Organisms mount the cellular stress response whenever environmental parameters exceed the range that is conducive to maintaining homeostasis. This response is critical for survival in emergency situations because it protects macromolecular integrity and, therefore, cell/organismal function. From an evolutionary perspective, the cellular stress response counteracts severe stress by accelerating adaptation via a process called stress-induced evolution. In this Review, we summarize five key physiological mechanisms of stress-induced evolution. Namely, these are stress-induced changes in: (1) mutation rates, (2) histone post-translational modifications, (3) DNA methylation, (4) chromoanagenesis and (5) transposable element activity. Through each of these mechanisms, organisms rapidly generate heritable phenotypes that may be adaptive, maladaptive or neutral in specific contexts. Regardless of their consequences to individual fitness, these mechanisms produce phenotypic variation at the population level. Because variation fuels natural selection, the physiological mechanisms of stress-induced evolution increase the likelihood that populations can avoid extirpation and instead adapt under the stress of new environmental conditions.
All living organisms exist under the stress of their environment. Stress, in this sense, refers to any environmental parameter exerting strain on biological systems (Kültz, 2020a). When organisms are well adapted to their environments, they harbor mechanisms that counteract imposed strain and therefore maintain homeostasis. Whenever environmental parameters change, organisms must adjust these mechanisms to uphold the balance between stress and the forces that oppose it. If the change in stress is minor enough, only the cellular homeostasis response (CHR) is needed for this adjustment. However, the capacity of the CHR may be exceeded depending on the magnitude of stress and how rapidly it arises. This threshold for stress tolerance is termed the ‘elastic limit’, and once it is surpassed, organisms must activate the cellular stress response (CSR) in order to survive (Kültz, 2020a; Call et al., 2017; Tian et al., 2012). Stress of this degree is becoming increasingly relevant and concerning to life on Earth amid climate change. As the atmosphere continues to collect greenhouse gases, numerous environmental factors, including the temperature, salinity and acidity of water, change globally and much more rapidly than during previous geological periods (Cheng et al., 2020; Hoegh-Guldberg et al., 2007; Karger et al., 2020). When populations are limited in their ability to migrate to more suitable environments, they must somehow adapt in order to remain viable.
Under these circumstances, the CSR can employ physiological mechanisms of stress-induced evolution (SIE). These are strategies by which individuals rapidly generate new heritable phenotypes. At the population level, SIE produces widespread phenotypic variation and therefore accelerates evolutionary processes. In one mechanism, stress triggers mutagenesis by causing both increased DNA damage and decreased DNA repair fidelity (Chatterjee and Walker, 2017). In a more flexible response, stress induces the alteration of epigenetic marks, including histone post-translational modifications (PTMs) and DNA methylation. These epigenetic marks modify the expression patterns of DNA. Therefore, even if an individual's sequence of DNA remains unchanged, expression patterns (and corresponding phenotypes) can be passed through generations. In a more radical response, stress can prompt the formation of structural genomic variants through either chromoanagenesis or transposable element (TE) activity. These processes can produce especially distinctive phenotypes by reorganizing gene regulatory networks, e.g. via activation or inhibition of cis-regulatory elements (CREs), modifying gene products, and creating and deleting genes (Lanciano and Mirouze, 2018; Mérot et al., 2020; Pellestor and Gatinois, 2020; Ye et al., 2018).
In this Review, we summarize key physiological mechanisms of SIE in eukaryotes. An emphasis is placed on animals for supporting examples. Throughout the article, we demonstrate on a molecular level how life experience can alter the phenotype of an individual and its progeny. Notably, these mechanisms may or may not increase an individual's fitness; oftentimes, they result in disease or sterility. Nonetheless, they facilitate the generation of phenotypic variation within populations, where individuals may develop novel solutions to compensate for stress. In doing so, these mechanisms increase the likelihood that populations will adapt under stress.
base excision repair
cellular homeostasis response
cyclobutane pyrimidine dimer
cellular stress response
differentially methylated region
heat shock protein 90
nucleotide excision repairs
non-homologous end joining
reactive oxygen species
translesion DNA synthesis
Stress triggers mutagenesis through increased DNA damage and decreased DNA repair fidelity
DNA damage is an unavoidable part of life. Even under ideal environmental conditions, DNA is continuously damaged by spontaneous alkylation, strand breaks, hydrolytic loss of nitrogenous bases and base conversion (Chakarov et al., 2014). In humans, it is estimated that 2×104 DNA damage events take place every day in each cell (Barzilai and Yamamoto, 2004). Damage, however, is not always detrimental, as the DNA damage response network has evolved to either repair DNA damage or tolerate it (Pilzecker et al., 2019). Only a fraction of DNA damage events lead to mutations that are retained and potentially inherited. In humans, despite the high frequency of DNA damage, rates of retained mutation are approximately 2.8×10−7 per base pair in somatic cells and 1.2×10−8 per base pair in the germline (Milholland et al., 2017).
Stress increases the rate of DNA damage, and therefore the rate of mutation, beyond what happens spontaneously. Diverse cellular stresses achieve this either directly or by secondarily stimulating the production of reactive oxygen species (ROS) in cells (Chakarov et al., 2014; Cheng et al., 2018; Kültz, 2005, 2020b). ROS can damage DNA by causing strand breaks or oxidizing nucleotides into a plethora of compounds, including thymine glycol and 8-oxo-deoxyguanosine (Grollman and Moriya, 1993; Honda et al., 2001; Sallmyr et al., 2008). Through alternative routes, stress can damage DNA by producing single-strand breaks (SSBs), double-strand breaks (DSBs), apurinic (AP) sites, deaminated cytosine, cyclobutane pyrimidine dimers (CPDs) and pyrimidine-pyrimidone photoproducts (6-4PP). In Table 1, we outline specific stresses that can produce these DNA lesions.
Cells attempt to repair all types of stress-induced DNA lesions. The strategy to repair DNA strand breaks depends on whether they are SSBs or DSBs. DSBs are especially mutagenic. When cells attempt to repair them, they can use the high-fidelity process of homologous recombination (HR), but most often they use the error-prone process of non-homologous end joining (NHEJ) (Chang et al., 2017). To address oxidized nucleotides, cells initiate base excision repair (BER) (Chatterjee and Walker, 2017). Nonetheless, approximately 2–5% of these lesions escape repair, and when they do, they often cause mutations from G:C to A:T (Chatterjee and Walker, 2017; Grollman and Moriya, 1993; Moriya, 1993). The remaining stress-induced lesions are often repaired through a combination of BER and nucleotide excision repair (NER). However, if the cell cycle progresses into S phase before the lesions can be repaired, DNA damage tolerance pathways are activated instead (Chatterjee and Walker, 2017; Duncan and Miller, 1980; Pilzecker and Jacobs, 2019). Translesion DNA synthesis (TLS) is a prominent mechanism of the DNA damage tolerance pathway, and it functions to ensure that DNA replication can proceed even when DNA lesions are present. TLS promotes mutagenesis by using low-fidelity DNA polymerases that lack corrective exonuclease activity (Gerlach et al., 1999; Masuda et al., 2016).
While DNA repair is naturally fallible, stress can further reduce its fidelity and thereby increase the retention of mutations. Heat stress, for example, can inhibit both the BER and NER systems (Kantidze et al., 2016). This inhibition compromises the repair of DNA damage inflicted by stress. Similarly, proteins required for mismatch repair are downregulated under the stresses of both hypoxia and toxins (Chatterjee and Walker, 2017; Mihaylova et al., 2003). The mechanism of DSB repair can also be altered by stress, ensuring that low-fidelity NHEJ is used for repair, e.g. during hypoxia and heat stresses (Galhardo et al., 2007; Kantidze et al., 2016).
Through these and many other mechanisms, stress increases the incidence and retention of mutations. Stress-induced mutagenesis is likely an adaptive strategy as it provides an avenue for a maladapted population to accumulate genetic diversity in response to environmental change. Selection can act on the resulting genetic variation, enabling the population to become better suited for stressful environments. These mutations are not entirely random. Stress-induced mutations accumulate at different rates in transcriptionally active versus silent genes because the susceptibility to DNA damage differs between corresponding euchromatin and heterochromatin (Makova and Hardison, 2015). This effect can accelerate evolution in genes that are actively involved in defining the phenotype of a specific cell type in a specific context. Altered cellular phenotypes, in turn, influence phenotypes at higher levels of organization, including the whole organism level.
Stress causes heritable (epigenetic) changes in histone PTMs
In the nucleus of eukaryotic organisms, DNA wraps around an octamer of the four core histones: H2A, H2B, H3 and H4 (Luger et al., 1997). These proteins are subject to a wide variety of PTMs (Zhao and Garcia, 2015). Histone PTMs are epigenetic marks that can modify the state of chromatin and influence gene expression. They can do this by altering the manner in which DNA is packaged, thus changing the accessibility of the DNA for proteins involved in transcription and repair (Norton et al., 1989). Histone PTMs also modulate the recruitment of histone reader proteins to specific genetic loci to carry out physiological functions, such as DNA repair, replication, transcription and chromosome condensation (Kouzarides, 2007).
Stress can alter the histone PTM landscape, which is the relative abundance and genomic distribution of all histone PTMs in a cell (Table 2). Histone PTMs are ‘written’ and ‘erased’ by histone modifying enzymes, but the catalytic activity of these enzymes can be modified under stress, e.g. through chemical inhibition or alteration of cosubstrate availability (Fan et al., 2015). Both of these strategies apply to the histone demethylase enzyme JmjC. Oxidative stress causes the iron in its catalytic center to be oxidized from Fe(II) to Fe(III), which inhibits its function and leads to histone hypermethylation (García-Giménez et al., 2021). Interestingly, hypoxia also represses the activity of this demethylase because JmjC requires oxygen as a cosubstrate (Hsu et al., 2021). At the same time, however, hypoxia-inducible factors transcriptionally upregulate JmjC to fine-tune the overall histone demethylation activity (Hsu et al., 2021). This example illustrates that the effects of stress on the regulation of histone PTMs are pervasive and highly complex.
By modifying the histone PTM landscape, stress can facilitate an appropriate physiological response, e.g. during temperature and salinity stresses. Heat stress increases the relative abundance of H3K27me3 in the adrenal gland of chickens (Gallus gallus domesticus) (Zheng et al., 2021). This epigenetic response is associated with increased glucocorticoid production, which assists in heat dissipation (Zheng et al., 2021). During cold stress, the relative abundance of H3K27me3 decreases in thale cress (Arabidopsis thaliana), and it does so specifically at the loci of two cold stress genes, leading to their activation (Yuan et al., 2013). On the contrary, stress-induced histone PTMs can be associated with maladaptive phenotypes. For example, people working in steel plants breathe in toxic particulate matter. As their time of employment increases, their levels of H3K4me2 and H3K9ac also increase. In this case, the histone PTM landscape is associated with an increased risk for lung cancer (Cantone et al., 2011).
Even once the stress has subsided, induced histone PTMs can be retained within individuals, via ‘intragenerational’ inheritance by mitosis (Alabert and Groth, 2012). When stress causes changes to histone PTMs in the germline, the epigenetic marks can be retained across generations (Fig. 1). This retention can occur through different processes. In one process sometimes called ‘intergenerational’ inheritance, stress directly induces histone PTMs in the gametes of exposed parents. Upon fertilization, gametes that carry the directly induced epigenetic marks become the next generation. In a second process often called ‘transgenerational’ inheritance, induced histone PTMs travel across multiple generations without the need for individuals inheriting them to be directly exposed to stress (Bošković and Rando, 2018; Mørkve Knudsen et al., 2018; Perez and Lehner, 2019; Woodhouse and Ashe, 2020). Transgenerational inheritance is especially relevant for SIE as it extends the time that natural selection can act on epigenetically mediated phenotypic variation. Heat stress, for example, was shown to increase the global acetylation levels of histones H3 and H4 in the brine shrimp (Artemia sp.). After heat stress subsided, the induced histone PTM landscape could be transmitted through three subsequent generations, and it was associated with enhanced tolerance to severe heat stress in the progeny (Norouzitallab et al., 2014).
While the mechanism of transgenerational epigenetic inheritance is not yet fully elucidated, individuals can directly receive modified histones from the gametes that form them. This process is relatively straightforward regarding maternal transfer, but epigenetic reprogramming represents a hurdle to paternal transfer. During spermatogenesis, histone proteins are replaced with protamines for an even tighter packaging of DNA (Bao and Bedford, 2016). Some species such as mice only retain 1–2% of histones in sperm; however, this value is widely variable between species (Champroux et al., 2018). For example, the percentage of retained histones is approximately 5–10% in humans (Champroux et al., 2018), 37% in nematode worms (Samson et al., 2014), 45% in marsupials (Soon et al., 1997) and 100% in lampreys and hagfish (Saperas et al., 1997). In this way, it is possible that some species have a much higher propensity for the transgenerational inheritance of histone PTMs.
Histone PTMs offer individuals a mechanism to rapidly modify gene expression patterns and their phenotypes to better tolerate their environment. Such altered phenotypes (and the underlying genotypes of corresponding individuals) are acted upon by natural selection and, therefore, represent targets of stress-induced adaptation. Selection on these targets may be prolonged over multiple generations because individuals exposed to stress can transmit histone PTMs, gene expression patterns and the resulting phenotypes they acquire to their progeny. The adaptive value of retaining phenotypes that confer tolerance to short periods of stress in the absence of persistent stress may seem questionable (Nilsson et al., 2018). However, what natural selection favors under such conditions are individuals with the ability to tolerate transient periods of stress best while also performing best during intermittent periods of low stress. For this reason, histone PTMs and corresponding gene expression patterns and phenotypes are reversible, and their persistence within a lineage can depend on the intensity and duration of stress experienced by their ancestors. In this way, epigenetic mechanisms can facilitate trial runs of new phenotypes and integrate stochasticity and periodicity in environmental conditions into the process of natural selection (Burggren, 2016; Walker and Burggren, 2020). Through this mechanism (and epigenetic inheritance of DNA methylation), natural selection assesses the adaptive value of corresponding phenotype variants in a particular lineage under variable environmental conditions over longer periods of time.
Stress alters heritable (epigenetic) DNA methylation patterns
DNA methylation is a heritable epigenetic mark characterized as a methyl group attached to the fifth carbon of cytosine. When DNA methylation occurs in a promoter, it typically silences the gene by preventing the binding of transcription factors and prompting the formation of heterochromatin. Conversely, when methylation occurs in an open reading frame, it typically activates the gene (Greenberg and Bourc'his, 2019; Jones, 2012; Moore et al., 2013). De novo DNA methylation is facilitated by the DNA methyltransferase enzymes DNMT3a and DNMT3b, which can be targeted to specific genes through the guidance of piwi-interacting RNA (Flores et al., 2013; Okano et al., 1999). Stress is well documented to induce de novo DNA methylation, leading to differentially methylated regions (DMRs). Owing to their influence on gene expression, DMRs can impact morphology, physiology, behavior and development (Angers et al., 2010).
Stress-induced DMRs have been reported across taxa, from plants to insects to humans (Ou et al., 2012; Shi et al., 2011; Martin and Fry, 2018). Through this epigenetic mechanism, the environment generates new phenotypes in individuals that, for better or worse, affect their fitness (Table 3). Many putatively adaptive responses have been observed. For example, the spiny chromis damselfish (Acanthochromis polyacanthus) was recently shown to accumulate 193 DMRs after exposure to increased temperature (Ryu et al., 2018). Those DMRs correlated with increased aerobic scope, which enhanced tolerance to heat stress (Ryu et al., 2018). Similarly, purple sea urchins (Strongylocentrotus purpuratus) that experienced upwelling conditions during gametogenesis induced DMRs in their progeny that were associated with increased body size (Strader et al., 2019; Wong et al., 2019). However, stress can sometimes also lead to transgenerational transmission of traits that reduce fitness. Ionizing radiation in zebrafish (Danio rerio), for example, was shown to induce 5658 DMRs; 19 of these were passed through one generation, and five were passed through two generations (Kamstra et al., 2018). In this case, the DMRs were localized to genes involved in cancer and apoptosis, which could help explain the developmental defects observed in the progeny inheriting these epigenetic marks (Kamstra et al., 2018).
Whether adaptive or maladaptive, phenotypes generated through stress-induced DMRs can be inherited within individuals and across generations. Within individuals, patterns of DNA methylation are often stably maintained through mitosis by the DNMT1 enzyme (Smith and Meissner, 2013). DNMT1 itself, however, has a relatively high error rate of approximately 5% (Bird, 2002). As a result, additional variation in DNA methylation patterns can emerge through time within an individual's cell population, which affects organismal phenotype. The mechanism of transgenerational inheritance of DNA methylation is not yet fully understood. A natural limitation to this process is that widespread reprogramming of DNA methylation takes place during gametogenesis and shortly after fertilization, but some genetic loci are protected during these events (Angers et al., 2010; Engmann and Mansuy, 2020). Even so, it has been observed on many occasions that stress-induced DMRs can be transferred through multiple generations, including in the examples mentioned above.
As an epigenetic mark, DNA methylation rapidly elicits phenotypic variation that can equip some individuals and their progeny to better cope with stress. Importantly, DNA methylation functions beyond an epigenetic mark as well, in a much more permanent manner. Namely, it increases rates of mutation by frequently causing cytosine to thymine transitions (Zhou et al., 2020; Yang et al., 2021; Holliday and Grigg, 1993). This pattern is so apparent that species with widespread DNA methylation exhibit global depletion of CpG dinucleotides, because this is where DNA methylation most often occurs (Gruenbaum et al., 1982). In humans, 60–80% of all CpG sites are methylated (Smith and Meissner, 2013). With such extensive DNA methylation, the human genome only has 20% of the expected amount of CpG dinucleotides, presumably because many cytosines in these sequences have been mutated into thymines (Bird, 1980). In contrast, fruit flies (Drosophila melanogaster), which display a very low level of DNA methylation, still have >90% of the expected amount of CpG sites (Capuano et al., 2014; Lyko, 2001; Bird, 1980). Because DNA methylation is targeted, C→T mutation can be targeted as well. Therefore, when stress induces DMRs, the resulting phenotypic advantages can potentially be fixed in a lineage by nonrandom mutation to specific genetic loci (Angers et al., 2010).
Stress impacts genome structure through chromoanagenesis
Of all the physiological mechanisms of SIE, changes to genome structure are the most dramatic. In a process called chromoanagenesis (also known as genome chaos), severe stress causes cells to rapidly shatter the genome and rearrange its contents (Heng and Heng, 2020). Structural genomic variants are the outcome of this process, and they can include any combination of copy number variants, chromosomal fusions, fissions, translocations, inversions and reshuffling (Mérot et al., 2020; Heng, 2009). These structural changes strongly affect organismal fitness by changing gene regulatory networks, altering gene dosage, functionally deleting genes or even creating new genes from previously non-coding DNA (Mérot et al., 2020; Pellestor and Gatinois, 2020; Ye et al., 2018). Most often, the effects of chromoanagenesis are deleterious. On rare occasions, however, the generated phenotypic diversity is lifesaving (Fig. 2).
Stress-induced changes in genome structure are well studied in the context of human disease. It has been discovered within the past 20 years that structural genomic variants are a universal feature of cancer, and they are frequently associated with additional diseases such as Alzheimer's (Heng, 2009; Horne et al., 2014). Using disease study systems, three categories of chromoanagenesis have been identified: chromothripsis, chromoanasynthesis and chromoplexy (Koltsova et al., 2019). Chromothripsis refers to a single event where one chromosome is shattered and randomly stitched back together. The process is triggered by a high load of DNA DSBs, which result under the pressure of numerous environmental stresses (Koltsova et al., 2019). Additional forces including telomere attrition, abortive apoptosis and mitotic errors also prompt chromothripsis (Pellestor and Gatinois, 2020). Chromoanasynthesis is a process that specifically leads to the generation of copy number variants, and it is triggered by DNA replication and repair errors (Koltsova et al., 2019). Finally, chromoplexy describes the reshuffling of several chromosomes over the course of multiple events, and it is often caused by replication stress, mitotic errors, and premature chromosome compaction (Shen, 2013).
Beyond causing disease states of somatic cells, chromoanagenesis proceeds within the germline and within embryos during early development (Pellestor and Gatinois, 2020). In this context, chromoanagenesis can be adaptive and lead to rapid speciation in asexually reproducing organisms and even in heterogametic species, as long as both parents experience compatible genome changes for sexual reproduction (Heng, 2009). Every type of structural genomic variant has been implicated in driving speciation (Campbell et al., 2018; Feulner and De-Kayne, 2017). Accordingly, both the morphology and number of chromosomes vary widely across taxa (Ferguson-Smith and Trifonov, 2007). For example, the number of chromosome pairs in eukaryotes ranges from 1 to 720 (Schubert and Vu, 2016; Khandelwal, 1990). In light of evolutionary history, chromoanagenesis could be a large contributor to this structural genomic variation because periods of major evolutionary change tend to occur during periods of severe stress. For example, the ‘Big Five’ mass extinctions and their subsequent events of adaptive radiation corresponded to large changes in temperature, sea level, volcanic and tectonic activity, and meteor impacts (Condamine et al., 2013). During such periods, eurytopic species are favored over stenotopic species, while the opposite is the case during long, stable geological periods (Kültz, 2003). Corresponding patterns of evolutionary history have been interpreted by the theory of punctuated equilibrium (Gould, 1982).
Structural genomic variants can be adaptive under various contexts (Table 4). When challenged by the widely used herbicide glyphosate, palmer amaranth (Amaranthus palmeri) developed a copy number variant that enabled resistance to the herbicide (Gaines et al., 2010). Similarly, the codling moth (Cydia pomonella) developed a sex-linked resistance to insecticides through a chromosome fusion (Nguyen et al., 2013). Chromosome inversions have been adaptive in the context of behavior, mating strategies and morphology (Wellenreuther and Bernatchez, 2018). For example, inversions produced cryptic color phenotypes in stick insects (Timema cristinae) and facilitated appropriate migratory behaviors in rainbow trout (Oncorhynchus mykiss) (Lindtke et al., 2017; Wellenreuther and Bernatchez, 2018).
Whether the process occurs in somatic cells or gametes, chromoanagenesis elicits major phenotypic changes by altering genome structure in individuals facing severe stress. Most of the time, the outcomes are deleterious – either a disease emerges, or individuals generate gametes that are incompatible with potential mates, rendering the individuals sterile. On the lucky occasion, structural genomic variants enable successful survival and reproduction, and they do so within one generation.
Stress affects the activity of TEs
TEs have long been considered an engine of evolutionary change fueled by stress (McClintock, 1984), and they make up a large portion of eukaryotic genomes. In mammals, approximately 40% of the genome is composed of TEs, and in plants, that value can be as high as 85% (Chénais et al., 2012). TEs are sequences of DNA, sometimes called ‘jumping genes’, that can readily move throughout the genome. The process of their transposition can proceed through ‘copy and paste’ or ‘cut and paste’ strategies. In the copy and paste strategy, class I TEs are transcribed into an RNA intermediate then reverse transcribed back into DNA at a new location. In the cut and paste strategy, many class II TEs have their DNA sequence broken out of its position, then relocated (Wicker et al., 2007). Oftentimes, all the information needed for transposition is encoded within the TE. If this is the case, then they are called autonomous TEs, and depending on their family, they encode enzymes such as reverse transcriptase, proteinase, RNase, integrase and transposase. Nonautonomous TEs have also evolved, and they lack some of the necessary components for transposition. As a result, they rely on autonomous TEs for their mobilization (Wicker et al., 2007).
When activated, TEs can quickly produce distinctive phenotypes by impacting gene expression, gene products and genome structure. The expression of genes can be affected when newly incorporated TEs provide CREs, change the context of existing CREs or alter the local epigenetic landscape (Chénais et al., 2012; Lanciano and Mirouze, 2018). Similarly, transposition can alter gene products when inserted TEs cause alternative transcription start sites, alternative splicing or premature termination. New exons and introns can even be created in the process (Lanciano and Mirouze, 2018). For transposition to occur, DNA DSBs are needed to cut out and insert TEs. This form of DNA damage increases rates of mutation, specifically at the sites of deletion and insertion (Biémont and Vieira, 2006). Furthermore, transposition-induced DSBs can produce structural genomic variants by feeding into the chromothripsis pathway, which leads to chromosome inversions and chromosome reshuffling (Fig. 2) (Pellestor and Gatinois, 2020). TEs generate additional structural genomic variants as a consequence of the high sequence similarity between TEs of the same family, in particular at their flanking sequences such as inverted terminal repeats. This similarity enables non-allelic homologous recombination, which can cause chromosome inversions, duplications, translocations and deletions (Kent et al., 2017).
Numerous stresses can alter TE activity, including cold and heat stresses, UV irradiation, salinity stress and pollution (Miousse et al., 2015; Rey et al., 2016). However, the pattern of alteration is context dependent. In response to stress, TEs may be activated, repressed, activated then repressed, or repressed then activated (Horváth et al., 2017). Furthermore, when TEs are activated, it can be at the transcriptional level, the transpositional level or both (Horváth et al., 2017). Epigenetic regulation is one major force that mediates this change (Biémont and Vieira, 2006). TEs are repressed under the control of DNA methylation and histone PTMs (Zemach et al., 2010). When stress alters these epigenetic marks, TEs can be released from repression and freed to transcribe their contents and/or mobilize to other parts of the genome (Pappalardo et al., 2021). Another stress-sensitive mechanism of TE activation involves the heat shock protein 90 (HSP90) family. While HSP90 silences TEs under optimal environmental conditions, moderate stress can limit this function when HSP90 is instead needed to protect protein conformation (Ryan et al., 2016). Notably, the limitation of available HSP90 also increases phenotypic diversity by releasing cryptic genetic variation from suppression (Paaby and Rockman, 2014). Therefore, HSP90 has been considered a key evolutionary capacitor (Rutherford and Lindquist, 1998).
Stress-induced changes in TEs have been observed across eukaryotic taxa (Table 5), and on many occasions, they have proven to be adaptive. For example, insecticide exposure has altered TE activity in insects. In the fruit fly (Drosophila melanogaster), this led to the overexpression of an insecticide detoxifying gene (Chung et al., 2007). In the common house mosquito (Culex pipiens), this led to the alternative splicing of a toxin receptor gene (Darboux et al., 2007). In both instances, the TEs induced by insecticides resulted in insecticide resistance. Similarly, climate has been shown to induce potentially adaptive TEs in the Asian tiger mosquito (Aedes albopictus). The frequency of TE insertions varies between a native population in a tropical environment and an invasive population in a temperate environment. In the invasive population, TEs of multiple families are inserted at higher frequencies, and they are positioned within the proximity of genes that likely facilitate overwintering (Goubert et al., 2017). Altered regulation of these genes could increase the fitness of mosquitoes living in colder climates.
Through the alteration of TE activity, stress generates rapid phenotypic variation. The variation can be significant because TE activation has the power to affect gene expression, gene products and genome structure. When these changes happen in the germline, they can be passed from parent to offspring indefinitely. This standard form of transmission is referred to as ‘vertical transfer’. However, ‘horizontal transfer’ of TEs can happen as well, where TEs jump between species. In the evolutionary history of vertebrates, for example, at least 975 events of horizontal transfer of TEs have occurred (Zhang et al., 2020).
Life experience and physiology shape evolution
Contrary to the principles of the modern synthesis of evolutionary theory, stress that an individual encounters throughout its lifetime is now known to induce heritable phenotypic variation (Burggren, 2014; Jablonka and Lamb, 2020; Noble, 2013; Skinner, 2015). This concept of SIE has been accepted for decades in regard to prokaryotes (Radman, 1975; Bjedov et al., 2003; Foster, 2007; Rosenberg et al., 2012). In prokaryotes, stress significantly increases rates of mutation, largely through the activation of the SOS system and RpoS stress response (Radman, 1975; Foster, 2007). Although these systems do not exist in eukaryotes, more recent studies have demonstrated that eukaryotes employ several powerful mechanisms to increase phenotypic variation in response to stress. Beyond the increased rates of mutation via DNA damage and lowered DNA repair fidelity that occur outside of the prokaryotic SOS and RpoS systems, variation is achieved through histone PTMs, DNA methylation, chromoanagenesis and TE activity.
In multicellular eukaryotes, the mechanisms of SIE can proceed in both the soma and the germline. Somatic cell evolution has been studied intensively in the context of disease (Anway et al., 2006; Heng, 2009; Rajesh Kumar et al., 2002), showing that the outcome of these mechanisms can be maladaptive. Considering that many multicellular organisms consist of millions, billions or even trillions of cells, e.g. 37 trillion cells in humans (Bianconi et al., 2013), the large population of cells provides a sufficient pool of beneficial alterations that selection can act on. A classic example of adaptive somatic cell evolution is the production of antibodies in vertebrates. After organisms are exposed to new antigens, the variable regions of immunoglobulin genes in B cells become hypermutated (Diaz and Flcjnik, 1998; Wysocki et al., 1986). This mechanism ultimately increases the affinity of antibodies to circulating antigens, thereby strengthening the immune system. Although these changes to somatic cells easily impact the fitness of individuals by affecting their ability to survive and reproduce, stress arguably has the strongest influence over organismal evolution when alterations happen within the germline. Each of the physiological mechanisms of SIE can proceed within the germline, although this happens less frequently than in somatic cells because germ cell chromatin is transcriptionally silent and better protected from damage (Bao and Yan, 2012; Engmann and Mansuy, 2020; Heng, 2009; Milholland et al., 2017). Nonetheless, critical windows of development exist where stress is more likely to induce stably transmitted epigenetic marks in the germline (Skinner, 2011). Embryonic gonadal sex determination is the first critical window, and gametogenesis is the second (Hanson and Skinner, 2016).
Despite the popularity of the idea that the soma and the germline are completely isolated, i.e. the Weismann Barrier (Weismann, 1890), this barrier can be bypassed through microvesicles. Microvesicles, in the form of either shedding vesicles or exosomes, are released from all cell types (Camussi et al., 2010; Doyle and Wang, 2019). Once released, they can remain in the extracellular matrix within the proximity of the cell of origin, or they can travel through biological fluids to reach distant target cells (Camussi et al., 2010). These microvesicles contain components of the origin cell, including RNA and proteins. By delivering both of these components, microvesicles have the power to epigenetically reprogram target cells (Engmann and Mansuy, 2020; Sharma, 2014). This important transfer of information can take place between two somatic cells, or between somatic and germ cells. A recent study clearly demonstrated this phenomenon in mice xenografted with human tumor cells. RNA from the xenografted cells traveled through the bloodstream in exosomes until being finally received by spermatozoa (Cossetti et al., 2014). Therefore, germ cells do not necessarily need to be directly altered by stress; it is possible for information from affected somatic cells to reach and modify the germline. Impressively, Charles Darwin essentially predicted the existence of microvesicles. He described them as ‘gemmules’ in 1868, before they could have possibly been detected (Noble, 2021).
Through all the physiological mechanisms discussed in this brief Review, eukaryotic organisms can establish heritable phenotypic variation in response to stress. Notably, DNA base mutation is not the only driver of this variation. Rapid phenotypic diversity can be achieved by histone PTMs, DNA methylation, chromoanagenesis and TE activity. The induced variation can be adaptive, maladaptive or neutral in specific contexts. In any case, it is produced at a time when homeostasis cannot be maintained, and the system is forced to explore novelty.
Conclusions and future perspectives
This Review summarizes five physiological mechanisms of SIE, which serve to generate novelty in populations experiencing altered environmental conditions. Owing to their widespread presence across the phylogenetic web of life, these mechanisms have likely been favored during evolution by conferring significant selective advantages that outweigh potential disadvantages, such as the increased susceptibility to pathologies. A better understanding of the profound implications of these mechanisms for cells, organisms and populations represents an exciting frontier in biology. Many open questions that should be of great interest to comparative physiologists remain, including the following. Is there a correlation between the prevalence of SIE mechanisms, incidence of proliferative disease and average lifespan across different species? How does the magnitude of stress impact the proportion of favorable to unfavorable phenotypes produced through SIE mechanisms in a population? To what extent has SIE driven punctuated equilibrium throughout evolutionary history? How does SIE impact ecosystem succession during geological periods of rapid environmental change? SIE represents an exciting new paradigm in comparative evolutionary physiology that challenges long-standing dogmas and stimulates the creative intellect of current and future physiologists. In this brief Review, we share our enthusiasm for this fascinating area of biology to inspire future research on SIE by a broader scientific community.
This work was supported by the National Science Foundation grant MCB-2127516 to D.K.
The authors declare no competing or financial interests.