Conserved LBL1-ta-siRNA and miR165/166-RLD1/2 modules regulate root development in maize

Angiospermic plant groups, such as dicotyledons and monocotyledons, display significant variation in their root system architecture (RSA), which comprises primary root (PR) and root branches (Rich and Watt, 2013; Yruela, 2015). Dicots develop a tap root system, which consists of a PR and lateral roots (LRs), whereas monocots have a fibrous root system that, in early growth phase, consists of embryonic PR and seminal root (SR), post-embryonic crown root (CR) and LRs (Rich and Watt, 2013; Yruela, 2015). However, in monocot maize, shoot-borne CRs dominate during the later growth phase, which are absent in eudicot Arabidopsis thaliana (Arabidopsis). Roots of monocot and dicot plants also differ anatomically in terms of quiescent centre (QC) cell number, cortical cells/cell files, root initials and vascular pattern (Hochholdinger and Zimmermann, 2008; Orman-Ligeza et al., 2013). Maize root possesses polyarch type of stelar organization consisting of several xylem poles alternating with the same number of phloem pole, which differs from that of Arabidopsis (Hochholdinger et al., 2018; Orman-Ligeza et al., 2013).

Despite having architectural and anatomical differences, recent studies have shown the role of some common genetic factors in regulation of shoot-borne CRs and root-borne LRs in monocots and dicots (Hochholdinger et al., 2018; Orman-Ligeza et al., 2013). Monocot specific root types, such as SRs and CRs, involve some level of distinct genetic regulation when compared with the eudicot Arabidopsis. Maize ROOTLESS CONCERNING CROWN AND SEMINAL LATERAL ROOTS (RTCS) and its rice homologue CROWN ROOTLESS1 (CRL1) mutants fails to develop CRs, though PRs and LRs remain unaffected (Inukai et al., 2005). On the other hand, the lateral rootless1 (lrt1) maize mutant lacks LR formation, while CRs remain normal (Hochholdinger and Feix, 1998). Although some LR regulatory genes are conserved between Arabidopsis and rice, about one quarter of CRL1-regulated genes are rice specific, indicating distinct regulation of CRs, SRs and LRs in monocot (Coudert et al., 2010; Zhao et al., 2009). This suggests that root type-specific regulation of developmental programme exists in maize and rice (Hochholdinger et al., 2018; Orman-Ligeza et al., 2013).

Besides protein-coding genes and phytohormones, several ncRNAs have been shown to regulate plant development by negatively regulating their target genes (Ambros et al., 2003; Chen, 2009; Petricka et al., 2012; Singh et al., 2018; Ubeda-Tomas et al., 2012). Molecular regulation of root growth and LR development is relatively well characterized in the eudicot model Arabidopsis compared with monocot plants (like rice and maize). microRNAs (miRNAs) and small interfering RNAs (siRNAs) constitute two major classes of endogenous small RNAs in plants. Ta-siRNAs are produced from TRANS ACTING siRNA (TAS) loci through the activity of specific miRNAs in both monocot and dicot (Nogueira et al., 2009; Yoshikawa et al., 2005). Maize LEAFBLADELESS1 (LBL1) is a homologue of Arabidopsis SUPPRESSOR OF GENE SILENCING3 (SGS3), which is hypothesized to be involved in the stabilization of the miRNA-cleaved transcripts generated from the TAS loci. Mutation in LBL1, as observed in the strong allele lbl1-rgd1, leads to abaxialized radial leaves and the retarded plant fails to enter reproductive state, which is partially different and much more severe than observed in the Arabidopsis sgs3-11 mutant (Peragine et al., 2004; Timmermans et al., 1998). A relatively weaker allele lbl1-ref also shows a leaf phenotype (Nogueira et al., 2007). This suggests the function of LBL1 in maize leaf development is slightly different from that of SGS3. Besides the TAS3-derived ta-siRNAs pathway, miR166 also contributes to the establishment of adaxial/abaxial leaf polarity by restricting the spatial expression domain of CLASS III HOMEODOMAIN-LEUCINE ZIPPER (HD-ZIP III) genes that specify adaxial fate (Husbands et al., 2009). The opposing activities of TAS3-derived tasiR-ARFs, which target AUXIN RESPONSE FACTOR3 (ARF3), and miR165/166 (Zm-miR165/166) specify the polarity of developing maize leaves (Nogueira et al., 2006). Expression of miR165/166 in the abaxial domain restricts the expression of ROLLED1 and ROLLED2 (RLD1/2), a member of HD-ZIP III gene family in maize, to the adaxial side of the leaf (Juarez et al., 2004a). A semi-dominant Rld1-Original (Rld1-O) mutation in a miR166 complementary site leads to the inability of miR166 to recognize and cleave the RLD1 transcript, resulting in the increased accumulation of its transcripts and adaxialized leaf fate (Juarez et al., 2004a). miR165/166-HD-ZIP III are known to have crosstalk with different phytohormones and their signalling (Dello Ioio et al., 2012; Singh et al., 2017). The phytohormone auxin is a major regulator of root growth and branching, both in monocot and dicot plants (Balzan et al., 2014; Benkova and Hejatko, 2009; Coudert et al., 2010). Auxin-mediated regulation of these processes is achieved through differential accumulation of auxin in various root cells and tissue types, which is attained by the spatiotemporal activity of several auxin biosynthesis genes and transporters (Orman-Ligeza et al., 2013). However, possible hormonal crosstalk with the maize ta-siRNA-ARF and miR165/166-RLD module in root development remains to be investigated.

Considering the known morphological, anatomical and genetic/molecular difference in root development between dicot and monocot models, we have anticipated distinct roles for small RNAs and their crosstalk in maize root development. In this study, we have uncovered distinct roles for the ta-siRNA pathway in maize root growth, branching and vascular patterning. We have addressed the crosstalk of ta-siRNA, miR165/166 and auxin signalling in maize root development. Additionally, we demonstrate some level of evolutionarily functional conservation of LBL1 and SGS3 between Arabidopsis thaliana and Zea mays.

To understand the possible molecular role of LBL1 in maize root development, we compared the phenotype of mutant lbl1-rgd1 roots at 7 days after germination (dag; 7-day-old) with that of wild type. We observed that lbl1-rgd1 has longer PR and a reduced number of lateral branches (SR, CR and LR) when compared with that of wild type (Fig. 1A). The primary root is about 72.61% longer in lbl1-rgd1 when compared with wild type (Fig. 1B). To determine whether increased PR length in lbl1-rgd1 was associated with a change in the root meristem size, we analysed longitudinal sections (LSs) of the PR. Meristem size was determined by calculating the distance from QC to the approximate first elongated cortical cell (Fig. S1A,B). The root meristem of lbl1-rgd1 was 34.09% longer in size than that of the wild type (Fig. S1B). We performed bromodeoxyuridine/5-bromo-2′-deoxyuridine (BrdU) staining, which marks nuclei of actively dividing cells of the root tip, to investigate the cell division activity in lbl1-rgd1 (Fig. S1C). We observed more actively dividing nuclei in the lbl1-rgd1 root meristem region, indicating increased cell division when compared with wild type (Fig. S1C). We further studied the expression levels of cell cycle regulators ZmCycB1, ZmCycD2, ZmCycD3 and ZmCycD4, and found that these genes were upregulated in lbl1-rgd1 PR tip (0-1 cm) when compared with wild type (Fig. S2A) (Hu et al., 2010). These results suggest that increased cell division activity contributes to the increase in root meristem size, and thus longer PR, in lbl1-rgd1.

Next, we analysed LRs and monocot-specific SRs and CRs in lbl1-rgd1. We observed that lbl1-rgd1 seedlings have 50% fewer CRs and 41.67% fewer SRs compared with wild type (Fig. 1C,D). The density of emerged LRs was also reduced in lbl1-rgd1 when compared with wild type by 67.31% (Fig. 1E). Root network ‘bushiness’ and area were reduced in lbl1-rgd1, which could be due to the lower number of SRs, CRs and LRs (Fig. S4). This experiment was performed on a 14-day-old plant; the parameters were checked using the GiA root online tool (Galkovskyi et al., 2012). As the number of emerged LRs was lower in the lbl1-rgd1 seedling, we checked the expression of the cell wall remodelling-related genes, such as ZmLAX1, ZmLAX2 and Zmα-EXP (homologues of AtEXP7 and AtEXP18), which have been hypothesized to be involved in LR development or emergence (Zhang et al., 2014) in Arabidopsis. We found that the expression levels of ZmLAX1, ZmLAX2 and Zmα-EXP were downregulated in the LR-forming region (1-2 cm) of the lbl1-rgd1 root (Fig. S2B). We further studied the expression levels of cell cycle regulators, ZmCycB1, ZmCycD2, ZmCycD3 and ZmCycD4, and found that these genes were downregulated in the lbl1-rgd1 LR region (1-2 cm) when compared with wild type (Fig. S2C). As lbl1-rgd1 showed altered root growth and branching, we compared the cellular anatomy between different regions of lbl1-rgd1 and wild-type root. To examine the root anatomy, we analysed the serial transverse sections (TSs) along the regions of the 7-day-old PRs that included 0-1 cm tip region (Fig. 1F) and 1-2 cm above tip region (Fig. S1D). The root tip (0-1 cm) of lbl1-rgd1 showed about 41.94% more metaxylem cells (Fig. 1H), and about 34.02% more cortical cell layers when compared with wild type (Fig. 1G). The average number of large metaxylem cells was 6.2 in wild type and 8.8 in lbl1-rgd1 in the 0-1 cm region of root tip (Fig. 1H). The average number of cortical cell layers was 6.3 in wild type and 8.4 in in 0-1 cm of lbl1-rgd1 root tip (Fig. 1G). As the density of emerged LRs was also reduced in lbl1-rgd1 when compared with wild type, we next asked whether LR emergence is affected in lbl1-rgd1. To study delay in LR emergence, first we analysed the root tip using a confocal microscope. We observed that in the 1-2 cm region of 7-day-old seedling roots, there were one emerged and two embedded (non-emerged) lateral root primordia (LRP) in lbl1-rgd1, whereas there were four emerged LRP in wild type (Fig. 1I). Comparative study of LR emergence between wild type and lbl1-rgd1 is also represented graphically (Fig. S5D). To further study the changes at a histological level, we made longitudinal sections of the 1-2 cm region of root and observed that LRP were emerged in wild type but not in lbl1-rgd1 (Fig. 1J). As the number of emerged nodal roots (CR and SR) was also reduced in lbl1-rgd1, we checked whether their emergence was affected. In transverse sections of the nodal region of 7-day-old seedlings, we observed three embedded (non-emerged) CRs in lbl1-rgd1, whereas in wild type we found two emerged and one non-emerged CR (as indicated by dark-red staining in Fig. S5). This suggests that the reduction in nodal roots in lbl1-rgd1 is due to delayed emergence. Increased cell numbers in most of the tissue layers indicate that the LBL1-mediated ta-siRNA pathway contributes to cell division in the vascular and cortical regions of the root.

As LBL1 is involved in ta-siRNA biogenesis, we investigated whether the production of TAS3-derived tasiR-ARFs was affected in lbl1-rgd1 root. We analysed the accumulation of tasiR-ARF using in situ hybridization in the PR tip and LR formation region of wild-type and lbl1-rgd1 root (Fig. 2A,B). Mature tasiR-ARFs were highly expressed in wild-type root meristem and developing LRP (Fig. 2A,B). However, in lbl1-rgd1, the accumulation of mature tasiR-ARFs was drastically reduced in both PR and LRP (Fig. 2A,B). This result suggests that the LBL1-mediated tasiR-ARF pathway contributes to maize root development.

As the accumulation of tasiR-ARFs was reduced in lbl1-rgd1 root, we have analysed the expression levels of ZmARF2 (predicted target and closest homologue of AtARF2; also referred to as ZmARF10 by Matthes et al., 2019) and ZmARF3a-e (ZmARF3a, ZmARF3b, ZmARF3c, ZmARF3d and ZmARF3e) genes by qRT-PCR; gene IDs are provided in Table S2 (Dotto et al., 2014). In comparison with wild type, the expression levels of ZmARF2, ZmARF3a, ZmARF3b, ZmARF3c, ZmARF3d and ZmARF3e were increased in lbl1-rgd1 root (Fig. 2C). Among ZmARF3a-e, ZmARF3b showed maximum upregulation (4.2-fold) in lbl1-rgd1 root (Fig. 2C). We further analysed the spatial expression pattern of ZmARF3b in PR and developing LRP in wild-type and lbl1-rgd1 root using in situ hybridization. The expression of ZmARF3b was significantly stronger in the PR and LRP of lbl1-rgd1 than in wild type, indicating the potential contribution of ZmARF3b to maize root growth and development (Fig. 2D,E).

As the distribution of auxin plays an important role in root growth and branching (Balzan et al., 2014; Benkova and Hejatko, 2009; Coudert et al., 2010; Orman-Ligeza et al., 2013), we asked whether auxin homeostasis was affected in lbl1-rgd1 root, which has altered RSA. To address this, we quantified endogenous auxin levels along different regions of the wild-type and lbl1-rgd1 roots. Maize root was divided into four regions – (a) (0-1 cm), (b) (1-2 cm), (c) (2-5 cm) and (d) (above 5 cm region) – starting from the root tip (Fig. 3A). We found reduced endogenous auxin levels in the 0-1 cm PR tip region of lbl1-rgd1 when compared with wild type (Fig. 3B,C). In the regions (b) (1-2 cm), (c) (2-5 cm) and (d) (above 5 cm), endogenous auxin level was higher in lbl1-rgd1 than in wild type (Fig. 3B). We performed the auxin immunolocalization to further confirm the change in endogenous auxin abundance in the root tip region. We found reduced auxin accumulation in the lbl1-rgd1 root tip, more specifically in columella, the QC area, cortical cell layers and the vascular region, in comparison with wild type (Fig. 3C). As the endogenous auxin level was reduced in the lbl1-rgd1 PR tip, we analysed the expression of genes involved in auxin biosynthesis and transport. The expression of ZmYUCCA1 (ZmYUC1), ZmYUC2 and ZmYUC3 were downregulated in the lbl1-rgd1 PR tip region (Fig. S3A). We analysed the transcript levels of auxin efflux carriers (transporter), ZmPINFORMED1 (ZmPIN1a, ZmPIN1b, ZmPIN1c and ZmPIN1d), ZmPIN2 and ZmPIN7 in lbl1-rgd1 and wild-type root tip. Expression analysis showed that ZmPIN1a, ZmPIN1d and ZmPIN7 were significantly upregulated in lbl1-rgd1 root (Fig. S3B). Furthermore, we have performed ZmPIN1 protein immunolocalization to substantiate the changes in auxin transport in the PR tip. We found that the accumulation of PIN1 protein increased in lbl1-rgd1 PR tip, with more abundance in the cortical cells (Fig. 3D). In LR primordia, the expression domain of ZmPIN1 protein expanded in lbl1-rgd1, in comparison with wild type (Fig. 3E).

As LBL1 regulates miR165/166 expression in maize shoot (Chitwood et al., 2007; Juarez et al., 2004a), and the miR165/166-HD-ZIP III module regulates root growth and vascular patterning in Arabidopsis (Carlsbecker et al., 2010; Miyashima et al., 2011; Singh et al., 2014; Turchi et al., 2015), we asked whether this regulatory module is conserved in maize root development. To address this, we studied the expression of miR165/166, RLD1 and RLD2 in lbl1-rgd1 root. We observed that miR165/166 was upregulated, and RLD1 and RLD2 were downregulated in lbl1-rgd1 root compared with wild type (Fig. 4A). Furthermore, we confirmed increased miR165/166 accumulation in lbl1-rgd1 root through in situ hybridization using a LOCKED NUCLEIC ACID (LNA) probe (Fig. 4B). To investigate whether the altered expression of miR165/166, RLD1/2 (HD-ZIP III) might have contributed to the root phenotype of lbl1-rgd1 in maize, we analysed the root growth and anatomy of the gain-of-function mutant Rld1-O, the transcripts of which are resistant to miR165/166-mediated cleavage. We observed that the primary root of Rld1-O was shorter than in wild type (Fig. 4C,D) with no significant difference in the CR and SR number, although occasional alteration was observed (Fig. 4C,E,F). Interestingly, there were six cortical cell layers in Rld1-O compared with eight in wild-type root (Fig. 4G,I). Moreover, Rld1-O showed reduced average number of metaxylems and cortical cells, in comparison with wild type (Fig. 4G,I). We also observed that the root diameter (0.68 mm) and stele diameter (0.42 mm) were much narrower in Rld1-O mutant than in wild-type root (0.84 mm) and stele (0.54 mm) (Fig. 4H,I). Our results suggest that the miR165/166-RLD1/2 module regulates root growth and vascular or anatomical patterning in maize and this module acts downstream of the LBL1/ta-siRNA pathway.

As both maize LBL1 and its Arabidopsis homologue SGS3 proteins show 65% of amino acid similarity (Nogueira et al., 2007), we asked whether they are functionally conserved. First, we carefully analysed the phenotypic similarities of lbl1-rgd1 and sgs3, in terms of leaf and root development. Consistent with a previous report, we observed that sgs3-11 leaves were elongated and downwardly curled (Fig. S3C) (Peragine et al., 2004). However, strong allele lbl1-rgd1 showed a severe phenotype with radialized and abaxialized leaves in maize (Timmermans et al., 1998). Two independent alleles, sgs3-11 and sgs3-13, which were genetically cleaned through backcrossing, showed longer PR and reduced length of LRs in Arabidopsis (Fig. 5A,B). The number of LRs was increased in sgs3 root (Fig. 5A). To determine whether both LBL1 and SGS3 are functionally conserved, we developed a complementation construct by expressing ZmLBL1 under the Arabidopsis SGS3 promoter (pAtSGS3:ZmLBL1) and transformed that into homozygous sgs3-11 plants, thus making sgs3-11/- (pAtSGS3:ZmLBL1). We found that the leaf curling phenotype of sgs3-11 was restored in the sgs3-11/- (pAtSGS3:ZmLBL1) line in Arabidopsis (Fig. S3C). As lbl1-rgd1 produced a longer PR, similar to sgs3, we investigated whether LBL1 could rescue the root growth phenotype of sgs3-11. Interestingly, we observed that the complementation line could restore the root length to wild-type levels (Fig. 5A,B). Additionally, we observed that the root meristem size of sgs3-11 was longer than in wild-type plants, whereas the root meristem size of sgs3-11/- (pAtSGS3:ZmLBL1) was restored to wild-type levels (Fig. 5C,D). Furthermore, we checked the expression level of the tasiR-ARF target genes AtARF2, AtARF3 and AtARF4 in wild type, sgs3-11 and the complementation line. In sgs3-11, AtARF2, AtARF3 and AtARF4 were upregulated due to absence of tasiR-ARFs production, whereas in the complementation line, AtARF2, AtARF3 and AtARF4 expression levels were restored to those in wild type (Fig. S3D). Thus, our results showed that LBL1 is able to rescue the leaf and root developmental defects of sgs3-11, possibly by restoring functional tasiR-ARF production in sgs3-11/- (pAtSGS3:ZmLBL1) plants. This indicates functional conservation of the LBL1- and SGS3-mediated ta-siRNA pathways between dicots and monocots.

The developmental pattern of different root types (PR, SR, CR and LR) is regulated by a partially independent genetic pathways. Molecular genetic evidences indicates that the developmental programmes of different root types are at least partially distinct between monocot and dicot plants (Orman-Ligeza et al., 2013). Our study underscores the role of small RNAs (ta-siRNA and a miRNA) in maize root development and their functional conservation with Arabidopsis. We show that maize LBL1, which is involved in ta-siRNA biogenesis, regulates root growth and branching possibly by modulating phytohormone auxin and miR165/166-RLD1/2 module.

LBL1 has been previously shown to regulate maize leaf polarity through ta-siR-ARF (Juarez et al., 2004b; Timmermans et al., 1998). In this study, we showed that the lbl1-rgd1 mutant produced longer PR and showed reduced number of SRs, CRs and LRs (Fig. 1A-E). The number of metaxylem and cortical cell layers was also reduced in lbl1-rgd1 root (Fig. 1F). This suggests that LBL1 regulates root growth, branching and anatomy in maize. BrdU staining and meristem cell count results indicate increased cell division in the lbl1-rgd1 root, which might result in enhanced root growth (Fig. S1C).

In maize, LBL1 is required for the biogenesis of TAS3 locus-derived tasiR-ARFs production, which targets transcripts of ZmARF3 family genes in shoot apex (Dotto et al., 2014; Nogueira et al., 2007). The leaf polarity defects observed in lbl1-rgd1 are caused by reduced tasiR-ARF accumulation and thus increased activity of ZmARF3 genes in vegetative apices (Dotto et al., 2014). Recently, five ZmARF3 genes (ZmARF3a-e) were shown to be the targets of tasiR-ARF, and the expression of all ZmARF3 genes, except ZmARF3b, was upregulated in lbl1-rgd1 vegetative apices, suggesting that ZmARF3b may not contribute to the leaf polarity defects in lbl1-rgd1 (Dotto et al., 2014). We showed that tasiR-ARF accumulation was reduced in root meristem and developing LRP of lbl1-rgd1, which correlate to the upregulated expression of ZmARF2 (a predicted target) and all five ZmARF3 (a-e) genes in lbl1-rgd1 root (Fig. 2C). Thus, impaired tasiR-ARF production and availability in lbl1-rgd1 leads to increased abundance of ZmARF3b and other ZmARF2/3 genes that contribute to the altered root phenotype. The upregulated expression of ZmARF2, a predicted target (not validated in maize through RNA ligase-mediated-rapid amplification of cDNA ends, RLM-RACE), indicates its potential contribution to root development. In contrast to this observation in maize, the mutation in TAS3 ta-siRNA pathway genes or their targets affects LR development, but not PR growth, in Arabidopsis (Marin et al., 2010; Yoon et al., 2010). Thus, our results suggest that LBL1-mediated balanced expression of tasiR-ARF and their target ZmARF3s contributes to proper root growth, anatomy and branching in maize. However, we cannot ignore the possible contribution of other ta-siRNAs and/or other phased siRNAs (phasiRNAs) to LBL1-mediated regulation of root development, which need to be explored further.

Interestingly in lbl1-rgd1 mutants, the misexpression of tasiR-ARF and target ZmARF3 is known to cause adaxial-abaxial polarity defect in leaf (Juarez et al., 2004a); however, this phenotype is not obvious in root development (current study). Moreover, some overlap in expression of tasiR-ARF and specific target ZmARF genes indicates their distinct regulatory interaction in root. It is also likely that, not only the spatial distribution, but also the maintenance of a balanced dose of ta-siRNA target is important for proper root development in maize.

The phytohormone auxin, which is biosynthesized in young apical regions and transported to different parts of the plant by various auxin influx (AUX/LAX) and efflux (PINs) carriers, plays a crucial role in root growth and branching (Orman-Ligeza et al., 2013; Overvoorde et al., 2010; Peret et al., 2009). The tissue-specific biosynthesis and transport leads to differential accumulation of auxin along various root tissues, which is required for proper growth of PR and LR formation, and emergence in Arabidopsis and rice (Coudert et al., 2010; Hochholdinger and Tuberosa, 2009; Orman-Ligeza et al., 2013; Peret et al., 2009). OsIAA11/13 and OsCRL1 are required for auxin-dependent formation of LR and CR, respectively, in rice (Inukai et al., 2005; Kitomi et al., 2012; Zhu et al., 2012). Our results showed that accumulation or distribution of endogenous auxin in various regions along the root was affected in lbl1-rgd1 (Fig. 3B,C), which could be caused by misregulation of biosynthesis and transporter genes. In lbl1-rgd1, reduced expression of the auxin biosynthesis genes ZmYUC2 and ZmYUC3 (Fig. S3A) altered expression of PIN family genes (Fig. S3B), and PIN1 protein (Fig. 3D) could have contributed to the change in auxin homeostasis observed in different regions of the root (Fig. S3A and B; Fig. 3D). As lbl-rgd1 has a leaf defect (Timmermans et al., 1998), we cannot rule out the partial contribution of a possible change in shoot-derived auxin and metabolite flow to root phenotype, which is an aspect for future study. However, LBL1 appears to make an important contribution to the local biosynthesis, transport and tissue-specific distribution of auxin along the zones of the root, which is known to play a pivotal role in root growth and branching (Ditengou et al., 2008; Peret et al., 2009; Singh et al., 2020). The reduction in auxin accumulation in PR (‘a’ region) of lbl1-rgd1 might contribute to the change in root growth, whereas altered distribution of auxin in the branch formation regions of root (‘c’ and ‘d’ regions) might be responsible for reduced emergence and growth of LRs or CRs (Fig. 3B,C). It has been shown that auxin distribution and maxima formation, and even the subcellular distribution, play crucial roles in root growth and branching (Ditengou et al., 2008). It will be interesting to address in future studies whether subcellular auxin distribution is affected in the LR-forming cells and vascular initials of lbl1-rgd1 root.

Cytokinins (CKs) regulate root growth and branching (LR and CR) by antagonistically affecting auxin transport and biosynthesis (Benkova and Hejatko, 2009; Laplaze et al., 2007; Peret et al., 2009; Rani Debi et al., 2005). OsCRL5 is induced by auxin and regulates CR initiation through repression of CK signalling in rice (Kitomi et al., 2011; Zhao et al., 2009). Auxin and overexpression of OsYUC1 induces the expression of WUSCHEL-RELATED HOMEOBOX11 (OsWOX11), which further interacts with ETHYLENE RESPONSE FACTOR3 (ERF3) and regulates CR initiation and elongation by modulating CK signalling in rice. This indicates the importance of auxin-CK crosstalk in root development (Zhao et al., 2015). We showed that the expression of maize homologue of Arabidopsis ISOPENTENYL TRANSFERASE genes (AtIPTs), which are involved in CK-biosynthesis, was altered in PR and branching regions of the lbl-rgd1 root (Fig. S3E,F). This altered CK biosynthesis might be regulating LR formation and emergence by antagonistically affecting auxin biosynthesis, transport and signalling (Fig. S3A,B; Fig. 2C). This indicates that an imbalance of auxin-CK homeostasis along the root could contribute to the altered root growth and branching phenotype observed in lbl-rgd1. It is possible that the LBL1-mediated ta-siRNA pathway helps in maintaining auxin homeostasis and auxin-CK balance, which is required for proper root growth and branching in maize through regulation of downstream pathway genes.

In Arabidopsis, LAX3 expression in the cells overlying LRP promotes auxin influx into these cells to induce cell-wall remodelling genes and facilitate lateral root emergence (Balzan et al., 2014; Peret et al., 2013; Swarup et al., 2008). ZmLAX1 and ZmLAX2 were downregulated in the root branching mutant rootless with undetectable meristems1 (rum1) in maize (Zhang et al., 2014). In both dicot and monocots, EXPANSIN (EXP) plays a role in root development (Cho and Cosgrove, 2002; Marowa et al., 2016). Reduced expression levels of ZmLAX1, ZmLAX2 and Zmα-EXP (Fig. S2B) may possibly contribute to delayed emergence of LRs or CRs in lbl1-rgd1 in maize, although this needs to be substantiated with further genetic studies.

The opposing activities of TAS3-derived tasiR-ARF and miR165/166 regulate leaf polarity in maize by negatively regulating target ARF2/3 and RLD1/RLD2, respectively (Chitwood and Timmermans, 2007; Juarez et al., 2004a,b). LBL1 regulates miR165/166 expression in maize shoot (Juarez et al., 2004a; Nogueira et al., 2009, 2007), and the miR165/166-HD-ZIP III module regulates root growth and vascular patterning in Arabidopsis (Carlsbecker et al., 2010; Dello Ioio et al., 2012; Miyashima et al., 2011; Singh et al., 2014; Turchi et al., 2015). We show that the LBL1 mediated ta-siRNA pathway regulates root growth and branching, and that the expression of miR165/166 was upregulated in a broader domain, whereas RLD1 and RLD2 were downregulated in lbl1-rgd1 root (Fig. 4A). Interestingly, Rld1-O plants, which produce RLD1 transcripts that are insensitive to miR165/166-mediated cleavage, developed a smaller and narrower primary root with a narrower stele and reduced ground tissues than in wild type, which is in contrast to the longer PR and broader stele and ground tissue of lbl1-rgd1 (Fig. 4C,D,G-I). The variation in reduction of expression could be due to very weak expression of RLD1 in root meristem, in comparison with RLD2. Thus, it is possible that RLD2 contributes more to the root phenotype of lbl1-rgd1 than to RLD1. Like REVOLUTA (AtREV) (a homologue of RLD1/2 in Arabidopsis), RLD1/2 may regulate root development by modulating the expression of downstream target genes as reported in Arabidopsis (Brandt et al., 2012). It would be interesting to further study the possible distinct and/or redundant functional contribution of RLD1 and RLD2 to maize root development.

In Arabidopsis, endodermis-derived miR165/166-mediated repression of HD-ZIP III (PHB) in the stele in a dose-dependent manner is required for the specification of xylem, pericycle and ground tissue (Carlsbecker et al., 2010; Miyashima et al., 2011; Ursache et al., 2014). High levels of PHB promote metaxylem specification (numbers) and increase in ground tissue in Arabidopsis root (Carlsbecker et al., 2010; Miyashima et al., 2011). However, unlike Arabidopsis, Rld1-O (with high RLD transcript) root showed less metaxylem and ground tissue (cortical cell layers) (Fig. 4G,I). Together, these results suggest that miR165/166 and HD-ZIP III (PHB and RLD1/2) play similar roles in primary root growth; however, they have distinct functions in vascular and ground tissue patterning in monocot (maize) and eudicot (Arabidopsis) plants.

In Arabidopsis, auxin modulates the expression of miR165/166 and targets, HD-ZIP III genes, and regulates root growth (Singh et al., 2017, 2014), whereas root-based auxin biosynthesis and polar auxin transport regulate vascular patterning (Ursache et al., 2014). In lbl1-rgd1 root, altered auxin biosynthesis and transport leading to altered auxin homeostasis along the root might cause upregulation of miR165/166 and downregulation of RLD1/2 (Fig. S3A,B; Fig. 3B-D; Fig. 4A,B). It is possible that auxin-modulated expression of miR165/166 and RLD1/2 is mediated by the tasiR-ARF target ZmARF2/3 genes, which were upregulated in lbl1-rgd1 root (Fig. 2C). In Arabidopsis, HD-ZIP III proteins regulate auxin biosynthesis and transport in a feedback loop (Baima et al., 1995, 2001; Brandt et al., 2012; Huang et al., 2014; Turchi et al., 2015). We cannot rule out the possibility that ZmARFs and RLD1/2 mediate feedback regulation of auxin transport and/or biosynthesis in maize, which would require further studies. Moreover, altered expression of IPT genes in lbl1-rgd1 root indicates that LBL1-mediated balanced activity of auxin and CK might be required for proper cell division and differentiation, and thus root growth (Fig. S3E,F). We have conceived a hypothetical model demonstrating the role of potential molecular players involved in LBL1-mediated maize root growth and branching (Fig. 6). Taken together, our results suggest that an orchestrated crosstalk between LBL1-mediated ta-siRNA and miR165/166-RLD1/2 modules is mediated by auxin and is required for root growth and vascular patterning in maize.

AtSGS3 and ZmLBL1 have a 65% amino acid similarity, and each of them has been implicated in leaf development involving the tasiR-ARF-ARF2/3 module in eudicot Arabidopsis and monocot maize, respectively (Nogueira et al., 2007; Peragine et al., 2004; Timmermans et al., 1998). In sgs3, leaves are downwardly curled with no polarity defect; however, lbl1-rgd1 leaves become radial and abaxialized (Peragine et al., 2004), which suggests an additional role for LBL1 in moncot leaf development (Dotto et al., 2014; Peragine et al., 2004; Timmermans et al., 1998). We show that ZmLBL1 can rescue the leaf defect of sgs3-11 in Arabidopsis (Fig. S3C), which suggests the partially conserved function of AtSGS3 and ZmLBL1 in leaf development.

Despite morphological and anatomical differences, monocot and dicot root developmental programmes use some similar and distinct factors, which might be attributed to their molecular evolution (Hochholdinger et al., 2018; Orman-Ligeza et al., 2013). The loss of function of SGS3 resulted in increased PR length and LR density, whereas the loss of function of LBL1 led to the increase in PR length, number of cortical and metaxylem cells, and reduction in the number of CR, SR and LRs (Figs 1 and 5). We show that ZmLBL1 can rescue the root defect of sgs3-11 in Arabidopsis, which suggests the conserved function of AtSGS3 and ZmLBL1 in root development.

Moreover, we show that the expression of tasiR-ARF targets (ARF2/3/4) is restored to wild-type levels in the sgs3-11/− (pAtSGS3:ZmLBL1) complementation line, which further confirms the functional conservation of ARF2/3/4 at the molecular level (Fig. S3D). Thus, our results suggest that the LBL1- and SGS3-mediated ta-siRNA pathways have conserved functions in root and leaf development in monocot and dicot. However, the phenotypic differences of lbl1 and sgs3-11 indicate their distinct function, at some level, that they might have acquired in the course of evolution through sub-functionalization. It is possible that crosstalk of ta-siRNA-ARFs and miR165/166-RLD1/2 modules and phytohormone signalling contribute to this molecular variation.

Generation of lbl1-rgd1 mutant population in maize (Zea mays) B73 inbred line and molecular nature of the mutation have been described previously (Nogueira et al., 2007; Timmermans et al., 1998). A heterozygous population in the B73 background, segregating wild type, lbl1-rgd1/− and lbl1-rgd1/+, was used for phenotypical analysis. As homozygous plants were sterile, the mutant was maintained in heterozygous condition. Rld1-O is a semi-dominant mutant in maize T43 inbred line and has been described previously (Juarez et al., 2004a; Timmermans et al., 1998). A heterozygous population in T43 background, segregating wild type and Rld-O were used for phenotypical analysis of root (Juarez et al., 2004a; Nelson et al., 2002; Timmermans et al., 1998). Arabidopsis thaliana (ecotype Col-0), and sgs3-11 and sgs3-13 mutants have been described previously (Peragine et al., 2004) and were procured from the Arabidopsis Biological Research Center (ABRC), USA. We have cleaned both the mutants by backcrossing to wild type (Col-0) three times.

Maize seeds were germinated in a composite soil (agropeat: vermiculite=3:1) and were grown in a green house chamber at 28±2°C with ∼70±5% relative humidity and under a photoperiod of 16 h/8 h (day/night) (with ∼300 µmoles/m²/s light). Age of the seedling was determined as days after germination (dag), which we have used throughout the text as day-old (e.g. 7 dag has been referred to as 7-day-old). Arabidopsis seeds were grown on half-strength Murashige and Skoog medium containing 0.8% agar and 1% sugar plates (Murashige and Skoog, 1962) for root growth assays, or on soil, as described previously (Singh et al., 2012). The above experiments were repeated at least thrice with replicates to ensure the reproducibility of observed phenotype.

For visualizing maize LR root, imaging was carried out using a SP5 confocal microscope (Leica); 0.1 mg/ml of PI was used for staining the root tissue for 30 min followed by vacuum infiltration. The selected dye has excitation/emission maxima of 535/617 nm. Meristem size was calculated by measuring the distance between the QC and TZ (transition zone) where cells start elongating. The number of cortical cells within the meristem (as above) was used to determine the meristem (cortical) cell number.

Auxin immunolocalization in maize root tissue was carried out by fixing the plant root samples in 4% paraformaldehyde in 100 mM Na₃PO₄ (pH 7.2), vacuum infiltrated and rotated overnight at 4°C. Tissue was dehydrated in graded series of ethanol for half to 1 h each, diluted in tert-butanol. Sample embedding was performed at 65°C by changing the paraffin wax twice in 1 day for a period of 3 days. Samples were sectioned using a rotary microtome (8-12 µM) and were placed on the glass slides. PIN1 antibody used for immunolocalization was procured from Dr Klaus Palme (Pasternak et al., 2015). A detailed protocol of the auxin immunolocalization has been described previously (Forestan and Varotto, 2013).

For quantitative real time PCR (qRT-PCR), gene-specific primers were designed using IDT software and were custom synthesized by Sigma Aldrich. cDNA synthesis and qRT-PCR have been described previously (Singh et al., 2020). ZmUBQ6 or 18s rRNA were used as an endogenous control. Dotto et al. have assigned ZmARF3a (GRMZM2G030710), ZmARF3b (GRMZM2G441325), ZmARF3c (GRMZM2G056120), ZmARF3d (GRMZM2G437460) and ZmARF3e (GRMZM5G874163) based on their homology to AtARF3. In the current study, we have assigned GRMZM2G338259 as ZmARF2 as it is the closest homologue of AtARF2 (Dotto et al., 2014). For the in silico predicted target (ZmARF2) and previously validated targets (ZmARF3a-e) of tasiR-ARF, primers were designed in the region flanking the tasiR-ARF target sites. At least two biological and three technical replicates were used. Relative expression of genes were calculated using the ΔΔC_t method, as described previously (Pfaffl, 2001; Singh et al., 2017, 2012). All primers are listed in Table S1.

Root tissues of 7-day-old maize (wild type and lbl1-rgd1) seedlings were harvested from different regions (a, b, c and d) (Fig. 3A), and extraction was carried out with 100% methanol (2.5 ml/gram fresh weight). Plant extracts and standard substances were resolved in the reverse phase C-18 column (Apollo C–18, Altech) with a HPLC system (Simadzu). A solvent gradient programme was optimized for indole-3-acetic acid (IAA) and indole-3-propionic acid (IPA) separation in the presence of 0.3% acetic acid. The elution profile was traced with a UV detector (Nakurte et al., 2012). For comparative auxin measurement, equivalent amounts of tissues were taken from four similar regions of wild-type and lbl1-rgd1 root, and extraction was carried out. Following HPLC, auxin normalization and measurement were carried out accordingly as described previously (Kim et al., 2006; Singh et al., 2020).

Different regions (along the root) of 7-day-old maize seedling (lbl1-rgd1 mutant and wild type) were dissected, fixed in fixative, embedded in paraplast (Sigma) and sectioned using a rotary microtome (Leica RM2265). Tissue fixation and processing have been described previously (Gautam et al., 2016). Embedding and block preparation was performed at 60°C. Successive transverse sections were made in each region of root. Longitudinal sections were made for root tip, including meristem. Sections were stained with 0.5% safranin stain (Sigma Aldrich), mounted on a glass slide and visualized under a bright-field Nikon 80i or Zeiss AxioImager2 microscope. Cell size, area and diameter were calculated using ImageJ (http://imagej.nih.gov/ij/) (Collins, 2007).

In situ hybridization was carried out for visualizing the tissue specific expression of both small RNAs and target transcripts. Small RNA localization was carried out using the previously published methods with modifications (Gautam et al., 2019; Javelle and Timmermans, 2012; Singh et al., 2014). Localization of the mRNA transcripts was carried out using a previously described method with modifications (Javelle and Timmermans, 2012; Sarkar et al., 2007; Singh et al., 2014). The elaborate method of tissue fixation/processing and in situ hybridization is provided in the supplementary Materials and Methods. LNA probes for miR165/166 (Eurogentec) were as described (Singh et al., 2017), and LNA probes for tasiR-ARF were used for in situ localization. Riboprobes of ZmARF3 were prepared and labelled with digoxigenin (DIG) through in vitro transcription using T7 RNA polymerase (Roche), according to the manufacturer's instructions.

Statistical analysis was carried out using one-way ANOVA: ***P<0.001, **P<0.01, *P<0.05.

For carrying out the BrdU staining reaction, we used the BrdU Staining Kit (Invitrogen, 93-3943). Longitudinal sections of 7-day-old maize root tips were mounted on glass slides, and incubated in BrdU Labeling Reagent (Thermo Fisher Scientific, 000103) for 24 h at room temperature. BrdU was incorporated into proliferating cells (S phase) and was marked by the staining of nuclei in blue. We have followed the entire protocol for the BrdU staining according to the manufacturer's instructions (http://tools.thermofisher.com/content/sfs/manuals/933944_Rev1009.pdf).

We acknowledge the Central Instrument Facility (NIPGR, New Delhi) for real-time PCR, use of a microtome, light/confocal microscopes, plant growth facility and other facilities. We sincerely thank Prof. Marja Timmermans (CSHL, USA; currently, University of Tubingen, Germany) for plant material support and valuable critical comments on the manuscript. We thank Prof. Klaus Palme for providing PIN1 antibody. We acknowledge the DBT-eLibrary Consortium (DeLCON) for providing access to e-resources. We also acknowledge ABRC, TAIR and Maize GDB for plant material/genomic sequence resources.