Insights into NEDD8 function and the regulation of its
conjugation system
Dissertation
zur Erlangung des akademischen Grades eines Doktors der
Naturwissenschaften (Dr. rer. nat.)
vorgelegt von
Dana Pagliarini
an der
Mathematisch‐Naturwissenschaftliche Sektion Fachbereich Biologie
Tag der mündlichen Prüfung: 07. Juni 2013 1. Referent: Prof. Dr. Martin Scheffner 2. Referent: Prof. Dr. Thomas U. Mayer
Für meine Eltern
Table of contents
Abbreviations i
Abstract ii
Zusammenfassung iii
1. Introduction 1
1.1 Ubiquitin‐proteasome system 1
1.1.1 Ubiquitin‐conjugation cascade 2
1.1.2 Modes of ubiquitination 4
1.1.3 Ubiquitin recycling and the proteasome 6
1.2 Ubiquitin‐like proteins (UBLs) 6
1.2.1 SUMO 7
1.2.2 Other UBLs 8
1.3 NEDD8 9
1.3.1 Substrates and functions of NEDD8 10
1.3.1.1 Cullins 10
1.3.1.2 NEDD8 and transcriptional regulation 12
1.3.1.3 Further substrates and functions of NEDD8 13
1.3.2 NEDD8‐conjugation cascade 14
1.3.2.1 APPBP1/UBA3, the NEDD8‐activating enzyme 16
1.3.2.2 NEDD8‐conjugating enzymes 18
1.4 Aim of the studies 21
2. Material and Methods 22
2.1 Material 22
2.1.1 Solutions and media 22
2.1.2 Chemicals and Reagents 24
2.1.3 Bacterial strains 25
2.1.4 Mammalian cell lines 26
2.1.5 Antibodies 26
2.1.6 Primers 27
2.1.7 Plasmids constructed and used in this study 28 2.1.8 Other plasmids used in this study 29
2.1.9 DNA‐ and protein markers 30
2.2 Methods 30
2.2.1 PCR and cloning 30
2.2.1.1 Polymerase chain reaction (PCR) 30
2.2.1.2 Gene synthesis 30
2.2.1.3 Site directed mutagenesis 30
2.2.1.4 Restriction digest 31
2.2.1.5 Agarose gel electrophoresis 31 2.2.1.6 Purification of DNA from agarose gels 31
2.2.1.7 Ligation 31
2.2.1.8 Transformation of DNA into chemical competent E. coli 31 2.2.1.9 Preparation of DNA in low and high scale 31 2.2.1.10 Measurement of DNA and RNA concentrations 32
2.2.1.11 DNA sequencing 32
2.2.2 Maintenance of bacterial cultures and mammalian cell lines 32 2.2.2.1 Bacterial cultivation and preparation of glycerol stocks 32 2.2.2.2 Maintenance of mammalian cell lines 32 2.2.2.3 Freezing of cells in liquid nitrogen 32 2.2.3 Protein expression and ‐purification 32 2.2.3.1 Expression and purification of GST‐fusion proteins in
E. coli 32
2.2.3.2 Expression and purification of His‐tagged proteins in
E. coli 33
2.2.3.3 Expression and purification of NEDD8 and ubiquitin for
click reaction 33 2.2.3.4 Expression and purification of PCNA for click reaction 33 2.2.3.5 In vitro translation 34
2.2.4 Protein analysis 34
2.2.4.1 Bradford assay 34
2.2.4.2 SDS‐polyacrylamide gel electrophoresis (SDS‐PAGE) 35 2.2.4.3 Coomassie Blue and colloidal Coomassie staining 35
2.2.4.4 Fluorography 35
2.2.4.5 Western Blot 35
2.2.5 In vitro assays 36
2.2.5.1 Methanol‐Chloroform precipitation of proteins 36 2.2.5.2 GST‐pulldown assay 36 2.2.5.3 Affinity chromatography 36 2.2.5.4 In vitro NEDDylation assay 37
2.2.5.5 Thioester assay 37
2.2.5.6 Cu(I)‐catalyzed Huisgen azide‐alkyne cycloaddition 37
2.2.5.7 Transient transfection 37 2.2.5.8 TNN cell lysis and ß‐galactosidase assay 38 2.2.5.9 Immunoprecipitation 38 2.2.5.10 Cycloheximide chase 38 2.2.5.11 Cellular fractionation 39 2.2.5.12 Preparation of total RNA 39 2.2.5.13 Reverse transcription 39 2.2.5.14 Antibody purification 39
2.2.6 In cellulo assays 40
2.2.6.1 In cellulo ubiquitination and NEDDylation assays 40 2.2.6.2 Immunofluorescence 40
3. Results 41
3.1 AutoNEDDylation of NEDD8‐conjugating enzymes increases the
affinity to their cognate E1 41
3.1.1 Ubc12 and Nce2 form a thioester bond with NEDD8 but not with ubiquitin
42 3.1.2 NEDD8‐conjugating enzymes are autoNEDDylated 43 3.1.3 HPNI mutants of the NEDD8‐conjugating enzymes are not impaired in
thioester formation but in autoNEDDylation 45 3.1.4 AutoNEDDylation of the NEDD8 E2 enzymes predominantly occurs in
their N terminus 46
3.1.5 AutoNEDDylation enhances the affinity of Ubc12 and Nce2 to the NEDD8
E1 enzyme APPBP1/UBA3 50
3.1.6 Fusion of NEDD8 to its E2s leads to an enhanced localization in the
nucleus 51
3.2 PCNA as a new substrate for the NEDD8‐conjugation pathway 54 3.2.1 PCNA interacts with NEDD8 and ubiquitin 54 3.2.2 PCNA is NEDDylated in cells being dependent on K164 56 3.2.3 PCNA NEDDylation depends on the activity of APPBP1/UBA3 in cells 58 3.2.4 NEDDylation of PCNA is enhanced by the E3 ligase Rad18 60 3.2.5 NEDD8‐conjugating enzymes do not bind to PCNA in cells
60
3.2.6 PCNA Y211F, a phosphorylation deficient mutant, is NEDDylated in cells 61 3.2.7 Cu(I)‐catalyzed Huisgen azide‐alkyne cycloaddition as a tool to study
functions of NEDDylated PCNA 62
3.3 A new isoform of the NEDD8‐conjugating enzyme Nce2 65 3.3.1 mRNA encoding Nce2 isoform 2 is expressed in HEK293T cells 67
3.3.2 Tertiary structure prediction for Nce2 isoform 2 reveals an unstructured, flexible C terminus
67 3.3.3 Nce2 isoform 2 forms a thioester bond with NEDD8 but not with
ubiquitin and is NEDDylated in vitro 68 3.3.4 Nce2 isoform 2 is NEDDylated and ubiquitinated in cells 70 3.3.5 Nce2 isoform 2 has a shorter half‐life than isoform1 and is degraded by
the proteasome 72
3.3.6 Nce2 isoform 1 and 2 differ in their subcellular localization 74 3.3.7 Development of an antibody specifically recognizing Nce2 isoform 2 75
4. Discussion 77
4.1 AutoNEDDylation as a regulatory mechanism of NEDD8‐conjugating enzymes 77 4.1.1 NEDD8‐conjugating enzymes are autoNEDDylated in their unique
N terminus
77 4.1.1.1 Indications for endogenous autoNEDDylation of Ubc12 and Nce2 77 4.1.1.2 The HPNI‐motif of Ubc12 and Nce2 is important for isopeptide
bond formation 78
4.1.1.3 The N termini of Ubc12 and Nce2 are crucial for an efficient
autoNEDDylation 79 4.1.2 Possible functions of autoNEDDylation 81 4.1.2.1 AutoNEDDylation as regulator of the subcellular localization of
Ubc12 and Nce2 82
4.1.2.2 Acetylation of NEDD8 E2 enzymes as a competitive modification
to NEDDylation 82
4.1.2.3 AutoNEDDylation enhances the affinity of NEDD8 E2s to
APPBP1/UBA3 83
4.1.2.4 Further possible functions and impacts of autoNEDDylation 85
4.2 Interplay between PCNA and the NEDD8 system 88
4.2.1 PCNA as an interaction partner of NEDD8 88 4.2.2 PCNA as a substrate for NEDD8 89 4.2.2.1 Evidence for NEDDylation of PCNA in vitro and in cellulo 89 4.2.2.2 Effects of MLN4924 on the NEDDylation of PCNA 91 4.2.2.3 Hints for possible functions of NEDDylated PCNA 92 4.2.2.4 Click reaction as tool to identify functions of monoNEDDylated
PCNA 94
4.3 A second isoform of Nce2 with individual properties 96 4.3.1 Nce2 isoform 2 as a splice variant with an extended C terminus 96
4.3.2 mRNA of the second isoform of Nce2 is present in HEK293T cells 96 4.3.3 Nce2 isoform 2 is active in vitro and can be NEDDylated 97 4.3.4 The cellular localization distinguishes Nce2 isoform 2 from isoform 1 99 4.3.5 A C‐terminal unstructured region promotes insolubility in bacteria and
rapid degradation in cells 100
4.3.6 Possible functions of Nce2 variants 101
5. References 105
Eidesstattliche Erklärung 124
Danksagung 125
Abbreviations
aa amino acid(s)
bp base pairs
cDNA complementary DNA DMSO Dimethylsulfoxide
dNTP Deoxynucleoside triphosphate DTT Dithiothreitol
EDTA Ethylenediaminetetraacetic acid E1 UBL‐activating enzyme
E2 UBL‐ conjugating enzyme
E3 UBL ligase
FCS Fetal calf serum
GST Glutathione‐S‐transferase HA‐tag Hemagglutinin‐tag His‐tag 6x Histidin‐tag IP Immunoprecipitation
IPTG Isopropyl‐β‐D‐thiogalactopyranoside kDa kilo Dalton
mRNA messenger RNA
Ni‐NTA Nickel‐nitrilotriacetic acid OD optical density
ONPG Ortho‐nitrophenyl‐β‐galactoside PBS Phosphate buffered saline PCR polymerase chain reaction rpm revolutions per minute RT reverse transcription SDS Sodium dodecyl sulfate UBL ubiquitin‐like protein
wt wild‐type
Abstract
NEDD8 belongs to the family of ubiquitin‐like proteins which share a common basic structure. In an enzymatic cascade, NEDD8 can be attached to other proteins via its C terminus (“NEDDylation”). NEDDylation plays an important role in various cellular processes including cell cycle, transcription or the regulation of protein stability. Previous studies revealed NEDD8 to have many yet uncharacterized substrates. During this work, the sliding clamp PCNA, which functions in replication and DNA damage repair, was identified as a new substrate for NEDDylation. PCNA is modified with NEDD8 at the same lysine (K164) residue that can be targeted for SUMOylation and ubiquitination. Rad18, the E3 ligase for ubiquitination of PCNA upon DNA damage, enhances monoNEDDylation of PCNA. We furthermore expressed NEDD8 and PCNA containing non‐natural amino acids that can be used for the formation of a triazole linkage via click reaction. This “monoNEDDylated” PCNA provides a tool for a functional characterization in the future.
In a further part of this study, regulation and function of the NEDD8‐conjugating enzymes Ubc12 and Nce2 were investigated. To date, not much is known about if and how the E2 enzymes of the NEDD8‐conjugation cascade are regulated. We here provide evidence that these enzymes are autoNEDDylated in the N‐terminal extension of the catalytic core domain, which is not present in most other E2 enzymes. This autoNEDDylation of Ubc12 and Nce2 appears to enhance their affinity to APPBP1/UBA3, the E1 enzyme of the NEDD8‐conjugation cascade. Using NEDD8‐E2 fusion proteins an increased efficiency in NEDD8 transfer from the E1 to the E2 enzyme compared to wt Ubc12 and Nce2 was observed. However, the exact function of autoNEDDylation in cells, which might be a change in substrate specificity, enhanced substrate NEDDylation or a switch‐off mechanism for the whole cascade, still needs to be determined.
To gain further insights into function and regulation of NEDD8‐conjugating enzymes, a second isoform of Nce2 was characterized which has recently been identified in a screen for splice variants. Nce2 isoform 2 mRNA was verified to be present in cells. To prove the existence of endogenous protein, an antibody specifically recognizing the second, but not the first isoform of Nce2 was developed. Tertiary structure prediction of the new isoform revealed an unstructured, flexible C terminus. Using overexpressed isoform 2, a difference in the subcellular localization of both isoforms was demonstrated. In contrast to Nce2 isoform 1, isoform 2 is probably not autoNEDDylated, but regulated via ubiquitination leading to a shorter half‐life. In future studies, the functionality and the role of Nce2 isoform 2 in cells need to be investigated.
In conclusion, identification of PCNA as a new substrate of NEDD8, evidence for a regulation of NEDD8‐conjugating enzymes by autoNEDDylation as well as the new isoform of Nce2 provide insights into yet unknown physiological functions of NEDD8.
Zusammenfassung
NEDD8 gehört zur Familie der Ubiquitin‐ähnlichen Proteine, die eine gemeinsame Grundstruktur aufweisen. In einer enzymatischen Kaskade kann NEDD8 über seinen C‐
Terminus an andere Proteine konjugiert werden („NEDDylierung“). Die NEDDylierung spielt eine wichtige Rolle in verschiedenen zellulären Prozessen wie z.B. dem Zellzyklus, der Transkription oder der Regulation der Proteinstabilität. Frühere Studien wiesen bereits auf die Existenz einiger weiterer, bisher noch nicht charakterisierter Substrate von NEDD8 hin. Im Rahmen dieser Arbeit konnte PCNA, das sowohl für die Replikation als auch für die DNA‐
Reparatur benötigt wird, als neues Substrat für NEDD8 identifiziert werden. PCNA wird an demselben Lysinrest (K164) monoNEDDyliert, der auch SUMOyliert und ubiquitiniert werden kann. Diese MonoNEDDylierung von PCNA wird durch die E3 Ligase Rad18 verstärkt, die auch die Ubiquitinierung von PCNA nach DNA‐Schäden vermittelt. Während dieser Studien wurden zudem NEDD8 und PCNA mit nicht‐natürlichen Aminosäuren exprimiert, die in der sog. „Click Reaktion“ eine Triazolbindung eingehen können. Das so gebildete „monoNEDDylierte“ PCNA kann zukünftig für eine funktionelle Charakterisierung verwendet werden.
Ein weiterer Teil dieser Arbeit beinhaltet die Untersuchung der Regulation und der Funktion der NEDD8‐konjugierenden Enzyme Ubc12 und Nce2. Bisher war nicht sehr viel darüber bekannt, ob und wie die E2‐Enzyme der NEDDylierungskaskade reguliert werden. Im Zuge dieser Arbeit konnte der Beweis erbracht werden, dass beide Enzyme autoNEDDyliert werden können. Diese AutoNEDDylierung findet dabei in der N‐terminalen Verlängerung der katalytischen Kerndomäne der E2s statt, die in den meisten anderen bekannten E2‐Enzymen nicht zu finden ist. Die AutoNEDDylierung von Ubc12 und Nce2 scheint die Affinität zu APPBP1/UBA3, dem E1‐
Enzym der NEDDylierungskaskade, zu verstärken. Unter Zuhilfenahme von NEDD8‐E2 Fusionsproteinen konnte gezeigt werden, dass der Transfer von NEDD8 von dem E1‐ zum E2‐
Enzym im Vergleich zu Ubc12 und Nce2 wt effizienter ist. Die genaue Funktion der AutoNEDDylierung der NEDD8 E2s in Zellen gilt es allerdings noch zu untersuchen.
Möglicherweise resultiert sie in einer Änderung der Substratspezifität, einer erhöhten SubstratNEDDylierung oder einem Abschaltmechanismus für die gesamte Kaskade.
Um weitere Kenntnisse über Funktion und Regulation von NEDD8‐konjugierenden Enzymen zu erlangen, wurde eine zweite Isoform von Nce2 charakterisiert, die kürzlich in einem Screen nach Splice‐Varianten entdeckt wurde. Die mRNA dieser Isoform konnte tatsächlich in Zellen nachgewiesen werden. Zwecks Detektion des endogenen Proteins wurde ein Antikörper entwickelt, der spezifisch die zweite, aber nicht die erste Isoform von Nce2 erkennt. Das erstellte Modell der Tertiärstruktur der Isoform 2 lässt einen unstrukturierten und flexiblen C‐Terminus erkennen. Mittels Überexpression von Isoform 2 konnte ein Unterschied in der subzellulären
Lokalisation beider Isoformen nachgewiesen werden. Überdies wird Isoform 2 im Gegensatz zur ersten Isoform wahrscheinlich nicht autoNEDDyliert, aber über Ubiquitinierung reguliert, die eine verkürzte Halbwertszeit des Proteins zur Folge hat. In zukünftigen Studien muss die Funktionalität und die Rolle der zweiten Isoform von Nce2 in der Zelle untersucht werden.
Die Identifikation von PCNA als neues Substrat von NEDD8, der Beweis einer Regulation der E2‐
Enzyme mittels AutoNEDDylierung sowie der Nachweis einer neuen Isoform von Nce2 ermöglichen bisher unbekannte Einblicke in die physiologische Rolle von NEDD8.
1. Introduction
Protein function, localization and stability are regulated by posttranslational modifications which involve for example small chemical groups such as acetyl‐, methyl‐ or phosphate moieties, or larger groups serving as membrane anchors such as palmitoyl‐ or myristoyl groups (Han and Martinage, 1992; Magee and Courtneidge, 1985). With the discovery of the protein ubiquitin and its ability to serve as posttranslational modification in the 1980s, a new chapter of the functions of these modifications has begun. Ubiquitin was found to target proteins for proteasomal degradation thereby offering the cell a second major pathway to regulate protein levels in addition to lysosomal degradation. In the meantime, several proteins with high structural similarity to ubiquitin were identified most of which act as posttranslational modifiers controlling various cellular functions (Hershko and Ciechanover, 1998; Kerscher et al., 2006).
1.1 Ubiquitin‐proteasome system
Ubiquitin is a small protein of 76 aa and 8.6 kDa which is highly conserved among eukaryotic organisms (Hershko and Ciechanover, 1998). Structurally, ubiquitin is characterized by a globular domain with a β‐grasp fold and a flexible C terminus (Figures 1 and 4B) (Vijay‐Kumar et al., 1987).
Figure 1. Crystal structure of ubiquitin revealing the characteristic βgrasp fold
Ribbon representation of human ubiquitin (protein data base 1UBQ; modeled with Pymol). The structure of ubiquitin is characterized by four antiparallel β‐sheets grasping an α‐helix. The C terminus of ubiquitin protrudes from the globular domain (Vijay‐Kumar et al., 1987).
C
N
1.1.1 Ubiquitin‐conjugation cascade
In humans, expression of ubiquitin from one of the four different genes first leads to the formation of an inactive precursor protein (Baker and Board, 1987; Lund et al., 1985; Wiborg et al., 1985). Being rapidly processed by ubiquitin‐specific proteases, the functionally important C‐
terminal double glycine motif of ubiquitin is exposed (Wilkinson, 1997).
In an enzymatic cascade, ubiquitin is covalently attached to other proteins via its C terminus (“ubiquitination”). Ubiquitination occurs through several consecutive steps catalyzed by three (or four) different classes of enzymes: ubiquitin‐activating enzymes (E1), ubiquitin‐conjugating enzymes (E2) and ubiquitin ligases (E3) (Figure 2). In some cases, an E4 enzyme may be required for the formation of ubiquitin chains (Hershko and Ciechanover, 1998; Hoppe, 2005).
In a first step, the carboxyl group of the C‐terminal glycine of ubiquitin forms a high energy thioester linkage with an active site cysteine residue of one of two E1 enzymes, UBA1 or UBA6.
This step is ATP‐dependent and involves the formation of a ubiquitin adenylate intermediate (Jin et al., 2007; Pelzer et al., 2007; Pickart, 2001). The E2 is then able to accept activated ubiquitin from its cognate E1, forming a thioester linkage. In the last step, which in most cases requires the presence of an E3, ubiquitin is transferred to the lysine residue of a substrate protein by the formation of an isopeptide bond (Pickart, 2001) (Figure 2). An asparagine residue in the conserved HPNI/V motif next to the catalytic cysteine of the E2 enzyme plays a crucial role for accomplishing this transfer, as it might be important for oxyanion intermediate stabilization during lysine attack (Wu et al., 2003b).
Structural and functional analyses indicate that E3s consist of at least two functional domains:
one domain (i.e. RING/RING‐like or HECT) is interacting with the cognate E2 while the other mediates the specific interaction with the substrate, thereby conferring substrate specificity on the whole conjugation cascade. Depending on their mode of action, E3 ligases can be divided into two major classes: HECT and RING/RING‐like ligases (Metzger et al., 2012). HECT ligases contain a HECT (Homologous to E6AP Carboxyl Terminus) domain which consists of an N‐terminal and a C‐terminal lobe. The N‐terminal lobe is necessary for the interaction with the E2 enzyme, whereas the C‐terminal lobe contains a catalytic cysteine which forms a thioester linkage with ubiquitin before transferring it to the substrate (Huang et al., 1999).
Figure 2. Ubiquitinconjugation cascade
Ubiquitin is first activated by the E1 and then transferred to one of a number of E2 enzymes. RING E3 ligases facilitate the conjugation of ubiquitin to the substrate by acting as adaptors between E2s and substrates. In contrast, HECT E3 ligases form thioester bonds with ubiquitin and subsequently transfer it to the substrate. By repeating these steps, the substrate can be modified with several ubiquitin moieties (modified from (Di Fiore et al., 2003)).
Being encoded by more than 600 genes in mammals, RING ligases represent the largest family of E3 ligases (Li et al., 2008). This type of ligase is characterized by a RING (Really Interesting New Gene) domain containing the consensus motif C‐X2‐C‐X(9‐39)‐C‐X(1‐3)‐H‐X(2‐3)‐C/H‐X2‐C‐X(4‐
48)‐C‐X2–C (X stands for any aa), which is stabilized by two zinc ions (Lovering et al., 1993). By bringing the E2 enzyme and the substrate into close proximity, RING ligases allow a direct transfer of ubiquitin from the E2 to the substrate.
The family of RING ligases comprises ligases that function as monomer, homo‐ or heterodimer or as large multi‐subunit complexes. Most of the known RING ligases possess intrinsic E3 ligase activity, but there are also RING domain containing proteins that do not show activity by themselves, e.g. Bard1, Bmi1 and HdmX. Heterodimerization of these E3 ligases with another RING ligase (Brca1, Ring1B and Hdm2, respectively), which involves the interaction of the RING domains, leads to the stimulation of ligase activity of the latter (Hashizume et al., 2001; Linares et al., 2003; Wang et al., 2004). Moreover, other RING E3 ligases like RNF4 or TRAF6 form
ATP AMP+PPi
homodimers to execute their function (Liew et al., 2010; Yin et al., 2009). Recent data about the mechanism of ubiquitin transfer by RNF4 sheds light on why dimerization is necessary for RING ubiquitin ligase activity. Dimerization of the RING domains facilitates the interaction with both components of the ubiquitin‐charged E2 at the same time: one monomer binds to ubiquitin and the other one to the E2. By altering the conformation of the active site of the E2, the E2‐ubiquitin thioester bond is then activated and ubiquitin can be transferred to the substrate (Plechanovova et al., 2011; Plechanovova et al., 2012).
Being composed of multiple subunits, the APC/C complex and cullin‐RING ligases such as SCF form an additional type of RING E3 ligases. APC/C, a complex consisting of at least twelve subunits including the RING domain containing protein APC11, plays a major role in cell cycle regulation, especially in mitotic progression. Dependent on the substrate adaptor bound, a specific set of substrates is recognized, ubiquitinated and degraded by the proteasome (reviewed in (Peters, 2006)). The SCF complex belongs to the largest family of ubiquitin E3 ligases known, the cullin‐RING ligases (see chapter 1.3.1.1). It consists of Cullin1 as scaffold protein, the RING ligase RBX1, Skp1 as adaptor protein and an exchangeable F‐box protein recognizing a specific substrate. Not only cell cycle inhibitors like p21 or p27, but also oncogenic proteins such as Cyclin E or c‐Myc turned out to be substrates for SCF complexes, underlining its important role in controlling the cell cycle (summarized in (Kitagawa et al., 2009)).
1.1.2 Modes of ubiquitination
In most cases, ubiquitination of substrates occurs via isopeptide bond formation between the C‐
terminal glycine residue of ubiquitin and the ‐amino group of a lysine residue in the substrate.
Furthermore, there are few publications showing that serine, threonine and cysteine residues as well as the N‐terminal amino group of some proteins are used for the attachment of ubiquitin (Cadwell and Coscoy, 2005; Ciechanover and Ben‐Saadon, 2004; Shimizu et al., 2010; Wang et al., 2007b).
In addition to mono‐ or multiubiquitination where single ubiquitin moieties are conjugated to one or several distinct residues in the substrate protein, respectively, ubiquitin is also capable of forming “chains” (polyubiquitination) (Figure 3). Monoubiquitination was found to be involved in DNA damage response and endocytosis (reviewed in (Hicke, 2001)). For instance, Rad6‐ and Rad18‐dependent ubiquitination of the sliding clamp PCNA leads to the recruitment of translesion synthesis polymerases that are crucial for inducing the DNA damage tolerance pathway (Hoege et al., 2002; Kannouche et al., 2004). Moreover, multiubiquitination plays a particular role in the internalization and lysosomal degradation of plasma membrane receptors, e.g. receptor tyrosine kinases (Haglund et al., 2003).
Each of the seven internal lysine residues of ubiquitin (K6, K11, K27, K29, K33, K48 and K63) can be used for isopeptide bond formation with the C‐terminal carboxyl group of another ubiquitin moiety and hence, for the formation of polyubiquitin chains (Figure 3). Best characterized and most abundant are K11‐, K48‐ and K63‐linked chains (Ye and Rape, 2009). As an example, the E3 ligase complex APC/C triggers degradation of its mitotic substrates via formation of K11‐linked ubiquitin chains (Jin et al., 2008). Furthermore, the ubiquitin‐
conjugating enzyme Ubc6 which is involved in ER‐associated degradation (ERAD), a protein quality control system mainly localized in the ER membrane, is ubiquitinated and degraded via K11‐linkage of ubiquitin in yeast (Xu et al., 2009). Nonetheless, K48‐linked chains are a more common signal to target proteins for proteasomal degradation (Chau et al., 1989). For a long time it was believed that a minimal chain length of four ubiquitin moieties linked via K48 is absolutely required for the recognition by the proteasome (Thrower et al., 2000). This hypothesis is contradicted by recent studies indicating that monoubiquitinated or multiubiquitinated proteins like PAX3 or p105, respectively, can also be recognized and degraded by the proteasome (Boutet et al., 2007; Kravtsova‐Ivantsiv et al., 2009; Shabek et al., 2012). In contrast to K11‐ and K48‐chains, K63‐linked chains are relevant for a variety of non‐
proteolytic cellular processes like endocytosis, DNA repair or activation of kinases (Deng et al., 2000; Duncan et al., 2006; Spence et al., 1995). Mixed and forked chains, as well as chains formed on internal lysine residues other than K11, K48 and K63 of ubiquitin are still under intense investigation (Figure 3).
Figure 3. Modes of ubiquitination
Ubiquitin is primarily attached to the ‐amino group of one or more lysine residues of a substrate (mono‐ and multiubiquitination, respectively) or, in rare cases, to the N terminus of a substrate (not shown). It contains seven internal lysine residues all of which can serve as an acceptor for another ubiquitin moiety, thereby forming ubiquitin chains (polyubiquitination). K11‐, K48‐, K63‐linked and linear chains, which are linked via C‐ and N terminus of two ubiquitin moieties, function in proteasomal degradation, DNA‐repair or intracellular signaling, whereas the function
1.1.3 Ubiquitin recycling and the proteasome
Processing of ubiquitin precursor proteins as well as recycling of ubiquitin is carried out by deubiquitinating enzymes (DUBs) which mainly exhibit cysteine protease activity. The two major classes of DUBs are UCHs (Ubiquitin COOH‐terminal Hydrolases) that preferentially cleave ubiquitin from substrates and USPs (Ubiquitin‐Specific Proteases) that additionally hydrolyze isopeptide bonds between two ubiquitin moieties. In addition, OUTs (otubain proteases), JAMM metalloproteases and MJDs were found to function as DUBs (summarized in (Sorokin et al., 2009)).
Some DUBs are associated with or even part of the 26S proteasome which degrades 80‐90 % of intracellular proteins (Rock et al., 1994). The 26S proteasome consists of the 20S core particle and two 19S regulatory particles. The core proteasome is composed of 14 α‐ and 14 β‐subunits forming four heptameric rings which build a channel in whose inner part the target protein is hydrolyzed. Three of the β‐subunits possess proteolytic activities ensuring efficient cleavage of the target: subunit β1 exhibits caspase‐like activity, β2 has trypsin‐like and β5 chymotrypsin‐
like activity. The regulatory particles contain subunits that fulfill three important functions:
interaction with the ubiquitinated substrate, cleaving ubiquitin from the substrate (by isopeptidases) and unfolding the target protein (by ATPases) (reviewed in (Murata et al., 2009;
Sorokin et al., 2009)).
1.2 Ubiquitin‐like proteins (UBLs)
The family of ubiquitin‐like proteins (UBLs) consists of more than a dozen members in mammals that are structurally characterized by the β‐grasp fold (Figures 1 and 4). UBLs use a similar conjugation mechanism as ubiquitin which involves the subsequent action of activating and conjugating enzymes and, in most cases, that of ligases. Functions of UBLs vary from regulation of protein stability, interactions or localization, autophagy and pre‐mRNA splicing to regulation of inflammation and development (see table 1) (summarized in (Herrmann et al., 2007; Kerscher et al., 2006; Schulman and Harper, 2009).
UBLs are conserved among eukaryotes. Nevertheless, it is speculated that they have arisen from a common ancestor in prokaryotes which plays a role in sulfur metabolism. The reason for this hypothesis lies in the existence of proteins in E. coli that reveal a ubiquitin‐like structure and exhibit parallels to the ubiquitin‐conjugation cascade. One of those proteins is MoaD, a sulfur carrier protein being involved in the biosynthesis of the molybdenum cofactor Moco (Rivers et al., 1993). The double glycine motif at the C terminus of MoaD is adenylated by MoeB, which shows similarities to the ubiquitin E1 enzyme. Adenylation of MoaD leads to its activation, subsequent formation of a thiocarboxylate at its C terminus and the insertion of sulfur into the
Moco precursor protein. Interestingly, the eukaryotic ubiquitin‐like protein URM1 might represent the connection to the prokaryotic system since it functions in sulfur transfer, too (reviewed in (Hochstrasser, 2009)).
Table 1. Ubiquitinlike proteins
Summary of the known ubiquitin‐like proteins in humans, their functions and enzymes involved in their conjugation;
NA: data not available; NC: no conjugation to other proteins (adapted from (Herrmann et al., 2007)).
1.2.1 SUMO
The small ubiquitin‐related modifier (SUMO), which is 11.5 kDa in size, is one of the best characterized ubiquitin‐like proteins. In vertebrates, there are four SUMO paralogues revealing the typical β‐grasp fold with a short N‐terminal extension (Bayer et al., 1998; Bohren et al., 2004;
Kamitani et al., 1998). In contrast to SUMO‐2 and SUMO‐3, the elongated N terminus of SUMO‐1 does not contain a SUMOylation site and thus, it is not able to form chains (Saitoh and Hinchey, 2000; Tatham et al., 2001). All SUMO family members need to be processed in order to expose
Ubiquitinlike protein E1, E2, E3 enzymes Functions
NEDD8 E1: APPBP1/UBA3
E2: Ubc12, Nce2 E3: e.g. Hdm2, RBX1/2
Transcriptional regulation, protein stability,
proteasomal degradation…
SUMO1‐3 E1: SAE1/SAE2
E2: Ubc9
E3: RanBP2, Pc2, PIAS
Transcriptional regulation, protein stability, protein localization…
ISG15 E1: UBE1L
E2: UbcH8 IFN‐induced immune response
FAT10 E1: UBA6
E2: USE1 Proteasomal degradation, apoptosis
LC3/GABARAP E1: ATG7
E2: ATG3 Autophagy
ATG12 E1: ATG7
E2: ATG10 Autophagy
FUB1 NA Regulation of leukocyte
functions
URM1 E1: Uba4 Oxidative stress response
UFM1 E1: Uba5
E2: Ufc1 Endoplasmic stress response
HUB1 NC Pre‐mRNA splicing
member which is probably not conjugated to substrates (Owerbach et al., 2005). According to their subcellular localization, SUMO‐1, ‐2 and ‐3 are able to carry out distinct functions: SUMO‐1 is mostly found at the nuclear membrane whereas SUMO‐2 is localized in nuclear bodies and nucleoli and SUMO‐3 in the cytoplasm and in nucleoli (reviewed in (Hay, 2005)).
The SUMO‐conjugation cascade starts with its activation by the heterodimeric E1 enzyme SAE1/SAE2 with SAE2 containing the active site cysteine. Ubc9, the SUMO conjugating enzyme which is particularly found in the nucleus, is able to directly transfer SUMO to substrates without involvement of an E3 enzyme (Desterro et al., 1999; Schwarz et al., 1998). Most SUMO substrates contain a special binding motif for Ubc9 consisting of the four amino acids [I/V/L]‐K‐X‐[D/E]
(X stands for any aa), which also contains the lysine residue that in turn is SUMOylated (Bernier‐
Villamor et al., 2002; Johnson, 2004). SUMO E3 ligases belong to the family of RanBP2‐, Pc2‐ or PIAS E3 ligases that have specific substrates and differ in their localization in the cell (Hay, 2005;
Johnson and Gupta, 2001; Kagey et al., 2003; Pichler et al., 2002). For SUMOylation, more than 100 substrates have been identified so far that are partially overlapping for SUMO‐1 and SUMO‐
2/‐3. These substrates can mainly be grouped into transcriptional regulators, nuclear envelope proteins, signaling proteins and cell membrane proteins (Rosas‐Acosta et al., 2005; Vertegaal et al., 2004). For instance, SUMOylation of RanGAP1, a factor which is necessary for nucleo‐
cytoplasmic transport, regulates its localization (Matunis et al., 1996). Another substrate of SUMO is PML which needs to be SUMOylated to initiate the formation of PML nuclear bodies that are involved in important cellular functions like transcriptional regulation (Shen et al., 2006;
Sternsdorf et al., 1997).
A unique feature of the SUMOylation cascade is the regulation of substrate specificity by autoSUMOylation of the E2 enzyme Ubc9. SUMOylation of Ubc9 occurs in its N‐terminal α‐helix and does not affect thioester bond formation with SUMO. Indeed, autoSUMOylation of Ubc9 leads to inhibited SUMOylation of RanGAP1 and enhanced SUMOylation of Sp100, which is at least partially due to a SUMO‐interaction motif (SIM) in Sp100 (Knipscheer et al., 2008).
1.2.2 Other UBLs
In the 1980s, ISG15 (Interferon‐Stimulated Gene product of 15 kDa) was the first ubiquitin‐like protein to be identified. ISG15 exhibits a dimeric ubiquitin structure. As the name already indicates, ISG15 expression is interferon‐inducible which is in line with its role in the induction of inflammation (D'Cunha et al., 1996; Haas et al., 1987; Korant et al., 1984).
FAT10 is characterized by two head‐to‐tail linked ubiquitin‐like domains which were shown to be able to directly interact with the proteasome. Accordingly, FAT10 offers a ubiquitin‐
independent way of delivering proteins for proteasomal degradation (Fan et al., 1996; Hipp et al., 2005). The double glycine motif at the C terminus of FAT10 is activated by UBA6, transferred
to USE1 and attached to other proteins as shown by the presence of FAT10 conjugates in cells (Aichem et al., 2010; Jin et al., 2007; Pelzer et al., 2007; Raasi et al., 2001). However, only four substrates of FAT10 have been identified to date, among these are USE1 which is autoFAT10ylated in a cis‐mechanism, the ubiquitin‐activating enzyme UBA1, p53, and the autophagosomal receptor p62 (Aichem et al., 2012; Aichem et al., 2010; Li et al., 2011; Rani et al., 2012).
Two UBLs, the LC3/GABARAP family and ATG12, are involved in autophagy, a conserved mechanism in eukaryotes to degrade macromolecules and organelles. Although showing no obvious homology to ubiquitin, they exhibit a similar structure and conjugating system (Ichimura et al., 2000; Sugawara et al., 2004; Suzuki et al., 2005). The LC3/GABARAP family (in mammals; Atg8 in yeast) is conjugated to phosphatidylethanolamine thereby initiating autophagosome formation. Autophagosomes are double‐membrane vesicles which deliver their content to the lysosome (Geng and Klionsky, 2008; Ichimura et al., 2000; Tanida et al., 2004).
Interestingly, aggregated ubiquitinated proteins were found to be targets for so called “selective”
autophagy. Some adaptor proteins are associated with the autophagosome via LC3 and interact with ubiquitinated proteins at the same time, thereby leading to lysosomal degradation of the respective protein (reviewed in (Kirkin et al., 2009)). Another UBL, ATG12, is conjugated constitutively to ATG5 which is necessary for completion of autophagosome formation (Geng and Klionsky, 2008; Ichimura et al., 2000).
1.3 NEDD8
NEDD8 (Neural precursor cell‐Expressed Developmentally Down‐regulated 8) was first discovered in a subset of genes that are down‐regulated during mouse brain development and is highly conserved from yeast to humans. With 60 % identity and 80 % homology to ubiquitin, NEDD8 forms its closest relative among the family of ubiquitin‐like proteins (Kumar et al., 1992;
Kumar et al., 1993). Human NEDD8 is expressed as an inactive, 81 aa precursor protein which needs to be processed in order to be conjugated to substrates (Kamitani et al., 1997). Similar to ubiquitin, the conjugation of NEDD8 involves three consecutive steps being catalyzed by the heterodimeric NEDD8 E1 enzyme APPBP1/UBA3, one of the two E2 enzymes Ubc12 or Nce2, and an E3 enzyme (Gong and Yeh, 1999; Huang et al., 2009). This process called “NEDDylation”
will be described in detail in section 1.3.2.
NEDD8 is a 9.1 kDa protein with a high structural similarity to ubiquitin. Its globular β‐grasp fold includes four antiparallel β‐sheets and one helix on top of them (Figure 4). In addition, surface charge distribution resembles that of ubiquitin: one face contains hydrophobic residues whereas the other one is acidic (Whitby et al., 1998).
A B
Figure 4. Crystal structure of NEDD8 (A) and overlay with ubiquitin and SUMO1 (B)
(A) The globular NEDD8 structure reveals a β‐grasp fold with four antiparallel β‐sheets and an α‐helix on top of them, being characteristic for ubiquitin and all UBLs (protein data base 1NDD; modeled using Pymol). The C‐terminal double glycine motif is marked in green.
(B) Overlay of ubiquitin (blue), SUMO‐1 (green) and NEDD8 (red). Ubiquitin and the UBLs SUMO‐1 and NEDD8 exhibit a highly similar structure characterized by the β‐grasp fold (Welchman et al., 2005).
Investigation of NEDD8 expression in different tissues using northern blot and immuno‐
cytochemical analysis revealed NEDD8 mRNA is enriched in brain and skeletal muscle, and that NEDD8 protein is predominantly found in the nucleus whereas ubiquitin is equally distributed in the cell (Kamitani et al., 1997).
In mice, knockout of the catalytic subunit of the NEDD8 E1 enzyme, UBA3, was shown to be lethal in utero at the periimplantation state. The reason for this phenomenon lies in the involvement of the NEDD8 pathway in cell cycle progression and morphogenesis, underlining its indispensability for early development (Tateishi et al., 2001). In S. cerevisiae, however, depletion of the NEDD8 homologue Rub1, as well as the respective E1 or the E2 enzymes does not affect normal cell growth (Liakopoulos et al., 1998).
1.3.1 Substrates and functions of NEDD8
1.3.1.1 Cullins
In 1998, Cullin4a was described as the first substrate of the NEDD8‐conjugation system (Osaka et al., 1998). In the meantime, it turned out that modification of all canonical cullins with NEDD8 is crucial to execute their function (Hori et al., 1999; Ohh et al., 2002). Cullins are components of multi‐protein complexes, the cullin‐RING ligases, which play important roles in cell growth, development, signal transduction, transcriptional control, genomic integrity and tumor suppression (see also 1.1.1). In mammals, there are six canonical cullins (Cullin1, Cullin2, Cullin3, Cullin4a, Cullin4b, Cullin5) and three atypical cullins (APC2, Cullin7 and PARC) known
N N
C C
which together build more than 500 distinct multi‐subunit complexes (Petroski and Deshaies, 2005a; Skaar et al., 2007; Zachariae et al., 1998).
Cullin‐RING ligases usually consist of cullins serving as scaffold proteins, substrate adaptors, which recognize and bind the respective substrate, and a RING domain protein with the catalytic activity to ubiquitinate the substrate. NEDDylation of the cullin subunit at a conserved lysine residue leads to a conformational change which in turn recruits the ubiquitin‐loaded E2 enzyme (Duda et al., 2008; Kawakami et al., 2001). In addition, NEDDylation seems to promote dimerization of cullin‐RING ligases through their substrate recognition subunit (Wimuttisuk and Singer, 2007). Recently, it was suggested that rotation of the RING domain of the E3 also plays a crucial role in the activation of the cullin‐RING complex (Calabrese et al., 2011). Both the NEDD8‐ and the ubiquitin E3 ligase activity of the complex are carried out by either RBX1 or RBX2 (Petroski and Deshaies, 2005a). Moreover, Dcn1 in yeast and DCNL proteins in humans serve as a scaffold‐type E3 ligases which interact with the respective cullin and the NEDD8 E2 at the same time thereby enhancing NEDDylation of cullins and thus, the ubiquitination activity of cullin‐RING complexes (Kurz et al., 2008; Monda et al., 2013; Yang et al., 2007).
As already mentioned, cullin complexes share a general composition. Dependent on the cullin, however, several different substrate adaptors can be used. So called SCF complexes consist of Cullin1, RBX1, the adaptor protein Skp1 and an F‐box protein which specifically interacts with the substrate to be degraded (Lyapina et al., 1998). One of the substrates for the SCF complex with Skp2 as F‐box protein is p21, an important regulator of cell proliferation and differentiation. Phosphorylation of p21 by Cdk2‐Cyclin E enhances its recognition and ubiquitination by SCFSkp2, leading to its rapid degradation (Bornstein et al., 2003). Additionally, not only cell cycle inhibitors but also cell cycle activators like Cyclin E are substrates for SCFSkp2 (Nakayama et al., 2000). In contrast to Cullin1, Cullin2 and Cullin5 assemble with elongin B/C as adaptor and a SOCS‐box containing protein as substrate receptor. A well‐studied substrate of the Cullin2‐RING ligase is HIFα, an important player in oxygen metabolism. A complex formed by HIFα and HIFβ under hypoxic conditions controls the expression of several genes like VEGF or erythropoietin (Maxwell, 2003). Oxygen‐dependent hydroxylation of HIFα leads to its recognition by the tumor suppressor protein pVHL, a substrate receptor of the Cullin2‐complex, and its subsequent ubiquitination and degradation (Ivan et al., 2001; Jaakkola et al., 2001;
Petroski and Deshaies, 2005a; Yu et al., 2001).
Deactivation of cullin‐RING complexes is achieved through removal of NEDD8 by the CSN5 subunit of the COP9 signalosome which belongs to the family of JAMM metalloproteases (Cope et al., 2002; Lyapina et al., 2001). Another mechanism to silence cullin‐RING complexes is the binding of Cand1 (Cullin‐Associated and NEDDylation‐Dissociated 1) which inhibits NEDDylation and the assembly of the whole complex by specifically binding to the region of the
that Cand1 also serves as an important exchange factor for cullin‐RING ligase adaptors (Pierce et al., 2013).
1.3.1.2 NEDD8 and transcriptional regulation
In the last decade, several new substrates of NEDD8 were described, giving insights into the role of NEDD8 in various cellular functions. Interestingly, a majority of these substrates is involved in transcriptional regulation.
The tumor suppressor protein p53, which plays a major role in the regulation of cell cycle arrest and apoptosis, is not only a substrate for ubiquitin but also for ubiquitin‐like proteins such as SUMO or NEDD8 (Kruse and Gu, 2009; Rodriguez et al., 1999; Scheffner et al., 1993; Xirodimas et al., 2004). NEDDylation of p53 by the RING ligase Hdm2 was shown to inhibit its transcriptional activity. In the case of p53, Hdm2 displays a dual specificity since it had also been described as an E3 ligase for the ubiquitination of p53 (Honda et al., 1997; Xirodimas et al., 2004).
Modification of p53 with NEDD8 is also promoted by FBXO11 and specifically inhibited by the histone acetyltransferase Tip60 (Abida et al., 2007; Dohmesen et al., 2008). Interestingly, C‐
terminal fusions of p53 with ubiquitin and NEDD8 to mimic its modification with these proteins showed that p53‐ubiquitin is rather found in the cytoplasm, whereas fusions of p53 with NEDD8 localize in the nucleus (Carter and Vousden, 2008).
Another member of the p53 family, TAp73, also serves as a substrate for Hdm2‐dependent NEDDylation. Modification of TAp73 with NEDD8 inhibits its transcriptional activity which can in part be explained by its localization to the cytoplasm (Watson et al., 2006).
In 2008, some ribosomal proteins were found to be modified with NEDD8 causing enhanced stability (Xirodimas et al., 2008). Further investigation of the ribosomal protein L11 revealed that its NEDDylation leads to a localization to the nucleolus (Sundqvist et al., 2009). Upon nucleolar stress, L11 is deNEDDylated and recruited to promoters of p53 regulated genes where it interacts with several co‐activators. In addition, binding of L11 to Hdm2 at these promoter sites inhibits interaction of p53 with Hdm2, thereby promoting transactivation of p53 target genes. Interestingly, ribosome biogenesis is not affected under deNEDDylation conditions in spite of reduced L11 levels (Mahata et al., 2011).
NEDDylation also seems to play an important role in the regulation of NFκB activity. On the one hand, suppression of the transcriptional activity of NFκB by BCA3 is dependent on its modification with NEDD8 (Gao et al., 2006). On the other hand, TRIM40‐catalyzed NEDDylation of IKKγ, an inhibitor of NFκB signaling, enhances the repression of NFκB (Noguchi et al., 2011).
Another protein which is regulated by NEDDylation is AICD, the intracellular domain of the amyloid precursor protein (APP). APP is found in plaques that accumulate in brains of Alzheimer patients. Cleaving of APP by secretases leads to the formation of AICD amongst others, whose role in Alzheimer development and progression is only poorly understood. Modification of AICD
with NEDD8 prevents the interaction with its co‐activator Fe65 and the histone acetyltransferase Tip60, resulting in an inhibition of the transactivator function for genes involved in e.g. cell growth and motility (Lee et al., 2008; Muller et al., 2008).
1.3.1.3 Further substrates and functions of NEDD8
Modification with NEDD8 does not only play roles in transcriptional regulation but also in the regulation of protein stability. The ribosomal protein L11 and the E3 ligase Hdm2 as well as PINK1, a protein involved in Parkinson´s disease, show an enhanced stability upon modification with NEDD8 (Choo et al., 2012; Xirodimas et al., 2004; Xirodimas et al., 2008).
In addition, NEDDylation regulates the stability of distinct receptors. Ubiquitination of EGFR is mediated by the RING‐ligase c‐Cbl, leading to its internalization and lysosomal degradation.
C‐Cbl is also capable of NEDDylating EGFR and therefore displays a dual specificity as Hdm2 does for modification of p53 (Oved et al., 2006; Xirodimas et al., 2004). Modification of EGFR with NEDD8 leads to an increased turnover rate caused by intensified ubiquitination (Oved et al., 2006). In the case of steroid hormone receptors, NEDD8 was even found to be required for their ubiquitination and degradation. Therefore, inactivation of the NEDD8 pathway might be involved in the development of steroid hormone dependent tumors (Fan et al., 2003; Fan et al., 2002).
Parkin, a RING‐type E3 ligase which is frequently mutated in patients suffering from Parkinson´s disease, reveals an enhanced activity upon NEDDylation. Substrates of parkin take part in diverse cellular functions like transcription, neurotransmission, synaptic function or cell cycle control (Choo et al., 2012; Walden and Martinez‐Torres, 2012).
As a component of the Cullin2/elongin B/C complex, pVHL is involved in the regulation of oxygen‐dependent ubiquitination of HIFα (see 1.3.1.1). Interestingly, pVHL is also required for fibronectin matrix assembly, independent of Cullin2 (Stickle et al., 2004). Toggling the binding of pVHL to the cullin complex is achieved by its modification with NEDD8: NEDDylation leads to its interaction with fibronectin whereas its association with Cullin2 is inhibited (Russell and Ohh, 2008).
IAPs (Inhibitor of Apoptosis) are often found to be overexpressed in cancer, thereby contributing to cell proliferation and survival. IAPs are RING ubiquitin E3 ligases that negatively regulate caspase activity and additionally have an influence on cellular survival functions. In Drosophila and humans, effector caspases were identified to act as substrates for NEDDylation by IAPs leading to their inactivation (Broemer et al., 2010). However, it was also supposed that IAPs themselves, rather than caspases, might be substrates for NEDD8 (Nagano et al., 2012).
Very recently, both reduced levels of Ubc12 and inhibition of APPBP1/UBA3 were discovered to impair T‐cell proliferation and cytokine production. Thereupon, NEDDylation of Shc, an adapter
protein between the antigen receptor of T‐cells and the Erk‐pathway, was identified as an important event in T‐cell receptor signaling (Jin et al., 2013).
In proteomic analyses, many further potential substrates for NEDD8 were identified which are mainly involved in mRNA splicing, DNA replication and repair, chromatin remodeling and proteasomal degradation (Jones et al., 2008; Xirodimas et al., 2008).
Finally, one well‐studied interaction partner of NEDD8 which targets NEDD8 and its conjugates for proteasomal degradation is NUB1 (NEDD8 Ultimate Buster1) (Kamitani et al., 2001). By interacting with the S5a subunit of the proteasome, NUB1 not only delivers NEDD8 but also the UBL FAT10 for degradation (Hipp et al., 2004; Tanji et al., 2005). Interaction of NEDD8 with NUB1 additionally results in inhibition of NEDDylation and enhances ubiquitination of p53, leading to its cytoplasmic localization (Liu and Xirodimas, 2010).
1.3.2 NEDD8‐conjugation cascade
The NEDD8‐conjugation cascade is very similar to the one of ubiquitin, involving three specialized types of enzymes which activate NEDD8 and transfer it to a substrate (Figure 5). In humans, NEDD8 is expressed as a precursor protein of 81 aa. The last five amino acids need to be cleaved off by the isopeptidase NEDP1/Den1/SENP8 in order to expose the C‐terminal double glycine motif which is crucial for NEDD8 attachment to substrates (Gan‐Erdene et al., 2003; Kamitani et al., 1997; Wu et al., 2003a). Alternatively, UCHL3, a deubiquitinating enzyme, is capable of processing both NEDD8 and ubiquitin (Wada et al., 1998).
In a first step of the conjugation pathway, NEDD8 is activated by the heterodimeric E1 enzyme APPBP1/UBA3 with UBA3 comprising the catalytically active cysteine residue. Activation includes the formation of a NEDD8 adenylate under consumption of ATP, and the subsequent covalent attachment of NEDD8 C‐terminal glycine to the catalytic cysteine of UBA3, generating a high‐energy thioester bond (Gong and Yeh, 1999). As soon as one NEDD8 is bound in a thioester, a second NEDD8 interacts with the adenylation site of APPBP1/UBA3 and can be activated.
Transfer of NEDD8 to the catalytic cysteine of one of the two conjugating enzymes Ubc12 or Nce2 is only accomplished if the E1 binds two NEDD8 proteins at the same time. In this so called transthiolation reaction, the E2 enzyme takes over NEDD8 from the E1 by the formation of a thioester linkage (Huang et al., 2009; Huang et al., 2007). Finally, an isopeptide bond is formed between the C‐terminal glycine of NEDD8 and a lysine residue of its substrate (Oved et al., 2006;
Wada et al., 1999; Xirodimas et al., 2004). This last step involves the action of an E3 ligase catalyzing the transfer of NEDD8. In contrast to ubiquitin of which two classes of E3 ligases are known, RING ligases seem to be the major protein family playing a role in the NEDDylation cascade (see 1.1.1) (reviewed in (Ardley and Robinson, 2005; Watson et al., 2011)). This type of ligase brings the E2 enzyme and the substrate into proximity thereby allowing the transfer of
NEDD8 to its substrate. However, a direct interaction between E2 and the RING domain of an E3 has only been shown for Ubc12 and the E3 ligases RBX1 and RNF 111 so far (Calabrese et al., 2011; Ma et al., 2013). In contrast to NEDD8 E1 and E2 enzymes, which are highly specific for NEDD8, all RING E3 ligases involved in NEDDylation display a dual specificity for ubiquitin and NEDD8 (Broemer et al., 2010; Huang et al., 2008; Morimoto et al., 2003; Oved et al., 2006;
Walden et al., 2003a; Xirodimas et al., 2004). Nevertheless, there is one unique E3 enzyme for NEDD8 without any known ligase domain: Dcn1/DCNL (yeast/humans), which acts as a scaffold‐
type E3 ligase enhancing NEDDylation of cullins (see 1.3.1.1) (Kurz et al., 2008; Monda et al., 2013).
Figure 5. NEDD8conjugation cascade
In a first step, NEDD8 is activated by its heterodimeric E1 enzyme APPBP1/UBA3 thereby forming a thioester bond.
Subsequently, NEDD8 is transferred to the catalytic cysteine of one of its E2 enzymes, Ubc12 or Nce2. Finally, NEDD8
DeNEDDylation of substrates is predominantly carried out by NEDP1 which cleaves NEDD8 both from cullins and other substrates (Chan et al., 2008; Mendoza et al., 2003). A specific deNEDDylating enzyme complex for cullins is the COP9 signalosome consisting of eight subunits including CSN5 as the catalytically active one (Cope et al., 2002; Lyapina et al., 2001). Moreover, USP21 is able to remove both NEDD8 and ubiquitin from substrates (Gong et al., 2000).
NEDD8 contains nine internal lysine residues at positions 4, 6, 11, 22, 27, 33, 48, 54 and 60, five of which are conserved in ubiquitin. In vitro, mainly K22, K48 and K54 are used for NEDD8 chain formation. K27 and K33 were also found to be modified with NEDD8 although they occurred with a lower abundance (Jeram et al., 2010; Jones et al., 2008). However, an existence of NEDD8 chains under normal cellular conditions still needs to be verified as proteomic analyses were performed using tryptic digest prior to mass spectrometry. Although a double glycine motif was identified at K11, K22, K48 and K60 of NEDD8, it is not clear whether NEDD8 or ubiquitin is attached since they are not distinguishable in this approach (Jones et al., 2008). Additionally, in human cells and in vitro NEDD8 can be attached to ubiquitin but seems to be a bad acceptor for itself or ubiquitin, arguing against the existence of NEDD8 chains (Singh et al., 2012; Whitby et al., 1998).
1.3.2.1 APPBP1/UBA3, the NEDD8‐activating enzyme
In eukaryotes, E1 enzymes consist of three conserved domains: an adenylation domain which is already present in a common E1 ancestor, MoeB, and two eukaryotic specific domains (reviewed in (Hochstrasser, 2000)). One of those evolutionary newer domains contains the catalytic cysteine and the other one, which is located at the C terminus, is involved in transferring the UBL to the E2. This so called “ubiquitin fold domain” (UFD) adopts a ubiquitin‐like structure and plays a crucial role in the interaction with the E2. Comparison of the heterodimeric APPBP1/UBA3 with the ubiquitin E1 UBA1 reveals sequence and structural homology of APPBP1 to the N‐terminal part of UBA1 and of UBA3 to its C‐terminal part (Hochstrasser, 2000;
Lake et al., 2001; Walden et al., 2003b).
The overall structure of APPBP1/UBA3 can be described as a canyon with a groove in the middle. A crossover loop connects the adenylation domain with the catalytic cysteine domain and divides the canyon into two clefts (Walden et al., 2003b). In the structure of APPBP1/UBA3 bound to NEDD8 and ATP, the globular domain of NEDD8 occupies one cleft and its flexible C terminus points towards ATP (Figure 6). Furthermore, there is a bipartite interaction between APPBP1/UBA3 and NEDD8. On the one hand, the conserved acidic surface of NEDD8 contacts a part of the catalytic cysteine domain of APPBP1 forming a polar interface. On the other hand, the hydrophobic surface of NEDD8 interacts with a conserved part of the adenylation domain in UBA3. Interestingly, all residues involved in these hydrophobic interactions are also present in