Generation of Herbicide-Resistant Soybean by Base Editing

Simple Summary Soybean is a prominent grain and oil crop in the world. The production of soybean is threatened by weed competition. Recently, CRISPR/Cas9-mediated base editing, which can cause irreversible nucleotide changes in target loci, has been widely used in basic plant science and breeding. However, until now, herbicide-resistant traits in soybean have not been created by the base editor. The objective of this study was to modify different AHAS homologous alleles by base editor to generate a novel AHAS-inhibiting herbicide-resistant soybean. Our results showed that the GmAHAS4 P180S mutants have apparent resistance to AHAS-inhibiting herbicides, which will be beneficial for weed control in soybean production. Abstract Weeds cause the largest yield loss in soybean production. The development of herbicide-resistant soybean germplasm is of great significance for weed control and yield improvement. In this study, we used the cytosine base editor (BE3) to develop novel herbicide-resistant soybean. We have successfully introduced base substitutions in GmAHAS3 and GmAHAS4 and obtained a heritable transgene-free soybean with homozygous P180S mutation in GmAHAS4. The GmAHAS4 P180S mutants have apparent resistance to chlorsulfuron, flucarbazone-sodium, and flumetsulam. In particular, the resistance to chlorsulfuron was more than 100 times that of with wild type TL-1. The agronomic performance of the GmAHAS4 P180S mutants showed no significant differences to TL-1 under natural growth conditions. In addition, we developed allele-specific PCR markers for the GmAHAS4 P180S mutants, which can easily discriminate homozygous, heterozygous mutants, and wild-type plants. This study demonstrates a feasible and effective way to generate herbicide-resistant soybean by using CRISPR/Cas9-mediated base editing.


Introduction
In fields, weeds not only compete with crops for light, growth space, and nutrients, but also spread pests and diseases and release toxic substances to reduce food production [1]; therefore, effective weed control is critical for world food security [2]. Currently, herbicides are extensively used to control weeds. However, improper use of herbicides can cause serious problems such as crop injury and the development of herbicide-resistant weeds [3]. Generating herbicide-resistant cultivars provides an effective solution [4]. On the one hand, herbicide-resistant varieties effectively reduced the risk of herbicide injury on crops; on the other hand, herbicide-resistant varieties with new mutations could introduce new herbicides to control recalcitrant weeds.
Soybean (Glycine max (L.) Merr.) is a prominent grain and oil crop in the world. Soybean is easily disturbed by weeds because the seeds are planted at a wide interval to form branches and to allow the canopy to expand fully in the later stage of growth. The late canopy closure makes soybean more prone to weeds than other crops [5]. Weeds cause the largest yield loss in soybean production. On a global basis, there is evidence that 37% of soybean yields is threatened by weed competition [6]. Therefore, creating herbicide resistant soybean germplasm is of great significance for weed control and yield improvement. However, the efficiency of using traditional breeding tools to create herbicide-resistant varieties is very low. Recently, a CRISPR/Cas9-mediated base editor which could efficiently and precisely perform C to T base editing in mammalian cells has been developed [7]. This tool was quickly applied to different plant species, generating several novel traits with significant agricultural values [8].
Acetohydroxy acid synthase (AHAS), a key enzyme for the biosynthesis of branchedchain amino acids including valine, leucine, and isoleucine, is the target of many herbicides with different chemical properties [9]. These types of herbicides, including imidazolinones (IMI), pyrimidinyl thiobenzoates (PTB), sulfonylaminocarbonyl triazolinones (SCT), sulfonylureas (SU), and triazolopyrimidines (TP), are widely used for weed control in crop fields, including soybean [10]. Thus, the creation of a herbicide-insensitive AHAS enzyme is of great value in agricultural application. Mutations in key amino acids of target enzymes can cause target site resistance, which is the most common mechanism leading to resistance to AHAS-inhibiting herbicides [11]. At present, in weed biotypes with field evolutionary resistance, at least 28 amino acid substitutions conferring resistance have been identified in eight conserved positions of AHAS enzymes [12]. Among them, the most common mutation site is the P197 codon (numbered according to the corresponding sequence of A. thaliana) [13]. Recently, the P197 site has been modified by the base editor to generate herbicide-resistant germplasm in many crops [4]. However, until now, herbicide-resistant traits have not been created by base editor in soybean.
In soybean, four homologs of AHAS have been identified, including GmAHAS1 (Glyma04g37270), GmAHAS2 (Glyma06g17790), GmAHAS3 (Glyma13g31470), and GmA-HAS4 (Glyma15g07860). C to T mutations were once successfully introduced to the P178 codon of GmAHAS1 via chemical mutagenesis [14], which conferred a moderate resistant level to sulfonylurea herbicides. However, it is unknown whether mutations at the P197 site (numbered according to the corresponding sequence of A. thaliana) of other homologous alleles will endow soybean resistance to herbicides. The objective of the present study was to modify different AHAS homologous alleles by base editing technology to create a novel herbicide-resistant soybean.

Vector Construction and Soybean Transformation
Close analysis revealed that the sequences of GmAHAS3 and GmAHAS4 in the P197 region are identical, and the sequence of GmAHAS2 is different from that of GmAHAS3 and GmAHAS4 with only one base (( Figure 1), so the P197 region of GmAHAS2 and GmAHAS3/4 were selected as the target sites for creating herbicide-resistant soybean by using a base editing system. These two selected target sequences (AGGTCCCCCGGCGCATGAT and AGGTCCCCCGCCGCATGAT) were cloned into pBSE901, respectively [15], in which sgRNA transcription was driven by the Arabidopsis U6 promoter and BE3 was driven by double 35S promoter. The bar gene, encoding phosphinothricin acetyltransferase (PAT) and conferring tolerance to phosphinothricin (PPT), was used as a selectable marker. These two binary vectors were then transformed into Glycine max (L.) cultivar Tianlong 1 (TL-1), as described previously [16].

Mutation and Transgene-Free Detection
For mutation detection, genomic DNA from the leaf tissue of T0 transgenic plants and their progenies was extracted using a modified CTAB method [17]. The target regions of GmAHAS1-4 were amplified by PCR using gene-specific primers and the products were subjected to Sanger sequencing.
To identify transgene-free plants, the gene-specific primer pairs U6-26p-F/U6-26t-R, Cas9-F/Cas9-R, and Bar-F/Bar-R were used to amplify the sgRNA, Cas9, and Bar genes, respectively. Plants-lacking all three genes concurrently, were considered to be free of transgenes.

Herbicide Tolerance Assay
Soybean seedlings were cultured in a greenhouse with 14 h light at 250 to 300 µmol photons/m 2 /s and 10 h dark at 25 °C with 60% humidity. Five representative herbicides from five AHAS inhibitors families, including bispyribac-sodium, imazapyr, chlorsulfuron, flucarbazone-sodium, and flumetsulam, were sprayed once at the V2 stage. The concentration gradient of each herbicide was 0, 1, 10, 100, and 1000 mg/L, respectively. Phenotypes and plant height of GmAHAS4 P180S mutants and wild-type plants were recorded 10 days after herbicide treatment. Each treatment was repeated at least three times.

Development of Allele-Specific PCR Markers
To distinguish GmAHAS4 P180S mutant from WT plants, two pairs of primers were designed. A common forward primer (GmAHAS4-P180-F) was GmAHAS4-specific and could differentiate between GmAHAS4 and other homologous genes. The two base editing sites were chosen to design at the 3′ end of the reverse primers. The wild type GmAHAS4 gene was amplified by GmAHAS4-P180-F/ GmAHAS4-180P-R, and the P180S mutant GmAHAS4 gene was amplified by GmAHAS4-P180-F/ GmAHAS4-180S-R.

Targeted Base Editing of GmAHAS Genes
A total of 8 T0 transgenic plants with GmAHAS2-targeting vector and 15 T0 trans-

Mutation and Transgene-Free Detection
For mutation detection, genomic DNA from the leaf tissue of T0 transgenic plants and their progenies was extracted using a modified CTAB method [17]. The target regions of GmAHAS1-4 were amplified by PCR using gene-specific primers and the products were subjected to Sanger sequencing.
To identify transgene-free plants, the gene-specific primer pairs U6-26p-F/U6-26t-R, Cas9-F/Cas9-R, and Bar-F/Bar-R were used to amplify the sgRNA, Cas9, and Bar genes, respectively. Plants-lacking all three genes concurrently, were considered to be free of transgenes.

Herbicide Tolerance Assay
Soybean seedlings were cultured in a greenhouse with 14 h light at 250 to 300 µmol photons/m 2 /s and 10 h dark at 25 • C with 60% humidity. Five representative herbicides from five AHAS inhibitors families, including bispyribac-sodium, imazapyr, chlorsulfuron, flucarbazone-sodium, and flumetsulam, were sprayed once at the V2 stage. The concentration gradient of each herbicide was 0, 1, 10, 100, and 1000 mg/L, respectively. Phenotypes and plant height of GmAHAS4 P180S mutants and wild-type plants were recorded 10 days after herbicide treatment. Each treatment was repeated at least three times.

Development of Allele-Specific PCR Markers
To distinguish GmAHAS4 P180S mutant from WT plants, two pairs of primers were designed. A common forward primer (GmAHAS4-P180-F) was GmAHAS4-specific and could differentiate between GmAHAS4 and other homologous genes. The two base editing sites were chosen to design at the 3 end of the reverse primers. The wild type GmAHAS4 gene was amplified by GmAHAS4-P180-F/GmAHAS4-180P-R, and the P180S mutant GmAHAS4 gene was amplified by GmAHAS4-P180-F/GmAHAS4-180S-R.

Targeted Base Editing of GmAHAS Genes
A total of 8 T0 transgenic plants with GmAHAS2-targeting vector and 15 T0 transgenic plants with GmAHAS3/4-targeting vector were obtained after screening by the PAT testing strips. Sanger sequencing results revealed that no gene editing occurred in 8 transgenic  Figure 2). Interestingly, most of these mutations are C to T, and in Line10, in addition to C to T, we also found C to A mutation. Furthermore, we found that these plants are all chimeric mutants, and the mutation levels at different sites in different plants are different. To assess the possible off-target events of the five base editing mutations, a total of nine putative off-target sites (including GmAHAS1 and GmAHAS2) were predicted by CRISPR-GE. After sequencing the PCR products of the potential offtarget sites, we did not detect any off-target events in these gene-edited plants ( Figure S1).  Figure 2). Interestingly, most of these mutations are C to T, and in Line10, in addition to C to T, we also found C to A mutation. Furthermore, we found that these plants are all chimeric mutants, and the mutation levels at different sites in different plants are different. To assess the possible off-target events of the five base editing mutations, a total of nine putative off-target sites (including GmAHAS1 and GmAHAS2) were predicted by CRISPR-GE. After sequencing the PCR products of the potential off-target sites, we did not detect any off-target events in these gene-edited plants ( Figure S1).

Identification of GmAHAS4 P180S Mutants
To test whether the base-edited mutations could be passed to the next generation efficiently, and to obtain heritable homozygous mutants, T1 progeny of five T0-edited plants were analyzed. Based on the sequencing results, only the mutation of Line 8 was stably inherited to the T1 generation, and a total of 23 wild-type plants (P180P/ P180P), 18 heterozygous mutants (P180P/P180S), and 14 homozygous mutants (P180S/P180S) were obtained ( Figure 3A). PCR amplification of Cas9, sgRNA, and Bar genes showed that a total of 15 T1 mutants lacked the exogenous T-DNA due to genetic segregation ( Figure 3B). To examine the stable transmission of P180S allele, T1 homozygous P180S mutants without

Identification of GmAHAS4 P180S Mutants
To test whether the base-edited mutations could be passed to the next generation efficiently, and to obtain heritable homozygous mutants, T1 progeny of five T0-edited plants were analyzed. Based on the sequencing results, only the mutation of Line 8 was stably inherited to the T1 generation, and a total of 23 wild-type plants (P180P/P180P), 18 heterozygous mutants (P180P/P180S), and 14 homozygous mutants (P180S/P180S) were obtained ( Figure 3A). PCR amplification of Cas9, sgRNA, and Bar genes showed that a total of 15 T1 mutants lacked the exogenous T-DNA due to genetic segregation ( Figure 3B). To examine the stable transmission of P180S allele, T1 homozygous P180S Biology 2023, 12, 741 5 of 11 mutants without transgenic components were cultured to T3 generation. Sequencing results showed that these T3 plants all contained homozygous P180S mutations ( Figure S2), and PCR amplification results showed that they did not contain transgenic components ( Figure S3). Meanwhile, no new editing events were detected in the target region of GmAHAS1-3 and seven putative off-target sites of the T3 P180S mutants ( Figure S4).
Biology 2023, 12, x 5 of 1 transgenic components were cultured to T3 generation. Sequencing results showed tha these T3 plants all contained homozygous P180S mutations ( Figure S2), and PCR amplifi cation results showed that they did not contain transgenic components ( Figure S3). Mean while, no new editing events were detected in the target region of GmAHAS1-3 and seve putative off-target sites of the T3 P180S mutants ( Figure S4).

Herbicide Tolerance of GmAHAS4 P180S Mutants
To determine whether the base editing mutants could endow soybean with herbicid resistance, the wild-type controls (TL-1), together with the transgene-free T3 homozygou GmAHAS4 P180S plants, were treated with different concentrations of bispyribac-so dium, imazapyr, chlorsulfuron, flucarbazone-sodium, and flumetsulam at the V2 stage As shown in Figure 4, there was no phenotypic difference between the GmAHAS4 P180 mutants and the wild-type controls (TL-1) without herbicide treatment. Compared wit the wild-type TL-1, the GmAHAS4 P180S mutants showed slight resistance to bispyribac sodium and no apparent resistance to imazapyr. As for chlorsulfuron, the wild-type TLcannot grow at a concentration of 1 mg/L, while the GmAHAS4 P180S mutant can sti maintain growth at 100 mg/L. As for flucarbazone-sodium, the wild-type plants die quickly at 100 mg/L, while the mutant survived at 1000 mg/L, although its growth wa severely inhibited. As for flumetsulam, both TL-1 and mutants could maintain growth a the concentration of 100 mg/L. When the concentration increased to 1000 mg/L, the wild

Herbicide Tolerance of GmAHAS4 P180S Mutants
To determine whether the base editing mutants could endow soybean with herbicide resistance, the wild-type controls (TL-1), together with the transgene-free T3 homozygous GmAHAS4 P180S plants, were treated with different concentrations of bispyribac-sodium, imazapyr, chlorsulfuron, flucarbazone-sodium, and flumetsulam at the V2 stage. As shown in Figure 4, there was no phenotypic difference between the GmAHAS4 P180S mutants and the wild-type controls (TL-1) without herbicide treatment. Compared with the wild-type TL-1, the GmAHAS4 P180S mutants showed slight resistance to bispyribac-sodium and no apparent resistance to imazapyr. As for chlorsulfuron, the wild-type TL-1 cannot grow at a concentration of 1 mg/L, while the GmAHAS4 P180S mutant can still maintain growth at 100 mg/L. As for flucarbazone-sodium, the wild-type plants died quickly at 100 mg/L, while the mutant survived at 1000 mg/L, although its growth was severely inhibited. As

Agronomic Traits of GmAHAS4 P180S Mutants
To evaluate the effect of the P180S allele on agronomic traits, the transgene-free T3 homozygous GmAHAS4 P180S soybean were grown in the greenhouse under natural growth conditions in Tianjin, China. As shown in Figure 5, the agronomic traits, including

Agronomic Traits of GmAHAS4 P180S Mutants
To evaluate the effect of the P180S allele on agronomic traits, the transgene-free T3 homozygous GmAHAS4 P180S soybean were grown in the greenhouse under natural growth conditions in Tianjin, China. As shown in Figure 5, the agronomic traits, including the morphology of plant and seed, seed number per plant, seed yield per plant, and hundred-grain weight, showed no significant differences between TL-1 and the P180S plants.
Biology 2023, 12, x 7 the morphology of plant and seed, seed number per plant, seed yield per plant, and h dred-grain weight, showed no significant differences between TL-1 and the P180S pla Furthermore, we designed allele-specific PCR markers to distinguish GmAH P180S mutants from the wild-type TL-1 plants. The upstream primer was designed cording to the specific sequence of GmAHAS4, and two base editing sites were desig to the 3' end of the downstream primer ( Figure 6A). The results showed that these prim could easily discriminate homozygous mutants, heterozygous mutants and wildplants ( Figure 6B). Furthermore, we designed allele-specific PCR markers to distinguish GmAHAS4 P180S mutants from the wild-type TL-1 plants. The upstream primer was designed according to the specific sequence of GmAHAS4, and two base editing sites were designed to the 3' end of the downstream primer ( Figure 6A). The results showed that these primers could easily discriminate homozygous mutants, heterozygous mutants and wild-type plants ( Figure 6B).

Discussion
Due to its simplicity, specificity, efficiency, and versatility, the CRISPR/Cas system has quickly become the preferred tool for plant genome engineering [19][20][21]. The application of the CRISPR/Cas system in plant genome editing mainly includes gene knockout, gene targeted insertion or replacement, and gene expression regulation [22]. Recently, as a new gene-editing strategy, CRISPR/Cas-mediated base editing, which can cause irreversible nucleotide changes in target loci without double stranded DNA cleavage or any donor template, has been widely used in basic plant science and breeding [23,24]. For example, C to G substitution with frequency of 1.6% to 3.9% and C to T substitution with frequency of 1.4% to 11.5% were detected in rice by using base editing. In addition, C to T substitution in SLR1 (S97L) resulted in a distinct semi dwarf phenotype [25]. To facilitate weed control, herbicide-resistant germplasm has been generated via the base editor in many crops, including rice [26], wheat [2], canola [27], tomato [26], and watermelon [15]. In the present study, we used the cytosine base editor BE3 to generate a novel AHASinhibiting herbicide-resistant soybean mutant. Our results showed that the GmAHAS4 P180S mutants have apparent resistance to chlorsulfuron, flucarbazone-sodium, and flumetsulam.
Since different types of AHAS-inhibiting herbicides bind to various sites of the herbicide binding domain, particular point mutation of the AHAS gene could confer tolerance to a specific range of AHAS-inhibiting herbicides [28]. Meanwhile, the herbicide resistance conferred by AHAS mutation has a dose effect [29]. In rice, the four different mutations in the P197 codon-P197A, P197S, P197Y, and P197F-exhibited different resistance patterns

Discussion
Due to its simplicity, specificity, efficiency, and versatility, the CRISPR/Cas system has quickly become the preferred tool for plant genome engineering [19][20][21]. The application of the CRISPR/Cas system in plant genome editing mainly includes gene knockout, gene targeted insertion or replacement, and gene expression regulation [22]. Recently, as a new gene-editing strategy, CRISPR/Cas-mediated base editing, which can cause irreversible nucleotide changes in target loci without double stranded DNA cleavage or any donor template, has been widely used in basic plant science and breeding [23,24]. For example, C to G substitution with frequency of 1.6% to 3.9% and C to T substitution with frequency of 1.4% to 11.5% were detected in rice by using base editing. In addition, C to T substitution in SLR1 (S97L) resulted in a distinct semi dwarf phenotype [25]. To facilitate weed control, herbicide-resistant germplasm has been generated via the base editor in many crops, including rice [26], wheat [2], canola [27], tomato [26], and watermelon [15]. In the present study, we used the cytosine base editor BE3 to generate a novel AHAS-inhibiting herbicideresistant soybean mutant. Our results showed that the GmAHAS4 P180S mutants have apparent resistance to chlorsulfuron, flucarbazone-sodium, and flumetsulam.
Since different types of AHAS-inhibiting herbicides bind to various sites of the herbicide binding domain, particular point mutation of the AHAS gene could confer tolerance to a specific range of AHAS-inhibiting herbicides [28]. Meanwhile, the herbicide resistance conferred by AHAS mutation has a dose effect [29]. In rice, the four different mutations in the P197 codon-P197A, P197S, P197Y, and P197F-exhibited different resistance patterns towards different AHAS-inhibiting herbicides. P197S and P197A showed themselves to be more sensitive to bispyribac than P197Y and P197F [28]. In wheat, the P197S conferred higher resistance to nicosulfuron than P197A [2]. In this study, the GmAHAS4 P180S mutants showed no apparent resistance to imazapyr, slight resistance to bispyribac-sodium, and apparent resistance to flumetsulam. The resistance to flucarbazone-sodium was more than 10 times and the resistance to chlorsulfuron was more than 100 times.
Gene editing efficiency and inheritance of genome modifications are two important considerations when using CRISPR/Cas9-mediated gene editing technology for basic research and molecular breeding [16]. Michno et al. reported that the integration and inheritance of CRISPR/Cas9 soybean lines include three patterns, nanemly, no transmission of mutations and transgenes, inheritance of transgenes located within the target sites but no mutations, and transmission of mutations but segregation of transgenes [30]. In CRISPR/Cas9-induced mutations, the loss of mutations and transgenes in subsequent generations seems to be common. Shi et al. reported that neither transgene nor mutations of two T0 generation soybean PDS gene editing lines were inherited by T1 progenies [17]. They found that the indel frequencies of the T0 generation were low in both strains, so the T0 plants were likely chimeric, and the mutation only existed in somatic cells, so it was not inherited into the T1 generation. In the present study, we did not obtain GmAHAS2 gene editing mutants in T0 transgenic plants. Meanwhile, although five GmAHAS3/4 gene editing mutants were obtained in the T0 transgenic plants, only one mutation was stably inherited to the next generation. The possible reasons for these phenomena mainly include the following aspects: (i) the genetic transformation efficiency of soybean was low [31], and the obtained transgenic plants were too few; (ii) the vector used in this study has low editing efficiency in soybean [32]; (iii) the T0 transgenic regenerated plants obtained by tissue culture were chimeric, and the mutations were only present in somatic cells [17]; thus, they cannot be inherited to the next generation. Therefore, systematic optimization of the soybean transgenic regeneration system and the gene-editing system to reduce the generation of chimeras and false-positive plants is of great significance for improving the efficiency of heritable gene editing mutation in soybean.

Conclusions
In summary, we have successfully introduced base substitutions in GmAHAS3 and GmAHAS4 by CRISPR/Cas-mediated base-editing and obtained a transgene-free herbicideresistant soybean with homozygous P180S mutation in GmAHAS4. This P180S mutation has no adverse effects on agronomic traits. Moreover, we successfully designed allelespecific markers to identify P180S mutations of GmAHAS4, which will help breed herbicideresistant soybean. In the future, multi-point mutation of the AHAS gene and simultaneous modification of different homologs will be beneficial to obtain soybean with multiple herbicide resistance, thus contributing to weed management in soybean production.