View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Developmental Cell, Vol. 9, 769–779, December, 2005, Copyright ª2005 by Elsevier Inc. DOI 10.1016/j.devcel.2005.10.007 The SUMO Pathway Is Essential for Nuclear Integrity and Segregation in Mice

Karim Nacerddine,1 Franc¸ois Lehembre,1,4,5 is covalently attached to lysines in substrate proteins Mantu Bhaumik,3,4,6 Je´roˆ me Artus,2 by a series of enzymatic reactions similar to those in- Michel Cohen-Tannoudji,2 Charles Babinet,2 volved in ubiquitination. SUMO is first activated by the Pier Paolo Pandolfi,3 and Anne Dejean1,* E1-activating , the Aos1/Uba2 heterodimer; sub- 1 Unite´ d’Organisation Nucle´ aire et Oncoge´ ne` se sequently transferred to the unique E2 enzyme, Ubc9; INSERM U 579 and conjugated to substrates in a reaction that is cata- 2 Unite´ de Biologie du De´ veloppement lyzed by a variety of E3 such as the PIAS family CNRS URA 2578 members (Johnson and Gupta, 2001; Sachdev et al., Institut Pasteur 2001; Takahashi et al., 2001), the RanBP2 nucleoporin 75724 Paris Cedex 15 (Pichler et al., 2002), or the polycomb protein Pc2 (Kagey France et al., 2003). SUMO is removed from substrates in re- 3 Molecular Biology Program actions catalyzed by a family of isopeptidases called Department of Pathology SENPs. Memorial Sloan-Kettering Cancer Center In S. cerevisiae, Ubc9-depleted cells arrest in G2/M New York, New York 10021 phase, causing the accumulation of large budded cells with a single nucleus, a short spindle, and replicated DNA (Seufert et al., 1995). By contrast, in S. pombe, hus5 Summary (Ubc9 homolog) deletion is viable, but cells are severely impaired in growth and exhibit high levels of abortive Covalent modification by SUMO regulates a wide range mitosis and chromosome missegregation (al-Khodairy of cellular processes, including transcription, cell cy- et al., 1995), a phenotype closely resembling that of cle, and chromatin dynamics. To address the biologi- pmt3 (SUMO homolog) loss (Tanaka et al., 1999). Ubc9 cal function of the SUMO pathway in mammals, we gen- or SUMO RNAi knockdown experiments in C. elegans re- erated mice deficient for the SUMO E2-conjugating sult in embryonic arrest after gastrulation and pleiotropic enzyme Ubc9. Ubc9-deficient embryos die at the early defects in larval development such as vulval eversion postimplantation stage. In culture, Ubc9 mutant blas- at the fourth larval stage and abnormal tail morphology tocysts are viable, but fail to expand after 2 days and (Jones et al., 2002). In D. melanogaster, the Ubc9 le- show of the inner cell mass. Loss of Ubc9 thal mutant, also known as lesswright, dominantly sup- leads to major chromosome condensation and segre- presses the nondisjunction and cytological defects of gation defects. Ubc9-deficient cells also show severe female meiotic mutations that affect spindle formation defects in nuclear organization, including nuclear en- (Apionishev et al., 2001). Another fly Ubc9 mutation, velope dysmorphy and disruption of nucleoli and PML termed semushi, blocks nuclear import of Bicoid during nuclear bodies. Moreover, RanGAP1 fails to accumu- embryogenesis and results in misregulation of the seg- late at the nuclear pore complex in mutant cells that mentation that are Bicoid targets with consequent show a collapse in Ran distribution. Together, these loss of the anterior half of the larval body structure (Epps findings reveal a major role for Ubc9, and, by implica- and Tanda, 1998). Conditional depletion of Ubc9 in tion, for the SUMO pathway, in nuclear architecture and chicken DT40 cells leads to an increase in the number of function, chromosome segregation, and embryonic vi- cells containing multiple or fragmented nuclei, with most ability in mammals. of these cells dying by apoptosis (Hayashi et al., 2002). To investigate the role of the SUMO system in mam- Introduction mals, we generated mice deficient for the Ubc9 protein. We find that Ubc9 is essential, and that Ubc9-deficient Posttranslational modification of proteins is an impor- embryos die early in development, subsequent to the tant regulatory mechanism of protein function. Modifica- blastocyst stage and prior to E7.5. In culture, mutant blas- tion by SUMO (small -related modifier) is un- tocysts are viable up to 2 days, but they subsequently usual in that the modifier itself is a small polypeptide. die of apoptosis. Mutant cells show chromosome de- Sumoylation has been involved in many different cell reg- fects and gross alterations in nuclear organization such ulatory processes, and the consequences of this modi- as disassembled nucleoli and PML Nuclear Bodies, mis- fication seem to be as diverse as its targets (for reviews, shapen nuclei, as well as mislocalized RanGAP1 and see Hay, 2005; Johnson, 2004; Mu¨ ller et al., 2001; Seeler Ran. Together, our findings indicate that Ubc9, and, by and Dejean, 2003). Sumoylation is a highly evolution- implication, the SUMO pathway, are crucial for proper ary conserved pathway from yeast to humans. SUMO nuclear architecture, for accurate chromosome segre- gation, and for embryonic viability.

*Correspondence: [email protected] Results 4 These authors contributed equally to this work. 5 ¨ Present address: Universitat Basel, Institut fu¨ r Biochemie und Disruption of the Murine Ubc9 Genetik, CH-4058 Basel, Switzerland. 6 Present address: Department of Pediatrics, Center for Advanced The murine Ubc9 gene consists of 7 exons that encode Biotechnology and Medicine, UMDNJ-Robert Wood Johnson Med- a 158 amino acid protein that is 100% identical to human ical School, Piscataway, New Jersey 08854. Ubc9 (Figure 1A). To inactivate the murine Ubc9 gene, Developmental Cell 770

Figure 1. Inactivation of the Ubc9 via A Probe 5' Probe 3' Cre-Mediated Gene Inversion 12 kb E E 1234567 (A) Diagram of the wild-type Ubc9 genomic Wild type allele (+) locus (+), the targeting vector, the targeted

X X X allele (t), the functional floxed allele (fl), and 5 kb 9 kb the inverted null allele (inv). The inverted allele E E E 23 4 5 6 is generated by simultaneous deletion of the Targeting vector Neo neo gene and inversion of the loxP-flanked exon 2 and 3 region. White boxes, 50 and 30 7.5 kb UTRs; black boxes, coding sequences; ar- E E E E 1 23 4 5 6 7 rowheads, loxP sites. E, EcoR1; X, Xba1; Targeted allele (t) Neo Neo, neomycin-resistant gene. The location 0 0 X X of the 5 and 3 Southern probes is given 15.3 kb along with the expected sizes in kilobases a b x PGK-Crem oligos (kb) of the hybridizing EcoR1 and Xba1 frag- c E ments. Arrows and letters indicate the posi- 1 23 4 5 6 7 Functional floxed allele (fl) tion of PCR primers used for genotyping. Matings with PGK-Crem transgenic mice and wild-type mice are indicated. (B) Southern blot analysis of wild-type (+/+)

1 23 4 5 6 7 and targeted (+/t) ES cells. EcoR1- and Xba1- Inverted null allele (inv) digested DNAs were hybridized with the 50 a b and 30 probes as shown. Expected sizes are oligos x Wild-type

c indicated in kb. 23 E 1 456 7 (C) Representative genotypic analysis of mice Stable inverted null generated by breeding Ubc9fl-inv/+/PGK-Crem allele (inv) animals with wild-type mice. DNA samples were subjected to PCR with the primer pairs a/b and a/c and Cre as shown in (A). The geno- B ES clonesC Mouse DNA D Testis proteins type of the Ubc9inv/+ mice harboring simulta- EcoR1 Xba1 a/ba/c Cre +/+ inv/+ Ubc9: neous deletion of the neo gene and stabilized +/+ +/t +/+ +/t +/+ +/t kb kb +/+ Ubc9-Sumo inversion of the loxP-flanked region sub- 15.3 sequent to the removal of the Cre transgene 12 fl fl/+ Ubc9 that were selected for the generation of + inv/inv 7.5 Ubc9 animals is boxed. The expected Tub inv/+ PCR products for the floxed (fl) and for the wild-type (+) alleles are indicated. +/+ 9 fl-inv/+ (D) Levels of Ubc9 in testis from Ubc9 and Ubc9inv/+ mice. Western blot was performed 5 +/+ with anti-Ubc9 and anti-tubulin (Tub) anti-

5' probe 5' probe 3' probe bodies as a control.

we used the flox-and-invert strategy in which the flank- Ubc9inv/+ mice harboring simultaneous deletion of the ing loxP sites are in opposite orientation, thus permitting neomycin gene and constitutive inversion of the loxP- Cre recombinase-mediated inversion. The advantage of flanked region (Figure 1C) were born at expected fre- this strategy is to provide an amenable system for the quencies, and they were viable and fertile and showed generation of mice expressing a quarter dose of Ubc9, no overt phenotype. Analysis of peripheral blood re- an approach successfully used in the context of lym- vealed an increase in both circulating lymphocytes phoid cell function (Lam and Rajewsky, 1998) and can- (Ubc9+/+, 4.03 6 0.53 3 103 cells/ml;Ubc9inv/+, 6.43 6 cer modeling (Roberts et al., 2002). The inverted allele, 1.16 3 103 cells/ml, p = 0.0007) and granulocytes designated Ubc9inv, was generated with a targeting vec- (Ubc9+/+, 1.03 6 0.54 3 103 cells/ml; Ubc9inv/+, 2.33 6 tor in which a DNA segment containing exons 2 and 3 0.93 3 103 cells/ml, p = 0.0272), as well as a slight but was flanked by two loxP sites in opposite orientation significant increase in glycemia (Ubc9+/+, 8.28 6 0.39 (Figure 1A). Three clonal Ubc9-targeted ES cell lines were mmol/l; Ubc9inv/+, 9.83 6 0.35 mmol/l, p = 0.0083). These produced, and appropriate integration was checked by alterations did not, however, compromise life span, at Southern blot (Figure 1B). When injected into blasto- least within the 2 year observation period. Western blot cysts, two of these yielded chimeras that gave germline analysis showed that the amount of both the free form transmission. Ubc9t/t homozygous males were crossed and the SUMO bound form of Ubc9 in various tissues with PGK-Crem females (Lallemand et al., 1998) that from Ubc9inv/+ animals is roughly half that expressed in express Cre from oogenesis onward to generate wild-type (Figure 1D and data not shown). These data in- Ubc9fl-inv/+/PGK-Crem animals. Since the Cre-mediated dicate that the Ubc9inv allele is null, and that no compen- inversion reaction is entirely reversible and functions satory mechanism arose in the hemizygotes. equally well in both directions, an equal ratio of floxed inverted (Ubc9inv) and noninverted (Ubc9fl) alleles will Expression from One Ubc9 Allele Is Sufficient be obtained in vivo as long as Cre is present in the cells for Normal Sumoylation (Figure 1A). To get stably inverted Ubc9 alleles, the Cre We next sought to determine if the general SUMO path- transgene was removed by crossing Ubc9fl-inv/+/PGK- way is affected in hemizygous mice. The global pattern Crem mice with wild-type animals. of SUMO-1 conjugates as detected by Western blot with Requirement of SUMO System in Mouse Development 771

A Table 1. Viability Analysis of Ubc9inv/inv Mice and Embryos +/+ inv/+ +/+ inv/+ +/+ inv/+ kDa Number with Genotypea 220 Stage +/+ inv/+ inv/inv Resorbedb Total mod. Live birth line 196 67 112 0 179 125 unmod. Live birth line 134 43 69 0 112 97 Embryos, E13.5 3 10 0 8 21 mod. Embryos, E9.5 7 16 0 12 35 * unmod. 66 * Embryos, E7.5 9 21 0 14 44 Blastocysts, E3.5 8 23 9 40 WB: Sumo-1 RanGAP1 PML Cultured blastocysts 37 96 47 180 46 a Genotypes were determined by PCR as described in Experimental Tubulin Tubulin Tubulin Procedures. b Empty deciduas were scored as ‘‘Resorbed.’’ Embryos from line 134 were selected for developmental studies. B Stresses: Control EtOH Osm Oxi Heat +/+ inv/+ +/+ inv/+ +/+ inv/+ +/+ inv/+ +/+ inv/+ kDa 220 jected to these stresses showed a similar accumulation of SUMO-2/3 conjugates in Ubc9inv/+ mice compared 125 Sumo-2/3 to wild-type animals, with heat shock showing more 97 conjugates dramatic changes than the other stresses (Figure 2B). Thus, Ubc9inv/+ mice have a normal capacity to respond 66 to acute sumoylation signals. However, it is noteworthy that, whereas the global pattern of SUMO-1 conjugates 20.5 inv/+ +/+ Free Sumo-2/3 seems unchanged between Ubc9 and Ubc9 mice 14.3 (Figure 2A), the steady-state level of SUMO-2/3 conju- inv/+ Sumo-2/3 gates appears slightly reduced in Ubc9 fibroblasts, whether subjected to stress or not, resulting in an in- 46 crease in the pool of free SUMO-2/3 in these cells (Figure Tubulin 2B). This suggests that conjugation of SUMO-2/3 to tar- Figure 2. Deletion of One Ubc9 Allele Has No Significant Effect on get proteins may be more subject to control by the Ubc9 Overall Sumoylation expression level than modification with SUMO-1. (A) Western blots for SUMO-1, RanGAP1, and PML from wild-type (+/+) and hemizygous (inv/+) mice testis lysates. The positions of Ubc9 Is Essential for Embryonic Viability unmodified and modified substrate proteins are indicated. Cross- To generate homozygous Ubc9-deficient mice, Ubc9inv/+ reacting bands with the anti-RanGAP1 and anti-PML mice were intercrossed. Of 291 live births analyzed by are marked by an asterisk. Tubulin levels were measured to ensure inv/inv equal loading. PCR, no homozygous Ubc9 mice were observed (B) Stress-induced modifications by SUMO-2/3 are similar in Ubc9+/+ when using either of the two founder lines. This finding and Ubc9inv/+ mice. Western blot of Ubc9+/+ and Ubc9inv/+ adult fibro- indicates that loss of Ubc9 leads to embryonic lethality blasts subjected to protein-damaging stimuli and blotted for SUMO- (Table 1). To determine when Ubc9inv/inv mice die in utero, 2/3 (upper panel). No stress, Control; ethanol, EtOH; osmotic stress, we analyzed embryos at embryonic days (E) 13.5, 9.5, Osm; oxidative stress, Oxi; and heat shock, Heat. The 18 kDa band in and 7.5. No homozygous mutant embryos were recov- all lanes (middle panel) corresponds to the nonconjugated pool of inv/inv SUMO-2/3. ered at any stage, indicating that Ubc9 embryos die prior to E7.5. However, the high number of deciduas- containing embryos in the final stage of resorption an anti-SUMO-1 was unchanged in Ubc9inv/+ (more than a quarter of total embryos) suggested that mice when compared to their normal littermates (Figure Ubc9inv/inv embryos survive earlier stages of develop- 2A). At the level of single SUMO substrates, such as ment and implant before degenerating. To test whether RanGAP1 and PML for which endogenous sumoylated Ubc9-deficient embryos indeed survive to the blastocyst forms are easily detectable, again, the ratio of modified stage, embryos at E3.5 were collected and genotyped. versus unmodified protein was identical in Ubc9inv/+ and Ubc9inv/inv blastocysts, which were undistinguishable wild-type animals, indicating that the sumoylation status from wild-type and heterozygous embryos (Figure 3A), of these two major substrates is not affected when were recovered at theexpected Mendelian ratio (Table 1). the level of Ubc9 expression is decreased by one-half Thus, Ubc9inv/inv embryos successfully develop to the (Figure 2A). blastocyst stage, undergo uterine implantation, and die While, in mammalian cells, most of SUMO-1 is conju- subsequently, presumably as cellular pools of mater- gated to its substrates, SUMO-2 and SUMO-3 exist pri- nally derived Ubc9 mRNA and protein decline. marily as free forms that are subject to massive conjuga- To further analyze the ability of mutant embryos to tion upon stress (Saitoh and Hinchey, 2000). To test the form blastocyst outgrowths in vitro, E3.5 blastocysts functionality of the SUMO-2/3 pathway upon cellular isolated from Ubc9inv/+ intercrosses were grown in cul- stress in hemizygous mice, we monitored the activation ture for 4 days (Figure 3A) and then genotyped. After of SUMO-2/3 conjugation in Ubc9inv/+ fibroblasts after 1 day in culture, Ubc9+/+, Ubc9inv/+, and Ubc9inv/inv blas- ethanol addition, osmotic stress, oxidative stress, and tocysts hatched from the zona pellucida and attached heat shock. Western blot analysis of cell extracts sub- onto the culture dishes, and trophoblast giant cells Developmental Cell 772

A E3.5 , t=0 Day 1 Day 2 Day 3 Day 4

TGC

+/+

ICM

TGC inv/+

ICM

TGC inv/inv

B 20

+/+ n=6 n=11

inv/+ 15 inv/inv * * C 10 Phase Contrast TUNEL Merge n=13 TGC n=7 n=7 5 n=7 n=8 n=8 n=21

ICM

Mean number of TUNEL positive cells per embryo Mean number 0 Day 1 Day 2 Day 3 Time of culture

Figure 3. Deletion of Ubc9 Leads to Apoptotic Death of Early Embryonic Cells (A) Outgrowth of Ubc9+/+, Ubc9inv/+, and Ubc9inv/inv embryos. E3.5 blastocysts from Ubc9inv/+ intercrosses were cultured in vitro for 4 days. Arrows point to the ICM, and trophoblastic giant cells (TGC) are indicated. Genotypes were determined by PCR afterward. The scale bars rep- resent 50 mm. (B) Mean number of TUNEL-positive apoptotic ICM cells per embryo. E3.5 blastocysts were isolated from Ubc9inv/+ intercrosses and cultured for 3 days. Statistical analyses (ANOVA) were performed by using the PLSD Fisher test. Error bars denote standard errors, and asterisks indicate significant difference, p < 0.01. n, number of embryos analyzed. (C) TUNEL staining of Ubc9inv/inv E3.5 blastocysts cultured for 2 days. Phase contrast, fluorescein (TUNEL), and merged images are shown. The arrow points to the ICM, and TGC is indicated. spread around a central inner cell mass (ICM). At isola- TUNEL-positive cells per embryo was low in wild-type tion and during the first day in culture, the ICM of mutant and heterozygous embryos, this number significantly blastocysts was morphologically indistinguishable from increased in mutant ICM cells after 2 days and remained that of wild-type and heterozygous blastocysts. How- high up to 3 days. Quantitation of apoptosis showed ever, at day 2 in culture, the ICM of Ubc9inv/inv embryos about 5-fold more apoptotic cells in Ubc9inv/inv ICM cells was disorganized and severely reduced in size, and vir- than in the corresponding wild-type and heterozygous tually no Ubc9inv/inv ICM cells persisted at day 4 (Figure cells at day 2 (Figure 3B). By contrast, only a few 3A). In contrast, the nondividing trophoblastic giant cells TUNEL-positive signals could be detected in the Ubc9- of Ubc9inv/inv embryos remained after 4 days in culture deficient trophoblastic cells, indicating that these cells and showed a normal appearance, suggesting that are not undergoing apoptosis (Figure 3C). We conclude they were still alive (Figure 3A). Altogether, the develop- that loss of Ubc9 ultimately results in ICM apoptotic cell mental defects of Ubc9 mutant embryos in vivo and death, whereas mutant future extraembryonic tissues in vitro indicate that the SUMO pathway is essential for escape this death pathway. growth of the ICM in the postimplantation phase of development. Loss of Ubc9 Leads to Mitotic Chromosome Defects Given the proposed role of sumoylation in chromosome Selective Apoptosis of Inner Cell Mass organization, we examined mitotic cells in E3.5 blasto- in Ubc9 Mutant Embryos cysts cultured for 2 days. The mitotic index of each em- To next investigate whether such ICM regression in bryo was determined by counting the fraction of cells Ubc9-deficient embryos was due to increased apopto- with condensed . Though the total num- sis, we performed a TUNEL assay on E3.5 blastocysts ber of cells in the mutant blastocysts was reduced by cultured for 1–3 days. While the average number of two-thirds when compared to that of the wild-type and Requirement of SUMO System in Mouse Development 773

A B C Figure 4. Analysis of Chromosome Defects 200 in Ubc9-Deficient Embryos n=17 4 400 n=19 Analysis was performed on unsynchronized n=17 350 n=56 E3.5 blastocysts cultured for 2 days and 150

) *

3 2 300 stained with DAPI. (A) Comparison of the total number of DAPI- 250 stained cells in Ubc9+ (Ubc9+/+ and Ubc9inv/+) 100 inv/inv 2 200 and Ubc9 embryos. Statistical analysis n=102 was performed by Mann-Whitney test. Error 150 n=18 bars denote standard errors, and the

* (%) Mitotic index 50 1 100 asterisk indicates significant difference, p <

Mean number of cells per embryo Mean number 0.0001. n, number of embryos analyzed.

Mean metaphase area (µm 50 (B) Mitotic index for embryos of Ubc9+ 0 0 0 (Ubc9+/+ and Ubc9inv/+) and Ubc9inv/inv geno- + inv/inv + inv/inv + inv/inv Ubc9 Ubc9 Ubc9 Ubc9 Ubc9 Ubc9 types. The mitotic index (% of total) was de- termined by counting cells with condensed DE FG chromosomes. Statistical analysis was per- formed by unpaired t test. Error bars denote standard errors; n, number of embryos ana- lyzed. The mean number of cells per Ubc9+ and Ubc9inv/inv embryo was 176 and 59, respectively. (C) Mean metaphase area (in mm2) for em- bryos of Ubc9+ (Ubc9+/+and Ubc9inv/+) and HIJ Ubc9inv/inv genotypes. The maximal area of randomly orientated metaphases was framed with rectangles and determined by Zeiss Axiovision 4.1 software. Statistical analysis was performed by using the Mann-Whitney test. Error bars denote standard errors, and the asterisk indicates significant difference, p < 0.0001. n, number of metaphases ana- lyzed. 41.6% vs. 16.6% 25% vs. 2.4% (D–G) DAPI-stained mitotic cells. (D) and (E) are wild-type embryos. (F) and (G) are KLM Ubc9inv/inv embryos. (D) and (F) show meta- phase, and (E) and (G) show anaphase. The scale bar represents 5 mm. (H–M) Chromosome spreads. (H) Typical chromosome spread from Ubc9+ embryos. (I–M) Examples of chromosome defects in Ubc9inv/inv embryos: (I) hypocondensation, (J) breaks (closed arrowheads) or sister chro- 20.8% vs. 4.8% 29.2% vs. 2.4% 45% vs. 2.4% matid separation (open arrowheads), (K) chromosome loss, (L) polyploidy, (M) multiple defects: polyploidy, breaks (closed arrow- heads), and chromatid separation (open arrowheads). The mean chromatid width was significantly higher in mutant compared to wild-type cell spreads (1.340 6 0.27 mm versus 1.108 6 0.26 mm, Mann-Whitney Test, p = 0.002). Metaphase chromosomes displaying a width greater than 1.372 mm, corresponding to the wild-type mean plus 1 SD width, were scored as hypocondensed. For Ubc9+, 42 cells from 14 blastocysts were scored. For Ubc9inv/inv, 24 cells from 12 blastocysts were scored. Ten chromatids per metaphase, from different chromosomes, were mea- sured to produce a chromatid mean width. Percentages of Ubc9inv/inv versus Ubc9+ for each defect are given with each panel. The scale bar rep- resents 10 mm. heterozygous embryos (Figure 4A), the mitotic index 4K). Furthermore, approximately one-third of the mutant was almost the same (Figure 4B). In wild-type ICM cells, cells exhibited polyploidy (Figure 4L), and almost half mitosis appeared normal at metaphase (Figure 4D), and of the cells displayed more than one of the above- chromosome masses separated properly in all ana- described defects (Figure 4M). The elevated frequency phase cells (Figure 4E). By contrast, two major types of mitotic defects in Ubc9-deficient ICM cells indicates of chromosome defects were observed in the mutant that sumoylation is critical for structure as well as for cells: (1) a significant increase in the size of nearly all proper segregation and inheritance of chromosomes. metaphase plates (Figures 4C and 4F), and (2) aberrantly segregating chromosomes in anaphase characterized Nuclear Organization Defects in Ubc9 Mutant Cells by the presence of anaphase bridges in w10% of mitotic We expected that the absence of Ubc9 would have an cells (Figure 4G). Such aberrant anaphase chromo- effect on the subcellular localization of the SUMO pro- somes were never observed in Ubc9inv/+ and Ubc9+/+ teins. In wild-type cells, SUMO-1, which is mostly conju- cells. Examination of DAPI-stained mitotic chromosome gated to its substrates, displayed its typical nuclear dif- spreads from wild-type and mutant blastocysts blocked fuse and speckled localization (Figures 5A and 5B). By with nocodazole further revealed that mutant cells har- contrast, in Ubc9-deficient cells, SUMO-1 became dif- bor significantly elevated numbers of chromosomes that fusely distributed both in the nucleus and the cytoplasm, are hypocondensed (Figure 4I), broken, or, in which sis- as expected for nonconjugated SUMO-1 (Figures 5C ter chromatids are separated (Figure 4J), or lost (Figure and 5D). A similar mislocalization of SUMO-1 has been Developmental Cell 774

+/+ inv/inv Figure 5. Abnormal Nuclear and Subnuclear Ubc9 Ubc9 Architecture in Ubc9-Deficient Cells Antibody Merge Antibody Merge Microscopy analysis of Ubc9+/+ and Ubc9inv/inv A B C D E3.5 blastocysts cultured for 2 days. (A–T) Cells were revealed with antibodies for the indicated nuclear proteins and merged with DAPI: (A–D) SUMO-1, (E–H) Lamin B1, (I–L) fibrillarin, (M–P) PML, and (Q–T) Daxx.

SUMO-1 In (G), note the dense foci of lamin B (arrow), and in (H), note the condensed chromatin masses that were not surrounded by a visible E FGH lamina (arrowhead). In (L), note the cap-like structures dissociated from condensed chro- matin areas (arrows). (U and V) Images from phase contrast micros- copy. The scale bar represents 10 mm. Lamin B1

I JKL

Fibrillarin

M NOP PML

Q RST Daxx

U V

observed in cells overexpressing subdomains of the for the PML Nuclear Bodies. When compared to the Ubc9-interacting RanBP2 nucleoporin (Saitoh et al., typical large nucleoli present in the wild-type cells, in 2002). more than 90% of mutant cells, fibrillarin is distributed To further assess the effect of Ubc9 ablation at the cel- in several foci, suggesting that functional nucleoli are not lular level, we examined the integrity of specific nuclear present (Figures 5I–5L). The cap distribution close to the and subnuclear structures in Ubc9+/+ and Ubc9inv/inv DAPI-negative structure indicates fragmented nucleolar embryos. Nuclear morphology was investigated by us- structures (arrows in Figure 5L). This disorganization, ing an anti-lamin B1 antibody. When compared to their which was further confirmed by phase contrast micros- wild-type counterparts, Ubc9-deficient cells exhibited copy (Figures 5U and 5V), is reminiscent of the nucleolar oversized interphase nuclei, which appeared to be highly disassembly that occurs when rDNA transcription is in- dysmorphic and showed invaginations, herniations, hibited (Hernandez-Verdun et al., 2002). Similarly, the and blebbing (Figures 5E–5H). In a subset of cells, con- PML Nuclear Bodies were disrupted in Ubc9inv/inv cells densed chromatin masses were present that were not (Figures 5M–5P). This feature recalls the pattern ob- surrounded by a visible lamina (arrowhead in Figure 5H). served in PML2/2 cells overexpressing a sumoylation- In these cells, lamin B1 labeling was either diffuse in the deficient PML protein (Zhong et al., 2000). To further cytoplasm or present in dense foci, evoking lamin ag- confirm the complete disruption of the PML Nuclear gregation (arrow in Figure 5G). Bodies, we analyzed the localization of the Daxx protein, We then used two other markers to study the fate a transcriptional repressor that is recruited to the Nu- of well-characterized subnuclear structures in Ubc9- clear Bodies via its association with SUMO-modified deficient cells, namely, fibrillarin for the nucleoli and PML PML (Ishov et al., 1999). In Ubc9+/+ cells, Daxx localized Requirement of SUMO System in Mouse Development 775

ICM RanGAP1 ICM Merge TGC RanGAP1 TGC Merge Figure 6. Alteration in RanGAP1 and Ran A B E F Distribution in Ubc9-Deficient Cells (A–P) Microscopy analysis of Ubc9+/+ and Ubc9+/+ Ubc9inv/inv E3.5 blastocysts cultured for 2 days. Cells were stained with antibodies for either (A, C, E, and G) RanGAP1 and merged with DAPI stain for (B, D, F, and H) nuclei or C D G H (I, K, M, and O) Ran and merged with DAPI stain for (J, L, N, and P) nuclei. Photos were Ubc9inv/inv taken from a single Ubc9+/+ or Ubc9inv/inv em- bryo at two different focal planes corre- sponding to the ICM (left panels) and TGC (right panels). The arrows in (G) point to ICM Ran ICM Merge TGC Ran TGC Merge Ubc9inv/inv TGC cells, which still showed I J M N association of RanGAP1 with the nuclear en- velope. The scale bar represents 20 mm. Ubc9+/+

K L O P

Ubc9inv/inv

specifically in the Nuclear Bodies together with PML embryos, except in the few trophoblastic giant cells that (Figure 5Q and data not shown). However, in Ubc9inv/inv retained accumulation of RanGAP1 at the NPC (Figures cells, Daxx was detected in a nuclear diffuse and patchy S1E–S1H). Given that the maternal pool of Ubc9 de- distribution characteristic of its condensed chromatin lo- creased to the same extent in the ICM and the tropho- cation (Figures 5S and 5T), similar to what was observed blastic Ubc9inv/inv cells, this suggests that the persis- in MEFs disrupted for PML Nuclear Bodies (Ishov et al., tence of RanGAP1 together with Ubc9 at the nuclear 1999). Altogether, these data indicate that impairment envelope in some of the trophoblastic cells likely results of the SUMO pathway induces drastic abnormalities in from a lower rate in nuclear envelope breakdown rather nuclear and subnuclear architecture that are likely detri- than from a slower loss of maternal Ubc9 in these cells. mental for cell survival. Thus, SUMO modification may be actively required for proper targeting of RanGAP1 upon nuclear envelope Mislocalization of RanGAP1 and Ran reformation. in Ubc9 Mutant Cells An essential factor in protein trafficking through the The first identified substrate for SUMO was RanGAP1, NPC is the Ran protein itself, which plays a key role in de- the GTPase-activating protein for Ran, which is required termining the bidirectionality of nuclear transport (Gor- for the bidirectional transport of proteins across the lich and Mattaj, 1996; Pemberton and Paschal, 2005). nuclear pore complex. Given that the association of To examine whether the failure of RanGAP1 to accumu- RanGAP1 with the nuclear pore complex (NPC) RanBP2 late at the NPC is associated with a perturbation in Ran protein was shown to depend on RanGAP1 sumoylation localization, we analyzed the subcellular distribution of (Mahajan et al., 1998; Matunis et al., 1998), we examined Ran in Ubc9 mutant cells. Whereas Ran was found to +/+ the consequences of the impairment of sumoylation on be predominantly located in the nucleus in Ubc9 cells RanGAP1 localization. Immunofluorescence micros- (Figures 6I and 6M), it was broadly distributed through- copy on Ubc9+/+ ICM cells and trophoblastic giant cells out the nucleus and the cytoplasm of all ICM and most of inv/inv predominantly revealed an intense nuclear rim staining the trophoblastic Ubc9 cells (Figures 6K and 6O). demonstrating the association of RanGAP1 with the nu- Given the central role of the Ran gradient in regulating clear envelope (Figures 6A, 6B, 6E, and 6F). As could be protein transport across the nuclear envelope, the high expected, in the null embryos, RanGAP1 remained dif- levels of cytosolic Ran in Ubc9-deficient cells strongly fusely distributed throughout the cytosol of ICM cells predict that these cells suffer profound defects in nucle- and failed to accumulate at the NPC (Figures 6C and ocytoplasmic trafficking. 6D). It is noteworthy that, while the mislocalization of RanGAP1 is evident in all cells from the ICM, some of Discussion the trophoblastic giant Ubc9inv/inv cells showed a nuclear rim staining (Figures 6G and 6H), indicating that these The SUMO modification pathway is a key regulator of nu- cells retained their ability to concentrate RanGAP1 at merous cellular activities, including transcriptional reg- the nuclear envelope. Consistent with previous findings ulation, nuclear transport, maintenance of genome in- (Zhang et al., 2002), Ubc9 was found to partially concen- tegrity, and signal transduction (Hay, 2005; Johnson, trate at the rim of the nuclear envelope in the wild-type 2004; Mu¨ ller et al., 2001; Seeler and Dejean, 2003). embryos (Figures S1B and S1D; see the Supplemental Here, we provide evidence that the SUMO system is es- Data available with this article online). By contrast, the sential for early mammalian development. Its involve- low amount of remaining maternal Ubc9 was dispersed ment in chromosome structure and function is demon- in the cytoplasm of every cell from Ubc9-deficient E5.5 strated by condensation defects during metaphase and Developmental Cell 776

defective anaphase movements. The major alterations sion at centromeric regions. Excess in sumoylation of in nuclear membrane morphology and subnuclear struc- topoisomerase II may in part account for these pheno- tures reveal a major role for SUMO modification in nu- types (Bachant et al., 2002). Similar conclusions were clear organization. Moreover, perturbation in RanGAP1 reached with a dominant-negative mutant of Ubc9 in and Ran distribution is predicted to severely affect nu- Xenopus egg extracts (Azuma et al., 2003). Most of the clear transport. Our results also show that loss of Ubc9 chromosome segregation defects associated with im- leads to apoptosis of the ICM. Therefore, Ubc9 and, by pairment of the SUMO pathway in lower eukaryotic cells implication, the SUMO pathway, is critical for early em- are associated with diverse phenotypes, including bryonic development, and lethality is most likely caused hypersensitivity to DNA damage, telomere length elon- by a combination of these defects. gation, and recombination defects. It remains to be de- termined whether other chromosome processes, in ad- Chromosome Defects in Ubc9-Deficient Cells dition to chromosome condensation and segregation, Several lines of evidence have substantiated a role for are also affected by Ubc9 loss in mammalian cells. SUMO modification in chromosome structure and func- tion, but the molecular basis of these effects is largely Loss of Sumoylation Leads to Abnormal unknown. Anaphase segregation defects of chromo- Nuclear Architecture somes as well as aneuploidy in Ubc9inv/inv ICM cells Lack of SUMO modification disrupted the structure of demonstrate a role for sumoylation in chromosome in- the nuclear envelope, producing grossly misshapen nu- heritance, whereas its involvement in chromosome clei with multiple large blebs and abnormal chromatin structureisillustratedbyhypocondensationduringmeta- distribution. Such alterations of the nuclear peripheral phase. These findings contrast with a previous study shape are reminiscent of the changes observed in lamin in which inactivation of Ubc9 in chicken DT40 cells failed B1-deficient murine embryo fibroblasts (Vergnes et al., to induce chromosome missegregation. Rather, cells 2004). A similar budding of the nuclear envelope can also with multiple nuclei suggested defects in cytokinesis be seen in cells committed to apoptosis (Buendia et al., (Hayashi et al., 2002). The elevated incidence of poly- 1999; Kihlmark et al., 2001). In cells programmed to die, ploidy we observe in Ubc9inv/inv cells may similarly reflect caspase-dependent cleavage of lamins has been con- a cytokinesis defect. Whether the absence of chromo- sidered as a prelude to nuclear breakdown, and morpho- some missegregation in chicken DT40 cells reflects a dif- logical modification of the nuclear envelope is recog- ference between embryonic cells and a transformed cell nized as a hallmark of apoptosis (Lazebnik et al., 1995). line is not known. Nevertheless, a very similar phenotype Thus, perturbation of the nuclear envelope in Ubc9- to that observed in murine Ubc9-deficient cells, includ- deficient cells may simply be the manifestation of apo- ing chromosome condensation defects and anaphase ptotic cell death. Alternatively, the state of lamin assem- bridges, was reported in D. melanogaster mutated for bly may also be involved in triggering apoptosis, as inhi- the Su(var)2-10 (PIAS homolog) gene (Hari et al., 2001). bition of the assembly of lamin B induces cell death In addition, S. pombe strains disrupted for the pmt3, (Steen and Collas, 2001). Whether lack of sumoylation hus5, and pli1 genes (SUMO, Ubc9, and PIAS homo- may result in a failure to correctly assemble the nuclear logs) show an increase in chromosome missegregation lamina remains to be determined. Moreover, recent find- (al-Khodairy et al., 1995; Tanaka et al., 1999; Xhemalce ings have shown that Ran and its cofactors RanBP2, et al., 2004). Similarly, inactivation of the ulp2/smt4 RanGAP1, and RCC1 are all essential for nuclear enve- SUMO isopeptidase gene in S. cerevisiae leads to chro- lope assembly in vivo (Askjaer et al., 2002; Bamba et al., mosome segregation defects and altered chromosome 2002). Thus, mislocalization of RanGAP1 and Ran in condensation. Notably, the mitotic-specific targeting of Ubc9-depleted cells might also contribute to the mor- the condensin complex to chromatin, in particular rDNA phological abnormalities of the nuclear envelope. chromatin, is impaired in ulp2 mutant strains (Strunnikov Our results also show that disruption of the SUMO et al., 2001). Given that altering either SUMO conjugation pathway results in nucleolar disassembly. Known for or deconjugation leads to similar chromosome pheno- several decades as the ribosome factory, the types, it is likely that a modification/demodification equi- was recently shown to be also involved in several other librium must be maintained for proper chromosome cellular processes, including chromatin architecture, in- structure and function. tranuclear trafficking, and control (reviewed in A possible role for sumoylation in chromosome con- Hernandez-Verdun et al., 2002). A possibility is that dis- densation is further supported by the recent finding ruption of nucleoli observed in Ubc9-deficient cells is that sumoylation of the budding yeast Ycs4 condensin a direct consequence of apoptotic cell death. However, subunit is stimulated by Cdc14, a phosphatase required a direct link between apoptosis and nucleolar break- for rDNA segregation, raising the possibility that Cdc14 down has not yet been established. If nucleolar disas- targets condensin to the rDNA by inducing sumoylation sembly can be seen in some apoptotic systems (Baran of one of its subunits (D’Amours et al., 2004). Moreover, et al., 2003), most studies have shown that both nucleo- Pds5, a protein important for sister chromatid cohesion lar size and density do not change under apoptosis, and condensation was found to be sumoylated, with so that ‘‘on the whole, the nucleolus is regarded as a nu- SUMO modification increasing during mitosis. Overex- clear domain very resistant to apoptosis’’ (Martelli et al., pression of Ulp2 suppresses the temperature sensitivity 2001). Apoptotis, therefore, seems an unlikely expla- of pds5 mutants, suggesting that Pds5 sumoylation pro- nation for nucleolar disruption in Ubc9 mutant cells. motes the dissolution of cohesion (Stead et al., 2003). The maintenance of the nucleolar organization relies on Conversely, in fission yeast, loss of Ulp2 leads to prema- the equilibrium between transcription, processing, and ture centromere separation and failure to maintain cohe- export of the ribosomal subunits, and inactivation of one Requirement of SUMO System in Mouse Development 777

of these processes ultimately leads to nucleolar break- is involved. The collapse in the Ran gradient we observed down. It is thus very likely that nucleolar disassembly in Ubc9-deficient cells is expected to result in severe de- observed in Ubc9-deficient cells rather reflects global fects in nuclear trafficking. defects in one or several of the events involved in ribo- Another possible function of RanGAP1 sumoylation is some production. Also, it is of note that several nucleolar its participation in the mitotic functions of Ran. During proteins, including RNA Pol-I, fibrillarin, nucleophosmin, mitosis, RanGAP1 is diffusely distributed throughout and some ribosomal proteins, were identified as puta- the cell, although a significant fraction localizes to the tive SUMO susbstrates in proteome-wide studies (Panse mitotic spindles with accumulation at the kinetochores et al., 2004; Rosas-Acosta et al., 2005). It is thus possible (Joseph et al., 2002; Matunis et al., 1996). Because that lack of sumoylation of these targets plays a role in SUMO modification is essential for RanGAP1 to be as- nucleolar defects that occur in Ubc9-depleted embryos. sociated with spindles (Joseph et al., 2002), it is also A major SUMO substrate, the PML protein, is a central tempting to speculate that Ubc9-deficient cells exhibit structural component of the PML Nuclear Bodies, the in- centromeric defects. Consistent with this notion, we tegrity of which is compromised in certain human dis- observed that, when compared to control cells, the en- eases (Borden, 2002). We show that lack of sumoylation richment of RanGAP1 at kinetochores was absent in leads to a drastic mislocalization of PML and Daxx, Ubc9-deficient cells that had been fixed without digito- indicative of the disruption of its associated Nuclear nin permeabilization (unpublished data). Bodies. These data are consistent with previous obser- The studies described here demonstrate that sumoy- vations made in PML2/2 cells in which expression of lation contributes one or several essential function(s) for a sumoylation-deficient PML mutant fails to induce the early mammalian development. Impairment of this pro- formation of Nuclear Bodies (Zhong et al., 2000). Alto- cess causes chromosome missegregation together with gether, these data demonstrate that sumoylation is es- major abnormalities in nuclear organization that results sential for formation of the Nuclear Bodies. An intriguing in apoptotic cell death. It is likely that the loss of nuclear point is that many Nuclear Body-associated proteins are architecture integrity and expected defects in nuclear SUMO modified, but whether, as for PML, sumoylation transport contribute to cell death. Definitive resolution is a requirement for Nuclear Body association is less of these issues will await the development of conditional clear (Seeler and Dejean, 2003). While these structures inactivation systems from which appropriate material for are not required for viability in mice (Wang et al., 1998), detailed mechanistic studies can be derived. they have been broadly involved in the control of cell growth, transcription, and apoptosis (Borden, 2002). It is thus tempting to speculate that disruption of Nuclear Experimental Procedures Bodies and/or the subsequent release of nonsumoylated Gene Targeting and Generation of Mutant Mice associated factors in Ubc9-deficient embryos have im- Ubc9 genomic clones were isolated from a mouse 129/Sv genomic portant effects on these fundamental processes. In library. To use the ‘‘flox-and-invert’’ strategy (Lam and Rajewsky, this regard, the fact that Daxx is delocalized from the Nu- 1998; Roberts et al., 2002), an inverted SpeI-flanked loxP sequence clear Bodies in the absence of sumoylation may be par- was inserted into an Xba I site upstream of exon 2, consequently de- ticularly relevant because it has been implicated in apo- stroying this site. CJ7 embryonic stem (ES) cells were electropo- rated, and transformed cells that survived G418 selection were ptosis (Salomoni and Pandolfi, 2002). screened for homologous recombination by Southern blotting by using a 50,a30, and a Neo probe (see below). Three positive ES Implications for Nuclear Transport cell clones were expanded and injected separately into C57BL6/J in Ubc9-Deficient Cells blastocysts to generate chimeras. Chimeras were crossed with C57BL6/J females to test the germline transmission of the Ubc9 tar- The first identified target for sumoylation, RanGAP1, pro- t vided a first link between nuclear transport and SUMO. geted allele (Ubc9 ). Two Ubc9 homologous recombinant founder lines were obtained (L134 and L196). Ubc9t/t males were mated with Conjugation to SUMO was shown to target cytosolic PGK-Crem females (Lallemand et al., 1998) to generate Ubc9fl-inv/+/ RanGAP1 to the NPC RanBP2 protein, and this asso- PGK-Crem mice carrying both deletion of the neomycin transgene ciation is essential for nuclear trafficking (Mahajan and reversible inversion of exons 2 and 3. These mice were mated et al., 1998; Matunis et al., 1998). Whether sumoylated to wild-type females in order to cross out the Cre allele. Ubc9inv/+ heterozygous mice carrying stabilized inverted allele were then in- RanGAP1 has to sit at the NPC for efficient transport re- inv/inv mains unclear, but the capacity of antibodies to block tercrossed to obtain Ubc9 embryos. nuclear import by inhibiting pore-associated RanGAP1 Genotyping (Mahajan et al., 1997) strongly supports this idea. ES Cells, Embryos, and Animals RanGAP1 is known to stimulate the GTPase activity of For Southern blot analysis, genomic DNA from cultured ES cells was Ran, thus accelerating the conversion of RanGTP to digested with EcoRI or XbaI and was hybridized with the indicated 0 0 RanGDP. A striking feature of the Ran system is the 5 ,3 (Figures 1A and 1B), or Neo probes. For PCR analysis, genomic asymmetric distribution of its constituents, including Ran DNA from tail clippings or whole embryos were used as template, primer ‘‘a’’ was used either with primer ‘‘b’’ to amplify the wild- itself, between nucleus and cytoplasm. Ran is predomi- type (+) (459 bp) and the floxed (fl) allele (537 bp), or with primer nantly found in the nucleus, and a steep gradient exists ‘‘c’’ to amplify the inverted (inv) allele (440 bp). (Figure 1A). Primers for RanGTP across the nuclear envelope (Gorlich and and conditions for nested PCR are available upon request. Mattaj,1996; Pemberton and Paschal, 2005). The changes we observed in Ran localization suggest that loss of Immunoblotting sumoylation may alter Ran cycling. One possibility is Cultured adult ear fibroblasts or freshly dissected testes were ex- tracted in buffer (50 mM Tris HCl [pH 7.5], 150 mM NaCl, 10 mM that the absence of RanGAP1 at the NPC would account EDTA, 0.5% sodium deoxycholate, 1% Nonidet P40, 0.1% SDS, for the increase in cytosolic RanGTP. It is also possible 20 mM N-ethyl maleimide [NEM], 1 mM dithiothreitol, 13 Protease that the lack of sumoylation of other transport factors inhibitor cocktail tablets [Complete, Roche]). Western blots were Developmental Cell 778

probed with antibodies against SUMO-1 (Zymed), SUMO-2/3 (rabbit ler and O. Bischof for critical reading of the manuscript. This work polyclonal antibody provided by M. Matunis), RanGAP1 (goat serum was supported by grants from the Association pour la Recherche provided by F. Melchior), PML (mouse polyclonal serum, provided sur le Cancer, the Fondation de France, the Pasteur-Negri-Weiz- by P.G. Pelicci), or a-tubulin for loading controls (Santa Cruz). mann Council, and the European Commission Framework Progam 6. K.N. was supported by the Ministe` re de la Recherche et la Tech- Culture of Primary Adult Fibroblasts nologie and Ligue Nationale Contre le Cancer. Adult fibroblasts were established from ear skin biopsies from +/+ inv/+ Ubc9 or Ubc9 mice as described (Me´ met et al., 1997). Stress Received: March 21, 2005 and drug treatments were performed as described (Saitoh and Hin- Revised: July 21, 2005 chey, 2000). Stimuli were: heat shock (10 min at 43ºC), osmotic Accepted: October 13, 2005 stress (normal medium plus 0.7 M NaCl for 15 min at 37ºC), oxidative Published: December 5, 2005 stress (100 mM H2O2 for 20 min at 37ºC), or exposure to 7% ethanol (20 min at 37ºC). Control samples were left untreated. References Blastocyst Recovery and Outgrowth Cultures al-Khodairy, F., Enoch, T., Hagan, I.M., and Carr, A.M. (1995). The 4- to 6-week-old heterozygous Ubc9inv/+ females were superovu- Schizosaccharomyces pombe hus5 gene encodes a ubiquitin conju- lated (5 IU of pregnant mare serum gonadotropin [Intervet Lab.], fol- gating enzyme required for normal mitosis. J. Cell Sci. 108, 475–486. lowed 48 hr later by 5 IU of human chorionic gonadotropin [Intervet Lab.]) and mated with heterozygous Ubc9inv/+ males. Females pos- Apionishev, S., Malhotra, D., Raghavachari, S., Tanda, S., and itive for the presence of the vaginal plug were sacrificed at 3.5 days Rasooly, R.S. (2001). The Drosophila UBC9 homologue lesswright postcoitus, and blastocysts were collected by uterine flush and mediates the disjunction of homologues in I. Genes Cells grown on 0.1% gelatin-coated chambered slides (Lab-Tek) in ES 6, 215–224. cell medium with antibiotics without leukemia inhibitory factor. Em- Askjaer, P., Galy, V., Hannak, E., and Mattaj, I.W. (2002). Ran GTPase bryo morphology was observed by phase contrast microscopy, and cycle and importins a and b are essential for spindle formation and blastocyst outgrowths were photographed each day. At the indi- nuclear envelope assembly in living Caenorhabditis elegans em- cated times, outgrowths were scraped off and lysed for genotyping bryos. Mol. Biol. Cell 13, 4355–4370. as described above. Azuma, Y., Arnaoutov, A., and Dasso, M. (2003). SUMO-2/3 regu- lates topoisomerase II in mitosis. J. Cell Biol. 163, 477–487. Pub- TUNEL Assay and Immunofluorescence Studies lished online November 3, 2003. 10.1083/jcb.200304088. DNA fragmentation-associated cell death of cultured blastocysts Bachant, J., Alcasabas, A., Blat, Y., Kleckner, N., and Elledge, S.J. was detected by the TdT-mediated dUTP-fluorescein nick end label- (2002). The SUMO-1 isopeptidase Smt4 is linked to centromeric ing (TUNEL) assay by using the In Situ Cell Death Detection kit, Pod cohesion through SUMO-1 modification of DNA topoisomerase II. (Roche), and the manufacturer’s protocol. Embryos were observed Mol. Cell 9, 1169–1182. by phase contrast and fluorescence microscopy; green-stained nu- clei were counted as TUNEL-positive cells. For immunofluorescence Bamba, C., Bobinnec, Y., Fukuda, M., and Nishida, E. (2002). The analysis, embryos were fixed in 4% PFA for 10 min; permeabilized GTPase Ran regulates chromosome positioning and nuclear enve- with 0.25% Triton X-100 for 10 min; incubated with antibodies lope assembly in vivo. Curr. Biol. 12, 503–507. against lamin B1 (affinity-purified rabbit serum, provided by J.C. Baran, V., Fabian, D., Rehak, P., and Koppel, J. (2003). Nucleolus in Courvalin), fibrillarin (purified human autoimmune serum A68, pro- apoptosis-induced mouse preimplantation embryos. Zygote 11, vided by D. Hernandez-Verdun), SUMO-1 (Zymed), PML (Upstate), 271–283. Daxx (Santa Cruz), RanGAP1 (goat serum provided by F. Melchior), Borden, K.L. (2002). Pondering the promyelocytic leukemia protein and Ran (BD Transduction Laboratories); and then incubated with (PML) puzzle: possible functions for PML nuclear bodies. Mol. Cell. appropriate secondary antibodies (Biorad, Molecular probes). Slides Biol. 22, 5259–5269. were mounted in VectaShield containing DAPI (Vector Laboratories). Brown, E.J. (2004). Analysis of cell cycle progression and genomic Structured illumination fluorescent images (optical sections) were integrity in early lethal knockouts. Methods Mol. Biol. 280, 201–212. acquired with a Zeiss Axiocam HRc camera (Zeiss Axiovision 4.1 software) on an Axioplan 2 imaging microscope-equipped Buendia, B., Santa-Maria, A., and Courvalin, J.C. (1999). Caspase- ApoTome system (Zeiss). After TUNEL assay or immunofluores- dependent proteolysis of integral and peripheral proteins of nuclear cence analysis, embryos were recovered and lysed for genotyping. membranes and nuclear pore complex proteins during apoptosis. J. Cell Sci. 112, 1743–1753. Preparation of Mitotic Spreads D’Amours, D., Stegmeier, F., and Amon, A. (2004). Cdc14 and con- E3.5 blastocysts were cultured for 42 hr, treated with 2.5 mM noco- densin control the dissolution of cohesin-independent chromosome dazole for 6 hr, processed for preparation of mitotic spreads as de- linkages at repeated DNA. Cell 117, 455–469. scribed by Brown (2004), and modified as follows: step 2, blastocyst Epps, J.L., and Tanda, S. (1998). The Drosophila semushi mutation cultures were trypsinized by adding 30 ml 0.5% trypsin/1 mM EDTA blocks nuclear import of bicoid during embryogenesis. Curr. Biol. and incubated at 37ºC for 15 min; and step 7, cells were rinsed once 8, 1277–1280. in 200 ml 75 mM KCl before hypotonic shock to allow for the com- Gorlich, D., and Mattaj, I.W. (1996). Nucleocytoplasmic transport. plete removal of PBS salt. Images were taken as described above. Science 271, 1513–1518. A total of 50 ml cell suspension from step 4 was washed twice with PBS and lysed for genotyping. Hari, K.L., Cook, K.R., and Karpen, G.H. (2001). The Drosophila Su(var)2–10 locus regulates chromosome structure and function and encodes a member of the PIAS . Genes Dev. 15, Supplemental Data 1334–1348. Hay, R.T. (2005). SUMO: a history of modification. Mol. Cell 18, 1–12. Supplemental Data including one figure are available at http:// www.developmentalcell.com/cgi/content/full/9/6/769/DC1/. Hayashi, T., Seki, M., Maeda, D., Wang, W., Kawabe, Y., Seki, T., Saitoh, H., Fukagawa, T., Yagi, H., and Enomoto, T. (2002). Ubc9 is essential for viability of higher eukaryotic cells. Exp. Cell Res. Acknowledgments 280, 212–221. We acknowledge Y. Lallemand, S. Lowe, M. Matunis, F. Melchior, Hernandez-Verdun, D., Roussel, P., and Gebrane-Younes, J. (2002). and P.G. Pelicci for providing us with reagents; A. de La Coste for Emerging concepts of nucleolar assembly. J. Cell Sci. 115, 2265– help in FACS analysis; N. Martin for sumoylation assays; and the 2270. dynamic imaging platform for assistance. We thank F. Coumailleau, Ishov, A.M., Sotnikov, A.G., Negorev, D., Vladimirova, O.V., Neff, N., J.-C. Courvalin, D. Hernandez, A. Lamond, V. Legagneux, I. Mattaj, Kamitani, T., Yeh, E.T., Strauss, J.F., 3rd, and Maul, G.G. (1999). PML and M.-G Mattei for advice and helpful discussions as well as J. See- is critical for ND10 formation and recruits the PML-interacting Requirement of SUMO System in Mouse Development 779

protein daxx to this nuclear structure when modified by SUMO-1. 56–72. Published online November 30, 2004. 10.1074/mcp. J. Cell Biol. 147, 221–234. M400149-MCP200. Johnson, E.S. (2004). Protein modification by SUMO. Annu. Rev. Sachdev, S., Bruhn, L., Sieber, H., Pichler, A., Melchior, F., and Biochem. 73, 355–382. Grosschedl, R. (2001). PIASy, a nuclear matrix-associated SUMO Johnson, E.S., and Gupta, A.A. (2001). An E3-like factor that pro- E3 , represses LEF1 activity by sequestration into nuclear motes SUMO conjugation to the yeast septins. Cell 106, 735–744. bodies. Genes Dev. 15, 3088–3103. Jones, D., Crowe, E., Stevens, T.A., and Candido, E.P. (2002). Func- Saitoh, H., and Hinchey, J. (2000). Functional heterogeneity of small tional and phylogenetic analysis of the ubiquitylation system in ubiquitin-related protein modifiers SUMO-1 versus SUMO-2/3. Caenorhabditis elegans: ubiquitin-conjugating enzymes, ubiquitin- J. Biol. Chem. 275, 6252–6258. activating enzymes, and ubiquitin-like proteins. Genome Biol. 3, Saitoh, H., Pizzi, M.D., and Wang, J. (2002). Perturbation of SUMOla- RESEARCH0002. tion enzyme Ubc9 by distinct domain within nucleoporin RanBP2/ Nup358. J. Biol. Chem. 277, 4755–4763. Joseph, J., Tan, S.H., Karpova, T.S., McNally, J.G., and Dasso, M. (2002). SUMO-1 targets RanGAP1 to kinetochores and mitotic spin- Salomoni, P., and Pandolfi, P.P. (2002). The role of PML in tumor dles. J. Cell Biol. 156, 595–602. suppression. Cell 108, 165–170. Kagey, M.H., Melhuish, T.A., and Wotton, D. (2003). The Polycomb Seeler, J.S., and Dejean, A. (2003). Nuclear and unclear functions of Protein Pc2 is a SUMO E3. Cell 113, 127–137. SUMO. Nat. Rev. Mol. Cell Biol. 4, 690–699. Kihlmark, M., Imreh, G., and Hallberg, E. (2001). Sequential degrada- Seufert, W., Futcher, B., and Jentsch, S. (1995). Role of a ubiquitin- tion of proteins from the nuclear envelope during apoptosis. J. Cell conjugating enzyme in degradation of S- and M-phase cyclins. Sci. 114, 3643–3653. Nature 373, 78–81. Lallemand, Y., Luria, V., Haffner-Krausz, R., and Lonai, P. (1998). Ma- Stead, K., Aguilar, C., Hartman, T., Drexel, M., Meluh, P., and Guacci, ternally expressed PGK-Cre transgene as a tool for early and uni- V. (2003). Pds5p regulates the maintenance of sister chromatid co- form activation of the Cre site-specific recombinase. Transgenic hesion and is sumoylated to promote the dissolution of cohesion. Res. 7, 105–112. J. Cell Biol. 163, 729–741. Published online November 17, 2003. 10.1083/jcb.200305080. Lam, K.P., and Rajewsky, K. (1998). Rapid elimination of mature autoreactive B cells demonstrated by Cre-induced change in Steen, R.L., and Collas, P. (2001). Mistargeting of B-type lamins at B cell antigen receptor specificity in vivo. Proc. Natl. Acad. Sci. the end of mitosis: implications on cell survival and regulation of USA 95, 13171–13175. lamins A/C expression. J. Cell Biol. 153, 621–626. Strunnikov, A.V., Aravind, L., and Koonin, E.V. (2001). Saccharomy- Lazebnik, Y.A., Takahashi, A., Moir, R.D., Goldman, R.D., Poirier, ces cerevisiae SMT4 encodes an evolutionarily conserved protease G.G., Kaufmann, S.H., and Earnshaw, W.C. (1995). Studies of the with a role in chromosome condensation regulation. Genetics 158, lamin proteinase reveal multiple parallel biochemical pathways dur- 95–107. ing apoptotic execution. Proc. Natl. Acad. Sci. USA 92, 9042–9046. Takahashi, Y., Kahyo, T., Toh, E.A., Yasuda, H., and Kikuchi, Y. Mahajan, R., Delphin, C., Guan, T., Gerace, L., and Melchior, F. (2001). Yeast Ull1/Siz1 is a novel SUMO1/Smt3 ligase for septin com- (1997). A small ubiquitin-related polypeptide involved in targeting ponents and functions as an adaptor between conjugating enzyme RanGAP1 to nuclear pore complex protein RanBP2. Cell 88, 97–107. and substrates. J. Biol. Chem. 276, 48973–48977. Mahajan, R., Gerace, L., and Melchior, F. (1998). Molecular charac- Tanaka, K., Nishide, J., Okazaki, K., Kato, H., Niwa, O., Nakagawa, terization of the SUMO-1 modification of RanGAP1 and its role in nu- T., Matsuda, H., Kawamukai, M., and Murakami, Y. (1999). Charac- clear envelope association. J. Cell Biol. 140, 259–270. terization of a fission yeast SUMO-1 homologue, pmt3p, required Martelli, A.M., Zweyer, M., Ochs, R.L., Tazzari, P.L., Tabellini, G., for multiple nuclear events, including the control of telomere length Narducci, P., and Bortul, R. (2001). Nuclear apoptotic changes: an and chromosome segregation. Mol. Cell. Biol. 19, 8660–8672. overview. J. Cell. Biochem. 82, 634–646. Vergnes, L., Peterfy, M., Bergo, M.O., Young, S.G., and Reue, K. Matunis, M.J., Coutavas, E., and Blobel, G. (1996). A novel ubiquitin- (2004). Lamin B1 is required for mouse development and nuclear in- like modification modulates the partitioning of the Ran-GTPase- tegrity. Proc. Natl. Acad. Sci. USA 101, 10428–10433. Published on- activating protein RanGAP1 between the cytosol and the nuclear line July 1, 2004. 10.1073/pnas.0401424101. pore complex. J. Cell Biol. 135, 1457–1470. Wang, Z.G., Delva, L., Gaboli, M., Rivi, R., Giorgio, M., Cordon- Matunis, M.J., Wu, J., and Blobel, G. (1998). SUMO-1 modifica- Cardo, C., Grosveld, F., and Pandolfi, P.P. (1998). Role of PML in tion and its role in targeting the Ran GTPase-activating protein, cell growth and the retinoic acid pathway. Science 279, 1547–1551. RanGAP1, to the nuclear pore complex. J. Cell Biol. 140, 499– Xhemalce, B., Seeler, J.S., Thon, G., Dejean, A., and Arcangioli, B. 509. (2004). Role of the fission yeast SUMO E3 ligase Pli1p in centromere Me´ met, S., Lilienbaum, A., and Israe¨ l, A. 1997. Rapid isolation of and telomere maintenance. EMBO J. 23, 3844–3853. Published on- mouse primary fibroblasts: a tool for the analysis of transgenic line September 9, 2004. 10.1038/sj.emboj.7600394. mice. Trends in Genetics, Technical Tips Online, T01046, 19/3/1997. Zhang, H., Saitoh, H., and Matunis, M.J. (2002). Enzymes of the Mu¨ ller, S., Hoege, C., Pyrowolakis, G., and Jentsch, S. (2001). SUMO modification pathway localize to filaments of the nuclear SUMO, ubiquitin’s mysterious cousin. Nat. Rev. Mol. Cell Biol. 2, pore complex. Mol. Cell. Biol. 22, 6498–6508. 202–210. Zhong, S., Muller, S., Ronchetti, S., Freemont, P.S., Dejean, A., and Panse, V.G., Hardeland, U., Werner, T., Kuster, B., and Hurt, E. Pandolfi, P.P. (2000). Role of SUMO-1-modified PML in nuclear body (2004). A proteome-wide approach identifies sumoylated substrate formation. Blood 95, 2748–2752. proteins in yeast. J. Biol. Chem. 279, 41346–41351. Published online July 30, 2004. 10.1074/jbc.M407950200. Pemberton, L.F., and Paschal, B.M. (2005). Mechanisms of receptor- mediated nuclear import and nuclear export. Traffic 6, 187–198. Pichler, A., Gast, A., Seeler, J.S., Dejean, A., and Melchior, F. (2002). The nucleoporin RanBP2 has SUMO1 E3 ligase activity. Cell 108, 109–120. Roberts, C.W., Leroux, M.M., Fleming, M.D., and Orkin, S.H. (2002). Highly penetrant, rapid tumorigenesis through conditional inversion of the Snf5. Cancer Cell 2, 415–425. Rosas-Acosta, G., Russell, W.K., Deyrieux, A., Russell, D.H., and Wilson, V.G. (2005). A universal strategy for proteomic studies of SUMO and other ubiquitin-like modifiers. Mol. Cell Proteomics 4,