Reexamining Povarov Reaction’s Scope and Limitation in the Generation of HCV-NS4A Peptidomimetics

Department of Pharmaceutical Chemistry, Faculty of Pharmacy, King Abdulaziz University, Jeddah 21589, Saudi Arabia Center for Artificial Intelligence in Precision Medicines, King Abdulaziz University, Jeddah 21589, Saudi Arabia Faculty of Pharmacy, Department of Pharmaceutical Chemistry, Al-Azhar University, Nasr City, Cairo, Egypt Department of Natural Products and Alternative Medicine, Faculty of Pharmacy, King Abdulaziz University, Jeddah 21589, Saudi Arabia Department of Pharmaceutics, Faculty of Pharmacy, King Abdulaziz University, Jeddah 21589, Saudi Arabia


Introduction
Hepatitis C virus (HCV) is a silent killer that starts with mild or even no symptoms, but in many cases, it advances to a life-long illness [1]. e HCV, when activated, attacks liver cells through its E2-envelop protein, maneuvers to penetrate the cell membrane and proceeds to disrupt cellular mechanisms, controls genetic expressions, and leverages cellular machinery to replicate [2,3]. Moreover, HCV can evade the innate immunity challenges and becomes a persistent infection [4]. Consequently, the liver starts to enlarge and develop inflammatory responses followed by necrosis and extensive damage of hepatic tissues leading to cirrhosis that ends with liver cancer [5]. In its 2019 report, the World Health Organization (WHO) estimated that more than 70 million people are infected with HCV regardless of the great progress in the control and treatment since 2013 [6]. e recent success of newly discovered direct antiviral agents (DAAs) led to a significant decline in HCV-related deaths concomitant to an increase in survival and recovery rate of patients suffering from the high viral count in their blood [7]. e approved DAAs up until now are directed to three targets: substrate-site inhibitor of NS3/4A protease, NS5A, and NS5B [8] (Figure 1). We have been interested in exploring NS4A as a target [9,10], aiming to add more drugs to the DAA arsenal that may curb the increasingly emerging resistance against available DAAs. Another important reason for targeting NS4A is that this 54-mer amino acid peptide is a common factor, albeit not structurally or functionally conserved, among all Flaviviridae family. is family of viruses comprises highly pathogenic viruses such as Zika, Dengue, and West Nile fever. In HCV, this small peptide is a multipurpose tool for HCV maturation and replication. It plays irreplaceable roles in activating the NS3 protein (both protease and RNA helicase functions), curbing host-cell immunological responses, and integrating the NS3 into the endoplasmic reticulum.
X-ray structures of NS3 complexed with NS4A revealed that the binding of the NS4A peptide initiates ordering of the N-terminal 28 residues of the NS3 protease into a β-strand and an α-helix [11,12]. It also causes local rearrangements, which are important for a catalytically favorable conformation at the active site. According to serine and alanine scanning studies, the hydrophobic residues Val-23′, Ile-25′, Gly-27′, Ile-29′, and Leu-31′ were found to be the most important to the enzyme activation, while the Arg-28′ was controversial [13,14]. As shown in Figure 1, the NS4A 21′-34′ peptide displays a brief bulge at Ile-25′ and Val-26′; otherwise, the peptide is an extended conformation. It forms main-chain hydrogen bonds with the A 0 and A 1 strands of the protease in an antiparallel fashion ( Figure 2). Another important conformational feature is that the kinky area of the bound NS4A 21′-34′ is almost planar, giving a good opportunity to design some cyclic aromatic structures that mimic this area and bind in the same manner, the NS3 protease. is conformation is conserved in all HCV NS3/4A protease crystal structures reported until now [9,10].
A few previous reports that investigated synthetic peptide variants of NS4A N-terminal provided strong grounds for leveraging this new target since some of these peptides could inhibit the function of NS3 protease in vitro [10]. Moreover, we succeeded in introducing 1H-imidazole-2,5-dicarboxamide derivatives as peptidomimetic that competed with NS4A and replaced it in binding assays with NS3 protease [9]. e imidazole, however, is a 5-membered monocyclic ring that mimicked the amide bond between Val-26′ and Gly-27′. Meanwhile, the flat kink of NS4A encompasses five bonds in this area (Figures 1 and 2 and Table 1).
To increase the binding potency of the heterocyclic nonpeptide mimics of NS4A, a bicyclic structure involving more bonds in this area was envisaged, because rigidification is an effective strategy to increase binding potency eliminating entropy factors of bond rotations. Figure 2(c) represents a promising bicyclic nucleus that established the same interactions of the mimicked planar region of the NS4A. e NH at the position of the THN nucleus could establish hydrogen bonds with the carbonyl of Ile-36. e carbonyl at position 4 of THN is ideal for replacing the carbonyl Val-26′. e isopropyl sidechain of Val-26′ was ignored for simplification  Table 1. purposes, because it was considered ineffective in the NS4A function [13]. e aromatic pyridine ring contains nitrogen, which acted as a hydrogen bond acceptor as a carbonyl replacement at groups R 1 aligning with Ile-25′ side chain; therefore, groups at this side chain are expected to be hydrophobic (Figure 1(b)). is pocket is composed of hydrophobic residues Trp-85, Pro-86, Ala-87, and Pro-88. It was in our focus as a good opportunity for obtaining tight binding peptidomimetics by making a library of compounds with R 2 � hydrophobic groups of different sizes and branching (Figure 2(c)).

Synthetic Plan of Target Compounds.
e target scaffold (THN derivatives, X � N) is an aza variant of 1,2,3,4-tetrahydroquinolines (X � CH), a nucleus commonly accessible via Povarov reaction (Scheme 1). e Povarov reaction (aka aza Diels-Alder or imino Diels-Alder) is a three-component reaction between an aromatic amine, an aldehyde, and an alkene in the presence of a Lewis or a Brønsted acid catalyst [15,16]. Povarov chemistry has been extensively utilized to construct several heterocyclic structures especially pyridinefused heterocycles [17]. Regardless of the fact that Povarov first reported the inverse electron demand 4 + 2 cycloaddition between a Schiff base (formed in situ) and electronrich alkene in the 1960s, the interest in this reaction increased vastly only in the past decade [18]. e most common and earliest heterocyclic system that has been efficiently prepared using the Povarov reaction is 1,2,3,4tetrahydroquinoline [19][20][21].
Planning for the synthesis of the final compounds, it was decided to start with suitably substituted arylamine 1 and react it with aldehyde 2 and the protected vinyl alcohol 3 to obtain the 1,7-naphtyridine 4. Having the core nucleus of the target compounds assembled, the crucial intermediate 5 (a deprotection product of 4) would be completed to the designed target compounds. Unfortunately, we encountered some difficulties in the early stages of the synthesis of the Povarov reaction step, as yields of Povarov products by using pyridyl amines were low. In some cases, the intended product was not observed in LC/ MS analysis. erefore, we thought of using aniline as an arylamine component that furnishes 1,2,3,4-tetrahydroquinoline (THQ) variants of the target scaffold (THN). THQ is a simpler structure, and it is more studied in previous reports [22]. Additionally, the THQ (X � CH) scaffold should be useful to investigate the utility of Povarov chemistry in our synthetic plans and study the structure-activity relationships. Results of the reactions of aniline with a variety of aldehydes and alkenes are shown in Scheme 2 and Table 2.

Limitations of the Alkene and the Aldehyde Components in the Povarov Reaction.
As illustrated in Scheme 2 and Table 2, the simplest arylamine, aniline (1a), was successfully reacted with benzaldehyde (2a) and acyclic vinylic ethers (3a-c) (Compounds 4a-c, . is success was expected, and it agrees with results previously reported elsewhere [22]. According to the design described above, the non-alkene portion acts as a temporary protective group (PG). It should be removed to obtain alcohol (intermediate 5) followed by oxidation to provide the ketone group in the target compound. Accordingly, we used the alkene component as vinyl ester, such as vinyl acetate or vinyl pivalate. We observed that the reaction stops at the stage of the Schiff base intermediate, as monitored by LC/MS. e failure of the Povarov reaction with vinyl esters confirms the ultra-sensitivity of this reaction to the electronic character of the alkene component. Our observation agrees with reports by Isambert and coworkers who reported that vinyl acetate does not advance the Povarov intermediate to cyclization via Friedel-Crafts mechanism as a termination step [23]. erefore, we decided to focus on the more suitable vinylic ethers. In particular, benzyl vinyl ether (BVE, 3d) has a removable group (considered as protective group, PG), but it was too expensive to consume in the chemistry development and optimization. Consequently, we started efforts to examine the utility of the Povarov chemistry in our synthesis using affordable alkenes such as ethyl vinyl ether (EVE, 3a), 3,4-dihydro-4H-pyran. (DHP, 2b), and 2,3-dihydrofuran (DHF, 2c). EVE (3a) was of special importance, because it is similar to BVE. e cyclic ethers DHF and DHP were used as positive references, because they are the most common alkenes used in the Povarov chemistry literature [19][20][21].
Ethyl glyoxylate (EtGlx, 2d) is an aldehyde with an ester tethering at C2 (R 3 ) furnished Povarov products 4j and 4k when reacting with the alkenes 3a,d (Entries 18-21). e use of phenylglyoxal (PhGlx, 2e) as the aldehyde component in this reaction with vinyl esters (as an alkene component) provided an extremely different reaction pathway that was described elsewhere [24]. However, PhGlx could react with electron-rich alkenes such as 3a-c as detected by LC/MS (Entries 22-28) to give the derivatives 4l-4m. Both 2-oxo aldehydes (EtGlx and PhGlx) are important for our targeted NS4A peptidomimetics as illustrated in Figures 2(b) and 2(c) for the design of target THN derivatives. We became particularly more optimistic when the Povarov reaction of aniline with EtGlx (2d) and BVE (3d) provided the MS peak of the product 4k, which featured a benzyl protecting group for the oxy group at position C4 of the THQ nucleus. In addition, 4k contains an ester group at position C2, which can be later manipulated into interesting screenable set of compounds for their inhibition of the target HCV-NS3. Accordingly, we decided to investigate the Povarov reaction with arylamine monomers (other than aniline 1a) that contain a protected amine substituent aiming to move forward with the synthetic plan illustrated in Scheme 1 (See Scheme 4 below). KSF) to try at this exploratory and chemistry development stage. CAN reaction setup is simple, does not require strict conditions (in terms of moisture and air control), and indeed its Povarov reported yields are high. e montmorillonite clays are green catalysts of natural origin that can be easily removed by filtration [25]. Results in Table 1 show that both catalysts provided similar yields depending on the reactants rather than the catalyst type.

Attempts to Broaden the Scope of the Povarov Reaction to
Synthesize Target Compounds. To move a step forward towards the synthesis of a screenable library of target compounds, we carried out Povarov reactions using mono-Boc-1,4-phenylenediamine (1b) and 5-amino-2-(N-Bocamino)-pyridine (1c) with a set of aldehydes (2a-c, 2f-g) along with the previously used activated alkenes 3a-d (Scheme 4, Table 3). e arylamines 1b-c carry a Boc-protected amine that could be later diversified by alkylation or acylation (R 1 -NH in Scheme 1).
To broaden the scope, we included a set of aldehydes encompassing benzaldehyde (2a) as a reference, two heteroaromatic aldehydes (2c, 2g), two aliphatic aldehydes (2b, 2f ), and a 2-oxoaldehyde (2d). e alkene component included the previously used monomers 3a-d. Unfortunately, aniline 1a was overall a better amine in the Povarov reaction than its variants 1b and 1c. e phenyl variant 1b provided the desired products 5a-c, albeit in low yields. e more important pyridylamine 1c could also show some positive results but failed to give the desired product MS peak in reactions with 2a/3b (Entries 35-38) or with 2c/3b (Entry 39), regardless of our attempts to push the reaction by increasing the amount of the catalyst (Entries 37-38). However, 1c could provide the required product MS peak with several combinations of reactants (Entries 40-53, Products 5f-m). It was noticed that CAN catalyst performed better than the M-KSF clay with this particular amine (2c). For instance, the product was detected in several cases when CAN was used as catalyst such as Entries 40, 48, 50, and 52 (Products 5g and 5k-m) but not M-KSF (Entries 42, 49, 51

R 3 O
Scheme 2: Povarov reactions using aniline (1a), aldehydes (2a-e), and alkenes (3a-f ) using a catalyst (Cat). R 1 is absent, because R 1 is assigned to arylamine monomers as mentioned in Scheme 1. In this scheme, only aniline was used as an arylamine.  Although there was a significant progress in the optimization efforts, and we wanted to move forward towards the target compounds described in Figure 2 and Scheme 1, we encountered a major problem in the process. We noticed that a big part of the products were lost in the purification step as described in Entries 29,40,43 and 50 (Table 3, Products 5a,f,g,l). Moreover, we noticed that many recovered pure products decomposed within few hours to form black tars. For instance, the product 5l (Entry 50) was collected in 96% overall purity (diastereomeric mixture of two peaks) (Figure 3(a)). We found that the purity of 5l decreases quickly within a few hours with concomitant formation of uncharacterized decomposition products (Figure 3(b)). e LC/MS peak of 5l completely disappeared after 24 hours of storage at room temperature (Figure 3(c)). It is worthy of hinting that storage under inert air and/or low temperature elongated the decomposition time, but the compounds quickly decompose before biological screening is completed or does another reaction towards the target compounds. erefore, among all compounds listed above, only four compounds were relatively stable upon storage (4c, 4f, 4m, and 5f ). Two of them made of the aliphatic 1-butanal (4f and 5f ), one from benzaldehyde (4c), and one from phenyl glyoxal (4m). One product was from the pyridyl amine building block (5f ), and three are from aniline (4c, 4f and 4m). e product 5a is one of the classic examples of Povarov chemistry that was used as positive reference in literature [26].
To conclude this chemistry investigation, our finding was that Povarov chemistry is a good tool in drug discovery that aniline or some substituted anilines are fine, but pyridine derivatives are not suitable. Aromatic aldehydes gave better yields than aliphatic aldehydes. e biggest downside of this chemistry is that either THQ or its aza analogue THN is not stable, and some have a very short shelf life. 4c, 4f, 4m, and 5f with HCV NS3 Enzyme. e protocol set previously by our research group to attest any compound has NS3 inhibition activities by binding to the NS4A binding site depended on first verifying that the compound has a significant affinity towards the NS3. If it passed this initial test, a competition assay with labeled NS4A would determine how potent the compound prevents the viral NS4A from binding to the NS3. A final in vitro test would investigate if compounds with confirmed high affinity inhibited NS3 by forming inactive complex [9,10]. Accordingly, the four compounds 4c, 4f, 4m, and 5f were tested for their binding affinity towards NS3 using the label-free Differential Scanning Light Scattering technique (DSLS).

Binding Assay of Compounds
is technique assesses the protein's thermal stability in terms of the formation of aggregates upon increasing the temperature gradually from 25°C to 90°C [27]. When the target protein binds a ligand, the aggregation temperature (T agg ) increases proportionally to the binding potency [28]. We performed this binding assay to measure the change in the aggregation temperature (ΔT agg ) upon mixing test compounds with NS3 and compare them  to the viral NS4A 21′-34′ (Table 4). Comparing the ΔT agg revealed that only compound 4m showed a weak affinity to the NS3 (22.6% of NS4A) (Figure 4). e other three compounds had nearly no affinity to the NS3. Compound 4f caused a drop in the thermal stability of the protein. It was not surprising that these compounds did not give any significant stability, because their structures had low similarity to the designed but failed compounds (Figure 2 and Scheme 1). e low affinity of the tested compounds was discouraging to pursue the biological screenings in the preset protocol. It is interesting to notice that the only compound that increased NS3 thermal stability was the 2-acyl analogue 4m (ΔT agg � 25% of the NS4A), indicating the significance of the oxo group to the binding.

Biological
Screening. All reagents used in the biological screenings were purchased from Sigma-Aldrich (UK) in molecular biology grade unless stated otherwise.

NS3 Protein.
(1) NS3 Constructs. A synthetic gene coding for the HCV NS3 domain of genotype 4a, the most abundant HCV in Saudi Arabia and Egypt [30], was synthesized by GenScript (Hong Kong), and the nucleotide sequence was optimized for E. coli codon usage. e synthetic gene was cloned as a NdeI-BamHI fragment into the expression vector pET-3a Novagen ® . e obtained construct was sequenced to confirm that we have the right clone and the gene is in the correct frame.
(4) Protein Purification. e produced protein was purified using equilibrated Ni-NTA beads, and the poly-histidine tag was not removed. In the process, cells were resuspended (1 g/ 5 mL) in buffer (50 mM HEPES, 0.3 M NaCl, 10% glycerol, 2 mM β-mercaptoethanol, pH8). Lysozymes were added (1 mg/mL) followed by a protease inhibitor cocktail tablet, and the suspension was sonicated. Cell lysate was centrifuged to collect the clear supernatant that contained the desired NS3 protein. e protein was purified using preequilibrated Ni-NTA beads (Qiagen, USA). Beads were washed with buffer (50 mM HEPES, 0.3 M NaCl, 10% glycerol, 2 mM β-mercaptoethanol, 20 mM imidazole, pH8) and eluted with another buffer (50 mM HEPES, 0.3 M NaCl, 10% glycerol, 2 mM β-mercaptoethanol, 350 mM imidazole, pH8). Fractions were collected and concentrated using Amicon Ultra-4 3000 MWCO centrifugal device (Millipore, Germany). Protein purity after the Ni-affinity purification step was not less than 70%. e purity, as estimated by SDS-PAGE, was sufficient to perform all investigations of this study, and the protein was stable for several hours at test conditions [30]. e concentration of NS3 in the final concentrate was measured using Nanodrop ™ nanoscale spectrophotometer. When needed, further purification of the protein was accomplished on Superdex 75 16/90 column (GE Healthcare, USA) equilibrated in 20 mM HEPES, 10 mM DDT, 200 mM NaCl, pH 7.6 run at a rate of 1 mL/min followed by SDS-PAGE for purity estimation.

NS4A.
e cofactor NS4A and the fluorescent fluorescein isothiocyanate NS4A (FITC-NS4A) were purchased from GenScript (Hong Kong). NS4A structure was identical to that of HCV genotype 4a, with two lysine residues added at both the N-and C-termini. us, the structure of NS4A used in this study was LL-G 21 SVVIVGRIVLSG 33 -LL.
We studied the binding of NS4A and its mutants with NS3 by DSLS using Stargazer-2 ™ (Harbinger Biotechnology and Engineering Corporation, Toronto, Canada). is method assesses protein stability by monitoring aggregate formation at controlled, gradually elevated temperatures [27]. NS3 domain stability upon binding to NS4A was measured by monitoring denatured protein aggregation upon increasing temperature from 25 to 85°C (0.5°C increments) at 600 nm.

DSLS Binding
Test. NS3 domain (15 µM) alone or mixed with the equimolar equivalent of tested MOC derivative was added to a binding buffer (20 mM HEPES, 10 mM DTT, 200 mM NaCl, pH 7.6) to a final volume 100 µL. e mixture was incubated at room temperature with gentle shaking for 2h. Afterward, 10 µL of the mixture was transferred into a clear bottomed Nunc 384-well plate and covered with 10 µL paraffin oil to minimize evaporation. Protein aggregation was monitored by tracking the change in scattered light that was detected by a Charged Coupled Device (CCD) camera. Snapshot images of the plate were taken every 0.5°C. e pixel intensities in a preselected region of each well were integrated using image analysis software to generate a value representative of the total amount of scattered light in that region. ese intensities were then plotted against temperature for each sample well and fitted to obtain the aggregation temperature (T agg ). Aggregation was monitored and analyzed to assess the effect of NS4A and its synthetic analogues on the stability of the NS3 as an indicator of binding. Each experiment was repeated 3 times. Statistical analysis was performed using GraphPad Prism v. 8.0 ® and Instat ® software. 2233RZ 120 Hz LCD Display ™ (3D ready) and Nvidia GeForce 3D Vision Glasses Kit ™ . e graphics of images were generated using PyMOL free software (https://pymol. org/2/).

Preparation of the Protein.
e 3-dimensional structure of NS3/4A protease [32] was downloaded from the Protein Data Bank (rcsb.org, Code: 1NS3), and its dimer structure was simplified to a monomer and optimized using Biopolymer > Prepare Structure tools. e Pep-15 was prepared by Biopolymer > Composition > Mutate Structure tools.

Data Availability
Spectra and biological data are available upon request by any party.

Conflicts of Interest
e authors declare no conflicts of interest.