Molecular and Genetic Therapies

Structure of the Genome

The genetic “code” is composed of deoxyribonucleic acid, a double helix with two complementary strands. Each strand is composed of a chain of nucleotides, the basic structural unit of DNA. A nucleotide is made up of a phosphate group, a sugar (deoxyribose), and an organic base. The bases are of four types: two pyrimidines: cytosine (C) and thymine (T), and two purines: adenine (A) and guanine (G). The double helix is formed by hydrogen bond pairing of the nucleotide bases of two DNA strands in an orderly manner, where A pairs with T of the complementary strand, and G with C (Watson-Crick base pairing). DNA is packed into chromosomes around histone proteins ( Fig. 11.1 ). The human genome consists of the full set of DNA included in the 23 pairs of chromosomes (one set inherited from each parent). Genes are the basic physical and functional units of the genome, composed of sequences of nucleotides that code for distinct proteins ( ). The human genome contains 19,000 to 20,000 genes ( ). An exon is the portion of the gene that is translated into protein. In each exon, a sequence of three nucleotide bases (the triplet codon) codes for one amino acid; the sequence of triplet codons determines the amino acid sequence of the protein. The unexpressed portions of a gene in between exons are termed “introns.” “Exome” refers to the sequence of all exons of all protein coding genes. Although comprising only 1%–2% of the genome, the exome carries an estimated 85% of mutations that cause single gene disorders ( ).

Inheritance Patterns

Mendelian disorders, named after Gregor Mendel, are disorders caused by single gene mutations (monogenic). They can be due to a change in a single base pair of a nucleotide coding for a protein (point mutations, substitutions), the loss (deletion) or gain (duplication) of a single or multiple base pairs. Many genetic NMDs fall under this category. An allele refers to an alternate form of a gene present at the same location on a chromosome. In the human genome, there are two alleles for each gene, one on each chromosome ( ). Genetic disorders may also be caused by variations in the number or structure of a chromosome. Aneuploidy refers to a missing (e.g., Turners syndrome, loss of an X chromosome) or an extra (e.g., trisomy 21) chromosome. Structural chromosomal variations include deletions, duplications, insertions, inversions, or translocations of a segment of a chromosome. Multifactorial diseases such as diabetes mellitus are caused by an interaction among genetic, environmental, and behavioral factors ( ).

The basic modes of inheritance for monogenic disorders are autosomal dominant (AD), autosomal recessive (AR), X-linked dominant, X-linked recessive (XL-R), and mitochondrial. AD disorders manifest when a mutation is present in one copy of the gene (heterozygous mutations). They tend to occur in every generation, although phenotypic variation and “mild” phenotypes result in the appearance of “skipped” generations (incomplete penetrance). The affected individual has an affected parent of either sex. AR disorders occur when mutations are present in both copies of the gene (homozygous mutations, with both parents unaffected, each carrying one copy of the mutated gene). X-linked disorders can be inherited either dominantly or in a recessive manner. Since males have only one X chromosome, either a dominant or recessive mutation will result in manifest disease. In females, recessive X-chromosome mutations require two copies for manifest disease and one copy results in carrier state ( ). Female carriers of XL-R mutations can have mild symptoms (manifest carriers). Other, more complex inheritance patterns such as compound heterozygosity (two different mutated alleles at a particular gene locus) and double heterozygosity (two different mutated alleles at two separate genetic loci) are also described. Mutations can occur spontaneously (de novo) in the absence of a family history ( ).

Trinucleotide repeats are sets of three nucleotides present in succession in varying copy numbers in the genome. Trinucleotide repeat expansions, where the number of triplets is greater than normal, underlie several neurologic disorders and NMD (e.g., DM1, with an unstable CTG trinucleotide repeat in the 3’ untranslated region of the dystrophia myotonica protein kinase gene on chromosome 19q) ( ). In FSHD type 1 (FSHD1), a contraction of the D4Z4 region on the Double Homeobox (DUX4) gene on chromosome 4 results in hypomethylation of the D4Z4 region and aberrant expression of DUX4 ( ).

Although the diagnosis of an inherited NMD is relatively simple in the presence of a classical phenotype and a family history, the diagnosis can be challenging when there is an overlap in the phenotype of more than one distinct disorder, as in LGMD. Diagnosis is also not straightforward in apparently sporadic cases, where the mode of inheritance may be any pattern, or due to a de novo dominant mutation.

Methods of Genetic Testing ( Table 11.1 )

A mutation is defined as a permanent change in the nucleotide sequence, whereas a polymorphism is defined as a variant with a frequency above 1%. These terms are often incorrectly assumed to imply they are disease causing and benign, respectively. Hence, the term “variant” has been recommended to replace both terms ( ). In this review, we use these terms interchangeably. Variants have further been classified into five groups, based on their relevance to Mendelian disorders: (i) pathogenic, (ii) likely pathogenic, (iii) uncertain significance, (iv) likely benign, and (v) benign ( ).

In 1977, Sanger described a DNA sequencing technique ( ) that was used to sequence the human genome in the Human Genome Project ( ). Briefly, the process involves incubating strands of the DNA sequence to be identified with radiolabeled primers (short single stranded nucleic acids that initiate DNA synthesis). In addition, DNA polymerase (an enzyme that synthesizes a strand of DNA complementary to the original strand), deoxynucleotide triphosphates (dNTPs; the building blocks of the complementary DNA strand), and dideoxynucleotide triphosphates (ddNTPs; which prevent addition of dNTPs by DNA polymerase and thus terminate DNA strand synthesis) are incorporated. There are four sets of sequencing reactions, each with a different ddNTP, carrying one of the four bases A, T, G, or C. In each sequence, as the reaction proceeds, synthesis of the complementary DNA strand is interrupted by a ddNTP and complementary DNA strands of different lengths are thus formed. These strands from each of the four reactions are separated by size using polyacrylamide gel electrophoresis. An autoradiograph of the DNA bands thus separated is used to read the corresponding DNA sequence off the gel, and the complementary sequence is then determined, providing the DNA sequence of the original strand ( ). Traditional genetic testing for NMD was based on gene-by-gene sequential testing, driven by clinical hypotheses and knowledge of the most common mutations for specific disorders. However, Sanger sequencing requires a large amount of the DNA template and is time consuming and expensive. It also lacks sensitivity for large deletions/duplications and for repeat expansions ( ).

Southern blotting detects deletions or duplications of one or more exons of the gene. DNA is digested by restriction enzymes that recognize a specific site. The fragments are sorted by size by agarose gel electrophoresis and transferred to a nylon or nitrocellulose membrane. Single-stranded DNA probes that are labeled with radioactive tracers are then added to the membrane. The probe hybridizes (associates) with the complementary fragment of interest. The labeled probes are detected by autoradiography ( ; ).

Fluorescent in situ hybridization (FISH) locates the position of specific DNA sequences on chromosomes and is useful for the analysis of chromosome deletions, duplications, and translocations using a probe that is targeted to the sequence of interest. Short sequences of single-stranded DNA (probes) are tagged with fluorescent markers to detect chromosomal DNA that has a complementary sequence. Locus-specific probes bind to complementary areas of a chromosome, such as a gene, or a repetitive sequence such as a centromere or telomere to detect abnormal duplication of a gene or chromosome ( .)

Single-nucleotide polymorphisms (SNPs) are the most common type of genetic variation and are composed of sequence differences at single nucleotide or base pair sites. They occur frequently in normal individuals. SNPs have been implicated as disease modifiers in several neuromuscular diseases, including DMD, SMA, ALS, and LGMD ( ). Copy number variants (CNVs) are changes in the number of copies of specific segments of DNA, either deletions or duplications. A CNV is defined as a DNA segment that is 1 kb or larger and presents in variable copy numbers in comparison to a reference genome ( ). Although found in normal individuals, they also contribute to diseases such as Alzheimer disease and Parkinson disease ( ; ). Oligonucleotides are short synthetic sequences of approximately 13–25 nucleotides ( ). DNA microarrays are slides that have multiple microscopic spots in defined positions, with each spot containing a known DNA sequence or gene. They are also referred to as gene chips or DNA chips ( ).

SNP microarray analysis (DNA probes that target SNPs) and oligonucleotide microarray analysis (short sequence probes of single-stranded DNA or ribonucleic acid [RNA] that hybridize to target complementary nucleic acid sequences) detect CNVs, small deletions, and duplications and uniparental disomy but not point mutations ( ; ; ). In contrast to comparative genomic hybridization (CGH) arrays, SNP arrays measure the absolute fluorescence probe intensities of the patient sample compared to the intensities of multiple normal control samples that are individually run, normalized, and combined to create a reference ( ).

CGH array is based on in situ hybridization of normal metaphase chromosome spreads with a test set and a reference set of whole genomic DNA that are differentially labeled using fluorescent markers, usually red and green. The two labeled sets are denatured into single-stranded DNA, mixed in a 1:1 ratio and hybridized in situ to normal metaphase chromosome spreads. The labeled DNA binds to target sequences in the metaphase spread in a concentration-dependent manner. CGH detects chromosomal imbalances: CNVs at the chromosomal level (changes in the number of chromosomes, ploidy) and deletions or duplications of large chromosome areas (extra or missing chromosomal material) ( ). For instance, in a deletion of chromosome 17p, there will be greater fluorescent labeling of the reference color than test color, and vice versa in the case of 17p duplication. Unlike FISH, which is based on the examination of a single target and requires prior knowledge of the region being tested, CGH can be used to quickly scan an entire genome for imbalances. Array CGH uses microarrays rather than metaphases. Since DNA probes are much smaller than metaphase chromosomes, array CGH has a higher resolution (i.e., ability to detect smaller changes). The resolution depends on the probe size and the genomic distance between two probes, since CNVs in the segment between two probes cannot be detected. This technique is equivalent to thousands of FISH assays and can detect submicroscopic chromosomal alterations.

Multiplex ligand-dependent probe amplification (MLPA) is useful in identifying large deletions or duplications that may be missed by gene sequencing. MLPA can analyze up to 50 DNA sequences in a single reaction and detect CNVs and small intragenic rearrangements. In MLPA, which is a multiplex polymerase chain reaction (PCR)-based assay, multiple probes that are specific for the exon of interest are used to evaluate the number of copies of the DNA sequence ( ). MLPA does not detect small mutations, inversions, translocations, and CNVs.

Conventional genetic testing for monogenic disorders involves several sequential steps including genome-wide linkage to identify mutations that cosegregate within affected individuals, positional cloning, and, finally, targeted candidate gene sequencing. Linkage is the tendency for genes and their resulting traits to be inherited together because of their proximity on the same chromosome. Linkage analysis estimates the distance between two genetic loci. The statistical probability of two genetic loci being close enough to be inherited together is the logarithm of the odds (LOD) score. A LOD score of 3 or greater generally assumes that the loci are linked. In screening affected and unaffected family members, linkage analysis identifies disease causing mutations by testing for a locus that cosegregates with a known genetic marker in affected but not in unaffected members ( ; ). Positional cloning then locates the position of the disease-associated gene along the chromosome ( ). Targeted sequencing of the identified candidate genes is then performed. In AR disorders, autozygosity mapping, which identifies regions of the genome that are homozygous in affected individuals but not in unaffected family members, is followed by gene sequencing to identify the causal mutation ( ; ). However, these techniques require several affected family members and are therefore not useful in rare disorders when only a few cases are available or in sporadic cases due to de novo mutations. Additional limits of these approaches include phenotypic heterogeneity (different phenotypes with a mutation in the same gene, e.g., mutations in the dysferlin gene causing both LGMD and Miyoshi myopathy) ( ), locus or genotypic heterogeneity (similar clinical presentations due to mutations in different genes, e.g., mutations in both anoctamin and dysferlin genes causing LGMD and Miyoshi myopathy phenotypes), clinical patterns merging over time, and incomplete penetrance ( ). Once the genetic mutation for a given disease is identified and reported in the literature, testing for common mutations becomes commercially available. However, as the number of genetic NMDs expands and genetic heterogeneity is increasingly recognized, single gene approaches are no longer efficient, and “reflex” panels sequentially testing for common mutations followed by less common ones are often performed.

In 2004, next generation sequencing (NGS) methods, also termed “massively parallel sequencing,” using high-throughput platforms were introduced, making it possible to sequence several thousands of pieces of DNA simultaneously. NGS platforms can complete genome sequencing within a few weeks at a fraction of the cost of Sanger sequencing. Whole genome sequencing (WGS), as the name implies, analyzes the entire genome, including coding and noncoding regions. NGS techniques can also be used for whole exome sequencing (WES) ( ). WES refers to sequencing of all exons of protein coding genes and may include exon-flanking intronic sequences to a varying extent ( ). WES is often sufficient to identify the genes underlying Mendelian disorders and reduces the time and costs associated with WGS ( ). Targeted WES is used to sequence a panel of genes that are relevant to a specific disease group.

In NGS, genomic DNA is extracted and broken down into short segments of about 350 bp. Adaptors, or priming sequences of DNA, are added to the ends of these fragments. The segments are then mixed with probes that are complementary to all known exons in the genome, which are consequently captured. The remaining DNA is washed away, leaving behind the enriched exome sequences for amplification followed by massive parallel sequencing of the enriched, amplified library. Fluorescent nucleotides are added during sequencing, resulting in their incorporation into strands of DNA. The incorporation of each fluorescent nucleotide into a sequence is detected using laser activation. Millions of short sequence reads are mapped to a reference genome, and variants within the genome are identified ( ; ). These variants are then filtered by various techniques. The first step is to cross-reference the variants against publicly available databases of exomes from unaffected individuals and thus exclude nonpathogenic variations. An example of such a reference databases is ClinVar ( ; ). The remaining variations are then filtered in several ways: the mode of inheritance, segregation within affected family members, etc. ( ). Because there are normal ethnic and racial variations in the genome, population-specific databases are being developed ( ; ; ).

Strategies for genetic testing depend upon the pretest probability of a diagnosis based on clinical features, family history, extramuscular manifestations such as cardiopulmonary involvement, etc. Sequencing of a single gene of interest or sequencing of selective exons of a gene is efficient if the pretest probability of a disease is high and the majority of pathogenic mutations are due to deletions or duplications. Approximately 5% of Mendelian disorders are attributable to gene deletions or duplications ( ). For example, in DMD, the large size of the dystrophin gene lends itself to a relatively high spontaneous mutation rate. The most common mutations are large (>1 exon) deletions (≈68%), large duplications (≈11%), followed by small deletions, insertions, and point mutations (≈20%) ( ; ). The first step in genetic testing of suspected dystrophinopathy is usually deletion/duplication analysis because of its ease and cost-efficiency, followed by sequence analysis of DMD coding regions and then whole-exome testing ( ). If a repeat expansion disorder is suspected, PCR-based methods and Southern blotting are the standard methods of diagnosis.

More frequently, clinical assessment narrows the differential diagnosis to a group of disorders rather than providing a single diagnosis. In this situation, Sanger sequencing of multiple individual genes is neither time nor cost effective, and the diagnostic strategy has rapidly moved toward targeted WES or WES. Targeted WES is now more easily available commercially. Panels of genes relevant to the disease group, e.g., “hereditary demyelinating neuropathy panel,” “neuromuscular panel,” etc., sequence the exons of several genes associated with those disorders effectively, efficiently, and cost-effectively. However, the usefulness of these panels depends on the number of genes tested. As more genes are discovered, panels have to be updated. The additive yield of WES in muscular dystrophies ranges from about 30% to 70% and averages at approximately 40%. In this group, WES may pick up unsuspected, treatable disorders such as acid maltase deficiency (Pompe disease) ( ; ; ; ; ; ; ; ; ; ; ; ). A diagnostic yield of up to 75% is described in populations with a high rate of consanguinity ( ; ). In a study of pediatric NMD, both targeted WES with a neuromuscular disease panel and WES were more efficient (diagnostic yield 42%–75% and 79%, respectively) compared to conventional algorithmic diagnosis (muscle biopsy with immunohistochemistry [IHC] and Western blot, followed by candidate gene sequencing) and significantly reduced the cost per diagnosis from US$16,495 to US$3,706 for the neuromuscular disease panel and US$5,646 for WES ( ).

The most common inherited peripheral neuropathy, CMT1A, is associated with duplication of the 17p PMP22 gene in over 70% of patients. Over 80% of cases are diagnosed by traditional methods ( ; ). Along with PMP-22 duplication, variants in the genes encoding gap junction 1 (GJB1), mitofuscin 2 (MFX2), and myelin protein-0 (MPZ) causing CMT-X, CMT-2A, and CMI-1B, respectively, account for >90% of genetically diagnosed neuropathies ( ; ; ; ). Despite this, only 60% of clinically diagnosed CMT patients receive a genetic diagnosis ( ; ; ). With targeted WES, a diagnostic yield ranging from 19% to 53% has been reported, including detection of known variants and descriptions of novel variants ( ; ; ; ; ; ). In axonal neuropathies with or without family history, the yield is much lower, approximately 2% without a family history and 7% with family history of neuropathy ( ). The yield depends upon the panel used, and custom panels may have to be created based on the clinical pattern. Other factors determining the yield include the number of genes in the panel and how exhaustive public reference databases are. The use of “trios,” in which the affected individual and their parents undergo WES, can be useful, when available, to segregate the variant between affected and unaffected members and to identify de novo mutations in AD disorders. An increase in diagnostic rate from 40% for single WES to 60% for trio WES has been reported ( ).

The chromosome 9 open reading frame 72 (C9orf72) hexanucleotide repeat expansion is the most frequent gene associated with ALS, present in approximately 39%–50% of fALS and 7% of sALS ( ; ). This mutation, along with mutations in cytosolic copper-zinc superoxide dismutase (SOD1; 20%–25% of fALS) ( ; ), transactive response DNA binding protein 43 (TARDP/TDP-43, 5%) ( ; ), and fused-in-sarcoma gene mutations (FUS; 5%) ( ; ), accounts for 60%–90% of all fALS and 10% of apparent sALS. There is variation in clinical practice regarding genetic testing in ALS; Sanger sequencing of individual genes is commercially available ( ). Recommendations are available regarding genetic testing of patients and presymptomatic genetic testing of family members ( ; ). Depending on the availability and cost, testing for C9orf72 repeat expansions may be offered to all patients with ALS of European descent with or without family history and in sALS with early onset (<50 years) because of the frequency of the mutation in this population, because of the clinical and counseling implications of the results, and because therapies for C9orf72 ALS are in the pipeline. In early-onset sALS or fALS with negative C9orf72 testing, targeted WES is the next step ( ).

It is important to understand the limitations of targeted WES. Incomplete enrichment or “coverage” of the sequences of interest is one such limitation. Coverage may be defined as depth, or the average number of reads that align to, or “cover,” known reference bases, or, more simply, how many times the base or mutation has been “read” by the sequencer. Uneven capture of nucleotide sequences can result in some exons with low coverage or in reads of DNA outside of the exome, that is, “off-target” reads. Low coverage (<5 reads per base) may result in a sequencing error being falsely interpreted as a mutation, or a true mutation being missed ( ). Minor misalignment of the sequenced DNA to the reference genome may cause small base substitutions or insertions/deletions to be missed, or suggest new mutations ( ). Disease-causing mutations in noncoding parts of the genome such as introns, regulatory components (promoters and enhancers), and splice sites will not be detected ( ). Approximately 15% of Mendelian disorders are attributed to such mutations ( ). Importantly, repeat disorders are generally not detected by exome sequencing. In the neuromuscular context, these include the more common adult muscular dystrophies, DM1, DM2, and FSHD. Cross-referencing using reference databases may omit pathogenic mutations because carriers who have not yet expressed the disease or have no symptoms due to reduced penetrance may be present in the reference panels ( ; ). The underrepresentation of Latin American, African, and Asian populations in reference databases compared to European populations limits the generalizability of these databases ( ; ). WES is not optimal to detect CNVs, although newer platforms have had some success in this regard ( ; ; ). Finally, a major challenge with WES is the detection of variants of unknown significance (VUSs), or variants that cannot be classified as either benign or pathogenic based on available information from reference databases and reported pathogenic mutations. These may represent rare, individual nonpathogenic variations or may be pathogenic but not yet reported. Further analysis is required to assess these variants: in silico prediction models, family screening of affected and unaffected members to segregate the variant in only affected members, and functional studies. Functional studies assess the significance of a VUS by evaluating its effect on the function of the protein product of the gene. This includes muscle biopsy with IHC and Western blot to assess for the reduction or absence of the protein product. Functional assays of the variant in animal models are used in the research setting but are not practical in the clinical setting.

Traditional evaluation of genetic NMD typically consists of muscle or nerve biopsy followed by genetic testing. This paradigm has shifted with the availability of NGS. For instance, because of the characteristic clinical picture of DMD or SMA, genetic testing is the preferred initial test almost universally. This eliminates the need for an invasive muscle biopsy in most of the patients. In a study of pediatric neuromuscular diseases, initial muscle or nerve biopsy provided a diagnostic yield of 34%. Genetic testing before biopsy showed pathogenic variants in 7.6%, and when genetic testing was directed by muscle biopsy findings, the yield rose to 47%. The authors concluded that the yield of genetic testing was higher when done after biopsy ( ). Other studies suggest the converse: that initial genetic testing is more efficient and cost effective ( ; ; ; ). Overall, in heterogeneous NMD, NGS should probably be the initial diagnostic test if available and affordable. The use of targeted WES vs. WES depends on the clinical phenotype and the number of genes in the targeted WES panels. Muscle biopsy IHC and Western blot provide information about functional defects not only in the setting of VUS but also when no pathogenic mutation is identified and will then direct further genetic testing for confirmation of the diagnosis. Muscle biopsy findings may also be useful to understand the pathogenesis of novel mutations. In dystrophinopathies, the distinction between DMD and BMD is a clinical one, because up to 10% of cases do not follow the reading frame hypothesis (i.e., in-frame deletions manifest a DMD phenotype and out-of-frame deletions manifest a BMD phenotype) ( ; ). The amount of dystrophin on biopsy is therefore a better correlate of the clinical phenotype and a surrogate outcome measure for clinical trials of therapies seeking to restore dystrophin.

WGS will likely increase diagnostic yield due to its ability to detect mutations in noncoding regions and chromosomal rearrangements, but the incremental yield compared to WES remains to be established. In a study of a heterogeneous group of genetic disorders in children, WGS had an additive yield of 17% over conventional genetic testing (41% and 24%, respectively). A part of this cohort also underwent research-based WES; WGS identified all the variants that were identified by WES and, in 26% of cases, detected pathogenic variants that WES did not ( ).

Pretest and posttest genetic counseling is an important part of genetic testing. Pretest counseling for WES should provide the patient with results that may be obtained: a diagnosis may be established, a diagnostic variant may not be detected, one or more VUS may be detected, or secondary findings may be reported. The probability of an established diagnosis as described in the literature should be discussed in order to set expectations. The need for subsequent testing including family screening and functional assays such as muscle biopsy or enzymatic assay (e.g., α-glucosidase assay for Pompe disease) should be discussed. Finally, the issue of incidental, known, or expected pathogenic variants (now termed “secondary findings”) has to be discussed. These are pathogenic variants unrelated to the clinical indication for testing. The American College of Medical Genetics and Genomics has published a minimum list of 59 “medically actionable” genes (i.e., disorders with either preventive or treatment strategies) that should be reported and informed to the patient. The three neurologic disorders in this list are tuberous sclerosis, neurofibromatosis, and Wilson disease. Patients should be counseled prior to testing regarding the reporting of secondary findings and should be given an opportunity to opt out of receiving these results ( ).