miR165/166 and the development of land plants


Author to whom all correspondence should be addressed.
Email: solan@bio.c.u-tokyo.ac.jp


In the developmental process of plant cells, cell fate is determined by expression of specific genes. Their expressions are directed by specific transcriptional regulators and positional signals that convey the locations of respective cells along various body axes, including proximal–distal, adaxial–abaxial and medial–lateral axes. Recently it was reported that some small regulatory RNAs work in a non-cell-autonomous fashion and provide neighboring cells with positional signals. Among such small RNAs, we will review unique biological features of microRNA 165/166 widely involved in plant development.

Characteristics of plant development

In the developmental process of plant cells, cell fate is determined by expression of specific genes. Their expressions are directed by specific transcriptional regulators and positional signals that convey the locations of respective cells along various body axes (Husbands et al. 2009), including proximal–distal, adaxial–abaxial and medial–lateral axes. Positional information is required for pattern formation in plants, as in animals. For instance, the adaxial side of leaves has tissues specialized for light harvesting, whereas tissues on the abaxial side are specialized for oxygen and carbon dioxide exchange with the atmosphere. Over recent decades, researchers have attempted to clarify how transcriptional regulation mechanisms are triggered, silenced, or switched off in particular circumstances. Such research has revealed the function of a group of small RNAs in controlling transcriptional regulation (Chitwood et al. 2009).

Physiological studies over a long history have shown that phytohormones like auxin and cytokinin are deeply involved in pattern formation in plant tissues (Nemhauser et al. 2006). Recently, it was reported that novel peptide hormones are synthesized in specific tissues and then secreted. Diffusion of these peptide hormones conveys positional information to surrounding cells (Leibfried et al. 2005). Local concentrations, gradients, or a certain balance of hormones are formed in this process. By some as yet unknown mechanism, expressions of some specific genes are induced, promoting differentiation into particular cell types as a result of appropriate positional information from adjacent cells of different types.

In plants, there are two main regions with a high rate of cell division; the shoot apical meristem (SAM) and the root apical meristem. The central zone of the SAM consists of a mass of undifferentiated stem cells that provide newly divided cells, while cells in the peripheral zones proliferate and differentiate into lateral organs along body polarities, like adaxial–abaxial axes (Zhao et al. 2010). In leaf polarity, the side facing the meristem is adaxial and the other side is abaxial.

Some transcriptional factors work in a non-cell-autonomous manner, that is, they function in cells that are several cells distant from where they were translated (Xu et al. 2011). Some small regulatory RNAs also work in a non-cell-autonomous fashion (Chitwood et al. 2009). Here, we review some of the interesting characteristics of one such species of RNA, microRNAs 165 and 166 (miR165/166). miR165/166s have received much attention because of their unique mechanism of action, their wide range of downstream targets, and their evolutionary conservation among land plants.

HD-ZIP III proteins and determination of adaxial/abaxial polarity

Members of the Class III HOMEODOMAIN-LEUCINE ZIPPER transcriptional factors (HD-ZIP III or Class III HD-ZIP) are plant-specific and involved in many plant development processes (Emery et al. 2003). The SAM consists of two types of tissues; a group of undifferentiated cells (stem cells) and their immediate derivatives. Stem cells divide continuously throughout the growing period to maintain the meristematic cells at shoot tips. Then, descendants of stem cells give rise to differentiated cells or tissues. Whether cells in the SAM are maintained or start to differentiate is specified and directed by the actions of HD-ZIP III proteins like PHABULOSA (PHB), PHAVOLUTA (PHV), REVOLUTA (REV), ATHB8, and ATHB15 (CORONA/INCURVATA4) (Emery et al. 2003; Prigge & Clark 2006).

The actions of these proteins contribute to maintain the SAM. They also function in the formation of axillary meristems, root lateral meristems, and determination of lateral organ polarity required for laminar outgrowth. They also affect the formation of leaf primordia surrounding the SAM (Fig. 1). In addition, their actions determine vascular patterning and differentiation in leaves, stems, and roots.

Figure 1.

 Stem apical meristem (SAM) and leaf primordium showing adaxial and abaxial sides. Transcriptional factors PHABULOSA (PHB), PHAVOLUTA (PHV), and REVOLUTA (REV) are expressed in the abaxial sides of leaf primordia, where they induce transcription of downstream genes to induce adaxialization. Conversely, miR165/166 is synthesized in abaxial sides and represses leaky translation of PHB, PHV, or REV HD-ZIP III genes to avoid adaxialization of the abaxial side.

PHV, and PHB play key roles in leaf development (McConnell et al. 2001). They, alone or together, establish the adaxial identity of the lateral organs. REV is important for auxiliary meristem formation and stem vascular patterning, while PHB and PHV play roles in laminar outgrowth. The three genes encoding these proteins are actually expressed in the SAM and in the adaxial domain of the lateral organs (Fig. 1). It seems that ATHB15 and ATHB8 have antagonistic functions to REV in one tissue type, but overlapping functions with REV in the other tissues (Prigge et al. 2005).

miR165/miR166 and HD-ZIP III proteins

MicroRNAs (miRNAs) are endogenous 21–24 nt RNAs. In plants and animals, the function of miRNAs is to regulate the expression of protein-coding genes by affecting mRNA translation or stability (Rhoades et al. 2002). Such mRNAs are known as target mRNAs or mRNAs of target genes. In plants, nearly half of such target genes are transcription factors (Mallory & Vaucheret 2006). In many cases, abnormal expression or loss of specific miRNA genes results in specific developmental defects. miRNAs contribute to gene regulation over time or phase changes during plant development.

Since the early stage of discovery of plant small RNAs, miR165/166 have been characterized as a relatively abundant class of miRNAs (Reinhart et al. 2002). miR165 and miR166 have almost identical base sequences except for C-U at base 17, that is:

  • miR165: 5′-ucggaccaggcuucauccccc-3′, and

  • miR166: 5′-ucggaccaggcuucauucccc-3′.

The Arabidopsis genome contains seven copies of miR166 genes and two copies of miR165 genes (Xie et al. 2005). Transcription of each gene yields a variety of different primary transcripts (pri-miRNAs) or miRNA precursors, which form foldback (hairpin) RNA structures. However, once the pri-miRNAs are specifically processed into approximately 21–22 nt fragments via the cleavage actions of the Dicer-like protein 1 (DCL1), HYPONASTIC LEAVES (HYL1) and SERRATE (SE) (Han et al. 2004; Kurihara & Watanbe 2004; Yang et al. 2006), mature miRNAs of identical sequences are formed (Xie et al. 2005). The miRNA* (complementary strand of miRNA) strand is degraded. These miRNAs subsequently associate with Argonaute proteins (AGO) to confer silencing. Both miRNAs can anneal to the same class of target mRNAs with sequences complementary to the miR165/166 sequences and repress their expression.

The mRNA sequences of PHB, PHV, REV, ATHB8, and ATHB15 genes all contain target sites of miR165/166 (Prigge & Clark 2006). The biological implications of this were not only revealed by the fact that these mRNAs have sequences complementary to miR165/166 in the coding region (Rhoades et al. 2002), but also because their specific cleavage products were actually detected. Using an in vitro wheat germ extract system, studies showed that PHV mRNA was cleaved at the predicted binding site (Tang et al. 2003; Mallory et al. 2004). If miR165/166s are synthesized/accumulated specifically in a particular cell, then they target and repress expression of the HD-ZIP III members, PHB, PHV, REV, ATHB8, or ATHB15. This results in formation of a robust network to regulate specific transcription factors in a quantitative and spatial manner leading to normal plant development. The activities of specific transcription factors result in the correct expressions of downstream genes for tissue development.

Several groups isolated dominant gain-of-function mutants of PHB and PHV (McConnell et al. 2001; Emery et al. 2003) with wholly adaxialized leaves. Characterization of these mutants showed that the causal base changes were within a narrow stretch of base sequences in each coding region. In these mutants, the position of the causal base change coincided with the region complementary to miR165/166. In addition, several dominant gain-of-function mutants of PHB or PHV contained a silent base change in the complementary site to miR165/166. Thus, it was speculated that loss of complementarity on the HD-ZIP III coding sequences hindered binding to miR165/166, allowing them to escape from the negative regulation triggered by this binding. To confirm this hypothesis, Mallory et al. (2004) introduced a silent base change in the miR165/166 complementary site in PHB mRNA. These silent mutations caused adaxialization of leaves, like phb-d mutants, and indicated that the 3′ region of the miRNA complementary site plays a critical role in the recognition of PHB mRNA by miR165/166. Next, a cleavage efficiency test of HD-ZIP III genes containing a silent mutation in the miRNA complementary site showed decreased cleavage efficiency. Taken together, these results suggest that the negative regulation by miR165/166 binding to PHB mRNA or HD-ZIP III genes is required for normal development.

An in situ hybridization analysis was performed to detect PHB expression in the SAM and leaf primordia regions of the wild-type and the phb-1d mutant. In the wild-type plant, PHB expression was restricted to the adaxial side of the leaf primordium. In striking contrast, in gain-of-function phb-1d mutants, PHB expression was observed throughout the SAM and primordia regions. By integrating these results described above, it was concluded that phb-1d mRNA partially resists miR165/166 inhibition and directs universal PHB expression in the SAM irrespective of the presence or absence of miR165/166. This resulted in adaxialization of both sides and the formation of peculiar leaves with only upper-type surfaces and no “back-side.” In the wild-type plant, miR165/166 was expressed on the abaxial side of the leaf primordium, repressing expression of PHB on that side. Otherwise, the tissues would have the adaxial character as the default (Fig. 1).

Since the first report of the complete genome sequence of Arabidopsis in 2000, the genomes of a dozen other plants have been reported. We are now at the stage where we can obtain a series of orthologous amino acid sequences and by comparing them, discuss possible functions of gene products and their evolutionary significance based on sequence conservation. It is not surprising that HD-ZIP III genes and miR165/166 genes have been found in gymnosperms, a fern, a lycopod, a moss, a liverwort and a hornwort, because they make an important contribution to maintain the SAM and determine adaxial/abaxial polarity in plant tissues. It is more surprising that the miR165/166 binding sites have been conserved in all HD-ZIP III genes of land plants over hundreds of millions of years (Floyd & Bowman 2004, 2006). RNA degradome analysis of 3′-end cleavage products possibly generated by miRNA-directed cleavage showed identical cleavage patterns in all tested vascular plants. This result indicates the importance of negative regulation by miR165/166. On the other hand, the HD-ZIP III gene is also present in members of the Charales, a sister group of land plants whose members show filamentous apical growth. However, there are five base changes on the target site of miR165/166 in the Charales HD-ZIP III gene (Floyd et al. 2006). Thus, miR regulation appears to be absent from members of the Charales.

We speculated that the negative regulation of HD-ZIP III genes by miR165/166 would have become the standard for development in all land plants. However, comparative analyses of HD-ZIP III expressions indicated that regulation programs for initiation, vascularization, and laminar growth processes differ among lycophytes, microphylls, and seed-plant megaphylls (Floyd & Bowman 2006). Such differences might be because of differences in the regulatory circuits of respective plants. Heterologous expression of an HD-ZIP III gene isolated from nonvascular moss imperfectly complemented the Arabidopsis rev mutant, especially in terms of specialization of organ polarity (Prigge & Clark 2006). Thus, negative regulation of HD-ZIP III by miR165/166 is highly conserved in land plants, and these HD-ZIP III functions show diversity among different plant groups.

ARGONAUTE 10 protein and SAM differentiation

Argonaute protein was named after small squid (Argonaute)-like phenotype of the mutant that lacked the corresponding gene function (Bohmert et al. 1998). Proteins of this family have the carboxy-terminal PAZ and PIWI domains and are found widely in eukaryotes (Vaucheret 2008). AGO proteins form a multiprotein complex known as the RNA-induced silencing complex (RISC) and play key roles in many RNA silencing pathways as immediate small RNA interacting factors. The RISC incorporates small RNAs as sequence guides to recognize target mRNA sequences with complementary sequences, and then represses the expression of target genes by cleaving their mRNA or repressing their translation (Vaucheret 2008).

It has been shown that AGO proteins are deeply involved in embryonic development and cell differentiation. Ten genes encoding AGO proteins are present in the Arabidopsis genome (Fagard et al. 2000; Vaucheret 2008). The functional diversity of AGO proteins has been characterized mainly from the profiles of bound small RNAs. Cell lysates were subjected to immunoprecipitation experiments with an antibody and small RNAs were analyzed by large-scale sequencing. AGO1 associates with all kinds of miRNAs and a variety of small interfering RNAs. AGO proteins have some preferences for the 5′-end base species as well as for RNAs with a particular base length (Zilberman et al. 2003; Qi et al. 2006; Zheng et al. 2007; Mi et al. 2008; Montgomery et al. 2008; Takeda et al. 2008). AGO1 preferentially binds to small RNAs with a 5′-uridine (U), whereas AGO2, AGO4, AGO6, and AGO9 preferentially bind to small RNAs with a 5′ adenine (A), and AGO5 to those with a 5′ cytosine (C). AGO7, which is a close relative of AGO1, associates mainly with miR390 and initiates production of trans-acting siRNA (tasiRNA) (Montgomery et al. 2008). Among the AGO proteins, AGO10 preferentially binds sRNAs that are 21-nt long with a 5′ U, similar to the preferential binding of AGO1. In contrast, AGO4, AGO6, and AGO9 all localize in nuclei, and bind to endogenous 24-nt sRNAs to silence genomic loci rich in repetitive DNA sequences, transposons, and heterochromatin regions (Vaucheret 2008).

Several genes, like SHOOT MERISTEMLESS (STM), are required for SAM initiation and maintenance (Barton & Poethig 1993). PINHEAD (PNH)/ZWILLE (ZLL) was reported to maintain the SAM, and is now known as AGO10. The ago10 mutant embryos form only defective SAMs that cannot supply new tissues and would just develop a single central organ in the end (McConnell & Barton 1995; Lynn et al. 1999; Moussian et al. 1989). Null alleles of ago10 mutants in the Ler ecotype often fail to maintain the SAM, instead resulting in the formation of a flat SAM, a radially symmetric small pin-shaped structure without a vascular strand, or with a trumpet-shaped leaf. Recently Zhu et al. (2011) reported that AGO10 has a unique property to bind miR165/166 preferentially over other miRNAs. They also showed that AGO10 competes with AGO1, the predominant component of the RISC in tissues, to specifically bind miR165/166 and attenuate its regular action to repress target HD-ZIP III proteins (Fig. 2). The loss-of-function ago10 mutation resulted in a significant increase in the amount of miR166 loaded into AGO1 instead of AGO10.

Figure 2.

 Competitive binding of miR165/166 between AGO1 and AGO10 in the stem apical meristem (SAM)and leaf primordial regions. Like AGO1, AGO10 binds miR165/166, but it does not show cleavage activity. It results in inhibition of AGO1 cleavage action of HD-ZIP III mRNAs, leading to their expression for adaxialization.

Zhu et al. (2011) constructed a series of base-substitution mutants by exchanging nucleotides of both miR166 and its complementary strand (miRNA*) to enable them to maintain base-parings and major-minor groove structures in the context of the miRNA/miRNA* duplex. When such miR166 mutants were co-expressed with AGO10, most miR166 mutants were efficiently loaded into AGO10, except those species with a U substitution at the 5′ end, confirming the strong 5′ base preference. Most mutations in this category did not affect their ability to associate with AGO10. They also constructed artificial miR166 duplexes using miR168a or 390b precursors and specifically, a miR166/168* duplex (miR168a duplex like-mispairing structure), and a miR166/390* duplex (miR390b duplex like-mispairing structure). They expressed these constructs in planta and checked whether miR166 was preferentially precipitated with AGO10. The miR166 molecules that were expressed from the context of miR168a and 390b precursors were recovered from AGO10 in similar quantities to that of wild-type constructs, but those from miR166/168* or miR166/390* were not. This result confirmed that the internal miR166/miR166* structure promotes the association between AGO10 and miR166, but the external structure does not.

Zhu et al. (2011) introduced the above miR166/390* duplex construct (as well as a wild-type miR166/166* construct) into the Ler ecotype of Arabidopsis to obtain constitutive transgenic plants. They found that miR166/390* (whether in the stem-loop context of wild-type or miR390-type) transformants showed a higher frequency of ago10 phenotypes than other miR166/166* transformants. If there is a situation in which AGO10 binds miR166 less efficiently and AGO1 binds it more efficiently, like in ago10 or miR166/390* mutants, then there is biased entrapment of miR166 in AGO1 over AGO10. Then, typical ago10 -like phenotypes will appear in shoot apexes.

Non-cell autonomous miR165/166 action in root stele organization

The root of Arabidopsis has a radial tissue organization that is generated through division of initial cells and subsequent acquisition of cell fates. In the central vascular cylinder, known as the stele, there are two water-conducting systems, the protoxylem and metaxylem, arranged centripetally. This pattern is formed via crosstalk between the stele and the surrounding endodermis tissues.

SHORT ROOT (SHR), a plant-specific GRAS-type transcriptional factor, is first produced in the stele before moving into abutting endodermis tissues. Next, SHR in endodermis tissues induces the SCARECROW (SCR) protein, another transcription factor (Nakajima et al. 2001) (Fig. 3). Then, the combined actions of SHR and SCR induce periclinal division of initial daughter cells to yield cells with two fates; cortex and endodermis (Di Laurenzio et al. 1996; Helariutta et al. 2000; Heidstra et al. 2004; Sena et al. 2004). Thus, the SHR/SCR complex regulates the formation of the radial pattern in the root. However, AGO1 also contributes to radial pattern formation in an SHR/SCR-independent pathway (Miyashima et al. 2009). In addition, the interaction between SHR and SCR induces MIR165a and MIR166b genes (Carlsbecker et al. 2010) (Fig. 3). After being transcribed, precursor miRNAs are processed into mature miR165 or miR166. miR165/166s induce degradation of target mRNAs of HD-ZIP III transcription factors. HD-ZIP III genes regulate vascular patterning and differentiation in the stele (Fig. 3).

Figure 3.

 Schematic representation of the Arabidopsis root meristem and stele in vertical and cross sections. Color codes indicate respective tissues. SHORT ROOT (SHR) transcriptional factor is initially synthesized in the stele, then moves to the endodermis and activates SCARECROW (SCR) expression. SHR–SCR then activates miR165/166 transcription. After maturation, miR165/166 spreads to adjacent tissues in a non-cell-autonomous manner. In vascular tissues, one HD-ZIP III transcriptional factor, PHABULOSA (PHB), is repressed at the translational level by miR165/166 to a lesser degree from outer tissues inwards. Thus, some PHB mRNA remains in the innermost section, while less remains in the adjacent section. This results in the innermost and adjacent sections differentiating into metaxylem and protoxylem, respectively.

Through genetic screening of abnormal vascular development, several researchers have succeeded in isolating mutants with short roots and with misplaced metaxylem differentiation in place of the protoxylem in the stele. Such mutants had point mutations in the miR165/166 target site in the coding region of PHB, one of the HD-ZIP III genes. One such allele, phb-7d, showed a striking similarity to the shr-2 phenotype, as indicated by various cell and tissue markers. However, the phb-7d defect was not complemented by pSHR::SHR::GFP (Carlsbecker et al. 2010). It was shown experimentally that the phb-7d-type PHB transcript was resistant to the miR165/166-mediated cleavage that was observed in wild-type plants. Consequently, phb-7d plants accumulated greater levels of PHB transcripts compared with wild-type plants, resulting in the defective differentiation of metaxylem instead of protoxylem.

miR165/166 molecules move from their site of production (the endodermis) to inner stele tissues in a non-cell-autonomous manner (Miyashima et al. 2011). To integrate the results, a model was proposed in which miR165/166 is transcribed in the endodermis region, and then enters inner tissues by an unknown mechanism. miR165/166 assists the degradation of HD-ZIP III mRNA by inducing RISC activities with some AGO proteins in the destination cells. Then, the rest of the miRNA moves further inward. As a result, miR165/166 acts to suppress PHB mRNA in the regions from the endodermis inward to the pericycle and the stele periphery in decreasing order.

miR166-specific locked nucleic acid (LNA) probe hybridization showed that the levels of HD-ZIP III mRNA in the root meristem were greatest at the center of the xylem axis and decreased towards the stele periphery. The athb8-1 cna-2 phb-13 rev-6 quadruple HD-ZIP III mutant showed over-differentiation of the protoxylem in place of the metaxylem. When all five HD-ZIP III genes were lost, no xylem was formed. Taken together, these results strongly suggest that suppression of HD-ZIP III mRNA in the stele periphery through the action of miR165/166 is necessary for proper pattern formation in the xylem.

miR165/166 offers a very unique and interesting example of how miRNAs can move from cell to cell and proportionally restrict expression of target mRNAs according to information gradients along various body axes.


miR165/166 is an abundant species of micro RNA in Arabidopsis. Its actions include a wide range of negative regulations of HD-ZIP III transcriptional factors. It has roles in determining the positional fate of leaf tissues (adaxial or abaxial) and in xylem differentiation in root stele tissues. It shows non-cell-autonomous actions in an as-yet-uncharacterized manner. The action of miR165/166 is still a unique topic in small RNA action, and is an example of a robust gene regulatory system that is well conserved among diverse groups of land plants.


We would like to thank the Ministry of Education, Culture, Sports, Science and Technology for Grants-in-Aid for Science Research on Priority Areas (Grant No. 23012008, 23120508), for our small RNA research in plants.