Chemical Synthesis Enables Structural Reengineering of Aglaroxin C Leading to Inhibition Bias for HCV Infection

Abstract

As a unique rocaglate (flavagline) natural product, aglaroxin C displays intriguing biological activity by inhibiting HCV viral entry. To further elucidate structure-activity relationships and diversify the pyrimidinone scaffold, we report a concise synthesis of aglaroxin C utilizing a highly regioselective pyrimidinone condensation. We have prepared more than forty aglaroxin C analogues utilizing various amidine condensation partners. Through biological evaluation of analogues, we have discovered two lead compounds, CMLD012043 and CMLD012044, which show preferential bias for the inhibition of HCV viral entry vs. translation inhibition. Overall, the study demonstrates the power of chemical synthesis to produce natural product variants with both target inhibition bias and improved therapeutic indexes.

Graphical Abstract

INTRODUCTION

Rocaglates (flavagline) natural products were first isolated from the dried roots and stems of Aglaia elliptifolia Merr. (family Meliaceae) in 1982.¹ Since then, over thirty rocaglate natural products have been identified with unique structures and intriguing biological activities.^{2, 3, 4} For instance, silvestrol (1, Figure 1)⁵ was identified as an excellent translation inhibitor for cancer chemotherapy,⁶ whereas other related natural products and analogues (2-4) displayed similar activities.⁷ In our previous studies, translation inhibition was found to be associated with inhibition of the DEAD box RNA helicase eIF4A.⁸ Owing to the interesting structures and biological activities of rocaglates, a number of total syntheses⁹ and medicinal chemistry studies^{10, 11} have been disclosed. Our group has a longstanding interest in the synthesis of rocaglates and derivatives as well as investigations of their biology.^{6, 7, 8, 9g, i, k o–q, 10a, c, e–f} In 2015, we reported the enantioselective synthesis of aglaiastatin (5) and aglaroxin C (6) through biomimetic kinetic resolution.^9pSubsequent studies revealed that 5 was a promising translation inhibitor. In contrast, 6, containing a fully substituted pyrimidi-none core, showed only moderate translation inhibition.^9p In separate studies, we discovered that 6 inhibited hepatitis C viral (HCV) entry into host cells at a low μM concentration, potentially through inhibition of the prohibitins (PHBs) as viral entry factors.^{12, 13} Notably, prohibitins 1 and 2 have been reported as general viral entry factors for other viruses including dengue virus type 2 (DENV-2)¹⁴ and Chikungunya.¹⁵ We were further encouraged by the observation that 6 did not induce cytotoxicity by translation inhibition at a concentration that is effective for inhibition of viral entry, providing a promising therapeutic in dex (TI) for potential treatment of HCV infection.¹²

Figure 1.

Representative Biologically Active Rocaglate Natural Products and Synthetic Analogues

HCV is a widespread viral pathogen. The recent estimated number of viral carriers is 143 million worldwide,¹⁶ and over 2% of the population in North America has been infected. In addition, chronic HCV infection causes severe liver diseases in carriers, including liver cancer and cirrhosis.¹⁷ In 2015, over a half million people died from diseases due to HCV infection.¹⁸ Recently, direct-acting antiviral agents (DAAs) have become available to cure HCV infection¹⁹ by inhibiting the function of non-structural (NS) viral proteins. However, treatment effects of DAAs vary across different genotypes of HCV, and HCV potentially may develop resistance to these agents.²⁰ For example, the NS3 protease inhibitor simeprevir only treats genotypes 1 and 4 of HCV, and several signature resistance mutations have been identified against this treatment.²¹ Sofosbuvir, a top NS5B inhibitor, failed in HCV treatment which was associated with resistance mutations (Figure SI1).²² Accordingly, discovery of alternative treatments for HCV is sorely needed. As PHB-mediated cellular signaling pathways¹² are required by all HCV genotypes to infect host cells, it is conceivable that a small molecule such as aglaroxin C (6) may block infection from multiple HCV genotypes.²³ Additionally, targeting the host component for viral entry may create higher genetic barriers against resistance.²⁴ Lastly, aglaroxin C and analogues may also be valuable for dissecting the mechanisms of HCV entry and may inhibit other viruses sharing the same entry mechanism.

Our goal was to reengineer the structure of aglaroxin C (6) to increase activity against HCV entry while minimizing translation inhibition, which may lead to undesired cytotoxicity.^{5, 6, 7, 8} As the oxidation state change between 6 and aglaiastatin (5) led to different biological profiles, we speculated that the pyrimidinone subunit of 6 may be important for inhibition of HCV viral entry vs. translation inhibition. Our first-generation synthesis of aglaroxin C (6) utilized a key intermediate, keto-rocaglate 9,²⁵ which was synthesized through the excited state intramolecular proton transfer (ESIPT) [3+2] cycloaddition between 3-hy-droxyflavone 7 and methyl cinnamate followed by α-ketol shift of the aglain intermediate 8 (Scheme 1).⁹ⁱ However, the late stage pyrimidinone synthesis in our first-generation synthetic route is not ideal for rapid analog synthesis, affording 6 in low yield on a multi-milligram scale (Scheme 1).²⁶ We also envisioned that stepwise pyrimidinone synthesis may suffer from poor functional group tolerance in analogues due to the use of both acidic and oxidative conditions. Moreover, use of tethered aminoacetals such as 10 in the ester-amide exchange to produce intermediate 11 considerably narrows the diversity of accessible analogues due to limited availability of such building blocks. We therefore considered a streamlined synthesis of aglaroxin analogues through late stage, one-step pyrimidinone formation (Scheme 2).

Scheme 1.

First-Generation Synthesis of Aglaroxin C Stepwise Pyrimidinone Formation

Scheme 2.

Retrosynthetic Analysis for Aglaroxin C Analogues

RESULTS AND DISCUSSION

Development of a Direct Pyrimidinone Formation.

Our second-generation synthetic route (Scheme 2), which relies on condensation of keto-rocaglate (9) with commercially available amidines 13, was expected to flexibly provide analogues (12) for biological experiments. Pyrimidinones have served as biologically important substructures in both drugs and natural products.^{27, 28} There are many well-established and practical methods for the synthesis of pyrimidinones and pyrimidines, including the Traube synthesis of purines,²⁹ which inspired us to consider late stage construction of the pyrimidinone core of aglaroxin analogues.³⁰ However, we anticipated that condensation between the structurally complex substrate 9 and amidines 13 would be challenging. In particular, the highly ionizable tertiary, benzylic alcohol adjacent to the ketoester moiety may provide unexpected reactivity under the reaction conditions.

Our initial model study for pyrimidinone formation employed benzamidine as a simplified condensation partner. Microwave conditions (Table 1, entry 1) produced minimal amounts of pyrimidinone 12a in an unsatisfactory yield of 28% (determined by ¹H NMR analysis) due to low conversion.^9p Thermal conditions (entry 2) resulted in full consumption of 9, while cyclopentapyrimidinedione 16 and 3,5-dimethox-yphloroglucinol were found as the major identified side products. The formation of fragmentation product 16 was found to correlate with the amount of 4-dimethylaminopyridine (DMAP) employed; reduction to 0.3 equivalents of DMAP minimized production of 16 (entry 3) but also led to formation of an unexpected cyclization product, oxazoline 15. The structure of 15 was confirmed by single crystal X-ray crystal structure analysis.²⁶ Reduced equivalents of the benzamidine reaction partner resulted in an incomplete reaction, with increased production of decarboxylation product 14 and retro-Nazarov products 17 (entries 4 & 5).²⁶ We next found that diluting the reaction eightfold improved the yield of 12a (entry 6). Increasing both temperature and reaction time under dilute conditions eliminated the production of 15, but also led to a lower yield of 12a in favor of significant amounts of 16 (entry 7). Gratifyingly, by reducing the reaction time to 45 minutes, we obtained a 90% NMR yield of 12a with minimal fragmentation to 16 (entry 8); ultimately, an 81% isolated yield of 12a was obtained on a 100-mg scale. Interestingly, we found that the model reaction gave the almost identical yield in absence of DMAP under the optimized conditions (entry 9), but we acheived better reproducibility when amidine hydrochloride salt was used with DMAP (vide infra).

Table 1.

Optimization of the Direct Pyrimidinone Formation

^aMicrowave, toluene as solvent;

^b10% of 9 recovered;

^c20% of 9 recovered, 24% yield of 17;

^d81% isolated yield of 12a;

^e78% isolated yield of 12a.

Proposed Mechanistic Pathway.

According to trends observed during reaction optimization, we propose the mechanistic pathway depicted in Scheme 3. We postulate that water generated from the pyrimidinone condensation may hydrolyze 9 to the β-keto-acid intermediate 21, which undergoes decarboxylation to ketone 14. When no nucleophile was used, 14 and retro-Nazarov product 17 were obtained. Presumably DMAP may promote intramolecular proton transfer to 19 followed by the extrusion of water to afford the retro-Nazarov precursor 20, whereas the formed water triggered decarboxylation of 9 (an alternative mechanism is shown in the SI). In the presence of the amidine, the desired formation of hemiaminals 22 and epi- 22appears to compete with decarboxylation and retro-Nazarov reaction, which can be enhanced through use of increased equivalents of the amidine. Based on the Katrizky mechanism determined for the Traube reaction of β-keto esters and amidines,³¹ two possible pathways may be envisioned for the formation of pyrimidinone 12a. In one mechanism, hemiaminal epi-22, generated from disfavored concave-addition of the amidine to 9,⁹ⁱ may cyclize to dihydropyrimidinone 25 followed by extrusion of water. In contrast, hemiaminal 22, obtained from amidine addition to the convex face of 9, has an anti-relationship between the aminal and ester thereby preventing direct cyclization. Instead, loss of water would produce the observed enamine 23, which we propose then cyclizes to 12a through extrusion of methanol generating imidoyl ketene 24 followed by a 6π-electrocyclization to pyrimidinone 12a.³² To support the formation of hemiaminal 22, the oxazoline byproduct 15 was formed through cyclization of the tertiary alcohol of 22 to the amidine carbon followed by extrusion of ammonia (blue arrow). The formation of 22 is also supported by our isolation of enamine 23 from a 250 mg scale reaction, where a 28% yield of 23 was found to precipitate after 10 minutes; we subsequently found that 23 can also be synthesized in 60% yield using 9 and benzamidine at 60 °C in toluene.²⁶ Interestingly, isolated 23 was found to solely produce 12a with no observed formation of 15 when 23 was resubjected to the reaction conditions. We rationalize that the sp³-hybridized hemiaminal carbon of 22 allows for a conformation necessary for intramolecular cyclization, whereas the sp²-hybridized enamine carbon of 23 prevents a similar cyclization. This conforms to the observation that elevated temperatures facilitate extrusion of water generating enamine 23 and minimize formation of byproduct 15. In contrast, 15 did not generate 12a under the reaction conditions. In a control experiment, DMAP was found to accelerate the fragmentation of 12a into 16 and 3,5-dimethoxyphloroglucinol.²⁶

Scheme 3.

Possible Mechanistic Pathways and products of Direct Pyrimidinone Formation.

Second-Generation Synthesis of Aglaroxin C.

After determining optimal conditions for direct pyrimidinone formation, we next sought to apply these conditions to synthesize aglaroxin C (6). In this case, the requisite pyrrolidin-2-imine was commercially available as an HCl salt (Table 2). Initial attempts using stepwise free-basing protocols failed due to the high-water solubility of the amidine; accordingly, we focused on in-situ free-basing conditions for optimization studies. Using excess base (entries 1 and 2), no conversion of substrate 9 was observed after refluxing for 12 h. As 9 exists as mixture of enol/keto isomers, presumably excess base favors formation of an unreactive enolate. In contrast, use of excess pyrrolidin-2-imine HCl salt (entry 3) did not induce pyrimidinone formation. We assume that DMAP is protonated by the amidine salt which negated its function. When a slight excess of amidine vs. Na-OMe (entry 4) was used, we obtained 6 in 19% isolated yield. Consistent with our earlier optimization efforts, use of increased concentration and 12 h reaction time resulted in a complex reaction mixture containing fragmentation products (entry 5) Finally, use of 3.0 equiv. of amidine salt and 2.95 equiv. of NaOMe along with a catalytic amount of DMAP (40 mol%, 130 °C) afforded 6 in 76% isolated yield on a 100-mg scale (entry 6).

Table 2.

Optimization of Second-Generation Synthesis of Aglaroxin C

^a0.025 M concentration, reflux 12 h;

^bno conversion;

^c0.2 M concentration, reflux 12 h;

^dcomplex reaction; only isolated the fragmentation products;

^e0.025 M concentration, 130°C 1 h;

^f76% isolated yield for aglaroxin C; 9% decarboxylated product also obtained.

Synthesis of Aglaroxin Analogues.

With reliable annulation protocols available using both amidines and amidine salts, we next synthesized a library of aglaroxin analogues (Table 3). In general, direct pyrimidinone formation was found to tolerate both aromatic and aliphatic amidines. We observed consistent yields for formation of C-substituted pyrimidinones (12b-12f). For instance, simple aliphatic amidines reacted with 9 to afford C-Me (12b) and C-Bn (12c) pyrimidinones. Hindered amidines such as iso-butanamidine and cyclohexylcarboxamidine afforded the desired products 12d and 12e in excellent yields. Interestingly, condensation of carbomethoxyformamidine with 9 was followed by an unexpected decarboxylation to produce the non-substituted product 12f in 53% yield. Moreover, we discovered that the pyrimidinone condensation could tolerate various functional groups, including ester 12g (41%), cyclopropane 12h (91%), and methoxyl methylene 12i (52%). Ester exchange was observed when ethyl amidinoacetate was employed, producing compound 12g. Additionally, guanidine-type reaction partners successfully underwent condensation, yielding products such as 12j (90% yield).

Table 3.

Synthese of Aglaroxin C Analogues Using Direct Pyrimidinone Formation^a

^aStandard conditions: 25 mg of 9, 3 equiv amidine, 0.3 equiv DMAP, m-xlyene (0.025 M), 130°C, 45 min; 2.95 equiv NaOme (1.0 M in MeOH) was used for amidine;

^b250 mg of 9 was used;

^cethyl 2-amidinoacetate HCI Salt was used, and 8% of the ethyl ester products was isolated;

^dm-xlyene (0.2 M)

Preliminary biological studies of the analog set indicated that C-aryl pyrimidinones consistently possessed good inhibition of HCV infection with low cytotoxicity (vide infra).²⁶ Accordingly, we synthesized additional C-aryl pyrimidinone analogues with a variety of substitution patterns, with an initial focus on para-substituted aryl benzamidines (Table 3, 12k-r). Generally, condensations tolerated both electron-rich and deficient benzamidines; however, we found that greater amounts of fragmentation products were generated using electron-deficient amidines (cf. compounds 12k-n 66–83% yield) with compound 12o, which was generated in 53% yield along with the corresponding fragmentation products. Compound 12m was further subjected to hydrogenolysis to afford the free phenol which may serve as a potential handle for further modifications.²⁶ As expected, analogues including para-halides (12p-12r) were synthesized in 70–80% yields. Such compounds may also allow for late stage functionalization via aminations and S_NAr reactions. We found that meta-substituted benzamidines were also workable, affording products (12s-12u) using the standard protocol in reasonable yields (71%−73%). Using the sterically hindered ortho-substituted benzamidines, 12v and 12w were synthesized in moderate yields (54% and 66%). Six C-heteroaryl substituted analogues (12x−12ac) were also synthesized; in these cases, we observed substantial fragmentation of the desired products. Only 12x and 12z were obtained in reasonable yields (92% and 77%, respectively), while compounds 12y and 12aa-12ac were obtained in yields averaging 50%.

As we found that aglaroxin C (6, Table 2) was synthesized regioselectively using an unsymmetrical amidine, we also tested additional unsymmetrical amidines in the pyrimidinone formation. Among all products formed, we found that the N- substituent was situated exclusively on the nitrogen adjacent to the pyrimidinone carbonyl (12ad-af). During the synthesis of 12ae, a small amount of oxazoline 15 was obtained as the only observable side product. This result supports our mechanistic proposal (cf. Scheme 3) wherein the less sterically hindered, unsubstituted nitrogen likely engages in initial hemiaminal formation with 9. Along these lines, we also synthesized 12ac, a ring-expanded analog of 6, in 69% yield. Unlike 12ad, adducts 12ae and 12af were synthesized in 24 and 33% yields, respectively; these reactions generated significant amounts of retro-Nazarov and decarboxylated products, suggestive of lower overall reactivities among the N-substituted amidines.²⁶ We also optimized the yield of 12ae to 39% by increasing the reaction concentration to 0.2 M. Finally, using exo-keto-rocaglate as the starting material, we synthesized compounds 12ah and 12ag, the C-3 epimers of 12a and 6, in 77 and 52% yields, respectively.

As a comparison to the direct condensation with unsymmet-rical amidines, we also attempted alkylations of 12a to produce compounds 12ai-aj (Table 4) 12al-an (see Supporting Information).²⁶ Unfortunately, all attempted alkylations displayed poor chemo-and regioselectivity, favoring the undesired O-al-kylation products such as O-methoxypyrimidine 12ai.

Table 4.

Structure-Activity Relationship of Racemic Aglaroxin C Analogues in the HCV infection inhibition and cytotoxicity Assays^a

^aAbsolute EC₅₀ and CC₅₀ were determined from fitted sigmoid curves;

^bAbsolute EC₅₀ indicates the concentration of compound that inhibits HCV infection by 50%. For its determination, HCVcc-Luc was added to Huh7.5.1 cells at 37 °C together with serially diluted compounds for 3 h prior to removal. Infected cells were incubated at 37 °C for an additional 48 h prior to luciferase assay (mean of n = 3; error bars, s.d.);

^cAbsolute CC₅₀ indicates the concentration of compound required to effect 50% cell death. The numbers of viable cells in culture were determined using the CellTiter-Glo Cell Viability Luminescent Assay kit according to the manufacturer’s (Promega) instruction (mean of n = 3; error bars, s.d.);

^dAUC_0.2–200 (cytotoxicity) is the integration of area under the cytotoxicity curve from 0.2 to 200 μM concentration;

^eAUC_0.2–20 (EC/CC) is the ratio (fold-difference) of the integraded area under curve (AUC) values from 0.2 to 20 μM concentration for the effective curve vs. the cytotoxicity curve;

^fn d = not determined; no detectable inhibition or cytotoxicity was observed;

^gn.d. = CC₅₀ was not determined; some cytotoxicity was observed, but cytotoxicity curve did not reach 50% inhibition.

BIOLOGICAL STUDIES

Structure-Activity Relationships.

We next evaluated the library of (±)−aglaroxin analogues and side products against HCV infection, and also tested their corresponding cytotoxicity in human hepatic (Huh) 7.5.1 cells.²⁶ In comparison to aglaroxin C (6), C-alkyl substituted pyrimidinones 12b-d and 12h (Table 3) exhibited increased inhibition against HCV infection, whereas 12e-g had decreased activities. Nevertheless, the C-aryl substituted products (12a, k-w) showed a promising increase of inhibition of HCV infection with similar or reduced cytotoxicity in comparison to 6. Excitingly, 12l (CMLD012043) and 12s (CMLD012044) showed a three-fold greater inhibition of HCV infection in comparison to 6, while 12l and 12s exhibited relatively low cytotoxicity to Huh-7.5.1 cells (less than 50% cell death at 200 μM). Thus, the two lead compounds (12l and 12s) provided excellent selectivity indexes (SI) in inhibiting HCV infection (vide infra). Notably, among all the C-heteroaryl substituted aglaroxin analogues (12x−12ac), only 12y and 12ab displayed slightly increased inhibition of HCV infection relative to 6. The six-membered ring analog 12ad and guanidine-type adducts 12j and 12af were found to have moderate antiviral activity.

To further understand structure-activity relationships (SAR) among aglaroxin analogues, we highlight ten compounds in Table 4 along with dose-response curves depicting their effectiveness in both HCV infection and cytotoxicity assays. As a benchmark compound, 6 was found to inhibit HCV infection with an EC₅₀ of 1.3 μM and showed a CC₅₀ of 12 μM. Of note, EC₅₀ and CC₅₀ values were calculated according to the fitted sigmoid curves, which are reported as absolute values (indicating the concentration of compounds providing 50% inhibition and 50% cell death, respectively). Next, we compared the EC₅₀ of cyclo-hexyl- (12e) and phenyl-substituted (12a) pyrimidinones and found that C-aryl substitutions led to improved potency against viral infection (420 nM vs. 2.5 μM). The C-3 epimer (12ah) of 12a was found to be inactive against HCV infection and no cytotoxicity. It is apparent that a syn-relationship of the two aryl rings on the cyclopenta[b]benzofuran core is crucial for HCV inhibition.

Based on the observation that N-methylated isomer 12ae was one-fold more active than 12a (EC₅₀ 0.2 μM vs. 0.42 μM) with similar cytotoxicities (CC₅₀ 7.3 μM vs. 8.1 μM),³³ we next utilized methylation to evaluate other sites for introduction of target identification tags. In contrast to N-methylation, we observed a decrease in inhibition of HCV infection (EC₅₀ = 9.2 μM) with the O-methylated pyrimidine 12ai. The doubly methylated product 12aj was also found inactive and non-toxic. Finally, we studied the impact of substituting the conjugated C-aryl group of the pyrimidinone (cf. 12l, 12o, and 12s). Compound 12o, with an electron-withdrawing CF₃-group, had higher cytotoxicity (CC₅₀ = 10 μM), and only a one-fold better EC₅₀ (0.24 μM) for HCV inhibition than 12l and 12s (EC₅₀ ~ 0.5 μM). The two lead compounds 12l and 12s displayed improved EC₅₀s in the viral infection assay but displayed low cytotoxicity; neither compound reached 50% cell death up to 200 M concentration (maximum cytotoxicity plateau: 40% and 35% cell death for 12l and 12s, respectively). Of note, we observed a similar plateauing of cytotoxicity using aglaroxin C (6) and its analogues (12e, 12a, and 12ae). In contrast, the 4’-trifluoro-methylphenyl analog 12o achieved close to 100% cell death. For a more reliable comparison of cytotoxicities among these compounds, we also calculated the area under the cytotoxicity curve (AUC) by integration,³⁴ an alternative method for accurately quantifying low cytotoxicity.³⁵ In this manner, AUC_0.2–200 analysis indicates that 12e, 12a, and 12ai share similar cytotoxicities to 6, whereas 12l and 12s exhibit relatively lower cytotoxicities. Extending this analysis to compare the SI among the analogues, we next calculated the AUC_0.2–20(EC/CC) ratios, determining that 12ae, 12l, 12o, and 12s had wider therapeutic windows (0.6 to greater than 2-fold increase in the SI) than 6. Based on the performance of 12l and 12s using these metrics, we utilized these two lead compounds for further biological investigations including mechanism studies.

To further characterize the mode of action of the two lead compounds 12l and 12s, we independently synthesized each of their respective enantiomers and investigated their biological activities in both the HCV viral infection and cytotoxicity assays (Table 5). As expected, only (+)−12l and (+)−12s displayed HCV inhibition, whereas the other enantiomers (−)−12l, and (−)−12s were found to be inactive and non-toxic.²⁶ Interestingly, we noticed that maximum cytotoxicity plateau of both active enantiomer (+)−12l and (+)−12s increased substantially compared to their racemic counterparts (±)−12l and (±)−12s. To verify this result, a secondary MTS cell viability assay was also performed on racemic compounds (±)−12k, (±)−12l, and (±)−12o, and showed nearly identical toxicity curves as were observed in the CellTiter-Glo assay.²⁶ While these results warrant further investigation, they suggest a potential rationale for the observed SI enhancement in which the “inactive” enantiomers may reduce or otherwise mitigate the cytotoxic effects of the active species through an as-yet undefined mechanism.

Table 5.

Chirality-Based Biological Profiles of Lead Compounds^a

compound	(±)−12l	(+)−12l	(−)−12l	(±)−12s	(+)−12s	(−)−12s
EC50b (μM)	0.5	0.3	n.d.e	0.5	0.5	n.d.e
CC50c (μM)	n.d.d	10	n.d.e	n.d.d	7.0	n.d.e

^aEC₅₀ and CC₅₀ were collected using the same assay in Table 4 and were calculated based on the fitted sigmoid curves;

^bEC₅₀ indicates the concentration of a compound that inhibits 50% HCV infection;

^cCC₅₀ indicates the concentration of a compound leads 50% cell death;

^dn.d. = CC₅₀ was not determined; some cytotoxicity was observed, but cytotoxicity curve did not reach 50% inhibition;

^en.d. = not determined; no detectable inhibition or cytotoxicity was observed.

Aglaroxin Analogues Do Not Affect HCV RNA Replication and Translation.

We subsequently evaluated the ability of the aglaroxin C analogues to inhibit HCV RNA replication and mRNA translation using an HCV replicon system which harbors a full-length HCV Genotype 1b RNA genome.³⁶ In replicon cells, HCV RNA replicates and is translated into viral proteins without forming infectious viruses. Hence, this system permits accurate assessment of any effects on viral RNA replication and translation. As shown in Figure 2, compounds (+)−12l and (+)−12s when used at 2 μM for 3 h did not inhibit HCV RNA replication or protein synthesis. This data implies that the observed inhibitory effects of (+)−12l and (+)−12s were unlikely due to inhibition of viral RNA replication or mRNA translation. To corroborate these findings, we assembled an in vitro translation inhibition assay where a bicistronic reporter mRNA was programmed for translation in Krebs-2 extracts (Figure 3).³⁷ In this system, translation of the Firefly luciferase (FLuc) translation is cap-dependent whereas Renilla luciferase (RLuc) translation is dependent upon the HCV internal ribosome entry site (IRES) (Fig. 3A). It was observed that translation inhibition of firefly luciferase by 6 and analogues 121 and 12s were minimal in comparison with CR-1–31-B (Figure 1, 4), a highly potent rocaglate translation inhibitor showing strong inhibition of cap-dependent protein synthesis (Figure 3B).^10c This data suggests that HCV mRNA translation is not inhibited by 6 and its analogues 12l and 12s.

Figure 2.

Aglaroxin C (6) and analogues (121 and 12s) neither inhibit HCV RNA replication nor protein translation. HCV replicon cell line 2-3+ was treated with indicated compounds for 3 h. Cells were incubated for an additional 48 (A & C) or 72 h (B & D). Cellular RNA or protein was isolated for quantifications of HCV viral RNA by real-time PCR (A & B) or non-structural protein NS5 by western blotting (C & D).

Figure 3.

Aglaroxin-type analogues displayed minimal translation inhibition in the comparison to (−)−4 (CR-1–31-B), (A) Schematic representation of the bicistronic reporter FF/HCV/Ren mRNA used to monitor translation. In this system, Firefly luciferase (FLuc) translation is cap-dependent whereas Renilla luciferase (RLuc) expression is HCV-IRES dependent. (B) Assessment of cap or HCV dependent translation of FF/HCV/Ren mRNA in the presence of 1 μM of the indicated compounds in Krebs-2 extracts. Results are presented relative to values obtained in the presence of DMSO and expressed as mean ± error of 2 biological replicates.

Aglaroxin Analogues Inhibit HCV Entry.

As aglaroxin C was previously reported to inhibit HCV entry,¹² we carried out two sets of experiments to test whether compounds 12l and 12s also inhibit viral entry. Firstly, we generated infectious lentiviral pseudotypes bearing glycoproteins of HCV (HCVpp), Chikungunya virus (CHIKVpp), Ebola virus (Ebolapp), and vesicular stomatitis virus (VSVpp). Compositionally, these pseudotyped viruses differ only in viral envelopes because they are packaged using the same lentiviral reporter construct with a separate construct expressing specified viral envelope protein. Hence, the only difference among these pseudoviral particles is the mode of entry, which is dictated by the particular viral glycoprotein found on the viral envelopes. Compounds 12l and 12s, as both the racemates and single (+)−enantiomers, specifically inhibited HCVpp and CHIKVpp but not Ebolapp and VSVpp (Figure 4, A and B). Interestingly, like rocaglamide, compounds (+)−12l and (+)−12s as well as their racemic versions also inhibited Dengue virus infection (Figure 4C). It is worth mentioning that HCV, Dengue virus, and CHIKV all utilize PHBs to enter cells; further studies are warranted to determine whether compounds 12l and 12s directly target PHBs. Secondly, to confirm inhibition of HCV entry, we performed time-of-addition experiments using the lead compounds 12l and 12s wherein compounds were added at different times relative to when the virus was added to cells. Similar to 6, 12l and 12s displayed maximal anti-HCV activity when added together with the virus, but partially lost their activity when added 3 h after infection was initiated (Figure 4D). This finding confirmed that 12l and 12s are preferentially inhibiting viral entry.

Figure 4.

Aglaroxin C (6) and its analogues (121 & 12s) inhibit viral entry. (A) Huh7.5.1 cells were infected by indicated pseudovirus in the presence of compounds (1 μM) for 3 h. After an additional 48 h infection, cells were lysed for luciferase assay. Relative infectivity was calculated by normalizing against the values obtained from cells treated with DMSO (arbitrarily set to 1.0) (mean ± SD, *p < 0.05). (B) Same as in a except (±)−121 and (±)−12s were added at 2 μM. (C) Huh7.5.1 cells were infected by Dengue virus (serotype 2, Thailand 16681 strain, MOI 0.1) treated with 2 μM compounds for 3 h. 48 h post-infection, RNA was isolated for quantitative RT-PCR analysis. Viral RNA levels were normalized against GAPDH levels. Data presented are representatives of two independent experiments. (D) (±)−6, (±)−12l, and (±)−12s inhibited HCV infection when added together with the virus. Details on this time-of-addition experiment can be found in Supporting Information.

CONCLUSION

In summary, we have developed a second-generation synthesis of aglaroxin C using late-stage, direct pyrimidinone formation of a keto-rocaglate scaffold. Using this method, we have used commercially available amidines as reaction partners to synthesize a library of over forty aglaroxin C analogues. Among newly synthesized analogues, we successfully demonstrated SAR for inhibition of HCV infection and identified two aryl pyrimidinone lead compounds, 12l and 12s, which have low cytotoxicities to Huh 7.5.1 cells. Additional biological studies with 12l and 12s indicate that the mechanism of inhibition of HCV infection is through inhibition of HCV viral entry, rather than by blocking viral replication and translation. Finally, 12l and 12s are also effective against infection of other viruses including Dengue and Chikungunya, both of which have been found to use prohibitins (PHBs) as an entry factor. These studies illustrate the power of chemical synthesis to bias inhibition of HCV viral entry vs. translation inhibition and improve properties including therapeutic index. Further studies toward target identification of 12l, 12s, and related compounds is currently in progress and will be reported in due course.