Haploid and Doubled Haploid Platforms for Wheat Improvement: Methods and Applications

Abstract

Haploid and doubled haploid (DH) technologies are important tools for accelerated wheat (Triticum aestivum L.) breeding, enabling the rapid production of fully homozygous lines and increasing the efficiency of genetic analysis of complex traits. This review presents a comprehensive analysis of the main approaches to producing haploid and DH wheat plants, with particular emphasis on androgenesis-mediated and chromosome elimination methods, including wheat-maize hybridisation. The biological basis of androgenesis is discussed in relation to stress-induced microspore reprogramming; however, the primary focus is on the methodological factors determining the efficiency of DH production, including the donor plant genotype, microspore development stage, pretreatment conditions, composition of the induction and regeneration media, and chromosome doubling. However, its widespread application remains limited by pronounced genotypic dependence, low responsiveness of many commercial varieties, albinism, and a lack of universally effective protocols. In contrast, distant hybridisation systems, particularly wheat-maize hybridisation, are generally characterised by greater reproducibility and less genotypic dependence, although they remain labour-intensive and require precise embryo rescue and chromosome doubling procedures. Overall, further progress in producing DH in wheat will be associated with the optimisation of protocols for difficult-to-respond genotypes and the integration of classical haploidisation systems with omics approaches, genomic selection, and genome-editing.

1. Introduction

Wheat (Triticum aestivum L.) is one of the most important food crops and plays a key role in ensuring global food security. However, increasing productivity under current conditions is constrained by the lengthy breeding cycle, the complex genetic architecture of most economically important traits, and the need to rapidly develop varieties resistant to abiotic and biotic stresses [1,2].

One of the most effective tools for accelerating breeding is the production of haploid and doubled haploid (DH) plants, which enables the generation of fully homozygous lines within a single generation. Compared with traditional multi-generation self-pollination schemes, DH technologies markedly shorten the breeding cycle, increase selection efficiency, and provide highly valuable material for gene mapping, QTL analysis, functional genomics, and other genetic studies [3,4,5].

Several approaches for haploid production in wheat have been developed, differing in their biological basis and technological application. The principal methods include in vitro androgenesis (anther culture and isolated microspore culture), gynogenesis, pollination with irradiated pollen, and distant hybridisation followed by chromosome elimination, particularly in the wheat × maize system [6]. The efficiency of these approaches is strongly influenced by genotype, explant developmental stage, pretreatment regime, nutrient medium composition, and regeneration capacity, which continue to limit the broad use of DH technologies in routine breeding practice [7,8].

Among the available systems, androgenesis occupies a special place as a model of stress-induced microspore reprogramming, in which development switches from the gametophytic to the sporophytic pathway. This process is important not only from an applied perspective but also as a model for investigating cellular reprogramming and the regulation of embryogenesis. At the same time, the practical value of androgenesis is determined primarily by the possibility of optimising protocols to increase embryogenesis and regeneration frequencies and to reduce the effects of genotypic dependence and albinism [9,10,11,12].

In parallel with the development of in vitro systems, substantial attention has also been paid to haploid production through chromosome elimination in interspecific and intergeneric crosses. Among these, the wheat × maize system remains one of the most established non-androgenic routes for DH production in wheat, while recent comparative evidence indicates that alternative pollen donors such as Imperata cylindrica may outperform maize in some materials. At the same time, these studies also show that induction efficiency remains strongly dependent on genomic background, ploidy level, and the specific donor combination, indicating that chromosome elimination systems, although highly useful, are not universally applicable [13].

Despite the rapid emergence of new molecular approaches, androgenesis and chromosome elimination remain the main practical and biologically informative systems for DH production in wheat. Nevertheless, recent studies from 2024 to 2026 have substantially expanded the conceptual framework of the field by showing that haploid induction can also be engineered through defined genetic factors [14,15,16,17]. These advances are important not because they replace classical systems in wheat breeding at present, but because they provide a mechanistic framework for understanding haploidisation, genome elimination, and the biological causes of efficiency differences among DH systems.

In particular, functional studies have identified several key genes associated with in vivo haploid induction, including MTL/ZmPLA1/NLD, DMP, PLD3, and CENH3, thereby establishing gene-based platforms for haploid induction in multiple crops [14,15,16]. In wheat, proof-of-concept studies have already shown that editing of TaMTL can induce maternal haploids, whereas editing of TaCENH3 can generate paternal haploids, demonstrating that gene-driven haploid induction is feasible in this species, although not yet broadly established as a routine breeding platform [16]. Additional recent work has also identified TaPLD as a promising candidate associated with ploidy instability and possible haploid induction in wheat [15].

These findings further indicate that haploid induction is not governed by a single molecular route. Instead, it may arise from several mechanistically distinct processes, including fertilization defects, genome elimination, and stress-associated genomic instability. For example, mutation of the wheat peroxidase gene TaPOD8 was recently shown to trigger paternal haploid induction through a reactive oxygen species burst associated with DNA damage and chromosome fragmentation in reproductive cells [17]. Although such systems are still limited by reduced fertility and incomplete breeding applicability, they are important because they reveal mechanistic links between stress responses, genomic instability, and haploid formation, which may also help explain the strong genotype dependence observed in classical DH systems [17].

The importance of DH approaches has therefore increased further with the expansion of genomic selection, omics platforms, and genome editing. Full homozygosity makes DH lines especially suitable for QTL mapping, GWAS, SNP genotyping, transcriptomic and epigenomic analyses because it reduces intraline variability and improves phenotyping reproducibility [18]. This is particularly relevant in polyploid wheat, where genomic complexity and homoeologous variation complicate both genetic analysis and breeding.

Recent work has also shown that DH technologies can be integrated with modern genome engineering. Haploid inducer-mediated genome editing systems, such as HI-Edit and IMGE, allow CRISPR/Cas reagents to be delivered through inducer lines, enabling editing in elite germplasm without direct transformation of recalcitrant target genotypes and facilitating the recovery of homozygous, transgene-free edited lines after chromosome doubling [16,19,20]. In parallel, morphogenic regulators such as TaWOX14 have been shown to improve regeneration and editing efficiency in wheat, thereby facilitating the functional testing of candidate haploid induction genes [15]. However, these emerging systems remain constrained by the same bottlenecks that affect classical methods, especially genotype dependence, variable regeneration capacity, and reduced fertility in some inducer backgrounds [12,15,16,17,18,19,20].

Thus, despite substantial progress, technologies for producing DH lines in wheat are still characterised by pronounced genotype dependence, inconsistent reproducibility of some protocols, and the need for further optimisation of individual steps. Several major knowledge gaps remain unresolved. There is still no broadly applicable system that simultaneously combines high efficiency, low genotype dependence, and operational simplicity across a wide range of wheat genotypes. Reliable molecular and epigenetic predictors of androgenic responsiveness remain insufficiently validated, and the translation of omics-based and gene-based insights into routine gains in DH efficiency is still limited. In addition, although recent studies have established the feasibility of integrating haploid induction with genome editing, the methodological robustness, breeding-scale applicability, and comparative value of these emerging systems relative to classical androgenesis and chromosome elimination approaches remain insufficiently assessed in wheat [14,15,16,17]. Therefore, this review aims to summarise the current approaches for producing DH wheat plants, with particular emphasis on androgenesis and chromosome elimination systems, examine the factors determining their efficiency, and assess their significance for accelerating breeding and advancing functional genomic research in the broader context of recent progress in haploid induction biology.

2. Androgenesis

Traditional methods for obtaining homozygous wheat lines are time-consuming and labor-intensive; therefore, haploid and DH production techniques are considered promising approaches for rapid genotype fixation and accelerating genetic progress in breeding [21]. The first haploid plants were described in 1921–1922 [22], and haploid wheat grains were first obtained in 1926 through interspecific hybridisation [23]. In the early 1970s, the first regenerants were obtained from wheat using anther culture [24]. Later, haploids were obtained using isolated microspore cultures and distant hybridisation [25].

Conventional gene stabilisation and the development of pure lines require self-pollination over 8–9 generations, whereas the DH method enables the production of genetically stable pure lines within approximately one year after the initial cross. DH technology accelerates breeding and early cultivar release, facilitates gene introgression from wild relatives through backcrossing, and provides completely homozygous material [26].

Haploid plants are mainly produced using four approaches: anther or microspore culture (androgenesis), culture of unfertilised ovules (gynogenesis), pollination with irradiated pollen, and interspecific and intergeneric hybridisation, followed by chromosome elimination. The choice of the most effective method is determined by the biological characteristics of the species [6]. Among these methods, androgenesis is one of the most widely used approaches for the rapid production of homozygous wheat plants with valuable agronomic traits.

Androgenesis, a haploid technique that enables the rapid generation of completely homozygous material, accelerates breeding through the fixation of recessive traits. It is based on the stress-induced reprogramming of microspores toward sporophytic development and the formation of haploid embryoids in vitro [27]. In contrast to conventional breeding, which requires multiple cycles of self-pollination, DH technology enables the generation of homozygous lines within two generations, substantially reducing the duration of breeding programs and increasing selection efficiency owing to the predominance of additive genetic variation [28]. Androgenesis has a high potential for DH production because each anther contains thousands of microspores, each capable of giving rise to a plant that is either haploid or spontaneously DH. Microspores are formed after meiosis and may undergo natural or induced genome doubling, making them highly valuable for genetic studies and breeding [29].

At the cellular level, embryogenic microspores are characterised by a centrally positioned nucleus, cytoplasmic condensation, and reconfiguration of the cytoskeleton [30,31], accompanied by changes in vacuolar structure and microtubule reorganisation, which promote symmetric cell division, distinct from the asymmetric division characteristic of pollen formation [10,11,32]. To better understand the biological basis of these cytological changes, it is important to consider the underlying stress-induced mechanisms of microspore reprogramming. Stress pretreatments used for androgenesis induction trigger a complex cellular response in microspores that is associated with stress perception, synthesis of small heat shock proteins, and reorganization of the cytoskeleton and its polarity axis, thereby redirecting development away from the normal gametophytic program [33]. At the same time, increasing evidence indicates that this transition is tightly linked to redox regulation. In particular, reactive oxygen species (ROS) are not merely stress by-products, but act as signalling components of microspore reprogramming, and a threshold level of ROS appears to be required for the initiation of embryogenic development, whereas excessive ROS accumulation leads to oxidative damage and reduced cell viability. Effective microspore embryogenesis therefore depends on a balance between ROS generation and antioxidative defence, including the activity of enzymes such as superoxide dismutase and catalase [34]. These stress- and redox-associated events are accompanied by cellular reprogramming, including re-entry into the cell cycle and symmetric division of the microspore nucleus, which represents one of the key morphological markers of androgenic induction [33]. At the molecular level, embryogenic transition is further associated with epigenetic regulation, including changes in DNA methylation, histone acetylation, and chromatin structure, which collectively modulate transcriptional activity and embryogenic competence. Consistent with this, activation of embryogenesis-related regulators such as WOX, BBM, LEC, and SERK has been associated with the acquisition of embryogenic potential and subsequent embryo formation in wheat in vitro systems [35]. Together, these coordinated stress-related, redox-dependent, epigenetic, and transcriptional processes lead to embryoid formation and regeneration of haploid or doubled haploid plants. The key stages of stress-induced microspore reprogramming are summarised in Figure 1.

Alongside these pronounced cytological changes, androgenesis is accompanied by profound molecular and epigenetic reprogramming, including chromatin remodelling and alterations in DNA methylation levels and histone modifications, leading to the repression of gametophyte-specific gene expression and the activation of embryogenic programs [9]. A recent comparative transcriptome analysis of two wheat genotypes with contrasting androgen responsiveness revealed that successful microspore embryogenesis is associated with more pronounced regulation of processes related to nucleosome assembly, chromatin remodelling, and organisation, as well as coordinated changes in the expression of genes involved in stress response, hormonal signalling, cytoskeletal and cell wall dynamics, and metabolic pathways [36]. These data suggest that androgenic competence in wheat depends on coordinated multi-level reprogramming rather than a single molecular trigger. Simultaneously, the study highlights that the molecular basis of recalcitrance remains insufficiently understood, and the identified associations still require functional validation before they can be translated into reliable markers or broadly applicable DH protocols.

However, despite this growing mechanistic understanding, the biological interpretation of microspore embryogenesis in wheat still requires caution. Much of the currently available evidence remains correlative, as many proposed regulators and pathways have been identified primarily through cytological observations or transcriptome profiling rather than direct functional validation in contrasting wheat genotypes. In addition, conclusions are often extrapolated from more responsive model systems or from a limited number of androgenesis-competent wheat lines, which restricts their general applicability. It is also important to note that improved embryogenic induction does not necessarily translate into higher frequencies of green fertile regenerants, indicating that embryogenic reprogramming, plant regeneration, and albinism control are biologically linked but not functionally equivalent processes. Thus, one of the central unresolved questions is not simply which molecular events accompany androgenesis, but which of them have predictive and causal value for improving reproducible DH production in breeding-relevant wheat germplasm.

Androgenesis has been reported to occur via several embryogenic pathways that differ in the type of primary division and the participation of vegetative and generative cells [37,38]. Nonetheless, all these pathways lead to the formation of multicellular structures morphologically similar to zygotic embryos [39]. Embryogenic competence in wheat is highest in late-uninucleate microspores, which are characterised by a large central vacuole and peripheral nucleus localisation [11].

Thus, androgenesis in wheat is a multi-level biological process that encompasses stress-induced cellular reprogramming, coordinated cytological transitions, and extensive molecular and epigenetic changes. Understanding these fundamental mechanisms is a key prerequisite for improving the efficiency of haploid and DH plant production and exploiting androgenesis as a platform for both fundamental and applied research.

2.1. Androgenesis-Based Haploid Production in Wheat

2.1.1. Biological Basis

Numerous endogenous and exogenous factors influence the embryogenic responses of anthers in culture. The genotype, physiological status, and growth conditions of the donor plants, pollen developmental stage, pretreatment of flower buds or anthers, culture medium composition, and in vitro culture conditions, together with their interactions, all significantly affect the anther response in vitro [38]. The main triggers of androgenesis include temperature effects, osmotic stress, carbon and nitrogen starvation, and chemical inducers [16,40,41,42,43,44,45]. Epigenetic regulation, including chromatin remodelling, changes in DNA methylation, and histone modifications, plays a significant role in this process, as further confirmed by the positive effect of epigenetic modulators on the regeneration rate of green plants [28,46,47].

In addition, the efficiency of androgenesis in wheat varies markedly among genotypes. Androgenic responsiveness, which is controlled by polygenic systems, involves additive, dominant, and epistatic effects, as well as interactions between nuclear and cytoplasmic genomes [11,24]. Moreover, components such as embryoid induction, plant regeneration, and albino frequency are inherited independently, further complicating the optimisation of universal protocols.

2.1.2. Pretreatment

The efficiency of producing DH plants through induced embryogenesis is largely determined by the pretreatment of the donor material using cold, heat, osmotic, or chemical stress applied to ears, flower buds, or isolated anthers. A key step in culturing isolated microspores is the induction of embryogenesis, typically through abiotic stress. A morphological marker of successful reprogramming is the formation of “stellate” microspores, characterised by fragmented vacuoles and a centrally located nucleus, followed by symmetrical cell division. The proportion of microspores with this morphology is directly correlated with the efficiency of embryoid formation and plant regeneration [6,11].

Cold stress is traditionally used in wheat and other cereals. For example, in poorly responding durum wheat genotypes, prolonged treatment at 4 °C for up to five weeks promotes the successful induction of embryogenesis and increases the regeneration capacity of green plants [48]. In wheat, cold pretreatment at 1–8 °C improves the androgenic response by increasing both the callus yield and spontaneous chromosome doubling rate. A systematic analysis showed that the efficiency of androgenesis is strongly dependent on the type and duration of stress pretreatment: for winter wheat, the optimal duration of cold pretreatment of ears is 28 days, while for spring wheat, it is 21 days, and the use of a 23% maltose gradient allows for efficient enrichment of suspensions with embryogenic microspores [6]. However, recent data indicate that cold treatment does not always induce expected physiological changes. A 2025 study showed that low-temperature pretreatment, alone or in combination with sodium selenate treatment, had virtually no effect on IAA, IAA-glutamic acid, and ox-IAA levels in the anthers of different wheat lines [49].

Another widely used approach is the induction of osmotic stress using mannitol, a non-metabolizable sugar that reduces the water potential of the culture medium. Mannitol treatment induces specific cellular responses, including the reorganisation of microspore microtubules, which form new cytoskeletal configurations and are associated with the induction of embryogenesis [50]. Experiments on durum wheat have shown that pretreatment of anthers with mannitol (0.3–0.7 M) at 4 °C has a pronounced but genotype-dependent effect on embryoid induction and green plant regeneration, and the composition of the culture medium and the duration of cold exposure also significantly affect the results [51]. Pretreatment with mannitol is widely used in practical schemes for androgenesis induction. For example, treatment of anthers with 0.7 M mannitol ensured a high yield of green plants: 407 green regenerants were obtained from 35 tested microspore isolates, of which approximately 67% spontaneously doubled their chromosomes [52].

In addition to osmotic pretreatment, other types of stress-inducing effects have been studied. For example, a combination of mannitol treatment and short-term exposure to n-butanol significantly enhanced embryogenesis. In the experiments of Dubas et al. (2021), treatment of Pavon wheat anthers with 0.2% n-butanol after mannitol stress significantly increased the number of formed embryoids and regenerants due to stress-induced fragmentation and reorganisation of microtubules [50]. Among alcohol-containing stress-inducing agents, n-butanol proved to be the most effective; in spring wheat, short-term addition of n-butanol (0.1–0.2%) to the induction medium significantly increased the formation of embryoids, green plants, and DH lines, whereas sec- and tert-butanol had no significant effect [53].

The use of chemical inducers is another important approach to enhance the efficiency of androgenesis. The application of an inducer containing 2-hydroxynicotinic acid, 2,4-D, and BA enabled the redirection of up to 50% of microspores toward the sporophytic developmental pathway, resulting in the production of 50–5500 green plants per spike across different wheat genotypes [44]. Zheng et al. (1999) showed that short-term treatment of freshly isolated wheat microspores with various inducers, including 2-hydroxynicotinic acid, benzotriazole, and other compounds, increased microspore survival and embryogenesis efficiency, with the most pronounced effect observed with 0.18 mM 2-hydroxynicotinic acid [43]. The most effective modern schemes involve the use of combinations of stress factors, such as mannitol-induced starvation, and chemical inducers, including 2-HNA, BAP, and 2,4-D. In sensitive genotypes, such systems, including the flask and fresh microspore methods, can redirect 20–50% of microspores to the sporophytic development pathway [11]. The combination of stressors is generally more effective than individual treatments. In wheat, a combination of starvation and heat stress induces the formation of embryogenic microspores, whereas in the absence of stress, mature and fertile pollen develops [10]. The practical effectiveness of the combined heat and chemical stress was also confirmed in DH production schemes. A modified protocol by Liu et al. [44], based on the use of 33 °C instead of cold treatment and a mixture of 2-HNA, 2,4-D, and BAP, provided higher rates of callus induction and green-plant regeneration.

In recent years, it has been shown that the use of biologically active compounds, including agents that disrupt microtubule organisation (colchicine, n-butanol), membrane-permeable substances such as dimethyl sulfoxide (DMSO), and epigenetic modifiers, in particular, histone deacetylase inhibitors, can significantly increase the frequency of microspore reprogramming in combination with classical stresses, such as cold, osmotic stress, and starvation [11,46,54]. Experimental data indicate that targeted influence on the epigenetic state of microspores, in particular using trichostatin A, can significantly increase the frequency of embryogenesis, emphasising the key role of epigenetic regulation in the formation of androgen competence [55]. For example, the use of trichostatin A in combination with mannitol stress led to a fourfold increase in the number of green DH plants compared with treatment with mannitol alone. Ultrastructural analysis revealed an increased proportion of microspores with stellate morphology and symmetrical division, whereas marker gene expression analysis confirmed the involvement of epigenetic regulation in early embryogenesis signalling pathways. In addition to TSA, other epigenetic modifiers are also being investigated, including scriptaid, BIX-01294, and sodium butyrate; however, their effects are often genotype- and dosage-dependent; therefore, further research is needed before their introduction into routine practice. The inclusion of trichostatin A in protocols further increases the frequency of microspore embryogenesis and green-plant regeneration [6,46].

The results of isolated microspore culture can also be improved by using antioxidants, cell proliferation factors, and ovary-pretreated medium, which collectively increase the number of embryoids and regenerated green plants. Moreover, the pretreatment stage is crucial for the development of androgenesis protocols. Castillo et al. [55] induced stress using a specific SM medium containing the macronutrients present in FHG medium along with 127.5 g/L mannitol, 40 mM CaCl₂, and 8 g/L Sea Plaque agarose. Embryogenesis was induced in MSMI medium supplemented with 2,4-D (1 mg/L) and BA (1 mg/L), and regeneration was performed in hormone-free J-25-8R medium.

Thus, the effectiveness of androgenesis in wheat is determined not by a single factor but by a combination of the type, duration, and intensity of stress pretreatment, the composition of the environment, and the use of additional regulators. Combination protocols that combine classical stress treatments with chemical and epigenetic modulators are the most promising; however, their effectiveness remains largely dependent on the genotype.

2.2. Anther Culture

2.2.1. Principle

Anther culture, a classical androgenic method, is one of the earliest technologies used for haploid production in wheat. This method is based on culturing intact anther-containing microspores at responsive developmental stages, predominantly the uninucleated microspore stage. In contrast to isolated microspore cultures, microspore development in this system occurs in the presence of anther tissue, including the anther wall and tapetal elements, which can substantially influence embryogenesis induction [7,10,11]. The anther culture workflow involves donor plant growth, selection of spikes at the late uninucleate stage, stress pretreatment, sterile anther excision, embryoid induction, regeneration, and subsequent assessment of chromosome doubling [56].

In anther culture, androgenesis is typically initiated by stress pretreatment, following which microspores exit the gametophytic pathway and form embryoids or callus-like structures. A notable feature of this approach is the potential contribution of somatic cells from anther tissue to regeneration, which may result in the production of diploid plants of somatic origin [11,57]. However, recent evidence indicates that individual-level protocol optimisation is required even in closely related wheat species. In the tetraploid wheat species Triticum timopheevii, low regeneration efficiency has long been a major limitation. In a study on DH plant production, the use of NPB-99 medium generated more embryoids than that of C17 and 190-2; nevertheless, even using NPB-99, green plant recovery was extremely low (approximately one green plant obtained per 100 anthers), and the frequency of albino plants reached 96% [58].

Using a simplified protocol for bread wheat, plant regeneration from microspore-derived embryoids was achieved using a single medium under a single culture regime [59]. After cold pretreatment, the anthers of the cultivar Pavon 76 were cultured on modified 85D12 medium, achieving responsiveness of up to 53% of anthers and production of up to 22 green and 25 albino plants per 100 anthers; approximately half of the green regenerants were fertile, indicating high levels of spontaneous chromosomal doubling, although strong genotypic dependence was observed [60]. Studies employing cold pretreatment have revealed high variability in wheat responses and the feasibility of obtaining productive DH lines. After 6–8 days at 4 °C, the anthers were cultured on N6 medium supplemented with 2,4-D, kinetin, and elevated sucrose, and regeneration was performed on MS medium containing NAA and kinetin. Callus formation was observed in all genotypes tested, and regenerants were obtained in most of them. Hybrids have outperformed their parental lines in many cases, some exhibiting heterosis, and the resulting DH lines have exhibited good yield potential under field conditions [59].

Adding Cu to the liquid induction medium AMC increases the proportion of green regenerants. The embryogenic response was strongly genotype-dependent; all spring wheat hybrids and approximately 63% of winter wheat hybrids produced embryoids and green plants, whereas the magnitude of the response varied considerably among genotypes [61]. Micronutrients in the induction medium, such as Cu(II) and Ag(I), can affect green regenerant yield through metabolic–epigenetic mechanisms, and their concentrations are associated with perturbations in S-adenosylmethionine (SAM) metabolism and de novo DNA methylation, reflecting the epigenetic influence of anther culture conditions on regeneration [47]. Spring wheat genotypes produced significantly more androgenic structures than winter wheat genotypes. On C17 medium, the optimal response depended on the composition of the growth regulators: spring-type genotypes responded better to auxin-based formulations (2,4-D with dicamba), whereas winter-type genotypes showed higher responses to a combination of 2,4-D and kinetin. All androgenesis-related parameters were strongly dependent on the genotype [29].

Although anther culture remains an accessible method for DH production in wheat, its efficiency depends on the microspore developmental stage, pretreatment regime, composition of the induction and regeneration media, and especially on the genotype of the donor plant. Contemporary studies emphasise the need for genotype-specific optimisation, reduction in albinism, and enhancement of spontaneous chromosome doubling to improve androgenic outcomes [25].

2.2.2. Advantages

Anther culture remains an attractive method owing to its relative technical simplicity and low equipment requirements. The absence of a microspore isolation step makes this method accessible to a wide range of laboratories and facilitates the implementation of androgenesis in breeding programs without requiring substantial infrastructure investment [7,11,28]. This method, which is widely applied in cereal breeding, has demonstrated higher cost-effectiveness for DH production than intergeneric hybridisation approaches. Heterozygous genotypes can produce larger numbers of androgenic haploids, and key androgenic parameters are controlled by both additive and non-additive genetic actions [62]. When responsive genotypes are used, anther culture enables the stable production of haploid and DH lines that are suitable for selection and subsequent field evaluation [44,58,59].

2.3. Isolated Microspore Culture

Principle

In isolated microspore culture–the most direct form of androgenesis–embryogenic development is induced directly within the microspore without the involvement of anther tissue. In contrast to anther culture, this approach eliminates the direct influence of somatic cells of the anther wall and tapetum, reduces the probability of somatic-origin regenerants, and allows each microspore to be regarded as an autonomous embryogenic unit [7,10,11,39,57].

Isolated microspore culture provides a unique experimental system for investigating the mechanisms of cellular reprogramming, as the transition from the gametophytic to sporophytic developmental pathway occurs synchronously and can be induced by controlling stress factors [45,46]. This method is thus widely used not only in applied breeding but also in fundamental studies on embryogenesis and epigenetic regulation of development. Simultaneously, the practical effectiveness of this method largely depends on the conditions of induction and regeneration.

Optimisation of the regeneration media and pretreatment conditions markedly increased green plant recovery. In particular, the use of regeneration media supplemented with ascorbic acid, in combination with mannitol pretreatment at 4 °C, resulted in the highest frequency of green plant regeneration across different wheat cultivars [63]. The roles of phytohormones and hormone-like compounds are actively being investigated. The addition of zearalenone in combination with growth regulators, such as 2,4-D, dicamba, or kinetin, to the induction medium enhances microspore embryogenesis efficiency and helps overcome genotype-dependent recalcitrance in wheat hybrids [64].

2.4. Major Limitations of Anther Culture and Isolated Microspore Culture in Wheat

Anther and isolated microspore cultures are the primary in vitro systems for androgenesis in wheat and are based on induced microspore switching. Despite the common biological principle, the effectiveness of these methods varies significantly depending on the genotype, pretreatment conditions of the donor material, nature of the stress factors, and composition of the culture media. The lack of a universal protocol has led to a wide variety of experimental designs for inducing embryogenesis and plant regeneration in wheat. A summary of the main pretreatment regimens, induction media, and regeneration systems used in anther and isolated microspore cultures is provided in Table 1.

Despite their high practical potential, both androgenesis systems in wheat have significant limitations. These include pronounced genotypic dependence, in which most genotypes are characterised by low or unstable androgenic responsiveness [11,61], as well as a high frequency of formation of albino regenerants, which sharply reduces the yield of viable green plants [9,44,65,66,67]. These factors significantly limit the widespread use of androgenesis in breeding programs. Moreover, anther and isolated microspore cultures differ in their method-specific limitations.

For anther culture, specific problems include the high frequency of somatic callus formation and the regeneration of diploid plants of somatic origin, which reduces the proportion of true haploids and requires additional ploidy testing [7,57]. In addition, anther culture is susceptible to somaclonal variability, including chromosomal aberrations, aneuploidy, and ploidy instability, which can negatively affect the agronomic value of the obtained lines [7,11,24]. The low frequency of haploid production and the need for complex multi-step protocol modifications further limit the application of this method; as a result, anther culture is more often used in laboratory studies and breeding programs with a limited number of responsive genotypes [26].

In contrast, isolated microspore culture has significant potential for the accelerated generation of haploid and DH lines, particularly when combined with molecular markers and genomic and epigenomic approaches. The ability to work with large microspore populations makes this method a promising tool for functional genomics and the study of the early stages of embryogenesis [7,48]. However, its application is also significantly limited, as most protocols demonstrate high efficiency mainly in spring genotypes, while winter wheat is often characterised by low responsiveness [6,11,68]. Additional specific problems include the lack of standardised protocols and the high sensitivity of the method to cold pretreatment conditions, enrichment of embryogenic microspores, and composition of the culture medium. For this reason, isolated microspore culture is often considered a complement rather than an alternative to the wheat × maize system [6,7,9].

The pronounced genotypic dependence of androgenesis in wheat is associated with the complex genetic control of androgen responsiveness, including additive, dominant, and epistatic effects, while individual components of the response can be inherited independently [69]. Modern transcriptomic data additionally show that differences between responsive and difficult-to-respond genotypes are formed as a result of a complex interaction between genetic, epigenetic, and environmental factors [36].

A significant limitation of the practical application of androgenesis in wheat is the high frequency of albino regenerants, which are incapable of normal autotrophic development and are therefore useless for agronomic production. Current data indicate that albinism is associated not only with cultivation conditions but also with disturbances during the early stages of chloroplast biogenesis. In particular, green plant regeneration correlates with a successful transition from NEP-dependent to PEP-dependent transcription during early plastid development, whereas disruption of this transition is associated with the regeneration of albino plants. The resistant Paledor genotype was found to have very low expression levels of 16S and 23S plastid rRNAs, as well as delayed GLK1 activation, indicating disturbances in plastid translation and regulation of chloroplastogenesis in the first weeks of in vitro culture [27]. At the same time, the literature shows that the causes of albinism are not limited to a single mechanism: in addition to disturbances in the expression of genes involved in chloroplast biogenesis, deletions and rearrangements of plastid DNA, as well as a deficiency of plastid ribosomes, have been discussed, and not all albino plants carry detectable changes in ptDNA [70]. This highlights the multifactorial nature of albinism and explains why this problem significantly limits the practical effectiveness of androgenetic agronomy in wheat.

In this regard, the current research aims to develop more universal protocols adapted to a wide range of genotypes. In particular, it has been shown that optimising the duration of cold pretreatment, using maltose gradients to enrich embryogenic microspores, and modifying the medium composition can increase the yield of embryoids and green plants in both spring and winter wheat [6,7,9].

Table 1. Overview of donor pretreatments, induction media compositions, and regeneration conditions in wheat anther culture (AC) and isolated microspore culture (IMC).

Genotype	Method	Pretreatment	Induction Medium	Regeneration Medium	Response (%)	Green Shoots (%)	Albino Shoots (%)	Reference
Triticum aestivum: Pavon, Caramba	AC	SM (0.7 M mannitol, 5 days, 25 °C, darkness) → MSMI + 0.2% n-butanol (2–3 h)	MSMIF2 + 6 mature ovaries → MSMIF4 + Ficoll	J25-8	not reported	not reported	not reported	[8]
Triticum spelta L.	AC	Cold pretreatment of donor shoots: 3–4 °C, 14 days	W14mf (300 anthers per Petri dish, 32 °C for 3 days → 28 °C, darkness, 8 weeks)	190-2Cu	not reported	not reported	not reported
Triticale	IMC	Cold pretreatment of spikes: 4 °C, 21 days (darkness, immersed in water)	CIMC with 20% maltose gradient → TSA 1 µM (10 min) → CIMC with PSK-α (0.1 µM) + 6 ovaries per dish	CIMC-4	not reported	not reported	not reported
Triticum aestivum	AC	Cold pretreatment of spikes: 4 °C, ~14 days (in water, wrapped with PVC, darkness)	P-4mf liquid medium (P-4 with 9% maltose; 10% Ficoll; potato extract); 32 °C for 3 days → 28 °C for 5–6 weeks	190-2Cu	not reported	not reported	not reported	[56], pp. 59–64
Triticum aestivum (Central and Eastern European genotypes, n = 32)	AC	Spike selection → surface sterilization with 20% NaOCl for 30 min	W14 medium containing colchicine (40 mg L−1) for 3 days at 29 °C in darkness → followed by W14 without colchicine	190-2 regeneration medium → 190-2 medium without plant growth regulators	6.8–82.3 (avg. 26.3)	1.8–43.8 (avg. 11.4)	not reported	[56], pp. 65–70
Winter wheat	AC	Cold pretreatment at 7 °C for ≤14 days	190-2 medium supplemented with 9% maltose, 2,4-D (1.5 mg L−1) and kinetin (0.5 mg L−1), without NAA; incubation at 28 °C in darkness	190-2	not reported	3.3–5.6	not reported	[56], pp. 71–76
Triticum aestivum	IMC	Cold pretreatment at 4 °C for up to 4 days → surface sterilization → one of the following treatments: (i) 0.4 M mannitol for 5 days at room temperature; or (ii) 0.4 M mannitol at 4 °C for 5–7 days; or (iii) water at 4 °C for 2–4 weeks	MMS4 medium; culture in liquid form or on filter paper with the addition of 4–6 ovaries per dish	MMS5	2000–4000 embryos/spike	40–50	genotype-dependent	[56], pp. 77–81
Triticum aestivum: Gemmeiza 7, Giza 164, Gemmeiza 9, Sids 4, Giza 168, Line-115	AC	Cold pretreatment of donor tillers: 4 °C for 6–8 days (in darkness, immersed in water)	N6 + 2 mg L−1 2,4-D + 1 mg L−1 kinetin + 90 g/L sucrose + 7 g/L agar	MS + 0.5 mg/L NAA + 0.5 mg L−1 kinetin + 30 g/L sucrose	0.67–18.0 (callus induction)	0–10	0–24	[59]
Spring wheat cv. Pavon 76	AC	Cold pretreatment of donor tillers: 4–8 °C for 1–2 weeks	Modified 85D12 medium + PAA + zeatin; 30 °C, 12 h photoperiod	not reported	5–15 (callus induction, genotype-dependent)	1–8	High	[60]
Triticum aestivum: Chris, Pavon 76, WED 202 16-2,Yecora Rojo, Calorwa, Waldro, Wawawai, and winter wheats Capo and Svilena	IMC	Chemical pretreatment of donor tillers: 2-HNA (0–1 g/L) ± 2,4-D (10−6 M) ± BAP (10−6 M), 33 °C for 48–72 h	NPB-99 liquid medium + PAA 1 mg/L + 2,4-D 0.2 mg/L + kinetin 0.2 mg/L + 90 g/L maltose + live ovaries (1 per mL)	190-2	20–50 (microspore reprogramming)	36–100	not reported	[44]
F1 hybrids	AC	Cold pretreatment at 4–6 °C for 5–10 days	Plated onto modified Potato II (P2) medium	190-2	not reported	not reported	not reported	[71]
Triticum aestivum	AC	Cold pretreatment at 4–6 °C for 5–10 days	Potato-2 in the dark and at 28–30 °C	190-2 + NAA (0.5 mg L−1) and kinetin (0.5 mg L−1)	1–15 (callus induction, genotype-dependent)	0–10	High	[72]
Iranian wheat cultivars and four segregating F 3 wheat lines	AC	Cold pretreatment at 4 °C for 7 days	P4 medium supplemented with 5 mg/L kinetin and 2 mg/L 2,4-D, containing 10% Ficoll	190-2 medium supplemented with 0.5 mg/L kinetin and 0.5 mg/L NAA	2–12 (callus induction, genotype-dependent)	0–6	High	[73]
Triticum aestivum	AC	Cold pretreatment of donor spikes at 4–6 °C for 6–10 days	Modified 85D12 liquid medium supplemented with 2 mg/L 2,4-D, 1 mg/L NAA, and 9% sucrose, with or without PEG 8000	R85D12 regeneration medium without plant growth regulators, containing 5 g/L wheat starch	5–20	2–10	High	[65]
Triticum aestivum	AC	Cold pretreatment of donor spikes at 4 °C for 2 days	Liquid P-4 induction medium supplemented with 2,4-D (0.5–4.0 mg/L; optimum ≈1.0 mg/L; high concentrations applied only for 10–15 days)	190-2 medium with 2.0 mg/L 2,4-D for less than 15 days, followed by transfer to medium with 0.2 mg/L 2,4-D	5–25 (callus induction)	2–12	High	[43]
Spring and winter wheat	IMC	Cold pretreatment of donor tillers at 4 °C for 21–28 days (in water, dark)	NPB-99 liquid medium supplemented with 2,4-D (0.2 mg/L), kinetin (0.2 mg/L) and PAA (1.0 mg/L)	B5-5 regeneration medium supplemented with kinetin (1.0 mg/L) and IAA (1.0 mg/L)	not reported	not reported	not reported	[6]
Triticum aestivum	AC	Anthers cultured for 5 days on medium containing 127.5 g/L mannitol + 5.9 g/L CaCl2·2H2O + FHG macronutrients, solidified with agarose (8 g/L)	MS3M medium (90 g/L maltose, 1 mg/L 2,4-D, 1 mg/L BA) supplemented with n-butanol (0.1–0.2%, 5 h) → medium replaced with MS3MF200 (62 g/L maltose + 200 g/L Ficoll, conditioned with 40 ovaries/6 mL, 5 days) → after 10–12 days addition of MS3MF300 (300 g/L Ficoll)	J25-8 regeneration medium, incubation in the dark at 25 °C for 2 days, then transferred to light	3–5× vs. control	3–5× vs. control	not reported	[53]
Triticum timopheevii	AC	Cold pretreatment of donor spikes at 4 °C for 14 days, in water, in darkness	Liquid NPB-99, or C17, or modified 190-2 (+ MS vitamins, biotin 1.5 mg/L, thiamine 0.9 mg/L, 2,4-D 2 mg/L, kinetin 0.5 mg/L); all media supplemented with arabinogalactan 10 mg/L, L-glutamine 500 mg/L, cefotaxime 100 mg/L, maltose 90 g/L, Ficoll 400 10%; media pre-conditioned with 10–15 wheat pistils per plate for 2 weeks at 26 °C in darkness; optional epigenetic treatment: NPB-99 + TSA 0.01 µM	190-4 shoot induction medium + kinetin 1.5 mg/L + NAA 0.5 mg/L, solidified with phytagel 3.5 g/L at 25 °C; red/blue LED light; 16 h photoperiod; subculture every 2 weeks; plantlet elongation on ½ MS salts + double B5 vitamins + activated charcoal 1 g/L (no PGRs)	1	1	75–96	[58]
Pavon, Caramba	AC	0.4 M mannitol with 5.9 g/L CaCl2·2H2O + FHG macronutrients + TSA 0.4 µM, 5 days, 25 °C, dark	OVPCM (MS3M with 1 mg/L 2,4-D; 1 mg/L BA; 200 g/L Ficoll), conditioned with 30 ovaries/10 mL for 5 days → after 10–12 days, add MS3M with 400 g/L Ficoll	J25-8	3–5× vs. control	3–5× vs. control	not reported	[46]
Triticum turgidum L.	IMC	Osmotic pretreatment of anthers: 0.7 M mannitol or 0.7 M mannitol + 30 mM CaCl2·2H2O, solidified with 8 g/L agarose; 5–6 days, 24 °C, dark (30 anthers per plate)	C17 with Ficoll-400 (300 g/L); 1 mg/L 2,4-D; 1 mg/L BAP	J25-8	not reported	not reported	not reported	[52]
Macon, Express, Pavón 76 and Chris	IMC	Osmotic pretreatment: 0.4 M mannitol, 7–10 days at 4 °C	MMS4	RM5 (MMS5 with 5 mg/L ascorbic acid) at 26 °C in the dark for 3 to 4 d, followed by transfer to a full-light growth chamber at 24 °C	not reported	not reported	not reported	[65]
Winter wheat F1 and F2 hybrids	AC	Cold pretreatment of donor spikes at 4 °C for 14 days, in water, in darkness	Liquid AMC c Cu2+ 2.5 mg/L at +29 °C in dark conditions	190-2	not reported	not reported	not reported	[61]
Spring wheat	AC	Cold pretreatment of donor spikes at 4 °C for 7 days, in water, in darkness	C17 with maltose 90 g/L; Gelrite 2.5 g/L; 2,4-D 1.0 mg/L; Dicamba 1.0 mg/L	MS with NAA 0.5 mg/L; Kinetin 0.5 mg/L; agar 0.6%	6.3	2.7	0.6–0.7	[29]
Winter wheat	AC	Cold pretreatment of donor spikes at 4 °C for 7 days, in water, in darkness	C17 with maltose 90 g/L; Gelrite 2.5 g/L; 2,4-D 1.5 mg/L; Kinetin 0.5 mg/L	MS with NAA 0.5 mg/L; Kinetin 0.5 mg/L; agar 0.6%	3.2	0.5	0.3
Chris, Pavon Spring wheat	AC	Cold pretreatment of donor spikes at 4 °C for ≤4 days, in water, in darkness	Modified 85D12 (liquid); 2 mg/L 2,4-D; 50 g/L sucrose (initiation) or 0.26 M maltose as carbon source	Modified 85D12 with 2 mg/L 2,4-D; 50 g/L wheat starch (instead of agar) + 90 g/L sucrose	genotype-dependent	0–10%	genotype-dependent	[74]

Table 1. Overview of donor pretreatments, induction media compositions, and regeneration conditions in wheat anther culture (AC) and isolated microspore culture (IMC).

Genotype	Method	Pretreatment	Induction Medium	Regeneration Medium	Response (%)	Green Shoots (%)	Albino Shoots (%)	Reference
Triticum aestivum: Pavon, Caramba	AC	SM (0.7 M mannitol, 5 days, 25 °C, darkness) → MSMI + 0.2% n-butanol (2–3 h)	MSMIF2 + 6 mature ovaries → MSMIF4 + Ficoll	J25-8	not reported	not reported	not reported	[8]
Triticum spelta L.	AC	Cold pretreatment of donor shoots: 3–4 °C, 14 days	W14mf (300 anthers per Petri dish, 32 °C for 3 days → 28 °C, darkness, 8 weeks)	190-2Cu	not reported	not reported	not reported
Triticale	IMC	Cold pretreatment of spikes: 4 °C, 21 days (darkness, immersed in water)	CIMC with 20% maltose gradient → TSA 1 µM (10 min) → CIMC with PSK-α (0.1 µM) + 6 ovaries per dish	CIMC-4	not reported	not reported	not reported
Triticum aestivum	AC	Cold pretreatment of spikes: 4 °C, ~14 days (in water, wrapped with PVC, darkness)	P-4mf liquid medium (P-4 with 9% maltose; 10% Ficoll; potato extract); 32 °C for 3 days → 28 °C for 5–6 weeks	190-2Cu	not reported	not reported	not reported	[56], pp. 59–64
Triticum aestivum (Central and Eastern European genotypes, n = 32)	AC	Spike selection → surface sterilization with 20% NaOCl for 30 min	W14 medium containing colchicine (40 mg L−1) for 3 days at 29 °C in darkness → followed by W14 without colchicine	190-2 regeneration medium → 190-2 medium without plant growth regulators	6.8–82.3 (avg. 26.3)	1.8–43.8 (avg. 11.4)	not reported	[56], pp. 65–70
Winter wheat	AC	Cold pretreatment at 7 °C for ≤14 days	190-2 medium supplemented with 9% maltose, 2,4-D (1.5 mg L−1) and kinetin (0.5 mg L−1), without NAA; incubation at 28 °C in darkness	190-2	not reported	3.3–5.6	not reported	[56], pp. 71–76
Triticum aestivum	IMC	Cold pretreatment at 4 °C for up to 4 days → surface sterilization → one of the following treatments: (i) 0.4 M mannitol for 5 days at room temperature; or (ii) 0.4 M mannitol at 4 °C for 5–7 days; or (iii) water at 4 °C for 2–4 weeks	MMS4 medium; culture in liquid form or on filter paper with the addition of 4–6 ovaries per dish	MMS5	2000–4000 embryos/spike	40–50	genotype-dependent	[56], pp. 77–81
Triticum aestivum: Gemmeiza 7, Giza 164, Gemmeiza 9, Sids 4, Giza 168, Line-115	AC	Cold pretreatment of donor tillers: 4 °C for 6–8 days (in darkness, immersed in water)	N6 + 2 mg L−1 2,4-D + 1 mg L−1 kinetin + 90 g/L sucrose + 7 g/L agar	MS + 0.5 mg/L NAA + 0.5 mg L−1 kinetin + 30 g/L sucrose	0.67–18.0 (callus induction)	0–10	0–24	[59]
Spring wheat cv. Pavon 76	AC	Cold pretreatment of donor tillers: 4–8 °C for 1–2 weeks	Modified 85D12 medium + PAA + zeatin; 30 °C, 12 h photoperiod	not reported	5–15 (callus induction, genotype-dependent)	1–8	High	[60]
Triticum aestivum: Chris, Pavon 76, WED 202 16-2,Yecora Rojo, Calorwa, Waldro, Wawawai, and winter wheats Capo and Svilena	IMC	Chemical pretreatment of donor tillers: 2-HNA (0–1 g/L) ± 2,4-D (10−6 M) ± BAP (10−6 M), 33 °C for 48–72 h	NPB-99 liquid medium + PAA 1 mg/L + 2,4-D 0.2 mg/L + kinetin 0.2 mg/L + 90 g/L maltose + live ovaries (1 per mL)	190-2	20–50 (microspore reprogramming)	36–100	not reported	[44]
F1 hybrids	AC	Cold pretreatment at 4–6 °C for 5–10 days	Plated onto modified Potato II (P2) medium	190-2	not reported	not reported	not reported	[71]
Triticum aestivum	AC	Cold pretreatment at 4–6 °C for 5–10 days	Potato-2 in the dark and at 28–30 °C	190-2 + NAA (0.5 mg L−1) and kinetin (0.5 mg L−1)	1–15 (callus induction, genotype-dependent)	0–10	High	[72]
Iranian wheat cultivars and four segregating F 3 wheat lines	AC	Cold pretreatment at 4 °C for 7 days	P4 medium supplemented with 5 mg/L kinetin and 2 mg/L 2,4-D, containing 10% Ficoll	190-2 medium supplemented with 0.5 mg/L kinetin and 0.5 mg/L NAA	2–12 (callus induction, genotype-dependent)	0–6	High	[73]
Triticum aestivum	AC	Cold pretreatment of donor spikes at 4–6 °C for 6–10 days	Modified 85D12 liquid medium supplemented with 2 mg/L 2,4-D, 1 mg/L NAA, and 9% sucrose, with or without PEG 8000	R85D12 regeneration medium without plant growth regulators, containing 5 g/L wheat starch	5–20	2–10	High	[65]
Triticum aestivum	AC	Cold pretreatment of donor spikes at 4 °C for 2 days	Liquid P-4 induction medium supplemented with 2,4-D (0.5–4.0 mg/L; optimum ≈1.0 mg/L; high concentrations applied only for 10–15 days)	190-2 medium with 2.0 mg/L 2,4-D for less than 15 days, followed by transfer to medium with 0.2 mg/L 2,4-D	5–25 (callus induction)	2–12	High	[43]
Spring and winter wheat	IMC	Cold pretreatment of donor tillers at 4 °C for 21–28 days (in water, dark)	NPB-99 liquid medium supplemented with 2,4-D (0.2 mg/L), kinetin (0.2 mg/L) and PAA (1.0 mg/L)	B5-5 regeneration medium supplemented with kinetin (1.0 mg/L) and IAA (1.0 mg/L)	not reported	not reported	not reported	[6]
Triticum aestivum	AC	Anthers cultured for 5 days on medium containing 127.5 g/L mannitol + 5.9 g/L CaCl2·2H2O + FHG macronutrients, solidified with agarose (8 g/L)	MS3M medium (90 g/L maltose, 1 mg/L 2,4-D, 1 mg/L BA) supplemented with n-butanol (0.1–0.2%, 5 h) → medium replaced with MS3MF200 (62 g/L maltose + 200 g/L Ficoll, conditioned with 40 ovaries/6 mL, 5 days) → after 10–12 days addition of MS3MF300 (300 g/L Ficoll)	J25-8 regeneration medium, incubation in the dark at 25 °C for 2 days, then transferred to light	3–5× vs. control	3–5× vs. control	not reported	[53]
Triticum timopheevii	AC	Cold pretreatment of donor spikes at 4 °C for 14 days, in water, in darkness	Liquid NPB-99, or C17, or modified 190-2 (+ MS vitamins, biotin 1.5 mg/L, thiamine 0.9 mg/L, 2,4-D 2 mg/L, kinetin 0.5 mg/L); all media supplemented with arabinogalactan 10 mg/L, L-glutamine 500 mg/L, cefotaxime 100 mg/L, maltose 90 g/L, Ficoll 400 10%; media pre-conditioned with 10–15 wheat pistils per plate for 2 weeks at 26 °C in darkness; optional epigenetic treatment: NPB-99 + TSA 0.01 µM	190-4 shoot induction medium + kinetin 1.5 mg/L + NAA 0.5 mg/L, solidified with phytagel 3.5 g/L at 25 °C; red/blue LED light; 16 h photoperiod; subculture every 2 weeks; plantlet elongation on ½ MS salts + double B5 vitamins + activated charcoal 1 g/L (no PGRs)	1	1	75–96	[58]
Pavon, Caramba	AC	0.4 M mannitol with 5.9 g/L CaCl2·2H2O + FHG macronutrients + TSA 0.4 µM, 5 days, 25 °C, dark	OVPCM (MS3M with 1 mg/L 2,4-D; 1 mg/L BA; 200 g/L Ficoll), conditioned with 30 ovaries/10 mL for 5 days → after 10–12 days, add MS3M with 400 g/L Ficoll	J25-8	3–5× vs. control	3–5× vs. control	not reported	[46]
Triticum turgidum L.	IMC	Osmotic pretreatment of anthers: 0.7 M mannitol or 0.7 M mannitol + 30 mM CaCl2·2H2O, solidified with 8 g/L agarose; 5–6 days, 24 °C, dark (30 anthers per plate)	C17 with Ficoll-400 (300 g/L); 1 mg/L 2,4-D; 1 mg/L BAP	J25-8	not reported	not reported	not reported	[52]
Macon, Express, Pavón 76 and Chris	IMC	Osmotic pretreatment: 0.4 M mannitol, 7–10 days at 4 °C	MMS4	RM5 (MMS5 with 5 mg/L ascorbic acid) at 26 °C in the dark for 3 to 4 d, followed by transfer to a full-light growth chamber at 24 °C	not reported	not reported	not reported	[65]
Winter wheat F1 and F2 hybrids	AC	Cold pretreatment of donor spikes at 4 °C for 14 days, in water, in darkness	Liquid AMC c Cu2+ 2.5 mg/L at +29 °C in dark conditions	190-2	not reported	not reported	not reported	[61]
Spring wheat	AC	Cold pretreatment of donor spikes at 4 °C for 7 days, in water, in darkness	C17 with maltose 90 g/L; Gelrite 2.5 g/L; 2,4-D 1.0 mg/L; Dicamba 1.0 mg/L	MS with NAA 0.5 mg/L; Kinetin 0.5 mg/L; agar 0.6%	6.3	2.7	0.6–0.7	[29]
Winter wheat	AC	Cold pretreatment of donor spikes at 4 °C for 7 days, in water, in darkness	C17 with maltose 90 g/L; Gelrite 2.5 g/L; 2,4-D 1.5 mg/L; Kinetin 0.5 mg/L	MS with NAA 0.5 mg/L; Kinetin 0.5 mg/L; agar 0.6%	3.2	0.5	0.3
Chris, Pavon Spring wheat	AC	Cold pretreatment of donor spikes at 4 °C for ≤4 days, in water, in darkness	Modified 85D12 (liquid); 2 mg/L 2,4-D; 50 g/L sucrose (initiation) or 0.26 M maltose as carbon source	Modified 85D12 with 2 mg/L 2,4-D; 50 g/L wheat starch (instead of agar) + 90 g/L sucrose	genotype-dependent	0–10%	genotype-dependent	[74]

In addition to the overall protocol design, the choice of basal culture medium is a critical determinant of the androgenic response and subsequent plant regeneration. The most commonly used basal medium formulations for wheat androgenesis via anther culture and isolated microspore culture include CIMC [8], N6 [75], MS [76], MMS4 and MMS5 [57], MSMI, MSMIF2, and MSMIF4 [8], C17 [77], 190-2 [78], P4 [79], NPB 99 [80], W14 [81], J25-8 [82], and 85D12 [83] (Supplementary Materials Table S1).

Thus, DH technologies are most valuable in breeding programs that require the rapid production of fully homozygous lines, accelerated fixation of valuable allelic combinations, and integration of molecular tools into breeding pipelines. More broadly, it has been shown that haploid approaches can be used not only to accelerate homozygosity fixation but also to optimize individual stages of hybrid breeding, including accelerated cytoplasmic transfer and reduced introgression time compared to traditional backcross schemes [84]. Simultaneously, even in crops where androgenesis has historically played a key role, the operational feasibility of the method is becoming increasingly important, and simpler or scalable systems may gradually displace classical androgenesis-based approaches from routine breeding practices [85]. This indicates that the practical significance of DH technologies in modern breeding programs is determined not only by their biological efficiency but also by their reproducibility, technological convenience, and possibility of integration into real breeding schemes.

In addition to biological and technological constraints, DH technology also has conceptual limitations related to the genetic architecture of complex traits. The rapid production of completely homozygous lines is highly effective for fixing alleles and analysing additive genetic effects; however, complex traits in wheat are influenced not only by additive components, but also by gene × gene and genotype × environment interactions, both of which can make substantial contributions to phenotypic variation and breeding progress [86]. Recent studies further indicate that non-additive effects, particularly additive-by-additive epistasis, may represent a major part of the genetic variance in wheat, and can even substantially exceed additive variance estimates when both components are modelled simultaneously. At the same time, accurate partitioning of additive and epistatic effects remains difficult because these components are often not orthogonal and cannot be estimated independently with high reliability [87]. Moreover, higher-order interactions are themselves biologically relevant, and failure to account for them may lead to underestimation of genetic effects and incomplete representation of trait architecture [88]. Therefore, although DH populations are highly valuable for rapid fixation, mapping, and selection, they do not necessarily capture the full complexity of traits strongly influenced by non-additive interactions, heterosis-related components, or environment-dependent genetic effects.

3. DH Production via Wheat × Maize Hybridisation

In 1984, Zenkteler and Nitzsche first reported that the pollination of hexaploid wheat with maize frequently results in embryoid formation [89]. This finding has attracted considerable interest among breeders, and Laurie and Bennett of the Plant Breeding Institute (Cambridge, UK) initiated systematic studies to verify these observations. Cytological analyses revealed that maize pollen germinated normally and penetrated the wheat embryo sac, where the wheat egg cells were fertilised by the maize sperm nucleus. A hybrid zygote containing 21 wheat and 10 maize chromosomes was formed. However, such hybrid zygotes are unstable because maize chromosomes fail to migrate properly to the spindle poles, likely owing to the progressive loss of centromere activity. Consequently, maize chromosomes are rapidly eliminated during the first mitotic divisions, ultimately leading to the formation of a haploid embryo carrying only the 21 wheat chromosomes [90,91,92,93]. However, the primary mechanisms and genes responsible for chromosomal elimination remain unclear. Several hypotheses have been proposed, including asynchrony in nucleoprotein synthesis, lack of synchrony between parental cell cycles, and parent-specific centromere inactivation [94].

The wheat × maize wide hybridisation method (the “Zea method”) is based on the pollination of emasculated wheat spikes with maize pollen, followed by the formation of a hybrid zygote and subsequent elimination of maize chromosomes [7,24,94,95,96]. In this method, donor plants are grown under controlled greenhouse conditions; the wheat spikes are emasculated prior to anthesis and pollinated with maize, after which auxin treatment is applied 24 h later; finally, approximately two weeks after pollination, the immature embryos are excised and cultured in vitro [8]. A schematic overview of the main stages of DH production in wheat via wheat × maize hybridisation is shown in Figure 2.

3.1. Haploid Production

In most laboratory protocols, six main steps are involved in haploid production: emasculation of wheat flowers; pollination with maize pollen; hormonal treatment to support embryo/seed development; embryo rescue; in vitro regeneration of haploid plants; and chromosome doubling to obtain fertile DH lines [94,95,96]. Post-pollination treatments reported in the literature include the immediate culture of pollinated spikes for three weeks; repeated application of 0.5 mg/L 2,4-D to pollinated spikes for 2–3 weeks; single or double injection or spraying with 100 mg/L 2,4-D into the internode or spikelets; treatment of flowers 24–30 h after pollination with a mixture of an auxin (picloram, 2,4-D, or 2,4,5-T) and 6-benzylaminopurine (6-BA) or with a combination of 2,4-D and GA₃; and application of dicamba (3,6-dichloro-o-anisic acid) solution [96].

Several studies have investigated the effects of hormonal treatments and crossing techniques on haploid production in wheat. Among the controllable factors, post-pollination auxin treatment has been considered the most effective in several studies. Typically, 2,4-D is used; however, it has been shown that substitution or optimisation of the auxin type and application method can substantially modify haploid yield at early developmental stages. For example, in durum wheat experiments, replacing 2,4-D with dicamba increased the number of DH plants per spike [97]. Changing the concentration of 2,4-D (20–100 mg/L) did not affect the efficiency of wheat × maize crosses; however, adding 100 mg/L AgNO₃ to 50 mg/L 2,4-D increased the embryo formation rate from 16.1% to 20.3%. The simplification of the crossing technique (pollination before anthesis, followed by immersion of spikes in a 2,4-D solution) reduces labour requirements and improves overall efficiency [98].

In durum wheat, DH induction efficiency depends mainly on the type of hormone applied; replacing 2,4-D with 50 mg/L dicamba increased the DH yield from 0.2 to 1.3 per spike (by increasing the number of developed kernels) and improved embryo germination [97]. Juzoń et al. (2022) examined the effects of different concentrations of 2,4-dichlorophenoxyacetic acid on DH production efficiency in oats crossed with maize via wide hybridisation; 29 oat genotypes were emasculated, pollinated with maize pollen, and treated with 2,4-D at concentrations of 50 and 100 mg/L [99]. Increasing the auxin concentration to 100 mg/L significantly enhanced haploid plant regeneration frequency and DH line production [99].

Khan et al. (2025) pollinated emasculated wheat spikes with maize pollen, followed by the early elimination of maize chromosomes in the hybrid zygote, in vitro culture of rescued embryos, and artificial chromosome doubling, demonstrating the application of this methodology for the rapid production of homozygous lines suitable for breeding and genetic analysis [100]. In bread wheat, DH lines were obtained by crossing F₁ hybrids with maize, followed by chromosomal doubling using colchicine. Of the 421 pollinated florets, 340 developed green parthenocarpic kernels; 70 embryos were rescued, 35 germinated, and 10 haploid plants were obtained. After colchicine treatment, eight of the nine plants survived, and one plant produced seeds as a DH plant [101].

The protocols emphasise the strong influence of donor plant health and stable greenhouse conditions, as stress caused by diseases, pests, or drought markedly reduces their efficiency. Any deterioration in the physiological status of plants negatively affects seed set, embryo formation, and the viability of early hybrid embryos [95,97]. Moreover, growing conditions, pollen sources, and careful handling during manipulation are critical factors influencing the success of haploid and DH production [96,97].

3.2. Advantages

The wheat × maize system, one of the most reproducible and efficient approaches for producing DH wheat, outperforms androgenesis for most genotypes, including lines recalcitrant to anther culture. Its key advantages include the high stability of the results, reduced genotype dependence, and the ability to rapidly obtain fully homozygous lines within 1–2 years [24,25,95,96,102]. These advantages have led to the widespread adoption of this technology. Since the successful establishment of wheat × maize hybridisation, the method has become a standard tool for developing bread and durum wheat cultivars, as well as populations for genetic mapping, and in recent years, many new genotypes have been generated using this approach [96].

The high efficiency of the system is also associated with its lower genetic specificity compared to the Hordeum bulbosum method or androgenesis, the absence of gametoclonal variation, and a simplified workflow, making this technique less labour-intensive from a biotechnological standpoint. In this approach, wheat donor plants are typically grown under controlled greenhouse and growth chamber conditions [25,98]. Unlike androgenesis-mediated systems, the wide hybridisation method has been characterised in several studies by its higher efficiency in producing green haploid plants, lower genotypic dependence, and the absence of albinism, making it an important alternative for DH production in wheat. However, this approach is not completely universal, as the efficiency of embryoid formation and plant regeneration also depends on the wheat and maize genotypes, and the protocol itself includes several mandatory steps, such as controlled pollination, auxin treatment, embryo rescue, and subsequent chromosome doubling [102]. Therefore, androgenesis and wide hybridisation systems should be considered complementary rather than interchangeable approaches, the choice of which is determined by the genotype of the source material, the available technical base, and the objectives of the breeding program.

Additional comparative data show that methods for producing DH in wheat differ not only in terms of regeneration efficiency but also in the genetic robustness of the resulting lines. A direct comparison of anther culture, corn pollination, and SSD revealed that segregation distortions occur significantly more frequently in androgenetically derived DH lines, whereas populations obtained by maize pollination are characterised by a significantly lower probability of segregation distortion and loss of desirable genotypes. This is partly explained by the fact that in anther culture, several plants of the same genotype can develop from a single callus, which represents an additional limitation of this method [103].

Comparative studies have shown that the differences between methods for producing DH in wheat concern not only the efficiency of induction but also the properties of the resulting material. Thus, in a direct comparison of lines obtained by maize pollination and anther culture, both systems provided high genetic stability and generally similar field productivity indicators, comparable to the corresponding SSD populations [104]. Therefore, the relative value of these methods in breeding should be assessed not only by induction efficiency but also by the quality and stability of the derived lines. Overall, the wheat × maize system is the most universal and effective tool for the accelerated production of wheat DH lines, making it a cornerstone of modern wheat breeding.

3.3. Major Limitations

Although wheat × maize hybridisation remains an effective method for producing DH wheat, it has several significant drawbacks that limit its widespread implementation. First, the technology is labour-intensive, as it involves multiple manual operations, and its major constraints are associated with the success of kernel and embryo formation, efficiency of embryo rescue and plant regeneration, and chromosome-doubling procedures [94,95]. Despite significant progress, the cost of this method remains high because of the need for specialised growth conditions, hormone treatment, and multiple in vitro steps [98]. An additional limitation is the lack of synchronisation between the flowering times of wheat and maize in regions where winter wheat is grown, because these crops are typically cultivated in rotation. Under natural conditions, their phenologies do not overlap, making large-scale haploid induction using maize pollen impossible [94].

Even when pollination is successful, the major biological barrier to wide hybridisation is abnormal endosperm development, which leads to seed abortion. Therefore, embryo rescue is considered an essential component of this technology, and optimisation of the embryo age at excision, culture medium composition, and culture conditions can substantially increase the recovery of regenerants [24,25,94,95]. Moreover, although this approach is generally considered less genotype-dependent than androgenesis, its effectiveness is still determined by both the wheat and maize genotypes and their interactions. Furthermore, the individual stages of the process, including pseudokernel formation, embryoid formation, and regeneration, are not completely coordinated, indicating a complex genetic basis for the effectiveness of wide hybridisation. Practically, this means that the effectiveness of the method can be improved by selecting more effective maize pollinators [105]. Alternative systems, such as microspore culture in maize, remain limited owing to strong genotype dependence; even under optimised protocols, elite lines either fail to respond or exhibit extremely low embryogenic competence [67]. Finally, the wheat × maize system has two critical biological constraints: low ratios of embryos to embryoless kernels and the absence of an endosperm in hybrid seeds, leading to embryo abortion if the seeds are left on the maternal plant. Overcoming these limitations requires the application of a set of post-pollination treatments, including hormonal applications and specific cultivation regimes [96].

Importantly, current views on the genetics and variability of haploid induction in maize emphasise the substantial variation among lines and hybrids, as well as strong sensitivity to environmental conditions, factors that indirectly explain why the correct choice of pollen donors and the stable production of high-quality pollen are critical in wheat × maize hybridisation [106]. Despite its high haploid induction efficiency and relative genotype universality, the wheat × maize method remains technologically complex, expensive, and biologically constrained.

4. Alternative Wide Hybridisation Systems: Imperata cylindrica × Hordeum bulbosum

4.1. Imperata cylindrica

In addition to the wheat × maize system, alternative wide hybridisation approaches have been proposed for haploid production in wheat, based on the use of other pollinator species capable of inducing uniparental chromosomal elimination. Among these, the most extensively studied systems involve the use of Hordeum bulbosum and Imperata cylindrica [92,94,95,96].

Imperata cylindrica (2n = 20) (cogon grass; Poaceae) is a perennial wild grass, flowering synchronously with wheat and triticale, and available in most wheat-growing regions, making it a convenient and reliable pollen source for haploid induction. Chromosome elimination mediated by I. cylindrica is considered an effective alternative to the wheat × maize system because of the low genotype dependency of the pollen donor, the absence of a requirement for greenhouse-grown pollen sources, the higher frequencies of haploid embryo and DH line production, reduced costs, and a lower risk of somaclonal variation [107,108,109]. Cytological studies have shown that in the wheat × I. cylindrica system, the endosperm does not develop, and Imperata chromosomes are eliminated as early as the first zygotic division, thereby enabling the formation of seeds containing haploid embryos [110].

Practical strategies to enhance the efficiency of this system include the selection of highly responsive parental genotypes, in vivo application of colchicine (0.02%), use of a morphological marker (for the absence of an endosperm) to identify embryo-containing pseudoseeds, and pollen conservation. Storage of pollen at −20 °C allows the application of this method in regions where I. cylindrica does not naturally occur and extends the DH breeding season by at least one month [110]. A schematic overview of the principal steps involved in DH production via the Imperata cylindrica × Hordeum bulbosum system is presented in Figure 3.

Comparative experiments demonstrated that I. cylindrica markedly outperformed maize in terms of embryo formation frequency (18.39% vs. 4.08%), with the Ic-ye genotype identified as the most efficient pollen donor. Among maize lines, HPMC-14, HPMC-53, HPMC-60, and HPMC-64 have been recognised as promising tester lines with positive general combining ability (GCA) for haploid induction traits [111]. These results are consistent with field-based comparisons of the wheat × maize and wheat × I. cylindrica systems, in which crossing with Imperata provided higher PFF, EFF, HRF, and HFF values. Moreover, line × tester and GCA analyses confirmed the significant effects of both the parental genotype and the superiority of the Ic-ye donor [112].

Based on protocol optimisation, the maximum yield of haploid plants was achieved by injecting a mixture of 50 mg/L 2,4-D + 50 mg/L dicamba, rescuing the embryos 24 days after pollination, and culturing them on ½ MS medium; AgNO₃ was not essential for embryo induction; and finally, the optimal colchicine concentrations for chromosome doubling were in the range of 0.08–0.075% [113]. The applied value of this technology for the targeted development of resistant lines has been demonstrated by the DH-1 line obtained from the cross F₁ (HS 542/China 84-40,022) × I. cylindrica. This line exhibited resistance to most races of stripe and leaf rusts, and molecular analysis confirmed the presence of Lr34, Lr26, and Lr32, where Lr34 provides durable adult-plant resistance, Lr26 is associated with the 1BL.1RS translocation, and Lr32 confers leaf rust resistance. These results indicate that DH-1 is a promising source material for resistance breeding [108].

Backcross generations have been demonstrated to broaden recombination and accelerate the development of mapping and breeding populations. In triticale × wheat derivatives, BC₁F₁ and BC₁F₂ statistically outperformed F₁–F₃ in terms of the key haploid induction parameters. BC₁F₂ exhibited a higher frequency of pseudo-seed formation, whereas BC₁F₁ provided a higher yield of haploid embryos with comparable regeneration efficiency [114]. Similarly, the evaluation of a wide range of generations (F₁, F₂, BC₁F₁–BC₁F₅) in triticale × wheat and wheat × rye derivatives has revealed that embryo formation and regeneration are largely genetically determined, and that BC₁F₁ and BC₁F₂ represent the most efficient generations [115]. In experiments using a broad genetic panel (16 hexaploid, 7 tetraploid, and wheat × rye-derived genotypes), I. cylindrica also exhibited the highest efficiency as a pollen donor; 19 new first-generation DH lines, with seed productivity comparable to that of the original cultivars, were obtained, thus confirming the maintenance of fertility in the derived lines [13].

Finally, the combining ability of the parental components has a significant effect on hybrid productivity and the success of DH induction. Line HS542 exhibited high general combining ability for grain, and the tester KLE/BER/2*FL-8/DONSK-POLL for thousand kernel weight. Certain combinations were characterised by strong specific combining ability while simultaneously producing haploid embryos through intergeneric hybridisation [116]. Collectively, these findings confirm that the wheat × I. cylindrica system represents a highly efficient and technologically flexible chromosome elimination platform suitable for both accelerated breeding (including disease resistance) and the rapid development of homozygous and mapping populations in wheat and its derivatives [107,109].

4.2. Hordeum bulbosum

Wild species of the genus Hordeum are of considerable interest for wheat improvement because they harbour valuable adaptive traits, including tolerance to drought and salinity, resistance to pests and diseases, and increased grain carotenoid content. In particular, Hordeum chilense Roem et Schultz. (2n = 2x = 14), endemic to Chile and Argentina, is regarded as a promising donor of such traits for the genetic improvement of Triticum aestivum [117].

A key technological breakthrough in the application of wide hybridisation was the discovery of uniparental chromosome elimination, which leads to the formation of maternal-type haploid embryos (via gynogenesis). The first convincing evidence was obtained in the Hordeum vulgare × H. bulbosum system, in which H. bulbosum chromosomes are selectively eliminated during hybrid seed development, resulting in the production of haploid barley plants. This approach was later adapted for haploid production in wheat, and the high frequency of haploid formation through chromosome elimination following the pollination of T. aestivum with the tetraploid H. bulbosum was first demonstrated by Barclay [118].

Despite its effectiveness, the use of H. bulbosum as a pollen donor for haploid induction in wheat is limited by barriers to genetic crossability. The major determinants are the dominant crossability inhibitor genes Kr1 and Kr2, located on chromosomes 5B and 5A, respectively, and Kr3 (5D) and Kr4 (1A). These genes act mainly by restricting the penetration of H. bulbosum pollen tubes into the wheat ovary, thereby reducing the seed set or causing complete incompatibility in the wheat × H. bulbosum system. Their effect appears to be stronger at the stage of pollen tube entry into the ovary tissues than at the level of pollen germination itself or the rate of pollen tube growth toward the micropyle. Consequently, many wheat cultivars, particularly those of European origin, exhibit low crossability with H. bulbosum. In contrast, the cultivar Chinese Spring is recognised as one of the most crossable genotypes because it carries recessive alleles that permit successful pollen tube penetration and fertilisation [62]. Importantly, substantial genetic variability exists within tetraploid H. bulbosum, enabling it to set seeds on wheat and produce seeds even from genotypes carrying strong crossability inhibitors. This variability opens up the possibility of partially overcoming such incompatibility by selecting H. bulbosum genotypes with high crossability [114].

In a broader context, chromosome elimination and haploid formation are recognised as widespread phenomena within the genus Hordeum, occurring not only in the H. vulgare × H. bulbosum combination but also in several other interspecific crosses in which the elimination of H. bulbosum chromosomes results in the formation of haploids of the other parent. However, despite the production of numerous interspecific and intergeneric hybrids involving Hordeum, targeted gene transfer remains relatively rare because the hybrids are frequently sterile, even after chromosome doubling [119].

The limitations of the wheat × H. bulbosum system have stimulated the development of alternative wide-crossing schemes for haploid production. In particular, the wheat × maize system enables fertilisation independently of the presence of Kr crossability inhibitors, while maize chromosomes are rapidly eliminated, necessitating early embryo rescue; a critical factor is post-pollination treatment with 2,4-D, which stimulates embryo development. This approach has been recognised as more universal for a wide range of wheat cultivars and hybrids (including durum wheat), in which Kr1 and Kr2 do not restrict crossing with maize [96,120]. In addition to maize, related taxa such as teosinte and Tripsacum, as well as sorghum and pearl millet, have also been evaluated as potential pollen donors; however, in practice, the wheat × maize and wheat × pearl millet systems are considered more stable because haploid embryo formation is less affected by genotype in these combinations [120].

Additional evidence for the stochastic nature of chromosome elimination was obtained by examining paternal chromosome retention in the progeny of a wheat × barley hybrid population. Using the T. aestivum system, plants were obtained after emasculation and pollination, injected with 2,4-D (100 ppm) one day after pollination, and the embryos were rescued 14–16 days after pollination. Analysis of 210 plants using chromosome-specific DNA markers for all seven barley chromosomes (1H–7H) revealed comparable proportions of haploids and true hybrids (20.5% and 19.5%, respectively), as well as a wide spectrum of hypoploids carrying 1–6 additional barley chromosomes, without preference for any particular chromosome. The absence of a statistically significant “preferential” elimination of individual chromosomes (1H–7H) indicates that the retention or loss of the paternal genome was largely random, which is important for planning introgression strategies and developing additional lines in cereal breeding [121].

In summary, there are many Triticum × Hordeum wide hybridisation systems that exhibit high potential for the accelerated production of homozygous lines through haploidisation and introgressive breeding. Nonetheless, the practical applicability of this approach is determined by genetic crossability barriers (the Kr loci) and the characteristics of chromosome elimination, whereas alternative systems–such as the wheat × maize system–provide greater universality with respect to wheat genotypes and are therefore considered more stable for broad implementation in breeding programs [96,120].

5. Haploid and DH Systems as Platforms for Omics Analyses in Wheat

Haploid and DH systems serve as convenient platforms for wheat omics studies, providing a stable genetic background and high accuracy for comparing molecular profiles across different conditions and genotypes. The main biotechnological approach for DH production is microspore embryogenesis, which entails the stress-induced reprogramming of immature male gametophytes, leading to embryoid formation and the subsequent regeneration of DH plants after genome doubling. Transcriptomic and small RNA analyses of the early stages of induction have revealed that cold stress triggers extensive reprogramming of gene expression from the uninucleate microspore stage to the first cell division. De novo assembly identified 29,388 contigs (20,224 transcripts), confirming the existence of a specific embryogenesis-related transcriptional program and the regulatory role of small RNAs [122]. The competence for such reprogramming is determined by the developmental stage: microspores and bicellular pollen display a more “sporophytic-like” expression profile, whereas mature pollen is characterised by the enrichment of pollen-specific transcripts associated with germination and pollen tube growth. This explains the critical importance of precise explant-stage selection in androgenesis and the interpretation of omics data [45].

The complete homozygosity of DH lines provides fundamental advantages in functional genomics. Compared with early selfing generations, DH lines offer uniformity and reproducibility in gene expression analyses across environments, enabling proper testing of landraces and open-pollinated varieties, and the haploid phase accelerates the elimination of deleterious alleles. The use of DH lines also facilitates variety registration and legal protection by meeting the criteria of novelty, distinctiveness, uniformity, and stability [9]. At the level of future strategic development, the DH approach provides a bridge between pan-omics and breeding, and the integration of transcriptomics, epigenomics, and regulatome layers (time-series and multi-tissue profiles) enables the construction of regulatory networks and genotype–phenotype maps for precision breeding. Moreover, advances in biotechnology, including genome editing (e.g., editing of CENH3 to induce paternal haploids), have translated omics discoveries into accelerated validation and practical improvements in wheat [123].

Omics studies have further demonstrated that epigenetic regulation can be deliberately exploited to enhance the efficiency of DH technologies. Treatment with the demethylating agent 5-azacytidine in combination with cold induction increased microspore embryogenesis in triticale in a genotype-dependent manner and was accompanied by proteomic signatures indicative of a switch in energy metabolism, activation of stress-protection systems, and elevated expression of embryogenic markers (TaTPD1-like, SERK2, and GSTF2), supporting the role of partial hypomethylation in triggering the embryogenic pathway [124]. In a broader applied context, multi-omics approaches have been used to dissect stress tolerance and quality traits. Under drought conditions, the integration of transcriptomics, proteomics, and metabolomics has been used to reveal regulatory networks (TF/miRNA/ABA signalling) and metabolic biomarkers (osmoprotectants and antioxidants), thereby supporting marker-assisted selection (MAS), QTL pyramiding, and genome editing strategies [125]. A multi-omics atlas of combined abiotic stresses has revealed that double and triple stress combinations cause the greatest yield penalties and elicit unique transcriptomic responses, whereas network analyses have identified hub regulators (e.g., TaWRKY33) as promising targets for breeding tolerance to multiple stresses [126]. Finally, multi-omic analysis of the amphidiploid T. durum × Aegilops sharonensis has demonstrated that interspecific hybridisation and polyploidisation profoundly reprogram carbon and amino acid metabolism and the balance of storage polymers in the grain, thus underscoring the value of haploid/polyploid platforms for the functional genomics of quality traits and the targeted utilisation of wild alleles [127]. Accordingly, DH-based mapping populations in wheat have been used to localize loci and QTLs for a wide range of agronomically important traits, including grain yield, grain weight, grain morphology-related traits, grain number, spike-related traits, plant height, tiller number, harvest index, biomass yield, days to heading, flowering and maturity, grain filling duration, pre-harvest sprouting tolerance, disease resistance, including Fusarium head blight, tolerance to drought and heat stress, nitrogen use efficiency, and root system architecture. Together, these examples illustrate that DH populations are not only a convenient experimental resource for genetic analysis but also an important bridge between omics-driven trait dissection and the identification of breeding-relevant loci and markers in wheat [128,129,130,131].

6. DH Populations in QTL Mapping

DH wheat has become one of the most convenient resources for QTL mapping and association analysis, because it exhibits complete homozygosity in the first generation, thus ensuring accurate phenotyping, reproducibility of locus effects, and stability of trait evaluation across multifactorial environments [1,132]. This is particularly important for examining quantitative traits related to yield, yield components, and adaptive responses, in which the contribution of genotype × environment interactions is often comparable to the genetic effect. In practice, the DH approach shortens breeding cycles and accelerates the accumulation of favourable alleles, while simultaneously generating convenient panels for QTL/MTA (marker–trait association) discovery and the subsequent deployment of markers in MAS/genomic selection strategies [3,19,133].

DH populations have been widely used to map the QTLs controlling yield and its components (biomass, spike architecture, and spikelet fertility), as well as tolerance to lodging, drought, and pre-harvest sprouting. DH populations provide higher power for QTL detection than segregating populations, particularly for traits with moderate-to-high heritability [134,135,136]. An additional advantage of the DH approach is that it can be used to identify stable QTLs that are reproducible across years and environmental conditions, which is particularly important for breeding adaptive cultivars. Such loci are often used as the basis for the development of molecular markers and their subsequent application in MAS [137,138].

In GWAS and QTL studies, DH lines are used in two complementary formats. The first involves association panels built from elite breeding material or regional cultivars, with a focus on “applied” variation that is already relevant to improvement programs. Notably, elite populations developed by breeders can retain sufficient genetic variability, with relatively weak population stratification but pronounced family structure; therefore, statistical models that account for relatedness (kinship), such as mixed linear models, are the most effective for reducing false associations [139].

Within this framework, a modern U.S. winter wheat DH panel (264 lines) was developed and evaluated for grain and biomass traits, and high-density SNP data were generated. After quality control, 59,482 markers were retained for further analysis. Association mapping using multiple statistical models, with stringent correction for multiple testing and large-scale phenotyping with detailed genomic annotation, has identified genomic regions associated with yield components and biomass [140]. Similarly, in collections of cultivated varieties (e.g., 102 Argentine hexaploid cultivars genotyped with a 35k SNP array), GWAS combined with haplotype block analysis revealed multiple regions associated with adaptation and yield components (including TGW, grain number, plant height, and heading date) and highlighted promising traits, such as fruiting efficiency, that are suitable for MAS under specific agro-ecological conditions [141].

The second format is classical QTL mapping in biparental DH populations, where the genetic interpretation of the effects is more straightforward, and the identified loci are convenient for validation and marker development. Numerous small-effect loci and stable “major” QTLs have been reported for productivity traits and grain morphometrics. For example, in the Superb × M321 DH population produced via maize pollination, genotyping with a 55k SNP array enabled the construction of a dense genetic map (15,001 SNPs; 2209.64 cM) and the detection of nine QTLs for TGW, among which two strong and stable loci (on 4DS and 6AL, respectively) explained a substantial proportion of phenotypic variance across environments and are considered valuable resources for MAS in high-yield breeding [44]. For yield components and spike architecture, studies in two Argentine DH populations (Baguette Premium 11 × BioINTA2002 and Baguette 19 × BioINTA2002) identified several chromosomal “hotspots” with overlapping confidence intervals for multiple traits (grain weight, grain number, and spike fertility-related traits). The presence of such pleiotropic or closely linked control regions suggests that entire regional “allelic packages”, rather than single QTLs, can be considered promising targets for pyramiding in breeding programs [135]. Importantly, DH genetic backgrounds greatly facilitate the analysis of pleiotropy and trade-offs among yield components (such as grain length and grain number), because they minimise within-line segregation and reduce background “noise” [132].

For stress-adaptive traits, DH populations are particularly valuable because they enable highly reproducible phenotyping under contrasting water and temperature regimes and allow direct comparison of QTL effects across environments. In the DH population Plainsman V. × Cappelle–Desprez, evaluated over three years under drought stress imposed by reduced irrigation and well-watered control conditions, 54 QTLs controlling 10 traits were identified. A subset of loci was detected under both drought stress and well-watered conditions (e.g., a TGW QTL on chromosome 1A), whereas QTLs for grain yield were predominantly expressed under drought stress conditions [134]. In a parallel line of research, resistance to pre-harvest sprouting, which is tightly associated with seed dormancy, has been mapped in DH populations as a trait that is strongly dependent on “inducing” environments. In two South African DH populations, three major additive-effect QTLs (on chromosomes 5 B and 7 B) were identified, although their reproducibility was observed specifically under environmental conditions favourable for pre-harvest sprouting [137]. A more refined dissection of seed dormancy was performed in the DH population Yangmai16 × Zhongmai895; stable QTLs were identified on chromosomes 3A and 3D (overlapping with QTL for grain colour) and at a potentially novel locus, 3D.2. For this locus, a KASP marker was developed and validated across a broad panel, illustrating the typical “QTL → diagnostic marker → deployment in MAS” pipeline [136].

Complex traits that require simultaneous consideration of agronomic trade-offs deserve particular attention. A representative example is the lodging resistance mediated by stem biomechanics. In the DH population AC Cadillac × Carberry, multifactorial QTL mapping across 16 traits revealed clusters of loci (hotspots) on several chromosomes and extensive co-localisation of stem-related QTLs with QTLs for grain traits and heading date. The Rht-B1b allele acts as a major driver of phenotypic variation, whereas certain genomic regions (e.g., on chromosome 6A) simultaneously increase stem-wall thickness, bending moment, and thousand-kernel weight, making them especially attractive targets for ideotype breeding and for the “safe” improvement of lodging resistance without compromising yield [138].

Currently, with the development of high-throughput genotyping platforms (SNP arrays, genotyping-by-sequencing, and sequencing-based approaches), DH populations are increasingly integrated into GWAS and QTL + GWAS strategies. The complete homozygosity of DH lines substantially reduces genotyping errors and simplifies the interpretation of allelic effects, which is particularly important when analysing large marker datasets [19,139].

Utilising DH material enables the effective integration of high-density SNP genotyping data with multi-trait phenotyping, including yield components, stress tolerance, and morphological characteristics. In several studies, DH populations have been used to refine the positions of previously identified QTLs and narrow candidate intervals, thereby facilitating the identification of candidate genes [132,140]. In the context of genomics-assisted breeding, DH lines are increasingly regarded as standard reference populations that provide data comparability across different breeding programs and research centres [8,19].

Methodological advances in wheat genetics have substantially increased the resolution and practical value of QTL analysis, particularly through the transition from sparse marker systems to dense SNP- and GBS-based maps and from simple statistical approaches to linear mixed models that account for relatedness, population structure, and multi-environment data [18,19]. However, the identification of QTLs in DH populations is only the first step toward their practical application. Their breeding value depends on the availability of marker systems suitable for validation, fine mapping, and deployment in selection programs. In this context, recent wheat genomic and multi-omics resources have increasingly integrated mapped loci with SNP arrays, PCR-based markers, and candidate genes, thereby enabling the transition from QTL detection to marker-assisted application. Therefore, SNP genotyping platforms represent a logical extension of DH-based QTL analysis in wheat [129,130,131].

7. DH Wheat in SNP Genotyping: SNP Arrays, KASP Markers, and Targeted Platforms as an Integrated Breeding–Genetic Framework

The fast acquisition of homozygous material is a cornerstone for accelerating bread wheat breeding and increasing the accuracy of the genetic dissection of complex traits. Haploid and DH plants enable the fixation of recombination in a single generation and substantially reduce genotype segregation effects during phenotyping and genotyping. However, the actual impact of DH technology depends on the efficiency with which genomic signals (QTLs/QTNs, introgressions, and functional alleles) are translated into applied marker systems. Consequently, DH-based breeding inherently requires the tight integration of SNP arrays with scalable KASP markers that are properly validated in polyploid wheat genomes [142,143].

The breeding logic of this integrated approach is well illustrated in accelerated selection schemes in which MAS is combined with haplodiploidisation. In the DH population derived from crossing Hidhab and Parula (donors of Lr34, Lr46, and Rht-B1), target alleles for disease resistance and reduced plant height were tracked using KASP markers in parallel with multidimensional phenotypic selection. This identified various ideotype DH lines. Among the 364 lines analysed, genotypes combining earliness, high yield, and the genetically confirmed Lr34 + Lr46 + Rht allele combination were selected. This example demonstrates that DH platforms provide an optimal framework for rapid allele pyramiding while maintaining a high selection intensity and substantially shortening the breeding cycle. Thus, DH materials not only accelerate the production of parental lines but also enhance genotyping efficiency because each allelic decision is fixed in a stable homozygous background, thus simplifying marker interpretation and deployment in breeding [144].

SNP arrays continue to play a central role in genetic determinant discovery, as they offer dense genome coverage and support QTL mapping and GWAS on the scale of thousands of markers. For instance, mapping of heading date, plant height, TGW, and spike length in a recombinant inbred line population (eh1 × Lunxuan987) using the Wheat55K SNP Array enabled the construction of a high-density genetic map (6505 SNPs condensed into 1097 loci) and identification of 37 QTL, including several stable and major-effect loci. A critical step toward practical application was the subsequent development of KASP markers based on RNA-seq data and their validation in an expanded panel of up to 400 lines, which allowed for the narrowing of physical intervals and progression toward candidate genes. This clearly identifies the “SNP array → QTL discovery → KASP validation” workflow as the most efficient, with the DH and inbred panels serving as the most suitable platforms for the fixation and testing of the detected effects in MAS pipelines [145].

Notably, the same “SNP screening → detection of stable signals → conversion to KASP” architecture is now applied not only to classical yield-related traits but also to complex adaptive responses, including heat tolerance. In a multi-environment study of 126 genotypes, multilocus GWAS based on 35 K SNP genotyping identified consensus QTNs for phenology, NDVI, and yield components, with several loci exhibiting pleiotropic effects. Importantly, the practical relevance of these associations was confirmed by the development and independent KASP validation of key SNP markers (e.g., AX-95018072/7A, AX-94946941/6B, and AX-95232570/1B). GWAS signals have thus been translated from the level of genomic statistics to applied tools for selecting stress-tolerant lines, which is particularly critical for DH material, in which stress tolerance and productivity must be evaluated simultaneously in a fixed genetic background [146].

The large-scale implementation of KASP in wheat faces a fundamental constraint imposed by polyploidy. The high homology and paralogy among the A, B, and D genomes create the risk of non-locus-specific amplification, false clustering, and erroneous allele calls. In this context, methodological studies addressing the conversion of chip-based SNPs into KASP assays are of great importance. Using 15 SNPs associated with a root biomass QTL on chromosome 5B as a model, it was shown that standard semi-automated primer design frequently failed to achieve locus specificity. Instead, successful strategies require multi-reference sequence alignment, confirmation of target sequences by Sanger sequencing, manual primer positioning, and mandatory validation of panels that include heterozygotes (including artificially generated sequences). A particularly important conclusion was that “null” alleles observed on SNP arrays can, in some cases, represent hybridisation or clustering artefacts; KASP combined with Sanger sequencing corrects the haplotype structure and increases the proportion of explained phenotypic variance. From an applied perspective, the DH material represents an ideal test platform for such procedures. Its homozygosity reduces cluster ambiguity, and any lack of locus specificity becomes especially evident when compared with biologically expected inheritance patterns [143].

Methods for the accurate detection and monitoring of alien chromatin are urgently required, particularly in pre-breeding programs that exploit introgression from wild relatives. DH introgression lines often carry not only the target segments but also complex recombination events and structural rearrangements, necessitating a high marker density and rigorous marker validation. In DH wheat × T. urartu introgression lines, analysis of marker profiles revealed recombination of introgressed segments with homoeologous chromosomes and probable translocations of alien segments to other chromosomes, while cytogenetic verification by genomic in situ hybridisation (GISH) confirmed the presence of recombinant chromosomes and detected aneuploidies (including loss of chromosome pairs 1D or 5D). This finding is methodologically important, as it indicates that neither markers nor cytogenetics alone can provide a complete picture; therefore, for DH introgression material, the minimal standard is the combined use of molecular markers and GISH [147].

Further progress along this line is associated with the development of chromosome-specific KASP panels based on wild-type donor genomics. For wheat × Amblyopyrum muticum DH lines, whole-genome sequencing of the wild species was performed, and a de novo pipeline was constructed to select SNPs located only in unique regions of the wheat genome; this resulted in the identification of ca. 38,000 SNPs suitable for marker design. Subsequently, hundreds of KASP assays were developed, and after filtering, a final set of 498 evenly distributed chromosome-specific markers was obtained. The application of this panel to 67 DH introgression lines enabled the precise identification of wheat chromosomes that had recombined or been replaced, and the high marker density allowed the detection of small introgressions that could not be detected using multicolour genomic in situ hybridisation (mcGISH). The workflow “wild-donor WGS → SNP discovery in unique regions → chromosome-specific KASP” is therefore emerging as a universal scheme for supporting DH introgression programs [148].

The implementation of functional markers is critically important for DH-based breeding, as these markers maximise “direct causality” in MAS and reduce the dependence on recombination. For example, sequencing of the TaSBEIII-A allele identified a novel nonsynonymous SNP (C/T; Pro → Ser), and a KASP marker was developed and validated across broad collections (in China and Pakistan and at the International Maize and Wheat Improvement Center [CIMMYT]) of diploid and tetraploid wheat. The T allele, which was statistically associated with increased TGW, exhibited positive selection signatures during breeding history. In the context of DH lines, such markers are especially valuable because they allow not only tracking of the QTL background, but also the fixing of favourable alleles in a homozygous state and their rapid advance into elite germplasm [149]. Reviews of KASP technology emphasise that converting functional SNPs/InDels into the KASP format represents one of the most rational strategies for reducing the cost of large-scale selection in wheat, particularly under increasing climatic risks and the need to accelerate the development of adaptive cultivars [142].

Finally, the practical implementation of DH genotyping requires the selection of an optimally targeted SNP platform that balances cost, cluster stability, and design flexibility. In a comparison of TaqMan, KASP, and rhAmp assays for functional SNPs/INDELs, KASP and rhAmp achieved a higher design success rate (49/50) than TaqMan (39/50) and provided superior allele discrimination. Although TaqMan produces more compact clusters, it is substantially more expensive per reaction. rhAmp proved to be the least costly platform, with KASP ranking second, supporting the continued role of KASP as a “workhorse” platform for MAS. Given that DH material typically requires thousands of genotyping reactions for breeding-scale screening, cost becomes a decisive factor [150].

In summary, these findings indicate that modern DH-based wheat breeding should be viewed as an integrated pipeline: SNP chips and high-density SNP arrays support QTL/QTN discovery and genome-wide mapping, and key loci are then converted into KASP markers, with mandatory verification of locus specificity in the polyploid genome and independent validation. In introgression programs, this pipeline is further complemented by cytogenetic verification or panels of chromosome-specific KASP markers developed from donor WGS data. Such an integrated framework ensures the accurate tracking of recombination events, detection of structural abnormalities, and reliable advancement of target segments in DH materials [143,147,148].

8. Integration of Haploid Technologies with Genome Editing in Wheat

In wheat breeding, the practical implementation of genome editing and transformation remains constrained not by CRISPR target design, but primarily by bottlenecks in the biotechnological pipeline–namely, low tissue-regeneration capacity, the strong genotype dependence of in vitro morphogenesis and transformation, and the long time required to obtain stable lines. Against this background, the effectiveness of even highly specific and technologically “simple” CRISPR/Cas systems in wheat culture is dictated by the availability of scalable delivery and regeneration protocols, the quality of the regulatory elements in expression cassettes, and the ability to rapidly fix edited genotypes as homozygous lines suitable for reproducible phenotyping and breeding evaluation [12,151]. Historically, progress in this area has been closely linked to the development of Agrobacterium-mediated transformation and platforms, such as PureWheat. Such work has demonstrated that optimising immature embryo stage co-cultivation regimes and selection conditions can yield high transformation efficiencies, at least in “model” genotypes. However, the transferability of such platforms to commercial cultivars remains limited, and technological solutions are required to reduce the genotype dependence [12].

In this context, increasing the universality of regeneration via the expression of morphogenetic regulators can act as a systemic “unlocking” module for the entire transformation/CRISPR pipeline. For instance, introducing the chimeric factor GRF4–GIF1 dramatically enhanced the regeneration capacity and accelerated the recovery of transgenic plants in wheat, triticale, and rice, with the chimera outperforming the other lines in terms of the expression of GRF4 and GIF1 separately and in combination, as well as other GRF/GIF combinations. Notably, regeneration was induced without the use of exogenous cytokinins, simplifying the protocol and potentially reducing the reliance on antibiotic selection. Importantly, this morphogenetic “accelerator” is not limited to transformation per se: when integrated with CRISPR–Cas9 (targeting Q/AP2L-A5), it enabled the production of fertile edited plants, and the construct was then segregated in the T1 generation. This approach facilitates the isolation of transgene-free edited lines and shifts the technology toward a more breeding-compatible format [152]. Collectively, these findings illustrate a fundamental principle: improving regeneration efficiency and broadening the range of transformable genotypes are decisive prerequisites, rather than auxiliary goals, for scalable genome editing in wheat [12,152].

In parallel with relieving the regeneration bottleneck, it is critical to optimise CRISPR cassette expression, as editing efficiency and mutation profiles both depend on the choice of the promoters driving sgRNA expression, cassette architecture (single vs. dual sgRNAs), and the nuclease employed, as well as on how accurately screening is configured under conditions of frequent mosaicism and polyploidy [12,151]. Agrobacterium-mediated delivery of CRISPR/Cas9 in wheat has been shown to yield heritable mutations across the T1–T3 generations, with an average editing efficiency of ca. 10%. Deletions, including large deletions, are common in wheat and may escape detection using standard PCR screening if the primer design is suboptimal. For example, a 1160 bp deletion in TaCKX2-D1 resulted in an increased grain number per spikelet, highlighting the applied value of deliberately generating large NHEJ-mediated deletions. Simultaneously, CRISPR cassette activity may persist in subsequent generations and generate new mutations, making the fixation of stable lines essential for reliable phenotyping [153]. Regulatory elements that control sgRNA expression have a decisive impact on editing efficiency. Among the OsU6a, TaU3, and TaU6 promoters, TaU3 exhibited the highest activity on a GUS target, while the use of two sgRNAs increased the overall mutation frequency and enabled the recovery of large deletions. Although alternative systems, such as xCas9 and LbCpf1, are also functional in wheat, the best performance is generally achieved using an optimised SpCas9 system. Notably, xCas9 exhibited the highest activity on NGG PAM sites, whereas LbCpf1 exhibited lower efficiency. Practically, such optimisation has enabled the efficient editing of TaMTL and subsequent haploid induction in T1, effectively turning CRISPR into a tool for generating haploid inducer lines for accelerated wheat breeding [154].

The logical continuation of this discussion is the integration of genome editing with HI-Edit and DH technologies to accelerate genotype fixation and the production of reproducible material. Following the earlier application of CENH3-dependent HI in many dicot model systems, it was long considered essential to show that such systems could be applied to cereals, and, in particular, to wheat. Editing of TaCENH3α enabled the generation of lines with paternal HI frequencies of several percent (ca. 7–8% in certain genotypic contexts) and revealed the strong dependence of HI efficiency on the specific allelic configuration of homeologs, together with associated trade-offs, such as reduced seed set, signs of genomic instability, and the possible occurrence of aneuploidy. These observations, which are consistent with the post-zygotic chromosome elimination model of HI, highlight the biological costs associated with the inducer phenotype [155].

Mechanistic studies on mMTL-mediated HI systems support the hypothesis that in edited lines, a late reactive oxygen species burst in pollen may precede DNA damage and potentially trigger chromosome fragmentation or elimination, thereby linking the HI phenomenon to male gametophyte physiology and stress-like signalling cascades. This connection is important for the further engineering of inducers and for managing the side effects on fertility and seed quality [156].

Against this background, the concept of using HI not only for accelerated production of DH material but also as a delivery platform for genome editing in transformation-recalcitrant genotypes appears particularly promising. In HI-Edit and related approaches (e.g., IMGE), Cas9/sgRNA is expressed in the inducer parent, editing occurs in the early zygote/embryo, and the inducer genome is subsequently eliminated, producing transgene-free but edited haploid progeny. In this way, a major regulatory and breeding bottleneck associated with residual transgenicity can be bypassed, and the introduction of edits into elite germplasms can be substantially accelerated [157,158,159].

Simultaneously, the available HI-Edit data indicate that its efficiency is highly dependent on the crop species, HI system employed, and temporal dynamics of paternal genome activation. Although substantial proportions of edited haploids have been reported in maize and Arabidopsis, the efficiencies in wide crosses (e.g., wheat × maize) can be low and sensitive to the Cas9-driving promoter, likely reflecting the rapidity of genome elimination and differences in the temporal window during which the editor can act [157,158]. An important methodological implication is that genetically stable materials for phenotyping often exhibit mosaicism in the primary transformants; therefore, the accelerated fixation of homozygosity via DH technologies is a critical bridge between genome editing and reproducible trait evaluation.

Despite this conceptual potential, the practical implementation of HI-based genome editing systems remains associated with several biological and technological constraints. Although haploid inducer-mediated genome editing (HI-Edit/IMGE) has demonstrated relatively high haploid induction efficiency in maize, typically reaching up to 15–20% depending on the genetic background, the overall performance of this system remains constrained by several biological and technical limitations. Editing efficiencies are generally low, and multiplex editing events occur at very low frequencies, indicating that efficient transmission of gene edits during haploid induction remains a major challenge [159].

In addition, the underlying biological mechanisms of haploid induction, including chromosome elimination and fertilization abnormalities, are still not fully understood and are associated with incomplete genome elimination and instability of the resulting lines [160]. This is consistent with observations of somatic chimerism and unintended introgression of inducer-derived DNA fragments, which may compromise genome integrity [159]. Importantly, the application of these approaches in wheat remains even more limited. Available data indicate that haploid induction efficiency typically ranges from approximately 6% to 15%, while the process is further constrained by strong genotype dependence, technical complexity, and limited reproducibility [161]. Taken together, these findings indicate that, despite its conceptual potential, the HI-Edit framework remains far from a universally efficient and robust tool in wheat breeding and requires substantial optimisation before routine application.

This logic is particularly evident in studies combining multiplex genome editing with DH production for the rapid assembly of panels of isogenic lines. In maize, the coupling of highly multiplexed CRISPR (targeting dozens of genes) with in vivo haploid induction and accelerated chromosome doubling via embryo rescue with chromosome doubling enabled the establishment of large arrays of homozygous allelic combinations suitable for replicated phenotyping within a few generations, which substantially reduced the amount of effort required, avoiding the need for many generations of crossing and intensive genotyping [20]. Although this was demonstrated in maize, it provides methodological support for the argument that, for complex traits and gene families, the “multiplex CRISPR → HI/DH → isogenic line panel” framework represents the most rational experimental architecture. This framework involves multiplex editing of several target loci or homeologs in primary transformants, followed by haploid induction or DH production to rapidly fix edited alleles in a homozygous state. The resulting DH-derived edited lines are then genotyped and sorted into an isogenic panel carrying different combinations of target mutations, which can subsequently be compared phenotypically to dissect the gene family function and complex trait architecture. This architecture is particularly relevant for wheat, owing to its polyploid nature and the need to assemble combinations across homeologs [20,151].

In wheat, however, precise editing must be considered together with two additional issues that directly affect the breeding value of edited materials, namely the persistence of genetically modified residues and the occurrence of unintended genomic changes. In conventional transformation-based systems, especially those relying on Agrobacterium-mediated delivery, edited plants may initially retain integrated transgenic components, and their removal often requires segregation in subsequent generations. This problem has practical solutions. In multi-target editing of wheat PPO gene families, for example, edited plants were advanced through crossing schemes to isolate progeny that had lost the transgene while retaining the desired edits, demonstrating that transgene-free edited lines can be recovered within breeding pipelines [162]. Additional progress has come from editing platforms that reduce or bypass stable DNA integration. Base editing, particularly cytosine base editors, enables precise nucleotide conversion without double-strand breaks or donor DNA and has been shown to facilitate the recovery of GM-free edited materials, thereby reducing concerns associated with residual transgenicity [163]. Likewise, HI-Edit/IMGE-type systems provide a conceptually attractive route to minimise transgenic residues through “targeted mutations through pollination”, in which editing occurs transiently after fertilisation and the inducer genome is subsequently eliminated, thus reducing the probability of stable transgene integration into elite germplasm. At the same time, these systems should not be described as entirely free of genomic side effects, because incomplete chromosome elimination, somatic chimerism, and introgression of inducer-derived DNA fragments have all been reported in at least a small proportion of materials [159].

A second issue concerns editing specificity. In wheat, considering that the technological barriers to transformation are compounded by the requirement for precise validation in a polyploid genome, it is important to emphasise the role of intermediate rapid-testing platforms (e.g., protoplast-based systems) in assessing sgRNA/Cas9 efficiency and specificity, followed by NGS-based evaluation of on-target and potential off-target sites. This approach reduces the risk of advancing ineffective or nonspecific constructs and renders the entire pipeline more controllable. Experimental evidence from wheat protoplast systems indicates that off-target mutation frequencies can be substantially lower than on-target editing frequencies and may in some cases be undetectable, although they cannot be excluded completely, particularly in a hexaploid genome with closely related homoeologous sequences [164]. At the same time, studies based only on in silico prediction tools, such as CRISPR-P, provide only partial reassurance, because low predicted off-target activity does not constitute genome-wide validation of editing specificity [162]. Thus, off-target assessment should ideally combine computational prediction with targeted or genome-scale sequencing validation, especially when edited lines are intended for breeding deployment.

These considerations are also important because unintended genomic variation is not unique to CRISPR-based systems. Wheat tissues maintained in vitro may themselves accumulate genomic instability, including DNA methylation changes, activation of transposable elements, polyploidy, chromosome-number variation, oxidative stress-associated DNA damage, strand breaks, and broader epigenetic alterations. Similarly, EMS-induced mutagenesis produces extensive unintended variation, including GC → AT substitutions, polymorphism, reduced genomic template stability, and signatures consistent with chromosomal damage and DNA methylation changes [165]. Against this background, genome editing should be viewed not as a completely error-free technology, but as a more controllable alternative to conventional mutagenesis and tissue-culture-driven variation, provided that construct design, validation, and downstream selection are rigorously managed.

In summary, the current evidence indicates that genome editing to shift wheat beyond the status of a recalcitrant crop does not depend on a single breakthrough, but rather on the convergence of multiple technological modules: morphogenetic enhancement of regeneration (GRF4–GIF1) to reduce genotype dependence and accelerate plant recovery [152]; optimisation of CRISPR architecture (sgRNA promoters, multi-sgRNA designs, and nuclease choice) to increase mutation frequency and predictability, including large deletions and strategic targets such as TaMTL for haploid induction [154,155]; careful management of genetically modified residues through segregation, DNA-free or low-integration editing strategies, and haploid inducer-mediated editing approaches; rigorous assessment of off-target and other unintended genomic changes through rapid-testing platforms and sequencing-based validation; and integration with HI/DH platforms and HI-Edit/IMGE strategies to provide a practical route to the rapid production of homozygous or transgene-free lines. This integrated framework aligns directly with the needs of breeding programs and the goals of functional genetics for quantitative traits [20,155,157,158,164,166].

9. Conclusions and Future Perspectives

Haploid and doubled haploid technologies remain among the most effective tools for accelerating wheat breeding because they enable the rapid production of fully homozygous lines, shorten breeding cycles, and support more reliable trait evaluation. In wheat, the two most practically relevant DH systems are androgenesis-based approaches (anther culture and isolated microspore culture) and wide hybridisation, especially the wheat × maize system. However, these systems are not interchangeable, as they differ in biological basis, reproducibility, technical demands, and suitability for different breeding contexts. Their practical value therefore depends on genotype, infrastructure, and breeding objectives.

Despite substantial progress, routine application of DH technologies in wheat is still constrained by major limitations, including strong genotype dependence, low responsiveness of many elite cultivars, frequent albinism, and inconsistent regeneration. These constraints mean that laboratory success does not always translate into breeding utility. For this reason, DH efficiency should be judged not only by the number of induced embryoids or regenerants, but also by the genetic stability, reproducibility, and breeding value of the resulting lines. In breeding practice, DH technologies are especially useful for the rapid fixation of favourable allelic combinations and for the development of aligned populations for field evaluation, mapping, QTL analysis, genomic selection, and improvement of complex traits. Future progress will depend on moving from highly genotype-specific protocols toward more robust and predictable systems. Particularly important directions include early prediction of androgenic response, identification of molecular markers of embryogenic competence, reduction in albinism, and adaptation of protocols for recalcitrant but agronomically valuable genotypes.

Further advances are likely to come from integrating DH platforms with genomics-assisted breeding and genome editing. Such integration can accelerate allele fixation and improve the strategic use of DH-derived material in breeding. Ultimately, progress in wheat DH technology should be measured not by the highest output achieved under isolated experimental conditions, but by the development of stable, scalable, and broadly applicable systems for diverse wheat germplasm.