Cytoskeletal dynamics in interphase, mitosis and cytokinesis analysed through Agrobacterium-mediated transient transformation of tobacco BY-2 cells

Authors


Author for correspondence:
H. Buschmann
Tel: +44 1603 450135
Email: henrik.buschmann@bbsrc.ac.uk

Summary

  • Transient transformation with Agrobacterium is a widespread tool allowing rapid expression analyses in plants. However, the available methods generate expression in interphase and do not allow the routine analysis of dividing cells. Here, we present a transient transformation method (termed ‘TAMBY2’) to enable cell biological studies in interphase and cell division.
  • Agrobacterium-mediated transient gene expression in tobacco BY-2 was analysed by Western blotting and quantitative fluorescence microscopy. Time-lapse microscopy of cytoskeletal markers was employed to monitor cell division. Double-labelling in interphase and mitosis enabled localization studies.
  • We found that the transient transformation efficiency was highest when BY-2/Agrobacterium co-cultivation was performed on solid medium. Transformants produced in this way divided at high frequency. We demonstrated the utility of the method by defining the behaviour of a previously uncharacterized microtubule motor, KinG, throughout the cell cycle.
  • Our analyses demonstrated that TAMBY2 provides a flexible tool for the transient transformation of BY-2 with Agrobacterium. Fluorescence double-labelling showed that KinG localizes to microtubules and to F-actin. In interphase, KinG accumulates on microtubule lagging ends, suggesting a minus-end-directed function in vivo. Time-lapse studies of cell division showed that GFP-KinG strongly labels preprophase band and phragmoplast, but not the metaphase spindle.

Introduction

Transient Agrobacterium-mediated transformation methods are being used widely in the plant sciences. Their main advantage is that results are obtained quickly as transgene expression can be observed after 1–3 d of incubation (Mathur et al., 1998; Sparkes et al., 2006; Li et al., 2009). The application of transient transformation methods is manifold. For instance, they enable high-throughput analyses of novel genes described by genome projects (Koroleva et al., 2005), as well as multiple parallel testing of mutant sequences (Buschmann et al., 2006). However, the available transient Agrobacterium-mediated transformation methods do not allow routine analysis of plant cells engaging in mitosis and cytokinesis. Therefore, experiments aimed at the rapid identification of protein function in plant cell division frequently experience technological bottlenecks. It is therefore highly desirable to establish an Agrobacterium-mediated transient transformation system enabling the cell biological analysis of cells from interphase through to mitosis and cytokinesis.

The tobacco BY-2 cell line – dubbed the ‘HeLa cell line of plants’ (Nagata et al., 1992) – is a model organism for the analysis of plant cell division. The cells are larger than those of Arabidopsis and are ideal for the microscopic observation of mitosis and cytokinesis. The BY-2 suspension line has further advantages: cultures are robust and easy to maintain, the morphology of the BY-2 cell is fairly constant, proteomic studies are possible through cDNA sequencing projects and protocols are available for cell cycle synchronization (Nagata et al., 2006). Stable genetic transformation of tobacco BY-2 through Agrobacterium is widely used and clonal transformants ready for analysis can be obtained in 4–8 wk (An, 1985; Koroleva et al., 2006). Transient transformation of BY-2 is, however, normally not achieved through Agrobacterium, but by applying technically rather demanding methods, including particle bombardment (Iida et al., 1991; Ito et al., 2001; Lee et al., 2003; Vetter et al., 2004; Maisch et al., 2009) or electroporation (Koscianska & Wypijewski, 2001; Miao & Jiang, 2007). We therefore investigated methods for the transient transformation of BY-2 with Agrobacterium.

Plant transformation via Agrobacterium requires the transfer of T-DNA from the bacterial cell to the plant host. The mechanism of DNA transfer, which requires cell-to-cell contact to be made by the T-pilus, is structurally related to the conjugational bridge formed between mating bacteria (Lessl & Lanka, 1994; Backert et al., 2008). In the laboratory, bacterial conjugation is usually achieved by mixing cells derived from liquid cultures and plating them on agar-based nonselective medium (Bohne et al., 1998; Piper & Farrand, 1999). After overnight incubation, the transconjugants are re-streaked on selective medium. Although it is possible to obtain conjugation in liquid medium with shaking, it has been shown for inter-Agrobacterium conjugation that efficiency is greatly enhanced when performed on solid medium (Piper & Farrand, 1999). We therefore surmised that the transient transformation of tobacco BY-2 cells could be enhanced when performed on solid medium (using agar or filters), and explored various parameters.

In this article, we describe the transient Agrobacterium-mediated transformation of BY-2 (TAMBY2) method and demonstrate its utility for cell biological studies by expressing various cytoskeletal marker genes in interphase and cell division. We apply the new method to characterize a novel C-terminal kinesin from Arabidopsis.

Materials and Methods

The transient Agrobacterium-mediated transformation of BY-2 (TAMBY2) method

A fresh tobacco BY-2 culture was initiated (5 ml per 100 ml) and grown for 3 d. After this time, the BY-2 culture typically yielded an optical density at 600 nm (OD600) of 1.10; however, slight variations in density did not seem to affect the results. A 50-ml aliquot of this culture was washed twice with sterile Paul’s medium. Paul’s medium contains Murashige and Skoog (MS) salts without vitamins (4.3 g l−1) (M0221, Duchefa, 2003 RV Haarlem, Netherlands) plus 1% (w/v) sucrose at pH 5.8. Decanting after the last washing step produced an approximately fivefold BY-2 cell density (final volume, 10 ml). Agrobacterium was grown overnight (10 ml) and then subcultured by adding 1 ml of cells to 5 ml of fresh LB medium (Luria-Bertani medium without glucose plus antibiotics). After several hours of growth, Agrobacterium was harvested (at OD600 of 0.8) by centrifugation at 9300 g (bench-top centrifuge) for 1 min. For each construct, 1 ml of Agrobacterium culture was pelleted in a 1.5-ml Eppendorf tube. After centrifugation, the supernatant was removed and the pellet was resuspended in a drop (30 μl) of Paul’s medium. Next, 1 ml of the fivefold-concentrated tobacco BY-2 culture was added. The tube was agitated for 5 min at 100 rpm to mix the bacteria with the plant cells. Aliquots of this suspension were then dropped on plates with Paul’s medium (no antibiotics). Solid Paul’s medium contained 0.5% (w/v) agar (Phytagel; Sigma P8169). The plates were sealed with Micropore tape (Phytagel, P8169; Sigma-Aldrich, Dorset, United Kingdom) and incubated at 27°C in the dark. All downstream analyses were performed after 2 or 3 d of incubation. For microscopy, cells were scraped off the agar surface and mixed on a microscopy slide with a drop of fresh culture medium, and analysed with a confocal microscope.

We used the Agrobacterium strains LBA4404 (pBBR1MCS virGN54D) and EHA105. Classical GV3101 was also employed, but yielded lower transformation rates. The transformation rate of EHA105 can be increased using acetosyringone. For this, 250 μM acetosyringone (final concentration) was added to the Agrobacterium culture (for several hours of growth until OD600 of 0.8 was reached) and also to the mixture of Agrobacterium with BY-2 (the mixture dropped onto the plates).

Comparison of liquid- and agar-based transformation

The co-culture experiments in liquid shown in Fig. 1(a,b) were performed exactly as the agar-based transformation, but without first concentrating the tobacco BY-2 cells. However, in separate experiments, and for direct comparison, liquid co-cultures were also assayed using fivefold-concentrated BY-2 cells. Results were entirely negative; however, it is noteworthy to mention that the BY-2 cells appeared to suffer from this treatment and assumed a grey colour. The results were therefore not included in Fig. 1.

Figure 1.

 Transient transformation is obtained only after co-cultivation of Agrobacterium with tobacco BY-2 on solid medium. (a–c) Microscopy was performed after 2 d of co-culture using strain LBA4404 carrying a 35S::GFP-tubulin construct. (a) Co-culture in liquid BY-2 medium. (b) Co-culture in liquid Paul’s medium. (c) Only when co-cultured on agar did cells express green fluorescent protein (GFP). Bars: (a–c) 100 μm.

Molecular biology

A full-length cDNA of the KinG kinesin was cloned by reverse transcription-polymerase chain reaction (RT-PCR) using the Pfx enzyme (Invitrogen), oligos KinG_attB1 (5′-GGGGACAAGTTTGTACAAAAAAGCAGGCTCTAAAATGAGCAGTCATCTGTCACAGGAC-3′) and KinG_attB2_stop (5′-GGGGACCACTTTGTACAAGAAAGCTGGGTCTCACCGCCTTCTATTGAGTGGTTTCG-3′) and Arabidopsis thaliana Landsberg erecta mRNA as a template. The PCR products were cloned into pDONR207 (Invitrogen) and sequenced. The sequence describing the experimental KinG cDNA was deposited in GenBank (accession number HM590591).The KinG cDNA was then transferred into binary pGFP-N-Bin (Chan et al., 2003) by GATEWAY technology. This binary vector confers CaMV35S-driven expression of N-terminal green fluorescent protein (GFP) fusions. The GFP-tubulin-α6 marker was produced by an identical strategy using attB1/2 adaptors attached to tubulin-specific oligonucleotides. The pGWB2 fusion with DsRED-talin was produced using a DsRED-talin construct kindly provided by O. R. Patharkar (Massachusetts Hospital, Boston, MA, USA). The marker was re-amplified with GATEWAY adaptors and cloned into pDONR207 (Invitrogen). The DsRED-talin fusion was then recombined into binary pGWB2 (kindly provided by T. Nakagawa, Shimane University, Shimane, Japan).

Western analyses were conducted by standard methods using an antibody directed against GFP from Abcam, Cambridge, UK.

Live cell microscopy

For long-term observations, tobacco BY-2 suspension cells were placed in a custom-made biochamber. The biochamber is designed to allow for efficient gas exchange whilst minimizing water loss (Koroleva et al., 2009). About 12–15 μl of medium with cells were placed in a small cavity (provided by a spacer) between a coverslip and the Lumox 25 membrane (http://www.greinerbioone.com). Confocal microscopy was performed at 23°C using the SP2 point scanner from Leica (Leica Microsystems, Milton Keynes, Buckinghamshire, United Kingdom) or a spinning disc microscope assembled by Visitech (VisiTech International, Sunderland, Tyne and Wear, United Kingdom). Image analyses were performed using Leica, Metamorph (Molecular Devices, Wokingham, United Kingdom) and ImageJ software.

Results

First, we attempted to achieve transient Agrobacterium-mediated transformation by co-cultivation with tobacco BY-2 in liquid cultures. For this, we used a GFP-tubulin construct driven by the CaMV35S promoter and transformation was assayed by fluorescence. Irrespective of the Agrobacterium or BY-2 concentration, attempts were unsuccessful, even when we explored different media (in this case, MS-based ‘Paul’s medium’: 1% (w/v) sucrose, without the auxin 2,4-Dichlorophenoxyacetic acid (2,4-D)). (Fig. 1a,b and data not shown). Because Agrobacterium to Agrobacterium transfer of DNA is enhanced on solid medium (Piper & Farrand, 1999), we asked whether co-cultivation of Agrobacterium together with tobacco BY-2 cells on agar-based medium would result in the transient transformation of plant cells. Agrobacterium harbouring the GFP-tubulin construct was mixed with BY-2 cells (for respective cell densities, see the Materials and Methods section) and droplets of this suspension were applied to agar plates without antibiotics (Fig. 2a). After 2 d, microscopy revealed that Agrobacterium had proliferated and produced a halo surrounding a cluster of BY-2 cells to which it was attached (Fig. 2b–d). The large number of BY-2 cells expressing a GFP signal indicated a high level of transformation (Fig. 1c). The production of transient Agrobacterium-mediated transformation of BY-2 by co-cultivation on solid medium is henceforth referred to as TAMBY2.

Figure 2.

 Co-culture of tobacco BY-2 with Agrobacterium on agar plates. (a–d) BY-2 and Agrobacterium interaction in the transient Agrobacterium-mediated transformation of BY-2 (TAMBY2) method. (a) Droplets of tobacco BY-2 cells with Agrobacterium placed on agar surface. (b) After 2 d, Agrobacterium proliferated and produced a corona. (c) Tobacco BY-2 cells visible at the edge of the corona. (d) Cells were scraped off the agar and visualized in bright field using a ×60 objective. Fluorescent cells exhibited Agrobacterium cells attached to their cell walls (arrowhead). Bars: (a) 1 cm; (d) 20 μm.

Next, we explored whether the type of Agrobacterium strain employed had an impact on the TAMBY2 method. When we compared the transformation efficiency of the same CaMV35S::GFP-tubulin construct in LBA4404 and EHA105, we found that the overall efficiency was similar, being 11.6% and 16.2%, respectively (Fig. 3a). The strain GV3101 was tested in separate trials. This strain also achieved transformation using the new method (Fig. 4a); however, the efficiency was lower than in the other strains. Acetosyringone had little effect on strains LBA4404 and GV3101; however, the transformation efficiency of strain EHA105 was increased to 24.8% (Fig. 3a). We also investigated whether the type of liquid medium had an effect on Agrobacterium-mediated transformation. Agar-based Paul’s medium was found to be superior to agar-based BY-2 medium, nearly doubling the transformation efficiency (Fig. 3b). The transformation efficiency was sufficiently high that transiently expressed GFP-tubulin could be easily detected in Western analyses using a GFP antibody (Fig. 3c).

Figure 3.

 Evaluation of the transient Agrobacterium-mediated transformation of BY-2 (TAMBY2) method using quantitative microscopy and Western blotting. (a) The Agrobacterium strains LBA4404 and EHA105 yielded similar transient transformation efficiencies. However, the EHA105 efficiency could be enhanced by acetosyringone (AcSy). (b) Co-cultivation on solid Paul’s medium was more efficient than on solid BY-2 medium. (c) Cells transiently expressing green fluorescent protein (GFP)-tubulin (+) were subjected to Western analysis. The fusion protein was easily detected by an antibody against GFP (arrowhead). A smaller degradation product was also recognized. Nonexpressing BY-2 served as a negative control (−). In the Western blotting experiment, expression was based on EHA105 without acetosyringone.

Figure 4.

 The transient transformation method was successfully tested using various plasmid backbones. Moreover, the cells obtained in this way divided with high frequency. (a–d) Type of vector and fusion protein expressed. (a) GFP-eIF4A localized to the cytoplasm and, to a lesser extent, to the nucleus. This construct was expressed using Agrobacterium strain GV3101. (b) The actin network labelled by DsRed-talin expression. (c) The actin network labelled by GFP-ABD2. (d) Cortical microtubules labelled by RFP-AIR9 expression. The pPZP200 derivative used here (pH7WGR2) stems from the Gent collection. (e–g) Double-labelling of the microtubule bundles of the preprophase band. (e) GFP-ZeMAP65-1. (f) RFP-AIR9. (g) Merge of green fluorescent protein (GFP) and red fluorescent protein (RFP) channels. (h) The transient transformation method often reveals mitotic figures. In this case, expression of GFP-tubulin highlights preprophase band, metaphase spindle and phragmoplast. Bars, 20 μm.

In further experiments, we asked whether the type of binary vector would greatly affect the transformation rate of TAMBY2. The backbone of Ti-plasmids carries important Vir genes; however, the plasmids used in this experiment vary in origin and sequence composition of the backbone. For this experiment, we employed fluorescence-labelled cytoskeletal and noncytoskeletal marker proteins (Mathur et al., 1999; Buschmann et al., 2006; Mao et al., 2006; Wang et al., 2008), each harboured by a different plasmid backbone. We successfully transiently transformed several pBIN19 derivatives (Chan et al., 2003), pGWB2 from the pGWB series (by courtesy of Tsuyoshi Nakagawa, Shimane University, Shimane, Japan), pCAMBIA1390 (Wang et al., 2008) and pH7WGR2 of the Ghent pPZP200 derivatives (Karimi et al., 2002) (Fig. 4a–d). The transformation efficiency was high in all cases. This demonstrated that a wide range of vectors are suitable for the new method. Generally, and independent of the specific vector, we found that the transiently transformed BY-2 cells showed various levels of expression, ranging from low to very high (see, for example, Fig. 1c).

We next explored double-labelling using red and green fluorescent markers. Double-labelling was achieved by two different approaches. One approach was to mix two separate Agrobacterium cultures each harbouring a plasmid with opposite coloured tag (e.g. GFP-ZeMAP65 and mRFP-AIR9). Equal volumes were mixed (500 μl plus 500 μl) and this mixture was used to transform wildtype BY-2 (in this case, 1 ml). Another approach was to use a single Agrobacterium clone harbouring two T-DNA vectors. We found that the Agrobacterium strain LBA4404 was capable of maintaining the two plasmids of the pH7WGR2 and pGFP-N-Bin backbones at the same time. For this the LBA4404 strain was transformed with both vectors and selected for hygromycin and kanamycin (Vogler et al., 2008). Transient transformation with double-transformed Agrobacterium lines appeared to be especially effective and nearly all fluorescent cells showed both colours. The two approaches of delivering two colours were employed to visualize several protein pairs, and we successfully analysed interphase as well as dividing cells (Fig. 4e–g; also see Fig. 5). During these experiments, we noticed that many transformed cells were engaging in mitosis and cytokinesis, and we estimate that 10% of the fluorescent cells were in late G2 or M phase (Fig. 4h). Double-labelling was also possible in dividing cells (Fig. 4e–g).

Figure 5.

 Transient expression of the calponin homology domain kinesin KinG in interphase and cell division using the transient Agrobacterium-mediated transformation of BY-2 (TAMBY2) method. (a) Cortical microtubule array labelled by green fluorescent protein (GFP)-KinG. Strong signals were often present in dots at crossover points (arrows) and at microtubule ends (arrowheads). Many of the KinG accumulations are mobile (see Supporting Information Movies S1, S2). (b–d) For kymograph analyses, data from the movies were re-plotted along the time axis using ImageJ software. The white arrow indicates time (128 s) and the black arrow indicates distance (7.3 μm). (b) Analysis of a growing and eventually depolymerizing microtubule. Typically, KinG produced a weak accumulation at the growing plus tip. The terminal signal was lost during catastrophe. (c) Highly fluorescent dots at the lagging end move slowly, reminiscent of minus end treadmilling. (d) Occasionally, fluorescent dots were seen that moved along the microtubules and coalesced into the lagging end. These dots may represent cargo. (e) A dividing cell expressing KinG observed from preprophase to cytokinesis. This time series is based on projections of z-sections. (f) Co-expression of GFP-KinG with red fluorescent protein (RFP)-tubulin shows co-localization (yellow signal in merge) of the kinesin with microtubules. However, some GFP-labelled filaments, especially those associated with the nucleus (n), were not labelled by RFP-tubulin. (g) Co-expression of GFP-KinG with DsRED-talin suggests co-localization (yellow signal in merge) of the kinesin with actin filaments. Bars: (a) 5 μm; (e–g) 20 μm.

Next, we aimed to establish whether TAMBY2 would produce healthy cells suitable for dynamic live cell imaging, including the analysis of cell division and cytokinesis. We chose to analyse a novel kinesin 14-type molecular motor from Arabidopsis, which we termed ‘KinG’ (At1g63640), and for which cell biological data are not available. This protein was identified in a forward proteomic screen for microtubule-associated proteins using Arabidopsis suspension cells as starting material (Korolev et al., 2005). The screen identified only a small number of kinesins; however, it was not necessary to add nonhydrolysable nucleotide analogues in order for these kinesins to attach to microtubules with high affinity (a similar behaviour was reported for some further plant kinesins) (Barroso et al., 2000; Xu et al., 2007). In view of its interesting biochemical microtubule-binding capacity, KinG was selected for further study. As we had previously generated stable BY-2 lines, as well as Arabidopsis plants, with a CaMV35S::GFP-KinG construct, we were able to compare the results obtained from transient and stable transformation. We found that transiently expressed KinG labels cortical microtubules of BY-2 interphase cells (Fig. 5a). The marker was found to accumulate as dots where microtubules crossed one another and at microtubule ends (Fig. 5a–d). KinG-labelled cortical microtubules were dynamic (Supporting Information Movies S1, S2). Minus- or plus-end dynamics were distinguished on the basis of their polymerization characteristics, i.e. leading (plus) ends showing mainly rapid growth and lagging (minus) ends showing slow depolymerization. We found that KinG accumulates at microtubule ends: weaker accumulation was seen at the plus tip (Fig. 5b); strong accumulation was seen at minus ends (Fig. 5c,d). In addition, we found that some dots labelled by GFP-KinG moved along microtubule tracks. In the kymograph, shown in Fig. 5(d), it can be seen how such dots move and coalesce into a microtubule lagging end. We next measured the speed of GFP-KinG-labelled plus ends and observed an average polymerization rate of 4.48 ± 1.02 μm min−1 (= 15), which is similar to that seen in stably transformed Arabidopsis lines using GFP-tubulin as a marker (e.g. Buschmann et al., 2009). The labelling pattern described above was also found in Arabidopsis plants stably transformed with the same construct. These data demonstrate that the new method allows the dynamics of the cortical microtubule array to be assessed and measured.

We further examined whether the TAMBY2 method is also suitable for extended live cell imaging required to follow cells through mitosis and cytokinesis. Here, imaging was performed using custom-made microscopy slides equipped with a gas-permeable/water-impermeable foil (see the Materials and Methods section). We selected transiently transformed cells containing a preprophase band labelled by GFP-KinG and initiated z-scans at regular intervals until cytokinesis was completed. In all cases, transiently transformed BY-2 cells formed mitotic microtubule arrays in a normal succession and with a normal timing of 75–100 min. GFP-KinG labelled microtubules of the preprophase band in a dotted fashion. Microtubule labelling disappeared after nuclear envelope breakdown and was not obvious at metaphase spindle stage (Fig. 5e, at 30 min); however, the tagged protein returned to the anaphase spindle and was present at all stages of phragmoplast development (Fig. 5e; Movies S3, S4). The behaviour of GFP-KinG in transiently transformed BY-2 cells was identical to that observed for stably transformed BY-2 lines in separate experiments (not shown).

In interphase cells, GFP-KinG labelled the cortical microtubule array, but was also frequently found in filaments that associated with the nuclear envelope (Fig. 5f,g). We transiently co-expressed GFP-KinG and red fluorescent protein (RFP)-TUA6 in BY-2 cells, and found that a subfraction of GFP-KinG filamentous structures was not labelled by RFP-tubulin, i.e. may not consist of microtubules (Fig. 5f). These filaments often emanated from the nucleus. KinG belongs to a group of 14-type kinesins that contain a conserved calponin homology domain at their N-terminus, implicated in actin binding. We therefore asked whether KinG also co-localizes with actin. GFP-KinG and the actin marker DsRED-talin were transiently co-transformed into BY-2. In cells expressing both proteins, we frequently observed co-localization of the KinG kinesin with the actin marker talin, suggesting that KinG binds to actin filaments in vivo (Fig. 5g). Taken together, these results indicate that the TAMBY2 method allows for the analysis of fluorescence-tagged proteins in interphase and cell division, as well as for co-localization studies using dual-colour labelling.

Discussion

Transient Agrobacterium-mediated transformation is increasingly being used in the plant sciences. Transient transformation methods are fast, allow the analysis of toxic gene products, serve to overcome gene silencing and allow the high-level overexpression of commercially relevant proteins (Marillonnet et al., 2005; Sparkes et al., 2006; Sainsbury & Lomonossoff, 2008; Li et al., 2009). Moreover, transient Agrobacterium-based transformation enables high-throughput approaches, such as the genome-based localization of GFP fusions in Arabidopsis (Koroleva et al., 2005). One great advantage of transformation with Agrobacterium is that specialized equipment is not required. This is to be contrasted with particle bombardment, which requires a particle gun and a large amount of DNA, resulting in a mixture of viable and nonviable cells that may or may not go on to divide. Electroporation of protoplasts, such as particle bombardment, also usually requires specialized vector systems. The use of Agrobacterium to deliver DNA is therefore highly advantageous and simplifies the set of vectors required for analysis. However, transient Agrobacterium-mediated transformation methods are not generally known for the analysis of dividing plant cells, but are usually directed at expression in interphase cells (Sparkes et al., 2006; Li et al., 2009). To our knowledge, only one previous investigation has used a transient Agrobacterium-based method to study cell biological aspects of plant cell division (Chan et al., 2005). However, we find that this transient transformation method of Arabidopsis Col-0 cells yields very few dividing cells (Mathur et al., 1998; Koroleva et al., 2005). This explains our motivation to develop a method for the Agrobacterium-based transient transformation of the tobacco BY-2 cell model. The TAMBY2 method presented in this article is based on a major model organism for studying plant cell division, tobacco BY-2, and routinely yields dividing cells with a high frequency. This allowed us to perform extended time-lapse studies and double-labelling experiments in dividing cells. This new method has the potential to enable high-throughput localization of GFP-tagged proteins in plant cell division.

A few investigations have utilized transient Agrobacterium-mediated expression in BY-2 cells and, in this case, the analyses have focused on β-glucuronidase detection or Northern blotting. Cell biological data, or the quantification of transformation efficiency on a per cell basis, were not presented (Yoshioka et al., 1996; Veena et al., 2003). Other reports describe difficulties when attempting to transiently transform BY-2 by co-cultivation with Agrobacterium. Efforts remained unsuccessful even when using the virulence inducer acetosyringone (Ditt et al., 2005; Koroleva et al., 2006). In view of these results, it is not surprising that no detailed analysis of a method has been published for the transient transformation of BY-2 by Agrobacterium. The results presented in this article demonstrate that Agrobacterium has a greater ability to transform BY-2 cells on solid medium when compared with cells suspended in agitated liquid medium (Fig. 1a–c). A similar result was obtained by a previous investigation, which showed that inter-Agrobacterium conjugational transfer of plasmid DNA was enhanced when performed on solid medium relative to liquid medium (Piper & Farrand, 1999). The authors speculated that this was dependent on the type of sex pilus formed, and that Agrobacterium may form short, rigid pili better suited for transfection on solid medium. It is assumed that Agrobacterium facilitates the same, or a highly similar, machinery to transfer T-DNAs to plant cells (Lessl & Lanka, 1994; Backert et al., 2008), and we therefore speculate that the Agrobacterium secretion system allows for better plant transformation in the semi-dry microenvironment of solid medium. Attachment of Agrobacterium to plant cells appears to be less effective when both cells are agitated in liquid medium. This is indirectly supported by our observation that, in A. thaliana Col-0 cells, where transformation by co-cultivation in liquid culture works reasonably well, transformation is mostly found inside cell clumps (Koroleva et al., 2005) (and H. Buschmann and J.H. Doonan, unpublished data). Such clumps are never formed in the tobacco BY-2 cell line and so co-cultivation on solid medium becomes crucial. Although incubation on solid medium was the most important adjustment leading to TAMBY2, other factors have the potential to increase the transformation efficiency. We found that switching to Paul’s medium (1% sucrose and no hormones) for the co-cultivation stage can improve efficiency. Importantly, all tested Agrobacterium strains work in principle, but we note that LBA4404 (which is now widely used in laboratories) and EHA105 are superior in performance. The efficiency of strain EHA105 can be boosted by the application of acetosyringone. Given the high transformation rates obtained by our new method, it is not surprising that we were able to use the same method to select stable transgenic BY-2 lines. For this, the transiently transformed cells were washed once with fresh medium and placed on selective BY-2 medium containing carbenicillin (500 μg ml−1) to destroy Agrobacterium. Transgenic calli were observed after 4–8 wk of incubation (data not shown).

The results demonstrate that the TAMBY2 method is well suited for the imaging of cytoskeletal behaviour in interphase, mitosis and cytokinesis. The rates of microtubule dynamics and the rate of progress through mitosis appear to be unaltered. As proof-of-principle of the new method, we analysed a novel C-terminal kinesin from Arabidopsis (At1g63640). The kinesin was originally identified in microtubule pulldowns (Korolev et al., 2005), although it also contains a calponin homology domain implicated in actin binding. Our results show that the overexpressed kinesin decorates cortical interphase microtubules. Animal kinesins with C-terminal motor domains usually show minus-end-directed movement. Our expression experiments provide evidence that KinG is a minus-end-directed motor in vivo. GFP-KinG strongly accumulates at lagging ends (minus ends) of interphase cortical microtubules. The simplest explanation for this accumulation is that the kinesin arrives at the minus end through the action of its own processive motor. In cell division, GFP-KinG associates with the preprophase band and with the phragmoplast. GFP-KinG normally does not strongly associate with microtubules of the metaphase spindle (Fig. 5e; Movies S3, S4), and it appears that this lack of microtubule association reflects a regulatory step. However, in a fraction of metaphase cells, we observed association with microtubules, and the spindles were found to be abnormally focused at their minus ends (see Fig. S1). This phenotype, which is likely to be produced by protein overexpression, is consistent with KinG bundling the minus ends of microtubules focused at the poles.

In addition to showing that the TAMBY2 method can be employed for the analysis of the dynamic properties of cytoskeletal proteins in cell division, the method has also been shown to be useful for double-labelling and co-localization studies in living cells. We found that GFP-KinG additionally labels filaments that are not labelled by the microtubule marker RFP-TUA6. Co-transformation with the actin marker DsRED-talin suggested co-localization of GFP-KinG with the actin network. In future studies, in order to confirm a possible co-localization of the kinesin with actin filaments, the talin marker could be replaced by other probes that cause less actin bundling. However, the suggestion of actin localization is consistent with the results obtained for other calponin homology domain kinesins from rice and cotton (Preuss et al., 2004; Frey et al., 2009, 2010; Xu et al., 2009), which bind to microtubules and actin in vivo. Recent cytoskeleton research has suggested functional interactions between microtubule and actin networks in plant cells (e.g. Wightman & Turner, 2008; Deeks et al., 2010; Smertenko et al., 2010), and it seems likely that calponin domain kinesins, such as KinG, play a role here.

In conclusion, this study shows that the TAMBY2 method of co-cultivation of Agrobacterium and tobacco BY-2 on solid medium allows for efficient transient transformation. The cells obtained in this way express fluorescent proteins and are well suited for live cell imaging. Proof-of-concept experiments using cytoskeletal marker proteins demonstrate the utility of the method not only for the analysis of interphase cells, but also for mitosis and cytokinesis. Together, this offers, for the first time, an efficient method for transient Agrobacterium-mediated protein expression in dividing plant cells.

Acknowledgements

We are grateful to Allan Downie for helpful discussions. We thank Max Bush and Grant Calder for technical assistance and Veronika Mitikova for the 35S::GFP-eIF4A construct in Agrobacterium strain GV3101. We are grateful to Tsuyoshi Nakagawa for his set of pGWB vectors, to O. R. Patharkar for the talin marker and to Elison Blancaflor for the GFP-ABD2 marker. This work was supported by a Biotechnology and Biological Sciences Research Council (BBSRC) grant to Clive W. Lloyd and John H. Doonan.

Ancillary