More information about the Ubiquitin-Proteasome Pathway .

	Saitoh H, Pu RT, Dasso M.	SUMO-1: wrestling with a new ubiquitin-related modifier, Trends Biochem. Sci. 22, (10), 374-376, 1997
	Kretz-Remy C, Tanguay RM.	SUMO/sentrin: protein modifiers regulating important cellular functions,  Biochem. Cell. Biol .  77, (4), 299-309, 1999
	Vierstra RD, Callis J.	Polypeptide tags, ubiquitous modifiers for plant protein regulation,  Plant. Mol. Biol . 41, (4), 435-442, 1999
	Melchior F.	SUMO--nonclassical ubiquitin, Annu. Rev. Cell. Dev. Biol. (16), 591-626, 2000
	Yeh ET, Gong L, Kamitani T.	Ubiquitin-like proteins: new wines in new bottles, Gene 2, 248 (1-2), 1-14, 2000
	Muller S, Hoege C, Pyrowolakis G, Jentsch S.	SUMO, ubiquitin's mysterious cousin, Nat. Rev. Mol. Cell. Biol. (2), 202-210, 2001
	Wilson VG and Rangasamy D.	Intracellular targeting of proteins by sumoylation, Exp. Cell Res. Nov. 2001
	Pichler A and Melchior F.	Ubiquitin-Related Modifier SUMO1 and Nucleocytoplasmic Transport, Traffic, 3, (6), 381-7, 2002.

The localization of Ahus5/AtUbc9 in a plant cell indicate that it has important functions also outside the nucleus.

The Ubc9/Hus5 proteins were first found in Saccharemyces cereviciae and Schizosaccharomyces pombe. At the time of their discovery, they were thought to be ordinary ubiquitin conjugating enzymes and the first reports focused on their involvement in regulating the cell cycle. In S. pombe they found that the Hus5 protein was required for normal mitosis and recovery from DNA damage or S-phase arrest, see Al-Khodairy F. et al. 1995 , which indicated that they could control the degradation of key components of the cell cyclus machinery. In S. cereviciae they found that B-type-cyclin degradation was mediated by the Ubc9 protein , and repression of Ubc9 synthesis prevents cell cycle progression at the G2 or early M phase, causing the accumulation of large budded cells with a single nucleus, Seufert W. et al. 1995 . These early reports indicated that Ubc9/Hus5 was a involved in the regulation of protein degradation, but later reports have suggested additional functions.
 
 
 

The consensus site for SUMOylation is: b(X)XXh KXE where b is a basic amino acid, h is hydrophobic and X is any amino acid. A shorter motif, YKxE, ( Y=large hydrophobic residue) has been proposed by Rodriguez M. S. et al. 2000 , but modification in vivo requires the additional presence of a nuclear localization signal. Thus, protein substrates with this consensus motif must be targeted to the nucleus to undergo SUMO-1 conjugation. A identical consensus motif for SUMOylation, Y KXE was proposed by Sampson D. A. et al. 2001 . In addition they showed that mutating residues in this recognition motif prevents Ubc9 interaction and SUMO modification. Is this recognition motif part of a "lysine-finger"? Studies of some of the regions with consensus sites suggests that they form an amphipathic helix (personal observation).
While SUMO-1 is not able to form polymeric chains, SUMO-2 (SMT3A/Sentrin-3) and SUMO-3 (SMT3B/Sentrin-2) apparently has this capacity. In a recent paper, Tatham M. H. et al. 2001 , showed that SUMO-2 and SUMO-3 contain the consensus sequence YKxE for SUMOylation, and together with Ubc9 are able to produce polymeric chains of SUMO-2/SUMO-3 in vitro and in vivo.
 
 
 

NF-k B / IkB (Cactus) and Ubc9

Before the discovery of SUMO/Smt3 the first reports suggested that in mammalian cells Hus5/Ubc9 was involved in the "ubiquitin" conjugation of IkBa, Tashiro K 1997 . The NF-kB/Rel proteins are sequestered in the cytoplasm when bound to IkB a. Tashiro and co-workers suggested that in response to external signals, IkBa is phosphorylated, multi-ubiquitinated, and degraded by proteasomes, thereby releasing NF- kB/Rel proteins to migrate to the nucleus. This is in contrast to the report by Desterro J.M. et al. 1998    which indicate that conjugation of Smt3 to I k B blocks ubiquitylation and therefore proteasomal degradation. Thus, by stabilizing IkB, the vertebrate Smt3 conjugation pathway results in downregulation of NFkB activity. A recent paper by Bhaskar V. et al. 2000 , suggest a different mechanism by which Rel family transcription factors are regulated in Drosophila. In a two hybrid screen using Dorsal, a member of the Rel family, as a bait they found that it bound DmUbc9. They also showed that Dorsal was conjugated Drosophila Smt3 and that this conjugation enhances nuclear uptake of Dorsal. The DmUbc9/DmSmt3 together opposes the inhibitory effects of Cactus (the Drosophila homolog to IkB) on Dorsal nuclear uptake. ( Cactus binds Dorsal and prevent its translocation to the nucleus Geisler R.  et al. 1992 ; Lemaitre B. et al. 1995 ). They suggest that Smt3 conjugation is an evolutionary conserved pathway for the regulation of subcellular localization of various proteins. Whether this just applies to arthropods or if this is also the case in vertebrates remains unclear.

In Arabidopsis a protein NIM1 shows homology to the mammalian transcription factor inhibitor IkB, Ryals J. et al. 1997 . The NIM1 (for noninducible immunity) gene product is involved in the signal transduction cascade leading to both systemic acquired resistance (SAR) and gene-for-gene disease resistance in Arabidopsis. This may indicate that SAR signaling pathway in plants is representative of an ancient and ubiquitous defense mechanism in higher organisms. So far no homologs of NF kB (proteins with the Rel-homology domain, RHD) has been found in plants, and with 99% of the Arabidopsis project now finished we probably have to conclude that these proteins have evolved just in animals (or has been lost in plants).

Ubc9 have also been found in other yeast two-hybrid screens and coprecipitation studies were MEKK1 and the type I TNF-alpha receptor was used as baits, Saltzman A. et al. 1998 . Both MEKK1 and TNF-alpha are known to regulate NFkappaB activity.
 
 

Drosophila Ubc9 and SUMO/smt3

In addition to the results from Bhaskar and co workers which show that Dorsal is regulated by DmUbc9/DmSmt3 conjugation system, a few other reports have appeared in recent years and further shows the versatility of the Ubc9/SUMO system in Drosophila. In a two hybrid experiment designed to find proteins interacting with heat shock proteins Joanisse D. R. et al. 1998 found that the Drosophila homologue of Ubc9/Hus5 interacts with a small heat shock protein, Hsp23. They also reports that DmUbc9 interacts with Drosophila and mammalian Hsp27. As with the human Ubc9 they find that the DmUbc9 is primary localized in the nucleus. Another Drosophila team have picked up DmUbc9 in an yeast two hybrid screen using the amino terminal domain of the Groucho corepressor as a bait, Ohsako S. et al. 1999 . The Drosophila Groucho transcriptional corepressor protein has previously been shown to interact with the DNA-binding bHLH domain of Enhancer of split, Hairy and Deadpan proteins, which proteins are involved in neurogenesis, segmentation and sex-determination, respectively.
    Long X. and Griffith L. C. 2000 have recently reported that one isoform of Drosophila neuronal calcium/ calmodulin-dependent protein kinase II, CaMKII, is conjugated to DmSUMO-1 in vivo. (As far as I know this is the first example of SUMOylation of a protein kinase, even though the human MEKK1 was found to interact with hUbc9 Saltzman and co-workers never showed that it was actually modified with SUMO-1). CaMKII is thought to be involved in the majority of the neuronal functions mediated by intracellular free Ca2+, and has been implicated in long-term potentiation, learning, and memory. In a study of the Ubc9/SUMO-1 conjugation system in Drosophila they found that SUMO-1 conjugation occurs in all developmental stages of Drosophila and in the adult central nervous system. The autors suggest that normal SUMO conjugation is essential in the differentiated nervous system. SUMOylation of CaMKII also suggest a new potential mechanism for regulating neuronal CaMKII function.
    In Drosophila Ubc9 is also known as the gene lwr ( lessWright ) Apionishev S. and Rasooly R. S. 1998 . lwr acts as a dominant suppressor of female specific meiotic mutations. In a suppression screen of a mutation in the nod locus, (which encompass a kinesin-like protein involved in female meiotic chromosome segregation), Apionishev S et al. 2001 identified mutations in lwr, the UBC9 homologue. lwrmutations dominantly suppress the nondisjunction and cytological defects of female meiotic mutations that affect spindle formation. In general Drosophila lwr mutations are recessive lethal and show pleiotropic effects. Tanda S. 2001 has for instance observed that lwr mutations result in a drastic increase in larval blood cell number. Another example of this pleiotropy is the semushi locus / semi gene Epps J. L. and Tanda S. 1998 . When mutated "semi" blocks nuclear import of Bicoid during early embryogenesis and results in misregulation of the segmentation genes that are Bicoid targets. Analysis of the semushi locus revealed that it actually encodes DmUbc9. A plausible explanation is that the "semi" mutation affect the nuclear pore complex, in particular RanGAP1 which is known to be modified with SUMO-1 Saitoh H. et al. 1997 . The pleiotropic effects observed when DmUbc9 function is disturbed are in many ways similar to what we observe in Arabidopsis plants expressing antisense constructs of AtUbc9. The severity of the phenotypes ranges from embryonic lethality to wild type plants.
    In a search for new genes that interact with the Ras signaling pathway that regulate morphogenesis in Drosophila, Schnorr J. D. et al. 2001 screened for lethal mutations that could enhance a weak Ras1 eggshell phenotype. In this screen they found that a P-element insertion in smt3 (SUMO-1) disrupted Ras signaling. Furthermore, they showed that smt3 interfere with functions downstream of the EGFR receptor. Maybe the ETS-1 transcription factor is a target or is this another "fuzzy link" to the Ras-p53 pathway?
    In a yeast two-hybrid screen set up to identify proteins that interact with septins, Shih H. P. et al. 2000 , identified both DmUbc9 and a homologue of yeast Uba2 (DmUba2). In S. cerevisiae Uba2 is an E1 activating enzyme. These results are interesting and are in good agreement with results from yeast, which shows that smt3 is localized to the septin ring Takahashi Y. et al. 1999 . See septins . In a recent report Shih H. P. et al. 2002 shows that DmUba2 shuttles between the embryonic cortex and nuclei during the syncytial blastoderm stage, indicating that SUMOylation occurs both inside and outside the nucleus. However they could not observe SUMOylation of septin proteins. They also report that DmSmt3/DmSUMO are enriched on chromosomes during mitosis, and that DmUba2 and DmSmt3/DmSUMO are localized to the nuclei during interphase but dispersed throughout the cells during mitosis.
 
 
 

Trancription factors modified with SUMO-1/smt3

Hus5/Ubc9 proteins have also been reported to bind to the ETS-1 transcription factor in the two hybrid screen, Hahn SL et al. 1997 . ETS-1 is regulated by the Ras and Ca2+ signaling pathways and is implicated in various physiological processes leading to cell growth, differentiation and apoptosis.

Another protein which has been found to interact with Ubc9 in a two-hybrid screen is activating transcription factor 2 (ATF2), Firestein R. et al. 1998 . Using ATF2 as a bait they pulled out Ubc9 as a interacting protein, and showed that ATF2 could be "ubiquitinated" in the presence of purified human UBC9. The transcription factor AP-2 is also modified with SUMO at a N-terminal conserved Lys residue. SUMOylation of AP-2 appear to downregulate its transcriptional activity, Eloranta J. J. and Hurst H. C. 2002 .

Muller S. and co-workers has recently shown that c-Jun is modified with SUMO-1. A single lysine residue in c-Jun, (Lys-229), is preferentially SUMOylated. They also show that a SUMO-1- deficient c-Jun (K229R mutant) has increased transactivation potential on an AP-1-containing promoter, suggesting that SUMO-1 negatively regulates c-Jun activity.  Kotaja N . et al. 2002 report that PIASxalpha enhance SUMOylation of c-Jun and other transcription factors, thereby confirming earlier reports that PIAS proteins function as zinc finger-dependent E3 SUMO protein ligases.

Another transcription factor which has shown to interact with human Ubc9 using the two-hybrid system, is the E2A transcription factor, Loveys D. A. et al. 1997 . They suggests that mUBC9 may be involved in the degradation of key nuclear proteins that regulate cell cycle progression. The transcription factor c-Myb is also reported to be SUMOylated Bies J. et al. 2002 . Mutational analysis of the protein showed that two lysine residues in the negative regulatory domain of c-Myb was modified with SUMO-1, and the stability of protein was increased. A mutant c-Myb protein which is not SUMOylated display an increased transactivation activity upon a Myb responsive promoter.

The androgen receptor (AR), a ligand-activated transcription factor belonging to the steroid receptor superfamily, do also interact with human Ubc9, Poukka et al. 1999 , but they claim that SMT3/SUMO-1 modification is not the main purpose of this interaction. The AR-Ubc9 interaction seems to activate the transcription of AR regulated genes, such that Ubc9 can be regarded as a coactivator. In another paper they show that when sumoylated Lys residues were substituted with other amino acids this enhanced the transcriptional activity of AR without influencing its transrepressing activity Poukka et al. 2000 . They also propose that reversible sumoylation is a mechanism for regulation of steroid receptor function, as the glucocorticoid, mineralocorticoid, and progesterone receptor all contain the sumoylation consensus sequence.
Another steroid transcription factor that bind Ubc9 is the glucocorticoid receptor (GR) Kaul S. et al. 2002 . In addition to GR they also show that the glucocorticoid modulatory elements GMEB-1 and -2 also interacts with Ubc9.

The transcriptional repressor TEL is a nuclear phosphoprotein, which gene is rearranged in several types of leukemias and fibrosarcomas. In a recent report Chakrabarti SR. et al. 2000 have shown that the N-terminal part of TEL interacts with Ubc9 and SUMO-1. Interestingly the SUMOylation of TEL results in a specific nuclear localization of the protein, at so called nuclear speckles, that they have named TEL bodies. They also show that the leukemia-associated fusion protein TEL/AML1 is SUMOylated but has a different nuclear localization than SUMOylated TEL alone. It appears once again that SUMOylation of nuclear proteins somehow targets proteins to specific sites in the nucleus. Interesting!! Maybe there exist special reseptors in the nuclear matrix that recruit Ubc9 modified proteins?

Human STRA13, a basic helix-loop-helix (bHLH) transcriptional factor, is also reported to interact with Ubc9 in a two-hybrid screen, Ivanova A. V., et al. 2001 . Stra13 (also known as Sharp2 and Dec1) is a transcriptional repressor which is related to the Drosophila Hairy, Enhancer of Split, and the mouse Hes1 proteins. The Stra13 gene is induced by retinoic acid and its gene product may play a key role in signaling pathways that lead to growth arrest and terminal differentiation by repression of target genes, probably through a interaction with the histone deacetylase HDAC1 Sun H and Taneja R. 2000. Ivanova and colleagues speculate that Ubc9 may be involved in the regulation of STRA13 protein level through ubiqutin dependent degradation, but I think it is more likely that Ubc9 regulate other functions such as "nuclear targeting".

Recently it was reported that the zebrafish paired-like:CVC homeobox protein VSX-1 interacts with Ubc9 Kurtzman A. L. and Schechter N. 2001 . VSX-1 is a paired-class homeobox protein that lack a second DNA binding domain, the paired-domain, and is closely related to the C. elegans ceh-10 gene Levine EM. et al. 1994 . The Vsx-1 gene is expressed during retinogenesis and plays an important role in bipolar cell differentiation. Even though the VSX-1 protein interact with Ubc9 it is not SUMOylated. Instead it appears as if the Ubc9 protein binds to the N-terminal NLS (nuclear localization signal) region and is required for the nuclear localization of VSX-1. The VSX-1 protein has however a consensus motif for SUMOylation inside the homeobox domain (personal observation). In a previous study Kurtzman A. L. et al. 2000 have found that VSX-1 is ubiquitinated and degraded via the ubiquitin/proteasome pathway. Is this a case where the ubiquitin and SUMOylation sites overlaps?

The zinc finger protein SALL1, which has been implicated in the development of the Townes-Brocks syndrome (TBS), is also modified with SUMO-1 Netzer C. et al. 2002. They observed that the SALL1 protein is localized to discrete nuclear foci and found that SALL1 interacted with Ubc9 and SUMO-1 in a yeast two hybrid system. Can SUMOylation of SALL1 explain its nuclear distribution, corresponding to PML nuclear bodies? The SALL1 protein interacts with PIN2/TRF1, a telomeric protein that negatively regulates telomere elongation, and it has been suggested that PIN2/TRF1 may be involved in the cellular response to double strand DNA breaks Kishi S. et al. 2001. Further more, SALL1 represses transcription by recruiting a histone deacetylase complex that include HDAC1, HDAC2, RbAp46/48, MTA-1, and MTA-2 Kiefer SM. et al. 2002. This is remarkably similar to the transcriptional repressor STRA13 (a Ubc9 interacting protein) which may act through HDAC1 as well Sun H and Taneja R. 2000. Furthermore, it has been shown that HDAC1 is SUMOylated at the C-terminal residues Lys-444 and Lys-476 David G. et al. 2002. Is this another example of a nuclear multiprotein complex were many of the proteins are modified by SUMO/smt3?

The GC box binding transcription factor Sp3, a transcription factor related to Sp1, is another target for SUMOylation Sapetschnig A. et al. 2002. PIAS1 (an E3-ligase) was shown to interact with Sp3 and enhance SUMO conjugation activity. Further more they showed that SUMOylation of Sp3 resulted in shutdown of its transcriptional activity. SUMOylation of Sp3 also affects its nuclear localization Ross S. et al. 2002. Sp3 modified with SUMO is relocalized to distinct nuclear bodies. A similar SUMO mediated relocalization have also been reported for proteins localized to PML nuclear bodies. Whether the transcriptional silencing of Sp3 activity is due to its interaction with histone deacetylases, some of which are SUMOylated, is unknown.

SUMOylation of histone deacetylases may be an important regulatory mechanism for transcriptional repression of genes. The class II histone deacetylase HDAC4 was recently shown to be SUMOylated Kirsh O. et al. 2002. An unmodified (not SUMOylated) HDAC4 protein had a reduced histone deacetylation activity and lower repression activity. They also suggests that SUMO modification occurs at the nuclear pore complex and is catalyzed by RanBP2, a SUMO E3-ligase.
 
 

Ubc9/SUMO and regulation of nucleocytoplasmic transport

A few years ago it was reported that a Xenopus homologue of RanGAP1 interacts with Ubc9, suggesting that Ubc9 may "interfere/regulate" with the nuclear transport machinery, Saitoh H. et al. 1997 . (Ran is a small GTPase required for nuclear transport in eukaryotic cells. Ran has also recently been reported to serve important functions during cytokinesis and is required for chromatin-induced mitotic spindle formation, Carazo-Salas R. E. et al. 1999 . Many Ran's have been found in plants / Arabidopsis, I my self found three Ran genes when I did a large screen for Ras/Rac/Rho members ). Another paper supporting these data reports that the human Ubc9 colocalizes with RanGAP1 at the nuclear envelope, Lee G.W. et al. 1998 . They also reports that the ubiquitin related protein SUMO-1 is covalently linked to RanGAP1, and that HsUbc9 forms thiolester conjugates with recombinant SUMO-1, but not with recombinant ubiquitin (which supports the finding by Desterro JM.). This should suggest that Ubc9 is functionally distinct from E2-type UBCs. A new report by Saitoh H. et al 1998 supports this view, and suggest that SUMO-1 conjugation promotes RanGAP1's interaction with RanBP2.
    Transport of the human Ubc9 protein into the nucleus is probably regulated by a importin ß-related transport receptor,  importin 13 (Imp13), which is controlled by the RanGTPase system Mingot J. M. et al. 2001. It has also been shown that human Ubc9 co-localizes with RanGAP1 at the nuclear envelope, Lee G.W. et al. 1998. The C-terminal part of the Ubc9 proteins have a basic region with homology to nuclear localization signals, but so far, (to my knowledge), there is no data on whether this region regulate nuclear import.
    Recently a 129 amino acid part from the human nucleoporin RanBP2/Nup358 protein was found to interact and bind to the Ubc9 protein with high affinity, both in vitro and in vivo Saitoh H. et al. 2001. When they transfected COS-7 cells with a construct expressing this Ubc9 binding element they observed mislocalization of SUMO-1 and SUMO2/3 and that a major part of Ubc9 was in the cytoplasm rather than the nucleus. They also observed mislocalization of other nuclear proteins such as the PML protein and RAD51. Controlled expression of this Ubc9 binding element may be used as a tool to study Ubc9 role in targeting proteins to the nucleus.
    I has been recently shown that Ubc9 is localized at the nuclear pore complex (NPC) together with SENP2 (a SUMO protease) Zhang H. et al. 2002. They show that the N-terminal part of SENP2, which has a FG repeat domain, interact with Nup153, a nucleoporin that is localized to the nucleoplasmic face of the pore. Using immunogold labeling they showed that Ubc9 is localized to both cytoplasmic and the nucleoplasmic filaments of the NPC. The interaction of SENP2 with the nuclear pore and its interaction with Nup153 was also observed by Hang J and Dasso M. 2002.

In another yeast two-hybrid study using bleomycin hydrolase (BH) as bait, the human Ubc9 protein was found as a potential partner Koldamova R. P. et al. 1998 . BH is a highly conserved cysteine proteinase and homologs can be found in many organisms. In yeast BH forms a complex with similarities to the 20 S proteasome. Could this be an indication that SMT3/SUMO-1 modification targets proteins to another "protein modification factory" where they undergo spesific changes?
 
 

Septins, CDC3-like proteins and SUMOylation

Initial yeast studies showed that SMT3/SUMO was localized to the septin ring Takahashi Y. et al. 1999 . They detected HA-tagged Smt3 mostly in nuclei but also at the mother-bud neck in a similar fashion as septin fibers. One of the possible "targets" was the Cdc3 protein, (CDC3, CDC10, CDC11, and CDC12 genes encode a family of related proteins, the septins, which are involved in cell division and the organization of the cell surface during vegetative growth). Further studies of yeast septins have indeed shown that they are modified with SMT3 (are SUMOylated). An important spin off from this work has been the identification of an E3-like factor that is essential for SUMOylation of the yeast septins. Two reports, one by Johnson E. S. and Gupta A. A. 2001 , the other by Takahashi Y. et al. 2001 , both identified Siz1/Ull1 as a novel SUMO1/Smt3-ligase. The Siz1p protein (SAP and Miz) is a member of the PIAS family (P rotein Inhibitor of A ctivated STAT) which all have a novel Ring-like domain. PIAS-like proteins are found in most eukaryots (plants, yeast and animals). The Ull1 mutant, which has a defective Siz1 gene, is unable to SUMOylate septins. SUMOylation of yeast septins is therefore dependent on three enzyme activities produced by the E1 (Sua1/Uba2-complex), the E2 (Ubc9), and the Siz1 E3-like protein. Strunnikov A. V. et al. 2001 , identified Siz1p as a as a bypass suppressor of the growth defect associated with the SMT4 disruption. Defects in SMT4/Ulp2 gene, which encode a Smt3-deconjugating enzyme, causes a prominent chromosome phenotype, and has been used to study chromosome condensation in S. cerevisiae. Yeast also have a paralog of Siz1, the Siz2 protein. Strunnikov and co-workers have suggested that the Siz1 and Siz2 proteins are involved in a novel pathway of chromosome maintenance, which make sense to me when you take into account all proteins involved in regulation of DNA repair and replication that is modified with SUMO.
 
 

Regulation of DNA replication and DNA repair

In another study, using the yeast two-hybrid system to find proteins which interact with the human RAD52 protein, Hus5/Ubc9 was identified as one of the interacting partners. Further studies showed that the Hus5/Ubc9 protein also interacted with RAD51, p53, and a ubiquitin-like protein UBL1, Shen Z. et al. 1996 . The interaction between RAD51 and Ubc9 was also observed by Kovalenko O. V. et al., 1996 in a similar two-hybrid setup using RAD51 as a bait. They also reported co-localization of RAD51 and Ubc9 in synaptonemal complexes which may suggest a role in meiosis. The RAD51 protein is a eukaryotic homolog of RecA and has been implicated in in homologous recombination and recombinational repair of DNA damage. Notably, the RAD52 homologue in S. pombe, RAD22, is modified with Pmt3 (SUMO-1/smt3 homologue), and can be observed in a complex with the RAD51 homologue Rhp51 Ho J. C. et al. 2001 . This suggest that SUMO/smt3 modification of RAD52 is more or less universal.
    The association between Ubc9 and p53 is particulary interesting and in 1999 it was shown that the human tumor suppressor p53 was "SUMOylated" ( Gostissa M. et al., 1999 ; Rodriguez M. S. et al., 1999 ). These research groups both show that an evolutionary conserved C-terminal lysine residue at position 386 is the major attachment site for SUMO-1. Overexpression of SUMO-1 in U2OS cells (osteoblast-like cells) leads to increased p53-dependent transcriptional activity. Whether this is due to increased protein stability or if SUMOylation affects it subcellular localization is unknown at present. The C-terminal Lys residue in p53 that is SUMOylated is also very close to the Serine residue at position 392 that is phosphorylated by casein kinase II. These C-terminal modifications, SUMOylation, phospholylation and acetylation, may therefore be closely intertwined and regulate p53 transcriptional activity. One should also keep in mind that the levels of p53 protein is regulated by ubiquitin conjugation and in normal cells, p53 protein is maintained at a low level by ubiquitin-mediated proteolysis. Ubiquitination of p53 requires the ubiquitin-activating enzyme Ubc5 and the MDM2 protein which acts as a ubiquitin protein ligase. Several C-terminal lysine residues in human p53 appear to be modified with ubiquitin Rodriguez M. S. et al. 2000 , and SUMOylation may therefore interfere with the p53 protein degradation pathway. A recent report by Kahyo and co-workers show that p53 interact with the PIAS1 protein, Kahyo T. et al., 2001 . A mutation in the ring finger domain resulted in lack of Ubc9 binding but it was still able to bind p53 and SUMO-1. A recent study by Schmidt D. and Muller S. 2002 , confirmed this result showing that PIAS1 and PIASxbeta, acted as specific E3-like ligases that promoted the SUMOylation of p53 (and c-Jun) in vitro and in vivo. Furthermore they reported that SUMOylation of p53 strongly repressed the transcriptional activity of p53.

Mdm2, an E3 ubiquitin ligase for the p53 tumor suppressor protein is also modified with SUMO-1, Buschmann T., et al., 2000 . The residue which is SUMOylated, Lys-446, is located within the RING finger domain of MDM2. They also shows that in vitro SUMOylation of Mdm2 abrogates its self-ubiquitination and increases its ubiquitin ligase activity toward p53. Thus, it appears that SUMOylation stabilize MDM2 and promote degradation / down regulation of p53. DNA damage, however, reduces MDM2 SUMOylation and thereby less p53 is targeted for degradation. Seems like a nice model. It has also been shown that covalent attachment of SUMO-1 to Mdm2 is dependent upon the activation of a heterodimeric Aos1/Uba2 enzyme (E1-activating enzyme) in addition to Ubc9. In a later study, the same group has identified the N-terminal part of MDM2 as the Ubc9 interacting region Buschmann T. et al. 2001. In cells exposed to UV irradiation the MDM2/Ubc9 interaction is reduced. They also show that a peptide corresponding to amino acid 40-59 from MDM2 is sufficient to prevent SUMOylation of MDM2 in vitro and in vivo. MDM2 was recently identified in a protein complex together with the histone deacetylase HDAC1 Ito A. et al. 2002, which in fact is another protein that can be modified with SUMO David G. et al. 2002. A dominant negative HDAC1 mutant promote acetylation and p53 stability, suggesting that deacetylation of p53 leave the active Lys residues open for ubiquitination. Acetylation of the C-terminal Lys residues in p53 (K320, K370, K372, K373, K381 and K382) does not overlap with the lysine modified by SUMO, K386, but overlaps with Lys residues that are ubiquitinatinated and therby preventing p53 degradation. The acetyltransferase CBP forms a trimeric complex together with p53 and PML and both p53 and PML are known to be modified with SUMO. Thus, several of the proteins that interact with p53 and p53 interacting proteins are SUMOylated. Coincidence or not?

    It is now also known that another protein of the p53 family, namely the p73alpha, is SUMOylated  Minty A. et al., 2000. They also show that SUMOylated p73 is more rapidly degraded by the proteasome than unmodified p73, but it appears that SUMOylation is not absolutely required for p73 degradation. The major target in p73 is the C-terminal lysine K627, while in p53 it is the C-terminal lysine K386 which is SUMOylated. In a two hybrid screen with the human p73 protein as a bait a they found a number of interesting proteins, among them Ubc9. Other p73 interacting proteins were, thymine DNA glycosylase and CHD3. Interestingly thymidine DNA glycosylase (see TDG) is also modified with SUMO-1 Hardeland U. et al. 2002. The CHD3 proteins are known as chromatin-remodeling factors involved in repression of transcription and a CHD3 homolog, PICKLE, has been found in Arabidopsis, where it regulate the transition from embryonic to vegetative development, Ogas J. et al., 1999. The CHD3 proteins have also been found in protein complexes containing HDAC1 and HDAC2 Humphrey GW. et al. 2001. Could this indicate that p73 also is regulated by acetylation / deacetylation in a similar fashion as p53? Earlier results have shown that acetylation of p73 by the acetyltransferase CBP/p300 is not essential for its transcriptional activity Zeng X. et al. 2000.

Poly(ADP-ribose) polymerase (PARP) have also been reported to bind to Hus5/Ubc9, Masson M. et al. 1997 . PARP has been suggested to play a regulatory role in vivo, in DNA replication and/or DNA repair based mainly on its capacity to bind to DNA strand breaks. Studies of cells from PARP (-/-) mice have later shown that PARP is crucial in the maintenance of genomic stability Rosenthal et al., 1999 .

Another protein which is known to be involved in DNA repair, topoisomerase I (topo-I), is SUMOylated in response to the inhibitor camptothecin Mao Y., et al. 2000 . Yeast with a temperature sensitive Ubc9 mutant was also shown to be hypersensitive to topoisomerase I mediated DNA damage induced by camptothecin. Over expression of SUMO-1 resulted in resistance to camptothecin. Previous studies have shown that topoisomerase I is down regulated by the 26S proteasome when cells are exposed to camptothecin Desai S. D., et al., 1997 . Whether SUMOylation / ubiquitin conjugation competes for the same Lys residue in topoisomerase I is unknown. It may be that the Ubc9/SUMO-1 pathway has evolved to process nuclear proteins that are damaged or have undergone specific conformational changes. The topo-I protein has a dynamic distribution in the nucleus, but is primary localized to the nucleolus. The localization of topo-1 can however be perturbed by treating cells with the topo I inhibitor topotecan, which results in a net movment of topo-1 out of the nucleolus. In a recent publication Mo Y. Y. et al. 2001 shows that this redistribution of topo I is dependent on SUMOylation of topo I. They also showed that heat-shock blocked SUMOylation of topo I and movement of topo I out of the nucleolus. They also report that overexpression of Ubc9 results in a redistribution of topo-1 (movement out of the nucleolus).
It has also been shown that topoisomerase II is SUMOylated when HeLa cells are exposed to the anticancer drug VM-26 (teniposide) and that there is a physical interaction between topoisomerase II and SUMO-1/UBC9 Mao Y., et al. 2000 .

The SUMOylation of p53, TOPO I and II, plus the interaction Ubc9 with RAD51 and RAD52 suggests that UBC9/smt3/SUMO are important regulators of DNA repair and replication. This link to DNA repair is also found in S.pombe where the HUS5 mutant (Ubc9 knockout) was found to display abnormally sensitivity to both inhibitors of DNA synthesis and to ionizing radiation al-Khodairy F. et al., 1995 . The HUS5 protein is required for efficient recovery from DNA damage or S-phase arrest and manifests a rapid death phenotype in combination with a temperature sensitive S phase and late S/G2 phase cdc mutants. The HUS5 deletion mutants also exhibit high levels of abortive mitoses, suggesting a role for hus5+ in chromosome segregation. Another S. pombe mutant rad31-1, which is sensitive to both UV and ionising radiation and exhibits a growth defect at 35 C, was identified as an ubiquitin activating protein Shayeghi M. et al., 1997 . The rad31 was not required for either the S/M or G2/M checkpoint, but double mutant analysis indicates that rad31 acts in a process which is defective in the checkpoint rad mutants and which involves HUS5. Interestingly the RAD31 protein has similarity to the AXR1 gene of Arabidopsis, which is required for various auxin responses. However, the AXR1 gene in Arabidopsis is closest related to amyloid beta precursor binding protein 1 (APP-binding protein 1), and may be one of the factors regulating SUMOylation in plants Chow N. et al. 1996, Ruegger M. et al. 1998.
Disruption of the pmt3 gene (SUMO-1/smt3 homologue) in S. pombe is not lethal but results in slower growth Tanaka K. et al. 1999 . Pmt3 defective cells are prone to aberrant mitosis and they have chromosomes with increased telomere lengths. Thus, the S. pombe SUMO-1 homologue may be involved in telomere maintenance. A Pmt3:GFP-fusion protein localized to the spindle pole body and again suggest involvement of Ubc9/SUMO-1 proteins in chromosome segregation.

Another piece of data that connect the Ubc9/Hus5 protein with DNA repair and replication comes from work with the Papilloma virus E1 protein. A two hybrid screen picked up Ubc9 as one of the E1 interacting proteins, and Ubc9 was found to interact specifically with E1 both in vitro and in vivo Rangasamy D. and Wilson V. G. 2000 . They also showed that the E1 protein was SUMOylated. One of the functions of the E1 protein is as a replication initiator that recognizes and binds to the viral origin and initiates DNA strand separation through its ATP-dependent helicase activity. In addition the E1 protein functions in viral DNA replication by recruiting several cellular proteins to the origin, including host DNA polymerase and RPA.

Another interesting DNA helicase which is SUMOylated is the Werner syndrome protein, WRN. In a JBC paper a few years ago Kawabe Y. et al. 2000, used the mouse WRN as a bait in the yeast two hybrid assay, and found Ubc9 and SUMO-1 as interacting proteins. The WRN protein is also known to interact with replication protein A (RPA), proliferating cell nuclear antigen (PCNA), TOPO I and p53. As previously mentioned both TOPO I and p53 are SUMOylated and it has recently been shown that also PCNA is SUMOylated Hoege C. et al. 2002. Is this an indication that SUMOylation serves as an address TAG in the nucleus, or does it extend the life time of the proteins? The helicase domain of WRN is similar to E. coli RecQ and the WRN protein is thought to be involved in DNA replication. Kawabe and co-workers also examined the interaction between Ubc9 and the Bloom syndrome gene product (BLM), another protein within the RecQ helicase family, and showed that it indeed interacted with Ubc9. When they expressed a GFP-BLM fusion protein which lacked the C-terminal region with the nuclear localization signal, this resulted in a cytoplasmic localization of the gene product and no SUMOylation Suzuki H. et al. 2001. Recent studies of the RecQ helicases has shown that they are involved in suppressing 'promiscuous' genetic recombination and in ensuring accurate chromosome segregation Karow J. K. et al., 2000. In E. coli, RecQ together with the UvrAB complex, have been implicated in the suppression of illegitimate recombination induced by UV radiation Hanada K. et al. 2000, suggesting that the functions of RecQ helicases is highly conserved.
Recently it was reported that PCNA, a DNA-polymerase sliding clamp involved in DNA synthesis and repair, is modified by SUMO-1 Hoege C. et al. 2002. The modified Lys residue (K63) can be modified with both SUMO and ubiquitin suggesting that the modification status of PCNA regulate its function.

A DNA methyltransferase, DNA (cytosine-5-)-methyltransferase 3 beta (Dnmt3b), has also been identified as a SUMOylated protein Kang E. S. et al. 2001 . Kang and co-workers identified Ubc9 and SUMO-1 as Dnmt3b-interacting proteins during a yeast two-hybrid screen. Dnmt3b is essential for de novo methylation and play an important role during development Okano M. et al. 1999 . Cytosine methylation is essential for the organization and stabilization of heterochromatin and mutations of the human Dnmt3b gene result in the developmental syndrome, ICF (immunodeficiency, centromeric heterochromatin instability, and facial anomalies).
Interestingly, a thymine-DNA glycosylase (TDG) is also modified by SUMO-1 and SUMO-2/3 Hardeland U. et al. 2002 . TDG is a is a mismatch-specific uracil / thymine-DNA glycosylase which is involved in base excision repair. Apparently the SUMOylation of TDG reduces the DNA substrate and AP site / (abasic site) binding affinity and may perturb with the controlled dissociation of TDG from the AP site. More results that connect excision repair with Ubc9 comes from a report by Yan MD. et al 2000. They report that the dual function reducing protein / DNA repair enzyme APE/redox factor-1 (Ref-1) is modified with SUMO-1. In a yeast two-hybrid screen after proteins binding to Ref-1/APE/APEX they identified Ubc9 as a partner. They also observed that over expression of Ubc9 in HeLa cells reduced the levels of Ref-1. In addition to the APE/Ref-1 AP endonuclease activity, this protein has also another activity that controls the redox status of a number of transcription factors Fritz G. 2000 and Evans AR. et al. 2000. A structural analysis of APE/Ref-1 like proteins in Arabidopsis show that they contain a N-terminal SAP motif (after SAF-A/B, Acinus and PIAS). (Remember, the yeast Siz1p protein, a ubiquitin-like protein ligase, contain two motifs: SAP and Miz). Many interesting links here...
The fact that two enzymes involved in excision repair so far have been found to be SUMOylated suggests that more factors involved in BER are SUMOylated, just wait and see.

Recently adenovirus E1B-55kDa was identified as a SUMO-1 modified protein and it was shown that SUMOylation is necessary for proper nuclear localization Endter C. et al. 2001 . Again SUMOylation appears to direct a protein to a specific "compartment" in the nucleus. The E1B-55kDa protein is known to interact with p53 and effectively block p53 mediated transactivation. This is thought to antagonize p53-induced apoptosis and cell cycle arrest.

The nucleotide binding protein Fhit (fragile histidine triad) is another tumor suppressor protein that appear to interact with Ubc9 Shi Y. et al. 2000 . Fhit is frequently inactivated in lung cancer and suppresses tumor formation by inducing apoptosis. Fhit has diadenosine triphosphatase activity and is encoded as a fusion protein with Nit, a member of the nitrilase superfamily Pekarsky Y. et al. 1998 . It would be interesting to know the cellular distribution of the Fhit protein, bearing in mind that the Ubc9 protein is mainly localized in the nucleus.

A report by Biggins S. et al. 2001 indicate that in yeast smt3 is an important factor during sister chromatid separation and segregation. In this study they microscopically screened a temperature-sensitive collection of budding yeast mutants that contain a GFP-marked chromosome and used this method to find genes involved in "chromosome behavior" during mitosis. From this analysis they found that SMT3 is required for chromosome segregation. In addition they report the isolation of WSS1 (weak suppressor of smt3), a high-copy smt3 suppressor. Interestingly this is a novel protein which is related to a Arabidopsis protein ( At1g55910 ) that contain zf-RanBP motifs (the homology to WSS1 do not include the zf-RanBP motifs). Keep in mind that SUMO-1 conjugation promotes RanGAP1's interaction with RanBP2 and that a 129 amino acid part from the human nucleoporin RanBP2/Nup358 protein interact and bind to the Ubc9 protein with high affinity. Also remember that the Ran GTPase regulate spindle assembly during mitosis (for a review see Melchior F. 2001 ). Some interesting links here.... The WSS1-homology region is conserved in most eukaryots. A new study by Fukagawa T et al. 2001 suggest that SUMO-1 / Ubc9 play important roles during cell cycle / cell division in animal cells. They have analyzed a number of temperature-sensitive mutants of the CENP-C protein, which is a conserved centromere protein, thought to be an important component in kinetochore assembly in vertebrate cells. One of the CENP-C mutants, ts4-11, displayed metaphase delay and chromosome missegregation. When ts4-11 cells were transfected with a human HeLa cell cDNA library maintained in a retroviral vector, they found that expression of the SUMO-1 gene were able to suppress the temperature-sensitive phenotype. Thus, in both yeast and animal cells smt3/SUMO-1 appear to have important functions during chromosome segregation.
In a recent report in Cell, Pichler A. et al. 2002 , show that the nucleoporin RanBP2/Nup358 also has SUMO1 E3-like activity. RanBP2 enhances the transfer of SUMO-1 from Ubc9 to the sp100 target protein, however the 33 kDa domain with catalytic activity has no homology to PIAS family proteins. The SUMOylation process appears to be located at the cytosolic part of the nuclear pore complex and suggest that modification and import may be linked processes, see commetary by Azuma Y. and Dasso M. 2002 .
 
 

PML / PML nuclear bodies and SUMOylation

PML, a zinc RING-finger nuclear phosphoprotein, is another protein which is modified with SMT3/SUMO-1, Duprez E et al. 1999 ; Ishov A. M. et al. 1999 ; Zhong S. et al. 2000 . PML was originally identified as part of a translocated chromosomal fusion product associated with acute promyelocytic leukemia (APL), and is now defined as a tumor suppressor gene. The PML protein can be induced by interferon and apperar to be essential for certain types of apoptosis. Duprez and co-workers found that the residues modified with SUMO-1 was part of the nuclear localisation signal (Lys487 or Lys490) and Ishov et al. 1999 showed that SUMOylated PML was essential for protein recruitment of to PML nuclear bodies. The PML nuclear body is a multiprotein complex that probably regulate gene transcription.
Recently it was reported that PML regulates p53 acetylation and premature senescence induced by oncogenic Ras, Pearson M. et al. 2000 . It was found that oncogenic Ras upregulates PML expression, and overexpression of PML induces senescence in a p53-dependent manner. Remember, p53 is acetylated at lysine 382 upon Ras expression, an event that is essential for its biological function. Interestingly they observed that activated Ras could induce re-localization of p53 and the formation of a trimeric complex containing p53, PML and the acetyltransferase CBP. Here again we have two proteins PML and p53 which is both regulated by SUMOylation, and are co-localized in the cell, is this a coincidence or...? Probably not! In a recent article Guo A. et al. 2000 show that PML acts as a transcriptional co-activator with p53 and show that there is a physical interaction both in vivo and in vitro between p53 and PML. Both proteins co-localizes within the PML nuclear body (PML-NB). Is the SUMO modification of p53 and PML the address tag? In recent paper Fogal V. et al. 2000 , have shown that p53 is recruited into nuclear bodies by a specific PML isoform (PML3). Binding of PML3 to p53 also appear to increase the trancriptional activity of p53, (based on its ability to activate p53 responsive promoters). PML-NBs are also known to contain a ubiquitin specific protease, HAUSP, Everett R. D. et al. 1997 , but apparently it does not remove SUMO-1 from substrates. Another abundant protein in PML-NBs which is SUMOylated is the SP100 protein Seeler J. S. et al. 2001 . The SP100 protein is known to interact with members of the HP1 family of nonhistone chromosomal proteins and the authors suggest that SUMOylation of SP100 stabilize these complexes.
    Several proteins in addition to these mentioned here are localized to PML-NBs, such as: INT6, Hsp70, RecQ helicase, BML and even the Rb protein has been found associated with PML in co-immunoprecipitation experiments Labbaye C. et al. 1999 . Whether more of the PML-NB localized proteins is SUMOylated is an open question. However, a "new" PML-NB associated protein has recently been found. Goodson M. L., et al. 2001 reports that the human heat shock factor 2 (HSF2) is localized to PML-NBs, and they also find, surprise surprise..., that it is modified with SUMO-1. A new twist to the story is that SUMOylation of HSF2 converts this protein to its active DNA-binding form. In a recent paper Hong Y. and co-workers reports that HSF1 is SUMOylated in response to heat stress and show that this modification is required for its localization in nuclear stress granules Hong Y. et al. 2001 . Mutation of Lys 298 (the SUMOylated residue) prevents HSF1 localization to these stress granules. Apparently both the HSF2 and HSF1 proteins need to be SUMOylated to acquire DNA binding ability.

    The human cytomegalovirus (HCMV) immediate-early transactivator 2 (IE2 / IE2-p86) is another protein that is modified with SUMO-1. Using a deletion mutant of IE2-p86 as a bait in a two hybrid screen, Hofman et al. 2000 found that it interacted with SUMO-1 as well as Ubc9. They also showed that there were two modification sites within IE2-p86, located in an immediate vicinity at amino acid positions 175 and 180. Interestingly they found that IE2-p86 was translocated to PML nuclear bodies (PML-NB's) independent of SUMOylation. A similar two hybrid screen with HCMV IE2 was performed by Ahn J. H. et al. 2001 , and they found that IE2 interacted with SUMO-1, SUMO-2 and SUMO-3 as well as Ubc9. Ahn and co-workers also identified the same two lysine residues at positions 175 and 180 as the major SUMO-1 conjugation sites. Another HCMV protein which is modified by SUMO-1 is the major immediate early protein IE1 Xu Y. et al. 2001 . This protein is an abundant 72-kDa nuclear phosphoprotein which is targeted to PML-NBs. Xu and co-workers found however that SUMOylation was not required for the targeting of IE1 to the PML-NBs, and the IE1 protein did not interact with either Ubc9 or SUMO-1 in yeast two hydrid experiments.

    Treatment of HeLa cells with with the proteasome inhibitor MG132 has remarkable effects on the localization of PML and SUMO-1 Mattsson K. et al. 2001 . After MG132 treatment these proteins were excluded from PML-NBs and accumulated instead in the nucleolus. The authors suggest that SUMO-1 and PML may be directed to the nucleoli under normal conditions and that the nucleolus may have a function in the regulation of proteasomal protein degradation.
    The Daxx protein was recently shown to interact with SUMO-1 (sentrin) and Ubc9, adding more support to the "address TAG / recruitment theory" Ryu S. W. et al. 2000 . Daxx was first identified as a novel signaling protein that bound specifically to the death domain in cell surface receptor Fas. However later work identified Daxx as an Nuclear domain 10 (ND10) associated adapter protein that can bind PML Ishov A. M. et al. 1999 . Ryu and co-workers results shows that the amino acids 625-740 of Daxx, known as the Fas binding region, overlaps with region which interacts with SUMO-1 and Ubc9. SUMOylation of Daxx may therefore regulate binding to Fas, PML or other proteins. The transcription factor Pax3 is probably regulated by Daxx and results suggests that Pax3 transcriptional activity can be repressed by Daxx. When Daxx is recruited to PML-NBs, Pax3 repression is relieved. In turn, Daxx recruitment to PML-NBs depends on a functional PML protein and the SUMOylation of PML, Lehembre, F. et al. 2001 . PML-NBs may therefore have important functions in sequestration of proteins regulating transcription factors such as Pax3. The transcriptional activity of p53 on the other hand appear to increase as p53 is recruited to PML-NBs. (It is a complicated world). Interestingly, the adaptor protein Daxx also bind another SUMOylated protein, the insulin-sensitive glucose transporter GLUT4 Lalioti V. S. et al. 2002 . Using confocal microscopy they found that Daxx was localized in both the nucleus and in punctate cytoplasmic structures which was organized in strings close to the plasma membrane (similar to location of Ubc9 in plant cells?.. se animation on top of this page). A new study by Jang M. S. et al. 2002, have tried to address the question whether SUMOylation is neccesary for the recruitment of Daxx to PML-NBs. Interestingly, they find that a SUMOylation defective Daxx co-localize with PML-NBs / PML oncogenic domains (PODs).
    It is however a striking observation that proteins in PML nuclear bodies (ND10 associated proteins) such as p53, PML, Sp100 and Daxx, are modified by SUMO-1. This can not be a coincidence. Experiments with primary PML(-/-) cells have shown that in the absence of PML, several ND10 associated proteins such as Sp100, CBP, ISG20 and Daxx, fail to accumulate in the nuclear bodies and acquire aberrant localization patterns Zhong S. et al. 2000 . PML is therefore required for the proper formation of the PML nuclear bodies and without SUMO modification it is not able to fulfill this task. The disruption of  PML-NBs by Epstein-Barr virus is also interesting as the viral proteins that interfere with PML-NB assembly also is SUMOylated. This was recently reported by Adamson A. L. and Kenney S. 2001 , which showed that expression of  the EBV immediate-early genes BZLF1 (Z) and BRLF1 (R) in EBV-positive cell lines resulted in PML-NB dis-assembly. They found that BZLF1 (Z) was SUMO-1 modified (through amino acid 12) and they speculate that PML-NB dispersion is due to a competition for SUMO-1.
Furthermore, the Epstein-Barr virus nuclear antigen, EBNA-3C, interact with SUMO-1 and SUMO-3 suggesting that a number of EBV proteins is involved in the disruption of PML-NBs Lin J. et al. 2002 .
    Another protein which recently was shown to be localized within the PML-NBs, and which probably is SUMOylated, is the dynamin like GTPase Mx1 Engelhardt O. G. et al. 2001 . Using the two-hybrid system they identified several Mx-1 binding proteins that was previously known to be localized within PML nuclear bodies, i.e. Sp100, Daxx, and Bloom's syndrome protein (BLM). In addition they also found SUMO-1 and SAE2 (SUMO-1 activating enzyme sub-unit 2) as interacting partners. Some of the interferon-induced Mx1 like proteins are known to have antiviral activity against a variety of viruses and accumulate in the nucleus following interferon treatment. A protein kinase, Mx-interacting protein kinase (PKM) / homeodomain-interacting protein kinase 2 (HIPK2), has also been localized to PML nuclear bodies and is another candidate as a SUMO-1 / Ubc9 interacting protein Trost M. et al. 2000 .
 
 

Plants and Ubc9 / smt3

The smt3/SUMO homologs in Arabidopsis has been characterized and results have shown that they are conjugated to a number of proteins Gosink M. M (poster 1998) . So far there is no publications of function or localization.
 

Hanania et al. 1999 , published a paper in Plant Journal where they reported the result from a two hybrid screen in tomato using ethylen-inducing xylanase (EIX) from the fungus Trichoderma viride. In this screen they found a SUMO-1 like protein, T-SUMO, which interacted with EIX. Transgenic plants expressing this T-SUMO cDNA in the sense orientation supress the production of ethylen induced by the xylanase. Plants expressing antisense T-SUMO however increase ethylen biosynthesis when they are challenged by the xylanase. Interesting!!!

Another interesting paper appeared in Plant Journal in 1996 by Xia G. et al. 1996 . Using the fission yeast Schizosaccharomyces pombe to find proteins in Arabidopsis which regulate cell shape, cell polarity  and cell cycle regulation, they came up with a number of interesting proteins and among them a Ubc9 homolog, (well it is NOT the same as our Ahus5, it turns out that the Arabidopisis fellows working with Ubc's have created their own nomenclature). Among the other interesting proteins they found were Rho/Rac like proteins (The Aracs) and Ran binding proteins.
 
 

Ubc9 structure

The three dimensional structure of human Ubc9 was published three years ago at 2.0 Å resolution, Tong H. et al. 1997 . Compared with the known crystal structures of Ubc1 and Ubc4, which regulate different cellular processes, Ubc9 has a 5-residue insertion that forms a very exposed tight beta-hairpin and a 2-residue insertion that forms a bulge in a loop close to the active site. The active Cys residue that is involved in binding SUMO is shown in the figure of human Ubc9 ( figure is copied from ).

Mossessova E. and  Lima C. D. 2000 have recently published the structure of the Ulp1/SUMO complex. The yeast Ulp1 protease is known to de-conjugate SUMO from targeted proteins.

The plant Ubc9 proteins are very similar to the human counterparts, as can be seen from the protein alignment at the bottom of this page. Using the Swiss PDB modeling tool I have made a preliminary modell of the AtUbc9 / Ahus5 protein (the Arabidopsis homolog to Ubc9). Compare with the figure above.
 

The spatial orientation of the AtUbc9 / Ahus5 protein is similar to the figure of human Ubc9 above. In the figure to left the active Cys. residue is shown in space-fill mode. In the figure to the right amino acids that differ from the human Ubc9 are shown as space-filled residues. Conservative amino acid changes are colored light blue, while more distinct amino acid changes are colored in green. The active Cys residue is marked in red and is conserved in all ubiquitin ligases. From the figure it can be seen that the pocket surrounding the active Cys is very well conserved. Parts of the C-terminal helix 4 is also relatively well conserved (helix in the centre-left in bottom of the figure).

In a recent study Lin D. et al. 2002 have analysed the substrate recognition site of human Ubc9 in more detail. Interestingly they found that the SUMOylated domain of p53 did not form any defined structure when bound to Ubc9. This is in contrast to the report from Bernier-Villamor V. et al. 2002 , which shows that distinct motifs is required for substrate binding and SUMO modification of p53, (see also commetary by Hochstrasser M. 2002 ).
 

SUMO/smt3 de-conjugation

The first enzyme / protease which had SUMO de-conjugating activity was the yeast protein Ulp1, Li S. J. et al. 1999 . One interesting finding they did was that the pattern of Smt3-coupled proteins in yeast changed throughout the cell cycle, and that specific conjugates accumulate in Ulp1 mutants. In a two hybrid screen using Ulp1 as a bait Takahashi Y. et al. 2000 , found two interacting proteins, NUP42 and Gle1, which are components of the nuclear pore complex (NPC). Is the Ulp1 protein located at the nuclear pore and for instance chop off the SUMO-1 / Smt3 tags when proteins are exported from the nucleus?
The yeast SMT4 mutant, first isolated as a high-copy-number suppressor of a defective centromere-binding protein, was recently shown to encode a second Smt3-deconjugating enzyme, Ulp2 Li S. J. and Hochstrasser M. 2000 . Cells lacking Ulp2, accumulate specific Smt3-protein conjugates and display a conjugate pattern that is distinct from that observed in a ulp1(ts) strain. The Ulp2 deficient strain is also hypersensitive to DNA-damaging agents and hydroxyurea. They also suggest a feedback mechanism that limits Smt3-protein ligation when Smt3 deconjugation by both Ulp1 and Ulp2 is compromised. The sequence similarity between Ulp1 and Ulp2 is limited and confined to a ~200 amino acids region (the Ulp Domain or UD), which includes all the residues forming the cysteine protease-like active site, poster at Yeast Genetics and Molecular Biology 2000 . The SMT4 mutant has a prominent chromosome phenotype and has been used to study chromosome condensation in S. cerevisiae Strunnikov A. V. et al. 2001 .
Interestingly an Arabidopsis homolog of Ulp1 exists, AAF26995 .

Several SUMO-1 specific proteases have been found in humans. The two proteases SENP1 and SENP2 (Sentrin-1 specific protease 1 & 2) was found by Gong L. et al., 2000 . They are both related to the Ulp1 protease. Their results suggest that sentrinized / SUMOylated PML was selectively affected by SENP1 whereas proteins residing in the cytosol was unaffected. The SENP2 protease is interestingly associated with the nuclear pore and suggest that it may be involved in regulating nucleocytoplasmic transport of proteins Hang J. and Dasso M. 2002.

Recently a SUMO-1-specific protease, SUSP1, was identified and cloned from human brain Kim K. I. et al., 2000 . With confocal microscopy using green fluorescent protein SUSP1 fusion they showed that SUSP1 is exclusively localized to the cytoplasm of NIH3T3 and HeLa cells. Another Smt3 spesific protease was recently reported. Using Smt3b as bait in the yeast two-hybrid screen they found a novel Smt3-specific isopeptidase, SMT3IP1, Nishida T. et al. 2000 . This protease is distantly related to budding yeast Saccharomyces cerevisiae Ulp1, human SENP1 or human SUSP1. This enzyme appear to have higher affinity to Smt3a and Smt3b than to SUMO-1 in vitro. Interestingly, it did not cleave ubiquitin from ubiquitinated p53 and the SMT3IP1 protein was localized almost exclusively at the nucleolus during interphase.