The fluid mosaic model describes the cell membrane as a two-layer sheet of lipid molecules with proteins drifting through it — not a rigid wall, but a fluid, patchy sea about 5 nanometers thick. Its defining condition is temperature: the membrane stays fluid and mosaic-like only within a living cell's working range, roughly 0 to 45 °C for most animal cells. The model, proposed by S. J. Singer and G. L. Nicolson in a 1972 Science paper, remains the standard description taught today, though later work added the cytoskeleton "fence" and lipid-raft details the original model missed.
What the Model Actually Says, Layer by Layer
Start with the numbers. A typical plasma membrane is 5 to 10 nm thick, and the phospholipid bilayer that forms its core is about 4 to 5 nm. The phospholipids are arranged tail-to-tail: hydrophilic phosphate heads face the watery interior and exterior of the cell, while hydrophobic fatty-acid tails point inward. Because the tails avoid water, the bilayer self-assembles in water without any enzyme or energy input.
The "fluid" half of the name is measurable. Individual phospholipids diffuse laterally within their own leaflet at rates of roughly 1 to 2 micrometers squared per second, according to measurements summarized in standard cell-biology texts such as Alberts et al.'s Molecular Biology of the Cell. A lipid can cross the entire length of a small bacterial cell in seconds. Flip-flopping between the two leaflets is rare — happening perhaps once per hour per lipid without help — which is why enzymes called flippases exist to move specific lipids across.
The "mosaic" half covers the proteins. Integral membrane proteins sit embedded through the bilayer, some spanning it once (single-pass) and others crossing it multiple times (multi-pass). Peripheral proteins attach to the surface. Proteins make up roughly half the mass of a typical plasma membrane, though the ratio varies wildly — the inner mitochondrial membrane runs closer to 75 percent protein because it carries out so much chemistry.
One detail the original 1972 model got wrong or omitted: the cytoskeleton. Below the membrane, a network of actin filaments (in animal cells) or cell-wall-linked structures (in bacteria) corrals proteins into patches roughly 30 to 300 nm across. That corralling is a big part of why real membranes look patchy rather than randomly mixed. Membrane proteins can't simply float anywhere — most are fenced in.
Proteins and Lipids Are Not Free to Wander Anywhere
Proteins don't just float anywhere. Two constraining systems matter. First, the cytoskeleton: actin-based "fences" and "pickets" restrict most proteins to zones of roughly 0.1 to 1 micrometer, so their effective diffusion is tens to hundreds of times slower than in an artificial lipid bilayer. Second, tight junctions between epithelial cells block proteins from crossing between the apical and basolateral surfaces — which is why your gut lining keeps digestive enzymes on one side and nutrient transporters on the other. That polarity is a membrane-level partition, not a chemical one.
Lipid rafts are the second patchiness mechanism. These are small, ordered regions enriched in cholesterol and saturated phospholipids, roughly 10 to 200 nm in diameter, that assemble and disassemble dynamically. According to research reviewed in Nature Reviews Molecular Cell Biology, raft lifetimes are often measured in milliseconds, though some stabilized rafts persist longer when proteins cross-link them. Rafts concentrate certain signaling proteins — Src-family kinases, GPI-anchored proteins, and some receptors — and are proposed as platforms for signal transduction.
The rafts are controversial. A 2006 review by Ken Jacobson and colleagues in Nature Cell Biology noted that raft size in living cells is far smaller than earlier estimates from model membranes, and some researchers argued rafts as stable entities are overhyped. The current consensus is that they exist but are transient and nanoscale. If you're citing raft size in an exam or paper, say "transient, nanoscale, cholesterol- and sphingolipid-enriched" rather than giving a fixed number.
The membrane doesn't just contain the cell—it ships cargo. Vesicles pinch off and fuse continuously, so a fibroblast may recycle its entire plasma membrane surface every 30 to 60 minutes. That turnover is why proteins must be anchored or corralled; a freely floating protein would be swept away.
How Fluidity Changes With Temperature, Cholesterol, and Tail Length
Fluidity is a physical property, and three factors move it. Temperature: as temperature rises, lipids gain kinetic energy and move faster; as it drops, they slow and the bilayer can enter a gel phase. For most animal cell membranes, the gel-to-fluid transition temperature lies well below the cell's working temperature, which keeps them fluid. But that transition temperature depends heavily on the fatty-acid tails.
Saturated tails pack tightly and raise the transition temperature; unsaturated tails have kinks that prevent tight packing and lower it. A phospholipid with two saturated 16-carbon tails transitions near 41 °C, while one with a saturated and an unsaturated tail may transition below 0 °C. That's why cold-water fish accumulate unsaturated fats in their membranes — the packing disruption keeps the membrane fluid in near-freezing water.
Cholesterol is the modulator. It inserts between phospholipids and does two opposite things depending on temperature. Above the transition temperature, cholesterol's rigid steroid ring restricts phospholipid movement and decreases fluidity. Below the transition temperature, it prevents tight packing and increases fluidity. That dual behavior — buffering the membrane against temperature swings — is why animal cells can tolerate wider temperature ranges than bacteria.
Organisms that live at temperature extremes adjust. Bacteria grown at low temperatures increase the proportion of unsaturated and short-chain fatty acids, a response called homeoviscous adaptation. A 2021 comparative study in Frontiers in Microbiology found psychrophilic (cold-loving) bacteria shifted their unsaturated-to-saturated fatty-acid ratio by factors of 2 to 4 when grown at 4 °C versus 20 °C. That's the source of "fluidity is regulated," not "fluidity is fixed."
Cholesterol Does Double-Duty, and That's the Trap
The cholesterol result above is worth spelling out because it's the most misremembered detail of the model. At body temperature (37 °C for humans), adding cholesterol makes a membrane stiffer; at low temperature, it makes it more fluid. "Cholesterol increases fluidity" or "cholesterol decreases fluidity" is wrong alone. It decreases fluidity at high temperature and increases it at low temperature, so the correct statement is that cholesterol buffers fluidity across a range.
The numbers matter. Animal plasma membranes contain roughly 20 to 25 percent cholesterol by mass, according to lipid-composition data summarized in Biochemistry by Berg, Tymoczko, and Stryer. That proportion is enough to broaden the gel-to-fluid transition and eliminate it as a sharp, cooperative event. Membranes with cholesterol don't "freeze" like a clear liquid crystal; they stiffen gradually and retain fluid behavior. That gradual response is why cells can survive sudden temperature drops.
Cholesterol also affects permeability. It reduces the transmembrane passage of small polar molecules like water and glycerol through the bilayer by filling packing gaps between phospholipids. That's why the bile duct epithelium, which contains relatively low cholesterol, is more permeable to water than many other epithelia. And herbs and plants don't use cholesterol — they use sterols like sitosterol, and fungi use ergosterol, which is why some antifungal drugs (the azoles) target ergosterol and leave cholesterol-containing human membranes comparatively alone.
Fluid Versus Rigid Membranes: The Two-State Comparison
Understanding the model requires knowing what it replaced. For decades, textbook diagrams showed the "Davson-Danielli model" — a lipid bilayer sandwiched between two layers of protein, like a protein-lipid-protein sandwich. That model dominated from 1935 to the late 1960s. Singer and Nicolson's 1972 model proposed instead that proteins are embedded in the fluid lipid, not coated on the outside. The table contrasts what each model predicts and where each fails.
| Feature | Danielli-Davson (1935) | Fluid Mosaic (Singer and Nicolson, 1972) |
|---|---|---|
| Protein location | Two continuous surface layers | Embedded (integral) or attached (peripheral) |
| Lipid state | Static | Fluid, laterally mobile |
| Predicted protein:lipid mass ratio | ~2:1 | ~1:1 (added protein surfaces) |
| Explains freeze-fracture EM | Poorly | Well (proteins appear as intramembrane particles) |
| Explains membrane fluidity | No | Yes |
| Status | Superseded | Standard model, amended by rafts and cytoskeleton |
The decisive evidence was freeze-fracture electron microscopy, developed in the late 1960s. When a membrane is frozen and split along its bilayer, proteins show up as particles scattered across the fracture faces — real intramembrane particles, not a continuous protein coat. That observation, combined with experiments by Frye and Edidin in 1970 on fused mouse and human cells showing that surface proteins mixed over minutes, killed the sandwich model. Singer and Nicolson's 1972 Science paper synthesized both lines into the model taught today.
The mosaic-model era also produced a rough failure list. It couldn't explain why some proteins stay fixed on one face of an epithelium (that's the tight-junction work). It couldn't explain the transient signaling microdomains (that's rafts). And it couldn't account for the cytoskeletal corrals (that's single-particle tracking, starting in the 1980s). If you see a diagram still showing a smooth, uniformly mixed mosaic, it's a teaching simplification from before the 1980s.
Semipermeability and the Transport Layer
The bilayer's hydrophobic core is why the fluid mosaic model does double duty in cell biology—it's the physical basis of semipermeability. Small nonpolar gases (oxygen, carbon dioxide) cross a lipid bilayer quickly; their permeability coefficients are on the order of 0.1 to 10 centimeters per second. Water crosses fairly fast, but not purely by lipid diffusion — aquaporins, discovered by Peter Agre in the 1990s (Nobel Prize 2003), provide dedicated channels that pass water at rates up to around 3 × 10⁹ molecules per subunit per second.
Ions and polar molecules are the hard case. A sodium ion has a lipid permeability of roughly 10⁻¹² centimeters per second — about 12 orders of magnitude lower than oxygen — so without transporters a signaling cell couldn't move ions across its own membrane. That's why cells spend a large fraction of their ATP budget on pumps. The sodium-potassium ATPase alone consumes roughly 20 to 25 percent of the body's resting metabolic energy, according to estimates in Molecular Biology of the Cell.
The model also explains why the membrane can fuse and split. Membrane fusion depends on the fluidity of the lipid bilayer. When a vesicle meets the plasma membrane, the two bilayers' hydrophobic surfaces merge, fuse, and re-form as one. If the membrane were rigid, vesicle trafficking would be impossible — and every secretion and endocytosis event would stall. That's a direct argument for the fluidity of the model.
Model Failure Cases: Where the Fluid Mosaic Breaks Down
Four edge conditions show where the model's simplifications stop working. Each is real and measurable.
Membrane rigidification by freeze or fixation. Standard cell fixation with paraformaldehyde at room temperature cross-links proteins and stiffens lipids. Under electron microscopy, fixed membranes show reduced fluidity and altered protein distribution — which is why high-resolution imaging often uses live-cell methods, not fixed preparations. Cold shock in warm-blooded animals. When tissue drops below the lipid transition temperature, ordered gel patches appear. They can crack, which is thought to be one mechanism of frostbite damage at the cell level. That's why cold-storage of organs ("cryopreservation") uses cryoprotectants to prevent ice formation in the bilayer. Bacterial membranes lacking cholesterol. Many bacteria have no cholesterol; their fluidity regulation depends entirely on fatty-acid composition and on hopanoids, which are bacterial sterol-like molecules. A bacteria's plasma membrane may remain fluid over a 0 to 45 °C range but not at 50 °C. Protein aggregation in disease. Misfolded membrane proteins, such as the mutant CFTR in cystic fibrosis, are caught by the protein quality-control machinery and degraded before they reach the surface. The fluid mosaic model explains the normal membrane; it doesn't predict why some proteins fail to insert.
The last case is worth a number. The ΔF508 mutation in the CFTR gene causes protein misfolding and retention in the endoplasmic reticulum. Roughly 1 in 2,500 to 3,000 newborns of Northern European descent are born with cystic fibrosis, according to Cystic Fibrosis Foundation figures. The membrane-level defect isn't a fluidity problem; it's a protein-folding problem. The model is a useful framework, not a complete explanation.
What Lives in the Fluidity Layer: Evidence and a Worked Example
Two early experiments carried the fluid mosaic model. The first was the 1970 Frye-Edidin cell-fusion experiment: mouse and human cells were fused with Sendai virus and labeled with fluorescent antibodies against each species' surface proteins. Within about 40 minutes, the labels intermixed completely. That mixing rate set a lower bound on lateral protein diffusion. The second was the 1972 freeze-fracture result showing intramembrane particles — thousands per square micrometer — consistent with embedded proteins. Together, those two pieces of evidence made the sandwich model untenable.
The model is also testable quantitatively. Single-particle tracking measures individual protein trajectories: a typical membrane protein moves diffusively but is confined to compartments of roughly 0.2 to 0.5 micrometers. Comparing the measured diffusion coefficient (around 0.1 micrometers squared per second for a labeled membrane protein in live cells, versus around 10 in a pure lipid bilayer) yields a confinement factor of roughly 100. That factor is the cytoskeletal-fence correction the original model didn't have. If you have to give one number for the model's evolution, that's the one.
A worked calculation makes the scale concrete. A single phospholipid molecule is about 0.5 nm across. The plasma membrane of a small intestinal epithelial cell covers roughly 1,000 micrometers squared. Divide by the area of one lipid head group (~0.5 nm²) and you get about 2 × 10⁹ lipid molecules — and roughly that many more in the inner leaflet, for around 4 × 10⁹ lipids in the whole plasma membrane. If each can diffuse at ~1 micrometer squared per second within its leaflet, a single lipid can traverse the cell in seconds. Proteins, corralled, need minutes. The model's fluidity and its patchiness are therefore the same physics at two scales.
Membrane Asymmetry: The Model's Two-Faced Detail
The fluid mosaic model is often drawn symmetrically, but real membranes are asymmetric. The outward-facing leaflet is enriched in phosphatidylcholine and sphingomyelin (cholesterol-rich, ordered), while the inward-facing leaflet is enriched in phosphatidylethanolamine and phosphatidylserine (charged, disordered). That asymmetry is maintained by flippases and scramblases. When phosphatidylserine appears on the outer surface — as happens in apoptosis — it acts as an "eat me" signal to macrophages. That one lipid flip triggers a whole immune response. The model didn't anticipate that; the discovery came in the 1990s.
Asymmetry affects how the membrane interacts with the world. Because the outer leaflet is more ordered, receptors that sit in it respond differently from those in the inner leaflet. The insulin receptor, which signals across the membrane, adopts an inactive conformation in ordered environments. That is part of what cholesterol-regulated signaling means. So the mosaic isn't just a patchwork of proteins; it's a patchwork of lipids too, and the two leaflets see different worlds.
For cells that need to recognize self versus non-self — immune cells, for example — the membrane's asymmetry is a functional switch. A passing T-cell reads the outer leaflet of a target cell directly. Flip that asymmetry, and recognition changes. The model's original "fluid" metaphor captured the movement but missed that the membrane faces outward on one side and inward on the other.
FAQ
What does "fluid mosaic" actually mean?
"Fluid" means lipid and protein molecules move laterally within the membrane, and "mosaic" means proteins are scattered through the lipid rather than forming a continuous surface layer. According to the 1972 Science paper by Singer and Nicolson, both properties together distinguish their model from the earlier sandwich model. The average lateral diffusion coefficient for a membrane lipid is about 1 micrometer squared per second. That's fast enough for a lipid to cross a typical cell in seconds.
Which part of the fluid mosaic model turned out wrong?
The original description missed the cytoskeleton and the membrane asymmetry. Specifically, protein diffusion in live cells is roughly 100 times slower than in a pure artificial lipid bilayer, according to single-particle tracking measurements. That factor of 100 comes from cytoskeletal fences and lipid corrals the 1972 paper didn't account for. Lipid-raft substructure, measured in the 1990s and 2000s, is the second amendment. Neither invalidates the model; both are refinements.
How does the fluid mosaic model break down at low temperatures?
Below about 0 °C, many phospholipids enter a gel phase and lose lateral mobility, so the mosaic freezes into patches. For cell membranes with high unsaturated lipid content, the transition can be well below 0 °C, which is why cold-water fish membranes stay fluid in near-freezing ocean water. Artificial cooling of mammalian tissue to 4 °C for transplant purposes requires cryoprotectants to prevent ice formation in the bilayer. Without them, the membrane cracks or hardens.
Is the fluid mosaic model the same as the lipid-bilayer model?
No — the lipid-bilayer model (from Gorter and Grendel, 1925) said only that membranes are two-layer sheets of lipid. The fluid mosaic model (1972) added that proteins are embedded in, not coated outside, the bilayer, and that the whole structure is dynamic. The bilayer is the substrate; the mosaic is how proteins sit in it. Davson-Danielli's 1935 "sandwich" was the hypothesis the mosaic model replaced.
How thick is a cell membrane, in nanometers?
Most plasma membranes measure 5 to 10 nm thick, with the pure phospholipid bilayer at about 4 to 5 nm. Adding the carbohydrate-rich glycocalyx on the outer surface can push the functional thickness to 20 nm or more. Those figures come from electron microscopy and X-ray diffraction studies summarized in standard cell biology references. Bacterial membranes tend to be similar in thickness but lack sterols in most species.
Does cholesterol make membranes more or less fluid?
Both, depending on temperature. At 37 °C, cholesterol reduces membrane fluidity by packing between phospholipids. Below the lipid transition temperature, it prevents tight packing and increases fluidity. That's why animals regulate cholesterol content across their tissues rather than keeping a single level. Animal plasma membranes run roughly 20 to 25 percent cholesterol by mass, and that proportion eliminates the sharp freezing transition.
Can membrane proteins cross from the outside face to the inside face?
Almost never by themselves. Integral membrane proteins are inserted into the bilayer during synthesis and flip across the bilayer at low rates — if at all — because the energy barrier for moving the hydrophilic domains through the hydrophobic core is huge. This is why they're called integral: they're topologically locked. Across one face, they're mobile; across the bilayer, they're effectively fixed. Specialized flippases move lipids, not proteins.
What's the practical use of the model in medicine?
It guides drug design for membrane permeability — drugs must be lipophilic enough to cross the bilayer but not so lipophilic that they accumulate in the membrane. The rule-of-five heuristic, developed by Christopher Lipinski at Pfizer in the 1990s, says compounds of molecular mass under 500 daltons and cLogP under 5 tend to show good membrane permeability. Those are empirical rules, not consequences of the fluid mosaic model, but the model gives them a physical basis.
The fluid mosaic model is a framework for how biological membranes work, not medical advice. If this discussion touches on pharmacology or clinical conditions, treat it as general educational information and confirm specifics with a licensed physician or pharmacist in your jurisdiction; for immediate danger or self-harm risk, contact local emergency services or a crisis line.