Genome-Wide Transcript Profiling Reveals an Auxin-Responsive Transcription Factor, OsAP2/ERF-40, Promoting Rice Adventitious Root Development

Abstract

Unlike dicots, the robust root system in grass species largely originates from stem base during postembryonic development. The mechanisms by which plant hormone signaling pathways control the architecture of adventitious root remain largely unknown. Here, we studied the modulations in global genes activity in developing rice adventitious root by genome-wide RNA sequencing in response to external auxin and cytokinin signaling cues. We further analyzed spatiotemporal regulations of key developmental regulators emerged from our global transcriptome analysis. Interestingly, some of the key cell fate determinants such as homeodomain transcription factor (TF), OsHOX12, no apical meristem protein, OsNAC39, APETALA2/ethylene response factor, OsAP2/ERF-40 and WUSCHEL-related homeobox, OsWOX6.1 and OsWOX6.2, specifically expressed in adventitious root primordia. Functional analysis of one of these regulators, an auxin-induced TF containing AP2/ERF domain, OsAP2/ERF-40, demonstrates its sufficiency to confer the adventitious root fate. The ability to trigger the root developmental program is largely attributed to OsAP2/ERF-40-mediated dose-dependent transcriptional activation of genes that can facilitate generating effective auxin response, and OsERF3–OsWOX11–OsRR2 pathway. Our studies reveal gene regulatory network operating in response to hormone signaling pathways and identify a novel TF regulating adventitious root developmental program, a key agronomically important quantitative trait, upstream of OsERF3–OsWOX11–OsRR2 pathway.

Introduction

The root system in plants is highly diversified and is required for several essential functions, such as anchorage, absorption of water and nutrients from the soil, storage of food materials and interaction between plants and soil microbes. Rice (Oryza sativa) adventitious roots [also called crown roots (CRs)] form a major functional component of the mature root system (Itoh et al. 2005, Rebouillat et al. 2009, Coudert et al. 2010, Orman-Ligeza et al. 2013). Arabidopsis lateral roots (LRs) develop from the xylem pole pericycle cells of the primary root (PR) at a regular interval (Péret et al. 2009, Lavenus et al. 2013); however, the origin for root branches in rice is diverged. The LRs in rice originate from endodermal and pericycle cells located opposite to protophloem, whereas the innermost ground meristem cells adjacent to vascular cylinder develop CRs at the stem base (Itoh et al. 2005, Rebouillat et al. 2009, Orman-Ligeza et al. 2013, Bellini et al. 2014, Mai et al. 2014). Though different root types in cereals display a gross morphological similarities, the regulatory developmental pathways are partially diverged and the double mutants of radicleless1 (ral1) and crown rootless5 (crl5) display an additive rootless phenotype (Kitomi et al. 2011b, Orman-Ligeza et al. 2013).

Transcription factors (TFs) and phytohormones are key regulators of various developmental program, including root architecture (Lavenus et al. 2013, Mai et al. 2014). TFs, such as ADVENTITIOUS ROOTLESS 1 (ARL1)/CROWN ROOTLESS 1 (OsCRL1), CROWN ROOTLESS 5 (OsCRL5), AP2/ETHYLENE-RESPONSIVE FACTOR3 (OsERF3) and WUSCHEL-Related Homeobox 11 (OsWOX11), CYTOKININ RESPONSE REGULATOR 2 (OsRR2) and QUIESCENT-CENTER-SPECIFIC HOMEOBOX (OsQHB), have been shown to play a crucial role during CR development in rice (Inukai et al. 2001, Kamiya et al. 2003b, Inukai et al. 2005, Liu et al. 2005, Zhao et al. 2009, Kitomi et al. 2011a,, Zhao et al. 2015). The entire process of LRs specification and emergence in Arabidopsis is regulated by different auxin modules that control the expression of many downstream TFs (De Rybel et al. 2010, Yadav et al. 2010, Lavenus et al. 2013). The expression of TFs known to control CR development in rice is regulated by auxin and/or cytokinin signaling pathways. CROWN-ROOTLESS4/OsGNOM1 (OsCRL4/OsGNOM1) encodes an ortholog of Arabidopsis GNOM1 that controls PINFORMED1 (PIN1)-mediated polar auxin transport (PAT), and crl4/osgnom1 mutants are defective in CR formation (Steinmann et al. 1999, Kitomi et al. 2008, Liu et al. 2009). In addition, auxin negatively regulates the expression of rice DEEPER ROOTING 1 (OsDRO1), a regulator of rice root growth angle resulting increased drought avoidance and crop yield (Uga et al. 2013). Furthermore, the cross talk between auxin and cytokinin is known during CR development in rice. Cytokinin Oxidase/Dehydrogenase 4 (OsCKX4) encoding a cytokinin-degrading enzyme integrates auxin and cytokinin signaling pathways in CRs (Gao et al. 2014). Auxin and cytokinin directly regulate expression of OsCKX4 in rice (Gao et al. 2014). Similarly, auxin-inducible genes such as OsCRL5, OsWOX11 and OsERF3 regulate the expression of various auxin and cytokinin responsive genes (Zhao et al. 2009, Kitomi et al. 2011a, Zhao et al. 2015). Although these studies have begun to provide insights into the making of rice root system, an interconnected gene network controlling rice root architecture yet remains obscure.

In this study, we have identified global genes regulated by auxin and cytokinin in rice crown tissues. Our transcriptome data have revealed TFs which are commonly and specifically regulated by auxin and cytokinin signaling pathways. RNA–RNA in situ hybridization shows that the expression of four TFs, OsAP2/ERF-40, OsHOX12, OsNAC39 and both splice variants of OsWOX6 is spatially restricted to developing CR primordia. Furthermore, functional studies of an auxin-responsive AP2-domain containing TF reveal that OsAP2/ERF-40 is sufficient to initiate root-specific developmental program from aerial node. Interestingly, it regulates auxin and OsERF3–OsWOX11–OsRR2 signaling pathways in a dose-dependent manner in the shoot tissues. Taken together, our studies not only identify the putative transcriptional regulators downstream of auxin and cytokinin signaling pathways but also demonstrate the unknown function of an auxin-responsive target and gene regulatory network in activating root-specific developmental program.

Results

Effects of auxin and cytokinin on CR primordia initiation and growth

Antagonistic interaction between auxin and cytokinin is well known for controlling root formation (Dello Ioio et al. 2008, Moubayidin et al. 2009). To study how external application of auxin and cytokinin affects CR initiation and growth in rice, wild-type seeds were grown in the presence of auxin (IAA) and cytokinin (6-BAP). The number of emerged CRs was counted on fifth and sixth day and length of primary and CRs on sixth day after germination. We observed that the number of CRs was increased upon exogenous auxin treatment, whereas it was reduced after cytokinin treatment (Fig. 1A, B). These observations suggest that auxin signaling is required to induce CR development but cytokinin inhibits its formation. In contrast to the opposite effects of auxin and cytokinin on CR number, we have observed a similar effect of these hormones on growth of the roots. Growth of both PR and CR was strongly reduced upon auxin and cytokinin treatments (Fig. 1C). The effects of these hormones on root initiation and growth are concentration-dependent, the 10 µM concentration showed severe effects compared with 1 µM. These data suggest strikingly opposite effects of auxin and cytokinin on CR initiation but similar effects on their growth.

Fig. 1

Effects of IAA and BAP on root number and growth. (A) Six-day-old rice seedling grown on 1/2 MS media supplemented with DMSO (mock), IAA (10 µM) and BAP (10 µM). The primary and LRs were removed from mock- and IAA-treated seedlings but present in BAP-treated plant. (B) Number of CRs on fifth and sixth day after germination from plants treated with two concentrations of IAA and BAP. Data are given as means ±SD (n = 30). (C) Growth of PR and CRs on sixth day after germination on IAA and BAP media. For PR, average length of PRs is plotted (n = 10). For CR data, average length of all CRs emerged from 10 plants is plotted (mock n = 44, 1 µM IAA n = 54, 10 µM IAA n = 66, 1 µM BAP n = 37, 10 µM BAP n = 3). (D–F) Cross-section at stem base of 6-day-old rice treated with mock (D), 10 µM IAA (E) and 10 µM BAP (F) to show CR primordia (red arrow). No CR primordium is seen upon BAP treatment. Significance is determined by the two-tailed Student’s t-test, ns, not significant, *P < 0.05, **P < 0.005. Bars: (A) 1 cm, (D–F) 100 µm.

Next, we analyzed the origin of the extra CRs upon auxin treatment by making 8-µm histological cross-sections of crown tissue. All auxin-induced CR primordia were initiated from the innermost ground tissues (Fig. 1D, E) suggesting that these tissues are specifically responsive to auxin signaling in initiating CR-specific developmental program. Reversibly, CRs are strongly inhibited upon exogenous cytokinin treatment, thus we studied whether lack of CR is because of the defect in CR primordia specification or their emergence. Cross-sections of 6-day-old crown tissue showed a complete absence of CR primordia in BAP-treated plants (Fig. 1F), revealing that cytokinin signaling is inhibitory for CR primordia establishment during very early stage of CR formation. These observations suggest that although both auxin and cytokinin signaling pathways have opposite roles during early stage of CR primordia establishment both signaling pathways are inhibitory during growth of emerged roots.

The global architecture of gene regulation by auxin and cytokinin in rice crown tissues

To investigate the molecular mechanism of hormonal regulation of CR development, we studied the modulation in the global architecture of gene expression by analyzing the genes whose transcript levels were deregulated upon auxin and cytokinin treatments in the crown tissues. We treated 6-day-old wild-type rice seedlings with DMSO (mock), IAA and BAP. The total RNA was extracted from the stem base (crown tissue) and was subjected to RNA sequencing. Differential gene expression analysis was performed using Cuffdiff to identify genes with significant differential expression (Fig. 2A). We observed that the expression level of 1,666 genes was induced and a total of 539 genes were reduced upon IAA treatment (Fig. 2B;Supplementary Tables S1, S2). Similarly, 1,569 genes were found to be upregulated and 850 genes were downregulated by BAP in rice crown tissue (Fig. 2B;Supplementary Tables S3, S4). Of these deregulated genes, a total of 644 genes (467 upregulated and 177 downregulated) were specifically regulated by IAA and 858 genes (376 upregulated and 482 downregulated) were BAP-specific (Fig. 2B;Supplementary Tables S4–S8). Consistent with the activating effects of auxin and inhibitory effects of cytokinin on CR initiation, we observed that a larger number of IAA-specific genes were induced, whereas the pattern was opposite for BAP-specific genes. Despite antagonistic effects of auxin and cytokinin during CR primordia specification, only eight genes showed opposite expression pattern upon IAA and BAP treatment (Fig. 2B;Supplementary Table S9), suggesting that both signaling pathways might be functioning largely independently during primordia establishment. A large number of genes were regulated by both IAA and BAP in similar pattern consistent with what was observed for some regulators of CRs, such as OsWOX11 and OsERF3 (Zhao et al. 2009, Zhao et al. 2015).

Fig. 2

Global analysis of genes regulated by IAA and BAP in rice crown tissues. (A) Heat-map for deregulated genes upon IAA and BAP treatment as compared with mock treatment (log₂ fold change ≥1.0, P-value <0.05 for both upregulated and downregulated). (B) Venn diagram showing common and unique genes regulated by IAA and BAP with at least log₂ fold change ≥1, P-value <0.05 (for both upregulated and downregulated). (C) Venn diagram showing common and unique TFs regulated by IAA and/or BAP (log₂ fold change ≥0.75, P-value <0.05 for both upregulated and downregulated). (D) GO enrichment for differentially expressed transcripts upon IAA and BAP treatment. Node size is proportional to the number of transcripts in that category.

Next, to gain deeper insights into the involvement of IAA- and BAP-regulated genes in various biological processes, we performed gene ontology (GO) enrichment analysis of differentially expressed transcripts upon IAA and BAP treatments. We observed that biological process GO terms related to organ development, hormonal transport and gene regulatory functions were specifically enriched in IAA-induced genes whereas IAA-repressed genes were mostly associated with transport, cell wall biogenesis, metabolic processes and responses to various stresses (Fig. 2D;Supplementary Fig. S1). This suggests that IAA-induced genes are involved in primary processes associated with organ development whereas IAA-repressed genes were more likely regulating secondary processes. On the other hand, biological processes involved in organogenesis are not enriched in genes specifically regulated by BAP (Fig. 2D;Supplementary Fig. S1), indicating that BAP does not regulate early/primary processes of root development to a larger extent.

Transcriptional regulation of auxin and cytokinin signaling pathways

TFs are the master regulators of developmental programs, and members from AP2/ERF (e.g. PLETHORA), homeobox (e.g. WOX) and GRAS gene family (e.g. SHORT-ROOT and SCARECROW) are known to regulate root meristem function and tissue patterning (Helariutta et al. 2000, Kamiya et al. 2003a, Cui et al. 2007, Zhao et al. 2009, Kitomi et al. 2011a, Horstman et al. 2014,, Zhao et al. 2015, Dolzblasz et al. 2016, Henry et al. 2017). Our analysis showed that the expression level of a significant fraction of TFs is regulated by auxin and cytokinin hormones. A total of 335 TFs were deregulated (264 up and 71 downregulated) upon IAA treatment. However, the expression of 371 TFs was under control of the BAP-mediated cytokinin signaling pathway, of which 250 were induced and 121 were repressed by BAP in rice stem base (Fig. 2C;Supplementary Fig. S2A, B, Table S10). Importantly, 65 deregulated TFs were putative downstream targets of auxin (51 induced and 14 repressed) specifically and 101 TFs are specifically regulated (37 induced and 64 repressed) by cytokinin signaling pathways in crown tissues (Fig. 2C;Supplementary Table S10).

We categorized deregulated TFs based on their predicted gene families (Supplementary Table S10). Interestingly, members of AP2/ERF, Aux/IAA, C2H2, C2H2-DOF, GRAS, Homeobox (HB) and MADS-box families, which are known regulators of plant development, were mostly induced by both auxin and cytokinin treatments and very few were repressed (Supplementary Table S10). Of these, AP2/ERF, DOF and GRAS families represent plant-specific TFs. Seven AP2-domain containing TFs (LOC_Os03g08500, LOC_Os09g28440, LOC_Os07g12510, LOC_Os09g35020, LOC_Os03g08460, LOC_Os01g59780, LOC_Os02g43820), two DOF-domain TFs (LOC_Os02g47810, LOC_Os03g42200) and five GRAS genes (LOC_Os07g36170, LOC_Os01g65900, LOC_Os01g62460, LOC_Os12g04200, LOC_Os11g04570) were specifically induced by IAA signaling pathway (Supplementary Table S10), indicating that they might have roles in activating CR-specific developmental program downstream of auxin signaling pathway. These data also suggested that auxin is activating expression of significant number of plant-specific transcriptional pathways.

Validation of deregulation and organ-specific expression pattern of genes regulated by auxin and cytokinin

To validate deregulation of putative downstream targets of auxin and cytokinin signaling, we selected few genes from different functional classes (Table 1). These genes mainly have predicted functions as TFs or hormonal/cell signaling pathways. For TFs, we selected two members of AP2/ERF, four members of homeobox, one each from NAC, bHLH and MYB classes. In rice, members of AP2/ERF (e.g. OsCRL5 and OsERF3) and homeobox domain containing TFs (e.g. OsWOX11) are known to regulate adventitious/CR development (Zhao et al. 2009, Kitomi et al. 2011a, Zhao et al. 2015). For hormonal signaling genes, one member from GA2Ox family, two CKX and one GH3 were selected. Our quantitative real-time PCR (qRT-PCR) analysis reveals that expression of 12 genes was induced and 3 genes were repressed upon IAA treatment to a similar extent as seen in RNA-Seq data (Fig. 3A). Similar co-relation was also observed between RNA-Seq and qRT-PCR data for six activated and four repressed genes upon BAP treatment (Fig. 3B).

Table 1

A list of genes selected for validation of their deregulation

Gene name	Gene ID	Annotation	Arabidopsis hits
OsAP2/ERF-27	LOC_Os03g08460	AP2 domain containing protein, expressed	AT1G72360
OsAP2/ERF-40	LOC_Os04g46400	AP2 domain containing protein, expressed	AT2G44940
OsHOX12	LOC_Os03g10210	Homeobox domain containing protein, expressed	AT4G36740
OsWOX10	LOC_Os08g14400	Homeobox domain containing protein, expressed	AT5G17810
OsWOX11	LOC_Os07g48560	Homeobox domain containing protein, expressed	AT3G03660
OsWOX6	LOC_Os03g20910	Homeobox domain containing protein, expressed	AT5G17810
OsNAC39	LOC_Os03g21030	No apical meristem protein, putative, expressed	AT5G18270
OsbHLH8	LOC_Os01g11910	Basic helix–loop–helix, putative, expressed	AT3G25710
OsMYBAS1	LOC_Os11g47460	MYB family TF, putative, expressed	AT5G59780
OsEF Hand	LOC_Os09g31000	EF hand family protein, expressed	AT2G44310
OsGA2Ox8	LOC_Os05g48700	Gibberellin 2-beta-dioxygenase, putative, expressed	AT1G30040
OsCKX5	LOC_Os01g56810	Cytokinin dehydrogenase precursor, putative, expressed	AT1G75450
OsCKX4	LOC_Os01g71310	Cytokinin dehydrogenase precursor, putative, expressed	AT2G41510
OsGH3.8	LOC_Os07g40290	OsGH3.8—Probable indole-3-acetic acid-amido synthetase, expressed	AT4G37390
OsTIP2-1	LOC_Os02g44080	Aquaporin protein, putative, expressed	AT5G47450
OsDRO1	LOC_Os09g26840	Expressed protein	AT1G72490

Fig. 3

Validation of IAA- and BAP-regulated genes. (A, B) Deregulated expression validation of selected putative targets of auxin (A) and cytokinin (B) signaling pathways by qRT-PCR in two biological replicates. The log₂ fold change data of qRT-PCR are compared with their deregulation level in RNA sequencing data. (C) qRT-PCR for expression pattern analysis of selected genes in emerged CR, stem base and PR. Data from two independent biological replicates are plotted. The log₂ fold change in (A, B) and relative expression in (C) were calculated from three technical replicates for each biological replicate and plotted with ±SD.

Next, we focused our study on TFs and analyzed the organ-specific expression of six putative TFs from different families with induced expression and one TF whose expression was repressed. Their expression was quantified in two biological replicates of emerged CR, stem base (crown tissue), and PR of 6-day-old seedling via qRT-PCR analysis. All of these genes were broadly expressed in these organs including stem base (Fig. 3C). However, the expression level of OsAP2/ERF-40 and OsWOX11 in the stem base was higher than other TFs (Fig. 3C). All these analyses together validate the deregulation of selected genes upon auxin and cytokinin treatment and show that these genes are expressed in the crown tissues, competent for CR formation.

Spatiotemporal expression patterns of TFs during CR development

As the expression of selected TFs was lower in the stem base as compared with emerged CRs and PRs, we surmise that their expression is restricted to only CR primordia in the stem base. To explore that, we studied tissue-specific expression pattern of 4 TFs (OsAP2/ERF-40, OsHOX12, OsNAC39 and both splice variants of OsWOX6) regulated by auxin and/or cytokinin in developing CR primordia using RNA–RNA in situ hybridization. The expression of OsAP2/ERF-40 and OsNAC39 is also regulated by OsCRL1, a known regulator of CR development (Coudert et al. 2015). OsHOX12 is a member of HD-Zip I family and members of this family play important roles during abiotic responses (Agalou et al. 2008, Zhao et al. 2014). OsWOX6 is a member of intermediate clade of WOX gene family and regulates rice tiller angles redundantly with OsWOX11 (Lian et al. 2014, Zhang et al. 2018). Its Arabidopsis homolog genes, AtWOX11 and AtWOX12 are involved in de novo root primordia initiation and organogenesis (Liu et al. 2014, Hu and Xu 2016).

DIG-labeled anti-sense RNA probes were hybridized on 8 µm cross-sections of wild-type rice stem base containing CR primordia at different stages. We observed that all these genes were specifically expressed in developing CR primordia and expression was not detected above background levels in other tissues (Fig. 4). The expression is initiated in the early CR primordia and is continued during later stages of CR differentiation and emergence. OsAP2/ERF-40 is expressed both early as well as in late stage CR primordia (Fig. 4A, B). OsHOX12 expression is uniform in the CR primordia during early stages but the expression is low at the base of the primordia in the later stages (Fig. 4C, D). The expression of OsNAC39 starts at very early stage of CR primordia specification (Fig. 4E, F). There was no significant difference in the expression pattern between two splice variants of OsWOX6 with splice variant-specific anti-sense probes (Fig. 4G, H;Supplementary Fig. S3A). Both variants are expressed at early as well as later stage primordia (Fig. 4G, H;Supplementary Fig. S3A). As control, cross-sections were hybridized with sense probes for all these genes and none showed any signal above the background levels (Supplementary Fig. S3B–E). This study confirms that the expression of these TFs is confined to developing CR primordia and suggests a strict necessity of their spatial regulation during CR development.

Fig. 4

Spatial expression pattern of selected TFs in rice stem base. (A, B) Tissue-specific expression pattern of OsAP2/ERF-40 during CR development, hybridized with anti-sense DIG-RNA probe. (C, D) Expression pattern of OsHOX12 in crown tissues. (E, F) Temporal and spatial expression pattern of OsNAC39 in rice stem base. (G, H) Tissue-specific transcript distribution of OsWOX6-2 in CR primordia. Arrow heads mark developing CR primordia and the expression is CR primordia-specific. Bars: (A–H) 20 µm.

Functional study of an auxin-responsive gene, OsAP2/ERF-40 in transgenic rice

AINTEGUMENTA-like (AIL) genes of AP2/ERF gene family that include double AP2-domain containing AINTEGUMENTA (ANT), BABY BOOM (BBM) and PLETHORA (PLT) genes, are key developmental regulators of plant regeneration, embryogenesis, root meristem establishment and maintenance, shoot apical meristem function, floral organ patterning and LR outgrowth in Arabidopsis (Aida et al. 2004, Galinha et al. 2007, Mudunkothge and Krizek 2012, Hofhuis et al. 2013, Horstman et al. 2014, Kareem et al. 2015, Du and Scheres 2017). The expression of PLT genes is regulated by auxin and they are functioning in genetically redundant manner (Krizek BA 2011, Pinon et al. 2013, Mähönen et al. 2014, Santuari et al. 2016). Rice PLT member, OsPLT8 regulates CR development (Kitomi et al. 2011a, Li and Xue 2011). Apart from the PLT genes, other members of the AP2/ERF gene family also regulate CR development in rice. For example, an auxin-inducible gene with single AP2-domain of AP2/ERF gene family, OsAP2-ERF-4 (also called OsERF3), functions during early and late stages of rice CR development (Rashid et al. 2012, Zhao et al. 2015).

To further investigate the role of auxin-regulated genes during root development, we selected yet another auxin-responsive TF of AP2/ERF gene family, OsAP2/ERF-40 (also called OsTOC168, LOC_Os04g46400) for functional study. OsAP2/ERF-40 is an intron-less gene encoding a single AP2-domain containing protein of CRT/DREB sub-family (Rashid et al. 2012). Its expression is induced by auxin in the rice stem base (Fig. 3A). We, first studied detailed temporal and spatial expression pattern of OsAP2/ERF-40 across various developmental stages of CR primordia and in emerged CRs (Fig. 5A–E). Prior to CR primordia establishment, its expression is low in the tissues peripheral to vascular cylinder (Fig. 5A). During later stages of development, its expression is specifically enhanced in the developing primordia (Fig. 5B–D). In emerged CRs, the expression of OsAP2/ERF-40 is restricted to the root meristem (Fig. 5E). These observations together prompted us to study the function of OsAP2/ERF-40 during adventitious root development.

Fig. 5

Effect of OsAP2/ERF-40 overexpression in transgenic rice lines. (A–E) Tempospatial expression pattern of OsAP2/ERF-40 during CR primordia development (A–D) and immerged CR (E). Cross-sections (A–C) and longitudinal sections (D) of rice stem base were hybridized with anti-sense riboprobes. Black arrow in (A) marks tissues peripheral to vascular cylinder and arrowhead in (B–D) marks CR primordia. (F–L) Phenotypes of OsAP2/ERF-40 overexpression in multiple transgenic lines. (F, I) Vector control plant (F) highlighting stem base (crown region) in box and enlarged in (I) that contains several nodes with unelongated internodes. (G) An OsAP2/ERF-40 overexpression line showing nodes with significantly elongated internodes. C, coleoptilar, nodes are numbered in roman. Few aerial nodes with unelongated internodes are highlighted in rectangle. (H, J–L) Adventitious root formation from aerial nodes in two independent transgenic lines at various stages of plant growth. Red arrows mark aerial roots. Aerial roots are emerged at every nodes including uppermost nodes just beneath the panicle in (H, J). Two consecutive nodes (c and i) of elongated internode are highlighted in (L) by arrow heads. Adventitious roots at node I of (L) elongate and support plant after penetrating the soil. Bars: (A) 100 µm; (B–E) 200 µm; (F–J, L) 1 cm.

Considering higher degree of genetic redundancy among the members of AP2/ERF gene family, we took mis-expression-based approach to study the function of OsAP2/ERF-40 in rice. We generated a construct for ectopic overexpression of OsAP2/ERF-40 by cloning full-length gene into a rice expression vector, pUN under maize Ubiquitin promoter (Supplementary Fig. S4A; Prasad et al. 2001). About 45 transgenic lines were generated with pUbi-OsAP2/ERF-40-nosT construct with a range of phenotypic severity. Only few transgenic lines attained reproductive phase in our growth condition and produced very few seeds. The phenotypic severity of these transgenic lines is corroborated with the extent of overexpression of OsAP2/ERF-40, the transgenic lines expressing high level of transgene (Supplementary Fig. S4B) display stronger phenotypes as compared with weak lines (Supplementary Fig. S5A, B).

Ectopic overexpression of OsAP2/ERF-40 promotes aerial shoot-borne root formation and acts synergistically with PAT machinery

Next, we analyzed OsAP2/ERF-40 overexpression lines and observed some auxin-related phenotypes in about 50–60% transgenic lines as compared with control plants. In stronger overexpression lines, stem internodes were significantly elongated, resulting in upward shift of the crown nodes (Fig. 5G, K, L). The crown region of wild-type rice has several compressed unelongated internodes (Fig. 5F, I) which were elongated by several folds upon OsAP2/ERF-40 overexpression (Fig. 5G, L). In addition to internode elongation, overexpression of OsAP2/ERF-40 also induced roots from the aerial nodes in stronger transgenic lines (Fig. 5H, J–L). Strikingly, we observed aerial root formation from all nodes, including the node of uppermost internode, just beneath the panicle (Fig. 5H, J). In wild-type plants, internode elongation occurs after transition from vegetative to reproductive phase but root formation was not observed at the aerial nodes under our growth condition, though root primordia were seen in these upper nodes (Supplementary Fig. S5C), consistent with previous reports (Lorbiecke and Sauter 1999, Itoh et al. 2005, Yadav et al. 2011, Yamaji and Ma 2014). These aerial roots when penetrating the soil support plant growth and development (Fig. 5L), suggesting that they are functional roots. These observations suggest that OsAP2/ERF-40 is sufficient to induce root-specific developmental program at aerial nodes. Similar root phenotypes are also seen when active pool of auxin is increased, either by overexpression of auxin biosynthesis gene, OsYUCCA1 (Yamamoto et al. 2007, Zhang et al. 2018) or downregulation of auxin-inactivating gene, OsMGH3 (Yadav et al. 2011). The phenotypic similarities of these lines with OsAP2/ERF-40 overexpression lines further support that OsAP2/ERF-40 controls auxin-mediated signaling in rice.

We next examined if OsAP2/ERF-40 overexpression acts synergistically with PAT machinery. Towards this, we pharmacologically inhibited the PAT by treating transgenic lines with suboptimal concentration of PAT inhibitor, 1-N-naphthylphthalamic acid (NPA). Treatment of wild-type seedling with various concentrations of NPA suggested that 50 nM NPA does not significantly affect root numbers or growth (Fig. 6A, B;Supplementary Fig. S6). We, therefore, germinated wild-type and OsAP2/ERF-40 overexpression lines on media supplemented with 50 nM of NPA. Unlike in wild-type plants, root growth was strongly inhibited by NPA in overexpression lines (Fig. 6C, D), suggesting that OsAP2/ERF-40 overexpression acts synergistically with PAT machinery to control CR development.

Fig. 6

Regulatory relationship of OsAP2/ERF-40 with auxin and OsERF3–OsWOX11–OsRR2 pathways. (A–D) Effects of NPA treatment (50 nM) on 7-day-old wild-type (A, B) and Ubi-OsAP2/ERF-40 plants (C, D). (E) Real-time qRT-PCR analysis of OsCRL4, OsYUCCA1, OsYUCCA6, OsIAA1 and OsGH3.8 in shoot tissues of a weak (L#7) and two strong overexpression lines (L#3 and L#1). (F) Quantitative overexpression of OsAP2/ERF-40 and expression levels of OsWOX11, OsERF3 and OsRR2 in shoot tissues of transgenic lines (L#7, L#3 and L#1). OsUBQ5 was used as an internal reference gene. Three technical replicates were measured to calculate log₂ fold change ±SD by normalizing with vector control lines. Bars: (A–D) 1 cm.

OsAP2/ERF-40 activates auxin signaling and OsERF3–OsWOX11–OsRR2 pathways in ectopic shoot tissues

The phenotypes of OsAP2/ERF-40 overexpression indicate that auxin signaling might be affected in the overexpression lines. To study this, we analyzed the effect of OsAP2/ERF-40 overexpression on expression levels of various components of the auxin signaling pathway. We determined expression levels of auxin biosynthetic genes (OsYUCCA1 and OsYUCCA6), regulator of auxin transport, OsCRL4/OsGNOM1 and early auxin-responsive genes (OsIAA1 and OsGH3.8) in OsAP2/ERF-40 overexpression lines. Our qRT-PCR analysis showed that the expression of OsCRL4/OsGNOM1, OsYUCCA1, OsYUCCA6, OsIAA1 and OsGH3.8 was induced in OsAP2/ERF-40 overexpression lines (Fig. 6E). We have observed that extent of induction of these genes depends on expression level of transgene, weak line L#7 has lower induction than strong lines (Fig. 6E). Thus, these data together suggest that OsAP2/ERF-40 regulates several steps of auxin signaling pathway (e.g. auxin biosynthesis, auxin distribution and its responses) in a dose-dependent manner.

In rice, physical and regulatory interactions among OsWOX11, OsERF3 and OsRR2 ensure CR initiation and elongation (Zhao et al. 2009, Zhao et al. 2015). OsERF3 induces expression of OsRR2 during CR initiation whereas represses OsRR2 expression together with OsWOX11 at later stage of CR elongation (Zhao et al. 2015). We, therefore, studied the regulatory interaction of OsAP2/ERF-40 with OsERF3–OsWOX11–OsRR2 pathways. We analyzed expression levels of OsERF3, OsWOX11 and OsRR2 in shoot tissues of a weak and two stronger OsAP2/ERF-40 overexpression lines by qRT-PCR analysis. We observed induced expression of OsERF3 in all three lines to a similar level whereas expression of OsWOX11 was induced to at lower extent in the weak transgenic lines whereas its expression was strongly induced in strong lines (Fig. 6F). On the other hand, the expression level of OsRR2 was not affected in the weak lines but was significantly repressed in the strong lines. These observations suggest that a strong induction of OsWOX11 is needed to repress the expression of OsRR2 in shoot tissues and OsAP2/ERF-40 is inducing root formation through OsERF3–OsWOX11–OsRR2 pathway.

Discussion

Regulation of root architecture by auxin and cytokinin signaling pathways

Auxin and cytokinin control root branching in higher plants by regulating the expression of various TFs (Coudert et al. 2010, Orman-Ligeza et al. 2013). The regulatory networks operating during rice CR specification and differentiation are largely conserved but some species-specific divergence also exists (Coudert et al. 2010, Orman-Ligeza et al. 2013). Modulation of auxin levels or mutation in the components or upstream regulators of auxin signaling pathway (e.g. OsYUCCA1, OsIAA23, OsGH3.8, OsCRL6, OsCRL4/OsGNOM1 or OsCAND1) displays defects in CR patterning and development (Yamamoto et al. 2007, Ding et al. 2008, Kitomi et al. 2008, Liu et al. 2009, Jun et al. 2011, Wang et al. 2011, Yadav et al. 2011, Wang et al. 2016, Zhang et al. 2018). These together suggest that proper auxin biosynthesis, distribution, activation and signaling are essentially required for CR primordia initiation and emergence. Our phenotypic analysis on the effects of auxin on root number and growth showed that IAA induces CR primordia formation but suppresses their growth. Importantly, the origin of additional CRs upon exogenous auxin treatment is same as the natural CRs, suggesting that only restricted tissues are competent to initiate CR-specific development program in response to auxin signaling.

Apart from auxin, cytokinin also plays crucial role in root development. In Arabidopsis, cytokinin suppresses LR initiation directly by acting at LR founder cells (Laplaze et al. 2007) whereas cytokinin inhibits initiation of LRs, but promotes their elongation in rice (Debi et al. 2005). Moreover, alteration in cytokinin metabolism also affects CR development in rice. For example, in a dominant mutant root enhancer1 (ren1-D) of OsCKX4 gene, CR numbers are increased (Gao et al. 2014). Our observation of decreased CR number upon cytokinin treatment was in line with the above observations. The effects of cytokinin on growth of rice CRs and LRs are opposite, it inhibits CRs growth (Fig. 1A, C) but stimulates LR elongation (Debi et al. 2005). The differential effects of cytokinin on rice CRs and LRs indicate root-type specific developmental output of cytokinin in rice. Cytokinin-regulated genes have been recently identified in shoots and roots of rice (Raines et al. 2016). Here, we have identified genes whose expression is regulated by the BAP-mediated signaling pathway in rice stem base, which contains developing CR primordia and showed that some of them are specifically regulated by BAP but not IAA.

Cross talk between hormones and TFs during CR formation

Auxin and cytokinin signaling pathways function through auxin response factors (ARFs) and cytokinin response regulators (RRs), respectively, to regulate their downstream genes including TFs and often cross talk during plant development (El-Showk et al. 2013, Lavy and Estelle 2016). OsCKX4 controls cytokinin homeostasis and integrates auxin and cytokinin signaling pathways in CRs (Gao et al. 2014). It is directly regulated by auxin through OsARF25 and cytokinin through OsRR2 and OsRR3 in roots (Gao et al. 2014). Individual and a cross talk between auxin and cytokinin signaling pathways also regulate expression of several TFs such as OsCRL1, OsCRL5, OsWOX11, OsQHB and OsERF3 during CR development (Kamiya et al. 2003b, Inukai et al. 2005, Zhao et al. 2009, Kitomi et al. 2011a, Orman-Ligeza et al. 2013, Mai et al. 2014, Zhao et al. 2015). The expression of rice OsCRL1 is positively regulated by ARF protein (Inukai et al. 2005). In Arabidopsis, ARF7 and 19 regulate expression of OsARL1/OsCRL1-related genes, LBD16/ASL18 and LBD29/ASL16 during LR development (Okushima et al. 2007).

Some of these TFs also integrate auxin and cytokinin signaling pathways. The expression of OsWOX11 and OsERF3 is induced by both, auxin and cytokinin that in turn regulates the expression of several auxin and cytokinin responsive genes (Zhao et al. 2009, Zhao et al. 2015). OsCRL5, OsWOX11 and OsERF3 activate expression of type-A cytokinin RRs which are negative regulators of cytokinin pathway and overexpression of OsRR1 partially complements crl5 mutants (Zhao et al. 2009, Kitomi et al. 2011a, Zhao et al. 2015). Our study provides global genes and TFs, which are commonly and specifically regulated by auxin and cytokinin. An over-representation of induced TFs upon auxin treatment, suggests activating regulatory roles of auxin during CR specification. Our experimental data also revealed that IAA, but not BAP, induces expression of some plant-specific TFs. Together these findings suggest that a cross talk between various TFs, auxin and cytokinin signaling during CR development.

Regulatory interactions between CR-specific TFs and auxin signaling

Most of the CR regulators act downstream of auxin signaling pathway. A LOB-domain TF, OsCRL1 is essential for auxin-mediated CR development as crl1 mutants do not develop CR, display decreased LRs and impaired root gravitropism (Inukai et al. 2005). Also a member of AP2/ERF TF family, OsCRL5 when mutated, develops fewer CRs (Kitomi et al. 2011a). Additive phenotypes of crl1 crl5 double mutant revealed that they function in different genetic pathways during CR initiation, suggesting roles of auxin in multiple genetic pathways in CRs (Kitomi et al. 2011a). Furthermore, WUSCHEL-related Homeobox TF, OsWOX11 is necessary and sufficient to activate the CR-specific developmental program (Zhao et al. 2009). Loss-of-function wox11 mutant or OsWOX11 downregulated rice lines develop fewer CRs with decreased root length whereas overexpression of OsWOX11 promotes ectopic root formation with increased root biomass in rice (Zhao et al. 2009). Another AP2/ERF gene, OsERF3 also regulates CR development and functions in cooperation with OsWOX11 (Zhao et al. 2015). Similar to OsWOX11, CR number is reduced in OsERF3 downregulated and increased in OsERF3 overexpression transgenic rice lines (Zhao et al. 2015).

Our studies identify several TFs such as OsHOX12, OsNAC39, OsWOX6 and OsAP2/ERF-40 that act downstream of auxin signaling pathway in crown tissues. It is important to note that neither the abundance of their expression in CR primordia nor their role on CR development was known previously. Our expression analysis and in situ hybridization studies revealed the expression patterns of several auxin-responsive TFs. Interestingly, they are specifically expressed in developing CR primordia, implicating their role during CR development. Strikingly, ectopic over expression of one of CR-specific TFs, OsAP2/ERF-40 was sufficient to trigger the emergence of adventitious root from aerial nodes. OsAP2/ERF-40 not only acts downstream of auxin signaling but can also activate expression of the regulator of auxin signaling (OsCRL4/OsGNOM1), auxin biosynthetic genes (OsYUCCA1 and OsYUCCA6) and auxin-responsive genes (OsIAA1 and OsGH3.8) suggesting a possible regulatory feedback loop between auxin signaling and OsAP2/ERF-40 activity. Our studies on pharmacological inhibition of PAT in OsAP2/ERF-40 overexpressing rice plants further revealed a synergistic interaction between OsAP2/ERF-40 and PAT machinery. Taken together our studies provide compelling evidence for multilayered intricate regulatory interactions between CR-specific TF OsAP2/ERF-40 and auxin signaling which is instrumental to shape up the architecture of CR development in rice.

Regulatory module comprising of TFs is instrumental for adventitious root formation

The aerial adventitious roots in OsAP2/ERF-40 overexpression lines are fully developed and functional, indicating that OsAP2/ERF-40 could regulate proper developmental program during outgrowth of adventitious roots. Interestingly, OsCRL5 and OsWOX11 function at different developmental stages through type-A cytokinin RRs. Activation of OsRR2 expression by OsCRL5 during CR initiation and repression of the same regulator by OsWOX11 during CR emergence and growth suggest stage-specific differential effects of cytokinin signaling pathway in CRs (Zhao et al. 2009, Kitomi et al. 2011a). OsERF3 is essential for CR development and functions differentially during CR primordia initiation and elongation through cytokinin signaling. It directly induces expression of OsRR2 during primordia initiation, whereas during primordia elongation, OsERF3 interacts with OsWOX11 and represses expression of OsRR2 (Zhao et al. 2015). Our expression analysis of OsERF3, OsWOX11 and OsRR2 in OsAP2/ERF-40 overexpression lines suggests a possible mechanism wherein OsAP2/ERF-40-mediated induction of OsERF3–OsWOX11–OsRR2 regulatory module could contribute to primordia elongation and emergence (Fig. 7).

Fig. 7

Schematic diagram demonstrating plausible regulatory interactions of OsAP2/ERF-40 with other regulators of CR development. The interactions highlighted by dashed arrows in saffron color are from this study.

Materials and Methods

Plant material, growth conditions, treatments and plasmid construction

Rice (O. sativa var IR-64) seeds were de-husked, surface sterilized and germinated on 1/2 MS media (Sigma-Aldrich, Bangalore, India) with 1% sucrose (Sigma-Aldrich) and 0.3% phytagel (Sigma-Aldrich), at 26°C in 16/8 h light/dark period for 6 d. For studying the effects of hormonal treatments on primary and CR development, IR-64 seeds were grown in above conditions with 10 µM IAA (Sigma-Aldrich) and 10 µM BAP (Sigma-Aldrich). Emerged CR numbers were counted from 30 seedlings on fifth and sixth day postgermination. To study the effects of IAA and BAP on CR growth, average length of all emerged CRs was measured on sixth day postgermination. For NPA treatment, wild-type and OsAP2/ERF-40 overexpression lines were grown on 1/2 MS media supplemented with 50 nM of NPA (Sigma-Aldrich). For generating ectopic overexpression construct for OsAP2/ERF-40, 1.14-kb full-length gene was PCR amplified on rice genomic DNA using primers FPI-AP2 and RP-AP2-BamHI and was cloned in pBluescript SK⁺ at EcoRV site as blunt and sequenced. Next, BamHI fragment of OsAP2/ERF-40 was subcloned into the plant expression vector pUN under maize Ubiquitin promoter to generate pUbi-OsAP2/ERF-40-nosT.

Induction, sampling and RNA extraction

For RNA sequencing and qRT-PCR analysis, 6-day-old 20 seedlings were treated with DMSO (Himedia, Mumbai, India) as mock, 10 µM IAA (Sigma-Aldrich) and 10 µM BAP (Sigma-Aldrich) for 3 h as at this time point early/primary auxin-responsive genes were significantly induced (Jain et al. 2006a,, Jain et al. 2006b). Each treatment was performed in three independent biological replicates. About 2-mm stem base containing crown tissues was collected under dissecting microscope. Total RNAs were extracted from crown tissues using Tri-Reagent (Sigma-Aldrich) following manufacturers protocol. The quantity of total RNA was measured on nanodrop and RNA integrity was checked on agarose gel. For molecular characterization of OsAP2/ERF-40 overexpression lines, total RNA was extracted from shoot tissues using Tri-Reagent (Sigma-Aldrich).

RNA sequencing and data analysis

A total of nine RNA sequencing libraries from three biological replicates of each sample (mock, IAA and BAP treatments) were constructed and sequenced on Illumina HiSeq 2000 platform as described previously (Garg et al. 2017) to generate at least 40 million reads (100 bp paired-end) for each sample. The raw data obtained were preprocessed using NGS QC Toolkit (Patel and Jain 2012) to remove low-quality reads and those containing adaptor/primer contaminations. The filtered reads from each sample were mapped on the rice genome using Tophat2 followed by assembly and quantification using Cufflinks as described (Garg et al. 2017). A consensus assembly of all the samples was generated using Cuffmerge. Differential gene expression analysis was performed using Cuffdiff and a list of genes with at least 2-fold change (log₂ fold change ≥1) and P-value <0.05 was considered as differentially expressed. A marginal but statistically significant change in the expression levels of TFs might have significant effects on their downstream targets; therefore, for TFs, at least 1.5-fold change (log₂ fold change 0.75) and P-value <0.05 were considered for differential expression. GO enrichment was performed using BiNGO plug-in of Cytoscape (version 3.3.0) with P-value ≤0.05. GO enrichment of each condition was further used to make comparative enrichment map via Cytoscape.

Reverse transcription and real-time PCR

Total RNAs were treated with DNase I (New England BioLabs, Ipswich, MA, USA) and precipitated after phenol and chloroform treatment. For validation of deregulation, oligo(dT)-primed cDNA was synthesized from 1 to 1.5 µg of total RNA using M-MuLV reverse transcriptase (NEB). A total of 10 ng cDNA were used for quantitative real-time PCR (qRT-PCR) using 250 nM of gene-specific primers and Power SYBR green master mix (Thermo Scientific, San Jose, CA, USA) in QuantStudio 3 machine (Thermo Scientific). Rice UBQ5-normalized ΔΔCt was used to calculate log₂ fold change in hormonal-treated samples as compared with mock-treated samples. For organ-specific gene expression analysis and expression of various genes in OsAP2/ERF-40 overexpression lines, cDNA was synthesized from 1 µg of total RNA using iScript cDNA synthesis kit (Bio-rad Laboratories, India) and qPCR was performed using iTAQ Universal SYBR Green Supermix (Bio-rad) as described above. A list of primer sequences is provided as Supplementary Table S11.

Histology

For histological and RNA in situ hybridization analysis, rice stem base was fixed in FAA (10% formaldehyde, 5% glacial acetic acid and 50% ethanol), dehydrated through ethanol series and embedded in paraffin (Sigma-Aldrich). Eight-micrometer cross-section was made using microtome (ThermoScientific), and sections were taken on poly-l-lysine-coated glass slides. For histology, the sections were dewaxed in xylene and stained with 1% toluidine blue (Sigma-Aldrich).

In situ hybridization

For preparing DIG-UTP-labeled riboprobes, 131-bp gene-specific region of OsHOX12, 133 bp of OsNAC39, 133 bp of OsWOX6-1 and 258 bp of specific sequences of OsWOX6-2 were cloned into pBluescript SK+ as a blunt in EcoRV site. The anti-sense probe for OsHOX12 and OsNAC39 was generated using EcoRI-lineralized pBluescript SK+ clones transcribed with T7 RNA Polymerase (Sigma-Aldrich) and HindIII-linearized clones for both splice variants of OsWOX6 using T3 RNA polymerase (NEB). OsAP2/ERF-40 anti-sense probe was prepared by in vitro transcription using T7 RNA Polymerase on OliI digested pBluescript SK+-dsOsAP2/ERF-40 clone. For making sense probe for OsAP2/ERF-40, OsHOX12 and OsNAC39, above clones were linearized with HindIII and transcribed with T3 RNA polymerase. For OsWOX6-2, sense probe was generated by EcoRI-digested pBluescript SK+ clone transcribed with T7 RNA Polymerase. Hybridization was performed on cross-sections as described by Prasad et al. (2005). The signal was developed using alkaline phosphatase-conjugated anti-DIG antibodies (Sigma-Aldrich) and NBT/BCIP substrate (Sigma-Aldrich). Sections were mounted in Entellan (Merck-Millipore, Darmstadt, Germany).

Rice transformation, generating transgenic lines and phenotyping

OsAP2/ERF-40 overexpression construct was mobilized to Agrobacterium tumefaciens LBA4404 and used to raise transgenic rice lines as described by Prasad et al. (2001). In brief, embryogenic calli of rice japonica var. TP309 was cocultivated with Agrobacterium harboring the construct and selected on NB6-SEL medium containing 50 mg/l hygromycin (Sigma-Aldrich). Actively proliferating calli were further transferred to NB6-RM media (50 mg/l hygromycin, 3 mg/l BAP and 0.5mg/l NAA) for shoot regeneration. Regenerated shoots were subsequently transferred to rooting media (1/2 MS with 25 mg/l hygromycin, 0.05 mg/l NAA). Hardening of regenerated plantlets was done in soilrite before transferring to clay for completing the life cycle. As OsAP2/ERF-40 overexpression lines with strong transgene expression did not produce fertile seeds, multiple independent stronger lines were phenotyped in T0 generation for their internode elongation and aerial root phenotypes.

Acknowledgments

We are thankful to Annapurna Bhattacharjee and Vikash K. Singh for their help with initial genomics experiments. Khrang Khrang Mushahary and Jiya Singh are acknowledged for their help in growing plants.

Funding

The work at S.R.Y. laboratory was supported by Science and Engineering Research Board (SERB), Department of Science and Technology (DST), Government of India, under Early Career Research Award scheme and Indian National Science Academy (INSA), India. M.J. acknowledges funding from the Department of Biotechnology (DBT), Government of India, under the National Bioscience Award for Career Development scheme. K.P. acknowledges fund support from Department of Biotechnology (DBT), Government of India. Infrastructural facilities were provided by Indian Institute of Technology, Roorkee and Jawaharlal Nehru University, New Delhi, India. University Grant Commission (UGC), India is acknowledged to provide fellowships to A.N., H.S. and A.K.D., Indian Institute of Technology, Roorkee to A.K., T.G. and Z.S. and Centre of Excellence in Bioinformatics at the School of Computational & Integrative Sciences, Jawaharlal Nehru University, New Delhi to U.S.

Disclosures

The authors have no conflicts of interest to declare.

Author notes

Ananya Neogy and Tushar Garg authors contributed equally.

Deceased.