Depletion of UBC9 Causes Nuclear Defects during the Vegetative and Sexual Life Cycles in Tetrahymena thermophila

Strains and cell culture.

Tetrahymena thermophila cell lines were obtained from the Tetrahymena Stock Center (Cornell University, Ithaca, NY). Cells were cultured in 1× SPP medium (2% proteose peptone, 0.1% yeast extract, 0.2% glucose, and 0.003% FeCl₃) at 30°C according to established procedures (25). Inbred wild-type strains B2086 (MPR1/MPR1 [MPR1; II]) and CU428 (mpr1-1/mpr1-1 [MPR1; mp-s VII]) were used for cell transformation and all other analyses. Star strains B*(VI) and B*(VII), which possess defective micronuclei but wild-type macronuclei, were used for generating UBC9 homozygous germ line knockout (KO) strains (26–29).

Construction of plasmids.

A Tetrahymena UBC9 knockout construct (pUBC9KO) was generated by ligating UBC9 flanking sequences into pMNBL, which contains a paromomycin-selectable cassette expressed via the metallothionein 1 (MTT1) promoter that is induced by CdCl₂ from T. thermophila (30). UBC9 upstream sequences (1,202 bp; positions 354009 to 355211 of scaffold 8254664) were amplified from a genomic DNA template from wild-type strain B2086 by PCR using Phusion DNA polymerase (Thermo Fisher Scientific Inc., Waltham, MA). The knockout cassette primers are shown in Table 1. The PCR product was cloned into the unique XhoI and BglII restriction sites of pMNBL. The corresponding downstream flanking sequences (1,404 bp; positions 356442 to 357846 of scaffold 8254664) were amplified and cloned into the unique BamHI and NotI restriction sites.

A GFP-UBC9 fusion construct was made in a pENTR Gateway plasmid (Life Technologies, Carlsbad, CA). The 997-bp coding region of the UBC9 gene (TTHERM_00522720, NW_002476431.1 [GI:229595896]) from the second codon (eliminating the initiating methionine) to the TGA stop codon was PCR amplified and cloned into the pENTR-D entry vector. The gene cassette in the entry vector was then inserted into a pBS-MTT-GFP-gtw destination vector (obtained from Doug Chalker, Washington University, St. Louis, MO) by using the LR recombinase in the Gateway cloning system (Life Technologies, Carlsbad, CA). Successful integration of the GFP-UBC9 fusion gene at the Tetrahymena RPL29 locus conferred cycloheximide resistance (12.5 μg/ml) (31).

Ubc9p-mCherry C-terminal fusions were made in plasmid pmCherryLAP-NEO2 as described previously (32). Approximately 1 kb of the 3′ end of the UBC9 coding region was ligated into the KpnI and NotI sites adjacent to the mCherryLAP (localization and affinity purification) tag, and 1 kb of the 3′ downstream sequence was placed adjacent to the NEO2 selectable marker, which confers paromomycin resistance.

The dominant negative UBC9 (pDN-UBC9) construct was generated by inserting an N-terminal His₆-FLAG-tagged version of UBC9 containing a mutation of the conserved cysteine in the 100th position to a serine residue (C100S). The gene was chemically synthesized by IDT (Coralville, IA) with BsiWI and ApaI restriction sites adjacent to the 5′ and 3′ coding regions, respectively. The coding region was then inserted into plasmid pBS-MTT-YFP-gtw (obtained from Doug Chalker, Washington University, St. Louis, MO) after digestion with BsiWI and ApaI. The construct was linearized by HindIII digestion and transformed into Tetrahymena. This construct integrates adjacent to the RPL29 locus by homologous recombination and confers resistance to cycloheximide (12.5 μg/ml).

Construction of UBC9 germ line knockout heterokaryons.

Plasmid pUBC9KO was purified by using a Qiagen Plasmid Maxi kit (Qiagen, Valencia, CA) and linearized with XhoI and NotI digestion. Gold particles (S550d DNAdel Gold carrier particles; Seashell Technology LLC) were coated with the DNA and introduced into mating B2086 and CU428 cells at 2.5 to 3.5 h postmixing (33) by biolistic bombardment, as described previously (34). Putative micronuclear UBC9 knockouts (ΔUBC9) were selected by growth in the presence of paromomycin (100 μg/ml) followed by 6-methylpurine (6-MP) (7.5 μg/ml). Paromomycin selects for insertion of the knockout construct in the genome, and 6-MP selects for cells that have completed conjugation and formed a new macronucleus. The heterozygous (ubc9::neo3/ubc9 [pm-r; paromomycin resistant]) transformants were test crossed to a wild-type strain to confirm the expected genetic segregation of 50% paromomycin resistance among progeny. After confirmation, the heterozygous cell line was mated to star strains B*(VI) and B*(VII) to generate two homozygous germ line knockout heterokaryon strains of different mating types, BVI ΔUBC9 (ubc9::neo3/ubc9::neo3 [VI; UBC9⁺ pm-s]) and BVII ΔUBC9 (ubc9::neo3/ubc9::neo3 [VII; UBC9⁺ pm-s; paromomycin sensitive]). Star strains contain defective micronuclei that do not contribute meiotic products in the cross and therefore result in endoreduplication of the haploid genome from the heterozygous micronuclear knockout strain and generation of a homozygous knockout MIC genome. Star crosses do not complete conjugation; they maintain their parental MAC, and the progeny of the star parent contains a wild-type MAC with a homozygous knockout MIC. BVI and BVII ΔUBC9 germ line knockout heterokaryons were crossed to obtain progeny that are complete ΔUBC9 homozygous homokaryon strains. These cells were used for phenotypic analyses with the initially generated complete micronuclear and macronuclear knockout lines described above. Elimination of the UBC9 gene was confirmed by genomic PCR (see Fig. 4C) and genetic crosses to wild-type strains (see Fig. 3).

Generation of a conditional UBC9 mutant line.

The pGFP-UBC9 construct that was inserted adjacent to the RPL29 locus was biolistically transformed (33) into mated heterokaryon homozygous ΔUBC9 cells at 8 h postmixing. Cells were selected with 12.5 μg/ml cycloheximide and induced with CdCl₂ to initiate the expression of the GFP-UBC9 fusion gene. Background cycloheximide-resistant cell lines could be generated by transformation of the parental heterokaryon cell lines, but these cells would have paromomycin-sensitive MACs. The cycloheximide-resistant GFP-UBC9 progeny of the cross were tested for paromomycin resistance to demonstrate successful mating and generation of the new UBC9 knockout MAC.

Isolation of DNA and RNA and RT-PCR.

Genomic DNA was isolated by phenol-chloroform extraction followed by isopropanol precipitation, as described previously (34). Wild-type Tetrahymena thermophila strain B2086 and conditional UBC9 KO strains were used for isolation of total RNA by using a Qiagen RNeasy minikit (Qiagen, Valencia, CA) with a QIAshredder instrument (Qiagen, Venlo, Germany) for cell homogenization. For reverse transcription-PCR (RT-PCR), ∼1 μg of RNA was used for first-strand cDNA synthesis (qScript cDNA supermix; Quanta Biosciences Inc., Gaithersburg, MD), using gene-specific primers that amplify wild-type UBC9 (UBC9WT5′UTRF and UBC9WT5′UTRR) (Table 1).

Fluorescence microscopy.

Approximately 1 ml (2 × 10⁵ cells) of vegetative or conjugating cells was concentrated to 10⁶ cells/ml by low-speed centrifugation (1,100 × g), fixed for 30 min with an equal volume of 4% paraformaldehyde, washed once in 1× phosphate-buffered saline (PBS) (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄ [pH 7.4]), and then stained with DAPI (4′,6′-diamidino-2-phenylindole) at 1.0 μg/ml for 10 min. Cells were then examined under a fluorescence microscope (Olympus BX51 TF microscope).

Growth rate.

UBC9 conditional mutant strains were first cultured overnight in SPP medium with 0.1 μg/ml CdCl₂, washed twice with 10 mM Tris-Cl (pH 7.5), and then transferred to 10 ml SPP medium at a concentration of 200 cells/ml in the presence or absence of 1.0 μg/ml CdCl₂. Cell populations were measured by direct cell counts at 0, 4, 8, 18, 24, and 48 h by using a hemocytometer.

Drug sensitivity.

UBC9 conditional mutant cells and wild-type B2086 cells were cultured in SPP medium in the presence of 1.0 μg/ml CdCl₂ or no CdCl₂ for 24 h, treated with 8 mM methyl methanesulfonate (MMS) or 2 mM cisplatin for 2 h, and then washed twice with 10 mM Tris-Cl (pH 7.5). Single cells were isolated in SPP drops containing 0.5 μg/ml CdCl₂. After 48 h, the number of cells in each drop was counted. Drops containing <10 cells were scored as nonviable, and drops with >500 cells were scored as viable.

Pairing efficiency.

Two different mating types of UBC9 conditional mutants and wild-type cells were cultured separately in SPP medium for 16 h, harvested at 1,100 × g, washed twice with 10 mM Tris-Cl (pH 7.5), and starved in the same buffer in the presence of 0.05 μg/ml or no CdCl₂ for 16 to 24 h at 30°C. Conditional mutant or wild-type cells of different mating types were mixed (time zero), and samples were taken at later time points, fixed with an equal volume of 4% paraformaldehyde, and then counted for pairing efficiency. One hundred cells were scored at 2, 4, 6, and 8 h postmixing. Pairing efficiency was calculated by counting the number of mating pairs, multiplying by 2 (2 cells per pair), and then dividing by the total number of cells.

Analysis of DN-UBC9 during conjugation.

Tetrahymena strains of different mating types expressing DN-UBC9 were grown in SPP medium for ∼24 h to a concentration of 2 × 10⁵ cells per ml. Cells were washed twice in 10 mM Tris (pH 7.5), suspended at 2 × 10⁵ cells per ml in the same solution for 24 h, and then mixed to initiate mating. At 6 hours postmixing, the culture was split into two equal portions and one half was induced with 0.1 μg/ml CdCl₂. At 24 h postmixing, 200 μl of cells was fixed with an equal volume of 4% paraformaldehyde and stained with DAPI for microscopy.

UBC9 is an essential gene.

We identified a single UBC9 homolog in the Tetrahymena genome. Searches of the Tetrahymena Genome Database (http://ciliate.org/index.php/home/welcome/) with UBC9 orthologs from Saccharomyces cerevisiae, Schizosaccharomyces pombe, and Drosophila melanogaster revealed the same top BLAST hit (TTHERM_00522720). Reciprocal searches of the S. pombe and D. melanogaster genomes using the Tetrahymena ortholog identified the corresponding UBC9 genes as the top BLAST hits. The Tetrahymena Ubc9p deduced amino acid sequence has ∼55% identity with UBC9 orthologs from yeast and Drosophila (Fig. 1). The amino acid sequence is 87% identical across the conserved catalytic domain (amino acids 92 to 100) and contains the conserved catalytic cysteine residue found in other Ubc9p proteins (35).

Previous studies showed that UBC9 orthologs are essential in budding yeast, nematodes, zebrafish, and mice (10, 15, 18, 36). An exception is S. pombe, where the deletion of hus5 (UBC9 ortholog) causes severe mitotic defects and slow growth but not death (14). To determine if UBC9 is essential in Tetrahymena, micronuclear (germ line) UBC9 knockout cell lines were constructed. The MIC knockout was constructed according to standard procedures by inserting the neo3 cassette (conferring paromomycin resistance) into one allele of the MIC UBC9 locus (Fig. 2A). This cell line was subsequently mated to star strains to generate cells with a homozygous knockout of UBC9 in the MIC and a wild-type MAC (Fig. 2B). Details of the construction of these homozygous MIC heterokaryons can be found in Materials and Methods (construction of UBC9 germ line knockout heterokaryons). Since the MIC is not transcribed, the knockout heterokaryon cell lines are phenotypically paromomycin sensitive (wild-type MAC), but PCR amplification generated the expected 480-bp product derived from the neo3 insertion in the UBC9 MIC gene (Fig. 2C). The homozygous ΔUBC9 heterokaryon cell lines were mated to wild-type strains and generated progeny that were 100% paromomycin resistant. This showed that the heterokaryon lines were fertile and homozygous for neo3 in the MIC (Fig. 3). The two heterokaryon lines were mated to each other to generate complete deletions of UBC9 in both the MIC and MAC of the progeny. Individual pairs were isolated and placed into drops with nutrient medium. The number of cells in each drop was counted when cell division was arrested (between 48 and 72 h). The average number of cells per drop for the 75% of cell lines that arrested and subsequently died was ∼420. Assuming geometric growth and starting with 2 cells per drop, we estimate that progeny died roughly 7 to 8 cell divisions after mating (Fig. 3). The remaining drops contained live cells after 72 h, but these cells could result from a failure to complete conjugation rather than the survival of progeny with a complete UBC9 deletion. To distinguish between these possibilities, surviving cells were tested for resistance to paromomycin. True progeny that successfully completed conjugation by exchanging genetic material and forming a new MAC should be paromomycin resistant. All drops were sensitive to paromomycin, which shows that they are parental lines that did not complete conjugation (Fig. 3). These results demonstrate that the UBC9 gene is essential for cell viability. Presumably, cells survive for several fissions after conjugation (with the formation of a new MAC) because Ubc9p from the parental cell is stable and is gradually diluted and/or degraded as the progeny cells divide. The results also demonstrate that there is no requirement for expression of the UBC9 gene from the developing MAC to complete conjugation.

Generation of a cell line with conditional Ubc9p expression.

The lethal phenotype of strains with a complete deletion of UBC9 afforded only limited opportunities for molecular analysis of Ubc9p function. To circumvent this problem, we created a cell line with conditional expression of Ubc9p. The strategy involved mating UBC9 homozygous heterokaryon strains and then transforming these cells with a GFP-UBC9 fusion gene regulated by the Tetrahymena thermophila metallothionein 1 (MTT1) gene promoter (Fig. 4A). The GFP-UBC9 transgene was introduced into the progeny of mated homozygous heterokaryon Tetrahymena cells at 8 h postmixing, when cells began to generate new MACs. Insertion of the transgene into the RPL29 locus of the new MAC conferred cycloheximide resistance (37). Transformants were selected with cycloheximide and induced with CdCl₂ as soon as progeny were placed into growth medium to initiate the expression of the GFP-UBC9 fusion gene. The resulting progeny cell lines were green fluorescent protein (GFP) positive and paromomycin resistant, as expected. As predicted by the localization of Ubc9p orthologs in other species, we observed nuclear GFP localization when cells were grown in the presence of CdCl₂ (Fig. 4B, middle and right). In order to verify GFP-UBC9 CdCl₂-dependent regulation, conditional mutants were grown in medium without CdCl₂ and in the presence of different concentrations of CdCl₂. Cells were fixed with 4% paraformaldehyde and examined by using fluorescence microscopy. As shown in Fig. 4B (left), no fluorescence signal was detected in the cells cultured without CdCl₂, consistent with the low level of MTT1 promoter activity in the absence of heavy metal. On the other hand, an increasing fluorescence signal was observed in the nucleus when the cells were cultured in 0.1 μg/ml or 1.0 μg/ml CdCl₂ (Fig. 4B, middle and right). RT-PCR was carried out to confirm the presence of RNA transcripts containing UBC9. Two sets of primers were designed, as indicated in Fig. 4C. The first set was used to detect UBC9 transcripts from the endogenous locus. The forward primer was in the 5′ untranslated region (UTR) of endogenous UBC9, and the reverse primer was inside the UBC9 coding region (Fig. 4C, arrowheads). The second set of primers was inside the UBC9 coding region (Fig. 4C, arrows). The band corresponding to the endogenous UBC9 transcript was observed only in wild-type Tetrahymena cells and was not present in ΔUBC9 conditional mutants (Fig. 4D, top). On the other hand, bands corresponding to transcripts from the UBC9 coding region were detected in both the wild type and conditional mutants (Fig. 4D, middle), which demonstrates that ΔUBC9 conditional mutants express only the GFP-UBC9 form and not wild-type UBC9. We therefore conclude that a conditional CdCl₂-dependent UBC9 expression strain was generated.

Depletion of UBC9 leads to reduced cell growth and nuclear defects.

To determine whether depletion of Ubc9p affects the Tetrahymena growth rate, equal numbers of cells (200 cells/ml) were inoculated into medium with or without 1.0 μg/ml CdCl₂, and cells were counted at intervals of 4, 8, 18, 24, and 48 h. As shown in Fig. 5, ΔUBC9 conditional cells had a lower growth rate than did wild-type cells in the absence of CdCl₂. However, the same cell lines cultured in CdCl₂ exhibited growth rates similar to those of wild-type cells in CdCl₂ during the first 24 h, and cell numbers then increased over the next 24 h to levels nearly as high as those of the wild type without CdCl₂ (Fig. 5). It should be noted that the conditional mutant grown in the absence of CdCl₂ also showed growth, most likely due to leaky expression from the MTT1 promoter, as we have seen in previous studies (24). In other species, Ubc9p has been shown to regulate cell cycle progression, including mitosis and chromosome segregation (14, 15, 38). Depletion of Ubc9p has been shown to cause chromosome missegregation in zebrafish (18). DAPI staining revealed that many Ubc9p-depleted cells lost detectable micronuclei (Fig. 6A, middle). Conditional mutants as well as wild-type cells, cultured with or without CdCl₂, were examined after 24 h. Micronuclei were counted, and a summary of the data is shown in Fig. 6B. Nearly half of the Ubc9p-depleted cells contained no DAPI-detectable micronuclei, compared with none of the wild-type cells, in the presence or absence of CdCl₂. We observed unequal partitioning of DNA to daughter micronuclei in 6 of 9 cells that were identified at late stages of mitosis, consistent with defects in the separation of sister chromatids (Fig. 6C). Unlike most eukaryotes, Tetrahymena cells do not require accurate chromosome segregation during mitosis because they do not rely on the MIC for gene expression during the vegetative life cycle. Cells can be maintained indefinitely with micronuclear aneuploidy (39). Surprisingly, we observed an increase in the number of cells having more than two MICs when the medium was supplemented with CdCl₂ (Fig. 6B). We decided to investigate possible dominant negative effects of overexpression of GFP-Ubc9p. We constructed cell lines from wild-type strains B2086 and CU428 by introducing a CdCl₂-inducible mutant of Ubc9p with a cysteine-to-serine mutation (C100S) in the conserved active site. These dominant negative UBC9 (DN-UBC9) cell lines were further modified by disruption of the macronuclear copy of the wild-type UBC9 gene with plasmid pUBC9KO to reduce the amount of wild-type protein. These cells were treated with 1.0 μg/ml CdCl₂ for 24 h to induce DN-Ubc9p expression. The cells should have shown decreased expression levels of catalytically active Ubc9p, yet rather than lacking micronuclei, they contained multiple MICs (Fig. 7A and B). To determine whether this phenotype was due to the overexpression of DN-Ubc9p or depletion of the endogenous copy, we counted the number of MICs using cell lines expressing DN-UBC9 without a knockout of wild-type UBC9. As shown in Fig. 7A, cells expressing DN-UBC9 alone displayed a multiple-MIC phenotype similar to that of DN-UBC9 KO cells (66% versus 72%). To determine whether overexpression of wild-type UBC9 had the same effect on the number of MICs, we scored cells expressing dual-tagged 6×His-, 3×FLAG-Ubc9p (6H3F-UBC9) that was also driven by the MTT1 promoter and induced by CdCl₂. Only 9% of cells had >2 MICs (Fig. 7A). This is similar to what we saw in wild-type cells in the presence of CdCl₂ (5%), which is apparently due to the effect of CdCl₂. This rather remarkable difference in phenotype between the depletion of Ubc9p and overexpression of defective Ubc9p cannot be readily explained. We examined the DN-UBC9 cell lines for defects in cytokinesis, but no obvious problems were identified. Individual cells were cultured in drops, and total cells were counted after 24 h, but no statistical differences between DN-UBC9 and wild-type cells could be found (data not shown). In addition, a time course experiment showed that >50% of induced DN-UBC9 cells contained >2 MICs within the first 9 h after the addition of CdCl₂ (see Fig. S3 in the supplemental material). This short time interval suggests that we should have detected a cytokinesis defect in cell division if it was responsible. No cells showed evidence of incomplete MAC separation between cells at division, analogous to the cut phenotype of yeast (40) (data not shown). There are possible alternative explanations for the increased number of MICs presented in the Discussion.

Depletion of Ubc9p results in hypersensitivity to DNA-damaging agents.

A role for SUMOylation in DNA damage repair, especially in response to double-strand breaks, has been established for mammalian species and yeast (41–45). To evaluate whether UBC9 is involved in DNA damage repair in Tetrahymena, we tested the effect of the DNA-damaging agents methyl methanesulfonate (MMS) and cisplatin on Ubc9p-depleted cells. ΔUBC9 conditional mutants and wild-type cells were cultured separately in SPP medium containing 1.0 μg/ml CdCl₂ or no CdCl₂. Cells were treated with 8 mM MMS or 2 mM cisplatin for 2 h and then washed twice with 10 mM Tris (pH 7.5). Single cells were placed into drops (∼15 μl) of SPP medium containing 0.5 μg/ml CdCl₂. CdCl₂ in the posttreatment drops provides conditions that are optimal for growth of conditional cell lines that would have poor growth without CdCl₂. After 48 h, the cells in each drop were counted and scored as viable if >500 cells were in a drop and nonviable if there were 0 to 10 cells. No drops had cell numbers between 10 and 500. As shown in Fig. 8, ΔUBC9 conditional cells cultured in the absence of CdCl₂ showed only 12% survival in MMS, compared with 90% survival when cultured with CdCl₂. The survival rate with cisplatin increased ∼15-fold (5% versus 80%) in the presence of CdCl₂. Wild-type cells showed little change in sensitivity regardless of the presence or absence of CdCl₂ (Fig. 8). This finding is consistent with a role of Ubc9p and SUMOylation in DNA break repair.

Ubc9p localizes to developing macronuclei.

We recently showed that a large increase in the amount of SUMOylated protein is observed during Tetrahymena sexual reproduction and that the increased signal is centered in the developing macronucleus (24). If this SUMOylation occurs within or around the developing MAC, then the signal for Ubc9p should also accumulate in the anlagen. To establish the location of Ubc9p during conjugation, we generated Tetrahymena cell lines in which mCherry is fused to the C terminus of UBC9. The transgene is created by homologous recombination in the MAC and is driven by the endogenous UBC9 promoter. Cells were starved and mixed to induce mating (see Materials and Methods), and samples were then fixed at various time points, stained with DAPI, and observed by using a fluorescence microscope (Fig. 9). Wild-type cells without an mCherry tag were used as controls. A faint signal was observed in the macronucleus of starved cells, indicating low expression levels of Ubc9p during starvation (Fig. 9). During early conjugation, the signal was observed exclusively in parental macronuclei during the pronuclear exchange and postzygotic division stages (T = 3, T = 4, and T = 6) (Fig. 9). At 8 h postmixing, the signal accumulated in the developing MAC (T = 8) (Fig. 9). The mCherry signal was much stronger in the developing macronuclei than in the DNA-rich parental MAC, which stained brightly with DAPI (Fig. 9, asterisks). The efficiency of integration of this construct was too low to successfully rescue the knockout heterokaryons; therefore, we were not able to demonstrate that Ubc9p-mCherry was fully functional. As an alternative approach, we used the GFP-UBC9 conditional mutants described above. These cells were mated to wild-type cells in the presence of CdCl₂. As shown in Fig. S1 in the supplemental material, GFP-Ubc9p localization was consistent with the Ubc9p-mCherry results. The protein was found exclusively in the parental MAC during meiotic prophase and postzygotic divisions. The signal then migrated to the developing MAC and eventually disappeared from the parental MAC. Although we cannot evaluate the relative signal strength in these experiments because the expression of GFP-Ubc9p is driven by CdCl₂, the results provide confidence in the localization of Ubc9p to the developing MAC, the site of developmentally regulated genome reorganization.

Ubc9p is required for pair formation in conjugating Tetrahymena cells.

The localization of Ubc9p-mCherry to the developing macronucleus suggested a role in conjugation; consequently, we mated our UBC9 conditional cell lines to assay for a conjugation-defective phenotype. ΔUBC9 conditional mutants of different mating types were cultured overnight in SPP medium with or without CdCl₂ to promote or reduce the expression of the GFP-UBC9 transgene. Cells were washed twice in 10 mM Tris (pH 7.5) and cultured in the same medium plus or minus CdCl₂. Wild-type cells served as controls. ΔUBC9 conditional cells from non-CdCl₂-treated cultures (Ubc9p depleted) were mixed to initiate mating. Separate cultures of CdCl₂-treated Ubc9p conditional lines or wild-type cells were also mixed to initiate cell pairing. Cells were evaluated at 2, 4, 6, and 8 h postmixing for pair formation. As shown in Fig. 10, UBC9 conditional mutant cultures that were not exposed to CdCl₂ (Ubc9p depleted) were unable to form mating pairs at 2 h postmixing and formed only 10% pairs after 8 h, well past the expected pairing period. In contrast, the same cell line supplemented with CdCl₂ was able to generate 40 to 60% mating pairs. Wild-type cells exhibited mating efficiencies of >90%. We consistently observed a slight loss of mating efficiency when wild-type cells were exposed to CdCl₂ (Fig. 10), so a large increase upon exposure to CdCl₂ is particularly significant. In addition to Ubc9p effects on pairing, other experiments showed a pairing defect for Uba2p- and Smt3p (SUMO)-depleted cell lines (24). When ΔUBC9 conditional cells cultured in the absence of CdCl₂ (UBC9 depleted) were mixed with wild-type cells, pairs formed and subsequently entered meiosis (see Fig. S2 in the supplemental material). After 8 h, 41% of cells from the conditional × wild-type cross remained in the crescent stage (meiotic prophase), compared with 0% of wild-type × wild-type pairs. The stall was temporary, and by 24 h, the UBC9-depleted cells progressed through meiosis. This phenotype indicates an additional role for UBC9 in meiosis. Cells expressing dominant negative Ubc9p showed normal pairing regardless of CdCl₂ treatment (data not shown).

Dominant negative Ubc9p results in arrest at the macronuclear development stage.

The inability of Ubc9p-depleted cells to form pairs prevented the analysis of later stages during conjugation. As an alternative approach, we examined matings between cells expressing the above-described DN-Ubc9p. DN-UBC9 cells of two different mating types were cultured in growth medium and subsequently starved without CdCl₂. Cells were then mixed to start conjugation. Half the volume of mixed cells was placed into a separate petri dish and supplemented with 0.1 μg/ml CdCl₂ at 6 h postmixing. At 24 h postmixing, samples were fixed, stained with DAPI, and examined by fluorescence microscopy. DN-UBC9 cells that were not treated with CdCl₂ progressed through anlagen formation normally, as did wild-type cells and cells expressing an affinity-tagged version of UBC9 (6H3F-UBC9) (Fig. 11, micrographs). DN-UBC9 cells that were treated with CdCl₂ (inducing the expression of DN-Ubc9p) contained a large fraction of cells that remained as mating pairs that retained their parental MAC (42% of total cells scored) (Fig. 11). To test whether an earlier addition of CdCl₂ would result in an earlier arrest in conjugation, a small volume of mixed DN-Ubc9p cells was placed into a separate petri dish and induced with 0.1 μg/ml CdCl₂ at the time of mixing. A similar fraction of cells also arrested with the same phenotype at 24 h (data not shown). These crosses using cells with excess catalytically defective Ubc9p demonstrate a role for Ubc9p in completion of conjugation.

Ubc9p regulates micronuclear chromosomal segregation.

Several results from our study are consistent with data from previous investigations of Ubc9: (i) complete deletion of UBC9 was lethal, a result consistent with findings for S. cerevisiae, Caenorhabditis elegans, and mammalian cells (10, 18, 36, 46); (ii) depletion of Ubc9p resulted in greater sensitivity to MMS and cisplatin, consistent with its role in DNA damage repair seen in other studies (41, 47); and (iii) disruption of Ubc9p function resulted in defective mitosis of the micronucleus. MICs were undetectable in a large fraction of UBC9-depleted cells during vegetative growth. Nine cells were identified during cell division and six displayed large differences in the intensity of DAPI-stained MICs such as those shown in Fig. 6C. The results are consistent with a missegregation of MIC chromosomes during mitosis. The requirement for SUMOylation in chromosome segregation is well established for other species (48–51). In budding yeast, deletion of UBC9 results in gross defects in chromosome structure and integrity as well as aberrant segregation and polyploidy (15). There is also evidence that SUMOylation plays critical roles in centromere function. The Saccharomyces cerevisiae Slx5/8 complex is the founding member of a recently defined class of SUMO-targeted ubiquitin ligases (STUbLs) (52). The Slx5/8 complex is preferentially located near centromeres, and deletion of either gene shows severe mitotic defects that include aneuploidy and spindle mispositioning (53). Defective chromosome segregation is also linked to defective SUMOylation of topoisomerase II (54) and the mechanism of kinetochore assembly and disassembly (reviewed in reference 51).

A defective mitotic phenotype was described previously for Tetrahymena by Cui and Gorovsky (55) after depletion of a centromeric H3 protein called Cna1p. The CNA1 gene encodes a centromere-specific histone H3 variant, Cna1p, also known as CenH3 (56–58), that is associated with centromeric DNA in place of the typical H3. CenH3 is required for functional centromeres and recruitment of other centromeric and spindle checkpoint proteins (59). A reduction in the MAC copy number of CNA1 results in MICs that are smaller than MICs in wild-type cells (55), rather than the loss-of-MIC phenotype observed for our Ubc9p depletion cells. These small MICs continue to undergo mitosis, but DNA is unequally distributed to daughter nuclei (55), which is similar to our observations for cells with reduced expression of Ubc9p.

Surprisingly, the overexpression of catalytically inactive dominant negative Ubc9p resulted in multiple MICs per cell. In chicken cells, the depletion of Ubc9p generated a fraction of cells with multiple nuclei (46), and the authors of that report suggested that a cytokinesis defect could be responsible. In our Tetrahymena DN-UBC9 cells, it is possible to generate multi-MIC cells without any effects on the genetic phenotype because the MIC is transcriptionally silent. However, we did not detect any defect in cytokinesis. Cell counts after 24 hours for wild-type, 6H3F-UBC9, and DN-UBC9 single cells in the presence or absence of CdCl₂ did not result in a statistically significant difference between cell lines (data not shown). A time course experiment performed after CdCl₂ induction showed that 50% of DN-UBC9 cells had >2 MICs within 9 h (see Fig. S3 in the supplemental material), suggesting that if a cytokinesis defect was responsible, it should have been detectable. We have considered alternatives to a defect in cytokinesis, and these include either a defective mitotic check point that allows reentry to mitosis or the formation of MICs with partial genomes. While we are unaware of any example of the former, there is some precedence for the latter in organisms with open mitosis that involves nuclear envelope breakdown and re-formation (reviewed in reference 60). Chromosomes that do not segregate with the majority of the genome can be enclosed in a nuclear envelope. This theory has the appeal of connecting segregation defects with both depletion and dominant negative UBC9 phenotypes, but there is no evidence for this phenomenon in organisms with closed mitosis such as Tetrahymena. Multiple MICs were previously reported for Tetrahymena cells, resulting from a MAC knockout of TGP1 (61), encoding a G-quartet DNA-binding protein, yet these cells showed normal division rates. It is possible that a novel mechanism for multiple MICs could be operating in Tetrahymena. Interestingly, Tgp1p has appeared as a candidate SUMO substrate and a Ubc9p-interacting protein in our recent proteomics studies (Q. Yang and A. M. Nasir, unpublished data). Together, our results indicate that Ubc9p plays an indispensable role in Tetrahymena mitosis through regulating chromosome segregation. Having a separate germ line nucleus that does not contribute to gene expression makes Tetrahymena a unique system for studies of mitotic chromosome segregation because other organisms are not able to survive with aneuploid nuclei.

Role of Ubc9p in Tetrahymena sexual reproduction.

Our initial interest in SUMOylation in Tetrahymena was stimulated by studies of the related ciliate Paramecium tetraurelia, where RNAi-induced silencing of UBA2 or SUMO prevented the excision of micronucleus-specific DNA elements during formation of the somatic macronucleus (23). For Tetrahymena, analysis of SUMOylation requirements during sexual reproduction has proven to be more complex. A large increase in the amount of SUMOylated substrates during MAC development (24) and the accumulation of Ubc9p (Fig. 9), SUMO, and Uba2p in MAC anlagen (24) are consistent with important roles during MAC development. However, depletion of the SUMO pathway protein Ubc9p (Fig. 10), Uba2p, or SUMO (24) prevents cell pairing, the first step in sexual reproduction. The observation that deficient SUMOylation inhibits cell pairing was surprising based on the expression profile of Tetrahymena SUMO pathway genes, yet in yeast, degradation of the mating-type factor α1 protein, important for the establishment of mating type, requires the STUbLs Slx5 and Slx8 (62). The mechanism required for SUMO-dependent pairing in Tetrahymena is not known, but the recent identification of the mating-type protein (63) provides an opportunity to examine whether it involves direct SUMOylation of the mating-type protein or an indirect signaling effect. This block in cell pairing at the first step of conjugation complicated efforts to investigate later stages, including MAC development.

To overcome this SUMOylation-dependent block, we expressed dominant negative Ubc9p after the start of conjugation. A large fraction of cells (36%) remained in pairs 24 h after the start of conjugation, and the parental MAC was not destroyed. Genetic analysis confirmed that crosses between cells expressing dominant negative Ubc9p resulted in fewer true progeny than for wild-type controls (data not shown). The results provide evidence for an additional SUMO-dependent step during conjugation. Failure to degrade the parental MAC is an uncommon mutant phenotype, but Tetrahymena cells treated with nicotinamide, an inhibitor of sirtuin histone deacetylase, arrest in this stage (64). These results suggest that increased acetylation of substrates results in retention of the parental MAC. Competition between SUMOylation and acetylation is known to alter the balance of sirtuin expression in human cells (65). Another example of parental MAC retention was observed after deletion of ATG8-2, a gene encoding a protein in the autophagy pathway (66). Although there is as yet no evidence for regulation of nuclear autophagy in Tetrahymena by SUMOylation, there is evidence that SUMOylation and acetylation regulate p53-mediated autophagy in mammalian cells (67). Considering the importance of SUMOylation in the nuclear events of mitosis, meiosis, DNA repair, and transcriptional control, we are not surprised to find that multiple steps of Tetrahymena conjugation are dependent on SUMOylation. We expect that our current efforts to identify SUMOylated substrates during sexual reproduction will reveal key targets that are required for genome reorganization during macronuclear development as well as cell pairing and nuclear degradation.