Zika virus cell tropism in the developing human brain and inhibition by azithromycin

Abstract

The rapid spread of Zika virus (ZIKV) and its association with abnormal brain development constitute a global health emergency. Congenital ZIKV infection produces a range of mild to severe pathologies, including microcephaly. To understand the pathophysiology of ZIKV infection, we used models of the developing brain that faithfully recapitulate the tissue architecture in early to midgestation. We identify the brain cell populations that are most susceptible to ZIKV infection in primary human tissue, provide evidence for a mechanism of viral entry, and show that a commonly used antibiotic protects cultured brain cells by reducing viral proliferation. In the brain, ZIKV preferentially infected neural stem cells, astrocytes, oligodendrocyte precursor cells, and microglia, whereas neurons were less susceptible to infection. These findings suggest mechanisms for microcephaly and other pathologic features of infants with congenital ZIKV infection that are not explained by neural stem cell infection alone, such as calcifications in the cortical plate. Furthermore, we find that blocking the glia-enriched putative viral entry receptor AXL reduced ZIKV infection of astrocytes in vitro, and genetic knockdown of AXL in a glial cell line nearly abolished infection. Finally, we evaluate 2,177 compounds, focusing on drugs safe in pregnancy. We show that the macrolide antibiotic azithromycin reduced viral proliferation and virus-induced cytopathic effects in glial cell lines and human astrocytes. Our characterization of infection in the developing human brain clarifies the pathogenesis of congenital ZIKV infection and provides the basis for investigating possible therapeutic strategies to safely alleviate or prevent the most severe consequences of the epidemic.

Keywords: Zika virus; azithromycin; cortical development; microcephaly.

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.

Tropism of ZIKV for radial glia in the developing human brain. Human cortical organotypic brain slices were infected with ZIKV-BR and cultured for 72 h. (A and B) Low-magnification overview of ZIKV infection detected by ENV (green) within the cortex. (A) ENV staining was analyzed with respect to region and cell type. CP, cortical plate; OSVZ, outer subventricular zone; SP, subplate; VZ, ventricular zone. (Scale bar, 100 μm.) (B) High magnification of A. Notably, ENV staining (arrowheads) appears to be preferentially enriched in the VZ and OSVZ. (Scale bars, 20 μM.) (C) Quantification of ENV+ cells by region (Top) and cell type (Bottom) at 13 to 14 pcw. n = 2; mean ± SD [SI Materials and Methods; an error bar is not shown where it is shorter than the line thickness (Top, CP; Bottom, SOX2)]. (D) Schematic summary of cell types observed to be susceptible to ZIKV infection (green) in the developing human brain during midneurogenesis. (E) High-magnification view of a ZIKV-infected radial glial cell in the OSVZ (arrow). oRG, outer radial glial. (Scale bar, 10 μm.) (F) Three-dimensional reconstruction of E, highlighting the intracellular presence of the ENV signal. (Scale bar, 10 μm.) (G) ENV and NS5 signal in OSVZ cells (arrowheads) suggested replicating ZIKV-PR. (Scale bars, 20 μm.) (H) Immature neurons (SATB2+, blue) infected with ZIKV (arrows). (Scale bar, 20 μm.) (I) Microglia (IBA1+) immunopositive for ENV. High magnification (Right) shows ENV+ microglia with amoeboid morphology (arrow), typical of activated microglia. (Scale bars, 10 μm.)

Fig. S1.

Characterization of ZIKV infection in cell culture. (A) Vero and U87 cells were infected with ZIKV at an MOI of 10. The supernatant was collected at 0, 8, 16, 28, 32, and 46 h postinfection and titered onto naïve Vero cells. The number of focus-forming units (FFU) at each time point was calculated by quantitative imaging of cells stained for ENV protein. n = 4; mean ± SEM. (B) U87 cells and hPSC-derived astrocytes were infected with ZIKV-PR at an MOI of 1. At 48 hpi, cells were immunostained for ENV protein and cellular DNA (DAPI, blue) and imaged. Note the nuclear exclusion and perinuclear accumulation of the envelope signal. (Scale bars, 100 µm.) (C) ENV protein staining in U87 cells requires permeabilization during the staining protocol. At 48 hpi, cells were fixed and permeabilized or left untreated before staining for ENV protein. Positive ENV protein staining was detected only in permeabilized conditions, suggesting intracellular presence of ENV protein. (Scale bar, 100 μm.) High magnification of boxed region (Right). (D) Treatment with the translational inhibitor cycloheximide robustly reduces staining of ENV protein in U87 cells following infection. U87 cells were treated with cycloheximide for 1 h and infected with ZIKV-PR at an MOI of 20 in the presence of the drug. At 48 hpi, cells were stained for ENV protein and imaged. (Scale bar, 100 µm.)

Fig. S2.

Radial glial tropism of ZIKV is conserved between strains. (A) Overview of organotypic slice culture infection with ZIKV-BR. Note the presence of many ENV+ cells in the ventricular zone. (Scale bar, 100 μm.) (B) Radial glial infection in the VZ, showing a high-density cluster of infected cells (arrows). (Scale bar, 100 μm.) (C–F) High-magnification examples of infected ventricular radial glia, with ZIKV-BR (C and D), ZIKV-PR (E), or ZIKV-CAM (F) showing SOX2+ nuclei (white arrows, C--F) and ENV+ cell body/processes (yellow arrowheads, C and D). (Scale bars, 10 μm.) (D) Three-dimensional reconstruction of the infected cell shown in C.

Fig. S3.

Minor increase in apoptosis following ZIKV-PR infection. (A and B) VZ (A) and OSVZ (B) of 15.5 pcw at 3 d postinfection (dpi) showing ENV and cleaved caspase-3 staining in a ZIKV-PR–infected slice (Top) compared with mock infection (Bottom). Merged images show the lack of overlap between ZIKV-PR–infected cells and cleaved caspase-3–positive cells. (Scale bars, 100 μm.) (C and D) Staining in slices from 17 pcw at 5 dpi with ZIKV-PR. (C, Top) Low magnification of a representative field showing DAPI, cleaved caspase-3, and ENV. The merged image shows a predominant lack of overlap between ENV-positive cells and cleaved caspase-3 cells. The dotted boxes highlight a rare cell with both ENV and cleaved caspase-3 signal, shown at higher magnification (Bottom, arrow), where the fragmented nucleus is evident in the DAPI signal. (Scale bars, 20 μm.) (D, Top) Low magnification of a representative field showing DAPI, cleaved caspase-3, and ENV. The merged image shows a lack of overlap between ENV and cleaved caspase-3. The dotted boxes highlight a cell with cleaved caspase-3 signal but no ENV, shown at higher magnification (Bottom, arrow), where the DAPI signal shows condensed chromatin at the periphery of the nuclear membrane. (Scale bars, 20 μm.) (E) Quantification of cell death. The percentage of cells in mock-infected slices or the percentage of ENV− and ENV+ cells in ZIKV-PR–infected slices that are immunopositive for cleaved caspase-3 at 5 dpi. The mean of biological replicates for each condition is shown by a horizontal line, at 0.9%, 0.7%, and 4.4%, respectively. In ZIKV-infected slices, 0.2 to 1% of all cells were ENV+; see SI Materials and Methods for cell numbers. Two independent donors at 17 pcw are included. One-way ANOVA with Tukey’s multiple comparisons test, **P ≤ 0.01.

Fig. 2.

ZIKV infects astrocytes in later stages of human brain development. (A) Low-magnification overview of ZIKV infection detected by ENV (green) within human organotypic cortical slices during late neurogenesis/gliogenesis. (Scale bar, 100 μm.) (B) Quantification of ENV+ cells by region (Top) and cell type (Bottom) at 20 to 22 pcw. n = 2; mean ± SD [SI Materials and Methods; an error bar is not shown where shorter than the line thickness (Top, VZ and second OSVZ)]. (C) Schematic summary of cell types observed to be susceptible to ZIKV infection (green). (D and E) Immunohistochemical analysis reveals ZIKV infection in astrocytes by positivity for ENV (arrows, D; arrowheads, E) or ENV and nonstructural protein NS5, indicating active viral replication (filled arrowheads, E). (Scale bars, 20 μm.) (F) Microglia colabeled with ENV (arrows). (Scale bars, 50 μm.) (G) ZIKV infection of oligodendrocyte precursor cells (OPCs, arrow). (Scale bars, 20 μm.) (H) Viral production in 19-pcw cortical slices, quantified by focus-forming assay from combined homogenized tissue and conditioned media at 4, 48, and 96 h postinfection. FFU, focus-forming units. Two independent biological replicates with two technical replicates for each time point; mean ± SEM; one-way ANOVA with Tukey’s multiple comparisons test, *P ≤ 0.05, **P ≤ 0.01; see also Fig. S4F. (I and J) Analysis of ZIKV-BR infection in the presence of AXL-blocking antibody in hPSC-derived astrocytes (SI Materials and Methods). Note the reduced ENV staining with AXL block compared with IgG control. (Scale bar, 100 μm.) (J) Quantification of the experiment represented in I; see also Fig. S5A; n = 3; mean ± SEM; one-way ANOVA with Tukey’s multiple comparisons test, **P ≤ 0.01. (K) ZIKV-PR infection after knockdown of AXL using U87-dCas9 lines expressing either GFP guide (g)RNA (nontargeting control) or AXL gRNAs (dCas9-mediated knockdown); see also Fig. S5B; two biological replicates in cell lines generated with independent transductions; mean ± SEM; two-way ANOVA with Tukey’s multiple comparisons test, n.s. (not significant), ***P ≤ 0.001.

Fig. S4.

Cellular tropism of ZIKV in the primary human cortex around midgestation. (A) ZIKV-PR infects immature astrocytes (arrow, Right) at the late second trimester. [Scale bars, 50 μm (Left) and 20 μm (Right).] (B) High-magnification example of a ZIKV-BR–infected astrocyte (arrow). (Scale bar, 10 μm.) (C) ZIKV-CAM–infected astrocytes (arrow) in a late second-trimester subplate. (Scale bar, 20 μm.) (D) Replication of ZIKV-PR in astrocytes. High-magnification examples of infected astrocytes with NS5 signal (filled arrowheads) demonstrating viral replication. Examples of rare ENV+ cells without appreciable NS5 staining are indicated with empty arrowheads. (Scale bar, 20 μm.) (E) Western blot analysis of flavivirus nonstructural protein NS3 in cells following ZIKV infection demonstrating viral replication. Whole-cell extracts from U87 cells (Left) and cultured cortical slices (17 pcw; Right) were separated by SDS/PAGE and immunoblotted with an antibody to flavivirus NS3 (∼60 kDa) and vinculin (loading control). (F) Viral production in cortical slices (19 pcw) from two donors, quantified by focus-forming assay from combined homogenized tissue and conditioned media at 4, 48, and 96 h postinfection; two technical replicates for each time point per donor; mean ± SD. see also Fig. 2H. (G) Microglia immunoreactive for ENV in late second-trimester brain exposed to ZIKV-BR. The asterisk indicates the panel shown also in Fig. 2F. (Scale bar, 50 μm.) (H) Microglia immunoreactive for ENV (arrow) in brain exposed to ZIKV-PR. (Scale bars, 20 μm.) (I) Low-magnification overview of oligodendrocyte precursor cells (OPCs) infected by ZIKV-BR. The asterisk indicates the panel shown also in Fig. 2G. (Scale bar, 100 μm.) (J) High-magnification example of a ZIKV-PR–infected neuron (arrow). (Scale bar, 10 μm.)

Fig. 3.

Azithromycin treatment inhibits ZIKV infection in glial cells. (A) U87 cells were treated with increasing concentrations of AZ and infected with ZIKV-PR at varying MOIs (0.01, 0.1, and 3, as indicated). The percentage of infected cells at 48 hpi was determined by flow cytometry of cells immunostained for ENV and normalized to untreated cells (for raw data, see Fig. S6A). EC50 values for AZ-mediated reduction of ZIKV infection were 5.1 µM for an MOI of 3 (n = 2), 2.9 µM for an MOI of 0.1 (n = 2), and 2.1 µM for an MOI of 0.01 (n = 2); mean ± SD. (B) Representative images of U87 cells treated with AZ and infected with ZIKV-PR at an MOI of 3 (as in A). At 48 hpi, cells were immunostained for ENV protein (green) and cellular DNA (DAPI, blue). (Scale bar, 100 μm.) (C) Rescue of cell viability with AZ. U87 cells were pretreated with AZ for 1 h and then infected with ZIKV-PR at an MOI of 10 in the presence of AZ. Cell viability was measured at 72 hpi using the CellTiter-Glo luminescence assay. The EC50 value for the AZ-mediated rescue of cell viability was 7.1 µM. The data point at the highest concentration of AZ (50 µM) showed reduced cell viability, likely due to drug toxicity (Fig. S6C). n = 2; mean ± SD. (D) Decrease of virus production with AZ treatment. U87 cells were pretreated with AZ for 1 h and then infected with ZIKV-PR at an MOI of 0.1 or 0.01 in the presence of AZ. Quantification of virus yield in conditioned media was performed by focus-forming assay at 0, 24, 48, and 72 hpi; n = 2 for each MOI; mean ± SD; two-way ANOVA with Tukey’s multiple comparisons testing, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Fig. S5.

AXL contributes to ZIKV infection. (A) AXL supports ZIKV infection of astrocytes. hPSC-derived astrocytes cultured in media containing growth factors (SI Materials and Methods) were infected with ZIKV-BR or ZIKV-PR at an MOI of 10. The bar chart shows the percentage of ENV+ cells at 24 hpi in the presence of IgG control or anti–AXL-blocking antibody; n = 3; mean ± SEM; one-way ANOVA with Tukey’s multiple comparisons test, **P ≤ 0.01, ***P ≤ 0.001; see also Fig. 2 I–K. (B) AXL knockdown in U87-dCas9-KRAB cells. Whole-cell extracts from U87-dCas9 cells expressing a GFP gRNA (nontargeting) or a pool of five AXL-targeting gRNAs were separated by SDS/PAGE and immunoblotted using antibodies against AXL or GAPDH (loading control); see also Fig. 2K. (C) U87 cells were infected with ZIKV-PR at an MOI of 20 in the presence of the AXL kinase inhibitor R428 at 1 or 3 μM or vehicle. Cells were immunostained for ENV protein at 48 hpi. (Scale bar, 100 μm.)

Fig. S6.

Azithromycin reduces ZIKV infection, replication, and virus-mediated cell death. (A) AZ effectively reduces ZIKV infection at different MOIs. U87 cells were treated with increasing concentrations of AZ and infected with ZIKV-PR at an MOI of 0.01, 0.1, or 3 in the presence of the drug. At 48 hpi, cells were stained for ENV protein, and the percentage of infection was determined by flow cytometry. EC50 values for AZ-mediated reduction of ZIKV infection are 5.1, 2.9, and 2.1 µM for the MOI of 3, 0.1, and 0.01, respectively. n = 2; mean ± SD. (B) The EC50 for AZ-dependent reduction of ZIKV infection is influenced by the viral dose. The EC50 values for AZ in A and H are plotted as a function of the baseline infection. (C) AZ shows mild toxicity in cultured cells. U87 cells were grown in media with the specified concentrations of AZ for 72 h. Cell viability and proliferation were evaluated using the CellTiter-Glo luminescence-based assay. The TC50 for AZ in U87 cells was calculated at 53 µM. n = 2; mean ± SD. (D) Rescue of cell viability with AZ. U87 cells were pretreated with AZ for 1 h and then infected with ZIKV-BR in the presence of AZ at an MOI of 10 for 72 h. Cell viability was measured using the CellTiter-Glo luminescence-based assay. The EC50 value for the AZ-mediated rescue of cell viability was 7.2 µM. n = 3; mean ± SEM. (E–G) AZ reduces high levels of infection in astrocytes. (E) hPSC-derived astrocytes were treated with the specified concentrations of AZ for at least 1 h. After treatment, cells were infected with ZIKV-PR at an MOI of 1 in the presence of the drug. At 48 hpi, cells were stained for ENV protein, and the percentage of infection was determined by immunohistochemistry with quantification. EC50 for AZ-mediated reduction of infection in this cell type was calculated at 15 µM (interpolated). n = 3; mean ± SEM. (F) AZ shows mild toxicity in cultured astrocytes. hPSC-derived astrocytes were grown in the presence of increasing concentrations of AZ for 72 h. Cell viability and proliferation were evaluated using the CellTiter-Glo luminescence-based assay. The TC50 for AZ in astrocytes was calculated at 44 µM. n = 2; mean ± SD. (G) hPSC-derived astrocytes were infected with ZIKV-PR at an MOI of 1 in the presence of AZ at the indicated concentrations. Cells were immunostained for ENV protein expression at 48 hpi and imaged. (Scale bar, 100 μm.) (H) Comparison of AZ, daptomycin, and sofosbuvir. U87 cells were treated with increasing concentrations of drug for 1 h before infection with ZIKV-PR at an MOI of 1 in the presence of the drug. At 48 hpi, cells were stained for ENV protein, and the percentage of infection was determined by flow cytometry. EC50 values for reduction of ZIKV infection were 2.2 µM for daptomycin, 6.5 µM for AZ, and 12.4 µM for sofosbuvir. Note that the reduction of infection at concentrations below 5 µM is similar between AZ and daptomycin, but the effect of daptomycin levels off with 46% infection remaining, whereas AZ continues to decrease infection to 4% at 20 µM. n = 2; mean ± SD.