Hsp90 and Complex Birth Defects: A Plausible Mechanism for the Interaction of Genes and Environment

Abstract

How genes and environment interact to cause birth defects is not well understood but key to developing new strategies to modify risk. The threshold model has been proposed to represent this complex interaction. This model stipulates that while environmental exposure or genetic mutation alone may not result in a defect, factors in combination increase phenotypic variability resulting in more individuals crossing the disease threshold where birth defects manifest. Many environmental factors that contribute to birth defects induce widespread cellular stress and misfolding of proteins. Yet, the impact of the stress response on the threshold model is not typically considered in discephering the etiology of birth defects. This mini-review will explore a potential mechanism for gene-environment interactions co-opted from studies of evolution. This model stipulates that heat shock proteins that mediate the stress response induced by environmental factors influence the number of individuals that cross disease thresholds resulting in increased incidence of birth defects. Studies in the field of evolutionary biology have demonstrated that heat shock proteins and Hsp90 in particular provide a link between environmental stress, genotype and phenotype. Hsp90 is a highly expressed molecular chaperone that assists a wide variety of protein clients with folding and conformational changes needed for proper function. Hsp90 also chaperones client proteins with potentially deleterious amino acid changes to suppress variation caused by genetic mutations. However, upon exposure to stress, Hsp90 abandons its normal physiological clients and is diverted to assist with the misfolded protein response. This can impact the activity of signaling pathways that involve Hsp90 clients as well as unmasking suppressed protein variation, essentially creating complex traits in a single step. In this capacity Hsp90 acts as an evolutionary capacitor allowing stored variation to accumulate and then become expressed in times of stress. This mechanism provides a substrate which natural selection can act upon at the population level allowing survival of the species with selective pressure. However, at the level of the individual, this mechanism can result in simultaneous expression of deleterious variants as well as reduced activity of a variety of Hsp90 chaperoned pathways and potentially shift phenotypic variability over the disease threshold resulting in birth defects.

The Complex Genetics of Birth Defects

Many non-syndromic birth defects are due to complex interactions between susceptible genomes and environmental factors. These include a spectrum of malformations of the central nervous system such as holoprosencephaly, anencephaly and spina bifida; eye malformations such as coloboma, anophthalmia and microphthalmia; as well as other malformations such as congenital heart defects, cleft lip and palate [17, 21, 26, 29, 30, 33, 55, 64, 76, 78, 84]. Neurological disorders such as Autism and Schizophrenia also exhibit complex pattern of inheritance [1, 3, 50]. Identification of the underlying genetic drivers of these complex diseases poses enormous challenges. While genome wide surveys have been successful in linking some genetic variants, these variants often only explain the underlying causes in a fraction of cases [4, 7, 11, 38, 42, 57]. Often plausible variants are also found in unaffected parents raising the question of why the offspring, but not the parent, exhibits the disease. It is generally accepted that the missing pieces in these cases lie in multifactorial genetics involving either multiple interacting genes and environmental factors. However, the complexity of the underlying causes makes pinpointing the key genetic changes that contribute to the malformation extremely difficult. The conundrum is further complicated by our limited understanding of how environmental factors interact with susceptible genomes. A greater understanding of this interaction is needed as well as how genetic variants might produce disease in one individual but not another. One potential mechanism that will be explored in this mini-review is co-opted from studies of the role of Hsp90 in evolution where Hsp90 has been shown to silence genetic variation acting as a capacitor to store genetic variation without impacting phenotype [67, 88]. As a capacitor, the chaperone activity of Hsp90 contributes to the robustness of the organism by masking genetic mutations allowing proper folding of mutant proteins so that they retain at least partial activity. The role of Hsp90 in stabilizing mutated oncoproteins that would otherwise be degraded is well established [24, 77]. However, when environmental stress is encountered mutations become leading to a release of the stored genetic variation and expression of phenotypic variance.

Hsp90 as an Evolutionary Capacitor

To understand how Hsp90 suppresses genetic variation and how this variation is unmasked upon exposure to stress, an appreciation of how heat shock proteins function as protein chaperones is required. Heat shock proteins act in different cellular compartments as chaperones to ensure proper folding and conformation of a variety of proteins. Heat shock proteins are utilized under physiological conditions, as well as in response to damage induced by stress [49, 74]. Multiple heat shock proteins are expressed in vertebrates and are named according to their molecular weight. Hsp90 is encoded by two genes: HSP90aa1 and HSP90ab1 corresponding to Hsp90α and Hsp90β. Hsp90β is expressed constitutively whereas expression of Hsp90α is increased by stress [14]. While, deletion of Hsp90ab1 results in embryonic lethality due to defects in development of the placenta, mutation of Hsp90aa1 does not result in embryonic phenotypes [28, 79]. Additional complexities related to differential expression and function of Hsp90α and Hsp90β as well as cochaperones, structure function and regulation are described in a number of recent review articles [6, 31, 36, 71, 92].

Hsp90 can provide an interface between stress responses and developmental networks influencing trait thresholds and expression of a wide variety of phenotypes. Many stresses encountered during pregnancy that are linked to structural birth defects cause protein misfolding and induce expression of Hsp90. These include heat shock as a result of maternal fever and oxidative stress that occurs with hot tub usage, diabetes or alcohol or arsenic exposure [18, 20, 39, 41, 49, 90]. In a non-stressed cell, heat shock proteins are already the most highly expressed proteins creating a reservoir in the event of stress. Hsp90 in particular comprises 1 to 2% of the bulk of cellular proteins which increases to 4 to 6% when the cell is exposed to stress [58]. Increased Hsp90 expression is mediated by heat shock factor (HSF) which binds to and activates transcription from heat shock elements within the promoter of Hsp90 [58]. In the absence of stress, HSF is inhibited by association with heat shock proteins, but with cellular stress, heat shock proteins dissociate from the HSF, translocate to the nucleus and activate transcription of a variety of proteins including heat shock proteins.

However, production of new Hsp90 in the face of cellular stress through increased transcription does not happen instantly. As transcription of Hsp90 is induced and then the newly transcribed transcripts are translated, the reservoir of highly abundant Hsp90 becomes diverted from physiological clients to assist with protein folding as well as degradation of misfolded proteins. Hsp90 interacts with around 2000 proteins or approximately 10% of the proteome that includes ~150 well characterized client interactions [66, 75, 89]. These clients are involved in developmentally important signaling pathways, apoptosis and almost every developmental process [69, 71, 74, 75, 88]. For example, Hsp90 is required for steroid hormone receptor function by facilitating intracellular transport, nuclear translocation, recycling and regulation of transcriptional output of the receptors [19]. Hsp90 is also required for stabilization and conformational maturation of a variety of protein kinases key for developmental pathways [73]. If the chaperone system cannot refold a protein, it then mediates targeting the misfolded protein to the proteasome for degradation [59, 63, 70]. Thus, inhibition of Hsp90 or diverting Hsp90 function by exposure to stress results in destabilization and targeting to the proteasome of a slew of proteins. Because Hsp90 has so many clients, diverting Hsp90 to stress induced protein misfolding can simultaneously reduce the activity of multiple signaling pathways decreasing the overall robustness of development.

Hsp90 chaperone activity also allows accumulation of mutations in client proteins by facilitating certain variant proteins to properly fold and retain at least some activity [88]. This can allow for accumulation of mutations in Hsp90 clients that would otherwise alter protein function. In this way Hsp90 can function as an “evolutionary capacitor” enabling genomes to store but not express accumulated genetic diversity without expression of the diversity [86, 88]. With exposure to stress, the large reserve of Hsp90 can be depleted as Hsp90 is redirected to chaperone proteins damaged by stress [13, 60]. Phenotypes arise when the level of stress overcomes the buffering capacity of the Hsp90 reserve. As Hsp90 is diverted to general protein folding and quality control functions, mutations become unmasked. This newly expressed variation may allow the organism to survive in a stressful environment conferring an advantage under selective pressure. In this way, Hsp90 allows for increased genetic diversity that evolution can act upon and provides a key link between environment stress and expression of genetic variation. As illustrated in Figure 1, unmasking genetic diversity in combination with reduced activity of signaling pathways that require Hsp90 chaperone activity for normal protein function has the potential to create complex traits in one step [65].

Figure 1. Diverting Hsp90 to the misfolded protein response with exposure to stress has the potential to create complex traits in a single step.

Hsp90 interacts with a wide variety of client proteins involved in developmental pathways and genomic stability. The chaperone activity of Hsp90 promotes functional conformations of these client proteins during normal development. Hsp90 also interacts with clients to mask function altering sequence variations. Upon exposure to stress, Hsp90 is diverted to the misfolded protein response abandoning its physiological clients. This results in reduced activity and targeting of clients to the proteasome. Proteins with sequence variants become unmasked resulting in altered function or targeting to the proteasome. The combined reduction in signaling pathways mediated by physiological clients and altered function of unmasked protein variants results in decanalization of phenotypes as well as genomic instability that can contribute to birth defects with exposure to environmental stress.

Hsp90 and Buffering of Environmental Exposures

Many developmental processes are quite robust with near normal results in the face of a wide variety of genetic and environmental perturbations. These traits are referred to as “canalized” [88]. Stability results from a combination of gene redundancy, network connectivity, feedback loops, modularity and miRNA buffering and contribute to buffering environmental stress [5, 12, 32, 35]. Perturbation of these stability enhancing mechanisms results in reduced phenotypic robustness and can release cryptic genetic variation. Of the multiple mechanisms proposed to buffer developmental processes, Hsp90 provides an attractive link between environmental and genetic perturbations. Hsp90 plays an important role in facilitating canalization and maintaining developmental stability by buffering the impact of small changes in environmental stress so that the resulting phenotype is not impacted. However, if the environmental challenge overcomes the buffering capacity of the Hsp90 reservoir, cryptic diversity can be unmasked as Hsp90 is diverted to deal with the stress response. Decanalization under these circumstances results in the gain of phenotypic diversity and provides genetic heterogeneity upon which natural selection can act within the population [88]. At the same time, newly expressed genetic variation may lead to birth defects, as unmasked variants disrupt protein function impacting survival and/or normal development. Thus, while the stored genetic diversity unmasked with stress may be advantageous in terms of the fitness of the population, on the individual level, deleterious variation can result in reduced fitness. In this fashion, decanalization of phenotypic diversity in response to stress can potentially explain the increased risk of birth defects with environmental exposures. This model also provides an attractive explanation for the frequent finding of candidate disease causing variants in unaffected family members, a phenomenon common in complex genetic disorders [11, 42, 57].

The threshold model is often evoked to explain the etiology of complex disorders [21, 22, 34, 91]. This model is based on the presence of a continuous spectrum of phenotypic variation within the population that does not result in disease until a certain threshold of liability is reached (Figure 2). In terms of the threshold model, decanalization would result in increased variance of complex traits without necessarily changing the mean of the population [68, 88]. This can lead to a greater number of individuals crossing the liability threshold to manifest a disease state or birth defect. However, if decanalization also unmasks deleterious variants the mean would also shift, resulting in even more individuals with phenotypes that cross the disease threshold.

Figure 2. Hsp90 and the threshold model for developmental defects.

Many developmental disorders exhibit discontinuous variation with individuals either being affected or not with a liability threshold (dotted line) above which the disorder is present. Canalized phenotypes (represented by the green distribution) exhibit a narrow phenotypic variability below the disease threshold which can be decanalized with stress or Hsp90 inhibition (represented by the red distribution). Decanalized and canalized traits have the same mean with increased variability resulting in a number of individuals falling above the disease threshold. Introduction of a masked variant would have no impact on the distribution of phenotypic variability as long as Hsp90 chaperone activity is engaged. However, with the induction of stress or Hsp90 inhibition the variant is unmasked shifting the mean of the population in addition to increasing the phenotypic variability as the phenotype becomes decanalized (represented by the blue distribution).

These ideas are well illustrated in studies of model organisms. In the 1950’s, Waddington published the results of a series of experiments describing his curious findings where exposure of Drosophila larva to heat shock results in a small fraction of progeny exhibiting the crossveinless phenotype with altered patterning of the wing veins or the ultrabithorax (Ubx) phenotype where the T2 thoracic segment is transformed to the T3 segment resulting in flies with two sets of wings [80–82]. Selection for the unmasked phenotype enriched the defect in the next generation. Interestingly, after a few generations, heat shock was no longer required to unmask the phenotype. Thus, in Waddington’s experiments, natural selection either changed the threshold of trait expression or multiple rare alleles that affected phenotypic penetrance were co-selected resulting in genetic assimilation of the trait.

The mechanism for the initial masking and then unmasking of variability was unknown and was termed “Waddington’s widget”. In 1998, Rutherford and Lindquist published results of a series of experiments demonstrating that Hsp90 fulfills the criteria of this unknown “widget” [67]. They found that masking of the crossveinless phenotype depends upon Hsp90 that can be unmasked by either exposure to stress, inhibitors of Hsp90 or genetic disruption of the Hsp90 locus. Furthermore, as in Waddington’s experiments, the particular phenotype expressed with Hsp90 inhibition depended upon the genetic background, arguing that the genetic diversity is already present. The observation that after multiple generations traits become fixed point to the oligogenetic nature of the selected traits.

Evidence for similar phenomena are found in a wide array of organisms including plants, zebrafish, yeast, worms, cavefish and even humans [10, 13, 27, 40, 45, 61, 65, 85]. For instance, silencing Hsp90 by genetic mutation, chemical inhibitors or heat shock in zebrafish embryos results in expression of a spectrum of underlying phenotypic variability [45]. Similar to the initial experiments in flies, the particular phenotype expressed with Hsp90 inhibition was dependent on strain background and the developmental stage of inhibition. Inhibition during early development (at the 30% epiboly stage of development) resulted in a wide variety of phenotypes including growth inhibition, reduced pigmentation, heterotaxy, pericardial oedema, as well as malformation of the heart, fin and notochord. Later exposure (50% epiboly) yielded more minor abnormalities, most notably completely penetrant eye phenotypes with variable severity. Interestingly, inhibition of Hsp90 in specific mutant background strains can modify phenotypic severity in a mutation dependent fashion. For example, Hsp90 can mask mutations in the zebrafish gene encoding the Nodal related Squint (Sqt). Exposure of mutant fish to heat shock or Hsp90 inhibition decanalized the phenotype increasing the penetrance of cyclopia [54]. Another study in zebrafish demonstrated Hsp90 inhibition can alter the severity of eye phenotypes unmasking cryptic genetic variants [85]. As in the fly experiments, frequency of phenotypes increased with selective breeding. Interestingly, genetic variants masked by Hsp90 had missense mutations that may impact proteins folding, whereas phenotypic severity in strains with nonsense variants were unaffected by Hsp90 inhibition.

In rodents, hyperthermia induces a wide range of defects dependent on exposure timing and dose. These included neural tube defects such as anencephaly, exencephaly, microcephaly, encephalocele and ocular defects such as microphthalmia, coloboma, as well as facial clefting, maxillary hypoplasia, omphalocele, renal agenesis, vertebrae and tail abnormalities [83]. While the involvement of Hsp90 in unmasking background variation has not been directly investigated in rodents, Hsp90 inhibition can relieve repression of endogenous retroviruses leading to altered expression of genes near viral integration sites [37]. In humans, disease severity in Fanconi anemia shows a dependence on the interaction between Hsp90 and variants in the FANCA gene. Mutations in FANCA that engaged Hsp90 are less pathogenic but if Hsp90 is inhibited, these mutations are unmasked, impairing the function of the FANCA protein [45]. Collectively these studies demonstrate that both stress and Hsp90 perturbation have similar outcomes in unmasking background specific traits and that some of the effects of environmental stressors could be through inhibition of Hsp90 activity.

Hsp90 Promotes Genomic Stability

Another mechanism that might contribute to birth defects with stress is the role of Hsp90 in maintaining genomic stability (Figure 1). Inhibition of Hsp90 can result in increased frequency of point mutations, insertions, deletions, microsatellite slippage events, somatic recombination and transposon activity [88]. Destabilization of the genome with Hsp90 inhibition can come from multiple sources. As mentioned previously, one of the physiological functions of Hsp90 is to silence transposons [56]. Thus, exposure to stress relieves Hsp90 inhibition, de-repressing transposable elements can generate new genetic variation to further genetic and phenotypic changes [72]. Hsp90 clients also include proteins involved in DNA maintenance and repair such as FANCA1, telomerase, BRCA1 and Rad proteins [88]. Since mutations induced by genomic instability are inherited in a mosaic fashion, the effect of Hsp90 inhibition on genomic instability during post-implantation development would have a greater impact on birth defects than on evolution since somatic mutations can have a devastating effect on embryonic development but are not usually inherited. It is not clear how much mosaicism contributes to birth defects but recent studies in humans indicate there is quite a bit of mosaicism in humans and increased mosaicism is associated with autism spectrum, congenital heart defects, as well as other developmental disorders [15, 23, 46, 48, 51, 87]. Interestingly, the number of de novo loss of function mutations detected in whole exome and whole genome projects is increased in a variety of complex birth defects [8, 25, 43, 47, 52, 62]. Moreover, heat stress plays an established role in many of these birth defects. For example, maternal fever that potentially activates Hsp90 has been linked to Autism when occurring in the 3^rd trimester or neural tube defects when exposure is in the 1^st trimester [20]. Both disorders are associated with an increased deleterious mutation load in affected individuals [11, 51]. Finally, the risk for cancer among children with birth defects is higher than in the general population [9, 16, 44]. Since cancer risk is linked to genomic instability [2, 53], this association provides further circumstantial evidence of genomic instability mechanisms and mosaicism in their etiology. Together these observations provide support for the idea that genomic instability is found in a variety of developmental disorders where stress exposures plays a role. Whether diverting Hsp90 away from its normal physiological chaperone functions and client proteins involved in genome stability to the stress response plays a role in increased genome instability in developmental disorders remains to be determined.

Conclusions

Exposure to heat and oxidative stress is a trigger for a variety of birth defects. Hsp90 plays a key role in chaperoning client proteins involved in developmental pathways and genomic stability as well as masking existing sequence variants. Exposure to stress causes Hsp90 to abandon these normal physiological functions to assist with the protein misfolding response. This shift results in simultaneous reductions in a wide variety of signaling pathways required for development, genomic instability and at the same time unmasking existing protein variants. The net effect of these changes would be to express complex traits in a single step. The impact of Hsp90 in mediating gene-environment interactions in on developmental disorders in humans and vertebrate model systems is relatively unexplored and could potentially account for many of the features of complex developmental disorders.

Highlights.

• Exposure to heat and oxidative stress is a key trigger of a variety of birth defects.

• Hsp90 plays a key role in chaperoning client proteins involved in developmental pathways and genomic stability as well as masking sequence variants.

• Exposure to stress causes Hsp90 to abandon its normal physiological functions to assist with general protein misfolding responses resulting in simultaneous reductions in a wide variety of signaling pathways required for development, genomic instability and unmasking of potentially deleterious protein variants resulting in the expression of complex traits in a single step.

• The impact of these mechanisms on developmental disorders in humans and vertebrate model systems is relatively unexplored and could potentially account for many of the features of complex developmental disorders.