The 102 nt ncRNA sequence of Arabidopsis thaliana U6-26 snRNA gene (NCBI accession: X52528.1) was used as a query to BLAST the Lotus japonicus genome database (http://www.kazusa.or.jp/lotus/). Nine putative U6-snRNA genes were identified and we carried out multiple sequence alignment among these genes (Figure S1). The LjU6-1 promoter was PCR-amplified (SgRNA-Kpn1-Spe1-F and U6-sgRNA-fusion-R) from wild type Lotus japonicus MG20 genomic DNA and the sgRNA region was PCR-amplified (U6-sgRNA-fusion-F and SgRNA-Xba1-Sal1-R) from the pBlueScript SK+-AtU6-26 sgRNA vector. The fusion fragment was then subcloned into pBlueScript SK+-AtU6-26 sgRNA vector (Feng et al., 2013) by replacing the AtU6-26 promoter element. The BbsI site in the LjU6 promoter region was mutated (primers Bbs1-mut-F and Bbs1-mut-R) before plasmid construction to facilitate downstream cloning. We used the web tool CRISPR-P (Lei et al., 2014) (http://cbi.hzau.edu.cn/crispr/) to select specific single-guide RNA sequences. The LjU6-sgRNA fragments between KpnI and SalI sites, together with the KpnI and EcoRI double-digested 2 × 35S-Cas9, were cloned into the pCAMBIA1300 vector. To construct the tissue-specific expression vector, the 2 × 35S constitutive promoter (SK+-35S-Cas9 vector) was replaced with LjLb2 promoter (1363 bp) using the XhoI and KpnI enzyme sites and the hygromycin-resistance gene (pCAMBIA1300 vector) was replaced with sGFP gene (XhoI-sGFP-Fand XhoI-sGFP-R) using the XhoI enzyme site. Primers and vectors used in this study are listed in supplemental Tables S1, S2, respectively.

The transient YF-FP-HR reporter assay was performed according to the method described by Feng et al. (2013). The Arabidopsis mesophyll protoplasts isolation and reporter plasmids transfection were conducted by a standard procedure (Yoo et al., 2007). 12~16 h after DNA transfection, fluorescence signals were analyzed by the “Microscopic Analysis” described below.

Microscopic analysis was performed using the Nikon SMZ18 and the Olympus FV1000 confocal laser-scanning fluorescence microscope. Imaging services were provided by the microscopic analysis facility of the State Key Laboratory of Agricultural Microbiology at Huazhong Agricultural University.

All the vectors were electroporated into Agrobacterium tumefaciens strain EHA105 and were used for transient expression in N benthamiana according to the protocol described by Sparkes et al. (2006). After 36 to 48 h post-infiltration, leaf tissues were harvested, immediately frozen in liquid nitrogen and stored at −80°C until use. Total RNA was isolated from tobacco leaves using TRIzol Reagent (Invitrogen) following the recommended protocol. One microgram total RNA was reverse transcribed using EasyScript one-step gDNA Removal and cDNA Synthesis Super Mix kit (TransGen Biotech, China). PCR amplification was performed using gene specific forward primers (YF-FP-sgRNA-F, SYMRK-sgRNA-F, Lb-sgRNA1-F, Lb-sgRNA2-F) and RT-sgRNA-R primer (sequences of these primers can be found in Table S1) under the following conditions: 95°C for 5 min; 31 cycles (94°C for 30 s, 58°C for 30 s, 72°C for 15 s); and 72°C for 5 min.

The Cas9 nuclease expression assay was performed according to the following procedure. About 100 mg of tobacco leaves were finely pulverized in liquid nitrogen with a cold mortar and pestle. The fine powder was transferred into 1.5 mL eppendorf tubes using a cold stainless-steel lab spoon and incubated with 400 μL protein extraction buffer containing 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 1% NP-40, 0.1% SDS, 1 × protease inhibitor cocktail (Roche) and 2 mM PMSF (phenylmethylsulfonyl fluoride, Sigma). The tubes were then shaken vigorously for 30~60 s until the powder was fully dissolved, incubated at 4°C for 30 min and then centrifuged at 13,000 g at 4°C for 15 min. About 300 μL supernatant was transferred into a new 1.5 mL tube and the protein concentration was quantified with the Pierce™ BCA Protein Assay Kit (Thermo Fisher). Equal amount of total proteins (~30 μg) were mixed with 4 × SDS loading buffer (40% Glycerol, 240 mM Tris/HCL pH 6.8, 8% SDS, 0.01% bromophenol blue, 10% β-mercaptoethanol), boiled for 10 min and were then separated on a 10% SDS polyacrylamide gel and transferred to PVDF membrane. Membranes were blocked with 5% non-fat dry milk in Tris-buffered saline (TBS, pH 7.2~7.4) for 2~4 h at room temperature. Blocked membranes were washed with TBST (TBS containing 0.1% Tween-20) for 5 min and incubated with anti-FLAG antibody (Sigma, 1:2000 dilution) in TBST for 2 h at room temperature. Membranes were then washed four times with TBST. HRP-conjugated goat anti-mouse secondary antibody (1 mg/ml, Proteintech Group Inc., China) was diluted 1:5000 in TBST and incubated with the membranes for 1 h at room temperature. Membranes were washed as above and chemiluminescence signals were detected with ChemiScope 6300 imaging system in accordance with the recommended instructions (Shanghai Clinx Science Instruments Co., Ltd, China).

A. tumefaciens-mediated stable transformation of Lotus japonicus MG20 was conducted by a standard procedure (Tirichine et al., 2005) with minor modifications. In brief, seeds were sterilized and germinated on plates with MS (Murashige & Skoog) medium (4.33 g/L Basal Salt Mixture (Sigma), 0.103 g/L vitamin mixtures (Catalog number M7150, Sigma), 30 g/L Sucrose (Sigma), 8 g/L Agar (Sigma), pH 6.0) for 4 d in the dark at 22°C, followed by 3 d under 16 h light/8 h darkness photocycle in a growth cabinet. The explants of the split hypocotyl were infected with A.tumefaciens EHA105 containing the appropriate plasmids for ~30 min. The co-cultivation plates (10 pieces of sterile filter paper soaked with 1/10 × Gamborg's B-5 liquid medium, Sigma) was incubated for 7 d at 21°C in the dark. After that, the explants were transferred to regeneration medium (Gamborg's B-5 medium, 3.1 g/L Basal Salt Mixture (Sigma), 1 × Vitamin solution (Catalog number G1019, Sigma), 20 g/L Sucrose (Sigma), 4 g/L Phytagel (Sigma), pH 5.5) containing 0.4 μg/mL 6-BA (Solarbio, China), 0.04 μg/mL NAA (Solarbio, China), 15 μg/mL hygromycin (Roche) and 300 μg/mL timentin (Solarbio, China) for callus formation. The transgenic calli were transferred to fresh regeneration medium every week for 1~2 months. After that, the green calli were constantly transferred to shoot induction medium (Gamborg's B-5 medium) containing 5 mM (NH₄)₂SO₄, 0.4 μg/mL 6-BA (Solarbio, China), 15 μg/mL hygromycin (Roche) and 300 μg/mL timentin (Solarbio, China) until shoots were 2~4 cm long. Individual shoots were transferred to root induction medium (Gamborg's B-5 medium) containing 0.1 μg/mL NAA, 15 μg/mL hygromycin and 300 μg/mL timentin and propagated until roots appeared.

Hairy root transformation of Lotus japonicus MG20 using the A. rhizogenes strain LBA1334 was described previously (Díaz et al., 2005). LBA1334 carrying the binary plasmids of interest was grown on Luria-Bertani (LB) plate supplemented with 50 μg/mL kanamycin (Teknova, USA) and 25 μg/mL rifampicin (Biosharp, China) for 2~3 days. To synchronize the emergence of hairy roots, 7-day-old Lotus MG20 seedlings were cut at the base of hypocotyls and moved to a conical flask containing resuspended Agrobacterium (final OD₆₀₀~ 0.6). After being soaked for half an hour in the flask, the hypocotyls were transferred onto MS plates (Sigma) and placed in the growth chamber for 5 days. The plants were transferred onto HRE (Hairy Root Emergence, Lotus japonicus Handbook, http://link.springer.com/book/10.1007/1-4020-3735-X) medium plates containing 300 μg/mL timentin and grown for 10~15 days at 22°C with 16 h light/8 h darkness and under white light intensity of 60~100 μmol photons m⁻² s⁻¹. Plants were then transferred to nutrition pots filled with vermiculite and perlite (2:1) and grown in a green house at 23°C with 16 h light/8 h darkness and under white light intensity of 60~100 μmol photons m⁻² s⁻¹. When the first true leaf appeared, plants were inoculated with Mesorhizobium loti strain MAFF303099 to assay nodulation phenotypes.

Genomic DNA was extracted using a method described by Li et al. (2010). PCR amplification was performed using gene specific primers under the conditions mentioned above. PCR products were digested with appropriate restriction enzymes and separated on a 1% agarose gel. Appropriate DNA bands were extracted from the gel, purified and cloned into the pMD19T-simple vector (Takara). Five individual clones were sequenced by Sanger sequencing using gene specific primers. For gene loci with no appropriate restriction enzymes, PCR products were directly sequenced by Sanger sequencing. If mutations were identified, each PCR product was cloned into the pMD19T-simple vector, individual E. coli colony was selected and then sequenced to confirm the mutation.

Five days post inoculation with M. loti, total RNA was isolated using Trizol Plants Total RNA Isolation Kit (TransGen Biotech, China) from at least 100 mg fresh root materials (~3 plants). Primescript RT Reagent kit (Takara) was used to synthesize first strand cDNAs. Reverse transcription-PCR (RT-PCR) amplification was performed using gene specific primers (SYMRK-RT-F and SYMRK-RT-R) under the following conditions: 95°C for 5 min; 31 cycles (94°C for 30 s, 58°C for 30 s, 72°C for 1 min 15 s); and 72°C for 5 min.

Real-time quantitative reverse transcription (qRT)-PCR was performed using the SYBR Select Master Mix reagent (Applied Biosystems) under the following conditions: Hold Stage, 50°C for 2 min, 95°C for 10 min; PCR Stage, 40 cycles (95°C for 15 s, 60°C for 1 min); and Melt curve Stage, 60°C to 95°C at the speed of 0.05°C/s. qRT-PCR reactions were performed using an ABI ViiA™ 7 Real-Time PCR System. The expression levels of LjSYMRK (with SYMRK-qRT-F and SYMRK-qRT-R primers) were calculated by the 2^−ΔΔCt method and normalized against the polyubiquitin gene (LjUBI, GenBank accession no. AW720576). For each genotype, at least three biological replicates and three technical replicates were analyzed.

Detailed procedures for protein immunoblot assay have been described in “Transient gene expression analysis in Nicotiana benthamiana.” For Lotus japonicus leghemoglobin western analysis, in brief, protein samples were harvested 1 month post inoculation. About 100~150 mg fresh nodules were finely pulverized in liquid nitrogen with a cold mortar and pestle. Protein samples were separated on a 12–15% SDS-PAGE gel. Western-blot analysis was performed using primary antibody against soybean leghemoglobin GmLba (a generous gift from Professor Yangrong Cao of Huazhong Agricultural University, China) and HRP-conjugated rabbit anti-goat secondary antibody. As a positive control, Glutathione S-transferase (GST) tagged LjLb3 was expressed in E. coli BL21. Cells were disrupted by high pressure homogenization (D-3L; PhD Technology International, MN, USA) and protein was purified by GST column following recommended procedures (Sangon Biotech, China).

To achieve efficient targeted gene editing, a single guide RNA (sgRNA) needs to be expressed to direct the Cas9 endonuclease to cleave corresponding DNA targets. Bortesi and Fischer (2015) compared various promoters that have been used to generate sgRNAs in different plants. U6 or U3 small nuclear RNA (snRNA) promoters (recognized by RNA polymerase III) were used most frequently. Based on the previous study in Arabidopsis by Feng et al. (2013), we chose the Lotus japonicus endogenous U6 promoter to drive the sgRNA expression. Nine putative U6-snRNA genes (Figure S1A) were identified in the Lotus genome based on the Arabidopsis U6-26 snRNA sequence. Both USE (upstream sequence element) and TATA-like box elements are conserved in the promoter region (Waibel and Filipowicz, 1990), which are indispensable for the transcription of snRNAs. Subsequently, we cloned an ~800 bp promoter region of LjU6-1 from the genomic DNA of Lotus ecotype MG20 (Miyakojima) (Figure S1B). A BbsI recognition site in the LjU6-1 gene promoter was mutated (G → C) so that the guide RNA can be inserted between two BbsI sites with annealed oligonucleotides. The CAMV 35S promoter (2 × 35S) was used to express the codon-optimized hSpCas9 (Feng et al., 2013). To test the efficacy of the engineered CRISPR/Cas9 system, we took advantage of a split yellow fluorescent protein (YFP) reporter system, YF-FP, in Arabidopsis protoplasts based on homologous recombination (HR) restoration of the fluorescence (Figure 1A). The 35S YFP and 35S YF-FP + CRISPR-Empty were used as positive and negative controls, respectively. After transient coexpression of the 35S YF-FP and CRISPR-YFP constructs, the YFP signal was readily detected using a confocal laser-scanning fluorescence microscope (Figure 1B). These results demonstrated that LjU6-1 promoter is functional in driving expression of the sgRNA gene and that the reconstructed CRISPR/Cas9 system is competent in generating precise DNA modification in plant cells.

We selected four Lotus japonicus symbiosis related genes as target loci, including the symbiosis receptor kinase LjSYMRK, and the symbiotic leghemoglobin genes LjLb1, LjLb2, LjLb3 (GenBank accession numbers: AF492655, AB042716, AB042717, and AB008224, respectively). Gene-specific guide sequences were designed by the web tool CRISPR-P, which allowed us to screen for highly specific editing sites within target DNA sequences and provided off-target prediction simultaneously (Lei et al., 2014). One guide sequence (SYMRK-sgRNA) was designed to target the third exon of LjSYMRK, while two guide RNAs were chosen to target the three LjLb genes simultaneously (Lb-sgRNA1 and Lb-sgRNA2) (Figures 2A, 4A, respectively). Successful expression of sgRNA and FLAG-tagged Cas9 was confirmed by reverse transcription PCR (RT-PCR) (Figure S2A) and Western blotting (Figure S2B) in the infiltrated tobacco leaves.

The transformed A.tumefaciens EHA105 harboring the SYMRK-sgRNA was used for stable transformation of Lotus japonicus. Twenty T0 transgenic plants were collected for PCR/RE (restriction enzyme) assay to detect mutations around the target site (Figure 2B). The enzyme-digestion pattern of 13 samples (65%) were the same as wide-type control, possibly suggesting that no mutations occurred in the target site. The other seven samples (35%) showed a mixture of PstI digested PCR fragements as well as intact, undigested PCR fragments (lines #1, #4, #8, #12, #15, #18, #20; Figure 2B), with the latter indicating the occurrence of Cas9/sgRNA-dependent mutations in the target site of some LjSYMRK alleles. The mixture of both digested and undigested bands suggested heterozygous alleles in the target gene. Interestingly, two of them (lines #18 and #20) were completely resistant to enzyme digestion. Sequencing analysis indicates these two transgenic lines contain homozygous mutations of the two SYMRK alleles in the genome, with identical 2 bp deletions at the site of Cas9 DNA cleavage. The 2 bp deletion near the PAM sequence was retained in the transcript and there was no evidence of alternative splicing in the knockout lines (Figure S3A). Quantitative measurement of LjSYMRK gene expression by qRT-PCR indicated no significant changes in line #18 and line #20, compared with WT and EMS61 mutant (Figure S3B). Thus, line #18 and #20 are most likely complete knockout in LjSYMRK gene because of the 2 bp deletion in exon 3.

PCR products of other 5 putative mutants were also cloned and individually sequenced (Figure 2C). We found that all these 5 plants contained various mutant alleles including diverse deletion, insertion and substitution mutations. Overall, our results demonstrated that the mutation frequency of the Lotus LjSYMRK gene using just one sgRNA was notably high (35%, 7 out of 20 T0 transgenic lines).

We next compared the symbiotic phenotype of the T1 progeny from line #18 and line #20, together with the LjSYMRK mutant EMS61 (Stracke et al., 2002) and MG20 WT plants. The mutant EMS61 was generated by EMS mutagenesis and contained a nonsense mutation in the SYMRK kinase domain. Seedlings were grown in a nutrient-poor soil medium containing vermiculite and perlite (volume ratio 2:1) supplied with one-half-strength Broughton and Dilworth (B&D) nutrient solution. Plants were inoculated with Mesorhizobium loti strain MAFF303099 when the first true leaf appeared. In contrast to the typical root hair curling phenotype of WT plants (Figure 3A), the root hairs of the T1 transgenic plants exhibited excessive swelling and branching after 3~6 days post inoculation with M. loti - a similar phenotype to that of the mutant EMS61 that contains loss of function SYMRK mutation (Figures 3B–D; Stracke et al., 2002). After 12 days post inoculation, we found no distinguishable nodule primordium or nodules formed on the mutant roots (Figures 3F–H) compared with the normal nodules produced by WT roots (Figure 3E).

The Lotus genome encodes three leghemoglobin genes with high sequence similarity. The three LjLb genes are all located on chromosome 5 (Figure 4A). LjLb-RNAi plants formed white nodules instead of leghemoglobin-rich pink nodules (Ott et al., 2005). Hence, the abundance of white nodules can be an easily scorable phenotype with which to estimate the efficiency of multi-gene modifications caused by CRISPR/Cas9. We transformed A. rhizogenes LBA1334 with the binary vector CRISPR-LbsgRNA1&2 and used these bacteria to obtain hairy root transformants of Lotus. LbsgRNA 1 targets identical sites (target site 1) in all three LjLbs genes (Figure 4A). Likewise, LbsgRNA 2 targets identical sites in LjLb1 and LjLb2 (target site 2) which is only 48 bp downstream of target site 1 (Figure 4A). The target site 2 on LjLb3 has two mismatches against the sgRNA 2 (Table S3).

70 transgenic plants were analyzed and 20 of them (~29%) formed white nodules after inoculated with M. loti. Genomic DNA was extracted and PCR was executed with gene specific forward primers (Lb1/2/3-F) and a common reverse primer (Lb-R) to determine mutation types. Unexpectedly, we found a mixture of simple and complex mutations in these plants, including deletions (representative plants #16, #18, #9, #11; Figures 4B,C, Figure S4A), insertions (representative plants #16, #18; Figure 4B), substitutions (representative plants #9, #11; Figure 4C, Figure S4A) and large fragment (~17 kb) deletion between LjLb1 and LjLb2 in plant #20 (Figure 4D). The moderate sized deletions in plants #18 and #9 (Figure 4C) appear to be rather precise deletions of the DNA sequences between the target site 1 and target site 2 of Cas9 cleavage sites in the Lb1 gene and the Lb2 gene, respectively.

Phenotypic analysis was performed on the representative triple mutant plant #11 (Figure 5), which exhibited significant reductions in shoot and root mass compared with a control plant harboring an empty CRIPSR vector 8 weeks post inoculation (Figure 5A). Control plants formed pink nodules (Figure 5B), while the triple mutant plant #11 produced only white nodules (Figure 5C). Moreover, the symbiosomes of plant #11 showed abnormal shape, seemed smaller and degraded compared with the radial-distributed, rod-shaped bacteroids and the regular-sized symbiosomes in control plants (Figures 5D,E). Overall, plant #11 demonstrated quite similar phenotypes to LjLb1/2/3 RNAi transgenic plants (Ott et al., 2005, 2009). DNA sequencing confirmed that all three LjLbs contained DNA deletions, i.e., a 105 bp deletion in LjLb3 and, as noted above, a 48 bp deletion in LjLb1 and a 47 bp deletion and 1 bp substitution in LjLb2 (Figure S4A). Immunoblot analysis also confirmed the absence of LjLb1/2/3 proteins in the white nodules of plant #11 (Figure S4B).

We also transformed A. tumefaciens EHA105 with the binary vector CRISPR-LbsgRNA1&2 and obtained stable transgenic lines. We have confirmed five ljlb mutants from ~135 T1 plants, with various mutation types of the three LjLb genes (Figure S5A). T2 generation plants of the LjLb triple mutant line3-A7 was assayed for the symbiotic phenotype. After 4 weeks post inoculation, vegetative growth of the line 3-A7 was significantly inhibited, with apparent chlorosis in the leaves, indicating the nitrogen deficiency phenotype (Figure S5B). Consistently, the triple mutant plants have small and white nodules caused by the absence of leghemoglobins (Figure S5C). In contrast, the MG20 WT plants form large and pink nodules which are effective in symbiotic nitrogen fixation (Figure S5D).

Constitutive promoters such as CaMV 35S, ubiquitin or actin promoters, are the most commonly used promoters to express Cas9 nuclease in various plants. However, these promoters occasionally cannot maintain relatively high level of gene expression in specific tissues (Feng et al., 2014). To compare the efficiency of nodule specific LjLb gene promoters in driving Cas9 expression for targeted gene modifications, we replaced the CaMV 35S promoter with the LjLb2 gene promoter (~1400 bp long) (Figure 6A). Four weeks after inoculation with M. loti, we found the percentage of white nodules of transgenic roots expressing p35SCas9-LbsgRNA1&2 (65 ± 6%) and of roots expressing pLjLb2Cas9-LbsgRNA1&2 (45 ± 7%) were both significantly increased (p < 0.01, Student's t-tests) compared with the respective control lines lacking sgRNA1&2 gene constructs (p35SCas9, 7 ± 2%; pLjLb2Cas9, 10 ± 2%; Figure 6B, bottom panel). The root lengths were similar among all four types of transgenic plants (Figure 6B, upper left panel). Consistent with the increased percentage of nitrogen-fixation-deficient white nodules in transgenic roots containing sgRNA1&2 gene constructs, the two types of corresponding transgenic plants showed reduced shoot length under nitrogen deficient conditions (Figure 6B, upper right panel).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.