Membrane Proteins

A membrane protein is any protein that is physically associated with a lipid bilayer, and biologists classify these associations into three mechanistically distinct categories. Integral membrane proteins (also called intrinsic proteins) have one or more stretches of sequence embedded within the hydrophobic core of the bilayer; most of these are transmembrane proteins that cross the membrane entirely, exposing domains on both sides, but some integral proteins are only partially embedded and do not cross all the way through. Peripheral membrane proteins (extrinsic proteins) never enter the hydrophobic core at all; instead they bind to the polar head groups of lipids or, more commonly, to the exposed hydrophilic surfaces of integral membrane proteins, held in place by electrostatic and hydrogen-bond interactions. Lipid-anchored proteins are a distinct third category: the protein itself may be entirely hydrophilic, with no embedded or amphipathic domain, but it is covalently attached to a lipid molecule — a fatty acid, a prenyl group, or a glycosylphosphatidylinositol (GPI) anchor — that inserts into one leaflet of the bilayer and holds the protein at the membrane surface. The defining chemical logic in all three cases is the same: a stable, permanent association with a lipid bilayer requires either a hydrophobic surface large enough to bury against the fatty acyl chains (integral and lipid-anchor strategies) or a strong enough noncovalent interaction with something that is already anchored (the peripheral strategy). Integral membrane proteins solve the hydrophobicity problem using transmembrane α-helices, typically 20–25 residues long and rich in nonpolar side chains (leucine, isoleucine, valine, phenylalanine), long enough to span the ~30 Å hydrophobic core of the bilayer. A minority of integral proteins, mostly in bacterial outer membranes, mitochondrial outer membranes, and chloroplast outer membranes, instead use β-barrels, closed cylinders of antiparallel β-strands whose alternating hydrophobic and hydrophilic side chains face outward into lipid and inward into an aqueous pore, respectively. A useful conceptual anchor is amphipathicity at the right length scale. A transmembrane helix must be hydrophobic along its entire ~20-residue length to match the thickness of the acyl-chain region; a peripheral protein's membrane-binding surface is instead a patch of a folded globular domain, often amphipathic or positively charged (binding to negatively charged phospholipid head groups such as phosphatidylserine); and a lipid anchor's hydrophobic module is not part of the polypeptide chain at all, but a chemically distinct lipid moiety added after (or during) translation. This distinction between "genetically encoded hydrophobic sequence" and "post-translationally added lipid" is the single most important conceptual fork in the whole topic, and it directly predicts how each class behaves when membranes are experimentally disrupted.

Step 1 — Integral proteins are built with hydrophobic transmembrane domains during synthesis. As a nascent polypeptide is translated on a ribosome bound to the endoplasmic reticulum, a signal-recognition-particle-mediated pathway threads it through the Sec61 translocon into the ER membrane. Stretches of about 20 consecutive hydrophobic residues, forming an α-helix, are laterally released from the translocon directly into the lipid bilayer rather than being pushed through into the ER lumen. A protein can contain one such helix (a single-pass or "bitopic" protein, e.g., glycophorin A) or many (a multi-pass or "polytopic" protein, e.g., the seven-transmembrane-helix architecture of G-protein-coupled receptors, or the twelve-helix glucose transporter GLUT1). Each transmembrane helix buries roughly 20 hydrophobic residues against the acyl chains, an energetically favorable arrangement that is also entropically stable — once inserted, the protein does not spontaneously leave, because extracting a helix into water would expose dozens of nonpolar side chains to solvent, an enormous energetic penalty (each buried CH2/CH3 contact is worth roughly 20–30 cal/mol of favorable hydrophobic transfer energy, and a 20-residue helix multiplies this into a large net stabilization).

Step 2 — β-barrel proteins fold outside the Sec pathway and insert via a dedicated machine. In Gram-negative bacteria, mitochondria, and chloroplasts, outer-membrane porins and related channels are synthesized without signal-anchor sequences and are instead threaded across an inner membrane (or imported into the organelle) in an unfolded state, then folded and inserted into the outer membrane by the BAM complex (β-barrel assembly machinery, bacteria) or the analogous SAM complex (mitochondria). The resulting barrel typically has 8 to 22 antiparallel β-strands; because β-strands alternate side chains between the two faces of the sheet, alternating residues point outward into lipid (hydrophobic) and inward into the aqueous channel (polar), an elegant solution that lets a single secondary-structure element satisfy two opposite chemical environments simultaneously.

Step 3 — Peripheral proteins bind afterward, noncovalently, often to exposed domains of integral proteins. Spectrin and ankyrin, the cytoskeletal proteins that give red blood cells their biconcave shape, do not touch the lipid bilayer directly at all; ankyrin binds to the cytoplasmic domain of the integral protein Band 3 (the anion exchanger), while spectrin forms a meshwork beneath the membrane connected to actin via protein 4.1. Other peripheral proteins, such as cytochrome c on the outer face of the inner mitochondrial membrane, bind directly to charged phospholipid head groups (cytochrome c binds cardiolipin) through electrostatic and hydrogen-bonding surfaces on the folded protein. Because these interactions are noncovalent — ionic bonds, hydrogen bonds, sometimes hydrophobic patches shallowly inserted into just the head-group region — they can be broken without disrupting the lipid bilayer itself.

Step 4 — Lipid-anchored proteins are covalently modified after translation to gain membrane affinity. Three chemistries dominate. Fatty acylation covalently attaches myristate (14-carbon, usually N-terminal glycine, via an amide bond, essentially irreversible) or palmitate (16-carbon, usually on internal cysteines, via a thioester bond that can be enzymatically removed, allowing reversible membrane association — important for signaling proteins like Ras that cycle on and off membranes). Prenylation attaches a farnesyl (15-carbon) or geranylgeranyl (20-carbon) isoprenoid group to a C-terminal cysteine within a CAAX motif, again anchoring small GTPases such as Ras and Rho family proteins to the cytosolic leaflet. GPI-anchoring occurs entirely within the ER lumen: a preformed glycosylphosphatidylinositol structure is attached to the C-terminus of a protein after a signal sequence is cleaved, and the whole GPI-anchored protein is then displayed exclusively on the extracellular (or organellar-lumenal-facing, ultimately outer) leaflet — GPI anchors never appear on the cytosolic face because they are built and attached inside the ER lumen. Alkaline phosphatase and the enzyme acetylcholinesterase are classic GPI-anchored proteins.

Step 5 — Orientation is fixed for the life of the protein. Once a transmembrane protein is inserted into the ER membrane with a particular topology (which domains face lumen vs. cytosol), that orientation is preserved through the whole secretory pathway — ER to Golgi to plasma membrane — because membrane proteins cannot flip across the bilayer (flipping would require dragging a hydrophilic domain through the hydrophobic core, a process with a prohibitively high energy barrier, unlike lipids, which can be flipped enzymatically by flippases). This is why a receptor's ligand-binding domain always ends up facing the extracellular space and its signaling domain always ends up in the cytosol — topology is set once, early, and never changes.

Step 6 — Detecting each class experimentally follows directly from the mechanism. Because integral proteins are held in place by burial in the hydrophobic core, only detergents (like Triton X-100 or SDS, which insert into and dissolve the bilayer, forming lipid–detergent micelles around the hydrophobic transmembrane domains) can release them without denaturing them beyond recognition. Because peripheral proteins are held on by ionic and hydrogen bonds, they can be released by high salt concentrations (which screen electrostatic interactions) or by shifting pH (which alters the protonation state of charged residues), without needing any detergent and without lysing the membrane. Because lipid-anchored proteins are attached by a covalent lipid bond, they resist both salt and mild pH change, but can be released by detergent (which competes for the lipid anchor's hydrophobic tail) or, diagnostically, by treatment with phospholipase C, which cleaves the phosphodiester bond within a GPI anchor and releases the entire protein intact into solution — a specific and elegant test unavailable for any other class.

The three-way classification of membrane proteins was not obvious from first principles; it emerged from a combination of biochemistry, sequence analysis, and electron microscopy carried out mostly in the 1970s, using the human red blood cell membrane as the workhorse system because erythrocytes can be lysed in hypotonic solution to yield pure plasma membrane "ghosts" free of any other organelle — a direct application of Cell Fractionation logic.

Extraction experiments (Steck, Fairbanks, Bretscher, and colleagues, early-to-mid 1970s). Ted Steck and coworkers systematically treated red blood cell ghost membranes with a graded series of conditions and then ran the extracted proteins on SDS-polyacrylamide gel electrophoresis (SDS-PAGE) to see which proteins came off. Mild treatments — washing with low-ionic-strength buffer, or with high salt (e.g., 1 M NaCl), or with alkaline pH (pH 11 EDTA buffer, later formalized by Fairbanks, Steck, and Wallach's 1971 protocol) — stripped spectrin, ankyrin, and actin off the membrane while leaving the membrane's overall lipid bilayer intact and still capable of resealing into vesicles. This showed these cytoskeletal proteins were peripheral: their attachment did not require disrupting the bilayer at all. In contrast, glycophorin and Band 3 (the anion exchanger) survived all of these mild treatments and could only be extracted using nonionic detergents such as Triton X-100, which visibly solubilized the lipid bilayer itself. This was the direct biochemical proof that glycophorin and Band 3 were integral, not peripheral — their release required destroying the very structure that held them, whereas peripheral proteins could be removed while that structure remained intact.

Sequence hydropathy and the discovery of the transmembrane helix motif. Once protein sequencing (and later cDNA sequencing) became routine, biologists noticed that integral membrane proteins identified biochemically contained conspicuous stretches of roughly 20 consecutive hydrophobic residues. Jaap Kyte and Russell Doolittle formalized this in 1982 with the hydropathy (hydropathicity) plot, assigning each amino acid a numerical hydrophobicity score and computing a running average across a sliding window (typically 19–21 residues) along the sequence; a strong positive peak of this length matching the ~30 Å bilayer thickness predicts a transmembrane helix. Applied retrospectively to glycophorin A — whose transmembrane domain had already been chemically mapped by direct protein sequencing and proteolytic mapping (Marchesi and colleagues, 1970s: a single ~23-residue hydrophobic stretch flanked by a sialic-acid-rich extracellular domain and a short cytoplasmic tail) — the hydropathy plot correctly picked out exactly the helix that biochemistry had already located, validating hydropathy analysis as a predictive tool that is still standard practice today.

Freeze-fracture electron microscopy (Branton, Deamer, Pinto da Silva, 1966 onward). A membrane sample is rapidly frozen and then physically fractured; because the two lipid leaflets are held together only by weak van der Waals forces, the fracture plane splits the bilayer down its hydrophobic middle, exposing the interior faces of both leaflets, which are then shadowed with platinum and imaged. Freeze-fracture images of red blood cell and many other plasma membranes revealed a dense scattering of roughly 8–10 nm bumps — intramembranous particles (IMPs) — studding the fractured internal faces. Because these particles appeared within the fracture plane, embedded in what should have been pure lipid, this was direct physical proof that integral proteins genuinely penetrate into the hydrophobic core of the bilayer rather than simply resting on its surface, exactly as the fluid-mosaic model proposed. Critically, mutant or genetically engineered membranes lacking specific integral proteins (or membranes treated to remove them) showed a corresponding reduction in IMP density, linking specific particles to specific known proteins.

Discovery of GPI anchors (Low and colleagues, early 1980s; Ferguson and Williams, mid-1980s). Martin Low showed that treating alkaline phosphatase or acetylcholinesterase — proteins that resisted detergent-free extraction like integral proteins, yet had no obvious hydrophobic transmembrane sequence upon protein sequencing — with a bacterial enzyme, phosphatidylinositol-specific phospholipase C (PI-PLC), released the entire protein into solution, intact and still enzymatically active, without any detergent. This was the discriminating experiment: it showed the membrane attachment was neither classic transmembrane insertion (no hydrophobic sequence was present) nor typical peripheral binding (salt and pH did nothing), but a distinct third mode — a cleavable lipid anchor — leading to the structural elucidation of the glycosylphosphatidylinositol anchor itself. Together, these lines of evidence — differential extraction sorted by SDS-PAGE, hydropathy analysis of sequences, direct visualization by freeze-fracture, and enzymatic release of lipid anchors — converged from completely independent methods on the same three-class picture, which is why the classification is considered solid, textbook-level cell biology rather than a provisional model.

Example 1 — Classifying an unknown protein from an extraction/hydropathy dataset. A membrane protein "X" is studied with three tests. (1) Washing the membrane with 1 M NaCl at neutral pH does not release protein X. (2) Treating with PI-PLC does not release protein X either. (3) Only 1% Triton X-100 releases protein X, and a hydropathy plot of its sequence shows seven distinct hydrophobic peaks of about 20–24 residues each, spaced apart by short hydrophilic loops, spanning a total sequence of 350 residues. Reasoning: failure to be released by salt rules out a purely peripheral protein (peripheral proteins come off with ionic strength/pH changes). Failure to be released by PI-PLC rules out a GPI-anchored protein. Requiring detergent to solubilize it is consistent only with an integral membrane protein, since only detergent can dissolve the hydrophobic core holding it in place. The hydropathy plot clinches it: seven hydrophobic peaks of the right length (~20 residues, matching the ~30 Å bilayer) predict seven transmembrane helices — this is a polytopic integral membrane protein, structurally consistent with a receptor of the seven-transmembrane GPCR family (rhodopsin, the archetype, has exactly this architecture: 348 residues folded into seven transmembrane α-helices).

Example 2 — Predicting the topology consequence of a signal-anchor mutation. A researcher studies a single-pass type I transmembrane receptor (N-terminus extracellular, C-terminus cytoplasmic, one 22-residue transmembrane helix, the normal topology for most cell-surface receptors including the LDL receptor and many receptor tyrosine kinases). They engineer a mutant in which 8 of the 22 hydrophobic residues within the transmembrane segment are mutated to charged residues (aspartate and lysine), reducing the length of the truly hydrophobic run to only 14 residues. Predicted consequence: a 14-residue hydrophobic stretch is too short to reliably span the ~30 Å hydrophobic core as a helix (an α-helix rises about 1.5 Å per residue, so 14 residues span only about 21 Å, and the interspersed charged residues additionally destabilize helix packing within lipid). The translocon is likely to fail to recognize this shortened, charge-interrupted segment as a stable transmembrane anchor. Two outcomes are plausible and have both been observed experimentally in analogous engineered constructs: either the protein is not stably inserted and is instead translocated entirely into the ER lumen and secreted (behaving like a signal sequence that is cleaved rather than a stop-transfer anchor), or, if some residual hydrophobicity is enough for marginal insertion, the protein may show mixed or unstable topology, with much lower total surface expression, since the translocon and downstream quality control machinery (the ER's protein-folding checkpoint) tend to retain and degrade improperly anchored proteins rather than release them to the plasma membrane. This illustrates that transmembrane topology is not a passive consequence of a protein's final fold, but is enforced at the moment of insertion by biophysical rules — length and hydrophobicity of the signal-anchor sequence — that can be experimentally violated and whose violation produces predictable, testable phenotypes.

Much of the definitive biochemical mapping of red blood cell membrane proteins — the extraction ladder that let biologists distinguish integral from peripheral proteins with rigor — was carried out by Theodore (Ted) Steck and colleagues at the University of Chicago in the early 1970s, building directly on the erythrocyte ghost preparation methods developed earlier by others and on Jonathan Singer and Garth Nicolson's 1972 fluid-mosaic model, which had proposed the integral/peripheral distinction conceptually but needed direct biochemical proof. Steck's systematic gradient of extraction conditions — mild low-ionic-strength washes, high-salt washes, alkaline EDTA extraction, and finally detergent solubilization, each analyzed by SDS-PAGE — turned a conceptual model into an operational, repeatable protocol that other labs could use on any membrane, not just red blood cells. This same period saw parallel, converging work: Vincent Marchesi's sequencing of glycophorin A pinned down the first fully mapped transmembrane domain at the amino acid level, Daniel Branton's freeze-fracture work gave the direct physical image of embedded intramembranous particles, and Martin Low's phospholipase experiments a decade later revealed the entirely separate GPI-anchoring mechanism. No single person "discovered" membrane protein classification; it was assembled from these converging biochemical, structural, and imaging threads across roughly fifteen years.

Membrane protein classification is not an abstract taxonomy — it directly shapes drug discovery, diagnostics, and biotechnology. G-protein-coupled receptors, seven-pass integral membrane proteins, are the target of roughly a third to a half of all marketed pharmaceuticals (beta-blockers, antihistamines, opioid analgesics), and their identification as polytopic integral proteins by hydropathy analysis was essential groundwork for the receptor cloning efforts of the 1980s–90s that made these drugs possible to design rationally. Ion channels — integral, often multi-pass proteins — are targets for anti-epileptics, local anesthetics, and cardiac antiarrhythmics, foreshadowing the mechanistic detail covered under Ion Channels and Carrier Proteins. GPI-anchored proteins have direct clinical relevance: paroxysmal nocturnal hemoglobinuria (PNH) is a disease in which a somatic mutation in the PIGA gene disables GPI-anchor synthesis, causing red blood cells to lack GPI-anchored complement-regulatory proteins (CD55, CD59); without these anchors, the cells are destroyed by the patient's own complement system, causing hemolysis and thrombosis. This is a direct, patient-level demonstration that the lipid-anchoring mechanism described above is not a laboratory curiosity but a load-bearing biological system. In biotechnology, the extraction logic discovered for red blood cells (detergent for integral proteins, salt/pH for peripheral proteins, phospholipase for GPI-anchored proteins) remains the operational basis for membrane protein purification used to prepare samples for cryo-electron microscopy and X-ray crystallography — the modern structural determination pipeline for essentially every drug-target membrane protein. Peripheral protein interactions with the cytoskeleton are central to understanding hereditary spherocytosis and elliptocytosis (mutations in spectrin, ankyrin, or Band 3 that weaken red-cell membrane mechanics), and lipid-anchoring chemistry (particularly prenylation of Ras) is directly targeted by farnesyltransferase inhibitor cancer drugs, which aim to prevent oncogenic Ras from reaching the plasma membrane where it must sit to signal. Finally, the fundamental concept of a fixed, unflippable transmembrane topology, established here, underlies every downstream discussion of transport proteins in Facilitated Diffusion and channel gating.

Ask the tutor: How does a cell decide, during translation, whether a hydrophobic stretch becomes a permanent transmembrane anchor versus a cleaved signal sequence?