M. polymorpha accession Takaragaike-1 (Tak-1; Ishizaki et al., 2008) was grown asexually on 1/2× Gamborg’s B5 medium containing 1.4% (w/v) agar at 22 °C under continuous white light. To induce antheridiophores, approximately two-week-old thalli were further cultivated on vermiculite soaked in 1:1000 Hyponex (HYPONeX JAPAN) at 22 °C under continuous white light for two weeks and then cultivated at 22 °C under continuous white light supplemented with far-red light. The transgenic plants used in this study are listed in Supplementary Table 1 .

For nomenclature of genes, proteins, and mutants of M. polymorpha, we followed Bowman et al. (2016). Gene IDs were obtained from MarpolBase (http://marchantia.info/), genome version 5.1 (Montgomery et al., 2020). The primer sequences used in this study are listed in Supplementary Table 2 .

To construct pENTR MpPGM1, CDS for MpPGM1 (Mp4g13750) was amplified using PCR from cDNA prepared from Tak-1 thalli and subcloned into pENTR/D-TOPO (Thermo Fisher Scientific). To construct pENTR MpPGM1-mTurquoise2 (monomeric Turquoise2), the CDS for mTurquoise2, amplified by PCR, was inserted into the AscI site of the pENTR MpPGM1 vector using the In-Fusion HD Cloning System (Clontech). To construct pENTR TP-sGFP, the sequence for the transit peptide (TP) of MpSIG2 (Mp4g13380; Kanazawa et al., 2013) was amplified via PCR from cDNA prepared from the Tak-1 thalli. The amplified fragments of TP_MpSIG2 and cDNA for sGFP were conjugated using PCR via homologous double crossover, and the amplified product was subcloned into pENTR/D-TOPO. The sequences flanked by the attL1 and attL2 sites in pENTR MpPGM1, pENTR MpPGM1-mTurquoise2, or pENTR TP-sGFP were introduced into pMpGWB308, pMpGWB103, or pMpGWB303 (Ishizaki et al., 2015), respectively, using the Gateway LR Clonase™ II Enzyme Mix (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Transformation of M. polymorpha was performed as previously described (Norizuki et al., 2022). Transformants were selected on plates containing 10 mg/L hygromycin B and 250 mg/L cefotaxime for the pMpGWB103 MpPGM1-mTurquoise2 vector, and 0.5 µM chlorsulfuron and 250 mg/L cefotaxime for the pMpGWB308 MpPGM1-Citrine and pMpGWB303 TP-sGFP vectors.

To observe thallus cells expressing MpPGM1-Citrine, five-day-old thalli were used.

To observe stage 0–4 spermatids, antheridia were fixed with 4% (w/v) paraformaldehyde and their cell walls were digested with 1% (w/v) Cellulase Onozuka RS (SERVA) and 0.25% (w/v) Pectolyase Y-23 (Kyowa Chemical Products), as previously described (Norizuki et al., 2022). The samples were incubated in PME buffer (50 mM PIPES, 5 mM EGTA, and 1 mM MgSO₄, adjusted to pH 6.8 using NaOH) containing 1 µg/mL Hoechst 33342 (Dojindo).

Spermatozoids (stage 5) were obtained by placing a mature antheridiophore upside-down on a drop of 20–30 μL water for 1 min and fixing in 4% (w/v) paraformaldehyde in PBS buffer (150 mM NaCl, 80 mM Na₂HPO₄, and 40 mM NaH₂PO₄, adjusted to pH 6.8, using NaOH) for 5 min. Fixed spermatozoids were centrifuged at 9,100 × g for 1 min, and the pellet was suspended in PBS buffer containing 1 µg/mL Hoechst 33342.

For simultaneous observation of plastid and vacuole markers in spermatids, antheridiophores were manually sectioned using a razor blade.

For immunostaining, fixed and cell-wall-digested antheridial cells were used, as previously described (Norizuki et al., 2022). As a primary antibody, the anti-double-stranded DNA (dsDNA) antibody (HYB331-01; Santa Cruz Biotechnology) was used at a 200× dilution. As a secondary antibody, Alexa Fluor™ 546 goat anti-mouse IgG (H+L) (Thermo Fisher Scientific), or goat anti-mouse IgG (H+L) Alexa Fluor™ Plus 488 (Thermo Fisher Scientific) were used at a 1000× dilution. The samples were mounted using ProLong™ Diamond Antifade Mountant (Thermo Fisher Scientific) and incubated for at least 24 h at room temperature in the dark.

For confocal microscopic observations, an LSM780 confocal microscope (Carl Zeiss) equipped with an oil immersion lens (×63, numerical aperture = 1.4) was used.

For TEM observation, wild-type and Mpatg5-1^ge antheridia were subjected to electron microscopy, as described previously (Norizuki et al., 2022).

Maximum-intensity projection images created from z-stacked images using ImageJ (version 1.50i; National Institute of Health) were used to calculate plastid circularity. To calculate the occupancy of starch granules in the area of the plastid, the total area of starch granules, which were observed as electron-dense structures, and the area of the plastid were calculated using ImageJ. The number of the large spherical vacuole and the plastid were counted using single-sectioned 22.49 μm × 22.49 μm images. To test the normality of the data, the Shapiro–Wilk test was performed using R (version 3.6.0; The R project), and samples were considered nonparametric when the p value was less than 0.05. For comparison between two groups, the Wilcoxon rank-sum test (for nonparametric samples) or Welch’s t test (for parametric samples) was performed using R software. For statistical analyses among three or more groups, the Steel–Dwass test was performed using R. Details of the statistical methods are indicated in each figure legend.

To observe how the plastid changes its shape during spermiogenesis, we generated transgenic M. polymorpha expressing fluorescently tagged MpPGM1. PGM genes encode phosphoglucomutase (PGM), and plastidal PGM (AtPGM1) is required for starch synthesis in plastids of Arabidopsis thaliana (Caspar et al., 1985; Periappuram et al., 2000). In M. polymorpha thallus cells, MpPGM1-Citrine was localized in chloroplasts, overlapping with autofluorescence from chlorophyll, confirming that MpPGM1-Citrine can be used as a plastid marker in M. polymorpha ( Figure 1B ). Using this marker, we observed plastid reorganization during spermiogenesis, which can be divided into 1 + 5 stages (stages 0–5) based on flagellar formation and nuclear shape (Minamino et al., 2022). In cells at stage 0, which included spermatogenous cells, spermatid mother cells, and spermatids before the appearance of flagella, tentaculate plastids were observed. At stages 1–4, one plastid with a spherical or lens-like shape was observed in each cell. Notably, at stage 4, an area with a weaker fluorescent signal from MpPGM1-Citrine was detected around the center of the plastid (the arrow in Figure 1C ), which could be the region where the posterior mitochondrion is embedded, as reported in a previous TEM observation (Carothers, 1975). In stage-5 cells, which are mature spermatozoids, the plastid exhibited a cylindrical shape with thin tubular extensions associated with the posterior end of the nucleus (arrowheads in Figure 1C ), as described previously (Carothers, 1975). Consistent with this observation, the circularity of the plastid increased between stages 0 and 1 and drastically decreased before stage 5 ( Figure 1D ). Similar results were obtained using another plastid marker, TP-sGFP, sGFP fused with the transit peptide (TP) of MpSIG2, which is a nucleus-encoded subunit of plastid-encoded plastid RNA polymerase (Kanazawa et al., 2013) ( Figure S1 ). These results indicate that dynamic morphological changes occur in the plastid during spermiogenesis in M. polymorpha.

We previously demonstrated that autophagy is required for reorganization of various organelles during spermiogenesis in M. polymorpha (Norizuki et al., 2022). To investigate whether autophagy is also involved in plastid reorganization during spermiogenesis, we observed plastids in spermatozoids of the Mpatg5-1^ge mutant, which is defective in autophagy (Norizuki et al., 2019; 2022). Whereas wild-type spermatozoids possessed one cylindrical plastid, more than 90% of Mpatg5-1^ge spermatozoids possessed one spherical plastid, and three out of the 54 Mpatg5-1^ge spermatozoids possessed two spherical plastids ( Figures 2A, B ). This result suggests that morphological changes in plastids involve autophagy. To gain further insight into the role of autophagy in plastid reorganization during spermiogenesis, we observed morphological changes in plastids during spermiogenesis in Mpatg5-1^ge spermatids. Compared with wild-type spermatids, striking differences were detected at stages 1 and 2; the plastids in Mpatg5-1^ge were complex in shape than the spherical wild-type plastids ( Figures 3A–E ). These results suggest that autophagy is involved in drastic morphological changes in the plastids during early spermiogenesis.

It has been reported that thylakoids are lost and starch deposition occurs as spermiogenesis proceeds, resulting in plastids mostly filled with starch granules in the spermatozoids of many plant species (Renzaglia and Garbary, 2001). Consistent with previous observations, the wild-type plastid at around stage 1-2 contained many starch granules that were observed as electron-dense structures using TEM. However, the plastids in Mpatg5-1^ge spermatids contained significantly fewer starch granules than wild-type plastids, suggesting that autophagy may be involved in starch synthesis regulation during spermiogenesis ( Figures 3F, G ). These results suggest that autophagy also plays important roles in plastid reorganization, which is required for shaping plastid and starch synthesis during spermiogenesis, although its role in the regulation of the number of plastids is minor.

We previously demonstrated that multiple organelles including mitochondria are degraded in the vacuole through autophagy during spermiogenesis in M. polymorpha (Norizuki et al., 2022). The observation that the autophagy-defective mutant exhibited impaired morphological transformation of the plastid and starch accumulation prompted us to investigate whether autophagy-dependent degradation of the plastid occurs during spermiogenesis. For this purpose, we investigated a plastid marker in spermatids, in which the vacuolar membrane was also visualized using mCitrine-MpVAMP71 (Kanazawa et al., 2016; Minamino et al., 2017). In wild-type spermatids, fluorescence from MpPGM1-mTurquoise2 was detected in the vacuole lumen, indicating that some of the plastid stroma was transported into the vacuole for degradation ( Figure 4A ). However, in Mpatg5-1^ge/cf spermatids (Norizuki et al., 2022), the fluorescent signal from MpPGM1-mTurquoise2 was not detected in the vacuole ( Figure 4A ). We also noted that the number of the large spherical vacuole detected in confocal images was significantly decreased in Mpatg5-1^ge/cf spermatids ( Figure 4B ). The mistargeting of MpPGM1-mTurquoise2 should not be due to the reduced number of the functional vacuole, because vacuolar targeting and degradation of the plasma membrane protein MpSYP12A and Golgi proteins occur in Mpatg5-1^ge/cf spermatids similar to wild-type spermatids (Norizuki et al., 2022). These results indicate that a portion of the plastid is degraded by autophagy during spermiogenesis in M. polymorpha.

Unlike nuclear DNA, plastid and mitochondrial DNA are uniparentally inherited, with minor exceptions (Kuroiwa, 2010). In bryophytes, elimination of plastid DNA has been reported to occur during spermiogenesis in the liverwort Dumortiera hirsuta and the hornwort Anthoceros punctatus (Izumi and Ono, 1999; Shimamura et al., 1999). Based on our finding that a portion of the plastid is degraded via autophagy during spermiogenesis in M. polymorpha ( Figure 4 ), we hypothesized that plastid DNA could be degraded through autophagy. To test this hypothesis, we detected organelle DNA using an dsDNA antibody. In addition to the strong signal in the nucleus representing nuclear DNA, we detected fluorescent signals in the cytoplasm of wild-type spermatids at stages 0–2. This signal overlapped with the plastid visualized using MpPGM1-Citrine, suggesting that this signal represents the plastid DNA ( Figure 5A ). Under our experimental conditions we did not detect mitochondrial DNA during spermiogenesis. In cells at stage 4, signal intensity from nuclear DNA was reduced compared with earlier stages, probably due to the restricted accessibility of the antibody to nuclear DNA resulting from chromatin compaction through the function of protamine ( Figure 5B ) (Kreitner, 1977; D’Ippolito et al., 2019). At this stage, the signal from plastid DNA was not detected, suggesting that elimination of plastid DNA occurs before stage 4 during spermiogenesis in M. polymorpha ( Figure 5B ). Contrary to our expectations, the removal of plastid DNA was also observed in Mpatg5-1^ge spermatids ( Figure 5B ). This result indicates that autophagy is dispensable for plastid DNA removal during spermiogenesis in M. polymorpha.