The COVID-19 pandemic is having a major impact on public health worldwide, and there is an urgent need for the creation of an armamentarium of effective therapeutics, including vaccines, biologics, and small-molecule therapeutics, to combat SARS-CoV-2 and emerging variants. Inspection of the virus life cycle reveals multiple viral- and host-based choke points that can be exploited to combat the virus. SARS-CoV-2 3C-like protease (3CLpro), an enzyme essential for viral replication, is an attractive target for therapeutic intervention, and the design of inhibitors of the protease may lead to the emergence of effective SARS-CoV-2-specific antivirals. We describe herein the results of our studies related to the application of X-ray crystallography, the Thorpe-Ingold effect, deuteration, and stereochemistry in the design of highly potent and nontoxic inhibitors of SARS-CoV-2 3CLpro.
The COVID-19 pandemic is having a major impact on public health worldwide, and there is an urgent need for the creation of an armamentarium of effective therapeutics, including vaccines, biologics, and small-molecule therapeutics, to combat SARS-CoV-2 and emerging variants. Inspection of the virus life cycle reveals multiple viral- and host-based choke points that can be exploited to combat the virus. SARS-CoV-2 3C-like protease (3CLpro), an enzyme essential for viral replication, is an attractive target for therapeutic intervention, and the design of inhibitors of the protease may lead to the emergence of effective SARS-CoV-2-specific antivirals. We describe herein the results of our studies related to the application of X-ray crystallography, the Thorpe-Ingold effect, deuteration, and stereochemistry in the design of highly potent and nontoxic inhibitors of SARS-CoV-2 3CLpro.
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is the etiological agent of
coronavirus disease (COVID-19).[1] The severity of the ongoing pandemic is
having a major impact on public health worldwide and is further exacerbated by the emergence
of more virulent strains.[2,3] Intense worldwide efforts to combat the virus have led to the successful
development of FDA-approved vaccines, and an array of potential therapeutics, such as
monoclonal antibodies, repurposed drugs, and others, are currently being evaluated in
clinical trials or are at various stages of clinical development.[4,5]The SARS-CoV-2 life cycle encompasses multiple viral- and host-based druggable targets that
can be exploited, including, for example, inhibitors that block virus entry and fusion, and
replication inhibitors targeting the 3C-like protease (3CLpro), papain-like protease and the
RNA-dependent RNA polymerase, among others. Attractive host-based targets include the
proteases transmembrane serine protease 2 (TMPRSS2), cathepsin L, and furin. Thus, the
development of small-molecule therapeutics that target host or viral targets essential for
viral replication is a potentially fruitful avenue of investigation.[6−9]The SARS-CoV-2 genome contains two open reading frames (ORF1a and ORF1b). Translation of
the genomic mRNA of ORF1a yields a polyprotein (pp1a), while a second polyprotein (pp1ab) is
produced by a ribosomal frameshift that joins ORF1a together with ORF1b. The two
polyproteins are processed by two cysteine proteases, a 3C-like protease (3CLpro) at 11
distinct cleavage sites and a papain-like protease (PLpro) at 3 distinct cleavage sites,
resulting in 16 mature nonstructural proteins. The two proteases are essential for viral
replication, making SARS-CoV-2 3CLpro an attractive target for therapeutic
intervention.[9−16]SARS-CoV-2 3CLpro is a homodimer with a catalytic Cys–His dyad
(Cys145–His41) and an extended binding cleft. The protease
displays a strong preference for a -Y–Z–Leu–Gln–X sequence,
corresponding to the residues
-P4–P3–P2–P1–P1′-,[17] where X is a small amino acid (Ser, Ala, Gly), Y is a small hydrophobic
amino acid, and Z is solvent-exposed and can tolerate polar or nonpolar amino acid
chains.[18]Our foray in this area has focused on the structure-guided design of inhibitors of
SARS-CoV-2 and MERS-CoV 3CLpro,[19−21] as well as
feline infectious peritonitis virus (FIPV) protease inhibitors.[22,23] We recently described the
structure-guided design of a dipeptidyl series of MERS-CoV and SARS-CoV-2 3CLpro inhibitors
incorporating in their structure a piperidine[20] or cyclohexyl[19] moiety capable of engaging in favorable binding interactions with the
S4 pocket. We furthermore demonstrated that members of the cyclohexyl series of
compounds improve survival in a mouse model of MERS-CoV infection.[19] In
this report, we established a cell-based assay to screen inhibitors against SARS-CoV-2
3CLpro, which is safe (BSL2) and fast (takes less than 24 h). Furthermore, we report the
results of structure-guided studies intended to interrogate the effects of stereochemistry,
conformation, and structure, including the systematic introduction of fluorine
(F-walk)[24,25] around
the structure of GC376[21−23] and the synthesis of
deuterated inhibitors,[26−28] to modulate
pharmacological activity, pharmacokinetic (PK) properties, and oral bioavailability.
Results and Discussion
Chemistry
The synthesis of compounds 1–24b/c entailed the use of a structurally
diverse set of precursor alcohols (Table ), some
of which were commercially available. Alcohols 12–16 were readily
synthesized from 4,4-difluorocyclohexane carboxylic acid via reduction to the
corresponding alcohol by treatment with carbonyl diimidazole and sodium
borohydride,[29] followed by oxidation with Dess–Martin
periodinane reagent to yield the aldehyde. Subsequent treatment with an array of Grignard
reagents generated alcohols 12 and 14–16
(Scheme /panel A). Alcohol 13 was
synthesized by reacting the methyl ester of 4,4-difluorocyclohexane carboxylic acid with
excess methyl magnesium iodide, followed by acidic workup (Scheme /panel A). Deuterated alcohols 9, 11,
20, and 22 were obtained by treatment of the precursor
carboxylic acid with carbonyl diimidazole followed by the addition of sodium
borodeuteride. All trans-substituted alcohols were synthesized by reducing the precursor
4-substituted cyclohexanone with sodium borohydride/CeCl3.[30]
Table 1
Alcohol Inputs
Scheme 1
Compounds 1–24b/c were readily obtained by reacting each
precursor alcohol with disuccinimidyl carbonate,[31] followed by coupling
with amino alcohol A. The resulting product was treated with
Dess–Martin periodinane to yield aldehydes 1–24b,
which were converted to the corresponding bisulfite adducts
1–24c upon treatment with sodium bisulfite (Scheme /panel B).[32] An
alternative synthesis was used in the case of compounds
6–8, 10–16,
23, and 24, which involved the reaction of the precursor
alcohol with (L) leucine methyl ester isocyanate, as described in detail
previously.[33] The synthesis of precursor amino alcohol A
was readily accomplished by coupling (L) Z-Leu with a glutamine surrogate, followed by
sequential reduction with LiBH4 and removal of the protective group
(H2/Pd) (Scheme /panel C).
Biochemical Studies
Enzyme Assays
The inhibitory activity of compounds 1–24b/c against
SARS-CoV-2 3CLpro in biochemical assays was determined as described in the Experimental Section. The IC50 values against SARS-CoV-2
and MERS-CoV-2 in the enzyme assays are summarized in Table , and they are the average of at least two determinations. Most
of the compounds potently inhibited SARS-CoV-2 3CLpro and displayed IC50
values that ranged between 0.13 and 1.25 μM. Compounds 15b and
15c were the most effective against SARS-CoV-2 3CLpro, with
IC50 values 0.13 and 0.17 μM, respectively. The inhibitory activity
of a select number of compounds against MERS-CoV 3CLpro was also investigated. The
compounds were found to be 3–5-fold more potent against MERS-CoV-3CLpro, with
IC50 values in the 40–150 nM range (Table ). Interestingly, compounds 15b and 15c
were the most effective against MERS-CoV-3CLpro as well, with IC50 values
0.04 and 0.05 μM, respectively. The broad spectrum of inhibitory activity
displayed by these compounds enhances their therapeutic potential.
Table 2
IC50 Values of Compounds 1–24b/c
against SARS-CoV-2 3CLpro, IC50 Values of Selected Compounds against
MERS-CoV 3CLpro, EC50 Values against SARS-CoV-2 3CLpro, and
CC50 Values of Selected Compounds
IC50’s taken from Rathnayake et al.[19]
EC50: 0.85 ± 0.1 and 0.70 ± 0.08 μM for
4c and 15c, respectively, from live SARS-CoV-2 in Vero
E6 cells.
IC50’s taken from Rathnayake et al.[19]EC50: 0.85 ± 0.1 and 0.70 ± 0.08 μM for
4c and 15c, respectively, from live SARS-CoV-2 in Vero
E6 cells.
Establishment of the Cell-Based Assay for SARS-CoV-2 3CLpro Inhibitors
We have previously reported EC50 values determined by incubating SARS-CoV-2
3CLpro inhibitors and Vero E6 cells that were inoculated with SARS-CoV-2 at
50–100 plaque forming units/well.[19,33] This cell-based assay requires a BSL3 facility and
takes at least 2–3 days. As an alternative method, we report herein a relatively
fast and safe cell-based assay system to screen SARS-CoV-2 3CLpro inhibitors using two
plasmids. A similar cell-based assay has been reported;[34] however, in
contrast, the present system utilizes the replication units of porcine respiratory and
reproductive syndrome virus (PRRSV)[35] to express SARS-CoV-2 3CLpro.
In this system, plasmid 1, pR-SARS-CoV-2 3CLpro, was used to express SARS-CoV-2 3CLpro,
whereas plasmid 2, pGlo-VRLQS, was used to express luciferase-VRLQS in HEK293T cells
(Figure /panel A). The expressed inactive
luciferase is activated by the catalytic mechanism of SARS-CoV-2 3CLpro in HEK293T
cells. Hence, the inhibition of SARS-CoV-2 3CLpro was measured as a function of firefly
luciferase activity (Figure /panel B). Plasmid
2 also contains intact Renilla luciferase gene as an expression control. As a negative
control, the inactive form of the 3CLpro was introduced to pR-SARS-CoV-2 3CLpro by the
mutagenesis, and the resulting plasmid was designated as pR-SARS-CoV-2 3CLpro C145A. The
mock-transfection was also used as a negative control. The expression of coronavirus
3CLpro was reported to induce cytotoxicity in the transfected
cells.[36,37] The
level of cytotoxicity in our assay system was about 7%, and the ratio of firefly to
Renilla luminescence was adjusted to reduce variability due to cytotoxicity and
transfection efficiency. The EC50 values of a select number of compounds,
including 1b/1c, 4b/4c, 8b/8c,
14b/14c, 15b/15c, 18b/18c,
21b/21c, and 23b/23c, were determined (Table
). Inhibition curves by each compound were consistent with
a dose-dependent mode and R2 > 0.8 (Figure /panel C). The IC50 and EC50
values of 15b and 15c were the lowest among tested compounds
listed in Table .
Figure 1
Generation of a cell-based assay for screening SARS-CoV-2 3CLpro inhibitors in
HEK293T cells. Panel (A) Plasmid 1; pR-SA2–3CLpro encodes SARS-CoV-2 3CLpro
from the PRRSV reverse genetics system. The gene of SARS-CoV-2 3CLpro is inserted
between ORF1b and 2a of PRRSV genome. Plasmid 2; pGlo-VRLQS encodes firefly
luciferase with coronavirus 3CLpro recognition sequences VRLQS. Active luciferase is
generated by the cleavage with CoV 3CLpro. Panel (B) Semiconfluent HEK293T cells
were transfected with two plasmids, and after overnight, various concentrations of
each compound are applied to the cells. The inhibition of SARS-CoV-2 3CLpro is
determined by measuring luciferase activity. Panel (C) Inhibition curves of selected
compounds, 1c, 4c, 8c, 14c,
15c, 18c, 21c, and 23c, using
the cell-based assay with pR-SA2–3CLpro and pGlo-VRLQS.
Generation of a cell-based assay for screening SARS-CoV-2 3CLpro inhibitors in
HEK293T cells. Panel (A) Plasmid 1; pR-SA2–3CLpro encodes SARS-CoV-2 3CLpro
from the PRRSV reverse genetics system. The gene of SARS-CoV-2 3CLpro is inserted
between ORF1b and 2a of PRRSV genome. Plasmid 2; pGlo-VRLQS encodes firefly
luciferase with coronavirus 3CLpro recognition sequences VRLQS. Active luciferase is
generated by the cleavage with CoV 3CLpro. Panel (B) Semiconfluent HEK293T cells
were transfected with two plasmids, and after overnight, various concentrations of
each compound are applied to the cells. The inhibition of SARS-CoV-2 3CLpro is
determined by measuring luciferase activity. Panel (C) Inhibition curves of selected
compounds, 1c, 4c, 8c, 14c,
15c, 18c, 21c, and 23c, using
the cell-based assay with pR-SA2–3CLpro and pGlo-VRLQS.For comparative purposes, we determined EC50 values using live SARS-CoV-2 in
Vero E6 cells and the established two-plasmid system of two compounds (4c,
15c) from this work and a previously published 3CLpro inhibitor, GC376.
Compounds 4c and 15c were selected on the basis of having the
lowest IC50 values in this series. The EC50 of GC376 was found to
be 0.23 ± 0.01 μM in live SARS-CoV-2 in Vero E6 cells.[33]
In the two-plasmid system, the EC50 of GC376 was determined to be 3.15 ±
0.67 μM (14-fold higher), and the R2 values of the
inhibition curves were >0.9. When the antiviral effects of compounds 4c
and 15c were examined from live SARS-CoV-2 in Vero E6 cells, the
EC50 values were 0.85 ± 0.1 and 0.70 ± 0.08 μM,
respectively (Table ). The results show that
while compounds in this series are cell-permeable, the EC50 values were
higher from the two-plasmid system (2-fold for 15c and 5-fold for
4c) than those by live SARS-CoV-2 in Vero E6 cells. The higher
EC50s may be due to various reasons including different cell types,
presence of transfection reagent, incubation time, and overexpression of the 3CLpro and
luciferase in HEK293T cells due to inhibition of cytotoxic 3CLpro in the two-plasmid
system.[36,37] Most
of the examined compounds showed minimal toxicity up to 100 μM; however, the
CC50 values for 4b/4c and 8b/8c were in the
40–60 μM range (Table ). Although
the EC50 values obtained from the two-plasmid system are higher than those
obtained from live SARS-CoV-2 in Vero E6 cells, considering the feasibility of
conducting the experiments under BSL2 laboratory conditions and the relatively short
amount of time required (24 h), the cell-based two-plasmid method could be a useful
initial screening tool for 3CLpro inhibitors against SARS-CoV-2.
X-ray Crystallographic Studies
A series of high-resolution cocrystal structures were determined to elucidate the
interaction of the inhibitors with the active site of SARS-CoV-2 3CLpro. Specifically, we
sought to confirm the mechanism of action, identify the structural determinants associated
with the binding of the inhibitors to the active site of the protease, and ultimately
harness the accumulated structural information and insights gained to further optimize
pharmacological activity and PK parameters. Three groups of inhibitor types were analyzed
with respect to their structural elements that interact within the S4 subsite
environment, which are (1) nonpolar substituents (Table , entries 1–9), (2) 4,4-difluorocyclohexyl groups that are connected to
a stereocenter (Table , entries 10–16),
and (3) fluorinated aryl compounds based on the structure of GC376 (Table
, entries 17–22).For all structures described below, the active sites contained prominent difference
electron density consistent with inhibitors covalently bound to Cys 145. Additionally, the
electron density was consistent with both the R and S enantiomers at the stereocenter
formed by covalent attachment of the Sγ atom of Cys 145 and were therefore modeled
as each enantiomer with 0.5 occupancy. The γ-lactam ring of the inhibitor forms
direct hydrogen bonds with Glu 166 and His163, and Glu 166 and Gln 189 form additional
H-bonds with the C=O and NH of the carbamate moiety in the inhibitor. The inhibitor
engages in hydrophobic interactions with the leucine side chain, which is snugly
accommodated in the S2 pocket. The cocrystal structure confirms that the
reaction of Cys145 with the aldehyde warhead results in the formation of a tetrahedral
hemithioacetal that is stabilized by a H-bond to His164.
Nonpolar Substituents
The structures of 5c, 1c, 3c, and
8b displayed well-defined electron densities and similar hydrogen bond
interactions, as shown in Figure . For all
structures, the nonpolar groups are mainly positioned within the S4 subsite
near a hydrophobic ridge formed by residues Leu 167, Pro 168, Gly 170, and Ala 191
(Figure ). However, the dimethylcyclohexyl
ring in 1c is too short to fully engage the hydrophobic ridge in the
S4 subsite (Figure B). The
addition of an n-propyl group in 3c permits further
engagement with the hydrophobic cleft, and the extra carbon atom in 8b
allows the propyl group to extend even further (Figure C,D). Superposition of 3c and 1c (Figure E) shows that the 4,4-dimethylcyclohexyl ring is
moved slightly out of the S4 subsite relative to the
n-propyl group in 3c. Additionally, the superposition of
3c and 8b revealed quite similar binding modes although the
n-propyl group of 8b is positioned deeper within the
S4 subsite (Figure F). Overall,
the similar binding modes and attendant high potency of the inhibitors are reflected in
their low IC50 values and similar potencies (Table , compounds 1–5b/c). With
respect to compound 8, it was envisaged that the corresponding deuterated
compound 9, found to be nearly equipotent to nondeuterated compound
8 (Table ), would likely
display improved PK properties due to its enhanced in vivo stability
arising from the greater strength of the C–D bond and the resulting deuterium
kinetic isotope effect.[26−28]
Figure 2
Binding mode of inhibitors containing nonpolar substituents. 5c (A/E),
1c (B/F), 3c (C/G), and 8b (D/H) with
SARS-CoV-2 3CLpro. Fo–Fc Polder omit map (A–D) contoured at 3σ.
Hydrogen bond interactions (E–H) are drawn as dashed lines. PDB IDs:
5c (7LZZ), 1c (7LZX), 3c (7LZY), and
8b (7LZT).
Figure 3
Surface representation showing the orientation of nonpolar groups near the
S4 subsite of SARS-CoV-2 3CLpro, with neighboring residues colored
yellow (nonpolar), cyan (polar), and white (weakly polar). 5c (A),
1c (B), 3c (C), and 8b (D). Superposition
of 3c (gray) and 1c (coral) (E). Superposition of
3c (gray) and 8b (magenta) (F). PDB IDs: 5c
(7LZZ), 1c (7LZX), 3c (7LZY), and 8b
(7LZT).
Binding mode of inhibitors containing nonpolar substituents. 5c (A/E),
1c (B/F), 3c (C/G), and 8b (D/H) with
SARS-CoV-2 3CLpro. Fo–Fc Polder omit map (A–D) contoured at 3σ.
Hydrogen bond interactions (E–H) are drawn as dashed lines. PDB IDs:
5c (7LZZ), 1c (7LZX), 3c (7LZY), and
8b (7LZT).Surface representation showing the orientation of nonpolar groups near the
S4 subsite of SARS-CoV-2 3CLpro, with neighboring residues colored
yellow (nonpolar), cyan (polar), and white (weakly polar). 5c (A),
1c (B), 3c (C), and 8b (D). Superposition
of 3c (gray) and 1c (coral) (E). Superposition of
3c (gray) and 8b (magenta) (F). PDB IDs: 5c
(7LZZ), 1c (7LZX), 3c (7LZY), and 8b
(7LZT).
4,4-Difluorocyclohexyl Compounds
In previous studies related to norovirus 3CLpro inhibitors, the strategic introduction
of a gem-dimethyl group into the inhibitor structure resulted in
enhanced potency by restricting the rotation around the nearby single bonds and lowering
the entropic penalty associated with binding.[38] Thus, we sought to
capitalize on this by synthesizing gem-dimethyl-substituted compound
13c and, additionally, achieve the same end by introducing a stereocenter
(12c). The structures of 12b, 13c, and
14c with SARS-CoV 3CLpro displayed well-defined electron densities and
the typically observed hydrogen bond interactions (Figure ). The 4,4-difluorocyclohexyl rings for all structures are
positioned near the hydrophobic cleft in the S4 subsite, as shown in Figure A–C. Superposition of these
structures revealed a nearly identical binding mode for 12b and
13c in which the 4,4-difluorocyclohexyl groups are positioned in the same
region within the S4 subsite (Figure D). For 14c, the benzyl ring is oriented in a wide cleft formed
by Asn 142 and Gln 189. However, the 4,4-difluorocyclohexyl ring of 13c
contacts residues Thr 190 and Ala 191 (3.0–3.2 Å) and forms new hydrogen
bond interactions with the backbone oxygen and nitrogen atoms, respectively (Figure E). This positions the
4,4-difluorocyclohexyl ring of 13c deeper into the S4 pocket and
results in a conformational change in the loop spanning Gln 189 to Gly 195 and Glu 166
to Gly 170 to accommodate the new interactions and avoid steric clash. This results in
the loss of the typical hydrogen bond between the side chain of Glu 166 and the
glutamine surrogate of 13c (Figures E and 5D), which may explain why the IC50 of
13c is ∼4-fold higher than those of 12b and
14c.
Figure 4
Binding mode of inhibitors containing a 4,4-difluorocyclohexyl group.
12b (A/D), 13c (B/E), and 14c (C/F) with
SARS-CoV-2 3CLpro. Fo–Fc Polder omit map (A–C) contoured at 3σ.
Hydrogen bond interactions (D–F) are drawn as dashed lines. PDB IDs:
12b (7LZU), 13c (7M00), and 14c (7M01).
Figure 5
Surface representation showing the orientation of the 4,4-difluorocyclohexyl groups
near the S4 subsite of SARS-CoV-2 3CLpro, with neighboring residues colored yellow
(nonpolar), cyan (polar), and white (weakly polar). 12b (A),
13c (B), and 14c (C). Superposition of 12b
(gold), 13c (coral), and 14c (gray) (D). The loop between
Gln 189 and Gly 195 is colored cyan for 13c and magenta for
12b/14c. Hydrogen bond interactions with Glu 166 are
indicated by the dashed lines. PDB IDs: 12b (7LZU), 13c
(7M00), and 14c (7M01).
Binding mode of inhibitors containing a 4,4-difluorocyclohexyl group.
12b (A/D), 13c (B/E), and 14c (C/F) with
SARS-CoV-2 3CLpro. Fo–Fc Polder omit map (A–C) contoured at 3σ.
Hydrogen bond interactions (D–F) are drawn as dashed lines. PDB IDs:
12b (7LZU), 13c (7M00), and 14c (7M01).Surface representation showing the orientation of the 4,4-difluorocyclohexyl groups
near the S4 subsite of SARS-CoV-2 3CLpro, with neighboring residues colored yellow
(nonpolar), cyan (polar), and white (weakly polar). 12b (A),
13c (B), and 14c (C). Superposition of 12b
(gold), 13c (coral), and 14c (gray) (D). The loop between
Gln 189 and Gly 195 is colored cyan for 13c and magenta for
12b/14c. Hydrogen bond interactions with Glu 166 are
indicated by the dashed lines. PDB IDs: 12b (7LZU), 13c
(7M00), and 14c (7M01).
Fluorinated Aryl Compounds
Positional analogue scanning is a widely used strategy for optimizing binding affinity,
selectivity, and physicochemical properties of lead compounds containing aromatic or
heteroaromatic rings.[24] For instance, the introduction of fluorine
(F-walk)[25] or nitrogen (N-walk)[39] is an
effective means for multiparameter optimization by leveraging the beneficial impact of
fluorine (or nitrogen) and minor structural changes. In an effort to determine the
effect of fluorine on the binding mode in the S4 subsite of GC376, the
structures of the fluorinated benzyl compounds 17c, 18c,
19b, 20b (deuterated analogue of 19b), and
21c were determined with SARS-CoV-2 3CLpro. The inhibitor o-fluorobenzyl
(17c) and m-fluorobenzyl (18c) compounds
displayed well-defined electron densities and similar hydrogen bond interactions, as
shown in Figure . Interestingly, the
o-fluorobenzyl ring of 17c adopts a conformation in which the fluorine atom
is directed away from Thr 190 and is instead positioned 3.38 Å from the backbone
oxygen atom of Glu 166 (Figure C). Conversely,
the fluorine atom in 18c is positioned between Thr 190/Ala 191 in the
S4 pocket and is 3.10 Å from the backbone nitrogen atom of Ala 191
(Figure D). The orientations of the fluorine
atoms in 17c and 18c relative to the hydrophobic ridge in the
S4 pocket are shown in Figure E,F.
Figure 6
Binding mode of inhibitors containing a fluorinated aromatic group.
17c (A/C) and 18c (B/D) with SARS-CoV-2 3CLpro.
Fo–Fc Polder omit map (A, B) contoured at 3σ. Hydrogen bond
interactions (C, D) are drawn as dashed lines. The 3.38 Å contact between the
F-atom of 17c and the backbone O-atom of Glu 166 is drawn as a solid
line in panel (C). Surface representation of 17c (E) and
18c (F) showing the orientation of the 4,4-difluorocyclohexyl groups
near the S4 subsite of SARS-CoV-2 3CLpro, with neighboring residues
colored yellow (nonpolar), cyan (polar), and white (weakly polar). PDB IDs:
17c (7M02) and 18c (7M03).
Binding mode of inhibitors containing a fluorinated aromatic group.
17c (A/C) and 18c (B/D) with SARS-CoV-2 3CLpro.
Fo–Fc Polder omit map (A, B) contoured at 3σ. Hydrogen bond
interactions (C, D) are drawn as dashed lines. The 3.38 Å contact between the
F-atom of 17c and the backbone O-atom of Glu 166 is drawn as a solid
line in panel (C). Surface representation of 17c (E) and
18c (F) showing the orientation of the 4,4-difluorocyclohexyl groups
near the S4 subsite of SARS-CoV-2 3CLpro, with neighboring residues
colored yellow (nonpolar), cyan (polar), and white (weakly polar). PDB IDs:
17c (7M02) and 18c (7M03).The compounds that contain a p-fluorobenzyl group 19b and its deuterated
analogue 20b not surprisingly adopt very similar binding modes and hydrogen
bond interactions, as shown in Figures S1 and S2. Interestingly, the inhibitor adopts two conformations
in which the p-fluorobenzyl ring is projected away from the S4 subsite in
subunit B and is positioned in the S4 pocket in subunit A. However, the
electron density for the p-fluorobenzyl ring is somewhat weaker in subunit A, which
suggests that the pose in subunit B is likely the predominant conformation. This may be
due to the fact that the fluorine atom does not form any contacts with polar atoms in
the S4 subsite and results in a conformation in which the aryl ring is
positioned out of the pocket, which is the same conformation observed for the parent
compound GC376.The perfluorinated compound 21c also displayed a well-defined difference
electron density consistent with the aryl ring in one conformation (Figure A). Interestingly, one of the o-fluorine atoms
interacts with the backbone oxygen of Glu 166 (3.08 Å), which is shorter than that
observed for 17c described above (3.38 Å). The other o-fluorine atom
is positioned 2.92 Å from the backbone N-atom of Thr 190 and 3.12 Å from the
side chain N-atom of Gln 189 (Figure B).
Similarly, the m-fluorine atom is positioned near the backbone nitrogen
atom of Ala 191 (3.40 Å), which is longer than the distance observed for
18c (3.10 Å). The pentafluorobenzyl ring is positioned on top of the
hydrophobic cleft within the S4-subsite (Figure C), unlike GC376 where the phenyl ring undergoes a hydrophobic
collapse with the γ-lactam ring and the inhibitor assumes a “paper
clip” shape.
Figure 7
Binding mode of 21c containing a perfluorinated aromatic group.
Fo–Fc Polder omit map contoured at 3σ (A). Hydrogen bond interactions
are drawn as dashed lines. Close contacts to the perfluorinated ring that are longer
than typical polar contact distances are drawn as solid lines (B). Surface
representation showing the orientation of 21c near the S4 subsite of
SARS-CoV-2 3CLpro, with neighboring residues colored yellow (nonpolar), cyan
(polar), and white (weakly polar) (C). PDB ID: 21c (7M04).
Binding mode of 21c containing a perfluorinated aromatic group.
Fo–Fc Polder omit map contoured at 3σ (A). Hydrogen bond interactions
are drawn as dashed lines. Close contacts to the perfluorinated ring that are longer
than typical polar contact distances are drawn as solid lines (B). Surface
representation showing the orientation of 21c near the S4 subsite of
SARS-CoV-2 3CLpro, with neighboring residues colored yellow (nonpolar), cyan
(polar), and white (weakly polar) (C). PDB ID: 21c (7M04).Finally, GC376 variants 23b/c and 24b/c were synthesized and
screened as mixtures of epimers. The aldehyde and bisulfite adduct compounds
23b/c were found to potently inhibit 3CLpro (IC50 0.15 and
0.18 μM, respectively), and these were 27- and 19-fold more potent than the
corresponding 24b/c aldehyde and bisulfite adducts, respectively. These
findings provide tentative validation of the design regarding the use of a chiral center
to attain directional control and augment binding interactions.
Conclusions
Effective management of SARS-CoV-2, the causative agent of the COVID-19 pandemic, would
require the availability of safe and effective vaccines (already realized), as well as the
availability of small-molecule therapeutics and prophylactics that target viral- and
host-based druggable targets. SARS-CoV-2 3CLpro is an attractive target for the development
of COVID-19 therapeutics because of its vital role in viral replication. An array of
approaches was utilized to optimize potency and physicochemical parameters, including
conformational and stereochemical control via the introduction of a
gem-dimethyl group (Thorpe–Ingold effect) or stereocenter,
deuteration, and fluorine, into the inhibitors. Virtually, all inhibitors were found to
display a submicromolar potency against SARS-CoV-2 and MERS-CoV 3CLpro, and the inhibitory
activities were confirmed by a newly established fast and safe cell-based assay.
Furthermore, several deuterated inhibitors, which are likely to exhibit improved
pharmacokinetics, were found to be equipotent with the corresponding nondeuterated
inhibitors. The fluorine-walk approach was applied to explore bioisosteric replacements for
the phenyl ring in GC376 by replacing one or more hydrogen atoms. The effects of these
modifications included unanticipated binding modes of the F-substituted phenyl ring and
modestly enhanced potency. The introduction of multiple fluorine atoms resulted in an
orientation that allowed the fluorine atoms to engage in H-bonding with residues in the
S4 pocket, although with suboptimal bond angles. High-resolution cocrystal
structures with an array of inhibitors unraveled the mechanism of action and provided
valuable insights regarding the binding of the inhibitors to the active site and the
identity of the structural determinants involved in binding. The 4,4-difluorocyclohexane
methyl moiety connected to the benzylic carbon, coupled with the directional control
imparted by the chiral center, resulted in a near-optimal fit in this series (Table , entry 15). Collectively, the results of the
studies described herein are significant and timely and provide an effective launching pad
for conducting further preclinical studies.
Experimental Section
General
Reagents and dry solvents were purchased from various chemical suppliers (Sigma-Aldrich,
Acros Organics, Chem-Impex, TCI America, Oakwood Chemical, APExBIO, Cambridge Isotopes,
Alpha Aesar, Fisher, and Advanced Chemblocks) and were used as obtained. Silica gel
(230–450 mesh) used for flash chromatography was purchased from Sorbent
Technologies (Atlanta, GA). Thin-layer chromatography was performed using Analtech silica
gel plates. Visualization was accomplished using UV light and/or iodine. NMR spectra were
recorded in CDCl3 or DMSO-d6 using a Varian XL-400
spectrometer. The purity of most of the aldehyde inhibitors was found to be ≥95%,
determined by absolute qNMR analysis using a Bruker AV III 500 equipped with a CPDUL
CRYOprobe and CASE autosampler at the KU NMR lab.[54] Dimethyl sulfone
TraceCERT was used as the internal calibrant. High-resolution mass
spectrometry (HRMS) was performed at the Wichita State University Mass Spectrometry lab
using an Orbitrap Velos Pro mass spectrometer (ThermoFisher, Waltham, MA) equipped with an
electrospray ion source.
Synthesis of Compounds
Preparation of Compounds 1–5a, 9a, and
17–22a
General Procedure
To a solution of alcohol (1 equiv) (Table )
in anhydrous acetonitrile (10 mL/g alcohol) were added
N,N′-disuccinimidyl carbonate (1.2 equiv)
and triethyl amine (TEA) (3.0 equiv), and the reaction mixture was stirred for 4 h at
room temperature. The solvent was removed in vacuo, and the residue
was dissolved in ethyl acetate (40 mL/g alcohol). The organic phase was washed with
saturated aqueous NaHCO3 (2 × 20 mL/g alcohol), followed by brine (20
mL/g alcohol). The organic layers were combined and dried over anhydrous
Na2SO4, filtered, and concentrated in vacuo
to yield the mixed carbonate, which was used in the next step without further
purification.To a solution of Leu–Gln surrogate amino alcohol (1.0 equiv) in dry methylene
chloride (10 mL/g of amino alcohol) was added TEA (1.5 equiv), and the reaction
mixture was stirred for 20 min at room temperature (solution 1). In a separate flask,
the mixed carbonate was dissolved in dry methylene chloride (10 mL/g of carbonate)
(solution 2). Solution 1 was added to solution 2, and the reaction mixture was stirred
for 3 h at room temperature. Methylene chloride was added to the organic phase (40
mL/g of carbonate) and then washed with saturated aqueous NaHCO3 (2 ×
20 mL/g alcohol), followed by brine (20 mL/g alcohol). The organic phase was dried
over anhydrous Na2SO4, filtered, and concentrated in
vacuo. The resultant crude product was purified by flash chromatography
(hexane/ethyl acetate) to yield dipeptidyl alcohol a as a white
solid.
To a solution of dipeptidyl alcohol a (1 equiv) in anhydrous
dichloromethane (300 mL/g dipeptidyl alcohol) kept at 0–5 °C under a
N2 atmosphere was added Dess–Martin periodinane reagent (3.0
equiv), and the reaction mixture was stirred for 3 h at 15–20 °C. The
organic phase was washed with 10% aq Na2S2O3 (2
×100 mL/g dipeptidyl alcohol), followed by saturated aqueous NaHCO3 (2
× 100 mL/g dipeptidyl alcohol), distilled water (2 × 100 mL/g dipeptidyl
alcohol), and brine (100 mL/g dipeptidyl alcohol). The organic phase was dried over
anhydrous Na2SO4, filtered, and concentrated in
vacuo. The resulting crude product was purified by flash chromatography
(hexane/ethyl acetate) to yield aldehyde b as a white solid.
To a solution of dipeptidyl aldehyde b (1 equiv) in ethyl acetate (10
mL/g of dipeptidyl aldehyde) was added absolute ethanol (5 mL/g of dipeptidyl
aldehyde) with stirring, followed by a solution of sodium bisulfite (1 equiv) in water
(1 mL/g of dipeptidyl aldehyde). The reaction mixture was stirred for 3 h at 50
°C. The reaction mixture was allowed to cool to room temperature and then
vacuum-filtered. The solid was thoroughly washed with absolute ethanol, and the
filtrate was dried over anhydrous sodium sulfate, filtered, and concentrated to yield
a white solid. The white solid was stirred with dry ethyl ether (3 ×10 mL/g of
dipeptidyl aldehyde), followed by careful removal of the solvent using a pipette and
dried using a vacuum pump for 2 h to yield dipeptidyl bisulfite adduct c
as a white solid.
Cloning and Expression of the 3CLpro of SARS-CoV-2 and FRET Enzyme Assays
The codon-optimized cDNA of full-length 3CLpro of SARS-CoV-2 (GenBank number
MN908947.3) fused with sequences encoding six histidines at the N-terminal was
synthesized by Integrated DNA (Coralville, IA). The synthesized gene was subcloned
into the pET-28a(+) vector. The expression and purification of SARS-CoV-2 3CLpro were
conducted following a standard procedure described previously.[19]
Briefly, a stock solution of an inhibitor was prepared in dimethyl sulfoxide (DMSO)
and diluted in assay buffer composed of 20 mM
4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer, pH 8, containing
NaCl (200 mM), ethylenediaminetetraacetic acid (EDTA) (0.4 mM), glycerol (60%), and 6
mM dithiothreitol (DTT). The SARS-CoV-2 3CLpro was mixed with serial dilutions of
compound or with DMSO in 25 μL of assay buffer and incubated at 37 °C for 1
h, followed by the addition of 25 μL of assay buffer containing substrate
(FAM-SAVLQ/SG-QXL520, AnaSpec, Fremont, CA). The substrate was derived from the
cleavage sites on the viral polyproteins of SARS-CoV. Fluorescence readings were
obtained using an excitation wavelength of 480 nm and an emission wavelength of 520 nm
on a fluorescence microplate reader (FLx800; Biotec, Winooski, VT) 1 h following the
addition of substrate. Relative fluorescence units (RFUs) were determined by
subtracting background values (substrate-containing well without protease) from the
raw fluorescence values, as described previously. The dose-dependent FRET inhibition
curves were fitted with a variable slope using GraphPad Prism software (GraphPad, La
Jolla, CA) to determine the IC50 values of the compounds. The expression
and purification of the 3CLpro of MERS-CoV, as well as the FRET enzyme assays, were
performed as described previously.[19−21,33]
Cell-Based Assay to Screen SARS-CoV-2 3CLpro Inhibitors
Two plasmids, pR-SARS-CoV-2 3CLpro and pGlo-VRLQS, were used for the system (Figure /panel A). First, the open reading frame of
SARS-CoV-2 3CLpro was cloned to the reverse genetics system of PRRSV with GFP[35] and designated as pR-SARS-CoV-2 3CLpro. The GFP gene was replaced with
SARS-CoV-2 3CLpro gene with AflII and MluI enzyme
sites. As a negative control, the inactive form of the 3CLpro was introduced to
pR-SARS-CoV-2 3CLpro by the mutagenesis, and resulting plasmid was designated as
pR-SARS-CoV-2 3CLpro C145A. Second, the plasmid was utilized with the pGloSensor
caspase-3/7 biosensor (Promega, Madison, WI), which contains a caspase-3/7 cleavage site
engineered in the Firefly luciferase gene. This plasmid also contains intact Renilla
luciferase gene as an expression control. After transfection of pGloSensor caspase-3/7
to cells, the Firefly luciferase is expressed as an inactive form, and in the presence
of caspase-3/7, it is activated after the cleavage. The caspase-3/7 cleavage site in
this plasmid was replaced with CoV 3CLpro recognition sequences, VRLQS, designated as
pGlo-VRLQS.[34] As a result, the expressed inactive luciferase is
activated by the cleavage with CoV 3CLpro in the cells (Figure /panel A). HEK293T cells were used for transfection and the
compound screening (Figure /panel B). One day
old HEK293T cells in 48-well plates were cotransfected with two plasmids, pR-SARS-CoV-2
3CLpro and pGlo-VRLQS, or pR-SARS-CoV-2 3CLpro C145A (serves as a negative control) and
pGlo-VRLQS. Following morning, the medium containing Mock-DMSO or serial concentrations
of each compound were replaced to the transfected cells and incubated at 37 °C for
6 h. Cell lysates were prepared for testing the levels of Firefly and Renilla
luciferases (Dual Luciferase assay kit, Promega) in a luminometer (Promega). The
expression levels of Firefly luciferase were normalized with Renilla luciferase levels.
The transfection of pR-SARS-CoV-2 3CLpro C145A and pGlo-VRLQS resulted in minimal levels
of Firefly luciferase, and this was applied to adjust all Firefly expression levels. The
inhibition curve (Figure /panel C) for each
compound was prepared, and the 50% effective concentration (EC50) values were
determined by GraphPad Prism software using a variable slope (GraphPad, La Jolla, CA).
Compounds 4c and 15c were selected and examined for the
antiviral effects with live SARS-CoV-2 in Vero E6 cells as described before,[33] and the EC50 was calculated by the same method described
above. A known SARS-CoV-2 3CLpro inhibitor, GC376, was used as the positive control in
each experiment. When GC376 and the selected compounds were examined if they inhibited
the replication of PRRSV, none of them resulted in the reduction of viral
replication.
HEK293T cells grown in 96-well plates were incubated with various concentrations
(1–100 μM) of each compound for 72 h. Cell cytotoxicity was measured by a
CytoTox 96 nonradioactive cytotoxicity assay kit (Promega), and the CC50
values were calculated using a variable slope by GraphPad Prism software.
Crystallization and Data Collection
Purified SARS-CoV-2 3CLpro[19] in 100 mM NaCl, 20 mM Tris pH 8.0 was
concentrated to 9.6 mg/mL (0.28 mM) for crystallization screening. All crystallization
experiments were set up using an NT8 drop-setting robot (Formulatrix Inc.) and UVXPO MRC
(Molecular Dimensions) sitting drop vapor diffusion plates at 18 °C. One hundred
nanoliters of protein and 100 nL of crystallization solution were dispensed and
equilibrated against 50 μL of the latter. A stock solution of 100 mM compound was
prepared in DMSO, and the SARS-CoV-2 3CLpro:compound complex was prepared by mixing 1
μL of the ligand (2 mM) with 49 μL (0.28 mM) of SARS2 3CLpro and incubating
on ice for 1 h. Crystals were obtained in 1–2 days from various conditions for
the following complexes: 8b: Proplex HT screen (Molecular Dimensions)
condition F7 (0.5 M ammonium sulfate, 100 mM MES pH 6.5); 12b,
13c, 14c, and 21c: Index HT screen (Hampton
Research) condition D10 (20% (w/v) poly(ethylene glycol) (PEG) 5000 monomethyl ether
(MME), 100 mM Bis–Tris pH 6.5); 19b and 20b: Proplex HT
screen (Rigaku Reagents) condition C5 (20% (w/v) PEG 4000, 100 mM Tris pH 8.0);
1c: Index HT screen (Hampton Research) condition F2 (20% (w/v) PEG 2000
MME, 100 mM Tris pH 8.5, 200 mM Trimethylamine N-oxide dihydrate); 3c:
Index HT screen (Hampton Research) condition F5 (17% (w/v) PEG 10 000, 100 mM
Bis–Tris pH 5.5, 100 mM ammonium acetate); 5c: Index HT screen
(Hampton Research) condition F1 (10% (w/v) PEG 3350, 100 mM Hepes pH 7.5, 200
L-proline); and 17c and 18c: Index HT screen (Rigaku Reagents)
condition H11 (30% (w/v) PEG 2000 MME, 100 mM potassium thiocyanate). Samples were
transferred to a fresh drop composed of 80% crystallization solution and 20% (v/v) PEG
200 and stored in liquid nitrogen. Crystals of SARS-CoV-2 3CLpro with 8b
were transferred to a cryoprotectant solution containing 80% crystallant and 20% (v/v)
glycerol prior to freezing. X-ray diffraction data were collected at the Advanced Photon
Source beamline except for the SARS-CoV-2 3CLpro complex with 14c, which
were collected at the National Synchrotron Light Source II (NSLS-II) AMX beamline
17-ID-1. All diffraction data were collected using a Dectris Eiger2 X 9M pixel array
detector.
Structure Solution and Refinement
Intensities were integrated using XDS[40,41] via Autoproc,[42] and the Laue class
analysis and data scaling were performed with Aimless.[43] Structure
solution was conducted by molecular replacement with Phaser[44] using a
previously determined structure of SARS2 3CLpro (PDB 6XMK[19]) as the
search model. Structure refinement and manual model building were conducted with
Phenix[45] and Coot,[46] respectively. Disordered
side chains were truncated to the point for which electron density could be observed.
Structure validation was conducted with Molprobity,[47] and figures
were prepared using the CCP4MG[48] package. Crystallographic data are
provided in Table S1.[49−53]
Authors: Chamandi S Dampalla; Athri D Rathnayake; Anushka C Galasiti Kankanamalage; Yunjeong Kim; Krishani Dinali Perera; Harry Nhat Nguyen; Matthew J Miller; Trent K Madden; Hunter R Picard; Hayden A Thurman; Maithri M Kashipathy; Lijun Liu; Kevin P Battaile; Scott Lovell; Kyeong-Ok Chang; William C Groutas Journal: J Med Chem Date: 2022-05-31 Impact factor: 8.039