Phenotypic Differences between Asian and African Lineage Zika Viruses in Human Neural Progenitor Cells

The mechanism by which ZIKV causes a range of neurological complications, especially congenital microcephaly, is not well understood. The fact that congenital microcephaly is associated with Asian lineage ZIKV strains raises the question of why this was not discovered earlier. One possible explanation is that Asian and African ZIKV strains differ in their abilities to infect cells of the CNS and to cause neurodevelopmental problems. Here, we show that Asian ZIKV strains infect and induce cell death in human neural progenitor cells—which are important target cells in the development of congenital microcephaly—less efficiently than African ZIKV strains. These features of Asian ZIKV strains likely contribute to their ability to cause chronic infections, often observed in congenital microcephaly cases. It is therefore likely that phenotypic differences between ZIKV strains could be, at least in part, responsible for the ability of Asian ZIKV strains to cause congenital microcephaly.

S ince the emergence of Zika virus (ZIKV) in 2015 in South America, infections have caused a wide spectrum of neurological diseases, such as Guillain-Barré syndrome, myelitis, meningoencephalitis, and in particular congenital microcephaly (1). Even though ZIKV was first detected in Uganda in 1947 in a rhesus monkey and has caused repeated outbreaks since 2007, not much was known about the pathogenesis of disease caused by ZIKV before the 2015 outbreak. Since then, several studies have shown that ZIKV can infect a variety of neuronal cells, but more insight into the pathogenesis of ZIKV-induced central nervous system (CNS) diseases is needed (2).
An important question that remains is whether the emergence of ZIKV in South America and the associated clinical findings are the result of genetic and phenotypic changes in the emerging ZIKV strain or whether they can be attributed to the introduction of ZIKV in a large naive population (3). Phylogenetically, two distinct lineages of ZIKV exist: the African lineage and the Asian lineage (4). The current outbreak strain belongs to the Asian lineage, and sequence analysis revealed that the virus has changed significantly over the last 50 years, both in nucleotide sequences and amino acid composition (4,5). The prototype ancestral ZIKV strain, MR766, of the African lineage has been used in many initial studies (6)(7)(8)(9), but recent in vitro and in vivo studies have shown some differences between African and Asian ZIKV strains (10)(11)(12)(13)(14). Whether there are also phenotypic differences between Asian ZIKV strains, caused by amino acid substitutions acquired just before the outbreak in South America, is currently unknown (5).
ZIKV has been shown to replicate and induce cell death in neuronal cells of fetal mice (5,15), as well as in human neural progenitor cells and brain organoids (6,7,11,12), a mechanism thought to play an important role in the pathogenesis of ZIKVinduced microcephaly. A recent study has shown that an African ZIKV strain might be able to infect human neural stem cells (hNSCs) and astrocytes more efficiently than Asian ZIKV strains (12). However, a comprehensive study on the replication kinetics and the ability to cause cell death of different African and Asian ZIKV strains is currently lacking (2).
To be able to detect phenotypic differences between Asian and African ZIKV strains or between recent Asian ZIKV strains, it is important to characterize and understand the in vitro replication kinetics-including the infection efficiency, burst size, and ability to cause cell death-of these viruses. Therefore, we determined the replication kinetics of two Asian ZIKV strains (isolated in 2013 and 2016) and two African ZIKV strains (isolated in 1947 and 1961) on induced pluripotent stem cell-derived human neural progenitor cells (hNPCs) and several human neural cell lines.

RESULTS
Phylogenetic and amino acid variance analysis of ZIKV strains selected in this study. Four ZIKV strains were included in this study (Fig. 1A). Two African strains, ZIKV MR766 (ZIKV AF-MR766 ) and Uganda 976 (ZIKV AF-976 ), were isolated in 1947 and 1961, respectively, and passaged on mouse brain tissue and Vero cells. The two Asian ZIKV strains included were H/PF/2013 (ZIKV AS-FP13 ) and ZIKVNL00013 (ZIKV AS-Sur16 ), which were isolated in 2013 and 2016, respectively, and passaged 4 times on Vero cells. A phylogenetic analysis of the complete genome of the selected strains with other ZIKV genomes shows their positions in the Asian or African lineages (Fig. 1A). There are over 50 amino acid (aa) differences between the African and Asian ZIKV strains that have previously been described (5). The amino acid differences between the Asian ZIKV strains were located in the NS1 (R67S; position 863), NS2B (S41T; position 1417), and NS5 (M60V; position 2634) proteins (Fig. 1B). Of these amino acid differences, the mutation at position 2634 is only observed in viruses isolated from the recent outbreak (4,5,16). The amino acid difference at position 1417 of ZIKV AS-Sur16 was not present in the original clinical isolate but was acquired during passaging on Vero cells (17).
Growth curves of Asian and African ZIKV strains on neuronal cells. Growth curves were determined for ZIKV AS-FP13 , ZIKV AS-Sur16 , ZIKV AF-MR766 , and ZIKV AF-976 by in vitro infections using low multiplicities of infection (MOI [0.1 and 0.01]) on SK-N-SH cells (human neuroblastoma cells), U87-MG cells (human glioblastoma cells), Vero cells, and hNPCs. Growth curves showed that all cells supported replication of all four ZIKV strains included, but virus titers were significantly lower for both Asian strains compared to ZIKV AF-MR766 strains on cells of the Vero, SK-N-SH, and U87-MG lines and hNPCs ( Fig. 2A and B). On hNPCs, ZIKV AF-MR766 grew faster to significantly higher titers than ZIKV AF-976 ( Fig. 2A and B). There were no evident differences in the growth curves between ZIKV AS-FP13 and ZIKV AS-Sur16 .
One-step growth curves of Asian and African ZIKV strains on neuronal cells. One-step growth curves (OSGCs) were assessed in vitro by using a high MOI (MOI of 10). Data from OSGC experiments on the different cell lines were used to calculate the percentage of infection and burst size (progeny virus produced per cell). OSGCs showed that baseline virus titers were higher for the African ZIKV strains than for the Asian ZIKV strains, and both African ZIKV strains grew to higher titers on all cells (Fig. 3A). In all cells, African ZIKV strains infected more cells than the Asian ZIKV strains ( Fig. 3B and D). The numbers of virus particles produced did not differ significantly between the different ZIKV strains. However, there was a trend toward burst size being higher in SK-N-SH cells (~100 to 400 infectious virus particles/cell) than in Vero cells (~20 to 40 infectious virus particles/cell), U87-MG cells (~50 to 150 infectious virus particles/cell), and hNPCs (~40 to 75 infectious virus particles/cell) (Fig. 3C).
Induction of cell death by Asian and African ZIKV strains. The ability of ZIKV AS-Sur16 , ZIKV AS-FP13 , and ZIKV AF-MR766 to cause cell death in hNPCs at 24, 48, and Infection with ZIKV AS-FP13 and ZIKV AS-Sur16 resulted in approximately 20% infection at 72 hpi, and up to 9% of cells were TUNEL positive, the latter comparable to control and BPL-treated cells. Immunofluorescent staining revealed that very few TUNELpositive cells were ZIKV infected ( Fig. 4A and B). In contrast, infection with ZIKV AF-MR766 resulted in 46% infection and 30% TUNEL-positive cells at 72 hpi (Fig. 4A). Immunofluorescent staining revealed that in ZIKV AF-MR766 -infected cells, the majority of TUNELpositive cells were also infected, indicating that ZIKV AF-MR766 is able to induce cell death early after infection in hNPCs (Fig. 4B).

DISCUSSION
This study on the in vitro replication of different ZIKV strains shows that African ZIKV strains replicated more efficiently in Vero, human glioblastoma, and human neuroblastoma cells and hNPCs than Asian ZIKV strains. In hNPCs, which are considered an important target cell type for the development of congenital microcephaly, African ZIKV strains induced cell death early after infection, which was not observed after infection with Asian ZIKV strains.  Overall there were few phenotypic differences between ZIKV AS-FP13 and ZIKV AS-Sur16 . This suggests that the mutations between these viruses, including position 2634 unique for ZIKV isolated from this outbreak, does not lead to large phenotypic changes, at least not in these cell lines. The fact that Vero and SK-N-SH cells permit efficient replication of Asian ZIKV strains supports the usage of these cells for virus isolation from clinical samples (18).
The replication kinetics and ability to cause cell death in hNPCs differed substantially between African and Asian ZIKV strains. Asian ZIKV strains infect and replicate less efficiently in hNPCs than the African ZIKV strains. This "reduced" replication is not an intrinsic feature of Asian ZIKV strains, since they replicate to high titers in Vero and SK-N-SH cells. One possible explanation for the increased ability of African ZIKV strains to infect hNPCs in this study could be that these strains have adapted to neural cells due to their passage history in mouse brain tissues (18) and that the 4-aa deletion in the E protein of these viruses contributes to the observed phenotype. However, similar results-high percentage of infection and induction of cell death in hNPCs-have been observed with a low-passage-number 1989 African ZIKV strain (ArB41644) (12). Upon sequencing, we did not find any deletion in the E protein (GenBank accession no. KY576904) of this low-passage-number African lineage ZIKV strain. Therefore, these studies together suggest that Asian ZIKV strains infect hNPCs less efficiently than African ZIKV strains, regardless of the passage history of the ZIKV strains. Both Asian lineage ZIKV strains do not seem to induce cell death early after infection, whereas ZIKV AF-MR766 does. This fits with previous observations, where more apoptotic nuclei were observed after infection with an African ZIKV strain than with an Asian ZIKV strain (12), which suggests that there are intrinsic differences between Asian and African ZIKV strains in their ability to cause cell death in hNPCs.
The observed phenotypic characteristics of Asian lineage ZIKV strains might contribute to their ability to cause chronic infection in tissues of the CNS (17-21). First, Asian ZIKV strains infect relatively few hNPCs. Second, Asian ZIKV strains release less than 40 infectious virus particles per infected hNPC, which is relatively low compared to other viruses, such as influenza virus and simian immunodeficiency virus (SIV) (22,23). A low burst size has previously also been associated with prolonged virus replication within the CNS for Japanese encephalitis virus, another flavivirus (24). Finally, Asian ZIKV strains do not seem to induce cell death early after infection in neural progenitor cells, which might result in chronic infection and replication within the CNS (19,20). This fits with a recent animal study using Stat2 Ϫ/Ϫ mice, which showed that African ZIKV strains induce short episodes of severe neurological symptoms followed by lethality, while Asian ZIKV strains manifest prolonged signs of neuronal malfunctions. Limited mortality was also only observed in one Asian ZIKV strain (25).
Taken together, we here show that African and Asian ZIKV strains differ in their abilities to infect and replicate in different neural cells, as well as their abilities to cause cell death early after infection. This implies that caution is necessary against extrapolation of experimental data obtained using historical African ZIKV strains to the current outbreak. In addition, the fact that Asian ZIKV strains infect only a minority of cells with a relatively low burst size together with the lack of early cell death induction might contribute to their ability to cause chronic infections within the CNS.
Next-generation sequencing. For genomic characterization of the virus strains, RNA was isolated from 140 l of the virus stocks with the QIAmp Viral Mini RNA kit (Qiagen, Germany). Subsequently, the product was eluted in 40 l double-distilled water. Viral metagenomic libraries were constructed with Phylogenetic analysis. Nearly full-length ZIKV genomes of 4 isolates (ZIKV AF-976 , ZIKV AF-MR766 , ZIKV AS-FP13 , and ZIKV AS-Sur16 ) and other reference sequences were obtained from the GenBank database. The sequences were aligned using ClustalW, and a phylogenetic tree was constructed by using the PhyML method in SeaView 4 (http://pbil.univ-lyon1.fr/software/seaview) with the approximate likelihood ratio test based on a Shimodaira-Hasegawa-like procedure which used general time reversible as a substitution model. Nearest-neighbor interchange, subtree pruning, and regrafting-based tree search algorithms were used to estimate tree topologies (28). The obtained tree was visualized by using FigTree version 1.3.1 (http://tree.bio.ed.ac.uk/software/figtree).
Replication kinetics of Zika virus strains. Replication kinetics of ZIKV strains ZIKV AS-976 , ZIKV AS-MR766 , ZIKV AS-FP13 , and ZIKV AS-Sur16 were studied in vitro by means of one-step growth curve (OSGC) experiments with a multiplicity of infection (MOI) of 10 and focal experiments (growth curves) with MOI of 0.1 and 0.01. Human neural progenitor cells and SK-N-SH, U87MG, and Vero cells were seeded into 96-well plates (2 ϫ 10 4 cells) (Greiner, USA). After 24 h, monolayers were inoculated with the different ZIKV strains or Vero cell culture medium as a control at an MOI of 10, 0.1, or 0.01 for 1 h at 37°C in 5% CO 2 . After 1 h of virus absorption, the inoculum was removed and cells were washed 3 times and replenished with fresh medium that contains 2% FCS (no FCS for hNPCs) and cultured for 24 or 72 h at 37°C for the OSGC and growth curve, respectively. For the OSGC, supernatant was collected every 2 h up to 24 h and subsequently stored at Ϫ80°C until virus titer determination. Cells were fixed in 4% paraformaldehyde (PFA) for 20 min at room temperature, washed with phosphate-buffered saline (PBS), and permeabilized and stored in 70% ethanol for immunofluorescent staining. For the growth curves, supernatant was collected at time points 0, 1, 12, 24, 48, and 72 hpi and stored at Ϫ80 until use. All growth curves and OSGCs were performed 3 times (biological replicates), and each growth curve included duplicate (technical replicates) measurements from which the average was used for future analysis.
Determination of virus titers. Virus titers (TCID 50 ) in the supernatant were determined by endpoint titrations on Vero cells. Tenfold serial dilutions were made and inoculated onto a monolayer of Vero cells. Cytopathic effect (CPE) was determined at 5 days postinfection (dpi), and virus titers were calculated using the Spearman-Kärber method (26). An initial 1:10 dilution of supernatant resulted in a detection limit of 10 1.5 TCID 50 /ml. . Zika virus-infected cells were identified by use of a Zeiss LSM 700 confocal laser scanning microscope fitted on an Axio observer Z1 inverted microscope (Zeiss). All images were processed using Zen 2010 software (Zeiss). Per sample, 5 high-power fields were photographed and scored blindly by three individuals to determine the percentages of infected and noninfected cells.
Calculation of percentage of infection and burst size. The percentage of infection and burst size were calculated from the OSGC experiment. The burst size is defined as the number of progeny virus particles produced per infected cell and was calculated as follows. The time at which half the number of progeny virus were released into the supernatant (50% effective concentration [EC 50 ] for dose-response curve) was determined, which was calculated by using a nonlinear regression analysis (sigmoidal dose-response, variable slope) in GraphPad Prism 6.0h using the infectious virus titer data from the OSGC measured over 24 h (2-h increments). At this time point, infected cells were fixed and stained for ZIKV (as described above), and the number of infected cells was calculated by counting virus-infected/ uninfected cells in 5 randomly chosen panels in duplicate by 3 blind assessors. The average number of infected cells from 5 panels was taken and corrected for the surface area of a single 96-well flat bottom plate. Next, the infectious virus titer over 24 h was calculated by subtracting time point 0 from 24 h, which then was divided by the number of infected cells, resulting in the number of progeny virus particles produced per infected cell.
TUNEL assay. Human neural progenitor cells (hNPCs) were cultured in a 24-well plate and inoculated with ZIKV AS-MR766 , ZIKV AS-FP13 , or ZIKV AS-Sur16 at an MOI of 3. In addition, ZIKV AS-MR766 was inactivated using ␤-propiolactone (BPL) (1:4,000 vol/vol; Sigma-Aldrich, USA) at 4°C for 48 to 72 h. Subsequently, BPL was inactivated for 24 h at 37°C. Both inactivated ZIKV AS-MR766 and Vero cell culture supernatant served as negative controls. Viruses and controls were allowed to absorb for 1 h, after which hNPCs were washed three times in hNPC medium. Subsequently the medium was replenished with fresh medium and cultured at 37°C for 24, 48, or 72 h. The numbers of dead cells were measured with a Sigma-Aldrich In situ cell death detection kit with fluorescein (Sigma-Aldrich, USA). Briefly, the cells were first fixed with 4% PFA and permeabilized with a 1:1 dilution of 1% Triton X-100 and 70% ethanol. Noninfected cells were treated with 180 IU/ml DNase (Roche Diagnostics, Mannheim, Germany) for 15 min at room temperature to serve as a positive control. The In situ cell death detection kit with fluorescein was used according to the manufacturer's instructions. Cells stained only with labeling solution were used as a negative control as suggested by the manufacturer. The number of TUNEL-positive cells was measured using a BD FACSCanto II (BD Biosciences, USA). Data were analyzed using FlowJo 10 software (Ashland, OR, USA). All experiments were performed three times (biological replicates), and each experiment included duplicate (technical replicate) measurements from which the average was calculated and used for further analysis.
Flow cytometry assay. Cells were infected the same way as described for the TUNEL assay. At time points 24, 48, and 72 h, cells were collected, fixed, and permeabilized using BD Cytofix/Cytoperm solution (BD Biosciences, USA) according to the manufacturer's instructions. Cells were blocked using 10% normal goat serum (NGS [Dako, Denmark]) for 10 min on ice. Subsequently, Zika virus was detected using mouse monoclonal antibody against anti-flavivirus group antigen (MAB10216, clone D1-4G2-4-15; Millipore, Germany) at a 1:200 dilution or mouse IgG2a isotype control (MAB003; Dako, Denmark) at a 1:50 dilution in BD Perm/Wash containing 2% NGS and incubated for 1 h on ice and in the dark. Cells were washed twice, and goat anti-mouse IgG2a conjugated with Alexa Fluor 488 (Life Technologies, Inc., The Netherlands) at a 1:250 dilution was incubated for 1 h in the dark and on ice. After incubation of the secondary antibody, cells were washed twice and resuspended in fluorescence-activated cell sorter (FACS) buffer. The percentage of infected cells was measured using a BD FACSCanto II (BD Biosciences, USA). Data were analyzed using FlowJo 10 software (Ashland, OR, USA). All experiments were performed three times (biological replicates), and each experiment included duplicate (technical replicate) measurements from which the average was calculated and used for further analysis.
Statistical analysis. The statistical analyses were performed using GraphPad Prism 6.0h software (La Jolla, CA) for Mac. Student's t test was used for comparison between two groups. For comparison between multiple groups, one-way analysis of variance (ANOVA) with Tukey's multiple-comparison test was used. P values of Յ0.05 were considered significant.
Accession number(s). Sequences of the E protein from African ZIKV strain ArB41644 have been submitted to GenBank under accession no. KY576904.