Transgene-based editing with CRISPR-Cas9 is presently a popular technique for introducing targeted mutations in model species and crops. Identification of transformant events relies on the introduction of selectable markers, typically genes encoding proteins that mediate resistance to antibiotics or herbicides. A selectable marker that can be obtained by editing, i.e., based on loss of function of an endogenous gene, would be advantageous for transgene-dependent as well as transgene-free genome editing approaches. Here, we developed a recessive marker that is caused by loss of function of an endogenous rice gene, i.e., LAT5/PUT3/PAR1, to confer resistance to the phytotoxin methylviologen (MV) as a method of selection. Note that, others used the LAT homolog PAR1 from Arabidopsis to establish a similar selection system (15). Also in parallel, Lyu et al. (30) demonstrated that disruption of three rice polyamine uptake transporter genes, namely LAT1/PUT1, LAT5/PUT3/PAR1, and LAT7/PUT2, confers resistance to MV. Our data indicate that disruption of a single polyamine uptake transporter gene in rice, i.e., LAT5/PUT3/PAR1, is sufficient to provide resistance to MV (Fig. 3C, Fig. 4C and Fig. 5C). RNA-seq experiments show FPKM (Fragments Per Kilobase of transcript per Million mapped reads) values of 25 in both leaves and roots and > 100 in calli for LAT5/PUT3/PAR1 (35)(S4A). By comparison, LAT1/PUT1 had substantially lower mRNA levels in leaves roots and calli (S4B). LAT7/PUT2 was also low in leaves and roots, but had FPKM values of about 24 in calli (S4C). The low base levels of the other two genes, namely LAT1/PUT1 and LAT7/PUT2, in calli and leaves may explain why clear resistance to MV was only observed for LAT5/PUT3/PAR1 (Fig. 3C, Fig. 4C and Fig. 5C); however alternative hypotheses are also conceivable, e.g., different substrate specificity, lower transport capacity or different subcellular localization.
Because disruption of LAT5/PUT3/PAR1 provides effective MV resistance (Fig. 3C, Fig. 4C and Fig. 5C), gRNAs that target regions that encode transmembrane helices can be used for selecting transformants in transgene-based and transgene-free genome editing experiments. We demonstrate MV-based selection in photosynthetically active rice tissues, e.g., germinating seeds (Fig. 3C) and young seedlings (Fig. 4C). MV resistance of green tissues is consistent with MV being known to transfer electrons from photosystem I to molecular oxygen, which leads to formation of cytotoxic reactive oxygen species (ROS) and photodestruction of chlorophyll (26, 37). Notably, MV-based selection was also successfully obtained for non-photosynthetically active rice calli (Fig. 5). Callus was grown photoautotrophically in a medium supplemented with 3% of sucrose. The high sucrose levels in the media appeared to prevent cytotoxic symptoms on the WT callus cells grown on the medium containing 0.1 µM MV (38)(Fig. 5B), while at 1 µM MV callus induction from WT mature embryos was inhibited effectively (Fig. 5C). Our results are in line with data from Zer et al. (39), who demonstrated that high MV concentrations inhibited growth of Phaseolus vulgaris cells during cultivation in darkness, likely by reducing DNA synthesis, and inhibiting the activity of enzymes involved in cellular defense against ROS. The deleterious effects of MV on non-photosynthetic cells was attributed to iron (39).
Since 1 µM MV blocked callus induction from WT mature embryos (Fig. 5C), MV-based selection of lat5/put3/par1 calli directly after transformation or transfection is possible. In the case of Arabidopsis, PAR1 null alleles generated by CRISPR-Cas9 were shown to confer resistance to 1 µM and 10µM MV (15). The authors stated that heritable transgene-free mutations at target loci were identified in the T1 generation. Apparently, although the inheritance of par1 was considered to be recessive at 10µM MV, intermediate inheritance was be observed at lower 1 µM MV concentrations. In our study reduction in MV concentrations enabled selection on 20-200-fold lower MV concentrations compared to Kong et al. (15), i.e., 0.05 µM and 0.1 µM in rice. At MV concentrations of 0.1 µM, a growth inhibition on neighboring calli can not be excluded but appears to be minor or negligible.
We surmise that lat5 can be used as a selectable marker for base and prime editing approaches via introduction of a premature stop codon in LAT5 using suitable gRNA / prime editing gRNA (pegRNA).
Since genome-editing, at least in rice, is highly efficient, frequently yielding biallelic mutations, e.g., high rates of multisite biallelic mutations were reported in several studies (31, 32, 40). Even prime editing efficiency using PE5max was recently shown to be highly efficient (7). Therefore, a recessive selection marker, although likely to reduce the total number of events, will only be disadvantageous relative to dominant markers if editing efficacy is low, e.g., due to guide RNA design, or due to low transformation efficiencies.
Identification of lines that do not carry a transgene is an essential prerequisite for the classification of edited lines to be treated equivalent to under regulations (41). One of the challenges is however that Agrobacterium-mediated transformation can lead to partial insertion of T-DNA copies (12). Another challenge is the inadvertent insertion of vector backbone fragments into the genome of transformed plants (42, 43). Genotyping of hornless bulls generated with the help of genome editing identified the unintended insertion of plasmid sequences into the genome (44). Thus, without further analyses, it is not possible to conclude that hygromycin-sensitive MV-resistant plants are transgene-free. Transgene-free editing based on transfection with CRISPR-enzyme-gRNA complexes or TALENs will be the new frontier to overcome the issues caused by Agrobacterium-mediated transformation.