Pseudo-Dipeptide Bearing α,α-Difluoromethyl Ketone Moiety as Electrophilic Warhead with Activity against Coronaviruses

The synthesis of α-fluorinated methyl ketones has always been challenging. New methods based on the homologation chemistry via nucleophilic halocarbenoid transfer, carried out recently in our labs, allowed us to design and synthesize a target-directed dipeptidyl α,α-difluoromethyl ketone (DFMK) 8 as a potential antiviral agent with activity against human coronaviruses. The ability of the newly synthesized compound to inhibit viral replication was evaluated by a viral cytopathic effect (CPE)-based assay performed on MCR5 cells infected with one of the four human coronaviruses associated with respiratory distress, i.e., hCoV-229E, showing antiproliferative activity in the micromolar range (EC50 = 12.9 ± 1.22 µM), with a very low cytotoxicity profile (CC50 = 170 ± 3.79 µM, 307 ± 11.63 µM, and 174 ± 7.6 µM for A549, human embryonic lung fibroblasts (HELFs), and MRC5 cells, respectively). Docking and molecular dynamics simulations studies indicated that 8 efficaciously binds to the intended target hCoV-229E main protease (Mpro). Moreover, due to the high similarity between hCoV-229E Mpro and SARS-CoV-2 Mpro, we also performed the in silico analysis towards the second target, which showed results comparable to those obtained for hCoV-229E Mpro and promising in terms of energy of binding and docking pose.


Introduction
In the last few decades, the recurring occurrence of viral pandemics worldwide, i.e., those caused by coronaviruses SARS-CoV (2003), MERS-CoV (2012), and SARS-CoV-2 (2019), indicated that outbreaks of emerging viruses are a severe threat for public health [1]. The current therapeutic arsenal to counter viral infections is confined to drugs that can be deployed against few viruses, mostly producing persistent infections, e.g., human immunodeficiency virus (HIV) [2], hepatitis B virus (HBV), hepatitis C virus (HCV) [3,4], some herpes viruses [5] and influenza virus. However, no specific antivirals have been approved for coronaviruses, with the sole exception of remdesivir (Veklury ® ), a repurposed broad-spectrum RNA-dependent RNA polymerase inhibitor which has been recently approved against SARS-CoV-2, the causative agent of COVID-19 [6][7][8]. Therefore, identifying new direct-acting antiviral agents with potent activity, non-overlapping resistance profiles, limited drug interactions, and minimal adverse effects is urgently needed to treat viral diseases, including COVID-19 [9] effectively. Part of our ongoing research is aimed at the discovery of small molecules endowed with an "electrophilic warhead" able to inactivate parasitic cysteine proteases by the formation of a reversible or irreversible covalent bond with the catalytic residue of the intended targets [9][10][11][12]. Peptide-based α-fluorinated ketones have long been considered compelling substrates in the drug discovery area to develop selective enzyme inhibitors and activitybased probes [13]. The α-fluorination of the methyl ketone functional group profoundly modifies the physicochemical properties of the related substrate and the reactivity of its carbonyl group [14]. The presence of one or more fluorine atoms in the terminal α position of a methyl ketone moiety enhances the electrophilicity of the carbonyl group, and the resulting fluorinated compound becomes more susceptible to undergo a nucleophilic attack by the catalytic residues of proteolytic enzymes, leading to a slow-binding competitive inhibition of the target by forming hemi(thio)ketal adducts [15][16][17]. Among the threeterminal fluoromethyl groups (-CH 2 F, -CHF 2 , and -CF 3 ) next to the carbonyl, -CHF 2 is the most interesting and yet least explored [18]. This difluorinated group exerts a substantial electron-withdrawing effect on the adjacent carbonyl, making it more reactive and acting as a weakly acidic and lipophilic hydrogen bond donor. It could be considered a valid bioisostere of the hydroxyl or thiol group [19][20][21][22]. To date, a considerable number of α,α-difluoromethyl ketones (DFMKs) have been prepared and evaluated mostly as protease inhibitors ( Figure 1). Non-peptidic DFMKs 1 and 2 have been proposed as reversible covalent inhibitors of the Anopheles gambiae acetylcholinesterase (AgAChE) [23], whereas compound 3 was investigated as an agonist of the γ-aminobutyric acid type B (GABA B ) receptor [24]. Compounds 4 and 5 are the only two examples of peptidic DFMKs reported in the literature so far [25,26]. They showed inhibitory activity against bovine pancreatic α-chymotrypsin and porcine pancreatic elastase. Quaternary DFMK 6 instead was investigated as an HIV-1 protease inhibitor [27].
Part of our ongoing research is aimed at the discovery of small molecules endowed with an "electrophilic warhead" able to inactivate parasitic cysteine proteases by the formation of a reversible or irreversible covalent bond with the catalytic residue of the intended targets [9][10][11][12]. Peptide-based α-fluorinated ketones have long been considered compelling substrates in the drug discovery area to develop selective enzyme inhibitors and activity-based probes [13]. The α-fluorination of the methyl ketone functional group profoundly modifies the physicochemical properties of the related substrate and the reactivity of its carbonyl group [14]. The presence of one or more fluorine atoms in the terminal α position of a methyl ketone moiety enhances the electrophilicity of the carbonyl group, and the resulting fluorinated compound becomes more susceptible to undergo a nucleophilic attack by the catalytic residues of proteolytic enzymes, leading to a slow-binding competitive inhibition of the target by forming hemi(thio)ketal adducts [15][16][17]. Among the three-terminal fluoromethyl groups (-CH2F, -CHF2, and -CF3) next to the carbonyl, -CHF2 is the most interesting and yet least explored [18]. This difluorinated group exerts a substantial electron-withdrawing effect on the adjacent carbonyl, making it more reactive and acting as a weakly acidic and lipophilic hydrogen bond donor. It could be considered a valid bioisostere of the hydroxyl or thiol group [19][20][21][22]. To date, a considerable number of α,α-difluoromethyl ketones (DFMKs) have been prepared and evaluated mostly as protease inhibitors ( Figure 1). Non-peptidic DFMKs 1 and 2 have been proposed as reversible covalent inhibitors of the Anopheles gambiae acetylcholinesterase (AgAChE) [23], whereas compound 3 was investigated as an agonist of the γ-aminobutyric acid type B (GABAB) receptor [24]. Compounds 4 and 5 are the only two examples of peptidic DFMKs reported in the literature so far [25,26]. They showed inhibitory activity against bovine pancreatic α-chymotrypsin and porcine pancreatic elastase. Quaternary DFMK 6 instead was investigated as an HIV-1 protease inhibitor [27]. On the other hand, both peptidic and non-peptidic trifluoromethyl ketones (TFMKs) have been widely investigated as protease inhibitors. Besides, a considerable number of biologically active peptidic structures bearing the C-terminal -COCF3 electrophilic warhead have been synthesized [13,28]. In particular, the dipeptidyl TFMK 7 (drawn in the predominant hydrated form; Figure 2) was evaluated against SARS-CoV M pro in vitro, and On the other hand, both peptidic and non-peptidic trifluoromethyl ketones (TFMKs) have been widely investigated as protease inhibitors. Besides, a considerable number of biologically active peptidic structures bearing the C-terminal -COCF 3 electrophilic warhead have been synthesized [13,28]. In particular, the dipeptidyl TFMK 7 (drawn in the predominant hydrated form; Figure 2) was evaluated against SARS-CoV M pro in vitro, and a good inhibitor activity was detected [29]. However, compound 7 and its derivatives were not investigated further. a good inhibitor activity was detected [29]. However, compound 7 and its derivatives were not investigated further. Taking 7 as reference and considering the topicality of the current health emergency, the interest of our ongoing research program aimed at the discovery of new antiviral agents [30][31][32] shifted on the synthesis of the pseudo-dipeptide Z-Leu-Homophe-CHF2 (8) and on the evaluation of its anti-coronavirus profile. Z-Leu-Homophe-CHF2 retains the overall chemical backbone of the SARS-CoV M pro inhibitor 7, with the exception of the Cterminal electrophilic warhead (i.e., DMK in place of TMK) and the Phe residue at the P1 site that was replaced by its superior homolog Homophe ( Figure 2).

Chemistry
The preparation of the targeted α,α-difluoromethyl ketone was achieved, levering on recent chemistry developed by the Pace's research group, dealing with the highly chemoselective delivery of α-substituted methyl-type carbanions to competent electrophilic platforms [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47]. In particular, such C1 synthons proved to be excellent nucleophilic elements for acylating Weinreb amides, as documented in extensive work focusing on the access to α-halogenated ketones, inter alia. At the outset of our synthetic plan, we were cognizant of the difficulties of applying such chemistry to complex peptidyl manifolds. However, the success of the tactic observed in the chemoenzymatic synthesis of Nelfinavir that we reported in 2018 made us confident on the feasibility of such an approach [47].
The chemoselective synthesis of the dipeptidyl α,α-difluoromethyl ketone 8 (Scheme 1) consists of a four-step synthetic procedure, starting from the commercially available Fmoc-Homophe-OH, which was converted into the corresponding Weinreb amide 10 with HN(Me)(OMe) HCl (DMHA). Then, the Fmoc group was removed through piperidine treatment and, the resulting free-amine 11 underwent a standard EDCI/HOBt coupling reaction with Z-Leu-OH to afford 12 in 65% overall yield. Finally, it was converted into 8 via the interrupted homologation with a formal CHF2 transfer agent, generated upon treatment of the commercially available TMSCHF2 with potassium tert-pentoxide (i.e., tert-amylate) as the activating factor [43].
Pleasingly, the difluoromethylation proceeded under full chemo control, leaving untouched the additional sensitive elements present (amides and carbamate). An excess of potassium tert-amylate was used (compared to TMSCHF2) to ensure the deprotonation of the acidic NH groups of the amide linkages. Under these conditions, the desired α,αdifluoroketone was prepared in excellent yield (87%) as an inseparable mixture of two Taking 7 as reference and considering the topicality of the current health emergency, the interest of our ongoing research program aimed at the discovery of new antiviral agents [30][31][32] shifted on the synthesis of the pseudo-dipeptide Z-Leu-Homophe-CHF 2 (8) and on the evaluation of its anti-coronavirus profile. Z-Leu-Homophe-CHF 2 retains the overall chemical backbone of the SARS-CoV M pro inhibitor 7, with the exception of the C-terminal electrophilic warhead (i.e., DMK in place of TMK) and the Phe residue at the P 1 site that was replaced by its superior homolog Homophe ( Figure 2).

Chemistry
The preparation of the targeted α,α-difluoromethyl ketone was achieved, levering on recent chemistry developed by the Pace's research group, dealing with the highly chemoselective delivery of α-substituted methyl-type carbanions to competent electrophilic platforms [33][34][35][36][37][38][39][40][41][42][43][44][45][46][47]. In particular, such C1 synthons proved to be excellent nucleophilic elements for acylating Weinreb amides, as documented in extensive work focusing on the access to α-halogenated ketones, inter alia. At the outset of our synthetic plan, we were cognizant of the difficulties of applying such chemistry to complex peptidyl manifolds. However, the success of the tactic observed in the chemoenzymatic synthesis of Nelfinavir that we reported in 2018 made us confident on the feasibility of such an approach [47].
The chemoselective synthesis of the dipeptidyl α,α-difluoromethyl ketone 8 (Scheme 1) consists of a four-step synthetic procedure, starting from the commercially available Fmoc-Homophe-OH, which was converted into the corresponding Weinreb amide 10 with HN(Me)(OMe) HCl (DMHA). Then, the Fmoc group was removed through piperidine treatment and, the resulting free-amine 11 underwent a standard EDCI/HOBt coupling reaction with Z-Leu-OH to afford 12 in 65% overall yield. Finally, it was converted into 8 via the interrupted homologation with a formal CHF 2 transfer agent, generated upon treatment of the commercially available TMSCHF 2 with potassium tert-pentoxide (i.e., tert-amylate) as the activating factor [43]. NMR spectroscopic data unambiguously supported the presence of the hydrated form of compound 8 as a mixture of two diastereomers (3:1 ratio). The 1 H NMR spectrum of 8a-b in methanol-d4 exhibited the characteristic signal ascribed to the -CHF2 group (5.69 ppm major isomer; 5.76 ppm minor isomer, 2 JH,F = 54.4 Hz) (see Figure S1). As expected, under the experimental conditions, the compound underwent the partial racemization at α position of -COCHF2. The 13 C NMR spectrum did not show any carbonyl signal when performed in methanol-d4; a peak at 96.8 ppm corresponding to the quaternary carbon of the hydrated ketone form was observed instead (see Figure S2). In markedly non-polar solvents such as benzene-d6, the equilibrium is slightly shifted toward the ketone form 9ab, as revealed by the carbonyl signal at 197.3/197.2 ppm in 13 C NMR spectrum (See Figure  S3). HRMS spectrum displays the peak ascribed to the hydrated form as sodiated species at 501.2176 m/z [M + Na] + and the peak ascribed to the ketone form as sodiated species at 483.2065 m/z [M + Na] + (see Figure S4). These experimental data suggest that the equilibrium hydrated form/carbonyl form is highly solvent-dependent.

Biological Evaluation
The ability of Z-Leu-Homophe-CHF2 to inhibit the replication of hCoV-229E, one of the four endemic human coronaviruses and a common cause of upper respiratory tract infections [51], was investigated by a CPE-based assay in normal human lung fibroblasts. To this end, MRC5 cells were infected at a low multiplicity of infection (MOI) to allow for multicycle replication. At an extended endpoint measurement (72 h post-infection, p.i.), a cell viability assay was used as a surrogate of CPE to measure inhibition of hCoV-229E replication [52]. As shown in Figure 3, significant concentration-dependent inhibition of the replication of hCoV-229 was observed in MRC5 cells treated with Z-Leu-Homophe-CHF2. The EC50 and EC90 were 12.9 ± 1.22 µM and 38.6 ± 0.67 µM, respectively. As expected, remdesivir, used as a control of a reference anti-CoV agent, inhibited hCoV-229 replication in the low nanomolar range (EC50 0.034 ± 0.0025 µM, CC50 >10 µM), as previously observed in the same type of antiviral assay [52]. Then, the Z-Leu-Homophe-CHF2 cytotoxicity was evaluated in different types of uninfected cells, such as the lung fibroblasts MRC5 and human embryonic lung fibroblasts (HELFs), and the lung epithelial cells A549 ( Figure S5). The cytotoxic concentrations (CC50) were 170 ± 3.79 µM, 307 ± 11.63 µM, and 174 ± 7.6 µM for A549, HELFs, and MRC5 cells, respectively, thus indicating that the Pleasingly, the difluoromethylation proceeded under full chemo control, leaving untouched the additional sensitive elements present (amides and carbamate). An excess of potassium tert-amylate was used (compared to TMSCHF 2 ) to ensure the deprotonation of the acidic NH groups of the amide linkages. Under these conditions, the desired α,α-difluoroketone was prepared in excellent yield (87%) as an inseparable mixture of two diastereomers (8a-b). Structurally, the carbonyl group was present in the hydrate form (gem-diol) in agreement with literature precedents for distinct (poly)α-fluoroketones [18,[48][49][50].
NMR spectroscopic data unambiguously supported the presence of the hydrated form of compound 8 as a mixture of two diastereomers (3:1 ratio). The 1 H NMR spectrum of 8a-b in methanol-d 4 exhibited the characteristic signal ascribed to the -CHF 2 group (5.69 ppm major isomer; 5.76 ppm minor isomer, 2 J H,F = 54.4 Hz) (see Figure S1). As expected, under the experimental conditions, the compound underwent the partial racemization at α position of -COCHF 2 . The 13 C NMR spectrum did not show any carbonyl signal when performed in methanol-d 4 ; a peak at 96.8 ppm corresponding to the quaternary carbon of the hydrated ketone form was observed instead (see Figure S2). In markedly non-polar solvents such as benzene-d 6 , the equilibrium is slightly shifted toward the ketone form 9a-b, as revealed by the carbonyl signal at 197.3/197.2 ppm in 13 C NMR spectrum (See Figure S3). HRMS spectrum displays the peak ascribed to the hydrated form as sodiated species at 501.2176 m/z [M + Na] + and the peak ascribed to the ketone form as sodiated species at 483.2065 m/z [M + Na] + (see Figure S4). These experimental data suggest that the equilibrium hydrated form/carbonyl form is highly solvent-dependent.

Biological Evaluation
The ability of Z-Leu-Homophe-CHF 2 to inhibit the replication of hCoV-229E, one of the four endemic human coronaviruses and a common cause of upper respiratory tract infections [51], was investigated by a CPE-based assay in normal human lung fibroblasts. To this end, MRC5 cells were infected at a low multiplicity of infection (MOI) to allow for multicycle replication. At an extended endpoint measurement (72 h post-infection, p.i.), a cell viability assay was used as a surrogate of CPE to measure inhibition of hCoV-229E replication [52]. As shown in Figure 3, significant concentration-dependent inhibition of the replication of hCoV-229 was observed in MRC5 cells treated with Z-Leu-Homophe-CHF 2 . The EC 50 and EC 90 were 12.9 ± 1.22 µM and 38.6 ± 0.67 µM, respectively. As expected, remdesivir, used as a control of a reference anti-CoV agent, inhibited hCoV-229 replication in the low nanomolar range (EC 50 0.034 ± 0.0025 µM, CC 50 >10 µM), as previously observed in the same type of antiviral assay [52]. Then, the Z-Leu-Homophe-CHF 2 cytotoxicity was evaluated in different types of uninfected cells, such as the lung fibroblasts MRC5 and human embryonic lung fibroblasts (HELFs), and the lung epithelial cells A549 ( Figure S5). The cytotoxic concentrations (CC 50 ) were 170 ± 3.79 µM, 307 ± 11.63 µM, and 174 ± 7.6 µM for A549, HELFs, and MRC5 cells, respectively, thus indicating that the antiviral activity of Z-Leu-Homophe-CHF 2 against hCoV-229E was not due to the cytotoxicity of the target cells themselves.

Computational
The 2.37 Å crystal structure resolution (PDB ID: 1P9U) of hCoV-229E M pro [53] shows that the protein comprises three domains ( Figure 4a). Domains I and II (residues 8 to 99 and 100 to 183, respectively) ( Figure 4c). have a high similarity with the structure of the 3C proteinases of picornavirus and chymotrypsin. The substrate-binding site is located in an empty area between these two domains. A long helix loop (residues 184 to 199) connects domain II to the C-terminal domain (domain III, residues 200 to 300). This globular cluster of five helices is responsible for the catalytic activity of M pro [54]. hCoV M pro forms a dimer (the contact interface is mainly located between the domain II of monomer A and the N-terminal residues of monomer B and has an area of~1300 Å 2 ) with the two monomers oriented in the opposite direction to each other. In the M pro dimer, the N-terminal amino acid residues are in close contact between domains II and III of the parent monomer and domain II of the other monomer, creating specific interactions suitable for binding this portion with high affinity after autocleavage. This mechanism would allow the catalytic site to bind other cleavage sites in the polyprotein. Cys144 and His41 (Figure 4b), in the active site of hCoV-229E M pro to form a catalytic dyad within the S1' pocket [53]. Molecular modeling studies were conducted to identify and evaluate the receptor/ligand recognition's key molecular interactions. Molecular docking was performed using the crystal structure of hCoV-229E M pro (PDB ID: 1P9U). The poses presenting the difluoro carbonyl group in the vicinity of the His41/Cys144 catalytic dyad were selected for further stability studies. All three selected poses show a different orientation of the phenethyl group and the isobutyl group (Figure 5a) within the binding site. In fact, in pose 1, the phenethyl group is inserted inside the S2 sub-pocket, while in pose 2 the phenethyl is arranged in the region overlying the S1' pocket and the isobutyl group occupies the S2 sub-pocket. Pose 3 reverses the positions of the two groups with respect to pose 1. In fact, the phenethyl group is incorporated inside the S1 sub-pocket and the isobutyl group inside the S2 pocket. The only difference between pose 2 and pose 3 lies in the position of the phenethyl group.
To verify the reliability of the docking results, molecular dynamic (MD) simulations were performed based on the results of the molecular docking. The first three Z-Leu-Homophe-CHF2 docking poses were evaluated in complex with hCoV-229E M pro for 5 ns of MD simulation. To explore the dynamic stability of both systems and to ensure the rationality of the sampling method, progression of the root mean square deviations (RMSD) from the starting structure was analyzed (Figure 5b). The superposition of the coordinates of each complex structure in a trajectory onto the initial structure allowed us to analyze the RMSD of ligand and protein in complex. The RMSD graph of the ligand (Figure 5c) indicates that the conformations of the complex in pose 1 reach equilibrium around 1.5 ns, while in pose 2 and pose 3, after 0.05 ns and 0.15 ns, respectively. Laying 2 provided a lower RMSD than the starting structure (2.11 Å), and the conformations oscillate on average around 1.3 Å. The protein structure in complex with the ligand in pose 2 also shows a more stable RMSD during MD simulation, reaching equilibrium after 1.5 ns. Molecular modeling studies were conducted to identify and evaluate the receptor/ligand recognition's key molecular interactions. Molecular docking was performed using the crystal structure of hCoV-229E M pro (PDB ID: 1P9U). The poses presenting the difluoro carbonyl group in the vicinity of the His41/Cys144 catalytic dyad were selected for further stability studies. All three selected poses show a different orientation of the phenethyl group and the isobutyl group (Figure 5a) within the binding site. In fact, in pose 1, the phenethyl group is inserted inside the S2 sub-pocket, while in pose 2 the phenethyl is arranged in the region overlying the S1' pocket and the isobutyl group occupies the S2 sub-pocket. Pose 3 reverses the positions of the two groups with respect to pose 1. In fact, the phenethyl group is incorporated inside the S1 sub-pocket and the isobutyl group inside the S2 pocket. The only difference between pose 2 and pose 3 lies in the position of the phenethyl group.
To verify the reliability of the docking results, molecular dynamic (MD) simulations were performed based on the results of the molecular docking. The first three Z-Leu-Homophe-CHF 2 docking poses were evaluated in complex with hCoV-229E M pro for 5 ns of MD simulation. To explore the dynamic stability of both systems and to ensure the rationality of the sampling method, progression of the root mean square deviations (RMSD) from the starting structure was analyzed (Figure 5b). The superposition of the coordinates of each complex structure in a trajectory onto the initial structure allowed us to analyze the RMSD of ligand and protein in complex. The RMSD graph of the ligand (Figure 5c) indicates that the conformations of the complex in pose 1 reach equilibrium around 1.5 ns, while in pose 2 and pose 3, after 0.05 ns and 0.15 ns, respectively. Laying 2 provided a lower RMSD than the starting structure (2.11 Å), and the conformations oscillate on average around 1.3 Å. The protein structure in complex with the ligand in pose 2 also shows a more stable RMSD during MD simulation, reaching equilibrium after 1.5 ns. Consequently, we focused on pose 2 for further study. Indeed, it has been previously shown that the S2 sub-pocket in the human coronavirus hCoV-229E M pro due to the limited size of the S2 pocket, the benzyl group (in the studied inhibitors) cannot enter deeply into this site, also due to the low plasticity of the pocket [55]. This could explain the better stability of the protein complex in pose 2.
Pose 2 at the binding site of hCoV-229E M pro forms more hydrogen bonds, with the main residue chain in the substrate-binding pocket, which also helps to lock the inhibitor within the media binding pocket by establishing H bonds with Gln163 and Glu165 at 2.03 Å and 2.36 Å, respectively. The isobutyl group inserts itself deeply inside the S2 hydrophobic pocket, stabilizing from hydrophobic interactions with Ile164 and Pro168, with the Consequently, we focused on pose 2 for further study. Indeed, it has been previously shown that the S2 sub-pocket in the human coronavirus hCoV-229E M pro due to the limited size of the S2 pocket, the benzyl group (in the studied inhibitors) cannot enter deeply into this site, also due to the low plasticity of the pocket [55]. This could explain the better stability of the protein complex in pose 2.
Pose 2 at the binding site of hCoV-229E M pro forms more hydrogen bonds, with the main residue chain in the substrate-binding pocket, which also helps to lock the inhibitor within the media binding pocket by establishing H bonds with Gln163 and Glu165 at 2.03 Å and 2.36 Å, respectively. The isobutyl group inserts itself deeply inside the S2 hydrophobic pocket, stabilizing from hydrophobic interactions with Ile164 and Pro168, with the phenethyl portion by the π-CH bond with Thr47 and π-alkyl with Ala44. The α,α-difluoromethyl ketone group is located at 3.46 Å from Cys144, stabilized by the H bond with Gln163 at 2.25 A. The H bonds between the fluorine atom and Gly142 and Cys144 at 2.59 Å and 2.52 Å could contribute together with Gln163 to stabilize the transition state during the nucleophilic attack on the carbonyl group. The large benzyl group extends into the S4 site, leaving this region of the ligand exposed to the solvent.
The lowest unoccupied molecular orbital (LUMO) and molecular electrostatic potential (MEP) were performed ( Figure 6), with semiempirical PM6 methods, to locate the atom most favorable to the nucleophilic attack by the thiolate ion. phenethyl portion by the π-CH bond with Thr47 and π-alkyl with Ala44. The α,α-difluoromethyl ketone group is located at 3.46 Å from Cys144, stabilized by the H bond with Gln163 at 2.25 A. The H bonds between the fluorine atom and Gly142 and Cys144 at 2.59 Å and 2.52 Å could contribute together with Gln163 to stabilize the transition state during the nucleophilic attack on the carbonyl group. The large benzyl group extends into the S4 site, leaving this region of the ligand exposed to the solvent. The lowest unoccupied molecular orbital (LUMO) and molecular electrostatic potential (MEP) were performed ( Figure 6), with semiempirical PM6 methods, to locate the atom most favorable to the nucleophilic attack by the thiolate ion. The carbonyl group of difluoromethyl ketone shows a LUMO density and a positive electrostatic potential, which are fundamental for the mechanism of the reaction. The LUMO delineates the areas most lacking in electrons, therefore subject to nucleophilic attack by the thiolate anion.
A 100 ns MD simulation was performed to analyze, in greater detail, any changes in the protein-ligand complex. The protein structure and the ligand reached equilibrium after 30 ns and a very similar average RMSD, 2.1 Å, and 2.3 Å for ligand and protein, respectively (see Figure S6). The flexibility of the different residues of hCoV-229E M pro , by calculating the root mean square oscillations (RMSFs) of Cα atoms, was analyzed. A relatively higher RMSF value is obtained around residue 50 (domain I), while more excellent stability occurs in domain II.
Furthermore, to understand the effect of Z-Leu-Homophe-CHF2 binding on the internal dynamics of hCoV-229E M pro , the dynamic cross-correlation matrix (DCCM) was calculated by using the coordinates of Cα atoms from the trajectories. Domain I (residues 8-99) displays correlated motions, while domain II (residues 100-183) shows highly anticorrelated movements. However, domain III (residues 201-300) has a lower strength of anti-correlation motions relative to domain I. Furthermore, it is evident from (See Figure  S6) that the inhibitor binding increases the strength of anti-correlated motions in the region of domain II, which indicates the residual motion of domain III. Overall, the binding of Z-Leu-Homophe-CHF2 with M pro creates a stable environment near the binding cavity.
Given the high similarity between hCoV-229E M pro and SARS-CoV-2 M pro , we decided to investigate the in silico affinity between Z-Leu-Homophe-CHF2 with SARS-CoV-2 M pro . However, there are many common features shared between the two types of proteases, particularly their almost absolute requirement for Gln in the S1 position of the substrate and space for only small amino acid residues as a common target for the design of broad-spectrum antiviral compounds. Since there is no human protease with a specificity for the Gln residue at the substrate's catalytic cleavage site, the interest of this viral target The carbonyl group of difluoromethyl ketone shows a LUMO density and a positive electrostatic potential, which are fundamental for the mechanism of the reaction. The LUMO delineates the areas most lacking in electrons, therefore subject to nucleophilic attack by the thiolate anion.
A 100 ns MD simulation was performed to analyze, in greater detail, any changes in the protein-ligand complex. The protein structure and the ligand reached equilibrium after 30 ns and a very similar average RMSD, 2.1 Å, and 2.3 Å for ligand and protein, respectively (see Figure S6). The flexibility of the different residues of hCoV-229E M pro , by calculating the root mean square oscillations (RMSFs) of Cα atoms, was analyzed. A relatively higher RMSF value is obtained around residue 50 (domain I), while more excellent stability occurs in domain II.
Furthermore, to understand the effect of Z-Leu-Homophe-CHF 2 binding on the internal dynamics of hCoV-229E M pro , the dynamic cross-correlation matrix (DCCM) was calculated by using the coordinates of Cα atoms from the trajectories. Domain I (residues 8-99) displays correlated motions, while domain II (residues 100-183) shows highly anticorrelated movements. However, domain III (residues 201-300) has a lower strength of anti-correlation motions relative to domain I. Furthermore, it is evident from (See Figure S6) that the inhibitor binding increases the strength of anti-correlated motions in the region of domain II, which indicates the residual motion of domain III. Overall, the binding of Z-Leu-Homophe-CHF 2 with M pro creates a stable environment near the binding cavity.
Given the high similarity between hCoV-229E M pro and SARS-CoV-2 M pro , we decided to investigate the in silico affinity between Z-Leu-Homophe-CHF 2 with SARS-CoV-2 M pro .
However, there are many common features shared between the two types of proteases, particularly their almost absolute requirement for Gln in the S1 position of the substrate and space for only small amino acid residues as a common target for the design of broadspectrum antiviral compounds. Since there is no human protease with a specificity for the Gln residue at the substrate's catalytic cleavage site, the interest of this viral target increases, as it increases the possibility that the inhibitors developed will not show host cell toxicity.
The high similarity of the binding site (Figure 7c) between the two enzymes produced comparable results in the energy of the binding and docking pose. Indeed, the free energy of binding (Table 1)  increases, as it increases the possibility that the inhibitors developed will not show host cell toxicity.
The high similarity of the binding site (Figure 7c) between the two enzymes produced comparable results in the energy of the binding and docking pose. Indeed, the free energy of binding (Table 1)    In order to evaluate the scoring function of AutoDock 4.2, the famous N3 inhibitor was docked as the experimental data of the inhibition constant (Ki), and the EC50 values are available, respectively for hCoV-229E M pro [56] and SARS-CoV-2 M pro [57]. The SARS2 experimental Ki value was not available due to the high inhibition potency. However, the results are shown in Table S1, demonstrating the high accuracy of the scoring function used by AutoDock 4.2.
By visualizing in detail the pose of Z-Leu-Homophe-CHF2 inside the pocket of SARS-CoV-2 M pro (Figure 7a), we can see that the difluoromethyl ketone is located at a distance of 3.38 A from Cys145, optimal for a nucleophilic attack on the carbonyl group. The oxygen atom of the aldehyde group also plays a crucial role in stabilizing the inhibitor's conformations, by forming a 2.16 Å hydrogen bond with the backbone of residue Cys143 at the S1 site. Furthermore, the amide bonds on the Z-Leu-Homophe-CHF2 chain form hydrogen bonds with the main chains of Cys145 (3.44 Å), Glu 166 (2.11 Å), and Gln 189 (1.94 Å), respectively. The isobutyl moiety of Z-Leu-Homophe-CHF2 in S2 inserts deeply into the S2 site, stacking with the imidazole ring of His 41. The side chains of Met49, Met165, Asp187, and Pro168, also surround the isobutyl group, producing extensive hydrophobic interactions (Figure 7b).
Furthermore, the difluoromethyl ketone moiety is further stabilized by the halogen interaction with Leu141 at 3.68 Å and the H-halogen bonds with Ser144 and Cys145 at 2.74  Table 1. Calculated binding energies (kcal/mol) and K i (µM) for the binding sites of HCoV-229E M pro and SARS-CoV-2 M pro for Z-Leu-Homophe-CHF 2 .

Enzyme
Calcd. ∆G Calcd. K i (µM) In order to evaluate the scoring function of AutoDock 4.2, the famous N3 inhibitor was docked as the experimental data of the inhibition constant (K i ), and the EC 50 values are available, respectively for hCoV-229E M pro [56] and SARS-CoV-2 M pro [57]. The SARS2 experimental K i value was not available due to the high inhibition potency. However, the results are shown in Table S1, demonstrating the high accuracy of the scoring function used by AutoDock 4.2.
By visualizing in detail the pose of Z-Leu-Homophe-CHF 2 inside the pocket of SARS-CoV-2 M pro (Figure 7a), we can see that the difluoromethyl ketone is located at a distance of 3.38 A from Cys145, optimal for a nucleophilic attack on the carbonyl group. The oxygen atom of the aldehyde group also plays a crucial role in stabilizing the inhibitor's conformations, by forming a 2.16 Å hydrogen bond with the backbone of residue Cys143 at the S1 site. Furthermore, the amide bonds on the Z-Leu-Homophe-CHF 2 chain form hydrogen bonds with the main chains of Cys145 (3.44 Å), Glu 166 (2.11 Å), and Gln 189 (1.94 Å), respectively. The isobutyl moiety of Z-Leu-Homophe-CHF 2 in S2 inserts deeply into the S2 site, stacking with the imidazole ring of His 41. The side chains of Met49, Met165, Asp187, and Pro168, also surround the isobutyl group, producing extensive hydrophobic interactions (Figure 7b). Furthermore, the difluoromethyl ketone moiety is further stabilized by the halogen interaction with Leu141 at 3.68 Å and the H-halogen bonds with Ser144 and Cys145 at 2.74 Å. During the MD simulation, the ligand is further stabilized by further water molecules bridging the residues Asn147 and Glu166 with the ester and amide groups of the ligand. The bulky benzyl group extends into the S4 site cavity, forming π-CH interactions with Gln189. Unlike the complex with E229, this region of the ligand is not exposed to the solvent.
A detailed analysis of the trajectories developed by MD simulation highlights the good stability of the protein-ligand complex. Z-Leu-Homophe-CHF 2 reaches equilibrium after 20 ns, while the protein structure seems to reach a slight flexion at 65 ns, remaining, however, at low RMSD levels. The DCCM plots (See Figure S7)  Despite the small size of the isobutyl substituent, this is possible because the S2 pocket of SARS-CoV-2 M pro is flexible enough to contract and enclose small substituents. This plasticity is expressed in a conformational change of residue Gln189, both in the main and side chains. As a consequence of these changes, the side-chain oxygen of Gln189 can accept a 1.94 Å hydrogen bond from the main-chain NH residue in the Z-Leu-Homophe-CHF 2 complex (Figure 8). However, the affinity of Z-Leu-Homophe-CHF 2 for the S2 pocket of hCoV-229E M pro is good because of an almost ideal match of size and not requiring conformational changes, which this enzyme would not be able to undergo because of the replacement of the flexible Gln189 by the more rigid proline [55]. The more remarkable plasticity of the S2 pocket can accommodate the isobutyl substituent in a deeper way. Thus, both effects could increase the stability, hence the affinity, of Z-Leu-Homophe-CHF 2 against SARS-CoV-2 M pro .
. 2020, 21, x FOR PEER REVIEW 10 of 17 The bulky benzyl group extends into the S4 site cavity, forming π-CH interactions with Gln189. Unlike the complex with E229, this region of the ligand is not exposed to the solvent.
A detailed analysis of the trajectories developed by MD simulation highlights the good stability of the protein-ligand complex. Z-Leu-Homophe-CHF2 reaches equilibrium after 20 ns, while the protein structure seems to reach a slight flexion at 65 ns, remaining, however, at low RMSD levels. The DCCM plots (See Figure S7) of the Z-Leu-Homophe-CHF2-SARS-CoV-2 M pro complex show a correlation region in the binding site domain and anti-correlation zones in domain II and domain III, confirming the analysis of RMSFs (see Figure S7), showing fluctuations between the residues 40-70 (domain II) and 210-240 (domain III).
Despite the small size of the isobutyl substituent, this is possible because the S2 pocket of SARS-CoV-2 M pro is flexible enough to contract and enclose small substituents. This plasticity is expressed in a conformational change of residue Gln189, both in the main and side chains. As a consequence of these changes, the side-chain oxygen of Gln189 can accept a 1.94 Å hydrogen bond from the main-chain NH residue in the Z-Leu-Homophe-CHF2 complex (Figure 8). However, the affinity of Z-Leu-Homophe-CHF2 for the S2 pocket of hCoV-229E M pro is good because of an almost ideal match of size and not requiring conformational changes, which this enzyme would not be able to undergo because of the replacement of the flexible Gln189 by the more rigid proline [55]. The more remarkable plasticity of the S2 pocket can accommodate the isobutyl substituent in a deeper way. Thus, both effects could increase the stability, hence the affinity, of Z-Leu-Homophe-CHF2 against SARS-CoV-2 M pro . Figure 8. Alignment of the binding site amino acids of hCoV-229E M pro (marine green) and SARS-CoV-2 M pro (orange) in complex with Z-Leu-Homophe-CHF2, represented as a hydrophobic surface (hydrophobic: brown, hydrophilic: blue, neutral: white). Sites S1, S1 ', S2, S4 of M pro are indicated.

General Experimental Information
Melting points were determined on a Reichert-Kofler hot-stage microscope (Vienna, Austria) and are uncorrected. Mass spectra were obtained on a Shimadzu QP 1000 instrument (EI, 70 eV) (Shimadzu Corporation, Kyoto, Japan) and on a Brüker maXis 4G instrument (ESI-TOF, HRMS) (Brüker Corporation, Billerica, MA, USA). 1  The center of the solvent signal was used as an internal standard, which was related to TMS with δ 4.87 ppm ( 1 H in CD 3 OD), δ 128.06 ppm ( 13 C in C 6 D 6 ) and δ 49.00 ppm ( 13 C in CD 3 OD). Absolute referencing via Ξ ratio was used for the 19 F NMR spectra. Spin-spin coupling constants (J) are given in Hz. In nearly all cases, the full and unambiguous assignment of all resonances was performed by combined application of standard NMR techniques, such as APT, HSQC, HMBC, HSQC-TOCSY, COSY, and NOESY experiments.
All the reactions were carried out under an inert atmosphere of argon. THF was distilled over Na/benzophenone. Chemicals were purchased from Sigma-Aldrich (Sigma-Aldrich, St. Louis, Missouri, USA), Acros (Thermo Fisher Scientific, NJ, USA), Alfa Aesar (Thermo Fisher, Kandel, Germany), Fluorochem (Hadfield Derbyshire, United Kingdom) and TCI Europe (Tokyo Chemical Industry, Tokyo, Japan). Solutions were evaporated under reduced pressure with a rotary evaporator. TLC was carried out on aluminum sheets precoated with silica gel 60F254 (Merchery-Nagel, Merk, Darmstadt, Germany); the spots were visualized under UV light (λ = 254 nm). (10) To a solution of Fmoc-D-Homophe-OH (1.0 g, 0.0025 mol, 1.0 equiv) in dry CPME (10 mL) was added 1,1 -carbonyldiimidazole (0.444 g, 0.0027 mol, 1.1 equiv) in one portion, and the resulting mixture was allowed to stir for 1 h at rt. Then, N,O-dimethylhydroxylamine hydrochloride (DMHA, 0.270 g, 0.0027 mol, 1.1 equiv) was added to the mixture, turning the solution a cloudy white. The reaction mixture was stirred at rt overnight, then quenched with 3 mL of a saturated aq. solution of NH 4 Cl, and the aqueous layer was extracted with ethyl acetate (2 × 10 mL). The combined organic layers were washed with a saturated aq. solution of NaHCO 3 (3 × 10 mL) and brine (3 × 10 mL) then, dried over anhydrous Na 2 SO 4 and the solvent was removed under reduced pressure to afford the corresponding Fmoc-D-Homophe Weinreb Amide 10, in 95% yield (1.05 g), as a yellow oil without any further purification. 1  To a solution of 10 in dry DMF (10 mL), piperidine (20%) was added and the resulting mixture was left stirring at rt overnight. The reaction mixture was dried under reduced pressure, and the crude product was purified via silica gel chromatography (DCM: MeOH 98:2) to afford compound 11 in 96% yield (0.504 g) as a yellow oil. 1 (12) Compound 11 (176 mg, 0.792 mmol, 1.2 equiv) and Z-Leu-OH (175 mg, 0.66 mmol, 1 equiv) were dissolved in dry DMF (8 mL) and EDCI (152 mg, 0.792 mmol, 1.2 equiv) and HOBt (107 mg, 0.792 mmol, 1.2 equiv) were added to the mixture at rt. The reaction was left stirring under Ar atmosphere at rt overnight. The reaction mixture was then diluted with ethyl acetate and was washed with saturated aq. solution of NH 4 Cl (3 × 10 mL), saturated aq. solution of NaHCO 3 (3 × 10 mL) and brine (3 × 10 mL). The organic layer was dried over anhydrous Na 2 SO 4 , filtered and concentrated under reduced pressure to afford 12 as a colorless liquid which readily crystallizes in a white solid upon trituration with diethyl ether (65% yield, 202 mg). 1  To a solution of compound 12 (120 mg, 0.255 mmol, 1 equiv) in dry THF (5 mL) cooled at 0 • C was added (difluoromethyl)trimethylsilane (0.18 mL, 1.275 mmol, 2 equiv) under Ar atmosphere. Then, potassium tert-pentoxide 0.9 M (1.25 mL, 1.122 mmol, 4.4 equiv) was added dropwise with good stirring at 0 • C during a period of 15 min. The reaction mixture was further stirred to reach rt within 3 h. After complete conversion of the starting material, the reaction mixture was quenched with saturated aqueous NH 4 Cl solution (3 mL) and extracted with ethyl acetate (3 × 5 mL). The organic layer was washed with brine (5 mL), dried over Na 2 SO 4 , filtered and concentrated under reduced pressure. The crude was purified via column chromatography on silica gel (DCM: MeOH 98:2) to afford the corresponding compounds 8a,b as diastereomeric mixture (3:1) in 87% yield (106 mg). 1

Cytotoxicity Assay
HELFs, MRC5, or A549 cells were seeded in 96-well plates (18,000 cells/well), and after 24 h, the cells were exposed to increasing concentrations of Z-Leu-Homophe-CHF 2 (8) or vehicle (DMSO) as control. After 72 h of incubation, the number of viable cells was determined using the CellTiter-Glo Luminescent assay (Promega) according to the specifications of the manufacturer.

Antiviral Assay
To evaluate the antiviral activity of different concentrations of Z-Leu-Homophe-CHF 2 or remdesivir (MedChem Express), MRC5 cells seeded in 48-well plates (30,000 cells/well) and then treated with different concentrations of the compounds during infection with hCoV-229E at 100 PFU/well. Following virus adsorption (2 h at 37 • C), viral inocula were removed, and cells were maintained in a medium containing the corresponding compounds, and 2% FBS. Cells treated with vehicle (DMSO) and cells infected and treated with vehicle were used as mock infection control for normalization and infection control, respectively. After 72 h p.i., cell viability was measured using CellTiter-Glo assay as a surrogate measurement of the viral cytopathic effect (CPE), as previously described [52]. The mean values for Z-Leu-Homophe-CHF 2 (8) and remdesivir were normalized to the mean values for the mock-infected control cells (DMSO-treated), and the concentration that produced 50% of reduced cell viability (EC 50 ) was determined by GraphPad Prism software, version 7.

Molecular Docking
Flexible ligand docking experiments, successfully used in our previous work [31,[59][60][61][62][63][64], were performed employing AutoDock 4.2.6 software implemented in YASARA (v. 20.10.4, YASARA Biosciences GmbH, Vienna, Austria) [65,66], using the three-dimensional crystal structure of hCoV-229E M pro (PDB ID: 1P9S), the three-dimensional crystal structure of SARS-CoV-2 M pro in complex with an inhibitor N3 PRD_002214 (PDB ID: 6LU7), both obtained from the Protein Data Bank (PDB, http://www.rcsb.org), and the Lamarckian genetic algorithm (LGA). The covalent bond between the Cys145 residue and the crystallized ligand has been eliminated. His41 and Cys145 residues were protonated and optimized using YASARA software. The maps were generated by the program AutoGrid (4.2.6) with a spacing of 0.375 Å and dimensions that encompass all atoms extending 5 Å from the surface of the structure of the crystallized ligands. The details were reported in the SI.

Molecular Dynamics Simulations
The MD simulations of the M pro /ligand complexes were performed with the YASARA Structure package (19.11.5) [64]. The details were reported in the SI.

Conclusions
To sum up, in this work, we applied the chemistry of α-fluorinated organometallic methyl type carbanions carried out in our labs, to the design of a peptide-based α,αdifluoromethyl ketone, having the -COCHF 2 moiety at the C-terminal of the peptidic framework as electrophilic warhead able to tackle cysteine proteases which are vital for coronaviruses' survival. Specifically, our pseudo-peptide, with the Z-Leu-Homophe-CHF 2 motif, was developed on the basis of a structurally related lead compound targeting SARS-CoV M pro with activity in the micromolar range. Biological assays performed on cells infected with hCoV-229E, one of the four human coronaviruses associated with respiratory distress, showed that our compound exerts a remarkable antiviral activity (EC 50 = 12.9 µM), with a shallow cytotoxicity profile evaluated in three different types of uninfected cells (A549, HELFs, and MRC5 cells). An exhaustive work of molecular docking and molecular dynamics indicated that this fluorinated pseudo-dipeptide might efficaciously bind to the intended cysteine target HCoV-229E M pro . Moreover, due to the high similarity between the M pro of hCoV-229E and SARS-CoV-2 (the causative agent of COVID-19), we also performed an in silico analysis towards SARS-CoV-2 M pro . The latter study showed that Z-Leu-Homophe-CHF 2 might bind SARS-CoV-2 M pro as effectively as for hCoV-229E M pro , suggesting that this fluorinated pseudo-peptide may be exploited as a new reference structure for the future design of druggable compounds with activity against coronaviruses.