The purpose of the URoL:Epigenetics1 program is to enable innovative research and promote multidisciplinary education and workforce training in the broad area of epigenetics,2 with a focus on understanding the relationships among epigenetic mechanisms associated with environmental change, organismal phenotype, and resultant robustness and adaptability of organisms and populations. NSF published two calls for URoL:Epigenetics proposals in 2018 and 2019 and offered two submission tracks: Track 1 for projects with a budget up to $500,000 and duration up to 3 years, and Track 2 for projects with a budget up to $3,000,000 and duration up to 5 years. By December 2022, NSF had funded 29 PIs under this program.
The charge from NSF to URoL:Epigenetics researchers was as follows:
Successful projects of the URoL:Epigenetics Program are expected to use complementary, interdisciplinary approaches to investigate how epigenetic phenomena lead to emergent properties that explain the fundamental behavior of living systems. Ultimately, successful projects should identify general principles (“rules”) that underlie biological phenomena within or across scales of size, complexity (e.g., molecular, cellular, organismal, population) and time (from sub-second to geologic) in taxa from anywhere within the tree of life, including humans. URoL:Epigenetics projects must integrate perspectives and research approaches from more than one research discipline (e.g., biology, chemistry, computer science, engineering, geology, mathematics, physics, social and behavioral sciences). The interdisciplinary scope of URoL:Epigenetics projects also provides unique training and outreach possibilities to train the next generation of scientists in a diversity of approaches and to engage society more generally.2
The workshop to discuss projects funded under URoL:Epigenetics was held on January 26, 2023. PIs from 6 of the 29 funded projects participated in the live discussion. These PIs were joined by three moderators—Connie Mulligan, Kunal Rai (MD Anderson Cancer Center), and Alexander Gimelbrant (Altius Institute)—all of whom conduct multidisciplinary epigenetics research that combines field, clinical, or wet lab work with computational methods. Twenty-seven individuals watched the live webcast, and 44 have viewed the recording as of May 10, 2023.
Mulligan summarized the key goals of the workshop. Participants were first asked to share their scientific findings and to consider what broader rules of life these findings might illuminate. Secondly, they were asked to discuss how they incorporated a multidisciplinary, systems-level approach into their project, how it contributed to their achievements, and the challenges that arose along with ideas for addressing those challenges in the future. Noting that
___________________
1 See https://www.nsf.gov/funding/pgm_summ.jsp?pims_id=505582 (accessed January 31, 2023).
2 Epigenetics refers to chemical or structural modifications to a cell’s DNA that change gene expression without directly altering the DNA sequence. Non-coding RNAs can also play an epigenetic role. Epigenetic changes may persist throughout the lifetime of an organism, and some epigenetic changes can be transmitted to future generations. Epigenetic modifications are influenced by the environment. A cell’s or organism’s epigenome consists of all its epigenetic modifications.
that “team science is a topic of research in itself,” Mulligan added that the final, cross-cutting workshop would be devoted to that topic. The third focus of the workshop was to discuss societal issues that were being addressed by these projects, as well as emerging societal concerns that might be addressed in the future. Finally, participants were asked to reflect on the main focus of URoL, which is to identify rules by which biological systems are governed that are generalizable across fields and scales.
Rai read selected comments submitted in response to a pre-workshop questionnaire by PIs from 10 funded projects, including 4 projects not represented among the workshop participants. These comments have been incorporated into the sections below, which are organized by topic. Gimelbrant and Mulligan moderated the discussion.
Participants described their research projects and briefly outlined their scientific findings to date. A brief overview of these research projects is provided here.
Keith Slotkin (Donald Danforth Plant Science Center; University of Missouri) described results from his group’s study of the epigenetic response to high carbon dioxide (CO2) in plants, which is now in its fourth year. When plants are stressed, they establish gene expression programs through epigenetic modifications that persist into subsequent generations. When C3 photosynthetic plants, which constitute 85% of all plant species, are exposed to the particular stress of high CO2 (which increases the input for sugar production for photosynthesis), they grow at an accelerated rate. Slotkin’s group has found that subsequent generations grown from seed will continue that accelerated pace of growth, even when the level of CO2 is reduced; this response is strongest in the first generation and diminishes over time. Using a range of DNA methylation3 mutants, Slotkin and colleagues have identified two distinct genetic pathways responsible for either establishment or persistence of accelerated growth following exposure to high CO2. Both pathways are needed to propagate this phenotype to the next generation. However, while de novo methylation of the plant genome is randomly distributed among cytosines in the target region, maintenance of methylation depends on sequence context, and it is that maintenance pathway that enables long-term persistence of the phenotype, which may last for centuries. By studying different methylated regions in the plant genome and honing in on specific regions, the researchers seek to identify the gene or genes responsible for expressing the rapid growth phenotype across generations.
The phenomenon of persistent rapid growth in later generations following high CO2 exposure was observed in all C3 plants tested, including the Physcomitrium moss that is used as a model for evolutionary precursors of vascular plants. In addition, the differentially methylated regions of the genome that are important for this phenomenon are enriched for genes encoding phosphatases and phospholipases,4 which have been implicated in the propagation of responses to other stresses across generations. Slotkin and colleagues are now probing connections between methylation state, plant morphology, and the 3D organization of the genome.
___________________
3 DNA methylation is one type of epigenetic modification. By studying plants that are defective in DNA methylation, this group seeks to understand how environmental stress leads to long-lasting epigenetic change.
4 Phosphatases and phospholipases are enzymes that play a role in the cellular response to stress.
Hollie Putnam (University of Rhode Island) studies the tight nutritional symbiosis5 that exists between coral cells and single-celled dinoflagellate algae that live endosymbiotically within the coral. The algae provide greater than 100% of the daily metabolic energy demand for the coral to grow, to calcify, and to build the reef structure. “They’re the solar cells, the ecosystem engineers for the entire coral reef,” said Putnam. This symbiotic system is very fragile, and disruptions to energy metabolism can cause a breakdown. Metabolism regulates production of essential co-factors, which enzymes that carry out epigenetic modification need in order to function. Therefore, under conditions of metabolic and environmental stress, Putnam said “our expectation … is that not only energetically is it more complicated to regulate yourself, but you don’t have those same … co-factors for the modifiers. So, we should see environmental-energetic-epigenetic linkages … that underlie changes in genome function to generate the phenome.”
Coral is not a model system, its genome and epigenome are less well-characterized than many other taxa, and it has a generation time ranging from years to decades. Work performed 20 years ago showed that corals that bleached in one location did not bleach again in that same location (although they bleached on the other side) following a subsequent stress event.6 To probe the epigenetics of this response, Putnam and colleagues began by characterizing seasonal changes in the epigenetic state of coral as it responds to changes in light and temperature. These findings contribute to the research team’s overall body of work, which also includes findings from another study7 on the effects of environmental exposures to coral offspring. In this study, they exposed brooding corals to the high summer temperature and ocean acidification conditions associated with high levels of atmospheric CO2.
Looking at parents and F1 juveniles from multiple species, the researchers have found evidence of cross-generational epigenetic inheritance, with a positive relationship between DNA methylation and gene expression, accompanied by positive phenotypic effects that endure through time. However, as temperatures cool with seasonal changes, external and internal feedbacks should prevent the corals from maintaining the same high levels of gene expression, which are metabolically wasteful. In this context, Putnam and colleagues are now trying to understand how these dynamic systems successfully navigate frequent, seasonal temperature fluctuations while being extremely sensitive to longer-term temperature increases. In the process of modeling how the epigenetic state of corals depends on their environmental history, the researchers have produced new tools for predicting equilibrium states (referred to as hysteresis) that demonstrate dependence on the history of the system. They have also developed tools to model how interactions among coral symbionts can modulate the bleaching response.
Margaret J. McFall-Ngai (NAS, Carnegie Institution for Science) raised the question of whether epigenetic strategy varies depending on whether a marine animal is able to tolerate a wide (eurythermal) versus narrow (stenothermal) temperature range. More important than range is the predictability of change, said Putnam; the more predictable the change, the more consistent the signal, and the easier it is to use energy-efficient regulatory mechanisms. This has been studied in American Samoa, where two pools have fluctuating temperatures, but one fluctuates
___________________
5 Symbiosis refers to a mutually beneficial relationship between two organisms. In endosymbiosis, one organism lives inside the other.
6 Coles SL, Brown BE. 2003. Coral bleaching—capacity for acclimatization and adaptation. Adv Mar Biol 46:183-223. https://doi.org/10.1016/s0065-2881(03)46004-5.
7 Not funded by the NSF URoL program.
more than the other.8 Corals in the pool with highly fluctuating temperatures front-load gene expression, essentially as a regulatory mechanism to deal with that high fluctuation. This gene expression is itself likely to be regulated by epigenetic mechanisms such as methylation or histone modification,9 whichever is more energetically efficient. “In corals in particular, there’s more and more literature coming out that says, ‘your environmental history really matters, and high-frequency temperature fluctuations really matter to reducing your bleaching response,’” she added, hypothesizing that layers of regulatory control will keep certain sets of genes front-loaded when they see a persistent signal.
Another endosymbiotic interaction, in this case in insects, is being studied by Julie Hotopp (University of Maryland). When mosquitos are infected with Wolbachia bacteria, they are less likely to become vectors for RNA viruses, including viruses that impact human health, such as dengue, chikungunya, and Zika. Researchers are trying to harness this system to eliminate dengue and other diseases in the wild.
One mechanism through which Wolbachia endosymbiosis changes the transmission of RNA viruses is through regulation of RNA modification enzymes in the host insect. Hotopp’s collaborators demonstrated that Wolbachia endosymbionts in the fruit fly Drosophila modify expression of the host Dnmt2 gene, which produces a protein that methylates RNA; this alters the RNA methylation profile of the virus, changing viral infectivity and persistence in the insect.
Until recently, epitranscriptomics focused on modifications to rRNA,10 tRNA, and viral genomic RNA, because it was hard to purify sufficient amounts of mRNA for this kind of analysis, said Hotopp. Hotopp has begun to study mRNA modifications in the insect-Wolbachia system, but it is a very new area. One of her colleagues anticipates major deliverables at the end of the project in the form of “wet lab protocols … and bioinformatics pipelines for the analysis of epitranscriptomic data sets from nanopore sequencing.”
Many proteins are being uncovered that can write, read, or erase chemical modifications on nucleic acids in many organisms, said Lydia Contreras (The University of Texas at Austin). Some of these proteins’ structures have been analyzed, revealing binding pockets and motifs that are conserved across species. In collaboration with a computing group that models the physiochemical properties of molecules, Contreras is developing a high throughput screen to elucidate patterns and, based on these, predict biological interactions that can then be tested in the lab. One of her models, the YTH family of proteins, reads the m6A chemical modification of RNA11 and responds by directing various regulatory processes. There are at least five different
___________________
8 Barshis DJ, Ladner JT, Oliver TA, Seneca FO, Traylor-Knowles N, Palumbi SR. 2013. Genomic basis for coral resilience to climate change. Proc Natl Acad Sci USA 110(4):1387-1392. https://doi.org/10.1073/pnas.1210224110.
9 Another epigenetic mechanism, in addition to DNA methylation, includes histone modification and other means of regulating gene expression without changing the genomic sequence. This refers to chemical modifications on histone proteins. DNA winds around histone proteins, giving chromosomes their structure. Histone modification can change this structure and thereby alter gene expression.
10 Epitranscriptomics refers to all reversible chemical modifications on RNA. rRNA = ribosomal RNA, tRNA = transfer RNA, and mRNA = messenger RNA. Viral genomic RNA is the genetic material of an RNA virus.
11 m6A methylation is a common type of RNA modification that can regulate RNA transcription, translation, and stability.
YTH proteins with different length and sequence but enough similarity to search for homologs in diverse organisms.
With hundreds of m6A modifications discovered to date, it would be very inefficient for each one to have its own protein dedicated to reading it. Therefore, using broad in silico searches, Contreras and colleagues are asking what exactly the various YTH proteins recognize, how promiscuous the binding pockets are, and whether they are flexible enough to permit one YTH protein to operate in multiple pathways and with multiple chemistries. The researchers are also comparing proteins with erasing versus reading properties. They have computationally predicted that YTH readers can recognize a diversity of modifications, some of which may not be adenine-based, with different strength. “We’re starting to understand that there are some basic rules in the chemistry, but there’s also quite a lot of flexibility in those rules. Those deviations have become very interesting into what they mean as far as biology,” she said, adding that emerging patterns are starting to shed light on the determinants of specific versus nonspecific RNA-protein interactions.
Unlike many insects, bees are capable of facultative thermogenesis, generating heat by shivering, which enables them to remain active in cold temperatures. In the mountains of Oregon where Jeffrey Lozier (University of Alabama) collects queen bumblebees, high elevations have extremely cold winter temperatures compared to lower elevations (although the heat is comparable in the summer), and different populations of the same bee species can be found in both settings. When these queen bees are transferred to the lab, individuals reared from queens collected in high elevation environments demonstrate greater cold tolerance than those reared from queens collected at lower, warmer elevations. Since the critical thermal maximum does not vary much across elevations, bee populations from high elevations have broader thermal tolerance overall; this could be plastic, adaptive, or epigenetic, said Lozier, who is trying to tease out epigenetic correlations by analyzing DNA methylation and transcriptomics.
Lozier found that bumblebees respond very quickly to cold shock at the molecular level by changing patterns of DNA methylation, which correlates with changes in metabolite production. Whole-genome sequences have been obtained from 700 individuals, enabling the researchers to distinguish differences in methylation patterns resulting from small genetic differences, namely single nucleotide polymorphisms (SNPs) versus those that are independent of genetics. Compared to other organisms, bumblebees show little genetic variation between populations. This relative homogeneity is an advantage because it reduces the noise of nucleotide variation, said Lozier. Additionally, by sampling transects at high and low elevations from multiple mountains in Oregon, the researchers hope to control for differences in local environmental conditions.
The researchers have observed consistent variation in DNA methylation and transcriptional profiles before and after cold stress. Methylation differences between populations are “somewhat clinal” and seemingly not entirely driven by nucleotide variation, suggesting that “there may be a population structure to epigenetics that’s similar to nucleotide variation,” said Lozier. Parallel to Putnam’s observations in corals, methylation of bee DNA is highly restricted to gene bodies,12 occurring primarily in the second and third exons with a little in the first exon, and spiking again at the 3’ end of the gene.
___________________
12 The gene body refers to the portion of a gene that is transcribed to make RNA.
The yeast Schizosaccharomyces pombe can undergo epigenetic change through mis-regulation of its heterochromatin, which constitutes regions of chromosome compression and gene silencing. Heterochromatin mis-regulation initiates a series of changes that land on a novel “epigenetic island,” which enables the yeast to adapt, and once adapted, to convey this state heritably to many subsequent generations, said Kaushik Ragunathan (Brandeis University). Ragunathan and colleagues designed a system, using auxin-inducible degron13 approaches, to trigger this process in a precise manner and watch adaptation occur as a function of time. The researchers found that stress was a key determinant of successful epigenetic adaptation by yeast cells, and the only way to get the necessary gene activation or silencing to occur was via activation of stress pathways.
Using whole genome analysis approaches, including RNA-seq14 and ChIP-seq, Ragunathan and colleagues followed the spectrum of choices that cells made before they reached the final adaptive state that supported continued growth and proliferation. Single molecule imaging enabled time-dependent tracking of proteins as they moved inside cells and revealed intermediate steps, letting the researchers “essentially watch every decision that this organism made as a function of time,” as epigenetic complexes disassembled, relocalized to new locations in the cell, and reassembled to enable adaptation. Observing the dynamic process unfold within the cell, they discovered that large multi-protein complexes involved in gene regulation did not assemble in solution and then land on an inert chromatin scaffold; rather, chromatin was an active catalyst that enabled complex formation.
Certain themes recurred throughout the scientific presentations. Participants considered how these themes played out in their respective systems, and how they might reflect more generalizable rules.
A common theme considered by the PIs was whether epigenetic changes that resulted from stress, whether in the form of DNA methylation, RNA methylation, or 3D changes in chromatin architecture, were adaptive or maladaptive. Slotkin considered the persistence of rapid growth in plants following exposure to high CO2 an adaptive trait similar to the phenomenon of priming, by which a plant exposed to stress early in life is better able to handle a recurrence of the same stress later, and offered this broadly applicable rule based on his research: “The increased rate of plant growth in high CO2 is heritable; it continues even when the levels of CO2 decline. This is true broadly across C3-photosynthetic plants.” He noted that agricultural companies intentionally stress plants prior to harvesting their seeds to sell to farmers, generating crops that are more resilient than those derived from unstressed parents.
___________________
13 A molecular method for achieving rapid depletion of a target protein in vivo. For more information, see https://www.nature.com/articles/s41467-020-19532-z (accessed May 1, 2023).
14 RNA-seq and ChIP-seq are methods for studying gene expression and physical interactions within chromatin, respectively. For more information, see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6981605 (accessed May 1, 2023).
Like plants, corals are sessile and therefore highly vulnerable to environmental change since they cannot run away. This makes them more dependent than other animals on mechanisms that enable rapid adaptation, said Putnam. In corals, phenotype and epigenetic patterning change with the seasons. Phenotypes associated with particular environmental signals diminish across generations and across seasons as feedbacks disappear. Climate change, with increasing storms and heat waves, has brought increasing unpredictability to seasonal changes. On the one hand, coral has offered a ray of hope, because those organisms that do not die from bleaching recover with greater resilience to future bleaching events. “What doesn’t kill you makes you stronger, and you ratchet through these events,” said Putnam. However, she noted, “these events are always changing, and they’re overlaid on top of the seasonal variation,” complicating the task of adaptation.
Mulligan provided a perspective drawn from studying the health outcomes and DNA methylation states of infants whose mothers endured trauma during their pregnancy. There is a tendency to argue that since the plasticity of the fetus enables it to prepare for post-uterine life by taking in intrauterine cues, these changes in methylation must be adaptive. However, she said, it is only adaptive if the post-uterine environment resembles the intrauterine one. Experiments that match up negative intrauterine environments with both positive and negative post-birth environments are just beginning, but they reflect an openness to the notion that the response may be maladaptive in some cases, she added.
The question of whether stress-induced changes are adaptive or maladaptive is relevant to bumblebees and many other insects, said Lozier. In general, upper thermal limits do not vary as much across large geographic gradients as lower thermal limits, and with climate change, upper thermal limits are going to become higher and more similar across the board. As the temperature increases and the same upper limit prevails everywhere, adaptation to the extreme temperature variation of cold environments could become maladaptive, creating “some conflicts as genotypes start shuffling around,” he posited.
Timescale needs to be considered when evaluating whether a particular change is maladaptive versus adaptive, said Ragunathan. A change can be identified as maladaptive initially, he explained, “then when time passes, whatever the timescale is, maybe we have a better understanding of what adaptation actually looks like.” S. pombe divides on a timescale of a few hours, so adaptation can be observed, Ragunathan said, “as a function of time. Essentially, we’re able to both fast forward and rewind the adaptive clock in order to figure out how epigenetic adaptation works.” This allows researchers to observe the impact of a particular epigenetic change as it plays out over many generations, potentially changing their perception of what constitutes adaptation. Human perspectives on climate change and its perturbations on temperature, CO2, and ocean acidification also influence perception, said Putnam, noting that in some cases the cell is able to buffer the impact of change.
In every system discussed, stress is likely a key cue for epigenetic adaptation, and in at least four of these—plant, coral, human, and bee—DNA methylation may be a key epigenetic element driving it. A heavy reliance on methylation would not come as a surprise given the comparative instability of histones and other components of the 3D architecture of chromosomes, noted Rai. In addition, Putnam raised the possibility that transcription of a gene might modulate its epigenetic status so that its up-regulation is inherited.
Putnam asked participants for their thoughts on the nature of stress hardening and stress priming through epigenetic mechanisms and how these might differ in more predictable systems, such as agriculture, versus those in the wild. “What is stress … what is it that matters, the external environment or the internal physicochemical environment? And … is this a direct epigenetic modifier that’s passed as information, or is it a physical chemical stimulus from a metabolite changing the pH of the internal condition?” In formulating a rule, one should consider how direct the effects of the endosymbiont are on the host on an even broader scale, said Contreras. Does the symbiont employ epigenetic communication in a direct way to change host physiology, or is it acting through an indirect effect on the environment that triggers a host epigenetic response? The latter possibility is difficult to rule out, said Hotopp.
In S. pombe, Ragunathan found that the stress response pathway must be activated in order for cells to activate the epigenetic mechanism (in this case altered organization of chromatin) needed for adaptation. This circled back to the topic of adaptive versus maladaptive changes and led him to ask, philosophically, whether stress itself should necessarily be perceived as always bad.
Across multiple scales and taxa, epigenetic systems investigated by URoL researchers have revealed a great deal of flexibility. At the molecular level, Contreras found that the overall aromatic nature of YTH binding pockets is conserved across many species. At the same time, the rules governing binding of YTH proteins to m6A modification on RNA are flexible, facilitating a combination of specific and nonspecific RNA–protein interactions.
At the cellular level, Putnam noted that non-genetic cellular memory can take multiple forms, including not just DNA methylation but also modifications to histones and RNA as well as 3D changes within the nucleus. “Multiple mechanisms could [produce] the same outcome, even if the exact stressor, magnitude, timing, etc., is not the same,” she said, citing a recent review on the topic.15
Unicellular organisms have the most flexibility, said Ragunathan; “each cell can actually roll the dice in a way that multicellular organisms cannot,” because the latter require coordination among tissues and cells. “It fundamentally comes down to a sampling problem, and sampling with a unicellular organism means if a million of them die, there’s still a million left behind.”
As one moves across taxa, and from single cells to complex organisms, there is tremendous variation in terms of the way organisms sample the environment, the way they transduce and respond to these signals, and the way that environmental selection operates on them, said Putnam. Nonetheless, there may be clusters of taxa that employ similar epigenetic mechanisms, perhaps reflecting similar modes of sampling or selection. For example, Putnam and colleagues have seen the same epigenetic mechanism (methylation in the gene body) employed across multiple coral taxa, in contrast to the epigenetic patterning of many vertebrates, where DNA methylation preferentially occurs in promoters.
___________________
15 Adrian-Kalchhauser I, Sultan SE, Shama LNS, Spence-Jones H, Tiso S, Keller Valsecchi CI, Weissing FJ. 2020. Understanding “non-genetic” inheritance: Insights from molecular-evolutionary crosstalk. Trends Ecol Evol 35(12):1078-1089. https://doi.org/10.1016/j.tree.2020.08.011. This review makes the important points that “inherited epigenetic factors and DNA sequence are not distinct but functionally interdependent; … [that] processes such as DNA methylation are not uniform mechanisms, but operate in a multiplicity of ways depending on both species and mode of induction; [and that] most epigenetic variants are not deterministic … but probabilistic.” See https://pubmed.ncbi.nlm.nih.gov/33036806 (accessed February 7, 2023).
Hotopp noted that researchers have observed flexibility at the organismal scale, among endosymbiotic viruses, bacteria, and insects, which demonstrate great diversity in both the organisms themselves and in their responses. Mulligan suggested that it might be possible to formulate a rule reflecting the types of epigenetic variation used across different organisms; for example, DNA methylation could be particularly important for sessile organisms like plants or coral, while single-celled organisms and those that are motile, with the potential for more rapid environmental change, might take advantage of more transient epigenetic mechanisms. One might also expect to see intergenic variation in the persistence of epigenetic modifications through successive generations, suggested Mulligan. Epigenetic mechanisms represent “memory on a different scale,” said Gimelbrant, and different adaptions operating on different clocks might employ different epigenetic mechanisms for their purpose. In a more general sense, these epigenetic mechanisms may play a role “not just in adaptation in a short time but also in the ability of species to evolve at different speeds and scales,” he added, citing papers in Arabidopsis16 and fungi.17 Flexibility of the epigenetic response could even extend to the community level, suggested Hotopp.
Rai and Gimelbrant noted that many species do not undergo much DNA methylation, including insects and Caenorhabditis elegans, suggesting that alternative epigenetic responses to stress may prevail in these orders. Slotkin’s team is probing one such alternative in plants by using Hi-C18 data to model interactions within and between chromosomes, illuminating chromosomal architecture, which can modulate gene expression. The researchers are trying to correlate this 3D information with plant phenotype and its epigenetic persistence into the next generation, as well as methylome19 and transcriptome data. Taken together, the goal is to understand how environmental change affects the regulation of plant growth at multiple levels.
In yeast, heterochromatin is an epigenetic modification capable of dynamic change and adaptation in response to stress, and the adaptive state can be conveyed to daughter cells for many generations. In addition, Ragunathan and colleagues have found that chromatin plays an active role in catalyzing the assembly of epigenetic complexes.
Hotopp noted that her research involving RNA modification does not fit neatly into the DNA-based epigenetic rubric.
Several presenters described systems in which one cell regulates gene expression in another via epigenetic mechanisms. In C3 plants, the nurse cell responsible for nourishing the germ cell has a dynamic transcriptome, “probably a lot of RNA modifications,” and it generates lots of small RNAs, said Slotkin. “We think those small RNAs are essentially being showered
___________________
16 Queitsch C, Sangster TA, Lindquist S. 2002. Hsp90 as a capacitor of phenotypic variation. Nature 417(6889):618-624. https://doi.org/10.1038/nature749.
17 Aramayo R, Selker EU. 2013. Neurospora crassa, a model system for epigenetics research. Cold Spring Harb Perspect Biol 5(10):a017921. https://doi.org/10.1101/cshperspect.a017921.
18 Hi-C is a method for studying the 3D conformation of genomes. For more information, see https://pubmed.ncbi.nlm.nih.gov/22652625 (accessed May 1, 2023).
19 The methylome is the set of all methylation modifications throughout the genome of a cell or organism. The transcriptome is the sum of all messenger RNAs.
down on the gametes, trying to establish modifications, trying to establish gene expression … to potentially silence transposable elements from the other gamete. We don’t really understand, but there’s this reprogramming that’s going on and it’s not from erasure in the gametes, it’s erasure in that nurse cell that’s right next to it.”
In the endosymbiotic systems studied by Putnam and Hotopp, a symbiotic microbe sends an epigenetic signal that modulates function in the host organism. In their metabolic role, the algae studied by Putnam regulate production of co-factors that are essential for epigenetic modifications within their coral host; these co-factors can transmit information about the environment to the host, with algae as the messenger.
Hotopp described a three-player system, where a bacterial endosymbiont regulates expression of a methyltransferase gene in the host insect, which then alters the RNA methylation profile of the virus. “With respect to the rules of life, this is possibly a mechanism by which microbes can alter their host, but also alter other microbes that are in the host as well,” said Hotopp. Rai suggested this could be a focus for future research. “We should understand how external organisms might be communicating to the host epigenome … that might be a novel area of epigenetic communication between different biological systems,” he said.
Hotopp and Putnam considered how their observations in endosymbiotic systems might translate to the wild and at scale. Regarding the host-microbe interactions that she studies in mosquitoes, Hotopp noted that both Wolbachia and RNA viruses undergo rapid diversification in the host insect, further complicating matters. “How much of that diversity is due to the epitranscriptome, we don’t have any feel for it at all … this is just too new of a finding,” she added.
In coral, the holobiont consists of the host along with its dinoflagellate, bacteria, viruses, and fungi; every one of these organisms alters the physicochemical environment depending on its morphology, noted Putnam. Given the short generation times of endosymbiotic microbes, “some are going to be evolving and adapting to be a different symbiont, but [living in] the same sort of tissue in the host.” Without understanding how these signals are perceived within the endosymbiotic system, it will be hard to formulate the rules governing how signal transduction modulates gene regulation, she added.
Although the evidence is currently lacking, Slotkin expressed his belief that his findings concerning the heritability of plant growth in high CO2 will scale to larger biomes, such as forests and grasslands. These biomes, he added, are “built out of plants, and as the levels of CO2 rise globally, plants adjust and their growth is affected.”
Several PIs noted that, while their results might not point to a particular “rule of life,” their work is aimed at advancing knowledge toward this goal. One wrote that their “modeling framework is serving as a quantitative tool to study biological systems … laying a foundation to dissect the mechanism of transcriptional silencing by ncRNAs.” A second researcher stated their goal “to identify rules that govern how RNA and DNA are modified in eukaryotic cells and the consequences for cell biology.” A third wrote that, while it is “premature to declare rules,” they are “making progress toward understanding rules that govern selection for phenotypic plasticity, including via epigenetic mechanisms; describe constraints on this plasticity; determine the
mechanisms of epigenetic modification; and govern resource exchange in mutualisms.” Additional rules in progress are mentioned throughout this proceedings.
Having observed that metabolite production occurs simultaneously with major transcriptional changes in bees recovering from cold shock, Lozier noted that he anticipates the possibility of relating specific metabolites to changes in specific genes, once his complete data set has been assembled.
PIs described examples of the challenges posed by multidisciplinary research, such as differences in lexicon, experimental methodologies, and theoretical approaches among scientists across disciplines. PIs highlighted strategies they have employed to overcome these challenges.
Several PIs said that keeping all the collaborators on the same page, or “herding cats,” was a challenge, which they are addressing through increased communication and relationship building within their groups.
One PI wrote that communication among team members was greatly aided by proximity, which enables important ad hoc hallway conversations. Communication can also be strengthened by regular group meetings and sub-meetings, sharing research folders on the cloud, and efforts on the part of project leaders to set goals and timelines with frequent check-ins. One team formed subgroups that met individually and then came together to report out and discuss integration. Many respondents mentioned that routine (monthly) meetings among PIs, often aided by tools like Zoom or Slack, are helpful for maintaining project integration. Increasing the frequency of in-person meetings “immediately changed our productivity,” wrote one PI. Although several PIs found that distance made collaboration more difficult, particularly in those instances when a member of the team switched institutions mid-grant, one PI, impressed by the ease of long-distance collaboration once he had adjusted to “Zoom life,” encouraged his trainees outside the project to form similar collaborations.
Two different PIs credited a track record of successful collaboration among their team members. “The most important factor in any successful collaboration is that each of the individual people like each other and get along well,” wrote a third PI, adding, “we had a shared sense of our scientific mission (and strongly overlapping interests).” A fourth cited the importance of “mutual trust across research teams, a commitment to helping and supporting each other, and our desire to find strong connections across our research areas. The collaborations were intentional.” Several PIs praised the diverse expertise of team members as critical for their project’s success.
Many researchers cited the diverse, field-specific perspectives afforded by multidisciplinary research as a major advantage of this project (see Figure 4-1). Nonetheless, “because of the lack of common vocabulary … you spend a large proportion of the grant just figuring out how to communicate,” said Putnam. This can be exacerbated when the biological
system under investigation lacks the known features of a model organism; examples from this discussion include coral, bees, and the insect-parasite-virus nexus.
For one PI, the different backgrounds of researchers made it difficult to reach an understanding of how the questions were formulated: “biological relevance is often not the first line of thinking for computational biophysicists,” they wrote, noting difficulties “reconciling in vitro assays with in vivo behavior” and replicating in silico results “with biological relevance.” A modeler described the challenge of analyzing data from non-model systems that lack standardized pipelines and pose species-specific challenges. This individual’s team had to rethink the assumptions brought to data collection and is “trying to figure out how to reduce the number of factors without oversimplifying the study.” One modeler wrote that “understanding the limitations of computing techniques is not always obvious … for experimentalists.” “As always, it was difficult to get the computer scientists and mathematicians to understand the biology,” said another PI, whose team created vocabulary lists to aid comprehension.
Efforts to form bridges between the various disciplines are most successful when they are “baked into the proposal,” said Mulligan. Teams used various approaches to help build bridges. One approach is to send students to work in different physical lab spaces. When a chemistry PhD student spent time in a biochemistry lab, “the conversations moved from being explanatory to being emergent in content,” wrote one PI. Another approach is to leverage areas of overlap.
Aware that the main challenge posed by multidisciplinary research is “getting to a common language for the integration across scales from empirical to theoretical,” Putnam and colleagues assembled a team of scientists with somewhat overlapping expertise, including “translator scientists” selected for their ability to span disciplines. These translators are able to rapidly share field-specific advancements with the rest of the group, which has worked well, she said.
Some teams used systems-level approaches capable of integrating data from diverse sources to bridge across disciplines. One example of this is a tool employed by Putnam’s team called a Dynamic Energy Budget (DEB) model, which represents the flows of energy and matter within an organism. The researchers are using the DEB model to generate a systems-level quantitative framework that can link epigenetic mechanisms to organismal and ecological performance. Contreras’s team has developed new tools or has optimized existing ones to form these bridges among the systems she studies, which have not been extensively studied either experimentally or computationally. Slotkin’s work incorporates dynamic 3D imaging of plants and Hi-C analysis of DNA structure to elucidate the genotype-phenotype connection related to changing growth rates at multiple levels of organization. “That goes beyond what I can do as a trained bench scientist, into the world of AI [artificial intelligence] and computer science and mathematical modeling,” he noted.
Participants discussed several additional factors that have contributed to the challenges of engaging collaborators, with multiple PIs citing the COVID-19 pandemic as having negatively affected team interactions. Lozier said that field research has been particularly challenging in recent years due to a combination of COVID-19 pandemic, which eliminated an entire field season, and climate change, which is making it harder to retrieve fresh queens. Delays in
fieldwork decoupled the timing for collaborators in research laboratories, leaving the chemists waiting to get samples and the computational biologists waiting for data to analyze. There are inherent difficulties to working with a field-based, non-model system, Lozier added, and these can make it difficult to keep the new PIs—those who offer interdisciplinary components, such as chemistry and computational science, but are not as invested as the bee specialists—engaged in the project.
The COVID-19 pandemic presented challenges to both students and faculty, noted Contreras, though Ragunathan wrote that focusing on supporting the trainees helped overcome some of this. Slotkin said that the pandemic was particularly tough on interdisciplinary projects, with some collaborating laboratories narrowing their focus to bread-and-butter projects and leaving the interdisciplinary URoL project by the wayside. “It’s everybody’s side thing besides the lead PI, which has my name down and my reputation at stake,” he said. Another researcher wrote that “because COVID inhibited our ability to meet together in person with regularity, we were especially prone to go off in different, discipline-specific directions.”
Based on their experiences, several participants proposed a few modifications to the structure of URoL grants to better support multidisciplinary team research.
Citing the challenges of cultivating a shared vocabulary and training diverse scientific perspectives on a common goal, one PI suggested that budgets could be scaled for a 7-year (as opposed to 5-year) timeline. “Projects that have to be built across different disciplines and labs need more time,” they wrote. Contreras agreed, stating that “when you’re working with multiple teams, multiple languages, [and] multiple disciplines, that standard NSF 3-year project model might be something we want to rethink.” Collaborating across institutions posed bureaucratic challenges for another PI, complicating tasks like managing budgets or sharing large data sets. Multidisciplinary research “absolutely takes more time … that has to be baked into the timeframe of the grant” rather than relying on no-cost extensions, said Mulligan.
As academic researchers, “we’re not farming out things to collaborators,” added Ragunathan; one of the key goals is training, and the “meta point” of having interdisciplinary teams is to train researchers in multiple disciplines. He said that this represents “a really new way of trying to think about doing science … [and] time is an important resource in realizing the full benefits of such a team.”
Grant money should be provided as a lump sum up front, suggested one researcher. This could provide flexibility for the team to decide how to spread it out over the years, for example, by funding data collection more heavily up front and data analysis later on. It could also potentially ease the difficulties encountered when PIs change institutions, which created budgetary problems for one team.
“The way the money was distributed creates difficulties for leadership,” wrote another lead PI. Each co-PI on their team received separate funding and the data analysis team began to draw on their grant money before data were available, using it to finish other projects that were short on funds due to the pandemic. Under the current funding arrangement, the lead PI has no ability to shift resources or pull funding from a group that is not performing the intended work.
Learning about other teams’ achievements is “exciting and stimulating,” said Putnam, who suggested extending the idea of translator scientists to forge connections among different projects, with a centralized location to share ideas and emerging rules of life. Facilitated interactions between funded teams could enable groups to share technology and methods and establish a cadre of “translators across what’s possible in science, even if your particular system isn’t quite there yet,” she added. Workshops bringing together multiple groups are a “missed educational opportunity” that could motivate and inspire graduate students, wrote Contreras.
Mulligan noted that large center grants from the American Heart Association include travel funds, not only for travel within groups but also for researchers from different centers to travel to one another. The grant review process even includes a meeting to evaluate the collaborative potential of the PIs. “That seemed to me to be very supportive of what you need to pull off this kind of multidisciplinary research,” she said.
“We would benefit from machine learning and artificial intelligence approaches to examine the seemingly endless number of interactions between epigenetics, environment, gene expression, and their feedbacks … through time,” said one PI, noting that current hypotheses connecting epigenetic state to gene regulation “only explain a fraction of the variance” and discoveries could be aided if they had the power to take nontargeted approaches to data collection and analysis.
Participants briefly discussed the impacts of their projects on society, which took two forms: obtaining the necessary knowledge to solve problems important to humankind; and informing the public about these biological concerns.
Several researchers drew a direct connection between their projects and the need to understand and ameliorate the effects of climate change on the ecosystem. “In a world where environments and ecologies are constantly under threat due to rapid climate change,” wrote Putnam, “our work provides frameworks for projecting what adaptive processes might look like in single celled and multicellular organisms in this uncertain future,” with the potential for “informing and providing direct conservation and restoration applications to improve reef ecosystem function for society … [such as] … to harness the inherent biological features of corals for applied approaches of stress hardening for reef restoration.” One of Putnam’s co-PIs
wrote that “we are better prepared to understand marine invertebrate epigenetics more broadly, including a number of species that are important food resources.”
Slotkin noted that “accelerated growth … puts stress on the fertilization, pest resistance, and other aspects of crop growth.” His results suggest that, even if atmospheric levels of CO2 could be reduced, plant growth rates would not immediately return to normal. Lozier’s work “has direct impacts on understanding biodiversity threats to beneficial insects that will result from climate change,” he wrote, and “understanding how species adapt to variations in cold and how those changes are inherited is important to species conservation.”
A colleague of Hotopp noted that their team’s work could contribute to global biocontrol efforts to use Wolbachia endosymbionts to eliminate pathogenic mosquito-borne RNA viruses. Beyond that, their findings “could have broad implications in host-microbe interactions,” and learning how to manipulate, modulate, or otherwise exploit host-microbe interactions could prove useful for a variety of societal needs relating to human health or agriculture.
Several investigators cited the contribution of their work to public understanding. “This project raises awareness about global change and the impact on coral reef ecosystems worldwide,” wrote a co-PI of Putnam. One team developed a video game in which players make predictions about the behavior of cells under warming conditions, with the aim of encouraging people to rethink their assumptions. Lozier’s team is deploying a video game to broaden pollinator education in grades K-12, highlighting aspects of bumblebee life history, thermal stress adaptation, and threats from climate change.
Participants discussed their strategies for mentoring students and other trainees engaged in multidisciplinary research, with the goal of developing a future generation of researchers who approach scientific inquiry in a way that crosses scales and scientific disciplines.
“The big challenge for students is the intellectual barrier between biological and quantitative sciences,” wrote one PI, who addresses this by fostering an inclusive laboratory environment; providing highly involved, hands-on training to graduate students; establishing effective channels of communication with constructive feedback; and encouraging collaboration among students, so that they can learn from one another. One mentor reported engaging students through active learning and critical thinking, while another noted that the experience of mentoring multiple trainees also helps train the mentor.
“Students need to understand the language of the other researchers on the team,” wrote a PI. In one group, students used Google and online sources to learn the other field’s terminology. As with the research itself, training students across institutions and disciplines “takes coordination and frequent meetings,” noted another PI. In large joint meetings, students on this team are taught to present their own work in a way that is both concise and broadly understandable.
Other respondents noted that the nature of the project forces trainees to think beyond their own fields. As previously mentioned, working in a collaborating laboratory stimulated students’ abilities to formulate questions across disciplines. One PI wrote, “The students … were forced to confront research in areas outside their comfort zone. The hope is that they get to launch their own careers by continuously challenging their own knowledge boundaries.”