Constraints on the adhesion of viscous threads spun by orb-weaving spiders: the tensile strength of glycoprotein glue exceeds its adhesion

The sticky viscous prey-capture threads spun by orb-weaving spiders of the Orbiculariae subclade Araneoidea are composed of regularly spaced droplets (Fig. 1), each containing a viscoelastic glycoprotein mass that confers thread adhesion (Peters, 1986; Vollrath et al., 1990; Vollrath and Tillinghast, 1991; Vollrath, 1992; Tillinghast et al., 1993; Sahni et al., 2010). These threads form the spiral component of an orb-web and are responsible for retaining insects that strike the web, providing spiders with more time to locate, run to and subdue prey before they escape (Chacón and Eberhard, 1980). Viscous threads are produced by a triad of spigots on each of a spider's paired posterior spinnerets. The single flagelliform gland spigot of this triad produces a supporting axial fiber and is flanked by two aggregate gland spigots, which coat this fiber with aqueous material (Coddington, 1989). The coated axial fibers from the two spinnerets merge to form a contiguous pair of fibers surrounded initially by a continuous sheath of viscous material. This hydrophilic material (Townley et al., 1991) absorbs atmospheric moisture soon after threads are spun in the early morning hours and condenses this moisture into a regular series of droplets (Edmonds and Vollrath, 1992), whose size and spacing differ greatly among species (Agnarsson and Blackledge, 2009; Opell and Hendricks, 2009).

Glycoprotein within each droplet confers thread adhesion (Peters, 1995; Vollrath and Tillinghast, 1991; Tillinghast et al., 1993; Opell and Hendricks, 2010). This glycoprotein adheres to a surface and, as force is applied to the thread, these viscoelastic molecules and the droplets that contain them elongate (Opell and Hendricks, 2010; Sahni et al., 2010). This allows the thread's elastic axial fibers (Blackledge and Hayashi, 2006a; Blackledge and Hayashi, 2006b) to sum adhesion from multiple droplets by implementing a suspension bridge mechanism (Opell and Hendricks, 2007; Opell et al., 2008; Opell and Hendricks, 2009). A thread's glycoprotein is encoded by the asg1 and asg2 genes, whose messenger RNAs are found exclusively in aggregate glands (Choresh et al., 2009). The upstream region of the 406 amino acid ASG1 protein has a high proportion of charged amino acids, which are though to be hydrophilic, and its repeating downstream region is similar to that of mucin, which is known to have adhesive properties. The upstream region of the 714 amino acid ASG2 protein is similar to known chitin-binding proteins, adapting it to adhere to insect exoskeleton, whereas its repeating downstream region has high proline content that resembles that of elastin and flagelliform spider silk, making it elastic. Thus, the properties of these proteins are consistent with the complementary adhesive, elastic and hydrophilic properties of a viscous thread's droplets.

Although we are beginning to understand how viscous threads generate adhesion, we are less certain about how their adhesion fails. Do elongating glycoprotein filaments rupture or do they pull free of a surface? At higher levels of organization, selection has favored thread release over thread rupture. The adhesion of viscous threads is always less than the tensile strength of their supporting axial fibers (Agnarsson and Blackledge, 2009), ensuring that these threads release before they break. If this functional property extends to finer levels of a thread's organization, then droplet release should result from glycoprotein pull-off rather than glycoprotein rupture.

Studies of viscous thread adhesion suggest or imply that glycoprotein pull-off is the case (Opell and Hendricks, 2007; Agnarsson and Blackledge, 2009; Opell and Hendricks, 2009). When a loaded thread is observed, it does appear to perform in this way, with outermost droplets successively elongating and releasing as force on the thread increases (Sahni et al., 2010) (T. A. Blackledge, personal communication; B.D.O., unpublished observations). Droplets that release from a surface reform and appear normal [fig. 8 in Opell and Hendricks (Opell and Hendricks, 2010)] and are again able to adhere to a surface and elongate when force is applied (Sahni et al., 2010) (B.D.O., unpublished observations). Moreover, a model of droplet performance that assumed glycoprotein pull-off was consistent with empirical data (Sahni et al., 2010). However, the small size of droplets leave open the possibility that the body of a contacting glycoprotein mass may separate from a thin contact layer rather than the entire mass pulling free of the surface. Indeed, studies comparing the stickiness of viscous threads note that they avoided the possibility of contamination by residues left from previously tested threads by using a new test surface or an unused portion of a test surface to measure the adhesion of each thread (e.g. Agnarsson and Blackledge, 2009; Opell and Hendricks, 2009; Sahni et al., 2010).

In this study we tested the hypothesis that a viscous thread releases its hold on a surface because its glycoprotein glue pulls from the surface and not because its elongating droplets break near their attachment to the surface. We did this by comparing the values obtained when three species' threads adhered to four smooth surfaces, which differed in their total surface energy and in the proportions of their dispersion and polar energy components. If thread release results from glycoprotein pull-off, then intraspecific differences in thread adhesion should be observed and these differences should be explained by the surface energies of the materials to which threads adhered.

We measured the stickiness of viscous threads spun by adult females of one member of the family Tetragnathidae, Leucauge venusta (Walckenaer 1842) (N=15), and two members of the family Araneidae, Metepeira labyrinthea (Hentz 1847) (N=11) and Micrathena gracilis (Walckenaer 1842) (N=12), all collected within 10 km of Blacksburg, Montgomery County, VA, USA. Leucauge venusta built webs less than 1 m above the ground near forest edges or in other partially shaded, moist habitats. Metepeira labyrinthea occupied dryer, more exposed habitats, where they built webs 1–2 m above the ground. Micrathena gracilis constructed their webs 1–2 m above the floor of moist forests. The droplet sizes, spacing and percent moisture composition of these species differed, as reported in the literature (Opell and Hendricks, 2009) and summarized in Table 1. In the context of temperate orb-weaving spiders, these three species have medium to small viscous droplets (Agnarsson and Blackledge, 2009; Opell and Hendricks, 2009).

Following techniques described previously (Opell and Hendricks, 2009), we collected orb-web samples on 17 cm diameter rings with Scotch® double-coated tissue tape (tape 4101T; 3M, St Paul, MN, USA) applied to their 5 mm wide rims to maintain threads at their native tensions. Threads from the outer third of orb-webs were sub-sampled on 4.8 mm square brass bars glued at 4.8 mm spacing to a microscope slide. Double-sided Scotch® tape (tape 665; 3M) on the upper surfaces of these supports held threads at their native tensions. After removing alternate spiral turns to prevent the contact plate used to measure stickiness from contacting multiple threads, we examined threads under a dissecting microscope to ensure that they were intact. The microscope slide thread sampler was then clamped to a mechanism that pressed a thread against the contact plate at a constant speed of 0.06 mm s^–1 until a force of 25.0 μN was exerted on the plate, at which time the direction of travel was immediately reversed, pulling the thread away from the contact plate at the same speed until the two separated. This 1891 μm wide (s.d.=22 μm, N=16) contact plate was attached to the lever arm of a load cell, enabling the maximum force registered before thread release to be observed and recorded.

On the day that an individual spider's threads were collected, we measured the stickiness of thread strands with contact plates surfaced with each of the four test materials within the same 1 h period. The 5 mm length of a contact plate allowed us to measure the stickiness of three strands before replacing its surface to ensure that a thread strand always contacted an unused region of the surface. The three measurements taken with each surface were averaged to produce a set of four stickiness values per individual spider.

The four test materials we used were: CS Hyde® Optically clear polytetrafluoroethylene (PTFE) (Teflon) FEP tape with silicone adhesive backing for the PTFE surface, Glad Cling Wrap® for the low-density polyethylene (LDPE) surface, Scotch Magic® tape (tape 810; 3M) for the acetate surface, and the shiny surface of Reynolds Wrap® aluminium foil for the aluminium surface. Material of each type was taken from the same roll or package to ensure consistency. All contact plate surfaces were prepared by the same individual to ensure consistency. As the formulation of these materials can differ and as it has long been known that the surface texture of a material affects its expressed surface energy (Wenzel, 1936), rather than using generic surface energy values we determined the dispersion (non-polar) surface energy (Υ^D_S; mJ m^–2) and polar surface energy (Υ^P_S) components of the surface energy (Υ_S) of each material following the methods described by Rathod and Hatzikiriakos (Rathod and Hatzikiriakos, 2004). We did this by measuring the contact angles of 5 μl droplets of water (HPLC/spectrophotometry grade Optima Water, Fisher, Pittsburgh, PA, USA) and 1-bromonaphthalene 96% (MDL no. MFCD00003868, Acros Organics, Fair Lawn, NJ, USA) that were placed onto each material using an Eppendorf® 10 μl micropipette. To achieve this we photographed droplets with a dissecting microscope on whose stage we positioned a front-surfaced, right-angle prism mounted to a plate that held a microscope slide to which the surface being evaluated was affixed (Fig. 2). These images were taken at a temperature of 21°C, which was the closest constant temperature to the 20°C standard for which the surfaces energies of the test liquids were reported (Table 2) that we could establish. Each of us used Image J (Rasband, 1997-2011) to measure the contact angle on each edge of a droplet independently and averaged these six values to obtain the contact angles shown in Table 3.

We used the JPM statistical analysis program (SAS Institute, Cary, NC, USA) to analyze data, considering P≤0.05 to be significant.

A comparison of our four surface energy values with those reported in the literature shows that our values are appropriate and confirms the usefulness of determining these values rather than using nominal values. For acetate, our 38.2 mJ m^–2 is identical to the reported value (Busscher and Arends, 1981). The reported 42.9 mJ m^–2 of aluminium (Hansen et al., 1993) corresponds well with our 45.4 mJ m^–2, obtained when we averaged the 38.8 mJ m^–2 measured on the dull side of aluminium foil and the 51.9 mJ m^–2 that we measured on the shiny side of the foil. Our 50.5 mJ m^–2 for the LDPE in Glad Cling Wrap® is greater that the 33 mJ m^–2 reported for polyethylene (Bezigian, 1992). However, to improve its adherence to containers, the surface energy of plastic food wrap film is typically increased by methods such as corona discharge treatment (Mangipudi and Tirrell, 1996). This process alters the film's surface properties and ‘as polar groups are introduced on the surface, the surface energy of PE can increase from about 33 to 55 mJ m^–2’ (Mangipudi and Tirrell, 1996). Other procedures, such as low pressure O₂ plasma treatment, can produce LDPE with surface energies as high as 69 mJ m^–2 (Sanchis et al., 2006). Consequently, we believe that our values are accurate. Finally, our Teflon value of 27.4 mJ m^–2 is greater than the reported 24 mJ m^–2 (Good, 1964), although, as expected, it was the lowest of the four values. We used Teflon tape that had an applied adhesive backing. When this tape was dispensed from a roll, the adhesive backing of one layer released freely from the exposed Teflon of the underling layer. However, it is possible that when this occurred the exposed Teflon surface was altered in a manner similar to that described above for polyethylene.

Thread adhesion differed both among species and among test surfaces (Fig. 4), despite the tendency for measurement error to increase as adhesion increased (Fig. 5). An ANOVA documented interspecific differences in thread stickiness and showed that the small differences in the relative humidity (RH) and temperature at which stickiness was measured did not contribute to this difference (model, P=0.0001; species, P=0.0001; RH, P=0.2734; temperature, P=0.2459), an observation that is consistent with earlier findings (Opell and Hendricks, 2009). The grand mean (±s.e.m.) adhesion of M. labyrinthea (119.04±3.85 μN) is 1.404 times that of L. venusta (84.77±3.86 μN) and 1.615 times that of M. gracilis (73.72±2.88 μN). Multiplying the adhesion values of L. venusta and M. gracilis by their respective indices standardizes adhesion across the three species. This is verified by a two-factor ANOVA, which examines the effects of species and test surface on standardized thread adhesion (model, P=0.0001; species, P=1.000; test material, P=0.0001). An ANOVA that included species, Υ^D_S and Υ^P_S was also significant, but with species and Υ^P_S failing to contribute to standardized adhesion (model, P=0.0001; species, P=1.000; Υ^D_S, P=0.0001; Υ^P_S, P=0.5369).

A regression of the three species' mean standardized adhesive values against the Υ^D_S values of the four test surfaces provides further support for the impact of surface energy on thread stickiness (Fig. 6). We attempted to document that the Υ^P_S values of the four test surfaces made a small contribution to thread adhesion by examining their relationship to the regression residuals of the 12 standardized stickiness values shown in Fig. 6, but found no relationship (P=0.9576). Metepeira labyrinthea threads appear to exhibit the greatest adhesion on PTFE and aluminium, M. gracilis threads the greatest values on acetate, and L. venusta threads the greatest values on LDPE. However, the apparent disparity in this ranking of the species' values has little meaning, as surface-by-surface ANOVAs document significant interspecific differences in standardized adhesion only on acetate surfaces (P=0.0067). On this surface, the order of these standardized adhesive values (from high to low: M. gracilis, M. labyrinthea and L. venusta) corresponds directly to the grand mean stickiness per droplet of these species (Table 1), suggesting that differences in droplet spacing contribute to interspecific differences in standardized thread stickiness.

Our finding that dispersion surface energy affected thread adhesion, combined with the results of a recent study of droplet performance (Sahni et al., 2010), provides convincing support for the hypothesis that threads release their hold on surfaces because glycoprotein pulls from the surface, not because it breaks or splits at its attachment to the surface. This conclusion suggests that, just as a viscous thread's stickiness is constrained by the tensile strength of its axial fibers (Agnarsson and Blackledge, 2009), the adhesion of a thread's glycoprotein is constrained by its tensile strength. We would expect the polar component of a surface's energy to interact with the water in viscous thread droplets. Therefore, our failure to document a significant contribution of Υ^P_S to thread adhesion, despite differences in Υ^P_S, the proportion of the test surfaces' Υ^D_S and Υ^P_S values, and the water content of the three species' droplets, is consistent with the finding that glycoprotein adhesion is much greater than the capillary forces resulting from water within thread droplets (Sahni et al., 2010).