Grasses belong to the family of Poaceae, which constitutes the most economically important plant family (Lee et al., 2020; Huang et al., 2022). Grasslands and meadows extend over vast portions of the planet, on land, and even under the sea (Lopez et al., 2022; McSteen and Kellogg, 2022). Their importance in Earth’s ecosystems goes beyond their use in fields and pastures. Grassy biomes comprise more than one-quarter of the planet’s land area. Grasses not only provide food, shelter, and building materials for animals and humans, but they also generate oxygen and store carbon (Strömberg and Staver, 2022). This storage, mainly subterranean, contributes towards the fertilization of soils and makes grasslands valuable sinks of CO₂ (Bengtsson et al., 2019; Terrer et al., 2021). Furthermore, grasses are considered more resilient to dryer and warmer conditions than trees. These facts suggest that in the climatic conditions predicted for the future, grasslands could be a better and more robust carbon sink than forests (Dass et al., 2018).

Grass crops provide the most essential dietary food sources globally. From these, forage grasses are the main component used to feed ruminant livestock (FAO, 2018; FAO, 2019). Grasses can be cultivated in less fertile lands compared to other crops. In these zones, normally associated with developing countries (Feller et al., 2012; Kwadzo and Quayson, 2021), animal husbandry and its derivates e.g., dairy production, remain essential (Capstaff and Miller, 2018; Moorby and Fraser, 2021). To cope with the predicted population growth and the consequential increase in food needs, high-yielding crops must be further developed (Raza et al., 2019). To reach food security, the strategies used must avoid causing negative environmental impacts. Synthetic nitrogen-based fertilizers have been important for reaching high yields, nevertheless, their production and usage are a source of massive generation and emission of greenhouse gases (GHGs) (Chai et al., 2019). It is well known that the high concentration of atmospheric GHGs is closely related to climate change. Therefore, the challenge is to increase farming efficiency while reducing the impact of agricultural activity on climate change (Rivero et al., 2021). Importantly, climate change not only directly affects crop productivity but also has indirect and socio-economic impacts, for instance soil fertility, need for irrigation, food demand, policy, rising costs (reviewed in (Raza et al., 2019)).

Grasses usage as forage and as reliable sinks of carbon emissions, call for an improvement in their biomass yield, and their resistance towards the new abiotic and biotic stresses caused by climate change (Giridhar and Samireddypalle, 2015). Especially, plants will have to cope with variations in temperature, water availability, and soil composition (Cushman et al., 2022). Said variations will generate stresses due to heat, cold, drought, and salinity conditions. A promising approach to provide grasses with stress resistance is using genome editing techniques (Bailey-Serres et al., 2019; Dhakate et al., 2022). The first attempts have been performed to use genome editing in forage grasses (Liu et al., 2020b; Weiss et al., 2020; Zhang et al., 2020; Cheng et al., 2021; Zhang et al., 2021; Zhu et al., 2021; Kumar et al., 2022; Wang et al., 2022). This is not an easy task due to their reproductive and genetic characteristics which are difficult to work with. The inability of forage grasses to self-pollinate hinders inbreeding. Additionally, forage grasses have high variability between the genetic background of different individuals. This provides them with a considerable gene pool, responsible for their adaptability and resilience towards environmental changes. Conversely, it creates difficulties for studies focused on identifying the genetic cause of traits or phenotypes of interest (Cropano et al., 2021; Muguerza et al., 2022). There are diverse ways of classifying grasses beyond their taxonomy. For instance, forage grasses can be divided into different types depending on their life cycle and ecotype. In the first case, according to the survival of the plant after going through its reproductive phase, grasses can be considered annual, biannual, or perennial. In terms of their ecotype, grasses can be separated into warm- or cool-season plants if their optimal growth happens during winter or summer, respectively. Importantly, warm-season grasses are C₄ plants, while cool-season grasses are C₃ plants (Moser and Hoveland, 1996; Moser et al., 2004).

In this review, we provide an overview of the main metabolic and molecular changes that plants suffer to cope with the effects of abiotic stress derived from climate change. Additionally, we summarize the actual state of genome editing applications in forage grasses. We propose how genome editing could be used to generate grass plants able to resist these abiotic stresses. Finally, we hypothesize how the new genetic resources and tools can be used to improve forage grass breeding that will help achieve food security in a sustainable way.

Extreme temperatures, uncommon precipitation patterns, and deterioration of soils are being observed due to climate change. These environmental consequences have a great impact on agriculture since plants are of sessile nature. The responses used by plants when encountering a stressor aim firstly to achieve acclimation to the new environment and later adaptation to it. Acclimation includes adjusting the physiology and metabolism of a plant to achieve a new state of homeostasis, while adaptation involves both phenotypic and genotypic alterations. Acclimation mediates quick responses to ensure the survival of a plant, whereas adaptation is considered an evolutionary and lengthy process whose goal is to preserve a population. Additionally, plants undergo epigenetic modifications when a stressing event happens (Guarino et al., 2022). Plants must cope with new and more extreme conditions, which lead to different abiotic and biotic stresses than those commonly present in their biomes (Sugimoto and Nowack, 2022). Abiotic stresses are those derived from the physical and chemical factors of an environment and are independent of living organisms (He et al., 2018). As a response to these environmental alterations, plants undergo morphological, metabolic, and physiological changes. In this review, we will focus on drought, salinity, cold, and heat stress responses at the cellular and molecular levels. These are not the only abiotic conditions that will vary due to climate change, but they represent some of the major alterations that will result from it (He et al., 2018; Villalobos-López et al., 2022). The stresses discussed in this review have a significant impact on the growth and development of plants, which is directly connected to crops’ yield and profitability (Bita and Gerats, 2013; Bulgari et al., 2019).

Even though the abiotic stresses will be described separately, in nature they tend to interact producing greater effects than individually. Therefore, plants normally must acclimate to a combination of stresses. This should not be ignored when designing strategies to improve crops’ tolerance to stress (Pascual et al., 2022).

The biggest challenge for agriculture nowadays is to obtain plants that are resilient to adverse environmental conditions, and at the same time provide enough yield to fulfill food and feed security in a sustainable way. In the case of perennial forage grasses, yield is determined by repeated harvesting of herbage over as many years as possible. Therefore, forage grass genotypes with improved survival and growth under abiotic stress conditions are needed.

Genome editing tools have proven to be useful for achieving such aims, especially the Nobel prize-winning discovery of application of RNA-directed Cas9 nuclease for genome editing (Gasiunas et al., 2012; Jinek et al., 2012) abbreviated as CRISPR-Cas9. Although this editing strategy was immediately applied in model and crop plants, almost ten years ago (Feng et al., 2013; Jiang et al., 2013; Li et al., 2013; Nekrasov et al., 2013; Shan et al., 2013), not much has been achieved in the forage grasses landscape. The European GMO database EUGENIUS lists only green foxtail (Setaria viridis (L.) P. Beauv.) line 193-31 that has been modified using CRISPR-Cas9 mediated mutagenesis. The expressed CRISPR-Cas9 system targeted the coding region of the S. viridis homolog of the Zea mays L. Indeterminate 1 (ID1) gene, which promotes flowering in maize. The deactivation of the homolog in S. viridis led to delayed flowering. In the knockout line 193-31, the CRISPR-Cas9 DNA construct was segregated away (GE Setaria viridis molecular characterization details, n.d).

To find out how many publications have been released showing edited genes in forage grasses, a search was carried out in the following databases: Scopus, Web of Science, Google scholar and PubMed. The search included the scientific or the common names of 47 grass species ( Supplementary Table 1 ) or the name of each of the 12 subfamilies of Poaceae and, in addition, one of the following terms: “CRISPR”, “genome editing”, “genome editing”. The outcome of the search is shown in Table 1 . The genome of only six species, three annual grasses and three perennial ones, all growing in temperate regions, has been targeted with CRISPR-Cas tools. Genome editing in S. viridis, a model plant for C4 grasses, has been reported three times. Most of the work has been done by knocking out a single gene using the easiest genome editing approach, i.e., CRISPR-Cas9.

CRISPR-Cas9 as a system for carrying out simple mutations (indels: insertions/deletions) that change the reading frame of a coding region and therefore generate knockouts, is straightforward and still mainly used for functional genomics. It consists of two main components: the Cas9 nuclease from Streptococcus pyogenes and the short guide RNA (gRNA) that targets the DNA sequence of interest. Designing the gRNA with precision enables the simultaneous mutations of all alleles of a gene in a polyploid plant, as it was the case for Panicum virgatum L. (tetraploid) and Lolium arundinaceum (allohexaploid, Table 1 ). Specific genes that have been knocked-out in forage grasses are related to flowering (phytochrome C—PHYC—of Setaria italica (L.) P.Beauv. and floral organ number 2—FON2—of S. viridis), tillering and branching (teosinte branched 1—tb1a and tb1b—of Panicum virgatum), meiosis (disrupted meitoic cDNA 1—DMC1—of Lolium multiflorum Lam.), haploid induction (matrilineal—MTL—of S. italica) and heat stress response (17.9 kDa class II heat shock protein—HSP17.9—of L. arundinaceum), apart from the phytoene desaturase (PDS) gene used as endogenous marker ( Table 1 and references therein). In most of the cases the cited publications discuss the targeted mutagenesis method and results obtained, but the phenotypic characterization of the mutants is limited and far away from field trials. Interestingly, not only classical CRISPR-Cas9 system has been used, but also CRISPR-Cas12a in the case of L. arundinaceum (Zhang et al., 2021) and CRISPR-Cas9_Trex2 in the case of S. viridis (Weiss et al., 2020).

The toolkit of CRISPR-Cas applications has expanded to around twenty different techniques that allow diverse targeted modifications in the genome (Villalobos-López et al., 2022; Capdeville et al., 2023). On the one hand, Cas enzymes from different bacteria have been characterized and adopted for use. That is the case for Cas12a (former Cpf1), an enzyme from the Lachnospiraceae bacterium ND2006 that cuts DNA strands distal from the sequence recognized by the nuclease (the PAM site), generating 4-5 nucleotide overhangs that enable an easy insertion of donor DNA sequences (Zetsche et al., 2015; Moreno-Mateos et al., 2017). Other modifications of the CRISPR-Cas9 system imply the co-expression or the fusion of different proteins to the Cas9 nuclease, in its original or mutated versions. CRISPR-Cas9_Trex2, for example, has the Trex2 exonuclease co-expressed with Cas9 for increasing the mutation efficiency (Čermák et al., 2017; Weiss et al., 2020). Importantly, an enzymatically inactive variant of Cas9, called “dead Cas9” (dCas9) that maintains its specific DNA binding ability, can be fused to transcription activators or repressors to regulate transcriptional levels of endogenous genes (Ding et al., 2022). Therefore, CRISPR-Cas tools are not only meant to inactivate genes and create loss-of-function mutants, but also gain-of-function mutants can be obtained. In addition, thanks to the Super Nova Tag (SunTag) system, the transcriptional regulation can be potentiated. The SunTag contains peptide repeats that bind several transcription factors for cooperatively activating a target gene (Tanenbaum et al., 2014). Moreover, a gene of interest may also be up- or downregulated epigenetically. For instance, CRISPR-dCas9 linked to DRM methyltransferase catalytic domain targets methylation to specific loci and thereby inactivates the target gene (Papikian et al., 2019).

An alternative way of inducing a change in the levels of expression of a gene is altering its promoter sequence. In fact, the promoter can be even swapped by another one that ensures e.g., higher levels of expression in a ubiquitous manner. Using CRISPR-Cas9 such a substitution is possible, as shown for the auxin-regulated gene involved in organ size 8 (ARGOS8) gene in maize, whose overexpression was associated with improved grain yield under field drought stress conditions (Shi et al., 2017).

It should be pointed out that yield and stress resistances are among the most difficult polygenic traits to improve through genetic engineering, but examples as the former one give hope that it can be achieved by CRISPR-Cas. Another example is the knockout via CRISPR-Cas9 of the main effect gene type-B response regulator 22 (OsRR22) that controls salt tolerance in rice. Obtained plants showed salt tolerance in growth chambers and no difference in agronomic traits compared to wild type plants in field trials under normal growth conditions (Zhang et al., 2019; Han et al., 2022a).

As explained in section 2, abiotic stress responses are complex, linked to different metabolic pathways and the genes involved in those mechanisms are mainly pleiotropic. Fishing out a specific key player, a master gene to be mutated, could be possible in some cases and it is worth trying. Since genome editing in grasses is in its early stages ( Table 1 ), we selected specific genes related to the four abiotic stresses discussed in this review and figured out if those target genes would need to be overexpressed or downregulated to gain tolerance to specific stresses. The suggested genes can be found in Table 2 . If a candidate gene was found in forage grasses or at least in a Poaceae species, that species was selected, but this was not possible in all cases. As shown in Table 2 , there are genes that are related to more than one stress response. For simplicity, it is not shown that, e.g., DHN11 seems to be also involved in cold and drought stresses and COR410 appears to be related to drought stress as well.

Section 2 mentioned that plants detect an increase in temperature (in the soil or air) when the structure and fluidity of their cell membranes change. Heat stress tends to make membranes more fluid (Niu and Xiang, 2018), which activates pathways through heat sensors like the CNGCs. In theory, an increased expression of stress receptors can lead to an improved response to stress. Consequently, the genes involved in the heat stress response signaling pathway can be upregulated by overexpressing a heat sensor coding gene. In A. thaliana, an overexpression of the SYTA gene resulted in higher germination and seedlings survival rates than in wild-type and knockout lines after heat stress exposition. Moreover, the overexpression plants presented higher expression of both HSPs and HSFs, together with lower levels of membrane lipid peroxidation than in non-overexpression lines (Yan et al., 2017). All these changes provide evidence that upregulating a heat stress sensor can improve the stress tolerance of a plant. Therefore, overexpressing a similar gene in grasses, like a homologous of rice OsCNGC14 or OsCNGC16 gene, could result in forage species with higher tolerance to heat stress. A similar approach can be followed by upregulating proteins present in plants in a basal state that are involved in the responses to abiotic pressures ( Figure 1 ). Kinase proteins are suitable for this goal since they are involved in most stress response pathways, regulating posttranslational modifications of other proteins as a response to both abiotic and biotic stress (Damaris and Yang, 2021). Therefore, overexpressing a gene from the SnRK2 family, a group of kinases specific to plants that have been shown to play important roles in abiotic stress regulation is an adequate approach (Zhang et al., 2016). The heterologous overexpression of the gene TaSnRK2.3 from wheat in Arabidopsis produced plants that had higher tolerance to drought conditions (Tian et al., 2013). Similarly, another study was able to overexpress the AcSnRK2.11 gene from Agropyron cristatum (L.) Gaertn., a forage grass species, in Nicotiana tabacum L. The overexpression plants had significantly higher survival rates than the wild-type ones after recovery periods from cold stress and presented significantly upregulated patterns of abiotic stress-related genes like dehydrins. Possibly, upregulation of these protein kinases could provide drought, cold and salinity stress tolerance to forage grasses plants.

On the other hand, negative regulators of abiotic stress responses are also suitable targets for abiotic stress tolerance improvement by downregulating them via genome editing ( Figure 1 ). Possible candidates for downregulation could be enzymes that degrade signaling molecules involved in stress response, like for example the inositol phosphatases (Jia et al., 2019). As previously mentioned in this review, the phosphatase SAL1 negatively regulates plants’ response to drought (Chan et al., 2016). Using the CRISPR-Cas9 system, scientists have already generated Tasal1 knockout mutant wheat with fewer and smaller stomata, that germinate and grow better under drought conditions (Abdallah et al., 2022). Likewise, modifying the expression of transcription factors related to abiotic stress is another alternative for producing tolerant plants. The transcription factors of the basic helix-loop-helix (bHLH) family have been shown to participate in abiotic stress regulation in different plant species (Guo et al., 2021). In rice, the OsbHLH024 gene seems to negatively regulate salinity tolerance. This was demonstrated by generating knockout plants using the CRISPR-Cas9 system. The mutated plants had an increased salinity tolerance when compared to the wild-type ones. Additionally, the knockout lines presented a reduced accumulation of sodium ions and ROS, but higher concentrations of potassium ions than the control plants. Finally, the expression of genes encoding ion transporter was upregulated in the knockout plants in comparison to the wild-type ones (Alam et al., 2022). All these variations suggest that the downregulation of homologues of the OsbHLH024 gene in grasses could provide them with salinity stress tolerance.

During the last years, innovative ways of inserting specific targeted mutations based on CRISPR-Cas have been developed, e.g., base- and prime editing and for now, some technical problems need to be overcome when applied to plants (Anzalone et al., 2020; Zhu et al., 2020; Hua et al., 2022). In these cases, the Cas9 nuclease is mutated in such a way that it acts as nickase, cutting only one strand of the targeted DNA. These strategies and the activation of homology- directed repair (HDR) instead of Non-Homologous End Joining (NHEJ), makes it possible to produce a wide range of mutations from single nucleotide changes and small indels to increasingly larger insertions and deletions, replacements or even to generate chromosomal rearrangements (Puchta et al., 2022; Villalobos-López et al., 2022).

The possibilities to induce targeted changes with CRISPR-Cas in the genome of crops, and specifically in forage grasses, are immense, not to mention the speed of obtaining the desired traits compared to conventional breeding techniques. In addition, genome editing can be easily multiplexed for targeting different sequences at one shot. Depending on the specific trait and species, there can be bottlenecks to be removed like specific ways of transforming a plant or availability and annotation quality of the reference genome. These obstacles are thought to be solved with technical advances, however in the case of grasses, important biological features need to be taken into consideration when aiming to combine genome editing with a breeding program. These challenges are elaborated in section 5. Here we briefly mention that also reproductive characteristics of grasses can be changed with genome editing.

Forage grasses have a strong gametophytic self-incompatibility (SI) system that makes inbreeding almost impossible. The two multi-allelic S and Z genes have since long been known to govern SI in grasses (Lundqvist, 1955; Cornish et al., 1979), and recently it was shown that two DUF247 genes are behind the S and Z loci (Manzanares et al., 2016; Herridge et al., 2022). With the sequences and molecular function of these genes known, they would be an obvious target for generating self-fertile knockout lines by genome editing. A similar approach has been used to develop self-compatibility in potato (Ye et al., 2018).

To obtain male sterile lines is also of importance in the case of forage grasses. The way has been paved by research in maize, where genes male sterility 1 (Ms1) and Ms45 have been targeted by CRISPR-Cas9 and male-sterile wheat lines for hybrid seed production have been obtained (Singh et al., 2018; Okada et al., 2019).

Fully homozygous doubled haploid lines can be generated by artificially inducing haploids with a knockout of MTL gene, as it has been done already in S. italica ( Table 1 ) (Cheng et al., 2021).

Finally, apomixis is present in several grass species, e.g., Poa pratensis, a species used both in lawns, pastures, and leys. Inducing apomixis in other forage grasses would be of importance for genetic fixation of hybrid vigor of parental line. Some steps towards achieving this aim have been taken already in rice. Mutations using CRISPR-Cas of several genes related to the abolishment of meiotic steps produced clonal diploid gametes. Then, parthenogenesis was induced by ectopic expression in the egg cell of BABY BOOM1 and clonal progeny was obtained (Khanday et al., 2019; Zhu et al., 2020).

Genetic variation is fundamental to crop improvement. Modern plant breeding started in the late 19^th century with the advent of cross-breeding which still is the backbone of most plant breeding efforts (Hickey et al., 2019; Gao, 2021). After the discovery that physical and chemical factors can lead to heritable changes in genetic material, random mutagenesis became a valuable tool for plant breeding to increase genetic diversity and to develop specific traits. With the discovery of recombinant DNA technology in the 1970s, the development of new combinations of genetic elements by splicing genes and regulatory elements from different species became possible. The discovery of Agrobacterium-mediated transformation enabled scientists to introduce these novel combinations of genes into plant genomes to produce new traits (Gao, 2021). While the introduction of transgenes into plant genomes has contributed enormously to the understanding of gene functions in plants, the commercial applications have been limited to mostly herbicide tolerance and insect resistance, which provide obvious advantages for farmers, but little direct, tangible benefits for consumers in developed countries. Only a few commercial applications of transgenic plants with improved yield and abiotic stress resistance are known. Wheat expressing the sunflower transcription factor HomeoBox 4 (HaHB4) has been shown to provide improved water use efficiency resulting in higher grain production (González et al., 2019). Wheat HB4 marketed by the company Bioceres Crop Solutions has been authorized for food and feed uses in a number of countries, such as Argentina, Australia, Brazil and United States, but its cultivation is approved only in Argentina (Argentina First to market with drought-resistant GM wheat, 2021; HB4 Wheat| GM Approval Database- ISAAA.org, n.d.). Maize MON87403 contains the ARABIDOPSIS THALIANA HOMEOBOX 17 (ATHB17) gene from A. thaliana encoding a transcription factor of the HD-Zip II family with reported increase in ear biomass at the early reproductive phase (Rice et al., 2014), which may provide an opportunity for increased grain yield under field conditions (Leibman et al., 2014). Maize MON87460 expresses the Bacillus subtilis cold shock protein B (CspB) resulting in increased grain yield under drought conditions (Nemali et al., 2015). Both GMO events have been assessed by the European Food Safety Authority (EFSA) (EFSA Panel on Genetically Modified Organisms (GMO) et al., 2012b; EFSA Panel on Genetically Modified Organisms (GMO) et al., 2018), and MON87460 was authorized for food and feed uses in the EU. Transformation techniques have been developed for most of the economically important forage and turf grass (Wang and Ge, 2006), however, very few transgenic forage grasses have been registered for commercial cultivation. The ISAAA GMO approval database lists only one transgenic event in creeping bentgrass (Agrostis stolonifera L.) with tolerance to glyphosate (ASR368) (Creeping Bentgrass (Agrostis stolonifera) GM Events | GM Approval Database - ISAAA.org, n.d.).

Even though commercial cultivation of GM crops has brought clear benefits to farmers and more indirect benefits to environment through reduced land and pesticides use (Brookes, 2019; Brookes, 2022), cultivation and use of transgenic plants for food and feed have been controversial in many regions of the world, and especially in Europe. Agronomic, environmental, human health, social and economic effects of transgenic crops have been comprehensively reviewed by the US National Academies of Sciences in 2016 (National Academies of Sciences, Engineering, and Medicine et al., 2016).

Genome editing became possible with advances in protein engineering which allowed production of site-directed nucleases (SDNs), such as zinc finger nucleases (ZFNs) and transcription activator-like effector nucleases (TALENs) (Shukla et al., 2009; Urnov et al., 2010). As outlined in section 3, genome editing has several advantages over the transgenic techniques including precision, lower number of off-target effects, more streamlined production, multiplex possibility, as well as potential for modification of many more different traits. A few examples include lower gluten content in wheat through simultaneous editing of alpha-gliadin genes (Sánchez-León et al., 2018), increased production of gamma-aminobutyric acid in tomato (Li et al., 2018) or increased accumulation of provitamin D3 in tomato (Li et al., 2022). The maize with an increased expression of ARGOS8 gene, as detailed in section 3, contained no exogenous DNA sequences, thus, theoretically, it could be exempt from GMO regulation depending on country-specific policies.

The increased precision, low off-target potential and the absence of exogenous DNA in some of the genome-edited plants suggested that genome editing would not be regulated similarly to GMOs. For example, in Japan Sanatech Seed has commercialized high gamma-aminobutyric acid tomato (Waltz, 2021). In the EU, however, the Court of Justice of the European Union (CJEU, case C-528/16) ruled that organisms resulting from mutagenesis techniques in legal aspects are GMOs and are subject to the regulations laid down by the Directive 2001/18/EC. This applies to mutagenesis techniques introduced since 2001, when the GMO Directive was adopted. Site-directed nucleases can modify plant genomes according to three scenarios, SDN-1, SDN-2 and SDN-3 (EFSA Panel on Genetically Modified Organisms (GMO) et al., 2012a), where only SDN-3 scenario results in transgenic plants, while under SDN-1 and SDN-2 scenarios no exogenous DNA is inserted into the genome. However, under the CJEU ruling, also the SDN-1 and SDN-2 techniques, including CRISPR-Cas fall under the GMO Directive, while chemical and radiation random mutagenesis remains exempt according to Annex IB of the Directive 2001/18/EC. The ruling provoked a strong response from both academia and biotech industry, which stressed that from a scientific point of view the application of GMO Directive to products created by a much more precise technique than random mutagenesis and transgenesis results in a disproportionate regulatory burden (Purnhagen et al., 2018; Urnov et al., 2018; Christiansen et al., 2019; Wasmer, 2019; Schulman et al., 2020). It was also noted that this ruling leads to a situation when two identical products with the same mutation resulting in, e.g., herbicide tolerance trait could be regulated in different ways. In addition, it would create an unsustainable situation with detection, since no technology can determine the origin of simple mutations, such as single nucleotide polymorphisms. Consequently, reliable detection methods for SDN-1 and SDN-2 products are problematic (European Network of GMO Laboratories (ENGL), 2019). This legal uncertainty makes genome-editing research in the EU less appealing, as seeking regulatory approval for gene-edited products would involve the same cumbersome procedure as for GMOs. So far there are no applications for regulatory approval involving gene-editing, although a few applications for authorization of products obtained with CRISPR-Cas9 in SDN-3 scenario, e.g., maize DP-915635-4 have been submitted to member states and are currently under review by EFSA (Maize DP-915635-4, n.d.).

According to the EU (Council Decision (EU) 2019/1904), the European Commission (EC) conducted a study involving input from the Member States and different stakeholders regarding the status of new genomic techniques (NGTs) including genome editing. Within this framework, the EC mandated EFSA to issue a scientific opinion on the risk assessment of plants produced by the SDN-1, SDN-2, and oligonucleotide-directed mutagenesis techniques. EFSA has assessed the safety of plants developed using SDN-1 and SDN-2 techniques and did not identify new hazards specifically linked to these techniques compared to both SDN-3 and conventional breeding. In addition, EFSA concluded that the existing Guidance for risk assessment of food and feed from GM plants and the Guidance on the environmental risk assessment of genetically modified plants are sufficient, but only partially applicable, to plants generated via SDN-1 and SDN-2 (EFSA Panel on Genetically Modified Organisms (GMO) et al., 2020; Rostoks, 2021). As part of the ongoing effort to update the EU GMO legislation upon EC request, EFSA recently produced an updated scientific opinion on cisgenesis and intragenesis (EFSA Panel on Genetically Modified Organisms (GMO) et al., 2022b). The EFSA scientific opinion concluded that no new risks were identified in cisgenic and intragenic plants obtained with NGTs, as compared with those already considered for plants obtained with conventional breeding and established genomic techniques, although only limited information on such plants was available. EFSA determined that the use of NGTs reduces the risks associated with potential unintended modifications of the host genome resulting in fewer requirements for the assessment of cisgenic and intragenic plants, due to site-specific integration of the added genetic material. However, there was no legal necessity to overhaul the GMO legislation, since the EFSA concluded that the current guidelines were partially applicable and sufficient. Importantly, the data requirements could be reduced on a case-by-case basis for the risk assessment of cisgenic or intragenic plants obtained through NGTs. While cisgenesis and intragenesis is just one of the possible approaches for forage grass breeding, EFSA also recently issued a statement on criteria for risk assessment of plants produced by targeted mutagenesis, cisgenesis and intragenesis (EFSA Panel on Genetically Modified Organisms (GMO) et al., 2022a). These criteria could be used by policy makers to design a more flexible and proportionate risk assessment framework for gene edited plants. Recently, several regulatory options have been proposed (Bratlie et al., 2019; Kearns et al., 2021; Gould et al., 2022). They range from maintaining the status quo (full risk assessment of genome edited organisms as GMOs) to product-based regulation or regulation based on the presence/absence of foreign DNA in the genome. These two options would be preferable for commercial deployment of genome edited crops, but they would require substantial reexamination of GMO Directive and authorization procedure. The EC is expected to present a new policy and/or legal proposal by the second quarter of 2023. Meanwhile, other jurisdictions around the world have already developed legal framework for genome edited plants, e.g., under Argentina NBT Resolution N° 21/2021, if a product (plant, animal or microorganism) does not have a new combination of genetic material, the product is non-GM and considered as conventional product (Goberna et al., 2022). Different regulatory approaches are summarized in a recent review (Entine et al., 2021).

Interestingly, the “EU GMO database of Deliberate Release into the environment of plants GMOs for any other purposes than placing on the market (experimental releases)” lists over 900 applications for field trials registered by the Member States since 2002 (Experimental releases of GM Plants, n.d.). Among those there is only one application for field trial of high fructan transgenic ryegrass in 2006, and there are no applications for field trials of genome edited forage grasses, although at least 14 field trials of plants edited with CRISPR-Cas9 have been authorized.

In conclusion, while there are a few basic studies on gene function in forage grasses using genome editing technique as described in section 3 of this review, these are yet to see commercial application. The main limiting factor for the investment in research and development of genome edited forage grasses is probably the regulatory uncertainty, especially in the EU. Although edited plants without foreign DNA in the genome are expected to receive the least amount of regulatory scrutiny, they are also less prone to show major changes in relevant traits. This is because gene knockouts or simple gene edits are unlikely to result in complex phenotypes, such as enhanced abiotic stress tolerance, higher yield or improved nutritional composition, especially considering the genetic complexity that has hindered progress in characterization of the genes underlying such traits in forage grasses. Nevertheless, as recent years have witnessed a dynamic development of genome editing tools and genotype-independent transformation approaches along with increasing genomic resources, the manipulation of plant responses may become possible to overcome abiotic stresses when combining modern techniques and good breeding management strategies.

Forage grasses are outbreeding species and highly heterozygous due to the strong gametophytic SI system. Inbred line development is thus very difficult with strong inbreeding depression as a result. Therefore, cultivars of forage grasses are usually synthetic populations (Posselt, 2010). Forage grass breeders usually start by phenotypic selection of superior candidate genotypes for traits with high heritability, e.g., heading date and disease resistance, among a large number of spaced plants. However, forage grasses are sown in swards and because yield and other traits will be affected by competition in the swards, such traits cannot be selected on single spaced plants. The candidate genotypes are therefore put in some form of progeny testing system, e.g., polycross to produce half-sib (HS) families or bi-parental crosses producing full-sib (FS) families, and selection for yield and forage quality traits are based the performance of such families in swards (genotypic selection). Synthetic populations/cultivars are constructed by crossing the best genotypes based on their performance in the progeny test or by mixing HS or FS families. The synthetic populations are further multiplied to obtain enough seed for establishing sward plots for testing in multi-location-year trials before the best candidate cultivars are being submitted to official variety testing. A typical breeding cycle will take 10-15 years before synthetic cultivars are available for farmers. With the advent of high-throughput molecular markers, whole-genome sequences, and genomic selection methods, the breeding cycle can be shortened (Rognli et al., 2021; Barre et al., 2022). Specifically, if genome editing is used for specific reproductive traits, like breaking down self-incompatibility, the forage grass breeding cycle could be shortened according to us by 4-5 years.

The success of a breeding program is very much dependent on the genetic variation present in the initial breeding material. Many agronomically important traits, like yield and adaptability to biotic and abiotic stresses, have been partly fixed within elite germplasm, however, they still exhibit large genetic variation and are thus of primary importance in breeding programs (Meyer et al., 2012; Swinnen et al., 2016). This variation might be employed for future improvements of crop productivity and tolerance to stress; however, landraces, closely related species and wild relatives can offer much wider and unexploited germplasm resources (Jonavičienė et al., 2009; Brozynska et al., 2016). Extensive studies of perennial ryegrass diversity among modern European cultivars revealed that modern cultivars are mostly related to ecotypes from north-western Europe (Blanco-Pastor et al., 2019), while most of the natural genetic variation remains unexploited. Later studies on the genetic structure of geographically diverse perennial ryegrass collection supported these findings and in addition showed that latitude was a prominent force shaping the diversity of wild-growing perennial ryegrass populations (Jaškūnė et al., 2020). Furthermore, the ecotypes exhibit biomass and seed yielding potential similar to cultivars (Bachmann-Pfabe et al., 2018; Jaškūnė et al., 2022b), suggesting that ecotypes could serve as valuable trait donors in breeding programs. Field testing of many L. perenne ecotypes and cultivars at several Nordic and Baltic locations identified tetraploid Baltic breeding lines and diploid ecotypes from Eastern Europe as being most winter hardy with stable performances across environments (Gylstrøm, 2020). None of the cultivars were among the most stable entries, and diploid ecotypes displayed a larger variation in heading date, regrowth, and winter survival than the cultivars. Thus, there is ample genetic variation still to be exploited within the genetic resources of perennial ryegrass. Induced polyploidization is also widely exploited in forage crop breeding as one of unconventional techniques to develop new superior yielding and abiotic stress tolerant breeding material (Akinroluyo et al., 2019; Akinroluyo et al., 2020; Rauf et al., 2021).

To utilize transgenes or gene-edits in grass breeding, first, efficient methods for introduction and regeneration in vitro need to be available in a range of independent genotypes. In principle, introgression of new genes can either be introduced into the parental clones of already existing varieties (variety-parent approach) or transferred into a new base population (population approach) (Posselt, 2010). Repeated backcrossing and an efficient selection system is needed to bring transgenes/gene-edits to homozygosity in the parental clones. A side-effect of this could be increased inbreeding depression due to linkage drags creating longer homozygous chromosomal segments. Traditional random insertion of transgenes in several genotypes that are intercrossed to construct synthetic cultivars is problematic due to the presence of multiple insertion sites, silencing and variable expression levels. The availability of complete genome sequences also of forage grass species, notably L. perenne (Nagy et al., 2022), and genome editing technologies, makes it possible to induce precise genome alterations. This will make it easier to develop synthetic cultivars of outbreeding crops like forage grasses with stable expression of genetic modifications.

Integration of transgenic traits in perennial grasses and the challenges associated with deployment and management of transgenic cultivars has been discussed by Badenhorst and colleagues as well as by Smith and Spangenberg (Badenhorst et al., 2016; Smith and Spangenberg, 2016). Using gene-drive technologies (Bier, 2022) would in principle be an efficient method for spreading gene-edits through breeding populations of grasses. However, the risk of gene flow between cultivars and to feral populations is high and would probably preclude practical use of such technologies.

A pertinent question is what the most important targets for genetic engineering in forage grasses would be. Genetic gain for yield has been modest due to the long breeding cycles and extensive field testing (Sampoux et al., 2011; McDonagh et al., 2016). The potential heterosis is only partially exploited in synthetic cultivars, and it is expected that great yield increased could be achieved if F1 hybrids, which has been very successfully exploited in maize, could be developed (Herridge et al., 2020). Self-incompatibility, inbreeding depression, and the lack of male-sterile lines for making hybrids are major obstacles for developing F1 hybrids. Inbreeding depression needs to be tackled to implement self-fertile lines in forage breeding programs. By generating a large number of self-fertile plants with diverse genetic backgrounds by gene-editing, and selecting genotypes with good seed set, the prospects of developing inbred lines in forage grasses have never been better. These lines could be used for F1 hybrid production and would also be very useful for functional studies. Other methods for capturing heterosis would be the development of facultative apomixis. The evolution of apomixis in natural populations and the challenges of utilizing apomixis in breeding has been reviewed recently (Hojsgaard and Hörandl, 2019).

In the current review, we focus on possible improvements of abiotic stress tolerance in forage grasses using new genome editing tools. The potential impact of climate change is described in relation to forage grass tolerance to four important abiotic stresses, such as heat, low temperature, drought and salinity. We propose approaches for editing the genome of grasses to regulate stress responses. Furthermore, we discuss the latest developments in the regulatory framework for genome editing, especially with regard to the EU, and identify factors affecting the application of genome editing techniques for the improvement of grasses. Finally, we address breeding strategies specific to the reproductive biology of forage grasses and identify how genome editing could be used to facilitate breeding and achieve food security in a sustainable way. In conclusion, we describe pathways for developing abiotic stress tolerance in forage grasses under climate change using genome editing technologies, provided that an appropriate legal framework is developed.

Conceptualization, FS-S, CS. writing—manuscript preparation, FS-S, NR, KJ, MS, OAR, MRK, GS and CS. writing—review and editing, FS-S and CS. funding acquisition, KJ, NR, OAR and CS. All authors have read and agreed to the published version of the manuscript. All authors contributed to the article and approved the submitted version.