A Plant Germline-Specific Integrator of Sperm Specification and Cell Cycle Progression

1The gametes of flowering plants are formed by discrete haploid gametophyte structures consisting of only a few cells that develop within the diploid reproductive floral organs. During spermatogenesis, each single haploid microspore divides asymmetrically to produce a larger vegetative cell that eventually gives rise to the pollen tube and a smaller germ, or generative, cell (Figure S1; reviewed in,). In contrast to germline cells in metozoans, angiosperm male germ cells do not undergo regenerative stem cell divisions, but divide once to form a pair of sperm cells. These sperm cells are delivered to the embryo sac via the pollen tube, where they fuse with egg and central cells to produce embryo and endosperm respectively. This process of double fertilization depends upon two functional sperm cells and is considered one of the major advances in the evolutionary success of flowering plants. Despite this importance, the molecular mechanisms underlying many component processes, including the production of both male and female gametes, remain largely unknown.

2Recent transcriptomic analysis of isolated Arabidopsis sperm cells shows that sperm cells express a distinct and diverse set of genes and there is evidence for extensive male germ cell gene expression in maize and lily,. Several male germline-specific genes have been characterized in Arabidopsis including AtMGH3, encoding a histone H3.3 variant,, AtGEX2, encoding a putative membrane associated protein, and AtGCS1 (HAP2), encoding a sperm cell surface protein required for fertilisation,. Homologues of AtGCS1 are found in many genera,, that include the green alga Chlamydomonas and the rat malarial parasite Plasmodium berghei, where they are required for gamete interactions and membrane fusion. Although gene expression in angiosperm sperm cells is extensive and essential for gamete functions little is known about its regulation. A transcriptional derepression mechanism, in which expression of male germline expressed genes is repressed in all non-germline cells by a protein called Germline Restrictive Silencing Factor (GRSF), has recently been proposed. A binding site for the GRSF protein was identified in the promoter region of the Lily male germline gene LGC1, and mutations in this sequence led to the ectopic activation of the LGC1 promoter in non-germline cells in lily and Arabidopsis. Although similar binding sites have been found in the promoter regions of several germline genes in Arabidopsis, including the germline-specific transcription factor gene DUO1, a functional role for these sites or of GRSF activity in regulating gene expression in Arabidopsis pollen has not been shown.

3Germ cell division resulting in the sperm cell pair in each pollen grain, is essential for double fertilization and recent data supports the capacity of both sperm cells to fertilize the egg cell in Arabidopsis. Several mutants have been described in which germ cell division is disrupted –. Mutations in the conserved cell cycle regulator CDKA1, and in the F-BOX protein FBL17 prevent germ cell division and result in mature pollen with a single germ cell. Defects in Chromosome Assembly Factor 1 (CAF1) can also disrupt germ cell division. Interestingly, the single germ cells in these mutants are capable of fertilization, with cdka1 and fbl17 mutant germ cells fertilizing the egg cell to produce an embryo that aborts early in development due to the lack of endosperm production. These mutations clearly demonstrate that germ cell division and specification can be uncoupled, but do not identify how these processes may be coordinated to produce twin sperm cells competent for double fertilization.

4DUO POLLEN1 (DUO1) is a unique male germ cell-specific R2R3 Myb protein that is also required for germ cell division in Arabidopsis. Unlike cdka1 and fbl17 single germ cells, duo1 germ cells do not lead to successful fertilization, suggesting that in addition to germ cell cycle defects, key features of gamete differentiation and function are impaired in duo1. Here we further characterize DUO1 as an essential, positive regulator of sperm cell production in plants. We use various molecular markers and ectopic expression assays to show that DUO1 is both necessary and sufficient for the expression of male germline genes. We show that DUO1 is required for the expression of the Arabidopsis G2/M regulator CyclinB1;1 (AtCycB1;1) in the male germline and that AtCycB1:1 can partially rescue defective germ cell division in duo1. Our findings reveal a novel integrative role for the germline-specific DUO1 protein, in cell specification and cell cycle progression necessary for twin sperm cell production. Furthermore, we show that restriction of DUO1 expression to the male germline is not dependent on a putative GRSF binding site but involves positive elements in the promoter.

5To investigate the potential role of DUO1 in regulating sperm specification we examined the expression of three male germline markers, AtMGH3, AtGEX2 and AtGCS1, in mutant duo1 pollen. We exploited marker lines with promoter regions of these germline genes linked to GFP. First we characterised the expression of these markers in a coordinated manner using confocal laser scanning microscopy (CLSM) throughout development of wild-type pollen (Figure 1A–C), and compared their profiles with the expression of a DUO1∶mRFP fusion protein under control of the DUO1 promoter (DUO1-DUO1::mRFP; Figure 1D). The expression of all three germ cell markers is undetectable in free microspores when DUO1 is not expressed (Figure 1, Panel 1). Fluorescence is first detected in the germ cell during or soon after engulfment by the vegetative cell, appearing at a similar time to the expression of DUO1 (Figure 1, Panel 2). As the pollen matures the level of GFP accumulates in germ cells before mitosis and remains high in mature sperm cells (Figure 1A–C, Panels 3–5). The accumulation of GFP in progressive stages is illustrated by the reduced autofluorescence signal arising from the pollen wall, reflecting the reduced exposure needed to capture a relatively unsaturated germ cell GFP signal. DUO1 expression persists during pollen development, although its abundance does not obviously increase in tricellular and mature pollen (Figure 1D). Our analysis shows that in common with AtMGH3 and AtGEX2, the expression of AtGCS1, previously thought to be sperm cell-specific in Arabidopsis, is detected in germ cells soon after asymmetric division (Figure 1C). The expression of all male germ cell markers shortly after the asymmetric division shows that sperm cell specification begins early after inception of the germline prior to passage of germ cells through mitosis.

6The three male germline markers were introduced into heterozygous duo1 plants that produce 50% wild type pollen and 50% mutant pollen, and GFP expression was scored. Virtually all the wild type pollen showed GFP fluorescence in twin sperm cells while there was no fluorescence, or rarely a weak GFP signal, in the single germ cell in duo1 pollen (Figure 1E–G, I–K; Table S1). When these markers were introduced into the cdka;1 mutant in which the arrested germ cell is able to fertilize the egg cell, fluorescence was observed in the single germ cells in mutant pollen (Figure 1E–G, Table S1). This result confirms that germ cell division and cell fate specification are uncoupled in cdka;1 mutant pollen, similar to the observed expression of germ cell markers in arrested but functional germ cells in CAF1 mutants. The absence of GFP in mutant duo1 germ cells demonstrates that DUO1 is necessary for the expression of several germline-expressed genes, and explains why duo1 pollen is infertile (it lacks proteins including AtGCS1 that are essential for fertilization). In contrast, when the DUO1 promoter was used to express a nuclear-targeted histone H2B::mRFP marker protein, fluorescence was detected in mutant duo1 germ cells, similar to its expression in wild type sperm cells and in cdka;1 germ cells (Figure 1H,L; Table S1), indicating that DUO1 promoter activation does not depend upon DUO1 itself.

7To independently confirm the regulation of germline genes by DUO1 we ectopically expressed DUO1 in seedlings, and in pollen vegetative cells, where AtMGH3, AtGEX2 and AtGCS1 are not normally expressed. As DUO1 contains a recognition site for microRNA159 we used a resistant DUO1 cDNA (mDUO1) with an altered nucleotide sequence at the miR159 binding site, but encoding the native amino acid sequence. Transgenic seedlings in which the mDUO1 cDNA was placed under the control of an estradiol inducible promoter showed mDUO1 induction when exposed to estradiol (Figure 2A). Expression of the male germline genes, AtMGH3, AtGEX2 and AtGCS1, was also induced, with high levels of transcripts present only in plants exposed to estradiol and containing mDUO1 (Figure 2A). Similarly, when a DUO1::mRFP fusion was ectopically expressed in pollen vegetative cells using the LAT52 promoter, we observed ectopic expression of the AtMGH3 marker in vegetative cell nuclei (Figure 2B,C; Table S2). Thus ectopic expression of DUO1 is sufficient for activation of germ cell-specific gene expression in a range of non-germline cells.

8The phenotype of duo1 shows that in addition to the activation of male germline genes, DUO1 is required for germ cell division. Mutant duo1 germ cells complete DNA synthesis (S) phase but fail to enter mitosis (M),, suggesting that DUO1 may regulate the expression of essential G2/M factors. As the Arabidopsis CDK regulatory subunit AtCycB1;1 shows enhanced expression at G2/M, and is expressed in developing pollen (Figure S2), we investigated AtCycB1;1 as a potential downstream target of DUO1. To monitor the expression of AtCycB1;1 we used the pCDG marker which contains the AtCycB1;1 promoter region and mitotic destruction box fused to the β-glucuronidase (GUS) reporter. First we analysed the marker in wild type pollen (Figure 3A–F). Individual pollen grains at different stages of development (as determined by DAPI staining) were analysed for GUS activity, which results in the formation of indigo microcrystals. Microspores and bicellular pollen shortly after mitosis contain numerous indigo crystals, with the number peaking close to mitosis (Figure 3A–C), indicating that expression of AtCycB1;1 is linked to asymmetric division. Expression is then abolished in bicellular pollen (Figure 3D). Close to germ cell mitosis, single indigo crystals are present specifically in germ cells (located by DAPI staining; Figure 3E) indicating expression of AtCycB1;1 in the germ cell before division. The protein is degraded after mitosis and is absent in tricellular pollen (Figure 3F).

9We then counted the number of pollen grains with GUS staining at different stages of development in wild type and heterozygous duo1 plants. In both wild type and heterozygous duo1 plants, polarized microspores and vegetative cells shortly after asymmetric division showed almost 100% staining, indicating expression of AtCycB1:1 (Figure 3G). Thereafter vegetative cell staining declined and was absent from late-bicellular stage pollen (Figure 3G). Germ cell staining was subsequently observed in ∼100% of pollen from wild type plants close to mitosis, but was reduced by approximately half in heterozygous duo1 plants at this stage (Figure 3H). As half of the pollen population is mutant in heterozygous duo1 plants, and wild type pollen show GUS staining, this reduction in staining is consistent with a lack of AtCycB1;1 expression in mutant duo1 pollen. This indicates that DUO1 is required for the expression of AtCycB1;1 in male germ cells.

10We then analysed the expression of AtCycB1;1 transcripts in seedlings after steroid induction of mDUO1. In contrast to the germline markers, AtCycB1;1 was expressed at a low level in seedlings not exposed to estradiol and the presence of DUO1 did not affect the level of AtCycB1;1 transcripts (Figure 2A). Thus, although DUO1 is required for germline expression of AtCycB1;1 the presence of DUO1 is not sufficient to induce AtCycB1;1 mRNA in seedlings. Transcription of the AtCycB1;1 gene is known to be regulated by a number of factors, including activators such as three repeat or other Myb proteins and TCP20 and repressors such as TOUSLED. Thus, DUO1 may be unable to overcome these controls in seedlings, and may affect AtCycB1;1 transcription in the male germline through an indirect mechanism or through effects on AtCycB1;1 protein stability.

11To investigate the role of AtCycB1;1 in the failure of duo1 male germ cells to enter mitosis we determined whether AtCycB1;1 is sufficient to rescue the germ cell mitosis defect in duo1 pollen. We used the DUO1 promoter to drive AtCycB1;1 expression in the male germline. The proportion of bicellular or tricellular pollen grains from heterozygous duo1 plants either not transformed or transformed with either of two control constructs (MGH3-AtCycB1;1::GFP, which is not expressed in mutant pollen, and LAT52-AtCycB1;1, which is expressed only in the vegetative cell) did not vary significantly from 50% (Chi2 p<0.05) (Figure 3I, Table S3). In contrast, in heterozygous duo1 plants transformed with DUO1-AtCycB1:1 the majority of lines (31/49) showed a significantly reduced frequency of bicellular pollen and a corresponding increase in tricellular pollen (Figure 3I, Table S3). This suggests that restoring AtCycB1;1 in duo1 mutant germ cells is sufficient to promote mitosis in a proportion of the population. Complementation was however incompletely penetrant, which may result from the use of the DUO1 promoter that may not produce native amounts of AtCycB1;1. It is also possible that other factors with a role in G2/M transition, such as other AtCycB family members that are also expressed during pollen development, may also be absent in duo1 pollen.

12To determine if the presence of DUO1-AtCycB1;1 in duo1 pollen restored only the ability to proceed through mitosis or germline specification as well, we analysed expression of the AtMGH3 and AtGCS1 markers in duo1 plants showing partial complementation (Figure 3J, Table S4). In contrast to plants without DUO1-AtCycB1;1 where almost all tricellular pollen expresses GFP, plants displaying partial complementation produce ∼10% of pollen that is tricellular but does not express the markers. As there is also a ∼10% decrease in bicellular pollen, this new class of tricellular pollen is most likely duo1 pollen in which the division defect has been complemented by the DUO1-AtCycB1;1 construct, but in which the markers have not been activated. Consistent with this, DUO1-AtCycB1;1 complemented duo1 pollen showed no male transmission (Table S5). Thus, complementation of the bicellular phenotype by AtCycB1;1 only affects cell division and does not restore expression of germline gene expression and sperm cell function.

13Closer examination of mature pollen grains ectopically expressing DUO1 in the vegetative cell revealed a distinctive morphology with reduced cytoplasmic density, larger vacuoles and numerous large cytoplasmic inclusions (Figure S3). This phenotype was only found in pollen containing vegetative nucleus GFP (Table S6) and analysis of pollen viability revealed up to 50% non-viable pollen with the aberrant pollen not being viable (Figure S3E, F, Table S7). Similar phenotypes are not seen in pollen of plants transformed with LAT52-H2B::GFP where the transgene is transmitted normally (data not shown). Furthermore, Arabidopsis plants constitutively expressing DUO1 (driven by the 35S promoter) show severe seedling patterning defects, twisted and curled leaves and floral defects. These phenotypes demonstrate the importance of restricting high level expression of DUO1 to male germ cells.

14Such restriction may partially rely upon degradation of DUO1 mRNA by microRNA159 in certain cell types but promoter elements are also likely to be important. As such, restriction of DUO1 expression to the male germline has been proposed to rely on the repressor protein GRSF due to a putative GRSF binding site in the DUO1 promoter. Mutagenesis of similar sequences in the LGC1 promoter led to ectopic activation of the LGC1 promoter in non-germ line cells in tobacco and Arabidopsis. However, when we specifically mutated the putative GRSF binding site in the DUO1 promoter this did not affect the germline-specific expression of DUO1 (Figure 4A–D). Moreover, sequences in the 150 bp proximal DUO1 promoter, excluding putative GRSF binding sites, were sufficient for germline-specific expression (Figure 4E). Although factors that bind to the lily LGC1 silencer appear to be present in non-germline cells in Arabidopsis, the germline-restricted activation of DUO1 does not appear to involve GRSF mediated repression. Since the DUO1 promoter appears to be active only after asymmetric division in the newly formed germ cell and that activation does not depend upon DUO1 itself (see Figure 1), activation of the DUO1 promoter may depend on proximal region-binding transcription factors that are inherited and/or segregated during asymmetric division of the microspore.

15We have shown that DUO1 is both necessary and sufficient for the expression of several male germline genes including AtGCS1 that is required for gamete fusion, thus DUO1 has a major role in the specification of functional male gametes. DUO1 is not involved in regulating microspore division and is first expressed in germ cells after asymmetric division. DUO1 is also required for the entry of male germ cells into mitosis and for the germline expression of the G2/M regulator AtCycB1;1. Thus, the germ cell programme under DUO1 control has an important role in regulating core cell cycle machinery specifically in the male germline. The discovery of the dual role of DUO1 as a positive regulator in male germline specification and cell cycle progression is a major advance in uncovering the molecular mechanisms involved in plant sexual reproduction. DUO1 is currently the only regulatory factor that has been shown to be required for gamete specification in plants. Recently we described an independent mechanism for male germ cell cycle regulation where the F-BOX protein FBL17 controls germ cell entry into S-phase via the degradation of the CDKA inhibitors KRP6 and 7. Taken together these data establish a molecular framework for twin sperm cell production in flowering plants (Figure 5).

16Further analysis will shed light on how DUO1 activates its targets, and how DUO1 itself is activated specifically in the male germline. The identification of the role of DUO1 in germ cell specification also provides an exciting platform to develop a detailed regulatory network for male gametogenesis and for comparative studies of the control of sperm cell production. DUO1 homologs are found throughout the land plants from the non-flowering plants Selaginella moellendorffii and Physcomitrella patens (moss) through to the monocots and dicots (Figure S4). Exploring the functional conservation of DUO1 in different species will reveal if DUO1 has a conserved role in male gamete production, in terms of both of germline mitosis and specification, where DUO1 may regulate the expression of a similar suite of genes such as the conserved GCS1 protein. Such studies may shed light on the evolution of regulatory mechanisms in plant germline development and their significance in double fertilization in flowering plants.

17Arabidopsis plants were grown at 21°C with a 16 h-light and 8 h-dark cycle or with 24 h light, with variable humidity. Experiments were conducted in the duo1-1 (in No-0) or the No-0 backgrounds, except for those involving the inducible ectopic expression of mDUO1 and analysis of the DUO1 promoter that were conducted in Col-0. The AtGCS1-AtGCS1::GFP, AtGEX2-GFP and CDG marker lines are also in Col-0. Plants were transformed with Agrobacterium tumefaciens (GV3101) using a standard floral dipping method. Transformants were selected either on Murashige and Skoog (MS) agar containing 50 µg/ml kanamycin or 20 µg/ml hygromycin or on soil with 30 µg/ml BASTA (glufosinate ammonium, DHAI PROCIDA) fed by sub-irrigation.

18Gateway single and multi-site construction (Invitrogen) was used to generate most vectors. DNA was amplified from genomic DNA, cDNA or plasmid DNA by PCR with high fidelity Phusion DNA polymerase (Finnzymes) and primers with suitable attachment site (attB) adapters (Table S8, attB adapters in italics). Full-length attB sites were added to each fragment in a second high fidelity PCR. For site-directed mutagenesis of the putative GRSF binding site in the DUO1 promoter a two-step recombinant PCR approach was taken. Two overlapping PCR fragments were generated containing the mutated sequence (underlined in Table S8) and the two fragments joined in a stitching PCR. PCR fragments were cloned into pDONR vectors (Invitrogen; pDONR207 for AtCycB1;1 cDNA or pDONR221 for H2B and DUO1 and mDUO1 cDNA, pDONRP4P1R for promoter regions and pDONRP2RP3 for GFP and RFP) via a BP reaction using BP Clonase II (Invitrogen). The product of BP reactions was transformed into alpha-select chemically competent cells (Bioline) and all clones were verified by sequencing.

19A multipart LR reaction using LR Clonase plus (Invitrogen) and the destination vector pK7m34GW was used to generate the AtMGH3 marker, AtMGH3-H2B::GFP. This contains the region upstream of the AtMGH3 coding region driving expression of a H2B::GFP fusion protein, with the H2B used to give a nuclear GFP signal. The GCS1-GCS1::GFP marker was constructed by inserting a PCR fragment of GFP (primers in Table S8) into an AflII site in the 16th exon of a Arabidopsis GCS1 genomic DNA fragment in the previously described binary vector. The AtGEX2-GFP marker, with the GEX2 promoter region driving GFP expression was kindly provided by Shelia McCormick.

20The vectors DUO1-DUO1::mRFP, DUO1-H2B::mRFP and LAT52-DUO1::mRFP were also generated using gateway multisite cloning and the vectors pK7m34GW or pB7m34GW. DUO1-DUO1::mRFP uses the DUO1 promoter region to drive expression of a DUO1::mRFP fusion (used to follow the DUO1 protein during pollen development) while DUO1-H2B::GFP uses the DUO1 promoter to produce a H2B::mRFP fusion protein (used to follow the activity of the DUO1 promoter in duo1 pollen). The LAT52 promoter is active in the vegetative cell so was used to ectopically express DUO1::mRFP in the vegetative cell. Vectors to analyse the DUO1 promoter region were also constructed using gateway multisite cloning.

21The DUO1 mRNA contains a functional recognition site for the microRNA miR159, so for inducible expression of DUO1 a miR159 resistant version of the DUO1 cDNA was used containing silent mutations in the miR159 binding site. This was cloned in the vector pMDC7 that contains the XVE estradiol inducible promoter system, using a single part LR reaction and LR Clonase II (Invitrogen).

22For experiments examining the ability of AtCycB1;1 to complement the duo1 division phenotype the vectors pB2GW7 and pH2GW7 were modified to contain the DUO1 and LAT52 promoters respectively. The 1.2 kb DUO1 and 609 bp LAT52 promoter fragments were amplified from cloned sequences using restriction tagged oligonucleotide primer pairs (Table S8). A single part gateway reaction was then used to clone AtCycB1;1 into the vectors creating DUO1-CycB1;1 and LAT52-CycB1;1. MGH3-CycB1;1::GFP was generated using a multipart gateway reaction.

23T2 seed from transgenic Col-O plants were grown in MSO plates containing 20 µg/ml hygromycin in standard conditions for 12 days. 25 seedlings were transferred to either control plants containing 0.002% v/v DMSO or induction plates contain 2 µM 17β-estradiol dissolved in DMSO. Plants were returned to the growth room for a further 24 h, before being snap frozen in liquid nitrogen.

24Pollen from ecotype Landsberg erecta at different stages of development was isolated and RNA extracted as described. For RT-PCR on seedling ectopically expressing mDUO1 and for DUO1 promoter analysis, RNA was extracted from frozen samples using the Qiagen RNeasy Kit. Samples of 750 ng or 1 µg of total RNA for pollen stages and seedlings, respectively, were reverse transcribed in a 20 µl reaction using Superscript II RNase H reverse transcriptase (Invitrogen) and an oligodT primer as per the manufactures instructions. For PCR amplification 1 µl of a 10× (pollen stages) or 5× (seedling) diluted cDNA was used in a 25 µl reaction using Biotaq DNA polymerase (Bioline) and 12.5 pmol of each primer (Table S9). PCR conditions were: 96°C for 1 min, 30 to 40 cycles at 96°C for 30 s, 55°C for 30 s, 72°C for 40 s followed by 5 min at 72°C. Histone H3 (At4g40040) was used as a control.

25Mature pollen was stained with DAPI (4′-6-Diamidino-2-phenylindole) as described previously. Staining for GUS activity was performed as described with inflorescences incubated in GUS buffer (100 mM sodium-phosphate, pH 7; 5 mM EDTA, 0.1% Triton X-100) with 1 mM X-gluc (5-bromo-4-chloro-3-indolyl b-D-glucuronide) and 0.5 mM K3Fe[CN]6, at 37°C for 1–3 days. Stained inflorescences were then cleared with 70% ethanol. Pollen was dissected out and stained with 0.8 µg/ml DAPI in GUS buffer. Phenotypic analysis of pollen was conducted on a Nikon TE2000-E inverted microscope (Nikon, Japan). Bright field and DIC images were captured with a Nikon-D100 camera (Model MH-18, Japan) and fluorescence images were captured with HAMAMATSU – ORCA-ER digital camera (Model C4742-95, Japan) using Openlab software version 5.0.2. (Improvision).

26For confocal laser scanning microscopy (CLSM) pollen from buds at different stages of development was teased out of the anther with a needle and mounted in 0.3 M mannitol and mature pollen was released directly into 0.3 M mannitol. Pollen was viewed with a Nikon TE2000-E inverted microscope and C1 confocal system using Melles Griot Argon Ion (emission 488 nm) and Melles Griot Helium-Neon (emission 543 nm) lasers, detection filters for GFP and RFP, and EZ-C1 control and imaging software.