Wolbachia Do Not Induce Reactive Oxygen Species-Dependent Immune Pathway Activation in Aedes albopictus

Aedes albopictus is a major vector of dengue (DENV) and chikungunya (CHIKV) viruses, causing millions of infections annually. It naturally carries, at high frequency, the intracellular inherited bacterial endosymbiont Wolbachia strains wAlbA and wAlbB; transinfection with the higher-density Wolbachia strain wMel from Drosophila melanogaster led to transmission blocking of both arboviruses. The hypothesis that reactive oxygen species (ROS)-induced immune activation plays a role in arbovirus inhibition in this species was examined. In contrast to previous observations in Ae. aegypti, elevation of ROS levels was not observed in either cell lines or mosquito lines carrying the wild-type Wolbachia or higher-density Drosophila Wolbachia strains. There was also no upregulation of genes controlling innate immune pathways or with antioxidant/ROS-producing functions. These data suggest that ROS-mediated immune activation is not an important component of the viral transmission-blocking phenotype in this species.


Introduction
Wolbachia are maternally transmitted, intracellular bacterial endosymbionts that can spread to high frequency in insect populations by manipulating their reproduction, in particular using patterns of sterility known as cytoplasmic incompatibility to give a reproductive advantage to was observed in wildtypes or a wMel transinfection, despite complete blocking of DENV transmission in the latter [5]. Viral inhibition did positively correlate with Wolbachia density [5], as also shown for in cell lines [26,27]. However, increased ROS levels were previously reported by flow cytometry and microscopy in naturally infected versus antibiotic-cured Ae. albopictus cell lines; upregulation of antioxidant genes was also observed [28]. The aim of this study was to reconcile these differing observations in Ae. albopictus, and relate them to work in Ae. aegypti, in order to better understand whether ROS-mediated immune activation may be an important component of arboviral inhibition in this system. Direct biochemical measurement of ROS was undertaken, and ROS-modulating and immune pathway regulator gene transcription examined, in both cell lines and mosquito lines carrying wild-type and transinfected Wolbachia strains.

Gene Transcription qRT-PCR Assays
RNA was extracted from pools of three 7-day old adult females or from cultures of Aa23 in 12-well plates, seeded from 250 k cells and incubated for 72 h. Samples were homogenised in 100 µL Trizol (Invitrogen-Life technologies, Paisley, UK) or Tri Reagent (Sigma-Aldrich, Gillingham, UK) per mosquito or 10 6 Aa23 cells using an electric motor driven pestle (Kimble Chase, Vineland, NJ, USA), or by grinding with borosilicate beads at 40 Hz for 4 min using a Qiagen TissueLyser LT (Qiagen, Manchester, UK). RNA was extracted per manufacturer's instructions and treated with DNase I (NEB) then purified by phenol-chloroform extraction. Samples were quantified and quality checked using Nanodrop analysis and those with high purity (A260:230 >1.90 and A260:280 ratios of >2.00) were used to synthesise cDNA from 1 µg RNA using the High Capacity RNA-to-cDNA Kit (Applied Biosystems-Life Technologies, Paisley, UK) as per manufacturer's instructions.
The Ae. albopictus transcriptome database albopictusexpression.org [32] was interrogated using BLAST to identify transcripts of immune pathway and ROS-related genes, and qRT-PCR primers were designed as per Table 1. All primers were tested using a template standard curve and achieved efficiencies of 1.8-2.1 (80%-110%). Standard curves were run on all plates to assess efficiency per run. Quantitative PCR was performed in a MJ Research PTC-200 DNA Engine using a Chromo4 detection system and Opticon Monitor Software v3.1 (all manufactured by Bio-Rad Laboratories, Hemel Hempstead, UK). SYBR Green EXPRESS qPCR Supermix Universal (Life Technologies, Paisley, UK) and the iQ SYBR Green supermix (Bio-Rad Laboratories) were used in 10 µL reaction as per manufacturers' instructions. Cycling conditions were 4 min at 95˝C followed by 40 cycles of 95˝C for 15 s and 59˝C for 30 s. Some independent repeats used SYBR Green JumpStart Taq ReadyMix (Sigma-Aldrich) on a Stratagene Mx3005P Real-Time PCR System (Agilent Technologies, Stockport, UK) using the same mix and thermocycler settings. Table 1. Oligonucleotide primers for qRT-PCR of immune pathway and ROS-related genes in Ae. albopictus.

Wolbachia Density qPCR Assay
A universal Culicidae Wolbachia density assay was designed, allowing absolute qPCR quantification of host and Wolbachia genome copy number based on a plasmid standard curve. A single copy nucleotide sequence conserved across mosquito species was identified: a 93 bp conserved regulatory element upstream of the homothorax (HTH) gene [33]. Primers were designed using PrimerBLAST to amplify this region across all sequenced Culicidae species and Wolbachia surface protein (Wsp) sequences in the insect-infecting Wolbachia strains for which genomes were available: Mos qHTH F2 TGGTCCTATATTGGCGAGCTA; Mos qHTH RTCGTTTTTGCAAGAAGGTCA; qWSPallF3-ATCTTTTATAGCTGGTGGTGGT [2], qWSPallR1-AAAGTCCCTCAACATCAACCC.
DNA was extracted from pools of three 7-day old adult females or from confluent cultures of Aa23 cell lines in 12-well plates. Samples were homogenised in 100 µL cetyltrimethyl ammonium bromide (CTAB) buffer per mosquito (0.1 M Tris HCl pH 8.0, 1 M NaCl, 10 mM EDTA, 1% CTAB). Following incubation at 95˝C for 30 min, the sample was left to cool and an equal volume of chloroform was added prior to centrifugation at 15,000ˆg for 25 min at 4˝C. The aqueous layer was transferred to a new tube and DNA was precipitated with an equal volume of isopropanol then washed twice with 70% ethanol prior to resuspension in TE buffer and quantification using a Nanodrop spectrophotometer. DNA was diluted to 50 ng/µL and 100 ng was added to a 10 µL qPCR reaction using SYBR Green EXPRESS qPCR Supermix Universal (Life Technologies) as per manufacturers' instructions.
Wolbachia genomes per host genome were determined from a linear model fitted to the known plasmid standard curve values. This data was tested for normality and homogeneity of variances using the Shapiro and Bartlett tests respectively. Normal data was then assessed using a two-way analysis of variance (ANOVA) followed by Tukey Honestly Significant Differences (Tukey HSD) post-hoc pairwise testing. Data which deviated from the normal distribution was analysed via a Kruskal-Wallis test using a Bonferroni adjustment for multiple comparisons.

Hydrogen Peroxide Assays
Hydrogen Peroxide H 2 O 2 is the most stable form of ROS and thus acts as a correlate of overall ROS levels, it also has a highly specific biochemical assay which overcomes some of the issues in interpreting data inherent in other ROS probes, some of which have been shown to be highly non-specific [34]. H 2 O 2 generation by live cells was assessed using an Amplex Red Hydrogen Peroxide Kit (Invitrogen). Cells in the exponential growth phase (four days post passage in 75 cm 2 flasks) were harvested in fresh Schneiders solution (+10% FBS, +1% penicillin-streptomycin) and viability was assessed by Trypan blue staining to ensure all lines had similar viability. 150,000 cells were plated per well of a flat bottomed 96-well microplate and allowed to settle for one hour in the dark at 25˝C. Media was then removed, cells were washed once with Phosphate-buffered saline (PBS) and 50 µL Amplex Buffer in PBS was added followed by 50 µL Amplex reaction solution (as per manufacturers' guidelines) immediately prior to insertion into a FLUOstar microplate reader (BMG LabTech, Aylesbury, UK).
Pools of three mosquitoes were collected on ice in 2 mL tubes containing 1000 µL PBS with 2 mg/mL 3-Amino-1,2,4-triazole (3-AT) and three borosilicate beads. Samples were homogenised at 50 Hz for 5 min in a TissueLyser LT (Qiagen) and centrifuged at 20,000ˆg for 15 min at 4˝C. The supernatant was transferred to a new tube and centrifuged again at 20,000ˆg for 10 min at 4˝C before 10 µL supernatant was added to a 96 well flat bottomed plate for the protein assay and 250 µL was added to a 10,000 k molecular weight cut-off (MWCO) Spinfilter Ultrafiltration centrifugal filter (VWR) and centrifuged at 14,000ˆg for 20 min. 50 µL of filtrate was added to a separate flat bottomed 96-well plate for the H 2 O 2 assay. Both plates were stored on ice until assays began.
200 µL Bradford Reagent (Sigma-Aldrich) was added to samples in the protein assay plate, which was then shaken at 100 rpm for 30 s and incubated at 25˝C in the dark for 10 min before being read at 595 nm using a FLUOstar microplate reader (BMG LabTech).
The Amplex Red H 2 O 2 Assay (Invitrogen) was used according to manufacturers' instructions with 50 µL Amplex reagent (containing 2ˆAmplex buffer) added to the 50 µL sample prepared above. The plate was then incubated in the FLUOstar microplate reader at 25˝C where it was read kinetically at 545/590 nm excitation/emission over 40 cycles of 60 s with shaking at 100 rpm every two cycles to reduce clumping. The 30 min reading was used after confirmation that the assay kinetics were linear at this time point. Readings from protein and H 2 O 2 samples were compared to standard curves of known concentration of albumin (Sigma Aldrich) and H 2 O 2 (Invitrogen) in the respective assays. H 2 O 2 was then normalised to protein content for adult assays and cell count for cell line assays (verified by post assay counts from randomly selected wells within each line). This data was tested for normality and homogeneity of variances using the Shapiro and Bartlett tests respectively then assessed using a two-way ANOVA followed by Tukey HSD pairwise testing or by Kruskal-Wallis tests.

Innate Immune Pathway Gene Transcription
The transcription of immune pathway regulator genes CAC, Rel1, CSP, Rel2, PIAS and STAT was assessed in the Wolbachia-uninfected Ae. albopictus cell line Aa23-T compared to the same cell line infected with wAlb, wMel and wMelPop using qRT-PCR. These transinfected lines differed significantly in density, respectively harbouring a mean of 2.5, 17.7 and 38.4 Wolbachia genomes per host genome ( Figure 1A; one-way ANOVA, F = 71.00, p < 0.001. Tukey HSD, p < 0.01 for all comparisons). Only one significant difference was found, a 5.66-fold increase in STAT expression in Aa23-T.wMel (pMCMC = 0.014) ( Figure 1B). This was not supported in the most densely infected line Aa23-T.wMelPop, which showed a non-significant STAT increase of 2.66-fold. There was no increase in expression of the Toll pathway transcription factor Rel1 or decrease in expression of its negative regulator CAC, both of which might be expected during Toll activation. this time point. Readings from protein and H2O2 samples were compared to standard curves of known concentration of albumin (Sigma Aldrich) and H2O2 (Invitrogen) in the respective assays. H2O2 was then normalised to protein content for adult assays and cell count for cell line assays (verified by post assay counts from randomly selected wells within each line). This data was tested for normality and homogeneity of variances using the Shapiro and Bartlett tests respectively then assessed using a two-way ANOVA followed by Tukey HSD pairwise testing or by Kruskal-Wallis tests.

Innate Immune Pathway Gene Transcription
The transcription of immune pathway regulator genes CAC, Rel1, CSP, Rel2, PIAS and STAT was assessed in the Wolbachia-uninfected Ae. albopictus cell line Aa23-T compared to the same cell line infected with wAlb, wMel and wMelPop using qRT-PCR. These transinfected lines differed significantly in density, respectively harbouring a mean of 2.5, 17.7 and 38.4 Wolbachia genomes per host genome ( Figure 1A; one-way ANOVA, F = 71.00, p < 0.001. Tukey HSD, p < 0.01 for all comparisons). Only one significant difference was found, a 5.66-fold increase in STAT expression in Aa23-T.wMel (pMCMC = 0.014) ( Figure 1B). This was not supported in the most densely infected line Aa23-T.wMelPop, which showed a non-significant STAT increase of 2.66-fold. There was no increase in expression of the Toll pathway transcription factor Rel1 or decrease in expression of its negative regulator CAC, both of which might be expected during Toll activation.  Immune gene transcription in the Ae. albopictus mosquito lines Uju.wAlb and Uju.wMel was compared to uninfected Uju.T, normalised to the housekeeping gene S17. Uju.wAlb had a mean Wolbachia infection density of 1.93 Wolbachia genomes per host genome, while Uju.wMel had a substantially higher infection density of 9.62 ( Figure 2). No consistent significant difference in CAC, Rel1 or Rel2 or TAK1 was observed.

H 2 O 2 Levels in Ae. albopictus Adults and Cell Lines
All Uju mosquito and Aa23-T cell lines were assayed for H 2 O 2 to determine if Wolbachia infection increased ROS production and if any effect correlated with Wolbachia density (Figure 3). Relative H 2 O 2 generation in Aa23-T cells across three independent experiments was significantly different between lines as analysed by Kruskal-Wallis testing for non-normal data (A: χ 2 = 43.80, df = 3, p-value = 1.67e´09. B: χ 2 = 41.89, df = 3, p-value = 4.23e´09. C: χ 2 = 26.04, df = 3, p-value = 9.36e´06). Post-hoc bootstrap ratio testing showed no consistent significant differences between uninfected and infected lines as denoted on Figure 4, with each infected line producing slightly more H 2 O 2 than uninfected in two out of three experiments but slightly less H 2 O 2 in the third. No overall trend towards increased H 2 O 2 in Wolbachia-infected cell lines was thus observed.

H2O2 Levels in Ae. albopictus Adults and Cell Lines
All Uju mosquito and Aa23-T cell lines were assayed for H2O2 to determine if Wolbachia infection increased ROS production and if any effect correlated with Wolbachia density (Figure 3). Relative H2O2 generation in Aa23-T cells across three independent experiments was significantly different between lines as analysed by Kruskal-Wallis testing for non-normal data (A: χ 2 = 43.80, df = 3, p-value = 1.67e−09. B: χ 2 = 41.89, df = 3, p-value = 4.23e−09. C: χ 2 = 26.04, df = 3, p-value = 9.36e−06). Post-hoc bootstrap ratio testing showed no consistent significant differences between uninfected and infected lines as denoted on Figure 4, with each infected line producing slightly more H2O2 than uninfected in two out of three experiments but slightly less H2O2 in the third. No overall trend towards increased H2O2 in Wolbachia-infected cell lines was thus observed.

ROS-Generating Enzyme and Antioxidant Gene Transcription
The lack of a detectable difference in H2O2 between lines could be because ROS production is similar or alternatively because there is a compensating increase in antioxidant enzyme expression to maintain redox homeostasis [28]. To differentiate between these hypotheses, transcription of a selection of antioxidant and ROS-generating genes were assessed by qRT-PCR.
In Ae. aegypti and a 28-fold upregulation of 'Duox2' (Entrez Gene ID 5569376) was observed [11], which was described as an additional ROS-producing enzyme. However, a BLAST comparison and conserved domain analysis of the 'Duox2' protein sequence demonstrated that the gene contains neither a peroxidase nor other domains which characterise Duox enzymes [35]. It is instead a DuoxA protein, sharing 68% amino acid identity with the D. melanogaster moladietz gene encoding Numb Interacting Protein (NIP), a maturation factor that is essential for the effective functioning of Duox and is thought to aid its transport to the plasma membrane [36,37]. There was no significant increase in transcription of either Duox or DuoxA in Wolbachia-infected Ae. albopictus ( Figure 5). There was also no consistent significant change in any antioxidants assayed ( Figure 6).

ROS-Generating Enzyme and Antioxidant Gene Transcription
The lack of a detectable difference in H 2 O 2 between lines could be because ROS production is similar or alternatively because there is a compensating increase in antioxidant enzyme expression to maintain redox homeostasis [28]. To differentiate between these hypotheses, transcription of a selection of antioxidant and ROS-generating genes were assessed by qRT-PCR.
In Ae. aegypti and a 28-fold upregulation of 'Duox2' (Entrez Gene ID 5569376) was observed [11], which was described as an additional ROS-producing enzyme. However, a BLAST comparison and conserved domain analysis of the 'Duox2' protein sequence demonstrated that the gene contains neither a peroxidase nor other domains which characterise Duox enzymes [35]. It is instead a DuoxA protein, sharing 68% amino acid identity with the D. melanogaster moladietz gene encoding Numb Interacting Protein (NIP), a maturation factor that is essential for the effective functioning of Duox and is thought to aid its transport to the plasma membrane [36,37]. There was no significant increase in transcription of either Duox or DuoxA in Wolbachia-infected Ae. albopictus ( Figure 5). There was also no consistent significant change in any antioxidants assayed ( Figure 6).     Figure 6. Transcription of CuZnSOD, Gpx and OXR1 in Uju.wAlb and Uju.wMel adult females, normalised to S17 transcription and to uninfected Uju.T. Points represent mean of 3-5 pools of three females and error bars denote Bayesian 95% credible intervals. Each bar within a gene category represents an independent experiment.

Discussion
No significant elevation of ROS levels or differential regulation of any of the innate immune pathway regulator genes or of antioxidant/ROS-modulating genes was demonstrated in Ae. albopictus, except that in cell lines STAT was sometimes upregulated. Overall, these results suggest that the hypothesis of ROS-induced Toll pathway activation is responsible for DENV inhibition in Ae. aegypti [11] does not apply in Ae. albopictus. In turn, this suggests that immune upregulation and priming may not be the predominant causal factor of Wolbachia-induced DENV inhibition. There was no evidence of immune priming in the presence of Wolbachia in the natural host of wMel and wMelPop, Drosophila melangaster, despite these strain imparting resistance to pathogenic viruses of Drosophila, suggesting that immune activation is not a mechanism of viral inhibition common to all species [38][39][40][41]. Immune activation may be a major contributor to Wolbachia-mediated DENV transmission blocking in Ae. aegypti but it is likely that the importance of immune pathways differs across different Wolbachia-host-virus interactions.
Increased ROS levels were previously reported by flow cytometry and microscopy in naturally infected and antibiotic-cured Ae. albopictus cell lines [28]. Both assays utilised 2',7'-dichlorodihydrofluorescein diacetate (H 2 DCFDA), which is non-specific and will also detect some RNS [42,43]. It is thought to require a transition metal catalyst such as free iron, therefore the localisation of the probe with Wolbachia-containing vacuoles reported [28] could be affected by cellular iron distribution and specifically the known sequestration of iron in insect vacuoles [44]. To avoid non-specificity and confounding factors, a H 2 O 2 -specific assay for reduced superoxide was used here, following consideration of alternative assays [42]. The higher variability of results between each independent experiment for the cell lines compared to the adult mosquito assays may be attributed to the heterogeneity of the Aa23 cell line, which contains different cell types; control of culture period and conditions, use of cells in the growth phase at 70%-80% confluency and frequent validation of Wolbachia infection via PCR and qPCR were employed to mitigate this inherent variability as far as possible.
The lack of immune activation in Ae. albopictus also suggests that there has been a co-evolved immune adaptation and potential attenuation of immune responses to Wolbachia in this host that has had a long natural history of Wolbachia infection. This is in contrast to the chronic immune upregulation that occurs in Ae. aegypti when transinfections of Wolbachia are made into this naïve host [2,3,10,11]. A potential role of ROS and innate immunity in controlling Wolbachia proliferation/infection density is also possible; it is known that innate immunity can play a role in controlling the Rickettsiales Anaplasma phagocytophilum and Ehrlichia chaffeensis [45,46] and endosymbionts from other phyla have been shown to be regulated by AMPs [47].
The primary causes of virus inhibition in species which do not display immune activation remains open.
Some alternative hypotheses have already been discounted, for example siRNA-mediated RNA-interference (RNAi) was not implicated in D. melanogaster Wolbachia-induced viral immunity [48]. It appears that any antiviral effect is Wolbachia density dependent, as inferred from the greater DENV resistance seen in wMelPop compared to wMel-infected Ae. aegypti and Ae. albopictus cell lines [26] and strengthened by later results showing that wAlbB, which does not induce DENV resistance in its low density natural infection of Ae. albopictus, can do so in a density dependent manner in Ae. albopictus cell lines [27]. A linear positive correlation between Wolbachia density and Defensin D (DEFD) transcription was observed [27]. However, given the lack of evidence for increased AMP production [5] and immune pathway signalling in Ae. albopictus presented here, alternative hypotheses which are also compatible with density dependence must be pursued. These include Wolbachia-mediated modulation of resources such as lipids required by arboviruses for their replication; indeed an effect of cholesterol levels on the inhibition phenotype has already been demonstrated in Drosophila [49].