium citrate pH 6.0

15 min

30 min

20 min

60 min

0 °C

0.15

1.9

37 °C

0 °C

CatL

CatS

1.12

37 °C

4

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Scheme 1. General synthesis for compounds 3a-3e and 4a-e. Reagents and conditions: a) HCl·H₂N-P₂-CO₂Me, K₂CO₃, MeOH, 50 °C, 18 h; b) 1 M Zn(BH₄)₂in DME,

ACN/MeOH, −40 °C, 3 h; c) H₂N-P₁-CN, HATU, DIPEA, DMF, r.t., 7–20 h; d) aryl/heteroaryl boronic acid, PdCl₂dppf, Na₂CO₃, DMF, 80 °C, 3 h.

temperature in which half of the protein is in the folded and unfolded

form, was determined along with the ΔH_Tm. The experiment was per-

formed in a diﬀerential scanning calorimetry- Nano DSC of the TA

Instruments. The instrument consists of two cells with a 600 µL of ca-

pacity, being one for the sample solution and the other for the re-

ference.

3. Results and discussion

3.1. Design, synthesis, structure activity relationships (SARs) and molecular

dynamics simulations

In a previous work, we performed an analysis of structure–activity

relationships (SARs) for nitrile-based Cz inhibitors that incorporate a P2

amide replacement based on triﬂuoroethylamine (derivatives of

Odanacatib, Fig. 1). Results unearthed two potent and selective in-

hibitors of Cz (Neq0659 and Neq0820, here numbered as 4d and 4e)

[8]. We observed a strong non-additive eﬀect when inverting the con-

ﬁguration of the P3 substituent and inserting benzyl at P1. Therefore,

we designed and synthesized four new derivatives (3e, 4a-4c) of

Neq0659 and Neq0820 that bear a meta substituent in P3 to evaluate

the inﬂuence of the P3/S3 interaction in the bimolecular recognition

process with diﬀerent CPs. In fact, lately we leveraged the aﬃnity of

dipeptidyl nitrile inhibitors against Cz and CatL by a single atom sub-

stitution in meta position of their P3 phenyl ring [12,13]. The general

synthesis of nitrile-based inhibitors bearing a triﬂuoroethylamine in P2

(3a-3e and 4a-4e) is described in Scheme 1 and chemical structures are

depicted in Fig. 2.

2.5.1. Cruzain reversible assay

Cruzain was thawed at room temperature for approximately 5 min.

After that, the protein was added in a dialysis membrane (Amicon

Ultra) of 15 mL and 10 kDa, along with a buﬀer solution of 100 mM

sodium acetate, 300 mM of NaCl and 5 mM of EDTA pH 5.5 and 1 mM

of β-mercaptoethanol (Sigma-Aldrich). The solution was centrifuged

(Eppendorf 5804R) at 4500 rcf at 4 °C for 35 min. Afterward, the so-

lution was transferred to a micro centrifuge tube and kept on ice. The

Cz concentration was measured in mg mL⁻¹using a DeNovixDS-11⁺

spectrometer; the measurement was repeated three times.

The samples for the reversibility assay were prepared with a ﬁnal

volume of 1000 µL. The sample solution was prepared with a Cz con-

centration of 30 µM (0.7 mg mL⁻¹), acetate buﬀer pH 5.5, 0.014% v/v

of Triton X-100, and 5% of DMSO. The reference cell was prepared with

the acetate buﬀer pH 5.5, 0.014% v/v of Triton X-100, and 5% of

DMSO. Both samples were degassed for 5 min before being added to the

equipment cells.

For a comprehensive approach to SARs, we included our in-house

synthesized dipeptidyl nitrile inhibitors (10a-10k), whose binding af-

ﬁnities (pK_i) for a panel of CPs were published later [7]. The general

synthesis of compounds 10a-10k is reported in Scheme 2 and their

chemical structures in Fig. 3.

The experiment was performed with sequential scans at 3 atm and

with a rate of 2 °C/min. First, the sample was heated from 0 to 90°, then

it was cooled to 0 °C and again heated to 90 °C. The experiment was

performed in duplicates.

The inhibition constants scaled to pK_ivalues were determined for

parasite cysteine proteases (Cz, LmCPB) and for human cysteine ca-

thepsins (CatK, CatL, CatS) and are reported in Table 3. Compounds

10a and 10c have previously been described as competitive inhibitors

by reversible binding to Cz [9]. Several new compounds are low na-

nomolar inhibitors for most of the CPs now reported. The utilization of

these CP inhibitors outspreads to candidates for antiprotozoal action

and to cysteine cathepsin inhibitors of interest in human diseases.

2.5.2. Thermal stability of cruzain with diﬀerent ligands

The dialysis of the protein was made in the same way described

before. The samples were prepared with a ﬁnal volume of 1000 µL. The

sample solution was prepared with a Cz concentration of 30 µM

(0.7 mg mL⁻¹), acetate buﬀer pH 5.5, 0.014% v/v of Triton X-100, 5%

of DMSO and with a ligand concentration of 120 µM. The reference cell

was prepared with the acetate buﬀer pH 5.5, 0.014% v/v of Triton X-

100, and 5% of DMSO. Both samples were degassed for 5 min before

being added to the equipment cells.

3.1.1. Structure activity relationships (SARs) and molecular dynamics

simulations of known potent cruzain inhibitors as Cross-Class CP inhibitors

Matched molecular pair (MMP) analysis provides the opportunity to

evaluate molecular modiﬁcations for signiﬁcant structure–activity re-

lationships (SAR). The analysis was performed in molecular pairs that

diﬀer only in positions where there is a promising chemical transfor-

mation. In this approach, transformations occur by identifying the

change from one functional group to another in which its re-

giochemistry is also considered. MMP analysis can be useful to interpret

bimolecular recognition process and the individual contributions of

The blank was made for all the compounds, where the protein was

removed from the sample cell. The experiment was performed in du-

plicates.

The experiment was performed with a single scan at 3 atm and with

a rate of 2 °C/min, where the samples were heated, ranging from 0 °C to

90 °C.

5

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 2. 2D Structure representation of compounds 3a-3e and 4a-4e.

each transformation in that process [26].

evaluated as shown in Fig. 4.

The MMP/SAR analysis [26] considered the diﬀerences in aﬃnity

related to a structural transformation, deﬁned as ΔpK_i. The ΔpK_iis

calculated through the diﬀerences in aﬃnity between two molecules

(e.g., ΔpK_i= pK_i[10c] – pK_i[10d]). ΔpK_ivalues inferior to 0.2 log

units were considered not signiﬁcant for aﬃnity, and such transfor-

mation is genuinely characterized as a bioisosteric replacement. For

selectivity, a ΔpK_iequal or higher than 1.0 log units was considered

signiﬁcant. As a starting point for the SAR analysis we have used

compound 10c as a prototypical scaﬀold, which has high structural

similarity with other dipeptidyl nitrile inhibitors co-crystalized with Cz

and CatL [9,27]. Therefore, 8 compounds with diﬀerent substituents at

the P1, P 2, and P3 positions and diﬀerent stereochemistries were

Structural modiﬁcation at the P3 position by altering 3-(tert-butyl)-

1-methyl-1H-pyrazole (10c) to the phenyl group (10a) led to a loss of

aﬃnity for almost all CPs except CatS. The diﬀerence in aﬃnity can be

explained, for instance, by a van der Waals interaction between the

methyl group of the 3-(tert-butyl)-1-methyl-1H-pyrazole and the Leu69

residue in the S3 pocket of CatL [27]. On the other hand, CatS contains

a Phe70 residue in the S3 pocket which can perform π-π stacking in-

teractions with the phenyl ring of the inhibitor without having enough

space to accommodate the bulkier 3-tert-butyl-N-methyl-pyrazole group

[28].

Modiﬁcation in P2 by changing Phe (10c) to Leu (10d) led to an

aﬃnity gain for Cz and CatS with the highest aﬃnity observed for CatK

Scheme 2. General synthesis of compounds 10a-10k. Reagents and conditions: a) Isolbutyl chloroformate, DIPEA, DMF, argon, −30 °C, 0.5 h, 2 M NH₄Cl, r.t., 20 h;

b) DIPEA, THF dry, TFAA, 0 °C → r.t., 2 h; c) HCO₂H, r.t., 5–18 h; d) N-boc-P2-CO₂H, HATU, DIPEA, DMF, r.t., 7–20 h; e) P3-CO₂H, HATU, DIPEA, DMF, r.t., 7–20 h.

6

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 3. 2D Structure representation of compounds 10a-10k.

Other structural modiﬁcations at the P1 position were investigated

among pairs of compounds, as shown in Fig. 5.

Table 3

Enzyme inhibition. pK_i= − log₁₀(K_i/M). ^aThe standard deviation was lower

than 15% for all reported pKi values. pKi values of compounds 10a-10k were

reported recently [7]. N/D = Not determined.

The modiﬁcation in P1 from 10k to 10j led to a strong gain in af-

ﬁnity for Cz, LmCPB and CatS, while for CatL and CatK there was no

meaningful diﬀerence in pK_ivalues. Noteworthy, an activity cliﬀ [30]

was observed for CatS when going from 10k to 10j. Transformation for

the inclusion of the benzyloxy group in P1 resulted in a CatS aﬃnity

gain that exceeded 2 log units.

pK_ivalues^a

Compound

Cz

LmCPB

CatK

CatL

CatS

3a

6.1

7.3

8.7

7.6

8.8

8.4

8.3

8.8

9.2

8.7

6.7

7.3

7.8

6.3

7.7

7.5

6.5

7.9

7.1

5.8

6.9

8.6

7.7

8.3

7.8

7.3

8.6

9.1

8.4

6.6

6.8

7.1

7.3

6.1

7.8

7.2

7.4

6.4

7.7

6.7

5.4

7.0

8.6

6.9

8.7

8.3

7.3

8.5

N/D

7.6

6.4

7.2

6.5

8.3

5.6

6.3

6.5

6.7

6.0

7.8

6.8

7.4

8.8

8.5

9.4

8.0

6.9

8.1

5.8

7.6

7.4

7.7

8.2

7.6

6.9

8.5

8.3

7.0

7.2

7.0

7.2

7.6

8.7

8.2

8.4

N/D

7.9

8.4

N/D

8.4

7.3

8.0

6.8

7.4

5.9

7.3

7.2

6.9

6.0

7.3

5.0

Recently, we have demonstrated that the description of covalent

and noncovalent states of covalent inhibitors of CatL are important to

describe the reversible covalent binding of dipeptidyl nitrile inhibitors

[12]. In this work, we perform MD simulations for 10k and 10j in-

hibitors in complex with CatS. We have selected these two compounds

for MD simulation, since they present the so-called activity cliﬀ. Note

that we have explored only the noncovalent state for both 10k-CatS and

10j-CatS complexes to investigate if P1 structural modiﬁcation can in-

ﬂuence the noncovalent binding mode of these ligands and conse-

quently inﬂuence the covalent bond formation with Cys25 residue in

the S1 pocket of CatS. Interestingly, after 60 ns of MD simulation the

10k left the active site of CatS, while 10j remain in the active site along

100 ns of MD simulation (Fig. 6). It is important to point out that the

nitrile group of 10j is close to Cys25 residue, which suggests that the

10j is well positioned in the active site of CatS (see Fig. S2) and this

ligand is more likely to form a covalent bond with Cys25 residue. On

the other hand, the formation of the covalent complex is less likely to

occur for the process involving CatS and 10k ligand. The binding pro-

cess of covalent inhibitors consists of multiple steps. CatS and its in-

hibitor form a non-covalent complex, followed by the approach of the

warhead to Cys25 and, ﬁnally, the formation of the covalent bond. We

thereby can assume that the noncovalent state of 10k with CatS does

not allow the formation of the covalent bond, yielding to a lower aﬃ-

nity (pK_i). Indeed, analyzing the pose of 10j at 15 ns (Fig. 6) the ben-

zyloxy group is buried into a region of aromatic amino acids (e.g.

Phe145, Trp186) and an arginine (Arg142) which can help to hold the

compound in the active site (see Fig. S4). The free hydroxyl group (10k)

is detrimental to the formation of the non-covalent complex, since it

does not provide any help to keep the compound close to Cys25.

Also, noteworthy, the 3b → 3c transformation resulted in steep

aﬃnity gain for the ﬁve CPs. To investigate the eﬀect of this transfor-

mation, we explored its matched molecular pair by the deconstruction

of one-digit nanomolar pan inhibitor 3c to the moderate CP inhibitor

10a through a double SAR cycle (Fig. 7). The double SAR cycle reveals

3b

3c

3d

3e

4a

4b

4c

4d

4e

10a

10b

10c

10d

10e

10f

10g

10h

10i

10j

10k

(+1.8 log units). CatL was negatively aﬀected as expected due to its

preference for aromatic residues in the P2 position.

Relevant structural modiﬁcations in P1 are observed for ﬁve com-

pounds. Not surprisingly, the (S) conﬁguration of the moiety in P1 was

preferred rather than the (R) conﬁguration, as also observed in a recent

study for amino acids substitution in P2 [29]. The substitution of the

cyclopropane for aromatic moieties as the (S)-benzyl (10f) increased

the aﬃnity values for all CPs. The exchange of the cyclopropane for the

isopropyl group (10 h) did not lead to any improvement.

The substitution of the cyclopropane to the pyridine group (10g)

maintains the aﬃnity for all the CPs. Despite no signiﬁcant change in

aﬃnity, pyridine raises the solubility of the inhibitor without modifying

the main hydrophobic interaction with the target protein. This ﬁrst

MMP/SAR (Fig. 4) revealed that Leu in P2 and (S)-benzyl moiety in P1

are privileged structures for the design of new cross-class inhibitors

amongst diﬀerent CPs.

7

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 4. MMPA/SAR summary for compound 10c. Values are reported as diﬀerences in pK_ivalues and are color-coded as red (negative), green (positive), grey (no

signiﬁcant diﬀerence, ΔpK_i< 0.2). Modiﬁcations in P1 are highlighted as orange, P2 blue and P3 purple. (For interpretation of the references to color in this ﬁgure

legend, the reader is referred to the web version of this article.)

a non-additive eﬀect on SAR, which is particularly valuable for testing

and validating diﬀerent interaction modes [31]. The values of inhibi-

tion constants are shown on the left side of compound 10a and on the

right side of 3c for all studied CPs. Table 4 presents all pK_ivalue dif-

ferences for the respective transformations shown in Fig. 7.

hand, the transformation [10b → 3b] produced a less negative eﬀect

for most of the CPs. The substitution of the Phe for Leu in P2 [3a → 3b],

while maintaining the CF₃group, produced an increase in aﬃnity of

almost 100 times. Therefore, the combination of the CF₃group with the

Leu at the P2 position has a superadditive eﬀect on aﬃnity for all the

CPs.

The upper part of the cycle consists of a semi-ﬂat SAR because there

are no signiﬁcant gains in aﬃnity. At the same time, all transformations

observed in this part of the cycle are superadditive. The structural

transformation related to the ΔpK_iof [10a → 10b] were below 1.0 log

units. The individual transformation of [10a → 3a] diminished the

aﬃnities of the ﬁve CPs. These losses are mainly observed for Cz,

LmCPB, CatK and CatL, but not for CatS whose decrease was only 0.1

log unit. This result suggests that the bioisosteric replacement of the

C]O bond to the CF3 group is better tolerated in CatS. On the other

For the lower part of the cycle, which leads to 3c from 3a, the

structural transformations for the ﬁve CPs were all additive as the

diﬀerences in ΔpK_iwere between

0.1/0.2 log units, all having been

considered real bioisosteric exchanges without increased aﬃnity. A

massive improvement in aﬃnity was obtained when observing the

lower part of the cycle. One could justify these phenomena by con-

sidering that when the CF₃is implemented, the mode of binding (MoB)

of the inhibitors change, allowing a better accommodation of P1, P2

Fig. 5. MMPA/SAR summary for compounds 10k

and 3b. Values are reported as diﬀerences in pKi and

are color-coded as green (positive) and grey (no

signiﬁcant diﬀerence, ΔpKi < 0.2). Modiﬁcations

in P1 are highlighted as orange. (For interpretation

of the references to color in this ﬁgure legend, the

reader is referred to the web version of this article.)

8

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 6. Mode of binding of compounds 10j

and 10k. Snapshot was taken at 95 ns of a

100 ns (non-covalent MD simulation).

Hydrogen bonds, when occur with Asp163

and Gly69, are represented in dashed lines.

Warhead is highlighted as blue. The dis-

tance (Å) ﬂuctuations between nitrile

carbon with sulfur of Cys25 were also re-

presented on the right side. The backbone of

CatS is light green. The main residues and

carbon chains are greys, nitrogen is blue,

oxygen is red, sulfur is yellow. a) 10k-CatS

non-covalent complex. b) 10j-CatS non-

covalent complex, the interaction of the ni-

trile carbon with sulfur from Cys25 is also

represented. (For interpretation of the re-

ferences to color in this ﬁgure legend, the

reader is referred to the web version of this

article.)

and P3 moieties in the subsite pockets leading to the strong cross-class

inhibitor (3c) with a K_iof about 3 nM for all the ﬁve CPs. Nonetheless,

it is worth mentioning that the chemical transformations implemented

in the dipeptidyl nitrile scaﬀold 10a that resulted in the elaboration of

the non-peptide inhibitor 3c, had an average aﬃnity gain of more than

60 times in all studied CPs. The most striking aﬃnity gain of over 100

times is not only observed for Cz and LmCPB (the two parasitic targets),

but also for CatK. It is also important to mention that the transformation

3a → 3c resulted in a gain of 2.6 log units leading 3c to a synergistic

gain of about 400 times over 3a after only a small cycle of molecular

modiﬁcations.

observed for the pair [4e → 3e] can be addressed by an halogen bond

interaction with the Gly61 [12]. It is noteworthy that while para 5-

phenylpyrimidine in P3 is a privileged moiety for maintaining aﬃnity

between protozoa and mammalian CPs, meta-substitution appears to

play a particular role amid cross-class inhibitors of diﬀerent CPs, such

as 3e and 4c. It is also striking that all CPs lost aﬃnity for compound 4b

that bears the 3,4,5-trimethoxy-1,1′-biphenyl moiety.

Molecular dynamics simulations were performed to better in-

vestigate the putative mode of binding of 4a, 4c, 4d and 4e derivatives

when bound to Cz. It is known from experimental data that dipeptidyl

nitrile-like molecules form hydrogen bonds to the backbone NH and

C]O groups of Gly66 and Asp161 in the Cz active site [9]. The mean

distances of these key interactions computed during 100 ns of MD for

4d in complex with Cz are depicted in Fig. 9, for the other ligands these

mean distances are shown in SI.

3.1.2. Structure activity relationships (SARs) and molecular dynamics

simulations of new Non-Peptidic nitrile based Cross-Class CP inhibitors

Recently, we have outlined that the para-biphenyl moiety that is in

the P3 position of a triﬂuoroamine derivative is a privileged structure

for leveraging Cz aﬃnity while gaining selectivity over CatL [8].

However, due to the undesirable physicochemical properties of the

compound, such as low water solubility and high lipophilicity, we

modiﬁed the aryl ring by a pyrimidine to introduce 4e (Fig. 8). To

improve its water solubility while maintaining aﬃnity for Cz, 4e is still

signiﬁcantly selective (1.1 log units) over CatL. Compound 4e was

thereby re-engineered to include the 5-phenylpyrimidine moiety giving

4c (Fig. 8). The synthesis was performed by following the same pro-

cedure for compound 4e [8]. None of the implemented modiﬁcations

led to a considerable increase in aﬃnity for protozoan CPs. On the other

hand, the meta-bromine (3e) in P3 produced a strong increment in af-

ﬁnity for both CatL and CatK. The aﬃnity cliﬀ for CatL inhibition

The average distances obtained in the MD simulations show that

Gly66 is positioned to form two hydrogen bonds with the hydrogen of

the triﬂuoroethylamine group and the oxygen of the amide bond. Too,

Asp161 interacts with the ligands through hydrogen bonding with the

hydrogen of the amide bond in P2/P1 position. Gly66 and Asp161 re-

sidues form hydrogen bonds to all ligands in the covalent complex and

maintain their stability during 100 ns of the molecular dynamics. This

result shows that ligands maintain the same key interactions at the

active site of Cz.

Fascinatingly, visual inspection of molecular dynamics simulations

brings up two putative modes of binding (MoB) for all four compounds.

In the ﬁrst MoB I (see Fig. 10 compound 4c), P3 is in contact with the

S3 surface and CF₃is exposed to the solvent. While in the MoB II (see

9

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 7. Double cooperativity cycle for transformations of compounds 10a to 3c for cruzain inhibition. Values are reported as diﬀerences in pK_ifor Cz inhibition.

Green arrows highlight positive trend while the red arrow depicts a negative trend. (For interpretation of the references to color in this ﬁgure legend, the reader is

referred to the web version of this article.)

simulation and therefore the MoB II with the formation of an intra-

molecular interaction is favorable (Fig. S5). Compound 4a, which bears

a meta biphenyl ring in P3 display a strong preference for the MoB II for

over 90% of the simulation where the biphenyl ring is fully solvent-

exposed while the hydrogen in position 2 of the aryl ring is pointing

towards the center of the phenyl ring in P1 with an angle of about 90°.

Similar behavior is displayed by compound 4d, which has a weak in-

teraction with the Ser61 when performing the MoB I, and therefore it

switched to MoB II after 30 ns forming intra-molecular interactions for

the rest of the simulation.

Table 4

ΔpK_ifrom the SAR analysis of Fig. 7. The notation [X → Y] indicates a struc-

tural modiﬁcation. Diagonal transformations are highlighted in orange.

Structural modiﬁcation

Cz

LmCPB

CatK

CatL

CatS

[X → Y]

[10a → 10b]

[10a → 3a]

[10a → 3b]

[10b → 3b]

[3a → 3b]

[3a → 3d]

[3a → 3c]

[3b → 3c]

[3d → 3c]

+ 0.6

−0.6

+0.6

0.0

+0.2

−0.8

+0.3

+0.1

+1.1

+1.9

+2.8

+1.7

+0.9

+0.8

−1.0

0.0

+0.3

−0.6

0.0

+0.7

−0.1

+0.3

−0.4

+0.4

+1.0

+1.5

+1.1

+0.5

− 0.8

+1.6

+1.5

+3.2

+2.2

+1.7

−0.3

+0.6

+1.7

+2.0

+1.4

+0.2

+1.2

+1.5

+2.6

+1.4

+1.1

Since the diﬀerence in pKi values amid these four compounds is

very small, it can be assumed that both putative MoBs contribute po-

sitively to Cz aﬃnity and are in equilibrium with each other. However,

we have recently published a strong non-additivity SAR for Cz inhibi-

tion [8] which can be explained by the presence of MoB I and MoB II.

Indeed, if we consider Neq0658, as molecular pair of compound 4d,

which can just adopt the MoB I (Fig. S12), due to its diﬀerent conﬁg-

uration of the carbon in P3 we experienced a drop of almost three log

unit in aﬃnity with a strong nonadditivity relationship in aﬃnity

(Fig. 12) [8]. Hence, new in-depth studies should be carried out to

better estimate the existence and inﬂuence of the MoB II over the

general bimolecular recognition process.

Fig. 10 compounds 4a, 4d and 4e), the CF3 group is straight towards

S3, while the P3 group is forming an intramolecular π stacking inter-

action between hydrogen in position 2 of the aryl ring at P3 and the

phenyl group at P1 (Fig. 10).

In general, the starting point of the simulation displays the MoB I for

all four compounds. However, compounds 4a, 4d, 4e, switch to MoB II

in < 40 ns and maintain this conformation for the rest of the simula-

tion, while compound 4c maintains the MoB I for 90% of the simulation

(Fig. 11). Hence, the main conﬁguration responsible for the binding

process interaction is MoB II for compounds 4a, 4d, 4e; while for

compound 4c the main bimolecular interaction arisen by the MoB I.

This diﬀerence can be arisen by the possibility of compound 4c, which

bears a pyrimidine ring in meta position, to interact with Ser61 by

hydrogen bonding when forming the MoB I but not when performing

the MoB II. Indeed, 4e which bears a pyrimidine ring in P3 but in para

position show not substantial interaction with the Ser61 over the

3.2. Thermodynamic cycle and ITC evaluation

As already described in our recent study [32], isothermal titration

calorimetry (ITC) is a technique used for the thermodynamic char-

acterization of interactions amongst biomolecules and small molecules,

which is vital for understanding the process of molecular recognition.

Therefore, we have performed an in-depth analysis of the thermo-

dynamic proﬁles for the key compounds with Cz to better estimate how

10

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 8. MMPA/SAR summary for compound 4e. Values are reported as diﬀerences in pKi and are color-coded as red (negative), green (positive), grey (no signiﬁcant

diﬀerence, ΔpKi < 0.2). (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

small chemical modiﬁcation aﬀected the SAR (Table 5) when the

binding energy is dissected in its enthalpic and entropic terms.

The binding energy ΔG is the result of enthalpy (ΔH) and entropy

(ΔS) contributions. Classically, enthalpy reﬂects the strength of the

interactions due to the formation of hydrogen bonds and van der Waals

interaction, while the binding entropy is a combination of the change in

the desolvation and conformational changes after the formation of the

bimolecular ligand-cruzain complex [33].

entropy-structure activity relationships. Fig. 13a shows that ΔG is

mirrored by both enthalpy and entropy, which is reﬂected by the lin-

earity between enthalpy and entropy (Fig. 13b): R²

= 0.928,

R²_adj= 0.921, with clear homoscedasticity seen in Fig. 13c residuals.

Beyond this classical interpretation of EEC, we can observe in the

analysis of the thermodynamic signatures (see Fig. 14a below) that

most of the compounds have, as depicted, a detrimental entropy con-

tribution. Nonetheless, both enthalpy and entropy play a favorable role

for the binding energy of compounds 10g and 4e (red bars), while

displaying an opposite compensation eﬀect which yields to the lowest

values of enthalpy contribution [34].

As can be seen from Table 5, there is a slight variation in the binding

energy values (ΔG) and all of them round to median of −9.8 (se = 0.2

and sd = 0.7). Thus, although all interactions are favorable for Cz-

inhibitor complex formation, optimization of ligands through binding

aﬃnity may not be trivial. Since ΔG binding aﬃnity is determined by

changes in enthalpy and entropy, it is desirable to establish any existing

correlations between enthalpy and entropy to evaluate any compensa-

tion (EEC) as well as the correlations arising from the enthalpy/

The aﬃnity K_a, which is governed by the Gibbs binding energy (ΔG:

K_a= e^{- ΔG/RT}), summed up by enthalpy and entropy (ΔG = ΔH - TΔS),

does not necessarily result in high aﬃnity when their favorable con-

tributions are observable (see Fig. 14a). Although we would much

prefer simultaneous optimization of enthalpy and entropy, it is clear

Fig. 9. Covalent molecular dynamics simulation for 4d-Cz complex and key interactions. a) Snapshot was taken at 85 ns of a 100 ns simulation for 4d-Cz complex and

key interactions in Cz active site. b) Hydrogen bond (Hb) distance (Å) ﬂuctuations between O₁of Gly66 and H of the ligand. c) Hb (Å) ﬂuctuation for H₂––O distance.

d) Hb (Å) ﬂuctuation between O₂of Asp161 and H₁of the ligand. Average values are shown in the box, along with the standard deviation in parenthesis. The

backbone of Cz is blue. The main residues and carbon chain of 4d are grey, ﬂuorine is green, nitrogen is blue, oxygen is red, sulfur is yellow. Hydrogen bonds are

depicted by dashed lines. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

11

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 10. Major putative mode of binding of compounds 4a, 4c, 4d, 4e with Cz. Compound 4c displays a major MoB I. Compounds 4d, 4e and 4a show a preference

for MoB II. Cz surface is depicted in grey. S2 pocket is light blue. S3 pocket is pink. The main residues and carbon chain of ligands are grey, ﬂuorine is green, nitrogen

is blue, oxygen is red, sulfur is yellow. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

from Fig. 14a that the present process is enthalpy-driven (blue bars).

The enthalpy contribution (median of −12.8, se = 0.7, sd = 2.6,

variance = 6.9, blue bars) is stronger than the entropy (median of 3.2,

se = 0.7, sd = 2.6, variance = 7.1, dark blue bars) within this series of

Cz inhibitors. Overall, the Gibbs binding energy remains almost the

same at the median of −9.9, se = 0.2, sd = 0.7, variance = 0.5 – light

blue bars.

(Fig. 14a, light and dark blue and red bars favorable proﬁle scheme –

the enthalpic and entropic parameters contribute positively to the

binding energy). Although the correlation may seem contradictory,

there is a tendency for an inverse correlation between lipophilicity and

topological polar surface area (TPSA). For instance, in a series of 443

cruzain inhibitors retrieved from the ChEMBL database [36], there is a

poor correlation between SlogP and TPSA whose R²is only −0.153. For

the case of a small but signiﬁcant series studied here, this trend appears

also to be valid since the value of R²is −0.376.

The binding entropy on the other hand, being dependent primarily

on the hydrophobic eﬀect seems to characterize the pattern observed

for cruzain inhibitors herein under study via ITC. Fig. 14(b, c, d) shows

the comparative proﬁles amongst compounds that present favorable

and unfavorable entropy for inhibitors that are the main directors of

bimolecular interaction. These Boxplots depict qualitatively a trend that

the two inhibitors with favorable entropy (Fe, cyan Boxplots) for the

composition of Gibbs binding energy, have higher lipophilicity, mole-

cular mass, and topological polar surface area than those observed for

inhibitors with detrimental entropy (De, green Boxplots).

The graph in Fig. 13b above shows the classic representation of the

correlation between −TΔS and ΔH. It is observed a clear linear de-

pendence that encompasses both the detrimental and favorable entropy

contributions (both processes being spontaneous at the studied tem-

perature), while the values of ΔG remains balanced throughout the

series (Table 5). Thus, we wanted to better evaluate such relationships

and the results are shown in Fig. 15.

Fig. 15a represents the observed changes in entropy (favorable-

unfavorable) for bimolecular interactions that are enthalpy-driven (fa-

vorable enthalpic contribution) in relation to Gibbs binding energy. The

bimolecular recognition pattern so depicted, aligns compounds 10g and

4e within the space comprised by favorable entropy inhibitors, with the

“similar” ΔG value centered at −9.5 kcal mol⁻¹. This equivalence in

binding energy escapes EEC, but the nature of the factors that aﬀect it

remains uncertain. The result observed in Fig. 15b suggests that the

ratio (ΔH + TΔS)/ΔG might be a measure of the enthalpic driving

force. The equal contribution of enthalpy and entropy to binding energy

must be zero and positive when the binding is enthalpy-driving.

Overall, the calculated ratio for the study compounds is positive but

It is striking that the binding thermodynamics can discriminate

between inhibitors with good and bad entropy contributions. The

binding aﬃnity observed also depends on the size of the ligands

(Fig. 14c) and this observation appears to be in line with medicinal

chemistry practice of adding new functional groups to improve aﬃnity

[35]. To the same extent, Gibbs binding energy is leveraged to its

median via lower enthalpy with higher entropy gain for inhibitors 10g

and 4e – as compared to the other inhibitors that present detrimental

entropy. Too, high lipophilicity and the highly polar character ex-

pressed by the large polar surface area of 10g and 4e amalgam favor-

ably towards the desired thermodynamic proﬁle of these Cz inhibitors

12

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 11. Distance (Å) ﬂuctuations between hydrogen located in P3 ring and the centroid of the benzene ring in P1. Interaction is represented inside the graphics. 4d

4e 4c 4a. Carbon chain is grey, ﬂuorine is green, nitrogen is blue, oxygen is red, sulfur is yellow. (For interpretation of the references to color in this ﬁgure legend, the

reader is referred to the web version of this article.)

tends to zero for inhibitors 10g and 4e. For the other inhibitors, the

values are greater than 1 and, thus, enthalpy overcompensates the

unfavorable entropic contribution (Fig. 15b). The dimensionless ratio

(ΔH + TΔS)/ΔG dubbed the Enthalpy–Entropy Index (I_{E – E}) is, there-

fore, used to indicate the enthalpy content of the binding interaction.

And, this may shed light directly towards thermodynamically-driven

optimization endeavor. For instance, the inhibitors 3a, 10a and 10 h

whose I_E-Eis greater than 2 (2.57, 2.24 and 2.02, respectively) do not

seem to surpass inhibitors 10g and 4e with I_E-Evalues of 0.705 and

0.499 in terms of their binding energies (ΔG) thereby conﬁrming the

trend that enthalpy optimization is not a trivial process [37].

other, K_iis dependent on the K_Mvalue but K_dis not. K_dmeasures the

balance between the target-ligand complex and its dissociated compo-

nents, the reason for which it is given by the rate of constant dis-

sociation for the complex and the association rate constant (k_-1/k₁). K_i,

on the other hand, measures the reduction in the enzyme catalytic ac-

tivity and, therefore, is dependent on the kinetic mechanism of in-

hibition, which, in the present case, is a competitive mechanism.

To initiate the correlation analysis between pK_iand pK_dwe have

built the plot of the kernel density estimate and histogram for both their

values (pK_iin blue and pK_din pink). As can be seen from Fig. 16a, there

is an overlapping density in both data. However, because hetero-

scedasticity is present, it reduces the precision of the estimates in OLS

linear regression. Also, the means are 7.02 and 7.30, with standard

deviations of 0.512 and 0.817, respectively, for pK_dand pK_ivalues. So,

we cannot employ ordinary least squares (OLS) to ﬁt such data [38].

The scatter plot is shown in Fig. 16b. In order to circumvent this matter,

we have employed a recursive least squares (RLS) method, which is an

The pharmacodynamic characterization of new drug candidates

presupposes the determination of target-binding kinetics. Since the

same set of ligands is used for ﬂuorimetric assays (K_i) and ITC (K_d), it is

reasonable to assume that there exists a correlation between the pK_iand

pK_dvalues. Nonetheless, it is important to note that there are signiﬁcant

biochemical diﬀerences between K_dand K_i. Although related to each

Fig. 12. MMPA for compound 4d. Values are reported as pKi for Cz inhibition.[8] Colored areas highlight the stereochemical modiﬁcations.

13

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Table 5

Thermodynamic parameters of Cz-ligand complexes obtained from ITC ex-

periments. pK_d= −log₁₀(K_d). Uncertainties are shown in parentheses. In blue

are highlighted 4e and 10g for presenting favorable entropy to Gibbs binding

energy.

N°

pK_d

ΔG (kcal mol⁻¹

)

ΔH (kcal mol⁻¹

)

−TΔS (kcal mol⁻¹

)

3a

6.1 (0.07)

7.3 (0.2)

7.5 (0.02)

7.1 (0.03)

6.6 (0.01)

7.5 (0.03)

7.2 (1.2)

7.6 (0.08)

6.2 (0.07)

7.0 (0.1)

7.5 (0.1)

6.5 (0.1)

7.2 (0.09)

−8.3 (0.02)

−10.0 (0.2)

−10.3 (0.0)

−9.7 (0.04)

−9.0 (0.02)

−10.2 (0.07)

−9.8 (0.5)

−10.4 (0.1)

−8.5 (0.07)

−9.5 (0.2)

−10.2 (0.2)

−8.8 (0.02)

−9.9 (0.1)

−14.9 (0.1)

−13.3 (0.3)

−13.5 (0.5)

−7.3 (1.5)

6.6 (0.1)

3.2 (0.1)

3.2 (0.5)

−2.4 (1.5)

5.6 (0.3)

4.3 (0.7)

2.4 (0.04)

5.3 (0.9)

3.4 (0.04)

−1.4 (0.7)

0.5 (0.4)

2.8 (1.2)

0.97 (0.6)

3b

4c

4e

10a

10b

10c

10d

10e

10g

10h

10i

10k

−14.6 (0.3)

−14.5 (0.7)

−12.2 (1.2)

−15.7 (0.08)

−11.9 (0.4)

−8.10 (0.5)

−10.7 (0.6)

−11.6 (1.1)

−10.8 (0.5)

expanding window version of OLS [39]. The RLS statspace regression

model results for ﬁtting data from Fig. 16b is: R²= 0.995 for a pK_d

slope of 1.04 ( 0.0220) with z = 48.4 and P > |z| = 0. The robust

linear model (RLM) [40] yields a similar result (P > |z| = 0 for

z = 111). The RLS model allows the calculation of the recursive re-

sidues with the statistics called the cumulative sum control chart

(CUSUM) [41]. The plotting of these statistics, Fig. 16c, indicates that

the CUSUM statistic (the blue line) does not expand outside the 5%

signiﬁcance limit.

3.2.1. Analysis of enthalpic and entropic contribution by thermodynamic

cycles

The double transformation 10a → 3b (Fig. 7 above) occurs via an

exchange of a carbonyl for a triﬂuoromethyl and phenylalanine group

for leucine with a gain of only 0.6 log units in Cz aﬃnity. However, this

result per se does not bring qualiﬁed information about the real nature

of the transformation in terms of the individual contributions of each

modiﬁed group. An elegant way to evaluate the thermodynamic con-

tributions to the molecular changes implemented on time is through the

analysis of the change in enthalpy and entropy caused by such changes.

As a matter of ease comparison, the relative diﬀerences (ΔΔX_n→m, X

= G, H, and S) of the thermodynamic properties for a given transfor-

mation are shown in Fig. 17. These diﬀerences make it possible to

analyze the distributions of the respective thermodynamic values for

the pairs of inhibitors in such a way that the enthalpy eﬀects and en-

tropy are analyzed diﬀerently from the above sources of correlation. In

addition, the sudden changes brought about by the transformations

oﬀer the opportunity to better assess the disturbance imprinted on the

binding energy via individual group contribution.

In the speciﬁc case, from 10a to 3b two molecular changes were

implemented. Thus, the analysis of enthalpic and entropic contributions

is performed through a thermodynamic cycle (Fig. 17). The two ligands

interact with similar binding aﬃnity with similarly putative modes of

interaction. Despite this, the Gibbs binding energy partition occurs

diﬀerently as seen in the thermodynamic cycle in Fig. 17.

Fig. 13. Enthalpy-entropy compensation (EEC) observed for Cz-inhibitor com-

plexes. (a) ΔG is mirrored by ΔH and −TΔS; (b) linear (R²

= 0.928,

R²_adj= 0.921) compensation between enthalpy and entropy: the most favorable

enthalpy interaction imposes some restriction on the movement of interacting

partners that results in an unfavorable entropic change. In line with it, the two

Cz inhibitors 10g and 4e with favorable entropy help to contribute to the

overall unchanged value of ΔG. In the box-legend to the right, De is for detri-

mental entropy whereas Fe stands for favorable entropy; (c) linear ﬁt of −TΔS

vs ΔH as in (b) but along with the residuals (the right box-legend of black to

blue), which represents the homoscedasticity of the residual entropy. (For in-

terpretation of the references to color in this ﬁgure legend, the reader is referred

to the web version of this article.)

Thermodynamic cycle analysis for the double transformation from

10a to 3b reveals a superadditive eﬀect in the free energy which is

driven mostly by the entropy change contribution (Fig. 17). The single

modiﬁcation in P2 from Phe to Leu (10a → 10b) leads to an indis-

tinguishable diﬀerence in enthalpy change and a positive eﬀect in the

entropy change which can be rationalized by an increase of rotational

bond in the molecule. On the other hand, the bioisosteric replacement

of the carbonyl group for the CF₃(10a → 3a) precludes any gain.

Transformation 3a → 3b results in a gain cliﬀ in the change of entropy

with a large loss in the enthalpic change. Cooperativeness is not seen in

the transformation 10a → 3b despite the detrimental contribution of

14

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 15. ITC representation of favorable vs unfavorable entropy. (a) ITC data for

the favorable enthalpic contribution of Cz inhibitors with favorable (com-

pounds 10g and 4e in the blue volume of the 95% conﬁdence ellipsoid) and

unfavorable entropy (ellipsoid in red) for the other inhibitors studied here; (b)

Plot of the Enthalpy–Entropy Index (I_{E – E}): (ΔH + TΔS)/ΔG ratio in red and ΔG

in blue bars. Red bars 4 and 10 are for Cz inhibitors 4e and 10g, respectively.

(For interpretation of the references to color in this ﬁgure legend, the reader is

referred to the web version of this article.)

the change in enthalpy. The decrease in the enthalpy change con-

tribution while improving the entropy change contribution may result

from a loss in planarity, loss of the oxygen as a hydrogen bond acceptor

and gain in ﬂexibility to better accommodate the inhibitor in the active

site.

On the contrary, the double transformation in P2 and P3 from 10a

to 10d (Fig. 18) unfolds a strong subadditive eﬀect in the free energy

change which is driven mostly by a detrimental entropy contribution.

Chemical substitution of the phenyl ring for the tert-butyl-N-methyl-

pyrazole in P3 (10a → 10c) increases the aﬃnity towards Cz through a

strong entropic eﬀect which could be addressed to the desolvation

process. However, for the same transformation in P3 (10b → 10d),

while having leucine in P2, we observed a detrimental entropy con-

tribution trend that yields to an insigniﬁcant change in the net energy.

A similar observation occurs for transformation 10c → 10d when

changing phenylalanine to leucine in P2. Nevertheless, both transfor-

mations 10a → 10b and 10a → 10c resulted in the gain of the entropic

variation. To this end, we can deﬁne that tert-butyl-N-methyl-pyrazole

in P3 increased the solubility of the inhibitor and formed positive in-

teraction with the S3 region, as shown for CatL [27]. However, this

double positive eﬀect is achieved just if Phe in P2 is substituted with the

less bulky and more ﬂexible Leu.

Fig. 14. (a) Thermodynamic signature to access the magnitude of the ther-

modynamic forces that contribute to the binding energy (ΔG), binding enthalpy

(ΔH) and binding entropy (−TΔS). The proﬁle is typically enthalpy-driven but

the enthalpy–entropy balance for cruzain inhibitors is dominated by detri-

mental contribution from entropy with the exception of 4e and 10g that

modulate bimolecular recognition via favorable enthalpy and entropy. (b, c, d)

Boxplots for some drug-like properties of the selected Cz inhibitors with fa-

vorable and unfavorable entropy, and favorable enthalpy contribution. (b)

SLogP = calculated logarithm of the partition coeﬃcient in octanol, (c)

ExactMW = molecular mass (Da) and (d) TPSA = topological polar surface

area (A²). Classes De and Fe are for the statistical binary classiﬁcation as det-

rimental entropy (De) and favorable entropy (Fe), respectively.

3.2.1.1. Insight of single modiﬁcation in P1 and P3 of Cz inhibitors by

ITC. Recently [42], we evaluated the mechanistic reasons for the P1

modiﬁcation in some dipeptidyl nitriles as Cz inhibitors, which resulted

in an enthalpy inhibition process with a detrimental contribution from

entropy. The inclusion of cyclopropane in carbon-1 of 2-

aminoacetonitrile was a detrimental enthalpic change with favorable

15

L. Cianni, et al.

BioorganicChemistry101(2020)104039

rigid cyclopropane in P1 leads to a greater degree of ﬂexibility and

subsite polar interaction. However, as already discussed [42], cyclo-

propane helps promote the Cys nucleophilic attack, by pointing the

nitrile closer to the nucleophilic thiol.

Finally, we have studied the regiochemistry eﬀect of the 5‐phe-

nylpyrimidine moiety in P3 when going from para (inhibitor 4e) to meta

(inhibitor 4c), Fig. 20. The change in P3 shows a mild increase in

binding aﬃnity because of an improvement in enthalpy counter-

balanced by detrimental enthalpy–entropy compensation. Molecular

dynamics simulations (see 3.1) explains this phenomenon by con-

sidering an improvement of the interaction of the pyrimidine ring with

Ser61 when it is in the meta position.

Due to solubility issues of some compounds we were not able to

analyze the aﬃnity by ITC. Therefore, we have used the Diﬀerential

Scanning Calorimetry (DSC) method to quantify the thermal stability of

Cz in the apo form and with a series of inhibitors (see 3.3).

3.3. DSC analysis

Experimental results obtained for Cz in the apo form and when in-

cubated with 13 compounds by diﬀerential scanning calorimetry (DSC)

are reported in Table 6.

First, a reversibility study of the thermal unfolding of Cz was per-

formed, seeking to identify if the process of unfolding was reversible or

irreversible. The result obtained is shown in Fig. 21.

By the sequential scans, it was determined that Cz has an irrever-

sible unfolding due to the non-appearance of a peak in the second

heating scan. This result is in agreement with previous studies made for

papain and other CPs of this family [43].

Having established the irreversible unfolding of the protein, the

assay of the native protein was carried out to determine its temperature

of melting (T_m) (see SI chapter S4). The presence of one peak is in-

dicative of the protein structure cooperativity. Cz has a two-domain

structure and because only one peak appeared in the scan, both do-

mains unfold cooperatively. A positive ΔCp was observed for Cz with a

mean value of approximately + 6.8 kcal mol⁻¹K⁻¹. A positive ΔCp

indicates that the Cp of the unfolding state is greater than the folded

state, which is mainly due to the exposure of hydrophobic amino acids

to the solvent [44]. Calorimetric enthalpy (ΔH^cal) was obtained by in-

tegrating the area under the peak before the baseline subtraction (see

Fig. S13); the value corresponds to the enthalpy at the Tm. The ΔH^cal

consists of endothermic and exothermic contributions correlated to the

bonds and interactions disrupted by the heating process. The Van’t Hoﬀ

enthalpy (ΔH^VH) was achieved when the ﬁt of the raw data was done

using a two-state scaled model equation. This equation considers the

equilibrium constant (K_eq), which takes both states of the protein in

perfect equilibrium.

Fig. 16. (a) A plot with a kernel density estimates and histogram for pK_i(in

blue) and pK_d(pink) data. (b) Scatter plot model of pK_iversus pK_d. Regression

for data in (b) was obtained from the recursive RLS model whose stats are:

R²= 0.995 for pK_d= 1.0402 ( 0.022) with z = 48.355 and P > |z| = 0.000.

The CUSUM statistic, the blue line in (c), does not expand outside the 5%

signiﬁcance bands. Models were run in python scripts. (For interpretation of the

references to color in this ﬁgure legend, the reader is referred to the web ver-

sion of this article.)

The ratio ΔH^cal/ ΔH^VHof the enthalpies was obtained aiming to

identify if the unfolding process of Cz follows a two-state model,

without the formation of intermediates. As shown in Table 6, the ratio

ΔH^cal/ ΔH^VHis very close to 1.0 and the standard deviation of the

analysis indicates that both enthalpies can be equal, and the process of

unfolding can be considered of two states with an equilibrium constant

equal to 1. So, as the enthalpy values are close, a thermal unfolding

without the presence of intermediates was considered for Cz [45].

Afterwards, the analysis proceeded to the investigation of the thermal

unfolding of Cz in the presence of diﬀerent ligands (Table 6).

The T_mvalues are higher than that obtained for the native protein,

showing that the ligands are interacting with the folded protein, thus

stabilizing the Cz structure. The positive shift indicates that the equi-

librium constant was displaced to the folded form due to the stabili-

zation of the Cz structure. Because of that, when Cz is bound to the

ligands the thermal unfolding probably occurred through the presence

of an intermediate and the two domains may unfold separately but not

independently as only one peak was observed.

entropic change. Thus, in this work, we studied changes in P1 to better

assess the role of other molecular fragments replacement. Structural

modiﬁcations in P1 starting from compound 10c display a clear trend: a

strong positive eﬀect in the entropic contributions (excluding 10e and

10i), which is balanced by a decrease in enthalpy (Fig. 19).

Inhibitors 10g, 10 h and 10k experienced almost complete EEC

thanks to the speciﬁc modiﬁcations implemented. On the other hand,

inhibitors 10e and 10i showed a decrease in binding energy. The 10g

and 10k inhibitors are two special cases considering that the transfor-

mation resulted in the largest increase in the TPSA, from 99.81 A²for

10c to 112.7 A²and 120.1 A²to 10g and 10k, respectively.

To explain this trend, we can consider that the substitution of the

In general, ΔT_mis in good agreement with the pK_ivalues. However,

16

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 17. Thermodynamic cycle analysis from compound 10a to 3b. The thermodynamics diﬀerences ΔΔX_n→m= ΔXm - ΔXn (X = G, H, S for the transformations

[n → m]) are color-coded as red (negative), green (positive), grey (no signiﬁcant diﬀerence Δ < 0.2). The values were calculated from the data in Table 5. (For

interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

because most of the ligands have pK_ibetween 7.0 and 8.0, the ΔT_mdid

not change signiﬁcantly (median of 10.5, se = 0.484, sd = 1.74, var-

iance = 3.04, range 6.1–13.5).

4. Conclusion

As we have heretofore observed, the literature is full of small mo-

lecule inhibitors of a single CP while little is known about cross-class

inhibitors amongst mammalian and protozoa CPs [1]. This work aims to

unveil new high-aﬃnity non-peptide inhibitors of protozoan CPs (Cz

and LmCPB), which in turn can be mammalian CPs (CatK, CatL, CatS)

high-aﬃnity inhibitors by unique structural modiﬁcations. We have

performed an MMP/SAR analysis which revealed that Leu in P2 and (S)-

benzyl moiety in P1 are privileged structures for the design of new

cross-class inhibitors. Among several high-aﬃnity CP inhibitors, com-

pound 3c, which exhibits the single-digit nanomolar K_ivalue (2 nM that

yields a pK_i~ 8.7) for all ﬁve studied CPs, was subsequently in-

vestigated through a double SAR cycle. The bioisosteric replacement of

The most potent ligand in the series, 3c, whose Cz aﬃnity (pK_i) is

8.7 was able to thermally stabilize the Cz structure with 13.5 °C. On the

other hand, 10a (pK_i= 6.7) displayed one of the lowest ΔT_mof the

series. These ﬁndings suggest that the DSC analysis can be used to es-

timate aﬃnity in the presence of a broad SAR. It is worth noting that

compound 10d has the second-highest ΔT_m, with a pK_i1.5 lower than

3c, nonetheless both displayed the highest enthalpy contribution in the

binding process (see chapter 3.2).

Fig. 18. Thermodynamic cycle analysis for 10a transformations to yield 10d. The thermodynamics diﬀerences ΔΔX_n→m= ΔXm - ΔXn (X = G, H, S for the

transformations [n → m]) are color-coded as red (negative), green (positive), grey (no signiﬁcant diﬀerence Δ < 0.2). The values were calculated from the data in

Table 5. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

17

L. Cianni, et al.

BioorganicChemistry101(2020)104039

Fig. 19. Thermodynamic MMP analysis of P1 modiﬁcation for compound 10c. The thermodynamics diﬀerences ΔΔX_n→m= ΔXm - ΔXn (X = G, H, S for the

transformations [n → m]) are color-coded as red (negative), green (positive), grey (no signiﬁcant diﬀerence Δ < 0.2). The values were calculated from the data in

Table 5. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

Fig. 20. Thermodynamic MMP analysis of 4e vs

4c. The thermodynamics diﬀerences ΔΔX_n→

= ΔXm - ΔXn (X = G, H, S for the transfor-

m

mations [n → m]) are color-coded as red (nega-

tive), green (positive), grey (no signiﬁcant dif-

ference Δ

< 0.2). The values were calculated

from the data in Table 5. (For interpretation of

the references to color in this ﬁgure legend, the

reader is referred to the web version of this ar-

ticle.)

Table 6

Experimental results obtained for 13 compounds when incubated with Cz and

analyzed by diﬀerential scanning calorimetry (DSC). The ΔTm was obtained by

subtracting the T_mof the incubated protein by the T_mof the native protein. The

conditions for the assays were the same used for native cruzain assays with the

only diﬀerence being the presence of the ligands. The mean values of the two

measurements are presented. Tm = midpoint thermal unfolding (also melting

temperature), ΔH^cal= calorimetric enthalpy, ΔH^VH= Van’t Hoﬀ enthalpy.

Ligand T_m(°C)

ΔT_m(°C) ΔH^Cal(kcal

ΔH^VH(kcal

ΔH^cal

/

mol⁻¹

)

mol⁻¹

)

ΔH^VH

Apo-Cz 66.4

0.1

–

100

127

170

152

115

107

128

105

117

75

5

113

187

159

168

133

193

191

189

196

198

188

202

178

212

12

2

0.9

0.7

1

3c

79.7

76.5

76.7

72.5

76.4

76.9

76.7

79.3

76.2

76.6

76.8

77.1

75.8

13.5

10.5

10.7

6.1

15

8

4a

3

4b

2.8

2

1.99

2

0.9

0.8

0.5

0.7

0.5

0.6

0.4

0.7

0.4

10a

10b

10c

10d

10f

10g

10h

10j

10k

10.2

10.7

10.5

13.1

10.0

10.4

10.6

10.9

9.6

6

12

13

7

28

6

2

12

9

74

3

2

Fig. 21. Reversibility assay performed for Cz. The blue line corresponds to the

ﬁrst heating scan, ranging from 0 to 90 °C, the green line corresponds to the

cooling scan and the red line to the second heating process. The peak of the

unfolding process is depicted with T_m= 66.4 °C. The scans were done at 3 atm

and with a rate of 2 °C per minute. (For interpretation of the references to color

in this ﬁgure legend, the reader is referred to the web version of this article.)

88

12

11

15

125

77

13

5

the carbonyl group in P2/P3 for the CF₃group and the presence of Leu

in the P2 position had a superadditive eﬀect on aﬃnity for all the CPs.

Likewise, if from one side the para biphenyl in P3 is a privileged

structure to gain selectivity amid protozoan and mammalian CPs, meta

18

L. Cianni, et al.

BioorganicChemistry101(2020)104039

phenyl substituent led to high-aﬃnity cross-class inhibitors of diﬀerent

CPs such as compounds 3e and 4c. Besides that, molecular dynamics

simulations for Cz reveal two possible modes of binding (MoB I and

MoB II) for compounds 4a, 4c, 4d, 4e which could explain the strong

nonadditivity SAR recently published [8].

P.H.J. Silva, C.B. Batista, C.H. Moraes, L.H.G. Franco, P.W. Freitas-Junior, A. Kenny,

A.C.B. Leitão, D. de Burtoloso, C.A.M. Vita, Synthesis and structure-activity re-

lationship of nitrile-based cruzain inhibitors incorporating a triﬂuoroethylamine-

based P2 amide replacement, Bioorg. Med. Chem. 27 (2019) 115083, , https://doi.

org/10.1016/j.bmc.2019.115083.

[9] L.A.A. Avelar, C.D. Camilo, S. de Albuquerque, W.B. Fernandes, C. Gonçalez,

P.W. Kenny, A. Leitão, J.H. McKerrow, C.A. Montanari, E.V.M. Orozco,

J.F.R. Ribeiro, J.R. Rocha, F. Rosini, M.E. Saidel, Molecular Design, Synthesis and

Trypanocidal Activity of Dipeptidyl Nitriles as Cruzain Inhibitors, PLoS Negl Trop

Dis. 9 (2015) e0003916, , https://doi.org/10.1371/journal.pntd.0003916.

[10] A.C.B. Burtoloso, S. de Albuquerque, M. Furber, J.C. Gomes, C. Gonçalez,

P.W. Kenny, A. Leitão, C.A. Montanari, J.C. Quilles, J.F.R. Ribeiro, J.R. Rocha, Anti-

trypanosomal activity of non-peptidic nitrile-based cysteine protease inhibitors,

PLoS Negl Trop Dis. 11 (2017) e0005343, , https://doi.org/10.1371/journal.pntd.

0005343.

The thermodynamic signatures for several compounds, obtained

using ITC, display a strong enthalpy–entropy compensation which was

surmounted by two compounds 10g and 4e. These compounds are the

ﬁrst cruzain inhibitors reported up to now to display a thermodynamic

signature wherein both entropy and enthalpy favorably contribute to

Gibbs binding energy. Finally, we also reported the ﬁrst DSC analysis of

Cz incubated with 13 diﬀerent compounds where ΔT_mis in good

agreement with the pK_ivalues spanning cruzain thermal stabilization

from 6 to 13 °C.

[11] J.Y. Gauthier, N. Chauret, W. Cromlish, S. Desmarais, L.T. Duong, J.-P. Falgueyret,

D.B. Kimmel, S. Lamontagne, S. Léger, T. LeRiche, C.S. Li, F. Massé, D.J. McKay,

D.A. Nicoll-Griﬃth, R.M. Oballa, J.T. Palmer, M.D. Percival, D. Riendeau,

J. Robichaud, G.A. Rodan, S.B. Rodan, C. Seto, M. Thérien, V.-L. Truong,

M.C. Venuti, G. Wesolowski, R.N. Young, R. Zamboni, W.C. Black, The discovery of

odanacatib (MK-0822), a selective inhibitor of cathepsin K, Bioorg. Med. Chem.

Lett. 18 (2008) 923–928, https://doi.org/10.1016/j.bmcl.2007.12.047.

[12] J. Lameira, V. Bonatto, L. Cianni, F. dos Reis Rocho, A. Leitão, C.A. Montanari,

Predicting the aﬃnity of halogenated reversible covalent inhibitors through relative

binding free energy, Phys. Chem. Chem. Phys. 21 (2019) 24723–24730, https://doi.

org/10.1039/C9CP04820K.

In conclusion, this work is aimed at oﬀering new insights into the

optimization strategy for the design, synthesis, and biochemical/bio-

physical evaluation of potent cross-class CP inhibitors. This may ulti-

mately aid advancing molecular designing eﬀorts directed to reversible

covalent inhibitors with direct implications for polypharmacology and

multi-target screenings perhaps to include other cysteine en-

dopeptidases.

[13] L. Cianni, G. Sartori, F. Rosini, D.V. De, G. Pires, B.R. Lopes, A. Leitão,

A.C.B. Burtoloso, C.A. Montanari, Leveraging the cruzain S3 subsite to increase

aﬃnity for reversible covalent inhibitors, Bioorg Chem. 79 (2018) 285–292,

https://doi.org/10.1016/j.bioorg.2018.04.006.

Declaration of Competing Interest

[14] J. Cai, J. Robinson, S. Belshaw, K. Everett, X. Fradera, M. van Zeeland, L. van

Berkom, P. van Rijnsbergen, L. Popplestone, M. Baugh, M. Dempster, J. Bruin,

W. Hamilton, E. Kinghorn, P. Westwood, J. Kerr, Z. Rankovic, W. Arbuckle,

D.J. Bennett, P.S. Jones, C. Long, I. Martin, J.C.M. Uitdehaag, T. Meulemans,

Triﬂuoromethylphenyl as P2 for ketoamide-based cathepsin S inhibitors, Bioorg.

Med. Chem. Lett. 20 (2010) 6890–6894, https://doi.org/10.1016/j.bmcl.2010.10.

012.

The authors declare that they have no known competing ﬁnancial

interests or personal relationships that could have appeared to inﬂu-

ence the work reported in this paper.

[15] B.T. Mott, R.S. Ferreira, A. Simeonov, A. Jadhav, K.-K.-H. Ang, W. Leister, M. Shen,

J.T. Silveira, P.S. Doyle, M.R. Arkin, J.H. McKerrow, J. Inglese, C.P. Austin,

C.J. Thomas, B.K. Shoichet, D.J. Maloney, Identiﬁcation and Optimization of

Inhibitors of Trypanosomal Cysteine Proteases: Cruzain, Rhodesain, and TbCatB, J.

Med. Chem. 53 (2010) 52–60, https://doi.org/10.1021/jm901069a.

[16] A. Jakalian, D.B. Jack, C.I. Bayly, Fast eﬃcient generation of high-quality atomic

charges. AM1-BCC model: II. Parameterization and validation, J. Comput. Chem. 23

(2002) 1623–1641, https://doi.org/10.1002/jcc.10128.

Acknowledgments

We are indebted to Fundação de Amparo à Pesquisa do Estado de

São Paulo – FAPESP (grant #2013/18009-4, #2018/03985-1 and grant

#2016/07946-5) for ﬁnancing this project. Too, we would also like to

thank the CAPES Drug Discovery program (Process #139/2015) that

allowed us to evaluate LmCPB inhibitors. We want to acknowledge the

National Council for Scientiﬁc and Technological Development (CNPq,

grant # 304030/2018-0) in Brazil for scholarships.

[17] W.L. Jorgensen, J. Chandrasekhar, J.D. Madura, R.W. Impey, M.L. Klein,

Comparison of simple potential functions for simulating liquid water, J. Chem.

Phys. 79 (1983) 926–935, https://doi.org/10.1063/1.445869.

[18] D.A. Case, T.E. Cheatham, T. Darden, H. Gohlke, R. Luo, K.M. Merz, A. Onufriev,

C. Simmerling, B. Wang, R.J. Woods, The Amber biomolecular simulation pro-

grams, J. Comput. Chem. 26 (2005) 1668–1688, https://doi.org/10.1002/jcc.

20290.

Appendix A. Supplementary data

[19] J.A. Maier, C. Martinez, K. Kasavajhala, L. Wickstrom, K.E. Hauser, C. Simmerling,

ﬀ14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters

from ﬀ99SB, J. Chem. Theory Comput. 11 (2015) 3696–3713, https://doi.org/10.

1021/acs.jctc.5b00255.

Supplementary data to this article can be found online at https://

doi.org/10.1016/j.bioorg.2020.104039.

[20] H.C. Andersen, Rattle: A “velocity” version of the shake algorithm for molecular

dynamics calculations, J. Comput. Phys. 52 (1983) 24–34, https://doi.org/10.

1016/0021-9991(83)90014-1.

References

[21] D.R. Roe, I.I.I. Thomas, E. Cheatham, PTRAJ and CPPTRAJ: Software for Processing

and Analysis of Molecular, Dynamics Trajectory Data (2013), https://doi.org/10.

1021/ct400341p.

[1] L. Cianni, C.W. Feldmann, E. Gilberg, M. Gütschow, L. Juliano, A. Leitão,

J. Bajorath, C.A. Montanari, Can Cysteine Protease Cross-Class Inhibitors Achieve

Selectivity? J. Med. Chem. (2019), https://doi.org/10.1021/acs.jmedchem.

9b00683.

[22] Gnuplot: An Interactive Plotting Program, (n.d.). http://www.gnuplot.info/docs_4.

0/gnuplot.html (accessed April 27, 2020).

[2] L. Kramer, D. Turk, B. Turk, The Future of Cysteine Cathepsins in Disease

Management, Trends Pharmacol. Sci. 38 (2017) 873–898, https://doi.org/10.1016/

j.tips.2017.06.003.

[23] E.F. Pettersen, T.D. Goddard, C.C. Huang, G.S. Couch, D.M. Greenblatt, E.C. Meng,

T.E. Ferrin, UCSF Chimera—A visualization system for exploratory research and

analysis, J. Comput. Chem. 25 (2004) 1605–1612, https://doi.org/10.1002/jcc.

20084.

[3] M. Sajid, S.A. Robertson, L.S. Brinen, J.H. McKerrow, Cruzain : the path from target

validation to the clinic, Adv. Exp. Med. Biol. 712 (2011) 100–115, https://doi.org/

10.1007/978-1-4419-8414-2_7.

[24] L. Cianni, C. Lemke, E. Gilberg, C. Feldmann, F. Rosini, F. dos Reis Rocho,

J.F.R. Ribeiro, D.Y. Tezuka, C.D. Lopes, S. de Albuquerque, J. Bajorath, S. Laufer,

A. Leitão, M. Gütschow, C.A. Montanari, Mapping the S1 and S1’ subsites of cy-

steine proteases with new dipeptidyl nitrile inhibitors as trypanocidal agents,

BioRxiv. (2019) 760736, https://doi.org/10.1101/760736.

[4] P.-A. Casgrain, C. Martel, W.R. McMaster, J.C. Mottram, M. Olivier, A. Descoteaux,

Cysteine Peptidase B Regulates Leishmania mexicana Virulence through the

Modulation of GP63 Expression, PLoS Pathog. 12 (2016) e1005658, , https://doi.

org/10.1371/journal.ppat.1005658.

[25] Enzyme Database - BRENDA, (n.d.). https://www.brenda-enzymes.org/ (accessed

April 27, 2020).

[5] D.N. Deaton, S. Kumar, Cathepsin K Inhibitors: Their Potential as Anti-Osteoporosis

Agents, in: Progress in Medicinal Chemistry, Elsevier, 2004: pp. 245–375. https://

doi.org/10.1016/S0079-6468(04)42006-2.

[26] C. Tyrchan, E. Evertsson, Matched Molecular Pair Analysis in Short: Algorithms,

Applications and Limitations, Comput. Struct. Biotechnol. J. 15 (2017) 86–90,

https://doi.org/10.1016/j.csbj.2016.12.003.

[6] J.R. Coura, P.A. Viñas, Chagas disease: a new worldwide challenge, Nature 465

(2010) S6–S7, https://doi.org/10.1038/nature09221.

[27] N. Asaad, P.A. Bethel, M.D. Coulson, J.E. Dawson, S.J. Ford, S. Gerhardt, M. Grist,

G.A. Hamlin, M.J. James, E.V. Jones, G.I. Karoutchi, P.W. Kenny, A.D. Morley,

K. Oldham, N. Rankine, D. Ryan, S.L. Wells, L. Wood, M. Augustin, S. Krapp,

H. Simader, S. Steinbacher, Dipeptidyl nitrile inhibitors of Cathepsin L, Bioorg.

Med. Chem. Lett. 19 (2009) 4280–4283, https://doi.org/10.1016/j.bmcl.2009.05.

071.

[7] L. Cianni, C. Lemke, E. Gilberg, C. Feldmann, F. Rosini, F.D.R. Rocho, J.F.R. Ribeiro,

D.Y. Tezuka, C.D. Lopes, S. de Albuquerque, J. Bajorath, S. Laufer, A. Leitão,

M. Gütschow, C.A. Montanari, I.C. Almeida, Mapping the S1 and S1’ subsites of

cysteine proteases with new dipeptidyl nitrile inhibitors as trypanocidal agents,

PLoS Negl Trop Dis 14 (2020) e0007755, https://doi.org/10.1371/journal.pntd.

0007755.

[28] P.K. Jadhav, M.A. Schiﬄer, K. Gavardinas, E.J. Kim, D.P. Matthews, M.A. Staszak,

D.S. Coﬀey, B.W. Shaw, K.C. Cassidy, R.A. Brier, Y. Zhang, R.M. Christie,

[8] J.C. Gomes, L. Cianni, J. Ribeiro, F. dos Reis Rocho, S. da Costa Martins,

19

L. Cianni, et al.

W.F. Matter, K. Qing, J.D. Durbin, Y. Wang, G.G. Deng, Discovery of Cathepsin S

BioorganicChemistry101(2020)104039

ChEMBL database in 2017, Nucleic Acids Res. 45 (2017) D945–D954, https://doi.

Inhibitor LY3000328 for the Treatment of Abdominal Aortic Aneurysm, ACS Med.

Chem. Lett. 5 (2014) 1138–1142, https://doi.org/10.1021/ml500283g.

[29] L.A.A. Avelar, C.D. Camilo, S. de Albuquerque, W.B. Fernandes, C. Gonçalez,

P.W. Kenny, A. Leitão, J.H. McKerrow, C.A. Montanari, E.V.M. Orozco,

J.F.R. Ribeiro, J.R. Rocha, F. Rosini, M.E. Saidel, Molecular Design Synthesis and

Trypanocidal Activity of Dipeptidyl Nitriles as Cruzain Inhibitors, PLOS Negl. Trop.

Dis. 9 (2015) e0003916, , https://doi.org/10.1371/journal.pntd.0003916.

[30] D. Stumpfe, J. Bajorath, Exploring Activity Cliﬀs in Medicinal Chemistry:

Miniperspective, J. Med. Chem. 55 (2012) 2932–2942, https://doi.org/10.1021/

jm201706b.

org/10.1093/nar/gkw1074.

[37] E. Freire, Do enthalpy and entropy distinguish ﬁrst in class from best in class? Drug

Discov. Today. 13 (2008) 869–874, https://doi.org/10.1016/j.drudis.2008.07.005.

[38] Y. Li, Z. Zhang, Q. Feng, D. Yi, F. Lu, An eﬃcacy evaluation method for non-normal

outcomes in randomized controlled trials, Sci. Rep. 9 (2019) 11393, https://doi.

org/10.1038/s41598-019-47727-y.

[39] K.J. Å ström, B. Wittenmark, Computer-controlled systems: theory and design, 3rd

ed, Prentice Hall, Upper Saddle River, N.J., 1997 .

[40] E. Ronchetti, C. Field, W. Blanchard, Robust Linear Model Selection by Cross-

Validation, J. Am. Stat. Assoc. 92 (1997) 1017–1023, https://doi.org/10.1080/

01621459.1997.10474057.

[31] C. Kramer, J.E. Fuchs, K.R. Liedl, Strong Nonadditivity as a Key Structure-Activity

Relationship Feature: Distinguishing Structural Changes from Assay Artifacts, J.

Chem. Inf. Model. 55 (2015) 483–494, https://doi.org/10.1021/acs.jcim.5b00018.

[32] I.M. Prokopczyk, J.F. Ribeiro, G.R. Sartori, R. Sesti-Costa, J.S. Silva, R.F. Freitas,

A. Leitão, C.A. Montanari, Integration of methods in cheminformatics and bioca-

lorimetry for the design of trypanosomatid enzyme inhibitors, Future Med. Chem. 6

(2014) 17–33, https://doi.org/10.4155/fmc.13.185.

[41] O.A. Grigg, V.T. Farewell, D.J. Spiegelhalter, Use of risk-adjusted CUSUM and

RSPRTcharts for monitoring in medical contexts, Stat. Methods Med. Res. 12 (2003)

147–170, https://doi.org/10.1177/096228020301200205.

[42] A.M. Dos Santos, L. Cianni, D. De Vita, F. Rosini, A. Leitão, C.A. Laughton,

J. Lameira, C.A. Montanari, Experimental study and computational modelling of

cruzain cysteine protease inhibition by dipeptidyl nitriles, Phys. Chem. Chem. Phys.

20 (2018) 24317–24328, https://doi.org/10.1039/C8CP03320J.

[43] S. Solís-Mendiola, A. Rojo-Domínguez, A. Hernández-Arana, Cooperativity in the

unfolding transitions of cysteine proteinases. Calorimetric study of the heat dena-

turation of chymopapain and papain, Biochimica et Biophysica, Acta (BBA) -

Protein Struct. Mol. Enzymol. 1203 (1993) 121–125, https://doi.org/10.1016/

0167-4838(93)90045-S.

[33] G. Calabrò, C.J. Woods, F. Powlesland, A.S.J.S. Mey, A.J. Mulholland, J. Michel,

Elucidation of Nonadditive Eﬀects in Protein-Ligand Binding Energies: Thrombin as

a Case Study, J. Phys. Chem. B. 120 (2016) 5340–5350, https://doi.org/10.1021/

acs.jpcb.6b03296.

[34] O.A. Khakhel’, T.P. Romashko, Extrathermodynamics: Varieties of Compensation

Eﬀect, J. Phys. Chem. A. 120 (2016) 2035–2040, https://doi.org/10.1021/acs.jpca.

6b00493.

[44] N.V. Prabhu, K.A. Sharp, HEAT CAPACITY IN PROTEINS, Annu. Rev. Phys. Chem.

56 (2005) 521–548, https://doi.org/10.1146/annurev.physchem.56.092503.

141202.

[35] G. Williams, G.G. Ferenczy, J. Ulander, G.M. Keserű, Binding thermodynamics

discriminates fragments from druglike compounds: a thermodynamic description of

fragment-based drug discovery, Drug Discov. Today. 22 (2017) 681–689, https://

doi.org/10.1016/j.drudis.2016.11.019.

[45] G. Bruylants, J. Wouters, C. Michaux, Diﬀerential Scanning Calorimetry in Life

Science: Thermodynamics, Stability, Molecular Recognition and Application in

Drug Design, CMC 12 (2005) 2011–2020, https://doi.org/10.2174/

0929867054546564.

[36] A. Gaulton, A. Hersey, M. Nowotka, A.P. Bento, J. Chambers, D. Mendez,

P. Mutowo, F. Atkinson, L.J. Bellis, E. Cibrián-Uhalte, M. Davies, N. Dedman,

A. Karlsson, M.P. Magariños, J.P. Overington, G. Papadatos, I. Smit, A.R. Leach, The

20

Article Doi

Optimization strategy of single-digit nanomolar cross-class inhibitors of mammalian and protozoa cysteine proteases

DOI: 10.1016/j.bioorg.2020.104039

Source and publish data:

Authors:

Article abstract of DOI:10.1016/j.bioorg.2020.104039

Full text of DOI:10.1016/j.bioorg.2020.104039

Products guided by the article

R&D Labs maybe for 105356-41-0

Relevant to this article

Hot Product