Improvement of a Genetic Transformation System and Preliminary Study on the Function of LpABCB21 and LpPILS7 Based on Somatic Embryogenesis in Lilium pumilum DC. Fisch

Auxin transport mediates the asymmetric distribution of auxin that determines the fate of cell development. Agrobacterium-mediated genetic transformation is an important technical means to study gene function. Our previous study showed that the expression levels of LpABCB21 and LpPILS7 are significantly up-regulated in the somatic embryogenesis (SE) of Lilium pumilum DC. Fisch. (L. pumilum), but the functions of both genes remain unclear. Here, the genetic transformation technology previously developed by our team based on the L. pumilum system was improved, and the genetic transformation efficiency increased by 5.7–13.0%. Use of overexpression and CRISPR/Cas9 technology produced three overexpression and seven mutant lines of LpABCB21, and seven overexpression and six mutant lines of LpPILS7. Analysis of the differences in somatic embryo induction of transgenic lines confirmed that LpABCB21 regulates the early formation of the somatic embryo; however, excessive expression level of LpABCB21 inhibits somatic embryo induction efficiency. LpPILS7 mainly regulates somatic embryo induction efficiency. This study provides a more efficient method of genetic transformation of L. pumilum. LpABCB21 and LpPILS7 are confirmed to have important regulatory roles in L. pumilum SE thus laying the foundation for subsequent studies of the molecular mechanism of Lilium SE.


Introduction
Auxin plays a critical role in plant development by controlling cell expansion, polarity, and organ patterning [1]. Auxin is mainly synthesized in expanding leaves along with shoot apical meristem and is transported to various parts of the plant [2] by polar auxin transport (PAT) [3]. Three crucial plasma membrane (PM) auxin transporters are involved in the PAT system, PIN-FORMED (PIN) auxin efflux facilitators [4], AUXIN1/LIKE AUX1 (AUX1/LAX) influx carriers [5], and ATP-binding cassette (ABC) transporter B family (ABCB) efflux/influx transporters [6]. PIN-LIKES (PILS) located at the endoplasmic reticulum (ER) has also been shown to be involved in auxin transport [7,8].
The ABCB subfamily is the second largest ABC protein subfamily in plants with 10,18,21, and 22 ABCB members in Physcomitrella patens, Selaginella moellendorfii, Arabidopsis thaliana, and Oryza sativa, respectively [9]. Some members of the ABCB subfamily have been confirmed to be medium), and the germination conditions of resistant calli. In this study, the genetic transformation efficiency is increased after the improvements, and important regulatory roles of LpABCB21 and LpPILS7 in L. pumilum SE are confirmed.

Construction of Expression Vectors
Complete CDSs (coding sequences) of LpABCB21 (4086 base pairs (bp)) and LpPILS7 (1461 bp) were inserted into the pRI101-ON vector by double enzymes digestion, and the generated vectors were named OE-LpABCB21 and OE-LpPILS7, respectively ( Figure 1A). The sequences information is listed in Supplementary File. efficiency is increased after the improvements, and important regulatory roles of LpABCB21 and LpPILS7 in L. pumilum SE are confirmed.

Construction of Expression Vectors
Complete CDSs (coding sequences) of LpABCB21 (4086 base pairs (bp)) and LpPILS7 (1461 bp) were inserted into the pRI101-ON vector by double enzymes digestion, and the generated vectors were named OE-LpABCB21 and OE-LpPILS7, respectively ( Figure 1A). The sequences information is listed in supplementary file.

Comparison of Genetic Transformation Systems before and after the Improvement
After two rounds of antibiotic screening, the transformed calli were cultured in somatic embryo induction medium for 30 days. The growth of resistant calli is shown in Figure 3. The results indicate A 20 bp sgRNA (small-guide RNA) was selected in the conserved domain of the CDS. The sgRNA sites in LpABCB21 and LpPILS7 are shown in Figure 2. Dual sgRNA expression cassette was successfully constructed by overlapping PCR (polymerase chain reaction). Dual sgRNA expression cassette was inserted into the pG3H-U3Ub vector using single enzyme digestion, and the corresponding vectors were named KO-LpABCB21 and KO-LpPILS7 ( Figure 1B).

Construction of Expression Vectors
Complete CDSs (coding sequences) of LpABCB21 (4086 base pairs (bp)) and LpPILS7 (1461 bp) were inserted into the pRI101-ON vector by double enzymes digestion, and the generated vectors were named OE-LpABCB21 and OE-LpPILS7, respectively ( Figure 1A). The sequences information is listed in supplementary file. A 20 bp sgRNA (small-guide RNA) was selected in the conserved domain of the CDS. The sgRNA sites in LpABCB21 and LpPILS7 are shown in Figure 2. Dual sgRNA expression cassette was successfully constructed by overlapping PCR (polymerase chain reaction). Dual sgRNA expression cassette was inserted into the pG3H-U3Ub vector using single enzyme digestion, and the corresponding vectors were named KO-LpABCB21 and KO-LpPILS7 ( Figure 1B).

Comparison of Genetic Transformation Systems before and after the Improvement
After two rounds of antibiotic screening, the transformed calli were cultured in somatic embryo induction medium for 30 days. The growth of resistant calli is shown in Figure 3. The results indicate

Comparison of Genetic Transformation Systems before and after the Improvement
After two rounds of antibiotic screening, the transformed calli were cultured in somatic embryo induction medium for 30 days. The growth of resistant calli is shown in Figure 3. The results indicate that a higher number of resistant calli masses were generated by the RC5.0 treatment (Re-suspension II and Co-cultivation II) (E-H) than that was generated by the RC5.8 (Re-suspension I and Co-cultivation I) (A-D), and all resistant calli masses had a larger volume and a higher number of clusters.  As shown in Table 1, the number of resistant calli masses obtained by the RC5.0 treatment increased by approximately 2-3-fold. Note: Re-suspended solution and co-culture medium had pH 5.8 (RC5.8) or 5.0 (RC5.0). m/n (%), m represents the number of newly formed calli, n represents the number of inoculated calli, and % represents the percentage of newly formed calli in the inoculated calli.
As shown in Figure    As shown in Table 1, the number of resistant calli masses obtained by the RC5.0 treatment increased by approximately 2-3-fold. Note: Re-suspended solution and co-culture medium had pH 5.8 (RC5.8) or 5.0 (RC5.0). m/n (%), m represents the number of newly formed calli, n represents the number of inoculated calli, and % represents the percentage of newly formed calli in the inoculated calli.
As shown in Figure

Identification of Transgenic Plants
As shown in Figure 5, the transgenic plants were identified by PCR. The NPTII gene fragment (817 bp) and the specific fragment between the vector and the target gene (1105 bp/1004 bp) were successfully cloned, and the positive plants were identified. After verification, 3 OE-LpABCB21 positive lines (L2, L11, and L12) and 7 OE-LpPILS7 positive lines (L5, L13, L17, L25, L27, L30, and L32) were obtained. All lines except L32 were obtained after the improvement.

Identification of Transgenic Plants
As shown in Figure 5, the transgenic plants were identified by PCR. The NPTII gene fragment (817 bp) and the specific fragment between the vector and the target gene (1105 bp/1004 bp) were successfully cloned, and the positive plants were identified. After verification, 3 OE-LpABCB21 positive lines (L2, L11, and L12) and 7 OE-LpPILS7 positive lines (L5, L13, L17, L25, L27, L30, and L32) were obtained. All lines except L32 were obtained after the improvement. As shown in Figure 6, the specific fragment (421 bp) between the vector and the sgRNA expression cassette was successfully cloned and the positive plants were identified. After verification, 24 KO-LpABCB21 positive plants were obtained; however, five plants were obtained before the improvement. 24 KO-LpPILS7 positive plants were obtained; however, only 5 plants were obtained before the improvement. As shown in Figure 6, the specific fragment (421 bp) between the vector and the sgRNA expression cassette was successfully cloned and the positive plants were identified. After verification, 24 KO-LpABCB21 positive plants were obtained; however, five plants were obtained before the improvement. 24 KO-LpPILS7 positive plants were obtained; however, only 5 plants were obtained before the improvement.
As shown in Figure 6, the specific fragment (421 bp) between the vector and the sgRNA expression cassette was successfully cloned and the positive plants were identified. After verification, 24 KO-LpABCB21 positive plants were obtained; however, five plants were obtained before the improvement. 24 KO-LpPILS7 positive plants were obtained; however, only 5 plants were obtained before the improvement.     Data were analyzed using oneway ANOVA using SPSS; the differences were compared using Fisher's least significant difference test (LSD) at the 5% significance level. The red vertical bars in C and F represent the edge of dual sgRNA.
Since LpABCB21 and LpPILS7 are membrane transporters, the hydrophilicity, hydrophobicity, and transmembrane domains of the mutant protein sequences were estimated by prediction. The results showed that hydrophilicity and hydrophobicity of the proteins were considerably affected by the frame shift mutation and protein mutations between dual sgRNA (B-D, G-J); however, these properties were affected to a lesser extent by the amino acid mutation in dual sgRNA (E, F, K, and L) ( Figure S1). The wild-type LpABCB21 protein sequence consists of 11 transmembrane domains (A). The frame shift mutation resulted in deletion of the third to eleventh transmembrane domains (B, C, D, and E). The protein mutation between dual sgRNA caused deletion of the third transmembrane domain; the distribution of amino acid residues to the inner and outer membrane prior to the missing transmembrane domain was also affected (F). The transmembrane domain of the amino acid mutant with mutation in dual sgRNA was identical to that of the wild-type protein (G and H). The wild-type LpPILS7 protein sequence consists of 10 transmembrane domains (I). The frame shift mutations resulted in deletion of the third to tenth transmembrane domains (K and M) or the fifth to tenth transmembrane domains (J and L). The protein mutation between dual sgRNA resulted in deletion of the fourth transmembrane domain (N). The transmembrane domain of the amino acid mutant with mutation in dual sgRNA was identical to that of the wild-type protein (O) ( Figure S2).
The assay of the expression levels in the overexpression lines showed that the level of LpABCB21 was significantly increased by 2.0-9.6-fold in 3 OE-LpABCB21 lines compared with that in wild-type plants ( Figure 8B). In 7 OE-LpPILS7 lines, the expression level of LpPILS7 was significantly increased (E), the expression levels of LpPILS7 in overexpression lines; (F), the protein sequence alignment of KO-LpPILS7 mutant lines. WT, wild-type plants; OE, overexpression lines; OE1, the expression level increases by more than 9-fold; OE2, the expression levels increases by 2-4-fold; OE3, the expression level did not change significantly; KO, mutant lines; KO1, frame shift mutations; KO2, mutation between dual sgRNA; KO3, mutations in dual sgRNA. Start, the emergence time of somatic embryo is marked in the image; 50 d, somatic embryo induction for 50 days. Data were analyzed using one-way ANOVA using SPSS; the differences were compared using Fisher's least significant difference test (LSD) at the 5% significance level. The red vertical bars in C and F represent the edge of dual sgRNA.
Since LpABCB21 and LpPILS7 are membrane transporters, the hydrophilicity, hydrophobicity, and transmembrane domains of the mutant protein sequences were estimated by prediction. The results showed that hydrophilicity and hydrophobicity of the proteins were considerably affected by the frame shift mutation and protein mutations between dual sgRNA (B-D, G-J); however, these properties were affected to a lesser extent by the amino acid mutation in dual sgRNA (E, F, K, and L) ( Figure S1). The wild-type LpABCB21 protein sequence consists of 11 transmembrane domains (A). The frame shift mutation resulted in deletion of the third to eleventh transmembrane domains (B, C, D, and E). The protein mutation between dual sgRNA caused deletion of the third transmembrane domain; the distribution of amino acid residues to the inner and outer membrane prior to the missing transmembrane domain was also affected (F). The transmembrane domain of the amino acid mutant with mutation in dual sgRNA was identical to that of the wild-type protein (G and H). The wild-type LpPILS7 protein sequence consists of 10 transmembrane domains (I). The frame shift mutations resulted in deletion of the third to tenth transmembrane domains (K and M) or the fifth to tenth transmembrane domains (J and L). The protein mutation between dual sgRNA resulted in deletion of the fourth transmembrane domain (N). The transmembrane domain of the amino acid mutant with mutation in dual sgRNA was identical to that of the wild-type protein (O) ( Figure S2).
The assay of the expression levels in the overexpression lines showed that the level of LpABCB21 was significantly increased by 2.0-9.6-fold in 3 OE-LpABCB21 lines compared with that in wild-type plants ( Figure 8B). In 7 OE-LpPILS7 lines, the expression level of LpPILS7 was significantly increased by 2.0-9.8-fold compared with that in wild-type plants except lines L5 and L25 ( Figure 8E).

Analysis of Genetic Transformation Efficiency
As shown in Table 2, the RC5.0 and Germination II conditions (after improvement) are better than RC5.8 and Germination I conditions (before improvement); the germination rate increased by 17.7-42.1%, the number of resistant plants increased by 2.7-6.4-fold, the number of positive lines increased by 3-6-fold, and the genetic transformation efficiency increased by 5.7-13.0%.

The Effects of LpABCB21 and LpABILS7 on SE
The somatic embryo was formed when the scales of the wild-type plants were cultured in somatic embryo induction medium for 21 days, and the somatic embryo induction rate was 100%. Comparison with somatic embryo induction in the wild-type plants indicated that LpABCB21 mainly regulated early SE and had a certain influence on regulation of late SE. After overexpression of LpABCB21, the emergence time of somatic embryos was advanced by 2-4 days. When the expression level increased by 2.0-2.8-fold, the somatic embryo induction efficiency was not changed. Although the expression levels in line L2 and L12 were significantly different, there were no differences in SE. When the expression level was significantly increased in line L11, somatic embryo induction efficiency decreased by approximately 30-40%, and the emergence time of somatic embryos was advanced by additional 2 days compared with that in line L2 and L12. LpABCB21 mutation delayed the emergence time of somatic embryos by 1-3 days, and adventitious buds were formed in the late stage. The number of adventitious buds formed by the frame shift mutant (KO1) was the highest, and the adventitious bud induction rate was 79.2% (19/24) ( Figure 8A). LpPILS7 mainly regulated late SE and had little effect on early SE. Only a few lines can advance (OE1: L30) or delay (KO1: L10 and L40) the emergence of somatic embryos by 1 day. Somatic embryo induction efficiency increased by approximately 10-30% when LpPILS7 expression levels increased by 2.0-9.8-fold, while LpPILS7 mutations decreased somatic embryo induction efficiency by approximately 10-60% ( Figure 8D). Although there were transgenic lines with significantly different expression levels in OE2, there were no significant differences in SE. If the expression level was significant increased by 9.8-fold (OE1), somatic embryo induction efficiency was further increased. The results of somatic embryo induction in various protein mutation modes indicated that proteins with the frame shift mutation (KO1) or the large fragment mutation (KO2) had severely damaged function of the protein; however, the insertion, replacement, and deletion of individual amino acids (KO3) had weaker effect on the function of the protein. For example, LpABCB21 mutant lines L18 and L29 had almost no effect on SE. The somatic embryo induction efficiency of LpPILS7 mutant line L5 was reduced by 10%, which was less than the effect of the KO1 and KO2 mutant lines.

Discussion
The pH of the medium was reported to be crucial for Agrobacterium virulence in in vitro tissue culture [41]. The optimal pH for Agrobacterium growth is 7.0 [42], but optimal infection of the host occurs at pH 4.8-5.5 [43,44]. In the presence of AS (acetosyringone), the induction of the vir gene is favored under the acidic conditions [32], while the expression of the vir gene is barely induced at neutral pH; the expression level of the vir gene is increased by 1.2-2.0-fold when pH drops from 7.0 to 5.5 [43]. In the co-culture of Agrobacterium tumefaciens and plant cells, the expression level of the vir gene changes depending on pH and reaches the maximum at pH 5.1 [43]; the expression level of the vir gene decreases when pH increases from 5.1 to 5.6 [41]. Genetic transformation efficiency is maximal when pH of the pre-culture or co-culture media is 5.0 [25,45]. The co-culture medium with pH 5.2 is more suitable for the genetic transformation of tomato (Solanum lycopersicum) cotyledon [46]. Our study further confirms that low pH can promote Agrobacterium-mediated genetic transformation. The number of resistant calli increased when pH of the re-suspension solution and co-culture medium decreased from 5.8 to 5.0 indicating that pH 5.0 is more suitable for the genetic transformation of L. pumilum than pH 5.8. However, a pronounced decrease in pH has an inhibitory effect on genetic transformation [35]. Since the genetic transformation efficiency of L. pumilum was compared only under the conditions with pH 5.0 and pH 5.8 in this experiment, additional improvement in the genetic transformation efficiency of Lilium should consider optimization of pH of the transformation process.
Calcium ions are the major second messenger in plants, and the calcium signaling pathway has been extensively studied [47,48]. Recent studies have shown that CaCl 2 (calcium chloride) can promote the induction of embryogenic calli (EC) and the germination of somatic embryos. Addition of 10 mM CaCl 2 increased the germination rate of somatic embryos by 51.79% [49]. The results of the present study indicate that an increase in the content of CaCl 2 from 0.44 g/L to 1.32 g/L can improve the germination coefficient of somatic embryos. However, the role of CaCl 2 in SE is genotype-and variety-dependent [49]. Therefore, additional screening of optimal dosages is necessary for various materials, and the mechanism of CaCl 2 regulation of the germination of somatic embryo requires additional investigation.
Previous studies have shown that ABCB21 and ABCB4 are bidirectional auxin transport carriers, which control the direction of auxin transport depending on auxin concentration to regulate the auxin level in the cells [14,16]. Recent studies in Arabidopsis have shown that ABCB21 can regulate the transport of auxin in cotyledons, root pericycle, and leaves to regulate normal growth [6]; however, the mechanism of SE regulation by ABCB21 remains unclear. ABCB21 has auxin influx or efflux activities under certain conditions, which is very similar to the pattern observed in ABCB4 [17]. A recent study has shown that ABCB21 initially has auxin uptake activity until intracellular auxin concentration reaches a certain threshold; then, the efflux activity is activated [6]. Our data indicate that various expression levels of LpABCB21 may also influence auxin transport rate. Several studies have shown that SE is more easily induced when the explants contain high auxin levels [50][51][52], which indirectly indicates that the auxin content in the explants reaches a high level in advance may promote earlier formation of somatic embryos. However, the ratio of auxin to cytokinin determines the development fate of the reproductive organs [53]. These considerations suggest that the content of exogenous auxin or cytokinin entering the explants and the content of endogenous cytokinin and auxin in the explants should be assessed when the exogenous hormone is used to change the original development state of the explants so that the ratio of auxin to cytokinin is always within the threshold range required for specific organ development. In this study, when exogenous auxin was constant, somatic embryos appeared earlier in the overexpression lines indicating that auxin content may increase to a higher level within a certain period of time to reach the auxin threshold required for SE to advance. An increase in the expression of LpABCB21 may shorten the time required for auxin to reach the threshold of SE to induce an additional advance in the emergence time of somatic embryo. However, if the auxin content is so high that it exceeded a certain threshold, the continued growth of somatic embryo is inhibited. Bielach found that cytokinin application on lateral root primordia delays the appearance of maximum auxin levels and lateral root growth but has no effect on the formation of the lateral root [54]. ABCB21 mutations and cytokinin treatment produce the same results [6]. These studies confirm the results of our experiments. The emergence time of somatic embryos was delayed in the LpABCB21 mutant lines, and there was no significant influence on the induction efficiency of somatic embryos. It is not clear whether this is the result of functionally redundancy of ABCB4 with ABCB21 during SE or if AUX/LAX1 and ABCB21 coordinate to regulate the exogenous auxin influx.
The function of PILS proteins in plant development is unclear; it is currently believed that PILS proteins can limit the rate of auxin signal transduction and response to external stimuli, such as light and temperature, to regulate auxin-dependent development [8,22]. Our results indicate that the mechanisms of LpPILS7 and LpABCB21 involved in auxin regulation may be similar; however, the proteins are associated with different patterns of auxin regulation. The reasons for this difference may be due to two factors. First, PILS protein is located at the ER and controls the accumulation of auxin in the cells. Second, the function of PILS protein mainly involves regulation of auxin signal through binding or releasing auxin and availability of auxin in the nucleus [8,22]; however, ABCB21 is located on the PM and is mainly involved in the balance of extracellular and intracellular levels of auxin. ABCB21 acts by determining the direction of auxin transport based on auxin concentration [14,17]. However, the ultimate result is the same. In in vitro tissue culture, the balance between exogenous and endogenous auxin levels can be adjusted so that the ratio between auxin and cytokinin can be maintained within the threshold range of specific requirements of a developmental process. When LpPILS7-overexpressimg and mutant lines induce somatic embryos, the effect on the emergence time of somatic embryos was very small. This phenomenon may be explained by very high rate of auxin transport [6]; hence, the process of exogenous hormone transport into the cells, which is probably determined by LpABCB21, is not affected. Our previous study demonstrated that LpPILS7 expression level is increased in the light culture, and somatic embryo induction efficiency in the light culture is better than that in the dark culture [40]. This experiment confirms that LpPILS7 can promote L. pumilum SE. PILS protein was suggested to bind auxin to limit nucleus auxin signaling [8,22]. PILS6 mutants have an increased auxin signal in the nucleus [8]. The results of this experiment suggest that PILS7 and PILS6 may have the same molecular mechanism of in regulation of auxin signals, which will be verified in the future studies.
In conclusion, this study describes a more efficient method of genetic transformation of L. pumilum improved by decreasing the pH of re-suspension solution and co-culture medium to 5.0 and increasing the CaCl 2 concentration of germination medium to 1.32 g/L; these results are important for subsequent functional genomics studies and molecular breeding in Lilium. Our findings confirm that LpABCB21 and LpPILS7 have important regulatory roles in L. pumilum SE and contribute to determination of the molecular mechanism of Lilium SE.

Plant Materials, Strains, and Vectors
L. pumilum EC were used as material. EC were obtained according to a previously described method [55]. Cultures were grown at 25 ± 1 • C under a 16/8 h photoperiod with a photosynthetic photon flux density of 36 µmol m −2 s −1 provided by cool white fluorescent light bulbs (Philips, Shanghai, China).

Construction of Overexpression and CRISPR/Cas9 Knockout Vectors
Primers with restriction enzyme cutting sites were designed using the Primer 5 software according to the sequences of the target genes and overexpression vectors. The sgRNA was designed according to the PAM/NGG domain on the sequences of the target genes. The sgRNA requires the content of GC% to be higher than 40%, absence of four consecutive thymines, and continuous pairing between sgRNA and target gene less than 8 bp. The primers for overexpression and knockout vector construction are listed in Table S1.
Previously, sequencing verified plasmids of LpABCB21 and LpPILS7 were used as templates to clone the CDS of LpABCB21 and LpPILS7. After the PCR product was purified, the CDS of the target genes and the sgRNA expression cassettes were inserted into pRI101-ON and pG3H-U3Ub, respectively, by enzymatic ligation.

Improvement of Agrobacterium-Mediated Genetic Transformation System
The genetic transformation method was performed as described previously [40]. In brief, EC were transferred to somatic embryo induction medium for proliferation and subcultured every 30 days until genetic transformation was performed. Light yellowish friable EC were selected as recipient materials. Agrobacterium expressing the vectors was grown in a liquid medium until the OD 600 value reached 0.6. Collected bacteria were subjected to Re-suspension I in the same volume to suspend the bacteria. After addition of 100 µmol AS, the EC that had been precultured in the pre-cultivation medium for 10 days were infected for 15 min. The EC were inoculated into Co-cultivation I for culture for 3 days in the dark, and then inoculated into Selection I and Selection II media for 2 weeks. Finally, transformed EC were inoculated into Germination I for germination. The results of preliminary experiments (Supplementary Table R1 and Figure R1) demonstrated that the germination efficiency of somatic embryos was the highest when non-transformed EC were inoculated in Germination II in the light culture (the light period was 16/8 h). To further improve the genetic transformation efficiency, the following aspects were improved. The original method was used as the control. The media used in the transformation process are shown in Appendix A Table A1.
Improvement of culture condition: The pH of the re-suspended solution and co-culture medium were changed to 5.0. The number of resistant calli was counted. Improvement of somatic embryo germination condition: EC were inoculated into Germination II for germination. Statistics of somatic embryo germination efficiency was assessed.

Identification of LpABCB21 and LpPILS7 Transgenic Plants
The genomic DNA of the leaves was extracted and used as a template for PCR identification. The primers for identification of transgenic plants are listed in Table S2. Specific primers for identification of overexpression in plants were designed based on the 35S promoter sequence (35S-F), the inserted DNA fragments (B21-R and S7-R), and the NPT II sequence (NTP II-F and NTP II-R). The expression levels in the overexpression plants were detected according to the previously described method [40]. Specific primers for identification of knockout plants were designed based on the OsU3 and TaU3 promoter (OsU3p and TaU3p) [56]. Specific primers for identification of the mutation mode were designed outside of the dual sgRNA (s7-F, s7-R, b21-F, and b21-R). DNAMAN software (version 7.0.2) was used to analyze the hydrophilicity and hydrophobicity. TMHMM (TransMembrane prediction using Hidden Markov Models) software (http://www.cbs.dtu.dk/services/TMHMM/) was used to analyze the transmembrane domains.

Identification of the Function of LpABCB21 and LpABILS7 in SE
The scales of the overexpression lines and mutant lines were used as explants. The scales were cut into small pieces of approximately 0.5 cm 2 and inoculated in the somatic embryo induction medium for dark culture. The wild-type plants were used as the control. The emergence time of somatic embryos was estimated. The results of somatic embryo induction were assessed after culture for 50 days.