Báo cáo khoa học: Protein interactions in the sumoylation cascade – lessons from X-ray structures pdf

REVIEW ARTICLE
Protein interactions in the sumoylation cascade – lessons
from X-ray structures
Zhongshu Tang
1,
*, Christina M. Hecker
2,
, Astrid Scheschonka
1
and Heinrich Betz
1
1 Department of Neurochemistry, Max-Planck-Institute for Brain Research, Frankfurt, Germany
2 Department of Biochemistry II, Johann-Wolfgang-Goethe-University, University Hospital, Frankfurt, Germany
Introduction
Sumoylation is a post-translational modification in
which a SUMO (small ubiquitin-like modifier) protein
is conjugated to the e-amino group of a lysine residue
of the substrate protein [1,2]. SUMO attachment has
been shown to occur for a large number of proteins
with important roles in many basic cellular processes,
and to be highly regulated in both the nucleus and
other cellular compartments [1–6]. Sumoylation is
mechanistically related to ubiquitination, a more exten-
sively studied protein modification reaction [7].
Although the process of ubiquitination has been recog-
nized for about 20 years and sumoylation for 10 years,
Keywords
Aos1-uba2; neddylation; Pc2; PIAS proteins;
RanBP2; RanGAP1; SUMO; sumoylation;
ubc9; ubiquitination
Correspondence

H. Betz, Department of Neurochemistry,
Max-Planck-Institute for Brain Research,
Deutschordenstrasse 46, 60528 Frankfurt,
Germany
Fax: +49 69 96769 441
Tel: +49 69 96769 220
E-mail:
Present addresses
*NIH ⁄ NEI, Bethesda, MD, USA
Department of Systemic Cell Biology, Max-
Planck-Institute for Molecular Physiology,
Dortmund, Germany
(Received 21 November 2007, revised 20
March 2008, accepted 11 April 2008)
doi:10.1111/j.1742-4658.2008.06459.x
Sumoylation is a multi-step protein modification reaction in which SUMO
(small ubiquitin-like modifier) proteins are covalently attached to lysine res-
idues of substrate proteins. Here, we compare the sequences and structures
of modifiers and enzymes involved in sumoylation with those of the related
ubiquitination and neddylation cascades. By using available structural data
on modifier ⁄ enzyme ⁄ substrate interactions, we discuss and model sumoy-
lation complexes that include SUMO-1 and the E1 and E2 enzymes Aos1-
uba2 and ubc9, or SUMO-1 and E2 together with the E3 ligase RanBP2
and its substrate RanGAP1. Their comparison provides insight into the
protein interactions underlying sumoylation, and suggests how SUMO
proteins may be translocated between enzymes during the various steps of
the protein modification reaction.
Abbreviations
APPBP1, APP binding protein 1; CtBP, C-terminus binding protein; CTD, C-terminal domain; E1, activating enzyme; E2, conjugating enzyme;
E3, ligase; HECT, homologous to E6AP C-terminus; IR, internal repeat; MIF-2, migration inhibiting factor 2; Nedd8, neural cell-expressed

developmentally down-regulation protein 8; Pc2, polycomb protein 2; PIAS, protein inhibitor of activated STAT; RanBP2, Ran binding
protein 2; RanGAP1, Ran GTPase-activating protein 1; RING, really interesting new gene; SAP, scaffold-associating region ⁄ Acinus ⁄ PIAS;
SENP, SUMO-1 ⁄ sentrin-specific peptidase; SIM, SUMO interaction motif; SMT3, suppressor of MIF-2; SP-RING, Siz ⁄ PIAS-RING; STAT,
signal transducer and activator of transcription; SUMO(A), SUMO non-covalently bound at the E1 adenylation site; SUMO(C), SUMO
conjugated to a substrate; SUMO(T1), SUMO linked to the catalytic cysteine of activating enzyme E1 via a thioester; SUMO(T2), SUMO
linked to the catalytic cysteine of conjugating enzyme E2 via a thioester; SUMO, small ubiquitin-like modifier; uba, ubiquitin activating
enzyme; ubc ⁄ E2, ubiquitin-conjugating enzyme; UBL, ubiquitin-like modifier; UFD, ubiquitin-fold domain; Ulps, ubiquitin-like protein protease.
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3003
understanding of the molecular mechanisms of these
modification reactions began to emerge only in the last
5 or 6 years, when high-resolution structures of the
proteins and protein complexes involved in ubiquitina-
tion, sumoylation and another ubiquitin-like modifica-
tion, neddylation [conjugation of Nedd8 (for neural
cell-expressed developmentally down-regulation pro-
tein 8)], were solved. This review compares the avail-
able structures of modifiers and enzymes involved in
the SUMO, ubiquitin and Nedd8 pathways (for refer-
ences, see Table 1) and discusses how these proteins
may interact during the sumoylation reaction. For fur-
ther details, on the function and cellular regulation of
sumoylation, please refer to previous reviews [1–6].
Sumoylation pathway
The enzymatic machinery that adds and removes
SUMO to and from substrate proteins resembles that
involved in ubiquitination [1,4]. Like ubiquitin, SUMO
proteins are expressed as precursors that need to be
proteolytically processed by C-terminal hydrolases to
expose a C-terminal Gly-Gly motif that is required for
conjugation. This processing is called SUMO matura-

tion (Fig. 1A). Two forms of Ulps (ubiquitin-like-
protein proteases) in Saccharomyces cerevisiae and
several forms of SENP (SUMO-1 ⁄ sentrin-specific pep-
tidase) proteins in human have been found to process
SUMO precursors [1,2,8].
The attachment of SUMO to substrate proteins
requires three enzymes, which catalyze distinct steps of
the conjugation reaction: SUMO-activating enzyme
(E1), SUMO-conjugating enzyme (E2) and SUMO
ligase (E3) [1,4] (Fig. 1A). The first step, called activa-
tion, includes two processes: adenylation and transfer
of SUMO to a cysteine within E1. As a result, a thio-
ester bond is formed between the carboxyl group of
the C-terminal glycine of SUMO and the E1 cysteine
residue; this step requires Mg
2+
-ATP. In the second
step, SUMO is transferred from the E1 active site to
another cysteine in the E2 enzyme, forming a SUMO–
E2 thioester intermediate. Finally, SUMO is attached
to the amino group of a lysine residue within the sub-
strate protein. This ligation reaction is assisted by an
E3 enzyme.
SUMO conjugation is a reversible process. The iso-
peptide bond between SUMO and the substrate can be
cleaved by members of the Ulp ⁄ SENP family [8]. This
process is called de-sumoylation (Fig. 1A). Both matu-
ration and de-sumoylation produce free mature SUMO
[1,2].
Components of the sumoylation

machinery
Sumoylation, neddylation and ubiquitination are all
mediated by three-enzyme cascades as shown in
Fig. 1B–D. For ubiquitination, a single ubiquitin gene
and approximately 10 E1, 100 E2 and 1000 E3
enzymes have been found in mammals [7,9]. In con-
trast, four SUMO isoforms (SUMO-1 to -4) have been
identified in humans, and conjugation involves single
E1 (Aos1-uba2) and E2 (ubc9) enzymes plus a growing
list of about 10 E3 candidates.
SUMO
The structures of yeast SUMO [called SMT3, suppre-
ssor of migration inhibiting factor 2 (MIF-2)], human
SUMO-1 ⁄ -2 ⁄ -3, human ubiquitin and human Nedd8
have all been determined [10–14]. Basically, all these
modifier polypeptides have a compact globular struc-
ture with a characteristic bbabbab fold and N- and
Table 1. Protein structures referred to in the review. All structures
except for those labeled by asterisks (by NMR) were resolved by
X-ray diffraction. All proteins are of human except for 1EUV (yeast)
1Z7L (mouse), 1U9A (mouse) and RanGAP1 (mouse) in 1KPS.
Protein or components of the
protein complex PDB No. Reference
APPBP1-uba3 1YOV 21
APPBP1-uba3, Nedd8 1R4N 21
APPBP1-uba3, ubc12N26 1TT5 29
APPBP1-uba3, Nedd8T,
Nedd8A, Mg
2+
-ATP, ubc12

2NVU 24
Cbl, ubch7 1FBV 31
E6AP, ubcH7 1C4Z 32
Nedd8 1NDD 10
Nedd8, Den1 1XT9 65
Sae1-Sae2, SUMO-1,
Mg
2+
-ATP (renamed as
Aos1-uba2, SUMO-1 in
the text)
1Y8R 15
SAP* 1V66 40
Senp2, SUMO-1 1TGZ 66
SMT3 1EUV 67
SUMO-1* 1A5R 11
SUMO-1, ubc9, RanBP2,
RanGAP1
1Z5S 19
SUMO-2 1WM3 12
SUMO-3 C47S* 1U4A 13
SUMO-1, SIM of PIASX* 2ASQ 20
uba1 1Z7L 23
ubc12, UFD of uba3 1Y8X 30
ubc9 1U9A 27
ubc9, RanGAP1 1KPS 28
ubc9, uba2 (Cys domain)* 2PX9 36
Ubiquitin 1UBQ 14
Interactions between sumoylation proteins Z. Tang et al.
3004 FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS

C-termini extending to opposite sides (Fig. 2A,B).
Superposition of the SUMO isoforms with ubiquitin
and Nedd8 shows that the peptide backbones are well
conserved (Fig. 2B,C). Many sumoylation sites lie in a
consensus motif FKXE ⁄ D, where F is a large hydro-
phobic amino acid, K is lysine, X is any amino acid,
and E ⁄ D is glutamate or aspartate. Yeast SMT3 and
mammalian SUMO-2 ⁄ 3 ⁄ 4 all contain such consensus
motifs at their N-termini, whereas SUMO-1 does not.
The consensus lysine sites of SMT3 and SUMO-2 ⁄ 3,
and the C-terminal GG motif are stringently conserved
[13]. Regions of low alignment are found along the a2
and b3 regions, which are located close to the N- and
C-termini, respectively (Fig. 2B,C). In addition, the
C-termini, the a2 helix and the loop between a1 and
b3 differ between modifiers (Fig. 2B).
Eleven residues of SUMO-1 have been shown to
have directly contact uba2, a subunit of the sumoyla-
β1
α1
β2
β3
β4
β5α2
SUMO-1
Nedd8
SUMO-2
ubiquitin
SMT3
SUMO-3

N-
C-
–5
N-
A
C
B
C-
1A5R/SUMO-1
1U4A/SUMO-3
1ABQ/ubiquiti
1NDD/Nedd8
1WM3/SUMO-2
1EUV/SMT3
Fig. 2. SUMO-1 structure and comparison with the other SUMO paralogues, ubiquitin and Nedd8, by the NCBI vector alignment search tool.
(A) Structure of SUMO-1 showing the typical bbabbab fold (determined in [11]). (B) Superposition of the backbones of various SUMO pro-
teins with ubiquitin and Nedd8, all shown in various colors. Regions conserved among all proteins are shown in red, non-conserved aligned
regions in light blue. The structure of the C-terminus of SUMO-1 changes upon protein binding. (C) Sequences of proteins aligned in (B).
Residues in capitals are aligned among all modifiers, including the strictly conserved residues in red and other aligned residues in light blue.
Residues indicated in lower case could not be aligned. Residues in light gray outside the boxed area are not included in the superposition
shown in (B). Secondary structure elements (a-helices and b-sheets) corresponding to SUMO-1 are indicated above the alignment. The )5
residues that confer selectivity for the corresponding E1 enzyme are highlighted by a yellow background (see also Fig. 3). Sequences are
from the NCBI Protein Data Bank, with accession numbers given on the left.
E1
E1
SUMO
S
SUMO
+
ATP

AMP + PPi
E2
SUMO
S
E2
E3
SUMO
substrate
Ulp
SH
SUMO
MATURATION
ACTIVATION
CONJUGATION
LIGATION
Aos1-uba2
ubc9
PIAS-1,
substrate
APPBP1-uba3
ubc12
Rbx1
substrate
SUMO-1/-2/-3/-4
Nedd8
DE-SUMOYLATION
uba1
ubc7H,
Cbl,
substrate

ubiquitin
ABCD
Fig. 1. Overview of sumoylation, neddylation and ubiquitination reactions. (A) The sumoylation pathway. Free SUMO is generated by either
proteolytic maturation of a SUMO precursor or de-sumoylation of a sumoylated substrate by Ulp ⁄ SENPs. For conjugation, SUMO is firstly
activated in an ATP-dependent reaction to form a thioester bond with the activating enzyme E1, then transferred to the conjugation enzyme
E2, and finally conjugated to a substrate that is recruited by a ligase E3. (B–D) Comparison of the enzyme cascades involved in sumoylation,
neddylation and ubiquitination reactions. (A) and (B) are modified from [1], and (C) and (D) from [29] and [7], respectively.
Z. Tang et al. Interactions between sumoylation proteins
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3005
tion E1 complex; seven of these residues lie in the very
C-terminal tail [15]. An exposed region formed by the
first two b-sheets and the first a-helix is the binding
site for enzymes and specific substrates, such as DNA
glycosylase, Ube2–25k, isopeptidase SENP2, protein
inhibitor of activated STAT (PIAS) and Ran binding
protein 2 (RanBP2) [16–20]. The N-terminal loops of
SUMO proteins are not homologous to those of
ubiquitin and Nedd8, and their functions are still
unknown.
E1
E1 activating enzymes facilitate the conjugation of
ubiquitin-like modifiers to substrate proteins through
adenylation, thioester formation within E1, and thioest-
er transfer from E1 to E2 [15]. Unlike the ubiquitin E1
enzyme uba1, which is a monomer, the E1 enzymes
involved in the SUMO and Nedd8 modification path-
ways are all heterodimers. SUMO E1 is composed of
Aos1 and uba2 (also called Sae1 and Sae2, respec-
tively), and Nedd8 E1 is composed of APPBP1 (APP
binding protein 1) and uba3. The domain structures of

Aos1 and APPBP1 resemble the N-termini, and those
of uba2 and uba3 resemble the C-termini, of the ubiqu-
itin E1 enzyme (Fig. 3A). SUMO-1 is recognized exclu-
sively by residues of uba2; no direct interactions have
been observed between SUMO-1 and the Aos1 subunit
[15]. Aos1 contains only a single domain, which partici-
pates in adenylation of SUMO. Uba2 includes three
domains: the catalytic cysteine domain, the adenylation
domain and the ubiquitin-fold domain (UFD), which is
structurally similar to ubiquitin and other ubiquitin-like
modifiers. The adenylation domain of uba2 has a typi-
cal Gly-X-Gly-X-X-Gly ATP-binding motif [21], which
is conserved among the ubiquitin-, SUMO- and Nedd8-
activating enzymes. The UFD of uba2 shows strong
interaction with E2, which is essential for recruiting E2
to E1 [15]. The C-terminal extension of uba2 contains a
unique nuclear localization signal that may be impor-
tant for enrichment of the SUMO machinery in the
nucleus [22]. APPBP1 has two domains: an adenylation
domain, which resembles Aos1, and part of a cysteine
domain, which resembles that of uba2.
Both subunits of SUMO E1 are conserved from
yeast to human [2]. The crystal structures of human
Aos1-uba2, human APPBP1-uba3, and of the partly
resolved mouse uba1, are closely related [15,21,23]
(Fig. 3C). From the side facing the ATP-binding motif,
Aos1-uba2 resembles a U-shaped complex (Fig. 3B). It
forms a large groove, with the adenylation domain at
the base, and the UFD and cysteine domains on both
sides. The catalytic cysteine residue lies in a long loop

between the cysteine domain and the adenylation
domain [15]. Of the 986 residues of Aos1-uba2, 663
can be precisely aligned with APPBP1-uba3, and 203
residues are conserved between both proteins
(Fig. 3D,E). Two APPBP1-uba3 structures have been
published: with and without Nedd8 bound at the cata-
lytic site [24]. They are largely identical, except for the
UFD [24]. Here, comparative structure analysis
revealed a turn of the UFD of approximately 120°
upon binding of Nedd8 to the cysteine domain [21,24].
This conformational change of E1 may be important
for passage of the modifiers through the conjugation
machinery [24].
Despite the similarities between the structures of all
ubiquitin-like modifiers and their corresponding E1s,
binding of the modifiers to their activating enzymes is
highly specific. This specificity is attributed to the
selective pairing of amino acid side chains upon inter-
action of the modifier and E1 [21]. The fifth-last ()5)
residue of the mature modifier proteins has been
shown to be particularly crucial, i.e. Glu93 in
SUMO-1, Gln89 ⁄ 88 ⁄ 89 in SUMO-2 ⁄ 3 ⁄ 4, Ala72 in
Nedd8, and Arg72 in ubiquitin (highlighted in yellow
in Fig. 2C). Matching E1 residues are Thr149 in
uba2, Arg190 in uba3 and Gln608 in uba1 (high-
lighted in purple in Fig. 3E). Interestingly, Ala72 of
Nedd8 has been found to be also essential for recog-
nition by the Nedd8-specific protease NEDP1 [25].
This suggests that a common mechanism may be used
to recruit both activating and processing enzymes to

ubiquitin-like modifiers.
Fig. 3. Domain organization, structures and alignment of various E1 activating enzymes. (A) Schematic representation of domains in human
Aos1-uba2, APPBP1-uba3 and uba1. (B) Structure of the Aos1-uba2 complex [15]. The protein complex is sub-divided into three domains
(indicated by dashed lines). The ATP-binding motif is highlighted by a dashed black circle, and the position of the catalytic cysteine residue is
shown in yellow. (C) Superposition of the polypeptide backbones of Aos1-uba2 and APPBP1-uba3. Residues conserved among all proteins
are shown in red and other aligned residues are shown in light blue. Divergent regions are shown in green for Aos1-uba2 and in pink for AP-
PBP1-uba3. The position of the ATP-binding motifs and the catalytic cysteine residues are indicated as in (B). (D,E) Alignment of the
sequences of Aos1 with APPBP1, and of uba2 with uba3, respectively. Residues in capital letters are aligned, including conserved residues
in red and other aligned residues in light blue. Lower-case residues indicate non-aligned sequences. Residues shown in light gray are not
shown in (C). A green background delineates the activation motif. Modifier-recognizing residues are shown on a purple background, and the
catalytic cysteine residues on a yellow background.
Interactions between sumoylation proteins Z. Tang et al.
3006 FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS
Adenylation domain
529
Aos1-uba2
APPBP1-uba3
uba1
442
1058
Cys216
Cys632
346
Cys173
Cys domain UFD domain
6401
1
Aos1
uba2
3abu1PBPPA

1
1
1
NLS
A
B
D
E
C
Z. Tang et al. Interactions between sumoylation proteins
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3007
E2
After activation, the SUMO molecule is transferred to
the conjugating enzyme E2. All SUMO proteins share
a common E2 enzyme, ubc9. It is highly conserved
from yeast to human [2]. Mouse ⁄ human ubc9 displays
high similarity to the human Nedd8 and ubiquitin E2
enzymes: 84% of its residues can be aligned with and
superpositioned to the Nedd8 E2 ubc12 and the ubiqu-
itin E2 ubcH7 (Fig. 4B,C). Common to all E2 enzymes
is a conserved 150-residue abbbbb(bb)aaa motif
named the ubc superfold; differences are found only in
their N- or C-terminal extensions [26–28]. The four
a-helices and six b-strands of ubc9 can be modeled
into a cubicle (Fig. 4A), in which a1, a2 and a3 ⁄ 4 are
each located on one side, and the anti-parallel b-sheet
formed by the b1–4 strands on another side. The two
remaining surfaces are covered by loose peptide
strands. In the sumoylation cascade, ubc9 functions as
a core component that interacts with nearly all other

proteins involved in the modification reaction. It pos-
sesses at least five protein interaction sites. (a) The a1
region contains the a1 helix and surrounding residues
of the b1b2 loop (Fig. 5A). A similar region in the
neddylation E2 enzyme ubc12 binds the UFD of uba3
[29,30] and thereby enables E2 recruitment. Similarly,
the UFD of uba2 has also been shown to strongly
interact with ubc9 [15]. Although structural data are
not available, it is likely that ubc9 and the UFD of
uba2 interact at the a1 region. (b) The region below
the a1 helix consists mainly of the N-terminal exten-
sion of a1 and loops between the b2 and b3 strands
and the b6 strand and the a2 helix, respectively
(Fig. 5B). In ubiquitin-conjugating enzymes, this is the
region that interacts with both the HECT (homolo-
gous to E6AP C-terminus) and RING (really interest-
ing new gene) domains of ubiquitin E3 ligases [31,32].
As discussed later, most sumoylation E3 ligases
contain RING domains. Thus, this region may be a
common site for E2–E3 interactions. (c) In the
SUMO-1–ubc9–RanBP2–RanGAP1 (Ran GTPase-
activating protein 1) complex, the four-stranded
b-sheet domain of ubc9 provides the binding site for
the IR1 (internal repeat 1) domain of RanBP2,
another suspected E3 ligase, in the sumoylation path-
way (Fig. 5C) [33]. (d) The a3 helix of the C-terminal
a3a4 helical region (Fig. 5D) binds RanGAP1, a spe-
cial substrate that is known to have a second binding
site for ubc9 [19,28,34]. (e) The ‘catalytic groove’ con-
tains the active site residue Cys93, and holds the

SUMO-1 C-terminus [35] (Fig. 5E). This region also
includes the a3 helix of E2 which mediates E2–UBL
(ubiquitin-like modifier) interactions in both the ubc9–
SUMO-1 and ubc12–Nedd8 complexes [19,24]. Cys93
A
B
C
Fig. 4. Structure and alignment of ubc9 with ubc12 and ubcH7. (A) Structure of ubc9 showing the typical ubc superfold as determined in
[27]. Secondary structure elements (a-helices and b-sheets) are indicated. (B) Superposition of the polypeptide backbones of ubc9, ubc12
and ubcH7. Residues conserved among all proteins are shown in red, non-conserved aligned residues are shown in light blue. Non-aligned
regions are shown in pink for ubc9, dark blue for ubcH9, and brown for ubc12. The positions of the catalytic cysteine residues are indicated
in yellow. (C) Corresponding sequence alignment. Sequences are from the NCBI Protein Data Bank, with accession numbers indicated on
the left. Residues aligned among all modifiers are given in capital letters, with fully conserved residues in red and other aligned residues in
light blue. Residues in lower case are not aligned. Secondary structure elements (a-helices and b-sheets) corresponding to ubc9 are
indicated above and below the sequences. Catalytic cysteine residues are indicated by black arrows.
Interactions between sumoylation proteins Z. Tang et al.
3008 FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS
is located close to the center of a long extended stretch
between the fourth b-strand and the second a-helix
[27]. Chemical shift perturbation experiments have
shown that the a3 helix of ubc9 displays affinity to the
Cys domain of uba2 [36]. This interaction, although
weak, seems to be important for the transfer of SUMO
from E1 to E2 [36].
In addition to the covalent bond formed between
SUMO and ubc9 during sumoylation, non-covalent
interactions between these proteins have recently been
characterized [37,38]. They occur between the a–b1
loop of ubc9 and the b1,3,5 strands of SUMO. This
assignment is distant from the E2 active site and

partially overlaps with the E1–UFD and RanBP1–IR1
A B
C
E F
H
G
D
Fig. 5. Protein interactions involved in sumoylation and related pathways. (A) E1–E2 interaction [29]. Ubc12 (blue) is bound to the UFD (yel-
low) of uba3. (B) E2–E3 interaction. UbcH7 (yellow) is bound to the RING domain (blue) of Cbl [31]. An interface is formed between ubcH7
a1 + L3 + L6 and the Cbl RING L1 + L2 + a. (C) E2–RanBP2 interaction. Ubc9 (blue) and RanBP2 (IR1 domain, yellow) are shown in the
SUMO-1–ubc9–RanBP2–RanGAP1 complex. The C-terminal IR1 motif contacts loop 1 and the b-sheet of ubc9 [19]. (D) E2–substrate inter-
action [19]. RanGAP1 binds ubc9 at its a3 region. (E) SUMO-1–E2 interaction [19]. Ubc9 (yellow) and SUMO-1 (blue), with the SUMO-1
b-sheet located near the ubc9 a2 helix. Note that SUMO-1 is not conjugated to ubc9 in this complex, but a strong interaction between ubc9
and SUMO-1 has been characterized [19]. (F) E1–SUMO-1 interaction at the adenylation site [15]. The SUMO-1 (blue) C-terminus interacts
close to an ATP (highlighted in red) which is bound at the adenylation motif (yellow). (G) E1–SUMO-1 interaction at the catalytic cysteine site
[24]. A covalent bond is formed between SUMO-1 (blue) and the catalytic cysteine (red). (H) RanBP2 (IR1 domain, yellow) interaction with
SUMO-1 [19]. The N-terminal IR1 domain binds SUMO-1 in a cleft between b2 and a1.
Z. Tang et al. Interactions between sumoylation proteins
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3009
interaction sites on ubc9. Interestingly, it seems that the
non-covalent interactions between SUMO and ubc9
promote SUMO chain formation, but seem not to be
important for SUMO–ubc9 thioester formation [37].
E3
The last step in sumoylation is the transfer of SUMO to
a substrate protein that is recruited by an E3 ligase.
Unlike SUMO and the E1 and E2 enzymes, E3 ligases
are quite diverse from yeast to mammals [1]. Ubiquitin
ligases can be broadly subdivided into two groups,
based on the presence of either a RING or HECT

domain. While both domains bind E2 enzymes at the
same position [31,32], HECT domain E3 ligases form a
covalent ubiquitin–thioester complex prior to conjuga-
tion of ubiquitin to its target, but RING domain E3
enzymes do not [7]. The most carefully analyzed group
of SUMO E3 ligases, the PIAS proteins, contain a modi-
fied RING domain and do not form thioesters with
SUMO [1,39]. HECT domain-containing E3 ligases for
SUMO have not been identified, but two other proteins
have been found to act as SUMO ligases: the polycomb
protein Pc2 and the nucleoporin RanBP2 [33].
PIAS proteins
PIAS proteins are characterized by a so-called SP-
RING (Siz⁄ PIAS-RING) domain, and bind both ubc9
and substrate proteins. However, they do not form a
covalent bond with the substrate but actually act as
adaptors. Within PIAS polypeptides, an N-terminal
SAP (scaffold-associating region ⁄ Acinus ⁄ PIAS)
domain is followed by the SP-RING domain, a
SUMO interaction motif (SIM), and a highly divergent
C-terminal domain (CTD) (Fig. 6A) [39]. The SAP
domain is known to bind DNA and proteins such as
tumor suppressor p53 [40] and lymphoid enhancer fac-
tor 1 [41]. The SIM domain has been implicated in
directly binding SUMO [42]. The SP-RING domain is
crucial for the interaction with E2 [1,5,43]. The CTD
has been repeatedly found to bind sumoylation sub-
strates; hence it is considered to be a substrate interac-
tion domain [1,42,44,45]. Except for the SP-RING
domain, no homology has been found between PIAS

proteins and single-chain ubiquitin E3 ligases. NSE2a
and TOPORS, two other proteins that contain RING
domains at positions different from those in PIAS pro-
teins, have recently been reported to function as
SUMO ligases; however, little is known about their
precise modes of action [46–48].
Of the functional domains of PIAS proteins, only
the human SAP domain has been analyzed at the
structural level. It is formed by a four-helical bundle
and is thought to bind DNA [40]. The structures of
the SP-RING domain and of the CTDs, the two
domains most important for E2 and substrate binding,
have not yet been analyzed. The only structural data
presently available are from the RING domain of the
ubiquitin E3 ligase Cbl which binds to the E2 enzyme
ubcH7 [31,32] (Fig. 5B); the latter was referred to
above when considering E2–E3 interactions.
Pc2
Pc2 was identified as a SUMO E3 for the transcrip-
tional co-repressors CtBP (C-terminal binding protein
of adenovirus E1A) and CtBP2, both in vivo and
in vitro [49,50]. However, the enhancement of CtBP
sumoylation by Pc2 in vitro is very modest, and
PIAS1, PIASx and RanBP2 also can promote SUMO
attachment to CtBP, suggesting that multiple factors
may be involved in CtBP sumoylation [1,50]. Pc2 is a
member of the polycomb group of proteins, which
were first identified in Drosophila as regulators of seg-
ment identity [51,52]. Pc2 contains various domains.
The ubc9 binding domain is located in a CTD frag-

ment (Fig. 6B, amino acids 401–469) that neighbors a
CtBP-binding fragment called the Pro-Ile-Asp-Leu-Ser
(PIDLR or PIDLS) motif (Fig. 6B, amino acids 469–
486). At present, no structural data are available for
the ubc9-binding and substrate interaction regions. A
separate domain in the N-terminal domain has been
reported to show E3 activity in vitro. In vivo, both the
A

B
C
Fig. 6. Domain organization of E3 ligases. (A) Typical PIAS proteins
contain an N-terminal SAP domain, an SP-RING domain, a SUMO
interaction motif (SIM) and a C-terminal domain (CTD) [39]. (B) Pc2
contains an uncharacterized E3 domain, a ubc9 binding domain and
a Pro-Ile-Asp-Leu-Ser motif involved in substrate binding [53].
(C) RanBP2, modified from [56]. RanBP2 harbors an N-terminal
leucine-rich domain, four RanBP1 repeats (R1–R4), a region contain-
ing eight zinc-finger motifs and the SUMO interaction domain
RanBP2DFG, which consists of two internal repeats (IR1 and IR2)
and a linker (M) domain [56].
Interactions between sumoylation proteins Z. Tang et al.
3010 FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS
N- and C-terminal domains appear to contribute to E3
activity [53].
RanBP2
The nuclear pore complex protein RanBP2 has been
found to catalytically enhance sumoylation of a
nuclear body component [54]. It directly interacts with
ubc9 and strongly enhances SUMO-1 transfer from

ubc9 to the SUMO-1 target Sp100 [54]. RanBP2 is a
multidomain protein (Fig. 6C) with interaction sites
for proteins including nuclear transport receptors, the
GTPase Ran, ubc9, and sumoylated RanGAP1 [55].
Its SUMO E3 ligase activity lies in a fragment called
RanBP2DFG, which contains two approximately
50-residue internal repeats (IR1 and IR2) separated by
a 25-residue spacer domain (M) [54,56]. The IR1 is a
UBL protein binding domain that promotes sumoyla-
tion of both SUMO-1 and SUMO-2 [33]. The human
IR1 sequence has been crystallized in a complex with
human SUMO-1, human ubc9 and part of human
RanGAP1 (residues 432–587). Its N-terminal region
interacts with SUMO-1 (Fig. 5H), but its C-terminal
portion binds ubc9 (Fig. 5C). Hence, it was proposed
that the RanBP2DFG region acts to position the
SUMO–E2 thioester in an optimal orientation, thereby
enhancing conjugation [19]. This shorter fragment does
not display the substrate specificity seen with full-
length RanBP2, suggesting that sequences outside the
ligase active site control substrate recognition, possibly
by interacting with other proteins.
Approximately 100 proteins have been demonstrated
to serve as sumoylation substrates. Most were initially
identified as binding partners of PIAS proteins and ⁄ or
ubc9 by protein–protein interaction assays, such as
yeast two-hybrid and ⁄ or GST pulldown, and con-
firmed later to be sumoylated. For only four proteins,
to our knowledge, has sumoylation been shown to be
specifically enhanced by RanBP2. These are RanGAP1

[57,58], Sp100 [54], promyelocytic leukaemia protein
[33] and histone deacetylase 4 [59]. However, RanBP2
does not directly interact with either RanGAP1 or
Sp100 in pulldown assays [54,60], and a direct interac-
tion between RanBP2 and the other two substrates has
not been shown either.
In conclusion, RanBP2 works in a way that is differ-
ent from that of PIAS proteins or Pc2.
Sumoylation complexes and
mechanism
An intriguing question is how the individual compo-
nents of the sumoylation machinery interact during the
various steps of the modification cycle. Above, we
describe interactions involving E2 enzymes (Fig. 5A–
E). Structural analysis suggests that other interactions
are also crucial for sumoylation to proceed.
1. SUMO–E1 interaction at the adenylation domain
(Fig. 5F). No covalent bond exists between E1 and
SUMO1 at this site. Rather, the C-terminal Gly-Gly
motif is modified here by an ATP residing in the ade-
nylation domain, resulting in SUMO activation [15].
2. SUMO–E1 interaction at the catalytic domain.
Here, the first thioester bond is formed. No structural
data are available yet for SUMO bound to Aos1-uba2
at its catalytic cysteine residue. However, the corres-
ponding complex has been resolved for the neddylation
pathway [24]. In the Nedd8–APPBP1–uba3–ubc12
complex, one Nedd8 molecule is bound covalently to
the cysteine site of E1 [24] (Fig. 5G).
3. SUMO–E3 interaction. For RanBP2, the interac-

tion with SUMO-1 has been analyzed (Fig. 5H) and
shown to require the N-terminal region of this E3
ligase, i.e. the IR1 domain [19].
Competitive binding experiments have disclosed that,
in the ubiquitination and neddylation cascades, the E2
enzymes must dissociate from E1 prior to E3-catalyzed
transfer of the modifier to the substrate [61,62]. This
implies that distinct protein complexes must form dur-
ing the activation, conjugation and ligation reactions.
Consistent with this view, two large UBL complexes
have been crystallized and structurally resolved: a ned-
dylation E1 modifier E2 complex comprising APPBP1–
uba3–Nedd8(A)–Nedd8(T)–Mg-A TP–ubc12(C111A)
(Protein Data Bank number 2NVU), and a sumoyla-
tion E2–E3–substrate complex containing ubc9, the
IR1 and M domains of RanBP2, SUMO-1 and a
RanGAP1 fragment containing the sumoylation site
(Protein Data Bank number 1Z5S) [19,24]. We have
used the 2NVU template to model the corresponding
sumoylation E1–E2 complex (Z. T., unpublished data);
the result is shown together with the published struc-
ture of the E2–E3–substrate complex in Fig. 7B.
In the modelled E1–E2–SUMO complex (Fig. 7B,
bottom), the E1 enzyme Aos1-uba2 forms the core
structure. The E2 enzyme ubc9 is positioned on top of
E1, with its a1 area attached to the E1 UFD. SUMO-1
is predicted to be associated with the E1–E2 complex
via at least three distinct sites, two of which have been
localized in the corresponding neddylation complex
structure (2NVU). At the adenylation domain of E1,

adenylated SUMO-1 [SUMO(A)] is formed, and cova-
lent attachment to Cys173 at the catalytic domain gen-
erates E1-conjugated ‘thiolated’ SUMO-1 [SUMO(T1)]
(Fig. 7B). Both domains are located on uba2. In
addition, SUMO-1 must interact with ubc9 to form
Z. Tang et al. Interactions between sumoylation proteins
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3011
the second thioester bond during conjugation; how-
ever, a template structure for E2-conjugated SUMO-1
[SUMO(T2)] is not yet available.
Within the E2–E3–substrate complex co-crystallized
with SUMO-1 (Fig. 7B, top), the fragment of Ran-
GAP1, a well-characterized sumoylation substrate,
binds to ubc9 at its a3 region. The other subdomains
of ubc9 extend freely, leaving sufficient space for bind-
ing of other interacting proteins. The partial E3
sequence of RanBP2 interacts with both SUMO-1 and
ubc9 via its IR1 domain. Conjugation of SUMO-1 to
the acceptor Lys524 of RanGAP1 generates the final
product of the sumoylation cascade [SUMO(C)].
The modeled and resolved structures discussed
above allow rationalization of the protein interactions
occurring during the various steps of the sumoylation
reaction. Accordingly, a free SUMO-1 molecule enters
the adenylation site of E1 where its C-terminal GG
motif accesses an ATP bound at the E1 ATP-binding
motif to allow formation of SUMO(A). SUMO(A) is
then transferred to the E1 catalytic cysteine; this
implies that a distance of approximately 30 A
˚

between
the adenylation and catalytic sites is bridged, with
a90° turn of the SUMO-1 molecule being required
(Fig. 7A,B). After thioester bond formation,
SUMO(T1) must be further transferred to the catalytic
cysteine of E2. The details of this second (and the first)
transfer reaction are unknown; however, it is assumed
that conformational changes accompanying the adeny-
lation and thiolation steps allow molecular movement
within the enzyme–substrate complex. Successful con-
jugation is then thought to result in detachment of the
E2–SUMO-1(T2) conjugate from E1, allowing interac-
tion with specific E3 enzymes and their substrates.
Within the resulting E2–E3–substrate complexes,
SUMO-1 is then ligated to the respective substrate
protein. Here, we referred to structural data obtained
for the RanGAP1-containing E2–E3 complex; it
should, however, be noted that RanGAP1 is unusual
in that it can interact directly with the E2 ubc9. In
A
B
C
E2-E3-substrate complex
Fig. 7. Structures of sumoylation complexes. (A) Various orientations of SUMO-1 that correspond to its positions at the various SUMO-1
binding sites of the sumoylation complexes. (B) Model of the E1–E2 and structure of the E2–E3–substrate sumoylation complexes, with
SUMO-1 molecules bound ⁄ attached at various sites [orientations as shown in (A); the other components are colour-coded as shown in (C)].
The positions of the ATP-binding motif of E1 (dashed red circle), the catalytic cysteine residues of E1 (filled red rectangle) and E2 (black
arrows) and the conjugated lysine residue of RanGAP1 (green arrow) are indicated. The E1–E2 model was compiled from published struc-
tures using
WINCOOT [63] and PYMOL [64]. (C) Individual sumoylation components: E1 Aos1-uba2 (yellow), E2 ubc9 (cyan), the IR1–M domain

fragment of the E3 RanBP2 (magenta) and substrate RanGAP1 (residues 435–587; brown). The complex structures displayed in (B) indicate
the following sequence of SUMO-1 interactions. Bottom: a free SUMO molecule (violet) enters the adenylation site of E1, where its C-termi-
nal GG motif accesses an ATP molecule (firebrick red, within the dashed red circle) bound at the ATP-binding motif of E1 (dashed red circle),
and thus becomes adenylated. SUMO(A) is then translocated (distance approximately 30 A
˚
,90° turn) to the E1 catalytic site and attached to
its reactive cysteine (red rectangle) to generate SUMO(T1). Thereafter, SUMO(T1) is transferred to the reactive cysteine of the catalytic site
of E1-bound E2 to form SUMO(T2); a structure showing SUMO-1 conjugated to Cys93 of E2 is not yet available. Conjugation results in
detachment of E2–SUMO-1 from E1, which renders E2 accessible for interaction with E3 and substrates. In the resulting SUMO-1–E2–E3–
substrate complex (top), SUMO-1 is covalently attached to RanGAP1 (ligation area indicated by dashed black rectangle).
Interactions between sumoylation proteins Z. Tang et al.
3012 FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS
general, substrates are recruited by E3 ligases, which,
by docking onto ubc9, act as adaptors for distinct
target proteins.
Perspectives
The details of SUMO conjugation as outlined above
have emerged from the structures and paired interac-
tions characterized so far. These include SUMO-1, the
full-length E1 and E2 enzymes, as well as partial
sequences of the E3 RanBP2 and the substrate Ran-
GAP1. Interactions within the SUMO-1–E1–E2 com-
plex have been inferred from data obtained on the
related neddylation proteins [19,28,31], and require val-
idation in future studies. It should be noted that the
presently available biochemical data suggest different
roles for the RanBP2, Pc2 and PIAS proteins,
although all these E3 enzymes have been reported to
enhance sumoylation of specific substrates. Clearly
novel structural data will be required to better under-

stand the mechanistic roles of these ligases.
Future studies should also address the question of
how SUMO proteins are translocated between their
various sites of interaction with the sumoylation
enzymes. First, SUMO must move from the adenyla-
tion domain to the catalytic site within E1. For the
neddylation machinery, the distance between these sites
has been calculated to be about 20 A
˚
, and is suggested
to be reduced by conformational changes of the E1
enzyme [24]. In the E1–E2–SUMO-1 complex modeled
here, the distance between bound SUMO(A) and
SUMO(T1) is predicted to be about 30 A
˚
, and the dis-
tance between the E1 and E2 reactive cysteines is
deduced to be about 20 A
˚
. As its C-terminal loop is
flexible, even comparatively low extents of SUMO-1
re-orientation may suffice to allow thioester bond for-
mation between these residues. In the crystallized E2–
E3–substrate complex, Lys524 of RanGAP1 lies only
about 10 A
˚
away from the E2 cysteine; hence only a
modest axial movement of SUMO(T2) might bring it
into close proximity to the substrate’s acceptor residue.
Clearly structural studies that provide information

about additional intermediates of the sumoylation cas-
cade are required to fully understand the precise mech-
anisms of the activation, conjugation and ligation
reactions.
In conclusion, the structural data summarized in this
review have greatly helped to advance our understand-
ing of sumoylation and related protein modifications.
However, several sub-steps of the underlying catalytic
events are still poorly understood, and the mechanisms
that coordinate these sub-steps and regulate their
efficiencies await future investigation.
Acknowledgements
We thank Dr Lin Liu National Institutes of Health ⁄
National Institute of Diabetes and Digestive and Kid-
ney Diseases (NIH ⁄ NIDDK, Bethesda, MD, USA) for
help with structural modeling. We are also grateful to
the unknown reviewers who greatly helped us to
improve the initial version of this review. Our work is
supported by the Max-Planck-Gesellschaft, Deutsche
Forschungsgemeinschaft, European Community and
Fonds der Chemischen Industrie.
References
1 Johnson ES (2004) Protein modification by SUMO.
Annu Rev Biochem 73, 355–382.
2 Melchior F (2000) SUMO nonclassical ubiquitin. Annu
Rev Cell Dev Biol 16, 591–626.
3 Dohmen RJ (2004) SUMO protein modification.
Biochim Biophys Acta 1695, 113–131.
4 Gill G (2004) SUMO and ubiquitin in the nucleus: dif-
ferent functions, similar mechanisms? Genes Dev 18,

2046–2059.
5 Melchior F, Schergaut M & Pichler A (2003) SUMO:
ligases, isopeptidases and nuclear pores. Trends Biochem
Sci 28, 612–618.
6 Kerscher O, Felberbaum R & Hochstrasser M (2006)
Modification of proteins by ubiquitin and ubiquitin-like
proteins. Annu Rev Cell Dev Biol 22, 159–180.
7 Pickart CM (2001) Mechanisms underlying ubiquitina-
tion. Annu Rev Biochem 70, 503–533.
8 Mukhopadhyay D & Dasso M (2007) Modification in
reverse: the SUMO proteases. Trends Biochem Sci 32,
286–295.
9 Hicke L, Schubert HL & Hill CP (2005) Ubiquitin-
binding domains. Nat Rev Mol Cell Biol 6, 610–621.
10 Whitby FG, Xia G, Pickart CM & Hill CP (1998) Crys-
tal structure of the human ubiquitin-like protein
NEDD8 and interactions with ubiquitin pathway
enzymes. J Biol Chem 273, 34983–34991.
11 Bayer P, Arndt A, Metzger S, Mahajan R, Melchior F,
Jaenicke R & Becker J (1998) Structure determination
of the small ubiquitin-related modifier SUMO–1. J Mol
Biol 280, 275–286.
12 Huang WC, Ko TP, Li SS & Wang AH (2004) Crystal
structures of the human SUMO–2 protein at 1.6 A
˚
and
1.2 A
˚
resolution: implication on the functional differ-
ences of SUMO proteins. Eur J Biochem 271, 4114–4122.

13 Ding H, Xu Y, Chen Q, Dai H, Tang Y, Wu J & Shi Y
(2005) Solution structure of human SUMO–3 C47S and
its binding surface for Ubc9. Biochemistry 44, 2790–2799.
14 Vijay-Kumar S, Bugg CE & Cook WJ (1987) Structure
of ubiquitin refined at 1.8 A
˚
resolution. J Mol Biol 194,
531–544.
Z. Tang et al. Interactions between sumoylation proteins
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3013
15 Lois LM & Lima CD (2005) Structures of the SUMO
E1 provide mechanistic insights into SUMO activation
and E2 recruitment to E1. EMBO J 24, 439–451.
16 Baba D, Maita N, Jee JG, Uchimura Y, Saitoh H,
Sugasawa K, Hanaoka F, Tochio H, Hiroaki H & Shir-
akawa M. (2005) Crystal structure of thymine DNA
glycosylase conjugated to SUMO–1. Nature 435, 979–
982.
17 Pichler A, Knipscheer P, Oberhofer E, van Dijk WJ,
Korner R, Olsen JV, Jentsch S, Melchior F & Sixma
TK. (2005) SUMO modification of the ubiquitin-conju-
gating enzyme E2–25K. Nat Struct Mol Biol 12, 264–
269.
18 Reverter D & Lima CD (2004) A basis for SUMO pro-
tease specificity provided by analysis of human Senp2
and a Senp2–SUMO complex. Structure 12, 1519–1531.
19 Reverter D & Lima CD (2005) Insights into E3 ligase
activity revealed by a SUMO–RanGAP1–Ubc9–Nup358
complex. Nature 435, 687–692.
20 Song J, Zhang Z, Hu W & Chen Y (2005) Small ubiqu-

itin-like modifier (SUMO) recognition of a SUMO
binding motif: a reversal of the bound orientation.
J Biol Chem 280, 40122–40129.
21 Walden H, Podgorski MS, Huang DT, Miller DW,
Howard RJ, Minor DL Jr, Holton JM & Schulman BA
(2003) The structure of the APPBP1–UBA3–NEDD8–
ATP complex reveals the basis for selective ubiquitin-
like protein activation by an E1. Mol Cell 12,
1427–1437.
22 Dohmen RJ, Stappen R, McGrath JP, Forrova H,
Kolarov J, Goffeau A & Varshavsky A (1995) An
essential yeast gene encoding a homolog of ubiquitin-
activating enzyme. J Biol Chem 270, 18099–18109.
23 Szczepanowski RH, Filipek R & Bochtler M (2005)
Crystal structure of a fragment of mouse ubiquitin-
activating enzyme. J Biol Chem 280, 22006–22011.
24 Huang DT, Hunt HW, Zhuang M, Ohi MD, Holton
JM & Schulman BA (2007) Basis for a ubiquitin-like
protein thioester switch toggling E1–E2 affinity. Nature
445, 394–398.
25 Shen LN, Liu H, Dong C, Xirodimas D, Naismith JH
& Hay RT (2005) Structural basis of NEDD8 ubiquitin
discrimination by the deNEDDylating enzyme NEDP1.
EMBO J 24, 1341–1351.
26 Jentsch S (1992) The ubiquitin-conjugation system.
Annu Rev Genet 26, 179–207.
27 Tong H, Hateboer G, Perrakis A, Bernards R & Sixma
TK (1997) Crystal structure of murine ⁄ human Ubc9
provides insight into the variability of the ubiquitin-
conjugating system. J Biol Chem 272, 21381–21387.

28 Bernier-Villamor V, Sampson DA, Matunis MJ & Lima
CD (2002) Structural basis for E2-mediated SUMO
conjugation revealed by a complex between ubiquitin-
conjugating enzyme Ubc9 and RanGAP1. Cell 108,
345–356.
29 Huang DT, Miller DW, Mathew R, Cassell R, Holton
JM, Roussel MF & Schulman BA (2004) A unique
E1–E2 interaction required for optimal conjugation of
the ubiquitin-like protein NEDD8. Nat Struct Mol Biol
11, 927–935.
30 Huang DT, Paydar A, Zhuang M, Waddell MB, Hol-
ton JM & Schulman BA (2005) Structural basis for
recruitment of Ubc12 by an E2 binding domain in
NEDD8’s E1. Mol Cell 17, 341–350.
31 Zheng N, Wang P, Jeffrey PD & Pavletich NP (2000)
Structure of a c–Cbl–UbcH7 complex: RING domain
function in ubiquitin-protein ligases. Cell 102, 533–
539.
32 Huang L, Kinnucan E, Wang G, Beaudenon S, Howley
PM, Huibregtse JM & Pavletich NP (1999) Structure
of an E6AP–UbcH7 complex: insights into ubiquiti-
nation by the E2–E3 enzyme cascade. Science 286,
1321–1326.
33 Tatham MH, Kim S, Jaffray E, Song J, Chen Y & Hay
RT (2005) Unique binding interactions among Ubc9,
SUMO and RanBP2 reveal a mechanism for SUMO
paralog selection. Nat Struct Mol Biol 12, 67–74.
34 Tatham MH, Kim S, Yu B, Jaffray E, Song J, Zheng J,
Rodriguez MS, Hay RT & Chen Y (2003) Role of an
N–terminal site of Ubc9 in SUMO–1, -2, and -3 binding

and conjugation. Biochemistry 42, 9959–9969.
35 Liu Q, Jin C, Liao X, Shen Z, Chen DJ & Chen Y
(1999) The binding interface between an E2 (UBC9)
and a ubiquitin homologue (UBL1). J Biol Chem 274,
16979–16987.
36 Wang J, Hu W, Cai S, Lee B, Song J & Chen Y (2007)
The intrinsic affinity between E2 and the Cys domain of
E1 in ubiquitin-like modifications. Mol Cell 27, 228–
237.
37 Knipscheer P, van Dijk WJ, Olsen JV, Mann M & Six-
ma TK (2007) Noncovalent interaction between Ubc9
and SUMO promotes SUMO chain formation. EMBO
J 26, 2797–2807.
38 Capili AD & Lima CD (2007) Structure and analysis of
a complex between SUMO and Ubc9 illustrates features
of a conserved E2–Ubl interaction. J Mol Biol 369,
608–618.
39 Sharrocks AD (2006) PIAS proteins and transcriptional
regulation – more than just SUMO E3 ligases? Genes
Dev 20, 754–758.
40 Okubo S, Hara F, Tsuchida Y, Shimotakahara S,
Suzuki S, Hatanaka H, Yokoyama S, Tanaka H,
Yasuda H & Shindo H. (2004) NMR structure of the
N–terminal domain of SUMO ligase PIAS1 and its
interaction with tumor suppressor p53 and A ⁄ T–rich
DNA oligomers. J Biol Chem 279, 31455–31461.
41 Sachdev S, Bruhn L, Sieber H, Pichler A, Melchior F &
Grosschedl R (2001) PIASy, a nuclear matrix-associated
SUMO E3 ligase, represses LEF1 activity by sequestra-
tion into nuclear bodies. Genes Dev 15, 3088–3103.

Interactions between sumoylation proteins Z. Tang et al.
3014 FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS
42 Minty A, Dumont X, Kaghad M & Caput D (2000)
Covalent modification of p73alpha by SUMO–1. Two-
hybrid screening with p73 identifies novel SUMO–1-
interacting proteins and a SUMO–1 interaction motif.
J Biol Chem 275, 36316–36323.
43 Kahyo T, Nishida T & Yasuda H (2001) Involvement
of PIAS1 in the sumoylation of tumor suppressor p53.
Mol Cell 8, 713–718.
44 Tang Z, El Far O, Betz H & Scheschonka A
(2005) PIAS1 interaction and sumoylation of metabo-
tropic glutamate receptor 8. J Biol Chem 280, 38153–
38159.
45 Nakagawa K & Yokosawa H (2002) PIAS3 induces
SUMO-1 modification and transcriptional repression of
IRF-1. FEBS Lett 530, 204–208.
46 Zhao X & Blobel G (2005) A SUMO ligase is part of a
nuclear multiprotein complex that affects DNA repair
and chromosomal organization. Proc Natl Acad Sci
USA 102, 4777–4782.
47 Potts PR & Yu H (2005) Human MMS21 ⁄ NSE2 is a
SUMO ligase required for DNA repair. Mol Cell Biol
25, 7021–7032.
48 Weger S, Hammer E & Heilbronn R (2005) Topors acts
as a SUMO–1 E3 ligase for p53 in vitro and in vivo.
FEBS Lett
49 Lin X, Sun B, Liang M, Liang YY, Gast A, Hildebrand
J, Brunicardi FC, Melchior F & Feng XH. (2003)
Opposed regulation of corepressor CtBP by SUMOyla-

tion and PDZ binding. Mol Cell 11, 1389–1396.
50 Kagey MH, Melhuish TA & Wotton D (2003) The poly-
comb protein Pc2 is a SUMO E3. Cell 113, 127–137.
51 Simon J, Chiang A & Bender W (1992) Ten different
Polycomb group genes are required for spatial control
of the abdA and AbdB homeotic products. Development
114, 493–505.
52 Simon JA & Tamkun JW (2002) Programming off and
on states in chromatin: mechanisms of Polycomb and
trithorax group complexes. Curr Opin Genet Dev 12,
210–218.
53 Kagey MH, Melhuish TA, Powers SE & Wotton D
(2005) Multiple activities contribute to Pc2 E3 function.
EMBO J 24, 108–119.
54 Pichler A, Gast A, Seeler JS, Dejean A & Melchior F
(2002) The nucleoporin RanBP2 has SUMO1 E3 ligase
activity. Cell 108, 109–120.
55 Pichler A, Knipscheer P, Saitoh H, Sixma TK & Mel-
chior F (2004) The RanBP2 SUMO E3 ligase is neither
HECT- nor RING-type. Nat Struct Mol Biol 11, 984–
991.
56 Matunis MJ & Pickart CM (2005) Beginning at the end
with SUMO. Nat Struct Mol Biol 12, 565–566.
57 Mahajan R, Delphin C, Guan T, Gerace L & Melchior
F (1997) A small ubiquitin-related polypeptide involved
in targeting RanGAP1 to nuclear pore complex protein
RanBP2. Cell 88, 97–107.
58 Matunis MJ, Coutavas E & Blobel G (1996) A novel
ubiquitin-like modification modulates the partitioning
of the Ran-GTPase-activating protein RanGAP1

between the cytosol and the nuclear pore complex.
J Cell Biol 135, 1457–1470.
59 Kirsh O, Seeler JS, Pichler A, Gast A, Muller S, Miska
E, Mathieu M, Harel-Bellan A, Kouzarides T, Melchior
F & Dejean A. (2002) The SUMO E3 ligase RanBP2
promotes modification of the HDAC4 deacetylase.
EMBO J 21, 2682–2691.
60 Matunis MJ, Wu J & Blobel G (1998) SUMO–1 modifi-
cation and its role in targeting the Ran GTPase-activat-
ing protein, RanGAP1, to the nuclear pore complex.
J Cell Biol 140, 499–509.
61 Eletr ZM, Huang DT, Duda DM, Schulman BA &
Kuhlman B (2005) E2 conjugating enzymes must
disengage from their E1 enzymes before E3-dependent
ubiquitin and ubiquitin-like transfer. Nat Struct Mol
Biol 12, 933–934.
62 Dye BT & Schulman BA (2007) Structural mechanisms
underlying posttranslational modification by ubiquitin-
like proteins. Annu Rev Biophys Biomol Struct 36,
131–150.
63 Emsley P & Cowtan K (2004) Coot: model-building
tools for molecular graphics. Acta crystallogr 60, 2126–
2132.
64 DeLano WL (2002) The PyMOL Molecular Graphics
System. DeLano Scientific, San Carlos, CA.
65 Reverter D, Wu K, Erdene TG, Pan ZQ, Wilkinson
KD & Lima CD (2005) Structure of a complex between
Nedd8 and the Ulp ⁄ Senp protease family member
Den1. J mol biol 345, 141–151.
66 Reverter D & Lima CD (2004) A basis for SUMO pro-

tease specificity provided by analysis of human Senp2
and a Senp2-SUMO complex. Structure 12, 1519–1531.
67 Mossessova E & Lima CD (2000) Ulp1-SUMO crystal
structure and genetic analysis reveal conserved interac-
tions and a regulatory element essential for cell growth
in yeast. Mol cell 5, 865–876.
Z. Tang et al. Interactions between sumoylation proteins
FEBS Journal 275 (2008) 3003–3015 ª 2008 The Authors Journal compilation ª 2008 FEBS 3015