Selectivity of E2-E3 Interactions in the Human Ubiquitin System

Home , UBE2A, UBE2B, UBE2D1, UBE2D2, UBE2D3, UBE2E1

Selectivity of E2-E3 interactions in the human ubiquitin system

Sjoerd Jan Leendert van Wijk

Cover illustration: Birds articulation by Adrian Hâncu, Moldavia (www.hancu.com) Invitation illustration: Radio birds by Adrian Hâncu, Moldavia (www.hancu.com)

Uitgeverij Boxpress B.V., Oisterwijk Printed by Proefschriftmaken.nl || printyourthesis.com ISBN 9789088911446

Selectivity of E2-E3 interactions in the human ubiquitin system

Selectiviteit van E2-E3 interacties in het humane ubiquitine systeem

(met een samenvatting in het Nederlands)

Proefschrift

ter verkrijging van de graad van doctor aan de Universiteit Utrecht op gezag van de rector magnificus, prof.dr. J.C. Stoof, ingevolge het besluit van het college voor promoties in het openbaar te verdedigen op donderdag 28 ja- nuari 2010 des ochtends te 10.30 uur

door

Sjoerd Jan Leendert van Wijk

geboren op 16 mei 1979 te Rotterdam

Promotor: Prof. dr. H. Th. M. Timmers

The research described in this thesis was financially supported by the Netherlands Pro- teomics Centre (NPC)

Printing of this thesis was financially supported by a generous gift of the J.E. Jurriaanse Stichting, Rotterdam, The Netherlands

There is grandeur in this view of life…

Charles Darwin

Table of contents

Chapter 1 9

Chapter 2 29

Chapter 3 65

Chapter 4 93

Chapter 5 113

Summery 125

Samenvatting 127

Curriculum Vitea 131

Publications 133

Dankwoord 135

Chapter 1

General Introduction

9 Flying insects, like butterflies, undergo remarkable changes before they are eventually equipped with wings and are able to fly. Via a four-staged lifecycle, starting as an egg, then a larva (or caterpillar), followed by the pupa stage, a butterfly becomes a full-grown adult and went through complete metamorphosis. During this morphological transition process, the genetic information of the butterfly, as coded by its DNA, remains the same, but dynamic transcriptional and translational processes define the type of proteins that are derived from that DNA. The set of proteins present at the egg-stage is different from the set that can be found in an adult butterfly and this can be clearly seen from how the organism appears in each of its developmental stages.

The butterfly clearly illustrates the dynamic behavior of proteins present at a certain point in time in an organism. Proteins are continuously formed and broken down and, in general, present in an equilibrium state in a healthy condition. The turnover of individual proteins heavily varies; some structural proteins have half-lives of several weeks, whereas certain transcription factors are degraded rapidly within minutes (1). In order to cope with this regulation, eukaryotic organisms have adopted two systems for protein degradation at the cellular level: autophagy and ubiquitination (2, 3). Autophagy involves protein uptake through the lysosomal machinery and occurs in many different forms. It is a highly regulated process, playing important roles in cellular homeostasis. As the predominant pathway to process macromolecular structures, autophagy is responsible for clearing entities like organelles and microorganisms (see, for example (4).

In contrast, the bulk of intracellular proteins is regulated by selective protein degradation or ubiquitination (5). Ubiquitination refers to the covalent attachment of one or multiple ubiquitin molecules to substrates and was initially identified as a post-translational mark that directs substrates for degradation by the 26S proteasome (3, 6). Although this housekeeping function is valid for some proteins, nowadays, ubiquitin is known to be involved in a much broader spectrum of processes, ranging from transcriptional regulation to apoptosis and cell cycle progression (5).

Ubiquitination Ubiquitin (Ub) is a small polypeptide of 76 amino acids that consists of a relatively globular core and a protruding C-terminal tail (7) (Figure 1.1A). Within the core region, seven lysine residues are present (K6, -11, -27, -29, -33, -48, -63), whereas the C-terminus contains two conserved glycine residues (Figure 1.1A). Polyubiquitin chains of various lengths and of different ubiquitin topologies have been found, constructed by covalent linkage of the glycine to one of the lysine residues. It is commonly accepted that the consequences of substrate modifications depend on the number of ubiquitin molecules added and how they are interconnected (Figure 1.1B).

10 A G76 G76 G76 G76

K27 K27 K33    K27 K11 90o 90o 90o K11

K6 K33 K48 K48 K6

K6 K48 K29 K63 K63 K63 K63

K6 K11 K27 K29 K33 K48 K63

N C Ubiquitin

Figure 1.1. The many faces of ubiquitin (A) Molecular structure of ubiquitin (PDB: 2bgf), indicated in magenta are the seven lysine residues. Also depicted are the C-terminal glycine residues (B) Seven lysine residues of ubiquitin and the functional consequences for substrates modified with corresponding lysine-linked chains.

The significance of the different lysine residues became clear with the observation that substrates with a chain of at least four K48-linked ubiquitin molecules are directed for degradation (8). The 26S proteasome contains several subunits that employ ubiquitin-binding motifs as receptors for ubiquitinated substrates and K48-linked tetra-ubiquitin stretches act as minimal binding signals for these receptors (8). On the other hand, K63-chains direct substrates for non-degradation purposes, like signaling and endocytosis (8-10). However, in some cases K48- and K63-chains fulfill opposing roles (reviewed in (11). For example, the Met4 transcription factor in the yeast Saccharomyces cerevisiae is involved in regulation of the methionine biosynthesis pathway. In line with loss of its transcriptional activity, Met4 becomes modified with K48-polyubiquitin chains (12). The ubiquitin-binding domain of Met4 binds the deposited K48-chain and thereby preventing itself from proteasomal clearing by restricting the length of the ubiquitin chain under the degradation-prone threshold (13). In addition, there are indications that K63-chains may be involved in targeting substrates for 26S proteasomal degradation (14) and for autophagy clearance (15).

Chains linked via a single, specific lysine residue are called homotypic, whereas in some cases multiple lysine linkages have been found in one chain (mixed-linkage chains) (5). The C- terminal glycine is used to anchor the ubiquitin chain to the substrate acceptor lysine (also,

11 alternative acceptor amino acids have been reported (16-18). Recently, linear polyubiquitination has been reported wherein chains are formed between the glycine and the N- terminus of ubiquitin (19, 20). In contrast to large chain-like modifications, substrates can be modified with a single ubiquitin molecule (mono-ubiquitination) or with multiple single ubiquitins (multi-ubiquitination) (5). Substrates decorated with either mono- or multi-ubiquitins are, in general, not degraded by the proteasome, but are involved in endocytosis, membrane trafficking, DNA repair and histone modification in transcriptional regulation (Figure 1.1B) (21, 22). Studies in yeast using epitope-tagged ubiquitin followed by mass spectrometry identified a relative lysine-usage abundance of K48 > K11 and K63 >> K6, K27, K29 and K33 (23).

The ubiquitin protein is completely conserved throughout the eukaryotic kingdom. It appears to be absent from bacteria and archae (although, recently a prokaryotic ubiquitin-like conjugation system has been described, called PUPylation (24)). Besides ubiquitin, other small proteins that resemble ubiquitin (ubiquitin-like proteins; UBLs) have been identified, like SUMO, NEDD8 and ISG15 (25). Although these UBLs display variable levels of sequence homology with ubiquitin, all UBLs have certain features in common. They all share a, more or less, conserved ubiquitin-fold. Most often, they become conjugated to lysine residues by their C-terminal glycine residues. Finally, the mechanisms of their activation and conjugation show remarkable similarities (25).

To ensure that ubiquitination takes place in a controlled manner in time and space, cells have evolved a tightly controlled machinery, which involves ATP-dependent activation and subsequent conjugation and the timed action of multiple types of enzymes (3). Ubiquitin itself is synthesized as an inactive precursor, either as tandem expressed ubiquitin polymers or as fusions with ribosomal proteins (26-28). Prior to activation, ubiquitin polymers need to be processed in order to reveal the C-terminal diglycines and to liberate individual ubiquitin molecules. Responsible for this is a family of enzymes called deubiquitinating enzymes (DUBs) (29).

Downstream activation and conjugation reactions are mediated by a cascade of enzymes that involve ubiquitin-activating enzymes (E1), ubiquitin-conjugating enzymes (E2s) and ubiquitin protein ligases (E3s) (30) (Figure 1.2). In the final step, E3 enzymes are selecting and recruiting appropriate targets (3). Depending on the type of polyubiquitin chain present, many ubiquitinated species end up as being processed by the 26S proteasome (8). The 26S proteasome is a large, multi-subunit barrel-shaped complex (20S-complex), topped-off with 19S activation complexes. Several 19S particle subunits specifically recognize and bind ubiquitinated substrates, followed by their unfolding (31). Specific DUBs cleave off ubiquitin chains and trim them down to individual molecules that can be used again for conjugation. Finally, the unfolded substrate enters the barrel-shaped 20S complex and is degraded in the catalytic centre of the proteasome (32).

12 E1 2

E2 35

E3 > 500

Figure 1.2. Hierarchical organization of the human ubiquitin-proteasome system In humans, two major ubiquitin E1s are activating ubiquitin (depicted is Uba1, PDB: 3cmm) that transfer the activated Ub/UBL to the family of E2 enzymes. In human, 35 active E2s have been identified (depicted is UbcH5B(UBE2D2), PDB: 2clw) that transfer the activated Ub/UBL to a group of > 500 E3 ligases (depicted is the CNOT4 RING-finger, PDB: 1ur6) via highly selective interactions between E2-E3 enzymes.

Ubiquitin-activating enzymes (E1) Prior to conjugation and when the C-terminus is enzymatically exposed, ubiquitin needs to be activated by the ubiquitin-activating enzyme (E1). The E1 binds ubiquitin and Mg 2+ and uses ATP to adenylate the C-terminus of ubiquitin at Gly76. The formed high-energy anhy- dride bond rapidly reacts with the sulfhydryl group of the E1-active site cysteine, forming a high-energy thioester bond (33, 34). A fully loaded E1 carries two ubiquitin molecules, one as adenylate and the other thioester bound to the active site. This makes the E1 a highly efficient enzyme; it generates sufficient amounts of activated ubiquitin molecules to be used by downstream enzymes, even while its concentration is often lower than the E2 concentration (3). Next, the thioester bound ubiquitin is transferred onto downstream enzymes (see be- low). Within the human genome, Ube1 is identified as the predominant E1 for ubiquitin and studies using a temperature-sensitive Ube1 mutant (ts85) revealed that it is an essential enzyme (30). An alternative, putative ubiquitin E1 has been identified, nUbe1L (novel Ube1- like), predominantly expressed in testis, where it presumably has roles in spermatogenesis and male fertility (35). However, no information concerning the role of nUbe1L in ubiquitin activation has been described yet. Recently, Uba6 has been characterized as a bona fide ubiquitin E1 in vertebrates and sea urchin. Uba6 specifically activated ubiquitin in vitro and in vivo. Instead of showing a broad E2 specificity, like the classical E1, Uba6 interacts with a more restricted selection of E2 enzymes, including the Uba6-specific E2 Use1(Uba6-specific E2 enzyme; UBE2Z) (36).

13 Ubiquitin-conjugating enzymes (E2) Upon activation, ubiquitin is transferred to a group of enzymes called ubiquitin-conjugating enzymes (E2s) (3). Members of the E2 family represent a group of highly conserved proteins, which accept the activated ubiquitin onto a catalytic cysteine residue within a highly conserved catalytic domain (UBC-fold) (37, 38). Besides this UBC-domain, less conserved N- and/or C-terminal extensions are present on certain E2 enzymes that mediate E2-specific processes. During evolution, expansion of the E2 family has occurred so lower organisms, like yeast, have smaller numbers of E2 enzymes (6). Besides this variation it appears that a minor subset of enzymes is conserved, and that other E2 enzymes are present in multiple isoforms or uniquely present only in higher organisms (39). The main function of E2 enzymes is to interact with ubiquitin-loaded E1s to accept ubiquitin, followed by binding to E3 ubiquitin protein ligases so that ubiquitin can be transferred to substrates (6). Structural clues concerning E1 and E3 binding indicate that both E1 and the E3 share partially the same binding interface on the E2 enzyme (33, 40).

E3 ubiquitin protein ligases (E3) The final step of ubiquitination is the transfer of ubiquitin from the E2 to substrates, mediated by ubiquitin protein ligases (E3) (3, 6). E3s fulfill two main functions; they recognize and bind the substrate and they recruit and interact with ubiquitin-loaded E2 enzymes. Be- sides LIM, FYVE and PHD-domain E3s, the majority of E3s belong either to the HECT (Homo- logous to E6-AP Carboxyl terminus)-family (41), the U-box family (42) or the RING (Really In- teresting New Gene)-family (43). The HECT-type E3 ligases contain a HECT-domain of approximately 350 residues (41). The domain was initially identified in the E6-AP protein that is able to interact with the human papilloma virus E6 protein in a complex that, upon viral infection, is able to induce p53 degradation (44). HECT-domain E3s are catalytically involved in ubiquitin conjugation, they are equipped with a catalytic cysteine that binds ubiquitin via a thioester (41). In contrast to HECTs, U-box and RING-type E3s facilitate ubiquitin transfer by acting as molecular scaffolds. They position both the E2 and the substrate in a catalytic favorable orientation so that the transfer of ubiquitin can take place (43, 45, 46). Both U-boxes and RINGs contain a domain that is characterized by eight cysteine or histidine residues, separated by a typical spacing, which folds into a specific cross-brace structure. Within RING-E3s the cross-brace is stabilized by interactions of the cysteine/histidine residues with two atoms of zinc, whereas in U-boxes the structure is formed via electrostatic interactions (43, 46, 47).

RING-type E3 ligases The human genome encodes over 600 RING or RING-like structures (48). At present it is not clear what exactly defines a RING-finger protein to be active as E3 ligase, but it is expected that at least a significant fraction of annotated RING-finger proteins will possess E3-like activities. RING-finger E3 ligases can be structurally classified as RING-H2 (with histidines at positions 4 and 5), RING-HC (with only one histidine at position 4) or RING-C domains (containing only zinc-chelating cysteines) (43). Several lines of evidence indicate that RING-finger

14 domains confer E2s with E2 interaction binding specificity (45, 49). In agreement with their function as molecular scaffolds RING-finger E3s accommodate their bound substrates at regions distinct than those that are occupied by E2 enzymes (43). Based on molecular, structural and mechanistic features RING-type E3 ligases can be classified into two main groups: multi-subunit or single RING E3 ligases.

The Skp1-Cul1-F-box (SCF) complex contains at least four subunits: Rbx1, Cul1, Skp1 and an F-box protein (50, 51). F-box proteins act like adaptor molecules and are involved in substrate recruitment and binding (52). They are bound to Skp1 via the F-box while Skp1 interacts with the N-terminus of Cul1. Within SCF-complexes, Rbx1 provides a RING E3 activity, which interacts with Cul1 to recruit and bind E2 enzymes (53). The Cul2-Elongin B-Elongin C (CBC)-complex is composed of Rbx1, Cul2, Elongin B, Elongin C and a SOCS-box protein and shows structural similarities with the SCF complex (54). The SOCS-box protein is the variable subunit of the complex and responsible for substrate selection. Among the SOCS-box proteins identified is the VHL (von Hippel-Lindau) protein, often mutated in von Hippel-Lindau disease (55, 56). VHL mediates degradation of the hypoxia-regulated transcription factor HIF1α that is involved in transcriptional regulation of hypoxia-inducible genes. Loss of VHL function leads to an accumulation of HIF1α, alterations in the transcription of hypoxia- responsive genes and tumor growth promotion (55). The anaphase-promoting complex/cyclosome (APC/C) is involved in the degradation of cyclin B via the APC2 and APC11 subunits, which are similar to Cul1 and Rbx1, respectively, and facilitates exit from M phase in the cell cycle (57-60). Besides these E3 complexes many other single-subunit E3 ligases have been identified, like c-Cbl, parkin and MDM2 (43). Irrespective of the complexities of the complexes build around RING-finger proteins, the RING-finger domain is responsible for recruitment and binding of the E2 enzyme in the vast majority of E3s. A central and very basic question in ubiquitin research is how the system controls substrate selectivity and timing of ubiquitin modification? And when the system has made this deci- sion, how does it decide what type of modification it has to create? In-depth studies concerning recognition of substrates by E3 ligases are sparse and hampered by the observation that for many substrates no E3 ligases have been identified yet. Given the high number of E3s in the human genome, only a small fraction is functionally annotated with respect to their targets. Matching a substrate with its cognate E3 is in many cases difficult since these interactions are often weak or transient. In order to address this issue, Elledge and coworkers monitored in vivo protein turnover rates of approximately 8000 proteins using a high- throughput combination of flow cytometry and microarray analysis (1). Employing this system in combination with a dominant-negative Cul1 subunit, Yen et al (61) identified many new substrates for the SCF-complex, providing evidence for the suitability of this system to screen for E3 targets.

Methods based on E3 sequence to predict substrates are held back by the observation, that domains involved in substrate recognition are mainly unknown and are predicted not to be

15 conserved among different RINGs. This hinders identification of these interfaces. This is complicated by the observation that substrate recognition may require additional post- translational modification, like phosphorylation. In some cases, however, substrates are re- cruited to E3s by small patches of amino acids arranged in a specific consensus, the so-called degrons. For example, the E3 ligase APCCdc20 recognizes substrates that have a D-box degron (R-X-X-L) (62), whereas APCCdh1 recognizes both the D-box and a KEN-motif (63). In other cases, SCFF-box E3s recruit their substrates only when phosphorylated at either a serine or a threonine (phosphodegrons). For example, SCF-TRCP recognizes the D-pS-G-X-X-pS motif while SCFFbw7 binds L-X-pT-P-P-X-pS (64).

More is known about what defines the type of inter-ubiquitin chain linkage to be assembled. In some cases E3s have been implicated as regulators of chain assembly. For example, the HECT E3 ligase E6AP has the ability to form K48 chains on its active site (65) and KIAA10 has been shown to catalyze both unanchored K48- and K29-linked chains (66). Also the yeast Rsp5 HECT E3 is involved in substrate ubiquitination via K48- or K63-linkage chains (67). Ad- ditionally, the BRCA1 tumor suppressor, when bound to its binding partner BARD1, harbors the ability to catalyze K6-connected ubiquitin chains (68), especially during S-phase and in response to replication stress and DNA damage (69). The RING-finger proteins Ring1B and Bmi1, which are part of the Polycomb Repressive Complex 1 (PRC1), are involved in monoubiquitination of histone H2A (70). Recently, it was reported that the active E3 subunit Ring1B mediates autoubiquitination with mixed K6-, K27-, and K48-linked polyubiquitin chains and that this auto-modification was a prerequisite for Ring1B to ubiquitinate H2A (71). In other cases, however E2s are directly implicated in lysine selection for chain construction. By selectively interacting with E3 enzymes, E2s employ their specific lysine selection characteristics to provide the system with flexibility in ubiquitin chain construction. A textbook example comes from yeast Ubc13(UBE2N) that has a unique feature to dimerize with one of the two inactive E2 variants, UEV-1A(UBE2V1) and MMS2(UBE2V2), to create E2 complexes capable of catalyzing K63-linked chains (10). To functionally accommodate the inactive E2 variants, Ubc13(UBE2N) has evolved a specialized β-sheet surface that specifically binds both variant E2s (72). The typical orientation that is than obtained between Ubc13(UBE2N) and the inactive E2s allows the specific access to K63 within ubiquitin (73). Back in 1999 it was already shown that UbcH5A(UBE2D1) has the ability to construct unanchored K48- and K29- linked polyubiquitin chains (74). In addition, UbcH5B(UBE2D2) has been shown bind ubiquitin non-covalently onto its β-sheet surface, yielding large UbcH5B(UBE2D2) assemblies that determine BRCA1/BARD1-mediated autoubiquitination (75). In other studies UbcH5A(UBE2D1) is capable of catalyzing anchored polyubiquitin chain to the CHIP and TRAF6 E3 ligases lacking any lysine specificity (76). When studying autoubiquitination of the CHIP U-box and the RING E3 ligases MuRF1 and MDM2, the UbcH5A-D(UBE2D1-4) family has the ability to synthesize all types of chains, but predominantly those linked via K48, K63 and K11. Surprisingly, these chains appeared to be bifurcated and composed of branched linkages via K6 + K11, K27 + K29, or K29 + K33 on the preceding Ub molecule. Interestingly, when

16 assaying UbcH5A(UBE2D1) in combination with the HECT E3s E6AP or NEDD4, homotypic chains are observed, exclusively linked via K48 (in the case of E6-AP) or even K63 chains in the case of NEDD8. Testing CHIP and MuRF1 with Ubc13(UBE2N)/UEV-1A(UBE2V1) produces solely K63-linkages, but when HIP2(UBE2K) was used, homotypic K48-chains were assembled on its substrate (77). The same HIP2(UBE2K) has also the ability to construct tricyclic unanchored polyubiquitin chains (78, 79). Finally, in a screen for interacting E2s the dimeric BRCA1/BARD1 E3 interacted with six previously unidentified E2s: UbcH6(UBE2E1), UbcH8(UBE2E2), UbcH9(UBE2E3), Ubc13(UBE2N), HIP2(UBE2K) and UBE2W. All these E2s were able to interact directly with the BRCA1 RING and were active in supporting autubiqui- tination. Among the interacting E2s, UbcH6(UBE2E1), UbcH8(UBE2E2), UbcH9(UBE2E3) and UBE2W were able to monoubiquitinate BRCA1, whereas Ubc13(UBE2N)-MMS2(UBE2V2) and HIP2(UBE2K) construct K63- and K48-linked chains, respectively, but require an acceptor ubiquitin at BRCA1. Furthermore, it was found that differences between mono- and polyubiquitination reside within the ability of the E2 to bind ubiquitin non-covalently, at a region distinctly from its active site (80).

Taken these things together, the capacity of the ubiquitin system to construct specific chains is not a simple consequence of combining specific E2 and E3 enzymes. It merely reflects a complex interplay between molecular characteristics of E2s, E3s and substrates and positioning of non-covalently bound ubiquitin molecules.

Two-hybrid systems to study binary protein-protein interactions In analogy with the dynamic interplay between instruments and vocals in symphonic orches- tras, proteins rarely act on their own. Interactions between proteins in networks are crucial for most biological processes, as exemplified earlier for the ubiquitin system. Varying from binary interactions to multi-subunit protein complexes, interactions are occurring by means of specific organizations of surface-exposed residues, arranged in certain geometrical properties on both protein surfaces. To study how complex, cellular systems are operating it is essential to first characterize these interactions. Once knowing the relationships between proteins, these interconnections can be assembled to construct large networks representing biological phenomena. In other words, musicians and vocalists need to listen and respond to each other in order to play a musical composition, like Monteverdi’s masterpiece Vespro Della Beata Vergine 1610.

A productive method to study binary protein-protein interaction in a high-throughput manner is yeast two-hybrid (Y2H) (81). Two-hybrid techniques rely on the observation that many transcription factors are composed of two different parts, a DNA binding domain (DBD) and a transcriptional activation domain (AD) (82). A protein of interest (bait) is fused to a DBD, whereas proteins to be tested for interaction are fused to the AD (preys). When two proteins interact, the DBD and the AD will come in close contact and reconstitute an active transcription factor. This will drive the expression of one or several reporter genes that can be eva-

17 luated on growth selection or on colorimetric basis (82). Since its initial discovery, the initial Y2H format has evolved in a myriad of branches. Variations in DBDs or ADs, reporter genes, host systems, types of read-out and platforms have been introduced, improving performance and applicability (for an overview, see (83). An advantage of yeast as host for two- hybrid systems is its sexual differentiation, which depends on mating between - and A-type haploid cells (84). Mediated by mating pheromones, growth of a projective extension (called a schmoo) is triggered after which the contents of both haploid cells will fuse to form a single diploid cell (84). This phenomenon opens an elegant possibility to reduce the amount of transformations and to adopt two-hybrid systems to arrays-based platforms (85-88). Array formats, in which a known cohort of prey cDNAs is constructed and the bait is screened over these arrays, has several advantages compared to classical library screening (88). First, the nature of the arrayed proteins is highly controlled, especially using the present ORFeome collections (89). Using arrays there are no risks that inappropriate fusions will be picked-up that are for example in incorrect reading frames or corresponds to non-coding DNAs. Ar- rayed proteins are normalized with respect to their representation, circumventing library- specific expression patterns and underrepresented preys like membrane proteins. Finally, pair-wise, array-based screens are believed to be more sensitive than library screens, probably because low-intensity signals can be more easily distinguished from background signals. Two-hybrid systems are fast and straightforward ways to explore the interactions and functions of individual proteins within large networks. They allow the detection of weak and transient interactions and can be used in combination with mutagenesis to modulate the interaction potential against a whole panel of prey-interacting partners at once.

Like every method, potential pitfalls of two-hybrid-based methods reside within its principle. Interactions are often assessed between over-expressed proteins, something that could increase the chance of detecting aspecific interactions. Many two-hybrid-systems force their baits and preys to interact within the nucleus, even if it is plausible that these proteins never reside within the nuclear compartment or physiologically never encounter each other at all. Bait and prey proteins do not always undergo proper post-translational modifications or become functionally folded. All these factors can potentially give rise to aspecific or false- positive or –negative interactions. False-positive interactions are defined as interactions that never occur under physiological conditions, whereas false-negatives are those protein interactions that are taking place endogenously but are not detecting using two-hybrid systems. This probability implies that observed interactions should be confirmed by other methods, but given the massive nature of many Y2H screens this remains a daunting task. Several strategies are described that can be used to assess the quality of yeast two-hybrid interaction data by sampling a fraction of interactions and study these in more detail. Alternative biochemical methods, like co-immunoprecipitations, GST pull-down assays or fluorescent complementation studies have been used to validate two-hybrid interactions. Besides those experimental techniques, the comparison of observed interactions with literature-curated interactions (LCIs), derived from high- and lower-throughput studies, has been

18 used extensively to validate protein interactions. At first sight this strategy seems straightforward, but recent analysis revealed that the quality of LCIs is often over-estimated (90). Another hurdle is platform specific, which is a fundamental problem in converging individual Y2H screens into genome-wide networks. At present, the two most frequently used yeast two-hybrid systems are based on either GAL4-AD/GAL4-BD or LexA-BD/B42-AD pairs. Unfortunately, many proteins are only interacting when tested in one of both two-hybrid systems, probably because AD- and BD-tags interfere with the ability of proteins to interact with each other. Since those Y2H interactions that are detected with both configurations are considered to be of higher confidence and often represent biologically relevant interactions, ideally each interaction should be assessed using both platforms (91-93). In practice, this is, for obvious reasons, not possible and the vast majority of protein-protein interactions are detected using only one of both systems. As a consequence, current protein-protein interaction data sets are characterized by a poor degree of overlap, which hampers the assemblage of realistic interaction networks. This basic problem is illustrated in a protein interaction map of cell cycle regulating proteins in Drosophila and its relation with other, genome-wide Dro- sophila data sets (94). In a screening approach, 1,814 interactions were detected among 488 proteins using the LexA-B42 system. Comparing those interactions with another Drosophila data set of 20,000 Gal4-based interactions (95), reveal an overlap of 28 interactions. Striking- ly, more than a quarter of the proteins tested in the LexA network were absent from the Gal4 proteome-wide map. Of the 106 proteins exhibiting interactions in the LexA-system, 60 failed to interact in the Gal4-based Y2H. The other way around is also true; 46 of the 152 LexA bait proteins failed to connect; yet 14 of these showed interactions in the Gal4 network. Thus, the low degree of overlap the two datasets is partly due to the ability of platforms to detect interactions with specific proteins (94).

Another trivial reason for the small overlap became clear after in-depth comparisons between different platforms using varying bait and prey proteins. This has been done for protein networks in Saccharomyces cerevisiae (91) and Caenorhabditis elegans (96) and the high-throughput interaction sets only covered for ~ 13% when compared with a reference set of interactions. Although displaying the same rate of true-positives, the small degree of overlap between detected interaction is a consequence of a lack of saturation among baits and prey proteins (97).

To date, several attempts have been made to characterize genome-wide protein-protein interaction networks in human (98, 99), fruitfly (94, 100), yeast (97, 101, 102), worms, (103), parasites (104, 105), bacteria (106, 107), viruses (108) and plants (109) in two-hybrid approaches. These efforts yielded thousands of interactions that will generate new starting points for future experiments. Despite these high-throughput efforts, in-depth analysis of E2-E3 interactions remains highly fragmented and largely untouched. Large-scale interactions maps do not provide a mechanistic basis for the underlying specificity of the interactions, which could be explained since bait and preys often represent a sample of a complete

19 proteome. Strikingly, within the literature-curated interactions deposited in several web- based databases, interactions between E2 and E3 enzymes remain underrepresented. This implies that large-scale, proteome-wide interactions studies are less suitable to detect E2-E3 interactions. Answering questions concerning E2-E3 interaction specificity would require a more in-depth evaluation of these networks. Scaling down bait and prey proteins to single domains increases the resolution of interaction studies. This approach, combined with mutational analysis, could provide insight in the attribution of individual residues on the selectivity of binary E2-E3 interactions in the ubiquitin-proteasome system.

Outline of this thesis Highly selective interactions between ubiquitin-conjugating enzymes and RING-type E3 ligases are crucial for the adequate and efficient action of ubiquitin and ubiquitin-like pathways. Within these cascades, E2 enzymes provide a connecting link between activation and the final covalent conjugation, thereby maintaining the integrity of the E1-E2-E3 hierarchic pyramid. It became increasingly clear that E2s are directly involved in specifying chain topology and thereby the fate of the substrate. Insights into the molecular basis of E2-E3 interaction specificity can direct future studies towards understanding ubiquitin network biology. Furthermore, structure-based altered-specificity E2-E3 mutant pairs can be used to identify substrates for E3 ligases. Finally, a detailed overview of E2-E3 complexes on the level of individual amino acids could support in screening of small-molecular weight compounds that modulate E2-E3 interactions and effects on substrates. However, despite these importance aspects, in-depth genome-wide insights into this interaction selectivity remain enigmatic and restricted to only a small subset of E2 and E3 enzymes. Therefore, the studies described in this thesis aim at unraveling the selectivity of E2-E3 interactions in the human ubiquitin- proteasome system, both on the level of individual E2 and E3 proteins as well as their molecular determinants of interaction specificity. A global interaction screen intended to identify binary interactions between 35 human E2 UBC-folds and 250 human RING-finger domains is described in Chapter 2. The observed 346 high-confidence interactions are in good agreement with literature-curated E2-E3 interactions and with GST pull-down verifications. As additional validation, E3-interaction interface of the K63-specific Ubc13(UBE2N) was changed by introducing two mutations so that its interface mimics that of the mixed-chain specific E2 UbcH5B(UBE2D2). As a result, synthetic E3 interactions were gained with the mutant Ubc13(UBE2N) that are shared with UbcH5B(UBE2D2) but not with wild-type Ubc13(UBE2N). Chapter 3 provides an in-depth overview of the evolutionary constraints influencing protein organization, enzyme regulation and selectivity towards E1s and E3s of the E2 superfamily in light of ubiquitin and ubiquitin-like conjugation pathways. An example of how E2 sequence conservation relates to differences in E3 selection is shown in Chapter 4, where dramatic differences in E3 interaction profile between two highly homologous E2s are presented. The low number of E3 interactions of UbcH8(UBE2E2), with residue E66 in Helix 1 and D113 in Loop 1, could be rescued by mutation to the UbcH6(UBE2E1) counterpart residues (E66D and D113E), thereby mimicking the E3 interactions of UbcH6(UBE2E1). Using a combination

20 of molecular dynamics simulations, protein threading and additional mutagenesis, models were generated between the UBC-folds of UbcH6(UBE2E1) and UbcH8(UBE2E2) with the RING-finger of TOPORS that explains how D113 is orientated in UbcH8(UBE2E2). Elongating the side chain of D113 by substitution with the glutamic acid of UbcH6(UBE2E1) enables UbcH8(UBE2E2) D113E to interact with the RING-finger of TOPORS, rescuing the shortness of D113 and thereby the interaction with TOPORS. Furthermore, an intramolecular network of salt-bridges of two strictly conserved residues (K117 and D145) that actively control the positioning of the D113 side-chains was studied. This thesis concludes with Chapter 5, in which the described results are discussed and reflected onto the theoretical framework of existing ubiquitin and network systems biology literature.

References 1. Yen, H. C. S., Xu, Q. K., Chou, D. M., Zhao, Z. M., and Elledge, S. J. (2008) Global Pro- tein Stability Profiling in Mammalian Cells. Science 322, 918-923 2. Levine, B., and Klionsky, D. J. (2004) Development by self-digestion: Molecular mechanisms and biological functions of autophagy. Dev Cell 6, 463-477 3. Dye, B. T., and Schulman, B. A. (2007) Structural mechanisms underlying posttransla- tional modification by ubiquitin-like proteins. Annu Rev Biophys Biomol Struct 36, 131-150 4. Levine, B. (2005) Eating oneself and uninvited guests: Autophagy-related pathways in cellular defense. Cell 120, 159-162 5. Ravid, T., and Hochstrasser, M. (2008) Diversity of degradation signals in the ubiquitin-proteasome system. Nat Rev Mol Cell Biol 9, 679-690 6. Pickart, C. M. (2001) Mechanisms underlying ubiquitination. Annu Rev Biochem 70, 503-533 7. Pickart, C. M., and Eddins, M. J. (2004) Ubiquitin: structures, functions, mechanisms. Biochim Biophys Acta 1695, 55-72 8. Thrower, J. S., Hoffman, L., Rechsteiner, M., and Pickart, C. M. (2000) Recognition of the polyubiquitin proteolytic signal. EMBO J 19, 94-102 9. Deng, L., Wang, C., Spencer, E., Yang, L., Braun, A., You, J., Slaughter, C., Pickart, C., and Chen, Z. J. (2000) Activation of the IkappaB kinase complex by TRAF6 requires a dimeric ubiquitin-conjugating enzyme complex and a unique polyubiquitin chain. Cell 103, 351-361 10. Hofmann, R. M., and Pickart, C. M. (1999) Noncanonical MMS2-encoded ubiquitin- conjugating enzyme functions in assembly of novel polyubiquitin chains for DNA repair. Cell 96, 645-653 11. Li, W., and Ye, Y. (2008) Polyubiquitin chains: functions, structures, and mechanisms. Cell Mol Life Sci 65, 2397-2406 12. Kaiser, P., Flick, K., Wittenberg, C., and Reed, S. I. (2000) Regulation of transcription by ubiquitination without proteolysis: Cdc34/SCFMet30-mediated inactivation of the transcription factor Met4. Cell 102, 303-314 13. Flick, K., Raasi, S., Zhang, H. W., Yen, J. L., and Kaiser, P. (2006) A ubiquitin-interacting motif protects polyubiquitinated Met4 from degradation by the 26S proteasome. Nat Cell Biol 8, 509-U144 14. Babu, J. R., Geetha, T., and Wooten, M. W. (2005) Sequestosome 1/p62 shuttles polyubiquitinated tau for proteasomal degradation. J Neurochem 94, 192-203 15. Olzmann, J. A., Li, L., Chudaev, M. V., Chen, J., Perez, F. A., Palmiter, R. D., and Chin, L. S. (2007) Parkin-mediated K63-linked polyubiquitination targets misfolded DJ-1 to aggresomes via binding to HDAC6. J Cell Biol 178, 1025-1038 16. Cadwell, K., and Coscoy, L. (2005) Ubiquitination on nonlysine residues by a viral E3 ubiquitin ligase. Science 309, 127-130

21 17. Ravid, T., and Hochstrasser, M. (2007) Autoregulation of an E2 enzyme by ubiquitin- chain assembly on its catalytic residue. Nat Cell Biol 9, 422-427 18. Li, W., Tu, D., Brunger, A. T., and Ye, Y. (2007) A ubiquitin ligase transfers preformed polyubiquitin chains from a conjugating enzyme to a substrate. Nature 446, 333-337 19. Iwai, K., and Tokunaga, F. (2009) Linear polyubiquitination: a new regulator of NF- kappa B activation. EMBO Rep 10, 706-713 20. Haas, A. L. (2009) Linear polyubiquitylation: the missing link in NF-kappa B signalling. Nat Cell Biol 11, 116-118 21. Ikeda, F., and Dikic, I. (2008) Atypical ubiquitin chains: new molecular signals - 'Protein modifications: Beyond the usual suspects' review series. EMBO Rep 9, 536- 542 22. Newton, K., Matsumoto, M. L., Wertz, I. E., Kirkpatrick, D. S., Lill, J. R., Tan, J., Dugger, D., Gordon, N., Sidhu, S. S., Fellouse, F. A., Komuves, L., French, D. M., Ferrando, R. E., Lam, C., Compaan, D., Yu, C., Bosanac, I., Hymowitz, S. G., Kelley, R. F., and Dixit, V. M. (2008) Ubiquitin chain editing revealed by polyubiquitin linkage-specific antibodies. Cell 134, 668-678 23. Peng, J., Schwartz, D., Elias, J. E., Thoreen, C. C., Cheng, D., Marsischky, G., Roelofs, J., Finley, D., and Gygi, S. P. (2003) A proteomics approach to understanding protein ubiquitination. Nat Biotechnol 21, 921-926 24. Pearce, M. J., Mintseris, J., Ferreyra, J., Gygi, S. P., and Darwin, K. H. (2008) Ubiquitin- like protein involved in the proteasome pathway of Mycobacterium tuberculosis. Science 322, 1104-1107 25. Hochstrasser, M. (2009) Origin and function of ubiquitin-like proteins. Nature 458, 422-429 26. Redman, K. L., and Rechsteiner, M. (1989) Identification of the Long Ubiquitin Exten- sion as Ribosomal Protein-S27a. Nature 338, 438-440 27. Finley, D., Bartel, B., and Varshavsky, A. (1989) The Tails of Ubiquitin Precursors Are Ribosomal-Proteins Whose Fusion to Ubiquitin Facilitates Ribosome Biogenesis. Na- ture 338, 394-401 28. Lund, P. K., B.M., M.-S., Simmons, J. G., Hoyt, E., D'Ercole, A. J., Martin, F., and Van Wyk, J. J. (1985) Nucleotide sequence analysis of a cDNA encoding human ubiquitin reveals that ubiquitin is synthesized as a precursor. J Biol Chem 260, 7609-7613 29. Nijman, S. M. B., Luna-Vargas, M. P. A., Velds, A., Brummelkamp, T. R., Dirac, A. M. G., Sixma, T. K., and Bernards, R. (2005) A genomic and functional inventory of deubiquitinating enzymes. Cell 123, 773-786 30. Scheffner, M., Nuber, U., and Huibregtse, J. M. (1995) Protein ubiquitination involving an E1-E2-E3 enzyme ubiquitin thioester cascade. Nature 373, 81-83 31. Murata, S., Yashiroda, H., and Tanaka, K. (2009) Molecular mechanisms of proteasome assembly. Nat Rev Mol Cell Biol 10, 104-115 32. Schwartz, A. L., and Ciechanover, A. (2009) Targeting Proteins for Destruction by the Ubiquitin System: Implications for Human Pathobiology. Ann Rev Pharmacol Toxicol 49, 73-96 33. Lee, I., and Schindelin, H. (2008) Structural insights into E1-catalyzed ubiquitin activation and transfer to conjugating enzymes. Cell 134, 268-278 34. Schulman, B. A., and Harper, J. W. (2009) Ubiquitin-like protein activation by E1 enzymes: the apex for downstream signalling pathways. Nat Rev Mol Cell Biol 10, 319- 331 35. Pelzer, C., Kassner, I., Matentzoglu, K., Singh, R. K., Wollscheid, H. P., Scheffner, M., Schmidtke, G., and Groettrup, M. (2007) UBE1L2, a novel E1 enzyme specific for ubiquitin. J Biol Chem 282, 23010-23014 36. Jin, J., Li, X., Gygi, S. P., and Harper, J. W. (2007) Dual E1 activation systems for ubiquitin differentially regulate E2 enzyme charging. Nature 447, 1135-1138 37. Burroughs, A. M., Jaffee, M., Iyer, L. M., and Aravind, L. (2008) Anatomy of the E2 ligase fold: Implications for enzymology and evolution of ubiquitin/Ub-like protein conjugation. J Struct Biol

22 38. Winn, P. J., Religa, T. L., Battey, J. N., Banerjee, A., and Wade, R. C. (2004) Determi- nants of functionality in the ubiquitin conjugating enzyme family. Structure 12, 1563- 1574 39. Michelle, C., Vourc'h, P., Mignon, L., and Andres, C. R. (2009) What Was the Set of Ubiquitin and Ubiquitin-Like Conjugating Enzymes in the Eukaryote Common Ances- tor? J Mol Evol 68, 616-628 40. Zheng, N., Wang, P., Jeffrey, P. D., and Pavletich, N. P. (2000) Structure of a c-Cbl- UbcH7 complex: RING domain function in ubiquitin-protein ligases. Cell 102, 533-539 41. Rotin, D., and Kumar, S. (2009) Physiological functions of the HECT family of ubiquitin ligases. Nat Rev Mol Cell Biol 10, 398-409 42. Patterson, C. (2002) A new gun in town: the U box is a ubiquitin ligase domain. Sci STKE 2002, PE4 43. Deshaies, R. J., and Joazeiro, C. A. (2009) RING domain E3 ubiquitin ligases. Annu Rev Biochem 78, 399-434 44. Scheffner, M., Huibregtse, J. M., Vierstra, R. D., and Howley, P. M. (1993) The HPV-16 E6 and E6-AP complex functions as a ubiquitin-protein ligase in the ubiquitination of p53. Cell 75, 495-505 45. Lorick, K. L., Jensen, J. P., Fang, S., Ong, A. M., Hatakeyama, S., and Weissman, A. M. (1999) RING fingers mediate ubiquitin-conjugating enzyme (E2)-dependent ubiquitination. Proc Natl Acad Sci U S A 96, 11364-11369 46. Freemont, P. S. (2000) RING for destruction? Curr Biol 10, R84-87 47. Ardley, H. C., and Robinson, P. A. (2005) E3 ubiquitin ligases. Essays Biochem 41, 15- 30 48. Li, W., Bengtson, M. H., Ulbrich, A., Matsuda, A., Reddy, V. A., Orth, A., Chanda, S. K., Batalov, S., and Joazeiro, C. A. (2008) Genome-wide and functional annotation of human E3 ubiquitin ligases identifies MULAN, a mitochondrial E3 that regulates the organelle's dynamics and signaling. PLoS ONE 3, e1487 49. Joazeiro, C. A., Wing, S. S., Huang, H., Leverson, J. D., Hunter, T., and Liu, Y. C. (1999) The tyrosine kinase negative regulator c-Cbl as a RING-type, E2-dependent ubiquitin- protein ligase. Science 286, 309-312 50. Schulman, B. A., Carrano, A. C., Jeffrey, P. D., Bowen, Z., Kinnucan, E. R., Finnin, M. S., Elledge, S. J., Harper, J. W., Pagano, M., and Pavletich, N. P. (2000) Insights into SCF ubiquitin ligases from the structure of the Skp1-Skp2 complex. Nature 408, 381-386 51. Feldman, R. M., Correll, C. C., Kaplan, K. B., and Deshaies, R. J. (1997) A complex of Cdc4p, Skp1p, and Cdc53p/cullin catalyzes ubiquitination of the phosphorylated CDK inhibitor Sic1p. Cell 91, 221-230 52. Skowyra, D., Craig, K. L., Tyers, M., Elledge, S. J., and Harper, J. W. (1997) F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex. Cell 91, 209-219 53. Zheng, N., Schulman, B. A., Song, L., Miller, J. J., Jeffrey, P. D., Wang, P., Chu, C., Koepp, D. M., Elledge, S. J., Pagano, M., Conaway, R. C., Conaway, J. W., Harper, J. W., and Pavletich, N. P. (2002) Structure of the Cul1-Rbx1-Skp1-F boxSkp2 SCF ubiquitin ligase complex. Nature 416, 703-709 54. Hu, J., McCall, C. M., Ohta, T., and Xiong, Y. (2004) Targeted ubiquitination of CDT1 by the DDB1-CUL4A-ROC1 ligase in response to DNA damage. Nat Cell Biol 6, 1003-1009 55. Ohh, M., Park, C. W., Ivan, M., Hoffman, M. A., Kim, T. Y., Huang, L. E., Pavletich, N., Chau, V., and Kaelin, W. G. (2000) Ubiquitination of hypoxia-inducible factor requires direct binding to the beta-domain of the von Hippel-Lindau protein. Nat Cell Biol 2, 423-427 56. Min, J. H., Yang, H., Ivan, M., Gertler, F., Kaelin, W. G., Jr., and Pavletich, N. P. (2002) Structure of an HIF-1alpha -pVHL complex: hydroxyproline recognition in signaling. Science 296, 1886-1889 57. Tyers, M., and Jorgensen, P. (2000) Proteolysis and the cell cycle: with this RING I do thee destroy. Curr Opin Genet Dev 10, 54-64

23 58. Turnell, A. S., Stewart, G. S., Grand, R. J., Rookes, S. M., Martin, A., Yamano, H., El- ledge, S. J., and Gallimore, P. H. (2005) The APC/C and CBP/p300 cooperate to regulate transcription and cell-cycle progression. Nature 438, 690-695 59. Rape, M., and Kirschner, M. W. (2004) Autonomous regulation of the anaphase- promoting complex couples mitosis to S-phase entry. Nature 432, 588-595 60. Reddy, S. K., Rape, M., Margansky, W. A., and Kirschner, M. W. (2007) Ubiquitination by the anaphase-promoting complex drives spindle checkpoint inactivation. Nature 446, 921-925 61. Yen, H. C. S., and Elledge, S. J. (2008) Identification of SCF Ubiquitin Ligase Substrates by Global Protein Stability Profiling. Science 322, 923-929 62. Glotzer, M., Murray, A. W., and Kirschner, M. W. (1991) Cyclin Is Degraded by the Ubiquitin Pathway. Nature 349, 132-138 63. Pfleger, C. M., and Kirschner, M. W. (2000) The KEN box: an APC recognition signal distinct from the D box targeted by Cdh1. Genes Dev 14, 655-665 64. Ang, X. L., and Wade Harper, J. (2005) SCF-mediated protein degradation and cell cycle control. Oncogene 24, 2860-2870 65. Scheffner, M., and Whitaker, N. J. (2003) Human papillomavirus-induced carcinoge- nesis and the ubiquitin-proteasome system. Sem Cancer Biol 13, 59-67 66. Wang, M., Cheng, D. M., Peng, J. M., and Pickart, C. M. (2006) Molecular determinants of polyubiquitin linkage selection by an HECT ubiquitin ligase. EMBO J 25, 1710- 1719 67. Kee, Y., Lyon, N., and Huibregtse, J. M. (2005) The Rsp5 ubiquitin ligase is coupled to and antagonized by the Ubp2 deubiquitinating enzyme. EMBO J 24, 2414-2424 68. Wu-Baer, F., Lagrazon, K., Yuan, W., and Baer, R. (2003) The BRCA1/BARD1 heterodimer assembles polyubiquitin chains through an unconventional linkage involving lysine residue K6 of ubiquitin. J Biol Chem 278, 34743-34746 69. Morris, J. R., and Solomon, E. (2004) BRCA1 : BARD1 induces the formation of conjugated ubiquitin structures, dependent on K6 of ubiquitin, in cells during DNA replication and repair. Hum Mol Genet 13, 807-817 70. Wang, H., Wang, L., Erdjument-Bromage, H., Vidal, M., Tempst, P., Jones, R. S., and Zhang, Y. (2004) Role of histone H2A ubiquitination in Polycomb silencing. Nature 431, 873-878 71. Ben-Saadon, R., Zaaroor, D., Ziv, T., and Ciechanover, A. (2006) The polycomb protein Ring1B generates self atypical mixed ubiquitin chains required for its in vitro histone H2A ligase activity. Mol Cell 24, 701-711 72. Eddins, M. J., Carlile, C. M., Gomez, K. M., Pickart, C. M., and Wolberger, C. (2006) Mms2-Ubc13 covalently bound to ubiquitin reveals the structural basis of linkage- specific polyubiquitin chain formation. Nat Struct Mol Biol 13, 915-920 73. VanDemark, A. P., Hofmann, R. M., Tsui, C., Pickart, C. M., and Wolberger, C. (2001) Molecular insights into polyubiquitin chain assembly: crystal structure of the Mms2/Ubc13 heterodimer. Cell 105, 711-720 74. Mastrandrea, L. D., You, J. X., Niles, E. G., and Pickart, C. M. (1999) E2/E3-mediated assembly of lysine 29-linked polyubiquitin chains. J Biol Chem 274, 27299-27306 75. Brzovic, P. S., Lissounov, A., Christensen, D. E., Hoyt, D. W., and Klevit, R. E. (2006) A UbcH5/ubiquitin noncovalent complex is required for processive BRCA1-directed ubiquitination. Mol Cell 21, 873-880 76. Windheim, M., Peggle, M., and Cohen, P. (2008) Two different classes of E2 ubiquitin- conjugating enzymes are required for the mono-ubiquitination of proteins and elon- gation by polyubiquitin chains with a specific topology. Biochem J 409, 723-729 77. Kim, H. T., Kim, K. P., Lledias, F., Kisselev, A. F., Scaglione, K. M., Skowyra, D., Gygi, S. P., and Goldberg, A. L. (2007) Certain pairs of ubiquitin-conjugating enzymes (E2s) and ubiquitin-protein ligases (E3s) synthesize nondegradable forked ubiquitin chains containing all possible isopeptide linkages. J Biol Chem 282, 17375-17386 78. Lee, H., Lee, E. K., Lee, S. H., Jeon, H. B., and Yoo, Y. J. (2007) Biological significance of E2-25K and unanchored polyubiquitin chains in ubiquitin proteasome system. FASEB J 21, A1022-A1022

24 79. Yao, T. T., and Cohen, R. E. (2000) Cyclization of polyubiquitin by the E2-25K ubiquitin conjugating enzyme. J Biol Chem 275, 36862-36868 80. Christensen, D. E., Brzovic, P. S., and Klevit, R. E. (2007) E2-BRCA1 RING interactions dictate synthesis of mono- or specific polyubiquitin chain linkages. Nat Struct Mol Biol 14, 941-948 81. Bader, S., Kuhner, S., and Gavin, A. C. (2008) Interaction networks for systems biology. FEBS Lett 582, 1220-1224 82. Fields, S., and Song, O. K. (1989) A Novel Genetic System to Detect Protein Protein Interactions. Nature 340, 245-246 83. Suter, B., Kittanakom, S., and Stagljar, I. (2008) Two-hybrid technologies in proteomics research. Curr Opin Biotechnol 19, 316-323 84. Bahn, Y. S., Xue, C. Y., Idnurm, A., Rutherford, J. C., Heitman, J., and Cardenas, M. E. (2007) Sensing the environment: lessons from fungi. Nat Rev Microbiol 5, 57-69 85. Rajagopala, S. V., Titz, B., and Uetz, P. (2007) Array-based yeast two-hybrid screening for protein-protein interactions. In Yeast Gene Analysis (Stansfield, I., and Stark, M.J.R., eds) 2th Edition, Academic Press, Oxford, England 86. Jin, F. L., Avramova, L., Huang, J., and Hazbun, T. (2007) A yeast two-hybrid smart- pool-array system for protein-interaction mapping. Nat Methods 4, 405-407 87. Zhong, J. H., Zhang, H. M., Stanyon, C. A., Tromp, G., and Finley, R. L. (2003) A strategy for constructing large protein interaction maps using the yeast two-hybrid system: Regulated expression arrays and two-phase mating. Genome Res 13, 2691-2699 88. Uetz, P. (2002) Two-hybrid arrays. Current Opinion in Chemical Biology 6, 57-62 89. Rual, J. F., Hill, D. E., and Vidal, M. (2004) ORFeome projects: gateway between genomics and omics. Curr Opin Chem Biol 8, 20-25 90. Cusick, M. E., Yu, H. Y., Smolyar, A., Venkatesan, K., Carvunis, A. R., Simonis, N., Rual, J. F., Borick, H., Braun, P., Dreze, M., Vandenhaute, J., Galli, M., Yazaki, J., Hill, D. E., Ecker, J. R., Roth, F. P., and Vidal, M. (2009) Literature-curated protein interaction datasets. Nat Methods 6, 39-46 91. von Mering, C., Krause, R., Snel, B., Cornell, M., Oliver, S. G., Fields, S., and Bork, P. (2002) Comparative assessment of large-scale data sets of protein-protein interactions. Nature 417, 399-403 92. Jansen, R., Yu, H. Y., Greenbaum, D., Kluger, Y., Krogan, N. J., Chung, S. B., Emili, A., Snyder, M., Greenblatt, J. F., and Gerstein, M. (2003) A Bayesian networks approach for predicting protein-protein interactions from genomic data. Science 302, 449-453 93. Fromont-Racine, M., Mayes, A. E., Brunet-Simon, A., Rain, J. C., Colley, A., Dix, I., De- courty, L., Joly, N., Ricard, F., Beggs, J. D., and Legrain, P. (2000) Genome-wide protein interaction screens reveal functional networks involving Sm-like proteins. Yeast 17, 95-110 94. Stanyon, C. A., Liu, G., Mangiola, B. A., Patel, N., Giot, L., Kuang, B., Zhang, H., Zhong, J., and Finley Jr., R. L. (2004) A Drosophila protein-interaction map centered on cell- cycle regulators. Genome Biol 5, R96 95. Giot, L., Bader, J. S., Brouwer, C., Chaudhuri, A., Kuang, B., Li, Y., Hao, Y. L., Ooi, C. E., Godwin, B., Vitols, E., Vijayadamodar, G., Pochart, P., Machineni, H., Welsh, M., Kong, Y., Zerhusen, B., Malcolm, R., Varrone, Z., Collis, A., Minto, M., Burgess, S., McDaniel, L., Stimpson, E., Spriggs, F., Williams, J., Neurath, K., Ioime, N., Agee, M., Voss, E., Fur- tak, K., Renzulli, R., Aanensen, N., Carrolla, S., Bickelhaupt, E., Lazovatsky, Y., DaSilva, A., Zhong, J., Stanyon, C. A., Finley, R. L., White, K. P., Braverman, M., Jarvie, T., Gold, S., Leach, M., Knight, J., Shimkets, R. A., McKenna, M. P., Chant, J., and Rothberg, J. M. (2003) A protein interaction map of Drosophila melanogaster. Science 302, 1727- 1736 96. Li, S., Armstrong, C. M., Bertin, N., Ge, H., Milstein, S., Boxem, M., Vidalain, P. O., Han, J. D., Chesneau, A., Hao, T., Goldberg, D. S., Li, N., Martinez, M., Rual, J. F., La- mesch, P., Xu, L., Tewari, M., Wong, S. L., Zhang, L. V., Berriz, G. F., and al, e. (2004) A map of the interactome network of the metazoan C. elegans. Science 303, 540-543

25 97. Ito, T., Chiba, T., Ozawa, R., Yoshida, M., Hattori, M., and Sakaki, Y. (2001) A comprehensive two-hybrid analysis to explore the yeast protein interactome. Proc Natl Acad Sci U S A 98, 4569-4574 98. Rual, J. F., Venkatesan, K., Hao, T., Hirozane-Kishikawa, T., Dricot, A., Li, N., Berriz, G. F., Gibbons, F. D., Dreze, M., Ayivi-Guedehoussou, N., Klitgord, N., Simon, C., Boxem, M., Milstein, S., Rosenberg, J., Goldberg, D. S., Zhang, L. V., Wong, S. L., Franklin, G., Li, S., Albala, J. S., Lim, J., Fraughton, C., Llamosas, E., Cevik, S., Bex, C., Lamesch, P., Sikorski, R. S., Vandenhaute, J., Zoghbi, H. Y., Smolyar, A., Bosak, S., Sequerra, R., Doucette-Stamm, L., Cusick, M. E., Hill, D. E., Roth, F. P., and Vidal, M. (2005) Towards a proteome-scale map of the human protein-protein interaction network. Nature 437, 1173-1178 99. Stelzl, U., Worm, U., Lalowski, M., Haenig, C., Brembeck, F. H., Goehler, H., Stroe- dicke, M., Zenkner, M., Schoenherr, A., Koeppen, S., Timm, J., Mintzlaff, S., Abraham, C., Bock, N., Kietzmann, S., Goedde, A., Toksoz, E., Droege, A., Krobitsch, S., Korn, B., Birchmeier, W., Lehrach, H., and Wanker, E. E. (2005) A human protein-protein interaction network: a resource for annotating the proteome. Cell 122, 957-968 100. Formstecher, E., Aresta, S., Collura, V., Hamburger, A., Meil, A., Trehin, A., Reverdy, C., Betin, V., Maire, S., Brun, C., Jacq, B., Arpin, M., Bellaiche, Y., Bellusci, S., Bena- roch, P., Bornens, M., Chanet, R., Chavrier, P., Delattre, O., Doye, V., Fehon, R., Faye, G., Galli, T., Girault, J. A., Goud, B., de Gunzburg, J., Johannes, L., Junier, M. P., Mi- rouse, V., Mukherjee, A., Papadopoulo, D., Perez, F., Plessis, A., Rosse, C., Saule, S., Stoppa-Lyonnet, D., Vincent, A., White, M., Legrain, P., Wojcik, J., Camonis, J., and Daviet, L. (2005) Protein interaction mapping: A Drosophila case study. Genome Res 15, 376-384 101. Ho, Y., Gruhler, A., Heilbut, A., Bader, G. D., Moore, L., Adams, S. L., Millar, A., Taylor, P., Bennett, K., Boutilier, K., Yang, L., Wolting, C., Donaldson, I., Schandorff, S., Shew- narane, J., Vo, M., Taggart, J., Goudreault, M., Muskat, B., Alfarano, C., Dewar, D., Lin, Z., Michalickova, K., Willems, A. R., Sassi, H., Nielsen, P. A., Rasmussen, K. J., Ander- sen, J. R., Johansen, L. E., Hansen, L. H., Jespersen, H., Podtelejnikov, A., Nielsen, E., Crawford, J., Poulsen, V., Sorensen, B. D., Matthiesen, J., Hendrickson, R. C., Gleeson, F., Pawson, T., Moran, M. F., Durocher, D., Mann, M., Hogue, C. W., Figeys, D., and Tyers, M. (2002) Systematic identification of protein complexes in Saccharomyces cerevisiae by mass spectrometry. Nature 415, 180-183 102. Uetz, P., Giot, L., Cagney, G., Mansfield, T. A., Judson, R. S., Knight, J. R., Lockshon, D., Narayan, V., Srinivasan, M., Pochart, P., Qureshi-Emili, A., Li, Y., Godwin, B., Conover, D., Kalbfleisch, T., Vijayadamodar, G., Yang, M., Johnston, M., Fields, S., and Roth- berg, J. M. (2000) A comprehensive analysis of protein-protein interactions in Saccha- romyces cerevisiae. Nature 403, 623-627 103. Adachi, K., Lakka, V., Zhao, Y., and Surrey, S. (2004) Ubiquitylation of nascent globin chains in a cell-free system. J Biol Chem 279, 41767-41774 104. Pavithra, S. R., Kumar, R., and Tatu, U. (2007) Systems analysis of chaperone networks in the malarial parasite Plasmodium falciparum. PLoS Comput Biol 3, 1701- 1715 105. LaCount, D. J., Vignali, M., Chettier, R., Phansalkar, A., Bell, R., Hesselberth, J. R., Schoenfeld, L. W., Ota, I., Sahasrabudhe, S., Kurschner, C., Fields, S., and Hughes, R. E. (2005) A protein interaction network of the malaria parasite Plasmodium falciparum. Nature 438, 103-107 106. Parrish, J. R., Yu, J., Liu, G., Hines, J. A., Chan, J. E., Mangiola, B. A., Zhang, H., Pacifico, S., Fotouhi, F., DiRita, V. J., Ideker, T., Andrews, P., and Finley, R. L., Jr. (2007) A proteome-wide protein interaction map for Campylobacter jejuni. Genome Biol 8, R130 107. Furstenthal, L., Swanson, C., Kaiser, B. K., Eldridge, A. G., and Jackson, P. K. (2001) Triggering ubiquitination of a CDK inhibitor at origins of DNA replication. Nat Cell Biol 3, 715-722 108. Calderwood, M. A., Venkatesan, K., Xing, L., Chase, M. R., Vazquez, A., Holthaus, A. M., Ewence, A. E., Li, N., Hirozane-Kishikawa, T., Hill, D. E., Vidal, M., Kieff, E., and Jo-

26 hannsen, E. (2007) Epstein-Barr virus and virus human protein interaction maps. Proc Natl Acad Sci U S A 104, 7606-7611 109. de Folter, S., Immink, R. G. H., Kieffer, M., Parenicova, L., Henz, S. R., Weigel, D., Bus- scher, M., Kooiker, M., Colombo, L., Kater, M. M., Davies, B., and Angenent, G. C. (2005) Comprehensive interaction map of the Arabidopsis MADS box transcription factors. Plant Cell 17, 1424-1433

Chapter 2

A comprehensive framework of E2-RING E3 interactions of the human ubiquitin-proteasome system

Molecular Systems Biology (2009) 5: 295

29 A comprehensive framework of E2-RING E3 interactions of the human ubiquitin-proteasome system

Sjoerd J.L. van Wijk1, Sjoerd J. de Vries2, Patrick Kemmeren1, Anding Huang2, Rolf Boelens2, Alexandre M. J. J. Bonvin2 and H. Th. Marc Timmers1

1 Department of Physiological Chemistry, Division of Biomedical Genetics, University Medical Center Utrecht, Universiteits- weg 100, 3584 CG Utrecht, the Netherlands 2 Department of NMR Spectroscopy, Bijvoet Center for Biomolecular Research, Utrecht University, Padualaan 8, 3584 CH, Utrecht, the Netherlands

Abstract Covalent attachment of ubiquitin to substrates is crucial to protein degradation, transcription regulation and cell signaling. Highly specific interactions between ubiquitin-conjugating enzymes (E2) and ubiquitin-protein E3 ligases fulfill essential roles in this process. We performed a global yeast-two hybrid screen to study the specificity of interactions between catalytic domains of the 35 human E2s with 250 RING-type E3s. Our analysis revealed over 300 high-quality interactions, uncovering a large fraction of new E2-E3 pairs. Both within the E2 and the E3 cohorts, several members were identified that are more versatile in their interaction behavior than others. We also found that the physical interactions of our screen compare well with reported functional E2-E3 pairs in in vitro ubiquitination experiments. For validation we confirmed the interaction of several versatile E2s with E3s in in vitro protein interaction assays and we used mutagenesis to alter the E3 interactions of the E2 specific for K63 linkages, Ubc13(UBE2N), towards the mixed chain linkage-specific UbcH5B(UBE2D2). Our data provide a detailed, genome-wide overview of binary E2-E3 interactions of the human ubiquitination system.

Introduction Modification of proteins with ubiquitin (ubiquitination) regulates the degradation of proteins by the 26S proteasome, but also serves a wide variety of other cellular processes, ranging from endocytosis to cell death (1). Complex and tightly regulated protein-protein interactions of three key enzymes, ubiquitin-activating enzymes (E1s), ubiquitin-conjugating enzymes (E2s) and ubiquitin protein ligases (E3s), allow that ubiquitin becomes activated and covalently linked to substrates (1, 2). These E1, E2 and E3 enzymes are operating in a hierarchical system. In the human genome, two E1s have been identified, responsible for the activation and transfer of ubiquitin (Ub) to the E2s (3, 4). In contrast, over 30 E2s have been identified, and these have in common a highly conserved 150-200 amino acid catalytic domain (5). This UBC domain consists of several alpha-helices, beta-sheets and variable loop regions, which are surrounding an active-site cysteine residue (6). Flanking this conserved region, additional N- and/or C-terminal extensions have been found that are involved in substrate selection, dimerization and additional processes (7-9). E2s are transferring the activated ubiquitin to E3s, characterized by the presence of either a HECT (Homologous to E6-AP Carboxy-Terminus) (10, 11), U-box (11) or Really Interesting New Genes (RING) domain (12).

30 The RING-finger domain is a highly conserved pattern of cysteine and/or histidine residues, chelating two atoms of zinc in a highly typical cross-brace structure. RING-type E3s can bind E2s through their RING-finger, but in contrast with HECT-type E3s, they catalyze the direct transfer of ubiquitin from the E2 to a substrate (12). Interactions between E2s and E3s are thought to be important since the interacting E2 dictates the type of inter-ubiquitin linkages and thereby determine the fate of the substrate. For example, ubiquitin chains linked through K48 (i.e. by UbcH5B(UBE2D2)) are targeted by the 26S proteasome for degradation (13), whereas Ubc13(UBE2N)-mediated K63-linked ubiquitin chains are involved in signaling (14).

A limited number of clues, obtained by experimental structures, showed that E3 interactions are occurring in the UBC fold (15-20). Large-scale, genome-wide protein-protein interaction (PPI) studies can make an important contribution in the understanding of the specificity of E2-E3 interactions. Aiming to map E2-E3 interactions on a genome-wide scale and to study the underlying selectivity of these binary complexes, we have used yeast two-hybrid (Y2H) screen, including 35 E2 UBC-folds and 250 E3 RING-finger domains. Our analysis revealed more than 300 binary E2-E3 interactions, of which a large proportion was previously not known. As expected we found that multiple E2 and E3 enzymes could interact with multiple E3s and E2s, respectively. Some of the E2s and E3s appeared to be more versatile than others and showed more interactions than the average number one would expect, thereby con- ferring to the concept of hubs in protein interaction networks. Comparing the physical E2-E3 interactions with reported E2-E3 pairs assayed in in vitro ubiquitination reactions revealed a high level of agreement. One of the highly interacting E2s, UBE2U, a previously unknown C- terminal extended, class III E2, showed the highest number of interactions. We validated the interaction between UBE2U and the E3 ligase MDM2 and also several interactions between hub E2s and E3s by independent GST pull-down analysis, indicating a high level of confidence of the found E2-E3 interactions. Based on the interaction network we generated mutants of Ubc13(UBE2N) to mimic the E3 interaction pattern of UbcH5B(UBE2D2). This resulted in new E3 interactions, which allows manipulation of ubiquitination pathways in living cells. The experimental interactions found in this study provide a global, high-resolution framework for structural and functional studies towards the organization and selectivity between E2 and E3 enzymes.

Results Central in the process of ubiquitination are the interactions between ubiquitin-conjugating enzymes (E2) and ubiquitin protein ligases (E3). In order to gain more insight in the selectivity of binary E2-E3 complexes, we devised a systematic and genome-wide yeast two-hybrid (Y2H) screen. Previous work by others and by our group showed that these interactions are highly selective, and that Y2H assays are powerful tools to study these specificities (18, 21). This screen was performed to generate a framework of binary E2-E3 interactions. Interac- tions between the individual UBC-folds and the RING-finger domains are the major contribu-

31 tors to E2-E3 interactions (22). Therefore, in a first approach we limited ourselves to screen for the interactions of human UBC-folds with RING-finger domains.

Annotating the human ubiquitin-conjugating enzymes It has been estimated previously that within the human genome, 30-40 E2s are present (1). In order to precisely annotate the complete human E2 superfamily, we searched human cDNA databases for candidates. Our search revealed 52 E2 family members that contain the UBC-fold (Supplementary Table I). Of these 52 E2s, we identified 16 pseudogenes for which no expressed protein-coding mRNAs could be found. We also identified an additional E2-like enzyme, UBE2N-like (UBE2NL), lacking the central ubiquitin-accepting cysteine. Among the remaining 35 enzymes, three E2s were annotated as putative E2s and no ubiquitin(-like) conjugating activity has been reported for these (HBUCE1/UbcH5D(UBE2D4), UBE2W and UBE2U) so far. Recently, it has been shown that NCE2(UBE2F) has the ability to act as a second NEDD8 E2 and prefers interaction with the Rbx2 RING-protein, adding divergence to the NEDDylation system (23). So at this moment, 31 E2 enzymes are described to be involved in the process of ubiquitin or ubiquitin-like protein conjugation. Among these E2s, there are well-known examples with biochemically active roles in protein ubiquitination, such as UbcH5B(UBE2D2) and UbcH7(UBE2L3) (24, 25). Apart from the wild-type E2s, we also included the UbcH5B(UBE2D2) K63E mutant, which previously has been shown to fail to interact with the wild-type CNOT4 RING-finger, but showed an altered-specificity interaction with the D48K/E49K double mutated CNOT4 (21). Using multiple sequence alignments, we determined the boundaries of the UBC-fold. These 36 UBC-folds were fused to the C- terminus of LexA, DNA sequence verified, transformed in MATα cells and arranged in a standardized array (BD-E2 array) (Figure 2.1B & D).

Selection of human RING-type E3s Recent genome-wide annotations of human E3 ubiquitin protein ligases have indicated that there are 300 RING-type and 9 U-box-type E3s present within the human genome (26). To select a representative collection of these domains, genome-wide database searches were initiated. First, we performed a SMART search using the curator-defined RING signature (SM00184) and this yielded 520 hits of which 314 of them were unique. Second, we performed a ScanProSite using ProSite pattern PS50089 against the human sequences deposited in SwissProt + Trembl (UniProt) generating a multiple sequence alignment of 627 hits in 611 UniProt sequences. Thirdly, we performed a PSI-BLAST search of the human CNOT4 sequence against SwissProt; sequences with “RING” in their name were retained during the process, with “ubiquitin ligase” in some steps retained and without these names removed in some steps. Finally, all sequences with an expectation of 1.1 and lower were selected and this yielded 217 hits. Finally, we constructed a list of 308 Ensembl entries containing InterPro domain IPR001841. Next, these entries were cleared of redundancies and converted to Uni- Prot codes and full-length protein sequences which were retrieved from SwissProt.

A D UBE2D2 UBE2A UBE2E2 UBE2J2 UBE2Q2 UBE2W UBE2A UBE2E2 UBE2J2 UBE2Q2 UBE2W UBE2D2

K63E K63E 1. 2. Spot 1-3: Biological LexA-E2 triplicates UBE2D2 UBE2B UBE2E3 HIP2 CDC34 USE1 UBE2B UBE2E3 HIP2 CDC34 USE1 UBE2D2

Spot 4: Empty LexA-vector CAF40 UBE2C UBE2F UBE2L3 UBE2R2 BIRC6 UBE2C UBE2F UBE2L3 UBE2R2 BIRC6 CAF40 3. 4.

CNOT2-15 UBE2D1 UBE2G1 UBE2L6 UBE2S UBE2D2 K63E UBE2D1 UBE2G1 UBE2L6 UBE2S UBE2D2 K63E CNOT2-15

B LexA in A UBE2D2 UBE2G2 UBE2M UBE2T MQ UBE2D2 UBE2G2 UBE2M UBE2T MQ LexA in A

EGY48 a UBE2D3 UBE2H UBE2N UBE2U MQ UBE2D3 UBE2H UBE2N UBE2U MQ EGY48 a UBE2E1 UBE2T UBER1 UBE2U UBE2N UBE2Z K63E UBE2D2 UBE2Q1 UBE2C UBE2L3 UBE2K UBEQ2 UBER2 UBE2G2 UBE2D2 UBE2D1 UBE2D4 UBE2D4 UBE2E3 UBE2W UBE2L6 UBE2H UBE2J2 UBE2A UBE2B UBE2E2 UBE2G1 UBE2J1 UBE2S UBE2I UBE2M UBE2F UBE2O BIRC6 UBE2V1 UBE2V2

LexA UBE2D2 UBE2D4 UBE2I UBE2O UBE2V1 MQ UBE2D4 UBE2I UBE2O UBE2V1 MQ UBE2D2

TUB1 UBE2D2 K63E UBE2E1 UBE2J1 UBE2Q1 UBE2V2 MQ UBE2E1 UBE2J1 UBE2Q1 UBE2V2 MQ UBE2D2 K63E C

Figure 2.1. Outline array-based, mating yeast two-hybrid matrix screen (A) Organization of LexA-E2 fusions as three independent bacterial triplicate clones and an empty LexA vector. (B) LexA-E2 protein expression levels. Total protein lysates of yeast cells transformed LexA-E2 fusions were resolved on SDS-PAGE, transferred to membranes and probed with either a LexA antibody or yeast tubulin (TUB1). (C) Overview of experimental yeast two-hybrid screening procedure. EGY48alpha cells were transformed with 36 LexA-E2 fusions and EGY48a cells were transformed with 250 B42-RING-fusions. Screening for pair-wise E2-E3 interactions was done by standardized array-based mating between EGY48alpha and EGY48a cells on YPD for 24 h. at 30oC, followed by diploid selection on SC HWU- for 48 h. at 30oC. Interactions were visualized by transferring diploids on SC HWU- X-Gal (colorimetric selection) or on SC HWUL- (auxotrophic selection) under both B42-inducing (galactose as main carbon source) and repressive (glucose) conditions. Digital images of the interactions were taken at three, 24-hour interval time points. (D) Overview of standardized LexA-E2 array. Each spot represents EGY48alpha cells transformed with the indicated LexA-E2 fusion construct. Spots with a red square contained LexA-UbcH5B(UBE2D2) and LexA-UbcH5B(UBE2D2) K63E constructs and were mated throughout the entire procedure with EGY48a cells transformed with B42-CNOT4 N63 as positive and negative interaction controls, respectively. Spots with a green square were transformed with LexA-Caf40 and LexA-CNOT2-15 and served as auto-activating controls. Blue squared spots represent either EGY48A cells transformed with LexA-empty vector and an untransformed EGY48alpha strain, as mating/growing controls.

To locate the RING signature, a RING-finger matching tool was written and from the 295 sequences, we retrieved 313 RING-finger sequences. All RING-fingers were aligned in a multiple sequence alignment using Clustal W. The resulting alignment was carefully corrected by hand, especially with respect to zinc-binding regions. Regions that could not be aligned with any degree of reliability, for example sequences containing more than four C/HXXC/H, were

33 removed from the alignment. In total, 313 RINGs were successfully aligned, 161 of them being classified as proper RING (matching the CXXC[X](10-51)C[X](1-3)H[X](2-5)CXXC[X](4- 60)CXXC signature), 16 as PHD (matching CXXC[X]( 8-31)C[X](2-4)C[X](2-5)HXXC[X](12- 60)CXXC) and 11 of them as FYVE (matching CXXC[X](12-40)CXXXXCXXC[X](13-80)CXXC). We identified 111 E3s being uncommon, as they match the following signature [C/H]XX[C/H][X](0-86)[C/H][X](0-5)[C/H][X](0-29)[C/H]XX[C/H][X](0-86)[C/H/D], wherein X could be any amino acid, except a C or an H. We also found 14 E3s being incomplete, belonging to none of the signatures. Together, 177 sequences are expected to act like E3 ligases, for 25 this is not expected, since they lack a complete RING signature and for the remaining 111 this remains unclear. Of these RING sequences we obtained clones for 250 human RING- finger/U-box E3 ligases (80.9 % coverage of total (250/309)), based on cDNA availability. The RING-finger/U-box domains were subcloned as C-terminal fusions with the B42 activation domain, under control of a galactose-inducible promoter. All B42-E3 constructs were verified by DNA sequencing and arrayed as a library of 250 B42-RING fusions in MATa yeast cells (AD- E3 array) (Supplementary Table II). Correct expression of the B42-fusions was evaluated using immunoblotting (data not shown).

Genome-wide E2-E3 yeast-two hybrid screen Systematic mating crosses between the BD-RING and AD-E2 arrays allowed scoring for putative interactions between the arrays of E2- and E3-domains. Each of the 36 LexA-E2 fusions was arrayed in three independent yeast transformants. These three spots were placed in a small square together with LexA-only, measuring E2-E3 interactions in triplicate parallel to the empty vector control (Figure 2.1A). EGY48a cells transformed with B42-RING fusions were spotted on top of the BD-E2 arrays. After mating, diploids were selected and screened for interactions based on colorimetric (LacZ-set) or auxotrophic (LEU2-set) selection. A semi- automated pipeline of digitalized plate photographs followed by spot size/color intensity quantification on three time points, allowed us to quantify interactions in an unbiased and sensitive manner (Figure 2.1C).

Selection of interactors Close to 10,000 interactions within the 36 x 250 E2-E3 matrix were assessed independently on LacZ and LEU2 reporter read-outs. It has been shown previously that, in general, interactions that activate both reporters are of higher confidence and more reproducible than interactions that activate only one reporter (27). Therefore our definition of an E2-E3 interaction is a galactose-dependent increase in both reporter activities. In order to select the shared interactions, we determined percentile values for both the LacZ- and the LEU2-set. Since any selection strategy has the potential risk of discarding true positive and including false- positive interactions, selection of the most suitable percentile cut-off was based on all E3 interactions for a given E2. Given the variability of spot-sizes in the LEU2 dataset, we con- cluded that interaction values from 50-200 pixels are the consequence of either noise (false- positives), absence of LEU2-reporter activation (true-negative interactions) or weak-

34 activating interactions. To avoid the risk of discarding true interactions, all interactors in the LEU2-set above the 75th percentile were selected (2238 signals). From the LacZ-set with a total of 577 blue signals we selected all interactors above the 50th percentile (562 interactors, 94.4 % of total). Combining these two sets resulted in 346 high-confidence interactors (15.5 % of LEU2-set and 61.6 % of the LacZ-set), which we named the LL-set (Figure 2.2A and Supplementary Table III). An overview of all the identified E2-E3 interactions scored by both LacZ staining and LEU2 selection is given in Supplementary Tables IV and V, respectively.

346 LacZ 216 p-value = 1892 LEU2 50 th percentile 1.7x10-81 75 th percentile

Figure 2.2. Topological characterization of the E2-E3 network (A) Venn-diagram showing the shared E2-E3 interactions in the 75th percentile selected LEU2-set and the 50th percentile selected LacZ-set and the overlap between them. (B) Distribution of the E3 connectivity for each individual E2. Depicted are the numbers of E3 interactions observed in the shared LacZ- and LEU2-set among all interactions of a single E2. (C) Distribution of the E2 connectivity for each individual E3, idem as panel (B).

Human E2-E3 domain-domain interactions Inspecting the connectivity, that is the numbers of E3 interactions for each E2 in the LL-set, we identified 20 E2s that were connected to at least one E3 and 16 E2s that did not show any E3 interaction (Figure 2.2B). Screening the E3 domains, we found 104 E3s that were

35 connected to at least one E2 and 147 E3s that did not show any E2 interactions. An overview of the E3 connectivity is shown in Figure 2.2C. The E2s showing the highest number of interactions were UBE2U (52 interactors), followed by UbcH5B(UBE2D2) (35), UbcH5C(UBE2D3) and HBUCE1/UbcH5D(UBE2D4) (33), Ubc13(UBE2N) (29), UbcH5A(UBE2D1) (28) and the K63E mutant of UbcH5B(UBE2D2) (27). Next, UBE2W, UbcH6(UBE2E1), UbcH9(UBE2E3), UBC7(UBE2G2), USE1(UBE2Z), UbcH8(UBE2L6) and NCUBE2(UBE2J2) showed 8-22 E2 interactions. Smaller numbers of interactions were found for UBCH(UBE2H) (three), UBE2Q(UBE2Q1) (two), UbcH10(UBE2C) (one), UbcH8(UBE2E2) (one) and UBE2Q2 (one) (Figure 2.2A). To investigate the possibility that differences in E3 interaction patterns were due to differences in LexA protein expression, immunoblot analysis was performed. Figure 2.1B shows that the majority of LexA-E2 fusions are expressed at similar levels and that only some E2s are expressed at low levels (UBE2Q2 and Ubc9(UBE2I)). Please note that we find one E3 interaction (PCGF2) for UBE2Q2 and, as expected, no interactions for the SUMO-specific Ubc9(UBE2I). Expression of NCE2(UBE2F) and HIP2(UBE2K) could not be detected, which correlates with the absence of E3 interactions. Of the other 13 non-interactors, five E2s (Ubc12(UBE2M), apollon(BIRC6), E2-230K(UBE2O), UEV-1A(UBE2V1) and MMS2(UBE2V2)) have not been linked to ubiquitin conjugation, but in some cases to conjugation of ubiquitin-like proteins. It is important to note that we do not observe an over-representation of a particular class of E2s, indicating that N- or C-terminal extensions to the UBC-fold are not confounding the results (data not shown). We also analyzed a potential effect of the position of the RING within the full-length E3 on interaction potential (Supplementary Table II). This indicated that N-terminal RING- domains score slightly lower in E2 interactions as compared C-terminal RINGs (Supplemen- tary Figure 1). Together, these analyses indicate that the interactions are distributed broadly over the different classes and types of E2 and E3 enzymes.

Topological characteristics of the E2-E3 interaction network The E2-E3 interaction map can be regarded as a network, in which individual E2s and E3s are considered as nodes (Figure 2.3), which are connected to each other via interactions. In total, we found 346 interactions between 20 E2 and 104 E3 domains, with an average of 2.8 interactions per domain (Figure 2.3). The connectivity of both the E2s and E3s interactions indicate the presence of several high and lesser connected E2 and E3 domains. Our analysis revealed that UBE2U, the UbcH5A-D(UBE2D1-4)-family of highly homologous E2s, Ubc13(UBE2N) and UbcH6(UBE2E1) and UbcH9(UBE2E3) showed a high number of E3 interactions (Figure 2.2B & 2.3). UBE2U is an E2 enzyme, which carries a C-terminal extension attached to its UBC-fold and thus belongs to the class III group of E2 enzymes. At present no interaction data for UBE2U or an ubiquitin (-like) conjugation activity has been described in the literature, which may be related it its restricted expression pattern in the urogenital tract (based on mRNA expression profile, data not shown). Members of the UbcH5A-D(UBE2D1- 4)-family are highly conserved E2 enzymes and homologous to yeast UBC4/5. These E2s are

36 involved in the degradation of misfolded and short-lived proteins and have been shown to be active in ubiquitination (28).

E2 hubs

E3 hubs

Figure 2.3. E2-E3 network A network graph of binary E2-E3 domain interactions involving 114 E2 and E3 domains linked via 257 interactions. Nodes (domains) are shown as circles and interactions between them as black lines. E2 domains are shown as red circles and E3 domains as blue circles. The size of each node is linear related to the number of links of that node. Red square depicts E2 hub proteins, blue square hub E3s. Figure was generated using Cytoscape v.2.6.1.

Ubc13(UBE2N) forms heterodimers with UEV-1A(UBE2V1) and MMS2(UBE2V2) and is involved in K63 ubiquitin chain assembly (14). UBC7(UBE2G2) is the human homolog of the yeast UBC7 and resides in the endoplasmatic reticulum (ER). UBC7(UBE2G2) has been reported to interact with the E3 ligase gp78 (AMFR) and this E2-E3 pair has a role in endoplasmatic reticulum associated degradation (ERAD) (29). Another characteristic of hub proteins is that they connect to proteins with only a few interactions (30). This is clearly illustrated by the E3 interaction patterns of UBE2U and UBC7(UBE2G2). On the other hand UbcH5A- D(UBE2D1-4) interacts with well-connected E3s.

37 The hierarchical nature of the ubiquitin-conjugation machinery allows a flux of activated ubiquitin from the activation by the E1 via E3s to downstream targets. The role of E2s in this process is to maintain the robustness of the system as a whole and to connect the activation of ubiquitin with its conjugation to substrates. Therefore, deletion of the E2s with high numbers of E3 interactions could have severe consequences for the complete ubiquitination network. To investigate this, we systematically removed (hub) E2s from the E2-E3 network and quantified the number of E3s that become unconnected (Supplementary Figure 2). By removing UBE2U, 32 E3s become unconnected, while 20 remain connected to another E2. In contrast, removing UbcH5B(UBE2D2) has only very mild effects, since many E3s interact with the homologous UbcH5A(UBE2D1), UbcH5C(UBE2D3) and UbcH5D(UBE2D4) and to a lesser extent with UbcH6(UBE2E1) and UbcH9(UBE2E3). Interestingly, removal of UBC7(UBE2G2) from the network leaves seven E3s unbound. Since this E2 is localized in the endoplasmatic reticulum and is involved in ERAD, it fulfills a central role in the network. Finally, deletion of USE1(UBE2Z) from the network disconnects two RING-domains from the recently-identified second E1 for ubiquitin, Uba6.

Apart from the wild-type E2s tested in this screen, the UbcH5B(UBE2D2) loop 1 mutant (K63E) was also included. As shown previously, contrary to wild-type UbcH5B(UBE2D2), the K63E mutant is not able to bind to wild-type CNOT4 (21). However, when mutating the UbcH5B(UBE2D2)-interacting residues in the CNOT4 RING-finger such that they swap their charges (D48K/E49K), complete rescue of interaction could be achieved. Interestingly, UbcH5B(UBE2D2) K63E does not show a dramatic change in its E3 interaction pattern compared to wt UbcH5B(UBE2D2) and several E3s are interacting with both wild-type and the mutant UbcH5B(UBE2D2). Strikingly, UbcH5B(UBE2D2) K63E gained interactions with LOC51136, NFX1 and RNF8 as compared to the wild-type enzyme. Inclusion of the D48K/E49K mutant of the CNOT4 RING-finger shows loss of binding with the UbcH5A- D(UBE2D1-4) family, UbcH6(UBE2E1), UbcH9(UBE2E3) and UBE2W and, as expected it displays specific binding with UbcH5B(UBE2D2) K63E. The mutant CNOT4 also showed a stronger interaction with Ubc13(UBE2N) as compared to the wild-type E3 (Figure 2.4).

The presence of highly connected E2s and E3s provides the network with robustness and resistance against small perturbations. This implies the presence of selective structural features involved in E2-E3 interaction specificity. To study the underlying molecular determinants, E2s were selected that exhibit more than one E3 interaction. First, the degree of overlap in E3 interaction between these E2s was quantified as the “hub overlap index” (Supple- mentary Figure 3). These numbers indicate the extent by which a certain E3 is used for E2- binding. The hub overlap index indicates that the homologous E2s, like UbcH5B(UBE2D2), UbcH5C(UBE2D3), UbcH5C(UBE2D4) and to a lesser extent UbcH5A(UBE2D1), cluster together in E3 interaction pattern, indicating that enzyme function is directly coupled to structural similarities. The same is true for the homologous UbcH6(UBE2E1) and UbcH9(UBE2E3) proteins, which share interactions with many E3s. Surprisingly, the highly similar

38 UbcH8(UBE2E2) displayed an E3-interaction pattern distinct from the profiles of UbcH6(UBE2E1) and UbcH9(UBE2E3). The E2s Ubc13(UBE2N) and UBE2W are highly connected, but they exhibit lower amounts of overlap. Unexpectedly, the highly-connected UBE2U does not cluster together with the other highly interacting E2s, but it interacts with many E3s, suggesting a general role in ubiquitination in the urogenital tract.

E49K

D48E

CNOT4-N63 CNOT4-N63 UBE2A A B UBE2B UBE2C UBE2D1

UBE2D2 K63E UBE2D2 UBE2D2 1810074P20RIK 1810074P20RIK UBE2D3 CADPS2 CADPS2 UBE2D4 CNOT4 UBE2E1 UBE2E2 DTX1 UBE2E3 E4A UBE2F KF-1 KF-1 UBE2G1 LOC51255 UBE2G2 UBE2H MAP3K1 MAP3K1 UBE2I NFX1 UBE2J1 MIB1 RING3 UBE2J2 RNF103 UBE2K UBE2L3 RNF11 RNF11 UBE2L6 RNF111 RNF111 UBE2M RNF130 RNF130 UBE2N UBE2O RNF150 RNF150 UBE2Q1 RNF167 RNF167 UBE2Q2 RNF43 RNF43 UBE2R1 UBE2R2 ZFP294 UBE2S ZNF364 ZNF364 UBE2T ZNRF1 UBE2U UBE2V1 ZNRF4 ZNRF4 UBE2V2 UBE2W UBE2Z BIRC6 UBE2D2 K63E

Figure 2.4. Mutant UbcH5B and mutant CNOT4-N63 interactions (A) Overview of RING-E3 interaction patterns between wild-type and UbcH5B(UBE2D2) K63E. E3s in blue interact with both wild-type and mutant UbcH5B(UBE2D2), in green only with wild-type UbcH5B(UBE2D2) and in pink only with the K63E mutant. (B) E2 interaction pattern of wild-type and mutant CNOT-N63. White indicates no interaction, green an interaction with wt CNOT4-N63, red with the mutant and blue with both.

39 To compare the hub overlap index with structural similarities between E2s, we calculated the percentages of sequence identity either for the complete UBC-fold or only for the predicted E3-interacting interface (Supplementary Figure 4). The distribution of sequence identity between E2 enzymes roughly equals the pattern of the hub overlap index. For most pairs, the sequence conservation clearly relates to the hub overlap, as indicated for the sub- clusters UbcH5A-D(UBE2D1-4), UbcH6(UBE2E1), UbcH9(UBE2E3), Ubc13(UBE2N) and UBE2W. This underscores a structural basis among these E2 enzymes for E2 enzymes. In contrast, UbcH8(UBE2E2) is highly similar to UbcH6(UBE2E1) and UbcH9(UBE2E3), but it displays little interaction overlap. UBE2T is interesting in this respect as it is equally similar to UbcH6(UBE2E1), UbcH8(UBE2E2) and UbcH9(UBE2E3), but the UBE2T interaction pattern only overlaps with that of UbcH8(UBE2E2). The similarity in the E3 interaction surfaces of UbcH8(UBE2L6) and the UbcH5A-D(UBE2D1-4) family is limited, but their E3-interaction profiles display a significant degree of overlap. Finally, the screen identified 11 E3 interactions for the USE1(UBE2Z) E2 enzyme. Recently, it has been shown that this E2 enzyme can become ubiquitin-loaded by the newly-identified E1 enzyme Uba6 in a manner discriminating against other E2 enzymes (4). The observed USE1(UBE2Z) interactions could provide a link between Uba6 and RING-E3 ligases.

Among the screened E3 domains, RNF150, ZNRF4, CADPS2 and MUL1 showed the highest number of E2 interactions (Figure 2.3). For RNF150, no E3 ligase activity has been described, but the protein belongs to the human Goliath family (31). Strikingly, two other Goliath family member E3s, RNF130 and RNF167 were also found to act as hub proteins, with only 52% sequence identity between the RING-fingers. The RNF130 protein (hGoliath) contains a protease-associated domain, a transmembrane domain and a RING-finger domain and is the human homolog of Drosophila’s g1, a zinc-finger protein, involved in embryonic development (32). RNF130 displayed auto-ubiquitination activity when interacting with UbcH5A(UBE2D1) and UbcH5C(UBE2D3), but not with a number of other E2s (32). It would be interesting to test whether RNF167 and RNF130 are also functional E3s. A second example of multiple, highly homologous family members acting as hub E3s come from the E3 ligases ZNRF4, ZNRF1 and ZNRF2, having 51% sequence identity between their RING-finger domains. These show ten, six and five E2 interactions, respectively. The ZNRF proteins are implicated in spermatogenesis and the establishment and maintenance of neuronal trans- mission and plasticity mediated by their E3 ligase activity (33, 34). CAPDS2 (Calcium dependent secretion activator 2) is involved in the exocytosis of neurotransmitter vesicles and cerebellar development (35). No E3 domain annotations could be found for this protein, and information about interacting E2s or ubiquitin-ligating activity is absent. MUL1 (MULAN) has been implicated recently in the establishment of mitochondrial dynamics and mitochondria- to-nucleus signaling, in a manner dependent on its RING-finger domain. MUL1 displayed au- to-ubiquitination activity when tested with Ubc4, the yeast homolog of the human UbcH5A- D(UBE2D1-4)-family (26).

40 Quality of the physical E2-E3 interaction network Comparing the E2-E3 interactions obtained in this screen with E2-E3 literature-curated interactions (LCIs) evaluates the quality of the interactions in the E2-E3 network. Physical interactions between E2 and E3 enzymes are required for efficient ubiquitination in vivo and in vitro and therefore for their catalytic function. Besides this, comparison the novel E2-E3 pairs generated by our screen with annotated functional E2-E3 pairs provides insight in the functionality of the E2-E3 pairs. We first selected the top 10% interacting E3s and compared their physical E2 interaction patterns with information obtained by low-throughput E2-E3 experiments. As shown in Figure 2.5, 25 hub E3s showed a total number of 209 E2 interactions covering 60% of the total number of E2-E3 two-hybrid interactions.

LCI + and Y2H + LCI - and Y2H - LCI + and Y2H - LCI - and Y2H + New interactions Protein E3 ligase Role active as UBE2A UBE2B UBE2C UBE2D1 UBE2D2 UBE2D3 UBE2D4 UBE2E1 UBE2E2 UBE2E3 UBE2F UBE2G1 UBE2G2 UBE2H UBE2I UBE2J1 UBE2J2 UBE2K UBE2L3 UBE2L6 UBE2M UBE2N UBE2O UBE2Q1 UBE2Q2 UBE2R1 UBE2R2 UBE2S UBE2T UBE2U UBE2V1 UBE2V2 UBE2W UBE2Z BIRC6 UBE2D2K63E Ref

RNF150 Unknown - ZNRF4 Unknown - CADPS2 Unknown - MUL1 Yes Li et al, 2008 RNF25 Yes Lorick et al, 1999 TOPORS Yes Rajendra et al, 2004 RNF111 Yes Levy et al, 2007 RNF167 Yes Yamada & Gorbsky, 2006 RNF165 Unknown - RNF130 Yes Guais et al, 2006 KF-1 Yes Lorick et al, 1999 ZNF364 Yes Burger et al, 2005 RNF43 Yes Sugiura et al, 2008 RNF13 Unknown - PJA2 Yes Yu et al, 2005 MAP3K1 Yes Lu et al, 2002 LOC51255 Yes Brophy et al, 2008 DTX1 Yes Takeyama et al, 2003 1810074P20RIK Unknown - SNURF Yes Tatham et al, 2008 AMFR Yes Fang et al, 2001 RNF26 Unknown - RNF11 Yes Subramaniam et al, 2003; Rual et al, 2005 RMND5B Unknown - ZNRF1 Yes Araki & Millbrandt, 2003

Figure 2.5. Quality of the E2-E3 network Physical E2-E3 interactions in relation to literature-curated, functional E2-E3 pairs. Hub node E3s were selected and interactions with E2s were scored when biochemically tested in in vitro ubiquitination assays as reported in literature.

For the published literature we recovered 64 E2-E3 pairs for the 25 hub E3s. 35 of the 64 pairs were positive for in vitro or in vivo ubiquitination activity and 29 were tested negative. Of the 35 positive literature interactions we confirmed 33 in our screen, which we regard as true positives ( Figure 2.5). We found only two pairs that were biochemically reported positive, but did not occur in the Y2H screen (false negatives). The observed literature interactions were mostly concerning the members of the UbcH5A-D(UBE2D1-4)-family and UbcH6(UBE2E1) and UbcH9(UBE2E3). To investigate the possibility that these E2s are more versatile in ubiquitination reactions, we compared reported E2-E3 combinations that were

41 negative in biochemical reactions. Of the 29 E2-E3 interactions reported negative we recovered 24 in our dataset (true negatives). We also found five pairs that were reported negative in ubiquitination assays, but showed an interaction in the Y2H screen (false positives). In conclusion, this indicates a specificity of 83% (24/(5+24)) and a sensitivity of 94% (33/(2+33)) between biochemically active E2-E3 pairs and pairs found interacting in our screen. In the above analysis we focused on the top 10% of interactors. Analysis of sensitivity and specificity of the complete data set is confounded by uncertainty of proper folding and/or localization of the Y2H proteins (Supplementary Table VI), which are potential pitfalls of Y2H screening. Therefore, we focused on the E2 and E3s, which display at least one interaction. This criterium retains 20 E2s and 104 E3s from our screen. Literature inspection of this set revealed 118 E2-E3 interactions of which 83 scored positive and 35 negative. Of the 83 positive interactions we recovered 48 interactions in our dataset. Of the 35 negative interactions 30 were also scored in Y2H interaction. This yields a sensitivity of 86% and a specificity of 58%. The verification rate of our screen (true positives plus true negatives/total interactions) is 66%, which compares well to global Y2H screens reported earlier (36, 37). Based on these data, we conclude that the detected E2-interactions for the hub E3s are in agreement with functional E2-E3 pairs reported in the literature, emphasizing the quality of interactions found in our screen.

Validation of E2-E3 interactions by in vitro binding Interestingly, we observed a high number of E3 interactions with an uncharacterized E2 annotated as UBE2U. The UBE2U protein (321 amino acids) is composed of a typical UBC- domain, with a C-terminal tail attached. No biological function has been described for this E2 yet, but microarray experiments revealed that UBE2U mRNA transcripts have been identified mainly in tissues belonging to the urogenital tract. The UBE2U gene is only present in mammalian genomes. We looked in detail, which E3-enzymes were able to interact with this E2. One of the interacting E3s was MDM2. MDM2 has been identified as a major regulator of the cellular level and activity of the transcriptional regulator p53 (38). The significance of MDM2 in controlling normal cellular behavior has become clear by the observation that in more than 50% of human tumors alterations in the p53 has been observed and in at least 7% of these cases, the MDM2 protein is affected (39). MDM2 has been identified as an E3 ligase, capable of ubiquitinating both itself and p53, targeting them for proteasomal degradation (40). For this action, MDM2 has a C-terminal Cys3-His2-Cys3 RING-finger, which was previously shown to be able to interact with UbcH5B(UBE2D2) (41). To investigate if other E2s identified in the two-hybrid screen can bind to MDM2, we focused on the MDM2-UBE2U interaction that appeared quite strong from the Y2H. In order to investigate the MDM2- UBE2U interaction in more detail, we performed GST pull-down assays with immobilized wild-type (WT) and catalytic mutant (C85A or C89A) GST-E2 UBC domains. As a negative control, GST-UbcH6(UBE2E1) was included. As shown in Figure 2.6A, MDM2 is captured by UBE2U and as expected also by UbcH5B(UBE2D2). Interestingly, UBE2U binds more efficiently to MDM2 than UbcH5B(UBE2D2). No binding could be observed with GST alone or GST-

42 UbcH6(UBE2E1), reflecting the specific nature of MDM2-E2 interactions. As expected, mutation of the catalytic cysteine did not affect the interactions. A Next, the GST-E2 enzymes were applied to lysates of human osteosarcoma cells (U2OS), which contain higher levels of MDM2 due to amplification or over-expression (42). Input Input GST GST-UbcH5B C85AGST-UbcH5B GST-UBE2U C89AGST-UBE2U GST-UbcH6 Figure 2.6B shows an efficient 80 pull-down of endogenous MDM2 with both GST-UBE2U 58 and GST-UBE2U C89A enzymes. Binding could also be observed when GST- B UbcH5B(UBE2D2) WT and C85 where used, but no bind- Input lysate A ing could be seen with GST alone or GST-UbcH6(UBE2E1).

10 % 10 % 5 % 2.5 % 1.25 % 0.625 The GST-E2 bound MDM2 frac- GST GST-UbcH5B C85AGST-UbcH5B GST-UBE2U C89AGST-UBE2U GST-UbcH6 tions were also tested by antibodies against p53. Figure 2.6B 175 shows that p53 could be de- 80 * MDM2 tected when MDM2 was captured on GST-E2 enzymes. 58 These results indicate that UBE2U can bind to MDM2, 58 confirming the Y2H results, and that the p53 substrate is part 46 of the captured complex. In order to validate the Y2H Figure 2.6. UBE2U physically interacts with MDM2 (A) HEK293T interactions for other E2 pro- cells were transiently transfected with myc-tagged MDM2, lysed teins we employed the GST and incubated with GST-E2s immobilized on gluthathion-agarose beads. Bound material was resolved on 7.5% SDS-PAGE and pro- pull-down approach for several teins were visualized after immunoblotting with antibodies against E3 nodes expressed in crude myc. (B) Untransfected U2OS cells were lysed and combined with bacterial lysates as 6xhistidine- GST-E2s as in Figure 2.6A. Immunoblotting was done using MDM2 tagged proteins. Sixty binary antibodies (upper panel) and using p53 antibodies after reprobing interactions were tested, (lower panel). Arrows indicate MDM2 signal, asterisks indicate which cover the spectru of background band. m different distributions between

43 hub and non-hub proteins. Figure 2.7 shows that 50 interactions were in agreement with the Y2H analysis: 23 interactions were positive in both assays (true positives) and 27 scored negative in both assays. Of the other ten interactions nine were only positive in the GST pull- down approach, while one interaction (PJA2-UBE2H) only scored positive in the Y2H screen. Comparison of the data sets indicates 96% specificity (TN/(FP+TN)=27/(1+27)) and 72% sensitivity (TP/(TP+FN)=23/(23+9)) (Figure 2.5) and with a verification rate of 83%, which is in agreement with previously reported verification rates of global Y2H screens (36, 37). To- gether, this underscores the high quality of our E2-E3 interaction dataset.

10% Input 10% GST GST-UBE2U GST-UBE2D2 GST-UBE2D3 GST-UBE2N GST-UBE2E1 GST-UBE2E3 GST-UBE2H GST-UBE2C GST-UBE2B GST-UBE2I GST-PD + Y2H + KF1 GST-PD - Y2H - GST-PD + Y2H - GST-PD - Y2H + SNURF

PJA2

CNOT4

*MDM2

UBE4B

CBB

Figure 2.7. GST pull-down analysis between (hub) E2 and E3 enzymes To validate yeast two-hybrid interactions, GST pull-down analysis was performed between ten E2 enzymes and six E3 ligases. Approximately 10 g GST-E2 enzyme were immobilized on G/A-beads and incubated with crude lysates expressing 6xhis-RING constructs. Bound proteins were resolved on SDS-PAGE, immunoblotted and proteins were visualized using anti- 6xhis antibodies. GST-E2 levels were visualized using Coomassie Brilliant Blue staining. An asterisk indicates aspecific signals.

44 Altering the E3-interaction specificity of the K63-specific Ubc13(UBE2N) Elucidation of the 3D structures of several E2-E3 complexes revealed that the E3 interaction surface of E2s is comprised of the N-terminal helix 1, loop 1 between b-strands 3 and 4 and loop 2 between helices 3 10 and 2 (Figure 2.8A) (16-18, 20, 43). We noted that the E3- interaction patterns of UbcH5B(UBE2D2) and of Ubc13(UBE2N) display a ~60% overlap ( Fig- ure 2.8D). This E2 pair is particularly interesting as the primary sequences are rather divergent and UbcH5B(UBE2D2) directs polyubiquitin conjugation via mixed lysine linkages (44), whereas Ubc13(UBE2N) forms K63-linked ubiquitin chains as a heterodimer with the catalytically inactive E2s, UEV-1A(UBE2V1) and MMS2(UBE2V2) (45, 46). To obtain more insight into the E3 interaction specificity of Ubc13(UBE2N) we decided to mutate residues in helix 1, loop 1 or loop 2. We reasoned that mutation of such residues into their UbcH5B(UBE2D2) counterparts could allow interactions, which would otherwise be specific to UbcH5B(UBE2D2). Comparing residues between UbcH5B(UBE2D2) and Ubc13(UBE2N) pointed to P5 and I9 in helix 1, E60 of loop 1 and Q100 in loop 2 as specificity candidates in Ubc13(UBE2N) (Figure 2.8A). We changed these positions into the UbcH5B(UBE2D2) residues and re-arrayed our library of 250 B42-E3(RING) clones to analyze the E3-interaction profiles of the Ubc13(UBE2N) mutants. First, we noted that loop 1 and loop 2 of Ubc13(UBE2N) are particularly sensitive to mutation as the E60T and Q100T mutants fail to interact with any E3 despite normal expression levels (Figure 2.8B & D). Interes- tingly, the E3 interaction pattern of P5L/I9H differs from both Ubc13(UBE2N) and UbcH5B(UBE2D2) (Figure 2.8B & C). This mutant still shares 13/29 interactions with wild type Ubc13(UBE2N) and gains seven new E3 interactions. Of these, five E3s (RMND5B, 1810074P20RIK, RNF150, MAP3K1 and MUL1) are shared with UbcH5B(UBE2D2) and two (RNF183 and CHFR) are unique to the P5L/I9H mutant (Figure 2.8C & Supplementary Table III). MAP3K1 and MUL1 (or MULAN) are particularly interesting as MAP3K1 has been identified as the E3 enzyme responsible for degradation of the ERK1/2 kinases involved in signal transduction pathways (47) and MUL1 has a role in mitochondrial dynamics (26). The P5L/I9H-specific interactor CHFR has been identified as the E3 ligase responsible for degradation of HDAC1, leading to upregulation of the Cdk inhibitor p21 and the metastasis mark- ers KAI1 and E-cadherin (48). These observations indicate that exogenous expression of P5L/I9H Ubc13(UBE2N) may result in switching from K48- or mixed linkage-chains to K63- linked polyubiquitin chains conjugated to substrates of the above-mentioned E3 ligases. In this respect it is important to note that the Ubc13(UBE2N) surface responsible for heterodimerization with UEV-1A(UBE2V1) and MMS2(UBE2V2) does not overlap with the E3 interaction surface comprised of helix 1, loop 1 and loop 2 (45, 46).

In conclusion, our mutational approach exploits the high-quality E2-E3 interaction dataset to design E2 enzymes with new E3-interaction specificities, which can be applied to alter ubiquitination pathways to manipulate protein function in cells.

H1 ß3 L1 ß4 310 L2 H2

* * * *

B UBE2D2 UBE2N UBE2N P5L I9H UBE2N E60T UBE2N Q100T

UBE2D2 UBE2N UBE2N P5L I9H UBE2N E60T UBE2N Q100T

D E

UBE2D2 5 13 UBE2D2 UBE2N I9H P5L UBE2N E60T UBE2N Q100T UBE2N 46 kDa UBE2N LexA 12 30 kDa 11 5 1 58 kDa UBE2N TUB1 2 46 kDa P5L I9H

Figure 2.8. Design of a Ubc13(UBE2N) derivative with a new E3 interaction specificity (A) Sequence alignment of UbcH5B(UBE2D2) and Ubc13(UBE2N). Secondary structure elements (indicated on top in purple and green for the UBE2D2-specific extension of helix 1) and regions involved in E3-RING interactions were derived from the E2 crystal structures (PDB codes 2ESK and 1JBB for UbcH5B(UBE2D2) and Ubc13(UBE2N), respectively). Amino acids are colored according to their physicochemical properties. (B) E3 interaction patterns of UbcH5B(UBE2D2), Ubc13(UBE2N) and Ubc13(UBE2N) mutants. The library of 250 yeast clones expressing B42- RING fusions were re-arrayed on three 96-well plates and mated with LexA-E2 expressing yeast clones. Diploids were selected and interactions were visualized as described in Figure 1. Pictures were taken after 72 hours incubation at 30˚C. Upper and lower three squares in the outer two columns of each plate represent plate controls. (C) Quantification of wild-type and mutant E2-E3 interactions. Spot staining (LacZ) and spot sizes (LEU2) of each E2-E3 interaction were quantified and corrected for background signals. Individual E2-E3 interactions are depicted as bars; black bars indicate no interaction and colored bars indicate E2-E3 interactions. (D) Venn- diagram showing the overlap in E3 interactions between the different E2 enzymes. (E) Expression of wild type and mutant E2 enzymes. Total cell lysates of yeast cells expressing the indicated LexA-E2 fusions were separated on 12.5% SDS-PAGE and analyzed by immunoblotting using antibodies against LexA or tubulin (TUB1) as loading control.

46 Discussion The selectivity of interactions between E2 enzymes and RING-type E3 ligases represent a central and crucial part of the ubiquitin-conjugation pathways in organisms. In order to explore the genome-wide landscape of E2-E3 complexes, we performed a comprehensive Y2H screen to study the selective behavior of these binary interactions. Specific interactions mediated by the E2 UBC-fold and the E3 RING-finger domain were tested between arrayed collections of 35 UBC-folds and 250 RING-finger domains. RING-finger E3 ligases are binding ubiquitin-loaded E2s on their RING domain on one end and substrates on the other acting like a scaffold (12). The few structures known for E2-E3 complexes indicate that the major determinants of E2-E3 binding and selectivity reside within these domains. Our screen focused on the molecular determinants of E2-E3 selectivity as the primary characteristic of specificity that does not take the secondary parameters like spatial and temporal interactions into account.

Interactions between human E2 and E3 enzymes Our database searches retrieved 35 human E2 enzymes that contain the highly conserved UBC-fold. Besides well-known E2 enzymes like UbcH10(UBE2C) (9), members of the UbcH5A- D(UBE2D1-4) subfamily (28) and Ubc13(UBE2N) (14) all having clear roles in ubiquitination, more divergent E2 enzymes have been identified. Among these E2s are NCE2(UBE2F), UBE2W, apollon(BIRC6), E2-230K(UBE2O) and UEV-1A(UBE2V1) and MMS2(UBE2V2). All functional E2s contain the UBC-fold, providing them with a molecular platform for RING- domain interaction and rendering them as catalysts in the process of ubiquitin (-like) conjugation (6). We also included the UbcH5B(UBE2D2) K63E mutant in the E2 arrays, evaluating the effects of this E2-E3 interaction mutant. The selection of RING-type E3 domains included in our screen covers the majority of the human E3 enzymes. Our results revealed that multiple E2s are able to interact with more than one E3 and vice versa. E2 enzymes that interact with many E3s were UBE2U, members of the UbcH5A- D(UBE2D1-4) family and Ubc13(UBE2N). UBE2U showed the highest levels of E3 interactions. The restricted expression pattern of UBE2U could explain why it has not been identified earlier as an E3(RING)-interactor. E2s like UEV-1A(UBE2V1) and MMS2(UBE2V2) did not show E3 interactions. Although they contain an UBC-fold that could potentially interact with E3s, they lack the catalytic cysteine residue required for ubiquitin acceptance (14). Some E2s are involved in the conjugation of ubiquitin-like proteins, such as Ubc12(UBE2M) and NCE2(UBE2F) for NEDD8 and Ubc9(UBE2I) for SUMO (2, 23, 49). Also E2-230K(UBE2O) and apollon(BIRC6) did not interact with E3s. Both apollon(BIRC6) and E2-230K(UBE2O) are thought to contain a chimeric E2/E3 domain, which would obviate the need for an exogenous E3 ligase (50). Other E2s failed to interact with E3s in our screen, like the human homologs of the yeast Rad6p, hHR6A(UBE2A) and hHR6B(UBE2B). A possible explanation is that the Rad6p is phosphorylated at serine 120 by CDK-1 and -2 (51) and this modification increases the E2 activity. Residue S120 is conserved in Rad6p homologs, but it is unclear if the modification increases the binding between the E2 and the E3 or whether it influences the

47 activity. Another E2, UbcH10(UBE2C) is required for degradation of mitotic cyclins and cell cycle progression. Recently, the N-terminal extension of UbcH10(UBE2C), which is missing in our construct, has been implicated in regulating substrate ubiquitination and the number of lysines ubiquitinated in these substrates via interaction with the APC E3 ligase (9). Some E3 enzymes were more connected than others. Strikingly, CADPS2 showed the highest number of interactions, but it has not been annotated as an E3 enzyme yet. This observation could indicate the identification of a new or deviant domain that could interact with E2 enzymes and potentially be involved in ubiquitination. Within the E3s with the highest number of E2 interactions, members of two closely related proteins families were found. The RNF130, RNF167 and RNF150 proteins belong to the Goliath family and are involved in apoptosis and embryonic development (31, 32, 52). These E3 showed a pattern of overlapping E2 interactions. Also ZNRF1, ZNRF2 and ZNRF4 were found among the highest interactors. The ZNRF proteins shared many interactions with the same E2s. These two protein families indicate that conserved E3s are clustered together to interact with the same panel of many E2s.

Hetero-dimeric complexes, like BRCA1-BARD1, MDM2-MDMX and BMI1-Ring1B, represent a special group of E3 enzymes. In those cases, one RING provides the complex with E3 activity, but only in association with the partnering RING. The active RING serves as a binding subunit for E2 enzymes, but the function of the other RING is often unclear. This situation is not represented in our two-hybrid set-up. In a recent yeast two-hybrid study, the BRCA1-BARD1 RING domains were fused and tested for E2 interactions (18). In this set-up, the E2 proteins UbcH5A(UBE2D1), UbcH5B(UBE2D2), UbcH5C(UBE2D3), UbcH6(UBE2E1), UbcH8(UBE2E2), UbcH9(UBE2E3), Ubc9(UBE2I), HIP2(UBE2K), UbcH7(UBE2L3), Ubc13(UBE2N) and UBE2W scored positive. In contrast, we only observe interactions with NCUBE2(UBE2J2), UBE2U and UBE2W for the isolated BRCA1 RING domain in the LacZ read-out. This clearly indicates that screening with fused RING dimers yields different results from screening with isolated RING domains. It remains, however, unclear what the underlying molecular determinants of these differences are and a potential role for the inactive dimerization partners needs to be established.

We have previously demonstrated that the CNOT4 D48K E49K mutant has lost its interactions with UbcH5A-D(UBE2D1-4), UbcH6(UBE2E1) and UbcH9(UBE2E3) (21). We show here that the RING-finger of CNOT4 is able to interact with UBE2W and that the mutant also loses this interaction. These observations indicate that the residues within UbcH5A-D(UBE2D1-4), UbcH6(UBE2E1), UbcH9(UBE2E3) and UBE2W required for RING interactions are coupled with respect to their involvement in determining RING specificity. The CNOT4 double mutant however gained a stronger signal with Ubc13(UBE2N) when compared to wild-type CNOT4. From this point of view, the RING-interacting residues in Ubc13(UBE2N) are likely to mediate binding in an opposite fashion, gaining a stronger RING association. Unexpectedly, residues at position 63 are not highly conserved between the CNOT4-interacting E2 domains, indicating a more complex recognition mode. The mutant RING-finger displays a specific interaction

48 with the UbcH5B(UBE2D2) K63E mutant, but not with the wild-type UbcH5B(UBE2D2), altering the specificity of interaction. When testing the UbcH5B(UBE2D2) K63E mutant, it gained an interaction with NFX1. Also the change in E3 interaction pattern of UbcH5B(UBE2D2) indicates that, although the E2-E3 interface is conserved, individual residues in this interface have different contributions in achieving interaction specificity among different pairs of E2s and E3s. Molecular modeling and docking approaches of E2-E3 pairs may provide more insight in the specific nature of these interactions. It is generally regarded that the UBC-fold of the E2 and the RING-finger domain of the E3s represent the interaction surfaces for E2- E3(RING) interactions (2, 53). However, in some cases residues outside these domains contribute as exemplified by the UbcH7(UBE2L3)-c-Cbl structure. The RING domain of c-Cbl packs onto the tyrosine kinase binding (TKB) domain and a connecting linker harbors several E2-interacting residues (16). The absence of the TKB domain and linker can explain why no c- Cbl interactions were recovered in our interaction screen. Another example is the Hsp70 interacting protein (CHIP), which exclusively interacts with the UBC-fold of UbcH5A(UBE2D1) via its RING-type U-box domain (17). No CHIP interactions were recovered in our screen. The false negatives of our screen as exemplified by c-Cbl and CHIP can be related to the absence of interaction residues outside the E3(RING) domain, misfolding and/or mislocalization of the E3s in our screen. Nevertheless, we describe 295 novel E2-E3 interactions and we identify 66 E3(RING) proteins as novel E2 interactors. Together, this represents an important step forward in our knowledge of the E2 and E3 proteins of the human ubiquitin system.

In a first attempt to understand E3-specificities in directing different types of ubiquitin- linkages we focused on the UbcH5B(UBE2D2)-Ubc13(UBE2N) E2 pair. Substitution of residues in the E3-interaction surfaces of the K63-specific Ubc13(UBE2N) yielded the P5L/I9H mutant (Figure 2.8). This mutant in the N-terminal helix of Ubc13(UBE2N) gained five E3 interactions shared with UbcH5B(UBE2D2) (specific for mixed-linkage chains and K48-linkages directing proteasomal degradation) and two unique interactions. Ubc13(UBE2N) is catalytically active as a heterodimeric E2 with UEV1-A(UBE2V1) or MMS2(UBE2V2). Importantly, the crystal structure of yeast Ubc13(UBE2N)/MMS2(UBE2V2) indicates that the P5L/I9H mutations will not interfere with heterodimer formation of Ubc13(UBE2N) (45, 46). Thus, the P5L/I9H mutant of Ubc13(UBE2N) may redirect K63- to K48/mixed-linked poly-ubiquitin chains on a selected set of UbcH5B(UBE2D2) substrates upon expression in cells. We expect that the Ubc13(UBE2N)-UbcH5B(UBE2D2) pair represents only the first example and that the E2-E3 interaction data harbor many more possibilities for redesigning ubiquitin pathways.

Topology of the E2-E3 network It is clear that E2-E3 interactions are not following a normal distribution. Both E2s and E3s showed a small number of highly connected proteins, followed by a larger group of enzymes with lesser interactions and so on. The identified highly interacting E2 and E3 proteins share many structural features within their complete UBC-fold and the E3-interacting interface residues. These high levels of sequence identity allow these hubs to interact with certain clus-

49 ters of RINGs. For some of these hubs, their biological function is expected to involve redundancy. Evolutionary duplication of, for example members of the UBE2D(UbcH5)-subgroup, allows these E2s to interact with the same set of RING-fingers and, possibly, to ubiquitinate the same substrates. E2 enzymes that are more divergent in their interface sequence, like Ubc13(UBE2N), have a different E3-interaction profile and are interacting with a different subcluster of RING E3s. Proteins with many connections are holding large sub-networks together, forming complete, interconnected networks (54). Deletion of these nodes has been shown to be associated with network robustness, pointing out that many hub proteins are essential for remaining the integrity of the network and the complete system (54). In the E2-E3 network, UBE2U, members of the UbcH5A-B(UBE2D1-4) family, UbcH6(UBE2E1) and UbcH9(UBE2E3), Ubc13(UBE2N) and UBC7(UBE2G2) behave as hub proteins. Inactivation of these E2 proteins could potentially have dramatic contributions to large regions of the E2-E3 network and to cellular ubiquitination. Since only a few substrates have been identified yet, we evaluated the in silico effect of deletion of hub E2s, by counting the numbers of E2s that become unconnected. Deletion of UBE2U, UBC7(UBE2G2) or Ubc13(UBE2N) has dramatic effects on the number of unconnected E3s, indicating that these hubs are indeed holding the E2-E3 network together. Strikingly, two of these three hubs have specific biological roles, Ubc13(UBE2N) is involved in K63-chain synthesis and UBC7(UBE2G2) is an ER-resident E2 involved in ERAD. In contrast, deletion of any of the UbcH5A-D(UBE2D1-4) or UbcH6(UBE2E1) has little effect on the network; all E3s that remain connected are already bound to a closely related E2 enzyme. The observation that members of the UbcH5A- D(UBE2D1-4) family are biologically implicated in the degradation of abnormally-folded and short-lived proteins could be related to the expansion of the UbcH5A-D(UBE2D1-4) family. Indeed, both the orthologs UBC4 and UBC5 in Saccharomyces cerevisiae are functionally re- dundant and deletion impairs growth, temperature tolerance, protein misfolding and stress responses (28). Interestingly, deletion of E2s known to conjugate UBLs, like Ubc9(UBE2I) (for SUMO) and Ubc12(UBE2M) and NCE2(UBE2F) (for NEDD8), does not give rise to any unconnected E3, indicating that these E2s are less important for maintaining the integrity of the ubiquitin network.

Physical E2-E3 interactions form the basis of ubiquitination activity and are required for an E2-E3 pair to become functional. By comparing the Y2H-data with E2-E3 pairs that are reported to be functional in ubiquitination, we showed that the physical interactions among hub proteins for the majority of E2-E3 pairs correlate very well with reported biochemically active E2-E3 pairs. In addition, comparisons of all detected Y2H interactions or those that are having only one or more interactions revealed good agreements with literature-curated E2- E3 pairs. These comparisons indicate that the ability of an E2-E3 pair to be functional in ubiquitination reactions is based on determinants of physical interaction. Also, physical interactions could be used to predict the functionality of an E2-E3 interaction. The independent validation of a subset of E2-E3 in GST pull-down binding experiments provide a good coverage

50 of the found yeast two-hybrid interactions and, combined with the above-mentioned, indicate a high quality of the E2-E3 interactions determined by yeast two-hybrid.

The E2-E3 network can be fitted into the overall organization of the ubiquitination process. The two known ubiquitin E1s are activating ubiquitin and interact with the E2s to transfer ubiquitin in a strictly hierarchical organization (3, 4). Then the E2-E3 interactions take place in their specific manner and finally the ubiquitin is transferred to the E3-bound substrate (2). Although no global information about E3-substrate specificity is available, some well- characterized E3s have been shown to mediate ubiquitination of multiple substrates (55-57). This stresses the need for genome-wide analysis of interactions between E3 enzymes and their substrates. But this may be hampered by the fact that for many E3s and substrates the binding sites are poorly characterized, a small amount of conservation among these sites is observed and the potential need for post-translational modifications prior to E3 recognition of the substrate. Besides this, the E1-E2-E3-substrate ubiquitination super-network can be regarded as a system that combines both hierarchical and the highly interconnected properties of E2-E3 interactions. Taken all together, we report a high-quality framework centered on the functional domains of E2 and E3 enzymes in the human ubiquitination system. Exhaustive screening of E2-E3 interactions provides insight in the arrangement of proteins in this complex biological system. These binary interactions can be used to initiate follow-up studies concerning the structural aspects of E2-E3 pairs, such as modeling and docking of E2-E3 interactions in order to gain more insight in the underlying specificity. Although not all of these interactions will occur physiologically, for example because there are barriers in space and time preventing the coming together of E2s and E3s, these results provide a highly detailed, global network of E2-E3 interactions and their role in ubiquitination as a whole system. The interactions obtained with the mutant E2 could initiate new lines of synthetic ubiquitin research aimed at designed modulation of the system. Together, these experimental interactions provide a rich source of information and can be used to guide future research in ubiquitination and E2-E3 selectivity.

Methods Plasmid procedures Full-length or EST cDNAs of human ubiquitin-conjugating enzymes and human RING-type E3 ubiquitin protein ligases were obtained from the Deutsches Ressourcenzentrum fur Genom- forschung GmbH (RZPD, Berlin, Germany). PCR primers (Biolegio, Nijmegen, The Nether- lands) were used to amplify the E2 UBC-fold and the E3 RING-finger catalytic domains (including 10 N-terminal and 10 C-terminal residues) and these domains were sub-cloned either into pEG202-NLS or in the pJG4-5 plasmid to generate E2 or RING constructs fused to the C-terminus of the E. coli LexA (DNA binding domain; DBD) and B42 (activation domain; AD) proteins, respectively. cDNAs encoding E2 enzymes were sub-cloned in pGEX-4T-1 to generate GST-fusions. Mutagenesis was performed using QuickChange II Site-Directed muta-

51 genesis (Stratagene), according to the manufacturer’s instructions. The mammalian expression plasmid myc-Mdm2 has been described previously (58). All constructs were completely verified by DNA sequencing. Additional information concerning cDNAs, amplification primers and cloning procedures is available upon request.

Antibodies Monoclonal antibodies recognizing LexA (2-12), GST (B14), MDM2 (SMP-14) and p53 (DO1) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA) and an HRP-coupled anti- penta-his antibody was purchased from Qiagen (Qiagen, Venlo, The Netherlands). Monoc- lonal antibodies against HA (12CA5) and CNOT4 (19A20) were purified from hybridomas as reported elsewhere (21).

Yeast manipulation and array generation Growth and manipulation of the parental yeast strain EGY48 (MAT alpha, his3, trp1, ura3, 6LexAop-LEU2) was done as described before (21). Mating type changing was carried out by using HO-endonuclease as described earlier (59). To generate a standardized array of yeast cells expressing all LexA-E2 fusions, three individual verified DB-E2 bacterial shuffle plasmids were co-transformed together with the LacZ reporter plasmid (LexAop)8 pSH18-34 in EGY48alpha, while AD-RING fusions were transformed in EGY48a by using a LiOAc/DMSO transformation in 96-well plates (60). Transformants were selected on solid synthetic complete (SC) medium without histidine (H) and uracil (U) in the case of EGY48alpha (LexA- UBC/pSH18-34) or without tryptophan (W) for EGY48a (B42-RING) by incubation at 30oC for 2-3 days. Single colonies were resuspended in sterile 40% glycerol and stored at -80oC. Cor- rect transformation with the appropriate fusion protein was verified using colony PCR as described before (59). In the E2 array, transformants of the EGY48alpha strains were arranged in a square of four spots, LexA-E2s were spotted in three independent biological transformants, combined with a transformant expressing LexA alone.

Semi high-throughput yeast two-hybrid analysis Individual EGY48a transformants bearing B42-RING fusions were inoculated in liquid SC W- medium, grown overnight at 30oC shaken at 230 rpm. Liquid cultures were transferred on solid SC W- by manual pinning in 384-format using manual pinning tools (VP384F), library copiers (VP381) and colony copiers (VP380) from V & P Scientific, Inc. (San Diego, CA, USA) and grown overnight to increase the amount of cells used for mating. In parallel, LexA-E2- transformed EGY48alpha cells were pinned from the frozen master stock plates on solid SC HU-. Mating was performed by pinning the LexA-E2 and the B42-RING transformants at exactly the same position on top of each other on solid, non-selective YPD medium for 24 hours at 30oC. Diploids were selected at 30oC for 48 hours by selecting mated cells on SC HWU- plates. To visualize putative E2-E3 interactions, pre-selected diploids were transferred to read-out plates. For colorimetric selection, diploids were pinned on SC HWU- plates supplemented with X-Gal, and for auxotrophic selection, diploids were transferred to SC HWUL-

52 plates. Both types of read-out were assessed under both inducing conditions, with galactose as the sole carbon source as well as under repressing conditions (with glucose as carbon source). Plates were incubated for a period of 24-96 hours and pictures of the yeast colonies were taken at fixed intervals.

Spot quantification analysis Digital images of yeast plates to quantify yeast spot size and blue staining intensity were carried out systematically at fixed intervals by using a standardized digital camera set-up. Spot size and blue staining intensity were quantified using a previously reported Java-based spot- size measuring software algorithm (61), according to the developer’s instructions and using an in-house written grid-based blue intensity measure algorithm, respectively.

Western blot verification of protein expression To confirm correct protein expression, total protein lysates were prepared (62). Briefly, individual, freshly pinned yeast colonies were grown in appropriate medium, overnight in 96- well plates at 30oC under continuously shaking. Cells were centrifuged, resuspended in 0.1 M NaOH and incubated for 5 minutes at room temperature. Next, cells were centrifuged and resuspended in 1x sample buffer, boiled for 5 minutes and centrifuged. Standard, proteins were separated on a 17.5 % SDS-PAGE gel, transferred to nitrocellulose membranes and probed with mouse anti-HA or anti-LexA antibodies and visualization was done with ECL according to the supplier’s instructions (Pierce).

Recombinant proteins Full-length or regions encoding the UBC-fold of individual E2s were subcloned in pGEX-4T-1 and expressed as GST-E2 fusion proteins in E. coli BL21(DE3) as described previously (21). Briefly, freshly transformed colonies were grown to mid-log phase in LB-medium, induced with 0.4 mM isopropyl-D-thiogalactopyranoside (IPTG) for 3 h at 30°C, and lysed in 25 ml ice- cold lysis buffer (300 mM KCl, 50 mM Tris–HCl pH 8.0, 2 mM EDTA 0.1% Triton X-100, 20% sucrose) containing 1 mM dithiothreitol (DTT), 0.5 mM phenylmethylsulfonyl fluoride (PMSF), protease inhibitor cocktail (Roche) and 25 mg/ml lysozyme. After freeze–thawing and sonification, lysates were centrifuged at 50 000 r.p.m. for 1 h at 4°C. Crude lysates were stored at -80oC. Constructs expressing recombinant human his-thioredoxin-RING fusions in the pLICHIS vector backbone were expressed in E. coli BL21(DE3) and protein expression except that IPTG induction was done overnight at 24oC in LB-medium supplemented with 100

M ZnCl2.

Mammalian cell culture Human embryonic kidney cells (HEK293T) and human osteosarcoma cells (U2OS) were maintained in DMEM supplemented with glutamine, penicillin/streptomycin and 10% FBS under o 5% CO2 at 37 C. 293T cells were grown to 60-70% confluency and transfections were done using FuGENE6 according to the suppliers’ instructions with myc-tagged MDM2. 24 hours

53 after transfection, cells were washed twice with ice-cold PBS, lysed in RIPA (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1.0% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM DTT, 0.5 mM PMSF and 1 µg/ml aprotonin, leupeptin and pepstatin), cleared by centrifugation at 14 krpm at 4oC. U2OS cells were treated with or without 20 M MG132 for 4 hours prior to lysis in RIPA-buffer.

GST pull-down assays GST-E2 proteins were immobilized on gluthathione-agarose beads (Sigma) and incubated with 5 g of His6-CNOT4-N78/His6-CNOT4-N227 or U2OS or 293T lysates for 2 hours at 4oC in G/A-buffer (50 mM potassium phosphate pH 6.6, 50 mM KCl, 0.1% NP-40, 10 M ZnCl2, 0.5 mM PMSF, 1 mM DTT, 1 g/ml aprotonin, leupeptin and pepstatin (All Sigma)). GST-E2- RING pull-down analysis was performed by incubation of GST-E2 proteins on beads with his- RING bacterial lysate (approximately 5 g of RING-fusion protein) for three hours at 4oC. Beads were washed three times with G/A-buffer, bound proteins resolved by SDS-PAGE and visualized with immunoblotting.

Acknowledgements The authors would like to thank Adrien S. L. Melquiond and members of the Timmers laboratory for helpful discussions. This work was funded by support from the Netherlands Proteo- mic Center (NPC) and the Netherlands Society for Scientific Research (700.50.034).

References 1. Pickart, C. M. (2001) Mechanisms underlying ubiquitination. Annu Rev Biochem 70, 503-533 2. Dye, B. T., and Schulman, B. A. (2007) Structural mechanisms underlying posttransla- tional modification by ubiquitin-like proteins. Annu Rev Biophys Biomol Struct 36, 131-150 3. Haas, A. L., Warms, J. V., Hershko, A., and Rose, I. A. (1982) Ubiquitin-activating enzyme. Mechanism and role in protein-ubiquitin conjugation. J Biol Chem 257, 2543- 2548 4. Jin, J., Li, X., Gygi, S. P., and Harper, J. W. (2007) Dual E1 activation systems for ubiquitin differentially regulate E2 enzyme charging. Nature 447, 1135-1138 5. Pickart, C. M. (2001) Mechanisms underlying ubiquitination. Annu Rev Biochem 70, 503-533 6. Burroughs, A. M., Jaffee, M., Iyer, L. M., and Aravind, L. (2008) Anatomy of the E2 ligase fold: Implications for enzymology and evolution of ubiquitin/Ub-like protein conjugation. J Struct Biol 7. Haldeman, M. T., Xia, G., Kasperek, E. M., and Pickart, C. M. (1997) Structure and function of ubiquitin conjugating enzyme E2-25K: the tail is a core-dependent activity element. Biochemistry 36, 10526-10537 8. Silver, E. T., Gwozd, T. J., Ptak, C., Goebl, M., and Ellison, M. J. (1992) A chimeric ubiquitin conjugating enzyme that combines the cell cycle properties of CDC34 (UBC3) and the DNA repair properties of RAD6 (UBC2): implications for the structure, function and evolution of the E2s. EMBO J 11, 3091-3098

54 9. Summers, M. K., Pan, B., Mukhyala, K., and Jackson, P. K. (2008) The unique N terminus of the UbcH10 E2 enzyme controls the threshold for APC activation and enhances checkpoint regulation of the APC. Mol Cell 31, 544-556 10. Bernassola, F., Karin, M., Ciechanover, A., and Melino, G. (2008) The HECT family of E3 ubiquitin ligases: multiple players in cancer development. Cancer Cell 14, 10-21 11. Hatakeyama, S., and Nakayama, K. I. (2003) U-box proteins as a new family of ubiquitin ligases. Biochem Biophys Res Commun 302, 635-645 12. Ardley, H. C., and Robinson, P. A. (2005) E3 ubiquitin ligases. Essays Biochem. 41, 15- 30 13. Finley, D., Sadis, S., Monia, B. P., Boucher, P., Ecker, D. J., Crooke, S. T., and Chau, V. (1994) Inhibition of proteolysis and cell cycle progression in a multiubiquitination- deficient yeast mutant. Mol Cell Biol 14, 5501-5509 14. Deng, L., Wang, C., Spencer, E., Yang, L., Braun, A., You, J., Slaughter, C., Pickart, C., and Chen, Z. J. (2000) Activation of the IkappaB kinase complex by TRAF6 requires a dimeric ubiquitin-conjugating enzyme complex and a unique polyubiquitin chain. Cell 103, 351-361 15. Poyurovsky, M. V., Priest, C., Kentsis, A., Borden, K. L., Pan, Z. Q., Pavletich, N., and Prives, C. (2007) The Mdm2 RING domain C-terminus is required for supramolecular assembly and ubiquitin ligase activity. EMBO J. 26, 90-101 16. Zheng, N., Wang, P., Jeffrey, P. D., and Pavletich, N. P. (2000) Structure of a c-Cbl- UbcH7 complex: RING domain function in ubiquitin-protein ligases. Cell 102, 533-539 17. Xu, Z., Kohli, E., Devlin, K. I., Bold, M., Nix, J. C., and Misra, S. (2008) Interactions between the quality control ubiquitin ligase CHIP and ubiquitin conjugating enzymes. BMC Struct Biol 8, 26 18. Christensen, D. E., Brzovic, P. S., and Klevit, R. E. (2007) E2-BRCA1 RING interactions dictate synthesis of mono- or specific polyubiquitin chain linkages. Nat Struct Mol Biol 14, 941-948 19. Schulman, B. A., Carrano, A. C., Jeffrey, P. D., Bowen, Z., Kinnucan, E. R., Finnin, M. S., Elledge, S. J., Harper, J. W., Pagano, M., and Pavletich, N. P. (2000) Insights into SCF ubiquitin ligases from the structure of the Skp1-Skp2 complex. Nature 408, 381-386 20. Huang, L., Kinnucan, E., Wang, G., Beaudenon, S., Howley, P. M., Huibregtse, J. M., and Pavletich, N. P. (1999) Structure of an E6AP-UbcH7 complex: insights into ubiquitination by the E2-E3 enzyme cascade. Science 286, 1321-1326 21. Winkler, G. S., Albert, T. K., Dominguez, C., Legtenberg, Y. I., Boelens, R., and Tim- mers, H. T. (2004) An altered-specificity ubiquitin-conjugating enzyme/ubiquitin- protein ligase pair. J Mol Biol 337, 157-165 22. Lorick, K. L., Jensen, J. P., Fang, S., Ong, A. M., Hatakeyama, S., and Weissman, A. M. (1999) RING fingers mediate ubiquitin-conjugating enzyme (E2)-dependent ubiquitination. Proc Natl Acad Sci USA 96, 11364-11369 23. Huang, D. T., Ayrault, O., Hunt, H. W., Taherbhoy, A. M., Duda, D. M., Scott, D. C., Borg, L. A., Neale, G., Murray, P. J., Roussel, M. F., and Schulman, B. A. (2009) E2- RING expansion of the NEDD8 cascade confers specificity to cullin modification. Mol Cell 33, 483-495 24. Ozkan, E., Yu, H., and Deisenhofer, J. (2005) Mechanistic insight into the allosteric activation of a ubiquitin-conjugating enzyme by RING-type ubiquitin ligases. Proc Natl Acad Sci USA 102, 18890-18895

55 25. Shimura, H., Hattori, N., Kubo, S., Mizuno, Y., Asakawa, S., Minoshima, S., Shimizu, N., Iwai, K., Chiba, T., Tanaka, K., and Suzuki, T. (2000) Familial Parkinson disease gene product, parkin, is a ubiquitin-protein ligase. Nat Genet 25, 302-305 26. Li, W., Bengtson, M. H., Ulbrich, A., Matsuda, A., Reddy, V. A., Orth, A., Chanda, S. K., Batalov, S., and Joazeiro, C. A. (2008) Genome-wide and functional annotation of human E3 ubiquitin ligases identifies MULAN, a mitochondrial E3 that regulates the organelle's dynamics and signaling. PloS ONE 3, e1487 27. Golemis, E. A., Serebriiski, I., Finley Jr., R. L., Kolonin, M. G., Gyuris, J., and Brent, R. (2001) Interaction trap/two-hybrid system to identify interacting proteins. Curr Pro- toc Mol Biol. Chapter 20 28. Seufert, W., and Jentsch, S. (1990) Ubiquitin-conjugating enzymes UBC4 and UBC5 mediate selective degradation of short-lived and abnormal proteins. EMBO J 9, 543- 550 29. Chen, B., Mariano, J., Tsai, Y. C., Chan, A. H., Cohen, M., and Weissman, A. M. (2006) The activity of a human endoplasmic reticulum-associated degradation E3, gp78, requires its Cue domain, RING finger, and an E2-binding site. Proc Natl Acad Sci USA 103, 341-346 30. Maslov, S., and Sneppen, K. (2002) Specificity and stability in topology of protein networks. Science 296, 910-913 31. Anandasabapathy, N., Ford, G. S., Bloom, D., Holness, C., Paragas, V., Seroogy, C., Skrenta, H., Hollenhorst, M., Fathman, C. G., and Soares, L. (2003) GRAIL: an E3 ubiquitin ligase that inhibits cytokine gene transcription is expressed in anergic CD4+ T cells. Immunity 18, 535-547 32. Guais, A., Siegrist, S., Solhonne, B., Jouault, H., Guellaen, G., and Bulle, F. (2006) h- Goliath, paralog of GRAIL, is a new E3 ligase protein, expressed in human leukocytes. Gene 374, 112-120 33. Fujii, T., Tamura, K., Copeland, N. G., Gilbert, D. J., Jenkins, N. A., Yomogida, K., Tana- ka, H., Nishimune, Y., Nojima, H., and Abiko, Y. (1999) Sperizin is a murine RING zinc- finger protein specifically expressed in Haploid germ cells. Genomics 57, 94-101 34. Araki, T., and Milbrandt, J. (2003) ZNRF proteins constitute a family of presynaptic E3 ubiquitin ligases. J Neurosci 23, 9385-9394 35. Sadakata, T., Kakegawa, W., Mizoguchi, A., Washida, M., Katoh-Semba, R., Shutoh, F., Okamoto, T., Nakashima, H., Kimura, K., Tanaka, M., Sekine, Y., Itohara, S., Yuzaki, M., Nagao, S., and Furuichi, T. (2007) Impaired cerebellar development and function in mice lacking CAPS2, a protein involved in neurotrophin release. J Neurosci 27, 2472- 2482 36. Rual, J. F., Venkatesan, K., Hao, T., Hirozane-Kishikawa, T., Dricot, A., Li, N., Berriz, G. F., Gibbons, F. D., Dreze, M., Ayivi-Guedehoussou, N., Klitgord, N., Simon, C., Boxem, M., Milstein, S., Rosenberg, J., Goldberg, D. S., Zhang, L. V., Wong, S. L., Franklin, G., Li, S., Albala, J. S., Lim, J., Fraughton, C., Llamosas, E., Cevik, S., Bex, C., Lamesch, P., Sikorski, R. S., Vandenhaute, J., Zoghbi, H. Y., Smolyar, A., Bosak, S., Sequerra, R., Doucette-Stamm, L., Cusick, M. E., Hill, D. E., Roth, F. P., and Vidal, M. (2005) Towards a proteome-scale map of the human protein-protein interaction network. Nature 437, 1173-1178 37. Li, S., Armstrong, C. M., Bertin, N., Ge, H., Milstein, S., Boxem, M., Vidalain, P. O., Han, J. D., Chesneau, A., Hao, T., Goldberg, D. S., Li, N., Martinez, M., Rual, J. F., La-

56 mesch, P., Xu, L., Tewari, M., Wong, S. L., Zhang, L. V., Berriz, G. F., and al, e. (2004) A map of the interactome network of the metazoan C. elegans. Science 303, 540-543 38. Momand, J., Zambetti, G. P., Olson, D. C., George, D., and Levine, A. J. (1992) The mdm-2 oncogene product forms a complex with the p53 protein and inhibits p53- mediated transactivation. Cell 69, 1237-1245 39. Momand, J., Jung, D., Wilczynski, S., and Niland, J. (1998) The MDM2 gene amplification database. Nucleic Acids Res. 26, 3453-3459 40. Nakamura, S., Roth, J. A., and Mukhopadhyay, T. (2000) Multiple lysine mutations in the C-terminal domain of p53 interfere with MDM2-dependent protein degradation and ubiquitination. Mol Cell Biol 20, 9391-9398 41. Saville, M. K., Sparks, A., Xirodimas, D. P., Wardrop, J., Stevenson, L. F., Bourdon, J. C., Woods, Y. L., and Lane, D. P. (2004) Regulation of p53 by the ubiquitin-conjugating enzymes UbcH5B/C in vivo. J Biol Chem 279, 42169-42181 42. Florenes, V. A., Maelandsmo, G. M., Forus, A., Andreassen, A., Myklebost, O., and Fodstad, O. (1994) MDM2 gene amplification and transcript levels in human sarco- mas: relationship to TP53 gene status. J Natl Cancer Inst 86, 1297-1302 43. Dominguez, C., Bonvin, A. M., Winkler, G. S., van Schaik, F. M., Timmers, H. T., and Boelens, R. (2004) Structural model of the UbcH5B/CNOT4 complex revealed by combining NMR, mutagenesis, and docking approaches. Structure 12, 633-644 44. Brzovic, P. S., Lissounov, A., Christensen, D. E., Hoyt, D. W., and Klevit, R. E. (2006) A UbcH5/ubiquitin noncovalent complex is required for processive BRCA1-directed ubiquitination. Mol Cell 21, 873-880 45. VanDemark, A. P., Hofmann, R. M., Tsui, C., Pickart, C. M., and Wolberger, C. (2001) Molecular insights into polyubiquitin chain assembly: crystal structure of the Mms2/Ubc13 heterodimer. Cell 105, 711-720 46. Hofmann, R. M., and Pickart, C. M. (1999) Noncanonical MMS2-encoded ubiquitin- conjugating enzyme functions in assembly of novel polyubiquitin chains for DNA repair. Cell 96, 645-653 47. Lu, Z., Xu, S., Joazeiro, C., Cobb, M. H., and Hunter, T. (2002) The PHD domain of MEKK1 acts as an E3 ubiquitin ligase and mediates ubiquitination and degradation of ERK1/2. Mol Cell 9, 945-956 48. Oh, Y. M., Kwon, Y. E., Kim, J. M., Bae, S. J., Lee, B. K., Yoo, C. H., Deshaies, R. J., and Seol, J. H. (2009) Chfr is linked to tumour metastasis through the downregulation of HDAC1. Nat Cell Biol 11, 295-302 49. Kerscher, O., Felberbaum, R., and Hochstrasser, M. (2006) Modification of proteins by ubiquitin and ubiquitin-like proteins. Annu Rev Cell Dev Biol 22, 159-180 50. Berleth, E. S., and Pickart, C. M. (1996) Mechanism of ubiquitin conjugating enzyme E2-230K: catalysis involving a thiol relay? Biochemistry 35, 1664-1671 51. Sarcevic, B., Mawson, A., Baker, R. T., and Sutherland, R. L. (2002) Regulation of the ubiquitin-conjugating enzyme hHR6A by CDK-mediated phosphorylation. EMBO J 21, 2009-2018 52. Bouchard, M. L., and Cote, S. (1993) The Drosophila melanogaster developmental gene g1 encodes a variant zinc-finger-motif protein. Gene 125, 205-209 53. Joazeiro, C. A., Wing, S. S., Huang, H., Leverson, J. D., Hunter, T., and Liu, Y. C. (1999) The tyrosine kinase negative regulator c-Cbl as a RING-type, E2-dependent ubiquitin- protein ligase. Science 286, 309-312

57 54. Jeong, H., Mason, S. P., Barabasi, A. L., and Oltvai, Z. N. (2001) Lethality and centrality in protein networks. Nature 411, 41-42 55. Tanaka, K., Suzuki, T., Hattori, N., and Mizuno, Y. (2004) Ubiquitin, proteasome and parkin. Biochim Biophys Acta 1695, 235-247 56. Lin, A. E., and Mak, T. W. (2007) The role of E3 ligases in autoimmunity and the regulation of autoreactive T cells. Curr Opin Immunol 19, 665-673 57. Boulton, S. J. (2006) BRCA1-mediated ubiquitylation. Cell Cycle 5, 1481-1486 58. van der Horst, A., de Vries-Smits, A. M., Brenkman, A. B., van Triest, M. H., van den Broek, N., Colland, F., Maurice, M. M., and Burgering, B. M. (2006) FOXO4 transcriptional activity is regulated by monoubiquitination and USP7/HAUSP. Nat Cell Biol 8, 1064-1073 59. Burke, D., Dawson, D., and Stearns, T. (2000) Methods in Yeast Genetics, a Cold Spring Harbor laboratory course manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor 60. Gietz, R. D., and Woods, R. A. (2002) Transformation of yeast by lithium ace- tate/single-stranded carrier DNA/polyethylene glycol method. Methods Enzym. 350, 87-96 61. Collins, S. R., Schuldiner, M., Krogan, N. J., and Weissman, J. S. (2006) A strategy for extracting and analyzing large-scale quantitative epistatic interaction data. Genome Biol. 7, R63 62. Kushnirov, V. V. (2000) Rapid and reliable protein extraction from yeast. Yeast 16, 857-860

58 Supplementary Figures Chapter 2

0 UBE2I BIRC6 UBE2F UBE2T UBE2A UBE2B UBE2C UBE2K UBE2S UBE2Z UBE2N UBE2U UBE2O UBE2H UBE2M UBE2W UBE2J1 UBE2J2 UBE2L3 UBE2L6 UBE2E1 UBE2E2 UBE2E3 UBE2R1 UBE2R2 UBE2V1 UBE2V2 UBE2D1 UBE2D2 UBE2D3 UBE2D4 UBE2G1 UBE2G2 UBE2Q1 UBE2Q2 UBE2D2 K63E UBE2D2 E2 enzyme

Supplementary Figure 1. Network robustness Importance of E3-connections of individual E2 nodes in the E2-E3 ubiquitination interaction network for system robustness. Individual E2 nodes were deleted from the network, followed by quantification of the number of unconnected E3s, as measure for the system’s fidelity.

UBE2U 100 81-100 UBE2D2 15 100 61-80 UBE2D3 15 94 100 41-60 UBE2D4 15 91 97 100 21-40 UBE2N 8 51 55 58 100 1-20 UBE2D1 13 80 85 85 59 100 0 UBE2W 8 57 61 58 38 57 100 UBE2E1 12 51 55 55 34 61 55 100 UBE2E3 10 46 48 48 28 50 50 79 100 UBE2G2 6 9 9 6 3 7 14 11 13100 UBE2Z 13 9 9 9 7 11 9 11 13 8 100 UBE2L6 8 20 21 21 7 25 27 21 19 0 18 100 UBE2J2 6 6 6 6 10 7 14 0 0 0 9 22 100 UBE2H0 6 3 3 7 4 5 0 0 0 9 0 0100 UBE2T2 6 6 6 3 7 9 0 0 0 0 11 0 0100 UBE2Q12 3 3 0 0 0 5 0 0 8 0 0 0 0 0100 UBE2C2 3 3 3 3 4 0 5 0 0 0 0 0 0 0 0100 UBE2E22 0 0 0 0 0 0 0 0 0 0 0 0 0 3350 0100 UBE2Q22 0 0 0 0 0 0 0 0 0 0 0 0 0 33 0 0100100 UBE2U UBE2D2 UBE2D3 UBE2D4 UBE2N UBE2D1 UBE2W UBE2E1 UBE2E3 UBE2G2 UBE2Z UBE2L6 UBE2J2 UBE2H UBE2T UBE2Q1 UBE2C UBE2E2 UBE2Q2

Supplementary Figure 2. Hub overlap index. Shared E3 binding partners were calculated for each E2 showing more than one interaction.

UBC-fold Interface

Supplementary Figure 3. E2 similarity indices Percentages sequence identity between all human E2 nodes in the E2-E3 network showing more than one interaction, displayed over the complete UBC-fold (left) or the predicted E3-binding interface, as determined by the CNOT4-UbcH5B model (PDB-code: 1ur6; right).

E2 p = 0.001

Number of E3 interactions N-terminal Internal C-terminal 120

0 48 66 31 1 27 12 12 100 2 5 3 6 3 3 2 1 80 4 1 0 2 5 0 1 1

6 1 0 3 60 7 0 2 3 8 0 6 2 9 0 4 3 40

10 0 1 2 interactions E2 of Number 11 0 1 0 Total 20

Total E3 85 98 66 249

0 Total E2 interactions (³ 0) 56 148 142 346 N-terminal Internal C-terminal Position of RING in full-length E3

Supplementary Figure 4. Influence of RING-domain position relative to the full-length E3 protein on E2 interacting potential Number of E2 interactions for N-, C-, or internal RING-finger domains. RING-domains were selected as N-, or C-terminal if the domain boundaries overlap with a maximum of 50 residues counted from the full-length’s E3 N- or C-terminus, respectively.

60 Supplementary tables Chapter 2

See http://www.nature.com/msb/journal/v5/n1/suppinfo/msb200955_S1.html for online supplementary tables 2-6

Supplementary Table 2 Overview of human RING-type E3 ligases Supplementary Table 3 Overview of binary human E2-E3 interactions Supplementary Table 4 Overview of all background-corrected E2-E3 interactions detected on LacZstaining Supplementary Table 5 Overview of all background-corrected E2-E3 interactions detected on LEU2-auxotrophy Supplementary Table 6 Overview of observed E2-E3 interactions compared with literature-curated, biochemically functional interactions

61 Supplementary Table 1. Overview of human ubiquitin-conjugating enzymes Entrez Gene ID 57448 997 7319 7320 11065 326105 7321 7322 7323 51619 606552 340561 7324 7325 10477 286480 140739 7326 7327 7328 392239 7329 51465 118424 3093 7330 7331 7332 7333 9246 283556 9040 606551 7334 389898 317770 63893 55585 92912 54926 27338 246719 440406 29089 148581 7335 170556 7336 55284 619457 65264 UniProt ID Q9NR09 P49427 P49459 P63146 O00762 P51668 P62837 P61077 Q9Y2X8 Q8IWF7 P51965 Q96LR5 Q969T4 Q969M7 P62253 P60604 P62256 P63279 Q9Y385 Q8N2K1 P61086 P68036 O14933 P61081 P61088 Q5JXB2 Q9C0C9 Q7Z7E8 Q8WVN8 Q712K3 Q16763 Q9NPD8 Q5VVX9 Q9UKL1 Q15819 Q96B02 Q9H832

ENSG00000184787 Ensembl ID ENSG00000115760 ENSG00000099804 ENSG00000077721 ENSG00000119048 ENSG00000175063 ENSG00000072401 ENSG00000131508 ENSG00000109332 ENSG00000078967 ENSG00000170142 ENSG00000182247 ENSG00000170035 ENSG00000170035 ENSG00000184182 ENSG00000132388 ENSG00000186591 ENSG00000103275 ENSG00000198833 ENSG00000160087 ENSG00000078140 ENSG00000185651 ENSG00000156587 ENSG00000131982 ENSG00000130725 ENSG00000177889 ENSG00000102069 ENSG00000175931 ENSG00000160714 ENSG00000140367 ENSG00000107341 ENSG00000108106 ENSG00000077152 ENSG00000177414 ENSG00000124208 ENSG00000169139 ENSG00000104343 ENSG00000159202 RefSeq ID NM_016252 NM_004359 NM_003336 NM_003337 NM_007019 NM_003338 NM_181838 NM_181893 NM_015983 NM_003341 NM_152653 NM_006357 NG_004721 NM_080678 NM_003342 NM_182688 NM_003344 NM_003345 NM_016021 NM_058167 NM_005339 NG_003101 NM_198157 NM_004223 NG_002490 NM_003969 NM_003348 NM_001012989 NM_022066 NM_017582 NM_173469 NM_017811 NM_014501 NG_001583 NM_014176 NM_152489 NM_021988 NM_003350 NM_001001481 NM_023079.3 AB051521 AA843379 Accession Numbers AF265555 L22005 AK223045 M74525 U73379 BC015997 L40146 U39318 BC004104 AL121755 BC040290 X92963 AK057886 AB017644 AF136176 BC010549 BC026288 BC008351 BC006277 D45050 AJ245898 AF296658 U58522 S81005 AJ000519 Y09515 AL161893 AF031141 AB012191 CR599143 D83004 AJ243666 BC017708 AK000426 BC004236 AF161499 BC029895 U39360 AL117334 X98091 AK001873 AC092393 Chromosome 2p22.3 19p13.3 Xq24 5q31.1 20q13.12 14q12 10q21.1 5q31.3 4q24 7p13 20p12.3 Xq21.1 3p24.2 3p24.2 2q31.3 Xp22.2 2q37.3 17p13.2 21q22.3 7q32 8q21 16p13.3 6q16.1 1p36.33 4p14 14q24.3 12q12 22q11.2 19q13.2 13q12.3 11q12.1 14q22.3 19q13.43 16p11.2 12q22 Xq27.3 14q11.2 17q25.1 1q22 15q24.2 9p11.2 19q13.43 17p11.2 17p11.2 1q32.1 1p31.3 20q13.2 20p13 8q11.21 8q21.11 1p21.3 17q21.31

MGC10481

E2(17)KB1

MMS2

CROC1

UBC3 KIAA1734

MGC8489

NCUBE1

UBC4/5, FLJ20419,

EDPF-1,

UEV1A,

FLJ13855

UBC13, UBE2R1,

NICE-5 CGI-76,

FLJ12878, UBCH5, UBC4

UBC12

CDC34B, FLJ25157 NCUBE2 L-UBC

CROC-1, DDVit-1,

UBC8

HHR6A HHR6B

HOYS7,

Aliases BRUCE E2-CDC34, UBC2, UBC2, UBCH10 UbcH5A, UbcH5B, UbcH5C HBUCE1 MGC42638 UbcH6 UbcH8, UbcH9 UbcM2 NCE2 UBC7 UBCH, UBC9 HSPC153, Ubc6p, HYPG UBEL1, UBCH7 UBCH8 UBCH7N3 hUbc12, UbcH-ben, E2-230K, PRO3094, DKFZp762C143 UBC3B, E2-EPF HSPC150 MGC35130 UEV-1, dJ687F11.3 UEV-2, FLJ11011 USE1, Previous Symbols SFT UBE2G HIP2 UBE2L5 UBE2L7 UBE2Q UBE2V pseudogene

1 yeast) yeast) yeast)

yeast) yeast) yeast)

yeast) yeast) yeast) yeast) yeast) yeast)

yeast) yeast) yeast) 1 2

pseudogene yeast)

1 1

1 1 2

homolog, homolog, homolog, homolog, homolog, homolog,

pseudogene

member member homolog) homolog) homolog, homolog,

homolog,

1 1 2

homolog,

(UBC4/5 (UBC4/5 (UBC4/5 (putative) pseudogene (UBC7 (UBC7 (UBC6 (UBC6 (UBC4/5 (UBC4/5 (UBC4/5 pseudogene pseudogene pseudogene

(putative) pseudogene (UBC12 pseudogene

(UBC13 pseudogene family family 1 2 3 4 5 N-terminal 1 2 (UBC8 (UBC8 (putative) (RAD6 (RAD6 2 pseudogene (UBC1 1 2 3 4 (putative) pseudogene pseudogene (putative)

1 2 3 4 5 6 7

(UBC9 J1 J2

variant variant variant cerevisiae)

E2A E2B E2C E2C E2D E2D E2D E2D E2D E2D E2E E2E E2E E2E E2F E2G E2G E2H E2H E2I E2, E2, E2K E2L E2L E2L E2L E2L E2L E2L E2M E2M E2N E2N-like E2N E2O E2Q E2Q E2R E2S E2S E2S E2T E2U E2 E2 E2 E2W E2W E2Z (S.

enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme enzyme

homolog

repeat-containing

cycle

IAP

division

Approved Name baculoviral cell ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating

Approved Symbol BIRC6 CDC34 UBE2A UBE2B UBE2C UBE2CP1 UBE2D1 UBE2D2 UBE2D3 UBE2D4 UBE2D5P UBE2DNL UBE2E1 UBE2E2 UBE2E3 UBE2E4P UBE2F UBE2G1 UBE2G2 UBE2H UBE2HP UBE2I UBE2J1 UBE2J2 UBE2K UBE2L1 UBE2L2 UBE2L3 UBE2L4 UBE2L5P UBE2L6 UBE2L7P UBE2M UBE2MP1 UBE2N UBE2NL UBE2NP1 UBE2O UBE2Q1 UBE2Q2 UBE2R2 UBE2S UBE2SP1 UBE2SP2 UBE2T UBE2U UBE2V1 UBE2V1P1 UBE2V2 UBE2W UBE2WP UBE2Z

Chapter 3

The family of ubiquitin-conjugating enzymes (E2s): deciding between life and death of proteins

Accepted for publication in FASEB Journal

65 The Family of Ubiquitin-Conjugating Enzymes (E2s): Deciding between Life and Death of Proteins

Sjoerd J.L. van Wijk and H. Th. Marc Timmers

Department of Physiological Chemistry, Division of Biomedical Genetics, University Medical Center Utrecht, Universiteitsweg 100, 3584 CG Utrecht, the Netherlands

Abstract The family of ubiquitin-conjugating (E2) enzymes is characterized by the presence of a highly conserved ubiquitin-conjugating (UBC) domain. These domains accommodate the ATP- activated ubiquitin (Ub) or ubiquitin-like (UBL) protein via a covalently linked thioester onto its active-site residue. E2 enzymes are acting via selective protein-protein interactions with the E1 and E3 enzymes, connecting activation to covalent modification. By doing so, E2s dif- ferentiate effects on downstream substrates, either with a single Ub/UBL molecule or as chain. While E3s are involved in substrate selection, E2s are the main determinants for selection of the lysine to construct ubiquitin chains, thereby directly controlling the cellular fate of the substrate. In humans, 35 active E2 enzymes have been identified so far, while other eukaryotic genomes harbor 16 to 35 E2 family members. Some E2s possess N- and/or C-terminal extensions that mediate E2-specific processes. During the last two decades strong support has been accumulated for the control of E2 enzymes in decisions concerning protein’s life or death. Here, we summarize current knowledge and recent developments on E2 enzymes with respect to structural characteristics and functions. From this we propose a shell-like model to rationalize the selectivity of these key enzymes in directing Ub/UBL-conjugation pathways.

Introduction Many intracellular proteins become covalently modified with ubiquitin (Ub) or ubiquitin-like proteins (UBLs) like SUMO, ISG15 or NEDD8 (1). These modifications determine protein turnover, regulation and molecular function, providing a complex regulation of the proteome. Both Ub and UBLs show highly conserved mechanisms in the way they become processed and covalently attached to their substrates (2). Before conjugation takes place, the initial activation of the modifier is carried out by activating-enzymes (E1s) in an ATP- dependent manner (extensively reviewed in (3). The E1 transfers the activated modifier to a family of enzymes called ubiquitin-conjugating enzymes (E2s) to form a high-energetic conjugate with Ub/UBL via a highly conserved catalytic cysteine residue. Ub/UBL-loaded E2s are now selectively interacting with protein ligases (E3s) that recruit and bind specific substrates. Besides Homologous to E6-AP C-Terminus (HECTs) (4) and Really Interesting New Gene (RING) E3s (5), U-box-containing proteins like CHIP (Carboxyl terminus of Hsc70-Interacting Protein) can also act as E3 enzymes (6). Where HECT E3s first accept the activated ubiquitin prior to substrate modification, RING and U-box E3s bind both the substrate and the ubiquitin-loaded E2, acting like scaffolds, providing an optimal conformation for ubiquitin transfer. In a few cases however, ubiquitin transfer can occur without the direct involvement of a

66 specialized E3 ligase, like in case of coupled mono-ubiquitination (7). Implications of this finding will not be discussed in this review. Ub/UBL-modifications occur in cycles wherein multiple modifiers are conjugated to substrates. Sometimes this results in chains, like for ubiquitin or SUMO (8), forming extended polymers of interconnected modifiers (9). Apart from chain conjugation, some substrates become decorated with only one Ub/UBL at a single residue, while others are multiply modified at more than one residue (10). Often, modification of substrates does not imply a permanent state. In the past few years, more and more deubiquitinases and ubiquitin-like proteases have been discovered which actively remove these small polypeptide tags, underscoring the dynamic nature of Ub/UBL conjugation (11).

Central in this enzymatic cascade is the family of ubiquitin-conjugating enzymes (E2) that represents a conserved group of enzymes that couple activation of the Ub/UBL to downstream conjugation events (2, 12). E2 enzymes are essential for adequate conjugation and, at least for ubiquitin, are directly influencing the type of lysine used to label substrates and thereby influencing the fate of the substrate. In this review, we discuss recent developments in the understanding of E2 enzyme function, regulation and specific E1 and E3 interactions and how selectivity of E2 enzymes towards Ub and UBLs critically decides between life and death of proteins.

E2 enzymes: a multi-tasking family? Members of the family of ubiquitin-conjugating enzymes (E2s) are characterized by the presence of a highly conserved 150-200 amino acid ubiquitin-conjugating catalytic (UBC) fold (2, 12). These domains of 14-16 kDa are approximately 35% conserved among different family members and provide a binding platform for E1s, E3s and the activated Ub/UBL (12). With- in this domain, a catalytic cysteine is embedded that accepts the activated Ub/UBL via a thioester bond, prior to interaction with E3 ligases and subsequent substrate conjugation (Figure 3.1). With the discovery of ubiquitin, E2s were identified that exclusively conjugate ubiquitin (2). At this moment, approximately a dozen of UBLs have been identified and each equipped with their own particular E2. Because of this specificity, Ub/UBL conjugation pathways are parallel. In some cases, however, extensive cross-talk between the pathways is occurring, like between the ubiquitin and ISG15-conjugation pathways (13) and between ubiquitin and SUMO (14). An overview of the predominant ubiquitin-like protein E2 enzymes, including their specific E1s and E3s is given in Figure 3.1. Small Ubiquitin-like MOdifier (SU- MO) becomes exclusively conjugated via Ubc9(UBE2I) (15). Ubc9 is highly conserved from yeast to mammals and displays high levels of similarity towards ubiquitin and UBL E2s. De- spite this similarity, Ubc9 only accepts activated SUMO from the dimeric SUMO E1 SAE1/SAE2 and transfers it to downstream targets via a handful of SUMO E3 ligases (16). In- terestingly, in some cases Ubc9(UBE2I) directly interacts with and modifies its substrate without the help of an E3 (17). This occurs via recognition and binding of Ubc9(UBE2I) to a lysine-rich specific SUMOylation motif in substrates (16) or through binding of Ubc9(UBE2I) to the phosphorylated form of MEF2A (18).

67 Ubiquitin SUMO NEDD8 ISG15 Ubl E1 ATP E1 (2): E1 (1): E1 (1): E1 (1): PPi Uba1 Aos1/Uba2 APPBP1/Uba3 UBE1L Uba6 E1 ~ Ubl E2 E1 E2 (30): E2 (1): E2 (2): E2 (3): E2 Ubl ~ UBE2A(hHR6A) UBE2L3(UbcH7) UBE2I(Ubc9) UBE2M(Ubc12) UBE2L6(UbcH8) ± E3 Substrate UBE2B(hHR6B) UBE2L6(UbcH8) ± UBE2F(NCE2) UBE2E1(UbcH6) ± E2 UBE2C(UbcH10) UBE2N(Ubc13) UBE2E2 ± UBE2D1(UbcH5A) UBE2O(E2-230K) E3 UBE2D2(UbcH5B) UBE2Q1(NICE-5) Substrate UBE2D3(UbcH5C) UBE2Q2 Ubl UBE2D4(HBUCE1) UBE2R1(CDC34) UBE2E1(UbcH6) ± UBE2R2(CDC34B) UBE2E2 UBE2S(E2-EPF) UBE2E3(UbcH9) UBE2T(HSPC150) UBE2G1(UBE2G) UBE2U* UBE2G2(UBC7) UBE2V1(UEV-1A) E3 UBE2H(UBCH) UBE2V2(MMS2) Substrate UBE2J1(NCUBE1) Ubl UBE2W Ubl UBE2J2(NCUBE2) UBE2Z(Use1) Ubl UBE2K(HIP2) BIRC6(apollon) Ubl DUB E3 E3 (>1000): E3 (4): E3 (2): E3 (2):

Ubl Substrate Single/multiple subunit RanBP2, Pc2, Rbx1 HERC5 Ubl Ubl RING, HECT, U-box, PHD PIAS-proteins, Rbx2 EFP Processed Ubl Topors

Figure 3.1. Overview of ubiquitin and ubiquitin-like conjugation pathways Schematic representation of the general Ub/UBL conjugation cascade in the outer left column. Before activation can take place, the Ub/UBL needs to become processed by deubiquitinating enzymes (DUBs) in order to expose the C-terminal diglycine residues. Next the Ub/UBL is activated in an ATP-dependent manner by an activating enzyme (E1), after which, via protein-protein interactions, the high-energetic Ub/UBL is transferred onto the catalytic cysteine of ubiquitin- conjugating enzymes (E2). In a final step, the activated Ub/UBL is transferred to substrates in a process catalyzed by protein ligases (E3). Different UBLs could have various effects on substrates (not depicted here). Outline of the major Ub/UBL conjugation pathways in higher eukaryotes. Besides Ub, the UBLs SUMO, NEDD8 and ISG15 each become activated by their cognate E1. The majority of Ub/UBLs use dedicated E2s for conjugation, but in some cases E2s are shared, for example for ubiquitin and ISG15 conjugation (indicated with ±). For E2s indicated with an *, no E2 activity has been described yet. Both UBE2L6 and UBE2E2 have UbcH8 as alternative protein name, but only UBE2L6 is aliased UbcH8. Finally, sets of devoted E3 mediate the covalent conjugation of the UBLs to the appropriate targets. Not shown are conjugation cascades for Atg8 and Atg12, FUBI, FAT10 and Urm1.

The NEDD8 (Neural precursor cell Expressed, Developmentally Down-regulated 8) protein becomes conjugated to subunits of the Skp1-cullin-F-box (SCF) complex and other cullin- containing E3 complexes (1, 19). As a result, the activity and recruitment of E2 enzymes to these complexes is altered, affecting SCF-mediated ubiquitination. The NEDD8 E2 Ubc12(UBE2M) is conserved among all eukaryotes and catalyzes NEDDylation of the Rbx1 RING-subunit of SCF-complexes (19). Recently, NCE2(UBE2F) has been identified as a second E2 for NEDDylation (20). Whereas Ubc12(UBE2M) displays interaction specificity towards

68 Rbx1, NCE2(UBE2F) selectively interacts with Rbx2. Both Ubc12(UBE2M)/Rbx1 and NCE2(UBE2F)/Rbx2 pairs have different effects on target modification since they conjugate different sets of cullin subunits (20). The identification of a second E2 for NEDD8 provides a first example of expansion within an UBL-conjugation system, resembling the broad variety of E2 enzymes within the Ub-conjugation machinery. Conjugation of the UBL Interferon- Stimulated Gene 15 (ISG15) is regulated by signal transduction pathways involved in innate immunity (21). ISG15 expression is upregulated when cells are treated with interferon or li- popolysaccharide as seen in viral infections, after which ISG15 can be found both in free and in conjugated forms (21). A dedicated ISG15 E1, UBE1L, charges the UBL and transfers it to UbcH8(UBE2L6) (22). Whereas UbcH8(UBE2L6) is identified as the main E2 involved in ISGy- lation (23), several other E2s, like UbcH6(UBE2E1) and UBE2E2 are also able to receive the activated ISG15 (24). Strikingly, all E2s involved in ISGylation are capable of ubiquitin conjugation, providing an example of UBL pathways that share common conjugation enzymes. Be- sides the abovementioned E2 enzymes, additional less-studied UBL E2s have been identified (which are not discussed in this review). Atg3, for example, is the E2-like enzyme for the conjugation of the Atg8 UBL and phosphatidylethanolamine (PE) and another UBL, Atg12, becomes conjugated via the Atg10 E2 (25). Both Atg8-PE and Atg12 conjugates are involved in autophagy, the major process for degradation of cytosolic macromolecular entities. Despite the identification of many UBL E2s, E2-activities for several other UBLs (FUBI (26), FAT10 (27) and Urm1 (28)) and several putative UBLs (like BUBL1 and -2, UBL-1, SF3A120 and oligoade- nylate synthetase, for an overview see (29)) remain to be identified.

Evolutionary organization, protein domain architecture and localization of E2s Evolutionary relationships The occurrence of ubiquitin and UBL pathways was thought to be restricted exclusively to eukaryotic species. Surprisingly, a similar pathway has been identified recently in prokaryotes. In a screen for interacting proteins of the proteosomal ATPase in Mycobacterium tu- berculosum (Mtb), Pearce et al. (30) identified a 6.9-kDa Pup (prokaryotic ubiquitin-like protein) protein. Pup becomes activated in a manner involving deamidation by Dop, followed by covalent linkage to lysine residues of acceptor substrates (such as the proteasomal subunit malonyl co-A acyl carrier protein (FabD)) and targeting them for proteasomal degradation (31). Proteasome-associated factor A (PafA) has been implicated in the conjugation (30), since pafA Mtb strains are absent in Pup-FabD conjugates and the pool of unmodified FabD is stabilized. Now, it is tempting to speculate about E2-like activities in prokaryotes resembling eukaryotic E2s. However, no enzymes bearing the UBC-fold have been identified in bacteria (32). Since PafA combines E2- and E3-like features in Mycobacteria, their features have possible become separated during eukaryotic evolution.

E2 enzymes are present in all eukaryotes, underscoring the significance of Ub/UBL systems in biology (33, 34). Genes encoding for E2 proteins are scattered throughout the genome and no general similarities in gene organization can be distinguished. The protein family is

69 expanded during evolution; lower eukaryotes have lower numbers of E2 enzymes than higher ones (for a recent overview: (33)). For example, in Saccharomyces cerevisae 16 E2 enzymes are found, whereas in human 35 active E2s are described (33). The variability in number of E2s between different organisms mostly results from gene duplication. For example, in yeast the almost identical E2s UBC4 and UBC5, involved in the degradation of short-lived and abnormally folded proteins, occur in higher eukaryotes as UbcH5A-D(UBE2D1-4) and UbcH6, -H8 and -H9 (UBE2E1-3) (35). The same holds true for yeast RAD6, CDC34 and UBC7, which have evolved into the hHR6A/B(UBE2A/UBE2B), CDC34 and CDC34B(UBE2R1/UBE2R2) and UBE2G/UBC7(UBE2G1/UBE2G2) pairs, respectively. Some E2s appear to be specific for higher species. The giant (528 kDa) E2 BIRC6(apollon) has not been found in yeast, whereas it is present in worms, flies, mice and human (36). The same is true for HSPC105(UBE2T), UBE2W and Use1(UBE2Z). Interestingly, UBE2U has only been identified in mammals. In general, a small subset of E2s is present in all eukaryotes, defining an old and evolutionary conserved panel of essential E2 enzymes, including the SUMO E2 Ubc9(UBE2I) and the NEDD8 E2 Ubc12(UBE2M). More specialized E2 enzymes exhibiting specific functions have evolved in higher organisms (33).

E2 enzyme architecture E2s are classified based on the existence of additional extensions to the catalytic core; some E2s only consist of the catalytic domain (class I), others have additional N- or C-terminal extensions (class II and III, respectively) or both (class IV) (Figure 3.2) (2, 34). These extensions are involved in functional differences between E2s, which involve differences in subcellular localization, stabilization of the interaction with E1 enzymes or modulation of the activity of the interacting E3. The class II E2 enzyme UbcH10(UBE2C), well-known to interact with the Apc11 RING subunit of the Anaphase-promoting complex (APC) involved in cell cycle regulation, contains a highly conserved N-terminal extension (37). This extension regulates ubiquitination activity and the number of substrate lysines that becomes modified by the E3 complex. The presence of this N-terminal extension also promotes ubiquitination of substrates lacking a clear degron, or signature that targets the substrate to become ubiquitinated. Fu- sion of the UbcH10(UBE2C) N-terminal extension to UbcH5B(UBE2D2), an E2 interactor of the APC-complex, does not deregulate ubiquitination, indicating a co-adaptation of the reg- ular E2-E3 interface and the N-terminal stretch (37). Cell Division Cycle 34 (CDC34/UBE2R2) is a class III E2 enzyme involved in G1-S phase transition of the eukaryotic cell cycle. CDC34(UBE2R2) fulfills this role by interacting with the multi-subunit SCF [Skp1 (S-phase kinase-associated protein 1)/cullin/F-box] E3 ligase complex (38, 39). Besides its UBC-fold, CDC34(UBE2R2) has an acidic tail that is essential for viability of yeast and cell cycle progression. It has been shown that certain residues within this acidic tail are phosphorylated by protein kinase CK2 and that this modification influences SCF-mediated Sic1 ubiquitination, cell cycle progression and possibly also localization of the E2 enzyme (40, 41).

70 UBC-fold Class I UBE2A (hHR6A) UBE2B (hHR6B) Class II UBE2C (UbcH10) UBE2D1 (UbcH5A) Class III UBE2D2 (UbcH5B) UBE2D3 (UbcH5C) Class IV UBE2D4 (HBUCE1) UBE2E1 (UbcH6) UBE2E2 UBE2E3 (UbcH9) UBE2F (NCE2) UBE2G1 (UBE2G) UBE2G2 (UBC7) UBE2H (UBCH) UBE2I (Ubc9) UBE2J1 (NCUBE1) UBE2J2 (NCUBE2) UBE2K (HIP2) UBE2L3 (UbcH7) UBE2L6 (UbcH8) UBE2M (Ubc12) UBE2N (Ubc13) UBE2O (E2-230K) UBE2Q1 (NICE-5) UBE2Q2 UBE2R1 (CDC34) UBE2R2 (CDC34B) UBE2S (E2-EPF) UBE2T (HSPC150) UBE2U UBE2V1 (UEV-1A) UBE2V2 (MMS2) UBE2W UBE2Z (Use1) BIRC6 (apollon) 0 100 200 Residues Figure 3.2. The family of human ubiquitin-conjugating enzymes Schematic overview of the superfamily of active, human E2 enzymes. The UBC-fold is indicated as dark-blue ellipse and extensions as blue lines. The UBA- domain of HIP2(UBE2K) is indicated as red line and the BIR-domain of apollon(BIRC6) is indicated as green line. Both UBE2L6 and UBE2E2 have UbcH8 as alternative protein name, but only UBE2L6 is aliased UbcH8. Different classes of E2 enzymes are indicated in different colors. Protein length in amino acids is indicated by the ruler.

Because of these extensions E2 enzymes differ dramatically in size. For example bona fide class I E2s, like UbcH5B(UBE2D2) and hHR6A(UBE2A), only consists of a bare UBC-fold. Other E2s, like class II Ubc12(UBE2M) and UbcH6(UBE2E2) and class III UBE2U can reach up to 200- 400 amino acids in size and contain non-conserved extensions. Two exceptionally large E2 enzymes have been identified. The first, class IV E2-230K(UBE2O) is a 1292 amino acid enzyme, involved in reticulocyte maturation and hematopoiesis (42). E2-230K(UBE2O) is able to directly ubiquitinate endogenous substrates without the involvement of any known E3. Furthermore, it has been shown that E2-230K(UBE2O) is sensitive for inorganic arsenite ad- duct formation. This unique feature is mediated via the catalytic cysteine residues, thereby inhibiting the transfer of ubiquitin from the E2 to the substrate when treated with arsenite (43). A second gigantic (528 kDa) class IV E2 apollon(BIRC6) combines an N-terminal baculoviral IAP (inhibitor of apoptosis) repeat (BIR) domain and a C-terminal UBC-fold. The essential apollon(BIRC6) enzyme has an anti-apoptotic role. Using its BIR- and UBC-domains, apollon(BIRC6) promotes the ubiquitination and proteasomal degradation of the pro-apoptotic protein SMAC and inhibition of caspase-9 (44). Recently, apollon(BIRC6) has been linked to the process of abscission, the actual separation of dividing cells by the removal of the midbody ring during cytokinesis (45). Apollon(BIRC6) has an essential role within this process, since its depletion from U2OS osteosarcoma cells leads to cytokinesis defects and cytokine-

71 sis-associated apoptosis. Although an enrichment of free ubiquitin colocalizes with apollon(BIRC6) at the midbody ring, it remain to be established how the apollon(BIRC6) protein itself and its E2 activity are involved in this. It is attractive to speculate about potential mechanisms and ubiquitination targets, especially since apollon(BIRC6) previously has been shown not to require an E3 activity to be active in ubiquitination (36). Given its large size, apollon(BIRC6) could act like a scaffold, bringing substrates and regulatory proteins in close contact, to allow ubiquitin transfer.

In human, the majority of E2 enzymes are actively involved in Ub/UBL conjugation. At present, it remains unclear whether UBE2U has conjugating activity towards Ub or UBLs. An overview of human E2s and their known functions is shown in Supplementary Table 1. Sev- eral E2 pseudogenes, like UBE2CP1, UBE2DNL, UBE2D3L, UBE2E4P, UBE2HP, UBE2L5P, UBE2L7P, UBE2MP1, UBE2NP1, UBE2SP1, UBE2SP2, UBE2V1P1 and UBE2WP have been found for which no mRNA or protein products were detected. Interestingly, also several ubiquitin-enzyme variants (UEVs) have been found, like UEV-1A(UBE2V1), MMS2(UBE2V2), UEV and lactate/malate dehydrogenase domains (UEVLD), fused toes homolog (AKTIP/FTH), UBE2N-like (UBE2NL) and tumor susceptibility gene 101 (TSG101). The E2 variant enzymes UEV-1A(UBE2V1) and MMS2(UBE2V2) have the UBC-fold, but lack the active site cysteine necessary for ubiquitin coupling (46). Both UBE2Vs are known to interact with Ubc13(UBE2N) to form dimers, which catalyze lysine 63 (K63)-linked ubiquitin chains. These chains can be recognized by the 26S proteasome in vitro, but it is currently unclear whether this results in proteasomal degradation in vivo (47-49). Interaction of Ubc13(UBE2N) with either UEV-1A(UBE2V1) or MMS2(UBE2V2) leads to differences in K63-linked ubiquitin chains (50). On a functional level, the Ubc13(UBE2N)-UEV-1A(UBE2V1) dimer is involved in nuclear factor-kappa B (NF-κB) activation but not in DNA damage repair, whereas the Ubc13(UBE2N)-MMS2(UBE2V2) dimers catalyze K63-chains during DNA-damage responses but do not affect NF-κB signaling (51). Ubc13(UBE2N) interacts with MMS2(UBE2V2) via a β- sheet surface, where the variant E2s interact with and position the ubiquitin in such way that Ubc13(UBE2N)-mediated K63 chain synthesis can take place (46). UEVLD(UEV3) has an N- terminal UBC-domain, lacking the essential cysteine, which is fused to a C-terminal sequence that shows homology with members of the lactate dehydrogenase family. It is unknown if UEVLD(UEV3) has the ability to interact with Ubc13(UBE2N), nor any involvement in UBL- conjugation has been described (52). AKT-interacting protein (AKTIP/FTH) is implicated in the trafficking and fusion of vesicles (53). The mouse homolog produces fused toes and thymic hyperplasia in heterozygous animals while homozygous knockouts die early in embryonic development. This protein is implicated in apoptosis, explaining the observed programmed cell death phenotypes in mice (54, 55). The protein directly interacts with serine/threonine kinase protein kinase B (PKB)/AKT and regulates kinase activity by enhancing the phosphorylation of regulatory sites in PKB (55). Finally, TSG101 has been shown to be involved in cell growth, differentiation and genome stability and in the trafficking of multivesicular bodies (56, 57).

Few E2s combine the UBC-fold with other protein domains (Figure 3.2). Besides the abovementioned apollon(BIRC6), huntingtin-interacting protein 2 (HIP2, UBE2K) is capable of modifying the huntingtin protein, implicated in Huntington’s disease (58). HIP2(UBE2K) is involved in aggregate formation of extended polyglutamine proteins and polyglutamine- induced apoptosis (58). Besides its UBC-fold, HIP2(UBE2K) has an ubiquitin-associated (UBA) domain, which is involved in mediating protein-protein interactions through binding of ubiquitin molecules (59). The UBA-domain of HIP2(UBE2K) resides in a specific 47-residue tail that enables the E2 to catalyze unanchored K48-ubiquitin chains, provided resistance against alkylating agents and strongly selects against interaction with non-mammalian E1 enzymes (60). In addition, HIP2(UBE2K) has been shown to catalyze cyclical conformation of ubiquitin chains of various lengths (47). However, a direct role of the UBA in this remains unclear.

E2 regulation by physical barriers The majority of E2s are ubiquitously expressed, indicating a general involvement of these enzymes in ubiquitin-conjugation. For most E2 enzymes, ESTs, mRNAs and proteins have been identified in a wide variety of tissues and cell types. Nevertheless, some E2 enzymes are found predominantly in certain tissues. For example, UBE2U is expressed in the urogenital tract, E2-230K(UBE2O) is mainly found in skeletal and heart muscle and HIP2(UBE2K) is highly expressed in specific regions of the brain. On a subcellular level, many E2 enzymes are found both in the nucleus and in the cytoplasm. It has been shown that covalent modification of certain E2s with ubiquitin on their catalytic cysteine mediates their import into the nucleus via importin-11 (61). Some E2 enzymes are exclusively restricted to specific cellular compartments. The most obvious examples are Ubc6 and Ubc7 (yeast homolog’s of human NCUBE1/2(UBE2J1/J2) and UBE2G/UBC7(UBE2G1/G2), respectively) , which are both localized in the lumen of the endoplasmic reticulum (ER) to participate in ER-associated degradation (ERAD) (62). These observations stress the importance of spatial and temporal regulation of E2 function and they provide additional layers of complexity for specific members of the E2 family.

Cross-talk between Ub/UBLs: regulation of E2 activity by post-translational modifications Another layer of complexity is added by observations that some E2s can become modified with Ub/UBLs themselves. This illustrates the intermingling of UBL-conjugation pathways and how this cross talk influences the performance of other UBL-networks. A typical example comes from the yeast Ubc7, which is ER-localized and normally sequestered at the ER- membrane via the transmembrane protein Cue7 (63). When Ubc7 protein levels exceed the levels of Cue7, Ubc7 becomes autoubiquitinated on its catalytic cysteine, mediated by the HECT E3 ligase Ufd4, followed by proteasomal degradation (64).

73 A

L1 B

CNOT4-UBE2D2 BRCA1/BARD1-UBE2D1 CHIP-UBE2N/UBE2V1 CHIP-UBE2D1 E6AP-UBE2L3 hot-spots E6AP-UBE2L3 c-Cbl-UBE2L3

Helix 1 Loop 1 Loop 2

Figure 3.3 The UBC-fold (A) Structural depiction of the UBC-fold of UbcH5B(UBE2D2) (PDB: 1w4u). Indicated are the conserved secondary structural elements (-helices in red, -sheets in blue and loops in yellow). Shown are the major sites of E1 and E3 interactions, composed of -helix 1, loop 1 and loop 2. Conserved catalytic cysteine is shown as magenta sphere. Molecular graphics images were generated using the UCSF Chimera package (http://www.cgl.ucsf.edu/chimera). (B) Overview of molecular contacts in E2s that mediate selectivity towards E3s. Alignment shows residues in Ubc13(UBE2N), UbcH5B(UBE2D2) and UbcH7(UBE2L3) and residues marked with a colored dot correspond with the equally colored E2-E3 structure. See main text for additional information and references).

This is consistent with the murine Ubc7 counterpart, Ube2g2, which also becomes modified with a chain of ubiquitin onto its catalytic cysteine (65). Apparently, this automodification requires two Ube2g2 molecules, each carrying an activated ubiquitin on their catalytic cysteines (65). Strikingly, preassembled K48-linked chains can be transferred from the active cysteine to a lysine residue within the E2 molecule itself or the lysine acceptor site in a substrate, but it is unclear whether this leads to proteosomal degradation in vivo (64). These observations provide an intriguing mechanism of how polymeric ubiquitin chains could be conjugated and transferred as a chain to an acceptor site. An example of how an Ub-E2 can become modified with a UBL is provided by Ubc13(UBE2N). This E2 becomes covalently modified with ISG15, thereby suppressing the ability to form ubiquitin thioesters (13, 66). ISGylation occurs at K92 and this residue is located in the assumed E1- and E3-binding sur-

74 face. Another example is the covalent modification of UBE2K(HIP2) by SUMO-modification on K14 in helix 1, which inhibits its interaction with the ubiquitin E1 (67). These examples testify of an extensive cross-regulation between different Ub/UBL-conjugation machineries, wherein several modifications are deposited on key enzymes, thereby altering their function and also the activity of the specific Ub/UBL-pathway as a whole.

   90o 90o 90o

Figure 3.4. E2-E3 complexes Ribbon representation overlaid with a transparent molecular surface of a representative E2-E3 complex. Shown is the model between the UbcH5B(UBE2D2) E2 enzyme and the RING-finger domain of the CNOT4 E3 ligase (PDB: 1ur6). Depicted in different colors are: UbcH5B(UBE2D2) (green), CNOT4 (dark blue), E1-E2 interaction interface; partially overlapping with the E2-E3 interaction interface (orange), E2- E3 interaction interface (red), catalytic cysteine (magenta), residues involved in binding the activated Ub/UBL (yellow) and conserved -sheet (light blue). Molecular graphics images were generated using the UCSF Chimera package (http://www.cgl.ucsf.edu/chimera).

Characteristic structural features of E2 enzymes The catalytic properties and molecular determinants of E2-interaction specificity regarding E1 and E3 enzymes are concentrated in the conserved UBC-fold (Figures 3.3A & 3.4). The

UBC domain consists of four α-helices, a short 310 helix, and a four-stranded, antiparallel - sheet (Figure 3.3A). The ß-sheet forms a central region with helix 2, bordered with helix 1 (H1) at one side and helices 3 and 4 at the other side (12, 33, 34). Located C-terminal to ß- strand 1 and 3 are two loop regions, L1 and L2. Although the global fold of the UBC-domain has been conserved, both loops display high levels of sequence variability, length and conformational freedom. These loop regions are relatively flexible and are involved in selection and binding of specific E3s (12). The catalytic cysteine lays in a highly conserved loop that connects ß-strand 4 with helix 2, forming a shallow groove of residues from the same loop and those of helix 2 and 3 (12). Within the UBC-fold, several regions mediate specialized functions. First, at one side of the UBC-domain, H1, L1 and L2 harbor the main determinants for E1 and E3 binding (Figure 3.3A & B). Second, residues surrounding the active site cysteine are organized in structural elements that facilitate processing of the activated Ub/UBL

75 and finally, some E2s have evolved a specific β-sheet surface that bind ubiquitin or other E2 enzymes (Figure 3.4).

E1-E2 interaction interface Information regarding E1-interacting regions of E2 enzymes comes from structural studies of the E1s and E2s for NEDD8 and SUMO. The dimeric Aos1/Uba2(SAE1/SAE2) complex carries out SUMO activation. Within this dimer, regions in Aos1 are homologous with the N- terminus of the ubiquitin E1 and domains in Uba2 share homology with the C-terminus of the ubiquitin E1, underlining the evolutionary relationship between these enzymes (68). Aos1 is mainly involved in adenylation of SUMO, whereas Uba2 contains three domains: the catalytic cysteine domain, the adenylation domain and the ubiquitin-fold domain (UFD). The UFD displays a strong interaction with E2s and is necessary for E2 recruitment and binding (68). It has been shown that Ubc9(UBE2I) interacts with the SUMO E1 via its H1 and residues in the ß1-ß2 loop (L1) (Figure 3.4). In addition to these contacts, chemical shift perturbation experiments have revealed that helix 3 of Ubc9(UBE2I) exhibit a weak, but important, affinity for the catalytic cysteine domain of the SUMO E1 (69). Additional clues came from the APP binding protein 1 (APPBP1)/UBA3 dimeric NEDD8 E1 complex (70). The interaction between the C-terminal domain from NEDD8s heterodimeric E1 (APPBP1-UBA3) and the catalytic core domain of Ubc12(UBE2M) pointed to the interaction interfaces between E1 and E2 enzymes (71-73). Mostly H1 and the L1 were involved in E1 binding indicated in orange in Figure 3.4. These E1-binding surfaces are conserved, since several mutations in helix 1 re- duces the ability to form thioesters with several E2s (74-76). In addition, mutation in the H1 of E2s leads to decreased transthioesterification with the ubiquitin E1 (77).

E2-E3 interaction interface To complete the process of conjugation, E2s interact with E3 ligases to transfer the activated UBL to the substrate lysine acceptor residues. E3 ligases can be classified according to the presence of a RING finger or HECT domain. The RING finger domain is formed by conserved patterns of cysteine and histidine residues that coordinate zinc atoms in a cross-brace structure (5). The HECT domain is a ~350 residues domain, which adopts a bilobal structure. In HECTs, the N-terminal lobe contains the E2-binding site and the C-terminal lobe harbors the catalytic centre. Both lobes are separated by a small linker stretch, forming a cleft with the active site (4, 78). RING finger proteins bind both Ub-loaded E2s and substrates (5). In contrast, HECT E3 ligases bind Ub/UBL-loaded E2s at their E2-binding domain and the Ub/UBL is first transferred to the HECT’s active-site cysteine, prior to conjugation to substrates (4). Al- though both domains are different in global structural appearance, E2 enzymes are employing conserved residues to interact with both (Figure 3.3B). The first structural clues of E2- RING E3 interactions came from the crystal structure of the E3 ligase c-Cbl RING and UbcH7(UBE2L3) (79). The tips of the two loop regions of UbcH7(UBE2L3) contact the RING via a shallow groove that contains an -helix and the two zinc-chelating loops (loop 1 and loop 2) of the RING domain. Although some E2 contacts are made with regions outside the

76 RING-finger, the E3 groove region contributes mostly to binding, mediated via F63 of L1 and P97 and A98 of L2 of UbcH7(UBE2L3). Besides groove contacts, the c-Cbl linker region interacts with H1 of UbcH7(UBE2L3). The central F63 residue in UbcH7(UBE2L3) characterizes common amino acids in E2s capable of interacting with c-Cbl and supporting c-Cbl-mediated ubiquitination. UbcH7(UBE2L3) is also able to interact with the E6AP HECT E3 ligase, involved in human papilloma virus-induced degradation of p53 and mutated in Angelman syndrome (78). When interacting with E6-AP, UbcH7(UBE2L3) L1 residues contribute most to E3 binding, clearly supported by contacts to H1 and L2 of the E3. Again, the F63 residue packs into the groove of the E3 and additional contacts in loop 1 are made via A59, P62 and E60. Within the E2 loop 2, contacts are made via P97, A97, K96 and K100 (78). Many of the residues of the E3-interaction interface of UbcH7(UBE2L3) are conserved in the closely-related E2 enzyme UbcH8(UBE2L6) that is also able to interact with E6AP. Alanine-scanning has been performed on the UbcH7-E6AP interface and binding constants for the E2-E3 complex revealed so-called hot-spots within the E2-E3 interface (80). Residues R5, R6 and K9 in helix 1, P62 and F63 in loop 1 and K96 and K100 in loop 2 behave as hotspots in the UbcH7-E6-AP interface (80) (Figure 3.3B). Structural clues regarding interactions with different E2 enzymes also come from the U-box E3 ligase CHIP (carboxy-terminus of heat-shock protein 70 (Hsp70) interacting protein. This E3 is involved in the chaperone-mediated degradation of misfolded and unstable proteins (81, 82). The U-box of CHIP has been shown to interact with UbcH5A- D(UBE2D1-4) and the Ubc13(UBE2N)-UEV-1A(UBE2V1) dimer, both physically and functionally (83, 84). The crystal structure of CHIP-UbcH5A(UBE2D1) reveals contacts between UbcH5A(UBE2D1) H1, L1 and L2 and a central groove in the CHIP U-box (83). The interaction between CHIP and Ubc13(UBE2N)-UEV-1A(UBE2V1) also involves E2 residues in H1, L1 and L2 (85). Thus, in most E2-E3 complexes, H1, L1 and L2 residues are buried in the central groove provided by the E3 domain. Correct orientation of the loop regions and important E2 side-chains organization is maintained by the presence of highly conserved proline residues. Strikingly, the so-called S-P-A motifs, three conserved residues located in L2 of many E2s, are important mediators for E2 interactions and the presence of this motif has been shown to predict productive interactions for the CHIP U-box (Figure 3.3B). The relevance of this motif has become clear when a BRCA1-BARD1 fusion construct was screened in yeast two-hybrid analysis against a panel of E2 enzymes (86). BRCA1-BARD1 is a dimeric E3 ligase, wherein BRCA1 provides the E2-binding RING that displays E3 activity but only when the BARD1 RING is present. The E3 dimer was found to interact with a subset of E2 enzymes and NMR chemical shift analysis in complex with UbcH5C(UBE2D3) revealed again residues in H1 and L2 as the most important contacts with BRCA1 (86). A combination of docking approaches, NMR analysis and mutagenesis provided insights in the structural organization of the CNOT4- UbcH5B(UBE2D2) interaction (87, 88). Also in this model, the H1, L1 and H2 residues are involved in E3-interaction (88). In conclusion, the structural comparisons between E1-E2 and E2-E3 complexes indicate overlapping surfaces are used on the E2. Thus, in order to interact with E3s, E2s must disengage first from the E1, providing a sequential mode of action (89) (Figure 3.4).

77 This suggests that differences in E1-E2 and E2-E3 interaction kinetics could have important implications for the enzymatic mechanism of substrate poly-ubiquitination. In case of a dis- tributive mechanism, one would predict that the E2 shuttles between E1 and E3 to accept and transfer the activated Ub/UBLs one-by-one on the substrate to extend the growing chain. In contrast, a processive mechanism would arise from a (prolonged?) E1-E2 interactions that lead to assembly of the Ub/UBL chain onto the E2, which is subsequently transferred in a single step to the substrate. This would stress the importance of dynamic differences between E1-E2 and E2-E3 interactions in the regulation of Ub/UBL chain assembly.

E2-Ub/UBL interactions When an E2 accepts the activated Ub/UBL from an E1, it needs to accommodate the active cysteine-bound labile, noncovalent E2~Ub/UBL thioester interaction. NMR analysis of the yeast homolog of the human HIP2(UBE2K) Ubc1~Ub thioester involves C-terminal surfaces on ubiquitin and E2 residues proximal to the active site (90) (Figure 3.4). The C-terminus of ubiquitin becomes somewhat extended and wraps around a specific part of the E2, occupy- ing a cleft formed by E2 side chains. Importantly, ubiquitin binding does not overlap with E2 regions that are involved in E3 binding (Figure 3.4). The main sites for Ub/UBL binding are the catalytic cysteine and surrounding residues, but some E2s have been shown to bind Ub/UBLs at other surfaces. Members of the UbcH5A-D(UBE2D1-4) E2 sub-family, for example, bind ubiquitin at their -sheet surface, which is distant from the catalytic site (91). A broad range of residues is involved in ubiquitin-binding, but none of them is involved in catalytic mechanisms or E1/E3 binding. Its implications for ubiquitination became clear when ubiquitin was bound to UbcH5C(UBE2D3) catalytic site cysteine. When this occurs, subtle structural induce a rearrangement of the -sheet surface, leading to the self-assembly of higher order complexes composed of non-covalent UbcH5C(UBE2D3)~ubiquitin molecules. This phenomenon has been shown to be important for BRCA1-BARD1-mediated ubiquitination (91). Additional examples of noncovalent-Ub/UBL interactions are provided by Ubc2b(yeast hHR6A/B-homolog)~Ub, Ubc4(yeast UBE2D1-4-homolog)~NEDD8 (92) and Ubc13(UBE2N)-MMS2(UBE2V2)~Ub (93). The significance of Ub/UBL binding at regions other than the catalytic centre is pointed out by the observation that these noncovalent arrangements are involved in lysine usage to construct polyubiquitin chains. The most obvious example here is the Ubc13(UBE2N)-MMS2(UBE2V2) complex, a dimeric E2 that shows prefe- rential catalysis of K63-polyubiquitin chain synthesis (50). The crystal structure of Ubc13(UBE2N)-MMS2(UBE2V2)~Ub (50, 94) indicated that the MMS2(UBE2V2) subunit binds ubiquitin and positions the K63 residue of ubiquitin to construct K63-linked chains (93, 95).

78 The Shell model for E2 structure Each E2 is evolutionary restricted to mediate interactions with cognate E1s and E3s of a specific Ub/UBL-system. This selectivity prevents the inappropriate conjugation of Ub/UBLs to substrates. Crucial are surface residues on E2s that recognize cognate E1 and E3 partners. Concerning NEDD8, Huang and co-workers (96) studied evolutionary boundaries that select against the use of alternative Ub/UBLs. Using alanine scanning of surface residues within Ubc12(UBE2M), regions on the E2 were systematically identified that mediate the transfer of activated ubiquitin via the ubiquitin E1. These mutants were active in the subsequent formation of an Ubc12(UBE2M)~Ub thioester, uncovering the selective obstacles for misconjuga- tion. They also identified regions in Ubc12(UBE2M) that dictate UBL selection in the E3 binding region, indicating a molecular mechanism that does not only provide selection against the unwanted E1, but also selection against the inappropriate targets.

   90o 90o 90o

Variable Conserved

B C L1 L2 H1 I Conserved Variable Cys II III

Figure 3.5. Connecting E2 structures with function: the shell-model (A & B) Evolutionary-diverged, mutational distances of amino acids in E2 enzymes concentrate at one side of the protein. Mutational distances among residues in UbcH5B(UBE2D2) orthologs were determined using the CONSURF-algorithm and spotted on the structure of UbcH5B(UBE2D2). Mutational divergence is according to the color key. See main text for more information. (C) Schematic representation of the shell-model for E2 enzymes. The first shell contains the catalytic cysteine and is connected through the second shell with the third shell that contains structural elements that mediate selective interactions with Ub/UBL E1s and E3s. Molecular graphics images were generated using the UCSF Chimera package (http://www.cgl.ucsf.edu/chimera).

79 This observation, together with altered-specificity mutants that exhibit modified binding towards E3s, can be used to create a model of how UBC-fold architecture relates to protein function.

We propose that each UBC-fold can be divided into three shells of amino acids. The first shell consists of the heart of the enzyme and is formed by the catalytic cysteine and conserved residues surrounding that, directly involved in UBL conjugation. Residues in the first shell are solvent accessible to accommodate the activated Ub/UBL. The second shell consists of the inner core structures, partially covering the first shell. Finally, the third shell, the actual surface residues, mediates the direct interactions with other proteins, like the E1 and the E3. We would like to point out, however, that some E2s can be used both for conjugation of Ub and of ISG15 (24). Importantly, the third shell does not cover the first shell (Figure 3.5A). The second shell has a dual role. It arranges residues in the third shell for E1 and E3 recognition and it positions the first shell in optimal position to the third shell, so that the C-terminus of the activated Ub/UBL can reach the catalytic centre within the first shell. This model can now be used to explain several properties of E2 enzymes. Obviously, it tells us why specificities of E2s can be altered in a directed fashion, like the synthetic specificity of the Ubc13(UBE2N) mutant that resembles UbcH5B(UBE2D2), simply by mutating surface (third shell) residues. In addition, mutation of second shell residues also lead to changes in E3 interactions, supporting a role for the latter in establishing the correct organization of third shell residues (Van Wijk et al, manuscript in preparation). As shown previously, when changing E2-E3 interaction specificity, catalytic mechanisms of E2s remain the same, indicating that the first shell is not critically dependent on or linked to how the third shell is organized.

The model also explains specificity among the different Ub/UBL pathways. As Schulman and colleagues showed earlier (20), residues on the surface (third shell) of Ubc12(UBE2M) select against interactions with the ubiquitin E1. Adaptation of the third shell allows now for an interaction with the E1, but nothing is altered in the first shell and still the altered Ubc12(UBE2M) (with its NEDD8 catalytic centre) process and binds ubiquitin on its catalytic cysteine. This illustrates that the first and the third shell can be disconnected. More importantly, it suggests that the ability of Ubc9(UBE2I), Ubc12(UBE2M) or UbcH8(UBE2L6) to accept SUMO, NEDD8 or ISG15, respectively, is solely dependent on residues within the third shell and not those within the catalytic centre. In other words, within the catalytic centre of a UBC, there is no preference for any type of Ub or UBL. However, the third shell, forcing each Ub/UBL-specific E2 to act like one, provides essential selectivity against each Ub/UBL E1 and E3. As a consequence, the model now explains how E2 enzymes changed under evolutionary pressure. This is illustrated in Figure 3.5A were all residues in UbcH5B(UBE2D2) were scored for their mutational distance, or how divergent the amino acid sequences of UbcH5B(UBE2D2) are among its orthologs. Most variation in residues can be seen in the third shell of UbcH5B(UBE2D2), covering H1, L1 and L2 as well as the -sheet surface and C- terminal -helices. Residues surrounding the active site appear to be more conserved but

80 are still able to be solvent accessible, since the third shell does not completely cover the entire E2 enzyme. So residues in both the first and the third shell are solvent exposed, but only those present in the first shell are evolutionary more conserved. Since the third shell needs to be surface exposed to mediate interactions with other proteins, relatively more sequence variability is observed in outer regions than within the inner core (Figure 3.5B). This can be predominantly seen in H1, both loop regions and in the specialized -sheet surface. These surfaces are evolutionary adapted to mediate selective E2-specific binding of Ub/UBL E1s and E3s, whereas the first, inner shell remains largely conserved. Also, extensions to the standard UBC-fold, which fall into the third shell class, are in general not conserved and even “transplantable” between different E2 enzymes. We believe that our shell model to link functional and structural aspects of E2s is useful to their central position in Ub/UBL- conjugation pathways.

E2 enzymes in systems-wide UBL conjugation Numerous factors are influencing the correct and adequate modification of substrates in pathways that are, instead of being parallel, interconnected at either the level of the E1, the E2, the E3 ligases or substrates. Critical interactions are taking place in a spatial, temporal and highly dynamic manner. To understand the complexity of these individual and combined pathways, the role of E2 enzymes should be addressed in multi-facetted, functional studies in living systems. By regarding ubiquitination as a system that uses input (free ubiquitin) and generates output (covalent substrate modification), selective changes can be introduced and their effects on the systems’ performance can be evaluated. Mass-spectrometric approaches using cells expressing tagged ubiquitin or any other UBL (either wild-type or specifically mutated) allow for the selective characterization of conjugated substrates under denaturing conditions. This strategy was employed in a global yeast screen wherein substrates were compared that were modified with either wild-type or K11R mutant polyubiquitin chains (97). Among the substrates that became specifically modified by K11-chains, the ubiquitin E2 Ubc6 (UBE2J1/J2 in human) was found. Ubc6 has been implicated in ERAD and this study showed that K11-constructed ubiquitin chains are critical for adequate ERAD function (97). This experimental system could provide an excellent starting point to investigate the effect of individual E2 deletions on the outcome of these processes. For example, knock-down of E2 enzymes or replacing wild-type by specifically mutated E2s, would allow the systematic investigation of E2 function in multiple aspects of the system’s performance. In another recent example, Makhnevych et al (98) studied the global function of SUMO in Saccharomyces cerevisiae by combining mass-spectrometric analysis, protein-protein interaction screens and the analysis of synthetic genetic interactions. Using these combined approaches, a network of SUMO conjugation was constructed in which SUMO conjugation and deconjugation was connected with 15 biological processes. The essential nature of SUMO became clear since many processes require SUMO conjugation and this study greatly expands the SUMO interactome. By generating functional interaction maps, the role of UBL-conjugation and E2 enzymes herein could be assessed on a high-resolution scale.

81 Our knowledge of the specificity of E2 enzymes towards E3 ligases is increasing. For the majority of E2 enzymes structures have been solved, but only few E2-E3 structures have actual- ly been solved yet. Homologous residues that add specificity to E2-E3 complexes are arranged in conserved patches. Our shell model for the organization of E2s helps to modify the E2-E3 interface to create synthetic interactions (re)directing specificity of ubiquitination reactions. The exact underlying selectivity of E2-E3 interactions on a global scale remains an important aspect of ubiquitin research. Therefore, additional E2-E3 complex structures will provide further insights into the selectivity questions. With the present structural information regarding E2 and E3 enzymes, it should be possible to use molecular docking approaches to obtain high-quality E2-E3 structures. These ensembles can help to study the effects of mutations on genome-wide scales or the development of therapeutic interventions targeting or modifying E2-E3 complexes. Until recently, information concerning E2-E3 interactions remained fragmented and only limited to a selected subset of E2-E3 enzyme pairs. Two large-scale E2-E3 interactions studies now provide global information of how these interactions are organized. Markson and co- workers (99) studied E2-E3 interactions using full-length human E2 bait proteins. Initially screening for E2 interactors in fetal brain and K562 myelogenous leukemia cell yeast two- hybrid library screens, followed by array-based two-hybrid assays, they identified 569 binary E2-E3 interactions. The observed interactions compare well with deposited physical interactions and mutagenesis of interface residues in 12 high-connected RING E3 confirmed that the vast majority of these selected interactions were susceptible to changes. Also, functional analysis of E2-E3 interactions revealed that a significant proportion of the observed interactions were active in in vitro ubiquitination. Using E2 and E3 structures as templates, homology modeling was done to predict new E2-E3 interactions. These theoretic interactions are in high agreement with observed two-hybrid interactions. In parallel, we have determined a network of E2-E3 interactions, centered on individual E2 and E3 domains (100). We identified 346 high-confidence positive binary E2-E3 interactions, of which a large proportion is new. A high degree of overlap was found when the observed two-hybrid interactions were compared to literature-curated biochemically functional E2-E3 pairs. Also, when the amino acid sequences of the highly connected E2 enzymes were compared with their individual interaction pattern, high levels of similarity were observed, indicating that the domain-based E2-E3 interactions result from differences in E2 enzyme residues. Using independent pull- down analysis, a large subset of E2-E3 interactions was verified, underscoring the quality of the high-confidence nature of the interactions. In addition, by using a mutagenesis approach, two interface residues of the K63-linkage specific E2 Ubc13(UBE2N) were changed into residues of UbcH5B(UBE2D2). This change now allows novel E3 interactions for the mutated Ubc13(UBE2N), which could lead to an increase in K63-linkages on certain UbcH5B(UDE2D2)-E3 substrates. This observation underscores the importance of highly selective surface regions on E2 that mediate interactions with E3 enzymes. These two global studies revealed the genome-wide landscape of binary E2-E3 interactions in the human ubiquitin-proteasome system. Information derived from these networks can

82 be used further to introduce mutations in E2/E3 interfaces directing altered-specificities and crossing evolutionary borders between Ub/UBL conjugation pathways. Additionally, these interactions can be used as a detailed starting point for constructing high-resolution models of E2-E3 interactions in a high-throughput manner, providing systems-wide insights into underlying interaction specificities.

Conclusions and future perspectives The involvement of E2 enzymes in Ub/UBL modification pathways reflects their central roles in processes like regulating protein degradation, function and localization, thereby controlling the biology of the eukaryotic cell. The mechanisms and specificities of E2s are becoming clearer. However, complexity is also increasing since many examples exist of interplay between Ub/UBL-pathways mediated via the different E2s, manipulating the activity of central factors and modulating the outcome of individual and combined processes. For many E2 enzymes, specific functions in single or in combined Ub/UBL conjugation pathways have been described. In addition, E2 protein architecture, function and specific interactions already provided useful information regarding Ub/UBL-conjugation. Still, many important questions remain to be answered. What exactly are the catalytic mechanisms of E2s in the final stages of conjugation? How is the transfer of the activated Ub/UBL via E2s taking place; does it involve some unidentified catalytically active residues on the E2 surface or is it solely driven by optimal spatial arrangements? Based on this, crystal structures of E2-E3-substrate-Ub/UBL complexes could provide extremely valuable information by capturing all aspects of the molecular steps of conjugation. These efforts would also clarify important issues concerning the dynamics of E2 enzyme structure in relation to Ub/UBL transfer to E2 ligases.

Since the discovery of E2 enzymes as regulators of the conjugation of ubiquitin and ubiquitin-like proteins, our knowledge concerning their catalytic systems, structural characteristics and biological functions is dramatically expanding. Despite the constant stream of new insights, many important pieces of the complex puzzle of ubiquitination and the role of E2 enzymes, as critical players herein, remain missing. An important trail lies undoubtedly in studying E2 enzymes in systems-wide studies in living cells. This approach could provide highly useful information about the role of different E2s in balancing between life and death of proteins.

Acknowledgements The authors thank Loes van de Pasch and Drs. Markus Kleinschmidt and Adrien Melquiond for critical reading of the manuscript. This work was funded by support from the Nether- lands Proteomic Center (NPC) and the Netherlands Society for Scientific Research (700.50.034).

83 References 1. Dye, B. T., and Schulman, B. A. (2007) Structural mechanisms underlying posttransla- tional modification by ubiquitin-like proteins. Annu Rev Biophys Biomol Struct 36, 131-150 2. Hofmann, R. M., and Pickart, C. M. (2001) In vitro assembly and recognition of Lys-63 polyubiquitin chains. J Biol Chem 276, 27936-27943 3. Schulman, B. A., and Harper, J. W. (2009) Ubiquitin-like protein activation by E1 enzymes: the apex for downstream signalling pathways. Nat Rev Mol Cell Biol 10, 319- 331 4. Rotin, D., and Kumar, S. (2009) Physiological functions of the HECT family of ubiquitin ligases. Nat Rev Mol Cell Biol 10, 398-409 5. Deshaies, R. J., and Joazeiro, C. A. (2009) RING domain E3 ubiquitin ligases. Annu Rev Biochem 78, 399-434 6. Ardley, H. C., and Robinson, P. A. (2005) E3 ubiquitin ligases. Essays Biochem. 41, 15- 30 7. Hoeller, D., Hecker, C. M., Wagner, S., Rogov, V., Doetsch, V., and Dikic, I. (2007) E3- independent monoubiquitination of ubiquitin-binding proteins. Mol Cell 26, 891-898 8. Golebiowski, F., Matic, I., Tatham, M. H., Cole, C., Yin, Y., Nakamura, A., Cox, J., Bar- ton, G. J., Mann, M., and Hay, R. T. (2009) System-wide changes to SUMO modification in response to heat shock. Sci Signal 2, ra24 9. Pickart, C. M., and Fushman, D. (2004) Polyubiquitin chains: polymeric protein signals. Curr Opin Chem Biol 8, 610-616 10. Ravid, T., and Hochstrasser, M. (2008) Diversity of degradation signals in the ubiquitin-proteasome system. Nat Rev Mol Cell Biol 9, 679-690 11. Love, K. R., Catic, A., Schlieker, C., and Ploegh, H. L. (2007) Mechanisms, biology and inhibitors of deubiquitinating enzymes. Nat Chem Biol 3, 697-705 12. Burroughs, A. M., Jaffee, M., Iyer, L. M., and Aravind, L. (2008) Anatomy of the E2 ligase fold: Implications for enzymology and evolution of ubiquitin/Ub-like protein conjugation. J Struct Biol 162, 205-218 13. Zou, W. G., Papov, V., Malakhova, O., Kim, K., Dao, C., Li, J., and Zhang, D. E. (2005) ISG15 modification of ubiquitin E2 Ubc13 disrupts its ability to form thioester bond with ubiquitin. Biochem Biophys Res Comm 336, 61-68 14. Tatham, M. H., Geoffroy, M. C., Shen, L., Plechanovova, A., Hattersley, N., Jaffray, E. G., Palvimo, J. J., and Hay, R. T. (2008) RNF4 is a poly-SUMO-specific E3 ubiquitin ligase required for arsenic-induced PML degradation. Nat Cell Biol 10, 538-546 15. Schwarz, S. E., Matuschewski, K., Liakopoulos, D., Scheffner, M., and Jentsch, S. (1998) The ubiquitin-like proteins SMT3 and SUMO-1 are conjugated by the UBC9 E2 enzyme. Proc Natl Acad Sci U S A 95, 560-564 16. Melchior, F. (2000) SUMO--nonclassical ubiquitin. Annu Rev Cell Dev Biol 16, 591-626 17. Lin, D., Tatham, M. H., Yu, B., Kim, S., Hay, R. T., and Chen, Y. (2002) Identification of a substrate recognition site on Ubc9. J Biol chem 277, 21740-21748 18. Mohideen, F., Capili, A. D., Bilimoria, P. M., Yamada, T., Bonni, A., and Lima, C. D. (2009) A molecular basis for phosphorylation-dependent SUMO conjugation by the E2 UBC9. Nat Struct Mol Biol 16, 945-952 19. Kawakami, T., Chiba, T., Suzuki, T., Iwai, K., Yamanaka, K., Minato, N., Suzuki, H., Shimbara, N., Hidaka, Y., Osaka, F., Omata, M., and Tanaka, K. (2001) NEDD8 recruits E2-ubiquitin to SCF E3 ligase. EMBO J 20, 4003-4012

84 20. Huang, D. T., Ayrault, O., Hunt, H. W., Taherbhoy, A. M., Duda, D. M., Scott, D. C., Borg, L. A., Neale, G., Murray, P. J., Roussel, M. F., and Schulman, B. A. (2009) E2- RING expansion of the NEDD8 cascade confers specificity to cullin modification. Mol Cell 33, 483-495 21. Staub, O. (2004) Ubiquitylation and isgylation: overlapping enzymatic cascades do the job. Sci STKE 2004, pe43 22. Durfee, L. A., Kelley, M. L., and Huibregtse, J. M. (2008) The basis for selective E1-E2 interactions in the ISG15 conjugation system. J Biol Chem 283, 23895-23902 23. Zhao, C., Beaudenon, S. L., Kelley, M. L., Waddell, M. B., Yuan, W., Schulman, B. A., Huibregtse, J. M., and Krug, R. M. (2004) The UbcH8 ubiquitin E2 enzyme is also the E2 enzyme for ISG15, an IFN-alpha/beta-induced ubiquitin-like protein. Proc Natl Acad Sci U S A 101, 7578-7582 24. Takeuchi, T., Iwahari, S., Saeki, Y., Sasajima, H., and Yokosawa, H. (2005) Link between the ubiquitin conjugation system and the ISG15 conjugation system: ISG15 conjugation to the UbcH6 ubiquitin E2 enzyme. J Biochem 138, 711-719 25. Geng, J., and Klionsky, D. J. (2008) The Atg8 and Atg12 ubiquitin-like conjugation systems in macroautophagy. 'Protein modifications: beyond the usual suspects' review series. EMBO Rep 9, 859-864 26. Rossman, T. G., Visalli, M. A., and Komissarova, E. V. (2003) fau and its ubiquitin-like domain (FUBI) transforms human osteogenic sarcoma (HOS) cells to anchorage- independence. Oncogene 22, 1817-1821 27. Hipp, M. S., Kalveram, B., Raasi, S., Groettrup, M., and Schmidtke, G. (2005) FAT10, a ubiquitin-independent signal for proteasomal degradation. Mol Cell Biol 25, 3483- 3491 28. Furukawa, K., Mizoshima, N., Noda, T., and Ohsumi, Y. (2000) A protein conjugation system in yeast with homology to biosynthetic enzyme reaction of prokaryotes. J Biol Chem 275, 7462-7465 29. Hochstrasser, M. (2009) Origin and function of ubiquitin-like proteins. Nature 458, 422-429 30. Pearce, M. J., Mintseris, J., Ferreyra, J., Gygi, S. P., and Darwin, K. H. (2008) Ubiquitin- like protein involved in the proteasome pathway of Mycobacterium tuberculosis. Science 322, 1104-1107 31. Striebel, F., Imkamp, F., Sutter, M., Steiner, M., Mamedov, A., and Weber-Ban, E. (2009) Bacterial ubiquitin-like modifier Pup is deamidated and conjugated to substrates by distinct but homologous enzymes. Nat Struct Mol Biol 16, 647-651 32. Iyer, L. M., Burroughs, A. M., and Aravind, L. (2008) Unraveling the biochemistry and provenance of pupylation: a prokaryotic analog of ubiquitination. Biology Direct 3, 45 33. Michelle, C., Vourc'h, P., Mignon, L., and Andres, C. R. (2009) What Was the Set of Ubiquitin and Ubiquitin-Like Conjugating Enzymes in the Eukaryote Common Ances- tor? J Mol Evol 68, 616-628 34. Winn, P. J., Religa, T. L., Battey, J. N., Banerjee, A., and Wade, R. C. (2004) Determi- nants of functionality in the ubiquitin conjugating enzyme family. Structure 12, 1563- 1574 35. Seufert, W., and Jentsch, S. (1990) Ubiquitin-conjugating enzymes UBC4 and UBC5 mediate selective degradation of short-lived and abnormal proteins. EMBO J. 9, 543- 550

85 36. Bartke, T., Pohl, C., Pyrowolakis, G., and Jentsch, S. (2004) Dual role of BRUCE as an antiapoptotic IAP and a chimeric E2/E3 ubiquitin ligase. Mol Cell 14, 801-811 37. Summers, M. K., Pan, B., Mukhyala, K., and Jackson, P. K. (2008) The unique N terminus of the UbcH10 E2 enzyme controls the threshold for APC activation and enhances checkpoint regulation of the APC. Mol Cell 31, 544-556 38. Goebl, M. G., Yochem, J., Jentsch, S., Mcgrath, J. P., Varshavsky, A., and Byers, B. (1988) The Yeast-Cell Cycle Gene Cdc34 Encodes a Ubiquitin-Conjugating Enzyme. Science 241, 1331-1335 39. Skowyra, D., Craig, K. L., Tyers, M., Elledge, S. J., and Harper, J. W. (1997) F-box proteins are receptors that recruit phosphorylated substrates to the SCF ubiquitin-ligase complex. Cell 91, 209-219 40. Coccetti, P., Tripodi, F., Tedeschi, G., Nonnis, S., Marin, O., Fantinato, S., Cirulli, C., Vanoni, M., and Alberghina, L. (2008) The CK2 phosphorylation of catalytic domain of Cdc34 modulates its activity at the G(1) to S transition in Saccharomyces cerevisiae. Cell Cycle 7, 1391-1401 41. Sadowski, M., Mawson, A., Baker, R., and Sarcevic, B. (2007) Cdc34 C-terminal tail phosphorylation regulates Skp1/cullin/F-box (SCF)-mediated ubiquitination and cell cycle progression. Biochem J 405, 569-581 42. Klemperer, N. S., Berleth, E. S., and Pickart, C. M. (1989) A novel, arsenite-sensitive E2 of the ubiquitin pathway: purification and properties. Biochemistry 28, 6035-6041 43. Berleth, E. S., and Pickart, C. M. (1996) Mechanism of ubiquitin conjugating enzyme E2-230K: catalysis involving a thiol relay? Biochemistry 35, 1664-1671 44. Hao, Y., Sekine, K., Kawabata, A., Nakamura, H., Ishioka, T., Ohata, H., Katayama, R., Hashimoto, C., Zhang, X., Noda, T., Tsuruo, T., and Naito, M. (2004) Apollon ubiquiti- nates SMAC and caspase-9, and has an essential cytoprotection function. Nat Cell Biol 6, 849-860 45. Pohl, C., and Jentsch, S. (2008) Final stages of cytokinesis and midbody ring formation are controlled by BRUCE. Cell 132, 832-845 46. Hofmann, R. M., and Pickart, C. M. (1999) Noncanonical MMS2-encoded ubiquitin- conjugating enzyme functions in assembly of novel polyubiquitin chains for DNA repair. Cell 96, 645-653 47. Kim, H. T., Kim, K. P., Lledias, F., Kisselev, A. F., Scaglione, K. M., Skowyra, D., Gygi, S. P., and Goldberg, A. L. (2007) Certain pairs of ubiquitin-conjugating enzymes (E2s) and ubiquitin-protein ligases (E3s) synthesize nondegradable forked ubiquitin chains containing all possible isopeptide linkages. J Biol Chem 282, 17375-17386 48. Kim, H. T., Kim, K. P., Uchiki, T., Gygi, S. P., and Goldberg, A. L. (2009) S5a promotes protein degradation by blocking synthesis of nondegradable forked ubiquitin chains. EMBO J 28, 1867-1877 49. Saeki, Y., Kudo, T., Sone, T., Kikuchi, Y., Yokosawa, H., Toh-e, A., and Tanaka, K. (2009) Lysine 63-linked polyubiquitin chain may serve as a targeting signal for the 26S proteasome. EMBO J 28, 359-371 50. VanDemark, A. P., Hofmann, R. M., Tsui, C., Pickart, C. M., and Wolberger, C. (2001) Molecular insights into polyubiquitin chain assembly: crystal structure of the Mms2/Ubc13 heterodimer. Cell 105, 711-720 51. Andersen, P. L., Zhou, H., Pastushok, L., Moraes, T., McKenna, S., Ziola, B., Ellison, M. J., Dixit, V. M., and Xiao, W. (2005) Distinct regulation of Ubc13 functions by the two ubiquitin-conjugating enzyme variants Mms2 and Uev1A. J Cell Biol 170, 745-755

86 52. Kloor, M., Bork, P., Duwe, A., Klaes, R., Doeberitz, M. V., and Ridder, R. (2002) Identi- fication and characterization of UEV3, a human cDNA with similarities to inactive E2 ubiquitin-conjugating enzymes. Biochim Biophys Acta 1579, 219-224 53. Xu, L., Sowa, M. E., Chen, J., Li, X., Gygi, S. P., and Harper, J. W. (2008) An FTS/Hook/p107(FHIP) Complex Interacts with and Promotes Endosomal Clustering by the Homotypic Vacuolar Protein Sorting Complex. Mol Biol Cell 19, 5059-5071 54. Lesche, R., Peetz, A., vanderHoeven, F., and Ruther, U. (1997) Ft1, a novel gene related to ubiquitin-conjugating enzymes, is deleted in the Fused toes mouse mutation. Mamm Genome 8, 879-883 55. Remy, I., and Michnick, S. W. (2004) Regulation of apoptosis by the Ft1 protein, a new modulator of protein kinase B/Akt. Mol Cell Biol 24, 1493-1504 56. Koonin, E. V., and Abagyan, R. A. (1997) TSG101 may be the prototype of a class of dominant negative ubiquitin regulators. Nat Genet 16, 330-331 57. Ruland, J., Sirard, C., Elia, A., MacPherson, D., Wakeham, A., Li, L., de la Pompa, J. L., Cohen, S. N., and Mak, T. W. (2001) p53 accumulation, defective cell proliferation, and early embryonic lethality in mice lacking tsg101. Proc Natl Acad Sci U S A 98, 1859-1864 58. de Pril, R., Fischer, D. F., Roos, R. A., and van Leeuwen, F. W. (2007) Ubiquitin- conjugating enzyme E2-25K increases aggregate formation and cell death in polyglutamine diseases. Mol Cell Neurosci 34, 10-19 59. Dikic, I., Wakatsuki, S., and Walters, K. J. (2009) Ubiquitin-binding domains - from structures to functions. Nat Rev Mol Cell Biol 10, 659-671 60. Haldeman, M. T., Xia, G., Kasperek, E. M., and Pickart, C. M. (1997) Structure and function of ubiquitin conjugating enzyme E2-25K: the tail is a core-dependent activity element. Biochemistry 36, 10526-10537 61. Plafker, S. M., Plafker, K. S., Weissman, A. M., and Macara, I. G. (2004) Ubiquitin charging of human class III ubiquitin-conjugating enzymes triggers their nuclear import. J Cell Biol 167, 649-659 62. Biederer, T., Volkwein, C., and Sommer, T. (1997) Role of Cue1p in ubiquitination and degradation at the ER surface. Science 278, 1806-1809 63. Chen, B., Mariano, J., Tsai, Y. C., Chan, A. H., Cohen, M., and Weissman, A. M. (2006) The activity of a human endoplasmic reticulum-associated degradation E3, gp78, requires its Cue domain, RING finger, and an E2-binding site. Proc Natl Acad Sci U S A 103, 341-346 64. Ravid, T., and Hochstrasser, M. (2007) Autoregulation of an E2 enzyme by ubiquitin- chain assembly on its catalytic residue. Nat Cell Biol 9, 422-427 65. Li, W., Tu, D., Brunger, A. T., and Ye, Y. (2007) A ubiquitin ligase transfers preformed polyubiquitin chains from a conjugating enzyme to a substrate. Nature 446, 333-337 66. Takeuchi, T., and Yokosawa, H. (2005) ISG15 modification of Ubc13 suppresses its ubiquitin-conjugating activity. Biochem Biophys Res Comm 336, 9-13 67. Pichler, A., Knipscheer, P., Oberhofer, E., van Dijk, W. J., Korner, R., Olsen, J. V., Jentsch, S., Melchior, F., and Sixma, T. K. (2005) SUMO modification of the ubiquitin- conjugating enzyme E2-25K. Nat Struct Mol Biol 12, 264-269 68. Lois, L. M., and Lima, C. D. (2005) Structures of the SUMO E1 provide mechanistic insights into SUMO activation and E2 recruitment to E1. EMBO J 24, 439-451

87 69. Wang, J., Hu, W., Cai, S., Lee, B., Song, J., and Chen, Y. (2007) The intrinsic affinity between E2 and the Cys domain of E1 in ubiquitin-like modifications. Mol Cell 27, 228- 237 70. Walden, H., Podgorski, M. S., Huang, D. T., Miller, D. W., Howard, R. J., Minor, D. L., Jr., Holton, J. M., and Schulman, B. A. (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 71. Huang, D. T., Hunt, H. W., Zhuang, M., Ohi, M. D., Holton, J. M., and Schulman, B. A. (2007) Basis for a ubiquitin-like protein thioester switch toggling E1-E2 affinity. Na- ture 445, 394-398 72. Huang, D. T., Miller, D. W., Mathew, R., Cassell, R., Holton, J. M., Roussel, M. F., and Schulman, B. A. (2004) A unique E1-E2 interaction required for optimal conjugation of the ubiquitin-like protein NEDD8. Nat Struct Mol Biol 11, 927-935 73. Huang, D. T., Paydar, A., Zhuang, M., Waddell, M. B., Holton, J. M., and Schulman, B. A. (2005) Structural basis for recruitment of Ubc12 by an E2 binding domain in NEDD8's E1. Mol Cell 17, 341-350 74. Pitluk, Z. W., McDonough, M., Sangan, P., and Gonda, D. K. (1995) Novel CDC34 (UBC3) ubiquitin-conjugating enzyme mutants obtained by charge-to-alanine scanning mutagenesis. Mol Cell Biol 15, 1210-1219 75. Sullivan, M. L., and Vierstra, R. D. (1991) Cloning of a 16-Kda Ubiquitin Carrier Protein from Wheat and Arabidopsis-Thaliana - Identification of Functional Domains by Invi- tro Mutagenesis. J Biol Chem 266, 23878-23885 76. McKenna, S., Spyracopoulos, L., Moraes, T., Pastushok, L., Ptak, C., Xiao, W., and Elli- son, M. J. (2001) Noncovalent interaction between ubiquitin and the human DNA repair protein Mms2 is required for Ubc13-mediated polyubiquitination. J Biol Chem 276, 40120-40126 77. Lee, I., and Schindelin, H. (2008) Structural insights into E1-catalyzed ubiquitin activation and transfer to conjugating enzymes. Cell 134, 268-278 78. Huang, L., Kinnucan, E., Wang, G., Beaudenon, S., Howley, P. M., Huibregtse, J. M., and Pavletich, N. P. (1999) Structure of an E6AP-UbcH7 complex: insights into ubiquitination by the E2-E3 enzyme cascade. Science 286, 1321-1326 79. Zheng, N., Wang, P., Jeffrey, P. D., and Pavletich, N. P. (2000) Structure of a c-Cbl- UbcH7 complex: RING domain function in ubiquitin-protein ligases. Cell 102, 533-539 80. Eletr, Z. M., and Kuhlman, B. (2007) Sequence determinants of E2-E6AP binding affinity and specificity. J Mol Biol 369, 419-428 81. Meacham, G. C., Patterson, C., Zhang, W., Younger, J. M., and Cyr, D. M. (2001) The Hsc70 co-chaperone CHIP targets immature CFTR for proteasomal degradation. Nat Cell Biol 3, 100-105 82. Jiang, J., Ballinger, C. A., Wu, Y., Dai, Q., Cyr, D. M., Hohfeld, J., and Patterson, C. (2001) CHIP is a U-box-dependent E3 ubiquitin ligase: identification of Hsc70 as a target for ubiquitylation. J Biol Chem 276, 42938-42944 83. Xu, Z., Kohli, E., Devlin, K. I., Bold, M., Nix, J. C., and Misra, S. (2008) Interactions between the quality control ubiquitin ligase CHIP and ubiquitin conjugating enzymes. BMC Struct. Biol. 8, 26 84. Qian, S. B., McDonough, H., Boellmann, F., Cyr, D. M., and Patterson, C. (2006) CHIP- mediated stress recovery by sequential ubiquitination of substrates and Hsp70. Na- ture 440, 551-555

88 85. Zhang, M., Windheim, M., Roe, S. M., Peggie, M., Cohen, P., Prodromou, C., and Pearl, J. H. (2005) Chaperoned ubiquitylation-crystal structures of the CHIP U box E3 ubiquitin ligase and a CHIP-Ubc13-Uev1a complex. Mol Cell 20, 525-538 86. Christensen, D. E., Brzovic, P. S., and Klevit, R. E. (2007) E2-BRCA1 RING interactions dictate synthesis of mono- or specific polyubiquitin chain linkages. Nat Struct Mol Biol 14, 941-948 87. Winkler, G. S., Albert, T. K., Dominguez, C., Legtenberg, Y. I., Boelens, R., and Tim- mers, H. T. (2004) An altered-specificity ubiquitin-conjugating enzyme/ubiquitin- protein ligase pair. J Mol Biol 337, 157-165 88. Dominguez, C., Bonvin, A. M., Winkler, G. S., van Schaik, F. M., Timmers, H. T., and Boelens, R. (2004) Structural model of the UbcH5B/CNOT4 complex revealed by combining NMR, mutagenesis, and docking approaches. Structure 12, 633-644 89. Eletr, Z. M., Huang, D. T., Duda, D. M., Schulman, B. A., and 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 90. Hamilton, K. S., Ellison, M. J., Barber, K. R., Williams, R. S., Huzil, J. T., McKenna, S., Ptak, C., Glover, M., and Shaw, G. S. (2001) Structure of a conjugating enzyme- ubiquitin thiolester intermediate reveals a novel role for the ubiquitin tail. Structure 9, 897-904 91. Brzovic, P. S., Lissounov, A., Christensen, D. E., Hoyt, D. W., and Klevit, R. E. (2006) A UbcH5/ubiquitin noncovalent complex is required for processive BRCA1-directed ubiquitination. Mol Cell 21, 873-880 92. Sakata, E., Yamaguchi, Y., Miyauchi, Y., Iwai, K., Chiba, T., Saeki, Y., Matsuda, N., Ta- naka, K., and Kato, K. (2007) Direct interactions between NEDD8 and ubiquitin E2 conjugating enzymes upregulate cullin-based E3 ligase activity. Nat Struct Mol Biol 14, 167-168 93. Eddins, M. J., Carlile, C. M., Gomez, K. M., Pickart, C. M., and Wolberger, C. (2006) Mms2-Ubc13 covalently bound to ubiquitin reveals the structural basis of linkage- specific polyubiquitin chain formation. Nat Struct Mol Biol 13, 915-920 94. Moraes, T. F., Edwards, R. A., McKenna, S., Pastushok, L., Xiao, W., Glover, J. N., and Ellison, M. J. (2001) Crystal structure of the human ubiquitin conjugating enzyme complex, hMms2-hUbc13. Nat Struct Mol Biol 8, 669-673 95. Lewis, M. J., Saltibus, L. F., Hau, D. D., Xiao, W., and Spyracopoulos, L. (2006) Struc- tural basis for non-covalent interaction between ubiquitin and the ubiquitin conjugating enzyme variant human MMS2. J Biomol NMR 34, 89-100 96. Huang, D. T., Zhuang, M., Ayrault, O., and Schulman, B. A. (2008) Identification of conjugation specificity determinants unmasks vestigial preference for ubiquitin within the NEDD8 E2. Nat Struct Mol Biol 97. Xu, P., Duong, D. M., Seyfried, N. T., Cheng, D. M., Xie, Y., Robert, J., Rush, J., Hoch- strasser, M., Finley, D., and Peng, J. (2009) Quantitative Proteomics Reveals the Func- tion of Unconventional Ubiquitin Chains in Proteasomal Degradation. Cell 137, 133- 145 98. Makhnevych, T., Sydorskyy, Y., Xin, X. F., Srikumar, T., Vizeacoumar, F. J., Jerarn, S. M., Li, Z. J., Bahr, S., Andrews, B. J., Boone, C., and Raught, B. (2009) Global Map of SUMO Function Revealed by Protein-Protein Interaction and Genetic Networks. Mol Cell 33, 124-135

89 99. Markson, G., Kiel, C., Hyde, R., Brown, S., Charalabous, P., Bremm, A., Semple, J., Woodsmith, J., Duley, S., Salehi-Ashtiani, K., Vidal, M., Komander, D., Serrano, L., Lehner, P., and Sanderson, C. M. (2009) Analysis of the human E2 conjugating enzymes protein interaction network. Genome Res Aug 26 100. van Wijk, S. J., de Vries, S. J., Kemmeren, P., Huang, A., Boelens, R., Bonvin, A. M., and Timmers, H. T. (2009) A comprehensive analysis of E2-RING E3 interactions of the human ubiquitin-proteasome system. Mol Syst Biol 5: 295

Supplementary Table 1 Multi-species overview of E2 enzymes Saccharomyces cerevisiae Saccharomyces ? ? UBC11 ? ? ? ? ? ubc-21 UBC1

13 UBC13

- Caenorhabditis elegans Caenorhabditis Drosophila melanogaster Drosophila

bendless ubc Mus musculus Mus

7330 7331 7333

Entrez Gene ID Gene Entrez

317770 326105 246719 440406 606552 170556 619457 286480 392239 283556 606551 UniProt ID UniProt Q8IWF7 340561 P60604 7327 Ube2g2 courtless ubc-14 UBC7

Ensembl ID Ensembl

ENSG00000184787

RefSeq ID RefSeq

NM_001012989 ENSG00000102069 Q5JXB2 389898 Ube2nl NG_001583 NG_002490 ENSG00000131982 Accession Numbers Accession

AB051521 NM_022066 ENSG00000175931 Q9C0C9 63893 Ube2o CG10254 ubc-17 ?

Chromosome 16q12.2 AK0233202p22.3 NM_022476 AF265555 NM_016252 ENSG00000166971 Q9H8T0 ENSG00000115760 Q9NR09 64400 Aktip 57448 Birc6 crossbronx ? bruce ? Xq27.3 19p13.3 L22005 NM_004359 ENSG00000099804 P49427 997 Cdc34 CG7656 ubc-3 CDC34 Xq24 AK223045 NM_003336 ENSG00000077721 P49459 7319 Ube2a UbcD6 ubc-1 RAD6 14q11.2 5q31.1 M74525 NM_003337 ENSG00000119048 P63146 7320 Ube2b UbcD6 ubc-1 RAD6 20q13.12 U73379 NM_007019 ENSG00000175063 O00762 11065 Ube2c vihar ? 17q25.1 14q12 1q2215q24.2 AJ243666 BC017708 NM_173469 NM_017582 ENSG00000140367 ENSG00000160714 Q8WVN8 Q7Z7E8 92912 Ube2q2 55585 Ube2q1 CG2924 CG2924 ubc-25 ubc-25 ? ? 5q31.3 L40146 NM_181838 ENSG00000131508 P62837 7322 Ube2d2 effete let-70 UBC4 17p11.2 17p11.2 19q13.43 BC004236 NM_014501 ENSG00000108106 Q16763 27338 Ube2s CG8188 ? 4q24 U39318 NM_181893 ENSG00000109332 P61077 7323 Ube2d3 effete let-70 UBC4 1q32.1 AF161499 NM_014176 ENSG00000077152 Q9NPD8 29089 Ube2t ? ? 7p13 BC004104 NM_015983 ENSG00000078967 Q9Y2X8 51619 Gm15361 effete let-70 UBC4 20p12.3 AL121755 1p31.3 BC029895 NM_152489 ENSG00000177414 Q5VVX9 148581 Ube2u ? ? Xq21.1 BC040290 20q13.2 U39360 NM_021988 ENSG00000124208 Q9UKL1 7335 Ube2v1 Uev1A uev-1 MMS2 3p24.2 X92963 NM_003341 ENSG00000170142 P51965 7324 Ube2e1 UbcD2 let-70 UBC5 20p13 AL117334 3p24.2 AK057886 NM_152653 ENSG00000182247 Q96LR5 7325 Ube2e2 UbcD2 let-70 UBC5 8q11.21 X98091 NM_003350 ENSG00000169139 Q15819 7336 Ube2v2 Uev1A uev-1 MMS2 11p15.1 AF503350 NM_018314 ENSG00000151116 Q8IX04 55293 2q31.3 AB017644 NM_006357 ENSG00000170035 Q969T4 10477 Ube2e3 UbcD2 let-70 UBC5 8q21.11 AK001873 NM_001001481 ENSG00000104343 Q96B02 55284 Ube2w CG7220 ubc-16 ? Xp22.2 AF136176 NG_004721 ENSG00000170035 1p21.3 AC092393 17q21.31 AA843379 NM_023079.3 ENSG00000159202 Q9H832 65264 Ube2z ? ? 2q37.3 BC010549 NM_080678 ENSG00000184182 Q969M7 140739 Ube2f CG7375 ubc-12 UBC12 17p13.2 BC026288 NM_003342 ENSG00000132388 P62253 7326 Ube2g1 CG40045 CG9602 UBC7 21q22.3 BC008351 NM_182688 7q32 BC006277 NM_003344 ENSG00000186591 P62256 7328 Ube2h Ubc-E2H ubc-8 UBC8 8q21 16p13.3 D45050 NM_003345 ENSG00000103275 P63279 7329 Ube2i lesswright ubc-9 UBC9 6q16.1 AJ245898 NM_016021 ENSG00000198833 Q9Y385 51465 Ube2j1 CG5823 ubc-6 UBC6 1p36.33 AF296658 NM_058167 ENSG00000160087 Q8N2K1 118424 Ube2j2 CG5823 ubc-26 UBC6 4p14 U58522 NM_005339 ENSG00000078140 P61086 3093 Ube2k UbcD4 ubc-20, 14q24.3 S81005 NG_003101 12q12 22q11.2 AJ000519 NM_198157 ENSG00000185651 P68036 7332 Ube2l3 UbcD10 ubc-18 ? 19q13.2 Y09515 13q12.3 AL161893 11q12.1 AF031141 NM_004223 ENSG00000156587 O14933 9246 Ube2l6 UbcD10 ? 14q22.3 19q13.43 AB012191 NM_003969 ENSG00000130725 P61081 9040 Ube2m CG7375 ubc-12 UBC12 16p11.2 CR599143 12q22 D83004 NM_003348 ENSG00000177889 P61088 7334 Ube2n

MGC10481 9p11.2 AK000426 NM_017811 ENSG00000107341 Q712K3 54926 Ube2r2 CG7656 ubc-3 CDC34

E2(17)KB1 10q21.1 BC015997 NM_003338 ENSG00000072401 P51668 7321 Ube2d1 effete let-70 UBC4

MMS2

CROC1

UBC3 KIAA1734

MGC8489

NCUBE1

UBC4/5, FLJ20419,

EDPF-1,

UEV1A,

FLJ13855

UBC13, UBE2R1,

NICE-5 CGI-76,

AKTIP FLJ12878, UBCH5, UBC4

UBC12

CDC34B, FLJ25157 NCUBE2

CROC-1, DDVit-1, L-UBC

UBC8

HHR6A HHR6B

HOYS7,

UEV3

FTS,

Aliases

BRUCE FT1, E2-CDC34, UBC2, UBC2, E2-230K, UBCH10 DKFZp762C143 UBC3B, UbcH5B, E2-EPF UbcH5C HSPC150 HBUCE1 MGC35130 MGC42638 UbcH6 dJ687F11.3 UbcH8, UEV-2, Attp, UbcH9 FLJ11011 UbcM2 USE1, NCE2 UBC7 UBCH, UBC9 HSPC153, Ubc6p, UBEL1, UBCH7 UBCH8 hUbc12, UbcH-ben, Previous Symbols Previous UBE2Q PRO3094, SFT UbcH5A, UBE2V UEV-1, UBE2G HIP2 HYPG UBE2L5 UBE2L7 UBCH7N3 pseudogene

1 yeast) yeast) yeast)

yeast) yeast) yeast)

yeast) yeast) yeast) yeast) yeast) yeast)

yeast) yeast) yeast) 1 2

pseudogene yeast)

1 1

1 2

homolog, homolog, homolog, homolog, homolog, homolog,

homolog, homolog, homolog, homolog, homolog, homolog, pseudogene

member member homolog)

homolog) homolog, homolog, homolog,

1 2

homolog,

domains

(UBC4/5 (UBC4/5 (UBC4/5 (putative) pseudogene (UBC7 (UBC7 (UBC6 (UBC6 (UBC4/5 (UBC4/5 (UBC4/5 pseudogene

pseudogene pseudogene pseudogene pseudogene pseudogene

(putative) pseudogene

(UBC12 pseudogene pseudogene family family

(UBC13 (RAD6 1 (putative) 2 3 4 5 N-terminal 1 2 (UBC8 (UBC8

(RAD6 pseudogene 2 (UBC1

pseudogene pseudogene (putative) 1 2 3 4 (putative)

1 2 3 4 5 6 7

(UBC9 J1 J2 6

variant variant variant cerevisiae) Q Q

E2N-like E2N E2A E2B E2O E2C E2C E2D E2 E2 E2R E2D E2S E2D E2S E2S E2T E2D E2U E2D E2D E2 E2E E2 E2E E2 E2E E2W E2E E2W E2Z E2F E2G E2G E2H E2H E2I E2, E2, E2K E2L E2L E2L E2L E2L E2L E2L E2M E2M E2N (S.

dehyrogenase

homolog

protein repeat-containing

cycle

IAP

lactate/malate

and

division

Approved Name Approved

ubiquitin-conjugating baculoviral cell AKT interacting ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating UEV ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating ubiquitin-conjugating Approved Symbol Approved UBE2NL BIRC6 CDC34 AKTIP UBE2NP1 UBE2A UBE2B UBE2O UBE2C UBE2CP1 UBE2D1 UBE2Q2 UBE2R2 UBE2Q1 UBE2D2 UBE2S UBE2D3 UBE2SP2 UBE2T UBE2SP1 UBE2D4 UBE2U UBE2D5P UBE2DNL UBE2V1 UBE2E1 UBE2V1P1 UBE2E2 UBE2V2 UBE2E3 UEVLD UBE2W UBE2E4P UBE2WP UBE2Z UBE2F UBE2G1 UBE2G2 UBE2H UBE2HP UBE2I UBE2J1 UBE2J2 UBE2K UBE2L1 UBE2L2 UBE2L3 UBE2L4 UBE2L5P UBE2L6 UBE2L7P UBE2M UBE2MP1 UBE2N

Chapter 4

A butterfly effect in protein-protein interactions: how minimal E2 sequence differences control global RING E3 interaction selectivity

Manuscript in preparation

93 A butterfly effect in protein-protein interactions: how minimal E2 sequence differences control global RING E3 interaction selectivity

Sjoerd J. L. van Wijk1,3, Adrien S. J. Melquiond2,3, H. Th. Marc Timmers1,4 and Alexandre M. J. J. Bonvin2,4 1Department of Physiological Chemistry, Division of Biomedical Genetics, University Medical Center Utrecht, Universiteits- weg 100, 3584 CG Utrecht, the Netherlands 2Department of NMR Spectroscopy, Bijvoet Center for Biomolecular Research, Utrecht University, Padualaan 8, 3584 CH Utrecht, the Netherlands.

3 these authors contributed equally 4 corresponding authors

Summary Sequence and structural homologies between proteins are primary predictors for function. We challenge this paradigm by reporting the astonishing effect of a simple and conserved aspartic to glutamic acid substitution on selective interactions in the ubiquitination pathway. The versatile UBC-fold of E2 ubiquitin-conjugating enzymes is at the center of this pathway as this domain channels ubiquitin from the activating E1 to the E3 ligase enzymes. We mapped the difference in E2-E3 interaction potential of two E2s, UbcH6(UBE2E1) and UbcH8(UBE2E2), to a single glutamic to aspartic acid substitution in loop 1 of the UBC-fold. By combining mutagenesis and molecular modeling with high-throughput analysis of protein-protein interactions we decipher how such a small change, a methylene group, can have such dramatic effects on the interaction potential of these enzymes, what we refer to as the “butterfly effect”. We show that a dynamic network of intra-molecular salt bridges controls the availability of an aspartic acid of UbcH8(UBE2E2) by shifting its equilibrium between a bridged and an open form critical for interaction with E3s. Mutation of single network components displaces this equilibrium resulting in a switch in E3 interaction profile. Together, our findings raise new challenges for bioinformatic and experimental studies of protein-protein interactions.

Results and Discussion The human genome encodes hundreds of E3 ligases characterized by conserved patterns of histidine and/or cysteine residues that are able to coordinate two Zn2+-ions (RING-finger domain) (1-3). When folded in a typical “cross-brace”-structure, RING-domains serve as binding platforms for E2 enzymes that dock onto RING-domains via their UBC-fold (1). Inte- ractions between E2-E3 enzymes link the upstream activation of ubiquitin by ubiquitin- activating enzymes (E1s) to downstream substrate conjugation mediated by ubiquitin protein ligases (E3s) (4). Although both UBC- and RING-domains show high levels of homology,

94 A

UbcH6(UBE2E1) UbcH8(UBE2E2 ) UbcH8(UBE2E2) UbcH8(UBE2E2) UbcH8(UBE2E2) E66D D113E S111T

Figure 4.1. Conserved key residues in two homologous E2 enzymes mediate global RING-type E3 interaction selectivity (A) Sequence alignments of full-length human UbcH6(UBE2E1), UbcH8(UBE2E2) and UbcH9(UBE2E3). Secondary structure elements are depicted schematically and are derived from the corresponding PDB-files (UbcH6(UBE2E1): 3bzh and UbcH8(UBE2E2): 1y6l). Helices are shown as purple cylinders, sheets as green rec- tangles and loops as black lines. E3-interface specifying regions helix 1 (H1), loop 1 (L1) and loop 2 (L2) are indicated. Residues within this predicted interface that differ between the E2s are indicated with arrows. Sequences absent in the LexA-E2 bait proteins are transparent. (B) E3 interaction profiles of wild-type and mutant UbcH6(UBE2E1) and UbcH8(UBE2E2). Mutagenesis was performed on LexA-E2 constructs, transformed in EGY48α, mated against the 250 human RING-finger domain arrays and transferred to X-Gal-supplemented selection medium. Pictures were taken 72 hours after transfer. See Materials section for more details.

95 interactions between E2s and E3s are highly selective (5-7). Experimental structures of c-Cbl- UbcH7(UBE2L3) (8), CHIP-UbcH5A(UBE2D1) (9), CHIP-Ubc13(UBE2N)/UEV-1A(UBE2V1) (10) and CNOT4- UbcH5B(UBE2D2) (11) designate residues present in helix 1 (H1), loop 1 (L1) and loop 2 (L2) in the UBC-fold as RING-interacting regions (7).

A reoccurring theme in protein biochemistry is the correlation between sequence-structure- function (12-14). We recently reported that the highly conserved UbcH6(UBE2E1)/UbcH8(UBE2E2) pair deviates from this principle and displays divergent E3 interaction patterns (6). UbcH6(UBE2E1) and UbcH8(UBE2E2) contain UBC-folds that differ at three E2-E3 interface positions, residue D66/E58 in H1 and residues T103/S111 and E105/D113 in L1, respectively (residue numbering according to Figure 4.1A) and a less conserved N-terminal extension of approximately 60 residues (Figure 4.1A). When testing these E2 enzymes for global RING-type E3 interactions, both UBC-folds exhibit deviating interaction patterns. UbcH6(UBE2E1) interacts with approximately 20-25 RING-finger domains, while UbcH8(UBE2E2) interacts with around 1-5 RINGs, depending on the cut-off scores used for interaction selection. These opposing E3 interaction patterns are in contrast with the high levels of sequence identity between the two E2 enzymes.

Surprisingly, none of the differences will change the overall charge of the molecule, since the negatively charged and polar residues are constant between the E2s. To understand how these minimal sequence differences are involved in the divergent interaction patterns, the three differential residues were studied in further detail, which are located in Helix 1 and in Loop 1 of the predicted E2-E3 interaction interface (Figure 4.1A). We tested their involvement in specifying RING-interactions by substituting the three differential residues in UbcH8(UBE2E2) with corresponding amino acids from UbcH6(UBE2E1), followed by interaction screening against a panel of 250 human RING-finger domains (Figure 4.1B). As antic- ipated, wild-type (wt) UbcH6(UBE2E1) is able to interact with many E3s, while UbcH8(UBE2E2) only interacts with a few RING-type E3s (Figure 4.1B). Surprisingly, the E66D and D113E UbcH8(UBE2E2) mutants interact in a highly similar manner as wt UbcH6(UBE2E1) (Figure 4.1B). Despite the minimal sequence alterations and nominal differences in physicochemical properties of the side-chains upon substitution, major alterations in physical E3-binding patterns between the two E2 enzymes were induced. In addition, UbcH8(UBE2E2) S111T did not resemble wt UbcH6(UBE2E1) (Figure 4.1B), indicating that not all residues at the E3-interaction interface are equally important. Thus, minimal differences in side-chain characteristics of a selective subset of interface residues induce dramatic changes in global physical E3 binding, we refer to this as the butterfly effect in protein- protein interactions (15).

To evaluate if the E3 interaction pattern of UbcH6(UBE2E1) could also be directed to that of UbcH8(UBE2E2), D58, T103 and E105 in UbcH6(UBE2E1) were substituted with the corresponding residues of UbcH8(UBE2E2) and tested against the RING-finger E3 arrays (Supple-

96 mentary Figure 4.1). All three reverse substitutions did not change the RING E3 interaction profile from wt UbcH6(UBE2E1) to UbcH8(UBE2E2), indicating that alterations in E3 interaction patterns induced by substitutions of conserved interface residues can only be directed in one way.

A C UbcH8(UBE2E2) D58/E66 T152A/S154A

T103/S111 Helix 1

UbcH6(UBE2E1) PDB: 3bzh UbcH8(UBE2E2) PDB: 1y6l Loop 1 Loop 2 E105/D113 B

S161

E66

Helix 1

UbcH8(UBE2E2) PDB: 1y6l

Loop 2 T159 Loop 1

Figure 4.2. Both solvent accessible and inaccessible E2 residues are involved in RING E3 interaction selectivity (A) Structural alignment of UbcH6(UBE2E2) (PDB: 3bzh, blue) and UbcH8(UBE2E2) (PDB:1y6l; pink). Shown are H1, L1 and L2 as well as the three differential residues in stick representation. (B) Crystal structure of UbcH8(UBE2E2). Residue E66 (white), as well as T152 and S154 (blue) are shown as spheres. (C) Mutant UbcH8(UBE2E2) T152A/S154A was screened for physical interactions against a panel of 250 human RING-finger domains. See Materials section for more details.

To investigate UbcH6(UBE2E1) and UbcH8(UBE2E2) specificity towards E3 ligases we analyzed their three-dimensional (3D) structures in more detail. The structures of UbcH6(UBE2E1) and UbcH8(UBE2E2) indicate how D58/E66 and E105/D113, respectively are organized (Figure 4.2A). The side-chain of H1 D58/E66 is not interface-exposed and is therefore most likely not involved in E3 binding. Structurally, these side-chains are in proximity with two closely positioned residues T144/S146 in UbcH6(UBE2E1) and T152/S154 in UbcH8(UBE2E2) (Figure 4.2B). These residues demarcate the S-P-A-L-T-I-S motif in Loop 2, which is highly conserved among different E2 enzymes and involved in specifying RING interactions (16). We tested the effects of these two residues on E3 interaction selectivity by mutating both amino acids to alanines.

97 A UbcH6(UBE2E1) UbcH8(UBE2E2) UbcH8(UBE2E2) D120E

   180o 180o 180o

B UbcH6(UBE2E1)-TOPORS

   90o 90o 90o

UbcH8(UBE2E2)-TOPORS

   90o 90o 90o

C UbcH6(UBE2E1)-topors UbcH8(UBE2E2)-topors

K33 E105 D113 K33

E120 D120

Figure 4.3. Structural approaches towards UbcH6(UBE2E1) and UbcH8(UBE2E2) D113E E2 enzymes (A) Sur- face electrostatic potentials of wt UbcH6(UBE2E1) and UbcH8(UBE2E2) and mutant UbcH8(UBE2E2) D113E, calculated with the APBS-plug-in in Pymol. These potentials were calculated on the minimized structures, no differences were observed on structures generated after 10 ns of molecular dynamics simulations. Two views of each protein (front- and back-sides) are shown. Blue color represents atoms with low electron density, red color represents residues with high electron density and white represents neutral and charged residues. Indi- cated with arrows are E105/D113 residues. (B) Structural representation of the UbcH6(UBE2E1)/UbcH8(UBE2E2)-TOPORS model after threading, structural alignment and refinement in water using HADDOCK. Shown are the E2 enzymes (purple), the RING-finger of TOPORS (pink) and zinc atoms (yellow). See Materials section for more details. Molecular graphics images were produced using the UCSF Chime- ra package (Pettersen et al, 2001) (C) Close-up of the E2-E3 interactions interface in the UbcH6(UBE2E1)/UbcH8(UBE2E2)-TOPORS model. Shown are the E2 enzymes (purple), the RING-finger of TO- PORS (pink) and zinc atoms (yellow). Zinc-binding cysteines and E2 residues E105/D113 and TOPORS residue K33 are shown as sticks and colored according to element.

98 Surprisingly, UbcH8(UBE2E2) T152/S154 interacts different as wt UbcH8(UBE2E2) in a pattern that shares many RING-interactions with UbcH6(UBE2E1), although the patterns are not equal (Figure 4.2C). This suggests that the T152/S154 residues are involved in positioning E66 and that substitution with alanines could destabilize E66 and potentially the complete Helix 1.Taken together, small differences between conserved residues at key positions in two highly homologous E2 enzymes specify E3 interactions, controlled by a network of intramolecular contacts between different positions with the E2 enzyme.

Next, we further analyzed the Loop 1 D-to-E substitution in UbcH8(UBE2E2). D113 in UbcH8(UBE2E2) is expected to be interface exposed and thus potentially directly involved in RING interactions. In order to explain how a single methylene-group difference can drive interaction patterns, this effect was studied in more detail. First, the electrostatic potentials of the E2 enzymes were calculated. The Poisson-Boltzmann electrostatic potential surfaces of both wt UbcH6(UBE2E1), UbcH8(UBE2E2) and UbcH8(UBE2E2) D113E indicate no differences, both for the minimized starting structures and for E2 structures after 10 ns of molecular dynamics simulation (Figure 4.3A). Secondly, the effect of D-to-E substitution on structural dynamics was evaluated using extended molecular dynamics trajectories (Supplementary Figure 4.2). The cross-RMSD calculated on all C and C (comparison values of each structure in the trajectory with respect to each other structure in the same trajectory), revealed that the RMSD remains low and no major conformational changes are occurring. Thus, no major structural rearrangements in the backbone and C traces of both E2 enzymes were observed when wt UbcH8(UBE2E2) and UbcH6(UBE2E1) and the UbcH8(UBE2E2) D113E mutant trajectories were run for 10 ns and compared side-by-side. These molecular dynamics simulations do not indicate electrostatic or major conformational changes that can explain the butterfly effect of a single methylene-group.

To investigate the effects on E3 interactions, an in-depth structure analysis was pursued concentrating on E2-E3 pairs. By using existing experimental E2-E3 structures as templates, combined with three-dimensional alignments, protein threading and water refinement, models were generated between UbcH6(UBE2E1) and UbcH8(UBE2E2) in complex with the RING-finger domain of TOPORS, one of the differentially interacting RING-type E3s (Figure 4.3B). Simulation runs of UbcH6(UBE2E1) and UbcH8(UBE2E2) with TOPORS did not yield steric clashes or overlapping residues. Scoring of the complexes with HADDOCK revealed significantly different values between these two pairs, despite the high levels of similarity. Differences were observed between HADDOCK’s scores of UbcH6(UBE2E1)-TOPORS (-96.0 +/- 4.3) as compared to UbcH8(UBE2E2)-TOPORS (-79.9 +/- 5.3). This was mainly due to variation in electrostatic interactions in both complexes (UbcH6(UBE2E1)-TOPORS: -248.6 ± 27.6 and UbcH8(UBE2E2)-TOPORS: -194.7 ±40.0). On the macromolecular level, both ensembles look very similar, indicating that no major alterations are occurring (Figure 4.3B).

99 A UbcH6(UBE2E1) UbcH8(UBE2E2)

Loop 1 Loop 1

E105 K117 D113 D137 D145 K109 bridged bridged

Loop 1 Loop 1

E105 D113 D137 D145 K117 K109 open open

B UbcH6(UBE2E1) UbcH8(UBE2E2)

K109 – E105 K117 – D113 K109 – D137 K117 – D145 Distance (Å)

Time (ns) Time (ns)

Figure 4.4. A network of intramolecular salt-bridges controls D/E-mediated RING selectivity (A) Molecular graphic images of the network of residues involved in the equilibrium state. UbcH6(UBE2E1) is depicted in blue and UbcH8(UBE2E2) in green. Loop 1 is indicated, as well as the residues involved in the intramolecular salt- bridge network. (B) Distances calculated between E105/D113-K109/K117 (blue) and K109/K117-D137/D145 (green) over a 10 ns trajectory in wt UbcH6(UBE2E1) and UbcH8(UBE2E2). Distances (in angstroms) are minimal distances between the nitrogen’s of the lysine side-chain and the oxygen’s of the glutamic acid (resp. aspartic acid) side-chains. Curves were smoothened using a running-average in a window of 120 ps. The standard deviation for each curve is plotted in dots.

100 The model of UbcH6(UBE2E1)-TOPORS indicate that E105 directly interacts a positive patch on the TOPORS side formed by lysine residue K33 and this leads to a salt bridge that would stabilize an intermolecular interaction between the two proteins. In contrast, when zooming into the interface, the complex between UbcH8(UBE2E2) and the TOPORS RING revealed that D113, sharing a similar side chain as E105 although shorter, is shielded away from K33 in TOPORS and is therefore insufficient to stabilize the interaction (Figure 4.3C). Next, the side-chain conformations of the amino acids at position 105 and 113 were evaluated in both wt E2 enzymes to unravel a potential effect of neighboring residues. Multiple extended trajectories were run for both the wt and mutant UbcH6(UBE2E1) and UbcH8(UBE2E2) and the side-chains of the residues at positions 105 and 113 in loop L1 were measured in terms of distance towards the side-chains of surrounding residues. Our in silico analysis revealed a subtle conformational equilibrium in loop L1 carrying the E105/D113 residues. Two different conformations of the side-chains of E105/D113 were observed in solution (Figure 4.4A). First, an “open” or binding competent conformation that leaves the side- chain of E105/D113 accessible for RING-interaction. Second, a “bridged” or binding incompetent conformation was observed, in which the side-chain of E105/D113 is shielded away from the RING E3 and orientated towards the E2 itself (Figure 4.4A). In both E2 enzymes, a central role in this equilibrium is provided by the side-chain of the adjacent residue K109/K117, which is conserved in both UbcH6(UBE2E1) and UbcH8(UBE2E2), respectively. The side-chains of K109/K117 are shielding E105/D113 by an intramolecular salt-bridge (Fig- ure 4.4A). K109/K117 is highly conserved among different members of the E2 super family and has previously been reported to be involved in establishing E2-E3 interaction selectivity.

Evaluation of in silico substitutions of either loop L1 E105 or D113 resulted in reverse populations and kinetics of exchange between UbcH6(UBE2E1) and UbcH8(UBE2E2). This is in agreement with the initial D113 to E113 substitution associated with changes in E3 interaction patterns in the two-hybrid system. According to this model, K109/K117 plays a crucial role in this equilibrium and is involved in mediating the shuttling between two intramolecular salt-bridges. The first is between K109/K117 and E105/D113, resulting in a “bridged” conformation, preventing E/D from contacting the RING. In the second, K109/K117 is captured by an intramolecular salt-bridge with D137/D145 and as a consequence E105/D113 is able to adopt an “open” conformation that allows an intermolecular salt-bridge with the RING (Fig- ure 4.4A). Residue D137/D145 is highly conserved, both in UbcH6(UBE2E1) and UbcH8(UBE2E2) and among many other members of the E2 super family. Interestingly, this amino acid is distant from the interface and does not interact direct with residues of the RING. The subtle, labile conformational equilibrium and the balance between open (binding competent) and bridged (binding incompetent) conformations are directly affected by the length of the side-chain of E105/D113. Accordingly, the key in controlling E3 interaction profiles between UbcH6(UBE2E1) and UbcH8(UBE2E2) resides in subtle intramolecular dynamics of residues in loop L1. Both conformations were sampled for both E2 enzymes in multiple trajectories. In wt UbcH8(UBE2E2), the population of the bridged, binding-incompetent con-

101 formation is high with a slow kinetic of exchange and this is associated with a poor interaction profile (Figure 4.4B). In contrast, in wt UbcH6(UBE2E1) the E105-K109 bridge fluctuates more frequent and displays, in contrast to UbcH8(UBE2E2), a faster exchange. Additionally, the distances between E105-K109 and K109-D139 in UbcH6(UBE2E1) and those between D113-K117 and K117-D145 in UbcH8(UBE2E2) are anti-correlated, although this effect is more profoundly in UbcH6(UBE2E1) indicating a more dynamic situation. These observations indicate that the distribution of differences in equilibrium populations and kinetics of exchange of the orientation of E105/D113 varies extensively between both E2 enzymes (Sup- plementary Figure 4.3A & B). The different clusters found describe both the conformational deviation of Loop 1 with respect to the rest of the E2 enzymes and to internal conformational changes of Loop 1. So, Loop 1 samples different states during the 10 ns trajectory, with a dynamical exchange between a “bridged” form, where the intramolecular E105-K109 salt bridge is formed (cluster 4, 63% of the trajectory) and open states (including the bridge K109-D137). This exchange is very rapid for UbcH6(UBE2E1) and could imply a low energy barrier between these different states. In contrast, the Loop 1 of UbcH8(UBE2E2) trajectory samples different states too, but displays a slow exchange, which implies a high energy barrier.

To test the different equilibrium states of UbcH6(UBE2E1) and UbcH8(UBE2E2), the molecular dynamics model was challenged experimentally using site-directed mutagenesis of residues involved in maintaining the dynamic balance of the intramolecular salt-bridge network. First, the function of K117 was assessed in UbcH8(UBE2E2). According to the simulations, K117 has a central role in contacting residues D113 and D145 alteratingly, thereby shielding away D113 for RING-contact. By substituting K117 with a histidine reverts the E3 interaction pattern of UbcH8(UBE2E2) to that of UbcH6(UBE2E1) (Figure 4.5A). By doing so, the central extended, positively-charged property of the K117 side-chain was replaced by the average- charged imidazole side-chain of histidine. As a consequence, K117H is no longer able to shield D113 and D145, leaving these residues fully accessible for contacting RING residues. On the other hand, UbcH8(UBE2E2) K117R interacts in a manner similar as wild-type UbcH8(UBE2E2) (Figure 4.5A). The arginine shares the basic side-chain with lysine and is therefore bridged in the intra-molecular salt-bridge with D113, shielding it away from the E2-E3 interface. In addition, preliminary results suggest that UbcH8(UBE2E2) K117D, K117A and K117Q interact similar as UbcH6(UBE2E1) (data not shown). The K117D substitution prevents intra-molecular salt-bridges with D113 and D145, enabling D113 to interact with the RING. The UbcH6(UBE2E1)-phenotype of the K117A mutation suggest that K117 itself is directly involved in RING contacts. Finally, UbcH8(UBE2E2) K117Q implies that the charge of the K117 side-chain is critical for maintaining the dynamic equilibrium, instead of the length of the side-chain. These observations confirm K117 as regulator of the side-chain conformation of D113 and D145.

102 A UbcH8(UBE2E2) UbcH8(UBE2E2) K117H K117R

B UbcH8(UBE2E2) UbcH8(UBE2E2) UbcH8(UBE2E2) D145K D145E D113A/D145E

UbcH6(UBE2E1) UbcH6(UBE2E1) C E105A K109R

Figure 4.5. E3 interaction profiles of UbcH8(UBE2E2) and UbcH6(UBE2E1) mutants (A) E3 interaction profiles of UbcH8(UBE2E2) K117H and K117R. Mutagenesis was performed on LexA-UbcH8(UBE2E2), transformed in EGY48α, mated against the 250 human RING-finger domain arrays and transferred to X-Gal-supplemented selection medium. Pictures were taken 72 hours after transfer. See Materials section for more details. (B) Idem as (A), shown are profiles of UbcH8(UBE2E2) D145K, D145E and D113A/D145E. (C) Idem as (A), shown are profiles of UbcH6(UBE2E1) E105A and K109R.

103 We next investigated the contribution of D145 in regulating the dynamic state of side-chain orientations. UbcH8(UBE2E2) D145K interact as wild-type UbcH8(UBE2E2) (Figure 4.5B), indicating that the similar charges of K117 and D145K induce bridging between K117 and D113 and thereby preventing RING-contacts. In contrast, UbcH8(UBE2E2) D145E actively shields K117 in a manner that prevents K117-D113 bridging, leading to the UbcH6(UBE2E1) interaction profile (Figure 4.5B). To evaluate a potential direct RING-interacting role for D113, the effect of D113A substitution was studied. Interestingly, combining D113A in the UbcH8(UBE2E2) D145E backbone reveals interactions with only a handful of E3s that are also bound by UbcH6(UBE2E1), but the UbcH8(UBE2E2) double mutant interacts different (Figure 4.5B). This suggests that D113A is not directly interacting with the majority of RING-finger domains, but only to a specific subset of RINGs. An alternative explanation could be that K117-D145E, although shielded by intramolecular salt-bridges, could still interact with some RING-fingers. In contrast, both UbcH6(UBE2E1) E105A and K109R interact similar as UbcH6(UBE2E1) wild-type (Figure 4.5C). This implies for UbcH6(UBE2E1) E105A that alternative residues are probably involved in RING-binding, something that could also be the case for UbcH6(UBE2E1) K124A, that did not alter interaction patterns of wild-type UbcH6(UBE2E1) (data not shown). On the other hand, UbcH6(UBE2E1) K109R suggests a mechanism of intra-molecular E105 bridging. Taken together, these experimental validations indicate a delicate way of positioning critical RING-interacting residues in UbcH8(UBE2E2) via an extended network of strictly conserved residues.

Intriguingly, mutants derived from UbcH6(UBE2E1) did not alter the interaction pattern towards the profile of UbcH8(UBE2E2). This can be explained by the observation that only single mutants in UbcH6(UBE2E1) were tested and that, although the high levels of sequence identity between UbcH6(UBE2E1) and UbcH8(UBE2E2), UbcH6(UBE2E1) requires combined mutations to direct the interaction patterns towards that of UbcH8(UBE2E2). An alternative explanation could come from the evolutionary relationship between those E2s and that UbcH8(UBE2E2) has deviated from UbcH6(UBE2E1) during evolution. It might be possible that evolutionary changes in E2 protein structure are irreversible and that the evolutionary route UbcH6(UBE2E1) has undertaken cannot be re-entered. Recently, Bridgham et al. (17) have studied the evolution of ligand specificity of the vertebrate glucocorticoid receptor and compared modern proteins with ancestral receptors. They first identified several restrictive mutations that specify ligand selectivity of the modern receptor. However, these mutations destabilized protein structures that are required for a functional ancestral receptor. The authors also identified ratchet-like epistatic substitutions that are required for changing the modern receptor into the ancestral state. Introduction of the restrictive mutations generated a non-functional protein, since mutation of these restrictive residues first does not en- hance the ancestral protein function. Only when the ratchet, ancestral background is res- tored, the restrictive mutations direct the specificity of the modern receptors towards the ancestral one. It could be possible that UbcH6(UBE2E1) requires a similar mechanism and

104 that initial substitution of residues making UbcH6(UBE2E1) more resembling UbcH8(UBE2E2) need additional ratchet-like factors.

In conclusion, these experiments indicate a butterfly effect in regulating E3 interactions between two highly conserved E2 enzymes. Mediated via a delicate network of salt-bridges, a central RING-interacting residue is in a dynamical equilibrium controlled by strictly conserved residues that actively control the positioning of the central loop L1 residue. These observations led to the postulation of a molecular switch theory, in which K109/K117 acts as a gate-keeper for the negatively charged E105/D113, and thus, for establishing the interaction. If this residue is longer, it can escape the effect of K109/K117 easier and thus making a much more favorable contact with the positive patch of TOPORS. If not, then it is trapped by K109/K117, which positions depends on D137/D145, thereby shielded from K33 of TOPORS and disabled to interact. Since the equilibrium residues are conserved in other E2 enzymes, this dynamic regulation might be a general manner of subtle orientation involved in establishing selectivity in physical E2-E3 interactions. The observation that minimal sequence differences between two highly similar proteins could have dramatic outcomes and might be indicative for revision of the widely used sequence-structure-function paradigm.

Materials and methods Plasmids and antibodies High-copy yeast-two hybrid (Y2H) shuttle plasmids expressing human UbcH6(UBE2E1) and UbcH8(UBE2E2) and RING-finger domains of 250 human RING-type E3 ubiquitin protein ligases as fusions with the E. coli LexA binding domain (BD) or with the B42 acidic blob protein activator domain (AD), respectively are described earlier (6). All amino acid substitutions were introduced using QuickChange II Site-Directed mutagenesis (Stratagene), according to the manufacturers’ instructions. Appropriate mutagenesis was validated using DNA sequencing. Monoclonal antibodies used were against LexA (2-12, Santa Cruz) and yeast tubulin (Li- fespan Biosciences, Seattle).

Molecular dynamic simulations Both wild-type and mutant UbcH6(UBE2E1) and UbcH8(UBE2E2) were used as starting structures. Wild-type UbcH6(UBE2E1) and UbcH8(UBE2E2) starting structures were directly obtained from the Protein Data Bank (PDB) with PDB-codes 3bzh and 1y6l, respectively. Selec- tive mutations were introduced with CNS, while keeping both the Cα and the Cβ to preserve the side-chain rotamer. To study E2 enzyme dynamics, eight molecular dynamics trajectories, covering a total time of 80 ns, were performed in explicit solvent conditions, using the GROMOS G53a6 force field (18) and GROMACS package (19). Each starting structure was solvated in a dodecahedral box with 10 Å distance between the solute and the box, using about 8700 SPC water molecules, 23 Na+ and 25 Cl- ions (1, depending on the system). Si- mulations were done using periodic boundary conditions at 300 K and constant pressure (1 atm.), using Berendsen coupling algorithm (20) for both temperature and pressure (coupling

105 constants of 0.1 and 1.0 ps, respectively). Bond lengths were constrained with the Linear Constraint Solver (LINCS) algorithm (21) and the time step for dynamics was 2 fs. The reaction-field method was used with a cut-off distance of 14 Å for electrostatic interactions and non-bonded interactions were updated every 10 fs (with a 9 Å cut-off distance for the short- range neighbor list). Each structure was equilibrated for 50 ps prior to molecular dynamics analysis.

Structural alignment and refinement of E2-E3 complexes with HADDOCK Binary complexes of E2/RING-E3 interactions are characterized by a common interface region and positioning of the UBC-fold and the RING-finger domain. The existence of a handful of complex structures is already solved and deposited in the PDB-databank. These structures are however biased towards functionality, e.g. structural ensembles are generated of biological active E2-E3 pairs. In contrast, there is no general consensus in determining functional from non-functional E2-E3 complexes from individual E2 and E3 domains. As such, data- driven docking per se seems not to be a trivial solution towards this problem. By using some of HADDOCK’s components - e.g. water refinement and scoring function - combined with a ‘threading’ approach it is possible to discriminate which pairs of E2s and E3s may interact biologically and which are artifacts. To predict the structures of the UbcH6(UBE2E1) and UbcH8(UBE2E2) in complex with the RING-finger domain of TOPORS, structural alignments of single structures of E2s and E3s were performed on the bound structures of existent E2-E3 complexes (PDB-codes: 1fbv, 1ur6, 3eb6, 2c2v and 2oxq). This was followed by refinement in water and scoring with HADDOCK’s force field where scoring function has been shown to discriminate between biologically relevant complexes and artifacts (unpublished observations Maaike Ransdorp, A.S.J.M and A.M.J.J.B.).

Electrostatic potentials of E2 enzymes Electrostatic potential surfaces were generated using the Adaptive Poisson-Boltzmann Solver package (APBS), which solves the Poisson-Boltzmann equation numerically (22). APBS was shown to evaluate efficiently through a continuum model the electrostatic interactions between molecular solutes in salty, aqueous media. This package was used directly in the Py- mol software as a plug-in to evaluate potentials of wild-type and mutant E2 enzymes.

Yeast two-hybrid analysis Manipulation of yeast cells and two-hybrid techniques are described previously (6). Briefly, wild-type and mutant LexA-E2 fusions were co-transformed together with the pSH18-34 LacZ reporter plasmid in EGY48 cells. To evaluate the effects of E2 mutants on human RING-type interactions, 250 sequence-verified B42-RING constructs were arrayed in EGY48a cells. Mat- ing between α and A cells was done on non-selective Yeast Peptone Dextrose (YPD)-medium (24 hours at 30 oC). Diploids were selected for 48 hour on SC HWU- medium by manual transfer of yeast spots using a 384-well replicator pinning tool (V & P Scientific, San Diego). Puta- tive interactions were assessed in triplicate by growing diploids on SC HWU- medium sup-

106 plemented with 5-bromo-4-chloro-3-indolyl -D-galactopyranoside (X-gal) or on SC HWUL-, with either galactose or glucose as main carbon source. Quantification of interactions was done as reported earlier (6).

References 1. Deshaies, R. J., and Joazeiro, C. A. (2009) RING domain E3 ubiquitin ligases. Annu Rev Biochem 78, 399-434 2. Joazeiro, C. A., Wing, S. S., Huang, H., Leverson, J. D., Hunter, T., and Liu, Y. C. (1999) The tyrosine kinase negative regulator c-Cbl as a RING-type, E2-dependent ubiquitin- protein ligase. Science 286, 309-312 3. Li, W., Bengtson, M. H., Ulbrich, A., Matsuda, A., Reddy, V. A., Orth, A., Chanda, S. K., Batalov, S., and Joazeiro, C. A. (2008) Genome-wide and functional annotation of human E3 ubiquitin ligases identifies MULAN, a mitochondrial E3 that regulates the organelle's dynamics and signaling. PLoS ONE 3, e1487 4. Hochstrasser, M. (2009) Origin and function of ubiquitin-like proteins. Nature 458, 422-429 5. Michelle, C., Vourc'h, P., Mignon, L., and Andres, C. R. (2009) What Was the Set of Ubiquitin and Ubiquitin-Like Conjugating Enzymes in the Eukaryote Common Ances- tor? J Mol Evol 68, 616-628 6. van Wijk, S. J., de Vries, S. J., Kemmeren, P., Huang, A., Boelens, R., Bonvin, A. M., and Timmers, H. T. (2009) A comprehensive analysis of E2-RING E3 interactions of the human ubiquitin-proteasome system. Mol Syst Biol 5 7. Burroughs, A. M., Jaffee, M., Iyer, L. M., and Aravind, L. (2008) Anatomy of the E2 ligase fold: Implications for enzymology and evolution of ubiquitin/Ub-like protein conjugation. J Struct Biol 162, 205-218 8. Zheng, N., Wang, P., Jeffrey, P. D., and Pavletich, N. P. (2000) Structure of a c-Cbl- UbcH7 complex: RING domain function in ubiquitin-protein ligases. Cell 102, 533-539 9. Xu, Z., Kohli, E., Devlin, K. I., Bold, M., Nix, J. C., and Misra, S. (2008) Interactions between the quality control ubiquitin ligase CHIP and ubiquitin conjugating enzymes. BMC Struct Biol 8, 26 10. Zhang, M., Windheim, M., Roe, S. M., Peggie, M., Cohen, P., Prodromou, C., and Pearl, J. H. (2005) Chaperoned ubiquitylation-crystal structures of the CHIP U box E3 ubiquitin ligase and a CHIP-Ubc13-Uev1a complex. Mol Cell 20, 525-538 11. Dominguez, C., Bonvin, A. M., Winkler, G. S., van Schaik, F. M., Timmers, H. T., and Boelens, R. (2004) Structural model of the UbcH5B/CNOT4 complex revealed by combining NMR, mutagenesis, and docking approaches. Structure 12, 633-644 12. Worth, C. L., Gong, S., and Blundell, T. L. (2009) Structural and functional constraints in the evolution of protein families. Nat Rev Mol Cell Biol 10, 709-720 13. Williams, S. G., and Lovell, S. C. (2009) The Effect of Sequence Evolution on Protein Structural Divergence. Mol Biol Evol 26, 1055-1065 14. Philippe, H., Casane, D., Gribaldo, S., Lopez, P., and Meunier, J. (2003) Heterotachy and functional shift in protein evolution. Iubmb Life 55, 257-265 15. Lee, B. C., Park, K., and Kim, D. (2008) Analysis of the residue-residue coevolution network and the functionally important residues in proteins. Proteins 72, 863-872 16. Christensen, D. E., Brzovic, P. S., and Klevit, R. E. (2007) E2-BRCA1 RING interactions dictate synthesis of mono- or specific polyubiquitin chain linkages. Nat Struct Mol Biol 14, 941-948

107 17. Bridgham, J. T., Ortlund, E. A., and Thornton, J. W. (2009) An epistatic ratchet con- strains the direction of glucocorticoid receptor evolution. Nature 461, 515-U578 18. Oostenbrink, C., Soares, T. A., van der Vegt, N. F. A., and van Gunsteren, W. F. (2005) Validation of the 53A6 GROMOS force field. Eur Biophys J 34, 273-284 19. Van der Spoel, D., Lindahl, E., Hess, B., Groenhof, G., Mark, A. E., and Berendsen, H. J. C. (2005) Gromacs: Fast, Flexible, and Free. J Comp Chem 26, 1701-1718 20. Berendsen, H. J. C., Postma, J. P. M., Van Gunsteren, W. F., Dinola, W. F., and Haak, J. R. (1984) Molecular-dynamics with coupling to an external bath. J Chem Phys 81, 3684-3690 21. Hess, B., Bekker, H., Berendsen, H. J. C., and Fraaije, J. G. E. M. (1997) LINCS: A linear constraint solver for molecular simulations. J Comp Chem 18, 1463-1472 22. Baker, N. A., Sept, D., Holst, M. J., and McCammon, J. A. (2001) Parallel adaptive solution of the Poisson-Boltzmann equation for large biomolecules. Abstr Paper Am Chem Soc Natl Meet 221, U437-U437

108 Supplementary Figures Chapter 4

UbcH6(UBE2E1) UbcH6(UBE2E1) UbcH6(UBE2E1) D58E E105D T144A/S146A

Supplementary Figure 1. E3 interaction profile of UbcH6(UBE2E1) mutants. E3 interaction profiles of UbcH6(UBE2E1) D58E, E105D and T152A/S154A. Mutagenesis was performed on wild-type LexA- UbcH6(UBE2E1), transformed in EGY48α, mated against the 250 human RING-finger domain arrays and transferred to X-Gal-supplemented selection medium. Pictures were taken 72 hours after transfer. See Materials section for more details.

109 A B UbcH6(UBE2E1) UbcH6(UBE2E1)

UbcH8(UBE2E2) D113E UbcH8(UBE2E2) C D UbcH6(UBE2E1) UbcH8(UBE2E2)

UbcH6(UBE2E1) UbcH8(UBE2E2) Supplementary Figure 2. Molecular dynamics simulation profiles of wt UbcH6(UBE2E1) and wt and D120E UbcH8(UBE2E2) (A) and (B) Cross-RMSDs calculated on all Cα + Cβ (comparison values of each structure in the trajectory with respect to each other structure in the same trajectory) to identify potential conformational changes along the trajectory. Shown are the trajectories between UbcH6(UBE2E1) and UbcH8(UBE2E2) D113E (A) and between UbcH6(UBE2E1) and UbcH8(UBE2E2) (B). (C) and (D) Idem as (A) and (B) but cross-comparing trajectories of UbcH6(UBE2E1) and UbcH8(UBE2E2), respectively.

110 A UbcH6(UBE2E1) – Loop 1

B UbcH8(UBE2E2) – Loop 1

Supplementary Figure 3. Differential dynamic exchanges in intramolecular salt-bridges in UbcH6(UBE2E1) and UbcH8(UBE2E2) (A) Clustering of Loop 1 during the trajectory of 10 ns for wt UbcH6(UBE2E1). Aligned are all conformations generated during the molecular dynamics simulation on the Cα’s of E2 enzymes, except those of Loop 1. Clustering was done by calculating the RMSD deviation between the aligned structures, only on Loop 1 (F117-P125, all atoms). (B) Idem as (A) but clustering is now done for UbcH8(UBE2E2).

111

Chapter 5

General Discussion

113 Complex biological networks depend on the selectivity and spatial and temporal regulation of protein-protein interactions. A typical example of such a system is the ubiquitin-proteasome system of which the outcome (substrate modification) is determined by selective interactions within the E1-E2-E3 pyramid. Substrates are modified in a myriad of ways, varying from single ubiquitin molecules to long bifurcated chains that are constructed via differential use of the ubiquitin lysines. All these different types of modifications, composed of individual ubiquitin building-blocks, direct substrates for either degradation or non-degradation purposes. In- dispensable for these functions are ubiquitin-conjugating enzymes (E2s) which actively control the type of modification deposited on substrates. The human genome encodes 35 E2 enzymes of which the majority interacts with ubiquitin protein ligases to come to the wealth of different shapes of ubiquitin. The studies presented in this thesis are aimed at studying E2-E3 interactions selectivity on a genome-wide scale in the human ubiquitin-proteasome system.

The E2-E3 networks For a long time the landscape of human E2-E3 interactions remained sparse and limited to a handful of binary pairs. Our efforts to construct a domain-based network revealed 346 high- confidence E2-E3 pairs of which a large proportion was previously unknown (Chapter 2) (1). Exhaustive screening identified many new interactions for several uncharacterized E2 and E3 enzymes. Comparing observed interactions with literature-curated, ubiquitination- productive E2-E3 pairs indicate a good degree of overlap, signifying physical E2-E3 interactions as useful predictors of biochemical functionality. Furthermore, a subset of interactions between high- and low-interacting E2s and E3s was experimentally verified in independent GST pull-down assays, confirming the quality of the yeast two-hybrid interactions.

Given the high levels of agreement between physical and biochemically functional E2-E3 pairs, a domain based-yeast two-hybrid approach is efficient in unraveling E2-E3 selectivity. This is in contrast to many high-throughput ORFeome-based two-hybrid screens (2-6) in which E2-E3 interactions remain underrepresented. It is expected that, apart from studying interactions in ubiquitination pathways, dedicated domain-centered interactions assays could provide high-resolution and more realistic networks than their massive genome-wide counterparts (7). Testing interactions between isolated domains could, despite extensive validation of DNA and protein quality of bait and preys, increase the risk of false-negatives since potentially essential sequences are missing (8, 9). Although it is well established that the minimal interacting units are the UBC-folds and RING-finger domains (10, 11), some E2-E3 pairs require additional sequences to interact, like c-Cbl-UbcH7 (12). This could explain why some proteins did not show any interaction. No overrepresentation of interactions was observed for E2s that have additional N- and/or C-terminal extensions to their UBC-fold nor did the position of the RING in the full-length E3 affect interaction potential. Additionally, some proteins require homo- or heterodimerization or post-translational modification to become biochemically active (13, 14), but how dimerization influences physical E2-E3 interactions remains to

114 be established. An example is BRCA1/BARD1, a dimeric E3 ligase of which the E3 activity of BRCA1 requires binding of the inactive partner BARD1. When tested as BARD1-bound dimer against a panel of E2 enzymes in yeast-two hybrid assays, BRCA1 interacts with a different set of E2 enzymes than the isolated RING-finger domain (1, 15).

At present it appears that alterations in E2 interface structure upon E3 binding are not common characteristics of all E2-E3 interactions. The structure of free UBE2L3(UbcH7) for example does not indicate changes when bound to the c-Cbl E3 (12). On the other hand, it has been shown that the RING-finger domain of CNOT4 stimulates the rate of UBE2D2(UbcH5B)~ubiquitin thioester release. Mutagenesis of UBE2D2(UbcH5B) revealed important roles for the E3-binding site and the active-site in mediating this effect, indicating allosteric communication between distant functional sites in UBE2D2(UbcH5B) (16). Interes- tingly, mutants that affect ubiquitin release from the E2 did not show alterations in E2-E3 interactions. This would suggest a role for E2-thioester bound ubiquitin in inducing (allosteric) changes in ubiquitin release, but not in the ability of E2 and E3 enzymes to interact. This is in agreement with observation that E2 enzymes in which the catalytic cysteine has been mutated to alanines can be used as dominant-negative E2 enzymes (17, 18).

Recently, Markson et al (19) presented an E2-E3 interaction map describing 535 binary interactions. Using a Gal4-based yeast two-hybrid system, both library- and array-based screen- ings were performed to map interactions between full-length E2 and E3 enzymes. Extending the experimentally derived network by homology modeling and energy calculations revealed >1300 additional E2-RING E3 interactions having more favorable predictable free energies than the c-Cbl-UbcH7(UBE2L3) complex. Furthermore, Markson and coworkers (19) used mutagenesis of conserved RING residues to verify experimentally derived interaction pairs as well as validation using in vitro (auto)ubiquitination assays of selected E2-E3 pairs.

Yeast-two hybrid screens are mostly performed using Gal4 DBD/AD or the LexA/B42 DBD/AD pairs and comparison of interactions observed in both types of assays reveals important information about biological relevance. Comparisons between the LexA-dataset and the Mark- son-dataset are ongoing (Van Wijk et al., manuscript in preparation). Preliminary pair-wise evaluations already revealed that both sets show an interaction overlap of approximately 26% (Christopher Sanderson, personal communication). This might seem low, but is in full agreement with previous observations concerning the overlap in interaction networks of human (3), yeast (4) and Drosophila (20) proteins (see Chapter 1). A major contributor to this effect is the bias of some interactions towards the type of system used, e.g. Gal4- or LexA- based (20), but also false-negatives and false-positive are confounding datasets. False-negative interactions are interactions that are not detected in two-hybrid assays, but that are occurring physiologically. Although the UBC-fold and the RING-finger domain are described as minimal E2-E3 interacting units, it might be possible that sequences outside these domains in full-length E2 and E3 enzymes contribute to E2-E3 interactions. This could

115 be as it is the case for the c-Cbl-UbcH7(UBE2L3) interaction (12). In addition, incorrect folding and absence of protein expression could lead to false-negatives. On the other hand, false-positive interactions are occurring in vitro, but are absent in vivo. An important drawback in screening for interactions between full-length E2 and E3 enzymes is the possibility that interactions are mediated via alternative domains instead of the UBC- fold and the RING-domain. Domains that bind ubiquitin or ubiquitin-like domains in proteins could lead to false-positive interactions. Some E2s, like the UBE2D(UbcH5)-family, show an intrinsic affinity for ubiquitin via their conserved -sheet surface (21) (discussed in Chapter 3) and these E2s may interact with protein domains that resemble ubiquitin in structure. Apart from this, several domains involved in ubiquitin signaling are present in E2s and E3s that potentially confound full-length E2-E3 interactions, either directly or indirectly mediated via free ubiquitin present in the host organism. The ubiquitin-associated (UBA) domain in UBE2K(HIP2) is able to interact with ubiquitin-like (UBL) domains (22), in a manner analog- ous to the dimeric HOIL-1L-HOIP E3 ligase complex (23, 24). The ubiquitin E2 variant (UEV) domain has been identified in the inactive variant E2s UBE2V1/UBE2V2 and in TSG101 (25, 26). Except for the two C-terminal helices, UEV-domains resemble the canonical UBC-fold but are also involved in ubiquitin binding (27). Finally, the Interacting with Ubiquitin Motif (UIM) has been identified in several RING-type E3 ligases, like Rabex-5 (28) and RNF168 (29).

The role of E2 ancestors Selectivity in the E2-E3 network originates from differences in sequence similarity that define the interface between UBC-folds and RING-fingers. To capture similarities in interaction pattern, the hub overlap index was used as a measure of how sequence divergent E2 enzymes may interact with the same set of RING E3s. Many E2 enzymes, like UbcH5A- D(UBE2D1-4), UbcH6(UBE2E1), UbcH9(UBE2E3) and to a lesser extend Ubc13(UBE2N) interact with similar sets of RINGs. To evaluate how these homologies in interaction pattern re- late to sequence divergence, E2 sequence similarities of these hub E2s were compared with their hub overlap index. For the abovementioned E2 enzymes, the patterns of binding overlaps with sequence homology (of both the complete UBC-fold and the RING-interacting regions). For some E2s, this can be explained by a common ancestor that diverged into E2 enzymes of which sequences are more similar, according to the sequence-structure-function paradigm (30). This homology among some E2 members could indicate a functional redundancy in the E2-E3 network and these E2s, except Ubc13(UBE2N), are involved in the general degradation of proteins (31). The expansion of this sub-group of E2s (compared to Saccha- romyces cerevisiae) could serve to protect the cell against inactivation of these vital functions. An exceptional case is Ubc13(UBE2N), an E2 that dimerizes with UEV-1A(UBE2V1) and MMS2(UBE2V2) to catalyze K63-linked polyubiquitin chains, involved in DNA repair and NF- κB signaling (25, 32, 33). Surprisingly, Ubc13(UBE2N) knock-out cells are viable and NF-κB pathways are unaffected in these cells (34), indicating that Ubc13(UBE2N) is not essential. It might be possible that other E2s dimerize with UEV-1A(UBE2V1)/MMS2(UBE2V2) and catalyze K63-chains but this seems unlikely since Ubc13(UBE2N) employs a highly specialized in-

116 teraction surface to bind the variant E2s (33). A more plausible explanation for this relatively mild phenotype could be that Ubc13(UBE2N)-catalyzed K63-chains merely serve as binding platforms for ubiquitin-binding proteins instead of directing substrates for proteasomal degradation (25, 32). These chains are structurally most resembling linear polyubiquitin chains constructed by the dimeric LUBAC E3 ligase (24, 35). Several proteins are identified that are able to bind both K63-linked and linear polyubiquitin chains and this could explain why deletion of Ubc13(UBE2N) does not lead to a severe phenotype (36). Other E2s, like UBE2U, UBE2W and Use1(UBE2Z) are more divergent in sequence as compared with other E2s and probably use specific characteristics to interact with many E3s (37).

Distributions of interactions in the ubiquitination system Interactions between E2 and E3 enzymes do not display a normal distribution pattern. In the E2-E3 network, a small fraction of proteins appeared to be more versatile in binding than the majority of proteins tested. The high-connected nodes, or hubs, followed by nodes with de- creasing numbers of interactions are approximately following a scale-free distribution (although statistical analysis of this principle is limited due to the small size of the network). Protein interaction networks often follow scale-free distributions, but hubs have also been found in technical, social and economic contexts (38). Scale-free properties of networks imply that a small number of high-connected hubs are holding the network together (39). Be- cause of this, these networks are highly resistant against inactivation of random nodes, but very sensitive to specific hub inactivation (39, 40). On the other hand, there are several reasons why a scale-free distribution among E2-E3 domains is surprising. The first is that despite the high levels of sequence and structure homologies, hub E2s are present. The UBC-fold needs to combine E1- and E3-interaction surfaces and is thereby under continuous evolutionary pressure to prevent interactions with other E1 and E3 and inappropriate conjugation of Ub or UBLs (30). A second reason lies in the basic scheme of the E1-E2-E3-substrate enzymatic pyramid (41). Selective E2-E3-substrate interactions are under direct control of the E1 since the E1 offers the activated Ub/UBLs to downstream enzymes (42, 43). Thus, the ubiquitin system combines both scale-free and hierarchical properties. This implies that inactivation of the E1 is detrimental for the complete system, since the E1 is an essential enzyme (44, 45).

Sequence correlates with structure and structure correlates with function (46). This is cer- tainly true for (the majority) of E2 enzymes (1). Mutagenesis of E2 and E3 enzymes has been applied to alter specificities of E2-E3 interactions. For example, Winkler and co-workers (47) created UbcH5B(UBE2D2)-CNOT4 interaction pairs with functional altered specificities. Also, Markson and colleagues (19) have extensively used RING-finger E3 mutagenesis of two conserved positions involved in E2-E3 complexes to validate the selectivity of E2-E3 interactions. Our sequence-based approach to redirect E3 selectivity from Ubc13(UBE2N) to UbcH5B(UBE2D2) by substitution of two residues, could have important biological implications (1). The Ubc13(UBE2N) mutant is expected to modify UbcH5B(UBE2D2)-targets with

117 K63-linked polyubiquitin chains that are normally conjugated with K48-linked or mixed- linkage chains. Therefore, substrates that are normally directed to the proteasome for degradation are being stabilized, at least theoretically. This opens new possibilities to study substrate identification, since substrates that are normally being degraded are expected to become stabilized when conjugated with K63-linked chains. Additionally, this alteration of interaction specificity could underscore the shell-model for E2 enzymes (Chapter 3) and illustrates how different pathways in ubiquitination can converge or diverge.

Shell-like properties of E2 enzymes The strict sequence-structure-function principle is illustrated when Ub- and UBL- conjugation systems are compared with each other (Chapter 3). Most of these pathways employ specific E2 enzymes to conjugate their cognate Ub/UBL. Connecting initial activation with downstream conjugation, UBC-folds are central nodes that distinguish between interactions with Ub- and UBL-specific E1 and E3 enzymes. To connect E2 structure with selective function, a model of E2 enzymes was proposed in which the catalytic core of E2 enzymes is not discriminating between Ub/UBLs usage, but rather that surface residues mediate selectivity with Ub/UBL E1s and E3s. This is clearly illustrated by Schulman and colleagues, who pioneered in altering interacting specificities between the ubiquitin and NEDD8-conjugation pathways (48). Although the E2 catalytic core (shell I) is solvent exposed, residues that make up this region are highly conserved in sequence. This is in contrast with many other proteins that, in general, have a relatively conserved core surrounded by less conserved, solvent accessible residues. According to this model, the first shell is connected to the third shell via buried residues that are not directly interacting with RING residues. Despite the lack of direct contact of these residues with the RING, there are indications that second shell residues have a supporting role in specifying E2-E3 interactions (Chapter 4). This suggests that E2-E3 selectivity is even more complex than assumed and that apart from solvent exposed residues, inner residues are also involved in E3 selection and binding.

E2s: deviating from a paradigm For many E2 pairs, the correlation between sequence, structure and function is a reoccurring theme. However, deviating from this paradigm is the homologous UbcH6(UBE2E1)/UbcH8(UBE2E2) pair. UbcH6(UBE2E1) behaves like a hub E2 while UbcH8(UBE2E2) displays considerable less E3 interactions even though both E2 differ on three residues in the predicted E3 interaction surface. UbcH6(UBE2E1)-mimicking substitutions of Helix 1 E66D and Loop 1 D113E in UbcH8(UBE2E2) reverts the poor interaction pattern of UbcH8(UBE2E2) to the rich profile of UbcH6(UBE2E1). So, very minimal differences lead to dramatic changes in interaction phenotype, something to which we refer to as the butterfly effects in protein interactions (49). The aspartic/glutamic acids 58/66 are at the C- terminal end of helix H1 and its side-chain is pointing towards the core of the E2 enzyme and probably not directly contacting the RING. Structures reveal that the side-chain of D58/E66 are in close proximity of T144/152 and S146/154 which are solvent exposed and known to

118 interact with E3s. Surprisingly, UbcH8(UBE2E2) T152A/S154A fully restores the pattern of UbcH6(UBE2E1). Up till now, this is the first example of a gain-of-function effect of alanine mutants in E2 enzymes. How is this possible? It could be that D58/E66 is a shell II-type of amino acid, involved in positioning the outer E3- interacting residues in optimal binding position. Speculating about the E66D substitution and the effects observed in the T152A/S154A mutant suggests a hinge-like mechanism that is involved in positioning helix H1. NMR analysis of UbcH5B(UBE2D2) already revealed that the stretch connecting H1 with the main part of the E2 enzyme is flexible. This could allow H1 to have some conformational freedom, influencing the distances between the three points of the E2-interaction triangle and thereby the possibility of RING interaction. Less speculative are the effects of loop L1 residue E105/D113 in which the single methylene- group difference is responsible for reverting the UbcH8(UBE2E2)-associated phenotype to that of UbcH6(UBE2E1). In Chapter 4, molecular dynamics was used to propose that the conserved E105/D113 is regulated by a dynamic intramolecular network of salt-bridges that determine a binding competent or ‘open’ and a binding incompetent or ‘bridged’ conformational state. Both conformations are in equilibriums that differ between UbcH6(UBE2E1) and UbcH8(UBE2E2). A crucial role in this balance is the strictly conserved side-chain of K109/117 that alternately binds E105/D113 and D137/145, in manner that is dynamically differs between UbcH6(UBE2E1) and UbcH8(UBE2E2). K109/117 of Loop 1 is present in many E2 enzymes and the homologous K63 in UbcH5B(UBE2D2) has been critically linked to interaction specificity in the CNOT4-UbcH5B(UBE2D2) interaction (47, 50). Residue D137/145 is also highly conserved among the E2 superfamily and together with K109/117 present in both UbcH6(UBE2E1) and UbcH8(UBE2E2), but in contrast, not directly involved in RING-binding (shell II-type of residue). The finding that protein-protein interaction can be directed, modified and redesigned is in agreement with earlier described attempts (51-53). Taken together, these findings indicate a common way of how interface and non-interface residues determine E2-E3 interaction specificity on a global scale.

Conclusively, the studies discussed here provide a solid starting point for in-depth bioinformatic and biochemical studies centered on E2 enzymes. Our current insights in E3- interacting residues in E2 enzymes open new possibilities to study altered specificities between non-homologous E2 enzymes to intermingle Ub/UBL- pathways. Increasing structural clues of E2 and E3 enzymes allow for genome-wide docking approaches that permits new predictions and eventually the development of rationalized design of small E2-E3 modulating small compounds. Residues that critically contribute directly or indirectly to interaction selectivity and the delicate ways of how these residues are influencing E2-E3 interactions are becoming more characterized. However, the exact subtle structural rearrangements in E2 enzymes upon Ub/UBL- and E3-binding remain to be determined. Also, the precise mechanisms of Ub/UBL-conjugation and -transfer are still enigmatic and solving E2-E3-substrate structures would be of great value. Additionally, systems-wide in vivo effects of individual E2 enzymes on E3-substrate recognition and chain assembly are urgently needed. However,

119 when new insights increase, new mechanisms of regulation will be identified that may require revision of old and established paradigms.

References 1. Van Wijk, S. J., de Vries, S. J., Kemmeren, P., Huang, A., Boelens, R., Bonvin, A. M., and Timmers, H. T. (2009) A comprehensive analysis of E2-RING E3 interactions of the human ubiquitin-proteasome system. Mol Syst Biol 5: 295 2. Stelzl, U., Worm, U., Lalowski, M., Haenig, C., Brembeck, F. H., Goehler, H., Stroe- dicke, M., Zenkner, M., Schoenherr, A., Koeppen, S., Timm, J., Mintzlaff, S., Abraham, C., Bock, N., Kietzmann, S., Goedde, A., Toksoz, E., Droege, A., Krobitsch, S., Korn, B., Birchmeier, W., Lehrach, H., and Wanker, E. E. (2005) A human protein-protein interaction network: a resource for annotating the proteome. Cell 122, 957-968 3. Rual, J. F., Venkatesan, K., Hao, T., Hirozane-Kishikawa, T., Dricot, A., Li, N., Berriz, G. F., Gibbons, F. D., Dreze, M., Ayivi-Guedehoussou, N., Klitgord, N., Simon, C., Boxem, M., Milstein, S., Rosenberg, J., Goldberg, D. S., Zhang, L. V., Wong, S. L., Franklin, G., Li, S., Albala, J. S., Lim, J., Fraughton, C., Llamosas, E., Cevik, S., Bex, C., Lamesch, P., Sikorski, R. S., Vandenhaute, J., Zoghbi, H. Y., Smolyar, A., Bosak, S., Sequerra, R., Doucette-Stamm, L., Cusick, M. E., Hill, D. E., Roth, F. P., and Vidal, M. (2005) Towards a proteome-scale map of the human protein-protein interaction network. Nature 437, 1173-1178 4. Uetz, P., Giot, L., Cagney, G., Mansfield, T. A., Judson, R. S., Knight, J. R., Lockshon, D., Narayan, V., Srinivasan, M., Pochart, P., Qureshi-Emili, A., Li, Y., Godwin, B., Conover, D., Kalbfleisch, T., Vijayadamodar, G., Yang, M., Johnston, M., Fields, S., and Roth- berg, J. M. (2000) A comprehensive analysis of protein-protein interactions in Saccha- romyces cerevisiae. Nature 403, 623-627 5. von Mering, C., Krause, R., Snel, B., Cornell, M., Oliver, S. G., Fields, S., and Bork, P. (2002) Comparative assessment of large-scale data sets of protein-protein interactions. Nature 417, 399-403 6. Li, S., Armstrong, C. M., Bertin, N., Ge, H., Milstein, S., Boxem, M., Vidalain, P. O., Han, J. D., Chesneau, A., Hao, T., Goldberg, D. S., Li, N., Martinez, M., Rual, J. F., La- mesch, P., Xu, L., Tewari, M., Wong, S. L., Zhang, L. V., Berriz, G. F., and al, e. (2004) A map of the interactome network of the metazoan C. elegans. Science 303, 540-543 7. Boxem, M., Maliga, Z., Klitgord, N., Li, N., Lemmens, I., Mana, M., de Lichtervelde, L., Mul, J. D., van de Peut, D., Devos, M., Simonis, N., Yildirim, M. A., Cokol, M., Kao, H. L., de Smet, A. S., Wang, H. D., Schlaitz, A. L., Hao, T., Milstein, S., Fan, C. Y., Tipsword, M., Drew, K., Galli, M., Rhrissorrakrai, K., Drechsel, D., Koller, D., Roth, F. P., Iakou- cheva, L. M., Dunker, A. K., Bonneau, R., Gunsalus, K. C., Hill, D. E., Piano, F., Taver- nier, J., van den Heuvel, S., Hyman, A. A., and Vidal, M. (2008) A protein domain- based interactome network for C-elegans early embryogenesis. Cell 134, 534-545 8. Huang, H., and Bader, J. S. (2009) Precision and recall estimates for two-hybrid screens. Bioinformatics 25, 372-378 9. Huang, H., Jedynak, B. M., and Bader, J. S. (2007) Where have all the interactions gone? Estimating the coverage of two-hybrid protein interaction maps. PLoS Comput Biol 3, 2155-2174 10. Joazeiro, C. A., Wing, S. S., Huang, H., Leverson, J. D., Hunter, T., and Liu, Y. C. (1999) The tyrosine kinase negative regulator c-Cbl as a RING-type, E2-dependent ubiquitin- protein ligase. Science 286, 309-312

120 11. Lorick, K. L., Jensen, J. P., Fang, S., Ong, A. M., Hatakeyama, S., and Weissman, A. M. (1999) RING fingers mediate ubiquitin-conjugating enzyme (E2)-dependent ubiquitination. Proc Natl Acad Sci U S A 96, 11364-11369 12. Zheng, N., Wang, P., Jeffrey, P. D., and Pavletich, N. P. (2000) Structure of a c-Cbl- UbcH7 complex: RING domain function in ubiquitin-protein ligases. Cell 102, 533-539 13. Xia, Y., Pao, G. M., Chen, H. W., Verma, I. M., and Hunter, T. (2003) Enhancement of BRCA1 E3 ubiquitin ligase activity through direct interaction with the BARD1 protein. J Biol Chem 278, 5255-5263 14. Linares, L. K., Hengstermann, A., Ciechanover, A., Muller, S., and Scheffner, M. (2003) HdmX stimulates Hdm2-mediated ubiquitination and degradation of p53. Proc Natl Acad Sci U S A 100, 12009-12014 15. Christensen, D. E., Brzovic, P. S., and Klevit, R. E. (2007) E2-BRCA1 RING interactions dictate synthesis of mono- or specific polyubiquitin chain linkages. Nat Struct Mol Biol 14, 941-948 16. Ozkan, E., Yu, H., and Deisenhofer, J. (2005) Mechanistic insight into the allosteric activation of a ubiquitin-conjugating enzyme by RING-type ubiquitin ligases. Proc Natl Acad Sci U S A 102, 18890-18895 17. Banerjee, A., Deshaies, R. J., and Chau, V. (1995) Characterization of a dominant negative mutant of the cell cycle ubiquitin-conjugating enzyme Cdc34. J Biol Chem 270, 26209-26215 18. Gu, H. D., and Roizman, B. (2003) The degradation of promyelocytic leukemia and Sp100 proteins by herpes simplex virus 1 is mediated by the ubiquitin-conjugating enzyme UbcH5a. Proc Natl Acad Sci U S A 100, 8963-8968 19. Markson, G., Kiel, C., Hyde, R., Brown, S., Charalabous, P., Bremm, A., Semple, J., Woodsmith, J., Duley, S., Salehi-Ashtiani, K., Vidal, M., Komander, D., Serrano, L., Lehner, P., and Sanderson, C. M. (2009) Analysis of the human E2 conjugating enzymes protein interaction network. Genome Res Aug 26 20. Stanyon, C. A., Liu, G., Mangiola, B. A., Patel, N., Giot, L., Kuang, B., Zhang, H., Zhong, J., and Finley Jr., R. L. (2004) A Drosophila protein-interaction map centered on cell- cycle regulators. Genome Biol 5, R96 21. Brzovic, P. S., Lissounov, A., Christensen, D. E., Hoyt, D. W., and Klevit, R. E. (2006) A UbcH5/ubiquitin noncovalent complex is required for processive BRCA1-directed ubiquitination. Mol Cell 21, 873-880 22. Wilson, R. C., Hughes, R. C., Flatt, J. W., Meehan, E. J., Ng, J. D., and Twigg, P. D. (2009) Structure of full-length ubiquitin-conjugating enzyme E2-25K (huntingtin- interacting protein 2). Acta Crystallogr Sect-F Struct Biol Cryst Commun 65, 440-444 23. Tokunaga, F., Sakata, S., Saeki, Y., Satomi, Y., Kirisako, T., Kamei, K., Nakagawa, T., Ka- to, M., Murata, S., Yamaoka, S., Yamamoto, M., Akira, S., Takao, T., Tanaka, K., and Iwai, K. (2009) Involvement of linear polyubiquitylation of NEMO in NF-kappa B activation. Nat Cell Biol 11, 123-U140 24. Kirisako, T., Kamei, K., Murata, S., Kato, M., Fukumoto, H., Kanie, M., Sano, S., Toku- naga, F., Tanaka, K., and Iwai, K. (2006) A ubiquitin ligase complex assembles linear polyubiquitin chains. EMBO J 25, 4877-4887 25. Hofmann, R. M., and Pickart, C. M. (1999) Noncanonical MMS2-encoded ubiquitin- conjugating enzyme functions in assembly of novel polyubiquitin chains for DNA repair. Cell 96, 645-653

121 26. Pornillos, O., Alam, S. L., Davis, D. R., and Sundquist, W. I. (2002) Structure of the Tsg101 UEV domain in complex with the PTAP motif of the HIV-1 p6 protein. Nat Struct Mol Biol 9, 812-817 27. Eddins, M. J., Carlile, C. M., Gomez, K. M., Pickart, C. M., and Wolberger, C. (2006) Mms2-Ubc13 covalently bound to ubiquitin reveals the structural basis of linkage- specific polyubiquitin chain formation. Nat Struct Mol Biol 13, 915-920 28. Lee, S., Tsai, Y. C., Mattera, R., Smith, W. J., Kostelansky, M. S., Weissman, A. M., Bo- nifacino, J. S., and Hurley, J. H. (2006) Structural basis for ubiquitin recognition and autoubiquitination by Rabex-5. Nat Struct Mol Biol 13, 264-271 29. Pinato, S., Scandiuzzi, C., Arnaudo, N., Citterio, E., Gaudino, G., and Penengo, L. (2009) RNF168, a new RING finger, MIU-containing protein that modifies chromatin by ubiquitination of histones H2A and H2AX. BMC Mol Biol 10, - 30. Michelle, C., Vourc'h, P., Mignon, L., and Andres, C. R. (2009) What Was the Set of Ubiquitin and Ubiquitin-Like Conjugating Enzymes in the Eukaryote Common Ances- tor? J Mol Evol 68, 616-628 31. Seufert, W., and Jentsch, S. (1990) Ubiquitin-conjugating enzymes UBC4 and UBC5 mediate selective degradation of short-lived and abnormal proteins. EMBO J. 9, 543- 550 32. Deng, L., Wang, C., Spencer, E., Yang, L., Braun, A., You, J., Slaughter, C., Pickart, C., and Chen, Z. J. (2000) Activation of the IkappaB kinase complex by TRAF6 requires a dimeric ubiquitin-conjugating enzyme complex and a unique polyubiquitin chain. Cell 103, 351-361 33. VanDemark, A. P., Hofmann, R. M., Tsui, C., Pickart, C. M., and Wolberger, C. (2001) Molecular insights into polyubiquitin chain assembly: crystal structure of the Mms2/Ubc13 heterodimer. Cell 105, 711-720 34. Yamamoto, M., Okamoto, T., Takeda, K., Sato, S., Sanjo, H., Uematsu, S., Saitoh, T., Yamamoto, N., Sakurai, H., Ishii, K. J., Yamaoka, S., Kawai, T., Matsuura, Y., Takeuchi, O., and Akira, S. (2006) Key function for the Ubc13 E2 ubiquitin-conjugating enzyme in immune receptor signaling. Nat Immunol 7, 962-970 35. Komander, D., Reyes-Turcu, F., Licchesi, J. D. F., Odenwaelder, P., Wilkinson, K. D., and Barford, D. (2009) Molecular discrimination of structurally equivalent Lys 63- linked and linear polyubiquitin chains. EMBO Rep 10, 466-473 36. Dikic, I., Wakatsuki, S., and Walters, K. J. (2009) Ubiquitin-binding domains - from structures to functions. Nat Rev Mol Cell Biol 10, 659-671 37. Winn, P. J., Religa, T. L., Battey, J. N., Banerjee, A., and Wade, R. C. (2004) Determi- nants of functionality in the ubiquitin conjugating enzyme family. Structure 12, 1563- 1574 38. Barabasi, A. L., and Oltvai, Z. N. (2004) Network biology: understanding the cell's functional organization. Nat Rev Genet 5, 101-113 39. Jeong, H., Mason, S. P., Barabasi, A. L., and Oltvai, Z. N. (2001) Lethality and centrality in protein networks. Nature 411, 41-42 40. Albert, R., Jeong, H., and Barabasi, A.-L. (2000) Error and attack tolerance of complex networks. Nature 206, 378-382 41. Hochstrasser, M. (2009) Origin and function of ubiquitin-like proteins. Nature 458, 422-429 42. Scheffner, M., Nuber, U., and Huibregtse, J. M. (1995) Protein ubiquitination involving an E1-E2-E3 enzyme ubiquitin thioester cascade. Nature 373, 81-83

122 43. Dye, B. T., and Schulman, B. A. (2007) Structural mechanisms underlying posttransla- tional modification by ubiquitin-like proteins. Annu Rev Biophys Biomol Struct 36, 131-150 44. Lee, I., and Schindelin, H. (2008) Structural insights into E1-catalyzed ubiquitin activation and transfer to conjugating enzymes. Cell 134, 268-278 45. Li, W., Tu, D. Q., Brunger, A. T., and Ye, Y. H. (2007) A ubiquitin ligase transfers preformed polyubiquitin chains from a conjugating enzyme to a substrate. Nature 446, 333-337 46. Worth, C. L., Gong, S., and Blundell, T. L. (2009) Structural and functional constraints in the evolution of protein families. Nat Rev Mol Cell Biol 10, 709-720 47. Winkler, G. S., Albert, T. K., Dominguez, C., Legtenberg, Y. I., Boelens, R., and Tim- mers, H. T. (2004) An altered-specificity ubiquitin-conjugating enzyme/ubiquitin- protein ligase pair. J Mol Biol 337, 157-165 48. Huang, D. T., Zhuang, M., Ayrault, O., and Schulman, B. A. (2008) Identification of conjugation specificity determinants unmasks vestigial preference for ubiquitin within the NEDD8 E2. Nat Struct Mol Biol 49. Lee, B. C., Park, K., and Kim, D. (2008) Analysis of the residue-residue coevolution network and the functionally important residues in proteins. Proteins 72, 863-872 50. Dominguez, C., Bonvin, A. M., Winkler, G. S., van Schaik, F. M., Timmers, H. T., and Boelens, R. (2004) Structural model of the UbcH5B/CNOT4 complex revealed by combining NMR, mutagenesis, and docking approaches. Structure 12, 633-644 51. Humphris, E. L., and Kortemme, T. (2007) Design of multi-specificity in protein interfaces. PLoS Comput Biol 3, 1591-1604 52. Kortemme, T. (2006) Prediction and design of molecular recognition. FASEB J 20, A1473-A1473 53. Kortemme, T., and Baker, D. (2004) Computational design of protein-protein interactions. Curr Opin Chem Biol 8, 91-97

123

Summary In eukaryotic organisms, nearly all biological processes rely on the covalent conjugation of one or multiple ubiquitin (Ub) molecules (ubiquitination). A triangular system composed of three types of key enzymes (ubiquitin-activating enzymes (E1s), ubiquitin-conjugating enzymes (E2s) and ubiquitin protein ligases (E3s) is responsible for the activation conjugation of Ub via several highly regulated protein interactions. First, E1s activate Ub in an ATP- dependent manner and subsequently interact with E2 enzymes followed by transfer of Ub to the catalytic cysteine of the E2s. Next, loaded E2 enzymes interact with E3 ligases to mediate the transfer to Ub and the covalent conjugation to substrates. Within this cascade, E3 ligases are the main determinants for substrate selection, while E2 enzymes mostly specify the outcome of ubiquitination reactions. Ubiquitin contains seven lysines and each of these can be used to build chains of different lengths. Apart from this, substrates can be modified with single ubiquitin molecules as well as with chains that are constructed via linkages between the N- and C-termini of Ub (linear ubiquitination). All these different forms and shapes have different effects on substrates, for example modification with a single ubiquitin is involved in transcription regulation and endocytosis, whereas conjugation with K48 or K63 chains targets substrates for proteasomal degradation or non-proteolytical purposes, respectively. E1-E2-E3 enzymes are operating in a triangular system. Apart from ubiquitin, several ubiquitin-like proteins (UBLs) have been identified, like SUMO, ISG15 and NEDD4, of which the majority is equipped with dedicated E1, E2 and E3 enzymes.

In humans, the ubiquitin system is characterized by two E1s, 35 E2s and over 500 E3s that cooperatively activate and conjugate ubiquitin. This expanded, pyramid-shaped system en- sures adequate regulation of ubiquitination in time and space. Within this enzymatic triangle, it is generally regarded that the two E1s interact with the cohort of E2 enzymes in a one- on-one fashion to transfer the activated Ub. In contrast, interactions between E2 and E3 enzymes are thought to be more selective, despite the relative high levels of sequence homology among the interaction domains of E2 and E3 family members.

In order to study E2-E3 interaction specificity, a comprehensive network of interactions between E2 and E3 domains is presented in Chapter 2 that describes the binary interactions between 35 UBC-folds and 250 E3 RING-finger domains generated with high-throughput yeast two-hybrid screens. Among the ~9000 pair-wise combinations tested, over 300 E2-E3 interactions were detected of which were unknown. Comparing the physical interactions from Y2H with literature-curated biochemically functional E2-E3 pairs reveals a good overlap, indicating the high quality of experimental interactions. Additional GST pull-downs between high- and low-interacting E2 and E3 domains also underscored the quality of the genome-wide E2-E3 network. A small fraction of domains tested displayed more interactors than one would expect, suggesting the presence of hub proteins in the E2-E3 network. This indicates a scale-free distribution of interactions between E2 and E3 enzymes, providing structural and functional redundancy and protection against network perturbations. Fur- thermore, E2 sequences and E3 binding potentials were studied between the K63-linkage specific E2 Ubc13(UBE2N) and UbcH5B(UBE2D2), which catalyzes mixed chain linkages. By selective substitution of E2-E3 interface residues in Ubc13(UBE2N) with the corresponding ones of UbcH5B(UBE2D2), the partial E3-interaction profile of UbcH5B(UBE2D2) could be established.

125 In Chapter 3, an in-depth overview of the evolutionary and structural diversity of the E2 enzyme super family is provided. Species-specific E2 diversity is discussed in relationship with E3 interaction selectivity and biochemical features, covering “true” ubiquitin E2s as well as E2s specific for UBLs. Furthermore, a shell-model is proposed, explaining how selectivity of E2 enzymes towards ubiquitin (-like modifiers) determines Ub/UBL conjugation pathways.

To illustrate the tight relationship between E2 sequence and E3 interaction potential in controlling E2-E3 interaction selectivity, Chapter 4 describes the comparison between the highly similar UbcH6(UBE2E1) and UbcH8(UBE2E2). Whereas UbcH6(UBE2E1) displays many E3 interactions, UbcH8(UBE2E2) only interacts with a few E3s. Substitution of an aspartic acid to a glutamic acid in UbcH8(UBE2E2), such that it resembles UbcH6(UBE2E1), alters the E3 interaction pattern to that of UbcH6(UBE2E1). By using a combination of molecular docking, molecular dynamics analysis and mutagenesis, a dynamic network of intra-molecular salt- bridges was unraveled that actively controls the availability of the aspartic acid present in UbcH8(UBE2E2). This network controls the equilibrium of this residue between an open and bridges state, thereby determining E3 interactions. Mutation of single network components alters this equilibrium towards a state of switches in E3 interactions.

Finally, Chapter 5 discusses these finding and reflect these observations against the current knowledge of E2-E3 complexes, network systems biology and structural biology.

126 Samenvatting Voor normale mensen is wetenschappelijk onderzoek vaak compleet onbegrijpelijk (niet in de laatste plaats door het gebruik van allerlei jargon…). Daarom voor iedereen een handlei- ding, in drie stappen, tot ubiquitinoloog…

Ons lichaam bestaat uit miljoenen cellen die het mogelijk maken om ons te laten lezen, schrijven en alle dingen te doen die we dagelijks doen. Om dit in goede banen te leiden, be- vatten de meeste cellen een celkern en in die celkern bevindt zich ons erfelijk materiaal, het DNA. Dit DNA is opgebouwd uit lange slierten van vier typen chemische stoffen (A, T, C en G) en de volgorde van die stoffen is grotendeels bepalend voor wie (of wat) we zijn. DNA op zichzelf is een redelijk saaie stof. Voordat wij er iets mee kunnen, moet er eerst een kopie van gemaakt worden. Deze kopie, het boodschapper RNA, wordt vervolgens gebruikt om te maken waar het echt om gaat, eiwitten. Eiwitten zijn, naast gekookt of gebakken bij het ontbijt, de belangrijkste stoffen die het leven mogelijk maken. Het eiwit hemoglobine vervoerd zuurstof in ons bloed en geeft het zijn rode kleur, onze huid en ons haar bestaat uit eiwitten, net als onze spieren. Eiwitten hebben veel verschillende functies, zo zijn er eiwitten die zorgen voor transport, eiwitten die structuur en vorm geven aan cellen en ook eiwitten die werken als kleine robotjes om zo bepaalde bio- chemische reacties te versnellen (enzymen). In feite zijn eiwitten te vergelijken met lange kettingen van verschillende kralen. In ons lichaam worden twintig soorten kralen gebruikt om deze kettingen te maken en deze noemt men aminozuren. Ook aminozuren zijn chemische stoffen, maar zij worden gebruikt als bouwstenen voor eiwitten. De volgorde waarin deze aminozuren in de ketting (het eiwit) terechtkomen wordt bepaald door de volgorde van de vier chemische stoffen A, T, C en G in ons DNA. Eiwitten zijn, over het algemeen, geen rechte kettingen, maar worden op een speciale manier in elkaar gevouwen. De manier waarop dit gebeurd wordt voor een groot deel bepaald door de chemische kenmerken van de verschillende aminozuren. Eenmaal gevouwen, dan bepaald de vorm van een eiwit hoe dat specifieke eiwit zal gaan werken. Zo zijn er eiwitten die er als een balletje uitzien, maar ook eiwitten die op springveren lijken, omdat een balletje of een springveer nodig is voor bepaalde processen. Als alles goed gaat, bepaald de volgorde van A, T, C, en G in ons DNA de aminozuurvolgorde en dus ook de vorm en functie van eiwitten. Maar het gaat niet altijd goed. Soms verandert de volgorde van A, T, C, of G in ons DNA (spontaan of doordat onze voorouders veranderingen in hun DNA hadden). Soms hebben deze veranderingen geen effect op de eiwitten die worden gevormd, maar soms ook wel en dan kunnen slechte eiwitten gevormd worden. De- ze slechte eiwitten zijn schadelijk en kunnen leiden tot allerlei ziekten, zoals kanker. Gelukkig hebben eiwitten hebben niet het eeuwige leven. Zodra ze worden aangemaakt en hun bepaalde functie verricht hebben, worden ze weer afgebroken. Om deze afbraak in goede banen te leiden (we willen niet dat een belangrijk gezond eiwit te vroeg wordt afgebroken of dat een slecht eiwit blijft bestaan, met alle gevolgen van dien!) is er een uitgebreid eiwit compostsysteem aanwezig in onze cellen, compleet met een shredder (proteasoom). Dit compostsysteem bestaat uit drie families van enzymen, die nauw samenwerken om een of een hele keten van kleine eiwitvlaggetjes (ubiquitine) aan te brengen op bepaalde eiwitten. Het ubiquitine vlaggetje is een klein eiwit wat veel voor komt door de hele cel en heeft 76 aminozuren en verspreid door de keten zijn zeven speciale aminozuren met de naam lysine aanwezig. Het compostsysteem bestaat uit drie groepen van speciale enzymen. In de mens komen twee ubiquitine-activerende enzymen (E1), 35 ubiquitine-conjugerende enzy-

127 men (E2) en meer dan 500 ubiquitine-eiwit ligases (E3) voor. Zowel de E2 en de E3 enzymen lijken onderling, wat betreft aminozuurvolgorde (de kralen in de ketting) en hoe deze eiwitten zich vouwen, heel erg op elkaar. Voordat het ubiquitine op andere eiwitten kan worden gezet, moet het eerst opgeladen worden. Hiervoor grijpt de E1 het kleine ubiquitine vast en laadt het op met behulp van een klein chemisch stofje (ATP). Vervolgens bindt de E1 aan een van de E2 enzymen en geeft het opgeladen ubiquitine over. De E2s (met daarop ubiquitine) binden vervolgens aan de E3 enzymen. E3 enzymen zijn speciaal omdat ze de E2s kunnen binden maar ook de eiwitten waarop een ubiquitine vlaggetje moet worden gezet. Zodra dit gebeurt, wordt het ubiquitine van de E2 op het doel eiwit gezet, bijeengehouden door de E3s.

Oorspronkelijk dacht men dat zodra eiwitten gemerkt werden met ubiquitine deze automa- tisch zouden worden versnipperd door de proteasoom tot afzonderlijke, kleinere eiwit deel- tjes en losse aminozuren. Tegenwoordig is bekend dat er verschillende vormen van eiwitvlaggetjes bestaan, variërend van een enkel vlaggetje tot zelfs hele ketens. Deze ketens worden opgebouwd door ubiquitine vlaggetjes op elkaar te zetten via een van de zeven lysine aminozuren. Eiwitten met ketens die zijn opgebouwd via bijvoorbeeld lysine 48 worden snel naar de shredder gestuurd om vermalen te worden terwijl ketens die zijn opgebouwd via lysine 63 over het algemeen niet worden versnipperd. In veel gevallen is de E2 bepalend voor welke ubiquitine-vorm er op de eiwitten gezet wordt. In deze gevallen is het dus ook de E2 die direct bepaald wat er met het eiwit gaat gebeuren. Daarom is het belangrijk te weten welke E2 met welke E3 een binding aan zal gaan.

In het compostsysteem gaat het E1 enzym één op één bindingen aan met alle E2 enzymen, maar hoe de hele groep E2s daarna bindingen aangaan met E3s was nog onbekend (in ieder geval wisten we dat voor een klein groepje E2s en E3s dit heel willekeurig verloopt). Deze willekeurigheid, of selectiviteit, is op zich vrij vreemd, want zowel E2s en E3s lijken heel erg op elkaar. Om dit te onderzoeken zijn in Hoofdstuk 2 de bindingen tussen grote groepen menselijke E2s en E3s onderzocht door gebruik te maken van een techniek die yeast two- hybrid genoemd wordt. Yeast two-hybrid (of gist twee-hybride) gebeurd in bakkersgist, hetzelfde type wat gebruikt wordt voor het maken van bier en brood. Deze techniek is erg ge- schikt om bindingen tussen twee eiwitten te onderzoeken. In feite is het te vergelijken met een zaklamp die werkt op twee batterijen. De zaklamp zal alleen gaan schijnen als je twee batterijen hebt die in de juiste volgorde tegen elkaar komen. In yeast two-hybrid zijn de batterijen vervangen door twee eiwitten (LexA en B42) en zodra deze vlak bij elkaar komen, kun je aan de gistcellen aflezen of er een binding tussen twee eiwitten is opgetreden (en dit kun je natuurlijk vergelijken met het licht uit de zaklamp). Alleen, LexA en B42 komen niet van- zelf bij elkaar. De truc is om aan LexA een E2 enzym en aan B42 een E3 enzym vast te maken. Als deze enzymen elkaar binden, komen ook de LexA en de B42 bij elkaar en zullen de gistcellen blauw kleuren en groeien op een speciaal dieet. In totaal zijn er 35 verschillende E2 enzymen aan LexA vastgemaakt en 250 E3 enzymen aan B42. Vervolgens zijn al deze “batterijen” in een matrix tegen elkaar uitgezet, zodat er 35 x 250 (een kleine 9000!) verschillende E2-E3 bindingen onderzocht zijn. Wat is nu de belangrijkste bevinding? We wisten al dat de aminozuurvolgorde en de vorm van zowel E2s als E3s sterk op elkaar lijken, maar toch blijken sommige enzymen meer bindingspartners te hebben dan wat je zou verwachten. Het blijkt dat er in de hele matrix maar enkele van zulke veelbinders zijn (deze veelbinders worden hubs genoemd) en dat het merendeel van de enzymen minder of geen bindingen vertoond.

128 Het blijkt zelfs dat een aantal van deze veelbinders een belangrijke rol vervullen in het bio- chemische proces van ubiquitinering. Wat betreft E2-E3 bindingen kan er onderscheid gemaakt worden tussen een fysieke binding en een functionele binding. Fysieke bindingen zijn getest met behulp van yeast two-hybrid en als deze bijna 9000 bindingen vergeleken worden met de bindingen die zijn beschreven in de literatuur (en die in staat zijn om daadwerkelijk ubiquitine te koppelen) dan blijkt dat er een hele goede overlap is. Dit houdt in dat fysieke binding een goede en natuurgetrouwe manier is om functionele E2-E3 paren te voorspellen. Ten slotte hebben we de aminozuurvolgorde en de E3-bindingen van twee specifieke E2 enzymen (UbcH5B en Ubc13) verder onderzocht. Hoewel deze E2s veel op elkaar lijken, maakt Ubc13 vooral K63-gelinkte ubiquitine-ketens, terwijl UbcH5B een mix van ubiquitine- verbindingen maakt. Door een aantal aminozuren in Ubc13 te vervangen door aminozuren van UbcH5B, is het gelukt om het patroon van E3 binding van Ubc13, (gedeeltelijk) te veranderen naar dat van UbcH5B. Dit kan belangrijke gevolgen hebben voor eiwitten die normaal gesproken K63-ketens krijgen, maar die nu een ander type keten ontvangen.

In Hoofdstuk 3 is een getailleerd overzicht van alle E2 enzymen die voorkomen in gist, plan- ten en dieren weergegeven. Hier wordt verder ingegaan op hoe E2s hun E3 herkennen en hoe deze binding op molecuul- en atoomniveau verloopt. Daarnaast worden de neefjes van ubiquitine (eiwitten die sterk op ubiquitine lijken en op een vergelijkbare manier aan eiwitten worden vastgemaakt) besproken. Het blijkt dat deze neefjes specifieke E1s, E2s en E3s gebruiken en in dit hoofdstuk wordt besproken hoe het komt dat een specifieke ubiquitine E2 niet het neefje herkend.

Vervolgens wordt in Hoofdstuk 4 dieper ingegaan op de relatie tussen de aminozuurvolgorde van E2 enzymen en hoe E3 enzymen gebonden worden door deze E2s. Uit Hoofdstuk 2 wisten we al dat twee E2s die zeer sterk op elkaar lijken, UbcH6 en UbcH8, hele uiteenlo- pende patronen van E3 binding opleverden. Door aminozuren van UbcH8 te veranderen in overeenkomende aminozuren van UbcH6, is het gelukt om met deze nieuwe UbcH8 hetzelfde patroon van E3 binding als dat van UbcH6. De verandering in chemische eigenschappen van het verwisselde aminozuur is heel erg klein en daarom is het zeer verrassend dat dit ge- beurde. Daarnaast hebben we op het niveau van afzonderlijke aminozuren door middel van uitgebreide computer modellen en berekeningen ontdekt dat omringende aminozuren be- palen hoe het verwisselde aminozuur E3 bindingen aangaat. Door nu deze omringende aminozuren te veranderen in UbcH8, is het gelukt om weer hetzelfde E3 patroon van UbcH6 te krijgen. Dit geeft aan dat het verwisselde aminozuur onder streng toezicht staat van omringende aminozuren. Ten slotte wordt in Hoofdstuk 5 alles nog even kritisch onder de loep genomen en worden de bevindingen die beschreven staan in dit proefschrift vergeleken met gegevens uit de vak- literatuur.

129

Curriculum vitae Sjoerd van Wijk werd op 16 mei 1979 geboren in Rotterdam. In 1996 behaalde hij zijn HAVO diploma aan de C.S.G. Angelus Merula in Spijkenisse. In datzelfde jaar begon hij de opleiding tot medisch-biologisch laboratoriumingenieur aan de Hogeschool Rotterdam welke met een afstudeerstage in de vakgroep Algemene Heelkunde (Universiteit Maastricht) onder leiding van Prof. Dr. W. A. Buurman werd afgesloten. Hier heeft hij onderzoek gedaan naar de pa- thofysiologie van ischemie/reperfusie schade in de nier in muis en rat. Vervolgens begon hij als researchanalist bij de vakgroep Humane Biologie onder leiding van Prof. Dr. R. P. Men- sink, waar hij betrokken was in het onderzoek naar lipide-homeostase in muizen en proef- personen. In 2002 startte bij met de opleiding Biologische Gezondheidskunde aan de Univer- siteit Maastricht, die hij combineerde met student-assistentschappen bij de vakgroepen Humane Biologie en Gezondheidsrisico Analyse en Toxicologie. Na een afstudeerstage bij de vakgroep Moleculaire Genetica onder leiding van Dr. J.-W. Voncken waarin onderzocht werd hoe Polycomb eiwit-complexen samengesteld zijn, studeerde hij in 2004 cum laude af. In datzelfde jaar begon hij als assistent in opleiding in het Laboratorium voor Fysiologische Chemie, Universitair Medisch Centrum Utrecht onder leiding van Prof. Dr. H. Th. M. Tim- mers. De resultaten van dat onderzoek staan beschreven in dit proefschrift. Vanaf 1 decem- ber 2009 is hij werkzaam als post-doc in het Instituut voor Biochemie II aan de medische fa- culteit van de Goethe Universiteit in Frankfurt am Main te Duitsland onder leiding van Prof. Dr. I. Dikic. Hier zal hij zich gaan bezighouden met systeem-wijd onderzoek naar ubiquitin- signaling.

131

Publications Sjoerd J.L. van Wijk & Geja J. Hageman (2005) Poly(ADP-ribose) polymerase-1 mediated caspase-independent cell death after ischemia/reperfusion. Free Rad. Biol. Med. 39, 81-90

Sjoerd J.L. van Wijk, Sjoerd J. de Vries, Patrick Kemmeren, Anding Huang, Rolf Boelens, Alex- andre M.J.J. Bonvin & H.Th. Marc Timmers (2009) A comprehensive framework of E2-RING E3 interactions of the human ubiquitin-proteasome system. Mol. Syst. Biol. 5, 295

Sjoerd J.L. van Wijk & H.Th. Marc Timmers (2009) The family of ubiquitin-conjugating enzymes (E2s): deciding between the life and death of proteins. FASEB J. (in press)

Sjoerd J.L. van Wijk*, Adrien S.J. Melquiond*, Alexandre M.J.J. Bonvin & H.Th. Marc Tim- mers. A butterfly effect in protein-protein interactions: how minimal differences in E2 sequence determine RING E3 interaction selectivity (Manuscript in preparation) * Equally contributing authors

133

Dankwoord Zo, dit is het dan… Ruim vier jaar promotieonderzoek en een proefschrift verder. Zoals ge- bruikelijk is in Nederland, wordt vaak na afloop van langdurige projecten de balans opge- maakt (en vaak komen daar de spreekwoordelijke lijken uit de kast…). Telt u even mee? In de afgelopen vijf jaar zijn er meer dan 500.000 spots gecontroleerd, om en nabij 400 liter medium gegoten en ontelbaar veel sequence-runs gecontroleerd (en ik kan nog wel even doorgaan...) Hoewel het op sommige momenten anders leek, lijkt dit proefschrift het gevolg van een uitgebreid netwerk van samenwerkende “nodes” in een “scale-free” netwerk.

Geachte promotor Timmers, beste Marc. Door jou ben ik in aanraking gekomen met ubiquitine, gist en Utrecht. Onze besprekingen gingen over van alles, van vakanties tot wolligheid en ingewikkelde E2 mutanten, maar uiteindelijk is het resultaat daar. Gedurende de afgelopen vijf jaar merkte ik dat jouw rol als begeleider steeds meer veranderde in die van advi- seur. Uiteindelijk denk ik dat ik (mede daardoor) redelijk zelfstandig geworden ben en daar- voor ben ik je dankbaar.

Geachte leden van de beoordelingscommissie, Prof. Klumperman, Sixma, Verrijzer en Boe- lens, dank voor het beoordelen van mijn manuscript, ondanks uw overvolle agenda’s. Prof. Bos en Boelens, dank u voor uw bijdragen in mijn AIO-commissie, uw kritische open aan- merkingen zijn zeer gewaardeerd.

Mijn (ex-)collega’s. Radhika, my dear roommate. Your Western’s are the nicest I’ve ever seen. Thank you for being such a nice colleague! I’m aware of all the troubles you’ve encoun- tered during your PhD, but I’m sure you will manage to finish everything nicely. Marcus, it was a pleasure to work, smoke or discuss with you. Keep up being critical, even now we are enrolled in our German exchange program... Marijke, ondanks alle bijnaampjes, grappen en grollen: bedankt voor alles (zowel voor het zijn van een goede, maar wel kleine ;-), kamerge- noot en de onbetwiste (maar kleine ;-)) sequence-koningin!). Gianpiero, Mr. G, thanks for being a nice colleague and a friend, the nice cooking and all the talks we had. I’m sure you PhD will be fine; make sure that you are doing what you want to do! I don’t know... Andree, strudel, eigenlijk kwam je te laat, maar ik ben superblij dat we nog half-half samengewerkt hebben. Zoals jij je iedere maandagmorgen naar het werk sleept, het is niet te geloven... Het was erg leuk om je collega te zijn! Heel veel succes met je proeven! Rick, min of meer het stokje overgenomen van Koen, heel veel succes met menin. Met jou ideeën weet ik zeker dat je een goede kant op zal gaan! Koen, veel succes in de kliniek! Petra, eindelijk klaar met het FRAP-verhaal, nu kun je de echt leuke dingen gaan doen! Met een beetje mazzel zien we elkaar op een microscopy-meeting? Succes!!! Florence, all the best in France! I’m sure that you guys will be super-parents! Pim, success met de stamcellen! Nikolay, good luck with the stem cells! I’m sure I will read nice stories of you guys somewhere! Hetty, moeder van het lab, nu zijn eindelijk de VMT-bakken leeg! Richard, veel success met je ITCs, peptide- syntheses en eiwitzuiveringen. Klaas, bedankt voor alle tips en trics van het pinnen, je ziet ‘t, zonder pinning tools, geen proefschrift! Succes in Engeland. Andrea, Christina en Marianne, bedankt voor alle kleine en grotere secretaresse-dingetjes waarmee jullie mij hebben geholpen. Zonder jullie was het niet gelukt. Marjoleine, bedankt voor het bestellen van honderden primers en gistplaten.

135 Dear colleagues from NMR, without your help I would have been lost! Sjoerd, ontzettend bedankt voor je hulp! Zonder jouw programmeer- en eiwitkennis zou hoofdstuk 2 er mis- schien heel anders uit hebben gezien! Anding, thanks for providing me with your RING-finger cDNAs and good luck finishing your PhD and I hope to see you soon at some ubiquitin- conference! Gert, dank voor je interesse en ik kijk uit naar je resultaten! Alexandre, thanks for your collaborations and your fruitful thoughts. It was nice to work with you. Rolf, bedankt voor alle goede ideeën. Adrien, without your help there would only be half a chapter 4! Thanks for the dynamics analysis and thinking about the E2 mutants. All the best with being a young father and I’m looking forward to future collaborations.

Collega’s van de Vermeulen-groep: Michiel, Pascal, Danny en Nelleke: bedankt voor alle lab- praatjes en goede tips! Collega’s van de Bos-, Burgering-, Kops en Medema-groepen: bedankt. Ook de Holstege-groep wil ik bedanken voor hun interesse. Loes, bedankt voor het proofreaden van hoofdstuk 3 en veel succes met je onderzoek! Ik kijk nu al uit naar je pa- pers! Sake en Joris, ik zal onze gist-gerelateerde praatjes missen, veel succes. Thanassis, good luck with your ubiquitin experiments! Patrick, bedankt voor de spotanalyse software!

Aalt-Jan, ondanks dat er in dit boekje niets staat over de E2 motieven, wil ik je toch bedanken voor de samenwerking. Laten we hopen dat we het toch nog ergens kunnen onderbren- gen!

Many thanks to my new Frankfurt collegues: Ivan, Birgit, Birgit, Rebecca, Yonathan, Sigrid, Farshid, Koraljka, Fumiyo, Panchali, David, Doris, Tobias, Oliver, Philip, Masuda, Nicola and Magda for giving me a new home.

Sandra, Marloes en de kleine Robin. Sandra en Marloes, ik ken jullie al heel lang, maar toch blijven jullie me verbazen. Wonen in Ede, trouwen en een kindje! Jullie zijn de beste vrien- den die je je kunt wensen! En San, ik ben blij dat we destijds achterliepen! Lieve kleine Ro- bin, jij bent het mooiste bewijs dat het gaat om liefde. Isabelle, Matthias en de kleine Amelie. Bedankt voor onze vriendschap. Hoewel we elkaar tegenwoordig niet zo heel vaak spreken, althans minder vaak als voorheen, is iedere keer weer zoals het was. Lieve kleine Amelie, jij kunt je geen betere ouders wensen!

Lieve schoonfamilie, Jo, Miet, Diana en Ruud, hoewel Utrecht en Nieuwstadt/Maastricht ver uit elkaar liggen, bedankt voor jullie continue interesse de afgelopen vijf jaar. Het is nu eindelijk voorbij!

Annabelle, Michel, Sara en Fenne, het is altijd leuk om jullie weer te zien. Bedankt voor alle lieve telefoontjes, verjaardagen en bezoekjes! Lieve Sara en Fenne, zodra ik jullie zie, vergeet ik alles!

Jannika en Stijn a.k.a. Shaniqua en de Don. Bedankt voor alle interesse, telefoontjes en etentjes. Ik hoop dat dat nog lang doorgaat!

Pieter en Petra, Pietje en Peetje, jullie ken ik al mijn hele leven en jullie hebben mij gevormd tot wie ik nu ben. Ik ben waarschijnlijk niet de eerste die zegt dat een mens het product is

136 van zijn genen en zijn opvoeding. In beide gevallen heb ik geluk gehad met jullie. Bedankt voor alles!

Tenslotte, lieve Ron. Mijn laatste woorden zijn en zullen voor jou zijn. Samen hebben we diepe dalen en hoge pieken bewandeld en ik heb er geen seconde spijt van dat ik je opmerk- te tijdens het roken. Ik heb heel erg veel respect voor hoe jij je leven leeft en hoe je, ondanks al mijn weekenden en avonden van werk (maar bovenal mijn laksheid), mij hebt gesteund en alles hebt gerelativeerd. Nu er 400 kilometers tussen ons liggen, ben je dichterbij dan ooit. En hoewel het hier anders lijkt, kom jij op de eerste plaats.

137

Won't you walk me through the Tiergarten? Won't you walk me through it all, darling? Doesn't matter if it is raining Won't you walk me through it all?

Even if the sun, it is blazing Even if the snow, it is raging All the elements, we must conquer To get to the other side of town

I have suffered shipwreck against your dark brown eyes I have run aground against your broken down smiles Believe me when I tell you I have no place to go But to go where the wild flowers grow and the stone gardens bloom.

Won't you walk me through the Tiergarten? Won't you walk me through it all, darling? Doesn't matter if it is raining, We'll get to the other side of town.

Rufus Wainwright - Tiergarten

139