Location: Home >> Detail
Crop Breed Genet Genom. 2026;8(2):e260015. https://doi.org/10.20900/cbgg20260015
,
Maud Fagny 2 ,
Rafael E. Venado 3,4 ,
Jia Shen 5,* ,
1
2
3
4
5
*
The study of genetics seeks to understand the molecular relationships that ensure the proper patterning of gene expression necessary for plant development and its responses to the environment that determine crop performance. Improvements in molecular genetics, biotechnology, and a deeper understanding of gene expression have given breeders access to major sources of phenotypic novelty in plants and allowed them to accelerate genetic gains by breeding locally adapted crops. Advances in genetics, and more recently the edition of epigenetic marks, cis-regulatory elements and gene editing have gained significant importance in crop breeding. Single-cell level studies have helped to elucidate cell heterogeneity and intercellular regulation in key tissues, enabling a more precise comprehension of the effects of plant developmental processes on phenotypes. The aim of this review is to highlight how genetic knowledge has fostered progress in crop improvement and selection. After a summary of the historical use of genetics in breeding strategies over the past 100 years, we discuss how the latest technological developments in genetics can be leveraged in complex quantitative trait improvement. Recent progress in single-cell analyses allow us to identify the subtle relationship between the genomic blueprint and cellular diversity, connecting this cellular genetic heterogeneity of a plant to specific phenotypes, that would be otherwise overlooked when using bulk cell analyses. High-throughput access to numerous epigenetic marks has also opened up new opportunities and challenges for crop genetic improvement. This progress has led to a better understanding of the role of cis-regulatory elements in determining some aspects of crop performance. Here we also discuss whether it is possible to leverage reliable variations in these regions in crop breeding, using targeted selection to manipulate traits in a wide range of crops. Finally, we outline advances in gene editing tools in crop improvement, and discuss how these could be implemented to accelerate breeding for crops adapted to new environments.
Plant breeding has profoundly evolved throughout the history of crop improvement and has incorporated timely co-evolving genetic progress knowledge, molecular biology and innovative techniques (Figure 1). Before Mendel’s law of heredity [1] and its rediscovery in the 1900s [2–4], breeding relied on classical phenotypic selection and pedigree methods. For instance, Vilmorin Co. in France introduced the pedigree method in the 1800s [5], where a few plants were selected to provide seeds to sow in rows for the next generation based on progeny testing principles. The development of genetics improved our understanding of chromosome and loci re-arrangement with a substantial impact on breeding methodologies.
Pure line selection theory [6], a system where the progeny of a single self-fertilized homozygous plant is obtained, helped demonstrate that phenotypic variation is present within natural accession, intrinsically separating out between-lines phenotypic variance by reducing within-line variation. One successful example of a variety developed using pure line selection was the sorghum line resistant to Dwarf Yellow Milo caused by root ascomycete fungus Periconia circinate [7] (pages 112–113). In 1930, F. A Wagner, at the Kansas Agriculture Research Station (USA), realized that two apparently phenotypically identical plants withstood the devastating Dwarf Yellow Milo disease differently and led to progeny-free offspring of the disease progeny. However, problems with the resistance mechanism were identified [8,9], and, despite the efforts of molecular geneticists [10,11], the resistance genes have not yet been characterized.
Bulk-population was originally proposed by Swedish geneticist Nilsson-Ehle [12]. This method takes advantage of natural selection: a large population of F2 segregants are sowed in a large plot, from which seeds are collected and sowed in subsequent years. Some artificial selection normally occurs by suppressing off-types plants. With his research on oats, Nilsson-Ehle shed light on a quantitative genetics theory at the same time as American geneticist East’s discoveries on maize [13]. Single seed descent (SSD) [14] is a method were segregants from a hybridization event are rapidly fixed by advancing one randomly selected seed per plant at the early stages through subsequent generations, as a modified pedigree method [15]). SSD is used to develop recombinant inbred lines (RILs), a breeding resource that takes advantage of several recombination events resulting from self-fertilization. Their analysis in applied quantitative genetic studies [16,17] has led to the identification of quantitative trait loci (QTLs) used for improving crops and developing resistant varieties. RILs are a powerful tool for genetic variation mapping. Together with next-generation sequencing (NGS), they have enabled rapid identification of candidate genes, for example, in maize for traits associated with the mimic lesion [18], in foxtail millet for morphological traits [19], in cucumber using genotyping-by-sequencing for downy mildew resistance [20] or as a RILs resource for diploid potato [21].
Figure 1. Diagram representing advances in genetics, molecular biology, epigenetics, single-cell level analysis, and gene editing, to better understand the layers of gene regulation for phenotypic expression as a result of a plant performance for variety development. Geneticists have elucidated and optimized traditional breeding. Similarly, breeding methods have provided a means to create population structures helping practical and theoretical quantitative genetic advances and gene mapping. At the same time, a breeding strategy is a breeder’s decision based on combined knowledge of all these genetic layers of information that lead to a complex regulatory interaction observed in a phenotype expressed within an environment. P = phenotype, G = genotype, E = environment; G × E = genotype by environment.
Backcross breeding [22,23] is another example of this co-evolution where plant breeders increase the genetic control of the populations in their breeding programs. In this method, a donor parent, generally with a specific trait (allele) to transfer, is backcrossed several times to an adapted variety to be improved, a process eventually ending with a self-pollination event to obtain a truly homozygous plant. Successful examples of backcross breeding include the US spring wheat cultivars, ‘Linkert’, resistant to the stem rust pathogen, Puccinia graminis f. sp. Tritici. Traditional backcross breeding has also benefited from incorporating the application of molecular biotechnologies (DNA markers) to practical breeding, which is known as marker assisted backcross (MABC). The use of MABC led to a five percent higher recurrent parent genome recovery in a BC2 generation compared to conventional backcross [24]. Continuous advances in understanding the genetic mechanisms underlying desirable traits, together with the development of molecular biology tools, have enabled the use of molecular markers in linkage disequilibrium with specific traits for marker-assisted selection (MAS). This approach has provided a major advantage in classical breeding because it is more cost-effective than conventional phenotypic selection. MAS has successfully improved crop breeding. In wheat, it has enabled the identification and use of markers associated with physiological traits [25], drought tolerance [25], and biotic resistance such as to Fusarium Head Blight [26]. It has also facilitated the simultaneous pyramiding of disease resistance and quality traits (e.g., glutenin) [27], the introgression of clubroot resistances in Brassica napus using Kompetitive Allele Specific PCR (KASP) markers [28], and the improvement of yield-related traits [29]. However, further genetics studies were needed to identify and characterize these genes (e.g., [30]) to better understand their defense mechanisms for optimal implementation in commercial breeding programs.
Hybrid breeding methods were initially proposed to breed cross-pollinated crops by combining two pure lines of maize [31], and later [32] suggested the use of double crosses to obtain a hybrid maize product. Hybrids attain high overall performance in the produced lines evaluated in a progeny after a cross with a set of testers. These crosses provide information on both specific and general combining abilities [33], reflecting the benefits of heterosis as a result of the response due to dominance effects [34].
Recurrent selection [35,36] can be used to improve inbred lines of each heterotic pool, or to develop hybrid varieties. The basis of this system relies on the self-pollination of an initial set of heterozygous lines that are evaluated for a desirable trait; plants with lower performance are discarded and the superior individual plants are sowed from the selfed seed. Then, all possible intercrosses are made, which serve as a source of the additional ‘recurrent’ cycles of selection. Genetic discoveries and advances in genomic capabilities have brought extraordinary improvement to hybrid development efficacy with the use of methods including cytoplasm male sterility [37,38], double haploid line development (e.g., in wheat [39,40] or maize [41]), or hybrid prediction performance of potential crosses (e.g., in rice [42,43] or Brassica napus [44]). These methods have been widely used to improve crops in the seed industry.
Bulk Segregant Analysis (BSA) is an effective method for identifying molecular markers associated with a phenotype of interest, and, compared with conventional mapping, is more cost-effective as it does not require genotyping and phenotyping of a complete mapping population [45]. In BSA two extreme phenotypes are characterized, and DNA bulks (groups) are pooled from individuals with similar phenotypes. These two groups can be derived from F2, F2:3 family populations, backcrossing (BC) populations, RILs, double haploids (DH) or multiparental populations [46]. With the advent of Next Generation Sequencing (NGS) technologies, BSA technology coupled with whole-genome sequencing (WGS) has been proven a powerful approach. By sequencing the two extreme phenotypes, BSA can identify genes located on different chromosomal regions associated with complex traits, for example, the control of amylose content in rice [47]. Advances in RNA-seq applied to the BSA breeding method, known as BSA-RNAseq, provides precise information on gene transcript abundance, offering stronger evidence of a gene’s causative role when its expression correlates with the phenotype of interest under specific conditions. Bulked segregant RNA-seq (BSA-RNA seq) has been successfully applied in different crops: e.g., to validate a stem resistance locus in a wild relative of wheat, Aegilops umbellulata [48], to identify a single variant (Pm5) conferring powdery mildew resistance in wheat [49], to study stay-green in maize [50] and to map Chilli veinal mottle virus resistance 4 (cvr4) in pepper [51]. This research has major implications for understanding crop adaptation and crop improvement.
Variations in chromosome number and structure from the basic crop genomes of a normal plant have been used to a lesser extent in cultivar development and have often been overlooked. Common types of aneuploidies include the addition or deletion of entire chromosomes or chromosome segments from the basic somatic number (2n) of a species. The disruption of chromosome pairing and segregation caused by the addition or deletion of a chromosome has been intentionally implemented for breeding purposes [52]. Examples include: hybrid seed production exploiting tertiary trisomic individuals in barley that contain the nuclear restorer locus male sterile (MS) in linkage with the breakpoint at the chromosome exchange position [53]; using monosomic lines in chromosome mapping, e.g., in Allium genetic studies [54,55]; or chromosome substitution, e.g., the transfer of rust resistance from Aegilops umbellulata to wheat [56], or the transfer of powdery mildew resistance into common wheat (Triticum aestivum L.) from its Aegilops geniculata relative [57]. Furthermore, another common type of aneuploidy is caused by the alteration of the chromosome structure known as translocation, where pieces of DNA segments move places within or between chromosomes. Translocations have also been suggested for inbred lines and hybrid seed production. Inbred lines can be obtained by crossing parents carrying different translocation events within a stock of multiple translocation lines [58,59]. For instance, in maize, hybrid seed production has exploited male sterility arising from heterozygous individuals of duplicate-deficient (Dp-Df) chromosomes [60], while a similar strategy was tested in hybrid wheat using the 4E-ms system [61]. Overall, chromosomal rearrangements can drive gene expression differences, leading to manifest phenotypic and metabolic differences [62], and are thus a source of genetic variation that can be exploited for crop breeding.
Advances in genetics and in particular, whole-genome sequencing (WGS) and the genome-wide mapping of genetic variations have helped improve breeding strategies and the development of powerful methods to predict yield/crop performance in the past 20 years. However, it has also shown its limits, and brought attention to the fact that how the genome is expressed is also an important component of crop performance (Figure 1). The cutting-edge technologies discussed in this article have given researchers access to gene expression patterns at single-cell level (see below) to partially decipher the epigenetic code governing gene regulation expression. Moreover, the integration of several data types has revealed complex regulatory networks underlying traits and highlighted the role of cis-regulatory elements. Finally, rapidly progressing editing techniques will allow breeders to integrate, in already elite germplasm, some advantageous (epi-)mutations, not only in genes but also in regulatory elements, and thus accelerate the timeframe for creating new, locally adapted genotypes (Figure 1).
Most phenotypic differences in heritable traits are considered to be associated with DNA structure variants that differ between individuals. Trait variations can, more than is generally believed, be the result of gene expression consequences of epi-genetic regulation that have long-term effects in terms of mitotic inheritance and meiotic inheritance (between generations) [63,64]. Thus, in addition to inheriting genetic information, cells receive information that is not encoded in the DNA’s nucleotide sequence. A key element is DNA packing in the nucleus and epigenetic marks intrinsically arranged in a hierarchical three-dimensional conformation [65]. Epigenetic marks facilitate interactions between proximal and distal DNA sequences [66], as well as between RNA and DNA, leading to transcriptional and translational regulations of gene expressions [67,68]. These epigenetic marks, which stably modify gene expression and subsequently contribute to phenotypic plasticity, generally include: (1) DNA methylation; (2) histone modifications (post-translational modification of histone proteins) and remodeling of chromatin; and (3) RNA processing and non-coding RNAs (ncRNAs).
DNA methylation entails the addition of a methyl group at the 5’carbon of cytosine. In plants, methylation has a tendency to concentrate at contexts followed by guanines forming symmetric CG and CHG, as well as in asymmetric CHH (where H denotes A, C, or T) contexts. CG dinucleotides in plants are uniformly distributed and not depleted genome-wide compared to mammals. This is because of a distinct DNA methylation architecture with additional 5mC contexts (CHG, CHH) that reduces mutational pressure across many contexts other than CG, in addition to active demethylation via the ROS1 and Demeter pathway [69]. DNA methylation is generally associated with repressive marks that mediate tissue-specific gene silencing. However, genome-wide DNA methylation maps have revealed unexpected local dynamics, particularly at regulatory regions, where active demethylation occurs [70]). Methylation can alter transcriptional processes by modifying CpG contexts of gene promoters, interfering with transcription factors and other DNA-binding proteins, and ultimately inhibiting gene expression by recognizing methyl-CpG binding protein (MBD) [71,72]. Methylation marks can have strong phenotypic effects and have influenced different crop domestication events, e.g., maize, cotton or lettuce [73–75]. A significant correlation has also been found between the methylated state and the geographic origins of a wheat collection [76]. Methylation also affects genes related to yield and other agronomically important traits. In maize, differentially methylated regions (DMR) in the ZMET2 (Zea methyltransferase2) gene affects yield and the number of husk layers and may have been a target of selection [77,78]. In rice, demethylated patterns of the DWARF1 (D1) gene promote the transition from dwarf to normal stature [63], while in wheat, overexpression of the histone acetyltransferase TaHAG1 improves wheat tolerance to heat [79]. Han et al. (2021) [80] showed that hyper DNA methylation of Teosinte branched1/cycloidea/proliferating cell factor (TCP) were linked to inbreeding depression in maize. DNA methylation can also occur in gene bodies, epi-marks where gene exons are methylated, termed as Gene Body Methylation (GBM). GBM primarily manifests as CG methylation as opposed to CHH or CHG contexts [81]. GBM occurs in evolutionary conserved and ubiquitously expressed genes, however, it is causal relationship is still unresolved.
Histone modifications are chemical modifications on N-terminal tails of histones that induce post-translational gene regulation by determining chromatin configuration, either promoting an open (active) or closed (repressive) configuration [82]. The main role of histone modifications is to provide a memory of transcriptional activity [83] as well as to stabilize/reinforce transcriptional state [84]. These modification sites can be acetylated (Hac), methylated (Hme), phosphorylated, ubiquitinated, glycosylated, and sumoylated [85]. Histone modifications are permissive/suppressive to a particular transcriptional outcome, and the various combinations of these marks will form what is often referred as the ‘histone code’. For instance, Hac promotes transcription by neutralizing the positive charge of histones, weakening the binding interaction with negatively charged DNA [86]. Hac regulating dynamics involve histone acetyltransferases (HATs) and histone deacetylases (HDACs), enzymes that respectively drive the expression and inhibition of a gene [86,87]. On the other hand, Hme occur primarily at the lysine (K) and arginine (R) residues on the N-terminal tails of histones H3 and H4 [88]; they are laid down by histone methyltransferases (HMTs) and are removed by histone demethylases (HDM). Histone modifications play key roles in shaping agronomic phenotypes and crop performance. For example, acetylation of lysine residues on histone H3 and H4 suggest that HATs were involved in drought stress responses in rice [89], wheat [90], and several other crops and vegetables [91,92]. Histones are also associated with complex traits such as heterosis, which is a cornerstone of hybrid breeding (Section “Genetics as a tool for improving breeding strategies”). In rice, inter-subspecific hybrids displayed allele-specific differences in histones H3K4me3 and H3K27me3 [93]. Similar findings in Arabidopsis identified epigenetic quantitative trait loci (QTLs) explaining up to 90% of heritable variation [94]. In addition, stress-induced changes of histone modifications have been shown to enhance plant adaptation, and could thus be targeted by breeding programs [95].
Finally, RNA processing and interplay with non-coding RNAs (ncRNAs) can shape the epigenetic landscape of genomes impacting genome integrity, and gene expression. For instance, modulation of RNA m6A methylation has improved crop yield in rice and potato [96], and has been associated with biotic and abiotic stresses [97,98]. RNA-processing epigenetic modifications contribute to the post-transcriptional regulation of gene expression and represent a promising strategy for crop improvement.
Advances in genetics driven by the use of NGS technologies, allowing sequencing of multiple plant genomes and transcriptomes, have led to impactful plant biology progress applied to agriculture. RNAseq, in particular, has become a powerful tool for investigating transcriptional variations across plant tissues and conditions, such as biotic and abiotic stresses [99,100]. However, RNAseq faces limitations, as traditional bulk methods often mask cell-specific responses due to tissue heterogeneity and developmental gradients. To overcome this limitation, single-cell RNA sequencing (scRNAseq, from protoplasts) and single-nucleus RNA sequencing (snRNAseq, from isolated nuclei) enable precise transcriptomic profiling of thousands of individual cells. These techniques reveal subtle biological variations often masked by bulk RNAseq methods [101] and have a bright future as complementary breeding methods.
Originally applied to mammalian systems, scRNAseq and snRNAseq have recently been adapted for plants. However, plant cells pose unique challenges due to their rigid cell walls, which complicate protoplast and nuclei isolation procedures from dense tissues. The first successful applications of single-cell sequencing in plants emerged in 2019 with Arabidopsis thaliana [102–106], followed by crops like maize (Zea mays) [107–109] and rice (Oryza sativa) [110,111]. Recent protocol improvements now facilitate studies in non-model plants, including orphan crops. In practice, isolated nuclei or protoplasts are typically processed using 10X Genomics droplet-based systems, although alternative platforms from other providers are also available [112]. After library preparation, the samples undergo sequencing, and their data are visualized using techniques like Uniform Manifold Approximation and Projection (UMAP) and t-distributed stochastic neighbor embedding (t-SNE). These methods cluster cells or nuclei with similar transcriptomic profiles. Cluster annotations, based on known marker genes, are then used to construct gene regulatory networks (GRNs) by cell type. As with other RNA-seq methods, identifying candidate genes typically requires additional validation through genetic experiments, such as knocking out or knocking down the candidate genes or their regulatory factors (Figure 2).
The applications of single-cell studies in plant breeding are expanding, notably in yield improvement and tissue-specific responses to environmental stresses. Cereals like maize and rice have been focal points for single-cell research, particularly in roots and leaves, where studies have identified key genes involved in tissue development [107–109,113]. For example, Li et al. (2022) [113] used scRNA-seq to characterize maize root cell populations responding to nitrate availability and identified cell-type-specific regulators associated with nitrogen uptake efficiency. Because nitrogen-use efficiency is a highly quantitative trait traditionally evaluated at the whole-plant level, single-cell approaches may improve the precision of breeding strategies targeting nutrient acquisition and resource allocation. Similarly, Xu et al. (2025) [114] discovered stem cell regulators associated with heritable yield traits such as cob diameter and ear length in maize. These findings illustrate how cellular-resolution analyses can uncover developmental programs contributing to complex agronomic phenotypes. Other studies have revealed regulators that could enhance nitrogen uptake efficiency by targeting root responses to nitrate [113] or for physiological traits such as fruit senescence [115,116].
Single-cell studies also extend to plant-biotic interactions, including beneficial relationships like root nodules and arbuscular mycorrhizal symbioses. Research in model legumes, such as Medicago truncatula [117–120] and Lotus japonicus [121], and crops like soybean [122,123] have identified regulators that could enable cereals to form symbiotic structures or improve their capacity for beneficial microbial associations [124]. Conversely, single-cell techniques are also applied to study negative interactions. For instance, Yue et al. (2024) [125] used scRNAseq to characterize tomato leaf responses to tomato chlorosis virus infection and identified ERF4 as a master regulator involved in chlorosis development. Functional validation demonstrated that disruption of ERF4 mitigated chlorosis symptoms, highlighting the potential of single-cell-guided gene editing approaches for crop improvement.
Figure 2. Single-cell omics in breeding. Workflow for asc/snRNAseq experiment in crop research. Plant tissue from economically important or orphan crops is used to isolate protoplasts or nuclei. mRNA transcripts from individual cells are captured using a microfluidic-based system, which encapsulate single cells with unique barcodes for high- throughput sequencing e.g., 10× Genomics. Downstream analysis identifies cell clusters, annotates them using known marker genes, and experimentally validates candidate genes. These genes can be utilized to enhance agronomically important traits through conventional breeding or precise tools like CRISPR. This approach can uncover novel genes involved in biological processes or abiotic and biotic interactions, advancing applications in crop improvement.
Orphan crops have also benefited from single-cell insights. Zhang et al. (2023) [126] profiled Manihot esculenta (Cassava) leaves, uncovering antioxidant compounds like flavonoids and carotenoids in mesophyll cells, while Li et al. (2024) explored metabolic regulation in Hylocereus undatus [115]. Other studies have highlighted single-cell applications in stress responses in Brassica rapa (Chinese cabbage) [127] and medicinal compound biosynthesis in Catharanthus roseus [128].
Beyond transcript abundance, single-cell technologies are increasingly integrated with epigenomic and cis-regulatory analyses. These approaches now allow the characterization of chromatin accessibility, transcription factor binding, and regulatory landscapes at cellular resolution. For example, Marand et al. (2021) generated a single-cell cis-regulatory atlas in maize that linked chromatin accessibility patterns to cell-type-specific gene expression programs [108]. Such studies demonstrate how single-cell approaches can identify cis-regulatory elements (CREs) and regulatory networks underlying agronomically important traits. These datasets provide valuable targets for both genome editing and cis-regulatory breeding strategies discussed in later sections of this review.
The challenge is whether single-cell analysis can be leveraged to improve complex quantitative traits. The classical single-locus genetic model has been expanded into a quantitative genetics framework [129–131], including epistasis interaction effects [132–135]. Integrating this framework with single-cell data offers the opportunity to account for cellular heterogeneity and intercellular regulatory effects in plants. In a single cell context, narrow sense heritability, defined as the proportion of additive genetic variance relative to total phenotypic variance would, in principle, be ‘narrower’ as the genetic variance would reflect differences within a population of individual cells rather than among bulk tissues or whole plants typically analyzed in population genetics. From a breeder’s equation perspective (Equation (1)), this refined resolution could potentially increase the response to selection, or genetic gain (ΔG), per breeding cycle (t). On the other hand, traditional quantitative genetics models that describe polygenic effects on phenotypic variance may represent an oversimplification, much like earlier single or multilocus models, when compared to the cellular complexity revealed by single-cell approaches. Experimental computational approaches on single-cell based quantitative genetics, such as Perturb-seq or MPRA approaches (e.g., Queitsch lab at University of Washington) will likely provide substantial advances to the multilocus quantitative model established in breeding.
Breeder’s equation, where R = response realized by selection, S = selection differential, i = standardized selection differential, selection accuracy (r), h2 = heritability, 𝜎𝑔 = genotypic standard deviation, and genetic gain (ΔG).
Single-cell technologies also provide an important conceptual bridge between genetics, epigenetics, cis-regulation, and gene editing. By resolving developmental trajectories and regulatory networks at cellular resolution, these approaches facilitate the identification of candidate genes, epigenetic states, and regulatory elements associated with agronomically important traits. These discoveries can subsequently be exploited through modern genome editing and molecular breeding strategies.
Despite its promise, single-cell omics technology in plants has limitations. Genes identified under controlled conditions require validation in field environments, and the high complexity and heterogeneity of plant tissues demand continued advances in computational and statistical methods. Future studies will benefit from integrating multi-omics approaches, including spatial transcriptomics and epigenomics, to better capture the dynamic cell-type cellular landscapes underlying plant environmental and developmental responses [136]. As these technologies continue to mature, hypothesis-driven single-cell studies will likely become an essential component of next-generation crop breeding programs.
Many attempts at identifying the mutations responsible for phenotypes of agronomic interest have historically focused on coding genes (Figure 1 and Section “Genetics as a tool for improving breeding strategies”). However, genome-wide association studies have, in the second part of the 2010s, highlighted the key role of non-coding genetic variation at CREs in determining yield and other traits of agronomical interest in many crop species [137–139], mainly influencing regulation of plant development [140,141]. The main candidates as carriers of causal mutations in such regions are cis-regulatory elements (CREs), short non-coding DNA sequence motifs bound by sequence-specific transcription factors (TFs) [142]. They include promoters that fix the transcription machinery, and distal elements (dCREs), also called enhancers and silencers, that interact with their target gene’s promoter in cis through a chromatin 3D loop, stabilizing or preventing the fixation of the transcription machinery.
Crop responses to biotic and abiotic stresses are also often largely determined by CRE [143]. Beyond promoters and distal CRE, genetic variations in the untranslated regions (UTRs) and introns can also affect integrated phenotypes through modification of the gene expression levels [139]. Targeting genetic variations within regulatory elements in crop breeding programs is thus a promising path to select for genotypes presenting enhanced yield, and most importantly, for phenotypes better adapted to future environments in the context of climate change (Table 1). This is partially reflected in the fact that breeding methods described in Section “Genetics as a tool for improving breeding strategies”, such as hybrid breeding or other types of population structure, lead to changes in allele frequencies at cis-expression quantitative traits loci (cis-eQTLs), genetic variants associated with nearby gene expression levels.
Recent advances in new breeding technologies, particularly CRISPR-based approaches have demonstrated that it is indeed possible to identify and modify functional CREs with reliable effect, resulting in measurable effects on target phenotype [144]. Overexpressing or deleting promoters of genes involved in developmental traits have enabled significative improvements to be measured in yield-related traits such as fruit size [145,146] and inflorescence type in tomatoes [147], heading date in rice [148] or grain yield in rice and maize [110,149]. By modifying promoters or 5’UTRs, it has been possible to improve other traits of agronomical interest in various crops, including modifying tomato fruit color [150], rice chalkiness [151], creating gradients of amylose content in rice [152–154], increasing sugar content in strawberries [155], and ascorbic acid levels in lettuce and lettuce [156]. These approaches have also been successful in improving stress-response-related phenotypes such as water use efficiency of rice [157], drought tolerance in maize [158,159], pest resistance in rice [160,161] and citrus [162]. Beyond genetic variations, epigenetic variations, particularly DNA methylation patterns at CRE (see Section “Leveraging epigenetics in epi-breeding”), play a role in important agronomical traits in several crops [163–165] and may serve as stress-response memory [166]. Thus, targeting CREs and harnessing these epigenetic variations represent promising avenues that deserve further investigation.
Table 1. Summary of cis-regulatory elements (CREs) described by crops, target gene, and their phenotypic effect.
What challenges do we face in our efforts to consistently implement cis-regulation in breeding? Most developmental traits and stress-response phenotypes often display a highly polygenic architecture and may involve a multitude of CREs. Two main concerns need to be overcome to improve breeding by targeting CREs. First, genetic variations at CREs may have a significant effect on the expression of their target gene(s), but they may have a small impact on the final phenotype value [140]. To this end, it will be necessary to select for beneficial alleles at different CREs at once. Second, pervasive pleiotropy may be a barrier. New CRISPR approaches (see Section “Gene editing technologies in breeding”) are aimed at modifying several regulatory regions at once to overcome the first caveat [167]. In addition, geneticists’ attempts at unveiling crops’ regulatory code will help breeders to identify the most relevant CREs [168,169]. Traits of agronomical interest are often partially correlated and share genetic bases [170], which means that it may be complex to select for one trait while not bringing deleterious consequences to another. Furthermore, progress in quantitative genetics approaches have allowed us to map these partial correlations between traits and the identification of their genetic basis [171,172]. Advances in machine learning have enabled the extraction of meaningful patterns from large and diverse datasets ranging from genomics to phenomics [169], coupled with progresses in long-read sequencing technologies [173,174] that provide better assemblies of intergenic regions and facilitate comparative genomics analyses of distal CREs [143], have greatly enhanced the identification of genomic targets of interest. Inference of these gene regulatory networks underlying the trait expression has also reduced the candidate pool of CREs [175,176]. Among the structured regulatory networks, we should narrow the candidate pools to a subset of CREs that are well-connected in pathways of interest but fairly isolated from others [177,178]. This increasing knowledge in determining, filtering and deciphering the importance of CREs in the molecular bases of crop phenotypes has allowed the selection of genomic and epigenetic alleles for breeding purposes. Therefore, cis-regulation CREs offer good future potential for genetic crop improvement, particularly in variety development in the context of climate change.
All of these advances in the understanding of the molecular bases of the complex phenotypes that determine crop performance are now beginning to be included in breeding programs. As highlighted in Section “Genetics as a tool for improving breeding strategies”, one of the aims of genetics in breeding is to combine these beneficial genetic variants to select for desired traits. However, dependence on spontaneous beneficial mutations is very time-consuming [179,180]. Consequently, methods for inducing genetic variation are gaining attention, as they have the capacity to rapidly generate novel mutations [181]. Traditional physical and chemical mutagenesis treatments can generate random mutations, but both require large-scale screening to discover which ones are valuable [182,183].
The emergence of gene editing technology has overcome the unpredictability of mutations by enabling precise, targeted modifications within specific genomic sequences [184,185]. Technologies such as single-cell sequencing (discussed in Section “Leveraging single-cell sequencing in breeding”) have elevated plant breeding from tissue level to single-cell level, greatly enhancing the resolution of molecular mechanisms underlying target traits. This provides targets for gene editing in breeding, which are then achieved through programmable molecular tools that recognize and interact with predetermined DNA sites. Early platforms, such as Zinc Finger (ZF) and Transcription Activator-Like Effector (TALEN) DNA-binding domains, required complex protein engineering for each new target. In contrast, the Clustered Regularly Interspaced Short Palindromic Repeats (CRISPR) system is distinguished by its simplicity and powerful programmability, facilitated by a guide RNA (gRNA) that is complementary to the target DNA sequence and a CRISPR-associated (Cas) nuclease. The Cas-gRNA complex locates its target by scanning for a protospacer adjacent motif (PAM) sequence, enabling highly specific DNA binding [186–189].
The canonical Cas nuclease primarily induces random small insertions and deletions (indels) at the target site, often resulting in gene knockouts that are valuable for functional genetics and breeding [190]. Gene editing technology has since evolved beyond simple nucleases. By fusing the CRISPR system with diverse effector domains such as deaminases, transposases, and recombinases, researchers can now introduce a wide spectrum of genomic variations. For instance, base editing permits precise single-nucleotide substitutions, while prime editing mediates all 12 possible base-to-base conversions, in addition to targeted insertions or deletions [191,192]. These advances are driving innovation in breeding procedures, offering superior efficiency and precision over conventional techniques (Table 2). Notably, these highly precise tools are well-suited for modifying the cis-regulatory elements detailed in Section “cis-regulation: cis-regulatory elements as opportunities for crop breeding”, allowing breeders to fine-tune gene expression for complex traits like quality and disease resistance rather than completely knocking out gene function.
Although gene editing has markedly enhanced the efficiency of genetic breeding, it has typically been limited to modifying a small number of genes per experiment. In the context of global climate change, which threatens crop resilience, there is a growing need for more transformative strategies. One promising approach is the de novo domestication of wild species, which naturally harbor robust resistance to biotic and abiotic stresses. Leveraging powerful gene-editing toolkits [193], breeders can now rapidly introduce domestication traits into wild relatives. Successful examples in tomato and rice demonstrate that gene editing can recapitulate centuries of natural domestication in a laboratory setting by simultaneously altering multiple key genes [194–196].
Table 2. Comparison of genetic editing tools, key mechanisms and main advantages.
A significant advantage of gene-edited crops is that their genomes can be modified without integrating foreign genetic material, resulting in products that may be indistinguishable from those developed through conventional breeding. Nevertheless, a primary concern for clinical and agricultural applications remains off-target effects—the introduction of unintended, potentially deleterious mutations at non-target genomic sites [197]. Significant progress has been made in mitigating this risk through the development of high-fidelity Cas variants, optimized sgRNA designs, prime editors, and anti-CRISPR proteins [198,199]. A future direction involves integrating the concepts of epi-breeding described in Section “Leveraging epigenetics in epi-breeding”. For instance, by exploiting epigenetic marks, “reader” proteins could be recruited to specific loci to modulate gene transcription by rewriting epigenetic states, thereby directing phenotypic outcomes without altering the underlying DNA sequence [88,200].
Regulatory frameworks are evolving in tandem with technological improvements. An increasing number of countries and regions have established policies that exempt certain gene-edited products from the stringent regulations applied to traditional genetically modified organisms (GMOs) [201]. For example, the United States and Japan have approved the commercial marketing of gene-edited crops, including high-GABA tomatoes and non-bitter mustard greens. This policy shift facilitates a potential global adoption of gene-edited crops. As technology continues to advance, it holds the promise of developing crop varieties that meet the long-standing demands for enhanced adaptability and resilience in a changing global environment.
Breeding methods have undoubtedly been essential in crop improvement. The increasing knowledge of plant genetics, together with advances in molecular technologies, has not only deepened our understanding of complex genetic relationships underpinning phenotypes but has also greatly enhanced the potential of traditional breeding methods. Since the rediscovery of Mendel’s laws, genetics has increasingly focused first on the efficient identification of genes, and then on their functional characterization (Figure 1). This progress, leveraged by quantitative genetics and functional genomics, has improved our understanding of key scientific questions such as domestication, hybrid breeding, heterosis, and molecular breeding. As an example, the SSD method [14] has greatly contributed to line development of self-pollinated crops and also laid the foundations for RIL development. Genome sequencing tools coupled with RIL populations shed light on genetic questions underlying a single gene and more complex traits [16,17,19,20].
Current crop breeding should shift, and in some cases is shifting quickly, from unpredictable discovery to a more predictable, engineering-oriented discipline. As discussed in this review, the core principles of traditional genetic breeding (Section “Genetics as a tool for improving breeding strategies”) have been improved with intrinsic genetic discoveries and are nowadays greatly enhanced by next-generation multi-omics and molecular tools. We believe that single-cell sequencing (Section “Leveraging single-cell sequencing in breeding”) will play an important role in the future of breeding, since it provides much higher cellular resolution and will allow breeders to pinpoint novel, cell-specific functional targets that will otherwise be missed by bulk tissue analyses (Figure 2). This approach will synergize with established single-locus and classical quantitative genetics to further unveil the relationships behind polygenic trait heritability. The expansion of classical quantitative models will be twofold. The ability to partition additive genetic variance nested at cellular population level offers new opportunities, as single-cell sequencing of a mutant can capture rare cell response with specific traits (e.g., rhizobia interactions, [120]). The possibility of optimizing resource allocation taking the cell into consideration rather than the confounding effect of the plant level phenotype should, in principle, enable more precise estimates of a trait’s narrow sense heritability. Thus, incorporating single-cell data into quantitative models could optimize the estimation of additive genetic variance within an individual’s cell population relative to total phenotypic variance, ultimately leading to increasing genetic gains over time. We believe that this strategy could enhance trait identification, as single-cell data can uncover key genes [107,108,113] and reveal distinct molecular phenotypes [114,115] that have remained hidden at population level across the genome.
Meanwhile, our growing understanding of dynamic epigenetic landscapes (Section “Leveraging epigenetics in epi-breeding”) creates further possibilities for breeding. Characterizing the genetic structure of a population is a requirement for crop improvement, in particular in hybrid breeding (see Section “Genetics as a tool for improving breeding strategies”). Similarly, the epigenetic structure characterization of a population will be a highly relevant factor to consider for the crossing matrix of breeding programs. In addition, significant research has focused on understanding the contribution of epigenetic marks’ to explain the mechanisms of heritable transcriptional states independently of DNA sequence [202] that lead to a better understanding of complex traits [94,203]. It is known that some epi-mutations are stable and heritable in offspring [64]. High levels of heritability are ideal for QTL detection; therefore, QTL identification of epigenetic traits (epiQTL) would have the properties to become targets of natural or artificial selection, and serve as markers for marker-assisted selection. On the other hand, specific DNA methyltransferases may have single targets. However, the potential partial-to-complete redundancy of different DNA methyltransferases for establishing and maintaining DNA methylation patterns in a locus or in a tissue-specific manner [204] can present a challenge. Finally, provided that DNA methylation patterns can be stably transmitted through meiosis and that large phenotypic variations may be due to epimutations, epigenome editing holds great promise by creating additional phenotypic variability in the same genetic background. One powerful but challenging possibility could be to breed cultivars with the potential to trigger a high response to certain environmental cues (e.g., drought), exhibiting a stand-by epigenomic state that would relax chromatin packing and induce favorable gene expression. We could couple gene editing tools by targeting specific epigenetic marks by reprograming epigenetic de novo functions in cells [88,200]. As an alternative to gene editing, we propose the direct application of proteins and RNAs to permanently modify the methylation status of a locus in a plant cell [205,206].
Importantly, single-cell approaches should not be viewed solely as transcriptomic tools, but also as extensions of classical quantitative genetics. Traditional breeding and quantitative genetic models typically evaluate phenotypes at the whole-plant or tissue level, often overlooking cellular heterogeneity and rare developmental responses. Single-cell sequencing provides an opportunity to partition biological variation at the cellular level and identify specific cell populations contributing disproportionately to quantitative traits. In principle, integrating single-cell information into quantitative genetic frameworks could improve estimates of additive genetic variance and narrow-sense heritability by reducing phenotypic noise generated by tissue heterogeneity.
cis-regulatory mechanisms (Section “cis-regulation: cis-regulatory elements as opportunities for crop breeding”) can change breeding strategies that have in some cases relied on gene knockouts. Instead, researchers are increasingly focusing on precise, spatiotemporal fine-tuning of gene expression, which helps optimize complex traits without causing unwanted linkage drag effects, e.g., yield penalties. We also envisage feasible and effective use of cis-regulation in modern breeding as an alternative to gene knockout approaches. Breeding leverages natural diversity as a source of resilience (abiotic and biotic stresses) for high-yielding varieties. However, the effective use of such natural variation resources and intrinsically the fine-tuning of the gene expression regulation underlying this natural phenotypic variation at the genome-wide level remains a challenge. Similarly, the identification of loci to target and also the functional variants that bring cis-element efficiency within the whole genome of cis-acting regulatory loci are key from a genome editing standpoint. Thus, there is a need for comprehensive maps of functional variation at transcription factor binding sites (TFBS), for example by quantifying haplotype-specific TF footprints [207]. A growing body of literature finds a likely relationship between siRNAs (via the RdDM pathway), inbreeding, and transcriptional activity of growth regulatory networks which would be interesting to investigate further for crop breeding purposes. A recent study of F1 parents-offspring in maize identified differentially trans-regulated accessible chromatin regions (ACR) related to heterosis inheritance [208], highlighting the importance of the cis-regulatory element in terms of chromatin accessibility inheritance and the expression patterns of heterotic phenotypes. The gap presented by these challenges is being closed both through advances in long-read sequencing technologies that more accurately assemble the intergenic regions in often complex genomes; and also, due to advances in molecular biology techniques that allow us to more precisely identify these cis-regulatory elements. As described in Section “cis-regulation: cis-regulatory elements as opportunities for crop breeding”, CREs are particularly relevant to regulating abiotic environmental stresses, and thus, selecting for genetic and epigenetic alleles at CREs will be a prospective breeding strategy to consider for crop adaptation to climate change.
Finally, gene-editing technologies have rapidly evolved in the past decade (see Section “Gene editing technologies in breeding”) and have proven their effectiveness in turning all these molecular possibilities into practical outcomes in favor of sustainable agriculture. For instance, we outlined their potential to modify entire cis-regulatory elements to show reliable regulatory variants as alternative ‘allele’ sources of crop improvement even when introduced in different genetic backgrounds.
In this review we revisited and proposed scientific advances in genetics, single-cell analysis, epigenetic marks, cis-regulation, and gene editing offering remarkable value both now and in future to improve traditional breeding strategies by combining genomic and functional knowledge with cutting-edge biotechnology to ensure a sustainable future for our society.
All data is used and presented in the manuscript.
EDM: Conceptualization, funding acquisition, writing—review and editing; MF: writing—review and editing; REV: writing—review and editing; JS: writing—review and editing.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
This work was supported by the COST Action RECROP (CA22157), European Cooperation in Science and Technology.
We thank Alexandre P. Marand (University of Michigan) for critical reading of the manuscript.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
99.
100.
101.
102.
103.
104.
105.
106.
107.
108.
109.
110.
111.
112.
113.
114.
115.
116.
117.
118.
119.
120.
121.
122.
123.
124.
125.
126.
127.
128.
129.
130.
131.
132.
133.
134.
135.
136.
137.
138.
139.
140.
141.
142.
143.
144.
145.
146.
147.
148.
149.
150.
151.
152.
153.
154.
155.
156.
157.
158.
159.
160.
161.
162.
163.
164.
165.
166.
167.
168.
169.
170.
171.
172.
173.
174.
175.
176.
177.
178.
179.
180.
181.
182.
183.
184.
185.
186.
187.
188.
189.
190.
191.
192.
193.
194.
195.
196.
197.
198.
199.
200.
201.
202.
203.
204.
205.
206.
207.
208.
Munaiz ED, Fagny M, Venado RE, Shen J. Genetics Has Enabled the Improvement of Breeding Strategies: Single-Cell Analysis, Epigenetics, cis-Regulation and Gene Editing to Exploit Genetic Variants Underlying Phenotypes. Crop Breed Genet Genom. 2026;8(2):e260015. https://doi.org/10.20900/cbgg20260015.
Copyright © Hapres Co., Ltd. Privacy Policy | Terms and Conditions
