Gained in translation: The power of digging deep into disease models

Abstract

Mutations affecting the SECISBP2 protein necessary for selenocysteine incorporation are linked to human disease, but with a wide range of clinical outcomes. To gain insight into this diversity, Zhao et al. dissect the phenotypic and molecular consequences of two specific mutations in the Secisbp2 gene that partially disrupt selenoprotein synthesis. They observe surprising tissue-dependent effects, emphasizing the complexities of translational science.

Introduction

Selenium is an essential trace element that gets incorporated into 25 proteins in humans as part of the amino acid selenocysteine (Sec)² (1). Because selenium and sulfur are in many respects biologically indistinguishable, the fact that Sec exists is not surprising, but how it gets incorporated into so few specific positions in the entire proteome is a matter of profound complexity that requires a multitude of factors to modify one of the fundamental principles of the genetic code.

Sec is incorporated at specific UGA codons, which would ordinarily specify translation termination (2). To accomplish this, the translation elongation reaction is catalyzed by a unique elongation factor that is specific for Sec (eEFSec), which binds specifically and solely to the Sec-specific tRNA, Sec-tRNA^Sec. Additionally, a unique RNA stem-loop structure, the so-called Sec insertion sequence (SECIS) element, serves as a signal to inform the ribosome which UGA codons are to specify Sec rather than termination. The SECIS element, found in the 3′-UTR of all selenoprotein mRNAs, interacts with SECIS-binding protein 2 (SECISBP2), which is essential for Sec incorporation in vitro (3), and elimination of the gene in mice results in early embryonic lethality (4).

Mutations in Secisbp2 are tightly linked to severe human disease (5). To date, a total of nine families have been found to harbor Secisbp2 mutations, resulting in a variety of phenotypes. The most common outcome is for children to have low thyroid hormone levels due to reduced expression of deiodinase, a Sec-containing enzyme required for production of mature thyroid hormone (6). However, the molecular basis by which mutations in Secisbp2 can lead to such diverse disease phenotypes has been unclear. To study this question, Zhao et al. (7) generated mice carrying two of the disease alleles. One of the mutations, R543Q (R540Q in humans), is found in the so-called Sec incorporation domain (SID) in SECISBP2, which is required for high affinity SECIS binding and eEFSec recruitment (8). The other, C696R (C691R in humans), is found in the conserved L7Ae RNA-binding domain, a domain also found in several ribosomal proteins and small nucleolar RNA–binding proteins. The authors were not able to generate homozygous mutant animals for either allele; this was expected for the C696R allele because the substitution inserts an amino acid with quite different properties in a well-conserved region of the protein and is present only as a compound heterozygous allele in humans. However, this was unexpected for the R543Q mutation because it is homozygous in the patient and causes only a mild phenotype.

The fact that the R543Q allele is apparently lethal in mice but not humans might give one pause about the applicability of the rodent model, but Zhao et al. anticipated they could still learn about how the mutant protein behaves by making tissue-specific mutant animals, focusing on liver and neurons. Consistent with the apparent lethality of an R543Q/R543Q genotype, the livers of mice harboring R543Q as the only copy of Secisbp2 were dramatically deficient in selenoprotein production. However, in vitro analysis of the Sec incorporation activity for the R543Q protein showed significant residual activity. What is the explanation for these paradoxical results? An insightful analysis of protein stability demonstrated that the R543Q mutant protein is significantly less stable in vitro and virtually undetectable in liver tissue. The C696R protein was also undetectable in vivo and completely inactive in vitro, thus copying the phenotype of a knockout mouse.

Analysis of the same alleles in a cortical neuron-specific mutant animal showed quite different results, with an apparently normal phenotype for the R543Q mutant despite the lower expression. Expression of the C696R allele led to a phenotype that mirrored the knockout animals. Upon investigating the R543Q mutant in more detail, the authors unexpectedly observed specific reduction of most selenoproteins, including SELENOW and SELENOM. To gain further insights into this discrepancy, the authors applied ribosome profiling combined with transcriptomics to analyze the translational efficiency for selenoprotein mRNAs, the modulation of global gene expression, and the translational efficiency of mRNAs derived from the mutant alleles. Whereas interpreting the peak patterns of ribosomal profiling data is difficult due to potential biases from library preparation (9), the calculation of translational efficiency at different locations in a given mRNA under different conditions is a robust measure of potential translational regulation, in this case either upstream or downstream of the UGA codon (10). The use of these cutting edge, genomic techniques demonstrated that selenoprotein mRNA read-through was less efficient in the R543Q mutant, and levels of selenoprotein mRNAs were lower, presumably due to degradation when translation is prematurely terminated. Beyond the specific changes in selenoprotein expression, the authors observed changes in gene expression associated with inflammation, allowing them to decipher a pathway that would explain an astrogliosis phenotype that arose in the mice. These results will likely be informative about the etiology of symptoms borne out of SECISBP2 deficiency syndrome (Fig. 1). Additionally, this work has revealed a specific location within the SECISBP2 protein that plays a role in regulating stability, opening up an entirely new line of investigation about how selenoprotein production might be regulated at the level of SECISBP2 stability.

Figure 1.

Application of methods that dig deeply into the molecular basis for allele-specific phenotypes can dramatically increase the value of model systems used to study human disease. In this case, two of the mutations in the SECISBP2 gene that lead to selenoprotein deficiency in humans were studied in a mouse model. The application of a broad spectrum of techniques measuring changes in selenoprotein and associated gene transcripts provides new insight into disease etiology.

Modeling human disease in rodents can be fraught with pitfalls. This may be particularly true in the case of rare diseases for which human data are limited, but there are no superior alternatives. As such, the rigor and diversity of the methodology becomes of paramount importance. The depth of the analysis from Zhao et al., which included immunoblotting, ribosome profiling, and transcriptomics, is a great example of what perhaps should become the state-of-the-art when it comes to phenotypic evaluation. When this type of approach is taken, the concept of translational science is essentially inverted such that studying human disease reveals just as much or more about fundamental mechanisms than about potential treatments.