A molecular toolkit to boost functional genomic studies in transformation-recalcitrant vegetable legumes

© The Author(s) 2023. Published by Oxford University Press on behalf of Nanjing Agricultural University. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Horticulture Research, 2023, 10: uhad064


Dear Editor
Legumes, the second-largest family of crops, contribute over one-third of human dietary proteins. Soybean (Glycine max L.), common bean (Phaseolus vulgaris L.), pea (Pisum sativum L.), and cowpea (Vigna unguiculata L.) are among the most widely cultivated crop legumes for grain and vegetable and are essential for food security globally. Their reference genomes have been decoded but, with the only exception of soybean, their functional genomics have lagged far behind genome assembly due to the transformation-recalcitrant nature of these species [1]. Hairy root transformation has been well established, but its usefulness in gene function investigation in aerial organs is limited. Virusinduced gene silencing (VIGS) has also found applications, but the most widely distributed bean pod mosaic virus-based system relies on the costly biolistic method [2]; other systems, like pea early browning virus-based VIGS, have not been proved for their universality. Genetic and omics studies have identified numerous quantitative trait loci (QTLs)/candidate genes governing various agriculturally important traits [1]. This necessitates the development of efficient and reproducible research tools for verification of gene function.
Here, we report on the development of a molecular toolkit encompassing in planta transient overexpression and RNA interference, virus-induced overexpression and multi-target gene editing, all based on convenient agroinoculation. The first tool is the Agrobacterium-mediated gene overexpression/silencing system, which is suitable for molecular assays like protein subcellular location and bimolecular f luorescence complementation (BiFC). From screening four genotypes, 1-week-old seedlings of the common bean cv. 'Honghuabaijia' were found to be optimal, and provided appropriate seedling size and leaf tenderness. Two Agrobacterium strains (GV3101 and EHA105) carrying the PCV-GFP or PBI121-GFP constructs (four combinations in total) were tested in a negative pressure vacuum pumping method. Brief ly, Agrobacterium was infiltrated into the abaxial surface of the leaves by negative pressure (0.08 MPa) for 3 minutes (Fig. 1a). At 60-96 hours post-infiltration (hpi), f luorescence in the infiltrated leaves was examined, which revealed widespread GFP f luorescence, with the leaves infected by GV3101 strain carrying PCV-GFP expressing the strongest signals ( Fig. 1b and c). Using this optimized protocol, we tested the robustness of the system using the BiFC assay. Genes encoding two known interactive proteins, TGB1 and TGB2, from cowpea mild mottle virus (CPMMV) were cloned and inserted into the PCV-NYFP and PCV-CYFP vectors (Table 1). Three days following vacuum infiltration with GV3101 carrying TGB1-NYFP and TGB2-CYFP, yellow f luorescence was seen around the cell membrane and in the nucleus, corroborating physical interaction of the two proteins (Fig. 1d). Notably, BiFC assay of the same proteins in the Nicotiana benthamiana heterogeneous system revealed spotted f luorescence signals [3]. Given that CPMMV naturally infects legumes but not N. benthamiana, the assay in the natural host species is considered to be more biologically meaningful. This method is also highly valuable for validating gene functions involved in abiotic stress, as recently reported by Fang et al. [4] for a heavy-metal responsive gene, PvXTH23, encoding xyloglucan endo-transglycosylase (Fig. 1e). The suitability and robustness of the method was also proved in cowpea (cv. A1212, Fig. 1k, l, and m) and pea (cv. 610, Fig. 1r and s). Agrobacterium could be well infiltrated into cowpea and pea leaves through a syringe, and thus the protocols were more simplified and f lexible in these crops.
We next demonstrated the adapted protocol of Agrobacteriummediated transient gene silencing (AMTS). We selected PvABA3, a gene involved in the last step of abscisic acid (ABA) biosynthesis and the regulation of stomatal closure, as the target [5]. To create a stem-loop structure, a 300-bp fragment of PvABA3 and its reverse complement fragment were amplified and inserted into the vector pFGC-5941 (Fig. 1b). Common beans were then infiltrated with Agrobacterium GV3101 harboring the PvABA3-RNAi or GFP-RNAi control vector. qRT-PCR showed that PvABA3 expression at 72 hpi was downregulated to ∼26% of that in the GFP-RNAi control (Fig. 1f). We observed a 1.5-fold increase in stomatal aperture in the PvABA3-RNAi-infiltrated leaves, supporting its role in negatively controlling stomatal opening. We checked the expressions of four known ABA-pathway downstream genes, Figure 1. The developed molecular toolkit for legume crops. a Negative pressure vacuum pumping for Agrobacterium strain infiltration. b Constructed vectors used in this study. c, k, l, r GFP f luorescence images of in planta transient overexpression; c shows confocal microscopy images of common bean leaves infiltrated with two Agrobacterium strains (GV3101 and EHA105) carrying the PCV-GFP or PBI121-GFP construct; k is a GFP image of infiltrated cowpea leaves under a f luorescence lamp; l and r are confocal microscopy images of cowpea and pea leaves infiltrated with PCV-GFP. d, m, s BiFC images of TGB1 and TGB2 interaction in common bean, cowpea, and pea leaves, respectively. e Phenotypes of transiently expressed common bean seedlings before and after CuCl 2 treatment. Pictures are reproduced from Fang et al. [4]. f, n, t Representative images of stomata (scale bars = 10 μm) from the control and ABA3-silenced common bean, cowpea, and pea plants, respectively, and the relative expression of ABA3 and ABA pathway downstream genes as well as stomatal apertures. g, o Confocal microscopy images of TRV-GFP-and TRV-PR1-GFP-infiltrated common bean and cowpea cells at 7, 11, and 15 dpi. h, p, u Phenotypes of TRV-GFP-and TRV-PR1-GFP-infiltrated common bean, cowpea, and pea plants inoculated with the pathogen BBWV2. 1-week-old common bean and cowpea plants were infiltrated with Agrobacterium, inoculated with BBWV2 at 7 dpi and imaged at 14 dpi. 2-week-old pea plants were infiltrated with Agrobacterium, inoculated with BBWV2 at 7 dpi and imaged at 21 dpi. i, q, v Levels of TRV and BBWV2 mRNA in leaves of TRV-GFP-and TRV-PR1-GFP-infiltrated common bean, cowpea, and pea plants. j Structure diagram, PvPDS sgRNA target positions and sequences, and the sequences of edited and non-edited PvPDS gene. Different letters in histograms indicate significant differences (P < .05). Unless specified, scale bars = 50 μm.

Gene ID Primer name Sequence 5 -3
Phvul.006G197400 PsSNRK2.3-RT-F ATCTTCGGTGCTTCATTCA PsSNRK2. 3-RT-R  AATACTGGACGCTGAGAAC  CPMMV-TGB1   TGB1-F  ATGAATGAACTGATCAGTAAAC  TGB1-R  TCACTCAGAGTTTGGATAGG  CPMMV-TGB2   TGB2-F  ATGCCACTGACTCCACCACC  TGB2-R  TCAGTGAACCCTATTGCAGA  TRV  TRV-RT-F  TCTACTTCGAACCGTGGCAG  TRV-RT-R  CCAACTCTCGCGTTGATTCG  BBWV2 BBWV2-RT-F TCAATTGCCAGGTAGCTCCG BBWV2-RT-R TTCCACCAATCCGCACAAGA including the transcription factor genes PvABF2 and PvABI5L and the kinase genes PvSNRK2.1 and PvSNRK2.3. Except for PvSNRK2.3, all genes were downregulated in the PvABA3-RNAiinfiltrated leaves (Fig. 1f), consistent with PvABA3 interference. This system was proved to also be functional in cowpea and pea with VuABA3 and PsABA3 as the target gene (Table 1), respectively, demonstrating great promise for studying gene functions in legumes ( Fig. 1n and t). Nevertheless, the short duration (60-96 h) of the transient systems restricts their use in studies where prolonged gene expression is required. Thus, we adapted a tobacco rattle virus-induced overexpression (TRV-VOX) system for continuous expression of transgenes, with no need for particle bombardment or harming the normal growth of plants. We cloned PvPR1, a known broad-spectrum resistance gene [6] to create TRV-RNA2-PvPR1-GFP ( Fig. 1b and Table 1). Agrobacterium GV3101 carrying TRV-RNA1/TRV-RNA2-PvPR1-GFP constructs were co-infiltrated into common bean leaves, with TRV-RNA1/TRV-RNA2-GFP as control. A devastating pathogen, broad bean wilt virus 2 (BBWV2), was inoculated at 7 days post-infiltration (dpi). After transformation, the GFP f luorescence became visible at 7 dpi and lasted at least until 15 dpi (Fig. 1g). Compared with the control, plants overexpressing PvPR1 displayed milder disease symptoms and lower pathogenic mRNA accumulation following BBWV2 infection ( Fig. 1h and i). The system was also tested in cowpea and pea by overexpressing VuPR1 (Fig. 1o, p and q) and PsPR1 ( Fig. 1u and v), respectively, where similar enhanced resistance to BBWV2 was observed. Notably, pea plants at an older age (2 weeks) and larger size were infiltrated using a syringe. We attempted to use the TRV-based VIGS system in the legume vegetables, but failed to see positive results despite detection of TRV in the inoculated leaves. We speculate that due to the incompatibility of TRV and the silencing machineries in legumes, the production of small RNAs was inefficient.
CRISPR/Cas-mediated gene editing has been reported in cowpea and pea [7]. To date, gene editing has not been utilized in common bean, and hence we established a high efficiency, multitarget gene editing system in this species. Four sgRNAs (Fig. 1j) were designed to target exons 2, 3, and 5 of the PvPDS gene, encoding a phytoene desaturase, which were then simultaneously built into the multiplex knockout vector pGmUbi-Cas9-4XsgR initially designed for soybean [8]. The vector was introduced into the Agrobacterium rhizogenes K599 strain. 1-week-old seedlings (cv. 'Biyuhonghua') were infected with the transformed K599 at the hypocotyl as described [9]. Four weeks later, the induced hairy roots were collected for PCR of the marker gene Bar, and the Bar gene could be detected in all seedlings. Sequencing results of the amplified fragment showed the target sequences were edited in 11 out of the 16 plants, demonstrating a high editing efficiency of 68.75% (Fig. 1j). This result demonstrated that the pGmUbi-Cas9-4XsgR system-based multiplex gene editing works effectively in common bean, holding great promise for genetic manipulation in legume crops.
Collectively, we developed and exemplified the use of an efficient and easily operational toolkit for transformationrecalcitrant vegetable legumes, which can largely break the long-standing bottleneck of functional studies. In planta transient overexpression, different from leaf disk infiltration [10], provides a means of studying protein subcellular localization and interactions, which outperforms heterogeneous systems; when coupled with AMTS, it is useful for rapid validation of gene functions. The TRV-VOX system adapted for legumes, in conjugation with existing VIGS approaches, offers a solution for gene functional studies in the longer term. The multiplex gene editing system outperforms single-target gene editing and will greatly facilitate gene knockout. Future work is needed to exploit the usefulness of this toolkit in more legume crops, such as chickpea and pigeon pea, to boost their functional genomics.