The Caenorhabditis elegans Microtubule-severing Complex MEI-1/MEI-2 Katanin Interacts Differently with Two Superficially Redundant β-Tubulin Isotypes

Abstract

The microtubule-severing protein complex katanin is required for a variety of important microtubule-base morphological changes in both animals and plants. Caenorhabditis elegans katanin is encoded by the mei-1 and mei-2 genes and is required for oocyte meiotic spindle formation and must be inactivated before the first mitotic cleavage. We identified a mutation, sb26, in the tbb-2 β-tubulin gene that partially inhibits MEI-1/MEI-2 activity: sb26 rescues lethality caused by ectopic MEI-1/MEI-2 expression during mitosis, and sb26 increases meiotic defects in a genetic background where MEI-1/MEI-2 activity is lower than normal. sb26 does not interfere with MEI-1/MEI-2 microtubule localization, suggesting that this mutation likely interferes with severing. Tubulin deletion alleles and RNA-mediated interference revealed that TBB-2 and the other germline enriched β-tubulin isotype, TBB-1, are redundant for embryonic viability. However, limiting MEI-1/MEI-2 activity in these experiments revealed that MEI-1/MEI-2 preferentially interacts with TBB-2–containing microtubules. Our results demonstrate that these two superficially redundant β-tubulin isotypes have functionally distinct roles in vivo.

INTRODUCTION

During cell division, precise segregation of chromosomes into the daughter cells relies on the dynamic organization of a bipolar spindle. With spindle microtubules nucleating from the centrosomes, a bipolar mitotic spindle forms as microtubules grow out and capture the chromosomes. Compared with mitosis of animal cells, female meiosis in most animals has a distinct mechanism to achieve the spindle bipolarity (reviewed in Schatten, 1994; Merdes and Cleveland, 1997). The oocyte lacks centrosomes, so spindle microtubules are nucleated around the meiotic chromatin and are then bundled into antiparallel arrays by multimeric plus end-directed motors such as Eg5 (Sharp et al., 1999). These microtubule arrays are driven with their minus ends out by plus end-directed microtubule motors, such as chromokinesin (Wang and Adler, 1995), Xklp 1 (Vernos et al., 1995), and Drosophila Nod (Afshar et al., 1995). Minus end-directed motors, such as cytoplasmic dynein or Drosophila Ncd then tether the spindle ends into focused poles (Gaglio et al., 1996; Matthies et al., 1996). Based on the morphological differences between the two spindle structures, microtubule dynamics in meiosis are likely also differentially regulated in accordance to the forces generated by motor proteins.

C. elegans female meiosis takes place in the fertilized zygote. Therefore, the zygote cytoplasm must support the formation of both the meiotic and the first mitotic spindle, which form within 20 min of one another (Kemphues et al., 1986). This along with other convenient genetic and molecular tools makes C. elegans an ideal system (Brenner, 1974) to study how these different types of spindles form. Previously, we showed that mei-1 and mei-2 encode two meiotic spindle-specific components. MEI-1 and MEI-2 are the C. elegans homologs of the p60 (catalytic) and p80 (localization) subunits of the sea urchin microtubule-severing complex katanin (Clark-Maguire and Mains, 1994a,b; Hartman et al., 1998; Srayko et al., 2000). Because MEI-1 and MEI-2, like sea urchin p60 and p80, disassembled interphase microtubules when coexpressed in HeLa cells, we proposed that they specifically regulate spindle microtubule dynamics and/or restrict microtubule length in meiosis, and thus are required for the formation of a bipolar spindle in the absence of centrosomes (Srayko et al., 2000). In mei-1 and mei-2 loss-of-function (lf) mutants, meiotic spindles fail to form properly but subsequent mitotic spindles are not affected (Mains et al., 1990a; Clandinin and Mains, 1993; Srayko et al., 2000). However, ectopic MEI-1/MEI-2 activity in mitosis, resulting either from the mei-1(ct46) gain-of-function (gf) mutation, or loss of the mei-1/mei-2 postmeiotic inhibitor mel-26, instead disrupt mitotic spindle structure (Mains et al., 1990a; Clark-Maguire and Mains, 1994a,b; Dow and Mains, 1998; Kurz et al., 2002). This ectopic microtubule-severing activity results in smaller, mispositioned mitotic spindles. Consistent with reduced microtubule length caused by ectopic microtubule-severing activity, mei-1(ct46gf) is phenocopied by low doses of the microtubule-destabilizing drug nocodazole (Strome and Wood, 1983; Hyman and White, 1987).

To better understand the role of MEI-1/MEI-2 katanin in spindle formation, we performed a screen for suppressors that rescue the lethality of ectopic katanin activity caused by the mei-1(ct46gf) mutation (Clandinin and Mains, 1993). Here, we describe the analysis of an extragenic suppressor, sb26, which is a missense allele of the β-tubulin gene tbb-2. tbb-2(sb26) genetically behaves as if it produces meiotic and mitotic microtubules that are resistant to katanin severing. Immunofluorescence with TBB-2–specific antibodies shows ubiquitous TBB-2 expression in microtubule structures throughout worm development. Furthermore, tubulin iso-type-specific RNA interference (RNAi) experiments demonstrate a redundant role for tbb-2 with the closely related gene tbb-1 during the early cleavage divisions. Finally, using a sensitized genetic background, we demonstrate that microtubules containing the TBB-2 β-tubulin isotype are preferred for MEI-1/MEI-2 activity.

MATERIALS AND METHODS

Nematode Strains and Culture Conditions

C. elegans was cultured under standard conditions (Brenner, 1974) and brood analysis was done as described by Mains et al. (1990b). Hatching rates were scored among 500-2000 embryos. The following genes and alleles were used: mei-2(ct98), mei-1(ct46), unc-29(e1072), tbb-2(sb26, gk129, gk130), tbb-1(gk207). tbb-2 deletion alleles gk129 and gk130 and tbb-1 deletion allele gk207 were provided by the C. elegans Gene Knockout Consortium (elegans.bcgsc.bc.ca/knockout.shtml). Both tbb-2 deletion alleles are predicted molecular nulls, removing some promoter region and most of exon I, including the start ATG without generating any downstream in-frame ATG. Both alleles were out-crossed at least three times by selecting for the healthiest strains that did not stain with anti-TBB-2 antibody. The tbb-1 allele gk207 is also a predicted molecular null, removing most of gene's promoter and part of its first exon. The strain was outcrossed six times and genotype was verified by polymerase chain reaction (PCR).

Genetic Mapping and Cloning of tbb-2(sb26)

Three-factor crosses placed sb26 between two cloned markers, pat-3 and mpk-1 on LGIII (data submitted to WormBase, www.wormbase.org), a region covered by 10 overlapping cosmids. The genomic sequence of tbb-2(sb26) was PCR amplified from sb26 homozygous mutants and sequenced from two independent PCR products as described by Srayko et al. (2000).

Antisera Production and Immunoblotting

PCR-amplified sequence corresponding to the last 20 amino acids of TBB-2 (residues 431–450) was inserted into the BamHI site of pGEX-3 × (Amersham Pharmacia, Uppsala, Sweden) to create a GST-fusion. Bacterially expressed protein was purified with a glutathione-Sephadex column (Pharmacia) and run on an SDS-PAGE gel. Immunization of rabbits was performed following Srayko et al. (2000). Crude sera were affinity purified against a column of a synthetic peptide (Alberta Peptide Institute, Edmonton, AB, Canada) corresponding to the amino acids 432–442 of TBB-2 (see Figure 3A). Western analyses was performed as described by Srayko et al. (2000), by using affinity purified TBB-2 antisera at 1/1000 dilution.

Figure 3.

TBB-2 specific antisera production. (A) Alignment of the C termini of TBB-2 and four other C. elegans β-tubulins. Black shading indicates identity to TBB-2. The box above shows the region used to generate antisera, whereas the shaded region of that box indicates the peptide used for antisera affinity purification. The sb26 mutation is indicated. (B) Immunoblots of wild-type, tbb-2 deletion alleles and germline absent (glp-1) worms with affinity purified anti-TBB-2 antibodies, demonstrating antibody specificity. Twenty adult hermaphrodites of each genotype were lysed and loaded per lane. Lane 1, wild type; lanes 2 and 3: tbb-2 deletion alleles gk129 and gk130, respectively; lane 4, glp-1. The arrow indicates the TBB-2 band.

Microscopy and Immunofluorescence

Embryos and dissected gonads were freeze-cracked and fixed with methanolacetone as described by Kemphues et al. (1986). Affinity-purified anti-TBB-2 antisera were used at 1/100 dilution for 1 h at 37°C. α-Tubulin localization was determined with either a mouse monoclonal antibody (Piperno and Fuller, 1985) at 1/100 dilution or the mouse DM 1A α-tubulin monoclonal antibody (Sigma-Aldrich, St. Louis, MO) at 1/200 dilution. Secondary antibodies (Jackson ImmunoResearch Laboratories, West Grove, PA) were fluorescein isothiocyanate-conjugated goat anti-rabbit, tetramethylrhodamine B isothiocyanate-conjugated goat anti-mouse, or Texas Red-conjugated goat anti-mouse. DNA was visualized with 4,6-diamidino-2-phenylindole (DAPI) as described in Srayko et al. (2000). Photographs were either taken with an Axioplan fluorescence microscope camera (Carl Zeiss, Thornwood, NY) with Techpan film (Eastman Kodak, Rochester, NY) developed at ASA100, or with a Hamamatsu ORCA-ER digital camera and deconvolved using Zeiss AxioVision software.

RNAi

tbb-1 and tbb-2 cDNA clones yk51f5 and yk84c10, respectively, were obtained from Dr. Yuji Kohara (University of Tokyo, Tokyo, Japan). Primers were designed to amplify the divergent C-terminal coding region and 3′-untranslated regions (UTRs) from each cDNA. Primers used for tbb-2 were forward: 5′-AAGCTTCACTTCACTCAGAT-3′ and reverse: 5′-GAAGCAACTGCCGAAGACGAC-3′. Primers used for tbb-1 were forward: 5′-AAGGTACCGCCGAAGACGAGCC-3′ and reverse: 5′-AATACGACTCACTATAG-3′ (T7). PCR fragments were cloned into pBluescript SK(+) (Stratagene, La Jolla, CA). RNA was transcribed using the Megascript system (Ambion, Austin, TX), and DNA templates were digested with DNase I. The resulting RNAs were purified and annealed as described by Fire (ftp.ciwemb.edu/PNF:byName:/FireLabWeb/) and 150 μg/ml was used for injection. The resulting embryos were dissected for fixation 24 h after injection.

RESULTS

tbb-2(sb26) Prevents Lethality Caused by Ectopic MEI-1/MEI-2 Expression

The mei-1(ct46gf) mutant shows a dominant, temperature-sensitive maternal-effect lethality due to mitotic spindle defects caused by ectopic MEI-1/MEI-2 severing activity (Mains et al., 1990a; Srayko et al., 2000). In a previous screen to identify interacting genes (Clandinin and Mains, 1993), we isolated an extragenic mei-1(ct46gf) suppressor, sb26. As shown below, sb26 is an allele of the β-tubulin gene tbb-2. Homozygous tbb-2(sb26) single mutant embryos develop the same as wild type, giving rise to fertile adults with wild-type brood sizes and hatching. However, although tbb-2(sb26) has no obvious mitotic phenotypes (Figure 1, A and B), it effectively suppresses (alleviates) the lethality caused by ectopic MEI-1/MEI-2 activity resulting from the mei-1(ct46gf) mutation (Table 1, lines 1–5). In mei-1(ct46gf) embryos, the first mitotic spindle is smaller than wild-type and is often mispositioned orthogonally to the anterior-posterior axis (Figure 1C). In contrast, when mei-1(ct46gf) is combined with the tbb-2(sb26) mutation, embryos have wild-type spindle morphology and orientation (Figure 1D). This result suggests that tbb-2(sb26) somehow interferes with the ectopic MEI-1/MEI-2 activity associated with the mei-1(ct46gf) mutation. To test whether tbb-2(sb26) also interferes with normal MEI-1/MEI-2 activity during meiosis, we looked at mutations that cause partial loss of mei-1and mei-2. On their own, mei-1 and mei-2 lf mutations result in defects in meiotic spindle formation, most likely due to decreased meiotic microtubule severing, and these defects were enhanced when combined with tbb-2(sb26) (Table 1, lines 10–11). Whereas the meiotic defect of a weak mei-2(lf) allele, ct98, results in 86% hatching at the nonpermissive temperature of 25°, lethality increased dramatically when combined with tbb-2(sb26), resulting in a 3.7% hatching under the same conditions. In C. elegans, nondisjunction of the X chromosome results in a normal XO male, and thus the frequency of males among the survivors is a measure of meiotic failure. Normally, ∼1/500 zygotes from a selfed hermaphrodite is a male (Hodgkin and Brenner, 1977), but among the mei-2(ct98); tbb-2(sb26) escapers at 25°, 17% were male (Table 1, line 11), likely indicating a failure in meiotic spindle function. Therefore, in either wild-type meiosis or abnormal mitosis as a result of ectopic MEI-1/MEI-2 expression, we find that tbb-2(+) is required for mei-1 and mei-2 function, suggesting that tbb-2(sb26) lowers the overall sensitivity of microtubules to MEI-1/MEI-2 activity.

Figure 1.

sb26 prevents the mitotic spindle defect caused by ectopic MEI-1/MEI-2 expression. Shown are indirect immunofluorescence images of anti-α-tubulin staining of the first mitotic spindles of wild-type (A), sb26 (B), mei-1(ct46), which causes ectopic MEI-1/MEI-2 expression (C) and mei1(ct46gf); tbb-2(sb26) double mutants (D). Embryos here and in all other figures are shown with anterior to the left. Bar, 10 μm.

Table 1.

Interaction of tbb-2 with mei-1 and mei-2 alleles

	% hatching (% malea)
Maternal genotype	15°	20°	25°
mei-1(ct46gf)b	23 (0)	1.5	0
tbb-2(sb26)	100 (0)	—c	97 (0.1)
mei-1(ct46gf); tbb-2(sb26)	96 (0.1)	92 (0.9)	67 (1.4)
mei-1(ct46gf)/+	—	38	1.2
mei-1(ct46gf)/+; tbb-2(sb26)/+	—	84	37
tbb-2(t1623)/+	—	96	74
mei-1(ct46gf)/+; tbb-2(t1623)/+	—	16	0.6
tbb-2(or362)/+	—	98	95
mei-1(ct46gf)/+; tbb-2(or362)/+	—	25	0
mei-2(ct98)d	96 (0)	97 (0.2)	74 (2.6)
mei-2(ct98); tbb-2(sb26)	89 (0.4)	64 (4.7)	3.7 (17)

^aPercentage of males among hatched embryos that survived to late larval stages, which is a measure of meiotic failure.

^bmei-1(ct46gf) is a semidominant temperature-sensitive maternal-effect lethal mutation that results in relatively normal meiotic MEI-1 function, but ectopic activity persists into mitosis.

^cNot determined.

^dmei-2(ct98) decreases but does not eliminate mei-2 activity.

The sb26 Mutation Alters Rather than Eliminates Gene Function

Several lines of evidence indicate that the tbb-2 allele sb26 represents an altered rather than a loss of gene function. A chromosomal deficiency that removes the tbb-2 locus, sDf130, failed to dominantly suppress the maternal-effect lethality of mei-1(ct46gf), although tbb-2(sb26)/+ did so (Table 2, lines 1–3). We also acquired two tbb-2 deletion alleles, both predicted to be molecular nulls, from the C. elegans Gene Knockout Consortium (see MATERIALS AND METHODS). These produced no detectable protein on immunoblots with TBB-2–specific antibodies (see below). In contrast to tbb-2(sb26), heterozygosity for a tbb-2 null allele decreased hatching when ectopic MEI-1/MEI-2 was present in mitosis (Table 2, line 4). Likewise, when the wild-type tbb-2 allele in mei-1(ct46gf)/+; tbb-2(sb26)/+ was removed and replaced with the molecular null gk129, suppression of the mei-1(ct46gf) embryonic lethality improved, although not to the extent of the complete suppression seen with sb26/sb26 (Table 2, compare lines 2, 5, and 6). This indicates that the wild-type tbb-2 allele interferes with the suppression by the sb26 allele. Finally, mei-1(ct46) lethality caused by the ectopic MEI-1/MEI-2 expression was rescued by transformation with tbb-2(sb26) but not with tbb-2(+) (see below). Together, these genetic studies clearly indicate that tbb-2(sb26) is a gf suppressor of mei-1(ct46gf).

Table 2.

Suppression of mei-1(ct46gf) by tbb-2(sb26) results from a gain-of-function mutation

Maternal genotype	% hatching (20°)
mei-1(ct46gf)/+	39
mei-1(ct46gf)/+; tbb-2(sb26)/+	75
mei-1(ct46gf)/+; sDf130/+	7.6a
mei-1(ct46gf)/+; tbb-2(gk129)b/+	12
mei-1(ct46gf)/+; tbb-2(sb26)	99
mei-1(ct46gf)/+; tbb-2(sb26)/tbb-2(gk129)	89

^aCorrected for the 43% zygotic lethality resulting from the deficiency.

^btbb-2(gk129) is a predicted molecular null allele. Another null allele, gk130, behaved similarly.

Other tbb-2 alleles that behave as gain-of-function mutations with mitotic spindle defects similar to mei-1(gf) are described by other groups (Ellis et al., 2004; Wright and Hunter, 2003). Neither tbb-2(t1623) nor tbb-2(or362), two missense tbb-2 alleles of such kind, suppresses mei-1(ct46gf), but rather acted as dominant enhancers (Table 1, lines 6–9). This likely results from the fact that these tbb-2 alleles and mei-1(ct46gf) both destabilize mitotic microtubules. Thus, tbb-2 interacts with mei-1 and mei-2 in an allele-specific manner.

tbb-2(sb26) Does Not Prevent Ectopic MEI-1/MEI-2 Localization

MEI-1 and MEI-2 localize exclusively to wild-type female meiotic spindles; both proteins are found throughout the spindle but are more concentrated at the poles and on chromatin (Clark-Maguire and Mains, 1994a; Srayko et al., 2000). In mutants resulting in ectopic MEI-1 and MEI-2 expression, both MEI-1 and MEI-2 localize to the analogous regions during mitosis (Figure 2, A and C), disrupting mitotic spindle function (Clark-Maguire and Mains, 1994a; Srayko et al., 2000). We asked whether the suppression of mitotic spindle defects by tbb-2(sb26) was due to absent (or reduced) MEI-1 and/or MEI-2 in the mitotic apparatus. Immunolocalization using anti-MEI-1 or MEI-2 showed that the gene products still persisted in the double mutants (Figure 2, B and D). Therefore, sb26 suppresses mei-1(ct46gf) despite the presence of ectopic MEI-1/MEI-2 protein. This suggests that sb26 interferes with MEI-1/MEI-2 function via a mechanism other than blocking localization or causing degradation of the complex and instead likely acts by inhibiting microtubule severing (see DISCUSSION).

Figure 2.

Mislocalization of MEI-1 and MEI-2 to mitotic centrosomes and chromosomes in mei-1(ct46); tbb-2(sb26) double mutant embryos. Indirect immunofluorescence images of anti-MEI-1 staining in mei-1(ct46) shows ectopic MEI-1 expression (A) and mei-1(ct46); tbb-2(sb26), which continues to show ectopic MEI-1 staining. Similar results were seen with anti-MEI-2 staining in mei-1(ct46) (C) and mei-1(ct46); tbb-2(sb26) (D). Both MEI-1 and MEI-2 localize to mitotic spindle microtubules, centrosomes (open arrowheads), and chromosomes (white arrowheads). The photograph in D was digitally deconvolved. Bar, 10 μm.

sb26 Has a Missense Mutation in the β-Tubulin Gene tbb-2

sb26 was genetically mapped (see MATERIALS AND METHODS) to a region containing the β-tubulin gene tbb-2 (C36E8.5) (Gremke, 1986). Sequencing of tbb-2 in sb26 mutants revealed a single G-to-A transition at the 3′ end of the coding region, resulting in a Glu to a Lys change (Figure 3A). Reverse transcription-PCR showed that a SL1 leader sequence (Krause and Hirsh, 1987) was trans-spliced onto the 5′ end of the tbb-2 transcript; sequencing confirmed the predicted gene structure in WormBase (www.wormbase.org). To eliminate the possibility that the mutation we found in tbb-2 was an unrelated ethylmethansulfonate-induced lesion very tightly linked to sb26, we transformed worms with tbb-2(+) and tbb-2(sb26) genomic constructs. One transgenic line of tbb-2(+) and three independent lines of tbb-2(sb26) were obtained. Wild-type worms carrying extrachromosomal arrays of either tbb-2(+) (sbEx156) or the mutant tbb-2(sb26) (e.g., sbEx145) showed essentially wild-type hatching rates (Table 3, lines 2 and 3), although both showed an incompletely penetrant protruding vulva phenotype. As expected from the genetic nature of tbb-2(sb26), only transgenic arrays made from sb26 mutant genomic DNA rescued the embryonic lethality of mei-1(ct46gf), with hatching rates increasing nearly 10-fold for tbb-2(sb26) compared with the control at 20° (Table 3, lines 1, 4, and 5). In addition, two independent TBB-2::GFP expressing lines failed to rescue mei-1(ct46gf) (our unpublished data). These results confirm that sb26 alters rather than eliminates gene function and that the tbb-2 mutation corresponds to sb26.

Table 3.

Suppression of mei-1(ct46gf) by tbb-2 extrachromosomal transgenic arrays

	% hatching
Maternal genotype	15°	20°
mei-1(ct46gf)	16	1.3
+; sbEx156[tbb-2(+)]	92	98
+; sbEx145[tbb-2(sb26)]	100	99
mei-1(ct46gf); sbEx156[tbb-2(+)]	16	2.2
mei-1(ct46gf); sbEx145[tbb-2(sb26)]	29	11

TBB-2 Is Widely Expressed throughout the Life Cycle

Because tbb-2(sb26) suppresses ectopic mei-1 activity during early embryogenesis, we expected tbb-2 to be a component of the meiotic and early mitotic spindles, where it perhaps facilitates MEI-1/MEI-2 microtubule severing. To explore the expression pattern of TBB-2, we generated specific polyclonal antisera directed to a nonconserved region of 11 amino acids in the C terminus (Figure 3A; see MATERIALS AND METHODS). Affinity-purified antisera recognized a single band at the expected size of 50 kDa on Western blots, and this band disappeared when blotting either of the two tbb-2 deletion strains (Figure 3B, lanes 1–3), demonstrating antisera specificity.

We predicted that TBB-2 is maternally provided and should be present during early embryogenesis. Female gonads were dissected from gravid hermaphrodites, fixed, and stained with TBB-2–specific antisera and costained with a generic monoclonal anti-α-tubulin antibody to visualize microtubule structures. TBB-2 was expressed throughout the female gonad, from the distal syncytial arm to maturing oocytes (Figure 4, A–C). We next examined TBB-2 expression in meiotic and early mitotic spindles, the developmental stage at which the genetic interactions take place. Figure 4, D–F, shows TBB-2 localization in a meiosis metaphase I spindle in a pattern identical to that of the α-tubulin staining. Digital deconvolution microscopy was used to examine in detail TBB-2 expression in mitotic spindles. As shown in Figure 4, G–I, TBB-2 was expressed in every subset of spindle microtubules recognized by the generic anti-α-tubulin antibody, including spindle midzone and astral microtubules.

Figure 4.

TBB-2 expression pattern. Dissected gonads and embryos were visualized by indirect immunofluorescence after staining DNA with DAPI (A and D), anti-α-tubulin (B, E, and H) and anti-TBB-2 (C, F, and I). TBB-2 was strongly expressed in the germline, starting from distal syncytial gonad arms (open arrowhead in C) and was also seen in maturing oocytes (white arrowhead) and the meiotic spindle (F). TBB-2 is present in all mitotic spindle microtubules recognized by the α-tubulin in the first cleavage spindle, as demonstrated by the digitally deconvolved merged image (G; blue: DAPI). Postembryonic TBB-2 expression was examined by anti-TBB-2 staining (K) and compared with anti-α-tubulin staining (J). TBB-2 is strongly expressed in adult neurons (K). Bars, 10 μm (C, I and K); 2 μm (F).

TBB-2 expression persisted in mitotic spindles and asters throughout embryogenesis (our unpublished data), beyond the end of the temperature-sensitive period of mei-1(ct46gf) (Mains et al., 1990a). Indeed, TBB-2 expression was not germline and embryo specific, because the protein was also expressed (albeit at lower levels) in a glp-1 mutant, which lack germline and fertilized embryos (Figure 3B, lane 4). Finally, TBB-2 was detected postembryonically in larval and adult neuronal tissues (Figure 4K). TBB-2 is therefore widely expressed throughout worm development. TBB-2(sb26) showed the same expression pattern as wild-type TBB-2.

tbb-2 and tbb-1 Are Partially Redundant during Early Development

Because sb26 results in an altered rather than a loss of gene activity, we used RNAi (Fire et al., 1998) to deplete its expression during embryonic development to better determine the gene's normal function. However, TBB-2 and the other β-tubulin that is expressed at high levels in the germline, TBB-1, share >85% DNA sequence identity within their coding regions. Therefore, full-length RNAi to either gene results in simultaneous inhibition of both gene's function (our unpublished data). Because sequence similarities among the 3′ coding and 3′-UTRs of these worm β-tubulins are <50%, we examined the effect of silencing tbb-2 by using dsRNA targeted to the last 210 base pairs of the transcript. This reduced TBB-2 expression to nearly undetectable levels by immunofluorescence (Figure 5F). However, 86% of tbb-2(RNAi) embryos hatched and grew to fertile adults (Table 4, line 1), similar to the level of lethality seen with either of the tbb-2 deletion strains (Table 4, line 2). These data suggest that other β-tubulins can compensate for loss of tbb-2.

Figure 5.

tbb-1 and tbb-2 are redundant during embryogenesis. Hermaphrodites were injected with gene-specific double-stranded RNA and were stained with DAPI (A, D, and G), anti-α-tubulin (B, E, and H) and anti-TBB-2 (C, F, and I) and visualized by indirect immunofluorescence. TBB-2 expression was reduced to near undetectable levels with tbb-2(RNAi) (F). However, microtubules as visualized with anti-α-tubulin staining were unaffected (E) compared with uninjected control (B). In embryos simultaneously treated with tbb-1 and tbb-2 dsRNA (G–I), no microtubules were detected by anti-α-tubulin antibody (H). Embryos arrested as one cell, with massive amounts of DNA (G). Bar, 10 μm.

Table 4.

TBB-1 and TBB-2 are redundant for embryonic development but are not equally preferred for katanin activity

Maternal genotype	% hatching at 25°
tbb-2(RNAi)	86a
tbb-2(gk129)b	75
tbb-1(RNAi)	92a
tbb-1(gk207)c	88
tbb-2(RNAi); tbb-1(RNAi)	0a
tbb-2(gk129); tbb-1(RNAi)	0.5
tbb-2(sb26); tbb-1(RNAi)	85
mei-2(ct98)d	74
mei-2(ct98); tbb-2(RNAi)	1.7
mei-2(ct98); tbb-2(gk129)	5.8
mei-2(ct98); tbb-1(RNAi)	71
mei-2(ct98); tbb-1(gk207)	60

^aPercentage of hatching at 20°.

^bgk129 is a predicted molecular null allele of tbb-2.

^cgk207 is a predicted molecular null allele of tbb-1.

^dmei-2(ct98) decreases but does not eliminate mei-2 activity.

Because tbb-1 is the only other β-tubulin expressed in early embryos at a significant level (Reinke et al., 2000; Baugh et al., 2003), we used RNAi to determine whether tbb-1 functions redundantly with tbb-2. Like tbb-2(RNAi), injection of dsRNA directed to the divergent 3′ coding and 3′-UTR of tbb-1 showed little embryonic lethality (Table 4, line 3), also similar to the level of lethality associated with a tbb-1 deletion strain, gk207 (Table 4, line 4). However, RNAi to both tbb-1 and tbb-2 resulted in 100% embryonic lethality and completely abolished spindle formation, and all embryos arrested without division (Table 4, line 5; and Figure 5, G–I). Oocyte meiosis also failed because no polar bodies were seen. Similar results were obtained by doing tbb-1(RNAi) in animals carrying a tbb-2 deletion (Table 4, line 6). Therefore, tbb-2 and tbb-1 function redundantly during embryonic development.

TBB-1–and TBB-2–containing Microtubules Interact Differently with MEI-1/MEI-2

During meiosis, microtubules fully resistant to MEI-1/MEI-2 katanin activity should result in meiotic defects similar to the complete loss of either mei-1 or mei-2. However, tbb-2(sb26) alone does not have any meiotic defects. It is possible that the MEI-1/MEI-2–interfering property of TBB-2(sb26) is simply masked by the presence of the functionally redundant TBB-1(+). Alternatively, TBB-2(sb26) may also permit some MEI-1/MEI-2 activity even in the absence of TBB-1(+). If the former is true, mei-1 and mei-2 lf phenotypes might be expected when TBB-1 is depleted by RNAi in the tbb-2(sb26) mutant. However, this only slightly increased the embryonic lethality (Table 4, line 7). Because no other β-tubulins are expressed at significant levels at this stage (Reinke et al., 2000; Baugh et al., 2003), microtubules containing TBB-2(sb26) as the only β-tubulin component are still partially sensitive to katanin activity, thus favoring the latter model described above. However, the high survival rate associated with removal of TBB-2 by RNAi or the tbb-2 null allele demonstrates that microtubules containing TBB-1 as the only β-tubulin isotype are also sensitive to MEI-1/MEI-2 (Table 4, lines 1–2).

In the absence of the sb26 mutation, it seems that microtubules are effective katanin substrates whether they contain only TBB-1 or only TBB-2 as the β-tubulin component. However, because mei-1 activity is present in excess during meiosis (Clandinin and Mains, 1993), microtubules containing either of the two β-tubulin isotypes could nonetheless differ in terms of their effectiveness of being MEI-1/MEI-2 katanin substrates. We took advantage of the partial lf mei-2 allele ct98, because in this sensitized background, differences in the genetic interactions between the different tubulin isotypes and katanin can be monitored more easily. As shown in Table 4 (lines 8–10), removal of TBB-2 by either a null mutation or RNAi in mei-2(ct98) animals substantially decreased hatching. In contrast, removal of TBB-1 by RNAi or the null mutation had very little effect (Table 4, lines 11–12). Therefore, MEI-1/MEI-2 interacts with microtubules more efficiently when TBB-2 is present, revealing a functional difference between the two β-tubulin isotypes. It is unlikely that this effect stems from TBB-2 being the more prevalent embryonic isotype because both tbb-1 and tbb-2 are expressed at similar levels at this stage (Reinke et al., 2000; Baugh et al., 2003), and Ellis et al. (2004) showed that TBB-1 expression is up-regulated in tbb-2 null mutants. Therefore, although TBB-1 and TBB-2 isotypes can each support embryogenesis on their own, these data demonstrate that MEI-1/MEI-2 katanin interacts with the two tubulins differently in vivo.

DISCUSSION

mei-1 and mei-2 encode subunits of the C. elegans homolog of microtubule-severing protein complex katanin (Srayko et al., 2000). Although MEI-1 and MEI-2 had been shown to disassemble interphase microtubules when coexpressed in HeLa cells, interactions of tbb-2(sb26) with mei-1 and mei-2 alleles provide the first genetic evidence that MEI-1 and MEI-2 indeed directly regulate meiotic spindle microtubule function, likely their length or dynamics, in C. elegans embryos. TBB-2 does not represent a tubulin specialized only for interactions with mei-1 and mei-2 because it is widely expressed throughout development, but TBB-2 does represent the β-tubulin that results in microtubules that are most susceptible to MEI-1/MEI-2 activity.

Our genetic evidence shows that the gf mutation tbb-2(sb26) inhibits MEI-1/MEI-2 katanin's activity on microtubules. This was demonstrated by the ability of tbb-2(sb26) to suppress mutations that cause ectopic mitotic MEI-1/MEI-2 expression and its enhancement of mutations that limit meiotic MEI-1/MEI-2 activity. However, sb26-containing microtubules are clearly not fully resistant to MEI-1/MEI-2 activity; otherwise tbb-2(sb26), which is wild type on its own, would be lethal as seen for null alleles of mei-1 or mei-2.

MEI-1 and MEI-2 still ectopically localize to mitotic spindles in mei-1(ct46gf); tbb-2(sb26) double mutants (Figure 2). These data, together with the known biochemical activity of MEI-1/MEI-2 katanin and the similarity of the mei-1(ct46gf) phenotype to that of embryos treated with microtubule-depolymerizing drugs (Strome and Wood, 1983; Hyman and White, 1987), strongly suggests that the sb26 tubulin mutation reduces the sensitivity of microtubules to the severing activity of katanin without interfering with other intrinsic properties of the MEI-1/MEI-2 complex, such as its intracellular localization. However, taxol-stabilized microtubules from sb26 and wild-type worms were at least qualitatively equally susceptible to human katanin severing in vitro (McNally, personal communication). This result was not surprising because completely resistant microtubules would be lethal, similar to mei-1 or mei-2 null mutations, and the in vitro assay may not be sensitive enough to detect subtle differences. An alternative way that tbb-2(sb26) could suppress ectopic MEI-1/MEI-2 severing is that it could result in intrinsically more stable microtubules. According to this model, sb26 microtubules might depolymerize more slowly in a cold environment (Hannak et al., 2002). However, wild-type and sb26 embryos showed similar length and organization of their mitotic spindles after different time periods of cold-induced depolymerization (our unpublished data). Furthermore, tbb-2(sb26) does not genetically interact with the lf zyg-9 allele b244 (our unpublished data), a mutation that results in a mei-1(ct46gf)-like mitotic phenotype, albeit due to loss of a microtubule-stabilizing protein (Kemphues et al., 1986; Matthews et al., 1998). This again suggests that sb26 specifically inhibits activity of the MEI-1/MEI-2 katanin complex rather than altering general microtubule dynamics.

The sb26 lesion is at the extreme C terminus of the β-tubulin protein, a region that is not essential for in vitro polymerization of microtubules (Lu and Ludueña, 1994). Interestingly, the small C-terminal fragments of both α- and β-tubulins are required for microtubule severing by katanin (McNally and Vale, 1993): when the C-terminal regions of the tubulins were removed by subtilisin digestion, katanin was able to bind the microtubules and hydrolyze ATP but could not disassemble the microtubules. This suggests that sb26 specifically disrupts a site required for katanin-mediated microtubule disassembly.

C. elegans has six β-tubulins. Among them, mec-7 is expressed exclusively in six touch neurons and two other neurons and is required only for touch sensitivity (Savage et al., 1989, 1994; Hamelin et al., 1992). Null alleles of the β-tubulin ben-1, which is also nonessential for development, confer resistance to the microtubule-depolymerizing drug benzimidazole (Driscoll et al., 1989). Among all the β-tubulins, microarray data indicates that only tbb-1 and tbb-2 mRNAs are expressed at a significant (and also at an equivalent) level in embryos (Reinke et al., 2000; Baugh et al., 2003). tbb-2–specific RNAi using the 3′ region resulted in only slight lethality (Table 4), consistent with the phenotype of tbb-2 deletion alleles. Similarly, depleting TBB-1 by either RNAi or a deletion allele also results in slight lethality (Table 4). However, simultaneous RNAi to both tbb-2 and tbb-1 resulted in 100% lethality and eliminated all microtubule structures (Figure 5), indicating that these tubulin isotypes function redundantly during embryonic development. Similar findings on developmental redundancies of α- and β-tubulins are also reported by Phillips et al. (personal communication) and Wright and Hunter (2003).

Even though katanin is involved in different microtubule-mediated developmental processes in a variety of species (Ahmad et al., 1999; Lohret et al., 1999; Webb et al., 2002; Bouquin et al., 2003), little is known about whether katanin prefers specific types of microtubules. Interestingly, although both tbb-1 and tbb-2 suffice for early C. elegans development and viability, MEI-1/MEI-2 katanin complex seems to prefer TBB-2 over TBB-1 for its activity (Table 4). The meiotic defects resulting from the weak lf mei-2(ct98) were enhanced by depleting tbb-2(+) by either RNAi or a deletion allele (Table 4). However, removal of the other β-tubulin isotype, TBB-1, by RNAi or the null mutation had no effect. Therefore, MEI-1/MEI-2 katanin severs spindle microtubules more efficiently when the TBB-2 β-tubulin isotype is present. Because the tbb-2 deletion allele alone does not give complete embryonic lethality, MEI-1/MEI-2 likely also uses other tubulin sites, β or α, in the absence of TBB-2, albeit less efficiently. The relatively high viability of animals lacking tbb-2 likely reflects the excess mei-1(+) activity normally present during meiosis (Clandinin and Mains, 1993), which can compensate for the decreased sensitivity of microtubules when TBB-2 is removed by mutations or RNAi.

Eukaryotic cells often express multiple α- and β-tubulin isotypes simultaneously, and these may also differ in their posttranslational modifications (Ludueña, 1998). Although different tubulin isotypes seem to be largely redundant, evidences for functional differences among tubulin isotypes do exist. For example, the budding and fission yeasts both express a pair of α-tubulin genes that are interchangeable for viability but the budding yeast isotypes differ in their effects on in vitro microtubule dynamics (Bode et al., 2003). Likewise, different mammalian β-tubulin isoforms have varying effects on in vitro microtubule dynamics and drug sensitivities (Panda et al., 1994; Derry et al., 1997). Our results demonstrate that different tubulin isotypes play different roles during katanin severing. Although at a superficial level tubulin isotypes may seem to be completely interchangeable, closer examination reveals that they indeed have specializations that may be important in nature.