Lawrence C Rome. American Scientist. Volume 85, Issue 4. Jul/Aug 1997.

Animals perform a tremendous variety of motor tasks. These include explosive movements that last just a few tens of milliseconds, swimming hundreds of miles and producing sounds at frequencies of several hundred hertz. These activities-each requiring very different outputs from muscles-cannot be performed by a single type of muscle. Instead, a given activity often requires a unique muscular design. For example, a fish’s swimming muscle cannot contract and relax quickly enough to produce highfrequency sounds, and sound-producing muscle generates too little force and requires too much energy to efficiently power swimming. As such, you might expect fundamental structural and physiological differences between one type of muscle and another.

Remarkably, microscopic examination reveals the same basic structure in all vertebrate skeletal muscles. Furthermore, physiological measurements show that these muscles operate by the same basic contractile mechanisms. Nevertheless, advanced techniques do reveal many subtle but crucial differences between muscles.

For the past 15 years, members of my laboratory and I have studied how different muscular systems produce unique motor outputs. In search of general design principles, one might begin with a system that produces a relatively straightforward function, such as jumping by a frog. For example, ranid frogsthe so-called true frogs-start from a crouched position and, in about 100 milliseconds, take off with a high-energy jump. This requires the muscle system to generate high mechanical power. In examining this system, as well as others, my colleagues and I began with a single question: What features make a muscular system well-adapted to a specific function?

Muscle Mechanics

Before moving to systems, let us begin with a single muscle. A muscle resembles a linear motor that can shorten actively, but it requires external forcessuch as a force from an antagonistic muscle-to lengthen. Like other motors, a muscle’s working characteristics depend on a particular set of features, many of which arise from a muscle’s molecular structure.

Starting at the macroscopic level, you probably think of a “muscle” as a structure like your biceps. Such a structure, however, consists of many muscle cells-called fibers-lying side by side. Each fiber in turn is made up of many myofibrils. A cylindrical myofibril consists of thousands of microscopic units called sarcomeres-each being about 2 micrometers long-that are connected end-toend. The components of a sarcomere include two types of myofilaments, which are made from proteins called actin and myosin. The thick filaments, which are made of myosin, form projections called crossbridges that generate force by ratcheting along the thin filaments, which are made of actin. These filaments interdigitate and pull past one another-creating the linear motor-like action that shortens the sarcomeres.

The force generated by a muscle depends on the number of force-generating crossbridges, which is a function of the overlap between the thick and thin filaments. The amount of overlap, and accordingly force, depends on the length of the sarcomeres: Long sarcomeres allow less overlap, and short ones allow more. This can be displayed by plotting the relative tension produced by a muscle versus the length of its sarcomeres, which produces a so-called sarcomere length-tension curve. Each muscle produces maximum force at optimal overlap, and the force declines to zero as sarcomere length becomes shorter or longer than the optimal value.

Does an animal’s muscular system operate in a way that maintains optimal overlap between the filaments? Despite many years of studying the structure and physiology of muscle cells, few investigators have examined how muscles actually operate in a moving animal. My colleagues and I have developed some techniques to explore this crucial aspect of muscular design.

For example, my graduate student Gordon Lutz and I studied a frog’s semimembranosus-a muscle that extends the hip. Using a high-speed motion-picture camera, we filmed a frog’s jump at 200 frames per second and then analyzed the film to determine the changes in the hip- and knee-joint angles during the movement. In separate experiments, we rotated a frog’s leg through those angles and measured the changes in length that the semimembranosus experienced during jumping. By using those lengths and a combination of physiological and microscopic methods, we determined the changes in the length of sarcomeres when a frog jumps.

We found that during a jump the length of the sarcomeres of a ranid frog’s semimembranosus shorten from 2.34 to 1.83 micrometers-points that span the plateau, and a bit beyond it, of the sarcomere length-tension curve. Despite operating off the plateau (where maximum force cannot be generated) during briefs portions of a jump, the frog’s semimembranosus still generates at least 90 percent of its maximum possible force throughout shortening, and that is very nearly optimal.

Perhaps surprisingly, operating beyond the plateau of the length-tension curve actually allows a frog to jump farther than if the muscle operated solely on the plateau. The distance a frog jumps depends on the total amount of mechanical work done by the muscle at takeoff, and work is the product of the force produced by a muscle and its change in length. The large change in the semimembranosus’s sarcomere length, which requires going beyond the edges of the sarcomere length-tension plateau, more than compensates for the slight decrement in muscle force during some parts of the muscle’s shortening. So the muscle generates more work, which leads to longer jumps. Furthermore, we found that-given the large change in the length of the sarcomere-this muscle operates near the theoretically perfect place on its sarcomere length-tension curve. In fact, increasing or decreasing the initial length of the sarcomere would result in less work being produced during a jump.

The velocity at which a muscle shortens also markedly affects its performance. Like a rotary motor in which torque and horsepower vary with motor speed, or revolutions per minute, the force and power produced by a muscle depend on its velocity of shortening (often given in units of muscle lengths per second). The mechanical power generated by a muscle is the product of the force produced and shortening velocity, and a muscle produces peak power at some intermediate velocity: about one-third of its maximum velocity of shortening. In our experiments on frogs, we found that the semimembranosus has a maximum shortening velocity of about 10.35 muscle lengths per second, and that it generates maximum power at a shortening velocity of about 3.44 muscle lengths per second. During a jump, this frog’s semimembranosus shortens at 3.43 muscle lengths per second-almost exactly where it generates maximum power.

As you might expect, a frog’s jumping system depends on many factors: joint configuration, structural elements (including bones and ligaments), the animal’s mass and many characteristics of the contracting muscles. Our findings of virtually optimal sarcomere length and muscle-shortening velocity-both of which can be called “optimal” only in relation to the rest of the system-indicate that the many elements in a frog’s jumping system are matched, making a frog a well-designed jumping machine when it takes its “crouched” position.

Activating Muscle

The above descriptions of muscle-generated force and power assume a maximum activation of the crossbridges between actin and myosin filaments. Such activation arises from a series of molecular steps driven by ionic calcium (Ca2+). In the absence of calcium, actin filaments are inhibited, thereby preventing myosin’s crossbridges from attaching and keeping the muscle relaxed. To “turn on” a muscle, calcium is released from an organelle called the sarcoplasmic reticulum, which surrounds the bundles of myofilaments. In the myoplasm, or space surrounding the myofilaments, the calcium binds to a protein called troponin, which is connected to the thin filament. This binding starts a process that removes the inhibition from the actin filaments and allows the myosin crossbridges to attach and generate force. Turning off a muscle requires the reverse process: Calcium gets pumped back into the sarcoplasmic reticulum, which reduces the calcium levels near the myofilaments. Then, calcium unbinds from troponin, and the myosin crossbridges detach from the actin. Perhaps surprisingly, turning off a muscle requires energy, because the calcium must be pumped up a concentration gradient into the sarcoplasmic reticulum.

If a frogs semimembranosus starts to shorten before being fully activated, it will generate less-than-maximum power. Achieving maximum power requires a combination of rapid activationwhich depends on the kinetics of calcium release, calcium’s binding to troponin and crossbridge attachment-and delayed movement by the frog-which depends on its mass and joint configuration. The significant question is: Does a frogs semimembranosus become maximally activated before leg extension begins and, thereby, generate maximum power during a jump?

To find out, we began with our measurements of the semimembranosus’ length changes during a jump. We also determined the nervous stimulation of the muscle during a jump by recording electromyograms-rather complex electrical patterns that indicate when a muscle is being stimulated by neurons. Next, we applied these in vivo length changes and stimulation pattern to an isolated muscle, in which we could also record the force that it produced. We found that, under conditions that mimic jumping, the isolated muscle produced the maximum possible force for a given velocity of shortening. In other words, these findings suggest that the semimembranosus is maximally activated during jumping, which requires matching between the molecular components that set the kinetics of activation and the overall biomechanics of jumping.

Cycles of Swimming

Although jumping by frogs provides a relatively simple system to study, we can learn even more about the design of muscular systems by investigating swimming in fish, because they produce a wide range of movements. For example, a carp (Cyprinus carpio) swimming steadily at 20 centimeters per second produces very little curvature in its backbone during a stroke, but during an escape response its backbone bends dramatically. In addition, one tail beat takes about 400 milliseconds during steady swimming, but during escape swimming the backbone goes from straight to an extreme curve in as little as 25 milliseconds. How can one type of muscle produce both behaviors when working optimally? It cannot. These two behaviors require two different groups of muscle fibers.

Different vertebrate skeletal muscles have different properties. In general, these muscles can be divided in two groups: red and white muscles. The red muscles contract rather slowly, but they fatigue slowly, too. These muscles look red because they contain high concentrations of myoglobin, an oxygen-storing protein, and mitochondria, which utilize oxygen and generate the energy-suppling compound (ATP) that powers contractions. (Red muscles make up “dark meat” in, say, chicken or fish.) White muscles contract rapidly, fatigue rapidly and lack high concentrations of myoglobin and mitochondria. (As you probably expect, white muscles make up “white meat,” such as a chicken breast and most of a fish.) Electromyographic recordings show that a carp uses red muscle alone during steady swimming and that white muscle gets recruited during higher-speed swimming and escapes.

To begin studying this system, my colleagues and I examined changes in sarcomere length during swimming. For a carp swimming at slow speeds, we found that red-muscle sarcomeres undergo cyclical length changes between 2.25 and 1.89 micrometers-lengths that produce the most force for these musdes. Although these muscles work almost perfectly for slow, steady swimming, they cannot provide extreme bending of the backbone. The red muscles lie parallel to a carp’s backbone, and they would need to shorten to a sarcomere length of about 1.4 micrometers for the backbone to bend as much as it does during an escape response. At that sarcomere length, the muscles produce very little force, and such excessive shortening can even damage the fibers.

The white muscles run in a helical orientation with respect to a carp’s backbone. This arrangement endows the white muscle with a fourfold higher gear ratio. That is, white muscle can produce a given backbone curvature with only one-quarter of the change in sarcomere length that would be required by the red muscles. You can think of the white muscles as being like a high gear on a bicycle, where one revolution of the pedals causes multiple turns of the rear wheel. In contrast, the red muscle works more like a low gear, where one pedal rotation may generate just one revolution of the wheel.

With its helical arrangement, white muscle can produce a carp’s escape response by shortening only to an average sarcomere length of about 1.82 micrometers, which produces at least 94 percent of the maximum force for those muscles. So this two-muscle system leads to nearly optimal force production despite the different demands of a carp’s two swimming behaviors.

A carp’s red and white muscles also shorten at different rates. The red muscles have a maximum velocity of shortening of 4.65 muscle lengths per second, and the white muscles can shorten 2.5 times faster, with a maximum velocity of shortening of 12.8 muscle lengths per second. During steady swimming the red muscle is used over shortening-velocity ranges of about 0.7 to 1.5 muscle lengths per second, which is where maximum power is generated. To power the escape response, however, the red muscle would have to shorten at 20 muscle lengths per second, which it clearly cannot do because that is four times faster than its maximum shortening velocity. Nevertheless, the fourfold higher gear ratio of the white muscle allows it to power the escape response by shortening at only 5 muscle lengths per second, which happens to be in the velocity range at which it produces its maximum power.

These results reveal that the red and white muscles form a two-gear system that powers very different movements. The red muscle powers slow movements, the white muscle powers fast movements, and both work at appropriate shortening velocities. The white muscle’s fourfold-higher gear ratio and 2.5 times-faster maximum shortening velocity allow it to produce 10-foldfaster movements than the red muscle. In fact, the rapid movement required for the escape response could not be generated at all without this appropriate combination of the gear ratio and shortening velocity.

Swimming Efficiently

Despite the similarities in optimal overlap and shortening velocity of myofilaments in a frogs jumping muscles and a fish’s swimming muscles, these muscular systems do differ in terms of their use and design. In a one-shot jump, a frog’s muscles have plenty of time to relax as the frog soars through the air, making the speed of relaxation essentially meaningless. During fish swimming, however, a muscle must relax after shortening. If an actively shortening muscle has to fight against muscle on the other side of the fish, which is being lengthened, then the work generated by the shortening muscle will be lost as heat, rather than being used to propel the fish. Since active muscle generates very high force when it lengthens, it is essential that a muscle be nearly fully relaxed by the end of shortening, so that it can be relengthened without resisting the shortening of opposing muscles.

The so-called workloop-a graph of the force produced versus muscle length-reveals how a muscle produces work during a cycle of shortening and relengthening. Beginning at the muscle’s maximum length, the muscle shortens and the area under the graph’s shortening phase represents the amount of positive work performed by the muscle. When the muscle reaches its minimum length and then is relengthened, the area under the bottom lengthening curve represents the amount of negative work being done. So the area contained inside the loop represents the net work performed-that is, the work that can be used for propelling a fish.

One might expect that a muscle would function most effectively by making its fibers fully active during shortening and fully relaxed during relengthening, because that would maximize the net work. However, there is a trade-off: A muscle capable of an instant switch from activated to relaxed requires a lot of energy, because a fastrelaxing muscle uses more energy for fast pumping of ionic calcium back into the sarcoplasmic reticulum and for fast cycling of crossbridges than does a slow-relaxing muscle. The question is: Are fish designed with fast-relaxing but energetically costly muscle or with slow-relaxing but energetically economical muscle?

Answering that question requires a determination of the nature of workloops in swimming fish. To do just that, my graduate student Doug Swank and I selected a fish called a scup (Stenotomus chrysops), which performs seasonal migrations of several hundred miles. In large part, we repeated the same strategy that was used on a jumping frog. First we determined both the changes in sarcomere length and the pattern of nervous stimulation to a scup’s red muscles during swimming, and then we used isolated red muscle bundles to measure the force that they produce during such cycles of length change and stimulation. We completed these procedures at four different points along a scup’s length.

Our experiments indicated that the nervous stimulation to a red muscle during swimming starts during the last half of lengthening and ceases just after the beginning of shortening. In other words, the muscles start activating during their relengthening stage and begin deactivating during the shortening stage or so-called power stroke. This ensures that the muscle is nearly relaxed prior to being relengthened. As a result, the muscles do not produce as much power as faster muscles that could remain maximally activated during shortening and then relax instantly at the end of shortening, but they probably use far less energy. Therefore, these muscles seem to be designed to generate power efficiently, rather than simply generating maximum power.

Our work on jumping by frogs and swimming by fish indicates fundamental differences in the design of these muscular systems. A frog’s jumping muscles generate maximum power during a single shortening stroke, and the force remains high throughout shortening. By contrast, the force generated by these fish swimming muscles declines dramatically during shortening, because of early cessation of the stimulus and subtle molecular processes. This is necessary, however, if a muscle is to be nearly relaxed prior to being relengtheneda requirement for power generation during cyclical activity. One common feature of these frog and fish muscles is that they both relax relatively slowly, but that cannot be said of all muscles.

Sound-Producing Muscles

A male toadfish (Opsanus tau) produces a boat-whistle mating call to attract females to his nest. He can call for many hours, producing the sound from 10 to 12 times each minute by contracting the muscles that surround a gas-filled swimbladder. The muscles contract as fast as 200 times per second, making them the fastest known vertebrate muscles. If these sonic muscles were replaced with typical slow-relaxing locomotory muscles, like the ones used by frogs to jump and by fish to swim, they would not have time to relax between stimuli, and thus would generate a constant force that would simply compress the bladder and prevent it from vibrating. Sonic muscles, then, must be specifically adapted to turn on and off rapidly To better understand what allows these muscles to turn on and off so rapidly, my postdoctoral associate Doug Syme, my colleagues at the University of Pennsylvania Stephen Hollingworth and Stephen Baylor, and I examined a variety of toadfish muscle fibers. We compared the contraction-relaxation capabilities of these muscles during a twitch, which is one contraction-relaxation cycle. For instance, we measured the so-called twitch half-width, where the half-width is the time it takes for the force to rise from 50 percent of its maximum magnitude, pass through the maximum level and return to the 50-percent level. Red muscle, which toadfish use for slow swimming, has a twitch halfwidth of about 500 milliseconds, compared to approximately 200 milliseconds for white muscle, which a toadfish uses for fast bursts of swimming, and about 10 milliseconds for the sonic muscle. So the sonic muscles contract and relax 50 times faster than the red muscle, and that requires fast deployment and removal of calcium, and quick attaching and detaching of crossbridges.

We measured the change in myoplasmic calcium during contractions by injecting muscle cells with a calcium-sensitive dye, furaptra, which changes its fluorescence when it binds to calcium. This allows us to determine the time course of calcium release and reuptake. Our results showed that sonic muscles produce the fastest calcium release and reuptake ever measured; its halfwidth is about 3.4 milliseconds at 16 degrees Celsius and just 1.5 milliseconds at 25 degrees Celsius. As one might expect, the white and red muscles have slower calcium cycles. A high density of calcium pumps in the sonic muscles provides them with their rapid calcium reuptake.

Working with Yale Goldman and Chris Cooke, two other colleagues at the University of Pennsylvania, we also measured the rate of detachment of crossbridges in the sonic muscle. It, too, was exceptionally fast, almost 100 times faster than in rabbit fast fibers (the beststudied fiber). This high detachment speed appears to result from a molecular modification of the myosin.

Another sound-producing musclethe shaker muscle of rattlesnakes in the genus Crotalus-has some characteristics that resemble those found in toadfish sonic muscles. (As an aside, these snakes shake their rattles as a warning to other animals, not to attract mates, as some people believe.) At 15 degrees Celsius, calcium can be released and recaptured in the shaker fibers with a halfwidth of just 4 or 5 milliseconds. Although that event is just 1 or 2 milliseconds slower than in a toadfish’s sonic muscles, the twitch halfwidth of the shaker muscle is about 25 milliseconds, or 2.5 times longer than in sonic muscle. This slower twitch likely reflects the effects of slower crossbridge detachment and, perhaps, slower unbinding of calcium from troponin. Despite the speed of calcium release and reuptake in shaker muscles, lack of speed in any other step makes the muscle a runner-up among the fastest of the fast. The limiting steps, however, speed up at higher temperatures, allowing the snake to rattle at an impressive 90 cycles per second at 35 degrees Celsius.

In the Future

We are beginning to understand how muscle is designed at the molecular level for different functions. Although a tremendous amount has been learned about muscle design over the past decade, many questions remain.

Some of these questions will be answered, in all likelihood, by improved technical capabilities. For instance, improving biophysical techniques-such as calcium-sensitive dyes-are revealing more details about the molecular design of muscles, which will enhance our ability to integrate from molecular components to whole animal locomotion. I also expect to see an increased use of modeling. Software is already available for building virtual animals with the appropriate anatomy (limb and muscle dimensions) and mass distribution. This software enables us to predict whole-animal movements from mechanical properties of muscle, or to see how molecular modifications might affect locomotory performance. Finally, genetic engineering may allow us to test theories of muscle design by altering specific components of a muscular system and studying the results. In the near future, I expect that we will know much more about how muscles are designed and how the designs evolved.