https://orcid.org/0000-0003-4417-905X
Figures
Abstract
The moon jelly Aurelia coerulea does not typically regenerate appendages. A previous study found that increasing nutrient availability promotes the activation of appendage regeneration. In this study, we found that changes in nutrient availability alter microbiome composition. Genome-scale metabolic modeling of the microbial symbionts suggests that the microbiome modulation reduces the abundance of a symbiont that competes with the host for metabolic resources. Antibiotic treatment that recapitulates the microbiome shift significantly promotes regeneration. Transcriptomic analysis of the host shows an upregulation of genes involved in the arginine biosynthetic pathway, including argininosuccinate synthase and carbamoyl phosphate synthase. Exogenous arginine supplementation promotes appendage regeneration. These findings suggest that microbiome composition, influenced by nutrient availability, contributes to the activation of host regeneration, and identify a role for arginine in promoting regeneration.
Citation: Ohdera A, Wang M, Mansbridge M, Yu C, Ginn R, DeMerit H, et al. (2026) Microbiome modulation in regeneration in jellyfish. PLoS One 21(8): e0355863. https://doi.org/10.1371/journal.pone.0355863
Editor: Phuping Sucharitakul, Tsinghua University, CHINA
Received: January 15, 2026; Accepted: July 27, 2026; Published: August 21, 2026
Copyright: © 2026 Ohdera et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: RNA sequencing reads are available in NCBI under the BioProject Accession PRJNA1248490. The GEMs (iAO314_Mariplasma and iAO833_Marinirickettsia), raw images, and scripts used for RNAseq and GEM analysis are available in the open repository CaltechDATA (https://doi.org/10.22002/wm422-yve83).
Funding: Funding for this project came from Caltech’s Division of Biology and Biological Engineering postdoctoral fellowship (to AO) (https://www.bbe.caltech.edu/), Caltech Center for Environmental Microbial Interactions Pilot grant (to AO) (http://microbiology.caltech.edu/), Max Factor Foundation Fund (to LG), Charles Trimble and the BBE Council’s Regeneration Fund (to LG), and Summer Research Connection (SRC) program Fund at Caltech (to MW, RG, IG, CL, AHK, GN, JO) (https://ctlo.caltech.edu/outreach/summerprograms/summer-research-connection). The funders did not play a role in the study design, data collection and analysis, decision to publish, nor the preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Recovering injured organs is energetically demanding. It is increasingly appreciated that a limiting factor in regeneration is nutrient availability. Ctenophores cease regenerative processes under nutrient-limited conditions [1]. Salamanders reared in low nutrient conditions show substantially reduced limb regeneration [2]. Nutrients are not only necessary for supporting normally occurring regeneration; increasing nutrient availability can promote activation of regeneration in poorly regenerating systems. In jellyfish, increasing nutrient availability promotes activation of appendage regeneration [3]. In tadpoles, which stop regenerating their tails during the refractory period, increased feeding restores tail regeneration [4]. In adult mice, which do not normally regrow digits from proximal amputations, increasing nutrients promotes activation of digit regrowth [3]. Given the role of nutrients in regeneration, we investigated whether the microbiome plays a role in mediating this relationship.
The microbiome can profoundly influence the host nutrient metabolism [5]. The microbiome can aid the breakdown of nutrients, produce molecules otherwise not available to the host, and increase amino acid availability [e.g., 6–10]. From flies to mice, the microbiome can compensate for undernutrition and rescue growth [11,12]. In parallel, the microbiome has also been linked to regenerative processes [13]. In planarians, a pathogenic shift in the microbiome inhibits regeneration [14]. In mice, taxa-specific signals from the gut influence regeneration processes in remote muscle and liver [15]. In frog tadpoles, antibiotic treatment inhibits tail regeneration, which can be rescued by supplying lipopolysaccharides [16]. While these studies connected microbiome and regeneration, few have looked into this connection in the context of varying nutrients. In this study, we began exploring the connections between nutrients, microbiome, and regeneration.
To investigate this, we used the moon jellyfish Aurelia coerulea as a model system. Aurelia is well-suited for this study because its regeneration responses vary across its life cycle (Fig 1a), allowing us to assay regeneration induction. In the sessile stage, similar to other cnidarian polyps such as hydras and sea anemones, Aurelia polyps readily regenerate body parts [17,18]. However, in the free-swimming stages, Aurelia ephyrae and medusae show limited regeneration [19,20]. This is clearly demonstrated in the ephyra (Fig 1b), whose eight discrete appendages, called arms, facilitate tracking of regeneration, or the lack of it. These arms function in swimming and feeding; they contract synchronously to produce fluid flow that propels the animal and captures prey [21]. In response to amputation (Fig 1c), Aurelia ephyrae rapidly reorganize the remaining body parts and arms, and regain radial symmetry (Fig 1d). The recovery of radial symmetry facilitates propulsion and feeding, and ultimately, continued survival [19]. While symmetrization is the primary response to injury, as noted above, it was found that increasing nutrient availability promotes ~40% of animals to begin regenerating partial appendages (ref. 3; reproduced here in Fig 1e and again later in Fig 4b). In addition to the established connection between nutrients and regeneration, the microbiome of Aurelia has been characterized. A recent study found that the Aurelia microbiome is dominated by two bacterial taxa [22]. Therefore, Aurelia offers a low complexity microbiome system to interrogate how the microbiome and nutrients interact to affect regeneration in a poorly regenerating animal.
(a) The moon jelly has a dimorphic life cycle. Sessile polyps reproduce asexually or, with favorable environmental cues, metamorphose into free-swimming juveniles, called ephyrae. Ephyrae grow into sexually reproductive medusae in typically 1–3 months. (b) The experiments in this study were performed on ephyrae, whose eight discrete swimming arms facilitate assessment of regeneration. (c-e) Ephyrae were amputated across the body (red line), removing three arms (c). Injured ephyrae typically reorganize existing arms and regain symmetry within 1–3 days (d). With increasing nutrient availability, as will be reproduced later, some ephyrae begin to regrow partial arms (e). Red arrows denote the arm regenerates. Regeneration remains partial even at 2–3 weeks after amputation, around which time, as part of the transition to medusa stage, bell tissues begin to develop between the arms, which complicates assessing further arm regrowth. Scale bar: 1 mm.
Materials and methods
Polyp culture of Aurelia coerulea (formerly Aurelia aurita sp. 1) used in this study originate from specimens collected by the Cabrillo Marine Aquarium off the coast of Long Beach, CA, USA (33o46’.2”N 118o0744.2W). Polyps were maintained in 32–35 ppt artificial seawater (ASW), at 20–22oC, with 12:12h light cycle, and fed every other day with 48-hr old Artemia (Brine Shrimp Direct, BSEP1.75Z), enriched with RGComplete (Reed Mariculture). To induce metamorphosis, polyps were treated overnight with 25 μM 5-methoxy-2-methyl-indole (Acros Organics, 10434575). Treated polyps were rinsed three times with ASW and fed every other day until metamorphosis began. Metamorphosis typically occurred ~10 days after induction. Newly strobilated ephyrae were fed L-size rotifers (Reed Mariculture) for two days prior to experiments.
Amputation experiments and treatment were performed using an X-acto knife fitted with a #17 blade following methods described previously [3]. Under a dissecting microscope, ephyrae held in a petri dish were amputated by cutting across the body to remove three arms (Fig 1c). Two-day old ephyrae were used for the amputation experiments. Amputated ephyrae were placed in 1 liter settling cones (Nalgene Imhoff, 1000−0010), with gentle aeration, at 25–40 ephyrae/cone and 15 mL of artificial seawater/ephyra. For control, low-food condition, ephyrae were fed ~ 5 rotifers per animal every other day. For the increased food condition, ephyrae were fed ~ 40 rotifers per animal. Rotifer density was determined by calculating the average number of rotifers in five replicates of 50 ul aliquots of well mixed culture to calculate the per milliliter density. The feeding level provided here should only be used as a starting estimate, because in our experience, the response to feeding can sensitively depend on polyp batches, as reflected in the variation across experiments from independent strobilation. We verified in a previous study that this is not due to genetic variations in the population, as the same variation occurs in clonal populations [3]. Accordingly, to establish the experiments, one would need to first calibrate what is low vs high feeding level in their own lab. For penicillin treatment, penicillin was added to the water to give a final concentration of 100 U/mL (G-Biosciences, RC-741). For arginine experiments, L-arginine methyl ester dihydrochloride was added to the water to give a final concentration of 200 μM (Chem-IMPEX International Inc., 03022). Penicillin and arginine treatments were performed with control or increased food condition, as specified in the main text. Water (along with everything in it, food, penicillin, or arginine) was refreshed every other day. Regeneration was assessed at 14 days post-amputation: individual ephyra were imaged, assessed for presence of a newly regenerated arm, and length of regenerated arm quantified using Fiji.
Statistical analysis
To assess the statistical significance in the regeneration experiments, we performed meta-analysis [23]. Meta-analysis was utilized because independent cohorts of animals (from different rounds of strobilation/hatching) can show different baseline rates of regeneration. Consequently, to accurately assess treatment effects, control and treatment groups needed to be set side-by-side using the same cohort of animals, and the treatment effect evaluated within each experiment. Meta-analysis was then performed to assess the reproducibility of the treatment effect across experiments.
Within each experiment, the treatment effect was quantified using standard metric. For the regeneration frequency data, which are in the form of 2 x 2 table, we used the Risk Ratio metric:
When regeneration frequency is zero in one of the datasets being compared, following the standard procedure, the constant 1 was used to compute the effect size. For the regenerate arm length data, which are quantitative, we used the Response Ratio metric:
The length is computed as percent of intact arm length within the same individuals, to account for variations of arm length across individuals. Datasets with zero regenerates in control were not included in the calculations. Having calculated the treatment effect sizes from multiple independent experiments, we used the restricted maximum likelihood model (REML), which enables taking into account heterogeneity across experiments, to assess the reproducibility of the treatment effects across experiments. The analysis was implemented using the metafor package in R ([24]; Response Ratio is called Ratio of Means in metafor).
Quantitative PCR
Control and treated unamputated ephyrae grown as described above were collected after 7 days and flash frozen. DNA extraction was performed using the DNeasy Blood and Tissue Kit (Qiagen, 69504), with a 3 hour proteinase K treatment. qPCR was performed with the QuantiTect SYBR Green PCR Kit (Qiagen, 204143) and StepOne Real-Time PCR system (Applied Biosystems, 4376357). Species-specific primers were used to target the bacterial 16S rDNA genes:
- Mariplasma forward: TGCGTAGATGGTGAAATTAGTC
- Mariplasma reverse: CGCTATTGGTGTTCCTTCATA
- Marinirickettsia forward: GTTATTTGTGAAAGCCCTAAGC
- Marinirickettsia reverse: GCCTTCGCTGTTGGTATT
To normalize the sample load across wells, the host gene Sox1 was used:
- Sox1 forward: CGCCTTGAAAGATAC CCTAAA
- Sox1 reverse: AGCACACCATTCTCCATTATA
Primer specificity was confirmed by gel analysis and Sanger sequencing the amplicon. The qPCR reactions were run with the following condition: an initial denaturation step at 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. A melt curve analysis was performed immediately after amplification to confirm product specificity.
To calculate the relative abundance of the bacteria, Ct values were first converted to raw DNA quantities and corrected to account for primer efficiency using the formula:
The DNA quantity was then normalized by the host Sox1 gene quantity, to account for loading variability. Subsequently, the Sox1-normalized DNA quantity was normalized by the mean of the control data. These are the relative bacterial abundances plotted in the main figures. Mariplasma-to-Marinirickettsia ratios were calculated by taking the ratio of Mariplasma and Marinirickettsia relative abundance within each biological replicate. To compute statistical significance, the non-parametric Kruskal-Wallis test was used.
Reconstruction of the genome-scale metabolic models (GEMs) was performed following the established workflow. First, an automated model reconstruction of the Mariplasma and Marinirickettsia genomes (NCBI Accession PRJNA975886) was performed with CarveMe (version 1.3.0) [25]. Next, the resulting model was further manually curated. Each reaction in the model was cross-referenced to the gene annotation and function. Gene functions were determined by BLASTp against the NCBI nr database. Based on the BLASTp results, reactions that fell into the following criteria were removed: (1) The reaction does not have functioning enzymatic annotation; (2) The reaction does not have a flux under optimum growth conditions (as determined with cobra_model.optimize(), where cobra_model is the name of the model input) and does not affect the feasibility of the model under a knock out simulation (using cobra_model.reactions.reactionID.knock_out(), where cobra_model is the name of the model input and reactionID is the name of the reaction); and (3) The reaction is not likely to be present for the species, as assessed from available experimental literature. The resulting, curated model was solved using Flux Balance Analysis (FBA) and Flux Variability Analysis (FVA) in COBRApy ver. 0.15.4 [26]. The models were solved to optimize the growth rate, using the default objective function. To achieve growth rates comparable to those experimentally observed for Mollicutes and Rickettsiales species (S1 Table), we varied the upper and lower bounds for metabolite uptake and secretion (for both models, the final values were, exchange_default_ub = 1000, exchange_default_lb = −10).
RNA sequencing and analysis
Animals were amputated as described above, and then reared in three conditions: control, increased food at 3–4 times the amount of control, and control food supplemented with penicillin. At five days after amputation, 5–7 animals were flash frozen in liquid nitrogen. RNA extraction was performed using the RNeasy Plus Mini kit (Qiagen, 74104), with the following modifications. RLT buffer containing β-mercaptoethanol was added to the frozen samples and the samples were homogenized on ice. Homogenized samples were centrifuged with the QIAshredder at 12,000 g for 2 minutes, and RNA was then purified with the RNeasy kit following the manufacturer’s protocol. DNA was removed from the samples using the DNA-free DNA Removal kit (ThermoFisher Scientific, AM1906), followed by poly-A selection using NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB, E7490S). Samples were submitted to the Caltech Millard and Muriel Jacobs Genetics and Genomics Core facility for library prep and 50 bp paired-end sequencing using NextSeq 2000 sequencer. Reads were trimmed and quality filtered using Trimmomatic (version 0.39; [27]). Trimmed reads were mapped to the Aurelia coerulea genome [28] using STAR (version 2.5.3a; [29]) to generate the counts table. Differential gene expression analysis was conducted using DESeq2 (version 1.34.0; [30]). The Wald test was used for hypothesis testing. Log2 fold change shrinkage was applied for pairwise contrasts with a false discovery rate cutoff of 0.05. Gene Ontology analysis was performed using the R package clusterProfiler (version 4.2.2; [31]), using Biological Process aspect and a q-value threshold of 0.05, and plotted using REVIGO [32].
Presence or absence of arginase across cnidarian groups was analyzed using available genomes [33–40], considering them as representative among currently annotated genomes. The cnidarian genomes were retrieved from the repositories specified in the source papers. For Hydra, protein models from version 2.0 were retrieved from the NHGRI repository (https://research.nhgri.nih.gov/hydra/). Gene models were queried with a representative set of arginine cycle genes from Nematostella (XP_032228695.1, XP_001635121.1, XP_048589032.1, XP_020896709.1, XP_001640743.2) using protein BLAST [41].
Results
Increasing nutrient abundance alters the microbiome composition
As characterized in previous work [22], the microbiome of Aurelia is dominated by two bacterial taxa, a Mollicutes (Candidatus Mariplasma lunae) and a Rickettsiales (Candidatus Marinirickettsia aquamalans). In the ephyra stage, the two bacteria encompass ~97% of the total bacterial abundance, at ~1:1 ratio. Although minor taxa may also contribute to host physiology, our analysis focused on the two dominant taxa.
To quantify these taxa, we utilized the metagenome-assembled genomes from our previous study [22] to design species-specific primers for quantitative PCR analysis (Fig 2). Measurements of the relative abundance of Mariplasma and Marinirickettsia in the control condition are shown in Fig 2a and 2b, respectively. Although the abundance of Mariplasma and Marinirickettsia vary across biological replicates, their relative composition within each biological replicate is more precise, as seen in the narrower spread of the Mariplasma-to-Marinirickettsia ratio (Fig 2c). Note that the designation of “control” feeding level is for simplicity, and does not presume the nutrient level in the wild. Nutrient level is likely to vary in the wild; we chose as the control a feeding level in the lab that best recapitulates a typical growth rate in wild populations (ephyrae mature to medusae over 1–3 months; [42]).
Three conditions were assessed: control level of feeding (see Methods for amount and frequency), increased level of food (~eight times higher than control food), and control food supplemented with 100 U/mL penicillin (Pen). Ephyrae were collected seven days after the start of the treatment. Quantitative PCR was performed using species-specific primers against the 16S rRNA genes. Each bacterium only has a single copy of a 16S rRNA gene. Each data point (black circle) is a measurement from a biological replicate (25 animals). From each biological replicate, we measured the relative abundance of Mariplasma (a), Marinirickettsia (b), and the ratio between two (c). Within each plot, the data are normalized to the average of the controls. Statistical analysis: Kruskal-Wallis test. ns: not significant. ** p < 0.001.
Having established the assay, we examined the effect of increasing nutrient availability. Increasing food does not affect Mariplasma abundance (Fig 2a). Increasing food also does not affect Marinirickettsia abundance with statistical significance, although there is a noticeable downward shift in the spread (Fig 2b). This downward shift in Marinirickettsia abundance leads to a statistically significant increase in the Mariplasma-to-Marinirickettsia ratio (Fig 2c, by a mean of two-fold; p < 0.001; Kruskal-Wallis test, Pen vs Control). These measurements show that the nutrient environment modulates microbiome composition.
Genome-scale metabolic reconstruction of Marinirickettsia and Mariplasma
The impacts of nutrient availability on the microbiome composition motivated us to build genome-scale metabolic models (GEMs) of the two dominant bacterial taxa (Fig 3). A GEM translates the genomic information into a network of metabolic reactions, which have proven to be a useful approach to assess biological capabilities of novel bacteria [43]. We built GEMs using the metagenome-assembled genomes of the bacteria [22]. Facilitating the GEM analysis is the small genomes of the bacteria (0.4 Mb for Mariplasma, 1 Mb for Marinirickettsia). The small genomes and consequently reduced numbers of genes mean that there are few alternative metabolic networks that the genome enables. We verified that the GEM solutions are robust to variation in parameters (S1 Fig).
(a, c) Each GEM diagram shows the reactions that form the top 90% of the total fluxes in the system. See S1 Table for a complete list of the GEM reactions and fluxes. In the Marinirickettsia GEM, Q8: Ubiquinone-8, Q8H2: Ubiquinol-8. (b, d) Shown are metabolites whose transport fluxes form the top 90% of the uptake and secretion. For instance, in the Marinirickettsia GEM, uptake of pyruvate constitutes 30% of all uptake fluxes (b); while in the Mariplasma GEM, ornithine secretion forms 25% of all secretion fluxes from the bacterium (d). The full GEM solutions can be found in S1 Table.
The Marinirickettsia GEM (Fig 3a) is dominated by fluxes of the tricarboxylic acid cycle (TCA), which accounts for 33% of the total fluxes, and electron transport chains, which account for an additional 39%. Marinirickettsia GEM does not possess a functioning glycolysis pathway. Consequently, to fuel its TCA cycle, Marinirickettsia is predicted to uptake pyruvate (Fig 3b). In addition to pyruvate as the top uptake, Marinirickettsia is predicted to sequester from the host a multitude of other metabolites, including the majority of amino acids and cofactors (S1 Table). In contrast to Marinirickettsia GEM, the Mariplasma GEM, while also predicted to uptake some metabolites from the host, also secrete nitrogenous metabolites (Fig 3c and 3d). Growth in the Mariplasma GEM is fueled by the arginine deiminase (ADI) pathway, whose fluxes account for 71% of total flux in the model (Fig 3c). The ADI pathway is conserved across multiple bacterial taxa [44]. It metabolizes arginine to produce ATP, and generates as by-products, ammonium and the amino acid ornithine. Since the Mariplasma genome encodes no genes for arginine biosynthesis, the GEM predicts that the bacterium uptakes arginine from the host (Fig 3d). Almost the entire arginine taken up (98%) is used to fuel the ADI pathway, with the remaining 2% assimilated for growth, consequently predicting secretion of ornithine and ammonium (Fig 3d).
Thus, the GEM analysis shows that the bacterium that is depleted in the regenerating condition, Marinirickettsia, is predicted to sequester high-energy metabolites from the host. On the other hand, Mariplasma, which remains unaffected in the regeneration condition, is predicted to sequester fewer metabolites from the host.
Reducing Marinirickettsia with an antibiotic promotes host regeneration
Based on the qPCR results and the GEM predictions, we tested whether reducing Marinirickettsia impacts host regeneration. Mariplasma belongs to the class Mollicutes, which is characterized by the lack of a cell wall [45]. To selectively target Marinirickettsia, we therefore tested penicillin that disrupts cell wall synthesis.
Penicillin treatment, as expected, does not affect Mariplasma abundance (see ‘Pen’ in Fig 2a). By contrast, penicillin treatment produces a marked decrease in Marinirickettsia (by 5-fold, Fig 2b), which correspondingly shifts the Mariplasma-to-Marinirickettsia ratio (Fig 2c). Next, to assess the effect of reducing Marinirickettsia on host regeneration, we performed amputation experiments and treated the amputated animals with penicillin (Fig 4a). Penicillin treatment strongly promotes regeneration (Fig 4b and 4c, and see more replications in S2 Table; p < 0.0001; meta-analysis, Pen vs Control): 65–75% of penicillin-treated animals showed regeneration activation even when the control group showed none (left panel in Fig 4b). Moreover, penicillin treatment shows an additive effect with increasing nutrients (right panel in Fig 4b). Although regeneration frequency varies from clutch to clutch, as also previously observed [3], penicillin treatment consistently increases the frequency of regeneration activation. Finally, penicillin treatment not only increases the frequency of regeneration, penicillin-treated animals can regenerate longer arms (morphometric analysis in S2 Fig).
(a-b) Ephyrae were amputated across the body (grey line) to remove three arms, and placed in regular seawater or seawater mixed with penicillin at 100 U/mL. Feeding was performed at control level or increased level, at 3–4 times higher than control. Regeneration was assessed two weeks after amputation. Each top and bottom pair of bars is an independent experiment, in which the two groups of animals were set up side by side (20–115 animals pooled in each group). Statistical significance: meta-analysis. ***p < 0.0001. The data used for the plot can be found in S2 Table. (c) Representative ephyrae in penicillin treatment. Red arrows: regenerating arms. Scale bar: 1 mm.
Regenerating animals upregulate genes related to arginine metabolism
Having analyzed the microbial side, we next investigated changes in the host transcriptional activities under regeneration-promoting conditions. To do this, we performed RNA sequencing (RNAseq). We compared three conditions: the low-regeneration condition (control food) and the two high-regeneration conditions (increased food and penicillin; Fig 5a). We analyzed two regeneration-promoting conditions because we reasoned that while each treatment may have non-specific effects, the overlap between the two would give a better assessment of regeneration-specific processes. Animals were collected five days after amputation for RNA extraction, just prior to onset of morphogenesis, to focus the analysis on the early metabolic processes that influence regeneration activation.
(a) Three conditions were tested: control food, increased food at eight times higher the control amount, and control food supplemented with 100 U/mL Penicillin. For each condition, four biological replicates were collected (6–15 animals in each). (b-c) Differentially expressed (DE) genes were determined using DEseq analysis. Statistically significant DE genes are those with p-value <0.05 in both pairwise comparisons (Increased food vs Control and Pen vs Control). (c) Color indicates z-score, the regularized transcript counts for each gene in log2 scale. Only annotated genes are shown in the heat map; see S3 Table for the full list of DE genes. (d) GO terms enriched in upregulated shared DE genes, plotted using Revigo. A q-value cutoff of 0.05 was used for the GO enrichment analysis. Circle size: GeneRatio, i.e., ratio of number of DE genes to total number of genes in the GO term. Circle color: statistical significance of the enrichment. (e) This diagram is based on the KEGG arginine biosynthesis pathway, curated for those genes that we have verified as present in the Aurelia genome. Arrow indicates an enzymatic reaction; enzyme catalyzing the reaction is shown in italics. Enzymes shown in red italics are upregulated. Blue cross: arginase is missing from the Aurelia genome (S3 Fig), but intriguingly, as suggested in the dashed box, Mariplasma can in theory complement this missing biochemical link; more on this in Discussion.
The RNAseq analysis identified 5324 differentially expressed (DE) genes in the increased-food condition and 1830 DE genes in the penicillin treatment (S3 Table). Of these, 1416 genes are differentially expressed in both increased-food and penicillin treatment (Fig 5b and 5c; the full list in S3 Table). On these shared DE genes, we performed gene ontology (GO) enrichment analysis. The top 25 GO terms enriched in the regenerating conditions are shown in Table 1. As expected, the DE genes are enriched for cell adhesion, extracellular matrix, and migration (Table 1)—processes that accompany early stages of morphogenesis [46–48]. In addition to morphogenetic processes, the enrichment analysis recovered processes in signaling (ERK and Wnt), redox and antioxidant metabolism, and immune processes.
Finally the enrichment analysis also recovered, notably, many upregulated DE genes associated with arginine metabolism (Fig 5d). The enriched GO terms include the arginine biosynthetic pathway itself, also known as the urea cycle (see bolded terms in Table 1). Looking closely into the genes that drive the enrichment in arginine metabolism, we find that two genes in the pathway that catalyze the ATP-dependent steps are upregulated (carbamoyl phosphate synthase and argininosuccinate synthase; red in Fig 5e). Arginine metabolism is intertwined with that of glutamine and glutamate—the three are often grouped together because of their roles in nitrogen balance, with their metabolism adjacent to the entry of nitrogen from inorganic forms. As shown in Fig 5e, upregulated in the regenerating conditions are not only genes in the arginine cycle itself, but also multiple genes in the adjacent metabolic pathways of glutamine and glutamate.
Arginine supplementation promotes activation of regeneration
In summary, transcriptomic analysis identifies a number of biological processes that are differentially regulated in the regenerating conditions, which can pave the way for next mechanistic studies. Of the processes enriched in the regenerating condition, we focus in this final section to further test arginine. The enrichment in arginine metabolism is interesting to us for a number of reasons. First, arginine is a non-essential amino acid that becomes essential during a period of stress, injury, and growth [49–51]—thus may plausibly be limiting in regeneration activation. Second, one of the two dominant symbionts in Aurelia, Mariplasma, which is not depleted in the regenerating conditions identified so far (Fig 2), coincidentally, is predicted to metabolize arginine (Fig 3).
To investigate the role of arginine, we tested administering arginine to the amputated animals. Amputated animals supplemented with arginine are indeed more likely to activate appendage regeneration (Fig 6a and 6b; p < 0.001, meta-analysis, Arg vs Control). As shown in Fig 6a (left panel), 34 ± 1% of the animals activated appendage regeneration even when the control groups showed none. Arginine treatment, similar to penicillin treatment, also shows an additive effect with increasing food (Fig 6a, right panel). Finally, we assessed how arginine supplementation affects the microbiome. As expected from arginine being the fuel for Mariplasma, arginine supplementation markedly increases the relative abundance of Mariplasma (p < 0.0001, Kruskal-Wallis test, Arg vs Control), while showing much less effect on Marinirickettsia (Fig 6c).
(a-b) Amputated ephyrae were fed control food, increased food (blue, at 3–4 times higher than control), or control food supplemented with 200 μM L-arginine (black). Regeneration was assessed two weeks after amputation. Each top and bottom pair of bars is an independent experiment, in which the two groups of animals were set up side by side (30–80 animals in each group). Statistical significance: meta-analysis. ** p < 0.001. The raw data can be found in S2 Table. (b) Representative ephyrae in arginine treatment. Red arrows: regenerating arms. Scale bar: 1 mm. (c) Relative abundance of Marinirickettsia and Mariplasma in control and arginine-supplemented condition, assessed using quantitative PCR, as described in Fig 2. Each data point (black circle) is a biological replicate (25 animals pooled). The data are plotted normalized to the average of the controls. Statistical analysis: Kruskal-Wallis test. ns: not significant. ** p < 0.001. *** p < 0.0001.
Discussion
It was previously found that increasing nutrient availability promotes activation of appendage regeneration in Aurelia. In this study, we find that increasing nutrient availability alters the microbiome composition. The microbiome composition in regenerating conditions can correspond to either a decrease in Marinirickettsia (Figs 2 and 4) or an increase in Mariplasma (Fig 6); both are interestingly similar in that they increase the Mariplasma-to-Marinirickettsia ratio. Since the antibiotic treatment that significantly targets Marinirickettsia can recapitulate regeneration induction (Fig 2), we propose that the change in microbiome composition is not a simple correlation, but meaningfully influences host regeneration. Finally, we analyzed the change in the host transcriptome with the microbiome composition that promotes regeneration, and identified, among the enriched processes, arginine metabolism. Exogenous arginine supplementation indeed promotes regeneration, even while increasing Mariplasma load.
How might arginine promote regeneration in Aurelia? In many animals, the body can normally synthesize enough arginine, except under a period of growth, stress, and injury [49–51]. In these energetically demanding periods, arginine becomes limiting, and exogenous supply of arginine from nutrients becomes essential. Arginine, and its related amino acid ornithine, plays multiple roles in tissue growth [e.g., 52,53]. Arginine is a building block of proteins and a signal that directly activates the TOR pathway, the major regulator of cell growth [54–56]. Finally, arginine can interconvert to ornithine, which is the precursor for polyamines, essential during proliferative periods and wound healing [57,58]. It will be interesting to test in the next studies whether one or more of these processes are involved in how arginine promotes regeneration.
Our data suggest that a reduction in Marinirickettsia is beneficial for host regeneration (Figs 2 and 4), Marinirickettsia is closely related to Rickettsia, Neorickettsia, and Wolbachia [22]; genera that house parasitic species [59–61]. Species of Rickettsiales can lack key enzymes necessary for glycolysis, and are thought to acquire pyruvate from the host [e.g., 62,63]. Consistent with many related species being parasitic, the GEM analysis in this study suggests that Marinirickettsia potentially competes with the host for resources, and predicts pyruvate as a top nutrient taken up by Marinirickettsia (Fig 3). Pyruvate is a critical ATP-producing molecule in animal cells, and glycolytic activity leading to pyruvate production can increase rapidly after injury [64,65]. Pyruvate can also play a role in regulating chromatin accessibility and transcriptional response [66]. Thus, we hypothesize that a reduction in Marinirickettsia may promote host regeneration by increasing the pool of available nutrients.
While our data implicate reduction in Marinirickettsia as beneficial for host regeneration, the role of Mariplasma is less clear. Mariplasma utilizes the arginine deiminase pathway (ADI) pathway to metabolize arginine (Fig 3), but an increase in the Mariplasma-to-Marinirickettsia ratio clearly correlates with increased regeneration (Figs 2 and 6). We are unable to test the metabolites Mariplasma secretes (i.e., ammonium and ornithine), since the slightest increases in their availability in water make the animals unhealthy. Since arginine supports regeneration activation (Fig 6), and Mariplasma metabolizes arginine to make ATP (Fig 3), one immediate hypothesis is that Mariplasma is competing with the host for this key metabolite. However, there is a hint from the genome analysis of a potentially intriguing role for Mariplasma. Examining the Aurelia arginine/ornithine cycle more closely, we find that the Aurelia genome encodes all the enzymes in the arginine cycle, except for one: arginase (see the blue cross in Fig 5e). The loss of arginase is surprising to us because the arginine/ornithine cycle is deeply conserved, whose evolutionary origin likely predates the emergence of Metazoa [67,68]. To verify the loss of arginase in Aurelia, we analyzed available cnidarian genomes, and find arginase in other cnidarian classes, but specifically missing in all surveyed genomes of scyphozoan jellies, the class to which Aurelia belongs (S3 Fig). Tantalizingly, the Mariplasma GEM metabolizes arginine and secretes ornithine as a by-product (the dashed box in Fig 4b). Thus, genomic analysis suggests that the Mariplasma ADI pathway can precisely complement a missing biochemical link in Aurelia, and may paradoxically support host arginine production. Indeed, as we previously surveyed (ref. 22 and references therein), while association with Marinirickettsia appears to be one-off, the association with Mariplasma, with >99% similarity in 16S rRNA sequence, recurs in Aurelia from multiple geographical locations, indicating the existence of an Aurelia Mariplasma. It will be interesting next to test if Aurelia and Mariplasma have indeed formed biochemical complementarity.
The ADI pathway is a deeply conserved pathway utilized by multiple bacterial taxa [44,69]. Thus, arginine metabolism along with the bacterial ADI pathway may be factors that can promote regeneration, potentially across species. To summarize, as the role of nutrients in promoting regeneration is increasingly identified in multiple species [1–4], the findings in this study can motivate further investigations into the role of microbiome composition in linking the nutrient environment and regeneration.
Supporting information
S1 Fig. The dominant fluxes of the Marinirickettsia and Mariplasma GEMs are robust to variation in the nutrient parameters and metabolic fluxes.
Shown in all the plots are the top 25 reactions, the full reaction list can be found in S1 Table. Flux is plotted as mmol/gram dry weight/hour (mmol/gDW/h). (a, c) In these simulations, the GEM was solved (using FBA analysis) at varying nutrient flux. 1x nutrient flux corresponds to the solution shown in the main figures. Shown are the top 25 reactions, the full reaction list can be found in S1 Table. The dominant fluxes in the GEMs remain dominant when the nutrient fluxes into the system are varied by ten fold. (b, d) In these simulations, the GEMs were solved using Flux Variability Analysis (FVA), which enables identification of, for each reaction, the range of feasible fluxes that support the optimal growth. Orange circle denotes the FBA solution; blue line is the FVA range. A large flux range means that we can change the flux of that reaction, and there are other ways the network can compensate for that change while still producing the optimal growth. For instance, in the Marinirickettsia GEM, a major means of proton import is predicted to occur through diffusion (Htex reaction). The FVA analysis tests the effects of varying this reaction. When the Htex reaction was set to zero, protons were instead imported via the Butyrate-Proton symporter (BUTt2r reaction) and the Glycine-Proton symporter (GLYt2r reaction) (not shown). When the Htex reaction was set at maximum flux (1000 mmol/gDW/hour), this was compensated by upregulation of the D-galacturonate-Proton symporter (GALURt2r reaction), which secretes excess protons. A small flux range means that the reaction is critical for supporting an optimal growth. For instance, most fluxes in the TCA reactions in Marinirickettsia GEM and all ADI reactions in the Mariplasma GEM are highly constrained, indicating that they cannot be varied without reducing growth.
https://doi.org/10.1371/journal.pone.0355863.s001
(TIFF)
S2 Fig. Measurement of arm regenerate length.
(a) Quantitation of arm length was performed in FIJI. Arm length was measured from the body margin to the rhopalium (see the red arrow). A threshold of 15% was used for calling a regenerating arm. (b) The length of arm regenerate is plotted as % relative to length of the intact arm. To measure the length of the intact arm, we measured the length of all five intact arms in the same individual, and then used the average value. Shown here are quantitations from 6 independent experiments, see S2 Table for more replications and statistical analysis. As clearly illustrated in Exp 3–6, penicillin-treated animals can regenerate longer arms than those stimulated by increased food availability.
https://doi.org/10.1371/journal.pone.0355863.s002
(TIFF)
S3 Fig. Arginase is missing in the scyphozoan lineage.
To assess the arginine biosynthesis pathway across the cnidarian phylogeny, we analyzed available genomes of cnidarian species The arginine biosynthesis pathway is driven by five genes, all analyzed here. Black indicates the gene is present; white indicates the gene is absent. Arginase is present in three cnidarian classes (Anthozoa, Cubozoa, and Hydrozoa) and appears to be specifically missing in Scyphozoa. Interestingly, a different gene, ornithine transcarbamylase is missing from both hydrozoan genomes analyzed. The arginine biosynthesis pathway may have undergone different evolutionary sculpting in different cnidarian lineages, which can be further tested as more cnidarian genomes become available.
https://doi.org/10.1371/journal.pone.0355863.s003
(TIFF)
S1 Table. GEM analysis of Ca. Marinirickettsia aquamalans and Ca. Mariplasma lunae.
https://doi.org/10.1371/journal.pone.0355863.s004
(XLSX)
S2 Table. Data from regeneration experiments.
https://doi.org/10.1371/journal.pone.0355863.s005
(XLSX)
Acknowledgments
We thank Gloria Bates and Yutian Li of Caltech, Sarah Bagby of Case Western University, and Cheryl Lewis Ames of Tohoku University for discussion and feedback on the manuscript. We thank Mitchell Aiken from the Center of Teaching, Learning, and Outreach at Caltech and Dana Huley from the Polytechnic School for coordinating summer research activities for the high school students (MW, RG, IG, CL, AHK, GN, JO), and the Flintridge Preparatory School and the Flintridge Prep Jelly Research Lab for facilitating some of the experiments for this work (by MW, RG, HD, and DH). We thank Sheila Kitchen at Texas A&M Galveston for the use of her laboratory space and equipment.
References
- 1. Bading KT, Kaehlert S, Chi X, Jaspers C, Martindale MQ, Javidpour J. Food availability drives plastic self-repair response in a basal metazoan- case study on the ctenophore Mnemiopsis leidyi A. Agassiz 1865. Sci Rep. 2017;7(1):16419. pmid:29180635
- 2. Peng Z-L, Yin B-X, Ren R-M, Liao Y-L, Cai H, Wang H. Altered metabolic state impedes limb regeneration in salamanders. Zool Res. 2021;42(6):772–82. pmid:34643071
- 3. Abrams MJ, et al. A conserved strategy for inducing appendage regeneration. eLife. 2021;10:e65092.
- 4. Williams MC, Patel JH, Kakebeen AD, Wills AE. Nutrient availability contributes to a graded refractory period for regeneration in Xenopus tropicalis. Dev Biol. 2021;473:59–70. pmid:33484704
- 5. Cani PD, Van Hul M, Lefort C, Depommier C, Rastelli M, Everard A. Microbial regulation of organismal energy homeostasis. Nat Metab. 2019;1(1):34–46. pmid:32694818
- 6. Macfarlane S, Macfarlane GT. Regulation of short-chain fatty acid production. Proc Nutr Soc. 2003;62(1):67–72. pmid:12740060
- 7. Li T-T, Chen X, Huo D, Arifuzzaman M, Qiao S, Jin W-B, et al. Microbiota metabolism of intestinal amino acids impacts host nutrient homeostasis and physiology. Cell Host Microbe. 2024;32(5):661-675.e10. pmid:38657606
- 8. Han Z, Zhao L, Hu Q, Hung I, Liu C, Liu S, et al. Gut microbiota-mediated modulation of host amino acid availability and metabolism. Gut Microbes. 2025;17(1):2552345. pmid:40878016
- 9. Jing TZ, et al. Most dominant roles of insect gut bacteria: digestion, detoxification, or essential nutrient provision? Microbiome. 2020;8(38).
- 10. Henderson G, Cox F, Ganesh S, Jonker A, Young W, Global Rumen Census Collaborators, et al. Rumen microbial community composition varies with diet and host, but a core microbiome is found across a wide geographical range. Sci Rep. 2015;5:14567. pmid:26449758
- 11. Schwarzer M, Gautam UK, Makki K, Lambert A, Brabec T, Joly A, et al. Microbe-mediated intestinal NOD2 stimulation improves linear growth of undernourished infant mice. Science. 2023;379(6634):826–33. pmid:36821686
- 12. Storelli G, Defaye A, Erkosar B, Hols P, Royet J, Leulier F. Lactobacillus plantarum promotes Drosophila systemic growth by modulating hormonal signals through TOR-dependent nutrient sensing. Cell Metab. 2011;14(3):403–14. pmid:21907145
- 13. Díaz-Díaz LM, Rodríguez-Villafañe A, García-Arrarás JE. The role of the microbiota in regeneration-associated processes. Front Cell Dev Biol. 2022;9:768783. pmid:35155442
- 14. Arnold CP, et al. Pathogenic shifts in endogenous microbiota impede tissue regeneration via distinct activation of TAK1/MKK/p38. eLife. 2016;5.
- 15. Hanna BS, Wang G, Galván-Peña S, Mann AO, Ramirez RN, Muñoz-Rojas AR, et al. The gut microbiota promotes distal tissue regeneration via RORγ+ regulatory T cell emissaries. Immunity. 2023;56(4):829-846.e8. pmid:36822206
- 16. Bishop TF, Beck CW. Bacterial lipopolysaccharides can initiate regeneration of the Xenopus tadpole tail. iScience. 2021;24(11):103281. pmid:34765912
- 17. Steinberg SN. The regeneration of whole polyps from ectodermal fragments of scyphistoma larvae of Aurelia aurita. Biol Bull. 1963;124:337–43.
- 18. Lesh-Laurie GE, Hujer A, Suchy P. Polyp regeneration from isolated tentacles of Aurelia scyphistomae: a role for gating mechanisms and cell division. Hydrobiologia. 1991;216–217(1):91–7.
- 19. Abrams MJ, Basinger T, Yuan W, Guo C-L, Goentoro L. Self-repairing symmetry in jellyfish through mechanically driven reorganization. Proc Natl Acad Sci U S A. 2015;112(26):E3365-73. pmid:26080418
- 20. Gong M, Ashok M, Helou A, Goentoro L. Jellyfish shape as a mechanical balance. Proc Natl Acad Sci U S A. 2025;122(13):e2412082122. pmid:40096622
- 21. Sullivan BK, Suchman CL, Costello JH. Mechanics of prey selection by ephyrae of the scyphomedusa Aurelia aurita. Mar Biol. 1997;130:213–22.
- 22. Ohdera AH, Mansbridge M, Wang M, Naydenkov P, Kamel B, Goentoro L. The microbiome of a Pacific moon jellyfish Aurelia coerulea. PLoS One. 2024;19(4):e0298002. pmid:38635587
- 23.
Borenstein M, Hedges L, Higgins J, Rothstein H. How a meta-analysis works. In: Introduction to meta-analysis. John Wiley & Sons, Ltd; 2009. p. 1–7.
- 24. Viechtbauer W. Conducting meta-analyses in R with the metafor Package. J Stat Soft. 2010;36(3).
- 25. Machado D, Andrejev S, Tramontano M, Patil KR. Fast automated reconstruction of genome-scale metabolic models for microbial species and communities. Nucleic Acids Res. 2018;46(15):7542–53. pmid:30192979
- 26. Ebrahim A, Lerman JA, Palsson BO, Hyduke DR. COBRApy: COnstraints-Based Reconstruction and Analysis for Python. BMC Syst Biol. 2013;7:74. pmid:23927696
- 27. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics. 2014;30(15):2114–20. pmid:24695404
- 28. Gold DA, Katsuki T, Li Y, Yan X, Regulski M, Ibberson D, et al. The genome of the jellyfish Aurelia and the evolution of animal complexity. Nat Ecol Evol. 2019;3(1):96–104. pmid:30510179
- 29. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. pmid:23104886
- 30. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. pmid:25516281
- 31. Yu G, Wang L-G, Han Y, He Q-Y. clusterProfiler: an R package for comparing biological themes among gene clusters. OMICS. 2012;16(5):284–7. pmid:22455463
- 32. Supek F, Bošnjak M, Škunca N, Šmuc T. REVIGO summarizes and visualizes long lists of gene ontology terms. PLoS One. 2011;6(7):e21800. pmid:21789182
- 33. Chapman JA, et al. The dynamic genome of Hydra. Nature. 2010;464:592–6.
- 34. Leclère L, Horin C, Chevalier S, Lapébie P, Dru P, Peron S, et al. The genome of the jellyfish Clytia hemisphaerica and the evolution of the cnidarian life-cycle. Nat Ecol Evol. 2019;3(5):801–10. pmid:30858591
- 35. Khalturin K, Shinzato C, Khalturina M, Hamada M, Fujie M, Koyanagi R, et al. Medusozoan genomes inform the evolution of the jellyfish body plan. Nat Ecol Evol. 2019;3(5):811–22. pmid:30988488
- 36. Kim H-M, Weber JA, Lee N, Park SG, Cho YS, Bhak Y, et al. The genome of the giant Nomura’s jellyfish sheds light on the early evolution of active predation. BMC Biol. 2019;17(1):28. pmid:30925871
- 37. Li Y, Gao L, Pan Y, Tian M, Li Y, He C, et al. Chromosome-level reference genome of the jellyfish Rhopilema esculentum. Gigascience. 2020;9(4):giaa036. pmid:32315029
- 38. Ohdera A, Ames CL, Dikow RB, Kayal E, Chiodin M, Busby B, et al. Box, stalked, and upside-down? Draft genomes from diverse jellyfish (Cnidaria, Acraspeda) lineages: Alatina alata (Cubozoa), Calvadosia cruxmelitensis (Staurozoa), and Cassiopea xamachana (Scyphozoa). Gigascience. 2019;8(7):giz069. pmid:31257419
- 39. Voolstra CR, Li Y, Liew YJ, Baumgarten S, Zoccola D, Flot J-F, et al. Comparative analysis of the genomes of Stylophora pistillata and Acropora digitifera provides evidence for extensive differences between species of corals. Sci Rep. 2017;7(1):17583. pmid:29242500
- 40. Zimmermann B, Montenegro JD, Robb SMC, Fropf WJ, Weilguny L, He S, et al. Topological structures and syntenic conservation in sea anemone genomes. Nat Commun. 2023;14(1):8270. pmid:38092765
- 41. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J Mol Biol. 1990;215(3):403–10. pmid:2231712
- 42. Lucas CH. Reproduction and life history strategies of the common jellyfish, Aurelia aurita, in relation to its ambient environment. Hydrobiologia. 2001;451(1–3):229–46.
- 43. O’Brien EJ, Monk JM, Palsson BO. Using genome-scale models to predict biological capabilities. Cell. 2015;161(5):971–87. pmid:26000478
- 44. Cunin R, Glansdorff N, Piérard A, Stalon V. Biosynthesis and metabolism of arginine in bacteria. Microbiol Rev. 1986;50(3):314–52. pmid:3534538
- 45. Dybvig K, Voelker LL. Molecular biology of mycoplasmas. Annu Rev Microbiol. 1996;50:25–57.
- 46. Borghi N, James Nelson W. Intercellular adhesion in morphogenesis: molecular and biophysical considerations. Curr Top Dev Biol. 2009;89:1–32. pmid:19737640
- 47. Rozario T, DeSimone DW. The extracellular matrix in development and morphogenesis: a dynamic view. Dev Biol. 2010;341(1):126–40. pmid:19854168
- 48. Aman A, Piotrowski T. Cell migration during morphogenesis. Dev Biol. 2010;341(1):20–33. pmid:19914236
- 49. Morris SM Jr. Arginine: beyond protein. Am J Clin Nutr. 2006;83:508S-512S.
- 50. Zhou M, Martindale RG. Arginine in the critical care setting. J Nutr. 2007;137(6 Suppl 2):1687S-1692S. pmid:17513448
- 51. Fung TS, Ryu KW, Thompson CB. Arginine: at the crossroads of nitrogen metabolism. EMBO J. 2025;44(5):1275–93. pmid:39920310
- 52. Russell D, Snyder SH. Amine synthesis in rapidly growing tissues: ornithine decarboxylase activity in regenerating rat liver, chick embryo, and various tumors. Proc Natl Acad Sci U S A. 1968;60(4):1420–7. pmid:4299947
- 53. Albaugh VL, Pinzon-Guzman C, Barbul A. Arginine-Dual roles as an onconutrient and immunonutrient. J Surg Oncol. 2017;115(3):273–80. pmid:27861915
- 54. Chantranupong L, Scaria SM, Saxton RA, Gygi MP, Shen K, Wyant GA, et al. The CASTOR proteins are arginine sensors for the mTORC1 pathway. Cell. 2016;165(1):153–64. pmid:26972053
- 55. Saxton RA, Sabatini DM. mTOR Signaling in growth, metabolism, and disease. Cell. 2017;168(6):960–76. pmid:28283069
- 56. Liu C, Zhang Y, Wang Y, Wu M, Li Y, Wei J, et al. CASTOR1 and CASTOR2 respond to different arginine levels to regulate mTORC1 activity. Mol Cell. 2026;86(2):362-375.e4. pmid:41506264
- 57. Gerner EW, Meyskens FL Jr. Polyamines and cancer: old molecules, new understanding. Nat Rev Cancer. 2004;4(10):781–92. pmid:15510159
- 58. Xuan M, Gu X, Li J, Huang D, Xue C, He Y. Polyamines: their significance for maintaining health and contributing to diseases. Cell Commun Signal. 2023;21(1):348. pmid:38049863
- 59. Salje J. Cells within cells: Rickettsiales and the obligate intracellular bacterial lifestyle. Nat Rev Microbiol. 2021;19(6):375–90. pmid:33564174
- 60. Schön ME, Martijn J, Vosseberg J, Köstlbacher S, Ettema TJG. The evolutionary origin of host association in the Rickettsiales. Nat Microbiol. 2022;7(8):1189–99. pmid:35798888
- 61. Castelli M, Nardi T, Gammuto L, Bellinzona G, Sabaneyeva E, Potekhin A, et al. Host association and intracellularity evolved multiple times independently in the Rickettsiales. Nat Commun. 2024;15(1):1093. pmid:38321113
- 62. Min C-K, Yang J-S, Kim S, Choi M-S, Kim I-S, Cho N-H. Genome-based construction of the metabolic pathways of Orientia tsutsugamushi and comparative analysis within the Rickettsiales order. Comp Funct Genomics. 2008;2008:623145. pmid:18528528
- 63. Driscoll TP, Verhoeve VI, Guillotte ML, Lehman SS, Rennoll SA, Beier-Sexton M, et al. Wholly Rickettsia! Reconstructed metabolic profile of the quintessential bacterial parasite of Eukaryotic cells. mBio. 2017;8(5):e00859-17. pmid:28951473
- 64. Brandão AS, Borbinha J, Pereira T, Brito PH, Lourenço R, Bensimon-Brito A, et al. A regeneration-triggered metabolic adaptation is necessary for cell identity transitions and cell cycle re-entry to support blastema formation and bone regeneration. Elife. 2022;11:e76987. pmid:35993337
- 65. Kübler IC, Kretzschmar J, Brankatschk M, Sandoval-Guzmán T. Local problems need global solutions: the metabolic needs of regenerating organisms. Wound Repair Regen. 2022;30(6):652–64. pmid:35596643
- 66. Mahmoud AI. Metabolic switches during development and regeneration. Development. 2023;150(20):dev202008. pmid:37883063
- 67. Allen AE, Dupont CL, Oborník M, Horák A, Nunes-Nesi A, McCrow JP, et al. Evolution and metabolic significance of the urea cycle in photosynthetic diatoms. Nature. 2011;473(7346):203–7. pmid:21562560
- 68. Dzik JM. Evolutionary roots of arginase expression and regulation. Front Immunol. 2014;5:544. pmid:25426114
- 69. Zúñiga M, Pérez G, González-Candelas F. Evolution of arginine deiminase (ADI) pathway genes. Mol Phylogenet Evol. 2002;25(3):429–44. pmid:12450748