Peanut (Arachis hypogaea L.) or groundnut is an important legume crop due to the seed being a rich source of edible oil, protein and fiber. The oil content is about 50%, of which oleic and linoleic acids account for 80% of the total fatty acid with the ratio of these components largely determining oil quality. Therefore, the conversion of monounsaturated oleic acid to polyunsaturated linoleic acid by the Fatty Acid Desaturase 2 (FAD2) enzyme is a key regulatory point. Peanut oil is naturally high in linoleic acid, making it prone to oxidation, which reduces oil stability leading to rancidity and poor flavor characteristics. However, linoleic acid is important for healthy plant growth and it is also an essential fatty acid for humans (Guan et al., 2012; Marangoni et al., 2020). High oleic acid is beneficial for improved shelf life through oxidative stability, improved flavor characteristics and consumption may lead to better cardiovascular health in humans (O’Byrne et al., 1997; Yu et al., 2008; Lim et al., 2017). The natural FAD2 knockout mutant line, F435, is devoid of FAD2 enzyme activity in seed and accumulates 80% oleic acid as observed by Norden et al., 1987. On the surface, these mutations may seem optimal for oil quality, however, introgression of these mutations into other peanut varieties is time consuming and knockout mutations may leave plants vulnerable during periods of stress (Zhang et al., 2012). Therefore, an ideal strategy to engineer oil composition without compromising plant health may involve downregulation of FAD2 gene expression in seed while minimizing pleiotropic effects in other vegetative tissues through promoter modification.

The FAD2 gene family has been extensively characterized from many plant species including peanut (Janila et al., 2016; Dar et al., 2017). Cultivated peanut is an allotetraploid (AABB, 2n = 4x = 40) and the genome contains two functional homeologous genes, AhFAD2A and AhFAD2B, derived from the diploid ancestors Arachis duranensis and Arachis ipaensis, respectively. FAD2 is the key enzyme that is responsible for biosynthesis of polyunsaturated acid in non-photosynthetic tissues, such as roots and developing seeds in oilseed plants (Miquel and Browse, 1992), while some FAD2 genes are targeted to stems and leaves (Cao et al., 2013; Lee et al., 2020). FAD2 genes typically have a conserved single large intron in the 5′-untranslated region (UTR), which has been suggested to have enhancer activity in mediating transcriptional regulation. This has been shown in sesame (Kim et al., 2006), and Brassica napus (Xiao et al., 2014) by deletion mapping approaches and Arabidopsis and olives by gene association studies (Menard et al., 2017; Salimonti et al., 2020).

Seed development, including embryogenesis and seed maturation, is regulated by a combination of hormonal, genetic and metabolic controls (Gutierrez et al., 2007). Differing clusters of cis-regulatory elements (CREs) in the promoters of seed associated proteins are bound by various combinations of transcription factors (TFs) including LEAFY COTYLEDON1 (LEC1), ABSCISIC ACID INSENSITIVE3 (ABI3), FUSCA3 (FUS3), and LEC2 (Parcy et al., 1997; Boulard et al., 2017; Jo et al., 2019; Tang et al., 2021). These interactions lead to formation of tightly controlled regulatory pathways in seed and other tissues making CREs potentially useful targets for altered gene expression. Among the CREs, the RY repeat element (Bäumlein et al., 1992; Fujiwara and Beachy, 1994) and 2S seed protein motif (Stålberg et al., 1996) are crucial for regulatory activity in a number of seed-specific promoters (Reidt et al., 2000; Chatthai et al., 2004).

RNA programmable CRISPR/Cas nucleases trigger sequence-specific double stranded breaks (DSB)s that are subject to error prone non-homologous end joining (NHEJ) repair or template dependent, homology-directed repair (HDR) pathways. NHEJ, the most frequently used repair mechanism in plants, often leading to insertions or deletions (indels) at the target site. This technology is used to generate gene knockouts when coding regions are targeted, therefore, this approach has been widely used for plant genome editing applications (Manghwar et al., 2019).

Our overarching goal was to utilize CRISPR/Cas9-mediated gene editing to modify cis-regulatory elements in the 5′ UTR and intron of FAD2 genes for functional characterization of these elements and to potentially generate seeds with increased oleic acid content without affecting the fatty acid composition in other plant tissues. We designed guide RNAs (gRNAs) specific to the RY repeat element and 2S seed protein motif of peanut FAD2 genes then transformed peanut tissues to explore the potential of modifying transcriptional regulation of FAD2 gene expression in seeds. Our findings demonstrated high frequency modification of FAD2 RY and 2S motifs by CRISPR/Cas9-mediated gene editing and alteration of the oil profile in seed.

The FAD2A (AH09G29670) and FAD2B (AH19G38370) gene models used in this study have a 5′ UTR with a single intron, coding sequence, and a 3′ UTR as shown in Figure 2. FAD2A has two transcript models and FAD2B has four transcript models with altered 5′ UTRs, and two transcriptional start sites predicted for each gene. We considered the transcript model AH09G29670.1 for FAD2A and AH19G38370.1 for FAD2B as a reference for this investigation. Mining of transcriptomic data (http://peanutgr.fafu.edu.cn/index.php) revealed co-expression of both FAD2 homeologous genes in different tissues including seed, leaf, root tip, root nodule, stem tip, inflorescence etc. (Sinha et al., 2020). FAD2 gene expression was significantly upregulated in peanut seeds, especially during advanced embryo development stages, specifically FAD2B had a relatively high expression in root nodules and seed tissues compared to FAD2A (Sinha et al., 2020).

In this study, several conserved cis-elements important for tissue specific responsiveness to hormonal and environmental cues were observed in the AhFAD2A and AhFAD2B upstream 5′UTRs (Supplementary File S1). Among them, the cis-regulatory RY element (CATGCATG) and 2S seed protein motifs (CAAACAC), which are implicated in seed specific gene expression. The initiator elements (INR) with sequences ‘TTCATTCT’ and ‘TTCATTTT’ responsible for basal transcription and light responsiveness were found. The cis-regulatory elements involved in stress response included Anaerobiosis Responsive Element (ARE), Wounding-Responsive Element (WUN motif), W box, G box, Anaeroconsensus among others were also identified. Additionally, the conserved elements for hormonal regulation included Abscisic acid Responsive Element (ABRE), Auxin Responsive Element (AuxRE), GA-responsive elements (GARE motif) etc. were predicted (Supplementary File S1). We chose to focus on the seed specific RY and 2S motifs for this study.

Complete single gRNA CRISPR/Cas9 constructs with a standard or extended gRNA scaffold were developed for the RY and 2S target sites then an Agrobacterium-mediated hairy root assay was employed to efficiently test the assembled constructs because traditional peanut transformation and regeneration methods are technically challenging and time consuming. Site-specific mutation frequencies and the type of DNA modifications were analyzed and recorded for each construct through PCR amplification of the target area followed by purification and Sanger sequencing.

For the single gRNA constructs, a total of 49 hairy roots for RY and 47 hairy roots for 2S were examined. Indels at the distal RY repeat motif ranged from +2 to −33 bp in AhFAD2A and +1 to −7 bp in AhFAD2B and the vast majority of edits were indels of less than 10 bp. The extended scaffold produced mutations in FAD2A at 20%, FAD2B at 10.3% and at both genes 9.68% of the time. The standard scaffold produced mutations in FAD2A at 37.5%, FAD2B at 50% and at both genes 22.22% of the time (Table 3). Most indels caused a disruption of the RY motif (Figure 3). Based on these figures, the standard gRNA scaffold performed better than the extended gRNA scaffold at RY.

The number of mutations detected at the proximal 2S motif were lower than that observed at the RY motif and the type of edits ranged from indels of −4 bp in FAD2A and +1 to −3 bp for FAD2B with most mutations causing disruption of the 2S motif (Figure 4). The extended scaffold produced mutations in FAD2A at 6.67%, FAD2B at 13.33% and in both genes at 4.35% of the time. The standard scaffold did not produce detectable mutations in FAD2A while 4.55% of mutations were at FAD2B (Table 3) Based on these figures, the extended scaffold performed better than the standard gRNA scaffold at 2S.

Dual gRNA editing constructs were generated that target both RY and 2S motifs using the validated targeting oligo sequences used in the single gRNA constructs. These constructs used the extended scaffold because they exhibited improved efficiency overall and were meant to test the efficiency of editing both targets simultaneously, targeting both genes simultaneously the possibility of generating large deletions between RY and 2S. To test for position effect in the gRNA order, two constructs were generated: one with a RY gRNA in the first position and 2S in the second position while the reverse order was used for the second construct.

Analysis of transformed hairy roots showed that overall editing efficiency at RY was greater than at 2S regardless of gRNA order; a result similar to that observed with the single gRNA study. This remained true for edits in both the FAD2A and FAD2B genes. Most indels at RY were less than 7 bp (80–90% in FAD2A and 50–75% in FAD2B) and a few (16–25%) deletions were greater than 40 bp (43 bp, 44 bp and 46 bp specifically) (Figure 3). Approximately 18–20% of the lines had RY edits in both FAD2A and FAD2B using either gRNA order (Table 3). Also noted was a single root containing a large deletion of 1,140 bp between the RY and 2S sites in FAD2B (Figure 3).

Using the dual editing constructs, it was noted that editing at the 2S motif was much less effective with only one edit of -9 bp in FAD2A when the RY gRNA was in the first position and no edits were observed when 2S was in the first position for FAD2A. The opposite was true for FAD2B where no edits were observed when the RY gRNA was in the first position, but a +1 edit and a −6 bp edit were observed when the 2S gRNA was in the first position (Figure 4). Additionally, the dual gRNA constructs produced no simultaneous 2S mutations in both FAD2A and FAD2B (Table 3).

To access the consequences of FAD2 gene editing on oil quality, Agrobacterium-mediated transformation through calyx tube injection was conducted. Single gRNA constructs with the extended scaffold targeting either RY or 2S were used to compare the mutation effect on oil quality at each motif. Approximate 650 calyx tubes were injected with the RY construct strain and 200 of calyx tubes were injected with the 2S construct strain. Following injection, a total of 120 potentially edited seeds were harvested. Among these, 26 seeds had an increased oleic acid content ranging from 55 to 70% when the RY motif was targeted and 7 seeds had an increased oleic acid content ranging from 55 to 60% when the 2S motif was targeted in compared to untransformed seeds with oleic acid concentration of 40–54% (Table 4). The changes of fatty acid in T0 seeds were listed in the Supplementary File S2. Although the apparent transformation efficiency was similar (27.66 RY vs 26.92% 2S) for both constructs, the resulting oleic acid content in the edited RY motif was higher than the edited 2S motif. It should be noted that the oleic acid content of two seeds for the RY motif was 66–70%, while the majority of presumed edits had a lower oleic acid content. Regardless, apparent mutations increased oleic acid concentration in these seed whether the RY or 2S motifs were targeted showing efficacy of this approach. However, amplifications of T₁ plant DNAs showed no mutations in the targets from either FAD2A or FAD2B, and all seeds harvested from T₁ plants derived from these T₀ edited seeds showed oleic acid content less than 55%. It might indicate that the CRISPR components were delivered into somatic cells rather than germ cells in T0 seeds using the calyx tube injection method.

Gene Editing has proven to be a powerful tool for generating gene knockouts for the study of gene function. Although less well studied, promoter editing has been successfully employed to engineer desirable traits such as disease resistance in rice (Oliva et al., 2019) and yield traits in tomato (Rodríguez-Leal et al., 2017). As demonstrated in this study and others, this approach has the potential to generate a wide spectrum of novel transcriptional alleles for genetic fine-tuning and optimized gene expression in crop plants.

The RY repeat motif is implicated in quantitative expression in seeds in diverse species (Baumlein et al., 1992; Lelievre et al., 1992; Dickinson et al., 1998) and is known to interact with B3 domain transcription factors involved in Abscisic acid (ABA) mediated transcription, embryogenesis and seed development (Reidt et al., 2000). Promoter studies by Ellerström et al. (1996) demonstrated that alteration of RY motifs in the Brassica napus nap A promoter resulted in changes in tissue specific gene expression. In addition, the 2S seed protein motif is conserved in the promoters of several seed storage protein genes and is implicated in higher activity of the nap A promoter (Stålberg et al., 1996). Deletion analysis specifically showed the 2S motif was important in regulation of seed protein accumulation while alterations to the promoter changed gene expression patterns (Stålberg et al., 1993). In this study, we investigated the feasibility and efficiency of simultaneously editing regulatory elements in two related peanut genes. Specifically, the RY and 2S motifs in the FAD2 genes were altered and an expected increase in oleic acid content of seed was detected suggesting reduced FAD2 gene expression as a consequence of targeted promoter alteration.

Initial testing of gene editing technology in peanut focused on determining the efficacy of CRISPR/Cas9 technology when single gRNAs were employed. Previously tested vector components were ligated together to create a basic vector platform that contained either a standard or an extended gRNA scaffold then targeting oligos for the RY or 2S motifs were added. Testing of constructs in a hairy root assay demonstrated the following: 1) editing efficiency at the distal RY motif was higher than at the proximal 2S motif, 2) the standard gRNA scaffold was more effective at RY than the extended gRNA scaffold, 3) the extended gRNA scaffold was more effective at 2S than the standard scaffold, 4) FAD2B was targeted more efficiently than FAD2A, and 5) the predominance of edits involving deletions less than 10 bp with a single gRNA target was consistent with earlier studies (Gisler et al., 2019). Taken together, these data demonstrate the efficacy of CRISPR/Cas9 mediated editing of peanut FAD2 promoter sequences and suggest that some positions within a promoter may be more accessible than others, use of different gRNA scaffolds may give some flexibility in target efficiency and there may be some difference in homeolog accessibility when targeting gene families.

To further explore the limits of promoter sequence editing, dual gRNA constructs, using the extended gRNA scaffold, were generated to target RY and 2S motifs simultaneously. Two versions of the vector were constructed with RY and 2S targeting oligos in the first and second positions respectively or in the reverse order. The results demonstrated differential editing rates at the target sites with RY being targeted at a higher rate than 2S. Again, FAD2A was targeted more frequently than FAD2B for the RY motif and FAD2B was targeted more frequently than FAD2A for the 2S motif. This could be related to the sequence context, local DNA methylation levels, or the chromatin configuration modulating accessibility as suggested by Liu et al., 2019. It should also be noted that this observation was independent of which targeting oligo was in the first position or second position.

Most indels created by the dual gRNA vectors were less than 12 bases while a few were larger. A few longer deletions (>40 bp) at a single site seems to suggest a plausible interaction effect of editing at closely linked sites possibly affecting the local chromatin context. This type of local opening up of chromatin can possibly modulate the timing and activity of the individual gRNAs. This seems analogous to the strategy involving coupling of proximal dead sgRNA (dsgRNA) to the functional sgRNA target, to open up the closed chromatin in rice cells and thereby enhancing editing efficiency (Liu et al., 2019). It should also be noted that one instance of a large deletion (1,140 bp) was detected, which demonstrates that it is possible to generate deletions between to gRNA targets albeit at a low frequency. Taken together this dual gRNA construct data demonstrates that multiple targets can be altered in a gene family in peanut, generating many different types of indels that may be useful for promoter modulation.

With the efficacy of single and dual gRNA constructs demonstrated in peanut, a test of practical application would demonstrate the usefulness of promoter modification. Fatty acids are essential components of plant cells and cell membranes, playing an important role in regulating different abiotic stress responses by modulating cell membrane properties (Iba, 2002; Kachroo and Kachroo, 2009). Thus, the desaturation of fatty acids and the number and position of double bonds in fatty acid chains influences the physical and physiological properties of membranes (Shanklin and Cahoon, 1998; Wu et al., 2009), which impacts proper plant growth and development (Traore and He, 2021). Therefore, general FAD2 loss of function mutations would be predicted to negatively impact cell membranes, leading to poor plant growth and development under abiotic stress conditions. This suggests that existing high oleic acid varieties resulting from mutations in the coding sequences of both FAD2A and FAD2B, while commercially useful, may be agriculturally detrimental (Lopez et al., 2000). CRISPR/Cas9 gene editing in peanut may afford a more nuance approach whereby the FAD2 gene activity in seed could be reduced while maintaining gene activity in other plant tissues.

A calyx tube transformation method was employed using the previously demonstrated RY and 2S single gRNA constructs to test if RY or 2S modification would impact oleic acid content in seed. A total of 120 seeds were generated through this method—94 for the RY construct and 26 for the 2S construct. The transformation efficiency for both constructs was relatively high at 27.66% for the RY construct and 26.92% for the 2S construct based on elevated oleic acid content relative to wild type. When tested by gas chromatograph method, the majority of the seeds had an oleic acid content ranging from 55% to 65%. While none of the 2S seeds tested above 61% oleic acid content, four of the RY edited seed were between 61 and 65% and two seeds were in the range of 66%–70% oleic acid. Taken together, this data suggests that editing of the RY or 2S motifs in the peanut FAD2 genes can impact oleic acid concentration in seed. Because the increased oleic acid content in T₀ seeds might be derived from somatic cell mutations, a further experiment will be needed to optimize the parameter for inoculation media, the suitable time and the location of the calyx tube for injection to improve the delivery of construct into reproductive cells with the calyx tube injection method.

Using gene editing to target cis-regulatory elements of the FAD2 gene promoters provides a promising approach to manipulate FAD2 gene expression in seeds and potentially minimize undesirable pleiotropic effects on other plant tissues while improving the fatty acid profile of seed. The use of CRISPR/Cas9 based promoter or enhancer editing may lead to the development of “cisgenic” plants with optimized FAD2 gene expression, which may serve as ideal breeding materials for trait introgression, without introducing potentially deleterious alleles and linkage drag, thereby accelerating the pace of cultivar development.