1. Introduction
RNA viruses constitute an immense threat to human health worldwide, as highlighted by the SARS-CoV-2 pandemic, which has caused more than 200 million infections and 4.2 million deaths as of August 2021 [
1]. Other key human pathogens include human immunodeficiency virus-1 (HIV-1), which currently affects 37.7 million people and has killed approximately the same number since the start of the epidemic [
2], and influenza A virus, which, apart from pandemics such as the 1918 “Spanish flu” that caused 50 million deaths [
3], continues to be a threat every year, despite being kept in check by a yearly vaccine approach in many countries. Finally, mosquito-borne infections by flaviviruses, which include dengue virus (DENV), Zika virus (ZIKV), West Nile virus (WNV), Japanese encephalitis virus (JEV) and yellow fever virus (YFV), account for a growing number of yearly infections; DENV alone causes more than 400 million infections every year [
4], whereas a large outbreak of >750,000 infections by ZIKV spanned 70 countries in 2015–2016 [
5,
6], with this number having grown to 87 countries with evidence of ZIKV transmission to date [
7].
Viruses exploit multiple components of the infected host cell in order to replicate efficiently and spread to infect new host cells. Importantly, all of the above viruses exploit members of the host importin (IMP) superfamily to transport a viral component into the nucleus, in most cases to subvert the host immune response [
8,
9]. DENV/ZIKV/WNV/JEV/YFV viruses, in particular, co-opt the host nuclear import-mediating IMPα/β1 heterodimer to transport viral non-structural protein 5 (NS5); in the case of DENV, viral attenuation can be achieved by mutations preventing NS5 nuclear import [
10], whilst small molecule inhibitors such as ivermectin targeting IMPα can inhibit infection by DENV/ZIKV/WNV [
11,
12,
13,
14]. Ivermectin can also inhibit infection by SARS-CoV-2 [
15,
16], influenza A [
17], HIV-1 [
11], and other RNA viruses (e.g., Venezuelan encephalitis virus [
18]).
We identified the synthetic retinoid N-(4-hydroxyphenyl) retinamide (4-HPR) through high-throughput screening (HTS) as a specific inhibitor of the binding of DENV NS5 to IMPα/β1 [
19]. Additionally, 4-HPR exhibited EC50 values of c. 1 μM against all four infectious DENV strains in vitro, but also in an ex vivo human peripheral blood mononuclear cell model of severe (antibody-dependent enhanced–ADE) infection [
19]. Furthermore, 4-HPR has since been shown to be efficacious against ZIKV [
20], WNV (Kunjin) [
21,
22] and YFV [
23]. Most importantly, a lethal mouse model of severe ADE DENV infection showed protection of 70% of infected mice through b.i.d (twice daily) administration with 20 mg/kg 4-HPR [
19]; however, no pharmacokinetic (PK) analysis was performed to inform the potential exposure required to achieve this important therapeutic outcome, and the 4-HPR formulation used is not suitable for human use.
Although 4-HPR has an established safety record from testing in Phase I and II clinical trials for various cancer indications, its relatively poor aqueous solubility remains a challenge [
24]. The standard human clinical formulation (the “HC formulation”) is a suspension of 4-HPR in corn oil with polysorbate 80 as a surfactant [
25], which can achieve plasma levels of up to 12.9 μM at a dose of 4000 mg/m
2/day in patients [
26]; patient compliance is a significant issue due to the large dose volumes required. As such, there is a clear need for an optimised formulation with enhanced solubility and in vivo PK properties.
In this study, we set out to evaluate the in vivo exposure profile of the 4-HPR formulation (mouse dengue efficacy formulation—MDE) shown to be effective in severe DENV infection [
19] and compare it to the profile of the HC formulation that is approved for human clinical use for other indications. Our PK analysis indicates that effective plasma concentrations can be achieved in mice with the HC formulation, but simulated exposure profiles indicate that trough concentrations are below the apparent therapeutic threshold in a b.i.d administration regime. We subsequently build on this to enhance 4-HPR in vivo exposure properties, developing and characterising novel self-emulsifying lipid-based formulations of 4-HPR, which show >3-fold improved exposure over the HC formulation in mice. We find that 4-HPR exposure was limited by both solubility and first-pass intestinal elimination but could be improved through inhibition of cytochrome P450 (CYP)-medicated intestinal metabolism. The results overall suggest that although the HC formulation achieves plasma 4-HPR levels above the predicted minimum effective plasma concentration for anti-DENV activity in a mouse model, self-emulsifying lipid-based formulations and CYP inhibition represent viable future options to increase exposure.
2. Materials and Methods
2.1. Materials
First, 4-HPR (C
26H
33NO
2—see
Figure 1—GM Pharma, FL, USA), diazepam (Sigma, St. Louis, MO, USA), 1-aminobenzotriazole (ABT; Honeywell Fluka, International Inc., Charlotte, NC, USA) were purchased from the sources indicated previously [
19]. The excipients used in formulation experiments were sourced as indicated; Capryol 90, dimethylsulphoxide (DMSO), Gelucire 44/14, Labrasol, Lauroglycol 90, Maisine 35-1, Maisine CC and Transcutol were from Gattefossé (Saint-Priest Cedex, France); Capmul MCM EP and Captex 355 EP/NF from ABITEC Corporation (Columbus, OH, USA); benzyl alcohol, corn oil, hydroxypropyl methylcellulose, polyethylene glycol (PEG) 400, polysorbate 80, RPMI-1640, soybean oil and Tween 85 from Merck KGaA (Darmstadt, Germany); Cremophor EL, Kolliphor RH40 from BASF SE (Ludwigshafen, Germany); ethanol from Ajax-Finechem, Scoresby, Australia); and foetal bovine serum (FBS) from CellSera Australia (Rutherford, Australia).
For in vitro digestion experiments, lipoid E PC S (phosphatidylcholine) was from Lipoid GmbH (Ludwigshafen, Germany), 4-bromophenylboronic acid, sodium taurodeoxycholate, porcine pancreatin, sodium chloride, sodium hydroxide, and Trizma maleate from Merck KGaA (Darmstadt, Germany), and calcium chloride dihydride from BDH Chemicals Australia (Kilsyth, Australia).
Ultrapure water was obtained from a Milli-Q purification system (Merck KGaA, Darmstadt, Germany). All other solvents were of HPLC grade (Merck KGaA, Darmstadt, Germany).
2.2. In Vivo PK Analysis
Studies with non-fasted male C57Bl/6 mice were conducted using established procedures in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes; study protocols were reviewed and approved by the Monash Institute of Pharmaceutical Sciences Animal Ethics Committee.
C57Bl/6 mice (22.6–28.7 g) having access to food and water ad libitum throughout the pre- and post-dose sampling period were dosed with 4-HPR either orally by gavage (2–10 mL/kg, depending upon formulation) or IV by bolus injection into the lateral tail vein (50 μL/animal, 2 mL/kg), and blood samples were collected up to 49 h post-dose by either submandibular bleed or terminal cardiac puncture (
n = 3–5 mice per timepoint for each formulation) with a maximum of four samples from each mouse. Blood samples were mixed with heparin, potassium fluoride and complete protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany), centrifuged and supernatant plasma was removed and stored at −80 °C until analysis by LC-MS. Specifics of formulations used in PK experiments are detailed in
Supplementary Table S1.
For in vivo CYP inhibition experiments, repeat oral doses of ABT (50 mg/kg b.i.d commencing 4 h before 4-HPR administration, 5 mL/kg by gavage) were administered according to a regimen shown to maintain ongoing CYP inhibition [
27]. Blood samples were collected from ABT-treated mice for up to 72 h after dosing 4-HPR.
For 4-HPR quantification, proteins were precipitated from plasma samples using acetonitrile (2-fold volume ratio) prior to analysis using a Waters Acquity Ultra-High Performance Liquid Chromatography (UHPLC) system coupled to a Waters Xevo TQ MS system (Waters Corporation, Milford, MA, USA). Chromatographic separation was performed using a C8 reverse phase column (Ascentis Express RP C8, 50 × 2.1 mm, 2.7 μm, Merck KGaA, Darmstadt, Germany) and a 3 μL injection volume. Gradient elution (methanol-water gradient with 0.05% formic acid) was performed at a flow rate of 0.4 mL/min, cycle time of 4 min, with detection by positive electrospray ionisation in multiple-reaction monitoring mode. Then, 4-HPR standards (10,000–0.5 ng/mL) were prepared in blank plasma, with diazepam as an internal standard (10 μL of 5 μg/mL in 50% acetonitrile/water per sample). For 4-HPR, the retention time was 2.97 min, m/z transition 392.23 > 161.15, cone voltage 30 V and CID 20 V.
2.3. PK Data Analysis and Profile Simulations
The dose administered to each mouse was calculated on the basis of the pre-dose body weight, the volume of formulation administered, and the 4-HPR concentration in the formulation. The plasma concentration versus time profile was defined by the average plasma concentration at each sample time, and PK parameters were calculated using non-compartmental methods (PKSolver Version 2.0) using the average dose administered to the dosing group.
For statistical analysis of exposure data, the 95% z-confidence interval for the dose-normalised AUC
0–30h was calculated in R Statistical Software (v4.1.2; R Core Team 2021) via Bailer’s method [
28], implemented as “batch” design in the package “PK” (Version 1.3.5) [
29]. Only data up to 30 h post-exposure were used for statistical comparisons to ensure that statistical comparisons were not biased by differences in the timeframe over which exposure data was monitored.
To illustrate the differences in 4-HPR plasma exposure profiles that would be expected in mice upon repeat-dose administration (b.i.d, 10 h/14 h at 20 mg/kg) of the different formulations, profiles were simulated using a simple compartmental PK approach. Initially, a linear 2-compartment model was fitted (PKSolver, Version 2.0) to the mean plasma concentration-time profile obtained after bolus IV administration of 4-HPR. The parameter estimates thus obtained (i.e., V1, k10, k12 and k21—see
Supplementary Table S2) were considered to define the systemic disposition of 4-HPR and were fixed during simulation of the oral profiles. Oral profiles were simulated using a differential equation-based compartmental model scripted in Berkeley Madonna (version 8.3.1.8). Systemic disposition parameters were fixed as described above, under the assumption that the post-absorption distribution and elimination of 4-HPR was independent of the formulation administered. Bioavailability was fixed at the value determined by non-compartmental analysis, and the first order rate constant for compound absorption was manually varied until the simulated concentration-time profile was in close agreement with the experimental concentration-time data obtained after single-dose administration of pre-dispersed lipid formulation. As all single-dose oral exposure profiles exhibited a similar Tmax, this value for the absorption rate constant was used for the simulation of all repeat-dose exposure profiles. These were obtained by setting the bioavailability (F) to the value determined for each formulation by non-compartmental analysis.
2.4. Solubility Studies
Excipients were weighed directly into glass vials at the indicated concentrations, mixed thoroughly, then transferred to Eppendorf tubes and loaded with an excess of 4-HPR. Formulations were then heated to 37 °C, vortexed extensively (initially every hour) and agitated with a stirrer and incubated at 37 °C in a shaking incubator. Periodically (every 1–3 days), undissolved drug was removed from the formulation by centrifugation (20 min at 2100× g, 37 °C) and 20 mg of supernatant was transferred to new tubes for quantification by UHPLC (stored at −20 °C), with the remaining formulation vortexed and agitated vigorously incubated at 37 °C in a shaking incubator until the next sampling. Samples were dissolved in 1 mL 1:1 chloroform:methanol, then diluted 1:100 in acetonitrile.
In this case, the UHPLC assay used a Shimadzu Nexera X2 UHPLC system (Shimadzu Corporation, Kyoto, Japan), with chromatographic separation on a C18 reverse phase column (Kinetex 2.6 μm C18 100 Å, LC Column 50 × 2.1 mm, Phenomenex Inc., Torrance, CA, USA) at 30 °C, with samples stored at 10 °C in the autosampler until injection (3 μL volume). Isocratic elution (79.95% acetonitrile, 19.95% H2O, 0.1% formic acid) was performed at a flow rate of 0.3 mL/min, with absorbance measured by the photodiode array detector. Then, 4-HPR standards (50–0.521 μg/mL) were created by successive serial dilution of a 50 μg/mL stock standard in acetonitrile, with area under curve at absorbance of 365 nm used to generate a standard curve using LabSolutions software V.5.82.
2.5. Plasma Protein Binding
Protein binding of 4-HPR was assessed in human (pooled, Innovative Research Inc., Novi, MI, USA) and mouse (collected in-house from male C57BL6 mice) plasma via Rapid Equilibrium Dialysis (RED) using pre-saturated dialysis units as previously [
30]. Briefly, RED inserts were pre-saturated overnight with 400 ng/mL 4-HPR in PBS, after which solutions were removed and discarded, and 300 µL aliquots of diluted human or mouse plasma (10%
v/
v in pH 7.4 PBS) containing 4-HPR (2000 ng/mL) were added to the donor chamber of the RED inserts (
n = 4/matrix) and dialysed against 500 µL of PBS (containing 40 ng/mL 4-HPR). During the 24 h dialysis period, the RED system was sealed with a gas impermeable film, and the plate was maintained at 37 °C under ambient atmosphere in an orbital plate shaker set at 800 rpm (ThermoMixer C; Eppendorf, Hamburg, Germany). At the end of the dialysis period, aliquots were taken from each donor and dialysate chamber to obtain post-dialysis measures of the total and unbound 4-HPR concentrations, respectively. The measured pH of pre- and post-dialysis donor matrix and dialysate buffer was within pH 7.4 ± 0.1. In parallel to the binding assessment, stability of 4-HPR in each diluted plasma matrix was assessed; there was no evidence of any degradation.
Donor and dialysate samples were processed and analysed using a matrix matching approach, and all samples were stored frozen at −80 °C until analysis by LC-MS. Compound binding was assessed on the basis of the measured concentrations in dialysate and donor samples at the end of the dialysis period, assuming that the system was at steady state by 24 h. As the binding assay was performed using diluted plasma, data were corrected for the dilution factor to give a binding value for neat plasma via an established approach which accounts for the shift in equilibria that occurs with protein dilution [
31].
4. Discussion
Previous work identified 4-HPR as a potent anti-dengue drug and demonstrated efficacy in an in vivo model of lethal DENV infection when delivered as the MDE suspension formulation [
19]. The present study analysed the PK properties of the MDE suspension formulation for the first time, with simulated exposure profiles for a twice-daily oral dosage of 4-HPR at 20 mg/kg in mice as administered in Fraser et al. [
19], indicating that the minimum plasma concentration under this dosage regime is approximately 1.0 μM. Importantly, this suggests that a trough concentration of 1 μM is sufficient for antiviral activity, which parallels the in vitro EC50 of 1–2 μM for both DENV and ZIKV [
9]. Carocci et al. [
21] showed that b.i.d administration of 4-HPR (albeit at a much higher dose of 180 mg/kg) can reduce DENV2 viremia > 50-fold in an AG129 mouse viremia model.
Although the MDE formulation is not suitable for human use, and the standard corn oil-based human clinical (HC) suspension formulation shows >3 times lower exposure at the same dose (20 mg/kg) here in mice, it should be noted that human clinical trials have used doses as high as 4000 mg/m
2/day with no significant toxicity (equivalent to approximately 100 mg/kg) [
26]. This suggests that the HC formulation would be able to safely achieve a therapeutic dose exposure, but it also highlights the need for a formulation with improved performance to reduce issues associated with patient compliance due to large dose volumes.
Various approaches have been investigated to improve 4-HPR exposure in a cancer context [
32,
39,
41,
42,
43,
44,
45,
46]. This study describes two novel self-emulsifying lipid-based formulations of 4-HPR with relatively high solubility that appear to have superior exposure profiles to the HC formulation. Simulated exposure profiles for twice-daily oral dosage of 4-HPR at 20 mg/kg in mice reveal a trough concentration of approximately 1.2 μM for lipid formulation 25 (
Figure 3B), thus allowing a 4-fold reduction in dosage to achieve the same exposure as the HC formulation. An alternative formulation using 4-HPR complexed with 2-hydroxypropyl-beta-cyclodextrin (nanofenretinide) similarly achieved higher (approximately 3-fold) exposure in mice than the HC formulation [
41]. A phase I human clinical study in paediatric neuroblastoma patients described that 4HPR/LYM-X-SORB (a free-flowing powder matrix of lysophosphatidylcholine, monoglycerides and free fatty acids at a molar ratio of 1:4:2) [
42], when mixed into a meal replacement drink, resulted in approximately 2-fold higher plasma concentration than that previously achieved at similar doses with the HC formulation [
24,
43]. Extrapolating to humans from our mouse PK data, formulations 23 and 25 may speculatively afford a potential improvement over the 4HPR/LYM-X-SORB formulation around 2-fold.
Intriguingly, the lipid-based as well as MDE formulations appear to encounter an upper limit on the bioavailability of less than 20%, with examinations of PK profiles of oral and IV 4-HPR in the absence or presence of the CYP-inhibitor, ABT, revealing a limited role for first pass hepatic elimination and implying significant intestinal metabolism. A likely mechanism for these differing contributions to first-pass elimination is the extremely high plasma protein binding of 4-HPR, which would attenuate the systemic hepatic clearance but may have a much lower effect on intestinal metabolism; this has not been previously reported. The data also lend support to the notion of pharmacologically modulating CYP-mediated metabolism of 4-HPR in patients through the use of drugs such as ketoconazole or ritonavir as a way of increasing 4-HPR exposure in existing formulations [
24,
43]. In this context, it will be important to consider drug interactions resulting from the use of CYP inhibitors, such as ritonavir, that have been well studied for various anti-human immunodeficiency virus-1 therapeutics [
47,
48], and more recently for nirmatrelvir (paxlovid) as an antiviral for SARS-CoV-2 [
49]. It will also be critical to take into account previous isolated studies in cultured mosquito cells [
50] and in an 81-year-old patient with a long history of hepatitis C infection [
51] that imply that CYP inhibitors may themselves have the potential to enhance DENV replication.
In summary, this study establishes that the existing HC formulation for 4-HPR with b.i.d administration can achieve in vivo 4-HPR exposure levels necessary for effective dengue virus prophylaxis in a mouse model. It also describes a successful lipid-based formulation approach for 4-HPR with improved in vitro and in vivo PK properties in mice that may allow dose volumes to be reduced to help increase patient compliance. Importantly, co-administration with CYP inhibitors is eminently worthy of further consideration in terms of the potential enhancement of 4-HPR efficacy in both cancer and flavivirus clinical settings.