Testing the efficiency of a modified vector for CRISPR/Cas9 gene editing system to target NCR169
To analyse the function of NCR genes, the CRISPR/Cas9 gene editing technology was used to generate composite Medicago truncatula plants mutated in hairy roots. In order to monitor the presence of CRISPR constructs, delivered into the genome of M. truncatula root cells using the Agrobacterium rhizogenes-mediated hairy-root system28, the gene coding for the DsRed fluorescent transformation marker was introduced into the CRISPR/Cas9 plant transformation vector pKSE401 resulting in the vector pKSE401-RR. The constitutively active red fluorescent marker allowed us to remove non-transgenic roots at the early stages to promote the growth of transgenic hairy roots. To test the efficiency of targeted mutagenesis of pKSE401-RR, we designed three different single guide RNA (sgRNA) constructs targeting the NCR169 gene (Fig. S1), previously proved to be essential for effective nitrogen-fixing symbiosis10. Each sgRNA (NCR169a, b and c) was designed directing to the first half of the coding sequence of NCR169 but the initial 5’ nucleotides for the protospacer sequences of NCR169b and NCR169c were C instead of the optimal G(/A) which is preferred by the AtU6 promoter (Fig. S1a;29). Nevertheless, we used these two suboptimal sgRNA constructs as well because the short, 189 bp long coding sequence of NCR169 constrained the available optimal target sequences. The three constructs carrying each sgRNAs were introduced into roots of wild-type M. truncatula 2HA genotype using A. rhizogenes-mediated hairy root transformation. At least ten transformed roots for each construct, detected by the DsRed fluorescent marker, were sampled and used for genomic DNA isolation. The region surrounding the targeted sequence in NCR169 was amplified by PCR. The sequence analysis of the amplicons detected only wild-type genotypes on all the NCR169b-transformed roots and roots transformed with the construct NCR169c were heterogeneous of wild-type and various mutant genotypes. However, the construct NCR169a generated mutations at higher efficiency and beside chimeric roots having both wild-type and distinct mutations, we could identify roots with homozygous or biallelic mutations for NCR169 (Fig. S1d). The results indicated that the presence of G at the 5’ end of the protospacer sequence (G(/A)N19NGG) is crucial and using C in this position greatly reduced the efficiency of genome editing which restricts the number of possible target sites in the case of small genes such as NCRs. In addition, the sequence analysis of amplicons containing the targeted region revealed that the sequences obtained by Sanger method composed of multiplex sequence traces around the break site (Fig. S1b). An online tool, TIDE, enabling the decomposition of mixed sequence signals was applied30 that predicted the types of the mutation and the size of the deletion or insertion. For instance, the decomposition of one of the amplicons of NCR169c disclosed the mixture of wild-type and CRISPR/Cas9 edited sequences containing a 7 bp deletion and/or a one bp insertion (Fig. S1c).
In agreement with a previous report22, we also found that different genome editing events occurred in distinct hairy roots of a single plant which necessitates the analysis of individual nodules separately. These findings led us to the conclusions that the effectiveness of genome editing and the reliability of the identification of the nature of genome editing events needed to be improved in our system. Therefore (i) we replaced the Arabidopsis U6-26 promoter with the M. truncatula U6 (MtU6.6) promoter driving sgRNA synthesis, (ii) we developed a preparation procedure allowing phenotypic and genotypic analysis of individual nodules, and (iii) applied an NGS amplicon sequencing approach to assess allele distribution in individual nodules.
Inducing mutations in NCR169 and NCR-new35 using the optimized CRISPR/Cas9-mediated genome editing system
Previous studies found that the M. truncatula U6 promoter enhanced the expression efficiency of guide or hairpin RNAs in legume species21,31. To increase the efficacy of targeted mutagenesis by CRISPR/Cas9, the Arabidopsis U6-26 promoter was replaced with MtU6.6 in the pKSE401-RR resulting in the vector pKSE466-RR. We inserted the sgRNAs NCR169a and another one, targeting the NCR-new35, of which requirement for effective symbiosis has been recently demonstrated13, into pKSE466-RR, respectively. To validate the efficiency of the pKSE466-RR vector construct, the empty vector and pKSE466-RR-NCR169a were introduced into wild-type M. truncatula 2HA plants using A. rhizogenes-mediated hairy root transformation and plants were inoculated with S. medicae WSM419. Nodules developed on DsRed fluorescent transformed roots were harvested 3–4 weeks post inoculation (wpi) with rhizobia, individually fixed and sections were stained with nucleic acid-binding dye SYTO13 (Fig. 1). Following image capturing, genomic DNA was purified from nodule sections and the regions surrounding the targeted sequences were amplified by PCR (Figs. 1 and 2).
The elongated nodules developed on empty vector-transformed roots were pink and showed the characteristic zonation of indeterminate nodules (Fig. 1a,d). The nodules on the roots of M. truncatula 2HA transformed with the pKSE466-RR-NCR169a were either pink and elongated, suggesting that these nodules were functional wild-type (Fig. 2g,i,k), or white undeveloped, implying that these were ineffective symbiotic nodules (Fig. 1b and 2b,d,f). To determine the nature of the mutations at the targeted region, the sequence analysis of the amplicons, generated from randomly selected elongated and undeveloped nodules, was carried out. The targeted amplicon sequences were determined using a next-generation sequencing (NGS) approach and the quantitative allele composition and the verification of genome editing mutations were analysed using the online tool CRISPResso232. The analysis revealed that all the white undeveloped nodules carried few base pair biallelic insertions or deletions at a high ratio and the very few reads of wild-type alleles identified in these nodules were probably derived from sporadic wild-type cells or the trace amount of wild-type contamination (Fig. 1b and 2a-f). The morphology and bacterial occupancy of these SYTO13-stained undeveloped nodules were very similar to the nodulation phenotype of the deletion mutant dnf7-2 showing no fluorescence in the nodule region proximal to the root (Fig. 1e). In addition, these plants, compared with empty-vector-transformed plants, showed the symptoms of nitrogen starvation (yellowish leaves and stunted growth habit) indicating the defect of symbiotic nitrogen fixation of these nodules (Fig. 1b). Contrary to these, the sequence analysis of the randomly selected elongated pink nodules showed that they either carried almost exclusively wild-type alleles (Fig. 2g,h) or they were heterozygous for example for a 4 bp deletion and the wild-type alleles (Fig. 2i,j). We also found a wild-type nodule that predominantly carried a homozygous deletion of 3 base pairs resulting in a deletion of the first single Glu residue of the mature peptide of NCR169 which mutation did not abolish the symbiotic function of NCR169 (Fig. 2k,l).
We also applied the developed CRISPR/Cas9 toolkit to target NCR-new35, a recently identified NCR gene which is crucial for the persistence of bacteroids in M. truncatula cv. Jemalong nodules13. The empty vector and the pKSE466-RR-NCRnew35 constructs were introduced into the roots of M. truncatula 2HA and M. truncatula ecotype R108 plants using A. rhizogenes-mediated hairy root transformation. Nodules on empty vector-transformed roots displayed the characteristic zonation of indeterminate nodules showing cells occupied by rhizobia (Fig. S2e,m). In contrast, undeveloped nodules were observed on roots of M. truncatula 2HA and R108 plants transformed with the editing construct of NCR-new35 (Fig. S2, S3g-r, 4e-h). The sequence analysis showed that undeveloped nodules, randomly collected from transformed roots of M. truncatula 2HA and R108 plants, carried biallelic or monoallelic homozygous mutations in the amplicons containing the target sequence in MtNCR-new35 (Fig. 4q,r). SYTO13-staining revealed that these nodules did not show the typical zonation and only their apical part fluoresced indicating that the region corresponding to the prospective nitrogen fixation zone was devoid of rhizobia (Fig. S2f-h, n-p). The morphology and bacterial occupancy of genome edited nodules were in agreement with the nodulation phenotype of Mtsym19 and Mtsym20 nodules defective in NCR-new3513. These results also revealed that NCR-new35 is crucial for the symbiotic interactions not only between M. truncatula 2HA and S. medicae WSM419 but between M. truncatula R108 and S. meliloti FSM-MA as well.
Selecting of NCR genes to be targeted by CRISPR/Cas9 editing
The M. truncatula genome contains a large gene family coding for more than 700 NCR peptides but only few of them, NCR169, NCR211, NCR247, NCR343 and NCR-new35, have been proved to be crucial for symbiosis10–15. The sequences of the essential NCR genes are highly variable, and their common feature is that they contain four cysteine residues in conserved positions. The peptide NCR247 is unique among the crucial and most of the other NCR peptides because it contains five residues between the third and fourth conserved cysteines compared with the canonical four internal residues13. In addition, no common feature could be identified among few NCR peptides found to be non-essential for symbiosis10,11,13. Therefore, it was not straightforward to ascertain which additional NCR peptides are essential and should be targeted. Despite all that, we have set up a selection pipeline to identify NCR genes to be targeted (Fig. 3). We have analysed several features of more than 700 predicted NCR genes identified in the M. truncatula genome9. The signal peptides of NCRs are crucial for delivering the peptides to the bacteroids4,6,7, therefore, (i) first we selected 655 NCR genes coding for peptides possessing a predicted signal peptide (Fig. 3a). The pool of NCR genes to be targeted (ii) were further enriched by selecting the encoded peptides with four cysteine residues in conserved positions. (iii) We have analysed the expression profile of the diminished list of 248 NCR genes in the database obtained by laser-capture microdissection of M. truncatula nodules33. We identified 11 NCRs with no reads, evidently seem to be pseudogenes, and 65 NCRs which are expressed at low level (< 10 000 reads) in nodules compared with other NCRs. The recently identified NCR-new35 shows the lowest expression level among the crucial NCRs13 and thus, we have selected genes from the 172 NCRs showing similar or higher expression level as NCR-new35 (Fig. 3c,d). (iv) These NCR genes were analysed to predict whether they had contained sequences which were amenable to CRISPR-based editing. The cysteine residues are essential for the function of NCR peptides in planta10,13, therefore we have prioritised those NCRs which could be targeted prior to the fourth conserved cysteines to generate presumably non-functional peptides. We have further searched the M. truncatula genome using an online prediction tool CRISPOR34 for off-target sites that could be recognized by sgRNAs designed for NCR genes. The isoeletric points (pI) of crucial NCR peptides vary between 4.78 and 10.15 implying that anionic, neutral and cationic NCR peptides could be essential for nitrogen-fixing symbiosis in M. truncatula. For this reason, the gene of NCR peptides with wide range of charges were selected to be edited (Fig. 3b). (v) Finally, we have selected four NCR genes, NCR068, NCR089, NCR128 and NCR161, to be targeted by CRISPR/Cas9 editing. The pI of the encoded mature peptides varied between 5.75 and 8.34 (Fig. 3b). NCR128 and NCR161 are induced in the late infection zone and showed expression in the interzone but NCR068 and NCR089 are preferentially expressed in the interzone and at lower level in the nitrogen fixation zone based on the LCM transcription data (Fig. 3c,33). The transcription activity of the four selected NCRs was lower than NCR169 and NCR211 but similar or higher than NCR-new35 (Fig. 3d).
Inducing mutations in selected NCR genes using the CRISPR/Cas9 system
To induce mutations in genes NCR068, NCR089, NCR128 and NCR161 by CRISPR/Cas9-mediated cleavage, sgRNA constructs targeting a single site of each gene were designed. The designed sgRNAs targeted either the upstream region of the gene close to the sequence coding for the first residues of the mature peptide or at least at a site prior the sequence coding for the last cysteine residue (Fig. 3e). These sgRNA constructs along with the control empty vector and the constructs targeting NCRnew-35 and NCR169 were introduced into M. truncatula 2HA roots using A. rhizogenes-mediated hairy root transformation. Transformed plants were inoculated with S. medicae WSM419 at 2–3 weeks after removal of non-transformed roots and the symbiotic phenotype of transformed plants were scored at 4–7 wpi (Fig. 4). The aerial part of plants transformed with empty vector or the constructs targeting NCR068 did not show the symptoms of nitrogen deficiency (Fig. 1a, c) and plants transformed with constructs targeting genes NCR089, NCR128 and NCR161 displayed the similar growth habit (data not shown) indicating the effective symbiotic nitrogen fixation capacity of these plants. Nodules formed on roots transformed with sgRNA constructs of the four selected NCRs were elongated and pink suggesting that they were functional nodules (Fig. 4j,l,n,p). The staining of nodule sections with SYTO13 revealed that the bacterial colonization of nodules targeted for mutagenesis of genes NCR068, NCR089, NCR128 and NCR161 was identical to the wild-type indeterminate nodules developed on empty-vector transformed roots (Fig. 4i-p). Contrary, 2HA plants carrying mutations in NCR169 or NCR-new35 in each hairy root displayed retarded growth, showed the symptoms of nitrogen starvation characteristically (Fig. 1b) and slightly elongated white nodules were found on their roots (Fig. 4f, S2 and S3). In addition, the occupancy of nodules mutagenized for NCR169 and NCRnew-35 showed the phenotype characteristic for the previously described deletion mutants10,13 (Fig. 1b, 4e-h, S2, S3).
To verify the result of CRISPR/Cas9 genome editing, the targeted regions were amplified from nodule sections following microscopic analysis and sequenced using a MiSeq platform. The sequence analysis of the amplicons was carried out with the online tool CRISPResso2. The Cas9-induced mutations were predominantly indels, with frequencies between 47 and 100% of total reads (Table 1.). The genotyping also revealed that individual nodules were often chimeric containing more than two alleles or the ratio of the two alleles deviated from the expected 1:1 ratio indicating that the tissues of these nodules on hairy roots originated from multiple founder cells with different genetic background. The identified indels generated frameshift mutations in the targeted NCR genes presumably producing malfunctioning peptides. Nevertheless, all the nodules carrying biallelic mutations in the targeted NCRs were elongated and pink, indicating the presence of leghemoglobin, and were fully colonized by rhizobia indicating that peptides NCR068, NCR089, NCR128 and NCR161 are not required for the effective symbiotic nitrogen fixation between M. truncatula 2HA and S. medicae WSM419 (Fig. 4i-p and S4).
The nodulation phenotype of stable mutant of NCR068 regenerated from gene edited hairy roots is not strain dependent
To analyse the symbiotic phenotype of genome editing mutants in more detail, we regenerated transgenic plants from hairy roots mutagenized for NCR genes. As a proof of concept, we cut root segments that had been transformed with the editing construct of NCR068 and stable transformants were regenerated using a slightly modified formerly published protocol35. DNA was purified from the leaves of hairy root-derived regenerated plants and the sequence analysis revealed that all the plants were homozygous containing a single adenine insertion in NCR068 (Fig. 5g). The one bp insertion generates a frame shift in the coding sequence of NCR068 and the mutant sequence encodes a chimera peptide containing 27 NCR068-specific and 23 non-NCR068-specific residues before termination. The fourth cysteine residue in the conserved position is lacking in the chimeric NCR068 peptide and based on the necessity of cysteines for the function of NCRs in planta10,13, we reckoned that the mutant NCR068 peptide does not function correctly.
To confirm the symbiotic phenotype of regenerated ncr068 mutant, T1 seeds were synchronously germinated and planted into substrate with no nitrogen content. Three wpi with S. medicae WSM419, ncr068 stable transformants did not show symptoms of nitrogen starvation and developed like wild type M. truncatula 2HA plants (Fig. 5). Nodules, formed on the roots of stable transformant ncr068 mutant plants were elongated similarly what we found on the hairy roots of composite plants induced by the transformation construct targeting NCR068 (Fig. 5c,e). The staining of ncr068 nodule sections with SYTO13 revealed infected cells in interzone and nitrogen fixation zone fully occupied with elongated bacteria, similarly as it was found in wild-type nodules and NCR068-targeted hairy roots (Fig. 1d,f, 5b,d). The nodulation phenotype of the regenerated mutant ncr068 confirmed that the peptide NCR068 is not essential for the nitrogen-fixing symbiosis between M. truncatula and S. medicae WSM419.
Different symbiotic efficiency has been reported for Sinorhizobium sp.36,37, therefore we assayed the strain-dependent nodulation phenotype of regenerated ncr068 mutant in response to S. meliloti strains 2011 and FSM-MA compared with S. medicae WSM419. The growth habit of wild-type M. truncatula 2HA and ncr068 mutant plants was similar following inoculation with any of the tested strains (Fig. 6). We observed a similar shoot dry weight of wild-type and ncr068 plants inoculated with S. meliloti FSM-MA or S. medicae WSM419 (Fig. S5). In agreement with published results38, the performance of wild-type and, in our study ncr068 plants as well, inoculated with S. medicae WSM419 superseded the plants inoculated with S. meliloti 2011 (Fig. S5). The dnf7-2 mutant showed the previously described ineffective symbiotic phenotype with all three tested rhizobia strains (Fig. 6). In line with the growth habit, the nodules developed on wt and ncr068 roots inoculated with the highly compatible strains S. medicae WSM419 and S. meliloti FSM-MA were cylindrical, pink and fully colonized by rhizobia but nodules elicited by the less effective strain S. meliloti 2011 were only slightly elongated (Fig. 6). In brief, no difference in plant development and nodule structure was noticed between wt and ncr068 plants with the tested rhizobia which suggests that the symbiotic phenotype of plants lacking NCR068 is not strain-dependent.