Evolution and Functional Differentiation of the Diaphragm Muscle of Mammals

Abstract

Symmorphosis is a concept of economy of biological design, whereby structural properties are matched to functional demands. According to symmorphosis, biological structures are never over designed to exceed functional demands. Based on this concept, the evolution of the diaphragm muscle (DIAm) in mammals is a tale of two structures, a membrane that separates and partitions the primitive coelomic cavity into separate abdominal and thoracic cavities and a muscle that serves as a pump to generate intra-abdominal (P_ab) and intra-thoracic (P_th) pressures. The DIAm partition evolved in reptiles from folds of the pleural and peritoneal membranes that was driven by the biological advantage of separating organs in the larger coelomic cavity into separate thoracic and abdominal cavities, especially with the evolution of aspiration breathing. The DIAm pump evolved from the advantage afforded by more effectively generation of both a negative P_th for ventilation of the lungs and a positive P_ab for venous return of blood to the heart and expulsive behaviors such as airway clearance, defecation, micturition, and child birth.

Didactic Synopsis

The DIAm separates abdominal and thoracic cavities; thus, it is a partition, and its evolution reflects that important role in isolating organs into separate thoracic and abdominal cavities. However, the DIAm is also a muscle, and is most often described as the principal pump muscle of inspiration. However, the DIAm also serves as a pump for generating both negative P_th and positive P_ab in other motor behaviors. Accordingly, the evolution of the DIAm is more complex and should be considered in the context of its dual physiological roles as a partition and muscular pump. In considering DIAm evolution, we adopt the guiding concept of symmorphosis or economy of design, where biological structures are not over designed for their functional roles. Thus, this is a tale of the evolution of two diaphragms, a partition and a muscular pump that separates thoracic and abdominal cavities but also affects generation of P_th and P_ab.

Introduction

When we think of the striking diversity of mammalian systems and observe the myriad of forms and ecological niches these species inhabit, child-like wonder stokes an instinct to imagine a plethora of unique adaptations in order to solve different challenges each species has for life on earth. Instinct could not be more wrong, as Kerr (1808–1890) entreats, “plus ça change, plus c’est la même chose” (“the more things change, the more they stay the same”). Immutable principles of comparative biology whittle away the superficial differences and we are left with the core constraints dictated by function. Symmorphosis is a concept introduced by Ewald Weibel and Charles Richard Taylor in 1981 [652] that codifies a biological design principle based on an economy of design, whereby the structure and function of integrative systems are linked, and no one component has excessive performance capacities above that which is necessary for the preservation of life. If a system has excess capacity, then it is likely to be involved in two distinct physiological processes [652, 653, 689, 690].

The DIAm is unique to mammals, and physiologists generally ascribe the main function of the DIAm as generating a negative P_th to drive airflow and fill the lungs during breathing (i.e., the principal muscle of inspiration). Accordingly, based on the symmorphosis concept, the DIAm should be primarily designed to accomplish ventilatory behaviors that have a high duty cycle (time active versus inactive) and are highly repetitive day in and day out. However, in most if not all mammals, the forces or transdiaphragmatic pressures (P_di) generated by the DIAm during ventilation represent less than half of the total force generating capacity of the DIAm. In addition, when fully activated the DIAm is susceptible to fatigue – not a good feature for a muscle primarily designed to accomplish ventilation. So this raises the question; is the DIAm over-designed just for ventilation or is it optimally designed to contribute to other physiological processes not just ventilation? In exploring the evolution of the DIAm across mammalian species, most investigators have considered only ventilatory demands. In this comprehensive review, we will consider not only the range of ventilatory behaviors across mammals but also differences in non-ventilatory behaviors of the DIAm as a partition and muscular pump to generate P_ab.

Symmorphosis: Linking Biological Structure with Functional Demands

Biological evolution is often considered in the context of structural and functional changes that afford some survival advantage. Structural biology spans from molecular, cellular, tissue, organ and whole organism levels. Similarly, physiology or function spans the full scale of structure. Symmorphosis is a theory of biology where structural design features (e.g., morphological properties) are matched to functional demands (i.e., range of physiological requirements) within an integrated system [652, 653, 689, 690]. The theory of symmorphosis was originally tested in the pulmonary system where diffusing capacity of the lungs (alveolar surface area) was compared to the maximum rate of O₂ consumption. As an integrated system, the theory also included the capacity for capillary diffusion and muscle O₂ consumption (mitochondrial density). In this regard, the DIAm, with its high duty cycle, is a major consumer of O₂ and this is reflected by the high mitochondrial density and oxidative capacity of at least some DIAm fibers [49–52, 163, 329, 606, 607, 625]. Though this concept has a proven utility in many biological systems, particularly when observing broad trends, there are a variety of limitations and exceptions to these trends both within and across species, a flaw readily acknowledged by the originators of the hypothesis [689]. For example, when linking structure and functional requirements in trained athletes and in the thoroughbred horse, lung structures remain unchanged, despite remarkable efficiency gains within the circulation (hematocrit, blood flow) and increased mitochondrial density in muscle [675], and obvious mismatch in the structure function relationship. Indeed, in horses, the structure function mismatch is so great that phenomenal gains in maximal VO₂ levels (~160 ml kg⁻¹ min⁻¹), peak cardiac output (0.8 L kg⁻¹ min⁻¹), splenic augmentation of blood volume and hematocrit (30% increases), values ~2–3 times higher than fit humans, are coupled with arterial hypoxemia, pulmonary hypertension, hypercapnia and pulmonary hemorrhage [675], a striking outlier from symmorphic principles and more resembling ‘dysmorphosis’, a term coined by Peter D. Wagner to describe serious mismatches between structure and function in some cases [674]. Sadly, the concept of dysmorphosis has never been fully developed or defined. The definition of dys – from the original Greek, meaning bad or difficult has implications for pathology or disease, and fails to capture the full gamut of the examples of ‘failed’ symmorphosis. Indeed, many examples of failed symmorphosis may be due to one system having multiple functions (e.g., the DIAm, as elaborated upon in this review), a function related to sexual selection (e.g., secondary sex characteristics such as the mane of a lion, the antlers of cervidae, peacock feathers and narwhal tusks) or merely an over-production. Perhaps the greatest example of failed symmorphosis is the venom of the taipan snake, which typically releases ~20 mg of venom per bite [696], with an intravenous LD₅₀ of ~0.01 μg g⁻¹ [646] and a subcutaneous LD₅₀ (analogous to the action of a bite) of ~0.1 μg g⁻¹ [457, 696]. Thus, with a single bite, enough venom is delivered to kill ~8000 mice, an egregious over-production!

In a more general presentation of the symmorphosis hypothesis, biological systems adhere to an “economy of design” such that no single parameter in the system has unnecessary excess capacity, beyond that which fulfills the range of physiological demands. Quantitative differences in design features and/or functional effects add to biological diversity and drive evolution. In this respect, symmorphosis postulates that the evolution of structural design and function are linked to the guiding principal of “enough but not too much” [652, 653, 689, 690]. Such an economy of design should also be present in the evolution of the DIAm, especially since this muscle is unique to and thereby distinguishes mammals. Indeed, this begs the questions: did the DIAm emerge through evolution to efficiently support only lung ventilatory requirements? Is the full capacity of the DIAm or any of its structural features overdesigned to achieve adequate ventilation? Is the DIAm also designed to efficiently generate P_ab? Hopefully, these questions will be at least partially addressed in this review.

Compared to humans, smaller mammals consume approximately six times more energy even when normalized for differences in body mass. Some outliers exist, with the Etruscan shrew having a submaximal O₂ consumption rate of 1000 ml kg min⁻¹, a rate ~30 times that of man [333], with rats ~8 times [144]. This basic observation has intrigued physiologists for more than 100 years. Part of this difference in energy consumption may relate to basal requirements for thermoregulation. It is well known that metabolic rate depends on body mass such that the surface-to-volume ratio of the body is ~2/3 [551]. However, in 1932, Kleiber reported that an animal’s O₂ consumption at rest scales to body mass with an ~3/4 ratio [353, 354]. While the absolute value of this scaling remains controversial, it is clear that in animals there is scaling between structural and functional properties that follows some power relationship to body mass (i.e., allometric scaling) and that this may represent a fundamental principle in biology.

In 1999, West and colleagues [693] introduced a fractal network model for the supply of O₂ to tissues via capillaries. Based on this fractal model, these investigators derived an allometric exponent for basal metabolic rate of 3/4 body mass. However, the 3/4 exponent describes a relationship for basal metabolic rate rather than a limitation on the O₂ supply. Taylor et al. [651] found that the maximal metabolic rate (MMR) during exercise scales with body mass with a higher ratio of 0.86 compared to the 0.75 predicted by the fractal model proposed by West and colleagues. Regardless of the absolute power relationships, the basic concept of such scaling relationships has intrigued physiologists for years [687], and there are a number of original and review articles on related subjects [124, 133, 134, 217, 297–299, 573, 574].

Various cell functions require energy (ATP largely supplied by mitochondria) including protein synthesis, active membrane ion transport, muscle work by contractile proteins (Figure 1). Clearly, during exercise the relative consumption of energy across cell functions shifts to a larger proportion by muscle work. At rest, muscle ATP consumption is minimal, while during peak power output of exercise, muscle ATP consumption markedly increases, while other the metabolic demands of other body functions may decrease. Thus, basal metabolic rate reflects the energy demands for maintaining processes such as thermoregulation in mammals. Blood flow to tissues is distributed based on need for energy substrates, ranging from basal requirements to maximum exercise. Thus, muscle energy demands range widely (an ~10–50-fold range depending on species) while muscles also account for a substantial proportion of total body mass. Since O₂ cannot be stored indefinitely for later use, the supply of O₂ from lung ventilation to arterial/capillary transport to cellular mitochondria must be able to match this range of muscle demand. Thus, the allometric scaling exponent for maximum metabolic rate varies across species depending on the relative metabolic requirements of muscle during exercise. In mitochondria, the maximum rate of ATP synthesis via oxidative phosphorylation does not vary with body mass when normalized for the surface area of the mitochondrial membrane [296]. Thus, at the mitochondrial level, the most athletic mammal on earth (the Rocky Mountain pronghorn antelope) consumes O₂ at similar rates to the mouse at maximum metabolic rate [402], though measurements of maximal consumption are confounded by different exercise types, muscles assessed and durations [535].

Figure 1:

Cells require energy substrates to function, even at basal levels of low metabolic work (A). The major energy substrate for ATP production is oxygen (O₂), supplies by gaseous exchange within the lung. Other major energy substrates include carbohydrates and fats, delivered via the arterial circulation from sources within the gastrointestinal tract and liver. At the level of the capillaries, cells uptake these energy substrates and excrete waste products of ATP generation, including carbon dioxide (CO₂), which is returned to the atmosphere during gaseous exchange. Metabolic work increases the requirement for ATP, and during muscle activity, gaseous exchange must be increased in order to cope with the increased demand for O₂ and removal of CO₂. B shows the allometric relationship of basal metabolic rate against body weight. Adapted from data within refs 105, 159, 229, 258, 451, 463, 568, 643 and 693.

Scaling of Lung Volumes Across Mammalian Species

The concept of symmorphosis was first established to compare the pathway for O₂ from the lung to the mitochondria in muscle [652], with critical assessments performed in showing the cellular aspects (capilliaries, blood and mitochondria) to be highly matched to demand, whereas lung structures had increased excess O₂ diffusion capacity, indicative of its limited plasticity in relation to exposure to external environments [689, 690]. Various experimental structural and functional characteristics of the lung have also been assessed in this regard [217, 433, 649, 651, 652]. In the mammalian lung, capillary blood is exposed to an extensive surface area for gas exchange. This large surface area for gas exchange is achieved through consecutive subdivision of airways into smaller and smaller airways ending in terminal units for gas exchange called alveoli. For comparison, a single sphere of 1 cm³ volume has a surface area of 4.8 cm², whereas a 1 cm³ volume of the lung has an alveolar surface area of 2,100 cm² due to the extensive subdivision into smaller alveoli [218]. Thus, the branching strategy of the mammalian lung can increase the surface area for gas exchange by ~430 fold. With ~250 million alveoli in the human lung, the gas exchange surface area is ~100 times greater than that of the body surface area, [686, 688]. Assuming an ~5 L lung volume in humans, with an average alveolar diameter of 250 μm, the lung alveoli have a gas exchange surface area of ~150 m² that is interfacing with ~213 cm³ of blood contained in lung capillaries [216].

Although by internal subdivision, the mammalian lung is very efficient in accomplishing gas exchange, the extent of subdivision of the lung is limited with respect to this as a strategy for increasing the respiratory surface area before structural integrity is compromised [412]. Importantly, the volume within the lung comprising large airways and vasculature remains steady across species [216, 225, 411, 523, 658], with parenchyma contributing ~84% of the total anatomical volume of the lung [658]. Subdivision of the lung results in a large number of small gas exchange units in a limited volume. These smaller gas exchange units with higher surface tension are susceptible to collapse at the air-tissue interface and require more work (energy) to inflate. Based on the Young-Laplace Law, the pressure required to inflate an alveolus is directly dependent on its surface tension and inversely related to its radius. Amongst mammals, the lungs of bats have the smallest diameter alveoli (~30 μm), whereas dugongs and manatees have the largest diameter alveoli (~1300 μm) [656]. The bat is astonishing in the sense that airborne locomotion is energetically taxing, requiring high levels of gas exchange in the lungs, a feat that is achieved through very small alveoli that are packed into large lungs (for its body weight) and a high hematocrit and high blood O₂ affinity [334, 412, 413]. In general, the diameter of alveoli is negatively correlated with metabolic O₂ consumption per unit of body weight, VO₂ [656]. The respiratory surface density, the surface area available for gas exchange divided by the lung parenchymal volume plotted against body weight shows a steady conservation in the amount of subdivision in the lung across mammalian species. Consequently, resting metabolic rate also correlates to total body weight in mammals [105, 159, 229, 258, 451, 463, 568, 643, 695] (Figure 1).

The initial proposal of symmorphosis dealt with gas exchange not only in the lung, but also at the tissue level where O₂ is utilized for energy production within mitochondria. In this respect, the gas exchange capacity within the lungs of mammalian species has an ~2 fold excess [689], with only the smallest mammals using their entire capacity during hypoxic conditions [648–650]. An important caveat for these assessments is that many of these measures are predicated on assumptions and estimations derived from anaesthetized preparations. Thus, the apparent ‘excess’ capacity of the lung gas exchange may actually reflect the short-term requirements for highly athletic behaviors in the wild. Overall, the apparent mismatch between structural and functional requirements may be interpreted to mean two things: 1) that in many cases considerable reserve capacity exists to tailor gas exchange requirements to increased energetic demands, be they developmental, aging, illness- or ecology-related and interspecies-related; or 2) that lung design is not fully determined by requirements for gas exchange, and that other factors such as biomechanical advantage and endocrine activities also drive lung design [412, 414, 415, 689, 690]. Indeed, dual roles are exactly the reason that the DIAm, at first glance, violates the principles of symmorphosis.

Symmorphosis: Evolution of the Diaphragm

The DIAm functions both as a partition, separating the thoracic and abdominal cavities, and also as a muscular pump generating negative P_th and positive P_ab. It is this dual functional role that should be accounted for in considering the evolution of the structural design features of the DIAm, and how these are matched to functional demands (i.e., the range of physiological roles in which the DIAm participates). Symmorphosis postulates that there should be no unnecessary, excess capacity in design features, so the evolution of the DIAm should be explored by considering how its structural design meets physiological demands. In this respect, we will first consider the partitioning role of the DIAm in forming separate abdominal and thoracic cavities. This partition prevents movement of abdominal organs into the thoracic space, especially during aspiration breathing. In addition, by reducing compartment volume, this partitioning improves the efficacy of P_th and P_ab generation.

We will also consider the role of the DIAm as a muscular pump for generating P_th and P_ab. In this context, the involvement of the DIAm in different motor behaviors requiring pressure generation must be considered. Certainly, it is well recognized that the DIAm is the major inspiratory muscle for lung ventilation. Since these respiratory efforts have a high duty cycle (i.e., time active versus inactive), the DIAm must be designed to avoid fatigue during inspiratory efforts. However, the DIAm is also involved in higher force, low duty cycle efforts associated with increased abdominal pressure generation. Thus, we suggest that the DIAm should be considered as two pumps, one used persistently for generating negative P_th for lung ventilation and the second used more infrequently for generating higher P_ab to promote venous return and for expulsive behaviors. Throughout this review, we refer to measurements obtained in a variety of species. We have endeavored to capture the entire breadth of available knowledge on the DIAm and many of the observations cited herein are not necessarily exemplars, but constitute the only available data.

Diaphragm as a Partition Between Abdominal and Thoracic Cavities

Formation of an internal body cavity is a major distinguishing feature in zoology and evolution, with vertebrates exhibiting a range of configurations (Figure 2). There was an important evolutionary sequence from acoelomate animals without internal body cavities to pseudocoelomate and finally to coelomate animals that have a true internal fluid filled body cavity lined with epithelial cells forming a double-layered (parietal and visceral) peritoneum, pleura or pericardium depending on location [446]. In fish, these cavities consist of a pericardial (containing the heart) and peritoneal (containing the visceral and urogenital organs) cavity [351, 366]. In amphibians and most reptiles, the body contains the pericardial cavity and the pleuroperitoneal cavity (comprising the lungs, visceral and urogenital organs) [351]. In some reptiles and all mammals, the pleural cavity (containing the lungs) and pericardial cavity are consolidated into the thoracic cavity [351]. Furthermore, in some mammalian species, there is a separation of the left and right lungs by the presence of a partition (the mediastinum) separating left and right pleural cavities [679]. The peritoneal cavity, also termed the abdominal cavity in mammals is completely partitioned from the rostral pleural cavities by the DIAm.

Figure 2:

The septum transversum is found in all vertebrates, and serves as a partition to separate the heart and pericardium from the peritoneum of the abdominal cavity. It forms in relation to the pericardium and folds caudally in association with the liver. In fish, the septum transversum divides the pericardial and peritoneal cavities. In amphibians and reptiles, it divides the pericardial cavity and the common pleuroperitoneal cavity, which contains the lungs, urogenital and visceral abdominal organs. In the mammals, the septum transversum, now muscularized in the form of the diaphragm muscle, extends the entire span of the body cavity, forming a separation between the collection of the pericardial and pleural (containing the lung) cavities and the peritoneal cavity (i.e. a thoracic cavity and an abdominal cavity). By partitioning separate abdominal and thoracic cavities, smaller relative abdominal and thoracic spaces are created, which improves the efficiency of P_ab and P_th generation. Note that in fish, amphibians and most reptiles, this partition is incomplete and less efficient. In advanced reptiles and all mammals, a complete separation of the thoracic and abdominal cavities is achieved, with this separation being muscularized in the case of mammals, the diaphragm muscle. After Kingsley [351].

Formation of a fluid filled coelomic cavity provided spatial and physiological independence of organs. The posterior portion of the coelomic cavity is lined by the peritoneum comprising a squamous epithelial cell layer on top of a mesenchymal layer with an intervening serous fluid-filled space that hosts blood vessels, lymphatics and nerves both on the outside wall (parietal peritoneum) and on the intestine (visceral peritoneum). In the anterior portion of the coelomic cavity, the pleura and the pericardium reflect an equivalent double layered epithelial lining that envelops not only the thoracic wall but also the lungs and heart, respectively.

The serous fluid-filled space between peritoneal layers allows movement of the viscera, especially the gastrointestinal tract, while providing support for the organs within the posterior coelomic cavity. Some organs such as the kidneys and portions of the gastrointestinal tract are positioned behind the peritoneum between the peritoneum and posterior coelomic cavity wall, and thus, they are classified as being retroperitoneal. Parts of the parietal and visceral peritoneum come together to form large folds that bind the internal organs together and to the walls of the posterior coelomic cavity. Examples include the mesenteric, omental and falciform ligaments. These peritoneal folds also form compartments within the coelomic cavity. The embryological development of the pericardial, pleural and peritoneal cavities clearly shows the relationship between the compartments of the body cavity and their partitions [496] (Figure 3). Thus, compartmentalization is a major evolutionary trait matching structure and function.

Figure 3:

Embryonic timeline of body cavity formation in humans. By week 5, pleuroperitoneal membranes are a pair of membranes which gradually separate the pleural and peritoneal cavities, produced as the pleural cavities expand by invading the body wall. The pleuropericardial membranes separate the developing heart from the developing lung buds. Pleuropericardial membranes initially appear as small folds or ridges projecting into the primitive undivided thoracic cavity. The folds contain the common cardinal veins which drain the primitive venous system into the sinus venosus of the primitive heart. During week 6, the edges of these membranous folds have fused with the dorsal mesentery of the esophagus and with the septum transversum to separate the pleural and pericardial cavities. As the heart descends and the pleural cavities expand, the membranes are drawn out in a mesentery like fold that extends from the lateral wall. By week 7, these membranes fuse with the mesoderm ventral to the esophagus, forming a single pericardial cavity and left and right pleural cavities. During week 8, the lung buds grow into the medial walls of the nascent pleural cavities; and the pleural cavities expand around the heart into the body wall. After Pansky [496].

The structure of the pleural and pericardial lining comprises two layers that form a serous fluid-filled space, which facilitates movement of the lungs and heart while providing structural support and partitioning [446]. Similar to the peritoneum, the merging of parietal and visceral pleura and pericardial membrane forms folds that delineate partitions within and across cavities. Importantly for the DIAm, the pleuroperitoneal fold contains the phrenic nerves emerging from the cervical spinal cord [107, 444, 503] (Figure 3). Myocytes that will fuse and form the DIAm also migrate along the pleuroperitoneal fold [107, 444, 503].

The parietal lining of the coelomic cavity forms a tight seal for pressure development by muscular contraction to enhance critical body functions. These functions included voiding of metabolic waste products (e.g., defecation and micturition), egg laying (procreation), pressure gradients for blood and lymph circulation and venous return of blood to the heart. There were distinct evolutionary advantages to develop mechanisms for increasing pressure generation within the coelomic cavity. For example, with the evolution of calcified (hard shelled) eggs in reptiles, extrusion required increased coelomic cavity pressure development [629]. Similarly, as animals grew in size, the requirements for voiding increased, again requiring increased coelomic cavity pressure generation. With increased energy demands and upright posture in some reptiles, coelomic cavity pressure was essential for venous return of blood to the heart [327]. The symmorphosis concept applied to DIAm evolution cannot exclude the evolutionary advantage gained by the impact of DIAm contraction on more efficient generation of P_ab. Thus, there is an evolutionary advantage afforded by the more efficient generation of positive P_ab and muscularization of the DIAm partition provided an evolutionary advantage in this respect.

The DIAm evolved from a membranous physical partition that separated a single coelomic cavity into separate abdominal and thoracic cavities [503]. As mentioned above, formation of peritoneal folds was a common feature for partitioning compartments within the coelomic cavity (Figure 2). In lower vertebrates, the hindgut of the digestive tract is contained within a single coelomic (abdominal) cavity but suspended by peritoneal folds. In air breathing vertebrates (e.g., amphibians), comingling of the lungs and digestive tract was fine when breathing was accomplished by buccal or positive pressure breathing (Figure 4). However, in reptiles, aspiration breathing evolved to facilitate the external circulation of air into the lungs for gas exchange [69, 503]. The strategy of aspiration breathing in reptiles was critical for more efficient external circulation of air into the lungs. However, the negative intra-coelomic pressure generated by rib cage muscles caused rostral movement of the digestive organs (stomach, liver, gut) into the thorax, which impeded expansion of the lungs. This led to the evolution of a physical partition of the abdominal and thoracic cavities in order to prevent movement of visceral organs so that they would not impinge on lung inflation during aspiration breathing (Figure 4).

Figure 4:

The evolution of air breathing occurred before the evolution of aspiration breathing. The simplified clade diagram shows that ray-finned fishes (actinoptarygii) employed buccal pump ventilation of the gills. Air breathing lungfish (dipnoi) and amphibians also employed this mechanism of ventilation, which involved buccal expansion to draw air into the oral cavity under negative pressure, followed by buccal compression, with a closed mouth and nares to inflate lungs under positive pressure. Turtles (testudines), scaled reptiles (lepidosauria), crocodiles (archosauria – also includes birds) and mammals all breathe using an aspiration mechanism, whereby the lungs are inflated with a negative intra-thoracic pressure. Importantly, aspiration breathing requires ribs being attached to the vertrebrae. The ribs serve to stabilize the coelomic cavity for walking and act as a bellows for aspiration. The evolution of aspiration breathing occurred some point between emergence of amphibians and reptiles, though in some reptiles, buccal breathing mechanisms are employed intermittently.

The septum transversum is related to the formation of the pericardial sac, cardiac tube and liver [351] (Figure 3). It serves as a partition to separate the heart and pericardium from the peritoneal cavity, containing the visceral and urogenital organs in lower vertebrates, including fish [343] (Figure 2). In animals with lungs, such as amphibians and reptiles, the septum transversum partitions the pericardial cavity and the pleuropertioneal cavity, containing the lungs, visceral and urogenital organs [351]. In fish, the septum transversum partition facilitates generation of a positive intra-coelomic pressure gradient when muscles of the coelomic cavity wall contract. This positive pressure is important in enhancing venous return of blood to the heart [343]. A positive intra-coelomic cavity pressure is also important for other body functions such as defecation, urination, egg laying and, in some cases, vomiting/regurgitation. In amphibians, contraction of coelomic wall muscles contributes to positive intra-coelomic cavity pressure generation [343]. Thus, muscular contraction in amphibians facilitates venous return of blood to the heart as well as other bodily functions that require a pressure gradient.

The first physical partition of the abdominal and thoracic cavities appeared in reptiles by the fusion of the septum transversum with the peritoneal folds of the post-hepatic septum and post-pulmonary septum [503] (Figure 2). In reptiles, the development of positive intra-coelomic cavity pressure is still important to enhance venous return of blood and to drive other bodily functions. Importantly, in most reptiles the lungs are comingling with visceral and urogenital organs within the pleuroperitoneal cavity [351]. By partitioning separate abdominal and thoracic cavities, smaller relative abdominal and thoracic spaces are created, which improves the efficiency of P_ab and P_th generation (Figure 2). However, in most extant reptiles, there is no separation of abdominal and thoracic cavities, but there is extensive development of the ribcage [343]. In these animals, intercostal muscle activation expands the coelomic cavity thereby developing a negative intra-coelomic pressure for lung inflation (aspiration breathing). This certainly is counterproductive for bodily functions that require a positive intra-coelomic pressure and thus, there was a need for conflict resolution through evolution.

In reptiles such as lizards with higher energetic demands due to their active predation lifestyle (e.g., tegu lizards), a separation of the lungs from liver and the rest of the abdominal compartment by the post-hepatic septum evolved, which prevents the stomach, liver and gut from impinging on lung expansion during inspiration [355–357, 359]. In more advanced reptiles such as varanoid lizards, chameleons, crocodilians and many turtles, a post-pulmonary septum, separating the lung from both the pericardium and the liver evolved, which partially stabilizes the lungs against the liver, providing a viscera-free compartment [155, 502]. Thus, in some reptiles and all mammals, the lungs are located in a thoracic cavity separated by a partition (post-pulmonary septa for reptiles and in the mammalian case, the DIAm) from the abdominal cavity (i.e., the thoracic cavity contains the peritoneal and pleural cavities and the abdominal cavity contains the peritoneal cavity) [351]. In some turtle species, the post-pulmonary septum contains skeletal muscle [226].

It is important to note that neither the bilateral skeletal muscle diaphragmaticus in the crocodile, nor the striatum pulmonalis in the turtle have a principal inspiratory function [503]. In crocodiles, piston action of the diaphragmaticus muscle against the liver increases the thoracic volume and decouples ventilation from locomotion restraints on chest wall expansions [169, 171, 211, 212, 473], and provides a functional advantage during increased metabolic demands of exercise, low temperature or hypercapnea [170, 263, 264, 469]. In turtles, when the striatum pulmonalis muscles contract, they increase P_ab for active expiration [213, 214]. In addition, these muscles contribute to buoyancy compensation by displacing the lungs and viscera when the animals are submerged, to help maintain desired pitch and yaw orientations [664]. Importantly, these muscles are not phylogenetically or ontogenetically related to the DIAm of mammals [503]. Furthermore, in crocodiles that have a post-pulmonary septum partitioning the thoracic and abdominal cavities, internal circulation of blood is enhanced, which allows some internal physiologic control of body temperature and thermo-stability [579, 584, 634]. It is tempting to speculate that the larger dinosaurs may have had similar muscularized partitions or piston-like apparatus to facilitate P_th and P_ab generation. Indeed detailed descriptions of this possibility have previously been discussed in the literature [547–549]. Alternatively, diaphragmaticus muscle like piston functions arguably require a kinetic pubis bone, found only in crocodilofoms [169], whereas dinosaurs have an immovable pubis [103, 270]. Furthermore, the undoubted relationship of sauropods to birds and the avian-like nature of their breathing [111], suggests that the evolution of the air sac, and the unidirectional crosscurrent gaseous exchange system [683, 684], evident in some alligators and lizards [167, 168, 172, 571], meant O₂ delivery was not a limiting factor in ever-more gigantic sauropods [501, 559–561], and helps overcome the dead-space constraints due to excessively long necks.

Of course, the diaphragm muscle is only one way to match respiratory function requirements to anatomical structure. Instead of inflating the thoracic cavity like mammals, the avian respiratory system instead decouples the pump (air sacs and diverticula of the postcranial skeleton) from the site of gaseous exchange (the parabronchi) [78]. Thus, the avian respiratory system allows for oxygenated air to flow constantly in a single direction throughout the entire breath cycle. The gaseous exchange at the parabronchi is also highly efficient, with the direction of pulmonary blood flow at right angles to the direction of airflow. These adaptations allow for the adequate oxygenation of blood at high altitudes during flight, including over the Himalayan mountain range [266, 576, 577] and compensate for the increased respiratory dead space created by the elaborate and elongated necks and beaks of some species [277]. Indeed, the defining components of an avian-like respiration system, namely the unidirectional airflow, air sacs, the pneumacity of bony structures and countercurrent gaseous exchange, are not interdependent, and may have arisen separately during evolution [559, 572], as evidenced by observations of unidirectional airflow in Nile crocodiles, despite the lack of postcranial pneumacity or air sacs [572]. The aforementioned unidirectional airflow in crocodiles and some other reptiles [168, 172, 571, 572] satisfies the first component, and modern phylogenetic classification schemes would thus indicate their likelihood in other sauropods [559, 654, 685]. The pneumacity of the postcranial skeleton would serve two purposes in dinosaurs: the first would serve to reduce the mass of the skeleton and return the center of mass towards the center of gravity, thus facilitating balance [41, 89, 685]; the second to allow for the growth of neck length, allowing for the exploitation of new ecological niches [559]. Regardless of whether or not dinosaur vertebrae were pneumatised or not, the extremely long neck does pose a respiratory dead space problem, particularly for tidal breathing. Regardless, in many sauropods, evolution towards continuous O₂ uptake, air sacs and pneumaticised postcranial skeletons would undoubtedly confer selective advantages for not only gigantism, but also for flight in pterosaurs, incidentally the largest flying vertebrates to grace the earth [104, 700]. Obviously, these adaptations are some of the defining features of avian species.

In summary, effective partitioning of the thoracic cavity from the abdominal cavity promotes P_ab generation and prevents impingement of abdominal viscera on the lungs during the inspiratory phase of aspiration breathing. Positive P_ab facilitates venous return of blood to the heart as well as several other important physiological functions. The development of a physical partition between abdominal and thoracic cavities enhances the development of P_ab. In reptiles, negative intra-coelomic cavity pressure generation during aspiration breathing creates another problem of incursion of gut viscera and impingement of lung expansion. A physical partition of thoracic and abdominal cavities solves this problem. Muscle action on this partition in crocodiles and turtles evolved to enhance positive P_ab generation and facilitate non-respiratory function, not to promote inspiratory efforts. By contrast, evolution of intrinsic DIAm in mammals serves both as a partition and an inspiratory pump, playing an important role in facilitating both P_ab and P_th generation. This increased efficiency of both P_ab and P_th generation and the functions they serve presents a major evolutionary advantage for mammals.

Diaphragm Muscle as a Pressure Pump

As noted above, the DIAm separates the abdominal and thoracic cavities of the body. As the DIAm contracts it moves caudally, creating a negative P_th and inspiratory airflow. This downward motion of the DIAm also produces a positive P_ab. The resulting transdiaphragmatic pressure (P_di = P_ab – P_th) reflects DIAm force generation (Figure 5).

Figure 5:

Anatomical schematic showing placement of the solid-state pressure catheters for measurement of esophageal intra-thoracic (P_th) and gastric intra-abdominal (P_ab) pressures in rodents. As the DIAm contracts it moves caudally, creating a negative P_th and inspiratory airflow and a positive P_ab. The resulting transdiaphragmatic pressure (P_di = P_ab – P_th) reflects DIAm force generation.

We assert that in the context of symmorphosis, evolution of the structure/function relationships in the DIAm involves a tale of two diaphragms, one serving as a partition separating thoracic and abdominal cavities and the second serving as a muscular pump for pressure generation. Even when considering the pump function of the DIAm, there are two evolutionary paths: 1) the DIAm as an inspiratory pump for the generation of persistent, non-fatiguing negative P_th to sustain the high duty cycle requirements of ventilating the lungs day in and day out; and 2) the DIAm as a pump for generating positive P_ab that are more infrequent (lower duty cycle) but typically require higher forces (i.e., requiring near maximal activation of the DIAm). Below, we consider the structural and functional design of two types of muscle fibers (or motor units): one a lower force but fatigue resistant set of muscle fibers (type I and IIa comprising slow-twitch - type S and fast-twitch fatigue resistant - type FR motor units) that are efficiently designed for generating negative P_th necessary to sustain breathing behaviors, and a second set of higher force but fatigable muscle fibers (type IIx and/or IIb comprising type fast-twitch fatigue intermediate – type FInt and fast-twitch fatigable – type FF motor units) that are optimally designed for short duration high force motor behaviors involving maximal P_ab generation.

In complex animals, the generation of positive P_ab and negative P_th impacts two circulations that are necessary to sustain life. The first “external” circulation is designed to intake and distribute raw materials from outside the body to within the body, where they can be processed and utilized by the organism. Generation of a negative pressure in the thoracic cavity facilitates this intake and thus, there is a distinct biological advantage. A primary example of this type of circulation is the respiratory system in aspiration breathing, where a negative P_th facilitates airflow from the environment into the lungs for gas exchange. Another example is the alimentary tract and swallowing, which shuttles food, water and nutrients to stomach and gut epithelia for absorption and digestion. The external circulation is also important in facilitating the voiding of waste products from the body. This is true in the respiratory tract, where positive P_ab helps move expired air from the lungs thereby eliminating carbon dioxide as the byproduct of metabolism. Generation of positive P_ab is also important in defecation and micturition and the elimination of waste products.

The second “internal” circulation distributes energy substrates (nutrients and O₂) to tissues throughout the body and eliminates metabolic waste products from the tissues delivering them to points of excretion. This internal circulation is provided by the cardiovascular and lymphatic systems. In the abdominal and thoracic cavities, these blood and lymphatic vessels are located in the peritoneal and pleural spaces and thus subject to pressure gradients (difference between P_ab and P_th). In the cardiovascular system, the heart acts as a pump to generate pressures in order to distribute oxygenated blood to the tissues. Venous blood returning to the heart, is lower pressure and thus proportionally more affected by gravity, which is mitigated by skeletal muscle contractions [20, 406] and a pressure gradient generated by a pressure differential between the thoracic and abdominal cavities. In mammals, the DIAm elegantly provides the partition necessary to simultaneously generate a negative P_th and a positive P_ab. Thus, both ventilation of the lungs and the ablution of abdominal contents is achieved by a single muscle, with multiple demands for P_th and P_ab generation.

The earliest strategy for distributing fresh air into the lungs was buccal breathing (Figure 4) [64, 66, 69, 399, 414, 415]. In this scenario, the oral cavity expands and compresses, with air forced into the lung under positive pressure. In ray-finned fishes, buccal breathing serves to perfuse the gill with water. In lungfish and the extant amphibians, it serves to inflate the lung [64, 66, 69, 399, 414, 415]. At some point in the early natural history of the amniotes (reptiles, mammals and birds), aspiration breathing evolved, a strategy whereby the lungs are inflated by the generation of a sub-atmospheric negative pressure to cause influx of atmospheric air [64, 66, 69, 399, 414, 415, 503]. The efficiency of this mechanism is enhanced by partitioning the thoracic and abdominal cavities (seen in some reptiles). Muscularization of this partition in the DIAm of mammals further increased pressure generation capacity and efficiency. There is no living intermediate form between buccal pump and aspiration breathing, and the abrupt shift from the cranial and hyobranchial musculature providing for ventilation remains puzzling [64, 66, 69]. However, this view has been challenged, with axial musculature involved in active exhalation maneuvers of salamanders [65, 67, 68, 70]. This suggests that aspiration breathing did not evolve suddenly, but in two phases, the first with axial musculature powering active exhalation (while buccal movements remain required for inhalation), followed by axial muscles being used for inhalation and exhalation. Further support of this notion is the buccal and gular breathing of some lizards and geckoes to assist in lung ventilation during and after bouts of locomotion [494]. Certainly, the interactive mechanisms for the neuromotor control required to coordinate the upper respiratory muscles were present in reptiles and amphibians before the emergence of muscularized DIAm, as evidenced by the conservation of rhythmic respiratory pattern generators in the medulla and brainstem of various animal kingdom phylogenetic lineages [531, 642, 665, 666, 697]. Furthermore, in the medulla and brainstem is where one finds the respiratory-related pre-motor neurons that project bilaterally to the phrenic motor pool [186], in many ways making the neuromotor control and ventilator function of the DIAm an extension of the rhythmic upper airway opening and closures related to buccal ventilation, while lung ventilation in amphibians is analogous to active expiration in aspiration breathers (Figure 4).

In many respects, the trunk and abdominal stabilization maneuvers are essential adaptations for both efficient mammalian respiration and locomotion, and there is an abundance of information for locomotor-respiratory coupling [500]. This coupling is facilitated by the ribs, which allow for improved energetics in locomotion and reduce the breathlessness apparent in the pausing of buccal breathers (between jumps) and in reptiles where the side-to-side locomotion impinges on the ability to inflate the lungs [110, 358]. In quadrupeds, there is synchrony integrating one breath per stride (1:1 ratio) [71], and in hopping mammals, such as wallabies a 1:1 pattern is also observed [32]. In humans during walking, locomotor-respiratory coupling is also in 1:1 [524]. However, while running, there is more variability, with synchronization following various patterns (4:1, 3:1, 2:1, 1:1, 5:2 and 3:2) [71]. Trained individuals exhibit markedly more stable coupling between respiratory patterns and locomotor patterns [437]. During locomotion, while parasternal activity remains associated with inspiration, activity of the interosseous intercostals becomes uncoupled from ventilation and instead the activity becomes associated with leg movement [92]. Thus, during locomotor tasks, the intercostal efficiency for ventilatory maneuvers is reduced, underlining the importance of locomotor–respiratory coupling of the DIAm during exercise.

Exercise increases muscle tissue demand for oxygen and thus increases the requirement for ventilation. Indeed, exercise is likely the greatest stressor on ventilation. Respiratory minute volumes for male adults during eupnea (quiet breathing) are ~6–10 L min⁻¹, assuming a tidal volume of ~500–1000 ml [1, 590] and a respiratory rate of 12–20 breaths min⁻¹ [1, 590]. The P_di generated by the DIAm during these events is ~4–8 cmH₂0 [3, 311, 603], which represents only ~3–5% of maximum P_di generation (~150 cmH₂O) [508, 657]. During exercise, minute ventilation increases to ~100–170 L min⁻¹, respiratory rates increasing to ~40–70 breaths min⁻¹ and tidal volumes increasing to ~3 L [46, 108, 250, 328, 435]. Using a modified version of Ohm’s Law, and assuming the resistance to airflow within the respiratory system is equal between eupnea and exercise, the requirement for P_di generation during maximum exercise is ~50 cmH₂0, representing only ~33% of maximum DIAm force production, and likely to be less given the increased contribution of the intercostals and other accessary inspiratory muscles [139, 699]. Indeed, ~30–40% of maximum P_di is in good agreement with the P_di generated during sustain airway occlusion in rodents [347], which likely represents the ceiling of DIAm pressure generation for ventilatory efforts. In humans such P_di may be achieved with inspiratory efforts associated with obstructive sleep apnea before arousal.

If the DIAm is only an “inspiratory muscle” as many believe, this raises the question of overdesign and violation of the symmorphosis hypothesis. It is clear that although exercise does utilize some of the ‘reserve capacity’ of the DIAm, these increased ventilatory requirements for P_di generation do not approach the maximum pressure generating capacity of the DIAm. However, expulsive/straining maneuvers require near maximum P_di generation and, we will endeavor to highlight the underappreciated non-ventilatory functions of the DIAm.

As the superior wall of the abdominal cavity, the DIAm is essential in the generation of positive P_ab in a variety of motor behaviors (Figure 6). The highest P_ab are generated during straining maneuvers that involve very high levels of DIAm activation with co-contraction of abdominal and external intercostal muscles and simultaneous closing of the glottis by coordinated activation of the lateral cricoarytenoid and transverse arytenoid muscles of the larynx [317]. Such straining maneuvers occur with vomiting [4, 5], defecation [203, 204] and parturition [271].

Figure 6:

The muscles of the thoracic wall provide for radial expansion of the chest wall (the external intercostal muscles), cranial expansion of the ribcage (parasternals, sternocleidomastoid and scalene) and caudal expansion of the thoracic cavity (the diaphragm muscle). In concert these provide for the generation of negative intra-thoracic pressures. The muscles of the abdominal walls include the lateral (internal intercostal, internal oblique, external oblique and transverse abdominal muscles) and ventral walls (the rectus abdominis and transverse thoracis), that serve to increase intra-abdominal pressured by compressing the abdomen. The cranial wall of the abdominal cavity is provided by the diaphragm muscle, which when activated reduces the cranial extent of the abdominal cavity, thus increasing intra-abdominal pressures.

Depending on the thoracic or abdominal nature of the expulsive behavior, and the orifice of ejection, an increase in P_ab may induce reflex activation of the anal or urethral sphincters. This reflex activation typically depends on the rate of rise of P_ab [317]. During coughing, a rapid increase in P_ab occurs, which leads to augmentation of anal sphincter activity, thus preventing incontinence [585–588]. By contrast, slower rate of increase in P_ab, such as rates observed during defecation, results in a decrease or the abolishment of anal sphincter activity [586]. Many of the non-ventilatory behaviors involving DIAm activation correspond to the generation of very high forces and thus extensive recruitment of phrenic motor neurons and DIAm motor units. Reflex straining maneuvers may be elicited by distension of the rectal wall, vaginal wall, bladder wall, or by stimulation of pelvic afferent nerves [203, 204]. In pregnant rats, stimulation of vaginal afferent nerves also elicits substantial DIAm activation and the generation of higher P_ab, leading to expulsion of the fetus [271]. Interestingly, when the abdominal muscles are paralyzed in quadriplegic patients, activation of the DIAm is still preserved during high-force efforts of coughing. This indicates that neither abdominal muscle contractions or reflexes are required to initiate cough-related DIAm contractions [165].

Expulsive straining behaviors require the generation of very high P_ab but these pressures do not need to be maintained for long-periods of time; thus, fatigue is not typically a factor. To generate higher P_ab participation of all abdominal wall muscles is required including the DIAm, which forms the superior wall of the abdomen. Synchronous co-contraction of antagonistic muscle groups prevents their relative shortening [7, 269]. This ensures that the lengths of these muscles are maintained within an optimal portion of their force-length relationship throughout activation. Preservation of optimal length explains why P_di pressures generated voluntarily in humans are highest during expulsive Valsalva maneuvers, where generation of an P_ab ranges from ~90 to 220 cmH₂O [109, 253, 591]. During the Valsalva maneuver, co-contraction of the DIAm and abdominal muscles prevents both the DIAm and abdominal muscles from shortening (i.e., maintaining maximum isometric condition) thereby maximizing DIAm force generation. In contrast, P_ab are lower in isolated voluntary maneuvers, where co-contraction of abdominal muscles with the DIAm does not occur [384]. Synchronous contraction of the DIAm and abdominal muscles during expulsive maneuvers also prevents cranial displacement of the abdominal organs toward the thoracic cavity.

The important role of DIAm activation in expulsive maneuvers is supported by a variety of clinical reports. In patients with severe DIAm weakness, difficulties with defecation are often concomitant [503]. In horses, the severing of the phrenic nerve leads to the accumulation of feces in the rectum [332]. In quadriplegic patients, DIAm pacing is associated with less difficulty during defecation [503]. In addition, diaphragmatic ruptures may occur during other high-force expulsive efforts such as coughing [100, 123, 227, 335, 540], vomiting [391], or during parturition [61, 255, 257, 275, 542, 550, 580, 713]. Indeed, the generation of higher P_ab during straining behaviors is a common factor in non-traumatic diaphragmatic hernia [157]. Among all expulsive straining behaviors, parturition inarguably requires the highest, most prolonged and repeated activations of the DIAm. During this effort, the coordinated and sustained contractions of the abdominal muscles and the DIAm generate incredibly high intra-uterine pressures, greater than ~130–220 mmHg (~170–300 cmH₂O) [84, 506, 521]. In humans, where the dimensions of the birth canal and the head of the fetus are very similar, hence there is a necessity for generation of high P_ab in addition to uterine contractions. In addition to humans, herniation of the DIAm during parturition has also been observed in other mammalian species including horses, buffaloes and goats [120, 630, 645]. Perhaps even more significant, DIAm fatigue may occur during the repetitive maximal contractions of normal human labor [476]. From an evolutionary standpoint, the development of a muscularized DIAm may have been a key adaptation in mammals to allow the bearing and birth of living offspring. Indeed, the high-force generating capacity of the DIAm may be a key facilitator favoring the evolution of increased cranial size in humans and thus the increased volume of the “disproportionate” human brain.

The DIAm is also active during non-respiratory activities that involve adjustments in trunk posture, including those that occur during rapid upper limb movements [279–281] and other motor activities such as weightlifting [9]. Anticipatory crural and costal DIAm contractions, initiated in response to visual stimulation, occur before rapid flexions of the shoulder [279]. Swift and repetitive motions of the upper limbs induce expiratory activity of DIAm relaxation, upon which a layer of phasic modulation is added at frequencies corresponding to both limb and respiratory movements [281]. Abdominal muscles (e.g., the transversus abdominis muscle) are subject to similar modulatory activities [279, 281], and likewise for pelvic floor muscles [283]. Importantly, the activation of pelvic floor muscles is linked to DIAm activity through polysynaptic reflex pathways that facilitate voiding behaviors. These reflexes are triggered by high-force expiratory (e.g., cough) or inspiratory (e.g., sniff) pressures that increase P_ab [102].

Intra-abdominal pressure increases during postural tasks [268], a result of the co-activation of the abdominal muscles and the DIAm, together with various pelvic muscles. A marked elevation of P_ab occurs during weight lifting behaviors and reflects the contribution of the DIAm to core strength and the tight coordination among the abdominal wall muscles [268]. Activation of the DIAm is the most important determinant of the maximum P_ab achieved, which is enhanced with a closed glottis. A rise in P_ab also increases spinal stiffness, reduces lumbar intervertebral motions [282, 595], and stabilizes the trunk, thereby optimizing posture and the efficiency of movement. As mentioned, rupture of the DIAm following high P_ab, closed glottis maneuvers may occur [434]. The mechanisms involved in such injuries range from DIAm avulsion from the muscle origin or insertion, the shearing of overstretched fibers or blunt/sudden force transmission through the abdominal viscera acting in a projectile manner.

Symmorphosis: Skeletal Muscle Design and Function

The structural properties of muscle are designed to efficiently accomplish force generation and contraction (change in length) while avoiding fatigue of these motor properties. The striated appearance of skeletal (and cardiac) muscle fibers is due to the presence of repeated sarcomeres in series arranged along the length of myofibrils [252, 398, 441] (Figure 7). The DIAm is a striated (skeletal) muscle with sarcomeric organization of contractile proteins. There is some controversy regarding the number of muscle fibers within the DIAm although very few studies have systematically examined this. Krnjevic and Miledi estimated that the rat DIAm comprises ~10,000 individual muscle fibers [373]. Since there are ~480 phrenic motor neurons in the rat spinal cord [190], this would represent a mean innervation ratio (muscle fibers per motor neuron) of ~21 fiber per motor neuron. However, the innervation ratio is likely closer to 100 fibers per motor unit [196], which would correspond to a total of ~48,000 muscle fibers in the rat diaphragm.

Figure 7:

Single multinucleated (green, propidium iodide myonuclear stain) diaphragm muscle fibers (A) exhibit striated sarcomeric structures (membrane stained with RH414, red). These striations comprise of the thin (actin) and thick filaments (myosin) seen in transmission electron micrographs (B). In response to Ca²⁺, overlapping filaments undego cross-bridge formation, driving the production of force. The mechanical cycling of cross bridges causes a force vector from the Z-disc towards the midline of the sarcomere (C). The electron micrograph of a skeletal muscle cross-section (D) shows the classical myofilament lattice spacing. The larger structures are the myosin filaments, each surrounded by six smaller actin filaments (E). Each actin filament is surrounded by three myosin filaments, and two actin filaments are shared by any given myosin filament. Thus arrangement allows for the doubled hexagonal crystalline array of the myofilament lattice, the distance between the center of the myosin filament and the center of an adjacent actin filament is 12 nm. The distance between the centers of adjacent myosin filaments is 40 nm. Adapted from elements within ref 608 and 615.

Similar to all skeletal muscle fibers, DIAm fibers are multinucleated, with nuclei located peripherally [237, 598, 615, 620]. Thus, each myonucleus controls gene expression within a restricted volume of the muscle fiber, termed the myonuclear domain [18, 430, 668]. Some investigators have suggested that myonuclear domain size is regulated via apoptotic elimination of myonuclei during atrophy and addition of myonuclei (satellite cell fusion) during hypertrophy. However, in the rat DIAm, we found that myonuclear domain size was not controlled during conditions of atrophy and hypertrophy and changed proportionately with muscle fiber cross-sectional area [233].

Although the basic sarcomeric structure of muscle fibers is similar, the sub-sarcomeric structural design of muscle fibers varies to enhance certain features in the form of different fiber types. In addition, the volume density of organelles such as the sarcoplasmic reticulum and mitochondria can vary across fibers enhancing performance features that are well aligned to functional demands.

Sarcomeric Structure of Striated Muscle

In eukaryotes, actin-based myosin motors evolved to facilitate a variety of movement-related functions including the trafficking of intracellular organelles, cellular division, and muscle contraction. Among the myosin motor protein superfamily, class-II myosins are distinguished by their ability to assemble into thick filaments [116]. Within the rod domain of the myosin heavy chain (MyHC) there is an interface for dimerization into a coiled-coil configuration [405], which contributes to a higher order assembly first into filaments [439, 440] and further into sarcomeres [122] found in skeletal and cardiac muscles (Figure 7).

Different MyHC isoforms are found that determine both the contractile and energetic properties of muscle fibers. In humans, skeletal muscle MyHC isoforms are encoded by six genes (chromosome 17p13) [596, 691, 692] and two genes (chromosome 14q12) encode cardiac muscle MyHC isoforms [410, 556]. In addition, in mammals three other class-II MyHC genes exist: 1) the smooth muscle MyHC gene that encodes, via alternative RNA splicing processes, four distinct proteins [25]; 2) non-muscle A MyHC; and 3) non-muscle B MyHC. These class-II MyHC proteins in non-muscle cells are responsible for a variety of actin-dependent motor functions [25, 344]. There is structural similarity between myofilaments comprising smooth muscle MyHC (thought to be the most primitive) and non-muscle MyHC [10], suggesting an evolutionary relationship among class-II MyHCs. It is likely that the evolution of smooth muscle MyHC and non-muscle MyHCs emerged from a common ancestral gene before the evolution of sarcomeric MyHC isoforms.

The sarcomere, which is the essential functional unit of myofibrils, consists of contractile proteins interspersed between Z-discs. The Z-discs are typically aligned orthogonally to the long axis of each myofibril and also aligned in parallel across myofibrils to form Z-disc (Figure 7). Thin filaments are anchored to the Z-discs and project toward a midline (M-line). Thick filaments are interposed between the thin filaments in a highly-organized, crystalline fashion with six actin filaments surrounding each myosin filament.

Thin filaments comprise polymerized actin molecules together with tropomyosin and troponin. Thick filaments comprise MyHC and myosin light chain (MyLC) molecules (together called the myosin head) with a longer myosin tail. Cross-bridges are formed by the chemical binding of myosin heads to the actin filament, and cross-bridge formation provides the molecular basis for force generation and contraction (i.e., shortening) [507] (Figure 7).

When muscle fibers contract, the ratcheting action of cross-bridge attachment and detachment (cross-bridge cycling) requires energy (ATP hydrolysis) and establishes the velocity of shortening depending on the external load (load-velocity relationship of striated muscle). The interactions between thick and thin filaments cause the Z-disc to move toward the midline of sarcomeres without a change in thick or thin filament length; first described as the sliding filament theory [313] (Figure 7).

Within a muscle fiber, the sarcoplasmic reticulum is a loose network between myofibrils that connects to T-tubules located near each Z-disc. The sarcoplasmic reticulum network serves as an intracellular storage depot for Ca²⁺, whose release can be triggered to initiate contraction through a process termed excitation-contraction coupling. Deep invaginations of the plasma membrane called T-tubules are juxtaposed to the sarcoplasmic reticulum at each Z-disc. These T-tubules transmit depolarization of the plasma membrane deep into the interior of the muscle fiber allowing the organization of myofibrils in parallel and larger diameter muscle fibers. With T-tubule depolarization, Ca²⁺ is released from internal stores (the sarcoplasmic reticulum). The resulting elevated of intracellular Ca²⁺ facilitates binding to troponin C (TnC) on the thin filament, thereby removing the steric hindrance of the binding site of the myosin head on the actin filament. This regulation of myosin attachment to actin and cross-bridge formation also involves troponin T (TnT), which binds the troponin complex to the tropomyosin molecule and troponin I (TnI) that actually blocks the actin binding site (Figure 8). Through this Ca²⁺ regulatory process, the attachment of myosin heads to actin and force generation is regulated, as reflected by a sigmoidal force-Ca²⁺ relationship (Figure 8).

Figure 8:

Excitation-contraction coupling is mediated by the sarcoplasmic reticulum, which acts as a store for Ca²⁺ (A). Within the T-tubules, depolarization waves activate dihydropyridine receptors (DHPRs; voltage sensitive L-type Ca2+ channels), which in turn induces an initial ryanodine receptor (RyR) mediated Ca²⁺ release. A positive feedback process induces further Ca²⁺-induced Ca²⁺ release. This process rapidly floods the cytosolic space surrounding contractile proteins with free Ca²⁺, eventually binding to troponin C, removing the steric hindrance and allowing for cross-bridge formation (B). This regulation of myosin attachment to actin also involves troponin T (TnT), which binds the troponin complex to the tropomyosin molecule and troponin I (TnI) that actually blocks the actin binding site. Release of Ca²⁺ to the cytosol is followed by the development of muscle fiber force (C). The Ca²⁺ binding of troponin complexes regulate the attachment of myosin heads to actin and thus regulate force generation, as reflected by a sigmoidal force-Ca²⁺ relationship. Adapted from elements within refs 224, 608 and 615.

Evolution of Excitation-Contraction Coupling

In striated muscle (both cardiac and skeletal), a process of excitation-contraction coupling evolved to control muscle contraction. In this process, the electrical charge stored on the membrane capacitor is discharged during an action potential. Muscle fiber action potentials are propagated along the length of the sarcolemma (muscle fiber membrane) and depolarization is passively transmitted down the T tubule where dihydropyridine receptors (DHPRs; voltage sensitive L-type Ca²⁺ channels) are located. In vertebrate skeletal muscle fibers, DHPRs are linked at tetrad structures to a subset of ryanodine receptors (RyRs), which are Ca²⁺ release channels located in the sarcoplasmic reticulum membrane. T-tubule depolarization activates DHPRs, which in turn induces an initial RyR-mediated Ca²⁺ release. The open probability of the RyRs is also sensitive to Ca²⁺ such that initial DHPR-mediated Ca²⁺ induces further Ca²⁺ release in a positive feedback process (i.e., Ca²⁺-induced Ca²⁺ release). This process rapidly floods the cytosolic space surrounding contractile proteins with free Ca²⁺. The elevated cytosolic Ca²⁺ binds to TnC leading to removal of steric hindrance and cross-bridge formation (Figure 8). In cardiac muscle and in body muscles of invertebrates, RyR and DHPRs are also present, but tetrads are not present and their interaction is indirect. Muscle fiber depolarization triggers opening of DHPRs and Ca²⁺ influx. This Ca²⁺ influx triggers Ca²⁺-induced Ca²⁺ release through RyRs in the sarcoplasmic reticulum.

Functional Properties of Striated Muscle

The two major inter-related functions of muscle are to generate force and cause shortening or contraction. Contraction of muscle typically occurs in opposing an external load. When the load equals or exceeds the force generation capacity of a muscle, no contraction occurs (isometric condition). With a lesser external load, the ability of the muscle to shorten depends on the force generated (force/load - velocity relationship; Figure 9). Thus, the generation of force is central to the overall function of muscle.

Figure 9:

In the DIAm, maximal power output is achieved at ~30% of maximal shortening velocity and ~30% of maximal force generation/load (A). The force a muscle fiber generates is related to its length (B). At sub-optimal lengths, not all actin and myosin elements are able to form cross-links, thus force is limited. At optimal length (L_o), maximal actin and myosin cross bridge formation is possible, thus force generation is maximal. Beyond this length, passive tension (stretch) causes reduction in the possible number of cross-bridges able to be formed and active force generation is reduced. Adapted from elements within refs 615, 619 and 620.

The force (F) generated by a muscle fiber is estimated by the following equation:

$$F=n•f•{\alpha }_{fs}$$

Where n is the MyHC concentration per half sarcomere, f is the force contributed by each cross-bridge and α_fs is the fraction of MyHC forming strongly bound cross-bridges. At optimal sarcomere length (i.e., greatest extent of overlap of myosin heads and actin binding sites), the primary determinant of α_fs is Ca²⁺-dependent regulation of the myosin head binding site on the thin filament (this forms the basis of the force-Ca²⁺ relationship in muscle fibers). At sub-optimal length, myosin heads cannot bind to actin; thus, α_fs is also reduced and less force is generated (Figure 9). This forms the basis of the force-length relationship of striated muscle (both skeletal and cardiac muscle). During muscle contraction sarcomere length changes thus affecting the overlap of thin and thick filaments, α_fs, the number of cross-bridge formed and the force generated. Thus, as sarcomere shortening velocity gets faster, force decreases and this dependency is reflected by the force-velocity relationship of muscle (Figure 9).

The force generated by an individual cross-bridge (f) is mainly determined by MyHC isoform composition. Myosin is a hexameric protein comprising two MyHC molecules (each weighing ~200 kDa), and four MyLC molecules (weighing ~17–20 kDa). The MyHC molecules are intertwined to form a double helix. A myosin head is at the end of each MyHC that acts as an enzyme to hydrolyze ATP during cross-bridge cycling. There are also two MyLCs at each myosin head; one “regulatory” and one “structural”. The functional role of the MyLC isoforms varies, but it is thought that they help stabilize the myosin head during contraction.

Several isoforms of MyHC exist, some present only during embryonic (MyHC_Emb) and neonatal development (MyHC_Neo). In the adult rat DIAm, MyHC isoforms MyHC_slow, MyHC_2A, MyHC_2X, MyHC_2B correspond to different fiber types [195, 610, 613, 625], a phenomenon conserved across all species studied, though the MyHC_2B isoform is not present adult human DIAm. Each DIAm fiber typically expresses only a single MyHC isoform, although some co-expression of MyHC_2X with MyHC_2B occurs in varying proportions in a variety of mammalian species. With muscle injury, it has been reported that expression of MyHC_Emb and MyHC_Neo reappears, possibly due to the fusion of satellite cells, into the injured fiber for repair. Differences in the relative proportions and expression of MyHC isoforms determine the contractile and fatigue properties of the DIAm across species.

Sir Andrew Huxley introduced the sliding filament theory of muscle contraction in 1957 [313]. Based on this general model, it is now widely accepted that cross-bridge cycling determines the mechanical properties of muscle fibers. When myosin heads strongly bind to actin, there is a “power stroke” that involves bending at the junction of the “head” and “neck” regions of the myosin molecule. As a result, the strong binding of the myosin head transitions to a weaker bond, and cross-bridges detach. During this power stroke force is generated and the sarcomere shortens depending on the external load. This process then repeats during cross-bridge cycling, and with each transition there is hydrolysis of one ATP molecule and chemical energy (ATP) is converted into mechanical energy (force and/or shortening). When ATP is removed, myosin and actin do not dissociate and consequently the muscle fiber stiffens (i.e., rigor mortis). It appears that the power stroke varies across MyHC isoforms, being greater in fibers comprising MyHC_2A, MyHC_2X, MyHC_2B isoforms compared to those comprising the MyHC_slow isoform. Thus, both the greater force per cross-bridge (f) and faster shortening velocity of “fast” muscle fibers relates to their MyHC isoform composition. The consumption of ATP for any MyHC isoform is described by the following equation:

$$ATP\text{Consumption}=n•b•{\alpha }_{fs}•g_{app}$$

Where n is MyHC concentration per half sarcomere, b is the number of half sarcomeres in series, α_fs is the fraction of MyHC that is strongly bound forming cross-bridges, and g_app is the apparent rate constant for cross bridge detachment. Velocity of shortening and g_app are dependent on external loading; thus, ATP consumption changes with external load and velocity of shortening reaching a maximum at peak power output of a muscle fiber [259, 261, 616]. This relationship was first quantitatively described by W. O. Fenn, who measured heat production in contracting muscle, and is known as the Fenn effect (Figure 10). The maximum velocity of the ATPase reaction can be determined using quantitative histochemistry [50, 52] and varies with DIAm fiber type due to MyHC concentration and differences in g_app [259–261]. During peak power output of DIAm fibers, the rate of ATP consumption is close to the maximum velocity of the ATPase reaction (Figure 10).

Figure 10:

Peak ATP consumption rates (2.7 nmol mm⁻³ s⁻¹) occur at peak power output for diaphragm muscle (top graph). Within different diaphragm muscle fiber types, ATP consumption varies according to MyHC concentration (increasing with increased MyHC) and the apparent rate of cross-bridge detachment (lower graph). Adapted from ref 543.

Cross-bridge cycling rates (shortening velocities) and maximum ATP consumption rates of muscle fibers are associated with the expression of different MyHC isoforms. Muscle fibers comprising MyHC_2B and MyHC_2X isoforms display faster rates of force development (faster f_app; Figure 11) and faster cross-bridge cycling rates and maximum shortening velocities (associated with faster g_app; Figure 11) and have the highest ATP hydrolysis (consumption) rates followed by fibers comprising MyHC_2A and MyHC_slow isoforms [259, 543, 619].

Figure 11:

Cross-bridges cycle between a strongly bound and an unbound state during force generation and contraction. Cross-bridge cycling determines rates of cross-bridge attachment (f_app) and detachment (g_app). The rate of force development (f_app) and cross-bridge cycling (g_app) in MyHC_2A-expressing fibers is greater than that of MyHC_SLOW- expressing fibers. Adapted from ref 608.

Consistent with the Fenn effect, ATP consumption within a muscle fiber increases with power output or work performed [177]. In order to fulfill the considerable range of ATP consumption demands across muscle fibers, a range of ATP production reserve capacity by mitochondria and glycolytic pathways is necessary. Due to the lower mitochondrial volume density in muscle fibers expressing MyHC_2X and MyHC_2B (type IIx and/or IIb) and thus the lower capacity for oxidative phosphorylation, the reserve capacity for ATP production oxidative phosphorylation in these fibers is lower compared to fibers expressing MyHC_2A and MyHC_slow [543, 621]. These type IIx and/or IIb fibers depend more on glycolytic pathways for ATP production, which has much lower functional reserve capacity. It is likely that the greater fatigue susceptibility of type IIx and/or IIb fibers is due to their higher rate of ATP consumption, fewer mitochondria and lower total reserve capacity for ATP production [195, 614].

Throughout life, skeletal muscle is constantly remodeling, adjusting to changes in activity, load, or innervation. Thus, within limits, muscle can change its structure and function to adapt to environmental conditions, natural or imposed. We see this commonly in sports, where athletes train to achieve muscle adaptations that optimize specific performance needs. This can be in the form of increasing the number and velocity of muscular contractions to promote mitochondrial biogenesis (increased oxidative capacity via increasing mitochondrial density and surface area) and endurance (e.g., marathon runners) or increasing the external loading on muscle contraction to increase force and power (e.g., body builders). Marathon runners and body builders obviously do not have the same physiques. Generally, muscles in marathon runners are smaller, weaker but with greater endurance. In contrast, the training regimen of body builders results in muscle hypertrophy and an increase in force by adding sarcomeres in parallel.

The importance of skeletal muscle remodeling extends far beyond exercise physiology to many clinical conditions and diseases. The strength and endurance of muscle changes throughout life, initially as mature muscle forms and alters through age-related sarcopenia. However, chronic diseases often induce cachexia or muscle wasting and weakening. There are also neuromuscular or muscular diseases such as muscular dystrophy and other congenital muscular disorders, amyotrophic lateral sclerosis (ALS), and spinal cord injury that impair muscle activation or muscle performance. Just as exercise varies for marathon runners versus body builders, there is not a single therapeutic approach for musculoskeletal diseases. In general, the structure and function of different motor unit and muscle fiber types in respiratory muscles are the same those of other skeletal muscles and the training effects are the same.

Evolution of Muscle and Myogenesis

The development of mechanisms for movement within cells, tissues and the whole organism is a key feature of evolution. Some motor proteins such as MyHC found in striated muscle were already present in unicellular organisms before the evolution of multicellular animals. Indeed, orthologs of MyHC were expressed in sponges, suggesting functional diversification of paralogs of myosin motor proteins before the origin of true striated muscles. Similarly, in jellyfish and other animals of the Cnidarian phylum, there were orthologs of Z-disc proteins, but they were not associated with striated muscles. Thus, it appears that vertebrate skeletal muscle evolved through the separate independent evolution of constituent proteins that may or may not have been part of a pre-existing, ancestral contractile apparatus. The combined components of striated muscle, including MyHC, filamentous actin, the troponin complex and titin appear in bilaterian Metazoan animals, which have three germ layers endoderm, mesoderm, and ectoderm. Muscle cells derive from the mesoderm layer in these animals.

All skeletal muscle fibers, both axial and limb, including those of the DIAm, undergo myogenesis consisting of two stages: 1) the determination phase, a temporal switch when myoblasts emerge from mesodermal progenitor cells; and 2) the terminal differentiation phase, when myoblasts fuse to form the nascent myotubes followed by the formation of myofibers that express contractile proteins in ordered sarcomeres (Figure 12).

Figure 12:

During the determination phase of myogenesis, stem cells divide and are committed to myoblasts, a process requiring expression of the regulatory factors MyoD and Myf-5. Subsequently, myoblast that are non-proliferative fuse to form primary myotubes, which eventually become muscle fibers that express MyHC_SLOW. Primary myotubes that have further fusion of non-proliferative myoblasts become secondary myotubes. These myotubes develop into muscle fibers capable of expressing MyHC_SLOW or any of the fast MyHC isoforms.

Potential mechanisms for myogenesis include the procession of an intrinsic genetic program, removal of inhibitory regulatory signals or the presence or promotion of positive extrinsic signals [19]. However, current knowledge indicates that myogenesis is programmed by the timed expression of muscle regulatory factors (MRFs), including MyoD [129], myogenin [73], Myf-5 [74] and MRF4 [534]. Furthermore, MRFs also appear to be involved in determining MyHC isoform expression in adult muscle fibers [306]. Thus, they are a parsimonious regulatory factor in both pre- and postnatal development.

The MRFs initiate myogenesis through their function as molecular switches, triggering the expression of muscle specific genes. This occurs via the binding of MRFs to the genomic control elements (E-boxes) of muscle specific genes [484]. These CANNTG sequence motifs are located in promoter regions of a variety of genes specific to skeletal muscle [82]. Removing the inhibitory signals that dampen MRF expression results in myoblast determination (commitment). In certain classes of helix-loop-helix (HLH) transcriptional regulators, termed Id factors (Id1-Id4), the formation of heterodimers with E2 products (HLH-containing proteins ubiquitously expressed) inhibits the activity of MRF. This prevents the activation of genes specific to skeletal muscle [38]. Calcineurin reduces Id inhibitory protein expression, and thus indirectly activates MyoD via removal of inhibition [199]. Twist proteins also prevent MRF-E protein heterodimer formation, and thus they inhibit MRF activity [638]. In a similar manner, another HLH-containing protein, Mist1 binds with MyoD to form an inactive heterodimer, and thus it inhibits myogenesis [394]. Nuclear factor 1X (Nfix) binds to MEF2A, and through this mechanism, it plays a role in the transition of embryonic to fetal MyHC isoform expression [445, 639]. Overall, the balance between pro-myogenic and anti-myogenic inhibitory signals determines the location and the developmental timing of skeletal muscle precursor cells.

The first MRF to be expressed in the DIAm and other hypaxial skeletal muscles is MyoD, which initiates a cascade involving subsequent expression of Myf-5, MRF4 and myogenin within skeletal muscle precursor cells [555]. The specific timing of specific MRF expression is important for the normal development of skeletal muscle. In proliferating myoblasts, MyoD and Myf-5 are highly expressed, underscoring their importance at this stage of myogenesis [129]. By contrast, lack of MyoD, MRF4 and Myf-5 has no known deleterious effect on the development of normal skeletal muscle [75, 552, 709]. However, in mutant mice lacking both Myf-5 and MyoD, the absence of myoblasts suggests that the expression of these particular MRFs is necessary for myoblast commitment [553]. By contrast, myoblast formation occurs in the absence of myogenin but myotubes and myofibers fail to develop [265, 471]. This suggests that myogenin is obligatory for the terminal differentiation of myoblasts into myotubes and subsequent formation of myofibers.

The irreversible exit of mononucleated proliferating myoblasts from the cell cycle occurs with terminal differentiation, when transformed myoblasts fuse into multinucleated myotubes and subsequently into myofibers with expression of contractile proteins. In adults some myoblasts persist as satellite cells with the ability to proliferate, and thus play a role in regenerative processes involved in injury repair [417]. These satellite cells remain susceptible to apoptosis until they are terminally differentiated [417]. In DIAm and other hypaxial muscles, terminal differentiation of myoblasts occurs only after migration to their final destination [417]. In the DIAm, myoblast migration begins at ~E10 and continues through ~E12 and during this time, the migrating myoblasts express Myf-5, MyoD and c-MET, a tyrosine protein kinase encoded by the MET gene [83]. As mentioned, the timed expression of Myf-5, MyoD and c-MET is important in the process of myoblast differentiation and fusion into myotubes.

Myoblast fusion to form multinucleated myotubes is essential for the subsequent formation of myofibers with expression of contractile proteins organized into sarcomeres. Phrenic nerve outgrowth and arrival at the DIAm is coincident with the period of myoblast migration and formation of myotubes (Figure 13) [425].

Figure 13:

The prenatal determination of phrenic nerve emergence, axon terminal vesicular release and neuromuscular junction specialization is independent of muscle cells until embryonic (E) day 13, where it contacts the primordial DIAm (A). Muscle development and synaptic formations then enter into a state of co-dependence with nerve (B), with the establishment of adult muscle fiber types, innervation ratios and the lack of polyneuronal innervation (synapse elimination) occurring postnatally (B). Adapted from ref 425.

However, terminal differentiation of myoblasts and formation of myotubes and myofibers is not dependent on innervation. Instead, myotube and myofibers formation is influenced by mechanical properties and proteins expressed in the cell membrane (e.g., β1-integrin), the cytoskeleton (e.g., actin and desmin), the basal lamina (e.g., muscle cell adhesion molecule - M-CAM), and the extracellular matrix (e.g., fibronectins, laminins, M-cadherin and neural cell adhesion molecule - N-CAM) [361, 371]. Thus, terminal differentiation of myoblasts and subsequent formation of myotubes and myofibers is a very complex process that involves interactions of the cells with their surrounding environment.

During embryonic and neonatal development, there is a transit0069on in the expression of MyHC isoforms with considerable co-expression of isoforms (Figure 12) [376, 377, 680, 681]. During the embryonic period, muscle fibers in the DIAm express an embryonic MyHC isoform (MyHC_Emb) as well as MyHC_Slow and MyHC_2A. This period corresponds with very low heterogeneity in the size of motor neurons innervating muscle fibers [191]. During the perinatal period, expression of the MyHC_Emb switches to a neonatal isoform (MyHC_Neo), with continued expression of MyHC_Slow and MyHC_2A. In the first two postnatal weeks in rodents, expression of the MyHC_Neo isoform in the DIAm gradually disappears and there is an emergence in the expression of MyHC_2X and MyHC_2B isoforms. Singular expression of MyHC isoforms in DIAm fibers, and thus the final proportions of fiber types does not occur until about 28 days of age [219]. Beginning at postnatal day 14 in the rodent DIAm, there is disproportionate growth of fibers expressing MyHC_2X and/or MyHC_2B isoforms (~2–3 fold greater increase in cross sectional area compared to fibers expressing MyHC_Slow and MyHC_2A isoforms) [331, 512]. This period also corresponds to an increase in specific force and velocity of shortening of the DIAm (Figure 14). It remains unclear whether specific innervation of DIAm fibers plays a role in the differential growth of type IIx and/or IIb fibers. However, it is of interest that this period corresponds with a period of growth of phrenic motor neurons [513, 517].

Figure 14:

Pre- and postnatal developmental changes in maximum DIAm specific force (normalized to muscle cross-sectional area) show a ~20-fold increase from embryonic to adult (A). The maximum velocity of DIAm shortening increases ~5-fold in the same timespan (B). Adapted from refs 222 and 219.

Diaphragm Muscle Structure

Diaphragm muscle structure serves two roles: as a partition separating the thoracic and abdominal cavities and as a muscular pump affecting both P_th and P_ab. Although the gross anatomy of the DIAm is generally similar across mammalian species, there are differences that reflect adaptation to specific environmental needs.

Gross Anatomy of the Diaphragm Muscle

Based on muscle fiber origins, the DIAm is typically separated into three major regions (Figure 15): 1) the sternal region in which muscle fibers originate from the posterior portion of xyphoid process and xyphisternal junction and insert into the central tendon; 2) the costal region in which fibers originate from the broad expanse of the lower rib cage (ribs 7–12 and their costal cartilage) and insert into the central tendon; and 3) the crural region in which fibers originate from the broad expanse of intervertebral fibrocartilages of the upper lumbar vertebrae, and the arcuate ligaments overlying the aorta, vena cava and psoas major and quadratus lumborum muscles, and insert into the central tendon.

Figure 15:

The abdominal view of the DIAm (A) clearly shows the vena cava and aortic vessels passing through the central tendon and aortic hiatus, respectively. The esophagus passes through the left and right crus of the DIAm, forming the esophageal sphincter. The crus inset on the lumbar vertebrae caudal to the thoracic vertebrae where the ribs articulate, helping to create the domed structure of the DIAm. After Downey [152]. There is a marked difference in phrenic nerve branching and neuromuscular junction (NMJ) innervation of the diaphragm in small and large species, such as rats and humans (B). In rats, the NMJ innervation is very linear. In humans and other larger species, there is less linearity to the NMJ innervations and more numerous and elaborate secondary phrenic nerve branching.

The DIAm comprises right and left sides (reflecting mammals as bilaterian Metazoan animals), which are generally symmetrical with some differences in the crural region based on structures that pass behind or through the DIAm. Obviously, there is some right/left asymmetry in DIAm structure due to the asymmetry of associated structures including the heart, lungs, esophagus, abdominal organs (e.g., liver) and major vessels.

The orientation of muscle fibers in the two sternal regions of the DIAm are parallel, such that unilateral contraction of one side leads to passive shortening of the opposite side [707]. By contrast, the radiating orientation of fibers in the left and right costal regions of the DIAm are generally in series, such that unilateral contraction of fibers on one side lead to passive lengthening of the contralateral side (Figure 15). In addition, the orientation of fibers in the costal regions of the DIAm are complicated by curvature to form a dome shape. Contraction of muscle fibers in the costal regions causes this curvature to flatten downward, thereby pushing on the abdominal cavity and increasing P_ab. Diaphragm contractile mechanics combines muscle tension and curvature with generation of pressures directed orthogonal to the axis of muscle fibers. In contrast, skeletal muscles of the limb exert forces parallel to the muscle fiber axis, with force production and joint movement in line with the direction of muscle fibers [302, 349, 390, 698].

The orientation of muscle fibers in the right and left crural regions of the DIAm is far more complex and displays the greatest asymmetry. The right side is larger and longer compared to the left side. The medial margins of the right and left crural regions encompass the esophagus and act as a lower esophageal sphincter during inspiratory contractions, decreasing the risk of gastric reflux. The descending aorta passes dorsal behind the right crural DIAm, whereas the inferior vena cava passes through the central tendon such that blood flow in both structures is unimpaired by DIAm contractions (Figure 15). Indeed, the negative transdiaphragmatic pressure generated during inspiration promotes an increase in venous return to the right atrium.

Blood flow is heterogeneous in the DIAm at rest, with the sternal portion of the DIAm receiving less blood flow than mid-costal and crural regions [511]. In dogs, increased blood flow is observed in the medial costal region of the DIAm, where the greatest shortening occurs [72]. This is consistent with the relationship between DIAm fiber power output and ATP consumption rate (Fenn effect; Fig x). During exercise, blood flow increases five-fold and heterogeneity persists, indicating that costal and crural regions of the DIAm may display differential loading, different shortening velocities and different contributions to inspiratory pressure generation [583]. However, there is no evidence of regional differences in fiber type proportions [527], or oxidative capacities [623].

Comparative Anatomy of the Diaphragm Muscle

In general, the gross morphology of the DIAm is conserved across species; however certain differences exist in relation to the relative area and shape of the central tendon, whether or not individual muscle fibers span the entire distance between the costal wall and central tendon, and the orientation of the DIAm in relation to the long axis of the body. These gross variations are likely due to various ecological niches inhabited by different mammals, indeed the successful adaptation of the mammalian clade to some of the most diverse environments on planet Earth, i.e., from pole to pole, from the oceans depths to the tallest peaks and from the driest desert to the wettest jungles is unique in vertebrates.

The central tendon is prevalent in all mammals, except for marine mammals [308, 316, 506], though the shape of the central tendon is variable across species, with pigs having a remarkably large proportion of tendon and ferrets having a relatively small proportion [503]. For example, most mammals have a “V”-shaped central tendon, though “I-” and “U-” shaped configurations do occur. On the right side, a foramen in the central tendon allows passage of the vena cava. Importantly, the relative proportion of central tendon to individual DIAm fiber length has little bearing on pressure generation, though it can have an effect on shortening.

In smaller mammals such as rats, mice and hamsters, DIAm fibers extend from their origin at the costal border to insertion at the central tendon, reaching lengths of ~20 mm [162, 234, 235, 245, 528], which may be a physiological limit of muscle fiber length. In larger species such as cats, dogs, monkeys and humans, DIAm fibers do not extend the full span between the costal margin and the central tendon but instead, have intramuscular tendinous origins and insertions [59, 232, 350, 466, 622]. These differences are evident in the pattern of neuromuscular junction distribution in the DIAm In rodents, neuromuscular junctions are distributed along a clear central line in the middle of the muscle (i.e., at the mid-point of each muscle fiber). In contrast, in larger mammals including humans, neuromuscular junctions are widely distributed as the mid-point of muscle fibers varies with intramuscular tendinous origins and insertions. The presence of intramuscular tendinous origins and insertions is also evident by the pattern of secondary branching of the phrenic nerve. In rodents this secondary branching pattern is very linear corresponding with the location of neuromuscular junctions. However, in larger mammals there is greater variation and elaboration of secondary branches, corresponding with the wider distribution of neuromuscular junctions [14, 59, 85, 152, 350, 386, 387, 443, 466, 578] (Figure 15).

Typically, the DIAm is oriented obliquely in the body, with the origin of fibers in costal and crural regions located more caudally than the origin of fibers in the sternal region. The central tendon insertion of fibers is located more rostral than origin of fibers in the costal and crural regions, thus fibers generally have a rostral projection. In many cases, the thickness of the diaphragm muscle in excised tissue does not match that of the in situ scenario, suggestive of potential differences in regional tensions and thus pressure generation, particularly in dogs [676]. Along the same line, horses exhibit increased length and thickness within the medial costal portion of the DIAm [510]. In horses and elephants, the orientation of the DIAm is more parallel [79, 389], with an almost complete parallel orientation (in relation to the long axis of the body) achieved in the whale in whales [503, 632, 633]. The structure and orientation of the DIAm in the manatee is highly atypical, with distinct hemi-diaphragms that are separated and positioned in the dorsal part of the body cavity, behind the heart, liver and gut, extending to the caudal end of the body cavity with no sternal attachments [541]. In the manatee, the septum transversum still partitions the heart from the liver, but is not muscularized, and actually these two structures are oriented orthogonally to each other. With the lungs distributed dorsally along the entire length of the body cavity, it is speculated that DIAm contractions can help maintain buoyancy [541].

In general, the largest pressures generated by the DIAm involve expulsive behaviors generating high P_ab following maximal DIAm contractions. By contrast, the elephant is likely to use near-maximal DIAm activation to fill its trunk with water, requiring a pressure of ~200–300 cm H₂O [79, 694]. The DIAm in the elephant is oriented obliquely between the thorax and abdomen, in contrast to the perpendicular orientation in most mammals, although it is unknown if this orientation serves to enhance pressure generation required for trunk water snorkel function or to alleviate gravitational stresses within the lung alveoli [79, 392, 393].

In diving and marine mammals, the design concerns are not related to gravitational stress, but to the considerable requirement for buoyancy control. In addition, some marine mammalian species dive ~2–3 km below the ocean’s surface [109, 495], and experience severe lung compression [48, 398]. As mentioned previously, pressure differentials between the intra-thoracic and intra-abdominal compartments affect both the cardiovascular and respiratory systems [400, 633]. In general, DIAm adaptations in diving marine mammals include a severe reduction in the size of the central tendon, increasing relative muscularization and helping to stabilize the DIAm [389, 400, 633] or alternatively, provide extra shortening potential for muscle fiber contractility [400]. In whales, the DIAm is oriented almost parallel to the long axis of the body [503, 632, 633], which in humpbacks is likely to account for some of their incredible stability in the water column without torso, flipper or fluke movement [210]. During the underwater activities of cetaceans (dolphins), both locomotor function [118, 495] and ambient oceanic pressure differentials load the DIAm [400]. By contrast, pinnipeds (seals), which do not load the DIAm as much during locomotion (generating thrust using hind flippers) [505], however they do experience marked oceanic pressures on dives. Accordingly, the DIAm of cetaceans has substantial subserosal collagen fibers, the amount of which correlated to locomotion velocities of different species [400]. This collagen layer stabilizes the DIAm during locomotion in cetaceans and prevents the intra-thoracic compartment from experiencing sinusoidal pressure waves due to spinal flexion and extension cycles [400].

Embryological Development of the Diaphragm Muscle

The DIAm derives from myoblasts that migrate along the pleuroperitoneal fold [24], which initially appears as bilateral folds of mesenchyme that project into the undivided thoracic cavity. The pleuroperitoneal fold ultimately fuses with the dorsal mesentery surrounding the esophagus and with the dorsal part of the septum transversum (formed from folds caudal to the developing heart and pericardium) to complete the partitioning of the thoracic and abdominal cavities (Figure 3).

With embryonic growth, there is a progressive caudal displacement of the pleuroperitoneal fold, with eventual attachments spanning from mid to lower thoracic levels. Myoblasts that will form the DIAm migrate along the pleuroperitoneal fold [417]. At the same time, the phrenic nerves emerge from the cervical spinal cord and grow following the trajectory of the septum transversum and migrating myoblasts [385, 388] (Figure 3). What chemical factor(s) drive or guide axonal growth of phrenic nerve is unclear, but the coincidental growth of phrenic nerve axons with formation of the pleuroperitoneal fold and migrating myoblasts suggests a linkage [12, 241]. Although the exact mechanisms remain elusive, branching of phrenic motor axons begins within the descending pleuroperitoneal fold at the same time that myoblasts migrate into the sternal, costal and crural regions of the DIAm [24]. Thus, somatotopic innervation of the DIAm is established early in embryonic development.

As mentioned above, the sternal and more ventral aspects of both costal and crural regions are innervated by phrenic motor neurons located in more rostral segments of the phrenic motor neuron pool in the cervical spinal cord, while more dorsal portions of the costal and crural regions are innervated by more caudal segments of the phrenic motor neuron pool [194] (Figure 16). This somatotopic pattern of DIAm innervation develops before initial synapse formation likely reflects the pattern of phrenic nerve outgrowth and branching on the pleuroperitoneal fold.

Figure 16:

The DIAm consists of four major anatomical divisions, the sternal, costal and crural muscular regions and the central tendon. The DIAm is innervated via the phrenic nerve, emanating from C₃-C₆. Innervation of the DIAm exhibits somatotopy, where phrenic motor neurons from the cranial regions of the phrenic motor pool innervate the more ventral sternal and ventral costal regions. Phrenic motor neurons from the more caudal portion of the phrenic motor neuron pool innervate the dorsal costal and dorsal crural regions of the DIAm. This figure has been amended from ref 608.

In rodents, the phrenic nerve emerges onto the pleuroperitoneal fold at about the 11^th day of embryonic development (E11) (Figure 13) [241]. As mentioned branching of phrenic motor axons occurs early with Schwann cells in close proximity, which actually appear to precede motor axon terminals. Guidance of Schwann cell and phrenic axon terminal outgrowth may involve a series of complex regulatory processes. Clearly, the phrenic nerve must initially target the pleuroperitoneal fold before contacting myoblasts and myotubes or myofibers. Subsequently, specialized axon terminals must form in association with myotubes and/or myofibers, where cholinergic receptor aggregation occurs in close proximity to form synapses. These events occur in a relatively short period of time within 2–3 days (E11-E14) in rodents, and the timing of the regulatory processes involved is critical. These events may be affected by muscle or nerve activity [30, 191, 192], mechanical properties the extracellular matrix, and/or the release of trophic factors and/or chemotactic substances by muscle or extracellular matrix [417, 487–492].

Agrin derived from motor neurons is important for development of neuromuscular junctions and thus, for development of excitation-contraction coupling mechanisms [29, 647]. Neuregulin [166, 563, 659, 712], expressed by motor neurons during embryonic development activates ErbB tyrosine kinase receptors located on DIAm myoblasts, myotubes and myofibers [267]. Neuregulin/ErbB signaling promotes protein synthesis in developing DIAm fibers [267] and also induces acetylcholine receptor expression and clustering [479, 480] at embryonic neuromuscular junctions via its effects on Schwann cells [325].

Diaphragm Muscle Innervation

In DIAm fibers, a series of electro- chemical- and mechanical transduction processes underlie neuromotor control of muscle activation by the nervous system. This excitation-contraction signaling process is begins with a nerve action potential (electrical signal), which is transduced at the neuromuscular junction to a chemical signal transmitted to receptors on the muscle fiber. This chemical signal then initiates another electrical signal (action potential) in the muscle fiber that is propagated and leads to another chemical signal (Ca²⁺ release from the sarcoplasmic reticulum), with subsequent binding of Ca²⁺ to TnC. This chemical signaling process results in cross-bridge formation and transduction to mechanical force or shortening (Figure 7). The interface between nerve and muscle fiber is the neuromuscular junctions, with specialized structures on the presynaptic nerve terminal and postsynaptic, motor endplate sides to mediate electrochemical transduction.

Neuromuscular Junctions

In animals, there are two different types of locomotor systems: one based on ciliary motion and the second that is based on muscle activation. Muscle-based locomotion is present in all metazoans except sponges (eumetazoan). Interestingly, sponges are the most primitive animals in the metazoan tree, and they lack a nervous system. The nervous system evolved in the eumetazoans (comprising cnidarians, ctenophores and bilaterians), and enabled these animals to sense their environment, process this information and then initiate a response that involved either neurosecretory or muscle-based (motor) systems. Response elements involving neurosecretion appear to have evolved first in early eumetazoan phylogeny. In contrast, the neural control of muscles and locomotion required the evolution of mechanisms mediating faster propagation of electrical signals, likely in bilaterian metazoan animals. Thus, there was a convergence in the evolution of striated muscle and neural control.

The nervous system controls skeletal muscle fiber force generation through neuromuscular synaptic transmission (Figure 17). Typically, multiple muscle fibers are innervated by a single motor neuron through branching of the motor axon. Collectively, the motor neuron and group of muscle fibers it innervates is called the motor unit [397, 594].

Figure 17:

Signaling between the phrenic nerve and the DIAm occurs across the neuromuscular junction. Neurofilament (red), labels phrenic nerve axons (presynaptic), with α-bungarotoxin labeling the postsynaptic acetylcholine receptors (green) on muscle fibers (with MyHC_2B expressing fibers labeled in blue) (A). The ultrastucture of the neuromuscular junction is illustrated using electron microscopy, with synaptic vesicles (acetylcholine) released at active zones via exocytosis into the junctional fold (B). Adapted from ref 425.

Neuromuscular synaptic transmission begins with the generation and propagation of an action potential in a motor neuron. The propagating action potential invades the presynaptic terminal, and the resulting depolarization activates voltage-dependent Ca²⁺ channels, resulting in Ca²⁺ influx, which triggers the release of synaptic vesicles containing acetylcholine (ACh).

At the presynaptic terminal, the influx of Ca²⁺ binds to synaptotagmin located on synaptic vesicles, which regulates fusion of synaptic vesicles to the presynaptic terminal membrane. Typically, during an action potential, a number of synaptic vesicles fuse to the presynaptic terminal membrane and the combined release of ACh diffuses across the synaptic cleft, i.e., the space between the presynaptic terminal and postsynaptic muscle membrane, where ACh binds to nicotinic cholinergic receptors (nAChR) located in a specialized region called the endplate. Binding of ACh to its receptor opens cation channels that results in depolarization of the muscle fiber membrane (an endplate potential - EPP). There can also be spontaneous synaptic vesicle release of ACh, and these events are reflected by a miniature end-plate potential (mEPP) [143, 173, 342]. The action potential mediated EPP is the quantal summation of mEPPs. The larger EPP depolarizes the sarcolemma surrounding the neuromuscular junction leading to the opening of voltage-gated Na⁺ channels and the generation of an action potential that propagates along the muscle fiber membrane. This electrical event can be measured using electromyography (EMG). Using discrete electrodes, the activation of a single motor unit can be measured which provides a window on the central nervous system control of muscle contraction.

Phrenic Motor Neurons

Diaphragm muscle fibers are innervated by phrenic motor neurons located within the ventral horn (lamina IX) of the cervical spinal cord, segments C₃-C₅ in rats [13, 431, 522, 637], echidnas [453], and humans [346]. While in the mouse the pool spans C₃-₆ [520] and C₄-₆ in cats [682] and rabbits [663]. In the rat, there are ~240 phrenic motor neurons on each side of the cervical spinal cord (Figure 18) [13, 190, 513, 522], bilaterally innervating the DIAm (providing a total of ~480 motor units).

Figure 18:

Phrenic motor neurons exist bilaterally in the cervical spinal cord (~C₃-₆) and are readily labeled by intrapleural injection of fluorescently-conjugated cholera-toxin subunit B (A). Phrenic motor neurons have extensive dendritic arborisations (B) and exhibit a large heterogeneity in somatic surface areas (C), passive electrical properties which largely determine motor unit recruitment order.

As mentioned above, phrenic innervation of the DIAm displays a somatotopic organization with more rostral cervical segments of the phrenic motor neuron pool innervating more ventral regions of the costal and crural regions of the DIAm, whereas more caudal segments innervate more dorsal portions of both the costal and crural regions (Figure 16) [622].

Adult phrenic motor neurons display a wide range of sizes, matching the range of DIAm motor unit types. In the rat, the somal surface areas of phrenic motor neurons range from 1,000 to 8,000 μm² with a median of 4,500 μm² (Figure 16) [190, 513, 515, 522]. In ~80% of cases, somal surface areas of phrenic motor neurons display a significant biomodal distribution, perhaps reflecting differences between phrenic motor neurons innervating type S and FR motor units (possibly including some relatively more fatigue resistant type FInt units) and larger phrenic motor neurons innervating type FInt and FF motor units.

During late embryonic and early postnatal development, ~50% of motor neurons within the mammalian spinal cord undergo pruning or programmed cell death [382]. Within the phrenic motor pool of the rat, this motor neuron loss occurs days before birth and numbers remain stable during late embryogenesis and early postnatal life within the mammalian spinal cord [12, 262]. In older animals, an age-related loss of phrenic motor neurons has been observed, and may underlie sarcopenia of old age [190, 347].

Embryological Development of the Diaphragm Muscle Innervation

Myoblast migration and myotubes formation is concurrent with the outgrowth of phrenic motor axons (Figure 13). Similarly, myotubes transformation into myofibers with contractile protein expression and sarcomeric organization is concurrent with the development of pre- and postsynaptic specialization, including the aggregation of cholinergic receptors. In the rodent (mouse and rat) DIAm, cholinergic receptors aggregate into clusters by ~E13-E14. This aggregation process for cholinergic receptors involves agrin secretion by the nerve terminal, which activates skeletal muscle specific kinase (MuSK) receptors at the postsynaptic membrane [215, 564, 565]. Following cholinergic receptor aggregation the postsynaptic membrane is further specialized by the formation of junctional folds (Figure 17).

In adult DIAm fibers, the morphology of neuromuscular junctions varies across different fiber types, with increased complexity and size at type IIx and/or IIb fibers compared to type I and IIa fibers (Figure 19) [421, 514].

Figure 19:

Different motor unit types exhibit different size and complexity of presynaptic axon terminal (red) and postsynaptic acetylcholine receptor (green) structures. Note that neuromuscular junctions of type IIx and/or IIb muscle fibers are markedly larger and more complex compared to neuromuscular junctions of type I or IIa muscle fibers.

At the time of synapse formation, each myotube or myofiber is contacted by multiple motor neurons (termed polyneuronal innervation) (Figure 20). Subsequently during late embryonic and early postnatal development, synapse elimination occurs and polyneuronal innervation disappears. The adult pattern of innervation of DIAm fibers by a single phrenic motor neuron is present by about the second postnatal week in the mouse and rat (Figure 13) [39, 40, 76, 504, 526, 564].

Figure 20:

At E16, many postsynaptic acetylcholine receptor clusters are not innervated yet by axons (arrows in A). At birth, polyneuronal innervation of one acetylcholine receptor cluster by two motor axons is clearly shown (arrow in B).

The mechanism(s) underlying perinatal synapse elimination remains unclear. A leading theory is that synapse elimination reflects a competition among motor neurons that depends on activity and/or efficacy of neurotransmission (Hebbian competition) [323, 504, 526]. This fundamental concept for synaptic neuroplasticity and assembly of cellular connections in the nervous system was introduced in 1949 by Donald Hebb, a Canadian psychologist/neuroscientist. The concept can be simply paraphrased as “cells that fire together, wire together”. In Hebb’s theory a synapse between two cells (e.g., a motor neuron and a muscle fiber) is strengthened if the synaptic transmission is effective as evidence by repeated and persistent stimulation of the postsynaptic cell. Accordingly, during the period of polyneuronal innervation of muscle fibers, the synaptic transmission of one motor neuron is more effective in activating the muscle fiber than other synapses and thus, this synapse is subsequently strengthened while other synapses are further weakened. Contrary to what some have argued, it is not a matter of a more active motor neurons persisting at the expense of less active motor neurons (i.e., purely activity dependent). Synapse elimination driven by activity alone during development is inconsistent with the fact that in adults, type FInt and FF motor units, with the largest motor neurons tending to have the largest innervation ratios (i.e., the number of muscle fibers innervated by a single motor neuron). In adults, these larger motor neurons are the least active [54, 195, 601]. Thus, during development, either the size and excitability of these motor neurons must change dramatically (transitioning from most to least active), or something other than activity alone accounts for the persistence of synaptic contacts of these motor neurons. However, differences in motor unit innervation ratios are entirely consistent with Hebb’s theory if there is something associated with improved synaptic efficacy that results in the strengthening of synaptic connections at larger motor units. One possibility is the release of neurotrophic factors (e.g., BDNF) that stabilizes the neuromuscular junction. Another possibility is the production of some trophic factor (“synaptomedin”) by the active muscle fiber that is then retrogradely transported by the presynaptic terminal and thus serves as an intracellular signal to facilitate and maintain the synaptic contact [28].

In mammals, including humans, respiratory movements are observed in the fetus indicating intact inspiratory drive to the DIAm and functional neuromuscular transmission. In the rat, fetal respiratory movements are present by E17 [244], but these movements may occur earlier since they are not easily detected. Functional synapses and excitation-contraction coupling in DIAm fibers (i.e., evoked intracellular Ca²⁺ and contractile responses to nerve stimulation) can be seen as early as E12.5 in rodents. It is possible that the transition of myotubes to myofibers and the establishment of the conventional sarcomeric organization in the DIAm may be influenced by mechanical responses evoked by neural signals.

It is not surprising that the efficacy of neuromuscular transmission in the fetal and neonatal DIAm differs from that in adults. For example, with repetitive phrenic nerve stimulation, neuromuscular transmission failure is more pronounced in the neonatal rat DIAm [34, 174, 193, 618]. It is possible that the more extensive branching of phrenic motor axons in the neonate due to polyneuronal innervation may increase susceptibility to branch point failure - a form of neuromuscular transmission failure [33, 45, 164, 193, 330, 432, 454, 618]. In addition, during the process of synapse elimination, weaker synapses are present that are more susceptible to neuromuscular transmission failure [193, 372, 374, 618]. In support, the frequency and amplitude of spontaneous mEPPs is reduced in the neonatal rat DIAm, and the amplitude of evoked EPPs is also reduced [34, 174, 193].

Motor Unit and Muscle Fiber Types in the Diaphragm Muscle

Diaphragm muscle force generation is dependent on the interplay between phrenic motor neurons and the muscle fibers they innervate. Motor units are commonly classified into four types (type S, FR, FInt and FF), according to the mechanical and fatigue properties of their constituent muscle fibers [195, 223, 224, 598, 608, 610–614] (Figure 21). Importantly, within an individual motor unit, all of the muscle fibers display homogeneous contractile protein expression and biochemical properties that define a specific muscle fiber type [195, 256, 477, 478]. This homogeneity of muscle fiber type within motor units has been confirmed in the adult cat DIAm [163, 331, 613, 622].

Figure 21:

Different DIAm motor unit types are distinguished by their intrinsic, mechanical and fatigue properties and are classified as type S, FR, FInt and FF (A). Within a particular motor unit, all muscle fibers are homogeneous, as evidenced by myosin heavy chain (MyHC) expression (B). In the DIAm of most species, type I and IIa diaphragm muscle fibers have smaller cross-sectional areas than those of type IIx and IIb fibers (C). Differences in specific force between different fiber types is related to the different MyHC content per half sarcomere and differing unitary forces produced by different MyHC isoforms (D). Adapted from elements within refs 186 and 223.

In modern classification schemes, different muscle fiber types are identified by the expression of different MyHC isoforms. Type I muscle fibers in slow-twitch (type S) motor units express MyHC_Slow. These type I muscle fibers have a high mitochondrial volume density and capacity for oxidative phosphorylation that contributes to their fatigue resistance [163, 614]. Furthermore, type I muscle fibers have smaller cross-sectional areas [395, 455, 513, 617, 708], and generate less force per cross-sectional area (specific force) [221–224]. Due to the enzymatic properties of the MyHC_Slow isoform, type I fibers display lower rates of ATP hydrolysis, slower cross-bridge cycling rats and slower maximum velocity of shortening compared to type II muscle fibers [221–224]. Fast-twitch, fatigue resistant (type FR) DIAm motor units comprise type IIa muscle fibers that express the MyHC_2A isoform [219, 220, 222, 223]. Type IIa muscle fibers have higher mitochondrial volume density and oxidative capacity comparable to type I fibers, which may account for their greater fatigue resistance in some species [259, 261, 621]. The cross-sectional areas of type IIa fibers is small, similar to that of type I fibers, but the specific force they generate is of similar magnitude to other type II muscle fibers [221–224]. The maximum ATP hydrolysis rate of type IIa fibers is lower than other type II fibers but higher than type I fibers (Figure 10) [625]. Accordingly, the maximum velocity of shortening of type IIa fibers is faster than type I fibers but slower than other type II fibers [221–224]. Type IIx and IIb DIAm fibers generally co-express MyHC_2X and/or MyHC_2B isoforms in varying proportions and comprise type FInt and FF motor units. In general, the higher proportionate expression of the MyHC_2B isoform in type IIx/IIb fibers, the greater is their susceptibility to fatigue. The relative co-expression of MyHC_2X and MyHC_2B isoforms in DIAm fibers also appears to underlie differences in mechanical and energetic properties [614]. Type IIx/IIb DIAm fibers have higher ATP hydrolysis rates and faster cross-bridge cycling rates and faster maximum shortening velocities than other DIAm fiber types [224, 259, 260]. Type IIx/IIb DIAm fibers also have both larger cross-sectional areas [246, 395, 455, 513, 597, 617, 708] and higher specific forces [221–224]; thus they have a greater contribution to muscle force generation (Figure 21).

An important factor in the force generation when recruited by a motor unit is the number of muscle fibers comprising the motor unit (innervation ratio). In limb muscles, the largest innervation ratios occur in type FInt and FF motor units and smaller ratios occur in type S and FR units [101, 336]. However, in the cat DIAm, there is no difference in the innervation ratio between motor unit types [194, 602, 604]. Taken together, the larger muscle fiber cross-sectional areas and the greater specific forces of type FInt and FF DIAm motor units result in markedly greater levels of force contributed by these motor units compared to type S and FR units [195, 423, 428, 609, 610, 612].

Although it was suggested that the costal and the crural regions of the DIAm were actually two separate muscles, with different embryological origin, and different in fiber type composition [140], this was subsequently shown to be incorrect in both respects [558, 622]. Firstly, all regions of the DIAm have been shown to derive from shared embryological origins [12, 24, 241, 242]. Secondly, systematic evaluations of the phrenic nerve innervation of the DIAm using glycogen depletion techniques clearly demonstrated considerable overlap in the cervical segmental innervation of DIAm fibers across sternal costal and crural regions [196, 603]. For a given species, the costal and crural regions of the DIAm have similar mixed fiber type composition. The relative fiber type proportions and cross-sectional areas for a given DIAm fiber type may vary across species [126, 251, 278, 348]. Similarly, DIAm fiber lengths and orientation may differ across species [558, 622]. Indeed, type I muscle fibers take up a very large proportion of DIAm fibers in some species of dolphin [141], whose rapid ventilation, ~90% lung volume exchange in under ~0.3 s [316, 368, 539], occurs mainly via epaxial, hypaxial and abdominal muscles [118]. For comparison to a terrestrial mammal, a galloping horse exchanges ~20% of its lung volume in 0.5 s [703], despite the horse being somewhat atypical, with a small tidal volume for its body size and exercise induced pulmonary hemorrhage [509]. For comparison, a trained human male during near maximal exercise will exchange ~50% of their total lung volume in ~1 s [250].

Like all skeletal muscles, DIAm force is related to the relative proportions of different fiber types (I, IIa, and IIx/IIb) and their cross sectional areas. A variety of conditions, including spinal cord injury, myasthenia gravis, spinal muscle atrophy, amyotrophic lateral sclerosis, corticosteroid treatment, mechanical ventilation and aging, lead to type-specific atrophy (reduced cross sectional area), force decline and fiber-type conversion or co-expression in the DIAm [186, 237, 416, 424, 426, 427, 429, 543]. Remarkably, in hibernating mammals, retention of type I muscle fibers and increased muscle mass is evident, despite uniform reduction in muscle mass and decreased protein levels in almost all other muscles, including those of mixed fiber types similar to DIAm [119].

From the standpoint of symmorphosis, it is interesting that the basic differences in contractile, fatigue and metabolic properties defining fiber types persist across species. Differences in the proportion and cross-sectional areas of these different fiber types will underlie difference in performance features of the DIAm across species. Accordingly, if there is a requirement for greater endurance of the DIAm (more fatigue resistance during ventilatory behaviors), the proportion of type I and IIa fibers and/or their cross-sectional areas may be greater in a particular species. Thus, the relative proportion of type I and type IIa fibers compared to type IIx and/or IIb fibers in the DIAm provides insight into its dual functional role in generating negative P_th versus P_ab, respectively. For example, in species where offspring size and maturity at birth are relatively greater (e.g., elephants, sea mammals), there may be a need for greater P_ab generation during childbirth. By contrast, in similarly sized animals where offspring are smaller and less precocial (e.g., monotremes and marsupials), there may not be as great of a demand for P_ab generation at childbirth. Accordingly, there may be difference in the proportion and size of type IIx and/or IIb fibers in the DIAm of these different species. However, in all of the mammalian species tested to date, the requirements of ventilation appear to be well met by the relative contribution of type I and IIa fibers in the DIAm.

The mean cross-sectional area of DIAm fibers scales with body weight, i.e., as body mass increases the size of all DIAm fibers increases [59, 126, 251, 278]. Sadly, many of these earlier observations did not determine cross-sectional areas specific to fiber type. When fiber type is taken into account, type IIx and/or IIb fibers have ~1.5–3 times greater cross-sectional areas than type I and type IIa fibers [222, 223]. Importantly, the specific force (force normalized to cross-sectional area) of different DIAm fiber types appears to be highly conserved across all species. In permeabilized rat DIAm fibers maintained at 15°C, the maximum specific force (at pCA 4.0 activation) of type I fibers is ~10 N cm⁻², type IIa fibers ~12 N cm⁻², and type IIx and/or IIB fibers ~15 N cm⁻². Considering the Q10 for force generation is ~2.0, at normal body temperature (i.e., ~37°C), specific forces would range between ~22–25 N cm⁻² in type I and IIa fibers to 32–35 N cm⁻² in type IIx and/or IIb fibers. These differences in specific force across fiber types are due in part to differences in contractile protein content per half sarcomere (n in the force equation) [222–224, 259, 621]. In the rat DIAm, we showed that type I and IIa fibers have lower MyHC content per half sarcomere compared to type IIx and/or IIb fibers. When isometric force is normalized for MyHC content per half sarcomere, the force per MyHC content of fibers expressing MyHC_slow is only ~ half that generated by all type II fibers regardless of type II MyHC isoform (i.e., MyHC_2A, MyHC_2X and/or MyHC_2B) (Figure 21). These results suggest that each MyHC head contributes a quantal level of force during cross-bridge cycling that only differs between slow (type I) and fast (type II) MyHC isoforms. Accordingly, the total amount of force contributed by a DIAm fiber depends on MyHC isoform (type I or II) and the fraction of MyHC heads that are attached to the actin filament to form cross-bridges (α_fs in force equation). Thus, differences in P_th and P_ab generation across species are likely to be associated with the thickness of the DIAm. For example, in elephants the DIAm is ~3 cm [79], an order of magnitude greater than that observed in humans, ~3 mm [21, 22], an order of magnitude difference. Additional considerations include proportions of muscle fiber types (i.e., motor unit types) within the DIAm, and the patterning of activation of costal and abdominal wall musculature when performing different maneuvers.

Neuromotor Control of the Diaphragm Muscle

Although much is still to be defined in regard to phrenic motor neuron inputs, it remains immutable that individual phrenic motor neurons are the final common output as well as the integrators of signals from pattern generators and other neural circuits. The DIAm motor unit remains the final executor of neuromotor control and produces motor force output across a range of ventilatory and higher force, non-ventilatory behaviors. It still remains unclear whether phrenic motor neurons constituting the different types of DIAm motor units receive diverse premotor inputs. Regardless, neuromotor control of the DIAm during different motor behaviors requires production of graded levels of force generation, a property dependent on recruitment and rate coding of motor units themselves (Figure 22).

Figure 22:

The initial model for force-frequency coding and DIAm motor unit recruitment was based on direct measurements performed in cats (A). Onset activation frequency for type S and FR motor units is ~8 Hz and maximal activation range is ~25 Hz (blue portion). For type FInt and FF motor units, onset is ~12 Hz and maximal activation ~60 Hz (green portion). The steepest portion of the force-frequency curve occurs between 10 and 30 Hz for all types of motor units in the diaphragm muscle (B). Individual motor unit recordings show that motor units with larger discharges are recruited after those with smaller discharges (C). Motor units are recruited in an orderly fashion with type S > FR > FInt > FF (D). Recruitment of type S and FR motor units is sufficient to accomplish ventilatory behaviors, including eupnea, response to hypoxia/hypercapnea and breathing against an occluded airway. To accomplish higher force expulsive/straining behaviors, such as coughing, sneezing, vomiting and defecation, recruitment of higher-force generating type FInt and FF motor units is necessitated. In general, diaphragm motor units operate in the steep portion of the frequency-coding curve (i.e., between 50–100% activation). Adapted from elements within refs 186.

Motor Unit Recruitment and Frequency Coding

The intrinsic size-dependent electrophysiological properties of motor neurons determine the recruitment order of motor units [87, 345, 610]. Mathematically, the relationship between motor neuron size and excitability (recruitment order) may be expressed as the following equation:

$$d{V}_m/dt=I_{syn}/C_m$$

Where dV_m/dt is the change in membrane potential with time, I_syn is the net synaptic inward (depolarizing) current and C_m is membrane capacitance. Thus, for a given I_syn, smaller motor neurons with a lower C_m have a greater dV_m/dt compared to larger motor neurons, reaching a threshold for action potential generation sooner (Figure 23).Smaller motor neurons have smaller C_m; thus, for a given amount of I_syn, the dV_m/dt is greater, reflecting greater excitability [179, 644, 704, 706]. Accordingly, if synaptic input is comparable across phrenic motor neurons, smaller more excitable motor neurons (presumably type S and FR) are recruited before larger motor neurons (presumably type FInt and FF) (Figure 22).

Figure 23:

The order of motor unit recruitment is related to the intrinsic properties of motor neurons. Of these the most important is motor neuron surface area, the size principle. Neuronal membrane acts as a capacitor, with total membrane capacitance (C) primarily determined by membrane surface area (A) and distance between the membrane lipid bilayer (d). As membrane bilayers are unchanged, a larger neuronal surface area (A₂ > A₁) will increase capacitance. For a given synaptic input (I_syn), the excitability of the membrane (dV_m/dt) is inversely related to neuronal capacitance. Thus, smaller motor neurons (type S and FR) with low capacitance are more excitable and recruited before larger motor neurons (type FInt and FF).

Motor unit recruitment order generally matches motor unit type, with initial recruitment of type S and FR motor units followed by type FInt and FF motor units (Figure 22) [86, 442, 644]. In agreement, it has been shown that motor units recruited first contribute less force compared to those units recruited later [704]. In the DIAm, it is very likely that type S and FR motor units are initially recruited to accomplish the lower force, high duty cycle (active vs. inactive) behavior of breathing. During rhythmic breathing, inspiration typically has a duty cycle of 30–40%, which would cause marked fatigue if type FInt and FF motor units were recruited. Higher force, expulsive sneezing and coughing behaviors of the DIAm have very short durations allowing sufficient time for recovery from any fatigue that might occur.

Based on these observations, a model of DIAm motor unit recruitment was formulated based on an orderly recruitment of type S, FR, FInt and FF motor units (Figure 22) [195, 423, 428, 581, 602, 604, 605, 610]. In the initial model developed for the cat DIAm, the force contribution of each motor unit type was based on direct measurements of motor unit mechanical properties [195, 598, 601, 604, 609, 610]. Subsequently, in the rat DIAm, the force contributed by different motor unit types was estimated based on measurements of Ca²⁺ activated specific force generated by single permeabilized muscle fibers of different types [221–224], the mean cross-sectional area [395, 455, 513, 617, 708] and proportions [163, 195, 598, 613, 614] of these muscle fiber types, and the assumption that innervation ratios are comparable across motor unit types [194, 602, 604]. In each of these models, ventilatory behaviors (i.e., eupnea and response to hypoxia/hypercapnea) were assumed to be accomplished by the recruitment of only fatigue resistant type S and FR motor units [423, 428, 599, 600, 602, 604, 610]. Based on these assumptions, it was estimated that in cats, ~23% of the total phrenic motor neuron pool is recruited during eupnea [326], corresponding to the relative proportions of type S and FR motor units in the DIAm [195, 610, 613]. To perform higher force, expulsive behaviors (i.e., coughing and sneezing), the recruitment of additional type FInt and FF motor units is required [423, 428, 599, 600, 602, 604, 610]. The progressions in force generated by DIAm motor units (type FF > FInt > FR > S) [195, 221–224, 609, 610] results in various slopes of force development during the sequential recruitment of motor units (Figure 22).

With increasing rates of activation, skeletal muscle force increases, reaching a plateau reflecting maximum tetanic force. As frequency of activation increases, force responses fuse into larger summated forces, thus the force-frequency response curve is sigmoidal. In type S and FR motor units, the force-frequency response is shifted leftward, such that tetanic fusion of force occurs at lower activation frequencies compared to type FInt and FF motor units (Figure 22) [195, 624]. Using fine wire electrodes implanted into the DIAm the discharge of motor unit action potentials was recorded during ventilatory behaviors [87, 367, 624]. For motor units recruited during these ventilatory behaviors, the onset discharge rate was lower compared to the peak discharge rate occurring later within the inspiratory burst [318, 624]. In the cat DIAm, motor unit discharge rates were compared to the force frequency response curves of different motor unit types [195]. Consistently, it was observed that the onset discharge rate of low threshold motor units (i.e., those recruited early during inspiration) was ~8–10 Hz with discharge peaking at ~25 Hz toward the end of inspiration. This range of discharge rates corresponds to the steepest part of the force-frequency response curves of type S and FR motor units. In contrast, DIAm motor units with higher recruitment thresholds (i.e., those recruited later during more forceful inspiratory efforts) had higher onset (~15 Hz) and peak (~60 Hz) discharge rates, which corresponded to the steep part of the force-frequency response curves for type FInt and FF motor units [624]. Based on these results, a modified model for DIAm motor unit recruitment was developed, which included frequency coding of different motor unit types (Figure 22) [422, 423, 581].

In the DIAm, motor units are likely recruited systematically depending on motor unit type (fatigue resistant to more fatigable) to accomplish a range of forces and different motor behaviors [423, 428, 582, 599, 600, 602, 604, 605]. Recruitment order is maintained even with increasing neural drive (I_syn), with a decrease in recruitment delay and an increase in discharge rate [582]. For those low threshold DIAm motor units that are recruited consistently across all ventilatory behaviors (i.e., eupnea and hypoxia-hypercapnia), onset discharge rates are comparable [581, 582]. We assume these low threshold DIAm motor units are type S and FR, which develop lower forces but are fatigue resistant, and thus, especially appropriate for high duty cycle repetitive motor behaviors such as breathing [367, 422, 423, 581]. In contrast, to achieve higher force, shorter duration (lower duty cycle) motor behaviors, it is more appropriate to recruit higher force but more fatigable type FInt and FF motor units. The size principle for motor unit recruitment in the DIAm has been confirmed [145, 326, 423, 581, 599, 600, 604, 605, 610, 620] and underlies neuromotor control of force generation across a range of motor behaviors.

Components of Neuromotor Control System

There are five main motor components of the neuronal circuitry of the phrenic/DIAm motor system: 1) a central pattern generator (for ventilatory – P_th generation; or expulsive/straining – P_ab generation - motor behaviors); 2) premotor neurons responsible for transmitting the output of the central pattern generator; 3) interneurons responsible for modulating or coordinating premotor neuron and/or phrenic motor neuron excitability – these interneurons also serve to integrate sensory feedback (e.g., chemoreceptive, lung stretch, propriospinal or other afferent inputs); 4) direct cortical premotor input to motor neurons via the corticospinal pathway; and 5) phrenic motor neurons as the final common output for generating the forces necessary for the desired motor behaviors (Figure 24).

Figure 24:

Neuromotor control of DIAm ventilatory and expulsive/straining behaviors requires cortical, brainstem and spinal cord centers. Ventilatory behaviors are the most-well characterized of these systems and require the recruitment of predominantly type S and FR motor units. Cortical pathways are able to modulate the eupnic rhythm by interactions with the ventilatory central pattern generator (CPG) or directly via synapses onto phrenic motor neurons (PhMNs). The ventilatory CPG activates brainstem premotor neurons that in turn innervate the PhMNs. Activity of PhMNs during ventilation is also modulated (directly and indirectly) by spinal cord ascending tracts and interneurons. Brainstem chemoreceptors and lung mechanoreceptors regulate the activity of premotor neurons, and act to increase premotor neuron discharge (and thus PhMN activity) during hypoxia/hypercapnia. In the case of expulsive/straining behaviors, the majority of control centers are located within the spinal cord, and recruitment of type FInt and FF motor units (higher-force producing units) is necessitated. Some cortical control of the PhMNs and spinal expulsive/straining CPG may be evident, but rectal and vaginal stretch receptors also elicit strong P_ab generation. There may be shared spinal premotor neurons within the spinal cord for PhMNs and abdominal muscle MNs, and a variety of ascending projections may facilitate the coordinated activity of all MNs involved in expulsive/straining maneuvers. Overall, expulsive behaviors result in near maximal co-contractions of the DIAm and abdominal wall muscles.

The central pattern generator for locomotion in the spinal cord was characterized much earlier than that for respiration. Indeed, initial studies were performed by the French physiologist Marie-Jean-Pierre Flourens, who in 1824 conducted brain and spinal cord lesion studies to determine the origin of motor behaviors. Subsequent work in the early 1900’s by Charles Sherrington, Thomas Graham Brown and others showed that stereotypical rhythmic motor behaviors such as walking were controlled by a spinal cord neuronal network, a central pattern generator.

Sherrington’s seminal research examining the effects of spinal cord transection showed that locomotor movements are modulated by spinal segmental reflexes [593]. However, it was the work of Thomas Graham Brown, who provided clear evidence for the local control of locomotion behaviors within the spinal cord [80]. In dogs with thoracic spinal cord transections preventing descending cortical control, spontaneous rhythmic locomotor patterns of the hindlimb were generated by a local spinal neuronal network, independent of peripheral input, i.e., these patterns were the result of a central pattern generator for locomotion [80].

It is likely that several distinct central pattern generators affect activation of the DIAm to accomplish motor behaviors, e.g., separate central pattern generators for breathing, sighing, sneezing, coughing, swallowing, vomiting, vocalization, defecation, etcetera (Figure 24). Of these, the neuronal circuitry responsible for generating the rhythmic behavior of breathing (ventilation) is the best characterized central pattern generator affecting phrenic motor neurons and DIAm activation. For ventilatory behaviors, this central pattern generator plays an indispensable role in determining the timing and duration of the different phases of the respiratory pattern.

Other central pattern generators affecting DIAm neuromotor control may also exist in the brainstem and spinal cord, e.g., for sighing [396], swallowing [146], coughing [56, 447], sneezing, vomiting [364, 447], defecation and parturition (Figure 24). The output of these other pattern generators is transmitted to premotor neurons and then phrenic and other motor neurons. In addition, the output of different central pattern generators that affect DIAm activation, such as those for swallowing or vomiting, must be integrated with the respiratory pattern generator that sets the baseline of activation and relaxation of DIAm activity.

Neuromotor Control for Intra-Thoracic Pressure Generation

The major functional demand for negative P_th generation is ventilation of the lungs for gas exchange. Lung ventilation consists of three main phases, inspiration, post inspiration and expiration [536, 537]. The precise mechanisms underlying the central pattern generator for respiration are still debated, but it is now generally agreed that the Pre-Bötzinger Complex (PreBötC) in the medulla provides the spontaneously active ‘kernel’ of neurons for the metronomic drive for the inspiratory phase of respiration, via interactions of various membrane channels (Figure 24) [35]. It now appears that the PreBötC is essential for inspiration [438], and is the prime source of inspiratory excitatory drive to respiratory premotor neurons [635] via a core subpopulation of glutamatergic (Glu) pacemaker cells that project bilaterally to premotor neurons in the medulla (Figure 24) [365]. Inhibitory neurotransmission at premotor neurons also plays an important role in modulating respiratory pattern generator outputs [42, 228].

The Bötzinger Complex actually located rostral to the PreBötC provides for switching from inspiration to expiration [536, 537], possibly via inhibitory inputs to the PreBötC [538]. Normally, expiration is passive; however, during increased respiratory efforts, active expiration involves activation of intercostal and abdominal muscle and this activation has a distinct central pattern generator located in the region of the retrotrapezoid nucleus (RTN) [322].

Premotor neurons providing monosynaptic drive to phrenic motor neurons during inspiration (ventilatory behavior) are located primarily in the ventrolateral medulla (ventral respiratory group – VRG), although there is some contribution from a dorsal respiratory group (DRG) in the dorsomedial medulla (Figure 24) [324, 493]. These descending excitatory (Glu) premotor inputs for inspiratory-related activation of phrenic motor neurons are located predominantly ipsilateral in the spinal cord [147, 160, 176], transmitted via bulbospinal pathways in the ventrolateral and ventromedial funiculi [175, 176, 635]. These excitatory premotor inputs for respiratory-related activation of phrenic motor neurons are generally thought to be widely distributed [113, 127], with recruitment of phrenic motor neurons (DIAm motor units) dependent on intrinsic electrophysiological properties of phrenic motor neurons rather than a specific pattern of premotor input.

It remains unclear whether the other central pattern generators that affect phrenic motor neuron and DIAm activation during other non-ventilatory motor behaviors share common or distinct premotor neurons. It is clear that the central pattern generator involved in breathing is different from that involved in generating the patterns of expulsive/straining motor behaviors of the DIAm. However, these different pattern generators may have a certain degree of overlap in their inputs to the medullary premotor neurons affecting phrenic motor neuron activation [554, 635]. For ventilatory behaviors, it is generally thought that phrenic motor neurons receive distributed descending medullary premotor input that is predominantly ipsilateral. However, retrograde tracing studies that examine the connectivity of phrenic motor neurons have indicated a variety of premotor inputs both from the brainstem and spinal cord [145, 160, 161, 303, 326, 403, 557]. Perhaps these connections reflect distinct premotor inputs mediating the output of central pattern generators that are different from those involved in respiratory drive to phrenic motor neurons. For higher force non-ventilatory behaviors of the DIAm, premotor input to larger (higher threshold) phrenic motor neurons may be different compared to premotor input mediating the output of the respiratory central pattern generator, which activates primarily smaller, lower threshold phrenic motor neurons (i.e., those innervating fatigue resistant type S and FR motor units). In support, we recently found that unilateral spinal cord hemisection at C₂, which silenced respiratory-related activation of the DIAm, predominantly affected Glu synaptic input to smaller phrenic motor neurons, while Glu synaptic input to larger phrenic motor neurons was far less affected. This differential effect of C₂ spinal hemisection on Glu input to phrenic motor neurons was consistent with the persistence of DIAm activation associated with higher force motor behaviors [62, 149–151, 207, 208, 236, 238, 239, 418, 562]. These results suggest that small versus large phrenic motor neurons may be involved in different motor behaviors – smaller phrenic motor neurons in ventilatory behaviors; larger phrenic motor neurons in expulsive/straining behaviors (Figure 24). If true, the larger phrenic motor neurons may be primarily influenced by central pattern generators and premotor neurons involved in expulsive/straining behaviors, which are located in the spinal cord (Figure 24).

Phrenic motor neurons also receive input from spinal cord interneurons (Figure 24), including those involved in proprioception. However, in stark contrast to muscles involved in locomotor behaviors [519], the DIAm has very few, if any, muscle spindles [156]. Thus, in contrast to well-studied limb muscles, direct muscle spindle proprioceptive feedback from the DIAm does not contribute substantially to modulation of phrenic motor neuron excitability [117, 321]. However, phrenic motor neurons do receive feedback from muscle spindles in intercostal muscles, and this input has primarily inhibitory effects on phrenic motor neuron excitability [158, 529]. In particular, an intercostal to phrenic reflex has been characterized, which suppresses phrenic nerve activity following strain on the chest wall [142, 530], an effect that appears to involve both disfacilitation of VRG premotor input [55, 530, 589] and interneuronal inhibition of phrenic motor neurons [36]. Additional local inhibition of phrenic motor neurons from interneurons within the spinal cord has been characterized [36, 37, 153, 154, 404]. Remodeling of local interneurons within the spinal cord may partially account for the spontaneous recovery of ventilatory and non ventilatory behaviors following cervical spinal cord injury [383, 711].

In humans, direct corticospinal inputs onto phrenic motor neurons allow for voluntary control of breathing [209, 407] and the interplay between ventilation and behaviors such as speech [592] (Figure 24). Neuromodulatory inputs (e.g., serotonergic – 5-HT) also affect changes in phrenic motor neuron excitability during sleep-waking states. For example, it is well documented that there is resilience to apnea during the waking state and an increased response to hypoxic and hypercapnic stimulation of ventilation [125, 178]. By contrast, sleep predisposes to episodes of apnea even during hypoxic, hypercapnic conditions [125, 631].

The depression of respiratory activity during sleep is remarkably similar to the depression of respiratory activity during diving [96, 97]. Apnea is a vital aspect of the diving response, which also includes bradycardia and peripheral vasoconstriction, in all air breathing vertebrates [15, 63, 88]. Mammals and other endotherms are constrained in their diving lengths and durations [63, 88], though many species are capable of extended dives ~2 hours in some seals and whales [276, 575]. Tolerance of hypoxia during extended dive durations are largely due to blood and tissue adaptations, including locomotor muscle fiber-type differences [626, 667], unrelated to the neural control of breathing [130, 369, 481, 636]. Obviously, limiting the chemoreceptor associated inspiratory drive is pertinent while submerged, yet, during sleep, seals experience bouts of apnea that can exceed 20 min, with no pathology or deleterious effects [97], and do not awaken in order to cease these bouts of apnea, with sleeping eupnea-apnea-eupnea cycles occurring uninterrupted [98]. Indeed, many seals sleep submerged, and rapid eye movement sleep of some seals involves complete cessation of ventilation, regardless of whether sleeping on land or submerged, perhaps due to muscle atonia of sleep extending to the DIAm in these species [97]. In longitudinal studies, the durations of apnea increased with age, correlating with the establishment of marked sinus arrhythmias [48, 99]. From a neural control standpoint, apneas in seals do not involve movements of the respiratory muscles, nor do they attempt to breathe [97]. In contrast, human apneas involve the movement of respiratory musculature and involve tissue damage, and apnea-related sudden infant death syndrome may involve impaired cardiorespiratory sinus arrhythmia development [97, 341]. Though many human apneas are obstructive, many are initiated centrally and related to arousal. The investigation of routine, repeated and extended periods of diving and sleeping apnea in the absence of deleterious consequences in seals and the random and pathological human episodes of apnea may provide more detailed information about the neuromotor control of ventilatory function.

Although much is still to be defined in regard to phrenic motor neuron inputs and the central pattern generation of ventilatory and other straining and expulsive motor behaviors of the DIAm, the individual phrenic motor neuron is the final integrator of these signals. The DIAm motor unit remains the final executor of neuromotor control and produces motor force output across a range of ventilatory and higher force, straining and expulsive behaviors. It remains unclear whether phrenic motor neurons innervating different types of DIAm motor units receive different premotor inputs. Regardless, neuromotor control of the DIAm during different motor behaviors requires production of different levels of force generation, a property dependent on recruitment and rate coding of motor units themselves.

Modulation of the respiratory pattern and/or of phrenic motor neuron activity may occur directly or indirectly in response to afferent inputs to phrenic motor neurons, from signaling initiated by mechanoreceptors in the lung and airway, peripheral and central chemoreceptors, to behavioral state influence mediated by serotonergic projections emanating from the raphe (Figure 24). Mechanoreceptors in the lung respond to lung inflation, are sensitive to the mechanical loading of breathing and prevent airway over-inflation by increasing their afferent activity – peaking at the end of inspiration [90, 474]. These afferent inputs exert effects on ventilatory phrenic motor neuron discharge indirectly, via vagal nerve inputs in the nucleus tractus solitarius (NTS) [43, 148]. Laryngeal mechanoreceptors also exert an indirect effect on phrenic motor neurons, decreasing inspiratory drive during upper airway collapse [566].

The carotid bodies, peripheral chemoreceptors, that respond to hypoxia and hypercapnea by increasing ventilation [2, 378, 379, 671] act indirectly via signaling to brainstem respiratory centers through the carotid sinus nerve [710]. Many brainstem areas contain central chemoreceptors [475], which are exquisitely sensitive to deviation of pH and pCO₂ [497], and act in response to hypercapnea to increase ventilation [254]. The field is full of intense debate regarding the particulars of the interactivity of peripheral and central chemoreceptors. Regardless of their precise activities, the indirect modulatory action of chemoreceptors (likely acting via medullary premotor neurons) leads to a heightening of the overall drive to phrenic motor neurons.

Serotinergic neurons located in the caudal raphe project to brainstem respiratory regions and to phrenic motor neurons [273], and thus may have effects on rhythmic central pattern generator and premotor outputs [380, 381] as well as on phrenic motor neuron excitation [401, 458]. Catecholamine modulation of respiration also occurs via activation of the α₁ or α₂ adrenoreceptors, enhancing or inhibiting, respectively respiratory rhythmic central pattern generation [272, 661, 669, 670].

Neuromotor Control for Intra-Abdominal Pressure Generation

As previously mentioned, activation of the DIAm is also involved in the generation of positive P_ab required for straining behaviors, such as vomiting, defecation and child birth. Expulsive behaviors that require near-maximal inspirations before perform effectively, such as coughing and sneezing, that are essential to maintaining a patent airway for breathing also require more forceful activation of the DIAm and generation of increased P_ab. Similar to inspiration, these expulsive maneuvers involve the coordinated activation of various abdominal wall muscles (including the DIAm), sphincter muscles, and upper airway muscles. For example, vomiting requires the coordinated contraction of the DIAm and other abdominal wall muscles, relaxation of pharyngeal muscles, protrusion of the tongue and the relaxation of the lower esophageal sphincters. By contrast, defecation, which involves the generation of P_ab ranging between 150–200 cm H₂O [93]) requires coordinated contraction of the DIAm and other abdominal wall muscles and pharyngeal muscles in order to close the glottus. A similar voluntary effort is the Valsalva maneuver, where after a deep inspiration, an individual attempts a forceful expiratory effort against a closed airway, thereby generating maximum positive P_ab ~90 to 220 CmH₂O [109, 253, 591]. This maneuver was introduced by a 17^th century Italian anatomist/physician, Antonia Valsalva who applied this procedure to clear the Eustachian tube. The reverse is the Müller’s maneuver, where, following a forced expiration, an individual attempts a forceful inspiratory effort against closed airway, thereby generating a maximum negative intrathoracic pressure. This maneuver was introduced by the 19^th century physiologist Johannes Peter Müller during which he demonstrated dramatic effects on the cardiovascular system due to increased venous blood return to the heart.

Extensive coordination between abdominal and pelvic musculature and the respiratory musculature have been observed during straining and voiding behaviors, including vomiting, defecation and child birth. Vomiting behavior is initiated by various stimuli, including motion, repugnant events, pregnancy, gastric irritation, toxins and drug effects [301]. Vomiting behavior is conserved across multiple species, including invertebrates, fish, amphibia, reptiles, birds and mammals [16, 17, 60, 114, 436]. Vomiting requires two phases; i) retching and ii) expulsion. The retching phase consists of cycles of co-contraction followed by simultaneous relaxation of DIAm and abdominal wall muscles [301]. During retching, P_th are reduced and P_ab are increased [300]. The expulsion phase consists of prolonged abdominal muscle contractions, coordinated with relaxation of the DIAm, and activation of intercostal muscles, the larynx and the pharynx. The maneuver occurs with a closed glottis and an elevated soft palate [301]. P_ab during vomiting have been estimated using gastric pressures, with positive pressures ranging from a mean of ~155 to a maximum of ~400 CmH₂O [315]. The initiation of vomiting involves the integration within the NTS of cortical, vestibular, area postrema and vagal afferents [23, 300, 301]. Due to the complicated orchestration of contractions of upper airway, DIAm, thoracic and abdominal musculature, and the remarkable range of stimuli provoking emesis, the control of vomiting is distributed throughout the medulla [301] and the phases are governed by a central pattern generator [364, 447]. Results from lesion, pharmacological and electrical stimulation studies show that regardless of how vomiting is elicited, the NTS is coordinating the response [138, 319, 363, 449, 450, 486, 525, 533], likely via axonal projections to motor control targets within the dorsal motor nucleus of the vagus, nucleus ambiguus and phrenic motor nucleus [319, 448, 450].

Defecation maneuvers often involve co-activation of the DIAm and the abdominal wall muscles. These straining efforts are characterized by the Valsalva maneuver, whereby maximal inspiratory efforts are followed by generation of P_ab against a closed glottis [202–205]. This coordination of the DIAm and abdominal musculature is likely to use the same pattern generation and reflex pathways that co-activate these muscles during vomiting behaviors, particularly projections from the nucleus retroambiguus [176, 284, 285, 289–291]. In spontaneous defecations, the EMG activities of the DIAm and abdominal muscles exceeds that of coughing [128]. In addition, projections from the nucleus retroambiguus onto Onuf’s motor nucleus within the sacral spinal cord ensures external anal sphincter opening during defecation, the only straining behavior where rectal continence is absent [180]. Currently, very little is known about the central control of defection, with the existence and location of a central pattern generator unknown, although central pattern generators have been observed for other pelvic functions, such as ejaculation [660]. Despite these limitations, though there is evidence for supraspinal brainstem defecation centers in the pons [200, 201, 206], though spinal cord lesion studies in rats and man indicate these supraspinal influences are superimposed over existing spinal reflexes [230, 409, 472, 672]. In accordance, studies have confirmed co-activation of DIAm and abdominal muscles following distension of the rectum [204], a reflex mediated by the Kölliker-Fuse nucleus, a well-described pneumotaxic center [44, 112, 702]. We assert that descending voluntary controls, in concert with central pattern generators within the spinal cord co-activate the DIAm and abdominal wall muscles to generate high P_ab (Figure 24).

Rhythmic straining behaviors of DIAm and abdominal wall muscles are elicited by distension of the vaginal walls [202, 205], similar to voiding behaviors initiated by bladder or colon distension. Currently there is an astonishing paucity of reliable studies into the neural and/or reflex control of increased P_ab generation. Hints as to the necessity of an ‘emotional nervous system’ [286–288], have shown that pathways within the periaquiductal grey region provide a level of supraspinal control for pelvic expulsive functions, including micturition and defecation, and sexual functions in females and males [286, 287, 314], however none have dealt specifically with parturition. Clearly, expulsive behaviors such as parturition may be facilitated by Valsalva maneuvers, however spontaneous expulsive contractions also occur during labor, with little knowledge of their neural control.

Overall, the neural control of low-force rhythmic motor behaviors (locomotion) and low to moderate intra-thoracic ventilatory pressures (eupnea and hypoxia/hypercapnea) are well established and increasingly defined in terms of specific central pattern generation circuitry. In our opinion, far less progress has been made in the characterization of the central pattern generators necessary to co-ordinate for high P_ab generation in expulsive and straining behaviors. These high-force behaviors are often impaired in neuromotor conditions, such as ALS, and during normal aging [181, 182, 190, 247, 248, 347]. Understanding the fundamentals of the neural control of expulsive behaviors will provide us with targets to alleviate some of the more deleterious consequences of neuromotor disease, injury or aging, many of which affect multiple components of the neuromotor system [185, 187–189, 347, 418–420, 516, 522, 640].

Development of Diaphragm Motor Behaviors

The maturation and adaptation of the respiratory system of newborn mammals following birth must be regarded as one of Nature’s all-time greatest achievements. The abrupt abandonment of the liquid uterine environment, to the air-breathing scenario is perilous. Indeed, not only is the change of environment a shock, but mammalian neonates generally (excepting some very small mammals, and the marsupials/monotremes) have a higher resting O₂ consumption (VO₂) requirement compared to adults of the same species [6, 27, 31, 47, 53, 121, 274, 307, 451, 459, 464, 465, 641, 655]. The exquisitely tailored structural and functional changes in mammalian neonates in adaptation to this shock begin in utero [240], with fetal breathing movements thought to prime neuromotor control of the respiratory muscles [12, 24, 243, 244, 362, 532], required for both ventilation and suckling in the newborn. The structural and functional dimensions of the first breath and the establishment of a steady and effective ventilatory behavior will be examined.

During in utero development, the lungs are not collapsed but maintained in a distended state [295, 459]. This distension is provided by fetal lung liquid [295], secreted by the pulmonary epithelium and serving to expand the lungs [485], particularly during the final trimester [295]. The amount of fetal lung fluid has direct effects on the development of lung size, with increased fluid increasing lung size [11, 91] and decreased fluid reducing lung size [11]. The latter has particularly disastrous consequences, as neonates born with pulmonary hypoplasia have a poor prognosis for morbidity and mortality [310, 360, 456]. Indeed, pulmonary hypoplasia as sequelae from congenital diaphragmatic hernia is the main concern for neonatal health following surgical correction of the defect [8, 470, 545, 546, 705]. Secretion of fetal lung liquid is tightly regulated. The inhibition of fetal lung liquid occurs by the action hormones such as adrenalin and vasopressin [77, 94, 95, 293, 498, 499, 677, 678, 701] and by O₂-dependent mechanisms, with hypoxaemia a particularly potent inhibitor [292, 294]. Fetal breathing movements are essential to the development of effective neuromotor circuits to control DIAm motor units [106, 131, 240] and to release episodically fetal lung liquid out of the trachea to be swallowed or absorbed by the amniotic fluid [295]. Episodic bouts of fetal breathing movements are essential for normal lung growth – possibly due to a relationship with fetal lung liquid [295].

Despite the essential function of fetal lung liquid during embryonic development, this accumulated fluid must be removed at birth for effective gaseous exchange to occur. Interestingly, by the onset of labor, the amount of fetal lung liquid accumulating in the lungs is ~30 ml kg⁻¹ [459, 482], very similar to the volume of air (functional residual capacity) present in the lung after birth ~20–25 ml kg⁻¹ [320, 544]. During labor, production of fetal lung liquid ceases, likely in response to adrenalin and vasopressin, due to an inhibitory effect that is exponentially more potent in late gestational ages [77, 94, 95, 293, 498, 499, 677, 678, 701]. Reabsorption is achieved through lymphatic drainage and diffusion into the pulmonary circulation [81, 231], both of which are enabled by the high permeability of the pulmonary endothelia and reduced vascular resistance of the pulmonary circulation in the newborn [312, 482]. It is important to note that fluid clearance during birth is not contingent on P_th generated by passage of the fetus through the pelvic canal in delivery, even though this does cause some expulsion of fetal lung fluid [58, 338, 567], caesarian deliveries display no appreciable rise in leftover fluids within the lung [26]. Overall, regardless of the mode of delivery, fetal lung liquid is reabsorbed during labor and the primary action of the newborn after birth is the first inspiratory breath.

A profound event in both the philosophical development of parents and the physiological development of the newborn is the first neonatal breath. The first breath involves the same stages as every breath thereafter, the inspiratory phase and the expiratory phase. The P_di pressures generated during the first breath (between 30 and 100 cmH₂O) approach that of maximal forces generated by adults, and dwarf the forces of neonatal steady-state ventilation (5–7 cmH₂O) established soon after birth [459]. There are many hypotheses proposed for the enlarged pressure generation of the first neonatal breath, ranging from the DIAm contractile forces being applied across a smaller surface area, motor control differences, i.e. the breathing centers not being attuned to inhibitory activity, and mechanical factors. The surface tension of the lung also plays a significant role in the larger forces required to inflate the lung during the first inspiration. Indeed, animal studies have shown that without surfactant, the volume inhaled with a first breath pressure generation of ~35 cmH₂0 is only ~10% of neonates with surfactant [375].

The first expiratory phase is often longer in duration than the later steady-state neonatal breathing and often involves periodic opening and closing of the upper respiratory tract, in order to promote fluid clearance from the lungs [340, 461]. There is a remarkable amount of air left in the lung following the first expiratory phase, which can be ~10–20% of the equivalent functional reserve capacity of a newborn a few days old [57, 340, 452, 461, 567, 673]. This air retention is primarily due to the formation of foam within the airways. In addition to the surface tension-overcoming properties of surfactant, it provides for the formation of foam within the newborn lung and allows for effective air retention in the newborn lung. The importance of surfactant is underscored by the observations that immature lungs, with less surfactant production, retain less air and have less foam than mature lungs [309, 370, 375, 569, 570].

The progressive establishment of a steady respiratory pattern occurs within a few hours after birth and requires three main adaptations; the clearing of pulmonary fluids, the matching of lung and respiratory mechanical properties to ventilation and the establishment of a functional reserve capacity. The clearing of the airways of pulmonary fluids is likely due to intermittent positive airway pressures (increased P_th) and fluid resorption [459]. Progressive increases in the compliance of the lungs, and progressive decreases in the resistance of the lungs and respiratory system occur concomitantly with the clearance of pulmonary fluids [339, 460]. Therefore, mechanical changes in the lung in the immediate hours after birth cause the pressures required to inflate the lung to steadily decrease [132, 308]. These changes allow the establishment of a functional reserve capacity [352, 452], ventilation-perfusion matching [518] and the normalization of blood gas partial pressures [483].

In the context of development, negative P_th development is inarguably essential for surviving the transition from the liquid-filled lung to the air-breathing scenario. If the principles of symmorphosis are to hold, than the more immature mammalian neonates, such as those found in marsupials with incredibly low basal metabolic rates, have extremely low requirement for P_di pressure generation and ventilatory capacity when compared to the more precocial mammalians. Importantly, the marsupials range from ~3–850 mg at birth (or hatching) [249, 662] are not uniformly all tiny species by at maturity, with further pouch gestation lasting up to ~7–11 months postpartum for larger wallaby and kangaroo species [662]. In newborn marsupials, the upper airway musculature (required for effective suckling behaviors [183, 184, 337]) and rib structures are well developed, with intercostal muscles innervated by motor neurons [304, 305]. However, the DIAm is a thin sheet and plays little, if any, role in the generation of negative P_th for respiration [304, 305], there is an impaired/lack of descending respiratory drive [197, 627], reduced chemosensitivity [628] and the lungs are similarly immature when compared to other mammals, with the extent of compartmentalization and vascularization ranging from canalicular to rudimentary [197, 408, 627]. Consequently, these muscular and alveolar deficiencies necessitate alternative modes of gaseous exchange, particularly via transcutaneous mechanisms [197, 198, 408, 462, 627]. Despite these developmental constraints, the athletic capacity of many marsupials scales with those of placental mammals (though there is large variability in this regard), including large tidal volumes [115, 135–137, 467, 468], providing further evidence for the overall utility of the symmorphosis concept, with markedly low neonatal marsupial ventilation requirements matched to functional capacity prior to exiting the pouch.

Conclusion

In this comprehensive review of the evolution of the DIAm, we have adopted the conceptual framework of symmorphosis by which the structural properties of the DIAm were matched to functional demands. In this context, the evolution of the DIAm should be viewed in relation to its two major functions: serving as a partition between the thoracic and abdominal cavities, and serving as a muscular pump to generate a negative P_th and a positive P_ab. These two functions are not exclusive but interrelated. In evolution, partitioning of the coelemic cavity into separate thoracic and abdominal cavities promoted the efficacy of aspiration breathing. Before partitioning the negative intra-coelemic pressure required to ventilate the lung caused incursion of the thoracic space by abdominal viscera. The impingement of viscera impeded lung expansion; thus the primordial DIAm membrane/partition promoted lung expansion and increased ventilation. Of course, improving respiratory efficiency by partitioning the abdominal and visceral compartments is not the only solution Nature has devised. An alternative is the development of the avian unidirectional air sac system.

Muscularization of the DIAm per se is unique to mammals. There were major evolutionary advantages in muscularizing the DIAm in order to provide an active pump for improved pressure generation in both intra-thoracic and intra-abdominal cavities. We often think of the DIAm as the major inspiratory muscle for ventilating the lung. This is certainly the case, and improved lung ventilation provided a major evolutionary advantage for mammals, especially considering the coincident evolution of the lung subdivision into alveoli to increase the surface area for gas exchange. However, even under extreme conditions, less than half of the force generating capacity of the DIAm is used to accomplish ventilatory behaviors. There is less consideration of the fact that the DIAm comprises the superior boundary of the abdominal cavity. Thus, DIAm contraction increases P_ab. Indeed, near maximum contraction of the DIAm occurs during straining behaviors that included defecation, vomiting, coughing sneezing and parturition. As with other skeletal muscles, contraction of the DIAm is under neural control. The motor unit, comprising a motor neuron and the muscle fibers it innervates is the fundamental unit of neuromotor control. Two groups of motor units are present in the DIAm to accomplish the range of motor functions involving P_th and P_ab generation. Fatigue resistant type S and FR motor units are well designed (with a high capacity for oxidative phosphorylation) to accomplish breathing behaviors that require lower forces but are repetitive and have a long duty cycle. In contrast, type FInt and FF motor units generate greater forces but are susceptible to fatigue. These motor units are well designed to accomplish higher force but more infrequent straining motor behaviors associated with increased P_ab generation.