Seeing Red, Selecting True: RUBY-Reported Seed Marker Streamlines CRISPR-Clean Rice Breeding

Abstract

See More

The Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR)-CRISPR-associated protein (Cas)9 genome-editing technology has become a cornerstone for generating knockout mutations in plant functional genomics. To obtain genetically stable CRISPR-edited plants, the removal of exogenous CRISPR constructs through genetic segregation is imperative. However, current transgene-free strategies lack universality and operational simplicity. Here, we developed a modular CRISPR toolkit integrated with the widely applicable visual RUBY marker. This system achieved 100% editing efficiency in three independent CRISPR-Cas9 editing events in rice (cv. Zhonghua 11), enabled rapid visual identification of transgene-free progeny, and may provide a framework for future adaptation of CRISPR vectors to other plant species. Our design significantly accelerates the identification of edited lines while bypassing laborious molecular validation steps.

Introduction

A critical challenge in CRISPR applications is the efficient recovery of transgene-free progeny. Current strategies relying on Mendelian segregation require labor-intensive PCR screening of T1 populations (He and Zhao 2020). While visual markers like anthocyanin reporters or fluorescent proteins (e.g., DsRed, Discosoma sp. Red) simplify selection, their application scope remains limited (Shimada et al. 2010). The Transgene Killer CRISPR (TKC) system achieves auto-excision through pollen grains-specific and seeds toxins (CMS2 and BARNASE) (He et al. 2018; Liu et al. 2022; Yang et al. 2022), yet its species-specificity restricts broader application. Recently, the chimeric RUBY marker, which produces visible red betalain pigments from endogenous tyrosine, has been effectively used to monitor gene expression and transformations in various plants, including rice, Arabidopsis, tobacco, bamboo, poplar, cotton, and so on (Ge et al. 2022; He et al. 2020; Li et al. 2023; Sun et al. 2023; Yuan et al. 2022). A gallery of RUBY plants is shown here (https://zhaolab.biosci.ucsd.edu/ruby/). Additionally, CRISPR-Cas9-mediated genome editing has revolutionized targeted mutagenesis in plants. Despite its widespread adoption, the development of species-specific CRISPR vectors remains technically demanding, particularly due to constraints imposed by restriction enzyme compatibility in traditional cloning systems. Even advanced modular platforms like GoldenBraid (Vazquez-Vilar et al. 2020) require substantial optimization for different species. There is an urgent need for truly modular vector architectures that permit seamless component exchange without compromising editing efficiency.

In this study, we present a RUBY-assisted Modular CRISPR-Cas9 toolkit (RMC) incorporating an endosperm-specific RUBY marker driven by the OsGluC promoter. This system enables: (1) Visual identification of transgene-free progeny in husked T1 caryopses (harvested from T0 plants); (2) High-efficiency editing through a tRNA-gRNA polycistronic design; (3) Species-agnostic adaptability via plug-and-play module replacement. We demonstrate its efficacy in rice through targeted knockout of OsCCD8 and OsLAZY, achieving 100% transgene elimination through segregation in selected lines.

Results

We first constructed a modularized CRISPR-Cas9 gene-editing vector, denoted as PMC-OsR41 (Fig. 1A). This vector contained three distinctive modules: a plant resistance marker module, a singular transcript CRISPR module consisting of a maize Ubiquitin promoter, Cas9 and a self-cleavable tRNAGly-assisted gRNA cassette (Tang et al. 2016), and a transgene-free module with a RUBY driven by OsGluC (Oryza sativa Glutelin C, Os02g0453600, a rice endosperm-specifically expressed glutelin gene) promoter (Fig. 1A; Supplementary Data Set 1; Qu et al. 2008).

To examine the genome editing efficiency of PMC-OsR41, two distinct target sites in OsCCD8 (Oryza sativa CAROTENOID CLEAVAGE DIOXYGENASE 8, Os01g0746400) and one target site in OsLAZY (Os11g0490600) were chosen, spacer designed and cloned into three CRISPR module, respectively. Following stable rice transformations (Hiei et al. 1994), transgene-positive T0 plants were identified by PCR amplification of the Cas9 cassette using the primers Cas9-VR-FP1 and NOS-RP1 (Fig. S1; Table S1). Specifically, 21 positive plants were obtained for target site 1 and 15 for target site 2 in OsCCD8, and 15 positive plants for target site 1 in OsLAZY (Fig. S1). To facilitate rapid and efficient identification of CRISPR-Cas9-induced insertions or deletions (InDels), sgRNAs were designed to target sequences containing restriction enzyme recognition sites. In detail, the target site 1 and site 2 in the OsCCD8 harbor a Sal I and an Xho I restriction sites situated proximally upstream to the PAM (Protospacer Adjacent Motif) sequence, respectively. In addition, the OsLAZY target contains a Sbf I site (Table S1). These designed targets allow for subsequent mutation screening through PCR followed by restriction enzyme digestion since InDels at the OsCCD8 and OsLAZY target sequences were anticipated to eliminate the corresponding restriction enzyme sites (He et al. 2018). Therefore, the potential mutations in these transgene-positive T0 plants were detected via PCR and restriction enzyme digestion. The results showed that all 51 T0 transgene-positive plants were fully resistant to the digestion, suggesting that OsCCD8 and OsLAZY were mutated in these plants (Fig. 1B). Then, the mutated loci of all these positive plants were Sanger sequenced. Consistent with the results of restriction enzyme digestion of the PCR products, every single T0 plant contained InDel mutations at the corresponding target site, existing as either homozygous or biallelic mutations (Fig. 1C, D; Fig. S3; Table S2). These results demonstrate that the PMC-OsR41 system achieved 100% editing efficiency across all three target sites tested (Fig. 1B; Fig. S1; Table S2).

Furthermore, phenotypical analysis revealed that T0 plants containing knockout mutations at OsCCD8 exhibited dwarfism and increased tiller numbers (Fig. 1E-G), while those with knockout mutations at OsLAZY displayed a tiller-spreading phenotype (Fig. 1E, H). These abnormalities are consistent with previously reported loss-of-function mutants of OsCCD8 and OsLAZY, supporting the validity of CRISPR editing approach (Arite et al. 2007; Li et al. 2007). It is worth noting that in both events of OsCCD8 editing, most T1 plants failed to boot and did not produce seeds (Fig. 1E–G). Only four out of 15 lines from the event of the OsCCD8 target site 2 successfully generated seeds. This observed booting failure and resulting sterility were not reported in previous studies and may be attributed to differences between our CRISPR-induced knockout mutations and the previously characterized SNP mutation (a T-to-C substitution resulting in a Leu112 to Pro amino acid change) described by Arite et al. (2007). Notably, all four seed-producing lines were biallelic mutants carrying weak in-frame mutations (Table S2), suggesting that less severe mutations preserved partial gene function, allowing booting and seed development. In contrast, T0 lines with presumed strong frameshift mutations failed to boot and set seeds, indicating that complete knockout of OsCCD8 disrupts fertility (Fig. 1E–G). Genotyping of their T1 progenies further confirmed the inheritance of both weak in-frame homozygous and frameshift mutations derived from the biallelic T0 plants (Table S3). These observations emphasize the importance of mutation type and severity in determining phenotypic outcomes and fertility, and highlight the necessity of generating true knockout alleles for rigorous functional analysis. Collectively, these findings demonstrate that the PMC-OsR41 system is highly effective for generating targeted knockout mutants in rice.

Next, the transgene-free effectiveness of the PMC-OsR41 was evaluated. Although the visual red color cannot be detected in unhusked T1 seeds (harvested from T0 transgenic edited plants), the husked T1 seeds can be easily classified into two groups—those with and without the red RUBY marker—according to Mendelian segregation, with an expected 3:1 ratio assuming a single-locus CRISPR T-DNA insertion (Fig. 1I). The OsGluC-driven endosperm-specific expression of RUBY enabled visual identification of T1 seeds harboring the PMC-OsR41 transgene through distinct red pigmentation. In contrast, normal-looking caryopses without red pigment would indicate that transgene-free progenies have successfully eliminated the transgenic fragment through segregation. Importantly, consistent with previous findings (Tian et al. 2020) and our own observations (Fig. 1I), RUBY/betanin accumulation did not negatively affect plant growth or seed setting. Due to booting failure and the absence of panicle development in most lines from the two events of OsCCD8 editing (Fig. 1E–G), we selected the four seed-producing lines from the OsCCD8-T2 event and 12 out of 15 lines from the OsLAZY-T1 event. In total, 16 independent lines were collected for subsequent transgene-free analysis. To evaluate the performance of this OsGluC::RUBY transgene-free module for visual selection of transgene-free progenies, we in total identified 170 T1 normal-looking caryopses, lacking red RUBY pigmentation, from those 16 independent T0 plants as shown in Fig. 1I. Following germination, PCR amplification of the Cas9 gene using the primers Cas9-VR-FP1 and NOS-RP1 was conducted to determine whether the CRISPR transgene had been segregated out in these T1 seedlings. As shown in Fig. S2, all of these T1 plants analyzed were transgene-free plants which did not contain the CRISPR construct, highlighting the effectiveness of our transgene-free module. To further validate the inheritance of CRISPR-induced mutations, we conducted genotyping of T1 progenies from 11 edited T0 lines, comprising four biallelic OsCCD8-T2 lines (the only fertile lines among the OsCCD8 edits), five biallelic, and two homozygous OsLAZY-T1 lines (Table S2; Table S3). All mutations were heritable. In progenies of biallelic lines, three genotypes were observed: two different homozygous types and one biallelic type. In contrast, progenies of homozygous lines remained uniformly homozygous. Interestingly, in one biallelic OsLAZY-T1 line, only homozygous progenies were detected, likely due to the limited number of plants analyzed (n = 15) (Table S2; Table S3). Collectively, these results robustly demonstrate that integration of the CRISPR system with the OsGluC::RUBY cassette provides an efficient and reliable visual genome editing platform, allowing rapid identification of transgene-free seeds and stable fixation of inherited mutations during CRISPR-mediated gene knockout applications (Fig. 1J).

It is worth noting that in some cases, editing efficiency may be low and only heterozygous edited T0 plants are recovered. In such scenarios, T1 seeds harvested from these T0 plants can be visually selected for normal-looking (transgene-free) seeds and genotyped to identify heterozygous or homozygous edited individuals, which can then be propagated. While propagation of RUBY-positive seeds is generally unnecessary, it remains feasible. Because the endosperm-specific expression of RUBY does not affect plant growth or seed production (Fig. 1I; Tian et al. 2020), red seeds can also be grown to produce the T2 generation, from which transgene-free homozygous edited seeds can again be visually selected from the panicle that contains both red and normal-looking seeds.

In our design, using the PMC-OsR41 as a backbone, CRISPR vectors for the various plant species can be readily developed by replacing specific modules (Supplementary Data Set 1). In plant species, the efficiency of CRISPR-Cas9-mediated genome editing and streamlined RUBY-based color visualization are critically dependent on key technical parameters, including: Cas9 promoter selection (constitutive or tissue-specific promoters), codon optimization of Cas9, gRNA cassette design, and spatiotemporal modulation of RUBY reporter expression to match target tissues. Consequently, our design provides an alternative to assess the editing efficiency associated with diverse combinations of these elements, facilitating the development of an optimized CRISPR vector for a specific plant species (Fig. 1A; Table S4; Supplementary Data Set 1). Future experimental validation of these constructs in other species will be essential to broaden the utility of this RMC toolkit (Table S4; Supplementary Data Set 1). Furthermore, our design is compatible with the construction procedure for multiplex gene editing, utilizing either PCR amplification or gene chemical synthesis with a Golden Gate protocol as reported by Xie et al. (2015).

In summary, our RUBY-integrated modular RMC toolkit addresses two critical bottlenecks in plant genome editing: (1) Streamlined vector optimization through component modularity; (2) Visual selection of transgene-free progeny without molecular screening. The system’s compatibility with diverse plant species and editing platforms positions it as a versatile solution for both basic research and biotechnological applications.

Fig. 1

A modularized CRISPR toolkit with RUBY-assisted transgene visibility. A A schematic diagram of modularized PMC-OsR41 CRISPR vector. OsGluC is a reported endosperm-specific promoter. RUBY is a triple chimera of betalain synthases (He et al. 2020). Target upload site is a 95-bp sequence with quadruple BsaI recognition sequence (GGTCTCN, N indicates A, G, C or T) which allows the upload of target in a one-step GoldenGate manner. Black triangles indicate four BsaI recognition sequences. The eight unique restriction enzyme sites are marked with arrows. B Mutation analysis of the target with restriction enzymatic reactions. From top to bottom, the five lanes of the marker on the right are 3 Kb, 2 Kb, 1 Kb, 750 bp and 500 bp, respectively. WT indicates the digested PCR product amplified with the wild type genome DNA. The digested PCR products of WT (Negative control at the most right lane) result in two bands, whereas all PCR products of positive transformants are resistant to digestion. Three targets are shown in Table S1. C The ratio of different mutation types of T0 positive transformants. D Sanger sequencing of the T0 positive transformants that showed in F and G, H, respectively. The three targets were underlined with the 3 bp PAM sequence in bold characters. Mutations are shown in red font. E–H Plant architecture of wild type (E), osccd8 knockout mutant generated from OsCCD8-Target 1 (F) and OsCCD8-Target 2 (G), oslazy knockout mutant generated from OsLAZY-Target 1 (H). Bar = 10 cm. I T1 seeds harvested from T0 positive transformants. Top, unhusked T1 seeds; middle, husked T1 RUBY-marked red transgenic seeds; bottom, husked T1 normal-looking transgene-free seeds. Bar = 1 cm. J A workflow for the application of PMC-OsR41 vector in rice

All gene sequences were downloaded from RAP-DB (https://rapdb.dna.affrc.go.jp/), TAIR (http://www.arabidopsis.org/), EnsemblPlants (http://plants.ensembl.org/index.html), and NCBI (https://www.ncbi.nlm.nih.gov/nuccore/?term=). The well-domesticated PMC-OsR41 vector was applied as a backbone for further optimization. A complete list of genes utilized in this study is provided in Table S4. All customized vectors were listed in Table S5 with detail sequences listed in Supplementary Data Set 1.

Materials and Methods

Plant Material and Growth Condition

Rice (Oryza sativa L. ssp. Japonica) cultivar Zhonghua 11 (ZH11) was used for stable transformation in the study. The wild type and transgenic plants of ZH11 were grown in the greenhouse at the Institute of Botany, the Chinese Academy of Sciences (Beijing, China), with about 60% humidity and 12 h-light / 12 h-dark cycle, with temperature as 30 ± 2 ℃ during the day, and 22 ± 2 ℃ during the night.

Target Design and Vector Construction

The editing targets in OsCCD8 (Os01g0746400) and OsLAZY (Os11g0490600) were designed with CRISPR-GE (http://skl.scau.edu.cn/home/; Xie et al. 2017). All primers used in this study were listed in Table S1. The construction procedure is described as our former report (Liu et al. 2022).

Rice Transformation and Transformation Rate

Stable transformation of rice was conducted by the commercial service provider BioRun Biotechnology Co., Ltd. (Wuhan, China). Transformation efficiency was calculated as the ratio of PCR-confirmed positive T0 transformants to the total number of regenerated T0 seedlings for each specific editing construct. Identification of positive transformants was performed through PCR-based genotyping, as shown in Figure S1. The transformation efficiencies for the three independent transformation events were 95% (21/22), 63% (15/24), and 47% (15/32), respectively.

Genotyping Analysis of CRISPR/Cas9 Editing Events

The target genes of all positive transformants were amplified using a 20 μL PCR reaction volume with KOD-FX DNA polymerase (Toyobo) for 40 cycles. Primers used for genotyping are listed in Table S1. For restriction enzyme digestion analysis, 2 μL of the PCR product was digested in a 20 μL reaction volume at 37 °C for 2 hours, followed by detection on a 1.2% agarose gel. All restriction enzymes were purchased from New England Biolabs. The target genes from all T0 positive transformants and transgene-free T1 progenies were PCR-amplified, Sanger sequenced, and the resulting chromatograms decoded using DSDecodeM (http://skl.scau.edu.cn/home/; Liu et al. 2015; Xie et al. 2017), as summarized in Table S2.

Observation and Verification of Transgene-Free Seeds

The T1 seeds were harvested from T0 positive transformants, dried in 37 ℃ dryer for a week, and then, husked. The normal-looking seeds without RUBY marked red color were selected as transgene-free seeds. Further genotypic verification with PCR amplification and Sanger sequencing of the target gene was conducted (Fig. S4), with the primers were listed in Table S1. All T1 seeds were observed with the naked eye and photographed under a dissecting microscope (SMZ800, Nikon).

Vector Domestication and Construction

All gene sequences were downloaded from RAP-DB (https://rapdb.dna.affrc.go.jp/), TAIR (http://www.arabidopsis.org/), EnsemblPlants (http://plants.ensembl.org/index.html), and NCBI (https://www.ncbi.nlm.nih.gov/nuccore/?term=). The well-domesticated PMC-OsR41 vector was applied as a backbone for further optimization. A complete list of genes utilized in this study is provided in Table S4. All customized vectors were listed in Table S5 with detail sequences listed in Supplementary Data Set 1.