Blood samples (buffy coats) from healthy donors were obtained anonymously from the Hemotherapy Service from the Hospital Universitário Clementino Fraga Filho (HUCFF) of Universidade Federal do Rio de Janeiro (UFRJ). The study protocol was approved by the Experimental Ethics Committee of UFRJ (Permit Number: 105/07) and the review board waived the need for informed patient consent.

A129 (deficient of IFNARI) mice were obtained from the mice facility of the Instituto de Microbiologia, Universidade Federal do Rio de Janeiro (IMPPG, UFRJ), Brazil. The animals were bred and housed according to institutional policies for animal care and usage and the protocol was approved by The Ethics Committee of Animal Care and Use (Comite de Etica no Uso de Animais-CEUA) from Centro de Ciencias da Saude, UFRJ (Permit Number: no104/16).

Vero cells were cultured in DMEM supplemented with L-glutamine and 5% fetal bovine serum (Life Technologies, Grand Island, NY) and maintained at 37°C with 5% CO₂. C6/36 mosquito cell line were cultured at 28°C in Leibovitz (L-15) medium (Life Technologies) supplemented with 10% of tryptose phosphate broth, 0.75% sodium bicarbonate, 0.2% of L-glutamine (Sigma-Aldrich, St Louis, MO), and 10% FBS (Life Technologies). Human brain microvascular endothelial cells (HBMEC) have been previously described (Nikolskaia et al., 2006) and were kindly given by Dr. Julio Scharfstein (Instituto de Biofísica Carlos Chagas Filho, UFRJ). The cells were cultured in M199, supplemented with L-glutamine, non-essential aminoacids, and 10% FBS (Life Technologies). Peripheral blood mononuclear cells (PBMC) were obtained after centrifugation of buffy coats samples over ficoll-hypaque gradients and cultured with RPMI supplemented with L-glutamine, and 10% FBS.

ZIKV strain MR766 (ZIKV_MR766; ATCC VR1838) was propagated in Vero or C6/36 cells, as indicated in each experiment. ZIKV_PE243 (gene bank ref. number KX197192) was isolated from a febrile case in the state of Pernambuco, Brazil, and was kindly given by Dr. Ernesto T.A. Marques Jr. (Centro de Pesquisas Aggeu Magalhães, FIOCRUZ, PE, Brazil; and Center for Vaccine Research, University of Pittsburgh, PA). Viruses were propagated in C6/36 cells and sequence analysis was performed after six passages. In some experiments, viruses were also propagated in Vero cells, as indicated in each experiment. ZIKV_BR−SP (gene bank ref. number KU497555; kindly given by Pedro Vasconcelos, Instituto Evandro Chagas, FIOCRUZ, PA, Brazil) was isolated from a mild ZIKV case, in the state of Paraíba, Brazil, and it was distributed as part of ZIKA FAPESP NETWORK after four passages in C6/36 cells (Faria et al., 2016). Viral titers were determined by plaque assay in Vero cells, as previously described (Coelho et al., 2017) and are indicated as PFU/ml. Supernatants of non-infected C6/36 or Vero cells cultured in the same conditions were used as mock controls. Inactivated virus (iZIKV) was obtained after U.V. exposition for 2 h and the inactivation was confirmed by qRT-PCR in Vero cells.

HBMECs were infected with ZIKV_MR766 or ZIKV_PE243 with a MOI of 1 for 2 h, at 37°C in 5% CO₂ atmosphere, for virus adsorption. As a control, the cells were incubated with supernatant of non-infected Vero or C6/36 cells (mock-infected). Cells were, then, washed with PBS and cultured with complete medium. After different time points, cells and supernatants were harvested and virus infection, cell survival, and cytokine secretion were evaluated as described below.

PBMCs were infected at the same conditions and, after 48 h, supernatants were harvested, and stored at −80°C. Supernatants from PBMC treated with mock, or infected with ZIKV_PE243 or ZIKV_MR766 were inactivated by U.V. radiation for 2 h and used as conditioned medium (iCM-PBMC). Virus inactivation was confirmed by qRT-PCR and plaque assay after 48 h infection of Vero cells.

HBMECs were mock-treated or infected with ZIKV_PE243 or ZIKV_MR766 at different MOIs. After 48 hpi, cells were blocked, permeabilized, and stained with anti-flavivirus 4G2 antibody (ATCC HB112), followed by AlexaFluor488-conjugated anti-mouse IgG (Life Technologies). HBMEC infection was then analyzed by immunofluorescence, using OLYMPUS IX81 microscopy.

Viral replication was also analyzed by qRT-PCR. Cells were treated as described and, after different time points, cells, and supernatants were harvested and RNA was isolated using TRIZOL reagent (Life Technologies), according to the manufacturer's instructions. Treatment with DNAse I (Ambion, Thermo Fischer) was performed to prevent genomic DNA contamination and first strand cDNA was synthesized using High-Capacity cDNA Archive Kit (Applied Biosystems), according to the manufacturer's instructions. cDNAs were subjected to quantitative real-time PCR for detection of viral RNA using a StepOnePlus Real-time PCR system and Taqman Master Mix Reagents (Applied Biosystems), using primers and probe specific for protein E sequence, as previously described (Lanciotti et al., 2008). cDNA obtained from virus samples ranging from 75,000 to 0.75 PFU/ml were used to construct a standard curve for estimating the genome copy number of ZIKV (RNA equivalent).

To evaluate secretion of infectious viral particles by HBMECs, cells were infected as described and, after different time points, the supernatants were harvested and titrated by plaque assay, using Vero cell line.

HBMECs were infected with ZIK_PE243 or ZIKV_MR766 propagated in C6/36 cells and determination of cell viability after different time points was carried out using XTT 2,3-Bis-(2-Methoxy-4-Nitro-5-Sulfophenyl)-2H-Tetrazolium-5-Carboxanilide (XTT) (Sigma-Aldrich, St. Louis, MO). Cells were incubated with XTT solution for 2–4 h and metabolization was evaluated by spectrophotometry at 450 nm OD. One percent Triton X100 was used as a positive control.

Annexin V and propidium iodide (AnnV/PI) staining was also performed in HBMECs-infected cells. Briefly, HBMECs were mock treated or infected with ZIKV_PE243 or ZIKV_MR766 at a MOI of 1. After different time points, cells were stained with FITC-AnnexinV and 2.5 μg/ml of PI and analyzed by flow cytometry. Alternatively, cells were incubated with Muse Annexin V and Dead cell kit (Millipore) and were also analyzed by flow cytometry. Samples acquisition and analysis were performed using FACSCanto equipment (BD Biosciences) and Flow Jo software. In addition, supernatants were harvested and release of lactate dehydrogenase was measured by LDH assay, according to manufacturer's protocol (Bioclin, RJ, Brazil).

Plasma membrane integrity and apoptosis induction were also evaluated by PI and TUNEL staining, respectively, and fluorescence microscopy analysis. HBMECs were seeded on transwell inserts (Costar—Corning®) and infected with ZIKV_PE243 or ZIKV_MR766, as described. After 72 hpi, we performed TUNEL staining (In Situ Cell Death Kit—Roche), following manufacturer's instructions, or the cells were incubated with Propidium Iodide (BD Biosciences). The fluorescence was evaluated using OLYMPUS IX81 microscopy. Staurosporin (50 μM; Sigma- Aldrich) and Triton X-100 (0.1%; J.T. Baker) were used as positive controls for TUNEL and PI staining, respectively.

HBMECs were infected as described and, the indicated time points, supernatants were harvested and cytokine concentrations were measured using a human cytokine 27-plex multiplex assay (Bio-Plex kit) and Bio-Plex® MAGPIX™ Multiplex equipment (Bio-Rad). The secretion of IL-6, IL-8, CCL5, and CXCL10 were confirmed by ELISA, according to manufacturer's protocol (Peprotech).

IFN-β, IFN-λ1, and IFN-λ4 expression in the cell lysates were measured at different time points post infection by qRT-PCR, and GAPDH expression was measured as a housekeeping control gene. RNA was extracted using TRIZOL reagent (Life Technologies), and first strand cDNA was synthesized using High-Capacity cDNA Archive Kit (Applied Biosystems), according to the manufacturer's instructions. Expression of type I and type III IFN mRNAs were measured using SYBR Green (Applied Biosystems), using the following primers: IFN-β sense: 5′-TAG CAC TGG CTG GAA TGA GA-3′, IFN-β antisense 5′-TCC TTG GCC TTC AGG TAA TG-3′; IFN-λ1 sense: 5′-GGG AAG CAG TTG CGA TTT AG-3′ and IFN-λ1 antisense 5′-GAT TTG AAC CTG CCA ATG TG-3′; IFN-λ4 sense: 5′-AGG GTC CTT AAC CGA CTG TG-3′ and IFN-λ4 antisense 5′-AAA CAA CCA ATG CGA TCA AA-3′; GAPDH sense 5′-GTG GAC CTG ACC TGC CGT CT-3′ and GAPDH antisense 5′-GGA GGA GTG GGT GTC GCT GT-3′. All qRT-PCR were performed with a standard PCR: the samples were subjected to 50°C for 2 min, 95°C for 10 min and 40 cycles of denaturation (95°C, 15 s), primer annealing (55°C, 30 s), and primer extension (60°C, 1 min). Next, the samples were subjected to a melt curve to eliminate primer dimers: 95°C, 15 s; 60°C, 1 min and 95°C, 15 s. Comparative CT method (ΔΔ Ct) was used to quantify gene expression levels with GAPDH used for normalization. Results are expressed as Mean ± SD. Kruskal–Wallis test One-way ANOVA was employed to compare differences between expressions of target genes with a significance level of 0.05.

HBMECs (5 × 10⁴ cells/well) were seed onto transwell inserts (Corning Costar, ME, USA; 0.4 μm membrane) and confluence was monitored everyday by measuring transendothelial electrical resistance (TEER) across cell monolayers using a Voltohmmeter (Millicell ERS-2). TEER was calculated after subtracting the resistance value in each experimental situation by the blank resistance of the membrane (without cells), and considering that resistance is inversely proportional to the area of the membrane. Resistance values were reported as “Ω/cm²” and the experiments were performed when a high resistance (>80 Ω/cm²) was reached (Mahad et al., 2006; Srinivasan et al., 2015). Cells were infected, from the apical side, with ZIKV_PE243 or ZIKV_MR766 (propagated in C6/36 cells) with a MOI of 1. As negative controls, cells were cultured with mock supernatants obtained from C6/36 cells. Staurosporin (STS; 10 μM; Sigma-Aldrich) was used as positive control. In some experiments, after virus adsorption, the cells were treated with Chloroquine diphosphate (50 μM; kindly supplied by FarManguinhos, Fiocruz, Rio de Janeiro, Brazil), or Nystatin (10 μM; Sigma-Aldrich), or Brefeldin A (BFA; 2 μg/ml; eBiosciences, San Diego, CA). After 72 h, supernatants were harvested and cells were incubated with FITC-conjugated BSA for 30 min. BSA extravasation was evaluated by measuring fluorescence intensity in the lower chamber, using spectrofotometer SpectraMax i3 (Molecular Devices, Lagerhausstrasse, Austria). The Permeability Coefficient (Pd) of albumin was calculated as: Pd = [A]/t × 1/A × V/[L], where [A] is the albumin concentration in lower chamber, t refers to time in seconds, A indicates the area of the membrane (in cm²), V is the volume of the bottom chamber, [L] is the albumin concentration in upper chamber. Data was normalized in relation to cells culture in culture medium only.

Virus RNA was measured in the upper and lower chamber of the transwell by qRT-PCR, as previously described. Also, the medium harvested from the lower chamber was inoculated into a Vero cell culture and virus RNA in the cell lysates and supernatants were measured by qRT-PCR. To confirm the presence of infectious virus in the abluminal chamber, Vero cells (8 × 10⁴ cells/well) were seeded in the lower chamber and, at 48 hpi, a real time PCR targeting viral negative strand was performed with the cell lysates. Briefly, intracellular viral RNA was extracted as described, and reverse transcription was performed using 835 forward primer instead of random primers (Lanciotti et al., 2008). Real time PCR was performed using TaqMan Universal Master Mix, as described.

In some experiments, HBMECs were also cultured with 50% conditioned medium obtained from PBMCs mock-treated or infected with ZIKV_PE243 or ZIKV_MR766 (iCM-PBMC) and cell permeability was evaluated as described.

HBMECs were cultured onto coverslips and cells were mock treated or infected as described. After 72 hpi, cells were fixed with 4% paraformaldehyde in PBS for 20 min, washed in PBS twice and permeabilized with 0.1% Triton X-100 (Sigma Aldrich) plus 3% bovine serum albumine (BSA—Sigma Aldrich) for 25 min. Cells were incubated with both an anti-flavivirus 4G2 antibody (ATCC HB112) and anti-ß-catenin (Sigma Aldrich) antibodies diluted in PBS−3% BSA overnight at 4°C. After washing in PBS, cells were incubated with the AlexaFluor488-conjugated anti-mouse IgG (4G2 staining), or with AlexaFluor594-conjugated goat anti-rabbit IgG (β-catenin staining) (Life technologies) for 40 min. The cells were then washed three times in PBS and incubated with DAPI for 5 min. Following thorough washing with PBS, the coverslips were mounted with prolong gold antifade reagent (Life technologies) and imaged on a Zeiss LSM 710 confocal.

A129 mice (4 weeks age) were infected with 2 × 10⁵ PFU of ZIKV_PE243 or ZIKV_MR766 by i.v route. Mock supernatants were used as negative control. To evaluate virus replication, mice brains were removed at 2 or 5 days p.i. and macerated. RNA isolation, cDNA, and qRT-PCR were performed as described previously.

A129 mice were infected with ZIKV_PE243 or ZIKV_MR766, as described. Two or five days post infection, mice were i.v. injected with 0.5% Evans blue solution (EB, 200 μL per mouse) (Vetec, Rio de Janeiro, BR). After 1 h, mice were perfused with PBS and their brains were carefully removed. Brains were weighted, placed in 1 mL formamide (Vetec) and kept at room temperature for 3 days for stain extraction. As a positive control for BBB permeabilization, we used C57BL/6 mice infected with Plasmodium berguei ANKA, an experimental model of cerebral malaria (Reis et al., 2012). Mice were intraperitoneally inoculated with 10⁵ infected erythrocytes and BBB permeability was assayed 7 days after infection. The amount of Evans blue in solution was measured by optical spectroscopy using SpectraMax i3 (Molecular Devices), and calculated using a standard curve. Blood brain barrier permeability was estimated as μg Evans Blue/mg tissue.

A129 mice were infected with ZIKV_PE243 or ZIKV_MR766, as described. Two or five days post infection, mice were deeply anesthetized and then transcardially perfused with ice-cold 0.9% saline, followed by 4% paraformaldehyde (PFA) in 0.1 M sodium phosphate buffer, pH 7.4. Brains were rapidly removed from skulls, postfixed in PFA for 1 d at 4°C, and cryoprotected in a PFA solution containing 20% (w/v) sucrose overnight. The frozen brains were then sectioned into 20 μm-thick coronal sections using a sliding microtome (Leica). Slices were collected in a cold cryoprotectant solution (0.05 M sodium phosphate buffer, pH 7.4, 30% ethylene glycol, 20% glycerol) and stored at −20°C. Free-floating sections were washed with 0.4% Triton X-100 in PBS (3 × 10 min) and then incubated for 30 min in a blocking solution containing 4% normal goat serum (Thermo Fisher Scientific) in PBS. Sections were washed with PBS (3 × 10 min), followed by an overnight incubation with a biotinylated goat anti-mouse IgG (H+L) antibody (1:500, Vector Laboratories). Binding was visualized using the peroxidase-based Vectastain ABC kit and 3,3′-diaminobenzidine (Vector Laboratories). Tissues were thereafter dehydrated through graded concentrations of alcohol, cleared in HistoChoice®Clearing agent (Sigma-Aldrich) and coverslipped with Organo/Limonene Mount™ (Sigma-Aldrich). Slides were scanned with a Pannoramic MIDI II scanner (3DHISTECH).

A129 mice were infected with ZIKV_PE243, as described. Five days post infection, mice were transcardially perfused with cold phosphate-buffered saline (PBS) solution followed by fresh ice-cold 4% formaldehyde (PFA). The brain samples were fixed in 4% PFA, and cryoprotected with sucrose 30%. Slides with coronal frozen brain sections (30 μm-thick) were fixed in acetone for 30 min, washed twice with PBS and slides were blocked with PBS supplemented with 10% FBS and 1% NDS (normal donkey serum) (blocking buffer), for 1 h. The tissues were then incubated with mouse antibodies anti-4G2 antibody, or anti-VE-cadherin (2 μg/ml; Santa Cruz Biotechnology), or anti-occludin (2.5 μg/ml; Invitrogen), all diluted in blocking buffer. Primary antibodies were incubated overnight, at 4°C, in humid chamber. After washing in PBS, cells were incubated with Alexa Fluor 488-conjugated anti-mouse IgG diluted in blocking buffer (2.5 μg/ml; Invitrogen), for 1 h, at 4°C, in humid chamber. Then, the slides were mounted in Prolong Gold Antifade with DAPI (Invitrogen) and imaged on a Zeiss Axio Observer Z1 microscope equipped with an Apotome module.

Data were analyzed using the GraphPad Prism software (GraphPad Software, San Diego, CA, USA). Comparisons among groups were performed by two way ANOVA; p < 0.05 were considered statistically significant.

Human brain microvascular cells were previously shown to be susceptible to ZIKV (Bayer et al., 2016); however, a productive infection was not fully addressed. We further investigated HBMEC infection by different strategies. Initially, cells were infected with different MOIs of ZIKV_PE243 or ZIKV_MR766 and, after 48 h, ZIKV antigen expression was analyzed by immunofluorescence, using anti-flavivirus 4G2 antibody. We observed that virus antigen was clearly detected after infection with either ZIKV strain, with 10 or 1 MOI (Figure 1A). The replication efficiency of ZIKV_PE243 and ZIKV_MR766, obtained from C6/36 or Vero cell lines, were then compared by performing kinetic measurements of virus RNA and infectious particles release. We observed that infection with both strains resulted in similar level of virus RNA release, independent of the virus source (Figures 1B,C). Analysis of the production of infectious particles by plaque assay confirmed that HBMEC were permissive to infection with both ZIKV strains (Figures 1D,E). Infection with viruses produced in Vero cells resulted in slightly increased replication of ZIKV_MR766; in addition, a significant drop in viral secretion was observed at 96 hpi (Figure 1E), which was not observed when cells were infected with viruses obtained from C6/36 cells (Figure 1D). Therefore, further investigation of the effects of ZIKV infection on HBMEC physiology was performed using C6/36-derived virus stocks.

We then evaluated whether infection with ZIKV would affect HBMEC physiology and survival. Initial analyses were performed by XTT metabolization assay at different time points post infection, normalized according to cell cultures with culture medium only. Infection with ZIKV_MR766 resulted in decreased XTT metabolization levels in comparison to mock-treated cells, from 48 hpi and thereafter, suggesting that HBMEC infection with this strain induced cell death. On the other hand, alteration in XTT metabolization by infection with ZIKV_PE243 was not detected until 72 hpi, indicating that the Brazilian strain was not associated to a severe cytophatic effect in these cells (Figure 2A). ZIKV-induced altered cell metabolism or viability depended on virus replication, since HBMEC culture with inactivated viruses did not result in any alteration of XTT metabolization. HBMECs cell death was also evaluated after infection with another Brazilian virus isolated (ZIKV_BR−SP), and the results obtained were very similar to the ones obtained with ZIKV_PE243 (Figure 2B).

Cell viability was also accessed by Annexin V/PI staining and LDH release. Flow cytometry analysis demonstrated that around 35% of cells infected with ZIKV_MR766 were AnnV/PI-positive after 72 hpi, in contrast to 20% of the cells infected with the Brazilian strain. About 70% of HBMECs were still alive even at 96 h after ZIKV_PE243 infection, when almost 60% of the cells infected with ZIKV_MR766 were dead (Figures 2C,D). This kinetic was consistent with the detection of LDH release, which was observed from 48 hpi; and peaked at 96 hpi (Figure 2E). Similarly, infection with ZIKV_MR766 resulted in higher levels of LDH activity detected in the supernatants, in comparison to infection with ZIKV_PE243.

To confirm that infection with ZIKV_PE243 was not associated to a remarkable cytopathic effect, HBMEC viability was also evaluated by different techniques, including flow cytometry analysis, based on 7AAD staining; fluorescence microscopy analysis, after PI and TUNEL staining; and cell counting with Trypan blue exclusion dye. Flow cytometry analysis demonstrated that dead cells could only be detected after 72 hpi with ZIKV_MR766, but not in mock-treated or ZIKV_PE243-infected cells (Supplementary Figure 1A). Fluorescence analysis did not indicate a significant TUNEL staining in any experimental situation, although PI staining could be detected in ZIKV infected cells (Supplementary Figure 1B). Finally, cell proliferation and viability was accessed by following the absolute cell numbers recovered after tripsinization of HBMECs infected with ZIKV_PE243 or ZIKV_MR766, in comparison to Vero cells. We observed that HBMECs infected with ZIKV_PE243 were still able to replicate during 48 h culture, although the absolute number of cells recovered after 48 h was a little lower (about 7–12%) than the mock-treated cells (Supplementary Figure 1C). HBMECs infected with ZIKV_MR766 also replicated, but at much lower efficiency. In contrast, the proportion of ZIKV-infected Vero cells recovered after 48 hpi was much lower (40–50%), in comparison to mock-treated cells (Supplementary Figure 1D). Taken together, these data corroborates with previous observation that HBMEC is more resistant to ZIKV-induced CPE than other cell types, such as Vero cells.

We have previously demonstrated that dengue virus stimulated HBMEC to produce increased levels of chemokines and proinflammatory cytokines, which might contribute to the inflammatory response observed during the disease (da Conceição et al., 2013). Here, we evaluated whether ZIKV was able to induce the same activation pattern. Supernatants of HBMECs infected with ZIKV_PE243 were harvested at different time points and levels of IL-6, IL-8, and CCL5 were measured by ELISA. We observed a significant increase in secretion of IL-6 and CCL5, but not IL-8, at 72 hpi, in comparison to mock treated cells (Figure 3A). To investigate whether HBMECs activation would be a unique feature of the Brazilian strains, we compared cytokine secretion induced by ZIKV_PE243 or ZIKV_MR766 by 27-plex multiplex analysis (Table 1). The cytokines that showed significant increased levels were confirmed by ELISA in additional experiments. ZIKV_PE243 and ZIKV_MR766 stimulated the secretion of IL-6 and CCL5 (Figures 3B,C), whereas CXCL10 secretion was not detected in any situation (data not shown). These data suggest that both viruses activated HBMECs, what may equally contribute to attraction of leukocytes, especially lymphocytes to BBB.

Expression of type I and type III interferons were also evaluated in infected cells by qRT-PCR. We observed that both virus strains stimulated IFN-β and IFN-λ1 production; however, ZIKV_MR766 promoted earlier and higher levels of IFN expression (Figures 3D,E). In addition, only ZIKV_MR766 produced IFN-λ4, although at low levels (Figure 3E). These data suggest that although ZIKV_MR766 induces increased CPE, virus replication was also associated to enhanced type I and III IFNs, what might contribute to the control of virus dissemination and endothelial lesion in vivo.

We finally analyzed whether ZIKV would alter BBB permeability. Cells were seed onto transwell inserts and infected with ZIKV_MR766 or ZIKV_PE243, with a MOI of 1. Mock supernatants and staurosporin (STS) were used as negative and positive controls, respectively. Transendothelial electrical resistance (TEER) was measured for 48 h and no significant differences were observed between ZIKV-infected and mock-treated cells (Figure 4A). After 72 hpi, the supernatants were harvested, cells were incubated with FITC-conjugated albumin for 30 min, and the fluorescence intensity was measured in the abluminal chamber. The permeability coefficient (Pd) was calculated and normalized in relation to cells cultures in culture medium only. We did not observe any significant difference in the levels of albumin extravasation at this time point, in comparison to mock-treated cells (Figure 4B).

Interestingly, in spite of the maintenance of endothelial monolayer integrity, virus RNA could be detected in the lower chamber of the transwell system, suggesting that ZIKV may cross BBB through basolateral virus release, transcytosis or paracytosis (Figure 4C). Conditioned medium obtained from the lower chamber of the transwell system was inoculated in ZIKV-susceptible Vero cells and, after 48 hpi, ZIKV RNA was measured in Vero cell lysates and supernatants. Intracellular and supernatant RNA virus were detected in the cultures, indicating that extravasated ZIKV was able to productively infect other cells (Figure 4D). To confirm the infectivity of the viruses crossing the monolayer, Vero cells were seeded in the lower chamber of the transwell system and the presence of virus negative strand RNA was accessed by qRT-PCR. As a negative and positive control, Vero cells were mock treated or directly infected with ZIKV. Virus negative strand RNA was detected in the Vero cell lysates, confirming that the virus crossed the HBMEC barrier, and replicated into the cells in the lower chamber (Figure 4E). These results confirm that ZIKV may cross endothelial cell monolayer without increasing permeability. To further evaluate endothelial monolayer integrity in the cultures, cells were stained with anti-flavivirus 4G2 antibody and anti-β-catenin. Immunofluorescence analysis did not indicate any cell junction disorganization in ZIKV-infected cultures, in comparison to mock-treated cells (Figure 4F).

We then asked whether inflammatory mediators produced by blood cells would be able to affect the permeability of ZIKV-infected HBMECs. PBMC were infected with either ZIKV_PE243 or ZIKV_MR766 for 72 h. Then, supernatants were harvested, and virus infectious particles were inactivated by U.V. radiation and used as PBMC conditioned medium (iCM-PBMC). HBMECs were mock treated or infected with ZIKV_PE243 or ZIKV_MR766, in the presence of absence of 50% the indicated iCM-PBMC. After 72 hpi, albumin extravasation was measured, as described before. Surprisingly, neither the supernatant obtained from ZIKV_MR766 nor ZIKV_PE243-infected cells were able to induce HBMECs permeability in any tested situation (Figure 5). These data indicate that rupture of BBB integrity is not the main pathway stimulated by ZIKV to invade the CNS.

Since virus extravasation was not related to cell death-mediated monolayer disruption, we investigated whether it was dependent on virus replication or transcytosis, by treating ZIKV-infected HBMECs with different pharmacological inhibitors of these pathways. Chloroquine is a weak base able to raise the pH of acidic compartments, which was previously shown to inhibit ZIKV-replication (Delvecchio et al., 2016); nystatin was reported to inhibit virus-induced transcytosis through caveola-mediated pathways (Harmon et al., 2012; Tugizov et al., 2013; Chanthick et al., 2016); brefeldin A (BFA) modulates vesicle traffic from endoplasmic reticulum to Golgi, leading to exocytosis inhibition (Fujiwara et al., 1988). Initially, we analyzed if the drugs would affect cell survival themselves, and none of them were cytotoxic at the concentrations used (Figure 6A). To investigate whether this drugs would affect virus extravasation, cells were cultured onto transwell plates, infected, and treated with them. HBMEC TEER was not altered by any drug during 48 h culture, confirming our previous data that ZIKV does not disturb cell resistance and demonstrating that the drugs did not induce cell permeability themselves (Figure 6B).

Finally, we investigated the effect of each drug in ZIKV replication and extravasation through HBMECs monolayers. Cells were infected with ZIKV_PE243, from the apical side, and treated with chloroquine, nystatin or brefelding A. The medium obtained from the luminal (upper) and abluminal (lower) chambers were harvested after 48 hpi, and virus RNA in each compartment was measured by qRT-PCR (Figures 6C,D). As expected, analysis of apical release of virus RNA demonstrated that chloroquine and BFA inhibited around 50% ZIKV infection. On the other hand, addition of nystatin did not decrease virus RNA release (Figure 6C). Treatment of ZIKV-infected HBMEC with chloroquine or BFA also resulted in around 50 and 65% decrease of virus RNA in the abluminal transwell chamber, respectively, indicating that virus replication and exocytosis or basolateral release were necessary for extravasation (Figure 6D). Interestingly, despite having no effect in virus replication and apical release, addition of nystatin to the cultures also decreased virus extravasation by around 50% (Figure 6D), suggesting that caveola-mediated transcytosis and/or basolateral release is an important pathway for ZIKV extravasation through brain endothelial cells.

To confirm that ZIKV was able to reach the brain without disrupting BBB, we evaluated BBB integrity in a mouse experimental model. For this purpose, we infected IFNAR-deficient A129 mice, which was previously demonstrated to be susceptible to ZIKV infection (Dowall et al., 2016; Lazear et al., 2016; Rossi et al., 2016). Mice were intravenously inoculated with ZIKV_PE243 or ZIKV_MR766 (2 × 10⁵ PFU). After 2 and 5 days p.i., virus RNA was quantified in the mice brains by qRT-PCR and BBB integrity was accessed by Evans Blue extravasation assay and measurement of endogenous IgG in the brains. ZIKV genome was highly detected in all the brains investigated, from mice infected with either ZIKV_PE243 or ZIKV_MR766, at 2 and 5 dpi (Figure 7A). BBB integrity was initially analyzed by inoculating Evans blue dye (EB), i.v., after 2 and 5 dpi. After 1 h, the brains were harvested and the amount of EB per mg tissue was measured. We could not detect a significant amount of EB in the brains obtained from ZIKV-infected mice, in comparison to mock-treated ones (Supplementary Figure 2). Since recent studies have questioned the sensibility of this method (Saunders et al., 2015), we performed another set of experiments, in which the presence of endogenous IgG was histologically evaluated in the brain sections. Corroborating with the previous data, no staining could be detected at 2 dpi in any of the experimental groups (Figure 7B). At 5 dpi, however, a diffuse staining pattern was observed in 1 out of 3 animals infected with ZIKV_MR766 (Figure 7B). In addition, small, focal areas of immunoglobulin G staining were detected in another animal infected with ZIKV_MR766 and in 3 out of 4 animals infected with ZIKV_PE243. No staining for immunoglobulin G was seen in the brain of control mice. These data supports the hypothesis that ZIKV is able to reach the CNS, without disrupting the BBB; however, blood brain barrier disruption may be a delayed event following ZIKV infection in young A129 mice.

Immunostaining for ZIKV envelope protein in the brain sections revealed robust staining in blood vessels (Figure 7C) and in the cells of plexus choroid of subventricular zone (SVZ) (Figure 7D), indicating that brain endothelial cells are actually infected by ZIKV. We also performed immunostaining of VE-cadherin and occludin proteins and we found no changes in the expression profile of these adherens and tight junctions markers in the brains of ZIKV-infected mice (Figures 7E,F), further suggesting that BBB was not remarkably affected.

Taken together, our data indicate that ZIKV productively infects HBMECs, and induces cellular activation. However, infection did not induce BBB permeability, in vitro nor in vivo, suggesting that ZIKV is able to reach the CNS without disrupting the BBB and that the invasion of CNS by ZIKV might be associated to transinfection or transcytosis from the endothelial cells of the BBB. However, the persistent CPE caused by the presence of ZIKV_MR766 and/or the subsequent inflammation triggered by the replication either virus strain in the brains may induce a slight disruption of BBB at later time points.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer KS and handling Editor declared their shared affiliation.