Progress and challenges in sorghum biotechnology, a multipurpose feedstock for the bioeconomy

Sorghum is a multipurpose crop resilient to suboptimal growth conditions. We highlight what genetic, genomic, and biotechnological resources are available for sorghum, with an emphasis on transformation technologies.

Abstract

Sorghum [Sorghum bicolor (L.) Moench] is the fifth most important cereal crop globally by harvested area and production. Its drought and heat tolerance allow high yields with minimal input. It is a promising biomass crop for the production of biofuels and bioproducts. In addition, as an annual diploid with a relatively small genome compared with other C₄ grasses, and excellent germplasm diversity, sorghum is an excellent research species for other C₄ crops such as maize. As a result, an increasing number of researchers are looking to test the transferability of findings from other organisms such as Arabidopsis thaliana and Brachypodium distachyon to sorghum, as well as to engineer new biomass sorghum varieties. Here, we provide an overview of sorghum as a multipurpose feedstock crop which can support the growing bioeconomy, and as a monocot research model system. We review what makes sorghum such a successful crop and identify some key traits for future improvement. We assess recent progress in sorghum transformation and highlight how transformation limitations still restrict its widespread adoption. Finally, we summarize available sorghum genetic, genomic, and bioinformatics resources. This review is intended for researchers new to sorghum research, as well as those wishing to include non-food and forage applications in their research.

Introduction

Sorghum [Sorghum bicolor (L.) Moench] is the world’s fifth largest cereal crop by acreage and production (FAOSTAT, https://www.fao.org/faostat/en/#data). It is an important staple food in the semi-arid tropics of Asia and Africa. Globally, sorghum is used for animal feed, fodder, and high-value products such as syrup and bioethanol. Harboring traits such as tolerance to drought, waterlogging, and salinity make it a highly productive crop in environmental conditions that restrict the cultivation of other cereals (Hadebe et al., 2017; Huang, 2018). Sorghum has also been the source of exciting advances in fundamental biology such as the discovery of a metabolon for dhurrin biosynthesis (Laursen et al., 2016) and a new gene and chemistry involved in conferring Striga resistance (Gobena et al., 2017). Although sorghum holds great promise, it is still underutilized. In this review, we will present the current state of research employing sorghum as a multipurpose feedstock for the bioeconomy, summarize available research tools with a focus on transformation and genetic engineering, and identify promising areas for future research.

Cultivated sorghum (Fig. 1A) can be classified into five basic races: bicolor, guinea, caudatum, kafir, and durra, which are differentiated by the phenotype of their mature panicles and spikelets (Harlan and Wet, 1972) (Fig. 1B). Sorghum bicolor (L.) Moench subsp. bicolor contains all the cultivated sorghum varieties (Dahlberg, 2000). Sorghum can also be classified based on its agronomic characteristics into forage, biomass, sweet, and grain types (Table 1). Forage sorghum is tall, and the biomass is used to feed livestock. Important traits of forage sorghum include digestibility, nutrient content, and palatability. Biomass sorghum is bred to maximize vegetative yields, with reports of up to 61 Mg ha^–1 (Snider et al., 2012), but, unlike forage sorghum, palatability is not a concern. Some of the original biomass breeding stock was derived from forage sorghum, so high-biomass sorghums can also be produced for forage (Venuto and Kindiger, 2008). Dedicated biomass sorghum is used to produce biofuels and chemicals from the lignocellulosic biomass (cell wall), fibers for biomaterials, and biogas via anaerobic digestion (Reddy and Yang, 2005; Wannasek et al., 2017; Silva and Vermerris, 2020). Sweet sorghum accumulates large amounts of soluble sugars (sucrose, glucose, and fructose) in its stems and was initially identified as an alternative sugar source in areas unsuitable for sugarcane production. Besides its use for syrup production, it can also be used for biofuel production and high-sugar forage (Rooney et al., 2007). Sorghum is typically a photoperiod-sensitive plant, requiring short days (8h/16h light/dark) to transition from the vegetative to the reproductive stage. Hybrid grain sorghum is photoperiod insensitive, meaning it can flower rapidly even in the summer in temperate regions, and therefore has shorter stature and reaches maturity earlier (Smith and Frederiksen, 2000). Grain sorghum is grown for its seeds and is used as a staple food mainly in the semi-arid tropics of Asia and Africa, as animal and poultry feed, as well as a sugar source for distillation into alcohol. Recently, grain sorghum has become more popular in other countries because of its health benefits, such as reducing rates of cardiovascular disease, obesity, and certain types of cancer (reviewed in Awika and Rooney, 2004). Certain genotypes contain 3–4 times more anthocyanin, a plant pigment which has antioxidant properties, compared with other grains (Awika et al., 2004). It is also a gluten-free alternative for people with celiac disease. However, in countries such as the USA, grain sorghum is primarily used to feed livestock and produce pet food, with approximately one-third of its production being directed to produce biofuels (United Sorghum Checkoff Program, https://www.sorghumcheckoff.com/).

Fig. 1.

Sorghum plant morphology (A) and panicle and spikelet phenotypes of the five basic races (B). The race bicolor is the most primitive of the cultivated races and has upright semi-open panicles, with long and clasping glumes. Commercially cultivated sorghum tends to be a mixture of these major races. The race guinea originated in humid regions of West Africa and has open, elongated panicles, which helps decrease mold infection. Caudatum originated in eastern Africa and has panicles ranging from compact to open, with shorter, asymmetric, glumes that expose the grain. On the other hand, kafir, which originated in southern Africa, has tighter and longer panicles. Durras have compact panicles and originated in southern Sahara.

Table 1.

Characteristics of sorghum groups

	Forage	Sweet	Grain	Biomass
Height (m)	1.8–3.6	>3	0.6–1.2	3.5–6
Traits	Single or multicut harvest, digestibility, nutrient content, palatability	Large amount of soluble sugars in stems	Photoperiod sensitive and insensitive, high grain yield	Photoperiod sensitive, dual-purpose, high lignocellulosic biomass
Uses	Livestock feed	Syrup and biofuel production, high-sugar forage	Seed as staple food in some regions, livestock feed and biofuel production	Biofuel, biogas, and biomaterial production

Independent of the usage type, sorghum is an attractive crop for cultivation in a wide range of environments (tropical, subtropical, and temperate regions) and in soils that are considered marginal for other food crops such as maize (Fu et al., 2016; Ameen et al., 2017). Sorghum can grow in mineral-rich soils with pH values that limit profitable cultivation of other crops (Smith and Frederiksen, 2000). It also requires less water and exhibits drought and waterlogging tolerance (Rosenow et al., 1983; Promkhambu et al., 2010; Varoquaux et al., 2019). As an adaptive mechanism, sorghum becomes dormant during severe drought conditions and resumes growth when re-exposed to water (reviewed by Assefa et al., 2010). Another post-flowering drought adaptation is known as non-senescence or stay-green (Borrell et al., 1999, 2000; Borrell and Hammer, 2000). The stay-green trait allows delayed remobilization of nitrogen in the leaves, maintaining photosynthetic activity and carbohydrate supply to the developing grain, which results in higher biomass and grain yield (Borrell et al., 1999, 2000; Borrell and Hammer, 2000). For instance, stay-green sorghum hybrids can produce 47% more post-anthesis biomass under drought conditions (Borrell et al., 2000). Along with drought tolerance, sorghum is also heat tolerant (Craufurd and Peacock, 1993; Nguyen et al., 2013), which is particularly relevant, considering climate change predictions that include reductions in rainfall and increases in temperature in many cereal-growing regions.

Sorghum as a multipurpose feedstock for the bioeconomy

While sorghum is an important staple food and forage crop globally, it has potential as a feedstock for renewable fuel and bioproducts (U.S. Department of Energy, 2016, https://www.energy.gov/eere/bioenergy/2016-billion-ton-report). For it to be a viable feedstock, agronomic and biomass compositional traits will likely need to be further developed to make the economics of the manufacturing processes comparable with those for fossil fuel-derived products (Baral et al., 2019; Yang et al., 2020, 2021). This is exemplified by the US Department of Energy Bioenergy Research Centers, which have been funded since 2007 to investigate all aspects of advanced biofuel production process (https://genomicscience.energy.gov/centers/). Sorghum is one of three DOE flagship biomass crops, and open research questions include biomass improvement and co-production of valuable chemicals. Sorghum’s versatility in multiple processing configurations is one of its key appeals (Stamenković et al., 2020). For example, biodiesel can be produced from sorghum grains after pressing and transesterifying lipids (Ved and Padam, 2013). Starch from the grain or the sucrose-rich juice from the stems of sweet sorghum can be used for fermentation into biofuels and bioproducts. Beyond this, sorghum, especially the photoperiod-sensitive varieties, can produce large amounts of aerial lignocellulosic biomass that can also be used as a sustainable and economically feasible feedstock for conversion. Because of sorghum’s versatility, designing an ideal sorghum ideotype is challenging (Yang et al., 2021). Instead, it is more likely that a range of sorghum varieties will continue to be developed, with their phenotype tuned to the desired downstream market.

Target traits for biomass improvement: the cell wall

The cell wall is a crucial organelle for cell structure and protection, and is made up mainly of cellulose, hemicelluloses, and lignin. Cellulose, which constitutes 25–35% of the sorghum biomass, is made of β-1,4-glucose chains, which in turn form crystalline fibrils via hydrogen bonding (Ioelovich, 2008; Polko and Kieber, 2019). Hemicelluloses are a collection of branched hetero-polysaccharides (Ebringerová et al., 2005), but in the sorghum cell wall, glucuronoarabinoxylans dominate, making up ~35% of the total biomass (Anglani, 1998; Xu et al., 2018). Lignins are complex branched polyphenolics, made up of monolignol subunits derived from phenylalanine and tyrosine metabolism, and are found only in some secondary cell walls (Boerjan et al., 2003). Sorghum lignin content varies between ~2% and 11% of dry matter depending on the cultivar (Brenton et al., 2016), and is a key factor affecting forage palatability and biorefinery efficiency.

During biomass processing in a biorefinery, cellulose, hemicelluloses, and soluble sugars can be converted to monosaccharides, which can then be utilized as a carbon source by microbes. Most microbes preferentially use hexose sugars (such as the glucose in cellulose) over pentose sugars (such as the xylose and arabinose in xylan), so biomass with a high hexose:pentose ratio, namely reduced xylan, is preferable (Brandon et al., 2020). Branched hemicelluloses such as xylan require multiple enzymes to hydrolyze them to monosaccharides, so hemicelluloses with fewer branches, or altered branch frequency, may also be preferable (Gao et al., 2020). However, the cell wall should not be weakened so much that the plant lodges in the field or is more susceptible to pathogens and pests. Susceptibility to lodging and diseases due to cell wall modifications have proved difficult to predict, with some plants with engineered walls being more resistant to pathogens (reviewed by Miedes et al., 2014).

In addition to sugar engineering, lignin can be modified for biomass improvement. An ideal biomass feedstock would have low lignin, since it both physically shields polysaccharides from polysaccharide-degrading enzymes and reduces enzymatic efficiency via non-specific binding. With the advent of designer lignins and use of microbes that can consume phenolics as a carbon source, monolignols are increasingly considered high-value intermediates for the production of important biochemicals (Eudes et al., 2014; Karlen et al., 2016; Baral et al., 2019). Therefore, the desired biomass phenotypes (the sorghum biomass ideotype) will vary depending on the final target product. As a plant breeding problem, this variability highlights the need for seed producers to be able to respond rapidly to needs in the supply chain beyond their direct market (farmers), as the bioeconomy develops.

The isolation of naturally occurring lignin mutants has already proved beneficial for commercial sorghum cultivars. Nineteen Brown midrib (Bmr) mutant loci have been identified in sorghum, though only 3–4 loci are considered of agronomic interest due to their lower lignin content and higher potential for biomass conversion (Porter et al., 1978; da Silva et al., 2018). Engineering approaches that re-route the lignin biosynthetic pathway have been demonstrated in a number of plant species (Fu et al., 2011; Eudes et al., 2014; Wilkerson et al., 2014; Yan et al., 2018). Restricting engineering to specific cell types has been successful in reducing lignin while avoiding stem weakness (Yan et al., 2018).

Beyond biomass: oils, bioproducts, and novel materials

In addition to being a source of starch and lignocellulose to produce biofuels, sorghum has the potential to function as a factory for other bioproducts or their precursors, and this will be important for the economic success of advanced biofuels. Compared with microbial production systems, in planta production of chemical compounds can reduce inputs, costs of post-production conversion steps, and the amount of pathway engineering needed (Yang et al., 2020). Proposed examples include pharmaceuticals (artemisinin and cannabidiol), materials (e.g. latex), insecticides (e.g. limonene), and plastic precursors [e.g. polyhydroxybutyrate (PHB)]. Modeling has shown that the added value of bioproducts can lower biofuel production costs to prices competitive with fossil fuels, as well as providing a better farmgate price for growers (Yang et al., 2020). Another promising route for sorghum metabolic engineering is to target triacylglycerol (TAG) accumulation in leaves for oil production, which can be used for biodiesel production. Although vegetative organs represent most of the above-ground biomass, leaves accumulate <1% lipids (Yang and Ohlrogge, 2009), so plant oil production relies on seeds rich in TAG. However, up to an 8.4% increase of TAG in leaf tissues has been achieved in sorghum by simultaneous overexpression of the genes encoding the maize transcription factor WRINKLED1, Umbelopsis ramanniana acyltransferase UrDGAT2a, and Sesamum indicum oil body protein OLEOSIN-L, providing a basis for further improvements in levels of extractable oil for commercial purposes (Vanhercke et al., 2019).

Lignin valorization is another attractive option to add value to compounds from waste products in a biorefinery (Mottiar et al., 2016). Potential high-value applications of lignin range from synthesis of lignin nanotubes for gene delivery (Ten et al., 2014) to development of lignin-based antibacterial products for pharmaceutical and biomedical industries, demonstrating the wide range of properties that can be exploited (Grossman et al., 2020). Lignin precursors have been re-routed in tobacco to produce intermediates that can be converted by an engineered microbial chassis to produce high-value compounds pyrogallol and cis,cis-muconic acid (Wu et al., 2017). Similar approaches could be applied to re-route higher levels of valuable intermediates in sorghum, although it will require better understanding of the regulation of cell wall biosynthesis pathways. Lignin valorization into phenolic compounds such as eugenol is also of great interest. Eugenol can be used in food, cosmetics, and pharmaceutical industries, and its high demand can lead to high market value (Martinez-Hernandez et al., 2019). Techno-economic analysis (TEA) and life cycle assessments (LCA) have shown that lignin valorization into eugenol and other methoxyphenols can reduce the cost of ethanol production by up to 23% and reduce greenhouse gas emissions by up to 78% compared with the petrochemical industry (Martinez-Hernandez et al., 2019). As demonstrated by this and other examples (Yang et al., 2020), TEA and LCA are important resources to guide decisions on which compounds should be targeted for genetic engineering, based on their economic value.

Finally, novel materials can be produced from biomass. For example, cellulose derived from lignocellulosic material can be broken into nanofibers, which have nanostructure favorable to high mechanical performance of nanofiber networks and composite materials (Sehaqui et al., 2010). Cellulose nanofibers are a great renewable material for the manufacturing of ultrafiltration membranes and can also be used as barrier layers in packaging material, among other useful applications (Forde et al., 2016). Additionally, both hemicelluloses and pectins have been suggested for use in a range of materials which include medical devices (Zheng et al., 2020), superconductors (Di Giacomo et al., 2015), and biodegradable packaging (Gouveia et al., 2019; Mendes et al., 2020). Collaborations between sorghum researchers and material scientists to develop new uses for biomass components or to engineer improvements are likely to be fruitful.

Barriers to using sorghum in biotechnology applications

There are three major barriers to the use of engineered sorghum: technical challenges around sorghum transformation, general societal concerns about engineered crops, and specific concerns about sorghum gene flow to weedy relatives. We will not dwell on the GMO issue here because it is reviewed in depth in the literature.(McHughen and Wager, 2010; National Academies of Sciences, Engineering, and Medicine, 2016; Wolt, 2017; Callaway, 2018; Spicer and Molnar, 2018; Waltz, 2018; Zhang et al., 2020).

Though we describe many examples of existing transgenic sorghum technology, to our knowledge, there is no transgenic sorghum grown commercially. One main reason for the limited use of transgenic sorghum in the USA is concerns about gene flow to its sexually compatible wild weedy relatives such as Johnsongrass (S. halepense), S. bicolor subsp. drummondii, and S. bicolor subsp. verticilliflorum via pollen dispersal and subsequent cross-pollination and hybridization. These wild relatives can easily hybridize with the cultivated sorghum to produce the noxious weed shattercane (Ejeta and Grenier, 2005). Strategies to limit gene flow, such as male sterility, could be implemented, as could agronomic strategies which monitor for compatible weedy species within the range of pollen flow. For example, in sorghum, it has been estimated that after 700 m, very little, if any, outcrossing would be expected (Schmidt and Bothma, 2006). It is also important to note that most of the discussed modifications would likely be considered ‘null’; that is, they would not be expected to give weedy relatives a selective advantage. This makes regulation more straightforward than traits such as herbicide tolerance.

The rapid development of transgenic sorghum varieties will be necessary to complement gains from traditional sorghum breeding, as humanity faces increasing challenges from climate change, degraded soils, and increased population. In the next section, we will give an overview of the transformation methods adopted for sorghum biotechnology thus far and discuss the main bottlenecks that need to be addressed to have efficiencies comparable with other grasses and move the field forward.

Sorghum transformation

The limited ability to transform sorghum is the major barrier to the widespread adoption of sorghum as a research model and as feedstock for the growing bioeconomy. Sorghum transformation is technically challenging, comparatively costly and time-consuming, and limited to a few genotypes. Sorghum is highly recalcitrant to tissue culture and transformation, mainly because of genotype-dependent responses, production of phenolic compounds, short-term plant regeneration ability, and acclimatization issues (the ability of plants to survive the transfer from in vitro culture to soil) (Maheswari et al., 2006; Altpeter et al., 2016). Here, we describe the achievements so far, and outline research questions that would help resolve existing barriers to sorghum engineering.

Since transgenic sorghum was first described (Casas et al., 1993), many improvements have been reported (Fig. 2). Casas and colleagues used immature embryos from the genotype P898012 to induce callus formation for particle bombardment, and obtained a transformation efficiency of 0.3% (Casas et al., 1993). Since then, the process has been improved using the genotype Tx430, and reached efficiencies of up to 46.6% (Belide et al., 2017). Using Agrobacterium tumefaciens to introduce the transgene via infection, transformation efficiency has increased from 9.7% in the initial studies (Zhao et al., 2000) to 33.2% (Wu et al., 2014). An important factor for tissue culture and, consequently, transformation success, is genotype selection. For the past 10 years, the grain sorghum inbred line Tx430 has been routinely used due to its consistently high callus induction and regeneration frequencies (Howe et al., 2006; Gurel et al., 2009; Liu and Godwin, 2012; Wu et al., 2014; Liu et al., 2015; Belide et al., 2017). However, Tx430 was directly compared with seven bioenergy parental sorghum lines using the protocols from Liu and Godwin (2012) and Wu et al. (2014). While Tx430 had high callus proliferation accompanied by low phenolic release, lines PI329311 and Rio had the best regeneration rates (Flinn et al., 2020).

Fig. 2.

Timeline of advances in sorghum transformation. Cat, chloramphenicol acetyltransferase; NptII, neomycin phosphotransferase II.

The explant source also plays a role in transformation efficiency. A variety of explants, such as immature and mature embryos, immature inflorescences, leaf discs, leaf whorls, and shoot meristems, have been used (Tables 2, 3). The most successful studies have used immature embryos due to their high embryogenic and regeneration competence (Tables 2, 3). However, the plant needs to reach the reproductive stage, which is limited to specific seasons or periods of time, and the narrow time window of 10–15 d in which the immature seeds need to be collected. To overcome these drawbacks, Silva et al. (2020) tested leaf whorls from the genotypes Tx430 and P898012, since this material can be collected throughout the year. The protocol also saves at least 4 weeks, as the explants can be collected around 30 d after emergence, compared with 70 d needed to collect immature embryos. Furthermore, the excision of leaf whorls is more technically straightforward than embryo isolation, allowing higher throughput.

Table 2.

Relevant literature regarding sorghum transformations and their use in research articles

References	Explant	Citationsa	Use in research article methods (References)
Electroporation
Ou-Lee et al., (1986)	Protoplasts	105	0
Battraw and Hall, (1991)	Protoplasts	36	0
Pollen sonication
Wang et al., (2007)	Pollen	17	0
Particle bombardment
Casas et al., (1993)	Immature embryos	125	5 (Casas et al., 1997; Emani et al., 2002; Jeoung et al., 2002; Grootboom et al., 2008; Kumar et al., 2011)
Casas et al., (1997)	Immature inflorescences	43	1 (Sato et al., 2004)
Able et al., (2001)	Immature embryos	43	0
Tadesse et al., (2003)	Immature embryos	49	2 (Grootboom et al., 2008; Brandão et al., 2012)
Grootbroom et al., (2010)	Immature embryos	25b	1 (Grootboom et al., 2014)
Raghuwanshi and Birch, (2010)	Immature embryos	28	0
Liu and Godwin, (2012)	Immature embryos	50	12 (Liu et al., 2013, 2015, 2019; Cardinal et al., 2016; Do et al., 2016; Belide et al., 2017; Lamont et al., 2017; Liu et al., 2017; Schnippenkoetter et al., 2017; Vanhercke et al., 2019; Flinn et al., 2020; Silva et al., 2020)
Brandão et al., (2012)	Immature inflorescences	3b	0
Visarada et al., (2014) c	Immature embryos and shoot buds	10	1 (Visarada et al., 2016)
Belide et al., (2017)	Immature embryos	8	1 (Vanhercke et al., 2019)
Agrobacterium-mediated transformation
Zhao et al., (2000)	Immature embryos	125	12 (Gao et al., 2005; Howe et al., 2006; Nguyen et al., 2007; Lu et al., 2009; Silva et al., 2011; Lipkie et al., 2013; Visarada et al., 2014; Wu et al., 2014; Cho et al., 2014; Elkonin et al., 2016; Assem et al., 2017; Kuriyama et al., 2019)
Carvalho et al., (2004)	Immature embryos	76b	2 (Nguyen et al., 2007; Raghuwanshi and Birch, 2010)
Gao et al., (2005)	Immature embryos	67	3 (Gurel et al., 2009; Liu et al., 2013; Chou et al., 2020)
Howe et al., (2006)	Immature embryos	79	12 (Kumar et al., 2011; Mall et al., 2011; Kumar et al., 2012; Jiang et al., 2013; Dwivedi et al., 2014; Elkonin et al., 2016; Scully et al., 2016; Peña et al., 2017a, b; Cuevas et al., 2016; Pan et al., 2018; Kempinski et al., 2019)
Nguyen et al., (2007)	Immature embryos	41	0
Gurel et al., (2009)	Immature embryos	71	4 (Singh et al., 2012; Liu et al., 2013; Hayta et al., 2019; Kuriyama et al., 2019)
Wu et al., (2014)	Immature embryos	64	7 (Cho et al., 2014; Magomere et al., 2016; Yamaguchi et al., 2016; Che et al., 2018; Kuriyama et al., 2019; Flinn et al., 2020; Aregawi et al., 2020, Preprint)
Yellisetty et al., (2015)	Shoot apical meristem - in planta	5	0
Do et al., (2016)	Immature embryos	18	3 (Mookkan et al., 2017; Do et al., 2018; Char et al., 2020)

Citation count checked on 12 November 20, based on CrossRef (source indicated when CrossRef was not available).

Citation count based on Google Scholar metrics

Tested both particle bombardment and Agrobacterium-mediated transformation

Table 3.

Main sorghum transformation methods, explants, genotypes, selectable markers, optimizations, and Agrobacterium strains, when appropriate, adopted for improvements in tissue culture and transformation efficiency

References	Explantsa	Genotypesa	Agrobacterium straina	SMa	Max. TE	Optimizations
Particle bombardment
Casas et al., (1993)	Immature embryos	P898012	–	Bar	0.29%	Genotypes (IS4225, CS3541, M91051, Tx430, P898012, P954035, SRN39, and Shanqui red)
Able et al., (2001)	Immature embryos	SA281	–	Bar	3 out of 4 tested events	Genotypes (M35-1, SA281, QL41, and P898012), explant (immature embryos and leaf segments), promoters (Act1, CaMV35S, and Ubi) and biolistic parameters (acceleration pressure, distance to target tissue from expulsion point, aperture of helium inlet valve)
Tadesse et al., (2003)	Immature embryos	Ethiopian accession ‘214856’	–	Npt	1.30%	Explants (immature and mature embryos, shoot tips, calli), promoters (Act1D, Adh1, CaMV35S, and Ubi1), selectable markers (Bar and Npt) and biolistics parameters (acceleration pressure, target distance, gap width and travel distance)
Casas et al., (1997)	Immature inflorescences	SRN39	–	Bar	2.61%	Genotypes (M91051, P898012, P954035, PP290, and SRN39), panicle length and biolistic parameters (particle size and material, DNA amount, acceleration pressure and target distance)
Grootbroom et al., (2008)	Immature embryos	P898012	–	Pmi	0.77%	Selectable markers (Bar and Pmi)
Raghuwanshi and Birch, (2010)	Immature embryos	Ramada	–	Hpt	0.09%	Genotypes (32 sweet sorghum), tissue culture media composition (increase of cytokinin), selectable markers (Hpt and NptII)
Liu and Godwin, (2012)	Immature embryos	Tx430	–	NptII	20.70%	Tissue culture media composition and biolistics parameters
Brandão et al., (2012)	Immature inflorescences	CMSXS102B	–	Bar	3.33%	Genotypes (nine accessions from Embrapa Maize and Sorghum National Research Center, Brazil), explant developmental stages (3–5cm in length), biolistics parameters (in osmotic medium, acceleration pressure, microcarriers flying distance)
Visarada et al., (2014)	Immature embryos	CS3541 and 296B	–	Bar	0.25%	Delivery method (Agrobacterium and particle bombardment), explant size, post-bombardment treatments
Belide et al., (2017)	Immature embryos	Tx430	–	NptII	46.60%	Tissue culture media composition (addition of lipoic acid), explant size, selectable markers (Bar and NptII), method of subculture post-bombardment
Silva et al., (2020)	Leaf whorls	Tx430	–	NptII	All 7 tested events	Genotypes (Tx430 and P898012) and tissue culture media composition (addition of activated charcoal and polyvinylpyrrolidone)
Agrobacterium-mediated transformation
Zhao et al., (2000)	Immature embryos	P898012	LBA4404	Bar	10.10%	Genotypes (P898012 and PHI391), source of explant (grown in the field or greenhouse), tissue culture conditions and media composition
Carvalho et al., (2004)	Immature embryos	P898012	LBA4404	Hpt	3.50%	Genotypes (Feterita Gesish, P898012, P967083, IS2329, Rio, Sugar drip, B-Wheatland, RTx430, and Candystripe), embryo selection, tissue culture media composition, tissue culture conditions, Agrobacterium inoculation methods
Gao et al., (2005)	Immature embryos	C401	EHA101	Pmi	3.30%	Genotypes (C401 and Pioneer 8505), tissue culture media composition
Howe et al. (2006)	Immature embryos	C2-97	NTL4	NptII	4.50%	Genotypes (Tx430 and C2-97), Agrobacterium strains (C58C1, LBA4404, EHA101, C58, and NTL4), selection agent (geneticin and paromomycin)
Nguyen et al., (2007)	Immature embryos	Sensako 85/1191	LBA4404	Hpt	5.00%	Explant pre-treatment, tissue culture conditions and media composition
Gurel et al., (2009)	Immature embryos	P898012	LBA4404	Pmi	8.30%	Genotypes (P898012, Tx430, 296B, and C401), explant pre-treatmemt, Agrobacterium strains (EHA101 and LBA4404), tissue culture media composition
Visarada et al., (2014)	Immature embryos and shoot buds	CS3541 and 296B	EHA105	Bar	0.23%	Delivery method (Agrobacterium and particle bombardment), explant type (shoot buds and immature embryos), decontamination treatments for removal of Agrobacterium
Wu et al., (2014)	Immature embryos	Tx430	AGL1	Pmi	33.20%	Agrobacterium strains (AGL1 and LBA4404), selectable markers (PAT and Pmi), tissue culture media composition (increased copper sulfate and plant hormone BAP)
Yellisetty et al., (2015)	Shoot apical ­meristem	SPV462	LBA4404	Hpt	36%b	In planta method development
Do et al., (2016)	Immature embryos	P898012	AGL1	Bar	14.20%	Genotypes (P898012, TBx623, Tx2737, Tx430, and Wheatland), Agrobacterium strains (AGL1, EHA101, and GV3101), promoters (CaMV35S, MAS, and Ubi), tissue culture conditions
Sato-Izawa et al., (2018)	Immature embryos	Tx430	GV2260	Hpt	1.90%	Explant pre-treatment and size

Selectable markers (SM): Bar, Bialaphos resistance; Hpt, Hygromycin phosphotransferase; Npt, Neomycin phosphotransferase; PAT, Phosphinothricin acetyltransferase; Pmi, Phosphomannose isomerase. Promoters: Act1D, Actin 1D; Adh1, Alcohol dehydrogenase isozyme 1; CaMV35S, Cauliflower mosaic virus 35S; MAS, Mannopine synthase; Ubi, Ubiquitin. Max. TE: maximum transformation efficiency. TE is generally defined as the total number of independent events regenerated divided by the total number of transformed explants, although it can be omitted or vary depending on the publication.

Results from the most successful transformations or optimized conditions.

Results from T₁ from selected positive T₀ plants.

Current transformation methods

To improve sorghum tissue culture and transformation, different genotypes, transformation methods, and explant sources have been tested over the years. The improvements resulted in reported increases in transformation efficiency from 0.3% to 46.6%, but these remain restricted to select genotypes, and hampered by the seasonality of explant availability. Thus far, four transformation methods have been reported for stable and transient gene expression in sorghum: electroporation; pollen-mediated transformation; particle bombardment; and the Agrobacterium-mediated method. Of these, particle bombardment and Agrobacterium-mediated transformation have been widely tested (Fig. 3). In this section, we summarize these methods, including the extent of their published usage following the initial report, which we use as a proxy for robustness, in Tables 2, 3. We will focus here on the two most commonly used methods: particle bombardment and Agrobacterium-mediated transformation.

Fig. 3.

Representation of transformation methods adopted for sorghum. GOI, gene of interest; CIM, callus induction media.

Particle bombardment

Particle bombardment, also called biolistics or the gene gun method, physically delivers DNA into intact cells or tissues. It is based on high-speed acceleration of DNA-coated gold or tungsten particles (Sanford et al., 1987). The method overcomes the host range restrictions faced when using Agrobacterium and viral vectors. Furthermore, since it does not introduce additional non-plant-derived DNA elements into the plant (as with Agrobacterium-mediated methods), it can simplify transgenic crop regulation. Particle bombardment has also been used for plastid transformation. Since plastids are maternally regulated, this can also aid control of gene flow (Svab et al., 1990; Svab and Maliga, 1993; Kumar et al., 2004; Dufourmantel et al., 2007; Lu et al., 2013; Li et al., 2016). The method of Liu and Godwin (2012), currently the most widely prescribed for sorghum transformation using particle bombardment (as judged by citation count in peer-reviewed literature outlined in Table 2), obtained a transformation efficiency of 20.7% with an optimized protocol using Tx430 immature embryos. Moreover, >90% of the transgenic plants exhibited normal growth and fertility. Adding lipoic acid to the medium and splitting the calli further enhanced the callus induction rate (Belide et al. 2017) (Table 3).

A major drawback of particle bombardment is the random integration of multiple copies of the transgene into the genome, which can lead to transgene rearrangements and silencing (reviewed by Kohli et al., 2003). However, optimization of the procedure can result in single or a low number of transgene copies (Yao et al., 2006). Random integration can also be mitigated by using approaches such as genomic safe harbors: sites in the genome that accommodate transgenes without unwanted interactions (Papapetrou and Schambach, 2016). For example, (Dong et al., 2020) have achieved targeted insertion of a 5.2kb carotenoid biosynthesis cassette at two pre-determined genomic safe harbors in rice. Therefore, this approach could potentially be applied to any crop species. Another issue with particle bombardment is the variable transformation efficiency among genotypes. For example, the elite parental lines CS3541 and 296B were transformed to increase stem borer resistance, but the highest transformation efficiency obtained was 0.25% (Visarada et al., 2014). Traditionally, the most extensively studied genotypes belong to the grain sorghum category, so reported advances are mostly applicable to that type of sorghum. For example, a comparison of 32 sweet sorghum varieties reported a transformation efficiency maximum of 0.09% (Raghuwanshi and Birch, 2010). To fully exploit sorghum as a multipurpose crop that supports the growing bioeconomy, it will be necessary to easily transform many sorghum types, including biomass and forage varieties. Finally, a recent study showed, using whole-genome sequencing, that particle bombardment can frequently induce large-scale genome damage and rearrangement (J. Liu et al., 2019). This can be problematic both for researchers, as this may impact phenotype, and potentially for regulators, as it may increase the risk of the crop being a food safety hazard. Therefore, Agrobacterium-mediated transformation, despite its drawbacks as discussed below, is still considered the preferred method of transformation by most.

Agrobacterium-mediated transformation

Agrobacterium tumefaciens mediated-transformation (reviewed by Gelvin, 2003) was initially used in eudicotyledonous plants, since monocotyledons are not natural hosts of A. tumefaciens. However, successful transformations of many monocots, such as barley, maize, rice, sorghum, and wheat, have now been achieved (Hiei et al., 1994; Ishida et al., 1996; Cheng et al., 1997; Zhao et al., 2000). Agrobacterium-mediated transformation is generally preferred when the goal is to produce plants with single- or low-copy inserts. This approach also has the advantage of resulting in minimal rearrangement of the integrated transgene.

The first reported use of Agrobacterium for stable sorghum transformation was from Zhao et al. (2000). Wu et al. (2014) optimized the resting and selection media by adding increased levels of copper sulfate and the plant hormone 6-benzylaminopurine (BAP) to generate high-quality, fast-growing, and regenerable transgenic calli. They also tested different Agrobacterium strains and selectable markers. Tx430 immature embryos infected by the Agrobacterium strain LBA4404 resulted in transformation efficiencies of up to 12.4% when the selectable marker adopted was Phosphomannose isomerase (Pmi), and 13.4% when Phosphinothricin acetyltransferase (PAT) was used. Using the strain AGL1 and Pmi selection, efficiencies of up to 33.2% were obtained, which is the most effective Agrobacterium protocol reported so far. The authors also point out that the size of T-DNA impacts the quality event frequency, as lower frequency was obtained when larger T-DNA was used (16.3kb versus 7.9kb). Quality events are defined as transformants with intact single copies of T-DNA integrated in the genome without the presence of a vector backbone.

Another optimization of transformation and regeneration with Agrobacterium was achieved by using standard binary vectors containing the Bar gene as the selectable marker under the control of a Mannopine synthase (MAS) promoter and the Agrobacterium strain AGL1 to transform immature embryos from P898012 (Do et al., 2016). Activities of modified Cauliflower mosaic virus 35S (CaMV35S), maize Ubiquitin (Zm-Ubi), and MAS promoters were evaluated, and the highest transformation efficiency was achieved using MAS. Additionally, transformation efficiency was significantly improved using a standard binary vector, while studies that achieved higher efficiencies, such as that of Wu et al. (2014), adopted superbinary vectors. Superbinary vectors have additional virulence genes from a Ti plasmid, which is beneficial for recalcitrant plants (Komari et al., 2006), but are challenging for vector construction, cloning, and transformation. The authors achieved a regeneration time frame of 7–12 weeks and an overall transformation efficiency of 14% (Do et al., 2016).

Improving Agrobacterium transformation efficiency with morphogenic regulators

To increase efficiency of transformation and expand the range of genotypes amenable to transformation, growth-stimulating morphogenic regulators have been used to induce somatic embryogenesis in monocots (Lowe et al., 2016; Mookkan et al., 2017; Nelson-Vasilchik et al., 2018). Morphogenic regulators are genes involved in developmental processes that control morphogenesis such as embryo and meristem development. Lowe et al. (2016) successfully introduced the morphogenic regulators Baby boom (Bbm) and Wuschel2 (Wu2) in maize, sorghum, and rice using the Agrobacterium strain LBA4404 and in sugarcane using the strain AGL1. Although morphogenic regulators promote the induction of somatic embryogenesis, they also cause calli necrosis, preventing the regeneration of transgenic plants (Lowe et al., 2016). To overcome this, a CRE/lox recombination system under the control of a desiccation-induced gene (Rab17) was used to remove the region of the expression cassette containing Bbm and Wu2. Transgenic calli are then subjected to desiccation prior to regeneration, allowing production of healthy transgenic plants. In sorghum, using Tx430 immature embryos as the starting material, the transformation efficiency improved from 1.9% to 18.3% when Bbm and Wu2 were introduced simultaneously (Lowe et al., 2016).

Although morphogenic regulators represented a significant improvement, higher transformation efficiencies of 33.2% (Wu et al., 2014) and 46.6% (Belide et al., 2017) were obtained with the genotype Tx430 using traditional methods. The most compelling argument for the morphogenic regulator method is the possibility of transforming genotypes that are currently recalcitrant to transformation. However, to date in sorghum, this approach has been reported mostly in transformable cultivars (Lowe et al., 2016; Mookkan et al., 2017). Mookkan et al. (2017) used the Agrobacterium strain EHA101 to transform immature embryos from sorghum genotype P898012 with a vector containing Bbm, Wu2, and the desiccation-inducible CRE/lox recombination system. They observed that calli transformed with Bbm and Wu2 reached up to 54.5% of green fluorescent protein (GFP) expression, while calli transformed with vectors without them did not show any GFP expression. Additionally, Nelson-Vasilchik et al. (2018) published a protocol using the genotype BTx623, besides the previously reported P898012, for Agrobacterium-mediated transformation with the same morphogenic regulators, the Rab17_pro:CRE/lox-inducible system, and Agrobacterium strains AGL1 and EHA101, and showed a regeneration rate of ~15%.

In planta transformation

Agrobacterium tumefaciens has also been used for in planta transformation in sorghum, which allows the introduction of DNA directly into intact plant tissue, removing the dependence on tissue culture and regeneration protocols. Yellisetty et al. (2015) reported an in planta transformation method where Agrobacterium strain LBA4404 was inoculated onto the shoot apical meristems of germinating sorghum seedlings, with transformation efficiencies of up to 36%. Despite these high reported efficiencies, the method has not been applied to further studies (Table 2). Haploid egg cell transformation by floral dipping is widely used for A. thaliana and has been applied to other Brassicaceae species, flax, and even a grass Setaria viridis (Liu et al., 2012; Martins et al., 2015). However, in planta transformation has not been established as a standard protocol for many species due to a lack of reproducibility (Hamada et al., 2017). Fundamental understanding of why some plants, such as A. thaliana and Camelina sativa, are susceptible to Agrobacterium haploid egg cell transformation would be an important step forward in plant science, as this could lead to application of this method to other species, such as sorghum.

Transient expression

The methods discussed above are mainly used for stable transformation, in which the genes are integrated into the host chromosomes and are inherited through subsequent generations. Stable transformation is particularly interesting if the goal is to engineer traits in the long term. However, for studies aiming at gene characterization, vector validation, or protein subcellular localization, especially in recalcitrant species such as sorghum, transient expression is a valuable and time-saving tool. It allows temporary expression of the introduced genes, which do not integrate into the host genome, but uses its transcriptional and translational machinery to synthesize the desired proteins. Transient expression generally reaches its maximum level between 18-48h after transformation and persists for a few days (Abel and Theologis, 1994). Agrobacterium-mediated transformation can also be successfully applied to transient expression. For example, Sharma et al. (2020) developed an in planta method using Agrobacterium for infiltration in leaves of 3- to 4-week-old sorghum, in which GFP expression was detected 3–4 d after infiltration. The method was also used to demonstrate clustered regularly interspaced short palindromic repeats (CRISPR)-mediated genome editing as a promising approach to test single-guide RNA (sgRNA) efficiencies in vivo (Liang et al., 2019).

Future needs for sorghum transformation

Although there has been progress, the technical challenges associated with sorghum tissue culture and transformation mean that efficiencies still lag behind most other monocot crops such as rice, which routinely reaches efficiencies of up to 90% (Hiei and Komari, 2008). To move the field forward, the main bottlenecks that need to be addressed are genotype dependence, prevention of transgene flow to wild relatives, and achieving higher transformation efficiency reproducibly. Overcoming these bottlenecks will allow the efficient application of synthetic biology principles, and the direct engineering of elite germplasm. In particular, this will enable the routine use of gene editing tools, including CRISPR/Cas systems and successful metabolic engineering for high-value traits. Here, we highlight some key areas for future research.

Genotype independence

Currently, the inbred lines Tx430 and P898012 are the most used genotypes for transformation due to their higher embryogenic capacity. This is limiting, particularly for commercial purposes, where the engineering of elite cultivars would be beneficial. What underpins these genotypic differences in transformability is not known. However, overexpression of morphogenic regulators such as Bbm and Wus2 has the potential not only to induce somatic embryogenesis in an expanded range of genotypes, but also to bypass or accelerate tissue culture via de novo meristem formation as demonstrated in eudicots (Maher et al., 2020). Assessment of other morphogenic regulators such as Leafy cotyledon1 (Lec1), Lec2, Monopteros (MP), Shoot meristemless (STM), hormone biosynthetic genes such as Isopentenyl transferase (Ipt), and their combinations are all promising strategies.

Besides using morphogenic regulators, genotype independence can be achieved by using tissues other than embryos, which has been achieved in barley, cotton, and rice (Dey et al., 2012; Ma et al., 2013; Han et al., 2021). For example, Han et al. 2021)successfully used microspores from barley anthers to induce callus formation for transformation and CRISPR/Cas gene editing. The diverse genetic backgrounds of the tested varieties indicated that the method was genotype independent and could be expanded to other species with established anther culture protocols. Additionally, shoot apices from 3- to 5-day-old seedlings have been used for Agrobacterium infection of cotton and rice for development of genotype-independent regeneration protocols (Dey et al., 2012; Ma et al., 2013).

Another approach for achieving genotype independence is to identify specific genes associated with tissue culture responses. Quantitative trait loci (QTL) mapping studies to identify genomic regions associated with callus induction and plant regeneration have been carried out in grasses such as barley, maize, and wheat (Amer et al., 1997; Mano and Komatsuda, 2002; Salvo et al., 2018). Although these studies concluded that tissue culture response is a complex polygenic trait, further investigation of specific candidate genes is needed, especially in sorghum, to identify genetic mechanisms that control somatic embryogenesis and efficient regeneration response.

Improved transformation efficiency

Successful introduction of a wide range of genes of interest into sorghum will depend on efficient tissue culture and transformation protocols. Currently, sorghum transformation typically uses indirect somatic embryogenesis, which goes through the callus stage. The maintenance of callus cultures is labor intensive and a lengthy process that can induce somaclonal variation. Direct somatic embryogenesis is an alternative that has been achieved in maize and sugarcane (Taparia et al., 2012; Lowe et al., 2018), and could be applied to sorghum to shorten tissue culture time and increase efficiency. As shown in maize, introduction of morphogenic regulators enables immature embryos to transition into somatic embryos in a few days and allows bypassing the callus stage (Lowe et al., 2018). Alternatively, the tissue culture method using leaf whorls reported by Silva et al. (2020) could be adapted to induce direct somatic embryogenesis as previously demonstrated in sugarcane (Desai et al., 2004; Taparia et al., 2012).

A promising alternative to somatic embryos is using embryogenic cell suspension cultures, which have been used to transform switchgrass, with high efficiency of 85% (Ondzighi-Assoume et al., 2019), and cotton, reaching transformation efficiency of ~19% (Ke et al., 2012). Efficient methods for maintaining sorghum cell cultures have potential to improve transformation efficiency by reducing somaclonal variation, decreasing false positives, and increasing the survival rate of transgenics. Additionally, cells with a synchronized cell cycle could be obtained, which may benefit CRISPR/Cas genome editing studies aiming for homology-directed repair (HDR). Cells have different abilities to repair double-stranded breaks using the non-homologous end joining (NHEJ) or HDR pathways, and the phase of the cell cycle plays a major role in the choice of the pathway (Heyer et al., 2010). The HDR pathway activity is restricted to the late S and G₂ phases of the cell cycle, while NHEJ occurs during the entire cell cycle . Therefore, cell suspension cultures can be a valuable tool not only to improve transformation efficiency, but also to increase genome editing efficiencies for targeting gene insertions, replacements, or stacking.

Other approaches to improve transformation efficiency involve the development of more efficient DNA delivery methods. Although progress has been made in Agrobacterium-mediated transformation, engineering strains with increased virulence and a wider host range will be necessary to boost efficiency. Optimizations to avoid overgrowth of Agrobacterium in the tissue culture selection media will also be relevant (Ahmed et al., 2018). Another promising strategy is the use of nanoparticles to deliver DNA, which has already been demonstrated in wheat and cotton leaves, resulting in strong protein expression (Demirer et al., 2019).

Genome editing

CRISPR/Cas-mediated genome editing can be applied broadly, including creating mutant collections of specific genes that have not been well characterized, creating variations for breeding purposes, and altering regulatory elements. CRISPR/Cas9-mediated gene editing in sorghum was first reported using Agrobacterium-mediated transformation to restore the function of an out-of-frame red fluorescence protein (DsRED2) through NHEJ (Jiang et al., 2013). Since then, CRISPR/Cas9 delivery by Agrobacterium has been adopted to mutate several sorghum genes (A. Li et al., 2018; Che et al., 2018; Char et al., 2020). To date, only one protocol for CRISPR/Cas9 genome editing using particle bombardment has been published (G. Liu et al., 2019).

Cas9 requires a 5ʹ-NGG-3ʹ protospacer adjacent motif (PAM) site upstream of the sgRNA-binding region in the genome. Other endonucleases, such as Cpf1 that targets T-rich regions (Zetsche et al., 2015), have not yet been exploited in sorghum. These alternative endonucleases broaden the range of sequences that can be targeted. In cases where the goal is generating precise point mutations, an alternative to the low-efficiency HDR pathway is using the CRISPR base editors (Komor et al., 2016). CRISPR base editors allow cytosine to thymine or adenine to guanine base editing, and have been widely adopted to introduce targeted substitutions in other crops such as rice and wheat to improve important agricultural traits, such as flowering time and herbicide resistance (C. Li et al., 2018; Kang et al., 2018; Zhang et al., 2019; Li et al., 2020).

Prevention of transgene flow

Valid concerns about transgene flow to sorghum’s sexually compatible wild weedy relatives have dampened commercial interest in engineered cultivars. Therefore, techniques that prevent transgene introgression or propagation through pollen should be prioritized. Alternatively, transgene-free methods for genome editing such as the delivery of a pre-assembled ribonucleoprotein (RNP) complex, which is done via protoplast transfection or particle bombardment, can be used (Woo et al., 2015; Svitashev et al., 2016; Liang et al., 2017). Particle bombardment would be the most suitable method for sorghum as it does not require plant regeneration from protoplasts, an ongoing challenge for sorghum tissue culture. Distinct methods adopted in other species also have potential in sorghum. For example, Zhang et al. (2016) generated transgene-free and homozygous wheat mutants in the T₁ generation by transiently expressing Cas9 in callus cells.

Another promising strategy to impede transgene flow into the wild would be the delivery of transgenes into chloroplasts to take advantage of their maternal inheritance. This avoids transgene transmission via pollen, closing a potential escape route into the environment (Daniell, 2002). Thus, chloroplast transformation would allow stable introduction of Cas9 into sorghum’s chloroplast genome to generate Cas9 lines that would not propagate the transgene via pollen.

Current genetic, genomic, and bioinformatic resources

Sorghum has several characteristics that make it an excellent potential model species for grass research. It is a diploid (2n=20), which makes it more amenable to genetic and genomic studies compared with polyploid bioenergy crops such as sugarcane. It also has a small genome size (~730 Mbp) compared with maize (2.5 Gbp), sugarcane (~10 Gbp), and wheat (~17 Gbp) (Paterson et al., 2009). Extensive variations across cultivated and wild species have been identified, suggesting a rich genetic source for adaptation and engineering (Tao et al., 2021). Additionally, sorghum is a C₄ grass with high nitrogen and water use efficiency (Ghannoum et al., 2011) and complements other grass models such as rice and Brachypodium, which are C₃ grasses. The wide genetic variation found within and among sorghum cultivars is also attractive as it can be exploited to improve the crop through breeding, population genetic, and quantitative genetic approaches (Satish et al., 2016). To support the adoption of a plant species as a research system, it is critical to have accessible resources, including germplasm collections, reference genome sequences with good quality functional annotations, and easy-to-use informatics tools that collate existing data. While sorghum does have some of these resources, there are still many gaps.

Genetic resources

The largest sorghum germplasm collection is maintained by the USDA Agricultural Research Service (ARS) National Plant Germplasm System and consists of >40 000 accessions from 114 countries, of which many regional specific subsets have been genetically characterized (Cuevas et al., 2017, 2018; Olatoye et al., 2018; Cuevas and Prom, 2020; Faye et al., 2021). The International Crops Research Institute for the Semi-Arid Tropics (ICRISAT) in India also has a large collection of 37 904 accessions (Morris et al., 2013; Cuevas et al., 2017). A third collection with >16 000 accessions is kept by the National Crop Genebank of China. Information and sources of seeds can be identified via databases such as USDA-ARS GRIN (https://npgsweb.ars-grin.gov/gringlobal/search.aspx), Eurisco (http://eurisco.ecpgr.org/), and Genesys (https://www.genesys-pgr.org/). Additional collections with particular relevance to the use of sorghum as a biomass crop include the biomass association panel (Brenton et al., 2016) and the nested association mapping population (Bouchet et al., 2017). These collections contain immense genetic diversity, which is essential for breeding programs that aim to develop cultivars better adapted to different conditions worldwide and also an important resource to elucidate molecular machineries that lead to traits of interest.

Furthermore, alleles not found in nature can be generated through mutagenesis (e.g. genotoxic chemicals or γ-irradiation) (Xin et al., 2008; Jiao et al., 2016; Chen et al., 2019) or, more recently, through genome editing. Mutant lines are being added to these germplasm collections to create an even more diverse community resource. Increasingly, these mutant populations are accompanied by whole-genome sequences, allowing researchers to take a reverse genetics approach to identifying gene function (Addo-Quaye et al., 2018).

Genomic resources

The first sorghum reference genome (from the grain sorghum BTx623) was generated using whole-genome shotgun sequencing in 2009 (Paterson et al., 2009), and placed ~98% of the genes in their chromosomal context. More recently, BTx623 version 3.1.1 was released with improved assembly and annotation (McCormick et al., 2018). The high-quality reference genome of the sweet sorghum ‘Rio’ was also recently released using Pacific Biosciences long-read sequencing (Cooper et al., 2019). The authors used it to explore the possible genomic differences between sorghum types, and revealed a high rate of non-synonymous and potential loss-of-function mutations in sweet sorghum. However, few changes in gene content and overall genome structure were observed (Cooper et al., 2019). Two additional genomes, BTx642 and RTx430, are also available on Phytozome (see below). An ongoing sorghum pan-genome project at the DOE Joint Genome Institute (JGI) will explore this information further (Mockler, 2016).

Bioinformatic resources

Several bioinformatic resources host sorghum data (links and references described in Table 4). Sorghum breeders and researchers can rely on bioinformatic resources such as Phytozome, the Plant Comparative Genomics portal of the DOE Joint Genome Institute (JGI) (Goodstein et al., 2012). This includes the latest sorghum reference genome (McCormick et al., 2018). Additionally, the Sorghum genome SNP database (SorGSD), a database with 62 million single nucleotide polymorphisms (SNPs) from 48 sorghum accessions, allows the user to search for synonymous and non-synonymous SNPs, their annotation, geographic origin, and breeding information (Luo et al., 2016). A valuable resource for sorghum improvement is the Sorghum Genomics Functional Gene Discovery Platform, which enables the identification of sorghum lines containing natural and chemical-induced variations in coding sequences (https://www.purdue.edu/sorghumgenomics/)(REF). The Sorghum Functional Genomics Database (SorghumFDB) also has a search feature with orthologous pairs in A. thaliana, rice, and maize, in addition to gene family classifications, gene annotations, loci conversions, miRNA and target gene information, and a genome browser (Tian et al., 2016). The PlantGDB, a resource for comparative plant genomics, has a section on sorghum (SbGDB), which includes gene structure annotation, sequence analysis tools, and annotated protein alignments. Also, sorghum metabolic network data can be found in SorghumbicolorCyc at the Plant Metabolic Network (PMN), a curated source of metabolic information from the literature and computational analyses (Schläpfer et al., 2017). Lastly, UniProt has sorghum protein sequences from genome sequencing projects (Saski et al., 2007; Paterson et al., 2009; Hawkins et al., 2021). These resources can assist researchers who are new to sorghum research to understand sorghum genome architecture and its variations, and to draw comparisons with other extensively studied species.

Table 4.

Bioinformatics resources available for sorghum research

Bioinformatic tool	Purpose	Website	Source	Reference
Phytozome	Reference genome and alignment searches	https://phytozome.jgi.doe.gov/pz/portal.html#!info?alias=Org_Sbicolor	Joint Genome Institute (JGI)	McCormick et al., (2018)
Plant Metabolic Network	Network of metabolic pathway data	https://plantcyc.org/content/sorghumbicolorcyc-7.0.1	Carnegie Institution for Science	Hawkins et al., (2021)
Gramene sorghum	All sorghum resources as statistics, germplasm resources, metabolic pathways	https://archive.gramene.org/species/sorghum/sorghum_intro.html	Cold Spring Harbor Laboratory and Cornell University	Tello-Ruiz et al., (2018)
Sorghum FDB - Functional Genomics Database	Integrated search for gene family classifications, gene annotations, miRNA and target gene information, orthologous pairs in Arabidopsis, rice, and maize, gene loci conversions and a genome browser	http://structuralbiology.cau.edu.cn/sorghum	Zhen Su’s group at China Agricultural University	Tian et al., (2016)
SbGDB	Sequence-centered genome view with focus on gene structure annotation	http://www.plantgdb.org/SbGDB/	Brendel group at Indiana University	–
Uniprot	Proteomic data	https://www.uniprot.org/proteomes/UP000000768	UniProt Consortium	–
Sorghum genomics - Functional Gene Discovery Platform	Search for lines containing natural and ems-induced variations in coding sequences	https://www.purdue.edu/sorghumgenomics#	Purdue University	–

Conclusion

Sorghum has a bright future as a multipurpose crop that is suited to the challenging growth conditions that climate change will bring. Its extensive genetic diversity combined with relatively recent and limited domestication means that it also has an excellent potential for further improvement. Sorghum can become a model system for other grass species, particularly in areas such as abiotic and biotic stress responses, plant–microbiome interactions, and evolution. We see transformation challenges as a major bottleneck to the development of sorghum as both a widely adopted research system and a key feedstock for the bioeconomy, and contend that research tackling this problem is a high priority.

Author contributions

TNS, JBT, JCM, and SYR: conceptualization; TNS and JCM: data curation; JCM, SYR, and JBT: funding acquisition; JCM and SYR: project administration; TNS, JBT, and JCM: writing—original draft; TNS, JBT, JD, SYR, and JCM: writing—review and editing.

Funding

This research was supported, in part, by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, Genomic Science Program grant no. DE-DE-SC0020366 (SYR and JCM) and DE-SC0018277 (SYR), the DOE Joint BioEnergy Institute (http://www.jbei.org) supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE-AC02-05CH11231 between the Lawrence Berkeley National Laboratory (JCM) and the U.S. Department of Energy, and the U.S. National Science Foundation grants IOS-1546838 (SYR) and MCB-1617020 (SYR).