Microbial production of megadalton titin yields fibers with advantageous mechanical properties

Abstract

Manmade high-performance polymers are typically non-biodegradable and derived from petroleum feedstock through energy intensive processes involving toxic solvents and byproducts. While engineered microbes have been used for renewable production of many small molecules, direct microbial synthesis of high-performance polymeric materials remains a major challenge. Here we engineer microbial production of megadalton muscle titin polymers yielding high-performance fibers that not only recapture highly desirable properties of natural titin (i.e., high damping capacity and mechanical recovery) but also exhibit high strength, toughness, and damping energy - outperforming many synthetic and natural polymers. Structural analyses and molecular modeling suggest these properties derive from unique inter-chain crystallization of folded immunoglobulin-like domains that resists inter-chain slippage while permitting intra-chain unfolding. These fibers have potential applications in areas from biomedicine to textiles, and the developed approach, coupled with the structure-function insights, promises to accelerate further innovation in microbial production of high-performance materials.

Fig. 1. The multi-scale structure of muscle and schematic representation of SI-based polymerization of the titin protein in E. coli.

a Muscle tissue (1) is composed of specialized, elongated (> 1 cm) cells called muscle fibers (2). Muscle fibers are packed with proteinaceous myofibrils (3) that span the entire length of the cell. Myofibrils are composed of repeating stacks of chemically controllable, contractile elements called sarcomeres. (4) Sarcomeres are composed primarily of three proteins: actin, myosin, and titin. Titin spans half the length of the sarcomere, anchoring the opposing Z- and M-lines, and consists of hundreds of repeating immunoglobulin (Ig) domains that are integral to the passive strength (i.e., resistance to deformation without energy input), damping capacity, and mechanical recovery of the macroscopic muscle fiber. The images of muscle fiber and myofibril are modified from the OpenStax Anatomy and Physiology Textbook Version 8.25, Published May 18, 2016 (OpenStax, CC BY 4.0 https://creativecommons.org/licenses/by/4.0) and the image of the sarcomere is modified from Giganti, D., Yan, K., Badilla, C.L. et al. Disulfide isomerization reactions in titin immunoglobulin domains enable a mode of protein elasticity. Nat Commun 9, 185 (2018). 10.1038/s41467-017-02528-7, (David Giganti, Kevin Yan, Carmen L. Badilla, Julio M. Fernandez & Jorge Alegre-Cebollada, CC BY 4.0 https://creativecommons.org/licenses/by/4.0). b To facilitate the production of UHMW titin polymer in vivo, a relatively small, genetically stable, 41.3 kDa rabbit soleus titin protein-coding sequence (4 Ig; purple) was flanked by complimentary SIs, gp41-1^C (Int^C; green) and gp41-1^N (Int^N; blue) (i). DNA sequence-recoded Int^C-4Ig-Int^N was produced in an engineered E. coli host under the control of inducible promoter P_LacO-1 (ii). The SI-flanked monomer protein was overexpressed in bioreactor cultures (iii) and polymerized intracellularly through successive rounds of SI-catalyzed intermolecular ligation to produce UHMW titin (iv). Purification and processing yielded microbially produced titin fibers that recapture the damping capacity and mechanical recovery of muscle along with high strength and toughness (v). Titin 4 Ig and SI structures were acquired using PBD accession numbers 3B43 and 6QAZ, respectively.

Fig. 2. Structural analyses of microbially produced UHMW titin protein and processed monofilament fibers.

a Circular dichroism spectrum for purified titin polymer in water. The inlaid pie graph indicates the results of spectral deconvolution by the BeStSel program. b STEM image of purified titin polymer. The scale bar is 50 nm. This image is representative of dozens of similar polymer molecules observed using STEM, 42 of which were selected for diameter measurements. c SEM image of a fracture cross-section of the spun titin fiber. The scale bar is 10 µm. This image is representative of 12 fibers that were observed using SEM. d Image of a textile net woven from the spun titin fibers. The white scale bar is 0.5 cm. Inset is a light microscopy image of an individual titin fiber, representative of all of the images taken for diameter measurements. The black scale bar is 40 µm. e FTIR analysis of as-spun and post-spin drawn UHMW titin polymer fibers. Averages of normalized spectra for each condition were overlaid. f Deconvolved β-sheet content of titin polymer fibers. For each fiber state, percentages were averaged for FTIR spectra acquired from three separate fibers. Error bars are the standard deviation of the three peak area calculations (see Methods, Supplementary Fig. 6). g Raman spectra of post-spin drawn titin polymer fibers oriented perpendicular (I_Y; pink line) or parallel (I_X; blue line) to the polarization of the incident laser. Spectra shown are the average of spectra acquired from three separate fibers. Standard deviations of the three measurements at each Raman shift are shown as black bars. The average ratio of the amide I peak (1670 cm⁻¹) intensity at 0° to that at 90° is shown above the spectrum as a measure of orientation sensitivity (see Methods, Supplementary Fig. 7). h–k Synchrotron-based wide-angle X-ray diffraction analysis of spun titin polymer fibers. Insets show the area selected for radial (h, i) or azimuthal (j, k) integration. h 1D radial intensity profile along the equator, with Gaussian fits for the (120) equatorial peak (dotted red), (200) equatorial peak (dotted blue), and two amorphous components (dotted gray). i 1D radial intensity profile along the meridian, with Gaussian fits for the (120) meridian peak (dotted red), (200) meridian peak (dotted blue), (002) peak (dotted pink), and two amorphous components (dotted gray). j, k Intensity as a function of the azimuthal angle at the radial position of the equatorial (120) peak (j) and (200) peak (k). The peaks are fitted as sums of two Gaussians, corresponding to crystalline (narrow) and amorphous (broad) distributions. In figure (j), small subsidiary peaks due to residual intensity from the (201) reflections (dotted purple) were treated as individual Gaussian functions. Source data are provided as a Source Data file.

Fig. 3. Mechanical testing of fibers spun from microbially produced UHMW titin reveals high toughness, damping capacity, and mechanical recovery reminiscent of natural muscle fibers.

a Stress-strain curves from tensile tests of 14 microbially produced UHMW titin fibers (polymer; purple) and the low MW, 4Ig titin (monomer; gold). b Box-plots displaying toughness measures extracted from the stress-strain curves for polymer (purple) and monomer (gold) fibers (n = 14 fibers for each protein; horizontal lines denote, from top to bottom, upper fence, Q3, median, Q1, and lower fence; × denotes mean; other data indicated with circles). ***P = 1.6 × 10⁻¹⁹, unpaired two-tailed t-test. c Toughness of microbially produced (blue), natural (green), and man-made (gray) materials compared to that of the UHMW titin fibers produced in this work (red). d Loading/unloading curves for microbially produced UHMW titin fibers acquired at increasing strains from 0.6–30%. e Average calculated damping capacity (blue curve) and damping energy (pink curve) at each strain tested in (d). Error bars are the standard deviation of the three fiber samples tested at each strain. f Stress-strain curves for microbially produced UHMW titin fibers subjected to 11 consecutive loading/unloading cycles with one minute of humid (95% RH) air treatment between cycles. The stress-strain curve of the first round is colored red. Following cycles use other colors. g Average calculated damping capacity (blue curve) and damping energy (pink curve) over consecutive cycles with humid air treatment between cycles. Error bars are the standard deviation of the values measured at each cycle number for the three fiber samples that were tested. Source data are provided as a Source Data file.

Fig. 4. Molecular dynamics simulation of uniaxial tensile testing of a model titin fiber.

a Representative uniaxial tensile stress-strain curves of the model titin fiber. b Snapshots of the molecular dynamics simulation of the titin fiber under tensile deformation. c, d Normalized atomic stress along the y-axis during extension of the titin fiber from 0 to 80% strain. Selected Ig-like domains are shown in dashed boxes. The Ig-like domain in the red and yellow boxes in (b) are shown in (c, d), respectively. e Average changes in intra- and inter-fibril (pink and blue, respectively) non-bonded energies (including Van der Waals, electrostatic, and hydrogen bonds) over the course of the simulation. Error bars are the standard deviation of three trials. f Average total number of intra-fibril backbone-backbone hydrogen bonds in the two selected Ig-like domains in (b–d) over the course of the tensile test. Error bars (shown as red or yellow bars) are the standard deviation of three trials. Source data are provided as a Source Data file.