Plant materials and growth conditions

For all experiments, the A. thaliana accession Columbia (Col-0) was used as the wild type. All A. thaliana transgenic lines were in a Columbia (Col-0) background, and the myb98 mutant was previously described [10]. The following transgenic lines were also previously described: RPS5Apro::H2B–tdTomato [11], RPS5Apro::tdTomato–LTI6b [12], RPS5Apro::H2B–sGFP [13], FGR8.0 [14], MYB98pro::GFP [10], MYB98pro::GFP–MYB98 [15], EC1.2pro::mtKaede [16], FWApro::FWA–GFP [17], and ABI4pro::H2B–tdTomato [18]. The transgenic lines used are listed in S1 Table.

Arabidopsis seeds were sown on plates containing half-strength Murashige and Skoog salts (Duchefa Biochemie, Haarlem, the Netherlands), 0.05% MES-KOH (pH 5.8), 1× Gamborg’s vitamin solution (Sigma, St Louis, Missouri, United States of America), and 1% agar. The plates were incubated in a growth chamber at 22°C under continuous lighting after cold treatments at 4°C for 2 to 3 days in the dark. Two-week-old seedlings were transferred to soil and grown at 21 to 25°C under long-day conditions (16-hour light/8-hour dark).

Plasmid construction

GPR1pro::H2B–mNeonGreen (coded as DKv1200) was constructed with the 2,568-bp upstream region of GPR1 (At3g23860) and the full-length coding region of H2B (HTB1: At1g07790) fused to mNeonGreen (Allele Biotechnology, San Diego, California, USA) with the (SGGGG)₂ linker, and the 1,959-bp downstream regions were cloned into the binary vector pPZP211 [19]. CDR1–LIKE2pro::CDR1–LIKE2–mClover (coded as DKv1023) was constructed using the 1,398-bp upstream region and the full-length coding region of CDR1–LIKE2 (At1g31450) fused to mClover with the (SGGGG)₂ linker and NOS terminator and cloned into the binary vector pPZP211. CDR1–LIKE1pro::CDR1–LIKE1–mClover (coded as DKv1024) was constructed using the 2,000-bp upstream region and the full-length coding region of CDR1–LIKE1 (At2g35615) fused to mClover with the (SGGGG)₂ linker and NOS terminator and then cloned into the binary vector pPZP211. Finally, CDR1pro::CDR1–mClover (coded as DKv1025) was constructed with the 1,577-bp upstream region and the full-length coding region of CDR1 (At5g33340) fused to mClover with the (SGGGG)₂ linker and NOS terminator and then cloned into the binary vector pPZP211.

To construct the multiple cell type–specific marker line with the nuclei marker (coded as DKv1110), the following sequences were cloned into the binary vector pPZP211, and NPTII was replaced with mCherry under the control of the At2S3 promoter from a pAlligator-derived binary vector [20]: EC1.1pro::SP–mTurquoise2–CTPP [21] (the 463-bp EC1.1 promoter was fused to mTurquoise2 that fused to the signal peptide (SP) sequence of EXGT–A1 (At2g06850) at the N-terminus and to a vacuolar sorting signal COOH-terminal propeptide (CTPP), and the HSP terminator); DD1pro::ermTFP1 (the 1,262-bp DD1 promoter (At1g36340) was fused to mTFP1 that was fused to the SP sequence of EXGT–A1 at the N-terminus and to an endoplasmic reticulum (ER) retention signal (HDEL) at the carboxyl terminus, and the OCS terminator); MYB98pro::mRuby3–LTI6b (the 1,610-bp MYB98 promoter and mRuby3 fused to the start codon of LTI6b (At3g05890) with the (SGGGG)₂ linker, and the HSP terminator); and AKVpro::H2B–mScarlet–I (the 2,949-bp upstream regions of AKV (At4g05440 [22]) and the full-length coding region of H2B (HTB1: At1g07790) fused to mScarlet–I with the (SGGGG)₂ linker). To enhance the expression level of H2B–mScarlet–I, the 5′ UTR of AtADH (alcohol dehydrogenase) was inserted between the AKV promoter and H2B coding sequence. Unfortunately, H2B–mScarlet–I was expressed not only in female gametophytes but also in sporophytic cells for this construct.

SBT4.13pro::SBT4.13–mClover (coded as pDM349) was a 2,040-bp upstream region, and the full-length coding region of SBT4.13 (At5g59120) was amplified and cloned into pPZP221Clo using a SmaI site [23].

MYB98pro::NLS–mRuby2 (coded as pDM371), a DNA fragment of NLS–mRuby2 (obtained from Addgene plasmid 40260), was amplified and then cloned into the pENTR/D-TOPO vector (Invitrogen, Japan) to generate pOR006. LR recombination between pDM286 [13] and pOR006 was performed using LR clonaseII (Invitrogen) to produce pDM371.

The binary vectors were introduced into the Agrobacterium tumefaciens strain EHA105. The floral dip or simplified Agrobacterium-mediated methods were used for Arabidopsis transformations [24]. The transgenic lines and primers are listed in S1 and S2 Tables, respectively.

Microscopy

To image female gametophyte development, we used 2 spinning disk confocal microscope systems following the settings of [25], with the following modification: For the live imaging of the in vitro female gametophyte development, the confocal images were acquired using an inverted fluorescence microscope (IX-83; Olympus, Tokyo, Japan), equipped with an automatically programmable XY stage (BioPrecision2; Ludl Electronic Products, Hawthorne, New York, USA), a disk-scan confocal system (CSU-W1; Yokogawa Electric, Tokyo, Japan), 488-nm and 561-nm LD lasers (Sapphire; Coherent, Santa Clara, USA), and an EMCCD camera (iXon3 888; Andor Technologies, South Windsor, Connecticut, USA). Time-lapse images were acquired with a 60× silicone oil immersion objective lens (UPLSAPO60XS, WD = 0.30 mm, NA = 1.30; Olympus) mounted on a Piezo focus drive (P-721; Physik Instrumente, Karlsruhe, Germany). We used 2 band-pass filters, 520/35 nm for the GFP and 593/46 nm for the tdTomato. The images were processed with Metamorph (Universal Imaging, Downingtown, USA) and Fiji [26] to create maximum-intensity projection images and to add color.

We also used an inverted confocal microscope system with a stable incubation chamber (CV1000; Yokogawa Electric) equipped with 488-nm and 561-nm LD lasers (Yokogawa Electric) and an EMCCD camera (ImagEM 1K C9100-14 or ImagEM C9100-13; Hamamatsu Photonics, Shizuoka, Japan). Time-lapse images were acquired with a 40× objective lens (UPLSAPO40×, WD = 0.18 mm, NA = 0.95; Olympus). We used the 2 band-pass filters, 520/35 nm for the GFP and 617/73 nm for the tdTomato.

The number of observations and microscope information for each construct are listed in S3 Table.

Isolation of female gametophyte cells

We used an inverted fluorescence microscope (IX-71; Olympus) equipped with a 3-charge-coupled device (CCD) digital camera (C7780; Hamamatsu Photonics). Images were acquired using a 40× objective lens (LUCPlanFl 40×, WD = 2.7–4 mm, NA = 0.60; Olympus). We emasculated the flowers at stage 12c [27]. Two days later, the unfertilized ovules of each cell marker line were treated with enzyme solution (1% cellulase [Worthington Biochemical Corporation, Freehold, USA], 0.3% macerozyme R-10 [Yakult, Tokyo, Japan], 0.05% pectolyase [Kyowa Kasei, Osaka, Japan], and 0.45 M mannitol [pH 7.0]). The target cells were collected during 20-minute to 1-hour period of the enzyme solution treatment; we used a micromanipulator (MN-4, MO-202U; Narishige, Tokyo, Japan) and micropipette (Picopipet HR; Nepa Gene, Chiba, Japan) with glass capillaries (G-1; Narishige), which were pulled with a micropipette puller (P-97; Sutter, Novato, USA) [28]. The frequency of released central cells were lower than that of egg cells. After 30 minutes of treatment with the enzyme solution, ovules without released central cells were gently pressed by glass capillary to help release from the ovules.

cDNA preparation and library construction for sequencing

The mRNA was extracted from 12 to 18 synergid, egg, and central cells with Dynabeads mRNA DIRECT Micro Kit according to the manufacturer’s instructions (Invitrogen, Carlsbad, USA). The extracted mRNA was amplified using Ovation RNA sequencing (RNA-seq) System V2 (NuGEN, San Carlos, USA). The RNA-seq libraries were prepared using a TruSeq RNA Sample Preparation Kit and Multiplexing Sample Preparation Oligonucleotide Kit according to the manufacturer’s instructions (Illumina, San Diego, USA). The libraries were sequenced on an Illumina GAIIx (Illumina) using 36-bp single-end reads.

RNA-seq data analysis

Reads were filtered using fastp (ver. 0.20.0 [29]). The cleaned reads were mapped to the Arabidopsis reference genome TAIR10, using HISAT2 (ver. 2.1.0 [30]). The expression level for each gene was quantified using the read count and transcripts per million (TPM) with Stringtie (ver. 2.1.1 [31,32]). Differentially expressed genes (DEGs) between the synergid cells of the wild-type and the myb98 mutant were identified by TCC with a false discovery rate of 0.01 (ver. 1.24.0 [33]). The TCC+baySeq (ver. 2.18.0) method with a false discovery rate of 0.01 was used for the identification of the DEGs among the synergid, egg, and central cells of the wild type [34]. Gene Ontology (GO) enrichment analysis was performed using g:Profiler (g:COSt; https://biit.cs.ut.ee/gprofiler/gost). The g:SCS threshold was set to 0.05.

Live imaging of the nuclear dynamics during female gametophyte development

The development of female angiosperm gametophytes in vivo occurred within multiple layers of the maternal tissues of the flower. To investigate their actual developmental time course, we performed live-cell imaging of the female gametophytes development using the previously developed in vitro ovule culture system for embryogenesis in Arabidopsis [25]. To observe the nuclear dynamics in female gametophyte development, we constructed GPR1pro::H2B–mNeonGreen::GPR1ter (Fig 1A, S1 Movie). GPR1 (GTP-BINDING PROTEIN RELATED1) was previously found to be expressed in the megaspore mother cells (i.e., at stage FG0) and the female gametophytes at FG1 to FG7 [35]. At FG1, the nucleus was located at the female gametophyte center (Fig 1A; 0:00). Approximately 3 hours after the observation, the nucleus divided into two during its first mitosis (Fig 1B; 3:15). At FG2, the 2 nuclei were positioned at the center of the female gametophyte. Approximately 8 hours after the start of FG2, the nuclei moved to opposite ends of the ovule (Fig 1A; 12:00), at which point the vacuole may appear (FG3) [27]. After the second mitosis, the nuclei divide to lie in an orthogonal line along the chalazal–micropylar axis (Fig 1A; 13:00, 13:25). The chalazal nuclei migrated along a line that was parallel to the chalazal–micropylar axis (Fig 1A; 14:20), while the micropylar nuclei migrated along the surface of the female gametophyte, not parallel to the chalazal–micropylar axis (Fig 1A; 15:45, 18:40). The micropylar nuclei tended to lie along the abaxial surface of the female gametophytes (78/94, 83%). After the end of the third mitosis, the polar nuclei migrated linearly, not along the surface of the female gametophyte, but toward each other to fuse (Fig 1A; 20:40, 24:00). We calculated the duration of each nuclear division from 78 movies of GPR1pro::H2B–mNeonGreen (n = 37), RPS5Apro::H2B–tdTomato (n = 29), and RPS5Apro::H2B–sGFP (n = 12) (Fig 1B). The duration of the second and third nuclear divisions was 11.8 ± 3.3 hours (mean ± standard deviation; n = 9, Fig 1B; FG2,3) and 7.7 ± 1.1 hours (n = 49, Fig 1B; FG4), respectively. After cellularization, it took 3.8 ± 1.1 hours (n = 71) and 11.8 ± 2.9 hours (n = 36, Fig 1B; FG5) after the third mitosis for the polar nuclei to attach and fuse, respectively. Thus, normal female gametophyte development was observed using the in vitro ovule culture system [27]. The Nitsch medium supplemented with 5% trehalose resulted in the highest percentage of ovule survival in vitro during seed development and after fertilization [25]. This medium also enabled us to perform live-cell imaging during female gametophyte development, prior to fertilization.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Live-cell imaging of the nuclear dynamics during female gametophyte development in Arabidopsis using an in vitro ovule culture system.

(A) Nuclei were labeled with GPR1pro::H2B–mNeonGreen. The numbers indicate time (hr:min) from the onset of observation. Images are representative of 37 time-lapse images from 3 independent transgenic lines. We succeeded in time-lapse recordings of the nuclear divisions in the isolated ovules from FG1 to FG6. FG1, uninucleate functional megaspore; FG2, 2-nucleate stage; FG3, 2 nuclei separated by a large central vacuole; FG4, 4-nucleate stage; FG5, 8-nucleate/7-celled stage; FG6, 7-celled with polar nuclei fused. Scale bar: 20 μm. (B) Durations of the nuclear divisions between the stages from FG2 to FG6. The interval times of the nuclear divisions for the female gametophyte development were analyzed for GPR1pro::H2B–mNeonGreen (n = 37), RPS5Apro::H2B–tdTomato (n = 29), and RPS5Apro::H2B–sGFP (n = 12). The underlying numerical data for B can be found in S1 Data. AC, antipodal cells; EC, egg cell; Pn, Polar nucleus; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g001

Live imaging of the plasma membrane formation during female gametophyte development

To analyze the relationship between the nuclear dynamics and plasma membrane formation during the cellularization, we observed their plasma membranes by labeling them with RPS5Apro::tdTomato–LTI6b and nuclei with RPS5Apro::H2B–sGFP (Fig 2A, S2 Movie). The female gametophytes were located at the center of the ovule in the early stages of development (Fig 2A; −11:20). The female gametophytes showed polar elongation toward the micropylar ends of the ovule (Fig 2A; −6:00, −5:30, −4:45). The fluorescent signals of the RPS5Apro::tdTomato–LTI6b were detected in the plasma membranes of the female gametophytes during cellularization (Fig 2A; 0:00, arrow). Cellularization of the egg and synergid cells finished after 25 minutes and 1 hour and 25 minutes, respectively (Fig 2A). The time differences between the cellularization of the egg and the synergid cells was 0.8 ± 0.2 hours (n = 14; Fig 2B). After the cellularization, the egg and synergid cells were elongated toward the chalazal end (Fig 2A; 3:50, 7:35). It took 3.7 ± 0.7 hours (n = 14) from the completion of the cellularization to the start of the elongation (Fig 2B). In the case of the micropylar end, the fluorescent signals of the tdTomato–LTI6b were detected at the side nearest the nuclei, which gives rise to the polar nucleus and the egg nucleus after cellularization (Fig 2A; 0:00). This fluorescent signal was elongated to the opposite sides of the cell membranes of the female gametophytes. The polar nuclei migrated toward the opposite sides along with the plasma membrane formation (Fig 2A; 0:00 to 1:25). In the case of the chalazal end, the fluorescent signals of the tdTomato–LTI6b were also detected between the polar nucleus and the antipodal nucleus (S2 Movie). Thus, the dynamics of the plasma membrane formation were similar at the micropylar and chalazal ends.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Live-cell imaging of cellularization and maturation during FG4–FG5.

(A) Nuclei and plasma membranes were labeled with RPS5Apro::H2B–sGFP (green) and RPS5Apro::tdTomato–LTI6b (magenta), respectively. Numbers indicate time (hr:min) from the detection of the fluorescent signal of the tdTomato–LTI6b, on the forming cell plate (arrow). Images are representative of 10 time-lapse images from 4 independent transgenic lines. (B) Differences in the time to completion of the cellularization between the egg cell and synergid cells (left) and the initiation of the cell elongation from the completion of cellularization (right) at the FG5 stage. (C) Plasma membranes were labeled with RPS5Apro::tdTomato–LTI6b. Numbers indicate time (hr:min) from the onset of observation. Images are representative of 5 time-lapse images from a transgenic line. The underlying numerical data for B can be found in S1 Data. Scale bar: 20 μm. AC, antipodal cells; CC, central cell; EC, egg cell; Pn, Polar nucleus; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g002

During the maturation of the female gametophyte cells at the FG5 and FG6 stages, the central cell showed polar elongation toward the chalazal end of the ovule (Fig 2C, S3 Movie). A bright field movie showed that the central cell became elongated by collapsing the chalazal regions of the ovule (S3 Movie). The direction of this elongation was the opposite to that of the FG2 to FG4 (Fig 2A; −11:20 to −4:45). As shown in S3 Movie, the antipodal cells appeared to be collapsing during the maturation of the central cell. However, we could not determine whether the antipodal cells degenerated or not, i.e., whether they reached FG7 (4-celled stage) or not [36] in the RPS5Apro::tdTomato–LTI6b. Although we could not observe the signature of FG7, such as the degeneration of the antipodal cells, our in vitro culture system could monitor the entire development of the female gametophyte.

Live imaging of cell fate specification during female gametophyte development

The transcriptome data of the mature ovules indicated that each female gametophyte cell had specific gene expression [37–39]. To investigate the initiation timing of the cell fate specification, we observed the egg cell marker, EC1.2pro::mtKaede [16], and the synergid cell marker, MYB98pro::GFP–MYB98 [15] (Fig 3A and 3B, S4 and S5 Movies). The fluorescent signals of EC1.2pro::mtKaede were detected in the egg cells before their elongation (Fig 3A; 0:00). Considering that the duration from egg cell cellularization to egg cell elongation was about 4 hours (Fig 2B), EC1.2 expression was initiated less than 4 hours after egg cell cellularization (Fig 2B). After 15 hours and 30 minutes, the fluorescent signals of ABI4pro::H2B–tdTomato were detected in the nucleus of the egg cell (Fig 3A; 21:10, arrowhead). Since MYB98 is an essential transcription factor for synergid cell function, the expression of MYB98 was predicted to begin after cellularization in the synergid cells; however, the fluorescent signals of MYB98pro::GFP–MYB98 were detected in not only the synergid cells but also in the egg cell and the central cell around cellularization (Fig 3B; 0:05, 0:50). As the cells mature, the GFP signals were decreased in the egg and central cells but were maintained in the synergid cells (Fig 3B; 1:50, 3:10).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Live-cell imaging of the cell fate specifications during FG4–FG5.

(A) The fluorescent signals of EC1.2pro::mtKaede were observed for the egg cell fate. Nuclei were labeled with ABI4pro::H2B–tdTomato (magenta). Numbers indicate time (hr:min) from the onset of observation. Images are representative of 4 time-lapse images from a transgenic line. Asterisk indicates the background signal in the ovule (0:00). Arrowhead indicates the fluorescent signal of ABI4pro::H2B–tdTomato. (B) The fluorescent signals of MYB98pro::GFP–MYB98 were observed for the synergid cell fate in myb98. Numbers indicate the time (hr:min) after the detection of GFP signals. Images are representative of 23 time-lapse images from a transgenic line. (C) Nuclei were labeled with EC1.1pro::NLS–3xDsRed2 (magenta) in the egg cells and LURE1.2pro::NLS–3xGFP (green) in the synergid cells, respectively, in FGR8.0. The plasma membranes were labeled with RPS5Apro::tdTomato–LTI6b (magenta). Numbers indicate the time (hr:min) after finishing the cell plate formation. Images are representative of 6 time-lapse images from a transgenic line. Arrow indicates the fluorescent signals of tdTomato–LTI6b on the forming cell plate. Arrowheads indicate the initiation of the expression of each cell-specific markers (6 hours 30 minutes). (D) Initiation of the expression of the cell-specific markers at FG5. The fluorescent signals for EC1.1pro::NLS–3xDsRed2 in the egg cells and LURE1.2pro::NLS–3xsGFP in the synergid cells were observed after completion of cell plate formation (D). (E) The fluorescent signals of EC1.1pro::SP–mTurquoise2–CTPP, DD1pro::ermTFP1, MYB98pro::mRuby3–LTI6b, and AKVpro::H2B–mScarlet–I were observed for the egg, antipodal, and synergid cell fates, and the nuclei movements, respectively. Numbers indicate time (hr:min) from the third mitosis. Images are representative of 2 time-lapse images from 2 independent transgenic lines. Arrowheads indicate the chromosomes during the third mitosis. Arrows indicate the initiation of the expression of the specific markers of the egg cell (cyan) and the antipodal cells (green) 1 hour and 40 minutes after cellularization. This timing was before the cell expansion and the polar nuclei migration. MYB98pro::mRuby3–LTI6b was detected 6 hours and 20 minutes after cellularization. The underlying numerical data for D can be found in S1 Data. Scale bars: 10 μm (A, B) and 20 μm (C, E). AC, antipodal cells; CC, central cell; EC, egg cell; Pn, Polar nucleus; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g003

To determine when the expression of each cell-specific marker began after cellularization, we utilized the female gametophyte-specific markers FGR8.0 [14] and RPS5Apro::tdTomato–LTI6b (Fig 3C, S6 Movie). Only the plasma membrane of female gametophyte cells was labeled, not sporophytic cells, in some RPS5Apro::tdTomato–LTI6b lines. After cellularization (Fig 3C; 0:00) and elongation of the egg and synergid cells (Fig 3C; 5:30), EC1.1pro::NLS–3xDsRed2 and LURE1.2pro::NLS–3xGFP signals were detected in the egg and synergid cells, respectively, in FGR8.0 (Fig 3C; 6:30). It took 5.9 ± 2.0 hours (n = 5) for the EC1.1pro::NLS–3xDsRed2 to be detected after the completion of cellularization (Fig 3D). Considering that the expression of EC1.2pro::mtKaede was initiated before egg cell elongation (Fig 3A), the detection of the NLS marker was slower than that of the mitochondrial marker. EC1.1 and EC1.2 genes encode the cysteine-rich proteins and are specifically expressed in the egg cell in Arabidopsis [40]. It is possible that EC1.1 (At1g76750) and EC1.2 (At2g21740) are activated differently. Recently, Eason and colleagues developed a sensor for transiently expressed proteins (STEP) to detect protein expression rapidly in Escherichia coli [41]. The fluorescence of a dim GFP (gSTEP) increased 11-fold by binding to STEPtag within seconds. It would be useful to analyze the expression timing of gene/protein in vivo.

To investigate the correlation between the timing of the expression of each cell-specific marker at FG5, we used the multiple cell type–specific marker line (Fig 3E, S7 Movie). We changed the target signals of the new markers from the NLS and the fluorescent proteins as detection may have been slow. The cell-specific markers of the egg cell (EC1.1pro::SP–mTurquoise2–CTPP) and the antipodal cells (DD1pro::ermTFP1) were expressed 1 hour and 40 minutes after cellularization (Fig 3E; 1:40). This was before the egg and synergid cell elongations and the polar nuclei migrations. These results suggested that each cell fate was specified almost immediately after cellularization at the 8-nucleate stage.

myb98 synergid cells showed aberrant morphology and subcellular dynamics

MYB98 is required for the formation of the filiform apparatus during the synergid cell differentiation and the expression of the AtLURE1 peptides to attract the pollen tube in the synergid cells [10,42]. However, MYB98pro::GFP–MYB98 was detected just after cellularization in the synergid, egg, and central cells (Fig 3B). To clarify the effects of the MYB98 transcription factor on the female gametophyte specifications, we observed the morphology and nuclear dynamics with the promoter activity of MYB98 in the synergid cells of the wild-type and myb98 mutant ovules (Fig 4, S8 and S9 Movies). The fluorescent signals of MYB98pro::NLS–mRuby2 were also detected in the egg and central cells, as well as the synergid cells of the wild-type and myb98 ovules. Although the nuclei were always located at the micropylar end of the synergid cells in the wild type (Fig 4A), they moved around in the synergid cells of myb98 (Fig 4B). The nuclei tracking over 14 hours also showed that the nuclei of myb98 moved closer to the chalazal end than in the wild type (Fig 4C). The large vacuoles occupied the chalazal end of the synergid cells in the wild type (Fig 4A). This polar distribution of the vacuole was disturbed in the synergid cells of myb98 (Fig 4B). In addition, the myb98 synergid cells were more elongated during maturation (Fig 4B; 2:50 to 8:20, S1 Fig). The results showed that the absence of MYB98 affected the morphology and cellular dynamics of the synergid cells in addition to the formation of the filiform apparatus [10].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Nuclear dynamics in the synergid cells of myb98.

(A) Nuclei of the synergid cells were labeled with MYB98pro::NLS–mRuby2 in the wild type. The numbers indicate the time (hr:min) from the onset of observation. Images are representative of 6 time-lapse images from a transgenic line. Dashed lines indicate the surface of the synergid cells at the micropylar end. (B) Nuclei of the synergid cells that were labeled with MYB98pro::NLS–mRuby2 in myb98. The fluorescent signals of MYB98pro::GFP were observed for the synergid cell fate. Images are representative of 12 time-lapse images from a transgenic line. The arrowheads indicate the vacuoles in the synergid cells. The arrows indicate the GFP signals in the egg cells. Scale bar: 20 μm. (C) Nuclei positions on the micropylar–chalazal axis were plotted in each synergid cell in the wild type and myb98 from S8 and S9 Movies. Each point indicates the time corresponding to the color bar. The leftmost point indicates the start time. The y-axis indicates the distance from the micropylar end of the synergid cell. The underlying numerical data for C can be found in S1 Data. EC, egg cell; Pn, Polar nucleus.

https://doi.org/10.1371/journal.pbio.3001123.g004

Previously, it has been reported that MYB98pro::GFP is expressed in all cells of the female gametophyte, except for the antipodal cells at FG5 [43]. The fluorescent signals of the GFP–MYB98 and NLS–mRuby2 were also detected in the synergid, egg, and central cells just after cellularization (Figs 3B and 4A). Except for the synergid cells, the fluorescent signals of GFP–MYB98 and NLS–mRuby2 were decreased as the cells matured. These results suggested that the synergid cell fate stabilized the gene expression of MYB98. The ectopic expression of MYB98pro::GFP and MYB98pro::NLS–mRuby2 was not detected after the restrictions of the expression in the synergid cells of myb98 mutant (Fig 4B). This suggested that the egg and central cells regularly maintain their cell fates. The promoter activity of MYB98 was normal in myb98 (Fig 4B), which also suggested that the maintenance, not initiation, of the synergid cell fate was defective in myb98. Considering these results, the positional information of the nuclei is essential for the initiation of the synergid cell fate.

Gene expression analysis of the female gametophyte cell

Previous studies have supported the lateral inhibition model for the differentiation of the female gametophyte cells. Although all cells in the female gametophyte have the gametic cell competence, the accessory cells like the synergid and antipodal cells are repressed in the gametic cell fate [4,6]. To investigate the gene expression profiles of the synergid cells in the wild-type and myb98 mutant, we established a method to isolate them in Arabidopsis. We treated the ovules in emasculated ovaries of the transgenic marker line for the synergid cells, MYB98pro::GFP, with enzyme solutions (Fig 5A). The protoplasts of the synergid cells were released from the ovules through their micropyles with enzyme treatment for 30 to 60 minutes (S2A Fig), and those with GFP signals were collected by micromanipulation (Fig 5A). Initially, the synergid cell–derived protoplasts were mostly associated with other GFP-negative ovular cells, probably due to insufficient cell wall digestion. To increase the efficiency of the single synergid cell isolation, we optimized the following 2 conditions. One was the calcium nitrate in the enzyme solution as the calcium ion was suggested to inhibit the degradation of the cell wall [44]. Subsequently, the removal of calcium ion from the enzyme solution decreased the adhesion of protoplasts and increased the frequency of the collectable synergid cells that were released as single cells (S2B Fig). The other condition was the pH of the enzyme solution. We found that the protoplasts began to decrease the GFP fluorescence in a short period and eventually ruptured after the cell surface gradually became rough, and this may be related to the decreases in viability. We performed the enzyme treatments at pH 5.0 to 9.0 and observed the GFP fluorescence as a vital indicator of the protoplast [45]. The rate of the GFP-positive synergid protoplasts was highest at pH 7.0, which was the best for the isolation of the synergid cells (Fig 2, S2 Fig). The optimized enzyme solutions allowed us to collect pure synergid cells with high efficiency (Fig 5C and 5D). To isolate other types of female gametophyte cells, we examined the enzyme solution treatment with the ovules of each marker line, EC1.2pro::mtKaede and FWApro::FWA–GFP, for the egg and central cells, respectively [16,17]. The protoplasts of the 2 gametic cells were also detached from their ovules through the micropyle (Fig 5F–5K).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Isolation of the female gametophyte cells.

(A) Scheme for the isolation of the female gametophyte cells. The ovules of the marker lines for the synergid (MYB98pro::GFP) (B), egg (EC1.2pro::mtKaede) (F), and central cells (FWApro::FWA–GFP) (I). We isolated each type of cell. Synergid cells in the wild-type (C, D) and myb98 mutant (E). (G, H) Egg cells. (J, K) Central cell. Scale bar: 20 μm. CC, central cell; EC, egg cell; mSY, synergid cell of myb98 mutant; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g005

We then performed RNA-seq to analyze the gene expression profiles of the collected the synergid, egg, and central cells in the wild type and the synergid cells in myb98 mutant (Fig 5E). RNA-seq data from these female gametophyte cells were mapped to the genome of Arabidopsis (TAIR version 10) with the published sequence data from the ovules at 12 hour after emasculation (HAE) [46] and 2-week-old seedlings [47]. There were 5,007 to 13,073 genes (TPM > 1) detected in each sample (Fig 6A; S4 and S5 Tables). Transcripts from the same cell type in different biological replicates were highly correlated (S3A Fig). The principal component analysis (PCA) indicated that PC1 (31.8%) and PC2 (15.0%) were sufficient for separating these samples into the 6 groups (Fig 6B, S3B and S3C Fig). These results suggested that our datasets had a high level of reproducibility and reflect the intermediary state of the myb98 synergid. We identified the DEGs among the central, egg, and synergid cells in the wild type (S7–S9 Tables) and between the synergid cells in the wild-type and myb98 mutant (S10 and S11 Tables). GO enrichment analysis of the DEGs that were specifically expressed in the wild-type female gametophyte cells indicated that the GO terms of each cell type were related to protein secretion and cell wall synthesis in the synergid cells; fertilization, DNA methylation, and small regulatory RNA in the egg cells; and photosynthesis and interaction with other cells (cell killing and defense response) in the central cells (Tables 1 and 2, S12–S14 Tables). These results are consistent with the known functions of each cell type. The synergid cells play a role in pollen tube guidance through the secretion of many peptides, including AtLUREs, the filiform apparatus, and thickened cell wall structure [42]. GO terms supported that RNA-based gene silencing and DNA methylation mechanisms have important roles in the egg cells. The central cells play a role in pollen tube guidance through regulating the expression of cysteine-rich peptides (CRPs) from the synergid and central cells [42,48,49]. As in the present study, photosynthesis has recently been reported as a central cell–specific GO term [50]. Including other GO terms, the central cells have high metabolite levels (S14 Table). Interestingly, several egg cell–specific genes were highly expressed in the mutant synergid (S4 Fig, S6 Table). We examined the expression patterns of the DEGs in the synergid dataset among all samples (Fig 6C). The cluster of mutant synergids was closer to that of the egg cells than the synergid cells in the wild type. These results also indicated that the expression pattern of the myb98 mutant synergid was partially changed to be egg cell–like.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. RNA-seq of the female gametophyte cells.

The biological replicates were sequenced for 2 SY, 2 CC, 3 EC, and 4 mSY cells. (A) Venn diagram of the expressed genes (TPM > 1) in each cell type. (B) The PCA analysis (PC1 vs. PC2) of all transcriptome data, the female gametophyte cells, ovules, and seedlings. (C) Heatmap of the DEGs between the synergid cells in the wild-type and the myb98 mutant. The underlying numerical data for (B, C) can be found in S1 Data. CC, central cell; EC, egg cell; mSY, synergid cell of myb98 mutant; OV, ovule; RNA-seq, RNA sequencing; SL, seedling; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. GO enrichment analysis of DEGs in the synergid cells.

https://doi.org/10.1371/journal.pbio.3001123.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. GO enrichment analysis of DEGs in the egg cell.

https://doi.org/10.1371/journal.pbio.3001123.t002

Thus, the RNA-seq of the female gametophyte cells identified many of the DEGs and the highly expressed genes in each type of cell (S5–S11 Tables). We compared the DEGs between the wild type and myb98 identified by this RNA-seq study with those identified by microarrays [38]. The number of up-regulated genes in myb98 was 204 and 40 from the RNA-seq and microarray studies, respectively (S2D Fig). The number of down-regulated genes in myb98 was 188 and 77 from the RNA-seq and microarray studies, respectively (S2E Fig). These results suggested that cell-specific RNA-seq had much higher sensitivity for the detection the DEGs than the microarrays because of the number of DEGs. Although 70 down-regulated genes in myb98 overlapped between RNA-seq and microarray data, only 4 up-regulated genes in myb98 overlapped (S2D and S2E Fig). The differences in the up-regulated genes of myb98 may be caused by the wild type background or the developmental stage for the sampling [38]. Furthermore, our RNA-seq revealed that the gene expression profiles of the myb98 mutant synergid changed partially to the egg cell–like (S6 Table). The RNA-seq analysis conducted here allowed for the isolation of single cell types and mutants and thus enabled the detection of cell-specific changes. Our results provide evidence of the utility of this method for the investigation of cell fate specification mechanisms.

MYB98 was reported as the gene that controlled the characteristic development of the synergid cells [10]. The myb98 synergid was like a deficient egg cell because an important factor for the synergid cell fate was lost. The expression of AtLURE1 was also decreased in the myb98 synergids [42]. In the present study, RKD1 and RKD2 and RKD2-induced genes were identified as egg cell DEGs, myb98 DEGs [51]. RKD1 and RKD2 were up-regulated in a replicate of myb98 synergid cells. RKD2-induced genes tended to be up-regulated in myb98 synergid cells. Moreover, EC1.5 was detected as a DEG in myb98 synergid cell. Other EC1s also tended to be up-regulated in myb98 synergid cells. The PCA analysis and the difference of gene expressions reflect the intermediary state of myb98 synergid (Fig 6B and 6C, S3B and S3C Fig). Further research is required to identify if the synergid cells of myb98 function as egg cells, synergid cells, or both.

Egg cell–specific markers were expressed in one of the synergid cells of myb98

To confirm the expression patterns of the egg cell–specific genes in myb98, we analyzed the CDR1–LIKE aspartyl proteases, which are highly expressed in the egg cells (S6 Table, S4 Fig). CONSTITUTIVE DISEASE RESISTANCE 1 (CDR1) was previously found to be involved in the peptide signaling of disease resistance [52]. The phylogenetic analysis showed that Arabidopsis contained 2 distinct groups of CDR1s: a CDR1–LIKE2 (At1g31450)/CDR1–LIKE1 group (At2g35615) and a CDR1 (At5g33340)/CDR1–LIKE3 (At1g64830) group [53] (S5 Fig). CDR1–LIKE2pro::CDR1–LIKE2–mClover (hereafter CDR1L2–mClover) and CDR1–LIKE1pro::CDR1–LIKE1–mClover were expressed only in the egg cells, while CDR1pro::CDR1–mClover was expressed in the central and antipodal cells (S5B Fig, S10 Movie). These localizations were consistent with the groupings of the CDR1s by the phylogenetic analysis. Although the fluorescent signals of CDR1L2–mClover were limited to the egg cell after cellularization in the wild type (Fig 7A, S11 Movie, Table 3; 100%, n = 14), the myb98 mutant had supernumerary cells with CDR1L2–mClover signals at the micropylar end (Fig 7B, S11 Movie, Table 3; 71%, n = 17). Initially, the CDR1L2–mClover signal was limited to a single cell at the egg cell position (Fig 7B; 0:00). However, 9 hours and 30 minutes after signal detection in the egg cell, the CDR1L2–mClover signal was also detected in one of the synergid cells (Fig 7B; 9:30). In most cases, one of the synergid cells expressed CDR1L2–mClover in myb98 (Table 3; 65%, n = 17).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Expression of EC–specific genes in the myb98 synergid cells.

(A, B) The expression of CDR1L2–mClover in the wild-type (A) and myb98 mutant (B) ovules. The numbers indicate the time (hr:min) from the onset of observation. Images are representative of 14 (A) and 17 (B) time-lapse images from 2 independent transgenic lines, respectively. The arrow indicates the CDR1L2–mClover signals in the synergid cell of myb98. (C, D) The expression patterns of SBT4.13pro::SBT4.13–mClover in the wild-type (C) and myb98 mutant (D) ovules. The numbers indicate the time (hr:min) from the first detection of SBT4.13–mClover. Images are representative of 36 (C) and 44 (D) time-lapse images from 3 (C) and 4 (D) independent transgenic lines, respectively. The fluorescent signals of SBT4.13–mClover were only detected in the egg cells of the wild type. (C). However, in the case of myb98, the fluorescent signals of SBT4.13–mClover were also detected in the synergid cells (D). (E) The expression patterns of the female gametophyte-specific markers in myb98. The numbers indicate the time (hr:min) from the onset of observation. At first, MYB98pro::mRuby3–LTI6b was detected in the 2 synergid cells (0:00). The arrows indicate the EC1.1pro::SP–mTurquoise2–CTPP expression in one of the synergid cells (3 hours and 00 min, 5 hours and 00 min). The arrowhead indicates no expression of EC1.1pro::SP–mTurquoise2–CTPP (3 hours and 00 min). The upper and lower panels indicate different z planes. Five ovules (50%, n = 10) showed EC1.1pro expression in the synergid cell, and the remaining 5 ovules were wild type–like. Scale bar: 20 μm. AC, antipodal cells; EC, egg cell; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Expression of EC-specific genes in the female gametophyte cells.

https://doi.org/10.1371/journal.pbio.3001123.t003

Previously, SBT4.13 was identified as an egg cell–specific gene [54]. We also analyzed the expression patterns of SBT4.13 in the wild-type and myb98 mutant ovules (Fig 7C and 7D). The fluorescent signal of SBT4.13pro::SBT4.13–mClover was detected only in the egg cell before its elongation (Fig 7C; 0:00 to 1:30, S12 Movie). This expression timing of SBT4.13pro::SBT4.13–mClover was similar to that of EC1.2pro::mtKaede (Fig 3A). The myb98 ovules showed 2 patterns of SBT4.13pro::SBT4.13–mClover in the female gametophyte (Fig 7D, S13 Movie). One was the expression of SBT4.13pro::SBT4.13–mClover in the synergid and the antipodal cells in addition to the egg cells of the myb98 ovules (S6A Fig; first half of S13 Movie; Table 3; 30%, n = 44). The other was the synergid and egg cells (Fig 7D; second half of S13 Movie, Table 3; 68%, n = 44). Similar to the results for CDR1L2–mClover, one of the synergid cells showed SBT4.13pro::SBT4.13–mClover expression in myb98 (Table 3; 61%, n = 44).

To determine whether egg cell–specific markers were expressed in 1 or 2 synergid cells more clearly, we observed the myb98 ovules in the multiple cell type–specific marker line (Fig 7E, S14 Movie). After detection of the MYB98pro::mRuby3–LTI6b signal in the 2 synergid cells, the signal of EC1.1pro::SP–mTurquoise2–CTPP was detected in one of the synergid cells (Fig 7E; 3:00, 5:00, 8:00). Thus, one of the synergid cells showed cell fate conversion to an egg cell in myb98. In myb98, SBT4.13pro::SBT4.13–mClover was expressed almost simultaneously in the egg and synergid cells during cell elongation, whereas CDR1L2–mClover was expressed in the synergid cells after egg cell maturation. The expression of SBT4.13pro::SBT4.13–mClover in the synergid cells from an early stage indicated that myb98 synergid cells had changed their cell fate from the early stage. These results suggest that the expression regulation of each egg cell–specific gene is different.

Although egg cell–specific genes were also expressed in one of the myb98 synergid cells, the myb98 pistils had only 1 embryo after fertilization, like the wild type (S7 Fig; 63 ovules from 10 pistils). This indicated that the synergid cells with the egg cell–specific genes were not functional for fertilization in myb98. The additional egg-like cells appeared to not be functional in lis, clo, ato, and wyr [6,8,9]. However, amp1 has twin embryos, and eostre has twin zygote-like cells, indicating that these additional egg-like cells are functional for fertilization [5,7]. These differences in the gene expression of mutants may provide clues as to the acquisition of egg cell functions.

Subcellular dynamics in female gametophyte development

To date, the female gametophyte of Arabidopsis has been analyzed only in fixed samples, so the actual developmental time course and subcellular dynamics were not known [27]. One of the major events that could not be seen in the fixed samples was that the vacuoles were dynamic in the female gametophytes. In the previous schematics, the vacuoles were drawn as large and only in the center of the cell [55]. When the polar nuclei migrated to fuse with each other at FG5, they were described as moving along the periphery of the female gametophyte to avoid the large vacuole [56]. However, the observations from the present study showed that the polar nuclei migrated linearly to fuse and adhere to the vacuole in the middle of the cell at shorter distances (Figs 1 and 8). This suggests that the vacuoles of the female gametophyte did not remain large and static, but changed shape dynamically. The dynamics of the vacuoles have been seen in Arabidopsis and tobacco BY-2 cultured cells, and this plasticity is due to actin filaments [57,58]. As actin filaments were also involved in the nuclear migrations during gamete fusion, the linear migration of the polar nuclei was also expected to involve them [20]. In the mature central cells after polar nuclei fusions, the nucleus of the central cells was located to the micropylar end, and the actin filaments played an important role in the positioning of the nucleus [59]. The vacuoles were located at the chalazal end of the synergid cells and the micropylar end of the egg cells, thus appearing to limit the nuclear migration (Fig 3A and 3B). In the case of the myb98 mutant, the vacuoles were dynamic, causing the nuclei to move around and not stay in one place (Fig 4B and 4C). It is considered that this nuclear movement promoted the expression of the egg cell markers in the synergid cells of myb98. The gene expression analysis showed an intermediary state between the synergid and egg cells in the myb98 synergid, suggesting that this nuclear movement may appear as a mixture of that seen in egg and synergid cells (Fig 6B and 6D, S3B and S3C Fig). Strong correlations between the nuclear position and the cell fate were shown in several mutants [5–9]. However, it remains unclear whether the nuclear position determines gene expression or if gene expression determines the nuclear position. Manipulation of nuclear behavior with the in vitro ovule culture systems will help to reveal the mechanisms of cell fate specifications in the development of the female gametophytes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Schematic illustration of the dynamics of female gametophyte development in Arabidopsis.

Yellow arrows show the direction of nuclear divisions. Blue arrows show the direction of cell elongation of the female gametophyte. Magenta arrows show polar nuclear migration at FG5. The time (mean ± standard deviation) was calculated from the movies. EC, egg cell; SY, synergid cell.

https://doi.org/10.1371/journal.pbio.3001123.g008

Cell–cell communication between the 2 synergid cells

An interesting phenotype of the myb98 mutant was in one of the 2 synergid cells that tended to be converted to an egg cell fate (Table 3; 92% for CDR1L2–mClover, n = 12, 88% for SBT4.13pro::SBT4.13–mClover, n = 43). Some mutants show similar phenotypes with additional egg cells [5–9]. In the amp1 mutant, 19% of the ovules showed EC1.1pro::HTA6–3GFP expression in both synergid cells, whereas 26% of the ovules showed this expression in only one of the synergid cells (notably, 45% of the ovules had no detectable fluorescent signal) [5]. The combination of cell–cell communication and flexible fate maintenance might allow for only one of the 2 synergid cells to become an egg cell. The synergid cells play an important role in pollen tube attraction through secreting peptides [60]. In the case of the ovule, which has been converted from both synergid cells to the egg cell fate, it cannot attract pollen tubes. Therefore, it is expected that plants may have a mechanism, which is independent of the MYB98, to retain not only the egg cell but also the synergid cell for pollen tube attraction and fertilization. Previously, we found that the laser disruption of the immature egg cells affects the cell differentiation for one of the synergid cells in Torenia fournieri [61]. Lateral inhibition from the egg cell restricts the egg cell fate in accessory cells [14]. Based on these findings, we speculate that egg cell failure induced a decrease in MYB98 expression. The results presented here raise 2 possibilities. One possibility is that the synergid cell acquired the egg cell fate, preventing the remaining synergid cell from obtaining the egg cell fate. However, we observed egg cell–specific gene expression in both synergid cells in only a few cases (11%, n = 55). In this case, the inhibition signal needs to be quickly transmitted to the remaining synergid cell. The other possibility is that the cell–cell communication between the 2 synergid cells decides which cells convert to the egg cell fate. Exploration and imaging of signal molecules involved in this cell–cell communication would help to reveal which of these possibilities occurs.

Our results suggested that the cell fate specifications are immediately initiated around the time of cellularization, depending on the positional information of the nucleus. Moreover, failure of cell fate maintenance, like that of the myb98 mutant, induced cell fate conversion from the adjacent accessory cells to gamete cell fates for fertilization. Previously, the existence of cell–cell communication between gametic and accessory cells, such as lateral inhibition from the egg cell to synergid cells, was proposed [4]. We propose that the synergid cells communicate with each other to determine their fate and behavior, and such flexibility compliments for the robustness of plant fertilization. Further studies, such as single-cell transcriptome profiling of the mutant synergids, will provide novel insights into the molecular mechanisms of the cell–cell communications in the cell fate specification of plants.