Introduction

Sesame (Sesamum indicum L.) is one of the oldest oil crops and is known as the “queen of oil seeds” because of its high oil content and quality (Yang and Huang 2009; Gebregergis et al. 2018). Sesame is an indispensable material for many traditional foods in many Asian countries such as China, Japan and Korea with extremely high nutritional value and a wide range of usages (Xu et al. 2018a, b). Furthermore, it has an important application value in medicine (Gharby et al. 2017). For example, sesame oil has been reported to possess anti-inflammatory and immunomodulatory functions (Hsu et al. 2005) and its lignan sesamin/sesamolin has anti-hypertensive and neuroprotective effects (Cheng et al. 2006; Nakano et al. 2010). In addition, sesame oil can also be used in other industrial sectors to make a synergist for pyrethrum insecticides, and cosmetics (Anilakumar et al. 2010).

Sesame production is mainly concentrated in Asia and Africa and the annual demand of sesame seeds is experiencing a steep increase (Dossa et al. 2017). Although the global sesame planting area is extending particularly in Africa, the productivity and yield are still very low, resulting in a huge gap between seed demand and supply (Sarkar et al. 2016). In fact, sesame is cultivated in harsh environments and its growth and development are greatly affected by the adverse conditions (Witcombe et al. 2008). In addition to its own genetic factors, abiotic or biotic stresses such as diseases, insects, drought, waterlogging are important factors leading to the low yield of sesame (Li et al. 2012; Dossa et al. 2019).

Root morphology and physiological characteristics are closely related to yield. Svacina et al. (2014) demonstrated that varieties with a larger root system showed a higher yield and better grain quality in barley. Similarly, Wang et al. (2015) found that the number of roots was closely related to the yield of maize. Studies on root traits are hampered by their complexity and the inaccessibility of the rhizosphere (Kuijken et al. 2015). However, with the development of root phenotyping platforms and root data analysis tools, our understanding of the root functions is increasing and the prospects for harnessing the potential of root systems to boost and stabilize crop yields are looming (Paez-Garcia et al. 2015). Liu et al. (2018) showed that the representative super hybrid rice cultivars had dense root and higher grain yield. In peanut, root dry weight, root volume, total root absorption area and active root uptake correlated with strong drought resistance (Li et al. 2014).

So far, there has been no specific performance in the sesame root breeding programmes although the root system is crucial for the plant functioning (Dossa et al. 2017). There are no data about the natural variation of sesame root traits and more importantly how the root traits affect the seed yield in sesame is unknown. To fill this gap, we designed this study and examined the morphological and anatomical variability of root traits in 40 sesame genotypes collected from different countries. We compared soil and nutrient solution plant culture systems to recommend an efficient approach to study sesame root traits at the large scale. In addition, we evaluated the contribution of root traits to seed yield in various environments. Also, we performed gene expression profiling in two contrasting sesame genotypes for root traits with the aim to reveal the candidate genes and biological pathways underlying the differential root phenotypes (Fig. 1). The goal of this work is to provide a comprehensive insight into the variability of sesame root systems at the morphological, anatomical and molecular levels for harnessing the potential of root traits in sesame breeding programmes.

Fig. 1
figure 1

Overview of the study

Full size image

Materials and methods

Plant materials

A total of 40 sesame (Sesamum indicum L.) genotypes were obtained from the Sesame Germplasm Resource Preserving Center, of the Oil Crops Research Institute-Chinese Academy of Agricultural Sciences (Wuhan, China). The genotypes originated in Asia, Africa and America (Table 1).

Table 1 Genotypes of Sesamum indicum collected from different countries used for root diversity analysis
Full size table

Plant growth in soil system

The trial was conducted in the Wuchang experimental station of Oil Crops Research Institute-Chinese Academy of Agricultural Sciences. Seeds were disinfected with absolute ethanol for 5 min and washed with 70% alcohol twice. Sterile seeds were grown in pots (25 cm diameter and 45 cm depth) filled with 6 kg of experimental soil composed of 1.5 kg vermiculite, 2 kg soil, 2 kg sand and 0.5 kg nutritive soil, and were maintained under shelter to avoid rainy days interfering. All plants were grown in a growth chamber under 16 h/8 h (day/night), 28 °C/25 °C (day/night), at a PPFD of 200 µmol m−2 s−1 and 60% relative humidity and were watered every 2 days to keep good growth conditions.

The experiment was carried out with five replications and two plants per pot, resulting in a total of 400 seedlings. The root of 45-day-old plants was delicately separated from the shoot and thoroughly washed and both samples were put into paper bags for later use.

Plant growth in hydroponic system

Round and uniform sesame seeds were sowed on gauze that was fixed to a plant germination tray (25 × 19 × 7 cm, length × width × height) filled with a quarter-strength modified Hoagland’s solution (Hoagland and Arnon 1950). The modified Hoagland’s solution consisted of: 5 mM Ca(NO3)2·4H2O, 6 mM KNO3, 2 mM NH4NO3, 2 mM MgSO4·7H2O, 1 mM KH2PO4, 0.5 mM K2SO4, 0.16 mM EDTA-Fe, 40 μM H3BO4, 9 μM MnCl2·4H2O, 0.6 μM ZnSO4·7H2O, 0.32 μM CuSO4·5H2O and 0.32 μM (NH4)6·Mo7O24·4H2O. The pH value was adjusted to 5.8 with NaOH or HCl using a portable PH meter. Seven days after sowing, uniform seedlings were selected and transplanted into blue plastic basins (34 × 26 × 12 cm, length × width × height) containing 160 seedlings of 40 genotypes (four replicated seedlings for each genotype). The experiment was performed under controlled environmental conditions with the temperature and relative humidity kept at 35 °C and 60%, respectively, and under long-day conditions (16 h day/8 h night). A randomized complete block design was used with three replications, resulting in a total of 480 seedlings. After the first week of germination, the seedlings were transferred into a half-strength Hoagland’s nutrient solution for three additional weeks. The nutrient solution was renewed once a week. One-month-old plants were harvested and the root was delicately separated from the shoot and both samples were put into paper bags for later use.

Root image scanning

At harvest, root samples were scanned using a desktop scanner (EPSON Perfection V800 Photo). Root images were analyzed using an image analysis system (WinRHIZO Pro software). Four root data including the total root length (RL, cm), root surface area (SA, cm2), root volume (RV cm3) and root number (RN) were extracted (see Abbreviations). The taproot length (MRL cm) and diameter per plant (MRD cm) were measured using the vernier caliper (only for soil-grown plants). Shoots and roots were then dried in a drying oven at 80 °C for 72 h and weighed to determine shoot dry weight (SDW g) and root dry weight (RDW g). Root–shoot ratio (RSR) data were then estimated (see Abbreviations).

Field trials and seed yield data recording

The 40 genotypes were planted in three different environments in China, including Wuhan, the Hubei province (30.57 N, 114.30 E), Nanning, the Guangxi province (23.17 N, 107.55 E) and Sanya, the Hainan province (109.187 E, 18.38 N). The field trials were conducted during the year 2017 using a randomized block design with three replications. Each plot had four rows of 2 m long spaced 0.4 m apart. At the four-leaf stage, seedlings were thinned down and eight evenly distributed plants in each row were retained. Five uniform plants for each genotype were randomly selected to collect the seed yield per plant data (g) after maturation.

Root phenotypic data analysis

Descriptive statistics of the experimental data were implemented in the R software (www.r-project.org). Normal distribution of data was determined using the Shapiro–Wilk W test, while homogeneity of variances was determined with the Bartlett test. ANOVA results were considered significant at P < 0.05 and mean comparisons were done using the Tukey HSD test. Agglomerative hierarchical clustering (AHC) was performed based on the Ward’s method using the Euclidean distances. Then, discriminant analysis (DA) was carried out to refine the clustering in order to sort out the different groups of genotypes according to the root trait characteristics. Pearson correlation coefficients of the root traits, the AHC and DA analyses were performed using the software XLSTAT version 2016. To compare the two methods of plant culture systems (Soil vs Hoagland), we employed the MANOVA with the Wilks lambda test.

Root anatomical experiment and data analysis

Roots from plants grown in soil were collected carefully by hand and then washed with deionized water to remove soil particles. Root anatomical characteristics were evaluated in samples. Tissue samples observed under microscopy were collected from 10 mm up of the taproot tip. The root tip was cut into slices, and cross-section was obtained. A few drops of dihydrochloride, DAPI dye solution were added to the prepared glass slides for staining. Pictures were taken using a LEICA microscope. For each root transverse slice, anatomical traits including root diameter, stele diameter, cortical thickness, mean diameter of cortical cells, cortical cell layer number, number of large xylem, number of small xylem were evaluated using the ImageJ software version 1.8.0_172. For each genotype, the average values of these anatomical traits were calculated from three cross sections.

RNAseq experiment and data analysis

Plant materials

Two genotypes including G546 (big root phenotype) and G259 (small root phenotype) were used for RNA-seq study. The whole root samples were harvested from three plants of the same genotype after 1 month growth in Hoagland nutrient solution as described above.

Total RNA isolation and sequencing

Total RNA of the six root samples (50 g with mixed taproot and lateral root) was extracted using an EASYspin Plus kit (Aidlab, Beijing, China). The cDNA libraries generated from RNA samples were pair-end sequenced on an Illumina Hiseq 4000 platform according to the methods described by Dossa et al. (2017).

RNA-seq data analysis

The raw data were first processed with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) to filter out adapters and low-quality sequences. Then, the clean reads were mapped to the sesame genome v.1.0 (https://www.ncbi.nlm.nih.gov/genome/?term=sesamum) (Wang et al. 2014) using HISAT (Kim et al. 2015). The RSEM package (Li and Dewey 2011) was used to calculate gene expression levels for each sample expressed as fragments per kilobase of transcript per million fragments mapped (FPKM). The gene expression levels in G546 were compared with those in G259 to identify the differentially expressed genes (DEG). The DEGs were detected as described by Tarazona et al. (2011) based on the parameters: fold change ≥ 2.00 and Probability ≥ 0.8 with a significant false discovery rate-adjusted P value (FDR) < 0.05 based on the three biological replicates. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses for the DEGs were performed using the clusterProfiler version 3.8.

Validation of gene expression using qRT-PCR

The qRT-PCR was performed on RNA extracted from root samples as described by Dossa et al. (2016) using the sesame gene Actin 7 (SIN_1006268) as the internal control. Specific primer pairs of ten selected genes were designed using the Primer Premier 5.0 (Lalitha 2000) (Table S1). Data are presented as relative transcript level based on the \(2^{{ - \Delta \Delta C_{\text{t}} }}\) method (Livak and Schmittgen 2001).

Results

Sesame root traits variation in soil culture system

Forty sesame genotypes were grown in pot containing 6 kg soil for 45 days and nine root traits were assessed (Table 1, see Abbreviations). Table 2 presents the mean value and range of the assayed root traits data. Overall, a large variation for root traits were observed among the sesame germplasm, which is cardinal for harnessing the potential of root traits in sesame improvement programmes. Pearson correlation analysis (R) of the root traits showed that the root surface area (SA) was highly and positively correlated to the total root length (RL), root volume (RV), root number (RN) and root dry weight (RDW), indicating that SA is a good trait that can be used to predict several root components in sesame and could be prioritized for a large-scale root phenotyping. Interestingly, a positive and strong correlation was found between shoot (SDW) and root dry weight (RDW), implying that the sesame plants with extensive aboveground parts tend to develop dense roots (Table 3).

Table 2 Variation of root in 40 sesame genotypes grown in soil culture system
Full size table
Table 3 Correlation analysis of root traits from soil-grown sesame plants
Full size table

Discriminant analysis of the genotypes based on the root phenotypic data resulted in three different clusters (Fig. 2a). Cluster I contained 22 genotypes and was characterized by the highest mean values for RL, SA, RV, RSR and moderate mean values for RDW, SDW and RN. Cluster II encompassed five genotypes with low mean values for RL, SA, RV, RDW, RSR, RDW, RSR, but moderate mean values for RN. Cluster III contained 13 genotypes and compared with the other clusters, and its genotypes had moderate mean values for RL, SA, RV, RSR and high mean values for RN, SDW and RDW. Cluster I could be designated as big root biomass genotypes, cluster II could be named as small root biomass genotypes and cluster III could be called as moderate root biomass genotypes.

Fig. 2
figure 2

Discriminant analysis of the studied population. a Soil culture system. Blue dots, big root biomass genotypes; green dots, small root biomass genotypes; red dots, moderate root biomass genotypes. b Hoagland solution culture. Green dots, big root biomass genotypes; blue dots, small root biomass genotypes; red dots, moderate root biomass genotypes

Full size image

Root traits variation in the Hoagland nutrient solution culture system

Hoagland nutrient solution is a plant culture system widely used to investigate root traits, particularly for a large-scale phenotyping. In the attempt to evaluate the response of sesame root under hydroponic growth condition, the 40 genotypes were grown in 1/4 strength Hoagland nutrient for 1 week and subsequently transferred into 1/2 strength Hoagland nutrient solution for an additional 3 weeks. Similar to the soil-grown culture, we observed a high variability for the root traits as summarized in Table 4. Discriminant analysis revealed three groups of genotypes based on their root phenotypic traits (Fig. 2b). Cluster I grouped 15 genotypes characterized by low RDW, SDW, RSR, RL, SA, RV and RN values. Cluster II encompassed 12 genotypes characterized by high RDW, RSR, RL, SA, RV and RN but moderate SDW values. Finally, cluster III gathered together 13 genotypes which have moderate RDW, SDW, RSR, RL, SA, RV, RN and high SDW. We, therefore, named cluster I as small root biomass genotypes, cluster II as big root biomass genotypes and cluster III as moderate root biomass genotypes.

Table 4 Variation of root traits in 40 sesame genotypes grown in Hoagland nutrient solution
Full size table

Comparing the grouping patterns of the genotypes from the two culture systems, we could observe that many genotypes were constitutively classified as big root biomass genotypes (G098, G221, G340, G349, G463, G546) or small root biomass genotypes (G259, G608, G698). This result indicates that highly contrasting genotypes for root traits could be clearly distinguished regardless of the growth medium. Next, the root traits measured from Hoagland solution culture and soil culture systems were individually compared using ANOVA. The results showed that except for SDW and RL, all the remaining traits varied significantly according to the plant culture systems (Table 5). In general, root trait values from the soil culture were significantly higher than those from the Hoagland solution culture. Only RN showed the opposite trend, suggesting that the Hoagland solution culture tends to increase the overall root number. By integrating all the root traits, the MANOVA test revealed that the two methods of plant culture induce significantly different root phenotypes in sesame.

Table 5 Comparison of root trait data from soil and Hoagland-grown plants
Full size table

Anatomical diversity of the sesame root

We selected nine genotypes, including seven with a big root biomass and two with a small root biomass (Fig. 2a) to investigate for the first time the diversity of anatomical features in sesame taproot. The variations of root diameter, stele diameter, cortex cell layer number, number of large xylem, number of small xylem are described in Table 6. Because of the high discrepancy of sample sizes among genotypes with big root biomass and small root biomass, we did not perform comparative statistical tests. Nonetheless, it could be noticed that the big root genotypes had globally higher values for root diameter, stele diameter, cortical thickness, number of large xylem and number of small xylem than the group of small root biomass genotypes. G546 had the thickest root (15.36 mm) followed by G256 (14.48 mm) and G340 (14.30 mm). The ratio of cortical thickness to root diameter of the big root genotypes was slightly smaller than that of the small root genotypes, while the ratio of the stele diameter to the root diameter showed the opposite trend. The reduced ratio of cortical thickness in favor to the stele diameter may be beneficial for nutrient and water transportation in root. The cortical cell layer numbers were similar among genotypes with big root or small root phenotypes, ranging from 5 to 7 layers. Furthermore, the total number of xylem vessels of big root sesame genotypes was larger than that of small roots genotypes. Overall, sesame harbors an important anatomical diversity, which seems to be linked to the global root biomass.

Table 6 Differences in anatomical characters of big root and small root
Full size table

Relationship between root biomass and seed yield in sesame

To determine the contribution of root biomass to seed yield in sesame, the 40 genotypes were planted in three different environments in China, including Wuhan, Nanning, and Sanya. The average seed yield per plant (g) was estimated for each genotype in each environment. We classified the genotypes according to the three clusters obtained from the DA based on the soil culture (Fig. 2a). As shown in Fig. 3, except for Sanya location (P < 0.3949), the genotypes with big root biomass had significantly higher seed yield than the genotypes with small root biomass [Nanning (P < 0.0006), Wuhan (P < 0.0283)], denoting that the root biomass positively contributes to increased seed yield in sesame. It is worth mentioning that all the genotypes did not grow well in Sanya location compared to the other two locations (as evidenced by the very low seed yield in Sanya) (Fig. 3) and this may explain the lack of significant differences of the seed yield between cultivars with different root types.

Fig. 3
figure 3

Seed yield per plant (g) of 40 sesame genotypes classified into three groups according to their root biomass. Data were recorded in three locations in China (Wuhan, Nanning and Sanya) and the P values represent Tukey HSD tests from ANOVA

Full size image

Global gene expression profiles in G546 and G259, two contrasting genotypes for root phenotype

Two genotypes including G546 with the biggest root biomass and G259 with the smallest root biomass (Fig. 4), constitutively identified from the soil-grown and Hoagland-grown culture systems, were used for RNA-seq analysis. With three biological replicates, transcriptome sequencing of the six root samples yielded a total of 33.54 Gb clean data with 93.22% of bases scoring Q30 and above (Table S2). Of the total clean reads, 82–93% were unique matches with the Sesamum indicum reference genome V1.0 (Wang et al. 2014) (Table S3). A total of 21,801 and 21,679 expressed genes in G546 and G259 were detected, respectively, resulting in 22,417 unique genes in sesame root. We identified 21,063 genes commonly expressed in both genotypes, while 738 and 616 genes were genotype-specific genes and may participate in the observed root phenotypes (Fig. 5a). Hierarchical clustering of the samples based on the number of fragments per kilobase of exon per million fragments mapped (FPKM) showed that all the biological replicates clustered together, indicating the high reliability of our sequencing data (Fig. 5b). The FPKM values range from 0.00333 to 19900.14667 in G546 and from 0.0033 to 26292.66 in G259. Overall, similar global gene expression levels could be observed in samples from the two genotypes (Fig. 5c).

Fig. 4
figure 4

Two sesame genotypes with contrasting root phenotypes detected after screening. a Phenotype of G546 and G259 based on soil culture system after 45 days. b Root morphology of G546 and G259 based on Hoagland solution culture after 1 month. c Scanned root of G546 and G259 based on soil culture system after 45 days

Full size image
Fig. 5
figure 5

Transcriptome sequencing in two sesame genotypes with contrasting root phenotypes. a Shared and specific expressed genes between G546 and G259. b Tree showing that the three replicates were very close and clearly separated the two genotypes. c The log10 FPKM values range in G546 and G259

Full size image

Differentially expressed genes between G546 and G259

We compared the gene expression profiles in roots of the two genotypes to detect the differentially expressed genes (DEGs), which underlay the differences in root phenotypes. We identified 1831 and 1066 up- and downregulated genes in G259, respectively. These DEGs contributed principally to the GO terms related to sequence-specific DNA-binding transcription factor activity, heme binding, protein kinase activity and iron binding (Fig. 6a), highlighting the key roles of transcription factors in the root phenotype in sesame. We further investigated the possible physiological and biochemical pathways involved in sesame root phenotypes through KEGG database. The DEGs between the two genotypes are enriched in several pathways, predominantly phenylpropanoid biosynthesis, starch and sucrose metabolism, stilbenoid, diarylheptanoid and gingerol biosynthesis, and flavonoid biosynthesis (Fig. 6b), indicating that these pathways are crucial for the root growth and development in sesame. The top up- and downregulated genes include several unannotated genes but also TFs and some structural genes, which could be functionally characterized using systems biology approaches (Fig. 7a). To validate our RNA-seq data, we selected five upregulated genes and five downregulated genes (Table S1) and performed qRT-PCR using the cDNAs from root of the two genotypes (G546 and G259). The qRT-PCR results were strongly correlated with the RNA-seq data (r2 = 0.8, Fig. S1). This result confirms well the high reliability of the RNA-seq data obtained in the present study.

Fig. 6
figure 6

Functional classification of the assembled unigenes. a Functional classification based on GO enrichment. b Pathway assignment based on the KEGG enrichment

Full size image
Fig. 7
figure 7

Differentially expressed genes between G546 and G259 a Top 20 up- and downregulated genes between G546 and G259. Values represent the fold changes in gene expression. b Major transcription factor families upregulated in G259. c Major transcription factor families downregulated in G259

Full size image

Enriched transcription factor families in the DEGs between G546 and G259

Transcription factors (TF) coordinate the expression levels of target genes to maintain a particular phenotype (Feng et al. 2018). To identify the main regulators of gene expression between the two genotypes, we investigated the enriched TF families among the DEGs. Prediction of the total unigenes encoding TF in sesame resulted in 58 TF families with a total of 2326 TFs. Among the upregulated genes in G259, 196 are TFs belonging to 35 families, predominantly ERF, WRKY, MYB and NAC, implying that these genes positively participate in the small root biomass phenotype (Fig. 7b). Likewise, 54 downregulated genes in G259 are TFs belonging to 21 families with bHLH, HD-ZIP, GRAS and LBD as the most enriched ones (Fig. 7c). These TFs may play key roles in the rapid root growth and high root biomass in sesame.

Discussion

The root system plays an important role in the growth and development of plants, including holding plants to the soil, absorbing water and nutrients for plant growth and acting as storage organs (Paez-Garcia et al. 2015). Illuminating the variability of sesame root system at the morphological, anatomical and molecular levels traits could provide desirable root characteristics and candidate genes for breeding programmes. Root traits have been extensively studied and harnessed to improved yield and resistance to biotic and abiotic stress in soybean (Fried et al. 2018), rice (Xu et al. 2018a, b), oilseed rape (Wang et al. 2017), wheat (Wasson et al. 2014), maize (Ju et al. 2018), etc. However, there has been no specific performance in the root breeding programmes in sesame. In this study, the selected 40 sesame genotypes are from various countries over the world (Table 1), so they have a good representativeness and genetic diversity, which can reflect the different root types of sesame. Through the statistical analysis of the assayed root traits in the 40 sesame genotypes, we found that sesame holds a large diversity (Table 2). This is very interesting because this natural variation will allow the detection of SNPs/candidate genes/QTLs, which can be directly employed in molecular breeding programmes (Wei et al. 2015; Dossa et al. 2019). Since the observed root variability in this study was mainly based on plants grown in controlled conditions, in future study, it is important to confirm these results in field non-controlled conditions.

The root research in crops has been mainly hampered by the difficulty to access the rhizosphere (Kuijken et al. 2015). Much time is required to wash plant roots from soil and separate the living plant roots from organic debris. In particular, for a large-scale plant root phenotyping, it is almost impossible to investigate accurately thousands of soil-grown plants. Therefore, various platforms and growth mediums have been proposed to study the root system, each with its advantages and drawbacks (Atkinson, et al. 2019). The effect of different culture conditions on the root systems of sesame was compared using two culture systems: Hoagland nutrient solution and soil culture (Fig. 1). MANOVA revealed that the two systems of plant culture induce significantly different root phenotypes in sesame, the root traits from the soil culture being significantly higher than those from the Hoagland solution culture except for the root number (Table 5). Our results are consistent with the study of You and Zhao (2016) who showed that hydroponic roots formed a growth characteristic that adapt to the water environment with smaller root diameter and lower root surface area. Similarly, Liu et al. (2017) reported in maize seedlings that the hydroponics system is characterized by a higher crown root number compared to the vermiculite culture system. In fact, the root structure changed significantly under hydroponic conditions in our experiment, with the formation of a large number of adventitious roots, leading to a higher number of roots in hydroponics culture than in soil culture. Young roots from hydroponics cultured sesame plants have no friction of soil particles and are not constrained by the soil resistance, so the epidermis gets thinner and their SA, RDW and RV are smaller than plants from the soil culture (Table 5). Importantly, the duration of the plant growth in soil was 45 days while in the Hoagland solution, it was 30 days and this may have also contributed to the higher values of root parameters in soil condition. Nonetheless, based on the clustering of the genotypes in the two methods (Fig. 2a, b), we found that similar genotypes were commonly clustered either in the small-root or in the big-root group, indicating that high-contrasting genotypes for root traits could be clearly distinguished regardless of the method used. In this regard, we can recommend the Hoagland solution culture for a large-scale root phenotyping in sesame.

Anatomical variation was associated with differences in functional efficiency and stress response of several crops, including maize (Zhu et al. 2010), Erianthus (Shiotsu et al. 2015) and rice (Singh et al. 2013). Kondo et al. (2000) reported that root diameter was highly correlated with stele diameter, which perfectly matched our results. Shiotsu et al. (2015) showed that compared with other cereal crops, Erianthus root were larger, correlating with a stronger water absorption and transportation. Because the genotypes with big root phenotypes had higher values for root and stele diameter and vessel number in the present study (Table 6), it is tempting to speculate that the sesame root biomass is tightly associated with the anatomical features. Selection and breeding for better root phenotypes based on the root anatomical features can improve acquisition of soil resources and crop production in marginal environments. Hence, the observed anatomical diversity may be essential in sesame root breeding programmes.

Mounting evidences demonstrate the positive contribution of root traits to crop yield (Ali et al. 2016; Sandhu et al. 2017; Xie et al. 2017; Beyer et al. 2019). Accordingly, in this study, we showed that sesame genotypes with a big root phenotype had significantly higher seed yield than those with a small root phenotype (Fig. 3). This extremely exciting result represents a solid proof that increasing sesame root biomass through breeding techniques will positively contribute to the seed yield improvement. Future investigations are needed to uncover the specific root traits to prioritize in sesame root breeding programmes. For example, in the present study, we found that the root surface area (SA) was highly and positively correlated with many other root traits (Table 3) and could, therefore, be the target in sesame root breeding programmes. However, how the big root biomass aids sesame plants to improve the seed yield is still elusive. We observed a positive and strong correlation between shoot and root dry weights traits (Table 3), suggesting that dense and big root biomass may help the plant to acquire more nutrient and water from the soil, which will ultimately promote the aboveground biomass formation. Since the shoot dry weight (SDW) is an easier trait to record compared to SA, SDW can be used as a proxy to detect sesame genotypes with dense root phenotype.

Transcriptome sequencing technology has been successfully applied to understand the molecular mechanisms underlying important agronomic traits in plants (Chen et al. 2018; Wan et al. 2018; Zeng et al. 2018; Arro et al. 2019; Wang et al. 2019; Xiao et al. 2019). In the present study, RNA-seq technology was employed to analyze the transcriptome of the genotypes G546 (big root phenotype) and G259 (small root phenotype). In organisms, different gene products coordinate with each other to perform biological functions, and annotation analysis of differentially expressed genes helps to interpret gene function (Cheng et al. 2018; Wu et al. 2018). The KEGG enrichment analysis of the differentially expressed genes between G546 and G259 highlighted important pathways involved in the differential root phenotypes, including phenylpropanoid biosynthesis, starch and sucrose metabolism and stilbenoid, diarylheptanoid and gingerol biosynthesis (Fig. 6b), which were reported to contribute to root growth and development in plants (Li and Zhang 2003; Sun et al. 2015; Jiang et al. 2017). Jiang et al. (2017) indicated that gingerols and diarylheptanoids were predominantly synthesized in fibrous roots of ginger and then transported to rhizomes, where they accumulate. Li and Zhang (2003) showed that development of sweet potato storage root coincided with starch accumulation. Also, most of the transcripts encoding sucrose synthase in Rehmannia glutinosa were upregulated in the developing tuberous roots (Sun et al. 2015). The GO functional classification of the differentially expressed genes showed that transcription factors (TF) are active in sesame root development (Fig. 6a). It is well known that TFs play an important role in the regulation of plant growth, evolution, and stress resistance (Sun et al. 2009; Gujjar et al. 2014; Feng et al. 2018). ERF and WRKY TFs were highly active in the small root genotype, while the bHLH and HD-ZIP were upregulated in the big root phenotype (Fig. 7b, c), denoting that some TF families participate in the growth and root biomass accumulation while some play the opposite role. The top five genes in this study that are upregulated in G546 with transcription factor activity include SIN_1000080 (bHLH), SIN_1008247 (B3), SIN_1010728 (ERF), SIN_1001533 (HSF), SIN_1022669 (B3), revealing that the mentioned genes may positively regulate the growth, development, and metabolism of sesame roots. The role of these key genes in root growth and development and the association with the seed yield performance in sesame remain to be further studied.

Author contribution statement

XZ and KD conceived, designed and supervised the experiment; RZ, MAM, DL, LQ, AL, JW, YG, MW, LS, JY, ZW, XZ, KD performed the experiments. KD and RS conducted data analyses; KD and RS wrote the manuscript; All authors read and approved the final version of the manuscript.