Wolbachia density changes seasonally amongst populations of the pale grass blue butterfly, Zizeeria maha (Lepidoptera: Lycaenidae)

Previous studies showed that the survival rate of Wolbachia decreases under high temperature in incubators. It is also known that a high density of Wolbachia in the host body reduces the host emergence rate, while low densities fail to change reproduction rates. However, few studies have examined the density of Wolbachia in hosts in the field. Here, we focus on Wolbachia infection of the pale grass blue butterfly, Zizeeria maha (Lepidoptera: Lycaenidae), which is distributed throughout the Japanese islands. We examined the rate and density of Wolbachia infection in the bodies of butterflies at thirteen locations in Japan. At seven of these places, we collected butterflies in different seasons to determine seasonal differences in the infection rate and density and found that Wolbachia density has seasonal differences within the same population. Moreover, to determine whether Wolbachia density has a geographical cline, we compared the infection density of Wolbachia amongst all geographical populations. In addition, we determined the sequences of Wolbachia wsp and host mtDNA CO1 haplotypes of all populations. The results showed that Wolbachia density increased in early summer and decreased in autumn. Further, the density of Wolbachia infecting the same strain of Z. maha varied amongst populations, although no tendency in geographical cline was observed.


Introduction
Wolbachia are endosymbionts belonging to the alpha-proteobacteria [1]. This bacterium is transmitted from mother to progeny [1] and causes a variety of reproductive manipulations in its hosts [2]. These manipulations were classified into four types; feminization of males, induction of thelytokous parthenogenesis, male killing, and cytoplasmic incompatibility (CI) [1][2][3]. Of these phenomena, CI is the most common manipulation by Wolbachia. CI affects sperm [4], and thus the infected males play an important role in the infected population. The eggs a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 derived from an infected male and an uninfected female do not hatch. As a result, Wolbachiainfected individuals dominate the host population.
External environmental conditions significantly affect the infective states of hosts including the density of the endosymbionts inhabiting the host body. For example, high temperatures are not salubrious for Wolbachia, even temperatures that they are normally exposed to in the fields. Some studies showed that high temperatures eliminated Wolbachia from Drosophila melanogaster, Tetranychus urticae, and other species, but these studies were conducted in the laboratory [3,[5][6][7][8]. Moreover, the density of Wolbachia infecting host bodies affects the success of the reproductive manipulation. In Culex pipiens, high Wolbachia density decreases the host emergence rate [9], but low Wolbachia density fails to manipulate host reproduction in Drosophila innubila [10].
Laboratory studies have shown that different geographical strains of some insects have different densities of Wolbachia. For examples, Mouton et al. (2007) compared Wolbachia densities between progenies of crosses of strains of a wasp with different densities of Wolbachia [11]. The F1 hybrid showed the mean Wolbachia density of their mothers. This suggests that Wolbachia density is maintained by the host genotype. On the other hand, it has been suggested that Wolbachia density changes with the season [10]. In addition, Unckless et al. (2009) suggested that Wolbachia density changes with the age of the host [10]. However, little is known about whether these phenomena occur in nature. Although Toju and Fukatsu (2011) reported that the Wolbachia infection rate in Curculio sikkimensis has a geographical cline, they did not reveal whether Wolbachia density has a geographical cline [12].
Previous laboratory studies suggested that infection rates and density of Wolbachia differ in host populations that inhabit temperate zones in summer [13]. In the present study, we hypothesized the existence of a geographical cline of Wolbachia density. This hypothesis expects that the abundance of Wolbachia is terminated or decreases in the host body in hot temperature regions, and Wolbachia density thereby varies amongst geographical populations. To verify this hypothesis, Zizeeria maha (Lepidoptera; Lycaenidae) is an ideal subject because the Wolbachia infection status has been studied in this butterfly, which inhabits the main island of Japan from north to south [13]. This makes it possible to verify the Wolbachia infection status of the butterfly in geographically separated populations. Furthermore, it is easy to compare the Wolbachia density of wild-caught butterflies, which are obtainable from spring to autumn. In the present study, we collected Z. maha butterflies across different latitudes and longitudes, and verified the rates of Wolbachia infection and density for the different populations using qPCR.
Studies of Wolbachia using multilocus sequence typing (MLST) have demonstrated the power to discriminate between Wolbachia strains [14]. Here, we also conducted MLST analysis and reproductive manipulation experiments to identify and characterize the effects of the Wolbachia strain that infects Z. maha.

Material and methods Sampling
We collected eight to 31 Z. maha males at thirteen points with different latitudes and longitudes in the Japanese temperate region  Fig 1. These collection sites covered an extensive habitat of Japan, except Hokkaido Island, the northern part of Japan, where this species is not distributed. We also did not collect from the Okinawa Islands where a sub-species is distributed.
We collected flying adults by butterfly net, and the collected samples were almost all males (more than 90%, T. Sumi, unpublished data) because almost all flying adults were male. Thoraces width of collected samples was measured their thorax width under a stereomicroscope. After that, the samples were stored in 99.5% ethanol. We investigated the Wolbachia strains that infected Z. maha.

DNA extraction
DNA was extracted from each specimen's abdomen (excluding the first to the third sternites). The whole abdomen including gonads and testes was crushed by a toothpick in an Eppendorf tube (1.5ul) with Chelex 1 100 (Bio-Rad) 98ul+Proteinase K (TaKaRa). The samples were incubated three hours at 56˚C (spinning samples from 6 to 8 times each 30 min) to dissolve protein, and then they were incubated in a heat block set at 100˚C for 3 min. After that, the sample was cooled to 5˚C, and stored at -20˚C.
We used CO1 primers [15] to confirm DNA by PCR. After confirmation of DNA, we used wsp primers to investigate Wolbachia infection [16]. To investigate their sequence, we used Wolbachia strains that infected Z. maha as host, Wolbachia MLST primers [13], CO1 [16], and wingless primers [17]. CO1 and wingless primers were used to investigate the phylogenetic relationships of samples. Wolbachia causes a selective sweep that shows confused genetic link between nuclear DNA and mitochondrial DNA. Accordingly, we selected wingless as a nuclear gene marker. Sequence analyses were carried out by Fasmac Co. Ltd. (Kanagawa, Japan). Wsp, CO1, and wingless gene sequences were assembled using DNA Baser version 4.2 (Heracles Bio-Soft, Pitesti, Romania), and aligned by Mega 5 (Accession numbers: LC125171-LC125181).
We developed qPCR primers from analyzed DNA sequences. We amplified the sample of Sendai No. 1, which was the standard of qPCR in this study, by PCR of each wsp or CO1 primer pair. After that, PCR products were diluted 10 times in 10 steps. For real-time qPCR, we used iQ SYBR Green Supermix (Biorad, Tokyo) and followed the supplier's protocol. Then, we obtained standard curves of CO1 and wsp from the standard sample. Next, we amplified each sample DNA by qPCR on CO1 and wsp primer pairs, and qPCR cycles were fitted by each standard curve. Then we obtained copies of each CO1 and wsp gene. We considered that the butterfly cells were approximated by CO1 gene copies. Therefore, to evaluate Wolbachia density, each gene copy number was calculated by a standard curve, and wsp was normalized to CO1. All primers are shown in electronic supplementary S2 Table. We always used Sendai No. 1 as the standard for creating this standard curve.
To compare Wolbachia infection rates amongst each geographical population, Fisher's exact test was used. Wolbachia densities in each population were tested by a Kruskal-Wallis rank sum test, and followed by a Steel-Dwass test (referenced by http://aoki2.si.gunma-u.ac.jp/ R/src/Steel-Dwass.R). To determine the relationship between Wolbachia density and collection season, we used a linear model from the eight points that were collected over several months. In each model, collection month and Wolbachia density were used as explanatory and response variables, respectively, for the eight points. Then, the models were tested by ANOVA. To investigate the correlation between latitude or longitude and Wolbachia density, we used the Pearson product-moment correlation coefficient. All statistical analyses were performed using R version 3.1.2 [18] Our butterfly collection sites did not require specific permissions to catch the invertebrates. Moreover, this butterfly is not protected by any law. Further, we do not conflict with any guideline of PLoS ONE because we used only the butterfly Z. maha.

Reproductive manipulation
To determine the type of Wolbchia infecting the butterfly, i.e., to examine the control of reproduction by Wolbachia infection, seven females collected from Okayama City in 2015 were allowed oviposit on the leaves of Oxalis corniculata. The larvae hatched from the eggs were divided into two groups, i.e., larvae were provided leaves grown in DW with 0.5% tetracycline (Tc) or without tetracycline (C), namely in DW only. Larvae were reared in the laboratory at 25˚C, RH 60%, and 16L:8D dark-light cycles, and a laboratory population was established. Adults emerging via pupae from these larvae were reared in a hydroponic culture system (Green Farm UH-A01E1: U-ING, Osaka, Japan) in which Oxalis corniculata were planted. The larvae of the 1st and 2nd generations were reared on food including tetracycline (Tc) or not including tetracycline (C). Then, the larvae of 3rd and 4th generations were reared on Oxalis corniculata alone to eliminate the effect of tetracycline.
At the fourth successive generation, reciprocal crossing was conducted; i.e., females of Tc groups were paired with Tc and C males, and females of C groups were paired with Tc and C males. Each pair was allowed to copulate for two weeks for all crosses. The numbers of eggs oviposited by these females (n = 3 for each cross) and hatching rates were counted.

Field population
MLST revealed that the Wolbachia strain did not differ amongst the Z. maha populations (LC125171-LC125176). The Wolbachia alleles revealed by the MLST analysis included coxA 14, fbpB 4, ftsZ 36, gatB 39, and hcpA 40. Therefore, the Wolbachia type was identified as ST41 by Wolbachia MLST Databases (https://pubmlst.org/wolbachia/). In the host genotype, the CO1 sequence differed only in Sendai (LC1251810). Similarly, the wingless sequence differed from other populations in Kagoshima (LC125179) and Tsukuba (LC12577). Numbers of infected adults with the numbers collected and tested for Wolbachia density, indexes of density of Wolbachia infection, and body sizes of Z. maha are shown in the S1 Table. Wolbachia infection rates were not significantly different amongst populations in all monthes examined (P > 0.05). However, Wolbachia infection density significantly differed amongst populations collected in October (S1 Fig, Tsukuba, 0.1, 0.  Nakatsu, P = 0.055, Saiki, P = 0.028). Then we examined whether the same seasonal change in Wolbachia density occurred in each population. Statistical analysis showed that Wolbachia density increased in early summer and decreased in autumn. However, there was a significant difference in the effects of month and population for each population, and the month × population interactions was also significantly different (Fig 3, month: P < 0.01, population: P < 0.01, month × population: P < 0.005).

Discussion
MLST conducted in the present study revealed that the Wolbachia strain infecting Z. maha was ST41. ST41 is prevalent in many butterfly species [19,20] and causes cytoplasmic incompatibility (CI), as shown in the present results by reproductive manipulation [21]. No seasonal difference was observed in the infection rates of Wolbachia. However, we found that Wolbachia densities fluctuate seasonally. The density was significantly higher from spring (Apr to June) to early summer (July) than in late summer (Aug), and it decreased during late summer and autumn (Sept and Oct) (Fig 3, S1 Table). Compared with median Wolbachia density in population by season, the highest difference was 89 times between from spring (24.08) to autumn (0.27) in Kagoshima, and from late summer to autumn the difference was around two times in Tsukuba and Okayama. Accordingly, the seasonal fluctuation of Wolbachia density in hosts might be affected by the environment around the host habitat. It has been reported that Wolbachia density varied significantly between seasons within a population by annual observation of D. innubila [10]. Unckless et al. (2009) suggested that individuals with a lower Wolbachia density were older than individuals that had a high density [10]. However, the adult longevity of Z. maha is short (about one to two weeks in laboratory conditions), and the change of generations is about one month [22]. Therefore, we do not think that only older individuals were collected in the late summer (the season of lowest Wolbachia density).
Although Wolbachia densities in Z. maha were significantly different amongst populations, a geographical cline of Wolbachia density was not observed. In the present result, the "geographical cline of Wolbachia density hypothesis" was dismissed. By median Wolbachia density, the highest population (Niimi) was 137 times higher than the lowest population (Tsukuba) in October. Surprisingly, Wolbachia density has an extremely wide range in natural population and at least difference of six times in density in the present study. On the other hand, Wolbachia density had small range in previous studies in the laboratory (e.g. Mouton et al., 2007). This result suggests that environmental factors are related to the seasonal changes of Wolbachia density in the Japanese temperate region. Hence, we suggest that temperature is the most significant factor in seasonal changes of Wolbachia density in populations in a temperate region.
Given that Wolbachia is a maternally-transmitted bacterium, seasonally-induced changes in Wolbachia density are likely more important in female insects than in males. However, we collected mostly males in this study, as described in the introduction. Therefore, additional studies using female samples are required. Because it is difficult to collect flying females in the field, analyzing samples of the larvae or pupae collected from fields is required in the future. However, we consider that it is suitable to use males in our study. Because of this, we revealed this Wolbachia strain was characterized as ST41, which causes CI. Furthermore, CI was caused by males.
An interaction between food quality and Wolbachia density was revealed by several studies in the laboratory. For example, Wolbachia density in Aedes albopictus was changed by the host juvenile diet [23], although the change in Wolbachia density was reported only in females [23]. Moreover, another study reported that food quality did not affect Wolbachia density in Drosophila innubila [10]. When Wolbachia density is high, it strongly manipulates host reproduction. Why a good nutrient condition did not affect Wolbachia density is unclear. If Wolbachia density in Z. maha males is not also changed by nutrient conditions, it suggests efficient Wolbachia transmission from mother and environment included seasonal nutrients affecting Wolbachia density (see S3 Fig). On the other hand, not only environmental conditions but genetic mechanisms of the host regulate Wolbachia density [11]. Furthermore, other endosymbionts may affect regulation of Wolbachia density [23]. In particular, it is possible that the difference in Wolbachia density amongst populations corresponds to competition between Wolbachia and other endosymbionts. In a future study, we hope to determine whether Z. maha is infected with other endosymbionts that compete with Wolbachia.
The present study confirmed that the rate of Wolbachia infection in Z. maha did not differ amongst populations in Japan, as reported in our previous study [13]. This result also suggests that Wolbachia strongly affect host reproduction of Z. maha. Furthermore, if Wolbachia infection strongly affects other aspects of host performance, it is shown in the following examples in Z. maha. For example, Charlat et al. (2007) reported that sudden male killing by Wolbachia did not manipulate Hypolimnas bolina. Similarly, Armadillidium vulgare have a resistance factor to feminization by Wolbachia [24]. These results suggest that if Z. maha does not suffer costs from Wolbachia infection, infection rates may change in the future. We considered further experimentation is required to unravel this issue. On the other hand, Wolbachia protect their host from virulent RNA viruses [25][26][27]. One reason for the high infection rate by Wolbachia of Z. maha is that it provides some benefit to the host.
In conclusion, our results provide the first evidence that Wolbachia density in Z. maha populations differs between seasons in the field and is higher in spring than in summer. However, the Wolbachia infection rate of this species did not change. This seasonal change may be prevalent in wild invertebrates infected by Wolbachia.