Multicomponent nature underlies the extraordinary mechanical properties of spider dragline silk

Significance

Artificial synthesis of spider silk has been actively pursued. However, until now, the natural mechanical properties of spider silk have been largely unreproducible. We thoroughly investigated the genomes and transcripts of four related species of orb-weaver spiders as well as the proteins in their silk threads. Then, in addition to spidroin, we found several low-molecular-weight proteins in common. Interestingly, the low-molecular-weight protein component of spider dragline silk doubled the tensile strength of artificial silk–based material. This discovery will greatly advance the industry and research on the use of protein-based materials.

Abstract

Dragline silk of golden orb-weaver spiders (Nephilinae) is noted for its unsurpassed toughness, combining extraordinary extensibility and tensile strength, suggesting industrial application as a sustainable biopolymer material. To pinpoint the molecular composition of dragline silk and the roles of its constituents in achieving its mechanical properties, we report a multiomics approach, combining high-quality genome sequencing and assembly, silk gland transcriptomics, and dragline silk proteomics of four Nephilinae spiders. We observed the consistent presence of the MaSp3B spidroin unique to this subfamily as well as several nonspidroin SpiCE proteins. Artificial synthesis and the combination of these components in vitro showed that the multicomponent nature of dragline silk, including MaSp3B and SpiCE, along with MaSp1 and MaSp2, is essential to realize the mechanical properties of spider dragline silk.

Spider silk is a typical natural high-performance structural protein and has potential for numerous applications as a protein biopolymer with biodegradability and biocompatibility (1). Spider silk has a tensile strength superior to that of steel, yet it is highly elastic, showing greater toughness than aramid fibers such as Kevlar (2); thus, it has received interest for use in industrial applications (1, 3). Orb-weaver spiders, especially those belonging to the family Araneidae, are often used as models in natural spider silk research. The average mechanical properties of their dragline silks reach ∼1 GPa for breaking strength, 30% for breaking strain, and 130 MJ/m³ for toughness (4). Numerous works have reported the use of recombinant protein and artificial fiber spinning to produce artificial spider silk in genetically optimized organisms (5–10), but it remains challenging to fully reproduce and rival the mechanical properties of natural silks (6, 11–13). The difficulties are multifactorial; the unusually large protein size and the repetitive nature of its sequence are inherent challenges for synthesis, and the natural conditions of spinning are only beginning to be fully uncovered (13–15). However, one primary reason that full reproduction has not been successful is that the previous recombinant approaches employed only MaSp1, only MaSp2, or at most a combination of the two (11, 16–18). In fact, it is often suggested that dragline silk is composed primarily of two components, MaSp1 and MaSp2 (5, 19–23). However, recent proteome analyses suggest the existence of additional components in spider silks, such as a cysteine-rich protein (CRP) in black widows (24–26). Genome and transcriptome analyses have identified many MaSp families (27, 28), and in the genus Araneus, it is known that dragline silk contains nearly equal amounts of MaSp3 and MaSp1/2 (28). Furthermore, low-molecular-weight (LMW) nonspidroin proteins, such as spider-silk constituting element (SpiCE), have been found by transcriptomic and proteomic analyses. SpiCE is a protein of unknown function that is commonly highly expressed in the silk gland and in spider silk. It is becoming apparent that dragline silk is a complex multicomponent material containing much more than MaSp1 and MaSp2.

To pinpoint the protein constituents of dragline silk through quantitative proteomics, a high-quality reference genome and full-length coding sequence annotation are essential to allow the correct and comprehensive identification of the proteins corresponding to the detected peptide fragments. Since spider fibroin genes are extremely long (∼10 kbp) and consist almost entirely of repeat sequences (29, 30), genome sequencing using PCR-free long reads is critical, and in order to eliminate false-positive annotations, predicted coding sequences need to be confirmed on the bases of conservation in closely related species and actual mRNA expression in the silk gland. Hence, we took a multiomics approach to quantitatively identify the dragline protein constituents and the genomes of four golden silk orb-weavers (subfamily Nephilinae): Trichonephila clavata, Trichonephila clavipes, Trichonephila inaurata madagascariensis, and Nephila pilipes. These spiders are reported to produce high-performance dragline silk with average toughness values of 169, 131, 285, and 292 MJ/m³, respectively (SI Appendix, Fig. S1). T. clavipes genome data have already been reported (27), but we chose to construct ab initio assemblies, including for this species, since the existing genome is based on PCR-amplified sequencing for fibroins, and some fibroin gene sequences remain incomplete. Moreover, the existing T. clavipes assembly is suspected to be substantially contaminated, as the entirety of the longest scaffold has been identified as bacterial in origin (31).

Results

Draft Genomes of Four Nephilinae Spiders.

Spider genomes are large and complex, and de novo sequencing is challenging. We extracted genomic DNA from dissected leg and cephalothorax tissue of adult female individuals and conducted hybrid sequencing with short- and long-read sequence technologies (32) (see Materials and Methods). The 10X Genomics linked sequencing produced 702.57 M reads (106.09 Gb), 1,090 M reads (164.60 Gb), 1,010.42 M reads (152.57 Gb), and 900.42 M reads (135.96 Gb) from T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes, respectively (SI Appendix, Dataset S1). PacBio sequencing for the T. clavata genome yielded 74.81 Gb of data with an average length of 8.24 kb and an N50 of 11.1 kb (SI Appendix, Dataset S1). Furthermore, 2.19 M reads (11.03 Gb), 4.70 M reads (41.18 Gb), 5.88 M reads (24.33 Gb), and 2.92 M reads (8.48 Gb) were generated by Nanopore sequencing of gDNA from T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes, respectively (SI Appendix, Dataset S1). Total coverage was ∼80×. Obtained sequence reads were assembled, polished, corrected, and contaminant eliminated (see Materials and Methods); the resulting the number of scaffolds and N50 length in the T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes draft genomes were 39,792 (N50 of 112,957 base pairs [bp]), 21,438 (N50 of 738,925 bp), 28,263 (N50 of 231,352 bp), and 132,899 (N50 of 292,147 bp), respectively (Table 1 and SI Appendix, Figs. S2 and S3). The completeness of each assembled result was assessed by BUSCO (33). The completeness scores were 90.6%, 93.0%, 89.8%, and 92.9% for the eukaryote lineage (Table 1), indicating the high quality of the obtained draft genomes. The tRNAs and ribosomal RNAs were annotated using tRNAscan-SE version 2.0 (34) and Barrnap (https://github.com/tseemann/barrnap) with default parameters. In total, 8,008, 11,433, 8,679, and 1,837 tRNAs and 248 (49 18S, 49 28S, 43 5.8S, and 107 5S), 106 (15 18S, 5 28S, 9 5.8S, and 77 5S), 128 (11 18S, 10 28S, 5 5.8S, and 102 5S), and 167 (8 18S, 8 28S, 6 5.8S, and 29 5S) rRNA copies were predicted in the T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes genomes, respectively. The T. clavipes genome improves upon the previous assembly (27) in quality and completeness (longest contig, 4.65 Mbp; N50 length, 738 kbp) (SI Appendix, Fig. S4 and Supplementary Text).

Table 1.

Assembly and annotation statistics of four spider draft genomes (T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes)

Statistics	T. clavata	T. clavipes	T. inaurata madagascariensis	N. pilipes
Genome	v4	v4	v4	v4
Estimated genome size (Gbp)*	2.46	2.72	2.32	2.12
Assembly size (Gbp)	2.50	2.87	2.51	2.69
Scaffold number	39,792	21,438	28,263	132,899
Average scaffold length (bp)	62,773	134,077	88,704	20,274
Longest scaffold length (bp)	1,797,358	4,645,559	2,667,260	4,220,928
Shortest scaffold length (bp)	1,010	150	1,150	451
N50 (bp) (# of scaffolds in N50)	112,957 (#5,596)	738,925 (#1,112)	231,352 (#2,910)	292,147 (#1,975)
N90 (bp) (# of scaffolds in N90)	22,952 (#25,683)	104,003 (#4,898)	40,466 (#12,765)	6,691 (#34,939)
BUSCO version 4.0.5 completeness (%)†	90.6	93.0	89.8	92.9
	C:90.6%[S:85.5%,D:5.1%],F:5.5%,M:3.9%,n:255	C:93.0%[S:92.2%,D:0.8%],F:2.7%,M:4.3%,n:255	C:89.8%[S:86.7%,D:3.1%],F:2.7%,M:7.5%,n:255	C:92.9%[S:89.8%,D:3.1%],F:3.5%,M:3.6%,n:255
Genes
Number of protein-coding genes	18,822	45,304	16,899	17,127
tRNAs	8,008	11,433	8,679	1,837
rRNAs	248	106	128	167
BUSCO version 4.0.5 completeness (%)†	91.4	87.5	91.0	93.4
	C:91.4%[S:89.0%,D:2.4%],F:5.9%,M:2.7%,n:255	C:87.5%[S:85.5%,D:2.0%],F:8.6%,M:3.9%,n:255	C:91.0%[S:89.8%,D:1.2%],F:4.3%,M:4.7%,n:255	C:93.4%[S:91.0%,D:2.4%],F:5.1%,M:1.5%,n:255

^*Genome size was estimated based on k-mer (GenomeScope).

^† BUSCO analysis was based on eukaryota lineage of protein-coding genes.

The number of protein-coding genes predicted by BRAKER was initially 70,418, 517,373, 51,581, and 72,271. Redundant genes were eliminated by CD-HIT-EST (35) clustering with a nucleotide identity of 97%. To obtain a functional gene set, we removed the genes with an expression level of less than 0.1 and unannotated genes. Finally, 18,822, 45,304, 16,899, and 17,127 functional protein-coding gene sets were obtained for T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes, respectively (Table 1). BUSCO (version 4.0.5) was used to determine the quality of our functional gene set using the eukaryote lineage, and the completeness scores were 91.4%, 87.5%, 91.0%, and 93.4% for the eukaryote lineage (Table 1).

Catalog of Spidroins in Four Nephilinae Spiders.

Spidroin genes are structured with a long repeat domain sandwiched between the nonrepetitive N/C-terminal domains and an extreme length of 10 kbp or more (29, 30, 36, 37). Short-read sequencing and PCR amplification technology alone often cause fragmentation, collapse, and chimerization of spidroin genes. Therefore, long-read sequencing using MinION or GridION is an essential technique for finding spidroins because these technologies can sequence a region covering the entire length of the spidroin-coding gene. However, the sequence of the repeat region is very delicate, and the sequence quality of long reads alone is not sufficiently high. We have contributed to the construction of the spidroin catalog using hybrid sequencing, which combines various sequencing technologies and has been reported previously (28, 32, 38). In a previous study, we reported the spidroin catalog, including complete spidroin coding sequence in A. ventricosus, and showed that known spidroin genes that were sequenced without using long reads were mostly short (28). Furthermore, the collection of partial or repeat-domain-collapsed spidroin genes omitted at least one gene family. The new gene family of MaSp in A. ventricosus was found in a comparison of full-length sequences. Here, we used a hybrid method that combines the gDNA sequencing reads used for genome assembly and the few hundred million reads obtained by cDNA and direct RNA sequencing to catalog the spidroin diversity in four Nephilinae spiders (SI Appendix, Dataset S1 and Supplementary Methods). Seven orthologous groups (MaSp: major ampullate spidroin, MiSp: minor ampullate spidroin, AcSp: aciniform spidroin, Flag: flagelliform spidroin, AgSp: aggregate spidroin, PySp: pyriform spidroin, and CySp: cylindrical/tubuliform spidroin) are known as the typical spidroins in Araneoid orb-weaving spiders. We obtained all seven groups contain 2 to 3 families or subfamilies from four Nephilinae draft genomes, five of which were full length (MaSp, MiSp, AcSp, CySp, and PySp) (see Fig. 1 and SI Appendix, Fig. S5 for details).

Fig. 1.

Catalog of spidroins in four Nephilinae spiders. This summary panel shows the spidroin sequence characteristics and structures obtained from four Nephilinae draft genomes. The icons in the first column represent spidroin groups. The second column represents the motif variety in the domains, as follows: β-sheet [A_n, (GA)_n, and AS], blue; β-turn (XQQ, GPGQQ, and GPGXX), yellow; spacer (SSX and TTX), green; 3₁₀ helix (GGX), orange; and MaSp motif (DGGR, GGYGGRF, and GGYGGL), red. The spidroin structure columns show the N/C-terminal (yellow and blue box) and repeat domains, and each structure is drawn to scale. The number, width, and color of stripes reflect the number, size, and motif identify of the repeats. Assembly gaps are represented by “N” (SI Appendix, Fig. S5 for details). T. clavata, T. clavipes, and N. pilipes images credit: Akio Tanikawa (University of Tokyo, Tokyo, Japan).

The spidroin genes were well conserved among Nephilinae spiders and followed the established phylogenetic relationship (SI Appendix, Figs. S6 and S7). Interestingly, the spidroin catalog revealed a group of major ampullate spidroins distinct from MaSp families 1 and 2 that form a unique clade within Nephilinae. The N-terminal phylogeny suggests that this group forms a clade with the MaSp family 3 sequence of Araneidae (SI Appendix, Fig. S8), reflected by the characteristic DGGR motif, yet the N terminus conservation is sufficiently low to define this as a unique group; thus, we named this gene MaSp3B based on a spidroin nomenclature (SI Appendix, Fig. S9 and Supplementary Text). Previously identified partial sequences, Sp-907 and Sp-74867 (27), revised as MaSp3_A and MaSp3_B (39), belong to this group.

Expression Profile of Spidroin and SpiCE.

With this high-quality reference of full-length spidroins, the existence of a large proportion of MaSp3B in the forcibly reeled dragline silk of T. clavata was immediately observable as a distinct band in the high-molecular-weight (HMW) fraction on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 2A). The abundance ratios of the proteins were then quantified by intensity-based absolute quantification (iBAQ) (40) with shotgun proteomics of selectively reeled dragline silks (SI Appendix, Fig. S10). These analyses showed consistent composition not only within the replicates but also among the four species analyzed (Fig. 2B). Among the spidroins, only MaSp family members were detected; MaSp1, 2, and 3 accounted for ∼90 wt% of the total in any spider species (Fig. 2B). Transcriptome analysis of silk glands confirmed the specific expression of MaSp3B in the major ampullate silk gland, similar to that of MaSp1/2 (Fig. 2C and SI Appendix, Fig. S11 and Datasets S2–S4). Both of these omics analyses also clearly demonstrated the consistent presence of nonspidroins, which was also confirmed by SDS-PAGE of the LMW fraction (Fig. 2A). Across the four Nephilinae species, these genes are consistently expressed as mRNAs at a level equivalent to that of MaSp genes in the major ampullate silk gland and at lower levels in the minor ampullate silk gland (Fig. 2C and SI Appendix, Fig. S11). Three such proteins highly expressed in both silk (as protein) and the silk gland (as mRNA) were annotated as SpiCEs, following the identification of nonspidroin constituents in A. ventricosus dragline silk (28). These SpiCEs are conserved in Nephilinae (SI Appendix, Fig. S12), but their sequences are considerably nonhomologous to those of A. ventricosus; therefore, we named them SpiCE-NMa1, 2, and 4 (SpiCE Nephilinae Major Ampullate). Among them, SpiCE-NMa2 and SpiCE-NMa4 were CRP type with high cysteine content. The most abundant nonspidroin protein among the dragline silks was SpiCE-NMa1 at ∼1 wt% (Fig. 2B).

Fig. 2.

Expression profile of spidroin and SpiCE in silk and silk glands. (A) SDS-PAGE of HMW and LMW fractions from 3.97 mg of T. clavata dragline silk dissolved in 80 µL Tris⋅HCl (pH 8.6) containing 1% Triton-X and 2% SDS. SDS-polyacrylamide gels (4%) of the HMW fraction were stained with SYPRO Ruby or Silver after electrophoresis for 80 min at 20 mA. HiMark Unstained Standard (LC5688, Life Technologies) was used as a molecular marker. SDS-polyacrylamide gels (12.5%) of the LMW fraction were stained with SYPRO Ruby or Silver after electrophoresis for 80 min at 20 mA. Broad Range Protein Molecular Weight Marker (V8491, Promega Technologies) was used as the molecular marker. (B) Average percentage of protein contents in dragline silk of T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes. Each SpiCE-NMa1 was indicated in yellow. (C) Expression profiles from the major ampullate silk glands in T. clavata (n = 6), T. clavipes (n = 7), and N. pilipes (n = 5). The x-axes represent the median expression level in transcripts per million. The y-axes show the highly expressed genes in each gland. The yellow boxes indicate SpiCE-NMa1. The photomicrographs show typical silk gland samples dissected from each spider (SI Appendix, Datasets S2–S4 for details).

SpiCE Doubles the Tensile Strength of Artificial Silk–Based Material.

To investigate the role of these proteins in the expression of the mechanical properties of Nephilinae dragline silks, we first produced composite films of recombinant MaSp family proteins. Recombinant MaSp was synthesized based on shortened amino acid sequences of ∼50 kDa (SI Appendix, Figs. S13 and S14). The transparency of silk-based films did not drastically decrease even when the composite films were prepared from two or three spidroins, suggesting that heterogeneous interactions among MaSps did not induce specific aggregation (Fig. 3A). Wide-angle X-ray scattering (WAXS) profiles demonstrated (210) and (040) reflections; these are indicative of the antiparallel β-sheet structure (41, 42). Thus, MaSp1, MaSp2, and MaSp3 predominantly form β-sheet structures, similar to polyalanine (Fig. 3B). The composite films of MaSps showed similar β-sheet structures, even with different protein compositions, that is, with MaSp1 used alone, with the mix of three families, and with the natural composition (Fig. 3B). We further tested the impact of SpiCE inclusion by producing composite films of recombinant MaSp and SpiCE-NMa1, which was ∼5 times more abundant than other SpiCEs (Fig. 2B). Recombinant SpiCE-NMa1 was added in the range of 0 ∼ 5 wt% based on proteome quantification to produce composite films, and the film transparency increased to 76.1 with 5 wt% SpiCE-NMa1 with MaSp compared to 72.5 with MaSp alone, suggesting a close interaction of the proteins in the composite film (Fig. 4A and SI Appendix, Fig. S15). Surprisingly, the addition of SpiCE-NMa1 dramatically changed the mechanical properties of the composite film. Tension tests showed a twofold increase in the tensile strength of the MaSp film containing SpiCE-NMa1 (Fig. 4 and SI Appendix, Figs. S16 and S17). The impact of SpiCE was also clearly observed in the recombinant composite silk mixed with recombinant MaSp and SpiCE-NMa1. Composite silk fibers with 2% SpiCE-NMa1 showed an increase in elongation at break as well as a decrease in tensile strength (Fig. 4C). These changes in mechanical properties indicate that SpiCE-NMa1 disturbed the strain-induced crystallization of silk molecules and plasticized the amorphous region. This can be because the amorphous region of silk fibers is relatively more oriented and aligned along the fiber axis than that of silk film (43), and hence, SpiCE-NMa1 cannot interact with silk molecules in the amorphous region of silk fibers, in contrast to the amorphous silk in film. A small percentage (∼2%) of SpiCE-NMa1, which interacts with the amorphous silk, in artificial silks could potentially be the molecular tool to modify and tune their mechanical properties.

Fig. 3.

Physical properties of the artificial MaSp composite film. (A) Seven films were produced by the combination of recombinant MaSp family proteins (MaSp1, MaSp2, and MaSp3). The sequences of recombinant proteins are indicated in SI Appendix, Fig. S13. T% indicates transparency at a wavelength of 500 nm. (B) The WAXS profiles of the films with different MaSp compositions showing the one-dimensional radial integration profiles and two-dimensional patterns.

Fig. 4.

SpiCEd film and silk. (A) Composite films mixing MaSp2 and SpiCE-NMa1. The stress–strain curve represents the typical tensile engineering stress–strain in MaSp2 and SpiCE composite films under different SpiCE concentrations (0, 1, 3, 5 wt%). The transparency (%) of these films is shown as T in each picture (SI Appendix, Figs. S14–S16 for details). (B) Recombinant spider silks mixing MaSp2 and SpiCE-NMa1. SEM images show the cylindrical and smooth surface and cross-section of each recombinant spider silk. (C) Stress–strain curve of the recombinant spider silks mixing MaSp2 and SpiCE-NMa1.

Discussion

MaSp3 is widely conserved in the large-web-forming group of the family Araneidae and has been found by proteomics to be a major component of dragline silk in A. ventricosus. The conservation of MaSp3 has been thought to be limited to closely related species of the genus Araneus (28), and it was believed that Nephilinae spiders do not have MaSp3 even though they also build large webs. However, our high-quality genome analyses demonstrate that MaSp3B, a distinct subtype of MaSp3 in the genus Araneus, is conserved in the genera Trichonephila and Nephila. Further, our analyses suggest that MaSp3B is one of the major components of dragline silk and may associate with other MaSps at equivalent stoichiometry. This finding demonstrates the limitation of the conventional MaSp1/2-based strategy for artificial silk synthesis and the importance of including other components, such as MaSp3, for the purpose of producing araneoid spider silk, which has particularly strong mechanical properties. The significance of the multicomponent nature of the mechanical property of dragline silk partly explains the robustness of the compositional percentages of spidroins previously reported for N. pilipes (44). Our proteome analysis showed compositional variability in line with the previous report, but the existence of MaSp3 and SpiCE prescribes the performance in these silks. Although the specific role of MaSp3 is still unclear, the key residues contributing to the N-terminal domain dimerization, such as D40, K65, E79, and E119, are conserved in MaSp1-3 (SI Appendix, Fig. S18). The N-terminal domain of MaSp changes its structure in a pH-dependent manner to form dimers, which are believed to be responsible for stable filament formation in ducts (45, 46). Each conserved amino acid site is associated with this dimerization (47). Therefore, it is likely that MaSp3, like other MaSp1 and MaSp2, functions in pH-sensitive dimerization of the N-terminal domain and contributes to silk formation.

N. pilipes shows higher composition compared to the other three species, which confirms the higher proline content documented in this species (48). Therefore, we analyzed the relationship between the amino acid composition of dragline silk and its mechanical properties. We reconfirmed the correlation between the mechanical properties and amino acid composition, such as glycine or proline, discussed in previous studies (48) (SI Appendix, Fig. S19). This result supports the previous claim that the evolutionary bias in spidroin may differ between the terminal region and the repeat region and that this variation in the amino acid composition may have contributed to some of the variability in physical properties among species.

The contribution of the LMW weight component of the spider dragline silk, SpiCE, to its mechanical properties was also confirmed in vitro, in that inclusion of SpiCE doubled the tensile strength of the artificial film, but the exact mechanism responsible for this phenomenon is still an area of future work. Accessory components of structural proteins are often used to stabilize the structure; for example, elastin, another known structural protein, binds with fibulin proteins to produce elastic fibers (49). Likewise, while a self-assembly system involving pH-induced fibrillization of the N-terminal domain and liquid-liquid phase separation (LLPS) of the C-terminal domain (14) is known to participate in spider silk fibril formation, SpiCE could serve to support the interaction between repeat regions rather than the terminal domains. According to SI Appendix, Fig. S17, the β-sheet structure and crystallinity of the films were not changed significantly. Overall, the SpiCE-NMa1 preferentially interacts with silk molecules in the amorphous region rather than those in the crystalline state, namely, β-sheet crystal structures. The intermolecular interactions of amorphous silk molecules via SpiCE-NMa1 decrease the molecular weight between crosslinking points, resulting in the enhancement of the strength and modulus of the composite films. Further observation of the interaction between SpiCE and spidroin with detailed biophysical analysis, such as NMR spectroscopy or atomic force microscopy scanning, is desirable to elucidate the functions of SpiCE. The evolutionary origin of SpiCE also requires further analyses, as these proteins seem to be specific to a narrow range of spider clades. Unlike MaSp3, the conservation of SpiCE has been confirmed only within very closely related groups (genus or subfamily at most), and CRP seems to be used instead of SpiCE in black widow spiders (26). The MaSps retain homology across different spider clades, but there is no sequence homology among SpiCEs, suggesting that MaSps and SpiCEs did not coevolve. Multiple lineage-specific duplications of MaSp paralogs are observed even among orb-weaving spiders (27, 28, 50–52), and the evolution of SpiCE seems to follow this lineage-specific evolutionary pattern.

Considering the contributions of MaSp3 and SpiCE to the mechanical performance of spider silk, it is essential to consider silk as not just a product of MaSp1 and MaSp2 but rather a multicomponent material that also includes MaSp3 and SpiCE. The contribution of multiple paralogs of MaSp1 and MaSp2, which is not covered in this work, may also contribute to the specific ecological adaptation of Nephilinae spiders. The availability of four closely related Nephilinae genomes and their gland transcriptomes as well as the dragline silk proteomes presented in this work will contribute to the research in this direction, as well as other phylogenomic work of arachnids.

Materials and Methods

Spider Sample and Rearing Environment.

T. clavata and N. pilipes samples were collected in Japan, and T. inaurata madagascariensis was collected in Madagascar. T. clavipes samples were purchased from Spider Pharm Inc. All the spiders used in this study were adult females. The identification of spider specimens was performed based on morphological characteristics and sequence identification of cytochrome c oxidase subunit 1 (COX1) in the Barcode of Life Data System (BOLD: http://barcodinglife.org). The spiders used for dissection or silk reeling were kept in a PET cup (129 × 97 mm, PAPM340: RISUPACK CO., LTD) in a climate-controlled laboratory with 25.1 °C, 57.8% humidity, and 12 to 12 h light–dark cycles. The spiders were fed every two days with crickets purchased from Mito-korogi Farm, and water was given daily by softly spraying inside the cup. The legs and cephalothoraxes were used for gDNA extraction, and abdomens and silk glands were used for RNA extraction. All of the tissues were dissected from adult female spiders. Dissected samples were immersed in liquid nitrogen (LN2) and stored at −80 °C until the next process.

HMW Genomic DNA Extraction.

Spider genomic DNA was extracted from dissected leg and cephalothorax tissue from adult female individuals using Genomic-tip 20/G (QIAGEN) following the manufacturer’s protocol. To obtain HMW genomic DNA, all steps were conducted as gently as possible. The legs and cephalothorax were separated from flash-frozen spider specimens (−80 °C) and then homogenized with a BioMasher II (Funakoshi) and mixed with 2 mL of Buffer G2 (QIAGEN) containing 200 µg/mL RNase A. After the addition of 50 µL Proteinase K (20 mg/mL), the lysate was incubated at 50 °C for 12 h on a shaker (300 rpm) and centrifuged at 5,000 × g for 5 min at 4 °C. The aqueous phase was loaded onto a pre-equilibrated QIAGEN Genomic-tip 20/G (QIAGEN) by gravity flow and washed three times. The DNA was eluted in high-salt buffer (Buffer QF) (QIAGEN). Using isopropanol precipitation, we desalted and concentrated the eluted DNA and resuspended it in 10 mM Tris⋅HCl (pH 8.5). The extracted genomic DNA was checked for quality using a TapeStation 2200 instrument with genomic DNA Screen Tape (Agilent Technologies) and quantified using a Qubit Broad Range dsDNA assay (Life Technologies). Size selection (>10 kb) was performed with a BluePippin with High Pass Plus Gel Cassette (Sage Science).

RNA Extraction, mRNA Selection, and cDNA Preparation.

Spiders and silk gland samples were stored in LN2 at −80 °C until RNA extraction. RNA isolation from the dissected abdomen or silk glands in the abdomen was conducted based on a spider transcriptome protocol (53). Flash-frozen dissected abdomen tissue was immersed in 1 mL TRIzol Reagent (Invitrogen) along with a metal cone and homogenized with the Multi-Beads Shocker (Yasui Kikai). Phase separation was performed by the addition of chloroform, and the upper aqueous phase containing RNA was purified automatically with an RNeasy Plus Mini Kit (QIAGEN) in a QIAcube instrument (QIAGEN). The RNA was quantified and checked for quality using a Qubit Broad Range RNA assay (Life Technologies) and NanoDrop 2000 (Thermo Fisher Scientific). The integrity was estimated by electrophoresis using a TapeStation 2200 instrument with RNA ScreenTape (Agilent Technologies). mRNA selection was performed using oligo d(T). For direct RNA sequencing, mRNA was prepared using the NucleoTrap mRNA Mini Kit (Clontech) from 500 ng of total RNA. cDNA was synthesized from mRNA isolated from 100 µg of total RNA by NEBNext Oligo d(T)₂₅ beads (skipping the Tris buffer wash step). The first- and second-strand cDNA were synthesized using ProtoScript II Reverse Transcriptase and NEBNext Second Strand Synthesis Enzyme Mix.

Library Preparation and Sequencing.

For cDNA, sequencing was carried out on the abdomens of T. clavata (n = 14), T. clavipes (n = 2), T. inaurata madagascariensis (n = 1), and N. pilipes (n = 14), in the major ampullate silk glands of T. clavata (n = 6), T. clavipes (n = 7), and N. pilipes (n = 5), and in the minor ampullate silk glands of T. clavata (n = 6), T. clavipes (n = 8), and N. pilipes (n = 6). All cDNA libraries were constructed according to the standard protocol of the NEBNext Ultra RNA Library Prep Kit for Illumina (New England BioLabs). The synthesized double-stranded cDNA was end-repaired using NEBNext End Prep Enzyme Mix before ligation with NEBNext Adaptor for Illumina. After USER enzyme treatment, cDNA was amplified by PCR with the following conditions: 20 µL cDNA, 2.5 µL Index Primer, 2.5 µL Universal PCR Primer, 25 µL NEBNext Q5 Hot Start HiFi PCR Master Mix 2X; 98 °C for 30 s and 12 cycles each of 98 °C for 10 s, 65 °C for 75 s, and 65 °C for 5 min. cDNA sequencing was conducted with a NextSEq 500 instrument (Illumina) using 150-bp paired-end reads with a NextSEq 500 High Output Kit (300 cycles).

The synthetic long-read sequencing using 10X Genomics was carried out in all four Nephilinae spiders. Purified genomic DNA fragments longer than 60 kb (10 ng) were used to prepare the 10X GemCode library with the Chromium instrument and Genome Reagent Kit version 2 (10X Genomics) following the manufacturer’s protocol. The 10X GemCode library sequencing was conducted with a NextSEq 500 instrument (Illumina) using 150-bp paired-end reads with a NextSEq 500 High Output Kit (300 cycles).

For PacBio sequencing, the library preparation was performed with gDNA fragments of more than 10 kb in size (selected by BluePippin) and sequenced in PacBio RSII using P6-C4 chemistry. PacBio library construction and sequencing were conducted at Genomic Information Research Center, Osaka University.

The gDNAs of T. clavata, T. clavipes, T. inaurata madagascariensis, and N. pilipes were also sequenced by Nanopore technology. Library preparation was completed following the 1D library protocol (SQK-LSK109, Oxford Nanopore Technologies). The quality of the libraries was estimated by TapeStation 2200 with D1000 Screen Tape (Agilent Technologies). Direct RNA sequencing was performed for T. clavata, T. clavipes, and N. pilipes. Libraries constructed from 500 ng of total RNA were prepared following the direct RNA sequencing protocol (SQK-RNA001, Oxford Nanopore Technologies). Sequencing was performed using a GridION instrument with Spot On Flow Cell Rev D (FLO-MIN106D, Oxford Nanopore Technologies). The base calls were performed by Guppy basecalling software (version 3.2.10+aabd4ec).

Genome Assembly.

We adopted a hybrid assembly strategy for spider draft genomes that combined the various reads produced by each sequencing technology. Natural long reads were produced by direct sequencing of HMW genomic DNA with Nanopore or PacBio technology. The synthetic long reads were generated using a combination of Illumina and 10X Genomics technologies. The details of the assembly methods used for each of the four spider genomes are described in SI Appendix, Supplementary Methods.

Contaminant Elimination and Genome Size Estimation.

To detect possible contaminants in the genome, we submitted each genome assembly to BlobTools analysis (54). The genome sequence was submitted to a Diamond (55) BLASTX search (–sensitive–max-target-sEqs. 1–evalue 1e-25) against the UniProt Reference proteome database (downloaded November 2018). Each contig was classified into a lineage with BlobTools taxify. We then mapped DNA-Seq reads to the genome by BWA MEM and conducted SAM to BAM conversion with SAMtools. These taxonomic annotation and coverage data were visualized by BlobTools following the protocol. Contigs that were classified as from bacteria, plants, or fungi were removed from the assembly, and the filtered genome assembly was assayed by Benchmarking Universal Single-Copy Orthologs (BUSCO) version 4.0.5 (33) (eukaryote lineage) to validate genome completeness. Genome heterozygosity, repeat content, and size were estimated based on the k-mer distribution with Jellyfish and GenomeScope (56).

Gene Prediction and Annotation.

Gene prediction was performed using a gene model created by cDNA-seq mapping data with HISAT2 and BRAKER (57, 58). cDNA-seq reads were mapped to the genome with HISAT2 (version 2.1.0), and the resulting SAM file was converted, sorted, and indexed by SAMtools (version 1.4). Repeat sequences were detected by RepeatModeler (1.0.11) and soft-masked by RepeatMasker (version 4.0.7). The soft-masked genome was submitted to gene prediction with BRAKER (version 2.1.4,–softmasking–gff3). The amino acid sequences were submitted to BLASTP or Diamond BLASTP searches against public databases (UniProt TrEMBL, UniProt Swiss-Prot). Redundant genes were eliminated by CD-HIT-EST (35) clustering with a nucleotide identity of 97%. To obtain a functional gene set, we removed the genes with an expression level of less than 0.1 and unannotated genes. BUSCO (version 4.0.5) was used to determine the quality of our functional gene set using the eukaryote lineage. The tRNAs were annotated using tRNAscan-SE version 2.0 (34) with default parameters. Ribosomal RNAs (rRNAs) were predicted using Barrnap (https://github.com/tseemann/barrnap).

Spidroin Gene Catalog.

We used a hybrid method that combines short- and long-read sequencing to catalog the spidroin diversity in four Nephilinae spiders. The short reads were obtained from Illumina sequencing of cDNA or gDNA libraries, and the long reads were obtained by Nanopore or PacBio sequencing of the gDNA library. The typically used de Bruijn graph algorithm is not suitable for the assembly of long repeat domains, so the SMoC (Spidroin Motif Collection) algorithm (28), developed based on the OLC (Overlap-Layout-Consensus) algorithm, was used. The SMoC algorithm collects as many patterns of repeat motifs as possible, scaffolds these repeat motifs based on long reads, and then provides the full-length spidroin gene sequence. SMoC first finds the N/C-terminal region (nonrepetitive region) contigs from the assembled contigs with a homology search. These terminal fragments were used as seeds for screening of the short reads obtained by an Illumina sequencer harboring an exact match of extremely large k-mers (∼100) up to the 5′-end. The obtained short reads were aligned on the seed sequence to construct a PWM (position weight matrix) on the 3′-side of the matching k-mer, and the seed sequence was extended based on the PWM until there was a split in the graph using stringent thresholds. Therefore, neighboring repeats are not resolvable. After the comprehensive collection of repeat motif patterns, scaffolding was performed using long reads. The long reads were obtained by direct sequencing, so the actual gene length was guaranteed. Finally, mapping the repeat motifs to the corresponding long read and curating the sequence yields the complete spidroin genes.

Phylogenetic Tree of Spidroin.

The phylogenetic tree in SI Appendix, Fig. S6 was constructed based on a core ortholog gene set obtained from BUSCO arthropoda_obd9 (33) using four Nephilinae draft genomes (this study) and the A. ventricosus draft genome (Ave_3.0) (28). The phylogenetic tree of spidroin (SI Appendix, Fig. S7) was constructed by FastTree (version 2.1.10) (59) as an approximately maximum-likelihood phylogenetic tree from aligned and trimmed N-terminal sequences (150 residues from the start codon) with MAFFT (version 7.273) (60) and trimAl (1.2rev59) (61). The accession numbers for known spidroin sequences are described SI Appendix, Supplementary Methods.

Gene Expression Analysis.

Gene expression profiling was conducted with multiple biological replicates from mRNA extracted from the abdomen or silk glands. The gene expression levels were quantified and normalized as transcripts per million by mapping processed reads to our assembled draft genome references with Kallisto version 0.42.1 (62) (SI Appendix, Datasets S2–S4).

Silk Collection.

The natural spider silks were sampled directly from adult female T. clavata restrained using two pieces of sponge and locked with rubber bands. Silk reeling from the spider was performed at a constant speed (1.28 m/m for 1 h) with a reeling machine developed by Spiber Inc. with six biological replicates and three technical replicates each. Since the spinnerets are very close together and many silks may be spun at the same time, simple silk reeling may involve minor ampullate silk as well. Therefore, to identify the proteins that were significantly present in minor ampullate silk and remove them as background, we separated the dragline and minor ampullate silks clearly from their spinnerets using a stereomicroscope (Leica S4E, Leica Microsystems GmbH) (SI Appendix, Fig. S10).

SDS-PAGE Analysis.

SDS-PAGE analysis of dragline silk was conducted at IDEA Consultants, Inc. The dragline silk reeled from T. clavate was dissolved in 2 mL of ionic liquid (1-butyl-3-methylimidazolium acetate) per mg. After vortexing for 2 min, the sample was incubated at 100 °C for 15 min. The ionic liquid solution was dialyzed into lysis buffer (6 M urea, 2 M thiourea, 2% CHAPS, 1% DTT) using a 3k MWCO cellulose membrane (Amicon Ultra, Millipore).

The HMW fraction of 0.33 mg dragline silk was dissolved in the ionic liquid and filtered through a 50k MWCO membrane (Amicon Ultra, Millipore), collected in 100 µL of lysis buffer, and incubated at 95 °C for 5 min. Then, 2.5 µL and 5 µL samples were applied to a 4% SDS-polyacrylamide gel. Electrophoresis was performed at 20 mA for 80 min in running buffer (25 mM Tris, 192 mM glycine, 0.1% SDS), and a HiMark Unstained Standard (LC5688, Life Technologies) was used as a molecular marker. After electrophoresis, the SDS gel was fixed in fixing solution (30% MeOH, 5% acetate acid), stained with SYPRO Ruby overnight, and washed with a wash solution (10% MeOH, 7% acetic acid). Silver staining was conducted after fixation with 50% MeOH and sensitization with 0.005% sodium thiosulfate (Fig. 2A).

A solution of 3.97 mg dragline silk dissolved in 80 µL Tris⋅HCl (pH 8.6), 1% Triton-X, and 2% SDS was vortexed, incubated for 1 h on ice, and freeze-dried. After resuspension, the total volume was applied to a 12.5% SDS-polyacrylamide gel. Electrophoresis was performed at 20 mA for 80 min, with Broad Range Protein Molecular Weight Markers (V8491, Promega) used as molecular markers. After electrophoresis, the SDS gel was fixed in fixing solution (30% MeOH, 5% acetic acid), stained with SYPRO Ruby overnight, and washed with a wash solution (10% MeOH, 7% acetic acid). Silver staining was conducted after fixation with 50% MeOH and sensitization with 0.005% sodium thiosulfate (Fig. 2A).

Liquid Chromatography Mass Spectrometry Analysis.

Approximately 1.0 mg of dragline silk was used for the liquid chromatography mass spectrometry (LC-MS) analysis. Using the iBAQ score, we identified the proteins specifically expressed in the minor ampullate silk based on significance (false discovery rate < 0.01) and removed them from the result of the dragline silk proteome analysis as contaminants. The percentage of each protein in silk was calculated based on the iBAQ multiplied by the amino acid length. The details of the LC-MS analysis used for each spider silk are described in SI Appendix, Supplementary Methods.

Recombinant Protein Expression and Purification

Sequence-optimized recombinant T. clavata MaSp family and SpiCE-NMa1 proteins were used for the composite film and silk. Targeting a size of ∼50 kDa, we reduced the number of repeat units to 14 for MaSp1, 9 for MaSp2, and 8 for MaSp3 while keeping the N- and C-terminal domains (SI Appendix, Fig. S13). The recombinant spidroin genes are composed of a 6x His tag (MHHHHHH), a linker (SSGSS), an HRV 3C protease recognition site (LEVLFQGP), an N-terminal domain, repeat units, and a C-terminal domain. The SpiCE-NMa1 gene is also composed of a 6x His tag, a linker (SSGSS), an HRV 3C protease recognition site, and a T. clavata comp1999 protein sequence. Each signal peptide was removed based on SignalP-5.0 (63). Nucleotide sequences were adjusted to the codon usage of Escherichia coli. Fragmented sequences were chemically synthesized by Fasmac Co., Ltd. The synthesized sequences were assembled with overlap extension PCR (64). Assembled sequences were cloned into pET-22b(+) and transformed into E. coli BLR (DE3). A 100-mL bacterial culture (5.0 g/L glucose, 4.0 g/L KH₂PO₄, 6.0 g/L yeast extract, and 0.1 g/L ampicillin) was grown at 30 °C to an optical density at 600nm (OD₆₀₀) of 5.0 and used for inoculation. The cells were cultured in a 10 L jar fermenter (Takasugi Seisakusho Co. Ltd.) containing 5.7 L of medium (12.0 g/L glucose, 9.0 g/L KH₂PO₄, 15 g/L yeast extract, 0.04 g/L FeSO₄ 7H₂O, 0.04 MnSO₄ 5H₂O, 0.04 CaCl₂ 2H₂O, GD-113; Nof Corporation) at 37 °C. The initial OD₆₀₀ was 0.05. After the initial glucose was consumed, a feeding solution containing 65.9 wt/vol% glucose was pumped into the fermenter at a feeding rate of 50 mL/h. The dissolved oxygen concentration was kept at 20% air saturation. The pH was kept at pH 6.9. After 24 h, protein expression was induced by adding isopropyl-β-D-thiogalactoside (IPTG) to a final concentration of 0.1 mM. The cells were harvested by centrifugation at 11,000 × g for 15 min at 24 h after induction. For the recombinant spidroins, the harvested cells were washed out with 20 mM Tris⋅HCl (pH 7.4) and centrifuged at 11,000 × g, and the supernatant was discarded. The cells were resuspended in 20 mM Tris⋅HCl (pH 7.4), 100 mM NaCl (pH 8.0) with 0.9 µg of DNase per wet cell weight (g), 82 µg of lysozyme per wet cell weight (g), and 0.2 mL of phenylmethylsulfonyl fluoride (PMSF) per wet cell weight (g). Cells were agitated overnight at 37 °C. The cells were resuspended in buffer solution (50 mM Tris⋅HCl, 100 mM NaCl [pH 8.0]) with 3 wt/vol% SDS, and the cell suspension was centrifuged at 11,000 × g for 30 min at room temperature. The pellets were dissolved in DMSO containing 1 M LiCl at 60 °C for 30 min. The solution was centrifuged at 11,000 × g for 30 min at room temperature. Ethanol precipitation was performed at room temperature. The mixture was centrifuged at 11,000 × g for 10 min. The precipitate was washed with reverse-osmosis purified (RO) water three times, followed by lyophilization. For SpiCE-NMa1, the harvested cells were suspended in sodium phosphate buffer (50 mM sodium phosphate, 300 mM NaCl [pH 7]) with 7.5 M urea, PMSF, and DTT and then lysed by a high-pressure homogenizer (GEA Niro Soavi), followed by centrifugation. The supernatant was filtered through a 0.44-µm filter and loaded onto a column packed with Nuvia IMAC (Bio-Rad). After loading, the resin was washed with 100 mM sodium phosphate buffer with 7.5 M urea, DTT, Triton-X100, and 15 mM imidazole and then eluted using a linear gradient of imidazole concentrations from 15 to 500 mM. The eluents were collected and dialyzed against buffer A (20 mM Tris⋅HCl, 7.5 M urea, 50 mM NaCl, DTT and Triton-X100 [pH 7.4]). The dialyzed sample was loaded onto a 5-mL Bio-Scale Macro-Prep High Q Column (Bio-Rad). The resin was washed with 20 mM Tris⋅HCl buffer containing 7.5 M urea, 80 mM NaCl, 1 mM DTT, and Triton-X100 (pH 7.4) and then eluted using a linear gradient of NaCl concentration from 80 mM to 1 M. The eluents were collected and loaded onto Nuvia IMAC resin again and eluted using a linear gradient of imidazole concentration from 15 to 500 mM. The eluents were dialyzed against RO water and then lyophilized.

Recombinant Films.

HFIP was used as the solvent for the film synthesis. Recombinant MaSps were mixed in 8 mL HFIP to a total of 100 mg and then dried at room temperature for 16 h in a dish (SI Appendix, Fig. S14). Composite films containing SpiCE-NMa1 were produced by mixing the MaSp2 solution with recombinant SpiCE-NMa1 at 1, 3, or 5 wt% each.

Recombinant Silks.

Recombinant MaSp2 protein of 22 wt% (1.66 g) was dissolved in 5.89 g DMSO (4 wt% LiCl) and incubated at 90 °C for 8 h. For composite silk containing SpiCE-NMa1, recombinant MaSp2 protein of 22 wt% and recombinant SpiCE-NMa1 of 0.012 g were dissolved in 4.2 g DMSO. The dope was aspirated with an N₂ pump and D = 0.1 mm needle at 70 °C and spun directly into baths containing MeOH/DMSO 30 vol/vol% (5 °C), MeOH (25 °C), and water (25 °C). Only MaSp2/SpiCE-NMa1 composite silk could be produced because of the low yield of recombinant MaSp1 and MaSp3 proteins.

WAXS Measurement.

The crystalline samples were characterized by a synchrotron WAXS measurement at the BL45XU beamline of SPring-8, Harima, Japan (43). The X-ray energy was set at 12.4 keV (wavelength: 0.1 nm), and the distance between the sample and the detector was ∼257 mm. For the diffraction patterns, an exposure time of 10 s was used. Using Fit2D software (65), the obtained data were converted into one-dimensional radial integration profiles and corrected by subtracting the background scattering. The degree of crystallinity was calculated from the area of the crystal peaks divided by the combined area of the crystal peaks and the amorphous halo by fitting the Gaussian function using Igor Pro-6.3.

Measurement of Mechanical Properties.

The tensile properties of the fibers were measured using an EZ-LX universal tester (Shimadzu) with a 1 N load cell at a strain rate of 10 mm/min (0.033 s⁻¹) at 25 °C and 48% relative humidity. For each tensile test, the cross-sectional area of an adjacent section of the fiber was calculated based on the SEM images. The transparency of the film was measured at selected wavelengths between 350 and 600 nm using a UV instrument. Each fiber was attached to a rectangular piece of cardboard with a 5-mm aperture using 95% cyanoacrylate.

Tensile tests of the recombinant silk fibers were performed using a mechanical testing apparatus (Automatic Single-Fiber Test System FAVIMAT+, Textechno H. Stein GmbH & Co. KG) with a 210 cN load cell at a strain rate of 10 mm/min (0.016 s⁻¹) at 20 °C and RH 65%. From each fiber sample, diameters were measured at different sites (n = 6) by microscopic observation (Nikon Eclipse LV100ND, lens 150 × 0) to calculate the cross-sectional area of an adjacent section of the fibers.

Data Availability

Genome assembly, transcriptome sequencing, and proteome analysis data have been deposited in DNA Data Bank of Japan (DDBJ) and jPOST. All raw reads and assembled sequence data have been uploaded to DDBJ under BioProject nos. PRJDB10007, PRJDB10126, PRJDB10127, PRJDB10128, PRJDB7399, and PRJDB11252. The whole genome sequence is available at the Whole-Genome Shotgun database in DDBJ under accession nos. BMAO01000001–BMAO01039792 (T. clavata), BMAU01000001–BMAU01021438 (T. clavipes), BMAV01000001–BMAV01028263 (T. inaurata madagascariensis), and BMAW01000001–BMAW01132898 (N. pilipes). The MS raw data and analysis files have been uploaded to the ProteomeXchange Consortium from the jPOST partner repository under accession nos. PXD024371, PXD024372, PXD024374, and PXD024376.