Optimization of 4-Anilinoquinolines as Dengue Virus Inhibitors

Emerging viral infections, including those caused by dengue virus (DENV) and Venezuelan Equine Encephalitis virus (VEEV), pose a significant global health challenge. Here, we report the preparation and screening of a series of 4-anilinoquinoline libraries targeting DENV and VEEV. This effort generated a series of lead compounds, each occupying a distinct chemical space, including 3-((6-bromoquinolin-4-yl)amino)phenol (12), 6-bromo-N-(5-fluoro-1H-indazol-6-yl)quinolin-4-amine (50) and 6-((6-bromoquinolin-4-yl)amino)isoindolin-1-one (52), with EC50 values of 0.63–0.69 µM for DENV infection. These compound libraries demonstrated very limited toxicity with CC50 values greater than 10 µM in almost all cases. Additionally, the lead compounds were screened for activity against VEEV and demonstrated activity in the low single-digit micromolar range, with 50 and 52 demonstrating EC50s of 2.3 µM and 3.6 µM, respectively. The promising results presented here highlight the potential to further refine this series in order to develop a clinical compound against DENV, VEEV, and potentially other emerging viral threats.


Introduction
Mosquito-borne viral infections, including those caused by the flavivirus dengue (DENV) and the alphavirus Venezuelan Equine Encephalitis virus (VEEV), represent a major public health concern [1][2][3][4]. Rapid urbanization and climate change have contributed to the expanding geographical range of DENV infections into the developed world [5] and have caused a significant increase in the number of annual infections, currently estimated at~400 million people in over 128 endemic countries [6,7]. The majority of symptomatic DENV infected patients experience mild illness, yet approximately 5-20%, particularly those with a secondary infection with a heterologous DENV serotype, progress into a potentially life-threatening disease known as severe dengue [8,9]. Currently, there are no approved antiviral therapies for DENV infection, and the development of an effective and safe DENV vaccine has been challenged by the need to generate a balanced protective immunity against the four distinct DENV serotypes.
The alphavirus VEEV is an important causative agent of neurological disease in Central and South America [10]. While most VEEV infections in humans are mild, approximately 14% of the patients develop encephalitis, often complicated by severe neurological deficits and sometimes death [11]. Beyond mosquito bites, VEEV can be transmitted deficits and sometimes death [11]. Beyond mosquito bites, VEEV can be transmitted via aerosol exposure and is considered a major bioterrorism threat, underscoring the importance of developing effective countermeasures [12]. However, there are currently no approved antiviral drugs nor licensed human vaccines available for VEEV infection. In the absence of approved antiviral therapies for DENV and VEEV infections, the management of infected patients largely relies on supportive care [12,13], resulting in continued morbidity and mortality.

Synthesis
Building on our previous work, to further explore the antiviral activity of the 4-anilinoquin(az)oline [14,32,33], we probed the structure activity relationships (SAR) with a series of focused libraries. We developed a series of hybrid molecules combining structural features of erlotinib and lead compound 4 to enhance the SAR in the current literature and improve the tractability of the quin(az)oline scaffold ( Figure 2).

Synthesis
Building on our previous work, to further explore the antiviral activity of the 4anilinoquin(az)oline [14,32,33], we probed the structure activity relationships (SAR) with a series of focused libraries. We developed a series of hybrid molecules combining structural features of erlotinib and lead compound 4 to enhance the SAR in the current literature and improve the tractability of the quin(az)oline scaffold ( Figure 2). deficits and sometimes death [11]. Beyond mosquito bites, VEEV can be transmitted via aerosol exposure and is considered a major bioterrorism threat, underscoring the importance of developing effective countermeasures [12]. However, there are currently no approved antiviral drugs nor licensed human vaccines available for VEEV infection. In the absence of approved antiviral therapies for DENV and VEEV infections, the management of infected patients largely relies on supportive care [12,13], resulting in continued morbidity and mortality.

Antiviral Screening
To evaluate the compounds for broad-spectrum antiviral activity we first studied their effect on DENV2-Rluc virus infection, a wild type virus carrying a Renilla luciferase reporter gene (virus production described in Materials and Methods). We measured the effect of compound treatment on overall infection in human hepatoma (Huh7) cells 48 hours following infection with DENV2 via luciferase assays and calculated the half-maximal effective concentration (EC50) relative to DMSO treated cells. In parallel, we tested the effect of these compounds on cell viability via an AlamarBlue assay in the DENVinfected Huh7 cells and measured the half-maximal cytotoxic concentration (CC50) values relative to DMSO treated cells.
In order to follow-up on the lead compound 4, we first screened a series of focused substitutions at the aniline meta-position (5-10) ( Table 1) [48]. The hydroxy analogue 5 was 3-fold less potent than the methoxy 4. The nitro substitution 6 was inactive at the top concentration, while the reduced amine 7, was 7-fold weaker with respect to 4 and 2-fold weaker with respect to the direct isosteric replacement 5. The monomethyl substituted amine 8 showed no improvement, while the dimethyl substitution showed a 2-fold increase in potency against DENV with respect to 7. The methanolic substitution 10, was 2fold weaker with respect to 5 and 8-fold with respect 4.

Antiviral Screening
To evaluate the compounds for broad-spectrum antiviral activity we first studied their effect on DENV2-Rluc virus infection, a wild type virus carrying a Renilla luciferase reporter gene (virus production described in Materials and Methods). We measured the effect of compound treatment on overall infection in human hepatoma (Huh7) cells 48 h following infection with DENV2 via luciferase assays and calculated the half-maximal effective concentration (EC 50 ) relative to DMSO treated cells. In parallel, we tested the effect of these compounds on cell viability via an AlamarBlue assay in the DENV-infected Huh7 cells and measured the half-maximal cytotoxic concentration (CC 50 ) values relative to DMSO treated cells.
In order to follow-up on the lead compound 4, we first screened a series of focused substitutions at the aniline meta-position (5-10) ( Table 1) [48]. The hydroxy analogue 5 was 3-fold less potent than the methoxy 4. The nitro substitution 6 was inactive at the top concentration, while the reduced amine 7, was 7-fold weaker with respect to 4 and 2-fold weaker with respect to the direct isosteric replacement 5. The monomethyl substituted amine 8 showed no improvement, while the dimethyl substitution showed a 2-fold increase in potency against DENV with respect to 7. The methanolic substitution 10, was 2-fold weaker with respect to 5 and 8-fold with respect 4.
To evaluate the compounds for broad-spectrum antiviral activity we first studied their effect on DENV2-Rluc virus infection, a wild type virus carrying a Renilla luciferase reporter gene (virus production described in Materials and Methods). We measured the effect of compound treatment on overall infection in human hepatoma (Huh7) cells 48 hours following infection with DENV2 via luciferase assays and calculated the half-maximal effective concentration (EC50) relative to DMSO treated cells. In parallel, we tested the effect of these compounds on cell viability via an AlamarBlue assay in the DENVinfected Huh7 cells and measured the half-maximal cytotoxic concentration (CC50) values relative to DMSO treated cells.
In order to follow-up on the lead compound 4, we first screened a series of focused substitutions at the aniline meta-position (5-10) ( Table 1) [48]. The hydroxy analogue 5 was 3-fold less potent than the methoxy 4. The nitro substitution 6 was inactive at the top concentration, while the reduced amine 7, was 7-fold weaker with respect to 4 and 2-fold weaker with respect to the direct isosteric replacement 5. The monomethyl substituted amine 8 showed no improvement, while the dimethyl substitution showed a 2-fold increase in potency against DENV with respect to 7. The methanolic substitution 10, was 2fold weaker with respect to 5 and 8-fold with respect 4.  We then screened a series of matched pair meta-methoxy and meta-hydroxy analogues (11)(12)(13)(14)(15)(16)(17)(18), changing the 6-position of the quinoline ring (Table 2) [48]. The 6-bromoquinoline methoxy analogue 11 was equipotent with the 6-triflurormethyl quinoline methoxy analogue 4. However, switching to the hydroxy 12 led to a 4-fold increase in potency against DENV (EC50 = 0.63 µM). The unsubstituted quinoline methoxy analogue 13 showed a 3-fold decrease in activity against DENV, while the hydroxy analogue 14 showed no activity at the top concentration tested. Switching to the electron donating 6methoxyquinoline methoxy analogue 15 was 5-fold less potent with respect to 4, while the hydroxy analogue was inactive (EC50 = >10 µM). Interestingly, both the 6-methylsulfone quinoline analogues 17 and 18 were inactive. Encouraged by the result of 3-((6-bromoquinolin-4-yl)amino)phenol 12, we followed up with a series of focused analogues (19-31) fixing the 6-bromoquinoline and altering the aniline ring substituent (Table 3) [48]. The direct nitro analogue 19 and reduced amine version 20 were both inactive (EC50 = > 10 µM). The substitution of the amine 20 with a  this position [49]. The tert-butoxy 26 2-fold less potent than 12, with the tert-butyl 27 near equipotent (EC50 = 0.94 µM), unfortunately both showed a level of toxicity (CC50 = 9.2 µM and CC50 = 6.0 µM respectively). We then explored how torsional strain on the methoxy orientation would affect activity (28)(29)(30)(31). The fused ring systems were overall less active against DENV, only 28 showed activity (EC50 = 1.7 µM) which was accompanied by light toxicity (CC50 = 8.6 µM). Predicting that we would identify active analogues based on the 2,3-and 3,4-connectivty, we followed up on the results of 28-31 with a series of different substitutions and fused heterocycles   (Table 4) [42]. The introduction of a fluorine to the distal orthoposition on the aniline ring of 30 produced compound 32 with good activity against DENV (IC50 = 1.9 µM). The ring opened version of 30, 33 was 4-fold weaker against DENV. The substitution of the catechol carbon linker of 30 with a gem-difluoro 34 afforded a compound with equipotency with the respective ring opened analogue 33 but was slightly more active than the parent 30. The removal of the ortho-substituted oxygen of 30 furnished 35, a compound with sub-micromolar activity against DENV (EC50 = 0.95 µM). However, the removal of the meta-substituted oxygen of 30 to afford 36 led to a 6-fold decrease in potency compared with 35. Aromatising 35 to yield 37 led to a 3-fold decrease in potency (EC50 = 2.7 µM), whereas aromatising 36 to afford 38 demonstrated a 3-fold increase in potency (EC50 = 1.9 µM). The formation of the respective isoxazole's of 37 and 38, affording 39 and 40, produced inactive compounds at the top concentration tested (EC50 = >10 µM). Moving the nitrogen of 39 to form the oxazole 41 led to a >3-fold increase in potency (EC50 = 3.2 µM). The reversed thiazole 42, was not active (EC50 = >10 µM), but removal of the nitrogen from the thiazole ring system afforded a thiophene 43 that had equipotent activity with oxazole 41. Predicting that we would identify active analogues based on the 2,3-and 3,4-connectivty, we followed up on the results of 28-31 with a series of different substitutions and fused heterocycles   (Table 4) [42]. An increase in diversity to produce 6-((6-bromoquinolin-4-yl)amino)isoindolin-1-one (52) led to an equipotent inhibitor compared with 49, though occupying a different chemical space. Several attempts to increase potency with 1,3-dihydro-2H-benzo[d]imidazol-2-one derivatives (53-55) and a fused cyclic sulfone (56) yielded compounds with efficacy only in the high micromolar range.                  A furazan substitution 44 only showed limited activity against DENV (EC50 = 9.2 whereas the introduction of a sulphur to construct a 1,2,5-thiadiazole 45 led to a 2 increase in potency compared with 44. The introduction of a methyl distal to the quin
While the mechanism of the antiviral activity on both viruses is being investigated, potentially it could be targeting multiple protein targets including both cellular kinases and other proteins. Several of these compounds were originally prepared as inhibitors targeting the ATP-binding site of human kinases including cyclin-G-associated kinase (GAK), serine/threonine-protein kinase 10 (STK10) and STE20-like serine/threonine-protein kinase (SLK) [42,47,48]. While our compounds could be targeting these kinases, it is also possible that other undefined proteins may participate in the mechanism of antiviral action. Multiple examples from previous literature show that compounds with structural features impeding interaction with a kinase hinge region, have antiviral activity [35]. It is also possible that the observed phenotypes may originate from modulation of other kinases or non-ATP binding proteins [51]. Lastly, beyond targeting cellular kinases, it is possible that these compounds target a viral protein. Indeed, related quinolinones were identified as VEEV inhibitors targeting the viral nonstructural protein (nsP2) [52,53].
This set of results presents a potential step towards the identification of a novel clinical compound to combat emerging viral diseases including DENV and VEEV infections and provides a medicinal chemistry trajectory to achieve this aim. In addition, several quinoline derivatives have been shown to inhibit alpha-and beta-coronaviruses [54], demonstrating the promising potential of quinoline group as part of a broader spectrum antiviral compound.

Chemistry Method
General Procedure for the Synthesis of 4-Anilinoquin(az)olines We suspended 4-chlo-quin(az)oline derivative (1.0 eq.), aniline derivative (1.1 eq.), in ethanol (10 mL) and refluxed for 18 h. The crude mixture was purified by flash chromatography using EtOAc:hexane followed by 1-5% methanol in EtOAc; After solvent removal under reduced pressure, the product was obtained as a free following solid or recrystallized from ethanol/water. Compounds 4-31 were synthesized as previous described [38,48], 32-51 were synthesized as previously described [42]. Representative supporting information provided (see Supplementary Materials).

Virus Construct
DENV2 (New Guinea C strain) [55,56] Renilla reporter plasmid used for virus production (DENV2-Rluc) was a gift from Pei-Yong Shi (The University of Texas Medical Branch). The plasmid encoding infectious VEEV (TC-83) with a nanoluciferase reporter and used for virus production (VEEV-TC-83-nLuc) was a gift from Dr. William B. Klimstra (Department of Immunology, University of Pittsburgh) [57].

Virus Production
DENV2-Rluc RNA was transcribed in vitro using mMessage/mMachine (Ambion Austin, TX, USA) kits. DENV2-Rluc virus was produced by electroporating RNA into BHK-21 cells, harvesting supernatants on day 10 post-electroporation and titering via standard plaque assays on BHK-21 cells. VEEV-TC-83-nLuc RNA was transcribed in vitro from cDNA plasmid template linearized with MluI via MegaScript Sp6 kit (Invitrogen #AM1330, Thermo Fisher Scientific, Waltham, MA, USA) and electroporated into BHK-21 cells. The VEEV-TC-83-nLuc virus was harvested from the supernatant 24 h post-electroporation, clarified from cell debris and the titer determined by standard plaque assay on Vero cells.

Infection Assays
Huh7 cells were treated with the inhibitors or DMSO. An hour later, the cells were infected with DENV2-Rluc virus in replicates (n = 5) at a multiplicity of infection (MOI) of 0.05. The inhibitors were present for the duration of the experiment. Overall infection was measured at 48 h post-infection using a Renilla luciferase substrate. U-87 MG cells were treated with the inhibitors or DMSO. One hour later, the cells were infected with VEEV-TC-83-nLuc virus in 5 replicates at MOI of 0.01, and overall infection was measured at 18 h post-infection via a nanoluciferase assay. The inhibitors were present for the duration of the experiment. The relative light units (RLUs) were normalized to DMSO treated cells (set as 100%). GraphPad Prism nonlinear regression (curve fit) was used to generate the graphs and EC 50 values.

Viability Assays
Viability was measured using AlamarBlue ® reagent (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's protocol. Fluorescence was detected at 560 nm on InfiniteM1000 plate reader (Tecan, Männedorf, Switzerland). The raw fluorescence values were normalized to DMSO treated cells (set as 100%). Graphpad Prism nonlinear regression (curve fit) was used to generate the graphs and CC 50 values.