How Spiders Spin Silk: The Strange Polymer Engineering of a 400-Million-Year-Old Material

A spider hanging from a strand of dragline silk is hanging from a fiber that, by mass, outperforms most engineered materials humans know how to make. Dragline silk has tensile strength comparable to high-grade steel (about 1.5 GPa, depending on the species), elongation at break of around 30 percent (matching nylon), and toughness — the energy absorbed before breaking — that exceeds Kevlar by a factor of three. It is made of proteins, at room temperature, from a feedstock that is essentially water. Spiders have been making it for around 400 million years.

For most of those 400 million years, the production process was opaque to anyone interested in copying it. Modern molecular biology has substantially closed the gap on understanding how silk is produced, but the gap on producing it artificially is still wide. Several well-funded companies have tried over the past two decades, with results that have so far underperformed both the biological material and the engineering specifications they were aiming for. The story is a useful case study in why biological materials are sometimes much harder to copy than the chemistry suggests they should be.

The molecular structure

Spider silk is a protein fiber. The proteins are called spidroins (spider fibroins), and they consist of long chains of amino acids arranged in a specific block structure. The chains contain alternating blocks of "crystalline" regions, rich in the amino acid alanine, and "amorphous" regions, rich in glycine. The crystalline regions form ordered structures called beta-sheets — flat sheets of protein chains held together by hydrogen bonds — and the amorphous regions form disordered, springy loops between the sheets.

This block structure is the source of silk's unusual combination of strength and elasticity. The beta-sheet crystals provide stiffness and tensile strength: they are densely packed, held together by many hydrogen bonds in parallel, and difficult to deform. The amorphous regions provide elasticity: when the fiber is stretched, the disordered chains uncoil and extend, absorbing energy. When the load is released, hydrogen bonding and entropic forces pull the amorphous regions back to their original conformation.

The proportion of crystalline to amorphous regions can be tuned by the spider. Dragline silk, used for the radial spokes of a web and as the lifeline a spider hangs from, has a higher crystalline fraction and is stiffer and stronger. Capture-spiral silk, used for the sticky catching threads, has a lower crystalline fraction and is much more elastic — it can stretch to several times its original length, which is what allows it to absorb the kinetic energy of a flying insect without breaking. The silk genes contain code for the protein, and the spider's silk-producing organ adjusts the production parameters to produce silks with different mechanical properties from the same underlying protein toolkit.

The production system

Spiders have multiple silk-producing glands, each producing a different kind of silk for a different purpose. The major ampullate gland produces dragline silk; the minor ampullate gland produces the auxiliary spiral; the flagelliform gland produces the elastic core of the capture spiral; the aggregate gland produces the sticky droplets that coat the capture spiral; the pyriform gland produces the cement that anchors silk to surfaces; the aciniform gland produces the wrapping silk for prey; the tubuliform gland (in females) produces the silk for egg sacs. A single orb-weaver spider can produce six or seven distinct silk types, each tuned to a specific mechanical role.

The silk-producing process inside each gland is the part that engineering has had trouble copying. The spidroin proteins are stored in the gland at extremely high concentrations — around 50 percent protein by mass, which would normally cause the proteins to aggregate and precipitate. The spider keeps them soluble using a combination of pH control, ion composition, and the specific structure of a "terminal domain" on each protein chain that prevents premature aggregation.

When the spider draws out silk through the spinneret, the protein solution undergoes a rapid transition: pH drops, ion composition changes, water is reabsorbed, and shear forces from the drawing action align the protein chains into the beta-sheet structures. The transition is essentially a controlled phase change, transforming a water-soluble protein concentrate into a solid fiber in milliseconds. The mechanical drawing — the spider physically pulling the silk out — is part of the process, not just delivery; the drawing alignment is what makes the beta-sheet structure form in the right direction.

The artificial-silk attempts

The obvious way to produce spider silk industrially is to express spidroin genes in some other organism, harvest the proteins, and spin them into fibers. This has been done. Multiple groups have produced recombinant spidroin proteins in bacteria, yeast, insect cells, plant cells, transgenic goats (whose milk contains spidroin), and transgenic silkworms. The proteins can be made; the question is whether the resulting fibers match natural silk.

So far, mostly not. Recombinant spider silks made by spinning recombinant proteins through artificial extrusion have generally achieved 10 to 50 percent of the strength and toughness of natural silk, depending on the species and method. The reasons fall into roughly four categories.

First, the recombinant proteins are typically shorter than the natural ones. Natural spidroins are around 3,000 to 4,000 amino acids long; recombinant systems struggle to produce proteins that long because the repetitive sequence is unstable in expression systems and tends to be deleted or truncated. Shorter proteins do not form the same beta-sheet networks and do not have the same mechanical properties.

Second, the storage-and-spinning process is hard to replicate. The natural process maintains the proteins at extreme concentrations using mechanisms that are still not fully understood, and the controlled phase change during spinning requires precise control over pH, ions, and shear that artificial extrusion has only partially reproduced. Several groups have made progress on this in recent years, including the use of microfluidic devices that simulate the spider's spinning duct.

Third, the natural silk's properties depend on the specific terminal domains of the proteins, the post-translational modifications, and possibly factors that have not yet been identified. The match between recombinant and natural protein is not always as close as it seems from the sequence comparison.

Fourth, the natural silk is composite — the dragline includes multiple proteins, the capture spiral has both fibrous and droplet components, and the web as a whole is an engineered structure. Reproducing one component is not the same as reproducing the working system.

The commercial state

As of 2026, several commercial-scale recombinant-silk operations exist. Bolt Threads in the United States, AMSilk in Germany, Spiber in Japan, and Kraig Biocraft (transgenic silkworms) are among the largest. Their products have made it into consumer items: AMSilk silk has been used in Adidas concept shoes and medical implants, Spiber silk has been used in North Face jackets, and Bolt Threads spent considerable money attempting a textile launch under the Microsilk brand that did not commercially succeed.

The commercial story is mostly about novel-protein-fibers-with-useful-properties rather than spider-silk-replacements. The recombinant materials have legitimate advantages over traditional fibers in some applications — they are biocompatible, they can be tuned chemically, they are made from renewable feedstocks — and they have found niches in medical devices, cosmetics, and high-end textiles. But they are not yet replacing high-performance synthetic fibers like Kevlar or carbon fiber in the structural applications where spider silk's mechanical properties would matter most.

The conservation question

The interest in artificial spider silk has incidentally drawn attention to the conservation status of the spiders themselves. Most large orb-weavers are not endangered, but the diversity of silk-producing spiders is enormous and not well-cataloged. Several species with unique silk properties are known only from a few specimens collected decades ago. As habitat loss continues, the chance that the next interesting variant of spider silk goes extinct before it is characterized is non-zero.

The Darwin's bark spider (Caerostris darwini), discovered in 2009 in Madagascar, produces dragline silk approximately twice as tough as any other known spider silk. The species lives near rivers and builds webs spanning the water; the high toughness is presumably adapted to the difficult anchor points and the strong winds in its habitat. The species is found in a narrow geographic range and is under pressure from deforestation; it is not currently classified as endangered, but the population is poorly characterized and the silk is not yet fully studied.

The deeper observation

Spider silk is a useful case study in the gap between understanding a biological material and reproducing it. The chemistry of the proteins has been understood in detail for decades. The structural biology of the beta-sheet crystals has been worked out. The mechanism of the gland storage and the spinning process has been substantially elucidated. Despite all of this, the artificial version is still a fraction of the natural performance, and the gap is most likely going to be closed not by better protein expression but by better process control — replicating the storage chemistry, the spinning geometry, and the post-spinning maturation that the spider does for free.

The broader pattern is that biology is often doing more work than the molecular structure suggests. The natural fiber is not just the protein sequence; it is the result of the protein, the storage chemistry, the spinning process, the post-spin maturation, and the assembly into a working structure. Reproducing any single component is not enough. The gap between reading a biological recipe and cooking it in a factory is consistently larger than chemistry alone would predict, and the closing of that gap is the actual engineering frontier in biomaterials.