A wAlbB Wolbachia Transinfection Displays Stable Phenotypic Effects across Divergent Aedes aegypti Mosquito Backgrounds

ABSTRACT

Aedes mosquitoes harboring intracellular Wolbachia bacteria are being released in arbovirus and mosquito control programs. With releases taking place around the world, understanding the contribution of host variation to Wolbachia phenotype is crucial. We generated a Wolbachia transinfection (wAlbB^Q) in Aedes aegypti and performed backcrossing to introduce the infection into Australian or Malaysian nuclear backgrounds. Whole Wolbachia genome sequencing shows that the wAlbB^Q transinfection is nearly identical to the reference wAlbB genome, suggesting few changes since the infection was first introduced to A. aegypti over 15 years ago. However, these sequences were distinct from other available wAlbB genome sequences, highlighting the potential diversity of wAlbB in natural Aedes albopictus populations. Phenotypic comparisons demonstrate the effects of wAlbB infection on egg hatching and nuclear background on fecundity and body size but no interactions between wAlbB infection and nuclear background for any trait. The wAlbB infection was stable at high temperatures and showed perfect maternal transmission and cytoplasmic incompatibility regardless of the host background. Our results demonstrate the stability of wAlbB across host backgrounds and point to its long-term effectiveness for controlling arbovirus transmission and mosquito populations.

IMPORTANCEWolbachia bacteria are being used to control the transmission of dengue virus and other arboviruses by mosquitoes. For Wolbachia release programs to be effective globally, Wolbachia infections must be stable across mosquito populations from different locations. In this study, we transferred Wolbachia (strain wAlbB) to Aedes aegypti mosquitoes with an Australian genotype and introduced the infection to Malaysian mosquitoes through backcrossing. We found that the phenotypic effects of Wolbachia are stable across both mosquito backgrounds. We sequenced the genome of wAlbB and found very few genetic changes despite spending over 15 years in a novel mosquito host. Our results suggest that the effects of Wolbachia infections are likely to remain stable across time and host genotype.

INTRODUCTION

Wolbachia bacteria are intracellular, maternally inherited bacteria found within about half of all insect species (1, 2). Wolbachia-infected Aedes aegypti mosquitoes are being released into field populations as a way of controlling mosquito populations and arbovirus transmission (3). Release programs involve strains of Wolbachia that have been introduced artificially into A. aegypti, which do not harbor Wolbachia naturally (4, 5). Most Wolbachia infections in mosquitoes induce cytoplasmic incompatibility, where Wolbachia-infected males cannot produce viable offspring with females that are not infected (6–8). Wolbachia infections can also suppress virus replication within the mosquitoes, limiting their ability to transmit dengue, Zika, and other arboviruses (9). Wolbachia can spread through target mosquito populations by inducing cytoplasmic incompatibility, which provides a frequency-dependent fitness advantage to Wolbachia-infected females (10). Wolbachia infections can also be used to suppress mosquito populations through cytoplasmic incompatibility (11–13) as well as deleterious host fitness effects (14).

A variety of Wolbachia strains originating from Drosophila and mosquitoes have been introduced into A. aegypti through microinjection (15, 16). These strains show diverse effects on host fitness, virus blocking, and cytoplasmic incompatibility (3). Population replacement releases currently involve two Wolbachia strains: wMel and wAlbB. Both strains have successfully been established in natural A. aegypti populations (10, 17–19) and have similar virus-blocking effects (16, 20). Population replacement programs have demonstrated efficacy against dengue virus transmission, with fewer dengue cases in areas where Wolbachia infections are at a high frequency in the A. aegypti population (18, 21–24). However, establishment success can vary dramatically in different locations, even with the same Wolbachia strain. For instance, wMel readily invaded Australian A. aegypti populations (10, 21, 22, 25) but did not reach high frequencies in many parts of Niterói and Rio de Janeiro in Brazil despite multiple rounds of release (24, 26).

Wolbachia release success depends in part on the properties of the Wolbachia strain, including host fitness costs, cytoplasmic incompatibility, maternal transmission, virus blocking, and environmental stability. The wAlbB Wolbachia strain occurs naturally in Aedes albopictus and was the first strain transferred to A. aegypti through microinjection (6). wAlbB has limited effects on dengue virus transmission in its native host (27, 28) but shows strong blocking against a range of arboviruses in A. aegypti (16, 29–31) and complete cytoplasmic incompatibility (6). wAlbB infection induces substantial host fitness costs in A. aegypti, including reduced tolerance to starvation (32, 33) and thermal stress (34), decreased quiescent egg viability (30, 35), and high rates of female infertility following egg storage (36). However, wAlbB is relatively stable at high temperatures compared to wMel (37), prompting its deployment in field trials in Kuala Lumpur, Malaysia. Releases of wAlbB led to stable population replacement in some trial locations, with a corresponding reduction in dengue virus transmission (18). Aedes aegypti mosquitoes carrying wAlbB have also been released in population suppression programs that rely on cytoplasmic incompatibility to reduce wild-female fertility (11, 12).

When a Wolbachia strain is released into a new environment, it is important to understand the factors that can affect its establishment. Wolbachia releases are taking place in genetically divergent mosquito populations (38, 39), and host genotype may influence release outcomes. Host genotype can influence Wolbachia’s effects on host fitness (40–43), cytoplasmic incompatibility (44, 45), and viral blocking (46, 47). Wolbachia releases in new locations often involve backcrossing to introduce the target Wolbachia infection into locally adapted mosquitoes (17, 19, 48). Since Wolbachia and mitochondria are coinherited, Wolbachia infections carry along their associated mitochondria when they invade (49, 50), resulting in mosquito populations with foreign mitochondria (17, 19). Mismatches between mitochondrial and nuclear genotypes in other insects can have severe deleterious effects (51–53). However, such effects have not been tested in mosquitoes.

In this study, we generated a new wAlbB transinfection in A. aegypti with an Australian mitochondrial haplotype and nuclear background, denoted wAlbB^Q. We then used antibiotic curing and backcrossing to generate populations to investigate the effects of wAlbB infection across two nuclear backgrounds. We found that the phenotypic effects were stable across backgrounds, with no evidence of deleterious effects when mitochondrial haplotypes and nuclear backgrounds are mismatched. Whole Wolbachia genome sequencing of the wAlbB^Q transinfection revealed very few changes compared to the reference genome. Our results indicate that wAlbB infections may remain stable in their effects in divergent mosquito populations and across time.

RESULTS

Genomic analysis of the wAlbB^Q transinfection.

We sequenced the whole Wolbachia genome of the wAlbB^Q transinfection generated in this study. Its genome is almost identical to the wAlbB reference genome (54), differing by only four single nucleotide variants (SNVs), suggesting that few genetic changes have occurred since wAlbB was first transferred to A. aegypti over 15 years ago (6). The genomes of the wAlbB variants from Hainan, China, and Florida were more divergent, each differing from the wAlbB reference genome by more than 100 single nucleotide polymorphisms (SNPs), substitutions, and small indels and by the presence of at least one large deletion (>100 bp) for the Hainan variant and at least four large deletions for the Florida variant (see Appendix S3 in the supplemental material).

We performed a phylogenomic analysis with a core set of protein-encoding gene ortholog sequences conserved as single-copy genes across various Wolbachia supergroup B infections, including four wAlbB variants and two supergroup A outgroup strains. The wAlbB variants formed a monophyletic group that shared a most recent common ancestor with the Wolbachia strains wNo, wMau, and wTpre from Drosophila simulans, Drosophila mauritiana, and Trichogramma pretiosum, respectively (Fig. 1A). A separate analysis of only the wAlbB sequences placed the wAlbB^Q transinfection in a cluster with the wAlbB reference variant, in agreement with the above-described sequencing results (Fig. 1B). Relative to the wAlbB reference and wAlbB^Q transfection genomes, the Hainan variant was slightly less divergent than the Florida variant, consistent with the pattern of nucleotide variation observed for the whole-genome sequences (Appendix S3).

FIG 1.

Phylogenomic analyses of wAlbB variants. Maximum likelihood trees were constructed with RAxML-HPC using a concatenated alignment of 459 gene orthologs, comprised of 382,520 nucleotide positions. Bootstrap values are 100 unless shown otherwise; bars indicate the number of substitutions per site. The wAlbB^Q transinfection is highlighted in purple. (A) For the analysis of Wolbachia from different host species, the alignment comprised 9,469 distinct patterns, and a rooted tree was built using two supergroup A sequences as outgroups. (B) For the analysis of wAlbB variants, an unrooted tree was built using an alignment with 23 distinct patterns.

Life-history traits are influenced by wAlbB infection and nuclear background.

To test the contributions of Wolbachia infection and nuclear background to mosquito life history, we generated 8 divergent populations using a combination of embryonic microinjection, antibiotic curing, and backcrossing (Table 1 and Fig. 2). Early attempts at wAlbB transinfection with an embryo homogenate method were unsuccessful, with a low frequency of wAlbB infection in surviving F₀ adults and no transmission to the subsequent generation. wAlbB was successfully introduced to mosquitoes with an Australian mitochondrial haplotype through cytoplasm transfer, with the resulting transinfection (wAlbB^Q) showing stable transmission by the third generation (Table S1). This population was derived from a single F₀ female, which produced a single positive F₁ female offspring out of 57 tested. Success rates with this approach were similar to those in previous studies (e.g., see reference 6), but direct comparisons between studies are not possible due to methodological differences. Following backcrossing, all 8 populations had cytochrome c oxidase subunit I (COI) sequences matching their expected mitochondrial haplotype (Table 1; Table S2).

TABLE 1.

Aedes aegypti populations used in this study

Population	Source(s) (reference)	Wolbachia infection type	Mitochondrial haplotype(s)a	Nuclear background(s) (estimated % of the population)	No. of replicate populations	Expt(s)
wAlbB Au	This study	wAlbBQ	1	Australian (100)	2	All phenotypic assays, genomics
wAlbB MY	This study	wAlbBQ	1	Malaysian (∼87.5), Australian (∼12.5)	2	All phenotypic assays
Uninfected Au	This study	Uninfected (tetracycline cured)	1	Australian (100)	2	All phenotypic assays
Uninfected My	This study	Uninfected (tetracycline cured)	1	Malaysian (∼87.5), Australian (∼12.5)	2	All phenotypic assays
wAlbB US	Z. Xi et al. (6), J. K. Axford et al. (35)	wAlbB	3	Australian (∼100)	1	Microinjection only
QL	Collected from Cairns, Australia, in 2018	Uninfected (native)	1, 4, 5	Australian (100)	1	Microinjection and backcrossing only
My	T. H. Ant et al. (16)	Uninfected (native)	2	Malaysian (100)	1	Backcrossing only
wMel	T. Walker et al. (8)	wMel	1	Australian (100)	1	Heat stress only

^aSee Table S2 in the supplemental material.

FIG 2.

Experimental design and backcrossing scheme. (A) Experimental design showing the division of populations before and after tetracycline curing and before backcrossing, with (1) and (2) denoting replicate populations. (B) Backcrossing scheme showing the introgression of the wAlbB infection and Australian mitochondria into a Malaysian nuclear background. wAlbB Au females were crossed to uninfected Malaysian males for three consecutive generations. Backcrossing was also conducted with uninfected Au females into a Malaysian background. nDNA, nuclear DNA. (Mosquito illustrations by Ana L. Ramírez [97].)

Cohorts of mosquitoes from each population were evaluated for life-history traits, including larval development time, survival to pupa, fecundity, egg hatch proportion, and body size (Fig. 3). Development time and survival to pupa were not significantly affected by Wolbachia infection or nuclear background (Table 2). Nuclear background had a substantial effect on fecundity (Table 2), with mosquitoes of a Malaysian (My) background laying more eggs than those of an Australian (Au) background (Fig. 3E). wAlbB infection reduced the proportion of eggs hatching, but there was no significant effect of nuclear background (Table 2 and Fig. 3F). Mosquitoes with a Malaysian background tended to have larger wings than mosquitoes with an Australian background (Fig. 3G and H), but this effect was significant only in males (Table 2). Sex ratios of pupae (Fig. 3D) did not deviate significantly from 1:1 (df = 9 by a chi-square test [all P > 0.05]) for any population. Overall, our results indicate clear effects of nuclear background on fecundity and wing length, with wAlbB infection inducing a cost to egg hatching in both nuclear backgrounds.

FIG 3.

Life-history parameters of wAlbB-infected and uninfected Aedes aegypti mosquitoes of Australian (Au) and Malaysian (My) backgrounds. Populations were evaluated for the following traits: larval development time for females (A) and males (B), percent survival to the pupal stage (C), sex ratio (D), female fecundity (E), egg hatch proportion (F), and wing length of females (G) and males (H). Data from two replicate populations were pooled for visualization. See Table 1 for a description of each population. Each point represents data averaged across a replicate container of 100 individuals (A to D) or data from individual mosquitoes (E to H). Medians and 95% confidence intervals are shown as black lines. Medians are also shown to the right of each data set.

TABLE 2.

GLMs for life-history parameters of wAlbB-infected and uninfected Aedes aegypti mosquitoes of Australian and Malaysian backgrounds^a

Trait and factor	Type III sum of squares	df	Mean square	F	P
Female development time
Wolbachia infection	0.016	1	0.016	1.279	0.265
Nuclear background	0.085	1	0.085	6.729	0.014
Error	0.468	37	0.013
Male development time
Wolbachia infection	0.032	1	0.032	1.175	0.285
Nuclear background	0.123	1	0.123	4.571	0.039
Error	0.995	37	0.027
Survival to pupa
Wolbachia infection	0.003	1	0.003	1.379	0.248
Nuclear background	0.000	1	0.000	0.177	0.677
Error	0.075	37	0.002
Fecundity
Wolbachia infection	5.663	1	5.663	0.020	0.889
Nuclear background	4,573.419	1	4,573.419	15.874	<0.0001
Error	42,351.708	147	288.107
Egg hatch proportion
Wolbachia infection	0.595	1	0.595	10.911	0.001
Nuclear background	0.060	1	0.060	1.096	0.297
Error	7.856	144	0.055
Female wing length
Wolbachia infection	0.007	1	0.007	0.920	0.339
Nuclear background	0.034	1	0.034	4.538	0.035
Error	1.163	157	0.007
Male wing length
Wolbachia infection	0.001	1	0.001	0.377	0.540
Nuclear background	0.088	1	0.088	27.554	<0.0001
Error	0.544	171	0.003

^aP values in boldface type are significant following Bonferroni correction (adjusted α value of 0.00556).

Quiescent egg viability depends on mosquito nuclear background and wAlbB infection.

Stored eggs from each population were hatched every 3 weeks to determine quiescent egg viability. wAlbB infection greatly reduced quiescent egg viability in both the Australian and Malaysian backgrounds (Fig. 4). By week 16, hatch proportions for wAlbB-infected populations approached zero, while hatch proportions for uninfected populations exceeded 40%. In uninfected populations, a linear model (LM) indicated a significant interaction between background and week [F_{(1, 118)} = 13.193 (P < 0.001)] due to the sharp decrease in viability in the Malaysian background. There was also an effect of week in this analysis [F_{(1, 118)} = 98.967 (P < 0.001)]. Eggs with an Australian background had higher hatch proportions (median, 0.819) than eggs with a Malaysian background (median, 0.445) by the end of the experiment. In wAlbB-infected populations where we did not consider the data from week 16, there was also a significant interaction between background and week [F_{(1, 93)} = 22.948 (P < 0.001)] due to the sharper decrease in viability in the Malaysian background along with an effect of week [F_{(1, 93)} = 368.012 (P < 0.001)].

FIG 4.

Quiescent egg viability of wAlbB-infected and uninfected Aedes aegypti mosquitoes of Australian (Au) and Malaysian (My) backgrounds. Data from two replicate populations were pooled for visualization. Symbols show median egg hatch proportions, while error bars show 95% confidence intervals.

Nuclear background influences wAlbB density.

We estimated Wolbachia density in whole adults from the wAlbB Au and wAlbB My populations (Fig. 5). Wolbachia density was higher in Australian mosquitoes than in Malaysian mosquitoes for both females [F_{(1, 82)} = 6.752 by a general linear (mixed-effects) model (GLM) (P = 0.011)] and males [F_{(1, 72)} = 7.956 (P = 0.006)]. Differences in Wolbachia density could not be explained by differential host gene amplification, with similar mosquito-specific crossing point (Cp) values across nuclear backgrounds [F_{(1, 163)} = 2.811 (P = 0.096)]. However, we found a substantial effect of sex [F_{(1, 163)} = 50.311 (P < 0.001)], with females having higher Cp values than males, likely reflecting their larger size.

FIG 5.

Female (A) and male (B) Wolbachia densities in wAlbB-infected Aedes aegypti mosquitoes of Australian (Au) and Malaysian (My) backgrounds. Data from two replicate populations were pooled for visualization. Each point represents the relative density for an individual averaged across 2 to 3 technical replicates. Medians and 95% confidence intervals are shown as black lines.

wAlbB shows complete maternal transmission regardless of nuclear background.

We observed complete maternal transmission of wAlbB in both the Australian and Malaysian nuclear backgrounds, with all 10 offspring from 10 females (100/100) in each population being infected (lower 95% confidence interval, 0.96).

wAlbB is stable under cyclical heat stress regardless of nuclear background.

To test the stability of wAlbB at high temperatures in the different nuclear backgrounds, we measured Wolbachia densities in adults after eggs were exposed to cyclical heat stress (29°C to 39°C) or held at 26°C. We found no significant effect of temperature [females, F_{(1, 56)} = 2.073 (P = 0.155); males, F_{(1, 56)} = 0.676 (P = 0.414)], nuclear background [females, F_{(1, 56)} = 2.228 (P = 0.141); males, F_{(1, 56)} = 1.341 (P = 0.252)], or interactions between temperature and nuclear background [females, F_{(1, 56)} = 1.276 (P = 0.264); males, F_{(1, 56)} = 0.749 (P = 0.391)], indicating that this infection is stable under heat stress regardless of nuclear background (Fig. 6). In contrast, wMel infection decreased in density by a median of 98.7% (across both sexes) under the same conditions (Fig. 6). In this experiment, host gene amplification was influenced substantially by sex [F_{(1, 116)} = 122.026 (P < 0.001)], temperature treatment [F_{(1, 116)} = 97.171 (P < 0.001)], and, to a lesser extent, nuclear background [F_{(1, 116)} = 7.025 (P = 0.009)], which may reflect differences in body size (but was not measured directly here).

FIG 6.

Wolbachia densities in females (A) and males (B) following exposure to cyclical heat stress during the egg stage. Eggs from the wAlbB Au, wAlbB My, and wMel populations were exposed to cyclical temperatures of 29°C to 39°C for 7 days or held at 26°C. Each point represents the relative density for an individual averaged across 2 to 3 technical replicates. Medians and 95% confidence intervals are shown as black lines.

wAlbB induces complete cytoplasmic incompatibility regardless of nuclear background and heat stress.

We tested the ability of wAlbB-infected males to induce cytoplasmic incompatibility. Uninfected females that were crossed to wAlbB Au or wAlbB My males produced no viable offspring (no eggs hatched), indicating complete cytoplasmic incompatibility (Fig. 7A). wAlbB-infected males continued to induce complete cytoplasmic incompatibility following exposure to cyclical temperatures of 29°C to 39°C for 7 days during the egg stage, regardless of nuclear background. In contrast, heat stress weakened cytoplasmic incompatibility induction by wMel-infected males, with all uninfected females in this cross producing viable offspring (Fig. 7A).

FIG 7.

Cytoplasmic incompatibility induction and compatibility restoration following exposure to cyclical heat stress during the egg stage. We performed crosses to test the ability of Wolbachia-infected males to induce cytoplasmic incompatibility with uninfected females (A) and Wolbachia-infected females to restore compatibility with Wolbachia-infected males (B). In each cross, males (A) or females (B) from the wAlbB Au, wAlbB My, and wMel populations were exposed to cyclical temperatures of 29°C to 39°C for 7 days (red circles) or held at 26°C (black circles) during the egg stage. Each point represents the hatch proportion of eggs from a single female. Medians and 95% confidence intervals are shown as black lines.

We also tested the ability of wAlbB-infected females to restore compatibility with wAlbB-infected males under heat stress (Fig. 7B). wAlbB-infected females exhibited high hatch proportions regardless of nuclear background or heat stress. Heat stress had no significant effect on the egg hatch proportions of wAlbB Au (Z = 1.512 [P = 0.131] by a Mann-Whitney U test) or wAlbB My (Z = 0.585 [P = 0.562]) females but greatly reduced the egg hatch proportion of wMel-infected females (Z = 5.114 [P < 0.001]). Overall, these results show that both cytoplasmic incompatibility induction and compatibility restoration are robust to nuclear background and heat treatment.

DISCUSSION

In this study, we generated a wAlbB transinfection (wAlbB^Q) in A. aegypti and evaluated its phenotypic effects in two nuclear backgrounds. wAlbB showed 100% maternal transmission, induced complete cytoplasmic incompatibility, and was stable at high temperatures across both Australian and Malaysian backgrounds. Life-history traits were driven largely by mosquito nuclear background, with Malaysian mosquitoes having larger wings and laying more eggs than Australian mosquitoes but with lower quiescent egg viability. However, wAlbB infection induced substantial costs to egg hatching, particularly in quiescent eggs. These effects were consistent across two replicate populations. Costs to quiescent egg viability are comparable to those in previous studies (35, 36) and are likely to make Wolbachia establishment more challenging in locations where larval habitats are intermittent. We found no clear deleterious effects of having mismatched mitochondrial and nuclear genomes, supporting the use of backcrossing as a method to introduce Wolbachia infections into target backgrounds for field release. Whole Wolbachia genome sequencing of the wAlbB^Q transinfection revealed very few changes compared to the reference genome. Our results point to the stability of wAlbB infections across time, environment, and host background. The wAlbB^Q transinfection generated in this study is therefore likely to retain desirable traits for arbovirus and mosquito control under a broad range of conditions.

wAlbB infections in A. aegypti typically induce host fitness costs (16, 30, 35, 36), but these effects may depend on both the host and Wolbachia variation. D. O. Carvalho et al. (43) recently demonstrated the importance of host background, where wAlbB showed larger deleterious effects in a Mexican background than in a Brazilian background. Here, we found that wAlbB infection reduced egg hatch proportions and dramatically reduced quiescent egg viability, but the effects were consistent across two mosquito backgrounds. Although the backgrounds used by D. O. Carvalho et al. (43) were different from ours, differences in outcomes could also be explained by methodological differences. We used antibiotic curing to ensure that mitochondrial haplotypes were matched between infected and uninfected lines, while D. O. Carvalho et al. (43) compared the wAlbB-infected populations to naturally uninfected populations. However, we also used fewer rounds of backcrossing, resulting in incomplete introgression into the Malaysian background. Controlling for mitochondrial haplotype is an important consideration when elucidating Wolbachia phenotypes given that it can have substantial effects on host fitness in other insects (51–53).

We performed whole-genome sequencing of our wAlbB^Q transinfection to detect potential evolutionary changes and found that its sequence is almost identical to that of the reference genome (54). The transinfection and reference genomes have a common origin: both wAlbB infections were derived from A. albopictus mosquitoes caught in Houston, TX, in 1986 (55). However, their histories are distinct: the reference sequence was derived from a wAlbB infection maintained in the A. aegypti Aa23 cell line for more than 2 decades (56), while the transinfection was introduced to A. aegypti over 15 years ago (6). The long-term stability of wAlbB is consistent with wMel and wMelPop-CLA infections in A. aegypti mosquitoes, which also show few long-term genomic (57, 58) and phenotypic (59, 60) changes following transinfection. These results suggest that Wolbachia infections will likely remain stable in their effects following field releases, broadly consistent with phenotypic data for a separate wAlbB transinfection following releases in Malaysia (61).

The wAlbB Wolbachia strain is widespread throughout natural populations of A. albopictus but is thought to exhibit low genetic diversity (62). Previous surveys have detected little or no genetic variation within wAlbB through a limited set of molecular markers (63–66), but a recent reanalysis of A. albopictus genome sequence data points to substantial variation within this strain (67). We compared the genome sequence of our wAlbB^Q transinfection to two other publicly available sequences from A. albopictus caught in St. Augustine, FL, in 2016 (68) and Haikou, Hainan, China, in 2016 (69) and found that all three were distinct from each other. These results emphasize the need to reevaluate wAlbB diversity in A. albopictus since variation within wAlbB could influence control interventions targeting this species. This also raises the question of whether independently derived wAlbB transinfections show phenotypic differences. For instance, the wAlbB transinfection generated by Xi et al. (6) is stable at high temperatures (37), while the transinfection generated by Ant et al. (16) that has been deployed in Malaysia (18) decreases in density under a similar temperature regime (16). Given the sequence variation within wAlbB, nomenclature for differentiating between variants (as for other Wolbachia strains such as wPip [70]) may be useful, particularly if they differ in their phenotypic effects.

Aedes albopictus is native to Asia and invaded the United States only in the 1980s (71). The initial incursion into the mainland United States is thought to have originated in Japan and to have been followed by a complex pattern of successive introductions from other locations (72–75). Given that the Houston strain of A. albopictus, from which the wAlbB^Q transinfection and the reference variant are derived, was collected only 1 year after the first report of A. albopictus in the mainland United States in 1985 (76), it is likely that this strain originated in Japan. Recent population genetic studies of A. albopictus have identified clear patterns of geographic differentiation among populations within the native Asian range of the species (39, 75). Consistent with this, our phylogenomic analysis separated the wAlbB variants into three main branches: a Chinese branch, a Florida branch (of unknown origin), and a probable Japanese branch. Understanding the extent of variation within wAlbB in natural populations could provide insights into the speciation and global spread of A. albopictus.

In conclusion, we have shown that the phenotypic effects of wAlbB infection tested here (and the associated mitochondrial DNA [mtDNA]) are stable across nuclear backgrounds, providing little evidence for nuclear genome-Wolbachia interactions or changes in wAlbB associated with multiple host transfers through microinjection. Our results have implications for the releases of Wolbachia-infected A. aegypti mosquitoes that are taking place in different mosquito backgrounds around the world. They suggest that one source of infection may serve releases in multiple locations, although adequate backcrossing remains important to ensure that the nuclear background is consistent with that of the target populations for genes under local selection, such as those involved in pesticide resistance (77).

MATERIALS AND METHODS

Strain production. (i) Mosquito strains and colony maintenance.

Aedes aegypti mosquitoes were reared in temperature-controlled insectaries at 26°C ± 1°C with a 12-h photoperiod according to methods described previously by P. A. Ross et al. (78). All populations were maintained at a census size of 400 individuals in BugDorm-4F2222 (13.8-liter) or BugDorm-1 (27-liter) cages (Megaview Science C Ltd., Taichung, Taiwan). Larvae were reared in trays filled with 4 liters of reverse-osmosis (RO) water and provided with fish food (Hikari tropical sinking wafers; Kyorin Food, Himeji, Japan) ad libitum throughout their development. Embryonic microinjection experiments were performed in a general insectary, where female mosquitoes (5 to 7 days old, starved for 24 h) were fed on the forearm of a human volunteer. Blood feeding of female mosquitoes on human volunteers was approved by the University of Melbourne Human Ethics Committee (approval number 0723847). All adult subjects provided informed written consent (no children were involved). All of the following experiments were conducted in a quarantine insectary; female mosquitoes were fed human blood via Hemotek membrane feeders (Hemotek Ltd., Blackburn Lancashire, Great Britain) according to methods described previously by V. Paris et al. (79). Human blood was sourced from the Red Cross (agreement number 16-10VIC-02) and refreshed monthly.

Several Wolbachia-infected and uninfected populations were used in this study (Table 1). The wAlbB infection originated from A. albopictus mosquitoes collected in Texas (55). wAlbB was introduced to A. aegypti through embryonic microinjection (6) and repeatedly backcrossed to an Australian nuclear background (35). We then introduced wAlbB into A. aegypti with an Australian mitochondrial haplotype and nuclear background (Queensland [QL]) by embryonic microinjection (see below). We have named this transinfection wAlbB^Q based on its mtDNA of Queensland origin.

(ii) Embryonic microinjection.

To generate wAlbB-infected A. aegypti mosquitoes with an Australian mitochondrial haplotype (wAlbB^Q), we transferred wAlbB from a donor population (6, 35) into uninfected Australian A. aegypti (QL). We used two approaches to obtain Wolbachia from the donor population. In the first approach, approximately 10 pairs of ovaries were dissected from females that were blood fed 4 to 5 days earlier, crushed gently with a pestle in sucrose-phosphate-glutamic acid (SPG) buffer in a 1.7-ml Eppendorf tube, and then processed according to methods described previously by Z. Xi and S. L. Dobson (80). In the second approach, the cytoplasm was removed from donor eggs and injected directly into the recipient embryo (81).

For all experiments, recipient eggs were collected by placing cups filled with larval rearing water and lined with filter paper (diameter of 90 mm) into cages of mosquitoes that were blood fed 4 to 5 days earlier. Filter papers were replaced every 0.5 to 1 h. Eggs (<1.5 h old, light gray) were lined up on filter paper and transferred to a coverslip with double-sided tape. Eggs were left to desiccate for 1 to 2 min before being covered in halocarbon oil 700 (Sigma-Aldrich, Castle Hill, NSW, Australia). Eggs were injected using a Minj-1000 microinjection system (Tritech Research, Los Angeles, CA, USA) and left in oil for 2 h. Eggs were gently removed from the oil using a fine paintbrush, rinsed in water, and placed on a moist piece of filter paper. Eggs were conditioned at ∼80% humidity and then hatched at 3 days postinjection by submerging filter papers in containers filled with RO water, a few grains of yeast, and one tablet of fish food. Hatching larvae were reared to adulthood, and F₀ females were crossed to males from an uninfected population (QL), blood fed, isolated for oviposition, and screened for wAlbB infection after producing viable offspring. This process was repeated for three further generations, with only progeny from Wolbachia-positive females contributing to the next generation. The population was closed (with no further backcrossing) once the wAlbB infection reached fixation. We used loop-mediated isothermal amplification (LAMP) assays for the rapid detection of wAlbB during microinjection experiments according to methods described previously by M. E. Jasper et al. (82).

(iii) Antibiotic curing.

To generate uninfected A. aegypti mosquitoes with matching mitochondrial haplotypes and nuclear backgrounds, we cured the wAlbB infection using antibiotic treatment, with untreated populations reared in parallel. wAlbB Au (generation 5 after microinjection) adults were fed 2 mg/ml tetracycline hydrochloride (≥95%; Sigma-Aldrich P/L, Castle Hill, NSW, Australia) in a 10% sucrose solution for 10 days before blood feeding. Larvae from the next generation were reared in a solution of 50 μg/liter tetracycline hydrochloride according to methods described previously by N. M. Endersby-Harshman et al. (83). This process was repeated for a total of three generations of adult treatment and two generations of larval treatment. After the third generation of adult treatment, 150 females from the treated population were blood fed and isolated for oviposition, with larvae from Wolbachia-negative mothers pooled to generate uninfected populations. To control for potential effects of drift or inbreeding (84), wAlbB-infected (untreated) and uninfected (tetracycline-cured) females from each population were crossed to wild-type uninfected males of the Australian background for one generation. The resulting populations were divided into two replicate populations for backcrossing (Fig. 2A). All populations were reared in the absence of tetracycline for four or more generations prior to experiments.

(iv) Backcrossing.

To generate populations with different combinations of nuclear background (Australian or Malaysian) and Wolbachia infection status, we performed backcrosses with an uninfected Malaysian population (Fig. 2B). wAlbB Au and uninfected Au females were crossed to uninfected Malaysian (My) males. Two replicate populations for each combination were backcrossed independently for a total of four populations. Each generation, 200 female pupae and 200 male pupae were separated by sex and left to emerge in separate cages. Sexes were confirmed following adult emergence, and females were aspirated into cages of males to mate. This process was repeated for two additional generations, resulting in an estimated 87.5% similarity to the target background. Four additional populations (two replicates each of wAlbB Au and uninfected Au) were not backcrossed. Instead, these were maintained at a census size of 400 individuals each generation. Before experiments commenced, all populations were screened for wAlbB infection as well as their COI haplotype (see below) to check for contamination during antibiotic curing and backcrossing.

DNA-level characterization of Wolbachia. (i) Whole-genome sequencing.

We sequenced the whole Wolbachia genome of the wAlbB^Q transinfection generated in this study after 11 generations posttransinfection. Genomic DNA was extracted from a pooled sample of 5 individuals (<24-h-old female adults) using a DNeasy blood and tissue kit (Qiagen, Hilden, Germany). Extracted DNA was randomly fragmented to a size of 350 bp; end polished, A tailed, and ligated with Illumina sequencing adapters using a NEBNext Ultra DNA library prep kit (New England BioLabs, Ipswich, MA, USA); and further PCR enriched with oligonucleotide primers P5 and P7. The PCR products that comprised the final libraries were purified (AMPure XP system; Beckman Coulter Life Sciences, Indianapolis, IN, USA) and subjected to quality control tests that included size distribution by an Agilent 2100 bioanalyzer (Agilent Technologies, CA, USA) and molarity measurement using real-time PCR. The libraries were then pooled and sequenced on a NovaSeq 6000 instrument (Illumina) using 2- by 150-bp chemistry by Novogene HK Company Ltd., Hong Kong.

(ii) Reference genome assembly.

Quality filtering of sequencing reads was performed with Trimmomatic (85), with the following parameter settings: leading = 3, trailing = 3, slidingwindow = 4:15, and minlen = 71. Reads were aligned to a wAlbB reference genome (GenBank accession no. CP031221.1 [54]) using the Burrows-Wheeler aligner (86) with the bwa mem algorithm and default parameter settings. Subsequent quality filtering and variant calling were performed with SAMtools and BCFtools (87, 88). PCR duplicates were excluded from downstream analyses by soft masking. Reads with a mapping quality (MAPQ) score of <25 were removed from the alignment, except for reads with a MAPQ score of 0, which were permitted to allow for mapping to repetitive regions. Genomic likelihoods were calculated using a maximum of 2,000 reads per position. For variant calling, ploidy was set to haploid. The variant call output was used to create a consensus nucleotide sequence, wherein genome positions with a coverage of <5 were masked as “N.” The consensus sequence was compared to the reference genome and two additional wAlbB genomes, originating from Hainan, China, and Florida (68, 69). Polymorphic nucleotide positions were identified by aligning the genome sequences with Geneious v9.1.8. Regions of the Hainan and Florida wAlbB genomes displaying anomalous patterns of nucleotide variation, suggestive of localized sequence contamination, were annotated by eye.

(iii) Phylogenomic analysis.

Genome assemblies and corresponding protein annotation data were obtained from GenBank (see Appendix S1 in the supplemental material). Thirteen representative supergroup B genomes from various arthropod hosts were included in the analysis, including three publicly available wAlbB genomes (54, 68, 69) and two supergroup A outgroup genomes (Appendix S1). Orthofinder v2.5.2 (89) was used to identify a core set of orthologs that were present as single-copy genes in all of the Wolbachia genomes included in the analysis (Appendix S2). Ortholog nucleotide sequences were aligned with MAFFT v7.475 (90) using the “auto” setting. Alignments were trimmed with Gblocks v0.91b (91) using the following settings: minimum number of sequences for a conserved position = seqs/2 + 1, minimum number of sequences for a flank = seqs/2 + 1, maximum number of contiguous nonconserved positions = 8, minimum block length = 5, and gaps allowed = none. Sixteen ortholog alignments were excluded from subsequent analyses due to possible sequence contamination in one sample (GenBank accession no. CP041923.1) (see above). For phylogenetic analysis, the individual orthogroup alignments were concatenated to form a single alignment comprised of 459 orthogroups and 382,520 nucleotide positions. Maximum likelihood (ML) trees were constructed with RAxML-HPC v8.2.12 (92) on XSEDE, using a general time reversible (GTR)-GAMMA model with rapid bootstrapping (100 inferences). Bootstrap scores were plotted onto the best-scoring ML tree. RAxML-HPC was accessed through the CIPRES Science Gateway (93).

(iv) COI sequencing.

To confirm the successful transfer of wAlbB infection to A. aegypti mosquitoes with an Australian mitochondrial haplotype, we performed COI sequencing on six adult females from the original wAlbB population described previously by Z. Xi et al. (6) as well as nine adult females from the wAlbB Au population generated through microinjection in the current study. To check for contamination following backcrossing, we performed COI sequencing from three adult females from each of the 8 populations used in the experiments (Table 1). We performed COI amplicon sequencing according to methods described previously by H. L. Yeap et al. (50). Samples were analyzed for a 750-bp region within the COI region (positions 1994 to 2743 under GenBank accession no. EU352212.1) using forward primer UEA5 (5′-AGTTTTAGCAGGAGCAATTACTAT-3′) and reverse primer UEA10 (5′-TCCAATGCACTAATCTGCCATATTA-3′) (94). PCR amplicons from individuals were sequenced in both the forward and reverse directions using Sanger sequencing (Macrogen, Inc., Geumcheongu, Seoul, South Korea). The sequenced 750-bp region was analyzed using Geneious 9.1.8 to investigate SNP variation among samples.

(v) Wolbachia detection and density.

We used real-time PCR assays (35, 95) to confirm the presence or absence of Wolbachia infection and estimate the relative density in whole adult females and males using the Roche LightCycler 480 system. Genomic DNA was extracted using 250 μl of 5% Chelex 100 resin (Bio-Rad Laboratories, Hercules, CA) and 3 μl of proteinase K (20 mg/ml) (Roche Diagnostics Australia Pty. Ltd., Castle Hill, New South Wales, Australia). Tubes were incubated for 60 min at 65°C and then for 10 min at 90°C. Three primer sets were used to amplify markers specific for mosquitoes (forward primer mRpS6_F [5′-AGTTGAACGTATCGTTTCCCGCTAC-3′] and reverse primer mRpS6_R [5′-GAAGTGACGCAGCTTGTGGTCGTCC-3′]), A. aegypti (aRpS6_F [5′-ATCAAGAAGCGCCGTGTCG-3′] and aRpS6_R [5′-CAGGTGCAGGATCTTCATGTATTCG-3′]), and wAlbB (wAlbB_F [5′-CCTTACCTCCTGCACAACAA-3′] and wAlbB_R [5′-GGATTGTCCAGTGGCCTTA-3′]). For mosquitoes carrying wMel infection, the Wolbachia density was determined using w1 primers (w1_F [5′-AAAATCTTTGTGAAGAGGTGATCTGC-3′] and w1_R [5′-GCACTGGGATGACAGGAAAAGG-3′] [95]). Relative Wolbachia densities were determined by subtracting the crossing point (Cp) value of the Wolbachia-specific marker from the Cp value of the mosquito-specific marker. Differences in Cp values were averaged across 2 to 3 consistent replicate runs and then transformed by 2ⁿ. For the maternal transmission experiment, we required only presence/absence data, so a single run was performed per sample (with testing on any negative samples repeated to confirm a lack of infection).

Experimental comparisons. (i) Life-history parameters.

To evaluate the effects of Wolbachia infection status and nuclear background on mosquito life history, we performed phenotypic assessments of the 8 populations generated through backcrossing (Table 1). Eggs (<1 week old) from each population were hatched in trays filled with 3 liters of RO water, a few grains of yeast, and one tablet of fish food. One day after hatching, 100 larvae were counted into trays filled with 500 ml of RO water, with 5 replicate trays per population. Larvae were provided with fish food ad libitum until pupation. To determine the average larval development time and survival to pupation for each tray, pupae were counted and separated by sex twice per day (in the morning and evening). Pupae were removed from larval development trays, pooled across replicate trays, and placed in an open container of water inside a BugDorm-4F2222 (13.8-liter) cage for adults to emerge.

Groups of adults were stored within 24 h of emergence for Wolbachia density and wing length measurements. One wing each from 20 males and 20 females per population was dissected and measured for length according to methods described previously by P. A. Ross et al. (32), as an estimate of body size, with damaged wings being excluded from the analysis. Adults from the four wAlbB-infected populations (20 females and 20 males per population) were screened by real-time PCR for Wolbachia density estimation (as described above). The remaining adults were provided with a 10% sucrose solution, which was then removed 24 h prior to blood feeding. Females (6 to 7 days old) were fed human blood via Hemotek membrane feeders. Twenty engorged females per population were isolated for oviposition in 70-ml specimen cups with mesh lids that were filled with 15 ml of larval rearing water and lined with a strip of sandpaper (Norton Master Painters P80 sandpaper; Saint-Gobain Abrasives Pty. Ltd., Thomastown, Victoria, Australia). Eggs were collected 4 days after blood feeding, partially dried, and then hatched 3 days after collection by submerging sandpaper strips in containers filled with RO water, a few grains of yeast, and one tablet of fish food. Female fecundity was determined by counting the total number of eggs on each sandpaper strip, while hatch proportions were determined by dividing the number of hatched eggs (with a clearly detached egg cap) by the total number of eggs per female. Females that did not lay eggs or that died before laying eggs were excluded from the analysis.

(ii) Quiescent egg viability.

Blood-fed females remaining in cages from the previous experiment were used to test quiescent egg viability. Six cups filled with larval rearing water and lined with sandpaper strips were placed inside each cage. Eggs were collected 5 days after blood feeding, partially dried, and then placed in a sealed chamber with an open container of a saturated potassium chloride solution to maintain a constant humidity of ∼84%. Eggs were checked under a dissecting microscope to ensure that no eggs had desiccated or hatched precociously prior to storage. When eggs were 1, 4, 7, 10, 13, and 16 weeks old, small sections of each sandpaper strip were removed and submerged in water with a few grains of yeast to hatch. Four to six replicate batches of eggs were hatched per population at each time point, with 40 to 125 eggs per batch. Eggs were left to hatch for at least 4 days before scoring due to potentially slow and variable hatching times under long-term storage. Hatch proportions were determined by dividing the number of hatched eggs (with a clearly detached egg cap) by the total number of eggs per female.

(iii) Cytoplasmic incompatibility and Wolbachia density following heat stress.

We measured Wolbachia density and cytoplasmic incompatibility in adults after being exposed to cyclical heat stress during the egg stage. Eggs were collected from Wolbachia-infected populations (one replicate population each from wAlbB Au, wAlbB My, and wMel). Four days after collection, batches of 40 to 60 eggs were tipped into 0.2-ml PCR tubes (12 replicate tubes per population) and exposed to cyclical temperatures of 29°C to 39°C for 7 days in Biometra TProfessional Trio 48 thermocyclers (Biometra, Göttingen, Germany) according to methods described previously by J. D. Kong et al. (96) and P. A. Ross et al. (34). Eggs of the same age from each population, as well as uninfected Au eggs, were kept at 26°C. Eggs held at 29°C to 39°C and 26°C were hatched synchronously, and larvae were reared at a controlled density (100 larvae per tray of 500 ml water). Pupae were separated by sex, and 15 males and 15 females per population and temperature treatment were stored in absolute ethanol within 24 h of emergence for Wolbachia density estimation [see “DNA-level characterization of Wolbachia. (v) Wolbachia detection and density,” above]. The remaining pupae were separated by sex and released into BugDorm-4S1515 (5.4-liter) cages (with pupae of each sex, temperature treatment, and population held in separate cages) for cytoplasmic incompatibility crosses.

We established two sets of crosses to (i) test the ability of Wolbachia-infected males to induce cytoplasmic incompatibility and (ii) test the ability of Wolbachia-infected females to restore compatibility with Wolbachia-infected males. In the first set, uninfected Au females (untreated) were crossed with Wolbachia-infected males from each temperature treatment by aspirating females into cages of males. In the second set, Wolbachia-infected females from each temperature treatment were crossed with wAlbB Au males, except for wMel females, which were crossed to wMel (untreated) males. Females (5 to 7 days old, starved for 24 h) were blood fed, and 20 females per cross were isolated for oviposition. Hatch proportions were determined by dividing the number of hatched eggs (with a clearly detached egg cap) by the total number of eggs per female. Females that did not lay eggs or that died before laying eggs were excluded from the analysis.

(iv) Wolbachia maternal transmission.

We tested the fidelity of Wolbachia maternal transmission in each nuclear background. Females from the wAlbB Au and wAlbB My populations (one replicate population each) were crossed to uninfected Au males. Females (5 to 7 days old, starved for 24 h) were blood fed, isolated for oviposition, and then stored in absolute ethanol after laying eggs. Offspring from each female were hatched, reared in trays of 500 ml RO water, and then stored 4 days after hatching. We tested 10 mothers and 10 offspring per mother for the presence of wAlbB using real-time PCR [see “DNA-level characterization of Wolbachia. (v) Wolbachia detection and density,” above].

(v) Statistical analysis.

All data were analyzed using SPSS Statistics version 24.0 for Windows (SPSS, Inc., Chicago, IL). Data sets were tested for normality with Shapiro-Wilk tests and transformed where appropriate (data for egg hatch proportions and survival to pupa were logit transformed). Life-history, quiescent egg viability, and Wolbachia density data were analyzed with general linear (mixed-effects) models (GLMs). We tested the effects of Wolbachia infection (wAlbB infected or uninfected), nuclear background (Au or My), and interactions between Wolbachia infection and nuclear background. Replicate populations were pooled for analysis when the effects of replicate population exceeded a P value of 0.1 in prior analyses. Where interaction terms were not significant, the terms were dropped, and the models were rerun without interactions. Data for each sex were analyzed separately. For quiescent egg viability, hatch proportions differed substantially between wAlbB-infected and uninfected populations. We therefore ran separate GLMs for wAlbB-infected and uninfected populations, with egg storage duration included as a continuous covariate for this trait. Replicate population (nested within nuclear background) was included as a random factor but was not significant in any instance. Quadratic egg storage duration was also included as a factor in the GLM in the case of a nonlinear relationship between egg hatch proportion and storage duration in these populations. We used chi-squared tests to determine whether sex ratios for each population differed significantly from 1:1. Bonferroni corrections were performed on P values when multiple traits were measured in the same experiment.

For Wolbachia density, untransformed data (i.e., differences in Cp values between Wolbachia and mosquito markers, before 2ⁿ transformation) were used to test for the effects of the nuclear background and the temperature treatment (26°C or 26°C to 39°C) as a factor. We performed separate GLMs on Cp values of mosquito-specific markers to test whether differences in Wolbachia densities could be explained by differential host DNA amplification. We were unable to perform direct comparisons between wMel and wAlbB strains due to the use of different markers for each strain; we therefore excluded wMel from the overall analysis but presented it graphically. Egg hatch proportions from cytoplasmic incompatibility crosses were analyzed with Mann-Whitney U tests.

Data availability.

Novel study data have been deposited in the NCBI under BioProject accession no. PRJNA751963, BioSample accession no. SAMN20571400, and GenBank accession no. CP080546.