The lue1 mutant was previously isolated in a bio-imaging screen for Arabidopsis mutants exhibiting inappropriate regulation of an AtGA20ox1 promoter-luciferase reporter fusion. Here we show that lue1 is allelic to fra2, bot1 and erh3, and encodes a truncated katanin-like microtubule-severing protein (AtKSS). Complementation of lue1 with the wild-type AtKSS gene restored both wild-type stature and luciferase reporter levels. Hormonal responses of lue1 to ethylene and gibberellins revealed inappropriate cortical microtubule reorientation during cell growth. Moreover, a fusion between the AtKSS protein and GFP decorated cortical microtubules. A yeast two-hybrid screen with AtKSS as the bait identified proteins related to those involved in microtubule processing, including a katanin p80 subunit and a kinesin ortholog. These results indicate that AtKSS is involved in microtubule dynamics in response to plant hormones.
The shape and growth orientation of plant cells are primarily controlled by cellulose microfibrils (CMF) of the cell wall and by microtubules (MT) of the cytoskeleton (reviewed in Kost et al., 1999; Azimzadeh et al., 2001; Wasteneys, 2002). The molecular mechanisms involved are being analyzed via the characterization of mutants with altered cell wall and MT organization. For example, both Arabidopsis rsw1 and kor mutants exhibit altered cellulose microfibril (CMF) organization and composition, and RSW1 and KOR, respectively, encode a cellulose synthase subunit (Arioli et al., 1998) and a membrane-bound endo (1-4)-β-D glucanase (Nicol et al., 1998; Zuo et al., 2000). More recently, the mor1 mutant was shown to be defective in MT organization. MOR1 encodes a new class of plant MT-interacting proteins (Whittington et al., 2001). Three other Arabidopsis allelic mutants, bot1, fra2 and erh3, exhibit disorganized CMT, leading to isotropic cell growth, inflorescence stem fragility and ectopic root hair (Bichet et al., 2001; Burk et al., 2001; Webb et al., 2002). These loci encode a 60 kDa microtubule-associated ATPase katanin ortholog designated AtKSS (McClinton et al., 2001). In animal cells, katanin is a heterodimer consisting of the 60 kDa ATPase that harbors MT-severing activity and an 80 kDa subunit that targets the heterodimer to centrosomes (Hartman et al., 1998; McNally et al., 2000). The in vitro MT-severing activity of AtKSS has recently been reported (Stoppin-Melllet et al., 2002).
Plant MT organization is under the control of external stimuli and endogenous signals, including hormones (reviewed by Shibaoka, 1994). For example, auxin, gibberellin (GA) and brassinosteroid treatments lead to a modification of cortical MT (CMT) orientation into a transverse array (Ishida and Katsumi, 1992; Baluska et al., 1993; Zandomeni and Schopfer, 1993). By contrast, ethylene (ET) and abscisic acid, an antagonist of GA, promote an oblique orientation. In the GA-deficient maize d5 dwarf, CMTs exhibit an oblique orientation that can be restored to the wild-type transverse orientation upon GA application, which results in normal growth (Baluska et al., 1993). Moreover, the use of a GA biosynthesis inhibitor leads to a CMT misorientation in wild-type root cells that is similar to that in d5. More recent work has confirmed the role of GA in reorienting the CMT network in root and leaf cells (Inada and Simmen, 2000; Wenzel et al., 2000). Despite these observations, molecular data linking GA responses to CMT organization remain sparse.
Two major genes affecting responses to GA have been identified in phenotypic screens for growth mutants. Dominant mutations of GAI/RGA, which encode GRAS proteins proposed to function as the metazoan STAT transcription factors (Richards et al., 2000), result in semi-dwarfism, an important agronomic trait. Recessive mutations in SPY, which encodes an O-linked N-acetylglucosamine transferase [OGT (Thornton et al., 1999)], result in elongated plants with a constitutive GA-response phenotype. In attempts to identify additional genes involved in GA responses, we developed a fusion genetic approach to identify trans-acting mutations affecting the expression of transgenes composed of the firefly luciferase reporter under the control of the promoter of AtGA20ox1, which encodes the biosynthetic GA20-oxidase (Meier et al., 2001). The AtGA20ox1 promoter was used because expression from it is regulated through negative feedback by active product GAs; therefore, cis-elements in the promoter may be targets for GA signaling pathways. This screen identified the recessive, semidwarf lue1 mutant. Lue1 exhibited constitutive, high levels of LUC reporter and AtGA20ox1 mRNA, as well as inappropriate feedback regulation of the endogenous AtGA20ox1 and At3ox1 biosynthetic genes by GA. Additionally, wild-type stature could not be rescued by GA applications. These results indicated that the sensitivity of lue1 to GA was altered at the levels of both GA biosynthetic feedback and vegetative cellular responses.
We show here that lue1 is allelic to fra2 and bot1. Complementation of lue1 with the wild-type AtKSS gene restored normal stature and luciferase reporter levels. Treatments of lue1 with ET and GA revealed inappropriate hormonal responses related to cell growth. A reporter fusion between AtKSS and GFP revealed that AtKSS decorates the CMT in a punctate pattern. Moreover, a yeast two-hybrid screen performed with AtKSS as the bait identified proteins related to those involved in MT growth and processing, including a katanin p80 and a large protein containing a kinesin-like domain. Potential links between GA signalling and MT organization are discussed.
Materials and Methods
Plant material and treatments
Seeds were surface-sterilized, plated on MS medium supplemented with 7% agar, 1% sucrose and appropriate hormones, then vernalized in the dark at 4°C for 4 days. For gibberellin treatments, 1-week-old seedlings were transferred to soil and grown under long-day conditions. Plants were sprayed twice a week with 10 μM GA3 or 10 μM GA4 (Sigma) until bolting. Lue1 mutant sensitivity to ET in the dark was investigated by plating wild-type and lue1 seeds on MS medium supplemented with 7% agar, 1% sucrose with or without 50 μM ACC (Sigma); thereafter plates were incubated vertically (to facilitate root measurements) at 21°C in the dark. Hypocotyl width was measured in the region exhibiting maximum diameter. Lue1 sensitivity to ET under light conditions was performed as described previously (Smalle et al., 1997), except that 50 μM ACC (Sigma) was used instead of gaseous ET. The ga1-1 mutant was obtained from the Nottingham Arabidopsis Stock Center. N.-H. Chua, Rockefeller University, kindly provided plants expressing the GFP-talin and microtubule-associated protein 4 (MP4)-GFP reporters that, respectively, decorate actin (Kost et al., 1998) and MT (Mathur and Chua, 2000). The lue1 mutant was crossed with the GFP-talin and GFP-MAP4 lines. F2 segregating seedlings were then scored for WT or lue1 phenotype before being subjected to confocal laser-scanning microscopy.
AtKSS gene cloning and lue1 complementation
The AtKSS gene was PCR-amplified with template DNA from BAC F5I6 and linker-primers 5′ACAAGCTTGTTGGTCCTGGCCAG-TCAGAC and 5′CTTAGATCTACATCCGGAGTCCTCCTTAGC. Products were digested with HindIII and BglII and subcloned in the HindIII and BamHI sites of the pCambia3300 vector (Cambia, Canberra) to produce C3300-AtKSS. This construct was introduced into Agrobacterium tumefaciens (PGV3101) by electroporation, which was used to transform the lue1 mutant by vacuum infiltration (Bechtold and Pelletier, 1998). T1 generation seedlings were selected in soil for phosphinothricin resistance expressed from the pCambia3300 T-DNA by spraying seedlings every 3 days with 10 mg/l Bastamycin (AgrEvo, Denmark).
AtKSS wild-type genomic DNA was PCR-amplified by RT-PCR using linker-primers 5′AGATCTGGGAAGTAGTAATTCGTTAGCGGGTC and 5′AGATCTCCAAACTCAGAGAGCCACTTCTCGTG. PCR was performed for 30 cycles using Vent DNA polymerase (NEB) and Arabidopsis Col0 genomic DNA as the template. The product was digested with BglII and cloned into the same site of pCAMBIA1304 (Cambia, Canberra) to produce the C1304-AtKSS-GFP-GUS fusion reporter. The AtKSS sequence and correct fusion open reading frame were confirmed by sequencing. This construct was transformed via the A. tumefaciens strain PGV3101 into lue1 and Arabidopsis Col ecotype by vacuum infiltration. T1 seedlings were selected for hygromycin resistance, carried on the pCAMBIA1304 T-DNA, on MS plates with 50 mg/l hygromycin B. Homozygous single insertion lines were selected from the T3 generation and approximately 20 lines analyzed further.
Equipment and protocols for LUC bioluminescence imaging were described previously (Meier et al., 2001). GFP was visualized using a Zeiss LSM 510 laser-scanning microscope applying the 488 nm line of the argon laser and the corresponding dichroic mirror and a 505-530 nm band-pass filter. For reference, chlorophyll fluorescence and a Nomarski image were recorded simultaneously.
Total RNA (10 μg) extracted with the RNAgent kit (Promega) was fractionated on standard formaldehyde gels and blotted onto Hybond-N+ membranes (Amersham). AtKSS mRNA levels were investigated by hybridization with a ribonucleic [32P]CTP antisense probe synthesized with T7 RNA polymerase (Ribokit, Promega) from a full-length cDNA cloned in the pGEM-Teasy vector (Promega). AtKSS primers for cDNA amplification were 5′GTTAGCGGGTCTACAA-GACCAC and 5′ACTCAGAGAGCCACTTCTCGTG. Hybridization and washing conditions were performed as recommended by the manufacturer.
Analysis of CMF orientation by polarizing microscopy
For polarizing microscopy, 5 mm segments from the lower part of the flowering stem were fixed in buffered 4% formaldehyde, embedded in paraffin wax and sectioned longitudinally at 8 μm. Positions of maximum extinction closest to the polarizing plane of the analyzer were determined visually by rotating the stage of the polarizing microscope (Frey-Wyssling, 1959). The angular absolute values of the difference between these positions and those of the long cell axis parallel to the analyzer plane were determined. The distinction between extinction parallel to the analyzer plane and to the polarizer plane was made by insertion of a red first order compensator (sensitive tint plate).
AtKSS protein interaction analyses
The full-length AtKSS cDNA bait was amplified by RT-PCR with linker-primers AAA1-2hyb-F (GAGGAATTCGTGGGAAGTAGTA-ATTCGTTAGCG) and AAA1-2hyb-R (GGGAGATCTTAAGCAGA-TCCAAACTCAGAGAGC). The product was digested with BamHI and BglII and subcloned in the BamHI site of pGBKT7 (Clontech Matchmaker System III) to produce pGBKT7-AtKSS, which was introduced into yeast strain PJ69A-4A. This bait strain was transformed with a cDNA library from mature leaf mRNA in vector pGAD10 (Clontech, FL4000AB). The yeast two-hybrid screen was performed according to the manufacturer's instructions for prototrophic growth on medium lacking tryptophan (TRP), leucine (LEU), histidine (HIS) and adenine (ADE) for 4 days at 30°C. A total of 15×106 transformants were screened to yield some 1000 positive clones. These clones were assayed for β-galactosidase activity, which eliminated roughly 50% of the clones. DNA was extracted from the remainder and approximately 100 clones were used as PCR templates with pGAD10 primers (Clontech, 9103-1) to size inserts and for sequencing. Sequencing allowed us to discard clones containing frame shifts between the GAL4-binding domain (GAL4-BD) and the prey clones or clones inserted in the reverse orientation. One clone from each of the remaining prey insert groups was mobilized in E. coli and re-sequenced. To confirm interactions, bait and prey plasmids were individually co-transformed into PJ69A-4A and re-evaluated for prototrophic growth and β-galactosidase activity. To confirm AtKSS interactions, prey cDNAs were subcloned into the T7 RNA polymerase promoter-containing pGAD-T7 plasmid (Clontech). Additionally, a truncated version (1.25B2) of the KTN P80.1 clone 1.25 was constructed by restriction digestion with BglII and subsequent cloning into the BamHI site of pGADT7. Correct orientation of the insert and frame were assessed by restriction digestion and sequencing. All pGADT7-based constructs were confirmed for interaction with AtKSS in directed yeast two-hybrid assays by co-transformation of the yeast PJ69A-4A with pGBKT7-AtKSS. Additionally, an empty pGBKT7 (i.e. without the AtKSS fusion) was co-transformed with the pGADT7-based vectors to confirm that both β-galactosidase activity and prototrophic growth require AtKSS-prey proteins interaction. To further confirm AtKSS interactions, bait and prey proteins were synthesized by in vitro transcription/translation using the T7-RNA-polymerase-based TnT kit (Promega, #L4610) in the presence of 35S-methionine and co-immunoprecipitated with protein-G-coupled Dynabeads (Dynal Biotech, #100.03/04) according to the manufacturer's instructions. An in-vitro-translated c-Myc-tagged lamin C (pGBKT7-Lam, Clontech) was used as a control in co-immunoprecipitation assays.
Lue1 was previously mapped with SSLP markers (Bell and Ecker, 1994) to the bottom of chromosome 1, south of marker nga692 (Meier et al., 2001). New SSLP markers were generated by comparison with Col0 and Ler ecotype DNA sequences available from Cereon Genomics (http://godot.ncgr.org/cereon/). The following primer combinations were used: F18B13-2-F (5′TTAATTATGGTTTCATGATCATGG) and F18B13-2-R (5′CTTTCCTTACACATCTTTCCTGC) from BAC F18B13; F23A5-2-F (5′CTCGAGATCTAGACATGGAGC) and F23A5-2-R (5′GTCTAGGTTCAACAATGCTGC) from BAC F23A5; F9K20-1-F (5′TCCTCCGCTTCCGATTGGTC) and F29K20-1-R (5′GGTACCGTCACGTTCGCCGT) from BAC F29K20; T8K14-1-F (5′CAATGCGCTCTGAATCTCTGAC) and T8K14-1-R (5′CCATTCACCCACTCTTGACTC) from BAC T8K14.
A nonsense mutation in AtKSS is responsible for the lue1 phenotype
We had previously mapped the lue1 mutation to the bottom of chromosome 1, south of SSLP marker nga692 (Meier et al., 2001). To more finely map the mutation, we designed SSLP markers following a sequence comparison between the Col0 and Ler ecotypes from Cereon Genomics (http://godot.ncgr.org/cereon/). This allowed the lue1 mutation to be located south of BAC T8K14. A search for genes putatively involved in cell elongation or GA responses in this region identified three candidates on BAC F516: two encoding At3ox1 homologs, and one encoding AtKSS (McClinton et al., 2001), a katanin p60 ortholog. Since the lue1 phenotype is not rescued by GA application (Meier et al., 2001), GA deficiency owing to loss of function of a At3ox1 homolog was not considered a likely cause of the dwarf stature of lue1 mutants. Moreover, polarizing microscopy of cell walls (below) indicated an abnormal CMF orientation in lue1. Hence, the AtKSS gene was a likely candidate, and it was PCR-amplified and sequenced from lue1. Sequencing identified a single base change in AtKSS producing a nonsense mutation at amino acid 394. Shortly after, the phenotype of the fra2 mutant was shown to be the result of another loss-of-function mutation in the AtKSS gene (Burk et al., 2001). In addition, recombinant AtKSS has recently been shown to posses ATP-dependent, microtubule-severing activity in vitro and is therefore a p60 katanin plant ortholog (Stoppin-Mellet et al., 2002).
To confirm that loss of AtKSS function was responsible for the lue1 mutant phenotype, the AtKSS gene, including 1.7 kb of 5′UTR and 1.3 kb of 3′UTR, was amplified from the Col0 ecotype and mobilized into lue1 via the vector C3300-AtKSS. Fig. 1A shows that the wild-type AtKSS gene rescued the lue1 dwarf phenotype, indicating that the nonsense mutation in AtKSS is responsible for the lue1 phenotype. Phenotypic rescue was observed for all bastamycin-resistant T1 plants. Moreover, herbicide resistance carried on the C3300-AtKSS construct was found to co-segregate with the wild-type phenotype in subsequent generations.
To check whether GA5-LUC reporter overexpression in lue1 (Meier et al., 2001) (Fig. 1B) would be restored to wild-type levels in lue1 plants complemented by the C3300-AtKSS construct, homozygous T3 generation seedlings were grown on MS plates and assayed for LUC activity in vivo. Strong reporter activity was detected in lue1 (Fig. 1C), whereas lue1 plants carrying C3300-AtKSS exhibited markedly reduced LUC levels (Fig. 1D). This indicates that the lue1 mutation in AtKSS is responsible for the GA5-LUC reporter overexpression observed in lue1.
Transcriptional regulation of AtKSS by GA
AtKSS mRNA levels in lue1 and wild-type transgenic (WT) seeds were assayed by RNA blot hybridization. This showed that AtKSS mRNA levels were markedly reduced in lue1 (Fig. 1E), which suggests that the truncated ORF of the lue1 AtKSS allele reduces the stability of the mutant mRNA. Alternatively, AtKSS could be involved in a feed-forward regulation of its own transcription, which would be impaired in lue1. Since lue1 exhibits altered regulation of both the GA5-LUC reporter and endogenous AtGA20ox1 gene (Meier et al., 2001), we investigated the expression of AtKSS upon GA3 treatment in the GA-deficient ga1-1 mutant (Koornneef and van der Veen, 1980). This showed that AtKSS mRNA accumulation levels were lower in ga1-1 than in wild-type Ler and could be restored to wild-type levels by GA3 treatment. These results indicate that GA levels modulate AtKSS mRNA accumulation.
Lue1 exhibits altered cell elongation responses to GA and ET
As lue1 exhibits altered AtGA20ox1 expression levels, the mutant might be affected in its responses to GA. To assess this, GA-related responses, including flowering induction and cell elongation, were compared in GA-treated WT seedlings and lue1 seedlings grown under long-day conditions (Fig. 2). GA treatments caused both lue1 and WT leaves to pale. Flowering could be promoted in lue1 by application of GA3 or GA4, although the effect of the latter was more pronounced (Fig. 2A). Flowering induction by GA was the same in both lue1 and WT (Fig. 2B). Indeed, GA applications reduced FT from 40 to 30 days for both WT and lue1 plants, indicating that the general sensitivity of lue1 to GA was not compromised. However, the mutant exhibited decreased stem elongation in response to GA (data not shown). This apparent insensitivity of lue1 to GA-responsive cell elongation was pronounced in leaves, such that neither blade nor petiole length was affected by GA treatment (Fig. 2C).
Preliminary germination tests revealed that hypocotyl hook formation is impaired in lue1. Since hooking is caused by differential cell elongation, which is regulated, at least in part, by ET, we investigated lue1 sensitivity to ET. WT and lue1 seeds were plated on MS supplemented with 50 μM ACC and allowed to germinate in the dark. As expected, WT seedlings exhibited a typical hook that could be increased by ACC treatment (Fig. 3A,B). By contrast, hypocotyl hook formation was impaired in lue1 control seedlings, whereas ACC treatment did not significantly induce hooking in the mutant (Fig. 3A,B). However, other ET-induced morphological changes were unaffected in lue1, including hypocotyl thickening (Fig. 3C) and hypocotyl and root shortening (Fig. 3D,E).
Lue1 sensitivity to ET was also investigated in seedlings germinated in the light, which has previously been shown to stimulate hypocotyl growth of seedlings grown on low nutrient medium (Smalle et al., 1997). As expected, hypocotyl growth was enhanced in WT seedlings grown in the presence of ACC (Fig. 3F). Similarly, lue1 hypocotyl growth was also induced, although apparently to a lesser extent. Both WT and lue1 root elongations were strongly reduced in the presence of ACC (Fig. 3G). Taken together, these results indicate that although lue1 is generally responsive to ET, the mutant exhibits inappropriate responses leading to cell growth orientation.
The lue1 mutant exhibits abnormal CMF and CMT orientation
Before cloning the AtKSS gene, the general isotropic cell growth observed in lue1 led us to investigate CMF and CMT orientation in the mutant. CMF orientation was measured in WT and lue1 hypocotyls by polarizing microscopy, which is based on the birefringence of CMF (Frey-Wyssling, 1959). Fig. 4A shows an example of polarizing microscopy of WT and lue1 pitted vessel cells. For WT, the minimum and maximum birefringences were obtained for a rotation angle close to 0° and 45°, respectively. This indicated an average transverse orientation of CMF compared to the main growth axis. By contrast, the minimum and maximum birefringent angles in lue1 were approximately 45° and 0°, indicating that CMF have an average orientation approaching 45°. Moreover, the generally lower intensity at maximum birefringence is interpreted as a more random orientation in the mutant cells, since staining of the wall showed no difference in wall thickness. Similar polarizing acquisitions were performed on xylem parenchyma, sclerenchyme and epidermal radial walls. Interestingly, most cell types exhibited deviations of CMF in lue1 compared with WT (Fig. 4B). Moreover, CMF orientation was more variable in lue1 than in WT cells, as clearly show in the standard deviations of measurements for pitted vessels. A Mann-Whitney rank sum test was performed to test whether the difference in median values of CMF orientation in WT and lue1 cells was statistically significant. This test confirmed the more random distribution of CMF in lue1 (P<0.001). These results have recently been confirmed and extended by microscopic analyses showing that the aberrant MT orientation caused by the fra2 mutation in AtKSS results in distorted deposition of cellulose microfibrils (Burk and Ye, 2002).
As it is generally thought that a CMT network orients the CMF cellulose polymers laid down just outside the plasmalemma, the altered CMF orientation in lue1 might reflect disorganized CMT. We therefore compared CMT network organization in lue1 and WT by crossing lue1 with transgenic plants expressing a translational fusion between GFP and the MT-associated protein4 [MP4 (Marc et al., 1998)]. This reporter fusion decorates CMT in Arabidopsis cells without interfering with cytoskeletal organization (Mathur and Chua, 2000). Confocal microscopy revealed a striking difference between WT and lue1 CMT in interphase cells. CMT in WT cells appeared ordered in transverse arrays, whereas the decorated CMT in lue1 exhibited a more random distribution (compare Fig. 4C with D). This result is in agreement with previous observations performed on bot1-5 and fra2 mutants cells (Bichet et al., 2001; Burk et al., 2001). These observations were confirmed for most cell types investigated, including those of the root, hypocotyl and cotyledon. However, no obvious difference between WT and lue1 CMT organization was detected in stomata (compare Fig. 4E with F), although lue1 stomata exhibited a reduced length, comparable to other cell types, when compared to WT. These results suggest that the role of AtKSS in stomatal cell differentiation and development is less pronounced than in other cell types.
The AtKSS protein decorates the CMT
To investigate the subcellular localization of AtKSS and its possible interaction with CMT, we generated transgenic Arabidopsis Co10 plants expressing a translational fusion between AtKSS and the GFP and GUS reporters (AtKSS-G-G; Fig. 5A). Interestingly, all herbicide-resistant transgenic plants expressing detectable levels of the GUS and GFP reporters phenocopied lue1 (Fig. 5B). These plants had the characteristic shorter and thicker organs of lue1, including leaves, flowers and siliques. Leaf trichomes were also mainly two branched and frequently distorted. By contrast, all herbicide-resistant transgenic plants that appeared to be wildtype had very weak GUS activities and no detectable GFP (data not shown). Confocal microscopy performed on intact tissue showed that the AtKSS-G-G protein fusion is targeted to the cytoplasm in epidermis root cells (Fig. 5D,E). No GFP was detected in organelles, vacuoles or nuclei. In root cells close to the root tip, no obvious pattern was observed, and GFP appeared as a blurry signal staining the entire cytoplasm. However, in epidermal root cells that were distant from the tip the GFP signal appeared associated with CMT (Fig. 5F,G). Similar CMT labeling by AtKSS-G-G was found in the hypocotyl (Fig. 5H), in the transition zone between root and hypocotyl (Fig. 5I,J) and in stomata (Fig. 5K). Closer examination revealed that GFP decorated CMT bundles while establishing apparent protein aggregations along its fibers (Fig. 5J). This punctate pattern of CMT labeling was particularly prominent in cells close to the transition zone between the root and hypocotyl.
AtKSS interacts with other MT-related proteins
The connection between AtKSS, CMT organization and GA prompted us to search for proteins that interact with AtKSS. A yeast two-hybrid screen was therefore performed with full-length AtKSS fused to the GAL4-DNA-binding domain (GAL4-BD) as bait. An Arabidopsis cv Co10 cDNA library from mature leaves was screened and resulted in the isolation of some 1000 clones, some of which were further characterized (Fig. 6A). Protein-protein interactions in yeast were confirmed by transforming the PJ69-4A strain (auxotrophic for TRP, LEU, HIS and ADE) with plasmids containing GAL4-BD-AtKSS (pGBT7-AtKSS; TRP marker) and GAL4-activation domain (pGAD10-prey; LEU marker). All transformants grew on SD medium lacking TRP, LEU, ADE and HIS (Fig. 6B, right). Additionally, they exhibited strong β-galactosidase activities (Fig. 6B, center). This indicates that the ADE2 and HIS3 genes, whose transcriptional control is dependent on GAL4-AD and GAL4-BD interaction, are expressed. By contrast, strains carrying empty pGBT7 lacking AtKSS and the pGAD10-prey plasmids exhibited neither auxotrophic growth on medium lacking the nutritional markers (Fig. 6C, right) norβ -galactosidase activities (Fig. 6C, center), confirming the interactions between AtKSS and prey proteins.
A katanin p80 subunit ortholog (Chromosome 1, BAC F11P17) was abundantly represented in the clones obtained from the library screen. Two different clones were obtained encoding the C-terminal domain of this protein, referred to here as KTN-p80.1 (accession number: AAB71474; Fig. 6A). This result is in agreement with previous observations that the C-terminal region of katanin p80 is required for interaction with the katanin p60 subunit in animal cells (Hartman et al., 1998; McNally et al., 2000). Co-immunoprecipitation assays confirmed the interaction between AtKSS and KTN-p80.1 in vitro (Fig. 6D). Database homology searches revealed the presence of three other Arabidopsis KTN-p80.1-related genes that are represented in EST databases. Although some of these predicted proteins exhibit low homology with the central region of KTN-p80.1, all share high similarities within their N-terminal WD40 repeats and within the C-terminal region of KTN-p80.1 and katanin p80 subunits from other organisms (Fig. 6E).
Another putative AtKSS-interacting protein (accession number, CAB89396; referred to here as KSN1) was isolated as two independent yeast clones from the library screen. KSN1 contains an ATP/GTP-binding site motif A (P-loop; residues 223-230) and a kinesin motor domain signature (residues 356-367). KSN1 was previously characterized as a cdc2a-interacting peptide [ accession number: AJ001729 (de Veylder et al., 1997)]. The KSN1 cDNA (clone 1.52) isolated in the two-hybrid screen spans residues 473-867, a region with significant similarities to another Arabidopsis protein represented by an EST (AB011479.1). Co-immunoprecipitation assays confirmed the interaction between AtKSS and the KSN1 peptide encoded by clone 1.52 (Fig. 6D).
It is well documented that external stimuli and hormones control MT dynamics, but less is known about the molecular mechanisms involved in such control. Here we show that the katanin p60 Arabidopsis protein AtKSS plays a role in CMT organization by genetic and cytological analyses of the lue1 mutant (Meier et al., 2001). This approach indicates that aspects of AtKSS function may be mediated by GA, in keeping with previous work on MT organization in GA-deficient mutants and other plants (Wenzel et al., 2000; Baluska et al., 1993; Inada and Shimmen, 2000) and that some form of feedback interaction results in effects on expression of GA biosynthetic genes. For example, complementation of lue1 with the wild-type AtKSS gene rescued the mutant phenotype and restored normal expression levels of the GA5-LUC reporter transgene in the mutant background. Moreover, progeny of a cross between lue1 and botl-5 (Bichet et al., 2001) exhibited both the dwarf stature and high GA5-LUC reporter phenotypes. This indicates that lue1 and bot1 are allelic and confirms that the GA5-LUC reporter is overexpressed in mutants lacking a functional AtKSS gene. These results suggest that MT function and/or AtKSS activity may be involved in feedback modulation of GA biosynthesis via the expression of AtGA20ox1 and perhaps of At3ox1 (Meier et al., 2001). An additional link between MT function and GA is suggested by the increase in AtKSS mRNA levels in the GA-deficient mutant ga1-1 following GA application.
An explanation for these effects is that GA affects a distinct signaling pathway that monitors and modulates cell growth. In this model, GA affects cell elongation that induces AtKSS and/or represses AtGA20ox1 indirectly via another pathway. Alternatively, GA and other growth signaling pathways may crosstalk, in which case shared or interacting components exist. Numerous reports have shown that GA signaling regulates the transcription of target genes and the post-translational control of certain proteins. For example, levels of the Arabidopsis GA signaling repressor RGA, a member of the GRAS protein family thought to act as transcription factors (Richards et al., 2000), are rapidly reduced upon GA application (Silverstone et al., 2001). In addition, SPY encodes an O-linked N-acetylglucosamine transferase (OGT) whose loss-of-function produces a constitutive GA-response mutant phenotype (Thornton et al., 1999). OGT addition of O-linked N-acetylglucosamine may regulate the activity of substrate proteins antagonistically to their modification by phosphorylation (Wells et al., 2001). Possible models of GA action therefore include the direct modification and resultant stabilization of RGA by SPY (Harberd et al., 1998) that is somehow counteracted by GA to derepress the expression of RGA downstream targets, potentially including AtKSS. This model does not, however, explain the effect that loss of AtKSS function has on the increase in expression of the GA biosynthetic genes AtGA20ox1 and At3ox1, whose expression is normally repressed by GA (Meier et al., 2001). This effect suggests that AtKSS and/or other MT-associated proteins indirectly affect GA feedback via regulatory pathways to integrate cytoskeletal organization and cell elongation. Although such pathways remain obscure, they may include signaling pathways related to brassinosteroids that also affect MT organization (Catterou et al., 2001) and that we have shown to increase the expression of AtGA20oxl (Bouquin et al., 2001).
The ectopic expression of a fusion reporter between AtKSS, GFP and GUS in wild-type Co10 plants resulted in a dwarf phenotype similar to that of lue1. This suggests that the AtKSS-G-G fusion functions as a dominant negative form of AtKSS that lacks all or some of its activity. This hypothesis is in agreement with the fact that the AtKSS-G-G reporter fusion failed to rescue the lue1 mutant (data not shown). However, since AtKSS-G-G decorates CMT, this reporter fusion is apparently correctly targeted and may compete with endogenous AtKSS protein for factors important for its function. If both the MT-interacting and katanin p80-interacting domains are functional in AtKSS-G-G, then the C-terminal ATPase domain required for MT-severing activity (Hartman and Vale, 1999) may not be fully functional in the fusion. Such a model is consistent with the fact that both the lue1 and fra2 mutations occur in the ATPase domain of AtKSS. In addition, we show that AtKSS mRNA levels are detectable in lue1, although at significantly lower levels than in WT. This suggests that the lue1 phenotype is primarily the result of an inactive katanin p60 rather than a lack of the protein. We note that the fusion between the two reporters and AtKSS was designed in the AtKSS C-terminus to avoid disturbing its N-terminal, MT-interacting domain or potential post-translational processing including glycosylation.
In all our observations, the AtKSS-G-G reporter was confined to the cytoplasm. More specifically, the protein was observed as a blurry signal in epidermal cells close to the root tip, whereas distinct CMT labeling patterns were detected in other cells such as in the transition zone between the root and hypocotyl, the hypocotyl and stomata. We also observed phragmoplast labeling by AtKSS-G-G, although the signal intensity appeared weaker and more random than in interphase cells (data not shown). Since mitosis and cytokinesis were apparently unaffected in the allelic bot1-5 and fra2 mutants (Bichet et al., 2001; Burk et al., 2001), these results indicate that the role of AtKSS in modulating MT dynamics is less marked during mitosis than during interphase. This may be due to a difference in protein targeting, although we cannot exclude transcriptional control of AtKSS. Recently the Caenorhabditis elegans Nedd8 ubiquitin-like protein modification pathway that regulates cell cycle progression was shown to negatively regulate katanin, thus allowing the formation of the mitotic spindle (Kurz et al., 2002). It is therefore possible that the Arabidopsis katanin is similarly targeted for ubiquitin-mediated degradation when assembly of the mitotic spindle is required.
In interphase cells, CMT labeling by the AtKSS-G-G reporter frequently appeared as a punctate pattern, suggesting that the reporter fusion aggregates along the CMT. Interestingly, C. elegans katanin p60 forms hexameric rings around MT (Hartman et al., 1998; Hartman and Vale, 1999). It is therefore likely that the AtKSS-G-G structures observed along the CMT correspond to aggregations of katanin rings. This would imply that the AtKSS-G-G ATPase domain binds ATP because katanin p60 oligomerization is an ATP-dependent process (Hartman and Vale, 1999). However, the presence of intact but mis-oriented CMT in plants expressing AtKSS-G-G suggests that this fusion may have reduced or no ATPase activity required for MT severing. For example, overexpression of human katanin p60 in HeLa cells results in disassembly of the interphase MT cytoskeleton (McNally et al., 2000), which is clearly not the case in plant cells that overexpress AtKSS-G-G. The importance of ATPase activity in vivo could be addressed with reporters based on GFP fused to the AtKSS N-terminus or on AtKSS forms mutated to block nucleotide hydrolysis and trap the enzyme in the ATP-bound state (Hartman and Vale, 1999).
Preliminary time-course observations of GFP-MAP4 reporter fluorescence indicated that wild-type CMT undergo rapid shrinkage and reorientation within minutes of transfer of seedlings from dark to light (data not shown). By contrast, modifications of the CMT network appeared to be slower in lue1. It is therefore probable that the abnormal CMT organization reflects the inability of lue1 CMT to respond rapidly to stimuli normally responsible for CMT reorientation and anisotropic cell growth. This would lead to the abnormal deposition of CMF in lue1 and contribute to the general organ fragility observed in fra2, bot1-5 and lue1, as well as their apparent insensitivity to hormone-mediated cell elongation.
A yeast two-hybrid screen with AtKSS as bait identified a katanin p80 subunit ortholog (KTN-p80.1) as an AtKSS interaction partner. This result confirms that AtKSS has a katanin-like function. The two KTN-p80.1 clones isolated (1.38 and 1.25) encode the C-terminus and implicate this region in the interaction with the katanin p60. More specifically, clone 1.25 encodes the last 222 residues of KTN-p80.1, including 101 amino acids conserved among katanin p80-like proteins. This domain is therefore sufficient to establish interactions between katanin heterodimers, as shown in directed yeast two-hybrid and co-immunoprecipitation assays using a truncated version (1.25B2) of clone 1.25. Sequence homology searches identified three other KTN-p80.1-related Arabidopsis genes with strong homology to other katanin p80s, particularly in the N-terminal WD40 repeats and in the C-terminal katanin-p60-interacting region. Since they are all represented in EST databases, it is surprising that only KTN-p80.1 was isolated in the two-hybrid screen. It may be that the other KTN-p80-encoding clones were under-represented in the cDNA library, and we therefore cannot exclude the possibility that AtKSS interacts with other KTN-p80.1-like proteins in planta. In animal cells, the katanin p80 subunit targets the katanin complex to centrosomes and regulates the MT-severing activity of the p60 subunit (Hartman et al., 1998; McNally et al., 2000). Database searches indicated that Arabidopsis has only one copy of AtKSS. However, several homologous proteins harboring the AAA ATPase domain, but lacking the N-terminal region that contains the MT- and katanin-p80-interacting domains, are present in Arabidopsis. Moreover, the genetic isolation of null mutant alleles of AtKSS suggests that there are no other genes with completely redundant functions. Therefore, the involvement of AtKSS in multiple and distinct MT-related activities may require tight control of its subcellular targeting. The fact that all katanin-P80-related proteins from Arabidopsis are highly similar within the WD40 domains and katanin-p60-interacting C-terminal regions, but otherwise exhibit low homologies in the central regions, suggests that the central region may be involved in differential targeting of the katanin heterodimers. For example, KTN p80.1 could target AtKSS to CMT, whereas another p80 form could target it to the phragmoplasm. This would explain why the other katanin p80s may be under-represented in the yeast two-hybrid library because there are less cells undergoing division than cells in interphase. GUS reporter fusions using various katanin p80 promoters may address this question. However, the fact that loss-of-function alleles of AtKSS are apparently not impaired in cytokinesis and mitosis indicates that this katanin p60 subunit plays a minor role in these cellular events. Katanin activity analogous to that required in the spindle pole of animal cells undergoing cell division may therefore be supplied by other MT-severing proteins in plant cells.
Another AtKSS-interacting protein identified here (KSN1) harbors an ATP/GTP-binding site motif A (P-loop) and a kinesin motor domain signature, although these regions were apparently not involved in the interaction with AtKSS. Kinesin motor enzymes hydrolyze ATP to generate force and movement along MT. Numerous studies have shown a role for kinesins in MT-associated activities such as vesicle transport along MT, mitosis and meiosis. Homology searches indicated that KSN1 is member of a small family that includes another protein exhibiting 77% identity with KSN1 (AB011479.1). An A-type cyclin-dependent kinase (CDK) designated Cdc2 was previously shown to interact with a KSN1 peptide (de Veylder et al., 1997). Interestingly, Cdc2 cosedimented with taxol-stabilized MT (Weingartner et al., 2001). Additionally, a functional Cdc2-GFP fusion decorated the anaphase spindle and phragmoplast. Subcellular localization of the Cdc2-GFP fusion was shown to be cell-cycle-dependent and tightly associated with the nucleus during interphase. Taken together with our inability to detect the AtKSS-G-G fusion in the nucleus, the association of AtKSS with KSN1 and Cdc2, if any, is likely to take place during mitosis. Alternatively, AtKSS and Cdc2a could compete for recruitment of KSN1, although we found no significant sequence homologies between AtKSS and Cdc2 to indicate conserved binding regions. Interestingly, immunological experiments with antibodies targeted either against the C-terminal region of Cdc2 or the PSTAIRE motif in the cyclin-binding domain found in A-type CDKs showed that Cdc2 was associated with CMT from various plants (Hemsley et al., 2001). This suggests that a multimeric protein complex including katanin heterodimers KSN1 and Cdc2a could be involved in CMT processing.
The growing number of MT-associated proteins (MAP) and MT regulatory proteins that are being characterized in plants is rapidly increasing our understanding of MT genesis and dynamics. Among these MAPs, AtKSS seems to play a central role, especially in the integration of hormonal signals that lead to anisotropic cell growth by severing CMT, allowing reorientation of MT growth and thus CMF deposition. Further work is required to elucidate the molecular mechanisms that lead to the transcriptional regulation of GA biosynthetic genes and the exact involvement of AtKSS in mitosis and cytokinesis.
We are grateful to Janni Kristensen for technical assistance. This research was funded by the European Union (grant number CT96-0062 and CT96-0621 to J.M.). T.B. was funded by a postdoctoral Marie Curie Research training grant.
- Accepted November 15, 2002.
- © The Company of Biologists Limited 2003