Keywords

14.1 Introduction

14.1.1 What Is the Ideotype for the Second Green Revolution?

Half a century ago, ideotype breeding using a semidwarf gene, sd1, resulted in the Green Revolution in rice (Khush 2001), but breeders and researchers still have to increase rice yield because rice is a staple food for nearly half of the world’s population. Thus, we need to consider what the ideotype for the second Green Revolution could be. Recent climate change is increasing the inequality of water and nutrient distribution in agricultural lands at a global level; global warming has caused serious drought damage in farmlands that rely on rainwater or that have limited access to irrigation (Scheffran and Battaglini 2011). What kinds of traits are needed to improve rice yield in this situation?

Root system traits should be such candidate traits to achieve the second Green Revolution. Root, which is the only organ absorbing water and nutrients from soil, is imperative for the survival of terrestrial plants, which cannot move around after germination. Optimal root development and distribution allow efficient acquisition of water and nutrients, which are heterogeneously distributed in soil (Gowda et al. 2011; Lynch 2013). For example, the topsoil tends to hold less water but more immobile nutrients such as phosphorus than does the subsoil. Nitrate, which is the main form of nitrogen under aerobic conditions, is leached by precipitation into subsoil. Therefore, root system architecture greatly affects the acquisition of water and nutrients from soil (Gewin 2010; Lynch 1995). Many wild species tend to have adequate root systems to adapt to severe stresses (Canadell et al. 1996); for example, drought-resistant plants tend to develop deeper root systems, which allow them to capture water from subsoil and thus avoid drought stress. Increased roots in subsoil would be also effective to avoid the negative impacts of drought on crop yield (de Dorlodot et al. 2007). Therefore, the genetic improvement of root system architecture has been regarded as an important approach to enhance crop production. However, it is more laborious and time-consuming to select root traits than aboveground traits. Therefore, many researchers are considering molecular breeding by using quantitative trait loci (QTLs) as one of the promising strategies for improving root system architecture (de Dorlodot et al. 2007; Yamamoto et al. 2014).

14.1.2 Prospects of Ideotype Breeding Using QTLs for Root System Architecture

In rice, a wide natural variation of root system architecture has been reported (Lafitte et al. 2001; O’Toole and Bland 1987; Uga et al. 2009). For example, typical upland rice has deeper, longer, and thicker roots than lowland rice (O’Toole and Bland 1987) and thus might be a useful breeding material to improve rice productivity. Root system architecture is a complex trait controlled by tens to hundreds of genes (Wachsman et al. 2015). Courtois et al. (2009) summarized 675 rice QTLs for 29 root parameters detected in 12 mapping populations reported in 24 published papers, but only one QTL associated with root system architecture has been isolated as a single gene in rice (Uga et al. 2013a).

The rice QTL DEEPER ROOTING 1 (DRO1) has been identified on chromosome 9 in recombinant inbred lines derived from a cross between the lowland cultivar “IR64” and the upland cultivar “Kinandang Patong” (Uga et al. 2011) and has been cloned (Uga et al. 2013a). “Kinandang Patong,” which has a functional allele of DRO1, has deep roots, whereas “IR64,” which has a nonfunctional allele, has shallow roots. In a near-isogenic line (Dro1-NIL) that carries DRO1 derived from “Kinandang Patong” in the genetic background of “IR64,” DRO1 increases root growth angle, resulting in deep rooting, but has a limited effect on other root and shoot traits. Field experiments using this unique line, which differs from “IR64” only in the increased root growth angle, have demonstrated that alteration of root system architecture improves rice productivity, as discussed in detail below (Fig. 14.1).

Fig. 14.1
figure 1

Schematic models of the effect of DRO1 on rice production and phytoremediation. (a) Drought avoidance (water uptake). In upland fields, where water is most abundant in subsoil, deep rooting caused by DRO1 allows plants to efficiently capture water. (b) Yield performance (nitrogen uptake). Under irrigated conditions in paddy fields, deep rooting caused by DRO1 allows plants to access nitrogen from the subsoil during reproductive stages. (c) Phytoremediation (Cd uptake). In a rain-fed paddy field after drainage, the bioavailable Cd concentration increases in the topsoil. Shallow rooting caused by dro1 allows plants to accumulate bioavailable Cd from the topsoil. (d) Lodging resistance. In wet paddy fields, deep rooting caused by DRO1 increases pushing resistance (an index of lodging resistance). Rice plants described in each part are IR64 on the left and Dro1-NIL on the right

Full size image

The yield performance of “IR64” and Dro1-NIL was compared under upland field conditions with no drought, moderate drought, or severe drought (Uga et al. 2013a). Under moderate drought in comparison with no drought, the grain weight of “IR64” decreased by nearly half, whereas that of Dro1-NIL was almost the same. Under severe drought, the grain weight of “IR64” was very low, whereas that of Dro1-NIL was more than 30% of that with no drought. This study suggests that deep rooting induced by DRO1 enhances drought avoidance, resulting in higher grain yield (Fig. 14.1a).

Comparison among cultivars with different root and shoot morphologies has suggested that deep roots increase grain yield in paddy fields (Kawata et al. 1978; Morita et al. 1988), but the genetic aspects of this effect have not yet been clear in previous studies. In paddy fields, Dro1-NIL showed about 10% higher grain yield than did “IR64,” irrespective of nitrogen treatment (Arai-Sanoh et al. 2014). There was no significant difference between “IR64” and Dro1-NIL in nitrogen content before heading, but nitrogen uptake was higher in Dro1-NIL than in “IR64” after heading. These results suggest that deep rooting induced by DRO1 enhances nitrogen uptake from lower soil layers, resulting in better grain filling (Fig. 14.1b).

Because root growth angle influences the efficiency of nitrogen absorption, it might also affect the uptake of other minerals such as heavy metals. In Cd-contaminated soil, the grain and straw Cd concentrations were significantly higher in “IR64” than in Dro1-NIL (Uga et al. 2015a). These results were opposite to those in the case of nitrogen uptake because the bioavailable Cd concentration was increased in the oxidized topsoil layer by withholding irrigation water during the vegetative growth period. Therefore, shallow roots could capture Cd from topsoil, resulting in a high concentration of Cd in rice plants (Fig. 14.1c). This suggests that, for phytoremediation, the allele occurring shallow rooting is a potential genetic resource for developing plants with high Cd accumulation. From the viewpoint of food safety, the allele giving deep rooting could be a useful resource to avoid absorbing the bioavailable Cd from topsoil.

Comparisons of different cultivars and examination of the effects of root pruning suggest that deep roots influence lodging resistance (Sakata et al. 2004; Terashima 1997; Terashima et al. 1994, 1995), although the genetic aspects of this effect are still unknown. Under wet paddy field conditions, Dro1-NIL had stronger pushing resistance (an index of lodging resistance) than “IR64” (Arai-Sanoh et al. 2014), suggesting that deep rooting induced by DRO1 improves lodging resistance (Fig. 14.1d).

These field experiments with Dro1-NIL confirm the potential of ideotype breeding for root system architecture, although further studies in other environments are needed. To design new root ideotypes that are adapted to diverse environmental stresses and to conduct ideotype breeding by using marker-assisted selection (Coudert et al. 2010), it will be necessary to update our understanding of the genetic mechanism associated with root system architecture. Information on gene networks involved in root formation has been accumulated for the model dicot plant Arabidopsis thaliana, but our knowledge of these aspects in rice is limited (Coudert et al. 2010; Rebouillat et al. 2009). In this chapter, we overview this knowledge and discuss the prospects of applying it to molecular breeding.

14.2 Root Formation

The first root of a rice plant, the radicle, is generated during embryogenesis. A radicle primordium originates endogenously from the embryo 4 days after pollination (DAP) (Itoh et al. 2005), whereas in Arabidopsis it is exogenously differentiated from the hypophysis (Dolan et al. 1993). After germination, the radicle is named the seminal root. Crown roots originate from the parenchyma cells adjacent to the peripheral cylinder of vascular bundles of the stem; therefore, crown roots are also named shoot-born roots (Fig. 14.2). Monocots develop a fibrous root system characterized by numerous crown roots, meanwhile dicots consist of only a main root system (Klepper 1992). A rice plant usually generates several hundreds of crown roots under field conditions (Klepper 1992). Lateral roots grow from seminal and crown roots (Fig. 14.2) and are responsible for taking up most water absorbed by the root system (Varney et al. 1993). In rice, two distinct types of lateral roots have been identified (Kawata and Shibayama 1965). L-type lateral roots are generally long and thick and are able to generate higher-order lateral branches, whereas S-type lateral roots are short, slender, and non-branching. Crown and lateral roots are classified as postembryonic roots because they are initiated after embryogenesis, whereas the seminal root (radicle) is generated during embryogenesis. Root hairs are tubular outgrowths of some root epidermal cells. Root hairs occupy most of the root surface area; they are thought to be important for water and nutrient uptake, anchorage, and interactions with soil microbes (Kim et al. 2007). Identification of genes associated with the formation of the different types of roots in rice has progressed together with the advances in molecular biology and DNA sequencing technology (Table 14.1).

Fig. 14.2
figure 2

Schematic view of the fibrous root system in rice

Full size image
Table 14.1 Root development-related genes in rice
Full size table

14.2.1 Root Apical Meristem

Except root hairs, each root type has a multicellular organization that can be described in terms of proximal-distal and radial polarity. All root cells are generated from stem cell daughters in the root apical meristem (RAM). Coordinated balance between cell division and differentiation is observed along the proximal-distal axis in the root tip, which can be divided into three zones according to cell division and elongation status: proximal division zone, transition zone, and distal elongation zone (Dolan et al. 1993). In rice, the root tip is formed by different types of cells arranged in concentric layers. The stele consists of the metaxylem, phloem, fibers, and pericycle and is surrounded by the endodermis, cortex, sclerenchyma, exodermis, and epidermis. In rice, these five cell layers are generated from single epidermis-endodermis structural initial cells by eight successive asymmetrical periclinal cell divisions following the first anticlinal division (Rebouillat et al. 2009).

The central region of RAM contains mitotically inactive cells, or the quiescent center (QC). The QC region in rice was examined by in situ hybridization with a probe for the rice CYCLIN-DEPENDENT KINASE (CDK) gene, a marker of cell division (Umeda et al. 1999). The analysis suggested that rice QC is large, unlike that of Arabidopsis, which consists of only four cells. In rice, the expression of a WUSCHEL (WUS)-type homeobox gene designated QUIESCENT-CENTER-SPECIFIC HOMEOBOX (QHB) was detected in the central cells of QC, similar to the expression of its Arabidopsis ortholog WUS-RELATED HOMEOBOX 5 (WOX5). WOX5 is involved in QC maintenance, suggesting that QHB plays a similar role in rice (Kamiya et al. 2003b; Sarkar et al. 2007). The phytohormone auxin is crucial for QC maintenance (Friml et al. 2002; Sabatini et al. 1999). A mutation in a member of the AUXIN (Aux)/INDOLE-3-ACETIC ACID (IAA) gene family, OsIAA23, causes defects in postembryonic QC maintenance due to the disintegration of the root cap and termination of root growth, suggesting the importance of auxin in rice QC maintenance (Ni et al. 2011). The GRAS family genes SCARECROW (SCR) and SHORT-ROOT (SHR) are also the key regulators of QC maintenance and root radial patterning (Di Laurenzio et al. 1996; Helariutta et al. 2000; Sabatini et al. 2003). OsSCR1 is specifically expressed in the endodermis, whereas OsSHR1 is expressed in the stele, similar to the patterns of their expression in Arabidopsis; OsSCR1 and OsSHR1 interact with each other when produced in yeast, similar to SCR and SHR in Arabidopsis (Cui et al. 2007; Kamiya et al. 2003a). These data suggest that OsSCR1 and OsSHR1 control the division of the epidermis-endodermis initial cells in rice. Concerning the outer cell layers (epidermis, exodermis, and sclerenchyma) in rice, a mutation in DEFECTIVE IN OUTER CELL LAYER SPECIFICATION 1 (DOCS1), which encodes a leucine-rich repeat receptor-like kinase (LRR-RLK), causes irregular epidermal cells with far fewer root hairs and transformation of some exodermal cells into additional sclerenchyma-like cells (Huang et al. 2012). The outer cell layers play an important role in protecting the inner root tissues from various stresses (Huang et al. 2009). Proper development of both inner and outer cell layers is essential for root development.

14.2.2 Radicle

A fertilized egg (zygote) undergoes iterative cell divisions to form a globular embryo with no apparent morphological differentiation until 3 DAP, and a radicle primordium is observed at 4 DAP together with a shoot apical meristem (Itoh et al. 2005). Therefore, radicle initiation is assumed to occur at the globular stage. The molecular mechanisms of radicle formation in rice and the key genes involved are presumed on the basis of experimental reports in Arabidopsis. Some rice mutants have defects in radicle formation; however, the causative genes have not yet been isolated.

Hong et al. (1995) reported three independent lines of radicleless (ral) mutants. One of them, ral1, is viable, although it has a reduced number of crown and lateral roots (Scarpella et al. 2003). The ral1 plants also have narrower leaves with vascular patterning distortions associated with a defective response to auxin, indicating that RAL1 has auxin-related functions. A mutant of Oryza sativa CONNECTED EMBRYO (OsCEM) produces multiple shoots and radicles (Yang and Hwa 2008). Endogenous indole-3-acetic acid (IAA) level in oscem embryos is lower than that in wild-type embryos despite no differences in vegetative stages. This result also strongly suggests the relationship between auxin and radicle formation. Multiple radicles are sometimes observed in the apical displacement 1 (apd1) mutant, in which the apical shoot region and scutellum are reduced and the basal region of the radicle is enlarged (Kinae et al. 2005). Formation of multiple radicles in apd1 may be a secondary effect of the aberrant apical-basal patterning of the embryo (Kinae et al. 2005).

Our knowledge of radicle formation is limited compared to that of other types of rice roots because there are few reports of radicle-related genes. Recently, microarray analysis of spatiotemporal gene expression patterns during early embryogenesis was performed (Itoh et al. 2016). Based on the datasets from that study, we listed putative radicle-related genes and assessed their functions by performing in situ hybridization (Fig. 14.3). The list contained 23 genes (including 14 encoding transcription factors), which were classified into three groups according to their expression patterns (Fig. 14.3): genes with radicle-specific expression from radicle initiation to its development (Group I, 14 genes); genes with radicle-specific expression during radicle development (Group II, 4 genes); and genes not showing radicle-specific expression (Group III, 5 genes). We then focused on the transcription factor genes in Group I (Fig. 14.3a). Four genes for APETALA 2 (AP2)/ETHYLENE RESPONSIVE FACTOR (ERF) transcription factors with two AP2 repeats (Os06g0657500, Os02g0614300, Os04g0504500, and Os01g0899800) were expressed in the center of the 3-DAP embryo where a radicle will be formed (Fig. 14.3a). They share sequence similarities with PLETHORA (PLT) homologs, the key factors in root meristem formation and maintenance (Aida et al. 2004; Galinha et al. 2007). They are assumed to have roles in radicle formation, especially meristem initiation and development because PLT protein dosage in RAM, which is determined by auxin, is translated into distinct cellular responses: high levels of PLT promote stem cell identity and maintenance, whereas low levels enhance cell division and differentiation (Galinha et al. 2007). Two NAC transcription factors, Os06g0530400 (OsNAC7) and Os02g0745300, share sequence similarities with SOMBRERO (SMB) and FEZ, which are involved in root cap development in Arabidopsis (Willemsen et al. 2008). Both Os06g0530400 and Os02g0745300 showed a root cap-like pattern in the basal region of the embryo (Fig. 14.3a), suggesting that both genes may be involved in root cap development in radicle formation.

Fig. 14.3
figure 3

Expression patterns of candidate genes essential for radicle formation in rice. (a) Group I: genes showing radicle-specific expression from radicle initiation to development. (b) Group II: genes showing radicle-specific expression during radicle development. (c) Group III: genes not showing radicle-specific expression. For each gene, a longitudinal section through a 3-DAP (days after pollination) to early 4-DAP embryo is shown in the top row and that through a late 4-DAP to 5-DAP embryo is shown in the bottom row. Signals were detected with DIG-NBT/BCIP. Bars = 100 μm. Of the 23 genes shown, 16 were selected as follows: 1st step, expression in the basal part of a 3-DAP embryo is >10 times that in the apical part according to microarray datasets in Itoh et al. (2016); 2nd step, high expression in the root of a 7-DAP embryo in Itoh’s datasets; 3rd step, high expression in the root in a mature plant and embryo in a seed according to RiceXPro (Sato et al. 2011). Five transcription factor genes that are coexpressed with these 16 genes and are highly expressed in the radicle of a 7-DAP embryo were also chosen. Additionally, two genes (OsSCR1 and QHB) previously reported to be expressed in the endodermis and central cells of QC were analyzed (Kamiya et al. 2003a, b)

Full size image

We also examined the expression patterns of Group I transcription factor genes in crown root primordia to check whether these patterns are similar to those in the radicle. Almost all of the genes examined had similar expression patterns, suggesting that they have the similar function during radicle and crown root formation (Fig. 14.4a). We also performed double-target in situ hybridization in crown root primordia, where the identities of cell layers are more easily distinguishable than in the radicle. First, we used the probe for OsSCR1 and AP2/ERFs. OsSCR1 mRNA was localized in the endodermis, including epidermis-endodermis initial cells, and in central QC cells (Fig. 14.4b). Os06g0657500 was expressed in the inner and outer layers of the endodermis without overlapping with the OsSCR1 signal (Fig. 14.4b). Os02g0614300 was also expressed in the inner and outer layers of the endodermis but overlapped with the OsSCR1 signal in the stem cell region (Fig. 14.4b). The expression patterns of two NACs were slightly different: the Os06g0530400 signal was observed in peripheral root cap cells and columella cells, whereas Os02g0745300 was expressed mainly in columella cells (Fig. 14.4c). Slight differences in the expression regions among the same gene family members might suggest the specialized function of each gene.

Fig. 14.4
figure 4figure 4

Expression patterns of transcription factor genes in crown root primordia of rice. (a) Cross sections through the nodes of 7-day-old plants. Upper sections, initiation to early developing stage of crown root primordia. Middle sections, developing stage of crown root primordia. Lower sections, late developing to emergence stage. Signals were detected using DIG-NBT/BCIP. Bars = 200 μm. (b) Double-target in situ hybridization with probes for OsSCR1 and AP2/ERF transcription factors. The OsSCR1 signal was detected with biotin-TSA-fluorescein (green fluorescence), and Os06g0657500 and Os02g0614300 signals were detected with DIG-Fast Red (red fluorescence). Arrowheads indicate the central cells of QC. (c) Double-target in situ hybridization with the probes for NAC transcription factors. The Os02g0745300 signal was detected with biotin-TSA-fluorescein, and the Os06g0530400 signal was detected with DIG-Fast Red

Full size image

14.2.3 Crown Roots

Monocot plants produce numerous crown roots from nodes, which form a fibrous root system. During crown root morphogenesis, three developmental stages can be clearly distinguished: initiation, development, and growth (Coudert et al. 2010; Itoh et al. 2005; Kitomi et al. 2011; Zhao et al. 2009). The regulation of crown root formation in rice and that of lateral root formation in Arabidopsis share several common characteristics. The molecular mechanism of crown root formation is relatively well analyzed compared with that of other root types (Fig. 14.5).

Fig. 14.5
figure 5

Proposed gene regulatory pathways of crown root formation in rice. *Genes characterized by using reverse genetic approaches (overexpression, knockdown, or both). Genes without asterisks are cloned and characterized by using forward genetic approaches

Full size image

Auxin is essential throughout root morphogenesis in these species, and auxin-related mutations lead to morphological abnormalities in rice crown roots and Arabidopsis lateral roots. Rice YUCCA 1 (OsYUC1) and CONSTITUTIVELY WILTED 1 (OsCOW1) encode flavin monooxygenases, the key enzymes in auxin biosynthesis (Woo et al. 2007; Yamamoto et al. 2007). Overexpression of OsYUC1 enhances crown root formation, whereas OsYUC1 antisense plants show severe growth retardation (Yamamoto et al. 2007). Alleles of OsCOW1 with the insertion of the Tos17 transposon or T-DNA decrease the root-to-shoot ratio by reducing crown and lateral root numbers (Woo et al. 2007). OsCOW1 was also reported as NARROW LEAF 7 (NAL7)/OsYUC8, which was identified in a mutant with narrow leaves (Fujino et al. 2008). The fish bone ( fib) mutant defective in crown and lateral root formation has a mutation in the gene encoding TRYPTOPHAN AMINOTRANSFERASE OF ARABIDOPSIS 1 (OsTAA1), an auxin biosynthetic enzyme (Yoshikawa et al. 2014). Several PIN-FORMED (PIN) genes, which encode auxin efflux carriers mediating polar auxin transport, also play a pivotal role in crown root formation. Downregulation of OsPIN1b and OsPIN3t (termed OsPIN10a in Wang et al. 2009; OsPIN3a in Miyashita et al. 2010), and overexpression of OsPIN2, reduces crown root number (Chen et al. 2012b; Xu et al. 2005; Zhang et al. 2012a). Overexpression of OsPINOID (OsPID), which controls auxin distribution by controlling subcellular localization of PINs, also causes abnormal crown root development (Morita and Kyozuka 2007). Crown root initiation is impaired in crown rootless4 (crl4)/osgnom1 mutant (Kitomi et al. 2008b; Liu et al. 2009). GNOM is a membrane-associated guanine-nucleotide exchange factor for the G protein ADP-ribosylation factor (Arf-GEF) and plays an important role in polar auxin transport by establishing coordinated polar localization of PIN1 in Arabidopsis (Geldner et al. 2003; Steinmann et al. 1999). Distortion of polar auxin transport and altered expression patterns of OsPINs were observed in crl4/osgnom1 mutants, indicating that polar auxin transport is required for crown root initiation in rice.

Besides auxin biosynthesis and polar transport, auxin perception and signal transduction are essential for crown root formation. A rice microRNA, miR393, affects crown root initiation and seminal root development through negative regulation of the homologs of Arabidopsis auxin receptors TRANSPORT INHIBITOR RESPONSE 1 (TIR1) and AUXIN SIGNALING F-BOX 2 (AFB2), OsTIR1, and OsAFB2 (Bian et al. 2012). Auxin signal is transmitted by a pathway mediated by Aux/IAA and AUXIN RESPONSE FACTOR (ARF) (Liscum and Reed 2002). Transgenic plants that produced constitutively active Aux/IAA, which was obtained by mutagenizing a conserved amino acid residue in the degradation-related domain (domain II) of OsIAA3 (OsIAA31 in Jain et al. 2006), have reduced crown root number (Nakamura et al. 2006). The crl1/adventitious rootless1 (arl1) mutants develop few crown roots; CRL1/ARL1 encodes a plant-specific LATERAL ORGAN BOUNDARIES DOMAIN (LBD)/ASYMMETRIC LEAVES2-LIKE (ASL) transcription factor, which acts downstream of the Aux/IAA and ARF-mediated auxin signaling pathway and whose expression is directly regulated by OsARF (Inukai et al. 2005; Liu et al. 2005). CRL6, which encodes a chromodomain helicase DNA-binding (CHD) family protein, is thought to influence crown root initiation and development through the Aux/IAA and ARF-mediated auxin signaling pathway because most of the 31 Aux/IAA genes are downregulated in the crl6 mutant (Wang et al. 2016). The phenotype of the crl2 mutant, impaired root gravitropism and crown root initiation, suggests that CRL2 might also be involved in auxin signaling, although the causal gene has not yet been identified (Inukai et al. 2001; Yamamoto et al. 2010). Most of the auxin-related crown root mutants mentioned above also show defects in lateral root formation and root hair development, indicating the importance of auxin in overall root morphogenesis.

Cytokinin also plays an important role in the regulation of root morphogenesis and is widely known to act antagonistically to auxin: root formation is promoted by auxin but is suppressed by cytokinin. In Arabidopsis, root meristem size is controlled by the balance between cell differentiation and division, which results from antagonistic regulation by auxin and cytokinin (Dello Ioio et al. 2007, 2008). This antagonistic regulation is also important in rice crown root formation. The phenotype of the dominant mutant root enhancer1 (ren1-D), which has an increased crown root number, is caused by the activation of a CYTOKININ OXIDASE/DEHYDROGENASE (CKX) family gene, OsCKX4 (Gao et al. 2014). CKXs are the only enzymes known to catalyze the irreversible degradation of cytokinin (Werner et al. 2003). Rice METALLOTHIONEIN 2b (OsMT2b) also has a role in the development of crown and lateral roots by influencing the endogenous cytokinin level (Yuan et al. 2008). Not only cytokinin content but also cytokinin signaling affects crown root formation. CRL5 encodes an AP2/ERF transcription factor AINTEGUMENTA (ANT), and its expression is induced by OsARFs (Kitomi et al. 2011). Auxin-induced CRL5 upregulates a type-A response regulator gene OsRR1, which suppresses cytokinin signaling and thus promotes crown root initiation. WOX11 activates crown root development by directly repressing OsRR2 (Zhao et al. 2009). Further analysis demonstrated that ERF3 interacts with WOX11 and promotes WOX11 binding to OsRR2 (Zhao et al. 2015a).

Cell division is essential for crown root formation because it contributes to the development of crown root primordia. The crl3 mutant produces a few crown root primordia consisting of vacuolated cells, whereas those in wild type consist of non-vacuolated cells (Kitomi et al. 2008a). Vacuolated cells divide in the early stage of crown root primordia development; however, cell division activity is gradually arrested, and primordia development is stopped in crl3. Overexpression of O. sativa ROOT ARCHITECTURE ASSOCIATED 1 (OsRAA1) increases the number of crown and lateral roots compared with control plants (Ge et al. 2004). OsRAA1 is an anaphase-promoting complex/cyclosome (APC/C)-targeted protein to block the cell cycle at the transition from metaphase to anaphase (Han et al. 2008). A mutation in rice CULLIN-ASSOCIATED AND NEDDYLATION-DISSOCIATED 1 (OsCAND1) causes a defect in the emergence of crown root primordia, although crown root initiation occurs normally (Wang et al. 2011). CAND1 is an SCFTIR1 E3 ubiquitin ligase involved in the degradation of Aux/IAA proteins in response to auxin in Arabidopsis (Chuang et al. 2004; Feng et al. 2004). OsCAND1 is involved in auxin signaling to maintain the G2/M cell cycle transition in the crown root meristem and consequently the emergence of crown roots (Wang et al. 2011).

14.2.4 Lateral Roots

Molecular mechanisms of lateral root formation are similar to that of crown root formation; therefore, most of the crown rootless mutants show lateral rootless phenotype as well. However, the differences between crown root and lateral root obviously exist: the sites of their initiation, the number of inner cell layers, physiological functions, and plasticity in response to environmental stimuli (Luquet et al. 2005; Rebouillat et al. 2009; Suralta et al. 2008). Although most of the mutants lacking crown roots also lack lateral roots, some mutants have abnormalities in lateral root formation without crown root defects. Analysis of such mutants might disclose lateral root-specific factors and schemes.

As mentioned above, auxin is a major player in lateral root formation. T-DNA-insertion mutants of the rice gene AUXIN RESISTANT 1 (OsAUX1), which is evolutionarily close to the members of the auxin influx carrier gene family AUX1/LIKE AUX 1 (LAX), have reduced lateral root number (Zhao et al. 2015b). The double mutant of nal2 and nal3 (nal2/3), which has mutations in two identical OsWOX3A/OsNARROW SHEATH (OsNS) genes located on chromosomes 11 and 12, respectively, produces fewer lateral roots than does the wild type (Cho et al. 2013). Reduced lateral root initiation in nal2/3 seems to be attributable to compromised distribution of endogenous IAA caused by altered expression of OsPIN1 and OsPIN2. Phenotypes of these mutants demonstrate that polar auxin transport mediated by auxin influx and efflux carrier proteins is important for lateral root formation. The phenotypes of some mutants also indicate the importance of auxin signaling mediated by Aux/IAA and ARF in lateral root formation. The gain-of-function mutants osiaa11 and osiaa13, which have stabilizing mutations in domain II of Aux/IAA proteins, have dramatically reduced lateral root number (Kitomi et al. 2012; Zhu et al. 2012). The rice cyclophilin 2 (oscyp2) mutant also shows impaired lateral root initiation (Kang et al. 2013). OsCYP2 is involved in Aux/IAA degradation by stimulating the activity of the SCFTIR ubiquitin E3 ligase complex. Auxin signaling is likely disturbed in these mutants because degradation of Aux/IAA proteins allows auxin-responsive transcription to be regulated by ARF proteins, which then act as transcriptional activators or repressors (Gray et al. 2001). A mutation in rice HEME OXYGENASE (OsHO1), which encodes an enzyme that catalyzes the degradation of heme into biliverdin IXa, Fe2+, and carbon monoxide, also affects lateral root formation in a manner dependent on auxin and stress-related signals (Chen et al. 2012a). Some mutants with auxin-related abnormalities also have defects in lateral root formation; these include lateral rootless 1 (lrt1), lrt2, auxin-resistant mutant 1 (arm1), and arm2 (Chhun et al. 2003; Faiyue et al. 2010; Hao and Ichii 1999; Wang et al. 2006b).

Cell cycle regulation is necessary for lateral root development. The mutant of O. sativa ORIGIN RECOGNITION COMPLEX SUBUNIT 3 (OsORC3) has a temperature-dependent defect in lateral root development (Chen et al. 2013). In OsORC3 knockdown plants, the emergence of lateral root primordia is blocked due to the perturbation of cell cycle-related gene expression in the primordia (Chen et al. 2013).

14.2.5 Root Hairs

Root hairs are long cylindrical outgrowths of individual root epidermal cells and are thus different from seminal, crown, and lateral roots. Root hair development includes three stages: cell fate determination, root hair initiation, and root hair elongation (Huang et al. 2013b). Vesicle trafficking, cytoskeleton reorganization, and cell wall loosening and synthesis are major driving forces for root hair elongation that depend on gene expression promoted by signals such as auxin, cellular pH, calcium ions, extracellular reactive oxygen species (ROS), and phosphatidylinositols (Libault et al. 2010).

A mutation in O. sativa SEC14-NODULIN DOMAIN PROTEIN (OsSNDP1), which encodes a phosphatidylinositol transfer protein, leads to short-branched root hairs (Huang et al. 2013b). Similar defects were reported in Arabidopsis mutants with defects in phospholipid metabolism and signaling (Kusano et al. 2008; Vincent et al. 2005), indicating the critical roles of phospholipids in root hair elongation. ROOT HAIRLESS 1 (RTH1)/O. sativa APYRASE 1 (OsAPY1), which encodes an enzyme that hydrolyzes NTPs and/or diphosphates, also affects root hair elongation (Yuo et al. 2009). Apyrases control the concentration of extracellular ATP, which functions as a signal molecule for growth control and is localized in the regions of active growth and cell expansion such as root hair tips (Roux and Steinebrunner 2007; Wu et al. 2007). Rice FORMIN HOMOLOGY 1 (OsFH1) is important for root hair elongation under submerged conditions (Huang et al. 2013a). Formins play critical roles in cytoskeleton organization by nucleating actin polymerization and elongation and bundling actin filaments, which drive tip growth (Paul and Pollard 2009). OsFH1 is assumed to have similar functions, although no null mutant with a defective root hair phenotype has been reported in Arabidopsis or rice (Deeks et al. 2005; Yi et al. 2005).

Factors involved in cell wall modification are closely related to root hair development. Root hair length is reduced in the rice short root hair 2 (srh2) mutant, which has a mutation in the XYLOGLUCAN XYLOSYLTRANSFERASE 1 (OsXXT1) gene (Wang et al. 2014a). Xyloglucan is not considered to be an important component of cell wall in grasses, including rice, because its content is below 5% (Vogel 2008). However, the srh2 mutant demonstrates the importance of xyloglucan in rice root hair development. Expansins, which are associated with cell wall loosening, permit turgor-driven cell elongation (Cosgrove 2000). Rice EXPANSIN A17 (OsEXPA17) and OsEXPA30 are also involved in root hair elongation because the osexpA17 mutant shows defects in root hair elongation, and these defects are partially complemented by OsEXPA30 (Yu et al. 2011). The promoters of OsEXPA17 and OsEXPA30 contain conserved root hair-specific cis-elements (RHEs), which are also found in root hair-specific genes and genes paralogous to AtEXPA7 (Kim et al. 2006). These EXPAs with RHEs are expressed in a root hair-specific manner in Arabidopsis and rice (Kim et al. 2007; Yu et al. 2011). OsEXPB5, which is so far found in Gramineae family and is absent in dicots, also has RHEs in the promoter region, and its expression is strongly associated with root hair initiation and elongation (Won et al. 2010); however, its in vivo function in root hair development has not been demonstrated. RTH2/O. sativa CELLULOSE SYNTHASE-LIKE D1 (OsCSLD1) is also required for root hair elongation (Kim et al. 2007; Yuo et al. 2011). Only OsCSLD1 is specifically expressed in roots, similar to root hair-specific genes with RHEs, whereas other OsCSLD subfamily members are expressed in both roots and shoots.

The root epidermis comprises hair cells (trichoblasts) and non-hair cells (atrichoblasts). In each plant species, root hair patterning belongs to one of three types according to the way how the fate of each cell is determined (Kim et al. 2006). In Type 1, hair cells can differentiate from any epidermal cell. In Type 2, the root epidermis consists of cells of two sizes, and only the short cells differentiate into hair cells. In Type 3, hair cells produce rows along the longitudinal root axis, resulting in a striped pattern. Root hair patterning in rice is Type 2, and the differences in size between mature hair and non-hair cells result from differential cell expansion relatively late in the development, after initiation of root hair growth (Kawata and Ishihara 1959; Kim and Dolan 2011). A mutation in the O. sativa ROOT HAIRLESS 1 (OsRHL1) gene, which encodes a basic helix-loop-helix (bHLH) transcription factor, results in very short root hairs (Ding et al. 2009). In the osrhl1 mutant, clearly short and long epidermal cells characteristic of Type 2 species are not observed, suggesting that OsRHL1 controls root hair elongation and epidermal cell patterning, similar to the bHLH gene ROOT HAIR DEFECTIVE 6-LIKE 4 (RSL4) in Arabidopsis (Ding et al. 2009; Yi et al. 2010).

14.3 Formation of Root System Architecture

The outline of the rice root system is formed by multiple crown roots developed from several phytomers; a phytomer is a nodal unit consisting of a leaf, an axillary bud, and crown roots (Rebouillat et al. 2009). Crown roots developed from the upper and lower regions of each node are called the upper and lower crown roots, respectively (Fig. 14.2). The lower crown roots have a larger diameter than the upper crown roots (Abe and Morita 1994). Another feature of lower crown roots is downward elongation, whereas upper crown roots elongate randomly in directions ranging from lateral to vertical, suggesting that they respond to gravity more weakly than do the lower crown roots (Abe and Morita 1994). Overall, the growth angle of each upper and lower crown root determines vertical distribution of the whole root system in the soil. Shallow and steep root growth angles favor root distribution in the topsoil and subsoil, respectively. The maximum length of each crown root restricts range of access for absorption of water and nutrients from the soil. Short roots result in compact root systems, whereas long roots produce large root systems.

The genetic mechanism of root system development in rice has been dissected mainly on the basis of QTL analysis; the first such study was reported by Champoux et al. (1995). Hundreds of QTLs with small to intermediate genetic effects on many root parameters that affect root system architecture have been detected in rice; such parameters include the growth angle, length, volume, and thickness of the roots (Rebouillat et al. 2009). However, the genetic mechanisms underlying these QTLs are poorly understood. On the other hand, several genes for root development have been isolated in rice mutants showing abnormal root phenotypes (Rebouillat et al. 2009; Wu and Cheng 2014). In this section, we discuss the genes and QTLs related to quantitative variation of root system architecture in rice (Table 14.1; Fig. 14.6).

Fig. 14.6
figure 6

Location of rice QTLs involved in root system architecture that are potentially useful for molecular breeding. Chromosome numbers are indicated above each linkage map. Black box indicates a region that contains a root QTL

Full size image

14.3.1 Root Growth Angle

Root growth angle is controlled by several environmental factors such as gravity, light, and water potential (Oyanagi et al. 1993; Uga et al. 2015a). Root gravitropism has been well studied in Arabidopsis (Baldwin et al. 2013; Morita 2010), but not in monocot plants including rice. Only two QTLs for the root gravitropic response have been reported in rice (Norton and Price 2009), but the underlying genes have not yet been isolated. DRO1, which was reported originally as a major QTL responsible for root growth angle, is also involved in gravitropism (Uga et al. 2013a). DRO1 is negatively regulated by auxin signaling downstream of Aux/IAA and ARF and is involved in cell elongation in the root tip, which causes gravitropic bending (Uga et al. 2013a). Under normal growth conditions, DRO1 is expressed around the RAM in the root tip and crown root primordia. For the response to gravitropic stimuli (i.e., rotation of the roots from the normal vertical axis to the horizontal axis), DRO1 transcripts in the outer cells of the distal elongation zone are repressed on the lower side than on the upper side of the roots by the redirected auxin flow to the lower side of the root, resulting in decreased cell elongation in the lower side relative to the upper side. This process contributes to asymmetric growth, leading to root gravitropic bending. Thus, QTLs for gravitropism should affect root growth angle, resulting in a natural variation of root system architecture.

Genes with high sequence similarity to DRO1 have been found in other monocots such as maize, sorghum, and barley, but their physiological and molecular functions are still unknown (Uga et al. 2013a). Recently, genes with low sequence similarity to DRO1 have been identified in dicots. The legume Medicago truncatula carrying mutations in NEGATIVE GRAVITROPIC RESPONSE OF ROOTS (NGR) shows a negative root gravitropic response (Ge and Chen 2016). Only triple mutants of three redundant AtNGR genes (At1g17400, At1g72490, At1g19115) in Arabidopsis also showed a similar negative root gravitropic response (Ge and Chen 2016). These NGR genes may be DRO1 homologs in the IGT family, the members of which have relatively low sequence similarity to each other but have conserved amino acid motifs (Guseman et al. 2017). These findings suggest that the functions of DRO1 and DRO1 homologs in root gravitropism are conserved in monocots and dicots. The IGT family also includes TILLER ANGLE CONTROL 1 (TAC1) and LAZY1, which control the branching angle of lateral shoot organs in both monocots and dicots (Guseman et al. 2017), suggesting that this gene family might be associated with the regulation of growth angle in shoot and root organs.

Many other QTLs for root growth angle have been reported in rice (Kitomi et al. 2015; Lou et al. 2015; Uga et al. 2012, 2013b, 2015b). DRO2 (Uga et al. 2013b), DRO3 (Uga et al. 2015b), DRO4 (Kitomi et al. 2015), and DRO5 (Kitomi et al. 2015) were detected in seven F2 mapping populations derived from a cross between several rice accessions with different root growth angles and “Kinandang Patong” as a donor line with a large root growth angle. Therefore, QTLs associated with root growth angle distinct from DRO1 exist in “Kinandang Patong.” qSOR1 (quantitative trait locus for SOIL SURFACE ROOTING 1) has been detected on chromosome 7 in recombinant inbred lines derived from a cross between “Gemdjah Beton,” a lowland rice accession with a high proportion of crown roots that run along or near the soil surface, and “Sasanishiki,” a lowland rice accession that does not form soil-surface roots (Uga et al. 2012). The “Gemdjah Beton” allele of qSOR1 causes many thick crown roots to elongate near the soil surface from the seedling stage. This phenotype is very unique because thick crown roots generally elongate downward. qSOR1 was fine-mapped to a 812-kb candidate region on chromosome 7 (Uga et al. 2012; Fig. 14.6). Lou et al. (2015) also reported QTLs for root growth angle on chromosomes 1, 2, 4, 7, and 10. Among them, three QTLs on chromosomes 2, 4, and 7 are located near the regions of DRO4, DRO2, and qSOR1, respectively. Cloning of these QTLs would deepen our understanding of the genetic mechanisms that determine root growth angle in rice.

14.3.2 Root Length

Maximal root length is determined by the rate and duration of root elongation. Root elongation is caused by cell division and elongation. Mutant analyses revealed that genes related to cell wall growth, cell expansion, and auxin signaling are involved in the division and elongation of root cells. Cell wall development affects root elongation and the maintenance of root structure. OsDGL1 encodes the dolichyl-diphosphooligosaccharide-protein glycosyltransferase 48-kDa subunit precursor (Qin et al. 2013). An ethyl methanesulfonate (EMS)-induced osdgl1 mutant has a defect in N-glycosylation, an altered composition of matrix polysaccharides in the cell wall, and cell death in the root, resulting in a decrease in root elongation without a decrease in the numbers of crown roots, lateral roots, or root hairs. OsMOGS encodes a putative mannosyl-oligosaccharide glucosidase and acts downstream of OsDGL1 during N-glycan processing in the endoplasmic reticulum (Wang et al. 2014b). An EMS-induced osmogs mutant has a decreased cell division and elongation in the root, resulting in short roots. OsMOGS is needed for cellulose biosynthesis and OsABCB-mediated auxin transport in rice (Wang et al. 2014b). Other genes associated with cell wall modification also control root elongation. ROOT GROWTH INHIBITING (RT)/OsGLU3, which encodes a membrane-anchored endo-1,4-β-d-glucanase, is involved in cell wall loosening necessary for root cell elongation (Inukai et al. 2012; Zhang et al. 2012b). The rt./osglu3 mutants have short-root phenotypes due to a decrease in longitudinal cell elongation without changes in root differentiation, root cell division, or shoot development. Interestingly, cellulose content in roots is increased in an ethylene imine-induced rt. mutant (Inukai et al. 2012) but is decreased in EMS-induced osglu3 mutants (Zhang et al. 2012b). To reconcile these contradictory findings, further studies are needed. OsGNA1 encodes glucosamine-6-P acetyltransferase, which is involved in de novo UDP-N-acetylglucosamine biosynthesis (Jiang et al. 2005). A T-DNA insertion osgna1 mutant has decreased root cell elongation caused by cell shrinkage, perhaps because of insufficient UDP-GlcNAc for protein N-glycosylation, which is necessary for plant development including cell wall synthesis (Lerouge et al. 1998; Lukowitz et al. 2001). Cell expansion occurs in turgor-driven cell elongation. Transgenic plants overexpressing OsEXPA8, which encodes a root-specific α-expansin (Shin et al. 2005), have increased seminal, crown, and lateral root length as well as plant height and increased leaf number and size caused by an increase in cell length in both shoot and root vascular bundles (Ma et al. 2013).

Auxin regulates cell fate determination and cell elongation (Tanaka et al. 2006). These effects are mostly mediated by ARFs. Loss-of-function Tos17 and T-DNA insertion mutants of osarf12, which is a member of ARFs (Wang et al. 2007), have short-root phenotypes due to a smaller elongation zone in seminal roots compared to the wild type (Qi et al. 2012). The short elongation zone is likely caused by a low auxin concentration. O. sativa SHORT POSTEMBRYONIC ROOTS 1 (OsSPR1) encodes a putative mitochondrial protein with an Armadillo-like repeat domain (Jia et al. 2011). EMS-induced osspr1 mutants have short-root phenotypes (decreased lengths of seminal, crown, and lateral roots) due to reduced cell elongation, whereas lateral root initiation and lateral root number are similar to those in the wild type. OsCYT-INV1/OsNIN8 encodes alkaline/neutral invertase and is homologous to AtCYT-INV1 in Arabidopsis (Ji et al. 2005, Jia et al. 2008). An EMS-induced Oscyt-inv1 mutant has a short-root phenotype due to a decreased cell length probably caused by hexose deficiency, as hexoses play various roles in cell elongation.

Despite isolation of several genes, our knowledge of the genetic mechanism controlling root length in rice is still limited compared to that for Arabidopsis. Reverse genetics should be a valuable approach to gain a better understanding of the gene network that regulates root length. Antisense transgenic plants with downregulated OsCKI1, which encodes putative casein kinase I, have short seminal roots and a low number of crown and lateral roots caused by reduced cell elongation (Liu et al. 2003). Examination of transgenic rice plants over- and under-expressing OsRPK1, which encodes an LRR-RLK, revealed that this gene affects seminal root length and crown root number by negatively regulating polar auxin transport (Zou et al. 2014).

Several QTLs for root length have been fine-mapped in rice, although none of them have been cloned (Fig. 14.6). qRL6.1, a QTL controlling root length at the seedling stage under hydroponic conditions, was mapped to a 337-kb interval on chromosome 6 (Obara et al. 2010). qRL7, a QTL affecting root length at the heading stage under hydroponic conditions, was mapped to a 657-kb interval on chromosome 7 (Wang et al. 2013). Recently, QUICK ROOTING 1 (QRO1) and QRO2 have been fine-mapped on chromosomes 2 and 6, respectively, in chromosome segment substitution lines derived from a cross between “IR64” and “Kinandang Patong” and grown under hydroponic conditions (Kitomi et al. in press, Fig. 14.6).

14.3.3 Other Root Traits

The combination of growth angle and length in seminal and crown roots is the main determinant of root system architecture in cereals (Abe and Morita 1994; Araki et al. 2002), although other root traits such as volume and thickness are also important.

Root volume affects root surface area and thus absorption of water and nutrients from soil (Gowda et al. 2011; Wang et al. 2006a), but rice genes that control root volume have not yet been isolated. qFSR4, a QTL for root volume per tiller, has been fine-mapped on chromosome 4 (Ding et al. 2011). The 38-kb qFSR4 candidate region has three open reading frames including NAL1 (Qi et al. 2008). The NAL1 gene is associated with polar auxin transport and controls leaf width. qFSR4 also affects flag leaf width. NAL1 may be the most promising candidate gene for qFSR4 because polar auxin transport affects root development and shoot growth.

Root thickness affects uptake of water and nutrients as well as root penetration ability (Gowda et al. 2011; Wang et al. 2006a). qRT9, a QTL for root thickness, has been fine-mapped to an 11.5-kb candidate region on chromosome 9 (Li et al. 2015) with only one annotated open reading frame, Os09g0455300, which encodes a putative bHLH transcription factor (OsbHLH120). Haplotype and expression analyses suggest that OsbHLH120 is the candidate gene for qRT9. For water and nutrient translocation, stele and xylem structures should be more important than root thickness (Uga et al. 2008). STELE TRANSVERSAL AREA 1 (STA1), a QTL controlling stele transversal area, has been fine-mapped to a 359-kb interval between SSR markers RM566 (14.70 Mb) and RM24334 (15.06 Mb) on chromosome 9 (Uga et al. 2008, 2010). qRT9 (17.13 Mb) and STA1 are located near DRO1 (16.31 Mb). As mentioned above, upland rice tends to have deeper and thicker roots than those of lowland rice (O’Toole and Bland 1987). The tight linkage of these QTLs should be associated with the phenotypic relationship between these root traits.

14.4 Conclusions

Several rice-specific genes controlling the root system have been found. However, the many rice genes homologous to Arabidopsis genes associated with the formation of the main root system are also involved in the formation of the fibrous root system. Thus, many parts of the genetic mechanism related to the root system have features common between monocots and dicots. On the other hand, much remains to be clarified about the difference between monocots and dicots in the natural variation of root system architecture because most related genes have not been cloned. Because dicots have DRO1 homologs, their genetic mechanism related to the natural variation in root system architecture might have many common features with that of monocots. Recent progress of forward and reverse genetic strategies, such as MutMap (Abe et al. 2012; Takagi et al. 2015) and TILLING (Suzuki et al. 2008; Till et al. 2007), and the CRISPR/Cas9-mediated genome editing system (Doudna and Charpentier 2014; Schaeffer and Nakata 2015) allow us to isolate genes and find new alleles easily. Using QTL cloning and these approaches, we would be able to elucidate the entire genetic network related to natural variation in rice root system architecture. Understanding of the genetic mechanism of root plasticity in response to environmental variation is also important for improving crop production under abiotic stresses, but it is difficult to obtain reliable phenotypic data and identify related genes or QTLs under field conditions. To resolve this issue, a reproducible root phenotyping platform with controlled soil water, nutrients, and temperature in which we can evaluate accurately a large number of plants is needed.