Matrotrophic viviparity constrains microbiome acquisition during gestation in a live‐bearing cockroach, Diploptera punctata

Abstract The vertical transmission of microbes from mother to offspring is critical to the survival, development, and health of animals. Invertebrate systems offer unique opportunities to conduct studies on microbiome‐development‐reproduction dynamics since reproductive modes ranging from oviparity to multiple types of viviparity are found in these animals. One such invertebrate is the live‐bearing cockroach, Diploptera punctata. Females carry embryos in their brood sac, which acts as the functional equivalent of the uterus and placenta. In our study, 16S rRNA sequencing was used to characterize maternal and embryonic microbiomes as well as the development of the whole‐body microbiome across nymphal development. We identified 50 phyla and 121 classes overall and found that mothers and their developing embryos had significantly different microbial communities. Of particular interest is the notable lack of diversity in the embryonic microbiome, which is comprised exclusively of Blattabacteria, indicating microbial transmission of only this symbiont during gestation. Our analysis of postnatal development reveals that significant amounts of non‐Blattabacteria species are not able to colonize newborn D. punctata until melanization, after which the microbial community rapidly and dynamically diversifies. While the role of these microbes during development has not been characterized, Blattabacteria must serve a critical role providing specific micronutrients lacking in milk secretions to the embryos during gestation. This research provides insight into the microbiome development, specifically with relation to viviparity, provisioning of milk‐like secretions, and mother–offspring interactions during pregnancy.

F I G U R E 1 Diploptera punctata reproduce by matrotrophic viviparity, this female D. punctata is giving birth, surrounded by her newly born nymphs Most, but not all, strains of Blattabacteria have an incomplete biosynthetic pathway for methionine Kambhampati, Alleman, & Park, 2013;López-Sánchez et al., 2008Patiño-Navarrete, Moya, Latorre, & Peretó, 2013;Sabree, Kambhampati, & Moran, 2009;Tokuda et al., 2013). This leads us to the question, do D. punctata embryos inherit only Blattabacteria, capable of methionine biosynthesis, from their mothers, or does the extended association between mother and offspring allow colonization of the embryonic microbiome by additional bacteria?
To address this question, this study determined the microbiome of D. punctata throughout development, characterizing the microbial communities inhabiting female D. punctata and their offspring across development using 16S rRNA gene sequencing. The information generated by this study will provide the first step in developing D. punctata as a model system to elucidate how intrauterine development and the prenatal microbiome affect later acquisition of microbial endosymbionts. Developing a new model system understanding microbial shifts during invertebrate matrotrophic viviparity will widen the evolutionary lens through which we view reproduction and the microbiome in viviparous animals.

| Animals
Colonies reared at the University of Cincinnati (UC) Department of Biological Sciences (Cincinnati, OH) were housed in a climate-controlled facility. Ambient temperature was held between 24-28°C, and relative humidity (RH) was held between 70%-80%. A 12:12-hr lightdark photoperiod was maintained for the duration of the experiment.
Animals were provided water and fed Old Roy Complete Nutrition brand dog food (Mars, Inc.) ad libitum. A second group of D. punctata were obtained from The Ohio State University (OSU) Biological Sciences Greenhouse (Columbus, OH) insect collection where they were reared in similar conditions with the exception of being fed a diet of Tetramin fish food (Spectrum Brands Pet). This second group was collected randomly from the OSU colony and brought to the UC laboratory, where they were housed separately from the UC colony under identical conditions and provided the same food and water sources as the UC colony for 1 week, when sacrificed for sample collection.

| Sample collection
Visibly pregnant females were selected from the colony for use in mother-embryo comparisons. Females were surface sterilized by rinsing for 1 min in each of the following solutions: 70% ethanol and 2% sodium hypochlorite. This was followed by four rinses in sterile phosphate-buffered saline (PBS; 81 mM Na 2 HPO 4 , 19 mM NaH 2 PO 4 , 150 mM NaCl, pH 7.4). Embryo broods were then dissected from the brood sac in sterile PBS by making two small incisions at the opening of the brood sac, one on each side, and removed using ethanol sterilized forceps. To determine the developmental stage of the embryos, a single embryo from the center of each brood was measured on a bleach sterilized ruler and designated as prelactation, early lactation, or late lactation based on its length (Table 1; Stay & Coop, 1973). Entire broods of embryos and individual mothers were then placed into separate 1.5-mL centrifuge tubes with silica beads and stored at −80°C until processing. While mother-embryo pairs were collected for all three trimesters, only late lactation pairs were utilized in this study. Nine mother-embryo pairs were collected from the UC colony for analysis and 12 from the OSU colony.
To characterize the postnatal development of the microbiome, visibly pregnant females were again selected from the colony and housed in individual containers with food and water ad libitum and monitored for active birthing. Nymphs were collected as neonates (identified by lack of cuticular melanization) or first instars (identified by melanization within 12 hr of birth). Second-, third-, and fourth-instar nymphs were sampled and identified by size and the presence of molts in the living quarters. Postnatal samples were collected only from the UC colonies. Upon collection, nymphs were surface sterilized using the methods described above and then stored in 1.5-ml centrifuge tubes with silica beads at −80°C until processing. Five neonates, seven first instars, nine second instars, nine third instars, and six fourth instars were utilized in this analysis.

| Genomic DNA preparation
Samples were homogenized in 1 µl of sterile 1× PBS, and DNA was extracted using a QIAGEN DNeasy Blood and Tissue Kit (Qiagen).
The homogenate (200 µl) was incubated with proteinase K (Qiagen) over night before continuing the provided protocol. DNA concentration and quality were measured using a NanoDrop 2000. All samples were diluted to 20 ng/µl for sequencing.

| 16S rRNA sequencing and bioinformatic analyses
The V4 hypervariable region of the bacterial 16S rRNA gene was PCR amplified using the 515f (GTGYCAGCMGCCGCGGTAA) and 806r (GGACTACNVGGGTWTCTAAT) universal primers ( rRNA gene sequences were removed using the classify.seqs and remove.lineage commands. Sequences were clustered using the cluster. split command at the taxonomic level 4, representing order. All further analyses were conducted using operational taxonomic unit (OTU) assignments generated in the above steps. Rarefaction curves were generated using the rarefaction.single and the number of observed OTUs (sobs), demonstrating adequate sequencing depth (Table S1). Alpha diversity was assessed using the inverse Simpson, and Shannon diversity metrics. NMDS and PCOA analyses were conducted using mothur.
Community composition was manually assessed for visualization at taxonomic level 5, representing bacterial families. Linear discriminant analysis effect size (LEfSe) as implemented in mothur was utilized to identify stage-specific OTUs across development (Segata et al., 2011); a p-value cutoff of 0.01 was utilized. In addition to mothur, we performed a second analysis of our data for validation purposes utilizing QIIME (v. When relative abundances calculated at the class level by both methods were compared, they were found to be significantly correlated ( Figure S1); consequently, results from mothur were reported.
Additional results from the QIIME analysis can be found in Data S1 and Data S2. test (Dinno, 2017), ggplot2 (Wickham, 2016), reshape2 (Wickham, 2007), RColorBrewer (Neuwirth, 2014), Rmisc (Hope, 2013), and wesanderson (Ram & Wickham, 2018).    Figure   S2). Additionally, it should be noted that these low abundance taxa show no consistency in representation across embryo samples with varying numbers of reads and OTUs (Tables S2 and S3). These findings were corroborated by secondary analyses completed using the Nephele implementation of QIIME, despite inherent differences in computational methods (Data S1). Embryos of both UC and OSU colonies did not differ significantly in diversity, evenness, and species richness ( Figure 3). However, microbial communities of embryos were less diverse and less so than mothers across both colonies  Table S5).

| Maternal and embryonic microbiomes
While the transmission of the cockroach-specific endosymbiont Blattabacteria is known to occur during oogenesis (Sacchi et al., 1996), surface sterilization of oothecae, and hatching into a sterile environment results in a microbiome exclusively composed of Blattabacteria, indicating any other bacteria must be acquired from food or feces (Pietri, Tiffany, & Liang, 2018). Such is the case in the intergenerational transfer of microbiota via proctodeal trophallaxis in Cryptocercus punctulatus and Mastotermes darwiniensis (McMahan, 1969 a trend that holds true at the order level as well (Table S7). At the family level, Blattabacteriaceae (42.877%) was again the most prominent taxon followed by unclassified bacteria (11.824%), unclassified Bacteroidetes (5.515%), and Porphyromonadaceae (4.506%; Figure 5, Table S7).
Postmelanization first instars, however, have a more diverse microbial      Figure 6). Neonates also did not differ from first instars but showed significant differences in both diversity metrics compared to second, third, and fourth instars as well as adult females. Second, third, and fourth instars, however, did not differ from each other or mothers in any diversity measure ( Figure 6). While AMOVA and HOMOVA analyses revealed slightly different relationships between the samples (Table S9), the analyses consistently showed that embryos and neonates differed from the other juvenile stages and adult females.
These results further support our hypothesis that D. punctata acquire microbial endosymbionts (outside of Blattabacteria), not through direct maternal transfer during gestation but in the days and weeks after birth, primarily during and after initial melanization during the first nymphal instar.
Investigations of developmental acquisition of the cockroach microbiome are rare; however, one study characterized the succession of the microbiota in the oviparous German cockroach, Blattella germanica (Carrasco et al., 2014). Contents of surface-sterilized oothecae contain exclusively Blattabacteria and whole bodies of first-instar nymphs that hatched from unsterilized oothecae contain predominantly Blattabacteria, but have begun to acquire other gut symbionts (Carrasco et al., 2014). Despite the difference in reproductive mode, we found similar results in the intrauterine developing embryos and neonatal D. punctata.
One previous study has attempted to characterize the microbiome of D. punctata mothers and embryos, concluding that there are significant amounts of non-Blattabacteria microbes in embryos (Ayayee, Keeney, Sabree, & Muñoz-Garcia, 2017). In direct contrast, our embryo samples from two independent colonies, including the colony used in the previous study, produced sequencing reads that were 99.5% assigned to Blattabacteriaceae. Two taxa identified to be significantly enriched in the embryonic microbiome by this previous study were Halomonadaceae and Shewanellaceae (Ayayee et al., 2017), neither of which were present in our maternal, embryo, or postnatal development samples. While our analysis using mothur did identify non-Blattabacteria sequences in embryonic samples, the extremely low abundances (<0.5% of total raw reads combined) suggest they are sequencing artifacts or misidentified and are not likely critical for embryos during gestation. This is further supported by our secondary analysis using the Nephele implementation of QIIME (Table S8, Data S1 and Data S2), which identified no taxa representing more than 0.2% of the embryonic community other than Blattabacteria. The fact that there is no consistency among low abundant taxa among embryo sample supports that bacteria, other than Blattabacteria, are not likely critical for the intrauterine stages. Because of our robust sampling method, including two separately housed colonies of D. punctata from separate institutional origins and use of two independent pipelines for analysis, we conclude that no bacterial transmission occurs after oogenesis during intrauterine development in D. punctata. Thus, Blattabacteria is the only bacterial component of the microbiome during intrauterine development. This is further supported by the lack of additional bacterial components in first-instar nymphs collected immediately after birth (=neonate). While we cannot eliminate rearing differences, our study indicates that other bacteria, beyond Blattabacteria, are not required for D. punctata development.
After determining that there was no significant gestational transmission of endosymbionts, we sought to characterize the microbial community across nymphal development. D. punctata juveniles have a minimum of three nymphal instars with females molting an additional time to a fourth-instar stage. Newborn, unmelanized first-instar nymphs did not differ in bacterial community from intrauterine developing embryos suggesting that significant bacterial transmission does not occur during the birthing process, unlike humans. However, by the time first instars fully develop a hardened cuticle they have developed a more diverse microbial community where Blattabacteria represents only 35% of the OTUs. This substantial increase is likely the results of food and water consumption that occurs following melanization. Across the remaining instars, the community continues to become more diverse; however, the changes become much less dramatic after the second-instar stage. These findings are again consistent with a previous study investigating the juvenile microbiome of B. germanica as well as in other egg-laying organisms such as burying beetle Nicrophorus vespilloides (Carrasco et al., 2014;Wang & Rozen, 2017). Consequently, we conclude that the microbial community is largely acquired during the first-and second-instar stages, likely from their environment where they cohabitate with both adult and other juvenile cockroaches, after they have started to feed and drink. There are continuously changes to the microbiome throughout the life of the animal, but these are minor compared to the initial acquisition in early developmental stages.
Consistently, these experiments find that animals unable to acquire microbes from their environment or mothers face severe disadvantages, often failing to progress from one instar to the next, unable to molt to adulthood or undergo pupation, or dying. One example of this is the inability of axenic mosquito larvae to reach adulthood (Coon et al., 2016(Coon et al., , 2014. In the dung beetle Onthophagus gazella, removal of a maternally provided fecal secretion, known as the pedestal, significantly reduces bacterial load in larvae hatched from surface-sterilized eggs (Schwab et al., 2016). While preventing microbiome acquisition in O. gazelle larvae does not result in mortality as in mosquitoes, it is associated with reduced larval mass, increased time to adulthood, smaller adult body size, and impaired dehydration tolerance (Schwab et al., 2016). In tsetse flies, Wigglesworthia glossinidia transmission via milk gland secretions is not only essential for B vitamin provisioning, but also immune function by influencing the expression of a specific odorant binding-protein (obp) in the larvae (Benoit et al., 2017;Weiss et al., 2011). Targeted elimination of this symbiont or the associated obp decreased the population of phagocytic hemocytes and reduced melanization ability (Benoit et al., 2017;Weiss et al., 2011). Symbiont community composition has also been implicated in insecticide resistance in the German cockroach (Pietri et al., 2018). Elimination of all bacteria from the cockroach except for Blattabacteria throughout development suggests that insecticide resistance are due to changes in non-Blattabacteria bacteria which are acquired after hatching (Pietri et al., 2018). These studies underscore the importance of developing a diverse and robust microbial community during early nymphal development, which we have found primarily occurs during the first instar of D. punctata.
The embryonic microbiome comprised exclusively of Blattabacteria is of interest relative to the intrauterine development of D. punctata embryos, as the milk-like secretion provided by mothers as the sole form of nutrition during development is largely devoid of two essential amino acids, methionine and tryptophan (Stay & Coop, 1974;Williford et al., 2004). Consequently, it has been suggested that these amino acids are acquired from bacterial endosymbionts (Williford et al., 2004). Bacterial symbionts commonly serve to supplement nutrients that may be lacking in the diet (Bermingham & Wilkinson, 2009;Douglas, 2017;Engel & Moran, 2013;Funkhouser & Bordenstein, 2013;Michalik, Szklarzewicz, Jankowska, & Wieczorek, 2014;Michalkova et al., 2014). Viviparous insects, such as tsetse flies, take advantage of endosymbionts to fill such nutritional gaps during development, mostly through the provisioning of B vitamins (Douglas, 2017;Snyder, Mclain, & Rio, 2012;Snyder & Rio, 2015;Wang et al., 2013). However, while Wolbachia is transmitted through the germ line before nutrient provisioning (Wang et al., 2013), other symbionts in these flies, such as Wigglesworthia and Sodalis, have been shown to be transmitted from mother to offspring during their extended gestation period (Denlinger & Ma, 1975;Douglas, 2017;Snyder et al., 2012;Snyder & Rio, 2015;Wang et al., 2013). The exclusively Blattabacterial composition of the microbiome in embryos suggests that this symbiont must be the source of these essential nutrients. However, previous studies characterizing the genome of Blattabacteria inhabiting other species of cockroaches have shown that only the strain belonging to the German cockroach (Blattella germanica) possesses the capability to synthesize methionine, one of the amino acids lacking in D. punctata milk, in any capacity Kambhampati et al., 2013;López-Sánchez et al., 2008Neef et al., 2011;Patiño-Navarrete et al., 2013;Sabree et al., 2012Sabree et al., , 2009Tokuda et al., 2013). Consequently, further investigation of this symbiotic relationship is required to understand the role of Blattabacteria during intrauterine development. Sequencing the genome of the D. punctata strain of Blattabacteria may reveal the presence of biosynthetic pathways that can provide amino acids required for prenatal development.
In conclusion, we provide a comprehensive survey of the microbial communities of mothers and their developing embryos along with succession of the microbiome community across postnatal development in D. punctata. This study provides evidence that, unlike other viviparous insects, there is no transmission of bacteria from mothers to offspring during their 63+ day pregnancy. Surprisingly, we also found no evidence that there is significant bacterial colonization of D. punctata during birth or within the few hours immediately following birth. Rather, a majority of the microbiome components are acquired, likely from their environment, throughout the full duration of the first-instar and melanization period. Further investigation will be required to further elucidate the specific mechanisms underlying nutrient provisioning by Blattabacteria during embryonic development in D. punctata, as well as the role of the microbiome during nymphal development.

DATA AVA I L A B I L I T Y S TAT E M E N T
Sequence data have been added to the NCBI Sequence Read Archive (SRA) database (PRJNA522760), additional output from analyses using the Nephele implementation of QIIME can be found in Data S1 and Data S2.