Evolutionary relationship of fat body endoreduplication and queen fecundity in termites

Abstract Endoreduplication or nuclear genome replication without cell division is widely observed in the metabolically active tissues of plants and animals. The fat body cells of adult female insects produce abundant yolk proteins and become polyploid, which is assumed to accelerate egg production. Recently, it was reported that in termites, endopolyploidy in the fat body occurs only in queens but not in the other females; however, the relationship between the fecundity and ploidy level in the fat body remains unclear. Termite queens exhibit a huge variation in their egg‐producing capacity among different species; queens in the species with a foraging lifestyle, in which workers leave the nest to forage outside, are much more fecund than those in the species living in a single piece of wood. In this study, we conducted ploidy analyses on three foragings and three wood‐dwelling termites via flow cytometry. In all the species, the fat body of queens contained significantly more polyploid cells than that of other nonreproductive females, considering their body size effect. However, the male fat body, which is not involved in yolk production, did not show consistency in polyploid cell numbers among the species studied. Moreover, highly fecund queens in foraging termites exhibit higher levels of endopolyploidy in their fat body than those with less fecundity in wood‐dwelling termites. These results suggest that endopolyploidy in the fat body of termite queens can boost their egg production, and the level of endopolyploidy in their fat body is linked to their fecundity. Our study provides a novel insight into the evolutionary relationship between endoreduplication and caste specialization in social insects.

& Johnston, 2015; Scholes, Suarez, & Paige, 2013;Scholes, Suarez, Smith, Johnston, & Paige, 2014). Scholes et al. (2013) investigated the degree of polyploidy within and among castes of four ant species with worker polymorphism and observed variation in ploidy levels among workers with different body sizes. Rangel et al. (2015) surveyed transition of polyploid levels in several body parts of worker honey bees and found age-related tissue-specific changes in endopolyploidy levels in the honey bees of different ages. However, there are very few reports on the relationship between endopolyploidy and the division of labor in reproduction.
Recently, we observed that the queens of the termite, Reticulitermes speratus Kolbe, exhibited higher endopolyploidy levels in their fat body cells in comparison with the nonreproductive females (Nozaki & Matsuura, 2016). Insect fat body is a multifunctional organ, which combines the roles of many tissues including the liver and fat tissues in vertebrates. This tissue is involved in the synthesis, storage, and secretion of lipids, proteins, and carbohydrates (Arrese & Soulages, 2010). During oogenesis, the female fat body produces abundant yolk protein precursors, namely vitellogenins (Bownes, 1986;Tufail & Takeda, 2008). The rate of vitellogenin synthesis from internal nutrients is probably one of the essential factors determining the capability of egg production.
It has long been known that in certain solitary insects, the fat body cells of sexually mature females become polyploid, presumably to promote vitellogenin synthesis (Dittmann, Kogen, & Hagedorn, 1989;Irvine & Brasch, 1981;Nair, Chen, & Wyatt, 1981). These studies suggest that the fat body endopolyploidy enhances the reproductive specialization of queens for egg production; however, other possibilities remain to be explored. For example, not vitellogenin synthesis, but simply body size differences among individuals can also affect the ploidy level in the fat body (Scholes et al., 2013). The ploidy levels of the male fat body, which is not involved in vitellogenin synthesis, should also be examined (Terrapon et al., 2014). Determining the ploidy levels in fat body cells of both the sexes of various termite species would provide further implications for the functional significance of endopolyploidy in caste specialization and queen fecundity in social insects.
In termites, queen fecundity varies greatly between lineages with different lifestyles. Wood-dwelling termites utilize a single piece of wood, both as food source and shelter (Figure 1a;Korb & Hartfelder, 2008;Korb et al., 2015), and form small colonies from several hundreds to a few thousands (reviewed in Nutting, 1969). Wood-dwelling lifestyle is considered phylogenetically basal (Korb et al., 2015), and a characteristic of damp-wood termites (Archotermopsidae), dry wood termites (Kalotermitidae), and the ancestral clade of subterranean termites (Rhinotermitidae). In these species, both queens and workers share the same developmental pathway, that is, workers are actually immature and all of them can potentially develop into reproductive individuals (Figure 1b;Korb & Hartfelder, 2008). On the other hand, the foraging termites are characterized by multiple-piece nesting ( Figure 1c) and large colony size of at most several millions (reviewed in Nutting, 1969;estimated in Evans, Lenz, & Gleeson, 1998;Evans, Lenz, & Gleeson, 1999). The workers leave the nest to forage outside for food materials at some point after colony formation (Korb et al., 2015). This lifestyle is observed in the advanced clade of Rhinotermitidae such as Coptotermes, Heterotermes, and Reticulitermes (Bourguignon et al., 2015), and in all of the higher termites (Termitidae). These species exhibit an early developmental bifurcation between the worker and reproductive termites (Figure 1d; Korb & Hartfelder, 2008), which might enhance the physiological discrepancy between them.
Queens of foraging species exhibit much higher fecundity than those of the wood-dwelling species (reviewed in Nutting, 1969;Weesner, 1969) and can unfold to expand their abdominal epicuticle of intersegmental membrane along with ovarian development (Myles, 1999). In the higher termites, it has been demonstrated that the abdomen of fully matured queens would be distended five to eight times its original size and their intersegmental membrane not only unfold but also grow continuously (Bordereau, 1982;Bordereau & Andersen, 1978). These variations in queen fecundity among different termite species are likely a reflection of the factors surrounding a particular lineage such as their lifestyle. For example, termite queens with high fecundity could be more favorable in foraging species, but probably not as much in wood-dwelling termites (Myles, 1999).
In this study, we investigated the relationship between reproduction and fat body polyploidization using three foraging species  (Table 1). First, we investigated the effect of caste (especially, queens vs. nonreproductive females) and body weight on the polyploid level in the fat body for each species. We also analyzed the endopolyploidy levels in the male fat body that is not involved in vitellogenin synthesis (Terrapon et al., 2014). Then, in order to elucidate the relationship between endopolyploidy and queen fecundity, we compared the relative ploidy levels of queens with the ploidy levels of conspecific workers between queens of the foraging and the wood-dwelling species.
F I G U R E 1 Life types of termites and developmental patterns. Wood-dwelling species nest in a single piece of wood that serves both as a shelter and food source (a) and their postembryonic development was linear (b). Foraging species are characterized by worker foraging outside of their nest and resultant multiple-pieces nesting (c). They display a bifurcated development between worker and reproductives (d) Seven mature colonies of Na. takasagoensis (A-G) were collected in the end of January 2016, from Iriomote Island, Okinawa Prefecture, Japan. This species is known to produce eggs even in the winter season (Matsuura & Yashiro, 2010). The carton nests were carefully transported to the laboratory and dissected, and termites were extracted. Termites were used for ploidy analysis immediately after extraction from the carton nest. Colonies A-E contained primary royal pairs (both king and queen); however, F and G did not. The last instar nymphs were involved in colonies A-C, and F and G, whereas not in colonies D and E. All collected royals were adult (alate-derived) reproductives, as previously reported (Hojo, Koshikawa, Matsumoto, & Miura, 2004).

| Coptotermes formosanus Shiraki (Rhinotermitidae; subterranean termite, foraging species)
In June 2013 and 2014, two C. formosanus colonies comprising newly emerged alates were collected from a coastal pine forest in Gobo City, Wakayama Prefecture, Japan and brought to our laboratory. The colonies were maintained in plastic containers at 25°C until the alates emerged and flew. After swarming, the alates were separated based on their sex, identified by the morphology of caudal sternite (Zimet & Stuart, 1982 Approximately 3 years later, each incipient colony was transferred into a plastic box containing additional wood chips and commercial culture soil (Tsuchitarou, Sumirin Nousan, Japan). In September 2018, five of the colonies were used for the ploidy analysis. All these five colonies comprised adult royals, and the queens' abdomens were expanded, and ovaries were fully developed. The number of individuals including workers and soldiers was more than 1,000 (from 1,033 to 3,147), and numerous egg piles were present in their nests; hence, we treated the royals from these colonies as fully matured reproductives.

| Reticulitermes speratus Kolbe (Rhinotermitidae; subterranean termite, foraging species)
From June to August 2018, five R. speratus colonies were collected from pine and cedar forests of Kyoto and Shiga Prefecture, Japan.
In R. speratus, egg production is seasonal (Matsuura, Kobayashi, & Yashiro, 2007). We used colonies collected during the egg-producing season. Rotten woods containing their colonies were transported to the laboratory, and the extracted termites were immediately used for ploidy analysis. Each colony contained multiple queens, which were nymph-derived neotenics, and one adult king, as reported in this species .  All sampled individuals were sexed based on their caudal sternite configuration (Zimet & Stuart, 1982) except for Na. takasagoensis. In this species, the sex of late instar nymphs and alates can be determined based on the shape of their abdominal sternites, whereas such a method is not applicable to the workers. In this study, we followed Hojo et al. (2004) and Miura, Roisin, and Matsumoto (1998), wherein major workers classified as females and minor workers as males. The soldiers were always differentiated from the male workers in mature colonies; thus, only male soldiers were available in this species (Hojo et al., 2004;Toga, Minakuchi, & Maekawa, 2017).

| Statistical analysis
To determine whether the endopolyploidy level correlated with the castes, the proportion of polyploid cells in the fat body, which was calculated by the nuclei count with C-value of 2C as diploid, and 4C, 8C and higher as polyploid, were compared among different castes within the species. Proportions of polyploid cells of each caste were analyzed using generalized linear mixed-effect models (GLMM) with binomial errors and a logit-link function, followed by Tukey's HSD post hoc test.
Simultaneously, to examine the effect of body size on endopolyploidy, we included the body weight of each individual into the model. In this analysis, the castes and body weight were treated as a fixed effect and the original colony was included as a random effect. To avoid problems of nonconvergence, we included the optimizer "bobyqa." We analyzed the males and females separately, because there must be functional differences between the sexes. Sex-specific difference in gene expression pattern in the fat body has in fact been described previously .

| Comparison of the proportion of polyploid cells among castes within species
Our flow cytometric analysis of female termites revealed that in all the six species studied, the queen fat body comprised a higher num-

ber of polyploid cells (4C and 8C) than diploid cells (2C), whereas
Relative polyploid level = log P q × 1 − P w P w × 1 − P q F I G U R E 3 Examples of nuclear DNA content analysis by flow cytometry. The first peak corresponds to the distribution of 2C-DNA nuclei, whereas the second and the third correspond to the distribution of 4C-and 8C-DNA nuclei, respectively. The ploidy was determined by analysis of sperm cells (haploid; 1C) of conspecific kings that of other nonreproductive females contained roughly the same number of diploid and polyploid cells (Figures 3 and 4). Notably, only in R. speratus, the queen fat body contained a few cells with 16C-DNA. In Na. takasagoensis, both castes and body weight had significant effects on the proportion of polyploid cells in the fat body (GLMM with a Wald chi-square test, caste: χ 2 = 27.021, df = 2, p < .001, body weight: χ 2 = 4.433, df = 1, p = .035). In C. formosanus, whereas the effect of castes was significant (GLMM with a Wald chi-square test, χ 2 = 71.8927, df = 2, p < .001), that of body weight was not significant (χ 2 = 2.7059, df = 1, p = .100). In R. speratus, the proportion of polyploid cells significantly differed among the castes (GLMM with a Wald chi-square test, χ 2 = 197.800, df = 2, p < .001), yet body weight did not have any significant effect (χ 2 = 0.714, df = 1, p = .398). In Ne. sugioi, both castes and body weight had significant effects on the polyploid ratio in the fat body of female individuals (GLMM with a Wald chi-square test, caste: χ 2 = 27.674, df = 2, p < .001, body weight: χ 2 = 23.508, df = 1, p < .001). In I. schwartzi, we found significant effects of both castes and body weight on the proportions of polyploid cells in the fat body (GLMM with a Wald chi-square test, castes: χ 2 = 263.827, df = 3, p < .001, body weight: χ 2 = 60.772, df = 1, p < .001). In Z. nevadensis, while body weight had a significant effect on the proportion of polyploid cells in their fat body (GLMM with a Wald chi-square test, χ 2 = 15.551, df = 1, p < .001), castes also had a significant effect (χ 2 = 216.074, df = 2, p < .001). In all species examined, queens presented significantly higher proportions of polyploid cells than that of other nonreproductives (Tukey's HSD, p < .05, Figure 4).

| Comparison of the relative polyploid level of queen in the fat body between foraging and wooddwelling termites
We found significant differences in the relative polyploid levels of queen fat body between the species (Figure 6). Queens of foraging species exhibited significantly higher polyploid levels than that exhibited by wood-dwelling species (LMM with a Wald chi-square test, χ 2 = 15.9351, df = 1, p < .001). On the other hand, differences in body weight between queens and workers did not have significant effect on polyploidy (χ 2 = 0.012, df = 1, p = .915).

| D ISCUSS I ON
We found that endoreduplication in the fat body is tightly linked to the reproductive division of labor and the queen fecundity in  Figure 6). These findings support the hypothesis that endopolyploidy in the queen fat body is exploited to boost their egg production as an adaptive strategy, whereas the direct and more detailed relationship between egg production and fat body endoreduplication needs to be examined.
Because age-related endopolyploidy has been reported in worker honeybees (Rangel et al., 2015), it is probably reasonable to expect that the fat body of long-lived individuals will have more endopolyploid cells than others in termites. In fact, queens in eusocial insects including termites exhibit long lives; their lifespan is more than 10-fold longer than that of workers and soldiers (Jemielity et al., 2005;Keller, 1998;Thorne, Breisch, & Haverty, 2002). In this study, however, the rate of endopolyploid cells in queen fat body was higher than or equal to that of termite kings, which have extraordinary longevity equal to or greater than the queens (Figures 4 and 5;Boomsma, Baer, & Heinze, 2005). Therefore, age cannot explain the high-level of endopolyploidy we observed in the termite queen fat body in our study. However, our data cannot exclude the effect of age on the fat body endoreduplication.
Polyploidy plays pivotal roles in the regulation of gene expression, cell size, and differentiation (Edgar et al., 2014;Lee et al., 2010;Neiman et al., 2015). It is possible that the fat body endoreduplication observed in this study is related to the cytological changes as described in previous studies (Costa-Leonardo et al., 2013;Han & Bordereau, 1982a, 1982bŠobotník, Weyda, Hanus, Cvačka, & Nebesářová, 2006).  Bordereau, 1982aBordereau, , 1982b Species termite Calcaritermes temnocephalus Silvestri will also provide a unique opportunity to test the hypothesis of endopolyploidy in the fat body and queen fecundity, because queens in the species are highly fecund and exhibit the morphological specialization, which is an exception in Kalotermitidae (Scheffrahn, 2011). The ploidy patterns of these species allow further progress in understanding the evolutionary relationship of extraordinary high fecundity of termite queens and fat body endoreduplication.
In this study, queen types were different among species; queens in Na. takasagoensis, C. formosanus, Ne. sugioi, and Z. nevadensis were all adults; however, all queens were neotenics in R. speratus. Both neotenic and adult queens were present in I. schwartzi. Several developmental and anatomical differences exist between these two queen types (Myles, 1999;Weesner, 1969); nevertheless, in our study, the fat body of all queens exhibited higher levels of endopolyploidy than conspecific nonreproductives, regardless of the queen types. This suggests that the queen types are unlikely to have much of an effect on the ploidy levels in their fat body, although the effect of queen types on endopolyploidy needs to be further examined in a species wherein both adult and neotenic queens are functional.
In conclusion, our study has provided a novel insight into the evolutionary linkage between endoreduplication and caste specialization in social insects. Based on the patterns observed in the study, we suggest the possibility that endopolyploidy in the queen fat body may enhance egg production, mechanisms of which should be assessed in future studies. In future, it will be worthwhile to determine whether polyploid fat body cells show increased expression of genes involved in vitellogenesis. In addition, it will be interesting to determine the cell types of termites that exhibit polyploidy and to explore the proximate mechanisms promoting endoreduplication in the fat body.

ACK N OWLED G M ENT
We thank T. Yashiro 18H05268).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

AUTH O R CO NTR I B UTI O N
Both authors designed and performed the research. T.N. analyzed the data and wrote the first draft. Both authors then contributed substantially to revisions and approved the final version.