A variety of cytoskeletal linking systems provide lateral connectivity between adjacent myofibrils, between myofibrils and the SR, and between peripheral myofibrils and the sarcolemma.² Perhaps the best-characterized of such networks is the desmin intermediate filament network; by binding to the C-terminus of nebulin at the Z-line periphery,³ desmin interconnects and mechanically aligns adjacent myofibrils.⁴ Desmin also links myofibrils to intracellular organelles, including mitochondria and nuclei.⁴ Furthermore, by binding to costamere-associated proteins, desmin connects peripheral myofibrils to the sarcolemma.⁴ In parallel, the giant protein obscurin binds to myomesin and titin in the M-line and small ankyrin 1.5 (sAnk1.5) in the SR, providing a molecular link between the myofibrils and SR.⁵ The molecular organization and function of the extrasarcomeric cytoskeleton is a rapidly evolving area of skeletal muscle research, extending far beyond the desmin and obscurin systems, with recent studies implicating multiple populations of actin filaments comprised of nonmuscle actin isoforms linking myofibrils to one another and to the sarcolemma.

The pristine regularity of skeletal muscle structure renders it an ideal system to study the intracellular sorting and isoform-specific functions of actin in the extrasarcomeric cytoskeleton. Although all actin isoforms share a minimum of 93% sequence identity, they exhibit distinct, nonoverlapping localizations and functions in cells.^6,7 In skeletal muscle, sarcomeric thin filaments consist of skeletal muscle α-actin (α_sk-actin). (The contractile machinery of cardiac and smooth muscle contain their own tissue-specific isoforms of α-actin, which need not concern us here).⁷ In contrast, the inner face of the sarcolemma contains a combination of nonmuscle actin isoforms: cytoplasmic β-actin (β_cyto-actin) and cytoplasmic γ-actin (γ_cyto-actin).^8,9 Our laboratory recently discovered that the SR is similar to the sarcolemma in that it also contains γ_cyto-actin,¹⁰ although whether β_cyto-actin coexists with γ_cyto-actin in the SR remains undetermined. (A second γ-actin isoform, smooth muscle γ-actin, is not expressed in skeletal muscle.⁷) The SR association of γ_cyto-actin echoes previous work that identified γ_cyto-actin colocalization with mitochrondria¹¹ and the appearance of a reticular γ_cyto-actin localization pattern when skeletal muscle is viewed in cross-section.¹² However, until recent work from our laboratory¹⁰ (see below), nothing was known about the in vivo functions of SR-associated γ_cyto-actin or its regulatory proteins.

The pointed ends of muscle and non-muscle actin filaments are capped by tropomodulin (Tmod) isoforms. Tmods are the only known proteins that cap pointed ends, and numerous studies have shown that Tmod capping of actin filaments inhibits actin monomer association/dissociation and stabilizes actin polymers in a tropomyosin (TM)-dependent manner.^13–15 Skeletal muscle contains three of the four mammalian Tmod isoforms: Tmod1, Tmod3 and Tmod4.^14,16–18 Like actin isoforms, Tmods are also precisely sorted within skeletal muscle fibers (Fig. 1). The α_sk-actin thin filament pointed ends are capped by a combination of Tmod1 and Tmod4,^16–18 which, when viewed by immunofluorescence microscopy, localize to stripes that define the periphery of the H-zone.^16–18 In contrast, SR-associated γ_cyto-actin filaments are predominantly capped by Tmod3 in a compartment whose immunofluorescence localization signature is a wide Z-line-flanking stripe and a narrow M-line stripe.¹⁰ A minor Z-line-flanking SR (or T-tubule) compartment also contains a combination of Tmod1 and Tmod4.¹⁰ Based on differences in the binding strengths of Tmods to muscle α- and β-TMs, we had hypothesized that sorting of Tmod1 to thin filaments depended on preferential binding of Tmod1 to the muscle TM-coated actin filaments in sarcomeres.¹⁶

Recent work from our laboratory has probed the functions of actin and Tmod isoforms in skeletal muscle by analyzing the skeletal muscle phenotype of Tmod1-null mice.^10,16 Surprisingly, α_sk-actin thin filament lengths are unchanged in the absence of Tmod1, because Tmod3 vacates its SR-associated compartment and translocates to the thin filament pointed ends,^10,16 capping the α_sk-actin thin filaments in combination with Tmod4, thereby maintaining both correct thin filament lengths and normal myofibril structure (Fig. 1).¹⁶ The altered localization of Tmod3 in Tmod1-null muscle indicates that competition between Tmod isoforms for the pointed ends of various TM-coated actin filaments can direct Tmod isoform sorting via the differential affinities of Tmods for α_sk-actin or γ_cyto-actin filament pointed ends, which, therefore, rules out a directed transport mechanism for Tmod localizations in skeletal muscle. Tmod1-null muscle is mechanically weak, which is evidenced directly by a ~20% decrease in isometric stress production of isolated muscles, and indirectly by depressed locomotor activity of animals.¹⁶ Muscle weakness in Tmod1-null mice is not attributable to an acute myopathy, seeing as Tmod1-null muscle does not exhibit pathological protein aggregation or an abnormally high frequency of centralized nuclei,¹⁶ which are hallmarks of murine models of nemaline myopathy and muscular dystrophy, respectively. Muscle weakness in Tmod1-null mice is also associated with selective fast fiber hypertrophy and muscle-wide shifts toward expression of faster myosin heavy chain isoforms.¹⁶ However, whether the observed fiber type reprogramming in Tmod1-null muscle is a cause or effect or muscle weakness remains uncertain.

What is the mechanism of muscle weakness in Tmod1-null mice? The key structural change in Tmod1-null muscle is not altered α_sk-actin thin filament lengths, but rather Tmod3 leaving its SR compartment, and, hence, a loss of γ_cyto-actin filament capping.^10,16 Thus, via a surprising and counterintuitive mechanism, deletion of Tmod1 in skeletal muscle enabled us to study the role of Tmod3-mediated capping of γ_cyto-actin in the SR.¹⁰ In wildtype muscle, Tmod3-capped γ_cyto-actin filaments link to the SR at the M-line via a direct interaction between Tmod3 and sAnk1.5,¹⁰ a 17-kDa splice variant of ankyrin-R with a hydrophobic transmembrane segment (Fig. 1).¹⁹ The Tmod3/γ_cyto-actin/sAnk1.5 complex in the SR also contains nonmuscle TM5NM1 and TM4, which have previously been shown to localize to T-tubules and the SR, respectively, with TM5NM1 being essential for correct excitation-contraction coupling and Ca²⁺ uptake in the T-tubules.²⁰ The Tmod3/γ_cyto-actin/sAnk1.5 complex does not appear to be associated with sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) or the ryanodine receptor.¹⁰ (Potential associations with other SR ion channels and pumps have not yet been examined.) Rather, Tmod3-capped γ_cyto-actin filaments appear to be a link in an SR-to-myofibril mechanical linking system, the nexus of which is the giant (~720 kDa) protein obscurin, which connects sAnk1.5 in the SR to myomesin and titin in the sarcomeric M-line, as described above.⁵

Tmod3 plays a key role in stabilizing γ_cyto-actin filaments in the SR. In Tmod1-null muscle, depletion of Tmod3 from the SR causes a partial dissociation of the Tmod3/γ_cyto-actin/sAnk1.5 complex (Fig. 1).¹⁰ While γ_cyto-actin, TM5NM1 and TM4 remain at both the M- and Z-lines, the relative proportion of each of these proteins in the M-line is attenuated by ~25%, as determined by quantitative analysis of confocal immunofluorescence images.¹⁰ Strikingly, sAnk1.5 also exhibits a novel ectopic Z-line localization in addition to its normal localization at the M-line.¹⁰ At the Z-line, sAnk1.5 may now bind to the N-terminal ZIg1 and ZIg2 domains of titin,²¹ possibly as part of a compensatory reinforcement mechanism. The mislocalization of sAnk1.5 to the Z-line that occurs when Tmod3 leaves the SR is not accompanied by changes in the localization of obscurin, indicating that the Tmod3/sAnk1.5 interaction is not required for the correct localization of obscurin but may enhance sAnk1.5 binding to obscurin.⁵ Additional biochemical studies are required to clarify the hierarchy of interactions that comprise this SR-to-myofibril linking system.

The rearrangement of the SR-associated γ_cyto-actin cytoskeleton when Tmod3 leaves the SR in the absence of Tmod1 alters both SR structure and function. Tmod3 loss from the SR in Tmod1-null muscle results in an increase in the frequency of abnormal SR swelling and a ~15% decrease in SR Ca²⁺ release.¹⁰ Tmod1-null muscle also exhibits an age-dependent increase in myofibril misalignment,¹⁰ highlighting the importance of Tmod3-capped γ_cyto-actin filaments in augmenting the maintenance of lateral linkages between adjacent myofibrils in skeletal muscle. Thus, based on these extrasarcomeric changes, the mechanism of muscle weakness in Tmod1-null mice appears to be a combination of Ca²⁺ handling and lateral force transmission defects rather than thin filament length regulatory defects.^10,16

A striking aspect of the regulation of SR morphology and physiology by Tmod3-capped γ_cyto-actin filaments is its conceptual homology to other membrane-associated Tmod/actin structures in nonmuscle cells, together with the morphological and physiological roles of these structures in such cells (Table 1). For example, in the canonical red blood cell (RBC) membrane, Tmod1 caps the pointed ends of short β_cyto-actin filaments, and Tmod1-capped β_cyto-actin filaments are, in turn, crosslinked by spectrin tetramers into a hexagonal lattice.^22,23 This lattice, with Tmod1/β_cyto-actin junctional complexes interconnected by spectrin strands, imparts the RBC membrane with the deformability and other mechanical characteristics required for uninterrupted passage of RBCs through the microvasculature. When Tmod1 is deleted, Tmod3 appears in RBCs, but at only 1/5^th of wild-type levels.²⁴ (Seeing as mature RBCs do not normally contain Tmod3, neither at the membrane nor in the cytosol, the appearance of Tmod3 in Tmod1-null RBCs must be a compensatory response that occurs during erythropoiesis.²⁴) Inadequate capping due to depressed total Tmod levels and the presence of a non-native Tmod isoform results in more heterogeneous β_cyto-actin filament lengths (albeit with normal average lengths), which, presumably, are a consequence of heightened actin dynamics.²⁴ This leads to defective connections between β_cyto-actin filaments and spectrin, generating an irregular spectrin network with larger and more variable pore sizes.²⁴ From a physiological perspective, these changes manifest as a mild, compensated spherocytic elliptocytosis with increased cellular osmotic fragility and reduced deformability.²⁴

A phenotype similar to the phenotype of Tmod1-null RBCs is observed in Tmod1-null lens fiber cells (Table 1).²⁵ Like RBCs, mature lens fiber cells are organelle-free cells under continual mechanical stimulation, with RBCs deforming during vascular transport and lens fiber cells enduring chronic stresses during lens growth, accommodation and focusing. Furthermore, both RBCs and lens fiber cells contain a spectrin/actin network that imparts robust mechanical properties to the cell membrane.²⁶ While both RBCs and lens fiber cells contain Tmod1, lens fiber cells are distinct from RBCs in that their Tmod1-capped actin filaments are comprised of both β_cyto-actin and γ_cyto-actin.^27,28 Deletion of Tmod1 causes actin filaments to partially dissociate or depolymerize from lens fiber cell membranes, relocalizing to the cytosol.²⁵ This is accompanied by a reduction in nonmuscle γ-TM (TM5NM) levels and a disruption in the spatial distribution of β2-spectrin along the fiber cell membranes.²⁵ In turn, these cytoskeletal changes cause the normally hexagonally packed lens fiber cells to exhibit abnormal membrane protrusion morphologies, irregular cell shapes and disordered packing organization.²⁵ Despite these striking cell-level morphological changes, Tmod1-null lenses do not exhibit impaired optical clarity or cataract formation.²⁵

Membrane morphological changes similar to those that occur after Tmod depletion in the SR, RBCs and lens fiber cells in vivo have also been observed in epithelial cells in vitro. In particular, a key study used shRNA to study the role of Tmod-capped actin filaments in confluent cultures of polarized epithelial cells (Table 1).²⁹ Similar to lens fiber cells, epithelial cells contain actin filaments that consist of a combination of β_cyto-actin and γ_cyto-actin.³⁰ However, unlike lens fiber cells but similar to the SR, Tmod3 is the exclusive Tmod isoform that caps the pointed ends of these membrane-associated actin filaments. After shRNA-mediated knockdown of Tmod3, both actin filaments and nonmuscle TM disappear from lateral membranes, and the localization of α2-spectrin is disrupted. Together, these changes manifest phenotypically as a reduction in cell height, but without any changes in cell polarity and the locations of tight junctions or adherens junctions, suggesting that β_cyto-actin and γ_cyto-actin depolymerization and TM dissociation downstream of Tmod3 knockdown renders polarized epithelial cells unable to fully support their own weight and maintain correct cell shape.²⁹

In a model that unifies the phenotypes of Tmod-depleted RBCs, lens fiber cells, epithelial cells and skeletal muscle SR, Tmod-capped actin filaments are tethered to their associated membranes via Tmods themselves or via additional filament-interacting proteins. When Tmod levels are reduced, actin monomers more readily associate/dissociate from actin filament pointed ends, leading to downstream cytoskeletal changes (e.g., TM dissociation from actin filaments and loss of spectrin attachment points) that impair membrane mechanics and physiology. Hence, Tmod-capped actin filaments function as essential “universal scaffolds” that enhance the mechanical properties of their associated membranes.

The precise biophysical mechanisms by which actin filaments enhance membrane properties remain unclear. We speculate that actin filament tethering to membranes modifies the local extensibility and fluidity of the membrane in a manner that optimizes membrane function. In RBCs, lens fiber cells and epithelial cells, actin filaments are crosslinked by spectrin strands, which, in turn, are tethered to the membrane by large ankyrins; thus, actin filaments in these cells have multiple attachment points and are tethered to the membrane indirectly via spectrin.³¹ In contrast, in the SR, Tmod3-capped γ_cyto-actin filaments are tethered to the membrane via a single-point attachment at the Tmod3/sAnk1.5 interface.¹⁰ It is likely that other horizontal attachments (analogous to the spectrin-mediated interconnections in RBCs, lens fiber cells and epithelial cells) may interconnect γ_cyto-actin filaments in the SR as well. Conversely, Tmods may also tether actin filaments directly to the membranes of RBCs, lens fiber cells, and/or epithelial cells via as-yet-unidentified binding partners (analogous to the scenario in the SR). Furthermore, by analogy to Tmod-capped actin filaments at the cell membranes of RBCs, lens fiber cells and epithelial cells, Tmods may also function in capping and stabilizing β_cyto-actin and γ_cyto-actin filaments along the inner face of the sarcolemma.^8,9,18 These possibilities require further exploration.