Repurposing of the anti-malaria drug chloroquine for Zika Virus treatment and prophylaxis

One of the major challenges of the current Zika virus (ZIKV) epidemic is to prevent congenital foetal abnormalities, including microcephaly, following ZIKV infection of pregnant women. Given the urgent need for ZIKV prophylaxis and treatment, repurposing of approved drugs appears to be a viable and immediate solution. We demonstrate that the common anti-malaria drug chloroquine (CQ) extends the lifespan of ZIKV-infected interferon signalling-deficient AG129 mice. However, the severity of ZIKV infection in these mice precludes the study of foetal (vertical) viral transmission. Here, we show that interferon signalling-competent SJL mice support chronic ZIKV infection. Infected dams and sires are both able to transmit ZIKV to the offspring, making this an ideal model for in vivo validation of compounds shown to suppress ZIKV in cell culture. Administration of CQ to ZIKV-infected pregnant SJL mice during mid-late gestation significantly attenuated vertical transmission, reducing the ZIKV load in the foetal brain more than 20-fold. Given the limited side effects of CQ, its lack of contraindications in pregnant women, and its worldwide availability and low cost, we suggest that CQ could be considered for the treatment and prophylaxis of ZIKV.

infected with Brazilian strain ZIKV (ZIKV BR , Brazil-ZKV2015), a common preclinical model for ZIKV research. However, the severity of disease precludes the use of AG129 mice for the investigation of vertical ZIKV transmission. To develop a suitable model for this purpose, we used SJL mice, which have a normal IFN signalling response and have previously been used for the study of ZIKV pathogenesis 4 . We found not only that SJL mice support chronic ZIKV BR infection but also that the virus can be transmitted vertically, making this a more relevant model of ZIKV infection in humans 14 . Notably, administration of CQ to pregnant SJL mice during mid-late gestation markedly reduced ZIKV BR infection in the foetal brain. Collectively, our data suggest that CQ could be effectively and readily employed for the treatment and prophylaxis of ZIKV infection in humans.

CQ protects human neural progenitors from ZIKV infection.
Human foetal NPCs are the major target of ZIKV in the developing brain 5,15 . To examine the effect of CQ in vitro, we infected monolayer cultures of primary human foetal NPCs with ZIKV BR (Brazilian strain ZKV2015) and cultured them in the absence or presence of up to 40 μM CQ. Consistent with work by others 12 , we found that CQ efficiently (90% inhibition at 6 µM) reduced ZIKV BR infection of primary human foetal NPCs. To mimic ZIKV BR infection in the context of the 3-dimensional architecture of the developing human brain, we examined neurospheres derived from human iPSCs (Fig. 1a). We found that CQ treatment reduced both the percentage of ZIKV BR -positive cells (Fig. 1b) and the level of apoptosis in the neurospheres with an IC 50 of ~10 μM ( Fig. 1c and d).
CQ attenuates acute ZIKV-induced mortality in AG129 mice. To corroborate the in vitro findings, we first examined AG129 mice, which lack receptors for type I (α/β) and type II (γ) IFNs and have previously been used to model ZIKV infection 16,17 . To test the prophylactic effects of CQ, mice were administered 50 mg/kg/ day CQ in drinking water for 2 days and then infected with ZIKV BR (2 × 10 3 PFU retro-orbitally). CQ treatment was continued at the same dose for 5 days and then at 5 mg/kg/day until the end of the experiment. Control mice received drinking water alone.
We observed that CQ extended the average lifespan of ZIKV-infected AG129 mice to 15 days (p < 0.01, log-rank Mantel-Cox test; Fig. 2a) and significantly attenuated ZIKV-induced weight loss (p < 0.01, unpaired t-test with Welch's correction; Fig. 2b). Overall animal health was assessed using a modified 6-point scoring system 15 , which showed that CQ-treated mice remained in good health and survived for longer than the vehicle-treated mice (80% vs 0% of animals alive on day 13, respectively) ( Fig. 2c and d). These results indicate that CQ attenuated disease severity in ZIKV-infected AG129 mice, which is considered the most severe model of ZIKV infection 16 . SJL mice support chronic infection with ZIKV. Mice deficient in IFN response genes, such as single knockout (Ifnar1) A129, double knockout (Ifnar1, Ifnar2) AG129, and triple knockout (Irf3, Irf5, Irf7) TKO mice 16 , succumb to ZIKV within a few days of infection making it difficult to investigate vertical transmission of ZIKV in such an aggressive disease model. Therefore, we explored the SJL mouse model, which we have previously used to study foetal transmission with high doses of ZIKV 4 . SJL males and females at 3 months of age were infected with ZIKV BR (1 × 10 8 PFU retro-orbitally), and circulating ZIKV RNA levels were analysed by qRT-PCR over the following 50 days. Our qRT-PCR assay is only 10-fold less sensitive than a laborious and time consuming plaque forming unit assay. Using qRT-PCR we could detect the levels of ZIKV as low as 10 plaque forming units per sample. In testing our samples, we did not record any ZIKV in the samples obtained from uninfected control mice. We found that the viral titres fluctuated over time in both males and females, ranging from 5 × 10 3 to 4 × 10 5 genome copies/µg total RNA. However, the mean titres were maintained between 10 4 and 10 5 genome copies/µg total RNA ( Fig. 3a and b). Previous work has shown that ZIKV inoculation of wild-type C57BL/6 mice treated with a single dose of IFNAR1-blocking monoclonal antibody leads to infection of and damage to the testes 18 . We therefore investigated ZIKV titres in the testes of chronically infected SJL mice (3 months post-infection) and found readily detectable levels (10 3 -10 4 ZIKV genome copies/1 µg testis RNA) (Fig. 3c). Collectively, these data indicate that SJL males and females support sustained ZIKV infection and display no signs of morbidity at 3 months post-infection. The mice therefore represent a physiologically relevant model for studying paternal and maternal vertical transmission of ZIKV BR .
Vertical and horizontal transmission of ZIKV in SJL mice. We examined horizontal transmission by infecting 3-month-old SJL males and females with ZIKV BR (10 8 PFU retro-orbitally) and allowing them to mate with uninfected mice of the opposite sex. After 14 days, the uninfected males were separated and bled and circulating viral titres were measured by qRT-PCR. To avoid stress during pregnancy dams were allowed to deliver and then bled on the next day and circulating viral titres were measured by qRT-PCR. Interestingly, we observed efficient transmission of ZIKV from infected males to females but not vice versa (Fig. 4a). This is strikingly similar to the mode of horizontal transmission in humans, where female-to-male transmission is relatively rare [19][20][21] . Because we could not detect any ZIKV transmission from infected female mice to uninfected males, we concluded other routes of transmission such as via saliva or ocular secretions are insignificant.
Our previous study investigated foetal development in SJL females directly infected with a 4 × 10 10 PFU/ml of ZIKV BR on E12.5 4 . Here, we investigated whether the virus could be transmitted vertically from ZIKV-infected dams and sires to their offspring through the natural breeding process. To this end, 3-month-old female and male SJL mice were infected with ZIKV BR (10 8 PFU retro-orbitally) and immediately allowed to breed with uninfected mice of the opposite sex. After regular delivery, the 1-day-old pups were euthanized and tissue samples were analysed for viral   RNA by qRT-PCR. We found that all dams efficiently transmitted ZIKV to their pups (Fig. 4b). Notably, transmission from the infected sires to their pups occurred in fewer animals was less efficient (Fig. 4b), possibly reflecting variable ZIKV titres in the semen and variations in the levels of sexual transmission of ZIKV from infected sires to dams. The molecular and cellular mechanisms of ZIKV infection during pregnancy are poorly understood 22 , and such knowledge is critical for the development of treatment to limit ZIKV infection during pregnancy. Our results thus demonstrate that SJL males and females can transmit ZIKV vertically through the natural mating process and thus represent a unique physiological mouse model for testing of drugs that could suppress vertical viral transmission.
CQ suppresses vertical transmission of ZIKV. Next, we examined the effect of CQ treatment on vertical transmission of ZIKV in SJL mice using our previously published protocol 4 . Pregnant SJL mice (2-3 months of age) were infected with ZIKV BR (2 × 10 5 PFU retro-orbitally) on day E12.5. This dose is sufficient to cause a robust ZIKV infection in SJL mice. Infected dams were provided with CQ (30 mg/kg/day in drinking water) starting on day E13.5 and were euthanized on E18.5, at which point maternal blood and foetal brain samples were collected and analysed by qRT-PCR (Fig. 5a). This lower, 30 mg/kg, dosage of CQ was specifically use to protect pregnant mice from potential negative effects of the drug. We found that treatment with CQ reduced by ~20-fold the ZIKV titre in both maternal blood (Fig. 5b) and foetal brain (Fig. 5c). To confirm these results, whole embryos were immunostained with an anti-ZIKV envelope protein antibody. Consistent with the qRT-PCR data, this analysis revealed a significant reduction in the ZIKV immunostaining intensity in the foetuses of CQ-treated pregnant mice compared with the untreated mice ( Fig. 5d and e).

Discussion
CQ has been used worldwide for more than half a century for anti-malaria prophylaxis and therapy without evidence of foetal harm [23][24][25][26] . CQ can cross the placental barrier and would be expected to reach similar concentrations in maternal and foetal plasma 27 . The side effects of CQ have been thoroughly evaluated in a malaria prophylaxis study (400 mg/week), which found no increase in the incidence of birth defects 11 . High CQ concentrations (up to 500 mg/day) were administered to pregnant women with severe lupus or rheumatoid arthritis. Although a few instances of spontaneous abortion were observed (likely a consequence of the disease itself), all term deliveries resulted in normal healthy newborns 28 , suggesting that high doses of CQ do not interfere with foetal development in humans. The dosages of CQ we employed in our study were either comparable or significantly lower relative to the acceptable and widely used dosages in humans. Studies in rodent models have found that brain concentrations of hydroxychloroquine (CQ analogue) are 4-30 times higher than plasma concentrations 29 , suggesting that it has a favourable pharmacokinetic profile for inhibition of ZIKV infection in NPCs. In arthritis patients, plasma CQ concentrations reached 10 µM after daily administration of 5 mg/kg/day for a week 30 . Given that the half-life of CQ in humans is approximately 40 days 31 , people treated with 5 mg/kg/day CQ for 7 days will build up over 30 mg/kg of CQ, which is comparable to the regimen used in our animal studies. CQ treatment can be associated with retinopathy, but the reported threshold dose in humans, 5.1 mg/kg/day 30 , thus allowing sufficient accumulation of CQ (see above). Moreover, eye disease was not detected in a study of more than 900 rheumatoid arthritis patients treated with up to 4.0 mg/kg/day CQ for an average of 7 years 30 . Therefore, a level of CQ sufficient to protect SJL mice from ZIKV could be safely build up in a human body in relatively short, 7 days, time period and then maintained for many weeks or months with a minimal intake of CQ. The pharmacokinetics of CQ thus make it an excellent candidate for prophylaxis in individuals at high risk of ZIKV infection (e.g., residents or visitors in ZIKV endemic areas). Our results demonstrate that CQ effectively reduces ZIKV infection in primary human foetal NPCs and in two mouse models, and that CQ at doses comparable to or less than those broadly used in humans can markedly reduce maternal and foetal infection.
ZIKV infects cells through receptor-mediated endocytosis and membrane fusion within acidic endosomes 32 . CQ is thought to affect acidification of the endosomes and thus obstructs fusion of the flaviviral envelope protein with the endosomal membrane 32 . Cellular proteases, including furin, are essential for cleavage of the flaviviral prM during viral egress 33 . This transition is pH-dependent, and alterations in the intracellular pH may result in the release of less infectious virions 34 . Clearly, additional studies are required to determine the precise pharmacological mechanism by which CQ counters ZIKV activity.
Neurosphere infection and treatment. Neurospheres were dissociated with Accutase (Thermo Fisher) and counted. For each assay, ~20 neurospheres/condition were treated as follows: uninfected (MOCK on figures), uninfected and treated with DMSO, infected with ZIKV BR at a multiplicity of infection of 1, or ZIKV infected and treated with CQ at 5, 20, or 40 µM. CQ was added during viral adsorption (1 h at 37 °C). Medium containing the appropriate concentration of CQ was changed after 2 days. At 96 h post-infection, neurospheres were transferred to polyornithine/laminin double-coated plates and maintained for 1 week to initiate neuronal maturation. Medium supplemented with CQ was changed every 2-3 days.
Histology and immunohistochemistry. Mouse embryos obtained on E18.5 were fixed for 48 h in 4% formaldehyde in phosphate-buffered saline (PBS), transferred to sucrose, and embedded in paraffin. Serial sections (5 μm) were cut along the sagittal axis of the embryo. Slides were deparaffinised and rehydrated using xylene and graded ethanol. Antigen retrieval was performed in a pressure cooker at 7.5 psi in 0.1 M Tris-HCl buffer (pH 9.0) for 15 min. Slides were rinsed with water 6 times at room temperature and washed for 5 min in PBS. Endogenous peroxidase activity was quenched by incubation in 3% hydrogen peroxide in PBS for 30 min at room temperature. Slides were incubated for 16-18 h at 4 °C with primary anti-Flavivirus Group Antigen (Millipore, #MAB10216) diluted 1:250 in Dako Antibody Diluent with Background Reducing Components (Agilent, #S3022). After rinsing in PBS 3 times for 5 min each, the slides were incubated with a horseradish peroxidase-conjugated goat anti-mouse secondary antibody (Abcam, #ab2891) for 30 min at room temperature. Slides were washed again in PBS, incubated for 3 min with DAB complex (ImmPACT DAB Peroxidase Substrate, Vector Laboratories,