Biological fabrication of cellulose fibers with tailored properties

See allHide authors and affiliations

Science  15 Sep 2017:
Vol. 357, Issue 6356, pp. 1118-1122
DOI: 10.1126/science.aan5830

More than just a cotton shirt

Responsive or functional fabrics include coatings or secondary materials with properties such as changing color with temperature or generating electricity with movement. The challenge is that anything added to a fabric can get washed or worn away. Hence, Natalio et al. opted to build the functionality directly into cotton grown in vitro, through the addition of glucose modified at the C2 position to the culture medium. By this process, fibers can be made that naturally fluoresce or have magnetic properties, for instance.

Science, this issue p. 1118


Cotton is a promising basis for wearable smart textiles. Current approaches that rely on fiber coatings suffer from function loss during wear. We present an approach that allows biological incorporation of exogenous molecules into cotton fibers to tailor the material’s functionality. In vitro model cultures of upland cotton (Gossypium hirsutum) are incubated with 6-carboxyfluorescein–glucose and dysprosium–1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid–glucose, where the glucose moiety acts as a carrier capable of traveling from the vascular connection to the outermost cell layer of the ovule epidermis, becoming incorporated into the cellulose fibers. This yields fibers with unnatural properties such as fluorescence or magnetism. Combining biological systems with the appropriate molecular design offers numerous possibilities to grow functional composite materials and implements a material-farming concept.

Smart or responsive textiles and wearable technologies are fabrics that react to external stimuli, such as a change in temperature or humidity (1, 2). Many synthetic polymers can be adapted to achieve high performance and multifunctionality by coating fiber surfaces by means of chemical treatments (35). Such fibers can have multiple applications (614), but they suffer from wear. Furthermore, the preferences of end consumers are for natural fibers because of issues of sensation, skin irritation, smoothness, and weight. This makes biological materials a prime choice, spurring intense economic interest and research aimed at a comprehensive understanding of the physiology and biochemistry of cotton fiber development (15). The identification of the intermembrane cellulose/sucrose synthase complex, its role in promoting glucose polymerization into cellulosic chains (β-1,4-glucans) (16, 17), and insights into epidermal cell differentiation mechanics were foundational in establishing the current cellulose biosynthesis pathways and cotton in vitro growth models (1820).

We used upland cotton (Gossypium hirsutum) ovules as an in vitro model for biological fabrication of cotton fiber composites with fluorescent and magnetic properties. We synthesized two glucose derivatives containing (i) a sugar moiety (glucose) that acts as a carrier capable of traveling from the vascular connection to the outermost cell layer of the ovule epidermis and (ii) a functional molecule (either an optical tracer or magnetic complex).

We started by exploring the uptake selectivity of pigments that either were devoid of glucose (6-carboxyfluorescein, kermesic acid, and indigo) or were glucose derivatives (carminic acid) in the millimeter-size fertilized ovules (21) (Fig. 1, A and B). Carminic acid is a kermesic acid glycoside—i.e., kermesic acid is bound to the anomeric carbon (C1) of the glucose moiety. The pigments were added to ovule’s medium and kept under standard conditions (32°C, 5% CO2) for 20 days. For the pigments devoid of glucose, we observed neither an intracellular coloration of the epidermal layer (fig. S1, right column) nor coloration of the fibers (fig. S1, left column). For glucose derivatives, namely carminic acid, we found red coloration of the epidermal layer (fig. S2A) but not of the fibers (fig. S2, B and C). These results allowed us to infer that the glucose moiety has a role in reaching the outermost epidermal layer and that C1 is essential for polymerization of glucose units into fibers.

Fig. 1 Biological incorporation of fluorescent-tagged glucose (6CF-Glc) into cotton fibers.

(A) A G. hirsutum flower grown under hydroponic conditions. (B) Schematic representation of the development of the in vitro cotton culture model. BT, Beasley-Ting. (C) Sequence of microscopic images taken under “white” illumination and standard conditions (32°C, 5% CO2) at 0, 4, 8, 12, 15, and 20 days. (D) Corresponding sequence of images taken under UV illumination (wavelength, 365 nm). (E) A representative ovule after day 20 of incubation with 6CF-Glc, showing yellow coloration of the fibers. (F) Similar to (E), but under UV illumination (365 nm), showing fluorescent fibers. (G) Binocular image of the fibers under UV illumination (365 nm). (H) CLSM 3D image stack of isolated fibers displaying a fluorescent tubular morphology. (I) High-magnification CLSM image of a fiber, showing fluorescent features in the cytosol. (J) Schematic representation of the transport of 6CF-Glc through the embryo to the outermost layer of the epidermis, uptake and metabolism by the cellulose-producing cells, and integration into the cotton fibers.

These results suggested that molecules carrying a functionality and a glucose moiety with free C1 and C4 positions, such as 6-carboxyfluorescein–glucose (6CF-Glc), could be transported from the external medium into the outermost epidermal layer of the ovule, intracellularly metabolized, and biologically integrated into the cellulose fibers. Using peptide chemistry, we synthesized 6CF-Glc by reacting the free amino group located at C2 of the acetyl-protected glucosamine moiety with the carboxylate from 6-carboxyfluorescein (fig. S3). We incubated 6CF-Glc with fertilized ovules under standard conditions and imaged their development under white and ultraviolet (UV) light (365 nm) (fig. S4). Figure 1C (i to vi) shows a sequence of images acquired under white light after 0, 4, 8, 12, 15, and 20 days, which demonstrate that the fibers acquired a yellow coloration coincident with the fluorescent signal (movies S1 and S2). In comparison with the ovules grown in the absence of pigment, we did not observe morphological differences from a simple visual inspection (fig. S5). On day 20 of incubation, the harvested ovules showed a yellow coloration of fibers (Fig. 1E) and strong fluorescence (Fig. 1, F and G). Using diffuse reflectance UV–visible spectroscopy, we calculated the uptake to be 4.8 ± 0.2% on the basis of the band centered at 497 nm and the respective calibration curve (fig. S6).

Imaging under confocal laser scanning microscopy (CLSM; emission wavelength, 490 nm) revealed tubular fluorescent morphology in the three-dimensional (3D) reconstruction (Fig. 1H), and the presence of fluorescent features in the cytosol (Fig. 1I) confirmed our hypothesis that 6CF-Glc is transported through the ovule tissue, metabolized by the epidermal cellulose-producing cells, and integrated into the cotton fibers (Fig. 1J).

Structural characterization of the fluorescent fibers containing 6CF-Glc, termed FGIO fibers (Fig. 2A), dispelled any suspicion of surface contamination during incubation or handling. FGIO wide-angle x-ray scattering (WAXS) patterns showed only one broad peak (Fig. 2B, upper panel, red), whereas the raw fiber WAXS patterns attributed to the native cellulose I-β crystal structure showed overlapping (110) and (Embedded Image) peaks at scattering vector length (q) = 1.05 and 1.18 Å−1, as well as the (200) peak at 1.62 Å−1 (Fig. 2B, upper panel, black) (22, 23). The absence of these peaks in the FGIO WAXS signal indicates that 6CF-Glc induces structural amorphization, which is corroborated by the lack of any evident angular dependence of the pattern obtained by azimuthal integration of the FGIO WAXS peak (Fig. 2B, lower panel). For both fiber types, the azimuthal profiles obtained from 2D small-angle x-ray scattering (SAXS) patterns (fig. S7, A and B) showed clear preferred orientation (fig. S8, blue and red)—with the raw fibers showing a higher degree of orientation (0.24) than FGIO fibers (0.12)—indicating the presence of oriented fibrous nanostructures in the size range of a few nanometers. Influence on the SAXS signal from cracks developing during fiber drying was eliminated by analyzing radial cuts of SAXS patterns of the aligned fibers (fig. S9).

Fig. 2 Structural characterization of fibers after biological integration of 6CF-Glc.

(A) Microscopic view of the yellow cotton fibers. (B) WAXS spectra of raw (black) and FGIO (red) fibers, showing induced structural amorphization. (C) DSC of raw (black) and FGIO (red) fibers. FGIO fibers had a lower temperature of crystallization (297°C) than raw fibers (324°C), in agreement with WAXS. (D) FT-IR ATR spectra of raw (black) and FGIO (red) fibers, showing a structural modification. The inset shows detailed spectral differences in the range between 1100 and 1900 cm−1. (E and F) Spectral deconvolution of the infrared spectrum in the OH band region (3000 to 3600 cm−1) for raw (E) and FGIO (F) fibers, showing the FGIO bands shifted toward higher wave numbers and the two new additional bands, reflecting differences between inter- and intrachain H-bond networks. (G) Force-displacement curves of single raw (black) and FGIO (red) fibers obtained during tensile testing. (H) Sequence of images from single-fiber tensile testing probed with a force-puller setup and recorded in situ until fracture (vi). White arrows indicate the direction of the force. a.u., arbitrary units.

Differential scanning calorimetry (DSC) showed two major endothermic peaks with minima at 297° and 400°C for FGIO fibers (Fig. 2C, red) and 324° and 418°C for raw fibers (Fig. 2C, black). They correspond to the thermal decomposition by depolymerization and volatilization of levoglucosans and to charcoal formation, respectively (24). In comparison with the raw fiber peaks, the FGIO peak located at 297°C is broader and less intense. The FGIO peak at 400°C is sharper and more intense. The 27°C temperature offset between the first endothermic peaks for FGIO and raw fibers indicates an increase in the amorphous cellulose content in FGIO fibers (24), in agreement with our WAXS results.

Fourier-transform infrared spectroscopy with attenuated total reflectance (FT-IR ATR) analysis of FGIO fibers showed a decrease in the intensity of the bands at 2850 and 2915 cm−1 (asymmetric CH2 stretching), 1727 cm−1 (C=O stretching from esters, waxes, fatty acids, and noncellulosic polysaccharides), 1422 cm−1 (CH2 symmetric deformation), and 1243 cm−1 (C=O stretching). Furthermore, a band appeared at 1204 cm−1 (COH in-plane bending and structural organization), and the bands at 1155 cm−1 (antisymmetric stretching of C–O–C glycosidic bonds), 1053 cm−1 (structural and secondary wall synthesis), and 896 cm−1 (β-1,4-glycosidic bonds) became sharper (21, 25). Scanning electron microscopy (SEM) imaging revealed comparable average thicknesses (1.7 ± 0.4 and 1.5 ± 0.3 μm) and widths (21 ± 5 and 24 ± 5 μm) for the raw and FGIO fibers, respectively (fig. S10, A and B), indicating that the modifications occur only at the structural level. Peak fitting and deconvolution of the region corresponding to the OH band (3000 to 3600 cm−1) revealed two intramolecular bonds and one intermolecular bond corresponding to O(2)H–O(6) (3465 cm−1), O(3)H–O(5) (3351 cm−1), and O(6)H–O(3) (3262 cm−1), respectively (Fig. 2E), for the raw fibers; for FGIO fibers, these bands were shifted toward higher wave numbers (Fig. 2F), revealing disturbances in the H-bond network between the cellulose chains (26). In addition, we found two new bands at lower wave numbers (3021 and 3139 cm−1), which we attributed to hydrogen bonding between 6CF-Glc and the cellulose chains, with expected implications for the fiber’s mechanical properties.

To verify this hypothesis, we conducted tensile tests on single fibers (fig. S11). Figure 2G shows typical force-displacement curves for raw (black) and FGIO (red) single fibers. Mean rupture forces were 16 ± 6 and 7 ± 3 mN, respectively. The standard deviation is attributed to variations in the cross-sectional area of the fibers. Because the cross-sectional area varies along the fiber and the rupture forces showed substantial scatter, we refrained from converting rupture forces into tensile strength in terms of force per unit area. However, given that raw and FGIO fibers have comparable average thicknesses and widths (fig. S10, A and B), and assuming the same degree of twisting, we are able to compare the rupture forces for raw and FGIO fibers under tensile loading. We found that fiber rupture occurred more often at the edges than in the middle section of the fibers, most likely because of stress concentration at the clamping points (table S1). During the initial phase of straining, we observed jumps in the force curves (Fig. 2G) owing to untwisting of the single fibers with increasing strain (Fig. 2H for FGIO fibers and fig. S12 for raw fibers). Despite the fact that we observed an increase in H bonds between the cellulose chains and 6CF-Glc and thus would expect an increase in rupture force, the global structural amorphization seems to have a predominant effect of mechanically weakening FGIO fibers.

Our second model exemplifies the application to inorganic compounds. We synthesized a water-soluble glucose macrocycle caging dysprosium [Glc-DOTA-Dy(III); DOTA, 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid; ground term 6H15/2, spin angular momentum S = 5/2, orbital angular momentum L = 5, total angular momentum J = 15/2, g-factor = 4/3, Curie constant C = 14.17 electromagnetic units (emu) K mol–1] to confer magnetic properties to the fibers. As for 6CF-Glc, we used peptide chemistry to synthesize Glc-DOTA-Dy(III) by reacting the free amino group of glucosamine located at C2 with DOTA-N-hydroxysuccinimide ester to form an amide bond. Afterward, Dy(III) was coordinated inside DOTA (Fig. 3A, fig. S13 for the complete synthesis, and fig. S14 for dc magnetic measurements) (27).

Fig. 3 Biological incorporation of a magnetic Dy(III)-based complex into cotton fibers.

(A) Structure of the Glc-DOTA-Dy(III) complex. (B) Sequence of microscopic images taken at 2, 6, 15, and 20 days under “white” illumination and standard conditions (32°C, 5% CO2). (C and D) Ovules at day 20. The incorporation of Glc-DOTA-Dy(III) does not result in (macro)morphological differences. (E) AFM image from an end-clamped single FiDy fiber, showing a relatively smooth surface. (F) WAXS spectra of raw (black) and FiDy (red) fibers, showing structural amorphization. (G) Sequence of images from tensile testing of FiDy fibers, probed with a force-puller setup and recorded in situ until fracture (iv). White arrows indicate the direction of the force. (H) Temperature versus χmass at 5 kOe for raw (black) and FiDy (red) fibers, showing diamagnetic and paramagnetic behavior below 30 K, respectively. The inset shows magnetization versus magnetic field (H) at 300 K of raw (black) and FiDy (red) fibers that remain unchanged throughout the temperature range. Blue points represent FiDy fibers at 2 K. FiDy exhibits superparamagnetic behavior between –20 and +20 kOe and diamagnetic behavior at 20 kOe < H < –20 kOe. (I) Schematic representation of the biological incorporation of Glc-DOTA-Dy(III) into the cellulose fibers.

The ovule development was monitored for 20 days after addition of Glc-DOTA-Dy(III) to the medium under standard conditions (Fig. 3B, i to iv). Visual inspection revealed no morphological differences relative to the control [Fig. 3C and fig. S15 for DOTA-Dy(III)]. Despite the yellow color of the complex, treated fibers remained white (Fig. 3D). Structural characterization was thus required to validate the integration of the Glc-DOTA-Dy(III) complex into the fibers.

By using inductively coupled plasma mass spectrometry, we detected the presence of Dy(III) at a concentration of 2.25% (weight/weight) in the fibers incubated with Glc-DOTA-Dy(III), termed FiDy fibers, whereas raw fibers and fibers incubated with DOTA-Dy(III) contained no detectable dysprosium. To eliminate the suspicion of surface contamination, we characterized the fiber structure at different levels. Atomic force microscopy (AFM) imaging revealed a distinct contrast in texture between FiDy and raw fibers. Whereas the raw fibers exhibited typical fibrillar morphology (fig. S16) with a surface roughness (root mean square, RMS) of about 12 nm for a 3-by-3-μm image, the FiDy fibers exhibited a relatively smooth surface (RMS roughness, 2.5 to 3.5 nm) with what appeared to be dewetted regions (holes) of a depth of 20 to 25 nm (Fig. 3E).

FiDy WAXS diffraction patterns showed a broad peak with a maximum at q = 1.35 Å−1 and a distinct shoulder at about 1.55 Å−1 (Fig. 3F, upper panel, red). Although the shoulder roughly agreed with the (200) peak seen in the raw fibers at 1.62 Å−1 (Fig. 3F, upper panel, black), the weaker (110) and (Embedded Image) peaks of the cellulose I-β crystal structure were absent, indicating partial amorphization and partial retention of the original polymeric nanocrystallites (22, 23). This is supported by the partial loss of preferential orientation, relative to raw fibers, seen in the azimuthal cuts of FiDy fibers (Fig. 3F, lower panel), indicating the presence of Dy(III) between the chains. Azimuthal profiles obtained from 2D SAXS patterns (fig. S7C) showed that the fibers retained a preferential orientation (fig. S8, green) and preserved a fibrous nanostructure in the size range of a few nanometers with a degree of orientation of 0.17—a value lower than that found for raw fibers (0.24) but higher than for FGIO fibers (0.12).

We conducted tensile tests on single FiDy fibers (Fig. 3G and fig. S17 for force-displacement curves and an SEM image after fracture). As with FGIO fibers, we observed jumps in the force curves during the initial phase of straining and avoided converting tensile strength to force per unit area. We found an average rupture force of 9 ± 4 mN for FiDy fibers. Although this value is smaller than the one found for raw fibers (16 ± 6 mN), if we consider FiDy fibers’ average width of 18 ± 6 μm and average thickness of 1 ± 0.3 μm (fig. S10C), and if we take into account that the estimated cross-sectional areas are about half that of the raw fibers, the rupture force should be reduced by a factor of 2 when compared with that for raw fibers. Unlike FGIO fibers, the presence of Glc-DOTA-Dy(III) and structural amorphization (Fig. 3F) do not seem to meaningfully affect the fiber’s mechanical properties. This similarity could challenge the assumption that Dy(III) was incorporated into the fibers and that the amorphization found in WAXS is a consequence of cellular stress response and not induced by Dy(III).

With this in mind, we investigated the Dy(III) environment by means of magnetic measurements, using a superconducting quantum interference device. Figure 3H shows the temperature (T) dependence of a mass magnetic susceptibility (χmass) plot measured at a magnetic field strength (H) of 5 kOe. We found a strong paramagnetic behavior above 30 K (Fig. 3H, red) and a product magnetic susceptibility of 14.12 cm3 K mol−1 at 300 K, decreasing to 10.97 cm3 K mol−1 at 2 K (fig. S18) because of thermal depopulation of the Stark sublevels of the Dy(III) ground state (27). In contrast, the raw fibers showed a diamagnetic behavior (negative magnetic moment) throughout the full temperature range (Fig. 3H, black). We measured magnetization versus magnetic field by sweeping the magnetic field between –60 and +60 kOe at T = 300 and 2 K (Fig. 3H, inset). At 300 K, both raw and FiDy fibers displayed a negative slope throughout the full range (black and red lines)—a feature typically associated with diamagnetic behavior. At 2 K, we found that fibers containing Glc-DOTA-Dy(III) had a superparamagnetic behavior at fields –20 kOe < H < 20 kOe, reaching magnetic saturation at these values. Additionally, fibers containing Glc-DOTA-Dy(III) displayed a negative slope at high magnetic fields (–60 kOe < H < –20 kOe and 20 kOe < H < 60 kOe), differing from that of the inorganic complex alone (Fig. 3H, inset, blue), thus confirming that the Glc-DOTA-Dy(III) is located in a different environment, as illustrated in Fig. 3I.

Here we demonstrate that exogenous molecules carrying a glucose moiety and functionality such as fluorescence or magnetism can be biologically incorporated into cotton cellulosic fibers by using G. hirsutum in vitro cultures. The biofabrication of composite functional materials need not be limited to cotton but could be expanded to other biological systems such as bacteria, bamboo, silk, and flax. With the appropriate molecular design, this approach offers numerous options for transforming raw materials into a wide range of functional and high-value end-products—i.e., material farming.

Supplementary Materials

Materials and Methods

Figs. S1 to S18

Table S1

References (2831)

Movies S1 and S2

References and Notes

  1. Acknowledgments: This work was financially supported by the Forschungsschwerpunkt Nano-strukturierte Materialien (NWG IV: Bioanorganische Chemie) program of the Ministerium für Wissenschaft und Wirtschaft des Landes Sachsen-Anhalt and the German Research Foundation (grants SPP 1486 and SFB 1420). We thank H. Cynus (Fraunhofer IZI, Halle, Germany) for help obtaining the cryogenic cross sections of cotton ovules, D. Strand (Medizinische Klinik, Universitätsmedizin der Johannes Gutenberg-Universität, Mainz, Germany) for help with CLSM imaging, Maren Müller [Max Planck Institute for Polymer Research (MPIP), Mainz, Germany] for help with SEM imaging, and N. Hoinkis (MPIP) for help with the tensile test and SEM measurements. G.F.-P. acknowledges financial support from Bundesministerium für Verkehr, Innovation und Technologie and Bundesministerium für Wissenschaft, Forschung und Wirtschaft, represented by Österreichische Forschungsförderungsgesellschaft, and the Styrian and the Tyrolean Provincial Government, represented by Steirische Wirtschaftsförderungsgesellschaft and Standortagentur Tirol, within the framework of the COMET Funding Programme. F.N. is inventor on patent WO2015051780 A1, held by Martin-Luther-Universität Halle-Wittenberg, that covers a method for producing plant fibers, particularly cotton fibers, doped with nanoparticles to create intelligent textiles by means of a sustainable, automated hydroponic system. The authors declare no competing financial interests.

Stay Connected to Science

Navigate This Article