Molecular dissection of dystrophin identifies the docking site for nNOS

At current count, we human beings have 20,687 protein-coding genes encoded within our DNA (1). The unfortunate reality is that we know very little about the functions of most of these proteins. Because new scientific discovery is built upon the foundation of prior work, it is perhaps obvious that knowledge gaps like these can have compounding effects to slow further scientific advancement. Indeed, imagine the state of molecular biology and biotechnology today if Watson and Crick’s publication of DNA structure had been reported just a few decades earlier. So too is the case for translational research: it is difficult to develop treatments for a genetic disorder when we do not understand the affliction’s pathogenic mechanisms. Thus, the more basic knowledge we possess about the function of a given disease gene and its protein product, the more ammunition we have to design rational therapies. In PNAS, Lai et al. from the Duan laboratory at the University of Missouri provide an exquisite example of a basic study on protein structure and function with potentially important translational implications for therapy of Duchenne muscular dystrophy (DMD) (2).

DMD is a devastating form of muscular dystrophy caused by mutations in the dystrophin gene (DMD). DMD holds many distinctions, but its size is particularly remarkable: the gene comprises 79 exons spread across 2.4 million base pairs, making it the largest human gene and roughly 90-times larger than average. The gene produces multiple protein isoforms, the most predominant of which is a leviathan consisting of 3,685 amino acids (aa) and measuring 427 kDa (3). Although the size and complexity of this gene and protein product may be impressive as a piece of biological trivia to those not working in the neuromuscular field, doing so here has a more immediate purpose: grasping the enormity of dystrophin is necessary to appreciate the impressive work published by Lai et al. in PNAS (2). In their work, the authors perform an exhaustive set of structure/function studies to identify a functionally important 10-aa microdomain buried within the dystrophin protein (representing a mere 0.2% of the entire molecule), and further show that this motif functioned only when properly framed by a specific set of four α-helical peptide sequences. In truth, this work built on prior studies by this group and others in the DMD field over many years, so a brief summary of dystrophin structure/function research will help put the current work in proper context.

Dystrophin mutations were identified as the cause of DMD in 1986 (4). The first clues about its function came from examinations of the primary amino acid sequence, which predicted a large, rod-shaped protein (3). Four major functional domains were identified: at the N terminus is the eponymous N-terminal domain, with the COOH-end harboring the cysteine-rich and aptly described C-terminal domain (Fig. 1) (3). These regions are separated by a long central rod, composed of 24 structurally similar spectrin-type repeats (STRs) and four flexible hinge regions (3) (Fig. 1). Binding sites for other proteins reside in the terminal regions of dystrophin, including motifs that direct critical interactions with cytoskeletal actin and a group of transmembrane, cytoplasmic, and extracellular proteins called the dystrophin glycoprotein complex. Thus, dystrophin links the internal cytoskeleton to the extracellular matrix. These early biochemical data led to the logical conclusion that dystrophin primarily functions to stabilize muscle membranes during contraction, thereby providing mechanical support to the cell. Within this context, the 24 STRs of the dystrophin central rod were often considered as a single unit, linked together to form a shock absorber or a spacer that separated the business ends of the molecule. Although this model still holds true, the field has evolved in the last 15 y or so. We now know that despite their same general structure, individual STRs are not always functionally equivalent, and many possess distinct functions beyond being a generic link in the rod-domain chain (5–10). The study published by Lai et al. in PNAS beautifully underscores this point (2).

Fig. 1.

(A) Structure of dystrophin and utrophin. See text for details. (B, Upper) This microdystrophin protects muscle membranes from damage during contraction and correctly localizes nNOS to the sarcolemma. (B, Lower) Structural depiction of the micro-dystrophin 4 repeat rod domain. Colored barrels indicate alpha helices from each respective STR. H1 and H4 represent hinge domains. The authors used this construct for domain swapping with comparable utrophin microdomains, the positions of which are indicated by roman numerals. The starred IX marks the location of the 10-aa nNOS binding site. This site functions only in a proper structural environment. CR, cystine-rich domain; CT, C-terminal domain; NT, N terminal domain.

The current study builds on previous work by this group, which showed that dystrophin STRs 16 and 17 were required for proper localization of neuronal nitric oxide synthase (nNOS) to the muscle membrane. Why is nNOS important in muscle and muscular dystrophy? Muscles comprise a significant portion of our body mass and require a large amount of energy for proper function. Metabolites and oxygen are brought to the muscles via the vasculature. nNOS primarily functions in regulating blood flow to ensure the metabolic needs of contracting muscles are met. Thus, DMD muscles that lack functional dystrophin, and subsequently have mis-localized nNOS, become susceptible to both mechanical damage and exacerbating blood-flow restrictions (called functional ischemia). Although the mis-localization of nNOS in DMD was described in 1995, the importance of this phenomenon was not fully appreciated until more recently, and the Duan laboratory has led the charge (2, 7, 11–14). This most recent report provides the most detailed characterization yet of the structural elements in dystrophin required for proper nNOS localization and function.

To accomplish this characterization, the authors cleverly use two major elements. The first element is the minimal dystrophin unit previously shown by this laboratory to bind nNOS: a microdystrophin they created called ΔR2-R15/ΔR18-R23/ΔC (Fig. 1). The second critical element is utrophin, an autosomal paralog of dystrophin that contains the same multidomain structure with a structurally homologous 22-repeat rod domain, but lacks the ability to bind nNOS (Fig. 1A). The Duan group then proceeds to methodically swap 14 different microdomains from utrophin into homologous regions of dystrophin, with the goal of identifying the critical regions for nNOS membrane localization. The authors use a viral vector to deliver their domain-swapped microdystrophin/utrophin chimeras to dystrophic mouse muscle (Fig. 1B). Of the 14 constructs they generate, only one (IX; figures 2 and 3 of ref. 2) disrupted the ability of microdystrophin to properly localize nNOS. This small 10-aa sequence, located in dystrophin repeat 17, thus represents the nNOS binding domain of dystrophin. With this sequence, the authors found the proverbial needle in the haystack.

Nevertheless, the authors dutifully sought to confirm their finding, and used a reciprocal experimental strategy to accomplish it. Lai et al. (2) reason that if this 10-aa microdomain is the minimal unit required for nNOS binding, swapping it into utrophin, which is otherwise lousy for nNOS, should give the protein nNOS binding ability. Thus, the authors constructed a microutrophin analogous to the ΔR2-R15/ΔR18-R23/ΔC microdystrophin, but with the nNOS binding motif from dystrophin R17. Following delivery to mouse muscle, they surprisingly found that nNOS was absent from the membrane, despite that fact that microutrophin was abundantly localized there. This finding suggests that additional sequences within R16-R17 are required to support nNOS binding. The authors thus set out to identify these additional sequences.

Each STR is composed of three α-helices. The first α-helix of R17 (R17α1) contains the nNOS binding site. The authors hypothesized that subtle structural differences in the flanking α-helices of R16-R17 could affect the way this R17α1 motif was positioned in 3D space, thereby affecting the ability or inability of nNOS to locate and properly bind the site. To address this question, the authors again used a domain-swapping strategy, but instead of using utrophin sequences, they inserted individual α-helices from dystrophin R18, which is not required for nNOS binding.

The structural elements required for proper nNOS localization should be included in any DMD therapy for which dystrophin restoration is the goal.

Thus, Lai et al. built multiple R16-R17 constructs in which each individual helix (of six possible) was replaced by the corresponding helix from R18 (figures 4 and 5 of ref. 2). Testing this strategy in an in vitro system failed to identify any effects of these substitutions, so the authors went back to their tried-and-true method of in vivo expression using viral vectors. Using this strategy, they found that the 10-aa nNOS binding motif required the α2 and α3 helices from both R16 and R17 for proper function.

This study is noteworthy for several reasons. First, it represents an incredibly large amount of painstaking work. Indeed, the authors generated 48 different constructs for delivery to mouse muscle. Second, Lai et al. encountered unexpected, and at first glance, negative results that could have signaled the end of the study (figures 3 and 4 of ref. 2). Impressively, the authors persisted, formed new hypotheses, and ultimately successfully identified the smallest known structural requirements for a specific dystrophin function to date. Finding a functional motif that is 10-aa long in a protein this size deserves some recognition. Third, the authors underscore the importance of mammalian in vivo assays for structure/function studies. Historically, dystrophin structure studies were performed in vitro, typically using peptide fragments generated in bacteria, yeast, or insect cells. The differential results reported by the Duan group using identical constructs in yeast versus mice should be “exhibit A” for anyone considering the best strategy for performing structure/function studies of any protein in the future.

Finally, this study has important therapeutic implications. Although DMD is caused by the absence of functional dystrophin, not all dystrophin mutations cause severe DMD phenotypes. A milder form of DMD, called Becker muscular dystrophy (BMD), arises from dystrophin mutations that typically cause in-frame, partial deletions of the rod domain. In these cases, smaller, subfunctional versions of dystrophin protein are produced and become properly localized at the sarcolemma. One BMD patient was famously ambulant well into his 60s, despite missing nearly two-thirds of the dystrophin rod domain (15). These truncated dystrophins in BMD patients formed the basis for therapeutic development of DMD using two strategies: dystrophin gene-replacement therapy and exon skipping. For the former, full-length dystrophin is too large to fit into currently available viral vectors, but miniature versions of dystrophin (i.e., microdystrophins) can be packaged and effectively delivered to muscle, with therapeutic benefit (6). For the latter, antisense oligonucleotides can be designed to skip over mutated exons, effectively converting a DMD transcript to a BMD one (16). The data from the Duan laboratory, reported in this paper and previous studies, demonstrates that the structural elements required for proper nNOS localization should be included in any DMD therapy for which dystrophin restoration is the goal.