Ethics and regulatory information

All experiments were conducted in compliance with national guidelines and with European Commission Directive 2000/54/EC of 18 September 2000 on the protection of workers from risks related to exposure to biological agents at work, and with Directives 2009/41/EC of 6 May 2009 and 98/81/EC of 12 June 1989 on the contained use of genetically modified micro-organisms. The generation of chimeric viruses and their use in experimental infections of mosquitoes were approved by the French Ministry of Higher Education, Research, and Technology (authorization number 8933, 4 August 2021) and the Institut Pasteur Dual Use Liaison Group (DURC 2021-03, 8 March 2022). Wild mosquito eggs were collected and exported with permission from local institutions and/or governments as required (Uganda: permit 2014-12-134, 2014; Cape Verde: authorization No 988, 2020).

Cells

Huh-7 and Vero E6 cells were maintained at 37 °C under 5% CO₂ in high-glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS; Gibco Thermo Fisher Scientific) and 1% penicillin/streptomycin (pen/strep; Gibco Thermo Fisher Scientific). C6/36 cells were maintained at 28 °C in Leibovitz’s L-15 medium (L15; Gibco Thermo Fisher Scientific) containing 10% FBS, 2% tryptose phosphate broth (TPB; Gibco Thermo Fisher Scientific), 1× non-essential amino acids (NEAA; Gibco Thermo Fisher Scientific), and 1% pen/strep.

Wild-type viruses

Wild-type ZIKV strain Kedougou2011, referred to as the iSenegal strain in this study, was isolated near Kedougou in 2011 from a pool of wild-caught mosquitoes⁶⁸. Wild-type ZIKV strain THA/2014/SV0127-14, referred to as the iThailand strain in this study, was isolated from a human serum sample⁶⁹. Virus stocks were prepared in C6/36 cells and the viral genome sequences were obtained by high-throughput sequencing as previously described^27,70. Wild-type ZIKV strains iSenegal and iThailand were converted by reverse genetics into rSenegal and rThailand strains, respectively, as described below for chimeric viruses.

Chimeric viruses

Parental and chimeric ZIKV strains were generated by circular polymerase extension reaction (CPER) following a published method⁷¹ with modifications. For each virus, a total of six ZIKV complementary DNA (cDNA) fragments covering the full-length viral genome were amplified utilizing PrimeSTAR GXL DNA Polymerase (TaKaRa Bio) and cloned into the pMW118 vector (listed in the Reagent Table). A CMV linker for ZIKV was then inserted into the pCR-Blunt II-TOPO vector, encoding sequences of the HDVr, Late SV40 pA signal, and CMV promoter. All DNA inserts were verified through Sanger sequencing. Following the cloning process (with some optimizations shown in Supplementary Fig. S1A), infectious cDNA was generated by assembling DNA fragments F1-F6, which encompass the entire ZIKV genome, and fragment F7, encoding HDVr, SV40 pA signal, and CMV promoter. The fragments were amplified, and the chimeric viral genomes were introduced using cloning plasmids as templates along with specific primer sets detailed in Supplementary Table S5. All the DNA fragments were designed to have complementary ends with a 49- to 95-nucleotide overlap. Equimolar amounts (0.1 pmol each) of the resulting DNA fragments F1-F7 were mixed in 50-μl reaction volumes of PrimeStar GXL with 2 μl of DNA polymerase. CPER was carried out with an initial 2 min of denaturation at 98 °C; 20 cycles of 10 s at 98 °C, 15 s at 55 °C, and 12 min at 68 °C; and a final extension for 12 min at 68 °C. The resulting CPER products, encoding the CMV promoter, full-length ZIKV genome sequence, followed by HDVr and SV40 poly(A) signal, were directly transfected into Huh-7 cells using Trans IT LT-1 (Mirus) following the manufacturer’s protocol. The culture supernatants from Huh-7 cells were collected and inoculated onto C6/36 cells. After three passages in C6/36 cells, the virus stocks were centrifuged to remove cell debris and stored at −80 °C for future use. Virus stock titers in plaque-forming units/ml (PFU/ml) were determined by plaque assay and full-genome sequences were confirmed by high-throughput sequencing as described below.

Viral genome sequencing

The full-length sequence of parental and chimeric ZIKV genomes was determined by Illumina sequencing as previously described⁷². Briefly, total RNA was extracted from a virus stock using QIAamp Viral RNA Mini Kit (Qiagen) and treated with Turbo DNase (Invitrogen). After the RNA purification with RNAClean XP beads (Beckman Coulter), cDNA was synthesized using M-MLV Reverse Transcriptase (Invitrogen), random hexameric primers (Roche) and RNaseOUT Recombinant Ribonuclease Inhibitor (Thermo Fisher Scientific) according to the manufacturer’s protocol. Double-stranded DNA (dsDNA) was produced with Second-Strand Synthesis Buffer (New England Biolabs), E. coli DNA ligase (New England Biolabs), E. coli DNA polymerase I (New England Biolabs) and E. coli RNase H (New England Biolabs), followed by DNA purification with AMPure XP beads (Beckman Coulter). The dsDNA was quantified by Qubit dsDNA HS assay kit (Invitrogen) and used for library preparation with a Nextera XT Library preparation kit (Illumina) according to the manufacturer’s instructions. The final libraries were checked with Bioanalyzer high sensitivity DNA analysis (Agilent) and sequenced on an Ilumina NextSeq 500 instrument (150 cycles, paired ends) with NextSeq 500/550 v2.0 Kit (Illumina). Adapters and low-quality sequences of raw reads were removed using Trimmomatic v0.39⁷³. The trimmed reads were assembled using megahit v1.2.9⁷⁴ with default parameters. The contigs were queried against the NCBI non-redundant protein database using DIAMOND v2.0.4⁷⁵, to look for potential contaminants in addition to the detected ZIKV genome. ZIKV scaffolds were constructed using the longest assembled contig and a viral sequence obtained in the previous study²⁷. The scaffolds were used to map the trimmed reads, using clc-assembly-cell v5.1.0. The consensus sequence generation and intrasample variant analysis were performed with ivar v1.0⁷⁶ using a minimum of 5× read depth of coverage for the consensus, and 500× with a 2% minimum threshold for minor variants. Samtools v1.10⁷⁷ was used to sort the aligned BAM files and generate alignment statistics. The mapping data was visually checked to confirm the accuracy of the obtained genomes using Geneious Prime 2023 (www.geneious.com). The raw reads were deposited in the Sequence Read Archive under bioproject PRJNA1199883, and the full-length consensus genome sequences of the viruses were deposited in GenBank under accession numbers PQ869243-PQ869266.

Plaque assay

Vero E6 cells were seeded in 24-well plates one day prior to virus inoculation. Ten-fold serial dilutions of the samples were prepared in DMEM and inoculated onto the confluent Vero E6 cells after removing the cell-culture supernatant. After 1-h incubation at 37 °C, the inoculum was replaced by DMEM containing 1.0% carboxymethylcellulose (Avantor, VWR), 1% FBS, and 1% pen/strep and the cells were incubated for 7 days at 37 °C. To count plaque numbers, the cells were fixed with 4% formaldehyde solution (Sigma) and stained with 0.2% crystal violet (Sigma).

Viral growth kinetics in vitro

C6/36 cells were seeded in 6-well plates one day prior to virus inoculation. After removing the cell-culture supernatant, the confluent cells were inoculated with ZIKV at a multiplicity of infection (MOI) of 0.01 for 1 h. The virus inoculum was replaced by fresh L15 medium containing 2% FBS, TPB, NEAA, and 1% pen/strep (2% FBS-L15 medium) and the cells were incubated at 28 °C. At 24, 48, 72, and 96 h post infection (h.p.i.), culture supernatants were collected, and infectious titers were determined by plaque assay as described above.

Virus attachment assay

C6/36 cells were seeded in 24-well plates one day before virus inoculation. The confluent cells were pre-chilled on ice for 15 min before the start of the assay. After removing the cell-culture supernatant, ZIKV was added to the cells at an MOI of 1 to allow virus attachment. After 1-h incubation on ice, the supernatant was collected for RNA extraction with QIAamp Viral RNA Mini Kit (Qiagen). The cells were rinsed twice with pre-chilled PBS before being lysed using TRIzol (Invitrogen). Total RNA was then extracted following the manufacturer’s protocol. Viral RNA was quantified by quantitative reverse transcription PCR (RT-qPCR) for ZIKV on total RNA using GoTaq Probe 1-step RT-qPCR kit (Promega), following the manufacturer’s protocol. The primers, probes, and gBlocks utilized in the RT-qPCR for ZIKV are listed in Supplementary Table S6.

Virus internalization assay

C6/36 cells were seeded in 24-well plates and pre-chilled on ice for 15 min before the start of the assay. As controls, cells were pre-treated with Dynasore (Sigma) at a final concentration of 100 μM in 2% FBS-L15 medium or with an equivalent dilution of DMSO for 1 h before chilling. After removing the cell-culture supernatant, ZIKV was added to the cells at an MOI of 1 and incubated on ice for 1 h to allow virus attachment. The cells were washed with PBS to remove unbound virus particles, and fresh 2% FBS-L15 medium containing the same concentrations of Dynasore or DMSO was added. The virus was then allowed to internalize into the cells for 3 h at 28 °C. After this period, the cells were washed with PBS and treated with protease E (5 mg/ml, Sigma) for 15 min on ice to remove any viruses remaining on the cell surface. The cells were washed 3 times with PBS and subsequently lysed with TRIzol (Invitrogen) for RNA extraction. The levels of viral RNA were quantified in total RNA by RT-qPCR using GoTaq 1-step RT-qPCR kit (Promega). Actin expression was assessed as an internal control using the same RT-qPCR method and the primer set listed in Supplementary Table S6. The efficiency of virus internalization, measured at 3 h.p.i, was determined by calculating the ratio of intracellular viral RNA to Actin expression. This ratio was normalized to the ratio of viral RNA to Actin expression from cells sampled at 0 h.p.i. without protease E treatment.

Quantification of ZIKV genomic and antigenomic RNA

C6/36 cells were seeded in 24-well plates and pre-chilled on ice for 15 min before the start of the assay. The confluent cells were inoculated with ZIKV at an MOI of 1 for one hour on ice to allow virus attachment. After one hour, the cells were washed with PBS and fresh 2% FBS-L15 medium was added. The cells were then incubated at 28 °C, and samples were collected at 0, 3, 6, 9, 12, 15, 18, 21, and 24 h.p.i. Each time point involved washing the cells with PBS, treating them with 5 mg/ml protease E for 15 min on ice, and lysing in TRIzol for total RNA extraction. Viral genomic and antigenomic RNAs were quantified using strand-specific RT-qPCR as previously described^78,79, with minor optimizations. The standard curve for antigenomic RNA was constructed by amplifying qPCR targets using primers listed in Supplementary Table S6, followed by reverse transcription using the MEGAscript T7 Transcription kit (Invitrogen) and purification using the MEGAclear Transcription Clean-Up kit (Invitrogen) as per the manufacturer’s guidelines. RNA concentrations were measured using a Nanodrop spectrophotometer, and 10-fold serial dilutions were made for standard curves. For specific incorporation of the tag sequence into cDNA, total RNA extracted from ZIKV-infected cells was reverse transcribed using M-MLV Reverse Transcriptase (Invitrogen) with RT-specific primers (Supplementary Table S6). Quantification of cDNA was then performed using the GoTaq probe qPCR Master Mix (Promega) and a set of strand-specific and tag-specific primers listed in Supplementary Table S6. At each time point, the ratio of viral antigenomic or genomic RNA to Actin expression was calculated as described above. To assess replication efficiency, relative viral antigenomic RNA was normalized against the ratio at 12 h.p.i., and relative genomic RNA against the ratio at 0 h.p.i.

Infectious titer/viral genome ratio

C6/36 cells were seeded in 24-well plates one day prior to virus inoculation. The confluent cells were inoculated with ZIKV at an MOI of 1 for one hour at 28 °C. After one hour, the inoculum was replaced by fresh 2% FBS-L15 medium. Cell-culture supernatant was sampled at 24, 48, 72, and 96 h.p.i. to determine the infectious titer by plaque assay and the concentration of viral genomic RNA by RT-qPCR, as described above.

Virus decay

ZIKV was prepared at a titer of 10⁶ PFU/ml and incubated at 28 °C for 96 h. Samples were collected after 0, 6, 12, 24, 48, 72, and 96 h, to determine infectious titer by plaque assay, as described above.

Cell viability assay

C6/36 cells were seeded in 24-well plates one day prior to virus inoculation. The confluent cells were inoculated with ZIKV at a MOI of 0.01 for 1 h at 28 °C. After the 1-h incubation, the inoculum was replaced by fresh 2% FBS-L15 medium. At the designated time points, the cell-culture supernatant was removed and CellTiter-Glo reagent (Promega) was added to the cells. Following a 10-min incubation at room temperature (20–25 °C), luminescence was measured using a GloMax 96 microplate luminometer (Promega) to quantify ATP, which is indicative of cell viability. For each viral infection, the cellular ATP levels at the indicated time points were normalized to the ATP levels of cells sampled at 0 h.p.i.

Mosquitoes

Ae. aegypti colonies were originally established from wild specimens caught in Colombia in 2017, Uganda in 2015, and Cape Verde in 2020, as previously described^17,50. Mosquitoes were reared under controlled insectary conditions (28°  ±  1 °C, 12-h light/dark cycle and 70% relative humidity). Prior to performing the experiments, their eggs were hatched synchronously in a vacuum chamber for one hour. Larvae were reared in dechlorinated tap water supplemented with a standard diet of TetraMin fish food (Tetra). Adults were kept in 30 × 30 × 30-cm BugDorm-1 insect cages (BugDorm) with permanent access to 10% sucrose solution. Mosquito experimental infections were performed with the 16^th-19^th, 23^rd, and 9^th laboratory generations of the colonies from Colombia, Uganda, and Cape Verde, respectively.

Mosquito oral exposure to ZIKV

Mosquitoes were orally exposed to ZIKV by membrane feeding in a biosafety level 3 containment facility. Briefly, 5- to 7-day-old female mosquitoes were starved for one day prior to the oral challenge. The infectious blood meal comprised a 2:1 mixture of washed rabbit erythrocytes (BCL) and ZIKV suspension, supplemented with 10 mM adenosine triphosphate (Sigma) and 0.1% sodium bicarbonate (Sigma). Mosquitoes were allowed to feed on the infectious blood meal for 15 min via a membrane-feeding apparatus (Hemotek Ltd.) with porcine intestine serving as the membrane. After feeding, fully engorged females were sorted on ice, transferred to 1-pint cardboard containers, and maintained under controlled conditions (28 °  ±  1 °C, 12-h light/dark cycle with 70% relative humidity) within a climatic chamber, with permanent access to 10% sucrose solution. The infectious titer of the blood meal was verified by plaque assay as described above.

Salivation assay

Mosquitoes were paralyzed with triethylamine (Sigma) for 5 min at 7, 10, and 14 days post infectious blood meal to collect saliva in vitro as previously described²⁷. Briefly, after removal of all legs, each mosquito’s proboscis was inserted into a 20-μl pipet tip containing 10 μl of FBS. The mosquitoes were allowed to salivate into this medium for 30 min. The saliva-containing FBS was then collected, combined with 40 μl of 2% FBS-DMEM supplemented with 4% Antibiotic-Antimycotic 100× (Life Technologies), and stored at −80 °C for subsequent titration by focus-forming assay, as described below. After salivation, the heads and bodies of each mosquito were dissected and individually transferred to 300 μl of squash buffer, composed of 10 mM Tris pH 8.0, 50 mM NaCl, and 1.27 mM EDTA pH 8.0 (all from Invitrogen), supplemented with 0.35 mg/ml proteinase K (Eurobio Scientific). Head and body samples were stored at −80 °C for subsequent testing by RT-PCR, as described below.

Focus-forming assay

Vero E6 cells were seeded in 96-well plates one day prior to virus inoculation. Serial 10-fold dilutions of the samples were prepared in DMEM (except for saliva samples that were used undiluted) and inoculated onto the confluent cells after removal of the cell-culture supernatant. Following an incubation period at 37 °C for 1 h, the inoculum was replaced with DMEM containing 1.0% carboxymethylcellulose, 1% FBS, 1% pen/strep, and 4% Antibiotic-Antimycotic 100×. The cells were further incubated for 5 days at 37 °C before being fixed with a 4% formaldehyde solution. For immunostaining, the fixed cells were permeabilized with 0.1% Triton X-100 in PBS (Sigma) for 10 min, blocked with 1% bovine serum albumin (Sigma) in PBS, and then incubated with a 1:1000 dilution of mouse anti-flavivirus group antigen monoclonal antibody clone D1-4G2-4-15 (Merck) in PBS for 1 h at room temperature (20–25 °C). Following three washes with PBS, the cells were incubated with Alexa Fluor 488-conjugated goat anti-mouse IgG (Life Technologies) at a 1:1000 dilution in PBS for 1 h at room temperature. Fluorescent foci were visualized using an EVOS FL fluorescence microscope (Thermo Fisher Scientific) equipped with appropriate barrier and excitation filters.

Detection of viral RNA by qualitative RT-PCR

To assess the presence of viral RNA in mosquito heads and bodies, the samples were homogenized for 30 s at 6000 rotations per minute in a Precellys 24 grinder (Bertin Technologies). A 100-μl aliquot of the homogenate was transferred to a PCR plate and subjected to crude RNA extraction by incubation for 5 min at 56 °C followed by 10 min at 98 °C. Total RNA was then used to synthesize cDNA using M-MLV Reverse Transcriptase, RNaseOUT Recombinant Ribonuclease Inhibitor, and random hexameric primers, following the manufacturer’s instructions. The resulting cDNA was amplified by PCR utilizing DreamTaq DNA polymerase (Thermo Fisher Scientific) with two sets of primers: pair 1 comprised ZIKV-PCR-F (5’-GTATGGAATGGAGATAAGGCCCA-3’) and ZIKV-PCR-R (5’-ACCAGCACTGCCATTGATGTGC-3’), and pair 2 consisting of ZIKV-PCR-F and ZIKV-PCR-R1 (5’-TCGTATTGCCAACCAGGCCAAAGC-3’). PCR cycling conditions were set as follows: an initial denaturation for 2 min at 95 °C, followed by 35 cycles of 30 s at 95 °C, 30 s at 55 °C, and 90 s at 72 °C, concluding with a final extension of 7 min at 72 °C. The PCR products were subsequently visualized via electrophoresis on a 1.5% agarose gel.

Mosquito dose-response curves

To estimate the 50% oral infectious dose (OID₅₀) of mosquitoes, dose-response curves were generated by preparing infectious blood meals containing 10⁴, 10⁵, and 10⁶ PFU/ml of ZIKV. Mosquitoes were orally exposed to ZIKV as described above. At 3- and 7-days post blood feeding, whole mosquito bodies were individually homogenized in 300 µl of 2% FBS-L15 supplemented with 4% Antibiotic-Antimycotic 100×. From each homogenate, 150 µl were used for total RNA extraction using the NucleoSpin96 kit (Macherey-Nagel), following the manufacturer’s protocol. Viral RNA detection was performed by qualitative RT-PCR as described above. The remaining homogenates were stored at −80 °C for subsequent titration by focus-forming assay, as described above. The OID₅₀ estimate was calculated from the dose-response curves using the drc package in R v.4.2.2 (www.r-project.org).

Statistical analyses

The prevalence of ZIKV infection, dissemination, and transmission in mosquitoes was analyzed by logistic regression as a function of experiment, ZIKV strain, and time. The initial statistical model included all their interactions, which were removed from the final model if their effect was non-significant (p > 0.05). Time was considered a continuous variable. To account for small, uncontrolled differences in virus concentration, the log₁₀-transformed blood meal titer was also included in the initial model as a covariate and removed from the final model if non-significant (p > 0.05). For in vitro assays, statistical significance was determined by two-tailed Student’s t test or one-way analysis of variance (ANOVA) followed by Dunnett’s test. Statistical analyses were performed in JMP v.14.0.0 (www.jmp.com) and R v.4.2.2 (www.r-project.org).

Model of in vitro viral dynamics

A logistic growth curve was employed to establish a model for predicting infectious viral titer (V_t) in cell culture at time t post infection defined as:

where s is the starting virus concentration, r is the growth rate, and k is the carrying capacity. Empirical data of viral growth kinetics in mosquito cells infected with the first set of chimeric viruses was used to approximate values for s, r, and k, enabling replication of dynamics observed in the laboratory experiments (Fig. 2C). To simplify the modeling process and avoid complex statistical fitting, s and k were approximated based on the viral titer at the start and end of the experiments, respectively. The growth rate r was adjusted to ensure a reasonable match between model predictions and empirical growth curves.

Model of in vivo viral dynamics

Following a previous study⁸⁰, the in vitro model of Eq. (1) was extended to simulate virus propagation within the mosquito midgut, hemocoel, and salivary glands. A probability of midgut infection (β) was incorporated to reflect the fact that only a few virions typically initiate infection out of thousands in the infectious blood meal^81,82. This approach introduces a stochastic element into the model to reflect demographic variability observed in oral experimental infections of mosquitoes. The model entails six probabilistic processes (Table 2) affecting three key stages: i) initial probability (β) of midgut infection from virus in the blood meal, before viral clearance from the blood meal (µ); ii) viral replication (r), constrained by a carrying capacity (k) in the midgut and hemocoel; and iii) viral ‘escape’ (λ) from the midgut to the hemocoel (M:H) and from the hemocoel to the salivary glands (H:S). G_v, M_v, H_v, and S_v represent the virus levels in the blood meal, midgut, hemocoel, and salivary glands, respectively. The stochastic dynamics were simulated using the tau-leap version of the Gillespie algorithm⁸³. In the simplest model scenario (Table 1), the parameters for r, k, and λ were assumed to be the same across tissues.

Sensitivity analysis of the in vivo model

A sensitivity analysis was conducted to explore the effects of varying parameter values on the probability of midgut infection as a function of starting virus concentration, and of the probability of systemic dissemination and transmission as a function of time. Empirical dose-response curves were used to inform on ranges of parameter values that could potentially produce results qualitatively to match those observed in the experiments. The stochastic model was run 30 times for each subset of parameters to simulate viral dynamics in 30 individual mosquitoes. Midgut infection simulations ran for 124 hourly time steps, whereas dissemination and transmission simulations ran for 360 hourly time steps. For midgut infection, Latin hypercube sampling generated 1000 parameter sets within the following ranges: μ from 1/72 to 1/12 h⁻¹; β from 0 to 0.0001; r from 0.001 to 0.1 h⁻¹; and k from 10³ to 10²⁰ PFU/ml. The proportion of mosquitoes developing a midgut infection was determined as the proportion of simulations where virus levels in the midgut exceeded zero (M_v > 0) at the end of the 124 hourly time steps. G_v was set at 10⁶ PFU/ml, adjusted by multiplying by 0.003, reflecting the average mosquito blood meal volume⁸⁴. Using the same sampling approach, 1000 parameter sets were generated for viral dissemination (hemocoel) and transmission potential (salivary glands), adopting the maximal and minimal values of μ and β that resulted in successful midgut infections from earlier simulations. Parameter ranges for r and k remained unchanged. λ was set between 10⁻⁸ and 10⁻⁵ h⁻¹. The outcome was determined as the proportion of simulations showing established infections in the hemocoel (H_v ≥ 1), or salivary glands (S_v ≥ 1), over intervals of 24 time steps (equivalent to per day), taking the last output for each daily time step. Evaluation threshold was set at 1000 PFU/ml to confirm infection establishment.

Simulating the in vivo data with the model

Initially, the model was used to reproduce, qualitatively, the dose-response curves of ZIKV infection prevalence observed experimentally. By adjusting β of at least one virion infecting the midgut epithelium, the model was calibrated to match empirical observations with the first set of chimeric viruses. Each value of β was tested over 124 hourly time steps in 100 simulations across varying G_v values from 10² and 10⁸ PFU/ml to quantify the proportion of established midgut infections, as described above. Following the results from the sensitivity analyses, μ of 1/72 h⁻¹, r of 0.04 h⁻¹, k of 10²⁰ PFU/ml, and λ of 0.00005 h⁻¹ were kept constant. Next, parameter values of β derived from dose-response curves in vivo and of r estimated from growth kinetics in mosquito cells were combined to simulate virus dissemination to the hemocoel and salivary glands. Simulations aimed to explore five different scenarios to understand patterns of virus spread within mosquitoes and to provide biologically plausible hypotheses to explain the experimental data (Table 1). Of note, the process of salivary gland infection and virus escape into the saliva were collapsed into S_v in the simulations, whereas the experiments detected virus presence in expectorated saliva. The baseline model (scenario 1, Table 1) was used as a simplistic framework where uniform infection and growth parameters were assumed to enable comparisons against four more elaborate scenarios. Differences in the infection processes are expected, such as variations in r across different cell types⁸⁵, cell-to-cell viral spread within the midgut, and virus propagation via freely moving hemocytes in the hemocoel^52,53, and distinct anatomical barriers between the midgut, hemocoel, and salivary glands^86,87. Additional model parameterizations were explored to represent the observed tissue-specific infection patterns more accurately. These modifications were aimed at modeling widespread infection across simulations consistently, though not necessarily achieving transmission in every single mosquito or each simulation (Fig. 4). The four more complex scenarios were designed to reflect variations in processes between tissues that could be consistent with the observed data. Parameterizations were specifically targeted to result in a disseminated infection across all simulations but not necessarily leading to salivary gland infection in every instance (Fig. 4C). From the baseline scenario, the impact of varying parameterizations was demonstrated by running the model 100 times with G_v at 10⁶ PFU/ml. Throughout the scenarios, μ at 1/72 h⁻¹, β at 10⁻³, the midgut growth rate (r_M) at 0.04 h⁻¹, the escape rate from the midgut to the hemocoel (λ_M:H) at 0.0005 h⁻¹, and k of all tissues at 10²⁰ PFU/ml, were kept constant. For scenarios two through five, where the value of one parameter was reduced by a tenth to illustrate the impact on model outputs, variations were introduced in the hemocoel growth rate (r_H) and the escape rate from hemocoel to salivary glands (λ_H:S), using Gamma distributions. For r_H, a shape (α) of 400 and a scale (θ) of 0.025 were used, resulting in an average of 0.04/10 and a variance of 0.00001. For λ_H:S, α of 2500 and θ of 0.0004 were utilized, achieving an average of 0.0005/10 and a variance of 0.00000002. In scenarios three and five, parameter values were randomly selected from these distributions for each simulation. Additionally, in scenario five, the variance of the Gamma distribution was varied, with simulations run using a mean of 0.00008 and variances of 10^−7.5, 10^−7.3, 10⁻⁷, and 10^−6.5. The five model scenarios were compared to identify the qualitative differences in model outputs among the biologically relevant model structures and to determine which scenario was qualitatively most similar to the observed results. Consequently, a qualitative comparison between observed data and modeled outputs was conducted visually and by assessing the relative differences between the proportion of mosquitoes with a disseminated infection and the proportion capable of transmitting.

Biosafety statement

All work on ZIKV was performed in biosafety level 3 (BSL3) facilities at Institut Pasteur in accordance with the European and French legislation. All experiments were conducted in compliance with national guidelines and with European Commission Directive 2000/54/EC of 18 September 2000 on the protection of workers from risks related to exposure to biological agents at work. All personnel working with ZIKV were trained with relevant safety and protocol-specific procedures.

Risk-benefit analysis

This study was undertaken to enhance understanding of ZIKV, which is an ongoing threat to global public health. This work included the generation of chimeric viruses by substituting the segments of parental ZIKV genomes and their use in experimental infections of mosquitoes. Such basic investigations are widely accepted to benefit public health by increasing our understanding of the mechanisms of transmissibility. The generation of chimeric viruses and their use in experimental infections of mosquitoes were approved by the French Ministry of Higher Education, Research, and Technology (authorization number 8933, 4 August 2021) and the Institut Pasteur Dual Use Liaison Group (DURC 2021-03, 8 March 2022).

The construction of chimeric ZIKV from parental strains is used to study how segments of the viral genome affect viral growth and in vivo transmissibility. Such chimeric viruses have been constructed for many different arboviruses over the last 10+ years and are generally accepted to be safe and low risk because the chimeric viruses do not become more dangerous than the natural isolates. The genomes of RNA viruses like ZIKV are highly optimized due to their ability to efficiently adapt to animal hosts and mosquito vectors. Natural selection has favored viral variants that are best suited to their ecological niches, balancing infectivity, pathogenicity, and transmissibility. As a result of this evolutionary fine-tuning, the likelihood that constructing chimeric ZIKV strains in the laboratory will yield a version of the virus with dramatically enhanced characteristics is low. Such changes would likely have been naturally selected for if they conferred a significant advantage, meaning that most potentially advantageous mutations have already been explored and either retained or discarded by evolutionary pressures.

Most studies using chimeric arboviruses reported to date reveal in-between or inferior phenotypes than their parental strains. In general, chimeric viruses with reduced infectivity and pathogenicity in animals and lower transmissibility in mosquitoes are unlikely to become endemic in the wild. While our work offers significant new insights into the genetic basis of mosquito-borne transmission of ZIKV, it does not provide a road map for creating a form of the virus that is more infectious, pathogenic, or transmissible than what already exists in nature.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.