N-Methyl-d-Aspartate (NMDA) Receptor Blockade Prevents Neuronal Death Induced by Zika Virus Infection

ABSTRACT Zika virus (ZIKV) infection is a global health emergency that causes significant neurodegeneration. Neurodegenerative processes may be exacerbated by N-methyl-d-aspartate receptor (NMDAR)-dependent neuronal excitoxicity. Here, we have exploited the hypothesis that ZIKV-induced neurodegeneration can be rescued by blocking NMDA overstimulation with memantine. Our results show that ZIKV actively replicates in primary neurons and that virus replication is directly associated with massive neuronal cell death. Interestingly, treatment with memantine or other NMDAR blockers, including dizocilpine (MK-801), agmatine sulfate, or ifenprodil, prevents neuronal death without interfering with the ability of ZIKV to replicate in these cells. Moreover, in vivo experiments demonstrate that therapeutic memantine treatment prevents the increase of intraocular pressure (IOP) induced by infection and massively reduces neurodegeneration and microgliosis in the brain of infected mice. Our results indicate that the blockade of NMDARs by memantine provides potent neuroprotective effects against ZIKV-induced neuronal damage, suggesting it could be a viable treatment for patients at risk for ZIKV infection-induced neurodegeneration.

Z ika virus (ZIKV) is a mosquito-borne pathogen that poses significant public health concerns due to the recent rapidly expanding outbreak. It is an emerging pathogen that belongs to the Flavivirus genus and Flaviviridae family, along with other clinically important arboviruses, including Dengue virus (DENV), Chikungunya virus (CHIKV), and West Nile virus (1)(2)(3). Since its first isolation in 1947 from a rhesus monkey in the Ziika Forest in Uganda (4), sporadic and benign human cases of ZIKV infection, most of them asymptomatic, have been reported in Africa and Asia (5), until the first serious outbreak that occurred in the population of Yap Island (Micronesia) in 2007 (6). This was followed by a major outbreak in French Polynesia in 2013 (7). Recently, ZIKV was introduced to the Western Hemisphere, causing an ongoing epidemic in South America, with millions of infections across Brazil, Colombia, and Venezuela (6,8). The first case of ZIKV infection in Brazil was reported in May 2015 (9), and ZIKV infection has been associated with severe neurological complications, including microcephaly and ophthalmological alterations, such as severe macular neuroretinal atrophy and foveal reflex loss in infants born from ZIKV-infected mothers (10)(11)(12)(13), uveitis (14), and direct virus-induced inflammatory polyneuropathy and Guillain-Barré syndrome (GBS) in adults (15,16). As a result, on February 2016, the World Health Organization announced that the ZIKV outbreak was a Public Health Emergency of International Concern (17). However, so far there is no available vaccine and treatment is only supportive (18).
The pathogenesis of ZIKV infection remains poorly understood and involves a complex interplay between viral and host factors. Recent studies have shown that ZIKV has extensive tropism to the central nervous system (CNS) and causes significant neurodegeneration, especially of neural progenitor cells (19)(20)(21)(22). These neurodegenerative effects appear to account for the neurological disorders associated with ZIKV infection (7,12,23).
Glutamate is the main excitatory neurotransmitter in the brain and plays a pivotal role during neurodegenerative processes (24)(25)(26). There are two types of glutamate receptors: ionotropic and metabotropic (27). Several studies indicate that glutamatergic overstimulation via activation of ionotropic glutamate receptors leads to excitotoxicity, which promotes neuronal calcium overload and, consequently, neurodegeneration (28). Here, we hypothesize that N-methyl-D-aspartate receptor (NMDAR) blockade by memantine can avoid the death of nearby neurons and decrease neurodegeneration and neuroinflammation associated with ZIKV infection. For instance, the NMDAR antagonist memantine has been approved by international regulatory agencies for the treatment of Alzheimer's disease (AD) (26). Moreover, the neuroprotective properties of memantine have been described in other contexts, such as cerebral ischemia (29). In AIDS, memantine appears to inhibit neuronal damage promoted by the viral protein gp120 (30)(31)(32)(33).
Here, using relevant in vitro and in vivo settings, we investigated whether NMDAR blockade could prevent neurodegeneration induced by ZIKV infection. Our data show that blocking NMDARs prevents neuronal death induced by ZIKV, suggesting that memantine may be a useful therapy to prevent neurological disorders in ZIKV-infected patients. mice with 4 ϫ 10 5 PFU of ZIKV induced lethality that was observed around the 6th and 7th days after ZIKV inoculation (see Fig. S1A in the supplemental material). Since around 80% of the IFN-␣/␤R ؊/؊ infected mice succumbed to the infection on day 6, all subsequent experiments were conducted at this time point. No lethality was observed for WT and IFN-␥R ؊/؊ mice after ZIKV inoculation (Fig. S1A). Disease manifestations, characterized by appearance of clinical signs such as ruffled fur, ataxia, partial or complete hind limb weakness or paralysis, and massive body weight loss (Fig. 1A), starting from day 5 after infection, were detectable only in IFN-␣/␤R ؊/؊ ZIKV-infected group. In contrast, MOCK-infected WT and IFN-␣/␤R ؊/؊ mice or WT infected mice did not present any body weight loss or signs of disease (Fig. 1A). Infection of IFN-␥R ؊/؊ mice, even with higher inocula, did not induce any signs of disease manifestation, as shown by the absence of lethality (Fig. S1A) or body weight loss (Fig. S1B), indicating that type II IFN deficiency alone is not sufficient to induce disease or death after ZIKV infection.
Next, we performed a series of experiments to characterize the clinical and inflammatory aspects at days 3 and 6 (peak of disease manifestation) after ZIKV infection in WT and IFN-␣/␤R ؊/؊ mice. No significant changes in platelet levels (Fig. 1B), hematocrit , and change in body weight was analyzed daily. Results are expressed as the percentage of initial weight loss. (B to I) Three or 6 days (peak of disease) after ZIKV inoculation, mice were culled and blood and tissues collected for the following analyses: (B) number of platelets, shown as platelets ϫ 10 3 /ml of blood; (C) hematocrit index, expressed as percentage of volume occupied by red blood cells; (D) total and differential cell counts on blood, represented as the number of differential cell counts (leukocytes, mononuclear cells, and neutrophils) normalized by the percentage of total cell counts; (E) neutrophil influx to the brain; (F to I) concentrations of CXCL1, CCL5, IL-1␤, and TNF-␣ in mouse brain, shown as picograms per 100 mg of brain tissue. All results are expressed as the mean Ϯ standard error of the mean (SEM) and are representative of at least two independent experiments. *, P Ͻ 0.05 compared to control uninfected mice (MOCK). WT, SV129 mice; KO (knockout), IFN-␣/␤R Ϫ/Ϫ mice. p.i., postinfection.
NMDAR Blockade Prevents ZIKV-Induced Neuronal Death ® index (Fig. 1C), or total and differential blood counts (Fig. 1D) were detected in ZIKV-infected WT mice compared to mock-infected mice. Levels of inflammatory parameters represented by levels of myeloperoxidase (MPO) activity, indicative of neutrophil recruitment to the brain (Fig. 1E), or production of chemokines or cytokines, as assessed by levels of CXCL1 (Fig. 1F), CCL5 (Fig. 1G), interleukin-1␤ (IL-1␤) (Fig. 1H), and tumor necrosis factor alpha (TNF-␣) (Fig. 1I) in the brain of ZIKV-infected WT mice, at both 3 and 6 days after ZIKV inoculation, were similar to those of mock-infected mice ( Fig. 1E to I). However, significant alterations of all of the parameters mentioned were detected in IFN-␣/␤R ؊/؊ mice at day 6 after ZIKV inoculation (Fig. 1C to I), with exception of platelet levels (Fig. 1B). More specifically, increased hematocrit index (Fig. 1C), elevated total and differential blood leukocyte counts, especially neutrophil numbers (Fig. 1D), and very high levels of MPO in brain tissue (Fig. 1E) were detected in ZIKV-infected IFN-␣/␤R ؊/؊ mice. Additionally, elevated levels of the inflammatory mediators CXCL1 (Fig. 1F), CCL5 (Fig. 1G), IL-1␤ (Fig. 1H), and TNF-␣ (Fig. 1I) were detected in the brain of ZIKV-infected IFN-␣/␤R ؊/؊ mice. Of note, no changes in the levels of the following mediators were recorded: brain-derived neurotrophic factor (BDNF), nerve growth factor (NGF), IL-10, or IL-6 (see Fig. S2 in the supplemental material). Therefore, our results show that infection of IFN-␣/␤R ؊/؊ mice with ZIKV induces several clinical and systemic inflammatory manifestations that impact directly on the CNS of these mice.
Several studies have shown that ZIKV is neurotropic (19)(20)(21)(22). Accordingly, studies conducted in mice detected elevated viral loads or elevated ZIKV RNA levels in the brain of adult ZIKV-infected IFN-␣/␤R ؊/؊ mice, even 28 days after infection (34,35). Our data showed that massive inflammation was found at the peak of ZIKV infection in the brain of ZIKV-infected IFN-␣/␤R ؊/؊ mice, as indicated by elevated levels of MPO activity (Fig. 1E) and inflammatory mediators ( Fig. 1F to I). Taking into account the virus neurotropism and its effects in the brain, we decided to search for the presence of virus in this target organ. Interestingly, elevated viral loads were recovered from the brain of ZIKV-infected IFN-␣/␤R ؊/؊ mice, starting on day 3 of infection and peaking on day 6 ( Fig. 2A). Additionally, ZIKV was also recovered from the optic nerve of ZIKV-infected IFN-␣/␤R ؊/؊ mice (Fig. 2B). Of note, no virus was detected in the brain or optic nerve of WT mice infected with ZIKV ( Fig. 2A and B).
ZIKV infection has been shown to cause substantial ophthalmic alterations in newborns of infected mothers (13) and in mice (37). To investigate possible ophthalmic alterations induced by ZIKV in vivo, intraocular pressure (IOP) of mice was evaluated. Results revealed significant increase in IOP on days 3 and 6 after ZIKV infection in both IFN-␣/␤R ؊/؊ and WT mice in comparison to mock-infected matched controls (Fig. 2C). Nevertheless, no change in the number of retinal ganglionar cells (RGC) following ZIKV infection of either WT or IFN-␣/␤R ؊/؊ mice was detected ( Fig. 2D and F).
In agreement with the inflammation (Fig. 1E to I) and elevated viral loads ( Fig. 2A) found in the CNS of ZIKV-infected IFN-␣/␤R ؊/؊ mice, elevated histopathological scores were observed in the brain of IFN-␣/␤R ؊/؊ mice ( Fig. 2E and F). Histopathological analysis of the brain of ZIKV-infected WT mice revealed mild gliosis and diffuse meningeal infiltration of polymorphonuclear and mononuclear leukocytes ( Fig. 2E and F). Mock WT and IFN-␣/␤R ؊/؊ mice did not show any histopathological sign of inflammation in the cerebral cortex and meninges ( Fig. 2E and F). In contrast, infection of IFN-␣/␤R ؊/؊ mice resulted in severe meningoencephalitis characterized by infiltration of leukocytes in the meninges (arrows), loss of hippocampal neurons, perivascular cuffs, gliosis, and focal areas with necrotic and apoptotic cells ( Fig. 2E and F). Corroborating these data, severe neurodegeneration, as assessed by fluoro-jade C staining, was found in the brain of ZIKV-infected IFN-␣/␤R ؊/؊ mice (  (Fig. S3B). Microgliosis, as measured by IBA-1 staining, was also increased in different brain sections from infected IFN-␣/ ␤R ؊/؊ mice ( Fig. 3C and D; Fig. S3C and D). Therefore, the data shown here confirm the neurotropism of ZIKV, which is associated with severe brain damage in IFN-␣/␤R ؊/؊ mice. Since WT mice were very resistant to ZIKV infection and showed only mild changes after virus inoculation, our next experiments were conducted only in IFN-␣/ ␤R ؊/؊ mice, which were highly susceptible to infection.
ZIKV induces massive neuronal damage of primary cultured neurons. Mechanisms underlying cell death induced by ZIKV have not been fully elucidated. To address this question, we prepared primary cultures of glial and neuronal cells from the brain of immunocompetent mice. We evaluated the ability of ZIKV to infect and replicate in these primary cultures ( Fig. 4A and B). ZIKV replication in glial cells was discrete with maximum replication levels detected at 48 h after infection, followed by reduction of viral loads at later time points (Fig. 4A). In contrast, viable virus was recovered from cell supernatant of ZIKV-infected undifferentiated neurons (5 days in vitro) at early time points (12 h), reaching a peak of replication at 48 and 72 h after ZIKV infection (multiplicity of infection [MOI] of 1) (Fig. 4B). Of note, virus replication in neurons was about 2-to 3-fold higher than that in glial cells, suggesting increased susceptibility of neuronal cells to ZIKV ( Fig. 4A and B). Neuronal death was time dependent, reaching a  Figure 4D shows representative images of mock-infected and ZIKV-infected neurons labeled with calcein acetoxymethyl ester (AM) (where green indicates live cells) and ethidium homodimer (where red indicates dead cells) after 72 h of infection. Therefore, our data show that ZIKV is able to infect and replicate in primary neurons and induce massive neuronal cell death.
NMDAR blockade prevents neuronal death induced by ZIKV. Neurodegeneration is a hallmark of neurodegenerative diseases such as Alzheimer's disease (AD) (38)(39)(40). In these neurodegenerative conditions, neurological alterations appear to be closely associated with NMDAR-dependent neuronal death. Thus, we decided to test whether memantine, an NMDAR antagonist used to treat Alzheimer's disease, can prevent neuronal cell death triggered by ZIKV infection. Our results indicate that memantine treatment was able to prevent neuronal cell death induced by ZIKV (Fig. 5A), especially at the highest tested dose (30 M). At these concentrations, memantine did not interfere with the ability of the virus to replicate in neuronal cells in vitro (Fig. 5B). To further evaluate whether NMDAR blockade was the underlying mechanism of neuroprotection, we tested the effects of three other NMDAR antago- nists. Dizocilpine (INN no. MK-801) is a noncompetitive antagonist of the NMDAR that, like memantine, binds inside the ion channel of the receptor, thus preventing the flow of ions, including Ca 2ϩ , through the channel. Figure 5C shows the dose-dependent effects of MK-801 treatment on ZIKV-infected neurons. Neuronal cell death was reduced at the 10 M dose of MK-801 and totally prevented at the highest dose tested (100 M). Agmatine, also known as (4-aminobutyl) guanidine, is an endogenous polyamine derived from enzymatic decarboxylation of L-arginine (41). Agmatine emerged as a neuromodulator and a promising agent to manage various central nervous system disorders by modulating the nitric oxide (NO) pathway, glutamate NMDARs, and oxidative stress (42). Figure 5D shows the effect of agmatine treatment on ZIKVinfected neurons. Interestingly, agmatine treatment prevented neuronal death in all tested doses (Fig. 5D).
NMDARs form tetrameric complexes that consist of several homologous subunits. The subunit composition of NMDARs is plastic, resulting in a large number of receptor subtypes. As each receptor subtype has distinct biophysical, pharmacological, and signaling properties, there is great interest in determining whether individual subtypes carry out specific functions in the CNS under pathological conditions (43). Transcriptome sequencing (RNA-seq) analysis performed between mock-infected and ZIKVinfected human cortical neural progenitor cells (hNPC) by Zhang et al. (44) revealed increased expression of the GRIN2B NMDAR subunit and the downstream FoxO1 gene (see Fig. S4A and B, respectively, in the supplemental material). Since primary neurons cultured for up to 9 days in vitro mostly express GluN2B-containing NMDARs (45), we tested whether ifenprodil, which is a specific inhibitor of GluN2B-containing NMDAR, was capable of preventing ZIKV-induced neuronal cell death. Similarly to what was NMDAR Blockade Prevents ZIKV-Induced Neuronal Death ® observed following memantine, MK-801, and agmatine treatment, Fig. 5E shows a dose-response effect of ifenprodil in preventing ZIKV neuronal death in vitro. Ifenprodil treatment at a dose of 0.0001 M has no effect on the percentage of dead cells in comparison to nontreated ZIKV-infected neurons. However, at doses of 0.001 and 0.01 M, treatment with ifenprodil was able to partially and fully prevent ZIKV-induced neuronal death, respectively (Fig. 5E); without any change in viral load culture supernatant levels (data not shown). Together, our results show that the blockade of NMDAR prevents ZIKV-induced neuronal cell death in vitro.
Memantine is a FDA-approved drug, has been widely used to treat patients with Alzheimer's disease, and thus was the drug of choice for further investigations. To test the neuroprotective effect of memantine in vivo, we conducted a series of experiments in IFN-␣/␤R ؊/؊ mice infected by ZIKV. Mice were treated with 30 mg/kg memantine, based on previous reports indicating that this dose was sufficient to improve a number of alterations exhibited by AD and stroke mouse models (46,47). Memantine treatment (30 mg/kg of body weight twice a day [b.i.d.]), starting on day 3 postinfection, failed to prevent the course of certain disease manifestations induced by ZIKV infection (see Fig. S5A and S5C to G in the supplemental material). ZIKV inoculation induced body weight loss starting at day 5 of infection in both vehicle-and memantine-treated groups (Fig. S5A). In accordance, ZIKV infection induced a similar increase in MPO activity (Fig. S5C) and production of chemokines and cytokines in the brain of vehicleand memantine-treated ZIKV-infected mice (Fig. S5D to G). However, total and differ- ® ential cell counts in the blood of ZIKV-infected mice treated with memantine were significantly reduced in comparison to infected vehicle-treated littermates (Fig. S5B).
Overall, memantine treatment of ZIKV-infected mice was unable to modify the course of clinical manifestations and inflammation in mouse brain.
Corroborating the in vitro virological data (Fig. 5B), viral loads in the brain and optic nerve of memantine-treated mice were similar to those of vehicle-treated mice (Fig. 6A  and B). Interestingly, memantine treatment was also able to prevent the increase of IOP shown after ZIKV infection (Fig. 6C) without affecting the number of retinal ganglionar cells (RGC) (Fig. 6D and F). Moreover, memantine treatment partially reversed the

NMDAR Blockade Prevents ZIKV-Induced Neuronal Death
® cortical and hippocampal damage, as well as the infiltration of leukocytes in the meninges induced by ZIKV infection, as assessed by histopathological scores observed in the brain of ZIKV-infected mice compared to vehicle-treated infected controls (Fig. 6E and F). In addition to reducing overall brain damage, memantine treatment also prevented ZIKV-induced neurodegeneration ( Fig. 7A and B; see Fig. S6A and B in the supplemental material) and greatly decreased microgliosis (Fig. 7C and D and Fig. S6C and D) in all brain substrates tested, including prefrontal and motor cortex, striatum, and hippocampus. Of note, no difference in all performed analyses was found between the control groups (vehicle-and memantine-treated IFN-␣/␤R Ϫ/Ϫ mice). Overall, our results show that memantine treatment prevents neuronal cell death induced by ZIKV infection without interfering with the ability of the virus to replicate in the host.

DISCUSSION
The rapidly expanding outbreaks of ZIKV and in particular its introduction in the Americas, including its unexpected association with neurological disorders, such as microcephaly and GBS, made this infection a matter of utmost public health concern, with an urgent need for therapies. In the present study, we evaluated the neuropro-  tective effect of the blockade of NMDAR by using an FDA-approved noncompetitive NMDAR antagonist drug (memantine) and other NMDAR blockers against ZIKVassociated neurodegeneration in vitro and in vivo. The major findings of this study can be summarized as follows. (i) ZIKV is able to infect and replicate in primary undifferentiated neurons and subsequently induce massive neuronal death. (ii) ZIKV induces neurodegeneration, microgliosis, and ophthalmologic disorders, such as increased intraocular pressure (IOP), in vivo. (iii) NMDAR blockade, by memantine, MK-801, agmatine sulfate, or ifenprodil, prevents neuronal death induced by ZIKV in vitro without affecting the ability of the virus to replicate in those cells. (iv) Neurodegeneration, microgliosis, and overall brain damage were ameliorated in ZIKV-infected mice, and increase in IOP levels was prevented by memantine treatment. These data, therefore, indicate an important neuroprotective effect of NMDAR blockade during ZIKV infection in vitro and in vivo.
In the present study, we characterized the disease induced by a clinical isolate of ZIKV (HS-2015-BA-01) from an outbreak that occurred in Brazil in 2015 in wild-type mice or type I or II IFN receptor-deficient mice. Our results are in accordance with the recent literature and show that IFN-␣/␤R Ϫ/Ϫ mice are susceptible to ZIKV infection (35). As expected, WT mice exhibited no weight loss, morbidity, or mortality associated with ZIKV infection. Interestingly and unexpectedly, deficiency in the IFN-␥ receptor alone did not confer increased susceptibility to infection by ZIKV, even when high inocula of the virus were administered (Fig. S1B). The finding that type II IFN alone is not sufficient to confer susceptibility in this experimental setting contrasts with our previous studies with a similar flavivirus, the dengue virus (48,49). We also characterized the inflammatory response induced by ZIKV in vivo. We found that ZIKV infection induces leukocytosis, especially of lymphocytes and neutrophils, at the peak of infection. Indeed, increased leukocyte recruitment to the brain of IFN-␣/␤R Ϫ/Ϫ mice was detected after 6 days of infection and was associated with elevated levels of inflammatory mediators, such as the chemokines CCL5 and CXCL1 and the cytokines IL-1␤ and TNF-␣. Interestingly, similar findings were observed in humans infected by ZIKV (50,51). The Centers for Disease Control and Prevention (CDC) reported the occurrence of leukocytosis, with predominance of neutrophils, in a patient infected by ZIKV during an outbreak in Puerto Rico in 2015 (50), and elevated levels of inflammatory mediators such as IL-1␤, IL-2, IL-4, IL-6, IL-9, IL-10, IL-13, IL-17, CCL3, CCL5, CXCL10, and vascular endothelial growth factor (VEGF) were found during both acute and convalescent phases of ZIKV (51). Those results show that ZIKV infection of IFN-␣/␤R Ϫ/Ϫ mice mimics some features of human clinical disease and may represent a useful platform for therapeutic testing in vivo.
Our data support the recent evidence of ZIKV cerebral tropism and its ability to promote neurodegeneration, especially of neural progenitor cells (19)(20)(21)(22). Here we described important histopathological alterations, such as massive neurodegeneration and microgliosis in the brain of ZIKV-infected mice. Bell and colleagues described similar results in 1971 after intracerebral infection of mice with ZIKV (52). These authors showed ZIKV replication in neurons and astroglial cells of Swiss Webster mice, and those findings were associated with cell damage, neurodegeneration, and necrosis, especially in the cortex and hippocampus (52). Accordingly, Wu and colleagues reported the occurrence of vertical ZIKV transmission experimentally, confirming the ability of ZIKV to target the radial glial cells, affecting the cortex development of offspring mice (53). Additionally, clinical evidence revealed the occurrence of ophthalmoscopic alterations in ZIKV-infected adults and infants born from ZIKV-infected mothers (13). Recently, Miner and colleagues have shown that ZIKV is able to infect several regions of the eye, including the retina, leading to apoptosis of neurons of the visual pathway (37). Our data also describes ophthalmic alterations induced by ZIKV infection in vivo. After infection, a significant increase in intraocular pressure (IOP) was observed in both WT and IFN-␣/␤R Ϫ/Ϫ mice, despite the presence of ZIKV only in IFN-␣/␤R Ϫ/Ϫ mice. However, no difference was found in RGC counts between the evaluated groups. The latter results may be explained by the short period of ZIKV infection. Our data suggest that infection by itself may be sufficient to elicit the increase of IOP levels in those mice. Corroborating our findings, Foureaux et al. demonstrated that elevated IOP levels found during the early phases of glaucoma induction in Wistar rats were independent of the reduction in the numbers of RGC, which occurred only after 15 days of glaucoma induction (54). Further studies are needed to investigate the mechanisms by which ZIKV induces the increase in IOP levels.
The development of antiviral drugs is essential for the treatment of viral infections, including by ZIKV (55). However, therapies capable of preventing the neurodegeneration induced by ZIKV may also be useful and be used in combination with antiviral drugs. Several studies indicate that activation of ionotropic glutamate receptors, especially NMDARs, plays an important role in the excitotoxic process that occurs in neurodegenerative diseases, such as Alzheimer's disease (28). It is still not clear how ZIKV promotes neuronal cell death. However, the neurons that die due to ZIKV infection may release glutamate and promote neurodegeneration of nearby cells, propagating neuronal cell death, as typically seen in neurodegenerative processes (56). Released glutamate can also activate microglial cells and trigger neuroinflammation, which will lead to further neuronal damage (57). Additionally, brain transcriptome analysis of ZIKV-infected babies with microcephaly revealed massive alterations in glutamatergic synapse and calcium regulatory pathways (unpublished data). Our results showed that memantine treatment was able to prevent neuronal cell death and microgliosis induced by ZIKV in vitro and in vivo, without affecting the ability of ZIKV to replicate in the host. The striking effect of memantine on ZIKV-mediated neuronal cell death suggests that this NMDAR antagonist not only could be blocking propagation of neurodegeneration, as we first hypothesized, but also could be interfering with the mechanism of neuronal cell death triggered by ZIKV itself. Moreover, our results strongly indicate that ZIKV-induced neuronal cell death depends on GluN2Bconataining NMDARs. Future experiments will be important to investigate how ZIKV promotes neuronal cell death and the involvement of NMDARs in this mechanism. Since NMDAR is the major excitatory receptor in the mammalian brain, blockade of this receptor could generate a series of adverse effects due to alterations in neurotransmission. However, memantine has been shown to be safe in patients, with only minor side effects reported. For the latter reasons, memantine is the only NMDAR antagonist approved for the treatment of AD. Mechanistically, NMDAR blockade by memantine leads to a decrease in excessive calcium entry, blocking excitotoxicity (58)(59)(60). Memantine is a noncompetitive NMDAR antagonist with relatively rapid off-rate from the channel. Moreover, the effects of the drug are strongly voltage dependent, leading to selective blockade of overactivated extrasynaptic NMDARs, which are associated with neurotoxicity (61)(62)(63)(64). Therefore, memantine blocks preferentially excessive NMDAR activity without disrupting physiological synaptic transmission, producing less adverse effects than other NMDAR antagonists (65). Notably, memantine is classified by the Food and Drug Administration (FDA) as a pregnancy drug category B, which means that memantine is probably safe to be used during pregnancy, making this drug a potential therapeutic tool to prevent and/or minimize ZIKV-related microcephaly in infected pregnant women.
In conclusion, the present study shows that ZIKV has tropism for the CNS and replicates preferentially in neurons, inducing neurodegeneration, neuroinflammation, and ophthalmologic disorders. ZIKV infection leads to massive neuronal damage via both direct replication in neuronal cells and possibly through increased excitotoxicity via overactivation of NMDARs in nearby cells. Although NMDAR blockade by memantine was very effective at preventing ZIKV-induced neuronal cell death and neurodegeneration, some aspects of disease in IFN-␣/␤R Ϫ/Ϫ mice, as seen by body weight loss and production of inflammatory mediators, were not modified by memantine treatment. This is likely due to the very high viral loads observed in infected mice. Once effective antiviral drugs are available for in vivo testing, it will be important to evaluate whether the combination of neuroprotective drugs, such as memantine, and antiviral drugs could be the ideal treatment for ZIKV infection. Hematological analysis. Blood was obtained from the cava vein in heparin-containing syringes at the indicated times after ZIKV infection. Platelets and the hematocrit index were analyzed as described in reference 66. The total leucocyte count was obtained by using a Neubauer chamber. Differential counts were subsequently quantified microscopically from blood smears of each mouse.

MATERIALS AND METHODS
Indirect detection of leukocytes in the brain. The extent of neutrophil accumulation in the brain of control and ZIKV-infected mice was measured by assaying myeloperoxidase activity, as previously described (67).
Measurement of cytokine/chemokine concentrations. The concentration of cytokines (IL-6, IL-10, TNF-␣, and IFN-␥), chemokines (CXCL1 and CCL5), and growth factors (NGF and BDNF) in the brain of control and ZIKV-infected mice was measured using commercially available antibodies and according to the procedures supplied by the manufacturer (R&D Systems, Minneapolis, MN).
Histopathological analysis. Brains from control and ZIKV-infected mice, treated or not treated with memantine, were collected and processed for hematoxylin and eosin (H&E) staining as described in reference 67. Histopathological scoring was performed according to criteria adapted from a previous study (67) by a researcher in a blind manner. For easy interpretation, the overall score totalized 12 points. Histopathological scoring was performed in cerebral cortex and hippocampus. Each area was graded as follows: 0, no damage; 1, minimal tissue destruction and/or mild inflammation/gliosis; 2, mild tissue destruction and/or moderate inflammation/gliosis; 3, definite tissue destruction (neuronal loss and parenchymal damage) and intense inflammation; 4, necrosis (complete loss of all tissue elements with associated cellular debris). Meningeal inflammation was graded following a 0-to 4-point scale, with 0 representing no inflammation and 1 to 4 corresponding to 1 to 4 cell layers of inflammation, respectively. The final score was calculated as a sum of cerebral cortex and hippocampus scores added to the score obtained from the meningeal inflammation analysis, totalizing a maximum of 12 points.
Histopathological analysis of the eyes was performed as previously described (54). The retinal ganglion cell (RGC) counting was performed in 6 histological slides from each eye sample covering the whole extension of the retina, including the area of the optic nerve, using an Olympus BX 41 microscope (Olympus, Irving, TX).
Immunohistochemistry IBA-1 and fluoro-jade C staining. Sections from hippocampus, striatum, and prefrontal and motor cortex of mice were assessed for microgliosis (IBA-1 staining) according to the procedures supplied by the manufacturer (Vector Elite kit) and neurodegeneration (fluoro-Jade C staining) as described in reference 68. Results represent the analysis of 2 images that were taken from each analyzed slice. We analyzed 3 slices per brain area per mouse and used 5 to 7 mice per experiment. We first analyzed one set of mice (2 mice per group), and then we repeated the same experiment (3 to 5 animals per group) to make sure our data could be reproduced. We obtained the same results in the two independent experiments, and thus, the graphed results represent the average obtained from these 5 to 7 mice. The images presented in the article are representative of one of those experiments.
IOP evaluation. IOP measurements were performed on days 0, 3, and 5 after ZIKV infection using a Tono-Pen Vet applanation tonometer (Reichert Technologies, NY) as described in reference 54.
Primary neurons and glial cell cultures. Neuronal cultures were prepared from the cortex and striatal regions of embryonic day 15 (E15) C57BL6j wild-type mouse embryo brains, and glial cells were obtained from the whole brain of newborn C57BL6j wild-type mice (1 to 3 days). After dissection, the brain tissue was submitted to trypsin digestion followed by cell dissociation using a fire-polished Pasteur pipette. Neuronal cells were plated onto poly-L-ornithine-coated dishes in neurobasal medium supplemented with N2 and B27 supplements, 2 mM GlutaMAX, 50 g/ml penicillin, and 50 g/ml streptomycin, incubated at 37°C and 5% CO 2 in a humidified incubator, and cultured for 5 days in vitro. Glial cells were NMDAR Blockade Prevents ZIKV-Induced Neuronal Death ® cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 1% streptomycin, and 0.3% amphotericin B and placed in 75-cm 2 disposable culture bottles, prepared by coating with 2% gelatin, followed for 14 days of incubation at 37°C and 5% CO 2 in a humidified incubator. Culture medium was replenishment every 4 days, and glial cells were plated onto poly-L-ornithine plates. Infection was performed with ZIKV (MOI of 1), followed by an adsorption period of 1 h. Residual virus was removed and titrated (represented by a dashed line on graphs). Wells were then washed twice with incomplete medium. Each well was replaced by a final volume of complete neurobasal medium in the presence or absence of the NMDAR antagonists. Titers are representative of virus accumulation over the whole period of infection, collected at different time points. Beside the kinetic experiments, all experiments evaluating the effects of NMDAR blockers on primary neurons were performed after 72 h of ZIKV infection. In some experiments, neuronal cultures were treated with 30 M memantine every 24 h.
Cell viability by LIVE/DEAD. Neuronal cell death was determined using a LIVE/DEAD assay kit. Briefly, neurons were stained with 2.0 M calcein acetoxymethyl ester (AM) and 2.0 M ethidium homodimer 1 for 15 min and the fractions of live (calcein AM-positive) and dead (ethidium homodimer 1-positive) cells were determined. Neurons were visualized by fluorescence microscopy. N1E-115 cell viability was analyzed as described previously (69).
Titration of virus. Virus loads in culture supernatant and mouse tissues (optic nerve and brain) were assessed by plaque assay in Vero cells as described in reference 66. Results were measured as PFU per gram of tissue weight or milliliter of supernatant.
Statistical analysis. Results are shown as the mean Ϯ standard error (SEM), except for viral loads, which were expressed as the median. Body weight was converted to a percentage, and weight loss/gain was calculated by subtracting the basal levels (obtained prior to infection) from those of control and infected mice. Differences were compared using analysis of variance (ANOVA) followed by Student-Newman-Keuls post hoc analysis. All analyses were performed using the GraphPad PRISM software 5.0 (GraphPad Software, Inc., USA). Results with a P value of Ͻ0.05 were considered significant.