Abstract
Climate change and globalization contribute to the expansion of mosquito vectors and their associated pathogens. Long spared, temperate regions have had to deal with the emergence of arboviruses traditionally confined to tropical regions. Chikungunya virus (CHIKV) was reported for the first time in Europe in 2007, causing a localized outbreak in Italy, which then recurred repeatedly over the years in other European localities. This raises the question of climate effects, particularly temperature, on the dynamics of vector-borne viruses. The objective of this study is to improve the understanding of the molecular mechanisms set up in the vector in response to temperature.
We combine three complementary approaches by examining Aedes albopictus mosquito gene expression (transcriptomics), bacterial flora (metagenomics) and CHIKV evolutionary dynamics (genomics) induced by viral infection and temperature changes.
We show that temperature alters profoundly mosquito gene expression, bacterial microbiome and viral population diversity. We observe that (i) CHIKV infection upregulated most genes (mainly in immune and stress-related pathways) at 20°C but not at 28°C, (ii) CHIKV infection significantly increased the abundance of Enterobacteriaceae Serratia marcescens at 28°C and (iii) CHIKV evolutionary dynamics were different according to temperature.
The substantial changes detected in the vectorial system (the vector and its bacterial microbiota, and the arbovirus) lead to temperature-specific adjustments to reach the ultimate goal of arbovirus transmission; at 20°C and 28°C, the Asian tiger mosquito Ae. albopictus was able to transmit CHIKV at the same efficiency. Therefore, CHIKV is likely to continue its expansion in the northern regions and could become a public health problem in more countries than those already affected in Europe.
Introduction
The mosquito Aedes albopictus is an invasive species that, in <40 years, has largely extended its distribution range from tropical and subtropical regions to much cooler temperate areas, increasing the risk of mosquito-borne viral diseases in regions so far spared, such as Europe and Northern America.1 Although mainly considered a secondary vector for several arthropod-borne viruses (arboviruses), Ae. albopictus has became a major vector of chikungunya virus (CHIKV; RNA (+) virus; Alphavirus, Togaviridae) in recent outbreaks.2,3 It was first detected in Europe in 19794 and then in 1990,5 Ae. albopictus is today present in >20 European countries.6
The circulation of arboviruses between vertebrate and invertebrate hosts is driven by the need for arthropod vectors to obtain a blood source to complete their life cycle. During a blood meal on a viraemic host, female mosquitoes ingest the virus along with the blood. Competent mosquitoes ensure an active viral replication resulting in virus infection, dissemination and transmission. Since mosquitoes are poikilothermic ectotherms, temperature is an overriding environmental factor affecting different mosquito life traits and biological processes, which may influence arbovirus transmission.7–10 However, the molecular mechanisms set up in response to temperature within the vector remain poorly understood.
The vectorial system results from the interaction of different compartments. The regulation of gene expression, particularly those related to immunity,11 the bacterial microbiota12 and the genetic properties of the virus itself13 are key parameters able to respond to environmental constraints in order to promote the functioning of the vectorial system. In the current climatic context and faced with the recurrence of arbovirus transmission events in Europe,14 we used the model Ae. albopictus infected with the CHIKV to investigate the effects of temperature (20 and 28°C) on mosquito physiology using transcriptomics and metagenomics, and viral populations using a genomic approach. With this multi-omics study, we attempted to provide an integrated view of the temperature-induced changes within the vector system.
Materials and methods
Two constant temperatures were used in experiments: 20 and 28°C. The temperature of 20°C was chosen as representative of the minimal mean monthly temperature required for CHIKV transmission15 and 28°C as a threshold below which the transmission of another emerging arbovirus in Europe, dengue virus, was not observed.16
Ethics statement
Mice were housed at the Institut Pasteur animal facilities (Paris) accredited by the French Ministry of Agriculture for performing experiments on live rodents. Work on animals was performed in compliance with French and European regulations on care and protection of laboratory animals (EC Directive 2010/63, French Law 2013–118, 6 February 2013). All experiments were approved by the Ethics Committee #89 and registered under the reference APAFIS (Autorisation de Projet utilisant des Animaux à des Fins Scientifiques) #6573-201 606 l412077987 v2.
Cells
Ae. albopictus cells (U4.4 and C6/36) were maintained in Leibovitz L-15 medium (Gibco, Thermo Fisher Scientific, Massachusetts, USA) with non-essential amino acids (1X) (Gibco, Thermo Fisher Scientific), penicillin/streptomycin (100 U/ml and 100 μg/ml, respectively) (Gibco, Thermo Fisher Scientific) and supplemented with 10% of fetal bovine serum (FBS) (Eurobio Scientific, Ulis, France). Aedes aegypti cells (Aag2) were maintained in Schneider’s medium (Gibco, Thermo Fisher Scientific) with L-glutamine (Gibco, Thermo Fisher Scientific), penicillin/streptomycin (100 U/ml and 100 μg/ml, respectively) and supplemented with 10% of FBS. U4.4 and Aag2 cell lines were used for CHIKV serial passages; U4.4 cells were also used for virus growth curves. C6/36 cell line is an RNAi-deficient cell line and was used for virus amplification and titration.
Viral clones
CHIKV (Alphavirus, Togaviridae) detected on La Réunion Island in 2005–0617 was an East-Central-South-African (ECSA) genotype presenting a substitution Ala>Val at position 226 of the E1 surface glycoprotein; the E1–226V variant was preferentially transmitted by Ae. albopictus.18,19 Two infectious CHIKV clones, derived from the LR2006-OPY1 CHIKV (La Réunion, ECSA genotype, DQ443544.2), were used in this study.20 The two clones differed by a point mutation at the amino acid 226 (Ala versus Val) of the E1 envelope glycoprotein. The viruses derived from E1–226V to E1–226A plasmids were rescued in BHK-21 cells (Baby Hamster Kidney, ATCC) and stored at −80°C prior use. Viral titres estimated by focus-forming assays were 109.15 focus-forming units per ml (FFU/ml) for the variant E1–226V and 108.34 FFU/ml for the variant E1–226A.
Mosquitoes
A. albopictus eggs were sampled in July 2018 in Montpellier, France, using ovitraps (1047 individuals were used to start the colony). The location of Montpellier was chosen because autochthonous CHIKV cases were detected there in 2014.21 In insectaries, eggs were hatched in 1 L of dechlorinated tap water. Larvae were split in pans of 200 individuals and were fed with yeast tablets (Gayelord Hauser). Water and food were renewed every 2–3 days. Immature stages were maintained at 25 ± 1°C. Emerging adults were reared in cages at 28 ± 1°C with a 12:12-light:dark cycle, 80% relative humidity and fed with a 10% sucrose solution ad libitum. To amplify the population, females were blood-fed three times a week on anaesthetized mice (OF1 mice, Charles River laboratories, France). The F4–6 generations in laboratory were used for oral infection experiments.
Virus titration
Monolayers of C6/36 cells were inoculated with 50 μl of diluted samples (viral production or mosquito saliva) in 96-well plates. After an incubation of 1 h at 28°C, cells were overlaid with 1:1 mix of 4% carboxymethyl cellulose and Leibovitz L-15 medium supplemented with 5% FBS and 1.5X of anti-biotic-anti-fungal solution (Life Technologies, California, USA). Following an incubation period of 3 days at 28°C, cells were fixed with 3.4% formaldehyde, washed three times in PBS 1X and stained using a hyperimmune ascetic fluid specific to CHIKV as the primary antibody and an Alexa Fluor 488 goat anti-mouse IgG as the secondary antibody (Life Technologies). The number of FFU was determined under a fluorescent microscope.
Mosquito oral infections
Boxes of ~60 10-to-15-day-old females were sorted and transferred to a biosafety level 3 (BSL-3) insectary (28 ± 1°C) for a 24-h period of acclimatization and starvation. Mosquitoes were then exposed to a CHIKV-free or CHIKV-infected (107 FFU/ml) blood meals. The blood meal contained 2/3 of washed rabbit erythrocytes, 1/3 of media or viral suspension and ATP as a phagostimulant at a final concentration of 10 mM. Fully engorged females were sorted and placed in climatic chambers set at 20°C or 28 ± 0.1°C with a 14:10 diurnal cycle, 80 to 90% relative humidity and access to 10% sucrose ad libitum. Saliva was collected using the forced salivation technique. Briefly, mosquito legs and wings were removed, and proboscis was inserted into a micropipette tip filled with 5 μl of FBS. After 45 min, the saliva-containing FBS was expelled in 45 μl of Leibovitz L-15 medium (Invitrogen, Massachusetts, USA). Following salivation, saliva samples and mosquito bodies were kept at −80°C until further processing. Mosquitoes were examined 9 days after CHIKV-infectious blood meals when the proportion of mosquitoes with viral particles detected in saliva (i.e. transmission efficiency; TE) showed an atypical profile, becoming higher at 20°C than at 28°C (Figure 4).
Mosquito transcriptome
Experimental conditions
Batches of mosquitoes were orally challenged with either the CHIKV infectious clone-derived E1–226V virus or a CHIKV-free blood meal. Mosquitoes from each group (CHIKV-uninfected or CHIKV-infected) were placed at 20 or 28°C for a 9-day incubation. Then, saliva of CHIKV-infected mosquitoes were collected (80 saliva; 40 from mosquitoes held at 20°C and 40 from those at 28°C) to identify individuals capable of transmitting the virus and to assess virus TE at each temperature. To avoid biases from salivation step, mosquitoes fed on a CHIKV-free blood meal were also subjected to salivation. The transcriptomic study was based on comparisons between four groups: (i) CHIKV-uninfected mosquitoes at 20°C, (ii) CHIKV-uninfected mosquitoes at 28°C, (iii) CHIKV-transmitting mosquitoes at 20°C and (iv) CHIKV-transmitting mosquitoes at 28°C (Figure S1).
RNA extraction, library construction and sequencing
Six mosquito bodies (without legs and wings) were individually processed for each experimental condition, giving a total of 24 samples. Samples were homogenized with silica glass beads for 30 s in lysis buffer RA1 and β-mercaptoethanol using a homogenizer (Precellys 24®). Total RNA was extracted from homogenates using a RNA extraction kit (Nucleospin® RNA, Macherey-Nagel, Düren, Germany). RNA was eluted in 40 μl RNase-free water and kept at −80°C until further use. RNA integrity and concentration were evaluated on Agilent 2100 Bioanalyzer (Agilent Technologies, California, USA).
PolyA fraction of total RNA was purified from 10 μg RNA using Dynabeads according to the manufacturer’s instructions (Thermo Fisher Scientific). Libraries were built using a TruSeq Stranded mRNA library Preparation kit (Illumina, California, USA) following the manufacturer’s protocol. Quality control was performed on an Agilent BioAnalyzer. Sequencing was performed on a NovaSeq 6000 (Illumina) by Novogene to obtain 150 base paired-end reads.
Data analysis—differential gene expression and function
To characterize changes in the RNA transcriptome in response to temperature and CHIKV infection, we compared the RNA expression profiles between the four groups described above. The RNA-seq analyses were performed with Sequana v0.9.16 (https://github.com/sequana/sequana_rnaseq),22 a pipeline based on Snakemake 5.8.1.23
Reads were trimmed from adapters using Cutadapt 2.10 then mapped to the Ae. albopictus genome (GCF_006496715.1) using STAR 2.7.3a.24 FeatureCounts 2.0.025 was used to produce the count matrix, assigning reads to features using corresponding primary annotation (release 102) with strand-specificity information. Quality control statistics were summarized using MultiQC 1.8.26 Statistical analysis on the count matrix was performed to identify differentially regulated genes. Differential expression testing was conducted using DESeq2 library 1.24.027 scripts based on SARTools 1.7.0,28 indicating the significance (Benjamini–Hochberg adjusted P-values, false discovery rate FDR < 0.05) and the effect size (fold-change) for each comparison.
For functional annotation, we developed a Python script to scan for intersections between the ‘LOC + GeneID’ from the RNA-Seq pipeline files and gene annotations of the reference genome AalbF229 to automatically assign scaffolds, coordinates and predicted functions to differentially expressed genes (DEGs).
To help in interpreting our results, we classified a selection of genes according to 5 classes and 20 subclasses of functional interest (File S2). We focused our analysis on thermal tolerance, stress-related pathways (oxidative stress, cytoskeleton, apoptosis/autophagy, signalling pathway-regulation, lysosome, stress), other pathways (venom & allergen, pathogen interaction, salivary proteins, others), non-specific immune pathways (ubiquitin-related proteins, serpins, serine proteases, lipid related proteins, lectins, others) and classical pathways (immune pathways, pathogen sensors, pathogen elimination). Special attention was given to DEGs coding for uncharacterized proteins and non-coding RNA genes.
Mosquito microbiota
Experimental conditions
The bacterial microbiota study was conducted using the same experimental design as the transcriptomic study (see section above) and based on comparisons between six groups: (i) CHIKV-uninfected mosquitoes at 20°C, (ii) CHIKV-uninfected mosquitoes at 28°C, (iii) CHIKV non-transmitting mosquitoes at 20°C, (iv) CHIKV non-transmitting mosquitoes at 28°C, (v) CHIKV-transmitting mosquitoes at 20°C and (vi) CHIKV-transmitting mosquitoes at 28°C (Figure S1).
DNA extraction from carcass and midgut
Mosquito bodies were surface sterilized by washing in successive baths: water (1 min), ethanol 70% (2× 5 min), bleach 10% (1 min) and water (1 min). Due to a problem of detection sensitivity, we decided to work on pools of mosquitoes; mosquito midguts were then dissected, and DNA extraction was performed on pools of six mosquito carcasses (bodies without legs, wings and midguts) and pools of six mosquito midguts. Each group corresponded to a pool of six mosquito carcasses and six midguts. Samples were homogenized with silica glass beads for 20 s in 180 μl of lysis buffer T1 and 25 μL of proteinase K using a homogenizer (Precellys 24®). Total DNA was extracted from homogenates following instructions of the DNA extraction kit manual (Nucleospin® Tissue, Macherey-Nagel). DNA was eluted in 40 μl elution buffer and kept at −20°C until analysis. DNA concentration was evaluated on Qubit 4 Fluorometer (Thermo Fisher Scientific).
16S rDNA, short-read Illumina sequencing
Both DNA samples from carcasses and midguts were sequenced on an Illumina NextSeq500 (Helixio). The entire 16S gene (1500 bp) was amplified using the Q5 High-Fidelity PCR kit (New England BioLabs) with the pair of primers 27F (AGAGTTTGATCMTGGCTCAG) and 1492R (CGGTTACCTTGTTACGACTT) (EzBioCloud). Libraries were made using the ‘NEBNext® Ultra™ II FS DNA Library Prep for Illumina’ kit (New England BioLabs) from 50 ng of DNA. Sequencing was performed in double reading on 150 bp according to the ‘Paired-end Sequencing’ protocol (Illumina). The quality of the run was checked during demultiplexing with the software ‘bcl2fastq 2.0’ (Illumina). Quality control was made on the concatenated Fastq files with the software, ‘FastQC v0.11.3’ (Babraham Institute).
Short-read Illumina processing, clustering and taxonomic assignment
Midgut and carcass short reads were processed separately by using the standard operating procedure available in mothur (v 1.44.3).30 Briefly, forward and reverse reads were joined into contigs. Contigs smaller than 150 bp, longer than 300 bp, containing Ns (ambiguous sequences) or containing homopolymers larger than 5 bases, were removed. Contigs were aligned on the SILVA v132 database. A filter was used to optimize the start and the end positions by choosing the 75th percentile for both, and then removed the overhangs at both ends. The sequences were classified using the default Naïve Bayesian Classifier31 and those classified as unknown, Chloroplast, Mitochondria, Archaea or Eukaryota were removed. The sequences were clustered into operational taxonomic units (OTUs) using the abundance-based greedy clustering algorithm with a cutoff of 0.05.32 At the end, the consensus taxonomy of each OTU was determined by looking at all the hierarchical taxonomies within an OTU and selecting the ones that best represents that OTU based on majority rule (up to 51%). Results were explored in R (v 4.0.3) (R Core Team, 2018) with the phyloseq package (v 1.37.0). Principal coordinates analysis (PCoA) of data was performed using a matrix of Bray–Curtis index, based on the relative abundance of clusters in samples. Spearman rank correlation between relative abundance of clusters and the first axis of the PCoA was calculated with the Hmisc package (v 4.6–0). Plots were created using the package ggplot2 (v 3.3.5).
16S rDNA, long-read nanopore sequencing
DNA samples from carcasses were sequenced by nanopore sequencing. Multiplexed nanopore ligation sequencing libraries were made (two by sample) by using the SQK-16S024 barcode kits, following the 16S Barcoding kit 1–24 protocol (Oxford Nanopore Technologies, Oxford, UK). MinION flow cells (Oxford Nanopore Technologies) were loaded with 75 μL of ligation library. The MinION instrument was run for ~8 h, until no further sequencing reads could be collected. Then, Fast5 files were preprocessed by base-calling using the High-Accuracy setting on the MinKNOW GUI (v 4.2.8) (Oxford Nanopore Technologies), and output sequence reads were saved as fastq files. Demultiplexing (barcode assignment) and trimming were then performed with Guppy (v 4.5.2). Reads were filtered to remove sequences shorter than 1250 bp, longer than 1950 bp and those with an average quality score lower than 7, using Nanofilt (v 2.8.0).33
Long read nanopore processing, clustering and taxonomic assignment
Filtered reads were passed on to NanoCLUST34 with −min_cluster_size 10 −cluster_sel_epsilon 0.5 −min_read_length 1400 and −max_read_length 1700 parameters. Briefly, NanoCLUST begins by clustering sequences based on 5-mers frequency, then creates a consensus sequence for each cluster and retrieves taxonomic information by using a blastn similarity search within a database. For this purpose, we created a BLAST database from the curated database, SILVA v132, which is the reference file used by mothur.35,36 An in-house script was used to retrieve TaxIDs that match the accession number of sequences in the file nucl_gb.accession2taxid available on NCBI (ftp://ftp.ncbi.nih.gov/pub/taxonomy/accession2taxid). If the accession number was not found, we used the file taxmap_slv_ssu_ref_132-corrected.txt (Silva repository) to retrieve the TaxID with the name of the species and the command name2taxid of the Taxonkit.37 Then, we used makeblastdb for creating the BLAST database based on the 213 119 SSU rRNA sequences. About 70% of the sequences of this new SSU-RNAs database have taxonomic information (TaxIDs) at genus level and half of them were at species level. Other sequences were at higher taxonomic level. However, due to a high misclassification rate of the BLAST hit taxonomy assignment at species level, we discarded the identification process with the sequence classifier of NanoCLUST. The Naïve Bayesian Classifier and the SILVA v132 reference file used by mothur were used to rapidly and accurately classify bacterial sequences.31 To control the accuracy of identification of species of interest, the consensus sequence and its closest relatives were aligned with muscle, and phylogenetic trees were built with PhyML, which was implemented in seaview (v 5.0.4).38
CHIKV experimental evolution
Serial passages in vitro
We used the two infectious clone-derived viruses described above (see section viral clones) to initiate serial passages on U4.4 and Aag2 cells (Figure S4). Subconfluent cells, prepared in 25 cm2 flasks, were infected in duplicates at a MOI of 0.1. After 1-h incubation at 28°C, cells monolayers were rinsed (with L-15 or Schneider medium depending on the cell line) and then covered with 5 ml of fresh medium supplemented with 5% FBS. Afterwards, cells were incubated at 20 or 28°C in climatic chambers (Binder, KB 53, Tuttlingen, Germany). Two days after, supernatants were harvested, centrifuged at 1000 g for 5 min and a solution of sodium bicarbonate (Gibco, Thermo Fisher Scientific) was added to adjust the pH to 7. From passage 2 to 40, passages were completed using 500 μl (1/10) of the previous passage as inoculum. From passage 31 to 40, the initial temperature was inverted; passages maintained at 20°C were placed at 28°C for 10 additional passages and vice versa. Genetic viral evolution was monitored by sequencing of passages 10, 20, 30 and 40. Titres of passages were assessed every five passages. Passages 30 (U4.4 cells) were amplified on C6/36 cells and used for mosquito oral infections and virus replication kinetics.
Analysis of viral populations
CHIKV evolutionary patterns associated with temperature were monitored by sequencing of parental viruses rescued from infectious clones (P0; E1–226V and E1–226A) and in vitro (P10, P20, P30 and P40) passages.
RNA extraction, amplicon library and sequencing
Viral RNAs were extracted using Nucleospin RNA II kit (Macherey-Nagel). Genomes were amplified using the reverse transcriptase Platinum® Taq High Fidelity polymerase enzyme (Thermo Fisher Scientific) with a set of primers presented in Table S4. PCR products were pooled in equimolar proportions. After Qubit quantification using Qubit® dsDNA HS Assay kit and Qubit 4.0 fluorometer (Thermo Fisher Scientific), amplicons were fragmented (sonication) into fragments of 200 bp long. Libraries were built adding barcode, for sample identification, and primers to fragmented DNA using AB Library Builder System (Thermo Fisher Scientific). To pool equimolarly the barcoded samples, a quantification step by real-time PCR using Ion Library TaqMan™ Quantitation kit (Thermo Fisher Scientific) was performed. An emulsion PCR of the pools and loading on 520 chip was achieved using the automated Ion Chef instrument (Thermo Fisher Scientific). Sequencing was performed using the S5 Ion torrent technology (Thermo Fisher Scientific) following manufacturer’s instructions. Low-quality reads/bases were filtered using Prinseq-lite version 0.20.4, using the following parameters: −min_len 150 −min_qual_mean 26 −ns_max_n 1 −derep 14 −trim_left_p 12 −trim_qual_left 26 −trim_right_p 16 −trim_qual_left 26 and, Shannon entropy (−lc.method dust −lc.threshold 50) to filter low complexity reads.39
Whole-genome sequence analysis of CHIKV populations
Single-nucleotide variant (SNV) with position, coverage, frequency and residue annotation (alternate nucleotide and amino acid, if changes occurred) was detected in viral populations by using the ViVan pipeline.40 A minimum coverage threshold of 500 was used for all the analyses. We set the minimum frequency threshold at 0.5% to distinguish ‘genuine’ substitutions and errors, because Ion Torrent PGM can detect substitutions occurring at frequencies ≥0.1%, but the total error rate (after trimming) of our samples ranges from 0.05 to 0.398% (i.e. a variant was considered to be ‘true’ if its frequency was greater than the cutoff). Virus RNA polymorphism was studied at two levels of SNV frequency in the population: from 0.03 to 0.1 for minor mutations and from 0.1 to 1 for main mutations. The analysis covered the entire viral genome excluding the first 18 nucleotides of the 5′UTR and the 88 nucleotides upstream the polyA tail.
Fast- and low-evolving CHIKV populations, hot-spots of synonymous mutations and amino acid changes in viral genomes were visualized by taking as reference the CHIKV variant E1–226A (P0-A) in SNPviewer (https://gitlab.pasteur.fr/plechat/snpviewer41,42). In addition, we performed group sample comparison to ensure reliability and specificity of the detection of SNVs. ViVan allowed us to compare viral populations sharing the same conditions of culture and temperature to highlight significant mutations (or punctual deletions) that were common to each replicate of one condition but were not found in any replicate of the other condition. Each SNV was selected as main or minor variant, if at least the frequency of one replicate reached the required threshold.
TE of viral variants by mosquitoes
Batches of mosquitoes were orally challenged with parental clone-derived stocks (E1–226V, E1–226A) and P30 viruses that were selected on U4.4 cells. Each batch was tested in duplicate. Following a CHIKV-infected blood meal, subsets of fully engorged females were placed at both temperatures, 20 or 28°C, regardless of the temperature at which the virus was maintained. Mosquito saliva were collected at 9 days post-infection (dpi) and titrated by focus fluorescent assay to evaluate transmission efficiencies.
Statistical analysis
Transmission rates and proportions were compared using Fisher’s exact test and saliva viral loads using Kruskall–Wallis non-parametric tests. The significance level for multiple tests was adjusted by the sequential method of Bonferroni.43 For bacterial communities, PCoA plots of Bray–Curtis dissimilarities were compared using Spearman correlation test. Tests were conducted using the STATA software (StataCorp LP, Texas, USA). P-values<0.05 were considered statistically significant.
Results
We experimentally infected Ae. albopictus mosquitoes with CHIKV which were then maintained at 20 or 28°C. At 9 dpi, mosquitoes were examined for changes in gene expression profiles and bacterial microbiota they host, along with the genetic changes in viral populations that were examined by in vitro experimental evolution.
Temperature strongly modifies gene expression profiles of CHIKV-transmitting mosquitoes
Regulation of gene expression is a common mechanism that organisms use to adapt their phenotypes and maintain fitness in response to environmental changes. We assessed mosquito transcriptional changes induced by temperature using a transcriptomic approach. Mosquitoes were exposed to a non-infectious or a CHIKV-infectious blood meal and examined 9 days after (Figure S1).
Responses to temperature of CHIKV-uninfected mosquitoes
First, we investigated the effect of temperature on gene expression of mosquitoes that received a non-infectious blood meal. We compared the transcriptome of mosquitoes exposed to a non-infectious blood meal and then held at 20 or 28°C. We identified a total of 296 DEGs; 125 were upregulated and 171 were downregulated at 20°C compared with 28°C (Figure 1A, Table S1, File S1). We found that the highest upregulated and downregulated transcripts corresponded to non-coding RNA genes (Figure 1B). The protein-coding gene with the highest enrichment at 20°C corresponded to the protein G12, whereas the most downregulated gene was the cytochrome P450 305a1 (Figure 1B). We defined five functional classes of genes to distribute DEGs according to their function, mainly associated with immunity, stress and thermal tolerance (File S2). At 20°C, most of the upregulated genes belonged to cytoskeleton dynamics (Figure 2A, File S1), whereas most of the downregulated genes were involved in oxidative stress (e.g. cytochrome P450) and non-specific immune responses (e.g. serine proteases) (Figure 2A, File S1). We observed that in absence of infection, temperature modulates the expression of different immune and stress-related pathways.
Overview of DEGs up/downregulated in Ae. albopictus mosquitoes. (A) Number of DEGs differentially expressed (in brackets). (B–E) Top 10 of DEGs ranked by Log2 fold-change in CHIKV-uninfected Ae. albopictus mosquitoes held at 20°C versus 28°C (B), CHIKV-transmitting mosquitoes held at 20°C versus 28°C (C), CHIKV-transmitting versus CHIKV-uninfected mosquitoes at 28°C (D), and CHIKV-transmitting versus CHIKV-uninfected mosquitoes at 20°C (E). Mosquitoes were examined 9 days after blood meal. RNA was extracted from six individual mosquitoes of each of the four experimental conditions, and sequencing was performed on a NovaSeq 6000 (Illumina) by Novogene to obtain 150 base paired-end reads. Reads were mapped onto the Ae. albopictus genome (GCF_006496715.1). Reads were characterized using corresponding primary annotation (release 102) with strand-specificity information. Functional annotation was done based on annotation of the reference genome AalbF2. LOC = GeneID in Ae albopictus genome (NCBI BioProject repository PRJNA484104,29). *indicates transcripts included in one of our defined classes of DEGs in Figure 3. ♦ indicates transcripts common to panels B and C
Functional classes and subclasses of DEGs up/downregulated in Ae. albopictus mosquitoes. (A) CHIKV-uninfected mosquitoes held at 20°C versus 28°C. (B) CHIKV-transmitting mosquitoes held at 20°C versus 28°C. (C) CHIKV-transmitting versus CHIKV-uninfected mosquitoes at 28°C. (D) CHIKV-transmitting versus CHIKV-uninfected mosquitoes at 20°C. Mosquitoes were examined 9 days after blood meal. Histograms indicate number of genes upregulated or downregulated in each subclass. For the lists of genes upregulated and downregulated, see Files S1 and S3–S5. For more details on classification, see File S2. * includes a transcript of the Top 10 DEGs
Responses to temperature of CHIKV-transmitting mosquitoes
Then, we analysed the effect of temperature on gene expression of mosquitoes exposed to a CHIKV-infectious blood meal, at 9 dpi. We compared the transcriptome of CHIKV-transmitting (i.e. mosquitoes with viral particles detected in saliva) held at 20°C versus 28°C (Figure 1A, Table S1, File S3). We identified a total of 906 DEGs; 552 were upregulated and 354 were down-regulated at 20°C compared with 28°C (Table S1). Transcripts with the highest upregulation and downregulation at 20°C were a non-coding RNA and an embryonic polarity protein (likely involved in Toll signalling pathway), respectively (Figure 1C). Interestingly, we found three genes strongly deregulated by temperature: an upregulated non-coding RNA (LOC109405538) and two downregulated collagenase proteins (Figure 1C; indicated by a diamond), which were also deregulated in CHIKV-uninfected mosquitoes (Figure 1B; indicated by a diamond). Genes involved in oxidative stress, cytoskeleton dynamics, signalling pathways, serine proteases and diverse immune pathways (RNAi, Toll…) were highly recruited in CHIKV-transmitting mosquitoes at 20°C (Figure 2B). We observed very distinct transcriptional profiles in CHIKV-transmitting mosquitoes at 20 and 28°C, with marked differences in immune and stress-related pathways.
Responses to CHIKV infection according to temperature
Then, we investigated the effect of temperature on gene expression of CHIKV-transmitting mosquitoes compared with CHIKV-uninfected mosquitoes. We compared gene expression profiles of CHIKV-transmitting mosquitoes versus CHIKV-uninfected mosquitoes at 20 and 28°C. We found that the total number of DEGs was more than three times higher during CHIKV infection at 20°C (602 DEGs) compared with CHIKV infection at 28°C (172 DEGs) (Table S2, Files S4 and S5). Genes were mostly upregulated (478 upregulated, 124 downregulated) by CHIKV infection at 20°C, whereas at 28°C, most genes were downregulated (115 downregulated, 57 upregulated) (Figure 1A, Table S2). The highest upregulated and downregulated genes corresponded to the WW domain-containing adapter protein (known to mediate protein–protein interactions without any well-defined function) and the cysteine-rich venom protein 6-like (a putative viral receptor in mosquitoes44) at 28°C (Figure 1D, File S4), whereas at 20°C, it was a GATOR complex protein MIOS-like involved in meiosis and a transmembrane 9 superfamily member protein involved in molecules transport, respectively (Figure 1E, File S5). When analysing mosquito responses to CHIKV infection by functional annotations of DEGs, we found that at 28°C, genes were poorly disrupted (Figure 2C, File S4), while at 20°C genes were highly upregulated by CHIKV infection: genes of oxidative stress (e.g. P450), serine proteases (e.g. CLIP, easter, grass, stubble, trypsin), and classical immune pathways including pathogen sensors (e.g. leucine-rich repeat proteins), pathogen elimination (e.g. melanization components, antimicrobial peptides) and canonical immune pathways (e.g. RNA interference and Toll pathway proteins) were the most disrupted gene classes (Figure 2D, File S5). We observed that CHIKV infection upregulated most genes at 20°C but not at 28°C.
Collectively, we demonstrated that temperature modulates mosquito gene expression and shapes responses to CHIKV infection.
Temperature alters bacterial communities structure in CHIKV-transmitting mosquitoes
To know if microbiota was involved in modulating transcriptomic responses, we analysed potential changes of mosquito bacterial microbiota induced by temperature using a metagenomics approach (Figure S1).
We sequenced the 16S rRNA gene from carcasses and midguts of pooled mosquitoes and obtained a total of 688 760 high-quality bacterial gene sequences for carcasses and 735 262 sequences for midguts by Illumina sequencing. These sequences were assigned to 2716 and 4977 OTUs at 95% similarities for carcasses and midguts, respectively.
Principal component analysis (PCoA) using Bray–Curtis dissimilarities revealed differences in bacterial communities structure in carcasses (axis 1 = 75.2% and axis 2 = 19.7% of the variation) and midguts (axis 1 = 46.4% and axis 2 = 42.7% of the variation) of CHIKV-uninfected (receiving a non-infectious blood meal), CHIKV non-transmitting (getting a CHIKV-infectious blood meal but unable to transmit at 9 dpi) and CHIKV-transmitting mosquitoes (getting a CHIKV-infectious blood meal and able to deliver virus in saliva at 9 dpi) (Figure S2A and B). In both carcasses and midguts, bacterial communities of CHIKV-transmitting mosquitoes at 28°C were distant from CHIKV-uninfected and CHIKV non-transmitting mosquitoes.
As expected, the symbiont Wolbachia was predominant in mosquito carcasses, ranging from 71 to 94% relative abundance (Figure 3A). OTUs of the family Acetobacteraceae including the genus Asaia were detected at 20 and 28°C, whereas OTUs from Enterobacteriaceae, including Serratia and Cedecea, were only found (at >1% of relative abundance) in bacterial flora of mosquitoes that received a CHIKV-infectious blood meal (CHIKV non-transmitting and CHIKV-transmitting mosquitoes) and held at 28°C (Figure 3A). Midguts exhibited more diverse bacterial communities; we detected a high diversity of Sphingomonadaceae, Rhizobiaceae and Micrococcaceae in CHIKV-uninfected mosquitoes held at 20°C, whereas at 28°C Asaia was predominant (Figure 3B). In mosquitoes that received a CHIKV-infectious blood meal, Wolbachia, Asaia and Acinetobacter were the majority at 20°C, whereas at 28°C Asaia predominated in CHIKV non-transmitting mosquitoes and Enterobacteriaceae, Serratia and Cedecea dominated in CHIKV-transmitting mosquitoes (Figure 3B). Overall, we observed combined effects of temperature and CHIKV infection on bacterial communities of mosquito midguts and carcasses; CHIKV infection tended to increase Enterobacteriaceae, Serratia and Cedecea, in CHIKV-transmitting mosquitoes at 28°C.
Relative abundance of bacterial taxa in Ae. albopictus carcasses and midgut. Each stacked bar illustrates the proportion of bacterial genera within a pool of six Ae. albopictus carcasses (A and C) or midguts (B). Sequencing was performed using Illumina (A and B) or Nanopore (C) technology. In frame A and B, OTUs with an overall relative abundance of at least 1% across all samples were represented in the plots. In frame C, two libraries were made and sequenced for each sample (two bars per experimental condition). Uninfected corresponds to mosquitoes that had a non-infectious blood meal. CHIKV non-transmitting mosquitoes refer to mosquitoes that had a CHIKV-infectious blood meal but were unable to transmit the virus with no viral particles detected in saliva at 9 days post-infection. CHIKV-transmitting mosquitoes represent mosquitoes that had a CHIKV-infectious blood meal and were able to transmit the virus, with viral particles detected in saliva at 9 days post-infection. The microbiota of mosquito carcasses and midguts were disrupted by infection and temperature. We detected Wolbachia and Asaia in microbiota of CHIKV-infected mosquitoes, but Serratia spp. were only found at 28°C. Complementary analysis of Nanopore data (see Material and Methods) allowed us to identify accurately Serratia marcescens
To better identify some bacteria of interest, we sequenced full length 16S rRNA genes of carcasses and obtained a total of 79 576 high-quality sequences which were grouped into 12 clusters based on the k-mer profiles they shared (Figure 3C). Full length 16S rRNA sequences of Acetobacteraceae were all identified as Asaia and those of Enterobacteriaceae were classified as Serratia, Enterobacter or Escherichia (Figure 3C). The consensus sequence representative of the Serratia cluster was assigned to Serratia marcescens by the naïve Bayesian classifier with bootstrap confidence estimate at 100%. Alignment of this consensus with four sequences retrieved as best Blast hits confirmed this identification (Figure S3).
According to long read analysis, microbial communities in carcasses of CHIKV-uninfected mosquitoes at 20 and 28°C were composed of two clusters of Wolbachia (representing together 88–97% of relative abundance), and one cluster of Asaia of lower abundance (2–12%) (Figure 3C). At 20°C, bacterial composition of CHIKV-infected mosquitoes remained substantially unchanged from that of uninfected mosquitoes (Figure 3C). Whereas at 28°C, in CHIKV non-transmitting mosquitoes, the relative abundances of Asaia and Wolbachia significantly decreased, whereas proportions of S. marcescens increased (36 and 40%; Figure 3C). Most notable changes were observed in CHIKV-transmitting mosquitoes with a relative abundance of S. marcescens reaching 83% (Figure 3C), which contributes to the shift in bacterial communities seen on PCoA plot (correlation coefficient between first axis of PCoA and relative abundance of Serratia: R = 0.8, Spearman correlation test) (Figure S2C). Overall, we observed a relative stability of the bacterial microbiota of mosquitoes infected by CHIKV at 20°C, whereas at 28°C, a significant increase of Serratia, namely S. marcescens, was detected in CHIKV-transmitting mosquitoes.
Temperature does not significantly modify CHIKV transmission by Ae. Albopictus
We evaluated mosquito TE of CHIKV infectious clone (E1–226V) at 20 and 28°C by detecting infectious viral particles in saliva of individual mosquitoes (Figure 4). Infectious saliva was detected from 3 dpi at 28°C (TE = 3.1%; 1/32), whereas at 20°C a little delay in TE was observed with first CHIKV-positive saliva detected at 6 dpi (TE = 2.9%; 1/35 mosquitoes). At 6 dpi, TE was significantly higher at 28°C (TE = 25%; 8/32 mosquitoes) than at 20°C (TE = 2.9%; 1/35 mosquitoes) (Fisher’s exact test: P < 0.05 after correction using Bonferroni’s method). Nevertheless, maximum TE was reached at 9 dpi at both temperatures, with TE peaking at 20°C (TE = 53.1%; 17/32 mosquitoes) without significant differences with TE at 28°C (TE = 32.2%; 10/31 mosquitoes) (Fisher’s exact test: P > 0.05; Figure 4).
CHIKV TE of Aedes albopictus mosquitoes maintained at 20 and 28°C. Using the forced salivation technique, saliva was collected from individual mosquitoes 3-, 6-, 9- and 14-days post-infection with E1–226V CHIKV infectious clone. Each point represents 31–35 mosquitoes analysed for a total of 259 mosquitoes. TE corresponds to the proportion of mosquitoes with infectious saliva among tested mosquitoes
Temperature influences CHIKV genetic evolution in mosquito cells
Temperature affects the selective environment and mutational diversity of arboviruses within mosquitoes. To characterize CHIKV evolutionary trajectories at 20 and 28°C, we performed serial passages of two infectious clone-derived CHIKV variants (E1–226V and E1–226A) on Ae. albopictus (U4.4) and Ae. aegypti (Aag2) cells (Figure S5). We counted a total of 1 604 125 350 and 2 283 422 120 nucleotides sequenced and 2570 and 4199 SNVs on U4.4 and Aag2 cells, respectively, suggesting that CHIKV infectious clone-derived viruses had mutation frequencies of 1.6 × 10−6 on U4.4 cells and 1.84 × 10−6 on Aag2 cells.
First, we examined distribution of SNVs along the viral genome and found a high density of SNVs (hotspot) in the envelope glycoprotein 2 (E2) (Figure 5). Main variants (frequency ≥ 10%) were mainly found at passages P20 and P30 (Figure 5). Interestingly, mutations in E2 mostly corresponded to non-synonymous substitutions leading to amino acid changes.
![Mapping of viral mutations acquired during passages on Ae. albopictus U4.4 cells (left) or Ae. aegypti Aag2 cells (right). E1–226V and E1–226A clones were serially passaged on cells maintained at 20 or 28°C: E1–226V at 20°C (A, E), E1–226V at 28°C (B, F), E1–226A at 20°C (C, G) and E1–226A at 28°C (D, H). SNV maps were generated using SNPviewer (https://gitlab.pasteur.fr/plechat/snpviewer). Viral genes were represented by different coloured bars schematizing coding regions of CHIKV genome. Two replicates of P10, P20, P30 and P40 were performed. (a) Blue dots refer to positions of minor single-nucleotide variations (SNVs) (frequency [3–10%]) and red dots to positions of main SNVs (≥10%). (b) Blue dots correspond to synonymous mutations and pink dots to non-synonymous mutations. SNV profiles were different according to cell type, virus variant and temperature](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/jtm/30/4/10.1093_jtm_taad062/1/m_taad062f5.jpeg?Expires=1747863111&Signature=0aqTIdKvfQNm8tqjAY1u3v2IhANVhL1hP5pqgjmns0aLR~zBfEU4bwgjDh0ukewfF95BSahnvYOqwxTKm-oga7-fdFnkA8goPs90Fo2WWRh-Hf9KmmalJG3raF6JQ8bP4P4U6VL99XChbiW-PiQhYtEoXURcJVR-VWnfYYRybhpWAMxxIpN2RJqZtJ9BTnuWHBjazPF5en0Z7I~guR5kQTXH9bsWADrWNMqWRL1fjrhIF22s-prlX4Ru1Skuh80QqHyzAxtQ6Y0d2bFMI3gIGhh3jgfKfU59ETqcFYL33nbMiEiNFabcsOOFaHqVZzYe~GcmuW154yhVQa0wit3Rzg__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Mapping of viral mutations acquired during passages on Ae. albopictus U4.4 cells (left) or Ae. aegypti Aag2 cells (right). E1–226V and E1–226A clones were serially passaged on cells maintained at 20 or 28°C: E1–226V at 20°C (A, E), E1–226V at 28°C (B, F), E1–226A at 20°C (C, G) and E1–226A at 28°C (D, H). SNV maps were generated using SNPviewer (https://gitlab.pasteur.fr/plechat/snpviewer). Viral genes were represented by different coloured bars schematizing coding regions of CHIKV genome. Two replicates of P10, P20, P30 and P40 were performed. (a) Blue dots refer to positions of minor single-nucleotide variations (SNVs) (frequency [3–10%]) and red dots to positions of main SNVs (≥10%). (b) Blue dots correspond to synonymous mutations and pink dots to non-synonymous mutations. SNV profiles were different according to cell type, virus variant and temperature
Then, we examined parallel evolution of replicates (main variants only; > 10%) in each combination of temperature (20 and 28°C), viral clones (E1–226V or E1–226A) and cell types (U4.4 or Aag2). We identified temperature-specific SNVs; if a SNV shared between replicates was present at 20°C and absent at 28°C, then this SNV was considered as a mutation associated with adaptation to 20°C and vice versa for SNVs specific to 28°C (Table 1). On Ae. albopictus U4.4 cells, we detected six temperature-specific SNVs, all specific to 28°C: two synonymous substitutions in nsP4 and four non-synonymous substitutions in E2 (Table 1). On Aag2 cells, we detected twice more SNVs specific to 20 or 28°C, which corresponded to three synonymous substitutions in regions encoding for nsP 1–3, 7 non-synonymous substitutions in E2 and 4 substitutions in 5′/3′ UTR (Table 1). Interestingly, we identified two SNVs that were strongly selected by temperature from P20 to P30; E2-T96I was selected at 28°C in both CHIKV variants (E1–226A and E1–226V) on U4.4 cells, but only in E1–226A viral variant on Aag2 cells, while E2-Y9H selected at 20°C in both CHIKV variants on Aag2 cells only.
List of temperature-specific SNVs detected on Ae. albopictus U4.4 cells and Ae. aegypti Aag2 cells, at passages P10, P20, P30 and, after temperature change, at P40.
| Temperature-specific SNVs selected on Ae. albopictus U4.4 cells | |||||||
|---|---|---|---|---|---|---|---|
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 5645 | nsP4 | T | 28 | 88.5 | E1–226V | 20 | |
| 5645 | nsP4 | T | 28 | 99.6 | E1–226V | 30 | |
| 6227 | nsP4 | T | 28 | 87.9 | E1–226V | 30 | |
| 8704 | E2 | A:380 | G55R | 28 | 56 | E1–226V | 30 |
| 8704 | E2 | A:380 | G55R | 28 | 13 | E1–226V | 40 (20 C– > 28 C) |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226V/Ac | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226A | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 13.7 | E1–226V | 30 |
| 8731 | E2 | C:389 | W64R | 28 | 21.8 | E1–226V | 40 (20 C– > 28 C) |
| 8828 | E2 | T:421–A:421 | T96I–T96K | 28C | 33.5—16.8 | E1–226V/Ac | 20 |
| 8828a | E2 | T:421—A:421 | T96I—T96K | 28 | 20.8—16.8 | E1–226A | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 33.5 | E1–226V | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 62.8 | E1–226V/Ac | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 62.8 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 40.5 | E1–226V | 30 |
| 9011 | E2 | C:482 | V157A | 28 | 19.5 | E1–226V | 30 |
| Temperature-specific SNVs selected on Ae. aegypti Aag2 cells | |||||||
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 19 | 5’UTR | A | 28 | 33 | E1–226A | 20 | |
| 20 | 5’UTR | A | 28 | 17.8 | E1–226A | 20 | |
| 25 | 5’UTR | T | 28 | 17.8 | E1–226A | 20 | |
| 202 | nsP1 | A | 20 | 13.7 | E1–226V | 20 | |
| 4020 | nsP2 | C | 28 | 12.6 | E1–226A | 20 | |
| 4851 | nsP3 | T | 28 | 11 | E1–226A | 20 | |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226A | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226V/Ac | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226V/Ac | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226A | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 46 | E1–226V | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 91.7 | E1–226A | 20 |
| 8828a | E2 | T:421 | T96I | 28 | 95.6 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 8.6 | E1–226A | 40 (20 C– > 28 C) |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226A | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 20.6 | E1–226V | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226V/Ac | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 14.2 | E1–226V | 30 |
| 8964 | E2 | G:466 | F141L | 28 | 84.9 | E1–226A | 30 |
| 9041 | E2 | C:492 | I167T | 20 | 23.4 | E1–226V | 20 |
| 9223 | E2 | A:553 | A228T | 20 | 79.9 | E1–226V | 20 |
| 9230 | E2 | T:555 | T230I | 20 | 12 | E1–226A | 30 |
| 11 395 | 3’NTR | A | 20 | 18 | E1–226A | 20 | |
| Temperature-specific SNVs selected on Ae. albopictus U4.4 cells | |||||||
|---|---|---|---|---|---|---|---|
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 5645 | nsP4 | T | 28 | 88.5 | E1–226V | 20 | |
| 5645 | nsP4 | T | 28 | 99.6 | E1–226V | 30 | |
| 6227 | nsP4 | T | 28 | 87.9 | E1–226V | 30 | |
| 8704 | E2 | A:380 | G55R | 28 | 56 | E1–226V | 30 |
| 8704 | E2 | A:380 | G55R | 28 | 13 | E1–226V | 40 (20 C– > 28 C) |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226V/Ac | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226A | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 13.7 | E1–226V | 30 |
| 8731 | E2 | C:389 | W64R | 28 | 21.8 | E1–226V | 40 (20 C– > 28 C) |
| 8828 | E2 | T:421–A:421 | T96I–T96K | 28C | 33.5—16.8 | E1–226V/Ac | 20 |
| 8828a | E2 | T:421—A:421 | T96I—T96K | 28 | 20.8—16.8 | E1–226A | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 33.5 | E1–226V | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 62.8 | E1–226V/Ac | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 62.8 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 40.5 | E1–226V | 30 |
| 9011 | E2 | C:482 | V157A | 28 | 19.5 | E1–226V | 30 |
| Temperature-specific SNVs selected on Ae. aegypti Aag2 cells | |||||||
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 19 | 5’UTR | A | 28 | 33 | E1–226A | 20 | |
| 20 | 5’UTR | A | 28 | 17.8 | E1–226A | 20 | |
| 25 | 5’UTR | T | 28 | 17.8 | E1–226A | 20 | |
| 202 | nsP1 | A | 20 | 13.7 | E1–226V | 20 | |
| 4020 | nsP2 | C | 28 | 12.6 | E1–226A | 20 | |
| 4851 | nsP3 | T | 28 | 11 | E1–226A | 20 | |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226A | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226V/Ac | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226V/Ac | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226A | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 46 | E1–226V | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 91.7 | E1–226A | 20 |
| 8828a | E2 | T:421 | T96I | 28 | 95.6 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 8.6 | E1–226A | 40 (20 C– > 28 C) |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226A | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 20.6 | E1–226V | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226V/Ac | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 14.2 | E1–226V | 30 |
| 8964 | E2 | G:466 | F141L | 28 | 84.9 | E1–226A | 30 |
| 9041 | E2 | C:492 | I167T | 20 | 23.4 | E1–226V | 20 |
| 9223 | E2 | A:553 | A228T | 20 | 79.9 | E1–226V | 20 |
| 9230 | E2 | T:555 | T230I | 20 | 12 | E1–226A | 30 |
| 11 395 | 3’NTR | A | 20 | 18 | E1–226A | 20 | |
aSNVs on both cell lines
bFor non-synonymous mutation, position in the structural polyprotein was indicated
cFor both E1–226V and E1–226A in italics, P40 obtained from P30 passaged 10 times at an alternative temperature.
Only single-nucleotide variations (SNVs) shared between replicates in condition A were considered and selected for analysis if not found in condition B. Conditions A and B referred to temperature (20 or 28°C). Genome position, nature of changes (synonymous mutation or non-synonymous mutation with AA change), maximum frequency observed in the two replicates were shown for each mutation. nt, nucleotide; AA, amino acid
List of temperature-specific SNVs detected on Ae. albopictus U4.4 cells and Ae. aegypti Aag2 cells, at passages P10, P20, P30 and, after temperature change, at P40.
| Temperature-specific SNVs selected on Ae. albopictus U4.4 cells | |||||||
|---|---|---|---|---|---|---|---|
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 5645 | nsP4 | T | 28 | 88.5 | E1–226V | 20 | |
| 5645 | nsP4 | T | 28 | 99.6 | E1–226V | 30 | |
| 6227 | nsP4 | T | 28 | 87.9 | E1–226V | 30 | |
| 8704 | E2 | A:380 | G55R | 28 | 56 | E1–226V | 30 |
| 8704 | E2 | A:380 | G55R | 28 | 13 | E1–226V | 40 (20 C– > 28 C) |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226V/Ac | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226A | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 13.7 | E1–226V | 30 |
| 8731 | E2 | C:389 | W64R | 28 | 21.8 | E1–226V | 40 (20 C– > 28 C) |
| 8828 | E2 | T:421–A:421 | T96I–T96K | 28C | 33.5—16.8 | E1–226V/Ac | 20 |
| 8828a | E2 | T:421—A:421 | T96I—T96K | 28 | 20.8—16.8 | E1–226A | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 33.5 | E1–226V | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 62.8 | E1–226V/Ac | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 62.8 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 40.5 | E1–226V | 30 |
| 9011 | E2 | C:482 | V157A | 28 | 19.5 | E1–226V | 30 |
| Temperature-specific SNVs selected on Ae. aegypti Aag2 cells | |||||||
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 19 | 5’UTR | A | 28 | 33 | E1–226A | 20 | |
| 20 | 5’UTR | A | 28 | 17.8 | E1–226A | 20 | |
| 25 | 5’UTR | T | 28 | 17.8 | E1–226A | 20 | |
| 202 | nsP1 | A | 20 | 13.7 | E1–226V | 20 | |
| 4020 | nsP2 | C | 28 | 12.6 | E1–226A | 20 | |
| 4851 | nsP3 | T | 28 | 11 | E1–226A | 20 | |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226A | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226V/Ac | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226V/Ac | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226A | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 46 | E1–226V | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 91.7 | E1–226A | 20 |
| 8828a | E2 | T:421 | T96I | 28 | 95.6 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 8.6 | E1–226A | 40 (20 C– > 28 C) |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226A | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 20.6 | E1–226V | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226V/Ac | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 14.2 | E1–226V | 30 |
| 8964 | E2 | G:466 | F141L | 28 | 84.9 | E1–226A | 30 |
| 9041 | E2 | C:492 | I167T | 20 | 23.4 | E1–226V | 20 |
| 9223 | E2 | A:553 | A228T | 20 | 79.9 | E1–226V | 20 |
| 9230 | E2 | T:555 | T230I | 20 | 12 | E1–226A | 30 |
| 11 395 | 3’NTR | A | 20 | 18 | E1–226A | 20 | |
| Temperature-specific SNVs selected on Ae. albopictus U4.4 cells | |||||||
|---|---|---|---|---|---|---|---|
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 5645 | nsP4 | T | 28 | 88.5 | E1–226V | 20 | |
| 5645 | nsP4 | T | 28 | 99.6 | E1–226V | 30 | |
| 6227 | nsP4 | T | 28 | 87.9 | E1–226V | 30 | |
| 8704 | E2 | A:380 | G55R | 28 | 56 | E1–226V | 30 |
| 8704 | E2 | A:380 | G55R | 28 | 13 | E1–226V | 40 (20 C– > 28 C) |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226V/Ac | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 25.3 | E1–226A | 20 |
| 8731 | E2 | C:389 | W64R | 28 | 13.7 | E1–226V | 30 |
| 8731 | E2 | C:389 | W64R | 28 | 21.8 | E1–226V | 40 (20 C– > 28 C) |
| 8828 | E2 | T:421–A:421 | T96I–T96K | 28C | 33.5—16.8 | E1–226V/Ac | 20 |
| 8828a | E2 | T:421—A:421 | T96I—T96K | 28 | 20.8—16.8 | E1–226A | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 33.5 | E1–226V | 20 |
| 8828 | E2 | T:421 | T96I | 28 | 62.8 | E1–226V/Ac | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 62.8 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 40.5 | E1–226V | 30 |
| 9011 | E2 | C:482 | V157A | 28 | 19.5 | E1–226V | 30 |
| Temperature-specific SNVs selected on Ae. aegypti Aag2 cells | |||||||
| Genome position | Region | nt changeb | AA change | Specific of temperature (°C) | Maximum Frequency (%) | CHIKV variant | Passage |
| 19 | 5’UTR | A | 28 | 33 | E1–226A | 20 | |
| 20 | 5’UTR | A | 28 | 17.8 | E1–226A | 20 | |
| 25 | 5’UTR | T | 28 | 17.8 | E1–226A | 20 | |
| 202 | nsP1 | A | 20 | 13.7 | E1–226V | 20 | |
| 4020 | nsP2 | C | 28 | 12.6 | E1–226A | 20 | |
| 4851 | nsP3 | T | 28 | 11 | E1–226A | 20 | |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226A | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 96.5 | E1–226V/Ac | 20 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226V/Ac | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 92.8 | E1–226A | 30 |
| 8566 | E2 | C:334 | Y9H | 20 | 46 | E1–226V | 30 |
| 8828a | E2 | T:421 | T96I | 28 | 91.7 | E1–226A | 20 |
| 8828a | E2 | T:421 | T96I | 28 | 95.6 | E1–226A | 30 |
| 8828 | E2 | T:421 | T96I | 28 | 8.6 | E1–226A | 40 (20 C– > 28 C) |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226A | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 20.6 | E1–226V | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 28.8 | E1–226V/Ac | 20 |
| 8831 | E2 | C:422 | M97T | 20 | 14.2 | E1–226V | 30 |
| 8964 | E2 | G:466 | F141L | 28 | 84.9 | E1–226A | 30 |
| 9041 | E2 | C:492 | I167T | 20 | 23.4 | E1–226V | 20 |
| 9223 | E2 | A:553 | A228T | 20 | 79.9 | E1–226V | 20 |
| 9230 | E2 | T:555 | T230I | 20 | 12 | E1–226A | 30 |
| 11 395 | 3’NTR | A | 20 | 18 | E1–226A | 20 | |
aSNVs on both cell lines
bFor non-synonymous mutation, position in the structural polyprotein was indicated
cFor both E1–226V and E1–226A in italics, P40 obtained from P30 passaged 10 times at an alternative temperature.
Only single-nucleotide variations (SNVs) shared between replicates in condition A were considered and selected for analysis if not found in condition B. Conditions A and B referred to temperature (20 or 28°C). Genome position, nature of changes (synonymous mutation or non-synonymous mutation with AA change), maximum frequency observed in the two replicates were shown for each mutation. nt, nucleotide; AA, amino acid
To test whether mutations selected at one temperature could revert when incubating viruses at the other temperature, we passaged P30 10 times on cells (P40) at 20 or 28°C (Figure S5). When examining SNVs, P40 variants accumulated many new substitutions at low frequency, especially in non-coding regions. All mutations selected at 20 or 28°C on U4.4 or Aag2 cells were lost upon temperature inversion, demonstrating their sensitivity to temperature (Table 1). When examining viral titres, we found that only viruses maintained at 20°C on U4.4 cells were impacted by temperature inversion; titres drastically dropped from P30 (20°C) to P31 (28°C) (Figure S5).
Besides temperature-specific mutations, we found several SNVs specific to a viral variant or to a cell type (Table S3). Interestingly, when the variant harbouring the Ae. albopictus-adaptive mutation E1–226V was passaged on Aag2 cells, it systematically switched to E1–226A from P10 at 28°C and from P20 at 20°C to reach a maximum frequency of 100% at P30 at both temperatures (Table S3). We tested whether temperature-selected mutations affected CHIKV transmission by mosquitoes. We exposed Ae. albopictus to CHIKV-infectious blood meals containing P30 passaged on U4.4 cells and maintained at 20 or 28°C. By examining transmission efficiencies, we found that Ae. albopictus did not better transmit CHIKV passaged 30 times on U4.4 cells at 20 or 28°C (Figure S6).
Collectively, we found that evolutionary dynamics of infectious clones derived CHIKV variants were different according to cell type (U4.4 or Aag2), CHIKV variant (E1–226V or E1–226A) and temperature (20 and 28°C).
Discussion
The dynamics and distribution of vector-borne diseases (VBDs) depend on the close interplay between the mosquito (genetic, immunophysiology, microbiota…), the pathogen and the environment.11,45–47 Nevertheless, the molecular basis of the relationship between VBD and temperature still remains highly theoretical and poorly elucidated.8,9,48,49 Together, our results indicate that Ae. albopictus gene expression, bacterial flora and CHIKV evolutionary dynamics are disrupted by temperature, and the combination of temperature-induced changes is associated with successful viral transmission. At 20°C, CHIKV infection was associated with a strong upregulation of genes involved in stress and immunity but had almost no impact on bacterial flora. Whereas at 28°C, the composition of mosquitoes bacterial microbiota was completely altered and gene expression was weakly impacted by CHIKV infection. These changes most likely create an environment that influences viral genetic diversity within the mosquito and promote temperature adaptation to ensure viral transmission.
It is largely admitted that global warming profoundly affects VBD as it necessarily calls for an arthropod vector which, in essence, is ectothermic.49–51 CHIKV exemplifies that VBD endemic to tropical regions can affect European countries52,53 once a competent vector such as Ae. albopictus has become well established.54 This situation raises many questions about the incidence of arboviral diseases in temperate regions, particularly in the context of climate change55(e.g. extension of favourable transmission periods which could increase the risk of diseases spread). Temperature plays a significant role in viral transmission according to a tripartite interaction between mosquito, virus and temperature.16,56,7
Upon infection, mosquitoes mount an immune response towards pathogens which implies major transcriptional changes, probably determinants of infection outcome and persistence establishment.57,58 Gene expression of Ae. aegypti in response to Zika virus11 and CHIKV46 infection varied significantly according to temperature. Here, we showed that Ae. albopictus transcriptional responses to CHIKV infection were quantitatively (more than 3-fold more DEGs at 20°C) and qualitatively very different at 20 and 28°C. Genes involved in oxidative stress (e.g. cytochrome P450, peroxidase, glutathione-S-transferases) and immune pathways (e.g. serine proteases, antimicrobial peptides, elements of small interfering RNA, Toll and melanization processes) were highly recruited at 20°C but not at 28°C, suggesting that temperature modulates gene expression in response to viral infection.
In addition, temperature is known to play a substantial role in shaping microbial populations in insects.12,59,60 At 20°C, CHIKV infection only induces minor changes in bacteria relative abundances, whereas at 28°C, substantial changes in bacterial composition were observed; we detected a significant decrease of Wolbachia abundance concomitantly to an increase of Serratia, both contributing to increase CHIKV infection.61 Using full length 16S rRNA gene sequencing technique, we identified the commensal bacterium S. marcescens; this bacterial species may facilitate virus entry in mosquito midgut by secreting a protein SmEnhancin that digests membrane-bound mucins on the gut epithelia.62 Serratia-positive Ae. aegypti and Ae. albopictus mosquitoes were predominant in dengue-endemic regions.62 Despite a strong antiviral response at 20°C and Serratia abundance in CHIKV-infected mosquitoes at 28°C, no significant difference in CHIKV transmission was observed between 20 and 28°C.
In 2005–06, the CHIKV outbreak on La Réunion Island was marked by the emergence of an epidemic variant carrying an amino acid substitution (alanine by valine) at the position 226 of the E1 envelope glycoprotein17; this mutation resulted in an enhanced transmission of the virus by Ae. albopictus.18,19 To what extent environmental temperature could influence the selection of such viral variants within the mosquito is a major knowledge gap. Using an in vitro selection system, we analysed CHIKV replication and diversity and detected SNVs hotspots in the E2 viral gene of serially passaged infectious CHIKV clones maintained at 20 and 28°C. The CHIKV envelope glycoprotein E2 is involved in cell attachment63; this highly variable region governs the ability of CHIKV to bind to receptors or attachment factors leaving multiple possibilities for viral entry.64 Our viral evolution experiment allowed us to identify two SNVs strongly specific to temperature, E2-T96I and E2-Y9H. E2-T96I was selected at 28°C in both CHIKV variants (E1–226A and E1–226V) maintained on U4.4 cells but only in E1–226A viral variant on Aag2 cells, suggesting some evolutionary limitations related to the virus on Aag2 cells. E2-Y9H was selected at 20°C in both CHIKV variants on Aag2 cells only, indicating that the emergence of this mutation was temperature- and cell type-dependent. This illustrates the potential of temperature to influence the evolutionary trajectories of viruses. Mutations selected after serial passages can be lost when reverting the temperature demonstrating that temperature governs in some extent viral genetic diversity. Moreover, the position E1–226 undergoes a switch from valine to alanine when passaged on Ae. aegypti Aag2 cells, at 20°C and 28°C, consistent with the fact that the valine at this position is an adaptation to Ae. albopictus. Nevertheless, despite significant genetic changes, CHIKV passaged 30 times on U4.4 cells at 20 or 28°C did not show any transmission advantage in Ae. albopictus mosquitoes.
To our knowledge, this is the first study to address the influence of temperature on virus–mosquito interactions by combining elements of both the virus and the mosquito (including the microbiota). It highlights the complexity of the vector system and opens the field to further explorations on the subject. Many perspectives and areas for improvement are possible; first larger sample sizes would allow to get more robust data, second, sampling at several times post-infection would give a clearer idea of the influence of temperature at different stages of infection and finally, the use of different virus strains and mosquito populations would better capture the diversity of vector systems in nature. In addition, the inclusion of temperature regimes with diurnal variations would provide information closer to the field conditions.65 It seems obvious that a suitable climate is necessary for a VBD to emerge and to persist. The large distribution of competent vectors is by itself a risk map predicting future arboviral emergences. In our model, temperature induced similar transmission profiles that corresponded to very distinct molecular processes. We are therefore dealing with a complex biological system whose plasticity and evolutionary potential allow it to respond to an environmental constraint like temperature. We conclude that temperature shapes the functioning of the vectorial system by making evolutionary adjustments at the levels of the vector, its microbiota and the arbovirus acquired during a blood meal, to reach a balance between the different protagonists leading to a successful arbovirus transmission.
Funding
This work was supported by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 731060 (Infravec2, Research Infrastructures for the control of vector-borne diseases), the Laboratoire d’Excellence ‘Integrative Biology of Emerging Infectious Diseases’ (grant n°ANR-10-LABX-62-IBEID) and the European Union’s Horizon 2020 research and innovation programme under ZIKAlliance grant agreement No 734548.
Acknowledgments
The authors thank P. Bousses for his help in mosquito collections, R. Loreto and S. Merkling for discussions on mosquito transcriptome and K. Vernick for his constant support as coordinator of the EU Infravec2 project. Sequencing for transcriptomic experiments were carried out by E. Turc, L. Motreff and L. Lemée (Biomics Platform, C2RT, Institut Pasteur) supported by France Génomique (ANR-10-INBS-09-09) and IBISA. JR was supported by the ERC RosaLind Starting Grant n°948135.
Authors’contribution
Rachel Bellone (Conceptualization [equal], Investigation [equal], Writing—original draft [equal], Writing—review & editing [equal]), Pierre Lechat (Formal analysis [equal]), Laurence Mousson (Investigation [equal]), Valentine Gilbart (Formal analysis [equal]), Géraldine Piorkowski (Formal analysis [equal]), Chloé Bohers (Investigation [equal]), Andres Merits (Methodology [equal]), Etienne Kornobis (Formal analysis [equal]), Julie Reveillaud (Formal analysis [equal]), Christophe Paupy (Methodology [equal]), Marie Vazeille (Investigation [equal]), Jean-Philippe Martinet (Investigation [equal]), Yoann Madec (Formal analysis [equal]), Xavier de Lamballerie (Formal analysis [equal]), Catherine Dauga (Formal analysis [equal], Writing—original draft-Supporting, Writing—review & editing [equal]), Anna-Bella Failloux (Conceptualization [equal], Funding acquisition [equal], Supervision [lead], Writing—original draft-Supporting, Writing—review & editing [lead]).
Conflict of interest
The authors declare that they have no conflict of interest.
Data Availability Statement
All data and code generated in this article are available in the main text or the supplementary materials.
References
Author notes
Catherine Dauga and Anna-Bella Failloux share last authorship
