Introduction

The organization and persistence of human society depend essentially on crop cultivation, which is the basis of food supply as well as many raw materials for industrial uses. Among the crops we currently cultivate around the globe, many are arboreal plants used to obtain a diverse range of products, such as fruits, nuts, oils, syrup, rubber, and lignocellulose-based products such as timber, fiber, charcoal, and cork. As perennial crops, their cultivation requires a long-term commitment of resources by the farmer, which cannot be easily changed according to market fluctuations. Beyond the intrinsic difficulties related to tending to a plant to harvest its commercial products, many tree species have been domesticated only recently and still display many non-fixed traits, particularly clonally propagated species (Spengler, 2019). For most tree crops, compared to annual crops, breeding is still in its infancy due mostly to the time required to surpass the juvenile phase of a tree (some requiring several years and even decades) (Neale et al. 2017) and the decision of defining varieties as clonally propagated genotypes selected from segregating populations.

Our ability to cultivate crops to provide resources to meet human needs is fundamental for the continued development of agriculture. Increasing yields and quality of primary products have been, to a large extent, the result of scientifically informed breeding, which was a pillar of the first Green Revolution. Since then, technological leaps have led to transformational improvements in our ability to modify plant traits through genetic manipulation. The recent CRISPR-based gene-editing technology and the advancements in plant molecular genetics and genomics are leading us toward the fourth generation of crop improvement, marked by the precise and fast breeding of traits informed by gene function and an enhanced ability to quickly tailor crops to specific demands of farmers and consumers while maintaining the sustainability of agricultural systems (Fernie and Yan 2019).

At this point, gene editing has already been applied to some arboreal plants, e.g., poplar (Populus tomentosa) (Fan et al. 2015), coffee (Coffea canephora) (Breitler et al. 2018), apple (Malus domestica) (Charrier et al. 2019), and pomelo (Citrus maxima) (Jia and Wang 2020). However, the challenges of carrying out progeny studies, genome projects, or even manipulating plants in vitro can hamper the broader application of this technology to tree species. Still, in envisioning the potential and expanding the capacity of CRISPR-based genome editing, as demonstrated by recent advances in base editing (Grünewald et al. 2020) and the development of prime editing (Anzalone et al. 2019; Lin et al. 2020), many possibilities remain to be explored for refining pipelines of tree genetic improvement, including the domestication of species that are not conventionally cultivated for commercial purposes.

When producers (in small farms or big companies) decide to cultivate tree crops, they expect the plants to commercially produce for many years or even decades to offset the initial investment and opportunity costs. Therefore, tree breeding programs must consider limited farm inputs and climate resilience for sustainable production. The use of gene editing to either directly solve breeding constraints or identify the causal genes of important traits will assist in the establishment of feasible roadmaps for cultivar improvement. Alongside the focus on elite varieties of established tree crops, a more diversified panel of species can now be achieved through modern domestication via gene editing of potential wild crops to innovate by increasing cultivation diversity and creating new commercial niches for primary products.

The examples are many and widespread across the globe of native species awaiting domestication or which are not explored at its fullest despite their great potential, such as pawpaw (Asimina triloba, Annonaceae) and black walnut (Juglans nigra, Juglandaceae) from the eastern USA; many Myrtaceae species native to South America, such as gabiroba and cambuci (Campomanesia spp.), savanna pear, uvaia, cagaita, and the Brazilian cherry or grumixama (all Eugenia spp.); red bush apple or bemburrtyak (Syzygium suborbiculare, Myrtaceae) from Australia; Malabar plum or jambolao (Syzygium cumini, Myrtaceae) and emblic (Phyllanthus emblica, Phyllanthaceae) from the Indian subcontinent; jackfruit (Artocarpus heterophyllus, Moraceae), star fruit (Averrhoa carambola, Oxalidaceae), the Sapindaceae rambutan (Nephelium lappaceum) and lychee (Litchi chinensis) from Asia; marula (Sclerocarya birrea, Anacardiaceae) and mongongo nut (Schinziophyton rautanenii, Euphorbiaceae) from Africa, to name a few species of edible fruits. Even though some of these species are commercially produced, cultivation could be facilitated mainly by taming several traits (e.g., plant size and architecture, removing toxic metabolites, fruit shape, uniform ripening, and improved post-harvest qualities).

Despite the initial hurdles of awaiting years for commercial production to start atop breeding difficulties, tree crops fit perfectly into the perspective of investing in perennial plants to achieve agriculture sustainability because they do not require annual soil disturbances, need much less energy to cultivate than annual crops, have a much lower carbon footprint, and contributing to carbon sequestration and storage (Crews et al. 2018; DeHaan et al. 2020). Therefore, investing in improving cultivated species and expanding tree crop diversity in agriculture is imperative to meet the current environmental needs while keeping up with the demands of an increasing population in the twenty-first century. Here we explore this scenario in light of plant genomics and gene editing technology capabilities.

The challenges of breeding of perennials: genomics and genetic modification of tree crops

Tree crops are defined mainly by three characteristics: an elongated, erect stem (often leading to a large plant) with a woody structure (lignified stem), and a perennial growth habit that produce food or industrial raw material (Neale et al. 2017). Humans cultivate a very long connection with trees by consuming their fruits and nuts already a significant dietary portion of hominids and hunter-gatherers. Even though this relationship existed for many millennia, the process of tree domestication differed from that of annual crops, mainly because the selection of genetic variants was followed by clonal (asexual) propagation in many cases, ultimately leading to the modern cultivated varieties without fixed mutations (Spengler 2019). This means that, due to the high level of heterozygosity in their genomes (often due to obligatory cross-pollination in allogamous species), the progeny from an ideotype tree may not exhibit desired phenotypes because the traits are segregating wildly.

Most commercial tree improvement programs seek to mine the natural genetic variation in wild populations for desirable traits. The main aim is to maximize genetic gain per time unit related to yield and sustainability with the highest economic efficiency. To this end, breeding programs go through a cyclical selection process from a genetic base (e.g., natural populations and germplasm banks) to establish a breeding population. Often, breeders cast a wide net to establish their base population from a broad set of provenances, such as landraces, hybrids, and exotic varieties. They continue to infuse their base population with diverse genotypes at each generation, aiming to introgress favorable or new traits and reduce inbreeding depression to select the “best” genotypes within a given cycle. The criteria for “best” genotypes are defined by the breeding goals and targeted hardiness zones (edaphoclimatic conditions).

Establishing the genetic merit (breeding value) of a given individual may take several years for tree species, given their lengthy sexual reproduction cycle and poor juvenile-mature correlation for most traits of commercial importance. The breeding cycle length is determined by the duration of the plant’s juvenile phase, which corresponds to the time required for the first flowering to occur and can take from three years (e.g., stone fruit and citrus) to up to more than a decade (e.g., persimmon, avocado, and olive) when planted from seeds (Nocker and Gardiner 2014; Moreno-Alías et al. 2010). This extended juvenile phase of arboreal plants compared to annual crops also hampers the development of high-resolution genetic maps, which are needed for establishing the position of markers on linkage groups or chromosomes.

Plenty of genomic resources for tree crops have been developed. The first tree to have its genome sequenced and assembled to the chromosome level was Populus trichocarpa (poplar, Tuskan et al. 2006), which was published just 6 years after the publication of the first angiosperm genome (Arabidopsis thaliana, Kaul et al. 2000). We have summarized 46 arboreal genome publications since poplar, among which the most recurring genus is Citrus (Table 1), and some recent examples comprise the oil-tea tree (Camellia chekiangoleosa, Shen et al. 2022), the tropical fruit tree, lychee (Litchi chinensis, HU et al. 2022) and a wild, caffeine-free coffee species (Coffea humblotiana, Raharimalala et al. 2021).

Table 1 Cultivated tree species with sequenced genomes
Full size table

Genomic information on tree crops allows for advances in understanding plant evolution (Wu et al. 2014; Rendón-Anaya et al. 2019; Liu et al. 2020b), the dynamics of gene families (Pinto et al. 2019; Zhang et al. 2021b), genome-wide exploration, association studies (Pinto et al. 2021; Fahrenkrog et al. 2017; Zhang et al. 2021a), and population genomics (Wang et al. 2020; Chhetri et al. 2019). Also, the sequencing and analysis of the genetic pools of crop relatives, called pan-genomics (Gao et al. 2019; Alonge et al. 2020; Jayakodi et al. 2020; Tao et al. 2021), is shedding new light on understanding the influence of structural variants on traits and advancing gene discovery. Among tree crops, such resource is available for poplar and apple (Pinosio et al. 2016; Sun et al. 2020), and opportunities are wide open for other species.

Using genomic information with recombinant DNA technologies allied to tissue culture will pave the way to innovative genetic engineering in tree crops. Efficient in vitro regeneration and genetic transformation protocols are established for many tree crops with sequenced genomes. However, due to technical difficulties of the genetic transformation and regeneration process, besides the nontechnical aspects related to regulatory agencies and consumer acceptance, trees make less than 3% of the total approved transgenic crops worldwide (ISAAA 2022). The recent technological expansion of genetic breeding provided by cheaper genome sequencing and efficient gene editing (Jinek et al. 2012; Nadakuduti and Enciso-Rodríguez 2021) is helping to balance the cost–benefit for tree biotechnology and genetic research and leading scientists and companies to embrace gene editing to create new tree varieties. We propose that the road be open for de novo domestication of wild trees (Zsögön et al. 2018; Gasparini et al. 2021) with commercial potential or improve traits of semi-domesticated arboreal species (Petersen et al. 2012; Sganzerla et al. 2021).

A brief analysis of the expanding potential of CRISPR-based gene editing for crop genetic improvement

The trajectory of crop improvement is divided into four generations: (i) crop domestication and breeding via phenotype-based selection performed by independent farmers (starting circa 10,000 BCE); (ii) scientific breeding by mating designs, hybrid breeding, statistical analyses, and the use of synthetic fertilizers and pesticides, marked by the high yielding dwarf plants that culminated in the first Green Revolution (1950–1960); (iii) biotechnology through tissue culture, transgenics (genetically-modified organisms, GMO), and marker-assisted breeding (second Green Revolution: 1980–2000s) mainly focused on farming and retailer traits, such as the long shelf life products (e.g., Flavr Savr tomato), resistance to insects (Bt crops), the herbicide glyphosate (roundup ready crops), and the papaya ringspot virus (e.g., the rainbow hybrid), in addition to the high-beta-carotene rice (golden rice); (iv) which led to the current stage of plant breeding, with the advent and ample adoption of genomics, precision breeding (including molecular assisted-breeding), mining of big data, gene editing, and the rise of artificial intelligence (cf. Cheng et al. 2021), and often referred to as the third Green Revolution.

The current stage of crop improvement is marked by efforts to increase the sustainability of agricultural systems imposed by social and environmental pressures (Fernie and Yan 2019), facilitate farming and trading, as well as improve consumer-oriented traits such as food nutrition (e.g., high-lycopene and high-anthocyanin tomatoes, optimal lipid composition in canola), size and shape (watermelon and pepper), aspect (e.g., the frivolous non-browning Arctic apple), and scent and aroma (e.g., tomatoes and ornamental roses). Often, traits of interest are genetically complex due to being multigenic, which require specific allelic sets to produce the desired phenotype—this is especially the case of drought, temperature and salt tolerances, and nutrient uptake and use efficiency.

The advent of the CRISPR-based gene editing technology occurred after a demonstration that the bacteria immunity system could be harnessed as a molecular tool to perform programmed genetic changes in higher organisms less than a decade ago (Jinek et al. 2012—for a comprehensive historical review, cf. Lander 2016). Since its adaptation as a molecular tool, the key ability of the CRISPR system to be explored and used is the programmable DNA double-strand break (DSB), which is catalyzed by the nuclease Cas9 [and other type-2 Cas nucleases, such as Cpf1 (Zetsche et al. 2015, 2017)], guided by a programmable single-stranded guide RNA (sgRNA or just guide RNA), designed to be complementary to a targeted genome sequence (Jinek et al. 2012). DSB is repaired by the innate DNA repair machinery of the cell through two different pathways, the nonhomologous end joining (NHEJ) and homology-directed repair (HDR) (Jiang and Doudna 2017). The error-prone NHEJ is the most used option in somatic plant cells (Rozov et al. 2019). Due to the repeated action of the Cas nuclease on the targeted site, the region tends to be misrepaired, leading to a permanent mutation.

The disruption in the genetic code caused by the DSB misrepair can be targeted to virtually any segment of an organism’s genome by designing a complementary sgRNA. When targeting an open reading frame (ORF) of a gene, it can potentially disrupt the protein function; in the case of a UTR-ORF (uORF), it may affect the translation of a gene downstream (principal ORF), and when the mutation is programmed to a promoter region, it can influence its transcriptional activity (Fig. 1). Despite being less common, somatic plant cells can also use the HDR pathway in response to DSB. Incorporating a selected DNA fragment into a desired genomic position is possible by delivering a donor DNA template and the CRISPR-Cas reagents (i.e., Cas enzyme and sgRNA). Although this approach is technically challenging, significant progress has been made in developing this knock-in strategy in plants (e.g., rice: Lu et al. 2020; Dong et al. 2020).

Fig. 1
figure 1

The versatility of CRISPR systems. Cas proteins complexed with sgRNAs promote DSB, which can either generate mutations through the endogenous nonhomologous end joining (NHEJ) pathway or induce gene insertions through the homology-directed repair (HDR) pathway when a template is available. An enzymatically dead protein (dCas9) can be harnessed to assist with varied functions via protein fusions, such as transcriptional regulation, epigenetic modification, DNA fragment localization, and enhanced meiotic recombination. A Cas9 in which only one DNase domain is inactivated functions as a nickase (nCas9) by catalyzing one-strand DNA breaks and is used for base editing and prime editing. Mutations targeted to promoter regions, open reading frames (ORFs), or untranslated ORFs (uORFs) can induce transcriptional changes, knock out or knockdown of gene functions, and translational rates, respectively

Full size image

Beyond the outstanding breeding potential of the original DSB-inducing CRISPR-Cas system, manipulating the catalytic Cas9 DNase domains (RuvC and HNH) made it possible to induce single-strand breaks (nicking) on the DNA by deactivating either of the domains (nickase: nCas9) or turning the nuclease into a programmable binding protein by eliminating both DNase activities (dead-Cas9: dCas9) (Qi et al. 2013; Ran et al. 2013) (Fig. 1). Fusions of nCas9 and dCas9 with other proteins, such as transcriptional activator or repressors, cytidine or adenine deaminases, methyltransferases, reporter proteins, and meiosis-specific endonucleases (e.g., Spo11), creates powerful tools to manipulate transcription, induce specific single-base substitutions (C → T and A → G), epigenetic modifications (DNA methylation and histone modifications), visualize DNA fragments on the chromosome context, and alter recombination rates during meiosis (Larson et al. 2013; Paixão et al. 2019; Qin et al. 2020; Hilton et al. 2015; Dreissig et al. 2017; Sarno et al. 2017) (Fig. 1). Moreover, different versions of Cas proteins can be exploited for the versatility of PAM sites (the DNA recognition motif for the Cas protein, such as 5′-NGG for Cas9 or 5′TTTV for Cpf1/Cas12a) and substrate affinity (e.g., the RNA editing capability of Cas13), making the CRISPR system a very versatile molecular tool for genetic engineering (Manghwar et al. 2019).

Recent breakthroughs revolving around the CRISPR-Cas system include a method to induce small programmable insertions on specific genomic locations, named prime editing (Anzalone et al. 2019), which has already been tested on wheat and rice (Lin et al. 2020). Prime editing is based on a nickase (nCas9) fused to a reverse transcriptase (RT) guided by a prime editing-gRNA (pegRNA) that guides the complex to the targeted genomic region and contains the sequence of interest to be incorporated into the DNA by the RT action after nCas9 performs the nicking. Furthermore, recent developments extend the toolkit of gene editing in plants, such as the C:G to G:C editing system (Chen et al. 2021), the compact CRISPR-Casɸ system (Pausch et al. 2020), and the development of near-PAMless CRISPR-based genome editors (Walton et al. 2020). In addition to the examples mentioned above of CRISPR-Cas gene editing approaches applied to crops, we refer the reader to recent reviews to deepen the comprehension of this technology (Wada et al. 2020; McCarty et al. 2020; Mishra et al. 2020; Moradpour and Abdulah 2020; Anzalone et al. 2020).

CRISPR technology is constantly being adapted and generating new tools, reflecting its remarkable adaptability and potential for molecular biology and plant breeding. Noteworthy systems include CRISPR-associated transposases (CAST), which could be harnessed on eukaryotes to drive DNA insertions guided by long RNAs efficiently, thus surpassing an existing bottleneck on genome editing, i.e., the controlled insertion of specific sequences to a specific location of the genome (Klompe et al. 2019; Strecker et al. 2019; Saito et al. 2021). The application of the CRISPR technology in plants is a multidisciplinary effort that requires advances in tissue culture (e.g., in vitro regeneration rates in recalcitrant species, especially trees; and the efficient protoplast manipulation protocols for allogamous or polyploid species), genetic transformation, genome sequencing technologies and bioinformatics, and vector manipulation based on efficient cloning techniques, such as Gibson assembly, type-II restriction enzyme-based cloning (golden gate and modular cloning), BP/LR recombination (gateway), ligation-independent cloning (LIC), or gene synthesis (e.g., Čermák et al. 2017; Hahn et al. 2020).

The expanding capacity of gene editing makes crop designing possible and will facilitate the domestication and breeding of arboreal species. Figure 2 shows some gene-editing approaches for performing targeted genetic modifications that could be applied to the molecular breeding of trees.

Fig. 2
figure 2

CRISPR applicability for generating stable gene-edited tree crops. A Editing of genes already functionally characterized in tree species. B Orthologs genes identified in tree genomes and gene editing. C Genes or QTLs associated with traits of agricultural importance with potential for gene editing. D Editing anti-crossover factor genes to increase meiotic recombination. The resulting decrease in linkage drag can lead to increased genetic diversity in population studies. E Editing domestication traits in wild species to create novel, resilient crops through de novo domestication. F Generating transgene-free edited plants via the introduction of pre-assembled Cas9 and sgRNA into protoplasts. G Genome editing protocols that do not rely on tissue culture have been proposed by using RNA virus-based vectors with mobile sgRNAs or manipulating genes that regulate plant meristem identity maintenance via infecting growing plants ex vitro

Full size image

CRISPR-based approaches applied to tree crops

Even though the technology is relatively recent, gene editing has already become feasible in tree crops (Table 2; Suppl. Table S1). The first CRISPR-edited arboreal plant was the Chinese white poplar (Populus tomentosa, Salicaceae), which genome was manipulated to knock out the PtoPDS gene, resulting in photobleached, albino-like plants as a proof-of-concept for gene editing (Fan et al. 2015). The same strategy was later applied to other trees for protocol development: citrus (Citrus trifoliate × C. sinensis, Rutaceae), apple tree (Malus × domestica and Malus prunifolia × M. pumila, Rosaceae), grapevine (Vitis vinifera, Vitaceae), coffee (Coffea canephora, Rubiaceae), kiwifruit (Actinidia spp. Actinidiaceae), and banana (Musa spp.) (Zhang et al. 2017; Charrier et al. 2019; Osakabe et al. 2018; Nakajima et al. 2017; Breitler et al. 2018; Wang et al. 2018a, 2018b; Ntui et al. 2020).

Table 2 Tree species with genomes manipulated via gene editing
Full size table

Aiming to optimize gene editing capabilities in plants, studies are being conducted to explore the influence of technical aspects related to the CRISPR system. In V. vinifera, VvU3 and VvU6 promoters (native from V. vinifera) proved to be worth integrating a CRISPR-based editing strategy, as an increase of more than threefold sgRNA expression was achieved when regulated by these promoters compared to AtU6 (from A. thaliana and commonly applied for many crops gene editing strategies). The higher sgRNA expression resulted in a higher mutation rate (e.g., 23.53% for the AtU6-based vector versus 43.24% for VvU6.2) (Ren et al. 2021). Similarly, the effect of using different promoters was tested in poplar, in which expression of Cas9 driven by the MAS promoter (from Mannopine synthase) led to higher mutation rates compared to the conventional double CaMV35S promoter (75% for MAS against 67.5% for 2xCaMV35S: An et al. 2021).

Base editing enables the exploration of already characterized single nucleotide variations to directly modify traits of a variety without the need for time-consuming crossing cycles. This is especially interesting and was already applied to tree crops, such as pear (Pyrus communis, Malabarba et al. 2020), apple (Malus domestica, Malabarba et al. 2020), and poplar (Populus alba × P. glandulosa, Li et al. 2021). For pear and apple, researchers explored the cytidine deaminase fused to a nickase (SpnCas9-PmCDA1-UGI) to induce base edits on the PDS and ALS genes as a proof-of-concept of the system feasibility, and they successfully obtained edited lines for both targets. Similarly, a broad study was performed to optimize the base editing strategy in P. alba × P. glandulosa, in which two cytosine base editor complexes were compared (PmCDA1-BE3 and A3A/Y130F-BE3) was tested alongside an adenine base editor (ABEmax_V1/V2). High rates (> 90%) of base editing were obtained for all tested systems in at least one of the targets of the study, even though the activity window varied significantly between the two cytosine editors.

Although some of its potentials has been well demonstrated, gene editing is still a relatively new technology that has yet to reveal its full potential in tree crops. Genetic transformation and in vitro regeneration processes remain challenging and time-consuming for many species. As a result, only a few reports have been published demonstrating the feasibility of gene editing in trees, in contrast to the many publications applying CRISPR systems to modify agricultural traits in annual crops. Still, an expanding number of papers describing mutations induced by CRISPR aim to rationally modify traits of interest in trees. One such example comprises efforts to cope with canker disease in citrus by mutating the coding region and the promoter of CsLOB1, which led to the development of resistant plants (Jia et al. 2017; Jia and Wang 2020; Jia et al. 2022). Also related to pathogen resistance, a WRKY transcription factor (VvWRKY52) was targeted in grape (V. vinifera). Its knockout resulted in a cv. Thompson seedless mutant lines with the reduced spreading of lesions caused by the inoculation of Botrytis cinerea (Wang et al. 2018a).

Recent works also demonstrate the capacity to tackle morphological traits in tree crops. In citrus (Citrus sinensis × Poncirus trifoliata), scientists identified the molecular basis of thorn development regulation and manipulated plant architecture by converting thorns into branches through the disruption of TI1 and TI2 transcription factors (Zhang et al. 2020). Moreover, the simultaneous knockout in the poplar of two paralogs orthologous to the rice TAC1 (Tiller Angle Control 1) led to the development of plants with upright-oriented leaves. This novel phenotype can be harnessed for commercial plantations to increase the density of individuals planted per area, increasing the overall yield (Fladung 2021). Since poplar is cultivated for wood and has potential for biofuel production, initial studies are also being conducted to investigate the genetic basis of lignin formation and the possibilities of manipulating its content, with recent examples of gene editing aimed to alter cell wall composition for increased saccharification efficiency (Jiang et al. 2021; Li et al. 2021).

Although not considered a crop due to lack of commercial exploration, it is worth highlighting gene-editing work conducted on the tropical tree Parasponia andersonii (also called Trema andersonii, Cannabaceae). This is a non-leguminous, nitrogen-fixing species with a recently developed genetic transformation protocol that is relatively fast and efficient for the application of CRISPR strategies (Zeijl et al. 2018). The main studies published so far regards the investigation of the genetic basis of nodule formation for symbiotic nitrogen fixation, which has been poorly explored in trees (Zeijl et al. 2018; Bu et al. 2020).

Potential gene targets in arboreal species for improving cultivation and sustainability

Traits of interest for tree crop cultivation are, in many cases, like annual crops (e.g., fruit size, root depth, and leaf shape), and the genetic basis governing these aspects are also most often conserved in different species. Thus, one approach to identifying genes for manipulating trees for a given trait is to use the knowledge gained from model species and annual crops. For that, in addition to sequence similarity, synteny and gene expression profiles should be added to improve the process of identifying functional orthologs.

For many fruit trees, larger fruit is a desirable trait. Scientists may be able to take advantage of studies in tomato (Solanum lycopersicum, Solanaceae), which is a model for fresh fruit (berry) development. For example, the AP2/ERF transcription factor excessive number of floral organs (ENO) directly inhibits the expression of WUSCHEL (WUS, a controller of meristem size) in the floral meristem, and its loss-of-function resulted in the development of a larger fruit (Yuste-Lisbona et al. 2020). If this gene is functionally conserved and its loss of function leads to little pleiotropic effects, it could be a good target for gene editing in fruit trees, including semi-domesticated species. Likewise, a well-conserved regulator of shoot apical meristem (SAM) development is the peptide signal CLV3, which mutations in either its coding region or promoter have been shown to lead to fruit increase in tomatoes (Zsögön et al. 2018; Li et al. 2018).

Another critical aspect in many crop species is photoperiod sensitivity, which is related to meristem determination and often dictates cultivation latitudes. In tomatoes, loss-of-function of the flowering repressor SELF PRUNING 5G (SP5G) gene, a member of the CETS family, promotes early, photoperiod-independent flowering and, consequently, an earlier harvest (Soyk et al. 2017; Zsögön et al. 2018; Li et al. 2018). Functional studies of the CETS family may lead to the regulation of flowering induction in arboreal plants by revealing a roadmap to manipulating juvenile phase length, thus accelerating breeding and speeding up commercial production. The simultaneous mutations in SP5G, SP (another CETS gene), and SlER (a leucine-rich receptor kinase) led to a very compact tomato plant (Kwon et al. 2020), which is another highly desired trait for many tree crops.

Growing tall is a common trait of trees and encompasses a significant advantage in the race for light. However, in an agricultural setting, crops can be neatly arranged and managed to minimize competition for light. Therefore, growing too tall can often lower profits due to the challenges of harvesting the commercial product. A strategy exploring dwarfism-inducer mutations could be interesting when height is a logistical problem. Fortunately, there are some examples of using CRISPR-based gene editing to achieve this goal in plants. In rice, directed mutations on the gene SD1 (Semi-Dwarf1) led to shorter plants (at least 30% of height reduction, depending on nitrogen supply) without significantly affecting yield in elite landraces (Hu et al. 2019). Similarly, targeting ERECTA homologs in cucurbits (Cucumis melo: melon; Cucurbita moschata: squash; and Cucumis sativus: cucumber) efficiently decreased stem length (Xin et al. 2022).

Although fruit weight and yield are critical, agricultural systems must also prepare for increased resource limitations (energy, fertilizers, and freshwater) and climatic instabilities (heat, rainfall patterns, and wind). Critical traits for this scenario include increased tolerance to drought, heat, and salinity, improved phosphate and nitrogen uptake and use, and increased resistance to pests and diseases. To support gene editing strategies focused on such traits, we compiled a list of target genes whose losses of function led to increased fitness in agricultural systems under environmental pressures (Suppl. Table S2). We also refer the reader to recent reviews on annual crops (Gonçalves et al. 2020; Janni et al. 2020; Huisman and Geurts 2020).

A recent report describes the influence of a B-type cyclin (NtCycB2) on glandular trichome formation of tobacco (Wang et al. 2022). Its knock-out by gene editing enhanced trichome formation and more compound exudation, resulting in stronger aphid resistance, a compelling achievement for pest management. Thus, potential functional orthologs of this gene could be targets for pest resistance in tree crops (Therezan et al. 2021).

Works describing induced mutations affecting abiotic stress tolerance have been recently published. The loss of function of Aluminium-activated Malate Transporter (ALMT and SlALMT15), which affects stomatal density in tomato leaves, enhanced tolerance to drought without penalties to the net photosynthetic rate in tomatoes (Ye et al. 2021). By the same token, the OsTT2 loss of function led to greater wax retention and increased heat tolerance (Kan et al. 2022). Moreover, two recent pieces exploring nutrient uptake using CRISPR-induced mutations led to increased uptake of iron (AtERF96 knockout) and phosphate (SlHA8 knockout) in Arabidopsis and tomato, respectively (Liu et al. 2020a; Yao et al. 2022).

Genome editing strategies for tree breeding to take off

Even though gene editing applied to tree species is challenging, partly because of the lack of knowledge on the genetic basis of many traits, some promising innovations involving CRISPR systems can facilitate the pipeline, such as the high-throughput mutagenesis recently applied to maize (Zea mays, Poaceae) (Liu 2020). This approach involved the application of multiplexed CRISPR-Cas9 on a batch pipeline with pooled transformation and barcoded deep sequencing, resulting in 118 mutated genes (412 different mutations) in the population. The same strategy could be applied to trees with sequenced genomes and efficient genetic transformation protocols (e.g., citrus and poplar) to identify causal genes for traits of interest without time-costly crossing (as in conventional functional genetics studies) or random mutation experiments.

Indeed, even though the candidate gene approach is interesting for unveiling causal genes coordinating traits of interest, it does not replace population segregation analyses. Such studies can reveal the genetic bases of novel traits without any previous knowledge of the genomic location of the variant. Fortunately for tree geneticists, some recent results demonstrate the capability of targeted mutagenesis for accelerated breeding. The analysis of an EMS-derived mutant collection showed that a disruption on helicases RECQ4 and FANCM, which are anti-crossover factors, resulted in over two-fold increase of meiosis recombination in three distantly related species: rice (Oryza sativa, Poaceae), tomato (Solanum lycopersicum, Solanaceae), and pea (Pisum sativum, Fabaceae) (Mieulet et al. 2018). The manipulation of crossover factors can increase recombination frequencies and faster develop high-definition genetic maps. In addition to accelerating the construction of genetic maps, the controlled recombination with directed nucleases promises a paradigm shift for breeding by increasing population diversity, fine-tuning introgressions during pre-breeding, and obtaining favorable haplotypes for breeding (Taagen et al. 2020).

The increase of genetic recombination frequencies is interesting for breeding strategies involving introgressions from related species to elite genotypes since higher recombination rates increase the chances of isolating the desired gene from nearby genetic elements, thus decreasing the linkage drag that could affect genetic gains. Indeed, the use of CRISPR-Cas9 to disrupt RECQ4 was successfully applied to tomatoes for this purpose. Interspecific tomato hybrids (S. lycopersicum × S. pimpinellifolium) with the loss-of-function for recq4 showed higher recombination frequencies than the wild type (Maagd et al. 2020). This approach paves the way for applications in the breeding of arboreal species, in which performing successive crossing cycles to attenuate linkage drag is difficult, costly, and time-consuming.

It is helpful to bear in mind that, while gene editing can be useful to create novel, improved genotypes, the presence of transgenic CRISPR cassettes can affect marketability, for example, by GMO labeling in some countries or refusal by a significant portion of consumers. Nonetheless, while gene-edited crop regulation is politically fluid and highly variable from country to country, the resulting variety does not necessarily contain a transgene—thus, technically being not a GMO (Metje-Sprink et al. 2020).

Segregating the transgene out of a genome requires at least one backcrossing cycle, representing many years for a tree. It can also be impractical for elite varieties that are not inbred lines, as with most tree crops. A promising alternative for such cases is the transgene-free strategy uses ribonucleoprotein (RNP) complexes. It requires the purified nuclease to be pre-assembled in vitro with a designed sgRNA. Both Cas9 and Cas12a (Cpf1) were successfully tested for this strategy. The Cas-sgRNA complex can be delivered into protoplasts via PEG-mediated osmotic shock (Woo et al. 2015; Kim et al. 2017; Brandt et al. 2020) or directly into explant somatic cells via shotgun (Liang et al. 2018; Dong et al. 2021). This method has been reported for apple, grapevine, and rubber tree (Hevea brasiliensis, Euphorbiaceae) (Malnoy et al. 2016; Osakabe et al. 2018; Fan et al. 2020). Delivery of plasmids containing Cas9 and sgRNAs directly to protoplasts by inducing the transient expression and selecting transgene-free events is also a possibility (Lin et al. 2018), albeit not reported on arboreal species yet.

The most considerable caveat of these transgene-free delivery methods is the need for efficient tissue culture and in vitro regeneration protocols, which have not been optimized for most tree species. Therefore, the ideal scenario for the broad application of CRISPR technology on tree species is a transgene- and tissue-culture-free gene-editing method. Two recent reports show that gene-edited plants can be obtained without tissue culture (Ellison et al. 2020; Maher et al. 2020). One approach infects ex vitro a transgenic plant (Nicotiana tabacum, Solanaceae) expressing the Cas9 protein with a tobacco rattle virus (TRV)-based vector containing sgRNAs cleverly fused with a translocable mRNA (FT gene product). This in planta transformation resulted in 60 to 100% of edited progeny when a single genomic site was targeted and up to 30% of edited progeny when three sites were targeted (Ellison et al. 2020). The second approach induced de novo development of plant meristems in Cas9-expressing plants by concomitantly expressing developmental regulators related to meristem identity maintenance (WUS2, IPT, STM, and BBM) and gene-editing reagents (sgRNAs targeting the PDS gene). Shoots regenerated from the de novo-induced meristems with fixed mutations, and some of them were transgene-free (Maher et al. 2020). These approaches highlight the potential of enabling gene editing tree species by circumventing transformation and in vitro regeneration.

In addition to editing the genome of elite varieties of cultivated species, we should also better explore the existing genetic diversity of plants by using non-domesticated species. By studying domestication traits and uncovering their genetic bases, we will be able to convert wild plants into novel domesticated crops by tackling those domestication traits yet preserving critical adaptive traits, such as strong levels of resistance to abiotic and biotic stresses. This concept has been proposed as de novo domestication (Zsögön et al. 2017; Gasparini et al. 2021; Curtin et al. 2022; Zsögön et al. 2022) and has been shown to be feasible for wild tomatoes (Zsögön et al. 2018; Li et al. 2018), as well as the semi-domesticated groundcherry (Physalis pruinosa, Solanaceae) (Lemmon et al. 2018). As the genetic bases of domestication traits become more understood and biotechnological tools are developed for perennial species, domesticating new plant species for agricultural purposes will become increasingly feasible, including arboreal species.

Alongside the technology development and the conceptualization of disruptive projects and perspectives on gene editing in tree species, improvements in legislation aspects around the globe must be performed to update the laws in the face of these new techniques. Some regulatory agencies (i.e., the Forest Stewardship Council (FSC), which certifies cellulosic products from responsibly grown forest crops around the globe) struggle to incorporate the new paradigm that arises with the advancement of genome editing technologies, like the fact that a crop can bear editions in its DNA at a specified target site without carrying any transgenes. Members of the scientific community are participating in the discussion regarding these matters (Strauss et al. 2019; Harfouche et al. 2021), and this is primordial to avoid misconceptions regarding genetic manipulation of crops and demonstrate the consequent possibilities of advancing scientific knowledge and increase the sustainability of modern agriculture responsibly.

Concluding remarks

The cultivation of arboreal species is challenging due to the long juvenile phase and the time required until harvest. New breeding strategies are being developed to cope with these drawbacks, and substantial advancements in the molecular design of tree crops are underway. The application of CRISPR-based gene editing technologies expands the horizon of genetic improvement programs and, given its versatility, along with the development of new cloning techniques, DNA sequencing technology, plant transformation platforms, molecular biology, and plant genetics, will enable advancing agriculture to the point of tailoring crops and domesticate de novo wild species to meet society needs. This is urgent, albeit particularly challenging, for the tree species we cultivate as crops since these plantations must be resilient to all agricultural threats that are forecast for the near future. Ergo, now is the time to plan for the coming decades, and the potential of gene editing for tree breeding awaits to be fully explored.