The passive elasticity of muscle is largely governed by the I-band part of the giant muscle protein titin1, 2, 3, 4, a complex molecular spring composed of a series of individually folded immunoglobulin-like domains as well as largely unstructured unique sequences5. These mechanical elements have distinct mechanical properties, and when combined, they provide the desired passive elastic properties of muscle6, 7, 8, 9, 10, 11, which are a unique combination of strength, extensibility and resilience. Single-molecule atomic force microscopy (AFM) studies demonstrated that the macroscopic behaviour of titin in intact myofibrils can be reconstituted by combining the mechanical properties of these mechanical elements measured at the single-molecule level8. Here we report artificial elastomeric proteins that mimic the molecular architecture of titin through the combination of well-characterized protein domains GB112 and resilin13. We show that these artificial elastomeric proteins can be photochemically crosslinked and cast into solid biomaterials. These biomaterials behave as rubber-like materials showing high resilience at low strain and as shock-absorber-like materials at high strain by effectively dissipating energy. These properties are comparable to the passive elastic properties of muscles within the physiological range of sarcomere length14 and so these materials represent a new muscle-mimetic biomaterial. The mechanical properties of these biomaterials can be fine-tuned by adjusting the composition of the elastomeric proteins, providing the opportunity to develop biomaterials that are mimetic of different types of muscles. We anticipate that these biomaterials will find applications in tissue engineering15 as scaffold and matrix for artificial muscles.
Figures at a glance
Figure 1: Force–extension curves of two polyproteins. 
a, (G–R)4. b, GRG5RG4R. The force peaks, characterized by a ΔLc of ~18 nm and an unfolding force of ~180 pN, result from the mechanical unfolding of GB1 domains. Stretching resilins does not result in any unfolding force peaks; instead we see a featureless spacer of length L0. The notable difference between the force–extension curves of (G–R)4 and GRG5RG4R is the shorter featureless spacer of GRG5RG4R, which is due to fewer resilin repeats in GRG5RG4R. Grey lines correspond to the worm-like chain model fits to the experimental data.
Figure 2: Mechanical properties of (G–R)4 and GRG5RG4R-based biomaterials. 
a, Photographs of moulded rings built from (G–R)4 (left, intact) and GRG5RG4R (right, after being loaded to failure in tensile test) under white light (middle panel) and ultraviolet illumination (top panel). b, c, Representative stress–strain curves of (G–R)4 (b) and GRG5RG4R (c) measured in PBS. For clarity, stress–strain curves are offset relative to one another. Final strains are shown on the curves. Insets show the superposition of the stress–strain curves at different strains. d, Resilience of GB1–resilin-based biomaterials decreases with the increase of strain. In contrast, biomaterials constructed from resilin do not show any appreciable hysteresis (data taken from ref. 13). e, GRG5RG4R-based biomaterials can recover hysteresis under residual stress. During stretching–relaxation experiments, when the biomaterial is partially relaxed to a strain above 35%, no recovery of hysteresis was observed. When the biomaterial was relaxed to below 35% strain, we started to observe partial recovery. The degree of recovery increased with the decrease of residual stress. For clarity, the initial stretching trace is coloured blue. The inset shows the experimental protocol of the partial relaxation experiments. The pulling speed used in the experiments was 25 mm min-1. Error bars indicate standard deviation of the data.
Figure 3: GB1–resilin-based biomaterials exhibit pronounced stress relaxation behaviours. 
a, Representative stress-relaxation curves of GRG5RG4R at varying strains. b, Relaxation rates of GRG5RG4R-based biomaterials depend upon the initial stress. The relaxation rates were obtained by fitting the stress-relaxation to a double-exponential equation: σ(t) = σ0 + A1exp(-k1t) + A2exp(-k2t), where σ(t) is the stress at time t, σ0 is the offset, A1 and A2 are decay amplitudes and k1 (filled squares) and k2 (open triangles) are relaxation rates. Error bars indicate fitting errors.
Figure 4: The macroscopic mechanical properties of GB1–resilin-based biomaterials can be fine-tuned by controlling the nanomechanical properties of the constituting elastomeric proteins at the single-molecule level. 
a, Force–extension curves of single GRG5RG4R molecules in PBS and in 4 M urea. The long featureless spacers observed in force-extension curves of GRG5RG4R in 4 M urea largely correspond to the stretching of mechanically labile, unfolded GB1 domains. The unfolding force of GB1 domains that remain folded in 4 M urea is also significantly reduced. Grey lines are WLC fits. b, Young’s modulus of GRG5RG4R-based biomaterial can be modulated by chemical denaturant urea. The conversion of folded GB1 domains into unfolded sequence leads to the dramatic decrease in Young’s modulus of the biomaterials in a urea-concentration-dependent manner. Error bars indicate standard deviation of the data.
