The preprophase band predicts the future cell division site. However, the mechanism of how a transient preprophase band fulfils this function is unknown. We have investigated the possibility that Golgi secretion might be involved in marking the preprophase band site. Observations on living BY-2 cells labeled for microtubules and Golgi stacks indicated an increased Golgi stack frequency at the preprophase band site. However, inhibition of Golgi secretion by brefeldin A during preprophase band formation did not prevent accurate phragmoplast fusion, and subsequent cell plate formation, at the preprophase band site. The results show that Golgi secretion does not mark the preprophase band site and thus does not play an active role in determination of the cell division site.
Microtubules play various essential roles during mitosis. In the sporophytic cells of higher plants, microtubules adopt at least three sequential configurations during cell division: the preprophase band (PPB); mitotic spindle; and phragmoplast (Goddard et al., 1994).
The PPB typically occurs as a band of microtubules encircling the G2/prophase nucleus (Mineyuki, 1999). The most intriguing property of the PPB is its location in the region of the cell cortex where the future cell plate inserts, thereby serving to predict the future cell division site (Mineyuki, 1999). Numerous studies have demonstrated that the phragmoplast (the cytokinetic apparatus which directs the formation of the nascent cell plate) fuses accurately at the PPB-defined cortical site (Gunning and Wick, 1985; Mineyuki, 1999; Pickett-Heaps and Northcote, 1966; Wick, 1991). In addition, it has been reported that aberrant initial trajectory of a phragmoplast with respect to the PPB site can be corrected, provided the expanding phragmoplast does not stray beyond a certain ‘critical zone’ (Galatis et al., 1984a; Galatis et al., 1984b). Thus the PPB appears to confer some property (or properties) to the adjacent cortical site that acts to guide/facilitate phragmoplast fusion at that site.
Much attention has been paid as to how a transient PPB can influence the operation of the future cytokinetic apparatus. One mechanism involves the role of the PPB in orientation of the early spindle axis (Cho and Wick, 1989; Cleary and Hardham, 1989; Wick and Duniec, 1984). The early spindle is always positioned perpendicular to the PPB site, and because the phragmoplast arises from remnants of the spindle, spindle axis orientation directly influences initial phragmoplast trajectory. However, in some cells the spindle rotates after formation, and consequently the initial phragmoplast trajectory is not directed towards the PPB site (Galatis et al., 1984a; Galatis et al., 1984b; Granger and Cyr, 2000b; Palevitz, 1986; Palevitz and Hepler, 1974). Nonetheless, in such cells the phragmoplast finds its way to the PPB site and inserts properly (Galatis et al., 1984a; Galatis et al., 1984b; Granger and Cyr, 2000b; Palevitz, 1986; Palevitz and Hepler, 1974). Therefore a second mechanism is likely to exist that ‘marks’ the PPB site such that the marking information persists throughout mitosis and can guide the trajectory of the expanding phragmoplast during cytokinesis. These two mechanisms are not mutually exclusive, and may represent complimentary and hence redundant activities that co-operate to ensure the accurate placement of the new cell plate.
We have investigated the role of Golgi stacks in marking the PPB site by generating a dual-marker BY-2 (Bright Yellow-2) tobacco cell line in which both microtubules and Golgi stacks can be observed in living cells. We tested the significance of Golgi secretion in determination of the cell division site by inhibiting Golgi secretion using brefeldin A (BFA). BFA is a fungal metabolite that affects the morphology and secretory activity of the plant Golgi apparatus (Driouich et al., 1993; Satiat-Jeunemaitre and Hawes, 1993). Our observations, using the dual-marker cell line, show that the PPB is associated with a small increase in Golgi stack frequency; however, BFA has no effect on the positional information laid down at the PPB site and therefore Golgi secretory activity does not play a role in marking the PPB site.
Generation of a microtubule- and Golgi stack-labeled cell line
BY-2 cells were used to determine whether there is a relationship between the PPB site and the distribution of Golgi stacks. Previously, a transgenic BY-2 cell line (designated BD2-5), labeled for microtubules, was generated (Granger and Cyr, 2000a). BD2-5 cells were transformed with an RFP-labeled Golgi marker protein to generate a tobacco cell line that would allow us to visualize both the PPB and Golgi stacks in living cells.
N-acetylglucosaminyltransferase I (Nag) is a resident enzyme of the Golgi organelle that participates in the glycosylation of secretory proteins, and has been used as a Golgi marker protein (Gleeson, 1998; Munro, 1998). The transmembrane-stem region of this protein is sufficient for retention in Golgi stacks of animal and plant cells (Burke et al., 1992; Burke et al., 1994; Essl et al., 1999; Grabenhorst and Conradt, 1999; Jaskiewicz et al., 1996; Nilsson et al., 1996; Opat et al., 2000; Tang et al., 1992). A chimeric gene was assembled that encodes a fusion protein consisting of the N-terminal transmembrane-stem region of an Arabidopsis Nag homologue with the Discosoma red fluorescent protein (RFP; Figure 1). A second chimeric gene encoding the rat sialyl transferase transmembrane-stem region (RatST) fused to RFP was also constructed. RatST has been shown to localize GFP to Golgi stacks (Boevink et al., 1998). BD2-5 cells were transformed with the chimeric genes to generate dual-marker cell lines. As a control, BD2-5 cells were also transformed with the RFP gene only.
RFP fluorescence in the BD2-5 cells expressing the Nag-RFP fusion protein (referred to as Nag-RFP cells) is visible as bright spots distributed in the perinuclear region, cytoplasmic strands, and cortex of the cytoplasm (Figure 2). RFP fluorescence in the Nag-RFP cells is excluded from the nucleus and vacuoles (Figure 2). This RFP fluorescence pattern is indistinguishable from the RFP fluorescence pattern of BD2-5 cells expressing RatST-RFP (Figure 2).
In sharp contrast to the Nag-RFP and RatST-RFP fluorescence patterns, fluorescence in BD2-5 cells expressing RFP alone appears predominantly in the nucleus with diffuse labeling of the cytoplasm (Figure 2). We rarely (<5% of the cells) observed punctate RFP fluorescence in the cells expressing RFP alone.
Biochemical, biophysical and crystallographic data have indicated that the mature form of RFP is tetrameric (Baird et al., 2000; Heikal et al., 2000; Yarbrough et al., 2001). Therefore it is possible that the punctate pattern of RFP fluorescence in the Nag-RFP cells is due to an intrinsic tendency of RFP to form large-order aggregates rather than RFP localization to Golgi stacks. However, the diffuse expression pattern of RFP alone does not support this alternative hypothesis.
Authenticity of Golgi stack labeling in the Nag-RFP cells
Observations on living Nag-RFP cells showed that the RFP spots display a complexity of motion and morphology that are identical to those previously reported for Golgi stacks (Boevink et al., 1998; Nebenführ et al., 1999). Movements of the RFP spots alternated between rapid translational motion (streaming) and Brownian motion (wiggling) (Fig. S1). Streaming activity was most frequently detected in cytoplasmic strands and cortical regions of the cytoplasm, with maximal velocities reaching 6 µm sec−1. Streaming of the RFP spots was also dependent on the presence of an intact actin cytoskeleton, as streaming of the RFP spots ceased after treatment of the Nag-RFP cells with 10 µm latrunculin B for 30 min (compare Figs S2 and S3).
The RFP spots in the Nag-RFP cells also displayed a distinct Golgi stack-like morphology. Most often the spots were visible as short, elongate images about 1 µm in length, but occasionally appeared comma-shaped or ring-like in morphology (Fig. S1). Time-lapse microscopy on living Nag-RFP cells revealed a transmutation of morphology of the same RFP spots (Supplementary material, Movie 1), indicating that the various forms of the RFP spots result from different viewing angles of the same spots.
We looked at the effect of brefeldin A (BFA) treatment on the RFP fluorescence pattern in Nag-RFP cells to determine the authenticity of Golgi stack labeling. Exposure of plant cells to BFA results in the aggregation and vesiculation of Golgi stacks to form so-called BFA compartments (Driouich et al., 1993; Essl et al., 1999; Satiat-Jeunemaitre and Hawes, 1992a; Satiat-Jeunemaitre and Hawes, 1992b; Satiat-Jeunemaitre and Hawes, 1993; Satiat-Jeunemaitre et al., 1996). Treatment of Nag-RFP cells with 30 µm BFA results in the clustering of RFP spots identical to the BFA compartments. The clustering effect is apparent after 15 min treatment (data not shown) and is clearly manifested by 1 h treatment (Figure 3b). Similar data were also obtained for the RatST-RFP cells (data not shown). Nag-RFP cells treated with ethanol (the solvent used to prepare the BFA stock solution) do not display clustering of RFP spots (Figure 3a). In addition, the clustered RFP spots revert back to the initial state on removal of BFA (Figure 3c). Importantly, control cells expressing RFP alone do not display clustering of RFP on BFA treatment (Figure 3d), indicating that the clustering of fluorescent spots in the Nag-RFP cells is not due to a non-specific effect of BFA on cytoplasmic organization. We conclude from these observations that the RFP spots in the Nag-RFP cells report the authentic location of Golgi stacks.
Relationship between PPB site and Golgi stack distribution
The Nag-RFP cell line allowed us to test unequivocally whether the PPB site is associated with an increased Golgi stack frequency. Observation of living Nag-RFP cells containing PPBs indicated that Golgi stacks do not accumulate exclusively at the PPB site, and only 10% of the cells had a clearly discernible increased Golgi stack frequency at the PPB site (Figure 4a). However, there appeared to be some degree of correlation between the PPB site and Golgi stack frequency in most cells (Figure 4b,c).
Fifty-two Nag-RFP cells with PPBs were examined to quantify the relationship between PPB site and Golgi stack frequency. The average width of a mature PPB in these cells was about 4 µm, and Golgi stack frequency was determined in a cortical area 6 µm wide in order to include the PPB and its immediate vicinity. As a comparison, Golgi stack frequency was also determined in a cortical area 6 µm wide and 6 µm away from the PPB site. As shown in Table 1, there is a small but significant increase in Golgi stack frequency associated with the PPB site.
Table 1. Correlation between PPB site and Golgi stack frequency
Golgi stack frequency is presented as the number of discrete Golgi stacks observed within a cortical area at the PPB site or a similar cortical area 6 µm away from the PPB site (non-PPB site). Results represent the mean value ±SD obtained from the observation of 52 Nag-RFP cells with PPBs. The P value was determined using Student's t-test.
Golgi secretory activity does not play a role in marking PPB position
The significant increase in Golgi stack frequency at the PPB site suggests that secretion through Golgi stacks might play a functional role in marking the PPB site. If this hypothesis is true, inhibition of Golgi secretion during PPB formation should prevent the accurate fusion of the expanding phragmoplast at the PPB site.
To test this hypothesis, BFA was used to inhibit Golgi secretion. BFA rapidly inhibits secretion of proteins and carbohydrates through the Golgi stacks in a variety of plant cells (Boevink et al., 1999; Driouich et al., 1993; Jones and Herman, 1993; Satiat-Jeunemaitre and Hawes, 1993). Recently, 20 µm BFA was also shown to inhibit phragmoplast expansion in BY-2 cells due to curtailment of the supply of Golgi stack-derived vesicles necessary for cell plate formation (Yasuhara and Shibaoka, 2000; Yasuhara et al., 1995). Treatment of Nag-RFP cells with 30 µm BFA results in the reduction of the phragmoplast expansion rate, and the extent of this reduction correlates with the extent of BFA treatment (Table 2). Exposure of cells to 30 µm BFA for 1 h reduces the phragmoplast expansion rate at least threefold, whereas exposure to 30 µm BFA for 2 h almost completely inhibits phragmoplast expansion (Table 2). These results indicate that 30 µm BFA exposure is sufficient to inhibit Golgi secretion in the Nag-RFP cells. In addition, the BFA effect is reversible, and the phragmoplast expansion rate recovers to control levels after 1 h BFA removal (Table 2). We note that other aspects of the cell cycle, such as PPB formation, PPB disappearance, nuclear migration, spindle formation and spindle elongation, are not affected by treatment with 30 µm BFA for 2 h (data not shown).
Table 2. Effect of brefeldin A on phragmoplast expansion rate
Phragmoplast expansion rate (µm min−1) ± SD
Nag-RFP cells were treated with 30 µm BFA for 1 or 2 h followed by time-lapse fluorescence microscopy (in the presence of BFA). The GFP marker was used to follow phragmoplast expansion at 10 sec intervals for 15–20 min. Control cells were treated with an equivalent volume of ethanol for 2 h. BFA recovery experiments were carried out by treating cells with 30 µm BFA for 2 h followed by removal of BFA by washing cells five times with medium without BFA. Cells were then incubated in the absence of BFA for 1 h, followed by microscopic observation. Numbers in parentheses indicate the number of cells used for each treatment.
P values for the null hypothesis were determined using Student's t-test: aP < 10−7; bP < 10−16; cP≈ 0.08.
The maximum time for PPB formation to the onset of mitosis is about 1 h (average time 50 ± 15 min, n = 11 cells) in the Nag-RFP cells. Nag-RFP cells were exposed to 30 µm BFA for 2–3 h, followed by the removal of BFA. During this treatment, cells in G2 progress normally into prophase accompanied by formation of the PPB. After BFA removal, cells with a mature PPB were chosen for observation and the progression of mitosis was followed using time-lapse microscopy. This treatment regime ensured that PPB formation in the chosen cells had occurred in the absence of Golgi secretion, and also allowed for normal phragmoplast expansion during cytokinesis.
The phragmoplasts in the BFA-treated cells always inserted at the PPB-defined sites, even if they started out with an aberrant trajectory (Figure 5b; compare Figs S4 and S5). This result was observed for 32 cells, even when the cells had been treated with 100 µm BFA.
It is possible that PPB formation leads to Golgi stack recruitment at the PPB site during G2/prophase, but that secretion through the Golgi stacks is triggered only on the onset of metaphase. In this scenario, a metaphase Golgi belt would mark the PPB site (Nebenführ et al., 2000). We have detected a Golgi belt at the PPB site during metaphase of control cells (Figure 6a). However, a Golgi belt was never detected at any of the mitotic stages in cells recovering from BFA treatment (Figure 6b). Hence we rule out the possibility that the metaphase Golgi belt may be involved in marking the PPB site.
The role of the PPB in determination of the cell division site has long been a subject of discussion and investigation (Gunning and Wick, 1985; Mineyuki, 1999; Wick, 1991); however, mechanistic details of PPB function remain sparse. One of the goals in understanding PPB function has been to determine the nature of the positional or marking information retained at the PPB site, which is proposed to function in determination of the future cell division site. We have investigated the role of Golgi stacks in providing such marking information, and have determined that Golgi secretory activity is not required for PPB marking in BY-2 cells.
The Nag-RFP dual-marker cells allowed us to specifically choose cells containing PPBs, and to directly determine the relationship between PPB site and Golgi stack distribution. Our results demonstrate that the PPB site is associated with a significant increase in Golgi stack frequency in BY-2 cells. However, it is not known whether this increase in Golgi stack frequency at the PPB site is due to specific recruitment of Golgi stacks at the PPB site, or is a consequence of the formation of a phragmosome. Nebenführ et al. (2000) reported the presence of a cortical Golgi belt in metaphase cells that was not detectable in interphase cells. However, as microtubules were not labeled in these cells, the authors were unable to directly determine the relationship between PPB site and Golgi stack distribution in interphase, and whether the Golgi belt in metaphase coincided with the PPB site. Our observations extend the observations of Nebenführ et al. (2000) insofar as an increased accumulation of Golgi stacks is indeed detectable at the PPB site.
A correlation between PPB site and increased Golgi stack frequency raises the possibility that the PPB is capable of attracting Golgi stacks in its vicinity, and that these stacks might participate in marking the PPB site. Therefore it became important to determine whether Golgi secretory activity is involved in marking the PPB site. We addressed this question by inhibiting Golgi secretion during PPB formation using BFA. BFA inhibits Golgi secretion in a number of plant cells, and inhibition of secretion is proposed to occur long before any detectable changes in Golgi morphology (Boevink et al., 1999; Driouich et al., 1993; Jones and Herman, 1993; Satiat-Jeunemaitre and Hawes, 1993). Inhibition of Golgi secretion during PPB formation did not prevent accurate phragmoplast fusion at the PPB site. Mechanism(s) that correct aberrant initial phragmoplast trajectory were also intact following BFA treatment. In addition, a metaphase Golgi belt was never detected in cells recovering from BFA treatment. Therefore recruitment of Golgi stacks at the PPB site does not have a functional role in terms of determination of the cell division site in BY-2 cells.
The BD2-5 tobacco cell line and culture technique have been described previously (Granger and Cyr, 2000a). The culture medium was supplemented with 100 mg l−1 kanamycin (and 40 mg l−1 hygromycin in the case of the dual-marker cell lines) prior to subculturing cells.
Construction of plant transformation plasmids
Gene segments encoding the transmembrane-stem regions of Arabidopsis N-acetylglucosaminyl transferase I (Nag; GenBank accession number AJ243198) and rat α-2,6-sialyltransferase (RatST) were amplified from an Arabidopsis thaliana var. Columbia floral bud cDNA library (gift from Dr June Nasrallah) and a cell expression plasmid (pSAST2; gift from Dr Sean Munro), respectively. Nag forward primer: 5′ AGTCGACA TGGCGAGGATCTCGTGTG 3′. Nag reverse primer: 5′ TCCGGTTGTTGCGGCAAGT TCTTCGTCCTGG 3′. RatST forward primer: 5′ AGTCGACATGATTCATACCAACTT G 3′. RatST reverse primer: 5′ TCCGGTTGTTGCGGCGGCCACTTTCTCCTG 3′. The red fluorescent protein (RFP) gene (Clontech, Palo Alto, CA) was also amplified using PCR (Forward primer: 5′ GCCGCAACAACCGGAGCCATGAGGTCTTCCAAG 3′; reverse primer: 5′ TGGATCCCTAAAGGAACAGATGGTG 3′) to allow the generation of Nag-RFP and RatST-RFP chimeric genes using two-step recombinant PCR (Higuchi, 1990). Cloned recombinant PCR products were sequenced to check the sequences of the chimeric genes.
The chimeric genes were introduced between the Cauliflower mosaic virus 35S promoter and nopaline synthase terminator sequences engineered in the pCAMBIA 1300 plant transformation vector (CAMBIA, Canberra, Australia). As a control, the RFP gene, by itself, was also introduced into this plant transformation vector.
Generation of dual-marker cell cultures
Nag-RFP, RatST-RFP and RFP plant transformation constructs were introduced into BD2-5 tobacco cells using Agrobacterium-mediated transformation (Granger and Cyr, 2000a). Selection of dual-marker cells was performed using 100 mg l−1 kanamycin and 40 mg l−1 hygromycin. Several calli (representing independent transformation events) were used to initiate cell cultures that were screened for GFP and RFP fluorescence.
Latrunculin B (LatB; Calbiochem, La Jolla, CA) was prepared as a 2 mm stock solution in dimethyl sulfoxide (DMSO). Brefeldin A (BFA; Sigma, St Louis, MO) was prepared as a 30 mm stock solution in absolute ethanol.
Cells were exposed to 10 µm LatB for 30 min, then observed in the presence of LatB. An equivalent volume of DMSO was used for the control treatments.
Cells were exposed to 30 or 100 µm BFA for varying periods (as indicated in the text) with the same results. An equivalent volume of ethanol was used for the control treatments. BFA treatment was stopped, when needed, by replacement of the BFA-containing medium with culture medium without BFA, five times. The BFA washout step took less than 5 min.
All observations were carried out on living cells, immobilized on poly l-lysine-coated coverslips, maintained in a humid chamber. All images were collected using a plan-Neofluar 40X (NA = 1.3) oil immersion objective (Zeiss, Thornwood, NY).
Confocal images were obtained using a Zeiss 410 laser scanning microscope (488 nm excitation, 515–565 nm emission for GFP, and 543 nm excitation, 590LP emission for RFP). Typically, 4 sec scan times and 4-line-averaging was applied for image acquisition. Quantification of Golgi stacks at the PPB and non-PPB sites was performed by manually counting the number of discrete Golgi stacks in a cortical region 6 µm wide. Clusters of irresolvable Golgi stacks were counted as a single stack.
For epifluorescence microscopy, a Zeiss Axiovert S100 TV microscope was used. Cells were typically illuminated with 10% light intensity from a variable-intensity mercury arc lamp (Zeiss AttoArc) and digital images captured with a Hamamatsu Orca CCD camera (Hamamatsu Corp, Bridgewater, NJ) controlled by ISee software (Inovision Corp, Durham, NC). GFP (450–490 nm excitation, 520–560 nm emission) and RFP (500–540 nm excitation, 570–610 nm emission) filter sets were used to discriminate between the two fluorochromes.
We thank Drs June Nasrallah and Sean Munro for kindly providing us with an Arabidopsis floral bud cDNA library and rat sialyl transferase cDNA, respectively, and Deborah Fisher for critical reading of the manuscript. This research was supported by USDA grant # 98-35304-6668.
The following material is available from http://www.blackwell-science.com/products/journals/suppmat/TPJ/TPJ_1202/TPJ_1202sm.htmFigure S1. Golgi stack movement. Time-lapse microscopy of Nag-RFP cells. Images were captured at 1 sec intervals for 5 mins. Figure S2. Latrunculin B-treated cells. Time-lapse microscopy of Nag-RFP cells treated with 10 mM latrunculin B for 30 mins. Cells were observed in the presence of latrunculin B. Images were captured at 1 sec intervals for 4 mins. Figure S3. DMSO-treated cells. Time-lapse microscopy of Nag-RFP cells treated with an equivalent volume of DMSO for 30 mins, as a control for the latrunculin B treatment. Cells were observed in the presence of DMSO. Images were captured at 1 sec intervals for 4 mins. Figure S4. BFA-treated cells. Nag-RFP cells were treated with 30 mM brefeldin A (BFA) for 2 h, followed by removal of BFA by washing the cells five times with medium not containing BFA. Images were captured at 1 min intervals until the end of cytokinesis. Figure S5. Control cell. Nag-RFB cells were treated with an equivalent volume of ethanol for 2 h, as a control for the brefeldin A treatment, followed by removal of ethanol by washing the cells five times with medium not containing ethanol. Images were captured at 1 min intervals until the end of cytokinesis.