Analysis of fw2.2 gene structure and phylogenetic analysis of Lycium ruthenicum
The full length of the DNA of Lycium ruthenicum fw2.2 is 2323bp, it is divided into three exon regions by introns, the lengths of which are exon 1 :259bp, exon 2:210bp, exon 3 :80bp. The cDNA sequence is 549bp in length, with a total coding of 182 amino acids. The fw2.2 gene of Lycium ruthenicum has a PLAC8 (Placenta-specific 8) conserved domain ( Fig. 1a), and belongs to the same family as fw2.2 of other plants. After BLASTp alignment on NCBI, some homologous protein sequences are obtained. Most of them were FWL/CNR family proteins, and few of them are PCR (Plant cadmium resistance) family proteins ( Fig. 1b). The two family proteins have the same conserved domain PLAC8 and similar structures. The selected Solanaceae fw2.2 homologous protein including Lycium ruthenicum fw2.2, Lycopersicon fw2.2, Physalis POS2, and Capsicum CNR1 are clustered in the same branch, and the amino acid sequence similarity is more than 85% ( Fig. 1C), indicating that fw2.2 protein is relatively conserved in Solanaceae. FW2.2-like genes have been renamed as the Cell Number Regulator (CNR) family, and it has a negative regulatory effect on cell number[17, 18]. Studies have shown that fw2.2 can explain 30% and 47% of the fruit size phenotypic variation in Lycopersicon pimpinellifolium and Lycopersicon pennellii, respectively[19]. The POS2 (physalis organ size 2) in Physalis floridana encodes a putative ortholog of fw2.2, which can regulate the cell cycle and has a negative effect on fruit size[20]. The function of CNR1 in pepper is unknown. Tomato fw2.2 and Physalis POS2 both have the functions of regulating cell division and affecting fruit size. It is speculated that Lycium ruthenicum fw2.2 which has a close phylogenetic relationship with them may also have similar functions.
Establishment of Lycium ruthenicum regeneration system
The seeds of Lycium ruthenicum are disinfected and inoculated on 1/2 MS medium, we have established a suitable regeneration system of Lycium ruthenicum after the induction and differentiation of callus, and the rooting of regenerated seedlings (Additional file 1 Table S1). Although the callus induction rate in each medium is 100% after 15 days of callus induction, the callus growth status is different. Among them, callus grows best on A3 medium (MS + 0.5 mg/L 6-BA + 0.5 mg/L NAA) with green appearance, loose structure, no browning, and low vitrification, which is the most suitable medium (Additional file 1 Table S1). The callus was subsequently transferred to the medium B1-B10 to induce differentiation, and the differentiation rate and multiplication coefficient were counted after 30 days. We found that the callus did not differentiate or the differentiation rate was low when it was cultured on the medium without 6-BA (B5) or the concentrations of 6-BA were higher (B1-B4 ). Both the differentiation rate and the multiplication coefficient were significantly increased on the medium supplemented with low concentration of 6-BA (less than 0.5mg/L). When the callus was cultured on B7 (MS + 0.2 mg/L 6-BA + 0.05 mg/L NAA) medium for cluster shoot induction, the differentiation rate and multiplication coefficient were the highest, which were (95.83 ± 2.23 )% and (7.06 ± 0.22)% respectively. The callus did not brown and the degree of vitrification was relatively low (Additional file 1 Table S1). The differentiated shoots were then inoculated on 1/2 MS medium without hormones, and rooting can be induced after 15 days.
Target site selection and sgRNA Design
The CRISPR online design tool (http://crispr.dbcls.jp/ ) was used to determine the location of the target, the genome of tomato ( Solanum lycopersicum genome, SL2.40 ), a species with high homology with Lycium ruthenicum, was used as a reference to improve target specificity. Based on the selection criteria that the PAM site sequence is NGG, the GC content is 40%-70%, the target sequence avoids spanning intron regions and the occurrence of more than 4 consecutive T bases, the top sgRNA were selected for the knockout experiments. The sgRNA used in this study are shown in Table S2. The position of fw2.2-1 and fw2.2-2 is 1857-1879 and 171-193 respectively, spanning 1664 bases.
Establishment of Lycium ruthenicum genetic transformation system and efficiency evaluation of CRISPR/Cas
Two single sgRNAs and a dual sgRNAs (sgRNA1 and sgRNA2) of fw2.2 were designed that target different sites after the whole sequence being cloned and phylogenetic analyzed. The target site of fw2.2-sgRNA1 and fw2.2-sgRNA2 is located in exon 2 and exon 1 respectively ( Fig. 2a). The high efficient Agrobacterium tumefaciens-mediated transformation of black wolfberry was performed here using the leaves under the condition of 0.2 Agrobacterium concentration (OD600), 10 min infection time, 200 μmol/L acetosyringone concentration and 2d co-culture time (Additional file 1 Table S3). The 40 mg/L hygromycin selection marker plus 200 mg/L carbenicillin were used to inhibit Agrobacterium growth. The entire experimental cycle took approximately 2 months from incubation to mutant identification with 2d co-cultivation, 15d calli production, 30d differentiation and sub-culture (Fig. 2b). The details of the transformation media can be found in Fig. 2c. The mean transformation efficiencies of these three lines were 2.66%, 1.18% and 5.33%, respectively.
In this study, a lot of resistant seedlings were produced. Twenty-one out of twenty-two, six out of eleven and fifteen out of sixteen bigger plants were detected with novel mutations in the sgRNA1, sgRNA2 and sgRNA1/sgRNA2 regions. As shown in Fig. 2d, the gene editing efficiency of fw2.2-1 target is high (95.45%), while that of fw2.2-2 target is low (54.55%). However, the editing efficiency of homozygous mutants (18.18%) and biallelic mutants (9.09%) is higher than those of fw2.2-1 (4.55% ).
The dual-sgRNA CRISPR/Cas9 system was highly reproducible and highly efficient since it could result in more reliable loss-of-function alleles that lack a large essential part of the gene[21]. Here we found the editing efficiency of fw2.2 in the homozygote/biallelic mutations altogether (56%) by the dual-sgRNA CRISPR/Cas9 system is more than twice (27%) of that by the sgRNA CRISPR/Cas9 system though the editing efficiency of the dual-sgRNA system is only 93.75%,which is a little less than that of (95.45%) the sgRNA1 system (Fig. 2e&f). It was also found that there was a 1281 bp segment deleted and 29bp insertion in the fw2.2 of a T0 plant (Fig. 2g), which is similar to the result such as 934-bp deletion mutation at the AtMIR169a locus of Arabidopsis[21] .
Expression analysis of fw2.2 in gene-edited seedlings by quantitative real-time PCR
We found that the expression of the fw2.2 gene changed significantly (Fig. 3). Among the selected gene-edited seedlings, the expression of fw2.2 gene decreased most significantly in X21, which had a large deletion of 1281 bp, and its expression was only 0.01. X9 is a homozygous mutant, the expression of fw2.2 gene in which was only 0.03, and followed by X17 and X20, which were 0.06 and 0.15 respectively. When a gene expression cassette is introduced into a genome by CRISPR/Cas9, the sgRNA target gene becomes inactive because of disruption of the gene, which probably influences the gene expression. These data suggest that CRISPR/Cas9 applied to fw2.2 had a significant effect on reducing gene expression , which probably influences the developmental characteristics of the modified strain if the inactive gene is generally involved in metabolism.