New insights into the evolution of spider silk proteins illuminated by long-read transcriptomes

Reviewer #1 (Public review):

Summary:

In this study, Zhu et al. address spider silk spidroin evolution using long-read transcriptomics across 12 spider species. The study provides a novel evolutionary framework for spidroin diversification, proposing the existence of two ancient ancestral templates, i.e., AS and GS, and tracing how these templates diversified into major spidroin classes observed in radiated spiders. The manuscript further focused on the evolutionary history of multiple known spidroin proteins, with some previous hypotheses being revised.

Strengths:

A major challenge in silk biology, the highly repetitive content, was well addressed in this study by full-length transcriptome sequencing. Also, the authors performed very detailed analyses on sequence features across a wide range of species. I therefore think the study is supported by sound levels of sampling, technology, and analysis.

Weaknesses:

The manuscript presents a lot of detail regarding various sequence features and derived claims, but these features are sometimes not friendly to an audience not working with spider silks. Also, the current figures are not very helpful for understanding those described patterns. I found many colorful, trivial elements in almost every figure, but how their organization supported the corresponding statement was often unclear to me. I recommend that the authors further improve the figure design, including presenting a schematic evolutionary history for those spider silk proteins.

Reviewer #2 (Public review):

Summary:

This paper utilizes long-read transcriptomics across 12 representative spider species to propose a new evolutionary framework for spider silk proteins (spidroins). By identifying ancestral templates in the most basal spider lineages, the authors trace how simple genetic materials diversified into the high-performance fibers used by modern spiders.

Strengths:

(1) The authors utilized PacBio ISO-Seq (long-read transcriptomics), which is essential for resolving the massive, highly repetitive sequences of spidroin genes that often cause gaps in traditional short-read assemblies.

(2) The researchers sampled 12 species representing the major nodes of spider evolution, including the basal Mesothelae, the Mygalomorphae (tarantulas), and the highly diverse Araneomorphae.

(3) The study successfully identified two distinct primordial spidroins in basal spiders: the AS-type (alanine-serine-rich) and the GS-type (glycine-serine-rich) proteins.

Weaknesses:

(1) The GS-Type "Base Gene" Paradox

The paper proposes that the GS-type gene (Liphistius sp._5400) in Liphistius (the most ancient spider lineage) is the prototype for all modern dragline silk. However, the data presented significantly undermines this conclusion.

Every functional spider silk protein requires N-terminal and C-terminal domains to control fiber assembly. The authors admit that neither the N- nor the C-terminal of this GS-type protein shows homology to any known spidroins. Because it lacks these domains, the authors explicitly state that it "may not assemble into typical silk fibers". The authors are identifying this as a "base gene" solely because it contains poly-GS motifs. Their logic is that because GS motifs are found in modern silk and other silk-producing insects, this must be the ancestor.

In the same spider, the AS-type gene (Liphistius sp._6705) does have recognizable C-terminal sequences and motifs similar to modern eggcase silk. This proves that "real" spidroins existed in Liphistius, making the claim that the non-homologous GS-type is a "spidroin ancestor" look like a misidentification of a general repetitive protein.

(2) Overstated Classification of FLAG in RTA Spiders

The authors identified a transcript in the RTA spider Heteropoda davidbowie (H.dav_6495) and labeled it a "Flag-like spidroin". This label is based on the repetitive internal motifs, which contain "GPGGX" and "GPG"-the classic building blocks of flagelliform capture silk. However, both the N- and C-termini of this gene are highly homologous to ampullate spidroins (MaSp), not typical Flag proteins. By calling it a "Flag-like spidroin" rather than a "MaSp with GPG motifs," the authors are forcing an evolutionary narrative. It is equally possible that this is simply a divergent Major Ampullate spidroin that evolved capture-like motifs, rather than a capture silk gene that "moved" into the ampullate gland.

The authors explicitly state, "Its origin could not be traced through sequence analysis". This admission directly contradicts the confidence with which they propose a "revised evolutionary trajectory".

Appraisal and Impact

This study provides a high-resolution map of spider silk evolution by utilizing long-read transcriptomics to bridge the gap between basal and derived lineages. By identifying the earliest known genetic templates for silk, the paper offers a significant leap forward in understanding how complex biological materials originate, though it raises critical questions about the functional definition of a "spidroin".

Reviewer #3 (Public review):

Summary:

In this study, Zhu et al. use long-read transcriptomes, with correction using short-read RNA-seq, from 12 spider species that span the major evolutionary lineages to investigate the diversification of spider silk proteins (spidroins). Here, they identify 60 spidroin sequences and propose that two highly divergent sequences found in the basal Liphistius sp., where one is an alanine-serine-rich (AS-type), and one is a glycine-serine-rich (GS-type), represent ancestral templates from which all major spidroin families diversified. Using separate phylogenetic analyses for N-terminal domains, C-terminal domains, and repetitive domains, the authors argue that the AS-type lineage remained relatively conserved and gave rise to tubuliform spidroins (TuSp) used in eggcase silk, while the GS-type lineage evolved into minor ampullate spidroins (MiSp) and may have provided the substrate for major ampullate spidroins (MaSp). In addition, they describe a specific flagelliform-like (flag) transcript in a basal clade spider, with MaSp-like terminal domains, and propose that Flag was co-opted into ampullate silk glands before being progressively lost in more derived retrolateral tibial apophysis (RTA) lineages.

Strengths:

The taxon sampling is a strength of this study, covering representative species at key nodes across spider evolution, from the earliest-diverging Mesothelae through Mygalomorphae and into the most derived Araneomorphae lineages, which enables the authors to make comparative inferences about ancestral states. Also, the use of long-read sequencing is well-suited to the problem since spidroin genes contain highly repetitive coding sequences that would be very hard to resolve by short-read assembly alone. Thus, retrieving 30 full-length sequences in this context is notable, and the assembly quality appears reasonable for transcriptomic resources, with BUSCO completeness values reported between 85% and 93% across species.

The decision to analyse N-terminal, C-terminal, and repetitive domains in separate phylogenetic trees is methodologically sound and yields a biologically interesting result: terminal domains show greater diversification in basal lineages than repetitive regions, suggesting that specialisation of silk gland microenvironments preceded compositional innovation in the repetitive sequences.

Weaknesses:

While the paper has strengths in providing a useful comparative resource and generating interesting hypotheses, several of the central evolutionary conclusions are not directly supported by the current data. There are three main elements that require further attention:

(1) The GS-type Liphistius sequence (Liphistius sp._5400) is central to the manuscript's model for the origin of GA-rich ampullate spidroins, but the authors describe it as a spidroin-like transcript whose N- and C-terminal regions lack homology to known spidroins and may not support typical silk-fiber assembly. Since its terminal domains are excluded from the phylogenetic analyses, the proposed scenario, GS-type to MiSp to MaSp, rests largely on repeat-region similarity. Supplementary materials provided in this study further indicate no predicted signal peptide, although this feature alone is not unique among the annotated silk proteins. The manuscript should therefore either provide a stronger justification for treating Liphistius sp._5400 as an ancestral spidroin or more consistently frame it as a spidroin-like, repeat-based intermediate. The distinction between repeat-region clustering and full functional homology should also be made explicit.

(2) The whole-body transcriptome approach is an important sampling limitation that is acknowledged here, where the authors note that they were unable to recover complete spidroin repertoires for each species. Because the newly generated data are not silk-gland-specific, the absence of a transcript in a given species should be interpreted with caution and not equated directly with gene absence. This is particularly relevant to the manuscript's proposed loss of Flag during RTA evolution. In the focal taxa, the inference combines one positive transcript in H. davidbowie with non-detection in H. diardi, while broader support comes from limited synteny-based absence in a small number of external genomes. Therefore, while the Flag-loss scenario could be plausible, it remains suggestive rather than conclusive without more targeted silk-gland sampling or broader genomic validation.

(3) The Flag co-option model is interesting, but as presented now, it is based on limited evidence: a single Flag-like transcript in H. davidbowie, the absence of detection in H. diardi, restricted synteny comparisons, and terminal-domain similarity to ampullate spidroins. The manuscript does not present proteomic evidence that this Flag-like protein is incorporated into ampullate silk fibers, nor does it show a series of pseudogenized or truncated Flag loci across derived RTA lineages. This is a plausible and interesting scenario, but it should be framed more consistently as a testable hypothesis rather than as an established evolutionary pathway.