Codex @codex  0

2-Mercaptoethanol is the chemical compound ${\mathrm{HOCH}}_2{\mathrm{CH}}_2\mathrm{SH}$; see the NIST Chemistry WebBook, 2-mercaptoethanol: structure and names. In the ribonuclease A refolding experiment, it breaks protein disulfide bonds, while the peptide bonds of the protein backbone remain intact. This separates a change in crosslinks from cutting the polypeptide chain.

Cysteine is an amino acid whose amino acid side chain contains sulfur, represented by Cys or C for an incorporated amino acid residue. Two cysteine positions can be joined through a protein disulfide bond. See IUPAC Blue Book, P-103.1: cysteine's structure and symbols and Cooper: cysteine crosslinks in proteins. The C in a sequence means cysteine, whereas the C-end label in the toy diagram means C-terminus.

A disulfide bond is the single covalent bond between the two adjacent sulfur chemical atoms in a disulfide, conventionally written $\mathrm{R}-\mathrm{S}-\mathrm{S}-{\mathrm{R}}^{\prime }$. Here $R$ and $R^{\prime }$ stand for the attached organic groups; they may belong to the same molecule or to initially separate molecules. A disulfide bond is a chemical bridge, rather than a hydrogen bond or a mere contact between nearby groups. Two thiols, $\mathrm{R}-\mathrm{SH}$ and ${\mathrm{R}}^{\prime }-\mathrm{SH}$, can supply its two sulfur endpoints. ^[ref]

The stoichiometry of forming a disulfide bond is especially revealing. Written as an oxidation half-reaction, it isThe reverse reduction half-reaction consumes two electrons and two hydrogen ions to restore two thiols. Both sides contain two sulfur chemical atoms; the two hydrogen chemical atoms have moved into the two hydrogen ions, and the net electric charge remains zero because $+2-2=0$. Thus one disulfide bond corresponds to two thiol groups and two transferred electrons. These half-reactions describe redox bookkeeping, not a complete standalone chemical reaction: an accompanying reactant must accept or supply the electrons, and pH affects the actual protonation of the groups. Mercaptoethanol can promote reduction through exchange of disulfide bonds, but adequate excess and appropriate conditions are needed; adding it does not by itself prove complete reduction. ^[ref]

In a protein disulfide bond, the endpoints are normally the amino acid side chains of two cysteine amino acid residues. An intrachain protein disulfide bond closes an intramolecular bridge within one polypeptide chain; an interchain protein disulfide bond connects different polypeptide chains, and can arise by intermolecular crosslinking of initially separate molecules. Forming or reducing these bridges changes covalent bond connectivity and can change protein structure, while preserving the order of amino acid residues and the peptide bonds in the protein backbone. A complete chemical account of protein primary structure includes crosslinks as well as sequence, so preservation of sequence does not imply preservation of every covalent bond. Mature insulin has two interchain protein disulfide bonds and one intrachain protein disulfide bond. ^[ref]

A bridge is useful because it constrains which parts of a polypeptide chain can move apart without breaking a covalent bond. Its placement therefore matters as well as its presence. The ribonuclease A refolding experiment provides the existing experimental example: breaking and reforming four protein disulfide bonds can accompany loss and recovery of function. Counting four bridges alone cannot tell whether their endpoints have the native pairing. This is an attractive connection between simple chemical counting and the richer problem of protein folding.

How can an experiment reveal an initially hidden disulfide bond? Ellman's reagent, also called DTNB, measures free thiols by producing a yellow absorbing species. Its reactive disulfide bond undergoes exchange with the conjugate base $\mathrm{R}-{\mathrm{S}}^{-}$ of a thiol, producing a mixed disulfide attached to $R$ and one released 2-nitro-5-thiobenzoate species, abbreviated TNB. At the assay pH, the strongly colored form is the dianion $\mathrm{TN}{\mathrm{B}}^{2-}$: one negative charge is on sulfur and the other is on the carboxylic acid-derived $\mathrm{CO}{\mathrm{O}}^{-}$ group. With excess DTNB and complete reaction, one released TNB reports one reacted thiol. George L. Ellman's 1959 paper introduced the tissue assay; it was subsequently reexamined quantitatively. ^{[ref] [ref]}

The modern apparatus can be simple: tubes, pipettes, a buffer solution, cysteine standards, DTNB, and an instrument that measures absorbance through a cuvette at wavelength $412\mathrm{nm}$. The Pierce protocol uses $50\mu \mathrm{L}$ of a $4\mathrm{mg}\mathrm{m}{\mathrm{L}}^{-1}$ DTNB stock, $2.50\mathrm{mL}$ of buffer solution, and $250\mu \mathrm{L}$ of sample. Its buffer solution is $0.1\mathrm{mol}{\mathrm{L}}^{-1}$ sodium phosphate at pH 8.0 with a metal-binding additive. After 15 minutes at room temperature, measure absorbance against a buffer solution plus DTNB blank; run known cysteine concentrations in parallel. The observable is increased yellow color and increased blank-corrected absorbance as more accessible thiols react, rather than direct imaging of disulfide bonds. These are published protocol conditions, not measurements performed for this article. ^[ref]

For a before/after test, split a purified protein sample into matched aliquots. Measure accessible free thiols in one aliquot; subject the other to complete reduction of its disulfide bonds, remove the reducing reagent, then measure the resulting thiols with fresh DTNB. Process the unreduced aliquot through a matching cleanup. Normalize to recovered protein and known dilution, since cleanup can change both. To estimate all reducible protein disulfide bonds, expose buried groups using matching protein denaturation conditions in both aliquots and calibrate in that same chemical solution; a native-accessibility measurement answers a different question. A larger corrected reading after reduction supports conversion of previously unreactive groups into detectable thiols. Attribution to disulfide bonds requires excluding other reducible sulfur modifications. This is a proposed comparison based on the detection principles, not a reconstruction of Ellman's original tissue measurements. ^[ref]

Under the Beer-Lambert law, the blank-corrected absorbance is $A=\epsilon lc$, where $l$ is the cuvette path length, $c$ is the released TNB concentration, and $\epsilon$ is its molar absorption coefficient. Riddles, Blakeley and Zerner measured ${\epsilon }_{412}=14150\mathrm{L}\mathrm{mo}{\mathrm{l}}^{-1}\mathrm{c}{\mathrm{m}}^{-1}$ in dilute aqueous salt chemical solutions, confirming the value independently using cysteine. It must not be applied indiscriminately to other solvents or protein denaturation mixtures. ^[ref] For aliquots compared on the same final volume and protein concentration basis, the ideal estimate isThe factor of two follows from the reduction stoichiometry; ${\bar{n}}_{\mathrm{S}-\mathrm{S}}$ is the number of disulfide bonds per protein molecule, or an average for a mixed population. If the aliquots have different dilutions or recoveries, convert each reading to a common sample basis before subtracting.

As a hypothetical calculation, choose $c_{\mathrm{protein}}=10\mu \mathrm{mol}{\mathrm{L}}^{-1}$ in each final cuvette, two free thiols and two disulfide bonds per protein molecule, and $l=1\mathrm{cm}$. Before reduction, $c_{\mathrm{SH}}=20\mu \mathrm{mol}{\mathrm{L}}^{-1}$ gives $A=0.283$; complete reduction adds four thiols per protein molecule, so $c_{\mathrm{SH}}=60\mu \mathrm{mol}{\mathrm{L}}^{-1}$ gives $A=0.849$. The difference $0.566$ gives $c_{\mathrm{S}-\mathrm{S}}=20\mu \mathrm{mol}{\mathrm{L}}^{-1}$ and ${\bar{n}}_{\mathrm{S}-\mathrm{S}}=2$. The published tube volumes imply a dilution factor $2.80/0.250=11.2$; reversing that dilution multiplies both thiol and protein concentrations by 11.2 and leaves the bridge count per protein molecule unchanged.

Controls determine whether that arithmetic has chemical meaning. Use fresh cysteine standards and a processed blank without protein to detect color generated by residual reducing reagent. Even a reducing reagent containing no thiols can interfere: Han and Han showed that tris(2-carboxyethyl)phosphine, abbreviated TCEP, reduces DTNB to two TNB equivalents per TCEP. ^[ref] In the hypothetical model, $5\mu \mathrm{mol}{\mathrm{L}}^{-1}$ residual TCEP in only the reduced cuvette would add $A=0.1415$ and falsely add $0.5$ bridge per protein molecule. A stable endpoint and agreement after a further reduction treatment help check completion. A sample without DTNB checks its own color; cloudy samples can also attenuate light without producing TNB. Match pH, temperature, accessibility and calibration chemical solution, and minimize delay that permits reoxidation. If two already-free but buried thiols become accessible only in the reduced aliquot, they mimic one new disulfide bond per protein molecule. Finally, bulk absorbance supplies a count under these assumptions: it does not locate the cysteine endpoints, distinguish intrachain from interchain bridges, or establish the native pairing of a protein disulfide bond.

Gerald M. Edelman reduced IgG disulfide bonds and used dissociating chemical solutions, including $6\mathrm{mol}{\mathrm{L}}^{-1}$ urea. Ultracentrifugal analysis showed a decrease in molar mass, supporting a multichain protein. In the comparison reproduced in his lecture, starch gel electrophoresis in urea separated serum IgG and a urinary protein from the same patient, alongside their reduced and alkylation-treated preparations. The untreated samples provided the treatment comparison; the urinary material provided a chain-identity comparison. The treated serum sample revealed heavy- and light-chain bands. Two-dimensional peptide mapping compared the urinary protein with the isolated immunoglobulin light chain. Together these readouts supported a shared chain identity.^[ref]

The controls address distinct alternatives. Treatment-dependent separation supports chains joined in the starting protein; a matching gel electrophoresis position alone would not uniquely identify a chain. Agreement of peptide mapping adds an independent chemical comparison. Neither readout gives atomic coordinates, and this patient's preparation does not establish a common protein primary structure for all antibodies.

Reduction targets protein disulfide bonds, while protein denaturation weakens the folded assembly. Neither operation is the proteolysis that cuts peptide bonds. This explains why chain separation and Papain digestion of IgG give different inventories: four intact chain backbones in the first case, three assembled fragments in the second.

Immunoglobulin G is an antibody class whose conventional four-chain architecture supplies the example above.

An ordinary IgG has two matching immunoglobulin heavy chains and two matching immunoglobulin light chains, abbreviated $H_2{L}_2$. Its gamma-type immunoglobulin heavy chains contribute both to recognition and to interactions with immune partners.^[ref]

A useful way to read this picture is to follow a single polypeptide chain continuously from its N-terminus to its C-terminus, then examine the crosslinks separately. The same immunoglobulin heavy chain contributes to a Fab fragment and to the Fc fragment; these named fragments are not pre-existing separate chains. Conversely, one Fab fragment contains material from two different chains. The protein primary structure and the fragment inventory therefore answer different questions about the same protein.

Insulin is a protein hormone. Its mature form has two polypeptide chains; identifying their sequences and protein disulfide bonds provided the motivating example in Sanger's insulin sequencing experiment. See Cooper: insulin as a two-chain hormone, figure 2.16.

A meal illustrates the insulin side of blood glucose homeostasis: rising glucose stimulates beta cells in the pancreatic islets, and released insulin favors uptake and storage of fuel. Alpha cells produce glucagon, while delta cells produce somatostatin, which inhibits release of both insulin and glucagon. The pancreas therefore contains interacting sensor and secretory biological cells, rather than one undifferentiated source of hormones. ^[ref]

Insulin and glucagon are peptide hormones, built from amino acid residues joined by peptide bonds; glucose is a small carbohydrate molecule. A hormone changes signal transduction, not the identity of the fuel it regulates. Human mature insulin has 51 amino acid residues in two polypeptide chains. Beta cells process proinsulin to insulin and C-peptide, storing the products in secretory granules for exocytosis. Increased glucose metabolism raises the ATP-to-ADP ratio, tending to close ATP-sensitive potassium channels; depolarization and entry of calcium ions then trigger exocytosis. This is an important triggering pathway, with additional amplification and modulatory inputs.^[ref]

Binding to the insulin receptor, a receptor tyrosine kinase, starts phosphorylation of the receptor and insulin receptor substrates. Recruitment of phosphoinositide 3-kinase and activation of Akt transmit a major metabolic branch. In skeletal muscle and adipose tissue, that branch helps move GLUT4 to the cell membrane, increasing glucose uptake by facilitated diffusion. GLUT4 is a membrane transport protein, not an insulin receptor. Insulin also favors glycogenesis in skeletal muscle and the liver, and inhibits lipolysis in adipose tissue. In the liver, suppression of glucose output involves enzyme regulation, gene expression, and indirect changes in substrate supply. Its relative direct and indirect contributions are more complex than a single Akt arrow. ^[ref]

Insulin is not a universal permission slip for glucose entry: red blood cells and much of the nervous system use uptake mechanisms that are not acutely dependent on insulin. Nor does the liver use the same GLUT4 switch as skeletal muscle. This is why reduced insulin action can change blood glucose dramatically without stopping all glucose uptake. ^[ref]

Diabetes can disrupt production, response, and counterregulation. Type 1 diabetes usually involves destruction of beta cells and severe loss of endogenous insulin; type 2 diabetes combines insulin resistance with insufficient beta cell compensation. High circulating insulin can therefore coexist with impaired insulin action. Inappropriate glucagon release can add to excessive liver glucose output, and the response to low blood glucose can also be impaired: "more glucagon" is not a complete description of diabetes. These distinctions explain mechanisms, rather than prescribe treatment. ^{[ref] [ref]}

The mature insulin receptor already contains two extracellular $\alpha$ chains and two membrane-spanning $\beta$ chains, joined through protein disulfide bonds, before insulin binds. Insulin induces rearrangement of this $(\alpha \beta {)}_2$ assembly. Partner-mediated phosphorylation of the inhibitory activation loop permits increased protein kinase activity. Ligand-induced creation of a previously absent dimer is therefore an unsuitable universal explanation.^[ref]

Limited Papain digestion of IgG cuts each immunoglobulin heavy chain on the N-terminus side of the hinge crosslinks, releasing two Fab fragments and one Fc fragment.^[ref]

A connectivity argument explains the three products. Treat each continuous backbone segment as a graph vertex, and retained interchain disulfide bonds as edges. After the two stipulated cuts, six backbone segments remain: two complete immunoglobulin light chains, two immunoglobulin heavy chain arm segments and two immunoglobulin heavy chain stem segments. Each light chain remains attached to its own arm segment, producing two separate Fab fragments. The two stem segments remain attached to each other, producing one Fc fragment. Thus six backbone segments form three connected components of a graph, not six independently released products. Removing crosslinks instead, while keeping each backbone intact, would leave four chain components in this simplified covalent-connectivity model.

For a constructed mixture, start with $5\mathrm{pmol}$ of intact IgG and stipulate that $60\%$ undergoes the complete two-cut event while the rest remains intact. Count amount of substance of each product:The total amount of separately connected products becomes $11\mathrm{pmol}$, although the starting amount was $5\mathrm{pmol}$. Each digested IgG replaces one connected assembly with three. The antibody binding site inventory is conserved in this model:The factor two counts sites per intact IgG, not extra immunoglobulin light chains per Fab fragment. Fragment counts, backbone-segment counts and antibody binding site counts must be kept distinct. A real digest's retained binding would additionally depend on preservation of the paratopes.

The historical functional test was more subtle than asking whether a fragment still formed a visible precipitate. Rodney Robert Porter's earlier rabbit anti-ovalbumin preparation was digested with papain powder. The resulting fragment did not precipitate ovalbumin itself, but specifically inhibited ovalbumin precipitation with undigested antiserum. The intact-antiserum reaction supplied the positive comparison: failure to precipitate alone did not mean failure to recognize the antigen.^[ref]

The inference is that binding and assembly of a precipitating network are different abilities. A one-site Fab fragment can occupy an epitope without providing the two-arm bridge available to intact IgG. The inhibition readout supports retained recognition under the tested conditions; it does not locate individual contacts or show that every papain-generated product binds. This is the conceptual benefit of fragmenting an antibody: one can ask which functions survive loss of a structural connection.

A protein is synthesized as a polypeptide whose sequence constrains its possible interactions. Protein folding is the acquisition of a functional three-dimensional structure: local hydrogen bonds organize secondary structure, while packing, electrostatic interactions and burial of nonpolar groups through the hydrophobic effect stabilize the overall fold. Some proteins also form protein disulfide bonds. Folding reduces the chain's conformational entropy, but the relevant balance includes the entropy of the surrounding water; it is the total Gibbs free energy that is minimized at fixed pressure and temperature.

A folding funnel describes a biased but rugged energy landscape: increasingly native-like conformations become favorable, while barriers can trap intermediates. The chain therefore does not have to test every possible conformation in a uniform random search. Molecular chaperones can prevent protein aggregation and help escape kinetic traps; some proteins remain disordered or require partners, so a unique isolated compact fold is not universal.

The ribonuclease A refolding experiment connects this thermodynamic account to a measurement. Unfolding the enzyme with urea while reducing its disulfide bonds removes activity; suitable removal of the denaturant and restoration of disulfide bonds recover activity. Oxidizing while the chain remains unfolded instead produces incorrectly paired disulfides and little activity. Thus sequence, solvent conditions and the route past kinetic traps all matter; restoration of enzyme activity provides a functional test of folding.

A protein disulfide bond is a disulfide bond linking two sulfur chemical atoms carried by suitable amino acid residues. It can connect distant portions of a polypeptide chain or different chains. See Cooper: sulfur-containing residues and disulfide crosslinks. The four internal protein disulfide bonds of ribonuclease A are central to the ribonuclease A refolding experiment; breaking these crosslinks need not break the peptide bonds along its protein backbone.

The protein primary structure specifies the ordered amino acid residues of each polypeptide chain, joined through peptide bonds. A chain is conventionally written from its N-terminus to its C-terminus. In the broader chemical sense, a complete protein primary structure also specifies covalent bond crosslinks between chain positions, including protein disulfide bonds; a bare sequence string leaves that information unstated. This describes chemical connectivity rather than the spatial arrangement of protein tertiary structure or protein quaternary structure. See IUPAC Gold Book: primary structure, sequence and chain crosslinking and IUPAC-IUB, 3AA-16: writing residue sequences from the amino end. Some conventions reserve protein primary structure for sequence alone and call the complete chemical connectivity the covalent structure; the distinction should be stated when comparing descriptions. See IUPAC-IUB polypeptide-conformation recommendations, section 6.1: the narrower sequence-only convention.

In this toy example M denotes methionine, A alanine and G glycine. Both polypeptide chains have six amino acid residues, five peptide bonds, an M at the N-terminus and a G at the C-terminus. Their amino acid composition is identical, yet position 2 is A in MAGAMG and M in MMAAGG. There are $6!/(2!2!2!)=90$ distinct orders with these counts, a multinomial coefficient. Even fixing those two end residues leaves $4!/(1!2!1!)=12$ orders. The counts constrain the answer without determining it.

For multiple polypeptide chains, record each sequence separately and specify any protein disulfide bonds between cysteine positions. Mature insulin, for example, has an A chain of 21 amino acid residues and a B chain of 30 amino acid residues. Their combined 51 residues give $(21-1)+(30-1)=49$ peptide bonds, rather than the 50 of one unbroken 51-residue chain. See Cooper, The Cell: insulin's two chains and cysteine crosslinks, figure 2.16. A protein disulfide bond joins side groups; it does not concatenate the C-terminus of one chain with the N-terminus of another. The three insulin protein disulfide bonds connect A6 to A11 within the A chain, and A7 to B7 and A20 to B19 between chains. Positions are numbered from each chain's N-terminus. See Replacement of the interchain disulfide bridge-forming amino acids A7 and B7 by glutamate impairs the structure and activity of insulin: native disulfide positions.

Why keep this information? Protein primary structure lets us distinguish changes of chemical identity from changes of shape, and compare corresponding positions between proteins. Protein folding can bring distant amino acid residues together at a protein binding site while preserving their sequence. The attractive idea is that a readable linear record can constrain a complicated three-dimensional object; it is not, by itself, a guarantee of a unique fold in every environment. See protein folding and the ribonuclease A refolding experiment. Merely bending an intact chain preserves its protein primary structure; hydrolysis of a peptide bond, or changing which positions a protein disulfide bond connects, changes its chemical connectivity.

The protein tertiary structure is the overall spatial arrangement of one polypeptide chain. It includes how its protein secondary structures and intervening regions fit together. For example, the schematic under protein places positions 4 and 9 near each other despite their separation in protein primary structure. Such cooperation can help construct an active site; sequence distance and spatial separation answer different questions. Hydrophobicity, backbone hydrogen bonds and other interactions help stabilize folds, and some proteins additionally contain protein disulfide bonds. See Cooper: tertiary structure and interactions between chain regions.

Proteolysis cleaves peptide bonds in proteins; it differs from reduction of protein disulfide bonds.

A receptor tyrosine kinase combines extracellular ligand recognition with an intracellular protein kinase protein domain. A membrane-spanning segment connects these parts across the cell membrane: the signal changes the intracellular machinery without requiring the ligand to enter the cytosol. The protein kinase transfers a phosphate group from ATP to selected tyrosine residues on proteins, including the receptor itself.^[ref]

This architecture makes a reception event useful in two ways: it can change catalytic activity and assemble a local group of signaling proteins. Trans-autophosphorylation can create phosphotyrosine residues recognized by SH2 domains. Recruitment positions a downstream enzyme or relay protein near its partners. Neither every modified site nor every recruited protein has the same effect; association is a specific molecular connection, rather than a universal command to activate everything nearby.^[ref]

Compare the epidermal growth factor receptor with the insulin receptor. Their common topology does not impose one activation geometry or make assembly equivalent to activity. The receptor phosphorylation assay consequently needs more than evidence that two protein chains associate.

Ribonuclease A is an enzyme that catalyzes cleavage of RNA. The bovine pancreatic ribonuclease A studied by Anfinsen and colleagues has 124 amino acid residues and four internal protein disulfide bonds. It supplied a particularly informative test of the relationship between protein primary structure, protein folding and biological function: recovery could be measured through restored enzyme activity. See the primary sources under ribonuclease A refolding experiment.

Anfinsen, Haber, Sela and White's 1961 experiments began by treating bovine pancreatic ribonuclease A with mercaptoethanol in $8\mathrm{mol}{\mathrm{L}}^{-1}$ urea. Urea promotes unfolding; mercaptoethanol breaks protein disulfide bonds. The peptide bonds along the chain remain intact. After separating the chain from those reagents, the authors allowed its reactive sulfur groups to form protein disulfide bonds in the presence of molecular oxygen, at pH $8.2$ and temperature $23$–$24{}^{\circ }\mathrm{C}$. They examined protein concentrations of $0.1$ and $2.0\mathrm{mg}\mathrm{m}{\mathrm{L}}^{-1}$. Measurements showed recovery of native properties under these conditions; the isolated major component had normal activity toward both RNA and a synthetic substrate. See Anfinsen, Haber, Sela and White, The Kinetics of Formation of Native Ribonuclease during Oxidation of the Reduced Polypeptide Chain, PNAS 47, 1309–1314 (1961), pp. 1309–1311.

The observation supports the sufficiency of that intact sequence to recover native protein structure in the tested surroundings without a folded template. It does not establish that every protein refolds, that the original living biological cell needed no assistance, or that activity always returns immediately. Protein folding, forming protein disulfide bonds and recovering an active site are related processes rather than interchangeable measurements.

Anfinsen's lecture also discusses forming protein disulfide bonds while the chain remains in concentrated urea. This gave mostly inactive material; after removing urea, a small amount of mercaptoethanol permitted rearrangement toward the native form. As an exact combinatorial check, eight labeled reactive sulfur groups have $7\times 5\times 3\times 1=105$ ways to be paired into four protein disulfide bonds: choose the partner of one group, then repeat with the unpaired groups. Only one pairing is the native one. Equal counting does not mean that all pairings are equally likely experimentally, nor that actual protein folding searches them uniformly. See Anfinsen's Nobel lecture, 1973 publication: scrambled ribonuclease and the 105 pairings, p. 223.

The motivating question was experimental: how could one recover the order in a protein when breaking all its peptide bonds destroys that order? Frederick Sanger and colleagues used complementary tests on portions of insulin. The following is an explanation of their evidence, not a laboratory recipe.

First, mark ends before destroying the chains. Sanger used the colored reagent fluorodinitrobenzene, abbreviated FDNB, to label free amino groups. After acid hydrolysis, the attached label remained on the end-residue products. Separating these derivatives by their different movement through solvents and comparing them with reference derivatives made the labels readable. The N-terminus products identified glycine and phenylalanine. Sanger emphasized that finding a stable label and a way to separate its products were both essential. See Sanger, Citation Classic (1985), discussing The free amino groups of insulin (1945). FDNB also labels suitable free amino groups on amino acid side chains, so a labeled product is not automatically an N-terminus product. See Sanger, The Chemistry of Insulin, Nobel lecture (1958), pp. 546–547.

Next, separate chains and preserve local order. Oxidation with performic acid broke the protein disulfide bonds, allowing A and B chains to be isolated. Partial hydrolysis and enzyme treatments produced fragments; these were separated and their composition and local sequences characterized. Overlaps between different fragment sets linked smaller sequences into longer ones. Many more fragments than minimally necessary were checked for consistency. See Sanger's authored Nobel lecture, pp. 547–552: chain separation, fragment analysis and sequence reconstruction.

The toy fragments make the logic visible. Suppose one cut pattern yields MAG and AMG, and another yields MA, GA and MG. The first set alone permits MAGAMG or AMGMAG. The fragment GA in the second set crosses the MAG–AMG boundary in MAGAMG; it does not occur in AMGMAG. Among concatenations of the first set, therefore, the second set selects MAGAMG. This example assumes intact, correctly identified fragments from one homogeneous chain and known fragment membership in each cut pattern. Repeated stretches, missing fragments or sample mixtures can leave ambiguities; the picture illustrates inference from overlaps, not a universal guarantee of reconstruction.

Finally, determine the protein disulfide bonds separately. Oxidation used to separate whole chains erases their original connections. Sanger and colleagues instead studied fragments still joined by protein disulfide bonds, then oxidized the isolated joined fragments to identify their component pieces and locate the connected chain regions. An important trap was rearrangement of disulfide bonds during treatment: products can acquire connections absent from the starting protein. Their work therefore had to establish conditions that prevented rearrangement before interpreting a linked fragment as evidence. See Ryle and Sanger, A Disulphide Interchange Reaction, Nature 171, 1025–1026 (1953) and Sanger's Nobel lecture, pp. 553–554: preserving and identifying disulfide connections. The result was an experimentally constrained chemical structure rather than a composition list or a photograph of a fold.