The Arabidopsis U–box/ARM repeat E3 ligase AtPUB4 influences growth and degeneration of tapetal cells, and its mutation leads to conditional male sterility


For correspondence (e-mail or


Pollen formation is a complex developmental process that has been extensively investigated to unravel underlying fundamental developmental mechanisms and for genetic manipulation of the male-sterility trait for hybrid crop production. Here we describe identification of AtPUB4, a U–box/ARM repeat-containing E3 ubiquitin ligase, as a novel player in male fertility in Arabidopsis. Loss of AtPUB4 function causes hypertrophic growth of the tapetum layer. The Atpub4 mutation also leads to incomplete degeneration of the tapetal cells and strikingly abnormal exine structures of pollen grains. As a result, although the Atpub4 mutant produces viable pollen, the pollen grains adhere to each other and to the remnants of incompletely degenerated tapetal cells, and do not properly disperse from dehisced anthers for successful pollination. We found that the male-sterility phenotype caused by the Atpub4 mutation is temperature-dependent: the mutant plants are sterile when grown at 22°C but are partially fertile at 16°C. Our study also indicates that the AtPUB4-mediated pathway acts in parallel with the brassinosteroid pathway in controlling developmental fates of the tapetal cells to ensure male fertility.


The life cycle of plants alternates between a diploid sporophyte generation and a haploid gametophyte generation. Animals generally form germ cells early in embryo development. However, in angiosperms such as Arabidopsis, the stem cells in the apical meristem strictly produce somatic organs for a long period to give rise to all vegetative organs before the shoot apical meristem undergoes floral transition to become an inflorescence meristem that then generates floral organs (Weigel and Jurgens, 2002; Walbot and Evans, 2003) . Floral organogenesis, germline specification, gametogenesis and subsequent fertilization involve important cellular processes that are coordinately regulated by various environmental and developmental signals and through complex cell–cell interactions (Yang and Sundaresan, 2000; Sieber et al., 2004; Yadegari and Drews, 2004; Ma, 2005).

An Arabidopsis flower has four basic organs, with stamens (the male reproductive organ) and the gynoecium (the female reproductive organ) wrapped in whorls of petals and sepals (Weigel and Meyerowitz, 1994). The stamens carry the pollen-bearing part, called the anther. An anther forms four lobes that contain a group of diploid microsporocytes or meiocytes. Each microsporocyte undergoes meiosis to give rise to four haploid microspores, each of which undergoes two mitotic divisions to produce a male gametophyte (pollen) with a vegetative cell and two sperm cells (McCormick, 2004; Ma, 2005). At the final stages of anther development, the septum and the stomium (somatic cells in the anther) undergo programmed cell death, leading to anther dehiscence and pollen release (Scott et al., 2004).

Development of functional pollen grains and their proper release are regulated through complex genetic networks. A large number of genes involved in these processes have been identified in the last decade, largely through molecular genetics approaches (McCormick, 2004; Ma, 2005; Wilson and Zhang, 2009). Studies have also revealed the importance of proper functioning of the tapetal cells in pollen development. The tapetum is the innermost layer of somatic tissues in the anther that surrounds developing pollen in anther locules. Tapetal cells produce and secrete various compounds that are either essential for pollen development or become components of the pollen outer wall, termed the exine (Ariizumi and Toriyama, 2011). The tapetal cells undergo programmed cell death at late stages of anther development. Many male-sterility mutants have been found to be associated with various defects in tapetum development, including lack of the tapetum layer and premature or delayed programmed cell death of tapetal cells (McCormick, 2004; Ma, 2005; Wilson and Zhang, 2009). Studying pollen development is not only important for addressing fundamental questions in developmental biology, but also for utilization of the male-sterility trait for hybrid crop production and for controlling transgene flow from genetically modified plants.

Regulation of protein activities via the ubiquitin-mediated protein degradation pathway is a common mechanism that modulates a wide range of biological processes (Moon et al., 2004; Zeng et al., 2006). In this pathway, ubiquitin units are attached to a targeted protein (ubiquitination). The ubiquitin tag directs the protein to the 26S proteasome for degradation, or changes protein conformation and activity (Vierstra, 2003). An ubiquitin–protein ligase (E3 ligase) is responsible for targeting a specific substrate by bringing the substrate and an E2 ubiquitin-conjugating enzyme into close proximity to catalyze the substrate's ubiquitination. The E3 ligases comprise a large number of proteins or protein complexes, most of which contain either a HECT domain or a RING/U–box domain (Vierstra, 2003; Moon et al., 2004).

The PUB (plant U–box) proteins belong to a family of U box-containing E3 ligases in plants (Azevedo et al., 2001). Among 64 predicted PUB proteins in Arabidopsis, 41 contain Armadillo (ARM) repeats (Mudgil et al., 2004; Samuel et al., 2006). The ARM repeats, first identified in the Drosophila segment polarity protein Armadillo, form a right-handed superhelix and mediate protein–protein interaction (Coates, 2003; Samuel et al., 2006). Known PUB/ARM proteins include ARM REPEAT CONTAINING 1, which is an S–locus receptor kinase-binding protein that is required for self-incompatibility in Brassica (Gu et al., 1998), PHOTOPERIOD-RESPONSIVE 1 from Solanum tuberosum (potato) (Amador et al., 2001), SPOTTED LEAF 11, which acts as a negative regulator of cell death and defense in Oryza sativa (rice) (Zeng et al., 2004), Arabidopsis AtPUB17 and its Nicotiana tabacum (tobacco) homolog ACRE276 (Avr9/Cf–9 rapidly elicited 276 ) both of which are positive regulators of cell death and defense (Yang et al., 2006), AtPUB12 and AtPUB13, which target the Arabidopsis pattern-recognition receptor FLS2 for degradation to attenuate the flagellin/FLS2-trigged immune response (Lu et al., 2011a), and PUB1 of Medicago truncatula, which is a negative regulator of nodulation meditated by LysM domain-containing receptor-like kinases 3 (LYK3) (Mbengue et al., 2010).

In this paper, we identify the Arabidopsis U–box/ARM protein AtPUB4 as a novel player in male fertility. The Atpub4 mutation causes hypertrophy and delayed degeneration of the tapetal cells. Atpub4 pollen grains were found to be viable; however, they were not properly released from anthers because of a defect in the tapetum layer. The male-sterility trait of Atpub4 is temperature-dependent, and is alleviated when the mutant plants are grown under a relatively lower temperature.


Disruption of AtPUB4 leads to male sterility

To investigate the biological function of the AtPUB4 (At2 g23140) gene, three putative T–DNA insertion lines of AtPUB4 (SALK_108269, SALK_054373 and SAIL_859H05) were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). Their T–DNA insertion sites were verified by PCR analysis and sequencing of the T–DNA flanking regions (Figure 1a). SALK108269 was previously found to contain another T–DNA insertion at At3 g06430 (AtPPR2) (Lu et al., 2011b). Progeny containing only the insertion at AtPUB4 but not at AtPPR2 were used in this study. Plants that are homozygous for the T–DNA insertion alleles were obtained from the other two lines. The insertion alleles of SALK_108269, SALK_054373 and SAIL_859H05 are referred to as pub4–1, pub4–2 and pub4–3, respectively, in this paper. The wild-type (WT) AtPUB4 allele is referred to as PUB4.

Figure 1.

Schematic representation of the AtPUB4 gene and the AtPUB4 protein. (a) Schematic representation of AtPUB4 and three T–DNA insertion alleles. Open boxes, filled boxes and lines represent untranslated regions (UTR), exons and introns, respectively. The T–DNA insertion sites in the three insertion lines are indicated. P1 and P2 represent the relative positions of the primers used in quantitative RT–PCR analysis to detect AtPUB4 transcripts. (b) Schematic representation of the U–box domain and the ARM repeats in AtPUB4. (c) Quantitative RT–PCR analysis of the AtPUB4 expression levels in the WT and three insertion lines using primers P1 and P2.

PUB4 encodes an 829 amino acid protein. The region between amino acids 572 and 775 is predicted to contain five ARM repeats (Figure 1b). The U–box domain is located in the region between amino acids 233 and 296. For the pub4–1 and pub4–3 alleles, the T–DNA insertion is in the 4th exon, whereas the insertion in pub4–2 is in the 2nd intron. Quantitative RT–PCR analysis using the PUB4 gene-specific primer pair P1 and P2 (Figure 1a and Table S1) did not detect the PUB4 transcript in any of these three lines (Figure 1c), indicating that all three lines are either knockout or severe knockdown lines.

The three pub4 lines were further analyzed for any discernible phenotypes. Plants were grown in growth rooms or growth chambers with standard growth conditions for Arabidopsis: 22°C, 50% humidity and a 15 h light/9 h dark photoperiod. The pub4–1 mutant produced slightly smaller rosettes, with narrower and shorter leaves than WT plants (Col–0 ecotype) (Figure 2a,b). The mutant plants bolted 1-5 days later, and their flowers had shorter petals than those of WT plants (Figure 2b–d and Figure S1). A more striking difference between the mutant and WT plants was displayed at the seed setting stage: the mutant plants were sterile as they did not produce seeds (Figure 2e). To determine whether the infertility is due to a male or female defect, we performed reciprocal crosses between WT and pub4–1 plants. Hand pollination of pub4–1 flowers using pollen from WT plants resulted in normal seed setting; however, when WT flowers were pollinated with pollen from pub4–1 plants, few seeds were produced. The results indicate that the pub4–1 mutation causes male sterility.

Figure 2.

The pub4 mutation leads to a smaller rosette size and sterility. (a) The pub4–1 mutant (pub4) showed narrower leaves and smaller rosette sizes than WT plants. (b) pub4–1 displayed a delay in flowering time. The plants were grown under a 15 h light/9 h dark photoperiod. (c, d) WT (c) and pub4–1 (d) inflorescences. Note the shorter petals in the pub4–1 flowers than in WT. (e) The pub4–1 plant displayed sterility. The shorter statue of the pub4 plant is due to a residual effect of later flowering. The plant on right is a pub4–1/pub4–1 plant carrying the AtPUB4 trangene (pub4Compl), which led to complementation of the pub4–1 phenotype. Scale bars = 1 cm.

The morphological and male-sterility phenotypes were associated with all three pub4 lines, strongly indicating that those phenotypes are caused by disruption of PUB4 by the T_DNA insertions. For further verification through genetic complementation, a 4.6 kb genomic DNA fragment containing the PUB4 gene and a 1.2 kb region upstream of its start codon was cloned from WT Arabidopsis genomic DNA by PCR and introduced into pub4–1 plants. Plants that were homozygous for the pub4–1 allele but carried the PUB4 transgene were obtained from 40 independent transgenic events, and 39 of these lines were indistinguishable from WT plants in terms of morphological and the fertility traits (Figure 2e), confirming that the smaller rosette size and the male-sterility phenotypes associated with the insertion lines do indeed result from disruption of PUB4. Further experiments were performed using the pub4–1 allele, and the results for the pub4 phenotype presented in this paper are from the pub4–1 mutant (abbreviated as pub4).

pub4 produces viable pollen grains

Among the progeny derived from self-pollinated heterozygous PUB4/pub4 plants that were examined, three-quarters (280 of 372 progeny) displayed a WT phenotype and one-quarter displayed the pub4 phenotype. PCR analysis further showed that all plants with the pub4 phenotype were homozygous for the pub4 insertion allele, whereas plants with a WT phenotype were either heterozygous or homozygous for the WT allele. In addition, we did not find any defect in male or female fertility or seed development in the heterozygous plants. These results indicate that the pub4 pollen produced in the heterozygous plants is fully functional and transmitted at a rate equal to that of WT pollen . Therefore, the male-sterility phenotype of pub4 is sporophytically determined by the recessive pub4 mutation.

To assess the defects in pollen grains produced by pub4 plants, anthers and pollen from pub4 and WT plants were examined with light microscopy and scanning electron microscopy (SEM). Before anther dehiscence, no obvious difference was observed on the anther surface between WT and pub4 (Figure 3a,b). Dehisced anthers from mature WT flowers had individual pollen grains visible on the surface (Figure 3c,e,g); however, pollen grains on the dehisced mutant anthers appeared to be compacted and stick together (Figure 3d,f,h). The mutant pollen grains also displayed abnormal spherical morphology instead of the normal oblong shape. In addition, the pub4 pollen grains did not display the regular surface structure and lacked visible furrows compared to WT pollen (Figure 3e,f).

Figure 3.

The pub4 mutation alters the morphology of pollen grains and affects pollen release. (a, b) SEM images of anther surfaces from WT (a) and pub4 (b) plants before anther dehiscence. (c, d) SEM images of dehisced anthers from WT (c) and pub4 (d) plants. (e, f) Close-up views showing abnormal morphology of the pub4 pollen grains (f) compared to WT pollen (e). (g, h) Freed pollen grains were visible on the surface of dehisced WT anthers (g), but pollen grains in the dehisced pub4 anther adhered to each other (h). Scale bars = 20 μm (a–f) and 200 μm (g, h).

To determine whether pub4 pollen is viable, Alexander staining (Alexander, 1969) was used to stain pollen grains from WT and pub4 plants. Both WT and pub4 pollen stained positive, as indicated by their purple color (Figure 4a). Figure 4(a) also shows that pollen from WT had a clean surface whereas cell residue was attached to pollen from the mutant after the pollen grains were squeezed out of the anthers. DAPI (4,6–diamidino-2–phenylindole) staining was then used to examine mutant and WT pollen at various developmental stages. The pub4 pollen was indistinguishable from WT pollen in terms of progression of mitotic division during microgametogenesis (Figure 4b,c). Like WT pollen, mature pub4 pollen contained two generative cells (sperm) (Figure 4c).

Figure 4.

The pub4 mutant produces viable pollen grains. (a) Both WT and pub4 pollen grains were stained purple by Alexander staining, an indication of viable pollen. Left panels: whole anthers stained with the Alexander solution; right panels: stained pollen grains squeezed out of anthers. (b) DAPI staining (left panels) and light microscopy images (right panels) for WT and pub4 pollen at the tetrad stage. (c) DAPI staining (left panels) and light microscopy images (right panels) for WT and pub4 pollen at the mature stage. (d) Normal seed setting of a silique (arrow) in a pub4 plant resulting from hand pollination of its stigma by an anther from the same flower that was mechanically broken to assist pollen release. (e) Pollen tube elongation revealed by aniline blue staining of WT and pub4 pistils following hand pollination. The images were taken 24 h after pollination. Scale bars = 20 μm (a–c), 1 mm (d) and 1 cm (e).

The above results indicate that pub4 pollen is viable, which prompted us to further determine whether pub4 pollen is capable of fertilizing ovules. We mechanically tore apart anthers from mature pub4 flowers and used them to pollinate stigmas of both WT and pub4 flowers by brushing against stigmas multiple times. The hand pollination approach led to nearly normal seed setting (Figure 4d), indicating that pub4 pollen is functional if it can be dispersed from anthers and thus given a chance to pollinate . Such a conclusion is further supported by observations of pollen germination and pollen tube elongation after hand pollination. Pistils hand-pollinated by WT and pub4 pollen grains were stained with aniline blue, which stains the callose in pollen tubes. As shown in Figure 4(e), pub4 pollen, like WT pollen, was able to germinate on the stigma, and pollen tubes reached to the farthest end of the ovule 24 h after hand pollination by both WT and pub4 pollen (Figure 4e). Taken together, our results show that the pub4 mutation causes inability of the pollen grains to undergo the normal pollination process because of the anther's failure to release pollen grains.

The pub4 mutation causes enlargement and delayed degeneration of tapetal cells and abnormal exine formation on pollen

To further determine the defect undermining normal functionality of anthers and pollen in pub4, mutant and WT anthers at various developmental stages (as defined by Sanders et al., 1999) were embedded with paraffin, sectioned, and analyzed by light microscopy. We also examined their ultrastructure in thin sections of ultra-rapidly frozen/freeze-substituted anthers. No obvious difference between the pub4 and WT anthers was found before stage 7 (the tetrad stage) (Figure 5a). At stage 8–9, the pub4 tapetal cells were abnormally enlarged compared to WT (Figure 5a–e). The tapetum is generally single-layered and pollen is scattered in WT anthers (Figure 5a,b); however, in pub4, some regions of the tapetum were multi-layered (Figure 5c and Figure S2) and pollen grains were tightly packed (Figure 5a,c). In addition, the tapetal cells in WT had more electron-dense tapetosome inclusions, visualized as dark spots (Figure 5d, arrows), than in pub4 (Figure 5e). The tapetal cells normally start to degenerate at stage 10–11, and disappear after stage 12 (Figure 5a,f); however, their degeneration was delayed and incomplete in the pub4 mutant (Figure 5a,g). Instead of randomly scattered pollen grains in the locules in WT, pollen grains in pub4 adhered to remnants of incompletely degenerated tapetum along the locule walls (Figure 5a,f,g).

Figure 5.

The pub4 mutation leads to abnormal expansion and incomplete degeneration of tapetal cells and abnormal exine formation. (a–g) Wax sections (a) and TEM images (b–g) of WT and pub4 anthers from stage 7 to stage 13 flowers. Abnormal expansion of the tapetum layer of pub4 anthers became apparent after stage 8–9 (a, c, e) compared with WT anthers (a, b, d). The tapetal cells were completely degenerated in WT after stage 12–13 (a, f); however, residue from degenerating tapetal cells remained at stage 12–13 in pub4 (a, g). Fully formed pollen was scattered in the locules in the WT anther (a, b, f). In the pub4 locules, pollen was tightly packed (a, c) and adhered to residue of incompletely degenerated tapetal cells (g). The arrows in (d) indicate electron-dense tapetosome inclusions. (h–j) TEM images showing close-up views of pollen outer wall structures, including baculae and tectum, in WT (h) and pub4 (i, j). The arrows in (i) and (j) indicate a sporopollenin droplet. (k, l) TEM images showing ectopic formation of sporopollenin droplets/exine-like structures (indicated by arrows and arrowheads ) on the surface of tapetal cells in pub4. The structures indicated by the arrows appear to be aggregates of sporopollenin droplets exocytosed by the tapetal cells. EN, endothecium; ML, middle cell layer; T, tapetal cell; P, pollen: Ba, bacula; Te, tectum; EP, epidermis. Scale bars = 20 μm (a), 10 μm (b–g), 0.5 μm (h, i) and 2 μm (j–l).

The pollen wall may be divided into two regions: the inner intine and outer exine. The exine includes the inner nexine and the outer sexine. Sexine has a three-dimensional structure comprising many columns termed baculae and a roof-like tectum (Figure 5h) that gives a reticulate appearance to the pollen surface. The exine structure on pub4 pollen was less electron-dense, and its structure was strikingly atypical and variable (Figure 5i,j). Some pollen had shorter and wider baculae and no obvious tectum structure (Figure 5i,j). In some cases, the thick bacula was topped with tectum and also spherical ‘droplets’ of exine (Figure 5i, arrows), probably derived from extracellular sporopollenin droplets secreted from the tapetum (Figure 5k,l). Exine deposition was relatively uniform on the WT pollen grains (Figure 5d), but was non-symmetrical on the pub4 pollen grains, with more exine deposition on the site bordered with tapetum (Figure 5e). Sporopollenin droplets formed not only on the pollen wall but also on the cell surface of the tapetal cells in pub4 anthers (Figure 5k, arrowheads and arrows , and Figure S3). In particular, the regions of tapetal cells proximal to pollen had aggregates of such sporopollenin droplets (Figure 5k,l, arrows , and Figure S3).

Expression patterns of PUB4

Quantitative RT–PCR results showed that PUB4 was expressed in all tissues examined, with the highest transcript level in flowers and flower buds (Figure 6a). We created a reporter gene construct in which the GUS reporter gene is under the control of the 1.2 kb promoter region of the PUB4 gene. Transgenic lines expressing the reporter gene were examined for GUS activity. Ten of twelve independent GUS transgenic lines displayed a similar GUS straining pattern. In young seedlings, GUS activity was detected in all organs (Figure 6b,j). Roots of young seedlings displayed strong GUS activity (Figure 6b), but GUS activity in older roots was concentrated more at the root tips (Figure 6f). In leaves, vascular tissues and guard cells exhibited stronger staining than other leaf cells (Figure 6c,d). Strong GUS staining was observed in trichomes of expanding leaves (Figure 6e). However, the pub4 mutation was not found to affect trichome development (Figure S4). In floral buds, anthers showed the highest GUS activity, although developing pistils were also stained (Figure 6g,h). Observation of wax-embedded sections of the anthers revealed that GUS activity was concentrated mainly in the tapetal cells, although the reporter gene was also expressed weakly in pollen and other anther tissues (Figure 6i).

Figure 6.

Expression patterns of PUB4. (a) Quantitative RT–PCR analysis of PUB4 transcript levels in various tissues of Arabidopsis plants. Se, whole seedlings; Rt, roots; St, stems; Le, leaves; Flo, flowers; Sili, siliques. (b–j) Histochemical localization of GUS activity in PUB4pro::GUS transgenic Arabidopsis plants. GUS staining is shown in a 10-day-old seedling (b), a 20-day-old seedling (c), a portion of an expanded leaf showing guard cells with high GUS activity (d), a portion of a young leaf showing trichomes with strong GUS staining (e), a root tip of a 10-day-old seedling (f), a gynoecium (g), an inflorescence (h) and a silique 6 days after pollination (j). In anthers, strong GUS activity was detected in the tapetal cells as seen in wax-embedded sections of the anther (i). Scale bars = 0.5 cm (b, c, h, j), 50 μm (d–f, i) and 1 mm (g).

To determine whether the lack of a functional PUB4 in tapetal cells is responsible for the male-sterility phenotype in pub4, we placed the coding region of the PUB4 gene under the control of the tapetum-specific promoter from the Arabidopsis A9 gene (Paul et al., 1992). The A9pro;PUB4 construct was introduced into pub4–1. Among 18 independent transgenic A9pro;PUB4/pub4 lines obtained, 15 showed normal male fertility, with individual pollen grains loosely attached to dehisced anthers as in WT, although their rosette phenotypes were still similar to the pub4 plants (Figure S1). The result indicates that the absence of normal PUB4 in the tapetum layer is the cause of male sterility associated with the pub4 mutation.

PUB4 localizes to the cytoplasm

To determine the subcellular localization of PUB4, we created made a protein fusion construct in which enhanced YFP (eYFP) was fused to the C–terminus of PUB4. The fusion gene, which is under the control of the CaMV 35S promoter, was expressed transiently in Arabidopsis protoplasts for observation of the subcellular localization of the eYFP signal by confocal microscopy. The eYFP signal was localized in the cytosol of Arabidopsis protoplasts (Figure 7a). Stable transgenic Arabidopsis plants carrying the 35S::PUB4-eYFP reporter were also generated, and this reporter was functional as it was able to complement the pub4–1 phenotype. The PUB4–eYFP fluorescence signal in the transgenic lines was again found in the cytosol (Figure 7b).

Figure 7.

PUB4–eYFP is localized in the cytosol. YFP signal in Arabidopsis protoplasts transiently expressing PUB4–eYFP or eYFP protein driven by the CaMV 35S promoter (a) and in root cells stably expressing PUB4–eYFP or eYFP protein driven by the CaMV 35S promoter (b). Scale bars = 20 μm.

PUB4 has E3 ligase activity

Several PUB sub-family genes have been found to encode functional E3 ubiquitin ligases. We initially tested whether PUB4 has E3 ligase activity by using purified PUB4 expressed in Escherichia coli. However, the E. coli-expressed PUB4 was found in inclusion bodies, and refolded PUB4 from inclusion bodies did not show E3 ligase activity in our assay. To overcome this hurdle, we generated transgenic Arabidopsis lines that express the PUB4 protein fused to a FLAG tag under the control of the CaMV 35S promoter. The fusion construct (35S::PUB4-FLAG) was capable of complementing the pub4–1 phenotype when introduced into the pub4–1 mutant, indicating that it is biologically functional. The PUB4–FLAG fusion protein was purified from leaves of the transgenic lines by affinity purification using anti-FLAG M2 affinity gel (Figure 8a). As U–box E3 ligases may undergo self-ubiquitination in the presence of E1 and E2 proteins, the purified PUB4–FLAG was subjected to E3 ubiquitination assay. As shown in Figure 8(b), PUB4–FLAG was ubiquitinated in the presence of E1, E2, ubiquitin and ATP. No ubiquitination was detected in the absence of E1, E2 or ubiquitin.

Figure 8.

PUB4 has E3 ligase activity. (a) The PUB4–FLAG recombinant protein was affinity-purified from transgenic plants stably transformed with the 35S::PUB4-FLAG fusion gene. The purified protein was separated by SDS–PAGE and stained with Coomassie Brilliant Blue R250 (Sigma-aldrich, (b) PUB4 ubiquitination assay. The ubiquitination reaction was performed in the presence of PUB4–FLAG, E1, E2, ubiquitin (Ub) and ATP. Each reaction was divided equally into two parts and subjected to SDS–PAGE. One immunoblot was probed with anti-ubiquitin antibody (upper panel) to detect ubiquitinated PUB4–FLAG protein, and the other was probed with anti-FLAG antibody (bottom panel) to detect total PUB4–FLAG (as a loading control).

The male-sterility phenotype caused by the pub4 mutation is temperature-dependent

Many male-sterility mutants have been found to be temperature-dependent (Mou et al., 2002; Yasuor et al., 2006; Masuda et al., 2007; Wiebbecke et al., 2012). We found that the male-sterility phenotype was alleviated when the pub4 plants were grown under a lower temperature than the regular temperature of 22°C. When plants were grown continuously at 16°C, the pub4 plants became partially fertile, with a mean seed yield approximately 10–30% that of WT plants (Figure 9a,b). At 16°C, the mutant anthers still showed abnormally expanded tapetum with packed pollen (Figure 9c). However, in mature pub4 flowers from plants grown at 16°C, some pollen grains were freed (Figure 9c), making it possible for them to be dispersed onto stigmas, resulting in partial fertility. Continuous low temperature is not necessary to alleviate the male-sterility phenotype of pub4 as pub4 plants were also partially fertile when grown at approximately 16°C during the night and 22°C during the day (see below).

Figure 9.

The male-sterility phenotype of pub4 is temperature-dependent. (a) Seed yield of WT and pub4 plants grown at 22 and 16°C. (b) pub4 plants were partially fertile when grown at 16°C night/22°C day. (c) Comparison of wax sections of WT and pub4 anthers from plants grown at 22 and 16°C. Images of anther sections from stage 10 and stage 14 are shown. The arrows point to free pollen grains.(d–h) Seed setting of WT, pub4, bri1–5 and the pub4 bri1–5 double mutant grown at 16°C night/22°C day. (e–h) Close-up views of the plants shown in (d). The bri1–5 mutant is in the Ws ecotype background. The Ws/Col–0 plant is an F1 hybrid between Ws and Col–0. Scale bars = 50 μm (c) and 1 cm (b,d–h).

The male-sterility phenotype of pub4 is enhanced by the bri1–5 mutation

It was previously reported that the defect in the brassinosteroid (BR) pathway caused enlargement and delayed death of tapetal cells and male sterility (Ye et al., 2010a), raising the possibility that the pub4 phenotype may be due to a defect in the BR pathway. However, pub4 does not show other phenotypes that are typical of a defective BR pathway, such as shorter hypocotyls, leaf petioles and stems. Exogenous application of BR did not alleviate the pub4 phenotype. We created the double mutants pub4 bes1–D and pub4 bzr1–D. bes1–D (bri1–EMS-suppressor 1–D) and bzr1–D (brassinazole resistant 1–D) are gain-of-function mutations that lead to constitutive activation of the BR signaling pathway (Wang et al., 2002; Yin et al., 2002). Neither bes1–D nor bzr1–D alleviated the pub4 male-sterility phenotype. Therefore, it is unlikely that the male-sterility phenotype caused by the pub4 mutation is due to a defective BR pathway. However, the male-sterility phenotype was found to be enhanced in the pub4 bri1–5 double mutant at low temperature (Figure 9d–h). bri1–5 (BR-insensitive 1–5) is a weak loss-of-function allele of the BRI1 locus that encodes the BR receptor (Noguchi et al., 1999), and bri1–5 plants have normal male sterility and seed setting at 22 or 16°C (Figure 9d–h). However, although pub4 was partially fertile at growth temperatures of 16°C night/22°C day, the pub4 bri1–5 double mutant was almost completely sterile (Figure 9d–h). Our results indicate that the BR pathway and the PUB4-mediated pathway act in parallel for normal functioning of the tapetum layer and pollen grains.


Formation and proper release of functional pollen involves complex developmental regulation, including interactions between sporophytic anther tissues and gametophytic tissues. Proper functioning of tapetal cells, in particular, is known to be vital for normal pollen development.

The pub4 mutation causes male sterility due to a sporophytic defect. PUB4 is expressed at a relatively high level in the tapetal cells. The tapetal cells in the pub4 anthers display abnormal expansion soon after the tetrad stage. The presence of hypertrophic tapetal cells in pub4 reduces the locule space and causes developing pollen grains to be packed against each other. The tapetal cells in pub4 also show incomplete degeneration, and pollen grains adhere to residue from incompletely degenerated tapetal cells. The pub4 pollen grains have a very different exine structure compared with WT pollen. Despite of these defects, pollen grains from pub4 plants were found to be viable, and were able to fertilize eggs when hand-pollinated. Our study shows that the male-sterility trait of pub4 results from the failure to release pollen grains from anthers caused by the tapetal dysfunction.

Various defects in the tapetum layer are associated with a number of male-sterile mutants, underlining the importance of this maternal tissue for production of functional pollen. In mutants that lack the tapetum layer, such as extra sporogenous cells/excess microsporocytes 1 (Canales et al., 2002; Zhao et al., 2002) and tapetal determinant1 (Yang et al., 2003), meiocytes initiate meiosis but are unable to complete cytokinesis, leading to extensive and early abortion. Mutations that cause abnormal development of the tapetum layer, including premature or delayed degeneration or hypertrophy of the tapetum layer, also lead to extensive pollen abortion, such as the Arabidopsis mutations dysfunctional tapetum 1 (dyt1) (Zhang et al., 2006), myb33 myb65 (Millar and Gubler, 2005), male sterility 1 (ms1) (Wilson et al., 2001; Ito and Shinozaki, 2002; Ariizumi et al., 2005; Vizcay-Barrena and Wilson, 2006) and myb80 (formerly myb103) (Higginson et al., 2003; Li et al., 2007; Zhang et al., 2007), and the rice mutations tapetal degeneration retardation (Li et al., 2006; Zhang et al., 2008), apoptosis inhibitor 5 (Li et al., 2011b), persistent tapetal cell 1 (Li et al., 2011a) and microspore and tapetum regulator 1 (Tan et al., 2012). These reports revealed that the tapetum layer plays important roles in almost all stages of pollen development, and its malfunction often leads to complete male sterility. The ability of the pub4 mutant to produce viable pollen despite abnormal tapetal cell development and exine formation suggests a relatively mild tapetal defect caused by the mutation, perhaps because the mutation affects tapetum development at a relatively late stage compared with many other tapetum-defective mutants that lead to more extensive pollen abortion. Our quantitative RT–PCR results also showed that the transcript levels of several genes that are known to regulate tapetal development and degeneration, including DYSFUNCTIONAL TAPETUM1 (DYT1), MALE STERILITY 1 (MS1), ABORTED MICROSPORES (AMS) and MYB103 (Wilson and Zhang, 2009), were not significantly altered by the pub4 mutation (Figure S5). In addition, from publicly available transcriptome data, it appears that the transcript level of PUB4 is not significantly altered in the dyt1 (Feng et al., 2012), ms1 (Yang et al., 2007), ams (Ma et al., 2012) or myb103 (Phan et al., 2011) mutants. These results suggest that PUB4 may regulate tapetum development by a distinct mechanism.

Development of a durable wall enables the spore/pollen to withstand desiccation and other physical damage (Wellman, 2004). The outer pollen wall, known as the exine, consists of a chemically resistant compound called sporopollenin that is made up of complex biopolymers mainly derived from long-chain fatty acids and phenolic compounds (Dominguez et al., 1999; Ariizumi and Toriyama, 2011; Shi et al., 2011). The tapetal cells provide precursors for sporopollenin formation. Pollen lacking the exine layer, such as that produced by the Arabidopsis male sterility 2 (Aarts et al., 1997; Chen et al., 2011) and no exine formation 1 mutants (Ariizumi et al., 2004) and the rice defective pollen wall mutant (Shi et al., 2011) are male-sterile. Many other mutations with altered exine have been identified (Ariizumi and Toriyama, 2011), and many of them also affect pollen viability, including Arabidopsis defective in exine patterning (Paxson-Sowders et al., 2001) and ruptured pollen grain 1 (Guan et al., 2008), suggesting that normal exine formation is critical for pollen viability. However, those mutations may not only affect the exine formation but also cause other defects that are the cause of pollen viability, as they were largely identified initially by the male-sterility phenotypes. Among 12 kaonashi (kns) mutants initially identified through their altered exine structures under SEM , the majority produce viable pollen and are fully fertile (Suzuki et al., 2008), arguing that a normal structure of the exine may not be so important for pollen viability. Pollen from the pub4 anthers was found to have unevenly distributed sporopollenin, probably because the packed pollen grains prevent even distribution of sporopollenin precursors. In addition, the outer sexine of the pub4 pollen has distorted structures of the tectum and the bacula. These alterations in the exine caused by pub4 apparently do not affect pollen viability; however, they may be one of the reasons why pollen grains stick together. Surprisingly, in the pub4 anthers, sporopollenin droplets or exine-like structures formed on the surface of the tapetal cells (Figure 5k,l and Figure S3). This ectopic deposition of sporopollenin or ectopic formation of exine may result from excessive sporopollenin exocytosed by the tapetal cells and/or may be due to a defect in releasing sporopollenin compounds to pollen.

Male-sterility phenotypes associated with a number of genic and cytoplasmic male-sterility mutants are affected by environmental conditions, including photoperiods, light intensity, humidity and temperature, and some of these environment-dependent mutants have been explored for hybrid crop production using the so-called two-line system (Weider et al., 2009; Ding et al., 2012; Zhou et al., 2012). In Arabidopsis, the myb35 myb65 mutant shows increased fertility under high light or low temperature (Millar and Gubler, 2005). The male sterility of the faceless pollen–1 mutant, which is defective in exine formation, may be rescued by growing under high humidity (Ariizumi et al., 2003). For pub4, a lower temperature (16°C) partially restores male fertility, apparently by enabling release of some pollen grains from anthers. It remains to be determined whether other environmental conditions affect the fertility of pub4.

Several brassinosteroid (BR) biosynthetic and signaling mutants have been found to have reduced male fertility (Ye et al., 2010a). The male defects associated with the BR mutants share similarity with those of the pub4 phenotypes, including and enlarged tapetum and defective release of pollen grains, which adhere to the anther wall. BR is also involved in primexine synthesis, as the transient defective exine 1 (tde1) mutant was found to comprise a mutation in the DE-ETIOLATED 2 (DET2) gene (Ariizumi et al., 2008). However, neither exogenous application of BR nor the bzr1–D or bes1–D mutation alleviated the male-sterility phenotype of pub4, suggesting that male sterility in pub4 is not due to a defect in the BR pathway. However, bri1–5 exaggerated the male-sterility phenotype of pub4 at 16 °C, suggesting that the BR pathway and the PUB4-mediated pathway act in parallel for normal functioning of the tapetum layer and pollen grains.

A tobacco PUB4 homolog (NtPUB4) was identified by its interaction in a yeast two-hybrid screen with CHRK1, a tobacco chitinase-related receptor-like kinase (Kim et al., 2003). Suppression of CHRK1 expression causes pleiotropic developmental abnormalities, including ectopic cell proliferation. CHRK1 suppression lines contained a higher level of cytokinin than wild-type plants (Lee et al., 2003). Whether NtPUB4 interacts with CHRK1 in planta has not been reported. NtPUB4 shares higher sequence similarity to AtPUB4 than to other PUB proteins in Arabidopsis, raising the possibility that they may be functional orthologs; however, NtPUB4 localized at the plasma membrane (Kim et al., 2003) whereas AtPUB4 is localized in the cytosol (Figure 7). The biological function of NtPUB4 remains unknown. Although the exact molecular mode of action of AtPUB4 has yet to be defined, our study indicates the importance of the ubiquitination pathway in regulating tapetum growth and degeneration to ensure proper development and release of pollen.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana plants in this study are all in the Columbia (Col) background, except bri1–5, which is in the Ws ecotype background. All T–DNA insertion lines were obtained from the Arabidopsis Biological Resource Center (Columbus, OH). Plants were grown in a growth room under the following conditions: a 15 h light/9 h dark cycle with a light intensity of 125 mol m−2 sec−1 provided by cool-white fluorescent bulbs, 50% humidity and 22°C. The plants grown at 16°C were subjected to the same growth conditions except the difference in temperature.

Identification of pub4 alleles by PCR

The T–DNA insertion lines were verified by PCR analysis using T–DNA left-border primers (LBa1 for pub4–1 and pub4–2 and pDAP101LB for pub4–3) and PUB4-specific primers. Table S1 provides information on all primers used in this study.

Genetic complementation

The genetic complementation methods are described in Methods S1.

Light microscopy and electron microscopy

For pollen viability staining, anthers/pollen grains were stained with Alexander staining solution as previously described (Alexander, 1969). DAPI staining of pollen nuclei was performed as described previously (Park et al., 1998), and stained pollen was observed using a 340–380 nm excitation filter and a 450–480 nm emission filter and a Nikon Eclipse 800 wide-field microscope (Nikon, Aniline blue staining of pollen tubes in pistils was performed as described previously (Ishiguro et al., 2001). The pre-emasculated wild-type flowers were pollinated with WT or pub4 pollen. Twenty-four hours after pollination, pistils were fixed in ethanol/acetic acid (3:1) for 2 h at room temperature, washed three times with distilled water at room temperature (3 min for each wash) and incubated in 8 m NaOH overnight. The pistil tissues were washed in distilled water and stained in aniline blue solution (0.1% aniline blue in 0.1 m K2HPO4/KOH buffer, pH 11) for 5 h in dark. The stained pistils were observed and photographed using a Nikon Eclipse 800 wide-field microscope.

SEM observation of unfixed pollen and leaf trichomes was performed using a Hitachi TM–1000 table-top SEM ( For tissue embedding in wax and sectioning of anthers, inflorescences were fixed in FAA solution (ethanol 50%, acetic acid 5%, formaldehyde 3.7%) (Zhao et al., 2002) and embedded in Paraplast+ (SPI Supplies, (SPI Supplies, 8 μm thick were cut using a Microm HM 355–S wax microtome (Richard-Allan Scientific,, stained using 0.05% toluidine blue, and observed using a Nikon Eclipse 800 wide-field microscope. For TEM observation of thin sections of anthers, flower buds were packed in specimen planchettes in 0.1 m PIPES buffer (pH 6.8) containing 0.15 m sucrose, and ultra-rapidly frozen in a HPM 010 high pressure freezer (BAL-TEC/Leica, Frozen samples were substituted in 2% osmium tetroxide in acetone for 5 days at −80°C, slowly warmed to room temperature, rinsed in acetone, and embedded in Spurr's resin. Thin sections were cut using a Leica UCT ultramicrotome, stained in uranyl and lead salts, and observed using a LEO 912 AB energy filter TEM (Zeiss,

GUS reporter assay

To create the PUB4pro::GUS construct, the 1.2 kb promoter region of PUB4 was amplified from Col–0 genomic DNA using primers PUB4proF and PUB4proR (Table S1), and inserted upstream of the GUS reporter gene in the binary vector pBAR-GUS (Lu et al., 2011b). Histochemical GUS activity staining was performed as previously described (Xia et al., 1996). For thin sectioning after GUS staining, the stained tissues were fixed in FAA at 4°C overnight, embedded in Paraplast+, sectioned at 8 μm thickness, and observed under a Nikon Eclipse 800 wide-field microscope.

RNA isolation and quantitative RT–PCR

Total RNA was isolated from plants using Trizol reagent (Invitrogen, according to the manufacturer's instructions. Isolated RNA was treated with RNase-free DNase and purified using Qiagen RNeasy mini columns according to the manufacturer's instructions ( First-strand cDNA was synthesized using SuperScript III reverse transcriptase (Invitrogen). The ACTIN2 transcript was used as an internal control to normalize the RNA quantity. Primers P1 and P2 (Table S1) were designed to determine the PUB4 gene expression levels in WT plants. Three biological replicates were included in quantitative RT–PCR analysis.

eYFP reporter assay for determining PUB4 subcellular localization

The PUB4 coding sequences was PCR-amplified from cDNA using the primer pair PUB4codF and the PUB4codR (Table S1), cloned into pCR–BluntII-TOPO (Invitrogen), and sub-cloned into the pCAM35S::eYFP binary vector (Lu et al., 2011b) to generate pCAM35S::PUB4-eYFP. Transient expression of pCAM35S::PUB4-eYFP and pCAM35S::eYFP in Arabidopsis mesophyll protoplast was performed as previously described (Yoo et al., 2007). The fusion construct was also transformed into WT Arabidopsis plants to create stable transgenic lines, and root cells of the transgenic plants were observed for eYFP localization. The eYFP signal was observed using a Nikon C1 confocal microscope (488 nm argon laser line, 500–530 nm bandpass (BP) detection for YFP, 650-700 nm bandpass (BP) detection for chlorophyll).

PUB4 expression, purification, and E3 activity assay

The PCR product containing the PUB4 coding sequence was cloned into pCR-Blunt II-TOPO vector (Invitrogen) and then sub-cloned into pBAR35S:: FLAG (Lu et al., 2011b). The resulting pBAR35S:: PUB4–FLAG vector was transformed to Agrobacterium and then into Arabidopsis to construct stable transgenic lines expressing the PUB4–FLAG recombinant protein driven by the CaMV 35S promoter. The PUB4–FLAG recombinant protein was affinity-purified from transgenic Arabidopsis leaves using anti-FLAG M2 affinity gel (Sigma ). Assessment of PUB4–FLAG ubiquitin E3 ligase activity was performed using an auto-ubiquitinylation kit (Enzo Lifesciences, according to the manufacturer's instructions; 20 μl of PUB4–FLAG was used in each 50 μl reaction mixture with the various components specified. After the reaction, the reaction mixture was equally divided and separated on two gels by SDS–PAGE. The ubiquitin conjugate was detected on one gel using the ubiquitin antibody, and the PUB4–FLAG protein was detected on the other gel using the anti-FLAG M2 antibody (Sigma).


We thank Hao Chen (Donald Danforth Plant Science Center) for technical assistance and helpful discussions. This work was supported by the Research Grants Council of Hong Kong (grant number HKBU26211 to Y.X.), and a Hong Kong Baptist University Faculty Research Grant (grant number FRG2/10-11/037 to Y.X.).