Spider Silk Fibers Spun from Soluble Recombinant Silk Produced in Mammalian Cells

See allHide authors and affiliations

Science  18 Jan 2002:
Vol. 295, Issue 5554, pp. 472-476
DOI: 10.1126/science.1065780


Spider silks are protein-based “biopolymer” filaments or threads secreted by specialized epithelial cells as concentrated soluble precursors of highly repetitive primary sequences. Spider dragline silk is a flexible, lightweight fiber of extraordinary strength and toughness comparable to that of synthetic high-performance fibers. We sought to “biomimic” the process of spider silk production by expressing in mammalian cells the dragline silk genes (ADF-3/MaSpII and MaSpI) of two spider species. We produced soluble recombinant (rc)–dragline silk proteins with molecular masses of 60 to 140 kilodaltons. We demonstrated the wet spinning of silk monofilaments spun from a concentrated aqueous solution of soluble rc–spider silk protein (ADF-3; 60 kilodaltons) under modest shear and coagulation conditions. The spun fibers were water insoluble with a fine diameter (10 to 40 micrometers) and exhibited toughness and modulus values comparable to those of native dragline silks but with lower tenacity. Dope solutions with rc–silk protein concentrations >20% and postspinning draw were necessary to achieve improved mechanical properties of the spun fibers. Fiber properties correlated with finer fiber diameter and increased birefringence.

The evolutionary survival of spiders has been tightly linked to the diversity, production, and use of silks (1–3). Orb-web spinning spiders have as many as seven highly specialized glands, each producing silk with different mechanical properties and functions (4–6). Dragline silk, used as the safety line and as the frame thread of the spider's web, is one of the strongest silks studied, being three times as tough as aramid fibers and five times stronger by weight than steel (7–10).

The protein core of dragline silk fibers is secreted as a mixture of two soluble proteins from specialized columnar epithelial cells of the major ampullate gland of orb-weaver spinning spiders (1,7). Partial cDNA clones encoding the two protein components of dragline silk have been isolated and characterized from two species of orb-web weaving spiders (Araneus diadematus: ADF-3 and ADF4; and Nephila clavipes: MaSpI and MaSpII) (11–13).

Dragline silk genes encode proteins that contain iterated peptide motifs (14) exhibiting a pattern of alternating Ala-rich, crystal-forming blocks (ASAAAAAA blocks) that impart the silk's mechanical properties (14–16) and Gly-rich amorphous blocks [GGYGPG, (GPGQQ)n], implicated in providing elasticity in the silk filament (17). The process by which spiders spin soluble silk protein into strong, liquid-crystal fibers is a product of 400 million years of evolution. Silk fibers begin to form within the spiders' silk gland from an acidified liquid crystalline-form solution (protein concentration of 30 to 50%) using very small forces followed by drawing in the air after the fiber has left the spiders' body (7). The unique mechanical properties of the spider silk filaments and an inability to domesticate spiders have driven numerous attempts to artificially manufacture spider dragline silks for industrial and medical applications. Recombinant (rc)–spider silk proteins have been produced in bacteria and yeast systems with limited success (18–22). The highly repetitive structure and the unusual mRNA secondary structure result in inefficient translation that limits the size of the silks produced (23). The feasibility of producing soluble rc–spider silks (up to 100 kD) by using synthetic genes (MaSpI) in transgenic plants has been reported recently (24). Several groups have attempted to spin silk protein fibers artificially. Most efforts have involved wet spinning of reconstituted Bombyx mori(25–27) or native spider silk (28, 29). Recombinant silk proteins have been spun with harsh solvents such as hexafluoroisopropanol (30) as diluents or dilute protein solutions in concentrated formic acid (21). However, the mechanical properties of the native spider silks have not been reproduced with either reconstituted or rc–spider silk proteins (23, 31).

Truncated synthesis has been a limiting factor in expressing silks of high molecular weight in Escherichia coli and Pichia(23). To investigate whether mammalian cell systems can efficiently overcome this limitation, we generated two series of constructs for expression of rc–spider silk proteins in mammalian cells using spider dragline silk gene cDNAs: the MaSpI or MaSpII cDNAs (12) and the ADF-3 cDNA (11,32) (Fig. 1). In addition, expression vectors were generated containing multimers of the dragline cDNAs [ADF-33, ADF-333, and MaSpI(2)] (32) (Fig. 1) to produce genes that encode proteins of similar size to those found in the spider major ampullate silk gland (23).

Figure 1

DNA expression constructs used to produce rc–dragline spider silks in mammalian cells.

Two cell lines, bovine mammary epithelial alveolar cells immortalized with large T (MAC-T) and baby hamster kidney (BHK) cells, were chosen as expression systems (32). Analysis of media from stable transfectants of ADF-3, MaSpI, and MaSpII constructs by Western blot (32) revealed prominent immunoreactive bands of the expected molecular size (Fig. 2A, lanes 1, 2, and 5; Fig. 2B, lane 1). To explore the relation between spider silk protein size and mechanical properties of spun fibers, we first evaluated the ability of the mammalian epithelial cells to produce rc–spider silk proteins of high molecular weight similar to that of silk proteins observed in the spider's silk glands (23). Analysis of conditioned media revealed the presence of rc–spider silk proteins of the predicted sizes (∼110 and ∼140 kD) (Fig. 2A, lanes 3 and 4; Fig. 2B, lane 2) produced from concatemers of the ADF-3 gene (ADF-33 and ADF-333) and a dimer of MaSpI, respectively. In all instances, the different expression vectors used enabled the secretion of soluble silk proteins in the media. Distinct spider silk proteins with molecular sizes of 120, 150, 190, 250, and 750 kD have been found in the ampullate gland of N. clavipes (23).

Figure 2

Western blot analysis of media from BHK cells transfected with DNA expression constructs containing (A) ADF-3 and MaSpII and (B) MaSpI. About 20 μl of conditioned media was loaded per lane. (A) Lane 1, ADF-3 His; lane 2, ADF-3; lane 3, ADF-33; lane 4, ADF-333; lane 5, MaSpII. (B) Lane 1, MaSpI; lane 2, MaSpI (2).

Expression of the secreted 110- and 140-kD spider silk proteins from BHK cells was much lower than that of the 60-kD monomer. This may be attributed to inefficient transcription due to the high secondary structure, insufficient secretion of the larger proteins, low copy numbers of the construct being transfected, or limitations of the cell translational machinery. During silk synthesis, the spider produces gland-specific pools of tRNAs for glycine and alanine to meet the increased demand for limiting amino acids (33). Given the unique amino acid composition of the silk proteins (e.g., MaSpII: 32% glycine, 16% alanine), it is possible that the aminoacyl-tRNA pools of the epithelial cells grown in vitro are depleted.

ADF-3 His and ADF-3 rc–spider silk protein (25 to 50 mg/liter each; corresponding to ∼20 μg per 106 cells per day) were produced in BHK cells with more than 12 g of material purified from conditioned culture media (32). The age of the hollow fiber reactor (∼3 months) correlated with the appearance of lower molecular weight spider silk proteins. The appearance of this protein “ladder” was probably due to termination errors of protein synthesis. Translational pausing, resulting in heterogeneous protein expression, has been reported in N. clavipes (16, 22) and B. mori(34).

ADF-3 was recovered from conditioned culture media by precipitation with 15 to 20% ammonium sulfate for an enrichment of at least 50% in a single step (32). The precipitated proteins, containing ADF-3, were readily redissolved in phosphate-buffered saline. In contrast, rc–spider silks produced in E. coli or yeast and precipitated similarly could only be redissolved in strong denaturing solvents such as hexafluoroisopropanol or guanidine hydrochloride (23). This difference in solubility may be attributed to the presence of the COOH-terminus in ADF-3 and MaSpII rc–spider silk proteins produced in mammalian cells, suggesting that the more hydrophilic COOH-terminal 100 amino acids [absent in other studies (23)] may increase the solubility of the secreted silks.

Purified ADF-3 migrated as a major band with an apparent molecular mass of 60 kD on silver-stained SDS–polyacrylamide gel electrophoresis (PAGE) gels under reducing conditions (Fig. 3A, lane 4) and was recognized by ADF-3–specific antibodies (Fig. 3B, lane 3). The purity of rc–spider silk ranged from 80 to 90% (32).

Figure 3

Purification of ADF-3 rc–spider silk secreted from mammalian cells. (A) Silver-stained SDS-PAGE gel. Lane 1, molecular size markers (kD); lane 2, solubilized proteins after ammonium sulfate precipitation of BHK conditioned media loaded onto an anion-exchange column; lane 3, flow-through protein fraction from anion-exchange column; lane 4, elution fraction of bound proteins from anion-exchange column. (B) Western blot analysis. Lanes 1 to 3: same as lanes 2 to 4 in (A).

The identity of the purified ADF-3 protein was further confirmed by NH2-terminal sequencing. It exhibited identity to the first six residues of ADF-3, confirming the predicted sequence and the faithful cleavage of the leader peptide at the expected site. Amino acid analysis further confirmed the identity and purity of the ADF-3 protein (32).

Fiber formation, appearance, and properties depended at least in part on spin dope concentration, purity, and coagulation bath composition. ADF-3 His fibers, derived from rc–spider silk protein purified from MAC-T and BHK cells, were spun from spin dopes at concentrations of 2.8 to 8.3% (w/v) rc–silk protein and 70 to 80% purity (35). Water-insoluble fibers were obtained with diameters of 8 to 23 μm. Because the produced fibers were brittle and difficult to manipulate, we focused on optimizing spinning processes using ADF-3 rc–spider silk protein.

Fibers derived from ADF-3 rc–spider silk protein purified from BHK cells were spun from spin dopes at protein concentrations of 10 to 28% (w/v) and protein purity ranging from 80 to 90% (35). Fibers with measurable mechanical properties were obtained with spin dope rc–silk protein concentrations >10%. However, the highest tensile properties were achieved at concentrations >23%. In these experiments, the water concentration in the coagulation bath was at least 20%. The diameters of as-spun fibers averaged 40 μm. The maximum postspinning draw achieved was up to five times the fiber's original length, and the average fiber diameter decreased to 20 μm. It was necessary to constrain the fiber ends while they were being drawn and air-dried, otherwise the fiber length would shorten partially relative to the drawn length, similar to a viscoelastic material. When the fibers were completely dry, further draw was no longer possible.

Immediately after spinning, fibers were subjected to either single or double draw. Single-drawn fibers were subjected to draw in 70 to 80% methanol (Table 1) (32). Double-drawn fibers were drawn in methanol and then further drawn in water. Fibers subjected to higher draw ratios displayed greater toughness, tenacity, and modulus values, suggesting the importance of a high draw ratio. Improvements in all three measurements were significant when the draw ratio increased from 3 to 5 (Fig. 4, A to C). Although tenacity values increased with increased draw, no difference was observed when fibers were subjected to single or double draw at a draw ratio > 3 (Fig. 4A). The average tenacity values improved from 0.026 to 1.65 grams per denier (gpd) with a draw ratio of 5, with a concomitant decrease of the fiber diameter. The highest tenacity value obtained for ADF-3 was 2.26 gpd (Table 1). This is lower than the reported values for dragline silk (7 to 11 gpd) (Table 1) (16, 29), yet similar to those measured for fibers spun from regenerated spider silk (29).Figure 5 represents typical stress/strain curves obtained from testing of ADF-3–based fibers (sample 1, Table 1). The maximum average toughness obtained from fibers drawn in 75% methanol was 0.645 gpd (Table 1). The highest average toughness value obtained for double-drawn ADF-3 fibers was 0.895 (Table 1) (32), with peak toughness measured as high as 1.60 gpd (Fig. 5). The highest average modulus obtained was 110.6 gpd, with a maximum modulus of 173 gpd recorded. These properties compared favorably with the toughness and modulus measured for nativeA. diadematus dragline silk (Table 1) (16,29).

Figure 4

Effect of postspinning draw on (A) tenacity, (B) toughness, and (C) modulus of ADF-3 rc–spider silk filaments. A fourfold draw improved both the tenacity and toughness of spun fibers for single- and double-drawn fibers. At a fivefold draw, the double-drawn fibers showed marked improvement in toughness. Double-drawn fibers with a fivefold draw have a higher modulus. In single-drawn fibers, the modulus levels off after a threefold draw.

Figure 5

Stress-strain curves from testing of ADF-3–based fibers (sample 1, Table 1). The curves represent analysis performed on seven different fiber fragments, each 1/2 inch in length.

Table 1

Comparison of the mechanical properties of rc–spider silk fibers with those of native silk. M, 75% methanol; W, water; NA, not applicable.

View this table:

The improved mechanical properties with postspinning draw of the rc-silk fibers agree with values reported for respun B. mori(27) and N. clavipes silks (29). In these studies, additional draw resulted in higher toughness and tenacity values. In our studies, the selection of solvent for draw was important; the mechanical properties of double-drawn fibers were superior to those of single-drawn fibers, presumably because of increased molecular orientation. Water can plasticize silks and stabilize the postspinning drawing process by packing and folding of the poly(alanine) domains into β sheets (29). The tenacity values of rc–spider silk fibers was lower than that of native dragline silk; however, from this and other studies (29), comparison of the effects of draw on diameter and tensile strength suggests that it should be possible to generate fibers similar to native dragline silk with further optimization of the spinning process.

A decrease in the diameter of the spun fiber with higher draw correlated with increased birefringence (Fig. 6). Upon drawing, birefringence in silks is positively correlated with molecular orientation, resulting in an increase in the proportion of the β-sheet crystalline areas and amorphous domains (36,37). A transition of birefringence was shown between the ADF-3 His and ADF-3 fibers (Fig. 6, B and D). ADF-3 His fibers were brittle, whereas ADF-3 fibers demonstrated appreciable mechanical properties.

Figure 6

Light microscopy of ADF-3 His and ADF-3 fibers spun from 12% and 22% (w/v) rc–spider silk protein solutions, respectively. Fibers exhibited birefringence and had an average fiber diameter of 20 μm. (A andC) Regular light microscopy for ADF-3 His and ADF-3 fibers, respectively. (B and D) Polarizing light microscopy with a 530-nm first-order red plate for ADF-3 His and ADF-3, respectively. (E) ADF-3 fiber spun from a 22% (w/v) rc–spider silk protein solution in methanol, then drawn fourfold in methanol and once in water. The fiber was observed under polarizing light with a 530-nm first-order red plate. Areas of elongation are apparent after the fiber has undergone mechanical testing.

Fibers observed microscopically after mechanical testing were generally uniform with distinct areas of elongation. These elongated areas showed a 12% diameter decrease and were not associated with the break point (Fig. 6E). This phenomenon resembles the deformation behavior of polyethylenes in which chains are pulled into the “elongated” area and a “necking” is propagated along the fiber (38).

Scanning electron microscopy (SEM) verified the ability to spin solid core fibers using our aqueous process. A scanning electron micrograph of an as-spun ADF-3 fiber stored at 50% humidity is shown in Fig. 7. The figure shows the cylindrical, smooth surface of the spun ADF-3 fiber, as well as the solid interior core of the fiber.

Figure 7

Scanning electron micrographs of an ADF-3 as-spun fiber. (A) Analysis of fiber surface (magnification, ×500); (B) analysis at a break point to examine the fiber interior core (×2000).

Spiders have perfected the process by which the non-Newtonian silk fluid/dope is transformed into fine silk filaments at ambient temperature and with low energy expenditure (37,39) by using shear-sensitive liquid crystalline silk secretions, high protein concentration, and low spinning speeds (7). It was previously thought that the silk protein sequence would be the primary factor determining fiber properties. We spun fibers from one (ADF-3) of the two proteins present in dragline silk with a molecular mass of 60 kD (1) and mechanical properties approaching those of native spider dragline silk. This result raises the question of the contribution of the second protein, ADF-4 or MaSpI, to the silk's fiber properties.

Conditions under which the fibers are spun and processed (for example by drawing) are critical in determining mechanical properties. Wet spinning of rc-silk as described here depends on a chemical transformation and several postspinning draw steps for achieving fiber properties presumably due to aligned microstructure (38). In contrast, spiders produce silks as a result of rapid physical transformation in a single processing step in parallel with drawing because postspinning draw is precluded (7, 37,39). The spinning parameters allow for a modular process design resulting in fiber formation diversity starting with a common primary silk protein sequence. It is anticipated that by varying the silk primary sequences and spinning parameters, fibers with a range of specifications can be produced (37). These fibers could be used in a variety of applications: for example, as very fine monofilament sutures in microsurgery or in uses requiring a high level of energy absorption and elongation similar to that provided by Nylon (37, 40). In contrast to the manufacture of aramid fibers, a process that requires highly concentrated sulfuric acid, production of rc–spider silk using water as the solvent offers obvious advantages (40). An additional benefit is that silk will ultimately degrade in aqueous environments, providing a “green” alternative to traditional synthetic fibers (40).

  • * To whom correspondence should be addressed. E-mail: alazaris{at}


View Abstract

Navigate This Article