Concept

The levels of protein structure

T-007Home BU-102Threads structure · energy

Statement

From amino-acid sequence to folded, functional shape.

Why it matters

This is the structural vocabulary the rest of the network's protein-related results depend on. protein-folding explains the thermodynamics that select one specific structure at this hierarchy's tertiary/quaternary levels; enzyme-lower-activation-energy presupposes a folded active site built from precisely positioned tertiary contacts; and dna-double-helix, lipids-amphipathic-assembly and carbohydrates-polysaccharides are this same unit's parallel accounts of how the other three macromolecule families' structure likewise determines their function. Proteins are the most structurally versatile of the four families specifically because they build a single covalent backbone (primary structure) up through four increasingly complex, non-covalent levels of organisation, each constrained by the one below it.

Understanding the hierarchy also explains, mechanistically, why a single amino-acid substitution can sometimes have a drastic functional effect and sometimes almost none at all: the consequence of a substitution depends entirely on which level of structure it disrupts.

Hypotheses

Each level of structure is built from, and constrained by, the level directly below it.Secondary structure cannot exist independently of primary structure — it is the local, regularly repeating backbone geometry a given sequence's backbone atoms are able to adopt; tertiary structure is the overall three-dimensional packing of the secondary-structure elements a chain contains; quaternary structure exists only for proteins built from more than one polypeptide chain, describing how those separately folded subunits associate. Without this dependency, "levels" would be an arbitrary classification rather than a genuine, causally ordered hierarchy. Secondary structure is stabilised specifically by backbone hydrogen bonding between the carbonyl oxygen and amide hydrogen of the peptide bond, not by side-chain interactions.This is what makes secondary structure geometrically regular and predictable across otherwise very different sequences: the backbone atoms and their hydrogen-bonding geometry are common to every residue regardless of side chain identity, while side-chain (R-group) interactions are instead the dominant forces stabilising tertiary structure, where geometry is sequence-specific and irregular.

Proof

\[ \text{Primary structure: the covalently bonded linear sequence of amino acids, joined by peptide bonds.} \]

Amino acids are joined in a condensation reaction between one residue's carboxyl group and the next residue's amino group, forming a rigid, planar peptide bond and releasing water; the resulting sequence, read from the free amino (N) terminus to the free carboxyl (C) terminus, is primary structure and is specified directly and entirely by the genetic code (translation-genetic-code) that encodes it. A

\[ \text{Secondary structure: regular, locally repeating backbone conformations (}\alpha\text{-helix, }\beta\text{-sheet) stabilised by backbone hydrogen bonds.} \]

The \(\alpha\)-helix forms when the backbone twists so that each residue's carbonyl oxygen hydrogen-bonds to the amide hydrogen four residues further along the chain, producing a right-handed coil with a characteristic period of \(3.6\) residues per turn. The \(\beta\)-sheet instead forms when extended chain segments (\(\beta\)-strands) lie alongside one another, hydrogen-bonding between adjacent strands' backbones rather than within a single stretch of chain; strands can run parallel or antiparallel to one another. A

\[ \text{Tertiary structure: the overall three-dimensional fold of a single polypeptide, stabilised chiefly by side-chain (R-group) interactions.} \]

Secondary-structure elements pack against one another and against less regular loop and turn regions, driven predominantly by burial of hydrophobic side chains away from water (protein-folding's hydrophobic effect), reinforced by hydrogen bonds and salt bridges between polar and charged side chains, van der Waals packing in the buried core, and, in some proteins, covalent disulfide bonds between cysteine residues. A

\[ \text{Quaternary structure: the non-covalent association of two or more independently folded polypeptide subunits into one functional complex.} \]

Not every protein has quaternary structure — it applies only to multi-subunit proteins, where each subunit first achieves its own tertiary fold and subunits then associate via the same categories of non-covalent interaction (and occasionally covalent disulfide links) responsible for tertiary structure, but now acting between chains rather than within one. Haemoglobin's four-subunit assembly (oxygen-dissociation-curve) is a standard example. A

\[ \text{A disruption at a given level propagates to every level above it, but not necessarily to levels below it.} \]

Because each level is built on the one below (Hypotheses), a change to primary structure (a mutation) can in principle alter secondary, tertiary, and quaternary structure; but altering conditions that disrupt tertiary structure (heat, extreme pH, denaturants) need not break the covalent peptide bonds of primary structure, which is why denaturation is generally reversible in principle (protein-folding) even though it can look, functionally, like total destruction of the protein. A

Result

\[ \text{Primary (sequence)}\ \to\ \text{Secondary (}\alpha\text{-helix, }\beta\text{-sheet)}\ \to\ \text{Tertiary (3D fold)}\ \to\ \text{Quaternary (multi-subunit assembly)} \]

Reading. Protein structure is organised into four nested levels, each built from and dependent on the one before it, with each level's stabilising forces shifting from purely covalent (primary) to backbone hydrogen bonding (secondary) to side-chain interactions (tertiary and quaternary).

Scope. Every protein has primary and (generically) some secondary structure; single-chain proteins lack quaternary structure by definition, and some short or intrinsically disordered regions of a chain lack stable, well-defined tertiary structure at all under physiological conditions.

Corollaries & converses

protein-folding supplies the thermodynamic mechanism (free-energy minimisation, driven substantially by the hydrophobic effect) that selects the specific tertiary structure this result describes only geometrically; the two results are complementary, structure and mechanism, rather than competing accounts.
enzyme-lower-activation-energy's active site is a tertiary-structure feature: residues that are far apart in the primary sequence are frequently brought close together in three-dimensional space once the chain is folded, which is precisely how a small number of catalytic residues can be positioned with the geometric precision an active site requires.
Converse: observing a specific, reproducible secondary-structure pattern (a run of backbone hydrogen bonds with a \(3.6\)-residue period, Step 2) is itself sufficient evidence to infer an \(\alpha\)-helix is present, without needing to examine side-chain identity at all — secondary structure is a backbone-level, sequence-independent geometric signature.

Fails without

Drop the level-dependency hypothesis and treat tertiary structure as though it could form independently of secondary structure: since tertiary structure is defined as the three-dimensional packing of secondary-structure elements (and connecting loops), there is no tertiary fold to speak of without first having some backbone (helix, sheet, or defined loop) geometry to pack — the hierarchy is not merely a naming convention but reflects a genuine physical dependency.
Assume side-chain (R-group) interactions, rather than backbone hydrogen bonding, stabilise secondary structure (second Hypothesis): this fails to explain why the \(\alpha\)-helix and \(\beta\)-sheet are found, essentially unchanged in geometry, across proteins built from wildly different amino-acid sequences — if side chains were the stabilising force, secondary structure would be sequence-specific and irregular in the way tertiary structure actually is, rather than a small set of standard, repeating backbone motifs common to nearly all proteins.

Common errors

Assuming quaternary structure is a universal fourth stage every protein passes through; many functional proteins are single-chain (monomeric) and simply have no quaternary structure at all (Result's Scope).
Confusing the peptide bond (Step 1, covalent, part of primary structure) with the hydrogen bonds that stabilise secondary and tertiary structure (Steps 2–3, non-covalent); only the former survives most denaturing conditions.
Assuming denaturation (loss of higher-order structure) necessarily breaks the primary sequence; in most cases only non-covalent interactions (and, if present, disulfide bonds) are disrupted, leaving the covalent peptide backbone (and hence primary structure) intact, which is why refolding can restore full activity (protein-folding, Anfinsen's experiment).
Attributing the \(\alpha\)-helix's stability mainly to side-chain packing rather than to the regular backbone hydrogen-bonding pattern that actually defines it (Fails without, second bullet).

Discussion

Linus Pauling and Robert Corey predicted the precise geometry of the \(\alpha\)-helix and \(\beta\)-sheet in 1951, working from careful modelling of peptide-bond bond lengths and angles rather than from a solved protein structure, before any full protein structure had been experimentally determined; both predictions were subsequently confirmed by x-ray-crystallography once such structures became available, John Kendrew's myoglobin structure (1958) being the first. The four-level hierarchy (primary/secondary/tertiary/quaternary) itself was proposed by Kaj Ulrik Linderstrøm-Lang in the early 1950s as an organising framework for exactly this kind of emerging structural data.

Not every functional region of every protein adopts a single, stable tertiary structure. Intrinsically disordered regions — stretches of sequence that remain conformationally flexible even under native, physiological conditions — are now recognised as functionally important in their own right, particularly in signalling and regulatory proteins, where flexibility itself (rather than a single fixed fold) can be the functionally relevant property, complicating any simple assumption that every protein sequence specifies exactly one rigid tertiary structure.

Common misconception: that primary structure alone (the amino-acid sequence) is "the" structure of a protein in any functionally meaningful sense. Sequence is necessary but not sufficient information for function: two very different sequences can converge on similar tertiary folds and similar functions, while, conversely, disrupting higher-order structure without changing the sequence at all (denaturation) is usually enough to abolish a protein's biological activity completely.

Worked examples

\[ \text{Sickle-cell haemoglobin: a single glutamate}\to\text{valine substitution at position 6 of the }\beta\text{-globin chain.} \]

This is a change to primary structure alone (Step 1): one residue, out of \(146\) in the \(\beta\)-chain, is substituted. The substituted valine is hydrophobic where glutamate was charged; because the mutation lies at a site exposed on the folded protein's surface, it creates a hydrophobic patch that, specifically when haemoglobin is deoxygenated, allows haemoglobin molecules to polymerise into rigid fibres — a quaternary-level (Step 4) consequence triggered entirely by a primary-structure change. A

\[ \text{One primary-structure substitution} \to \text{altered surface chemistry} \to \text{abnormal quaternary-level polymerisation} \]

Reading. A single change at the lowest level of the hierarchy can, depending on where in the folded structure it falls, propagate all the way to the highest level, exactly as Step 5 predicts.

Scope. Not every substitution has this severity; a substitution at a solvent-exposed, structurally unimportant site (Fails without's logic, applied in reverse) frequently has little to no detectable effect on tertiary or quaternary structure.

Further example. A researcher is studying the stability of a small, single-domain protein (molecular weight \(14\,000\ \text{g mol}^{-1}\)) in a dilute solution. The protein exists in an equilibrium between a native, folded state (N) and an unfolded, random-coil state (U). At \(25^\circ\text{C}\) (\(298\ \text{K}\)), the fraction of unfolded protein in the solution is measured to be \(0.01\) (1%). Using the standard Gibbs free energy equation, calculate the standard free energy of unfolding, \(\Delta G^\circ_{\text{unfolding}}\), for this protein. Assume the gas constant \(R = 8.314\ \text{J K}^{-1}\text{mol}^{-1}\).

\[ \text{Fraction unfolded } (f_U) = 0.01 \implies \text{Fraction native } (f_N) = 1 - 0.01 = 0.99 \]

The total population is the sum of the native and unfolded states. Since the fraction unfolded is given as 1%, the remaining 99% must be in the native, folded conformation. A

\[ K_{\text{eq}} = \frac{[\text{U}]}{[\text{N}]} = \frac{0.01}{0.99} \approx 0.010101 \]

The equilibrium constant for the unfolding reaction (\(\text{N} \rightleftharpoons \text{U}\)) is the ratio of the concentration (or fraction) of the product (unfolded) to the reactant (native). A

\[ \Delta G^\circ = -RT \ln(K_{\text{eq}}) \]

This is the standard thermodynamic relationship linking the standard Gibbs free energy change to the equilibrium constant at a given temperature \(T\). A

\[ \Delta G^\circ = -(8.314\ \text{J K}^{-1}\text{mol}^{-1})(298\ \text{K}) \ln(0.010101) \]

Substituting the known values for \(R\), \(T\), and \(K_{\text{eq}}\) into the equation. A

\[ \ln(0.010101) \approx -4.5951 \]

Calculating the natural logarithm of the equilibrium constant. A

\[ \Delta G^\circ = -(2477.572) \times (-4.5951) \approx 11385\ \text{J mol}^{-1} = 11.4\ \text{kJ mol}^{-1} \]

Multiplying the terms yields the free energy in Joules per mole, which is then converted to kilojoules per mole for standard biological reporting. A

\[ \Delta G^\circ_{\text{unfolding}} \approx 11.4\ \text{kJ mol}^{-1} \]

Reading. A positive \(\Delta G^\circ_{\text{unfolding}}\) indicates that the native state is thermodynamically more stable than the unfolded state under these conditions. The magnitude (\(11.4\ \text{kJ mol}^{-1}\)) is typical for small, stable proteins, reflecting the balance of stabilizing interactions (hydrophobic effect, hydrogen bonds) versus the entropic cost of folding.

Scope. This calculation assumes ideal solution behavior and that the two-state model (N \(\rightleftharpoons\) U) is a valid approximation for this small domain. It does not account for the presence of intermediate states or the effects of denaturants, which would alter the apparent \(K_{\text{eq}}\).

Problems

A mutation substitutes one hydrophobic residue on a protein's fully solvent-exposed surface for a different hydrophobic residue of similar size. Predict, using the hierarchy, whether this substitution is more or less likely to disrupt tertiary structure than the sickle-cell substitution in Worked Example 1, and explain why.
Solution
Considerably less likely to disrupt structure. The sickle-cell substitution introduced a hydrophobic residue at a site that had been charged and solvent-exposed, creating a novel sticky patch (Worked Example 1). A hydrophobic-to-hydrophobic substitution at an already solvent-exposed site changes little about the local chemistry that site presents to the surrounding water or to other protein surfaces, and does not affect the buried hydrophobic core packing that stabilises tertiary structure (Step 3), so it is expected to be comparatively well tolerated.
Explain, using Step 2, why the period of the \(\alpha\)-helix (\(3.6\) residues per turn) being a non-integer number is functionally significant, rather than merely an odd numerical curiosity.
Solution
Because \(3.6\) is not a whole number, side chains projecting outward from successive turns of the helix are not stacked directly above one another but are instead offset around the helix's circumference, spreading side chains from the helix at a range of rotational positions. This allows an \(\alpha\)-helix to present, for example, a hydrophobic face on one side and a hydrophilic face on the other along its length (an amphipathic helix), a geometric consequence of the specific non-integer periodicity that a helix with an exact integer period of residues per turn would not produce.
A protein is treated with a reagent that reduces (breaks) all of its disulfide bonds, but no denaturant is applied and the protein's overall fold is otherwise undisturbed. Which level(s) of structure, if any, are directly affected? Explain.
Solution
Disulfide bonds are one of the side-chain (R-group) interactions stabilising tertiary structure (Step 3), so breaking them directly affects tertiary structure (and, for proteins where interchain disulfides link separate subunits, potentially quaternary structure, Step 4). Primary structure (Step 1, the peptide backbone) and the backbone hydrogen bonding underlying secondary structure (Step 2) are unaffected by disulfide reduction alone, since disulfides are a distinct, side-chain-based covalent interaction, not part of either of those two levels.
A protein has a molecular weight of \(50\,000\ \text{g mol}^{-1}\). Assuming an average residue mass of \(110\ \text{g mol}^{-1}\), estimate the number of amino acid residues in the polypeptide chain. If the protein is entirely composed of \(\alpha\)-helices, and each turn of an \(\alpha\)-helix contains \(3.6\) residues, calculate the total number of helical turns.
Solution
First, calculate the number of residues: \[ \text{Number of residues} = \frac{\text{Molecular Weight}}{\text{Average Residue Mass}} = \frac{50\,000}{110} \approx 454.5 \] Since the number of residues must be an integer, we round to the nearest whole number, which is \(455\) residues. Next, calculate the number of helical turns: \[ \text{Number of turns} = \frac{\text{Total Residues}}{\text{Residues per Turn}} = \frac{455}{3.6} \approx 126.4 \] Thus, the protein would contain approximately \(126\) full turns of \(\alpha\)-helix.
A buried leucine in a protein core is mutated to alanine. Leucine and alanine residues occupy about 167 and 89 \(\text{\AA}^3\) respectively. Estimate the cavity created, and explain why such “large-to-small” mutations typically destabilise a protein by several kJ mol\(^{-1}\).
Solution
The cavity is about \(167 - 89 = 78\ \text{\AA}^3\), roughly the volume of three methylene groups. The folded core is packed as tightly as a crystal; removing side-chain atoms leaves a cavity that the rest of the structure can only partly close. The protein loses the van der Waals contacts those atoms made and some of the hydrophobic driving force, while the unfolded state is hardly affected. Measured leucine-to-alanine mutations in cores (for example in T4 lysozyme) destabilise by roughly 5–20 kJ mol\(^{-1}\), increasing with the size of the cavity, which is comparable to the whole stability of many small proteins.