Blockade of dengue virus infection and viral cytotoxicity in neuronal cells in vitro and in vivo by targeting endocytic pathways

Dengue virus (DENV) infection in neuronal cells was speculated to trigger neuropathy. Herein, we determined the blockade of DENV infection by targeting endocytic pathways in vitro and in vivo. In DENV-infected mouse brains, we previously showed that viral proteins are expressed in neuronal cells around the hippocampus with accompanying neurotoxicity. DENV caused infection, including entry, double-stranded (ds)RNA replication, protein expression, and virus release, followed by cytotoxicity in the mouse neuronal Neuro-2a cell line. Pharmacologically blocking clathrin-mediated endocytosis of the DENV retarded viral replication. Targeting vacuolar-type H+-ATPase (V-ATPase)-based endosomal acidification effectively blocked the DENV replication process, but had no direct effect on viral translation. Blockade of the clathrin- and V-ATPase-based endocytic pathways also attenuated DENV-induced neurotoxicity. Inhibiting endosomal acidification effectively retarded DENV infection, acute viral encephalitis, and mortality. These results demonstrate that clathrin mediated endocytosis of DENV followed by endosomal acidification-dependent viral replication in neuronal cells, which can lead to neurotoxicity.

neurological dysfunction, and multiple organ failure 1,2 . Although DENV is not a characteristic neurotropic virus, dengue is widely considered to be one of the leading causes of neurological manifestations, including encephalitis, encephalopathy, dengue-associated neuromuscular complications, and dengue-associated neuro-ophthalmic complications [10][11][12] . However, the pathogenesis of dengue-associated neurological complications is not fully understood due to a lack of appropriate evaluations in previous studies. A retrospective 2-year study reported that dengue patients may show various persistent neurological complications 13 . DENV is rapidly replicated and causes viremia during the acute phase of infection, and it was speculated that acute viral encephalitis and central nervous system (CNS) inflammation may facilitate dengue-associated neurological complications.
Intracranial infection with non-adapted DENV induces lethality in immunocompetent mice following limb paralysis, seizures, and encephalitis [14][15][16] . Expressions of viral antigens, double-stranded (ds)RNA, and virus particles are found in DENV-infected brains. Importantly, in vivo DENV infection in mouse brains causes neuronal damage in the pyramidal layer of Cornu Ammonis (CA) areas of the hippocampus 14,16 . Based on viral antigen identification, neurons can be infected by DENV; however, viral receptors and infectious routes have not been well addressed. The targeting of dopamine receptors, which are expressed in hippocampal neurons 17 , retarded DENV infection in vitro 18,19 and in vivo 19 . Studies showed induction of neuronal cell death, particularly apoptosis, by DENV infection both in vitro [20][21][22][23] and in vivo 14,16,24 . This study attempted to investigate the infectious route of DENV in neuronal cells and evaluate potential antiviral strategies by targeting endocytic pathways of DENV infection in vitro and in vivo.

DENV initiates infection in neuronal cells in vitro, including viral entry, RNA replication, protein expression, and viral release.
In DENV-infected brains of 7-day ICR mice 16 , we noted that NS3-positive cells were present within the Iba-1-negative cell population, suggesting the ability of DENV to infect non-microglia in vivo. Previous work 15,24 showed that DENV can infect neuronal cells in vivo; however, the effects of DENV on neuronal cells remain poorly understood. To demonstrate infection efficacy in Neuro-2a cells, we performed fluorescent DENV staining followed by fluorescent imaging (Fig. 1A) and a flow cytometric analysis (Fig. 1B). Results showed viral binding/entry at 2 h post-inoculation. To confirm this finding, we used confocal microscopy to evaluate the intracellular localization of fluorescence-stained DENV in Neuro-2a cells (Fig. 1C). To investigate DENV replication in neuronal cells, we used immunostaining, which demonstrated significant viral dsRNA expression in DENV-infected Neuro-2a cells (p < 0.05, Fig. 1D). We also performed Western blotting (Fig. 1E) and a quantitative polymerase chain reaction (qPCR) (Fig. 1F). Our results confirmed viral NS1 protein expression in Neuro-2a cells 24 h post-infection. A plaque assay, which was performed to determine viral replication and release, showed significant (p < 0.05) infection of Neuro-2a cells by DENV (Fig. 1G).
Following infection, DENV can cause neurotoxicity as shown by histopathological changes and apoptotic staining 16 . Changes in cell morphology, cell growth, and cytotoxicity were used to evaluate the effects of DENV on neuronal cells. Microscopic observations revealed a significant change in Neuro-2a cells after DENV infection (Fig. 1H). The MTT (Fig. 1I) and lactate dehydrogenase (LDH; Fig. 1J) assays showed that DENV caused cell growth inhibition and cytotoxicity, particularly when cells were incubated with a high multiplicity of infection (MOI). These results indicated that DENV can infect neuronal cells and cause neurotoxicity in vitro.

D2R mediates DENV2 infection in Neuro-2a cells.
Currently, the targeting of dopamine receptors retards DENV infection in vitro 18,19 and in vivo 19 , suggesting a potential role of dopamine receptor-mediated DENV infection. We performed immunostaining that revealed expression of dopamine receptor D2 (D2R), but not D4R, in Neruo-2a cells (Supplemental Fig. 1A). Additionally, immunostaining showed the expression of D2R in isolated NeuN-positive hippocampal neurons, a neuronal nuclear antigen that is commonly used as a biomarker for neurons (Supplemental Fig. 1B). Fluorescent DENV staining followed by fluorescent imaging showed significant viral binding/entry in isolated hippocampal neurons (Supplemental Fig. 1C). Plaque assays confirmed that significant DENV replication occurred in isolated hippocampal neurons (Supplemental Fig. 1D).
To verify the essential role of D2R in mediating DENV infection, we used metoclopramide (MCP) to pharmacologically decrease DENV binding/entry (Supplemental Fig. 1E) and viral replication (Supplemental Fig. 1F) in Neuro-2a cells, as respectively shown by fluorescent imaging and plaque assays. These findings indicated that DENV caused infection of neuronal cells in a D2R-mediated manner.

DENV2 infects Neuro-2a cells via clathrin-mediated endocytosis. DENV infects cells via distinct
entry pathways for DENV internalization, including clathrin-mediated and clathrin-independent endocytosis 3,25,26 . However, no further evidence has revealed the endocytic pathway of DENV in neuronal cells. To investigate the involvement of (clathrin-mediated) endocytosis, the pharmacological inhibitors, CPZ and Pitstop 2, were utilized as previously described 4 . We performed fluorescent DENV staining followed by fluorescent imaging (Fig. 2A) and a flow cytometric analysis (Fig. 2B). Treatment with CPZ and Pitstop 2 had significantly reduced viral binding/entry in Neuro-2a cells at 2 h post-inoculation. In confirmation of this finding, immunostaining demonstrated a significant decrease in viral dsRNA expression in DENV-infected Neuro-2a cells co-treated with CPZ and Pitstop 2 (Fig. 2C). We also performed plaque assays to study viral replication and release, and results showed significant (p < 0.01) inhibition in DENV-infected Neuro-2a cells (Fig. 2D). Treatment with CPZ and Pitstop 2 did not affect cell growth or cytotoxicity, but they caused an inhibition of acidification following DENV infection. These results show clathrin-regulated DENV binding/entry followed by viral replication in neuronal cells in vitro.

Blockade of endosomal acidification attenuates DENV2 infection in Neuro-2a cells.
Once the dengue viral receptor mediates endocytosis, V-ATPase facilitates endosomal acidification for an uncoating of the viral genome 3 . A previous study showed that the dengue prM protein interacts with V-ATPase to facilitate viral entry and egression 27 . We used BafA1 and ConA, which are inhibitors of V-ATPase 28 , to decrease endosomal acidification. Immunostaining demonstrated a blockade of viral dsRNA expression in DENV-infected Neuro-2a cells pretreated with BafA1 (p < 0.05, Fig. 3A). We also performed Western blotting, a qPCR, and plaque assays to respectively confirm that inhibiting endosomal acidification reduced viral NS1 protein expression (  All quantitative data are shown as the mean ± SD of three independent experiments. *p < 0.05, **p < 0.01, and ***p < 0.001. ns, not significant. and NH 4 Cl, reduced acidification. These results indicate that inhibiting endosomal acidification attenuated DENV infection in neuronal cells.

Inhibition of endosomal acidification reduces DENV2 infection independent of translation.
We confirmed the inhibitory role of endosomal acidification only in viral genome uncoating but not in other steps of the viral cell cycle, by using firefly luciferase activity in BHK-D2-Fluc-SGR-Neo-1 cells, and found that treatment with BafA1 caused no direct inhibitory effects on viral translation or replication (Fig. 4A) or cytotoxicity in cells (Fig. 4B). However, consistent with results in Neuro-2a cells, BafA1 caused blockade of DENV2 replication in parental BHK-21 cells, as shown by plaque assays (Fig. 4C). These results indicate that the blockade of endosomal acidification had no direct effects on viral inhibition via alterations of viral translation and replication.

Targeting endocytic pathways reduces in vitro neurotoxicity induced by DENV infection.
Infection with DENV causes neuronal cell apoptosis [20][21][22][23] . To analyze roles of endocytic pathways involved in DENV infection in neuronal cells, we showed that pretreatment with CPZ, Pitstop2 (for blocking clathrin-mediated endocytosis), BafA1, and ConA (for inhibiting acidification) effectively abolished DENV-induced changes in cell morphology (Fig. 5A), cell growth inhibition (Fig. 5B), and cytotoxicity (Fig. 5C). Treatment with MCP (for inhibiting D2R) was used as a positive control, as D2R mediates DENV binding/entry. These results indicate the involvement of endocytic pathways in DENV-induced neurotoxicity in vitro.
Inhibiting endosomal acidification decreases DENV infection, neural impairment, and mortality in suckling mice. Our previous animal model of DENV infection used immunocompetent mice of the ICR strain to induce viral replication in the brain causing acute encephalitis 16 . BafA1 was administered at days 0 and 1 post-infection. We exploited a series of methods, including a Western blot analysis of the NS1 and NS4B viral proteins (Fig. 6A) and plaque assays for detecting virus replication (Fig. 6B). According to our results, DENV caused significant infection and replication in mouse brains at 7 days post-infection, as BafA1 inhibited viral protein expression and replication. Mice with neurological changes were then evaluated as previously described 16 . We monitored time-dependent changes in clinical scores, which were graded according to the severity of illness as follows: 0 for healthy; 1 for minor illness, including weight loss, reduced mobility, and a hunchback body orientation; 2 for limbic seizures; 3 for moving with difficulty and anterior limb or posterior limb weakness; 4 for paralysis; and 5 for death. A significant increase in clinical scores (Fig. 6C) had occurred in DENV-infected mice compared to mock-infected mice by 7 days post-infection. The survival rate of DENV-infected mice had decreased by days 8 or 9 post-infection, and all of the mice had died by days 9 or 10 post-infection (Fig. 6D). Either co-or post-treatment with BafA1 significantly reduced DENV-induced disease progression and mortality. The data indicated that inhibiting endosomal acidification abolished encephalitic DENV infection in our model, leading to neural impairment following viral replication.

Discussion
Neurological complications of DENV infection are now classified as some of the hallmarks of severe dengue. In addition to our in vitro study showing DENV-induced MOI-and time-dependent neurotoxicity, our previous study indicated that DENV caused in vivo infection in neuronal cells among hippocampal lesions and induced neuronal cell apoptosis 16 . We further showed that a clathrin-regulated endocytic pathway controls viral entry at an early step of infection in neuronal cells. Following endocytosis, lysosomal acidification is essential for DENV genome uncoating and replication in infected neuronal cells. Regarding no further studies showing the blockade of endocytic pathways for anti-DENV actions in mice, this study demonstrated that inhibiting the endocytic pathways of DENV infection decreased viral replication and attenuated DENV-induced neurotoxicity as well as acute viral encephalitis.
In general, neurological complications associated with dengue diseases are unusual. Treatment with PCZ, a D2R antagonist that has been approved for treating nausea, vomiting, and headaches in humans, confers anti-dengue effects in vitro and in vivo 19 . As to viral entry, PCZ can also act as an inhibitor of clathrin-mediated endocytosis. The roles of PCZ in targeting D2R and clathrin during DENV infection need further investigation. In addition to PCZ, the anti-psychotics, CPZ and Pitstop 2, were used in this study to confirm the essential role of clathrin-mediated endocytosis in DENV entry. CPZ is also an antagonist of D2R 29 . We showed that D2R was expressed in DENV-infected neuronal cells and mediated DENV entry as a viral receptor. All of these studies indicated that neuronal cells can be targets of DENV infection through potential viral receptor D2R-mediated and clathrin-regulated viral entry. Serotype 2 of DENV was used in our work as well as in a previous study 19 ; however, further studies have to dissect the different routes of clathrin-dependent and -independent endocytosis by other, different serotypes of DENV 4 . Importantly, according to in vivo and in vitro immunostaining, D2R was expressed in hippocampal neurons, consistent with a previous study 17 . The pathogenic effects of DENV-infected hippocampal neurons remain unclear in dengue encephalitis. The affinity of viral infection and neurotoxicity, especially in the hippocampal regions, is of interest for further studies.
In addition to developing vaccines and viral inhibitors that precisely target viral proteins, which are essential for viral binding/entry, replication, and assembly/release, identifying specific virus-host interactions, such as viral receptors, endocytic pathways, and viral assembly, could be useful for anti-infective therapies 3, 5, 6 . Although the search for antivirals to combat DENV infection is critical, there are no currently accepted antiviral drugs for treating dengue patients. Previous studies showed that the administration of chloroquine, a lysosomotropic agent, exerts a modest antiviral effect by interfering with endosomal fusion and furin-dependent virus maturation in vitro 30 and in vivo 31 . It was speculated that chloroquine and its analogue, hydroxychloroquine, could be used to treat low pH-dependent viruses, such as dengue, chikungunya, influenza, and Ebola, at the initial phase of infection 32,33 . However, a randomized controlled trial of chloroquine failed to inhibit viremia, antigenemia, and cytokine or T cell responses in dengue patients 34 . Further studies may be needed to evaluate its therapeutic efficacy and treatment route.
By small interfering RNA screening, several human membrane trafficking genes that mediate the infectious entry of DENV were identified 35,36 . Importantly, knockdown of V-ATPase reduced DENV infection and replication in both arthropod cells 36 and human cells 35 . Studies in arthropod vectors or cells showed blockade of the  [36][37][38] . According to our study, targeting V-ATPase-based endosomal acidification using BafA1 and ConA, which bind to the proteolipid ring of the V 0 domain of V-ATPase 28 , resulted in a notable antiviral response against DENV replication in vitro and in vivo. Consistent with a previous study as demonstrated in vitro 35 , our study further showed both in vitro and in vivo antiviral effects of BafA1 treatment. These findings provide evidence to strengthen the preclinical importance of BafA1-based anti-dengue therapy.
Viral infections require an acidic pH for infectivity, generally during the process of viral genome uncoating following endocytosis. BafA1 can block V-ATPase-based endosomal acidification, and the blockade of endocytic pathways by BafA1 treatment can be demonstrated by detecting viral genome uncoating, protein expression, replication, and virus release 39 . In our study, treatment with BafA1 significantly decreased DENV RNA replication, protein expression, and virus release in vitro and in vivo. These results confirmed the importance of endocytic pathways for DENV infection. However, the antiviral efficiency of BafA1 could also include targeting the autophagic process 40 , as autophagy-regulated energy production is required during DENV replication 41 . However, in this study, as demonstrated using a stable luciferase reporter DENV, BafA1 treatment did not affect viral translation, which suggests that its antiviral action presumably occurs by means other than by directly affecting viral replication-associated autophagy. Furthermore, V-ATPases are not only found within membranes of endosomes, lysosomes, and secretory vesicles, but are also found in plasma membranes 28 . It was speculated that BafA1 treatment may also affect upstream and downstream aspects of the endocytic pathways of DENV infection, such as viral entry or the secretion of infectious virions.
Neurotoxicity can be caused by DENV infection in brains of experimental mice 14,16,24 , and DENV infection has caused fatalities in dengue patients 42 . Several in vitro studies showed the ability of DENV infection to trigger neuronal cell death [20][21][22][23] . We confirmed the cytotoxic effects caused by DENV infection in neuronal cells in vivo and in vitro. After either a long time post-infection or a higher MOI, DENV caused neuronal cell growth inhibition accompanied by cytotoxicity. In response to viral replication, cellular stress, including endoplasmic reticular stress, oxidative stress, and cytotoxic factor release (such as tumor necrosis factor-α), may promote the apoptosis of neuronal cells. A direct effect of DENV infection on cytotoxicity in neuronal cells was speculated in vitro; however, a bystander effect such as CNS inflammation, which is probably caused by DENV infection of microglia and astrocytes, may also contribute to neurotoxicity in vivo.
In conclusion, targeting the endocytic pathways of DENV infection can reduce viral replication and cytotoxicity in neuronal cells in vitro and in vivo. These findings shed light on the development of antiviral therapeutics against DENV infection, particularly in the brain. In addition to DENV, targeting an Axl-mediated clathrin-based endocytic pathway conferred protection against the emerging flavivirus Zika infection in human glial cells 43 . Similar to flaviviruses, endosomal acid-dependent viral entry may also determine Zika membrane fusion during early infection and could be targeted for further antiviral therapy. For flaviviral encephalitis, in addition to a direct cytotoxic effect on neuronal cells, immunopathogenic issues, including innate and adaptive immune responses in the CNS, are also involved in encephalitic development 44  Isolation of neuronal cells. Hippocampal neurons were dissected from E16.5 C57BL/6 mouse embryos (Jackson Laboratory, Bar Harbor, MA) and suspended in 0.02% trypsin-EDTA (Invitrogen) at 37 °C for 10 min according to a previous study 45 .

DENV infection in vitro and in vivo.
The in vitro and in vivo infectious procedures were carried out according to our previous studies 16,46 . Seven-day-old ICR strain suckling mice were inoculated intracerebrally with 2.5 × 10 5 plaque-forming units (PFU) and intraperitoneally with 7.5 × 10 5 PFU of DENV2 (PL046), which was combined with or without BafA1 (1 mg/kg) treatment.