CRISPR/Cas9-Mediated Mutagenesis of Carotenoid Cleavage Dioxygenase (CCD) Genes in Sorghum Alters Strigolactone Biosynthesis and Plant Biotic Interactions

Strigolactones are a class of carotenoid-derived plant hormones that act as signals in planta for a diverse range of plant developmental processes such as shoot branching, stem elongation, and root architecture (Brewer et al. 2013; Umehara et al. 2008). Strigolactones are also exuded from plant roots and act as signals to microorganisms and parasitic plants in soils and promote plant adaption to nutrient availability and responses to biotic and abiotic stresses such as drought and high salinity (Mostofa et al. 2018; Van Ha et al. 2014). Strigolactones were originally discovered as germination stimulants of parasitic weeds such as Striga lutea in the family Orobanchaceae (Cook et al. 1966). Parasitic plants that colonize roots are completely reliant on the host plant for water and nutrients (Bouwmeester et al. 2003). In addition to acting as a germination signal for the seeds of Striga spp., studies have also revealed that strigolactones stimulate the hyphal branching of arbuscular mycorrhizal fungi (AMF) which, in turn, enhances the symbiotic association between AMF and the host plants (Akiyama et al. 2005). Currently, the importance of strigolactones in plant development and interactions with other organisms has brought them into the spotlight of plant biology research.

To date, more than 30 natural strigolactones have been identified in the root exudates of different plant species (Yoneyama 2020). The various strigolactones can be classified into canonical and noncanonical strigolactones based on their distinct chemical structures. Canonical strigolactones can be further classified into strigol-type and orobanchol-type based on the C-ring orientation (Yoneyama and Brewer 2021). Some plant species such as Arabidopsis and maize produce both canonical and noncanonical strigolactones (Charnikhova et al. 2017; Kohlen et al. 2011). Some plants such as tomato and pea only produce orobanchol-type strigolactones (Wang and Bouwmeester 2018). Studies focused on strigolactone biosynthesis and the strigolactone signaling pathway have characterized orthologs of carotenoid cleavage dioxygenase 7 (CCD7) and CCD8 genes in many plant species such as Arabidopsis, rice, tomato, tobacco, and maize (Arite et al. 2012; Dutta et al. 2019; Gao et al. 2018; Tan et al. 2003; Vogel et al. 2010). CCD7 and CCD8 are two key proteins involved in the first stages of strigolactone biosynthesis (Alder et al. 2012; Waters et al. 2017). CCD7 catalyzes 9-cis-β-carotene to produce β-ionone and 10′-apo-β-carotenal, which is then cleaved by CCD8, yielding carlactone, the biosynthetic precursor of strigolactones (Butt et al. 2018; Schwartz et al. 2004). Carlactone is further catalyzed by a member of the more axillary growth 1 (MAX1) family of cytochrome P450 enzymes and subsequent enzymes to produce canonical or noncanonical strigolactones (López-Ráez et al. 2017; Waters et al. 2017). Strigolactone-deficient ccd7 or ccd8 mutants display more branching or high tillering phenotypes in Arabidopsis, rice, and pea (Al-Babili and Bouwmeester 2015). In addition, the root exudates from ccd7 or ccd8 mutant plants of pea and rice show impaired ability to stimulate parasitic weed germination (Al-Babili and Bouwmeester 2015). In roots of sorghum, rice, and wheat, the exudation of strigolactones is highly induced under phosphorus (P)- and nitrogen (N)-limiting conditions (Yoneyama et al. 2012). Over many years, the synthesis and role of strigolactones in plant development and communication with soil organisms has been characterized, with many other functional details in important crop species yet to be elucidated.

Sorghum (Sorghum bicolor (L.) Moench) is a major cereal crop grown for food and feed worldwide. It possesses a significant genetic diversity for traits of agronomic importance such as yield, and yield-related traits such as stand count, number of productive tillers, panicle weight, and so forth. Sorghum produces both strigol-type and orobanchol-type strigolactones. The composition and quantity of strigolactones such as orobanchol, 5-deoxystrigol, sorgomol, strigol, and sorgolactone differs among sorghum cultivars (Mohemed et al. 2018), which further highlights the wide genetic diversity in this species. Sorghum belongs to the Poaceae family, in which a genomic and phylogenetic analysis has been performed on the CCD gene family. Of the major cereal crops, including maize (Zea mays), rice (Oryza sativa), and sorghum (S. bicolor), the clustering based on a phylogenetic tree identified one single copy of the SbCCD7 gene and six copies of SbCCD8 genes (SbCCD8a, SbCCD8b, SbCCD8c, SbCCD8c-like, SbCCD8d, and SbCCD8d-like) in sorghum (Vallabhaneni et al. 2010). However, the enzymatic functions of the SbCCD genes have not been demonstrated.

One of the major biotic constraints to sorghum yield, especially in dry and warm regions such as Africa, are the parasitic weeds of the genus Striga (Muchira et al. 2021). Previous studies have assessed the correlation between Striga germination and strigolactone profile in the root exudates of different sorghum varieties and found that variation in the composition and proportion of strigolactones produced in root exudates of sorghum influences the susceptibility to Striga (Mohemed et al. 2018; Yoneyama et al. 2015). Some studies have found that sorghum genotypes such as Tetron, IS9830, and Wad Baco that produce a higher proportion of orobanchol and a lower proportion of 5-deoxystrigol had low germination-stimulating activity of Striga (Mohemed et al. 2018), making them less susceptible to Striga colonization. A Striga-resistant sorghum cultivar (SRN39) mainly exudes orobanchol, which is partly responsible for this resistance (Yoneyama et al. 2010) and is due to a mutation in the low germination stimulant 1 (LGS1) locus, causing changes in sorghum exudate from mostly 5-deoxystrigol to orobanchol (Gobena et al. 2017). Although there have been conclusions made about the importance of 5-deoxystrigol to orobanchol in Striga resistance, this appears to be dependent on the Striga ecotype being tested, which was recently demonstrated in a genomic approach to Striga resistance (Bellis et al. 2020). In that study, the Kenyan ecotypes of Striga were shown to be equally responsive in their germination to 5-deoxystrigol to orobanchol (Bellis et al. 2020). These previous results highlight the importance of further studies on the associations between sorghum host genotype, strigolactone profiles, the roles of SbCCD7 and SbCCD8 genes in strigolactone production, and the Striga ecotype variation in response to different strigolactones.

Strigolactone perception by parasites in the family Orobanchaceae and AMF highlights the importance of strigolactones as a signal for soil organisms to find host roots, and suggests that other soil organisms might use strigolactone as a cue. The rapid decay of strigolactones and their very low bioactive concentrations may be characteristics making them a good cue for other organisms to find plant roots in soil (Yoneyama 2019). Researchers have begun to identify the role of strigolactones in communication between plants and other soil microorganisms that result in important interactions (Andreo-Jimenez et al. 2015). In addition to the importance in the symbiosis of AMF, strigolactones exert beneficial effects on legume–Rhizobium interaction and nodule formation (De Cuyper et al. 2015). Activation of the key strigolactone biosynthesis genes such as D27, CCD7, or CCD8 by interactions of plants with Rhizobium has also been reported (Liu et al. 2013; van Zeijl et al. 2015). Moreover, strigolactones have been indicated to influence defense responses of lower plants against disease-causing fungal or bacterial pathogens. For example, ccd7 and ccd8 knockout mutants in moss (Bryophyta spp.) were more susceptible to the necrotrophic fungus Sclerotinia sclerotiorum, and the resistance to this fungus could be restored by application of GR24, a synthetic analog of strigolactones (Decker et al. 2017). In Arabidopsis, mutant plants with disruption of the MAX2 gene, which plays essential roles in strigolactone signaling and regulates strigolactone perception, had decreased resistance to the pathogenic bacteria Pectobacterium carotovorum and Pseudomonas syringae (Piisilä et al. 2015). The reduced Striga germination allele in sorghum LGS1 that changes exudates of strigolactones shows population genetic evidence of potential tradeoffs that are currently unknown (Bellis et al. 2020) but could involve interactions with microbes. Considering the biological and ecological importance of strigolactones and their signaling pathways in plant–microbe interactions, a deeper investigation will help to fully elucidate their potential roles.

Therefore, in this study, two SbCCD genes from sorghum were characterized. Although we aimed to study the consequences of the knockouts in all of the SbCCD genes, we were only successful in obtaining gene edits that knocked out two (SbCCD8a and SbCCD8b) of the four isoforms of SbCCD8 and were not able to obtain plants with knockout edits in SbCCD7. The roles of two SbCCD8 genes in strigolactone production, sorghum plant growth and development, resistance to the parasitic weed Striga, and plant–microbe interactions were investigated through CRISPR/Cas9-mediated gene knockout.

MATERIALS AND METHODS

Vector construction.

Phytozome (https://phytozome-next.jgi.doe.gov/) was used to retrieve gene sequences from Sorghum bicolor (L.). CRISPR single-guide RNAs (sgRNAs) were designed based on the encoding sequences of SbCCD genes using CRISPR-P (http://crispr.hzau.edu.cn/cgi-bin/CRISPR2/CRISPR), a web application tool for synthetic sgRNA design of the CRISPR system in many plants species. A phylogenetic analysis on the carotenoid dioxygenase gene family in maize (Zm), sorghum (Sb), and rice (Os) revealed that SbCCD7 clustered separately from the other SbCCD genes, and was phylogenetically closer to ZmCCD7 and OsCCD7 (Vallabhaneni et al. 2010). One CRISPR sgRNA construct was designed to target the single-gene copy of CCD7 (Booker et al. 2004). One CRISPR sgRNA construct was designed to target SbCCD8b because it previously was identified to play a role in apocarotenoid biogenesis and CCD8b genes form a separate phylogenetic group (Vallabhaneni et al. 2010). The expression profile of the CCD gene family in S. bicolor retrieved from the Phytozome database also showed that SbCCD8b was highly expressed in the roots compared with that of other SbCCD genes at different plant growth stages (Supplementary Fig. S1) (McCormick et al. 2018). The other three SbCCD8 isoforms, including SbCCD8a, SbCCD8d, and SbCCD8d-like, were targeted with a single CRISPR sgRNA construct, in which SbCCD8a and SbCCD8d-like were targeted by one sgRNA derived from a highly conserved sequence, and SbCCD8d was targeted by another sgRNA. The three sets of sgRNA construct were then separately assembled into a CRISPR/Cas9 binary vector, pBUN421 (Addgene plasmid number 62204), using BsaI restriction enzyme and T4 ligase (New England Biolabs) with the methods as previously described by Xing et al. (2014). The resulting binary vectors were mobilized into Agrobacterium tumefaciens and used for Agrobacterium-mediated transformation of the sorghum genotype Tx430 (Howe et al. 2006).

Mutation detection.

To detect CRISPR/Cas9-edited mutations at the CCD gene locus, a restriction enzyme site loss method was used (Nekrasov et al. 2013). Briefly, the partial fragment of CCD gene encompassing the sgRNA target site was amplified by PCR from leaf genomic DNA of the wild type Tx430 and transgenic plants (primer sequences were listed in Supplementary Table S1). Then, approximately 50 ng of PCR product was digested with 0.2 μl of CpoI (for CCD7), AvaII (for CCD8b), or Cfr13I (for CCD8a, CCD8d, and CCD8d-like) nuclease (10 units μl⁻¹; New England Biolabs) in a 30-μl reaction at 37°C for 1 h. Digested PCR products were separated and visualized by running on a 1.2% agarose gel. Homozygous mutations were determined based on the abolition of a restriction enzyme site within the target region. The mutations in the target sequence were further verified by Sanger sequencing.

Strigolactones detection.

Sorghum seeds of the wild type and mutant were germinated in soil in the greenhouse. Five days after emergence, the seedlings were transferred into hydroponic containers filled with 1 liter of full-strength Hoagland nutrient solution and grown in a growth chamber (16 h of light at 26°C and 8 h of darkness at 18°C, relative humidity approximately 60%). The nutrient solution was aerated continuously and refreshed every 2 days. To stimulate strigolactone exudate production after 2 weeks, nitrogen-deficient conditions were applied by replacing the solutions with nutrient solution containing low-nitrogen Hoagland solution (10% ammonium and nitrate). Fresh low-nitrogen nutrient solution was replaced after 7 days and, after 24 h, plants were removed from the solutions and tissues were dried in a 65°C oven for 3 days, at which point root and leaf tissue was weighed. The nutrient solutions containing the root exudates were filtered through a Nalgene Rapid-Flow 75-mm Bottle Top Filter (Thermo Fisher Scientific) and the strigolactones were purified using reverse-phase chromatography on a 5-ml C18 SEPAK cartridge column (Octadecyl 500 mg; JT Baker). The C18 SEPAK column was conditioned with 5 ml of methanol three times, followed by washing three times with 5 ml of water. Next, 300 ml of the nutrient solution containing root exudates was passed through the preequilibrated C18 SEPAK cartridge column. The strigolactone fraction was first eluted with 5 ml of 60% acetone, followed by 5 ml of 100% acetone. The resulting 10 ml of solvent was collected and dried down using nitrogen gas. The dried samples were stored at −80°C before liquid chromatography–tandem mass spectrometry (LC/MS-MS). Strigolactones were identified and the levels of strigolactones were quantified using targeted LC/MS-MS. Briefly, the samples were resuspended in 30% methanol and strigolactones were separated using a ZORBAX Eclipse Plus C18 column (2.1 by 100 mm; Agilent) running at a flow rate of 0.45 ml/min. The gradient of the mobile phases A (0.1% formic acid in water) and B (0.1% formic acid/90% acetonitrile) was as follows: 30% B for 1 min, to 100% B in 4 min, hold at 100% B for 3 min, to 30% B in 0.5 min. The column compartment was set at 40°C. The Shimadzu LC system used was interfaced with a Sciex QTRAP 6500+ mass spectrometer equipped with a TurboIonSpray electrospray ion source. Analyst software (version 1.6.3) was used to control sample acquisition and data analysis. The QTRAP 6500+ mass spectrometer was tuned and calibrated according to the manufacturer's recommendations. The strigolactones (orobanchol, strigol, and 5-deoxystrigol) were detected using MRM transitions that were optimized using standards. The MRM transitions, compound settings, and instrument settings have been previously published (Lopez-Guerrero et al. 2022).

Striga germination assay.

To determine the effect of mutations on preattachment resistance of sorghum against Striga hermonthica, both wild-type and transgenic plants (homozygous T3 generation) were grown in a greenhouse at the Pennsylvania State University (PSU) with the temperatures of 27°C (day) and 24°C (night) without additional lighting, and no shading or humidity control. Plants were grown in 164-ml cone-tainers (Ray Leach, Stuewe & Sons, Inc.) in sand, with three cotton balls at the bottom of the pot. Five replicate plants per genotype (three genotypes included) were grown, and there were three technical Striga replicates per sorghum plant and two Striga populations (Kenyan and Malian), for a total of 120 Striga germination assays. A positive control with GR24 (a strigolactone analogue) at 0.1 ppm and negative water controls were also included. Plants were top-watered with 20 ml of solution every other day, initially at high nutrition for 2 weeks, followed by a week of low-phosphorus fertilizer, then a week of watering only without fertilizer. Whole plants were then harvested and sand was carefully washed off roots. Root systems were suspended in water with the shoots intact in covered glass flasks in darkness for 48 h. The solution in the flask containing root exudates was then applied to preconditioned Striga seeds (11 days in water at 30°C in darkness). Striga germination was assayed 3 days later following the protocol of Bellis and Kelly (https://www.protocols.io/view/striga-hermonthica-germination-assay-2wdgfa6). S. hermonthica seeds were collected in 2016 from a sorghum field in Kibos, Kenya (0°40′ S, 34°49′ E) and in 2018 from a sorghum field in Siby, Mali (12°23′ N, 8°20′ W) and imported to the PSU quarantine facility under Animal and Plant Health Inspection Service (APHIS) Plant Protection and Quarantine permit P526P-18-01678.

Field site and sample collection.

A field study was conducted in Mead, NE, U.S.A. (41°13′34′′ N, 96°29′18′′ W) in the growing season of 2020 with the transgenic sorghum lines under an APHIS permit. Two homozygous transgenic events (Sbccd8a and Sbccd8b) and the wild type Tx430 were planted on 3 June 2020, with four replicate plots in a randomized complete block design. Each plot was 4.57 by 2.29 m, containing three rows (Supplementary Fig. S2A). A 46-0-0 nitrogen fertilizer was applied to the field at 89 kg/ha. The field was irrigated on 12 June, 29 June, and 3 September 2020. Bulk soil, soil near roots, rhizosphere, root endosphere, and leaf samples were collected at three growth stages, including the growth point differentiation stage (vegetative stage) boot stage, and grain-fill stage on 7 July, 3 August, and 31 August 2020, respectively (Supplementary Fig. S2B). Sampling was conducted as previously described (McPherson et al. 2018). Two plants were excavated from two random locations in each plot to a depth of 20 cm and bulked together as one replicate sample. Bulk soil was taken between plant rows at a depth of 30 cm using a shovel. Soil near roots (approximately 1 cm of soil surrounding the root surface) was shaken off and collected into a Ziploc bag. Different sections of the roots, including crown, seminal, and primary roots, were collected in a 50-ml falcon tube containing 35 ml of phosphate buffer. The tube was vortexed for 2 min to wash off the soil adhered to roots (rhizosphere sample). Root samples were removed from the tube, blotted dry, and collected in a new 50-ml falcon tube. All soil, rhizosphere, and root samples were kept on ice during sampling. Root samples were surface sterilized with 1% bleach and 70% ethanol, and stored at −80°C before DNA extraction. Rhizosphere samples were filtered through a 100 μm filter and spun down by centrifugation. The supernatant was discarded, and the rhizosphere was resuspended in 2 ml of phosphate buffer and transferred to an Eppendorf tube for final spin. Supernatant was removed and the rhizosphere pellet was stored at −20°C until DNA extraction. At the end of the growing season, the panicles were harvested and dried, and seeds were weighed to compare grain yields between the wild-type and mutant plants.

DNA extractions, library construction, and Illumina sequencing.

DNA was isolated from root samples using the MagMAX Plant DNA Kit (Thermo Fisher Scientific) and from rhizosphere and soil samples using the MagAttract PowerSoil DNA KF Kit (Qiagen) by automated DNA extraction using a KingFisher Flex (Thermo Fisher Scientific). The V4 region of the 16S ribosomal RNA (rRNA) gene was amplified by PCR using a dual-index sequencing strategy (Kozich et al. 2013) with AccuPrime Pfx DNA polymerase (Invitrogen, Carlsbad, CA, U.S.A.). For root samples, peptide nucleic acid (PNA) oligonucleotides (PNA Bio, Inc.) were used in the PCR assays to suppress amplification of plant mitochondrial and plastid 16S rRNA sequences (Lundberg et al. 2013). The 16S rRNA amplicon libraries were prepared as previously described (Hao et al. 2021). In each sequencing library, a blank DNA extraction control was used as a negative control and genomic DNA from microbial mock community B (even, low concentration) v5.1L 16S rRNA gene sequencing (BEI Resources) was used as the positive control. Libraries of internal transcribed spacer (ITS) amplicons were generated using locus-specific primers targeting the ITS1-1F region (ITS1-1F-F: 5′-CTTGGTCATTTAGAGGAAGTAA-3′ and ITS1-1F-R: 5′-GCTGCGTTCTTCATCGATGC-3′). Sequencing libraries were quantified and quality checked using a high-sensitivity DNA kit on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, U.S.A.). The 16S rRNA libraries were paired-end sequenced on the Illumina MiSeq platform with the MiSeq 600 cycles v3 kit and the ITS paired-end sequencing was done on a NovaSeq platform.

16S rRNA and ITS sequence data processing.

The 16S rRNA and ITS sequencing data were processed using USEARCH (version 10.0.240) (Edgar 2010) and QIIME (Quantitative Insights into Microbial Ecology, version 1.9.1) (Caporaso et al. 2010). Briefly, paired-end sequencing reads were demultiplexed and sequence pairs were merged using USEARCH, followed by quality filtering and clustering with simultaneous chimera removal using the UNOISE algorithm (implemented within USEARCH) into amplicon sequence variant (ASVs) with an identity threshold of 100%. ASVs were taxonomically annotated using the Ribosomal Database Project classifier (Wang et al. 2007) against the SILVA rRNA database (Quast et al. 2012) and UNITE general FASTA release, version 8.0 (Lücking et al. 2021) for 16S rRNA and ITS data, respectively. Low-abundance ASVs (<2 total counts across all samples), mitochondria, and chloroplast sequences were identified and discarded from the dataset. The microbial α diversity was evaluated by calculating the observed ASVs and Shannon diversity index. All samples were rarefied to an equal sampling depth (rarefaction values for all tissue types at all plant growth stages are listed in Supplementary Table S2). Bray-Curtis dissimilarity matrices between the wild-type and mutant plant samples were calculated to evaluate the microbial β diversity. Data were visualized using the ggplot2 package (v2.2.1) of R (Wickham 2009).

Root architecture analysis.

To study root morphology, sorghum plants of the T2 wild type and Sbccd8a and Sbccd8b mutant plants were grown in 15.24 cm round pots filled with calcined clay (eight replicate pots of each genotype) (Diamond Pro Calcined Clay Drying Agent). Two sorghum seeds were sown in each pot and plants were thinned to one uniform seedling on the third day after emergence. Plants were kept in the greenhouse with a photoperiod of 16 and 8 h at 21 and 18°C (day and night), respectively. All pots were randomly placed and periodically rerandomized. Plants were watered daily, and full-strength Hoagland nutrient solution was applied twice a week. When the fifth leaf was fully expanded, roots were carefully removed from the pots and rinsed with water to remove calcined clay. The washed roots were placed on an Epson Perfection V800 scanner bed and carefully flattened with tweezers, and the root system images were captured by scanning using the SilverFast 8 SE software for Epson. Root architectural traits were analyzed using DIRT (digital imaging of root traits), an online high-throughput computing platform for root architecture from digital images (Das et al. 2015). Fresh shoots and roots were weighed separately for fresh mass, then dried in a 65°C oven for 3 days for dry mass measurements.

Statistical analyses.

Differences in microbial α diversity were determined using the Wilcoxon test adjusted for false discovery rate (FDR) implemented in R. Comparisons of Bray-Curtis dissimilarities or Weighted Unifrac dissimilarities were conducted using permutational multivariate analysis of variance with the “adonis” function in the vegan R package (Oksanen et al. 2018). Pairwise comparisons between the wild-type and mutant plants were conducted using Student's t test. Pearson's χ² tests were conducted to evaluate the association of relative abundance of the microbes and the amount of orobanchol. Differences in the relative abundance of microbes at specific taxonomic levels were assessed with Welch's t test with Bonferroni correction using STAMP (structural time series analyzer, modeler and predictor) software (Parks et al. 2014). P < 0.05 was considered statistically significant.

RESULTS

sgRNA design for targeting the CCD genes in sorghum.

To effectively knock out the expression of SbCCD genes, two CRISPR sgRNA constructs were designed to target CCD7 and CCD8b individually. A third sgRNA construct was designed to target the other three CCD8 isoforms, including SbCCD8a, SbCCD8d, and SbCCD8d-like, in which SbCCD8a and SbCCD8d-like were targeted by one sgRNA derived from a highly conserved sequence, and SbCCD8d was targeted by another sgRNA. There were five, two, and one independent T0 transgenic events obtained from the CCD7, CCD8b, and CCD8 isoform (CCD8a, CCD8d, and CCD8d-like) constructs, respectively. The most highly shoot-tillering phenotype was observed in the ccd8b line (Fig. 1A). To verify target site mutations in the CCD genes, a restriction enzyme site loss assay was performed (Fig. 1B), showing that an approximately 618-bp PCR undigested fragment encompassing the target site amplified from the T1 ccd8b mutant when subjected to Eco47I digestion. In contrast, digestion of the product amplified from the wild type contained two cleaved bands with expected sizes (496 and 122 bp) (Fig. 1B). Sequencing confirmed 1-bp deletion and 1-bp insertion in the ccd8a and ccd8b edited lines, respectively (Fig. 1C), resulting in changes in the open reading frames that introduced a premature stop codon (Supplementary Fig. S3). No homozygous mutations were identified in the CCD7, CCD8d, and CCD8d-like transgenic events (data not shown). In field experiments, the yields of the wild type of sorghum and the CCD8a and CCD8b mutants were measured. The grain yield of the T2 ccd8b mutant plants were greatly reduced compared with the wild type but changes in yield were not observed in the ccd8a mutants (Fig. 1D).

Download as PowerPoint

Strigolactone deficiency altered the root architecture of sorghum plants.

To evaluate the effect of strigolactone deficiency on sorghum root morphological traits and biomass, plants were sampled at the fifth leaf stage and the root architecture traits were compared with the wild type. Root images taken by scanning were converted into black-and-white images, and significant changes in biomass were measured in the mutant lines compared with the wild type (Fig. 2A and B). The ccd8b roots were smaller than the wild type whereas the ccd8a roots were larger than the wild type (n = 8 for each genotype) (Fig. 2B). Root architectural traits were analyzed and three root phenes, including the projected root area (pixel counts), number of root tip paths, and the number of skeleton nodes, were derived using DIRT software. This analysis showed that the ccd8b mutant roots had significantly reduced projected area, indicative of root mass distribution, while the ccd8a mutant roots had significantly increased projected area compared with that of the wild type (n = 8 for each genotype) (Fig. 2C). In addition, both number of root tip paths (Fig. 2D) and skeleton nodes (Fig. 2D) of the root system were reduced in the ccd8b mutant (n = 8 for each genotype). No differences in the number of root tip paths and skeleton nodes were observed between the wild type and the ccd8a mutant. Because the roots were flattened, some root morphological traits such as the width of root system, root top angle, and so on were not further analyzed.

Download as PowerPoint

Reduced strigolactone production and striga stimulation in sorghum mutant plants.

To determine the effect of altered production of strigolactones in root exudates of sorghum on the susceptibility to Striga spp., a Striga germination stimulation assay was conducted. It showed that the ccd8b mutant plants were resistant to colonization by Kenyan Striga compared with the wild type, whereas the wild type and mutants both showed resistance to Malian Striga (Fig. 3A). Previous experiments have shown that the Kenyan Striga germinates at an approximately equal rate in response to 5-deoxystrigol versus orobanchol, whereas the Malian Striga population germinates at a much higher rate in response to 5-deoxystrigol versus orobanchol (Bellis et al. 2020). The amounts of strigolactones produced in root exudates of sorghum were measured by LC/MS-MS. Consistent with the known difference in Striga populations, only orobanchol was detected in the root exudates of the wild type of sorghum, and other types of strigolactones were not detectable (5-deoxystrigol, orobanchol, and strigol were monitored for detection). The amount of orobanchol varied significantly among different mutants compared with the wild type. Root exudates of the ccd8a mutant had significantly reduced amounts of orobanchol compared with that of the wild type (Fig. 3B). Orobanchol was undetectable in the root exudates of the ccd8b mutant (Fig. 3C), and these exudates failed to induce any germination of Kenyan Striga (Fig. 3A). The ccd8a mutant did not induce significantly different amounts of germination in the Kenyan Striga compared with the wild type. It is worth noting that the wild type values of orobanchol varied among the two comparisons (Fig. 3B and C) because each mutant was analyzed for strigolactone in root exudates at different times with a wild type control. These results confirmed that knockout of the CCD8a and CCD8b genes altered the production of strigolactones as well as the sorghum susceptibility to the Kenyan population of Striga whereas all of the mutants and the wild type were resistant and did not promote the germination of the Malian population of Striga, which is consistent with Tx430 carrying the lgs1 allele (Bellis et al. 2020).

Download as PowerPoint

Orobanchol impacts on bacterial diversity and community composition.

Bacterial diversity and community composition were investigated using 16S rRNA amplicon sequencing in three compartments, including the root endosphere, rhizosphere, and soil, at three plant developmental stages in a replicated field trial. At grain-fill stage, significantly decreased richness and diversity in the bacterial community were observed in rhizosphere samples of ccd8b mutant, in which orobanchol was undetectable in root exudates. The diversity of the bacterial community as determined by the Shannon diversity index was significantly lower in the ccd8b mutant (Fig. 4A). Bacterial species richness was significantly decreased in the ccd8b mutant as determined by observed ASVs (Fig. 4B). In addition, canonical analysis of principal coordinates based on a Bray-Curtis dissimilarity matrix revealed that rhizosphere bacterial community composition was significantly affected by sorghum genotype in a comparison with the wild type and mutants (P < 0.01, 4.0% variation explained) (Fig. 4C). Pairwise comparison of Bray-Curtis dissimilarities showed that the bacterial community composition was significantly higher in the ccd8b mutant compared with the wild type (P < 0.05), and not between the wild type and the ccd8a mutant (Fig. 4D). Furthermore, the changes in relative abundance of genera in the rhizosphere of the ccd8b mutant at grain-fill stage compared with the wild type showed that the relative abundances of 42 genera were significantly different (Supplementary Fig. S4). The top 20 genera whose abundance was most significantly changed compared with the wild type is shown in Figure 4E. Among these 20 genera, only six increased in relative abundance with the knockout of the CCD8b gene and the abolition of orobanchol concentrations in the exudates, including Massilia, Chitinophaga, Niastella, RB41, Nocardia, and Kribbella. The genera that significantly decreased in relative abundance in the ccd8b mutant included RB41, Chthoniobacter, Pirellula, Pedosphaeraceae, and Haliangium, as well as nine others. In addition, 10 bacterial genera were identified to be significantly (Pearson's χ² test, FDR < 0.01) associated with the amount of orobanchol in the ccd8b mutant, and all 10 bacteria were enriched in the ccd8b mutant (Supplementary Fig. S5A; Supplementary Table S3). These results indicated that the knockout of the CCD8b gene causing the abolition of orobanchol root exudates had significant effects on the rhizosphere bacterial diversity and community composition at the grain-fill stage.

Download as PowerPoint

Orobanchol in root exudates influences fungal taxa abundance.

In addition to bacterial communities, fungal diversity and community composition were also investigated using ITS amplicon sequencing in three compartments of root, rhizosphere, and soil at all three plant growth stages. Global changes in the fungal diversity or community composition were not observed between the wild-type and mutant plants using either the weighted UniFrac or unweighted UniFrac distance matrix, or Bray-Curtis dissimilarities (Supplementary Table S4). Because the ccd8b mutants exhibited some changes in root and shoot morphology and bacterial community composition and it is known that strigolactones affect mycorrhizal fungi in soil, the abundance of fungal genera was evaluated. STAMP analysis revealed significant changes in the abundance of specific fungal genera in the rhizosphere at all three growth stages, including the growth point differentiation (Fig. 5A), boot (Fig. 5B), and grain-fill (Fig. 5C) stages for the ccd8b mutant only. At the growth point differentiation stage, only one genus (Pyrenochaetopsis) was higher in relative abundance in the ccd8b mutant than in the wild type. Seven other genera, including Solicoccozyma, Arthrinium, Auxarthron, and others, were lower in relative abundance in the ccd8b mutant (Fig. 5A). At boot stage, 13 genera were differentially abundant when comparing the wild type with the ccd8b mutant. In all, 3 of the genera (i.e., Didymellaceae, Ceratobasidium, and Dothideales) increased in abundance in the ccd8b mutant, while the other 10 genera decreased in relative abundance (Fig. 5B). At the grain-fill stage, four genera Disciseda, Hymenoscyphus, Calvatia, and Chaetosphaeriaceae all decreased in relative abundance in the ccd8b mutant (Fig. 5C). In addition, five fungal genera were identified to be significantly (Pearson's χ² test, FDR < 0.01) associated with the amount of orobanchol in the ccd8b mutant, and all of these fungi were enriched in the ccd8b mutant (Supplementary Fig. S5B; Supplementary Table S3). Changes in fungal relative abundance in the ccd8b mutant in the other two compartments (soil and root) at all three plant growth stages were also assessed by STAMP analysis and are shown in Supplementary Fig. S6 and Supplementary Fig. S7, respectively.

Download as PowerPoint

DISCUSSION

Extensive studies in the last decade have demonstrated the essential roles of strigolactones in the regulation of plant developmental processes and plant responses to biotic and abiotic stresses, as well as stimulating the hyphal branching of AMF (Yoneyama and Brewer 2021). To study the impacts of strigolactones in sorghum, we created two SbCCD8 sorghum CRISPR-edited lines. As previously reported for other plant species such as Arabidopsis, pea, and rice (Umehara et al. 2008), strigolactone-deficient sorghum mutants SbCCD8 displayed four to six times more tillering, shorter stature, and decreased grain yields compared with that of the wild type Tx430. Tillering and dwarfing are important agronomic traits affecting crop yield. For example, the number of tillers in rice and wheat has a major (positive or negative) impact on grain yield, which might be manipulated by application or modulation of strigolactones levels. The direct impact of strigolactones on yield has been documented in a limited number of crop species. A recent study using a panel of 147 rice accessions showed that a partial loss-of-function allele of the DWARF17 (D17) gene, which encodes CCD7, contributed to increased tiller numbers and improved grain yield in rice (Wang et al. 2020). Although strigolactones were not measured in this rice study, the well-characterized strigolactone-deficient d17 rice mutant had decreased production of strigolactones (Umehara et al. 2008). In sorghum, plants with increased numbers of tillers may have reduced grain yield because of poor genetic gain, reduced grain weight, or the presence of nondormant axillary buds (Chen et al. 2018; Mwamahonje and Maseta 2018), which is similar to our findings with the ccd8b mutant, where excessive tillering reduced grain yield. Production of zucchini squash (Cucurbita pepo) and sweet orange (Citrus sinensis) has also been increased by strigolactone application under normal growth conditions (Pokluda et al. 2018; Zheng et al. 2018). However, the underlying mechanism of the improved production was not elucidated in those studies. Our data contribute additional specific insights into the molecular mechanisms of the strigolactone biosynthesis pathway in the regulation of plant development by strigolactones. In addition to shoot morphology, altered root architecture was observed in the CCD-knockout mutant plants. It is worth noting that data on the role of strigolactones in the regulation of root development differ according to plant species and growth conditions (Wu et al. 2022). For example, in both Arabidopsis and pea, strigolactone-deficient mutants exhibit more adventitious roots (Rasmussen et al. 2012), while strigolactone-deficient rice mutants have fewer crown roots than wild-type plants (Sun et al. 2014). The exact mechanism of the regulation of root development by the strigolactone signaling is unclear, and the complex crosstalk of strigolactones with other phytohormones such as ethylene, cytokinin, and auxin might contribute to determining differences in root architecture (Aliche et al. 2020).

Parasitic weeds of the genus Striga pose a major constraint for cereal production and food security in sub-Saharan Africa, where they are widespread and prevalent in a broad range of hosts, and where resources for their control are lacking (Mwangangi et al. 2021).Yield losses due to Striga infestation of more than 8.6 million tons of sorghum and millet and 2.1 million tons of maize have been reported, resulting in an estimated annual loss of approximately U.S.$7 billion in revenues (Ejeta and Gressel 2007; Gressel et al. 2004). The ccd8b mutant plants with undetectable levels of the orobanchol in the root exudates were resistant to colonization by the Kenyan Striga population that is equally responsive to orobanchol and 5-deoxystrigol (Bellis et al. 2020). The resistance of both the wild type Tx430 and the ccd8 mutants to colonization by the Malian Striga population is consistent with previous studies on the LGS1 gene. The sorghum Tx430 variety used for the transformation in this study has a deletion of the LGS1 gene on chromosome 5 leading to resistance to certain ecotypes of Striga (Yoda et al. 2021). LGS1 plays a critical role in the regulation of 5-deoxystrigol and in regulating the stimulant activity of Striga germination by sorghum (Gobena et al. 2017). Wild-type sorghum varieties that contain a functioning LGS1 produce 5-deoxystrigol and a low level of orobanchol, whereas lgs1 mutants only produce orobanchol but not 5-deoxystrigol, leading to enhanced resistance to the Malian population of Striga (Gobena et al. 2017) due to it being stimulated to germinate by 5-deoxystrigol. Consistent with the presence of the lgs1 mutation in Tx430, we observed that only orobanchol and not 5-deoxystrigol was detectable in both the wild type and ccd mutant plants (Yoda et al. 2021). Therefore, further research will be needed on the impacts of these CRISPR constructs in sorghum lines that produce both 5-deoxystrigol and orobanchol. Although the ccd8a mutant produced significantly lower amounts of orobanchol, it did not have significant effects on Kenyan Striga germination, possibly because the magnitude of the effects of this mutation on orobanchol production was not large enough.

Transcriptome, phylogenetic, and large-scale genomic sequencing studies have been conducted in different plant species (Priya et al. 2016, 2019; Vallabhaneni et al. 2010) to identify species-specific CCD genes with unknown biological functions. However, in-depth characterization of this gene family and the role its members play in different organisms are still lacking. To our knowledge, ours is the first study to characterize members of this gene family in sorghum, which is an important cereal crop, especially in Africa, where sorghum yields are greatly affected by Striga colonization and nutrient limitations. The CCD7 and CCD8 subfamilies are involved in the core pathway of strigolactone biosynthesis, which leads to the formation of carlactone (Alder et al. 2012). All CCD7 genes identified thus far are single copies, whereas CCD8 has two, four, and six copies in maize, rice, and sorghum, respectively (Vallabhaneni et al. 2010; Wang and Bouwmeester 2018). Mohemed et al. (2018) reported that the expression pattern of CCD7 and CCD8 (Sb03g034400.1; CCD8b in the present study) genes varies among different sorghum cultivars differing in strigolactone profile and Striga resistance. Specifically, the expression levels of CCD7 were higher in Striga-resistant genotypes than in Striga-susceptible genotypes. In contrast, the expression levels of CCD8 were higher in Striga-susceptible genotypes (Mohemed et al. 2018). The expression profiles of the CCD gene family in Sorghum bicolor were publicly available based on data from a deep transcriptome analysis, in which the sorghum CCD7 and CCD8 gene expression was measured in major plant tissue types across the juvenile, vegetative, and reproductive stages of development (McCormick et al. 2018). This expression profile showed that the CCD family genes are mainly expressed in the root, consistent with their function in synthesizing the root exudate, strigolactone. In addition, among the CCD family genes, CCD8b was highly expressed in the roots compared with that of other CCD genes at different plant growth stages. Therefore, our observation that the ccd8b mutants showed obvious phenotypes is consistent with the expression profile of this gene family (McCormick et al. 2018). In the present study, we demonstrated an effective approach to generate CCD knockout mutants with modified strigolactone production in sorghum using CRISPR/Cas9 system-mediated mutagenesis. Unfortunately, all of the independent T0 transgenic sorghum lines targeting the CCD7, CCD8d, or CCD8d-like gene were biallelic and not homozygous mutants, which is commonly seen in T0 plants presenting somatic mutations (Bari et al. 2019), and self-pollination in later generations did not generate homozygous mutants for these three genes. The different mutation efficiencies observed due to the CCD sgRNA constructs might be due to the variability in Cas9 activity. A deeper knowledge of the CCD gene network and gene functions will be necessary to elucidate the role that each isoform plays in strigolactone biosynthesis in sorghum.

The effects of the root exudation of strigolactones on the interactions of plants with other organisms in soils such as root and rhizosphere microbes are not yet fully understood. In the present study, we analyzed the bacterial and fungal microbial communities recruited to the soil, rhizosphere, and root of the wild type Tx430 and two sorghum ccd mutants across three plant developmental stages, aiming to test the hypothesis that altered production of orobanchol and structurally different strigolactones affected the global root, rhizosphere, and soil microbiome. A few recent studies using overexpression or knockout mutants for genes related to strigolactone biosynthesis or perception have demonstrated the role of strigolactones in influencing the composition of the rhizosphere microbial communities in Arabidopsis, soybean, and rice (Carvalhais et al. 2019; Kim et al. 2022; Liu et al. 2020; Nasir et al. 2019). In Arabidopsis, the max4 mutant with impaired production of strigolactones showed changed composition of rhizosphere fungal but not bacterial communities (Carvalhais et al. 2019). Overexpression of the strigolactone biosynthesis gene MAX1d and signaling perception genes MAX2a and D14 in soybean roots significantly altered the rhizosphere bacterial communities but not the fungal communities, including the symbiotic arbuscular fungal family Glomeraceae (Liu et al. 2020). Similar trends were also observed in rice, in which mutants defective in CCD7 and D14 genes demonstrated distinct differences in the rhizosphere bacterial community composition but less obvious effects on fungi (Nasir et al. 2019). A more recent study assessed the microbial communities recruited to the roots and rhizosphere of 16 rice genotypes with structurally different strigolactones in the root exudates. The authors suggested that orobanchol in rice roots, even though the amount is approximately 10-fold lower than the other two forms of endogenous strigolactones (4-deoxyorobanchol and methoxy-5-deoxystrigol isomer 1), might have a stronger effect on the composition of both bacterial and fungal communities (Kim et al. 2022). In sorghum, our study showed a significant effect on the diversity, richness and, composition of bacterial communities in the rhizosphere compartment at plant grain-fill stage due to the loss-of-function of CCD8b. In agreement with observations in rice and soybean with modulation of the strigolactone signaling and biosynthesis pathway (Liu et al. 2020; Nasir et al. 2019), disruption of CCD8b affects individual fungal taxa without a significant change at the whole-community level. Considering the contradictory observations in various studies, as well as the multifunctional features of strigolactones, further investigations are necessary to explore whether strigolactones interact with other phytohormones, or whether strigolactones influence plant–microbe interactions only at specific stage of plant development, as shown in this study.

The relative abundance of many bacterial and fungal taxa was altered by changes in the production of strigolactones in our study. Similar to our study with sorghum, the abundance of TM7, known as Saccharibacteria, was also shown to be negatively associated with the level of orobanchol in the rhizosphere compartment of rice (Kim et al. 2022). The endophytic bacterium Chitinophagaceae, which was reported to be present in wheat (Triticum aestivum) when the plants experience nitrogen deficiency (Pagé et al. 2019), was found to be negatively associated with orobanchol amount in rice (Kim et al. 2022) and our ccd8b mutant sorghum plants. Among all of the differentially abundant bacteria genera that we identified in the rhizosphere, RB41 was highly abundant in both the wild type and ccd8b mutant, with a significant reduction in ccd8b mutant compared with the wild type. RB41 belongs to the Acidobacteria phylum, which was shown to be more abundant in sorghum cultivar SRN-39, which exudes more orobanchol than six other genotypes (Schlemper et al. 2017), supporting the idea that the abundance of RB41 is positively associated with the amount of orobanchol. Acidobacteria was also significantly depleted in strigolactone-deficient rice mutants (d17 and d14) compared with the wild-type plants (Nasir et al. 2019). In addition, the majority of taxa that changed in abundance in the ccd8b mutant tended to decrease in abundance, consistent with the hypothesis that strigolactones are important in the recruitment of microbes. Regarding fungi, although we did not observe fungi compositional changes to strigolactone deficiency at the community level, we found significant changes at the individual taxa level in all three compartments in the ccd mutants. Furthermore, there was no difference in the relative abundance of bacterial or fungal genera between the other mutant (ccd8a) and the wild type, and a χ² test also failed to detect any relationship between the amount of orbanchol and the relative abundance of microbial or fungal taxa, suggesting the small effect of the ccd8a mutant or the low power of the current dataset. Due to the fact that many different forms of strigolactone exist in plants, and the production of strigolactones is strongly influenced by plant species as well as environmental conditions, additional mutants in different sorghum genotypes will be required to fully elucidate the interactions between plant root exudation of strigolactones and soil microbes.

In summary, the CRISPR/Cas9 system was used to induce targeted and heritable mutations of the CCD8 genes involved in strigolactone biosynthesis in sorghum. Knockout of the CCD8 genes in S. bicolor var. Tx430 abolished the production of the orobanchol form of strigolactone, and knockout of CCD8b reduced grain yield and altered the shoot and root architecture and sorghum susceptibility to the parasitic plant Striga. In addition, changes in root exudation of orobanchol affected the rhizosphere bacterial diversity and community composition and the abundance of specific fungal taxa. The fungal taxa that were differentially enriched due to changes in the exudation of orobanchol highlight the diverse roles of this plant hormone root exudate in plant–microbe interactions. Our findings also shed more light on a potentially sustainable approach for the control of the parasitic weed Striga through manipulation of strigolactone production. Further research should underpin the specific actions of different exudate mixtures of strigolactones produced by important crop plants on multiple Striga populations to gain in-depth knowledge of the functions that this plant hormone exuded from roots have on plant interactions with other soil organisms. In future, we envision that this research will lead to new insights into the application of strigolactones for the optimal recruitment of beneficial microbes and parasitic weed control for agriculture.

Data availability.

16S rRNA and ITS gene amplicon data are available at the NCBI Sequence Read Archive under accession number PRJNA832242.

ACKNOWLEDGMENTS

We thank the Nebraska Center for Biotechnology-Plant Transformation Core for the transformation of sorghum, P. Tenopir for maintaining and monitoring the permitted field for the transgenic sorghum experiments, S. Link for planting and maintaining plants for molecular analysis, and M. Butler for her help with the field work.