Abstract
Endosymbiotic Wolbachia bacteria infect divergent arthropod and nematode hosts. Many strains cause cytoplasmic incompatibility (CI) that kills uninfected embryos fertilized by Wolbachia-modified sperm. Infected embryos are protected from CI, promoting Wolbachia spread to high equilibrium frequencies balanced by imperfect maternal transmission. CI strength varies widely in nature and tends to decrease as males age. Understanding the causes of CI-strength variation is crucial to explain Wolbachia prevalence in host populations. Here, we investigate how fast and why CI strength decreases with male age in two model systems: wMel in Drosophila melanogaster and wRi in D. simulans. Average wMel CI strength decreases rapidly (19%/ day), and wRi CI strength decreases slowly (6%/ day) as males age; thus, within three days, wMel-infected males do not cause CI, whereas twelve-day-old wRi-infected males still cause minor, yet significant, CI. We tested if reductions in Wolbachia densities or CI gene expression as males age could explain this pattern. Indeed, wRi densities and CI gene expression decrease in testes as males age, but wMel densities and CI gene expression surprisingly increase with male age as CI strength decreases. Phage WO lytic activity and wMel Octomom copy number—an ampliconic gene region that influences wMel proliferation—do not explain age-dependent Wolbachia densities. However, the expression of Relish, an essential gene in the Drosophila immune deficiency pathway, strongly correlates with wMel densities. Together, these results suggest that testes-wide Wolbachia density and CI gene expression are insufficient to explain age-dependent CI strength across strains and that Wolbachia density is variably impacted by male age across Wolbachia-host associations. We hypothesize that host immunity may underlie variation in age-dependent density dynamics. More broadly, the rapid decline of wMel CI strength during the first week of D. melanogaster life likely contributes to wMel frequency variation observed on several continents.
Introduction
Reproductive parasites manipulate host reproduction to facilitate their maternal transmission. These endosymbiotic microbes may kill or feminize males or induce parthenogenesis to bias sex ratios in favor of females [1]. More frequently, reproductive parasites cause cytoplasmic incompatibility (CI) that reduces embryonic viability when aposymbiotic females mate with symbiont-bearing males (Fig. 1A) [2]. Females harboring a comparable symbiont are compatible with CI-causing symbiotic males of the same strain, providing symbiont-bearing females a relative advantage that encourages symbiont spread to high frequencies in host populations [3–6]. Divergent Cardinium [7], Rickettsiella [8], Mesenet [9], and Wolbachia [10] endosymbionts cause CI. Of these, Wolbachia are the most common, infecting 40-65% of arthropod species [11, 12]. Wolbachia cause CI in at least ten arthropod orders [2], and pervasive CI directly contributes to Wolbachia spread and its status as one of the most common endosymbionts in nature.
Within host populations, Wolbachia frequencies are governed by their effects on host fitness [13–16], the efficiency of maternal transmission [17–19], and CI strength (% embryonic death) [3, 5]. CI strength varies from very weak to very strong and produces relatively low and high infection frequencies, respectively. For example, wYak in Drosophila yakuba causes weak CI (∼15%) and tends to occur at intermediate and often variable frequencies (∼40-88%) in west Africa [18, 20]. Conversely, wRi in D. simulans causes strong CI (∼90%) and occurs at high and stable frequencies (e.g., ∼93% globally) [4,21–23]. In D. melanogaster, wMel CI strength is relatively weak [24–26], contributing to infection frequencies that vary considerably on multiple continents [27–31]. In contrast, wMel usually causes complete CI (no eggs hatch) in transinfected Aedes aegypti mosquitoes [32–35]. Vector control groups use this strong CI to either suppress mosquito populations through the release of wMel-infected males [36–40] or to drive pathogen-blocking wMel to high and stable frequencies to inhibit pathogen spread [32,41,42].
Despite CI’s importance for explaining Wolbachia prevalence in natural systems and reducing human disease transmission in transinfected mosquito systems, the mechanistic basis of CI-strength variation remains unresolved. Two hypotheses are plausible. First, the bacterial density model predicts that CI is strong when bacterial density is high (Fig. 1B) [43]. Indeed, Wolbachia densities positively covary with CI strength across Drosophila-Wolbachia associations [44, 45] and with variable CI within strains [33,34,46–52]. Second, the CI gene expression hypothesis predicts that higher CI gene expression contributes to stronger CI (Fig. 1B) [53]. In Drosophila, two genes (cifA and cifB) associated with Wolbachia’s temperate bacteriophage (WO) induce CI when expressed in testes [53–57], and one gene (cifA) rescues CI when expressed in ovaries [56–58]. CI strength covaries with transgenic cif expression in D. melanogaster [53, 57], and natural cif expression covaries with CI strength in Habrobracon ectoparasitoid wasps [59]. Bacterial density may explain CI strength via cif expression but may not perfectly align with CI strength since Wolbachia variably express cifs across conditions that impact CI strength [53]. Thus, the bacterial density and cif expression hypotheses are not mutually exclusive. It remains unknown if cif expression is responsible for CI-strength variation and if it covaries with Wolbachia density in natural Drosophila-Wolbachia associations.
If symbiont density is a crucial factor governing CI strength, what governs the change in density? There are several plausible drivers of Wolbachia density variation. First, phage WO is a temperate phage capable of cell lysis in some Wolbachia strains [59–62]. Lytic phage form particles that burst through the bacterial cell membrane, killing the bacterial host. The phage density model proposes that as phage densities increase, Wolbachia densities decrease (Fig. 1B) [46]. Temperature-induced phage lysis covaries with lower Wolbachia densities and CI strength in some parasitoid wasps [46, 59], though it is unknown if phage lysis influences Wolbachia densities in any other systems. Second, wMel Wolbachia have a unique ampliconic gene region composed of eight genes termed “Octomom” [63, 64]. Octomom copy number varies among wMel Wolbachia between host generations and positively covaries with Wolbachia densities (Fig. 1B), but effects of Octomom-dependent Wolbachia densities on CI have not been investigated. Third, theory predicts that selection favors the evolution of host suppressors [6], as observed for male killing [65, 66]. Indeed, CI strength varies considerably across host backgrounds [20,25,35,67], supporting a role for host genotype in CI-strength variation. The genetic underpinnings and mechanistic consequences of host suppression remain unknown, but two models have been proposed [2]. The defensive model suggests that host CI targets diverge to prevent interaction with cif products, and the offensive model suggests that host products directly interfere with Wolbachia density or the proper expression of cif products (e.g., through immune regulation) (Fig. 1B). Only a taxon-restricted gene of Nasonia wasps has been functionally determined to contribute to Wolbachia density variation [68]; thus considerable work is necessary to uncover host determinants of variation in Wolbachia density. Since Wolbachia densities significantly contribute to several phenotypes [47, 69], investigation of the causes of Wolbachia density variation are sorely needed.
CI strength within Wolbachia-host systems covaries with several factors, including temperature [25,33,34,46,59], male mating rate [70, 71], male development time [72], rearing density [72], nutrition [73], paternal grandmother age [26], and male age [3,16,23,25,70]. Changes in CI strength with male age are particularly notable. Older males cause weaker CI in wMel-infected D. melanogaster [25] and wRi-infected D. simulans [3,16,23]. Age-dependent CI seems particularly strong for wMel [3,16,23,25], although the precise rates of CI-strength decline have not been estimated. While several factors might contribute to age-dependent CI strength, the precise mechanistic underpinnings of this phenotype remain unknown.
Here, we investigate how fast and why Wolbachia densities and CI strengths vary with male age in two model Wolbachia that diverged 0.6-6 million years ago [74]: wMel in D. melanogaster and wRi in D. simulans. First, how fast does CI strength decrease with male age? Second, is Wolbachia density consistently correlated with age-related CI strength, as predicted by the bacterial density model? If so, does phage WO lysis, Octomom copy number, or host immune gene expression correlate with density? Third, does cif expression consistently correlate with CI strengths and Wolbachia densities, as predicted by the CI gene expression model? This study is the first to test the cif expression hypothesis in either system with age and is the highest resolution investigation of Wolbachia density variation across age to date. Our results suggest that testes-wide Wolbachia densities and cif expression alone do not explain age-dependent CI-strength relationships across Wolbachia-host associations. While phage WO and Octomom copy number do not covary with the age-dependent Wolbachia density variation we observe, immune expression in D. melanogaster positively correlates with wMel densities. We discuss how these data contribute to our understanding of the causes of age-dependent CI strength and Wolbachia density variation and the consequences for Wolbachia prevalence in nature.
Results
How much does CI strength vary with age?
CI manifests as embryonic lethality (Fig. 1A). As such, we measured CI strength as the percent of embryos that hatch from a mating pair’s clutch of offspring. Our experiments use males of different ages to test the impact of male age on CI strength. Here, we define age as days since eclosion where males paired with females the day they eclosed are considered 0-days-old. For wMel, we measured CI strength daily across the first three days of male age (Fig. 2A) and separately every two days across the first eight days of male age (Fig. 2B). This design enabled us to determine the rate of CI decline and the ages where males no longer cause significant CI. Crossing uninfected D. melanogaster females and males yields high levels of compatibility (Fig. 2A; 95% confidence interval of the mean = 74 - 93%). Young 0-day-old wMel-infected males induce strong CI when mated with uninfected females (95% interval = 9 - 27%). wMel-infected females significantly rescue CI caused by infected 0-day-old males (95% interval = 87 - 92%, P = 1.74E-12). Crosses using older 1- (95% interval = 31 - 51%), 2- (95% interval = 53 - 73%), and 3-day-old (95% interval = 69 - 83%) infected males trend toward progressively weaker CI (Fig. 2A). Average wMel CI strength decreases daily by 19.3%: 22.8% from 0- to 1- day-old males, 21.8% from 1- to 2-day-old, and 13.4% from 2- to 3-day-old. Crosses between uninfected females and 3-day-old males (95% interval = 69 - 83%) do not cause significant CI, with egg hatch similar to the compatible uninfected (95% interval = 74 - 93%; P = 0.35) and rescue (95% interval = 87 - 92%; P = 0.19) crosses. This highlights the rapid decline of wMel CI strength with D. melanogaster male age.
In the age group that includes older males (Fig. 2B), the uninfected cross also yields high compatibility (95% interval = 72 - 88%). 0-day-old infected males cause strong CI when crossed with uninfected females (95% interval = 8 - 15%), and infected females significantly rescue 0-day-old CI (95% interval = 83 - 91%; P = 2.51E-12). Older 2- (95% interval = 59 - 73%), 4- (95% interval = 66 - 83%), 6- (95% interval = 76 - 92%), and 8-day-old (95% interval = 77 - 91%) infected males cause weaker CI as males age (Fig 2B). CI crosses using 4-day-old or older males do not significantly differ in egg hatch from the compatible uninfected cross (P = 1 in all cases). These data suggest that average wMel CI strength decreases by approximately 19.3% each day as D. melanogaster males age, but this rate of decrease slows each day, such that CI is no longer statistically detectable once males are 3-days-old.
Next, we assess age-dependent CI in wRi-infected D. simulans (Fig. 2C). As expected, uninfected D. simulans females and males are compatible (95% interval = 74 - 94%). Young 0- day-old wRi-infected males cause strong CI when mated with uninfected females (95% interval = 0 - 1%), and infected females significantly rescue 0-day-old CI (95% interval = 59 - 84%; P = 1.83E-10). Older 4- (95% interval = 21 - 39%), 8- (95% interval = 54 - 64%), and 12-day-old (95% interval = 64 - 82%) infected males induce progressively weaker CI as males age. Average wRi CI strength decreases by about 6.0% per day: 29.1% (7.3%/ day) from 0-day-old to 4-day-old males, 29.0% (7.3%/ day) from 4-day-old to 8-day-old, and 14.0% (3.5%/ day) from 8-day-old to 12-day-old. These data support a strong effect of D. simulans male age on wRi CI strength, but the daily decrease is more than three times slower than what we observe for wMel CI strength decline as D. melanogaster males age.
What causes CI strength to vary with age?
The bacterial density and CI gene expression hypotheses are both proposed to explain CI-strength variation. These hypotheses predict that Wolbachia density and/or cif expression positively covary with CI strength. To elucidate the causes of declining CI strength with male age, we test both hypotheses in the context of rapidly declining wMel CI strength and more slowly declining wRi CI strength in D. melanogaster and D. simulans, respectively.
Bacterial density differentially covaries with age between species
We tested the bacterial density hypothesis by dissecting testes from siblings of flies used in our CI assays above, extracting DNA, and measuring the relative abundance of a single-copy Wolbachia gene (FtsZ) relative to a single-copy ultraconserved element (UCE) [75] of Drosophila via qPCR. We selected a random infected sample from the youngest 0-day-old age group as the reference for all fold change analyses within each experiment. Surprisingly, 0-day-old D. melanogaster testes have low wMel density (Fig. 3A; 95% interval = 0.53 - 1.01), and older 2- (95% interval = 0.92 - 1.11), 4- (95% interval = 0.96 - 1.72), 6- (95% interval = 1.17 - 1.49), and 8-day-old (95% interval = 1.19 - 1.51) infected testes have progressively higher wMel densities (Fig. 3A). wMel densities are significantly different among age groups according to a Kruskal-Wallis test (Fig. 3A; P = 1.1E-03). To test for a correlation between wMel densities and CI strength, we performed Pearson (rp) and Spearman (rs) correlations on the relationship between wMel fold change against median hatch rates from the associated age groups. Indeed, wMel densities are significantly positively correlated with decreasing CI strength (Table S3; rp = 0.75, P = 5.5E-06; rs = 0.77, P = 2.3E-06). wMel densities also covary with age (Fig. S1; P = 0.02) and correlate with decreasing CI strength (Table S3; rp = 0.64, P = 7.7E-04; rs = 0.64, P = 7.4E-04) in the younger 0-, 1-, 2-, and 3-day-old D. melanogaster age group.
Next, we tested the bacterial density model in wRi-infected D. simulans. In contrast to wMel, wRi-infected 0-day-old (95% interval = 0.82 - 1.36) D. simulans testes have the highest wRi densities that consistently decrease in 4- (95% interval = 0.41 - 0.83), 8- (95% interval 0.41 - 0.83), and 12-day-old (95% interval = 0.24 - 0.40) testes (Fig. 3B). wRi densities are significantly different among D. simulans age groups (P = 3.9E-04) and are significantly negatively correlated with decreasing CI strength (Table S3; rp = −0.84, P = 2.4E-07; rs = −0.89, P = 6.9E-09).
In conclusion, these data fail to support the bacterial density hypothesis for age-dependent CI-strength variation in wMel-infected D. melanogaster but support the hypothesis in wRi-infected D. simulans. Thus, testes-wide Wolbachia densities alone cannot explain age-dependent CI across Wolbachia-host associations, suggesting that other factors must contribute to these patterns. Next, we investigate if cif expression covaries with age-dependent CI.
cif expression varies with age, but the direction differs between strains
cif expression is the proximal mechanistic force hypothesized to control CI-strength variation within Wolbachia-host associations [2, 53]. cif loci are classified into five different phylogenetic clades called ‘Types’ [53,76–78]. wMel has a single pair of Type I cifs, and wRi has two identical pairs closely related to the wMel copy plus a divergent Type 2 pair [53]. Since wMel density increases as CI strength decreases, we predicted that cifwMel[T1] expression would decrease in age relative to the host. Since wMel densities increase with male age, wMel would need to express cifwMel[T1] at lower levels in older males. Contrary to our first prediction, the relative expression of cifAwMel[T1] to D. melanogaster β Spectrin (βspec), a Drosophila membrane protein with invariable expression with age (see Materials and Methods for details), is low in 0-day-old infected males (95% = 1.1 - 1.6) and consistently increases in 2- (95% interval = 1.5 - 3.2), 4- (95% interval = 1.9 - 2.3), 6- (95% interval = 2.1 - 2.8), and 8-day-old (95% interval = 0.9 - 3.8) testes (Fig. 4A). Relative expression of cifAwMel[T1] to βspec significantly varies across male age (P = 8.4E-03) and is significantly positively correlated with decreasing CI strength (Table S3; rp = 0.61, P = 6.4E-04; rs = 0.59, P = 9.7E-04). Comparably, relative expression of cifBwMel[T1] to βspec also significantly increases with male age (Fig. S2A; P = 7.3E-03). Moreover, analysis of raw quantification cycle (Cq) variation with age supports increased cifAwMel[T1] (Fig. S2C; P = 3.1E-04) and cifBwMel[T1] (Fig. S2D; P = 1.1E-03) expression; βspec Cq does not vary with age (Fig. S2E; P = 0.1) and FtsZ Cq significantly decreases with age (Fig. S2F; P = 1.3E-04). Thus, we report for the first time that testes-wide cif expression is not sufficient to explain CI-strength variation, leading us to reject the hypothesis that testes-wide cifwMel[T1] expression can explain age-dependent wMel CI strength.
However, relative expression of cifAwMel[T1] to wMel FtsZ is highest in 0-day-old infected-D. melanogaster testes (95% interval = 0.9 - 1.1), and consistently decreases in 2- (95% interval = 0.7 - 0.8), 4- (95% interval = 0.7 - 0.9), 6- (95% interval = 0.6 - 0.7), and 8-day-old (95% interval = 0.4 - 0.9) testes (Fig. 4B). Relative expression of cifAwMel[T1] to wMel FtsZ significantly varies with age (P = 2.9E-03) and is significantly correlated with decreasing CI strength (Table S3; rp = −0.8, P = 4.0E-07; rs = −0.7, P = 3.5E-05). Similarly, relative expression of cifBwMel[T1] to wMel FtsZ does not significantly covary with age (Fig. S2B; P = 0.3), but is significantly correlated with decreasing CI strength (Table S3; rp = −0.42, P = 3.7E-02; rs = −0.46, P = 2.2E- 02). These data are in line with prior reports that wMel expression of cifAwMel[T1] and cifBwMel[T1] decrease as males age [53].
We also tested if the relative expression of cifAwMel[T1] to cifBwMel[T1] varied with age. Intriguingly, cifA/BwMel[T1] relative expression does not significantly covary with age (Fig. 4C; P = 0.09), but is positively correlated with decreasing CI strength (Table S3; rp = −0.61, P = 1.3E-03; rs = −0.46, P = 0.021). In summary, these data suggest that cifwMel[T1] expression per wMel decreases as males age, that cifAwMel[T1] expression decreases marginally faster than cifBwMel[T1], and that overall cifwMel[T1] expression increases relative to the host as males age and CI strength decreases. This is the first report that CI strength is decoupled from Wolbachia densities and cif expression in testes.
Next, we investigated the cif expression hypothesis in wRi. We predicted that cifwRi[T1] and/or cifwRi[T2] expression would decrease relative to host expression. Since wRi density decreases with age, cif expression per wRi would not need to change to accomplish this shift in relative expression. As predicted, relative expression of cifAwRi[T1] to D. simulans βspec is highest in infected 0-day-old (95% interval = 0.7 - 1.7) testes, and declines in 4- (95% interval = 0.1 - 0.4), 8- (95% interval = 0.3 - 0.7), and 12-day-old (95% interval = 0.2 - 0.3) testes (Fig. 4D). Relative expression of cifAwRi[T1] to D. simulans βspec significantly covaries with age (P = 1.2E-03) and is significantly correlated with decreasing CI strength (Table S3; rp = −0.76; rs = - 0.88). Similarly, relative expression of cifBwRi[T1] (Fig. S3A; P = 2.3E-03), cifAwRi[T2] (Fig. S3C; P = 1.9E-03), and cifBwRi[T2] (Fig. S3E; P = 1.2E-03) to D. simulans βspec also decreases with age and each are significantly correlated with decreasing CI strength (Table S3). As with wMel-infected D. melanogaster testes, relative expression of cifAwRi[T1] to wRi FtsZ significantly covaries with male age (Fig. 4E; P = 4.1E-02) and is significantly correlated with decreasing CI strength (Table S3; rp = −0.47, P = 0.032; rs = −0.47, P = 0.033). However, 0- (95% interval = 0.9 - 1.2), 4- (95% interval = 0.9 - 1.2), and 8-day-old (95% interval = 0.8 - 1.2) testes have similar expression patterns, suggesting that expression in 12-day-old (95% interval = 0.5 - 0.9) testes drives this significant difference; however, a Dunn’s test was unable to identify significantly different pairs (Fig. 4E). Conversely, cifBwRi[T1] (Fig. S3B; P = 0.6), cifAwRi[T2] (Fig. S3D; P = 0.2), and cifBwRi[T2] (Fig. S3F; P = 0.2) expression relative to wRi FtsZ did not vary with age or decreasing CI strength (Table S3).
Finally, as with wMel, we investigated the relationship between cifA and cifB expression in wRi across age and found similar results where cifAwRi[T1] expression relative to cifBwRi[T1] expression does not significantly vary with male age (Fig. 4F; P = 0.2) but does significantly correlate with decreasing CI strength (Table S3; rp = −0.44, P = 0.045; rs = −0.46, P = 0.035). Relative expression of cifAwRi[T1] to cifAwRi[T2] expression does not covary with age (Fig. S3G; P = 0.6) or decreasing CI strength (Table S3; rp = 0.01, P = 0.96; rs = −0.05, P = 0.84). Analysis of raw Cq values supports decreasing cifAwRi[T1] (Fig. S3H; P = 1.0E-03), cifBwRi[T1] (Fig. S3I; P = 8.1E-04), cifAwRi[T2] (Fig. S3J; P = 1.8E-03), and cifBwRi[T2] (Fig. S3K; P = 1.7E-03) expression with male age; D. simulans βspec Cq does not vary with age (Fig. S3L; P = 0.6) and wRi FtsZ Cq significantly increases with age (Fig. S3M; P = 8.9E-04). In summary, cifwRi expression significantly decreases with age in wRi testes, cifAwRi[T1] expression decreases marginally faster than cifBwRi[T1] expression, and there is a small decrease in cifAwRi[T1] expression relative to wRi but other cifwRi loci do not follow similar trends.
In conclusion, we find that wMel cif expression does not explain age-dependent CI-strength variation. More specifically, wMel’s expression of cif genes decreases with age [53], relative wMel and wRi cifA-to-cifB expression varies marginally with age, and cif expression dynamics vary considerably across male age and differ between wMel- and wRi-infected hosts.
What causes Wolbachia density to vary with age?
We find that testes-wide Wolbachia density significantly increases with male age in wMel-infected D. melanogaster and significantly decreases with male age in wRi-infected D. simulans. The causes of age-dependent Wolbachia density variation have not been explored. We test three plausible hypotheses. Namely, that phage lytic activity, Octomom copy number, or host immune expression may govern age-dependent Wolbachia densities.
Phage density does not covary with age-dependent Wolbachia density
The phage density model predicts that Wolbachia density negatively covaries with phage lytic activity [46]. Since phage lysis corresponds with increased phage copy number [46, 59], we tested the phage density model by measuring the relative abundance of phage to Wolbachia FtsZ using qPCR. wMel and wRi each harbor a unique set of phage haplotypes: wMel has two phages (WOMelA and WOMelB), and wRi has four (WORiA-C, WORiB is duplicated) [79]. We monitored WOMelA and WOMelB of wMel simultaneously using primers that target homologs present in a single copy in each phage. Conversely, we monitored WORiA, WORiB, and WORiC separately since shared homologs are too diverged to make suitable qPCR primers that match multiple phage haplotypes.
First, we evaluate the phage density model for wMel. We predicted the relative abundance of WOMelA/B to decrease with D. melanogaster male age since wMel density increases with age. However, there is no change in WOMelA/B abundance relative to wMel FtsZ as males age (Fig. 5A; P = 0.3), while WOMelA/B abundance relative to D. melanogaster UCE increases similar to wMel density (Fig. S4A; P = 3.0E-04). Relative phage abundance is not significantly correlated with decreasing wMel CI strength (Table S3; rp = −0.065, P = 0.75; rs = 0.17, P = 0.39). Similarly, WOMelA/B significantly varies with age relative to UCE (Fig. S4B; P = 0.049) but not wMel FtsZ (Fig. S4C; P = 0.15) in the 0-, 1-, 2-, and 3-day-old age experiment.
Next, we predicted that WORi phage abundance would increase with decreasing wRi densities across D. simulans male age if governed by the phage density model. As with wMel in D. melanogaster, relative WORiB to wRi FtsZ abundance does not significantly covary with male age (Fig. 5B; P = 0.053) or correlate with decreasing CI strength (Table S3; rp = 0.032, P = 0.88; rs = 0.12, P = 0.58). Relative WORiB to D. simulans UCE abundance increases with age, similar to wRi density (Fig. S4D; P = 4.4E-04). Comparably, WORiA (Fig. S4E; P = 0.3) and WORiC (Fig. S4F; P = 0.4) abundance relative to wRi did not vary with male age. These data suggest that phage WO is unrelated to age-dependent Wolbachia density variation in wMel and wRi.
Octomom does not vary with age-dependent wMel density
The relative abundance of Octomom to Wolbachia genes positively covaries with wMel density [64, 80]. We tested if Octomom copy number variation correlates with age-dependent wMel density variation using qPCR. Only wMel encodes all eight Octomom genes, and Octomom amplification is rapid and unstable, commonly changing between generations. We found that the relative abundance of an Octomom gene (WD0509) to wMel FtsZ does not covary with male age (Fig. 5C; P = 0.53) or correlate with decreasing CI strength in the older age group (Table S3; rs = −0.19, P = 0.36; rs = 0.1, P = 0.61). Similar results were observed in 0-, 1-, 2-, and 3-day-old wMel-infected males (Fig. S5; Table S3). We conclude that Octomom copy number is unrelated to the age-dependent increase in wMel densities.
Relish expression is positively correlated with age-dependent wMel, but not wRi, densities
Theory predicts that natural selection favors the evolution of host genes that suppress CI [6]. Manipulation of Wolbachia densities is one mechanism that may drive CI suppression [2]. Since the immune system is designed to control bacterial loads, we investigated the role of the host immune system in Wolbachia density variation across male age. The immune deficiency (Imd) pathway is broadly involved in defense against gram-negative bacteria like Wolbachia [81]. Bacteria activate the Imd pathway by interacting with peptidoglycan recognition proteins which start a signal cascade that results in the expression of the NF-κB transcription factor Relish (Rel). Relish then activates antimicrobial peptide production.
We predicted that D. melanogaster Relish expression and wMel density would be correlated if the Imd pathway is involved in wMel density regulation. Indeed, relative expression of Relish to βspec is lowest in 0-day-old (95% interval = 0.9 - 1.1) infected testes and consistently increases in 2- (95% interval = 1.1 - 1.8), 4- (95% interval = 1.3 - 1.7), 6- (95% interval = 1.9 - 2.3), and 8-day-old (95% interval = 1.5 - 3.9) testes (Fig. 5D). Relative expression of Relish to βspec significantly varies among age groups (P = 6.1E-4) and is significantly positively correlated with wMel FtsZ to βspec within testes samples (Fig. 5E; rp = 0.77, P = 2.5E-06; rs = 0.87, P = 1.4E-06).
Conversely, relative expression of D. simulans Relish to βspec does not significantly covary with age (Fig. 5F; P = 0.7), but remains positively correlated with the relative expression of wRi FtsZ to βspec within testes samples according to Pearson, but not Spearman, analyses (Fig. 5G; rp = 0.55, P = 0.012; rs = 0.13, P = 0.59). In summary, Relish expression is positively correlated with age-dependent wMel densities in D. melanogaster, but less so in wRi-infected D. simulans, supporting a role for the Imd pathway in the regulation of at least wMel density variation. Importantly, since wMel and wRi density are differentially associated with immune expression, Imd activity may represent a novel mechanism separating the age-dependent density dynamics in these systems. These data highlight that age-dependent Wolbachia density variation may have multiple mechanistic underpinnings.
Discussion
Within Wolbachia-host systems, several factors influence CI strength [25,26,33,34,46,59,70–73], but male age can be particularly impactful [3,16,23,25]. Our results elucidate how fast and why CI strength declines as males age. First, we estimate that CI-strength decreases rapidly for wMel-infected D. melanogaster (19%/ day), becoming statistically insignificant when males reach three days old. In contrast, wRi causes intense CI that declines more slowly (6%/ day), resulting in statistically significant CI through at least the first 12 days of D. simulans male life. Second, testes-wide Wolbachia densities and cif expression increase in wMel-infected D. melanogaster and decrease in wRi-infected D. simulans as males age and CI weakens, indicating that testes-wide bacterial density and CI gene expression cannot fully account for age-dependent CI strength across host-Wolbachia associations. Third, while WO phage activity and Octomom copy number cannot explain Wolbachia density variation, D. melanogaster immune expression covaries with wMel densities, suggesting the host immune system may contribute to age-dependent Wolbachia density in D. melanogaster, but much less so in D. simulans. We discuss how our discoveries inform the basis of age-dependent CI-strength variation, how multiple mechanistic underpinnings likely govern age-dependent Wolbachia densities, and how age-dependent CI may contribute to Wolbachia frequency variation observed in nature.
Testes-wide Wolbachia density and CI gene expression do not fully explain age-dependent CI-strength variation
Since CI strength decreases with age for both wMel-infected D. melanogaster and wRi-infected D. simulans, we predicted that Wolbachia densities and cif expression would also decrease with age. Indeed, wRi densities and cif expression are highest in young males and decrease significantly with age, supporting both the bacterial density and cif expression hypotheses for wRi. However, the opposite is true for wMel—both wMel densities and cif expression increase with male age as CI strength decreases, indicating that testes-wide Wolbachia density and cif expression are insufficient to explain age-dependent CI-strength variation in wMel-infected D. melanogaster. Despite support that CI strength is linked to Wolbachia density and cif expression across and within systems [33,34,44–47,53,59], these observations add to a growing body of literature suggesting Wolbachia densities in adult testes [26, 72] and, for the first time, cif expression, are insufficient to explain CI-strength variation broadly. We discuss three hypotheses to explain the disconnect between testes-wide Wolbachia density, cif expression, and CI strength with male age.
First, the localization and density of Wolbachia and cif products within specific cells in testes may more accurately predict CI strength. Indeed, the proportion of infected spermatocyte cysts covaries with CI strength in natural and transinfected combinations of CI-inducing Wolbachia and D. melanogaster, D. simulans, D. yakuba, D. teissieri, and D. santomea [44, 45]. Intriguingly, two wRi-infected D. simulans strains whose Wolbachia cause variable CI did not have different Wolbachia densities according to qPCR, but the number of infected sperm cysts covaries with CI between strains [82]. Thus, testes-wide Wolbachia densities may not reflect the cyst infection frequency, but it is unknown how generalizable this discrepancy is across or within Wolbachia-host associations with variable CI strengths. It seems plausible that while wMel densities increase in the testes as males age, the proportion of infected spermatocytes could decrease. Notably, since wMel infections increase drastically as males age, a considerable shift in localization and density dynamics would be necessary. Microscopy assays are required for future work to test if Wolbachia and cif localization explains wMel age-dependent CI-strength variation.
Second, age-dependent CI may be governed by developmental constraints of CI-susceptibility. For instance, the paternal grandmother age effect, where sons of older virgin females cause stronger CI than sons of younger females, covaries with Wolbachia densities in embryos but not in adult males [26]. Intriguingly, temperature-sensitive CI-strength variation in Cardinium-infected Encarsia wasps is also decoupled from symbiont densities, but CI strongly correlates with pupal development time [83, 84]. Cardinium CI effectors likely have more time to interact with host targets at critical stages of pupal development when slowed by cool temperatures, despite lower Cardinium density [83, 84]. These studies suggest that sperm are modified in spermatogenesis before adult eclosion, and that variation in symbiont densities during early development can contribute to CI-strength variation. If modified sperm are primarily produced during pupal or larval development, then younger adult males would have a higher proportion of CI-modified sperm than older males in their seminal vesicle since older males continue to produce sperm as adults. Since CI strength decreases faster in D. melanogaster than in D. simulans, this hypothesis predicts that adult D. simulans sperm production is slower and/or that CI modification occurs for an extended time. Functional work is necessary to determine if CI modification is developmentally restricted.
Finally, age-dependent CI may be related to the availability of CI-effector targets with male age and not the abundance of cif products. Indeed, the number of genes transcribed by D. melanogaster increases from 7,000 in embryos to over 12,000 in adult males, and nearly a third of genes are not expressed until 3rd instar [85]. As adult males age, the number of transcribed genes continues to vary, though less so than during metamorphosis [85]. These data support the possibility that host targets of CI may vary in abundance as males age. However, since transgenic cif expression can significantly enhance CI strength above wild-type levels [53], there are circumstances when natural cif expression is not high enough to saturate all targets—it is unknown if similar experimental approaches can strengthen age-dependent CI. More work will be necessary to determine the host genes that modify CI and how those factors vary in expression relative to CI strength.
Age-dependent bacterial density covaries with immune expression, not phage or Octomom
We report a strong relationship between male age and Wolbachia densities that differ between systems: densities decrease in wRi-infected D. simulans and increase in wMel-infected D. melanogaster. These findings add to a growing body of literature reporting age-dependent variation in Wolbachia densities across age in different tissues and sexes [44, 86], but the basis of this variation remains unexplored. We investigated the cause(s) of this variation for the first time. First, we tested whether phage or Octomom covary with Wolbachia densities. Despite prior reports that phage WO of Nasonia and Habrobracon Wolbachia can regulate temperature-dependent Wolbachia densities [46, 59] and that Octomom copy number correlates with wMel densities [64, 80], we found that neither covaries with age-dependent Wolbachia densities in testes.
We next asked whether host genes regulate age-dependent Wolbachia densities. Wolbachia are gram-negative bacteria and encode the genes necessary to synthesize peptidoglycan, which can activate the host Imd pathway to produce antimicrobial peptides (AMPs) for immune defense [87, 88]. Thus, host immune genes were attractive candidates for the regulation of Wolbachia densities. Here, we report that Relish expression, which activates AMP production in the Imd pathway [81], increases with D. melanogaster male age and strongly correlates with increased wMel densities. Conversely, Relish does not vary with D. simulans male age and is only very weakly correlated with wRi densities. Relish expression is the only factor we investigated that differentiates the density dynamics of these strains and is an exciting candidate gene for host manipulation of Wolbachia density dynamics. To our knowledge, this is the first report that host immunity covaries with Wolbachia density. We propose two non-exclusive hypotheses to explain the relationship between wMel densities and Relish expression.
First, wMel rapidly proliferates as males age and elicit an immune response proportional to their infection density. Since established Wolbachia are bound in host-derived membranes [89], wMel may largely evade the host immune response [11]. Indeed, AMP gene expression only covaries with infection state in transinfected [90–93], and not established infections [94–97], suggesting that Wolbachia can be targeted by Imd but adapt to avoid its effects. Thus, the Drosophila immune system may be attempting, but unable, to control age-dependent Wolbachia densities. This hypothesis does not explain differences between wMel and wRi densities since it assumes age-dependent wMel densities increase independent of Imd expression. Thus, an alternative mechanism unrelated to immune expression may contribute to variation in age-dependent Wolbachia densities across species.
Second, Imd expression increases independent of Wolbachia infection but impacts Wolbachia densities. Indeed, aging in D. melanogaster is associated with increased expression of AMPs, Relish, and other immune genes [98–104], and age covaries with increased gut microbial loads [98–100,105–107]. Why gut bacterial loads increase with D. melanogaster age remains unknown; but age-dependent immune expression may damage the epithelium, lead to dysbiosis through differential effects on gut microbial members, alter gut tissue renewal and differentiation, and/or cause cellular inflammation [81, 108]. To our knowledge, we report the first case where endosymbiont densities increase with age-dependent immune expression, suggesting that the cause(s) of age-dependent bacterial proliferation apply to more than gut microbes. Such age-dependent immune expression may be host restricted since Relish expression was essentially invariable with age in D. simulans males and only weakly correlated with wRi densities. Functional and cell biological assays are needed to reinforce the relationship between host immunity, other novel host factors, and age-dependent Wolbachia densities. Mapping additional host factors that modulate Wolbachia densities will be particularly useful.
Age-dependent CI strength could contribute to Wolbachia frequency variation in nature
We can consider our estimates of age-dependent CI strength in the context of an idealized discrete-generation model of Wolbachia frequency dynamics first proposed by Hoffmann et al. (1990). This model incorporates imperfect maternal transmission (μ), Wolbachia effects on host fitness (F), and the proportion of embryos that hatch in a CI cross relative to compatible crosses (H) [3]. Across all experiments, CI strength (sh = 1 - H) progressively decreases as males age (Table S2): wMel CI strength decreases quickly (Day 0 sh = 0.860; Day 8 sh = −0.007) and wRi CI strength decreases relatively slowly (Day 0 sh = 0.991; Day 8 sh = 0.244). Small negative values of sh indicate that the CI cross has a slightly higher egg hatch than the compatible crosses.
wRi occurs globally at high and relatively stable infection frequencies, consistent with generally strong CI [4, 22], while wMel varies in frequency on several continents. In eastern Australia, wMel frequencies range from ∼90% in the tropical north to ∼30% in the temperate south [30]. While the factors that maintain this cline are unresolved, mathematical modeling suggests clinal differences in CI strength likely contribute [30]. For example, CI must be essentially nonexistent (sh << 0.05) to explain relatively low wMel frequencies observed in temperate Australia, assuming little imperfect transmission (μ = 0.01 - 0.026) [109]. Conversely, with μ = 0.026 and similarly low-to-nonexistent CI (sh ≤ 0.055), large and positive wMel effects on host fitness (F ∼ 1.3) are required to explain higher wMel frequencies observed in the tropics. Though, explaining higher tropical frequencies becomes easier with stronger CI (sh > 0.05) or more reliable wMel maternal transmission (μ < 0.026) (Kriesner et al. 2016).
So what is wMel CI strength in nature? Field-collected males from near the middle of the Australian cline to the northern tropics cause very weak (sh ∼ 0.05) to no CI (Hoffmann et al. 1998). These, and other data from the middle of the cline [25], led Kriesner et al. (2016) to conjecture that the plausible range of sh in subtropical/tropical Australian populations is sh = 0 - 0.05, but < 0.1. In our study, only 6-(sh = −0.006) and 8-day-old (sh = −0.007) wMel-infected males exhibited CI weaker than sh = 0.1, suggesting that field-collected males causing little or no CI [109] are older than 4 days. Though, interactions among male age, temperature, remating, and other factors likely contribute to weaker CI in younger males [25,33,34,46,59,70,71]. Future analyses aimed at disentangling the contributions of male age and other factors to CI-strength variation are sorely needed. These estimates, along with estimates of Wolbachia transmission rate variation across genetic and abiotic contexts [18], are ultimately required to better understand Wolbachia frequency variation in host populations [18,20,30,110,111].
Conclusions
Our results highlight that testes-wide Wolbachia densities and cif expression are insufficient to explain age-dependent CI strength and that no single mechanism is likely to explain age-dependent Wolbachia densities. While age-dependent CI strength in wRi aligns with the bacterial density and CI gene expression hypotheses without the need to consider other factors, wMel CI strength cannot be explained by either of these hypotheses. We propose that localization, development, and/or host genetic variation contribute to this relationship. Moreover, wMel densities increase and wRi decrease as their respective hosts age. Neither phage WO nor Octomom explain age-dependent Wolbachia density, but variation in these systems covaries with the expression of the immune gene Relish. This represents the first report that the host immune system may contribute to variation in Wolbachia density in a natural Wolbachia-host association. This work motivates an extensive analysis of Wolbachia and cif expression in the context of localization and development, and a thorough investigation of the relationship between host genes and Wolbachia density and CI phenotypes. Finally, Incorporating the age-dependency of CI into future modeling efforts may help improve our ability to explain temporally and spatially variable Wolbachia infection frequencies, as incorporating temperature effects on wMel-like Wolbachia transmission has [18,20,112]. Ultimately this will help explain Wolbachia’s status as the most prevalent endosymbionts in nature.
Materials and Methods
Fly lines
All fly lines used in this study are listed in Table S4. Uninfected flies were derived via tetracycline treatment in prior studies [14, 53]. Tetracycline cleared lines were used in experiments over a year after treatment, avoiding the effects of antibiotic treatment on mitochondria [113]. We regularly confirmed infection status by using PCR to amplify the Wolbachia surface protein (wsp). An arthropod-specific 28S rDNA was also amplified and served as a control for DNA quality [20, 74]. DNA was extracted for infection checks using a squish buffer protocol. Briefly, flies were homogenized in 50 uL squish buffer per fly (100mL 1M Tris-HCL, 0.0372g EDTA, 0.1461g NaCl, 90 mL H2O, 150uL Proteinase K), incubated at 65°C for 45m, incubated at 94°C for 4m, centrifuged for 2m, and the supernatant was used immediately for PCR.
Fly care and maintenance
Flies were reared in vials with 10mL of food made of cornmeal (32.6%), dry corn syrup (32%), malt extract (20.6%), inactive yeast (7.8%), soy flour (4.5%), and agar (2.6%). Fly stocks were maintained at 23°C between experiments. Flies used for virgin collections were reared at 25°C, virgin flies were stored at 25°C, and experiments were performed at 25°C. Flies were always kept on a 12:12 light:dark cycle. Flies were anesthetized using CO2 for virgin collections and dissections. During hatch-rate assays, flies were mouth aspirated between vials.
Hatch-rate assays
CI manifests as embryonic death. We measured CI as the percentage of embryos that hatch into larva. Flies used in hatch rates were derived from vials where flies were given ∼24hr to lay to control for rearing density [72]. In the morning, virgin 6-8 day females were added individually to vials containing a small ice cream spoon filled with fly food. Spoon fly food was prepared as described above, but with blue food coloring added, 0.1g extra agar per 100mL of food, and fresh yeast smeared on top. After 4-5hr of acclimation, a single virgin male was added to each vial. The age of virgin males varied by experiment and cross. Paternal grandmother age was not controlled, but paternal grandmothers were non-virgin when setting up vials for fathers. Since Wolbachia densities associated with older paternal grandmothers are reduced upon mating [26], we do not expect variation in paternal grandmother Wolbachia densities across experiments or conditions. Vials with paired flies were incubated overnight at 25°C. Flies were then aspirated into new vials with a fresh spoon. Vials were incubated for another 24hr before flies were removed via aspirating. Embryos were counted on spoons immediately after flies were removed. After 48hr, the number of remaining unhatched eggs were counted. The percentage of embryos that hatched was calculated.
Relative abundance assays
Siblings from hatch-rate assays were collected for DNA extractions. Virgin males were anesthetized and testes were dissected in chilled phosphate-buffered saline (PBS). Five pairs of testes were placed into a single 1.5mL Eppendorf tube and stored at −80°C until processing. All tissue was collected the day after the hatch-rate setup. Tissue was homogenized using a pestle, and the DNeasy Blood and Tissue kit (Qiagen) was used to extract and purify DNA.
qPCR was used to measure the relative abundance of the host, Wolbachia, phage WO, and Octomom products. Samples were tested in triplicate using Powerup SYBR Green Master Mix (Applied Biosystems), which contains a ROX passive reference dye. Unless otherwise noted, all primers were designed using Primer3 v2.3.7 in Geneious Prime [114]. Host primers target an ultraconserved element (UCE) Mid1 identified previously [75]. Phage genes were also identified from prior works [79]. Primers for wMel’s phages target both WOMelA (WD0288) and WOMelB (WD0634), while those for wRi are unique to a single phage haplotype. WORiA, WORiB, and WORiC were measured with wRi_012460, wRi_005590/wRi_010250, and wRi_006880 primers, respectively. Only wMel has all eight Octomom genes (WD0507-WD0514) [63]. We measured wMel Octomom copy number using primers targeting WD0509. Primer sequences and PCR conditions are listed in Table S5. Fold difference was calculated as 2-ΔΔCt for each comparison. A random sample in the youngest age group was selected as the reference.
Gene expression assays
Siblings from hatch-rate assays were collected for RNA extractions. Virgin males were anesthetized, and testes were dissected in chilled RNase-free PBS. Fifteen pairs of testes were placed into a single 2mL tube with 200 uL of Trizol and four 3 mm glass beads. Tissue was kept on ice between dissections. Samples were then homogenized using a TissueLyser II (Qiagen) at 25Hz for 2m, centrifuged, and stored at −80°C until processing. All tissue was collected the day after the hatch-rate setup.
Samples were thawed, 200uL of additional Trizol was added, and tissue was further homogenized using a TissueLyser II at 25Hz for 2m. RNA was extracted using the Direct-Zol RNA Miniprep kit (Zymo Research) following the manufacturer’s recommendations, but with an extra wash step. On-column DNase treatment was not performed. The ‘rigorous’ treatment protocol from the DNA-free kit (Ambion) was used to degrade DNA in RNA samples. Samples were confirmed DNA-free using PCR and gel electrophoresis for an arthropod-specific 28S rDNA [20, 74]. The Qubit RNA HS Assay Kit (Invitrogen) was used to measure RNA concentration. Samples within an experiment were diluted to the same concentration. RNA was converted to cDNA using SuperScript IV VILO Master Mix (Invitrogen) with either 200ng or 500ng of total RNA per reaction depending on the experiment. qRT-PCR was performed using 1ng of cDNA per reaction using Powerup SYBR Green Master Mix (Applied Biosystems). All samples were tested in triplicate.
Primers for expression included host reference, Wolbachia reference, cif, and host immune genes. Primers to Drosophila genes for qRT-PCR were selected from FlyPrimerBank [115]. Since Drosophila expression patterns change with age [85], a host gene that is invariable with male age was selected to act as a reference gene for relative expression analyses. We selected an invariable gene using the Drosophila Gene Expression Tool (DGET) to retrieve modENCODE gene expression data for ribosome and cytoskeletal genes [116]. DGET reports expression as Reads Per Kilobase of transcript, per million mapped reads (RPKM), and included data for adult males 1, 5, and 30 days after eclosion. β-spec (1 Day = 81 RPKM, 5 Day = 80, 30 Day = 79) was selected because it is largely invariable across age. Our results confirm invariable expression across male age (Fig. S2E; Fig. S3L). D. melanogaster and D. simulans are identical across βspec primer binding sequences. All other primers were designed using Primer3 in Geneious Prime [114] and are listed in Table S5. Fold difference was calculated as 2-ΔΔCt for each comparison. A random sample in the youngest age group was selected as the reference.
Statistical analyses
All statistics were performed in R [117]. Hatch rate, relative abundance, and expression assays were analyzed using a Kruskal-Wallis followed by a Dunn’s multiple comparisons test. Kruskal-Wallis and Dunn’s P-values are reported in Table S1. Correlations between hatch rate and expression or relative abundance measures were performed using Pearson and Spearman correlations in GGPubR [118]. Correlation statistics are reported in Table S3. 95% confidence intervals were calculated using the classic MeanCI function in DescTools [119]. 95% BCa intervals were calculated using boot.ci in boot [120]. Samples with fewer than ten embryos laid were excluded from hatch-rate analyses. Samples with a Cq standard deviation exceeding 0.4 between triplicate measures were excluded from qPCR and qRT-PCR analyses. Figures were created using GGPlot2 [121], and figure aesthetics were edited in Affinity Designer 1.8 (Serif Europe, Nottingham, UK).
Data availability
All data are made publicly available in the supplement of this manuscript.
Supporting Information
Acknowledgments
We thank Michael Turelli for helpful feedback and members of the Cooper Lab for providing assistance throughout this study: Will Conner for help identifying phage gene targets for qPCR, Mike Hague for support with BCa estimates of H, Kelley Van Vaerenberghe and John Statz for review of earlier versions of the manuscript, and Tim Wheeler for assisting in the laboratory. This work was supported by a National Institutes of Health R35 GM124701 to BSC and a National Science Foundation Postdoctoral Research Fellowship DBI-2010210 to JDS. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors(s) and do not necessarily reflect the views of the National Institutes of Health or the National Science Foundation.
References
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.
- 29.
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.
- 38.
- 39.
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.
- 49.
- 50.
- 51.
- 52.↵
- 53.↵
- 54.
- 55.
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.
- 61.
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.
- 92.
- 93.↵
- 94.↵
- 95.
- 96.
- 97.↵
- 98.↵
- 99.
- 100.↵
- 101.
- 102.
- 103.
- 104.↵
- 105.↵
- 106.
- 107.↵
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.↵
- 114.↵
- 115.↵
- 116.↵
- 117.↵
- 118.↵
- 119.↵
- 120.↵
- 121.↵