More than DNA methylation: does pleiotropy drive the complex pattern of evolution of Dnmt1?

DNA methylation is an important chromatin modification that can stably alter gene expression in cells and maintain genome integrity in plants and vertebrates. The function of DNA methylation outside of these well-studied systems, however, is unclear. Insects, in particular, represent an understudied group. Variation in the level of DNA methylation and gains and losses in the maintenance methyltransferase, DNMT1, across the insect tree of life suggests that there is much we don’t understand about DMNT1 function and evolution. One constant across the studies examining patterns of Dnmt1 expression in insects is that expression is consistently high in reproductive tissues compared to somatic tissue. The explanation for this has been that DNMT1 is required in tissues that have high levels of cell division. Our previous study found that downregulation of Dnmt1 expression in the milkweed bug Oncopeltus fasciatus results in the expected reduction of DNA methylation, no global changes in gene expression reflecting changes in DNA methylation, and the loss of the ability to produce viable oocytes. Here, we show that females treated with ds-Dnmt1 RNA during larval development have a more extreme phenotype; they lack oocytes entirely but develop a normal somatic ovary. Our results indicate a specific role for DNMT1 in the formation of gametes and are consistent with data from other systems, including Tribolium castaneum, a species does not have DNA methylation. We propose that DNMT1 has multiple functional roles in addition to methylating DNA, which explains its complex patterns of evolution, and suggests that previous inferences of causation from associations are premature.

Oncopeltus fasciatus, an evolution and development model system with molecular genetic resources, 49 a methylated genome and the full complement of DNA methyltransferases. Using parental RNAi, we 50 have previously shown that knockdown of Dnmt1 in sexually mature females results in embryo arrest 51 (Bewick et al. 2019). Eventually, females injected with ds-Dnmt1 stopped laying eggs and their 52 ovarian structure was significantly disrupted. We further found that reduction of DNA methylation 53 was not directly associated with changes to gene expression patterns within the ovary and that DNA 54 methylation patterns in the somatic tissues, the gut and muscles, were unaffected by the knockdown. 55 Our parental RNAi experiments raise the question as to why the morphological defects of Dnmt1 females injected with ds-Boule lay eggs that don't develop, but after a few clutches, females stop 105 producing eggs. Boule is a widely conserved gene required for reproduction across the bilateral 106 animals (Shah et al. 2010). Mutations in Boule cause arrest in meiosis prior to metaphase in 107 Drosophila melanogaster males and Caenorhabditis elegans females (Karashima et al. 2000). Thus, 108 Boule provided an excellent comparison for evaluating the hypothesis that Dnmt1 is required for 109 oogenesis, and perhaps the transition from oogonia to oocytes in O. fasciatus. We examined this 110 hypothesis by testing for the following predictions arising from our overall hypothesis. First, Dnmt1 111 should be most highly expressed when and where oogenesis is occurring. Second, downregulating 112 Dnmt1 expression during a critical developmental stage for oogenesis would result in the loss of 113 oocyte production. Third, downregulating Dnmt1 expression will affect reproductive function 114 without affecting somatic function and lifespan. 115

116
Animal husbandry 117 To produce nymphs of known age and social history, O. fasciatus adults from our mass colony were 118 randomly mated, and resulting eggs were removed and stored in individual plastic containers. Upon 119 hatching, nymphs were transferred into 2-cup Rubbermaid containers supplied with organic, raw 120 sunflower seeds and deionized water. Nymph colonies were housed under 12 hr: 12 Table 1. Primer sequences used to generate RNAi for injection and to quantify expression levels using quantitative Real Time PCR Primers to produce PCR products for transcription reaction to produce ds-RNA: lowercase letters indicate T7 promoter sequences required for transcription, uppercase letters indicate gene-specific sequences.

Gene
Sense primer Anti-sense primer To examine expression levels of Dnmt1 across development, groups of nymphs were staged and flash 152 frozen in liquid nitrogen, and stored at -80°C. The earliest stages of nymphs are difficult to sex 153 accurately, so the 2 nd and 3 rd instar nymphs were not separated by sex. By the 4 th instar stage of 154 development, females can be identified, so only female 4 th and 5 th instar nymphs were collected. To 155 ensure that there was a sufficient amount of tissue for RNA isolation, smaller nymphs were pooled 156 for each sample. A sample of 2 nd , 3 rd , 4 th and 5 th instar nymphs contained 10, 5, 4, and 3 individuals, 157 respectively. The other two developmental stages we collected were newly emerged females, 158 collected within 24 hours of adult emergence, and sexually mature females, collected 7-10 days post-159 adult emergence. All adult females were housed in a container without males to ensure that they had 160 not mated prior to RNA isolation. A sample of adults consisted of a single individual. Total RNA 161 was extracted and cDNA was synthesized as described above.
Expression levels of Dnmt1, Boule and Vasa were quantified by quantitative real-time PCR 163 (qRT-PCR). Boule and Vasa primers were designed using the O. fasciatus genome as a reference 164 (Panfilio et al. 2019) and actin and GAPDH were used as endogenous reference genes (Table 1). 165 Expression levels were quantified by qRT-PCR as described above. We used ANOVA followed by 166 Tukey-Kramer HSD to compare all pairs in JMP Pro 14.1 to determine significant differences in 167 levels of gene expression among the different stages. 168 To examine for tissue specific expression of Dnmt1, we dissected adult, virgin females 7-10 169 days post adult emergence. We flash froze ovaries, gut, and thorax (muscle) from individual females. 170 Total RNA was extracted and complementary DNA (cDNA) was synthesized as described above. 171 Expression of Dnmt1 was quantified and analyzed as described for the development series. 172   As predicted, injecting earlier in development resulted in reduced DNA methylation of all 230 tissues due to an increased number of mitotic division cycles within the somatic tissues between 231 treatment and sampling. In both somatic tissues we tested, gut and muscle, and in the ovary, 232 methylated CpG levels went from around 12.5% in controls to around 5% in Dnmt1 knockdowns 233   a germarium containing trophocytes, oogonia, primary oocytes, and prefollicular tissue, the 258 vitellarium where oocytes mature, and the pedicel that connects the germarium to the oviduct and in 259 which the developing oocytes mature ( Figure 5A; see also Bonhag and Wick 1953). Low 260 magnification images revealed that the somatic cells of the ovary, including the terminal filament, the 261 Figure 5. The structure of the somatic ovary, including the trophocytes, which develop from oogonia via mitotic division, was relatively normal in both Dnmt1 and Boule knockdown females. (A) Overview of the tip of an ovariole from a control female with the significant regions marked. The terminal tip (asterix) anchors the ovariole to the body wall. The majority of the germarium contains trophocytes that surround a trophic core. At the base of the germarium there is a group of cells with smaller nuclei. These cells are oogonia and primary ooctyes and pre-follicular cells that will envelope the developing oocyte as it moves into the vitellarium. During early stages of oocyte development, ooctyes remain connected to the trophocytes through trophic cords (arrowheads). Low magnification images (2X) of ovaries from control (B), Dnmt1 knockdown (C), and Boule knockdown females stained with DAPI show that the germarium and the pedicel cells were present in the two knockdowns. However, no oocytes or their associated follicle cells ever entered the vitellarium.  Figure 6D). Although the appearance of the control and ds-Boule germarium tissue was 281 highly repeatable, there was significant variation in the amount of both cell death, evidenced by 282 highly condensed and fragmented nuclei, and tissue disruption in the germarium of ds-Dnmt1 treated 283 females (Supplementary materials Figure S1). Despite the specific expression of the phenotype, all 284 The other cell type located in this region is the pre-follicular cells. To determine if these cells 289 were also affected by Dnmt1 knockdown, we examined cell division patterns by labeling the 290 ovarioles with anti-phosphohistone H3 antibody, which labels dividing cells. The pre-follicular cells 291 divide to provide follicle cells that will envelope the developing oocyte as it moved down through the 292 ovariole. In our control females, division of the pre-follicle cells was apparent at the junction between 293 the trophocytes and developing ovaries ( Figure 7A). In the Boule knockdown females, the pre-294 follicular cells were particularly clear as they were not obscured by the developing ooctyes (  Within the control ovary, the pre-follicular cells are present in the tissues that surrounded the ooctyes at the early stages of development and the majority of the nuclei that stained positive for pHH3 were present within this region of the germarium. In the Boule knockdown ovaries (C), the prefollicular cells were very apparent as they were not obscured by any developing ooctyes. While the structure of the germarium in the Dnmt1 knockdown females made it difficult to identify particular cell types, there was evidence that prefollicular cells remained in these germarium, as there are nuclei of the right size at the base end of the germarium and these were frequently stained for cell division (yellow arrows), although they were somewhat obscured by the condensed nuclei within this region of the germarium. All images were stained with DAPI and antiphosphohistone H3 antibody and imaged at 20X with 0.7 optical zoom. Dnmt1 knockdown does not significantly affect lifespan. We predicted that if Dnmt1 plays a role 300 in somatic function, we would see reduction in lifespan in Dnmt1 downregulated females due to the 301 reduction of Dnmt1 and DNA methylation observed in the somatic tissues as indicated by our results 302 with gut and muscle tissue. To control for any impact on lifespan due to reduced reproductive effort, 303 we included ds-Boule treated females in our lifespan analysis. We found no statistically significant 304 differences in lifespan among the three treatments (Figure 7 gut and muscle. However, rates of cell division in adult insects tend to be low and so the somatic 344 tissues assayed did not lose DNA methylation. Thus, the lack of phenotype could have been due to 345 the experimental design in which the timing of knockdown relative to assay was insufficient to allow 346 any effect to be manifest. This study was specifically designed to address this issue; we treated 347 nymphs to allow sufficient rounds of cell division to enable reduction of DNA methylation in both 348 reproductive tissues and somatic tissues. We did indeed observe that the levels of DNA methylation 349 were reduced over 2-fold in all the tissues tested. However, despite this over 2-fold reduction in DNA Dnmt1 gene in its genome (Bewick et al. 2016). When Dnmt1 is knocked down during pupal methylation can be lost without a phenotype and that DNMT1 plays a functional role in an insect that 358 lacks DNA methylation make it likely that the functional role of DNMT1 is not always mediated 359 through DNA methylation. 360

Quantification of DNA methylation
Our results establish that Dnmt1 has a specific function in germ cells. The comparison with 361 the Boule knockdown females indicates that Dnmt1 is required for maintenance of oogonia or 362 primary oocytes. In the Boule knockdown females, the overall structure of the germarium is 363 preserved, although no ooctyes ever form. Given that Boule is required for entry into meiosis 364 (Eberhart et al. 1996), one likely interpretation of our data is that healthy oogonia are waiting for the 365 signal that it is time to divide meiotically to form primary ooctyes, but that signal never comes. The 366 tissue disruption and evidence of cell death in the Dnmt1 knockdown females, however, indicate that 367 Dnmt1 is required to maintain oogonia and/or primary ooctyes and that if Dnmt1 function is missing, 368 these cells are not viable. While we are not able to differentiate among oogonia and primary ooctyes 369 in our samples, the fact that the trophic tissue develops normally provides a clue as to where Dnmt1 370 is required. During the first, second, and third instar stages of development, the germ cells within the 371 somatic ovary increase in number through mitosis (Wick and Bonhag). In the fourth instar stage, 372 oogonia either divide mitotically to form trophocytes or divide meiotically to form ooctyes. 373 We suggest that Dnmt1 may be required for proper progression through meiosis or for 374 stability of the germ cells. In Dnmt1 knockdown females, oogonia that have not been able to advance 375 properly through meiosis will die or be targeted for destruction. Alternatively, oogonia that divide to 376 form primary ooctyes may require Dnmt1 to maintain genome integrity. In the absence of Dnmt1 377 newly born oocytes may degenerate. We currently are not able to determine if this is due to the loss 378 of methylation or a pleiotropic function of DNMT1. While it is clear that DNA methyltransferases 379 are associated with the reproductive cells in insects, the role of DNA methylation in germ cell effect on somatic function. It is likely, however, that there is less flexibility in the pathways required 405 for maintaining a stable germline (Maklakov and Immler 2016), which is expected to be strongly 406 related to fitness and therefore under strong stabilizing selection. If Dnmt1 has a pleiotropic function 407 in germline stability, independent of DNA methylation, it suggests that alternative or multiple 408 pathways for stabilization of the germline genome must exist. Those pathways are yet to be 409 discovered but examining insects that lack Dnmt1 (Bewick et al. 2018) provides a potential model. 410

411
We have demonstrated that Dnmt1 functions to maintain a healthy germline in female O. fasciatus. 412 Dnmt1 knockdown, while resulting in a reduction in DNA methylation across both the germ and 413 soma, only has a morphological phenotype in reproductive cells. We propose that Dnmt1 has a 414 pleiotropic function independent of DNA methylation in the germ cells and could be required to 415 maintain the genome integrity of germ cells or be required to progress through meiosis. These results 416 raise a number of questions that need to be addressed in future experiments. 417

418
The datasets analyzed for this study will be made available by the authors through the publicly 419 available Dryad Digital Repository (https://datadryad.org/stash/) 420

430
The authors declare that the research was conducted in the absence of any commercial or financial 431 relationships that could be construed as a potential conflict of interest. 432