Diel patterning in the bacterial community associated with the sea anemone Nematostella vectensis

Abstract Microbes can play an important role in the physiology of animals by providing essential nutrients, inducing immune pathways, and influencing the specific species that compose the microbiome through competitive or facilitatory interactions. The community of microbes associated with animals can be dynamic depending on the local environment, and factors that influence the composition of the microbiome are essential to our understanding of how microbes may influence the biology of their animal hosts. Regularly repeated changes in the environment, such as diel lighting, can result in two different organismal responses: a direct response to the presence and absence of exogenous light and endogenous rhythms resulting from a molecular circadian clock, both of which can influence the associated microbiota. Here, we report how diel lighting and a potential circadian clock impacts the diversity and relative abundance of bacteria in the model cnidarian Nematostella vectensis using an amplicon‐based sequencing approach. Comparisons of bacterial communities associated with anemones cultured in constant darkness and in light:dark conditions revealed that individuals entrained in the dark had a more diverse microbiota. Overall community composition showed little variation over a 24‐hr period in either treatment; however, abundances of individual bacterial OTUs showed significant cycling in each treatment. A comparative analysis of genes involved in the innate immune system of cnidarians showed differential expression between lighting conditions in N. vectensis, with significant up‐regulation during long‐term darkness for a subset of genes. Together, our studies support a hypothesis that the bacterial community associated with this species is relatively stable under diel light conditions when compared with static conditions and that particular bacterial members may have time‐dependent abundance that coincides with the diel photoperiod in an otherwise stable community.


| INTRODUC TI ON
Animals and other eukaryotes associate with diverse microbial communities that are known to have distinct and sometimes essential roles in the development, physiology, and life history of various species (Fraune & Bosch, 2010;Kohl & Dearing, 2012;Macke, Tasiemski, Massol, Callens, & Decaestecker, 2017;McFall-Ngai & Ruby, 2000;Sommer & Backhed, 2013). The members that compose host-associated microbial communities often shift depending on the local environmental conditions (Carrier & Reitzel, 2017), the presence of particular species that may facilitate or limit the colonization by other microbes (Vega & Gore, 2017), and the expression of the immune system by the host (Nyholm & Graf, 2012;Thaiss, Zmora, Levy, & Elinav, 2016). Over the past few decades, sequence-based approaches have broadened our understanding of diverse interactions between hosts and associated microbial communities (O'Brien, Webster, Miller, & Bourne, 2019). Specifically, these studies have provided insight into the relative proportions of microbes that are stably symbiotic or transient with a host when experiencing variable environmental conditions (Shade & Handelsman, 2012), including external factors (e.g., temperature, nutrients) or host-regulation (e.g., immune system). Determining how these factors impact host-associated microbial communities in general, and how they affect specific OTUs (operational taxonomic units), would provide a better understanding of how complex microbial communities vary for eukaryotes.
Light is an environmental factor that influences many organisms through a combination of two principal responses. First, light can significantly impact the physiology and survival of an organism following direct exposure, and photosynthetically active wavelengths may impact the function of microbial partners. The result can be positive for increasing growth of certain microbes (e.g., cyanobacteria), where photons are harvested for the production of photosynthates.
Light can also have negative effects by causing damage that can inhibit growth, particularly for short wavelength portions of the light spectrum (Dai et al., 2012). Secondly, animal-associated bacterial communities can, in turn, shift following responses by the host due to an entrained endogenous pathway (the circadian clock). Circadian rhythms are critical internal regulatory systems that allow organisms to anticipate daily changes in their environment and adjust biological processes appropriately. Using an endogenous centralized clock, cycles of about 24-hr are entrained and maintained by exogenous cues (Zeitgebers) that modulate temporal rhythms through a series of transcription-translation feedback loops (Dunlap, 1999). Previous work with vertebrates suggests that the circadian clock is an important regulator of the immune system, which can impact portions of the bacterial community throughout a day. In humans (Huang, Ramsey, Marcheva, & Bass, 2011) and mice (Leone et al., 2015), the gut microbiota is time-of-day-dependent and is hypothesized to modulate the regulation of host metabolism and immunity (Keller et al., 2009;Liang, Bushman, & FitzGerald, 2015;Thaiss et al., 2014).
However, the interaction between light-dependent responses that influence the host's behavior and an endogenous circadian clock remains unknown.
Our knowledge of the connections between diel lighting, circadian rhythms, and symbiotic microbiota remains limited in invertebrates, and especially those in aquatic habitats. One of the best-studied examples is the mutualism between the squid Euprymna scolopes and bacterium Aliivibrio fischeri (formally Vibrio fischeri; Boettcher, Ruby, & McFall-Ngai, 1996;Heath-Heckman et al., 2013;Wier et al., 2010), where the expression of a light-sensitive cryptochrome (escry1) in the host has been linked to the presence of A. fischeri. While the Euprymna-Aliivibrio system provides a host-focused view of circadian-related symbioses, decades of work on the coral microbiome have provided additional context for the evolutionary ecology of daily cycles exhibited by both the holobiont and by each partner (Hoadley, Vize, & Pyott, 2016). Symbiodiniaceae, a eukaryotic endosymbiotic mutualist of corals and other invertebrates, demonstrates diel periodicity of photosynthetic processes in the free-living and mutualistic states, suggesting that symbiotic partners maintain their own circadian clocks and, perhaps, contribute to that of the holobiont Sorek, Díaz-Almeyda, Medina, & Levy, 2014;Sorek, Yacobi, Roopin, Berman-Frank, & Levy, 2013).
Aside from these systems, oscillations of individual members or communities of host-associated microbiota in marine invertebrates are poorly understood. Further, how much of the holobiont rhythmicity is due to a direct response to environmental cues (e.g., light) or is driven by endogenous mechanisms remains an active area of research (Brady, Willis, Harder, & Vize, 2016;Leach, Macrander, Peres, & Reitzel, 2018;Oren et al., 2015;Vize, 2009).
Nematostella vectensis, an infaunal sea anemone that lives in shallow estuaries, is an emerging model for studying the host-associated microbial communities and circadian biology of cnidarians. Similar to corals, N. vectensis exhibits nocturnal patterns in behavior [e.g., circadian locomotion and body expansion (Hendricks, Byrum, & Meyer-Bernstein, 2012;Oren et al., 2015)], gene expression [e.g., immunity and stress tolerance (Leach et al., 2018)], and metabolism (Maas, Jones, Reitzel, & Tarrant, 2016). Unlike corals, N. vectensis does not associate with zooxanthellate and, thus, exhibits rhythmicity independent of the eukaryotic mutualist of corals and other cnidarians. Previous studies have shown that the bacterial community associated with N. vectensis is diverse in natural habitats (Har et al., 2015), variable across development (Mortzfeld et al., 2016), and significantly dissimilar for individuals from different geographic locations (Mortzfeld et al., 2016), which together support this species as a system to study animal and bacterial interactions (Fraune, Foret, & Reitzel, 2016). The innate immune system is a combination of molecular mechanisms that may explain the variation in bacteria associated with cnidarians (Bosch et al., 2009). Genomic and transcriptomic resources for a number of anthozoan and hydrozoan species have identified numerous genes predicted to be involved in cnidarian immunity (Miller et al., 2007;Reitzel, Sullivan, Traylor-Knowles, & Finnerty, 2008). Based on sequence similarity and experimental characterization, cnidarians have many components of a traditionally defined innate immune system, including the Toll-like and NOD-like receptors for microbial recognition (Bosch et al., 2009;Brennan et al., 2017), at least three complement families [e.g., C3, Bf, and MASP (Kimura, Sakaguchi, & Nonaka, 2009), MyD88 and other proteins for intracellular signal transduction (Franzenburg et al., 2012), and Nf-κB along with other Rel-related proteins for transcriptional regulation of effector genes (Sullivan et al., 2009;Wolenski et al., 2011)]. These studies support the hypothesis that the cnidarian-bilaterian ancestor had a rich and complex innate immune system. The regulation of the cnidarian system and how environmental changes may modulate the expression of components of this pathway remain unstudied.
Using high-throughput sequencing of the 16S rDNA gene to represent the bacterial communities associated with N. vectensis, we tested two hypotheses regarding if and how diel lighting influences the anemone-associated bacterial community. First, we tested whether the bacterial community of N. vectensis exhibits diel oscillations synchronous with light:dark cycling, and second, whether individual OTUs were differentially abundant after host exposure to light:dark cycles. Here, we identify compositional differences between anemones exposed to light:dark cycles or constant darkness.
Further, these data reveal specific bacterial OTUs that exhibit diel patterns of abundance in either light regime. By assessing bacterial abundance across diel and constant conditions, our research sheds light on the potential of microbial interactions in the regulation of host anemone cyclic behavior and physiology measured in other studies (Hendricks et al., 2012;Maas et al., 2016;Oren et al., 2015).

| Animal culturing and experimental conditions
Adult Nematostella vectensis derived from the original "Maryland strain" (Putnam et al., 2007) were used for these experiments.
Individuals were reared in a single dish at room temperature (~20-25°C) and ambient lighting conditions (as described in Hand & Uhlinger, 1994). In preparation for the experiment, adults from the common garden conditions were split into two glass dishes and transferred to an incubator at 25°C. Individuals were fed freshly hatched brine shrimp (Artemia sp.) haphazardly three times weekly, and water was replaced following feeding using 200 ml of 15 ppt artificial saltwater.
To simulate diel light conditions, full-spectrum LED lights (MINGER) were set to 12-hr light:12-hr dark (LD) cycles, with lights on at 11:30 a.m. (ZT = 0) and lights off (ZT = 12) at 11:30 p.m. Each dish of individuals was assigned to LD or DD (constant darkness) and was subsequently kept in their respective conditions for 30 days ( Figure 1). During this entrainment period, individuals were kept on the same feeding and water change schedule as previous. Feeding and water changes occurred during "daytime" hours (between ZT = 0 and ZT = 12). To eliminate the potential for light contamination in DD animals, dishes were wrapped in tin foil during the entrainment and sampling periods; however, dishes were briefly removed from the incubator and were handled in a dark room for feeding and water changes.
Two days prior to sampling, individuals from each light regime were split into four glass dishes per condition (LD 1-4; DD 1-4) with 200 ml of fresh, 15 ppt artificial seawater and were starved.
Using the filtered biom table and "biom summarize-table" function to count total sequences per sample, the rarefaction depth of 1,080 (see Dryad for Figure  Faith's phylogenetic distance, and observed OTUs) was calculated using alpha_diversity.py and compared statistically using Student's t test in JMP. Beta diversity was calculated using unweighted and weighted UniFrac (Lozupone & Knight, 2005), compared using principal coordinate analyses (PCoA) with jackknifed_beta_diversity.py, visualized using make_2d_plots.py, and stylized for presentation in Adobe Illustrator CS6. UniFrac distances were then compared statistically using an analysis of similarity (ANOSIM) in QIIME as part of compare_categories.py. Community composition was generated using summarize_taxa_through_plots.py and stylized using Prism 7 (GraphPad Software) and Adobe Illustrator CS6. Differential abundance of OTUs between light regimes was tested using the DESeq2_nbinom algorithm as part of differential_abundance.py.
Lastly, the shared or "core" community was determined using com-  A step-by-step listing of the informatic pipeline, including QIIME scripts, used to convert and process raw reads are available on Dryad in file "Supplemental Note."

| Identification of oscillating OTUs
Using the filtered biom table, we identified oscillating OTUs using the R statistical package JTK_Cycle (version 3.1; Hughes, Hogenesch, & Kornacker, 2010). Specifically, we used the script described by Hughes et al. (2010), setting the parameters to select significantly cycling OTUs between 20 and 28 hr (JTK_Cycle, p < .05; "per" = 24) and shifts in peak expression (JTK_Cycle, "lag") between LD and DD samples were compared (Zeitgeber Time: ZT). JTK_Cycle does not classify units into rhythmic categories; therefore, we compared read counts for OTUs based on their periodicity values ("per") and significance (p-value). ANOVA and post hoc Tukey tests were performed with GraphPad Prism between timepoints and within treatments.

| RE SULTS
To characterize the variation in the bacterial community associated with N. vectensis when cultured in two treatments, light:dark (LD) and constant dark (DD), we used 16S rDNA sequencing to compare microbial diversity and abundance in each light regime.

| Composition of the bacterial community
The bacterial communities associated with N. vectensis were best explained by the presence/absence of a diel photoperiod, so we next  Table S3).
Differences in the bacterial communities of LD and DD entrained anemones were due, in part, to the differential abundance of 37 bacterial OTUs (

| Shared taxa between LD and DD
A shared (or "core") bacterial community for N. vectensis in LD and DD was determined for different proportions of shared OTUs. At a core level of 60%, 70%, 80%, 90%, and 100% (i.e., bacterial phylotypes found in at least "N"% of samples), we observed that 141, 93, 63, 38, and nine phylotypes (see Dryad for Figure S2), respectively, were shared between LD and DD conditions. At core levels 60% and 70%, we observed that the taxonomic representation (but not composition) of these communities was distinct but converged at a core level of 80% (see Dryad for Figure S3).

| Patterns of abundance in LD and DD
Using JTK_cycle (Hughes et al., 2010), we identified 26 bacterial OTUs in LD that showed rhythmic cycling (p < 0.05), five of which exhibited a 24-hr periodicity with peak abundance at either ZT = 20 or ZT = 22 (per = 24; Figure 3; Table 2). Of these five OTUs, four were from the bacterial order Rhodobacterales and the other was from Alteromonadales (Table 2). In DD, 16 bacterial OTUs showed rhythmic cycling (p < 0.05), and five of these exhibited 24-hr periodicity with peak abundance ranging between ZT = 2 and ZT = 20 (per = 24; Figure 3; Table 2). Unlike the cycling OTUs in LD, OTUs in DD with 24-hr cycling were from four disparate bacterial orders: Chlamydiales, Spirochaetales, Oceanospirillales, and CL500-15 (Table 2). When comparing the ten bacterial OTUs that exhibited a 24-hr periodicity to the core community, we observed that OTUs "LD 1," "LD 2," and "LD 5" were specific to LD, while OTU "DD 2" was specific to DD. Moreover, OTUs "LD 2" and "LD 5" were observed at the 80% core level, while the OTU "DD 2" was only detected at the 60% core level.

| D ISCUSS I ON
Photoperiods and circadian clocks are an integral part of diverse biological processes for animals, ranging from immune performance to metabolism to host-microbe associations (Heath-Heckman, 2016; Hubbard et al., 2018;Liang et al., 2015;Zarrinpar, Chaix, Yooseph, & Panda, 2014). In traditional mammalian systems, such as mice and humans, the composition of the gut microbiota of individuals entrained to light:dark or constant darkness differs for particular taxonomic groups of bacteria (Deaver, Eum, & Toborek, 2018;Wu et al., 2018). Thus, the photoperiod may influence compositional dynamics of host-associated bacterial communities in other animals, and the responses may also involve an endogenous circadian clock.  Table 2).
The research we report here suggests that diel lighting may impact a fraction of the microbial community, but this effect appears to be relatively small compared with differences in the associated microbiota over developmental stages or in natural populations (Har et al., 2015;Mortzfeld et al., 2016). For N. vectensis and other animals, the role of the animal host, the environment, and the resident microbiota may play in shaping host-microbe interactions remains fragmentary. One hypothesis for the observed shift in community-level composition reported here is that the lack of a photoperiod drives ecological (or stochastic) drift in the microbes associated with N. vectensis. Generally, in ecological systems, species diversity is expected to increase as environmental heterogeneity increases up to a point as described in Curd, Martiny, Li, and Smith (2018). Our results do not show a positive relationship between environmental variation (i.e., diel lighting) and community diversity; rather, we measured greater community-level variability in constant conditions (Table 2).
A second hypothesis for community-level shifts over a diel light period is that changes in bacteria are attributed to physiological differences between individuals in light:dark and constant darkness (i.e., gene expression, behavior, metabolism; Leach et al., 2018;Maas et al., 2016;Peres et al., 2014;Reitzel et al., 2010;Roopin & Levy, 2012 Table 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2 Z T 6 Z T 1 0 Z T 1 4 Z T 1 8 Z T 2 2 Z T 2 6 Z T 3 0 Z T 3 4 Z T 3 8 Z T 4 2 Z T 2   (Augustin et al., 2017;Bosch et al., 2009;Franzenburg et al., 2012;Fraune et al., 2015). While previous research has shown that N. vectensis has a circadian clock based on behavioral, physiology, and molecular measurements (see Section 1), genes likely to be involved in innate immunity have little differential TA B L E 2 Taxonomy of OTUs with significant 24-hr periodicity associated with adult Nematostella vectensis in light:dark (LD) and constant darkness ( --TA B L E 3 Differential gene expression of candidate cnidarian innate immune genes between light:dark (LD) and constant darkness (DD) determined by DESeq2. NOD genes were arbitrarily given numbers and the order matches the order in Table S1 from Lange et al. (2011)  expression over a 24-hr period, at least when measured in whole animal homogenates. As a preliminary investigation using previously published transcriptomic data (Leach & Reitzel, 2019), we compared the expression of candidate cnidarian immune genes from anemones sampled in LD and DD conditions. The genes selected include the hypothesized principal innate immune genes (e.g., Toll-like receptors, Nf-κB) from Miller et al. (2007) and Brennan et al. (2017), the NOD-like receptors from Lange et al. (2011) and Yuen, Bayes, and Degnan (2014), and the complement genes identified by Kimura et al. (2009) (Table 3). These transcriptomic comparisons showed only a small portion of the genes predicted to be involved in the cnidarian innate immune system (7 out of the 34 surveyed) to be differently expressed between LD and DD (Table 3). Of these seven, four genes were up-regulated in constant dark conditions (compared to LD) and included predicted members of the cnidarian multi-complement pathway (e.g., NvC3-1, NvBF1, and NvBF2) and NOD-like receptors.
The function of any of these genes in N. vectensis is unknown, but

ACK N OWLED G M ENTS
We thank the Reitzel Laboratory for continual support, Jason

AUTH O R CO NTR I B UTI O N S
WBL and AMR designed and conceived the study. WBL and TJC performed the experiment. WBL and TJC generated and analyzed the data. WBL, TJC, and AMR wrote and approved the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data and supplemental files will be archived in the public archive