Arrays of TAL effector nucleases targeting the silkworm genome were designed using TAL Effector Nucleotide Targeter 2.0 (https://tale-nt.cac.cornell.edu/) (Doyle et al., 2012). The construction of TALEN vectors were performed by Golden-Gate assembly according to the instruction provided by the manufacturer of the Golden Gate TALEN and TAL effector kit 2.0 (Addgene number: 1000000024) (Cermak et al., 2011). All plasmids were verified by Sanger sequencing.

The HR donor contained a left homology arm and right homology arm amplified and sequenced from the silkworm genome; the TALEN target site was placed at the 5′end of the left arm and the 3′end of the right arm. The CP coding sequence was seamlessly designed behind the left arm with the poly(A) sequence of FibH. Those above fragments were synthesized and cloned into the pUC-GW-Amp vector (GENEWIZ) to generate a pUC-L-CP-R plasmid. The HR3-IE1-EGFP-SV40 cassette was cut from the pBac[IE1-EGFP]-2×PySp1-A.arg plasmid (stored in our laboratory) using AflⅡ restriction endonuclease and cloned into the pUC-L-CP-R vector to generate a pUC-FibL–CP–EGFP HR-mediated donor. The donor vector was verified by Sanger sequencing.

TALEN vectors were linearized by XbaⅠ restriction endonuclease digestion and purified with phenol/chloroform/isoamylalcohol (25:24:1); then, TALEN mRNAs were synthesized in vitro with an mMESSAGE mMACHINE™ T7 Kit (Invitrogen), respectively. The donor vector and TALEN mRNAs were purified with phenol/chloroform/isoamylalcohol (25:24:1) and dissolved in 1× phosphate-buffered saline (1×PBS buffer). TALEN mRNA (300 ng/μL) and the donor vector (200 ng/μL) were mixed and injected into Lan10 silkworm embryos within 8 h after oviposition. The injected eggs were incubated at 25 °C, 90% humidity, until hatching. The G0 larvae were reared with fresh mulberry leaves until the cocoon stage. G0 moths were mated with wild-type (WT) individuals, and positive larvae in G1 were screened under a green fluorescence microscope (Olympus, Tokyo, Japan).

Genomic DNA was extracted from the PSGs of positive and WT individuals using an Ezup Column Animal Genomic DNA Purification Kit (Sangon Biotech), and the 5′and 3′junction were confirmed by PCR amplification. Sanger sequencing was performed directly on the amplification products. Quantitative real-time PCR (qRT-PCR) analysis was performed as previously described (You et al., 2018). Briefly, total RNA was extracted from PSGs of positive and WT individuals on the third day of the fifth instar larvae using TRIzol reagent (Invitrogen). The cDNA was synthesized using HiScript III RT SuperMix for qPCR (+ gDNA wiper) (Vazyme). Relative transcript levels of FibL and FibL–CP were determined by qRT-PCR using ChamQ Universal SYBR qPCR Master Mix (Vazyme). The PCR reaction was performed according to the manufacturer’s instructions, all of the primers used are listed in the (Supplementary Table S1), and three independent replicates of each sample were used.

Proteins were extracted according to a previously described method with some modifications (Wu et al., 2021). Briefly, silk gland and cocoon shell proteins were extracted with sodium dodecyl sulfate (SDS) buffer (1:40, wt/vol) containing 5% (vol/vol) β-mercaptoethanol, and the supernatants were collected by centrifuging after incubation at 37 °C for 8 h. The extracted proteins were quantified using a Modified BCA Protein Assay Kit (Sangon Biotech); then, western blotting analysis was performed. FibL and FibL–CP proteins were detected using FibL polyclonal antibodies as primary antibodies (1:3,000 dilution, GenScript), and horseradish-peroxidase-conjugated goat anti-rabbit IgG was used as the secondary antibody (1:5,000 dilution, Sangon Biotech). The signal was detected using highly sensitive Plus ECL Reagent (Sangon Biotech). ImageJ software was used to quantify band intensities.

Ten cocoons were randomly chosen from each of the WT and FibL–CP strains, and three pieces of silk were drawn from each cocoon and tested. Silk fibers were prepared and tested as described previously (Perez-Rigueiro et al., 2000; Zhao et al., 2007; Zhao et al., 2021). Briefly, the diameter of each silk fiber was measured using a three-dimensional digital microscope (Keyence), and the cross-sectional area of the silk was calculated as the average of five measurements. Tensile tests were performed on each silk fiber using an AGS-J Universal Test instrument (Shimadzu) with a cell load of 5 N. The toughness of the silk fiber was calculated by integrating the area under the stress–strain curves using mathematical software (Origin 2019b).

The three-dimensional structure of the FibL monomer (AFDB accession: AF-P21828-F1) and the C-terminal domain (CTD) of Fib-H (AFDB accession: AF-Q1KS45-F1) were directly obtained from the AlphaFold Protein Structure Database, and the three-dimensional structure of FibL–CP was predicted using the I-TASSER online server (Yang and Zhang, 2015; Zheng et al., 2021; Zhou et al., 2022). Molecular docking of the CTD of FibH to the FibL and FibL–CP proteins, respectively, was performed using the HADDOCK online server (Wassenaar et al., 2012; van Zundert et al., 2016). The best docking results were visualized and analyzed using PyMOL software.

The experimental data were analyzed with Student’s t-test (*p < 0.05, **p < 0.01, ***p < 0.001, and n.s for p > 0.05). Correlation analysis and linear regression analysis were performed using Excel software. In the figures, error bars indicate the mean ± standard error of the mean.

We constructed a TAL effector by Golden-Gate assembly and ligated it into the VK006-06 vector (ViewSolid Biotech), which contains a T7 promoter for in vitro transcription. Two TALEN vectors containing different FOK Ⅰ enzyme form heterodimers to improve the efficiency of DNA cutting. In the 3′untranslated region of FibL, we designed two target sites. Three days after injection of TALEN mRNAs into silkworm eggs, the genomic DNA was extracted, and a DNA fragment near the targeting site was amplified for sequencing. One target site tested positive (Figure 1), whereas the other was inactive (Supplementary Figure S1). We used the positive target site for subsequent experiments.

To achieve fusion expression of FibL and CP, we designed a donor vector containing the CP gene (Figure 2), and an 834-bp 5′-homologous arm and an 812-bp 3′-homologous arm were designed to facilitate HR. The CP gene was seamlessly inserted at the end of FibL to achieve FibL–CP fusion expression. Early studies have reported that linear DNA can improve the efficiency of HR (Wang et al., 2014; Takasu et al., 2016). Therefore, TALEN targeting sequences were added to both sides of the donor vector to achieve linearization in vivo. The TALEN mRNAs and donor plasmid were mixed and microinjected into silkworm eggs. Under the repair template we provided, HDR repair occurred, and the CP gene was site-directed inserted into FibL.

In the second instar larvae of G0, many lumps of fluorescence on the body surface of silkworm could be observed under the green fluorescence microscope (Figure 3A). G0 moths were crossed with WT, and positive larvae of the first instar of G1 were screened under a fluorescence microscope (Figure 3B). We screened six EGFP-positive G1 broods; the transformation efficiency was 9.2% (Table 1). Green fluorescence could be seen throughout the lifecycle (Figure 3C). Genomic DNA was extracted from mated G1-positive moths, and the 5′and 3′junctions were amplified. Sanger sequencing was performed directly on the amplification products; the results showed that the insertion was precise and seamless (Figure 3D and Supplementary Figure S2). The primers used for amplification are listed in Supplementary Table S1. The FibL–CP heterozygous animals showed successful spinning and cocooning, but the cocoon shells were a little smaller than those of the WT (Figure 3E). We weighed the cocoon shells of the FibL–CP and WT silkworms. In both females and males, the average cocoon shell weight of the FibL–CP silkworm was markedly decreased compared with that of the WT. However, the average weights of female and male pupae did not show obvious changes between the FibL–CP and WT silkworms (Figure 4). FibL–CP homozygotes showed abnormal spinning and cocooning (Supplementary Figure S3). These results indicate that insertion of the CP gene may affect the spinning of silkworm.

FibL has long been of interest as one of the components of silk protein, and FibH, another component of silk protein, has previously been modified by gene editing (Xu et al., 2018; Junpeng et al., 2023; Takasu et al., 2023). However, there has rarely report of site-specific knock-in of FibL by HDR. Here, the CP gene from silkworm was inserted into exon 7 of FibL by TALEN-mediated HDR. The relative mRNA levels of FibL and FibL–CP in the PSG of the FibL–CP transgenic silkworm were determined by qRT-PCR. In FibL–CP animals, abundant expression of the FibL–CP gene was detected, whereas we did not detect any expression of this gene in WT animals (Figure 5B); therefore, endogenous FibL transcriptional levels were markedly decreased in the FibL–CP heterozygous silkworm compared with WT (Figure 5A). Subsequently, we extracted total protein from the PSG of the FibL–CP and WT strains, and detected the expression of FibL–CP protein. The theoretical molecular weight of FibL–CP protein is 57.587 kDa. In the FibL–CP heterozygote, both FibL–CP and FibL proteins were expressed; by contrast, only FibL–CP protein was expressed in the homozygote, and only FibL protein was expressed in the WT animal (Figure 5C). Grayscale analysis of the bands showed that the content of FibL–CP protein was comparable with that of the native FibL in the heterozygote (Supplementary Figure S4). In addition, we extracted total protein from the cocoon shell. SDS–polyacrylamide gel electrophoresis (PAGE) and western blotting analysis revealed that FibL–CP protein failed to be secreted into the cocoon shell (Figure 5D). The homozygote showed an abnormal cocoon shell owing to the absence of FibL (Supplementary Figure S3), with a phenotype similar to that of fibroin-deficient line Nd-s ^D (Takei et al., 1987). These results indicate that the fusion of FibL with CP was successfully transcribed and translated as a new protein in the PSG of FibL–CP strains. Simultaneously, the fusion of CP may have influenced the fibrosis of silk protein and affected spinning and cocooning in FibL–CP strains.

To further verify the effects of a reduction in FibL levels on silk fiber formation, the silk properties of the FibL–CP heterozygote were measured. The average diameter of FibL–CP silk fibers was 25.9 µm, compared with 25.3 µm for the WT; that is, there was no significant difference between the two strains (Figure 6A). However, the FibL–CP silk fibers had an average breaking strength of 129.1 MPa, compared with 199.4 MPa for the WT, a significant decreased. FibL–CP silk fibers could extend up to 21.9% on average, making them slightly more elastic than their WT counterparts (Figure 6B; Supplementary Table S2). SDS-PAGE and western blotting analysis of proteins from the cocoon shells showed that expression of FibL was reduced in the FibL–CP strain, with no FibL detected at all in the homozygote. The new FibL–CP protein containing complete FibL was not detected in the cocoon shells of all strains (Figure 5D). These results demonstrate the importance of FibL in the formation and spinning of silk fibers (Tanaka et al., 1999; Inoue et al., 2000; Inoue et al., 2004). FibL–CP protein was expressed in the PSG but not secreted into the cocoon shell, possibly owing to either the physicochemical properties of CP or the destruction of FibL.

In silkworm, Cys-172 of FibL forms a disulfide bond with Cys-c20 (20th from the C terminus) of FibH, which is required for the intracellular transport and secretion of silk fibroin (Takei et al., 1987; Tanaka et al., 1999). A partial deletion of the FibL gene that caused failure to form the disulfide bond between FibL and FibH resulted in the production of the Nd-s ^D mutant silkworm (Takei et al., 1987). Here, the homozygote of the FibL–CP strain had a similar phenotype to Nd-s ^D , and we hypothesized that this could also be owing to disruption of the disulfide bond between FibL–CP and FibH. Hence, we used molecular docking models to predict the disulfide bond between FibL–CP and FibH. All predicted docking models were clustered by the HADDOCK server; the lower the energy, the more stable the binding. The model with highest score from the low-energy cluster was selected for analysis (Supplementary Figure S5). In the model of FibL–CP docking with the CTD of FibH, we observed that the fusion of CP protein prevented the CTD of FibH from entering the region near Cys-172 of the FibL–CP protein, which would lead to a change in the distance between Cys-172 and Cys-c20. We measured the distance between these two cysteine residues and found that it was 5.9 Å in the WT, but 38.6 Å in the FibL–CP docking model (Figure 7). This significant increase in distance may prevent the formation of the disulfide bond between FibL–CP and FibH, resulting in failure of silk fibroin secretion (Tanaka et al., 1999). Hence, we speculated that the impaired secretion of silk fibroin would further reduce the weight and size of the cocoon shells (Figure 3E; Figure 4A).