The past, present, and future of coral heat stress studies

Abstract The global loss and degradation of coral reefs, as a result of intensified frequency and severity of bleaching events, is a major concern. Evidence of heat stress affecting corals through loss of symbionts and consequent coral bleaching was first reported in the 1930s. However, it was not until the 1998 major global bleaching event that the urgency for heat stress studies became internationally recognized. Current efforts focus not only on examining the consequences of heat stress on corals but also on finding strategies to potentially improve thermal tolerance and aid coral reefs survival in future climate scenarios. Although initial studies were limited in comparison with modern technological tools, they provided the foundation for many of today's research methods and hypotheses. Technological advancements are providing new research prospects at a rapid pace. Understanding how coral heat stress studies have evolved is important for the critical assessment of their progress. This review summarizes the development of the field to date and assesses avenues for future research.

The El Niño Southern Oscillation (ENSO) incident of 1998 was the first mass bleaching event recorded by the Hotspot program of the US National Oceanic and Atmospheric Administration (Liu et al., 2014;Strong, Barrientos, & Duda, 1996). The program predicted, weeks in advance, which geographical regions would experience bleaching due to increased sea surface temperatures. These findings supported the growing evidence that climate change was having severe impacts on marine ecosystems, fueling the need for a better understanding of coral symbiotic relationships. The cumulative observations of the ENSO events in the 1980s and 1990s revealed that the consequences of temperature stress and coral bleaching were much greater than imagined. These consequences include increased mortality, decreased reproduction, reduced reef productivity, and changes in community structure (Hoegh-Guldberg, 1999). Nonetheless, growing reef monitoring efforts have revealed potential adaptation and acclimatization strategies of corals and their symbionts, providing hope for their survival under a changing climate. This has moved scientists into a new avenue of research and provided a novel outlook on coral conservation.
The growing recognition of coral reefs' high socioeconomic importance (Black & Bloom, 1984;Carte, 1996;Hoegh-Guldberg, 1999;Jameson, McManus, & Spalding, 1995) has propagated interest in understanding reef systems, their functions, and how to aid in their survival. Our understanding of coral-algal symbiosis and environmental stress responses has progressed significantly, with rapid technological advancement enabling further insight into the complex dynamics of this relationship. While physiological measurements were, and are, fundamental in understanding coral's responses to stressors, understanding underlying molecular mechanisms of coral symbiosis, acclimatization, and adaptation is critical if we want to aid coral reef survival. This review assesses coral heat stress studies and their development over time. Based on the past and present progress, we provide suggestions on the future directions of this field.
Understanding mechanisms and indicators of thermal susceptibility has thus been a central focus of coral heat stress studies. The early development of stress response biomarkers enabled a monitoring system and provided a basis for comparison.

| Proteins: the first molecular insights
The first biomarkers to be confidently established in coral heat stress studies were heat-shock proteins (HSPs). Prior heat stress studies in other cnidarians, such as Hydra (Bosch, Krylow, Bode, & Steele, 1988), Anemonia viridis (Miller, Brown, Sharp, & Nganro, 1992), and the jellyfish Aurelia aurita (Black & Bloom, 1984), showed the presence and the increased abundance of HSPs in thermal stress response. The first 70-kDa HSP homologue in corals was characterized in Goniopora dijboutiensis (Sharp, Miller, Bythell, & Brown, 1994). Additional studies on other coral and anemone species extended the repertoire of HSP homologues, establishing them as viable biomarker candidates (Black, Voellmy, & Szmant, 1995;Branton, MacRae, Lipschultz, & Wells, 1999;Wiens et al., 2000). Thermal stress has also been shown to increase reactive oxygen species (ROS) in corals and algal symbionts, consequently triggering antioxidant mechanisms. Antioxidant pathways control cell-level toxicity and, thus, cellular stress and damage during stress events. Antioxidant-related protein's ability to maintain ROS levels at nontoxic levels made them interesting targets for biomarker development. Although proteins such as superoxide dismutase (SOD), catalase (CAT), and ascorbate peroxidase (ASPX) had been shown to play critical roles in coral's antioxidative stress response (e.g., Dykens, Shick, Benoit, Buettner, & Winston, 1992;Lesser & Garcia, 1997;Lesser, 1996;Lesser et al., 1990), their use as biomarkers only developed in the early 2000s.
Proteins known to be important for antioxidant mechanisms such as B-crystallin, copper/zinc SOD (Cu/ZnSOD), manganese SOD (MnSOD), ubiquitin, lipid peroxide (LPO), and total glutathione (GSH) were detected and measured in O. faveolata (Downs, Mueller, Phillips, Fauth, & Woodley, 2000). Proteins showed a higher abundance in heat-and light-stressed corals under laboratory conditions as well as during natural bleaching events (Downs et al., 2000;Lesser, 1996). These findings confirmed the coral host was experiencing oxidative stress, as a consequence of the symbionts' damaged photosystem II (PSII). Additionally, ROS was shown to compromise host cell integrity and consequently induce bleaching (Hoegh-Guldberg, 1999). Follow-up studies showed that oxidative stress caused by symbionts also triggered host nitric oxide production through the activation of NF-kB, leading to cell death and bleaching (Perez & Weis, 2006).
An increased repertoire and understanding of these biomarkers enabled the comparison of stress responses between coral species.
Interspecies comparisons revealed ambient HSP70 levels differed between two species and multi-HSP expression was an indicator of improved thermal response (Robbart, Peckol, Scordilis, Curran, & Brown-Saracino, 2004). However, variations in thermal tolerance were observed not only between species but also within species and even within single colonies (Brown, Downs, Dunne, & Gibb, 2002;Brown, Dunne, Goodson, & Douglas, 2000;Cook, Logan, Ward, Lcukhurst, & Berg, 1990;Goreau & Macfarlane, 1990;Jokiel & Coles, 1990). A single coral colony can grow into a large structure where parts of the colony may experience differences in competition, light regime, and temperature. Thus, single large corals can often experience a variety of conditions. Use of both HSP and oxidative stress-related biomarkers showed that large coral colonies had significant differences in thermal stress experience across sections (Brown et al., 2002). Further studies focused on understanding tissue-specific expression of biomarkers (Richier et al., 2003), their general characterization (Plantivaux et al., 2004), and potential role in symbiosis (Richier, 2005). These studies contributed to the gen-  provided insight into new pathways involved in heat stress response and stress mitigation, such as actin (cytoskeleton structure), ferritin (oxidative stress), ribosomal proteins (protein biosynthesis), and Rab7 (membrane trafficking). With the combined efforts of these initial microarray studies, genes were identified from a variety of cellular pathways, providing evidence that the breakdown of symbiosis was a result of multiple interactions (Dunn, Schnitzler, & Weis, 2007). Nonetheless, microarrays had their limitations in that only a number of known sequences could be studied. The development of next-generation sequencing (NGS), which allowed the total mRNA F I G U R E 1 Number of publications on cnidarian heat stress response using -omics tools. Publication records with keywords were recorded from Web of Knowledge and plotted according to year of publication (https ://apps.webof knowl edge.com/). Since the development of the first cnidarian microarray in 2005 and technological advancement of various molecular sequencing platforms, the application of -omics tools has increased steadily. Keywords used to determine studies were separated into independent variables (or) within three categories donated by (and). The keywords used were as follows: "heat stress or temperature stress or thermal stress" and "coral or anemone or anthozoan" and "gene expression or transcriptome or transcriptomics or proteomics or genomics or genome" content of an organism (the transcriptome) to be sequenced, overcame these shortcomings and provided new insight into the molecular layers of organisms.

| The rise of transcriptomics
Whole-genome, transcriptome, and proteome sequencing, collectively known as -omics tools, have opened the field to new possibilities, hypotheses, and information regarding heat stress resilience of coral holobionts. The possibility of whole-mRNA sequencing propelled transcriptome studies in a variety of corals (Meyer et al., 2009;Schwarz et al., 2008;Traylor-Knowles et al., 2011) as well as in Symbiodiniaceae species (Barshis, Ladner, Oliver, & Palumbi, 2014;Bayer et al., 2012;Rosic et al., 2015). Though not comprehensively discussed in this review, a growing number of studies are applying proteomics in cnidarians (Barneah, Benayahu, & Weis, 2006;Cziesielski et al., 2018;Drake et al., 2013;Oakley et al., 2016;Ramos-Silva et al., 2013). Omics methods were previously only utilized in a handful of studies, but are currently one of the most common tools applied in the field ( Figure 1) and thus the primary focus of the following sections.
Although research endeavors have mainly focused on the cnidarian host and algal symbiont, there has also been growing recognition of another important holobiont component: the microbiome.
Evidently, thermal resilience cannot solely be attributed to only one of the members of the holobiont. However, since this review focuses on the cnidarian host, we have only briefly touched on the other members. Growing evidence suggests that the host genotype is capable of local adaptation and acclimation Hawkins, Krueger, Wilkinson, Fisher, & Davy, 2015). Host genotype response variability is particularly important, as studies have shown improved tolerance in corals with previous exposure, indicating that resilience may be heritable (Dixon et al., 2015;Howells et al., 2016). The concept of pre-exposure has gained increasing attention in recent years, as it may provide a crucial platform for coral survival in light of global change.

| Learning from experience: Life history and pre-exposure to stress provide platforms for coral resilience
Environmental history can significantly impact coral's response to elevated temperatures and their overall tolerance to extreme events (Hawkins & Warner, 2017;Krueger et al., 2017;Rivest, Kelly, DeBiasse, & Hofmann, 2018). The hypothesis that prior heat exposure could improve a coral's response to follow-up stress events was proposed early on (Coles & Jokiel, 1978;Jokiel & Coles, 1977;Middlebrook, Anthony, Hoegh-Guldberg, & Dove, 2010;Middlebrook et al., 2008). Experiencing sublethal doses of thermal stress can provide a new acclimated baseline for subsequent stress events, by setting in place physiological and molecular mechanisms crucial in heat stress response (Ainsworth et al., 2016;Berry & Gasch, 2008). These observations indicate corals' potential to acclimatize to new environmental conditions. A recent large-scale observational study, based on a model of the Great Barrier Reef's sea surface temperatures (SST), showed that prestress events occurred prior to the main stress, serving as a physiological preparation (Ainsworth et al., 2016). This study further validated its observations with in situ heat stress studies on Acropora aspera, reporting significant differences in gene expression profiles between pre-exposure and control conditions. Not only can preconditioned corals show transcriptional differences (Barshis et al., 2013;Bellantuono, Granados-Cifuentes, Miller, Hoegh-Guldberg, & Rodriguez-Lanetty, 2012;, but they also have the capacity to maintain higher Symbiodiniaceae densities under stress (Bay & Palumbi, 2015;Palumbi et al., 2014). Additionally, some studies suggest that preconditioned corals could potentially utilize the same genes but achieve larger magnitude in gene expression change (Bay & Palumbi, 2015;Kenkel & Matz, 2016).
A growing number of studies suggest that epigenetic mechanisms may play critical roles in the acclimatization process of corals.
Epigenetic modifications such as DNA methylation, the addition of methyl groups to specific sites on a genome, and histone modifications, packaging proteins that bind DNA to condense it into chromosomes, are currently severely understudied in corals (Eirin-Lopez & Putnam, 2019). Differential expression of transcripts can be a consequence of changes in DNA methylation distribution in response to stressors (Dixon et al., 2015). Hence, changes in DNA methylation sites have been linked to transcriptional plasticity, which may facilitate response mechanisms to a previously encountered stressor (Dimond & Roberts, 2016;Liew et al., 2018;Putnam, Davidson, & Gates, 2016). Although complex gene regulation through histone modifications is conserved in cnidarians (Schwaiger et al., 2014), knowledge regarding the role of histones in coral acclimatization and adaptation is lacking. Epigenetics and preconditioning appear to be promising mechanisms for coral adaptation and survival. However, it requires a significantly greater understanding before these mechanisms can be successfully utilized to their full potential.

| WHERE S HOULD WE G O?
Potential directions for future work are plentiful. Living in the -omics age also means continuous possibilities to venture into new research avenues. Nonetheless, considering the past and present progress of coral heat stress studies, certain subjects stand out, which will require significant attention if we hope to increase our understanding of coral thermotolerance, and aid in their survival.

| Reference genomes and model organisms
High-quality reference genome assemblies are the key to informative molecular genetic studies. In general, the availability of reference genomes will also promote venturing into new fields of interest such as comparative genotyping and epigenetics. Hydra magnipapillata (Chapman et al., 2010) and Nematostella vectensis (Sullivan et al., 2006) genomes have been stable reference points, providing many evolutionarily conserved cnidarian genes that could be utilized in transcriptomic studies. However, unlike corals, neither of these two cnidarians associate with endosymbionts of the family Symbiodiniaceae. Thus, it was necessary to develop high-quality genomes for corals. The first coral genome of Acropora digitifera (Shinzato et al., 2011) initiated studies on conserved mechanisms and an estimation of the depth of divergence between corals and other cnidarians. However, A. digitifera lies in the complex clade of the scleractinians, thus only representing a portion of corals.
Phylogenetic analyses of robust and complex corals indicated that these clades separated at least 245 mya (Simpson, Kiessling, Mewis, Baron-Szabo, & Müller, 2011), leaving enough time for divergence and the evolution of clade-specific traits and adaptations. For some time, there was a severe lack in robust clade coral genomes that was only recently remedied. The Stylophora pistillata genome provided the first genomic resource for the robust clade .
Reference genomes of a diverse range of corals will provide further insight into their biology and enable the development of new molecular tools. Yet, only four fully sequenced genomes are currently available (Acropora digitifera (Shinzato et al., 2011), Acropora millepora (Ying et al., 2019), Pocillopora damicornis (Cunning, Bay, Gillette, Baker, & Traylor-Knowles, 2018), and Stylophora pistillata ). The Reef Future Genomics Consortium  recognized the urgency of this problem. They defined a set of 10 coral species for which to investigate physiological differences and identified a framework of molecular datasets that are anticipated to provide new insight into coral's adaptive capabilities.
Although development of new reference genomes is required, the progress of these needs to occur simultaneously with the optimization and development of genetic tools for existing sequenced genomes.
While there is a strong interest in making more coral genomes available (Liew, Aranda, & Voolstra, 2016), there is also the proposition of a coral model organism , such as the small sea anemone Aiptasia pallida (sensu Exaiptasia pallida). Having a model organism allows stronger international efforts to gain an integrative understanding of cnidarian biology by allowing studies to be combined and directly comparable. Additionally, working on the same established model organism could speed up the development of molecular tools.

| Integrative analysis and secondary validation
There is no doubt that transcriptomics has provided invaluable insight into stress response in corals. However, the main limitation of transcriptomics is that it does not necessarily reflect the physiological response. Hence, coral heat stress research requires molecular and physiological measurements to be incorporated together to fully understand thermotolerance.
Formation of mRNA is only the first step in a long chain of regulatory mechanisms leading to the final protein (Baumgarten et al., 2018). Through these, a single mRNA can potentially translate into thousands of proteins and be controlled by a number of regulatory mechanisms at posttranscriptional and posttranslational levels. The analysis of mRNA is seldom a representation of the protein content in the organism, which is frequently reflected in the poor correlation reported between mRNA and protein expression Griffin et al., 2002;Lee et al., 2003).
Discrepancies between the transcriptome and proteome cause concern not only for interpretation of data but also for the development of new biomarkers. Whereas previous biomarkers were established based on protein extraction and identification, current markers are suggested predominantly on the presence of mRNA. In particular, combined transcriptome-proteome approaches have the capacity for complementing one another (Seliger et al., 2009), allowing for data integration to provide a better understanding of a system or its current situation (Gomez-Cabrero et al., 2014). A narrow assortment of papers utilizes proteomic analysis to elaborate on fundamental coral biology such as symbiosis, larval development, and calcification (Barneah et al., 2006;Drake et al., 2013;Oakley et al., 2016;Ramos-Silva et al., 2013), but work related to proteomic stress response in corals is sparse Matthews et al., 2017;Weston et al., 2015). While technology and analytical tools are quickly progressing, large-scale studies on proteins are not as feasible as for nucleic acids (Graves & Haystead, 2002). Secondly, the application of proteomics is not as standardized as that for mRNA-seq.
Methods are being developed for integrative analysis of multi-omics data to illustrate more comprehensive pictures of the molecular systems (Bersanelli et al., 2016). Achieving data integration is a difficult challenge that has not been simplified by the rapidly increasing amount of data.
With the growing use of -omics tools, it is important that the targeted biological question should drive the use of these tools instead of embarking on a frenzy of large-scale sequencing. Those that choose to focus their work on -omics should consider physiologically validating their observations. One omic layer might not represent the other, but if the phenotype does not support molecular findings, a reassessment of conclusions drawn may be advisable.
Additionally, -omics users should consider creating clear hypotheses that can be incorporated and tested by physiologists or others.
The important lesson learned from such tools is that identifying long lists of genes and proteins often generates more questions, which, when fully utilized, can lead to new research and progression in the field.

| The holistic holobiont
The term coral holobiont comprises the totality of the coral symbiotic relations including, but not limited to, endosymbiotic zooxanthellae, bacteria, archaea, viruses, and fungi. All are part of what collectively is termed the microbiome, and each plays a role in the response of the holobiont. Research can often be targeted to a specific symbiont of interest. As much as we need to simplify the system into individual parts in order to confidently discern the role of each player, however, we must also remember that it is intricately connected. Ultimately, the goal is to understand system requirements and describe the relationship between each component to unveil the mechanisms of thermal tolerance.
Cnidarian host and Symbiodiniaceae temperature response are often pursued as separate fields. Recent studies have increasingly been combining coral and Symbiodiniaceae responses in their hypothesis testing. A growing understanding of the metabolic hostsymbiont relationship has encouraged the use of methods that allow the measuring of these dynamics, such as metabolomics (Cui et al., 2019;Matthews et al., 2017) or NanoSIMS (Krueger et al., 2018). These have shown that the metabolic balance between the two partners is not only sensitive to environmental stressors but specifically fine-tuned (Cui et al., 2019;Matthews et al., 2017;Nielsen, Petrou, & Gates, 2018). The coral microbiome has also been shown to significantly impact the thermal stress response mechanisms of the coral as well as on their symbionts (Littman et al., 2010;Pogoreutz et al., 2018;Ziegler et al., 2017).
Evidently, interactions between the different partners of the holobiont are extremely important to consider when attempting to understand the overall response. Our understanding of coral temperature and bleaching tolerance hypotheses increasingly acknowledges the difficulty of discerning the role of one partner from the other.
Targeting the complexity of the individual components of the holobiont was recognized by the ReFuGe Consortium , and explores the sequencing of the various components of the coral holobiont. Integrative approaches are required to efficiently compare and contrast not only different molecular layers, but also the response and interaction of different members of the system. This interplay is particularly important in light of current aims at aiding corals in surviving rapid climate changes.

| CON CLUS ION
Realization of the economic benefit of reef systems coupled with undeniable evidence of climate change impacts has fueled the field of coral heat stress studies. The emergence of new research methods such as transcriptomics has led to a continuous expansion of knowledge in the field. However, rapid advancements in technology perpetuate the increase in data generated, which may distract from developing mechanistic understanding. In these times, it is important to reflect upon the path that research endeavors have taken, build upon these, and expand in directions suitable for the long-term goal of understanding how we may possibly ensure a future for coral reefs.
Gathering knowledge on the intricate system of the coral holobiont and combining the different parts to allow deeper insight into the overall response mechanisms will require a collaborative effort.
Only then can we hope to find successful ways to aid corals in acclimatizing and adapting to the rapidly altering environment.

ACK N OWLED G M ENTS
We would like to thank Alicia Schmidt-Roach for her help in editing the manuscript and providing feedback.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R S' CO NTR I B UTI O N S
MJC conceived the idea and wrote the first draft of the manuscript.
SSR designed the figures. MA provided guidance and structure. All authors read and approved the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
There are no data to be accessed or deposited.