Elongate cavities and skin–core structure in Nephila spider silk observed by electron microscopy



Major ampullate silk fibres from the orb-weaving spider Nephila madagascariensis were analysed by transmission electron microscopy. The fibres have a thin outer layer surrounding a column of apparently homogeneous material which contains elongate cavities orientated parallel to the silk fibre axis. The cavities appear similar to ‘elongate vacuolar droplets’ observed in the silk of Antheraea silkmoth larvae. The overall skin–core structure is probably the result of a rheological pattern originating in the two secreting regions recognized in Nephila silk glands; the cavities indicate material inhomogeneities.


Silk is a fibrous protein produced in glandular organs and processed into fibres used outside the silk-producing animal's body ( Rudall & Kenchington, 1971). Best known is the silk from the larvae of the domesticated silkmoth Bombyx mori and other moth larvae, e.g. Antheraea spp. The silks of spiders and other animals have received less attention than those of lepidoptera, and the interpretation of data on spider silk relies heavily on generalizations based on the study of lepidopteran silk. Yet among all silk-producing animals the spiders are outstanding in their wide use of this material, shown by their ability to produce several kinds of silk differing in material properties and used throughout their lifespan ( Vollrath, 1992). Research into spider silk has focused on silk of the tropical genus of orb-weaving spiders Nephila, especially the silk from the major ampullate glands from which protein sequences have been published ( Xu & Lewis, 1990; Hinman & Lewis, 1992; Beckwitt & Arcidiacono, 1994).

The silk from the major ampullate glands of orb-weaving spiders is used for the radii and frame of the orb web and for the dragline, which the spiders lay out when walking about. The silk protein is secreted from the major ampullate gland and passed through a tapering duct to the anterior of the three pairs of spinnerets located near the tip of the spider's abdomen. The gland consists of a tail and a lumen, which are both lined by a secreting epithelium but no cuticle, as opposed to the duct which is lined with cuticle. Below the cuticle, the duct exhibits an epithelium with numerous microvillae ( Kovoor & Zylberberg, 1972), across which water is withdrawn from the liquid protein solution during silk production ( Tillinghast et al., 1984 ). Near the spinning orifice at the spinneret, a muscle-controlled valve has been found ( Wilson, 1962), adding yet another feature to this complex glandular organ, which induces the aqueous solution of proteins to undergo an irreversible phase change to a water-insoluble solid fibre. This change involves an intermediate liquid crystalline state ( Kerkam et al., 1991 ; Willcox et al., 1996 ).

The molecular structure of spider silk fibres has been studied by X-ray diffraction ( Warwicker, 1960; Becker et al., 1994 ), and consensus has it that the protein molecules exist in a rather extended conformation with the protein main chains orientated parallel to the fibre axis ( Lucas & Rudall, 1968). NMR spectroscopy confirms this to be either β-sheets ( Simmons et al., 1996 ) or 31 helices ( Kümmerlen et al., 1996 ).

On supramolecular levels the structural organization seems less clear. X-ray diffraction data of Nephila major ampullate silk are interpreted in terms of the homogeneous two-phase inclusion model proposed as the general scheme for silks ( Gosline et al., 1986 ; Vollrath, 1992). Crystallites of a small size are assumed to be embedded in an amorphous matrix. The crystallites in Bombyx silk are estimated on the basis of X-ray investigations of solutions of silk-gland content to be 60 × 20 Å in lateral dimensions and 200 Å in length ( Kratky et al., 1964 ). We are not aware of any experimental studies demonstrating similar small crystallites in spider silk. Larger, irregularly shaped crystallites, ≈ 70–100 nm in diameter, have been identified in Nephila major ampullate silk by electron microscopy, and a hypothesis for the formation of these has been proposed on the basis of molecular modelling ( Thiel et al., 1994 ; Thiel & Viney, 1995).

Several fibrillar configurations have been observed in silks ( Zahn, 1952; Dobb et al., 1967 ; Mahoney et al., 1994 ; Putthanarat & Eby, 1996; Vollrath et al., 1996 ) but the existence of protein fibrils as substructures of spider silk fibres is still an open question. A recent hypothetical structural model based on light microscopical observations of urea super-contracted silk includes helically twisted fibrils ( Vollrath et al., 1996 ). This model invokes a structural organization of a skin–core morphology such as observed by light microscopy on water-contracted silk ( Work, 1984) and by AFM on native, though epoxy-embedded and cryosectioned, silk ( Li et al., 1994 ).

In the present study we investigate further the microstructure of Nephila major ampullate silk, especially the identity of the twisted fibrils and the question of skin–core morphology.

Materials and methods

Major ampullate silk was drawn mechanically at a speed of 1 cm s−1 from adult female Nephila madagascariensis. CO2 was used to anaesthetize the spiders for a couple of minutes during mounting in the reeling machine, but they were fully awake during reeling. The silk was cut in 2–4 mm sections and submersed for 24 h in a solution containing 4 m urea, 0.5 m NaCl and 50 m m Tris.Cl, pH 8, at room temperature. This treatment, which induced a swelling of about 50% radially, was not essential, but promoted the success of the following procedure. The silk was fixed in 2% glutaraldehyde in 0.1 m cacodylate buffer, pH 7.4, for 6 days, post-fixed for 4.5 h with 1% osmium tetroxide, en bloc stained with 0.5% uranyl acetate for 25 h, dehydrated in ethanol and embedded in Epon. The first 4 h of infiltration were performed in vacuum. Additional fixation procedures only using osmium and avoiding aqueous solutions prior to dehydration were performed after Afzelius (1959). Sectioning was done with a diamond knife and the sections were stained with uranyl acetate and lead citrate. In a standard Philips EM208 transmission electron microscope at 80 kV, in which the pictures were taken, the sections were very stable, showing no sign of special sensitivity to the beam.

The number of cavities was counted using an unbiased 5 × 5 cm counting frame ( Gundersen et al., 1988 ) on micrographs of three random cross-sections (electron microscope magnification 16 000×, photographically enlarged three times). On each section the number of cavities was counted in 14 randomly placed frames. To avoid edge effects, no frame was placed closer than 8% of the radial distance to the outer surface. The cross-sectional diameter of cavities was measured on the same micrographs using an electronic vernier calliper.


In cross-section the fibres had a circular profile, with a thin outer layer of higher electron density surrounding the inner material showing slightly lower electron density and numerous cavities of very low electron density ( Fig. 1a). In cross-section these cavities had a circular profile and the inner surface of the cavities seems to be rough ( Fig. 1b). The two cross-sections of the same fibre shown in Fig. 1(c,d) are at least 300 nm apart. Almost all of the cavities are present in both sections, but some cavities are present only in one of the sections. This indicates an average length of the cavities in the micrometre range. When sectioned at other angles the cavities had an elongated profile always orientated parallel to the silk fibre axis, independent of the direction of sectioning ( Fig. 2a). Examination showed that the cavities were also present in silk: (a) not treated with urea, (b) fixed only with osmium (c) fixed only with osmium and not exposed to aqueous solutions prior to dehydration.

Figure 1.

. Cross-sections of Nephila madagascariensis major ampullate silk fibre. The broad arrow in all micrographs is normal to the edge of the microtome knife. (a) Complete cross-section showing circular profile of the fibre. The material is very hard to infiltrate; sections often show folds. (b) Cross-section showing the cavities. Note the circular profile. (c,d) Two cross-sections of the same fibre at least 300 nm apart. Almost all of the cavities are present in both sections (examples marked with arrows), but some cavities are present in only one of the sections (examples marked with arrowheads).

Figure 2.

. Oblique angle sections of spider silk fibres. The broad arrow in all micrographs is normal to the edge of the microtome knife. (a) Note the elongate shape and aligned orientation of the electron light cavities, independent of sectioning direction indicated by knife marks. (b) Detail of the outer layer at the end of a fibre. The surface layer is present only along the natural surface, facing right, not along the artificial fibre surface originating from initial cutting of the silk in pieces, facing down. Since this initial cutting took place before any other treatments, any introduced artefacts like precipitation from solutions would be expected on both surfaces. Since this is not the case, the radial difference in electron density must be considered to reflect differences in the silk fibre. (c) Between the outer layer and the inner material a third layer of mixed character is possibly identifiable. (d) On the surface of the fibre very delicate triple-layered structures were present.

The thin outer layer was present only along natural surfaces of the fibre ( Fig. 2b). Between the outer layer and the inner material is possibly a third layer of porous appearance ( Fig. 2c). The fibre surface showed very delicate triple-layered structures ( Fig. 2d).

The spatial distribution of cavities within the inner material did not differ significantly from random in any of the sections (Table 1). The cavities located more than 2/3 radius distance from the centre were found to have significantly larger diameters than cavities located less than 1/3 radius distance from the centre (Table 2).

Table 1.  . Test of random distribution of cavities. Results of a X2 (variance to mean ratio) test for agreement with a Poisson series. If the cavities are randomly distributed within the inner material, the distribution of number of cavities in a counting frame will be a Poisson distribution. In none of the sections is the 0-hypothesis of randomness rejected (Zar, 1984). Thumbnail image of
Table 2.  . Diameter of cavities in inner and outer regions. Because of the large sample size and unequal variance, the normal approximation of a one-tailed Mann–Whitney test with tied ranks ( Zar, 1984) is used to test the 0-hypothesis: ‘Mean diameter of cavities in group 1 ≥ Mean diameter of cavities in group 2’. This is rejected on the 0.025 level, which means that the cavities in group 1, close to the centre of the fibre, have significantly smaller diameters than the cavities in group 2, further away from the centre. Thumbnail image of


The observation of a thin outer layer differing in electron density from the inner material is consistent with a proposed skin–core model ( Work, 1984; Li et al., 1994 ; Vollrath et al., 1996 ). In the major ampullate gland two secreting regions are identified ( Kovoor, 1986). The products of these regions differ in histochemical staining affinities, and are believed to be different proteins. In the gland, the protein secreted in the distal region, the one closest to the exit, surrounds the protein secreted in the proximal region ( Kovoor, 1986). This organization of the liquid precursive material is probably the origin of the skin–core structure.

Elongate cavities have not been described in spider silk before. However, similar cavities, named ‘elongated tubular vacuoles’, are known in silk from the moth Antheraea (but not in silk from Bombyx mori) ( Akai et al., 1988 ). Earlier, it was believed that these cavities were an intermicellar system ‘composed of submicroscopic capillaries, all connected by longitudinal cleavage’ ( Frey-Wyssling, 1948). In Antheraea the cavities were later shown to originate from droplets in the liquid silk material. The droplets are present in the protein solution in the posterior part of the silk gland, and appear to originate from lysosomes, which are emptied into the gland lumen ( Akai et al., 1993 ). During the passage through the spinneret of the moth larvae, the spherical droplets, along with the protein mass, are stretched, and thus form elongate cavities in the solid fibre ( Akai et al., 1988 ).

TEM images of the content of silk gland ducts of Bombyx and Nephila show structures described as holes. They have a diameter of the order of 100 nm in Nephila glandular silk ( Willcox et al., 1996 ). The holes were interpreted as artefacts originating from ice crystals caused by imperfect cryofixation of the specimens. Their absence in sections of Bombyx glands, which Willcox et al. ascribe to the use of liquid nitrogen instead of liquid propane as cryogen, as well as the coordinated shape of the holes, makes this interpretation less likely. Instead, the finding of ‘holes’ using yet another preparation technique further supports their identity as intrinsic to Nephila silk. We think the objects giving rise to the holes observed by Willcox et al. (1996 ) could very well be the precursors of the cavities observed by us. Ultrastructural investigations of the Nephila silk gland, especially the secretion process, are needed to show if a process similar to the secretion of droplets in Antheraea is responsible for the cavities in Nephila major ampullate silk.

The significantly smaller diameters of cavities in the centre of the Nephila fibre could indicate that this part of the fibre is stretched more than the outer part, if one assumes the volume of a typical cavity to be independent of its location in the fibre. On the other hand, cavity volume could vary according to position and be affected by the process of extrusion, e.g. the radial flow gradient in the spinneret lumen, which again is affected by spinning speed, viscosity of the liquid, etc. In this case the cavities (and their effect on the material properties of the fibre) could possibly be controlled by the spider; knowledge of the composition of the material in the cavities could provide information about the way in which the properties of the fibres are altered. Finally, the size distribution of cavities might be altered by swelling. However, the swelling process induced by urea proceeds from a cut end of the fibre ( Vollrath et al., 1996 ). This indicates that urea does not penetrate the fibre from the side, and a radial size difference caused by urea is therefore not expected.

In this study we could not observe the fibrillar lattice work which Vollrath et al. (1996 ) described from light microscopy studies of urea-treated silk, nor the irregularly shaped crystals described by Thiel et al. (1994 ) using electron diffraction. This is probably no surprise, since the LM observations depend on staining methods not applicable in EM and since Thiel et al. (1994 ) did not find any crystals using standard TEM. We nevertheless confirm that Nephila major ampullate silk fibres have a heterogeneous structure consisting of several structural elements. This should be considered when modelling the mechanical properties and interpreting surface data obtained from specimens prepared by abrading the fibre surface or longitudinal cleavage ( Mahoney et al., 1994 ; Termonia, 1994; Putthanarat & Eby, 1996). Last but not least, a heterogeneous microstructure does not appear to be a unique feature of Nephila spider silk ( Kovoor, 1987; Peters & Kovoor, 1989; Stubbs et al., 1992 ).


We thank Karen Thomsen and Else-Merete Løcke for excellent technical assistance and Hans Jørgen Gundersen for helpful discussions. We are grateful to Bjørn Afzelius and Svend Olav Andersen for their perceptive comments on the manuscript.