{
  "id": "myosin",
  "title": "Myosin",
  "lang": "en",
  "source_lang": "en",
  "summary": "Myosins are ATP-powered molecular motors that interact with actin to generate force, drive muscle contraction, transport cargo, and organize cells.",
  "category": "science",
  "subcategory": "biology",
  "type": "concept",
  "aliases": [
    "Myosin motor proteins",
    "Myosin superfamily"
  ],
  "generated_at": "2026-11-03",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/myosin/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/myosin/",
      "json": "https://aiwiki.page/zh/myosin.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "Myosin is a superfamily of proteins that convert the chemical energy of adenosine triphosphate (ATP) into mechanical force through interactions with actin filaments. Myosins are best known as the motors responsible for muscle contraction, but their activities extend throughout the eukaryotic world. They help divide cells, move intracellular cargo, shape membranes, maintain tissue tension, and support specialized structures involved in hearing, vision, and nutrient absorption.\n\nThe defining feature of a myosin is a conserved motor domain that binds actin and couples ATP turnover to structural changes. Other regions vary extensively, allowing different myosins to assemble into contractile filaments, walk along actin, attach membranes to the cytoskeleton, or remain under tension as molecular anchors. Consequently, myosin is not a single muscle protein but a diverse family of related molecular machines.\n\nUnderstanding myosin requires connecting several scales: atomic rearrangements within its catalytic site, nanometre-scale steps along actin, collective force generation by molecular ensembles, and the movement of whole cells and tissues. The same basic motor architecture underlies both a heartbeat and the delivery of a vesicle to the cell surface, although its regulation and mechanical performance differ greatly between these settings."
    },
    {
      "heading": "Diversity, classification, and evolutionary distribution",
      "text": "Myosins are classified primarily through the evolutionary relationships of their motor domains, with tail organization and cellular function providing additional distinctions. Classes are conventionally numbered, often with Roman numerals: myosin I, myosin II, myosin V, and so forth. The numbering is a historical naming system, not a ranking of complexity or a sequence of evolutionary appearance. Individual organisms may possess several members of one class and lack many others. Human cells express roughly 40 myosin heavy-chain genes distributed among about a dozen classes.\n\nMyosin II comprises the familiar filament-forming motors of skeletal, cardiac, and smooth muscle, as well as nonmuscle myosin II. These proteins typically have two motor heads and a long coiled-coil tail. The term “conventional myosin” historically refers to this class. Other classes are collectively called unconventional myosins, but that designation describes their difference from the originally characterized muscle motor rather than an exceptional or marginal biological role. Myosin I proteins, for example, are widespread membrane-associated motors, while myosin V proteins are important intracellular cargo transporters.\n\nThe diversity of myosins is especially apparent when animals are compared with fungi, plants, and protists. Budding yeast uses class I motors in membrane internalization, a class II motor in cell division, and class V motors for polarized transport. Land plants lack the conventional animal muscle system but possess class XI motors that drive rapid organelle movement and cytoplasmic streaming. Animal cells contain specialized classes associated with structures such as sensory hair bundles and intestinal microvilli. Some eukaryotic lineages have lost myosins, showing that these motors are widespread but not universal.\n\nComparative phylogenetic analysis indicates that substantial myosin diversification occurred early in eukaryotic evolution. However, the exact complement present in the last common ancestor of living eukaryotes remains dependent on evolutionary models and taxon sampling. Myosin motor domains share structural features with other nucleotide-dependent motor systems, particularly kinesins, but they operate on different cytoskeletal tracks. Kinesin and dynein generally move on microtubules, whereas myosins use actin.\n\nClassification also has practical limits. Similar tail domains can appear in different evolutionary contexts, and the same class can contain motors with markedly different kinetics or regulatory mechanisms. Alternative products of a single gene may differ in cargo binding, tissue distribution, or mechanical behavior. A class designation therefore identifies an evolutionary family, but does not by itself determine whether a motor is a fast transporter, a tension-bearing anchor, or a component of a contractile filament."
    },
    {
      "heading": "Molecular architecture and assembly",
      "text": "Most myosin heavy chains contain three broad functional regions: a motor domain, a neck or lever arm, and a tail. The motor domain is usually near the amino terminus and contains approximately 700–800 amino-acid residues. It includes the nucleotide-binding pocket, surfaces that contact actin, and internal structural elements that transmit changes between those sites. Although its amino-acid sequence varies among classes, its overall three-dimensional organization is strongly conserved.\n\nThe nucleotide pocket contains a phosphate-binding loop and conserved switch regions that respond to ATP, ADP, and inorganic phosphate. The actin-binding face includes several loops and subdomains whose relative positions change during the motor cycle. A cleft between major portions of the motor changes configuration as actin affinity changes. Near the opposite end of the domain, the relay region and converter transmit structural rearrangements toward the lever arm. This arrangement allows a relatively small movement within the motor core to produce a larger displacement at the distal end of the neck.\n\nThe neck commonly contains one or more IQ motifs, short sequence elements that bind light chains or calmodulin. These associated proteins stabilize an extended helical region, making it suitable for transmitting force. Muscle myosin II normally carries an essential light chain and a regulatory light chain on each heavy chain. Myosin V has a longer neck, typically containing six IQ motifs per heavy chain. Neck length is an important contributor to movement geometry, but step size also depends on motor coordination, flexibility, and the structure of the actin track.\n\nTail regions provide much of the superfamily’s functional diversity. In myosin II, two heavy chains associate through an extended coiled coil. Their tails assemble into bipolar filaments with motor heads projecting from opposite ends, enabling the filament to pull oppositely oriented actin filaments toward one another. A skeletal muscle myosin II molecule is approximately 500 kilodaltons in total mass, including two heavy chains and four light chains, and its extended tail is roughly 150 nanometres long. These dimensions describe a representative architecture rather than a universal myosin design.\n\nOther tails contain lipid-binding regions, cargo-binding domains, or modules that interact with adaptor proteins. Many class I myosins have a single motor and bind the cell membrane through their tails. Myosin V generally forms a two-headed dimer with globular cargo-binding domains. Myosin VI places its motor near the amino terminus but contains a distinctive insert that redirects its lever arm, contributing to its unusual direction of travel. Some myosins fold into inhibited conformations in which tail regions contact the heads, linking molecular architecture directly to regulation."
    },
    {
      "heading": "The actin–ATPase cycle",
      "text": "Myosin is an enzyme whose ATPase activity is strongly influenced by actin. Its mechanical operation depends not simply on ATP breakdown, but on coordinating nucleotide binding, hydrolysis, product release, and changing affinity for the filament. The familiar cross-bridge cycle is a useful framework, although individual myosin classes differ substantially in which reaction is slowest and how strongly mechanical load alters each transition.\n\nIn a nucleotide-free state, a myosin head binds actin tightly. This strong-binding state is often called rigor. ATP binding to the motor reduces its affinity for actin and promotes detachment. Once detached, ATP hydrolysis produces an ADP–inorganic-phosphate state in which the products remain bound to the motor. Structural rearrangement during this phase primes the lever arm for a subsequent force-producing interaction. ATP hydrolysis therefore does not immediately produce the main working stroke against actin.\n\nThe primed motor encounters another actin-binding site, initially through comparatively weak interactions. Progression toward strong binding is coupled to rearrangements in the actin-binding interface, phosphate release, and the force-generating structural transition. These events are tightly linked, and their precise temporal ordering depends on the experimental conditions and the motor being studied. The working stroke moves the lever arm relative to the actin-bound motor domain. ADP is then released, returning the motor toward a nucleotide-free strong-binding state until another ATP binds.\n\nA schematic representation is:\n\n\\[A\\!\\cdot\\!M + ATP \\rightleftharpoons A + M\\!\\cdot\\!ATP \\rightleftharpoons A + M\\!\\cdot\\!ADP\\!\\cdot\\!P_i \\rightleftharpoons A\\!\\cdot\\!M\\!\\cdot\\!ADP + P_i \\rightleftharpoons A\\!\\cdot\\!M + ADP,\\]\n\nwhere \\(A\\) represents actin, \\(M\\) myosin, and \\(P_i\\) inorganic phosphate. This notation compresses several biochemical and structural intermediates into a small number of states; it is not a complete kinetic model.\n\nThe cycle is powered by the difference in Gibbs free energy between ATP and its hydrolysis products under cellular conditions. For a complete cycle tightly coupled to one ATP molecule, the available chemical free-energy decrease places an upper bound on the net mechanical work. Actual work depends on load, displacement, coupling efficiency, and dissipative losses. Product concentrations, especially ADP and phosphate, can alter cycle behavior.\n\nThis mechanism explains why ATP is required for relaxation as well as contraction. After death, ATP depletion prevents normal detachment of strongly bound actomyosin cross-bridges, contributing to rigor mortis. In living muscle, ATP also powers calcium handling and other processes needed to restore the resting state. Myosin’s ATP requirement therefore cannot be understood solely as fuel for an isolated pulling stroke."
    },
    {
      "heading": "Directionality, processivity, and force generation",
      "text": "An actin filament has structural polarity: its subunits are oriented consistently, creating distinct barbed and pointed ends. Most characterized myosins move toward the barbed, or plus, end. Myosin VI is the best-established exception, moving toward the pointed, or minus, end. Its reverse direction arises from changes in how motor-domain motions are transmitted through the converter and lever arm, rather than from a different source of chemical energy.\n\nA central kinetic property is the duty ratio, the fraction of the ATPase cycle spent strongly attached to actin. Many fast skeletal muscle myosin II motors have relatively low duty ratios. A single head therefore spends much of its cycle detached or weakly attached, but large ensembles maintain continuous force because some heads remain engaged while others recover. Processive transport motors such as myosin V have high duty ratios and coordinated heads, allowing one molecule to take many successive steps without fully leaving the filament.\n\nDimeric myosin V typically advances its cargo-binding region in steps of about 36 nanometres, close to the long-pitch helical repeat of actin. Its two heads move predominantly in a hand-over-hand pattern. During an individual stepping event, a trailing head can move approximately twice that distance as it passes the leading head. Internal strain helps coordinate nucleotide transitions, making premature detachment of both heads less likely. These dimensions differ from the roughly 5–10-nanometre working strokes commonly associated with individual muscle myosin II interactions.\n\nMechanical load changes the rates of biochemical transitions. Resisting force can slow ADP release or stabilize particular attached states, depending on the motor. Some myosins consequently become better anchors under tension, whereas others are adapted for rapid movement under low load. A motor’s stall force—the opposing force at which its average forward velocity becomes zero—is generally on the piconewton scale, but there is no single stall force applicable to all myosins. Measurements depend on motor identity, geometry, nucleotide conditions, and the number of active heads.\n\nThe relationship between movement and force can be expressed through mechanical work:\n\n\\[W=\\int F\\,dx.\\]\n\nFor an approximately constant opposing load, this becomes \\(W \\approx Fd\\), where \\(d\\) is displacement. A force of 3 piconewtons acting through 10 nanometres corresponds to \\(3\\times10^{-20}\\) joules. Such estimates place molecular events on an energy scale comparable with the usable chemical energy from ATP hydrolysis.\n\nMovement is not purely a rigid lever swinging through a fixed trajectory. Brownian motion, elastic strain, filament geometry, and biased binding all contribute. The lever arm constrains and amplifies structural change, while thermal motion allows detached heads to search for accessible binding sites. Myosin mechanics therefore combines directed conformational changes with stochastic molecular motion."
    },
    {
      "heading": "Myosin in skeletal and cardiac muscle",
      "text": "In striated muscle, myosin II forms the thick filaments of the sarcomere, the repeating contractile unit between adjacent Z-discs. Actin-containing thin filaments extend inward from the Z-discs, while thick filaments occupy the central region. A typical vertebrate thick filament is approximately 1.6 micrometres long. Myosin heads project from its backbone and interact with overlapping thin filaments. During shortening, the filaments slide past one another; neither actin nor myosin filaments need to shorten substantially.\n\nThe bipolar organization of a thick filament is essential. Myosin heads on its two halves have opposite overall orientations, so their movement draws thin filaments from both ends toward the sarcomere’s center. The central bare zone contains relatively few projecting heads. Proteins including titin and myosin-binding protein C help organize the contractile system and influence its mechanical behavior. The orderly alignment of many sarcomeres allows molecular forces to sum along and across muscle fibers.\n\nActivation in skeletal and cardiac muscle is controlled mainly through the thin filament. Calcium binds troponin, causing rearrangements that alter the position of tropomyosin on actin and increase access for strong myosin binding. In skeletal muscle, an action potential triggers calcium release from the sarcoplasmic reticulum through coupling between membrane voltage sensors and calcium-release channels. In cardiac muscle, calcium entry through surface-membrane channels triggers further calcium release from intracellular stores. The final actomyosin interaction is similar, but the upstream control differs.\n\nMyosin isoforms contribute to differences in contraction speed, energetic cost, and fatigue resistance. Slow skeletal fibers generally express motors with slower ATP turnover and shortening kinetics than fast fibers. Human ventricular muscle predominantly contains β-cardiac myosin, encoded by MYH7, whereas MYH6 encodes α-cardiac myosin, which has a greater role in atrial tissue. Isoform distribution varies with species, developmental stage, and physiological condition; rodent ventricular patterns cannot simply be transferred to adult humans.\n\nMuscle force also depends on how many heads are available to bind actin. In relaxed thick filaments, heads can adopt folded, interacting arrangements that suppress activity. A biochemically defined super-relaxed state has particularly slow ATP turnover, while a disordered-relaxed population turns over ATP more rapidly. These biochemical populations are related to, but not strictly synonymous with, particular structural conformations. Changes in head availability provide a layer of regulation beyond calcium control of the thin filament.\n\nAt the tissue scale, myosin contributes to the force–velocity relationship: an actively shortening muscle generally produces less force at higher shortening speed. Under load, cross-bridge transitions and attachment lifetimes constrain movement. During lengthening contractions, attached cross-bridges and passive structural elements resist stretch, producing behavior that cannot be described by unloaded motor speed alone."
    },
    {
      "heading": "Smooth muscle, cell shape, and division",
      "text": "Smooth muscle uses myosin II but does not arrange its contractile machinery into the regular sarcomeres of skeletal and cardiac muscle. Actin filaments attach to dense bodies and membrane-associated dense plaques, while myosin filaments interact with the surrounding actin network. This organization permits contraction over a broad range of cell shapes and lengths. Smooth muscle supports functions such as regulation of blood-vessel diameter, movement through the gastrointestinal tract, and contraction of the urinary bladder.\n\nIts principal activation mechanism acts directly on myosin. An increase in intracellular calcium promotes formation of a calcium–calmodulin complex, which activates myosin light-chain kinase. This kinase phosphorylates the regulatory light chain, especially at serine 19, increasing actin-activated ATPase activity and promoting an active filament-associated state. Myosin light-chain phosphatase reverses the modification. Force therefore depends not only on calcium concentration but also on the balance between kinase and phosphatase activities.\n\nSignaling pathways can change the sensitivity of contraction to calcium. The RhoA–Rho-associated kinase pathway, for example, can inhibit myosin light-chain phosphatase and increase regulatory light-chain phosphorylation at a given calcium concentration. Sustained smooth-muscle force can also be maintained with relatively low ATP consumption. The classical latch-bridge model explains part of this economy through slowly detaching, dephosphorylated attached cross-bridges, although sustained force involves additional structural and regulatory processes.\n\nNonmuscle myosin II uses related machinery to generate tension outside muscle. Mammals have three major nonmuscle heavy-chain paralogs, encoded by MYH9, MYH10, and MYH14, commonly called myosin IIA, IIB, and IIC. These proteins differ in kinetics, expression, and localization. They assemble into bipolar minifilaments, often containing different paralogs, that contract actin networks. Regulatory light-chain phosphorylation promotes activity and assembly, while heavy-chain modifications and binding proteins provide further control.\n\nDuring cell division, myosin II participates in the actomyosin contractile ring that constricts the cleavage furrow in many animal cells and fungi. Positioning depends on signals coordinated with the mitotic apparatus, including localized Rho-family signaling. Ring constriction is not simply a miniature sarcomere shortening: filament organization, turnover, membrane attachment, and cortical flows all contribute. Final separation of the daughter cells requires additional membrane-remodeling machinery.\n\nIn migrating cells, myosin II generates traction, retracts the rear, and organizes stress fibers. At cell–cell junctions, it helps maintain and remodel tissue tension. The cortical actin–myosin network also controls cell stiffness and shape during development. These functions depend on the spatial distribution of contractility, not merely its total magnitude: increased tension at one junction can change neighboring cells’ geometry, while coordinated contraction across an epithelial sheet can bend or invaginate a tissue."
    },
    {
      "heading": "Cargo transport and specialized cellular structures",
      "text": "Unconventional myosins connect actin-based movement to intracellular organization. Myosin V is a major example: its tail binds cargo directly or through adaptor complexes, while its coordinated heads support processive transport. Different myosin V paralogs participate in the movement and positioning of secretory vesicles, pigment granules, and recycling compartments. Cargo engagement can also relieve autoinhibition, coupling motor activation to the presence of an appropriate transport target.\n\nMelanosome transport illustrates cooperation between motors and adaptors. In pigment cells, the small GTPase Rab27a and the adaptor melanophilin recruit myosin Va to melanosomes. Long-range movement can involve microtubule motors, while actin-associated myosin helps distribute and retain organelles in peripheral regions. This division of labor is more accurate than treating actin and microtubules as mutually exclusive transport systems. Cargo can switch tracks, engage several motors, or remain mechanically tethered.\n\nClass I myosins frequently couple membranes to actin. Their tails can bind acidic membrane lipids, while their motor domains engage cortical filaments. In yeast and many animal cells, class I motors contribute to endocytosis, where actin-generated forces help deform or internalize membrane. Other class I proteins regulate membrane tension or organize specialized surface projections. Depending on the motor and load, their role may resemble a dynamic tether more than a long-distance transporter.\n\nSensory cells use highly specialized myosins. In inner-ear hair cells, stereocilia are actin-filled projections arranged in mechanically sensitive bundles. Myosin VIIA and myosin XV contribute to the organization and maintenance of these structures, and defects in their genes can impair hearing. Myosin XV helps deliver proteins involved in stereocilia growth to their tips. Myosin III proteins also participate in stereocilia organization. Although myosin Ic has been extensively studied in relation to hair-cell adaptation, the molecular mechanisms of adaptation differ among preparations and cannot be reduced to a single universally established motor model.\n\nAt the intestinal brush border, myosin Ia contributes to interactions between microvillar actin bundles and the surrounding membrane. Myosin VI, moving toward actin pointed ends, participates in selected endocytic and membrane-trafficking pathways and can function as a force-bearing anchor. Its behavior depends strongly on oligomerization, cargo adaptors, and mechanical context. Myosin X localizes to filopodial tips and helps organize or transport components involved in these slender actin-rich protrusions.\n\nPlant myosin XI motors support particularly rapid organelle movement and cytoplasmic streaming. Such movement redistributes materials within large cells and influences interactions among organelles, the endoplasmic reticulum, and the actin network. These examples show that “cargo” may mean a discrete vesicle, an extended membrane system, or a protein assembly whose positioning alters the architecture of the cell itself."
    },
    {
      "heading": "Discovery and experimental investigation",
      "text": "The name myosin dates to Wilhelm Kühne’s work on muscle proteins in 1864. Early preparations did not distinguish all the molecular components now recognized in the contractile apparatus. A decisive advance came in the early 1940s through work by Albert Szent-Györgyi and colleagues, including Brunó Straub, which separated actin from myosin and clarified the ATP-dependent behavior of actomyosin. This established muscle contraction as a tractable problem in biochemistry, rather than an unexplained property of intact tissue.\n\nIn 1954, two independent studies—one by Andrew Huxley and Rolf Niedergerke, the other by Hugh Huxley and Jean Hanson—provided the foundation of the sliding filament theory. Their observations showed that contraction changes filament overlap rather than requiring major shortening of the filaments themselves. Subsequent cross-bridge models linked muscle mechanics to attachment and detachment of myosin heads. Studies of nonmuscle cells later expanded the subject beyond muscle; the identification of myosin I in Acanthamoeba in the 1970s helped establish unconventional myosins as a distinct field.\n\nStructural methods revealed how the motor might work. A 1993 structure of skeletal muscle myosin subfragment 1, reported by Ivan Rayment and colleagues, provided an atomic framework for the motor and associated light-chain-binding region. X-ray crystallography subsequently captured numerous nucleotide-dependent conformations. Cryo-electron microscopy made it possible to examine myosin bound to actin and to resolve increasingly detailed arrangements within thick filaments. Static structures, however, must be interpreted alongside kinetic measurements because a structural snapshot does not establish the order or rate of transitions.\n\nBiochemical assays measure ATP turnover, nucleotide binding, and product release. Rapid-mixing experiments can separate reaction steps occurring on millisecond timescales. In vitro motility assays attach myosin to a surface and observe fluorescent actin filaments gliding over it, or immobilize actin and track moving motors. These experiments reveal velocity, directionality, and processivity, but surface attachment, motor density, and assay geometry can affect the result.\n\nOptical tweezers use focused laser light to trap microscopic beads, allowing forces and displacements associated with individual motor interactions to be measured. Fluorescence localization and other single-molecule methods track stepping, while force-clamp experiments test how imposed load changes attachment lifetimes. These techniques established that myosin behavior is intrinsically stochastic and that averages can conceal distinct kinetic states.\n\nCellular approaches complement purified systems. Genetic disruption, targeted changes in motor or tail domains, fluorescent tagging, and selective inhibitors reveal where particular myosins act and which functions require ATPase activity. A persistent experimental challenge is connecting purified-motor properties to crowded, regulated cells, where actin architecture, cargo adaptors, phosphorylation, mechanical resistance, and multiple interacting motors operate simultaneously."
    },
    {
      "heading": "Human disease, pharmacology, and unresolved mechanisms",
      "text": "Changes in myosin genes can affect organs in ways that reflect the specialized functions of individual motors. Pathogenic variants in MYH7, encoding β-cardiac myosin heavy chain, are associated with inherited cardiomyopathies and some skeletal muscle disorders. MYH6 variants are also associated with cardiac disease. The consequences are not explained by a simple rule that every variant increases or decreases motor speed: altered head availability, actin interaction, ATP turnover, stability, and filament incorporation can each contribute.\n\nMYH9-related disease results from pathogenic variants in the gene encoding nonmuscle myosin IIA. It is characterized by large platelets and a reduced platelet count, with some affected individuals also developing kidney disease, hearing loss, or cataracts. The syndrome illustrates how one widely expressed motor can be required in several mechanically and structurally different tissues. Disease features depend on the particular variant and other biological factors.\n\nUnconventional myosins account for additional well-established genetic disorders. Biallelic MYO5A variants can cause Griscelli syndrome type 1, involving pigment dilution and neurological abnormalities. MYO7A variants can cause Usher syndrome type 1B, which combines congenital hearing impairment with progressive retinal degeneration, as well as some forms of nonsyndromic hearing loss. MYO15A variants are another established cause of inherited nonsyndromic hearing loss. These associations connect cargo transport and sensory-cell architecture directly to human physiology.\n\nMyosin is also a pharmacological target. Blebbistatin, widely used in laboratory studies, inhibits several class II myosins by stabilizing a motor state associated with the ATPase cycle; it is not a general inhibitor of every myosin class. Experimental interpretation requires attention to isoform sensitivity and to the compound’s light sensitivity and phototoxic effects. Other research compounds target different myosin classes, but biochemical selectivity does not automatically establish selectivity in an intact cell.\n\nMavacamten is a cardiac myosin inhibitor approved in the United States in 2022 for specified adults with symptomatic obstructive hypertrophic cardiomyopathy. It reduces actomyosin interactions and favors less active myosin populations. Its clinical use illustrates the importance of regulating the number of available force-generating heads, rather than only changing calcium signaling. Clinical effects and risks depend on cardiac function, drug exposure, and interactions with other medicines.\n\nSeveral mechanistic questions remain active. The precise coupling among phosphate release, actin-binding changes, and the working stroke is still being resolved across different classes and loads. Researchers also distinguish between structural folded-head arrangements and biochemically measured slow-turnover states, whose relationship is not identical in every preparation. At the cellular scale, understanding how motors select cargo, negotiate branched or bundled actin networks, share load, and respond to local tension requires models that integrate molecular kinetics with the changing organization of living cells."
    }
  ],
  "links": [
    "protein",
    "adenosine-triphosphate",
    "actin",
    "eukaryote",
    "cell",
    "cytoskeleton",
    "phylogenetics",
    "evolution",
    "microtubule",
    "gene",
    "amino-acid",
    "cell-membrane",
    "enzyme",
    "hydrolysis",
    "gibbs-free-energy",
    "brownian-motion",
    "action-potential",
    "cell-division",
    "endocytosis",
    "biochemistry",
    "x-ray-crystallography",
    "cryo-electron-microscopy"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}