This journal is c The Royal Society of Chemistry 2012 Mol. BioSyst., 2012, 8, 1723–1729 1723

Cite this: Mol. BioSyst., 2012, 8, 1723–1729

Systematic determinations of SUMOylation activation intermediates and

dynamics by a sensitive and quantitative FRET assay

Yang Songaand Jiayu Liao*

abc

Received 10th November 2011, Accepted 12th March 2012

DOI: 10.1039/c2mb05465e

Ubiquitination and SUMOylation are multi-step cascade reactions, in which small protein

modifiers are activated by E1 activating enzyme, transferred to E2 conjugating enzyme, and

conjugated to substrates mediated by the E3 ligase in vivo. The structural and biochemical bases

for the cascade reactions have been elucidated by several studies. However, the reaction

intermediates and dynamics of these peptide modifiers among the enzymes have not been

completely elucidated. Here we report detailed investigations of SUMOylation dynamics and

interaction switches of SUMO1 among its ligases using FRET technology. These studies show

that, while SUMO1 and the E1 subunit Aos1 or Uba2 have no intrinsic affinity for each other,

the adenylation of SUMO1 carried out by Aos1 requires the presence of Uba2, and subsequently

conformational changes trigger the interaction of SUMO1 and Uba2 for a thioester bond

formation. The reaction intermediates among SUMO1 and its ligases are indirectly revealed by

FRET signals generated by each pair. Furthermore, the transfer of SUMO1 from Uba2 to E2

enzyme, Ubc9, depends on the formation of a thioester bond between SUMO1 and Ubc9, and

requires non-covalent interaction between Ubc9 and Uba2, but not between Ubc9 and SUMO1.

These interaction switches provide the physical and biochemical bases for the SUMO activation

and a transfer cascade required for SUMO activation.

Introduction

Post-translational modification by Ubiquitin (Ub) and

Ubiquitin-like (Ubls) proteins, such as SUMO and NEDD8,

plays critical roles in many physiological and pathological

processes in eukaryotes, such as signal transductions, genome

integrity maintenance, protein transport, cell cycles, immunity

and tumorigenesis.1–7 The conjugation of Ubiquitin, SUMO

and NEDD8 to target substrates shares an evolutionary

conserved but distinct enzyme cascade that involves the sequential

actions of E1, E2 and E3 ligases. For example, after maturation

of SUMO from pre-SUMO by the cleavage processing of SENPs,

SUMO is conjugated to its substrate by a multienzyme-catalyzed

cascade involving E1, E2 and E3 ligases (Fig. 1). The heterodimer

E1, Aos1/Uba2, activates Ub/Ubls in two major steps. First,

Aos1 catalyzes adenylation of the SUMO at its C-terminus using

ATP, with the help ofMg2+. Second, a catalytic cysteine of Uba2

forms a covalent thioester bond with the C-terminus of SUMOby

replacing the adenylate group.8,9 Then, a transthiolation reaction

takes place between E1 and E2, resulting in the formation of a

Fig. 1 SUMOylation and deSUMOylation cascade. The pre-SUMO is

processed into a mature form by a cleavage of C-terminal peptide by

SENPs. The matured SUMO is then activated by heterodimer E1 ligase,

Aos1/Uba2, in two steps, adenylation and thioester bond formation. The

E1-activated SUMO is transferred to E2, Ubc9, to form thioester with

Ubc9 in a transthiolation reaction. With the help of E3 ligases, such as

PIAS, the SUMO is conjugated to the substrate at the Lys residue.

aDepartment of Bioengineering, Bourns College of Engineering,University of California at Riverside, 900 University Avenue,Riverside, CA 92521, USA. E-mail: jiayu.liao@ucr.edu;Fax: +1 951-827-6416; Tel: +1 951-827-6240

b The Center for Bioengineering Research, Bourns College ofEngineering, University of California at Riverside,900 University Avenue, Riverside, CA 92521, USA

c Institute for Integrative Genome Biology, Bourns College ofEngineering, University of California at Riverside,900 University Avenue, Riverside, CA 92521, USA

MolecularBioSystems

Dynamic Article Links

www.rsc.org/molecularbiosystems PAPER

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online / Journal Homepage / Table of Contents for this issue

1724 Mol. BioSyst., 2012, 8, 1723–1729 This journal is c The Royal Society of Chemistry 2012

thioester bond between Ubc9 catalytic cysteine and SUMO.8 In

combination with a variety of E3s, such as PIAS family of

ligase, which are generally considered to determine substrate

specificity in vivo, SUMO is finally conjugated to lysine residues

of substrates.1–3 The SUMOylated substrate can also be

deSUMOylated by the same family of protease, SENPs (Fig. 1).

Several studies have revealed insights into the biochemical

and structural bases of Ub/Ubl with E1 and E2 enzyme

recognition and successive steps in substrate conjugation.10,11

First, biochemical studies show that E1 and E2s have different

affinities for each other in their free or Ub/Ubl thioester-linked

form. Free E1 has low affinity for E2s, while double Ub-loaded

E1 binds E2s at higher affinity.11,12 These differential affinities

between different enzyme forms may contribute to the Ub/Ubl

progression along the E1–E2–E3 cascade. Another study suggests

that there is an intrinsic non-covalent interaction between

SUMO and its E2, Ubc9, and this non-covalent interaction

is required for polySUMOylation of substrates but not for

monoSUMOylation.13 Second, structures of E1 of NEDD8,

SUMO or Ubiquitin alone or in complexes with substrate,

respectively, display similar domain orientations.14–17 Interestingly,

the structure of an activation complex of NEDD8 containing

heterodimeric E1, double-loaded NEDD8 (one bound non-

covalently in the adenylation packet and one linked to an E1

catalytic Cys domain via the thioester bond), and the NEDD8

E2 conjugating enzyme Ubc12 bound to the UFD domain

revealed two E2 binding sites on E1.17 One binding site comes

from conformational change of E1 after binding by NEDD8,

and the other binding site comes from E1 and NEDD8

directly. The speculation that a conformational change of E1

upon the binding of Ubl may provide a mechanism to bring E1

and E2 catalytic cysteine domains to a close proximity has been

further demonstrated by a recent structural study of SUMO–E1

adenylate and tetrahedral intermediate analogues.18 A striking

conformational change of E1 has been observed after the thioester

bond formation between SUMO and E1. This conformational

change may potentially bring two catalytic Cys domains of E1

and E2 within 3.5 A which will facilitate the transfer of SUMO

from E1 to E2.18 Although much knowledge about the Ub/Ubl

handoff among ligases has been gained from these studies, the

reaction intermediates and dynamics in the sequential formation

of transient complexes along the E1–E2–E3 conjugation cascade

have not been observed in a real dynamic mode. In addition,

mechanistic mechanisms of Ubl activation and switches between

E1 and E2 ligases have not been completely determined. Here,

we determined the intermediates and reaction dynamics of

SUMO1 with its E1s and E2 ligases and revealed the mechanistic

mechanisms of SUMO activation and switches among ligases

using the sensitive FRET technology.

Results

Design of the FRET-based system for monitoring the

SUMOylation cascade and FRET index definition

To observe and determine intermediates resulting from non-

covalent or covalent conjugation of SUMO1 with E1 or E2,

dynamics during Ub/Ubl adenylation, thioester bond formations,

and SUMO transfer between E1 and E2 in real time, we used a

Forster Resonance Energy Transfer (FRET) technique, which

involves the FRET pair CyPet- and YPet-labeled SUMO1,

Aos1 (E1 subunit), Uba2 (E1 subunit), Ubc9 (E2 subunit), the

C-terminal domain (420–587) of RanGAP1 (RanGAP1C,

substrate) and their mutant proteins which lose different

activities in the SUMOylation cascade19–21 (Fig. 2A). When

a CyPet-tagged protein is mixed with a YPet-tagged protein

in vitro, interaction between two proteins can draw the FRET

pair close to each other and result in two emission peaks at 475 nm

and 530 nm when the mixture is excited at 414 nm. We recently

developed a new spectrum analysis method for FRET signal

analysis.22 In this method, the emission signal at 530 nm emission

(Emtotal) is divided into three components: direct emission from

the donor CyPet (CyPet [cont]), direct emission from the acceptor

YPet (YPet [cont]), and FRET emission from YPet (EmFRET)

(Fig. 2B). The direct emission of CyPet and YPet at 530 nm

(Ex = 414 nm) is proportional to their emission at 475 nm

(Ex = 414 nm) (FLDD) and 530 nm (Ex = 475 nm) (FLAA)

(Fig. 2C) by a constant ratio, namely x and y, which are 0.378 and

0.026, respectively. Therefore, we can calculate the CyPet (cont)

and YPet (cont), and subtract them from Emtotal to derive

EmFRET. In order to use FRET to monitor the change of

protein–protein interaction status, we define the FRET index r

to be the ratio of FRET emission (EmFRET) to the emission at

530 nm (Ex = 475 nm) (FLAA) (Fig. 2D). Since FLAA is

contributed by direct emission of YPet, the FRET index can

indicate the relative percentage of YPet that receives energy from

excited CyPet, and therefore the FRET index can be used as

quantitative measurement of the FRET signal.22 When we mix

the proteins involved in the SUMOylation process and initiate the

conjugation cascade, the formation of protein complexes will lead

to an increase of the FRET index whereas disruption of protein

complexes will result in a decrease of the FRET index.

Revealing intermediates of SUMO/E1 ligase and dynamics

The structures of E1 identified residues from both Aos1 and

Uba2 that contact Mg2+/ATP and the Ub/Ubl C-terminal

Fig. 2 Design of the FRET assay for protein interactions in SUMOyla-

tion and FRET index definition. (A) CyPet-tagged SUMO1 interaction

YPet-tagged SUMO ligases and substrate. (B) Analysis of fluorescence

emission spectra of the mixtures of CyPet and YPet-tagged proteins when

excited at 414 nm. (C) The emission of FRET acceptor YPet proteins when

excited at 475 nm. (D) Definition of the FRET index.

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online

This journal is c The Royal Society of Chemistry 2012 Mol. BioSyst., 2012, 8, 1723–1729 1725

glycine.14–16,23 We speculated that there would be intrinsic

interactions between SUMO1 and its E1, Aos1/Uba2. We first

determined the interactions and intermediates of SUMO1 with

its heterodimeric E1, which initiate the SUMOylation cascade

and catalyze two critical reactions, adenylation with the help

of ATP and thioester bond formation with one of the E1

subunits, Uba2. To our surprise, when SUMO1 and Aos1

were fused with CyPet and YPet, respectively, the FRET assay

showed that there was no intrinsic interaction between Aos1

and SUMO1 even in the presence of ATP (Fig. 3A). Only a

weak interaction was observed when only Uba2, but not ATP,

was added. However, a strong interaction was observed after

both Uba2 and ATP were added (Fig. 3A). This suggests that

SUMO recognition and/or adenylation of SUMO peptide by

Aos1 cannot take place by itself. Similar results were observed

when SUMO1 and Uba2 were labeled with the CyPet and

YPet pair, respectively. In consistence with the above results,

none or only weak interaction was observed in the absence or

presence of Aos1 alone. However, a strong interaction and an

interaction intermediate were observed only after Aos1 and

ATP were added to the reaction mixture either at the starting

point or after a period of waiting (Fig. 3B). These results

strongly suggest that there are no intrinsic affinities among

SUMO1 and its E1, and a strong interaction is induced

by Aos1 and Uba2 after they bind to ATP and catalyze

the adenylation reaction and thioester bond formation. Inter-

estingly, the FRET peak between SUMO1 and Uba2 decays

much faster than that between SUMO1 and Aos1, indicating

that either the SUMO1–Uba2 thioester bond may not be very

stable, or there could be a conformation change of the

SUMO1–Uba2 intermediate that increases the distance between

the FRET pair. These results demonstrate that striking

dynamic movements occur between E1 and SUMO1 during

adenylation and thioester bond formation, which are consistent

with the recent results of E1 and SUMO1 intermediate crystal

complexes.18

After adenylation, the cysteine from the catalytic domain of

Uba2 will form a thioester bond with the C-terminal Gly of

SUMO1 peptide, forming a high energy covalent intermediate

before it is transferred to the cysteine catalytic domain of E2.

We then investigated whether the interaction between SUMO1

and Uba2 was due to the covalent thioester intermediate

formation or just non-covalent interactions induced by E1

remodelling. A mutant of Uba2 with its catalytic cysteine

changed to serine (C173S) was generated. The interaction

between Uba2 (C173S) mutant and SUMO1 was strong and

more stable than its wild type counterpart (Fig. 3C), indicating

that the formation of thioester covalent intermediate does not

contribute significantly to the interaction observed above. This is

in major contrast to the SUMO1 interaction with its E2, Ubc9,

where thioester covalent interaction contributes significantly to

the interaction (see below). Compared to the transient interaction

between wild type Uba2 and SUMO1, the more stable

complex between Uba2 (C173S) mutant and SUMO1 suggests

that SUMO1 binds to the Aos1/Uba2 complex after being

activated by Aos1 and will not be released if the thioester bond

is not formed.

Wild-type activation complexes containing an E1, a Ubl and

an E2 are not stable, because thioester-bonding Ubl is readily

passed from the E1’s catalytic cysteine to the catalytic cysteine

of E2, and the intermediate of Ubl and E2 is not stable either,

because it is readily passed to substrates. Therefore, an

activation complex trapping strategy was used to determine

the crystal structure of the NEDD8 pathway: APPBP1–UBA3

two NEDD8s (NEDD8 (T) thioester-linked to UBA3 and

NEDD8 (A) noncovalently associated at the adenylation

active site), Mg2+/ATP, and a catalytic cysteine-to-alanine

(C111A) mutant of Ubc12. It would be interesting to determine

how SUMO1–Uba2 active complex forms in the presence of

Ubc9. Consistent with the previous result, the SUMO1–Uba2

active complex reached a peak concentration in the presence of

Aos1 and ATP, and then decayed to the background fluorescence

of two separated proteins over thirty minutes (Fig. 3D, filled

circles). In the presence of Ubc9, the FRET signal of SUMO1–

Uba2 active complex did not reach the peak in contrast to that

it does in the absence of Ubc9, but gradually increased over

time, suggesting a quick transfer of thioester–SUMO1 to Ubc9

(Fig. 3D, open circles). Interestingly, in the presence of catalytic

deficient mutant of Ubc9 (C93S), the interaction between

SUMO1 and Uba2 became more stable, but eventually decayed

(Fig. 3D, filled squares). These results suggest that Ubc9 stabilizes

the SUMO1–Uba2 active complex and two intermediates are

formed in the presence of wild type or catalytic cysteine mutant

E2, suggesting a critical role of E2’s catalytic cysteine in inducing

the transfer of SUMO1 from Uba2.

Determinations of two levels of interaction between SUMO and

E2 ligase, Ubc9

SUMO1 has been shown to interact strongly with Ubc9 in a

non-covalent manner and form a thioester bond with catalytic

cysteine of Ubc9.13 To determine whether these two interactions

can be distinguished by FRET technology, we analyzed the

interaction between SUMO1 and Ubc9, which were tagged with

CyPet and YPet, respectively. In the absence of E1, SUMO1

Fig. 3 Intermediate formation and stability of SUMO1 with E1

ligase subunits, Aos1 and Uba2. (A) Specific binding of Aos1 to

SUMO1 requires its dimerization with Uba2 and ATP. (B) Specific

binding of SUMO1 to Uba2 requires its adenylation mediated by Aos1

and ATP. (C) Thioester bond formation is not required for the

interaction between SUMO1 and Uba2. (D) Ubc9 stabilizes the

Uba2 and SUMO1 intermediate complex.

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online

1726 Mol. BioSyst., 2012, 8, 1723–1729 This journal is c The Royal Society of Chemistry 2012

can interact with Ubc9 non-covalently. To demonstrate that

this is a truly non-covalent interaction, we added increasing

amounts of non-tagged SUMO1 to the mixture of CyPet–

SUMO1 and YPet–Ubc9. The FRET signal gradually decreased

with the increasing amounts of SUMO1, indicating that the

complex was disrupted by the competitor (Fig. 4A). In the presence

of E1 (Aos1 and Uba2) and ATP, SUMO1 interacted with and

formed a thioester-mediated covalent complex with Ubc9, shown

as a peak of FRET index which can be clearly distinguished from

the baseline indicating the non-covalent interaction of SUMO1 and

Ubc9 (Fig. 4B, filled circles). This thioester-mediated complex

could not form in the absence of ATP and E1. In contrast, the

catalytic cysteine mutant of Ubc9 (C93S), which could not

form a thioester bond with SUMO1, could only interact with

SUMO1 and form a non-covalent complex (Fig. 4B filled

squares). Furthermore, the thioester-mediated covalent complex

can only be formed in the presence of E1 and E2, but not the

catalytic cysteine mutant Uba2 (Fig. 4C). These results suggest

that the coordination of enzyme catalyzed cascade reaction is

required and two levels of interaction between Ubc9 and

SUMO1 can be distinguished by the FRET assay.

However, it remains unclear whether the intrinsic affinity

between Ubc9 and SUMO1 is required for SUMO1 peptide

transfer from E1 to E2, or Ubc9–SUMO thioester bond

formation, or whether it is required merely to maintain the

stable covalent active complex of SUMO1 and Ubc9. To

assess the roles of non-covalent interaction between Ubc9

and SUMO1, mutations of Ubc9 or SUMO1 that would

abolish the non-covalent interaction with each other could

provide tools addressing this question. The previous studies

have shown that the mutation of R17E on Ubc9 abolished its

interaction activity with both E1 and SUMO1 and the mutation of

E67R orG68Y on SUMO1 either completely or partially abolished

the interaction with Ubc9.13 We therefore generated a mutant

Ubc9 (R17E) and a double mutant of SUMO1 (E67R/G68Y).

The mutant Ubc9 (R17E) and the catalytic cysteine mutant of

Ubc9 were obtained from bacterial cells (Fig. 5A). Both Ubc9

(R17E) and SUMO1 (E67R/G68Y) lost their intrinsic affinity

to its partners, respectively (Fig. 5B). Most strikingly, the

SUMO1 (E67R/G68Y) still could interact with and form a

strong thioester covalent intermediate with Ubc9, as good, if

not better, as wild-type SUMO1 (Fig. 5C, filled triangles). In

comparison, SUMO1 could only form a non-covalent complex

with the catalytic cysteine mutant of Ubc9 (C93S) or in the absence

of ATP (Fig. 5C, filled squares and empty circles). Notably,

SUMO1 (E67R/G68Y) formed the active covalent complex with

Ubc9 directly from free status, while wild-type SUMO1 started

from a non-covalent intermediate stage as evidenced by the FRET

index around 0.1, a signal for the non-covalent complex (Fig. 5C,

filled triangles). Not too surprisingly, Ubc9 (R17E) failed to form a

thioester covalent intermediate with SUMO1 (Fig. 5C, empty

squares). This probably is due to its inability to interact with

Uba2, suggesting a critical role of the interaction between Uba2

and Ubc9 for the transfer of SUMO1 from E1 to E2. These results

indicate that the non-covalent interaction between SUMO1 and

Ubc9 is not required for its switch from E1 to E2 or the formation

of thioester covalent intermediate, while as the interaction of Ubc9

and Uba2 is critical for the switch.

SUMO conjugation requires thioester bond switch and physical

interaction between E1 and E2

We next turned our attention to investigate the significances of

these molecular interactions and intermediates of SUMO1/E1

or SUMO1/E2 in the SUMOylation cascade as substrate

conjugation. We were able to determine the final SUMO

conjugation product by determining the FRET signal from

CyPet–SUMO1 and YPet–RanGAP1c, indicating E1’s and E2’s

substrate tolerance of tagged SUMO in the cascade reaction

(Fig. 6A, filled circles). Interestingly, SUMO1 (E67R/G68Y)

could be conjugated to RanGAP1c in the presence of E1, E2

Fig. 4 Non-covalent and covalent intermediates of SUMO1 with E2

ligase, Ubc9. (A) SUMO1 non-covalently interacts with Ubc9 in the

absence of E1. (B) FRET reveals non-covalent and covalent inter-

actions between SUMO1 and Ubc9 in the presence of E1 and ATP.

(C) Thioester bond-mediated interaction of SUMO1 and Ubc9

requires E1, ATP, and the formation of a high-energy bond between

SUMO1 and Uba2.

Fig. 5 Non-covalent interaction between Ubc9 and SUMO1 is not

required for its transfer from E1 to E2. (A) Protein gel of mutant Ubc9

and wild-type Ubc9 expressed in bacterial cell. (B) Mutation of

SUMO1 at E67R/G68Y or mutation of Ubc9 at R17E disrupts the

non-covalent interaction between SUMO1 and Ubc9. (C) Non-covalent

interaction of Ubc9 with Uba2 but not SUMO1 is required to transfer

SUMO1 from E1 to E2.

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online

This journal is c The Royal Society of Chemistry 2012 Mol. BioSyst., 2012, 8, 1723–1729 1727

and ATP (Fig. 6A, filled squares). This result further suggests

that the non-covalent interaction between SUMO1 and Ubc9 is

not required for its activation by E2, and thereafter, conjugation

to substrates. The catalytic cysteine mutations of Uba2 (C173S)

and Ubc9 (C93S), respectively, abolish the SUMO1 activation

and substrate conjugation (Fig. 6B), indicating the key roles of

thioester intermediate formation between SUMO1 and Uba2 or

Ubc9 for its final conjugation to the substrate. It has been

suggested that the interaction between E1 and E2 is important

for SUMO–Ubc9 thioester formation.24 We went further to

investigate whether this interaction is required for the SUMO

activation cycle. Several mutations and mutation combinations

in Ubc9, including R17E, R13E/F22A, R13E/H20D/F22A,

were chosen based on their co-crystal structure and previous

studies.13 All these mutations abolished the SUMO conjugation

to RanGAP1c (Fig. 5C). Wild-type of SUMO1 and mutant

SUMO1 (E67R/G68Y) were able to conjugate to the substrate,

while all the mutations in E1 and E2 ligases, including Uba2

(C173S), Ubc9 (C93S), Ubc9 (R17E), Ubc9 (R13E/F22A), and

Ubc9 (R13E/H20D/F22A), failed to conjugate SUMO1 to

RanGAP1c (Fig. 6C). These results suggest that intermediates

of SUMO1/E1 and SUMO1/E2, and interactions between Uba2

and Ubc9, are critical for SUMO1 activation along the E1–E2

pathway and thereafter final substrate conjugation.

Discussion

Our results revealed the thioester intermediates and interaction

affinity switches of SUMO1-E1 and -E2 enzymes using dynamic

FRET technology in real time. In order for the SUMOylation

cascade to proceed, sequential formation of transient complexes

during E1–E2–E3 conjugation cascades is mediated by distinct

transient interactions and catalytic activities by the different

players.25 The reaction dynamics revealed by the FRET signal

have shown marked both conformational changes and interactive

partner switches that accompany adenylation and thioester bond

formation. The FRET signal between SUMO1 and Aos1 is only

observed in the presence of another E1 subunit, Uba2 and ATP,

indicating that heterodimer formation of E1s is absolutely

required for the adenylation reaction to take place. The

interaction of SUMO1 and Aos1 needs a signal directly from

Uba2, suggesting a conformational change of Aos1 in the

presence of Uba2. The gradual decrease of FRET signal

between CyPet–SUMO1 and YPet–Uba2 suggests that the

thioester intermediate of SUMO1 and Uba2, in the presence of

Aos1 and ATP, is not stable (Fig. 2B and C). Interestingly, a

stable complex is observed between SUMO1 and the Uba2

cysteine mutant (C173S), which would not form a thioester

intermediate with SUMO1 (Fig. 2C), indicating the thioester

bond formation is required for the leaving of SUMO1 from

Uba2. This difference between the two SUMO1/Uba2 complexes

also suggests that a conformational change of Uba2 takes place

after the SUMO1/Uba2 thioester intermediate formation and

this conformational change is required for the leaving of SUMO1

in the presence or absence of E2, Ubc9 (Fig. 1C and D). This

conclusion is also supported by the fact that intrinsic inter-

action affinity between SUMO1 and Ubc9 is not required for

the switch of SUMO1 from Uba2 to Ubc9 (Fig. 4B and C).

This switch between E1 and E2 seems solely promoted by

conformational changes of E1 triggered by thioester inter-

mediate formation and proximity of Ubc9 Cysteine to

SUMO1 C-terminus. These results are consistent with recent

co-crystal studies of both SUMO1 with its E1 and NEDD8

with its E1s and E2.17,18

Our studies also demonstrate an interesting example that

the tracking of FRET signals can reveal molecular interactions

and dynamics of the reaction cascade involving multiple

proteins. To the best of our knowledge, this is the first time

that the Ubl-ligase thioester intermediate and dynamics of

transfer can be observed in real time. The quantitative estimation

of different affinities for the transient complexes in the conjugation

cascade is hard to determine using other approaches because

the thioester intermediates are chemically not stable. They

could be captured by co-crystals using non-hydrolysable adenylate

mimics.18 But this approach cannot reveal intermediates and

reaction dynamics in real time. The E1 heterodimer-required

interaction with SUMO1 is uniquely detected by FRET

technology. We also observed that the affinity of SUMO1–

Ubc9 thioester is much higher than that of SUMO1–Uba2

thioester (FRET index in Fig. 2 and 5). This may serve the

mechanism of the SUMO peptide switch from E1 to E2. The

observations of reaction intermediates using FRET technology

provide another novel way to track dynamics/kinetics and

sequential order of an enzymatic reaction, and the FRET-

based reaction analysis technology can also be applied to

multicomponent-involved biochemical reactions.

The intermediate detections and dynamics analysis of the

SUMOylation cascade provides frameworks to understand not

only the SUMO activation cycle but also other Ub/Ubls cascades.26

Whether all the Ubls utilize the same mechanisms for activations or

not needs to be determined. In our studies, we have not included the

E3 ligase because in in vitro studies, E1s and E2 seem sufficient for

substrate conjugation.More complete understanding of the detailed

molecular mechanisms underlying the functional interactions

among Ubls, its E3 ligase and the substrate circuitry awaits

future experimentation.

Fig. 6 The stable substrate conjugate requires thioester bond switch

between E1 and E2. (A) SUMO1 conjugation to RanGAP1 does not

require the non-covalent interaction between SUMO1 and Ubc9.

(B) SUMO1 conjugation to RanGAP1 requires thioester bond

formations between SUMO1 and Uba2 or Ubc9. (C) Western blotting

confirms the inability of mutant Ubc9 to mediate the SUMO conjuga-

tion of RanGAP1.

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online

1728 Mol. BioSyst., 2012, 8, 1723–1729 This journal is c The Royal Society of Chemistry 2012

Materials and methods

DNA constructs

The open reading frames of CyPet and YPet were amplified by

using primers containing NheI or SalI sites, respectively, and

the open reading frames of SUMO1, Aos1, Uba2, Ubc9 and

the C-terminal domain (420–587) of RanGAP1 (RanGAP1C)

were amplified by PCR using primers containing SalI or NotI

sites, respectively. All PCR products were cloned into the

pCRII-TOPO vector (Invitrogen). The fragments encoding

SUMO1, RanGAP1C and SUMO ligases were digested with

SalI/NotI and the purified fragments were inserted into

pCRII-CyPet or pCRII-YPet plasmids which had been linearized

by SalI and NotI. After the sequences were confirmed, the

cDNAs encoding fusion proteins were digested with NheI/NotI

and cloned into the NheI/NotI sites of the pET28(b) vector

(Novagen). To get non-fluorescent tagged proteins, the open

reading frames of SUMO1 and SUMO enzymes were extracted

by SalI/NotI digestion and cloned into the SalI/NotI sites of the

pET28(b) vector. The mutant forms of SUMO1, Uba2 and Ubc9

were created by PCR site-directed mutagenesis and cloned into

the pET28(b) vector, following the protocol described above.

Protein expression and purification

BL21 (DE3) Escherichia coli (E. coli) cells were transformed

with pET28 vectors encoding SUMO1, Aos1, Uba2, Ubc9,

RanGAP1C, their mutants, and fluorescent protein-labeled

versions. The transformed bacteria were inoculated in 2 � YT

medium and the expression of polyhistidine-tagged recombinant

proteins was induced with 0.1 mM IPTG at 25 1C overnight.

The recombinant proteins were purified by Ni2+-NTA agarose

beads (QIAGEN) followed by gel filtration HPLC using the

Superdex75 10/300 GL column on a HPLC purification system

(GE Healthcare, AKTAt purifier). Protein purity was later

confirmed by SDS-PAGE followed by Coomassie blue staining

and their concentrations were determined by the Coomassie Plus

Protein Assay (Pierce).

FRET measurements

After the proteins labeled by CyPet or YPet are mixed in the

presence or absence of other protein members or cofactors,

their fluorescence was determined by the fluorescent plate

reader Flexstation II384 (Molecular Devices) in a 384-well

plate over a period of time. The emission intensities at three

wavelengths were collected: 475 nm and 530 nm at an excitation

wavelength of 414 nm, and 530 nm at an excitation wavelength

of 475 nm. After the emission intensities were corrected by

subtraction of background fluorescence from the plate, the

FRET indexes (see below) at different time points were calculated

and compared.

Protein interaction and SUMOylation assay

All reagents were purchased from Research Products Interna-

tional Corp. unless specified. For the interaction of SUMO1

with E1, 20 pmole CyPet–SUMO1 and 20 pmole YPet–Aos1

or YPet–Uba2 were mixed in the presence or absence of

20 pmole Uba2 or Aos1, respectively, in a buffered solution

containing 50 mM Tris–HCl (pH 7.4), 1 mM DTT and 4 mM

MgCl2 with a total volume of 36 mL. Samples were incubated

at 37 1C and fluorescence emission was monitored after 4 mL10 mM ATP or H2O was added at time zero. For the mutant

Uba2 interaction assay, 20 pmole CyPet–SUMO1, 20 pmole

wild type or mutant YPet–Uba2 and 10 pmole Aos1 were

mixed in the buffer mentioned above in a total volume of 36 mL.For the interaction of SUMO1 and Uba2 in the presence of

Ubc9, 10 pmole CyPet–SUMO1, 10 pmole YPet–Uba2 and

10 pmole Aos1 were mixed in the presence or absence of

100 pmole wild type or mutant Ubc9 in the buffer mentioned

above. Samples were measured by the FRET index after 4 mL of

10 mM ATP was added.

For the interaction disruption assay of SUMO1 and Ubc9,

0.4 mg CyPet–SUMO1 and 0.5 mg YPet–Ubc9 were mixed in

20 mL PBS. 0 to 2 mg nonfluorescence-labeled SUMO1 was

added to the mixture and the FRET indexes were determined.

In the interaction assay of SUMO1 and Ubc9, 20 pmole

CyPet–SUMO1, 20 pmole wild type or mutant YPet–Ubc9,

10 pmole Aos1 and 20 pmole Uba2 were mixed in a buffered

solution containing 50 mM Tris–HCl (pH 7.4), 1 mM DTT

and 4 mM MgCl2 with a total volume of 36 mL. For the

mutant Uba2 assay, 10 pmole wild type or mutant Uba2 was

added. Samples were incubated at 37 1C and fluorescence

emission was monitored after 4 mL 10 mM ATP or H2O was

added at time zero.

Five mg of purified wild type of Ubc9, Ubc9 (C35S) and

Ubc9 (R21E) proteins were loaded onto 10% SDS-PAGE,

separated by electrophoresis and stained by Coomassie blue

staining.

For the interaction assays of wild type or mutant SUMO1

and mutant or wild type Ubc9, respectively, 20 pmole wild

type or mutant CyPet–SUMO1 and 20 pmole wild type or mutant

YPet–Ubc9 were mixed in the presence or absence of 10 pmole

Aos1 and 10 pmole Uba2 in the buffer mentioned above.

For the SUMO1 conjugation assay of mutant SUMO1

peptide, 0.4 mg wild type or mutant CyPet–SUMO1, 0.5 mgYPet–RanGAP1C, 0.05 mg Aos1, 0.1 mg Uba2 and 0.5 mgUbc9 were mixed in the buffer mentioned above, containing

50 mM Tris–HCl (pH 7.4), 1 mMDTT, and 4 mMMgCl2 with

a total volume of 36 mL. Samples were incubated at 37 1C and

fluorescence emission was monitored after 4 mL 10 mM ATP

or H2O was added at time zero. Samples were measured by

FRET index at different time points.

For the SUMO1 conjugation assay of mutant Ubc9 or

Uba2, SUMO1 peptide 0.4 mg CyPet–SUMO1, 0.5 mgYPet–RanGAP1C, 0.05 mg Aos1 were mixed in the presence

or absence of 0.1 mg wild type or mutant Uba2 and 0.5 mg wildtype or mutant Ubc9 in the buffer mentioned above.

The wild type or mutant form of 0.5 mg SUMO1, 0.05 mgFLAG-tagged RanGAP1C, 0.05 mg Aos1, 0.1 mg Uba2 and

0.5 mg Ubc9 were mixed in a buffered solution containing

50 mM Tris–HCl (pH 7.4), 1 mM DTT, 4 mM MgCl2 and

1 mM ATP with a total volume of 40 mL. Samples were

incubated at 37 1C for 1 hour. After they were separated by

10% SDS-PAGE and transferred to a PVDF membrane, the

RanGAP1C proteins were recognized by incubation with

mouse anti-FLAG antibody (Sigma Aldrich) followed by

HRP-labeled goat anti-mouse IgG antibody (Sigma Aldrich).

The chemiluminescence was developed using the Supersignals

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online

This journal is c The Royal Society of Chemistry 2012 Mol. BioSyst., 2012, 8, 1723–1729 1729

West Dura Extended Duration Substrate (Thermo Scientific) and

captured by the Biospectrums imaging system (UVP, LLC).

Competing interests statement

The authors declare that they have no competing financial interests.

Acknowledgements

We are very grateful to Prof. Victor Rodgers for the valuable

advices on reaction rates. We would like to thank Prof. Jerome

Schultz and Dr Elizabeth Gillard for careful comments on our

manuscript. We thank Prof. Xuemei Chen for allowing us to

use the HPLC system. We thank the members in Liao’s group

for very close collaborative work and help with the work. This

work was supported by the National Institutes of Health

Grant (AI076504) to J.L.

References

1 A. Hershko and A. Ciechanover, The ubiquitin system, Annu. Rev.Biochem., 1998, 67, 425–479.

2 O. Kerscher, R. Felberbaum andM. Hochstrasser, Modification ofproteins by ubiquitin and ubiquitin-like proteins, Annu. Rev. CellDev. Biol., 2006, 22, 159–180.

3 F. Melchior, SUMO—nonclassical ubiquitin, Annu. Rev. Cell Dev.Biol., 2000, 16, 591–626.

4 K. D. Sarge and O. K. Park-Sarge, Sumoylation and humandisease pathogenesis, Trends Biochem. Sci., 2009, 34(4), 200–205.

5 K. I. Kim and S. H. Baek, SUMOylation code in cancer developmentand metastasis, Mol. Cells, 2006, 22(3), 247–253.

6 A. D. Sharrocks, PIAS proteins and transcriptional regulation—more than just SUMO E3 ligases? Genes Dev., 2006, 20(7),754–758.

7 S. Bergink and S. Jentsch, Principles of ubiquitin and SUMOmodifications in DNA repair, Nature, 2009, 458(7237), 461–467.

8 E. S. Johnson and G. Blobel, Ubc9p is the conjugating enzyme forthe ubiquitin-like protein Smt3p, J. Biol. Chem., 1997, 272(43),26799–26802.

9 E. S. Johnson, I. Schwienhorst, R. J. Dohmen and G. Blobel, Theubiquitin-like protein Smt3p is activated for conjugation to otherproteins by an Aos1p/Uba2p heterodimer, EMBO J., 1997, 16(18),5509–5519.

10 H. Walden, M. S. Podgorski and B. A. Schulman, Insights into theubiquitin transfer cascade from the structure of the activatingenzyme for NEDD8, Nature, 2003, 422(6929), 330–334.

11 Z. M. Eletr, D. T. Huang, D. M. Duda, B. A. Schulman andB. Kuhlman, E2 conjugating enzymes must disengage from theirE1 enzymes before E3-dependent ubiquitin and ubiquitin-liketransfer, Nat. Struct. Mol. Biol., 2005, 12(10), 933–934.

12 A. L. Haas, P. M. Bright and V. E. Jackson, Functional diversityamong putative E2 isozymes in the mechanism of ubiquitin-histoneligation, J. Biol. Chem., 1988, 263(26), 13268–13275.

13 P. Knipscheer, W. J. van Dijk, J. V. Olsen, M. Mann andT. K. Sixma, Noncovalent interaction between Ubc9 and SUMOpromotes SUMO chain formation, EMBO J., 2007, 26(11),2797–2807.

14 H. Walden, et al., The structure of the APPBP1-UBA3–NEDD8-ATP complex reveals the basis for selective ubiquitin-like proteinactivation by an E1, Mol. Cell, 2003, 12(6), 1427–1437.

15 L. M. Lois and C. D. Lima, Structures of the SUMO E1 providemechanistic insights into SUMO activation and E2 recruitment toE1, EMBO J., 2005, 24(3), 439–451.

16 I. Lee and H. Schindelin, Structural insights into E1-catalyzedubiquitin activation and transfer to conjugating enzymes, Cell(Cambridge, Mass.), 2008, 134(2), 268–278.

17 D. T. Huang, et al., Basis for a ubiquitin-like protein thioesterswitch toggling E1–E2 affinity, Nature, 2007, 445(7126), 394–398.

18 S. K. Olsen, A. D. Capili, X. Lu, D. S. Tan and C. D. Lima, Activesite remodelling accompanies thioester bond formation in theSUMO E1, Nature, 2010, 463(7283), 906–912.

19 A. W. Nguyen and P. S. Daugherty, Evolutionary optimization offluorescent proteins for intracellular FRET, Nat. Biotechnol., 2005,23(3), 355–360.

20 B. N. Giepmans, S. R. Adams, M. H. Ellisman and R. Y. Tsien, Thefluorescent toolbox for assessing protein location and function,Science, 2006, 312(5771), 217–224.

21 J. Zhang, R. E. Campbell, A. Y. Ting and R. Y. Tsien, Creatingnew fluorescent probes for cell biology, Nat. Rev. Mol. Cell Biol.,2002, 3(12), 906–918.

22 Y. Song, V. Madahar and J. Liao, Development of the FRETAssay into Quantitative and High-throughput Screening TechnologyPlatforms for Protein–Protein Interactions, Ann. Biomed. Eng., 2010,39(4), 1224–1234.

23 M. W. Lake, M. M. Wuebbens, K. V. Rajagopalan andH. Schindelin, Mechanism of ubiquitin activation revealed by thestructure of a bacterial MoeB–MoaD complex, Nature, 2001,414(6861), 325–329.

24 J. Wang, et al., The intrinsic affinity between E2 and the Cysdomain of E1 in ubiquitin-like modifications, Mol. Cells, 2007,27(2), 228–237.

25 Z. Tang, C. M. Hecker, A. Scheschonka and H. Betz, Proteininteractions in the sumoylation cascade, lessons from X-ray structures,FEBS J., 2008, 275(12), 3003–3015.

26 D. C. Schwartz and M. Hochstrasser, A superfamily of proteintags, ubiquitin, SUMO and related modifiers, Trends Biochem.Sci., 2003, 28(6), 321–328.

Publ

ishe

d on

30

Mar

ch 2

012.

Dow

nloa

ded

on 1

6/08

/201

3 20

:14:

14.

View Article Online