Malignant Hyperthermia in the Post-Genomics Era: New Perspectives on an Old Concept

Sheila Riazi, M.Sc., M.D.; Natalia Kraeva, Ph.D.; Philip M. Hopkins, M.D., F.R.C.A.

Disclosures

Anesthesiology. 2018;128(1):168-180. 

In This Article

New Genetic Technology and Mh

Next-generation sequencing enables fast and cost-efficient sequencing of all protein coding regions—exons—in the human genome. The 1000 Genomes Project (2008 to 2015)[35] has created the largest public catalogue of human genetic variation through analysis of whole-genome sequencing data of thousands of individuals from multiple populations.

Prevalence of RYR1 and CACNA1S Variants in the General Population

Analysis of next-generation sequencing variation databases for MH genes by two recent studies has corroborated previous observations of high levels of allelic heterogeneity within RYR1 and CACNA1S compared to other genes of the genome. Based on the NHLBI Exome Sequencing Project data set, which includes variation data from 6,500 exomes (http://evs.gs.washington.edu/EVS/; accessed April 8, 2017), it was estimated that CACNA1S and RYR1 are more genetically diverse than the average gene in either African American or in European-American populations, i.e., both these genes have a high level of natural variation.[36] Comparably high levels of variation in both RYR1 and CACNA1S were detected by a second study where the authors evaluated exome sequencing data from the ClinSeq data set on a cohort of 870 volunteers not selected for MH susceptibility.[37] Furthermore, using RYR1 and CACNA1S variation data for this unselected cohort, the authors assessed the prevalence of MH susceptibility in the general population. Based on an allele frequency of less than 1%, genotype–phenotype data, and the primary literature, they found that only 19% of the RYR1 variants identified in this cohort unselected for MH were probably benign, whereas 6% were pathogenic, and 75% were variants of unknown significance. Of the pathogenic RYR1 variants, three have been previously reported in association with MH: such a high prevalence of MH–associated RYR1 variants has not been replicated in the much larger sample presented in the ExAC Browser. For CACNA1S, 20% of the variants were defined as benign, and 80% were defined as variants of unknown significance.

These studies showed that a large fraction of the RYR1 variants found in databases were rare, with a frequency of 0.00001 or less. It is noteworthy that an abundance of rare variants is a feature common to other genetic disorders. The exome sequencing of thousands of patients with monogenic disorders has revealed that about 80% of the identified variants are unique, seen only in one proband, and 96% of the variants were found three times or less.[38] Similar to variation data for RYR1 and CACNA1S, variants of unknown significance comprised about 70% of the variants identified in Mendelian genes.

Search for MH–associated Novel Genes and Variants Using Next Generation Sequencing

When applied to studies of rare monogenic or oligogenic diseases, whole exome sequencing allows unbiased, not based on any a priori hypothesis, screening of the coding sequence of all of a patient's genes. On the other hand, whole exome sequencing generates a large number of variants in multiple genes whose relevance to a specific phenotype is often difficult to ascertain. At present, the use of next-generation sequencing-based targeted sequencing of a restricted panel of genes associated with a disease phenotype seems to be a more practical approach. Targeted sequencing has a higher coverage (up to 99%) and accuracy and therefore higher sensitivity than whole exome sequencing. Targeted sequencing panels can be also supplemented with Sanger sequencing for regions poorly covered by next-generation sequencing. This gives an added level of coverage and reduces the potential for false-negative results.

Targeted sequencing of panels of genes implicated in excitation–contraction coupling, alongside whole exome sequencing, holds great promise for finding a genetic cause of MH in cases where no RYR1 and CACNA1S variants were found. Four recently published studies used next-generation sequencing-based whole exome sequencing and targeted sequencing of panels of genes with a potential involvement in excitation–contraction coupling, skeletal muscle calcium homeostasis, or immune response, as well as targeted RYR1 and CACNA1S gene sequencing to search for MH–associated variants and novel MH genes in cohorts of unrelated MH–susceptible patients.[34,36,39,40] However, these first studies using next-generation sequencing for identification of MH–associated variants did not result in the discovery of novel candidate genes. They confirmed the findings of previous studies, where Sanger sequencing was used for MH variant screening, namely, that variants in RYR1 and, to a lesser degree in CACNA1S, are associated with MH in the majority of MH cases. Known MH pathogenic variants comprised about 30% of the identified variants, and the remaining variants were variants of unknown significance. These studies also found that up to 50% of MH–susceptible individuals do not carry potentially pathogenic variants in either MH gene, corroborating previous evidence.

Rare variants in several additional genes (CACNB1, CASQ1, SERCA1, CASQ2, and KCNA1) encoding proteins involved in calcium homeostasis in skeletal muscle have been identified using next-generation sequencing[40] and Sanger sequencing.[41,42] However, these variants will remain variants of unknown significance until functional assays are developed to validate their role in MH susceptibility.

Current Approaches to Characterization of Potentially Pathogenic RYR1 and CACNA1S Variants

To increase the sensitivity and specificity of genetic testing for MH and expand the number of pathogenic variants that can be used in clinical genetic testing, all MH–associated variants have to be validated at the genetic level, as well as functionally.[33] Advances in our current knowledge about genetic variation within MH genes have prompted changes in the approach to genetic characterization of the variants. Two approaches, genetic and functional characterization, are used in combination to assess the pathogenic effect of novel variants.

Genetic Characterization

Today, instead of genotyping hundreds of control individuals to exclude common variants or neutral polymorphisms, estimation of the variant allele population frequency can be done by searching publicly available variation databases containing data from sequencing more than 67,000 of human exomes (e.g., the Short Genetic Variations database [http://www.ncbi.nlm.nih.gov/SNP; accessed April 8, 2017], the 1000 Genomes project [http://browser.1000genomes.org; accessed April 8, 2017], the NHLBI Grand Opportunity Exome Sequencing Project [https://esp.gs.washington.edu/drupal/; accessed April 8, 2017], the exome variant server [http://evs.gs.washington.edu/EVS/; accessed April 8, 2017], and the Exome Aggregation Consortium Browser [http://exac.broadinstitute.org]; accessed April 8, 2017). Pathogenic variants are likely to have a minor allele frequency not higher than 0.001.[34] However, because the majority of identified variants of unknown significance in RYR1 are rare and have frequencies lower than 0.001, a low frequency of a variant cannot serve as a predictor of its pathogenicity.

The effect of a variant on protein function and/or stability can be assessed using bioinformatics prediction software tools, such as SIFT,[43] PolyPhen-2,[44] and CADD.[45] These prediction tools use protein sequence information and annotations to protein functional domains to compute predictions with relatively low false-positive and false-negative error rates.[46] However, such predictions should be taken with caution. Different prediction tools use different prediction algorithms and different input data sets (disease-associated mutation sets and neutral variation sets identified in the same protein and available from variation databases, such as the Human Gene Mutation database, Online Mendelian Inheritance in Man database, and the Short Genetic Variations database), and their predictions might be discordant. In addition, the bioinformatics tools are based on imperfect algorithms and on imperfect databases.[47] A recent study compared the predicted and actual consequences of missense mutations and found that half of the de novo or low-frequency missense mutations found by genome sequencing and inferred as deleterious correspond to nearly neutral variants that have little impact on the clinical phenotype of individual cases.[48] Similarly, a significant proportion of RYR1 sequence variants in the human gene mutation database classified as "disease-causing mutations" was found to be benign, probably benign, or as being of unknown pathogenicity.[37] The sensitivity of commonly used bioinformatics prediction tools for RYR1 and CACNA1S has been estimated[46] at 84 to 100% with specificity of 25 to 83%. Therefore, other approaches such as segregation analysis and functional studies are necessary to accurately differentiate clinically relevant variants from neutral variants.

Another challenge in genetic characterization of a variant of unknown significance is the small size of the families. It is not always possible to perform a meaningful analysis of segregation of the variant with the disease phenotype (susceptibility to malignant hyperthermia, MHS) and to generate sufficient statistical power even when combining data from several families carrying the same variant.[33,34]

Functional Characterization

Functional characterization of candidate RYR1 variants remains a key component of their validation. MH–associated RYR1 variants are dominant gain-of-function variants. They render the RyR1 channels hypersensitive to depolarization and pharmacologic agonists or lead to greater depolarization-induced Ca2+ influx into the muscle cell.[2,32] The effect of each MH candidate variant on RyR1 function should be assayed in one of the recombinant in vitro expression systems, HEK293 cells, or myotubes of the dyspedic/dysgenic mouse (RYR1/CACNA1S knock-out).[49,50] These systems use expression of a rabbit or human RYR1 cDNA construct with incorporated variants and measure the properties of expressed channels. The advantage of in vitro systems is the defined cDNA and the standardized genetic background of the recipient cell line. In view of the large number of private familial variants found to date, the revised European Malignant Hyperthermia Group guidelines have removed the need for mandatory description of the variant in more than one family, if functional characterization is done using the more rigorous genetic manipulation of heterologous or homologous expression systems.[33]

Systems using ex vivo expression utilize tissues from MHS patients with characterized RYR1 variants such as myotubes, microsomal sarcoplasmic reticulum preparations from muscle biopsies, or lymphoblasts.[33] Assays of RyR1 function in ex vivo systems are controversial, because they assume that the identified gene variant is the only variant present, when this may not be the case. The compromise presented in the European Malignant Hyperthermia Group guideline is the stipulation that ex vivo analyses should be done on samples from at least two unrelated patients with the same variant to reduce the likelihood of confounding genetic factors.

The first knock-in mouse models of MH carrying RYR1 variants analogous to the MH pathogenic human variants Tyr522Ser[51] and Arg163Cys[52] and a mouse model of central core disease carrying an equivalent of the human uncoupling central core disease mutation Ile4898Thr[53,54] allowed in vitro and in vivo functional studies of these mutations in fully differentiated adult muscle fibers. However, generation of mouse models for validation of each of more than 150 MH–associated variants is not realistic. To circumvent this obstacle, a promising novel approach has been developed to study the function of RyR1 mutant channels.[55] Using localized in vivo electroporation, Lefebvre et al.[55] have expressed constructs of RyR1 N-terminally tagged with enhanced green fluorescent protein carrying MH variants in fully differentiated normal mouse muscle fibers and found that the results were consistent with those obtained for MH variants in previous studies. They showed that expression of the RyR1 channels carrying MH mutations, Tyr523Ser, Arg615Cys, or Arg2163His, was associated with an increased Ca2+ release in response to depolarization, whereas expression of the central core disease mutant, Ile4897Thr, resulted in a reduction of Ca2+ release compared to nonexpressing regions of the same muscle cell. These results indicate that in vivo expression in adult mouse muscles might serve as a novel technique for assessment of functional properties of mutant RyR1s.

Some of the RYR1 variants have been already functionally and genetically characterized and found to be likely or very likely pathogenic. The revised European Malignant Hyperthermia Group guidelines recommend an individual carrying a potentially MHS–associated variant to be considered as being at risk for MH until contracture testing can be performed.

Cryoelectron Microscopy and X-ray Crystallography Contribution to Functional Assessment of RYR1 Variants

Recent determination of crystal structures for the N-terminal domains together with the development of cryoelectron microscopy (cryo-EM) maps of the full-length RyR at nanometer resolution allowed elucidation of the three-dimensional architecture and domain organization of RyR1 and facilitated modeling interactions between its N-terminal, central, and C-terminal domains. These studies revealed how small conformational changes in the cytoplasmic domains, induced by the binding of RyR regulators, are transmitted to the C-terminal domain regulating the channel opening.[55–58] They showed that RyR1 channel opening coincides with subtle changes in the cytoplasmic domain that affect interfaces between individual RyR1 subunits. Mapping the MH/central core disease variant hot spots and individual disease-related variants onto the high resolution structure of RyR1 domains helped reveal mechanisms by which disease-related RYR1 variants might disrupt RyR1 function.[59,60] Because most of the N-terminal MH–associated variants are mapped onto the interfaces between N-terminal domains or at interfaces between subunits in the tetrameric channel, it is probable that those variants weaken the interdomain interactions, thus lowering the energetic barrier to channel opening.[61–63]

The high resolution (near 3Å) cryo-EM images of the transmembrane region that contains the ion conducting pore revealed the presence of the six membrane-spanning helices (S1 to S6) of each RyR1 subunit and allowed mapping of the majority of central core disease-associated variants to the pore-forming domain[64,65] (Figure 1).

Figure 1.

(A) A schematic illustration of the N-terminal domains docked in the pseudoatomic model of the RyR1 tetramer. The RyR1 N-terminal domain (NTD), corresponding to malignant hyperthermia (MH) hot spot 1, is composed of three subdomains: A, B, and C. Interactions among the domains A, B, and C on the same RyR1 subunit together with the interactions between domains of the neighboring subunits are involved in the global conformational RyR1 transmissions that control effector induced channel gating. The variants impair the domain–domain interactions and thus would cause the RyR1 channel dysfunction. (B) A schematic model of domain organization in a RyR1 monomer, composed of the NTD, the central domain, and the channel domain. Each domain consists of several interconnected subdomains. The channel domain consists of the six transmembrane fragments (S1 to S6), and pore helices with the selectivity filter (SF), the linker helix of S4 and S5, the voltage sensor–like domain (VSL), and the C-terminal domain (CTD). Binding Ca2+ to the central domain initiates a cascade of conformational transmissions via allosteric intradomain and interdomain interactions from the central domain to the NTD, to the VSL, the CTD and S6, ultimately inducing opening of the channel.64 Together with the amino acids forming the ion channel—the pore helix and the selectivity filter (amino acids 4,894 to 4,900); the S4–S5 linker (amino acid 4,830 to 4,841); Gly4934, which serves as a "hinge" for the outward movement of the helix S6, and the CTD (amino acids 4,957 to 5,033) are all critical for RyR1 channel gating (modified with permission from Wei et al.).64

Ramachandran et al.,[65] by using homology modeling and high resolution cryo-EM data, succeeded in identification a novel interface between the pore-lining helix (amino acids 4,912 to 4,948) and a S4–S5 linker helix (amino acids 4,830 to 4,841) and showed that this interface controls RyR gating. They built structural models for the RyR1 membrane-spanning domains based on the alignment between RyR1 and two other ion channels with known crystal structures, docked the structural models onto the cryo-map, and showed the close fit between them, indicating that their structural model is suited to model interactions involved in RyR1 gating. Using the models of membrane-spanning domains in the open and closed state, they showed the S4–S5 linker helix interacts with the S6 helix and thus plays a role in gating. Based on this model, they computationally predicted the effect of several variants within S4–S5 linker on RyR1 gating. They further expressed in HEK-293 cells recombinant wild-type and mutant RyR1 channels carrying the same variants and used single-channel experiments to characterize the channels. The effect of each of the variants on channel gating (activating for some and deactivating for other variants) was similar to that predicted by their structural models, thus confirming the role of the S4–S5 linker helix in RyR1 gating.

Generation of structural models of the open and closed states of RyR1 facilitated comparison of computational impact predictions of C-terminal variants with the results of single-channel experiments and identification of amino acid residues in the predicted pore-lining helix and a linker helix that are important for channel gating.[65,66]

Future studies will further refine the three-dimensional RyR1 structure and elucidate the complex molecular mechanisms involved in RyR1 channel regulation and function, thus allowing a more reliable computational prediction of the functional impact of newly discovered variants.

Incidental Findings

Whole exome sequencing generates a large number of variants in multiple genes, and some of those variants might be of clinical relevance to a condition that is different from the original clinical condition for which whole exome sequencing was offered. Such variants are called incidental findings. The American College of Medical Genetics and Genomics has included MH genes RYR1 and CACNA1S among the list of clinically relevant genes, whose potentially pathogenic variants should be reported as incidental findings.[67] Reporting of incidental findings, however, has created clinical and ethical challenges.[47,67–69] The American College of Medical Genetics and Genomics guidelines on reporting the incidental findings emphasize the need for accurate assessment of clinical and research evidence supporting a variant's pathogenicity before reporting it to a patient. The guidelines also caution against excessive reliance on in silico predictions of pathogenicity in the diagnostic context. In light of the prevalence of rare variants in RYR1 and CACNA1S and the difficulty in assessing their pathogenicity, it is likely that a significant number of patients undergoing whole exome sequencing for non–MH indications will be labeled as potentially susceptible to MH when only a small minority will be at risk.

Comments

3090D553-9492-4563-8681-AD288FA52ACE

processing....