Fibrous Proteins

Fibrous protein chemistry is now central to biomaterials engineering: recombinant spider silk proteins are produced in bacteria, goats, and silkworms for high-strength, biodegradable sutures, ballistic-resistant textiles, and drug-delivery microspheres, exploiting the same crystalline/amorphous β-sheet balance quantified in Example 3. Collagen — as gelatin, as injectable dermal filler, and as tissue-engineering scaffolds — is used clinically at enormous scale, and understanding Gly-X-Y register and hydroxyproline stabilization directly informs why collagen-mimetic peptides need proline/hydroxyproline-rich flanking sequences to fold at all. Mutations disrupting the Gly-X-Y repeat cause the heritable brittle-bone disease osteogenesis imperfecta, and mutations in keratin coiled-coil rod domains (particularly at conserved boundary residues that anchor helix-helix packing) cause epidermolysis bullosa simplex, a blistering skin disorder — both diseases are direct, structurally explicable consequences of the mechanisms derived above, and both are active targets for antisense and gene-therapy approaches. In industry, understanding disulfide crosslink chemistry in keratin underlies the entire hair-styling and wool-processing industries (permanent waving, hair straightening), while wound-healing research exploits collagen's self-assembly and crosslinking (lysyl oxidase-mediated pyridinoline formation) to engineer scar-reducing dressings. More broadly, the sequence-to-mechanics logic of fibrous proteins — periodic repeats dictating quaternary assembly and hence bulk material properties — is a template actively used in de novo protein design, where synthetic heptad repeats are engineered to build custom nanoscale fibers, hydrogels, and self-assembling biomaterials, extending principles first read off Astbury's X-ray films of hair and horn.

Ask the tutor: How does the α-helix-to-β-sheet transition under mechanical stretching actually happen at the hydrogen-bond level in a keratin fiber?

Fibrous proteins are built from three recurring architectural solutions, each a direct geometric consequence of a specific sequence periodicity superimposed on standard secondary structure.

The coiled coil (keratins, myosin, tropomyosin, fibrinogen). An ordinary α-helix has 3.6 residues per turn and a helical rise of 1.5 Å per residue. If a sequence carries a seven-residue (heptad) repeat, denoted positions a-b-c-d-e-f-g, with hydrophobic residues preferentially at a and d, those two positions trace out a stripe that winds around the helix slightly slower than the helix itself completes a turn (7 residues span barely less than 2 turns of an ideal helix). Because 3.6 residues/turn does not divide evenly into a 7-residue repeat, the hydrophobic stripe would drift around the helix cylinder if the helix stayed perfectly straight — so instead the whole helix bends slightly and wraps around a partner helix, restoring a straight hydrophobic seam. The result is two right-handed α-helices wound around each other with a slow left-handed supertwist: a coiled coil. This is precisely the structural logic of the knobs-into-holes packing described by Francis Crick in 1953 — a side chain (knob) from one helix nestles into a diamond-shaped pocket (hole) formed by four side chains on the partner helix.

The collagen triple helix. Each collagen chain is a left-handed helix with an unusual geometry — about 3 residues per turn (much tighter than the α-helix) — stabilized not by intrachain H-bonds but by steric packing of a repeating Gly-X-Y tripeptide, where X is frequently proline and Y is frequently 4-hydroxyproline (Hyp). Three such chains wind around a common axis with a right-handed supertwist, opposite in handedness to each individual chain, giving mechanical interlocking analogous to a three-strand rope. Glycine — the only amino acid with a side chain small enough (a single H atom) — must sit at every third position because that position points into the crowded center of the triple helix; any larger residue would clash. Interchain hydrogen bonds run from the backbone NH of glycine in one chain to the backbone C=O of the X residue in a neighboring chain.

Stacked β-sheets (silk fibroin, β-keratins). Silk fibroin's crystalline domains are built from the repeat (Gly-Ala-Gly-Ala-Gly-Ser)ₙ. In the antiparallel β-sheet each chain zigzags with alanine methyl groups and glycine hydrogens alternating on the two faces of the sheet; sheets stack via these small side chains with extremely tight interplanar spacing (~3.5 Å on the glycine face, ~5.7 Å on the alanine face), maximizing van der Waals contact and H-bonding density. This gives silk's crystalline β-sheet regions their extraordinary stiffness, while amorphous, glycine-poor linker segments between crystallites provide the extensibility that keeps the fiber from being brittle.

The common theme across all three motifs is repeat-driven propagation: a short sequence period (7 residues for keratin, 3 for collagen, 2 for silk's Gly-Ala core) is repeated dozens to hundreds of times, so that a single local packing rule — hydrophobic seam, glycine-in-the-core, small-side-chain sheet stacking — is enforced continuously along a fiber that can run for microns. There is no single tertiary fold to speak of; the "fold" is the fiber.

The mechanical logic of fibrous proteins can be expressed in a small number of quantitative relationships linking sequence period, helix geometry, and mechanical stiffness.

1. The heptad repeat and coiled-coil supercoiling. For an α-helix with helical twist $\theta_0$ (degrees per residue) and a heptad period of 7 residues, the underwinding per heptad relative to an integral number of turns determines the coiled-coil superhelical pitch:

where: $3.6$ is the number of residues per turn of an unconstrained α-helix.

where: $7\theta_0$ is the total twist accumulated over one heptad if the helix were straight; $2 \times 360^\circ$ is the twist of exactly two full turns (the nearest integral match to a 7-residue span). A perfectly periodic hydrophobic seam at a and d requires the seam to return to the same azimuthal position every 7 residues, i.e., every 2 turns exactly — but the helix naturally overwinds by 20° per heptad relative to that target, so the whole helix must supercoil by an equal and opposite amount to compensate, producing the left-handed superhelix (opposite in sense to the right-handed α-helix backbone).

2. Collagen triple-helix thermal stability and the Hyp effect. The melting temperature of the triple helix increases roughly linearly with hydroxyproline content because the 4R-hydroxyl group locks the pyrrolidine ring in a C-γ-exo pucker that pre-organizes the backbone φ,ψ angles required for the triple helix, and stabilizes an ordered, bridging water network:

where: $T_{m,0}$ is the melting temperature of a Gly-Pro-Pro control triple helix (no hydroxylation, roughly 24–30 °C for short synthetic peptides), $[\text{Hyp}]$ is the mole fraction of Y-position hydroxyproline, and $k_{\text{Hyp}}$ is an empirical stabilization constant (tens of °C per full substitution in model peptides). Physiological collagen, fully hydroxylated, denatures (unwinds to random coil, "gelatin") near 37–40 °C in vivo — just above body temperature, consistent with a fold poised for regulated turnover.

3. Force-extension behavior of an entropic vs. enthalpic fiber. A random-coil, rubber-like protein (e.g., elastin) obeys entropy-dominated elasticity, while a crystalline β-sheet fiber (silk) or a triple helix (collagen) is enthalpy-dominated, stretching by bond distortion rather than conformational entropy loss. The freely-jointed chain gives the low-force (small extension $x$ relative to contour length $L$) force law:

where: $f$ is applied force, $k_BT$ is thermal energy, $L$ is contour length, $\ell_p$ is the persistence length, and $x$ is end-to-end extension — this is the regime relevant to elastin and unfolded silk linker domains, giving low-modulus, highly extensible behavior, in contrast to the near-linear, high-modulus, low-extensibility stress-strain curve of crystalline collagen or silk β-sheet, where stiffness comes from stretching covalent bonds and H-bonds rather than reducing conformational entropy.

Example 1 (Foundational): Rise and residues-per-turn bookkeeping in an α-keratin coiled-coil rod. The central rod domain of a human hair keratin monomer (e.g., keratin 1/10 heterodimer) is approximately 310 residues long. Treating it as an ideal coiled-coil α-helix with axial rise per residue of 1.485 Å (slightly compressed from the free α-helix value of 1.5 Å, due to the superhelical twist — a well-documented ~1% axial compaction), find the contour length of the rod domain and compare to the ~46 nm rod length measured by electron microscopy and X-ray fiber diffraction of keratin intermediate filaments. Contour length:

where $N$ is residue count and $d$ is the compacted axial rise per residue. This matches the experimentally measured ~46–48 nm rod length essentially exactly, confirming both the residue count (from sequence) and the compaction factor (from fiber diffraction) are mutually consistent. As a limiting-case check: if we instead used the uncompacted free-helix rise of 1.5 Å (i.e., ignoring the coiled-coil supertwist entirely, the "as-if-straight" limit), we would predict

only about 1% longer — showing the supercoiling correction is a small but measurable effect on total contour length, exactly the ~1% axial shortening reported for coiled-coils relative to isolated α-helices, and not the dominant length-determining factor (that role belongs to residue count).

Example 2 (Applied): Estimating tensile stress in an Achilles tendon fiber bundle from collagen packing density. The Achilles tendon is essentially parallel bundles of type I collagen fibrils. A typical fibril has a diameter of ~100 nm, and tendon collagen fibrils are packed at roughly 60–70% volume fraction within the tendon cross-section (the rest being ground substance and water). Human Achilles tendons rupture at forces of roughly 3500–5000 N in vigorous activity (documented cadaveric and in vivo tensiometry studies), and a typical Achilles tendon cross-sectional area is about 60 mm² (0.6 cm²). Estimate the ultimate tensile stress the tendon material bears, and compare to the known ultimate tensile strength of collagenous tendon (commonly cited ~50–100 MPa). Using the midpoint failure force of 4250 N and cross-sectional area of 60 mm² = $60\times10^{-6}\ \text{m}^2$:

where $\sigma$ is engineering stress, $F$ is failure force, $A$ is cross-sectional area. This value (70.8 MPa) falls squarely inside the experimentally cited 50–100 MPa range for tendon ultimate tensile strength, showing that a simple force/area calculation using independently measured rupture force and cross-sectional area reproduces the accepted material property — a useful cross-check connecting whole-tendon mechanical testing to the fibril-level collagen chemistry described above. As a limiting case: if the tendon cross-section were doubled (120 mm², e.g., a larger individual with a thicker Achilles) at the same rupture force, stress at failure would halve to ~35.4 MPa, below the typical failure threshold — correctly predicting that thicker tendons should be more resistant to a given absolute force, matching the clinical observation that tendon cross-sectional area (not just collagen "quality") is protective against rupture.

Example 3 (Exam/competition level): Silk fibroin extensibility from crystalline/amorphous fraction — a two-component (Takayanagi-type) model. Bombyx mori silk fibroin fiber can be modeled mechanically as crystalline β-sheet domains (Young's modulus $E_c \approx 20$ GPa, essentially inextensible under normal working loads) connected in series with amorphous, more flexible linker domains (effective modulus $E_a \approx 1.5$ GPa, representing disordered Gly-rich and charged-residue segments). Suppose a silk fiber is 40% crystalline by volume fraction ($\phi_c = 0.40$) and the crystalline and amorphous regions are arranged as a series (isostress) composite along the fiber axis — a reasonable first approximation for a fiber loaded along its draw direction where crystallites and amorphous linkers alternate end-to-end. Estimate the overall fiber modulus, and the fraction of total extension contributed by the amorphous domains at a given applied stress. For a series (isostress) composite, the reciprocal rule of mixtures applies:

where: $\phi_c = 0.40$, $\phi_a = 1-\phi_c = 0.60$, $E_c = 20$ GPa, $E_a = 1.5$ GPa.

This is close to experimentally reported native silk fiber moduli (~5–12 GPa for dragline silk, somewhat higher than this simplified estimate because real silk fibers are not perfectly isostress in series — parallel load-sharing through hydrogen-bonded crosslinks between crystallites raises the effective modulus above the naive series bound — but the calculation correctly captures the qualitative rule that the soft phase dominates a series composite's compliance). Now find the fraction of total strain borne by the amorphous phase at fixed applied stress $\sigma$. In an isostress composite, both phases feel the same $\sigma$, so each phase's strain is $\varepsilon_i = \sigma/E_i$, and total elongation is the length-weighted sum. Assuming crystalline and amorphous regions occupy fractional lengths equal to their volume fractions ($\phi_c$, $\phi_a$) along the fiber axis:

The amorphous phase's contribution to total strain is $\phi_a \sigma/E_a = 0.400\,\sigma$ GPa$^{-1}$, so its fractional share of total extension is:

Despite comprising only 60% of the fiber's volume, the amorphous domains account for over 95% of the total extension under load — quantitatively explaining why silk's extensibility (its toughness, i.e., energy absorbed before failure, which for dragline silk substantially exceeds Kevlar on a per-mass basis) is governed by the amorphous linker fraction, while its ultimate strength is governed by the crystalline β-sheet fraction. As a limiting case, if $\phi_c \to 1$ (a hypothetical all-crystalline silk with no amorphous linkers), $E_{\text{fiber}} \to E_c = 20$ GPa and the amorphous strain fraction correctly collapses to zero — recovering a stiff but brittle material, consistent with the observed embrittlement of silk fibers that are over-processed to maximize crystallinity at the expense of toughness.

Fibrous protein structure owes its modern mechanistic understanding chiefly to William Astbury, a Leeds-based X-ray crystallographer working in the 1930s who first distinguished the two fundamental "conformational states" of keratin fibers by X-ray diffraction: an unstretched, coiled form he called α-keratin, giving a diffraction pattern with a strong ~5.1 Å meridional reflection, and a stretched, extended form he called β-keratin, with a very different pattern consistent with an extended chain conformation. Astbury correctly inferred that stretching hair or wool converts one regular, periodic structure into another — years before Pauling, Corey, and Branson would derive the atomic-resolution α-helix model in 1951 using precise bond geometry and rigorously planar peptide units, finally explaining Astbury's 5.1 Å reflection as an artifact of a coiled-coil superhelix (the true helical repeat is ~1.5 Å per residue along the local helix axis, and the 5.4 Å "α-pattern" spacing Astbury measured reflects the superhelical pitch component, not a single uncoiled helix repeat — a subtlety that took another two decades, and Francis Crick's 1953 "knobs-into-holes" coiled-coil packing model, to fully resolve). Astbury's parallel X-ray studies of collagen (with Street, 1931) and silk fibroin similarly established the triple-helical and stacked-sheet repeat spacings that later crystallographic and fiber-diffraction work (Ramachandran and Kartha's 1954 collagen triple-helix model; Marsh, Corey, and Pauling's silk β-sheet model) refined into the atomic-detail pictures used throughout this article.