- © 2019 American Society of Plant Biologists. All Rights Reserved.
Abstract
Maternal cells play a critical role in ensuring the normal development of embryos, endosperms, and seeds. Mutations that disrupt the maternal control of embryogenesis and seed development are difficult to identify. Here, we completely deleted four MICRORNA167 (MIR167) genes in Arabidopsis (Arabidopsis thaliana) using a clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein9 (Cas9) genome-editing technology. We found that plants with a deletion of MIR167A phenocopied plants overexpressing miRNA167-resistant versions of Auxin Response Factor6 (ARF6) or ARF8, two miRNA167 targets. Both the mir167a mutant and the ARF overexpression lines were defective in anther dehiscence and ovule development. Serendipitously, we found that the mir167a (♀) × wild type (♂) crosses failed to produce normal embryos and endosperms, despite the findings that embryos with either mir167a+/− or mir167a−/− genotypes developed normally when mir167a+/− plants were self-pollinated, revealing a central role of MIR167A in maternal control of seed development. The mir167a phenotype is 100% penetrant, providing a great genetic tool for studying the roles of miRNAs and auxin in maternal control. Moreover, we found that mir167a mutants flowered significantly later than wild-type plants, a phenotype that was not observed in the ARF overexpression lines. We show that the reproductive defects of mir167a mutants were suppressed by a decrease of activities of ARF6, ARF8, or both. Our results clearly demonstrate that MIR167A is the predominant MIR167 member in regulating Arabidopsis reproduction and that MIR167A acts as a maternal gene that functions largely through ARF6 and ARF8.
The formation of a normal Arabidopsis (Arabidopsis thaliana) seed requires the coordinated development of three genetically distinct components: seed coat, embryo, and endosperm (Ingram, 2010). The seed coat develops completely from maternal cells, whereas two-thirds of the genetic material in the endosperm is maternal. Genetic screens for mutants defective in one of the three components have identified many genes that affect embryogenesis, endosperm development, and the formation and integrity of the seed coat (Lafon-Placette and Köhler, 2014; Li and Li, 2015). Very few genes that exert maternal control of embryogenesis and seed development have been identified (Ray et al., 1996; Grossniklaus et al., 1998; Mizzotti et al., 2012; Zhang et al., 2017). The first reported maternal effect gene, SHORT INTEGUMENT, encodes DICER, which is an essential player in processing microRNA (miRNA) precursors and other small RNAs (Ray et al., 1996; Golden et al., 2002). A homozygous sin1 mutant embryo develops normally if it is generated from a self-pollinating sin1 heterozygous plant. About 90% of embryos with sin1 or sin1+/− genotypes are defective in embryogenesis when produced by pollination with a sin1 homozygous plant with pollen from wild-type or sin1 heterozygous plants (Ray et al., 1996). The sin1 studies suggest that small RNAs may play important roles in maternal control of embryogenesis and seed development. Another example of maternal effects was uncovered when either Mitogen activated Protein Kinase 6 or its upstream kinases MPK Kinase 4 (MKK4)/MKK5 are disrupted (Zhang et al., 2017). A significant fraction (6%–35%) of mpk6 or mkk4 mkk5 double mutants had embryos that burst out of the seed coats or had wrinkled seeds. Overall, it is still very difficult to study maternal effects because of a lack of proper genetic materials.
MiRNAs are a class of ∼21-nucleotide small RNAs that regulate diverse developmental processes in both plants and animals (Lee et al., 1993; Wightman et al., 1993; Bartel, 2004). It is generally believed that miRNAs down-regulate the expression of target genes via cleavage of target mRNA or inhibition of target mRNA translation (Bartel, 2009; Shukla et al., 2011). The biological functions of miRNAs were mainly inferred from overexpressing MICRORNA (MIR) genes or miRNA-resistant versions of the target genes. Whereas results from such experiments are informative, variations in phenotypes and stability are often encountered due to differences in expression levels. Moreover, a miRNA often has multiple targets, and overexpression of a miRNA-resistant version of a target may not lead to a complete picture of the miRNA functions. To unambiguously define the physiological roles of miRNAs, characterization of complete loss-of-function mir mutants is needed. Most MIR genes belong to gene families whose members presumably have redundant functions, making it difficult to obtain plants that lack a particular miRNA. Furthermore, miRNAs are produced from small genes, and the chances for isolating T-DNA or transposon insertional mutants of MIR genes are limited. In addition, miRNAs function by base pairing with their target mRNAs. A point mutation in miRNAs may not completely abolish their functions. Therefore, very few studies in plants lacking a particular miRNA have been reported (Baker et al., 2005; Allen et al., 2007; Sieber et al., 2007; Liu et al., 2010). With the advancement of clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein9 (Cas9) gene-editing technology, it is now feasible to systematically generate mir knockout mutants to study their roles in regulating plant growth and development.
We are interested in understanding how auxin controls various plant developmental processes. Auxin is perceived by the TRANSPORT INHIBITOR RESPONSE1/AUXIN SIGNALING F-BOX PROTEIN (TIR1/AFB) and AUXIN/INDOLE-3-ACETIC ACID (Aux/IAA) complexes (Dharmasiri et al., 2005; Kepinski and Leyser, 2005). Degradation of Aux/IAA repressors frees Auxin Response Factors (ARFs) for transcriptional activities. Interestingly, several key components of auxin signaling pathways are targets of miRNAs. The miRNA393 targets the mRNAs encoding the auxin receptors TIR1/AFBs (Jones-Rhoades and Bartel, 2004; Vidal et al., 2010; Si-Ammour et al., 2011; Windels et al., 2014). The miRNA160 targets the mRNAs for several ARFs, including ARF10, ARF16, and ARF17 (Mallory et al., 2005; Wang et al., 2005; Liu et al., 2007). ARF3 and ARF4 are targets of trans-acting short-interfering RNAs, which require miRNA390 for their biogenesis (Fahlgren et al., 2006; Marin et al., 2010). The miRNA167 has been reported to directly regulate auxin signaling and auxin homeostasis in Arabidopsis. The miRNA167 targets the mRNAs encoding the ARF6 and ARF8 transcription factors (Ru et al., 2006; Wu et al., 2006; Yang et al., 2006). The miRNA167 also regulates the expression levels of IAA-ALA RESISTANT3 (IAR3), which encodes an auxin-amidohydrolase and plays a role in auxin conjugation/deconjugation with amino acids in Arabidopsis (Kinoshita et al., 2012). Regulation of auxin signaling by miRNAs is evolutionarily conserved across plant species. For example, the interaction between miRNA167 and ARF6/8 is conserved among Arabidopsis, tomato (Solanum lycopersicum), and soybean (Glycine max; Yang et al., 2006; Gutierrez et al., 2012; Liu et al., 2014; Wang et al., 2015). An understanding of the molecular mechanisms of miRNAs in auxin signaling may provide guides for improving agriculturally important traits.
The Arabidopsis genome harbors four MIR167 genes (MIR167A, MIR167B, MIR167C, and MIR167D). Overexpression of MIR167A using the Cauliflower mosaic virus 35S promoter mimicked the phenotypes of arf6 arf8 double mutants (Wu et al., 2006). Interestingly, overexpression of MIR167B and MIR167C only caused mild phenotypes, whereas 35S:MIR167D plants were very similar to wild-type plants (Wu et al., 2006). Overexpression of miRNA167-resistant versions of ARF6 or ARF8 caused pleiotropic phenotypes, including small leaves and sterile flowers, suggesting that MIR167s are important for Arabidopsis development (Wu et al., 2006). Herein, we report the construction of knockout mutants of the four MIR167 genes in Arabidopsis. We found that mir167a plants were defective in anther dehiscence, ovule development, and seed development, phenotypes that were observed in plants that overexpress the miRNA167-resistant versions of ARF6 or ARF8 (referred to as mARF6 or mARF8, respectively; Wu et al., 2006). However, unlike the mARF overexpression lines, which had small leaves and normal flowering time, mir167a single mutants had normal leaf size but flowered much later than wild-type plants. More importantly, we found that the mir167a (♀) × wild type (♂) crosses failed to produce normal embryos and endosperms. It was very clear that embryos with either mir167a+/− or mir167a−/− genotypes resulting from the selfing of mir167a+/− plants developed normally, revealing a central role of MIR167A in maternal control of seed development. The mir167a phenotype is 100% penetrant, providing a valuable genetic material for studying the roles of miRNAs and auxin in maternal control of embryo and seed development. Another unexpected finding was that mir167bcd triple mutants were very similar to wild-type plants, indicating that MIR167A plays more predominant roles during plant reproductive development. MiRNA167 has several targets, but we show that miRNA167 regulates plant development mainly through controlling the expression levels of ARF6 and ARF8.
RESULTS
Deletion of MIR167A Completely Abolishes Fertility
To assess the roles of MIR167A in plant development and auxin signaling, we deleted the MIR167A gene using CRISPR/Cas9 gene-editing technology (Fig. 1A). We used two guide RNAs that targeted the promoter and the region downstream of the mature miRNA167A (Fig. 1A) in order to remove the entire MIR167A gene. Plants that harbored a deletion of a 1,170-bp fragment, including the region corresponding to the mature miRNA167A, failed to produce transcripts of the MIR167A gene (Supplemental Fig. S1) and displayed a completely sterile phenotype (Fig. 1B).
Plants without MIR167A are defective in anther dehiscence, pollen germination, and female gametophyte development. A, Generation of the deletion mutations of MIR167A using CRISPR/Cas9 gene-editing technology. Two Cas9 target sequences are shown, and the protospacer adjacent motif sites are highlighted in red. The green box represents the primary sequence of MIR167A. The small red box refers to the location of the mature MIR167A. The black line represents the promoter of the MIR167A gene. The locations of the genotyping primers are indicated. B, The sterile phenotype of mir167a mutants. Wild-type (WT) Columbia-0 (Col-0; left) and the mir167a mutant (right) are shown at 43 d after germination. Bars = 5 cm. C, The floral organs of Col-0 and the mir167a mutant. Bars = 1 mm. D, The anthers of Col-0 (left) and mir167a (middle [0 min] and right [30 min]) plants dissected from opened flowers. Bars = 500 μm. E, In vitro germination of wild-type (left) and mir167a (right) pollen. Bars = 100 μm. F to H, Mature ovules from wild type Col-0 (F) and mir167a (G and H) plants. EN, Egg cell nucleus; SEN, secondary endosperm nucleus; SN, synergid nucleus. The yellow stars indicate the leakage of embryo sac. Bars = 50 μm.
Plants without MIR167A Produce Normal Pollen But Fail to Dehisce
We analyzed the mir167a flowers to determine the causes of the observed sterility. Flowers of mir167a plants contained all of their floral organs, and all of the floral organs appeared normal except for the anthers (Fig. 1C). The anthers of the mir167a mutants were much larger than those of the wild type (Fig. 1C). Moreover, the mir167a anthers failed to dehisce (Fig. 1C) and no pollen grains were released (Fig. 1C; Supplemental Fig. S1C), which was probably the cause of the observed sterility of mir167a plants. In contrast, anther dehiscence releases pollen to the pistils when wild-type flowers are open (Fig. 1C; Supplemental Fig. S1C).
We investigated whether mir167a plants were defective in male gametophytic development. The mir167a anthers produced normal-looking tricellular integrated pollen grains (Supplemental Fig. S1). Moreover, when we analyzed the morphology of mir167a anthers with a dissecting microscope, we discovered that the mir167a anthers dehisced 30 min after they were detached from the flower (Fig. 1D). The pollen grains appeared normal. We conducted in vitro germination assays to determine the viability of the mir167a pollen. About 80% (n > 300) of wild-type pollen grains germinated normally, whereas less than 40% (n > 300) of the mir167a mutant pollen grains were able to germinate (Fig. 1E). Our results indicated that MIR167A is required for anther dehiscence and plays a critical role for pollen maturation.
MIR167A Affects Ovule Development
The seven-celled female gametophyte (embryo sac) consists of one egg cell, one central cell, two synergid cells, and three antipodal cells. During development, the three antipodal cells degenerate. Before pollination, the female gametophyte in wild-type plants consists of one egg cell, one central cell, and two synergid cells (Fig. 1F). However, only 27.2% (n = 33) of the mir167a embryo sacs contained the three cell types (Fig. 1G), and the other embryo sacs lacked any nucleus and contained degenerated remains (n = 33; Fig. 1H). In wild-type ovules, the embryo sac was enclosed by two layers of integuments: the inner and outer integuments (Fig. 1F). In contrast, the integuments in mir167a ovules failed to completely enclose the embryo sac. The mir167a embryo sac appeared to extrude out from the micropylar end (Fig. 1, G and H).
The mir167a (♀) × Wild Type (♂) Crosses Fail to Produce Normal Embryos
We pollinated mir167a pistils with wild-type pollen to investigate the function of MIR167A in embryogenesis and seed development. The wild-type pollen germinated on the stigmas of mir167a plants (Supplemental Fig. S2). After manual pollination with wild-type pollen, the siliques of mir167a plants became elongated, as we expected (Fig. 2A). However, the seeds that generated from the pollination of mir167a pistils with wild-type pollen were small and shriveled (Fig. 2, A–C). Wild-type embryos developed into the globular stage at 2 or 4 d after pollination (Fig. 2D; Supplemental Table S1). Some embryos from the mir167a (♀) × wild type (♂) crosses developed normally, but a significant number of ovules did not contain embryos at all within the same time window (Fig. 2, E and F; Supplemental Table S1). By 5 to 6 d after pollination, wild-type embryos had already developed into the heart or torpedo stage, whereas the observed embryos from the mir167a (♀) × wild type (♂) crosses were abnormal and arrested at the globular stages (Fig. 2, G and H; Supplemental Table S1).
Defects in seed development in mir167a plants after being pollinated with wild-type (WT) pollen. A, Pollination of mir167a flowers with wild-type pollen produced elongated siliques but only produced shriveled seeds. White triangles indicate the elongation of siliques resulting from pollination. mir167a (♀) × wild type (♂) crosses resulted in progeny with shriveled seeds in siliques (left), whereas wild-type seeds are full and smooth (right). Bars = 1 cm (left) and 1 mm (right). B and C, Seeds produced from wild-type plants (B) and mir167a plants pollinated with wild-type pollen (C). Bars = 1 mm. D to M, Embryo development in wild-type plants (D, G, I, and K) and mir167a plants after being pollinated with wild-type pollen (E, F, H, J, L, and M). DAP, Days after pollination. Bars = 50 μm (D–J) and 100 μm (K–M).
In the following days (7–16 d) after pollination, wild-type ovules developed into the cotyledon stage and then developed into mature seeds, while the embryos from the mir167a (♀) × wild type (♂) crosses were arrested at different stages, such as the globular, heart-shaped, or cotyledon stage (Fig. 2, K and L; Supplemental Table S1). Few embryos from the mir167a (♀) × wild type (♂) crosses reached the bent cotyledon stage, but such embryos were noticeably abnormal and were phenotypically distinguishable from those of the wild type by their shorter and asymmetric cotyledons (Fig. 2L). Embryos from mir167a (♀) × wild type (♂) crosses had the mir167a+/− genotype, yet such embryos were not able to develop normally. The fact that both mir167a−/− and mir167a+/− embryos from selfing mir167a+/− plants developed normally indicated that maternal tissues, such as the seed coat, also need MIR167A for normal functions.
MIR167A Acts as a Maternal Gene for Embryogenesis
Reciprocal crosses between wild-type plants and mir167a+/− plants revealed that the mir167a allele transmitted normally from male (102%) and female (79.4%) gametophytes (Table 1; Supplemental Table S2), suggesting that the pollen immaturity and the ovule abortion phenotypes of mir167a plants were mainly due to sporophytic effects. We also analyzed the progeny generated from selfing a mir167a+/− plant. The segregation pattern (wild type:mir167a+/−:mir167a = 36:64:31) also followed Mendelian genetics (Table 1). Selfing mir167a+/− plants or reciprocal crosses between mir167a+/− plants and wild-type plants produced normal seeds (Supplemental Table S2), which could germinate and develop into normal adult plants. A seed is made from the maternal sporophyte, zygote embryo, and endosperm. Endosperm is composed of two-thirds maternal genes and one-third paternal genes. The detailed genotypes of the three components of seeds produced by each cross are shown in Table 1. Embryos could develop normally as long as there was a functional MIR167A copy in the maternal sporophyte, regardless of the genotype of the endosperm and the embryo (crosses 1–8; Table 1). In cross 8, even the mir167a−/− embryo and mir167a−/−/- endosperm displayed normal embryogenesis. However, the mir167a (♀) × wild type (♂) crosses only produced nonfunctional and shriveled seeds (Fig. 2). As shown in cross 9, mir167a+/− heterozygous embryos developed abnormally when they were borne on a mir167a−/− homozygous mother. The only difference between cross 4 (or 7) and cross 9 was the genotype of the maternal sporophyte. These results unambiguously indicated that MIR167A acts as a maternal gene in regulating embryo development. The number of cell layers of the seed coat from mir167a−/− (♀) × wild type (♂) crosses is the same as that of the wild type (Fig. 2, I and J). The most inner layer of the seed coat from the seeds of mir167a (♀) × wild type (♂) crosses seemed less vacuolated and had less intense staining (Fig. 2, I and J). We noticed that the mir167a integuments, which eventually develop into the seed coat, were short and might also contribute to the observed shriveled seed phenotype. Moreover, endosperm cellularization after double fertilization of mir167a plants with wild-type pollen often failed (Fig. 2, I and J), suggesting that the developmental defects of the seeds from mir167a (♀) × wild type (♂) crosses may be caused by a combination of defects in the seed coat and endosperm development.
For each cross group (except for mir167a−/− × Col-0), we genotyped all the germinated seedlings for each (zygote) genotype.
Deletion of MIR167A Leads to Late Flowering
Flowering represents the transition from vegetative growth to reproductive growth, and flowering time is an important trait. We observed that mir167a plants (33 d after germination; n = 10) flowered significantly later than wild-type plants (22 d after germination; n = 18). Under our growth conditions, wild-type plants flowered when they had 12 rosette leaves. However, mir167a plants did not flower until they had at least 17 rosette leaves (Fig. 3, A–C). To better understand the molecular basis of the late-flowering phenotype of the mir167a mutants, we analyzed the expression levels of several flowering time genes by reverse transcription quantitative PCR (RT-qPCR). Among these genes, FLOWERING LOCUS C (FLC), MADS AFFECTING FLOWERING (MAF4), and MAF5, which encode floral repressors, repress floral integrator genes such as FLOWERING LOCUS T (FT) and SUPPRESSOR OF OVEREXPRESSION OF CONSTANS (SOC1) to delay flowering. SQUAMOSA PROMOTER BINDING PROTEIN-LIKE3 (SPL3) and SPL5 not only act downstream of FT but also function in parallel with FT to promote flowering (Wang et al., 2009; Lal et al., 2011). MYB33 activates components of the GA pathway to promote flowering (Gocal et al., 2001; Achard et al., 2004). Consistent with the late-flowering phenotype observed in mir167a plants, the expression levels of FLC, MAF4, and MAF5 were increased in mir167a plants, whereas the expression levels of FT, SOC1, SPL5, SPL3, and MYB33 were reduced in mir167a plants (Fig. 3D).
The late-flowering phenotype of mir167a plants and the expression pattern of MIR167A. A and B, The number of rosette leaves that wild-type (WT; A) and mir167a (B) plants had when they started to flower. C, Quantification of the number of rosette leaves needed before plants started to flower. The values of the wild type and mir167a represent means together with sd (n ≥ 20). The asterisks indicate a statistical difference between wild-type and mir167a plants (**, P < 0.01 by two-sided Student’s t test). D, Relative gene expression levels determined by RT-qPCR analysis. RNA was extracted from 20-d-old leaves. RT-qPCR results were normalized to TUBULIN BETA CHAIN2. Relative expression levels are shown as means together with sd from three biological repeats. E to J, The expression patterns of proMIR167A::GFP in the anther (E–G) and ovule (H–J). In E, an enlarged image (right) shows the expression of GFP in the sporophytic cells of an anther. Bars = 1 cm (A and B), 100 μm (E–G), 20 μm (enlarged image of E), and 50 μm (H–J).
MIR167A Is Expressed during Flower Development
Results from reciprocal crosses between wild-type plants and mir167a+/− plants indicated that the floral defects of mir167a plants were mainly due to sporophytic effects (Supplemental Table S2). We constructed transgenic plants that express the green fluorescent protein (GFP) marker under the control of the MIR167A promoter. Consistent with our genetic results, we detected abundant expression of GFP in the sporophytic cells of both the anther and the ovule (Fig. 3, E–J). The GFP signal was observed in several layers of anther wall cells. This observation was consistent with the dehiscence defects of mir167a anthers. We also detected GFP signals in filaments during pollen development (Fig. 3, E and G). GFP was visible in ovules, but the signal was particularly strong in the epidermis of ovules (Fig. 3, H–J). These expression patterns correlated well with the observed defects in the elongation of ovule integuments in mir167a plants.
Complementation of mir167a Mutants with the Wild-Type MIR167A Gene
To provide genetic evidence that the deletion of MIR167A was responsible for the observed developmental defects, we isolated a second mutant allele, mir167a-c2 (where c stands for CRISPR; Supplemental Fig. S3), that harbored a deletion of 844 bp and a random insertion of 31 bp (Supplemental Fig. S3). The mir167a-c2 allele also lacked the entire region corresponding to the mature miRNA167A. The mir167a-c2 plants displayed the same phenotypes as those of mir167a described above, suggesting that the observed phenotypes of the mir167a plants were caused by deletion of the MIR167A gene. We also carried out complementation experiments. A genomic fragment that included both the 2.4-kb upstream sequence of the MIR167A gene and the primary sequence of MIR167A was used for the complementation experiments. The construct (proMIR167A::MIR167A) was transformed into mir167a+/− plants. Five independent lines of mir167a that contained the proMIR167a::MIR167a transgene displayed normal fertility (Supplemental Fig. S3) and flowered at the same time as the wild type, further demonstrating that the mir167a mutation was responsible for the observed developmental phenotypes.
MIR167A Functions through Regulating ARF6 and ARF8
Two ARFs, ARF6 and ARF8, have been shown to be targets of miRNA167 in several species (Wu et al., 2006; Glazińska et al., 2014; Wang et al., 2015). Overexpression of miRNA167-resistant versions of ARF6 or ARF8 caused pleiotropic phenotypes, arrested growth of ovule integuments, and anther dehiscence defects, which were similar to what we observed in the mir167a mutants (Fig. 1, C, D, G, and H; Wu et al., 2006). We hypothesized that some of the developmental defects in mir167a plants might be caused by ectopic accumulation of ARF6 and ARF8. To test this hypothesis, we crossed the mir167a mutants to arf6 arf8 double mutants. As shown in Figure 4, the anther dehiscence defects of mir167a plants were largely rescued in the arf6, arf6 arf8+/−, and arf6+/− arf8 backgrounds (Fig. 4, D–F). Both arf6 arf8 and mir167a plants were not able to undergo dehiscence (Fig. 4, B and C), but the underlying causes were different. The arf6 arf8 double mutants did not make sufficient jasmonic acid, and the anther dehiscence defects of the arf6 arf8 mutants were rescued by exogenous jasmonic acid applications (Zhang et al., 2017). We observed that the anthers of mir167a arf6 arf8 triple mutants did not dehisce, which was similar to that in the arf6 arf8 double mutant (Fig. 4, B and G).
Mutations in ARF6 and ARF8 partially rescue the defects of anther dehiscence, pollen germination, and sterility of the mir167a mutant. A to F, The anther morphology of wild-type (WT; A), arf6 arf8 double mutants (B), mir167a (C), mir167a arf6 arf8+/− (D), mir167a arf6+/− arf8 (E), and mir167a arf6 (F) plants. The yellow stars indicate the dehiscence of anthers. Note that the anther dehiscence defects of mir167a were largely rescued in mir167a arf6 arf8+/−, mir167a arf6+/− arf8, and mir167a arf6 plants. G, Quantification of the dehiscent anthers. All the anthers from eight opened flowers for each genotype were observed and calculated. The values represent means together with sd. The asterisks indicate statistical differences between different genotypes (**, P < 0.01 by two-sided Student’s t test). For mir167a and arf6 arf8 plants, the percentages of dehiscent anthers were compared with that of the wild type. For mir167a arf6, mir167a arf6 arf8+/−, and mir167a arf8 arf6+/− plants, the percentages of dehiscent anthers were compared with that of the mir167a mutant. H, Pollen germination rates of wild-type, mir167a, and mir167a arf6 arf8+/− plants. The data represent means together with error based on more than 300 pollen grains for each genotype per experiment from three independent experiments; the error bars indicate sd. The asterisks indicate statistical differences between the wild type and mir167a (**, P < 0.01 by two-sided Student’s t test) and mir167a and mir167a arf6 arf8+/ (*, P < 0.05 by two-sided Student’s t test). I, Opened siliques at 7 d after pollination in mir167a (left) and mir167a arf6 arf8+/− (right) plants. Bars = 1 mm (A–F and I).
We conducted in vitro pollen germination assays to determine the viability of the pollen released from mir167a arf6 arf8+/− plants. The germination rate of mir167a was less than 40% (n > 300; Fig. 1E), but it was improved to about 70% (n > 300) in the arf6 arf8+/− background (Fig. 4H). Moreover, we observed that the siliques from self-pollination of mir167a arf6 arf8+/− plants always contained a few seeds (Fig. 4I), which was never observed in the mir167a mutants. Our results indicated that the developmental defects of anther, ovule, and seed maturation in mir167a mutants were likely caused by overexpression of ARF6 and ARF8.
MIR167B, MIR167C, and MIR167D Play Minor Roles in Arabidopsis Development
We used CRISPR/Cas9 technology to generate deletion mutants of MIR167B, MIR167C, and MIR167D (Supplemental Fig. S4). Unlike the mir167a plants, single mutants mir167b, mir167c, and mir167d did not display any obvious developmental defects under our growth conditions. We generated mir167bcd triple mutants, but the triple mutants behaved very similarly to wild-type plants (Fig. 5, A and E).
MIR167B, MIR167C, and MIR167D play minor roles in Arabidopsis reproduction development. A and B, mir167bcd (A) and mir167abcd (B) mutants 43 d after germination. The mir167abcd mutants were sterile, while the development of mir167bcd was similar to that of the wild type. C, Morphology of an adult shoot of mir167abcd pollinated with wild-type pollen. White arrows indicate the pollinated siliques. D and E, The rosette leaves of mir167abcd (D) and mir167bcd (E) when the plants started to flower. F, The numbers of rosette leaves developed when the plants started to flower. The values of mir167bcd and mir167abcd represent means together with sd (n ≥ 20). The asterisks indicate a statistical difference between mir167bcd and mir167abcd (**, P < 0.01 by two-sided Student’s t test). G, The severe defects in the embryo sac of the mir167abcd quadruple mutant. Bars = 5 cm (A–C), 1 cm (D and E), and 50 μm (G).
We analyzed mir167abcd quadruple mutants to determine whether MIR167B, MIR167C, and MIR167D have overlapping functions with MIR167A. Similar to mir167a plants, the mir167abcd quadruple mutants flowered later (33 d after germination; n = 10) and were sterile (Fig. 5, B, D, and F). Compared with the mir167a single mutants, the ovules of mir167abcd quadruple mutants displayed more severe defects. All of the embryo sacs we observed were abnormal, although sometimes one or two nuclei could be observed in some embryo sacs (Fig. 5G; n = 30). In addition to the cell differential defects of the embryo sac, the outer integuments in mir167abcd plants were much shorter than those of mir167a mutants (Figs. 1, G and H, and 5G). When pollinated with wild-type pollen, mir167abcd gynecia did not elongate much compared with mir167a or wild-type plants (silique lengths were as follows: the wild type, 18–20 mm; mir167a pollinated with wild-type pollen, 16–18 mm; and mir167abcd pollinated with wild-type pollen, 15–16 mm [n > 20]; Figs. 2A and 5C). These results suggested that MIR167B, MIR167C, and MIR167D indeed have overlapping functions with MIR167A in regulating ovule and seed development.
DISCUSSION
In this study, we uncovered several functions of MIR167s by analyzing the loss-of-function mutants in Arabidopsis. MIR167s regulate anther dehiscence, ovule development, and fertility, as plants that overexpress MIR167A or mARF6 and mARF8 are defective in those processes (Wu et al., 2006). Our mir167a mutants largely phenocopied the mARF overexpression plants in those processes (Wu et al., 2006). Moreover, we show that MIR167A is critical for flowering time control, a process that was not observed in the mARF overexpression lines.
An unexpected finding was that MIR167A has a true maternal effect. The first reported maternal gene required for embryo development in Arabidopsis is DICER-LIKE 1 (DCL1)/SHORT INTEGUMENT/SUSPENSOR 1/CARPEL FACTORY (Ray et al., 1996). DCL1 is a primary enzyme involved in miRNA biogenesis in plants (Kurihara and Watanabe, 2004; Fang and Spector, 2007). SERRATE, one of the accessory proteins of DCL1 required for miRNA biogenesis, was also reported to act as a maternal effector for embryo development (Prigge and Wagner, 2001). Although about 100 families of miRNAs have been identified in Arabidopsis (Voinnet, 2009), it was not known which miRNA acts downstream of DCL1 in controlling the maternal sporophytic effect for embryogenesis. In this study, we found that MIR167A acts as a maternal gene for embryo development. It has been reported that DCL1 is responsible for MIR167 processing (Dong et al., 2008; Liu et al., 2012). Combined with the similar phenotypes and the discovered function of DCL1 in miRNA processing, it is likely that miRNA167A is the key DCL1-processed miRNA involved in controlling embryo development from maternal sporophytic tissues. Interestingly, the defects in embryogenesis and seed development in mir167a mutants are 100% penetrant, whereas dcl1 mutants were not 100% penetrant. The difference in penetrance may be due to the nonnull nature of dcl1 mutations. Alternatively, the difference may suggest that other DCL proteins may also be involved in processing miRNA167 precursors. Nevertheless, the mir167a mutants described here provide a valuable genetic material for studying maternal control.
The development of the maternal seed integument, the embryo, and the endosperm needs to be properly coordinated (Ingram, 2010). Fertilization of the central cell leads to the production of auxin, which is subsequently transported to the maternal tissues to drive seed coat development (Figueiredo et al., 2016). Recently, it has also been reported that auxin biosynthesis increased in the integuments at early embryogenesis and that this maternally produced auxin is required for embryo development (Robert et al., 2018). Without miRNA167 in the integuments, it is expected that ARF6 and ARF8 mRNAs will overaccumulate, causing an imbalance of auxin signaling and disruption of seed development. In addition to the influence from maternal tissues, MIR167A and MIR167B were also detected in the globular stage embryos (Armenta-Medina et al., 2017), suggesting that miRNA167 may also regulate the auxin response in embryos and contribute to normal embryo development. It is intriguing to explore whether miRNA167 or auxin or both actually move from maternal cells to embryos and endosperms, given that both miRNAs and auxin are mobile. The mir167a mutants reported here will be a valuable resource for studying auxin signaling in seed development as well as for studying how the development of the seed coat, embryo, and endosperm is coordinated.
It is interesting that MIR167A plays a more predominant role in Arabidopsis reproductive development than the other three MIR167 genes. The phenotypic differences among the mir167 mutants could result from their different expression patterns. We observed significant expression of MIR167A in the anther wall and ovule epidermis (Fig. 3), whereas the other three MIR167 genes did not show strong expression in those cells (Wu et al., 2006). Differences in regulatory elements in the primary MIR167 sequences, which may impact transcription efficiency or affect the maturation of miRNAs, may also partially account for the observed phenotypic differences in mir167 mutants. Consistent with this hypothesis, it was reported that overexpression of MIR167A, but not the other three MIR167 genes, under the control of the same Cauliflower mosaic virus 35S promoter caused strong phenotypes similar to those in arf6 arf8 double mutants (Wu et al., 2006).
MiRNA167 has at least three target genes in Arabidopsis: ARF6, ARF8, and IAR3. In rapeseed (Brassica napus), the NRAMP1b gene, which encodes a metal transporter, is also a target for miRNA167 (Meng et al., 2017). Our results suggest that ARF6 and ARF8 are the major targets of miRNA167 in anther and ovule development. The phenotypes of mir167a mutants are very similar to those of mARF6/8 overexpression plants. Both mir167a mutants and mARF plants displayed indehiscent anthers and shorter integuments (Fig. 1; Wu et al., 2006). The anther indehiscent and sterility phenotypes were partially rescued in mir167a arf6 arf8+/− and mir167a arf8 arf6+/− mutants (Fig. 4), suggesting that miRNA167s mainly function through the ARFs. It is clear that the expression levels of ARF6 and ARF8 should be tightly regulated. Too much (as in mir167a plants) or too little (as in arf6 arf8 mutants) expression of ARF6/8 will lead to defects in anther dehiscence.
MATERIALS AND METHODS
Plant Materials and Growth Conditions
Arabidopsis (Arabidopsis thaliana) plants used in this study were in the Col-0 genetic background. The mir167 mutants were generated using CRISPR/Cas9 gene-editing technology as described previously (Gao and Zhao, 2014; Gao et al., 2015, 2016). The two mir167a alleles harbor 1,170- and 844-bp deletions, respectively (Supplemental Fig. S3). The mir167a mutants were genotyped by PCR using the 167A-Genotyping 1 (GT1) and 167A-GT2 primer pair (Supplemental Table S3), which amplify about a 1.6-kb fragment from the wild type and a much smaller fragment from the mir167a mutants. The mir167b mutant had a 612-bp deletion, which completely removed the MIR167B coding region (Supplemental Fig. S4). The mir167b mutant was genotyped using three primers, 167B-GT1, 167B-GT2, and 167B-GT3. The 167B-GT1 and 167B-GT2 pair generates 1.6- and 1-kb fragments for the wild type and the mutant, respectively. To further clarify the homozygosity of the mir167b mutants, we used 167B-GT1 and 167B-GT3, which only amplify a fragment from the wild type. The mir167c mutant contained a 697-bp deletion (Supplemental Fig. S4), which was genotyped using 167C-GT1, 167C-GT2, and 167C-GT3 primers. The 167C-GT1 and 167C-GT2 pair produce 2- and 1.3-kb fragments for the wild type and the mir167c mutant, respectively. The 167C-GT1 and 167C-GT3 pair only amply the wild type for about a 1.4-kb fragment. The mir167d mutant had a 1.6-kb deletion (Supplemental Fig. S4). We used 167D-GT1 and 167D-GT2 to determine whether a plant contained the mutation. We used 167D-GT2 and 167D-GT3 to determine the zygosity of the mir167d mutant. All of the genotyping primers are listed in Supplemental Table S3.
Plants were grown at 22°C under long-day conditions (16 h of light/8 h of dark).
Plasmid Construction and Complementation of the mir167a Mutant
We amplified an ∼1.5-kb upstream sequence from genomic DNA of Col-0 into pEasy-Blunt for sequence verification (see Supplemental Table S3 for ProMIR167A-F and ProMIR167A-R primers) and then subcloned it into the modified vector pCAMBIA1300-GFP to form the proMIR167a::GFP construct. The 3.8-kb MIR167A genomic sequence was amplified and recombined into pCAMBIA1300-nos to form the proMIR167A::MIR167A constructs (see Supplemental Table S3 for MIR167A-F and MIR167A-R primers). The proMIR167A::GFP and proMIR167A::MIR167A plasmids were transformed into Col-0 and mir167a+/− plants by the floral dip method (Clough and Bent, 1998). The T1 seeds were grown on one-half-strength Murashige and Skoog medium containing 15 µg mL−1 hygromycin for selecting transgenic plants. For proMIR167A::MIR167A transgenic plants, we then genotyped the transgenic plants for the mir167a mutation using 167A-CGT3/CGT2 (see Supplemental Table S3 for 167A-CGT3 and 167A-CGT2). For proMIR167A::GFP transgenic plants, a LSM 5 Pascal confocal laser scanning microscope (Zeiss)was used to visualize the GFP expression pattern. Anthers and ovules at each stage were dissected from flower buds. The fluorescence was excited at 488 nm and collected at 515 to 530 nm.
Pollen Germination in Vitro and in Pistils
For in vitro pollen germination, open flowers were dehydrated for 2 h at room temperature. Anthers were dissected and pollen grains were released onto agar medium by dipping the anthers on the surface of agar plates, which contained 18% (w/v) Suc, 0.01% (w/v) boric acid, 1 mm MgSO4, 1 mm CaCl2, 1 mm Ca(NO3)2, and 0.5% (w/v) agar. Six hours after germination at 24°C, pollen grains were examined and photographed using an Olympus BX51 digital microscope. For the pollen germination ratio, more than 300 pollen grains of each genotype were analyzed with a light microscope. This experiment was carried out three times. The mean values and sd were calculated based on these three repeats.
For pollen germination in pistils, wild-type pollen was hand pollinated onto the pistils of Col-0 and mir167 mutants in the afternoon of the first day. In the morning of the second day (18–19 h after pollination), the pollinated pistils were collected and fixed in Carnoy’s solution (ethanol:acetic acid = 3:1) for 2 to 3 h. After rinsing in 0.01 mol L−1 phosphate-buffered saline four times (each for 5 min), 7 m NaOH was used to soften the pistils for 4 h. After rinsing in 0.01 mol L−1 phosphate-buffered saline four times (each for 5 min), the pistils were incubated in 0.1% (w/v) Aniline Blue for 30 min. Finally, we used a UV channel to monitor the pollen tube growth and took photographs using an Olympus BX51 digital microscope.
Late-Flowering Phenotype Analysis
The days to flowering after germination were calculated from more than 10 plants of each genotype (the wild type, mir167a, mir167bcd, and mir167abcd). For determining the rosette leaf numbers at bolting of wild-type, mir167a, mir167bcd, and mir167abcd plants, at least 20 plants for each genotype were scored.
RNA Extraction and qPCR
Twenty-day-old leaves of the wild type and mir167a mutants were used to detect the transcripts of flowering regulatory genes. For analyzing the transcriptional level of MIR167 genes, inflorescences were collected from wild-type, mir167a, and mir167bcd plants. Arabidopsis leaves or inflorescences were used to extract total RNA via TRIzol Reagent (Invitrogen). Total RNA was then used for first-strand cDNA synthesis according to the manufacturer’s instructions (TransGen Biotech). qPCR was performed using an ABI PRISM 7300 detection system (Applied Biosystems) with the SYBR Green Realtime PCR Master Mix (Toyobo). The relative expression level of each gene was normalized to TUBULIN BETA CHAIN2 and averaged over three biological repeats. The relevant primer sequences are listed in Supplemental Table S3.
Microscopic Analysis
The protocol used to analyze female gametophyte development was modified from the previously reported protocol (Christensen et al., 1997). Briefly, pistils were fixed in a solution of 4% (v/v) glutaraldehyde in 12.5 mm sodium cacodylate buffer (pH 6.9) for 2 h at room temperature. Next, the pistils were dehydrated in a series of ethanol (v/v) solutions (10%, 20%, 40%, 60%, 80%, and 95%), each for 10 min. The pistils were incubated in 95% (v/v) ethanol solution overnight. The next day, the pistils were dehydrated in 100% ethanol two times, each for 10 min. Then the tissues were cleared in the benzyl benzoate:benzyl alcohol (2:1) solution for 20 min. Ovules were dissected and mounted in immersion oil and examined using a Carl Zeiss LSM5 Pascal confocal laser scanning microscope. Excitation wavelength was 488 nm, and the emission wavelength was 515 to 530 nm.
The modified pseudo-Schiff propidium iodide staining was performed as described previously to observe the development of embryos (Truernit et al., 2008). Briefly, the tips of pistils were cut and fixed in fixative solution (50% [v/v] methanol and 10% [v/v] acetic acid) for 12 h at 4°C. Pistils were transferred to 80% (v/v) ethanol solution for 5 min at 80°C. The materials were then put back into the fixative solution for 1 h at 4°C. The pistils were retained in a solution of 1% SDS and 0.2 n NaOH overnight. After washing, the pistils were incubated in bleach solution (2.5% NaClO), rinsed with water, and then incubated in 1% periodic acid for 40 min. After another wash, the pistils were transferred into Schiff reagent with propidium iodide for 2 h (100 mm sodium metabisulfite and 0.15 n HCl; PI concentration was 100 mg mL−1). The pistils were then cleared in chloral hydrate solution (4 g of chloral hydrate, 1 mL of glycerol, and 2 mL of water) overnight at room temperature. After being kept in Hoyer’s solution for 3 d, pistils were observed using a Carl Zeiss LSM5 Pascal confocal laser scanning microscope. Excitation wavelength was 488 nm, and the emission wavelength was 515 to 530 nm.
For the observation of endosperm cellularization in embryos, Col-0 and mir167a seeds (5 d after fertilization) were fixed and embedded as described (Zhang et al., 2007). Semithin sections (1 μm) of seeds were stained with Toluidine Blue and photographed using an Olympus BX51 digital microscope.
Accession Numbers
Sequence data from this article can be found in the GenBank/European Molecular Biology Laboratory data libraries under accession numbers MIR167A, AT3G22886; MIR167B, AT3G63375; MIR167C, AT3G04765; MIR167D, AT1G31173; ARF6, AT1G30330; ARF8, AT5G37020; FT, AT1G65480; SOC1, AT2G45660; SPL5, AT3G15270; SPL3, AT2G33810; MYB33, AT5G06100; CO, AT5G15840; FLC, AT5G10140; MAF4, AT5G65070; and MAF5, AT5G65080.
Supplemental Data
The following supplemental materials are available.
Supplemental Figure S1. Pollen mitosis I and II can be carried out in mir167a mutants.
Supplemental Figure S2. Wild-type pollen geminated and grew normally on the pistils of mir167 mutants.
Supplemental Figure S3. Genetic complementation of the mir167a mutants.
Supplemental Figure S4. The CRISPR mutants of MIR167B, MIR167C, and MIR167D genes.
Supplemental Table S1. Various embryo developmental defects in mir167a plants pollinated with wild-type pollen.
Supplemental Table S2. The transmission efficiency of male and female gametophytes of mir167a.
Supplemental Table S3. Primers used in this study.
Footnotes
↵1 This work was supported by the National Institute of General Medical Sciences (R01GM114660 to Y.Z.), the National Natural Science Foundation of China (31401030 to X.Y.), and the Science and Technology Commission of Shanghai Municipality (18DZ2260500).
↵2 These authors contributed equally to the article.
↵[OPEN] Articles can be viewed without a subscription.
- Received February 1, 2019.
- Accepted March 5, 2019.
- Published March 13, 2019.