The P. patens genome has one SABRE gene that encodes for a 2736 aa protein. To disrupt SABRE, we used CRISPR-Cas9-mediated homology-directed repair (HDR) to insert a 363 bp cassette with stop codons in all three possible frames into exon 2 of the SABRE locus (Figure 1—figure supplement 1). The mutant allele results in a frame shift starting at amino acid 38 followed by a premature stop codon after amino acid 47. For consistency, all ∆sabre lines generated in this study contained the same mutation. To ensure that the mRNA expressed from the genome-edited allele was altered as predicted, we amplified the 5′ end of the SABRE cDNA isolated from wild type and ∆sab plants. We found that the cDNA from ∆sabre was larger due to insertion of the stop codon cassette (Figure 1—figure supplement 1A). Sequencing revealed that the transcript contained the expected in-frame stop codon.

P. patens protonemal tissue expands exclusively by polarized growth with cell division occurring in the apical cell of the filament. Subapical cells re-enter the cell cycle once a protrusion emerges generating a new filament, with the branching cell dividing at the base of the emerging protrusion. Compared to wild type, ∆sabre plants had smaller and more compact protonemata (Figure 1A). To quantify this difference, we regenerated plants from single protoplasts and measured the overall size 7 days after protoplasting (Figure 1B). Compared to protonemata in control plants, we found that ∆sabre protonemata were 60% smaller (Figure 1C). As protonemata age, some of the protrusions switch fates to bud-like structures that expand in three dimensions, ultimately resulting in the development of leafy shoots, known as gametophores, the adult tissues. While protonemal filaments increase in size exclusively by polarized expansion at the tip of the apical cell, gametophores expand by diffuse growth (Figure 1D). We also observed that ∆sabre gametophores are 36% shorter than wild type (Figure 1D, E). The phyllids, leaf-like structures that emanate from the gametophore, were 22% smaller than wild type (Figure 1F). Rhizoids are polarized-growing filaments that grow from the base of the gametophore anchoring it in the soil. ∆sabre rhizoids were 70% shorter than wild type rhizoids, a decrease in size comparable to the polarized-growing protonemata (Figure 1G). Time-lapse imaging demonstrated that while early developmental patterning of ∆sabre gametophores was not altered, cell expansion was significantly delayed (Figure 1—figure supplement 2, Video 1). Together, these data demonstrate that the single SABRE gene regulates both polarized and diffuse growth in P. patens.

To determine whether smaller organ size resulted from changes in underlying cell size, we imaged protonemata and gametophore development at higher resolution (Figure 2). During protonemal development, the apical stem cell divides, leaving behind a subapical cell that does not elongate anymore. Thus, we measured the length of protonemal subapical cells. While ∆sabre cells were 30% shorter than wild type (Figure 2A), the decrease of overall plant size was 60% (Figure 1C). This discrepancy could be due to defects in the rate of growth or how often a particular filament is actively growing. To distinguish between these two possibilities, we measured cell expansion rates in actively growing apical cells as determined by time-lapse imaging. While on average actively growing ∆sabre cells grew only 20% slower than control cells, the reduction in growth was not statistically significant. In contrast, time-lapse imaging revealed that over the same time period ∆sabre protonemal filaments often grew significantly less than a comparable wild type filament due to the fact that for a large portion of the time-lapse acquisition the ∆sabre cell was not growing (Figure 2B, Video 2). As a result, we reasoned that additional mechanisms, such as frequent long pauses in growth, likely contributed to the decrease in plant size.

Interestingly, ∆sabre protonemal cells often changed direction during growth (Figure 2B, Video 2). Both actin and microtubules contribute to maintaining the direction of polarized growth. Cytoplasmic microtubules polymerize towards the cell tip where their plus ends then focus onto an apically localized actin spot (Hiwatashi et al., 2014; Wu et al., 2018). When actin filaments are disrupted, microtubules no longer focus below the tip (Wu and Bezanilla, 2018) and growth is inhibited. When microtubules are disrupted, the actin spot randomly appears and disappears throughout the cell, with expansion occurring in areas that accumulate actin, ultimately resulting in a loss of directional growth (Wu and Bezanilla, 2018; Yamada and Goshima, 2018). Since defects in growth directionality in ∆sabre cells resembled wild type cells lacking microtubules (Doonan et al., 1988; Wu and Bezanilla, 2018), we wondered if SABRE influenced the apical microtubule focus. We found that similar to wild type, the actin and microtubule foci in ∆sabre cells were persistently present at the tip (Figure 2—figure supplement 1A, B). Thus, the observed defects in growth directionality and rate appear to be independent of actin and microtubules. Actin and microtubules also form dynamic networks at the cell cortex in protonemal cells. Using variable angle epifluorescence microscopy (VAEM), we imaged microtubules and actin in wild type and ∆sabre plants and quantitatively compared the global dynamics of these networks (Figure 2—figure supplement 1C, D). We measured the correlation coefficient of the intensity of the GFP-tubulin and Lifeact-GFP (labeling the actin cytoskeleton) signal at all pixel locations for all temporal intervals (Figure 2—figure supplement 1C, D). As the microtubule and actin filaments change their position due to polymerization, depolymerization, and translocation, the correlation coefficient decreases with larger temporal increments (Vidali et al., 2010). A fast decay in the correlation coefficient is indicative of very rapid changes in global filament dynamics. Using this analysis, we did not observe obvious changes in microtubule organization or dynamics at the cortex (Figure 2—figure supplement 1C). However, cortical actin exhibited a slight increase in dynamics (Figure 2—figure supplement 1D).

Diffuse growing tissues in the P. patens ∆sabre mutants were stunted similar to what was observed in the Arabidopsis sabre mutant. Time-lapse imaging of expanding phyllids (Figure 2, Video 3) revealed that due to defective cell expansion (Figure 2C, yellow and orange lines), ∆sabre grew significantly less within the same time window compared to wild type (Figure 2C, magenta and blue lines and dashes). In mature phyllids, the final cell size was also smaller in ∆sabre (Figure 2C, brown lines). We quantified cell size in fully expanded phyllids and discovered that cell area at the base and tip of phyllids was reduced in ∆sabre (Figure 2D–F), consistent with the overall stunted gametophore stature (Figure 1D–F). In particular, cell area at the base of phyllids, which are generally the largest cells in the phyllid, was more affected than cells at the tip of the phyllid in ∆sabre (Figure 2E), suggesting that in the absence of SABRE cells may not be able to expand beyond a certain size. To measure changes in shapes of individual cells, we fit a bounding ellipse to each cell and quantified the ratio between the major and minor axes. Cells were less elongated in ∆sabre both at the tip and the base of the phyllids compared to wild type (Figure 2F).

During cytokinesis, plant cells use the phragmoplast, a microtubule-based structure, to build a new cell plate that physically separates the daughter cells. In protonemata, the phragmoplast forms between the two daughter nuclei, expanding perpendicular to the axis of the filament and eventually fusing with the plasma membrane. Phragmoplast microtubules are a dense bipolar array of microtubules that direct late secretory vesicles to the midzone where they fuse to form a membrane-encapsulated cell plate (Smertenko et al., 2018). During the early stage known as the disc phragmoplast, interdigitated microtubules with their plus ends at the cell equator are arranged in a spindle-like structure to establish the phragmoplast. As the cell plate grows, the microtubules label the outer edge of the expanding cell plate, known as the ring phragmoplast. Eventually the developing cell plate reaches the existing side wall and fuses with it, at which point the microtubule array dissipates (Boruc and Van Damme, 2015; Smertenko, 2018). In ∆sabre protonemata, we discovered that while mitosis is unaltered (e.g., Figure 3A, 0–12 min), phragmoplast microtubules were present for significantly longer than in control cells (Figure 3A, Video 4). We found that starting at the time of disc phragmoplast establishment (Figure 3A, 12 min), all control cell phragmoplast microtubules disappeared within 28–50 min (N = 10). However, for all ∆sabre cells (N = 10), microtubules were still present for at least 50 min after disc phragmoplast establishment. To test whether the delay in microtubule disassembly observed in ∆sabre phragmoplasts resulted from defects in microtubule dynamics, we performed photobleaching on phragmoplast microtubules labeled with GFP-tubulin (Figure 3—figure supplement 1A). Microtubule fluorescence recovered from photobleaching similarly in control and ∆sabre cells (Figure 3—figure supplement 1B), indicating that SABRE does not regulate microtubule dynamics during phragmoplast expansion.

Besides the lengthy delay in phragmoplast microtubule disassembly in protonemata, we observed that a fraction of cells contained dark brown material, which accumulated near the cell division plane (Figure 3B–D). We used numerous dyes for cell wall components to attempt to stain the brown material to get a hint of its composition, but none of these dyes stained. Time-lapse imaging revealed that brown material deposition was slow (Figure 3D, Figure 3, Video 5), but always initiated during a cell division event. Cells with brown material would either reinitiate growth after a long recovery time or die (Figure 3, Video 5). Cell death suggests that defects in cell plate formation resulted in loss of cell integrity. Interestingly, brown material appeared more frequently in cells with a large diameter, characteristic of the first few cells in plants regenerated from protoplasts (Figure 3B, C). Cells with a large diameter have inherently more degrees of freedom for orienting the phragmoplast. Furthermore, phragmoplast expansion and insertion occurs over a longer distance in these cells. We also noticed that ∆sabre protoplasts regenerated inefficiently compared to wild type, likely because many protoplasts died during the first cell division, with brown material deposited at the cell division site (Figure 3B, yellow arrow). Considering that neither cell length nor average growth rate accounted for the 60% reduction in protonemal area (Figures 1 and 2), we reasoned that the developmental delay caused by delays and failures in cytokinesis, coupled with frequent pauses in growth, likely accounts for the rest of the decrease in protonemal plant size.

To determine how SABRE impacts cell growth and division, we generated fluorescent fusions of SABRE to analyze its subcellular distribution. We used CRISPR-Cas9-mediated HDR to insert sequences encoding fluorescent proteins upstream of the stop codon of the native SABRE locus (Figure 4—figure supplement 1A). We demonstrated that tagging SABRE at its C terminus with either three tandem mEGFPs (Vidali et al., 2009) or mNeonGreens did not influence its function as we observed no growth defects in plants carrying only the tagged SABRE allele (Figure 4—figure supplement 2A). We focused on imaging the mNeonGreen fusion protein (hereafter, SAB-3mNG) because it was brighter than the GFP fusion. Confocal microscopy revealed that SAB-3mNG formed small puncta at the cell cortex, whose density was highest near the tip of the apical cell (Figure 4A). Even with mNeonGreen, the SAB-3mNG signal was weak and acquiring Z-stacks was challenging. To increase the signal, we inserted the maize ubiquitin promoter (a strong promoter) before the SABRE-3mNG start codon at the native genomic locus (Figure 4—figure supplement 1B). Plants with overexpressed SAB-3mNG (hereafter, OE-SAB-3mNG) had more SABRE puncta, but a similar localization pattern (Figure 4—figure supplement 2C). Disruption of this edited allele by inserting the stop codon cassette used to generate ∆sabre plants resulted in no fluorescence, indicating that ∆sabre alleles do not produce a protein (Figure 4—figure supplement 2D). Importantly, OE-SAB-3mNG plants were indistinguishable from SAB-3mNG plants, demonstrating that overexpression did not have any adverse consequences for protonemal growth (Figure 4—figure supplement 2B). During cell division, OE-SAB-3mNG formed discrete puncta decorating the entire developing cell plate (Figure 4B). Due to weak signals in both SABRE-3mNG and OE-SAB-3mNG, autofluorescence from the chloroplasts was prominently visible in the mNeonGreen channel for both lines.

During cell division, microtubules form the phragmoplast, actin interacts with microtubules and guides the expanding phragmoplast (Buschmann and Müller, 2019; Müller, 2019; Wu and Bezanilla, 2014), while the ER threads through the developing cell plate to build plasmodesmata – plant-specific channels that connect adjacent plant cells (Sager and Lee, 2018; Tilney et al., 1991). Since ∆sabre plants have severe defects in protonemal cell division, we generated SAB-3mNG and OE-SAB-3mNG in moss lines possessing microtubules (mCherry-tubulin), actin (Lifeact-mRuby), and ER (ER luminal marker SP-mCherry-KDEL) markers, enabling inquiry of SABRE behavior in the context of known phragmoplast structures. Initially, to maximize the SABRE signal, we used VAEM to image SABRE simultaneously with these markers at the cell cortex. Consistent with the finding that microtubules were not affected in ∆sabre plants (Figure 2—figure supplement 1A, C) and that SABRE localizes to the nascent cell plate even in late phragmoplasts that lack microtubules (Figure 4B), we found that cortical SABRE puncta did not associate with microtubules (Figure 4C, Figure 4, Video 6). We observed limited overlap between SABRE and actin at the cell cortex (Figure 4C, Figure 4, Video 6). However, most surprisingly, we discovered that cortical SABRE dynamically associated with ER tubules (Figure 4C, Figure 4—figure supplement 3A, B,, Video 6). In a kymograph along one tubule highlighted by the blue line, there were two events where SAB-3mNG dots translocated along the tubule (Figure 4E, arrowheads). More examples are shown in Figure 4—figure supplement 3A, B.

To quantify the degree of overlap between SABRE, actin, and microtubules, we used NIS-Elements software to calculate the Pearson’s correlation coefficient between SABRE and the ER, actin, or microtubules. Pearson’s correlation coefficients were highest between ER and SABRE (Figure 4—figure supplement 3D). Since the Pearson’s correlation coefficient measures the correlation between signal intensities, these measurements were sensitive to overexpression and all correlation coefficients uniformly increased when SABRE was overexpressed (Figure 4—figure supplement 3E). To quantify co-localization independent of signal intensity, we measured the SABRE fraction that did not overlap with the ER, actin, or microtubules in the first 50 frames of a time-lapse acquisition (Figure 4—figure supplement 3C). We derived the overlapping SABRE fraction by subtracting the non-overlapping fraction from 1. Using this method, we found that on average 57% of SABRE puncta area overlapped with cortical ER tubules (Figure 4E). To determine if this overlap fraction occurs by chance, we measured the overlap using a flipped ER image (Figure 4—figure supplement 3C). The average overlap fraction plummeted to 29% when the ER image was flipped. Notably, 23% was the overlap measured with actin and was independent of whether the actin image was flipped (Figure 4E), suggesting that overlap with actin is coincidental. There was even less overlap with microtubules (Figure 4E), which increased to the same level as the overlap with actin when the microtubule image was flipped. Overexpression of SABRE increases the number of SABRE puncta at the cortex. However, importantly overexpression did not significantly affect the ER overlap (Figure 4E). With a higher density, the overlap with actin and microtubules increased to 25–30% and was the same even with flipped actin and microtubule images, suggesting that this degree of overlap occurs by chance.

Since ∆sabre exhibited serious defects in cell plate maturation and associated with the ER at the cell cortex, we further investigated the timing of SABRE and ER localization during cell division. After mitosis, the disc phragmoplast was labeled with thin strands of ER parallel to the microtubules and very little ER was present in the phragmoplast midzone (Figure 5—figure supplement 1A, first time point). In the ring phragmoplast, microtubules were shorter and expanded with the phragmoplast edge (Figure 5—figure supplement 1), while the ER strands parallel to the microtubules became more defined and the ER accumulated in the midzone perpendicular to the microtubules lining the future cell plate (Figure 5—figure supplement 1A). Cell plate maturation occurs during the late phragmoplast stage (Smertenko et al., 2017), coincident with an increase in the ER signal (third and later time points of Figure 5A, B, Figure 5—figure supplement 1). Confocal time-lapse imaging revealed that OE-SAB-3mNG localized to the midzone of the ring phragmoplast weakly, and the signal strengthened as the phragmoplast fully expanded and inserted into the side wall (Figure 5A, B, Figure 5—figure supplement 1B). The strongest OE-SAB-3mNG signal correlated with the timing of cell plate maturation (Figure 5A, B, Video 7) and remained until the ER signal visibly split in two on either side of the new cell plate (last two time points of Figure 5A).

Given the striking association between SABRE puncta and the ER tubules at the cell cortex, we wondered if loss of SABRE function might impact ER in tip-growing cells and during cell division. With an ER luminal marker, SP-GFP-KDEL, we examined the overall ER structure in the cytoplasm. We discovered that ∆sabre cells contained abnormal ER aggregates in the cytoplasm in protonemata, which might underlie defects in polarized growth and directionality (Figure 6A, Video 8). Initially during cell division ER localization was normal (0 min time point of Figure 6B, Video 9). However, during the transition to an increase in ER parallel to the cell plate, the cell plate ER signal noticeably buckled in ∆sabre, while in control cells the ER was straight (Figure 6B, magenta arrowheads, Video 9). Interestingly buckling in ∆sabre mutants occurred coincident with the timing of maximal SABRE accumulation on the cell plate in control cells (Figure 5). Note that while twisting of the ER was obvious in the center of the cell plate, the edges adjacent to the side wall remained fixed, indicating that phragmoplast guidance mechanisms and the phragmoplast insertion site were not affected (Figure 6B, Figure 6, Video 9). For ∆sabre cells that divided relatively normally, the twisted ER, which was observed in all ∆sabre cells (N = 31), eventually straightened out 20–30 min later (Figure 6B, Figure 6, Video 9).

In addition to ER buckling, we noticed that nuclei in ∆sabre apical cells exhibited aberrant motility during cell division. With the ER outlining the re-formed nuclear envelope after mitosis, we observed that in control cells the apical daughter nucleus (Figure 6—figure supplement 1A) moved apically (phase I, blue arrows, see also Figure 6C), basally (phase II, orange arrows, see also Figure 6C), and then resumed apical migration (phase III, white arrows, see also Figure 6C) as has been described previously (Yamada and Goshima, 2018). In ∆sabre cells, the initial apical movement of the nucleus was delayed in 39% of the 31 imaged cells (Figure 6—figure supplement 1B, green arrowheads, Video 10), resulting in a close association between the nucleus and the cell plate, and coinciding with ER buckling (Figure 6B, Figure 6—figure supplement 1B, magenta arrowheads, Video 10). While the phase I apical nuclear movement was normal in the remaining 61% of imaged ∆sabre cells, the subsequent basal movement was dramatically exaggerated in all cells (Figure 6—figure supplement 1, red arrowheads, Video 10).

To further quantify nuclear migration defects, we disrupted SABRE in a line expressing a nuclear localized GFP (NLS-GFP-GUS) and a plasma membrane marker (SNAP-TM-mCherry) (van Gisbergen et al., 2018), enabling imaging of cell division and nuclear movement before and after cell division (Figure 6C, Video 11). In control cells, basal migration (phase II) was often subtle or sometimes even appeared to be a stationary phase (Figure 6C; Yamada and Goshima, 2018). In contrast, ∆sabre basal (phase II) nuclear movement was extreme (Figure 6C, Video 11). In many cases, the nucleus migrated so far back that it appeared to deform as it smashed up against the new cell plate (Figure 6C, Video 11). To quantify the defect in basal movement, we measured the distance between the nucleus and the cell plate when the nucleus was closest to the cell plate (basal position), normalizing the basal position to the apical cell length at that time point. Despite shorter cells in ∆sabre, the relative basal nuclear position was significantly smaller than in control cells (Figure 6D). Beyond the immediate nuclear migration defects after cytokinesis, we also observed less consistent velocity and directionality during interphase in ∆sabre, represented by the zigzagging trajectory in the kymograph (Figure 6C, yellow arrows, Video 11). We tracked the nucleus and quantified the ratio between the final nuclear displacement and the total distance traveled in interphase. We discovered that ∆sabre nuclear movement was less linear and the ratio was smaller (Figure 6E) compared to wild type. The average instantaneous velocity of the nucleus in ∆sabre was expectedly smaller since the cells were not always actively growing (Figure 6F). Given that SABRE influences the ER and that the nuclear envelope is contiguous with the ER, defects in nuclear migration likely result from altered ER function.

To characterize key components of the cell division machinery in ∆sabre cells that failed to form a normal cell plate and resulted in cell death, we imaged the ER, Lifeact-GFP labeling actin and GFP-tubulin labeling microtubules. We observed the ER signal transition from defined tubules (Figure 7A, 40 min) in the phragmoplast to a diffuse signal (Figure 7A, 50 min), correlating with the onset of division failure. As the brown material accumulated, it was outlined by the ER signal (Figure 7A, Figure 7, Video 12). Both Lifeact-GFP and GFP-tubulin signals were relatively normal up until the phragmoplast had expanded to the parental cell membrane. However, at that point Lifeact-GFP became uneven around the phragmoplast with large gaps appearing near the presumptive cell plate. Lifeact-GFP lingered near the cell plate as the brown material generated a gap in the fluorescence gradually ‘invading’ the cytoplasm, which was accompanied by flashes of Lifeact-GFP fluorescence (Figure 7B, 50 min and after). In young plants regenerating from protoplasts, phragmoplasts labeled with GFP-tubulin expand across a larger distance, often finishing insertion on one side of the cell and then extending to the other (Figure 7C). In ∆sabre cells that accumulated brown material, phragmoplast microtubules on the outer edge became disoriented once they reached the parental plasma membrane (Figure 7C, 160 min). Later, deposition of brown material formed a gap between the remaining phragmoplast microtubules (Figure 7C). Sometimes cell plate defects resulted in immediate death. These cells lysed quickly after division and exhibited similar trends to cells that accumulated brown material; phragmoplast expansion was normal until it reached the side wall, at which point the cell lysed (Figure 7—figure supplement 1). Taken together our data suggest that phragmoplast expansion is normal in ∆sabre. However, during cell plate maturation, defects arise in ∆sabre cells; the ER becomes diffuse and both the actin and microtubule cytoskeletons remain associated with the cell plate.

To follow membrane and cell wall remodeling events occurring during cell division, we imaged cell division in the presence of the lipophilic dye, FM4-64, and the callose-specific dye, aniline blue. FM4-64 labels endocytic membranes (Aniento and Robinson, 2005; Jelínková et al., 2010; Klima and Foissner, 2008; Kutsuna and Hasezawa, 2002; Tse et al., 2004; Ueda et al., 2004; van Gisbergen et al., 2008), which are readily incorporated into the membrane surrounding the nascent cell plate early in cytokinesis as clathrin-mediated endocytosis is integral for remodeling the tubular membrane network during early phragmoplast formation (Dhonukshe et al., 2006; Lam et al., 2008; Zhang et al., 2011). However, once the phragmoplast has expanded across the entire mother cell, FM4-64 labeling decreases, consistent with a transition to different membrane trafficking machinery employed specifically during cell plate maturation and membrane fusion (Drakakaki, 2015; Park et al., 2014; Smertenko et al., 2017). In seed plants, callose is deposited early accumulating during the ring stage of the phragmoplast and reaching a peak just prior to fusion of the membrane encapsulating the nascent cell plate with the parental plasma membrane (Samuels et al., 1995). By imaging actively dividing wild type cells, we observed that aniline blue fluorescence steadily increased following a characteristic decrease in FM4-64 staining (Figure 8A, Figure 8, Video 13). Notably, aniline blue only stained cell plates that had fully expanded, suggesting that during phragmoplast expansion, callose within the membranous tubular network is not accessible to aniline blue in the extracellular environment. However, as FM4-64 levels diminished, which occurred during cell plate maturation and fusion of the cell plate membrane with the parental plasma membrane, aniline blue could access the callose, leading to the observed increase in staining (Figure 8A). ∆sabre cells that divided relatively normally (contained no brown material) exhibited a similar decrease in FM4-64 intensity (Figure 8B, solid lines). However, the FM4-64 signal continued to decrease for an additional 10 min, before aniline blue began to rise. At this time, wild type cells would have been fully stained with aniline blue. Further, the rate of increase in aniline blue was significantly slower in ∆sabre cells (Figure 4B, dashed lines), suggesting that it either took longer for aniline blue to diffuse into the cell plate or that callose was deposited later in ∆sabre cells.

To distinguish between defects in membrane fusion or delays in callose deposition, we analyzed images of cell division in protonemata labeled with a nuclear GFP and the membrane marker SNAP-TM-mCherry (Figure 6C). In these lines, loss of GFP fluorescence from the nucleus signaled nuclear envelope break down and the onset of prometaphase. In contrast to FM4-64, SNAP-TM-mCherry did not accumulate on the nascent cell plate (Figure 6, Figure 8C, Video 11). Instead, SNAP-TM-mCherry appeared at the cell plate 30 min after nuclear envelope break down, the timing of which coincides with the end of phragmoplast expansion (Figure 3A). At the same time, SNAP-TM-mCherry disappeared from the plasma membrane adjacent to the cell plate (Figure 8C, white arrowheads), suggesting that SNAP-TM-mCherry accumulates at the cell plate by diffusing into the cell plate membrane from the parental plasma membrane once membrane fusion has occurred, rather than by delivery to the cell plate via exocytosis. By measuring the intensity of SNAP-TM-mCherry at the plasma membrane adjacent to the cell plate and in the middle of the cell plate, we quantified the kinetics of SNAP-TM-mCherry as it disappeared from the plasma membrane and appeared in the cell plate. Within 20 min after nuclear envelope break down, SNAP-TM-mCherry fluorescence decreased at the plasma membrane accompanied by an increase in the cell plate SNAP-TM-mCherry fluorescence. Interestingly ∆sabre cells exhibited the same SNAP-TM-mCherry kinetics as wild type, suggesting that diffusion from the parental plasma membrane was not impaired and thus membrane fusion is unaffected in ∆sabre.

To determine why cells that accumulated brown material often lysed, we used aniline blue staining to image accessible callose in cells with brown material. We found that compared to normal cell plates, which exhibited donut-shaped callose enrichments (Figure 8E), cells with brown material could be grouped into three categories: 27% stained weakly with aniline blue, 20% had a partial aniline blue ring, and 53% had large chunks of aniline blue staining near the brown material (Figure 8F; N = 15 cells). Time-lapse imaging in a ∆sabre cell that accumulated dark material and abnormal chunks of callose revealed accumulation of callose near the stalled microtubules that subsequently disappeared and reaccumulated on the other side of the cell (Figure 8G, Video 14), suggesting that ∆sabre cells exhibit defects in callose secretion and remodeling of callose during cell plate maturation.

All genomic modifications in this study were performed using CRISPR-Cas9-mediated HDR. In brief, two plasmids were generated: a CRISPR plasmid that contains the protospacer(s) and Cas9 ultimately generating the double-stranded break(s) at the designated genomic site(s) and a homology plasmid that provides the template in addition to the sequence being inserted (knockout cassette, fluorescent protein sequence, or promoter) for DNA repair. The two plasmids were co-transformed into moss protoplasts and transformants were regenerated from single protoplasts. All plasmids were constructed using the methods and modular vectors described in Mallett et al., 2019. Primers used to generate these plasmids along with the corresponding plasmid products and primers used for subsequent genotyping are listed in Supplementary file 1. Plasmids were transformed into a variety of moss lines stably expressing fluorescently labeled markers. These lines and the corresponding new lines generated in this study are listed in Supplementary file 1. Note that for generating SABRE-3mNG tag we initially transformed pMH-SAB-C and pGEM-SAB-3mNG into moss lines expressing Lifeact-mRuby and mCherry-tubulin. To introduce ER labeling, we used CRISPR-mediated HDR to swap Lifeact-mRuby into mCherry-KDEL (Supplementary file 1). All other fluorescent labeled lines are as described before: GFP-tubulin (Wu and Bezanilla, 2018), Lifeact-GFP (van Gisbergen et al., 2012), SP-GFP-KDEL (Chang et al., 2020), NLS-GFP-GUS/SNAP-TM-mCherry (van Gisbergen et al., 2018, p. 10), mCherry-tubulin (Burkart et al., 2015), and Lifeact-mRuby (van Gisbergen et al., 2020).

To generate the three tandem mNeonGreen (3mNG) tag, we amplified mNG codon optimized for budding yeast with DC625 and DC626 (Supplementary file 1) incorporating a BamHI site just upstream of the ATG. This PCR product was introduced into pDONR221-P4rP3r using BP reaction (Invitrogen) generating pENTR-R4R3-mNG. A second mNG codon optimized for P. patens was amplified without the stop codon and incorporating BamHI upstream and BglII downstream. This product was ligated into pENTR-mNG in-frame upstream using the BamHI site to create pENTR-R4R3-2XmNG. The third mNG codon optimized for budding yeast was similarly amplified and ligated into pENTR-R4R3-2XmNG. The resulting pENTR R4R3 Cterm BamHI 3xSc_Pp_Sc_mNeon plasmid (Supplementary file 1) was used to create the homology plasmid pGEM-SAB-3mNG according to Mallett et al., 2019. Similarly, to knock-in the stronger constitutive maize ubiquitin promoter, pENTR-R4R3-Ubiquitin-pro was generated by inserting the ubiquitin promoter into pDONR221P4rP3r with BP reaction (Supplementary file 1), and the resulting entry clone was used to make the homology plasmid for moss transformation.

Moss tissue was cultured on PpNH₄ medium (1.03 mM MgSO₄, 1.86 mM KH₂PO₄, 3.3 mM Ca(NO₃)₂, 2.72 mM (NH₄)₂-tartrate, 45 μM FeSO₄, 9.93 μM H₃BO₃, 220 nM CuSO₄, 1.966 μM MnCl₂, 231 nM CoCl₂, 191 nM ZnSO₄, 169 nM KI, and 103 nM Na₂MoO₄) supplied with 0.7% agar, plated on Petri dishes. After propagation by blending, 5–7-day-old tissue was protoplasted and then transformed as previously described (Liu and Vidali, 2011). For all the HDR transformations, 7.5 µg of the CRISPR/protospacer plasmid and 7.5 µg of the homology plasmid were co-transformed into 150 µL protoplasts at a concentration of 2,000,000 protoplasts/mL. Transformed protoplasts were resuspended in liquid plating medium (PpNH₄ plus 8.5% mannitol and 10 mM CaCl₂), plated and regenerated on PRM-B media (PpNH₄ plus 6% mannitol and 10 mM CaCl₂) with 0.8% agar. A layer of cellophane was placed on top of the PRM-B plates, and protoplasts were plated on top of the cellophane. After 4 days on PRM-B, the cellophane was transferred to PpNH₄ supplied with antibiotic for selection. Then, 15 µg/mL hygromycin was used for selection of transformed protoplasts. Plants were grown on selection for a week before moving to PpNH₄ media for subsequent culturing and genotyping.

Growth assays were used to quantify protonemal area. Tissue regenerated from protoplasts was used to synchronize plant growth. Protoplasts were isolated, plated, and regenerated as described above. After 4 days on PRM-B, they were transferred to PpNH₄ and allowed to grow for another 3 days. Seven days after protoplasting, plants were imaged with a Nikon SMZ25 stereomicroscope equipped with a color camera (Nikon digital sight DS-Fi2). Plants were transferred from the plate to a slide and stained with 0.1 mg/mL calcofluor. Calcofluor fluorescence was imaged with a violet filter cube (excitation 420/25, dichroic 455, emission 460 longpass). Subapical cell length was measured manually using these images. Quantification of plant area was carried out using the methods modified from Vidali et al., 2007. In brief, colored images were converted to a single red color image. Single plants were selected and highlighted by cropping and thresholding above a certain intensity value. Plant area was calculated based on the thresholded images. For each experiment, plant area was normalized to the average area of control plants.

Total mRNA was extracted from fresh tissue of both wild type and ∆sabre. Total cDNA was generated with oligo-dT primers using extracted mRNA as the template. Primers DC65 and DC191 (Supplementary file 1) were used to amplify the CDS encoding the N terminal portion of the SABRE protein (Figure 1—figure supplement 1A). Amplified fragments were cloned into pGEM-T Easy (Promega) after A tailing. The cloned fragments were then confirmed with Sanger sequencing.

Brightfield time-lapse microscopy was performed using a Nikon Ti microscope equipped with a 0.8 NA ×20 objective. Plants were cultured in continuous light in PDMS microfluidic devices with liquid Hoagland’s medium as previously described (Bascom et al., 2016). For imaging protonemal tissue, ground tissue was loaded into microfluidic devices and allowed to grow at least 2–3 days before imaging. Gametophores emerged from protonemata naturally 2–3 weeks after loading ground tissue. Multiple XY positions were acquired, and a either a single focal plane or a Z-stack was acquired for each position. Between each time point, white light remained on to provide light for plant growth. Mono-color brightfield images were acquired with Nikon DS-Qi2 camera. Colored brightfield images were acquired with Nikon DS-Vi1 camera. Extended-depth-of-focus (EDF) images were created for the Z-stacks with NIS-Elements (Nikon). Protonemata growth rate was measured by manually tracing the growing tip during active growing period in NIS-Elements software.

For short-term imaging, moss protonemal tissue was mounted on an agar pad on a slide, submerged in Hoagland’s medium, and sealed with a coverslip. For protonemata staining, FM4-64 (15 µM), aniline blue (20 µg/mL), and Fast Scarlet (50 µg/mL) were dissolved in Hoagland’s medium used for tissue mounting. For imaging tissue regenerated from protoplasts, regenerating protoplasts were removed from the cellophane, loaded into microfluidic devices, and immediately imaged. To synchronize cell divisions, ground tissue was loaded into microfluidic devices and allowed at least 4 days to grow. The microfluidic device was placed in far red light for 3–4 days before exposure to white light and imaging. Confocal imaging was performed using a Nikon A1R laser scanning confocal with a 1.3 NA ×40 or 1.49 NA ×60 oil immersion objective (Nikon). Laser illumination at 405 nm was used for exciting aniline blue dye, 488 nm was used for exciting mNeonGreen, GFP, and chlorophyll autofluorescence; 561 nm for mRuby2, mCherry, FM4-64, Fast Scarlet, and propidium iodide. Emission filters were 525/50 nm for mNeonGreen/GFP and aniline blue; 595/50 nm for mRuby2, mCherry, FM4-64, Fast Scarlet, and propidium iodide. For chlorophyll autofluorescence emission, light passed through a long-pass filter allowing wavelengths larger than 640 nm to pass. Image acquisition was controlled by Nikon NIS-Elements software (Nikon). In between each time point, transmitted white light was on providing light for plant growth. 3D reconstruction was done using 3D volume viewer with maximum projection rendering method in NIS-Elements, and contrast for slices at different Z positions was adjusted individually to compensate for loss of signal in tissue further away from the objective using 3D lookup table function. Deconvolution was carried out with NIS-Elements (Nikon) with the 2D deconvolution default settings.

To quantify cell size in gametophores, mature phyllids were removed from gametophore and mounted in a droplet of staining solution (15 µg/mL propidium iodide dissolved in liquid Hoagland’s medium) between a slide and coverslip. Confocal images were captured for quantification. In Fiji (Schindelin et al., 2012), fluorescent images were processed using enhance contrast, subtract background, and smooth and median filter. The processed images were then converted to a binary mask and put through binary process, Close>Dilate>Close>Skeletonize>Dilate, to outline the edges of the cells. Images were then inverted to highlight the cell area and subsequently quantified using the analyze particle function. After quantification, incorrect cells (fused or broken) were manually removed.

VAEM microscopy was performed using a Nikon Ti-E inverted microscope equipped with a TI-TIRF-PAU illuminator, using a Nikon 1.49 NA ×100 oil immersion TIRF objective. Also, 6–8-day-old plants regenerated from protoplasts were mounted between a coverslip and an agar pad on top of a slide, prepared right before imaging. GFP and mNeonGreen were illuminated with a 488 nm laser, while mRuby2 and mCherry were excited with a 561 nm laser, the emission passed through a 525/50 filter for GFP/mNeonGreen and 610/75 for mRuby2/mCherry. Images were simultaneously captured with two Andor 897 EMCCD cameras. Image acquisition was controlled by Nikon NIS-Elements software. All data was processed with enhanced contrast (0.1% pixel saturation), subtract background and smoothing in Fiji using default settings.

To analyze SABRE co-localization with either ER, actin, or microtubules, Pearson’s correlation coefficients were calculated using established algorithms in the NIS-Elements software package. To measure co-localization using an intensity-independent method, we measured the fraction SABRE area overlapping with the either ER, actin, or microtubules. Time-lapse images were processed as described above and as described in Figure 4—figure supplement 3C. Specifically, using Fiji, the SABRE channel was thresholded with the MaxEntropy method uniformly to select area with intensity between 36,000 and 65,535, then converted to a binary mask. The total SABRE area was acquired by measuring the white area of this mask. The ER, actin, or microtubule channel was filtered using unsharp mask with a radius = 3 and mask = 0.6, thresholded with the Otsu method between 28,000 and 65,535, and converted to a binary mask. The binary mask image was skeletonized and dilated to create the ER, actin, or microtubule skeleton. The skeletonized image was then flipped vertically. Then original or flipped skeleton was subtracted from the SABRE binary mask to extract the proportion of SABRE area not overlapping with ER, actin, or microtubules. These processes were done for the first 50 frames of the time-lapse video with 120 ms interval, and the average of each frame was calculated as a data point for each cell. The area of SABRE that did overlap was derived from the above number, which equals to one minus the fraction that did not overlap.

Cortical actin and microtubule dynamics were quantified by measuring the decay of the correlation coefficient over time, as previously described (Vidali et al., 2010). Briefly, two frames within the time-lapse video that were 1, 2, 3 … to N (largest frame number present in the video) frames apart were paired, and the correlation coefficient between every pixel position in those two images was calculated. The numbers for the same frame numbers apart were then averaged between different cells to generate the data point, and standard error of the means was calculated to generate the error bar. The quicker the decrease in the correlation coefficient as the temporal spacing between the frames increased indicates more dynamic movement of the fluorescence signal.

Phragmoplast microtubule photobleaching experiments were conducted using a Nikon A1R laser scanning confocal microscope with 1.49 NA ×60 oil immersion objective. Actively growing plant tissue was mounted on an agar pad between the slide and coverslip, then imaged immediately with a 488 nm laser to identify actively dividing cells at the phragmoplast expansion stage. A 3 µm × 3 µm square region of interest (ROI) was placed in the center of phragmoplast, ensuring the entire ROI was filled with phragmoplast microtubules. Photobleaching was carried out using a 405 nm laser at 10% power for 1 s, after six frames (5 s) of normal imaging. Imaging continued after photobleaching for 2 min to capture fluorescence recovery. The average intensity in the ROI was measured for each frame, then normalized to the average of the value from the first six frames.

A moss line with the nuclear marker NLS-GFP-GUS and the plasma membrane marker SNAP-TM-mCherry (van Gisbergen et al., 2018) was imaged with time-lapse confocal microscopy. A Z-stack was taken every 5 min for the apical cells of several filaments. A segmented line was drawn manually along the axis of growth to generate the kymograph. To make the trends on the kymograph easier to label (Figure 6C), we stretched the kymograph image by increasing the Y axis twofold. To measure the basal nuclear position, using the straight-line tool in Fiji we measured the distance between the middle of the cell plate and the basal edge of the nucleus when it was closest to the cell plate in the kymograph. The distance was divided by the cell length at the same time point, to generate the relative basal nuclear position. To analyze nuclear migration, we isolated movie fragments of nuclear apical migrating periods, tracked the nuclear GFP signal with the TrackMate plugin in Fiji (Tinevez et al., 2017), with a spot diameter of 10 µm. Displacement, distance, and velocity were calculated. Nuclear migration displacement was defined as the straight-line distance between the initial and final positions of the nucleus. Total migration distance was calculated as the sum of displacement between each frame. Average instantaneous velocity was calculated between each frame.