A Zika virus protein expression screen in Drosophila to investigate targeted host pathways during development

Expression of most ZIKV proteins or protein complexes causes lethality or morphological phenotypes when ectopically expressed in Drosophila

To determine how ZIKV proteins might hijack host pathways to cause disease, we sought to express each protein in a wide variety of tissues throughout development and in the adult fly using the GAL4/UAS system (Fig. 1B). To generate transgenic flies, we used sequence information from the PRVABC-59 strain isolated in Puerto Rico (Lanciotti et al., 2016) and synthesized and cloned individual protein-coding sequences into a UAS transgenic vector (pGW-HA.attB) (Bischof et al., 2013) using Gibson assembly (Gibson et al., 2009). The ZIKV genome is translated as one polyprotein and subsequently cleaved at established viral and host protease cleavage sites (Fishburn et al., 2022). We therefore designed expression constructs based on the major individual proteins formed following polyprotein processing. Each protein of interest was tagged with a C-terminal 3×HA tag for visualization and expression analysis, and careful consideration was taken to include proper signal sequences for transmembrane proteins (Shah et al., 2018). Transgenic strains were generated using the ɸC31 transgenesis technique, allowing integration of all constructs into an identical genomic locus of the fly at the established attP-docking site (VK37) at 22A3 on the second chromosome (Venken et al., 2006). We generated transgenic lines that allow expression of six proteins encoded in the ZIKV genome as single proteins: Capsid (C), NS1, NS2A, NS4A, NS4B and NS5. We also generated a construct that combines prM with E (prM::E) and NS2B with NS3 (NS2B::NS3). The former transgene was generated because prM is known to act as a co-chaperone for proper folding of E, and the two proteins form heterodimers that are critical for flaviviruses to form infectious particles (Yu et al., 2008). The latter transgene was generated because NS2B and NS3 form a heterodimer, which functions as a key viral protease that processes the viral precursor polyprotein (Falgout et al., 1991). In addition, as there are proteolytic cleavage sites on both sides of the short peptide 2K located between NS4A and NS4B (Sun et al., 2017), we generated constructs for NS4A and NS4B with and without 2K (NS4A, NS4A::2K, 2K::NS4B and NS4B). We also generated constructs for NS1 and prM::E with altered amino acids that have been implicated in infectability or pathogenicity. These constructs express NS1^cam and prM::E^cam, which carry single amino acid changes that are found in a less pathogenic strain of ZIKV, FSS13025, isolated in Cambodia (Liu et al., 2017).

NS1^cam carries an alanine (A) in position 188 instead of the valine (V) in the NS1 in the Puerto Rican strain (NS1^pr). We generated this transgene because the p.A188V mutation in NS1 increases the infectivity of ZIKV in mosquitos (Liu et al., 2017) and inhibits interferon β production in human cells (Xia et al., 2018). prM::E^cam carries a serine (S) in position 139 of the E protein instead of the asparagine (N) in E protein in the the Puerto Rican strain (E^pr). This construct was generated because a p.S139N mutation increased infectability of the virus in human and mouse neural progenitor cells and caused a more severe microcephaly phenotype in mouse models (Yuan et al., 2017). In total, we generated 12 transgenic lines (Table S1).

First, we crossed eight transgenic lines (UAS-C, UAS-prM::E, UAS-NS1, UAS-NS2A, UAS-NS2B::NS3, UAS-NS4A, UAS-NS4B and UAS-NS5) from the Puerto Rican strain to a variety of GAL4 lines to express each viral or fusion protein in a variety of tissues at different developmental timepoints. This would allow us to cast a wide net to assess host pathways that ZIKV might hijack during infection and disease progression. We tested three ubiquitous drivers with variable strengths [α-Tubulin at 84B (αTub84B or Tub)-GAL4, Actin (Act5C or Act)-GAL4 and daughterless (da)-GAL4] and three tissue-specific drivers [nubbin (nub)-GAL4 for the wing, pannier (pnr)-GAL4 for the dorsal thorax and glass multiple reporter (GMR)-GAL4 for the eye] to assess the functional consequences of overexpressing the ZIKV proteins on viability and gross morphology. We found that seven out of eight lines (prM::E, NS1, NS2A, NS2B::NS3, NS4A, NS4B and NS5) caused scorable phenotypes with at least one of these drivers (Figs 1C and 2, Table S1). Ubiquitous expression revealed that some viral proteins (prM::E, NS1, NS4A, NS4B and NS5) caused lethality, suggesting that these proteins likely impact essential cellular pathways. The lethal stage induced by some ZIKV proteins is presented in Table S2. Overexpression of N2A caused increase in mechanosensory bristles, whereas NS2B::NS3 overexpression caused reduction of mechanosensory bristles on the dorsal thorax of the fly, indicating that these proteins affect pathways that are required for peripheral nervous system development (Figs 1C and 2B-C) (Schweisguth, 2015; Schweisguth et al., 1996). Overexpression of NS4A using pnr-GAL4 caused a dorsal thorax closure defect (Figs 1C and 2D), which is a phenotype often seen when cell migration or cell communication through JNK and TGF-β/BMP signaling is defective (Agnès et al., 1999; Martin-Blanco et al., 2000). Overexpression of NS4A or NS4B using GMR-GAL4 caused a rough eye morphology phenotype (Figs 1C and 3D-H), suggesting a developmental defect of the compound eye, which can be caused by disruption of many pathways and biological processes (Kumar, 2001; Thomas and Wassarman, 1999). In contrast, although the development of the wing depends on many signaling pathways and cellular processes (Bier, 2005), none of the lines examined, including those expressing the ZIKV proteins that cause lethality when overexpressed ubiquitously, caused an obvious wing morphological defect when the ZIKV proteins were expressed using nub-GAL4. Therefore, ZIKV proteins appear to affect specific proteins, pathways or cellular processes to induce a phenotype upon overexpression in flies in a context-specific manner, rather than through inducing a general cellular toxicity.

The site of 2K peptide cleavage can affect the function of NS4A and NS4B

Between the NS4A and NS4B proteins, there is a linker sequence that is referred to as the 2K peptide (Fig. 1A) (Lin et al., 1993). The site between NS4A and 2K is cleaved by the viral NS2B::NS3 protease and the site between 2K and NS4B is cleaved by one or more host proteases, respectively (Sun et al., 2017). Because the two cleavage events occur sequentially (Lin et al., 1993) and 2K has been shown to alter NS4A function (Miller et al., 2007; Roosendaal et al., 2006), we tested whether expressing a fusion protein of NS4A and 2K (NS4A::2K) or 2K and NS4B (2K::NS4B) has a different effect than expressing these two proteins without this peptide (Fig. 3A,B).

Ectopic expression of NS4A without the 2K peptide caused multiple scorable phenotypes with diverse GAL4 lines, including lethality with Act-GAL4, Tub-GAL4 and da-GAL4 (Figs 1C and 3C). Expression with GMR-GAL4 produced animals with rough eye phenotypes (Figs 1C and 3C,E) and pnr-GAL4 expression led to dorsal thorax closure (Figs 1C, 2D and 3C). When we expressed NS4A::2K with the same GAL4 drivers, we observed mostly weaker phenotypes compared to those seen upon expression of NS4A alone (Fig. 3C,F). Expression of NS4A caused lethality in most animals with few escapers using pnr-GAL4 at 29°C, but lethality was not observed when NS4A::2K was expressed using this driver and temperature. However, ubiquitous expression of NS4A::2K with da-GAL4 caused lethality during pupal stages, which is in contrast to NS4A causing early larval lethality using the same driver. Similarly, viable flies that expressed NS4A::2K using Act-GAL4 were observed, which were not seen when NS4A was expressed using the same driver. In summary, the 2K peptide can sometimes suppress the function of NS4A depending on the temperature, context and GAL4 drivers being used.

Similar to NS4A, ectopic expression of NS4B without the 2K peptide using multiple drivers affected viability and development. Expression of NS4B with all three ubiquitous drivers (Act-GAL4, Tub-GAL4 and da-GAL4) caused lethality, whereas expression in the developing eye using GMR-GAL4 induced rough eyes (Figs 1C and 3C,G).

To explore why GMR-GAL4 expression of ZIKV NS4A and NS4B proteins caused rough eye phenotypes, we assessed cell division and cell death in developing eye discs. We counted the number of phospho-histone H3-positive cells, indicating a dividing cell, in the eye disc and found no significant changes (Fig. S1A-E). To mark cells undergoing apoptosis, we used activated Death Caspase 1 (Dcp1) and again found no significant changes at this stage (Fig. S1F-J). Therefore, the rough eye phenotype is likely be caused by other factors during morphogenesis or cell death occurring at later stages of development.

We also observed that addition of 2K affected the severity of phenotypes with some drivers (Fig. 3C,H). In three cases, the addition of 2K made the phenotype more severe. For example, using da-GAL4, expression of 2K::NS4B caused embryonic lethality, whereas expression of NS4B caused late larval lethality. Similarly, expression of 2K::NS4B caused a much more severe eye developmental defect compared to that seen upon expression of NS4B using GMR-GAL4 (Fig. 3C,G,H). However, overexpression of 2K::NS4B did not change the lethality observed when NS4B was expressed using Act-GAL4 or Tub-GAL4 and made the phenotype caused by another driver weaker (Rh1-GAL4, see below). Therefore, although addition of the 2K peptide to NS4B enhances its function in some contexts, it can also be irrelevant or may decrease its function in other circumstances.

Multiple ZIKV proteins reduce brain volume when expressed in Drosophila

As ZIKV infection is associated with severe microcephaly, we set out to define the set of viral proteins that might target important processes during brain development. Previously, we showed that expression of ZIKV NS4A::2K using ubiquitous (Act-GAL4) drivers or drivers specific to neural stem cells and their progeny [inscuteable (insc)-GAL4)] reduces the volume of third instar larval brains (Link et al., 2019; Shah et al., 2018). To determine whether other transgenes also have similar biological functions, we expressed each protein or fusion protein with Act-GAL4 or insc-GAL4. Upon ubiquitous expression with Act-GAL4 at 25°C, NS1, NS4A, NS4B, 2K::NS4B and NS5 caused lethality prior to the late third instar larval stage, so brains could not be assessed (Fig. 4). Expression of NS4A::2K using Act-GAL4 caused significant brain size reduction compared to that in controls overexpressing mCD8::GFP, a neutral transmembrane protein (Fig. 4G,I; one-way ANOVA and multiple comparisons posttest), whereas expression of NS2B::NS3 caused a milder but more significant brain volume defect (Fig. 4F,I). Using insc-GAL4 at 29°C, we found that expression of NS2B::NS3, NS4A, NS4A::2K and 2K::NS4B caused significant brain volume defects (Fig. 4A-E,H).

To determine possible reasons why ZIKV proteins cause significant brain volume reduction, we assessed cell division and cell death in insc>NS2B::NS3-, insc>NS4A- and insc>2K::NS4B-expressing animals. We chose a single timepoint in late third instar development for these experiments and, as a result, they provide a snapshot in time in a cell type-specific population. We counted the number of phospho-histone H3-positive neuroblasts in the central brain and found a non-significant decrease with NS4A and a significant decrease with 2K::NS4B expression (Fig. S2A-D). To mark cells undergoing apoptosis, we use activated Dcp1 and found a non-significant increase in both NS2B::NS3 and NS4A (Fig. S2E-H). In summary, these data not only support previous observations that NS4A expression (with or without 2K) in Drosophila causes reduced brain volume by interacting with the ANKLE2 pathway (rescue experiment data shown in Fig. S3) (Link et al., 2019; Shah et al., 2018), but also identify NS2B::NS3 and 2K::NS4B as previously unreported negative regulators of brain size when expressed in Drosophila.

Immunostaining of epitope tags attached to ZIKV proteins expressed in Drosophila neuroblasts reveal different patterns of subcellular localization

ZIKV proteins are synthesized on the rough endoplasmic reticulum (ER), and replication and assembly of the virus takes place primarily in this organelle (Cortese et al., 2017). However, there are several studies that suggest that some viral proteins function in other subcellular organelles. For example, NS5 can function in the nucleus to modulate host cell division and gene expression (Conde et al., 2020; Kesari et al., 2020; Li et al., 2022; Petit et al., 2021; Zhao et al., 2021). We assessed the expression and subcellular localization pattern of these proteins using immunofluorescence staining of the C-terminal 3×HA tag and confocal microscopy. We dissected late third instar larval brains expressing a ZIKV transgene using insc-GAL4 and immunostained them with antibodies against Deadpan (Dpn), a nuclear marker of neuroblasts (neural stem cells), and the HA tag, which labels each viral protein expressed (Fig. 5C-N). Many viral proteins [Capsid, prM::E^pr, NS1^pr, NS2A, NS4A::2K and NS4B] were localized in a pattern that was suggestive of ER localization (Fig. 5C,D,F,H,J,M), as indicated by the ER protein Calnexin 99A (Cnx99A) (Fig. 5A), consistent with previous observations in mammalian cells (Hou et al., 2017). Some proteins exhibited distinct patterns, including NS4A and 2K::NS4B that showed a punctate pattern (Fig. 5K,L), consistent with previous studies in mammalian cells (Miller et al., 2007; Roosendaal et al., 2006; Shah et al., 2018). NS5 showed nuclear localization (Fig. 5N), also consistent with previous studies in mammalian cells (Petit et al., 2021). It is interesting to note that the presence or absence of the 2K peptide on NS4A and NS4B, which can alter the phenotypic strength (Fig. 3), caused a change in the subcellular localization of these proteins (Fig. 5J-M).

Expression of some ZIKV proteins in post-differentiated photoreceptor cells causes electrophysiological defects

Adult humans infected with ZIKV can present with neurological symptoms (Mehta et al., 2018), whereas adult mice exposed to this virus can exhibit long-term neuropathological defects (Nem de Oliveira Souza et al., 2018). Thus, it is possible that ZIKV infects and impacts the function of the post-developmental nervous system. To test whether ZIKV proteins affect post-differentiated neurons, we expressed each protein in the fly photoreceptor cells using Rhodopsin 1 (Rh1)-GAL4 at 25°C. Photoreceptors in insects are light-sensory neurons that project axons into the lamina or medulla layers of the adult brain. Rh1 [also known as neither inactivation nor afterpotential E (ninaE)] encodes a rhodopsin that is expressed in six out of eight photoreceptors (R1 to R6) that make up an ommatidia, the basic unit of a compound eye. The function of photoreceptor neurons can be assessed using electroretinogram (ERG) analyses, which are a cumulative measure of neuronal function in response to light from the Drosophila eye and brain (Ugur et al., 2016). Live flies were immobilized on glass slides with glue, a recording electrode was placed touching the eye, but not penetrating the tissue, and a reference electrode was placed behind the eye. The fly was acclimated to a dark room and a light was flashed on for a brief period, usually 1 s. When the light was flashed on, an ERG trace that consists of an on-transient spike followed by depolarization was recorded. When the light was turned off, an off-transient spike was observed, followed by repolarization of the potential to the baseline (Fig. 6). Depolarization/repolarization represents the phototransduction response of photoreceptors to light stimulus, whereas the on/off transient spikes correspond to signal transmission between pre- and post-synaptic neurons in the brain (Dolph et al., 2011). In mutants with degenerative phenotypes, the reduction in depolarization amplitude is often documented, which may be accompanied by reduction or loss of on/off transients (Deal and Yamamoto, 2018).

Adult animals expressing each viral protein were aged for 30 days post eclosion in a 12 h/12 h light/dark cycle at 25°C. For most ZIKV proteins, expression in photoreceptor cells did not affect ERG patterns. However, prM::E expression caused a strong decrease in ERG depolarization amplitude, indicating photoreceptor function defects (compare Fig. 6A to Fig. 6B,D).

In addition, expression of NS4A, but not NS4A::2K, caused loss of ERG on/off transient spikes in aged animals (Figs 3C and 6C,E,F,J-L). To determine whether this phenotype was age dependent, we tested the ERG on flies expressing NS4A using Rh1-GAL4 that were only aged for 5 days post eclosion. Interestingly, we did not observe a difference in the on/off transient spikes or depolarization in these young animals (Fig. 6G-I). Therefore, the ERG defect caused by expression of NS4A is likely to be degenerative over time in adulthood rather than developmental or physiological in nature. Indeed, examination of the ultrastructure of the retina using transmission electron microscopy revealed that overexpression of NS4A causes significant reduction in the number of photoreceptors accompanied by many abnormal vacuoles (Fig. 6M,N). Finally, Expression of NS4B, but not 2K::NS4B, mildly decreased on transient amplitude (Figs 3C and 6D-F). Importantly, expression of other ZIKV proteins did not show these phenotypes (Fig. 6D-F), indicating that prM::E, NS4A and NS4B specifically inhibit the function or integrity of mature neurons.

ZIKV proteins encoded by different strains show functional differences when tested in Drosophila

Finally, we explored whether the Drosophila system could be used to test the functional consequences of mutations that were acquired in protein-coding genes during ZIKV evolution. The more recent Puerto Rican ZIKV strain, which is more closely related to strains responsible for recent outbreaks in the Americas, causes high rates of fetal microcephaly, similar to those seen in outbreaks in French Polynesia (Cauchemez et al., 2016). However, before these outbreaks, ZIKV infection was not associated with microcephaly. As the Cambodian strain was the last virus of the Asian lineage not associated with severe microcephaly (Duong et al., 2017), it is useful for comparative investigations. We focused on two mutations, one in NS1 and the other in E, as they had both been previously suggested to contribute to disease severity using other experimental systems (Wang et al., 2017; Xia et al., 2018; Yuan et al., 2017). The NS1 protein from the Puerto Rican strain (NS1^pr) carries a p.A188V mutation compared to the NS1 protein from the more ancient Cambodian strain (NS1^cam) (Delatorre et al., 2017). This mutation has been found to enhance the evasion of ZIKV from the host immune system based on studies performed in mice and human cells (Xia et al., 2018). When expressed using Act-GAL4, Tub-GAL4 or pnr-GAL4, NS1^pr caused lethality (Figs 1C and 7A). However, expression of NS1^cam was only semi-lethal using the same drivers tested under the same conditions (Fig. 7A). This suggests that NS1^pr has stronger function compared to NS1^cam, consistent with the notion that the Puerto Rican isolate is more pathogenic than the Cambodian isolate.

The E protein from the Puerto Rican strain (prM::E^pr) carries a p.S139N mutation compared to the E protein from the Cambodian strain (prM::E^cam). This mutation has been associated with increased infectivity and severity of microcephaly when tested in mice and human cells (Yuan et al., 2017). prM::E^pr caused semi-lethality when expressed with Tub-GAL4 (Figs 1C and 7) and exhibited an ERG defect when expressed with Rh1-GAL4 (Fig. 6). Although we did not observe a difference between prM::E^pr and prM::E^cam when these proteins were expressed using Tub-GAL4, we identified functional differences when pnr-GAL4 and Rh1-GAL4 were used. In contrast to prM::E^pr, which did not show any defect when driven with pnr-GAL4, prM::E^cam expression using this driver caused lethality (Fig. 7A). Similarly, expression of prM::E^cam caused a stronger ERG phenotype compared to that of prM::E^pr using Rh1-GAL4 (Fig. 7B-G). These data indicate that the prM::E^cam is more deleterious compared to prM::E^pr, which was opposite from what we expected. Regardless, these data show that the Drosophila assay system can be used to identify functional differences of mutant viral proteins.

Drosophila melanogaster strains and culture

The following fly lines were used: PBac{y[+mDint2] w[+mC]=UAS-Capsid^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]=UAS-prM::E^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]=UAS-prM::E^cam.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]=UAS-NS1^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]=UAS-NS1^cam.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]=UAS-NS2A^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]=UAS-NS2B::NS3^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]= UAS-NS4A^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC]= UAS-NS4A::2K^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC] =UAS-2K::NS4B^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC] =UAS-NS4B^pr.HA}VK00037 (this study), PBac{y[+mDint2] w[+mC] =UAS-NS5^pr.HA}VK00037 (this study), P{UASt-CD8-GFP} (Lee and Luo, 2001), Act-GAL4 (P{Act5C-GAL4}17bFO1) (Ito et al., 1997), insc-GAL4 (P{w[+mW.hs]=GawB}insc[Mz1407]) (Luo et al., 1994), Tub-GAL4 P{w[+mC]=tubP-GAL4}LL7 [α-Tubulin at 84B (αTub84B)] (Lee and Luo, 1999), pnr-GAL4 P{w[+mW.hs]=GawB}pnr[MD237] (RRID: BDSC_3039), GMR-GAL4 (P{w[+mC]=GAL4-ninaE.GMR}12) (RRID: BDSC_1104) and da-GAL4 (P{w[+mW.hs]=GAL4-da.G32}UH1) (Wodarz et al., 1995). All flies were maintained at room temperature (∼22°C) and grown on standard cornmeal and molasses medium in plastic vials. Crosses were performed at the temperatures indicated (25°C or 29°C). All studies contained male and female flies. Each cohort was repeated at least three times, sometimes more. Each cohort had multiple crosses. Brain volume measurements were conducted in late third instar larvae (judged by gut clearance and extruding spiracles) and all other assessments occurred in adults. All novel Drosophila lines were deposited to the Bloomington Drosophila Stock Center.

Generation of transgenic constructs

To generate ZIKV expression constructs, the ZIKV sequence was acquired from the strain PRVABC-59 (GenBank: KX377337.1), an Asian lineage isolated from Puerto Rico in 2015. Constructs were designed as described in Shah et al. (2018). All sequences were codon optimized for expression in Drosophila using the Codon Optimization Tool (Integrated DNA Technologies; https://www.idtdna.com/CodonOpt). For initial fragments (Capsid, prM::E, NS1, NS2A, NS2B::NS3, NS4A, 2K::NS4B and NS5), double-stranded, linear gene fragments were generated using gBlocks (Integrated DNA Technologies). ZIKV fragments were synthesized with additional regions of homology that correspond to sequences in the destination vector (pGW-HA.attB, a gift from Konrad Basler, Institute of Molecular Life Sciences, University of Zurich; Bischof et al., 2013) at the 5′ (5′-TCAAAAGGTATAGGAACTTCAACCGGTCAAC-3′, Kozak sequence underlined) and 3′ (5′-GTACCTCGAAGTTCCTATTCTCTACTTAGTATA-3′) ends. AgeI and KpnI restriction enzymes were used to remove the gateway compatible cassette from pGW-HA.attB, removing the attR1, CmR, ccdB and attR2 regions. UAS expression constructs were generated by combining the linearized pGW-HA.attB backbone and each individual ZIKV fragment in Gibson assembly reactions (Gibson Assembly Cloning Kit, E5510S, New England Biolabs). To generate the NS4A::2K construct, the NS4A and 2K peptide constructs (with 25 bp overlapping regions included in primers) were PCR amplified, gel purified and used in a three-fragment Gibson assembly reaction. To generate the NS4B construct without the 2K peptide, the NS4B construct was PCR amplified, gel purified and assembled with pGW-HA.attB using Gibson assembly. To generate prM::E and NS1 constructs containing single amino acid changes seen in the Cambodian strain (FSS13025), variant products were synthesized and cloned into pGW-HA.attB using Gibson assembly as above. All constructs were inserted into the VK37 landing site on the second chromosome in D. melanogaster by microinjection into a strain expressing ɸC31 integrase in the germline (y[1] M{RFP[3xP3.PB] GFP[E.3xP3]=vas-int.Dm}ZH-2A w[*]; PBac{y[+]-attP-3B}VK00037) (Bischof et al., 2007; Venken et al., 2006). Transgenic lines were selected by the white⁺ marker in a white⁻ background, and stable stocks were established through standard procedures (Harnish et al., 2019).

Crossing schemes and scoring of gross phenotypes

Wing, eye and thorax phenotypes were explored by driving the UAS-ZIKV cDNA lines with nub-GAL4, GMR-GAL4 and pnr-GAL4, respectively. Virgin females from the GAL4 stocks were collected from stock bottles kept at 25°C. Male flies were collected from the UAS stocks. Virgin females were isolated for 24 h before being crossed, and three to four females with three to five males were placed in the same vial containing standard cornmeal and molasses medium. Crosses were initially maintained at 25°C. The crosses were transferred into new vials after 4-6 days and maintained at 25°C. After 4-6 days, these parental flies were again transferred to new vials and maintained at 29°C. New flies were added if there was an insufficient number for the cross at 29°C. Flies were collected 6-8 days after flies had begun to eclose. An example of the number of flies assessed for each cross is given in Table S3. F1 progeny were scored for gross morphological phenotypes after being anesthetized with carbon dioxide. If phenotypes were observed, the flies were sacrificed via overnight freezing at −20°C and imaged the following day. The wings were dissected dry in a siliconized dish and imaged at 5× magnification. To assess lethality, the rate of flies carrying balancer chromosomes in the F1 generation was compared to the expected rate to determine whether flies were eclosing according to Mendelian ratios. Semi-lethality was noted when the observed number of protein-expressing flies was less than 75% of the expected number based on Mendelian ratios.

Immunostaining of larval brains

Late third instar (judged based on gut clearance and extruding spiracles) larval brains were dissected in PBS and fixed with 4% paraformaldehyde in PBS containing 0.3% Triton X-100 for 20 min. For immunostaining, brains were blocked in PBS containing 0.3% Triton X-100, 1% bovine serum albumin (BSA) and 5% normal goat serum and incubated in primary antibody in PBS containing 0.3% Triton X-100 and 1% BSA overnight. The primary antibodies included rat anti-Deadpan (Bier et al., 1992) (Abcam, ab195172, 1:250 or 1:500), mouse anti-Prospero (Spana and Doe, 1995) (Developmental Studies Hybridoma Bank, MR1A, 1:1000), mouse anti-Cnx99A (Riedel et al., 2016) (Developmental Studies Hybridoma Bank, Cnx99A 6-2-1, 1:100), rabbit anti-phospho-histone H3 (Ser10) (Millipore Sigma, 06-570, 1:1000), rabbit anti-cleaved Drosophila Dcp1 (Asp215) (Cell Signaling Technology, 9578, 1:100 to 1:500) and mouse anti-HA (Covance, 901501, 1:1000) with donkey secondary antibodies from Jackson ImmunoResearch (712-605-153, 715-295-151 and 711-295-152) used at 1:500. Brains were mounted with double-sided tape spacers and imaged using a confocal microscope (Leica SP8) with 2 or 3 µm sections through the entire brain lobe.

Brain volume measurement

Brains from third instar larvae were stained and mounted with tape spacers and imaged using a Leica SP8 microscope with 2 or 3 µm sections through the entire brain lobe. The resulting stacks were analyzed using the Surfaces function in Imaris (Bitplane) to quantify brain lobe volume (in μm³). One lobe from each brain was imaged and a total of ten brains was analyzed per genotype or condition. Brain lobe volumes are displayed as individual values with the line representing the median. Statistical significance was determined using one-way ANOVA with multiple comparisons posttest calculated using GraphPad Prism. Brain volumes are displayed as total brain volume (μm³). A post hoc power analysis was conducted using RStudio (2023.09.01+494). The average of all treatment averages was used to calculate the effect size. n=7.8 animals needed for 80% confidence at P=0.05.

More than five Dpn⁺ central brain neuroblasts were assessed and imaged for high-quality images for subcellular localization. However, numerous lower magnification images were acquired and assessed (every brain lobe for volume, cell division and cell death; n>20)

ERG recordings

Rh1-GAL4 was crossed to each ZIKV expression line, progeny were collected and aged at 25°C with a 12 h/12 h light/dark cycle (500 lux) for 30 days. ERG recordings were performed as follows: flies were immobilized on glass slides with Elmer's glue, a glass reference electrode filed with 100 mM NaCl was inserted behind the eye and a sharp recording electrode filled with 100 mM NaCl was placed on the eye. Light flashes of 1 s followed by 1 s recovery were delivered using a halogen lamp. Eight to ten flies were tested for each genotype. Data were recorded and analyzed using AxoScope pClamp (Molecular Devices).

Transmission electron microscopy

Retinas were imaged following standard electron microscopy procedures. Whole heads were dissected and fixed in 2% paraformaldehyde and 2.5% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.2) overnight on a rotator, rinsed with water three times, and post fixed with 1% aqueous osmium tetroxide. Samples were rinsed again three times with water and moved through an ethanol dehydration series (30-100%) with propylene oxide as the final wash. Samples were gradually permeated with three ratios of propylene oxide and Embed 812 (Electron Microscopy Sciences) and finally three changes of pure resin overnight on a rotator. Retinas were embedded in flat silicone molds and cured at 62°C in an oven for 3 days. Samples were then sectioned, stained with 1% uranyl acetate for 10 min, stained with lead citrate for 1 min, then examined by transmission electron microscopy using a JEOL JEM 1010 transmission electron microscope (JEOL USA) at 80 kV. An AMT XR16 CCD camera with exposure of 1100 ms, gain 2.7, bin 1 and sigma 1.65 was used.