Co-purification of sperm cell and vegetative nuclei

In order to improve and simplify the SC-sorting method and to additionally co-purify the VN from the same genetic background, we generated a transgenic line expressing distinct fluorescent proteins in both nuclei. The ACT11 promoter driving histone H2B fused to mRFP was used as a VN marker[17]. A homozygous plant harboring the ACT11p::H2B-mRFP transgene in Col-0 background was crossed with a sperm-specific marker line of the same ecotype, harboring a MGH3p::MGH3-eGFP construct, which encodes a male germline-specific histone variant[18, 19]. MGH3 is expressed as early as bicellular pollen, and is highly abundant in the sperm nuclei of mature pollen (Figure1A, B). Progeny resulting from this cross were allowed to segregate by self-pollination, in order to obtain stable double homozygous plants (Figure1C). Studying the expression pattern of both transgenes throughout pollen development revealed that ACT11p::H2B-mRFP is initially expressed in the generative cell (GC) of bicellular pollen, but preferentially expressed in the VN at the mature pollen stage (Figure1A, B). We did not observe any pollen phenotypes in homozygous lines for both transgenes, and the plants are fully fertile, thus indicating that ectopic expression of an additional H2B gene in the VN and an extra copy of MGH3 in the sperm cells did not result in any significant change in chromatin structure and dynamics in both nuclei.

Pollen grains were collected from open flowers by vortexing with Galbraith buffer[20] (see Materials and Methods), and released pollen was then disrupted by additional vortexing with glass beads. Enriched filtrates containing SC nuclei and VN released from broken pollen were co-purified by FACS based on their distinct fluorescence properties (Figure2A). We inspected purity of sorted samples by microscopy (Figure2B) and RT-PCR on SC and VC-specific transcripts. MGH3 is only expressed in the SC and was not detected in the sorted VN fraction, while VEX1, a VC-expressed gene, was equally not detected in the SC fraction (Figure2D), indicative of two pure populations. Re-analyzing sorted populations stained with 4',6-diamidino-2-phenylindole (DAPI) confirmed that the SC and VN populations were consistently more than 99% pure (Figure2C). Furthermore, the ratio SC/VN before sorting was consistently 2:1, demonstrating a good recovery for both types of nuclei after pollen disruption (data not shown).

As Galbraith’s buffer was originally conceived to sort and analyze nuclei by flow cytometry[20], we tested our FACS method using a sperm extraction buffer (SEB) that has been previously used to sort intact and viable sperm cells[2]. In order to test sperm cell viability before and after sorting, we stained the cells with SYTOX Orange dye that stains DNA of dead cells. As expected, sorted GFP positive events prepared in Galbraith buffer stained 100%, while only 19% of the cells stained after isolation in SEB (Figure3). This result indicates that whenever intact and viable sperm cells are required for downstream applications, the cells should be prepared in SEB instead of Galbraith buffer. We could also confirm that this buffer is suitable for VN sorting, although the efficiency is significantly lower (data not shown). SYTOX staining can be additionally used to exclude dead sperm cells from sorted fraction.

The procedure was optimised to obtain higher yields of VN and SC nuclei/cells, maximising the number of sorted events per second, though the whole procedure can be scaled-down to approximately 150mg of fresh weight material in a 2 ml eppendorf tube. 0.8-1.5 million SC can be obtained from a 10 ml flower batch, while a 2 ml batch of open flowers yields approximately 100.000 sperm cells.

Microspore purification by FACS

To understand the genetic and epigenetic mechanisms that control VN and SC differentiation, analysis of their precursor microspore cell is essential. Available methods to isolate Arabidopsis microspores relied on Percoll density gradients[21], and would not provide a sufficiently pure fraction suitable for genomic analysis at the DNA level, besides the problem of relatively low yield. Previous attempts to isolate microspores by flow cytometry were successfully applied in Brassica napus[22], however, to our knowledge these same methods were not tried in Arabidopsis. We used closed flower buds of wild type plants to collect all different stages of pollen development, by grinding in pollen extraction buffer (PEB) (see details in Materials and Methods). The method to purify SC and VN revealed that intact pollen grains have a high amount of autofluorescence (population D in Figure2A). As such, we explored this property in addition to the smaller size of microspores in comparison with other stages of pollen development (Figure4A). The sorted population was analyzed by microscopy and confirmed to contain mostly microspores (97.7%) (Figure4B and C). A very small fraction (2.3%) of early bicellular pollen was also observed (Figure4C). As these are probably cells that have just gone through the first pollen mitosis, they have identical size and autofluorescent properties as microspores, and are therefore impossible to distinguish with our FACS settings.

Plant material, growth conditions and transgene cloning

Transgenic Arabidopsis plants were used for both sperm cell and pollen isolation. Plants were sown on soil and grown for 8 weeks in short-day conditions (8 h light at 21°C-23°C) and then transferred to long-day conditions (16 h light) to induce flowering. MGH3p::MGH3-eGFP was obtained by cloning 1.2 kb upstream of the MGH3 transcriptional start site, together with the MGH3 genomic sequence without the stop codon, into the pMDC107 vector[25] by gateway cloning. The MGH3 coding sequence contains all endogenous introns, which might help for stable transcription and accumulation to high levels. Primers used for MGH3p::MGH3-eGFP cloning are presented in Additional file1. Double homozygous plants harbouring MGH3p::MGH3-eGFP and ACT11p::H2B-mRFP[17] transgenes, were obtained by crossing individual homozygous lines. This seed stock is available from ABRC under stock number CS67829.

Purification of sperm cells, vegetative nuclei and sperm nuclei

Open flowers were collected into a 50 ml falcon tube, filled roughly with a volume of 10 ml of fresh material. Flowers were vortexed at medium speed (Fisher Scientific, TOP-mix 3 IKA, speed 2000) in 10 ml of Galbraith buffer (45 mM MgCl₂, 30 mM Sodium Citrate, 20 mM MOPS, 1% Triton-100, pH to 7.0) for 3 minutes, at room temperature. This crude fraction was then filtered through a Miracloth mesh (Calbiochem) to remove flower parts and centrifuged for 1 minute at 2600 g to pellet pollen. The supernatant was carefully removed with minimal disturbance and the pollen enriched pellet was resuspended in approximately 1.5 ml of fresh buffer. This pollen enriched fraction was then transferred to a 1.5 ml eppendorf tube containing 100 μl of acid-washed glass beads (425–600 μm, Sigma), and vortexed continuously at maximum speed (2500) for 4 minutes in order to break mature pollen grains. The fraction containing the released nuclei was then filtered through a 28μm mesh (SEFAR) to exclude unbroken and hydrated pollen. The nuclei enriched solution was ready for FACS at this point, which included debris from broken pollen. Hydrated pollen that remained intact and thus was retained in the filter can be recovered by washing the 28 μm mesh in new buffer and used in a second extraction step with the glass beads.

If intact and viable SC are desired for downstream applications (e.g. RNA extraction or in vitro manipulation) the whole procedure should be performed with sperm extraction buffer (SEB) (1.3 mM H₃BO₃, 3.6 mM CaCl₂, 0.74 mM KH₂PO₄, 438 mM sucrose, 5.83 mM MgSO₄, 7 mM MOPS at pH 6), which assures that 75-80% of the sorted sperm cells are viable and intact.

Purification of Microspores from young flower buds

Closed flower buds (approximately 20 ml) were gently ground using mortar and pestle in 10 ml of pollen extraction buffer (PEB) (10 mM CaCl₂, 2 mM MES, 1 mM KCl, 1% H₃BO₃, 10% Sucrose, pH 7.5) in order to release the spores. This crude fraction was initially filtered through Miracloth (Calbiochem) to remove bigger debris, and concentrated by centrifugation (800 g, 5 min) in 15 ml falcon tubes. The resulting yellowish pellet enriched in microspores was resuspended in 1.5 ml of PEB, and filtered through a 20 μm mesh (Partec, CellTrics) before FACS to further enrich for microspores and exclude tricellular and mature pollen.

Fluorescence activated cell sorting

Fluorescent activated cell sorting was carried out with a MoFlo (Beckman Coulter, Fort Collins, USA) with a 488 nm laser (200 mW air-cooled Sapphire, Coherent) at 140 mW used for scatter measurements (Low angle or Forward Scatter, and High angle or Side Scatter; FSC and SSC, respectively) and for GFP excitation, and a 561 nm laser (50 mW DPSS, CrystaLaser) at 38 mW for RFP excitation. GFP and RFP were detected using a 530/40 nm and a 630/75 nm bandpass filters, respectively. FSC was used for triggering, and threshold had to be low to avoid missing the sperm cell population, whose size is on average 2.5 μm in diameter. Phosphate Buffer Saline (PBS) was used as sheath, and run at a constant pressure of 400 kPa (~60 psi). Frequency of drop formation was approximately 96,000 Hz. Even though pollen was present in the sample (broken and intact) it did not interfere with drop formation or break-off, as we were able to sort pollen under the same conditions as for SC and VN. Sorting rates were typically 2 million SC and 1 million VN per hour, i.e. an average rate of 500 SC and 250 VN per second, respectively.

Viability tests were performed with approximately 50.000 sorted SC and VN by staining with SYTOX Orange (Molecular Probes, Invitrogen) at a final concentration of 25nM. Sperm cells negative for SYTOX Orange and positive for GFP were considered viable (intact membrane) and considered compromised or to represent bare nuclei (only the nuclei stained), if positive for both SYTOX and GFP. SYTOX Orange was excited with the 488 nm laser and detected with a 580/20 nm bandpass filter in the Moflo.

Purity was determined by running aliquots of sorted cells in a CyAn ADP flow cytometer (Beckman Coulter, Fort Collins, USA). A 488 nm laser was used to excite both GFP and RFP, detected with 530/30 and 616/21 nm bandpass filters, respectively. 1 μM DAPI (Sigma) was added to the sorted cells and incubated on ice for 5 min, to discriminate between nuclei containing cells and electronic noise. DAPI was excited with a 405 nm laser and detected with a 450/40 nm bandpass filter.

Microspores were also purified by FACS using a Moflo high-speed cell sorter. The machine was used in a standard configuration, using a 100 μm ceramic nozzle with PBS running at a constant pressure of 200 kPa (~30 psi), and a drop-drive frequency of approximately 30,000 Hz. The 488 nm laser line was used for scatter measurements and autofluorescence excitation, which was detected in the GFP channel using a 530/40 nm bandpass filter. Microspores and other stages of pollen development were identified by their elevated high angle scatter (SSC) and autofluorescence properties (observed in the GFP channel). Within this population, microspores were selected by their characteristic smaller size, captured by a diminished low angle scatter (FSC) and time-of-flight (Pulse Width).

RT-PCR

Semi-quantitative RT-PCR was performed with total RNA isolated from approximately 100,000 cells/nuclei that were sorted directly into Tri Reagent LS (Sigma). First-strand cDNA (Oligo-dT primed) was synthesized in 25 μl reactions using the MLV reverse transcriptase - RNaseH minus (Promega) according to manufacturer instructions. 2 μl of non-diluted cDNA was used as a template for 30 PCR cycles. Primers used are listed in Additional file1.