Mitogen-activated protein kinase 4 is involved in the regulation of mitotic and cytokinetic microtubule transitions in Arabidopsis thaliana


  • Martina Beck,

    1. Institut für Zelluläre und Molekulare Botanik, Universität Bonn, Kirschallee 1, D53115, Bonn, Germany
    Search for more papers by this author
    • These authors contributed equally to this work as first authors.

  • George Komis,

    1. Institut für Zelluläre und Molekulare Botanik, Universität Bonn, Kirschallee 1, D53115, Bonn, Germany
    2. Institute of General Botany, Faculty of Biology, University of Athens, GR15784, Greece
    Search for more papers by this author
    • These authors contributed equally to this work as first authors.

  • Anja Ziemann,

    1. Institut für Zelluläre und Molekulare Botanik, Universität Bonn, Kirschallee 1, D53115, Bonn, Germany
    Search for more papers by this author
  • Diedrik Menzel,

    1. Institut für Zelluläre und Molekulare Botanik, Universität Bonn, Kirschallee 1, D53115, Bonn, Germany
    Search for more papers by this author
  • Jozef Šamaj

    1. Centre of the Region Haná for Biotechnological and Agricultural Research, Department of Cell Biology, Faculty of Science, Palacký University, 783 01 Olomouc, Czech Republic
    Search for more papers by this author

Author for correspondence:
Jozef Šamaj
Tel: 00420 585 634978


  • A mitogen-activated protein kinase kinase kinase (MAPKKK) double mutant, Arabidopsis homologue of nucleus and phragmoplast associated kinase (anp) anp2anp3, and the mitogen-activated protein kinase (MAPK) 4 mutant mpk4 of Arabidopsis thaliana show prominent cytokinetic defects. This prompted the analysis of mitotic and cytokinetic progression as a function of MAPK signalling. Mutants were compared with wild types untreated or treated with the specific MAPKK inhibitor PD98059.
  • This study included phenotype analysis, expression analysis of the MPK4 promoter, immunofluorescent localization of MPK4, tubulin and MAP65-1, and time-lapse microscopic visualization of the mitotic microtubule (MT) transitions in control, mutant and inhibitor-treated cells.
  • Mutant and inhibitor-treated cells showed defects in mitosis and cytokinesis, including aberrant spindle and phragmoplast formation and drastically delayed or abortive mitosis and cytokinesis. As a result, bi- and multinucleate cells were formed, ultimately disturbing the vegetative tissue patterning. MPK4 was localized to all stages of the expanding phragmoplast, in a pattern similar to that of its putative substrate MAP65-1.
  • In this study, MPK4 is shown to be involved in the regulation of mitosis/cytokinesis through modulation of the cell division plane and cytokinetic progression.


Cytokinesis, the cytoplasmic partitioning of the daughter nuclei following mitosis, is a temporally and spatially controlled process (Smith, 2001; Müller et al., 2009). In plants, cytokinesis may be symmetric or asymmetric depending on the spatial coordination of the daughter wall deposition. In this way, cytokinesis contributes to tissue patterning, the generation of specialized cell types and overall plant form development (Smith, 2001). Daughter wall formation is driven by the delivery of vesicles containing polysaccharidic material to the midplane. These fuse to form a callosic cell plate which develops into the daughter wall (Otegui et al., 2005). Vesicle delivery to the nascent cell plate is driven by the phragmoplast, a microtubule (MT)-based apparatus, which forms at the end of anaphase. The phragmoplast comprises a bipolar set of anti-parallel MTs emanating from the surface of the daughter nuclei and interdigitating in the midplane by virtue of their plus ends (Otegui et al., 2005). Phragmoplast expansion and function are regulated at three levels: MT dynamics, vesicle delivery, and the direction of expansion (Edamatsu, 2001; Hasezawa & Kumagai, 2002; Sasabe & Machida, 2006; Van Damme & Geelen, 2008).

Regulation of MT dynamics depends on structural, MT-associated proteins (MAPs) and regulatory proteins, such as many kinase species, including mitogen-activated protein kinases (MAPKs). The latter regulate the activity of MAPs associated with MT dynamics and organization, by phosphorylating serine and threonine residues involved in MT binding (e.g. Sasabe & Machida, 2006).

Spatial alignment of phragmoplast expansion is established early during the course of mitosis and is hallmarked by the presence of a cortical MT band called the preprophase band (PPB; Hasezawa & Kumagai, 2002). The PPB occupies cortical sites, where the expanding phragmoplast will meet the parent walls (Van Damme & Geelen, 2008). The PPB is also accompanied by a set of positive and negative markers, some of which reside at the PPB site throughout mitosis. These probably represent tools of the machinery that ‘drags’ the phragmoplast margin to the correct position. They include the presence of a cortical actin PPB (Panteris, 2008), in some cases a cortical endoplasmic reticulum (ER) band (Zachariadis et al., 2001), a cortical actin exclusion zone during mitosis (Panteris, 2008), a cortical exclusion zone for the kinesin CDKA1 associated (KCA1) (Vanstraelen et al., 2006), the TANGLED protein (Walker et al., 2007; Lloyd & Buschmann, 2007), the RanGAP1 (Ras related nuclear protein GTPase activating protein) activator of the small GTPase Ran (Xu et al., 2008), and the POK1 (phragmoplast orienting kinesin 1) and POK2 kinesins (Müller et al., 2006).

MAPKs controlling cytokinesis have been reported in Medicago sativa (Bögre et al., 1999), Nicotiana tabacum (tobacco; Calderini et al., 1998; Nishihama et al., 2002) and Arabidopsis thaliana (Krysan et al., 2002). The involvement of MAPKs in phragmoplast expansion has been thoroughly studied in tobacco. In this case, a protein complex comprising nucleus- and phragmoplast-localized MAP3 kinase 1 (NPK1), the NQK1 (next to NPK1) MAP2K and the NRK1 (next to NQK1) MAPK, is transported by two kinesin-related proteins, namely NACK1 (NPK1-activating kinesin-like protein 1) and NACK2. This signalling pathway, which is termed the PQR pathway of NPK1, NQK1 and NRK1, targets two MT-associated proteins, namely MAP65-1a and MAP65-1b, at the phragmoplast midzone (Nishihama & Machida, 2001; Soyano et al., 2002; Takahashi et al., 2004). Mutants of any member of this PQR pathway show similar cytokinetic defects. Also, a double null mutant of the A. thaliana homologues of nucleus- and phragmoplast-associated protein kinase, ANP2 and ANP3 (Arabidopsis homologues of NPK1; Krysan et al., 2002), and a double null mutant of HINKEL and STUD (homologues of NACK1 and NACK2; Yang et al., 2003; Tanaka et al., 2004) show abnormal cytokinesis that sometimes fails to complete.

In a previous study, we reported MT defects in the interphase cortical system of MAPKKK and MAPK mutants, namely anp2anp3 and mpk4 (mitogen-activated protein kinase 4), leading in both cases to aberrant cell growth and organ development (Beck et al., 2010). In that study we noted cytokinetic phenotypes of the mpk4 mutant (Beck et al., 2010) reminiscent to those of anp2anp3 (Krysan et al., 2002), with features such as the occurrence of cell wall stubs. However, neither of these studies provided information on the progression of mitosis and cytokinesis in these mutants. In the present study we used time-lapse confocal laser scanning microscopy (CLSM) to provide a detailed description of the temporal dynamics of mitotic spindle and phragmoplast progression in the mpk4 and the anp2anp3 mutants, as well as in dividing cells of wild-type plants treated with a MAPKK inhibitor. These data suggest that there is temporal delay and deregulation of mitotic and cytokinetic progression in both mutants.

Materials and Methods

Plant material and treatments

T-DNA insertional mutants corresponding to the ANP2/ANP3 and MPK4 genes (Krysan et al., 2002; SALK_056245, respectively) raised on a Wassilevskaja (Ws) and a Columbia (Col-0) ecotype background were used (Krysan et al., 2002; Nakagami et al., 2006). Mutant and wild-type Arabidopsis thaliana (L.) Heynh seeds, untransformed or transformed with the 35S:GFP:MBD construct for visualization of MTs in vivo (Marc et al., 1998), were germinated on Phytagel-solidified half-strength Murashige and Skoog (MS) medium without vitamins but supplemented with 1% w/v sucrose. Plants used in this study were 4–7 d old. For inhibitor treatments, DMSO stock solutions of respective inhibitors were diluted either in liquid half-strength MS medium or in water.

Reagents, antibodies and inhibitors

All common chemicals used in the study were purchased from Roth (Roth, Karlsruhe, Germany), Sigma or Serva (Serva Carl Benz Str. 7, Heidelberg, Germany). For wholemount immunolabelling and immunoblot analysis, the following primary antibodies were used: rabbit polyclonal anti-MPK4 (Sigma), anti-pERK (phosphorylated Extracellular Signal Related Kinase) (Cell Signaling, Frankfurt am Mein, Germany, or Santa Cruz, Heidelberg, Germany), anti-MAP65-1 (gift of Dr Andrej Smertenko) and mouse or rat monoclonals against α-tubulin (Abcam PLC, 330 Cambridge Science Park, Cambridge, UK, and Serotec, Morphosys AbD GmbH, Düsseldorf, Germany; clones DM1a and YOL1/34, respectively). Secondary antibodies were anti-mouse or anti-rabbit immunoglobulin G (IgG) coupled to horseradish peroxidase (Cell Signaling) in the case of immunoblotting, and anti-rabbit or anti-rat IgGs conjugated with Alexa Fluor 488 or 546 (Invitrogen) in the case of immunofluorescence labelling. To inhibit MAPK activation, the MAPKK inhibitor PD98059 (Calbiochem, Merck Chemicals Ltd., Beeston, Nottingham, UK) was used. This was prepared as a 10 mM DMSO stock solution and diluted in half-strength MS medium or water to obtain the working solutions.

Phenotype analysis and detection of cytokinetic defects

In order to compare cytokinetic deficiencies in the mpk4 mutant and Col-0, control seedlings were fixed and embedded in Spurr resin according to standard procedures. Semithin sections collected using a Reichert Ultracut microtome (Leica Microsysteme Vertrieb GmbH, Wetzlar, Germany) were heat-adhered to glass slides and briefly stained with 1% w/v toluidine blue in 1% w/v aqueous borax. Semithin sections were examined using a Leica DMRBE upright microscope and a Leica DFC290 camera (Leica Microsystems, Wetzlar, Germany) operated through the IrfanView freeware ( Cells were scored for multinuclearity and the occurrence of incomplete cell walls. Alternatively, plasma membranes of cells within root tissue were stained by incubating plant material for 10 min in 4 μM aqueous solutions of the red fluorescing styryl dye FM4-64 or AM4-65 (Sigma).

Tissue-specific MPK4 expression analysis

Genomic DNA of A. thaliana Col-0 was used to amplify the 2-kb putative promoter upstream of At4g01370 encoding MPK4 using the following primer pair: 5′-gc gga tcc gaa gaa gaa caa tgc tcg-3′ and 5′-gc ccc ggg cgg agc aaa att cct cac-3′. After BamHI and SmaI restriction sites were inserted, the promoter sequence was cloned into a ΔGUS (β-glucuronidase) vector (Topping et al., 1994). Following sequencing verification, clones were transformed with Agrobacterium tumefaciens strain GV3101 and subsequently used to generate transgenic plants using the floral dipping method (Clough & Bent, 1998). Transformed plants were screened for kanamycin resistance and subsequently used to detect promoter activity at the tissue level according to published procedures (De Block & Debrouwer, 1992).

Immunofluorescence labelling of roots

Whole roots of mutant and wild-type seedlings were prepared for whole-mount immunodetection of MPK4, pERK, MAP65-1 and tubulin according to previously published procedures (Collings & Wasteneys, 2005; Beck et al., 2010; Müller et al., 2010). Alternatively, the standard root-squash approach was selected as it allowed better penetration of antibodies (Wick et al., 1981), especially for the detection of weakly labelled antigens (e.g. those detected using the pERK antibody). The specificity of the antibodies was determined by standard immunoblotting.

Live-cell imaging and quantitative analysis of mitotic/cytokinetic MT transitions

MTs of both mutant and wild-type plants were visualized using the microtubule-binding domain (MBD) of MAP4, fused to the green fluorescent protein (GFP). In brief, a 35S::GFP:MBD construct (Marc et al., 1998) was used to transform mutant and wild-type inflorescences by the floral dip method. Stably transformed plants were selected for fluorescence studies with the aid of a binocular microscope and then used for live-cell imaging. Time-lapse videos of mitotic/cytokinetic MT transitions were obtained using an Olympus FV1000 confocal laser scanning microscope (CLSM) equipped with an argon laser (at a wavelength 488 nm) and operated with the fv10-asw1.7 software (Olympus, Hamburg, Germany).

For quantitative analysis of mitotic/cytokinetic progression, time-lapse videos were segmented to individual time frames by uploading uncompressed avi files generated by the fv10-asw1.7 software to ImageJ (Research Services Branch, National Institute of Mental Health, Bethesda, Maryland, USA). Duration calculations per phase were determined manually and imported to MS Excel 2007 (Microsoft Corp.) to extract averages. Spindle and phragmoplast abnormalities were scored by visual inspection.

Mitotic onset was set at prophase, at which stage cells have a clear prophase spindle and PPB remnants. Earlier stages were excluded because of the indefinite duration of PPB formation and maturation.

The starting point of cytokinesis was set at the point immediately following the completion of anaphase and the formation of the characteristic telophase interzonal MT system. The end point was set exactly at the point at which the last remnants of the phragmoplast disappeared and the cortical MT array started to emerge, irrespective of whether the cell plate had fused with the parent walls. Thus, abortive cytokinesis in mutant or inhibitor-treated wild-type cells is considered to be any cytokinesis that failed to produce a full cell plate at the time of cortical MT array reinstatement.

Protein extraction and immunoblot analysis

Total proteins were extracted from liquid-nitrogen pulverized A. thaliana whole seedlings according to previously published procedures (Müller et al., 2010). Following quantification according to the Bradford method, proteins were mixed with Laemmli sample buffer and resolved in 10% sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE). Fractionated proteins were blotted on polyvinylidene fluoride (PVDF) membranes and processed according to standard protocols for enhanced chemiluminescence detection (Komis et al., 2004).


Expression pattern of MPK4 is consistent with its putative cytokinetic role

Tissue-specific analysis of MPK4 expression was carried out in PromMPK4:GUS transgenic plants. Although the MPK4 promoter was active in almost all root tissues, prominent GUS staining was present especially at the root tip containing meristematic cells (Supporting Information Fig. S1a). MPK4 promoter activity was also localized to other root regions with meristematic activity, including lateral root initial cells and lateral root primordia (Fig. S1b,c).

Cytokinetic defects, and bi- and multinucleate cells in anp2anp3 and mpk4 roots

Previous studies (Krysan et al., 2002; Beck et al., 2010) reported that the anp2anp3 and mpk4 mutants show MT-related phenotypes, such as cytokinetic defects and radial cell swelling of growing root, cotyledon, hypocotyl and leaf cells. Although such phenotypes are evident in all vegetative tissues of both mutants, further analysis of phenotypes caused by cytokinetic aberrancies was restricted to the root.

To gain a general overview of the root tissue pattern, wild-type and mutant roots were incubated with 4 μM of the styryl dye FM4-64 or AM4-65; these dyes are incorporated into the plasma membrane and thereby outline the cell periphery and the nascent cell plates (Dhonukshe et al., 2006). Whereas wild-type roots showed an orderly arrangement of cell files, tissue patterns were visibly disturbed in both the mpk4 (compare Fig. 1b with Fig. 1a) and the anp2anp3 (compare Fig. 1d with Fig. 1c) mutants. Closer inspection of FM4-64/AM4-65-stained roots showed the occurrence of several incomplete cell plates and cell wall stubs in both mutants (Fig. 1e,f). Co-staining of roots with DAPI (4′,6-diamidino-2-phenylindole) and FM4-64/AM4-65 revealed that the majority of enlarged cells, including those with cell wall stubs, were bi- or multinucleate (Fig. 2, Table 1). The cytokinetic phenotype was much stronger in the mpk4 mutant. Therefore, root anatomy and cellular structure were also examined in toluidine blue-stained semithin sections (this dye stains both cell walls and nuclei dark blue) (Fig. S2). Also in this case, several multinucleate, giant cells were found, all of which possessed incomplete cell walls. For the quantification of multinucleate cells in mutants compared with their respective control, cells in one plane at a distance of 180–200 μm from the root tip were inspected, and those with at least two nuclei were counted as multinucleate. In the anp2anp3 mutant c. 8.6 ± 4.4% (mean ± standard error) of the investigated cells possessed more than one nucleus, and in the mpk4 mutant the percentage of multinucleate cells increased c. 3-fold (27.6 ± 7.1%; Table 1). No multinucleate cells were observed in either of the control lines, Ws and Col-0 (Figs 2a,c, S2a).

Figure 1.

 Fluorescence imaging of Arabidopsis thaliana Columbia (Col-0), mitogen-activated protein kinase (mpk)4, Wassilevskaja (Ws) and Arabidopsis homologue of nucleus and phragmoplast associated kinase (anp)2anp3 roots labelled with the plasma membrane marker FM4-64. (a) Root tissue pattern of Col-0 showing the typical cross-wall arrangement in cells organized in regular cell files. (b) Disturbed tissue pattern in the root of the mpk4 mutant. Some cell plates appear incomplete (arrows). (c) Normal arrangement of root cell files in a Ws root. (d) Aberrant tissue pattern in the anp2anp3 mutant with cells showing cell wall stubs (arrows). (e, f) Higher magnifications of FM4-64-stained anp2anp3 (e) and mpk4 (f) roots showing aberrant and incomplete cell plate formation (arrows). Bars: (a–d) 10 μm; (e, f) 5 μm.

Figure 2.

 Bi- and multinucleate cells in Arabidopsis thaliana anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) and mitogen-activated protein kinase (mpk)4 mutants. Plasma membranes were labelled with AM1-65 (red) and nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI) (blue), except for (e) where FM4-64 staining (red) was merged with the Differential Interference Contrast (DIC) channel in order to visualize nuclei. Nuclei are marked with stars. (a, c) Wild-type (WT) (Columbia (Col-0) and Wassilevskaja (Ws)) plants with regularly formed cell files. Note that each cell contains only one nucleus. (b, d) Both mutants showed high numbers of binucleate (b, d) and multinucleate (e–i) cells. Bars, 5 μm.

Table 1.   Quantification of multinucleate cells in Arabidopsis thaliana anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) and mitogen-activated protein kinase (mpk)4 mutants
Mean number (±SD) of cells% of multinucleate cellsMean number (± SD) of cellsMean number (± SD) of multinucleate cellsAverage % (± SD) of multinucleate cells
  1. Cells in one plane per root tip (the number of root tips was seven) were counted, and the mean number (± SD) of all cells and multinucleate cells (defined as cells possessing more than one nucleus) is given in the table. Both controls (Wassilevskaja (Ws) and Columbia (Col-0)) showed no multinucleate cells, whereas in the mutants 8.6% (anp2anp3) and 27.6% (mpk4) of cells had two or more nuclei. The reduced mean number of cells per root-tip plane in both mutants in comparison with Ws and Col-0 controls is a result of the enlargement of multinucleate cells.

128.7 ± 17.8
90.0 ± 23.7
6.9 ± 2.8
8.6 ± 4.4
117.6 ± 10,0
63.3 ± 19.9
13.0 ± 2.5
27.6 ± 7.1

Immunofluorescence visualization of mitotic spindles and phragmoplasts in anp2anp3 and mpk4 roots is consistent with a multinucleate phenotype

The observations reported in the previous section indicated that there were defects associated with the dynamics of phragmoplast expansion. To gain further insights into the mitotic MT organization of anp2anp3 and mpk4 mutants, roots were processed for whole-mount tubulin immunolocalization and compared with the control. An overview of the wild-type root meristem in 5-d-old seedlings showed the normal complement of mitotic stages (Fig. S3a). These included PPBs of variable width (Fig. S3c,d), mitotic spindles of different mitotic stages (i.e. prophase, metaphase and anaphase; Fig. S3e,f) and phragmoplasts (Fig. S3g,h).

As stated previously, root patterning appeared disturbed due to the disturbance of both cell expansion which changes from longitudinal to lateral but also to malorientation of the cell division plane, which in the control is mostly perpendicular to the longitudinal root axis (Fig. S4). Indeed, a survey of tubulin immunolabelled root meristems of both mutants showed the formation of oblique cell plates (cf. Fig. S3 with Fig. 3). Most strikingly, the number of mitotic figures appeared to be increased in both mutants (compare Fig. S3a with Fig. 3a,d), resulting in mitotically synchronized cell clusters at the same or similar mitotic stages (compare Fig. 3b,c,h with Fig. 3g). When such clusters of mitoses are found in wild-type roots, they normally consist of asynchronous cell groups (Fig. 3g). Finally, in accordance to the phenotype analysis described above for mpk4, or those previously published for anp2anp3 (Krysan et al., 2002), we found bi- and multinucleate cells in both respective mutants (Figs 3b,e,f, outlined cells; S4).

Figure 3.

 Organization of mitotic and cytokinetic microtubule (MT) arrays in 4′,6-diamidino-2-phenylindole (DAPI)-counterstained mitotic cells of Arabidopsis thaliana anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) and mitogen-activated protein kinase (mpk)4 mutants and of the Columbia (Col-0) control. (a) Overview of the anp2anp3 root meristem showing the over-accumulation of mitotic and cytokinetic figures (boxed area). (b) Higher magnification of the boxed area of (a), showing an eight-cell cluster with phragmoplasts and preprophase bands (PPBs). (c) Another example of accumulation of mitotic figures. (d) Low magnification of the mpk4 root meristem. (e) Higher magnification of the rightmost boxed area of (d), showing a multinucleate cell with multiple phragmoplasts (arrows) and a binucleate cell with two metaphase spindles (bracket). (f) Higher magnification of the leftmost boxed area of (d), showing another multinucleate cell with numerous phragmoplasts (arrows). (g) A file of asynchronously dividing cells in the Col-0 root. From left to right are an anaphase spindle, a bipolar prophase spindle and a robust PPB, a broad PPB, and finally an early phragmoplast. (h) A four-cell file of mpk4 cells synchronized at the early preprophase stage showing broad PPBs (asterisks). Bars: (a, d) 20 μm; (b, c, e–h) 5 μm.

Live-cell imaging of mitosis in control wild-type plants

To monitor the dynamics of mitosis and cytokinesis, wild-type and mutant plants were stably transformed with a 35S:GFP:MBD construct in order to visualize MTs and follow their rearrangements using time-lapse live imaging. Although the dynamic behaviour of mitotic and cytokinetic MT arrays was tissue-independent, in most cases we used dividing epidermal cells of cotyledons, which allow higher resolution. We were thus able to monitor the transitions of MT reorganization from prophase to metaphase to the anaphase spindle to the telophase interzonal MT system, and finally to the cytokinetic phragmoplast in Col-0 cells, in times ranging between c. 30 and 50 min (Fig. 4, Video S1). Quantitative analysis of both mitosis and cytokinesis in wild-type plants showed that the duration of mitotic stages, and, in particular, the duration of cytokinetic progression were essentially similar to previously published values (summarized in Table 2; cf. e.g. Mao et al., 2005), whereas in 23 visually inspected cells of Ws and 18 cells of Col-0 (both representing wild-type controls) no plants were found with abnormal mitotic/cytokinetic features. With particular reference to cytokinesis, it must be noted that its duration largely depends on the distance the nascent cell plate has to grow. For most cases, however, in order to allow direct comparisons with mutant and inhibitor-treated cells, we used cells like that depicted in Fig. 4, in which the duration of cytokinesis was 42.2 ± 2.3 min for Ws, and 46.3 ± 2.2 min for Col-0 (Table 2).

Figure 4.

 Time-lapse imaging of Arabidopsis thaliana wild-type mitosis in a dividing cotyledon epidermal cell of Columbia (Col-0), stably transformed with the 35S::GFP:MBD microtubule (MT) marker. Transitions of MT rearrangements are observed as the cell progresses from metaphase (frame 1), to anaphase (frame 2), to telophase (frame 3), and finally to cytokinesis (frames 4–6). Note the gradual expansion of the phragmoplast towards the parent walls (arrows in frames 4–6). Frames were chosen from Supporting Information Video S1. Bottom left numbers: time (in min and s) starting from the beginning of the observation. Bar, 5 μm.

Table 2.   Duration of discrete mitotic stages in Arabidopsis thaliana Wassilevskaja (Ws), anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) Columbia (Col-0), mitogen-activated protein kinase (mpk)4 and PD98059-treated Col-0 dividing cotyledon epidermal cells
  1. Duration (min ± SE) was calculated from time-lapse mitotic sequences of 35S:GFP:MBD-transformed cells. The preprophase duration is designated as ‘not applicable’ (na) because of the high variability of the duration of this stage. Metaphase and anaphase durations in anp2anp3, mpk4 and PD98059-treated cells were also very variable (roughly ranging from 30 min to over 2 h) but in every case exceeded 30 min on average. Numbers of cotyledon cells were: Ws, = 12; anp2anp3, = 8; Col-0, = 10; mpk4, = 6 and PD98059, = 8 from three to five individual plants except for PD98059 where only two plants were used. Statistical comparisons were made using Student’s t-test.

  2. †, < 0.05 compared with Ws; ‡, < 0.05 compared with Col-0; *, > 0.05 compared with Ws; ¶, > 0.05 compared with Col-0; **, < 0.05 compared with Ws; §, < 0.05 compared with Col-0.

Prophase/prometaphase15.7 ± 3.823.2 ± 7.1†17.9 ± 5.237.3 ± 9.2‡29.1 ± 3.5‡
Metaphase3.1 ± 0.3Variable, > 30 min4.1 ± 0.2Variable, > 30 minVariable, > 30 min
Anaphase4.6 ± 1.1Variable, > 30 min4.7 ± 1.8Variable, > 30 minVariable, > 30 min
Telophase1.9 ± 0.42.8 ± 0.2*2.0 ± 0.52.0 ± 0.3¶2.1 ± 1.1¶
Cytokinesis42.2 ± 2.368.7 ± 5.4**46.3 ± 2.274.3 ± 6.2§69.8 ± 6.2§

Live-cell imaging of mitosis in the anp2anp3 mutant

Live-cell imaging of the mitotic/cytokinetic progression in cotyledon cells of the anp2anp3 mutant transformed with the 35S:GFP:MBD construct showed various abnormalities in their mitotic spindle and phragmoplast rearrangements. In many cells, although the spindle structure resembled that of Ws cells, there was a significant mitotic delay resulting in the pausing of spindle rearrangements for indefinite periods (Fig. 5b, Video S2a,b). Evaluation of the duration of discrete mitotic stages such as metaphase, anaphase and cytokinesis revealed that these were significantly prolonged in anp2anp3 cells in comparison to the Ws control (Table 2). Eventually such cells may abort mitosis before the formation of a cytokinetic phragmoplast. Alternatively, they may reinstate the cortical MT system (Fig. 5a) or enter cytokinesis, which in most cases fails to be completed (Figs 5b, 6a,b; Video S3). Variations in these defects included the occurrence of multinucleate cells undergoing two (Fig. S5) or multiple mitoses (Fig. S6). It was also noteworthy that, in the case of dividing multinucleate cells, individual spindles or phragmoplasts could fuse to single entities (Fig. S7).

Figure 5.

 Time-lapse observations of defective mitotic microtubule (MT) transitions in dividing cotyledon epidermal cells of the Arabidopsis thaliana anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) mutant stably transformed with the 35S::GFP:MBD MT marker. (a) Selected frames from a time-lapse observation of a cell possessing a mini-spindle and a large spindle (arrows) captured from metaphase onwards. Both spindles transit to an atypical telophase system (frames 4 and 5) which eventually dissolves to yield cortical MTs (frame 6). (b) Selected frames from Supporting Information Videos S2(a,b), showing a cell with delayed mitotic and cytokinetic progression. Asterisks in the first three frames denote remnants of the preprophase band (PPB) indicating that the cell was at late preprophase/prophase at the onset of the recording. Asterisks in the last six frames denote the phragmoplast, while the stippled line delineates the parent wall of the neighbouring cell. The last frame is taken from Video S2(b), to emphasize the delay of phragmoplast expansion. It is noteworthy that, since the appearance of the phragmoplast (39 min 39 s), nearly 52 min have elapsed and still the phragmoplast has not fused with the parent walls. Bottom left numbers show time (in min and s). Bars: (a) 10 μm; (b) 5 μm.

Figure 6.

 Time-lapse observations of abortive phragmoplast expansion in dividing cotyledon epidermal cells of the Arabidopsis thaliana anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) mutant stably transformed with the 35S::GFP:MBD microtubule (MT) marker. (a) Top view of an expanding phragmoplast which fails to reach completion. (b) Side view of an expanding phragmoplast. The phragmoplast disappears before reaching the right-most parent wall. See also Supporting Information Video S3. In both (a) and (b), cell boundaries are delineated in the first frames. Bottom left numbers: time (min, s). Bar, 5 μm.

In some cases, mitotic and cytokinetic defects were accompanied by spindle or phragmoplast malformation, while in other cases the major problem was the long delay in mitotic and/or cytokinetic progression. Temporal analysis of anp2anp3 cytokinetic cells with structurally normal phragmoplasts revealed marked delays in cell plate expansion, which took on average 68.7 ± 5.4 min, significantly longer than the duration calculated for Ws cells of 42.2 ± 2.3 min (= 0.0027) (Table 2). Visual assessment of mitotic and cytokinetic figures in anp2anp3 showed that six of 34 cells aborted their mitosis, rapidly reversing to the interphase state, as judged by the reinstatement of a cortical MT system, and 26 of 34 cells failed to complete cytokinesis at any stage of phragmoplast expansion, while 20 of 34 cells displayed double or multiple spindles/phragmoplasts in a single cytoplasmic compartment (Table 3).

Table 3.   Summary of cell populations with aberrant mitotic/cytokinetic characteristics in Arabidopsis thaliana wild-types (Wassilevskaja (Ws) and Columbia (Col-0)), mutants (anp2anp3 (Arabidopsis homologue of nucleus and phragmoplast associated kinase) and mitogen-activated protein kinase (mpk)4) and wild-type treated with PD98059
 Number and % of cells showing mitotic/cytokinetic aberrancies as a result of impaired MAPK signalling
NormalAbortive mitosis (failure to form interzonal system/phragmoplast)Abortive cytokinesis (single or multiple)Multiple mitoses (abortive or complete)
  1. n, number of cells undergoing mitosis/cytokinesis. The term ‘abortive mitosis’ means mitosis that stopped before the formation of a phragmoplast, resulting in the disorganization of the spindle and the reinstatement of a cortical MT system (e.g. Fig. 5a). The term ‘abortive cytokinesis’ refers to the situation where the phragmoplast expands to some extent and disorganizes prematurely, failing to bring the cell plate to completion (e.g. Fig. 6). The term ‘multiple mitoses’ relates to cells showing double or multiple spindles in the same compartment (e.g. Supporting Information Fig. S7).

  2. MAPK, mitogen-activated protein kinase.

Ws (= 23)23 (100)0 (0)0 (0)0 (0)
anp2anp3 (= 34)2 (5.9)6 (17.7)26 (76.5)20 (58.8)
Col-0 (= 18)18 (100)0 (0)0 (0)0 (0)
mpk4 (= 21)0 (0)6 (28.6)21 (100)5 (23.8)
PD98059 (= 15)4 (26.7)3 (20)8 (53.3)5 (33.3)

Live-cell imaging of mitosis in the mpk4 mutant

Mitotic and cytokinetic defects of mpk4 cotyledon epidermal cells (Fig. 7) were more frequent than those observed in anp2anp3. Up to 27.6 ± 7.1% of the cells were multinucleate (Table 1), undergoing asynchronous (Fig. 7a) or synchronous (Fig. 7b) multiple mitoses. In general, metaphase, anaphase and cytokinesis were prolonged in mpk4 cells when compared with Col-0 cells (Table 2). In particular, cytokinesis from start to completion (full cell plate formation) or to premature dissolution of the phragmoplast (incomplete cell plate formation) lasted 74.3 ± 6.2 min. This contrasts with the mean duration of 46.3 ± 2.2 min for Col-0 (= 0.01) (Table 2). As in the case of the anp2anp3 mutant, aberrant mitotic figures were excluded from this analysis. Such figures were found in six cells (out of 21 in total) which showed a failure to transit from mitosis to cytokinesis, and in 21 (of 21 cells) which showed incomplete phragmoplast expansion, with five of these 21 cells containing two or more nuclei (Table 3).

Figure 7.

 Mitotic progression in dividing cotyledon epidermal cells of the Arabidopsis thaliana mitogen-activated protein kinase (mpk)4 mutant stably transformed with the 35S::GFP:MBD microtubule (MT) marker. (a) A multinucleate cotyledon epidermal cell of mpk4 undergoing asynchronous multiple mitoses. All the phragmoplasts have disassembled before reaching the parent walls. At the onset of the observation, the cell shows an atypical spindle, two mini-spindles, a normal-sized phragmoplast (block arrow) and a mini-phragmoplast (thin arrow). (b) Another dividing cell of mpk4 with three synchronously expanding phragmoplasts, all of which dissolve before the completion of cytokinesis. Bottom left numbers: time (min, s). Bars, 10 μm.

Mitotic and cytokinetic disorders of anp2anp3 and mpk4 are mimicked by pharmacological MAPK inhibition

To test the possibility that the mitotic/cytokinetic defects described in the anp2anp3 and the mpk4 mutants are related to MAPK activity, we treated cotyledons of wild-type plants stably transformed with the 35S::GFP:MBD construct with the potent and specific MAPKK inhibitor PD98059. Probably because of its broad specificity for any putative MAPK involved in the mitotic/cytokinetic progress, the inhibitor had broader effects than those observed in either mutant. The most adverse effect of PD98059 was the induction of abnormal spindle formation, leading to abortive mitosis. In Fig. 8 (a), a multinucleate cell with three abnormal spindles is shown. It is arrested in the anaphase state and is gradually reinstating the cortical MT system. Otherwise, delayed mitoses were routinely encountered, whereby spindles persisted at stages that normally progress rapidly (e.g. compare Fig. 4 with Fig. 8b). An example can be seen in Video S5, which shows a binucleate cell undergoing synchronous mitosis, representing an extreme case of metaphase arrest with a duration exceeding 2 h. Additionally, Table 2 shows that the duration of every mitotic stage was significantly prolonged in PD98059-treated cells compared with their untreated counterparts. Importantly, cytokinesis lasted 69.8 ± 6.2 min compared with 46.3 ± 2.2 min (= 0.011) in Col-0 (Table 2).

Figure 8.

 The effects of PD98059 in Arabidopsis thaliana cotyledon epidermal cell mitosis. (a). A single dividing cotyledon epidermal cell of Columbia (Col-0), stably transformed with the 35S::GFP:MBD microtubule (MT) marker following treatment with 20 μM PD98059 for 6 h. The cell contains four atypical spindles. Two of them appear normal-sized (wide arrows) while the other two are probably mini-spindles (narrow arrows) joined at one pole. Upon failure to complete mitosis, spindles gradually disintegrate, and the cortical MT system reforms without the formation of a phragmoplast. (b) A cell undergoing synchronous mitosis with two spindles. Both spindles persist at metaphase for a period exceeding 2 h. Frames were selected from Supporting Information Video S5. Bottom left numbers show time (in min and s). Bar, 10 μm.

Immunolocalization of MPK4 to phragmoplasts in the root meristem of A. thaliana

In view of the mitotic and cytokinetic defects observed in the mutant and inhibitor treated cells, we speculated that MPK4 might colocalize with mitotic MT arrays in order to control MT organization. To investigate this possibility, we performed MPK4 intracellular localization using immunofluorescence staining in root whole-mounts co-labelled with anti-tubulin antibodies. The specificity of the anti-MPK4 antibody (recognizing a single 43-kDa band on immunoblots from crude A. thaliana extracts) was shown previously (Beck et al., 2010). It is noteworthy that, when the anti-MPK4 antibody was used for immunofluorescence labelling of mpk4 mutant roots, no specific detection at the phragmoplasts, cell plates or elsewhere in cells could be observed, further supporting the specificity of this antibody (Fig. S8).

In interphase and pre-mitotic cells, some MPK4 was localized to cytoplasmic spots (Fig. 9a). These perhaps represented localization of MPK4 to vesicular subcellular compartments, which were recently characterized for MPK6 (Müller et al., 2010). In cells undergoing cytokinesis, MPK4 colocalized with MTs in the phragmoplast (Fig. 9b). In post-cytokinetic cells, where remnants of the phragmoplast were still visible, MPK4 localized to spots along the nascent cell plate (Fig. 9c). By using an antibody that specifically recognizes the dually phosphorylated, activated form of MAPKs (anti-pERK), robust phragmoplast decoration could be observed (Fig. 9d). pERK detection could be eliminated by treating roots with PD98059 (Fig. 9e), but this treatment did not affect MPK4 localization to the phragmoplast (Fig. 9h), suggesting that it probably does not depend on the phosphorylation status of MPK4. As was shown by Beck et al. (2010), this treatment inhibits the dual phosphorylation and activation of MPK4. The anti-pERK antibody was tested by immunoblotting on whole protein extracts from Col-0 seedlings (Fig. S8). This anti-pERK antibody detected multiple bands, including one at 43 kDa (presumably corresponding to dually phosphorylated MPK4; Fig. S9), reflecting the multitude of MAPKs encoded by the A. thaliana genome.

Figure 9.

 Immunofluorescent colocalization of MPK4 (a–c) and phosphorylated Extracellular Signal Related Kinases (pERKs) (d–f) with microtubules (MTs) in the Arabidopsis thaliana root meristem. (a) Low magnification of the root meristem showing accumulation of mitogen-activated protein kinase (MPK)4 in cells possessing preprophase bands (PPBs) (asterisks). (b) Higher magnification of a cytokinetic cell showing MPK4 labelling along the entire length of phragmoplast MTs (arrows). (c) A post-cytokinetic cell showing a punctuate pattern of MPK4 localization along the nascent cell plate (arrows). (d) pERK localization along a cytokinetic phragmoplast. A prominent signal is also detected in the daughter nuclei (arrows). (e) Absence of pERK signal from the phragmoplast of a cytokinetic cell following treatment with 20 μM PD98059 for 2 h. (f) Prominent staining of MPK4 along the phragmoplast MTs of a cytokinetic cell following treatment with 20 μM PD98059 for 2 h. Bars: (a) 10 μm; (b–f), 5 μm.

Immunolocalization of MAP65-1 to PPBs and phragmoplasts in the root meristem of A. thaliana

Previous in vitro studies have shown that MAP65-1 is phosphorylated by MPK4 (Smertenko et al., 2006). Further biochemical co-precipitation and co-immunoprecipitation studies revealed physical association of MAPKs, such as MPK4 and MPK6, with MAP65-1 in A. thaliana (Beck et al., 2010; Müller et al., 2010). In this study, correlative co-immunolocalization of MAP65-1 and MTs in root whole-mount preparations revealed that MAP65-1 colocalized with both PPB (Fig. S10a–c) and phragmoplast MTs (Fig. S10e,f). In the latter case, MAP65-1 colocalized with phragmoplast MTs similarly to MPK4 (compare Fig. 9e,f with Fig. 9b). This suggested accumulation of both MAP65-1 and MPK4 at the same cytokinetic structures represented by phragmoplasts. The specificity of the anti-MAP65-1 antibody as well as interaction between MPK4 and MAP65-1 and putative MPK4-dependent MAP65-1 phosphorylation were tested previously (Beck et al., 2010, and references therein). Thus, MAP65-1 cannot be excluded as one of putative substrates of MPK4 during cytokinesis.


In the present study, one major stress-activated MAPK of A. thaliana, namely MPK4, is shown to be involved in the regulation of mitotic and cytokinetic progression in A. thaliana. Although the involvement of MAPKs in mitosis and cytokinesis is firmly established in plants, the potential role of MPK4 in mitotic/cytokinetic regulation is reported here for the first time. This role is deduced from live-cell imaging and characterization of mitosis and cytokinesis in two mutants, namely anp2anp3 and mpk4, both of which show unique abnormalities. Although these mutants show similar mitotic/cytokinetic aberrations, it seems that such defects are much more frequent in the mpk4 mutant. This prompts us to suggest that, although ANP2/ANP3 signalling is probably coupled to MPK4 to control MT dynamics and organization, in the absence of ANP2 and ANP3, MPK4 may be activated by redundant MAPKKKs such as MEKK1/MEKK2 (MAP kinase kinase kinase) (Su et al., 2007), thus overcoming the ANP2/ANP3 deficiency. Actually, the phenotype reported for mekk1 mutants (e.g. Gao et al., 2008), which is similar to that of mpk4 reported here, may support this notion. Nevertheless, further studies are necessary to investigate the possibility of such an alternative mechanism of MPK4 activation.

Expression and localization of MPK4 are consistent with its potential role in cell division

Transcripts of MPK4 are abundant in the root, but also in other organs (Beck et al., 2010). However, tissue-specific visualization of MPK4 promoter activity revealed that it was particularly increased in the primary root tip meristem as well as in meristematic cells of lateral roots. Furthermore, immunolocalization studies showed that MPK4 abundance was increased in phragmoplast MT arrays, adding to previous data showing direct or indirect MT–MPK4 interactions (Beck et al., 2010). Thus, MPK4 is a novel MAPK localizing to the phragmoplast. Other such MAPKs include p34Ntf6 and NRK1 in tobacco and MMK3 (medicago MAPK) in Medicago sativa, each of which is actively involved in cytokinetic progression (Calderini et al., 1998; Bögre et al., 1999), and also MPK6 in A. thaliana, which is involved in the correct orientation of the cell division plane (Müller et al., 2010). Finally, SIMK (stress inducible MAPK) from M. sativa was shown to associate with actin filaments in root hairs (Šamaj et al., 2002) and with developing phragmoplasts in dividing cells (Šamaj et al., 2004). In the latter case, the association was more extensive with taxol treatment, inducing overstabilization of MTs.

The recruitment of MPK4 to the phragmoplast is likely to be activation-independent

Immunolocalization with the pERK antibody showed prominent decoration of the phragmoplast which was effectively eliminated upon inhibition of MAPKK activity by PD98059. However, this particular treatment had no effect on the localization of MPK4, which apparently remained unaffected.

In plants, members of the PQR pathway localize to phragmoplast MTs by virtue of the interaction between the NPK1 MAPKKK and two kinesins, namely NACK1 and NACK2 (Ishikawa et al., 2002; Nishihama et al., 2002). In this case, such an interaction also activates NPK1 by releasing it from an autoinhibitory state (Nishihama et al., 2002). It is not currently possible to speculate on the mechanism underlying the recruitment of MPK4 to the phragmoplast because the interactome of MPK4 has not yet been fully characterized (Popescu et al., 2009). A report by Smertenko et al. (2006) demonstrated an interaction of MPK4 with MAP65-1, but it could not be determined whether this interaction mediates the binding of MPK4 to the phragmoplast MTs. Nevertheless, our immunolocalization studies have revealed that both proteins are localized to the phragmoplast with very similar patterns. Importantly, Smertenko et al. (2006) showed that the MPK4–MAP65-1 interaction is functional and results in the phosphorylation of MAP65-1. In a previous study, we found reduced levels of phosphorylated MAP65-1 in the mpk4 mutant (Beck et al., 2010).

Functional implications of MAPK–MT interactions in A. thaliana

Live imaging revealed unique aspects of mitotic and cytokinetic disturbances in the anp2anp3 and mpk4 mutants and more generalized effects of the inhibitor PD98059. Most dividing cells of the anp2anp3 and mpk4 mutants and PD98059-treated wild-type cells showed defective organization of the mitotic spindle and phragmoplast. In most cases, however, the major differences were in the duration of the mitotic stages, particularly those following metaphase.

For this reason, we postulate that the importance of MPK4 and of MAPKs in general lies mainly in the temporal control of mitotic changes in MT organization and dynamics. The above MT defects observed in mutant and inhibitor treated mitotic cells are in agreement with the suggestion by Smertenko et al. (2006) that MAPK signalling plays a significant role in the regulation of spindle form and dynamics, especially after the end of metaphase. The putative MAPK substrate MAP65-1 also becomes localized to mitotic MT arrays after metaphase and especially during telophase and cytokinesis (Mao et al., 2005; Smertenko et al., 2006). It is thought that the phosphorylation of MAP65-1 by CDKs (cyclin dependent kinases) and MAPKs reduces the affinity of MAP65-1 for the MT lattice up to the end of metaphase, thereby explaining its absence from the mitotic spindle before anaphase. These observations also suggest that the mode of regulation contributed by MAPKs is mainly related to MT dynamics. Studies in mammals showed that certain proteins regulating MT dynamics during mitosis are coupled to MAPK signalling. Examples include the stathmin/Op18 MT catastrophe-promoting factor, which is targeted by the p38 MAPK (Parker et al., 1998), and MAP2, which is phosphorylated by ERKs (Hoshi et al., 1992; Morishima-Kawashima & Kosik, 1996). In the above examples, as well as in other cases involving MT crosslinking proteins, phosphorylation renders MAPs incapable of MT binding, thus reducing or eliminating their effect on MT dynamics and cross-linking.

Members of the MAP65 protein family are likely candidate targets of MPK4

Previous studies have associated members of the MAP65 family with MAPK signalling (Sasabe et al., 2006; Smertenko et al., 2006). MAP65-1 from A. thaliana is phosphorylated by MPK4 and MPK6 (Smertenko et al., 2006) and also by the heterologous NRK1 of tobacco. Similarly, MAP65-2 and MAP65-3 are also phosphorylated by NRK1 (Sasabe & Machida, 2006). The tobacco homologue of MAP65-1, which is targeted and phosphorylated by NRK1 (Sasabe & Machida, 2006), is involved to the regulation of MT dynamics during phragmoplast expansion. However, this does not seem to be the case for A. thaliana MAP65-1, as it is mainly involved in MT bundling (Smertenko et al., 2006).

Our previous study (Beck et al., 2010) showed the up-regulation and reduced phosphorylation of MAP65-1 as well as down-regulation of MAP65-3 in the mpk4 mutant. It was also shown that MAP65-1 co-immunoprecipitated with MPK4, while MPK4 precipitated with heterologous (bovine brain) pre-assembled MTs, when these were added to a pre-cleared A. thaliana supernatant (Beck et al., 2010). The MPK4–MAP65-1 interaction and the subsequent phosphorylation of MAP65-1 by MPK4 (as documented in Smertenko et al., 2006; Beck et al., 2010) can explain the mitotic and cytokinetic aberrations observed in this study. Artificial overexpression of MAP65-1 driven by the 35S tobacco mosaic virus promoter induced the formation of abnormal spindles and phragmoplasts, as a result of extensive MT crosslinking (Mao et al., 2005). Additionally, multiphosphomimetic analogues of MAP65-1 decreased MT-binding efficiency, while the knock-down of MAP65-3 displayed a robust cytokinetic phenotype (Müller et al., 2004; Smertenko et al., 2006). However, the MAP65-3/PLEIADE mutation produced cytokinetic defects restricted to the root and did not affect other organs (Müller et al., 2002). By contrast, both the anp2anp3 and the mpk4 mutants showed widespread cytokinetic defects.

The purpose of the present study was to describe in detail the cytokinetic defects in the mpk4 mutant which were previously mentioned but not characterized (Beck et al., 2010). Live imaging of mitotic spindles and phragmoplasts showed that these defects were dependent on MT organization and dynamics. Comparing the mpk4 mutant with a known MAPKKK double mutant, anp2anp3, with definite cytokinetic defects (Krysan et al., 2002), and taking into consideration other similarities in MT organization between the two mutants, it seems likely that a pathway integrating ANP2/ANP3, MPK4 (this study)/MPK6 (Müller et al., 2010) and perhaps MAP65-1 as one of the MAPK substrates, might be globally important for the organization and dynamics of mitotic MTs in vegetative organs. The involvement of MPK4 in mitotic/cytokinetic progression in relation to dynamic MT rearrangements further expands the complexity of MAPK signalling in plant cells.


This work was supported by a graduate student fellowship from the University of Bonn to Martina Beck, an Alexander von Humboldt post-doctoral fellowship to George Komis, DFG grants (nos SA 1564/1-2 and SA 1564/2-3) and structural research grants from the EU and the Czech Republic to the Centre of the Region Haná for Biotechnological and Agricultural Research, Faculty of Science, Palacký University, Olomouc, Czech Republic (grant no. ED0007/01/01). We gratefully acknowledge the donation of mutant seeds and antibodies by Heribert Hirt (University of Vienna, Vienna, Austria), Patrick Krysan (University of Wisconsin, Madison, WI, USA) and Andrei Smertenko (University of Durham, Durham, UK). We would like to thank Dr Peter Barlow (University of Bristol, Bristol, UK) for English editing and correction. Thanks are extended to Ursula Mettbach and Claudia Heym for expert technical assistance.